From 36d22d82aa202bb199967e9512281e9a53db42c9 Mon Sep 17 00:00:00 2001 From: Daniel Baumann Date: Sun, 7 Apr 2024 21:33:14 +0200 Subject: Adding upstream version 115.7.0esr. Signed-off-by: Daniel Baumann --- intl/icu/source/common/BUILD.bazel | 1214 +++++ intl/icu/source/common/Makefile.in | 207 + intl/icu/source/common/appendable.cpp | 74 + intl/icu/source/common/bmpset.cpp | 741 +++ intl/icu/source/common/bmpset.h | 164 + intl/icu/source/common/brkeng.cpp | 304 ++ intl/icu/source/common/brkeng.h | 277 + intl/icu/source/common/brkiter.cpp | 538 ++ intl/icu/source/common/bytesinkutil.cpp | 161 + intl/icu/source/common/bytesinkutil.h | 88 + intl/icu/source/common/bytestream.cpp | 85 + intl/icu/source/common/bytestrie.cpp | 441 ++ intl/icu/source/common/bytestriebuilder.cpp | 512 ++ intl/icu/source/common/bytestrieiterator.cpp | 214 + intl/icu/source/common/caniter.cpp | 586 ++ intl/icu/source/common/capi_helper.h | 97 + intl/icu/source/common/characterproperties.cpp | 419 ++ intl/icu/source/common/chariter.cpp | 100 + intl/icu/source/common/charstr.cpp | 273 + intl/icu/source/common/charstr.h | 193 + intl/icu/source/common/charstrmap.h | 55 + intl/icu/source/common/cmemory.cpp | 138 + intl/icu/source/common/cmemory.h | 900 +++ intl/icu/source/common/common.rc | 110 + intl/icu/source/common/common.vcxproj | 404 ++ intl/icu/source/common/common.vcxproj.filters | 1298 +++++ intl/icu/source/common/common_uwp.vcxproj | 539 ++ intl/icu/source/common/cpputils.h | 97 + intl/icu/source/common/cstr.cpp | 54 + intl/icu/source/common/cstr.h | 60 + intl/icu/source/common/cstring.cpp | 341 ++ intl/icu/source/common/cstring.h | 126 + intl/icu/source/common/cwchar.cpp | 55 + intl/icu/source/common/cwchar.h | 58 + intl/icu/source/common/dictbe.cpp | 1503 +++++ intl/icu/source/common/dictbe.h | 433 ++ intl/icu/source/common/dictionarydata.cpp | 242 + intl/icu/source/common/dictionarydata.h | 191 + intl/icu/source/common/dtintrv.cpp | 63 + intl/icu/source/common/edits.cpp | 804 +++ intl/icu/source/common/emojiprops.cpp | 220 + intl/icu/source/common/emojiprops.h | 90 + intl/icu/source/common/errorcode.cpp | 42 + intl/icu/source/common/filteredbrk.cpp | 736 +++ intl/icu/source/common/filterednormalizer2.cpp | 363 ++ intl/icu/source/common/hash.h | 267 + intl/icu/source/common/icudataver.cpp | 31 + intl/icu/source/common/icuplug.cpp | 884 +++ intl/icu/source/common/icuplugimp.h | 93 + intl/icu/source/common/loadednormalizer2impl.cpp | 418 ++ intl/icu/source/common/localebuilder.cpp | 468 ++ intl/icu/source/common/localefallback_data.h | 1350 +++++ intl/icu/source/common/localematcher.cpp | 826 +++ intl/icu/source/common/localeprioritylist.cpp | 240 + intl/icu/source/common/localeprioritylist.h | 115 + intl/icu/source/common/localsvc.h | 27 + intl/icu/source/common/locavailable.cpp | 270 + intl/icu/source/common/locbased.cpp | 55 + intl/icu/source/common/locbased.h | 107 + intl/icu/source/common/locdispnames.cpp | 902 +++ intl/icu/source/common/locdistance.cpp | 415 ++ intl/icu/source/common/locdistance.h | 151 + intl/icu/source/common/locdspnm.cpp | 1136 ++++ intl/icu/source/common/locid.cpp | 2734 ++++++++++ intl/icu/source/common/loclikely.cpp | 1415 +++++ intl/icu/source/common/loclikelysubtags.cpp | 683 +++ intl/icu/source/common/loclikelysubtags.h | 121 + intl/icu/source/common/locmap.cpp | 1315 +++++ intl/icu/source/common/locmap.h | 40 + intl/icu/source/common/locresdata.cpp | 220 + intl/icu/source/common/locutil.cpp | 275 + intl/icu/source/common/locutil.h | 39 + intl/icu/source/common/lsr.cpp | 114 + intl/icu/source/common/lsr.h | 82 + intl/icu/source/common/lstmbe.cpp | 856 +++ intl/icu/source/common/lstmbe.h | 88 + intl/icu/source/common/messageimpl.h | 65 + intl/icu/source/common/messagepattern.cpp | 1233 +++++ intl/icu/source/common/mlbe.cpp | 270 + intl/icu/source/common/mlbe.h | 116 + intl/icu/source/common/msvcres.h | 25 + intl/icu/source/common/mutex.h | 77 + intl/icu/source/common/norm2_nfc_data.h | 1154 ++++ intl/icu/source/common/norm2allmodes.h | 405 ++ intl/icu/source/common/normalizer2.cpp | 572 ++ intl/icu/source/common/normalizer2impl.cpp | 2806 ++++++++++ intl/icu/source/common/normalizer2impl.h | 987 ++++ intl/icu/source/common/normlzr.cpp | 529 ++ intl/icu/source/common/parsepos.cpp | 23 + intl/icu/source/common/patternprops.cpp | 230 + intl/icu/source/common/patternprops.h | 98 + intl/icu/source/common/pluralmap.cpp | 44 + intl/icu/source/common/pluralmap.h | 292 + intl/icu/source/common/propname.cpp | 334 ++ intl/icu/source/common/propname.h | 212 + intl/icu/source/common/propname_data.h | 1984 +++++++ intl/icu/source/common/propsvec.cpp | 529 ++ intl/icu/source/common/propsvec.h | 178 + intl/icu/source/common/punycode.cpp | 590 ++ intl/icu/source/common/punycode.h | 120 + intl/icu/source/common/putil.cpp | 2506 +++++++++ intl/icu/source/common/putilimp.h | 621 +++ intl/icu/source/common/rbbi.cpp | 1286 +++++ intl/icu/source/common/rbbi_cache.cpp | 701 +++ intl/icu/source/common/rbbi_cache.h | 203 + intl/icu/source/common/rbbicst.pl | 459 ++ intl/icu/source/common/rbbidata.cpp | 476 ++ intl/icu/source/common/rbbidata.h | 212 + intl/icu/source/common/rbbinode.cpp | 373 ++ intl/icu/source/common/rbbinode.h | 127 + intl/icu/source/common/rbbirb.cpp | 361 ++ intl/icu/source/common/rbbirb.h | 237 + intl/icu/source/common/rbbirpt.h | 296 + intl/icu/source/common/rbbirpt.txt | 337 ++ intl/icu/source/common/rbbiscan.cpp | 1285 +++++ intl/icu/source/common/rbbiscan.h | 167 + intl/icu/source/common/rbbisetb.cpp | 694 +++ intl/icu/source/common/rbbisetb.h | 147 + intl/icu/source/common/rbbistbl.cpp | 270 + intl/icu/source/common/rbbitblb.cpp | 1810 +++++++ intl/icu/source/common/rbbitblb.h | 232 + intl/icu/source/common/resbund.cpp | 399 ++ intl/icu/source/common/resbund_cnv.cpp | 57 + intl/icu/source/common/resource.cpp | 22 + intl/icu/source/common/resource.h | 295 + intl/icu/source/common/restrace.cpp | 133 + intl/icu/source/common/restrace.h | 147 + intl/icu/source/common/ruleiter.cpp | 162 + intl/icu/source/common/ruleiter.h | 233 + intl/icu/source/common/schriter.cpp | 119 + intl/icu/source/common/serv.cpp | 968 ++++ intl/icu/source/common/serv.h | 996 ++++ intl/icu/source/common/servlk.cpp | 188 + intl/icu/source/common/servlkf.cpp | 152 + intl/icu/source/common/servloc.h | 551 ++ intl/icu/source/common/servls.cpp | 296 + intl/icu/source/common/servnotf.cpp | 122 + intl/icu/source/common/servnotf.h | 125 + intl/icu/source/common/servrbf.cpp | 96 + intl/icu/source/common/servslkf.cpp | 123 + intl/icu/source/common/sharedobject.cpp | 62 + intl/icu/source/common/sharedobject.h | 184 + intl/icu/source/common/simpleformatter.cpp | 325 ++ intl/icu/source/common/sources.txt | 199 + intl/icu/source/common/sprpimpl.h | 130 + intl/icu/source/common/static_unicode_sets.cpp | 245 + intl/icu/source/common/static_unicode_sets.h | 140 + intl/icu/source/common/stringpiece.cpp | 116 + intl/icu/source/common/stringtriebuilder.cpp | 618 +++ intl/icu/source/common/uarrsort.cpp | 274 + intl/icu/source/common/uarrsort.h | 103 + intl/icu/source/common/uassert.h | 71 + intl/icu/source/common/ubidi.cpp | 3036 +++++++++++ intl/icu/source/common/ubidi_props.cpp | 254 + intl/icu/source/common/ubidi_props.h | 148 + intl/icu/source/common/ubidi_props_data.h | 950 ++++ intl/icu/source/common/ubidiimp.h | 484 ++ intl/icu/source/common/ubidiln.cpp | 1347 +++++ intl/icu/source/common/ubiditransform.cpp | 530 ++ intl/icu/source/common/ubidiwrt.cpp | 650 +++ intl/icu/source/common/ubrk.cpp | 361 ++ intl/icu/source/common/ubrkimpl.h | 15 + intl/icu/source/common/ucase.cpp | 1739 ++++++ intl/icu/source/common/ucase.h | 464 ++ intl/icu/source/common/ucase_props_data.h | 998 ++++ intl/icu/source/common/ucasemap.cpp | 1038 ++++ intl/icu/source/common/ucasemap_imp.h | 282 + .../source/common/ucasemap_titlecase_brkiter.cpp | 134 + intl/icu/source/common/ucat.cpp | 78 + intl/icu/source/common/uchar.cpp | 708 +++ intl/icu/source/common/uchar_props_data.h | 4011 ++++++++++++++ intl/icu/source/common/ucharstrie.cpp | 414 ++ intl/icu/source/common/ucharstriebuilder.cpp | 443 ++ intl/icu/source/common/ucharstrieiterator.cpp | 215 + intl/icu/source/common/uchriter.cpp | 367 ++ intl/icu/source/common/ucln.h | 91 + intl/icu/source/common/ucln_cmn.cpp | 124 + intl/icu/source/common/ucln_cmn.h | 78 + intl/icu/source/common/ucln_imp.h | 182 + intl/icu/source/common/ucmndata.cpp | 393 ++ intl/icu/source/common/ucmndata.h | 117 + intl/icu/source/common/ucnv.cpp | 2922 ++++++++++ intl/icu/source/common/ucnv2022.cpp | 3973 ++++++++++++++ intl/icu/source/common/ucnv_bld.cpp | 1689 ++++++ intl/icu/source/common/ucnv_bld.h | 296 + intl/icu/source/common/ucnv_cb.cpp | 261 + intl/icu/source/common/ucnv_cnv.cpp | 182 + intl/icu/source/common/ucnv_cnv.h | 323 ++ intl/icu/source/common/ucnv_ct.cpp | 646 +++ intl/icu/source/common/ucnv_err.cpp | 486 ++ intl/icu/source/common/ucnv_ext.cpp | 1143 ++++ intl/icu/source/common/ucnv_ext.h | 481 ++ intl/icu/source/common/ucnv_imp.h | 139 + intl/icu/source/common/ucnv_io.cpp | 1360 +++++ intl/icu/source/common/ucnv_io.h | 127 + intl/icu/source/common/ucnv_lmb.cpp | 1388 +++++ intl/icu/source/common/ucnv_set.cpp | 70 + intl/icu/source/common/ucnv_u16.cpp | 1579 ++++++ intl/icu/source/common/ucnv_u32.cpp | 1253 +++++ intl/icu/source/common/ucnv_u7.cpp | 1491 +++++ intl/icu/source/common/ucnv_u8.cpp | 944 ++++ intl/icu/source/common/ucnvbocu.cpp | 1413 +++++ intl/icu/source/common/ucnvdisp.cpp | 88 + intl/icu/source/common/ucnvhz.cpp | 625 +++ intl/icu/source/common/ucnvisci.cpp | 1635 ++++++ intl/icu/source/common/ucnvlat1.cpp | 756 +++ intl/icu/source/common/ucnvmbcs.cpp | 5723 ++++++++++++++++++++ intl/icu/source/common/ucnvmbcs.h | 605 +++ intl/icu/source/common/ucnvscsu.cpp | 2045 +++++++ intl/icu/source/common/ucnvsel.cpp | 823 +++ intl/icu/source/common/ucol_data.h | 89 + intl/icu/source/common/ucol_swp.cpp | 615 +++ intl/icu/source/common/ucol_swp.h | 58 + intl/icu/source/common/ucptrie.cpp | 601 ++ intl/icu/source/common/ucptrie_impl.h | 285 + intl/icu/source/common/ucurr.cpp | 2727 ++++++++++ intl/icu/source/common/ucurrimp.h | 78 + intl/icu/source/common/udata.cpp | 1460 +++++ intl/icu/source/common/udatamem.cpp | 161 + intl/icu/source/common/udatamem.h | 61 + intl/icu/source/common/udataswp.cpp | 473 ++ intl/icu/source/common/udataswp.h | 404 ++ intl/icu/source/common/uelement.h | 94 + intl/icu/source/common/uenum.cpp | 189 + intl/icu/source/common/uenumimp.h | 155 + intl/icu/source/common/uhash.cpp | 1066 ++++ intl/icu/source/common/uhash.h | 811 +++ intl/icu/source/common/uhash_us.cpp | 26 + intl/icu/source/common/uidna.cpp | 921 ++++ intl/icu/source/common/uinit.cpp | 74 + intl/icu/source/common/uinvchar.cpp | 627 +++ intl/icu/source/common/uinvchar.h | 219 + intl/icu/source/common/uiter.cpp | 1108 ++++ intl/icu/source/common/ulayout_props.h | 46 + intl/icu/source/common/ulist.cpp | 270 + intl/icu/source/common/ulist.h | 50 + intl/icu/source/common/uloc.cpp | 2228 ++++++++ intl/icu/source/common/uloc_keytype.cpp | 536 ++ intl/icu/source/common/uloc_tag.cpp | 2869 ++++++++++ intl/icu/source/common/ulocimp.h | 378 ++ intl/icu/source/common/umapfile.cpp | 530 ++ intl/icu/source/common/umapfile.h | 59 + intl/icu/source/common/umath.cpp | 26 + intl/icu/source/common/umutablecptrie.cpp | 1852 +++++++ intl/icu/source/common/umutex.cpp | 224 + intl/icu/source/common/umutex.h | 313 ++ intl/icu/source/common/unames.cpp | 2108 +++++++ intl/icu/source/common/unicode/appendable.h | 239 + intl/icu/source/common/unicode/brkiter.h | 670 +++ intl/icu/source/common/unicode/bytestream.h | 307 ++ intl/icu/source/common/unicode/bytestrie.h | 568 ++ intl/icu/source/common/unicode/bytestriebuilder.h | 193 + intl/icu/source/common/unicode/caniter.h | 214 + intl/icu/source/common/unicode/casemap.h | 497 ++ intl/icu/source/common/unicode/char16ptr.h | 313 ++ intl/icu/source/common/unicode/chariter.h | 734 +++ intl/icu/source/common/unicode/dbbi.h | 48 + intl/icu/source/common/unicode/docmain.h | 237 + intl/icu/source/common/unicode/dtintrv.h | 164 + intl/icu/source/common/unicode/edits.h | 531 ++ intl/icu/source/common/unicode/enumset.h | 69 + intl/icu/source/common/unicode/errorcode.h | 144 + intl/icu/source/common/unicode/filteredbrk.h | 152 + intl/icu/source/common/unicode/icudataver.h | 43 + intl/icu/source/common/unicode/icuplug.h | 391 ++ intl/icu/source/common/unicode/idna.h | 330 ++ intl/icu/source/common/unicode/localebuilder.h | 309 ++ intl/icu/source/common/unicode/localematcher.h | 708 +++ intl/icu/source/common/unicode/localpointer.h | 595 ++ intl/icu/source/common/unicode/locdspnm.h | 211 + intl/icu/source/common/unicode/locid.h | 1272 +++++ intl/icu/source/common/unicode/messagepattern.h | 949 ++++ intl/icu/source/common/unicode/normalizer2.h | 771 +++ intl/icu/source/common/unicode/normlzr.h | 816 +++ intl/icu/source/common/unicode/parseerr.h | 94 + intl/icu/source/common/unicode/parsepos.h | 237 + intl/icu/source/common/unicode/platform.h | 881 +++ intl/icu/source/common/unicode/ptypes.h | 130 + intl/icu/source/common/unicode/putil.h | 183 + intl/icu/source/common/unicode/rbbi.h | 745 +++ intl/icu/source/common/unicode/rep.h | 266 + intl/icu/source/common/unicode/resbund.h | 498 ++ intl/icu/source/common/unicode/schriter.h | 187 + intl/icu/source/common/unicode/simpleformatter.h | 341 ++ intl/icu/source/common/unicode/std_string.h | 41 + intl/icu/source/common/unicode/strenum.h | 281 + intl/icu/source/common/unicode/stringoptions.h | 190 + intl/icu/source/common/unicode/stringpiece.h | 343 ++ intl/icu/source/common/unicode/stringtriebuilder.h | 426 ++ intl/icu/source/common/unicode/symtable.h | 119 + intl/icu/source/common/unicode/ubidi.h | 2211 ++++++++ intl/icu/source/common/unicode/ubiditransform.h | 326 ++ intl/icu/source/common/unicode/ubrk.h | 647 +++ intl/icu/source/common/unicode/ucasemap.h | 388 ++ intl/icu/source/common/unicode/ucat.h | 160 + intl/icu/source/common/unicode/uchar.h | 4169 ++++++++++++++ intl/icu/source/common/unicode/ucharstrie.h | 623 +++ intl/icu/source/common/unicode/ucharstriebuilder.h | 193 + intl/icu/source/common/unicode/uchriter.h | 393 ++ intl/icu/source/common/unicode/uclean.h | 262 + intl/icu/source/common/unicode/ucnv.h | 2056 +++++++ intl/icu/source/common/unicode/ucnv_cb.h | 164 + intl/icu/source/common/unicode/ucnv_err.h | 465 ++ intl/icu/source/common/unicode/ucnvsel.h | 193 + intl/icu/source/common/unicode/uconfig.h | 466 ++ intl/icu/source/common/unicode/ucpmap.h | 158 + intl/icu/source/common/unicode/ucptrie.h | 645 +++ intl/icu/source/common/unicode/ucurr.h | 466 ++ intl/icu/source/common/unicode/udata.h | 440 ++ intl/icu/source/common/unicode/udisplaycontext.h | 173 + intl/icu/source/common/unicode/uenum.h | 209 + intl/icu/source/common/unicode/uidna.h | 776 +++ intl/icu/source/common/unicode/uiter.h | 709 +++ intl/icu/source/common/unicode/uldnames.h | 307 ++ intl/icu/source/common/unicode/uloc.h | 1393 +++++ intl/icu/source/common/unicode/umachine.h | 459 ++ intl/icu/source/common/unicode/umisc.h | 62 + intl/icu/source/common/unicode/umutablecptrie.h | 240 + intl/icu/source/common/unicode/unifilt.h | 136 + intl/icu/source/common/unicode/unifunct.h | 132 + intl/icu/source/common/unicode/unimatch.h | 168 + intl/icu/source/common/unicode/uniset.h | 1793 ++++++ intl/icu/source/common/unicode/unistr.h | 4784 ++++++++++++++++ intl/icu/source/common/unicode/unorm.h | 476 ++ intl/icu/source/common/unicode/unorm2.h | 606 +++ intl/icu/source/common/unicode/uobject.h | 324 ++ intl/icu/source/common/unicode/urename.h | 1975 +++++++ intl/icu/source/common/unicode/urep.h | 157 + intl/icu/source/common/unicode/ures.h | 911 ++++ intl/icu/source/common/unicode/uscript.h | 724 +++ intl/icu/source/common/unicode/uset.h | 1290 +++++ intl/icu/source/common/unicode/usetiter.h | 323 ++ intl/icu/source/common/unicode/ushape.h | 476 ++ intl/icu/source/common/unicode/usprep.h | 274 + intl/icu/source/common/unicode/ustring.h | 1685 ++++++ intl/icu/source/common/unicode/ustringtrie.h | 97 + intl/icu/source/common/unicode/utext.h | 1603 ++++++ intl/icu/source/common/unicode/utf.h | 225 + intl/icu/source/common/unicode/utf16.h | 734 +++ intl/icu/source/common/unicode/utf32.h | 25 + intl/icu/source/common/unicode/utf8.h | 882 +++ intl/icu/source/common/unicode/utf_old.h | 1201 ++++ intl/icu/source/common/unicode/utrace.h | 506 ++ intl/icu/source/common/unicode/utypes.h | 730 +++ intl/icu/source/common/unicode/uvernum.h | 191 + intl/icu/source/common/unicode/uversion.h | 187 + intl/icu/source/common/unifiedcache.cpp | 554 ++ intl/icu/source/common/unifiedcache.h | 556 ++ intl/icu/source/common/unifilt.cpp | 71 + intl/icu/source/common/unifunct.cpp | 28 + intl/icu/source/common/uniquecharstr.h | 98 + intl/icu/source/common/uniset.cpp | 2355 ++++++++ intl/icu/source/common/uniset_closure.cpp | 358 ++ intl/icu/source/common/uniset_props.cpp | 1139 ++++ intl/icu/source/common/unisetspan.cpp | 1522 ++++++ intl/icu/source/common/unisetspan.h | 157 + intl/icu/source/common/unistr.cpp | 2030 +++++++ intl/icu/source/common/unistr_case.cpp | 250 + intl/icu/source/common/unistr_case_locale.cpp | 56 + intl/icu/source/common/unistr_cnv.cpp | 417 ++ intl/icu/source/common/unistr_props.cpp | 77 + .../icu/source/common/unistr_titlecase_brkiter.cpp | 57 + intl/icu/source/common/unistrappender.h | 90 + intl/icu/source/common/unorm.cpp | 280 + intl/icu/source/common/unormcmp.cpp | 640 +++ intl/icu/source/common/unormimp.h | 488 ++ intl/icu/source/common/uobject.cpp | 105 + intl/icu/source/common/uposixdefs.h | 77 + intl/icu/source/common/uprops.cpp | 843 +++ intl/icu/source/common/uprops.h | 455 ++ intl/icu/source/common/ures_cnv.cpp | 78 + intl/icu/source/common/uresbund.cpp | 3462 ++++++++++++ intl/icu/source/common/uresdata.cpp | 1510 ++++++ intl/icu/source/common/uresdata.h | 574 ++ intl/icu/source/common/uresimp.h | 397 ++ intl/icu/source/common/ureslocs.h | 27 + intl/icu/source/common/usc_impl.cpp | 361 ++ intl/icu/source/common/usc_impl.h | 139 + intl/icu/source/common/uscript.cpp | 162 + intl/icu/source/common/uscript_props.cpp | 309 ++ intl/icu/source/common/uset.cpp | 681 +++ intl/icu/source/common/uset_imp.h | 71 + intl/icu/source/common/uset_props.cpp | 143 + intl/icu/source/common/usetiter.cpp | 152 + intl/icu/source/common/ushape.cpp | 1728 ++++++ intl/icu/source/common/usprep.cpp | 871 +++ intl/icu/source/common/ustack.cpp | 62 + intl/icu/source/common/ustr_cnv.cpp | 256 + intl/icu/source/common/ustr_cnv.h | 51 + intl/icu/source/common/ustr_imp.h | 155 + intl/icu/source/common/ustr_titlecase_brkiter.cpp | 237 + intl/icu/source/common/ustr_wcs.cpp | 535 ++ intl/icu/source/common/ustrcase.cpp | 1900 +++++++ intl/icu/source/common/ustrcase_locale.cpp | 94 + intl/icu/source/common/ustrenum.cpp | 398 ++ intl/icu/source/common/ustrenum.h | 87 + intl/icu/source/common/ustrfmt.cpp | 59 + intl/icu/source/common/ustrfmt.h | 19 + intl/icu/source/common/ustring.cpp | 1535 ++++++ intl/icu/source/common/ustrtrns.cpp | 1451 +++++ intl/icu/source/common/utext.cpp | 2880 ++++++++++ intl/icu/source/common/utf_impl.cpp | 329 ++ intl/icu/source/common/util.cpp | 435 ++ intl/icu/source/common/util.h | 267 + intl/icu/source/common/util_props.cpp | 217 + intl/icu/source/common/utrace.cpp | 504 ++ intl/icu/source/common/utracimp.h | 391 ++ intl/icu/source/common/utrie.cpp | 1234 +++++ intl/icu/source/common/utrie.h | 793 +++ intl/icu/source/common/utrie2.cpp | 663 +++ intl/icu/source/common/utrie2.h | 955 ++++ intl/icu/source/common/utrie2_builder.cpp | 1483 +++++ intl/icu/source/common/utrie2_impl.h | 175 + intl/icu/source/common/utrie_swap.cpp | 348 ++ intl/icu/source/common/uts46.cpp | 1494 +++++ intl/icu/source/common/utypeinfo.h | 32 + intl/icu/source/common/utypes.cpp | 227 + intl/icu/source/common/uvector.cpp | 566 ++ intl/icu/source/common/uvector.h | 386 ++ intl/icu/source/common/uvectr32.cpp | 335 ++ intl/icu/source/common/uvectr32.h | 306 ++ intl/icu/source/common/uvectr64.cpp | 214 + intl/icu/source/common/uvectr64.h | 279 + intl/icu/source/common/wintz.cpp | 331 ++ intl/icu/source/common/wintz.h | 36 + 425 files changed, 245471 insertions(+) create mode 100644 intl/icu/source/common/BUILD.bazel create mode 100644 intl/icu/source/common/Makefile.in create mode 100644 intl/icu/source/common/appendable.cpp create mode 100644 intl/icu/source/common/bmpset.cpp create mode 100644 intl/icu/source/common/bmpset.h create mode 100644 intl/icu/source/common/brkeng.cpp create mode 100644 intl/icu/source/common/brkeng.h create mode 100644 intl/icu/source/common/brkiter.cpp create mode 100644 intl/icu/source/common/bytesinkutil.cpp create mode 100644 intl/icu/source/common/bytesinkutil.h create mode 100644 intl/icu/source/common/bytestream.cpp create mode 100644 intl/icu/source/common/bytestrie.cpp create mode 100644 intl/icu/source/common/bytestriebuilder.cpp create mode 100644 intl/icu/source/common/bytestrieiterator.cpp create mode 100644 intl/icu/source/common/caniter.cpp create mode 100644 intl/icu/source/common/capi_helper.h create mode 100644 intl/icu/source/common/characterproperties.cpp create mode 100644 intl/icu/source/common/chariter.cpp create mode 100644 intl/icu/source/common/charstr.cpp create mode 100644 intl/icu/source/common/charstr.h create mode 100644 intl/icu/source/common/charstrmap.h create mode 100644 intl/icu/source/common/cmemory.cpp create mode 100644 intl/icu/source/common/cmemory.h create mode 100644 intl/icu/source/common/common.rc create mode 100644 intl/icu/source/common/common.vcxproj create mode 100644 intl/icu/source/common/common.vcxproj.filters create mode 100644 intl/icu/source/common/common_uwp.vcxproj create mode 100644 intl/icu/source/common/cpputils.h create mode 100644 intl/icu/source/common/cstr.cpp create mode 100644 intl/icu/source/common/cstr.h create mode 100644 intl/icu/source/common/cstring.cpp create mode 100644 intl/icu/source/common/cstring.h create mode 100644 intl/icu/source/common/cwchar.cpp create mode 100644 intl/icu/source/common/cwchar.h create mode 100644 intl/icu/source/common/dictbe.cpp create mode 100644 intl/icu/source/common/dictbe.h create mode 100644 intl/icu/source/common/dictionarydata.cpp create mode 100644 intl/icu/source/common/dictionarydata.h create mode 100644 intl/icu/source/common/dtintrv.cpp create mode 100644 intl/icu/source/common/edits.cpp create mode 100644 intl/icu/source/common/emojiprops.cpp create mode 100644 intl/icu/source/common/emojiprops.h create mode 100644 intl/icu/source/common/errorcode.cpp create mode 100644 intl/icu/source/common/filteredbrk.cpp create mode 100644 intl/icu/source/common/filterednormalizer2.cpp create mode 100644 intl/icu/source/common/hash.h create mode 100644 intl/icu/source/common/icudataver.cpp create mode 100644 intl/icu/source/common/icuplug.cpp create mode 100644 intl/icu/source/common/icuplugimp.h create mode 100644 intl/icu/source/common/loadednormalizer2impl.cpp create mode 100644 intl/icu/source/common/localebuilder.cpp create mode 100644 intl/icu/source/common/localefallback_data.h create mode 100644 intl/icu/source/common/localematcher.cpp create mode 100644 intl/icu/source/common/localeprioritylist.cpp create mode 100644 intl/icu/source/common/localeprioritylist.h create mode 100644 intl/icu/source/common/localsvc.h create mode 100644 intl/icu/source/common/locavailable.cpp create mode 100644 intl/icu/source/common/locbased.cpp create mode 100644 intl/icu/source/common/locbased.h create mode 100644 intl/icu/source/common/locdispnames.cpp create mode 100644 intl/icu/source/common/locdistance.cpp create mode 100644 intl/icu/source/common/locdistance.h create mode 100644 intl/icu/source/common/locdspnm.cpp create mode 100644 intl/icu/source/common/locid.cpp create mode 100644 intl/icu/source/common/loclikely.cpp create mode 100644 intl/icu/source/common/loclikelysubtags.cpp create mode 100644 intl/icu/source/common/loclikelysubtags.h create mode 100644 intl/icu/source/common/locmap.cpp create mode 100644 intl/icu/source/common/locmap.h create mode 100644 intl/icu/source/common/locresdata.cpp create mode 100644 intl/icu/source/common/locutil.cpp create mode 100644 intl/icu/source/common/locutil.h create mode 100644 intl/icu/source/common/lsr.cpp create mode 100644 intl/icu/source/common/lsr.h create mode 100644 intl/icu/source/common/lstmbe.cpp create mode 100644 intl/icu/source/common/lstmbe.h create mode 100644 intl/icu/source/common/messageimpl.h create mode 100644 intl/icu/source/common/messagepattern.cpp create mode 100644 intl/icu/source/common/mlbe.cpp create mode 100644 intl/icu/source/common/mlbe.h create mode 100644 intl/icu/source/common/msvcres.h create mode 100644 intl/icu/source/common/mutex.h create mode 100644 intl/icu/source/common/norm2_nfc_data.h create mode 100644 intl/icu/source/common/norm2allmodes.h create mode 100644 intl/icu/source/common/normalizer2.cpp create mode 100644 intl/icu/source/common/normalizer2impl.cpp create mode 100644 intl/icu/source/common/normalizer2impl.h create mode 100644 intl/icu/source/common/normlzr.cpp create mode 100644 intl/icu/source/common/parsepos.cpp create mode 100644 intl/icu/source/common/patternprops.cpp create mode 100644 intl/icu/source/common/patternprops.h create mode 100644 intl/icu/source/common/pluralmap.cpp create mode 100644 intl/icu/source/common/pluralmap.h create mode 100644 intl/icu/source/common/propname.cpp create mode 100644 intl/icu/source/common/propname.h create mode 100644 intl/icu/source/common/propname_data.h create mode 100644 intl/icu/source/common/propsvec.cpp create mode 100644 intl/icu/source/common/propsvec.h create mode 100644 intl/icu/source/common/punycode.cpp create mode 100644 intl/icu/source/common/punycode.h create mode 100644 intl/icu/source/common/putil.cpp create mode 100644 intl/icu/source/common/putilimp.h create mode 100644 intl/icu/source/common/rbbi.cpp create mode 100644 intl/icu/source/common/rbbi_cache.cpp create mode 100644 intl/icu/source/common/rbbi_cache.h create mode 100755 intl/icu/source/common/rbbicst.pl create mode 100644 intl/icu/source/common/rbbidata.cpp create mode 100644 intl/icu/source/common/rbbidata.h create mode 100644 intl/icu/source/common/rbbinode.cpp create mode 100644 intl/icu/source/common/rbbinode.h create mode 100644 intl/icu/source/common/rbbirb.cpp create mode 100644 intl/icu/source/common/rbbirb.h create mode 100644 intl/icu/source/common/rbbirpt.h create mode 100644 intl/icu/source/common/rbbirpt.txt create mode 100644 intl/icu/source/common/rbbiscan.cpp create mode 100644 intl/icu/source/common/rbbiscan.h create mode 100644 intl/icu/source/common/rbbisetb.cpp create mode 100644 intl/icu/source/common/rbbisetb.h create mode 100644 intl/icu/source/common/rbbistbl.cpp create mode 100644 intl/icu/source/common/rbbitblb.cpp create mode 100644 intl/icu/source/common/rbbitblb.h create mode 100644 intl/icu/source/common/resbund.cpp create mode 100644 intl/icu/source/common/resbund_cnv.cpp create mode 100644 intl/icu/source/common/resource.cpp create mode 100644 intl/icu/source/common/resource.h create mode 100644 intl/icu/source/common/restrace.cpp create mode 100644 intl/icu/source/common/restrace.h create mode 100644 intl/icu/source/common/ruleiter.cpp create mode 100644 intl/icu/source/common/ruleiter.h create mode 100644 intl/icu/source/common/schriter.cpp create mode 100644 intl/icu/source/common/serv.cpp create mode 100644 intl/icu/source/common/serv.h create mode 100644 intl/icu/source/common/servlk.cpp create mode 100644 intl/icu/source/common/servlkf.cpp create mode 100644 intl/icu/source/common/servloc.h create mode 100644 intl/icu/source/common/servls.cpp create mode 100644 intl/icu/source/common/servnotf.cpp create mode 100644 intl/icu/source/common/servnotf.h create mode 100644 intl/icu/source/common/servrbf.cpp create mode 100644 intl/icu/source/common/servslkf.cpp create mode 100644 intl/icu/source/common/sharedobject.cpp create mode 100644 intl/icu/source/common/sharedobject.h create mode 100644 intl/icu/source/common/simpleformatter.cpp create mode 100644 intl/icu/source/common/sources.txt create mode 100644 intl/icu/source/common/sprpimpl.h create mode 100644 intl/icu/source/common/static_unicode_sets.cpp create mode 100644 intl/icu/source/common/static_unicode_sets.h create mode 100644 intl/icu/source/common/stringpiece.cpp create mode 100644 intl/icu/source/common/stringtriebuilder.cpp create mode 100644 intl/icu/source/common/uarrsort.cpp create mode 100644 intl/icu/source/common/uarrsort.h create mode 100644 intl/icu/source/common/uassert.h create mode 100644 intl/icu/source/common/ubidi.cpp create mode 100644 intl/icu/source/common/ubidi_props.cpp create mode 100644 intl/icu/source/common/ubidi_props.h create mode 100644 intl/icu/source/common/ubidi_props_data.h create mode 100644 intl/icu/source/common/ubidiimp.h create mode 100644 intl/icu/source/common/ubidiln.cpp create mode 100644 intl/icu/source/common/ubiditransform.cpp create mode 100644 intl/icu/source/common/ubidiwrt.cpp create mode 100644 intl/icu/source/common/ubrk.cpp create mode 100644 intl/icu/source/common/ubrkimpl.h create mode 100644 intl/icu/source/common/ucase.cpp create mode 100644 intl/icu/source/common/ucase.h create mode 100644 intl/icu/source/common/ucase_props_data.h create mode 100644 intl/icu/source/common/ucasemap.cpp create mode 100644 intl/icu/source/common/ucasemap_imp.h create mode 100644 intl/icu/source/common/ucasemap_titlecase_brkiter.cpp create mode 100644 intl/icu/source/common/ucat.cpp create mode 100644 intl/icu/source/common/uchar.cpp create mode 100644 intl/icu/source/common/uchar_props_data.h create mode 100644 intl/icu/source/common/ucharstrie.cpp create mode 100644 intl/icu/source/common/ucharstriebuilder.cpp create mode 100644 intl/icu/source/common/ucharstrieiterator.cpp create mode 100644 intl/icu/source/common/uchriter.cpp create mode 100644 intl/icu/source/common/ucln.h create mode 100644 intl/icu/source/common/ucln_cmn.cpp create mode 100644 intl/icu/source/common/ucln_cmn.h create mode 100644 intl/icu/source/common/ucln_imp.h create mode 100644 intl/icu/source/common/ucmndata.cpp create mode 100644 intl/icu/source/common/ucmndata.h create mode 100644 intl/icu/source/common/ucnv.cpp create mode 100644 intl/icu/source/common/ucnv2022.cpp create mode 100644 intl/icu/source/common/ucnv_bld.cpp create mode 100644 intl/icu/source/common/ucnv_bld.h create mode 100644 intl/icu/source/common/ucnv_cb.cpp create mode 100644 intl/icu/source/common/ucnv_cnv.cpp create mode 100644 intl/icu/source/common/ucnv_cnv.h create mode 100644 intl/icu/source/common/ucnv_ct.cpp create mode 100644 intl/icu/source/common/ucnv_err.cpp create mode 100644 intl/icu/source/common/ucnv_ext.cpp create mode 100644 intl/icu/source/common/ucnv_ext.h create mode 100644 intl/icu/source/common/ucnv_imp.h create mode 100644 intl/icu/source/common/ucnv_io.cpp create mode 100644 intl/icu/source/common/ucnv_io.h create mode 100644 intl/icu/source/common/ucnv_lmb.cpp create mode 100644 intl/icu/source/common/ucnv_set.cpp create mode 100644 intl/icu/source/common/ucnv_u16.cpp create mode 100644 intl/icu/source/common/ucnv_u32.cpp create mode 100644 intl/icu/source/common/ucnv_u7.cpp create mode 100644 intl/icu/source/common/ucnv_u8.cpp create mode 100644 intl/icu/source/common/ucnvbocu.cpp create mode 100644 intl/icu/source/common/ucnvdisp.cpp create mode 100644 intl/icu/source/common/ucnvhz.cpp create mode 100644 intl/icu/source/common/ucnvisci.cpp create mode 100644 intl/icu/source/common/ucnvlat1.cpp create mode 100644 intl/icu/source/common/ucnvmbcs.cpp create mode 100644 intl/icu/source/common/ucnvmbcs.h create mode 100644 intl/icu/source/common/ucnvscsu.cpp create mode 100644 intl/icu/source/common/ucnvsel.cpp create mode 100644 intl/icu/source/common/ucol_data.h create mode 100644 intl/icu/source/common/ucol_swp.cpp create mode 100644 intl/icu/source/common/ucol_swp.h create mode 100644 intl/icu/source/common/ucptrie.cpp create mode 100644 intl/icu/source/common/ucptrie_impl.h create mode 100644 intl/icu/source/common/ucurr.cpp create mode 100644 intl/icu/source/common/ucurrimp.h create mode 100644 intl/icu/source/common/udata.cpp create mode 100644 intl/icu/source/common/udatamem.cpp create mode 100644 intl/icu/source/common/udatamem.h create mode 100644 intl/icu/source/common/udataswp.cpp create mode 100644 intl/icu/source/common/udataswp.h create mode 100644 intl/icu/source/common/uelement.h create mode 100644 intl/icu/source/common/uenum.cpp create mode 100644 intl/icu/source/common/uenumimp.h create mode 100644 intl/icu/source/common/uhash.cpp create mode 100644 intl/icu/source/common/uhash.h create mode 100644 intl/icu/source/common/uhash_us.cpp create mode 100644 intl/icu/source/common/uidna.cpp create mode 100644 intl/icu/source/common/uinit.cpp create mode 100644 intl/icu/source/common/uinvchar.cpp create mode 100644 intl/icu/source/common/uinvchar.h create mode 100644 intl/icu/source/common/uiter.cpp create mode 100644 intl/icu/source/common/ulayout_props.h create mode 100644 intl/icu/source/common/ulist.cpp create mode 100644 intl/icu/source/common/ulist.h create mode 100644 intl/icu/source/common/uloc.cpp create mode 100644 intl/icu/source/common/uloc_keytype.cpp create mode 100644 intl/icu/source/common/uloc_tag.cpp create mode 100644 intl/icu/source/common/ulocimp.h create mode 100644 intl/icu/source/common/umapfile.cpp create mode 100644 intl/icu/source/common/umapfile.h create mode 100644 intl/icu/source/common/umath.cpp create mode 100644 intl/icu/source/common/umutablecptrie.cpp create mode 100644 intl/icu/source/common/umutex.cpp create mode 100644 intl/icu/source/common/umutex.h create mode 100644 intl/icu/source/common/unames.cpp create mode 100644 intl/icu/source/common/unicode/appendable.h create mode 100644 intl/icu/source/common/unicode/brkiter.h create mode 100644 intl/icu/source/common/unicode/bytestream.h create mode 100644 intl/icu/source/common/unicode/bytestrie.h create mode 100644 intl/icu/source/common/unicode/bytestriebuilder.h create mode 100644 intl/icu/source/common/unicode/caniter.h create mode 100644 intl/icu/source/common/unicode/casemap.h create mode 100644 intl/icu/source/common/unicode/char16ptr.h create mode 100644 intl/icu/source/common/unicode/chariter.h create mode 100644 intl/icu/source/common/unicode/dbbi.h create mode 100644 intl/icu/source/common/unicode/docmain.h create mode 100644 intl/icu/source/common/unicode/dtintrv.h create mode 100644 intl/icu/source/common/unicode/edits.h create mode 100644 intl/icu/source/common/unicode/enumset.h create mode 100644 intl/icu/source/common/unicode/errorcode.h create mode 100644 intl/icu/source/common/unicode/filteredbrk.h create mode 100644 intl/icu/source/common/unicode/icudataver.h create mode 100644 intl/icu/source/common/unicode/icuplug.h create mode 100644 intl/icu/source/common/unicode/idna.h create mode 100644 intl/icu/source/common/unicode/localebuilder.h create mode 100644 intl/icu/source/common/unicode/localematcher.h create mode 100644 intl/icu/source/common/unicode/localpointer.h create mode 100644 intl/icu/source/common/unicode/locdspnm.h create mode 100644 intl/icu/source/common/unicode/locid.h create mode 100644 intl/icu/source/common/unicode/messagepattern.h create mode 100644 intl/icu/source/common/unicode/normalizer2.h create mode 100644 intl/icu/source/common/unicode/normlzr.h create mode 100644 intl/icu/source/common/unicode/parseerr.h create mode 100644 intl/icu/source/common/unicode/parsepos.h create mode 100644 intl/icu/source/common/unicode/platform.h create mode 100644 intl/icu/source/common/unicode/ptypes.h create mode 100644 intl/icu/source/common/unicode/putil.h create mode 100644 intl/icu/source/common/unicode/rbbi.h create mode 100644 intl/icu/source/common/unicode/rep.h create mode 100644 intl/icu/source/common/unicode/resbund.h create mode 100644 intl/icu/source/common/unicode/schriter.h create mode 100644 intl/icu/source/common/unicode/simpleformatter.h create mode 100644 intl/icu/source/common/unicode/std_string.h create mode 100644 intl/icu/source/common/unicode/strenum.h create mode 100644 intl/icu/source/common/unicode/stringoptions.h create mode 100644 intl/icu/source/common/unicode/stringpiece.h create mode 100644 intl/icu/source/common/unicode/stringtriebuilder.h create mode 100644 intl/icu/source/common/unicode/symtable.h create mode 100644 intl/icu/source/common/unicode/ubidi.h create mode 100644 intl/icu/source/common/unicode/ubiditransform.h create mode 100644 intl/icu/source/common/unicode/ubrk.h create mode 100644 intl/icu/source/common/unicode/ucasemap.h create mode 100644 intl/icu/source/common/unicode/ucat.h create mode 100644 intl/icu/source/common/unicode/uchar.h create mode 100644 intl/icu/source/common/unicode/ucharstrie.h create mode 100644 intl/icu/source/common/unicode/ucharstriebuilder.h create mode 100644 intl/icu/source/common/unicode/uchriter.h create mode 100644 intl/icu/source/common/unicode/uclean.h create mode 100644 intl/icu/source/common/unicode/ucnv.h create mode 100644 intl/icu/source/common/unicode/ucnv_cb.h create mode 100644 intl/icu/source/common/unicode/ucnv_err.h create mode 100644 intl/icu/source/common/unicode/ucnvsel.h create mode 100644 intl/icu/source/common/unicode/uconfig.h create mode 100644 intl/icu/source/common/unicode/ucpmap.h create mode 100644 intl/icu/source/common/unicode/ucptrie.h create mode 100644 intl/icu/source/common/unicode/ucurr.h create mode 100644 intl/icu/source/common/unicode/udata.h create mode 100644 intl/icu/source/common/unicode/udisplaycontext.h create mode 100644 intl/icu/source/common/unicode/uenum.h create mode 100644 intl/icu/source/common/unicode/uidna.h create mode 100644 intl/icu/source/common/unicode/uiter.h create mode 100644 intl/icu/source/common/unicode/uldnames.h create mode 100644 intl/icu/source/common/unicode/uloc.h create mode 100644 intl/icu/source/common/unicode/umachine.h create mode 100644 intl/icu/source/common/unicode/umisc.h create mode 100644 intl/icu/source/common/unicode/umutablecptrie.h create mode 100644 intl/icu/source/common/unicode/unifilt.h create mode 100644 intl/icu/source/common/unicode/unifunct.h create mode 100644 intl/icu/source/common/unicode/unimatch.h create mode 100644 intl/icu/source/common/unicode/uniset.h create mode 100644 intl/icu/source/common/unicode/unistr.h create mode 100644 intl/icu/source/common/unicode/unorm.h create mode 100644 intl/icu/source/common/unicode/unorm2.h create mode 100644 intl/icu/source/common/unicode/uobject.h create mode 100644 intl/icu/source/common/unicode/urename.h create mode 100644 intl/icu/source/common/unicode/urep.h create mode 100644 intl/icu/source/common/unicode/ures.h create mode 100644 intl/icu/source/common/unicode/uscript.h create mode 100644 intl/icu/source/common/unicode/uset.h create mode 100644 intl/icu/source/common/unicode/usetiter.h create mode 100644 intl/icu/source/common/unicode/ushape.h create mode 100644 intl/icu/source/common/unicode/usprep.h create mode 100644 intl/icu/source/common/unicode/ustring.h create mode 100644 intl/icu/source/common/unicode/ustringtrie.h create mode 100644 intl/icu/source/common/unicode/utext.h create mode 100644 intl/icu/source/common/unicode/utf.h create mode 100644 intl/icu/source/common/unicode/utf16.h create mode 100644 intl/icu/source/common/unicode/utf32.h create mode 100644 intl/icu/source/common/unicode/utf8.h create mode 100644 intl/icu/source/common/unicode/utf_old.h create mode 100644 intl/icu/source/common/unicode/utrace.h create mode 100644 intl/icu/source/common/unicode/utypes.h create mode 100644 intl/icu/source/common/unicode/uvernum.h create mode 100644 intl/icu/source/common/unicode/uversion.h create mode 100644 intl/icu/source/common/unifiedcache.cpp create mode 100644 intl/icu/source/common/unifiedcache.h create mode 100644 intl/icu/source/common/unifilt.cpp create mode 100644 intl/icu/source/common/unifunct.cpp create mode 100644 intl/icu/source/common/uniquecharstr.h create mode 100644 intl/icu/source/common/uniset.cpp create mode 100644 intl/icu/source/common/uniset_closure.cpp create mode 100644 intl/icu/source/common/uniset_props.cpp create mode 100644 intl/icu/source/common/unisetspan.cpp create mode 100644 intl/icu/source/common/unisetspan.h create mode 100644 intl/icu/source/common/unistr.cpp create mode 100644 intl/icu/source/common/unistr_case.cpp create mode 100644 intl/icu/source/common/unistr_case_locale.cpp create mode 100644 intl/icu/source/common/unistr_cnv.cpp create mode 100644 intl/icu/source/common/unistr_props.cpp create mode 100644 intl/icu/source/common/unistr_titlecase_brkiter.cpp create mode 100644 intl/icu/source/common/unistrappender.h create mode 100644 intl/icu/source/common/unorm.cpp create mode 100644 intl/icu/source/common/unormcmp.cpp create mode 100644 intl/icu/source/common/unormimp.h create mode 100644 intl/icu/source/common/uobject.cpp create mode 100644 intl/icu/source/common/uposixdefs.h create mode 100644 intl/icu/source/common/uprops.cpp create mode 100644 intl/icu/source/common/uprops.h create mode 100644 intl/icu/source/common/ures_cnv.cpp create mode 100644 intl/icu/source/common/uresbund.cpp create mode 100644 intl/icu/source/common/uresdata.cpp create mode 100644 intl/icu/source/common/uresdata.h create mode 100644 intl/icu/source/common/uresimp.h create mode 100644 intl/icu/source/common/ureslocs.h create mode 100644 intl/icu/source/common/usc_impl.cpp create mode 100644 intl/icu/source/common/usc_impl.h create mode 100644 intl/icu/source/common/uscript.cpp create mode 100644 intl/icu/source/common/uscript_props.cpp create mode 100644 intl/icu/source/common/uset.cpp create mode 100644 intl/icu/source/common/uset_imp.h create mode 100644 intl/icu/source/common/uset_props.cpp create mode 100644 intl/icu/source/common/usetiter.cpp create mode 100644 intl/icu/source/common/ushape.cpp create mode 100644 intl/icu/source/common/usprep.cpp create mode 100644 intl/icu/source/common/ustack.cpp create mode 100644 intl/icu/source/common/ustr_cnv.cpp create mode 100644 intl/icu/source/common/ustr_cnv.h create mode 100644 intl/icu/source/common/ustr_imp.h create mode 100644 intl/icu/source/common/ustr_titlecase_brkiter.cpp create mode 100644 intl/icu/source/common/ustr_wcs.cpp create mode 100644 intl/icu/source/common/ustrcase.cpp create mode 100644 intl/icu/source/common/ustrcase_locale.cpp create mode 100644 intl/icu/source/common/ustrenum.cpp create mode 100644 intl/icu/source/common/ustrenum.h create mode 100644 intl/icu/source/common/ustrfmt.cpp create mode 100644 intl/icu/source/common/ustrfmt.h create mode 100644 intl/icu/source/common/ustring.cpp create mode 100644 intl/icu/source/common/ustrtrns.cpp create mode 100644 intl/icu/source/common/utext.cpp create mode 100644 intl/icu/source/common/utf_impl.cpp create mode 100644 intl/icu/source/common/util.cpp create mode 100644 intl/icu/source/common/util.h create mode 100644 intl/icu/source/common/util_props.cpp create mode 100644 intl/icu/source/common/utrace.cpp create mode 100644 intl/icu/source/common/utracimp.h create mode 100644 intl/icu/source/common/utrie.cpp create mode 100644 intl/icu/source/common/utrie.h create mode 100644 intl/icu/source/common/utrie2.cpp create mode 100644 intl/icu/source/common/utrie2.h create mode 100644 intl/icu/source/common/utrie2_builder.cpp create mode 100644 intl/icu/source/common/utrie2_impl.h create mode 100644 intl/icu/source/common/utrie_swap.cpp create mode 100644 intl/icu/source/common/uts46.cpp create mode 100644 intl/icu/source/common/utypeinfo.h create mode 100644 intl/icu/source/common/utypes.cpp create mode 100644 intl/icu/source/common/uvector.cpp create mode 100644 intl/icu/source/common/uvector.h create mode 100644 intl/icu/source/common/uvectr32.cpp create mode 100644 intl/icu/source/common/uvectr32.h create mode 100644 intl/icu/source/common/uvectr64.cpp create mode 100644 intl/icu/source/common/uvectr64.h create mode 100644 intl/icu/source/common/wintz.cpp create mode 100644 intl/icu/source/common/wintz.h (limited to 'intl/icu/source/common') diff --git a/intl/icu/source/common/BUILD.bazel b/intl/icu/source/common/BUILD.bazel new file mode 100644 index 0000000000..47d3d24bf5 --- /dev/null +++ b/intl/icu/source/common/BUILD.bazel @@ -0,0 +1,1214 @@ +# © 2021 and later: Unicode, Inc. and others. +# License & terms of use: http://www.unicode.org/copyright.html + +# This file defines Bazel targets for a subset of ICU4C "common" library header and source files. +# The configuration of dependencies among targets is strongly assisted by the +# file in depstest that maintains such information, at +# icu4c/source/test/depstest/dependencies.txt . + +load("@rules_cc//cc:defs.bzl", "cc_binary", "cc_library") + +package( + default_visibility = ["//visibility:public"], +) + +# When compiling code in the `common` dir, the constant +# `U_COMMON_IMPLEMENTATION` needs to be defined. See +# https://unicode-org.github.io/icu/userguide/howtouseicu#c-with-your-own-build-system . + +# If linker errors occur, then this may be a sign that the dependencies were +# not specified correctly. Use dependencies.txt in depstest for assistance. See +# https://stackoverflow.com/q/66111709/2077918 . + +cc_library( + name = "headers", + hdrs = glob([ + "unicode/*.h", # public + "*.h", # internal + ], + # Instead of using these checked-in files, our Bazel build process + # regenerates them and then uses the new versions. + # Same list of .h files as in icu4c/source/data/unidata/clean.sh. + exclude = ["norm2_nfc_data.h", "propname_data.h", "*_props_data.h"], + ), + # We need to add includes in order to preserve existing source files' + # include directives that use traditional paths, not paths relative to + # Bazel workspace: + # https://stackoverflow.com/a/65635893/2077918 + includes = ["."], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "platform", + srcs = [ + "cmemory.cpp", + "uobject.cpp", + "cstring.cpp", + "cwchar.cpp", + "uinvchar.cpp", + "charstr.cpp", + "unistr.cpp", + "appendable.cpp", + "stringpiece.cpp", + "ustrtrns.cpp", + "ustring.cpp", + "ustrfmt.cpp", + "utf_impl.cpp", + "putil.cpp", + "ucln_cmn.cpp", + "udataswp.cpp", + "umath.cpp", + "umutex.cpp", + "sharedobject.cpp", + "utrace.cpp", + ], + deps = [ + ":headers", + # omit other deps b/c they are sys symbols + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], + linkopts = ["-ldl"], +) + +cc_library( + name = "utrie", + srcs = ["utrie.cpp"], + deps = [":platform"], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "utrie2", + srcs = ["utrie2.cpp"], + deps = [":platform"], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "utrie2_builder", + srcs = ["utrie2_builder.cpp"], + deps = [ + ":utrie", + ":utrie2", + ":platform", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "ucptrie", + srcs = ["ucptrie.cpp"], + deps = [":platform"], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "umutablecptrie", + srcs = ["umutablecptrie.cpp"], + deps = [":ucptrie"], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "bytestrie", + srcs = ["bytestrie.cpp"], + deps = [":platform"], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "bytestriebuilder", + srcs = ["bytestriebuilder.cpp"], + deps = [ + ":bytestrie", + ":stringtriebuilder", + ":sort", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "stringtriebuilder", + srcs = ["stringtriebuilder.cpp"], + deps = [ + ":uhash", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "uhash", + hdrs = [ + "uhash.h", + ], + srcs = [ + "uhash.cpp", + ], + deps = [ + ":headers", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "errorcode", + hdrs = [ + ], + srcs = [ + "errorcode.cpp", + ], + includes = ["."], + deps = [ + ":platform", + ":utypes", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "utypes", + srcs = [ + "utypes.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "uniset", + srcs = [ + "uniset.cpp", + "unifilt.cpp", + "unisetspan.cpp", + "bmpset.cpp", + "util.cpp", + "unifunct.cpp", + "usetiter.cpp", + ], + includes = ["."], + deps = [ + ":patternprops", + ":uvector", + ":headers", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "patternprops", + srcs = [ + "patternprops.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "propsvec", + srcs = [ + "propsvec.cpp", + ], + includes = ["."], + deps = [ + ":sort", + ":utrie2_builder", + ":headers", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "propname", + srcs = [ + "propname.cpp", + "propname_data.h", + ], + includes = ["."], + deps = [ + ":bytestrie", + ":headers", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +# Note: The cc_library target names "uvector32" and "uvector64" match the +# dependencies.txt group names, but the filenames are "uvectr32.*"/"uvectr64.*". +cc_library( + name = "uvector32", + srcs = [ + "uvectr32.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":platform", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "uvector64", + srcs = [ + "uvectr64.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":platform", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "sort", + srcs = [ + "uarrsort.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "uvector", + srcs = [ + "uvector.cpp", + ], + includes = ["."], + deps = [ + ":platform", + ":sort", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "breakiterator", + srcs = [ + "brkiter.cpp", + "brkeng.cpp", + "dictbe.cpp", + "dictionarydata.cpp", + "filteredbrk.cpp", + "lstmbe.cpp", + "mlbe.cpp", + "rbbi.cpp", + "rbbi_cache.cpp", + "rbbidata.cpp", + "rbbinode.cpp", + "rbbirb.cpp", + "rbbiscan.cpp", + "rbbisetb.cpp", + "rbbistbl.cpp", + "rbbitblb.cpp", + "ubrk.cpp", + ], + includes = ["."], + deps = [ + ":bytestrie", + ":headers", + ":normlzr", + ":resourcebundle", + ":schriter", + ":service_registration", + ":ucharstrie", + ":ucharstriebuilder", + ":uhash", + ":uniset_core", + ":uniset_props", + ":ustack", + ":utext", + ":utrie2_builder", + ":uvector32", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "bytesinkutil", + srcs = [ + "bytesinkutil.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":bytestream", + ":edits", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "bytestream", + srcs = [ + "bytestream.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":platform", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "canonical_iterator", + srcs = [ + "caniter.cpp", + ], + deps = [ + ":normalizer2", + ":usetiter", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "characterproperties", + srcs = [ + "characterproperties.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":emojiprops", + ":ucptrie", + ":umutablecptrie", + ":uniset_core", + ":uprops", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "chariter", + srcs = [ + "chariter.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":platform", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "edits", + srcs = [ + "edits.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":icu_utility", + ":platform", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "filterednormalizer2", + srcs = [ + "filterednormalizer2.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":normalizer2", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "hashtable", + srcs = [ + "uhash_us.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":uhash", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "icu_utility", + srcs = [ + "util.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":patternprops", + ":platform", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "loadednormalizer2", + srcs = [ + "loadednormalizer2impl.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":normalizer2", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "locale_display_names", + srcs = [ + "locdispnames.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":locresdata", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "locresdata", + srcs = [ + "locresdata.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":resourcebundle", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "normlzr", + srcs = [ + "normlzr.cpp", + ], + includes = ["."], + deps = [ + ":filterednormalizer2", + ":headers", + ":schriter", + ":uniset_props", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "parsepos", + srcs = [ + "parsepos.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":platform", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "resourcebundle", + srcs = [ + "localebuilder.cpp", + "locavailable.cpp", + "locbased.cpp", + "locid.cpp", + "loclikely.cpp", + "locmap.cpp", + "resbund.cpp", + "resource.cpp", + "uloc.cpp", + "uloc_tag.cpp", + "uloc_keytype.cpp", + "uresbund.cpp", + "uresdata.cpp", + "wintz.cpp", + ], + includes = ["."], + deps = [ + ":bytesinkutil", + ":errorcode", + ":headers", + ":propname", + ":sort", + ":stringenumeration", + ":ucol_swp", + ":udata", + ":uhash", + ":uscript_props", + ":uvector", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "schriter", + srcs = [ + "schriter.cpp", + "uchriter.cpp", + ], + includes = ["."], + deps = [ + ":chariter", + ":headers", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "service_registration", + srcs = [ + "locutil.cpp", + "serv.cpp", + "servlk.cpp", + "servlkf.cpp", + "servls.cpp", + "servnotf.cpp", + "servrbf.cpp", + "servslkf.cpp", + ], + includes = ["."], + deps = [ + ":hashtable", + ":headers", + ":locale_display_names", + ":resourcebundle", + ":uvector", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "stringenumeration", + srcs = [ + "uenum.cpp", + "ustrenum.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":platform", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "ubidi_props", + srcs = [ + "ubidi_props.cpp", + "ubidi_props_data.h", + ], + includes = ["."], + deps = [ + ":headers", + ":utrie2", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "ucase", + srcs = [ + "ucase.cpp", + "ucase_props_data.h", + ], + includes = ["."], + deps = [ + ":headers", + ":utrie2", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "uchar", + srcs = [ + "uchar.cpp", + "uchar_props_data.h", + ], + includes = ["."], + deps = [ + ":headers", + ":utrie2", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "emojiprops", + srcs = [ + "emojiprops.cpp", + "emojiprops.h", + ], + includes = ["."], + deps = [ + ":headers", + ":ucharstrie", + ":ucharstrieiterator", + ":ucptrie", + ":udata", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "ucharstrie", + srcs = [ + "ucharstrie.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":platform", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "ucharstriebuilder", + srcs = [ + "ucharstriebuilder.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":sort", + ":stringtriebuilder", + ":ucharstrie", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "ucharstrieiterator", + srcs = [ + "ucharstrieiterator.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":ucharstrie", + ":uvector32", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "ucol_swp", + srcs = [ + "ucol_swp.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":utrie_swap", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "udata", + srcs = [ + "restrace.cpp", + "ucmndata.cpp", + "udata.cpp", + "udatamem.cpp", + "umapfile.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":icu_utility", + ":platform", + ":uhash", + "//icu4c/source/stubdata", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "uiter", + srcs = [ + "uiter.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":platform", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "ulist", + srcs = [ + "ulist.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":platform", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "unames", + srcs = [ + "unames.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":uchar", + ":udata", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "unifiedcache", + srcs = [ + "unifiedcache.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":platform", + ":uhash", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "uniset_core", + srcs = [ + "bmpset.cpp", + "unifilt.cpp", + "unifunct.cpp", + "uniset.cpp", + "unisetspan.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":icu_utility", + ":patternprops", + ":uvector", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "uniset_closure", + srcs = [ + "uniset_closure.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":uniset_core", + ":unistr_case_locale", + ":unistr_titlecase_brkiter", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "uniset_props", + srcs = [ + "uniset_props.cpp", + "ruleiter.cpp", + ], + includes = ["."], + deps = [ + ":characterproperties", + ":headers", + ":parsepos", + ":propname", + ":resourcebundle", + ":unames", + ":uniset_core", + ":unistr_case", + ":uprops", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "unistr_case", + srcs = [ + "unistr_case.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":ustring_case", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "unistr_case_locale", + srcs = [ + "unistr_case_locale.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":unistr_case", + ":ustring_case_locale", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "unistr_titlecase_brkiter", + srcs = [ + "unistr_titlecase_brkiter.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":ustr_titlecase_brkiter", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "uprops", + srcs = [ + "uprops.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":emojiprops", + ":loadednormalizer2", + ":normalizer2", + ":ubidi_props", + ":ucase", + ":uchar", + ":unistr_case", + ":ustring_case", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "uscript_props", + srcs = [ + "uscript_props.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":platform", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "uset", + srcs = [ + "uset.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":platform", + ":uniset_core", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "uset_props", + srcs = [ + "uset_props.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":uniset_closure", + ":uniset_core", + ":uniset_props", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "usetiter", + srcs = [ + "usetiter.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":platform", + ":uniset_core", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "ustack", + srcs = [ + "ustack.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":uvector", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "ustr_titlecase_brkiter", + srcs = [ + "ustr_titlecase_brkiter.cpp", + ], + includes = ["."], + deps = [ + ":breakiterator", + ":headers", + ":ucase", + ":ustring_case_locale", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "ustring_case", + srcs = [ + "ustrcase.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":ucase", + ":uchar", + ":edits", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "ustring_case_locale", + srcs = [ + "ustrcase_locale.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":resourcebundle", + ":ustring_case", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "utext", + srcs = [ + "utext.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":ucase", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +cc_library( + name = "utrie_swap", + srcs = [ + "utrie_swap.cpp", + ], + includes = ["."], + deps = [ + ":headers", + ":udata", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) + +# This target depends on a header file that contains NFC/NFD normalization data. +# This header file is generated by a script (generate.sh) that invokes the gennorm2 binary. +# See the Unicode update change log (changes.txt). +cc_library( + name = "normalizer2", + srcs = [ + "norm2_nfc_data.h", # generated by gennorm2 + "normalizer2.cpp", + "normalizer2impl.cpp", + ], + includes = ["."], + hdrs = [ + "normalizer2impl.h", + ], + deps = [ + ":headers", + ], + local_defines = [ + "U_COMMON_IMPLEMENTATION", + ], +) diff --git a/intl/icu/source/common/Makefile.in b/intl/icu/source/common/Makefile.in new file mode 100644 index 0000000000..67bcf00349 --- /dev/null +++ b/intl/icu/source/common/Makefile.in @@ -0,0 +1,207 @@ +# Copyright (C) 2016 and later: Unicode, Inc. and others. +# License & terms of use: http://www.unicode.org/copyright.html +#****************************************************************************** +# +# Copyright (C) 1999-2016, International Business Machines +# Corporation and others. All Rights Reserved. +# +#****************************************************************************** +## Makefile.in for ICU - icuuc.so +## Stephen F. Booth + +## Source directory information +srcdir = @srcdir@ +top_srcdir = @top_srcdir@ + +top_builddir = .. + +## All the flags and other definitions are included here. +include $(top_builddir)/icudefs.mk + +## Build directory information +subdir = common + +# for service hook +LOCALSVC_CPP=localsvc.cpp +SVC_HOOK_INC=$(top_builddir)/common/svchook.mk + +## Extra files to remove for 'make clean' +CLEANFILES = *~ $(DEPS) $(IMPORT_LIB) $(MIDDLE_IMPORT_LIB) $(FINAL_IMPORT_LIB) $(SVC_HOOK_INC) + +## Target information + +TARGET_STUBNAME=$(COMMON_STUBNAME) + +ifneq ($(ENABLE_STATIC),) +TARGET = $(LIBDIR)/$(LIBSICU)$(TARGET_STUBNAME)$(ICULIBSUFFIX).$(A) +endif + +ifneq ($(ENABLE_SHARED),) +SO_TARGET = $(LIBDIR)/$(LIBICU)$(TARGET_STUBNAME)$(ICULIBSUFFIX).$(SO) +ALL_SO_TARGETS = $(SO_TARGET) $(MIDDLE_SO_TARGET) $(FINAL_SO_TARGET) $(SHARED_OBJECT) + +ifeq ($(ENABLE_SO_VERSION_DATA),1) +SO_VERSION_DATA = common.res +endif + +ifeq ($(OS390BATCH),1) +BATCH_TARGET = $(BATCH_COMMON_TARGET) +BATCH_LIBS = $(BATCH_LIBICUDT) -lm +endif # OS390BATCH + +endif # ENABLE_SHARED + +ALL_TARGETS = $(TARGET) $(ALL_SO_TARGETS) $(BATCH_TARGET) + +DYNAMICCPPFLAGS = $(SHAREDLIBCPPFLAGS) +DYNAMICCFLAGS = $(SHAREDLIBCFLAGS) +DYNAMICCXXFLAGS = $(SHAREDLIBCXXFLAGS) +CFLAGS += $(LIBCFLAGS) +CXXFLAGS += $(LIBCXXFLAGS) +ifeq ($(OS390BATCH),1) +CFLAGS += -WI +CXXFLAGS += -WI +endif + +CPPFLAGS += -I$(srcdir) $(LIBCPPFLAGS) $(CPPFLAGSICUUC) +# we want DEFS here +DEFS += -DU_COMMON_IMPLEMENTATION +LDFLAGS += $(LDFLAGSICUUC) + +# for plugin configuration +CPPFLAGS += "-DDEFAULT_ICU_PLUGINS=\"$(libdir)/icu\" " + +# for icu data location +ifeq ($(PKGDATA_MODE),common) +CPPFLAGS += "-DU_ICU_DATA_DEFAULT_DIR=\"$(ICUDATA_DIR)\"" +endif + +# $(LIBICUDT) is either stub data or the real DLL common data. +LIBS = $(LIBICUDT) $(DEFAULT_LIBS) + +SOURCES = $(shell cat $(srcdir)/sources.txt) +OBJECTS = $(SOURCES:.cpp=.o) + +## Header files to install +HEADERS = $(srcdir)/unicode/*.h + +STATIC_OBJECTS = $(OBJECTS:.o=.$(STATIC_O)) + +DEPS = $(OBJECTS:.o=.d) + +-include Makefile.local + +-include $(SVC_HOOK_INC) + + +## List of phony targets +.PHONY : all all-local install install-local clean clean-local \ +distclean distclean-local install-library install-headers dist \ +dist-local check check-local check-exhaustive + +## Clear suffix list +.SUFFIXES : + +## List of standard targets +all: all-local +install: install-local +clean: clean-local +distclean : distclean-local +dist: dist-local +check: all check-local + +check-exhaustive: check + +all-local: $(ALL_TARGETS) + +install-local: install-headers install-library + +install-library: all-local + $(MKINSTALLDIRS) $(DESTDIR)$(libdir) +ifneq ($(ENABLE_STATIC),) + $(INSTALL-L) $(TARGET) $(DESTDIR)$(libdir) +endif +ifneq ($(ENABLE_SHARED),) +# For MinGW, do we want the DLL to go in the bin location? +ifeq ($(MINGW_MOVEDLLSTOBINDIR),YES) + $(MKINSTALLDIRS) $(DESTDIR)$(bindir) + $(INSTALL-L) $(FINAL_SO_TARGET) $(DESTDIR)$(bindir) +else + $(INSTALL-L) $(FINAL_SO_TARGET) $(DESTDIR)$(libdir) +ifneq ($(FINAL_SO_TARGET),$(SO_TARGET)) + cd $(DESTDIR)$(libdir) && $(RM) $(notdir $(SO_TARGET)) && ln -s $(notdir $(FINAL_SO_TARGET)) $(notdir $(SO_TARGET)) +ifneq ($(FINAL_SO_TARGET),$(MIDDLE_SO_TARGET)) + cd $(DESTDIR)$(libdir) && $(RM) $(notdir $(MIDDLE_SO_TARGET)) && ln -s $(notdir $(FINAL_SO_TARGET)) $(notdir $(MIDDLE_SO_TARGET)) +endif +endif +endif +ifneq ($(IMPORT_LIB_EXT),) + $(INSTALL-L) $(FINAL_IMPORT_LIB) $(DESTDIR)$(libdir) +ifneq ($(IMPORT_LIB),$(FINAL_IMPORT_LIB)) + cd $(DESTDIR)$(libdir) && $(RM) $(notdir $(IMPORT_LIB)) && ln -s $(notdir $(FINAL_IMPORT_LIB)) $(notdir $(IMPORT_LIB)) +endif +ifneq ($(MIDDLE_IMPORT_LIB),$(FINAL_IMPORT_LIB)) + cd $(DESTDIR)$(libdir) && $(RM) $(notdir $(MIDDLE_IMPORT_LIB)) && ln -s $(notdir $(FINAL_IMPORT_LIB)) $(notdir $(MIDDLE_IMPORT_LIB)) +endif +endif +endif + +$(SVC_HOOK_INC): + @echo generating $@ + @-test -f $(top_srcdir)/common/$(LOCALSVC_CPP) && ( echo "have $(LOCALSVC_CPP) - U_LOCAL_SERVICE_HOOK=1" ; \ + echo 'CPPFLAGS +=-DU_LOCAL_SERVICE_HOOK=1' > $@ ; \ + echo 'OBJECTS += $(LOCALSVC_CPP:%.cpp=%.o)' >> $@ \ + ) ; true + @echo "# Autogenerated by Makefile" >> $@ + +install-headers: + $(MKINSTALLDIRS) $(DESTDIR)$(includedir)/unicode + @for file in $(HEADERS); do \ + echo "$(INSTALL_DATA) $$file $(DESTDIR)$(includedir)/unicode"; \ + $(INSTALL_DATA) $$file $(DESTDIR)$(includedir)/unicode || exit; \ + done + +dist-local: + +clean-local: + test -z "$(CLEANFILES)" || $(RMV) $(CLEANFILES) + $(RMV) $(OBJECTS) $(STATIC_OBJECTS) $(ALL_TARGETS) $(SO_VERSION_DATA) + +distclean-local: clean-local + $(RMV) Makefile icucfg.h $(SVC_HOOK_INC) + +check-local: + +Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status $(SVC_HOOK_INC) + cd $(top_builddir) \ + && CONFIG_FILES=$(subdir)/$@ CONFIG_HEADERS= $(SHELL) ./config.status + +ifneq ($(ENABLE_STATIC),) +$(TARGET): $(STATIC_OBJECTS) + $(AR) $(ARFLAGS) $(AR_OUTOPT)$@ $^ + $(RANLIB) $@ +endif + +ifneq ($(ENABLE_SHARED),) +$(SHARED_OBJECT): $(OBJECTS) $(SO_VERSION_DATA) + $(SHLIB.cc) $(LD_SONAME) $(OUTOPT)$@ $^ $(LIBS) +ifeq ($(ENABLE_RPATH),YES) +ifneq ($(wildcard $(libdir)/$(MIDDLE_SO_TARGET)),) + $(warning RPATH warning: --enable-rpath means test programs may use existing $(libdir)/$(MIDDLE_SO_TARGET)) +endif +endif + +ifeq ($(OS390BATCH),1) +$(BATCH_TARGET):$(OBJECTS) + $(SHLIB.cc) $(LD_SONAME) $(OUTOPT)$@ $^ $(BATCH_LIBS) +endif # OS390BATCH +endif # ENABLE_SHARED + +ifeq (,$(MAKECMDGOALS)) +-include $(DEPS) +else +ifneq ($(patsubst %clean,,$(MAKECMDGOALS)),) +-include $(DEPS) +endif +endif + diff --git a/intl/icu/source/common/appendable.cpp b/intl/icu/source/common/appendable.cpp new file mode 100644 index 0000000000..c0fbcc6530 --- /dev/null +++ b/intl/icu/source/common/appendable.cpp @@ -0,0 +1,74 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2011-2012, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: appendable.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010dec07 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" +#include "unicode/appendable.h" +#include "unicode/utf16.h" + +U_NAMESPACE_BEGIN + +Appendable::~Appendable() {} + +UBool +Appendable::appendCodePoint(UChar32 c) { + if(c<=0xffff) { + return appendCodeUnit((char16_t)c); + } else { + return appendCodeUnit(U16_LEAD(c)) && appendCodeUnit(U16_TRAIL(c)); + } +} + +UBool +Appendable::appendString(const char16_t *s, int32_t length) { + if(length<0) { + char16_t c; + while((c=*s++)!=0) { + if(!appendCodeUnit(c)) { + return false; + } + } + } else if(length>0) { + const char16_t *limit=s+length; + do { + if(!appendCodeUnit(*s++)) { + return false; + } + } while(s>6; // Named for UTF-8 2-byte lead byte with upper 5 bits. + int32_t trail=start&0x3f; // Named for UTF-8 2-byte trail byte with lower 6 bits. + + // Set one bit indicating an all-one block. + uint32_t bits=(uint32_t)1<>6; + int32_t limitTrail=limit&0x3f; + + if(lead==limitLead) { + // Partial vertical bit column. + while(trail0) { + do { + table[trail++]|=bits; + } while(trail<64); + ++lead; + } + if(lead=0x100) { + break; + } + do { + latin1Contains[start++]=1; + } while(start0x80) { + if(start<0x80) { + start=0x80; + } + break; + } + } + + // Set table7FF[]. + while(start<0x800) { + set32x64Bits(table7FF, start, limit<=0x800 ? limit : 0x800); + if(limit>0x800) { + start=0x800; + break; + } + + start=list[listIndex++]; + if(listIndex0x10000) { + limit=0x10000; + } + + if(start>=6; + bmpBlockBits[start&0x3f]|=0x10001<<(start>>6); + start=(start+1)<<6; // Round up to the next block boundary. + minStart=start; // Ignore further ranges in this block. + } + if(start>6, limit>>6); + } + + if(limit&0x3f) { + // Mixed-value block of 64 code points. + limit>>=6; + bmpBlockBits[limit&0x3f]|=0x10001<<(limit>>6); + limit=(limit+1)<<6; // Round up to the next block boundary. + minStart=limit; // Ignore further ranges in this block. + } + } + } + + if(limit==0x10000) { + break; + } + + start=list[listIndex++]; + if(listIndex(~(0x10001<<0xd)); // Lead byte 0xED. + bits=1<<0xd; + for(i=32; i<64; ++i) { // Second half of 4k block. + bmpBlockBits[i]=(bmpBlockBits[i]&mask)|bits; + } + } else { + mask= static_cast(~(0x10001<<0xd)); // Lead byte 0xED. + for(i=32; i<64; ++i) { // Second half of 4k block. + bmpBlockBits[i]&=mask; + } + } +} + +int32_t BMPSet::findCodePoint(UChar32 c, int32_t lo, int32_t hi) const { + /* Examples: + findCodePoint(c) + set list[] c=0 1 3 4 7 8 + === ============== =========== + [] [110000] 0 0 0 0 0 0 + [\u0000-\u0003] [0, 4, 110000] 1 1 1 2 2 2 + [\u0004-\u0007] [4, 8, 110000] 0 0 0 1 1 2 + [:Any:] [0, 110000] 1 1 1 1 1 1 + */ + + // Return the smallest i such that c < list[i]. Assume + // list[len - 1] == HIGH and that c is legal (0..HIGH-1). + if (c < list[lo]) + return lo; + // High runner test. c is often after the last range, so an + // initial check for this condition pays off. + if (lo >= hi || c >= list[hi-1]) + return hi; + // invariant: c >= list[lo] + // invariant: c < list[hi] + for (;;) { + int32_t i = (lo + hi) >> 1; + if (i == lo) { + break; // Found! + } else if (c < list[i]) { + hi = i; + } else { + lo = i; + } + } + return hi; +} + +UBool +BMPSet::contains(UChar32 c) const { + if((uint32_t)c<=0xff) { + return (UBool)latin1Contains[c]; + } else if((uint32_t)c<=0x7ff) { + return (UBool)((table7FF[c&0x3f]&((uint32_t)1<<(c>>6)))!=0); + } else if((uint32_t)c<0xd800 || (c>=0xe000 && c<=0xffff)) { + int lead=c>>12; + uint32_t twoBits=(bmpBlockBits[(c>>6)&0x3f]>>lead)&0x10001; + if(twoBits<=1) { + // All 64 code points with the same bits 15..6 + // are either in the set or not. + return (UBool)twoBits; + } else { + // Look up the code point in its 4k block of code points. + return containsSlow(c, list4kStarts[lead], list4kStarts[lead+1]); + } + } else if((uint32_t)c<=0x10ffff) { + // surrogate or supplementary code point + return containsSlow(c, list4kStarts[0xd], list4kStarts[0x11]); + } else { + // Out-of-range code points get false, consistent with long-standing + // behavior of UnicodeSet::contains(c). + return false; + } +} + +/* + * Check for sufficient length for trail unit for each surrogate pair. + * Handle single surrogates as surrogate code points as usual in ICU. + */ +const char16_t * +BMPSet::span(const char16_t *s, const char16_t *limit, USetSpanCondition spanCondition) const { + char16_t c, c2; + + if(spanCondition) { + // span + do { + c=*s; + if(c<=0xff) { + if(!latin1Contains[c]) { + break; + } + } else if(c<=0x7ff) { + if((table7FF[c&0x3f]&((uint32_t)1<<(c>>6)))==0) { + break; + } + } else if(c<0xd800 || c>=0xe000) { + int lead=c>>12; + uint32_t twoBits=(bmpBlockBits[(c>>6)&0x3f]>>lead)&0x10001; + if(twoBits<=1) { + // All 64 code points with the same bits 15..6 + // are either in the set or not. + if(twoBits==0) { + break; + } + } else { + // Look up the code point in its 4k block of code points. + if(!containsSlow(c, list4kStarts[lead], list4kStarts[lead+1])) { + break; + } + } + } else if(c>=0xdc00 || (s+1)==limit || (c2=s[1])<0xdc00 || c2>=0xe000) { + // surrogate code point + if(!containsSlow(c, list4kStarts[0xd], list4kStarts[0xe])) { + break; + } + } else { + // surrogate pair + if(!containsSlow(U16_GET_SUPPLEMENTARY(c, c2), list4kStarts[0x10], list4kStarts[0x11])) { + break; + } + ++s; + } + } while(++s>6)))!=0) { + break; + } + } else if(c<0xd800 || c>=0xe000) { + int lead=c>>12; + uint32_t twoBits=(bmpBlockBits[(c>>6)&0x3f]>>lead)&0x10001; + if(twoBits<=1) { + // All 64 code points with the same bits 15..6 + // are either in the set or not. + if(twoBits!=0) { + break; + } + } else { + // Look up the code point in its 4k block of code points. + if(containsSlow(c, list4kStarts[lead], list4kStarts[lead+1])) { + break; + } + } + } else if(c>=0xdc00 || (s+1)==limit || (c2=s[1])<0xdc00 || c2>=0xe000) { + // surrogate code point + if(containsSlow(c, list4kStarts[0xd], list4kStarts[0xe])) { + break; + } + } else { + // surrogate pair + if(containsSlow(U16_GET_SUPPLEMENTARY(c, c2), list4kStarts[0x10], list4kStarts[0x11])) { + break; + } + ++s; + } + } while(++s>6)))==0) { + break; + } + } else if(c<0xd800 || c>=0xe000) { + int lead=c>>12; + uint32_t twoBits=(bmpBlockBits[(c>>6)&0x3f]>>lead)&0x10001; + if(twoBits<=1) { + // All 64 code points with the same bits 15..6 + // are either in the set or not. + if(twoBits==0) { + break; + } + } else { + // Look up the code point in its 4k block of code points. + if(!containsSlow(c, list4kStarts[lead], list4kStarts[lead+1])) { + break; + } + } + } else if(c<0xdc00 || s==limit || (c2=*(limit-1))<0xd800 || c2>=0xdc00) { + // surrogate code point + if(!containsSlow(c, list4kStarts[0xd], list4kStarts[0xe])) { + break; + } + } else { + // surrogate pair + if(!containsSlow(U16_GET_SUPPLEMENTARY(c2, c), list4kStarts[0x10], list4kStarts[0x11])) { + break; + } + --limit; + } + if(s==limit) { + return s; + } + } + } else { + // span not + for(;;) { + c=*(--limit); + if(c<=0xff) { + if(latin1Contains[c]) { + break; + } + } else if(c<=0x7ff) { + if((table7FF[c&0x3f]&((uint32_t)1<<(c>>6)))!=0) { + break; + } + } else if(c<0xd800 || c>=0xe000) { + int lead=c>>12; + uint32_t twoBits=(bmpBlockBits[(c>>6)&0x3f]>>lead)&0x10001; + if(twoBits<=1) { + // All 64 code points with the same bits 15..6 + // are either in the set or not. + if(twoBits!=0) { + break; + } + } else { + // Look up the code point in its 4k block of code points. + if(containsSlow(c, list4kStarts[lead], list4kStarts[lead+1])) { + break; + } + } + } else if(c<0xdc00 || s==limit || (c2=*(limit-1))<0xd800 || c2>=0xdc00) { + // surrogate code point + if(containsSlow(c, list4kStarts[0xd], list4kStarts[0xe])) { + break; + } + } else { + // surrogate pair + if(containsSlow(U16_GET_SUPPLEMENTARY(c2, c), list4kStarts[0x10], list4kStarts[0x11])) { + break; + } + --limit; + } + if(s==limit) { + return s; + } + } + } + return limit+1; +} + +/* + * Precheck for sufficient trail bytes at end of string only once per span. + * Check validity. + */ +const uint8_t * +BMPSet::spanUTF8(const uint8_t *s, int32_t length, USetSpanCondition spanCondition) const { + const uint8_t *limit=s+length; + uint8_t b=*s; + if(U8_IS_SINGLE(b)) { + // Initial all-ASCII span. + if(spanCondition) { + do { + if(!latin1Contains[b] || ++s==limit) { + return s; + } + b=*s; + } while(U8_IS_SINGLE(b)); + } else { + do { + if(latin1Contains[b] || ++s==limit) { + return s; + } + b=*s; + } while(U8_IS_SINGLE(b)); + } + length=(int32_t)(limit-s); + } + + if(spanCondition!=USET_SPAN_NOT_CONTAINED) { + spanCondition=USET_SPAN_CONTAINED; // Pin to 0/1 values. + } + + const uint8_t *limit0=limit; + + /* + * Make sure that the last 1/2/3/4-byte sequence before limit is complete + * or runs into a lead byte. + * In the span loop compare s with limit only once + * per multi-byte character. + * + * Give a trailing illegal sequence the same value as the result of contains(FFFD), + * including it if that is part of the span, otherwise set limit0 to before + * the truncated sequence. + */ + b=*(limit-1); + if((int8_t)b<0) { + // b>=0x80: lead or trail byte + if(b<0xc0) { + // single trail byte, check for preceding 3- or 4-byte lead byte + if(length>=2 && (b=*(limit-2))>=0xe0) { + limit-=2; + if(containsFFFD!=spanCondition) { + limit0=limit; + } + } else if(b<0xc0 && b>=0x80 && length>=3 && (b=*(limit-3))>=0xf0) { + // 4-byte lead byte with only two trail bytes + limit-=3; + if(containsFFFD!=spanCondition) { + limit0=limit; + } + } + } else { + // lead byte with no trail bytes + --limit; + if(containsFFFD!=spanCondition) { + limit0=limit; + } + } + } + + uint8_t t1, t2, t3; + + while(s=0xe0) { + if(b<0xf0) { + if( /* handle U+0000..U+FFFF inline */ + (t1=(uint8_t)(s[0]-0x80)) <= 0x3f && + (t2=(uint8_t)(s[1]-0x80)) <= 0x3f + ) { + b&=0xf; + uint32_t twoBits=(bmpBlockBits[t1]>>b)&0x10001; + if(twoBits<=1) { + // All 64 code points with this lead byte and middle trail byte + // are either in the set or not. + if(twoBits!=(uint32_t)spanCondition) { + return s-1; + } + } else { + // Look up the code point in its 4k block of code points. + UChar32 c=(b<<12)|(t1<<6)|t2; + if(containsSlow(c, list4kStarts[b], list4kStarts[b+1]) != spanCondition) { + return s-1; + } + } + s+=2; + continue; + } + } else if( /* handle U+10000..U+10FFFF inline */ + (t1=(uint8_t)(s[0]-0x80)) <= 0x3f && + (t2=(uint8_t)(s[1]-0x80)) <= 0x3f && + (t3=(uint8_t)(s[2]-0x80)) <= 0x3f + ) { + // Give an illegal sequence the same value as the result of contains(FFFD). + UChar32 c=((UChar32)(b-0xf0)<<18)|((UChar32)t1<<12)|(t2<<6)|t3; + if( ( (0x10000<=c && c<=0x10ffff) ? + containsSlow(c, list4kStarts[0x10], list4kStarts[0x11]) : + containsFFFD + ) != spanCondition + ) { + return s-1; + } + s+=3; + continue; + } + } else { + if( /* handle U+0000..U+07FF inline */ + b>=0xc0 && + (t1=(uint8_t)(*s-0x80)) <= 0x3f + ) { + if((USetSpanCondition)((table7FF[t1]&((uint32_t)1<<(b&0x1f)))!=0) != spanCondition) { + return s-1; + } + ++s; + continue; + } + } + + // Give an illegal sequence the same value as the result of contains(FFFD). + // Handle each byte of an illegal sequence separately to simplify the code; + // no need to optimize error handling. + if(containsFFFD!=spanCondition) { + return s-1; + } + } + + return limit0; +} + +/* + * While going backwards through UTF-8 optimize only for ASCII. + * Unlike UTF-16, UTF-8 is not forward-backward symmetrical, that is, it is not + * possible to tell from the last byte in a multi-byte sequence how many + * preceding bytes there should be. Therefore, going backwards through UTF-8 + * is much harder than going forward. + */ +int32_t +BMPSet::spanBackUTF8(const uint8_t *s, int32_t length, USetSpanCondition spanCondition) const { + if(spanCondition!=USET_SPAN_NOT_CONTAINED) { + spanCondition=USET_SPAN_CONTAINED; // Pin to 0/1 values. + } + + uint8_t b; + + do { + b=s[--length]; + if(U8_IS_SINGLE(b)) { + // ASCII sub-span + if(spanCondition) { + do { + if(!latin1Contains[b]) { + return length+1; + } else if(length==0) { + return 0; + } + b=s[--length]; + } while(U8_IS_SINGLE(b)); + } else { + do { + if(latin1Contains[b]) { + return length+1; + } else if(length==0) { + return 0; + } + b=s[--length]; + } while(U8_IS_SINGLE(b)); + } + } + + int32_t prev=length; + UChar32 c; + // trail byte: collect a multi-byte character + // (or lead byte in last-trail position) + c=utf8_prevCharSafeBody(s, 0, &length, b, -3); + // c is a valid code point, not ASCII, not a surrogate + if(c<=0x7ff) { + if((USetSpanCondition)((table7FF[c&0x3f]&((uint32_t)1<<(c>>6)))!=0) != spanCondition) { + return prev+1; + } + } else if(c<=0xffff) { + int lead=c>>12; + uint32_t twoBits=(bmpBlockBits[(c>>6)&0x3f]>>lead)&0x10001; + if(twoBits<=1) { + // All 64 code points with the same bits 15..6 + // are either in the set or not. + if(twoBits!=(uint32_t)spanCondition) { + return prev+1; + } + } else { + // Look up the code point in its 4k block of code points. + if(containsSlow(c, list4kStarts[lead], list4kStarts[lead+1]) != spanCondition) { + return prev+1; + } + } + } else { + if(containsSlow(c, list4kStarts[0x10], list4kStarts[0x11]) != spanCondition) { + return prev+1; + } + } + } while(length>0); + return 0; +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/bmpset.h b/intl/icu/source/common/bmpset.h new file mode 100644 index 0000000000..fad0310038 --- /dev/null +++ b/intl/icu/source/common/bmpset.h @@ -0,0 +1,164 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2007, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: bmpset.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2007jan29 +* created by: Markus W. Scherer +*/ + +#ifndef __BMPSET_H__ +#define __BMPSET_H__ + +#include "unicode/utypes.h" +#include "unicode/uniset.h" + +U_NAMESPACE_BEGIN + +/* + * Helper class for frozen UnicodeSets, implements contains() and span() + * optimized for BMP code points. Structured to be UTF-8-friendly. + * + * Latin-1: Look up bytes. + * 2-byte characters: Bits organized vertically. + * 3-byte characters: Use zero/one/mixed data per 64-block in U+0000..U+FFFF, + * with mixed for illegal ranges. + * Supplementary characters: Binary search over + * the supplementary part of the parent set's inversion list. + */ +class BMPSet : public UMemory { +public: + BMPSet(const int32_t *parentList, int32_t parentListLength); + BMPSet(const BMPSet &otherBMPSet, const int32_t *newParentList, int32_t newParentListLength); + virtual ~BMPSet(); + + virtual UBool contains(UChar32 c) const; + + /* + * Span the initial substring for which each character c has spanCondition==contains(c). + * It must be s0 and spanCondition==0 or 1. + * @return The string pointer which limits the span. + */ + const uint8_t *spanUTF8(const uint8_t *s, int32_t length, USetSpanCondition spanCondition) const; + + /* + * Span the trailing substring for which each character c has spanCondition==contains(c). + * It must be length>0 and spanCondition==0 or 1. + * @return The start of the span. + */ + int32_t spanBackUTF8(const uint8_t *s, int32_t length, USetSpanCondition spanCondition) const; + +private: + void initBits(); + void overrideIllegal(); + + /** + * Same as UnicodeSet::findCodePoint(UChar32 c) const except that the + * binary search is restricted for finding code points in a certain range. + * + * For restricting the search for finding in the range start..end, + * pass in + * lo=findCodePoint(start) and + * hi=findCodePoint(end) + * with 0<=lo<=hicontains(c); +} + +int32_t +UnhandledEngine::findBreaks( UText *text, + int32_t /* startPos */, + int32_t endPos, + UVector32 &/*foundBreaks*/, + UBool /* isPhraseBreaking */, + UErrorCode &status) const { + if (U_FAILURE(status)) return 0; + UChar32 c = utext_current32(text); + while((int32_t)utext_getNativeIndex(text) < endPos && fHandled->contains(c)) { + utext_next32(text); // TODO: recast loop to work with post-increment operations. + c = utext_current32(text); + } + return 0; +} + +void +UnhandledEngine::handleCharacter(UChar32 c) { + if (fHandled == nullptr) { + fHandled = new UnicodeSet(); + if (fHandled == nullptr) { + return; + } + } + if (!fHandled->contains(c)) { + UErrorCode status = U_ZERO_ERROR; + // Apply the entire script of the character. + int32_t script = u_getIntPropertyValue(c, UCHAR_SCRIPT); + fHandled->applyIntPropertyValue(UCHAR_SCRIPT, script, status); + } +} + +/* + ****************************************************************** + */ + +ICULanguageBreakFactory::ICULanguageBreakFactory(UErrorCode &/*status*/) { + fEngines = 0; +} + +ICULanguageBreakFactory::~ICULanguageBreakFactory() { + if (fEngines != 0) { + delete fEngines; + } +} + +U_NAMESPACE_END +U_CDECL_BEGIN +static void U_CALLCONV _deleteEngine(void *obj) { + delete (const icu::LanguageBreakEngine *) obj; +} +U_CDECL_END +U_NAMESPACE_BEGIN + +const LanguageBreakEngine * +ICULanguageBreakFactory::getEngineFor(UChar32 c) { + const LanguageBreakEngine *lbe = nullptr; + UErrorCode status = U_ZERO_ERROR; + + static UMutex gBreakEngineMutex; + Mutex m(&gBreakEngineMutex); + + if (fEngines == nullptr) { + LocalPointer engines(new UStack(_deleteEngine, nullptr, status), status); + if (U_FAILURE(status) ) { + // Note: no way to return error code to caller. + return nullptr; + } + fEngines = engines.orphan(); + } else { + int32_t i = fEngines->size(); + while (--i >= 0) { + lbe = (const LanguageBreakEngine *)(fEngines->elementAt(i)); + if (lbe != nullptr && lbe->handles(c)) { + return lbe; + } + } + } + + // We didn't find an engine. Create one. + lbe = loadEngineFor(c); + if (lbe != nullptr) { + fEngines->push((void *)lbe, status); + } + return U_SUCCESS(status) ? lbe : nullptr; +} + +const LanguageBreakEngine * +ICULanguageBreakFactory::loadEngineFor(UChar32 c) { + UErrorCode status = U_ZERO_ERROR; + UScriptCode code = uscript_getScript(c, &status); + if (U_SUCCESS(status)) { + const LanguageBreakEngine *engine = nullptr; + // Try to use LSTM first + const LSTMData *data = CreateLSTMDataForScript(code, status); + if (U_SUCCESS(status)) { + if (data != nullptr) { + engine = CreateLSTMBreakEngine(code, data, status); + if (U_SUCCESS(status) && engine != nullptr) { + return engine; + } + if (engine != nullptr) { + delete engine; + engine = nullptr; + } else { + DeleteLSTMData(data); + } + } + } + status = U_ZERO_ERROR; // fallback to dictionary based + DictionaryMatcher *m = loadDictionaryMatcherFor(code); + if (m != nullptr) { + switch(code) { + case USCRIPT_THAI: + engine = new ThaiBreakEngine(m, status); + break; + case USCRIPT_LAO: + engine = new LaoBreakEngine(m, status); + break; + case USCRIPT_MYANMAR: + engine = new BurmeseBreakEngine(m, status); + break; + case USCRIPT_KHMER: + engine = new KhmerBreakEngine(m, status); + break; + +#if !UCONFIG_NO_NORMALIZATION + // CJK not available w/o normalization + case USCRIPT_HANGUL: + engine = new CjkBreakEngine(m, kKorean, status); + break; + + // use same BreakEngine and dictionary for both Chinese and Japanese + case USCRIPT_HIRAGANA: + case USCRIPT_KATAKANA: + case USCRIPT_HAN: + engine = new CjkBreakEngine(m, kChineseJapanese, status); + break; +#if 0 + // TODO: Have to get some characters with script=common handled + // by CjkBreakEngine (e.g. U+309B). Simply subjecting + // them to CjkBreakEngine does not work. The engine has to + // special-case them. + case USCRIPT_COMMON: + { + UBlockCode block = ublock_getCode(code); + if (block == UBLOCK_HIRAGANA || block == UBLOCK_KATAKANA) + engine = new CjkBreakEngine(dict, kChineseJapanese, status); + break; + } +#endif +#endif + + default: + break; + } + if (engine == nullptr) { + delete m; + } + else if (U_FAILURE(status)) { + delete engine; + engine = nullptr; + } + return engine; + } + } + return nullptr; +} + +DictionaryMatcher * +ICULanguageBreakFactory::loadDictionaryMatcherFor(UScriptCode script) { + UErrorCode status = U_ZERO_ERROR; + // open root from brkitr tree. + UResourceBundle *b = ures_open(U_ICUDATA_BRKITR, "", &status); + b = ures_getByKeyWithFallback(b, "dictionaries", b, &status); + int32_t dictnlength = 0; + const char16_t *dictfname = + ures_getStringByKeyWithFallback(b, uscript_getShortName(script), &dictnlength, &status); + if (U_FAILURE(status)) { + ures_close(b); + return nullptr; + } + CharString dictnbuf; + CharString ext; + const char16_t *extStart = u_memrchr(dictfname, 0x002e, dictnlength); // last dot + if (extStart != nullptr) { + int32_t len = (int32_t)(extStart - dictfname); + ext.appendInvariantChars(UnicodeString(false, extStart + 1, dictnlength - len - 1), status); + dictnlength = len; + } + dictnbuf.appendInvariantChars(UnicodeString(false, dictfname, dictnlength), status); + ures_close(b); + + UDataMemory *file = udata_open(U_ICUDATA_BRKITR, ext.data(), dictnbuf.data(), &status); + if (U_SUCCESS(status)) { + // build trie + const uint8_t *data = (const uint8_t *)udata_getMemory(file); + const int32_t *indexes = (const int32_t *)data; + const int32_t offset = indexes[DictionaryData::IX_STRING_TRIE_OFFSET]; + const int32_t trieType = indexes[DictionaryData::IX_TRIE_TYPE] & DictionaryData::TRIE_TYPE_MASK; + DictionaryMatcher *m = nullptr; + if (trieType == DictionaryData::TRIE_TYPE_BYTES) { + const int32_t transform = indexes[DictionaryData::IX_TRANSFORM]; + const char *characters = (const char *)(data + offset); + m = new BytesDictionaryMatcher(characters, transform, file); + } + else if (trieType == DictionaryData::TRIE_TYPE_UCHARS) { + const char16_t *characters = (const char16_t *)(data + offset); + m = new UCharsDictionaryMatcher(characters, file); + } + if (m == nullptr) { + // no matcher exists to take ownership - either we are an invalid + // type or memory allocation failed + udata_close(file); + } + return m; + } else if (dictfname != nullptr) { + // we don't have a dictionary matcher. + // returning nullptr here will cause us to fail to find a dictionary break engine, as expected + status = U_ZERO_ERROR; + return nullptr; + } + return nullptr; +} + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ diff --git a/intl/icu/source/common/brkeng.h b/intl/icu/source/common/brkeng.h new file mode 100644 index 0000000000..240dc8f4d3 --- /dev/null +++ b/intl/icu/source/common/brkeng.h @@ -0,0 +1,277 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** + ************************************************************************************ + * Copyright (C) 2006-2012, International Business Machines Corporation and others. * + * All Rights Reserved. * + ************************************************************************************ + */ + +#ifndef BRKENG_H +#define BRKENG_H + +#include "unicode/utypes.h" +#include "unicode/uobject.h" +#include "unicode/utext.h" +#include "unicode/uscript.h" + +U_NAMESPACE_BEGIN + +class UnicodeSet; +class UStack; +class UVector32; +class DictionaryMatcher; + +/******************************************************************* + * LanguageBreakEngine + */ + +/** + *

LanguageBreakEngines implement language-specific knowledge for + * finding text boundaries within a run of characters belonging to a + * specific set. The boundaries will be of a specific kind, e.g. word, + * line, etc.

+ * + *

LanguageBreakEngines should normally be implemented so as to + * be shared between threads without locking.

+ */ +class LanguageBreakEngine : public UMemory { + public: + + /** + *

Default constructor.

+ * + */ + LanguageBreakEngine(); + + /** + *

Virtual destructor.

+ */ + virtual ~LanguageBreakEngine(); + + /** + *

Indicate whether this engine handles a particular character for + * a particular kind of break.

+ * + * @param c A character which begins a run that the engine might handle + * @return true if this engine handles the particular character and break + * type. + */ + virtual UBool handles(UChar32 c) const = 0; + + /** + *

Find any breaks within a run in the supplied text.

+ * + * @param text A UText representing the text. The + * iterator is left at the end of the run of characters which the engine + * is capable of handling. + * @param startPos The start of the run within the supplied text. + * @param endPos The end of the run within the supplied text. + * @param foundBreaks A Vector of int32_t to receive the breaks. + * @param status Information on any errors encountered. + * @return The number of breaks found. + */ + virtual int32_t findBreaks( UText *text, + int32_t startPos, + int32_t endPos, + UVector32 &foundBreaks, + UBool isPhraseBreaking, + UErrorCode &status) const = 0; + +}; + +/******************************************************************* + * LanguageBreakFactory + */ + +/** + *

LanguageBreakFactorys find and return a LanguageBreakEngine + * that can determine breaks for characters in a specific set, if + * such an object can be found.

+ * + *

If a LanguageBreakFactory is to be shared between threads, + * appropriate synchronization must be used; there is none internal + * to the factory.

+ * + *

A LanguageBreakEngine returned by a LanguageBreakFactory can + * normally be shared between threads without synchronization, unless + * the specific subclass of LanguageBreakFactory indicates otherwise.

+ * + *

A LanguageBreakFactory is responsible for deleting any LanguageBreakEngine + * it returns when it itself is deleted, unless the specific subclass of + * LanguageBreakFactory indicates otherwise. Naturally, the factory should + * not be deleted until the LanguageBreakEngines it has returned are no + * longer needed.

+ */ +class LanguageBreakFactory : public UMemory { + public: + + /** + *

Default constructor.

+ * + */ + LanguageBreakFactory(); + + /** + *

Virtual destructor.

+ */ + virtual ~LanguageBreakFactory(); + + /** + *

Find and return a LanguageBreakEngine that can find the desired + * kind of break for the set of characters to which the supplied + * character belongs. It is up to the set of available engines to + * determine what the sets of characters are.

+ * + * @param c A character that begins a run for which a LanguageBreakEngine is + * sought. + * @return A LanguageBreakEngine with the desired characteristics, or 0. + */ + virtual const LanguageBreakEngine *getEngineFor(UChar32 c) = 0; + +}; + +/******************************************************************* + * UnhandledEngine + */ + +/** + *

UnhandledEngine is a special subclass of LanguageBreakEngine that + * handles characters that no other LanguageBreakEngine is available to + * handle. It is told the character and the type of break; at its + * discretion it may handle more than the specified character (e.g., + * the entire script to which that character belongs.

+ * + *

UnhandledEngines may not be shared between threads without + * external synchronization.

+ */ + +class UnhandledEngine : public LanguageBreakEngine { + private: + + /** + * The sets of characters handled. + * @internal + */ + + UnicodeSet *fHandled; + + public: + + /** + *

Default constructor.

+ * + */ + UnhandledEngine(UErrorCode &status); + + /** + *

Virtual destructor.

+ */ + virtual ~UnhandledEngine(); + + /** + *

Indicate whether this engine handles a particular character for + * a particular kind of break.

+ * + * @param c A character which begins a run that the engine might handle + * @return true if this engine handles the particular character and break + * type. + */ + virtual UBool handles(UChar32 c) const override; + + /** + *

Find any breaks within a run in the supplied text.

+ * + * @param text A UText representing the text (TODO: UText). The + * iterator is left at the end of the run of characters which the engine + * is capable of handling. + * @param startPos The start of the run within the supplied text. + * @param endPos The end of the run within the supplied text. + * @param foundBreaks An allocated C array of the breaks found, if any + * @param status Information on any errors encountered. + * @return The number of breaks found. + */ + virtual int32_t findBreaks( UText *text, + int32_t startPos, + int32_t endPos, + UVector32 &foundBreaks, + UBool isPhraseBreaking, + UErrorCode &status) const override; + + /** + *

Tell the engine to handle a particular character and break type.

+ * + * @param c A character which the engine should handle + */ + virtual void handleCharacter(UChar32 c); + +}; + +/******************************************************************* + * ICULanguageBreakFactory + */ + +/** + *

ICULanguageBreakFactory is the default LanguageBreakFactory for + * ICU. It creates dictionary-based LanguageBreakEngines from dictionary + * data in the ICU data file.

+ */ +class ICULanguageBreakFactory : public LanguageBreakFactory { + private: + + /** + * The stack of break engines created by this factory + * @internal + */ + + UStack *fEngines; + + public: + + /** + *

Standard constructor.

+ * + */ + ICULanguageBreakFactory(UErrorCode &status); + + /** + *

Virtual destructor.

+ */ + virtual ~ICULanguageBreakFactory(); + + /** + *

Find and return a LanguageBreakEngine that can find the desired + * kind of break for the set of characters to which the supplied + * character belongs. It is up to the set of available engines to + * determine what the sets of characters are.

+ * + * @param c A character that begins a run for which a LanguageBreakEngine is + * sought. + * @return A LanguageBreakEngine with the desired characteristics, or 0. + */ + virtual const LanguageBreakEngine *getEngineFor(UChar32 c) override; + +protected: + /** + *

Create a LanguageBreakEngine for the set of characters to which + * the supplied character belongs, for the specified break type.

+ * + * @param c A character that begins a run for which a LanguageBreakEngine is + * sought. + * @return A LanguageBreakEngine with the desired characteristics, or 0. + */ + virtual const LanguageBreakEngine *loadEngineFor(UChar32 c); + + /** + *

Create a DictionaryMatcher for the specified script and break type.

+ * @param script An ISO 15924 script code that identifies the dictionary to be + * created. + * @return A DictionaryMatcher with the desired characteristics, or nullptr. + */ + virtual DictionaryMatcher *loadDictionaryMatcherFor(UScriptCode script); +}; + +U_NAMESPACE_END + + /* BRKENG_H */ +#endif diff --git a/intl/icu/source/common/brkiter.cpp b/intl/icu/source/common/brkiter.cpp new file mode 100644 index 0000000000..41e4e0dff5 --- /dev/null +++ b/intl/icu/source/common/brkiter.cpp @@ -0,0 +1,538 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 1997-2015, International Business Machines Corporation and +* others. All Rights Reserved. +******************************************************************************* +* +* File brkiter.cpp +* +* Modification History: +* +* Date Name Description +* 02/18/97 aliu Converted from OpenClass. Added DONE. +* 01/13/2000 helena Added UErrorCode parameter to createXXXInstance methods. +***************************************************************************************** +*/ + +// ***************************************************************************** +// This file was generated from the java source file BreakIterator.java +// ***************************************************************************** + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/rbbi.h" +#include "unicode/brkiter.h" +#include "unicode/udata.h" +#include "unicode/ures.h" +#include "unicode/ustring.h" +#include "unicode/filteredbrk.h" +#include "bytesinkutil.h" +#include "ucln_cmn.h" +#include "cstring.h" +#include "umutex.h" +#include "servloc.h" +#include "locbased.h" +#include "uresimp.h" +#include "uassert.h" +#include "ubrkimpl.h" +#include "utracimp.h" +#include "charstr.h" + +// ***************************************************************************** +// class BreakIterator +// This class implements methods for finding the location of boundaries in text. +// Instances of BreakIterator maintain a current position and scan over text +// returning the index of characters where boundaries occur. +// ***************************************************************************** + +U_NAMESPACE_BEGIN + +// ------------------------------------- + +BreakIterator* +BreakIterator::buildInstance(const Locale& loc, const char *type, UErrorCode &status) +{ + char fnbuff[256]; + char ext[4]={'\0'}; + CharString actualLocale; + int32_t size; + const char16_t* brkfname = nullptr; + UResourceBundle brkRulesStack; + UResourceBundle brkNameStack; + UResourceBundle *brkRules = &brkRulesStack; + UResourceBundle *brkName = &brkNameStack; + RuleBasedBreakIterator *result = nullptr; + + if (U_FAILURE(status)) + return nullptr; + + ures_initStackObject(brkRules); + ures_initStackObject(brkName); + + // Get the locale + UResourceBundle *b = ures_openNoDefault(U_ICUDATA_BRKITR, loc.getName(), &status); + + // Get the "boundaries" array. + if (U_SUCCESS(status)) { + brkRules = ures_getByKeyWithFallback(b, "boundaries", brkRules, &status); + // Get the string object naming the rules file + brkName = ures_getByKeyWithFallback(brkRules, type, brkName, &status); + // Get the actual string + brkfname = ures_getString(brkName, &size, &status); + U_ASSERT((size_t)size=sizeof(fnbuff)) { + size=0; + if (U_SUCCESS(status)) { + status = U_BUFFER_OVERFLOW_ERROR; + } + } + + // Use the string if we found it + if (U_SUCCESS(status) && brkfname) { + actualLocale.append(ures_getLocaleInternal(brkName, &status), -1, status); + + char16_t* extStart=u_strchr(brkfname, 0x002e); + int len = 0; + if (extStart != nullptr){ + len = (int)(extStart-brkfname); + u_UCharsToChars(extStart+1, ext, sizeof(ext)); // nul terminates the buff + u_UCharsToChars(brkfname, fnbuff, len); + } + fnbuff[len]=0; // nul terminate + } + } + + ures_close(brkRules); + ures_close(brkName); + + UDataMemory* file = udata_open(U_ICUDATA_BRKITR, ext, fnbuff, &status); + if (U_FAILURE(status)) { + ures_close(b); + return nullptr; + } + + // Create a RuleBasedBreakIterator + result = new RuleBasedBreakIterator(file, uprv_strstr(type, "phrase") != nullptr, status); + + // If there is a result, set the valid locale and actual locale, and the kind + if (U_SUCCESS(status) && result != nullptr) { + U_LOCALE_BASED(locBased, *(BreakIterator*)result); + locBased.setLocaleIDs(ures_getLocaleByType(b, ULOC_VALID_LOCALE, &status), + actualLocale.data()); + } + + ures_close(b); + + if (U_FAILURE(status) && result != nullptr) { // Sometimes redundant check, but simple + delete result; + return nullptr; + } + + if (result == nullptr) { + udata_close(file); + if (U_SUCCESS(status)) { + status = U_MEMORY_ALLOCATION_ERROR; + } + } + + return result; +} + +// Creates a break iterator for word breaks. +BreakIterator* U_EXPORT2 +BreakIterator::createWordInstance(const Locale& key, UErrorCode& status) +{ + return createInstance(key, UBRK_WORD, status); +} + +// ------------------------------------- + +// Creates a break iterator for line breaks. +BreakIterator* U_EXPORT2 +BreakIterator::createLineInstance(const Locale& key, UErrorCode& status) +{ + return createInstance(key, UBRK_LINE, status); +} + +// ------------------------------------- + +// Creates a break iterator for character breaks. +BreakIterator* U_EXPORT2 +BreakIterator::createCharacterInstance(const Locale& key, UErrorCode& status) +{ + return createInstance(key, UBRK_CHARACTER, status); +} + +// ------------------------------------- + +// Creates a break iterator for sentence breaks. +BreakIterator* U_EXPORT2 +BreakIterator::createSentenceInstance(const Locale& key, UErrorCode& status) +{ + return createInstance(key, UBRK_SENTENCE, status); +} + +// ------------------------------------- + +// Creates a break iterator for title casing breaks. +BreakIterator* U_EXPORT2 +BreakIterator::createTitleInstance(const Locale& key, UErrorCode& status) +{ + return createInstance(key, UBRK_TITLE, status); +} + +// ------------------------------------- + +// Gets all the available locales that has localized text boundary data. +const Locale* U_EXPORT2 +BreakIterator::getAvailableLocales(int32_t& count) +{ + return Locale::getAvailableLocales(count); +} + +// ------------------------------------------ +// +// Constructors, destructor and assignment operator +// +//------------------------------------------- + +BreakIterator::BreakIterator() +{ + *validLocale = *actualLocale = 0; +} + +BreakIterator::BreakIterator(const BreakIterator &other) : UObject(other) { + uprv_strncpy(actualLocale, other.actualLocale, sizeof(actualLocale)); + uprv_strncpy(validLocale, other.validLocale, sizeof(validLocale)); +} + +BreakIterator &BreakIterator::operator =(const BreakIterator &other) { + if (this != &other) { + uprv_strncpy(actualLocale, other.actualLocale, sizeof(actualLocale)); + uprv_strncpy(validLocale, other.validLocale, sizeof(validLocale)); + } + return *this; +} + +BreakIterator::~BreakIterator() +{ +} + +// ------------------------------------------ +// +// Registration +// +//------------------------------------------- +#if !UCONFIG_NO_SERVICE + +// ------------------------------------- + +class ICUBreakIteratorFactory : public ICUResourceBundleFactory { +public: + virtual ~ICUBreakIteratorFactory(); +protected: + virtual UObject* handleCreate(const Locale& loc, int32_t kind, const ICUService* /*service*/, UErrorCode& status) const override { + return BreakIterator::makeInstance(loc, kind, status); + } +}; + +ICUBreakIteratorFactory::~ICUBreakIteratorFactory() {} + +// ------------------------------------- + +class ICUBreakIteratorService : public ICULocaleService { +public: + ICUBreakIteratorService() + : ICULocaleService(UNICODE_STRING("Break Iterator", 14)) + { + UErrorCode status = U_ZERO_ERROR; + registerFactory(new ICUBreakIteratorFactory(), status); + } + + virtual ~ICUBreakIteratorService(); + + virtual UObject* cloneInstance(UObject* instance) const override { + return ((BreakIterator*)instance)->clone(); + } + + virtual UObject* handleDefault(const ICUServiceKey& key, UnicodeString* /*actualID*/, UErrorCode& status) const override { + LocaleKey& lkey = static_cast(const_cast(key)); + int32_t kind = lkey.kind(); + Locale loc; + lkey.currentLocale(loc); + return BreakIterator::makeInstance(loc, kind, status); + } + + virtual UBool isDefault() const override { + return countFactories() == 1; + } +}; + +ICUBreakIteratorService::~ICUBreakIteratorService() {} + +// ------------------------------------- + +// defined in ucln_cmn.h +U_NAMESPACE_END + +static icu::UInitOnce gInitOnceBrkiter {}; +static icu::ICULocaleService* gService = nullptr; + + + +/** + * Release all static memory held by breakiterator. + */ +U_CDECL_BEGIN +static UBool U_CALLCONV breakiterator_cleanup() { +#if !UCONFIG_NO_SERVICE + if (gService) { + delete gService; + gService = nullptr; + } + gInitOnceBrkiter.reset(); +#endif + return true; +} +U_CDECL_END +U_NAMESPACE_BEGIN + +static void U_CALLCONV +initService() { + gService = new ICUBreakIteratorService(); + ucln_common_registerCleanup(UCLN_COMMON_BREAKITERATOR, breakiterator_cleanup); +} + +static ICULocaleService* +getService() +{ + umtx_initOnce(gInitOnceBrkiter, &initService); + return gService; +} + + +// ------------------------------------- + +static inline UBool +hasService() +{ + return !gInitOnceBrkiter.isReset() && getService() != nullptr; +} + +// ------------------------------------- + +URegistryKey U_EXPORT2 +BreakIterator::registerInstance(BreakIterator* toAdopt, const Locale& locale, UBreakIteratorType kind, UErrorCode& status) +{ + ICULocaleService *service = getService(); + if (service == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + return service->registerInstance(toAdopt, locale, kind, status); +} + +// ------------------------------------- + +UBool U_EXPORT2 +BreakIterator::unregister(URegistryKey key, UErrorCode& status) +{ + if (U_SUCCESS(status)) { + if (hasService()) { + return gService->unregister(key, status); + } + status = U_MEMORY_ALLOCATION_ERROR; + } + return false; +} + +// ------------------------------------- + +StringEnumeration* U_EXPORT2 +BreakIterator::getAvailableLocales() +{ + ICULocaleService *service = getService(); + if (service == nullptr) { + return nullptr; + } + return service->getAvailableLocales(); +} +#endif /* UCONFIG_NO_SERVICE */ + +// ------------------------------------- + +BreakIterator* +BreakIterator::createInstance(const Locale& loc, int32_t kind, UErrorCode& status) +{ + if (U_FAILURE(status)) { + return nullptr; + } + +#if !UCONFIG_NO_SERVICE + if (hasService()) { + Locale actualLoc(""); + BreakIterator *result = (BreakIterator*)gService->get(loc, kind, &actualLoc, status); + // TODO: The way the service code works in ICU 2.8 is that if + // there is a real registered break iterator, the actualLoc + // will be populated, but if the handleDefault path is taken + // (because nothing is registered that can handle the + // requested locale) then the actualLoc comes back empty. In + // that case, the returned object already has its actual/valid + // locale data populated (by makeInstance, which is what + // handleDefault calls), so we don't touch it. YES, A COMMENT + // THIS LONG is a sign of bad code -- so the action item is to + // revisit this in ICU 3.0 and clean it up/fix it/remove it. + if (U_SUCCESS(status) && (result != nullptr) && *actualLoc.getName() != 0) { + U_LOCALE_BASED(locBased, *result); + locBased.setLocaleIDs(actualLoc.getName(), actualLoc.getName()); + } + return result; + } + else +#endif + { + return makeInstance(loc, kind, status); + } +} + +// ------------------------------------- +enum { kKeyValueLenMax = 32 }; + +BreakIterator* +BreakIterator::makeInstance(const Locale& loc, int32_t kind, UErrorCode& status) +{ + + if (U_FAILURE(status)) { + return nullptr; + } + + BreakIterator *result = nullptr; + switch (kind) { + case UBRK_CHARACTER: + { + UTRACE_ENTRY(UTRACE_UBRK_CREATE_CHARACTER); + result = BreakIterator::buildInstance(loc, "grapheme", status); + UTRACE_EXIT_STATUS(status); + } + break; + case UBRK_WORD: + { + UTRACE_ENTRY(UTRACE_UBRK_CREATE_WORD); + result = BreakIterator::buildInstance(loc, "word", status); + UTRACE_EXIT_STATUS(status); + } + break; + case UBRK_LINE: + { + char lb_lw[kKeyValueLenMax]; + UTRACE_ENTRY(UTRACE_UBRK_CREATE_LINE); + uprv_strcpy(lb_lw, "line"); + UErrorCode kvStatus = U_ZERO_ERROR; + CharString value; + CharStringByteSink valueSink(&value); + loc.getKeywordValue("lb", valueSink, kvStatus); + if (U_SUCCESS(kvStatus) && (value == "strict" || value == "normal" || value == "loose")) { + uprv_strcat(lb_lw, "_"); + uprv_strcat(lb_lw, value.data()); + } + // lw=phrase is only supported in Japanese and Korean + if (uprv_strcmp(loc.getLanguage(), "ja") == 0 || uprv_strcmp(loc.getLanguage(), "ko") == 0) { + value.clear(); + loc.getKeywordValue("lw", valueSink, kvStatus); + if (U_SUCCESS(kvStatus) && value == "phrase") { + uprv_strcat(lb_lw, "_"); + uprv_strcat(lb_lw, value.data()); + } + } + result = BreakIterator::buildInstance(loc, lb_lw, status); + + UTRACE_DATA1(UTRACE_INFO, "lb_lw=%s", lb_lw); + UTRACE_EXIT_STATUS(status); + } + break; + case UBRK_SENTENCE: + { + UTRACE_ENTRY(UTRACE_UBRK_CREATE_SENTENCE); + result = BreakIterator::buildInstance(loc, "sentence", status); +#if !UCONFIG_NO_FILTERED_BREAK_ITERATION + char ssKeyValue[kKeyValueLenMax] = {0}; + UErrorCode kvStatus = U_ZERO_ERROR; + int32_t kLen = loc.getKeywordValue("ss", ssKeyValue, kKeyValueLenMax, kvStatus); + if (U_SUCCESS(kvStatus) && kLen > 0 && uprv_strcmp(ssKeyValue,"standard")==0) { + FilteredBreakIteratorBuilder* fbiBuilder = FilteredBreakIteratorBuilder::createInstance(loc, kvStatus); + if (U_SUCCESS(kvStatus)) { + result = fbiBuilder->build(result, status); + delete fbiBuilder; + } + } +#endif + UTRACE_EXIT_STATUS(status); + } + break; + case UBRK_TITLE: + { + UTRACE_ENTRY(UTRACE_UBRK_CREATE_TITLE); + result = BreakIterator::buildInstance(loc, "title", status); + UTRACE_EXIT_STATUS(status); + } + break; + default: + status = U_ILLEGAL_ARGUMENT_ERROR; + } + + if (U_FAILURE(status)) { + return nullptr; + } + + return result; +} + +Locale +BreakIterator::getLocale(ULocDataLocaleType type, UErrorCode& status) const { + U_LOCALE_BASED(locBased, *this); + return locBased.getLocale(type, status); +} + +const char * +BreakIterator::getLocaleID(ULocDataLocaleType type, UErrorCode& status) const { + U_LOCALE_BASED(locBased, *this); + return locBased.getLocaleID(type, status); +} + + +// This implementation of getRuleStatus is a do-nothing stub, here to +// provide a default implementation for any derived BreakIterator classes that +// do not implement it themselves. +int32_t BreakIterator::getRuleStatus() const { + return 0; +} + +// This implementation of getRuleStatusVec is a do-nothing stub, here to +// provide a default implementation for any derived BreakIterator classes that +// do not implement it themselves. +int32_t BreakIterator::getRuleStatusVec(int32_t *fillInVec, int32_t capacity, UErrorCode &status) { + if (U_FAILURE(status)) { + return 0; + } + if (capacity < 1) { + status = U_BUFFER_OVERFLOW_ERROR; + return 1; + } + *fillInVec = 0; + return 1; +} + +BreakIterator::BreakIterator (const Locale& valid, const Locale& actual) { + U_LOCALE_BASED(locBased, (*this)); + locBased.setLocaleIDs(valid, actual); +} + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ + +//eof diff --git a/intl/icu/source/common/bytesinkutil.cpp b/intl/icu/source/common/bytesinkutil.cpp new file mode 100644 index 0000000000..a32254a7db --- /dev/null +++ b/intl/icu/source/common/bytesinkutil.cpp @@ -0,0 +1,161 @@ +// © 2017 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// bytesinkutil.cpp +// created: 2017sep14 Markus W. Scherer + +#include "unicode/utypes.h" +#include "unicode/bytestream.h" +#include "unicode/edits.h" +#include "unicode/stringoptions.h" +#include "unicode/utf8.h" +#include "unicode/utf16.h" +#include "bytesinkutil.h" +#include "charstr.h" +#include "cmemory.h" +#include "uassert.h" + +U_NAMESPACE_BEGIN + +UBool +ByteSinkUtil::appendChange(int32_t length, const char16_t *s16, int32_t s16Length, + ByteSink &sink, Edits *edits, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return false; } + char scratch[200]; + int32_t s8Length = 0; + for (int32_t i = 0; i < s16Length;) { + int32_t capacity; + int32_t desiredCapacity = s16Length - i; + if (desiredCapacity < (INT32_MAX / 3)) { + desiredCapacity *= 3; // max 3 UTF-8 bytes per UTF-16 code unit + } else if (desiredCapacity < (INT32_MAX / 2)) { + desiredCapacity *= 2; + } else { + desiredCapacity = INT32_MAX; + } + char *buffer = sink.GetAppendBuffer(U8_MAX_LENGTH, desiredCapacity, + scratch, UPRV_LENGTHOF(scratch), &capacity); + capacity -= U8_MAX_LENGTH - 1; + int32_t j = 0; + for (; i < s16Length && j < capacity;) { + UChar32 c; + U16_NEXT_UNSAFE(s16, i, c); + U8_APPEND_UNSAFE(buffer, j, c); + } + if (j > (INT32_MAX - s8Length)) { + errorCode = U_INDEX_OUTOFBOUNDS_ERROR; + return false; + } + sink.Append(buffer, j); + s8Length += j; + } + if (edits != nullptr) { + edits->addReplace(length, s8Length); + } + return true; +} + +UBool +ByteSinkUtil::appendChange(const uint8_t *s, const uint8_t *limit, + const char16_t *s16, int32_t s16Length, + ByteSink &sink, Edits *edits, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return false; } + if ((limit - s) > INT32_MAX) { + errorCode = U_INDEX_OUTOFBOUNDS_ERROR; + return false; + } + return appendChange((int32_t)(limit - s), s16, s16Length, sink, edits, errorCode); +} + +void +ByteSinkUtil::appendCodePoint(int32_t length, UChar32 c, ByteSink &sink, Edits *edits) { + char s8[U8_MAX_LENGTH]; + int32_t s8Length = 0; + U8_APPEND_UNSAFE(s8, s8Length, c); + if (edits != nullptr) { + edits->addReplace(length, s8Length); + } + sink.Append(s8, s8Length); +} + +namespace { + +// See unicode/utf8.h U8_APPEND_UNSAFE(). +inline uint8_t getTwoByteLead(UChar32 c) { return (uint8_t)((c >> 6) | 0xc0); } +inline uint8_t getTwoByteTrail(UChar32 c) { return (uint8_t)((c & 0x3f) | 0x80); } + +} // namespace + +void +ByteSinkUtil::appendTwoBytes(UChar32 c, ByteSink &sink) { + U_ASSERT(0x80 <= c && c <= 0x7ff); // 2-byte UTF-8 + char s8[2] = { (char)getTwoByteLead(c), (char)getTwoByteTrail(c) }; + sink.Append(s8, 2); +} + +void +ByteSinkUtil::appendNonEmptyUnchanged(const uint8_t *s, int32_t length, + ByteSink &sink, uint32_t options, Edits *edits) { + U_ASSERT(length > 0); + if (edits != nullptr) { + edits->addUnchanged(length); + } + if ((options & U_OMIT_UNCHANGED_TEXT) == 0) { + sink.Append(reinterpret_cast(s), length); + } +} + +UBool +ByteSinkUtil::appendUnchanged(const uint8_t *s, const uint8_t *limit, + ByteSink &sink, uint32_t options, Edits *edits, + UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return false; } + if ((limit - s) > INT32_MAX) { + errorCode = U_INDEX_OUTOFBOUNDS_ERROR; + return false; + } + int32_t length = (int32_t)(limit - s); + if (length > 0) { + appendNonEmptyUnchanged(s, length, sink, options, edits); + } + return true; +} + +CharStringByteSink::CharStringByteSink(CharString* dest) : dest_(*dest) { +} + +CharStringByteSink::~CharStringByteSink() = default; + +void +CharStringByteSink::Append(const char* bytes, int32_t n) { + UErrorCode status = U_ZERO_ERROR; + dest_.append(bytes, n, status); + // Any errors are silently ignored. +} + +char* +CharStringByteSink::GetAppendBuffer(int32_t min_capacity, + int32_t desired_capacity_hint, + char* scratch, + int32_t scratch_capacity, + int32_t* result_capacity) { + if (min_capacity < 1 || scratch_capacity < min_capacity) { + *result_capacity = 0; + return nullptr; + } + + UErrorCode status = U_ZERO_ERROR; + char* result = dest_.getAppendBuffer( + min_capacity, + desired_capacity_hint, + *result_capacity, + status); + if (U_SUCCESS(status)) { + return result; + } + + *result_capacity = scratch_capacity; + return scratch; +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/bytesinkutil.h b/intl/icu/source/common/bytesinkutil.h new file mode 100644 index 0000000000..929c71fbee --- /dev/null +++ b/intl/icu/source/common/bytesinkutil.h @@ -0,0 +1,88 @@ +// © 2017 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// bytesinkutil.h +// created: 2017sep14 Markus W. Scherer + +#ifndef BYTESINKUTIL_H +#define BYTESINKUTIL_H + +#include "unicode/utypes.h" +#include "unicode/bytestream.h" +#include "unicode/edits.h" +#include "cmemory.h" +#include "uassert.h" + +U_NAMESPACE_BEGIN + +class ByteSink; +class CharString; +class Edits; + +class U_COMMON_API ByteSinkUtil { +public: + ByteSinkUtil() = delete; // all static + + /** (length) bytes were mapped to valid (s16, s16Length). */ + static UBool appendChange(int32_t length, + const char16_t *s16, int32_t s16Length, + ByteSink &sink, Edits *edits, UErrorCode &errorCode); + + /** The bytes at [s, limit[ were mapped to valid (s16, s16Length). */ + static UBool appendChange(const uint8_t *s, const uint8_t *limit, + const char16_t *s16, int32_t s16Length, + ByteSink &sink, Edits *edits, UErrorCode &errorCode); + + /** (length) bytes were mapped/changed to valid code point c. */ + static void appendCodePoint(int32_t length, UChar32 c, ByteSink &sink, Edits *edits = nullptr); + + /** The few bytes at [src, nextSrc[ were mapped/changed to valid code point c. */ + static inline void appendCodePoint(const uint8_t *src, const uint8_t *nextSrc, UChar32 c, + ByteSink &sink, Edits *edits = nullptr) { + appendCodePoint((int32_t)(nextSrc - src), c, sink, edits); + } + + /** Append the two-byte character (U+0080..U+07FF). */ + static void appendTwoBytes(UChar32 c, ByteSink &sink); + + static UBool appendUnchanged(const uint8_t *s, int32_t length, + ByteSink &sink, uint32_t options, Edits *edits, + UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return false; } + if (length > 0) { appendNonEmptyUnchanged(s, length, sink, options, edits); } + return true; + } + + static UBool appendUnchanged(const uint8_t *s, const uint8_t *limit, + ByteSink &sink, uint32_t options, Edits *edits, + UErrorCode &errorCode); + +private: + static void appendNonEmptyUnchanged(const uint8_t *s, int32_t length, + ByteSink &sink, uint32_t options, Edits *edits); +}; + +class U_COMMON_API CharStringByteSink : public ByteSink { +public: + CharStringByteSink(CharString* dest); + ~CharStringByteSink() override; + + CharStringByteSink() = delete; + CharStringByteSink(const CharStringByteSink&) = delete; + CharStringByteSink& operator=(const CharStringByteSink&) = delete; + + void Append(const char* bytes, int32_t n) override; + + char* GetAppendBuffer(int32_t min_capacity, + int32_t desired_capacity_hint, + char* scratch, + int32_t scratch_capacity, + int32_t* result_capacity) override; + +private: + CharString& dest_; +}; + +U_NAMESPACE_END + +#endif //BYTESINKUTIL_H diff --git a/intl/icu/source/common/bytestream.cpp b/intl/icu/source/common/bytestream.cpp new file mode 100644 index 0000000000..bd870cd3b2 --- /dev/null +++ b/intl/icu/source/common/bytestream.cpp @@ -0,0 +1,85 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// Copyright (C) 2009-2011, International Business Machines +// Corporation and others. All Rights Reserved. +// +// Copyright 2007 Google Inc. All Rights Reserved. +// Author: sanjay@google.com (Sanjay Ghemawat) + +#include "unicode/utypes.h" +#include "unicode/bytestream.h" +#include "cmemory.h" + +U_NAMESPACE_BEGIN + +ByteSink::~ByteSink() {} + +char* ByteSink::GetAppendBuffer(int32_t min_capacity, + int32_t /*desired_capacity_hint*/, + char* scratch, int32_t scratch_capacity, + int32_t* result_capacity) { + if (min_capacity < 1 || scratch_capacity < min_capacity) { + *result_capacity = 0; + return nullptr; + } + *result_capacity = scratch_capacity; + return scratch; +} + +void ByteSink::Flush() {} + +CheckedArrayByteSink::CheckedArrayByteSink(char* outbuf, int32_t capacity) + : outbuf_(outbuf), capacity_(capacity < 0 ? 0 : capacity), + size_(0), appended_(0), overflowed_(false) { +} + +CheckedArrayByteSink::~CheckedArrayByteSink() {} + +CheckedArrayByteSink& CheckedArrayByteSink::Reset() { + size_ = appended_ = 0; + overflowed_ = false; + return *this; +} + +void CheckedArrayByteSink::Append(const char* bytes, int32_t n) { + if (n <= 0) { + return; + } + if (n > (INT32_MAX - appended_)) { + // TODO: Report as integer overflow, not merely buffer overflow. + appended_ = INT32_MAX; + overflowed_ = true; + return; + } + appended_ += n; + int32_t available = capacity_ - size_; + if (n > available) { + n = available; + overflowed_ = true; + } + if (n > 0 && bytes != (outbuf_ + size_)) { + uprv_memcpy(outbuf_ + size_, bytes, n); + } + size_ += n; +} + +char* CheckedArrayByteSink::GetAppendBuffer(int32_t min_capacity, + int32_t /*desired_capacity_hint*/, + char* scratch, + int32_t scratch_capacity, + int32_t* result_capacity) { + if (min_capacity < 1 || scratch_capacity < min_capacity) { + *result_capacity = 0; + return nullptr; + } + int32_t available = capacity_ - size_; + if (available >= min_capacity) { + *result_capacity = available; + return outbuf_ + size_; + } else { + *result_capacity = scratch_capacity; + return scratch; + } +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/bytestrie.cpp b/intl/icu/source/common/bytestrie.cpp new file mode 100644 index 0000000000..532ea9e9c0 --- /dev/null +++ b/intl/icu/source/common/bytestrie.cpp @@ -0,0 +1,441 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2011, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: bytestrie.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010sep25 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" +#include "unicode/bytestream.h" +#include "unicode/bytestrie.h" +#include "unicode/uobject.h" +#include "cmemory.h" +#include "uassert.h" + +U_NAMESPACE_BEGIN + +BytesTrie::~BytesTrie() { + uprv_free(ownedArray_); +} + +// lead byte already shifted right by 1. +int32_t +BytesTrie::readValue(const uint8_t *pos, int32_t leadByte) { + int32_t value; + if(leadByte=kMinValueLead) ? + valueResult(node) : USTRINGTRIE_NO_VALUE; + } +} + +UStringTrieResult +BytesTrie::branchNext(const uint8_t *pos, int32_t length, int32_t inByte) { + // Branch according to the current byte. + if(length==0) { + length=*pos++; + } + ++length; + // The length of the branch is the number of bytes to select from. + // The data structure encodes a binary search. + while(length>kMaxBranchLinearSubNodeLength) { + if(inByte<*pos++) { + length>>=1; + pos=jumpByDelta(pos); + } else { + length=length-(length>>1); + pos=skipDelta(pos); + } + } + // Drop down to linear search for the last few bytes. + // length>=2 because the loop body above sees length>kMaxBranchLinearSubNodeLength>=3 + // and divides length by 2. + do { + if(inByte==*pos++) { + UStringTrieResult result; + int32_t node=*pos; + U_ASSERT(node>=kMinValueLead); + if(node&kValueIsFinal) { + // Leave the final value for getValue() to read. + result=USTRINGTRIE_FINAL_VALUE; + } else { + // Use the non-final value as the jump delta. + ++pos; + // int32_t delta=readValue(pos, node>>1); + node>>=1; + int32_t delta; + if(node=kMinValueLead ? valueResult(node) : USTRINGTRIE_NO_VALUE; + } + pos_=pos; + return result; + } + --length; + pos=skipValue(pos); + } while(length>1); + if(inByte==*pos++) { + pos_=pos; + int32_t node=*pos; + return node>=kMinValueLead ? valueResult(node) : USTRINGTRIE_NO_VALUE; + } else { + stop(); + return USTRINGTRIE_NO_MATCH; + } +} + +UStringTrieResult +BytesTrie::nextImpl(const uint8_t *pos, int32_t inByte) { + for(;;) { + int32_t node=*pos++; + if(node=kMinValueLead) ? + valueResult(node) : USTRINGTRIE_NO_VALUE; + } else { + // No match. + break; + } + } else if(node&kValueIsFinal) { + // No further matching bytes. + break; + } else { + // Skip intermediate value. + pos=skipValue(pos, node); + // The next node must not also be a value node. + U_ASSERT(*pos=0) { + // Remaining part of a linear-match node. + if(inByte==*pos++) { + remainingMatchLength_=--length; + pos_=pos; + int32_t node; + return (length<0 && (node=*pos)>=kMinValueLead) ? + valueResult(node) : USTRINGTRIE_NO_VALUE; + } else { + stop(); + return USTRINGTRIE_NO_MATCH; + } + } + return nextImpl(pos, inByte); +} + +UStringTrieResult +BytesTrie::next(const char *s, int32_t sLength) { + if(sLength<0 ? *s==0 : sLength==0) { + // Empty input. + return current(); + } + const uint8_t *pos=pos_; + if(pos==nullptr) { + return USTRINGTRIE_NO_MATCH; + } + int32_t length=remainingMatchLength_; // Actual remaining match length minus 1. + for(;;) { + // Fetch the next input byte, if there is one. + // Continue a linear-match node without rechecking sLength<0. + int32_t inByte; + if(sLength<0) { + for(;;) { + if((inByte=*s++)==0) { + remainingMatchLength_=length; + pos_=pos; + int32_t node; + return (length<0 && (node=*pos)>=kMinValueLead) ? + valueResult(node) : USTRINGTRIE_NO_VALUE; + } + if(length<0) { + remainingMatchLength_=length; + break; + } + if(inByte!=*pos) { + stop(); + return USTRINGTRIE_NO_MATCH; + } + ++pos; + --length; + } + } else { + for(;;) { + if(sLength==0) { + remainingMatchLength_=length; + pos_=pos; + int32_t node; + return (length<0 && (node=*pos)>=kMinValueLead) ? + valueResult(node) : USTRINGTRIE_NO_VALUE; + } + inByte=*s++; + --sLength; + if(length<0) { + remainingMatchLength_=length; + break; + } + if(inByte!=*pos) { + stop(); + return USTRINGTRIE_NO_MATCH; + } + ++pos; + --length; + } + } + for(;;) { + int32_t node=*pos++; + if(nodekMaxBranchLinearSubNodeLength) { + ++pos; // ignore the comparison byte + if(nullptr==findUniqueValueFromBranch(jumpByDelta(pos), length>>1, haveUniqueValue, uniqueValue)) { + return nullptr; + } + length=length-(length>>1); + pos=skipDelta(pos); + } + do { + ++pos; // ignore a comparison byte + // handle its value + int32_t node=*pos++; + UBool isFinal=(UBool)(node&kValueIsFinal); + int32_t value=readValue(pos, node>>1); + pos=skipValue(pos, node); + if(isFinal) { + if(haveUniqueValue) { + if(value!=uniqueValue) { + return nullptr; + } + } else { + uniqueValue=value; + haveUniqueValue=true; + } + } else { + if(!findUniqueValue(pos+value, haveUniqueValue, uniqueValue)) { + return nullptr; + } + haveUniqueValue=true; + } + } while(--length>1); + return pos+1; // ignore the last comparison byte +} + +UBool +BytesTrie::findUniqueValue(const uint8_t *pos, UBool haveUniqueValue, int32_t &uniqueValue) { + for(;;) { + int32_t node=*pos++; + if(node>1); + if(haveUniqueValue) { + if(value!=uniqueValue) { + return false; + } + } else { + uniqueValue=value; + haveUniqueValue=true; + } + if(isFinal) { + return true; + } + pos=skipValue(pos, node); + } + } +} + +int32_t +BytesTrie::getNextBytes(ByteSink &out) const { + const uint8_t *pos=pos_; + if(pos==nullptr) { + return 0; + } + if(remainingMatchLength_>=0) { + append(out, *pos); // Next byte of a pending linear-match node. + return 1; + } + int32_t node=*pos++; + if(node>=kMinValueLead) { + if(node&kValueIsFinal) { + return 0; + } else { + pos=skipValue(pos, node); + node=*pos++; + U_ASSERT(nodekMaxBranchLinearSubNodeLength) { + ++pos; // ignore the comparison byte + getNextBranchBytes(jumpByDelta(pos), length>>1, out); + length=length-(length>>1); + pos=skipDelta(pos); + } + do { + append(out, *pos++); + pos=skipValue(pos); + } while(--length>1); + append(out, *pos); +} + +void +BytesTrie::append(ByteSink &out, int c) { + char ch=(char)c; + out.Append(&ch, 1); +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/bytestriebuilder.cpp b/intl/icu/source/common/bytestriebuilder.cpp new file mode 100644 index 0000000000..876e0dfa16 --- /dev/null +++ b/intl/icu/source/common/bytestriebuilder.cpp @@ -0,0 +1,512 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2012, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: bytestriebuilder.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010sep25 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" +#include "unicode/bytestrie.h" +#include "unicode/bytestriebuilder.h" +#include "unicode/stringpiece.h" +#include "charstr.h" +#include "cmemory.h" +#include "uhash.h" +#include "uarrsort.h" +#include "uassert.h" +#include "ustr_imp.h" + +U_NAMESPACE_BEGIN + +/* + * Note: This builder implementation stores (bytes, value) pairs with full copies + * of the byte sequences, until the BytesTrie is built. + * It might(!) take less memory if we collected the data in a temporary, dynamic trie. + */ + +class BytesTrieElement : public UMemory { +public: + // Use compiler's default constructor, initializes nothing. + + void setTo(StringPiece s, int32_t val, CharString &strings, UErrorCode &errorCode); + + StringPiece getString(const CharString &strings) const { + int32_t offset=stringOffset; + int32_t length; + if(offset>=0) { + length=(uint8_t)strings[offset++]; + } else { + offset=~offset; + length=((int32_t)(uint8_t)strings[offset]<<8)|(uint8_t)strings[offset+1]; + offset+=2; + } + return StringPiece(strings.data()+offset, length); + } + int32_t getStringLength(const CharString &strings) const { + int32_t offset=stringOffset; + if(offset>=0) { + return (uint8_t)strings[offset]; + } else { + offset=~offset; + return ((int32_t)(uint8_t)strings[offset]<<8)|(uint8_t)strings[offset+1]; + } + } + + char charAt(int32_t index, const CharString &strings) const { return data(strings)[index]; } + + int32_t getValue() const { return value; } + + int32_t compareStringTo(const BytesTrieElement &o, const CharString &strings) const; + +private: + const char *data(const CharString &strings) const { + int32_t offset=stringOffset; + if(offset>=0) { + ++offset; + } else { + offset=~offset+2; + } + return strings.data()+offset; + } + + // If the stringOffset is non-negative, then the first strings byte contains + // the string length. + // If the stringOffset is negative, then the first two strings bytes contain + // the string length (big-endian), and the offset needs to be bit-inverted. + // (Compared with a stringLength field here, this saves 3 bytes per string for most strings.) + int32_t stringOffset; + int32_t value; +}; + +void +BytesTrieElement::setTo(StringPiece s, int32_t val, + CharString &strings, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return; + } + int32_t length=s.length(); + if(length>0xffff) { + // Too long: We store the length in 1 or 2 bytes. + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return; + } + int32_t offset=strings.length(); + if(length>0xff) { + offset=~offset; + strings.append((char)(length>>8), errorCode); + } + strings.append((char)length, errorCode); + stringOffset=offset; + value=val; + strings.append(s, errorCode); +} + +int32_t +BytesTrieElement::compareStringTo(const BytesTrieElement &other, const CharString &strings) const { + // TODO: add StringPiece::compare(), see ticket #8187 + StringPiece thisString=getString(strings); + StringPiece otherString=other.getString(strings); + int32_t lengthDiff=thisString.length()-otherString.length(); + int32_t commonLength; + if(lengthDiff<=0) { + commonLength=thisString.length(); + } else { + commonLength=otherString.length(); + } + int32_t diff=uprv_memcmp(thisString.data(), otherString.data(), commonLength); + return diff!=0 ? diff : lengthDiff; +} + +BytesTrieBuilder::BytesTrieBuilder(UErrorCode &errorCode) + : strings(nullptr), elements(nullptr), elementsCapacity(0), elementsLength(0), + bytes(nullptr), bytesCapacity(0), bytesLength(0) { + if(U_FAILURE(errorCode)) { + return; + } + strings=new CharString(); + if(strings==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } +} + +BytesTrieBuilder::~BytesTrieBuilder() { + delete strings; + delete[] elements; + uprv_free(bytes); +} + +BytesTrieBuilder & +BytesTrieBuilder::add(StringPiece s, int32_t value, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return *this; + } + if(bytesLength>0) { + // Cannot add elements after building. + errorCode=U_NO_WRITE_PERMISSION; + return *this; + } + if(elementsLength==elementsCapacity) { + int32_t newCapacity; + if(elementsCapacity==0) { + newCapacity=1024; + } else { + newCapacity=4*elementsCapacity; + } + BytesTrieElement *newElements=new BytesTrieElement[newCapacity]; + if(newElements==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return *this; // error instead of dereferencing null + } + if(elementsLength>0) { + uprv_memcpy(newElements, elements, (size_t)elementsLength*sizeof(BytesTrieElement)); + } + delete[] elements; + elements=newElements; + elementsCapacity=newCapacity; + } + elements[elementsLength++].setTo(s, value, *strings, errorCode); + return *this; +} + +U_CDECL_BEGIN + +static int32_t U_CALLCONV +compareElementStrings(const void *context, const void *left, const void *right) { + const CharString *strings=static_cast(context); + const BytesTrieElement *leftElement=static_cast(left); + const BytesTrieElement *rightElement=static_cast(right); + return leftElement->compareStringTo(*rightElement, *strings); +} + +U_CDECL_END + +BytesTrie * +BytesTrieBuilder::build(UStringTrieBuildOption buildOption, UErrorCode &errorCode) { + buildBytes(buildOption, errorCode); + BytesTrie *newTrie=nullptr; + if(U_SUCCESS(errorCode)) { + newTrie=new BytesTrie(bytes, bytes+(bytesCapacity-bytesLength)); + if(newTrie==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } else { + bytes=nullptr; // The new trie now owns the array. + bytesCapacity=0; + } + } + return newTrie; +} + +StringPiece +BytesTrieBuilder::buildStringPiece(UStringTrieBuildOption buildOption, UErrorCode &errorCode) { + buildBytes(buildOption, errorCode); + StringPiece result; + if(U_SUCCESS(errorCode)) { + result.set(bytes+(bytesCapacity-bytesLength), bytesLength); + } + return result; +} + +void +BytesTrieBuilder::buildBytes(UStringTrieBuildOption buildOption, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return; + } + if(bytes!=nullptr && bytesLength>0) { + // Already built. + return; + } + if(bytesLength==0) { + if(elementsLength==0) { + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return; + } + uprv_sortArray(elements, elementsLength, (int32_t)sizeof(BytesTrieElement), + compareElementStrings, strings, + false, // need not be a stable sort + &errorCode); + if(U_FAILURE(errorCode)) { + return; + } + // Duplicate strings are not allowed. + StringPiece prev=elements[0].getString(*strings); + for(int32_t i=1; ilength(); + if(capacity<1024) { + capacity=1024; + } + if(bytesCapacity(uprv_malloc(capacity)); + if(bytes==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + bytesCapacity=0; + return; + } + bytesCapacity=capacity; + } + StringTrieBuilder::build(buildOption, elementsLength, errorCode); + if(bytes==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } +} + +BytesTrieBuilder & +BytesTrieBuilder::clear() { + strings->clear(); + elementsLength=0; + bytesLength=0; + return *this; +} + +int32_t +BytesTrieBuilder::getElementStringLength(int32_t i) const { + return elements[i].getStringLength(*strings); +} + +char16_t +BytesTrieBuilder::getElementUnit(int32_t i, int32_t byteIndex) const { + return (uint8_t)elements[i].charAt(byteIndex, *strings); +} + +int32_t +BytesTrieBuilder::getElementValue(int32_t i) const { + return elements[i].getValue(); +} + +int32_t +BytesTrieBuilder::getLimitOfLinearMatch(int32_t first, int32_t last, int32_t byteIndex) const { + const BytesTrieElement &firstElement=elements[first]; + const BytesTrieElement &lastElement=elements[last]; + int32_t minStringLength=firstElement.getStringLength(*strings); + while(++byteIndex0); + return i; +} + +int32_t +BytesTrieBuilder::indexOfElementWithNextUnit(int32_t i, int32_t byteIndex, char16_t byte) const { + char b=(char)byte; + while(b==elements[i].charAt(byteIndex, *strings)) { + ++i; + } + return i; +} + +BytesTrieBuilder::BTLinearMatchNode::BTLinearMatchNode(const char *bytes, int32_t len, Node *nextNode) + : LinearMatchNode(len, nextNode), s(bytes) { + hash=static_cast( + static_cast(hash)*37u + static_cast(ustr_hashCharsN(bytes, len))); +} + +bool +BytesTrieBuilder::BTLinearMatchNode::operator==(const Node &other) const { + if(this==&other) { + return true; + } + if(!LinearMatchNode::operator==(other)) { + return false; + } + const BTLinearMatchNode &o=static_cast(other); + return 0==uprv_memcmp(s, o.s, length); +} + +void +BytesTrieBuilder::BTLinearMatchNode::write(StringTrieBuilder &builder) { + BytesTrieBuilder &b=static_cast(builder); + next->write(builder); + b.write(s, length); + offset=b.write(b.getMinLinearMatch()+length-1); +} + +StringTrieBuilder::Node * +BytesTrieBuilder::createLinearMatchNode(int32_t i, int32_t byteIndex, int32_t length, + Node *nextNode) const { + return new BTLinearMatchNode( + elements[i].getString(*strings).data()+byteIndex, + length, + nextNode); +} + +UBool +BytesTrieBuilder::ensureCapacity(int32_t length) { + if(bytes==nullptr) { + return false; // previous memory allocation had failed + } + if(length>bytesCapacity) { + int32_t newCapacity=bytesCapacity; + do { + newCapacity*=2; + } while(newCapacity<=length); + char *newBytes=static_cast(uprv_malloc(newCapacity)); + if(newBytes==nullptr) { + // unable to allocate memory + uprv_free(bytes); + bytes=nullptr; + bytesCapacity=0; + return false; + } + uprv_memcpy(newBytes+(newCapacity-bytesLength), + bytes+(bytesCapacity-bytesLength), bytesLength); + uprv_free(bytes); + bytes=newBytes; + bytesCapacity=newCapacity; + } + return true; +} + +int32_t +BytesTrieBuilder::write(int32_t byte) { + int32_t newLength=bytesLength+1; + if(ensureCapacity(newLength)) { + bytesLength=newLength; + bytes[bytesCapacity-bytesLength]=(char)byte; + } + return bytesLength; +} + +int32_t +BytesTrieBuilder::write(const char *b, int32_t length) { + int32_t newLength=bytesLength+length; + if(ensureCapacity(newLength)) { + bytesLength=newLength; + uprv_memcpy(bytes+(bytesCapacity-bytesLength), b, length); + } + return bytesLength; +} + +int32_t +BytesTrieBuilder::writeElementUnits(int32_t i, int32_t byteIndex, int32_t length) { + return write(elements[i].getString(*strings).data()+byteIndex, length); +} + +int32_t +BytesTrieBuilder::writeValueAndFinal(int32_t i, UBool isFinal) { + if(0<=i && i<=BytesTrie::kMaxOneByteValue) { + return write(((BytesTrie::kMinOneByteValueLead+i)<<1)|isFinal); + } + char intBytes[5]; + int32_t length=1; + if(i<0 || i>0xffffff) { + intBytes[0]=(char)BytesTrie::kFiveByteValueLead; + intBytes[1]=(char)((uint32_t)i>>24); + intBytes[2]=(char)((uint32_t)i>>16); + intBytes[3]=(char)((uint32_t)i>>8); + intBytes[4]=(char)i; + length=5; + // } else if(i<=BytesTrie::kMaxOneByteValue) { + // intBytes[0]=(char)(BytesTrie::kMinOneByteValueLead+i); + } else { + if(i<=BytesTrie::kMaxTwoByteValue) { + intBytes[0]=(char)(BytesTrie::kMinTwoByteValueLead+(i>>8)); + } else { + if(i<=BytesTrie::kMaxThreeByteValue) { + intBytes[0]=(char)(BytesTrie::kMinThreeByteValueLead+(i>>16)); + } else { + intBytes[0]=(char)BytesTrie::kFourByteValueLead; + intBytes[1]=(char)(i>>16); + length=2; + } + intBytes[length++]=(char)(i>>8); + } + intBytes[length++]=(char)i; + } + intBytes[0]=(char)((intBytes[0]<<1)|isFinal); + return write(intBytes, length); +} + +int32_t +BytesTrieBuilder::writeValueAndType(UBool hasValue, int32_t value, int32_t node) { + int32_t offset=write(node); + if(hasValue) { + offset=writeValueAndFinal(value, false); + } + return offset; +} + +int32_t +BytesTrieBuilder::writeDeltaTo(int32_t jumpTarget) { + int32_t i=bytesLength-jumpTarget; + U_ASSERT(i>=0); + if(i<=BytesTrie::kMaxOneByteDelta) { + return write(i); + } else { + char intBytes[5]; + return write(intBytes, internalEncodeDelta(i, intBytes)); + } +} + +int32_t +BytesTrieBuilder::internalEncodeDelta(int32_t i, char intBytes[]) { + U_ASSERT(i>=0); + if(i<=BytesTrie::kMaxOneByteDelta) { + intBytes[0]=(char)i; + return 1; + } + int32_t length=1; + if(i<=BytesTrie::kMaxTwoByteDelta) { + intBytes[0]=(char)(BytesTrie::kMinTwoByteDeltaLead+(i>>8)); + } else { + if(i<=BytesTrie::kMaxThreeByteDelta) { + intBytes[0]=(char)(BytesTrie::kMinThreeByteDeltaLead+(i>>16)); + } else { + if(i<=0xffffff) { + intBytes[0]=(char)BytesTrie::kFourByteDeltaLead; + } else { + intBytes[0]=(char)BytesTrie::kFiveByteDeltaLead; + intBytes[1]=(char)(i>>24); + length=2; + } + intBytes[length++]=(char)(i>>16); + } + intBytes[length++]=(char)(i>>8); + } + intBytes[length++]=(char)i; + return length; +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/bytestrieiterator.cpp b/intl/icu/source/common/bytestrieiterator.cpp new file mode 100644 index 0000000000..65f54be48a --- /dev/null +++ b/intl/icu/source/common/bytestrieiterator.cpp @@ -0,0 +1,214 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2012, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: bytestrieiterator.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010nov03 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" +#include "unicode/bytestrie.h" +#include "unicode/stringpiece.h" +#include "charstr.h" +#include "uvectr32.h" + +U_NAMESPACE_BEGIN + +BytesTrie::Iterator::Iterator(const void *trieBytes, int32_t maxStringLength, + UErrorCode &errorCode) + : bytes_(static_cast(trieBytes)), + pos_(bytes_), initialPos_(bytes_), + remainingMatchLength_(-1), initialRemainingMatchLength_(-1), + str_(nullptr), maxLength_(maxStringLength), value_(0), stack_(nullptr) { + if(U_FAILURE(errorCode)) { + return; + } + // str_ and stack_ are pointers so that it's easy to turn bytestrie.h into + // a public API header for which we would want it to depend only on + // other public headers. + // Unlike BytesTrie itself, its Iterator performs memory allocations anyway + // via the CharString and UVector32 implementations, so this additional + // cost is minimal. + str_=new CharString(); + stack_=new UVector32(errorCode); + if(U_SUCCESS(errorCode) && (str_==nullptr || stack_==nullptr)) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } +} + +BytesTrie::Iterator::Iterator(const BytesTrie &trie, int32_t maxStringLength, + UErrorCode &errorCode) + : bytes_(trie.bytes_), pos_(trie.pos_), initialPos_(trie.pos_), + remainingMatchLength_(trie.remainingMatchLength_), + initialRemainingMatchLength_(trie.remainingMatchLength_), + str_(nullptr), maxLength_(maxStringLength), value_(0), stack_(nullptr) { + if(U_FAILURE(errorCode)) { + return; + } + str_=new CharString(); + stack_=new UVector32(errorCode); + if(U_FAILURE(errorCode)) { + return; + } + if(str_==nullptr || stack_==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + int32_t length=remainingMatchLength_; // Actual remaining match length minus 1. + if(length>=0) { + // Pending linear-match node, append remaining bytes to str_. + ++length; + if(maxLength_>0 && length>maxLength_) { + length=maxLength_; // This will leave remainingMatchLength>=0 as a signal. + } + str_->append(reinterpret_cast(pos_), length, errorCode); + pos_+=length; + remainingMatchLength_-=length; + } +} + +BytesTrie::Iterator::~Iterator() { + delete str_; + delete stack_; +} + +BytesTrie::Iterator & +BytesTrie::Iterator::reset() { + pos_=initialPos_; + remainingMatchLength_=initialRemainingMatchLength_; + int32_t length=remainingMatchLength_+1; // Remaining match length. + if(maxLength_>0 && length>maxLength_) { + length=maxLength_; + } + str_->truncate(length); + pos_+=length; + remainingMatchLength_-=length; + stack_->setSize(0); + return *this; +} + +UBool +BytesTrie::Iterator::hasNext() const { return pos_!=nullptr || !stack_->isEmpty(); } + +UBool +BytesTrie::Iterator::next(UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return false; + } + const uint8_t *pos=pos_; + if(pos==nullptr) { + if(stack_->isEmpty()) { + return false; + } + // Pop the state off the stack and continue with the next outbound edge of + // the branch node. + int32_t stackSize=stack_->size(); + int32_t length=stack_->elementAti(stackSize-1); + pos=bytes_+stack_->elementAti(stackSize-2); + stack_->setSize(stackSize-2); + str_->truncate(length&0xffff); + length=(int32_t)((uint32_t)length>>16); + if(length>1) { + pos=branchNext(pos, length, errorCode); + if(pos==nullptr) { + return true; // Reached a final value. + } + } else { + str_->append((char)*pos++, errorCode); + } + } + if(remainingMatchLength_>=0) { + // We only get here if we started in a pending linear-match node + // with more than maxLength remaining bytes. + return truncateAndStop(); + } + for(;;) { + int32_t node=*pos++; + if(node>=kMinValueLead) { + // Deliver value for the byte sequence so far. + UBool isFinal=(UBool)(node&kValueIsFinal); + value_=readValue(pos, node>>1); + if(isFinal || (maxLength_>0 && str_->length()==maxLength_)) { + pos_=nullptr; + } else { + pos_=skipValue(pos, node); + } + return true; + } + if(maxLength_>0 && str_->length()==maxLength_) { + return truncateAndStop(); + } + if(node0 && str_->length()+length>maxLength_) { + str_->append(reinterpret_cast(pos), + maxLength_-str_->length(), errorCode); + return truncateAndStop(); + } + str_->append(reinterpret_cast(pos), length, errorCode); + pos+=length; + } + } +} + +StringPiece +BytesTrie::Iterator::getString() const { + return str_ == nullptr ? StringPiece() : str_->toStringPiece(); +} + +UBool +BytesTrie::Iterator::truncateAndStop() { + pos_=nullptr; + value_=-1; // no real value for str + return true; +} + +// Branch node, needs to take the first outbound edge and push state for the rest. +const uint8_t * +BytesTrie::Iterator::branchNext(const uint8_t *pos, int32_t length, UErrorCode &errorCode) { + while(length>kMaxBranchLinearSubNodeLength) { + ++pos; // ignore the comparison byte + // Push state for the greater-or-equal edge. + stack_->addElement((int32_t)(skipDelta(pos)-bytes_), errorCode); + stack_->addElement(((length-(length>>1))<<16)|str_->length(), errorCode); + // Follow the less-than edge. + length>>=1; + pos=jumpByDelta(pos); + } + // List of key-value pairs where values are either final values or jump deltas. + // Read the first (key, value) pair. + uint8_t trieByte=*pos++; + int32_t node=*pos++; + UBool isFinal=(UBool)(node&kValueIsFinal); + int32_t value=readValue(pos, node>>1); + pos=skipValue(pos, node); + stack_->addElement((int32_t)(pos-bytes_), errorCode); + stack_->addElement(((length-1)<<16)|str_->length(), errorCode); + str_->append((char)trieByte, errorCode); + if(isFinal) { + pos_=nullptr; + value_=value; + return nullptr; + } else { + return pos+value; + } +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/caniter.cpp b/intl/icu/source/common/caniter.cpp new file mode 100644 index 0000000000..64a3c65d29 --- /dev/null +++ b/intl/icu/source/common/caniter.cpp @@ -0,0 +1,586 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ***************************************************************************** + * Copyright (C) 1996-2015, International Business Machines Corporation and + * others. All Rights Reserved. + ***************************************************************************** + */ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_NORMALIZATION + +#include "unicode/caniter.h" +#include "unicode/normalizer2.h" +#include "unicode/uchar.h" +#include "unicode/uniset.h" +#include "unicode/usetiter.h" +#include "unicode/ustring.h" +#include "unicode/utf16.h" +#include "cmemory.h" +#include "hash.h" +#include "normalizer2impl.h" + +/** + * This class allows one to iterate through all the strings that are canonically equivalent to a given + * string. For example, here are some sample results: +Results for: {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D}{COMBINING DOT ABOVE}{COMBINING CEDILLA} +1: \u0041\u030A\u0064\u0307\u0327 + = {LATIN CAPITAL LETTER A}{COMBINING RING ABOVE}{LATIN SMALL LETTER D}{COMBINING DOT ABOVE}{COMBINING CEDILLA} +2: \u0041\u030A\u0064\u0327\u0307 + = {LATIN CAPITAL LETTER A}{COMBINING RING ABOVE}{LATIN SMALL LETTER D}{COMBINING CEDILLA}{COMBINING DOT ABOVE} +3: \u0041\u030A\u1E0B\u0327 + = {LATIN CAPITAL LETTER A}{COMBINING RING ABOVE}{LATIN SMALL LETTER D WITH DOT ABOVE}{COMBINING CEDILLA} +4: \u0041\u030A\u1E11\u0307 + = {LATIN CAPITAL LETTER A}{COMBINING RING ABOVE}{LATIN SMALL LETTER D WITH CEDILLA}{COMBINING DOT ABOVE} +5: \u00C5\u0064\u0307\u0327 + = {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D}{COMBINING DOT ABOVE}{COMBINING CEDILLA} +6: \u00C5\u0064\u0327\u0307 + = {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D}{COMBINING CEDILLA}{COMBINING DOT ABOVE} +7: \u00C5\u1E0B\u0327 + = {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D WITH DOT ABOVE}{COMBINING CEDILLA} +8: \u00C5\u1E11\u0307 + = {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D WITH CEDILLA}{COMBINING DOT ABOVE} +9: \u212B\u0064\u0307\u0327 + = {ANGSTROM SIGN}{LATIN SMALL LETTER D}{COMBINING DOT ABOVE}{COMBINING CEDILLA} +10: \u212B\u0064\u0327\u0307 + = {ANGSTROM SIGN}{LATIN SMALL LETTER D}{COMBINING CEDILLA}{COMBINING DOT ABOVE} +11: \u212B\u1E0B\u0327 + = {ANGSTROM SIGN}{LATIN SMALL LETTER D WITH DOT ABOVE}{COMBINING CEDILLA} +12: \u212B\u1E11\u0307 + = {ANGSTROM SIGN}{LATIN SMALL LETTER D WITH CEDILLA}{COMBINING DOT ABOVE} + *
Note: the code is intended for use with small strings, and is not suitable for larger ones, + * since it has not been optimized for that situation. + *@author M. Davis + *@draft + */ + +// public + +U_NAMESPACE_BEGIN + +// TODO: add boilerplate methods. + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(CanonicalIterator) + +/** + *@param source string to get results for + */ +CanonicalIterator::CanonicalIterator(const UnicodeString &sourceStr, UErrorCode &status) : + pieces(nullptr), + pieces_length(0), + pieces_lengths(nullptr), + current(nullptr), + current_length(0), + nfd(*Normalizer2::getNFDInstance(status)), + nfcImpl(*Normalizer2Factory::getNFCImpl(status)) +{ + if(U_SUCCESS(status) && nfcImpl.ensureCanonIterData(status)) { + setSource(sourceStr, status); + } +} + +CanonicalIterator::~CanonicalIterator() { + cleanPieces(); +} + +void CanonicalIterator::cleanPieces() { + int32_t i = 0; + if(pieces != nullptr) { + for(i = 0; i < pieces_length; i++) { + if(pieces[i] != nullptr) { + delete[] pieces[i]; + } + } + uprv_free(pieces); + pieces = nullptr; + pieces_length = 0; + } + if(pieces_lengths != nullptr) { + uprv_free(pieces_lengths); + pieces_lengths = nullptr; + } + if(current != nullptr) { + uprv_free(current); + current = nullptr; + current_length = 0; + } +} + +/** + *@return gets the source: NOTE: it is the NFD form of source + */ +UnicodeString CanonicalIterator::getSource() { + return source; +} + +/** + * Resets the iterator so that one can start again from the beginning. + */ +void CanonicalIterator::reset() { + done = false; + for (int i = 0; i < current_length; ++i) { + current[i] = 0; + } +} + +/** + *@return the next string that is canonically equivalent. The value null is returned when + * the iteration is done. + */ +UnicodeString CanonicalIterator::next() { + int32_t i = 0; + + if (done) { + buffer.setToBogus(); + return buffer; + } + + // delete old contents + buffer.remove(); + + // construct return value + + for (i = 0; i < pieces_length; ++i) { + buffer.append(pieces[i][current[i]]); + } + //String result = buffer.toString(); // not needed + + // find next value for next time + + for (i = current_length - 1; ; --i) { + if (i < 0) { + done = true; + break; + } + current[i]++; + if (current[i] < pieces_lengths[i]) break; // got sequence + current[i] = 0; + } + return buffer; +} + +/** + *@param set the source string to iterate against. This allows the same iterator to be used + * while changing the source string, saving object creation. + */ +void CanonicalIterator::setSource(const UnicodeString &newSource, UErrorCode &status) { + int32_t list_length = 0; + UChar32 cp = 0; + int32_t start = 0; + int32_t i = 0; + UnicodeString *list = nullptr; + + nfd.normalize(newSource, source, status); + if(U_FAILURE(status)) { + return; + } + done = false; + + cleanPieces(); + + // catch degenerate case + if (newSource.length() == 0) { + pieces = (UnicodeString **)uprv_malloc(sizeof(UnicodeString *)); + pieces_lengths = (int32_t*)uprv_malloc(1 * sizeof(int32_t)); + pieces_length = 1; + current = (int32_t*)uprv_malloc(1 * sizeof(int32_t)); + current_length = 1; + if (pieces == nullptr || pieces_lengths == nullptr || current == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + goto CleanPartialInitialization; + } + current[0] = 0; + pieces[0] = new UnicodeString[1]; + pieces_lengths[0] = 1; + if (pieces[0] == 0) { + status = U_MEMORY_ALLOCATION_ERROR; + goto CleanPartialInitialization; + } + return; + } + + + list = new UnicodeString[source.length()]; + if (list == 0) { + status = U_MEMORY_ALLOCATION_ERROR; + goto CleanPartialInitialization; + } + + // i should initially be the number of code units at the + // start of the string + i = U16_LENGTH(source.char32At(0)); + // int32_t i = 1; + // find the segments + // This code iterates through the source string and + // extracts segments that end up on a codepoint that + // doesn't start any decompositions. (Analysis is done + // on the NFD form - see above). + for (; i < source.length(); i += U16_LENGTH(cp)) { + cp = source.char32At(i); + if (nfcImpl.isCanonSegmentStarter(cp)) { + source.extract(start, i-start, list[list_length++]); // add up to i + start = i; + } + } + source.extract(start, i-start, list[list_length++]); // add last one + + + // allocate the arrays, and find the strings that are CE to each segment + pieces = (UnicodeString **)uprv_malloc(list_length * sizeof(UnicodeString *)); + pieces_length = list_length; + pieces_lengths = (int32_t*)uprv_malloc(list_length * sizeof(int32_t)); + current = (int32_t*)uprv_malloc(list_length * sizeof(int32_t)); + current_length = list_length; + if (pieces == nullptr || pieces_lengths == nullptr || current == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + goto CleanPartialInitialization; + } + + for (i = 0; i < current_length; i++) { + current[i] = 0; + } + // for each segment, get all the combinations that can produce + // it after NFD normalization + for (i = 0; i < pieces_length; ++i) { + //if (PROGRESS) printf("SEGMENT\n"); + pieces[i] = getEquivalents(list[i], pieces_lengths[i], status); + } + + delete[] list; + return; +// Common section to cleanup all local variables and reset object variables. +CleanPartialInitialization: + if (list != nullptr) { + delete[] list; + } + cleanPieces(); +} + +/** + * Dumb recursive implementation of permutation. + * TODO: optimize + * @param source the string to find permutations for + * @return the results in a set. + */ +void U_EXPORT2 CanonicalIterator::permute(UnicodeString &source, UBool skipZeros, Hashtable *result, UErrorCode &status) { + if(U_FAILURE(status)) { + return; + } + //if (PROGRESS) printf("Permute: %s\n", UToS(Tr(source))); + int32_t i = 0; + + // optimization: + // if zero or one character, just return a set with it + // we check for length < 2 to keep from counting code points all the time + if (source.length() <= 2 && source.countChar32() <= 1) { + UnicodeString *toPut = new UnicodeString(source); + /* test for nullptr */ + if (toPut == 0) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + result->put(source, toPut, status); + return; + } + + // otherwise iterate through the string, and recursively permute all the other characters + UChar32 cp; + Hashtable subpermute(status); + if(U_FAILURE(status)) { + return; + } + subpermute.setValueDeleter(uprv_deleteUObject); + + for (i = 0; i < source.length(); i += U16_LENGTH(cp)) { + cp = source.char32At(i); + const UHashElement *ne = nullptr; + int32_t el = UHASH_FIRST; + UnicodeString subPermuteString = source; + + // optimization: + // if the character is canonical combining class zero, + // don't permute it + if (skipZeros && i != 0 && u_getCombiningClass(cp) == 0) { + //System.out.println("Skipping " + Utility.hex(UTF16.valueOf(source, i))); + continue; + } + + subpermute.removeAll(); + + // see what the permutations of the characters before and after this one are + //Hashtable *subpermute = permute(source.substring(0,i) + source.substring(i + UTF16.getCharCount(cp))); + permute(subPermuteString.remove(i, U16_LENGTH(cp)), skipZeros, &subpermute, status); + /* Test for buffer overflows */ + if(U_FAILURE(status)) { + return; + } + // The upper remove is destructive. The question is do we have to make a copy, or we don't care about the contents + // of source at this point. + + // prefix this character to all of them + ne = subpermute.nextElement(el); + while (ne != nullptr) { + UnicodeString *permRes = (UnicodeString *)(ne->value.pointer); + UnicodeString *chStr = new UnicodeString(cp); + //test for nullptr + if (chStr == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + chStr->append(*permRes); //*((UnicodeString *)(ne->value.pointer)); + //if (PROGRESS) printf(" Piece: %s\n", UToS(*chStr)); + result->put(*chStr, chStr, status); + ne = subpermute.nextElement(el); + } + } + //return result; +} + +// privates + +// we have a segment, in NFD. Find all the strings that are canonically equivalent to it. +UnicodeString* CanonicalIterator::getEquivalents(const UnicodeString &segment, int32_t &result_len, UErrorCode &status) { + Hashtable result(status); + Hashtable permutations(status); + Hashtable basic(status); + if (U_FAILURE(status)) { + return 0; + } + result.setValueDeleter(uprv_deleteUObject); + permutations.setValueDeleter(uprv_deleteUObject); + basic.setValueDeleter(uprv_deleteUObject); + + char16_t USeg[256]; + int32_t segLen = segment.extract(USeg, 256, status); + getEquivalents2(&basic, USeg, segLen, status); + + // now get all the permutations + // add only the ones that are canonically equivalent + // TODO: optimize by not permuting any class zero. + + const UHashElement *ne = nullptr; + int32_t el = UHASH_FIRST; + //Iterator it = basic.iterator(); + ne = basic.nextElement(el); + //while (it.hasNext()) + while (ne != nullptr) { + //String item = (String) it.next(); + UnicodeString item = *((UnicodeString *)(ne->value.pointer)); + + permutations.removeAll(); + permute(item, CANITER_SKIP_ZEROES, &permutations, status); + const UHashElement *ne2 = nullptr; + int32_t el2 = UHASH_FIRST; + //Iterator it2 = permutations.iterator(); + ne2 = permutations.nextElement(el2); + //while (it2.hasNext()) + while (ne2 != nullptr) { + //String possible = (String) it2.next(); + //UnicodeString *possible = new UnicodeString(*((UnicodeString *)(ne2->value.pointer))); + UnicodeString possible(*((UnicodeString *)(ne2->value.pointer))); + UnicodeString attempt; + nfd.normalize(possible, attempt, status); + + // TODO: check if operator == is semanticaly the same as attempt.equals(segment) + if (attempt==segment) { + //if (PROGRESS) printf("Adding Permutation: %s\n", UToS(Tr(*possible))); + // TODO: use the hashtable just to catch duplicates - store strings directly (somehow). + result.put(possible, new UnicodeString(possible), status); //add(possible); + } else { + //if (PROGRESS) printf("-Skipping Permutation: %s\n", UToS(Tr(*possible))); + } + + ne2 = permutations.nextElement(el2); + } + ne = basic.nextElement(el); + } + + /* Test for buffer overflows */ + if(U_FAILURE(status)) { + return 0; + } + // convert into a String[] to clean up storage + //String[] finalResult = new String[result.size()]; + UnicodeString *finalResult = nullptr; + int32_t resultCount; + if((resultCount = result.count()) != 0) { + finalResult = new UnicodeString[resultCount]; + if (finalResult == 0) { + status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + } + else { + status = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + //result.toArray(finalResult); + result_len = 0; + el = UHASH_FIRST; + ne = result.nextElement(el); + while(ne != nullptr) { + finalResult[result_len++] = *((UnicodeString *)(ne->value.pointer)); + ne = result.nextElement(el); + } + + + return finalResult; +} + +Hashtable *CanonicalIterator::getEquivalents2(Hashtable *fillinResult, const char16_t *segment, int32_t segLen, UErrorCode &status) { + + if (U_FAILURE(status)) { + return nullptr; + } + + //if (PROGRESS) printf("Adding: %s\n", UToS(Tr(segment))); + + UnicodeString toPut(segment, segLen); + + fillinResult->put(toPut, new UnicodeString(toPut), status); + + UnicodeSet starts; + + // cycle through all the characters + UChar32 cp; + for (int32_t i = 0; i < segLen; i += U16_LENGTH(cp)) { + // see if any character is at the start of some decomposition + U16_GET(segment, 0, i, segLen, cp); + if (!nfcImpl.getCanonStartSet(cp, starts)) { + continue; + } + // if so, see which decompositions match + UnicodeSetIterator iter(starts); + while (iter.next()) { + UChar32 cp2 = iter.getCodepoint(); + Hashtable remainder(status); + remainder.setValueDeleter(uprv_deleteUObject); + if (extract(&remainder, cp2, segment, segLen, i, status) == nullptr) { + continue; + } + + // there were some matches, so add all the possibilities to the set. + UnicodeString prefix(segment, i); + prefix += cp2; + + int32_t el = UHASH_FIRST; + const UHashElement *ne = remainder.nextElement(el); + while (ne != nullptr) { + UnicodeString item = *((UnicodeString *)(ne->value.pointer)); + UnicodeString *toAdd = new UnicodeString(prefix); + /* test for nullptr */ + if (toAdd == 0) { + status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + *toAdd += item; + fillinResult->put(*toAdd, toAdd, status); + + //if (PROGRESS) printf("Adding: %s\n", UToS(Tr(*toAdd))); + + ne = remainder.nextElement(el); + } + } + } + + /* Test for buffer overflows */ + if(U_FAILURE(status)) { + return nullptr; + } + return fillinResult; +} + +/** + * See if the decomposition of cp2 is at segment starting at segmentPos + * (with canonical rearrangement!) + * If so, take the remainder, and return the equivalents + */ +Hashtable *CanonicalIterator::extract(Hashtable *fillinResult, UChar32 comp, const char16_t *segment, int32_t segLen, int32_t segmentPos, UErrorCode &status) { +//Hashtable *CanonicalIterator::extract(UChar32 comp, const UnicodeString &segment, int32_t segLen, int32_t segmentPos, UErrorCode &status) { + //if (PROGRESS) printf(" extract: %s, ", UToS(Tr(UnicodeString(comp)))); + //if (PROGRESS) printf("%s, %i\n", UToS(Tr(segment)), segmentPos); + + if (U_FAILURE(status)) { + return nullptr; + } + + UnicodeString temp(comp); + int32_t inputLen=temp.length(); + UnicodeString decompString; + nfd.normalize(temp, decompString, status); + if (U_FAILURE(status)) { + return nullptr; + } + if (decompString.isBogus()) { + status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + const char16_t *decomp=decompString.getBuffer(); + int32_t decompLen=decompString.length(); + + // See if it matches the start of segment (at segmentPos) + UBool ok = false; + UChar32 cp; + int32_t decompPos = 0; + UChar32 decompCp; + U16_NEXT(decomp, decompPos, decompLen, decompCp); + + int32_t i = segmentPos; + while(i < segLen) { + U16_NEXT(segment, i, segLen, cp); + + if (cp == decompCp) { // if equal, eat another cp from decomp + + //if (PROGRESS) printf(" matches: %s\n", UToS(Tr(UnicodeString(cp)))); + + if (decompPos == decompLen) { // done, have all decomp characters! + temp.append(segment+i, segLen-i); + ok = true; + break; + } + U16_NEXT(decomp, decompPos, decompLen, decompCp); + } else { + //if (PROGRESS) printf(" buffer: %s\n", UToS(Tr(UnicodeString(cp)))); + + // brute force approach + temp.append(cp); + + /* TODO: optimize + // since we know that the classes are monotonically increasing, after zero + // e.g. 0 5 7 9 0 3 + // we can do an optimization + // there are only a few cases that work: zero, less, same, greater + // if both classes are the same, we fail + // if the decomp class < the segment class, we fail + + segClass = getClass(cp); + if (decompClass <= segClass) return null; + */ + } + } + if (!ok) + return nullptr; // we failed, characters left over + + //if (PROGRESS) printf("Matches\n"); + + if (inputLen == temp.length()) { + fillinResult->put(UnicodeString(), new UnicodeString(), status); + return fillinResult; // succeed, but no remainder + } + + // brute force approach + // check to make sure result is canonically equivalent + UnicodeString trial; + nfd.normalize(temp, trial, status); + if(U_FAILURE(status) || trial.compare(segment+segmentPos, segLen - segmentPos) != 0) { + return nullptr; + } + + return getEquivalents2(fillinResult, temp.getBuffer()+inputLen, temp.length()-inputLen, status); +} + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_NORMALIZATION */ diff --git a/intl/icu/source/common/capi_helper.h b/intl/icu/source/common/capi_helper.h new file mode 100644 index 0000000000..54b1db9e33 --- /dev/null +++ b/intl/icu/source/common/capi_helper.h @@ -0,0 +1,97 @@ +// © 2018 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +#ifndef __CAPI_HELPER_H__ +#define __CAPI_HELPER_H__ + +#include "unicode/utypes.h" + +U_NAMESPACE_BEGIN + +/** + * An internal helper class to help convert between C and C++ APIs. + */ +template +class IcuCApiHelper { + public: + /** + * Convert from the C type to the C++ type (const version). + */ + static const CPPType* validate(const CType* input, UErrorCode& status); + + /** + * Convert from the C type to the C++ type (non-const version). + */ + static CPPType* validate(CType* input, UErrorCode& status); + + /** + * Convert from the C++ type to the C type (const version). + */ + const CType* exportConstForC() const; + + /** + * Convert from the C++ type to the C type (non-const version). + */ + CType* exportForC(); + + /** + * Invalidates the object. + */ + ~IcuCApiHelper(); + + private: + /** + * While the object is valid, fMagic equals kMagic. + */ + int32_t fMagic = kMagic; +}; + + +template +const CPPType* +IcuCApiHelper::validate(const CType* input, UErrorCode& status) { + if (U_FAILURE(status)) { + return nullptr; + } + if (input == nullptr) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + auto* impl = reinterpret_cast(input); + if (static_cast*>(impl)->fMagic != kMagic) { + status = U_INVALID_FORMAT_ERROR; + return nullptr; + } + return impl; +} + +template +CPPType* +IcuCApiHelper::validate(CType* input, UErrorCode& status) { + auto* constInput = static_cast(input); + auto* validated = validate(constInput, status); + return const_cast(validated); +} + +template +const CType* +IcuCApiHelper::exportConstForC() const { + return reinterpret_cast(static_cast(this)); +} + +template +CType* +IcuCApiHelper::exportForC() { + return reinterpret_cast(static_cast(this)); +} + +template +IcuCApiHelper::~IcuCApiHelper() { + // head off application errors by preventing use of of deleted objects. + fMagic = 0; +} + + +U_NAMESPACE_END + +#endif // __CAPI_HELPER_H__ diff --git a/intl/icu/source/common/characterproperties.cpp b/intl/icu/source/common/characterproperties.cpp new file mode 100644 index 0000000000..978e6761ce --- /dev/null +++ b/intl/icu/source/common/characterproperties.cpp @@ -0,0 +1,419 @@ +// © 2018 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// characterproperties.cpp +// created: 2018sep03 Markus W. Scherer + +#include "unicode/utypes.h" +#include "unicode/localpointer.h" +#include "unicode/uchar.h" +#include "unicode/ucpmap.h" +#include "unicode/ucptrie.h" +#include "unicode/umutablecptrie.h" +#include "unicode/uniset.h" +#include "unicode/uscript.h" +#include "unicode/uset.h" +#include "cmemory.h" +#include "emojiprops.h" +#include "mutex.h" +#include "normalizer2impl.h" +#include "uassert.h" +#include "ubidi_props.h" +#include "ucase.h" +#include "ucln_cmn.h" +#include "umutex.h" +#include "uprops.h" + +using icu::LocalPointer; +#if !UCONFIG_NO_NORMALIZATION +using icu::Normalizer2Factory; +using icu::Normalizer2Impl; +#endif +using icu::UInitOnce; +using icu::UnicodeSet; + +namespace { + +UBool U_CALLCONV characterproperties_cleanup(); + +constexpr int32_t NUM_INCLUSIONS = UPROPS_SRC_COUNT + (UCHAR_INT_LIMIT - UCHAR_INT_START); + +struct Inclusion { + UnicodeSet *fSet = nullptr; + UInitOnce fInitOnce {}; +}; +Inclusion gInclusions[NUM_INCLUSIONS]; // cached getInclusions() + +UnicodeSet *sets[UCHAR_BINARY_LIMIT] = {}; + +UCPMap *maps[UCHAR_INT_LIMIT - UCHAR_INT_START] = {}; + +icu::UMutex cpMutex; + +//---------------------------------------------------------------- +// Inclusions list +//---------------------------------------------------------------- + +// USetAdder implementation +// Does not use uset.h to reduce code dependencies +void U_CALLCONV +_set_add(USet *set, UChar32 c) { + ((UnicodeSet *)set)->add(c); +} + +void U_CALLCONV +_set_addRange(USet *set, UChar32 start, UChar32 end) { + ((UnicodeSet *)set)->add(start, end); +} + +void U_CALLCONV +_set_addString(USet *set, const char16_t *str, int32_t length) { + ((UnicodeSet *)set)->add(icu::UnicodeString((UBool)(length<0), str, length)); +} + +UBool U_CALLCONV characterproperties_cleanup() { + for (Inclusion &in: gInclusions) { + delete in.fSet; + in.fSet = nullptr; + in.fInitOnce.reset(); + } + for (int32_t i = 0; i < UPRV_LENGTHOF(sets); ++i) { + delete sets[i]; + sets[i] = nullptr; + } + for (int32_t i = 0; i < UPRV_LENGTHOF(maps); ++i) { + ucptrie_close(reinterpret_cast(maps[i])); + maps[i] = nullptr; + } + return true; +} + +void U_CALLCONV initInclusion(UPropertySource src, UErrorCode &errorCode) { + // This function is invoked only via umtx_initOnce(). + U_ASSERT(0 <= src && src < UPROPS_SRC_COUNT); + if (src == UPROPS_SRC_NONE) { + errorCode = U_INTERNAL_PROGRAM_ERROR; + return; + } + U_ASSERT(gInclusions[src].fSet == nullptr); + + LocalPointer incl(new UnicodeSet()); + if (incl.isNull()) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + USetAdder sa = { + (USet *)incl.getAlias(), + _set_add, + _set_addRange, + _set_addString, + nullptr, // don't need remove() + nullptr // don't need removeRange() + }; + + switch(src) { + case UPROPS_SRC_CHAR: + uchar_addPropertyStarts(&sa, &errorCode); + break; + case UPROPS_SRC_PROPSVEC: + upropsvec_addPropertyStarts(&sa, &errorCode); + break; + case UPROPS_SRC_CHAR_AND_PROPSVEC: + uchar_addPropertyStarts(&sa, &errorCode); + upropsvec_addPropertyStarts(&sa, &errorCode); + break; +#if !UCONFIG_NO_NORMALIZATION + case UPROPS_SRC_CASE_AND_NORM: { + const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode); + if(U_SUCCESS(errorCode)) { + impl->addPropertyStarts(&sa, errorCode); + } + ucase_addPropertyStarts(&sa, &errorCode); + break; + } + case UPROPS_SRC_NFC: { + const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode); + if(U_SUCCESS(errorCode)) { + impl->addPropertyStarts(&sa, errorCode); + } + break; + } + case UPROPS_SRC_NFKC: { + const Normalizer2Impl *impl=Normalizer2Factory::getNFKCImpl(errorCode); + if(U_SUCCESS(errorCode)) { + impl->addPropertyStarts(&sa, errorCode); + } + break; + } + case UPROPS_SRC_NFKC_CF: { + const Normalizer2Impl *impl=Normalizer2Factory::getNFKC_CFImpl(errorCode); + if(U_SUCCESS(errorCode)) { + impl->addPropertyStarts(&sa, errorCode); + } + break; + } + case UPROPS_SRC_NFC_CANON_ITER: { + const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode); + if(U_SUCCESS(errorCode)) { + impl->addCanonIterPropertyStarts(&sa, errorCode); + } + break; + } +#endif + case UPROPS_SRC_CASE: + ucase_addPropertyStarts(&sa, &errorCode); + break; + case UPROPS_SRC_BIDI: + ubidi_addPropertyStarts(&sa, &errorCode); + break; + case UPROPS_SRC_INPC: + case UPROPS_SRC_INSC: + case UPROPS_SRC_VO: + uprops_addPropertyStarts((UPropertySource)src, &sa, &errorCode); + break; + case UPROPS_SRC_EMOJI: { + const icu::EmojiProps *ep = icu::EmojiProps::getSingleton(errorCode); + if (U_SUCCESS(errorCode)) { + ep->addPropertyStarts(&sa, errorCode); + } + break; + } + default: + errorCode = U_INTERNAL_PROGRAM_ERROR; + break; + } + + if (U_FAILURE(errorCode)) { + return; + } + if (incl->isBogus()) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + // Compact for caching. + incl->compact(); + gInclusions[src].fSet = incl.orphan(); + ucln_common_registerCleanup(UCLN_COMMON_CHARACTERPROPERTIES, characterproperties_cleanup); +} + +const UnicodeSet *getInclusionsForSource(UPropertySource src, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return nullptr; } + if (src < 0 || UPROPS_SRC_COUNT <= src) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + Inclusion &i = gInclusions[src]; + umtx_initOnce(i.fInitOnce, &initInclusion, src, errorCode); + return i.fSet; +} + +void U_CALLCONV initIntPropInclusion(UProperty prop, UErrorCode &errorCode) { + // This function is invoked only via umtx_initOnce(). + U_ASSERT(UCHAR_INT_START <= prop && prop < UCHAR_INT_LIMIT); + int32_t inclIndex = UPROPS_SRC_COUNT + (prop - UCHAR_INT_START); + U_ASSERT(gInclusions[inclIndex].fSet == nullptr); + UPropertySource src = uprops_getSource(prop); + const UnicodeSet *incl = getInclusionsForSource(src, errorCode); + if (U_FAILURE(errorCode)) { + return; + } + + LocalPointer intPropIncl(new UnicodeSet(0, 0)); + if (intPropIncl.isNull()) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + int32_t numRanges = incl->getRangeCount(); + int32_t prevValue = 0; + for (int32_t i = 0; i < numRanges; ++i) { + UChar32 rangeEnd = incl->getRangeEnd(i); + for (UChar32 c = incl->getRangeStart(i); c <= rangeEnd; ++c) { + // TODO: Get a UCharacterProperty.IntProperty to avoid the property dispatch. + int32_t value = u_getIntPropertyValue(c, prop); + if (value != prevValue) { + intPropIncl->add(c); + prevValue = value; + } + } + } + + if (intPropIncl->isBogus()) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + // Compact for caching. + intPropIncl->compact(); + gInclusions[inclIndex].fSet = intPropIncl.orphan(); + ucln_common_registerCleanup(UCLN_COMMON_CHARACTERPROPERTIES, characterproperties_cleanup); +} + +} // namespace + +U_NAMESPACE_BEGIN + +const UnicodeSet *CharacterProperties::getInclusionsForProperty( + UProperty prop, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return nullptr; } + if (UCHAR_INT_START <= prop && prop < UCHAR_INT_LIMIT) { + int32_t inclIndex = UPROPS_SRC_COUNT + (prop - UCHAR_INT_START); + Inclusion &i = gInclusions[inclIndex]; + umtx_initOnce(i.fInitOnce, &initIntPropInclusion, prop, errorCode); + return i.fSet; + } else { + UPropertySource src = uprops_getSource(prop); + return getInclusionsForSource(src, errorCode); + } +} + +U_NAMESPACE_END + +namespace { + +UnicodeSet *makeSet(UProperty property, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return nullptr; } + LocalPointer set(new UnicodeSet()); + if (set.isNull()) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + if (UCHAR_BASIC_EMOJI <= property && property <= UCHAR_RGI_EMOJI) { + // property of strings + const icu::EmojiProps *ep = icu::EmojiProps::getSingleton(errorCode); + if (U_FAILURE(errorCode)) { return nullptr; } + USetAdder sa = { + (USet *)set.getAlias(), + _set_add, + _set_addRange, + _set_addString, + nullptr, // don't need remove() + nullptr // don't need removeRange() + }; + ep->addStrings(&sa, property, errorCode); + if (property != UCHAR_BASIC_EMOJI && property != UCHAR_RGI_EMOJI) { + // property of _only_ strings + set->freeze(); + return set.orphan(); + } + } + + const UnicodeSet *inclusions = + icu::CharacterProperties::getInclusionsForProperty(property, errorCode); + if (U_FAILURE(errorCode)) { return nullptr; } + int32_t numRanges = inclusions->getRangeCount(); + UChar32 startHasProperty = -1; + + for (int32_t i = 0; i < numRanges; ++i) { + UChar32 rangeEnd = inclusions->getRangeEnd(i); + for (UChar32 c = inclusions->getRangeStart(i); c <= rangeEnd; ++c) { + // TODO: Get a UCharacterProperty.BinaryProperty to avoid the property dispatch. + if (u_hasBinaryProperty(c, property)) { + if (startHasProperty < 0) { + // Transition from false to true. + startHasProperty = c; + } + } else if (startHasProperty >= 0) { + // Transition from true to false. + set->add(startHasProperty, c - 1); + startHasProperty = -1; + } + } + } + if (startHasProperty >= 0) { + set->add(startHasProperty, 0x10FFFF); + } + set->freeze(); + return set.orphan(); +} + +UCPMap *makeMap(UProperty property, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return nullptr; } + uint32_t nullValue = property == UCHAR_SCRIPT ? USCRIPT_UNKNOWN : 0; + icu::LocalUMutableCPTriePointer mutableTrie( + umutablecptrie_open(nullValue, nullValue, &errorCode)); + const UnicodeSet *inclusions = + icu::CharacterProperties::getInclusionsForProperty(property, errorCode); + if (U_FAILURE(errorCode)) { return nullptr; } + int32_t numRanges = inclusions->getRangeCount(); + UChar32 start = 0; + uint32_t value = nullValue; + + for (int32_t i = 0; i < numRanges; ++i) { + UChar32 rangeEnd = inclusions->getRangeEnd(i); + for (UChar32 c = inclusions->getRangeStart(i); c <= rangeEnd; ++c) { + // TODO: Get a UCharacterProperty.IntProperty to avoid the property dispatch. + uint32_t nextValue = u_getIntPropertyValue(c, property); + if (value != nextValue) { + if (value != nullValue) { + umutablecptrie_setRange(mutableTrie.getAlias(), start, c - 1, value, &errorCode); + } + start = c; + value = nextValue; + } + } + } + if (value != 0) { + umutablecptrie_setRange(mutableTrie.getAlias(), start, 0x10FFFF, value, &errorCode); + } + + UCPTrieType type; + if (property == UCHAR_BIDI_CLASS || property == UCHAR_GENERAL_CATEGORY) { + type = UCPTRIE_TYPE_FAST; + } else { + type = UCPTRIE_TYPE_SMALL; + } + UCPTrieValueWidth valueWidth; + // TODO: UCharacterProperty.IntProperty + int32_t max = u_getIntPropertyMaxValue(property); + if (max <= 0xff) { + valueWidth = UCPTRIE_VALUE_BITS_8; + } else if (max <= 0xffff) { + valueWidth = UCPTRIE_VALUE_BITS_16; + } else { + valueWidth = UCPTRIE_VALUE_BITS_32; + } + return reinterpret_cast( + umutablecptrie_buildImmutable(mutableTrie.getAlias(), type, valueWidth, &errorCode)); +} + +} // namespace + +U_NAMESPACE_BEGIN + +const UnicodeSet *CharacterProperties::getBinaryPropertySet(UProperty property, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return nullptr; } + if (property < 0 || UCHAR_BINARY_LIMIT <= property) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + Mutex m(&cpMutex); + UnicodeSet *set = sets[property]; + if (set == nullptr) { + sets[property] = set = makeSet(property, errorCode); + } + return set; +} + +U_NAMESPACE_END + +U_NAMESPACE_USE + +U_CAPI const USet * U_EXPORT2 +u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode) { + const UnicodeSet *set = CharacterProperties::getBinaryPropertySet(property, *pErrorCode); + return U_SUCCESS(*pErrorCode) ? set->toUSet() : nullptr; +} + +U_CAPI const UCPMap * U_EXPORT2 +u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { return nullptr; } + if (property < UCHAR_INT_START || UCHAR_INT_LIMIT <= property) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + Mutex m(&cpMutex); + UCPMap *map = maps[property - UCHAR_INT_START]; + if (map == nullptr) { + maps[property - UCHAR_INT_START] = map = makeMap(property, *pErrorCode); + } + return map; +} diff --git a/intl/icu/source/common/chariter.cpp b/intl/icu/source/common/chariter.cpp new file mode 100644 index 0000000000..16f3b1e14d --- /dev/null +++ b/intl/icu/source/common/chariter.cpp @@ -0,0 +1,100 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1999-2011, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ + +#include "unicode/chariter.h" + +U_NAMESPACE_BEGIN + +ForwardCharacterIterator::~ForwardCharacterIterator() {} +ForwardCharacterIterator::ForwardCharacterIterator() +: UObject() +{} +ForwardCharacterIterator::ForwardCharacterIterator(const ForwardCharacterIterator &other) +: UObject(other) +{} + + +CharacterIterator::CharacterIterator() +: textLength(0), pos(0), begin(0), end(0) { +} + +CharacterIterator::CharacterIterator(int32_t length) +: textLength(length), pos(0), begin(0), end(length) { + if(textLength < 0) { + textLength = end = 0; + } +} + +CharacterIterator::CharacterIterator(int32_t length, int32_t position) +: textLength(length), pos(position), begin(0), end(length) { + if(textLength < 0) { + textLength = end = 0; + } + if(pos < 0) { + pos = 0; + } else if(pos > end) { + pos = end; + } +} + +CharacterIterator::CharacterIterator(int32_t length, int32_t textBegin, int32_t textEnd, int32_t position) +: textLength(length), pos(position), begin(textBegin), end(textEnd) { + if(textLength < 0) { + textLength = 0; + } + if(begin < 0) { + begin = 0; + } else if(begin > textLength) { + begin = textLength; + } + if(end < begin) { + end = begin; + } else if(end > textLength) { + end = textLength; + } + if(pos < begin) { + pos = begin; + } else if(pos > end) { + pos = end; + } +} + +CharacterIterator::~CharacterIterator() {} + +CharacterIterator::CharacterIterator(const CharacterIterator &that) : +ForwardCharacterIterator(that), +textLength(that.textLength), pos(that.pos), begin(that.begin), end(that.end) +{ +} + +CharacterIterator & +CharacterIterator::operator=(const CharacterIterator &that) { + ForwardCharacterIterator::operator=(that); + textLength = that.textLength; + pos = that.pos; + begin = that.begin; + end = that.end; + return *this; +} + +// implementing first[32]PostInc() directly in a subclass should be faster +// but these implementations make subclassing a little easier +char16_t +CharacterIterator::firstPostInc() { + setToStart(); + return nextPostInc(); +} + +UChar32 +CharacterIterator::first32PostInc() { + setToStart(); + return next32PostInc(); +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/charstr.cpp b/intl/icu/source/common/charstr.cpp new file mode 100644 index 0000000000..f76cc8a4dc --- /dev/null +++ b/intl/icu/source/common/charstr.cpp @@ -0,0 +1,273 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2015, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: charstr.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010may19 +* created by: Markus W. Scherer +*/ + +#include + +#include "unicode/utypes.h" +#include "unicode/putil.h" +#include "charstr.h" +#include "cmemory.h" +#include "cstring.h" +#include "uinvchar.h" +#include "ustr_imp.h" + +U_NAMESPACE_BEGIN + +CharString::CharString(CharString&& src) noexcept + : buffer(std::move(src.buffer)), len(src.len) { + src.len = 0; // not strictly necessary because we make no guarantees on the source string +} + +CharString& CharString::operator=(CharString&& src) noexcept { + buffer = std::move(src.buffer); + len = src.len; + src.len = 0; // not strictly necessary because we make no guarantees on the source string + return *this; +} + +char *CharString::cloneData(UErrorCode &errorCode) const { + if (U_FAILURE(errorCode)) { return nullptr; } + char *p = static_cast(uprv_malloc(len + 1)); + if (p == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memcpy(p, buffer.getAlias(), len + 1); + return p; +} + +int32_t CharString::extract(char *dest, int32_t capacity, UErrorCode &errorCode) const { + if (U_FAILURE(errorCode)) { return len; } + if (capacity < 0 || (capacity > 0 && dest == nullptr)) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return len; + } + const char *src = buffer.getAlias(); + if (0 < len && len <= capacity && src != dest) { + uprv_memcpy(dest, src, len); + } + return u_terminateChars(dest, capacity, len, &errorCode); +} + +CharString &CharString::copyFrom(const CharString &s, UErrorCode &errorCode) { + if(U_SUCCESS(errorCode) && this!=&s && ensureCapacity(s.len+1, 0, errorCode)) { + len=s.len; + uprv_memcpy(buffer.getAlias(), s.buffer.getAlias(), len+1); + } + return *this; +} + +int32_t CharString::lastIndexOf(char c) const { + for(int32_t i=len; i>0;) { + if(buffer[--i]==c) { + return i; + } + } + return -1; +} + +bool CharString::contains(StringPiece s) const { + if (s.empty()) { return false; } + const char *p = buffer.getAlias(); + int32_t lastStart = len - s.length(); + for (int32_t i = 0; i <= lastStart; ++i) { + if (uprv_memcmp(p + i, s.data(), s.length()) == 0) { + return true; + } + } + return false; +} + +CharString &CharString::truncate(int32_t newLength) { + if(newLength<0) { + newLength=0; + } + if(newLength(uprv_strlen(s)); + } + if(sLength>0) { + if(s==(buffer.getAlias()+len)) { + // The caller wrote into the getAppendBuffer(). + if(sLength>=(buffer.getCapacity()-len)) { + // The caller wrote too much. + errorCode=U_INTERNAL_PROGRAM_ERROR; + } else { + buffer[len+=sLength]=0; + } + } else if(buffer.getAlias()<=s && s<(buffer.getAlias()+len) && + sLength>=(buffer.getCapacity()-len) + ) { + // (Part of) this string is appended to itself which requires reallocation, + // so we have to make a copy of the substring and append that. + return append(CharString(s, sLength, errorCode), errorCode); + } else if(ensureCapacity(len+sLength+1, 0, errorCode)) { + uprv_memcpy(buffer.getAlias()+len, s, sLength); + buffer[len+=sLength]=0; + } + } + return *this; +} + +CharString &CharString::appendNumber(int32_t number, UErrorCode &status) { + if (number < 0) { + this->append('-', status); + if (U_FAILURE(status)) { + return *this; + } + } + + if (number == 0) { + this->append('0', status); + return *this; + } + + int32_t numLen = 0; + while (number != 0) { + int32_t residue = number % 10; + number /= 10; + this->append(std::abs(residue) + '0', status); + numLen++; + if (U_FAILURE(status)) { + return *this; + } + } + + int32_t start = this->length() - numLen, end = this->length() - 1; + while(start < end) { + std::swap(this->data()[start++], this->data()[end--]); + } + + return *this; +} + +char *CharString::getAppendBuffer(int32_t minCapacity, + int32_t desiredCapacityHint, + int32_t &resultCapacity, + UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + resultCapacity=0; + return nullptr; + } + int32_t appendCapacity=buffer.getCapacity()-len-1; // -1 for NUL + if(appendCapacity>=minCapacity) { + resultCapacity=appendCapacity; + return buffer.getAlias()+len; + } + if(ensureCapacity(len+minCapacity+1, len+desiredCapacityHint+1, errorCode)) { + resultCapacity=buffer.getCapacity()-len-1; + return buffer.getAlias()+len; + } + resultCapacity=0; + return nullptr; +} + +CharString &CharString::appendInvariantChars(const UnicodeString &s, UErrorCode &errorCode) { + return appendInvariantChars(s.getBuffer(), s.length(), errorCode); +} + +CharString &CharString::appendInvariantChars(const char16_t* uchars, int32_t ucharsLen, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return *this; + } + if (!uprv_isInvariantUString(uchars, ucharsLen)) { + errorCode = U_INVARIANT_CONVERSION_ERROR; + return *this; + } + if(ensureCapacity(len+ucharsLen+1, 0, errorCode)) { + u_UCharsToChars(uchars, buffer.getAlias()+len, ucharsLen); + len += ucharsLen; + buffer[len] = 0; + } + return *this; +} + +UBool CharString::ensureCapacity(int32_t capacity, + int32_t desiredCapacityHint, + UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return false; + } + if(capacity>buffer.getCapacity()) { + if(desiredCapacityHint==0) { + desiredCapacityHint=capacity+buffer.getCapacity(); + } + if( (desiredCapacityHint<=capacity || buffer.resize(desiredCapacityHint, len+1)==nullptr) && + buffer.resize(capacity, len+1)==nullptr + ) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return false; + } + } + return true; +} + +CharString &CharString::appendPathPart(StringPiece s, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return *this; + } + if(s.length()==0) { + return *this; + } + char c; + if(len>0 && (c=buffer[len-1])!=U_FILE_SEP_CHAR && c!=U_FILE_ALT_SEP_CHAR) { + append(getDirSepChar(), errorCode); + } + append(s, errorCode); + return *this; +} + +CharString &CharString::ensureEndsWithFileSeparator(UErrorCode &errorCode) { + char c; + if(U_SUCCESS(errorCode) && len>0 && + (c=buffer[len-1])!=U_FILE_SEP_CHAR && c!=U_FILE_ALT_SEP_CHAR) { + append(getDirSepChar(), errorCode); + } + return *this; +} + +char CharString::getDirSepChar() const { + char dirSepChar = U_FILE_SEP_CHAR; +#if (U_FILE_SEP_CHAR != U_FILE_ALT_SEP_CHAR) + // We may need to return a different directory separator when building for Cygwin or MSYS2. + if(len>0 && !uprv_strchr(data(), U_FILE_SEP_CHAR) && uprv_strchr(data(), U_FILE_ALT_SEP_CHAR)) + dirSepChar = U_FILE_ALT_SEP_CHAR; +#endif + return dirSepChar; +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/charstr.h b/intl/icu/source/common/charstr.h new file mode 100644 index 0000000000..7749a804b9 --- /dev/null +++ b/intl/icu/source/common/charstr.h @@ -0,0 +1,193 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2001-2015, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* Date Name Description +* 11/19/2001 aliu Creation. +* 05/19/2010 markus Rewritten from scratch +********************************************************************** +*/ + +#ifndef CHARSTRING_H +#define CHARSTRING_H + +#include "unicode/utypes.h" +#include "unicode/unistr.h" +#include "unicode/uobject.h" +#include "cmemory.h" + +U_NAMESPACE_BEGIN + +// Windows needs us to DLL-export the MaybeStackArray template specialization, +// but MacOS X cannot handle it. Same as in digitlst.h. +#if !U_PLATFORM_IS_DARWIN_BASED +template class U_COMMON_API MaybeStackArray; +#endif + +/** + * ICU-internal char * string class. + * This class does not assume or enforce any particular character encoding. + * Raw bytes can be stored. The string object owns its characters. + * A terminating NUL is stored, but the class does not prevent embedded NUL characters. + * + * This class wants to be convenient but is also deliberately minimalist. + * Please do not add methods if they only add minor convenience. + * For example: + * cs.data()[5]='a'; // no need for setCharAt(5, 'a') + */ +class U_COMMON_API CharString : public UMemory { +public: + CharString() : len(0) { buffer[0]=0; } + CharString(StringPiece s, UErrorCode &errorCode) : len(0) { + buffer[0]=0; + append(s, errorCode); + } + CharString(const CharString &s, UErrorCode &errorCode) : len(0) { + buffer[0]=0; + append(s, errorCode); + } + CharString(const char *s, int32_t sLength, UErrorCode &errorCode) : len(0) { + buffer[0]=0; + append(s, sLength, errorCode); + } + ~CharString() {} + + /** + * Move constructor; might leave src in an undefined state. + * This string will have the same contents and state that the source string had. + */ + CharString(CharString &&src) noexcept; + /** + * Move assignment operator; might leave src in an undefined state. + * This string will have the same contents and state that the source string had. + * The behavior is undefined if *this and src are the same object. + */ + CharString &operator=(CharString &&src) noexcept; + + /** + * Replaces this string's contents with the other string's contents. + * CharString does not support the standard copy constructor nor + * the assignment operator, to make copies explicit and to + * use a UErrorCode where memory allocations might be needed. + */ + CharString ©From(const CharString &other, UErrorCode &errorCode); + + UBool isEmpty() const { return len==0; } + int32_t length() const { return len; } + char operator[](int32_t index) const { return buffer[index]; } + StringPiece toStringPiece() const { return StringPiece(buffer.getAlias(), len); } + + const char *data() const { return buffer.getAlias(); } + char *data() { return buffer.getAlias(); } + /** + * Allocates length()+1 chars and copies the NUL-terminated data(). + * The caller must uprv_free() the result. + */ + char *cloneData(UErrorCode &errorCode) const; + /** + * Copies the contents of the string into dest. + * Checks if there is enough space in dest, extracts the entire string if possible, + * and NUL-terminates dest if possible. + * + * If the string fits into dest but cannot be NUL-terminated (length()==capacity), + * then the error code is set to U_STRING_NOT_TERMINATED_WARNING. + * If the string itself does not fit into dest (length()>capacity), + * then the error code is set to U_BUFFER_OVERFLOW_ERROR. + * + * @param dest Destination string buffer. + * @param capacity Size of the dest buffer (number of chars). + * @param errorCode ICU error code. + * @return length() + */ + int32_t extract(char *dest, int32_t capacity, UErrorCode &errorCode) const; + + bool operator==(StringPiece other) const { + return len == other.length() && (len == 0 || uprv_memcmp(data(), other.data(), len) == 0); + } + bool operator!=(StringPiece other) const { + return !operator==(other); + } + + /** @return last index of c, or -1 if c is not in this string */ + int32_t lastIndexOf(char c) const; + + bool contains(StringPiece s) const; + + CharString &clear() { len=0; buffer[0]=0; return *this; } + CharString &truncate(int32_t newLength); + + CharString &append(char c, UErrorCode &errorCode); + CharString &append(StringPiece s, UErrorCode &errorCode) { + return append(s.data(), s.length(), errorCode); + } + CharString &append(const CharString &s, UErrorCode &errorCode) { + return append(s.data(), s.length(), errorCode); + } + CharString &append(const char *s, int32_t sLength, UErrorCode &status); + + CharString &appendNumber(int32_t number, UErrorCode &status); + + /** + * Returns a writable buffer for appending and writes the buffer's capacity to + * resultCapacity. Guarantees resultCapacity>=minCapacity if U_SUCCESS(). + * There will additionally be space for a terminating NUL right at resultCapacity. + * (This function is similar to ByteSink.GetAppendBuffer().) + * + * The returned buffer is only valid until the next write operation + * on this string. + * + * After writing at most resultCapacity bytes, call append() with the + * pointer returned from this function and the number of bytes written. + * + * @param minCapacity required minimum capacity of the returned buffer; + * must be non-negative + * @param desiredCapacityHint desired capacity of the returned buffer; + * must be non-negative + * @param resultCapacity will be set to the capacity of the returned buffer + * @param errorCode in/out error code + * @return a buffer with resultCapacity>=min_capacity + */ + char *getAppendBuffer(int32_t minCapacity, + int32_t desiredCapacityHint, + int32_t &resultCapacity, + UErrorCode &errorCode); + + CharString &appendInvariantChars(const UnicodeString &s, UErrorCode &errorCode); + CharString &appendInvariantChars(const char16_t* uchars, int32_t ucharsLen, UErrorCode& errorCode); + + /** + * Appends a filename/path part, e.g., a directory name. + * First appends a U_FILE_SEP_CHAR or U_FILE_ALT_SEP_CHAR if necessary. + * Does nothing if s is empty. + */ + CharString &appendPathPart(StringPiece s, UErrorCode &errorCode); + + /** + * Appends a U_FILE_SEP_CHAR or U_FILE_ALT_SEP_CHAR if this string is not empty + * and does not already end with a U_FILE_SEP_CHAR or U_FILE_ALT_SEP_CHAR. + */ + CharString &ensureEndsWithFileSeparator(UErrorCode &errorCode); + +private: + MaybeStackArray buffer; + int32_t len; + + UBool ensureCapacity(int32_t capacity, int32_t desiredCapacityHint, UErrorCode &errorCode); + + CharString(const CharString &other) = delete; // forbid copying of this class + CharString &operator=(const CharString &other) = delete; // forbid copying of this class + + /** + * Returns U_FILE_ALT_SEP_CHAR if found in string, and U_FILE_SEP_CHAR is not found. + * Otherwise returns U_FILE_SEP_CHAR. + */ + char getDirSepChar() const; +}; + +U_NAMESPACE_END + +#endif +//eof diff --git a/intl/icu/source/common/charstrmap.h b/intl/icu/source/common/charstrmap.h new file mode 100644 index 0000000000..64d5fd1254 --- /dev/null +++ b/intl/icu/source/common/charstrmap.h @@ -0,0 +1,55 @@ +// © 2020 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// charstrmap.h +// created: 2020sep01 Frank Yung-Fong Tang + +#ifndef __CHARSTRMAP_H__ +#define __CHARSTRMAP_H__ + +#include +#include "unicode/utypes.h" +#include "unicode/uobject.h" +#include "uhash.h" + +U_NAMESPACE_BEGIN + +/** + * Map of const char * keys & values. + * Stores pointers as is: Does not own/copy/adopt/release strings. + */ +class CharStringMap final : public UMemory { +public: + /** Constructs an unusable non-map. */ + CharStringMap() : map(nullptr) {} + CharStringMap(int32_t size, UErrorCode &errorCode) { + map = uhash_openSize(uhash_hashChars, uhash_compareChars, uhash_compareChars, + size, &errorCode); + } + CharStringMap(CharStringMap &&other) noexcept : map(other.map) { + other.map = nullptr; + } + CharStringMap(const CharStringMap &other) = delete; + ~CharStringMap() { + uhash_close(map); + } + + CharStringMap &operator=(CharStringMap &&other) noexcept { + map = other.map; + other.map = nullptr; + return *this; + } + CharStringMap &operator=(const CharStringMap &other) = delete; + + const char *get(const char *key) const { return static_cast(uhash_get(map, key)); } + void put(const char *key, const char *value, UErrorCode &errorCode) { + uhash_put(map, const_cast(key), const_cast(value), &errorCode); + } + +private: + UHashtable *map; +}; + +U_NAMESPACE_END + +#endif // __CHARSTRMAP_H__ diff --git a/intl/icu/source/common/cmemory.cpp b/intl/icu/source/common/cmemory.cpp new file mode 100644 index 0000000000..e59d4b0efe --- /dev/null +++ b/intl/icu/source/common/cmemory.cpp @@ -0,0 +1,138 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2002-2015, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* File cmemory.c ICU Heap allocation. +* All ICU heap allocation, both for C and C++ new of ICU +* class types, comes through these functions. +* +* If you have a need to replace ICU allocation, this is the +* place to do it. +* +* Note that uprv_malloc(0) returns a non-nullptr pointer, +* and that a subsequent free of that pointer value is a NOP. +* +****************************************************************************** +*/ +#include "unicode/uclean.h" +#include "cmemory.h" +#include "putilimp.h" +#include "uassert.h" +#include + +/* uprv_malloc(0) returns a pointer to this read-only data. */ +static const int32_t zeroMem[] = {0, 0, 0, 0, 0, 0}; + +/* Function Pointers for user-supplied heap functions */ +static const void *pContext; +static UMemAllocFn *pAlloc; +static UMemReallocFn *pRealloc; +static UMemFreeFn *pFree; + +#if U_DEBUG && defined(UPRV_MALLOC_COUNT) +#include +static int n=0; +static long b=0; +#endif + +U_CAPI void * U_EXPORT2 +uprv_malloc(size_t s) { +#if U_DEBUG && defined(UPRV_MALLOC_COUNT) +#if 1 + putchar('>'); + fflush(stdout); +#else + fprintf(stderr,"MALLOC\t#%d\t%ul bytes\t%ul total\n", ++n,s,(b+=s)); fflush(stderr); +#endif +#endif + if (s > 0) { + if (pAlloc) { + return (*pAlloc)(pContext, s); + } else { + return uprv_default_malloc(s); + } + } else { + return (void *)zeroMem; + } +} + +U_CAPI void * U_EXPORT2 +uprv_realloc(void * buffer, size_t size) { +#if U_DEBUG && defined(UPRV_MALLOC_COUNT) + putchar('~'); + fflush(stdout); +#endif + if (buffer == zeroMem) { + return uprv_malloc(size); + } else if (size == 0) { + if (pFree) { + (*pFree)(pContext, buffer); + } else { + uprv_default_free(buffer); + } + return (void *)zeroMem; + } else { + if (pRealloc) { + return (*pRealloc)(pContext, buffer, size); + } else { + return uprv_default_realloc(buffer, size); + } + } +} + +U_CAPI void U_EXPORT2 +uprv_free(void *buffer) { +#if U_DEBUG && defined(UPRV_MALLOC_COUNT) + putchar('<'); + fflush(stdout); +#endif + if (buffer != zeroMem) { + if (pFree) { + (*pFree)(pContext, buffer); + } else { + uprv_default_free(buffer); + } + } +} + +U_CAPI void * U_EXPORT2 +uprv_calloc(size_t num, size_t size) { + void *mem = nullptr; + size *= num; + mem = uprv_malloc(size); + if (mem) { + uprv_memset(mem, 0, size); + } + return mem; +} + +U_CAPI void U_EXPORT2 +u_setMemoryFunctions(const void *context, UMemAllocFn *a, UMemReallocFn *r, UMemFreeFn *f, UErrorCode *status) +{ + if (U_FAILURE(*status)) { + return; + } + if (a==nullptr || r==nullptr || f==nullptr) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + pContext = context; + pAlloc = a; + pRealloc = r; + pFree = f; +} + + +U_CFUNC UBool cmemory_cleanup() { + pContext = nullptr; + pAlloc = nullptr; + pRealloc = nullptr; + pFree = nullptr; + return true; +} diff --git a/intl/icu/source/common/cmemory.h b/intl/icu/source/common/cmemory.h new file mode 100644 index 0000000000..3705c2dfd0 --- /dev/null +++ b/intl/icu/source/common/cmemory.h @@ -0,0 +1,900 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1997-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* File CMEMORY.H +* +* Contains stdlib.h/string.h memory functions +* +* @author Bertrand A. Damiba +* +* Modification History: +* +* Date Name Description +* 6/20/98 Bertrand Created. +* 05/03/99 stephen Changed from functions to macros. +* +****************************************************************************** +*/ + +#ifndef CMEMORY_H +#define CMEMORY_H + +#include "unicode/utypes.h" + +#include +#include +#include "unicode/localpointer.h" +#include "uassert.h" + +#if U_DEBUG && defined(UPRV_MALLOC_COUNT) +#include +#endif + +// uprv_memcpy and uprv_memmove +#if defined(__clang__) +#define uprv_memcpy(dst, src, size) UPRV_BLOCK_MACRO_BEGIN { \ + /* Suppress warnings about addresses that will never be NULL */ \ + _Pragma("clang diagnostic push") \ + _Pragma("clang diagnostic ignored \"-Waddress\"") \ + U_ASSERT(dst != NULL); \ + U_ASSERT(src != NULL); \ + _Pragma("clang diagnostic pop") \ + U_STANDARD_CPP_NAMESPACE memcpy(dst, src, size); \ +} UPRV_BLOCK_MACRO_END +#define uprv_memmove(dst, src, size) UPRV_BLOCK_MACRO_BEGIN { \ + /* Suppress warnings about addresses that will never be NULL */ \ + _Pragma("clang diagnostic push") \ + _Pragma("clang diagnostic ignored \"-Waddress\"") \ + U_ASSERT(dst != NULL); \ + U_ASSERT(src != NULL); \ + _Pragma("clang diagnostic pop") \ + U_STANDARD_CPP_NAMESPACE memmove(dst, src, size); \ +} UPRV_BLOCK_MACRO_END +#elif defined(__GNUC__) +#define uprv_memcpy(dst, src, size) UPRV_BLOCK_MACRO_BEGIN { \ + /* Suppress warnings about addresses that will never be NULL */ \ + _Pragma("GCC diagnostic push") \ + _Pragma("GCC diagnostic ignored \"-Waddress\"") \ + U_ASSERT(dst != NULL); \ + U_ASSERT(src != NULL); \ + _Pragma("GCC diagnostic pop") \ + U_STANDARD_CPP_NAMESPACE memcpy(dst, src, size); \ +} UPRV_BLOCK_MACRO_END +#define uprv_memmove(dst, src, size) UPRV_BLOCK_MACRO_BEGIN { \ + /* Suppress warnings about addresses that will never be NULL */ \ + _Pragma("GCC diagnostic push") \ + _Pragma("GCC diagnostic ignored \"-Waddress\"") \ + U_ASSERT(dst != NULL); \ + U_ASSERT(src != NULL); \ + _Pragma("GCC diagnostic pop") \ + U_STANDARD_CPP_NAMESPACE memmove(dst, src, size); \ +} UPRV_BLOCK_MACRO_END +#else +#define uprv_memcpy(dst, src, size) UPRV_BLOCK_MACRO_BEGIN { \ + U_ASSERT(dst != NULL); \ + U_ASSERT(src != NULL); \ + U_STANDARD_CPP_NAMESPACE memcpy(dst, src, size); \ +} UPRV_BLOCK_MACRO_END +#define uprv_memmove(dst, src, size) UPRV_BLOCK_MACRO_BEGIN { \ + U_ASSERT(dst != NULL); \ + U_ASSERT(src != NULL); \ + U_STANDARD_CPP_NAMESPACE memmove(dst, src, size); \ +} UPRV_BLOCK_MACRO_END +#endif + +/** + * \def UPRV_LENGTHOF + * Convenience macro to determine the length of a fixed array at compile-time. + * @param array A fixed length array + * @return The length of the array, in elements + * @internal + */ +#define UPRV_LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0])) +#define uprv_memset(buffer, mark, size) U_STANDARD_CPP_NAMESPACE memset(buffer, mark, size) +#define uprv_memcmp(buffer1, buffer2, size) U_STANDARD_CPP_NAMESPACE memcmp(buffer1, buffer2,size) +#define uprv_memchr(ptr, value, num) U_STANDARD_CPP_NAMESPACE memchr(ptr, value, num) + +U_CAPI void * U_EXPORT2 +uprv_malloc(size_t s) U_MALLOC_ATTR U_ALLOC_SIZE_ATTR(1); + +U_CAPI void * U_EXPORT2 +uprv_realloc(void *mem, size_t size) U_ALLOC_SIZE_ATTR(2); + +U_CAPI void U_EXPORT2 +uprv_free(void *mem); + +U_CAPI void * U_EXPORT2 +uprv_calloc(size_t num, size_t size) U_MALLOC_ATTR U_ALLOC_SIZE_ATTR2(1,2); + +/** + * Get the least significant bits of a pointer (a memory address). + * For example, with a mask of 3, the macro gets the 2 least significant bits, + * which will be 0 if the pointer is 32-bit (4-byte) aligned. + * + * uintptr_t is the most appropriate integer type to cast to. + */ +#define U_POINTER_MASK_LSB(ptr, mask) ((uintptr_t)(ptr) & (mask)) + +/** + * Create & return an instance of "type" in statically allocated storage. + * e.g. + * static std::mutex *myMutex = STATIC_NEW(std::mutex); + * To destroy an object created in this way, invoke the destructor explicitly, e.g. + * myMutex->~mutex(); + * DO NOT use delete. + * DO NOT use with class UMutex, which has specific support for static instances. + * + * STATIC_NEW is intended for use when + * - We want a static (or global) object. + * - We don't want it to ever be destructed, or to explicitly control destruction, + * to avoid use-after-destruction problems. + * - We want to avoid an ordinary heap allocated object, + * to avoid the possibility of memory allocation failures, and + * to avoid memory leak reports, from valgrind, for example. + * This is defined as a macro rather than a template function because each invocation + * must define distinct static storage for the object being returned. + */ +#define STATIC_NEW(type) [] () { \ + alignas(type) static char storage[sizeof(type)]; \ + return new(storage) type();} () + +/** + * Heap clean up function, called from u_cleanup() + * Clears any user heap functions from u_setMemoryFunctions() + * Does NOT deallocate any remaining allocated memory. + */ +U_CFUNC UBool +cmemory_cleanup(void); + +/** + * A function called by uhash_remove, + * uhash_close, or uhash_put to delete + * an existing key or value. + * @param obj A key or value stored in a hashtable + * @see uprv_deleteUObject + */ +typedef void U_CALLCONV UObjectDeleter(void* obj); + +/** + * Deleter for UObject instances. + * Works for all subclasses of UObject because it has a virtual destructor. + */ +U_CAPI void U_EXPORT2 +uprv_deleteUObject(void *obj); + +#ifdef __cplusplus + +#include +#include "unicode/uobject.h" + +U_NAMESPACE_BEGIN + +/** + * "Smart pointer" class, deletes memory via uprv_free(). + * For most methods see the LocalPointerBase base class. + * Adds operator[] for array item access. + * + * @see LocalPointerBase + */ +template +class LocalMemory : public LocalPointerBase { +public: + using LocalPointerBase::operator*; + using LocalPointerBase::operator->; + /** + * Constructor takes ownership. + * @param p simple pointer to an array of T items that is adopted + */ + explicit LocalMemory(T *p=nullptr) : LocalPointerBase(p) {} + /** + * Move constructor, leaves src with isNull(). + * @param src source smart pointer + */ + LocalMemory(LocalMemory &&src) noexcept : LocalPointerBase(src.ptr) { + src.ptr=nullptr; + } + /** + * Destructor deletes the memory it owns. + */ + ~LocalMemory() { + uprv_free(LocalPointerBase::ptr); + } + /** + * Move assignment operator, leaves src with isNull(). + * The behavior is undefined if *this and src are the same object. + * @param src source smart pointer + * @return *this + */ + LocalMemory &operator=(LocalMemory &&src) noexcept { + uprv_free(LocalPointerBase::ptr); + LocalPointerBase::ptr=src.ptr; + src.ptr=nullptr; + return *this; + } + /** + * Swap pointers. + * @param other other smart pointer + */ + void swap(LocalMemory &other) noexcept { + T *temp=LocalPointerBase::ptr; + LocalPointerBase::ptr=other.ptr; + other.ptr=temp; + } + /** + * Non-member LocalMemory swap function. + * @param p1 will get p2's pointer + * @param p2 will get p1's pointer + */ + friend inline void swap(LocalMemory &p1, LocalMemory &p2) noexcept { + p1.swap(p2); + } + /** + * Deletes the array it owns, + * and adopts (takes ownership of) the one passed in. + * @param p simple pointer to an array of T items that is adopted + */ + void adoptInstead(T *p) { + uprv_free(LocalPointerBase::ptr); + LocalPointerBase::ptr=p; + } + /** + * Deletes the array it owns, allocates a new one and reset its bytes to 0. + * Returns the new array pointer. + * If the allocation fails, then the current array is unchanged and + * this method returns nullptr. + * @param newCapacity must be >0 + * @return the allocated array pointer, or nullptr if the allocation failed + */ + inline T *allocateInsteadAndReset(int32_t newCapacity=1); + /** + * Deletes the array it owns and allocates a new one, copying length T items. + * Returns the new array pointer. + * If the allocation fails, then the current array is unchanged and + * this method returns nullptr. + * @param newCapacity must be >0 + * @param length number of T items to be copied from the old array to the new one; + * must be no more than the capacity of the old array, + * which the caller must track because the LocalMemory does not track it + * @return the allocated array pointer, or nullptr if the allocation failed + */ + inline T *allocateInsteadAndCopy(int32_t newCapacity=1, int32_t length=0); + /** + * Array item access (writable). + * No index bounds check. + * @param i array index + * @return reference to the array item + */ + T &operator[](ptrdiff_t i) const { return LocalPointerBase::ptr[i]; } +}; + +template +inline T *LocalMemory::allocateInsteadAndReset(int32_t newCapacity) { + if(newCapacity>0) { + T *p=(T *)uprv_malloc(newCapacity*sizeof(T)); + if(p!=nullptr) { + uprv_memset(p, 0, newCapacity*sizeof(T)); + uprv_free(LocalPointerBase::ptr); + LocalPointerBase::ptr=p; + } + return p; + } else { + return nullptr; + } +} + + +template +inline T *LocalMemory::allocateInsteadAndCopy(int32_t newCapacity, int32_t length) { + if(newCapacity>0) { + T *p=(T *)uprv_malloc(newCapacity*sizeof(T)); + if(p!=nullptr) { + if(length>0) { + if(length>newCapacity) { + length=newCapacity; + } + uprv_memcpy(p, LocalPointerBase::ptr, (size_t)length*sizeof(T)); + } + uprv_free(LocalPointerBase::ptr); + LocalPointerBase::ptr=p; + } + return p; + } else { + return nullptr; + } +} + +/** + * Simple array/buffer management class using uprv_malloc() and uprv_free(). + * Provides an internal array with fixed capacity. Can alias another array + * or allocate one. + * + * The array address is properly aligned for type T. It might not be properly + * aligned for types larger than T (or larger than the largest subtype of T). + * + * Unlike LocalMemory and LocalArray, this class never adopts + * (takes ownership of) another array. + * + * WARNING: MaybeStackArray only works with primitive (plain-old data) types. + * It does NOT know how to call a destructor! If you work with classes with + * destructors, consider: + * + * - LocalArray in localpointer.h if you know the length ahead of time + * - MaybeStackVector if you know the length at runtime + */ +template +class MaybeStackArray { +public: + // No heap allocation. Use only on the stack. + static void* U_EXPORT2 operator new(size_t) noexcept = delete; + static void* U_EXPORT2 operator new[](size_t) noexcept = delete; +#if U_HAVE_PLACEMENT_NEW + static void* U_EXPORT2 operator new(size_t, void*) noexcept = delete; +#endif + + /** + * Default constructor initializes with internal T[stackCapacity] buffer. + */ + MaybeStackArray() : ptr(stackArray), capacity(stackCapacity), needToRelease(false) {} + /** + * Automatically allocates the heap array if the argument is larger than the stack capacity. + * Intended for use when an approximate capacity is known at compile time but the true + * capacity is not known until runtime. + */ + MaybeStackArray(int32_t newCapacity, UErrorCode status) : MaybeStackArray() { + if (U_FAILURE(status)) { + return; + } + if (capacity < newCapacity) { + if (resize(newCapacity) == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + } + } + } + /** + * Destructor deletes the array (if owned). + */ + ~MaybeStackArray() { releaseArray(); } + /** + * Move constructor: transfers ownership or copies the stack array. + */ + MaybeStackArray(MaybeStackArray &&src) noexcept; + /** + * Move assignment: transfers ownership or copies the stack array. + */ + MaybeStackArray &operator=(MaybeStackArray &&src) noexcept; + /** + * Returns the array capacity (number of T items). + * @return array capacity + */ + int32_t getCapacity() const { return capacity; } + /** + * Access without ownership change. + * @return the array pointer + */ + T *getAlias() const { return ptr; } + /** + * Returns the array limit. Simple convenience method. + * @return getAlias()+getCapacity() + */ + T *getArrayLimit() const { return getAlias()+capacity; } + // No "operator T *() const" because that can make + // expressions like mbs[index] ambiguous for some compilers. + /** + * Array item access (const). + * No index bounds check. + * @param i array index + * @return reference to the array item + */ + const T &operator[](ptrdiff_t i) const { return ptr[i]; } + /** + * Array item access (writable). + * No index bounds check. + * @param i array index + * @return reference to the array item + */ + T &operator[](ptrdiff_t i) { return ptr[i]; } + /** + * Deletes the array (if owned) and aliases another one, no transfer of ownership. + * If the arguments are illegal, then the current array is unchanged. + * @param otherArray must not be nullptr + * @param otherCapacity must be >0 + */ + void aliasInstead(T *otherArray, int32_t otherCapacity) { + if(otherArray!=nullptr && otherCapacity>0) { + releaseArray(); + ptr=otherArray; + capacity=otherCapacity; + needToRelease=false; + } + } + /** + * Deletes the array (if owned) and allocates a new one, copying length T items. + * Returns the new array pointer. + * If the allocation fails, then the current array is unchanged and + * this method returns nullptr. + * @param newCapacity can be less than or greater than the current capacity; + * must be >0 + * @param length number of T items to be copied from the old array to the new one + * @return the allocated array pointer, or nullptr if the allocation failed + */ + inline T *resize(int32_t newCapacity, int32_t length=0); + /** + * Gives up ownership of the array if owned, or else clones it, + * copying length T items; resets itself to the internal stack array. + * Returns nullptr if the allocation failed. + * @param length number of T items to copy when cloning, + * and capacity of the clone when cloning + * @param resultCapacity will be set to the returned array's capacity (output-only) + * @return the array pointer; + * caller becomes responsible for deleting the array + */ + inline T *orphanOrClone(int32_t length, int32_t &resultCapacity); + +protected: + // Resizes the array to the size of src, then copies the contents of src. + void copyFrom(const MaybeStackArray &src, UErrorCode &status) { + if (U_FAILURE(status)) { + return; + } + if (this->resize(src.capacity, 0) == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + uprv_memcpy(this->ptr, src.ptr, (size_t)capacity * sizeof(T)); + } + +private: + T *ptr; + int32_t capacity; + UBool needToRelease; + T stackArray[stackCapacity]; + void releaseArray() { + if(needToRelease) { + uprv_free(ptr); + } + } + void resetToStackArray() { + ptr=stackArray; + capacity=stackCapacity; + needToRelease=false; + } + /* No comparison operators with other MaybeStackArray's. */ + bool operator==(const MaybeStackArray & /*other*/) = delete; + bool operator!=(const MaybeStackArray & /*other*/) = delete; + /* No ownership transfer: No copy constructor, no assignment operator. */ + MaybeStackArray(const MaybeStackArray & /*other*/) = delete; + void operator=(const MaybeStackArray & /*other*/) = delete; +}; + +template +icu::MaybeStackArray::MaybeStackArray( + MaybeStackArray && src) noexcept + : ptr(src.ptr), capacity(src.capacity), needToRelease(src.needToRelease) { + if (src.ptr == src.stackArray) { + ptr = stackArray; + uprv_memcpy(stackArray, src.stackArray, sizeof(T) * src.capacity); + } else { + src.resetToStackArray(); // take ownership away from src + } +} + +template +inline MaybeStackArray & +MaybeStackArray::operator=(MaybeStackArray && src) noexcept { + releaseArray(); // in case this instance had its own memory allocated + capacity = src.capacity; + needToRelease = src.needToRelease; + if (src.ptr == src.stackArray) { + ptr = stackArray; + uprv_memcpy(stackArray, src.stackArray, sizeof(T) * src.capacity); + } else { + ptr = src.ptr; + src.resetToStackArray(); // take ownership away from src + } + return *this; +} + +template +inline T *MaybeStackArray::resize(int32_t newCapacity, int32_t length) { + if(newCapacity>0) { +#if U_DEBUG && defined(UPRV_MALLOC_COUNT) + ::fprintf(::stderr, "MaybeStackArray (resize) alloc %d * %lu\n", newCapacity, sizeof(T)); +#endif + T *p=(T *)uprv_malloc(newCapacity*sizeof(T)); + if(p!=nullptr) { + if(length>0) { + if(length>capacity) { + length=capacity; + } + if(length>newCapacity) { + length=newCapacity; + } + uprv_memcpy(p, ptr, (size_t)length*sizeof(T)); + } + releaseArray(); + ptr=p; + capacity=newCapacity; + needToRelease=true; + } + return p; + } else { + return nullptr; + } +} + +template +inline T *MaybeStackArray::orphanOrClone(int32_t length, int32_t &resultCapacity) { + T *p; + if(needToRelease) { + p=ptr; + } else if(length<=0) { + return nullptr; + } else { + if(length>capacity) { + length=capacity; + } + p=(T *)uprv_malloc(length*sizeof(T)); +#if U_DEBUG && defined(UPRV_MALLOC_COUNT) + ::fprintf(::stderr,"MaybeStacArray (orphan) alloc %d * %lu\n", length,sizeof(T)); +#endif + if(p==nullptr) { + return nullptr; + } + uprv_memcpy(p, ptr, (size_t)length*sizeof(T)); + } + resultCapacity=length; + resetToStackArray(); + return p; +} + +/** + * Variant of MaybeStackArray that allocates a header struct and an array + * in one contiguous memory block, using uprv_malloc() and uprv_free(). + * Provides internal memory with fixed array capacity. Can alias another memory + * block or allocate one. + * The stackCapacity is the number of T items in the internal memory, + * not counting the H header. + * Unlike LocalMemory and LocalArray, this class never adopts + * (takes ownership of) another memory block. + */ +template +class MaybeStackHeaderAndArray { +public: + // No heap allocation. Use only on the stack. + static void* U_EXPORT2 operator new(size_t) noexcept = delete; + static void* U_EXPORT2 operator new[](size_t) noexcept = delete; +#if U_HAVE_PLACEMENT_NEW + static void* U_EXPORT2 operator new(size_t, void*) noexcept = delete; +#endif + + /** + * Default constructor initializes with internal H+T[stackCapacity] buffer. + */ + MaybeStackHeaderAndArray() : ptr(&stackHeader), capacity(stackCapacity), needToRelease(false) {} + /** + * Destructor deletes the memory (if owned). + */ + ~MaybeStackHeaderAndArray() { releaseMemory(); } + /** + * Returns the array capacity (number of T items). + * @return array capacity + */ + int32_t getCapacity() const { return capacity; } + /** + * Access without ownership change. + * @return the header pointer + */ + H *getAlias() const { return ptr; } + /** + * Returns the array start. + * @return array start, same address as getAlias()+1 + */ + T *getArrayStart() const { return reinterpret_cast(getAlias()+1); } + /** + * Returns the array limit. + * @return array limit + */ + T *getArrayLimit() const { return getArrayStart()+capacity; } + /** + * Access without ownership change. Same as getAlias(). + * A class instance can be used directly in expressions that take a T *. + * @return the header pointer + */ + operator H *() const { return ptr; } + /** + * Array item access (writable). + * No index bounds check. + * @param i array index + * @return reference to the array item + */ + T &operator[](ptrdiff_t i) { return getArrayStart()[i]; } + /** + * Deletes the memory block (if owned) and aliases another one, no transfer of ownership. + * If the arguments are illegal, then the current memory is unchanged. + * @param otherArray must not be nullptr + * @param otherCapacity must be >0 + */ + void aliasInstead(H *otherMemory, int32_t otherCapacity) { + if(otherMemory!=nullptr && otherCapacity>0) { + releaseMemory(); + ptr=otherMemory; + capacity=otherCapacity; + needToRelease=false; + } + } + /** + * Deletes the memory block (if owned) and allocates a new one, + * copying the header and length T array items. + * Returns the new header pointer. + * If the allocation fails, then the current memory is unchanged and + * this method returns nullptr. + * @param newCapacity can be less than or greater than the current capacity; + * must be >0 + * @param length number of T items to be copied from the old array to the new one + * @return the allocated pointer, or nullptr if the allocation failed + */ + inline H *resize(int32_t newCapacity, int32_t length=0); + /** + * Gives up ownership of the memory if owned, or else clones it, + * copying the header and length T array items; resets itself to the internal memory. + * Returns nullptr if the allocation failed. + * @param length number of T items to copy when cloning, + * and array capacity of the clone when cloning + * @param resultCapacity will be set to the returned array's capacity (output-only) + * @return the header pointer; + * caller becomes responsible for deleting the array + */ + inline H *orphanOrClone(int32_t length, int32_t &resultCapacity); +private: + H *ptr; + int32_t capacity; + UBool needToRelease; + // stackHeader must precede stackArray immediately. + H stackHeader; + T stackArray[stackCapacity]; + void releaseMemory() { + if(needToRelease) { + uprv_free(ptr); + } + } + /* No comparison operators with other MaybeStackHeaderAndArray's. */ + bool operator==(const MaybeStackHeaderAndArray & /*other*/) {return false;} + bool operator!=(const MaybeStackHeaderAndArray & /*other*/) {return true;} + /* No ownership transfer: No copy constructor, no assignment operator. */ + MaybeStackHeaderAndArray(const MaybeStackHeaderAndArray & /*other*/) {} + void operator=(const MaybeStackHeaderAndArray & /*other*/) {} +}; + +template +inline H *MaybeStackHeaderAndArray::resize(int32_t newCapacity, + int32_t length) { + if(newCapacity>=0) { +#if U_DEBUG && defined(UPRV_MALLOC_COUNT) + ::fprintf(::stderr,"MaybeStackHeaderAndArray alloc %d + %d * %ul\n", sizeof(H),newCapacity,sizeof(T)); +#endif + H *p=(H *)uprv_malloc(sizeof(H)+newCapacity*sizeof(T)); + if(p!=nullptr) { + if(length<0) { + length=0; + } else if(length>0) { + if(length>capacity) { + length=capacity; + } + if(length>newCapacity) { + length=newCapacity; + } + } + uprv_memcpy(p, ptr, sizeof(H)+(size_t)length*sizeof(T)); + releaseMemory(); + ptr=p; + capacity=newCapacity; + needToRelease=true; + } + return p; + } else { + return nullptr; + } +} + +template +inline H *MaybeStackHeaderAndArray::orphanOrClone(int32_t length, + int32_t &resultCapacity) { + H *p; + if(needToRelease) { + p=ptr; + } else { + if(length<0) { + length=0; + } else if(length>capacity) { + length=capacity; + } +#if U_DEBUG && defined(UPRV_MALLOC_COUNT) + ::fprintf(::stderr,"MaybeStackHeaderAndArray (orphan) alloc %ul + %d * %lu\n", sizeof(H),length,sizeof(T)); +#endif + p=(H *)uprv_malloc(sizeof(H)+length*sizeof(T)); + if(p==nullptr) { + return nullptr; + } + uprv_memcpy(p, ptr, sizeof(H)+(size_t)length*sizeof(T)); + } + resultCapacity=length; + ptr=&stackHeader; + capacity=stackCapacity; + needToRelease=false; + return p; +} + +/** + * A simple memory management class that creates new heap allocated objects (of + * any class that has a public constructor), keeps track of them and eventually + * deletes them all in its own destructor. + * + * A typical use-case would be code like this: + * + * MemoryPool pool; + * + * MyType* o1 = pool.create(); + * if (o1 != nullptr) { + * foo(o1); + * } + * + * MyType* o2 = pool.create(1, 2, 3); + * if (o2 != nullptr) { + * bar(o2); + * } + * + * // MemoryPool will take care of deleting the MyType objects. + * + * It doesn't do anything more than that, and is intentionally kept minimalist. + */ +template +class MemoryPool : public UMemory { +public: + MemoryPool() : fCount(0), fPool() {} + + ~MemoryPool() { + for (int32_t i = 0; i < fCount; ++i) { + delete fPool[i]; + } + } + + MemoryPool(const MemoryPool&) = delete; + MemoryPool& operator=(const MemoryPool&) = delete; + + MemoryPool(MemoryPool&& other) noexcept : fCount(other.fCount), + fPool(std::move(other.fPool)) { + other.fCount = 0; + } + + MemoryPool& operator=(MemoryPool&& other) noexcept { + // Since `this` may contain instances that need to be deleted, we can't + // just throw them away and replace them with `other`. The normal way of + // dealing with this in C++ is to swap `this` and `other`, rather than + // simply overwrite: the destruction of `other` can then take care of + // running MemoryPool::~MemoryPool() over the still-to-be-deallocated + // instances. + std::swap(fCount, other.fCount); + std::swap(fPool, other.fPool); + return *this; + } + + /** + * Creates a new object of typename T, by forwarding any and all arguments + * to the typename T constructor. + * + * @param args Arguments to be forwarded to the typename T constructor. + * @return A pointer to the newly created object, or nullptr on error. + */ + template + T* create(Args&&... args) { + int32_t capacity = fPool.getCapacity(); + if (fCount == capacity && + fPool.resize(capacity == stackCapacity ? 4 * capacity : 2 * capacity, + capacity) == nullptr) { + return nullptr; + } + return fPool[fCount++] = new T(std::forward(args)...); + } + + template + T* createAndCheckErrorCode(UErrorCode &status, Args &&... args) { + if (U_FAILURE(status)) { + return nullptr; + } + T *pointer = this->create(args...); + if (U_SUCCESS(status) && pointer == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + } + return pointer; + } + + /** + * @return Number of elements that have been allocated. + */ + int32_t count() const { + return fCount; + } + +protected: + int32_t fCount; + MaybeStackArray fPool; +}; + +/** + * An internal Vector-like implementation based on MemoryPool. + * + * Heap-allocates each element and stores pointers. + * + * To append an item to the vector, use emplaceBack. + * + * MaybeStackVector vector; + * MyType* element = vector.emplaceBack(); + * if (!element) { + * status = U_MEMORY_ALLOCATION_ERROR; + * } + * // do stuff with element + * + * To loop over the vector, use a for loop with indices: + * + * for (int32_t i = 0; i < vector.length(); i++) { + * MyType* element = vector[i]; + * } + */ +template +class MaybeStackVector : protected MemoryPool { +public: + template + T* emplaceBack(Args&&... args) { + return this->create(args...); + } + + template + T *emplaceBackAndCheckErrorCode(UErrorCode &status, Args &&... args) { + return this->createAndCheckErrorCode(status, args...); + } + + int32_t length() const { + return this->fCount; + } + + T** getAlias() { + return this->fPool.getAlias(); + } + + const T *const *getAlias() const { + return this->fPool.getAlias(); + } + + /** + * Array item access (read-only). + * No index bounds check. + * @param i array index + * @return reference to the array item + */ + const T* operator[](ptrdiff_t i) const { + return this->fPool[i]; + } + + /** + * Array item access (writable). + * No index bounds check. + * @param i array index + * @return reference to the array item + */ + T* operator[](ptrdiff_t i) { + return this->fPool[i]; + } +}; + + +U_NAMESPACE_END + +#endif /* __cplusplus */ +#endif /* CMEMORY_H */ diff --git a/intl/icu/source/common/common.rc b/intl/icu/source/common/common.rc new file mode 100644 index 0000000000..020abacc0d --- /dev/null +++ b/intl/icu/source/common/common.rc @@ -0,0 +1,110 @@ +// Do not edit with Microsoft Developer Studio Resource Editor. +// It will permanently substitute version numbers that are intended to be +// picked up by the pre-processor during each build. +// Copyright (C) 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// Copyright (c) 2001-2010 International Business Machines +// Corporation and others. All Rights Reserved. +// +#include "msvcres.h" + +#define APSTUDIO_READONLY_SYMBOLS +///////////////////////////////////////////////////////////////////////////// +// +// Generated from the TEXTINCLUDE 2 resource. +// +#include +///////////////////////////////////////////////////////////////////////////// +#undef APSTUDIO_READONLY_SYMBOLS + +///////////////////////////////////////////////////////////////////////////// +// + +LANGUAGE LANG_NEUTRAL, SUBLANG_NEUTRAL +#pragma code_page(1252) + +#ifdef APSTUDIO_INVOKED +///////////////////////////////////////////////////////////////////////////// +// +// TEXTINCLUDE +// + +1 TEXTINCLUDE +BEGIN + "msvcres.h\0" +END + +2 TEXTINCLUDE +BEGIN + "#include \0" +END + +3 TEXTINCLUDE +BEGIN + "\r\n" + "\0" +END + +#endif // APSTUDIO_INVOKED + + +///////////////////////////////////////////////////////////////////////////// +// +// Version +// +#define STR(s) #s +#define CommaVersionString(a, b, c, d) STR(a) ", " STR(b) ", " STR(c) ", " STR(d) "\0" + +VS_VERSION_INFO VERSIONINFO + FILEVERSION U_ICU_VERSION_MAJOR_NUM, U_ICU_VERSION_MINOR_NUM, U_ICU_VERSION_PATCHLEVEL_NUM, U_ICU_VERSION_BUILDLEVEL_NUM + PRODUCTVERSION U_ICU_VERSION_MAJOR_NUM, U_ICU_VERSION_MINOR_NUM, U_ICU_VERSION_PATCHLEVEL_NUM, U_ICU_VERSION_BUILDLEVEL_NUM + FILEFLAGSMASK 0x3fL +#ifdef _DEBUG + FILEFLAGS 0x1L +#else + FILEFLAGS 0x0L +#endif + FILEOS VOS__WINDOWS32 + FILETYPE VFT_DLL + FILESUBTYPE 0x0L +BEGIN + BLOCK "StringFileInfo" + BEGIN + BLOCK "00000000" + BEGIN + VALUE "Comments", ICU_WEBSITE "\0" + VALUE "CompanyName", ICU_COMPANY "\0" + VALUE "FileDescription", ICU_PRODUCT_PREFIX " Common DLL\0" + VALUE "FileVersion", CommaVersionString(U_ICU_VERSION_MAJOR_NUM, U_ICU_VERSION_MINOR_NUM, U_ICU_VERSION_PATCHLEVEL_NUM, U_ICU_VERSION_BUILDLEVEL_NUM) + VALUE "LegalCopyright", U_COPYRIGHT_STRING "\0" +#ifdef _DEBUG + VALUE "OriginalFilename", "icuuc" U_ICU_VERSION_SHORT "d.dll\0" +#else + VALUE "OriginalFilename", "icuuc" U_ICU_VERSION_SHORT ".dll\0" +#endif + VALUE "PrivateBuild", "\0" + VALUE "ProductName", ICU_PRODUCT "\0" + VALUE "ProductVersion", CommaVersionString(U_ICU_VERSION_MAJOR_NUM, U_ICU_VERSION_MINOR_NUM, U_ICU_VERSION_PATCHLEVEL_NUM, U_ICU_VERSION_BUILDLEVEL_NUM) + VALUE "SpecialBuild", "\0" + END + END + BLOCK "VarFileInfo" + BEGIN + VALUE "Translation", 0x000, 0000 + END +END + +///////////////////////////////////////////////////////////////////////////// + + + +#ifndef APSTUDIO_INVOKED +///////////////////////////////////////////////////////////////////////////// +// +// Generated from the TEXTINCLUDE 3 resource. +// + + +///////////////////////////////////////////////////////////////////////////// +#endif // not APSTUDIO_INVOKED + diff --git a/intl/icu/source/common/common.vcxproj b/intl/icu/source/common/common.vcxproj new file mode 100644 index 0000000000..2b4cc05357 --- /dev/null +++ b/intl/icu/source/common/common.vcxproj @@ -0,0 +1,404 @@ + + + + {73C0A65B-D1F2-4DE1-B3A6-15DAD2C23F3D} + + + DynamicLibrary + false + MultiByte + + + + + + + + + + + + <_ProjectFileVersion>10.0.30319.1 + .\$(Platform)\$(Configuration)\ + .\$(Platform)\$(Configuration)\ + + .\x86\$(Configuration)\ + .\x86\$(Configuration)\ + + true + false + + + + + $(OutDir)\icuuc.tlb + + + U_ATTRIBUTE_DEPRECATED=;U_COMMON_IMPLEMENTATION;U_PLATFORM_USES_ONLY_WIN32_API=1;%(PreprocessorDefinitions) + false + Level3 + Default + ProgramDatabase + $(OutDir)/icuuc.pch + $(OutDir)/ + $(OutDir)/ + $(OutDir)/icuuc.pdb + + + + icudt.lib;%(AdditionalDependencies) + .\..\..\$(IcuLibOutputDir);%(AdditionalLibraryDirectories) + + + + + + RBBI_DEBUG;%(PreprocessorDefinitions) + true + MultiThreadedDebugDLL + + + ..\..\$(IcuBinOutputDir)\icuuc$(IcuMajorVersion)d.dll + .\..\..\$(IcuLibOutputDir)\icuucd.pdb + ..\..\$(IcuLibOutputDir)\icuucd.lib + + + + + + MultiThreadedDLL + true + + + ..\..\$(IcuBinOutputDir)\icuuc$(IcuMajorVersion).dll + .\..\..\$(IcuLibOutputDir)\icuuc.pdb + ..\..\$(IcuLibOutputDir)\icuuc.lib + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/intl/icu/source/common/common.vcxproj.filters b/intl/icu/source/common/common.vcxproj.filters new file mode 100644 index 0000000000..28a5d90342 --- /dev/null +++ b/intl/icu/source/common/common.vcxproj.filters @@ -0,0 +1,1298 @@ + + + + + {8049805d-3f8b-4731-ac8a-aa19b3cbea45} + + + {7668dce2-7846-4b48-8bd4-4ef781671b62} + + + {f08dc85c-73ea-47b5-a867-b67589e6bebf} + + + {51edbf4b-fe36-401e-9571-0472d9dae727} + + + {3fd9d1ea-6efc-4768-a43d-811379162322} + + + {f74f43ea-8ea9-49a9-a097-e2aa5a409368} + + + {c7b856da-4f50-42c5-a741-8995df25dc6b} + + + {c76cfce3-cfe6-4de0-aa37-44c599377e2f} + + + {9abf6a02-9db0-4ecd-a8d9-bbb28073a424} + *.c,*.h + + + {b481ca81-7bc8-4399-a220-27fc29a7b74d} + + + {923074b3-0112-4faf-b8c7-59feb85a0835} + + + {17948745-0376-4851-bf12-7dabf94622e7} + + + {b32b9445-b6dc-4800-94a6-28641e586cd4} + + + {ea285df7-8e35-4b10-ae2a-86cb32c0d477} + + + {2c52192a-7b17-4c3b-998e-ca025063bd3c} + + + + + bidi + + + bidi + + + bidi + + + bidi + + + bidi + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + collation + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + formatting + + + formatting + + + formatting + + + formatting + + + formatting + + + formatting + + + formatting + + + formatting + + + idna + + + idna + + + idna + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + normalization + + + normalization + + + normalization + + + normalization + + + normalization + + + normalization + + + normalization + + + normalization + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + registration + + + registration + + + registration + + + registration + + + registration + + + registration + + + registration + + + sprep + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + collections + + + collections + + + collections + + + formatting + + + strings + + + break iteration + + + conversion + + + collections + + + collections + + + collections + + + collections + + + collections + + + properties & sets + + + collections + + + locales & resources + + + break iteration + + + registration + + + bidi + + + formatting + + + data & memory + + + + + bidi + + + bidi + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + collation + + + collation + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + collections + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + conversion + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + formatting + + + formatting + + + formatting + + + formatting + + + formatting + + + formatting + + + formatting + + + formatting + + + idna + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + normalization + + + normalization + + + normalization + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + registration + + + registration + + + registration + + + registration + + + sprep + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + configuration + + + locales & resources + + + configuration + + + properties & sets + + + bidi + + + properties & sets + + + properties & sets + + + formatting + + + break iteration + + + properties & sets + + + break iteration + + + bidi + + + formatting + + + data & memory + + + data & memory + + + + + configuration + + + + + bidi + + + bidi + + + break iteration + + + break iteration + + + break iteration + + + break iteration + + + collections + + + collections + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + configuration + + + conversion + + + conversion + + + conversion + + + conversion + + + data & memory + + + data & memory + + + data & memory + + + data & memory + + + formatting + + + formatting + + + formatting + + + formatting + + + idna + + + idna + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + locales & resources + + + normalization + + + normalization + + + normalization + + + normalization + + + normalization + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + properties & sets + + + sprep + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + strings + + + collections + + + collections + + + formatting + + + strings + + + collections + + + collections + + + collections + + + collections + + + collections + + + data & memory + + + configuration + + + collections + + + strings + + + diff --git a/intl/icu/source/common/common_uwp.vcxproj b/intl/icu/source/common/common_uwp.vcxproj new file mode 100644 index 0000000000..5df0d57a7d --- /dev/null +++ b/intl/icu/source/common/common_uwp.vcxproj @@ -0,0 +1,539 @@ + + + + + Debug + Win32 + + + Debug + x64 + + + Debug + ARM + + + Debug + ARM64 + + + Release + Win32 + + + Release + x64 + + + Release + ARM + + + Release + ARM64 + + + + {C10CF34B-3F79-430E-AD38-5A32DC0589C2} + DynamicLibrary + en-US + + + DynamicLibrary + false + MultiByte + + + + + + + + + + + + <_ProjectFileVersion>10.0.30319.1 + .\$(Platform)\$(Configuration)UWP\ + .\$(Platform)\$(Configuration)UWP\ + + .\x86\$(Configuration)UWP\ + .\x86\$(Configuration)UWP\ + + true + false + + + + + true + true + $(OutDir)\icuuc.tlb + + + ..\..\include;..\common;%(AdditionalIncludeDirectories) + + + U_ATTRIBUTE_DEPRECATED=;_CRT_SECURE_NO_DEPRECATE;U_COMMON_IMPLEMENTATION;U_PLATFORM_USES_ONLY_WIN32_API=1;%(PreprocessorDefinitions) + true + + + true + false + true + Level3 + true + Default + NotUsing + false + /utf-8 %(AdditionalOptions) + $(OutDir)/icuuc.pch + $(OutDir)/ + $(OutDir)/ + $(OutDir)/icuuc.pdb + ProgramDatabase + + + 0x0409 + ../common;%(AdditionalIncludeDirectories) + + + true + false + true + + + true + vccorlib.lib;msvcrt.lib + + icudt.lib;onecore.lib;%(AdditionalDependencies) + + + + + + NDEBUG;%(PreprocessorDefinitions) + + + NDEBUG;%(PreprocessorDefinitions) + MultiThreadedDLL + + + NDEBUG;%(PreprocessorDefinitions) + + + true + vccorlib.lib;msvcrt.lib;vcruntime.lib;%(AdditionalDependencies) + ..\..\$(IcuBinOutputDir)\icuuc$(IcuMajorVersion).dll + .\..\..\$(IcuLibOutputDir)\icuuc.pdb + ..\..\$(IcuLibOutputDir)\icuuc.lib + + + + + + _DEBUG;%(PreprocessorDefinitions) + + + _DEBUG;%(PreprocessorDefinitions) + MultiThreadedDebugDLL + Disabled + EnableFastChecks + true + + + _DEBUG;%(PreprocessorDefinitions) + + + true + vccorlibd.lib;msvcrtd.lib;vcruntimed.lib;%(AdditionalDependencies) + ..\..\$(IcuBinOutputDir)\icuuc$(IcuMajorVersion)d.dll + .\..\..\$(IcuLibOutputDir)\icuucd.pdb + ..\..\$(IcuLibOutputDir)\icuucd.lib + + + + + + Win32 + + + WIN32;%(PreprocessorDefinitions) + + + + .\..\..\lib;%(AdditionalLibraryDirectories) + + + + + + X64 + + + WIN64;WIN32;%(PreprocessorDefinitions) + + + MachineX64 + + .\..\..\lib64;%(AdditionalLibraryDirectories) + + + + + + ARM + + + ARM;WIN32;%(PreprocessorDefinitions) + + + MachineARM + + .\..\..\libARM;%(AdditionalLibraryDirectories) + + + + + + ARM64 + + + ARM64;WIN32;%(PreprocessorDefinitions) + + + MachineARM64 + + .\..\..\libARM64;%(AdditionalLibraryDirectories) + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/intl/icu/source/common/cpputils.h b/intl/icu/source/common/cpputils.h new file mode 100644 index 0000000000..2eda999a65 --- /dev/null +++ b/intl/icu/source/common/cpputils.h @@ -0,0 +1,97 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1997-2011, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: cpputils.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +*/ + +#ifndef CPPUTILS_H +#define CPPUTILS_H + +#include "unicode/utypes.h" +#include "unicode/unistr.h" +#include "cmemory.h" + +/*==========================================================================*/ +/* Array copy utility functions */ +/*==========================================================================*/ + +static +inline void uprv_arrayCopy(const double* src, double* dst, int32_t count) +{ uprv_memcpy(dst, src, (size_t)count * sizeof(*src)); } + +static +inline void uprv_arrayCopy(const double* src, int32_t srcStart, + double* dst, int32_t dstStart, int32_t count) +{ uprv_memcpy(dst+dstStart, src+srcStart, (size_t)count * sizeof(*src)); } + +static +inline void uprv_arrayCopy(const int8_t* src, int8_t* dst, int32_t count) + { uprv_memcpy(dst, src, (size_t)count * sizeof(*src)); } + +static +inline void uprv_arrayCopy(const int8_t* src, int32_t srcStart, + int8_t* dst, int32_t dstStart, int32_t count) +{ uprv_memcpy(dst+dstStart, src+srcStart, (size_t)count * sizeof(*src)); } + +static +inline void uprv_arrayCopy(const int16_t* src, int16_t* dst, int32_t count) +{ uprv_memcpy(dst, src, (size_t)count * sizeof(*src)); } + +static +inline void uprv_arrayCopy(const int16_t* src, int32_t srcStart, + int16_t* dst, int32_t dstStart, int32_t count) +{ uprv_memcpy(dst+dstStart, src+srcStart, (size_t)count * sizeof(*src)); } + +static +inline void uprv_arrayCopy(const int32_t* src, int32_t* dst, int32_t count) +{ uprv_memcpy(dst, src, (size_t)count * sizeof(*src)); } + +static +inline void uprv_arrayCopy(const int32_t* src, int32_t srcStart, + int32_t* dst, int32_t dstStart, int32_t count) +{ uprv_memcpy(dst+dstStart, src+srcStart, (size_t)count * sizeof(*src)); } + +static +inline void +uprv_arrayCopy(const char16_t *src, int32_t srcStart, + char16_t *dst, int32_t dstStart, int32_t count) +{ uprv_memcpy(dst+dstStart, src+srcStart, (size_t)count * sizeof(*src)); } + +/** + * Copy an array of UnicodeString OBJECTS (not pointers). + * @internal + */ +static inline void +uprv_arrayCopy(const icu::UnicodeString *src, icu::UnicodeString *dst, int32_t count) +{ while(count-- > 0) *dst++ = *src++; } + +/** + * Copy an array of UnicodeString OBJECTS (not pointers). + * @internal + */ +static inline void +uprv_arrayCopy(const icu::UnicodeString *src, int32_t srcStart, + icu::UnicodeString *dst, int32_t dstStart, int32_t count) +{ uprv_arrayCopy(src+srcStart, dst+dstStart, count); } + +/** + * Checks that the string is readable and writable. + * Sets U_ILLEGAL_ARGUMENT_ERROR if the string isBogus() or has an open getBuffer(). + */ +inline void +uprv_checkCanGetBuffer(const icu::UnicodeString &s, UErrorCode &errorCode) { + if(U_SUCCESS(errorCode) && s.isBogus()) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + } +} + +#endif /* _CPPUTILS */ diff --git a/intl/icu/source/common/cstr.cpp b/intl/icu/source/common/cstr.cpp new file mode 100644 index 0000000000..b87597e450 --- /dev/null +++ b/intl/icu/source/common/cstr.cpp @@ -0,0 +1,54 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2015-2016, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: charstr.cpp +*/ +#include "unicode/utypes.h" +#include "unicode/putil.h" +#include "unicode/unistr.h" + +#include "cstr.h" + +#include "charstr.h" +#include "uinvchar.h" + +U_NAMESPACE_BEGIN + +CStr::CStr(const UnicodeString &in) { + UErrorCode status = U_ZERO_ERROR; +#if !UCONFIG_NO_CONVERSION || U_CHARSET_IS_UTF8 + int32_t length = in.extract(0, in.length(), static_cast(nullptr), static_cast(0)); + int32_t resultCapacity = 0; + char *buf = s.getAppendBuffer(length, length, resultCapacity, status); + if (U_SUCCESS(status)) { + in.extract(0, in.length(), buf, resultCapacity); + s.append(buf, length, status); + } +#else + // No conversion available. Convert any invariant characters; substitute '?' for the rest. + // Note: can't just call u_UCharsToChars() or CharString.appendInvariantChars() on the + // whole string because they require that the entire input be invariant. + char buf[2]; + for (int i=0; i +#include +#include "unicode/utypes.h" +#include "cmemory.h" +#include "cstring.h" +#include "uassert.h" + +/* + * We hardcode case conversion for invariant characters to match our expectation + * and the compiler execution charset. + * This prevents problems on systems + * - with non-default casing behavior, like Turkish system locales where + * tolower('I') maps to dotless i and toupper('i') maps to dotted I + * - where there are no lowercase Latin characters at all, or using different + * codes (some old EBCDIC codepages) + * + * This works because the compiler usually runs on a platform where the execution + * charset includes all of the invariant characters at their expected + * code positions, so that the char * string literals in ICU code match + * the char literals here. + * + * Note that the set of lowercase Latin letters is discontiguous in EBCDIC + * and the set of uppercase Latin letters is discontiguous as well. + */ + +U_CAPI UBool U_EXPORT2 +uprv_isASCIILetter(char c) { +#if U_CHARSET_FAMILY==U_EBCDIC_FAMILY + return + ('a'<=c && c<='i') || ('j'<=c && c<='r') || ('s'<=c && c<='z') || + ('A'<=c && c<='I') || ('J'<=c && c<='R') || ('S'<=c && c<='Z'); +#else + return ('a'<=c && c<='z') || ('A'<=c && c<='Z'); +#endif +} + +U_CAPI char U_EXPORT2 +uprv_toupper(char c) { +#if U_CHARSET_FAMILY==U_EBCDIC_FAMILY + if(('a'<=c && c<='i') || ('j'<=c && c<='r') || ('s'<=c && c<='z')) { + c=(char)(c+('A'-'a')); + } +#else + if('a'<=c && c<='z') { + c=(char)(c+('A'-'a')); + } +#endif + return c; +} + + +#if 0 +/* + * Commented out because cstring.h defines uprv_tolower() to be + * the same as either uprv_asciitolower() or uprv_ebcdictolower() + * to reduce the amount of code to cover with tests. + * + * Note that this uprv_tolower() definition is likely to work for most + * charset families, not just ASCII and EBCDIC, because its #else branch + * is written generically. + */ +U_CAPI char U_EXPORT2 +uprv_tolower(char c) { +#if U_CHARSET_FAMILY==U_EBCDIC_FAMILY + if(('A'<=c && c<='I') || ('J'<=c && c<='R') || ('S'<=c && c<='Z')) { + c=(char)(c+('a'-'A')); + } +#else + if('A'<=c && c<='Z') { + c=(char)(c+('a'-'A')); + } +#endif + return c; +} +#endif + +U_CAPI char U_EXPORT2 +uprv_asciitolower(char c) { + if(0x41<=c && c<=0x5a) { + c=(char)(c+0x20); + } + return c; +} + +U_CAPI char U_EXPORT2 +uprv_ebcdictolower(char c) { + if( (0xc1<=(uint8_t)c && (uint8_t)c<=0xc9) || + (0xd1<=(uint8_t)c && (uint8_t)c<=0xd9) || + (0xe2<=(uint8_t)c && (uint8_t)c<=0xe9) + ) { + c=(char)(c-0x40); + } + return c; +} + + +U_CAPI char* U_EXPORT2 +T_CString_toLowerCase(char* str) +{ + char* origPtr = str; + + if (str) { + do + *str = (char)uprv_tolower(*str); + while (*(str++)); + } + + return origPtr; +} + +U_CAPI char* U_EXPORT2 +T_CString_toUpperCase(char* str) +{ + char* origPtr = str; + + if (str) { + do + *str = (char)uprv_toupper(*str); + while (*(str++)); + } + + return origPtr; +} + +/* + * Takes a int32_t and fills in a char* string with that number "radix"-based. + * Does not handle negative values (makes an empty string for them). + * Writes at most 12 chars ("-2147483647" plus NUL). + * Returns the length of the string (not including the NUL). + */ +U_CAPI int32_t U_EXPORT2 +T_CString_integerToString(char* buffer, int32_t v, int32_t radix) +{ + char tbuf[30]; + int32_t tbx = sizeof(tbuf); + uint8_t digit; + int32_t length = 0; + uint32_t uval; + + U_ASSERT(radix>=2 && radix<=16); + uval = (uint32_t) v; + if(v<0 && radix == 10) { + /* Only in base 10 do we conside numbers to be signed. */ + uval = (uint32_t)(-v); + buffer[length++] = '-'; + } + + tbx = sizeof(tbuf)-1; + tbuf[tbx] = 0; /* We are generating the digits backwards. Null term the end. */ + do { + digit = (uint8_t)(uval % radix); + tbuf[--tbx] = (char)(T_CString_itosOffset(digit)); + uval = uval / radix; + } while (uval != 0); + + /* copy converted number into user buffer */ + uprv_strcpy(buffer+length, tbuf+tbx); + length += sizeof(tbuf) - tbx -1; + return length; +} + + + +/* + * Takes a int64_t and fills in a char* string with that number "radix"-based. + * Writes at most 21: chars ("-9223372036854775807" plus NUL). + * Returns the length of the string, not including the terminating NUL. + */ +U_CAPI int32_t U_EXPORT2 +T_CString_int64ToString(char* buffer, int64_t v, uint32_t radix) +{ + char tbuf[30]; + int32_t tbx = sizeof(tbuf); + uint8_t digit; + int32_t length = 0; + uint64_t uval; + + U_ASSERT(radix>=2 && radix<=16); + uval = (uint64_t) v; + if(v<0 && radix == 10) { + /* Only in base 10 do we conside numbers to be signed. */ + uval = (uint64_t)(-v); + buffer[length++] = '-'; + } + + tbx = sizeof(tbuf)-1; + tbuf[tbx] = 0; /* We are generating the digits backwards. Null term the end. */ + do { + digit = (uint8_t)(uval % radix); + tbuf[--tbx] = (char)(T_CString_itosOffset(digit)); + uval = uval / radix; + } while (uval != 0); + + /* copy converted number into user buffer */ + uprv_strcpy(buffer+length, tbuf+tbx); + length += sizeof(tbuf) - tbx -1; + return length; +} + + +U_CAPI int32_t U_EXPORT2 +T_CString_stringToInteger(const char *integerString, int32_t radix) +{ + char *end; + return uprv_strtoul(integerString, &end, radix); + +} + +U_CAPI int U_EXPORT2 +uprv_stricmp(const char *str1, const char *str2) { + if(str1==nullptr) { + if(str2==nullptr) { + return 0; + } else { + return -1; + } + } else if(str2==nullptr) { + return 1; + } else { + /* compare non-nullptr strings lexically with lowercase */ + int rc; + unsigned char c1, c2; + + for(;;) { + c1=(unsigned char)*str1; + c2=(unsigned char)*str2; + if(c1==0) { + if(c2==0) { + return 0; + } else { + return -1; + } + } else if(c2==0) { + return 1; + } else { + /* compare non-zero characters with lowercase */ + rc=(int)(unsigned char)uprv_tolower(c1)-(int)(unsigned char)uprv_tolower(c2); + if(rc!=0) { + return rc; + } + } + ++str1; + ++str2; + } + } +} + +U_CAPI int U_EXPORT2 +uprv_strnicmp(const char *str1, const char *str2, uint32_t n) { + if(str1==nullptr) { + if(str2==nullptr) { + return 0; + } else { + return -1; + } + } else if(str2==nullptr) { + return 1; + } else { + /* compare non-nullptr strings lexically with lowercase */ + int rc; + unsigned char c1, c2; + + for(; n--;) { + c1=(unsigned char)*str1; + c2=(unsigned char)*str2; + if(c1==0) { + if(c2==0) { + return 0; + } else { + return -1; + } + } else if(c2==0) { + return 1; + } else { + /* compare non-zero characters with lowercase */ + rc=(int)(unsigned char)uprv_tolower(c1)-(int)(unsigned char)uprv_tolower(c2); + if(rc!=0) { + return rc; + } + } + ++str1; + ++str2; + } + } + + return 0; +} + +U_CAPI char* U_EXPORT2 +uprv_strdup(const char *src) { + size_t len = uprv_strlen(src) + 1; + char *dup = (char *) uprv_malloc(len); + + if (dup) { + uprv_memcpy(dup, src, len); + } + + return dup; +} + +U_CAPI char* U_EXPORT2 +uprv_strndup(const char *src, int32_t n) { + char *dup; + + if(n < 0) { + dup = uprv_strdup(src); + } else { + dup = (char*)uprv_malloc(n+1); + if (dup) { + uprv_memcpy(dup, src, n); + dup[n] = 0; + } + } + + return dup; +} diff --git a/intl/icu/source/common/cstring.h b/intl/icu/source/common/cstring.h new file mode 100644 index 0000000000..3a14e4216c --- /dev/null +++ b/intl/icu/source/common/cstring.h @@ -0,0 +1,126 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1997-2012, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* File CSTRING.H +* +* Contains CString interface +* +* @author Helena Shih +* +* Modification History: +* +* Date Name Description +* 6/17/98 hshih Created. +* 05/03/99 stephen Changed from functions to macros. +* 06/14/99 stephen Added icu_strncat, icu_strncmp, icu_tolower +* +****************************************************************************** +*/ + +#ifndef CSTRING_H +#define CSTRING_H 1 + +#include "unicode/utypes.h" +#include "cmemory.h" +#include +#include +#include + +#define uprv_strcpy(dst, src) U_STANDARD_CPP_NAMESPACE strcpy(dst, src) +#define uprv_strlen(str) U_STANDARD_CPP_NAMESPACE strlen(str) +#define uprv_strcmp(s1, s2) U_STANDARD_CPP_NAMESPACE strcmp(s1, s2) +#define uprv_strcat(dst, src) U_STANDARD_CPP_NAMESPACE strcat(dst, src) +#define uprv_strchr(s, c) U_STANDARD_CPP_NAMESPACE strchr(s, c) +#define uprv_strstr(s, c) U_STANDARD_CPP_NAMESPACE strstr(s, c) +#define uprv_strrchr(s, c) U_STANDARD_CPP_NAMESPACE strrchr(s, c) +#define uprv_strncpy(dst, src, size) U_STANDARD_CPP_NAMESPACE strncpy(dst, src, size) +#define uprv_strncmp(s1, s2, n) U_STANDARD_CPP_NAMESPACE strncmp(s1, s2, n) +#define uprv_strncat(dst, src, n) U_STANDARD_CPP_NAMESPACE strncat(dst, src, n) + +/** + * Is c an ASCII-repertoire letter a-z or A-Z? + * Note: The implementation is specific to whether ICU is compiled for + * an ASCII-based or EBCDIC-based machine. There just does not seem to be a better name for this. + */ +U_CAPI UBool U_EXPORT2 +uprv_isASCIILetter(char c); + +// NOTE: For u_asciiToUpper that takes a UChar, see ustr_imp.h + +U_CAPI char U_EXPORT2 +uprv_toupper(char c); + + +U_CAPI char U_EXPORT2 +uprv_asciitolower(char c); + +U_CAPI char U_EXPORT2 +uprv_ebcdictolower(char c); + +#if U_CHARSET_FAMILY==U_ASCII_FAMILY +# define uprv_tolower uprv_asciitolower +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY +# define uprv_tolower uprv_ebcdictolower +#else +# error U_CHARSET_FAMILY is not valid +#endif + +#define uprv_strtod(source, end) U_STANDARD_CPP_NAMESPACE strtod(source, end) +#define uprv_strtoul(str, end, base) U_STANDARD_CPP_NAMESPACE strtoul(str, end, base) +#define uprv_strtol(str, end, base) U_STANDARD_CPP_NAMESPACE strtol(str, end, base) + +/* Conversion from a digit to the character with radix base from 2-19 */ +/* May need to use U_UPPER_ORDINAL*/ +#define T_CString_itosOffset(a) ((a)<=9?('0'+(a)):('A'+(a)-10)) + +U_CAPI char* U_EXPORT2 +uprv_strdup(const char *src); + +/** + * uprv_malloc n+1 bytes, and copy n bytes from src into the new string. + * Terminate with a null at offset n. If n is -1, works like uprv_strdup + * @param src + * @param n length of the input string, not including null. + * @return new string (owned by caller, use uprv_free to free). + * @internal + */ +U_CAPI char* U_EXPORT2 +uprv_strndup(const char *src, int32_t n); + +U_CAPI char* U_EXPORT2 +T_CString_toLowerCase(char* str); + +U_CAPI char* U_EXPORT2 +T_CString_toUpperCase(char* str); + +U_CAPI int32_t U_EXPORT2 +T_CString_integerToString(char *buffer, int32_t n, int32_t radix); + +U_CAPI int32_t U_EXPORT2 +T_CString_int64ToString(char *buffer, int64_t n, uint32_t radix); + +U_CAPI int32_t U_EXPORT2 +T_CString_stringToInteger(const char *integerString, int32_t radix); + +/** + * Case-insensitive, language-independent string comparison + * limited to the ASCII character repertoire. + */ +U_CAPI int U_EXPORT2 +uprv_stricmp(const char *str1, const char *str2); + +/** + * Case-insensitive, language-independent string comparison + * limited to the ASCII character repertoire. + */ +U_CAPI int U_EXPORT2 +uprv_strnicmp(const char *str1, const char *str2, uint32_t n); + +#endif /* ! CSTRING_H */ diff --git a/intl/icu/source/common/cwchar.cpp b/intl/icu/source/common/cwchar.cpp new file mode 100644 index 0000000000..20c7d71e0f --- /dev/null +++ b/intl/icu/source/common/cwchar.cpp @@ -0,0 +1,55 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2001, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: cwchar.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2001may25 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" + +#if !U_HAVE_WCSCPY + +#include "cwchar.h" + +U_CAPI wchar_t *uprv_wcscat(wchar_t *dst, const wchar_t *src) { + wchar_t *start=dst; + while(*dst!=0) { + ++dst; + } + while((*dst=*src)!=0) { + ++dst; + ++src; + } + return start; +} + +U_CAPI wchar_t *uprv_wcscpy(wchar_t *dst, const wchar_t *src) { + wchar_t *start=dst; + while((*dst=*src)!=0) { + ++dst; + ++src; + } + return start; +} + +U_CAPI size_t uprv_wcslen(const wchar_t *src) { + const wchar_t *start=src; + while(*src!=0) { + ++src; + } + return src-start; +} + +#endif + diff --git a/intl/icu/source/common/cwchar.h b/intl/icu/source/common/cwchar.h new file mode 100644 index 0000000000..8fd041a1b9 --- /dev/null +++ b/intl/icu/source/common/cwchar.h @@ -0,0 +1,58 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2001, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: cwchar.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2001may25 +* created by: Markus W. Scherer +* +* This file contains ICU-internal definitions of wchar_t operations. +* These definitions were moved here from cstring.h so that fewer +* ICU implementation files include wchar.h. +*/ + +#ifndef __CWCHAR_H__ +#define __CWCHAR_H__ + +#include +#include +#include "unicode/utypes.h" + +/* Do this after utypes.h so that we have U_HAVE_WCHAR_H . */ +#if U_HAVE_WCHAR_H +# include +#endif + +/*===========================================================================*/ +/* Wide-character functions */ +/*===========================================================================*/ + +/* The following are not available on all systems, defined in wchar.h or string.h. */ +#if U_HAVE_WCSCPY +# define uprv_wcscpy wcscpy +# define uprv_wcscat wcscat +# define uprv_wcslen wcslen +#else +U_CAPI wchar_t* U_EXPORT2 +uprv_wcscpy(wchar_t *dst, const wchar_t *src); +U_CAPI wchar_t* U_EXPORT2 +uprv_wcscat(wchar_t *dst, const wchar_t *src); +U_CAPI size_t U_EXPORT2 +uprv_wcslen(const wchar_t *src); +#endif + +/* The following are part of the ANSI C standard, defined in stdlib.h . */ +#define uprv_wcstombs(mbstr, wcstr, count) U_STANDARD_CPP_NAMESPACE wcstombs(mbstr, wcstr, count) +#define uprv_mbstowcs(wcstr, mbstr, count) U_STANDARD_CPP_NAMESPACE mbstowcs(wcstr, mbstr, count) + + +#endif diff --git a/intl/icu/source/common/dictbe.cpp b/intl/icu/source/common/dictbe.cpp new file mode 100644 index 0000000000..0e420c67c5 --- /dev/null +++ b/intl/icu/source/common/dictbe.cpp @@ -0,0 +1,1503 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** + ******************************************************************************* + * Copyright (C) 2006-2016, International Business Machines Corporation + * and others. All Rights Reserved. + ******************************************************************************* + */ + +#include + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "brkeng.h" +#include "dictbe.h" +#include "unicode/uniset.h" +#include "unicode/chariter.h" +#include "unicode/resbund.h" +#include "unicode/ubrk.h" +#include "unicode/usetiter.h" +#include "ubrkimpl.h" +#include "utracimp.h" +#include "uvectr32.h" +#include "uvector.h" +#include "uassert.h" +#include "unicode/normlzr.h" +#include "cmemory.h" +#include "dictionarydata.h" + +U_NAMESPACE_BEGIN + +/* + ****************************************************************** + */ + +DictionaryBreakEngine::DictionaryBreakEngine() { +} + +DictionaryBreakEngine::~DictionaryBreakEngine() { +} + +UBool +DictionaryBreakEngine::handles(UChar32 c) const { + return fSet.contains(c); +} + +int32_t +DictionaryBreakEngine::findBreaks( UText *text, + int32_t startPos, + int32_t endPos, + UVector32 &foundBreaks, + UBool isPhraseBreaking, + UErrorCode& status) const { + if (U_FAILURE(status)) return 0; + (void)startPos; // TODO: remove this param? + int32_t result = 0; + + // Find the span of characters included in the set. + // The span to break begins at the current position in the text, and + // extends towards the start or end of the text, depending on 'reverse'. + + int32_t start = (int32_t)utext_getNativeIndex(text); + int32_t current; + int32_t rangeStart; + int32_t rangeEnd; + UChar32 c = utext_current32(text); + while((current = (int32_t)utext_getNativeIndex(text)) < endPos && fSet.contains(c)) { + utext_next32(text); // TODO: recast loop for postincrement + c = utext_current32(text); + } + rangeStart = start; + rangeEnd = current; + result = divideUpDictionaryRange(text, rangeStart, rangeEnd, foundBreaks, isPhraseBreaking, status); + utext_setNativeIndex(text, current); + + return result; +} + +void +DictionaryBreakEngine::setCharacters( const UnicodeSet &set ) { + fSet = set; + // Compact for caching + fSet.compact(); +} + +/* + ****************************************************************** + * PossibleWord + */ + +// Helper class for improving readability of the Thai/Lao/Khmer word break +// algorithm. The implementation is completely inline. + +// List size, limited by the maximum number of words in the dictionary +// that form a nested sequence. +static const int32_t POSSIBLE_WORD_LIST_MAX = 20; + +class PossibleWord { +private: + // list of word candidate lengths, in increasing length order + // TODO: bytes would be sufficient for word lengths. + int32_t count; // Count of candidates + int32_t prefix; // The longest match with a dictionary word + int32_t offset; // Offset in the text of these candidates + int32_t mark; // The preferred candidate's offset + int32_t current; // The candidate we're currently looking at + int32_t cuLengths[POSSIBLE_WORD_LIST_MAX]; // Word Lengths, in code units. + int32_t cpLengths[POSSIBLE_WORD_LIST_MAX]; // Word Lengths, in code points. + +public: + PossibleWord() : count(0), prefix(0), offset(-1), mark(0), current(0) {} + ~PossibleWord() {} + + // Fill the list of candidates if needed, select the longest, and return the number found + int32_t candidates( UText *text, DictionaryMatcher *dict, int32_t rangeEnd ); + + // Select the currently marked candidate, point after it in the text, and invalidate self + int32_t acceptMarked( UText *text ); + + // Back up from the current candidate to the next shorter one; return true if that exists + // and point the text after it + UBool backUp( UText *text ); + + // Return the longest prefix this candidate location shares with a dictionary word + // Return value is in code points. + int32_t longestPrefix() { return prefix; } + + // Mark the current candidate as the one we like + void markCurrent() { mark = current; } + + // Get length in code points of the marked word. + int32_t markedCPLength() { return cpLengths[mark]; } +}; + + +int32_t PossibleWord::candidates( UText *text, DictionaryMatcher *dict, int32_t rangeEnd ) { + // TODO: If getIndex is too slow, use offset < 0 and add discardAll() + int32_t start = (int32_t)utext_getNativeIndex(text); + if (start != offset) { + offset = start; + count = dict->matches(text, rangeEnd-start, UPRV_LENGTHOF(cuLengths), cuLengths, cpLengths, nullptr, &prefix); + // Dictionary leaves text after longest prefix, not longest word. Back up. + if (count <= 0) { + utext_setNativeIndex(text, start); + } + } + if (count > 0) { + utext_setNativeIndex(text, start+cuLengths[count-1]); + } + current = count-1; + mark = current; + return count; +} + +int32_t +PossibleWord::acceptMarked( UText *text ) { + utext_setNativeIndex(text, offset + cuLengths[mark]); + return cuLengths[mark]; +} + + +UBool +PossibleWord::backUp( UText *text ) { + if (current > 0) { + utext_setNativeIndex(text, offset + cuLengths[--current]); + return true; + } + return false; +} + +/* + ****************************************************************** + * ThaiBreakEngine + */ + +// How many words in a row are "good enough"? +static const int32_t THAI_LOOKAHEAD = 3; + +// Will not combine a non-word with a preceding dictionary word longer than this +static const int32_t THAI_ROOT_COMBINE_THRESHOLD = 3; + +// Will not combine a non-word that shares at least this much prefix with a +// dictionary word, with a preceding word +static const int32_t THAI_PREFIX_COMBINE_THRESHOLD = 3; + +// Elision character +static const int32_t THAI_PAIYANNOI = 0x0E2F; + +// Repeat character +static const int32_t THAI_MAIYAMOK = 0x0E46; + +// Minimum word size +static const int32_t THAI_MIN_WORD = 2; + +// Minimum number of characters for two words +static const int32_t THAI_MIN_WORD_SPAN = THAI_MIN_WORD * 2; + +ThaiBreakEngine::ThaiBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status) + : DictionaryBreakEngine(), + fDictionary(adoptDictionary) +{ + UTRACE_ENTRY(UTRACE_UBRK_CREATE_BREAK_ENGINE); + UTRACE_DATA1(UTRACE_INFO, "dictbe=%s", "Thai"); + UnicodeSet thaiWordSet(UnicodeString(u"[[:Thai:]&[:LineBreak=SA:]]"), status); + if (U_SUCCESS(status)) { + setCharacters(thaiWordSet); + } + fMarkSet.applyPattern(UnicodeString(u"[[:Thai:]&[:LineBreak=SA:]&[:M:]]"), status); + fMarkSet.add(0x0020); + fEndWordSet = thaiWordSet; + fEndWordSet.remove(0x0E31); // MAI HAN-AKAT + fEndWordSet.remove(0x0E40, 0x0E44); // SARA E through SARA AI MAIMALAI + fBeginWordSet.add(0x0E01, 0x0E2E); // KO KAI through HO NOKHUK + fBeginWordSet.add(0x0E40, 0x0E44); // SARA E through SARA AI MAIMALAI + fSuffixSet.add(THAI_PAIYANNOI); + fSuffixSet.add(THAI_MAIYAMOK); + + // Compact for caching. + fMarkSet.compact(); + fEndWordSet.compact(); + fBeginWordSet.compact(); + fSuffixSet.compact(); + UTRACE_EXIT_STATUS(status); +} + +ThaiBreakEngine::~ThaiBreakEngine() { + delete fDictionary; +} + +int32_t +ThaiBreakEngine::divideUpDictionaryRange( UText *text, + int32_t rangeStart, + int32_t rangeEnd, + UVector32 &foundBreaks, + UBool /* isPhraseBreaking */, + UErrorCode& status) const { + if (U_FAILURE(status)) return 0; + utext_setNativeIndex(text, rangeStart); + utext_moveIndex32(text, THAI_MIN_WORD_SPAN); + if (utext_getNativeIndex(text) >= rangeEnd) { + return 0; // Not enough characters for two words + } + utext_setNativeIndex(text, rangeStart); + + + uint32_t wordsFound = 0; + int32_t cpWordLength = 0; // Word Length in Code Points. + int32_t cuWordLength = 0; // Word length in code units (UText native indexing) + int32_t current; + PossibleWord words[THAI_LOOKAHEAD]; + + utext_setNativeIndex(text, rangeStart); + + while (U_SUCCESS(status) && (current = (int32_t)utext_getNativeIndex(text)) < rangeEnd) { + cpWordLength = 0; + cuWordLength = 0; + + // Look for candidate words at the current position + int32_t candidates = words[wordsFound%THAI_LOOKAHEAD].candidates(text, fDictionary, rangeEnd); + + // If we found exactly one, use that + if (candidates == 1) { + cuWordLength = words[wordsFound % THAI_LOOKAHEAD].acceptMarked(text); + cpWordLength = words[wordsFound % THAI_LOOKAHEAD].markedCPLength(); + wordsFound += 1; + } + // If there was more than one, see which one can take us forward the most words + else if (candidates > 1) { + // If we're already at the end of the range, we're done + if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) { + goto foundBest; + } + do { + if (words[(wordsFound + 1) % THAI_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) > 0) { + // Followed by another dictionary word; mark first word as a good candidate + words[wordsFound%THAI_LOOKAHEAD].markCurrent(); + + // If we're already at the end of the range, we're done + if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) { + goto foundBest; + } + + // See if any of the possible second words is followed by a third word + do { + // If we find a third word, stop right away + if (words[(wordsFound + 2) % THAI_LOOKAHEAD].candidates(text, fDictionary, rangeEnd)) { + words[wordsFound % THAI_LOOKAHEAD].markCurrent(); + goto foundBest; + } + } + while (words[(wordsFound + 1) % THAI_LOOKAHEAD].backUp(text)); + } + } + while (words[wordsFound % THAI_LOOKAHEAD].backUp(text)); +foundBest: + // Set UText position to after the accepted word. + cuWordLength = words[wordsFound % THAI_LOOKAHEAD].acceptMarked(text); + cpWordLength = words[wordsFound % THAI_LOOKAHEAD].markedCPLength(); + wordsFound += 1; + } + + // We come here after having either found a word or not. We look ahead to the + // next word. If it's not a dictionary word, we will combine it with the word we + // just found (if there is one), but only if the preceding word does not exceed + // the threshold. + // The text iterator should now be positioned at the end of the word we found. + + UChar32 uc = 0; + if ((int32_t)utext_getNativeIndex(text) < rangeEnd && cpWordLength < THAI_ROOT_COMBINE_THRESHOLD) { + // if it is a dictionary word, do nothing. If it isn't, then if there is + // no preceding word, or the non-word shares less than the minimum threshold + // of characters with a dictionary word, then scan to resynchronize + if (words[wordsFound % THAI_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) <= 0 + && (cuWordLength == 0 + || words[wordsFound%THAI_LOOKAHEAD].longestPrefix() < THAI_PREFIX_COMBINE_THRESHOLD)) { + // Look for a plausible word boundary + int32_t remaining = rangeEnd - (current+cuWordLength); + UChar32 pc; + int32_t chars = 0; + for (;;) { + int32_t pcIndex = (int32_t)utext_getNativeIndex(text); + pc = utext_next32(text); + int32_t pcSize = (int32_t)utext_getNativeIndex(text) - pcIndex; + chars += pcSize; + remaining -= pcSize; + if (remaining <= 0) { + break; + } + uc = utext_current32(text); + if (fEndWordSet.contains(pc) && fBeginWordSet.contains(uc)) { + // Maybe. See if it's in the dictionary. + // NOTE: In the original Apple code, checked that the next + // two characters after uc were not 0x0E4C THANTHAKHAT before + // checking the dictionary. That is just a performance filter, + // but it's not clear it's faster than checking the trie. + int32_t num_candidates = words[(wordsFound + 1) % THAI_LOOKAHEAD].candidates(text, fDictionary, rangeEnd); + utext_setNativeIndex(text, current + cuWordLength + chars); + if (num_candidates > 0) { + break; + } + } + } + + // Bump the word count if there wasn't already one + if (cuWordLength <= 0) { + wordsFound += 1; + } + + // Update the length with the passed-over characters + cuWordLength += chars; + } + else { + // Back up to where we were for next iteration + utext_setNativeIndex(text, current+cuWordLength); + } + } + + // Never stop before a combining mark. + int32_t currPos; + while ((currPos = (int32_t)utext_getNativeIndex(text)) < rangeEnd && fMarkSet.contains(utext_current32(text))) { + utext_next32(text); + cuWordLength += (int32_t)utext_getNativeIndex(text) - currPos; + } + + // Look ahead for possible suffixes if a dictionary word does not follow. + // We do this in code rather than using a rule so that the heuristic + // resynch continues to function. For example, one of the suffix characters + // could be a typo in the middle of a word. + if ((int32_t)utext_getNativeIndex(text) < rangeEnd && cuWordLength > 0) { + if (words[wordsFound%THAI_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) <= 0 + && fSuffixSet.contains(uc = utext_current32(text))) { + if (uc == THAI_PAIYANNOI) { + if (!fSuffixSet.contains(utext_previous32(text))) { + // Skip over previous end and PAIYANNOI + utext_next32(text); + int32_t paiyannoiIndex = (int32_t)utext_getNativeIndex(text); + utext_next32(text); + cuWordLength += (int32_t)utext_getNativeIndex(text) - paiyannoiIndex; // Add PAIYANNOI to word + uc = utext_current32(text); // Fetch next character + } + else { + // Restore prior position + utext_next32(text); + } + } + if (uc == THAI_MAIYAMOK) { + if (utext_previous32(text) != THAI_MAIYAMOK) { + // Skip over previous end and MAIYAMOK + utext_next32(text); + int32_t maiyamokIndex = (int32_t)utext_getNativeIndex(text); + utext_next32(text); + cuWordLength += (int32_t)utext_getNativeIndex(text) - maiyamokIndex; // Add MAIYAMOK to word + } + else { + // Restore prior position + utext_next32(text); + } + } + } + else { + utext_setNativeIndex(text, current+cuWordLength); + } + } + + // Did we find a word on this iteration? If so, push it on the break stack + if (cuWordLength > 0) { + foundBreaks.push((current+cuWordLength), status); + } + } + + // Don't return a break for the end of the dictionary range if there is one there. + if (foundBreaks.peeki() >= rangeEnd) { + (void) foundBreaks.popi(); + wordsFound -= 1; + } + + return wordsFound; +} + +/* + ****************************************************************** + * LaoBreakEngine + */ + +// How many words in a row are "good enough"? +static const int32_t LAO_LOOKAHEAD = 3; + +// Will not combine a non-word with a preceding dictionary word longer than this +static const int32_t LAO_ROOT_COMBINE_THRESHOLD = 3; + +// Will not combine a non-word that shares at least this much prefix with a +// dictionary word, with a preceding word +static const int32_t LAO_PREFIX_COMBINE_THRESHOLD = 3; + +// Minimum word size +static const int32_t LAO_MIN_WORD = 2; + +// Minimum number of characters for two words +static const int32_t LAO_MIN_WORD_SPAN = LAO_MIN_WORD * 2; + +LaoBreakEngine::LaoBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status) + : DictionaryBreakEngine(), + fDictionary(adoptDictionary) +{ + UTRACE_ENTRY(UTRACE_UBRK_CREATE_BREAK_ENGINE); + UTRACE_DATA1(UTRACE_INFO, "dictbe=%s", "Laoo"); + UnicodeSet laoWordSet(UnicodeString(u"[[:Laoo:]&[:LineBreak=SA:]]"), status); + if (U_SUCCESS(status)) { + setCharacters(laoWordSet); + } + fMarkSet.applyPattern(UnicodeString(u"[[:Laoo:]&[:LineBreak=SA:]&[:M:]]"), status); + fMarkSet.add(0x0020); + fEndWordSet = laoWordSet; + fEndWordSet.remove(0x0EC0, 0x0EC4); // prefix vowels + fBeginWordSet.add(0x0E81, 0x0EAE); // basic consonants (including holes for corresponding Thai characters) + fBeginWordSet.add(0x0EDC, 0x0EDD); // digraph consonants (no Thai equivalent) + fBeginWordSet.add(0x0EC0, 0x0EC4); // prefix vowels + + // Compact for caching. + fMarkSet.compact(); + fEndWordSet.compact(); + fBeginWordSet.compact(); + UTRACE_EXIT_STATUS(status); +} + +LaoBreakEngine::~LaoBreakEngine() { + delete fDictionary; +} + +int32_t +LaoBreakEngine::divideUpDictionaryRange( UText *text, + int32_t rangeStart, + int32_t rangeEnd, + UVector32 &foundBreaks, + UBool /* isPhraseBreaking */, + UErrorCode& status) const { + if (U_FAILURE(status)) return 0; + if ((rangeEnd - rangeStart) < LAO_MIN_WORD_SPAN) { + return 0; // Not enough characters for two words + } + + uint32_t wordsFound = 0; + int32_t cpWordLength = 0; + int32_t cuWordLength = 0; + int32_t current; + PossibleWord words[LAO_LOOKAHEAD]; + + utext_setNativeIndex(text, rangeStart); + + while (U_SUCCESS(status) && (current = (int32_t)utext_getNativeIndex(text)) < rangeEnd) { + cuWordLength = 0; + cpWordLength = 0; + + // Look for candidate words at the current position + int32_t candidates = words[wordsFound%LAO_LOOKAHEAD].candidates(text, fDictionary, rangeEnd); + + // If we found exactly one, use that + if (candidates == 1) { + cuWordLength = words[wordsFound % LAO_LOOKAHEAD].acceptMarked(text); + cpWordLength = words[wordsFound % LAO_LOOKAHEAD].markedCPLength(); + wordsFound += 1; + } + // If there was more than one, see which one can take us forward the most words + else if (candidates > 1) { + // If we're already at the end of the range, we're done + if (utext_getNativeIndex(text) >= rangeEnd) { + goto foundBest; + } + do { + if (words[(wordsFound + 1) % LAO_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) > 0) { + // Followed by another dictionary word; mark first word as a good candidate + words[wordsFound%LAO_LOOKAHEAD].markCurrent(); + + // If we're already at the end of the range, we're done + if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) { + goto foundBest; + } + + // See if any of the possible second words is followed by a third word + do { + // If we find a third word, stop right away + if (words[(wordsFound + 2) % LAO_LOOKAHEAD].candidates(text, fDictionary, rangeEnd)) { + words[wordsFound % LAO_LOOKAHEAD].markCurrent(); + goto foundBest; + } + } + while (words[(wordsFound + 1) % LAO_LOOKAHEAD].backUp(text)); + } + } + while (words[wordsFound % LAO_LOOKAHEAD].backUp(text)); +foundBest: + cuWordLength = words[wordsFound % LAO_LOOKAHEAD].acceptMarked(text); + cpWordLength = words[wordsFound % LAO_LOOKAHEAD].markedCPLength(); + wordsFound += 1; + } + + // We come here after having either found a word or not. We look ahead to the + // next word. If it's not a dictionary word, we will combine it with the word we + // just found (if there is one), but only if the preceding word does not exceed + // the threshold. + // The text iterator should now be positioned at the end of the word we found. + if ((int32_t)utext_getNativeIndex(text) < rangeEnd && cpWordLength < LAO_ROOT_COMBINE_THRESHOLD) { + // if it is a dictionary word, do nothing. If it isn't, then if there is + // no preceding word, or the non-word shares less than the minimum threshold + // of characters with a dictionary word, then scan to resynchronize + if (words[wordsFound % LAO_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) <= 0 + && (cuWordLength == 0 + || words[wordsFound%LAO_LOOKAHEAD].longestPrefix() < LAO_PREFIX_COMBINE_THRESHOLD)) { + // Look for a plausible word boundary + int32_t remaining = rangeEnd - (current + cuWordLength); + UChar32 pc; + UChar32 uc; + int32_t chars = 0; + for (;;) { + int32_t pcIndex = (int32_t)utext_getNativeIndex(text); + pc = utext_next32(text); + int32_t pcSize = (int32_t)utext_getNativeIndex(text) - pcIndex; + chars += pcSize; + remaining -= pcSize; + if (remaining <= 0) { + break; + } + uc = utext_current32(text); + if (fEndWordSet.contains(pc) && fBeginWordSet.contains(uc)) { + // Maybe. See if it's in the dictionary. + // TODO: this looks iffy; compare with old code. + int32_t num_candidates = words[(wordsFound + 1) % LAO_LOOKAHEAD].candidates(text, fDictionary, rangeEnd); + utext_setNativeIndex(text, current + cuWordLength + chars); + if (num_candidates > 0) { + break; + } + } + } + + // Bump the word count if there wasn't already one + if (cuWordLength <= 0) { + wordsFound += 1; + } + + // Update the length with the passed-over characters + cuWordLength += chars; + } + else { + // Back up to where we were for next iteration + utext_setNativeIndex(text, current + cuWordLength); + } + } + + // Never stop before a combining mark. + int32_t currPos; + while ((currPos = (int32_t)utext_getNativeIndex(text)) < rangeEnd && fMarkSet.contains(utext_current32(text))) { + utext_next32(text); + cuWordLength += (int32_t)utext_getNativeIndex(text) - currPos; + } + + // Look ahead for possible suffixes if a dictionary word does not follow. + // We do this in code rather than using a rule so that the heuristic + // resynch continues to function. For example, one of the suffix characters + // could be a typo in the middle of a word. + // NOT CURRENTLY APPLICABLE TO LAO + + // Did we find a word on this iteration? If so, push it on the break stack + if (cuWordLength > 0) { + foundBreaks.push((current+cuWordLength), status); + } + } + + // Don't return a break for the end of the dictionary range if there is one there. + if (foundBreaks.peeki() >= rangeEnd) { + (void) foundBreaks.popi(); + wordsFound -= 1; + } + + return wordsFound; +} + +/* + ****************************************************************** + * BurmeseBreakEngine + */ + +// How many words in a row are "good enough"? +static const int32_t BURMESE_LOOKAHEAD = 3; + +// Will not combine a non-word with a preceding dictionary word longer than this +static const int32_t BURMESE_ROOT_COMBINE_THRESHOLD = 3; + +// Will not combine a non-word that shares at least this much prefix with a +// dictionary word, with a preceding word +static const int32_t BURMESE_PREFIX_COMBINE_THRESHOLD = 3; + +// Minimum word size +static const int32_t BURMESE_MIN_WORD = 2; + +// Minimum number of characters for two words +static const int32_t BURMESE_MIN_WORD_SPAN = BURMESE_MIN_WORD * 2; + +BurmeseBreakEngine::BurmeseBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status) + : DictionaryBreakEngine(), + fDictionary(adoptDictionary) +{ + UTRACE_ENTRY(UTRACE_UBRK_CREATE_BREAK_ENGINE); + UTRACE_DATA1(UTRACE_INFO, "dictbe=%s", "Mymr"); + fBeginWordSet.add(0x1000, 0x102A); // basic consonants and independent vowels + fEndWordSet.applyPattern(UnicodeString(u"[[:Mymr:]&[:LineBreak=SA:]]"), status); + fMarkSet.applyPattern(UnicodeString(u"[[:Mymr:]&[:LineBreak=SA:]&[:M:]]"), status); + fMarkSet.add(0x0020); + if (U_SUCCESS(status)) { + setCharacters(fEndWordSet); + } + + // Compact for caching. + fMarkSet.compact(); + fEndWordSet.compact(); + fBeginWordSet.compact(); + UTRACE_EXIT_STATUS(status); +} + +BurmeseBreakEngine::~BurmeseBreakEngine() { + delete fDictionary; +} + +int32_t +BurmeseBreakEngine::divideUpDictionaryRange( UText *text, + int32_t rangeStart, + int32_t rangeEnd, + UVector32 &foundBreaks, + UBool /* isPhraseBreaking */, + UErrorCode& status ) const { + if (U_FAILURE(status)) return 0; + if ((rangeEnd - rangeStart) < BURMESE_MIN_WORD_SPAN) { + return 0; // Not enough characters for two words + } + + uint32_t wordsFound = 0; + int32_t cpWordLength = 0; + int32_t cuWordLength = 0; + int32_t current; + PossibleWord words[BURMESE_LOOKAHEAD]; + + utext_setNativeIndex(text, rangeStart); + + while (U_SUCCESS(status) && (current = (int32_t)utext_getNativeIndex(text)) < rangeEnd) { + cuWordLength = 0; + cpWordLength = 0; + + // Look for candidate words at the current position + int32_t candidates = words[wordsFound%BURMESE_LOOKAHEAD].candidates(text, fDictionary, rangeEnd); + + // If we found exactly one, use that + if (candidates == 1) { + cuWordLength = words[wordsFound % BURMESE_LOOKAHEAD].acceptMarked(text); + cpWordLength = words[wordsFound % BURMESE_LOOKAHEAD].markedCPLength(); + wordsFound += 1; + } + // If there was more than one, see which one can take us forward the most words + else if (candidates > 1) { + // If we're already at the end of the range, we're done + if (utext_getNativeIndex(text) >= rangeEnd) { + goto foundBest; + } + do { + if (words[(wordsFound + 1) % BURMESE_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) > 0) { + // Followed by another dictionary word; mark first word as a good candidate + words[wordsFound%BURMESE_LOOKAHEAD].markCurrent(); + + // If we're already at the end of the range, we're done + if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) { + goto foundBest; + } + + // See if any of the possible second words is followed by a third word + do { + // If we find a third word, stop right away + if (words[(wordsFound + 2) % BURMESE_LOOKAHEAD].candidates(text, fDictionary, rangeEnd)) { + words[wordsFound % BURMESE_LOOKAHEAD].markCurrent(); + goto foundBest; + } + } + while (words[(wordsFound + 1) % BURMESE_LOOKAHEAD].backUp(text)); + } + } + while (words[wordsFound % BURMESE_LOOKAHEAD].backUp(text)); +foundBest: + cuWordLength = words[wordsFound % BURMESE_LOOKAHEAD].acceptMarked(text); + cpWordLength = words[wordsFound % BURMESE_LOOKAHEAD].markedCPLength(); + wordsFound += 1; + } + + // We come here after having either found a word or not. We look ahead to the + // next word. If it's not a dictionary word, we will combine it with the word we + // just found (if there is one), but only if the preceding word does not exceed + // the threshold. + // The text iterator should now be positioned at the end of the word we found. + if ((int32_t)utext_getNativeIndex(text) < rangeEnd && cpWordLength < BURMESE_ROOT_COMBINE_THRESHOLD) { + // if it is a dictionary word, do nothing. If it isn't, then if there is + // no preceding word, or the non-word shares less than the minimum threshold + // of characters with a dictionary word, then scan to resynchronize + if (words[wordsFound % BURMESE_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) <= 0 + && (cuWordLength == 0 + || words[wordsFound%BURMESE_LOOKAHEAD].longestPrefix() < BURMESE_PREFIX_COMBINE_THRESHOLD)) { + // Look for a plausible word boundary + int32_t remaining = rangeEnd - (current + cuWordLength); + UChar32 pc; + UChar32 uc; + int32_t chars = 0; + for (;;) { + int32_t pcIndex = (int32_t)utext_getNativeIndex(text); + pc = utext_next32(text); + int32_t pcSize = (int32_t)utext_getNativeIndex(text) - pcIndex; + chars += pcSize; + remaining -= pcSize; + if (remaining <= 0) { + break; + } + uc = utext_current32(text); + if (fEndWordSet.contains(pc) && fBeginWordSet.contains(uc)) { + // Maybe. See if it's in the dictionary. + // TODO: this looks iffy; compare with old code. + int32_t num_candidates = words[(wordsFound + 1) % BURMESE_LOOKAHEAD].candidates(text, fDictionary, rangeEnd); + utext_setNativeIndex(text, current + cuWordLength + chars); + if (num_candidates > 0) { + break; + } + } + } + + // Bump the word count if there wasn't already one + if (cuWordLength <= 0) { + wordsFound += 1; + } + + // Update the length with the passed-over characters + cuWordLength += chars; + } + else { + // Back up to where we were for next iteration + utext_setNativeIndex(text, current + cuWordLength); + } + } + + // Never stop before a combining mark. + int32_t currPos; + while ((currPos = (int32_t)utext_getNativeIndex(text)) < rangeEnd && fMarkSet.contains(utext_current32(text))) { + utext_next32(text); + cuWordLength += (int32_t)utext_getNativeIndex(text) - currPos; + } + + // Look ahead for possible suffixes if a dictionary word does not follow. + // We do this in code rather than using a rule so that the heuristic + // resynch continues to function. For example, one of the suffix characters + // could be a typo in the middle of a word. + // NOT CURRENTLY APPLICABLE TO BURMESE + + // Did we find a word on this iteration? If so, push it on the break stack + if (cuWordLength > 0) { + foundBreaks.push((current+cuWordLength), status); + } + } + + // Don't return a break for the end of the dictionary range if there is one there. + if (foundBreaks.peeki() >= rangeEnd) { + (void) foundBreaks.popi(); + wordsFound -= 1; + } + + return wordsFound; +} + +/* + ****************************************************************** + * KhmerBreakEngine + */ + +// How many words in a row are "good enough"? +static const int32_t KHMER_LOOKAHEAD = 3; + +// Will not combine a non-word with a preceding dictionary word longer than this +static const int32_t KHMER_ROOT_COMBINE_THRESHOLD = 3; + +// Will not combine a non-word that shares at least this much prefix with a +// dictionary word, with a preceding word +static const int32_t KHMER_PREFIX_COMBINE_THRESHOLD = 3; + +// Minimum word size +static const int32_t KHMER_MIN_WORD = 2; + +// Minimum number of characters for two words +static const int32_t KHMER_MIN_WORD_SPAN = KHMER_MIN_WORD * 2; + +KhmerBreakEngine::KhmerBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status) + : DictionaryBreakEngine(), + fDictionary(adoptDictionary) +{ + UTRACE_ENTRY(UTRACE_UBRK_CREATE_BREAK_ENGINE); + UTRACE_DATA1(UTRACE_INFO, "dictbe=%s", "Khmr"); + UnicodeSet khmerWordSet(UnicodeString(u"[[:Khmr:]&[:LineBreak=SA:]]"), status); + if (U_SUCCESS(status)) { + setCharacters(khmerWordSet); + } + fMarkSet.applyPattern(UnicodeString(u"[[:Khmr:]&[:LineBreak=SA:]&[:M:]]"), status); + fMarkSet.add(0x0020); + fEndWordSet = khmerWordSet; + fBeginWordSet.add(0x1780, 0x17B3); + //fBeginWordSet.add(0x17A3, 0x17A4); // deprecated vowels + //fEndWordSet.remove(0x17A5, 0x17A9); // Khmer independent vowels that can't end a word + //fEndWordSet.remove(0x17B2); // Khmer independent vowel that can't end a word + fEndWordSet.remove(0x17D2); // KHMER SIGN COENG that combines some following characters + //fEndWordSet.remove(0x17B6, 0x17C5); // Remove dependent vowels +// fEndWordSet.remove(0x0E31); // MAI HAN-AKAT +// fEndWordSet.remove(0x0E40, 0x0E44); // SARA E through SARA AI MAIMALAI +// fBeginWordSet.add(0x0E01, 0x0E2E); // KO KAI through HO NOKHUK +// fBeginWordSet.add(0x0E40, 0x0E44); // SARA E through SARA AI MAIMALAI +// fSuffixSet.add(THAI_PAIYANNOI); +// fSuffixSet.add(THAI_MAIYAMOK); + + // Compact for caching. + fMarkSet.compact(); + fEndWordSet.compact(); + fBeginWordSet.compact(); +// fSuffixSet.compact(); + UTRACE_EXIT_STATUS(status); +} + +KhmerBreakEngine::~KhmerBreakEngine() { + delete fDictionary; +} + +int32_t +KhmerBreakEngine::divideUpDictionaryRange( UText *text, + int32_t rangeStart, + int32_t rangeEnd, + UVector32 &foundBreaks, + UBool /* isPhraseBreaking */, + UErrorCode& status ) const { + if (U_FAILURE(status)) return 0; + if ((rangeEnd - rangeStart) < KHMER_MIN_WORD_SPAN) { + return 0; // Not enough characters for two words + } + + uint32_t wordsFound = 0; + int32_t cpWordLength = 0; + int32_t cuWordLength = 0; + int32_t current; + PossibleWord words[KHMER_LOOKAHEAD]; + + utext_setNativeIndex(text, rangeStart); + + while (U_SUCCESS(status) && (current = (int32_t)utext_getNativeIndex(text)) < rangeEnd) { + cuWordLength = 0; + cpWordLength = 0; + + // Look for candidate words at the current position + int32_t candidates = words[wordsFound%KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd); + + // If we found exactly one, use that + if (candidates == 1) { + cuWordLength = words[wordsFound % KHMER_LOOKAHEAD].acceptMarked(text); + cpWordLength = words[wordsFound % KHMER_LOOKAHEAD].markedCPLength(); + wordsFound += 1; + } + + // If there was more than one, see which one can take us forward the most words + else if (candidates > 1) { + // If we're already at the end of the range, we're done + if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) { + goto foundBest; + } + do { + if (words[(wordsFound + 1) % KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) > 0) { + // Followed by another dictionary word; mark first word as a good candidate + words[wordsFound % KHMER_LOOKAHEAD].markCurrent(); + + // If we're already at the end of the range, we're done + if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) { + goto foundBest; + } + + // See if any of the possible second words is followed by a third word + do { + // If we find a third word, stop right away + if (words[(wordsFound + 2) % KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd)) { + words[wordsFound % KHMER_LOOKAHEAD].markCurrent(); + goto foundBest; + } + } + while (words[(wordsFound + 1) % KHMER_LOOKAHEAD].backUp(text)); + } + } + while (words[wordsFound % KHMER_LOOKAHEAD].backUp(text)); +foundBest: + cuWordLength = words[wordsFound % KHMER_LOOKAHEAD].acceptMarked(text); + cpWordLength = words[wordsFound % KHMER_LOOKAHEAD].markedCPLength(); + wordsFound += 1; + } + + // We come here after having either found a word or not. We look ahead to the + // next word. If it's not a dictionary word, we will combine it with the word we + // just found (if there is one), but only if the preceding word does not exceed + // the threshold. + // The text iterator should now be positioned at the end of the word we found. + if ((int32_t)utext_getNativeIndex(text) < rangeEnd && cpWordLength < KHMER_ROOT_COMBINE_THRESHOLD) { + // if it is a dictionary word, do nothing. If it isn't, then if there is + // no preceding word, or the non-word shares less than the minimum threshold + // of characters with a dictionary word, then scan to resynchronize + if (words[wordsFound % KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) <= 0 + && (cuWordLength == 0 + || words[wordsFound % KHMER_LOOKAHEAD].longestPrefix() < KHMER_PREFIX_COMBINE_THRESHOLD)) { + // Look for a plausible word boundary + int32_t remaining = rangeEnd - (current+cuWordLength); + UChar32 pc; + UChar32 uc; + int32_t chars = 0; + for (;;) { + int32_t pcIndex = (int32_t)utext_getNativeIndex(text); + pc = utext_next32(text); + int32_t pcSize = (int32_t)utext_getNativeIndex(text) - pcIndex; + chars += pcSize; + remaining -= pcSize; + if (remaining <= 0) { + break; + } + uc = utext_current32(text); + if (fEndWordSet.contains(pc) && fBeginWordSet.contains(uc)) { + // Maybe. See if it's in the dictionary. + int32_t num_candidates = words[(wordsFound + 1) % KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd); + utext_setNativeIndex(text, current+cuWordLength+chars); + if (num_candidates > 0) { + break; + } + } + } + + // Bump the word count if there wasn't already one + if (cuWordLength <= 0) { + wordsFound += 1; + } + + // Update the length with the passed-over characters + cuWordLength += chars; + } + else { + // Back up to where we were for next iteration + utext_setNativeIndex(text, current+cuWordLength); + } + } + + // Never stop before a combining mark. + int32_t currPos; + while ((currPos = (int32_t)utext_getNativeIndex(text)) < rangeEnd && fMarkSet.contains(utext_current32(text))) { + utext_next32(text); + cuWordLength += (int32_t)utext_getNativeIndex(text) - currPos; + } + + // Look ahead for possible suffixes if a dictionary word does not follow. + // We do this in code rather than using a rule so that the heuristic + // resynch continues to function. For example, one of the suffix characters + // could be a typo in the middle of a word. +// if ((int32_t)utext_getNativeIndex(text) < rangeEnd && wordLength > 0) { +// if (words[wordsFound%KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) <= 0 +// && fSuffixSet.contains(uc = utext_current32(text))) { +// if (uc == KHMER_PAIYANNOI) { +// if (!fSuffixSet.contains(utext_previous32(text))) { +// // Skip over previous end and PAIYANNOI +// utext_next32(text); +// utext_next32(text); +// wordLength += 1; // Add PAIYANNOI to word +// uc = utext_current32(text); // Fetch next character +// } +// else { +// // Restore prior position +// utext_next32(text); +// } +// } +// if (uc == KHMER_MAIYAMOK) { +// if (utext_previous32(text) != KHMER_MAIYAMOK) { +// // Skip over previous end and MAIYAMOK +// utext_next32(text); +// utext_next32(text); +// wordLength += 1; // Add MAIYAMOK to word +// } +// else { +// // Restore prior position +// utext_next32(text); +// } +// } +// } +// else { +// utext_setNativeIndex(text, current+wordLength); +// } +// } + + // Did we find a word on this iteration? If so, push it on the break stack + if (cuWordLength > 0) { + foundBreaks.push((current+cuWordLength), status); + } + } + + // Don't return a break for the end of the dictionary range if there is one there. + if (foundBreaks.peeki() >= rangeEnd) { + (void) foundBreaks.popi(); + wordsFound -= 1; + } + + return wordsFound; +} + +#if !UCONFIG_NO_NORMALIZATION +/* + ****************************************************************** + * CjkBreakEngine + */ +static const uint32_t kuint32max = 0xFFFFFFFF; +CjkBreakEngine::CjkBreakEngine(DictionaryMatcher *adoptDictionary, LanguageType type, UErrorCode &status) +: DictionaryBreakEngine(), fDictionary(adoptDictionary), isCj(false) { + UTRACE_ENTRY(UTRACE_UBRK_CREATE_BREAK_ENGINE); + UTRACE_DATA1(UTRACE_INFO, "dictbe=%s", "Hani"); + fMlBreakEngine = nullptr; + nfkcNorm2 = Normalizer2::getNFKCInstance(status); + // Korean dictionary only includes Hangul syllables + fHangulWordSet.applyPattern(UnicodeString(u"[\\uac00-\\ud7a3]"), status); + fHangulWordSet.compact(); + // Digits, open puncutation and Alphabetic characters. + fDigitOrOpenPunctuationOrAlphabetSet.applyPattern( + UnicodeString(u"[[:Nd:][:Pi:][:Ps:][:Alphabetic:]]"), status); + fDigitOrOpenPunctuationOrAlphabetSet.compact(); + fClosePunctuationSet.applyPattern(UnicodeString(u"[[:Pc:][:Pd:][:Pe:][:Pf:][:Po:]]"), status); + fClosePunctuationSet.compact(); + + // handle Korean and Japanese/Chinese using different dictionaries + if (type == kKorean) { + if (U_SUCCESS(status)) { + setCharacters(fHangulWordSet); + } + } else { // Chinese and Japanese + UnicodeSet cjSet(UnicodeString(u"[[:Han:][:Hiragana:][:Katakana:]\\u30fc\\uff70\\uff9e\\uff9f]"), status); + isCj = true; + if (U_SUCCESS(status)) { + setCharacters(cjSet); +#if UCONFIG_USE_ML_PHRASE_BREAKING + fMlBreakEngine = new MlBreakEngine(fDigitOrOpenPunctuationOrAlphabetSet, + fClosePunctuationSet, status); + if (fMlBreakEngine == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + } +#else + initJapanesePhraseParameter(status); +#endif + } + } + UTRACE_EXIT_STATUS(status); +} + +CjkBreakEngine::~CjkBreakEngine(){ + delete fDictionary; + delete fMlBreakEngine; +} + +// The katakanaCost values below are based on the length frequencies of all +// katakana phrases in the dictionary +static const int32_t kMaxKatakanaLength = 8; +static const int32_t kMaxKatakanaGroupLength = 20; +static const uint32_t maxSnlp = 255; + +static inline uint32_t getKatakanaCost(int32_t wordLength){ + //TODO: fill array with actual values from dictionary! + static const uint32_t katakanaCost[kMaxKatakanaLength + 1] + = {8192, 984, 408, 240, 204, 252, 300, 372, 480}; + return (wordLength > kMaxKatakanaLength) ? 8192 : katakanaCost[wordLength]; +} + +static inline bool isKatakana(UChar32 value) { + return (value >= 0x30A1 && value <= 0x30FE && value != 0x30FB) || + (value >= 0xFF66 && value <= 0xFF9f); +} + +// Function for accessing internal utext flags. +// Replicates an internal UText function. + +static inline int32_t utext_i32_flag(int32_t bitIndex) { + return (int32_t)1 << bitIndex; +} + +/* + * @param text A UText representing the text + * @param rangeStart The start of the range of dictionary characters + * @param rangeEnd The end of the range of dictionary characters + * @param foundBreaks vector to receive the break positions + * @return The number of breaks found + */ +int32_t +CjkBreakEngine::divideUpDictionaryRange( UText *inText, + int32_t rangeStart, + int32_t rangeEnd, + UVector32 &foundBreaks, + UBool isPhraseBreaking, + UErrorCode& status) const { + if (U_FAILURE(status)) return 0; + if (rangeStart >= rangeEnd) { + return 0; + } + + // UnicodeString version of input UText, NFKC normalized if necessary. + UnicodeString inString; + + // inputMap[inStringIndex] = corresponding native index from UText inText. + // If nullptr then mapping is 1:1 + LocalPointer inputMap; + + // if UText has the input string as one contiguous UTF-16 chunk + if ((inText->providerProperties & utext_i32_flag(UTEXT_PROVIDER_STABLE_CHUNKS)) && + inText->chunkNativeStart <= rangeStart && + inText->chunkNativeLimit >= rangeEnd && + inText->nativeIndexingLimit >= rangeEnd - inText->chunkNativeStart) { + + // Input UText is in one contiguous UTF-16 chunk. + // Use Read-only aliasing UnicodeString. + inString.setTo(false, + inText->chunkContents + rangeStart - inText->chunkNativeStart, + rangeEnd - rangeStart); + } else { + // Copy the text from the original inText (UText) to inString (UnicodeString). + // Create a map from UnicodeString indices -> UText offsets. + utext_setNativeIndex(inText, rangeStart); + int32_t limit = rangeEnd; + U_ASSERT(limit <= utext_nativeLength(inText)); + if (limit > utext_nativeLength(inText)) { + limit = (int32_t)utext_nativeLength(inText); + } + inputMap.adoptInsteadAndCheckErrorCode(new UVector32(status), status); + if (U_FAILURE(status)) { + return 0; + } + while (utext_getNativeIndex(inText) < limit) { + int32_t nativePosition = (int32_t)utext_getNativeIndex(inText); + UChar32 c = utext_next32(inText); + U_ASSERT(c != U_SENTINEL); + inString.append(c); + while (inputMap->size() < inString.length()) { + inputMap->addElement(nativePosition, status); + } + } + inputMap->addElement(limit, status); + } + + + if (!nfkcNorm2->isNormalized(inString, status)) { + UnicodeString normalizedInput; + // normalizedMap[normalizedInput position] == original UText position. + LocalPointer normalizedMap(new UVector32(status), status); + if (U_FAILURE(status)) { + return 0; + } + + UnicodeString fragment; + UnicodeString normalizedFragment; + for (int32_t srcI = 0; srcI < inString.length();) { // Once per normalization chunk + fragment.remove(); + int32_t fragmentStartI = srcI; + UChar32 c = inString.char32At(srcI); + for (;;) { + fragment.append(c); + srcI = inString.moveIndex32(srcI, 1); + if (srcI == inString.length()) { + break; + } + c = inString.char32At(srcI); + if (nfkcNorm2->hasBoundaryBefore(c)) { + break; + } + } + nfkcNorm2->normalize(fragment, normalizedFragment, status); + normalizedInput.append(normalizedFragment); + + // Map every position in the normalized chunk to the start of the chunk + // in the original input. + int32_t fragmentOriginalStart = inputMap.isValid() ? + inputMap->elementAti(fragmentStartI) : fragmentStartI+rangeStart; + while (normalizedMap->size() < normalizedInput.length()) { + normalizedMap->addElement(fragmentOriginalStart, status); + if (U_FAILURE(status)) { + break; + } + } + } + U_ASSERT(normalizedMap->size() == normalizedInput.length()); + int32_t nativeEnd = inputMap.isValid() ? + inputMap->elementAti(inString.length()) : inString.length()+rangeStart; + normalizedMap->addElement(nativeEnd, status); + + inputMap = std::move(normalizedMap); + inString = std::move(normalizedInput); + } + + int32_t numCodePts = inString.countChar32(); + if (numCodePts != inString.length()) { + // There are supplementary characters in the input. + // The dictionary will produce boundary positions in terms of code point indexes, + // not in terms of code unit string indexes. + // Use the inputMap mechanism to take care of this in addition to indexing differences + // from normalization and/or UTF-8 input. + UBool hadExistingMap = inputMap.isValid(); + if (!hadExistingMap) { + inputMap.adoptInsteadAndCheckErrorCode(new UVector32(status), status); + if (U_FAILURE(status)) { + return 0; + } + } + int32_t cpIdx = 0; + for (int32_t cuIdx = 0; ; cuIdx = inString.moveIndex32(cuIdx, 1)) { + U_ASSERT(cuIdx >= cpIdx); + if (hadExistingMap) { + inputMap->setElementAt(inputMap->elementAti(cuIdx), cpIdx); + } else { + inputMap->addElement(cuIdx+rangeStart, status); + } + cpIdx++; + if (cuIdx == inString.length()) { + break; + } + } + } + +#if UCONFIG_USE_ML_PHRASE_BREAKING + // PhraseBreaking is supported in ja and ko; MlBreakEngine only supports ja. + if (isPhraseBreaking && isCj) { + return fMlBreakEngine->divideUpRange(inText, rangeStart, rangeEnd, foundBreaks, inString, + inputMap, status); + } +#endif + + // bestSnlp[i] is the snlp of the best segmentation of the first i + // code points in the range to be matched. + UVector32 bestSnlp(numCodePts + 1, status); + bestSnlp.addElement(0, status); + for(int32_t i = 1; i <= numCodePts; i++) { + bestSnlp.addElement(kuint32max, status); + } + + + // prev[i] is the index of the last CJK code point in the previous word in + // the best segmentation of the first i characters. + UVector32 prev(numCodePts + 1, status); + for(int32_t i = 0; i <= numCodePts; i++){ + prev.addElement(-1, status); + } + + const int32_t maxWordSize = 20; + UVector32 values(numCodePts, status); + values.setSize(numCodePts); + UVector32 lengths(numCodePts, status); + lengths.setSize(numCodePts); + + UText fu = UTEXT_INITIALIZER; + utext_openUnicodeString(&fu, &inString, &status); + + // Dynamic programming to find the best segmentation. + + // In outer loop, i is the code point index, + // ix is the corresponding string (code unit) index. + // They differ when the string contains supplementary characters. + int32_t ix = 0; + bool is_prev_katakana = false; + for (int32_t i = 0; i < numCodePts; ++i, ix = inString.moveIndex32(ix, 1)) { + if ((uint32_t)bestSnlp.elementAti(i) == kuint32max) { + continue; + } + + int32_t count; + utext_setNativeIndex(&fu, ix); + count = fDictionary->matches(&fu, maxWordSize, numCodePts, + nullptr, lengths.getBuffer(), values.getBuffer(), nullptr); + // Note: lengths is filled with code point lengths + // The nullptr parameter is the ignored code unit lengths. + + // if there are no single character matches found in the dictionary + // starting with this character, treat character as a 1-character word + // with the highest value possible, i.e. the least likely to occur. + // Exclude Korean characters from this treatment, as they should be left + // together by default. + if ((count == 0 || lengths.elementAti(0) != 1) && + !fHangulWordSet.contains(inString.char32At(ix))) { + values.setElementAt(maxSnlp, count); // 255 + lengths.setElementAt(1, count++); + } + + for (int32_t j = 0; j < count; j++) { + uint32_t newSnlp = (uint32_t)bestSnlp.elementAti(i) + (uint32_t)values.elementAti(j); + int32_t ln_j_i = lengths.elementAti(j) + i; + if (newSnlp < (uint32_t)bestSnlp.elementAti(ln_j_i)) { + bestSnlp.setElementAt(newSnlp, ln_j_i); + prev.setElementAt(i, ln_j_i); + } + } + + // In Japanese, + // Katakana word in single character is pretty rare. So we apply + // the following heuristic to Katakana: any continuous run of Katakana + // characters is considered a candidate word with a default cost + // specified in the katakanaCost table according to its length. + + bool is_katakana = isKatakana(inString.char32At(ix)); + int32_t katakanaRunLength = 1; + if (!is_prev_katakana && is_katakana) { + int32_t j = inString.moveIndex32(ix, 1); + // Find the end of the continuous run of Katakana characters + while (j < inString.length() && katakanaRunLength < kMaxKatakanaGroupLength && + isKatakana(inString.char32At(j))) { + j = inString.moveIndex32(j, 1); + katakanaRunLength++; + } + if (katakanaRunLength < kMaxKatakanaGroupLength) { + uint32_t newSnlp = bestSnlp.elementAti(i) + getKatakanaCost(katakanaRunLength); + if (newSnlp < (uint32_t)bestSnlp.elementAti(i+katakanaRunLength)) { + bestSnlp.setElementAt(newSnlp, i+katakanaRunLength); + prev.setElementAt(i, i+katakanaRunLength); // prev[j] = i; + } + } + } + is_prev_katakana = is_katakana; + } + utext_close(&fu); + + // Start pushing the optimal offset index into t_boundary (t for tentative). + // prev[numCodePts] is guaranteed to be meaningful. + // We'll first push in the reverse order, i.e., + // t_boundary[0] = numCodePts, and afterwards do a swap. + UVector32 t_boundary(numCodePts+1, status); + + int32_t numBreaks = 0; + // No segmentation found, set boundary to end of range + if ((uint32_t)bestSnlp.elementAti(numCodePts) == kuint32max) { + t_boundary.addElement(numCodePts, status); + numBreaks++; + } else if (isPhraseBreaking) { + t_boundary.addElement(numCodePts, status); + if(U_SUCCESS(status)) { + numBreaks++; + int32_t prevIdx = numCodePts; + + int32_t codeUnitIdx = -1; + int32_t prevCodeUnitIdx = -1; + int32_t length = -1; + for (int32_t i = prev.elementAti(numCodePts); i > 0; i = prev.elementAti(i)) { + codeUnitIdx = inString.moveIndex32(0, i); + prevCodeUnitIdx = inString.moveIndex32(0, prevIdx); + // Calculate the length by using the code unit. + length = prevCodeUnitIdx - codeUnitIdx; + prevIdx = i; + // Keep the breakpoint if the pattern is not in the fSkipSet and continuous Katakana + // characters don't occur. + if (!fSkipSet.containsKey(inString.tempSubString(codeUnitIdx, length)) + && (!isKatakana(inString.char32At(inString.moveIndex32(codeUnitIdx, -1))) + || !isKatakana(inString.char32At(codeUnitIdx)))) { + t_boundary.addElement(i, status); + numBreaks++; + } + } + } + } else { + for (int32_t i = numCodePts; i > 0; i = prev.elementAti(i)) { + t_boundary.addElement(i, status); + numBreaks++; + } + U_ASSERT(prev.elementAti(t_boundary.elementAti(numBreaks - 1)) == 0); + } + + // Add a break for the start of the dictionary range if there is not one + // there already. + if (foundBreaks.size() == 0 || foundBreaks.peeki() < rangeStart) { + t_boundary.addElement(0, status); + numBreaks++; + } + + // Now that we're done, convert positions in t_boundary[] (indices in + // the normalized input string) back to indices in the original input UText + // while reversing t_boundary and pushing values to foundBreaks. + int32_t prevCPPos = -1; + int32_t prevUTextPos = -1; + int32_t correctedNumBreaks = 0; + for (int32_t i = numBreaks - 1; i >= 0; i--) { + int32_t cpPos = t_boundary.elementAti(i); + U_ASSERT(cpPos > prevCPPos); + int32_t utextPos = inputMap.isValid() ? inputMap->elementAti(cpPos) : cpPos + rangeStart; + U_ASSERT(utextPos >= prevUTextPos); + if (utextPos > prevUTextPos) { + // Boundaries are added to foundBreaks output in ascending order. + U_ASSERT(foundBreaks.size() == 0 || foundBreaks.peeki() < utextPos); + // In phrase breaking, there has to be a breakpoint between Cj character and close + // punctuation. + // E.g.［携帯電話］正しい選択 -> ［携帯▁電話］▁正しい▁選択 -> breakpoint between ］ and 正 + if (utextPos != rangeStart + || (isPhraseBreaking && utextPos > 0 + && fClosePunctuationSet.contains(utext_char32At(inText, utextPos - 1)))) { + foundBreaks.push(utextPos, status); + correctedNumBreaks++; + } + } else { + // Normalization expanded the input text, the dictionary found a boundary + // within the expansion, giving two boundaries with the same index in the + // original text. Ignore the second. See ticket #12918. + --numBreaks; + } + prevCPPos = cpPos; + prevUTextPos = utextPos; + } + (void)prevCPPos; // suppress compiler warnings about unused variable + + UChar32 nextChar = utext_char32At(inText, rangeEnd); + if (!foundBreaks.isEmpty() && foundBreaks.peeki() == rangeEnd) { + // In phrase breaking, there has to be a breakpoint between Cj character and + // the number/open punctuation. + // E.g. る文字「そうだ、京都」->る▁文字▁「そうだ、▁京都」-> breakpoint between 字 and「 + // E.g. 乗車率９０％程度だろうか -> 乗車▁率▁９０％▁程度だろうか -> breakpoint between 率 and ９ + // E.g. しかもロゴがＵｎｉｃｏｄｅ！ -> しかも▁ロゴが▁Ｕｎｉｃｏｄｅ！-> breakpoint between が and Ｕ + if (isPhraseBreaking) { + if (!fDigitOrOpenPunctuationOrAlphabetSet.contains(nextChar)) { + foundBreaks.popi(); + correctedNumBreaks--; + } + } else { + foundBreaks.popi(); + correctedNumBreaks--; + } + } + + // inString goes out of scope + // inputMap goes out of scope + return correctedNumBreaks; +} + +void CjkBreakEngine::initJapanesePhraseParameter(UErrorCode& error) { + loadJapaneseExtensions(error); + loadHiragana(error); +} + +void CjkBreakEngine::loadJapaneseExtensions(UErrorCode& error) { + const char* tag = "extensions"; + ResourceBundle ja(U_ICUDATA_BRKITR, "ja", error); + if (U_SUCCESS(error)) { + ResourceBundle bundle = ja.get(tag, error); + while (U_SUCCESS(error) && bundle.hasNext()) { + fSkipSet.puti(bundle.getNextString(error), 1, error); + } + } +} + +void CjkBreakEngine::loadHiragana(UErrorCode& error) { + UnicodeSet hiraganaWordSet(UnicodeString(u"[:Hiragana:]"), error); + hiraganaWordSet.compact(); + UnicodeSetIterator iterator(hiraganaWordSet); + while (iterator.next()) { + fSkipSet.puti(UnicodeString(iterator.getCodepoint()), 1, error); + } +} +#endif + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ + diff --git a/intl/icu/source/common/dictbe.h b/intl/icu/source/common/dictbe.h new file mode 100644 index 0000000000..a2c761bdc3 --- /dev/null +++ b/intl/icu/source/common/dictbe.h @@ -0,0 +1,433 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** + ******************************************************************************* + * Copyright (C) 2006-2014, International Business Machines Corporation * + * and others. All Rights Reserved. * + ******************************************************************************* + */ + +#ifndef DICTBE_H +#define DICTBE_H + +#include "unicode/utypes.h" +#include "unicode/uniset.h" +#include "unicode/utext.h" + +#include "brkeng.h" +#include "hash.h" +#include "mlbe.h" +#include "uvectr32.h" + +U_NAMESPACE_BEGIN + +class DictionaryMatcher; +class MlBreakEngine; +class Normalizer2; + +/******************************************************************* + * DictionaryBreakEngine + */ + +/** + *

DictionaryBreakEngine is a kind of LanguageBreakEngine that uses a + * dictionary to determine language-specific breaks.

+ * + *

After it is constructed a DictionaryBreakEngine may be shared between + * threads without synchronization.

+ */ +class DictionaryBreakEngine : public LanguageBreakEngine { + private: + /** + * The set of characters handled by this engine + * @internal + */ + + UnicodeSet fSet; + + public: + + /** + *

Constructor

+ */ + DictionaryBreakEngine(); + + /** + *

Virtual destructor.

+ */ + virtual ~DictionaryBreakEngine(); + + /** + *

Indicate whether this engine handles a particular character for + * a particular kind of break.

+ * + * @param c A character which begins a run that the engine might handle + * @return true if this engine handles the particular character and break + * type. + */ + virtual UBool handles(UChar32 c) const override; + + /** + *

Find any breaks within a run in the supplied text.

+ * + * @param text A UText representing the text. The iterator is left at + * the end of the run of characters which the engine is capable of handling + * that starts from the first character in the range. + * @param startPos The start of the run within the supplied text. + * @param endPos The end of the run within the supplied text. + * @param foundBreaks vector of int32_t to receive the break positions + * @param status Information on any errors encountered. + * @return The number of breaks found. + */ + virtual int32_t findBreaks( UText *text, + int32_t startPos, + int32_t endPos, + UVector32 &foundBreaks, + UBool isPhraseBreaking, + UErrorCode& status ) const override; + + protected: + + /** + *

Set the character set handled by this engine.

+ * + * @param set A UnicodeSet of the set of characters handled by the engine + */ + virtual void setCharacters( const UnicodeSet &set ); + + /** + *

Divide up a range of known dictionary characters handled by this break engine.

+ * + * @param text A UText representing the text + * @param rangeStart The start of the range of dictionary characters + * @param rangeEnd The end of the range of dictionary characters + * @param foundBreaks Output of C array of int32_t break positions, or 0 + * @param status Information on any errors encountered. + * @return The number of breaks found + */ + virtual int32_t divideUpDictionaryRange( UText *text, + int32_t rangeStart, + int32_t rangeEnd, + UVector32 &foundBreaks, + UBool isPhraseBreaking, + UErrorCode& status) const = 0; + +}; + +/******************************************************************* + * ThaiBreakEngine + */ + +/** + *

ThaiBreakEngine is a kind of DictionaryBreakEngine that uses a + * dictionary and heuristics to determine Thai-specific breaks.

+ * + *

After it is constructed a ThaiBreakEngine may be shared between + * threads without synchronization.

+ */ +class ThaiBreakEngine : public DictionaryBreakEngine { + private: + /** + * The set of characters handled by this engine + * @internal + */ + + UnicodeSet fEndWordSet; + UnicodeSet fBeginWordSet; + UnicodeSet fSuffixSet; + UnicodeSet fMarkSet; + DictionaryMatcher *fDictionary; + + public: + + /** + *

Default constructor.

+ * + * @param adoptDictionary A DictionaryMatcher to adopt. Deleted when the + * engine is deleted. + */ + ThaiBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status); + + /** + *

Virtual destructor.

+ */ + virtual ~ThaiBreakEngine(); + + protected: + /** + *

Divide up a range of known dictionary characters handled by this break engine.

+ * + * @param text A UText representing the text + * @param rangeStart The start of the range of dictionary characters + * @param rangeEnd The end of the range of dictionary characters + * @param foundBreaks Output of C array of int32_t break positions, or 0 + * @param status Information on any errors encountered. + * @return The number of breaks found + */ + virtual int32_t divideUpDictionaryRange( UText *text, + int32_t rangeStart, + int32_t rangeEnd, + UVector32 &foundBreaks, + UBool isPhraseBreaking, + UErrorCode& status) const override; + +}; + +/******************************************************************* + * LaoBreakEngine + */ + +/** + *

LaoBreakEngine is a kind of DictionaryBreakEngine that uses a + * dictionary and heuristics to determine Lao-specific breaks.

+ * + *

After it is constructed a LaoBreakEngine may be shared between + * threads without synchronization.

+ */ +class LaoBreakEngine : public DictionaryBreakEngine { + private: + /** + * The set of characters handled by this engine + * @internal + */ + + UnicodeSet fEndWordSet; + UnicodeSet fBeginWordSet; + UnicodeSet fMarkSet; + DictionaryMatcher *fDictionary; + + public: + + /** + *

Default constructor.

+ * + * @param adoptDictionary A DictionaryMatcher to adopt. Deleted when the + * engine is deleted. + */ + LaoBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status); + + /** + *

Virtual destructor.

+ */ + virtual ~LaoBreakEngine(); + + protected: + /** + *

Divide up a range of known dictionary characters handled by this break engine.

+ * + * @param text A UText representing the text + * @param rangeStart The start of the range of dictionary characters + * @param rangeEnd The end of the range of dictionary characters + * @param foundBreaks Output of C array of int32_t break positions, or 0 + * @param status Information on any errors encountered. + * @return The number of breaks found + */ + virtual int32_t divideUpDictionaryRange( UText *text, + int32_t rangeStart, + int32_t rangeEnd, + UVector32 &foundBreaks, + UBool isPhraseBreaking, + UErrorCode& status) const override; + +}; + +/******************************************************************* + * BurmeseBreakEngine + */ + +/** + *

BurmeseBreakEngine is a kind of DictionaryBreakEngine that uses a + * DictionaryMatcher and heuristics to determine Burmese-specific breaks.

+ * + *

After it is constructed a BurmeseBreakEngine may be shared between + * threads without synchronization.

+ */ +class BurmeseBreakEngine : public DictionaryBreakEngine { + private: + /** + * The set of characters handled by this engine + * @internal + */ + + UnicodeSet fEndWordSet; + UnicodeSet fBeginWordSet; + UnicodeSet fMarkSet; + DictionaryMatcher *fDictionary; + + public: + + /** + *

Default constructor.

+ * + * @param adoptDictionary A DictionaryMatcher to adopt. Deleted when the + * engine is deleted. + */ + BurmeseBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status); + + /** + *

Virtual destructor.

+ */ + virtual ~BurmeseBreakEngine(); + + protected: + /** + *

Divide up a range of known dictionary characters.

+ * + * @param text A UText representing the text + * @param rangeStart The start of the range of dictionary characters + * @param rangeEnd The end of the range of dictionary characters + * @param foundBreaks Output of C array of int32_t break positions, or 0 + * @param status Information on any errors encountered. + * @return The number of breaks found + */ + virtual int32_t divideUpDictionaryRange( UText *text, + int32_t rangeStart, + int32_t rangeEnd, + UVector32 &foundBreaks, + UBool isPhraseBreaking, + UErrorCode& status) const override; + +}; + +/******************************************************************* + * KhmerBreakEngine + */ + +/** + *

KhmerBreakEngine is a kind of DictionaryBreakEngine that uses a + * DictionaryMatcher and heuristics to determine Khmer-specific breaks.

+ * + *

After it is constructed a KhmerBreakEngine may be shared between + * threads without synchronization.

+ */ +class KhmerBreakEngine : public DictionaryBreakEngine { + private: + /** + * The set of characters handled by this engine + * @internal + */ + + UnicodeSet fEndWordSet; + UnicodeSet fBeginWordSet; + UnicodeSet fMarkSet; + DictionaryMatcher *fDictionary; + + public: + + /** + *

Default constructor.

+ * + * @param adoptDictionary A DictionaryMatcher to adopt. Deleted when the + * engine is deleted. + */ + KhmerBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status); + + /** + *

Virtual destructor.

+ */ + virtual ~KhmerBreakEngine(); + + protected: + /** + *

Divide up a range of known dictionary characters.

+ * + * @param text A UText representing the text + * @param rangeStart The start of the range of dictionary characters + * @param rangeEnd The end of the range of dictionary characters + * @param foundBreaks Output of C array of int32_t break positions, or 0 + * @param status Information on any errors encountered. + * @return The number of breaks found + */ + virtual int32_t divideUpDictionaryRange( UText *text, + int32_t rangeStart, + int32_t rangeEnd, + UVector32 &foundBreaks, + UBool isPhraseBreaking, + UErrorCode& status) const override; + +}; + +#if !UCONFIG_NO_NORMALIZATION + +/******************************************************************* + * CjkBreakEngine + */ + +//indicates language/script that the CjkBreakEngine will handle +enum LanguageType { + kKorean, + kChineseJapanese +}; + +/** + *

CjkBreakEngine is a kind of DictionaryBreakEngine that uses a + * dictionary with costs associated with each word and + * Viterbi decoding to determine CJK-specific breaks.

+ */ +class CjkBreakEngine : public DictionaryBreakEngine { + protected: + /** + * The set of characters handled by this engine + * @internal + */ + UnicodeSet fHangulWordSet; + UnicodeSet fDigitOrOpenPunctuationOrAlphabetSet; + UnicodeSet fClosePunctuationSet; + + DictionaryMatcher *fDictionary; + const Normalizer2 *nfkcNorm2; + MlBreakEngine *fMlBreakEngine; + bool isCj; + + private: + // Load Japanese extensions. + void loadJapaneseExtensions(UErrorCode& error); + // Load Japanese Hiragana. + void loadHiragana(UErrorCode& error); + // Initialize fSkipSet by loading Japanese Hiragana and extensions. + void initJapanesePhraseParameter(UErrorCode& error); + + Hashtable fSkipSet; + + public: + + /** + *

Default constructor.

+ * + * @param adoptDictionary A DictionaryMatcher to adopt. Deleted when the + * engine is deleted. The DictionaryMatcher must contain costs for each word + * in order for the dictionary to work properly. + */ + CjkBreakEngine(DictionaryMatcher *adoptDictionary, LanguageType type, UErrorCode &status); + + /** + *

Virtual destructor.

+ */ + virtual ~CjkBreakEngine(); + + protected: + /** + *

Divide up a range of known dictionary characters handled by this break engine.

+ * + * @param text A UText representing the text + * @param rangeStart The start of the range of dictionary characters + * @param rangeEnd The end of the range of dictionary characters + * @param foundBreaks Output of C array of int32_t break positions, or 0 + * @param status Information on any errors encountered. + * @return The number of breaks found + */ + virtual int32_t divideUpDictionaryRange( UText *text, + int32_t rangeStart, + int32_t rangeEnd, + UVector32 &foundBreaks, + UBool isPhraseBreaking, + UErrorCode& status) const override; + +}; + +#endif + +U_NAMESPACE_END + + /* DICTBE_H */ +#endif diff --git a/intl/icu/source/common/dictionarydata.cpp b/intl/icu/source/common/dictionarydata.cpp new file mode 100644 index 0000000000..80b6d82d56 --- /dev/null +++ b/intl/icu/source/common/dictionarydata.cpp @@ -0,0 +1,242 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2014-2016, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* dictionarydata.h +* +* created on: 2012may31 +* created by: Markus W. Scherer & Maxime Serrano +*/ + +#include "dictionarydata.h" +#include "unicode/ucharstrie.h" +#include "unicode/bytestrie.h" +#include "unicode/udata.h" +#include "cmemory.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +U_NAMESPACE_BEGIN + +const int32_t DictionaryData::TRIE_TYPE_BYTES = 0; +const int32_t DictionaryData::TRIE_TYPE_UCHARS = 1; +const int32_t DictionaryData::TRIE_TYPE_MASK = 7; +const int32_t DictionaryData::TRIE_HAS_VALUES = 8; + +const int32_t DictionaryData::TRANSFORM_NONE = 0; +const int32_t DictionaryData::TRANSFORM_TYPE_OFFSET = 0x1000000; +const int32_t DictionaryData::TRANSFORM_TYPE_MASK = 0x7f000000; +const int32_t DictionaryData::TRANSFORM_OFFSET_MASK = 0x1fffff; + +DictionaryMatcher::~DictionaryMatcher() { +} + +UCharsDictionaryMatcher::~UCharsDictionaryMatcher() { + udata_close(file); +} + +int32_t UCharsDictionaryMatcher::getType() const { + return DictionaryData::TRIE_TYPE_UCHARS; +} + +int32_t UCharsDictionaryMatcher::matches(UText *text, int32_t maxLength, int32_t limit, + int32_t *lengths, int32_t *cpLengths, int32_t *values, + int32_t *prefix) const { + + UCharsTrie uct(characters); + int32_t startingTextIndex = (int32_t)utext_getNativeIndex(text); + int32_t wordCount = 0; + int32_t codePointsMatched = 0; + + for (UChar32 c = utext_next32(text); c >= 0; c=utext_next32(text)) { + UStringTrieResult result = (codePointsMatched == 0) ? uct.first(c) : uct.next(c); + int32_t lengthMatched = (int32_t)utext_getNativeIndex(text) - startingTextIndex; + codePointsMatched += 1; + if (USTRINGTRIE_HAS_VALUE(result)) { + if (wordCount < limit) { + if (values != nullptr) { + values[wordCount] = uct.getValue(); + } + if (lengths != nullptr) { + lengths[wordCount] = lengthMatched; + } + if (cpLengths != nullptr) { + cpLengths[wordCount] = codePointsMatched; + } + ++wordCount; + } + if (result == USTRINGTRIE_FINAL_VALUE) { + break; + } + } + else if (result == USTRINGTRIE_NO_MATCH) { + break; + } + if (lengthMatched >= maxLength) { + break; + } + } + + if (prefix != nullptr) { + *prefix = codePointsMatched; + } + return wordCount; +} + +BytesDictionaryMatcher::~BytesDictionaryMatcher() { + udata_close(file); +} + +UChar32 BytesDictionaryMatcher::transform(UChar32 c) const { + if ((transformConstant & DictionaryData::TRANSFORM_TYPE_MASK) == DictionaryData::TRANSFORM_TYPE_OFFSET) { + if (c == 0x200D) { + return 0xFF; + } else if (c == 0x200C) { + return 0xFE; + } + int32_t delta = c - (transformConstant & DictionaryData::TRANSFORM_OFFSET_MASK); + if (delta < 0 || 0xFD < delta) { + return U_SENTINEL; + } + return (UChar32)delta; + } + return c; +} + +int32_t BytesDictionaryMatcher::getType() const { + return DictionaryData::TRIE_TYPE_BYTES; +} + +int32_t BytesDictionaryMatcher::matches(UText *text, int32_t maxLength, int32_t limit, + int32_t *lengths, int32_t *cpLengths, int32_t *values, + int32_t *prefix) const { + BytesTrie bt(characters); + int32_t startingTextIndex = (int32_t)utext_getNativeIndex(text); + int32_t wordCount = 0; + int32_t codePointsMatched = 0; + + for (UChar32 c = utext_next32(text); c >= 0; c=utext_next32(text)) { + UStringTrieResult result = (codePointsMatched == 0) ? bt.first(transform(c)) : bt.next(transform(c)); + int32_t lengthMatched = (int32_t)utext_getNativeIndex(text) - startingTextIndex; + codePointsMatched += 1; + if (USTRINGTRIE_HAS_VALUE(result)) { + if (wordCount < limit) { + if (values != nullptr) { + values[wordCount] = bt.getValue(); + } + if (lengths != nullptr) { + lengths[wordCount] = lengthMatched; + } + if (cpLengths != nullptr) { + cpLengths[wordCount] = codePointsMatched; + } + ++wordCount; + } + if (result == USTRINGTRIE_FINAL_VALUE) { + break; + } + } + else if (result == USTRINGTRIE_NO_MATCH) { + break; + } + if (lengthMatched >= maxLength) { + break; + } + } + + if (prefix != nullptr) { + *prefix = codePointsMatched; + } + return wordCount; +} + + +U_NAMESPACE_END + +U_NAMESPACE_USE + +U_CAPI int32_t U_EXPORT2 +udict_swap(const UDataSwapper *ds, const void *inData, int32_t length, + void *outData, UErrorCode *pErrorCode) { + const UDataInfo *pInfo; + int32_t headerSize; + const uint8_t *inBytes; + uint8_t *outBytes; + const int32_t *inIndexes; + int32_t indexes[DictionaryData::IX_COUNT]; + int32_t i, offset, size; + + headerSize = udata_swapDataHeader(ds, inData, length, outData, pErrorCode); + if (pErrorCode == nullptr || U_FAILURE(*pErrorCode)) return 0; + pInfo = (const UDataInfo *)((const char *)inData + 4); + if (!(pInfo->dataFormat[0] == 0x44 && + pInfo->dataFormat[1] == 0x69 && + pInfo->dataFormat[2] == 0x63 && + pInfo->dataFormat[3] == 0x74 && + pInfo->formatVersion[0] == 1)) { + udata_printError(ds, "udict_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized as dictionary data\n", + pInfo->dataFormat[0], pInfo->dataFormat[1], pInfo->dataFormat[2], pInfo->dataFormat[3], pInfo->formatVersion[0]); + *pErrorCode = U_UNSUPPORTED_ERROR; + return 0; + } + + inBytes = (const uint8_t *)inData + headerSize; + outBytes = (outData == nullptr) ? nullptr : (uint8_t *)outData + headerSize; + + inIndexes = (const int32_t *)inBytes; + if (length >= 0) { + length -= headerSize; + if (length < (int32_t)(sizeof(indexes))) { + udata_printError(ds, "udict_swap(): too few bytes (%d after header) for dictionary data\n", length); + *pErrorCode = U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + } + + for (i = 0; i < DictionaryData::IX_COUNT; i++) { + indexes[i] = udata_readInt32(ds, inIndexes[i]); + } + + size = indexes[DictionaryData::IX_TOTAL_SIZE]; + + if (length >= 0) { + if (length < size) { + udata_printError(ds, "udict_swap(): too few bytes (%d after header) for all of dictionary data\n", length); + *pErrorCode = U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + + if (inBytes != outBytes) { + uprv_memcpy(outBytes, inBytes, size); + } + + offset = 0; + ds->swapArray32(ds, inBytes, sizeof(indexes), outBytes, pErrorCode); + offset = (int32_t)sizeof(indexes); + int32_t trieType = indexes[DictionaryData::IX_TRIE_TYPE] & DictionaryData::TRIE_TYPE_MASK; + int32_t nextOffset = indexes[DictionaryData::IX_RESERVED1_OFFSET]; + + if (trieType == DictionaryData::TRIE_TYPE_UCHARS) { + ds->swapArray16(ds, inBytes + offset, nextOffset - offset, outBytes + offset, pErrorCode); + } else if (trieType == DictionaryData::TRIE_TYPE_BYTES) { + // nothing to do + } else { + udata_printError(ds, "udict_swap(): unknown trie type!\n"); + *pErrorCode = U_UNSUPPORTED_ERROR; + return 0; + } + + // these next two sections are empty in the current format, + // but may be used later. + offset = nextOffset; + nextOffset = indexes[DictionaryData::IX_RESERVED2_OFFSET]; + offset = nextOffset; + nextOffset = indexes[DictionaryData::IX_TOTAL_SIZE]; + offset = nextOffset; + } + return headerSize + size; +} +#endif diff --git a/intl/icu/source/common/dictionarydata.h b/intl/icu/source/common/dictionarydata.h new file mode 100644 index 0000000000..8751e502d5 --- /dev/null +++ b/intl/icu/source/common/dictionarydata.h @@ -0,0 +1,191 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2014, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* dictionarydata.h +* +* created on: 2012may31 +* created by: Markus W. Scherer & Maxime Serrano +*/ + +#ifndef __DICTIONARYDATA_H__ +#define __DICTIONARYDATA_H__ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/utext.h" +#include "unicode/udata.h" +#include "udataswp.h" +#include "unicode/uobject.h" +#include "unicode/ustringtrie.h" + +U_NAMESPACE_BEGIN + +class UCharsTrie; +class BytesTrie; + +class U_COMMON_API DictionaryData : public UMemory { +public: + static const int32_t TRIE_TYPE_BYTES; // = 0; + static const int32_t TRIE_TYPE_UCHARS; // = 1; + static const int32_t TRIE_TYPE_MASK; // = 7; + static const int32_t TRIE_HAS_VALUES; // = 8; + + static const int32_t TRANSFORM_NONE; // = 0; + static const int32_t TRANSFORM_TYPE_OFFSET; // = 0x1000000; + static const int32_t TRANSFORM_TYPE_MASK; // = 0x7f000000; + static const int32_t TRANSFORM_OFFSET_MASK; // = 0x1fffff; + + enum { + // Byte offsets from the start of the data, after the generic header. + IX_STRING_TRIE_OFFSET, + IX_RESERVED1_OFFSET, + IX_RESERVED2_OFFSET, + IX_TOTAL_SIZE, + + // Trie type: TRIE_HAS_VALUES | TRIE_TYPE_BYTES etc. + IX_TRIE_TYPE, + // Transform specification: TRANSFORM_TYPE_OFFSET | 0xe00 etc. + IX_TRANSFORM, + + IX_RESERVED6, + IX_RESERVED7, + IX_COUNT + }; +}; + +/** + * Wrapper class around generic dictionaries, implementing matches(). + * getType() should return a TRIE_TYPE_??? constant from DictionaryData. + * + * All implementations of this interface must be thread-safe if they are to be used inside of the + * dictionary-based break iteration code. + */ +class U_COMMON_API DictionaryMatcher : public UMemory { +public: + DictionaryMatcher() {} + virtual ~DictionaryMatcher(); + // this should emulate CompactTrieDictionary::matches() + /* @param text The text in which to look for matching words. Matching begins + * at the current position of the UText. + * @param maxLength The max length of match to consider. Units are the native indexing + * units of the UText. + * @param limit Capacity of output arrays, which is also the maximum number of + * matching words to be found. + * @param lengths output array, filled with the lengths of the matches, in order, + * from shortest to longest. Lengths are in native indexing units + * of the UText. May be nullptr. + * @param cpLengths output array, filled with the lengths of the matches, in order, + * from shortest to longest. Lengths are the number of Unicode code points. + * May be nullptr. + * @param values Output array, filled with the values associated with the words found. + * May be nullptr. + * @param prefix Output parameter, the code point length of the prefix match, even if that + * prefix didn't lead to a complete word. Will always be >= the cpLength + * of the longest complete word matched. May be nullptr. + * @return Number of matching words found. + */ + virtual int32_t matches(UText *text, int32_t maxLength, int32_t limit, + int32_t *lengths, int32_t *cpLengths, int32_t *values, + int32_t *prefix) const = 0; + + /** @return DictionaryData::TRIE_TYPE_XYZ */ + virtual int32_t getType() const = 0; +}; + +// Implementation of the DictionaryMatcher interface for a UCharsTrie dictionary +class U_COMMON_API UCharsDictionaryMatcher : public DictionaryMatcher { +public: + // constructs a new UCharsDictionaryMatcher. + // The UDataMemory * will be closed on this object's destruction. + UCharsDictionaryMatcher(const char16_t *c, UDataMemory *f) : characters(c), file(f) { } + virtual ~UCharsDictionaryMatcher(); + virtual int32_t matches(UText *text, int32_t maxLength, int32_t limit, + int32_t *lengths, int32_t *cpLengths, int32_t *values, + int32_t *prefix) const override; + virtual int32_t getType() const override; +private: + const char16_t *characters; + UDataMemory *file; +}; + +// Implementation of the DictionaryMatcher interface for a BytesTrie dictionary +class U_COMMON_API BytesDictionaryMatcher : public DictionaryMatcher { +public: + // constructs a new BytesTrieDictionaryMatcher + // the transform constant should be the constant read from the file, not a masked version! + // the UDataMemory * fed in here will be closed on this object's destruction + BytesDictionaryMatcher(const char *c, int32_t t, UDataMemory *f) + : characters(c), transformConstant(t), file(f) { } + virtual ~BytesDictionaryMatcher(); + virtual int32_t matches(UText *text, int32_t maxLength, int32_t limit, + int32_t *lengths, int32_t *cpLengths, int32_t *values, + int32_t *prefix) const override; + virtual int32_t getType() const override; +private: + UChar32 transform(UChar32 c) const; + + const char *characters; + int32_t transformConstant; + UDataMemory *file; +}; + +U_NAMESPACE_END + +U_CAPI int32_t U_EXPORT2 +udict_swap(const UDataSwapper *ds, const void *inData, int32_t length, void *outData, UErrorCode *pErrorCode); + +/** + * Format of dictionary .dict data files. + * Format version 1.0. + * + * A dictionary .dict data file contains a byte-serialized BytesTrie or + * a UChars-serialized UCharsTrie. + * Such files are used in dictionary-based break iteration (DBBI). + * + * For a BytesTrie, a transformation type is specified for + * transforming Unicode strings into byte sequences. + * + * A .dict file begins with a standard ICU data file header + * (DataHeader, see ucmndata.h and unicode/udata.h). + * The UDataInfo.dataVersion field is currently unused (set to 0.0.0.0). + * + * After the header, the file contains the following parts. + * Constants are defined in the DictionaryData class. + * + * For the data structure of BytesTrie & UCharsTrie see + * https://icu.unicode.org/design/struct/tries + * and the bytestrie.h and ucharstrie.h header files. + * + * int32_t indexes[indexesLength]; -- indexesLength=indexes[IX_STRING_TRIE_OFFSET]/4; + * + * The first four indexes are byte offsets in ascending order. + * Each byte offset marks the start of the next part in the data file, + * and the end of the previous one. + * When two consecutive byte offsets are the same, then the corresponding part is empty. + * Byte offsets are offsets from after the header, + * that is, from the beginning of the indexes[]. + * Each part starts at an offset with proper alignment for its data. + * If necessary, the previous part may include padding bytes to achieve this alignment. + * + * trieType=indexes[IX_TRIE_TYPE] defines the trie type. + * transform=indexes[IX_TRANSFORM] defines the Unicode-to-bytes transformation. + * If the transformation type is TRANSFORM_TYPE_OFFSET, + * then the lower 21 bits contain the offset code point. + * Each code point c is mapped to byte b = (c - offset). + * Code points outside the range offset..(offset+0xff) cannot be mapped + * and do not occur in the dictionary. + * + * stringTrie; -- a serialized BytesTrie or UCharsTrie + * + * The dictionary maps strings to specific values (TRIE_HAS_VALUES bit set in trieType), + * or it maps all strings to 0 (TRIE_HAS_VALUES bit not set). + */ + +#endif /* !UCONFIG_NO_BREAK_ITERATION */ +#endif /* __DICTIONARYDATA_H__ */ diff --git a/intl/icu/source/common/dtintrv.cpp b/intl/icu/source/common/dtintrv.cpp new file mode 100644 index 0000000000..011ad8136f --- /dev/null +++ b/intl/icu/source/common/dtintrv.cpp @@ -0,0 +1,63 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/******************************************************************************* +* Copyright (C) 2008, International Business Machines Corporation and +* others. All Rights Reserved. +******************************************************************************* +* +* File DTINTRV.CPP +* +******************************************************************************* +*/ + + + +#include "unicode/dtintrv.h" + + +U_NAMESPACE_BEGIN + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(DateInterval) + +//DateInterval::DateInterval(){} + + +DateInterval::DateInterval(UDate from, UDate to) +: fromDate(from), + toDate(to) +{} + + +DateInterval::~DateInterval(){} + + +DateInterval::DateInterval(const DateInterval& other) +: UObject(other) { + *this = other; +} + + +DateInterval& +DateInterval::operator=(const DateInterval& other) { + if ( this != &other ) { + fromDate = other.fromDate; + toDate = other.toDate; + } + return *this; +} + + +DateInterval* +DateInterval::clone() const { + return new DateInterval(*this); +} + + +bool +DateInterval::operator==(const DateInterval& other) const { + return ( fromDate == other.fromDate && toDate == other.toDate ); +} + + +U_NAMESPACE_END + diff --git a/intl/icu/source/common/edits.cpp b/intl/icu/source/common/edits.cpp new file mode 100644 index 0000000000..3348145d48 --- /dev/null +++ b/intl/icu/source/common/edits.cpp @@ -0,0 +1,804 @@ +// © 2017 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// edits.cpp +// created: 2017feb08 Markus W. Scherer + +#include "unicode/edits.h" +#include "unicode/unistr.h" +#include "unicode/utypes.h" +#include "cmemory.h" +#include "uassert.h" +#include "util.h" + +U_NAMESPACE_BEGIN + +namespace { + +// 0000uuuuuuuuuuuu records u+1 unchanged text units. +const int32_t MAX_UNCHANGED_LENGTH = 0x1000; +const int32_t MAX_UNCHANGED = MAX_UNCHANGED_LENGTH - 1; + +// 0mmmnnnccccccccc with m=1..6 records ccc+1 replacements of m:n text units. +const int32_t MAX_SHORT_CHANGE_OLD_LENGTH = 6; +const int32_t MAX_SHORT_CHANGE_NEW_LENGTH = 7; +const int32_t SHORT_CHANGE_NUM_MASK = 0x1ff; +const int32_t MAX_SHORT_CHANGE = 0x6fff; + +// 0111mmmmmmnnnnnn records a replacement of m text units with n. +// m or n = 61: actual length follows in the next edits array unit. +// m or n = 62..63: actual length follows in the next two edits array units. +// Bit 30 of the actual length is in the head unit. +// Trailing units have bit 15 set. +const int32_t LENGTH_IN_1TRAIL = 61; +const int32_t LENGTH_IN_2TRAIL = 62; + +} // namespace + +void Edits::releaseArray() noexcept { + if (array != stackArray) { + uprv_free(array); + } +} + +Edits &Edits::copyArray(const Edits &other) { + if (U_FAILURE(errorCode_)) { + length = delta = numChanges = 0; + return *this; + } + if (length > capacity) { + uint16_t *newArray = (uint16_t *)uprv_malloc((size_t)length * 2); + if (newArray == nullptr) { + length = delta = numChanges = 0; + errorCode_ = U_MEMORY_ALLOCATION_ERROR; + return *this; + } + releaseArray(); + array = newArray; + capacity = length; + } + if (length > 0) { + uprv_memcpy(array, other.array, (size_t)length * 2); + } + return *this; +} + +Edits &Edits::moveArray(Edits &src) noexcept { + if (U_FAILURE(errorCode_)) { + length = delta = numChanges = 0; + return *this; + } + releaseArray(); + if (length > STACK_CAPACITY) { + array = src.array; + capacity = src.capacity; + src.array = src.stackArray; + src.capacity = STACK_CAPACITY; + src.reset(); + return *this; + } + array = stackArray; + capacity = STACK_CAPACITY; + if (length > 0) { + uprv_memcpy(array, src.array, (size_t)length * 2); + } + return *this; +} + +Edits &Edits::operator=(const Edits &other) { + if (this == &other) { return *this; } // self-assignment: no-op + length = other.length; + delta = other.delta; + numChanges = other.numChanges; + errorCode_ = other.errorCode_; + return copyArray(other); +} + +Edits &Edits::operator=(Edits &&src) noexcept { + length = src.length; + delta = src.delta; + numChanges = src.numChanges; + errorCode_ = src.errorCode_; + return moveArray(src); +} + +Edits::~Edits() { + releaseArray(); +} + +void Edits::reset() noexcept { + length = delta = numChanges = 0; + errorCode_ = U_ZERO_ERROR; +} + +void Edits::addUnchanged(int32_t unchangedLength) { + if(U_FAILURE(errorCode_) || unchangedLength == 0) { return; } + if(unchangedLength < 0) { + errorCode_ = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + // Merge into previous unchanged-text record, if any. + int32_t last = lastUnit(); + if(last < MAX_UNCHANGED) { + int32_t remaining = MAX_UNCHANGED - last; + if (remaining >= unchangedLength) { + setLastUnit(last + unchangedLength); + return; + } + setLastUnit(MAX_UNCHANGED); + unchangedLength -= remaining; + } + // Split large lengths into multiple units. + while(unchangedLength >= MAX_UNCHANGED_LENGTH) { + append(MAX_UNCHANGED); + unchangedLength -= MAX_UNCHANGED_LENGTH; + } + // Write a small (remaining) length. + if(unchangedLength > 0) { + append(unchangedLength - 1); + } +} + +void Edits::addReplace(int32_t oldLength, int32_t newLength) { + if(U_FAILURE(errorCode_)) { return; } + if(oldLength < 0 || newLength < 0) { + errorCode_ = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + if (oldLength == 0 && newLength == 0) { + return; + } + ++numChanges; + int32_t newDelta = newLength - oldLength; + if (newDelta != 0) { + if ((newDelta > 0 && delta >= 0 && newDelta > (INT32_MAX - delta)) || + (newDelta < 0 && delta < 0 && newDelta < (INT32_MIN - delta))) { + // Integer overflow or underflow. + errorCode_ = U_INDEX_OUTOFBOUNDS_ERROR; + return; + } + delta += newDelta; + } + + if(0 < oldLength && oldLength <= MAX_SHORT_CHANGE_OLD_LENGTH && + newLength <= MAX_SHORT_CHANGE_NEW_LENGTH) { + // Merge into previous same-lengths short-replacement record, if any. + int32_t u = (oldLength << 12) | (newLength << 9); + int32_t last = lastUnit(); + if(MAX_UNCHANGED < last && last < MAX_SHORT_CHANGE && + (last & ~SHORT_CHANGE_NUM_MASK) == u && + (last & SHORT_CHANGE_NUM_MASK) < SHORT_CHANGE_NUM_MASK) { + setLastUnit(last + 1); + return; + } + append(u); + return; + } + + int32_t head = 0x7000; + if (oldLength < LENGTH_IN_1TRAIL && newLength < LENGTH_IN_1TRAIL) { + head |= oldLength << 6; + head |= newLength; + append(head); + } else if ((capacity - length) >= 5 || growArray()) { + int32_t limit = length + 1; + if(oldLength < LENGTH_IN_1TRAIL) { + head |= oldLength << 6; + } else if(oldLength <= 0x7fff) { + head |= LENGTH_IN_1TRAIL << 6; + array[limit++] = (uint16_t)(0x8000 | oldLength); + } else { + head |= (LENGTH_IN_2TRAIL + (oldLength >> 30)) << 6; + array[limit++] = (uint16_t)(0x8000 | (oldLength >> 15)); + array[limit++] = (uint16_t)(0x8000 | oldLength); + } + if(newLength < LENGTH_IN_1TRAIL) { + head |= newLength; + } else if(newLength <= 0x7fff) { + head |= LENGTH_IN_1TRAIL; + array[limit++] = (uint16_t)(0x8000 | newLength); + } else { + head |= LENGTH_IN_2TRAIL + (newLength >> 30); + array[limit++] = (uint16_t)(0x8000 | (newLength >> 15)); + array[limit++] = (uint16_t)(0x8000 | newLength); + } + array[length] = (uint16_t)head; + length = limit; + } +} + +void Edits::append(int32_t r) { + if(length < capacity || growArray()) { + array[length++] = (uint16_t)r; + } +} + +UBool Edits::growArray() { + int32_t newCapacity; + if (array == stackArray) { + newCapacity = 2000; + } else if (capacity == INT32_MAX) { + // Not U_BUFFER_OVERFLOW_ERROR because that could be confused on a string transform API + // with a result-string-buffer overflow. + errorCode_ = U_INDEX_OUTOFBOUNDS_ERROR; + return false; + } else if (capacity >= (INT32_MAX / 2)) { + newCapacity = INT32_MAX; + } else { + newCapacity = 2 * capacity; + } + // Grow by at least 5 units so that a maximal change record will fit. + if ((newCapacity - capacity) < 5) { + errorCode_ = U_INDEX_OUTOFBOUNDS_ERROR; + return false; + } + uint16_t *newArray = (uint16_t *)uprv_malloc((size_t)newCapacity * 2); + if (newArray == nullptr) { + errorCode_ = U_MEMORY_ALLOCATION_ERROR; + return false; + } + uprv_memcpy(newArray, array, (size_t)length * 2); + releaseArray(); + array = newArray; + capacity = newCapacity; + return true; +} + +UBool Edits::copyErrorTo(UErrorCode &outErrorCode) const { + if (U_FAILURE(outErrorCode)) { return true; } + if (U_SUCCESS(errorCode_)) { return false; } + outErrorCode = errorCode_; + return true; +} + +Edits &Edits::mergeAndAppend(const Edits &ab, const Edits &bc, UErrorCode &errorCode) { + if (copyErrorTo(errorCode)) { return *this; } + // Picture string a --(Edits ab)--> string b --(Edits bc)--> string c. + // Parallel iteration over both Edits. + Iterator abIter = ab.getFineIterator(); + Iterator bcIter = bc.getFineIterator(); + UBool abHasNext = true, bcHasNext = true; + // Copy iterator state into local variables, so that we can modify and subdivide spans. + // ab old & new length, bc old & new length + int32_t aLength = 0, ab_bLength = 0, bc_bLength = 0, cLength = 0; + // When we have different-intermediate-length changes, we accumulate a larger change. + int32_t pending_aLength = 0, pending_cLength = 0; + for (;;) { + // At this point, for each of the two iterators: + // Either we are done with the locally cached current edit, + // and its intermediate-string length has been reset, + // or we will continue to work with a truncated remainder of this edit. + // + // If the current edit is done, and the iterator has not yet reached the end, + // then we fetch the next edit. This is true for at least one of the iterators. + // + // Normally it does not matter whether we fetch from ab and then bc or vice versa. + // However, the result is observably different when + // ab deletions meet bc insertions at the same intermediate-string index. + // Some users expect the bc insertions to come first, so we fetch from bc first. + if (bc_bLength == 0) { + if (bcHasNext && (bcHasNext = bcIter.next(errorCode)) != 0) { + bc_bLength = bcIter.oldLength(); + cLength = bcIter.newLength(); + if (bc_bLength == 0) { + // insertion + if (ab_bLength == 0 || !abIter.hasChange()) { + addReplace(pending_aLength, pending_cLength + cLength); + pending_aLength = pending_cLength = 0; + } else { + pending_cLength += cLength; + } + continue; + } + } + // else see if the other iterator is done, too. + } + if (ab_bLength == 0) { + if (abHasNext && (abHasNext = abIter.next(errorCode)) != 0) { + aLength = abIter.oldLength(); + ab_bLength = abIter.newLength(); + if (ab_bLength == 0) { + // deletion + if (bc_bLength == bcIter.oldLength() || !bcIter.hasChange()) { + addReplace(pending_aLength + aLength, pending_cLength); + pending_aLength = pending_cLength = 0; + } else { + pending_aLength += aLength; + } + continue; + } + } else if (bc_bLength == 0) { + // Both iterators are done at the same time: + // The intermediate-string lengths match. + break; + } else { + // The ab output string is shorter than the bc input string. + if (!copyErrorTo(errorCode)) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + } + return *this; + } + } + if (bc_bLength == 0) { + // The bc input string is shorter than the ab output string. + if (!copyErrorTo(errorCode)) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + } + return *this; + } + // Done fetching: ab_bLength > 0 && bc_bLength > 0 + + // The current state has two parts: + // - Past: We accumulate a longer ac edit in the "pending" variables. + // - Current: We have copies of the current ab/bc edits in local variables. + // At least one side is newly fetched. + // One side might be a truncated remainder of an edit we fetched earlier. + + if (!abIter.hasChange() && !bcIter.hasChange()) { + // An unchanged span all the way from string a to string c. + if (pending_aLength != 0 || pending_cLength != 0) { + addReplace(pending_aLength, pending_cLength); + pending_aLength = pending_cLength = 0; + } + int32_t unchangedLength = aLength <= cLength ? aLength : cLength; + addUnchanged(unchangedLength); + ab_bLength = aLength -= unchangedLength; + bc_bLength = cLength -= unchangedLength; + // At least one of the unchanged spans is now empty. + continue; + } + if (!abIter.hasChange() && bcIter.hasChange()) { + // Unchanged a->b but changed b->c. + if (ab_bLength >= bc_bLength) { + // Split the longer unchanged span into change + remainder. + addReplace(pending_aLength + bc_bLength, pending_cLength + cLength); + pending_aLength = pending_cLength = 0; + aLength = ab_bLength -= bc_bLength; + bc_bLength = 0; + continue; + } + // Handle the shorter unchanged span below like a change. + } else if (abIter.hasChange() && !bcIter.hasChange()) { + // Changed a->b and then unchanged b->c. + if (ab_bLength <= bc_bLength) { + // Split the longer unchanged span into change + remainder. + addReplace(pending_aLength + aLength, pending_cLength + ab_bLength); + pending_aLength = pending_cLength = 0; + cLength = bc_bLength -= ab_bLength; + ab_bLength = 0; + continue; + } + // Handle the shorter unchanged span below like a change. + } else { // both abIter.hasChange() && bcIter.hasChange() + if (ab_bLength == bc_bLength) { + // Changes on both sides up to the same position. Emit & reset. + addReplace(pending_aLength + aLength, pending_cLength + cLength); + pending_aLength = pending_cLength = 0; + ab_bLength = bc_bLength = 0; + continue; + } + } + // Accumulate the a->c change, reset the shorter side, + // keep a remainder of the longer one. + pending_aLength += aLength; + pending_cLength += cLength; + if (ab_bLength < bc_bLength) { + bc_bLength -= ab_bLength; + cLength = ab_bLength = 0; + } else { // ab_bLength > bc_bLength + ab_bLength -= bc_bLength; + aLength = bc_bLength = 0; + } + } + if (pending_aLength != 0 || pending_cLength != 0) { + addReplace(pending_aLength, pending_cLength); + } + copyErrorTo(errorCode); + return *this; +} + +Edits::Iterator::Iterator(const uint16_t *a, int32_t len, UBool oc, UBool crs) : + array(a), index(0), length(len), remaining(0), + onlyChanges_(oc), coarse(crs), + dir(0), changed(false), oldLength_(0), newLength_(0), + srcIndex(0), replIndex(0), destIndex(0) {} + +int32_t Edits::Iterator::readLength(int32_t head) { + if (head < LENGTH_IN_1TRAIL) { + return head; + } else if (head < LENGTH_IN_2TRAIL) { + U_ASSERT(index < length); + U_ASSERT(array[index] >= 0x8000); + return array[index++] & 0x7fff; + } else { + U_ASSERT((index + 2) <= length); + U_ASSERT(array[index] >= 0x8000); + U_ASSERT(array[index + 1] >= 0x8000); + int32_t len = ((head & 1) << 30) | + ((int32_t)(array[index] & 0x7fff) << 15) | + (array[index + 1] & 0x7fff); + index += 2; + return len; + } +} + +void Edits::Iterator::updateNextIndexes() { + srcIndex += oldLength_; + if (changed) { + replIndex += newLength_; + } + destIndex += newLength_; +} + +void Edits::Iterator::updatePreviousIndexes() { + srcIndex -= oldLength_; + if (changed) { + replIndex -= newLength_; + } + destIndex -= newLength_; +} + +UBool Edits::Iterator::noNext() { + // No change before or beyond the string. + dir = 0; + changed = false; + oldLength_ = newLength_ = 0; + return false; +} + +UBool Edits::Iterator::next(UBool onlyChanges, UErrorCode &errorCode) { + // Forward iteration: Update the string indexes to the limit of the current span, + // and post-increment-read array units to assemble a new span. + // Leaves the array index one after the last unit of that span. + if (U_FAILURE(errorCode)) { return false; } + // We have an errorCode in case we need to start guarding against integer overflows. + // It is also convenient for caller loops if we bail out when an error was set elsewhere. + if (dir > 0) { + updateNextIndexes(); + } else { + if (dir < 0) { + // Turn around from previous() to next(). + // Post-increment-read the same span again. + if (remaining > 0) { + // Fine-grained iterator: + // Stay on the current one of a sequence of compressed changes. + ++index; // next() rests on the index after the sequence unit. + dir = 1; + return true; + } + } + dir = 1; + } + if (remaining >= 1) { + // Fine-grained iterator: Continue a sequence of compressed changes. + if (remaining > 1) { + --remaining; + return true; + } + remaining = 0; + } + if (index >= length) { + return noNext(); + } + int32_t u = array[index++]; + if (u <= MAX_UNCHANGED) { + // Combine adjacent unchanged ranges. + changed = false; + oldLength_ = u + 1; + while (index < length && (u = array[index]) <= MAX_UNCHANGED) { + ++index; + oldLength_ += u + 1; + } + newLength_ = oldLength_; + if (onlyChanges) { + updateNextIndexes(); + if (index >= length) { + return noNext(); + } + // already fetched u > MAX_UNCHANGED at index + ++index; + } else { + return true; + } + } + changed = true; + if (u <= MAX_SHORT_CHANGE) { + int32_t oldLen = u >> 12; + int32_t newLen = (u >> 9) & MAX_SHORT_CHANGE_NEW_LENGTH; + int32_t num = (u & SHORT_CHANGE_NUM_MASK) + 1; + if (coarse) { + oldLength_ = num * oldLen; + newLength_ = num * newLen; + } else { + // Split a sequence of changes that was compressed into one unit. + oldLength_ = oldLen; + newLength_ = newLen; + if (num > 1) { + remaining = num; // This is the first of two or more changes. + } + return true; + } + } else { + U_ASSERT(u <= 0x7fff); + oldLength_ = readLength((u >> 6) & 0x3f); + newLength_ = readLength(u & 0x3f); + if (!coarse) { + return true; + } + } + // Combine adjacent changes. + while (index < length && (u = array[index]) > MAX_UNCHANGED) { + ++index; + if (u <= MAX_SHORT_CHANGE) { + int32_t num = (u & SHORT_CHANGE_NUM_MASK) + 1; + oldLength_ += (u >> 12) * num; + newLength_ += ((u >> 9) & MAX_SHORT_CHANGE_NEW_LENGTH) * num; + } else { + U_ASSERT(u <= 0x7fff); + oldLength_ += readLength((u >> 6) & 0x3f); + newLength_ += readLength(u & 0x3f); + } + } + return true; +} + +UBool Edits::Iterator::previous(UErrorCode &errorCode) { + // Backward iteration: Pre-decrement-read array units to assemble a new span, + // then update the string indexes to the start of that span. + // Leaves the array index on the head unit of that span. + if (U_FAILURE(errorCode)) { return false; } + // We have an errorCode in case we need to start guarding against integer overflows. + // It is also convenient for caller loops if we bail out when an error was set elsewhere. + if (dir >= 0) { + if (dir > 0) { + // Turn around from next() to previous(). + // Set the string indexes to the span limit and + // pre-decrement-read the same span again. + if (remaining > 0) { + // Fine-grained iterator: + // Stay on the current one of a sequence of compressed changes. + --index; // previous() rests on the sequence unit. + dir = -1; + return true; + } + updateNextIndexes(); + } + dir = -1; + } + if (remaining > 0) { + // Fine-grained iterator: Continue a sequence of compressed changes. + int32_t u = array[index]; + U_ASSERT(MAX_UNCHANGED < u && u <= MAX_SHORT_CHANGE); + if (remaining <= (u & SHORT_CHANGE_NUM_MASK)) { + ++remaining; + updatePreviousIndexes(); + return true; + } + remaining = 0; + } + if (index <= 0) { + return noNext(); + } + int32_t u = array[--index]; + if (u <= MAX_UNCHANGED) { + // Combine adjacent unchanged ranges. + changed = false; + oldLength_ = u + 1; + while (index > 0 && (u = array[index - 1]) <= MAX_UNCHANGED) { + --index; + oldLength_ += u + 1; + } + newLength_ = oldLength_; + // No need to handle onlyChanges as long as previous() is called only from findIndex(). + updatePreviousIndexes(); + return true; + } + changed = true; + if (u <= MAX_SHORT_CHANGE) { + int32_t oldLen = u >> 12; + int32_t newLen = (u >> 9) & MAX_SHORT_CHANGE_NEW_LENGTH; + int32_t num = (u & SHORT_CHANGE_NUM_MASK) + 1; + if (coarse) { + oldLength_ = num * oldLen; + newLength_ = num * newLen; + } else { + // Split a sequence of changes that was compressed into one unit. + oldLength_ = oldLen; + newLength_ = newLen; + if (num > 1) { + remaining = 1; // This is the last of two or more changes. + } + updatePreviousIndexes(); + return true; + } + } else { + if (u <= 0x7fff) { + // The change is encoded in u alone. + oldLength_ = readLength((u >> 6) & 0x3f); + newLength_ = readLength(u & 0x3f); + } else { + // Back up to the head of the change, read the lengths, + // and reset the index to the head again. + U_ASSERT(index > 0); + while ((u = array[--index]) > 0x7fff) {} + U_ASSERT(u > MAX_SHORT_CHANGE); + int32_t headIndex = index++; + oldLength_ = readLength((u >> 6) & 0x3f); + newLength_ = readLength(u & 0x3f); + index = headIndex; + } + if (!coarse) { + updatePreviousIndexes(); + return true; + } + } + // Combine adjacent changes. + while (index > 0 && (u = array[index - 1]) > MAX_UNCHANGED) { + --index; + if (u <= MAX_SHORT_CHANGE) { + int32_t num = (u & SHORT_CHANGE_NUM_MASK) + 1; + oldLength_ += (u >> 12) * num; + newLength_ += ((u >> 9) & MAX_SHORT_CHANGE_NEW_LENGTH) * num; + } else if (u <= 0x7fff) { + // Read the lengths, and reset the index to the head again. + int32_t headIndex = index++; + oldLength_ += readLength((u >> 6) & 0x3f); + newLength_ += readLength(u & 0x3f); + index = headIndex; + } + } + updatePreviousIndexes(); + return true; +} + +int32_t Edits::Iterator::findIndex(int32_t i, UBool findSource, UErrorCode &errorCode) { + if (U_FAILURE(errorCode) || i < 0) { return -1; } + int32_t spanStart, spanLength; + if (findSource) { // find source index + spanStart = srcIndex; + spanLength = oldLength_; + } else { // find destination index + spanStart = destIndex; + spanLength = newLength_; + } + if (i < spanStart) { + if (i >= (spanStart / 2)) { + // Search backwards. + for (;;) { + UBool hasPrevious = previous(errorCode); + U_ASSERT(hasPrevious); // because i>=0 and the first span starts at 0 + (void)hasPrevious; // avoid unused-variable warning + spanStart = findSource ? srcIndex : destIndex; + if (i >= spanStart) { + // The index is in the current span. + return 0; + } + if (remaining > 0) { + // Is the index in one of the remaining compressed edits? + // spanStart is the start of the current span, first of the remaining ones. + spanLength = findSource ? oldLength_ : newLength_; + int32_t u = array[index]; + U_ASSERT(MAX_UNCHANGED < u && u <= MAX_SHORT_CHANGE); + int32_t num = (u & SHORT_CHANGE_NUM_MASK) + 1 - remaining; + int32_t len = num * spanLength; + if (i >= (spanStart - len)) { + int32_t n = ((spanStart - i - 1) / spanLength) + 1; + // 1 <= n <= num + srcIndex -= n * oldLength_; + replIndex -= n * newLength_; + destIndex -= n * newLength_; + remaining += n; + return 0; + } + // Skip all of these edits at once. + srcIndex -= num * oldLength_; + replIndex -= num * newLength_; + destIndex -= num * newLength_; + remaining = 0; + } + } + } + // Reset the iterator to the start. + dir = 0; + index = remaining = oldLength_ = newLength_ = srcIndex = replIndex = destIndex = 0; + } else if (i < (spanStart + spanLength)) { + // The index is in the current span. + return 0; + } + while (next(false, errorCode)) { + if (findSource) { + spanStart = srcIndex; + spanLength = oldLength_; + } else { + spanStart = destIndex; + spanLength = newLength_; + } + if (i < (spanStart + spanLength)) { + // The index is in the current span. + return 0; + } + if (remaining > 1) { + // Is the index in one of the remaining compressed edits? + // spanStart is the start of the current span, first of the remaining ones. + int32_t len = remaining * spanLength; + if (i < (spanStart + len)) { + int32_t n = (i - spanStart) / spanLength; // 1 <= n <= remaining - 1 + srcIndex += n * oldLength_; + replIndex += n * newLength_; + destIndex += n * newLength_; + remaining -= n; + return 0; + } + // Make next() skip all of these edits at once. + oldLength_ *= remaining; + newLength_ *= remaining; + remaining = 0; + } + } + return 1; +} + +int32_t Edits::Iterator::destinationIndexFromSourceIndex(int32_t i, UErrorCode &errorCode) { + int32_t where = findIndex(i, true, errorCode); + if (where < 0) { + // Error or before the string. + return 0; + } + if (where > 0 || i == srcIndex) { + // At or after string length, or at start of the found span. + return destIndex; + } + if (changed) { + // In a change span, map to its end. + return destIndex + newLength_; + } else { + // In an unchanged span, offset 1:1 within it. + return destIndex + (i - srcIndex); + } +} + +int32_t Edits::Iterator::sourceIndexFromDestinationIndex(int32_t i, UErrorCode &errorCode) { + int32_t where = findIndex(i, false, errorCode); + if (where < 0) { + // Error or before the string. + return 0; + } + if (where > 0 || i == destIndex) { + // At or after string length, or at start of the found span. + return srcIndex; + } + if (changed) { + // In a change span, map to its end. + return srcIndex + oldLength_; + } else { + // In an unchanged span, offset within it. + return srcIndex + (i - destIndex); + } +} + +UnicodeString& Edits::Iterator::toString(UnicodeString& sb) const { + sb.append(u"{ src[", -1); + ICU_Utility::appendNumber(sb, srcIndex); + sb.append(u"..", -1); + ICU_Utility::appendNumber(sb, srcIndex + oldLength_); + if (changed) { + sb.append(u"] ⇝ dest[", -1); + } else { + sb.append(u"] ≡ dest[", -1); + } + ICU_Utility::appendNumber(sb, destIndex); + sb.append(u"..", -1); + ICU_Utility::appendNumber(sb, destIndex + newLength_); + if (changed) { + sb.append(u"], repl[", -1); + ICU_Utility::appendNumber(sb, replIndex); + sb.append(u"..", -1); + ICU_Utility::appendNumber(sb, replIndex + newLength_); + sb.append(u"] }", -1); + } else { + sb.append(u"] (no-change) }", -1); + } + return sb; +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/emojiprops.cpp b/intl/icu/source/common/emojiprops.cpp new file mode 100644 index 0000000000..f2b3f4762f --- /dev/null +++ b/intl/icu/source/common/emojiprops.cpp @@ -0,0 +1,220 @@ +// © 2021 and later: Unicode, Inc. and others. +// License & terms of use: https://www.unicode.org/copyright.html + +// emojiprops.cpp +// created: 2021sep04 Markus W. Scherer + +#include "unicode/utypes.h" +#include "unicode/uchar.h" +#include "unicode/ucharstrie.h" +#include "unicode/ucptrie.h" +#include "unicode/udata.h" +#include "unicode/ustringtrie.h" +#include "unicode/utf16.h" +#include "emojiprops.h" +#include "ucln.h" +#include "ucln_cmn.h" +#include "umutex.h" +#include "uset_imp.h" + +U_NAMESPACE_BEGIN + +namespace { + +EmojiProps *singleton = nullptr; +icu::UInitOnce emojiInitOnce {}; + +UBool U_CALLCONV emojiprops_cleanup() { + delete singleton; + singleton = nullptr; + emojiInitOnce.reset(); + return true; +} + +void U_CALLCONV initSingleton(UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return; } + singleton = new EmojiProps(errorCode); + if (singleton == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + } else if (U_FAILURE(errorCode)) { + delete singleton; + singleton = nullptr; + } + ucln_common_registerCleanup(UCLN_COMMON_EMOJIPROPS, emojiprops_cleanup); +} + +// TODO: turn this into a shared helper function +// Requires the major version to match, and then requires at least the minor version. +UBool udata_isAcceptableMajorMinor( + const UDataInfo &info, const char16_t *dataFormat, uint8_t major, uint8_t minor) { + return + info.size >= 20 && + info.isBigEndian == U_IS_BIG_ENDIAN && + info.charsetFamily == U_CHARSET_FAMILY && + info.dataFormat[0] == dataFormat[0] && + info.dataFormat[1] == dataFormat[1] && + info.dataFormat[2] == dataFormat[2] && + info.dataFormat[3] == dataFormat[3] && + info.formatVersion[0] == major && + info.formatVersion[1] >= minor; +} + +} // namespace + +EmojiProps::~EmojiProps() { + udata_close(memory); + ucptrie_close(cpTrie); +} + +const EmojiProps * +EmojiProps::getSingleton(UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return nullptr; } + umtx_initOnce(emojiInitOnce, &initSingleton, errorCode); + return singleton; +} + +UBool U_CALLCONV +EmojiProps::isAcceptable(void * /*context*/, const char * /*type*/, const char * /*name*/, + const UDataInfo *pInfo) { + return udata_isAcceptableMajorMinor(*pInfo, u"Emoj", 1, 0); +} + +void +EmojiProps::load(UErrorCode &errorCode) { + memory = udata_openChoice(nullptr, "icu", "uemoji", isAcceptable, this, &errorCode); + if (U_FAILURE(errorCode)) { return; } + const uint8_t *inBytes = (const uint8_t *)udata_getMemory(memory); + const int32_t *inIndexes = (const int32_t *)inBytes; + int32_t indexesLength = inIndexes[IX_CPTRIE_OFFSET] / 4; + if (indexesLength <= IX_RGI_EMOJI_ZWJ_SEQUENCE_TRIE_OFFSET) { + errorCode = U_INVALID_FORMAT_ERROR; // Not enough indexes. + return; + } + + int32_t i = IX_CPTRIE_OFFSET; + int32_t offset = inIndexes[i++]; + int32_t nextOffset = inIndexes[i]; + cpTrie = ucptrie_openFromBinary(UCPTRIE_TYPE_FAST, UCPTRIE_VALUE_BITS_8, + inBytes + offset, nextOffset - offset, nullptr, &errorCode); + if (U_FAILURE(errorCode)) { + return; + } + + for (i = IX_BASIC_EMOJI_TRIE_OFFSET; i <= IX_RGI_EMOJI_ZWJ_SEQUENCE_TRIE_OFFSET; ++i) { + offset = inIndexes[i]; + nextOffset = inIndexes[i + 1]; + // Set/leave nullptr if there is no UCharsTrie. + const char16_t *p = nextOffset > offset ? (const char16_t *)(inBytes + offset) : nullptr; + stringTries[getStringTrieIndex(i)] = p; + } +} + +void +EmojiProps::addPropertyStarts(const USetAdder *sa, UErrorCode & /*errorCode*/) const { + // Add the start code point of each same-value range of the trie. + UChar32 start = 0, end; + uint32_t value; + while ((end = ucptrie_getRange(cpTrie, start, UCPMAP_RANGE_NORMAL, 0, + nullptr, nullptr, &value)) >= 0) { + sa->add(sa->set, start); + start = end + 1; + } +} + +UBool +EmojiProps::hasBinaryProperty(UChar32 c, UProperty which) { + UErrorCode errorCode = U_ZERO_ERROR; + const EmojiProps *ep = getSingleton(errorCode); + return U_SUCCESS(errorCode) && ep->hasBinaryPropertyImpl(c, which); +} + +UBool +EmojiProps::hasBinaryPropertyImpl(UChar32 c, UProperty which) const { + if (which < UCHAR_EMOJI || UCHAR_RGI_EMOJI < which) { + return false; + } + // Note: UCHAR_REGIONAL_INDICATOR is a single, hardcoded range implemented elsewhere. + static constexpr int8_t bitFlags[] = { + BIT_EMOJI, // UCHAR_EMOJI=57 + BIT_EMOJI_PRESENTATION, // UCHAR_EMOJI_PRESENTATION=58 + BIT_EMOJI_MODIFIER, // UCHAR_EMOJI_MODIFIER=59 + BIT_EMOJI_MODIFIER_BASE, // UCHAR_EMOJI_MODIFIER_BASE=60 + BIT_EMOJI_COMPONENT, // UCHAR_EMOJI_COMPONENT=61 + -1, // UCHAR_REGIONAL_INDICATOR=62 + -1, // UCHAR_PREPENDED_CONCATENATION_MARK=63 + BIT_EXTENDED_PICTOGRAPHIC, // UCHAR_EXTENDED_PICTOGRAPHIC=64 + BIT_BASIC_EMOJI, // UCHAR_BASIC_EMOJI=65 + -1, // UCHAR_EMOJI_KEYCAP_SEQUENCE=66 + -1, // UCHAR_RGI_EMOJI_MODIFIER_SEQUENCE=67 + -1, // UCHAR_RGI_EMOJI_FLAG_SEQUENCE=68 + -1, // UCHAR_RGI_EMOJI_TAG_SEQUENCE=69 + -1, // UCHAR_RGI_EMOJI_ZWJ_SEQUENCE=70 + BIT_BASIC_EMOJI, // UCHAR_RGI_EMOJI=71 + }; + int32_t bit = bitFlags[which - UCHAR_EMOJI]; + if (bit < 0) { + return false; // not a property that we support in this function + } + uint8_t bits = UCPTRIE_FAST_GET(cpTrie, UCPTRIE_8, c); + return (bits >> bit) & 1; +} + +UBool +EmojiProps::hasBinaryProperty(const char16_t *s, int32_t length, UProperty which) { + UErrorCode errorCode = U_ZERO_ERROR; + const EmojiProps *ep = getSingleton(errorCode); + return U_SUCCESS(errorCode) && ep->hasBinaryPropertyImpl(s, length, which); +} + +UBool +EmojiProps::hasBinaryPropertyImpl(const char16_t *s, int32_t length, UProperty which) const { + if (s == nullptr && length != 0) { return false; } + if (length <= 0 && (length == 0 || *s == 0)) { return false; } // empty string + // The caller should have delegated single code points to hasBinaryProperty(c, which). + if (which < UCHAR_BASIC_EMOJI || UCHAR_RGI_EMOJI < which) { + return false; + } + UProperty firstProp = which, lastProp = which; + if (which == UCHAR_RGI_EMOJI) { + // RGI_Emoji is the union of the other emoji properties of strings. + firstProp = UCHAR_BASIC_EMOJI; + lastProp = UCHAR_RGI_EMOJI_ZWJ_SEQUENCE; + } + for (int32_t prop = firstProp; prop <= lastProp; ++prop) { + const char16_t *trieUChars = stringTries[prop - UCHAR_BASIC_EMOJI]; + if (trieUChars != nullptr) { + UCharsTrie trie(trieUChars); + UStringTrieResult result = trie.next(s, length); + if (USTRINGTRIE_HAS_VALUE(result)) { + return true; + } + } + } + return false; +} + +void +EmojiProps::addStrings(const USetAdder *sa, UProperty which, UErrorCode &errorCode) const { + if (U_FAILURE(errorCode)) { return; } + if (which < UCHAR_BASIC_EMOJI || UCHAR_RGI_EMOJI < which) { + return; + } + UProperty firstProp = which, lastProp = which; + if (which == UCHAR_RGI_EMOJI) { + // RGI_Emoji is the union of the other emoji properties of strings. + firstProp = UCHAR_BASIC_EMOJI; + lastProp = UCHAR_RGI_EMOJI_ZWJ_SEQUENCE; + } + for (int32_t prop = firstProp; prop <= lastProp; ++prop) { + const char16_t *trieUChars = stringTries[prop - UCHAR_BASIC_EMOJI]; + if (trieUChars != nullptr) { + UCharsTrie::Iterator iter(trieUChars, 0, errorCode); + while (iter.next(errorCode)) { + const UnicodeString &s = iter.getString(); + sa->addString(sa->set, s.getBuffer(), s.length()); + } + } + } +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/emojiprops.h b/intl/icu/source/common/emojiprops.h new file mode 100644 index 0000000000..777b7b11a8 --- /dev/null +++ b/intl/icu/source/common/emojiprops.h @@ -0,0 +1,90 @@ +// © 2021 and later: Unicode, Inc. and others. +// License & terms of use: https://www.unicode.org/copyright.html + +// emojiprops.h +// created: 2021sep03 Markus W. Scherer + +#ifndef __EMOJIPROPS_H__ +#define __EMOJIPROPS_H__ + +#include "unicode/utypes.h" +#include "unicode/ucptrie.h" +#include "unicode/udata.h" +#include "unicode/uobject.h" +#include "uset_imp.h" + +U_NAMESPACE_BEGIN + +class EmojiProps : public UMemory { +public: + // @internal + EmojiProps(UErrorCode &errorCode) { load(errorCode); } + ~EmojiProps(); + + static const EmojiProps *getSingleton(UErrorCode &errorCode); + static UBool hasBinaryProperty(UChar32 c, UProperty which); + static UBool hasBinaryProperty(const char16_t *s, int32_t length, UProperty which); + + void addPropertyStarts(const USetAdder *sa, UErrorCode &errorCode) const; + void addStrings(const USetAdder *sa, UProperty which, UErrorCode &errorCode) const; + + enum { + // Byte offsets from the start of the data, after the generic header, + // in ascending order. + // UCPTrie=CodePointTrie, follows the indexes + IX_CPTRIE_OFFSET, + IX_RESERVED1, + IX_RESERVED2, + IX_RESERVED3, + + // UCharsTrie=CharsTrie + IX_BASIC_EMOJI_TRIE_OFFSET, + IX_EMOJI_KEYCAP_SEQUENCE_TRIE_OFFSET, + IX_RGI_EMOJI_MODIFIER_SEQUENCE_TRIE_OFFSET, + IX_RGI_EMOJI_FLAG_SEQUENCE_TRIE_OFFSET, + IX_RGI_EMOJI_TAG_SEQUENCE_TRIE_OFFSET, + IX_RGI_EMOJI_ZWJ_SEQUENCE_TRIE_OFFSET, + IX_RESERVED10, + IX_RESERVED11, + IX_RESERVED12, + IX_TOTAL_SIZE, + + // Not initially byte offsets. + IX_RESERVED14, + IX_RESERVED15, + IX_COUNT // 16 + }; + + // Properties in the code point trie. + enum { + // https://www.unicode.org/reports/tr51/#Emoji_Properties + BIT_EMOJI, + BIT_EMOJI_PRESENTATION, + BIT_EMOJI_MODIFIER, + BIT_EMOJI_MODIFIER_BASE, + BIT_EMOJI_COMPONENT, + BIT_EXTENDED_PICTOGRAPHIC, + // https://www.unicode.org/reports/tr51/#Emoji_Sets + BIT_BASIC_EMOJI + }; + +private: + static UBool U_CALLCONV + isAcceptable(void *context, const char *type, const char *name, const UDataInfo *pInfo); + /** Input i: One of the IX_..._TRIE_OFFSET indexes into the data file indexes[] array. */ + static int32_t getStringTrieIndex(int32_t i) { + return i - IX_BASIC_EMOJI_TRIE_OFFSET; + } + + void load(UErrorCode &errorCode); + UBool hasBinaryPropertyImpl(UChar32 c, UProperty which) const; + UBool hasBinaryPropertyImpl(const char16_t *s, int32_t length, UProperty which) const; + + UDataMemory *memory = nullptr; + UCPTrie *cpTrie = nullptr; + const char16_t *stringTries[6] = { nullptr, nullptr, nullptr, nullptr, nullptr, nullptr }; +}; + +U_NAMESPACE_END + +#endif // __EMOJIPROPS_H__ diff --git a/intl/icu/source/common/errorcode.cpp b/intl/icu/source/common/errorcode.cpp new file mode 100644 index 0000000000..e7ac43b527 --- /dev/null +++ b/intl/icu/source/common/errorcode.cpp @@ -0,0 +1,42 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2009-2011, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: errorcode.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2009mar10 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" +#include "unicode/errorcode.h" + +U_NAMESPACE_BEGIN + +ErrorCode::~ErrorCode() {} + +UErrorCode ErrorCode::reset() { + UErrorCode code = errorCode; + errorCode = U_ZERO_ERROR; + return code; +} + +void ErrorCode::assertSuccess() const { + if(isFailure()) { + handleFailure(); + } +} + +const char* ErrorCode::errorName() const { + return u_errorName(errorCode); +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/filteredbrk.cpp b/intl/icu/source/common/filteredbrk.cpp new file mode 100644 index 0000000000..a705b037c4 --- /dev/null +++ b/intl/icu/source/common/filteredbrk.cpp @@ -0,0 +1,736 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2014-2015, International Business Machines Corporation and +* others. All Rights Reserved. +******************************************************************************* +*/ + +#include "unicode/utypes.h" +#if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION + +#include "cmemory.h" + +#include "unicode/filteredbrk.h" +#include "unicode/ucharstriebuilder.h" +#include "unicode/ures.h" + +#include "uresimp.h" // ures_getByKeyWithFallback +#include "ubrkimpl.h" // U_ICUDATA_BRKITR +#include "uvector.h" +#include "cmemory.h" +#include "umutex.h" + +U_NAMESPACE_BEGIN + +#ifndef FB_DEBUG +#define FB_DEBUG 0 +#endif + +#if FB_DEBUG +#include +static void _fb_trace(const char *m, const UnicodeString *s, UBool b, int32_t d, const char *f, int l) { + char buf[2048]; + if(s) { + s->extract(0,s->length(),buf,2048); + } else { + strcpy(buf,"nullptr"); + } + fprintf(stderr,"%s:%d: %s. s='%s'(%p), b=%c, d=%d\n", + f, l, m, buf, (const void*)s, b?'T':'F',(int)d); +} + +#define FB_TRACE(m,s,b,d) _fb_trace(m,s,b,d,__FILE__,__LINE__) +#else +#define FB_TRACE(m,s,b,d) +#endif + +/** + * Used with sortedInsert() + */ +static int32_t U_CALLCONV compareUnicodeString(UElement t1, UElement t2) { + const UnicodeString &a = *(const UnicodeString*)t1.pointer; + const UnicodeString &b = *(const UnicodeString*)t2.pointer; + return a.compare(b); +} + +/** + * A UVector which implements a set of strings. + */ +class UStringSet : public UVector { + public: + UStringSet(UErrorCode &status) : UVector(uprv_deleteUObject, + uhash_compareUnicodeString, + 1, + status) {} + virtual ~UStringSet(); + /** + * Is this UnicodeSet contained? + */ + inline UBool contains(const UnicodeString& s) { + return contains((void*) &s); + } + using UVector::contains; + /** + * Return the ith UnicodeString alias + */ + inline const UnicodeString* getStringAt(int32_t i) const { + return (const UnicodeString*)elementAt(i); + } + /** + * Adopt the UnicodeString if not already contained. + * Caller no longer owns the pointer in any case. + * @return true if adopted successfully, false otherwise (error, or else duplicate) + */ + inline UBool adopt(UnicodeString *str, UErrorCode &status) { + if(U_FAILURE(status) || contains(*str)) { + delete str; + return false; + } else { + sortedInsert(str, compareUnicodeString, status); + if(U_FAILURE(status)) { + return false; + } + return true; + } + } + /** + * Add by value. + * @return true if successfully adopted. + */ + inline UBool add(const UnicodeString& str, UErrorCode &status) { + if(U_FAILURE(status)) return false; + UnicodeString *t = new UnicodeString(str); + if(t==nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; return false; + } + return adopt(t, status); + } + /** + * Remove this string. + * @return true if successfully removed, false otherwise (error, or else it wasn't there) + */ + inline UBool remove(const UnicodeString &s, UErrorCode &status) { + if(U_FAILURE(status)) return false; + return removeElement((void*) &s); + } +}; + +/** + * Virtual, won't be inlined + */ +UStringSet::~UStringSet() {} + +/* ----------------------------------------------------------- */ + + +/* Filtered Break constants */ +static const int32_t kPARTIAL = (1<<0); //< partial - need to run through forward trie +static const int32_t kMATCH = (1<<1); //< exact match - skip this one. +static const int32_t kSuppressInReverse = (1<<0); +static const int32_t kAddToForward = (1<<1); +static const char16_t kFULLSTOP = 0x002E; // '.' + +/** + * Shared data for SimpleFilteredSentenceBreakIterator + */ +class SimpleFilteredSentenceBreakData : public UMemory { +public: + SimpleFilteredSentenceBreakData(UCharsTrie *forwards, UCharsTrie *backwards ) + : fForwardsPartialTrie(forwards), fBackwardsTrie(backwards), refcount(1) { } + SimpleFilteredSentenceBreakData *incr() { + umtx_atomic_inc(&refcount); + return this; + } + SimpleFilteredSentenceBreakData *decr() { + if(umtx_atomic_dec(&refcount) <= 0) { + delete this; + } + return 0; + } + virtual ~SimpleFilteredSentenceBreakData(); + + bool hasForwardsPartialTrie() const { return fForwardsPartialTrie.isValid(); } + bool hasBackwardsTrie() const { return fBackwardsTrie.isValid(); } + + const UCharsTrie &getForwardsPartialTrie() const { return *fForwardsPartialTrie; } + const UCharsTrie &getBackwardsTrie() const { return *fBackwardsTrie; } + +private: + // These tries own their data arrays. + // They are shared and must therefore not be modified. + LocalPointer fForwardsPartialTrie; // Has ".a" for "a.M." + LocalPointer fBackwardsTrie; // i.e. ".srM" for Mrs. + u_atomic_int32_t refcount; +}; + +SimpleFilteredSentenceBreakData::~SimpleFilteredSentenceBreakData() {} + +/** + * Concrete implementation + */ +class SimpleFilteredSentenceBreakIterator : public BreakIterator { +public: + SimpleFilteredSentenceBreakIterator(BreakIterator *adopt, UCharsTrie *forwards, UCharsTrie *backwards, UErrorCode &status); + SimpleFilteredSentenceBreakIterator(const SimpleFilteredSentenceBreakIterator& other); + virtual ~SimpleFilteredSentenceBreakIterator(); +private: + SimpleFilteredSentenceBreakData *fData; + LocalPointer fDelegate; + LocalUTextPointer fText; + + /* -- subclass interface -- */ +public: + /* -- cloning and other subclass stuff -- */ + virtual BreakIterator * createBufferClone(void * /*stackBuffer*/, + int32_t &/*BufferSize*/, + UErrorCode &status) override { + // for now - always deep clone + status = U_SAFECLONE_ALLOCATED_WARNING; + return clone(); + } + virtual SimpleFilteredSentenceBreakIterator* clone() const override { return new SimpleFilteredSentenceBreakIterator(*this); } + virtual UClassID getDynamicClassID() const override { return nullptr; } + virtual bool operator==(const BreakIterator& o) const override { if(this==&o) return true; return false; } + + /* -- text modifying -- */ + virtual void setText(UText *text, UErrorCode &status) override { fDelegate->setText(text,status); } + virtual BreakIterator &refreshInputText(UText *input, UErrorCode &status) override { fDelegate->refreshInputText(input,status); return *this; } + virtual void adoptText(CharacterIterator* it) override { fDelegate->adoptText(it); } + virtual void setText(const UnicodeString &text) override { fDelegate->setText(text); } + + /* -- other functions that are just delegated -- */ + virtual UText *getUText(UText *fillIn, UErrorCode &status) const override { return fDelegate->getUText(fillIn,status); } + virtual CharacterIterator& getText() const override { return fDelegate->getText(); } + + /* -- ITERATION -- */ + virtual int32_t first() override; + virtual int32_t preceding(int32_t offset) override; + virtual int32_t previous() override; + virtual UBool isBoundary(int32_t offset) override; + virtual int32_t current() const override { return fDelegate->current(); } // we keep the delegate current, so this should be correct. + + virtual int32_t next() override; + + virtual int32_t next(int32_t n) override; + virtual int32_t following(int32_t offset) override; + virtual int32_t last() override; + +private: + /** + * Given that the fDelegate has already given its "initial" answer, + * find the NEXT actual (non-excepted) break. + * @param n initial position from delegate + * @return new break position or UBRK_DONE + */ + int32_t internalNext(int32_t n); + /** + * Given that the fDelegate has already given its "initial" answer, + * find the PREV actual (non-excepted) break. + * @param n initial position from delegate + * @return new break position or UBRK_DONE + */ + int32_t internalPrev(int32_t n); + /** + * set up the UText with the value of the fDelegate. + * Call this before calling breakExceptionAt. + * May be able to avoid excess calls + */ + void resetState(UErrorCode &status); + /** + * Is there a match (exception) at this spot? + */ + enum EFBMatchResult { kNoExceptionHere, kExceptionHere }; + /** + * Determine if there is an exception at this spot + * @param n spot to check + * @return kNoExceptionHere or kExceptionHere + **/ + enum EFBMatchResult breakExceptionAt(int32_t n); +}; + +SimpleFilteredSentenceBreakIterator::SimpleFilteredSentenceBreakIterator(const SimpleFilteredSentenceBreakIterator& other) + : BreakIterator(other), fData(other.fData->incr()), fDelegate(other.fDelegate->clone()) +{ +} + + +SimpleFilteredSentenceBreakIterator::SimpleFilteredSentenceBreakIterator(BreakIterator *adopt, UCharsTrie *forwards, UCharsTrie *backwards, UErrorCode &status) : + BreakIterator(adopt->getLocale(ULOC_VALID_LOCALE,status),adopt->getLocale(ULOC_ACTUAL_LOCALE,status)), + fData(new SimpleFilteredSentenceBreakData(forwards, backwards)), + fDelegate(adopt) +{ + if (fData == nullptr) { + delete forwards; + delete backwards; + if (U_SUCCESS(status)) { + status = U_MEMORY_ALLOCATION_ERROR; + } + } +} + +SimpleFilteredSentenceBreakIterator::~SimpleFilteredSentenceBreakIterator() { + fData = fData->decr(); +} + +void SimpleFilteredSentenceBreakIterator::resetState(UErrorCode &status) { + fText.adoptInstead(fDelegate->getUText(fText.orphan(), status)); +} + +SimpleFilteredSentenceBreakIterator::EFBMatchResult +SimpleFilteredSentenceBreakIterator::breakExceptionAt(int32_t n) { + int64_t bestPosn = -1; + int32_t bestValue = -1; + // loops while 'n' points to an exception. + utext_setNativeIndex(fText.getAlias(), n); // from n.. + + //if(debug2) u_printf(" n@ %d\n", n); + // Assume a space is following the '.' (so we handle the case: "Mr. /Brown") + if(utext_previous32(fText.getAlias())==u' ') { // TODO: skip a class of chars here?? + // TODO only do this the 1st time? + //if(debug2) u_printf("skipping prev: |%C| \n", (char16_t)uch); + } else { + //if(debug2) u_printf("not skipping prev: |%C| \n", (char16_t)uch); + utext_next32(fText.getAlias()); + //if(debug2) u_printf(" -> : |%C| \n", (char16_t)uch); + } + + { + // Do not modify the shared trie! + UCharsTrie iter(fData->getBackwardsTrie()); + UChar32 uch; + while((uch=utext_previous32(fText.getAlias()))!=U_SENTINEL) { // more to consume backwards + UStringTrieResult r = iter.nextForCodePoint(uch); + if(USTRINGTRIE_HAS_VALUE(r)) { // remember the best match so far + bestPosn = utext_getNativeIndex(fText.getAlias()); + bestValue = iter.getValue(); + } + if(!USTRINGTRIE_HAS_NEXT(r)) { + break; + } + //if(debug2) u_printf("rev< /%C/ cont?%d @%d\n", (char16_t)uch, r, utext_getNativeIndex(fText.getAlias())); + } + } + + //if(bestValue >= 0) { + //if(debug2) u_printf("rev<+/%C/+end of seq.. r=%d, bestPosn=%d, bestValue=%d\n", (char16_t)uch, r, bestPosn, bestValue); + //} + + if(bestPosn>=0) { + //if(debug2) u_printf("rev< /%C/ end of seq.. r=%d, bestPosn=%d, bestValue=%d\n", (char16_t)uch, r, bestPosn, bestValue); + + //if(USTRINGTRIE_MATCHES(r)) { // matched - so, now what? + //int32_t bestValue = iter.getValue(); + ////if(debug2) u_printf("rev< /%C/ matched, skip..%d bestValue=%d\n", (char16_t)uch, r, bestValue); + + if(bestValue == kMATCH) { // exact match! + //if(debug2) u_printf(" exact backward match\n"); + return kExceptionHere; // See if the next is another exception. + } else if(bestValue == kPARTIAL + && fData->hasForwardsPartialTrie()) { // make sure there's a forward trie + //if(debug2) u_printf(" partial backward match\n"); + // We matched the "Ph." in "Ph.D." - now we need to run everything through the forwards trie + // to see if it matches something going forward. + UStringTrieResult rfwd = USTRINGTRIE_INTERMEDIATE_VALUE; + utext_setNativeIndex(fText.getAlias(), bestPosn); // hope that's close .. + //if(debug2) u_printf("Retrying at %d\n", bestPosn); + // Do not modify the shared trie! + UCharsTrie iter(fData->getForwardsPartialTrie()); + UChar32 uch; + while((uch=utext_next32(fText.getAlias()))!=U_SENTINEL && + USTRINGTRIE_HAS_NEXT(rfwd=iter.nextForCodePoint(uch))) { + //if(debug2) u_printf("fwd> /%C/ cont?%d @%d\n", (char16_t)uch, rfwd, utext_getNativeIndex(fText.getAlias())); + } + if(USTRINGTRIE_MATCHES(rfwd)) { + //if(debug2) u_printf("fwd> /%C/ == forward match!\n", (char16_t)uch); + // only full matches here, nothing to check + // skip the next: + return kExceptionHere; + } else { + //if(debug2) u_printf("fwd> /%C/ no match.\n", (char16_t)uch); + // no match (no exception) -return the 'underlying' break + return kNoExceptionHere; + } + } else { + return kNoExceptionHere; // internal error and/or no forwards trie + } + } else { + //if(debug2) u_printf("rev< /%C/ .. no match..%d\n", (char16_t)uch, r); // no best match + return kNoExceptionHere; // No match - so exit. Not an exception. + } +} + +// the workhorse single next. +int32_t +SimpleFilteredSentenceBreakIterator::internalNext(int32_t n) { + if(n == UBRK_DONE || // at end or + !fData->hasBackwardsTrie()) { // .. no backwards table loaded == no exceptions + return n; + } + // OK, do we need to break here? + UErrorCode status = U_ZERO_ERROR; + // refresh text + resetState(status); + if(U_FAILURE(status)) return UBRK_DONE; // bail out + int64_t utextLen = utext_nativeLength(fText.getAlias()); + + //if(debug2) u_printf("str, native len=%d\n", utext_nativeLength(fText.getAlias())); + while (n != UBRK_DONE && n != utextLen) { // outer loop runs once per underlying break (from fDelegate). + SimpleFilteredSentenceBreakIterator::EFBMatchResult m = breakExceptionAt(n); + + switch(m) { + case kExceptionHere: + n = fDelegate->next(); // skip this one. Find the next lowerlevel break. + continue; + + default: + case kNoExceptionHere: + return n; + } + } + return n; +} + +int32_t +SimpleFilteredSentenceBreakIterator::internalPrev(int32_t n) { + if(n == 0 || n == UBRK_DONE || // at end or + !fData->hasBackwardsTrie()) { // .. no backwards table loaded == no exceptions + return n; + } + // OK, do we need to break here? + UErrorCode status = U_ZERO_ERROR; + // refresh text + resetState(status); + if(U_FAILURE(status)) return UBRK_DONE; // bail out + + //if(debug2) u_printf("str, native len=%d\n", utext_nativeLength(fText.getAlias())); + while (n != UBRK_DONE && n != 0) { // outer loop runs once per underlying break (from fDelegate). + SimpleFilteredSentenceBreakIterator::EFBMatchResult m = breakExceptionAt(n); + + switch(m) { + case kExceptionHere: + n = fDelegate->previous(); // skip this one. Find the next lowerlevel break. + continue; + + default: + case kNoExceptionHere: + return n; + } + } + return n; +} + + +int32_t +SimpleFilteredSentenceBreakIterator::next() { + return internalNext(fDelegate->next()); +} + +int32_t +SimpleFilteredSentenceBreakIterator::first() { + // Don't suppress a break opportunity at the beginning of text. + return fDelegate->first(); +} + +int32_t +SimpleFilteredSentenceBreakIterator::preceding(int32_t offset) { + return internalPrev(fDelegate->preceding(offset)); +} + +int32_t +SimpleFilteredSentenceBreakIterator::previous() { + return internalPrev(fDelegate->previous()); +} + +UBool SimpleFilteredSentenceBreakIterator::isBoundary(int32_t offset) { + if (!fDelegate->isBoundary(offset)) return false; // no break to suppress + + if (!fData->hasBackwardsTrie()) return true; // no data = no suppressions + + UErrorCode status = U_ZERO_ERROR; + resetState(status); + + SimpleFilteredSentenceBreakIterator::EFBMatchResult m = breakExceptionAt(offset); + + switch(m) { + case kExceptionHere: + return false; + default: + case kNoExceptionHere: + return true; + } +} + +int32_t +SimpleFilteredSentenceBreakIterator::next(int32_t offset) { + return internalNext(fDelegate->next(offset)); +} + +int32_t +SimpleFilteredSentenceBreakIterator::following(int32_t offset) { + return internalNext(fDelegate->following(offset)); +} + +int32_t +SimpleFilteredSentenceBreakIterator::last() { + // Don't suppress a break opportunity at the end of text. + return fDelegate->last(); +} + + +/** + * Concrete implementation of builder class. + */ +class SimpleFilteredBreakIteratorBuilder : public FilteredBreakIteratorBuilder { +public: + virtual ~SimpleFilteredBreakIteratorBuilder(); + SimpleFilteredBreakIteratorBuilder(const Locale &fromLocale, UErrorCode &status); + SimpleFilteredBreakIteratorBuilder(UErrorCode &status); + virtual UBool suppressBreakAfter(const UnicodeString& exception, UErrorCode& status) override; + virtual UBool unsuppressBreakAfter(const UnicodeString& exception, UErrorCode& status) override; + virtual BreakIterator *build(BreakIterator* adoptBreakIterator, UErrorCode& status) override; +private: + UStringSet fSet; +}; + +SimpleFilteredBreakIteratorBuilder::~SimpleFilteredBreakIteratorBuilder() +{ +} + +SimpleFilteredBreakIteratorBuilder::SimpleFilteredBreakIteratorBuilder(UErrorCode &status) + : fSet(status) +{ +} + +SimpleFilteredBreakIteratorBuilder::SimpleFilteredBreakIteratorBuilder(const Locale &fromLocale, UErrorCode &status) + : fSet(status) +{ + if(U_SUCCESS(status)) { + UErrorCode subStatus = U_ZERO_ERROR; + LocalUResourceBundlePointer b(ures_open(U_ICUDATA_BRKITR, fromLocale.getBaseName(), &subStatus)); + if (U_FAILURE(subStatus) || (subStatus == U_USING_DEFAULT_WARNING) ) { + status = subStatus; // copy the failing status +#if FB_DEBUG + fprintf(stderr, "open BUNDLE %s : %s, %s\n", fromLocale.getBaseName(), "[exit]", u_errorName(status)); +#endif + return; // leaves the builder empty, if you try to use it. + } + LocalUResourceBundlePointer exceptions(ures_getByKeyWithFallback(b.getAlias(), "exceptions", nullptr, &subStatus)); + if (U_FAILURE(subStatus) || (subStatus == U_USING_DEFAULT_WARNING) ) { + status = subStatus; // copy the failing status +#if FB_DEBUG + fprintf(stderr, "open EXCEPTIONS %s : %s, %s\n", fromLocale.getBaseName(), "[exit]", u_errorName(status)); +#endif + return; // leaves the builder empty, if you try to use it. + } + LocalUResourceBundlePointer breaks(ures_getByKeyWithFallback(exceptions.getAlias(), "SentenceBreak", nullptr, &subStatus)); + +#if FB_DEBUG + { + UErrorCode subsub = subStatus; + fprintf(stderr, "open SentenceBreak %s => %s, %s\n", fromLocale.getBaseName(), ures_getLocale(breaks.getAlias(), &subsub), u_errorName(subStatus)); + } +#endif + + if (U_FAILURE(subStatus) || (subStatus == U_USING_DEFAULT_WARNING) ) { + status = subStatus; // copy the failing status +#if FB_DEBUG + fprintf(stderr, "open %s : %s, %s\n", fromLocale.getBaseName(), "[exit]", u_errorName(status)); +#endif + return; // leaves the builder empty, if you try to use it. + } + + LocalUResourceBundlePointer strs; + subStatus = status; // Pick up inherited warning status now + do { + strs.adoptInstead(ures_getNextResource(breaks.getAlias(), strs.orphan(), &subStatus)); + if(strs.isValid() && U_SUCCESS(subStatus)) { + UnicodeString str(ures_getUnicodeString(strs.getAlias(), &status)); + suppressBreakAfter(str, status); // load the string + } + } while (strs.isValid() && U_SUCCESS(subStatus)); + if(U_FAILURE(subStatus)&&subStatus!=U_INDEX_OUTOFBOUNDS_ERROR&&U_SUCCESS(status)) { + status = subStatus; + } + } +} + +UBool +SimpleFilteredBreakIteratorBuilder::suppressBreakAfter(const UnicodeString& exception, UErrorCode& status) +{ + UBool r = fSet.add(exception, status); + FB_TRACE("suppressBreakAfter",&exception,r,0); + return r; +} + +UBool +SimpleFilteredBreakIteratorBuilder::unsuppressBreakAfter(const UnicodeString& exception, UErrorCode& status) +{ + UBool r = fSet.remove(exception, status); + FB_TRACE("unsuppressBreakAfter",&exception,r,0); + return r; +} + +/** + * Jitterbug 2974: MSVC has a bug whereby new X[0] behaves badly. + * Work around this. + * + * Note: "new UnicodeString[subCount]" ends up calling global operator new + * on MSVC2012 for some reason. + */ +static inline UnicodeString* newUnicodeStringArray(size_t count) { + return new UnicodeString[count ? count : 1]; +} + +BreakIterator * +SimpleFilteredBreakIteratorBuilder::build(BreakIterator* adoptBreakIterator, UErrorCode& status) { + LocalPointer adopt(adoptBreakIterator); + + LocalPointer builder(new UCharsTrieBuilder(status), status); + LocalPointer builder2(new UCharsTrieBuilder(status), status); + if(U_FAILURE(status)) { + return nullptr; + } + + int32_t revCount = 0; + int32_t fwdCount = 0; + + int32_t subCount = fSet.size(); + + UnicodeString *ustrs_ptr = newUnicodeStringArray(subCount); + + LocalArray ustrs(ustrs_ptr); + + LocalMemory partials; + partials.allocateInsteadAndReset(subCount); + + LocalPointer backwardsTrie; // i.e. ".srM" for Mrs. + LocalPointer forwardsPartialTrie; // Has ".a" for "a.M." + + int n=0; + for ( int32_t i = 0; + i-1 && (nn+1)!=ustrs[i].length()) { + FB_TRACE("partial",&ustrs[i],false,i); + // is partial. + // is it unique? + int sameAs = -1; + for(int j=0;jadd(prefix, kPARTIAL, status); + revCount++; + FB_TRACE("Added partial",&prefix,false, i); + FB_TRACE(u_errorName(status),&ustrs[i],false,i); + partials[i] = kSuppressInReverse | kAddToForward; + } else { + FB_TRACE("NOT adding partial",&prefix,false, i); + FB_TRACE(u_errorName(status),&ustrs[i],false,i); + } + } + } + for(int i=0;iadd(ustrs[i], kMATCH, status); + revCount++; + FB_TRACE(u_errorName(status), &ustrs[i], false, i); + } else { + FB_TRACE("Adding fwd",&ustrs[i], false, i); + + // an optimization would be to only add the portion after the '.' + // for example, for "Ph.D." we store ".hP" in the reverse table. We could just store "D." in the forward, + // instead of "Ph.D." since we already know the "Ph." part is a match. + // would need the trie to be able to hold 0-length strings, though. + builder2->add(ustrs[i], kMATCH, status); // forward + fwdCount++; + //ustrs[i].reverse(); + ////if(debug2) u_printf("SUPPRESS- not Added(%d): /%S/ status=%s\n",partials[i], ustrs[i].getTerminatedBuffer(), u_errorName(status)); + } + } + FB_TRACE("AbbrCount",nullptr,false, subCount); + + if(revCount>0) { + backwardsTrie.adoptInstead(builder->build(USTRINGTRIE_BUILD_FAST, status)); + if(U_FAILURE(status)) { + FB_TRACE(u_errorName(status),nullptr,false, -1); + return nullptr; + } + } + + if(fwdCount>0) { + forwardsPartialTrie.adoptInstead(builder2->build(USTRINGTRIE_BUILD_FAST, status)); + if(U_FAILURE(status)) { + FB_TRACE(u_errorName(status),nullptr,false, -1); + return nullptr; + } + } + + return new SimpleFilteredSentenceBreakIterator(adopt.orphan(), forwardsPartialTrie.orphan(), backwardsTrie.orphan(), status); +} + + +// ----------- Base class implementation + +FilteredBreakIteratorBuilder::FilteredBreakIteratorBuilder() { +} + +FilteredBreakIteratorBuilder::~FilteredBreakIteratorBuilder() { +} + +FilteredBreakIteratorBuilder * +FilteredBreakIteratorBuilder::createInstance(const Locale& where, UErrorCode& status) { + if(U_FAILURE(status)) return nullptr; + LocalPointer ret(new SimpleFilteredBreakIteratorBuilder(where, status), status); + return (U_SUCCESS(status))? ret.orphan(): nullptr; +} + +FilteredBreakIteratorBuilder * +FilteredBreakIteratorBuilder::createInstance(UErrorCode &status) { + return createEmptyInstance(status); +} + +FilteredBreakIteratorBuilder * +FilteredBreakIteratorBuilder::createEmptyInstance(UErrorCode& status) { + if(U_FAILURE(status)) return nullptr; + LocalPointer ret(new SimpleFilteredBreakIteratorBuilder(status), status); + return (U_SUCCESS(status))? ret.orphan(): nullptr; +} + +U_NAMESPACE_END + +#endif //#if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION diff --git a/intl/icu/source/common/filterednormalizer2.cpp b/intl/icu/source/common/filterednormalizer2.cpp new file mode 100644 index 0000000000..5eafcb6d78 --- /dev/null +++ b/intl/icu/source/common/filterednormalizer2.cpp @@ -0,0 +1,363 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2009-2012, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: filterednormalizer2.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2009dec10 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_NORMALIZATION + +#include "unicode/edits.h" +#include "unicode/normalizer2.h" +#include "unicode/stringoptions.h" +#include "unicode/uniset.h" +#include "unicode/unistr.h" +#include "unicode/unorm.h" +#include "cpputils.h" + +U_NAMESPACE_BEGIN + +FilteredNormalizer2::~FilteredNormalizer2() {} + +UnicodeString & +FilteredNormalizer2::normalize(const UnicodeString &src, + UnicodeString &dest, + UErrorCode &errorCode) const { + uprv_checkCanGetBuffer(src, errorCode); + if(U_FAILURE(errorCode)) { + dest.setToBogus(); + return dest; + } + if(&dest==&src) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + return dest; + } + dest.remove(); + return normalize(src, dest, USET_SPAN_SIMPLE, errorCode); +} + +// Internal: No argument checking, and appends to dest. +// Pass as input spanCondition the one that is likely to yield a non-zero +// span length at the start of src. +// For set=[:age=3.2:], since almost all common characters were in Unicode 3.2, +// USET_SPAN_SIMPLE should be passed in for the start of src +// and USET_SPAN_NOT_CONTAINED should be passed in if we continue after +// an in-filter prefix. +UnicodeString & +FilteredNormalizer2::normalize(const UnicodeString &src, + UnicodeString &dest, + USetSpanCondition spanCondition, + UErrorCode &errorCode) const { + UnicodeString tempDest; // Don't throw away destination buffer between iterations. + for(int32_t prevSpanLimit=0; prevSpanLimitreset(); + } + options |= U_EDITS_NO_RESET; // Do not reset for each span. + normalizeUTF8(options, src.data(), src.length(), sink, edits, USET_SPAN_SIMPLE, errorCode); +} + +void +FilteredNormalizer2::normalizeUTF8(uint32_t options, const char *src, int32_t length, + ByteSink &sink, Edits *edits, + USetSpanCondition spanCondition, + UErrorCode &errorCode) const { + while (length > 0) { + int32_t spanLength = set.spanUTF8(src, length, spanCondition); + if (spanCondition == USET_SPAN_NOT_CONTAINED) { + if (spanLength != 0) { + if (edits != nullptr) { + edits->addUnchanged(spanLength); + } + if ((options & U_OMIT_UNCHANGED_TEXT) == 0) { + sink.Append(src, spanLength); + } + } + spanCondition = USET_SPAN_SIMPLE; + } else { + if (spanLength != 0) { + // Not norm2.normalizeSecondAndAppend() because we do not want + // to modify the non-filter part of dest. + norm2.normalizeUTF8(options, StringPiece(src, spanLength), sink, edits, errorCode); + if (U_FAILURE(errorCode)) { + break; + } + } + spanCondition = USET_SPAN_NOT_CONTAINED; + } + src += spanLength; + length -= spanLength; + } +} + +UnicodeString & +FilteredNormalizer2::normalizeSecondAndAppend(UnicodeString &first, + const UnicodeString &second, + UErrorCode &errorCode) const { + return normalizeSecondAndAppend(first, second, true, errorCode); +} + +UnicodeString & +FilteredNormalizer2::append(UnicodeString &first, + const UnicodeString &second, + UErrorCode &errorCode) const { + return normalizeSecondAndAppend(first, second, false, errorCode); +} + +UnicodeString & +FilteredNormalizer2::normalizeSecondAndAppend(UnicodeString &first, + const UnicodeString &second, + UBool doNormalize, + UErrorCode &errorCode) const { + uprv_checkCanGetBuffer(first, errorCode); + uprv_checkCanGetBuffer(second, errorCode); + if(U_FAILURE(errorCode)) { + return first; + } + if(&first==&second) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + return first; + } + if(first.isEmpty()) { + if(doNormalize) { + return normalize(second, first, errorCode); + } else { + return first=second; + } + } + // merge the in-filter suffix of the first string with the in-filter prefix of the second + int32_t prefixLimit=set.span(second, 0, USET_SPAN_SIMPLE); + if(prefixLimit!=0) { + UnicodeString prefix(second.tempSubString(0, prefixLimit)); + int32_t suffixStart=set.spanBack(first, INT32_MAX, USET_SPAN_SIMPLE); + if(suffixStart==0) { + if(doNormalize) { + norm2.normalizeSecondAndAppend(first, prefix, errorCode); + } else { + norm2.append(first, prefix, errorCode); + } + } else { + UnicodeString middle(first, suffixStart, INT32_MAX); + if(doNormalize) { + norm2.normalizeSecondAndAppend(middle, prefix, errorCode); + } else { + norm2.append(middle, prefix, errorCode); + } + first.replace(suffixStart, INT32_MAX, middle); + } + } + if(prefixLimit 0) { + int32_t spanLength = set.spanUTF8(s, length, spanCondition); + if (spanCondition == USET_SPAN_NOT_CONTAINED) { + spanCondition = USET_SPAN_SIMPLE; + } else { + if (!norm2.isNormalizedUTF8(StringPiece(s, spanLength), errorCode) || + U_FAILURE(errorCode)) { + return false; + } + spanCondition = USET_SPAN_NOT_CONTAINED; + } + s += spanLength; + length -= spanLength; + } + return true; +} + +UNormalizationCheckResult +FilteredNormalizer2::quickCheck(const UnicodeString &s, UErrorCode &errorCode) const { + uprv_checkCanGetBuffer(s, errorCode); + if(U_FAILURE(errorCode)) { + return UNORM_MAYBE; + } + UNormalizationCheckResult result=UNORM_YES; + USetSpanCondition spanCondition=USET_SPAN_SIMPLE; + for(int32_t prevSpanLimit=0; prevSpanLimit +#ifdef __MVS__ /* defined by z/OS compiler */ +#define _POSIX_SOURCE +#include /* 12 Nov 2011 JAM iscics() function */ +#endif +#include "charstr.h" + +using namespace icu; + +#ifndef UPLUG_TRACE +#define UPLUG_TRACE 0 +#endif + +#if UPLUG_TRACE +#include +#define DBG(x) fprintf(stderr, "%s:%d: ",__FILE__,__LINE__); fprintf x +#endif + +/** + * Internal structure of an ICU plugin. + */ + +struct UPlugData { + UPlugEntrypoint *entrypoint; /**< plugin entrypoint */ + uint32_t structSize; /**< initialized to the size of this structure */ + uint32_t token; /**< must be U_PLUG_TOKEN */ + void *lib; /**< plugin library, or nullptr */ + char libName[UPLUG_NAME_MAX]; /**< library name */ + char sym[UPLUG_NAME_MAX]; /**< plugin symbol, or nullptr */ + char config[UPLUG_NAME_MAX]; /**< configuration data */ + void *context; /**< user context data */ + char name[UPLUG_NAME_MAX]; /**< name of plugin */ + UPlugLevel level; /**< level of plugin */ + UBool awaitingLoad; /**< true if the plugin is awaiting a load call */ + UBool dontUnload; /**< true if plugin must stay resident (leak plugin and lib) */ + UErrorCode pluginStatus; /**< status code of plugin */ +}; + + + +#define UPLUG_LIBRARY_INITIAL_COUNT 8 +#define UPLUG_PLUGIN_INITIAL_COUNT 12 + +/** + * Remove an item + * @param list the full list + * @param listSize the number of entries in the list + * @param memberSize the size of one member + * @param itemToRemove the item number of the member + * @return the new listsize + */ +static int32_t uplug_removeEntryAt(void *list, int32_t listSize, int32_t memberSize, int32_t itemToRemove) { + uint8_t *bytePtr = (uint8_t *)list; + + /* get rid of some bad cases first */ + if(listSize<1) { + return listSize; + } + + /* is there anything to move? */ + if(listSize > itemToRemove+1) { + memmove(bytePtr+(itemToRemove*memberSize), bytePtr+((itemToRemove+1)*memberSize), memberSize); + } + + return listSize-1; +} + + + + +#if U_ENABLE_DYLOAD +/** + * Library management. Internal. + * @internal + */ +struct UPlugLibrary; + +/** + * Library management. Internal. + * @internal + */ +typedef struct UPlugLibrary { + void *lib; /**< library ptr */ + char name[UPLUG_NAME_MAX]; /**< library name */ + uint32_t ref; /**< reference count */ +} UPlugLibrary; + +static UPlugLibrary staticLibraryList[UPLUG_LIBRARY_INITIAL_COUNT]; +static UPlugLibrary * libraryList = staticLibraryList; +static int32_t libraryCount = 0; +static int32_t libraryMax = UPLUG_LIBRARY_INITIAL_COUNT; + +/** + * Search for a library. Doesn't lock + * @param libName libname to search for + * @return the library's struct + */ +static int32_t searchForLibraryName(const char *libName) { + int32_t i; + + for(i=0;i= libraryMax) { + /* Ran out of library slots. Statically allocated because we can't depend on allocating memory.. */ + *status = U_MEMORY_ALLOCATION_ERROR; +#if UPLUG_TRACE + DBG((stderr, "uplug_openLibrary() - out of library slots (max %d)\n", libraryMax)); +#endif + return nullptr; + } + /* Some operating systems don't want + DL operations from multiple threads. */ + libraryList[libEntry].lib = uprv_dl_open(libName, status); +#if UPLUG_TRACE + DBG((stderr, "uplug_openLibrary(%s,%s) libEntry %d, lib %p\n", libName, u_errorName(*status), libEntry, lib)); +#endif + + if(libraryList[libEntry].lib == nullptr || U_FAILURE(*status)) { + /* cleanup. */ + libraryList[libEntry].lib = nullptr; /* failure with open */ + libraryList[libEntry].name[0] = 0; +#if UPLUG_TRACE + DBG((stderr, "uplug_openLibrary(%s,%s) libEntry %d, lib %p\n", libName, u_errorName(*status), libEntry, lib)); +#endif + /* no need to free - just won't increase the count. */ + libraryCount--; + } else { /* is it still there? */ + /* link it in */ + uprv_strncpy(libraryList[libEntry].name,libName,UPLUG_NAME_MAX); + libraryList[libEntry].ref=1; + lib = libraryList[libEntry].lib; + } + + } else { + lib = libraryList[libEntry].lib; + libraryList[libEntry].ref++; + } + return lib; +} + +U_CAPI void U_EXPORT2 +uplug_closeLibrary(void *lib, UErrorCode *status) { + int32_t i; + +#if UPLUG_TRACE + DBG((stderr, "uplug_closeLibrary(%p,%s) list %p\n", lib, u_errorName(*status), (void*)libraryList)); +#endif + if(U_FAILURE(*status)) return; + + for(i=0;i=pastPlug) { + return pluginCount; + } else { + return (d-pluginList)/sizeof(pluginList[0]); + } +} + + +U_CAPI UPlugData * U_EXPORT2 +uplug_nextPlug(UPlugData *prior) { + if(prior==nullptr) { + return pluginList; + } else { + UPlugData *nextPlug = &prior[1]; + UPlugData *pastPlug = &pluginList[pluginCount]; + + if(nextPlug>=pastPlug) { + return nullptr; + } else { + return nextPlug; + } + } +} + + + +/** + * Call the plugin with some params + */ +static void uplug_callPlug(UPlugData *plug, UPlugReason reason, UErrorCode *status) { + UPlugTokenReturn token; + if(plug==nullptr||U_FAILURE(*status)) { + return; + } + token = (*(plug->entrypoint))(plug, reason, status); + if(token!=UPLUG_TOKEN) { + *status = U_INTERNAL_PROGRAM_ERROR; + } +} + + +static void uplug_unloadPlug(UPlugData *plug, UErrorCode *status) { + if(plug->awaitingLoad) { /* shouldn't happen. Plugin hasn't been loaded yet.*/ + *status = U_INTERNAL_PROGRAM_ERROR; + return; + } + if(U_SUCCESS(plug->pluginStatus)) { + /* Don't unload a plug which has a failing load status - means it didn't actually load. */ + uplug_callPlug(plug, UPLUG_REASON_UNLOAD, status); + } +} + +static void uplug_queryPlug(UPlugData *plug, UErrorCode *status) { + if(!plug->awaitingLoad || !(plug->level == UPLUG_LEVEL_UNKNOWN) ) { /* shouldn't happen. Plugin hasn't been loaded yet.*/ + *status = U_INTERNAL_PROGRAM_ERROR; + return; + } + plug->level = UPLUG_LEVEL_INVALID; + uplug_callPlug(plug, UPLUG_REASON_QUERY, status); + if(U_SUCCESS(*status)) { + if(plug->level == UPLUG_LEVEL_INVALID) { + plug->pluginStatus = U_PLUGIN_DIDNT_SET_LEVEL; + plug->awaitingLoad = false; + } + } else { + plug->pluginStatus = U_INTERNAL_PROGRAM_ERROR; + plug->awaitingLoad = false; + } +} + + +static void uplug_loadPlug(UPlugData *plug, UErrorCode *status) { + if(U_FAILURE(*status)) { + return; + } + if(!plug->awaitingLoad || (plug->level < UPLUG_LEVEL_LOW) ) { /* shouldn't happen. Plugin hasn't been loaded yet.*/ + *status = U_INTERNAL_PROGRAM_ERROR; + return; + } + uplug_callPlug(plug, UPLUG_REASON_LOAD, status); + plug->awaitingLoad = false; + if(!U_SUCCESS(*status)) { + plug->pluginStatus = U_INTERNAL_PROGRAM_ERROR; + } +} + +static UPlugData *uplug_allocateEmptyPlug(UErrorCode *status) +{ + UPlugData *plug = nullptr; + + if(U_FAILURE(*status)) { + return nullptr; + } + + if(pluginCount == UPLUG_PLUGIN_INITIAL_COUNT) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + + plug = &pluginList[pluginCount++]; + + plug->token = UPLUG_TOKEN; + plug->structSize = sizeof(UPlugData); + plug->name[0]=0; + plug->level = UPLUG_LEVEL_UNKNOWN; /* initialize to null state */ + plug->awaitingLoad = true; + plug->dontUnload = false; + plug->pluginStatus = U_ZERO_ERROR; + plug->libName[0] = 0; + plug->config[0]=0; + plug->sym[0]=0; + plug->lib=nullptr; + plug->entrypoint=nullptr; + + + return plug; +} + +static UPlugData *uplug_allocatePlug(UPlugEntrypoint *entrypoint, const char *config, void *lib, const char *symName, + UErrorCode *status) { + UPlugData *plug = uplug_allocateEmptyPlug(status); + if(U_FAILURE(*status)) { + return nullptr; + } + + if(config!=nullptr) { + uprv_strncpy(plug->config, config, UPLUG_NAME_MAX); + } else { + plug->config[0] = 0; + } + + if(symName!=nullptr) { + uprv_strncpy(plug->sym, symName, UPLUG_NAME_MAX); + } else { + plug->sym[0] = 0; + } + + plug->entrypoint = entrypoint; + plug->lib = lib; + uplug_queryPlug(plug, status); + + return plug; +} + +static void uplug_deallocatePlug(UPlugData *plug, UErrorCode *status) { + UErrorCode subStatus = U_ZERO_ERROR; + if(!plug->dontUnload) { +#if U_ENABLE_DYLOAD + uplug_closeLibrary(plug->lib, &subStatus); +#endif + } + plug->lib = nullptr; + if(U_SUCCESS(*status) && U_FAILURE(subStatus)) { + *status = subStatus; + } + /* shift plugins up and decrement count. */ + if(U_SUCCESS(*status)) { + /* all ok- remove. */ + pluginCount = uplug_removeEntryAt(pluginList, pluginCount, sizeof(plug[0]), uplug_pluginNumber(plug)); + } else { + /* not ok- leave as a message. */ + plug->awaitingLoad=false; + plug->entrypoint=0; + plug->dontUnload=true; + } +} + +static void uplug_doUnloadPlug(UPlugData *plugToRemove, UErrorCode *status) { + if(plugToRemove != nullptr) { + uplug_unloadPlug(plugToRemove, status); + uplug_deallocatePlug(plugToRemove, status); + } +} + +U_CAPI void U_EXPORT2 +uplug_removePlug(UPlugData *plug, UErrorCode *status) { + UPlugData *cursor = nullptr; + UPlugData *plugToRemove = nullptr; + if(U_FAILURE(*status)) return; + + for(cursor=pluginList;cursor!=nullptr;) { + if(cursor==plug) { + plugToRemove = plug; + cursor=nullptr; + } else { + cursor = uplug_nextPlug(cursor); + } + } + + uplug_doUnloadPlug(plugToRemove, status); +} + + + + +U_CAPI void U_EXPORT2 +uplug_setPlugNoUnload(UPlugData *data, UBool dontUnload) +{ + data->dontUnload = dontUnload; +} + + +U_CAPI void U_EXPORT2 +uplug_setPlugLevel(UPlugData *data, UPlugLevel level) { + data->level = level; +} + + +U_CAPI UPlugLevel U_EXPORT2 +uplug_getPlugLevel(UPlugData *data) { + return data->level; +} + + +U_CAPI void U_EXPORT2 +uplug_setPlugName(UPlugData *data, const char *name) { + uprv_strncpy(data->name, name, UPLUG_NAME_MAX); +} + + +U_CAPI const char * U_EXPORT2 +uplug_getPlugName(UPlugData *data) { + return data->name; +} + + +U_CAPI const char * U_EXPORT2 +uplug_getSymbolName(UPlugData *data) { + return data->sym; +} + +U_CAPI const char * U_EXPORT2 +uplug_getLibraryName(UPlugData *data, UErrorCode *status) { + if(data->libName[0]) { + return data->libName; + } else { +#if U_ENABLE_DYLOAD + return uplug_findLibrary(data->lib, status); +#else + return nullptr; +#endif + } +} + +U_CAPI void * U_EXPORT2 +uplug_getLibrary(UPlugData *data) { + return data->lib; +} + +U_CAPI void * U_EXPORT2 +uplug_getContext(UPlugData *data) { + return data->context; +} + + +U_CAPI void U_EXPORT2 +uplug_setContext(UPlugData *data, void *context) { + data->context = context; +} + +U_CAPI const char* U_EXPORT2 +uplug_getConfiguration(UPlugData *data) { + return data->config; +} + +U_CAPI UPlugData* U_EXPORT2 +uplug_getPlugInternal(int32_t n) { + if(n <0 || n >= pluginCount) { + return nullptr; + } else { + return &(pluginList[n]); + } +} + + +U_CAPI UErrorCode U_EXPORT2 +uplug_getPlugLoadStatus(UPlugData *plug) { + return plug->pluginStatus; +} + + + + +/** + * Initialize a plugin from an entrypoint and library - but don't load it. + */ +static UPlugData* uplug_initPlugFromEntrypointAndLibrary(UPlugEntrypoint *entrypoint, const char *config, void *lib, const char *sym, + UErrorCode *status) { + UPlugData *plug = nullptr; + + plug = uplug_allocatePlug(entrypoint, config, lib, sym, status); + + if(U_SUCCESS(*status)) { + return plug; + } else { + uplug_deallocatePlug(plug, status); + return nullptr; + } +} + +U_CAPI UPlugData* U_EXPORT2 +uplug_loadPlugFromEntrypoint(UPlugEntrypoint *entrypoint, const char *config, UErrorCode *status) { + UPlugData* plug = uplug_initPlugFromEntrypointAndLibrary(entrypoint, config, nullptr, nullptr, status); + uplug_loadPlug(plug, status); + return plug; +} + +#if U_ENABLE_DYLOAD + +static UPlugData* +uplug_initErrorPlug(const char *libName, const char *sym, const char *config, const char *nameOrError, UErrorCode loadStatus, UErrorCode *status) +{ + UPlugData *plug = uplug_allocateEmptyPlug(status); + if(U_FAILURE(*status)) return nullptr; + + plug->pluginStatus = loadStatus; + plug->awaitingLoad = false; /* Won't load. */ + plug->dontUnload = true; /* cannot unload. */ + + if(sym!=nullptr) { + uprv_strncpy(plug->sym, sym, UPLUG_NAME_MAX); + } + + if(libName!=nullptr) { + uprv_strncpy(plug->libName, libName, UPLUG_NAME_MAX); + } + + if(nameOrError!=nullptr) { + uprv_strncpy(plug->name, nameOrError, UPLUG_NAME_MAX); + } + + if(config!=nullptr) { + uprv_strncpy(plug->config, config, UPLUG_NAME_MAX); + } + + return plug; +} + +/** + * Fetch a plugin from DLL, and then initialize it from a library- but don't load it. + */ +static UPlugData* +uplug_initPlugFromLibrary(const char *libName, const char *sym, const char *config, UErrorCode *status) { + void *lib = nullptr; + UPlugData *plug = nullptr; + if(U_FAILURE(*status)) { return nullptr; } + lib = uplug_openLibrary(libName, status); + if(lib!=nullptr && U_SUCCESS(*status)) { + UPlugEntrypoint *entrypoint = nullptr; + entrypoint = (UPlugEntrypoint*)uprv_dlsym_func(lib, sym, status); + + if(entrypoint!=nullptr&&U_SUCCESS(*status)) { + plug = uplug_initPlugFromEntrypointAndLibrary(entrypoint, config, lib, sym, status); + if(plug!=nullptr&&U_SUCCESS(*status)) { + plug->lib = lib; /* plug takes ownership of library */ + lib = nullptr; /* library is now owned by plugin. */ + } + } else { + UErrorCode subStatus = U_ZERO_ERROR; + plug = uplug_initErrorPlug(libName,sym,config,"ERROR: Could not load entrypoint",(lib==nullptr)?U_MISSING_RESOURCE_ERROR:*status,&subStatus); + } + if(lib!=nullptr) { /* still need to close the lib */ + UErrorCode subStatus = U_ZERO_ERROR; + uplug_closeLibrary(lib, &subStatus); /* don't care here */ + } + } else { + UErrorCode subStatus = U_ZERO_ERROR; + plug = uplug_initErrorPlug(libName,sym,config,"ERROR: could not load library",(lib==nullptr)?U_MISSING_RESOURCE_ERROR:*status,&subStatus); + } + return plug; +} + +U_CAPI UPlugData* U_EXPORT2 +uplug_loadPlugFromLibrary(const char *libName, const char *sym, const char *config, UErrorCode *status) { + UPlugData *plug = nullptr; + if(U_FAILURE(*status)) { return nullptr; } + plug = uplug_initPlugFromLibrary(libName, sym, config, status); + uplug_loadPlug(plug, status); + + return plug; +} + +#endif + +static UPlugLevel gCurrentLevel = UPLUG_LEVEL_LOW; + +U_CAPI UPlugLevel U_EXPORT2 uplug_getCurrentLevel() { + return gCurrentLevel; +} + +static UBool U_CALLCONV uplug_cleanup() +{ + int32_t i; + + UPlugData *pluginToRemove; + /* cleanup plugs */ + for(i=0;iawaitingLoad) { + if(pluginToLoad->level == UPLUG_LEVEL_LOW) { + if(currentLevel > UPLUG_LEVEL_LOW) { + pluginToLoad->pluginStatus = U_PLUGIN_TOO_HIGH; + } else { + UPlugLevel newLevel; + uplug_loadPlug(pluginToLoad, &subStatus); + newLevel = uplug_getCurrentLevel(); + if(newLevel > currentLevel) { + pluginToLoad->pluginStatus = U_PLUGIN_CHANGED_LEVEL_WARNING; + currentLevel = newLevel; + } + } + pluginToLoad->awaitingLoad = false; + } + } + } + for(i=0;iawaitingLoad) { + if(pluginToLoad->level == UPLUG_LEVEL_INVALID) { + pluginToLoad->pluginStatus = U_PLUGIN_DIDNT_SET_LEVEL; + } else if(pluginToLoad->level == UPLUG_LEVEL_UNKNOWN) { + pluginToLoad->pluginStatus = U_INTERNAL_PROGRAM_ERROR; + } else { + uplug_loadPlug(pluginToLoad, &subStatus); + } + pluginToLoad->awaitingLoad = false; + } + } + +#if UPLUG_TRACE + DBG((stderr, " Done Loading Plugs. Level: %d\n", (int32_t)uplug_getCurrentLevel())); +#endif +} + +/* Name of the plugin config file */ +static char plugin_file[2048] = ""; +#endif + +U_CAPI const char* U_EXPORT2 +uplug_getPluginFile() { +#if U_ENABLE_DYLOAD && !UCONFIG_NO_FILE_IO + return plugin_file; +#else + return nullptr; +#endif +} + + +// uplug_init() is called first thing from u_init(). + +U_CAPI void U_EXPORT2 +uplug_init(UErrorCode *status) { +#if !U_ENABLE_DYLOAD + (void)status; /* unused */ +#elif !UCONFIG_NO_FILE_IO + CharString plugin_dir; + const char *env = getenv("ICU_PLUGINS"); + + if(U_FAILURE(*status)) return; + if(env != nullptr) { + plugin_dir.append(env, -1, *status); + } + if(U_FAILURE(*status)) return; + +#if defined(DEFAULT_ICU_PLUGINS) + if(plugin_dir.isEmpty()) { + plugin_dir.append(DEFAULT_ICU_PLUGINS, -1, *status); + } +#endif + +#if UPLUG_TRACE + DBG((stderr, "ICU_PLUGINS=%s\n", plugin_dir.data())); +#endif + + if(!plugin_dir.isEmpty()) { + FILE *f; + + CharString pluginFile; +#ifdef OS390BATCH +/* There are potentially a lot of ways to implement a plugin directory on OS390/zOS */ +/* Keeping in mind that unauthorized file access is logged, monitored, and enforced */ +/* I've chosen to open a DDNAME if BATCH and leave it alone for (presumably) UNIX */ +/* System Services. Alternative techniques might be allocating a member in */ +/* SYS1.PARMLIB or setting an environment variable "ICU_PLUGIN_PATH" (?). The */ +/* DDNAME can be connected to a file in the HFS if need be. */ + + pluginFile.append("//DD:ICUPLUG", -1, *status); /* JAM 20 Oct 2011 */ +#else + pluginFile.append(plugin_dir, *status); + pluginFile.append(U_FILE_SEP_STRING, -1, *status); + pluginFile.append("icuplugins", -1, *status); + pluginFile.append(U_ICU_VERSION_SHORT, -1, *status); + pluginFile.append(".txt", -1, *status); +#endif + +#if UPLUG_TRACE + DBG((stderr, "status=%s\n", u_errorName(*status))); +#endif + + if(U_FAILURE(*status)) { + return; + } + if((size_t)pluginFile.length() > (sizeof(plugin_file)-1)) { + *status = U_BUFFER_OVERFLOW_ERROR; +#if UPLUG_TRACE + DBG((stderr, "status=%s\n", u_errorName(*status))); +#endif + return; + } + + /* plugin_file is not used for processing - it is only used + so that uplug_getPluginFile() works (i.e. icuinfo) + */ + pluginFile.extract(plugin_file, sizeof(plugin_file), *status); + +#if UPLUG_TRACE + DBG((stderr, "pluginfile= %s len %d/%d\n", plugin_file, (int)strlen(plugin_file), (int)sizeof(plugin_file))); +#endif + +#ifdef __MVS__ + if (iscics()) /* 12 Nov 2011 JAM */ + { + f = nullptr; + } + else +#endif + { + f = fopen(pluginFile.data(), "r"); + } + + if(f != nullptr) { + char linebuf[1024]; + char *p, *libName=nullptr, *symName=nullptr, *config=nullptr; + int32_t line = 0; + + + while(fgets(linebuf,1023,f)) { + line++; + + if(!*linebuf || *linebuf=='#') { + continue; + } else { + p = linebuf; + while(*p&&isspace((int)*p)) + p++; + if(!*p || *p=='#') continue; + libName = p; + while(*p&&!isspace((int)*p)) { + p++; + } + if(!*p || *p=='#') continue; /* no tab after libname */ + *p=0; /* end of libname */ + p++; + while(*p&&isspace((int)*p)) { + p++; + } + if(!*p||*p=='#') continue; /* no symname after libname +tab */ + symName = p; + while(*p&&!isspace((int)*p)) { + p++; + } + + if(*p) { /* has config */ + *p=0; + ++p; + while(*p&&isspace((int)*p)) { + p++; + } + if(*p) { + config = p; + } + } + + /* chop whitespace at the end of the config */ + if(config!=nullptr&&*config!=0) { + p = config+strlen(config); + while(p>config&&isspace((int)*(--p))) { + *p=0; + } + } + + /* OK, we're good. */ + { + UErrorCode subStatus = U_ZERO_ERROR; + UPlugData *plug = uplug_initPlugFromLibrary(libName, symName, config, &subStatus); + if(U_FAILURE(subStatus) && U_SUCCESS(*status)) { + *status = subStatus; + } +#if UPLUG_TRACE + DBG((stderr, "PLUGIN libName=[%s], sym=[%s], config=[%s]\n", libName, symName, config)); + DBG((stderr, " -> %p, %s\n", (void*)plug, u_errorName(subStatus))); +#else + (void)plug; /* unused */ +#endif + } + } + } + fclose(f); + } else { +#if UPLUG_TRACE + DBG((stderr, "Can't open plugin file %s\n", plugin_file)); +#endif + } + } + uplug_loadWaitingPlugs(status); +#endif /* U_ENABLE_DYLOAD */ + gCurrentLevel = UPLUG_LEVEL_HIGH; + ucln_registerCleanup(UCLN_UPLUG, uplug_cleanup); +} + +#endif + + diff --git a/intl/icu/source/common/icuplugimp.h b/intl/icu/source/common/icuplugimp.h new file mode 100644 index 0000000000..9df309204e --- /dev/null +++ b/intl/icu/source/common/icuplugimp.h @@ -0,0 +1,93 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2009-2015, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* FILE NAME : icuplugimp.h +* +* Internal functions for the ICU plugin system +* +* Date Name Description +* 10/29/2009 sl New. +****************************************************************************** +*/ + + +#ifndef ICUPLUGIMP_H +#define ICUPLUGIMP_H + +#include "unicode/icuplug.h" + +#if UCONFIG_ENABLE_PLUGINS + +/*========================*/ +/** @{ Library Manipulation + */ + +/** + * Open a library, adding a reference count if needed. + * @param libName library name to load + * @param status error code + * @return the library pointer, or NULL + * @internal internal use only + */ +U_CAPI void * U_EXPORT2 +uplug_openLibrary(const char *libName, UErrorCode *status); + +/** + * Close a library, if its reference count is 0 + * @param lib the library to close + * @param status error code + * @internal internal use only + */ +U_CAPI void U_EXPORT2 +uplug_closeLibrary(void *lib, UErrorCode *status); + +/** + * Get a library's name, or NULL if not found. + * @param lib the library's name + * @param status error code + * @return the library name, or NULL if not found. + * @internal internal use only + */ +U_CAPI char * U_EXPORT2 +uplug_findLibrary(void *lib, UErrorCode *status); + +/** @} */ + +/*========================*/ +/** {@ ICU Plugin internal interfaces + */ + +/** + * Initialize the plugins + * @param status error result + * @internal - Internal use only. + */ +U_CAPI void U_EXPORT2 +uplug_init(UErrorCode *status); + +/** + * Get raw plug N + * @internal - Internal use only + */ +U_CAPI UPlugData* U_EXPORT2 +uplug_getPlugInternal(int32_t n); + +/** + * Get the name of the plugin file. + * @internal - Internal use only. + */ +U_CAPI const char* U_EXPORT2 +uplug_getPluginFile(void); + +/** @} */ + +#endif + +#endif diff --git a/intl/icu/source/common/loadednormalizer2impl.cpp b/intl/icu/source/common/loadednormalizer2impl.cpp new file mode 100644 index 0000000000..768564edc8 --- /dev/null +++ b/intl/icu/source/common/loadednormalizer2impl.cpp @@ -0,0 +1,418 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2014, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* loadednormalizer2impl.cpp +* +* created on: 2014sep03 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_NORMALIZATION + +#include "unicode/udata.h" +#include "unicode/localpointer.h" +#include "unicode/normalizer2.h" +#include "unicode/ucptrie.h" +#include "unicode/unistr.h" +#include "unicode/unorm.h" +#include "cstring.h" +#include "mutex.h" +#include "norm2allmodes.h" +#include "normalizer2impl.h" +#include "uassert.h" +#include "ucln_cmn.h" +#include "uhash.h" + +U_NAMESPACE_BEGIN + +class LoadedNormalizer2Impl : public Normalizer2Impl { +public: + LoadedNormalizer2Impl() : memory(nullptr), ownedTrie(nullptr) {} + virtual ~LoadedNormalizer2Impl(); + + void load(const char *packageName, const char *name, UErrorCode &errorCode); + +private: + static UBool U_CALLCONV + isAcceptable(void *context, const char *type, const char *name, const UDataInfo *pInfo); + + UDataMemory *memory; + UCPTrie *ownedTrie; +}; + +LoadedNormalizer2Impl::~LoadedNormalizer2Impl() { + udata_close(memory); + ucptrie_close(ownedTrie); +} + +UBool U_CALLCONV +LoadedNormalizer2Impl::isAcceptable(void * /*context*/, + const char * /* type */, const char * /*name*/, + const UDataInfo *pInfo) { + if( + pInfo->size>=20 && + pInfo->isBigEndian==U_IS_BIG_ENDIAN && + pInfo->charsetFamily==U_CHARSET_FAMILY && + pInfo->dataFormat[0]==0x4e && /* dataFormat="Nrm2" */ + pInfo->dataFormat[1]==0x72 && + pInfo->dataFormat[2]==0x6d && + pInfo->dataFormat[3]==0x32 && + pInfo->formatVersion[0]==4 + ) { + // Normalizer2Impl *me=(Normalizer2Impl *)context; + // uprv_memcpy(me->dataVersion, pInfo->dataVersion, 4); + return true; + } else { + return false; + } +} + +void +LoadedNormalizer2Impl::load(const char *packageName, const char *name, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return; + } + memory=udata_openChoice(packageName, "nrm", name, isAcceptable, this, &errorCode); + if(U_FAILURE(errorCode)) { + return; + } + const uint8_t *inBytes=(const uint8_t *)udata_getMemory(memory); + const int32_t *inIndexes=(const int32_t *)inBytes; + int32_t indexesLength=inIndexes[IX_NORM_TRIE_OFFSET]/4; + if(indexesLength<=IX_MIN_LCCC_CP) { + errorCode=U_INVALID_FORMAT_ERROR; // Not enough indexes. + return; + } + + int32_t offset=inIndexes[IX_NORM_TRIE_OFFSET]; + int32_t nextOffset=inIndexes[IX_EXTRA_DATA_OFFSET]; + ownedTrie=ucptrie_openFromBinary(UCPTRIE_TYPE_FAST, UCPTRIE_VALUE_BITS_16, + inBytes+offset, nextOffset-offset, nullptr, + &errorCode); + if(U_FAILURE(errorCode)) { + return; + } + + offset=nextOffset; + nextOffset=inIndexes[IX_SMALL_FCD_OFFSET]; + const uint16_t *inExtraData=(const uint16_t *)(inBytes+offset); + + // smallFCD: new in formatVersion 2 + offset=nextOffset; + const uint8_t *inSmallFCD=inBytes+offset; + + init(inIndexes, ownedTrie, inExtraData, inSmallFCD); +} + +// instance cache ---------------------------------------------------------- *** + +Norm2AllModes * +Norm2AllModes::createInstance(const char *packageName, + const char *name, + UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return nullptr; + } + LoadedNormalizer2Impl *impl=new LoadedNormalizer2Impl; + if(impl==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + impl->load(packageName, name, errorCode); + return createInstance(impl, errorCode); +} + +U_CDECL_BEGIN +static UBool U_CALLCONV uprv_loaded_normalizer2_cleanup(); +U_CDECL_END + +#if !NORM2_HARDCODE_NFC_DATA +static Norm2AllModes *nfcSingleton; +static icu::UInitOnce nfcInitOnce {}; +#endif + +static Norm2AllModes *nfkcSingleton; +static icu::UInitOnce nfkcInitOnce {}; + +static Norm2AllModes *nfkc_cfSingleton; +static icu::UInitOnce nfkc_cfInitOnce {}; + +static UHashtable *cache=nullptr; + +// UInitOnce singleton initialization function +static void U_CALLCONV initSingletons(const char *what, UErrorCode &errorCode) { +#if !NORM2_HARDCODE_NFC_DATA + if (uprv_strcmp(what, "nfc") == 0) { + nfcSingleton = Norm2AllModes::createInstance(nullptr, "nfc", errorCode); + } else +#endif + if (uprv_strcmp(what, "nfkc") == 0) { + nfkcSingleton = Norm2AllModes::createInstance(nullptr, "nfkc", errorCode); + } else if (uprv_strcmp(what, "nfkc_cf") == 0) { + nfkc_cfSingleton = Norm2AllModes::createInstance(nullptr, "nfkc_cf", errorCode); + } else { + UPRV_UNREACHABLE_EXIT; // Unknown singleton + } + ucln_common_registerCleanup(UCLN_COMMON_LOADED_NORMALIZER2, uprv_loaded_normalizer2_cleanup); +} + +U_CDECL_BEGIN + +static void U_CALLCONV deleteNorm2AllModes(void *allModes) { + delete (Norm2AllModes *)allModes; +} + +static UBool U_CALLCONV uprv_loaded_normalizer2_cleanup() { +#if !NORM2_HARDCODE_NFC_DATA + delete nfcSingleton; + nfcSingleton = nullptr; + nfcInitOnce.reset(); +#endif + + delete nfkcSingleton; + nfkcSingleton = nullptr; + nfkcInitOnce.reset(); + + delete nfkc_cfSingleton; + nfkc_cfSingleton = nullptr; + nfkc_cfInitOnce.reset(); + + uhash_close(cache); + cache=nullptr; + return true; +} + +U_CDECL_END + +#if !NORM2_HARDCODE_NFC_DATA +const Norm2AllModes * +Norm2AllModes::getNFCInstance(UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { return nullptr; } + umtx_initOnce(nfcInitOnce, &initSingletons, "nfc", errorCode); + return nfcSingleton; +} +#endif + +const Norm2AllModes * +Norm2AllModes::getNFKCInstance(UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { return nullptr; } + umtx_initOnce(nfkcInitOnce, &initSingletons, "nfkc", errorCode); + return nfkcSingleton; +} + +const Norm2AllModes * +Norm2AllModes::getNFKC_CFInstance(UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { return nullptr; } + umtx_initOnce(nfkc_cfInitOnce, &initSingletons, "nfkc_cf", errorCode); + return nfkc_cfSingleton; +} + +#if !NORM2_HARDCODE_NFC_DATA +const Normalizer2 * +Normalizer2::getNFCInstance(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFCInstance(errorCode); + return allModes!=nullptr ? &allModes->comp : nullptr; +} + +const Normalizer2 * +Normalizer2::getNFDInstance(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFCInstance(errorCode); + return allModes!=nullptr ? &allModes->decomp : nullptr; +} + +const Normalizer2 *Normalizer2Factory::getFCDInstance(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFCInstance(errorCode); + return allModes!=nullptr ? &allModes->fcd : nullptr; +} + +const Normalizer2 *Normalizer2Factory::getFCCInstance(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFCInstance(errorCode); + return allModes!=nullptr ? &allModes->fcc : nullptr; +} + +const Normalizer2Impl * +Normalizer2Factory::getNFCImpl(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFCInstance(errorCode); + return allModes!=nullptr ? allModes->impl : nullptr; +} +#endif + +const Normalizer2 * +Normalizer2::getNFKCInstance(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFKCInstance(errorCode); + return allModes!=nullptr ? &allModes->comp : nullptr; +} + +const Normalizer2 * +Normalizer2::getNFKDInstance(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFKCInstance(errorCode); + return allModes!=nullptr ? &allModes->decomp : nullptr; +} + +const Normalizer2 * +Normalizer2::getNFKCCasefoldInstance(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFKC_CFInstance(errorCode); + return allModes!=nullptr ? &allModes->comp : nullptr; +} + +const Normalizer2 * +Normalizer2::getInstance(const char *packageName, + const char *name, + UNormalization2Mode mode, + UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return nullptr; + } + if(name==nullptr || *name==0) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + const Norm2AllModes *allModes=nullptr; + if(packageName==nullptr) { + if(0==uprv_strcmp(name, "nfc")) { + allModes=Norm2AllModes::getNFCInstance(errorCode); + } else if(0==uprv_strcmp(name, "nfkc")) { + allModes=Norm2AllModes::getNFKCInstance(errorCode); + } else if(0==uprv_strcmp(name, "nfkc_cf")) { + allModes=Norm2AllModes::getNFKC_CFInstance(errorCode); + } + } + if(allModes==nullptr && U_SUCCESS(errorCode)) { + { + Mutex lock; + if(cache!=nullptr) { + allModes=(Norm2AllModes *)uhash_get(cache, name); + } + } + if(allModes==nullptr) { + ucln_common_registerCleanup(UCLN_COMMON_LOADED_NORMALIZER2, uprv_loaded_normalizer2_cleanup); + LocalPointer localAllModes( + Norm2AllModes::createInstance(packageName, name, errorCode)); + if(U_SUCCESS(errorCode)) { + Mutex lock; + if(cache==nullptr) { + cache=uhash_open(uhash_hashChars, uhash_compareChars, nullptr, &errorCode); + if(U_FAILURE(errorCode)) { + return nullptr; + } + uhash_setKeyDeleter(cache, uprv_free); + uhash_setValueDeleter(cache, deleteNorm2AllModes); + } + void *temp=uhash_get(cache, name); + if(temp==nullptr) { + int32_t keyLength= static_cast(uprv_strlen(name)+1); + char *nameCopy=(char *)uprv_malloc(keyLength); + if(nameCopy==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memcpy(nameCopy, name, keyLength); + allModes=localAllModes.getAlias(); + uhash_put(cache, nameCopy, localAllModes.orphan(), &errorCode); + } else { + // race condition + allModes=(Norm2AllModes *)temp; + } + } + } + } + if(allModes!=nullptr && U_SUCCESS(errorCode)) { + switch(mode) { + case UNORM2_COMPOSE: + return &allModes->comp; + case UNORM2_DECOMPOSE: + return &allModes->decomp; + case UNORM2_FCD: + return &allModes->fcd; + case UNORM2_COMPOSE_CONTIGUOUS: + return &allModes->fcc; + default: + break; // do nothing + } + } + return nullptr; +} + +const Normalizer2 * +Normalizer2Factory::getInstance(UNormalizationMode mode, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return nullptr; + } + switch(mode) { + case UNORM_NFD: + return Normalizer2::getNFDInstance(errorCode); + case UNORM_NFKD: + return Normalizer2::getNFKDInstance(errorCode); + case UNORM_NFC: + return Normalizer2::getNFCInstance(errorCode); + case UNORM_NFKC: + return Normalizer2::getNFKCInstance(errorCode); + case UNORM_FCD: + return getFCDInstance(errorCode); + default: // UNORM_NONE + return getNoopInstance(errorCode); + } +} + +const Normalizer2Impl * +Normalizer2Factory::getNFKCImpl(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFKCInstance(errorCode); + return allModes!=nullptr ? allModes->impl : nullptr; +} + +const Normalizer2Impl * +Normalizer2Factory::getNFKC_CFImpl(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFKC_CFInstance(errorCode); + return allModes!=nullptr ? allModes->impl : nullptr; +} + +U_NAMESPACE_END + +// C API ------------------------------------------------------------------- *** + +U_NAMESPACE_USE + +U_CAPI const UNormalizer2 * U_EXPORT2 +unorm2_getNFKCInstance(UErrorCode *pErrorCode) { + return (const UNormalizer2 *)Normalizer2::getNFKCInstance(*pErrorCode); +} + +U_CAPI const UNormalizer2 * U_EXPORT2 +unorm2_getNFKDInstance(UErrorCode *pErrorCode) { + return (const UNormalizer2 *)Normalizer2::getNFKDInstance(*pErrorCode); +} + +U_CAPI const UNormalizer2 * U_EXPORT2 +unorm2_getNFKCCasefoldInstance(UErrorCode *pErrorCode) { + return (const UNormalizer2 *)Normalizer2::getNFKCCasefoldInstance(*pErrorCode); +} + +U_CAPI const UNormalizer2 * U_EXPORT2 +unorm2_getInstance(const char *packageName, + const char *name, + UNormalization2Mode mode, + UErrorCode *pErrorCode) { + return (const UNormalizer2 *)Normalizer2::getInstance(packageName, name, mode, *pErrorCode); +} + +U_CFUNC UNormalizationCheckResult +unorm_getQuickCheck(UChar32 c, UNormalizationMode mode) { + if(mode<=UNORM_NONE || UNORM_FCD<=mode) { + return UNORM_YES; + } + UErrorCode errorCode=U_ZERO_ERROR; + const Normalizer2 *norm2=Normalizer2Factory::getInstance(mode, errorCode); + if(U_SUCCESS(errorCode)) { + return ((const Normalizer2WithImpl *)norm2)->getQuickCheck(c); + } else { + return UNORM_MAYBE; + } +} + +#endif // !UCONFIG_NO_NORMALIZATION diff --git a/intl/icu/source/common/localebuilder.cpp b/intl/icu/source/common/localebuilder.cpp new file mode 100644 index 0000000000..c1e1f2ad68 --- /dev/null +++ b/intl/icu/source/common/localebuilder.cpp @@ -0,0 +1,468 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +#include + +#include "bytesinkutil.h" // CharStringByteSink +#include "charstr.h" +#include "cstring.h" +#include "ulocimp.h" +#include "unicode/localebuilder.h" +#include "unicode/locid.h" + +U_NAMESPACE_BEGIN + +#define UPRV_ISDIGIT(c) (((c) >= '0') && ((c) <= '9')) +#define UPRV_ISALPHANUM(c) (uprv_isASCIILetter(c) || UPRV_ISDIGIT(c) ) + +constexpr const char* kAttributeKey = "attribute"; + +static bool _isExtensionSubtags(char key, const char* s, int32_t len) { + switch (uprv_tolower(key)) { + case 'u': + return ultag_isUnicodeExtensionSubtags(s, len); + case 't': + return ultag_isTransformedExtensionSubtags(s, len); + case 'x': + return ultag_isPrivateuseValueSubtags(s, len); + default: + return ultag_isExtensionSubtags(s, len); + } +} + +LocaleBuilder::LocaleBuilder() : UObject(), status_(U_ZERO_ERROR), language_(), + script_(), region_(), variant_(nullptr), extensions_(nullptr) +{ + language_[0] = 0; + script_[0] = 0; + region_[0] = 0; +} + +LocaleBuilder::~LocaleBuilder() +{ + delete variant_; + delete extensions_; +} + +LocaleBuilder& LocaleBuilder::setLocale(const Locale& locale) +{ + clear(); + setLanguage(locale.getLanguage()); + setScript(locale.getScript()); + setRegion(locale.getCountry()); + setVariant(locale.getVariant()); + extensions_ = locale.clone(); + if (extensions_ == nullptr) { + status_ = U_MEMORY_ALLOCATION_ERROR; + } + return *this; +} + +LocaleBuilder& LocaleBuilder::setLanguageTag(StringPiece tag) +{ + Locale l = Locale::forLanguageTag(tag, status_); + if (U_FAILURE(status_)) { return *this; } + // Because setLocale will reset status_ we need to return + // first if we have error in forLanguageTag. + setLocale(l); + return *this; +} + +static void setField(StringPiece input, char* dest, UErrorCode& errorCode, + UBool (*test)(const char*, int32_t)) { + if (U_FAILURE(errorCode)) { return; } + if (input.empty()) { + dest[0] = '\0'; + } else if (test(input.data(), input.length())) { + uprv_memcpy(dest, input.data(), input.length()); + dest[input.length()] = '\0'; + } else { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + } +} + +LocaleBuilder& LocaleBuilder::setLanguage(StringPiece language) +{ + setField(language, language_, status_, &ultag_isLanguageSubtag); + return *this; +} + +LocaleBuilder& LocaleBuilder::setScript(StringPiece script) +{ + setField(script, script_, status_, &ultag_isScriptSubtag); + return *this; +} + +LocaleBuilder& LocaleBuilder::setRegion(StringPiece region) +{ + setField(region, region_, status_, &ultag_isRegionSubtag); + return *this; +} + +static void transform(char* data, int32_t len) { + for (int32_t i = 0; i < len; i++, data++) { + if (*data == '_') { + *data = '-'; + } else { + *data = uprv_tolower(*data); + } + } +} + +LocaleBuilder& LocaleBuilder::setVariant(StringPiece variant) +{ + if (U_FAILURE(status_)) { return *this; } + if (variant.empty()) { + delete variant_; + variant_ = nullptr; + return *this; + } + CharString* new_variant = new CharString(variant, status_); + if (U_FAILURE(status_)) { return *this; } + if (new_variant == nullptr) { + status_ = U_MEMORY_ALLOCATION_ERROR; + return *this; + } + transform(new_variant->data(), new_variant->length()); + if (!ultag_isVariantSubtags(new_variant->data(), new_variant->length())) { + delete new_variant; + status_ = U_ILLEGAL_ARGUMENT_ERROR; + return *this; + } + delete variant_; + variant_ = new_variant; + return *this; +} + +static bool +_isKeywordValue(const char* key, const char* value, int32_t value_len) +{ + if (key[1] == '\0') { + // one char key + return (UPRV_ISALPHANUM(uprv_tolower(key[0])) && + _isExtensionSubtags(key[0], value, value_len)); + } else if (uprv_strcmp(key, kAttributeKey) == 0) { + // unicode attributes + return ultag_isUnicodeLocaleAttributes(value, value_len); + } + // otherwise: unicode extension value + // We need to convert from legacy key/value to unicode + // key/value + const char* unicode_locale_key = uloc_toUnicodeLocaleKey(key); + const char* unicode_locale_type = uloc_toUnicodeLocaleType(key, value); + + return unicode_locale_key && unicode_locale_type && + ultag_isUnicodeLocaleKey(unicode_locale_key, -1) && + ultag_isUnicodeLocaleType(unicode_locale_type, -1); +} + +static void +_copyExtensions(const Locale& from, icu::StringEnumeration *keywords, + Locale& to, bool validate, UErrorCode& errorCode) +{ + if (U_FAILURE(errorCode)) { return; } + LocalPointer ownedKeywords; + if (keywords == nullptr) { + ownedKeywords.adoptInstead(from.createKeywords(errorCode)); + if (U_FAILURE(errorCode) || ownedKeywords.isNull()) { return; } + keywords = ownedKeywords.getAlias(); + } + const char* key; + while ((key = keywords->next(nullptr, errorCode)) != nullptr) { + CharString value; + CharStringByteSink sink(&value); + from.getKeywordValue(key, sink, errorCode); + if (U_FAILURE(errorCode)) { return; } + if (uprv_strcmp(key, kAttributeKey) == 0) { + transform(value.data(), value.length()); + } + if (validate && + !_isKeywordValue(key, value.data(), value.length())) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + to.setKeywordValue(key, value.data(), errorCode); + if (U_FAILURE(errorCode)) { return; } + } +} + +void static +_clearUAttributesAndKeyType(Locale& locale, UErrorCode& errorCode) +{ + // Clear Unicode attributes + locale.setKeywordValue(kAttributeKey, "", errorCode); + + // Clear all Unicode keyword values + LocalPointer iter(locale.createUnicodeKeywords(errorCode)); + if (U_FAILURE(errorCode) || iter.isNull()) { return; } + const char* key; + while ((key = iter->next(nullptr, errorCode)) != nullptr) { + locale.setUnicodeKeywordValue(key, nullptr, errorCode); + } +} + +static void +_setUnicodeExtensions(Locale& locale, const CharString& value, UErrorCode& errorCode) +{ + // Add the unicode extensions to extensions_ + CharString locale_str("und-u-", errorCode); + locale_str.append(value, errorCode); + _copyExtensions( + Locale::forLanguageTag(locale_str.data(), errorCode), nullptr, + locale, false, errorCode); +} + +LocaleBuilder& LocaleBuilder::setExtension(char key, StringPiece value) +{ + if (U_FAILURE(status_)) { return *this; } + if (!UPRV_ISALPHANUM(key)) { + status_ = U_ILLEGAL_ARGUMENT_ERROR; + return *this; + } + CharString value_str(value, status_); + if (U_FAILURE(status_)) { return *this; } + transform(value_str.data(), value_str.length()); + if (!value_str.isEmpty() && + !_isExtensionSubtags(key, value_str.data(), value_str.length())) { + status_ = U_ILLEGAL_ARGUMENT_ERROR; + return *this; + } + if (extensions_ == nullptr) { + extensions_ = Locale::getRoot().clone(); + if (extensions_ == nullptr) { + status_ = U_MEMORY_ALLOCATION_ERROR; + return *this; + } + } + if (uprv_tolower(key) != 'u') { + // for t, x and others extension. + extensions_->setKeywordValue(StringPiece(&key, 1), value_str.data(), + status_); + return *this; + } + _clearUAttributesAndKeyType(*extensions_, status_); + if (U_FAILURE(status_)) { return *this; } + if (!value.empty()) { + _setUnicodeExtensions(*extensions_, value_str, status_); + } + return *this; +} + +LocaleBuilder& LocaleBuilder::setUnicodeLocaleKeyword( + StringPiece key, StringPiece type) +{ + if (U_FAILURE(status_)) { return *this; } + if (!ultag_isUnicodeLocaleKey(key.data(), key.length()) || + (!type.empty() && + !ultag_isUnicodeLocaleType(type.data(), type.length()))) { + status_ = U_ILLEGAL_ARGUMENT_ERROR; + return *this; + } + if (extensions_ == nullptr) { + extensions_ = Locale::getRoot().clone(); + if (extensions_ == nullptr) { + status_ = U_MEMORY_ALLOCATION_ERROR; + return *this; + } + } + extensions_->setUnicodeKeywordValue(key, type, status_); + return *this; +} + +LocaleBuilder& LocaleBuilder::addUnicodeLocaleAttribute( + StringPiece value) +{ + CharString value_str(value, status_); + if (U_FAILURE(status_)) { return *this; } + transform(value_str.data(), value_str.length()); + if (!ultag_isUnicodeLocaleAttribute(value_str.data(), value_str.length())) { + status_ = U_ILLEGAL_ARGUMENT_ERROR; + return *this; + } + if (extensions_ == nullptr) { + extensions_ = Locale::getRoot().clone(); + if (extensions_ == nullptr) { + status_ = U_MEMORY_ALLOCATION_ERROR; + return *this; + } + extensions_->setKeywordValue(kAttributeKey, value_str.data(), status_); + return *this; + } + + CharString attributes; + CharStringByteSink sink(&attributes); + UErrorCode localErrorCode = U_ZERO_ERROR; + extensions_->getKeywordValue(kAttributeKey, sink, localErrorCode); + if (U_FAILURE(localErrorCode)) { + CharString new_attributes(value_str.data(), status_); + // No attributes, set the attribute. + extensions_->setKeywordValue(kAttributeKey, new_attributes.data(), status_); + return *this; + } + + transform(attributes.data(),attributes.length()); + const char* start = attributes.data(); + const char* limit = attributes.data() + attributes.length(); + CharString new_attributes; + bool inserted = false; + while (start < limit) { + if (!inserted) { + int cmp = uprv_strcmp(start, value_str.data()); + if (cmp == 0) { return *this; } // Found it in attributes: Just return + if (cmp > 0) { + if (!new_attributes.isEmpty()) new_attributes.append('_', status_); + new_attributes.append(value_str.data(), status_); + inserted = true; + } + } + if (!new_attributes.isEmpty()) { + new_attributes.append('_', status_); + } + new_attributes.append(start, status_); + start += uprv_strlen(start) + 1; + } + if (!inserted) { + if (!new_attributes.isEmpty()) { + new_attributes.append('_', status_); + } + new_attributes.append(value_str.data(), status_); + } + // Not yet in the attributes, set the attribute. + extensions_->setKeywordValue(kAttributeKey, new_attributes.data(), status_); + return *this; +} + +LocaleBuilder& LocaleBuilder::removeUnicodeLocaleAttribute( + StringPiece value) +{ + CharString value_str(value, status_); + if (U_FAILURE(status_)) { return *this; } + transform(value_str.data(), value_str.length()); + if (!ultag_isUnicodeLocaleAttribute(value_str.data(), value_str.length())) { + status_ = U_ILLEGAL_ARGUMENT_ERROR; + return *this; + } + if (extensions_ == nullptr) { return *this; } + UErrorCode localErrorCode = U_ZERO_ERROR; + CharString attributes; + CharStringByteSink sink(&attributes); + extensions_->getKeywordValue(kAttributeKey, sink, localErrorCode); + // get failure, just return + if (U_FAILURE(localErrorCode)) { return *this; } + // Do not have any attributes, just return. + if (attributes.isEmpty()) { return *this; } + + char* p = attributes.data(); + // Replace null terminiator in place for _ and - so later + // we can use uprv_strcmp to compare. + for (int32_t i = 0; i < attributes.length(); i++, p++) { + *p = (*p == '_' || *p == '-') ? '\0' : uprv_tolower(*p); + } + + const char* start = attributes.data(); + const char* limit = attributes.data() + attributes.length(); + CharString new_attributes; + bool found = false; + while (start < limit) { + if (uprv_strcmp(start, value_str.data()) == 0) { + found = true; + } else { + if (!new_attributes.isEmpty()) { + new_attributes.append('_', status_); + } + new_attributes.append(start, status_); + } + start += uprv_strlen(start) + 1; + } + // Found the value in attributes, set the attribute. + if (found) { + extensions_->setKeywordValue(kAttributeKey, new_attributes.data(), status_); + } + return *this; +} + +LocaleBuilder& LocaleBuilder::clear() +{ + status_ = U_ZERO_ERROR; + language_[0] = 0; + script_[0] = 0; + region_[0] = 0; + delete variant_; + variant_ = nullptr; + clearExtensions(); + return *this; +} + +LocaleBuilder& LocaleBuilder::clearExtensions() +{ + delete extensions_; + extensions_ = nullptr; + return *this; +} + +Locale makeBogusLocale() { + Locale bogus; + bogus.setToBogus(); + return bogus; +} + +void LocaleBuilder::copyExtensionsFrom(const Locale& src, UErrorCode& errorCode) +{ + if (U_FAILURE(errorCode)) { return; } + LocalPointer keywords(src.createKeywords(errorCode)); + if (U_FAILURE(errorCode) || keywords.isNull() || keywords->count(errorCode) == 0) { + // Error, or no extensions to copy. + return; + } + if (extensions_ == nullptr) { + extensions_ = Locale::getRoot().clone(); + if (extensions_ == nullptr) { + status_ = U_MEMORY_ALLOCATION_ERROR; + return; + } + } + _copyExtensions(src, keywords.getAlias(), *extensions_, false, errorCode); +} + +Locale LocaleBuilder::build(UErrorCode& errorCode) +{ + if (U_FAILURE(errorCode)) { + return makeBogusLocale(); + } + if (U_FAILURE(status_)) { + errorCode = status_; + return makeBogusLocale(); + } + CharString locale_str(language_, errorCode); + if (uprv_strlen(script_) > 0) { + locale_str.append('-', errorCode).append(StringPiece(script_), errorCode); + } + if (uprv_strlen(region_) > 0) { + locale_str.append('-', errorCode).append(StringPiece(region_), errorCode); + } + if (variant_ != nullptr) { + locale_str.append('-', errorCode).append(StringPiece(variant_->data()), errorCode); + } + if (U_FAILURE(errorCode)) { + return makeBogusLocale(); + } + Locale product(locale_str.data()); + if (extensions_ != nullptr) { + _copyExtensions(*extensions_, nullptr, product, true, errorCode); + } + if (U_FAILURE(errorCode)) { + return makeBogusLocale(); + } + return product; +} + +UBool LocaleBuilder::copyErrorTo(UErrorCode &outErrorCode) const { + if (U_FAILURE(outErrorCode)) { + // Do not overwrite the older error code + return true; + } + outErrorCode = status_; + return U_FAILURE(outErrorCode); +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/localefallback_data.h b/intl/icu/source/common/localefallback_data.h new file mode 100644 index 0000000000..d2eb7eaafd --- /dev/null +++ b/intl/icu/source/common/localefallback_data.h @@ -0,0 +1,1350 @@ +// © 2022 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// +// Internal static data tables used by uresbund.cpp +// WARNING: This file is mechanically generated by the CLDR-to-ICU tool +// (see tools/cldr/cldr-to-icu/src/main/java/org/unicode/tool/cldrtoicu/generator/ResourcFallbackCodeGenerator.java). +// DO NOT HAND EDIT!!! + +#ifdef INCLUDED_FROM_URESBUND_CPP + +//====================================================================== +// Default script table +const char scriptCodeChars[] = + "Aghb\0Ahom\0Arab\0Armi\0Armn\0Avst\0Bamu\0Bass\0Batk\0Beng\0Bopo\0" + "Brah\0Cakm\0Cans\0Cari\0Cham\0Cher\0Chrs\0Copt\0Cprt\0Cyrl\0Deva\0" + "Egyp\0Elym\0Ethi\0Geor\0Gong\0Gonm\0Goth\0Gran\0Grek\0Gujr\0Guru\0" + "Hang\0Hani\0Hans\0Hant\0Hebr\0Hluw\0Hmnp\0Ital\0Java\0Jpan\0Kali\0" + "Kana\0Kawi\0Khar\0Khmr\0Kits\0Knda\0Kore\0Lana\0Laoo\0Lepc\0Lina\0" + "Linb\0Lisu\0Lyci\0Lydi\0Mand\0Mani\0Marc\0Medf\0Merc\0Mlym\0Modi\0" + "Mong\0Mroo\0Mtei\0Mymr\0Narb\0Newa\0Nkoo\0Nshu\0Ogam\0Olck\0Orkh\0" + "Orya\0Osge\0Ougr\0Pauc\0Phli\0Phnx\0Plrd\0Prti\0Rjng\0Rohg\0Runr\0" + "Samr\0Sarb\0Saur\0Sgnw\0Sinh\0Sogd\0Sora\0Soyo\0Syrc\0Tagb\0Takr\0" + "Tale\0Talu\0Taml\0Tang\0Tavt\0Telu\0Tfng\0Thaa\0Thai\0Tibt\0Tnsa\0" + "Toto\0Ugar\0Vaii\0Wcho\0Xpeo\0Xsux\0Yiii\0"; + +const char dsLocaleIDChars[] = + "aaf\0aao\0aat\0ab\0abh\0abl\0abq\0abv\0acm\0acq\0acw\0acx\0adf\0" + "adp\0adx\0ady\0ae\0aeb\0aec\0aee\0aeq\0afb\0agi\0agj\0agx\0ahg\0" + "aho\0ahr\0aib\0aij\0ain\0aio\0aiq\0ajp\0ajt\0akk\0akv\0alk\0all\0" + "alr\0alt\0alw\0am\0ams\0amw\0ani\0anp\0anr\0anu\0aot\0apc\0apd\0" + "aph\0aqc\0ar\0arc\0arq\0ars\0ary\0arz\0as\0ase\0ask\0atn\0atv\0" + "auj\0auz\0av\0avd\0avl\0awa\0awn\0axm\0ayh\0ayl\0ayn\0ayp\0az_IQ\0" + "az_IR\0az_RU\0azb\0ba\0bal\0bap\0bax\0bbl\0bcq\0bdv\0bdz\0be\0" + "bee\0bej\0bfb\0bfq\0bft\0bfu\0bfw\0bfy\0bfz\0bg\0bgc\0bgd\0bgn\0" + "bgp\0bgq\0bgw\0bgx\0bha\0bhb\0bhd\0bhe\0bhh\0bhi\0bhj\0bhm\0bhn\0" + "bho\0bht\0bhu\0biy\0bjf\0bji\0bjj\0bjm\0blk\0blt\0bmj\0bn\0bns\0" + "bo\0bph\0bpx\0bpy\0bqi\0bra\0brb\0brd\0brh\0brk\0brv\0brx\0bsh\0" + "bsk\0bsq\0bst\0btd\0btm\0btv\0bua\0bwe\0bxm\0bxu\0byh\0byn\0byw\0" + "bzi\0cbn\0ccp\0cde\0cdh\0cdi\0cdj\0cdm\0cdo\0cdz\0ce\0cgk\0chg\0" + "chm\0chr\0chx\0cih\0cja\0cji\0cjm\0cjy\0ckb\0ckt\0clh\0clw\0cmg\0" + "cna\0cnp\0cog\0cop\0cpg\0cr\0crh\0crj\0crk\0crl\0crm\0csh\0csp\0" + "csw\0ctd\0ctg\0ctn\0ctt\0cu\0cuu\0cv\0czh\0czk\0daq\0dar\0dcc\0" + "ddo\0def\0deh\0der\0dgl\0dhi\0dhn\0dho\0dhw\0dka\0dlg\0dmf\0dmk\0" + "dml\0dng\0dnu\0dnv\0doi\0dox\0dre\0drh\0drq\0drs\0dry\0dso\0dty\0" + "dub\0duh\0dus\0dv\0dwk\0dwz\0dz\0dzl\0ecr\0ecy\0egy\0eky\0el\0" + "emg\0emu\0enf\0enh\0era\0esg\0esh\0ett\0eve\0evn\0fa\0fay\0faz\0" + "fia\0fmu\0fub\0gan\0gaq\0gas\0gau\0gbj\0gbk\0gbl\0gbm\0gbz\0gdb\0" + "gdo\0gdx\0gez\0ggg\0ggn\0gha\0ghe\0ghr\0ght\0gig\0gin\0gjk\0gju\0" + "gld\0glh\0glk\0gmv\0gmy\0goe\0gof\0gok\0gom\0gon\0got\0gra\0grc\0" + "grt\0gru\0gu\0gvr\0gwc\0gwf\0gwt\0gyo\0gzi\0ha_CM\0ha_SD\0hac\0" + "hak\0har\0haz\0hbo\0hdy\0he\0hi\0hii\0hit\0hkh\0hlb\0hlu\0hmd\0" + "hmj\0hmq\0hnd\0hne\0hnj\0hnj_AU\0hnj_CN\0hnj_FR\0hnj_GF\0hnj_LA\0" + "hnj_MM\0hnj_SR\0hnj_TH\0hnj_US\0hnj_VN\0hno\0hoc\0hoh\0hoj\0how\0" + "hoy\0hpo\0hrt\0hrz\0hsn\0hss\0htx\0hut\0huy\0huz\0hy\0hyw\0ii\0" + "imy\0inh\0int\0ior\0iru\0isk\0itk\0itl\0iu\0iw\0ja\0jad\0jat\0" + "jbe\0jbn\0jct\0jda\0jdg\0jdt\0jee\0jge\0ji\0jje\0jkm\0jml\0jna\0" + "jnd\0jnl\0jns\0jog\0jpa\0jpr\0jul\0jun\0juy\0jya\0jye\0ka\0kaa\0" + "kap\0kaw\0kbd\0kbu\0kby\0kca\0kdq\0kdt\0ket\0kex\0key\0kfa\0kfb\0" + "kfc\0kfd\0kfe\0kfh\0kfi\0kfk\0kfm\0kfp\0kfq\0kfr\0kfs\0kfx\0kfy\0" + "kgj\0kgy\0khb\0khf\0khg\0khn\0kht\0khv\0khw\0kif\0kim\0kip\0kjg\0" + "kjh\0kjl\0kjo\0kjp\0kjt\0kk\0kk_AF\0kk_CN\0kk_IR\0kk_MN\0kkf\0" + "kkh\0kkt\0kle\0klj\0klr\0km\0kmj\0kmz\0kn\0ko\0koi\0kok\0kpt\0" + "kpy\0kqd\0kqy\0kra\0krc\0krk\0krr\0kru\0krv\0ks\0ksu\0ksw\0ksz\0" + "ktb\0ktl\0ktp\0ku_LB\0kuf\0kum\0kv\0kva\0kvq\0kvt\0kvx\0kvy\0" + "kxc\0kxf\0kxk\0kxl\0kxm\0kxp\0kxv\0ky\0ky_CN\0kyu\0kyv\0kyw\0" + "kzh\0lab\0lad\0lae\0lah\0lbc\0lbe\0lbf\0lbj\0lbm\0lbo\0lbr\0lcp\0" + "lep\0lez\0lhm\0lhs\0lif\0lis\0lkh\0lki\0lmh\0lmn\0lo\0loy\0lpo\0" + "lrc\0lrk\0lrl\0lsa\0lsd\0lss\0luk\0luu\0luv\0luz\0lwl\0lwm\0lya\0" + "lzh\0mag\0mai\0man_GN\0mby\0mde\0mdf\0mdx\0mdy\0mfa\0mfi\0mgp\0" + "mhj\0mid\0mjl\0mjq\0mjr\0mjt\0mju\0mjv\0mjz\0mk\0mkb\0mke\0mki\0" + "mkm\0ml\0mlf\0mn\0mn_CN\0mni\0mnj\0mns\0mnw\0mpz\0mr\0mra\0mrd\0" + "mrj\0mro\0mrr\0ms_CC\0mtm\0mtr\0mud\0muk\0mut\0muv\0muz\0mvf\0" + "mvy\0mvz\0mwr\0mwt\0mww\0my\0mym\0myv\0myz\0mzn\0nan\0nao\0ncd\0" + "ncq\0ndf\0ne\0neg\0neh\0nei\0new\0ngt\0nio\0nit\0niv\0nli\0nlm\0" + "nlx\0nmm\0nnp\0nod\0noe\0nog\0noi\0non\0nos\0npb\0nqo\0nsd\0nsf\0" + "nsk\0nst\0nsv\0nty\0ntz\0nwc\0nwx\0nyl\0nyq\0oaa\0oac\0oar\0oav\0" + "obm\0obr\0odk\0oht\0oj\0ojs\0okm\0oko\0okz\0ola\0ole\0omk\0omp\0" + "omr\0oon\0or\0ort\0oru\0orv\0os\0osa\0osc\0osi\0ota\0otb\0otk\0" + "oty\0oui\0pa\0pa_PK\0pal\0paq\0pbt\0pcb\0pce\0pcf\0pcg\0pch\0" + "pci\0pcj\0peg\0peo\0pgd\0pgg\0pgl\0pgn\0phd\0phk\0phl\0phn\0pho\0" + "phr\0pht\0phv\0phw\0pi\0pka\0pkr\0plk\0pll\0pmh\0pnt\0ppa\0pra\0" + "prc\0prd\0prp\0prt\0prx\0ps\0psh\0psi\0pst\0pum\0pwo\0pwr\0pww\0" + "pyx\0qxq\0raa\0rab\0raf\0rah\0raj\0rav\0rbb\0rdb\0rei\0rhg\0rji\0" + "rjs\0rka\0rki\0rkt\0rmi\0rmt\0rmz\0rsk\0rtw\0ru\0rue\0rut\0rwr\0" + "ryu\0sa\0sah\0sam\0sat\0saz\0sbn\0sbu\0sck\0scl\0scp\0sct\0scu\0" + "scx\0sd\0sd_IN\0sdb\0sdf\0sdg\0sdh\0sds\0sel\0sfm\0sga\0sgh\0" + "sgj\0sgr\0sgt\0sgw\0sgy\0shd\0shi\0shm\0shn\0shu\0shv\0si\0sia\0" + "sip\0siy\0siz\0sjd\0sjp\0sjt\0skb\0skj\0skr\0slq\0smh\0smp\0smu\0" + "smy\0soa\0sog\0soi\0sou\0spt\0spv\0sqo\0sqq\0sqt\0sr\0srb\0srh\0" + "srx\0srz\0ssh\0sss\0sts\0stv\0sty\0suz\0sva\0swb\0swi\0swv\0syc\0" + "syl\0syn\0syr\0syw\0ta\0tab\0taj\0tbk\0tcn\0tco\0tcx\0tcy\0tda\0" + "tdb\0tdd\0tdg\0tdh\0te\0tes\0tg\0tg_PK\0tge\0tgf\0th\0the\0thf\0" + "thi\0thl\0thm\0thq\0thr\0ths\0ti\0tig\0tij\0tin\0tjl\0tjo\0tkb\0" + "tks\0tkt\0tmk\0tmr\0tnv\0tov\0tpu\0tra\0trg\0trm\0trw\0tsd\0tsf\0" + "tsj\0tt\0tth\0tto\0tts\0tvn\0twm\0txg\0txo\0tyr\0tyv\0ude\0udg\0" + "udi\0udm\0ug\0ug_KZ\0ug_MN\0uga\0ugh\0ugo\0uk\0uki\0ulc\0unr\0" + "unr_NP\0unx\0ur\0urk\0ush\0uum\0uz_AF\0uz_CN\0uzs\0vaa\0vaf\0" + "vah\0vai\0vas\0vav\0vay\0vgr\0vmd\0vmh\0wal\0wbk\0wbq\0wbr\0wlo\0" + "wme\0wne\0wni\0wsg\0wsv\0wtm\0wuu\0xal\0xan\0xas\0xco\0xcr\0xdq\0" + "xhe\0xhm\0xis\0xka\0xkc\0xkj\0xkp\0xlc\0xld\0xly\0xmf\0xmn\0xmr\0" + "xna\0xnr\0xpg\0xpi\0xpm\0xpr\0xrm\0xrn\0xsa\0xsr\0xss\0xub\0xuj\0" + "xve\0xvi\0xwo\0xzh\0yai\0ybh\0ybi\0ydg\0yea\0yej\0yeu\0ygp\0yhd\0" + "yi\0yig\0yih\0yiv\0ykg\0yna\0ynk\0yoi\0yoy\0yrk\0ysd\0ysn\0ysp\0" + "ysr\0ysy\0yud\0yue\0yue_CN\0yug\0yux\0ywq\0ywu\0zau\0zba\0zch\0" + "zdj\0zeh\0zen\0zgb\0zgh\0zgm\0zgn\0zh\0zh_AU\0zh_BN\0zh_GB\0zh_GF\0" + "zh_HK\0zh_ID\0zh_MO\0zh_PA\0zh_PF\0zh_PH\0zh_SR\0zh_TH\0zh_TW\0" + "zh_US\0zh_VN\0zhd\0zhx\0zkb\0zko\0zkt\0zkz\0zlj\0zln\0zlq\0zqe\0" + "zrp\0zum\0zyg\0zyn\0zzj\0"; + +const int32_t defaultScriptTable[] = { + 0, 320, // aaf -> Mlym + 4, 10, // aao -> Arab + 8, 150, // aat -> Grek + 12, 100, // ab -> Cyrl + 15, 10, // abh -> Arab + 19, 425, // abl -> Rjng + 23, 100, // abq -> Cyrl + 27, 10, // abv -> Arab + 31, 10, // acm -> Arab + 35, 10, // acq -> Arab + 39, 10, // acw -> Arab + 43, 10, // acx -> Arab + 47, 10, // adf -> Arab + 51, 540, // adp -> Tibt + 55, 540, // adx -> Tibt + 59, 100, // ady -> Cyrl + 63, 25, // ae -> Avst + 66, 10, // aeb -> Arab + 70, 10, // aec -> Arab + 74, 10, // aee -> Arab + 78, 10, // aeq -> Arab + 82, 10, // afb -> Arab + 86, 105, // agi -> Deva + 90, 120, // agj -> Ethi + 94, 100, // agx -> Cyrl + 98, 120, // ahg -> Ethi + 102, 5, // aho -> Ahom + 106, 105, // ahr -> Deva + 110, 10, // aib -> Arab + 114, 185, // aij -> Hebr + 118, 220, // ain -> Kana + 122, 345, // aio -> Mymr + 126, 10, // aiq -> Arab + 130, 10, // ajp -> Arab + 134, 10, // ajt -> Arab + 138, 575, // akk -> Xsux + 142, 100, // akv -> Cyrl + 146, 260, // alk -> Laoo + 150, 320, // all -> Mlym + 154, 100, // alr -> Cyrl + 158, 100, // alt -> Cyrl + 162, 120, // alw -> Ethi + 166, 120, // am -> Ethi + 169, 210, // ams -> Jpan + 173, 480, // amw -> Syrc + 177, 100, // ani -> Cyrl + 181, 105, // anp -> Deva + 185, 105, // anr -> Deva + 189, 120, // anu -> Ethi + 193, 45, // aot -> Beng + 197, 10, // apc -> Arab + 201, 10, // apd -> Arab + 205, 105, // aph -> Deva + 209, 100, // aqc -> Cyrl + 213, 10, // ar -> Arab + 216, 15, // arc -> Armi + 220, 10, // arq -> Arab + 224, 10, // ars -> Arab + 228, 10, // ary -> Arab + 232, 10, // arz -> Arab + 236, 45, // as -> Beng + 239, 455, // ase -> Sgnw + 243, 10, // ask -> Arab + 247, 10, // atn -> Arab + 251, 100, // atv -> Cyrl + 255, 10, // auj -> Arab + 259, 10, // auz -> Arab + 263, 100, // av -> Cyrl + 266, 10, // avd -> Arab + 270, 10, // avl -> Arab + 274, 105, // awa -> Deva + 278, 120, // awn -> Ethi + 282, 20, // axm -> Armn + 286, 10, // ayh -> Arab + 290, 10, // ayl -> Arab + 294, 10, // ayn -> Arab + 298, 10, // ayp -> Arab + 302, 10, // az_IQ -> Arab + 308, 10, // az_IR -> Arab + 314, 100, // az_RU -> Cyrl + 320, 10, // azb -> Arab + 324, 100, // ba -> Cyrl + 327, 10, // bal -> Arab + 331, 105, // bap -> Deva + 335, 30, // bax -> Bamu + 339, 125, // bbl -> Geor + 343, 120, // bcq -> Ethi + 347, 385, // bdv -> Orya + 351, 10, // bdz -> Arab + 355, 100, // be -> Cyrl + 358, 105, // bee -> Deva + 362, 10, // bej -> Arab + 366, 105, // bfb -> Deva + 370, 505, // bfq -> Taml + 374, 10, // bft -> Arab + 378, 540, // bfu -> Tibt + 382, 385, // bfw -> Orya + 386, 105, // bfy -> Deva + 390, 105, // bfz -> Deva + 394, 100, // bg -> Cyrl + 397, 105, // bgc -> Deva + 401, 105, // bgd -> Deva + 405, 10, // bgn -> Arab + 409, 10, // bgp -> Arab + 413, 105, // bgq -> Deva + 417, 105, // bgw -> Deva + 421, 150, // bgx -> Grek + 425, 105, // bha -> Deva + 429, 105, // bhb -> Deva + 433, 105, // bhd -> Deva + 437, 10, // bhe -> Arab + 441, 100, // bhh -> Cyrl + 445, 105, // bhi -> Deva + 449, 105, // bhj -> Deva + 453, 10, // bhm -> Arab + 457, 480, // bhn -> Syrc + 461, 105, // bho -> Deva + 465, 490, // bht -> Takr + 469, 105, // bhu -> Deva + 473, 105, // biy -> Deva + 477, 480, // bjf -> Syrc + 481, 120, // bji -> Ethi + 485, 105, // bjj -> Deva + 489, 10, // bjm -> Arab + 493, 345, // blk -> Mymr + 497, 515, // blt -> Tavt + 501, 105, // bmj -> Deva + 505, 45, // bn -> Beng + 508, 105, // bns -> Deva + 512, 540, // bo -> Tibt + 515, 100, // bph -> Cyrl + 519, 105, // bpx -> Deva + 523, 45, // bpy -> Beng + 527, 10, // bqi -> Arab + 531, 105, // bra -> Deva + 535, 235, // brb -> Khmr + 539, 105, // brd -> Deva + 543, 10, // brh -> Arab + 547, 10, // brk -> Arab + 551, 260, // brv -> Laoo + 555, 105, // brx -> Deva + 559, 10, // bsh -> Arab + 563, 10, // bsk -> Arab + 567, 35, // bsq -> Bass + 571, 120, // bst -> Ethi + 575, 40, // btd -> Batk + 579, 40, // btm -> Batk + 583, 105, // btv -> Deva + 587, 100, // bua -> Cyrl + 591, 345, // bwe -> Mymr + 595, 100, // bxm -> Cyrl + 599, 330, // bxu -> Mong + 603, 105, // byh -> Deva + 607, 120, // byn -> Ethi + 611, 105, // byw -> Deva + 615, 535, // bzi -> Thai + 619, 535, // cbn -> Thai + 623, 60, // ccp -> Cakm + 627, 520, // cde -> Telu + 631, 105, // cdh -> Deva + 635, 155, // cdi -> Gujr + 639, 105, // cdj -> Deva + 643, 105, // cdm -> Deva + 647, 175, // cdo -> Hans + 651, 45, // cdz -> Beng + 655, 100, // ce -> Cyrl + 658, 540, // cgk -> Tibt + 662, 10, // chg -> Arab + 666, 100, // chm -> Cyrl + 670, 80, // chr -> Cher + 674, 105, // chx -> Deva + 678, 105, // cih -> Deva + 682, 10, // cja -> Arab + 686, 100, // cji -> Cyrl + 690, 75, // cjm -> Cham + 694, 175, // cjy -> Hans + 698, 10, // ckb -> Arab + 702, 100, // ckt -> Cyrl + 706, 10, // clh -> Arab + 710, 100, // clw -> Cyrl + 714, 475, // cmg -> Soyo + 718, 540, // cna -> Tibt + 722, 175, // cnp -> Hans + 726, 535, // cog -> Thai + 730, 90, // cop -> Copt + 734, 150, // cpg -> Grek + 738, 65, // cr -> Cans + 741, 100, // crh -> Cyrl + 745, 65, // crj -> Cans + 749, 65, // crk -> Cans + 753, 65, // crl -> Cans + 757, 65, // crm -> Cans + 761, 345, // csh -> Mymr + 765, 175, // csp -> Hans + 769, 65, // csw -> Cans + 773, 400, // ctd -> Pauc + 777, 45, // ctg -> Beng + 781, 105, // ctn -> Deva + 785, 505, // ctt -> Taml + 789, 100, // cu -> Cyrl + 792, 255, // cuu -> Lana + 796, 100, // cv -> Cyrl + 799, 175, // czh -> Hans + 803, 185, // czk -> Hebr + 807, 105, // daq -> Deva + 811, 100, // dar -> Cyrl + 815, 10, // dcc -> Arab + 819, 100, // ddo -> Cyrl + 823, 10, // def -> Arab + 827, 10, // deh -> Arab + 831, 45, // der -> Beng + 835, 10, // dgl -> Arab + 839, 105, // dhi -> Deva + 843, 155, // dhn -> Gujr + 847, 105, // dho -> Deva + 851, 105, // dhw -> Deva + 855, 540, // dka -> Tibt + 859, 100, // dlg -> Cyrl + 863, 310, // dmf -> Medf + 867, 10, // dmk -> Arab + 871, 10, // dml -> Arab + 875, 100, // dng -> Cyrl + 879, 345, // dnu -> Mymr + 883, 345, // dnv -> Mymr + 887, 105, // doi -> Deva + 891, 120, // dox -> Ethi + 895, 540, // dre -> Tibt + 899, 330, // drh -> Mong + 903, 105, // drq -> Deva + 907, 120, // drs -> Ethi + 911, 105, // dry -> Deva + 915, 385, // dso -> Orya + 919, 105, // dty -> Deva + 923, 155, // dub -> Gujr + 927, 105, // duh -> Deva + 931, 105, // dus -> Deva + 935, 530, // dv -> Thaa + 938, 385, // dwk -> Orya + 942, 105, // dwz -> Deva + 946, 540, // dz -> Tibt + 949, 540, // dzl -> Tibt + 953, 150, // ecr -> Grek + 957, 95, // ecy -> Cprt + 961, 110, // egy -> Egyp + 965, 215, // eky -> Kali + 969, 150, // el -> Grek + 972, 105, // emg -> Deva + 976, 105, // emu -> Deva + 980, 100, // enf -> Cyrl + 984, 100, // enh -> Cyrl + 988, 505, // era -> Taml + 992, 135, // esg -> Gonm + 996, 10, // esh -> Arab + 1000, 200, // ett -> Ital + 1004, 100, // eve -> Cyrl + 1008, 100, // evn -> Cyrl + 1012, 10, // fa -> Arab + 1015, 10, // fay -> Arab + 1019, 10, // faz -> Arab + 1023, 10, // fia -> Arab + 1027, 105, // fmu -> Deva + 1031, 10, // fub -> Arab + 1035, 175, // gan -> Hans + 1039, 385, // gaq -> Orya + 1043, 155, // gas -> Gujr + 1047, 520, // gau -> Telu + 1051, 385, // gbj -> Orya + 1055, 105, // gbk -> Deva + 1059, 155, // gbl -> Gujr + 1063, 105, // gbm -> Deva + 1067, 10, // gbz -> Arab + 1071, 385, // gdb -> Orya + 1075, 100, // gdo -> Cyrl + 1079, 105, // gdx -> Deva + 1083, 120, // gez -> Ethi + 1087, 10, // ggg -> Arab + 1091, 105, // ggn -> Deva + 1095, 10, // gha -> Arab + 1099, 105, // ghe -> Deva + 1103, 10, // ghr -> Arab + 1107, 540, // ght -> Tibt + 1111, 10, // gig -> Arab + 1115, 100, // gin -> Cyrl + 1119, 10, // gjk -> Arab + 1123, 10, // gju -> Arab + 1127, 100, // gld -> Cyrl + 1131, 10, // glh -> Arab + 1135, 10, // glk -> Arab + 1139, 120, // gmv -> Ethi + 1143, 275, // gmy -> Linb + 1147, 540, // goe -> Tibt + 1151, 120, // gof -> Ethi + 1155, 105, // gok -> Deva + 1159, 105, // gom -> Deva + 1163, 520, // gon -> Telu + 1167, 140, // got -> Goth + 1171, 105, // gra -> Deva + 1175, 95, // grc -> Cprt + 1179, 45, // grt -> Beng + 1183, 120, // gru -> Ethi + 1187, 155, // gu -> Gujr + 1190, 105, // gvr -> Deva + 1194, 10, // gwc -> Arab + 1198, 10, // gwf -> Arab + 1202, 10, // gwt -> Arab + 1206, 105, // gyo -> Deva + 1210, 10, // gzi -> Arab + 1214, 10, // ha_CM -> Arab + 1220, 10, // ha_SD -> Arab + 1226, 10, // hac -> Arab + 1230, 175, // hak -> Hans + 1234, 120, // har -> Ethi + 1238, 10, // haz -> Arab + 1242, 185, // hbo -> Hebr + 1246, 120, // hdy -> Ethi + 1250, 185, // he -> Hebr + 1253, 105, // hi -> Deva + 1256, 490, // hii -> Takr + 1260, 575, // hit -> Xsux + 1264, 10, // hkh -> Arab + 1268, 105, // hlb -> Deva + 1272, 190, // hlu -> Hluw + 1276, 415, // hmd -> Plrd + 1280, 50, // hmj -> Bopo + 1284, 50, // hmq -> Bopo + 1288, 10, // hnd -> Arab + 1292, 105, // hne -> Deva + 1296, 195, // hnj -> Hmnp + 1300, 260, // hnj_AU -> Laoo + 1307, 260, // hnj_CN -> Laoo + 1314, 260, // hnj_FR -> Laoo + 1321, 260, // hnj_GF -> Laoo + 1328, 260, // hnj_LA -> Laoo + 1335, 260, // hnj_MM -> Laoo + 1342, 260, // hnj_SR -> Laoo + 1349, 260, // hnj_TH -> Laoo + 1356, 195, // hnj_US -> Hmnp + 1363, 260, // hnj_VN -> Laoo + 1370, 10, // hno -> Arab + 1374, 105, // hoc -> Deva + 1378, 10, // hoh -> Arab + 1382, 105, // hoj -> Deva + 1386, 170, // how -> Hani + 1390, 105, // hoy -> Deva + 1394, 345, // hpo -> Mymr + 1398, 480, // hrt -> Syrc + 1402, 10, // hrz -> Arab + 1406, 175, // hsn -> Hans + 1410, 10, // hss -> Arab + 1414, 575, // htx -> Xsux + 1418, 105, // hut -> Deva + 1422, 185, // huy -> Hebr + 1426, 100, // huz -> Cyrl + 1430, 20, // hy -> Armn + 1433, 20, // hyw -> Armn + 1437, 580, // ii -> Yiii + 1440, 285, // imy -> Lyci + 1444, 100, // inh -> Cyrl + 1448, 345, // int -> Mymr + 1452, 120, // ior -> Ethi + 1456, 505, // iru -> Taml + 1460, 10, // isk -> Arab + 1464, 185, // itk -> Hebr + 1468, 100, // itl -> Cyrl + 1472, 65, // iu -> Cans + 1475, 185, // iw -> Hebr + 1478, 210, // ja -> Jpan + 1481, 10, // jad -> Arab + 1485, 10, // jat -> Arab + 1489, 185, // jbe -> Hebr + 1493, 10, // jbn -> Arab + 1497, 100, // jct -> Cyrl + 1501, 540, // jda -> Tibt + 1505, 10, // jdg -> Arab + 1509, 100, // jdt -> Cyrl + 1513, 105, // jee -> Deva + 1517, 125, // jge -> Geor + 1521, 185, // ji -> Hebr + 1524, 165, // jje -> Hang + 1528, 345, // jkm -> Mymr + 1532, 105, // jml -> Deva + 1536, 490, // jna -> Takr + 1540, 10, // jnd -> Arab + 1544, 105, // jnl -> Deva + 1548, 105, // jns -> Deva + 1552, 10, // jog -> Arab + 1556, 185, // jpa -> Hebr + 1560, 185, // jpr -> Hebr + 1564, 105, // jul -> Deva + 1568, 385, // jun -> Orya + 1572, 385, // juy -> Orya + 1576, 540, // jya -> Tibt + 1580, 185, // jye -> Hebr + 1584, 125, // ka -> Geor + 1587, 100, // kaa -> Cyrl + 1591, 100, // kap -> Cyrl + 1595, 225, // kaw -> Kawi + 1599, 100, // kbd -> Cyrl + 1603, 10, // kbu -> Arab + 1607, 10, // kby -> Arab + 1611, 100, // kca -> Cyrl + 1615, 45, // kdq -> Beng + 1619, 535, // kdt -> Thai + 1623, 100, // ket -> Cyrl + 1627, 105, // kex -> Deva + 1631, 520, // key -> Telu + 1635, 245, // kfa -> Knda + 1639, 105, // kfb -> Deva + 1643, 520, // kfc -> Telu + 1647, 245, // kfd -> Knda + 1651, 505, // kfe -> Taml + 1655, 320, // kfh -> Mlym + 1659, 505, // kfi -> Taml + 1663, 105, // kfk -> Deva + 1667, 10, // kfm -> Arab + 1671, 105, // kfp -> Deva + 1675, 105, // kfq -> Deva + 1679, 105, // kfr -> Deva + 1683, 105, // kfs -> Deva + 1687, 105, // kfx -> Deva + 1691, 105, // kfy -> Deva + 1695, 105, // kgj -> Deva + 1699, 105, // kgy -> Deva + 1703, 500, // khb -> Talu + 1707, 535, // khf -> Thai + 1711, 540, // khg -> Tibt + 1715, 105, // khn -> Deva + 1719, 345, // kht -> Mymr + 1723, 100, // khv -> Cyrl + 1727, 10, // khw -> Arab + 1731, 105, // kif -> Deva + 1735, 100, // kim -> Cyrl + 1739, 105, // kip -> Deva + 1743, 260, // kjg -> Laoo + 1747, 100, // kjh -> Cyrl + 1751, 105, // kjl -> Deva + 1755, 105, // kjo -> Deva + 1759, 345, // kjp -> Mymr + 1763, 535, // kjt -> Thai + 1767, 100, // kk -> Cyrl + 1770, 10, // kk_AF -> Arab + 1776, 10, // kk_CN -> Arab + 1782, 10, // kk_IR -> Arab + 1788, 10, // kk_MN -> Arab + 1794, 540, // kkf -> Tibt + 1798, 255, // kkh -> Lana + 1802, 105, // kkt -> Deva + 1806, 105, // kle -> Deva + 1810, 10, // klj -> Arab + 1814, 105, // klr -> Deva + 1818, 235, // km -> Khmr + 1821, 105, // kmj -> Deva + 1825, 10, // kmz -> Arab + 1829, 245, // kn -> Knda + 1832, 250, // ko -> Kore + 1835, 100, // koi -> Cyrl + 1839, 105, // kok -> Deva + 1843, 100, // kpt -> Cyrl + 1847, 100, // kpy -> Cyrl + 1851, 480, // kqd -> Syrc + 1855, 120, // kqy -> Ethi + 1859, 105, // kra -> Deva + 1863, 100, // krc -> Cyrl + 1867, 100, // krk -> Cyrl + 1871, 235, // krr -> Khmr + 1875, 105, // kru -> Deva + 1879, 235, // krv -> Khmr + 1883, 10, // ks -> Arab + 1886, 345, // ksu -> Mymr + 1890, 345, // ksw -> Mymr + 1894, 105, // ksz -> Deva + 1898, 120, // ktb -> Ethi + 1902, 10, // ktl -> Arab + 1906, 415, // ktp -> Plrd + 1910, 10, // ku_LB -> Arab + 1916, 260, // kuf -> Laoo + 1920, 100, // kum -> Cyrl + 1924, 100, // kv -> Cyrl + 1927, 100, // kva -> Cyrl + 1931, 345, // kvq -> Mymr + 1935, 345, // kvt -> Mymr + 1939, 10, // kvx -> Arab + 1943, 215, // kvy -> Kali + 1947, 120, // kxc -> Ethi + 1951, 345, // kxf -> Mymr + 1955, 345, // kxk -> Mymr + 1959, 105, // kxl -> Deva + 1963, 535, // kxm -> Thai + 1967, 10, // kxp -> Arab + 1971, 385, // kxv -> Orya + 1975, 100, // ky -> Cyrl + 1978, 10, // ky_CN -> Arab + 1984, 215, // kyu -> Kali + 1988, 105, // kyv -> Deva + 1992, 105, // kyw -> Deva + 1996, 10, // kzh -> Arab + 2000, 270, // lab -> Lina + 2004, 185, // lad -> Hebr + 2008, 105, // lae -> Deva + 2012, 10, // lah -> Arab + 2016, 280, // lbc -> Lisu + 2020, 100, // lbe -> Cyrl + 2024, 105, // lbf -> Deva + 2028, 540, // lbj -> Tibt + 2032, 105, // lbm -> Deva + 2036, 260, // lbo -> Laoo + 2040, 105, // lbr -> Deva + 2044, 535, // lcp -> Thai + 2048, 265, // lep -> Lepc + 2052, 100, // lez -> Cyrl + 2056, 105, // lhm -> Deva + 2060, 480, // lhs -> Syrc + 2064, 105, // lif -> Deva + 2068, 280, // lis -> Lisu + 2072, 540, // lkh -> Tibt + 2076, 10, // lki -> Arab + 2080, 105, // lmh -> Deva + 2084, 520, // lmn -> Telu + 2088, 260, // lo -> Laoo + 2091, 105, // loy -> Deva + 2095, 415, // lpo -> Plrd + 2099, 10, // lrc -> Arab + 2103, 10, // lrk -> Arab + 2107, 10, // lrl -> Arab + 2111, 10, // lsa -> Arab + 2115, 185, // lsd -> Hebr + 2119, 10, // lss -> Arab + 2123, 540, // luk -> Tibt + 2127, 105, // luu -> Deva + 2131, 10, // luv -> Arab + 2135, 10, // luz -> Arab + 2139, 535, // lwl -> Thai + 2143, 535, // lwm -> Thai + 2147, 540, // lya -> Tibt + 2151, 175, // lzh -> Hans + 2155, 105, // mag -> Deva + 2159, 105, // mai -> Deva + 2163, 360, // man_GN -> Nkoo + 2170, 10, // mby -> Arab + 2174, 10, // mde -> Arab + 2178, 100, // mdf -> Cyrl + 2182, 120, // mdx -> Ethi + 2186, 120, // mdy -> Ethi + 2190, 10, // mfa -> Arab + 2194, 10, // mfi -> Arab + 2198, 105, // mgp -> Deva + 2202, 10, // mhj -> Arab + 2206, 295, // mid -> Mand + 2210, 105, // mjl -> Deva + 2214, 320, // mjq -> Mlym + 2218, 320, // mjr -> Mlym + 2222, 105, // mjt -> Deva + 2226, 520, // mju -> Telu + 2230, 320, // mjv -> Mlym + 2234, 105, // mjz -> Deva + 2238, 100, // mk -> Cyrl + 2241, 105, // mkb -> Deva + 2245, 105, // mke -> Deva + 2249, 10, // mki -> Arab + 2253, 535, // mkm -> Thai + 2257, 320, // ml -> Mlym + 2260, 535, // mlf -> Thai + 2264, 100, // mn -> Cyrl + 2267, 330, // mn_CN -> Mong + 2273, 45, // mni -> Beng + 2277, 10, // mnj -> Arab + 2281, 100, // mns -> Cyrl + 2285, 345, // mnw -> Mymr + 2289, 535, // mpz -> Thai + 2293, 105, // mr -> Deva + 2296, 535, // mra -> Thai + 2300, 105, // mrd -> Deva + 2304, 100, // mrj -> Cyrl + 2308, 335, // mro -> Mroo + 2312, 105, // mrr -> Deva + 2316, 10, // ms_CC -> Arab + 2322, 100, // mtm -> Cyrl + 2326, 105, // mtr -> Deva + 2330, 100, // mud -> Cyrl + 2334, 540, // muk -> Tibt + 2338, 105, // mut -> Deva + 2342, 505, // muv -> Taml + 2346, 120, // muz -> Ethi + 2350, 330, // mvf -> Mong + 2354, 10, // mvy -> Arab + 2358, 120, // mvz -> Ethi + 2362, 105, // mwr -> Deva + 2366, 345, // mwt -> Mymr + 2370, 195, // mww -> Hmnp + 2374, 345, // my -> Mymr + 2377, 120, // mym -> Ethi + 2381, 100, // myv -> Cyrl + 2385, 295, // myz -> Mand + 2389, 10, // mzn -> Arab + 2393, 175, // nan -> Hans + 2397, 105, // nao -> Deva + 2401, 105, // ncd -> Deva + 2405, 260, // ncq -> Laoo + 2409, 100, // ndf -> Cyrl + 2413, 105, // ne -> Deva + 2416, 100, // neg -> Cyrl + 2420, 540, // neh -> Tibt + 2424, 575, // nei -> Xsux + 2428, 105, // new -> Deva + 2432, 260, // ngt -> Laoo + 2436, 100, // nio -> Cyrl + 2440, 520, // nit -> Telu + 2444, 100, // niv -> Cyrl + 2448, 10, // nli -> Arab + 2452, 10, // nlm -> Arab + 2456, 105, // nlx -> Deva + 2460, 105, // nmm -> Deva + 2464, 565, // nnp -> Wcho + 2468, 255, // nod -> Lana + 2472, 105, // noe -> Deva + 2476, 100, // nog -> Cyrl + 2480, 105, // noi -> Deva + 2484, 435, // non -> Runr + 2488, 580, // nos -> Yiii + 2492, 540, // npb -> Tibt + 2496, 360, // nqo -> Nkoo + 2500, 580, // nsd -> Yiii + 2504, 580, // nsf -> Yiii + 2508, 65, // nsk -> Cans + 2512, 545, // nst -> Tnsa + 2516, 580, // nsv -> Yiii + 2520, 580, // nty -> Yiii + 2524, 10, // ntz -> Arab + 2528, 355, // nwc -> Newa + 2532, 105, // nwx -> Deva + 2536, 535, // nyl -> Thai + 2540, 10, // nyq -> Arab + 2544, 100, // oaa -> Cyrl + 2548, 100, // oac -> Cyrl + 2552, 480, // oar -> Syrc + 2556, 125, // oav -> Geor + 2560, 410, // obm -> Phnx + 2564, 345, // obr -> Mymr + 2568, 10, // odk -> Arab + 2572, 575, // oht -> Xsux + 2576, 65, // oj -> Cans + 2579, 65, // ojs -> Cans + 2583, 165, // okm -> Hang + 2587, 170, // oko -> Hani + 2591, 235, // okz -> Khmr + 2595, 105, // ola -> Deva + 2599, 540, // ole -> Tibt + 2603, 100, // omk -> Cyrl + 2607, 340, // omp -> Mtei + 2611, 325, // omr -> Modi + 2615, 105, // oon -> Deva + 2619, 385, // or -> Orya + 2622, 520, // ort -> Telu + 2626, 10, // oru -> Arab + 2630, 100, // orv -> Cyrl + 2634, 100, // os -> Cyrl + 2637, 390, // osa -> Osge + 2641, 200, // osc -> Ital + 2645, 205, // osi -> Java + 2649, 10, // ota -> Arab + 2653, 540, // otb -> Tibt + 2657, 380, // otk -> Orkh + 2661, 145, // oty -> Gran + 2665, 395, // oui -> Ougr + 2669, 160, // pa -> Guru + 2672, 10, // pa_PK -> Arab + 2678, 405, // pal -> Phli + 2682, 100, // paq -> Cyrl + 2686, 10, // pbt -> Arab + 2690, 235, // pcb -> Khmr + 2694, 345, // pce -> Mymr + 2698, 320, // pcf -> Mlym + 2702, 320, // pcg -> Mlym + 2706, 105, // pch -> Deva + 2710, 105, // pci -> Deva + 2714, 520, // pcj -> Telu + 2718, 385, // peg -> Orya + 2722, 570, // peo -> Xpeo + 2726, 230, // pgd -> Khar + 2730, 105, // pgg -> Deva + 2734, 370, // pgl -> Ogam + 2738, 200, // pgn -> Ital + 2742, 105, // phd -> Deva + 2746, 345, // phk -> Mymr + 2750, 10, // phl -> Arab + 2754, 410, // phn -> Phnx + 2758, 260, // pho -> Laoo + 2762, 10, // phr -> Arab + 2766, 535, // pht -> Thai + 2770, 10, // phv -> Arab + 2774, 105, // phw -> Deva + 2778, 460, // pi -> Sinh + 2781, 55, // pka -> Brah + 2785, 320, // pkr -> Mlym + 2789, 10, // plk -> Arab + 2793, 345, // pll -> Mymr + 2797, 55, // pmh -> Brah + 2801, 150, // pnt -> Grek + 2805, 105, // ppa -> Deva + 2809, 230, // pra -> Khar + 2813, 10, // prc -> Arab + 2817, 10, // prd -> Arab + 2821, 155, // prp -> Gujr + 2825, 535, // prt -> Thai + 2829, 10, // prx -> Arab + 2833, 10, // ps -> Arab + 2836, 10, // psh -> Arab + 2840, 10, // psi -> Arab + 2844, 10, // pst -> Arab + 2848, 105, // pum -> Deva + 2852, 345, // pwo -> Mymr + 2856, 105, // pwr -> Deva + 2860, 535, // pww -> Thai + 2864, 345, // pyx -> Mymr + 2868, 10, // qxq -> Arab + 2872, 105, // raa -> Deva + 2876, 105, // rab -> Deva + 2880, 105, // raf -> Deva + 2884, 45, // rah -> Beng + 2888, 105, // raj -> Deva + 2892, 105, // rav -> Deva + 2896, 345, // rbb -> Mymr + 2900, 10, // rdb -> Arab + 2904, 385, // rei -> Orya + 2908, 430, // rhg -> Rohg + 2912, 105, // rji -> Deva + 2916, 105, // rjs -> Deva + 2920, 235, // rka -> Khmr + 2924, 345, // rki -> Mymr + 2928, 45, // rkt -> Beng + 2932, 20, // rmi -> Armn + 2936, 10, // rmt -> Arab + 2940, 345, // rmz -> Mymr + 2944, 100, // rsk -> Cyrl + 2948, 105, // rtw -> Deva + 2952, 100, // ru -> Cyrl + 2955, 100, // rue -> Cyrl + 2959, 100, // rut -> Cyrl + 2963, 105, // rwr -> Deva + 2967, 220, // ryu -> Kana + 2971, 105, // sa -> Deva + 2974, 100, // sah -> Cyrl + 2978, 440, // sam -> Samr + 2982, 375, // sat -> Olck + 2986, 450, // saz -> Saur + 2990, 10, // sbn -> Arab + 2994, 540, // sbu -> Tibt + 2998, 105, // sck -> Deva + 3002, 10, // scl -> Arab + 3006, 105, // scp -> Deva + 3010, 260, // sct -> Laoo + 3014, 490, // scu -> Takr + 3018, 150, // scx -> Grek + 3022, 10, // sd -> Arab + 3025, 105, // sd_IN -> Deva + 3031, 10, // sdb -> Arab + 3035, 10, // sdf -> Arab + 3039, 10, // sdg -> Arab + 3043, 10, // sdh -> Arab + 3047, 10, // sds -> Arab + 3051, 100, // sel -> Cyrl + 3055, 415, // sfm -> Plrd + 3059, 370, // sga -> Ogam + 3063, 100, // sgh -> Cyrl + 3067, 105, // sgj -> Deva + 3071, 10, // sgr -> Arab + 3075, 540, // sgt -> Tibt + 3079, 120, // sgw -> Ethi + 3083, 10, // sgy -> Arab + 3087, 10, // shd -> Arab + 3091, 525, // shi -> Tfng + 3095, 10, // shm -> Arab + 3099, 345, // shn -> Mymr + 3103, 10, // shu -> Arab + 3107, 10, // shv -> Arab + 3111, 460, // si -> Sinh + 3114, 100, // sia -> Cyrl + 3118, 540, // sip -> Tibt + 3122, 10, // siy -> Arab + 3126, 10, // siz -> Arab + 3130, 100, // sjd -> Cyrl + 3134, 105, // sjp -> Deva + 3138, 100, // sjt -> Cyrl + 3142, 535, // skb -> Thai + 3146, 105, // skj -> Deva + 3150, 10, // skr -> Arab + 3154, 10, // slq -> Arab + 3158, 580, // smh -> Yiii + 3162, 440, // smp -> Samr + 3166, 235, // smu -> Khmr + 3170, 10, // smy -> Arab + 3174, 515, // soa -> Tavt + 3178, 465, // sog -> Sogd + 3182, 105, // soi -> Deva + 3186, 535, // sou -> Thai + 3190, 540, // spt -> Tibt + 3194, 385, // spv -> Orya + 3198, 10, // sqo -> Arab + 3202, 260, // sqq -> Laoo + 3206, 10, // sqt -> Arab + 3210, 100, // sr -> Cyrl + 3213, 470, // srb -> Sora + 3217, 10, // srh -> Arab + 3221, 105, // srx -> Deva + 3225, 10, // srz -> Arab + 3229, 10, // ssh -> Arab + 3233, 260, // sss -> Laoo + 3237, 10, // sts -> Arab + 3241, 120, // stv -> Ethi + 3245, 100, // sty -> Cyrl + 3249, 105, // suz -> Deva + 3253, 125, // sva -> Geor + 3257, 10, // swb -> Arab + 3261, 170, // swi -> Hani + 3265, 105, // swv -> Deva + 3269, 480, // syc -> Syrc + 3273, 45, // syl -> Beng + 3277, 480, // syn -> Syrc + 3281, 480, // syr -> Syrc + 3285, 105, // syw -> Deva + 3289, 505, // ta -> Taml + 3292, 100, // tab -> Cyrl + 3296, 105, // taj -> Deva + 3300, 485, // tbk -> Tagb + 3304, 540, // tcn -> Tibt + 3308, 345, // tco -> Mymr + 3312, 505, // tcx -> Taml + 3316, 245, // tcy -> Knda + 3320, 525, // tda -> Tfng + 3324, 105, // tdb -> Deva + 3328, 495, // tdd -> Tale + 3332, 105, // tdg -> Deva + 3336, 105, // tdh -> Deva + 3340, 520, // te -> Telu + 3343, 205, // tes -> Java + 3347, 100, // tg -> Cyrl + 3350, 10, // tg_PK -> Arab + 3356, 105, // tge -> Deva + 3360, 540, // tgf -> Tibt + 3364, 535, // th -> Thai + 3367, 105, // the -> Deva + 3371, 105, // thf -> Deva + 3375, 495, // thi -> Tale + 3379, 105, // thl -> Deva + 3383, 535, // thm -> Thai + 3387, 105, // thq -> Deva + 3391, 105, // thr -> Deva + 3395, 105, // ths -> Deva + 3399, 120, // ti -> Ethi + 3402, 120, // tig -> Ethi + 3406, 105, // tij -> Deva + 3410, 100, // tin -> Cyrl + 3414, 345, // tjl -> Mymr + 3418, 10, // tjo -> Arab + 3422, 105, // tkb -> Deva + 3426, 10, // tks -> Arab + 3430, 105, // tkt -> Deva + 3434, 105, // tmk -> Deva + 3438, 480, // tmr -> Syrc + 3442, 60, // tnv -> Cakm + 3446, 10, // tov -> Arab + 3450, 235, // tpu -> Khmr + 3454, 10, // tra -> Arab + 3458, 185, // trg -> Hebr + 3462, 10, // trm -> Arab + 3466, 10, // trw -> Arab + 3470, 150, // tsd -> Grek + 3474, 105, // tsf -> Deva + 3478, 540, // tsj -> Tibt + 3482, 100, // tt -> Cyrl + 3485, 260, // tth -> Laoo + 3489, 260, // tto -> Laoo + 3493, 535, // tts -> Thai + 3497, 345, // tvn -> Mymr + 3501, 105, // twm -> Deva + 3505, 510, // txg -> Tang + 3509, 550, // txo -> Toto + 3513, 515, // tyr -> Tavt + 3517, 100, // tyv -> Cyrl + 3521, 100, // ude -> Cyrl + 3525, 320, // udg -> Mlym + 3529, 0, // udi -> Aghb + 3533, 100, // udm -> Cyrl + 3537, 10, // ug -> Arab + 3540, 100, // ug_KZ -> Cyrl + 3546, 100, // ug_MN -> Cyrl + 3552, 555, // uga -> Ugar + 3556, 100, // ugh -> Cyrl + 3560, 535, // ugo -> Thai + 3564, 100, // uk -> Cyrl + 3567, 385, // uki -> Orya + 3571, 100, // ulc -> Cyrl + 3575, 45, // unr -> Beng + 3579, 105, // unr_NP -> Deva + 3586, 45, // unx -> Beng + 3590, 10, // ur -> Arab + 3593, 535, // urk -> Thai + 3597, 10, // ush -> Arab + 3601, 150, // uum -> Grek + 3605, 10, // uz_AF -> Arab + 3611, 100, // uz_CN -> Cyrl + 3617, 10, // uzs -> Arab + 3621, 505, // vaa -> Taml + 3625, 10, // vaf -> Arab + 3629, 105, // vah -> Deva + 3633, 560, // vai -> Vaii + 3637, 105, // vas -> Deva + 3641, 105, // vav -> Deva + 3645, 105, // vay -> Deva + 3649, 10, // vgr -> Arab + 3653, 245, // vmd -> Knda + 3657, 10, // vmh -> Arab + 3661, 120, // wal -> Ethi + 3665, 10, // wbk -> Arab + 3669, 520, // wbq -> Telu + 3673, 105, // wbr -> Deva + 3677, 10, // wlo -> Arab + 3681, 105, // wme -> Deva + 3685, 10, // wne -> Arab + 3689, 10, // wni -> Arab + 3693, 130, // wsg -> Gong + 3697, 10, // wsv -> Arab + 3701, 105, // wtm -> Deva + 3705, 175, // wuu -> Hans + 3709, 100, // xal -> Cyrl + 3713, 120, // xan -> Ethi + 3717, 100, // xas -> Cyrl + 3721, 85, // xco -> Chrs + 3725, 70, // xcr -> Cari + 3729, 100, // xdq -> Cyrl + 3733, 10, // xhe -> Arab + 3737, 235, // xhm -> Khmr + 3741, 385, // xis -> Orya + 3745, 10, // xka -> Arab + 3749, 10, // xkc -> Arab + 3753, 10, // xkj -> Arab + 3757, 10, // xkp -> Arab + 3761, 285, // xlc -> Lyci + 3765, 290, // xld -> Lydi + 3769, 115, // xly -> Elym + 3773, 125, // xmf -> Geor + 3777, 300, // xmn -> Mani + 3781, 315, // xmr -> Merc + 3785, 350, // xna -> Narb + 3789, 105, // xnr -> Deva + 3793, 150, // xpg -> Grek + 3797, 370, // xpi -> Ogam + 3801, 100, // xpm -> Cyrl + 3805, 420, // xpr -> Prti + 3809, 100, // xrm -> Cyrl + 3813, 100, // xrn -> Cyrl + 3817, 445, // xsa -> Sarb + 3821, 105, // xsr -> Deva + 3825, 100, // xss -> Cyrl + 3829, 505, // xub -> Taml + 3833, 505, // xuj -> Taml + 3837, 200, // xve -> Ital + 3841, 10, // xvi -> Arab + 3845, 100, // xwo -> Cyrl + 3849, 305, // xzh -> Marc + 3853, 100, // yai -> Cyrl + 3857, 105, // ybh -> Deva + 3861, 105, // ybi -> Deva + 3865, 10, // ydg -> Arab + 3869, 320, // yea -> Mlym + 3873, 150, // yej -> Grek + 3877, 520, // yeu -> Telu + 3881, 415, // ygp -> Plrd + 3885, 185, // yhd -> Hebr + 3889, 185, // yi -> Hebr + 3892, 580, // yig -> Yiii + 3896, 185, // yih -> Hebr + 3900, 580, // yiv -> Yiii + 3904, 100, // ykg -> Cyrl + 3908, 415, // yna -> Plrd + 3912, 100, // ynk -> Cyrl + 3916, 210, // yoi -> Jpan + 3920, 535, // yoy -> Thai + 3924, 100, // yrk -> Cyrl + 3928, 580, // ysd -> Yiii + 3932, 580, // ysn -> Yiii + 3936, 580, // ysp -> Yiii + 3940, 100, // ysr -> Cyrl + 3944, 415, // ysy -> Plrd + 3948, 185, // yud -> Hebr + 3952, 180, // yue -> Hant + 3956, 175, // yue_CN -> Hans + 3963, 100, // yug -> Cyrl + 3967, 100, // yux -> Cyrl + 3971, 415, // ywq -> Plrd + 3975, 415, // ywu -> Plrd + 3979, 540, // zau -> Tibt + 3983, 10, // zba -> Arab + 3987, 170, // zch -> Hani + 3991, 10, // zdj -> Arab + 3995, 170, // zeh -> Hani + 3999, 525, // zen -> Tfng + 4003, 170, // zgb -> Hani + 4007, 525, // zgh -> Tfng + 4011, 170, // zgm -> Hani + 4015, 170, // zgn -> Hani + 4019, 175, // zh -> Hans + 4022, 180, // zh_AU -> Hant + 4028, 180, // zh_BN -> Hant + 4034, 180, // zh_GB -> Hant + 4040, 180, // zh_GF -> Hant + 4046, 180, // zh_HK -> Hant + 4052, 180, // zh_ID -> Hant + 4058, 180, // zh_MO -> Hant + 4064, 180, // zh_PA -> Hant + 4070, 180, // zh_PF -> Hant + 4076, 180, // zh_PH -> Hant + 4082, 180, // zh_SR -> Hant + 4088, 180, // zh_TH -> Hant + 4094, 180, // zh_TW -> Hant + 4100, 180, // zh_US -> Hant + 4106, 180, // zh_VN -> Hant + 4112, 170, // zhd -> Hani + 4116, 365, // zhx -> Nshu + 4120, 100, // zkb -> Cyrl + 4124, 100, // zko -> Cyrl + 4128, 240, // zkt -> Kits + 4132, 100, // zkz -> Cyrl + 4136, 170, // zlj -> Hani + 4140, 170, // zln -> Hani + 4144, 170, // zlq -> Hani + 4148, 170, // zqe -> Hani + 4152, 185, // zrp -> Hebr + 4156, 10, // zum -> Arab + 4160, 170, // zyg -> Hani + 4164, 170, // zyn -> Hani + 4168, 170, // zzj -> Hani +}; + +//====================================================================== +// Parent locale table +const char parentLocaleChars[] = + "az_Arab\0az_Cyrl\0bal_Latn\0blt_Latn\0bm_Nkoo\0bs_Cyrl\0byn_Latn\0" + "cu_Glag\0dje_Arab\0dyo_Arab\0en_001\0en_150\0en_AG\0en_AI\0en_AT\0" + "en_AU\0en_BB\0en_BE\0en_BM\0en_BS\0en_BW\0en_BZ\0en_CC\0en_CH\0" + "en_CK\0en_CM\0en_CX\0en_CY\0en_DE\0en_DG\0en_DK\0en_DM\0en_Dsrt\0" + "en_ER\0en_FI\0en_FJ\0en_FK\0en_FM\0en_GB\0en_GD\0en_GG\0en_GH\0" + "en_GI\0en_GM\0en_GY\0en_HK\0en_IE\0en_IL\0en_IM\0en_IN\0en_IO\0" + "en_JE\0en_JM\0en_KE\0en_KI\0en_KN\0en_KY\0en_LC\0en_LR\0en_LS\0" + "en_MG\0en_MO\0en_MS\0en_MT\0en_MU\0en_MV\0en_MW\0en_MY\0en_NA\0" + "en_NF\0en_NG\0en_NL\0en_NR\0en_NU\0en_NZ\0en_PG\0en_PK\0en_PN\0" + "en_PW\0en_RW\0en_SB\0en_SC\0en_SD\0en_SE\0en_SG\0en_SH\0en_SI\0" + "en_SL\0en_SS\0en_SX\0en_SZ\0en_Shaw\0en_TC\0en_TK\0en_TO\0en_TT\0" + "en_TV\0en_TZ\0en_UG\0en_VC\0en_VG\0en_VU\0en_WS\0en_ZA\0en_ZM\0" + "en_ZW\0es_419\0es_AR\0es_BO\0es_BR\0es_BZ\0es_CL\0es_CO\0es_CR\0" + "es_CU\0es_DO\0es_EC\0es_GT\0es_HN\0es_MX\0es_NI\0es_PA\0es_PE\0" + "es_PR\0es_PY\0es_SV\0es_US\0es_UY\0es_VE\0ff_Adlm\0ff_Arab\0fr_HT\0" + "ha_Arab\0hi_Latn\0ht\0iu_Latn\0kk_Arab\0ks_Deva\0ku_Arab\0ky_Arab\0" + "ky_Latn\0ml_Arab\0mn_Mong\0mni_Mtei\0ms_Arab\0nb\0nn\0no\0no_NO\0" + "pa_Arab\0pt_AO\0pt_CH\0pt_CV\0pt_FR\0pt_GQ\0pt_GW\0pt_LU\0pt_MO\0" + "pt_MZ\0pt_PT\0pt_ST\0pt_TL\0root\0sat_Deva\0sd_Deva\0sd_Khoj\0" + "sd_Sind\0shi_Latn\0so_Arab\0sr_Latn\0sw_Arab\0tg_Arab\0ug_Cyrl\0" + "uz_Arab\0uz_Cyrl\0vai_Latn\0wo_Arab\0yo_Arab\0yue_Hans\0zh_Hant\0" + "zh_Hant_HK\0zh_Hant_MO\0"; + +const int32_t parentLocaleTable[] = { + 0, 1023, // az_Arab -> root + 8, 1023, // az_Cyrl -> root + 16, 1023, // bal_Latn -> root + 25, 1023, // blt_Latn -> root + 34, 1023, // bm_Nkoo -> root + 42, 1023, // bs_Cyrl -> root + 50, 1023, // byn_Latn -> root + 59, 1023, // cu_Glag -> root + 67, 1023, // dje_Arab -> root + 76, 1023, // dyo_Arab -> root + 92, 85, // en_150 -> en_001 + 99, 85, // en_AG -> en_001 + 105, 85, // en_AI -> en_001 + 111, 92, // en_AT -> en_150 + 117, 85, // en_AU -> en_001 + 123, 85, // en_BB -> en_001 + 129, 92, // en_BE -> en_150 + 135, 85, // en_BM -> en_001 + 141, 85, // en_BS -> en_001 + 147, 85, // en_BW -> en_001 + 153, 85, // en_BZ -> en_001 + 159, 85, // en_CC -> en_001 + 165, 92, // en_CH -> en_150 + 171, 85, // en_CK -> en_001 + 177, 85, // en_CM -> en_001 + 183, 85, // en_CX -> en_001 + 189, 85, // en_CY -> en_001 + 195, 92, // en_DE -> en_150 + 201, 85, // en_DG -> en_001 + 207, 92, // en_DK -> en_150 + 213, 85, // en_DM -> en_001 + 219, 1023, // en_Dsrt -> root + 227, 85, // en_ER -> en_001 + 233, 92, // en_FI -> en_150 + 239, 85, // en_FJ -> en_001 + 245, 85, // en_FK -> en_001 + 251, 85, // en_FM -> en_001 + 257, 85, // en_GB -> en_001 + 263, 85, // en_GD -> en_001 + 269, 85, // en_GG -> en_001 + 275, 85, // en_GH -> en_001 + 281, 85, // en_GI -> en_001 + 287, 85, // en_GM -> en_001 + 293, 85, // en_GY -> en_001 + 299, 85, // en_HK -> en_001 + 305, 85, // en_IE -> en_001 + 311, 85, // en_IL -> en_001 + 317, 85, // en_IM -> en_001 + 323, 85, // en_IN -> en_001 + 329, 85, // en_IO -> en_001 + 335, 85, // en_JE -> en_001 + 341, 85, // en_JM -> en_001 + 347, 85, // en_KE -> en_001 + 353, 85, // en_KI -> en_001 + 359, 85, // en_KN -> en_001 + 365, 85, // en_KY -> en_001 + 371, 85, // en_LC -> en_001 + 377, 85, // en_LR -> en_001 + 383, 85, // en_LS -> en_001 + 389, 85, // en_MG -> en_001 + 395, 85, // en_MO -> en_001 + 401, 85, // en_MS -> en_001 + 407, 85, // en_MT -> en_001 + 413, 85, // en_MU -> en_001 + 419, 85, // en_MV -> en_001 + 425, 85, // en_MW -> en_001 + 431, 85, // en_MY -> en_001 + 437, 85, // en_NA -> en_001 + 443, 85, // en_NF -> en_001 + 449, 85, // en_NG -> en_001 + 455, 92, // en_NL -> en_150 + 461, 85, // en_NR -> en_001 + 467, 85, // en_NU -> en_001 + 473, 85, // en_NZ -> en_001 + 479, 85, // en_PG -> en_001 + 485, 85, // en_PK -> en_001 + 491, 85, // en_PN -> en_001 + 497, 85, // en_PW -> en_001 + 503, 85, // en_RW -> en_001 + 509, 85, // en_SB -> en_001 + 515, 85, // en_SC -> en_001 + 521, 85, // en_SD -> en_001 + 527, 92, // en_SE -> en_150 + 533, 85, // en_SG -> en_001 + 539, 85, // en_SH -> en_001 + 545, 92, // en_SI -> en_150 + 551, 85, // en_SL -> en_001 + 557, 85, // en_SS -> en_001 + 563, 85, // en_SX -> en_001 + 569, 85, // en_SZ -> en_001 + 575, 1023, // en_Shaw -> root + 583, 85, // en_TC -> en_001 + 589, 85, // en_TK -> en_001 + 595, 85, // en_TO -> en_001 + 601, 85, // en_TT -> en_001 + 607, 85, // en_TV -> en_001 + 613, 85, // en_TZ -> en_001 + 619, 85, // en_UG -> en_001 + 625, 85, // en_VC -> en_001 + 631, 85, // en_VG -> en_001 + 637, 85, // en_VU -> en_001 + 643, 85, // en_WS -> en_001 + 649, 85, // en_ZA -> en_001 + 655, 85, // en_ZM -> en_001 + 661, 85, // en_ZW -> en_001 + 674, 667, // es_AR -> es_419 + 680, 667, // es_BO -> es_419 + 686, 667, // es_BR -> es_419 + 692, 667, // es_BZ -> es_419 + 698, 667, // es_CL -> es_419 + 704, 667, // es_CO -> es_419 + 710, 667, // es_CR -> es_419 + 716, 667, // es_CU -> es_419 + 722, 667, // es_DO -> es_419 + 728, 667, // es_EC -> es_419 + 734, 667, // es_GT -> es_419 + 740, 667, // es_HN -> es_419 + 746, 667, // es_MX -> es_419 + 752, 667, // es_NI -> es_419 + 758, 667, // es_PA -> es_419 + 764, 667, // es_PE -> es_419 + 770, 667, // es_PR -> es_419 + 776, 667, // es_PY -> es_419 + 782, 667, // es_SV -> es_419 + 788, 667, // es_US -> es_419 + 794, 667, // es_UY -> es_419 + 800, 667, // es_VE -> es_419 + 806, 1023, // ff_Adlm -> root + 814, 1023, // ff_Arab -> root + 828, 1023, // ha_Arab -> root + 836, 323, // hi_Latn -> en_IN + 844, 822, // ht -> fr_HT + 847, 1023, // iu_Latn -> root + 855, 1023, // kk_Arab -> root + 863, 1023, // ks_Deva -> root + 871, 1023, // ku_Arab -> root + 879, 1023, // ky_Arab -> root + 887, 1023, // ky_Latn -> root + 895, 1023, // ml_Arab -> root + 903, 1023, // mn_Mong -> root + 911, 1023, // mni_Mtei -> root + 920, 1023, // ms_Arab -> root + 928, 934, // nb -> no + 931, 934, // nn -> no + 937, 934, // no_NO -> no + 943, 1023, // pa_Arab -> root + 951, 1005, // pt_AO -> pt_PT + 957, 1005, // pt_CH -> pt_PT + 963, 1005, // pt_CV -> pt_PT + 969, 1005, // pt_FR -> pt_PT + 975, 1005, // pt_GQ -> pt_PT + 981, 1005, // pt_GW -> pt_PT + 987, 1005, // pt_LU -> pt_PT + 993, 1005, // pt_MO -> pt_PT + 999, 1005, // pt_MZ -> pt_PT + 1011, 1005, // pt_ST -> pt_PT + 1017, 1005, // pt_TL -> pt_PT + 1028, 1023, // sat_Deva -> root + 1037, 1023, // sd_Deva -> root + 1045, 1023, // sd_Khoj -> root + 1053, 1023, // sd_Sind -> root + 1061, 1023, // shi_Latn -> root + 1070, 1023, // so_Arab -> root + 1078, 1023, // sr_Latn -> root + 1086, 1023, // sw_Arab -> root + 1094, 1023, // tg_Arab -> root + 1102, 1023, // ug_Cyrl -> root + 1110, 1023, // uz_Arab -> root + 1118, 1023, // uz_Cyrl -> root + 1126, 1023, // vai_Latn -> root + 1135, 1023, // wo_Arab -> root + 1143, 1023, // yo_Arab -> root + 1151, 1023, // yue_Hans -> root + 1160, 1023, // zh_Hant -> root + 1179, 1168, // zh_Hant_MO -> zh_Hant_HK +}; + + +#endif // INCLUDED_FROM_URESBUND_CPP diff --git a/intl/icu/source/common/localematcher.cpp b/intl/icu/source/common/localematcher.cpp new file mode 100644 index 0000000000..5f8c703df7 --- /dev/null +++ b/intl/icu/source/common/localematcher.cpp @@ -0,0 +1,826 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// localematcher.cpp +// created: 2019may08 Markus W. Scherer + +#include "unicode/utypes.h" +#include "unicode/localebuilder.h" +#include "unicode/localematcher.h" +#include "unicode/locid.h" +#include "unicode/stringpiece.h" +#include "unicode/uloc.h" +#include "unicode/uobject.h" +#include "cstring.h" +#include "localeprioritylist.h" +#include "loclikelysubtags.h" +#include "locdistance.h" +#include "lsr.h" +#include "uassert.h" +#include "uhash.h" +#include "ustr_imp.h" +#include "uvector.h" + +#define UND_LSR LSR("und", "", "", LSR::EXPLICIT_LSR) + +/** + * Indicator for the lifetime of desired-locale objects passed into the LocaleMatcher. + * + * @draft ICU 65 + */ +enum ULocMatchLifetime { + /** + * Locale objects are temporary. + * The matcher will make a copy of a locale that will be used beyond one function call. + * + * @draft ICU 65 + */ + ULOCMATCH_TEMPORARY_LOCALES, + /** + * Locale objects are stored at least as long as the matcher is used. + * The matcher will keep only a pointer to a locale that will be used beyond one function call, + * avoiding a copy. + * + * @draft ICU 65 + */ + ULOCMATCH_STORED_LOCALES // TODO: permanent? cached? clone? +}; +#ifndef U_IN_DOXYGEN +typedef enum ULocMatchLifetime ULocMatchLifetime; +#endif + +U_NAMESPACE_BEGIN + +LocaleMatcher::Result::Result(LocaleMatcher::Result &&src) noexcept : + desiredLocale(src.desiredLocale), + supportedLocale(src.supportedLocale), + desiredIndex(src.desiredIndex), + supportedIndex(src.supportedIndex), + desiredIsOwned(src.desiredIsOwned) { + if (desiredIsOwned) { + src.desiredLocale = nullptr; + src.desiredIndex = -1; + src.desiredIsOwned = false; + } +} + +LocaleMatcher::Result::~Result() { + if (desiredIsOwned) { + delete desiredLocale; + } +} + +LocaleMatcher::Result &LocaleMatcher::Result::operator=(LocaleMatcher::Result &&src) noexcept { + this->~Result(); + + desiredLocale = src.desiredLocale; + supportedLocale = src.supportedLocale; + desiredIndex = src.desiredIndex; + supportedIndex = src.supportedIndex; + desiredIsOwned = src.desiredIsOwned; + + if (desiredIsOwned) { + src.desiredLocale = nullptr; + src.desiredIndex = -1; + src.desiredIsOwned = false; + } + return *this; +} + +Locale LocaleMatcher::Result::makeResolvedLocale(UErrorCode &errorCode) const { + if (U_FAILURE(errorCode) || supportedLocale == nullptr) { + return Locale::getRoot(); + } + const Locale *bestDesired = getDesiredLocale(); + if (bestDesired == nullptr || *supportedLocale == *bestDesired) { + return *supportedLocale; + } + LocaleBuilder b; + b.setLocale(*supportedLocale); + + // Copy the region from bestDesired, if there is one. + const char *region = bestDesired->getCountry(); + if (*region != 0) { + b.setRegion(region); + } + + // Copy the variants from bestDesired, if there are any. + // Note that this will override any supportedLocale variants. + // For example, "sco-ulster-fonipa" + "...-fonupa" => "sco-fonupa" (replacing ulster). + const char *variants = bestDesired->getVariant(); + if (*variants != 0) { + b.setVariant(variants); + } + + // Copy the extensions from bestDesired, if there are any. + // C++ note: The following note, copied from Java, may not be true, + // as long as C++ copies by legacy ICU keyword, not by extension singleton. + // Note that this will override any supportedLocale extensions. + // For example, "th-u-nu-latn-ca-buddhist" + "...-u-nu-native" => "th-u-nu-native" + // (replacing calendar). + b.copyExtensionsFrom(*bestDesired, errorCode); + return b.build(errorCode); +} + +LocaleMatcher::Builder::Builder(LocaleMatcher::Builder &&src) noexcept : + errorCode_(src.errorCode_), + supportedLocales_(src.supportedLocales_), + thresholdDistance_(src.thresholdDistance_), + demotion_(src.demotion_), + defaultLocale_(src.defaultLocale_), + withDefault_(src.withDefault_), + favor_(src.favor_), + direction_(src.direction_) { + src.supportedLocales_ = nullptr; + src.defaultLocale_ = nullptr; +} + +LocaleMatcher::Builder::~Builder() { + delete supportedLocales_; + delete defaultLocale_; + delete maxDistanceDesired_; + delete maxDistanceSupported_; +} + +LocaleMatcher::Builder &LocaleMatcher::Builder::operator=(LocaleMatcher::Builder &&src) noexcept { + this->~Builder(); + + errorCode_ = src.errorCode_; + supportedLocales_ = src.supportedLocales_; + thresholdDistance_ = src.thresholdDistance_; + demotion_ = src.demotion_; + defaultLocale_ = src.defaultLocale_; + withDefault_ = src.withDefault_, + favor_ = src.favor_; + direction_ = src.direction_; + + src.supportedLocales_ = nullptr; + src.defaultLocale_ = nullptr; + return *this; +} + +void LocaleMatcher::Builder::clearSupportedLocales() { + if (supportedLocales_ != nullptr) { + supportedLocales_->removeAllElements(); + } +} + +bool LocaleMatcher::Builder::ensureSupportedLocaleVector() { + if (U_FAILURE(errorCode_)) { return false; } + if (supportedLocales_ != nullptr) { return true; } + LocalPointer lpSupportedLocales(new UVector(uprv_deleteUObject, nullptr, errorCode_), errorCode_); + if (U_FAILURE(errorCode_)) { return false; } + supportedLocales_ = lpSupportedLocales.orphan(); + return true; +} + +LocaleMatcher::Builder &LocaleMatcher::Builder::setSupportedLocalesFromListString( + StringPiece locales) { + LocalePriorityList list(locales, errorCode_); + if (U_FAILURE(errorCode_)) { return *this; } + clearSupportedLocales(); + if (!ensureSupportedLocaleVector()) { return *this; } + int32_t length = list.getLengthIncludingRemoved(); + for (int32_t i = 0; i < length; ++i) { + Locale *locale = list.orphanLocaleAt(i); + if (locale == nullptr) { continue; } + supportedLocales_->adoptElement(locale, errorCode_); + if (U_FAILURE(errorCode_)) { + break; + } + } + return *this; +} + +LocaleMatcher::Builder &LocaleMatcher::Builder::setSupportedLocales(Locale::Iterator &locales) { + if (ensureSupportedLocaleVector()) { + clearSupportedLocales(); + while (locales.hasNext() && U_SUCCESS(errorCode_)) { + const Locale &locale = locales.next(); + LocalPointer clone (locale.clone(), errorCode_); + supportedLocales_->adoptElement(clone.orphan(), errorCode_); + } + } + return *this; +} + +LocaleMatcher::Builder &LocaleMatcher::Builder::addSupportedLocale(const Locale &locale) { + if (ensureSupportedLocaleVector()) { + LocalPointer clone(locale.clone(), errorCode_); + supportedLocales_->adoptElement(clone.orphan(), errorCode_); + } + return *this; +} + +LocaleMatcher::Builder &LocaleMatcher::Builder::setNoDefaultLocale() { + if (U_FAILURE(errorCode_)) { return *this; } + delete defaultLocale_; + defaultLocale_ = nullptr; + withDefault_ = false; + return *this; +} + +LocaleMatcher::Builder &LocaleMatcher::Builder::setDefaultLocale(const Locale *defaultLocale) { + if (U_FAILURE(errorCode_)) { return *this; } + Locale *clone = nullptr; + if (defaultLocale != nullptr) { + clone = defaultLocale->clone(); + if (clone == nullptr) { + errorCode_ = U_MEMORY_ALLOCATION_ERROR; + return *this; + } + } + delete defaultLocale_; + defaultLocale_ = clone; + withDefault_ = true; + return *this; +} + +LocaleMatcher::Builder &LocaleMatcher::Builder::setFavorSubtag(ULocMatchFavorSubtag subtag) { + if (U_FAILURE(errorCode_)) { return *this; } + favor_ = subtag; + return *this; +} + +LocaleMatcher::Builder &LocaleMatcher::Builder::setDemotionPerDesiredLocale(ULocMatchDemotion demotion) { + if (U_FAILURE(errorCode_)) { return *this; } + demotion_ = demotion; + return *this; +} + +LocaleMatcher::Builder &LocaleMatcher::Builder::setMaxDistance(const Locale &desired, + const Locale &supported) { + if (U_FAILURE(errorCode_)) { return *this; } + Locale *desiredClone = desired.clone(); + Locale *supportedClone = supported.clone(); + if (desiredClone == nullptr || supportedClone == nullptr) { + delete desiredClone; // in case only one could not be allocated + delete supportedClone; + errorCode_ = U_MEMORY_ALLOCATION_ERROR; + return *this; + } + delete maxDistanceDesired_; + delete maxDistanceSupported_; + maxDistanceDesired_ = desiredClone; + maxDistanceSupported_ = supportedClone; + return *this; +} + +#if 0 +/** + * Internal only! + * + * @param thresholdDistance the thresholdDistance to set, with -1 = default + * @return this Builder object + * @internal + * @deprecated This API is ICU internal only. + */ +@Deprecated +LocaleMatcher::Builder &LocaleMatcher::Builder::internalSetThresholdDistance(int32_t thresholdDistance) { + if (U_FAILURE(errorCode_)) { return *this; } + if (thresholdDistance > 100) { + thresholdDistance = 100; + } + thresholdDistance_ = thresholdDistance; + return *this; +} +#endif + +UBool LocaleMatcher::Builder::copyErrorTo(UErrorCode &outErrorCode) const { + if (U_FAILURE(outErrorCode)) { return true; } + if (U_SUCCESS(errorCode_)) { return false; } + outErrorCode = errorCode_; + return true; +} + +LocaleMatcher LocaleMatcher::Builder::build(UErrorCode &errorCode) const { + if (U_SUCCESS(errorCode) && U_FAILURE(errorCode_)) { + errorCode = errorCode_; + } + return LocaleMatcher(*this, errorCode); +} + +namespace { + +LSR getMaximalLsrOrUnd(const XLikelySubtags &likelySubtags, const Locale &locale, + UErrorCode &errorCode) { + if (U_FAILURE(errorCode) || locale.isBogus() || *locale.getName() == 0 /* "und" */) { + return UND_LSR; + } else { + return likelySubtags.makeMaximizedLsrFrom(locale, errorCode); + } +} + +int32_t hashLSR(const UHashTok token) { + const LSR *lsr = static_cast(token.pointer); + return lsr->hashCode; +} + +UBool compareLSRs(const UHashTok t1, const UHashTok t2) { + const LSR *lsr1 = static_cast(t1.pointer); + const LSR *lsr2 = static_cast(t2.pointer); + return *lsr1 == *lsr2; +} + +} // namespace + +int32_t LocaleMatcher::putIfAbsent(const LSR &lsr, int32_t i, int32_t suppLength, + UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return suppLength; } + if (!uhash_containsKey(supportedLsrToIndex, &lsr)) { + uhash_putiAllowZero(supportedLsrToIndex, const_cast(&lsr), i, &errorCode); + if (U_SUCCESS(errorCode)) { + supportedLSRs[suppLength] = &lsr; + supportedIndexes[suppLength++] = i; + } + } + return suppLength; +} + +LocaleMatcher::LocaleMatcher(const Builder &builder, UErrorCode &errorCode) : + likelySubtags(*XLikelySubtags::getSingleton(errorCode)), + localeDistance(*LocaleDistance::getSingleton(errorCode)), + thresholdDistance(builder.thresholdDistance_), + demotionPerDesiredLocale(0), + favorSubtag(builder.favor_), + direction(builder.direction_), + supportedLocales(nullptr), lsrs(nullptr), supportedLocalesLength(0), + supportedLsrToIndex(nullptr), + supportedLSRs(nullptr), supportedIndexes(nullptr), supportedLSRsLength(0), + ownedDefaultLocale(nullptr), defaultLocale(nullptr) { + if (U_FAILURE(errorCode)) { return; } + const Locale *def = builder.defaultLocale_; + LSR builderDefaultLSR; + const LSR *defLSR = nullptr; + if (def != nullptr) { + ownedDefaultLocale = def->clone(); + if (ownedDefaultLocale == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + def = ownedDefaultLocale; + builderDefaultLSR = getMaximalLsrOrUnd(likelySubtags, *def, errorCode); + if (U_FAILURE(errorCode)) { return; } + defLSR = &builderDefaultLSR; + } + supportedLocalesLength = builder.supportedLocales_ != nullptr ? + builder.supportedLocales_->size() : 0; + if (supportedLocalesLength > 0) { + // Store the supported locales in input order, + // so that when different types are used (e.g., language tag strings) + // we can return those by parallel index. + supportedLocales = static_cast( + uprv_malloc(supportedLocalesLength * sizeof(const Locale *))); + // Supported LRSs in input order. + // In C++, we store these permanently to simplify ownership management + // in the hash tables. Duplicate LSRs (if any) are unused overhead. + lsrs = new LSR[supportedLocalesLength]; + if (supportedLocales == nullptr || lsrs == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + // If the constructor fails partway, we need null pointers for destructibility. + uprv_memset(supportedLocales, 0, supportedLocalesLength * sizeof(const Locale *)); + for (int32_t i = 0; i < supportedLocalesLength; ++i) { + const Locale &locale = *static_cast(builder.supportedLocales_->elementAt(i)); + supportedLocales[i] = locale.clone(); + if (supportedLocales[i] == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + const Locale &supportedLocale = *supportedLocales[i]; + LSR &lsr = lsrs[i] = getMaximalLsrOrUnd(likelySubtags, supportedLocale, errorCode); + lsr.setHashCode(); + if (U_FAILURE(errorCode)) { return; } + } + + // We need an unordered map from LSR to first supported locale with that LSR, + // and an ordered list of (LSR, supported index) for + // the supported locales in the following order: + // 1. Default locale, if it is supported. + // 2. Priority locales (aka "paradigm locales") in builder order. + // 3. Remaining locales in builder order. + supportedLsrToIndex = uhash_openSize(hashLSR, compareLSRs, uhash_compareLong, + supportedLocalesLength, &errorCode); + if (U_FAILURE(errorCode)) { return; } + supportedLSRs = static_cast( + uprv_malloc(supportedLocalesLength * sizeof(const LSR *))); + supportedIndexes = static_cast( + uprv_malloc(supportedLocalesLength * sizeof(int32_t))); + if (supportedLSRs == nullptr || supportedIndexes == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + int32_t suppLength = 0; + // Determine insertion order. + // Add locales immediately that are equivalent to the default. + MaybeStackArray order(supportedLocalesLength, errorCode); + if (U_FAILURE(errorCode)) { return; } + int32_t numParadigms = 0; + for (int32_t i = 0; i < supportedLocalesLength; ++i) { + const Locale &locale = *supportedLocales[i]; + const LSR &lsr = lsrs[i]; + if (defLSR == nullptr && builder.withDefault_) { + // Implicit default locale = first supported locale, if not turned off. + U_ASSERT(i == 0); + def = &locale; + defLSR = &lsr; + order[i] = 1; + suppLength = putIfAbsent(lsr, 0, suppLength, errorCode); + } else if (defLSR != nullptr && lsr.isEquivalentTo(*defLSR)) { + order[i] = 1; + suppLength = putIfAbsent(lsr, i, suppLength, errorCode); + } else if (localeDistance.isParadigmLSR(lsr)) { + order[i] = 2; + ++numParadigms; + } else { + order[i] = 3; + } + if (U_FAILURE(errorCode)) { return; } + } + // Add supported paradigm locales. + int32_t paradigmLimit = suppLength + numParadigms; + for (int32_t i = 0; i < supportedLocalesLength && suppLength < paradigmLimit; ++i) { + if (order[i] == 2) { + suppLength = putIfAbsent(lsrs[i], i, suppLength, errorCode); + } + } + // Add remaining supported locales. + for (int32_t i = 0; i < supportedLocalesLength; ++i) { + if (order[i] == 3) { + suppLength = putIfAbsent(lsrs[i], i, suppLength, errorCode); + } + } + supportedLSRsLength = suppLength; + // If supportedLSRsLength < supportedLocalesLength then + // we waste as many array slots as there are duplicate supported LSRs, + // but the amount of wasted space is small as long as there are few duplicates. + } + + defaultLocale = def; + + if (builder.demotion_ == ULOCMATCH_DEMOTION_REGION) { + demotionPerDesiredLocale = localeDistance.getDefaultDemotionPerDesiredLocale(); + } + + if (thresholdDistance >= 0) { + // already copied + } else if (builder.maxDistanceDesired_ != nullptr) { + LSR suppLSR = getMaximalLsrOrUnd(likelySubtags, *builder.maxDistanceSupported_, errorCode); + const LSR *pSuppLSR = &suppLSR; + int32_t indexAndDistance = localeDistance.getBestIndexAndDistance( + getMaximalLsrOrUnd(likelySubtags, *builder.maxDistanceDesired_, errorCode), + &pSuppLSR, 1, + LocaleDistance::shiftDistance(100), favorSubtag, direction); + if (U_SUCCESS(errorCode)) { + // +1 for an exclusive threshold from an inclusive max. + thresholdDistance = LocaleDistance::getDistanceFloor(indexAndDistance) + 1; + } else { + thresholdDistance = 0; + } + } else { + thresholdDistance = localeDistance.getDefaultScriptDistance(); + } +} + +LocaleMatcher::LocaleMatcher(LocaleMatcher &&src) noexcept : + likelySubtags(src.likelySubtags), + localeDistance(src.localeDistance), + thresholdDistance(src.thresholdDistance), + demotionPerDesiredLocale(src.demotionPerDesiredLocale), + favorSubtag(src.favorSubtag), + direction(src.direction), + supportedLocales(src.supportedLocales), lsrs(src.lsrs), + supportedLocalesLength(src.supportedLocalesLength), + supportedLsrToIndex(src.supportedLsrToIndex), + supportedLSRs(src.supportedLSRs), + supportedIndexes(src.supportedIndexes), + supportedLSRsLength(src.supportedLSRsLength), + ownedDefaultLocale(src.ownedDefaultLocale), defaultLocale(src.defaultLocale) { + src.supportedLocales = nullptr; + src.lsrs = nullptr; + src.supportedLocalesLength = 0; + src.supportedLsrToIndex = nullptr; + src.supportedLSRs = nullptr; + src.supportedIndexes = nullptr; + src.supportedLSRsLength = 0; + src.ownedDefaultLocale = nullptr; + src.defaultLocale = nullptr; +} + +LocaleMatcher::~LocaleMatcher() { + for (int32_t i = 0; i < supportedLocalesLength; ++i) { + delete supportedLocales[i]; + } + uprv_free(supportedLocales); + delete[] lsrs; + uhash_close(supportedLsrToIndex); + uprv_free(supportedLSRs); + uprv_free(supportedIndexes); + delete ownedDefaultLocale; +} + +LocaleMatcher &LocaleMatcher::operator=(LocaleMatcher &&src) noexcept { + this->~LocaleMatcher(); + + thresholdDistance = src.thresholdDistance; + demotionPerDesiredLocale = src.demotionPerDesiredLocale; + favorSubtag = src.favorSubtag; + direction = src.direction; + supportedLocales = src.supportedLocales; + lsrs = src.lsrs; + supportedLocalesLength = src.supportedLocalesLength; + supportedLsrToIndex = src.supportedLsrToIndex; + supportedLSRs = src.supportedLSRs; + supportedIndexes = src.supportedIndexes; + supportedLSRsLength = src.supportedLSRsLength; + ownedDefaultLocale = src.ownedDefaultLocale; + defaultLocale = src.defaultLocale; + + src.supportedLocales = nullptr; + src.lsrs = nullptr; + src.supportedLocalesLength = 0; + src.supportedLsrToIndex = nullptr; + src.supportedLSRs = nullptr; + src.supportedIndexes = nullptr; + src.supportedLSRsLength = 0; + src.ownedDefaultLocale = nullptr; + src.defaultLocale = nullptr; + return *this; +} + +class LocaleLsrIterator { +public: + LocaleLsrIterator(const XLikelySubtags &likelySubtags, Locale::Iterator &locales, + ULocMatchLifetime lifetime) : + likelySubtags(likelySubtags), locales(locales), lifetime(lifetime) {} + + ~LocaleLsrIterator() { + if (lifetime == ULOCMATCH_TEMPORARY_LOCALES) { + delete remembered; + } + } + + bool hasNext() const { + return locales.hasNext(); + } + + LSR next(UErrorCode &errorCode) { + current = &locales.next(); + return getMaximalLsrOrUnd(likelySubtags, *current, errorCode); + } + + void rememberCurrent(int32_t desiredIndex, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return; } + bestDesiredIndex = desiredIndex; + if (lifetime == ULOCMATCH_STORED_LOCALES) { + remembered = current; + } else { + // ULOCMATCH_TEMPORARY_LOCALES + delete remembered; + remembered = new Locale(*current); + if (remembered == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + } + } + } + + const Locale *orphanRemembered() { + const Locale *rem = remembered; + remembered = nullptr; + return rem; + } + + int32_t getBestDesiredIndex() const { + return bestDesiredIndex; + } + +private: + const XLikelySubtags &likelySubtags; + Locale::Iterator &locales; + ULocMatchLifetime lifetime; + const Locale *current = nullptr, *remembered = nullptr; + int32_t bestDesiredIndex = -1; +}; + +const Locale *LocaleMatcher::getBestMatch(const Locale &desiredLocale, UErrorCode &errorCode) const { + if (U_FAILURE(errorCode)) { return nullptr; } + int32_t suppIndex = getBestSuppIndex( + getMaximalLsrOrUnd(likelySubtags, desiredLocale, errorCode), + nullptr, errorCode); + return U_SUCCESS(errorCode) && suppIndex >= 0 ? supportedLocales[suppIndex] : defaultLocale; +} + +const Locale *LocaleMatcher::getBestMatch(Locale::Iterator &desiredLocales, + UErrorCode &errorCode) const { + if (U_FAILURE(errorCode)) { return nullptr; } + if (!desiredLocales.hasNext()) { + return defaultLocale; + } + LocaleLsrIterator lsrIter(likelySubtags, desiredLocales, ULOCMATCH_TEMPORARY_LOCALES); + int32_t suppIndex = getBestSuppIndex(lsrIter.next(errorCode), &lsrIter, errorCode); + return U_SUCCESS(errorCode) && suppIndex >= 0 ? supportedLocales[suppIndex] : defaultLocale; +} + +const Locale *LocaleMatcher::getBestMatchForListString( + StringPiece desiredLocaleList, UErrorCode &errorCode) const { + LocalePriorityList list(desiredLocaleList, errorCode); + LocalePriorityList::Iterator iter = list.iterator(); + return getBestMatch(iter, errorCode); +} + +LocaleMatcher::Result LocaleMatcher::getBestMatchResult( + const Locale &desiredLocale, UErrorCode &errorCode) const { + if (U_FAILURE(errorCode)) { + return Result(nullptr, defaultLocale, -1, -1, false); + } + int32_t suppIndex = getBestSuppIndex( + getMaximalLsrOrUnd(likelySubtags, desiredLocale, errorCode), + nullptr, errorCode); + if (U_FAILURE(errorCode) || suppIndex < 0) { + return Result(nullptr, defaultLocale, -1, -1, false); + } else { + return Result(&desiredLocale, supportedLocales[suppIndex], 0, suppIndex, false); + } +} + +LocaleMatcher::Result LocaleMatcher::getBestMatchResult( + Locale::Iterator &desiredLocales, UErrorCode &errorCode) const { + if (U_FAILURE(errorCode) || !desiredLocales.hasNext()) { + return Result(nullptr, defaultLocale, -1, -1, false); + } + LocaleLsrIterator lsrIter(likelySubtags, desiredLocales, ULOCMATCH_TEMPORARY_LOCALES); + int32_t suppIndex = getBestSuppIndex(lsrIter.next(errorCode), &lsrIter, errorCode); + if (U_FAILURE(errorCode) || suppIndex < 0) { + return Result(nullptr, defaultLocale, -1, -1, false); + } else { + return Result(lsrIter.orphanRemembered(), supportedLocales[suppIndex], + lsrIter.getBestDesiredIndex(), suppIndex, true); + } +} + +int32_t LocaleMatcher::getBestSuppIndex(LSR desiredLSR, LocaleLsrIterator *remainingIter, + UErrorCode &errorCode) const { + if (U_FAILURE(errorCode)) { return -1; } + int32_t desiredIndex = 0; + int32_t bestSupportedLsrIndex = -1; + for (int32_t bestShiftedDistance = LocaleDistance::shiftDistance(thresholdDistance);;) { + // Quick check for exact maximized LSR. + if (supportedLsrToIndex != nullptr) { + desiredLSR.setHashCode(); + UBool found = false; + int32_t suppIndex = uhash_getiAndFound(supportedLsrToIndex, &desiredLSR, &found); + if (found) { + if (remainingIter != nullptr) { + remainingIter->rememberCurrent(desiredIndex, errorCode); + } + return suppIndex; + } + } + int32_t bestIndexAndDistance = localeDistance.getBestIndexAndDistance( + desiredLSR, supportedLSRs, supportedLSRsLength, + bestShiftedDistance, favorSubtag, direction); + if (bestIndexAndDistance >= 0) { + bestShiftedDistance = LocaleDistance::getShiftedDistance(bestIndexAndDistance); + if (remainingIter != nullptr) { + remainingIter->rememberCurrent(desiredIndex, errorCode); + if (U_FAILURE(errorCode)) { return -1; } + } + bestSupportedLsrIndex = LocaleDistance::getIndex(bestIndexAndDistance); + } + if ((bestShiftedDistance -= LocaleDistance::shiftDistance(demotionPerDesiredLocale)) <= 0) { + break; + } + if (remainingIter == nullptr || !remainingIter->hasNext()) { + break; + } + desiredLSR = remainingIter->next(errorCode); + if (U_FAILURE(errorCode)) { return -1; } + ++desiredIndex; + } + if (bestSupportedLsrIndex < 0) { + // no good match + return -1; + } + return supportedIndexes[bestSupportedLsrIndex]; +} + +UBool LocaleMatcher::isMatch(const Locale &desired, const Locale &supported, + UErrorCode &errorCode) const { + LSR suppLSR = getMaximalLsrOrUnd(likelySubtags, supported, errorCode); + if (U_FAILURE(errorCode)) { return 0; } + const LSR *pSuppLSR = &suppLSR; + int32_t indexAndDistance = localeDistance.getBestIndexAndDistance( + getMaximalLsrOrUnd(likelySubtags, desired, errorCode), + &pSuppLSR, 1, + LocaleDistance::shiftDistance(thresholdDistance), favorSubtag, direction); + return indexAndDistance >= 0; +} + +double LocaleMatcher::internalMatch(const Locale &desired, const Locale &supported, UErrorCode &errorCode) const { + // Returns the inverse of the distance: That is, 1-distance(desired, supported). + LSR suppLSR = getMaximalLsrOrUnd(likelySubtags, supported, errorCode); + if (U_FAILURE(errorCode)) { return 0; } + const LSR *pSuppLSR = &suppLSR; + int32_t indexAndDistance = localeDistance.getBestIndexAndDistance( + getMaximalLsrOrUnd(likelySubtags, desired, errorCode), + &pSuppLSR, 1, + LocaleDistance::shiftDistance(thresholdDistance), favorSubtag, direction); + double distance = LocaleDistance::getDistanceDouble(indexAndDistance); + return (100.0 - distance) / 100.0; +} + +U_NAMESPACE_END + +// uloc_acceptLanguage() --------------------------------------------------- *** + +U_NAMESPACE_USE + +namespace { + +class LocaleFromTag { +public: + LocaleFromTag() : locale(Locale::getRoot()) {} + const Locale &operator()(const char *tag) { return locale = Locale(tag); } + +private: + // Store the locale in the converter, rather than return a reference to a temporary, + // or a value which could go out of scope with the caller's reference to it. + Locale locale; +}; + +int32_t acceptLanguage(UEnumeration &supportedLocales, Locale::Iterator &desiredLocales, + char *dest, int32_t capacity, UAcceptResult *acceptResult, + UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return 0; } + LocaleMatcher::Builder builder; + const char *locString; + while ((locString = uenum_next(&supportedLocales, nullptr, &errorCode)) != nullptr) { + Locale loc(locString); + if (loc.isBogus()) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + builder.addSupportedLocale(loc); + } + LocaleMatcher matcher = builder.build(errorCode); + LocaleMatcher::Result result = matcher.getBestMatchResult(desiredLocales, errorCode); + if (U_FAILURE(errorCode)) { return 0; } + if (result.getDesiredIndex() >= 0) { + if (acceptResult != nullptr) { + *acceptResult = *result.getDesiredLocale() == *result.getSupportedLocale() ? + ULOC_ACCEPT_VALID : ULOC_ACCEPT_FALLBACK; + } + const char *bestStr = result.getSupportedLocale()->getName(); + int32_t bestLength = (int32_t)uprv_strlen(bestStr); + if (bestLength <= capacity) { + uprv_memcpy(dest, bestStr, bestLength); + } + return u_terminateChars(dest, capacity, bestLength, &errorCode); + } else { + if (acceptResult != nullptr) { + *acceptResult = ULOC_ACCEPT_FAILED; + } + return u_terminateChars(dest, capacity, 0, &errorCode); + } +} + +} // namespace + +U_CAPI int32_t U_EXPORT2 +uloc_acceptLanguage(char *result, int32_t resultAvailable, + UAcceptResult *outResult, + const char **acceptList, int32_t acceptListCount, + UEnumeration *availableLocales, + UErrorCode *status) { + if (U_FAILURE(*status)) { return 0; } + if ((result == nullptr ? resultAvailable != 0 : resultAvailable < 0) || + (acceptList == nullptr ? acceptListCount != 0 : acceptListCount < 0) || + availableLocales == nullptr) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + LocaleFromTag converter; + Locale::ConvertingIterator desiredLocales( + acceptList, acceptList + acceptListCount, converter); + return acceptLanguage(*availableLocales, desiredLocales, + result, resultAvailable, outResult, *status); +} + +U_CAPI int32_t U_EXPORT2 +uloc_acceptLanguageFromHTTP(char *result, int32_t resultAvailable, + UAcceptResult *outResult, + const char *httpAcceptLanguage, + UEnumeration *availableLocales, + UErrorCode *status) { + if (U_FAILURE(*status)) { return 0; } + if ((result == nullptr ? resultAvailable != 0 : resultAvailable < 0) || + httpAcceptLanguage == nullptr || availableLocales == nullptr) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + LocalePriorityList list(httpAcceptLanguage, *status); + LocalePriorityList::Iterator desiredLocales = list.iterator(); + return acceptLanguage(*availableLocales, desiredLocales, + result, resultAvailable, outResult, *status); +} diff --git a/intl/icu/source/common/localeprioritylist.cpp b/intl/icu/source/common/localeprioritylist.cpp new file mode 100644 index 0000000000..e5ba0a3c77 --- /dev/null +++ b/intl/icu/source/common/localeprioritylist.cpp @@ -0,0 +1,240 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// localeprioritylist.cpp +// created: 2019jul11 Markus W. Scherer + +#include "unicode/utypes.h" +#include "unicode/localpointer.h" +#include "unicode/locid.h" +#include "unicode/stringpiece.h" +#include "unicode/uobject.h" +#include "charstr.h" +#include "cmemory.h" +#include "localeprioritylist.h" +#include "uarrsort.h" +#include "uassert.h" +#include "uhash.h" + +U_NAMESPACE_BEGIN + +namespace { + +int32_t hashLocale(const UHashTok token) { + auto *locale = static_cast(token.pointer); + return locale->hashCode(); +} + +UBool compareLocales(const UHashTok t1, const UHashTok t2) { + auto *l1 = static_cast(t1.pointer); + auto *l2 = static_cast(t2.pointer); + return *l1 == *l2; +} + +constexpr int32_t WEIGHT_ONE = 1000; + +struct LocaleAndWeight { + Locale *locale; + int32_t weight; // 0..1000 = 0.0..1.0 + int32_t index; // force stable sort + + int32_t compare(const LocaleAndWeight &other) const { + int32_t diff = other.weight - weight; // descending: other-this + if (diff != 0) { return diff; } + return index - other.index; + } +}; + +int32_t U_CALLCONV +compareLocaleAndWeight(const void * /*context*/, const void *left, const void *right) { + return static_cast(left)-> + compare(*static_cast(right)); +} + +const char *skipSpaces(const char *p, const char *limit) { + while (p < limit && *p == ' ') { ++p; } + return p; +} + +int32_t findTagLength(const char *p, const char *limit) { + // Look for accept-language delimiters. + // Leave other validation up to the Locale constructor. + const char *q; + for (q = p; q < limit; ++q) { + char c = *q; + if (c == ' ' || c == ',' || c == ';') { break; } + } + return static_cast(q - p); +} + +/** + * Parses and returns a qvalue weight in millis. + * Advances p to after the parsed substring. + * Returns a negative value if parsing fails. + */ +int32_t parseWeight(const char *&p, const char *limit) { + p = skipSpaces(p, limit); + char c; + if (p == limit || ((c = *p) != '0' && c != '1')) { return -1; } + int32_t weight = (c - '0') * 1000; + if (++p == limit || *p != '.') { return weight; } + int32_t multiplier = 100; + while (++p != limit && '0' <= (c = *p) && c <= '9') { + c -= '0'; + if (multiplier > 0) { + weight += c * multiplier; + multiplier /= 10; + } else if (multiplier == 0) { + // round up + if (c >= 5) { ++weight; } + multiplier = -1; + } // else ignore further fraction digits + } + return weight <= WEIGHT_ONE ? weight : -1; // bad if > 1.0 +} + +} // namespace + +/** + * Nothing but a wrapper over a MaybeStackArray of LocaleAndWeight. + * + * This wrapper exists (and is not in an anonymous namespace) + * so that we can forward-declare it in the header file and + * don't have to expose the MaybeStackArray specialization and + * the LocaleAndWeight to code (like the test) that #includes localeprioritylist.h. + * Also, otherwise we would have to do a platform-specific + * template export declaration of some kind for the MaybeStackArray specialization + * to be properly exported from the common DLL. + */ +struct LocaleAndWeightArray : public UMemory { + MaybeStackArray array; +}; + +LocalePriorityList::LocalePriorityList(StringPiece s, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return; } + list = new LocaleAndWeightArray(); + if (list == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + const char *p = s.data(); + const char *limit = p + s.length(); + while ((p = skipSpaces(p, limit)) != limit) { + if (*p == ',') { // empty range field + ++p; + continue; + } + int32_t tagLength = findTagLength(p, limit); + if (tagLength == 0) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + CharString tag(p, tagLength, errorCode); + if (U_FAILURE(errorCode)) { return; } + Locale locale = Locale(tag.data()); + if (locale.isBogus()) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + int32_t weight = WEIGHT_ONE; + if ((p = skipSpaces(p + tagLength, limit)) != limit && *p == ';') { + if ((p = skipSpaces(p + 1, limit)) == limit || *p != 'q' || + (p = skipSpaces(p + 1, limit)) == limit || *p != '=' || + (++p, (weight = parseWeight(p, limit)) < 0)) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + p = skipSpaces(p, limit); + } + if (p != limit && *p != ',') { // trailing junk + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + add(locale, weight, errorCode); + if (p == limit) { break; } + ++p; + } + sort(errorCode); +} + +LocalePriorityList::~LocalePriorityList() { + if (list != nullptr) { + for (int32_t i = 0; i < listLength; ++i) { + delete list->array[i].locale; + } + delete list; + } + uhash_close(map); +} + +const Locale *LocalePriorityList::localeAt(int32_t i) const { + return list->array[i].locale; +} + +Locale *LocalePriorityList::orphanLocaleAt(int32_t i) { + if (list == nullptr) { return nullptr; } + LocaleAndWeight &lw = list->array[i]; + Locale *l = lw.locale; + lw.locale = nullptr; + return l; +} + +bool LocalePriorityList::add(const Locale &locale, int32_t weight, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return false; } + if (map == nullptr) { + if (weight <= 0) { return true; } // do not add q=0 + map = uhash_open(hashLocale, compareLocales, uhash_compareLong, &errorCode); + if (U_FAILURE(errorCode)) { return false; } + } + LocalPointer clone; + UBool found = false; + int32_t index = uhash_getiAndFound(map, &locale, &found); + if (found) { + // Duplicate: Remove the old item and append it anew. + LocaleAndWeight &lw = list->array[index]; + clone.adoptInstead(lw.locale); + lw.locale = nullptr; + lw.weight = 0; + ++numRemoved; + } + if (weight <= 0) { // do not add q=0 + if (found) { + // Not strictly necessary but cleaner. + uhash_removei(map, &locale); + } + return true; + } + if (clone.isNull()) { + clone.adoptInstead(locale.clone()); + if (clone.isNull() || (clone->isBogus() && !locale.isBogus())) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return false; + } + } + if (listLength == list->array.getCapacity()) { + int32_t newCapacity = listLength < 50 ? 100 : 4 * listLength; + if (list->array.resize(newCapacity, listLength) == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return false; + } + } + uhash_putiAllowZero(map, clone.getAlias(), listLength, &errorCode); + if (U_FAILURE(errorCode)) { return false; } + LocaleAndWeight &lw = list->array[listLength]; + lw.locale = clone.orphan(); + lw.weight = weight; + lw.index = listLength++; + if (weight < WEIGHT_ONE) { hasWeights = true; } + U_ASSERT(uhash_count(map) == getLength()); + return true; +} + +void LocalePriorityList::sort(UErrorCode &errorCode) { + // Sort by descending weights if there is a mix of weights. + // The comparator forces a stable sort via the item index. + if (U_FAILURE(errorCode) || getLength() <= 1 || !hasWeights) { return; } + uprv_sortArray(list->array.getAlias(), listLength, sizeof(LocaleAndWeight), + compareLocaleAndWeight, nullptr, false, &errorCode); +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/localeprioritylist.h b/intl/icu/source/common/localeprioritylist.h new file mode 100644 index 0000000000..41e9d3ea08 --- /dev/null +++ b/intl/icu/source/common/localeprioritylist.h @@ -0,0 +1,115 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// localeprioritylist.h +// created: 2019jul11 Markus W. Scherer + +#ifndef __LOCALEPRIORITYLIST_H__ +#define __LOCALEPRIORITYLIST_H__ + +#include "unicode/utypes.h" +#include "unicode/locid.h" +#include "unicode/stringpiece.h" +#include "unicode/uobject.h" + +struct UHashtable; + +U_NAMESPACE_BEGIN + +struct LocaleAndWeightArray; + +/** + * Parses a list of locales from an accept-language string. + * We are a bit more lenient than the spec: + * We accept extra whitespace in more places, empty range fields, + * and any number of qvalue fraction digits. + * + * https://tools.ietf.org/html/rfc2616#section-14.4 + * 14.4 Accept-Language + * + * Accept-Language = "Accept-Language" ":" + * 1#( language-range [ ";" "q" "=" qvalue ] ) + * language-range = ( ( 1*8ALPHA *( "-" 1*8ALPHA ) ) | "*" ) + * + * Each language-range MAY be given an associated quality value which + * represents an estimate of the user's preference for the languages + * specified by that range. The quality value defaults to "q=1". For + * example, + * + * Accept-Language: da, en-gb;q=0.8, en;q=0.7 + * + * https://tools.ietf.org/html/rfc2616#section-3.9 + * 3.9 Quality Values + * + * HTTP content negotiation (section 12) uses short "floating point" + * numbers to indicate the relative importance ("weight") of various + * negotiable parameters. A weight is normalized to a real number in + * the range 0 through 1, where 0 is the minimum and 1 the maximum + * value. If a parameter has a quality value of 0, then content with + * this parameter is `not acceptable' for the client. HTTP/1.1 + * applications MUST NOT generate more than three digits after the + * decimal point. User configuration of these values SHOULD also be + * limited in this fashion. + * + * qvalue = ( "0" [ "." 0*3DIGIT ] ) + * | ( "1" [ "." 0*3("0") ] ) + */ +class U_COMMON_API LocalePriorityList : public UMemory { +public: + class Iterator : public Locale::Iterator { + public: + UBool hasNext() const override { return count < length; } + + const Locale &next() override { + for(;;) { + const Locale *locale = list.localeAt(index++); + if (locale != nullptr) { + ++count; + return *locale; + } + } + } + + private: + friend class LocalePriorityList; + + Iterator(const LocalePriorityList &list) : list(list), length(list.getLength()) {} + + const LocalePriorityList &list; + int32_t index = 0; + int32_t count = 0; + const int32_t length; + }; + + LocalePriorityList(StringPiece s, UErrorCode &errorCode); + + ~LocalePriorityList(); + + int32_t getLength() const { return listLength - numRemoved; } + + int32_t getLengthIncludingRemoved() const { return listLength; } + + Iterator iterator() const { return Iterator(*this); } + + const Locale *localeAt(int32_t i) const; + + Locale *orphanLocaleAt(int32_t i); + +private: + LocalePriorityList(const LocalePriorityList &) = delete; + LocalePriorityList &operator=(const LocalePriorityList &) = delete; + + bool add(const Locale &locale, int32_t weight, UErrorCode &errorCode); + + void sort(UErrorCode &errorCode); + + LocaleAndWeightArray *list = nullptr; + int32_t listLength = 0; + int32_t numRemoved = 0; + bool hasWeights = false; // other than 1.0 + UHashtable *map = nullptr; +}; + +U_NAMESPACE_END + +#endif // __LOCALEPRIORITYLIST_H__ diff --git a/intl/icu/source/common/localsvc.h b/intl/icu/source/common/localsvc.h new file mode 100644 index 0000000000..3364019513 --- /dev/null +++ b/intl/icu/source/common/localsvc.h @@ -0,0 +1,27 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +*************************************************************************** +* Copyright (C) 2006 International Business Machines Corporation * +* and others. All rights reserved. * +*************************************************************************** +*/ + +#ifndef LOCALSVC_H +#define LOCALSVC_H + +#include "unicode/utypes.h" + +#if defined(U_LOCAL_SERVICE_HOOK) && U_LOCAL_SERVICE_HOOK +/** + * Prototype for user-supplied service hook. This function is expected to return + * a type of factory object specific to the requested service. + * + * @param what service-specific string identifying the specific user hook + * @param status error status + * @return a service-specific hook, or NULL on failure. + */ +U_CAPI void* uprv_svc_hook(const char *what, UErrorCode *status); +#endif + +#endif diff --git a/intl/icu/source/common/locavailable.cpp b/intl/icu/source/common/locavailable.cpp new file mode 100644 index 0000000000..0ea2093988 --- /dev/null +++ b/intl/icu/source/common/locavailable.cpp @@ -0,0 +1,270 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 1997-2013, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: locavailable.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010feb25 +* created by: Markus W. Scherer +* +* Code for available locales, separated out from other .cpp files +* that then do not depend on resource bundle code and res_index bundles. +*/ + +#include "unicode/errorcode.h" +#include "unicode/utypes.h" +#include "unicode/locid.h" +#include "unicode/uloc.h" +#include "unicode/ures.h" +#include "cmemory.h" +#include "cstring.h" +#include "ucln_cmn.h" +#include "uassert.h" +#include "umutex.h" +#include "uresimp.h" + +// C++ API ----------------------------------------------------------------- *** + +U_NAMESPACE_BEGIN + +static icu::Locale* availableLocaleList = nullptr; +static int32_t availableLocaleListCount; +static icu::UInitOnce gInitOnceLocale {}; + +U_NAMESPACE_END + +U_CDECL_BEGIN + +static UBool U_CALLCONV locale_available_cleanup() +{ + U_NAMESPACE_USE + + if (availableLocaleList) { + delete []availableLocaleList; + availableLocaleList = nullptr; + } + availableLocaleListCount = 0; + gInitOnceLocale.reset(); + + return true; +} + +U_CDECL_END + +U_NAMESPACE_BEGIN + +void U_CALLCONV locale_available_init() { + // This function is a friend of class Locale. + // This function is only invoked via umtx_initOnce(). + + // for now, there is a hardcoded list, so just walk through that list and set it up. + // Note: this function is a friend of class Locale. + availableLocaleListCount = uloc_countAvailable(); + if(availableLocaleListCount) { + availableLocaleList = new Locale[availableLocaleListCount]; + } + if (availableLocaleList == nullptr) { + availableLocaleListCount= 0; + } + for (int32_t locCount=availableLocaleListCount-1; locCount>=0; --locCount) { + availableLocaleList[locCount].setFromPOSIXID(uloc_getAvailable(locCount)); + } + ucln_common_registerCleanup(UCLN_COMMON_LOCALE_AVAILABLE, locale_available_cleanup); +} + +const Locale* U_EXPORT2 +Locale::getAvailableLocales(int32_t& count) +{ + umtx_initOnce(gInitOnceLocale, &locale_available_init); + count = availableLocaleListCount; + return availableLocaleList; +} + + +U_NAMESPACE_END + +// C API ------------------------------------------------------------------- *** + +U_NAMESPACE_USE + +/* ### Constants **************************************************/ + +namespace { + +// Enough capacity for the two lists in the res_index.res file +const char** gAvailableLocaleNames[2] = {}; +int32_t gAvailableLocaleCounts[2] = {}; +icu::UInitOnce ginstalledLocalesInitOnce {}; + +class AvailableLocalesSink : public ResourceSink { + public: + void put(const char *key, ResourceValue &value, UBool /*noFallback*/, UErrorCode &status) override { + ResourceTable resIndexTable = value.getTable(status); + if (U_FAILURE(status)) { + return; + } + for (int32_t i = 0; resIndexTable.getKeyAndValue(i, key, value); ++i) { + ULocAvailableType type; + if (uprv_strcmp(key, "InstalledLocales") == 0) { + type = ULOC_AVAILABLE_DEFAULT; + } else if (uprv_strcmp(key, "AliasLocales") == 0) { + type = ULOC_AVAILABLE_ONLY_LEGACY_ALIASES; + } else { + // CLDRVersion, etc. + continue; + } + ResourceTable availableLocalesTable = value.getTable(status); + if (U_FAILURE(status)) { + return; + } + gAvailableLocaleCounts[type] = availableLocalesTable.getSize(); + gAvailableLocaleNames[type] = static_cast( + uprv_malloc(gAvailableLocaleCounts[type] * sizeof(const char*))); + if (gAvailableLocaleNames[type] == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + for (int32_t j = 0; availableLocalesTable.getKeyAndValue(j, key, value); ++j) { + gAvailableLocaleNames[type][j] = key; + } + } + } +}; + +class AvailableLocalesStringEnumeration : public StringEnumeration { + public: + AvailableLocalesStringEnumeration(ULocAvailableType type) : fType(type) { + } + + const char* next(int32_t *resultLength, UErrorCode&) override { + ULocAvailableType actualType = fType; + int32_t actualIndex = fIndex++; + + // If the "combined" list was requested, resolve that now + if (fType == ULOC_AVAILABLE_WITH_LEGACY_ALIASES) { + int32_t defaultLocalesCount = gAvailableLocaleCounts[ULOC_AVAILABLE_DEFAULT]; + if (actualIndex < defaultLocalesCount) { + actualType = ULOC_AVAILABLE_DEFAULT; + } else { + actualIndex -= defaultLocalesCount; + actualType = ULOC_AVAILABLE_ONLY_LEGACY_ALIASES; + } + } + + // Return the requested string + int32_t count = gAvailableLocaleCounts[actualType]; + const char* result; + if (actualIndex < count) { + result = gAvailableLocaleNames[actualType][actualIndex]; + if (resultLength != nullptr) { + *resultLength = static_cast(uprv_strlen(result)); + } + } else { + result = nullptr; + if (resultLength != nullptr) { + *resultLength = 0; + } + } + return result; + } + + void reset(UErrorCode&) override { + fIndex = 0; + } + + int32_t count(UErrorCode&) const override { + if (fType == ULOC_AVAILABLE_WITH_LEGACY_ALIASES) { + return gAvailableLocaleCounts[ULOC_AVAILABLE_DEFAULT] + + gAvailableLocaleCounts[ULOC_AVAILABLE_ONLY_LEGACY_ALIASES]; + } else { + return gAvailableLocaleCounts[fType]; + } + } + + private: + ULocAvailableType fType; + int32_t fIndex = 0; +}; + +/* ### Get available **************************************************/ + +static UBool U_CALLCONV uloc_cleanup() { + for (int32_t i = 0; i < UPRV_LENGTHOF(gAvailableLocaleNames); i++) { + uprv_free(gAvailableLocaleNames[i]); + gAvailableLocaleNames[i] = nullptr; + gAvailableLocaleCounts[i] = 0; + } + ginstalledLocalesInitOnce.reset(); + return true; +} + +// Load Installed Locales. This function will be called exactly once +// via the initOnce mechanism. + +static void U_CALLCONV loadInstalledLocales(UErrorCode& status) { + ucln_common_registerCleanup(UCLN_COMMON_ULOC, uloc_cleanup); + + icu::LocalUResourceBundlePointer rb(ures_openDirect(nullptr, "res_index", &status)); + AvailableLocalesSink sink; + ures_getAllItemsWithFallback(rb.getAlias(), "", sink, status); +} + +void _load_installedLocales(UErrorCode& status) { + umtx_initOnce(ginstalledLocalesInitOnce, &loadInstalledLocales, status); +} + +} // namespace + +U_CAPI const char* U_EXPORT2 +uloc_getAvailable(int32_t offset) { + icu::ErrorCode status; + _load_installedLocales(status); + if (status.isFailure()) { + return nullptr; + } + if (offset > gAvailableLocaleCounts[0]) { + // *status = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + return gAvailableLocaleNames[0][offset]; +} + +U_CAPI int32_t U_EXPORT2 +uloc_countAvailable() { + icu::ErrorCode status; + _load_installedLocales(status); + if (status.isFailure()) { + return 0; + } + return gAvailableLocaleCounts[0]; +} + +U_CAPI UEnumeration* U_EXPORT2 +uloc_openAvailableByType(ULocAvailableType type, UErrorCode* status) { + if (U_FAILURE(*status)) { + return nullptr; + } + if (type < 0 || type >= ULOC_AVAILABLE_COUNT) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + _load_installedLocales(*status); + if (U_FAILURE(*status)) { + return nullptr; + } + LocalPointer result( + new AvailableLocalesStringEnumeration(type), *status); + if (U_FAILURE(*status)) { + return nullptr; + } + return uenum_openFromStringEnumeration(result.orphan(), status); +} + diff --git a/intl/icu/source/common/locbased.cpp b/intl/icu/source/common/locbased.cpp new file mode 100644 index 0000000000..adcf2f843c --- /dev/null +++ b/intl/icu/source/common/locbased.cpp @@ -0,0 +1,55 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2004-2014, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* Author: Alan Liu +* Created: January 16 2004 +* Since: ICU 2.8 +********************************************************************** +*/ +#include "locbased.h" +#include "cstring.h" + +U_NAMESPACE_BEGIN + +Locale LocaleBased::getLocale(ULocDataLocaleType type, UErrorCode& status) const { + const char* id = getLocaleID(type, status); + return Locale((id != 0) ? id : ""); +} + +const char* LocaleBased::getLocaleID(ULocDataLocaleType type, UErrorCode& status) const { + if (U_FAILURE(status)) { + return nullptr; + } + + switch(type) { + case ULOC_VALID_LOCALE: + return valid; + case ULOC_ACTUAL_LOCALE: + return actual; + default: + status = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } +} + +void LocaleBased::setLocaleIDs(const char* validID, const char* actualID) { + if (validID != 0) { + uprv_strncpy(valid, validID, ULOC_FULLNAME_CAPACITY); + valid[ULOC_FULLNAME_CAPACITY-1] = 0; // always terminate + } + if (actualID != 0) { + uprv_strncpy(actual, actualID, ULOC_FULLNAME_CAPACITY); + actual[ULOC_FULLNAME_CAPACITY-1] = 0; // always terminate + } +} + +void LocaleBased::setLocaleIDs(const Locale& validID, const Locale& actualID) { + uprv_strcpy(valid, validID.getName()); + uprv_strcpy(actual, actualID.getName()); +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/locbased.h b/intl/icu/source/common/locbased.h new file mode 100644 index 0000000000..45738863b5 --- /dev/null +++ b/intl/icu/source/common/locbased.h @@ -0,0 +1,107 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2004-2014, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* Author: Alan Liu +* Created: January 16 2004 +* Since: ICU 2.8 +********************************************************************** +*/ +#ifndef LOCBASED_H +#define LOCBASED_H + +#include "unicode/locid.h" +#include "unicode/uobject.h" + +/** + * Macro to declare a locale LocaleBased wrapper object for the given + * object, which must have two members named `validLocale' and + * `actualLocale' of size ULOC_FULLNAME_CAPACITY + */ +#define U_LOCALE_BASED(varname, objname) \ + LocaleBased varname((objname).validLocale, (objname).actualLocale) + +U_NAMESPACE_BEGIN + +/** + * A utility class that unifies the implementation of getLocale() by + * various ICU services. This class is likely to be removed in the + * ICU 3.0 time frame in favor of an integrated approach with the + * services framework. + * @since ICU 2.8 + */ +class U_COMMON_API LocaleBased : public UMemory { + + public: + + /** + * Construct a LocaleBased wrapper around the two pointers. These + * will be aliased for the lifetime of this object. + */ + inline LocaleBased(char* validAlias, char* actualAlias); + + /** + * Construct a LocaleBased wrapper around the two const pointers. + * These will be aliased for the lifetime of this object. + */ + inline LocaleBased(const char* validAlias, const char* actualAlias); + + /** + * Return locale meta-data for the service object wrapped by this + * object. Either the valid or the actual locale may be + * retrieved. + * @param type either ULOC_VALID_LOCALE or ULOC_ACTUAL_LOCALE + * @param status input-output error code + * @return the indicated locale + */ + Locale getLocale(ULocDataLocaleType type, UErrorCode& status) const; + + /** + * Return the locale ID for the service object wrapped by this + * object. Either the valid or the actual locale may be + * retrieved. + * @param type either ULOC_VALID_LOCALE or ULOC_ACTUAL_LOCALE + * @param status input-output error code + * @return the indicated locale ID + */ + const char* getLocaleID(ULocDataLocaleType type, UErrorCode& status) const; + + /** + * Set the locale meta-data for the service object wrapped by this + * object. If either parameter is zero, it is ignored. + * @param valid the ID of the valid locale + * @param actual the ID of the actual locale + */ + void setLocaleIDs(const char* valid, const char* actual); + + /** + * Set the locale meta-data for the service object wrapped by this + * object. + * @param valid the ID of the valid locale + * @param actual the ID of the actual locale + */ + void setLocaleIDs(const Locale& valid, const Locale& actual); + + private: + + char* valid; + + char* actual; +}; + +inline LocaleBased::LocaleBased(char* validAlias, char* actualAlias) : + valid(validAlias), actual(actualAlias) { +} + +inline LocaleBased::LocaleBased(const char* validAlias, + const char* actualAlias) : + // ugh: cast away const + valid((char*)validAlias), actual((char*)actualAlias) { +} + +U_NAMESPACE_END + +#endif diff --git a/intl/icu/source/common/locdispnames.cpp b/intl/icu/source/common/locdispnames.cpp new file mode 100644 index 0000000000..3ec1b81c81 --- /dev/null +++ b/intl/icu/source/common/locdispnames.cpp @@ -0,0 +1,902 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 1997-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: locdispnames.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010feb25 +* created by: Markus W. Scherer +* +* Code for locale display names, separated out from other .cpp files +* that then do not depend on resource bundle code and display name data. +*/ + +#include "unicode/utypes.h" +#include "unicode/brkiter.h" +#include "unicode/locid.h" +#include "unicode/uenum.h" +#include "unicode/uloc.h" +#include "unicode/ures.h" +#include "unicode/ustring.h" +#include "bytesinkutil.h" +#include "charstr.h" +#include "cmemory.h" +#include "cstring.h" +#include "putilimp.h" +#include "ulocimp.h" +#include "uresimp.h" +#include "ureslocs.h" +#include "ustr_imp.h" + +// C++ API ----------------------------------------------------------------- *** + +U_NAMESPACE_BEGIN + +UnicodeString& +Locale::getDisplayLanguage(UnicodeString& dispLang) const +{ + return this->getDisplayLanguage(getDefault(), dispLang); +} + +/*We cannot make any assumptions on the size of the output display strings +* Yet, since we are calling through to a C API, we need to set limits on +* buffer size. For all the following getDisplay functions we first attempt +* to fill up a stack allocated buffer. If it is to small we heap allocated +* the exact buffer we need copy it to the UnicodeString and delete it*/ + +UnicodeString& +Locale::getDisplayLanguage(const Locale &displayLocale, + UnicodeString &result) const { + char16_t *buffer; + UErrorCode errorCode=U_ZERO_ERROR; + int32_t length; + + buffer=result.getBuffer(ULOC_FULLNAME_CAPACITY); + if(buffer==0) { + result.truncate(0); + return result; + } + + length=uloc_getDisplayLanguage(fullName, displayLocale.fullName, + buffer, result.getCapacity(), + &errorCode); + result.releaseBuffer(U_SUCCESS(errorCode) ? length : 0); + + if(errorCode==U_BUFFER_OVERFLOW_ERROR) { + buffer=result.getBuffer(length); + if(buffer==0) { + result.truncate(0); + return result; + } + errorCode=U_ZERO_ERROR; + length=uloc_getDisplayLanguage(fullName, displayLocale.fullName, + buffer, result.getCapacity(), + &errorCode); + result.releaseBuffer(U_SUCCESS(errorCode) ? length : 0); + } + + return result; +} + +UnicodeString& +Locale::getDisplayScript(UnicodeString& dispScript) const +{ + return this->getDisplayScript(getDefault(), dispScript); +} + +UnicodeString& +Locale::getDisplayScript(const Locale &displayLocale, + UnicodeString &result) const { + char16_t *buffer; + UErrorCode errorCode=U_ZERO_ERROR; + int32_t length; + + buffer=result.getBuffer(ULOC_FULLNAME_CAPACITY); + if(buffer==0) { + result.truncate(0); + return result; + } + + length=uloc_getDisplayScript(fullName, displayLocale.fullName, + buffer, result.getCapacity(), + &errorCode); + result.releaseBuffer(U_SUCCESS(errorCode) ? length : 0); + + if(errorCode==U_BUFFER_OVERFLOW_ERROR) { + buffer=result.getBuffer(length); + if(buffer==0) { + result.truncate(0); + return result; + } + errorCode=U_ZERO_ERROR; + length=uloc_getDisplayScript(fullName, displayLocale.fullName, + buffer, result.getCapacity(), + &errorCode); + result.releaseBuffer(U_SUCCESS(errorCode) ? length : 0); + } + + return result; +} + +UnicodeString& +Locale::getDisplayCountry(UnicodeString& dispCntry) const +{ + return this->getDisplayCountry(getDefault(), dispCntry); +} + +UnicodeString& +Locale::getDisplayCountry(const Locale &displayLocale, + UnicodeString &result) const { + char16_t *buffer; + UErrorCode errorCode=U_ZERO_ERROR; + int32_t length; + + buffer=result.getBuffer(ULOC_FULLNAME_CAPACITY); + if(buffer==0) { + result.truncate(0); + return result; + } + + length=uloc_getDisplayCountry(fullName, displayLocale.fullName, + buffer, result.getCapacity(), + &errorCode); + result.releaseBuffer(U_SUCCESS(errorCode) ? length : 0); + + if(errorCode==U_BUFFER_OVERFLOW_ERROR) { + buffer=result.getBuffer(length); + if(buffer==0) { + result.truncate(0); + return result; + } + errorCode=U_ZERO_ERROR; + length=uloc_getDisplayCountry(fullName, displayLocale.fullName, + buffer, result.getCapacity(), + &errorCode); + result.releaseBuffer(U_SUCCESS(errorCode) ? length : 0); + } + + return result; +} + +UnicodeString& +Locale::getDisplayVariant(UnicodeString& dispVar) const +{ + return this->getDisplayVariant(getDefault(), dispVar); +} + +UnicodeString& +Locale::getDisplayVariant(const Locale &displayLocale, + UnicodeString &result) const { + char16_t *buffer; + UErrorCode errorCode=U_ZERO_ERROR; + int32_t length; + + buffer=result.getBuffer(ULOC_FULLNAME_CAPACITY); + if(buffer==0) { + result.truncate(0); + return result; + } + + length=uloc_getDisplayVariant(fullName, displayLocale.fullName, + buffer, result.getCapacity(), + &errorCode); + result.releaseBuffer(U_SUCCESS(errorCode) ? length : 0); + + if(errorCode==U_BUFFER_OVERFLOW_ERROR) { + buffer=result.getBuffer(length); + if(buffer==0) { + result.truncate(0); + return result; + } + errorCode=U_ZERO_ERROR; + length=uloc_getDisplayVariant(fullName, displayLocale.fullName, + buffer, result.getCapacity(), + &errorCode); + result.releaseBuffer(U_SUCCESS(errorCode) ? length : 0); + } + + return result; +} + +UnicodeString& +Locale::getDisplayName( UnicodeString& name ) const +{ + return this->getDisplayName(getDefault(), name); +} + +UnicodeString& +Locale::getDisplayName(const Locale &displayLocale, + UnicodeString &result) const { + char16_t *buffer; + UErrorCode errorCode=U_ZERO_ERROR; + int32_t length; + + buffer=result.getBuffer(ULOC_FULLNAME_CAPACITY); + if(buffer==0) { + result.truncate(0); + return result; + } + + length=uloc_getDisplayName(fullName, displayLocale.fullName, + buffer, result.getCapacity(), + &errorCode); + result.releaseBuffer(U_SUCCESS(errorCode) ? length : 0); + + if(errorCode==U_BUFFER_OVERFLOW_ERROR) { + buffer=result.getBuffer(length); + if(buffer==0) { + result.truncate(0); + return result; + } + errorCode=U_ZERO_ERROR; + length=uloc_getDisplayName(fullName, displayLocale.fullName, + buffer, result.getCapacity(), + &errorCode); + result.releaseBuffer(U_SUCCESS(errorCode) ? length : 0); + } + + return result; +} + +#if ! UCONFIG_NO_BREAK_ITERATION + +// ------------------------------------- +// Gets the objectLocale display name in the default locale language. +UnicodeString& U_EXPORT2 +BreakIterator::getDisplayName(const Locale& objectLocale, + UnicodeString& name) +{ + return objectLocale.getDisplayName(name); +} + +// ------------------------------------- +// Gets the objectLocale display name in the displayLocale language. +UnicodeString& U_EXPORT2 +BreakIterator::getDisplayName(const Locale& objectLocale, + const Locale& displayLocale, + UnicodeString& name) +{ + return objectLocale.getDisplayName(displayLocale, name); +} + +#endif + + +U_NAMESPACE_END + +// C API ------------------------------------------------------------------- *** + +U_NAMESPACE_USE + +/* ### Constants **************************************************/ + +/* These strings describe the resources we attempt to load from + the locale ResourceBundle data file.*/ +static const char _kLanguages[] = "Languages"; +static const char _kScripts[] = "Scripts"; +static const char _kScriptsStandAlone[] = "Scripts%stand-alone"; +static const char _kCountries[] = "Countries"; +static const char _kVariants[] = "Variants"; +static const char _kKeys[] = "Keys"; +static const char _kTypes[] = "Types"; +//static const char _kRootName[] = "root"; +static const char _kCurrency[] = "currency"; +static const char _kCurrencies[] = "Currencies"; +static const char _kLocaleDisplayPattern[] = "localeDisplayPattern"; +static const char _kPattern[] = "pattern"; +static const char _kSeparator[] = "separator"; + +/* ### Display name **************************************************/ + +static int32_t +_getStringOrCopyKey(const char *path, const char *locale, + const char *tableKey, + const char* subTableKey, + const char *itemKey, + const char *substitute, + char16_t *dest, int32_t destCapacity, + UErrorCode *pErrorCode) { + const char16_t *s = nullptr; + int32_t length = 0; + + if(itemKey==nullptr) { + /* top-level item: normal resource bundle access */ + icu::LocalUResourceBundlePointer rb(ures_open(path, locale, pErrorCode)); + + if(U_SUCCESS(*pErrorCode)) { + s=ures_getStringByKey(rb.getAlias(), tableKey, &length, pErrorCode); + /* see comment about closing rb near "return item;" in _res_getTableStringWithFallback() */ + } + } else { + bool isLanguageCode = (uprv_strncmp(tableKey, _kLanguages, 9) == 0); + /* Language code should not be a number. If it is, set the error code. */ + if (isLanguageCode && uprv_strtol(itemKey, nullptr, 10)) { + *pErrorCode = U_MISSING_RESOURCE_ERROR; + } else { + /* second-level item, use special fallback */ + s=uloc_getTableStringWithFallback(path, locale, + tableKey, + subTableKey, + itemKey, + &length, + pErrorCode); + if (U_FAILURE(*pErrorCode) && isLanguageCode && itemKey != nullptr) { + // convert itemKey locale code to canonical form and try again, ICU-20870 + *pErrorCode = U_ZERO_ERROR; + Locale canonKey = Locale::createCanonical(itemKey); + s=uloc_getTableStringWithFallback(path, locale, + tableKey, + subTableKey, + canonKey.getName(), + &length, + pErrorCode); + } + } + } + + if(U_SUCCESS(*pErrorCode)) { + int32_t copyLength=uprv_min(length, destCapacity); + if(copyLength>0 && s != nullptr) { + u_memcpy(dest, s, copyLength); + } + } else { + /* no string from a resource bundle: convert the substitute */ + length=(int32_t)uprv_strlen(substitute); + u_charsToUChars(substitute, dest, uprv_min(length, destCapacity)); + *pErrorCode=U_USING_DEFAULT_WARNING; + } + + return u_terminateUChars(dest, destCapacity, length, pErrorCode); +} + +typedef int32_t U_CALLCONV UDisplayNameGetter(const char *, char *, int32_t, UErrorCode *); + +static int32_t +_getDisplayNameForComponent(const char *locale, + const char *displayLocale, + char16_t *dest, int32_t destCapacity, + UDisplayNameGetter *getter, + const char *tag, + UErrorCode *pErrorCode) { + char localeBuffer[ULOC_FULLNAME_CAPACITY*4]; + int32_t length; + UErrorCode localStatus; + const char* root = nullptr; + + /* argument checking */ + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + if(destCapacity<0 || (destCapacity>0 && dest==nullptr)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + localStatus = U_ZERO_ERROR; + length=(*getter)(locale, localeBuffer, sizeof(localeBuffer), &localStatus); + if(U_FAILURE(localStatus) || localStatus==U_STRING_NOT_TERMINATED_WARNING) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + if(length==0) { + // For the display name, we treat this as unknown language (ICU-20273). + if (getter == uloc_getLanguage) { + uprv_strcpy(localeBuffer, "und"); + } else { + return u_terminateUChars(dest, destCapacity, 0, pErrorCode); + } + } + + root = tag == _kCountries ? U_ICUDATA_REGION : U_ICUDATA_LANG; + + return _getStringOrCopyKey(root, displayLocale, + tag, nullptr, localeBuffer, + localeBuffer, + dest, destCapacity, + pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayLanguage(const char *locale, + const char *displayLocale, + char16_t *dest, int32_t destCapacity, + UErrorCode *pErrorCode) { + return _getDisplayNameForComponent(locale, displayLocale, dest, destCapacity, + uloc_getLanguage, _kLanguages, pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayScript(const char* locale, + const char* displayLocale, + char16_t *dest, int32_t destCapacity, + UErrorCode *pErrorCode) +{ + UErrorCode err = U_ZERO_ERROR; + int32_t res = _getDisplayNameForComponent(locale, displayLocale, dest, destCapacity, + uloc_getScript, _kScriptsStandAlone, &err); + + if (destCapacity == 0 && err == U_BUFFER_OVERFLOW_ERROR) { + // For preflight, return the max of the value and the fallback. + int32_t fallback_res = _getDisplayNameForComponent(locale, displayLocale, dest, destCapacity, + uloc_getScript, _kScripts, pErrorCode); + return (fallback_res > res) ? fallback_res : res; + } + if ( err == U_USING_DEFAULT_WARNING ) { + return _getDisplayNameForComponent(locale, displayLocale, dest, destCapacity, + uloc_getScript, _kScripts, pErrorCode); + } else { + *pErrorCode = err; + return res; + } +} + +static int32_t +uloc_getDisplayScriptInContext(const char* locale, + const char* displayLocale, + char16_t *dest, int32_t destCapacity, + UErrorCode *pErrorCode) +{ + return _getDisplayNameForComponent(locale, displayLocale, dest, destCapacity, + uloc_getScript, _kScripts, pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayCountry(const char *locale, + const char *displayLocale, + char16_t *dest, int32_t destCapacity, + UErrorCode *pErrorCode) { + return _getDisplayNameForComponent(locale, displayLocale, dest, destCapacity, + uloc_getCountry, _kCountries, pErrorCode); +} + +/* + * TODO separate variant1_variant2_variant3... + * by getting each tag's display string and concatenating them with ", " + * in between - similar to uloc_getDisplayName() + */ +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayVariant(const char *locale, + const char *displayLocale, + char16_t *dest, int32_t destCapacity, + UErrorCode *pErrorCode) { + return _getDisplayNameForComponent(locale, displayLocale, dest, destCapacity, + uloc_getVariant, _kVariants, pErrorCode); +} + +/* Instead of having a separate pass for 'special' patterns, reintegrate the two + * so we don't get bitten by preflight bugs again. We can be reasonably efficient + * without two separate code paths, this code isn't that performance-critical. + * + * This code is general enough to deal with patterns that have a prefix or swap the + * language and remainder components, since we gave developers enough rope to do such + * things if they futz with the pattern data. But since we don't give them a way to + * specify a pattern for arbitrary combinations of components, there's not much use in + * that. I don't think our data includes such patterns, the only variable I know if is + * whether there is a space before the open paren, or not. Oh, and zh uses different + * chars than the standard open/close paren (which ja and ko use, btw). + */ +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayName(const char *locale, + const char *displayLocale, + char16_t *dest, int32_t destCapacity, + UErrorCode *pErrorCode) +{ + static const char16_t defaultSeparator[9] = { 0x007b, 0x0030, 0x007d, 0x002c, 0x0020, 0x007b, 0x0031, 0x007d, 0x0000 }; /* "{0}, {1}" */ + static const char16_t sub0[4] = { 0x007b, 0x0030, 0x007d , 0x0000 } ; /* {0} */ + static const char16_t sub1[4] = { 0x007b, 0x0031, 0x007d , 0x0000 } ; /* {1} */ + static const int32_t subLen = 3; + static const char16_t defaultPattern[10] = { + 0x007b, 0x0030, 0x007d, 0x0020, 0x0028, 0x007b, 0x0031, 0x007d, 0x0029, 0x0000 + }; /* {0} ({1}) */ + static const int32_t defaultPatLen = 9; + static const int32_t defaultSub0Pos = 0; + static const int32_t defaultSub1Pos = 5; + + int32_t length; /* of formatted result */ + + const char16_t *separator; + int32_t sepLen = 0; + const char16_t *pattern; + int32_t patLen = 0; + int32_t sub0Pos, sub1Pos; + + char16_t formatOpenParen = 0x0028; // ( + char16_t formatReplaceOpenParen = 0x005B; // [ + char16_t formatCloseParen = 0x0029; // ) + char16_t formatReplaceCloseParen = 0x005D; // ] + + UBool haveLang = true; /* assume true, set false if we find we don't have + a lang component in the locale */ + UBool haveRest = true; /* assume true, set false if we find we don't have + any other component in the locale */ + UBool retry = false; /* set true if we need to retry, see below */ + + int32_t langi = 0; /* index of the language substitution (0 or 1), virtually always 0 */ + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + if(destCapacity<0 || (destCapacity>0 && dest==nullptr)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + { + UErrorCode status = U_ZERO_ERROR; + + icu::LocalUResourceBundlePointer locbundle( + ures_open(U_ICUDATA_LANG, displayLocale, &status)); + icu::LocalUResourceBundlePointer dspbundle( + ures_getByKeyWithFallback(locbundle.getAlias(), _kLocaleDisplayPattern, nullptr, &status)); + + separator=ures_getStringByKeyWithFallback(dspbundle.getAlias(), _kSeparator, &sepLen, &status); + pattern=ures_getStringByKeyWithFallback(dspbundle.getAlias(), _kPattern, &patLen, &status); + } + + /* If we couldn't find any data, then use the defaults */ + if(sepLen == 0) { + separator = defaultSeparator; + } + /* #10244: Even though separator is now a pattern, it is awkward to handle it as such + * here since we are trying to build the display string in place in the dest buffer, + * and to handle it as a pattern would entail having separate storage for the + * substrings that need to be combined (the first of which may be the result of + * previous such combinations). So for now we continue to treat the portion between + * {0} and {1} as a string to be appended when joining substrings, ignoring anything + * that is before {0} or after {1} (no existing separator pattern has any such thing). + * This is similar to how pattern is handled below. + */ + { + char16_t *p0=u_strstr(separator, sub0); + char16_t *p1=u_strstr(separator, sub1); + if (p0==nullptr || p1==nullptr || p1(p1 - separator); + } + + if(patLen==0 || (patLen==defaultPatLen && !u_strncmp(pattern, defaultPattern, patLen))) { + pattern=defaultPattern; + patLen=defaultPatLen; + sub0Pos=defaultSub0Pos; + sub1Pos=defaultSub1Pos; + // use default formatOpenParen etc. set above + } else { /* non-default pattern */ + char16_t *p0=u_strstr(pattern, sub0); + char16_t *p1=u_strstr(pattern, sub1); + if (p0==nullptr || p1==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + sub0Pos = static_cast(p0-pattern); + sub1Pos = static_cast(p1-pattern); + if (sub1Pos < sub0Pos) { /* a very odd pattern */ + int32_t t=sub0Pos; sub0Pos=sub1Pos; sub1Pos=t; + langi=1; + } + if (u_strchr(pattern, 0xFF08) != nullptr) { + formatOpenParen = 0xFF08; // fullwidth ( + formatReplaceOpenParen = 0xFF3B; // fullwidth [ + formatCloseParen = 0xFF09; // fullwidth ) + formatReplaceCloseParen = 0xFF3D; // fullwidth ] + } + } + + /* We loop here because there is one case in which after the first pass we could need to + * reextract the data. If there's initial padding before the first element, we put in + * the padding and then write that element. If it turns out there's no second element, + * we didn't need the padding. If we do need the data (no preflight), and the first element + * would have fit but for the padding, we need to reextract. In this case (only) we + * adjust the parameters so padding is not added, and repeat. + */ + do { + char16_t* p=dest; + int32_t patPos=0; /* position in the pattern, used for non-substitution portions */ + int32_t langLen=0; /* length of language substitution */ + int32_t langPos=0; /* position in output of language substitution */ + int32_t restLen=0; /* length of 'everything else' substitution */ + int32_t restPos=0; /* position in output of 'everything else' substitution */ + icu::LocalUEnumerationPointer kenum; /* keyword enumeration */ + + /* prefix of pattern, extremely likely to be empty */ + if(sub0Pos) { + if(destCapacity >= sub0Pos) { + while (patPos < sub0Pos) { + *p++ = pattern[patPos++]; + } + } else { + patPos=sub0Pos; + } + length=sub0Pos; + } else { + length=0; + } + + for(int32_t subi=0,resti=0;subi<2;) { /* iterate through patterns 0 and 1*/ + UBool subdone = false; /* set true when ready to move to next substitution */ + + /* prep p and cap for calls to get display components, pin cap to 0 since + they complain if cap is negative */ + int32_t cap=destCapacity-length; + if (cap <= 0) { + cap=0; + } else { + p=dest+length; + } + + if (subi == langi) { /* {0}*/ + if(haveLang) { + langPos=length; + langLen=uloc_getDisplayLanguage(locale, displayLocale, p, cap, pErrorCode); + length+=langLen; + haveLang=langLen>0; + } + subdone=true; + } else { /* {1} */ + if(!haveRest) { + subdone=true; + } else { + int32_t len; /* length of component (plus other stuff) we just fetched */ + switch(resti++) { + case 0: + restPos=length; + len=uloc_getDisplayScriptInContext(locale, displayLocale, p, cap, pErrorCode); + break; + case 1: + len=uloc_getDisplayCountry(locale, displayLocale, p, cap, pErrorCode); + break; + case 2: + len=uloc_getDisplayVariant(locale, displayLocale, p, cap, pErrorCode); + break; + case 3: + kenum.adoptInstead(uloc_openKeywords(locale, pErrorCode)); + U_FALLTHROUGH; + default: { + const char* kw=uenum_next(kenum.getAlias(), &len, pErrorCode); + if (kw == nullptr) { + len=0; /* mark that we didn't add a component */ + subdone=true; + } else { + /* incorporating this behavior into the loop made it even more complex, + so just special case it here */ + len = uloc_getDisplayKeyword(kw, displayLocale, p, cap, pErrorCode); + if(len) { + if(len < cap) { + p[len]=0x3d; /* '=', assume we'll need it */ + } + len+=1; + + /* adjust for call to get keyword */ + cap-=len; + if(cap <= 0) { + cap=0; + } else { + p+=len; + } + } + /* reset for call below */ + if(*pErrorCode == U_BUFFER_OVERFLOW_ERROR) { + *pErrorCode=U_ZERO_ERROR; + } + int32_t vlen = uloc_getDisplayKeywordValue(locale, kw, displayLocale, + p, cap, pErrorCode); + if(len) { + if(vlen==0) { + --len; /* remove unneeded '=' */ + } + /* restore cap and p to what they were at start */ + cap=destCapacity-length; + if(cap <= 0) { + cap=0; + } else { + p=dest+length; + } + } + len+=vlen; /* total we added for key + '=' + value */ + } + } break; + } /* end switch */ + + if (len>0) { + /* we added a component, so add separator and write it if there's room. */ + if(len+sepLen<=cap) { + const char16_t * plimit = p + len; + for (; p < plimit; p++) { + if (*p == formatOpenParen) { + *p = formatReplaceOpenParen; + } else if (*p == formatCloseParen) { + *p = formatReplaceCloseParen; + } + } + for(int32_t i=0;i0; + } + } + } + + if(*pErrorCode == U_BUFFER_OVERFLOW_ERROR) { + *pErrorCode=U_ZERO_ERROR; + } + + if(subdone) { + if(haveLang && haveRest) { + /* append internal portion of pattern, the first time, + or last portion of pattern the second time */ + int32_t padLen; + patPos+=subLen; + padLen=(subi==0 ? sub1Pos : patLen)-patPos; + if(length+padLen <= destCapacity) { + p=dest+length; + for(int32_t i=0;i0) { + /* true length is the length of just the component we got. */ + length=haveLang?langLen:restLen; + if(dest && sub0Pos!=0) { + if (sub0Pos+length<=destCapacity) { + /* first component not at start of result, + but we have full component in buffer. */ + u_memmove(dest, dest+(haveLang?langPos:restPos), length); + } else { + /* would have fit, but didn't because of pattern prefix. */ + sub0Pos=0; /* stops initial padding (and a second retry, + so we won't end up here again) */ + retry=true; + } + } + } + + ++subi; /* move on to next substitution */ + } + } + } while(retry); + + return u_terminateUChars(dest, destCapacity, length, pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayKeyword(const char* keyword, + const char* displayLocale, + char16_t* dest, + int32_t destCapacity, + UErrorCode* status){ + + /* argument checking */ + if(status==nullptr || U_FAILURE(*status)) { + return 0; + } + + if(destCapacity<0 || (destCapacity>0 && dest==nullptr)) { + *status=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + + /* pass itemKey=nullptr to look for a top-level item */ + return _getStringOrCopyKey(U_ICUDATA_LANG, displayLocale, + _kKeys, nullptr, + keyword, + keyword, + dest, destCapacity, + status); + +} + + +#define UCURRENCY_DISPLAY_NAME_INDEX 1 + +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayKeywordValue( const char* locale, + const char* keyword, + const char* displayLocale, + char16_t* dest, + int32_t destCapacity, + UErrorCode* status){ + + + /* argument checking */ + if(status==nullptr || U_FAILURE(*status)) { + return 0; + } + + if(destCapacity<0 || (destCapacity>0 && dest==nullptr)) { + *status=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* get the keyword value */ + CharString keywordValue; + { + CharStringByteSink sink(&keywordValue); + ulocimp_getKeywordValue(locale, keyword, sink, status); + } + + /* + * if the keyword is equal to currency .. then to get the display name + * we need to do the fallback ourselves + */ + if(uprv_stricmp(keyword, _kCurrency)==0){ + + int32_t dispNameLen = 0; + const char16_t *dispName = nullptr; + + icu::LocalUResourceBundlePointer bundle( + ures_open(U_ICUDATA_CURR, displayLocale, status)); + icu::LocalUResourceBundlePointer currencies( + ures_getByKey(bundle.getAlias(), _kCurrencies, nullptr, status)); + icu::LocalUResourceBundlePointer currency( + ures_getByKeyWithFallback(currencies.getAlias(), keywordValue.data(), nullptr, status)); + + dispName = ures_getStringByIndex(currency.getAlias(), UCURRENCY_DISPLAY_NAME_INDEX, &dispNameLen, status); + + if(U_FAILURE(*status)){ + if(*status == U_MISSING_RESOURCE_ERROR){ + /* we just want to write the value over if nothing is available */ + *status = U_USING_DEFAULT_WARNING; + }else{ + return 0; + } + } + + /* now copy the dispName over if not nullptr */ + if(dispName != nullptr){ + if(dispNameLen <= destCapacity){ + u_memcpy(dest, dispName, dispNameLen); + return u_terminateUChars(dest, destCapacity, dispNameLen, status); + }else{ + *status = U_BUFFER_OVERFLOW_ERROR; + return dispNameLen; + } + }else{ + /* we have not found the display name for the value .. just copy over */ + if(keywordValue.length() <= destCapacity){ + u_charsToUChars(keywordValue.data(), dest, keywordValue.length()); + return u_terminateUChars(dest, destCapacity, keywordValue.length(), status); + }else{ + *status = U_BUFFER_OVERFLOW_ERROR; + return keywordValue.length(); + } + } + + + }else{ + + return _getStringOrCopyKey(U_ICUDATA_LANG, displayLocale, + _kTypes, keyword, + keywordValue.data(), + keywordValue.data(), + dest, destCapacity, + status); + } +} diff --git a/intl/icu/source/common/locdistance.cpp b/intl/icu/source/common/locdistance.cpp new file mode 100644 index 0000000000..fb22fe79ed --- /dev/null +++ b/intl/icu/source/common/locdistance.cpp @@ -0,0 +1,415 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// locdistance.cpp +// created: 2019may08 Markus W. Scherer + +#include "unicode/utypes.h" +#include "unicode/bytestrie.h" +#include "unicode/localematcher.h" +#include "unicode/locid.h" +#include "unicode/uobject.h" +#include "unicode/ures.h" +#include "cstring.h" +#include "locdistance.h" +#include "loclikelysubtags.h" +#include "uassert.h" +#include "ucln_cmn.h" +#include "uinvchar.h" +#include "umutex.h" + +U_NAMESPACE_BEGIN + +namespace { + +/** + * Bit flag used on the last character of a subtag in the trie. + * Must be set consistently by the builder and the lookup code. + */ +constexpr int32_t END_OF_SUBTAG = 0x80; +/** Distance value bit flag, set by the builder. */ +constexpr int32_t DISTANCE_SKIP_SCRIPT = 0x80; +/** Distance value bit flag, set by trieNext(). */ +constexpr int32_t DISTANCE_IS_FINAL = 0x100; +constexpr int32_t DISTANCE_IS_FINAL_OR_SKIP_SCRIPT = DISTANCE_IS_FINAL | DISTANCE_SKIP_SCRIPT; + +constexpr int32_t ABOVE_THRESHOLD = 100; + +// Indexes into array of distances. +enum { + IX_DEF_LANG_DISTANCE, + IX_DEF_SCRIPT_DISTANCE, + IX_DEF_REGION_DISTANCE, + IX_MIN_REGION_DISTANCE, + IX_LIMIT +}; + +LocaleDistance *gLocaleDistance = nullptr; +UInitOnce gInitOnce {}; + +UBool U_CALLCONV cleanup() { + delete gLocaleDistance; + gLocaleDistance = nullptr; + gInitOnce.reset(); + return true; +} + +} // namespace + +void U_CALLCONV LocaleDistance::initLocaleDistance(UErrorCode &errorCode) { + // This function is invoked only via umtx_initOnce(). + U_ASSERT(gLocaleDistance == nullptr); + const XLikelySubtags &likely = *XLikelySubtags::getSingleton(errorCode); + if (U_FAILURE(errorCode)) { return; } + const LocaleDistanceData &data = likely.getDistanceData(); + if (data.distanceTrieBytes == nullptr || + data.regionToPartitions == nullptr || data.partitions == nullptr || + // ok if no paradigms + data.distances == nullptr) { + errorCode = U_MISSING_RESOURCE_ERROR; + return; + } + gLocaleDistance = new LocaleDistance(data, likely); + if (gLocaleDistance == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + ucln_common_registerCleanup(UCLN_COMMON_LOCALE_DISTANCE, cleanup); +} + +const LocaleDistance *LocaleDistance::getSingleton(UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return nullptr; } + umtx_initOnce(gInitOnce, &LocaleDistance::initLocaleDistance, errorCode); + return gLocaleDistance; +} + +LocaleDistance::LocaleDistance(const LocaleDistanceData &data, const XLikelySubtags &likely) : + likelySubtags(likely), + trie(data.distanceTrieBytes), + regionToPartitionsIndex(data.regionToPartitions), partitionArrays(data.partitions), + paradigmLSRs(data.paradigms), paradigmLSRsLength(data.paradigmsLength), + defaultLanguageDistance(data.distances[IX_DEF_LANG_DISTANCE]), + defaultScriptDistance(data.distances[IX_DEF_SCRIPT_DISTANCE]), + defaultRegionDistance(data.distances[IX_DEF_REGION_DISTANCE]), + minRegionDistance(data.distances[IX_MIN_REGION_DISTANCE]) { + // For the default demotion value, use the + // default region distance between unrelated Englishes. + // Thus, unless demotion is turned off, + // a mere region difference for one desired locale + // is as good as a perfect match for the next following desired locale. + // As of CLDR 36, we have . + LSR en("en", "Latn", "US", LSR::EXPLICIT_LSR); + LSR enGB("en", "Latn", "GB", LSR::EXPLICIT_LSR); + const LSR *p_enGB = &enGB; + int32_t indexAndDistance = getBestIndexAndDistance(en, &p_enGB, 1, + shiftDistance(50), ULOCMATCH_FAVOR_LANGUAGE, ULOCMATCH_DIRECTION_WITH_ONE_WAY); + defaultDemotionPerDesiredLocale = getDistanceFloor(indexAndDistance); +} + +int32_t LocaleDistance::getBestIndexAndDistance( + const LSR &desired, + const LSR **supportedLSRs, int32_t supportedLSRsLength, + int32_t shiftedThreshold, + ULocMatchFavorSubtag favorSubtag, ULocMatchDirection direction) const { + BytesTrie iter(trie); + // Look up the desired language only once for all supported LSRs. + // Its "distance" is either a match point value of 0, or a non-match negative value. + // Note: The data builder verifies that there are no <*, supported> or rules. + int32_t desLangDistance = trieNext(iter, desired.language, false); + uint64_t desLangState = desLangDistance >= 0 && supportedLSRsLength > 1 ? iter.getState64() : 0; + // Index of the supported LSR with the lowest distance. + int32_t bestIndex = -1; + // Cached lookup info from XLikelySubtags.compareLikely(). + int32_t bestLikelyInfo = -1; + for (int32_t slIndex = 0; slIndex < supportedLSRsLength; ++slIndex) { + const LSR &supported = *supportedLSRs[slIndex]; + bool star = false; + int32_t distance = desLangDistance; + if (distance >= 0) { + U_ASSERT((distance & DISTANCE_IS_FINAL) == 0); + if (slIndex != 0) { + iter.resetToState64(desLangState); + } + distance = trieNext(iter, supported.language, true); + } + // Note: The data builder verifies that there are no rules with "any" (*) language and + // real (non *) script or region subtags. + // This means that if the lookup for either language fails we can use + // the default distances without further lookups. + int32_t flags; + if (distance >= 0) { + flags = distance & DISTANCE_IS_FINAL_OR_SKIP_SCRIPT; + distance &= ~DISTANCE_IS_FINAL_OR_SKIP_SCRIPT; + } else { // <*, *> + if (uprv_strcmp(desired.language, supported.language) == 0) { + distance = 0; + } else { + distance = defaultLanguageDistance; + } + flags = 0; + star = true; + } + U_ASSERT(0 <= distance && distance <= 100); + // Round up the shifted threshold (if fraction bits are not 0) + // for comparison with un-shifted distances until we need fraction bits. + // (If we simply shifted non-zero fraction bits away, then we might ignore a language + // when it's really still a micro distance below the threshold.) + int32_t roundedThreshold = (shiftedThreshold + DISTANCE_FRACTION_MASK) >> DISTANCE_SHIFT; + // We implement "favor subtag" by reducing the language subtag distance + // (unscientifically reducing it to a quarter of the normal value), + // so that the script distance is relatively more important. + // For example, given a default language distance of 80, we reduce it to 20, + // which is below the default threshold of 50, which is the default script distance. + if (favorSubtag == ULOCMATCH_FAVOR_SCRIPT) { + distance >>= 2; + } + // Let distance == roundedThreshold pass until the tie-breaker logic + // at the end of the loop. + if (distance > roundedThreshold) { + continue; + } + + int32_t scriptDistance; + if (star || flags != 0) { + if (uprv_strcmp(desired.script, supported.script) == 0) { + scriptDistance = 0; + } else { + scriptDistance = defaultScriptDistance; + } + } else { + scriptDistance = getDesSuppScriptDistance(iter, iter.getState64(), + desired.script, supported.script); + flags = scriptDistance & DISTANCE_IS_FINAL; + scriptDistance &= ~DISTANCE_IS_FINAL; + } + distance += scriptDistance; + if (distance > roundedThreshold) { + continue; + } + + if (uprv_strcmp(desired.region, supported.region) == 0) { + // regionDistance = 0 + } else if (star || (flags & DISTANCE_IS_FINAL) != 0) { + distance += defaultRegionDistance; + } else { + int32_t remainingThreshold = roundedThreshold - distance; + if (minRegionDistance > remainingThreshold) { + continue; + } + + // From here on we know the regions are not equal. + // Map each region to zero or more partitions. (zero = one non-matching string) + // (Each array of single-character partition strings is encoded as one string.) + // If either side has more than one, then we find the maximum distance. + // This could be optimized by adding some more structure, but probably not worth it. + distance += getRegionPartitionsDistance( + iter, iter.getState64(), + partitionsForRegion(desired), + partitionsForRegion(supported), + remainingThreshold); + } + int32_t shiftedDistance = shiftDistance(distance); + if (shiftedDistance == 0) { + // Distinguish between equivalent but originally unequal locales via an + // additional micro distance. + shiftedDistance |= (desired.flags ^ supported.flags); + if (shiftedDistance < shiftedThreshold) { + if (direction != ULOCMATCH_DIRECTION_ONLY_TWO_WAY || + // Is there also a match when we swap desired/supported? + isMatch(supported, desired, shiftedThreshold, favorSubtag)) { + if (shiftedDistance == 0) { + return slIndex << INDEX_SHIFT; + } + bestIndex = slIndex; + shiftedThreshold = shiftedDistance; + bestLikelyInfo = -1; + } + } + } else { + if (shiftedDistance < shiftedThreshold) { + if (direction != ULOCMATCH_DIRECTION_ONLY_TWO_WAY || + // Is there also a match when we swap desired/supported? + isMatch(supported, desired, shiftedThreshold, favorSubtag)) { + bestIndex = slIndex; + shiftedThreshold = shiftedDistance; + bestLikelyInfo = -1; + } + } else if (shiftedDistance == shiftedThreshold && bestIndex >= 0) { + if (direction != ULOCMATCH_DIRECTION_ONLY_TWO_WAY || + // Is there also a match when we swap desired/supported? + isMatch(supported, desired, shiftedThreshold, favorSubtag)) { + bestLikelyInfo = likelySubtags.compareLikely( + supported, *supportedLSRs[bestIndex], bestLikelyInfo); + if ((bestLikelyInfo & 1) != 0) { + // This supported locale matches as well as the previous best match, + // and neither matches perfectly, + // but this one is "more likely" (has more-default subtags). + bestIndex = slIndex; + } + } + } + } + } + return bestIndex >= 0 ? + (bestIndex << INDEX_SHIFT) | shiftedThreshold : + INDEX_NEG_1 | shiftDistance(ABOVE_THRESHOLD); +} + +int32_t LocaleDistance::getDesSuppScriptDistance( + BytesTrie &iter, uint64_t startState, const char *desired, const char *supported) { + // Note: The data builder verifies that there are no <*, supported> or rules. + int32_t distance = trieNext(iter, desired, false); + if (distance >= 0) { + distance = trieNext(iter, supported, true); + } + if (distance < 0) { + UStringTrieResult result = iter.resetToState64(startState).next(u'*'); // <*, *> + U_ASSERT(USTRINGTRIE_HAS_VALUE(result)); + if (uprv_strcmp(desired, supported) == 0) { + distance = 0; // same script + } else { + distance = iter.getValue(); + U_ASSERT(distance >= 0); + } + if (result == USTRINGTRIE_FINAL_VALUE) { + distance |= DISTANCE_IS_FINAL; + } + } + return distance; +} + +int32_t LocaleDistance::getRegionPartitionsDistance( + BytesTrie &iter, uint64_t startState, + const char *desiredPartitions, const char *supportedPartitions, int32_t threshold) { + char desired = *desiredPartitions++; + char supported = *supportedPartitions++; + U_ASSERT(desired != 0 && supported != 0); + // See if we have single desired/supported partitions, from NUL-terminated + // partition strings without explicit length. + bool suppLengthGt1 = *supportedPartitions != 0; // gt1: more than 1 character + // equivalent to: if (desLength == 1 && suppLength == 1) + if (*desiredPartitions == 0 && !suppLengthGt1) { + // Fastpath for single desired/supported partitions. + UStringTrieResult result = iter.next(uprv_invCharToAscii(desired) | END_OF_SUBTAG); + if (USTRINGTRIE_HAS_NEXT(result)) { + result = iter.next(uprv_invCharToAscii(supported) | END_OF_SUBTAG); + if (USTRINGTRIE_HAS_VALUE(result)) { + return iter.getValue(); + } + } + return getFallbackRegionDistance(iter, startState); + } + + const char *supportedStart = supportedPartitions - 1; // for restart of inner loop + int32_t regionDistance = 0; + // Fall back to * only once, not for each pair of partition strings. + bool star = false; + for (;;) { + // Look up each desired-partition string only once, + // not for each (desired, supported) pair. + UStringTrieResult result = iter.next(uprv_invCharToAscii(desired) | END_OF_SUBTAG); + if (USTRINGTRIE_HAS_NEXT(result)) { + uint64_t desState = suppLengthGt1 ? iter.getState64() : 0; + for (;;) { + result = iter.next(uprv_invCharToAscii(supported) | END_OF_SUBTAG); + int32_t d; + if (USTRINGTRIE_HAS_VALUE(result)) { + d = iter.getValue(); + } else if (star) { + d = 0; + } else { + d = getFallbackRegionDistance(iter, startState); + star = true; + } + if (d > threshold) { + return d; + } else if (regionDistance < d) { + regionDistance = d; + } + if ((supported = *supportedPartitions++) != 0) { + iter.resetToState64(desState); + } else { + break; + } + } + } else if (!star) { + int32_t d = getFallbackRegionDistance(iter, startState); + if (d > threshold) { + return d; + } else if (regionDistance < d) { + regionDistance = d; + } + star = true; + } + if ((desired = *desiredPartitions++) != 0) { + iter.resetToState64(startState); + supportedPartitions = supportedStart; + supported = *supportedPartitions++; + } else { + break; + } + } + return regionDistance; +} + +int32_t LocaleDistance::getFallbackRegionDistance(BytesTrie &iter, uint64_t startState) { +#if U_DEBUG + UStringTrieResult result = +#endif + iter.resetToState64(startState).next(u'*'); // <*, *> + U_ASSERT(USTRINGTRIE_HAS_VALUE(result)); + int32_t distance = iter.getValue(); + U_ASSERT(distance >= 0); + return distance; +} + +int32_t LocaleDistance::trieNext(BytesTrie &iter, const char *s, bool wantValue) { + uint8_t c; + if ((c = *s) == 0) { + return -1; // no empty subtags in the distance data + } + for (;;) { + c = uprv_invCharToAscii(c); + // EBCDIC: If *s is not an invariant character, + // then c is now 0 and will simply not match anything, which is harmless. + uint8_t next = *++s; + if (next != 0) { + if (!USTRINGTRIE_HAS_NEXT(iter.next(c))) { + return -1; + } + } else { + // last character of this subtag + UStringTrieResult result = iter.next(c | END_OF_SUBTAG); + if (wantValue) { + if (USTRINGTRIE_HAS_VALUE(result)) { + int32_t value = iter.getValue(); + if (result == USTRINGTRIE_FINAL_VALUE) { + value |= DISTANCE_IS_FINAL; + } + return value; + } + } else { + if (USTRINGTRIE_HAS_NEXT(result)) { + return 0; + } + } + return -1; + } + c = next; + } +} + +UBool LocaleDistance::isParadigmLSR(const LSR &lsr) const { + // Linear search for a very short list (length 6 as of 2019), + // because we look for equivalence not equality, and + // because it's easy. + // If there are many paradigm LSRs we should use a hash set + // with custom comparator and hasher. + U_ASSERT(paradigmLSRsLength <= 15); + for (int32_t i = 0; i < paradigmLSRsLength; ++i) { + if (lsr.isEquivalentTo(paradigmLSRs[i])) { return true; } + } + return false; +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/locdistance.h b/intl/icu/source/common/locdistance.h new file mode 100644 index 0000000000..51b777e627 --- /dev/null +++ b/intl/icu/source/common/locdistance.h @@ -0,0 +1,151 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// locdistance.h +// created: 2019may08 Markus W. Scherer + +#ifndef __LOCDISTANCE_H__ +#define __LOCDISTANCE_H__ + +#include "unicode/utypes.h" +#include "unicode/bytestrie.h" +#include "unicode/localematcher.h" +#include "unicode/locid.h" +#include "unicode/uobject.h" +#include "lsr.h" + +U_NAMESPACE_BEGIN + +struct LocaleDistanceData; + +/** + * Offline-built data for LocaleMatcher. + * Mostly but not only the data for mapping locales to their maximized forms. + */ +class LocaleDistance final : public UMemory { +public: + static const LocaleDistance *getSingleton(UErrorCode &errorCode); + + static int32_t shiftDistance(int32_t distance) { + return distance << DISTANCE_SHIFT; + } + + static int32_t getShiftedDistance(int32_t indexAndDistance) { + return indexAndDistance & DISTANCE_MASK; + } + + static double getDistanceDouble(int32_t indexAndDistance) { + double shiftedDistance = getShiftedDistance(indexAndDistance); + return shiftedDistance / (1 << DISTANCE_SHIFT); + } + + static int32_t getDistanceFloor(int32_t indexAndDistance) { + return (indexAndDistance & DISTANCE_MASK) >> DISTANCE_SHIFT; + } + + static int32_t getIndex(int32_t indexAndDistance) { + // assert indexAndDistance >= 0; + return indexAndDistance >> INDEX_SHIFT; + } + + /** + * Finds the supported LSR with the smallest distance from the desired one. + * Equivalent LSR subtags must be normalized into a canonical form. + * + *

Returns the index of the lowest-distance supported LSR in the high bits + * (negative if none has a distance below the threshold), + * and its distance (0..ABOVE_THRESHOLD) in the low bits. + */ + int32_t getBestIndexAndDistance(const LSR &desired, + const LSR **supportedLSRs, int32_t supportedLSRsLength, + int32_t shiftedThreshold, + ULocMatchFavorSubtag favorSubtag, + ULocMatchDirection direction) const; + + UBool isParadigmLSR(const LSR &lsr) const; + + int32_t getDefaultScriptDistance() const { + return defaultScriptDistance; + } + + int32_t getDefaultDemotionPerDesiredLocale() const { + return defaultDemotionPerDesiredLocale; + } + +private: + // The distance is shifted left to gain some fraction bits. + static constexpr int32_t DISTANCE_SHIFT = 3; + static constexpr int32_t DISTANCE_FRACTION_MASK = 7; + // 7 bits for 0..100 + static constexpr int32_t DISTANCE_INT_SHIFT = 7; + static constexpr int32_t INDEX_SHIFT = DISTANCE_INT_SHIFT + DISTANCE_SHIFT; + static constexpr int32_t DISTANCE_MASK = 0x3ff; + // tic constexpr int32_t MAX_INDEX = 0x1fffff; // avoids sign bit + static constexpr int32_t INDEX_NEG_1 = 0xfffffc00; + + LocaleDistance(const LocaleDistanceData &data, const XLikelySubtags &likely); + LocaleDistance(const LocaleDistance &other) = delete; + LocaleDistance &operator=(const LocaleDistance &other) = delete; + + static void initLocaleDistance(UErrorCode &errorCode); + + UBool isMatch(const LSR &desired, const LSR &supported, + int32_t shiftedThreshold, ULocMatchFavorSubtag favorSubtag) const { + const LSR *pSupp = &supported; + return getBestIndexAndDistance( + desired, &pSupp, 1, + shiftedThreshold, favorSubtag, ULOCMATCH_DIRECTION_WITH_ONE_WAY) >= 0; + } + + static int32_t getDesSuppScriptDistance(BytesTrie &iter, uint64_t startState, + const char *desired, const char *supported); + + static int32_t getRegionPartitionsDistance( + BytesTrie &iter, uint64_t startState, + const char *desiredPartitions, const char *supportedPartitions, + int32_t threshold); + + static int32_t getFallbackRegionDistance(BytesTrie &iter, uint64_t startState); + + static int32_t trieNext(BytesTrie &iter, const char *s, bool wantValue); + + const char *partitionsForRegion(const LSR &lsr) const { + // ill-formed region -> one non-matching string + int32_t pIndex = regionToPartitionsIndex[lsr.regionIndex]; + return partitionArrays[pIndex]; + } + + int32_t getDefaultRegionDistance() const { + return defaultRegionDistance; + } + + const XLikelySubtags &likelySubtags; + + // The trie maps each dlang+slang+dscript+sscript+dregion+sregion + // (encoded in ASCII with bit 7 set on the last character of each subtag) to a distance. + // There is also a trie value for each subsequence of whole subtags. + // One '*' is used for a (desired, supported) pair of "und", "Zzzz"/"", or "ZZ"/"". + BytesTrie trie; + + /** + * Maps each region to zero or more single-character partitions. + */ + const uint8_t *regionToPartitionsIndex; + const char **partitionArrays; + + /** + * Used to get the paradigm region for a cluster, if there is one. + */ + const LSR *paradigmLSRs; + int32_t paradigmLSRsLength; + + int32_t defaultLanguageDistance; + int32_t defaultScriptDistance; + int32_t defaultRegionDistance; + int32_t minRegionDistance; + int32_t defaultDemotionPerDesiredLocale; +}; + +U_NAMESPACE_END + +#endif // __LOCDISTANCE_H__ diff --git a/intl/icu/source/common/locdspnm.cpp b/intl/icu/source/common/locdspnm.cpp new file mode 100644 index 0000000000..73fe531c54 --- /dev/null +++ b/intl/icu/source/common/locdspnm.cpp @@ -0,0 +1,1136 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2016, International Business Machines Corporation and +* others. All Rights Reserved. +******************************************************************************* +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_FORMATTING + +#include "unicode/locdspnm.h" +#include "unicode/simpleformatter.h" +#include "unicode/ucasemap.h" +#include "unicode/ures.h" +#include "unicode/udisplaycontext.h" +#include "unicode/brkiter.h" +#include "unicode/ucurr.h" +#include "cmemory.h" +#include "cstring.h" +#include "mutex.h" +#include "ulocimp.h" +#include "umutex.h" +#include "ureslocs.h" +#include "uresimp.h" + +#include + +/** + * Concatenate a number of null-terminated strings to buffer, leaving a + * null-terminated string. The last argument should be the null pointer. + * Return the length of the string in the buffer, not counting the trailing + * null. Return -1 if there is an error (buffer is null, or buflen < 1). + */ +static int32_t ncat(char *buffer, uint32_t buflen, ...) { + va_list args; + char *str; + char *p = buffer; + const char* e = buffer + buflen - 1; + + if (buffer == nullptr || buflen < 1) { + return -1; + } + + va_start(args, buflen); + while ((str = va_arg(args, char *)) != 0) { + char c; + while (p != e && (c = *str++) != 0) { + *p++ = c; + } + } + *p = 0; + va_end(args); + + return static_cast(p - buffer); +} + +U_NAMESPACE_BEGIN + +//////////////////////////////////////////////////////////////////////////////////////////////////// + +// Access resource data for locale components. +// Wrap code in uloc.c for now. +class ICUDataTable { + const char* path; + Locale locale; + +public: + ICUDataTable(const char* path, const Locale& locale); + ~ICUDataTable(); + + const Locale& getLocale(); + + UnicodeString& get(const char* tableKey, const char* itemKey, + UnicodeString& result) const; + UnicodeString& get(const char* tableKey, const char* subTableKey, const char* itemKey, + UnicodeString& result) const; + + UnicodeString& getNoFallback(const char* tableKey, const char* itemKey, + UnicodeString &result) const; + UnicodeString& getNoFallback(const char* tableKey, const char* subTableKey, const char* itemKey, + UnicodeString &result) const; +}; + +inline UnicodeString & +ICUDataTable::get(const char* tableKey, const char* itemKey, UnicodeString& result) const { + return get(tableKey, nullptr, itemKey, result); +} + +inline UnicodeString & +ICUDataTable::getNoFallback(const char* tableKey, const char* itemKey, UnicodeString& result) const { + return getNoFallback(tableKey, nullptr, itemKey, result); +} + +ICUDataTable::ICUDataTable(const char* path, const Locale& locale) + : path(nullptr), locale(Locale::getRoot()) +{ + if (path) { + int32_t len = static_cast(uprv_strlen(path)); + this->path = (const char*) uprv_malloc(len + 1); + if (this->path) { + uprv_strcpy((char *)this->path, path); + this->locale = locale; + } + } +} + +ICUDataTable::~ICUDataTable() { + if (path) { + uprv_free((void*) path); + path = nullptr; + } +} + +const Locale& +ICUDataTable::getLocale() { + return locale; +} + +UnicodeString & +ICUDataTable::get(const char* tableKey, const char* subTableKey, const char* itemKey, + UnicodeString &result) const { + UErrorCode status = U_ZERO_ERROR; + int32_t len = 0; + + const char16_t *s = uloc_getTableStringWithFallback(path, locale.getName(), + tableKey, subTableKey, itemKey, + &len, &status); + if (U_SUCCESS(status) && len > 0) { + return result.setTo(s, len); + } + return result.setTo(UnicodeString(itemKey, -1, US_INV)); +} + +UnicodeString & +ICUDataTable::getNoFallback(const char* tableKey, const char* subTableKey, const char* itemKey, + UnicodeString& result) const { + UErrorCode status = U_ZERO_ERROR; + int32_t len = 0; + + const char16_t *s = uloc_getTableStringWithFallback(path, locale.getName(), + tableKey, subTableKey, itemKey, + &len, &status); + if (U_SUCCESS(status)) { + return result.setTo(s, len); + } + + result.setToBogus(); + return result; +} + +//////////////////////////////////////////////////////////////////////////////////////////////////// + +LocaleDisplayNames::~LocaleDisplayNames() {} + +//////////////////////////////////////////////////////////////////////////////////////////////////// + +#if 0 // currently unused + +class DefaultLocaleDisplayNames : public LocaleDisplayNames { + UDialectHandling dialectHandling; + +public: + // constructor + DefaultLocaleDisplayNames(UDialectHandling dialectHandling); + + virtual ~DefaultLocaleDisplayNames(); + + virtual const Locale& getLocale() const; + virtual UDialectHandling getDialectHandling() const; + + virtual UnicodeString& localeDisplayName(const Locale& locale, + UnicodeString& result) const; + virtual UnicodeString& localeDisplayName(const char* localeId, + UnicodeString& result) const; + virtual UnicodeString& languageDisplayName(const char* lang, + UnicodeString& result) const; + virtual UnicodeString& scriptDisplayName(const char* script, + UnicodeString& result) const; + virtual UnicodeString& scriptDisplayName(UScriptCode scriptCode, + UnicodeString& result) const; + virtual UnicodeString& regionDisplayName(const char* region, + UnicodeString& result) const; + virtual UnicodeString& variantDisplayName(const char* variant, + UnicodeString& result) const; + virtual UnicodeString& keyDisplayName(const char* key, + UnicodeString& result) const; + virtual UnicodeString& keyValueDisplayName(const char* key, + const char* value, + UnicodeString& result) const; +}; + +DefaultLocaleDisplayNames::DefaultLocaleDisplayNames(UDialectHandling dialectHandling) + : dialectHandling(dialectHandling) { +} + +DefaultLocaleDisplayNames::~DefaultLocaleDisplayNames() { +} + +const Locale& +DefaultLocaleDisplayNames::getLocale() const { + return Locale::getRoot(); +} + +UDialectHandling +DefaultLocaleDisplayNames::getDialectHandling() const { + return dialectHandling; +} + +UnicodeString& +DefaultLocaleDisplayNames::localeDisplayName(const Locale& locale, + UnicodeString& result) const { + return result = UnicodeString(locale.getName(), -1, US_INV); +} + +UnicodeString& +DefaultLocaleDisplayNames::localeDisplayName(const char* localeId, + UnicodeString& result) const { + return result = UnicodeString(localeId, -1, US_INV); +} + +UnicodeString& +DefaultLocaleDisplayNames::languageDisplayName(const char* lang, + UnicodeString& result) const { + return result = UnicodeString(lang, -1, US_INV); +} + +UnicodeString& +DefaultLocaleDisplayNames::scriptDisplayName(const char* script, + UnicodeString& result) const { + return result = UnicodeString(script, -1, US_INV); +} + +UnicodeString& +DefaultLocaleDisplayNames::scriptDisplayName(UScriptCode scriptCode, + UnicodeString& result) const { + const char* name = uscript_getName(scriptCode); + if (name) { + return result = UnicodeString(name, -1, US_INV); + } + return result.remove(); +} + +UnicodeString& +DefaultLocaleDisplayNames::regionDisplayName(const char* region, + UnicodeString& result) const { + return result = UnicodeString(region, -1, US_INV); +} + +UnicodeString& +DefaultLocaleDisplayNames::variantDisplayName(const char* variant, + UnicodeString& result) const { + return result = UnicodeString(variant, -1, US_INV); +} + +UnicodeString& +DefaultLocaleDisplayNames::keyDisplayName(const char* key, + UnicodeString& result) const { + return result = UnicodeString(key, -1, US_INV); +} + +UnicodeString& +DefaultLocaleDisplayNames::keyValueDisplayName(const char* /* key */, + const char* value, + UnicodeString& result) const { + return result = UnicodeString(value, -1, US_INV); +} + +#endif // currently unused class DefaultLocaleDisplayNames + +//////////////////////////////////////////////////////////////////////////////////////////////////// + +class LocaleDisplayNamesImpl : public LocaleDisplayNames { + Locale locale; + UDialectHandling dialectHandling; + ICUDataTable langData; + ICUDataTable regionData; + SimpleFormatter separatorFormat; + SimpleFormatter format; + SimpleFormatter keyTypeFormat; + UDisplayContext capitalizationContext; +#if !UCONFIG_NO_BREAK_ITERATION + BreakIterator* capitalizationBrkIter; +#else + UObject* capitalizationBrkIter; +#endif + UnicodeString formatOpenParen; + UnicodeString formatReplaceOpenParen; + UnicodeString formatCloseParen; + UnicodeString formatReplaceCloseParen; + UDisplayContext nameLength; + UDisplayContext substitute; + + // Constants for capitalization context usage types. + enum CapContextUsage { + kCapContextUsageLanguage, + kCapContextUsageScript, + kCapContextUsageTerritory, + kCapContextUsageVariant, + kCapContextUsageKey, + kCapContextUsageKeyValue, + kCapContextUsageCount + }; + // Capitalization transforms. For each usage type, indicates whether to titlecase for + // the context specified in capitalizationContext (which we know at construction time) + UBool fCapitalization[kCapContextUsageCount]; + +public: + // constructor + LocaleDisplayNamesImpl(const Locale& locale, UDialectHandling dialectHandling); + LocaleDisplayNamesImpl(const Locale& locale, UDisplayContext *contexts, int32_t length); + virtual ~LocaleDisplayNamesImpl(); + + virtual const Locale& getLocale() const override; + virtual UDialectHandling getDialectHandling() const override; + virtual UDisplayContext getContext(UDisplayContextType type) const override; + + virtual UnicodeString& localeDisplayName(const Locale& locale, + UnicodeString& result) const override; + virtual UnicodeString& localeDisplayName(const char* localeId, + UnicodeString& result) const override; + virtual UnicodeString& languageDisplayName(const char* lang, + UnicodeString& result) const override; + virtual UnicodeString& scriptDisplayName(const char* script, + UnicodeString& result) const override; + virtual UnicodeString& scriptDisplayName(UScriptCode scriptCode, + UnicodeString& result) const override; + virtual UnicodeString& regionDisplayName(const char* region, + UnicodeString& result) const override; + virtual UnicodeString& variantDisplayName(const char* variant, + UnicodeString& result) const override; + virtual UnicodeString& keyDisplayName(const char* key, + UnicodeString& result) const override; + virtual UnicodeString& keyValueDisplayName(const char* key, + const char* value, + UnicodeString& result) const override; +private: + UnicodeString& localeIdName(const char* localeId, + UnicodeString& result, bool substitute) const; + UnicodeString& appendWithSep(UnicodeString& buffer, const UnicodeString& src) const; + UnicodeString& adjustForUsageAndContext(CapContextUsage usage, UnicodeString& result) const; + UnicodeString& scriptDisplayName(const char* script, UnicodeString& result, UBool skipAdjust) const; + UnicodeString& regionDisplayName(const char* region, UnicodeString& result, UBool skipAdjust) const; + UnicodeString& variantDisplayName(const char* variant, UnicodeString& result, UBool skipAdjust) const; + UnicodeString& keyDisplayName(const char* key, UnicodeString& result, UBool skipAdjust) const; + UnicodeString& keyValueDisplayName(const char* key, const char* value, + UnicodeString& result, UBool skipAdjust) const; + void initialize(); + + struct CapitalizationContextSink; +}; + +LocaleDisplayNamesImpl::LocaleDisplayNamesImpl(const Locale& locale, + UDialectHandling dialectHandling) + : dialectHandling(dialectHandling) + , langData(U_ICUDATA_LANG, locale) + , regionData(U_ICUDATA_REGION, locale) + , capitalizationContext(UDISPCTX_CAPITALIZATION_NONE) + , capitalizationBrkIter(nullptr) + , nameLength(UDISPCTX_LENGTH_FULL) + , substitute(UDISPCTX_SUBSTITUTE) +{ + initialize(); +} + +LocaleDisplayNamesImpl::LocaleDisplayNamesImpl(const Locale& locale, + UDisplayContext *contexts, int32_t length) + : dialectHandling(ULDN_STANDARD_NAMES) + , langData(U_ICUDATA_LANG, locale) + , regionData(U_ICUDATA_REGION, locale) + , capitalizationContext(UDISPCTX_CAPITALIZATION_NONE) + , capitalizationBrkIter(nullptr) + , nameLength(UDISPCTX_LENGTH_FULL) + , substitute(UDISPCTX_SUBSTITUTE) +{ + while (length-- > 0) { + UDisplayContext value = *contexts++; + UDisplayContextType selector = (UDisplayContextType)((uint32_t)value >> 8); + switch (selector) { + case UDISPCTX_TYPE_DIALECT_HANDLING: + dialectHandling = (UDialectHandling)value; + break; + case UDISPCTX_TYPE_CAPITALIZATION: + capitalizationContext = value; + break; + case UDISPCTX_TYPE_DISPLAY_LENGTH: + nameLength = value; + break; + case UDISPCTX_TYPE_SUBSTITUTE_HANDLING: + substitute = value; + break; + default: + break; + } + } + initialize(); +} + +struct LocaleDisplayNamesImpl::CapitalizationContextSink : public ResourceSink { + UBool hasCapitalizationUsage; + LocaleDisplayNamesImpl& parent; + + CapitalizationContextSink(LocaleDisplayNamesImpl& _parent) + : hasCapitalizationUsage(false), parent(_parent) {} + virtual ~CapitalizationContextSink(); + + virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/, + UErrorCode &errorCode) override { + ResourceTable contexts = value.getTable(errorCode); + if (U_FAILURE(errorCode)) { return; } + for (int i = 0; contexts.getKeyAndValue(i, key, value); ++i) { + + CapContextUsage usageEnum; + if (uprv_strcmp(key, "key") == 0) { + usageEnum = kCapContextUsageKey; + } else if (uprv_strcmp(key, "keyValue") == 0) { + usageEnum = kCapContextUsageKeyValue; + } else if (uprv_strcmp(key, "languages") == 0) { + usageEnum = kCapContextUsageLanguage; + } else if (uprv_strcmp(key, "script") == 0) { + usageEnum = kCapContextUsageScript; + } else if (uprv_strcmp(key, "territory") == 0) { + usageEnum = kCapContextUsageTerritory; + } else if (uprv_strcmp(key, "variant") == 0) { + usageEnum = kCapContextUsageVariant; + } else { + continue; + } + + int32_t len = 0; + const int32_t* intVector = value.getIntVector(len, errorCode); + if (U_FAILURE(errorCode)) { return; } + if (len < 2) { continue; } + + int32_t titlecaseInt = (parent.capitalizationContext == UDISPCTX_CAPITALIZATION_FOR_UI_LIST_OR_MENU) ? intVector[0] : intVector[1]; + if (titlecaseInt == 0) { continue; } + + parent.fCapitalization[usageEnum] = true; + hasCapitalizationUsage = true; + } + } +}; + +// Virtual destructors must be defined out of line. +LocaleDisplayNamesImpl::CapitalizationContextSink::~CapitalizationContextSink() {} + +void +LocaleDisplayNamesImpl::initialize() { + LocaleDisplayNamesImpl *nonConstThis = (LocaleDisplayNamesImpl *)this; + nonConstThis->locale = langData.getLocale() == Locale::getRoot() + ? regionData.getLocale() + : langData.getLocale(); + + UnicodeString sep; + langData.getNoFallback("localeDisplayPattern", "separator", sep); + if (sep.isBogus()) { + sep = UnicodeString("{0}, {1}", -1, US_INV); + } + UErrorCode status = U_ZERO_ERROR; + separatorFormat.applyPatternMinMaxArguments(sep, 2, 2, status); + + UnicodeString pattern; + langData.getNoFallback("localeDisplayPattern", "pattern", pattern); + if (pattern.isBogus()) { + pattern = UnicodeString("{0} ({1})", -1, US_INV); + } + format.applyPatternMinMaxArguments(pattern, 2, 2, status); + if (pattern.indexOf((char16_t)0xFF08) >= 0) { + formatOpenParen.setTo((char16_t)0xFF08); // fullwidth ( + formatReplaceOpenParen.setTo((char16_t)0xFF3B); // fullwidth [ + formatCloseParen.setTo((char16_t)0xFF09); // fullwidth ) + formatReplaceCloseParen.setTo((char16_t)0xFF3D); // fullwidth ] + } else { + formatOpenParen.setTo((char16_t)0x0028); // ( + formatReplaceOpenParen.setTo((char16_t)0x005B); // [ + formatCloseParen.setTo((char16_t)0x0029); // ) + formatReplaceCloseParen.setTo((char16_t)0x005D); // ] + } + + UnicodeString ktPattern; + langData.get("localeDisplayPattern", "keyTypePattern", ktPattern); + if (ktPattern.isBogus()) { + ktPattern = UnicodeString("{0}={1}", -1, US_INV); + } + keyTypeFormat.applyPatternMinMaxArguments(ktPattern, 2, 2, status); + + uprv_memset(fCapitalization, 0, sizeof(fCapitalization)); +#if !UCONFIG_NO_BREAK_ITERATION + // Only get the context data if we need it! This is a const object so we know now... + // Also check whether we will need a break iterator (depends on the data) + UBool needBrkIter = false; + if (capitalizationContext == UDISPCTX_CAPITALIZATION_FOR_UI_LIST_OR_MENU || capitalizationContext == UDISPCTX_CAPITALIZATION_FOR_STANDALONE) { + LocalUResourceBundlePointer resource(ures_open(nullptr, locale.getName(), &status)); + if (U_FAILURE(status)) { return; } + CapitalizationContextSink sink(*this); + ures_getAllItemsWithFallback(resource.getAlias(), "contextTransforms", sink, status); + if (status == U_MISSING_RESOURCE_ERROR) { + // Silently ignore. Not every locale has contextTransforms. + status = U_ZERO_ERROR; + } else if (U_FAILURE(status)) { + return; + } + needBrkIter = sink.hasCapitalizationUsage; + } + // Get a sentence break iterator if we will need it + if (needBrkIter || capitalizationContext == UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE) { + status = U_ZERO_ERROR; + capitalizationBrkIter = BreakIterator::createSentenceInstance(locale, status); + if (U_FAILURE(status)) { + delete capitalizationBrkIter; + capitalizationBrkIter = nullptr; + } + } +#endif +} + +LocaleDisplayNamesImpl::~LocaleDisplayNamesImpl() { +#if !UCONFIG_NO_BREAK_ITERATION + delete capitalizationBrkIter; +#endif +} + +const Locale& +LocaleDisplayNamesImpl::getLocale() const { + return locale; +} + +UDialectHandling +LocaleDisplayNamesImpl::getDialectHandling() const { + return dialectHandling; +} + +UDisplayContext +LocaleDisplayNamesImpl::getContext(UDisplayContextType type) const { + switch (type) { + case UDISPCTX_TYPE_DIALECT_HANDLING: + return (UDisplayContext)dialectHandling; + case UDISPCTX_TYPE_CAPITALIZATION: + return capitalizationContext; + case UDISPCTX_TYPE_DISPLAY_LENGTH: + return nameLength; + case UDISPCTX_TYPE_SUBSTITUTE_HANDLING: + return substitute; + default: + break; + } + return (UDisplayContext)0; +} + +UnicodeString& +LocaleDisplayNamesImpl::adjustForUsageAndContext(CapContextUsage usage, + UnicodeString& result) const { +#if !UCONFIG_NO_BREAK_ITERATION + // check to see whether we need to titlecase result + if ( result.length() > 0 && u_islower(result.char32At(0)) && capitalizationBrkIter!= nullptr && + ( capitalizationContext==UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE || fCapitalization[usage] ) ) { + // note fCapitalization[usage] won't be set unless capitalizationContext is UI_LIST_OR_MENU or STANDALONE + static UMutex capitalizationBrkIterLock; + Mutex lock(&capitalizationBrkIterLock); + result.toTitle(capitalizationBrkIter, locale, U_TITLECASE_NO_LOWERCASE | U_TITLECASE_NO_BREAK_ADJUSTMENT); + } +#endif + return result; +} + +UnicodeString& +LocaleDisplayNamesImpl::localeDisplayName(const Locale& loc, + UnicodeString& result) const { + if (loc.isBogus()) { + result.setToBogus(); + return result; + } + UnicodeString resultName; + + const char* lang = loc.getLanguage(); + if (uprv_strlen(lang) == 0) { + lang = "root"; + } + const char* script = loc.getScript(); + const char* country = loc.getCountry(); + const char* variant = loc.getVariant(); + + UBool hasScript = uprv_strlen(script) > 0; + UBool hasCountry = uprv_strlen(country) > 0; + UBool hasVariant = uprv_strlen(variant) > 0; + + if (dialectHandling == ULDN_DIALECT_NAMES) { + char buffer[ULOC_FULLNAME_CAPACITY]; + do { // loop construct is so we can break early out of search + if (hasScript && hasCountry) { + ncat(buffer, ULOC_FULLNAME_CAPACITY, lang, "_", script, "_", country, (char *)0); + localeIdName(buffer, resultName, false); + if (!resultName.isBogus()) { + hasScript = false; + hasCountry = false; + break; + } + } + if (hasScript) { + ncat(buffer, ULOC_FULLNAME_CAPACITY, lang, "_", script, (char *)0); + localeIdName(buffer, resultName, false); + if (!resultName.isBogus()) { + hasScript = false; + break; + } + } + if (hasCountry) { + ncat(buffer, ULOC_FULLNAME_CAPACITY, lang, "_", country, (char*)0); + localeIdName(buffer, resultName, false); + if (!resultName.isBogus()) { + hasCountry = false; + break; + } + } + } while (false); + } + if (resultName.isBogus() || resultName.isEmpty()) { + localeIdName(lang, resultName, substitute == UDISPCTX_SUBSTITUTE); + if (resultName.isBogus()) { + result.setToBogus(); + return result; + } + } + + UnicodeString resultRemainder; + UnicodeString temp; + UErrorCode status = U_ZERO_ERROR; + + if (hasScript) { + UnicodeString script_str = scriptDisplayName(script, temp, true); + if (script_str.isBogus()) { + result.setToBogus(); + return result; + } + resultRemainder.append(script_str); + } + if (hasCountry) { + UnicodeString region_str = regionDisplayName(country, temp, true); + if (region_str.isBogus()) { + result.setToBogus(); + return result; + } + appendWithSep(resultRemainder, region_str); + } + if (hasVariant) { + UnicodeString variant_str = variantDisplayName(variant, temp, true); + if (variant_str.isBogus()) { + result.setToBogus(); + return result; + } + appendWithSep(resultRemainder, variant_str); + } + resultRemainder.findAndReplace(formatOpenParen, formatReplaceOpenParen); + resultRemainder.findAndReplace(formatCloseParen, formatReplaceCloseParen); + + LocalPointer e(loc.createKeywords(status)); + if (e.isValid() && U_SUCCESS(status)) { + UnicodeString temp2; + char value[ULOC_KEYWORD_AND_VALUES_CAPACITY]; // sigh, no ULOC_VALUE_CAPACITY + const char* key; + while ((key = e->next((int32_t *)0, status)) != nullptr) { + value[0] = 0; + loc.getKeywordValue(key, value, ULOC_KEYWORD_AND_VALUES_CAPACITY, status); + if (U_FAILURE(status) || status == U_STRING_NOT_TERMINATED_WARNING) { + return result; + } + keyDisplayName(key, temp, true); + temp.findAndReplace(formatOpenParen, formatReplaceOpenParen); + temp.findAndReplace(formatCloseParen, formatReplaceCloseParen); + keyValueDisplayName(key, value, temp2, true); + temp2.findAndReplace(formatOpenParen, formatReplaceOpenParen); + temp2.findAndReplace(formatCloseParen, formatReplaceCloseParen); + if (temp2 != UnicodeString(value, -1, US_INV)) { + appendWithSep(resultRemainder, temp2); + } else if (temp != UnicodeString(key, -1, US_INV)) { + UnicodeString temp3; + keyTypeFormat.format(temp, temp2, temp3, status); + appendWithSep(resultRemainder, temp3); + } else { + appendWithSep(resultRemainder, temp) + .append((char16_t)0x3d /* = */) + .append(temp2); + } + } + } + + if (!resultRemainder.isEmpty()) { + format.format(resultName, resultRemainder, result.remove(), status); + return adjustForUsageAndContext(kCapContextUsageLanguage, result); + } + + result = resultName; + return adjustForUsageAndContext(kCapContextUsageLanguage, result); +} + +UnicodeString& +LocaleDisplayNamesImpl::appendWithSep(UnicodeString& buffer, const UnicodeString& src) const { + if (buffer.isEmpty()) { + buffer.setTo(src); + } else { + const UnicodeString *values[2] = { &buffer, &src }; + UErrorCode status = U_ZERO_ERROR; + separatorFormat.formatAndReplace(values, 2, buffer, nullptr, 0, status); + } + return buffer; +} + +UnicodeString& +LocaleDisplayNamesImpl::localeDisplayName(const char* localeId, + UnicodeString& result) const { + return localeDisplayName(Locale(localeId), result); +} + +// private +UnicodeString& +LocaleDisplayNamesImpl::localeIdName(const char* localeId, + UnicodeString& result, bool substitute) const { + if (nameLength == UDISPCTX_LENGTH_SHORT) { + langData.getNoFallback("Languages%short", localeId, result); + if (!result.isBogus()) { + return result; + } + } + langData.getNoFallback("Languages", localeId, result); + if (result.isBogus() && uprv_strchr(localeId, '_') == nullptr) { + // Canonicalize lang and try again, ICU-20870 + // (only for language codes without script or region) + Locale canonLocale = Locale::createCanonical(localeId); + const char* canonLocId = canonLocale.getName(); + if (nameLength == UDISPCTX_LENGTH_SHORT) { + langData.getNoFallback("Languages%short", canonLocId, result); + if (!result.isBogus()) { + return result; + } + } + langData.getNoFallback("Languages", canonLocId, result); + } + if (result.isBogus() && substitute) { + // use key, this is what langData.get (with fallback) falls back to. + result.setTo(UnicodeString(localeId, -1, US_INV)); // use key ( + } + return result; +} + +UnicodeString& +LocaleDisplayNamesImpl::languageDisplayName(const char* lang, + UnicodeString& result) const { + if (uprv_strcmp("root", lang) == 0 || uprv_strchr(lang, '_') != nullptr) { + return result = UnicodeString(lang, -1, US_INV); + } + if (nameLength == UDISPCTX_LENGTH_SHORT) { + langData.getNoFallback("Languages%short", lang, result); + if (!result.isBogus()) { + return adjustForUsageAndContext(kCapContextUsageLanguage, result); + } + } + langData.getNoFallback("Languages", lang, result); + if (result.isBogus()) { + // Canonicalize lang and try again, ICU-20870 + Locale canonLocale = Locale::createCanonical(lang); + const char* canonLocId = canonLocale.getName(); + if (nameLength == UDISPCTX_LENGTH_SHORT) { + langData.getNoFallback("Languages%short", canonLocId, result); + if (!result.isBogus()) { + return adjustForUsageAndContext(kCapContextUsageLanguage, result); + } + } + langData.getNoFallback("Languages", canonLocId, result); + } + if (result.isBogus() && substitute == UDISPCTX_SUBSTITUTE) { + // use key, this is what langData.get (with fallback) falls back to. + result.setTo(UnicodeString(lang, -1, US_INV)); // use key ( + } + return adjustForUsageAndContext(kCapContextUsageLanguage, result); +} + +UnicodeString& +LocaleDisplayNamesImpl::scriptDisplayName(const char* script, + UnicodeString& result, + UBool skipAdjust) const { + if (nameLength == UDISPCTX_LENGTH_SHORT) { + langData.getNoFallback("Scripts%short", script, result); + if (!result.isBogus()) { + return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageScript, result); + } + } + if (substitute == UDISPCTX_SUBSTITUTE) { + langData.get("Scripts", script, result); + } else { + langData.getNoFallback("Scripts", script, result); + } + return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageScript, result); +} + +UnicodeString& +LocaleDisplayNamesImpl::scriptDisplayName(const char* script, + UnicodeString& result) const { + return scriptDisplayName(script, result, false); +} + +UnicodeString& +LocaleDisplayNamesImpl::scriptDisplayName(UScriptCode scriptCode, + UnicodeString& result) const { + return scriptDisplayName(uscript_getName(scriptCode), result, false); +} + +UnicodeString& +LocaleDisplayNamesImpl::regionDisplayName(const char* region, + UnicodeString& result, + UBool skipAdjust) const { + if (nameLength == UDISPCTX_LENGTH_SHORT) { + regionData.getNoFallback("Countries%short", region, result); + if (!result.isBogus()) { + return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageTerritory, result); + } + } + if (substitute == UDISPCTX_SUBSTITUTE) { + regionData.get("Countries", region, result); + } else { + regionData.getNoFallback("Countries", region, result); + } + return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageTerritory, result); +} + +UnicodeString& +LocaleDisplayNamesImpl::regionDisplayName(const char* region, + UnicodeString& result) const { + return regionDisplayName(region, result, false); +} + + +UnicodeString& +LocaleDisplayNamesImpl::variantDisplayName(const char* variant, + UnicodeString& result, + UBool skipAdjust) const { + // don't have a resource for short variant names + if (substitute == UDISPCTX_SUBSTITUTE) { + langData.get("Variants", variant, result); + } else { + langData.getNoFallback("Variants", variant, result); + } + return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageVariant, result); +} + +UnicodeString& +LocaleDisplayNamesImpl::variantDisplayName(const char* variant, + UnicodeString& result) const { + return variantDisplayName(variant, result, false); +} + +UnicodeString& +LocaleDisplayNamesImpl::keyDisplayName(const char* key, + UnicodeString& result, + UBool skipAdjust) const { + // don't have a resource for short key names + if (substitute == UDISPCTX_SUBSTITUTE) { + langData.get("Keys", key, result); + } else { + langData.getNoFallback("Keys", key, result); + } + return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageKey, result); +} + +UnicodeString& +LocaleDisplayNamesImpl::keyDisplayName(const char* key, + UnicodeString& result) const { + return keyDisplayName(key, result, false); +} + +UnicodeString& +LocaleDisplayNamesImpl::keyValueDisplayName(const char* key, + const char* value, + UnicodeString& result, + UBool skipAdjust) const { + if (uprv_strcmp(key, "currency") == 0) { + // ICU4C does not have ICU4J CurrencyDisplayInfo equivalent for now. + UErrorCode sts = U_ZERO_ERROR; + UnicodeString ustrValue(value, -1, US_INV); + int32_t len; + const char16_t *currencyName = ucurr_getName(ustrValue.getTerminatedBuffer(), + locale.getBaseName(), UCURR_LONG_NAME, nullptr /* isChoiceFormat */, &len, &sts); + if (U_FAILURE(sts)) { + // Return the value as is on failure + result = ustrValue; + return result; + } + result.setTo(currencyName, len); + return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageKeyValue, result); + } + + if (nameLength == UDISPCTX_LENGTH_SHORT) { + langData.getNoFallback("Types%short", key, value, result); + if (!result.isBogus()) { + return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageKeyValue, result); + } + } + if (substitute == UDISPCTX_SUBSTITUTE) { + langData.get("Types", key, value, result); + } else { + langData.getNoFallback("Types", key, value, result); + } + return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageKeyValue, result); +} + +UnicodeString& +LocaleDisplayNamesImpl::keyValueDisplayName(const char* key, + const char* value, + UnicodeString& result) const { + return keyValueDisplayName(key, value, result, false); +} + +//////////////////////////////////////////////////////////////////////////////////////////////////// + +LocaleDisplayNames* +LocaleDisplayNames::createInstance(const Locale& locale, + UDialectHandling dialectHandling) { + return new LocaleDisplayNamesImpl(locale, dialectHandling); +} + +LocaleDisplayNames* +LocaleDisplayNames::createInstance(const Locale& locale, + UDisplayContext *contexts, int32_t length) { + if (contexts == nullptr) { + length = 0; + } + return new LocaleDisplayNamesImpl(locale, contexts, length); +} + +U_NAMESPACE_END + +//////////////////////////////////////////////////////////////////////////////////////////////////// + +U_NAMESPACE_USE + +U_CAPI ULocaleDisplayNames * U_EXPORT2 +uldn_open(const char * locale, + UDialectHandling dialectHandling, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return 0; + } + if (locale == nullptr) { + locale = uloc_getDefault(); + } + return (ULocaleDisplayNames *)LocaleDisplayNames::createInstance(Locale(locale), dialectHandling); +} + +U_CAPI ULocaleDisplayNames * U_EXPORT2 +uldn_openForContext(const char * locale, + UDisplayContext *contexts, int32_t length, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return 0; + } + if (locale == nullptr) { + locale = uloc_getDefault(); + } + return (ULocaleDisplayNames *)LocaleDisplayNames::createInstance(Locale(locale), contexts, length); +} + + +U_CAPI void U_EXPORT2 +uldn_close(ULocaleDisplayNames *ldn) { + delete (LocaleDisplayNames *)ldn; +} + +U_CAPI const char * U_EXPORT2 +uldn_getLocale(const ULocaleDisplayNames *ldn) { + if (ldn) { + return ((const LocaleDisplayNames *)ldn)->getLocale().getName(); + } + return nullptr; +} + +U_CAPI UDialectHandling U_EXPORT2 +uldn_getDialectHandling(const ULocaleDisplayNames *ldn) { + if (ldn) { + return ((const LocaleDisplayNames *)ldn)->getDialectHandling(); + } + return ULDN_STANDARD_NAMES; +} + +U_CAPI UDisplayContext U_EXPORT2 +uldn_getContext(const ULocaleDisplayNames *ldn, + UDisplayContextType type, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return (UDisplayContext)0; + } + return ((const LocaleDisplayNames *)ldn)->getContext(type); +} + +U_CAPI int32_t U_EXPORT2 +uldn_localeDisplayName(const ULocaleDisplayNames *ldn, + const char *locale, + char16_t *result, + int32_t maxResultSize, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return 0; + } + if (ldn == nullptr || locale == nullptr || (result == nullptr && maxResultSize > 0) || maxResultSize < 0) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + UnicodeString temp(result, 0, maxResultSize); + ((const LocaleDisplayNames *)ldn)->localeDisplayName(locale, temp); + if (temp.isBogus()) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + return temp.extract(result, maxResultSize, *pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +uldn_languageDisplayName(const ULocaleDisplayNames *ldn, + const char *lang, + char16_t *result, + int32_t maxResultSize, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return 0; + } + if (ldn == nullptr || lang == nullptr || (result == nullptr && maxResultSize > 0) || maxResultSize < 0) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + UnicodeString temp(result, 0, maxResultSize); + ((const LocaleDisplayNames *)ldn)->languageDisplayName(lang, temp); + return temp.extract(result, maxResultSize, *pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +uldn_scriptDisplayName(const ULocaleDisplayNames *ldn, + const char *script, + char16_t *result, + int32_t maxResultSize, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return 0; + } + if (ldn == nullptr || script == nullptr || (result == nullptr && maxResultSize > 0) || maxResultSize < 0) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + UnicodeString temp(result, 0, maxResultSize); + ((const LocaleDisplayNames *)ldn)->scriptDisplayName(script, temp); + return temp.extract(result, maxResultSize, *pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +uldn_scriptCodeDisplayName(const ULocaleDisplayNames *ldn, + UScriptCode scriptCode, + char16_t *result, + int32_t maxResultSize, + UErrorCode *pErrorCode) { + return uldn_scriptDisplayName(ldn, uscript_getName(scriptCode), result, maxResultSize, pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +uldn_regionDisplayName(const ULocaleDisplayNames *ldn, + const char *region, + char16_t *result, + int32_t maxResultSize, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return 0; + } + if (ldn == nullptr || region == nullptr || (result == nullptr && maxResultSize > 0) || maxResultSize < 0) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + UnicodeString temp(result, 0, maxResultSize); + ((const LocaleDisplayNames *)ldn)->regionDisplayName(region, temp); + return temp.extract(result, maxResultSize, *pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +uldn_variantDisplayName(const ULocaleDisplayNames *ldn, + const char *variant, + char16_t *result, + int32_t maxResultSize, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return 0; + } + if (ldn == nullptr || variant == nullptr || (result == nullptr && maxResultSize > 0) || maxResultSize < 0) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + UnicodeString temp(result, 0, maxResultSize); + ((const LocaleDisplayNames *)ldn)->variantDisplayName(variant, temp); + return temp.extract(result, maxResultSize, *pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +uldn_keyDisplayName(const ULocaleDisplayNames *ldn, + const char *key, + char16_t *result, + int32_t maxResultSize, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return 0; + } + if (ldn == nullptr || key == nullptr || (result == nullptr && maxResultSize > 0) || maxResultSize < 0) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + UnicodeString temp(result, 0, maxResultSize); + ((const LocaleDisplayNames *)ldn)->keyDisplayName(key, temp); + return temp.extract(result, maxResultSize, *pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +uldn_keyValueDisplayName(const ULocaleDisplayNames *ldn, + const char *key, + const char *value, + char16_t *result, + int32_t maxResultSize, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return 0; + } + if (ldn == nullptr || key == nullptr || value == nullptr || (result == nullptr && maxResultSize > 0) + || maxResultSize < 0) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + UnicodeString temp(result, 0, maxResultSize); + ((const LocaleDisplayNames *)ldn)->keyValueDisplayName(key, value, temp); + return temp.extract(result, maxResultSize, *pErrorCode); +} + +#endif diff --git a/intl/icu/source/common/locid.cpp b/intl/icu/source/common/locid.cpp new file mode 100644 index 0000000000..70a794ae07 --- /dev/null +++ b/intl/icu/source/common/locid.cpp @@ -0,0 +1,2734 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ********************************************************************** + * Copyright (C) 1997-2016, International Business Machines + * Corporation and others. All Rights Reserved. + ********************************************************************** +* +* File locid.cpp +* +* Created by: Richard Gillam +* +* Modification History: +* +* Date Name Description +* 02/11/97 aliu Changed gLocPath to fgDataDirectory and added +* methods to get and set it. +* 04/02/97 aliu Made operator!= inline; fixed return value +* of getName(). +* 04/15/97 aliu Cleanup for AIX/Win32. +* 04/24/97 aliu Numerous changes per code review. +* 08/18/98 stephen Changed getDisplayName() +* Added SIMPLIFIED_CHINESE, TRADITIONAL_CHINESE +* Added getISOCountries(), getISOLanguages(), +* getLanguagesForCountry() +* 03/16/99 bertrand rehaul. +* 07/21/99 stephen Added U_CFUNC setDefault +* 11/09/99 weiv Added const char * getName() const; +* 04/12/00 srl removing unicodestring api's and cached hash code +* 08/10/01 grhoten Change the static Locales to accessor functions +****************************************************************************** +*/ + +#include + +#include "unicode/bytestream.h" +#include "unicode/locid.h" +#include "unicode/localebuilder.h" +#include "unicode/strenum.h" +#include "unicode/stringpiece.h" +#include "unicode/uloc.h" +#include "unicode/ures.h" + +#include "bytesinkutil.h" +#include "charstr.h" +#include "charstrmap.h" +#include "cmemory.h" +#include "cstring.h" +#include "mutex.h" +#include "putilimp.h" +#include "uassert.h" +#include "ucln_cmn.h" +#include "uhash.h" +#include "ulocimp.h" +#include "umutex.h" +#include "uniquecharstr.h" +#include "ustr_imp.h" +#include "uvector.h" + +U_CDECL_BEGIN +static UBool U_CALLCONV locale_cleanup(); +U_CDECL_END + +U_NAMESPACE_BEGIN + +static Locale *gLocaleCache = nullptr; +static UInitOnce gLocaleCacheInitOnce {}; + +// gDefaultLocaleMutex protects all access to gDefaultLocalesHashT and gDefaultLocale. +static UMutex gDefaultLocaleMutex; +static UHashtable *gDefaultLocalesHashT = nullptr; +static Locale *gDefaultLocale = nullptr; + +/** + * \def ULOC_STRING_LIMIT + * strings beyond this value crash in CharString + */ +#define ULOC_STRING_LIMIT 357913941 + +U_NAMESPACE_END + +typedef enum ELocalePos { + eENGLISH, + eFRENCH, + eGERMAN, + eITALIAN, + eJAPANESE, + eKOREAN, + eCHINESE, + + eFRANCE, + eGERMANY, + eITALY, + eJAPAN, + eKOREA, + eCHINA, /* Alias for PRC */ + eTAIWAN, + eUK, + eUS, + eCANADA, + eCANADA_FRENCH, + eROOT, + + + //eDEFAULT, + eMAX_LOCALES +} ELocalePos; + +U_CDECL_BEGIN +// +// Deleter function for Locales owned by the default Locale hash table/ +// +static void U_CALLCONV +deleteLocale(void *obj) { + delete (icu::Locale *) obj; +} + +static UBool U_CALLCONV locale_cleanup() +{ + U_NAMESPACE_USE + + delete [] gLocaleCache; + gLocaleCache = nullptr; + gLocaleCacheInitOnce.reset(); + + if (gDefaultLocalesHashT) { + uhash_close(gDefaultLocalesHashT); // Automatically deletes all elements, using deleter func. + gDefaultLocalesHashT = nullptr; + } + gDefaultLocale = nullptr; + return true; +} + + +static void U_CALLCONV locale_init(UErrorCode &status) { + U_NAMESPACE_USE + + U_ASSERT(gLocaleCache == nullptr); + gLocaleCache = new Locale[(int)eMAX_LOCALES]; + if (gLocaleCache == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + ucln_common_registerCleanup(UCLN_COMMON_LOCALE, locale_cleanup); + gLocaleCache[eROOT] = Locale(""); + gLocaleCache[eENGLISH] = Locale("en"); + gLocaleCache[eFRENCH] = Locale("fr"); + gLocaleCache[eGERMAN] = Locale("de"); + gLocaleCache[eITALIAN] = Locale("it"); + gLocaleCache[eJAPANESE] = Locale("ja"); + gLocaleCache[eKOREAN] = Locale("ko"); + gLocaleCache[eCHINESE] = Locale("zh"); + gLocaleCache[eFRANCE] = Locale("fr", "FR"); + gLocaleCache[eGERMANY] = Locale("de", "DE"); + gLocaleCache[eITALY] = Locale("it", "IT"); + gLocaleCache[eJAPAN] = Locale("ja", "JP"); + gLocaleCache[eKOREA] = Locale("ko", "KR"); + gLocaleCache[eCHINA] = Locale("zh", "CN"); + gLocaleCache[eTAIWAN] = Locale("zh", "TW"); + gLocaleCache[eUK] = Locale("en", "GB"); + gLocaleCache[eUS] = Locale("en", "US"); + gLocaleCache[eCANADA] = Locale("en", "CA"); + gLocaleCache[eCANADA_FRENCH] = Locale("fr", "CA"); +} + +U_CDECL_END + +U_NAMESPACE_BEGIN + +Locale *locale_set_default_internal(const char *id, UErrorCode& status) { + // Synchronize this entire function. + Mutex lock(&gDefaultLocaleMutex); + + UBool canonicalize = false; + + // If given a nullptr string for the locale id, grab the default + // name from the system. + // (Different from most other locale APIs, where a null name means use + // the current ICU default locale.) + if (id == nullptr) { + id = uprv_getDefaultLocaleID(); // This function not thread safe? TODO: verify. + canonicalize = true; // always canonicalize host ID + } + + CharString localeNameBuf; + { + CharStringByteSink sink(&localeNameBuf); + if (canonicalize) { + ulocimp_canonicalize(id, sink, &status); + } else { + ulocimp_getName(id, sink, &status); + } + } + + if (U_FAILURE(status)) { + return gDefaultLocale; + } + + if (gDefaultLocalesHashT == nullptr) { + gDefaultLocalesHashT = uhash_open(uhash_hashChars, uhash_compareChars, nullptr, &status); + if (U_FAILURE(status)) { + return gDefaultLocale; + } + uhash_setValueDeleter(gDefaultLocalesHashT, deleteLocale); + ucln_common_registerCleanup(UCLN_COMMON_LOCALE, locale_cleanup); + } + + Locale *newDefault = (Locale *)uhash_get(gDefaultLocalesHashT, localeNameBuf.data()); + if (newDefault == nullptr) { + newDefault = new Locale(Locale::eBOGUS); + if (newDefault == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return gDefaultLocale; + } + newDefault->init(localeNameBuf.data(), false); + uhash_put(gDefaultLocalesHashT, (char*) newDefault->getName(), newDefault, &status); + if (U_FAILURE(status)) { + return gDefaultLocale; + } + } + gDefaultLocale = newDefault; + return gDefaultLocale; +} + +U_NAMESPACE_END + +/* sfb 07/21/99 */ +U_CFUNC void +locale_set_default(const char *id) +{ + U_NAMESPACE_USE + UErrorCode status = U_ZERO_ERROR; + locale_set_default_internal(id, status); +} +/* end */ + +U_CFUNC const char * +locale_get_default() +{ + U_NAMESPACE_USE + return Locale::getDefault().getName(); +} + + +U_NAMESPACE_BEGIN + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(Locale) + +/*Character separating the posix id fields*/ +// '_' +// In the platform codepage. +#define SEP_CHAR '_' +#define NULL_CHAR '\0' + +Locale::~Locale() +{ + if ((baseName != fullName) && (baseName != fullNameBuffer)) { + uprv_free(baseName); + } + baseName = nullptr; + /*if fullName is on the heap, we free it*/ + if (fullName != fullNameBuffer) + { + uprv_free(fullName); + fullName = nullptr; + } +} + +Locale::Locale() + : UObject(), fullName(fullNameBuffer), baseName(nullptr) +{ + init(nullptr, false); +} + +/* + * Internal constructor to allow construction of a locale object with + * NO side effects. (Default constructor tries to get + * the default locale.) + */ +Locale::Locale(Locale::ELocaleType) + : UObject(), fullName(fullNameBuffer), baseName(nullptr) +{ + setToBogus(); +} + + +Locale::Locale( const char * newLanguage, + const char * newCountry, + const char * newVariant, + const char * newKeywords) + : UObject(), fullName(fullNameBuffer), baseName(nullptr) +{ + if( (newLanguage==nullptr) && (newCountry == nullptr) && (newVariant == nullptr) ) + { + init(nullptr, false); /* shortcut */ + } + else + { + UErrorCode status = U_ZERO_ERROR; + int32_t lsize = 0; + int32_t csize = 0; + int32_t vsize = 0; + int32_t ksize = 0; + + // Check the sizes of the input strings. + + // Language + if ( newLanguage != nullptr ) + { + lsize = (int32_t)uprv_strlen(newLanguage); + if ( lsize < 0 || lsize > ULOC_STRING_LIMIT ) { // int32 wrap + setToBogus(); + return; + } + } + + CharString togo(newLanguage, lsize, status); // start with newLanguage + + // _Country + if ( newCountry != nullptr ) + { + csize = (int32_t)uprv_strlen(newCountry); + if ( csize < 0 || csize > ULOC_STRING_LIMIT ) { // int32 wrap + setToBogus(); + return; + } + } + + // _Variant + if ( newVariant != nullptr ) + { + // remove leading _'s + while(newVariant[0] == SEP_CHAR) + { + newVariant++; + } + + // remove trailing _'s + vsize = (int32_t)uprv_strlen(newVariant); + if ( vsize < 0 || vsize > ULOC_STRING_LIMIT ) { // int32 wrap + setToBogus(); + return; + } + while( (vsize>1) && (newVariant[vsize-1] == SEP_CHAR) ) + { + vsize--; + } + } + + if ( newKeywords != nullptr) + { + ksize = (int32_t)uprv_strlen(newKeywords); + if ( ksize < 0 || ksize > ULOC_STRING_LIMIT ) { + setToBogus(); + return; + } + } + + // We've checked the input sizes, now build up the full locale string.. + + // newLanguage is already copied + + if ( ( vsize != 0 ) || (csize != 0) ) // at least: __v + { // ^ + togo.append(SEP_CHAR, status); + } + + if ( csize != 0 ) + { + togo.append(newCountry, status); + } + + if ( vsize != 0) + { + togo.append(SEP_CHAR, status) + .append(newVariant, vsize, status); + } + + if ( ksize != 0) + { + if (uprv_strchr(newKeywords, '=')) { + togo.append('@', status); /* keyword parsing */ + } + else { + togo.append('_', status); /* Variant parsing with a script */ + if ( vsize == 0) { + togo.append('_', status); /* No country found */ + } + } + togo.append(newKeywords, status); + } + + if (U_FAILURE(status)) { + // Something went wrong with appending, etc. + setToBogus(); + return; + } + // Parse it, because for example 'language' might really be a complete + // string. + init(togo.data(), false); + } +} + +Locale::Locale(const Locale &other) + : UObject(other), fullName(fullNameBuffer), baseName(nullptr) +{ + *this = other; +} + +Locale::Locale(Locale&& other) noexcept + : UObject(other), fullName(fullNameBuffer), baseName(fullName) { + *this = std::move(other); +} + +Locale& Locale::operator=(const Locale& other) { + if (this == &other) { + return *this; + } + + setToBogus(); + + if (other.fullName == other.fullNameBuffer) { + uprv_strcpy(fullNameBuffer, other.fullNameBuffer); + } else if (other.fullName == nullptr) { + fullName = nullptr; + } else { + fullName = uprv_strdup(other.fullName); + if (fullName == nullptr) return *this; + } + + if (other.baseName == other.fullName) { + baseName = fullName; + } else if (other.baseName != nullptr) { + baseName = uprv_strdup(other.baseName); + if (baseName == nullptr) return *this; + } + + uprv_strcpy(language, other.language); + uprv_strcpy(script, other.script); + uprv_strcpy(country, other.country); + + variantBegin = other.variantBegin; + fIsBogus = other.fIsBogus; + + return *this; +} + +Locale& Locale::operator=(Locale&& other) noexcept { + if ((baseName != fullName) && (baseName != fullNameBuffer)) uprv_free(baseName); + if (fullName != fullNameBuffer) uprv_free(fullName); + + if (other.fullName == other.fullNameBuffer || other.baseName == other.fullNameBuffer) { + uprv_strcpy(fullNameBuffer, other.fullNameBuffer); + } + if (other.fullName == other.fullNameBuffer) { + fullName = fullNameBuffer; + } else { + fullName = other.fullName; + } + + if (other.baseName == other.fullNameBuffer) { + baseName = fullNameBuffer; + } else if (other.baseName == other.fullName) { + baseName = fullName; + } else { + baseName = other.baseName; + } + + uprv_strcpy(language, other.language); + uprv_strcpy(script, other.script); + uprv_strcpy(country, other.country); + + variantBegin = other.variantBegin; + fIsBogus = other.fIsBogus; + + other.baseName = other.fullName = other.fullNameBuffer; + + return *this; +} + +Locale * +Locale::clone() const { + return new Locale(*this); +} + +bool +Locale::operator==( const Locale& other) const +{ + return (uprv_strcmp(other.fullName, fullName) == 0); +} + +namespace { + +UInitOnce gKnownCanonicalizedInitOnce {}; +UHashtable *gKnownCanonicalized = nullptr; + +static const char* const KNOWN_CANONICALIZED[] = { + "c", + // Commonly used locales known are already canonicalized + "af", "af_ZA", "am", "am_ET", "ar", "ar_001", "as", "as_IN", "az", "az_AZ", + "be", "be_BY", "bg", "bg_BG", "bn", "bn_IN", "bs", "bs_BA", "ca", "ca_ES", + "cs", "cs_CZ", "cy", "cy_GB", "da", "da_DK", "de", "de_DE", "el", "el_GR", + "en", "en_GB", "en_US", "es", "es_419", "es_ES", "et", "et_EE", "eu", + "eu_ES", "fa", "fa_IR", "fi", "fi_FI", "fil", "fil_PH", "fr", "fr_FR", + "ga", "ga_IE", "gl", "gl_ES", "gu", "gu_IN", "he", "he_IL", "hi", "hi_IN", + "hr", "hr_HR", "hu", "hu_HU", "hy", "hy_AM", "id", "id_ID", "is", "is_IS", + "it", "it_IT", "ja", "ja_JP", "jv", "jv_ID", "ka", "ka_GE", "kk", "kk_KZ", + "km", "km_KH", "kn", "kn_IN", "ko", "ko_KR", "ky", "ky_KG", "lo", "lo_LA", + "lt", "lt_LT", "lv", "lv_LV", "mk", "mk_MK", "ml", "ml_IN", "mn", "mn_MN", + "mr", "mr_IN", "ms", "ms_MY", "my", "my_MM", "nb", "nb_NO", "ne", "ne_NP", + "nl", "nl_NL", "no", "or", "or_IN", "pa", "pa_IN", "pl", "pl_PL", "ps", "ps_AF", + "pt", "pt_BR", "pt_PT", "ro", "ro_RO", "ru", "ru_RU", "sd", "sd_IN", "si", + "si_LK", "sk", "sk_SK", "sl", "sl_SI", "so", "so_SO", "sq", "sq_AL", "sr", + "sr_Cyrl_RS", "sr_Latn", "sr_RS", "sv", "sv_SE", "sw", "sw_TZ", "ta", + "ta_IN", "te", "te_IN", "th", "th_TH", "tk", "tk_TM", "tr", "tr_TR", "uk", + "uk_UA", "ur", "ur_PK", "uz", "uz_UZ", "vi", "vi_VN", "yue", "yue_Hant", + "yue_Hant_HK", "yue_HK", "zh", "zh_CN", "zh_Hans", "zh_Hans_CN", "zh_Hant", + "zh_Hant_TW", "zh_TW", "zu", "zu_ZA" +}; + +static UBool U_CALLCONV cleanupKnownCanonicalized() { + gKnownCanonicalizedInitOnce.reset(); + if (gKnownCanonicalized) { uhash_close(gKnownCanonicalized); } + return true; +} + +static void U_CALLCONV loadKnownCanonicalized(UErrorCode &status) { + ucln_common_registerCleanup(UCLN_COMMON_LOCALE_KNOWN_CANONICALIZED, + cleanupKnownCanonicalized); + LocalUHashtablePointer newKnownCanonicalizedMap( + uhash_open(uhash_hashChars, uhash_compareChars, nullptr, &status)); + for (int32_t i = 0; + U_SUCCESS(status) && i < UPRV_LENGTHOF(KNOWN_CANONICALIZED); + i++) { + uhash_puti(newKnownCanonicalizedMap.getAlias(), + (void*)KNOWN_CANONICALIZED[i], + 1, &status); + } + if (U_FAILURE(status)) { + return; + } + + gKnownCanonicalized = newKnownCanonicalizedMap.orphan(); +} + +class AliasData; + +/** + * A Builder class to build the alias data. + */ +class AliasDataBuilder { +public: + AliasDataBuilder() { + } + + // Build the AliasData from resource. + AliasData* build(UErrorCode &status); + +private: + void readAlias(UResourceBundle* alias, + UniqueCharStrings* strings, + LocalMemory& types, + LocalMemory& replacementIndexes, + int32_t &length, + void (*checkType)(const char* type), + void (*checkReplacement)(const UnicodeString& replacement), + UErrorCode &status); + + // Read the languageAlias data from alias to + // strings+types+replacementIndexes + // The number of record will be stored into length. + // Allocate length items for types, to store the type field. + // Allocate length items for replacementIndexes, + // to store the index in the strings for the replacement script. + void readLanguageAlias(UResourceBundle* alias, + UniqueCharStrings* strings, + LocalMemory& types, + LocalMemory& replacementIndexes, + int32_t &length, + UErrorCode &status); + + // Read the scriptAlias data from alias to + // strings+types+replacementIndexes + // Allocate length items for types, to store the type field. + // Allocate length items for replacementIndexes, + // to store the index in the strings for the replacement script. + void readScriptAlias(UResourceBundle* alias, + UniqueCharStrings* strings, + LocalMemory& types, + LocalMemory& replacementIndexes, + int32_t &length, UErrorCode &status); + + // Read the territoryAlias data from alias to + // strings+types+replacementIndexes + // Allocate length items for types, to store the type field. + // Allocate length items for replacementIndexes, + // to store the index in the strings for the replacement script. + void readTerritoryAlias(UResourceBundle* alias, + UniqueCharStrings* strings, + LocalMemory& types, + LocalMemory& replacementIndexes, + int32_t &length, UErrorCode &status); + + // Read the variantAlias data from alias to + // strings+types+replacementIndexes + // Allocate length items for types, to store the type field. + // Allocate length items for replacementIndexes, + // to store the index in the strings for the replacement variant. + void readVariantAlias(UResourceBundle* alias, + UniqueCharStrings* strings, + LocalMemory& types, + LocalMemory& replacementIndexes, + int32_t &length, UErrorCode &status); + + // Read the subdivisionAlias data from alias to + // strings+types+replacementIndexes + // Allocate length items for types, to store the type field. + // Allocate length items for replacementIndexes, + // to store the index in the strings for the replacement variant. + void readSubdivisionAlias(UResourceBundle* alias, + UniqueCharStrings* strings, + LocalMemory& types, + LocalMemory& replacementIndexes, + int32_t &length, UErrorCode &status); +}; + +/** + * A class to hold the Alias Data. + */ +class AliasData : public UMemory { +public: + static const AliasData* singleton(UErrorCode& status) { + if (U_FAILURE(status)) { + // Do not get into loadData if the status already has error. + return nullptr; + } + umtx_initOnce(AliasData::gInitOnce, &AliasData::loadData, status); + return gSingleton; + } + + const CharStringMap& languageMap() const { return language; } + const CharStringMap& scriptMap() const { return script; } + const CharStringMap& territoryMap() const { return territory; } + const CharStringMap& variantMap() const { return variant; } + const CharStringMap& subdivisionMap() const { return subdivision; } + + static void U_CALLCONV loadData(UErrorCode &status); + static UBool U_CALLCONV cleanup(); + + static UInitOnce gInitOnce; + +private: + AliasData(CharStringMap languageMap, + CharStringMap scriptMap, + CharStringMap territoryMap, + CharStringMap variantMap, + CharStringMap subdivisionMap, + CharString* strings) + : language(std::move(languageMap)), + script(std::move(scriptMap)), + territory(std::move(territoryMap)), + variant(std::move(variantMap)), + subdivision(std::move(subdivisionMap)), + strings(strings) { + } + + ~AliasData() { + delete strings; + } + + static const AliasData* gSingleton; + + CharStringMap language; + CharStringMap script; + CharStringMap territory; + CharStringMap variant; + CharStringMap subdivision; + CharString* strings; + + friend class AliasDataBuilder; +}; + + +const AliasData* AliasData::gSingleton = nullptr; +UInitOnce AliasData::gInitOnce {}; + +UBool U_CALLCONV +AliasData::cleanup() +{ + gInitOnce.reset(); + delete gSingleton; + return true; +} + +void +AliasDataBuilder::readAlias( + UResourceBundle* alias, + UniqueCharStrings* strings, + LocalMemory& types, + LocalMemory& replacementIndexes, + int32_t &length, + void (*checkType)(const char* type), + void (*checkReplacement)(const UnicodeString& replacement), + UErrorCode &status) { + if (U_FAILURE(status)) { + return; + } + length = ures_getSize(alias); + const char** rawTypes = types.allocateInsteadAndCopy(length); + if (rawTypes == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + int32_t* rawIndexes = replacementIndexes.allocateInsteadAndCopy(length); + if (rawIndexes == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + for (int i = 0; U_SUCCESS(status) && ures_hasNext(alias); i++) { + LocalUResourceBundlePointer res( + ures_getNextResource(alias, nullptr, &status)); + const char* aliasFrom = ures_getKey(res.getAlias()); + UnicodeString aliasTo = + ures_getUnicodeStringByKey(res.getAlias(), "replacement", &status); + if (U_FAILURE(status)) return; + + checkType(aliasFrom); + checkReplacement(aliasTo); + + rawTypes[i] = aliasFrom; + rawIndexes[i] = strings->add(aliasTo, status); + } +} + +/** + * Read the languageAlias data from alias to strings+types+replacementIndexes. + * Allocate length items for types, to store the type field. Allocate length + * items for replacementIndexes, to store the index in the strings for the + * replacement language. + */ +void +AliasDataBuilder::readLanguageAlias( + UResourceBundle* alias, + UniqueCharStrings* strings, + LocalMemory& types, + LocalMemory& replacementIndexes, + int32_t &length, + UErrorCode &status) +{ + return readAlias( + alias, strings, types, replacementIndexes, length, +#if U_DEBUG + [](const char* type) { + // Assert the aliasFrom only contains the following possibilities + // language_REGION_variant + // language_REGION + // language_variant + // language + // und_variant + Locale test(type); + // Assert no script in aliasFrom + U_ASSERT(test.getScript()[0] == '\0'); + // Assert when language is und, no REGION in aliasFrom. + U_ASSERT(test.getLanguage()[0] != '\0' || test.getCountry()[0] == '\0'); + }, +#else + [](const char*) {}, +#endif + [](const UnicodeString&) {}, status); +} + +/** + * Read the scriptAlias data from alias to strings+types+replacementIndexes. + * Allocate length items for types, to store the type field. Allocate length + * items for replacementIndexes, to store the index in the strings for the + * replacement script. + */ +void +AliasDataBuilder::readScriptAlias( + UResourceBundle* alias, + UniqueCharStrings* strings, + LocalMemory& types, + LocalMemory& replacementIndexes, + int32_t &length, + UErrorCode &status) +{ + return readAlias( + alias, strings, types, replacementIndexes, length, +#if U_DEBUG + [](const char* type) { + U_ASSERT(uprv_strlen(type) == 4); + }, + [](const UnicodeString& replacement) { + U_ASSERT(replacement.length() == 4); + }, +#else + [](const char*) {}, + [](const UnicodeString&) { }, +#endif + status); +} + +/** + * Read the territoryAlias data from alias to strings+types+replacementIndexes. + * Allocate length items for types, to store the type field. Allocate length + * items for replacementIndexes, to store the index in the strings for the + * replacement regions. + */ +void +AliasDataBuilder::readTerritoryAlias( + UResourceBundle* alias, + UniqueCharStrings* strings, + LocalMemory& types, + LocalMemory& replacementIndexes, + int32_t &length, + UErrorCode &status) +{ + return readAlias( + alias, strings, types, replacementIndexes, length, +#if U_DEBUG + [](const char* type) { + U_ASSERT(uprv_strlen(type) == 2 || uprv_strlen(type) == 3); + }, +#else + [](const char*) {}, +#endif + [](const UnicodeString&) { }, + status); +} + +/** + * Read the variantAlias data from alias to strings+types+replacementIndexes. + * Allocate length items for types, to store the type field. Allocate length + * items for replacementIndexes, to store the index in the strings for the + * replacement variant. + */ +void +AliasDataBuilder::readVariantAlias( + UResourceBundle* alias, + UniqueCharStrings* strings, + LocalMemory& types, + LocalMemory& replacementIndexes, + int32_t &length, + UErrorCode &status) +{ + return readAlias( + alias, strings, types, replacementIndexes, length, +#if U_DEBUG + [](const char* type) { + U_ASSERT(uprv_strlen(type) >= 4 && uprv_strlen(type) <= 8); + U_ASSERT(uprv_strlen(type) != 4 || + (type[0] >= '0' && type[0] <= '9')); + }, + [](const UnicodeString& replacement) { + U_ASSERT(replacement.length() >= 4 && replacement.length() <= 8); + U_ASSERT(replacement.length() != 4 || + (replacement.charAt(0) >= u'0' && + replacement.charAt(0) <= u'9')); + }, +#else + [](const char*) {}, + [](const UnicodeString&) { }, +#endif + status); +} + +/** + * Read the subdivisionAlias data from alias to strings+types+replacementIndexes. + * Allocate length items for types, to store the type field. Allocate length + * items for replacementIndexes, to store the index in the strings for the + * replacement regions. + */ +void +AliasDataBuilder::readSubdivisionAlias( + UResourceBundle* alias, + UniqueCharStrings* strings, + LocalMemory& types, + LocalMemory& replacementIndexes, + int32_t &length, + UErrorCode &status) +{ + return readAlias( + alias, strings, types, replacementIndexes, length, +#if U_DEBUG + [](const char* type) { + U_ASSERT(uprv_strlen(type) >= 3 && uprv_strlen(type) <= 8); + }, +#else + [](const char*) {}, +#endif + [](const UnicodeString&) { }, + status); +} + +/** + * Initializes the alias data from the ICU resource bundles. The alias data + * contains alias of language, country, script and variants. + * + * If the alias data has already loaded, then this method simply returns without + * doing anything meaningful. + */ +void U_CALLCONV +AliasData::loadData(UErrorCode &status) +{ +#ifdef LOCALE_CANONICALIZATION_DEBUG + UDate start = uprv_getRawUTCtime(); +#endif // LOCALE_CANONICALIZATION_DEBUG + ucln_common_registerCleanup(UCLN_COMMON_LOCALE_ALIAS, cleanup); + AliasDataBuilder builder; + gSingleton = builder.build(status); +#ifdef LOCALE_CANONICALIZATION_DEBUG + UDate end = uprv_getRawUTCtime(); + printf("AliasData::loadData took total %f ms\n", end - start); +#endif // LOCALE_CANONICALIZATION_DEBUG +} + +/** + * Build the alias data from resources. + */ +AliasData* +AliasDataBuilder::build(UErrorCode &status) { + LocalUResourceBundlePointer metadata( + ures_openDirect(nullptr, "metadata", &status)); + LocalUResourceBundlePointer metadataAlias( + ures_getByKey(metadata.getAlias(), "alias", nullptr, &status)); + LocalUResourceBundlePointer languageAlias( + ures_getByKey(metadataAlias.getAlias(), "language", nullptr, &status)); + LocalUResourceBundlePointer scriptAlias( + ures_getByKey(metadataAlias.getAlias(), "script", nullptr, &status)); + LocalUResourceBundlePointer territoryAlias( + ures_getByKey(metadataAlias.getAlias(), "territory", nullptr, &status)); + LocalUResourceBundlePointer variantAlias( + ures_getByKey(metadataAlias.getAlias(), "variant", nullptr, &status)); + LocalUResourceBundlePointer subdivisionAlias( + ures_getByKey(metadataAlias.getAlias(), "subdivision", nullptr, &status)); + + if (U_FAILURE(status)) { + return nullptr; + } + int32_t languagesLength = 0, scriptLength = 0, territoryLength = 0, + variantLength = 0, subdivisionLength = 0; + + // Read the languageAlias into languageTypes, languageReplacementIndexes + // and strings + UniqueCharStrings strings(status); + LocalMemory languageTypes; + LocalMemory languageReplacementIndexes; + readLanguageAlias(languageAlias.getAlias(), + &strings, + languageTypes, + languageReplacementIndexes, + languagesLength, + status); + + // Read the scriptAlias into scriptTypes, scriptReplacementIndexes + // and strings + LocalMemory scriptTypes; + LocalMemory scriptReplacementIndexes; + readScriptAlias(scriptAlias.getAlias(), + &strings, + scriptTypes, + scriptReplacementIndexes, + scriptLength, + status); + + // Read the territoryAlias into territoryTypes, territoryReplacementIndexes + // and strings + LocalMemory territoryTypes; + LocalMemory territoryReplacementIndexes; + readTerritoryAlias(territoryAlias.getAlias(), + &strings, + territoryTypes, + territoryReplacementIndexes, + territoryLength, status); + + // Read the variantAlias into variantTypes, variantReplacementIndexes + // and strings + LocalMemory variantTypes; + LocalMemory variantReplacementIndexes; + readVariantAlias(variantAlias.getAlias(), + &strings, + variantTypes, + variantReplacementIndexes, + variantLength, status); + + // Read the subdivisionAlias into subdivisionTypes, subdivisionReplacementIndexes + // and strings + LocalMemory subdivisionTypes; + LocalMemory subdivisionReplacementIndexes; + readSubdivisionAlias(subdivisionAlias.getAlias(), + &strings, + subdivisionTypes, + subdivisionReplacementIndexes, + subdivisionLength, status); + + if (U_FAILURE(status)) { + return nullptr; + } + + // We can only use strings after freeze it. + strings.freeze(); + + // Build the languageMap from languageTypes & languageReplacementIndexes + CharStringMap languageMap(490, status); + for (int32_t i = 0; U_SUCCESS(status) && i < languagesLength; i++) { + languageMap.put(languageTypes[i], + strings.get(languageReplacementIndexes[i]), + status); + } + + // Build the scriptMap from scriptTypes & scriptReplacementIndexes + CharStringMap scriptMap(1, status); + for (int32_t i = 0; U_SUCCESS(status) && i < scriptLength; i++) { + scriptMap.put(scriptTypes[i], + strings.get(scriptReplacementIndexes[i]), + status); + } + + // Build the territoryMap from territoryTypes & territoryReplacementIndexes + CharStringMap territoryMap(650, status); + for (int32_t i = 0; U_SUCCESS(status) && i < territoryLength; i++) { + territoryMap.put(territoryTypes[i], + strings.get(territoryReplacementIndexes[i]), + status); + } + + // Build the variantMap from variantTypes & variantReplacementIndexes. + CharStringMap variantMap(2, status); + for (int32_t i = 0; U_SUCCESS(status) && i < variantLength; i++) { + variantMap.put(variantTypes[i], + strings.get(variantReplacementIndexes[i]), + status); + } + + // Build the subdivisionMap from subdivisionTypes & subdivisionReplacementIndexes. + CharStringMap subdivisionMap(2, status); + for (int32_t i = 0; U_SUCCESS(status) && i < subdivisionLength; i++) { + subdivisionMap.put(subdivisionTypes[i], + strings.get(subdivisionReplacementIndexes[i]), + status); + } + + if (U_FAILURE(status)) { + return nullptr; + } + + // copy hashtables + auto *data = new AliasData( + std::move(languageMap), + std::move(scriptMap), + std::move(territoryMap), + std::move(variantMap), + std::move(subdivisionMap), + strings.orphanCharStrings()); + + if (data == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + } + return data; +} + +/** + * A class that find the replacement values of locale fields by using AliasData. + */ +class AliasReplacer { +public: + AliasReplacer(UErrorCode status) : + language(nullptr), script(nullptr), region(nullptr), + extensions(nullptr), variants(status), + data(nullptr) { + } + ~AliasReplacer() { + } + + // Check the fields inside locale, if need to replace fields, + // place the the replaced locale ID in out and return true. + // Otherwise return false for no replacement or error. + bool replace( + const Locale& locale, CharString& out, UErrorCode& status); + +private: + const char* language; + const char* script; + const char* region; + const char* extensions; + UVector variants; + + const AliasData* data; + + inline bool notEmpty(const char* str) { + return str && str[0] != NULL_CHAR; + } + + /** + * If replacement is neither null nor empty and input is either null or empty, + * return replacement. + * If replacement is neither null nor empty but input is not empty, return input. + * If replacement is either null or empty and type is either null or empty, + * return input. + * Otherwise return null. + * replacement input type return + * AAA nullptr * AAA + * AAA BBB * BBB + * nullptr || "" CCC nullptr CCC + * nullptr || "" * DDD nullptr + */ + inline const char* deleteOrReplace( + const char* input, const char* type, const char* replacement) { + return notEmpty(replacement) ? + ((input == nullptr) ? replacement : input) : + ((type == nullptr) ? input : nullptr); + } + + inline bool same(const char* a, const char* b) { + if (a == nullptr && b == nullptr) { + return true; + } + if ((a == nullptr && b != nullptr) || + (a != nullptr && b == nullptr)) { + return false; + } + return uprv_strcmp(a, b) == 0; + } + + // Gather fields and generate locale ID into out. + CharString& outputToString(CharString& out, UErrorCode status); + + // Generate the lookup key. + CharString& generateKey(const char* language, const char* region, + const char* variant, CharString& out, + UErrorCode status); + + void parseLanguageReplacement(const char* replacement, + const char*& replaceLanguage, + const char*& replaceScript, + const char*& replaceRegion, + const char*& replaceVariant, + const char*& replaceExtensions, + UVector& toBeFreed, + UErrorCode& status); + + // Replace by using languageAlias. + bool replaceLanguage(bool checkLanguage, bool checkRegion, + bool checkVariants, UVector& toBeFreed, + UErrorCode& status); + + // Replace by using territoryAlias. + bool replaceTerritory(UVector& toBeFreed, UErrorCode& status); + + // Replace by using scriptAlias. + bool replaceScript(UErrorCode& status); + + // Replace by using variantAlias. + bool replaceVariant(UErrorCode& status); + + // Replace by using subdivisionAlias. + bool replaceSubdivision(StringPiece subdivision, + CharString& output, UErrorCode& status); + + // Replace transformed extensions. + bool replaceTransformedExtensions( + CharString& transformedExtensions, CharString& output, UErrorCode& status); +}; + +CharString& +AliasReplacer::generateKey( + const char* language, const char* region, const char* variant, + CharString& out, UErrorCode status) +{ + out.append(language, status); + if (notEmpty(region)) { + out.append(SEP_CHAR, status) + .append(region, status); + } + if (notEmpty(variant)) { + out.append(SEP_CHAR, status) + .append(variant, status); + } + return out; +} + +void +AliasReplacer::parseLanguageReplacement( + const char* replacement, + const char*& replacedLanguage, + const char*& replacedScript, + const char*& replacedRegion, + const char*& replacedVariant, + const char*& replacedExtensions, + UVector& toBeFreed, + UErrorCode& status) +{ + if (U_FAILURE(status)) { + return; + } + replacedScript = replacedRegion = replacedVariant + = replacedExtensions = nullptr; + if (uprv_strchr(replacement, '_') == nullptr) { + replacedLanguage = replacement; + // reach the end, just return it. + return; + } + // We have multiple field so we have to allocate and parse + CharString* str = new CharString( + replacement, (int32_t)uprv_strlen(replacement), status); + LocalPointer lpStr(str, status); + toBeFreed.adoptElement(lpStr.orphan(), status); + if (U_FAILURE(status)) { + return; + } + char* data = str->data(); + replacedLanguage = (const char*) data; + char* endOfField = uprv_strchr(data, '_'); + *endOfField = '\0'; // null terminiate it. + endOfField++; + const char* start = endOfField; + endOfField = (char*) uprv_strchr(start, '_'); + size_t len = 0; + if (endOfField == nullptr) { + len = uprv_strlen(start); + } else { + len = endOfField - start; + *endOfField = '\0'; // null terminiate it. + } + if (len == 4 && uprv_isASCIILetter(*start)) { + // Got a script + replacedScript = start; + if (endOfField == nullptr) { + return; + } + start = endOfField++; + endOfField = (char*)uprv_strchr(start, '_'); + if (endOfField == nullptr) { + len = uprv_strlen(start); + } else { + len = endOfField - start; + *endOfField = '\0'; // null terminiate it. + } + } + if (len >= 2 && len <= 3) { + // Got a region + replacedRegion = start; + if (endOfField == nullptr) { + return; + } + start = endOfField++; + endOfField = (char*)uprv_strchr(start, '_'); + if (endOfField == nullptr) { + len = uprv_strlen(start); + } else { + len = endOfField - start; + *endOfField = '\0'; // null terminiate it. + } + } + if (len >= 4) { + // Got a variant + replacedVariant = start; + if (endOfField == nullptr) { + return; + } + start = endOfField++; + } + replacedExtensions = start; +} + +bool +AliasReplacer::replaceLanguage( + bool checkLanguage, bool checkRegion, + bool checkVariants, UVector& toBeFreed, UErrorCode& status) +{ + if (U_FAILURE(status)) { + return false; + } + if ( (checkRegion && region == nullptr) || + (checkVariants && variants.size() == 0)) { + // Nothing to search. + return false; + } + int32_t variant_size = checkVariants ? variants.size() : 1; + // Since we may have more than one variant, we need to loop through them. + const char* searchLanguage = checkLanguage ? language : "und"; + const char* searchRegion = checkRegion ? region : nullptr; + const char* searchVariant = nullptr; + for (int32_t variant_index = 0; + variant_index < variant_size; + variant_index++) { + if (checkVariants) { + U_ASSERT(variant_index < variant_size); + searchVariant = (const char*)(variants.elementAt(variant_index)); + } + + if (searchVariant != nullptr && uprv_strlen(searchVariant) < 4) { + // Do not consider ill-formed variant subtag. + searchVariant = nullptr; + } + CharString typeKey; + generateKey(searchLanguage, searchRegion, searchVariant, typeKey, + status); + if (U_FAILURE(status)) { + return false; + } + const char *replacement = data->languageMap().get(typeKey.data()); + if (replacement == nullptr) { + // Found no replacement data. + continue; + } + + const char* replacedLanguage = nullptr; + const char* replacedScript = nullptr; + const char* replacedRegion = nullptr; + const char* replacedVariant = nullptr; + const char* replacedExtensions = nullptr; + parseLanguageReplacement(replacement, + replacedLanguage, + replacedScript, + replacedRegion, + replacedVariant, + replacedExtensions, + toBeFreed, + status); + replacedLanguage = + (replacedLanguage != nullptr && uprv_strcmp(replacedLanguage, "und") == 0) ? + language : replacedLanguage; + replacedScript = deleteOrReplace(script, nullptr, replacedScript); + replacedRegion = deleteOrReplace(region, searchRegion, replacedRegion); + replacedVariant = deleteOrReplace( + searchVariant, searchVariant, replacedVariant); + + if ( same(language, replacedLanguage) && + same(script, replacedScript) && + same(region, replacedRegion) && + same(searchVariant, replacedVariant) && + replacedExtensions == nullptr) { + // Replacement produce no changes. + continue; + } + + language = replacedLanguage; + region = replacedRegion; + script = replacedScript; + if (searchVariant != nullptr) { + if (notEmpty(replacedVariant)) { + variants.setElementAt((void*)replacedVariant, variant_index); + } else { + variants.removeElementAt(variant_index); + } + } + if (replacedExtensions != nullptr) { + // DO NOTHING + // UTS35 does not specify what should we do if we have extensions in the + // replacement. Currently we know only the following 4 "BCP47 LegacyRules" have + // extensions in them languageAlias: + // i_default => en_x_i_default + // i_enochian => und_x_i_enochian + // i_mingo => see_x_i_mingo + // zh_min => nan_x_zh_min + // But all of them are already changed by code inside ultag_parse() before + // hitting this code. + } + + // Something changed by language alias data. + return true; + } + // Nothing changed by language alias data. + return false; +} + +bool +AliasReplacer::replaceTerritory(UVector& toBeFreed, UErrorCode& status) +{ + if (U_FAILURE(status)) { + return false; + } + if (region == nullptr) { + // No region to search. + return false; + } + const char *replacement = data->territoryMap().get(region); + if (replacement == nullptr) { + // Found no replacement data for this region. + return false; + } + const char* replacedRegion = replacement; + const char* firstSpace = uprv_strchr(replacement, ' '); + if (firstSpace != nullptr) { + // If there are are more than one region in the replacement. + // We need to check which one match based on the language. + // Cannot use nullptr for language because that will construct + // the default locale, in that case, use "und" to get the correct + // locale. + Locale l = LocaleBuilder() + .setLanguage(language == nullptr ? "und" : language) + .setScript(script) + .build(status); + l.addLikelySubtags(status); + const char* likelyRegion = l.getCountry(); + LocalPointer item; + if (likelyRegion != nullptr && uprv_strlen(likelyRegion) > 0) { + size_t len = uprv_strlen(likelyRegion); + const char* foundInReplacement = uprv_strstr(replacement, + likelyRegion); + if (foundInReplacement != nullptr) { + // Assuming the case there are no three letter region code in + // the replacement of territoryAlias + U_ASSERT(foundInReplacement == replacement || + *(foundInReplacement-1) == ' '); + U_ASSERT(foundInReplacement[len] == ' ' || + foundInReplacement[len] == '\0'); + item.adoptInsteadAndCheckErrorCode( + new CharString(foundInReplacement, (int32_t)len, status), status); + } + } + if (item.isNull() && U_SUCCESS(status)) { + item.adoptInsteadAndCheckErrorCode( + new CharString(replacement, + (int32_t)(firstSpace - replacement), status), status); + } + if (U_FAILURE(status)) { return false; } + replacedRegion = item->data(); + toBeFreed.adoptElement(item.orphan(), status); + if (U_FAILURE(status)) { return false; } + } + U_ASSERT(!same(region, replacedRegion)); + region = replacedRegion; + // The region is changed by data in territory alias. + return true; +} + +bool +AliasReplacer::replaceScript(UErrorCode& status) +{ + if (U_FAILURE(status)) { + return false; + } + if (script == nullptr) { + // No script to search. + return false; + } + const char *replacement = data->scriptMap().get(script); + if (replacement == nullptr) { + // Found no replacement data for this script. + return false; + } + U_ASSERT(!same(script, replacement)); + script = replacement; + // The script is changed by data in script alias. + return true; +} + +bool +AliasReplacer::replaceVariant(UErrorCode& status) +{ + if (U_FAILURE(status)) { + return false; + } + // Since we may have more than one variant, we need to loop through them. + for (int32_t i = 0; i < variants.size(); i++) { + const char *variant = (const char*)(variants.elementAt(i)); + const char *replacement = data->variantMap().get(variant); + if (replacement == nullptr) { + // Found no replacement data for this variant. + continue; + } + U_ASSERT((uprv_strlen(replacement) >= 5 && + uprv_strlen(replacement) <= 8) || + (uprv_strlen(replacement) == 4 && + replacement[0] >= '0' && + replacement[0] <= '9')); + if (!same(variant, replacement)) { + variants.setElementAt((void*)replacement, i); + // Special hack to handle hepburn-heploc => alalc97 + if (uprv_strcmp(variant, "heploc") == 0) { + for (int32_t j = 0; j < variants.size(); j++) { + if (uprv_strcmp((const char*)(variants.elementAt(j)), + "hepburn") == 0) { + variants.removeElementAt(j); + } + } + } + return true; + } + } + return false; +} + +bool +AliasReplacer::replaceSubdivision( + StringPiece subdivision, CharString& output, UErrorCode& status) +{ + if (U_FAILURE(status)) { + return false; + } + const char *replacement = data->subdivisionMap().get(subdivision.data()); + if (replacement != nullptr) { + const char* firstSpace = uprv_strchr(replacement, ' '); + // Found replacement data for this subdivision. + size_t len = (firstSpace != nullptr) ? + (firstSpace - replacement) : uprv_strlen(replacement); + if (2 <= len && len <= 8) { + output.append(replacement, (int32_t)len, status); + if (2 == len) { + // Add 'zzzz' based on changes to UTS #35 for CLDR-14312. + output.append("zzzz", 4, status); + } + } + return true; + } + return false; +} + +bool +AliasReplacer::replaceTransformedExtensions( + CharString& transformedExtensions, CharString& output, UErrorCode& status) +{ + // The content of the transformedExtensions will be modified in this + // function to NUL-terminating (tkey-tvalue) pairs. + if (U_FAILURE(status)) { + return false; + } + int32_t len = transformedExtensions.length(); + const char* str = transformedExtensions.data(); + const char* tkey = ultag_getTKeyStart(str); + int32_t tlangLen = (tkey == str) ? 0 : + ((tkey == nullptr) ? len : static_cast((tkey - str - 1))); + CharStringByteSink sink(&output); + if (tlangLen > 0) { + Locale tlang = LocaleBuilder() + .setLanguageTag(StringPiece(str, tlangLen)) + .build(status); + tlang.canonicalize(status); + tlang.toLanguageTag(sink, status); + if (U_FAILURE(status)) { + return false; + } + T_CString_toLowerCase(output.data()); + } + if (tkey != nullptr) { + // We need to sort the tfields by tkey + UVector tfields(status); + if (U_FAILURE(status)) { + return false; + } + do { + const char* tvalue = uprv_strchr(tkey, '-'); + if (tvalue == nullptr) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return false; + } + const char* nextTKey = ultag_getTKeyStart(tvalue); + if (nextTKey != nullptr) { + *((char*)(nextTKey-1)) = '\0'; // NUL terminate tvalue + } + tfields.insertElementAt((void*)tkey, tfields.size(), status); + if (U_FAILURE(status)) { + return false; + } + tkey = nextTKey; + } while (tkey != nullptr); + tfields.sort([](UElement e1, UElement e2) -> int32_t { + return uprv_strcmp((const char*)e1.pointer, (const char*)e2.pointer); + }, status); + for (int32_t i = 0; i < tfields.size(); i++) { + if (output.length() > 0) { + output.append('-', status); + } + const char* tfield = (const char*) tfields.elementAt(i); + const char* tvalue = uprv_strchr(tfield, '-'); + if (tvalue == nullptr) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return false; + } + // Split the "tkey-tvalue" pair string so that we can canonicalize the tvalue. + *((char*)tvalue++) = '\0'; // NUL terminate tkey + output.append(tfield, status).append('-', status); + const char* bcpTValue = ulocimp_toBcpType(tfield, tvalue, nullptr, nullptr); + output.append((bcpTValue == nullptr) ? tvalue : bcpTValue, status); + } + } + if (U_FAILURE(status)) { + return false; + } + return true; +} + +CharString& +AliasReplacer::outputToString( + CharString& out, UErrorCode status) +{ + out.append(language, status); + if (notEmpty(script)) { + out.append(SEP_CHAR, status) + .append(script, status); + } + if (notEmpty(region)) { + out.append(SEP_CHAR, status) + .append(region, status); + } + if (variants.size() > 0) { + if (!notEmpty(script) && !notEmpty(region)) { + out.append(SEP_CHAR, status); + } + variants.sort([](UElement e1, UElement e2) -> int32_t { + return uprv_strcmp((const char*)e1.pointer, (const char*)e2.pointer); + }, status); + int32_t variantsStart = out.length(); + for (int32_t i = 0; i < variants.size(); i++) { + out.append(SEP_CHAR, status) + .append((const char*)(variants.elementAt(i)), + status); + } + T_CString_toUpperCase(out.data() + variantsStart); + } + if (notEmpty(extensions)) { + CharString tmp("und_", status); + tmp.append(extensions, status); + Locale tmpLocale(tmp.data()); + // only support x extension inside CLDR for now. + U_ASSERT(extensions[0] == 'x'); + out.append(tmpLocale.getName() + 1, status); + } + return out; +} + +bool +AliasReplacer::replace(const Locale& locale, CharString& out, UErrorCode& status) +{ + data = AliasData::singleton(status); + if (U_FAILURE(status)) { + return false; + } + U_ASSERT(data != nullptr); + out.clear(); + language = locale.getLanguage(); + if (!notEmpty(language)) { + language = nullptr; + } + script = locale.getScript(); + if (!notEmpty(script)) { + script = nullptr; + } + region = locale.getCountry(); + if (!notEmpty(region)) { + region = nullptr; + } + const char* variantsStr = locale.getVariant(); + CharString variantsBuff(variantsStr, -1, status); + if (!variantsBuff.isEmpty()) { + if (U_FAILURE(status)) { return false; } + char* start = variantsBuff.data(); + T_CString_toLowerCase(start); + char* end; + while ((end = uprv_strchr(start, SEP_CHAR)) != nullptr && + U_SUCCESS(status)) { + *end = NULL_CHAR; // null terminate inside variantsBuff + variants.addElement(start, status); + start = end + 1; + } + variants.addElement(start, status); + } + if (U_FAILURE(status)) { return false; } + + // Sort the variants + variants.sort([](UElement e1, UElement e2) -> int32_t { + return uprv_strcmp((const char*)e1.pointer, (const char*)e2.pointer); + }, status); + + // A changed count to assert when loop too many times. + int changed = 0; + // A UVector to to hold CharString allocated by the replace* method + // and freed when out of scope from his function. + UVector stringsToBeFreed([](void *obj){ delete ((CharString*) obj); }, + nullptr, 10, status); + while (U_SUCCESS(status)) { + // Something wrong with the data cause looping here more than 10 times + // already. + U_ASSERT(changed < 5); + // From observation of key in data/misc/metadata.txt + // we know currently we only need to search in the following combination + // of fields for type in languageAlias: + // * lang_region_variant + // * lang_region + // * lang_variant + // * lang + // * und_variant + // This assumption is ensured by the U_ASSERT in readLanguageAlias + // + // lang REGION variant + if ( replaceLanguage(true, true, true, stringsToBeFreed, status) || + replaceLanguage(true, true, false, stringsToBeFreed, status) || + replaceLanguage(true, false, true, stringsToBeFreed, status) || + replaceLanguage(true, false, false, stringsToBeFreed, status) || + replaceLanguage(false,false, true, stringsToBeFreed, status) || + replaceTerritory(stringsToBeFreed, status) || + replaceScript(status) || + replaceVariant(status)) { + // Some values in data is changed, try to match from the beginning + // again. + changed++; + continue; + } + // Nothing changed. Break out. + break; + } // while(1) + + if (U_FAILURE(status)) { return false; } + // Nothing changed and we know the order of the variants are not change + // because we have no variant or only one. + const char* extensionsStr = locale_getKeywordsStart(locale.getName()); + if (changed == 0 && variants.size() <= 1 && extensionsStr == nullptr) { + return false; + } + outputToString(out, status); + if (U_FAILURE(status)) { + return false; + } + if (extensionsStr != nullptr) { + changed = 0; + Locale temp(locale); + LocalPointer iter(locale.createKeywords(status)); + if (U_SUCCESS(status) && !iter.isNull()) { + const char* key; + while ((key = iter->next(nullptr, status)) != nullptr) { + if (uprv_strcmp("sd", key) == 0 || uprv_strcmp("rg", key) == 0 || + uprv_strcmp("t", key) == 0) { + CharString value; + CharStringByteSink valueSink(&value); + locale.getKeywordValue(key, valueSink, status); + if (U_FAILURE(status)) { + status = U_ZERO_ERROR; + continue; + } + CharString replacement; + if (uprv_strlen(key) == 2) { + if (replaceSubdivision(value.toStringPiece(), replacement, status)) { + changed++; + temp.setKeywordValue(key, replacement.data(), status); + } + } else { + U_ASSERT(uprv_strcmp(key, "t") == 0); + if (replaceTransformedExtensions(value, replacement, status)) { + changed++; + temp.setKeywordValue(key, replacement.data(), status); + } + } + if (U_FAILURE(status)) { + return false; + } + } + } + } + if (changed != 0) { + extensionsStr = locale_getKeywordsStart(temp.getName()); + } + out.append(extensionsStr, status); + } + if (U_FAILURE(status)) { + return false; + } + // If the tag is not changed, return. + if (uprv_strcmp(out.data(), locale.getName()) == 0) { + out.clear(); + return false; + } + return true; +} + +// Return true if the locale is changed during canonicalization. +// The replaced value then will be put into out. +bool +canonicalizeLocale(const Locale& locale, CharString& out, UErrorCode& status) +{ + AliasReplacer replacer(status); + return replacer.replace(locale, out, status); +} + +// Function to optimize for known cases without so we can skip the loading +// of resources in the startup time until we really need it. +bool +isKnownCanonicalizedLocale(const char* locale, UErrorCode& status) +{ + if ( uprv_strcmp(locale, "c") == 0 || + uprv_strcmp(locale, "en") == 0 || + uprv_strcmp(locale, "en_US") == 0) { + return true; + } + + // common well-known Canonicalized. + umtx_initOnce(gKnownCanonicalizedInitOnce, + &loadKnownCanonicalized, status); + if (U_FAILURE(status)) { + return false; + } + U_ASSERT(gKnownCanonicalized != nullptr); + return uhash_geti(gKnownCanonicalized, locale) != 0; +} + +} // namespace + +// Function for testing. +U_CAPI const char* const* +ulocimp_getKnownCanonicalizedLocaleForTest(int32_t* length) +{ + *length = UPRV_LENGTHOF(KNOWN_CANONICALIZED); + return KNOWN_CANONICALIZED; +} + +// Function for testing. +U_CAPI bool +ulocimp_isCanonicalizedLocaleForTest(const char* localeName) +{ + Locale l(localeName); + UErrorCode status = U_ZERO_ERROR; + CharString temp; + return !canonicalizeLocale(l, temp, status) && U_SUCCESS(status); +} + +/*This function initializes a Locale from a C locale ID*/ +Locale& Locale::init(const char* localeID, UBool canonicalize) +{ + fIsBogus = false; + /* Free our current storage */ + if ((baseName != fullName) && (baseName != fullNameBuffer)) { + uprv_free(baseName); + } + baseName = nullptr; + if(fullName != fullNameBuffer) { + uprv_free(fullName); + fullName = fullNameBuffer; + } + + // not a loop: + // just an easy way to have a common error-exit + // without goto and without another function + do { + char *separator; + char *field[5] = {0}; + int32_t fieldLen[5] = {0}; + int32_t fieldIdx; + int32_t variantField; + int32_t length; + UErrorCode err; + + if(localeID == nullptr) { + // not an error, just set the default locale + return *this = getDefault(); + } + + /* preset all fields to empty */ + language[0] = script[0] = country[0] = 0; + + // "canonicalize" the locale ID to ICU/Java format + err = U_ZERO_ERROR; + length = canonicalize ? + uloc_canonicalize(localeID, fullName, sizeof(fullNameBuffer), &err) : + uloc_getName(localeID, fullName, sizeof(fullNameBuffer), &err); + + if(err == U_BUFFER_OVERFLOW_ERROR || length >= (int32_t)sizeof(fullNameBuffer)) { + U_ASSERT(baseName == nullptr); + /*Go to heap for the fullName if necessary*/ + fullName = (char *)uprv_malloc(sizeof(char)*(length + 1)); + if(fullName == 0) { + fullName = fullNameBuffer; + break; // error: out of memory + } + err = U_ZERO_ERROR; + length = canonicalize ? + uloc_canonicalize(localeID, fullName, length+1, &err) : + uloc_getName(localeID, fullName, length+1, &err); + } + if(U_FAILURE(err) || err == U_STRING_NOT_TERMINATED_WARNING) { + /* should never occur */ + break; + } + + variantBegin = length; + + /* after uloc_getName/canonicalize() we know that only '_' are separators */ + /* But _ could also appeared in timezone such as "en@timezone=America/Los_Angeles" */ + separator = field[0] = fullName; + fieldIdx = 1; + char* at = uprv_strchr(fullName, '@'); + while ((separator = uprv_strchr(field[fieldIdx-1], SEP_CHAR)) != 0 && + fieldIdx < UPRV_LENGTHOF(field)-1 && + (at == nullptr || separator < at)) { + field[fieldIdx] = separator + 1; + fieldLen[fieldIdx-1] = (int32_t)(separator - field[fieldIdx-1]); + fieldIdx++; + } + // variant may contain @foo or .foo POSIX cruft; remove it + separator = uprv_strchr(field[fieldIdx-1], '@'); + char* sep2 = uprv_strchr(field[fieldIdx-1], '.'); + if (separator!=nullptr || sep2!=nullptr) { + if (separator==nullptr || (sep2!=nullptr && separator > sep2)) { + separator = sep2; + } + fieldLen[fieldIdx-1] = (int32_t)(separator - field[fieldIdx-1]); + } else { + fieldLen[fieldIdx-1] = length - (int32_t)(field[fieldIdx-1] - fullName); + } + + if (fieldLen[0] >= (int32_t)(sizeof(language))) + { + break; // error: the language field is too long + } + + variantField = 1; /* Usually the 2nd one, except when a script or country is also used. */ + if (fieldLen[0] > 0) { + /* We have a language */ + uprv_memcpy(language, fullName, fieldLen[0]); + language[fieldLen[0]] = 0; + } + if (fieldLen[1] == 4 && uprv_isASCIILetter(field[1][0]) && + uprv_isASCIILetter(field[1][1]) && uprv_isASCIILetter(field[1][2]) && + uprv_isASCIILetter(field[1][3])) { + /* We have at least a script */ + uprv_memcpy(script, field[1], fieldLen[1]); + script[fieldLen[1]] = 0; + variantField++; + } + + if (fieldLen[variantField] == 2 || fieldLen[variantField] == 3) { + /* We have a country */ + uprv_memcpy(country, field[variantField], fieldLen[variantField]); + country[fieldLen[variantField]] = 0; + variantField++; + } else if (fieldLen[variantField] == 0) { + variantField++; /* script or country empty but variant in next field (i.e. en__POSIX) */ + } + + if (fieldLen[variantField] > 0) { + /* We have a variant */ + variantBegin = (int32_t)(field[variantField] - fullName); + } + + err = U_ZERO_ERROR; + initBaseName(err); + if (U_FAILURE(err)) { + break; + } + + if (canonicalize) { + if (!isKnownCanonicalizedLocale(fullName, err)) { + CharString replaced; + // Not sure it is already canonicalized + if (canonicalizeLocale(*this, replaced, err)) { + U_ASSERT(U_SUCCESS(err)); + // If need replacement, call init again. + init(replaced.data(), false); + } + if (U_FAILURE(err)) { + break; + } + } + } // if (canonicalize) { + + // successful end of init() + return *this; + } while(0); /*loop doesn't iterate*/ + + // when an error occurs, then set this object to "bogus" (there is no UErrorCode here) + setToBogus(); + + return *this; +} + +/* + * Set up the base name. + * If there are no key words, it's exactly the full name. + * If key words exist, it's the full name truncated at the '@' character. + * Need to set up both at init() and after setting a keyword. + */ +void +Locale::initBaseName(UErrorCode &status) { + if (U_FAILURE(status)) { + return; + } + U_ASSERT(baseName==nullptr || baseName==fullName); + const char *atPtr = uprv_strchr(fullName, '@'); + const char *eqPtr = uprv_strchr(fullName, '='); + if (atPtr && eqPtr && atPtr < eqPtr) { + // Key words exist. + int32_t baseNameLength = (int32_t)(atPtr - fullName); + baseName = (char *)uprv_malloc(baseNameLength + 1); + if (baseName == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + uprv_strncpy(baseName, fullName, baseNameLength); + baseName[baseNameLength] = 0; + + // The original computation of variantBegin leaves it equal to the length + // of fullName if there is no variant. It should instead be + // the length of the baseName. + if (variantBegin > baseNameLength) { + variantBegin = baseNameLength; + } + } else { + baseName = fullName; + } +} + + +int32_t +Locale::hashCode() const +{ + return ustr_hashCharsN(fullName, static_cast(uprv_strlen(fullName))); +} + +void +Locale::setToBogus() { + /* Free our current storage */ + if((baseName != fullName) && (baseName != fullNameBuffer)) { + uprv_free(baseName); + } + baseName = nullptr; + if(fullName != fullNameBuffer) { + uprv_free(fullName); + fullName = fullNameBuffer; + } + *fullNameBuffer = 0; + *language = 0; + *script = 0; + *country = 0; + fIsBogus = true; + variantBegin = 0; +} + +const Locale& U_EXPORT2 +Locale::getDefault() +{ + { + Mutex lock(&gDefaultLocaleMutex); + if (gDefaultLocale != nullptr) { + return *gDefaultLocale; + } + } + UErrorCode status = U_ZERO_ERROR; + return *locale_set_default_internal(nullptr, status); +} + + + +void U_EXPORT2 +Locale::setDefault( const Locale& newLocale, + UErrorCode& status) +{ + if (U_FAILURE(status)) { + return; + } + + /* Set the default from the full name string of the supplied locale. + * This is a convenient way to access the default locale caching mechanisms. + */ + const char *localeID = newLocale.getName(); + locale_set_default_internal(localeID, status); +} + +void +Locale::addLikelySubtags(UErrorCode& status) { + if (U_FAILURE(status)) { + return; + } + + CharString maximizedLocaleID; + { + CharStringByteSink sink(&maximizedLocaleID); + ulocimp_addLikelySubtags(fullName, sink, &status); + } + + if (U_FAILURE(status)) { + return; + } + + init(maximizedLocaleID.data(), /*canonicalize=*/false); + if (isBogus()) { + status = U_ILLEGAL_ARGUMENT_ERROR; + } +} + +void +Locale::minimizeSubtags(UErrorCode& status) { + if (U_FAILURE(status)) { + return; + } + + CharString minimizedLocaleID; + { + CharStringByteSink sink(&minimizedLocaleID); + ulocimp_minimizeSubtags(fullName, sink, &status); + } + + if (U_FAILURE(status)) { + return; + } + + init(minimizedLocaleID.data(), /*canonicalize=*/false); + if (isBogus()) { + status = U_ILLEGAL_ARGUMENT_ERROR; + } +} + +void +Locale::canonicalize(UErrorCode& status) { + if (U_FAILURE(status)) { + return; + } + if (isBogus()) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + CharString uncanonicalized(fullName, status); + if (U_FAILURE(status)) { + return; + } + init(uncanonicalized.data(), /*canonicalize=*/true); + if (isBogus()) { + status = U_ILLEGAL_ARGUMENT_ERROR; + } +} + +Locale U_EXPORT2 +Locale::forLanguageTag(StringPiece tag, UErrorCode& status) +{ + Locale result(Locale::eBOGUS); + + if (U_FAILURE(status)) { + return result; + } + + // If a BCP 47 language tag is passed as the language parameter to the + // normal Locale constructor, it will actually fall back to invoking + // uloc_forLanguageTag() to parse it if it somehow is able to detect that + // the string actually is BCP 47. This works well for things like strings + // using BCP 47 extensions, but it does not at all work for things like + // legacy language tags (marked as “Type: grandfathered” in BCP 47, + // e.g., "en-GB-oed") which are possible to also + // interpret as ICU locale IDs and because of that won't trigger the BCP 47 + // parsing. Therefore the code here explicitly calls uloc_forLanguageTag() + // and then Locale::init(), instead of just calling the normal constructor. + + CharString localeID; + int32_t parsedLength; + { + CharStringByteSink sink(&localeID); + ulocimp_forLanguageTag( + tag.data(), + tag.length(), + sink, + &parsedLength, + &status); + } + + if (U_FAILURE(status)) { + return result; + } + + if (parsedLength != tag.size()) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return result; + } + + result.init(localeID.data(), /*canonicalize=*/false); + if (result.isBogus()) { + status = U_ILLEGAL_ARGUMENT_ERROR; + } + return result; +} + +void +Locale::toLanguageTag(ByteSink& sink, UErrorCode& status) const +{ + if (U_FAILURE(status)) { + return; + } + + if (fIsBogus) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + ulocimp_toLanguageTag(fullName, sink, /*strict=*/false, &status); +} + +Locale U_EXPORT2 +Locale::createFromName (const char *name) +{ + if (name) { + Locale l(""); + l.init(name, false); + return l; + } + else { + return getDefault(); + } +} + +Locale U_EXPORT2 +Locale::createCanonical(const char* name) { + Locale loc(""); + loc.init(name, true); + return loc; +} + +const char * +Locale::getISO3Language() const +{ + return uloc_getISO3Language(fullName); +} + + +const char * +Locale::getISO3Country() const +{ + return uloc_getISO3Country(fullName); +} + +/** + * Return the LCID value as specified in the "LocaleID" resource for this + * locale. The LocaleID must be expressed as a hexadecimal number, from + * one to four digits. If the LocaleID resource is not present, or is + * in an incorrect format, 0 is returned. The LocaleID is for use in + * Windows (it is an LCID), but is available on all platforms. + */ +uint32_t +Locale::getLCID() const +{ + return uloc_getLCID(fullName); +} + +const char* const* U_EXPORT2 Locale::getISOCountries() +{ + return uloc_getISOCountries(); +} + +const char* const* U_EXPORT2 Locale::getISOLanguages() +{ + return uloc_getISOLanguages(); +} + +// Set the locale's data based on a posix id. +void Locale::setFromPOSIXID(const char *posixID) +{ + init(posixID, true); +} + +const Locale & U_EXPORT2 +Locale::getRoot() +{ + return getLocale(eROOT); +} + +const Locale & U_EXPORT2 +Locale::getEnglish() +{ + return getLocale(eENGLISH); +} + +const Locale & U_EXPORT2 +Locale::getFrench() +{ + return getLocale(eFRENCH); +} + +const Locale & U_EXPORT2 +Locale::getGerman() +{ + return getLocale(eGERMAN); +} + +const Locale & U_EXPORT2 +Locale::getItalian() +{ + return getLocale(eITALIAN); +} + +const Locale & U_EXPORT2 +Locale::getJapanese() +{ + return getLocale(eJAPANESE); +} + +const Locale & U_EXPORT2 +Locale::getKorean() +{ + return getLocale(eKOREAN); +} + +const Locale & U_EXPORT2 +Locale::getChinese() +{ + return getLocale(eCHINESE); +} + +const Locale & U_EXPORT2 +Locale::getSimplifiedChinese() +{ + return getLocale(eCHINA); +} + +const Locale & U_EXPORT2 +Locale::getTraditionalChinese() +{ + return getLocale(eTAIWAN); +} + + +const Locale & U_EXPORT2 +Locale::getFrance() +{ + return getLocale(eFRANCE); +} + +const Locale & U_EXPORT2 +Locale::getGermany() +{ + return getLocale(eGERMANY); +} + +const Locale & U_EXPORT2 +Locale::getItaly() +{ + return getLocale(eITALY); +} + +const Locale & U_EXPORT2 +Locale::getJapan() +{ + return getLocale(eJAPAN); +} + +const Locale & U_EXPORT2 +Locale::getKorea() +{ + return getLocale(eKOREA); +} + +const Locale & U_EXPORT2 +Locale::getChina() +{ + return getLocale(eCHINA); +} + +const Locale & U_EXPORT2 +Locale::getPRC() +{ + return getLocale(eCHINA); +} + +const Locale & U_EXPORT2 +Locale::getTaiwan() +{ + return getLocale(eTAIWAN); +} + +const Locale & U_EXPORT2 +Locale::getUK() +{ + return getLocale(eUK); +} + +const Locale & U_EXPORT2 +Locale::getUS() +{ + return getLocale(eUS); +} + +const Locale & U_EXPORT2 +Locale::getCanada() +{ + return getLocale(eCANADA); +} + +const Locale & U_EXPORT2 +Locale::getCanadaFrench() +{ + return getLocale(eCANADA_FRENCH); +} + +const Locale & +Locale::getLocale(int locid) +{ + Locale *localeCache = getLocaleCache(); + U_ASSERT((locid < eMAX_LOCALES)&&(locid>=0)); + if (localeCache == nullptr) { + // Failure allocating the locale cache. + // The best we can do is return a nullptr reference. + locid = 0; + } + return localeCache[locid]; /*operating on nullptr*/ +} + +/* +This function is defined this way in order to get around static +initialization and static destruction. + */ +Locale * +Locale::getLocaleCache() +{ + UErrorCode status = U_ZERO_ERROR; + umtx_initOnce(gLocaleCacheInitOnce, locale_init, status); + return gLocaleCache; +} + +class KeywordEnumeration : public StringEnumeration { +private: + char *keywords; + char *current; + int32_t length; + UnicodeString currUSKey; + static const char fgClassID;/* Warning this is used beyond the typical RTTI usage. */ + +public: + static UClassID U_EXPORT2 getStaticClassID() { return (UClassID)&fgClassID; } + virtual UClassID getDynamicClassID() const override { return getStaticClassID(); } +public: + KeywordEnumeration(const char *keys, int32_t keywordLen, int32_t currentIndex, UErrorCode &status) + : keywords((char *)&fgClassID), current((char *)&fgClassID), length(0) { + if(U_SUCCESS(status) && keywordLen != 0) { + if(keys == nullptr || keywordLen < 0) { + status = U_ILLEGAL_ARGUMENT_ERROR; + } else { + keywords = (char *)uprv_malloc(keywordLen+1); + if (keywords == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + } + else { + uprv_memcpy(keywords, keys, keywordLen); + keywords[keywordLen] = 0; + current = keywords + currentIndex; + length = keywordLen; + } + } + } + } + + virtual ~KeywordEnumeration(); + + virtual StringEnumeration * clone() const override + { + UErrorCode status = U_ZERO_ERROR; + return new KeywordEnumeration(keywords, length, (int32_t)(current - keywords), status); + } + + virtual int32_t count(UErrorCode &/*status*/) const override { + char *kw = keywords; + int32_t result = 0; + while(*kw) { + result++; + kw += uprv_strlen(kw)+1; + } + return result; + } + + virtual const char* next(int32_t* resultLength, UErrorCode& status) override { + const char* result; + int32_t len; + if(U_SUCCESS(status) && *current != 0) { + result = current; + len = (int32_t)uprv_strlen(current); + current += len+1; + if(resultLength != nullptr) { + *resultLength = len; + } + } else { + if(resultLength != nullptr) { + *resultLength = 0; + } + result = nullptr; + } + return result; + } + + virtual const UnicodeString* snext(UErrorCode& status) override { + int32_t resultLength = 0; + const char *s = next(&resultLength, status); + return setChars(s, resultLength, status); + } + + virtual void reset(UErrorCode& /*status*/) override { + current = keywords; + } +}; + +const char KeywordEnumeration::fgClassID = '\0'; + +KeywordEnumeration::~KeywordEnumeration() { + uprv_free(keywords); +} + +// A wrapper around KeywordEnumeration that calls uloc_toUnicodeLocaleKey() in +// the next() method for each keyword before returning it. +class UnicodeKeywordEnumeration : public KeywordEnumeration { +public: + using KeywordEnumeration::KeywordEnumeration; + virtual ~UnicodeKeywordEnumeration(); + + virtual const char* next(int32_t* resultLength, UErrorCode& status) override { + const char* legacy_key = KeywordEnumeration::next(nullptr, status); + while (U_SUCCESS(status) && legacy_key != nullptr) { + const char* key = uloc_toUnicodeLocaleKey(legacy_key); + if (key != nullptr) { + if (resultLength != nullptr) { + *resultLength = static_cast(uprv_strlen(key)); + } + return key; + } + // Not a Unicode keyword, could be a t, x or other, continue to look at the next one. + legacy_key = KeywordEnumeration::next(nullptr, status); + } + if (resultLength != nullptr) *resultLength = 0; + return nullptr; + } +}; + +// Out-of-line virtual destructor to serve as the "key function". +UnicodeKeywordEnumeration::~UnicodeKeywordEnumeration() = default; + +StringEnumeration * +Locale::createKeywords(UErrorCode &status) const +{ + StringEnumeration *result = nullptr; + + if (U_FAILURE(status)) { + return result; + } + + const char* variantStart = uprv_strchr(fullName, '@'); + const char* assignment = uprv_strchr(fullName, '='); + if(variantStart) { + if(assignment > variantStart) { + CharString keywords; + CharStringByteSink sink(&keywords); + ulocimp_getKeywords(variantStart+1, '@', sink, false, &status); + if (U_SUCCESS(status) && !keywords.isEmpty()) { + result = new KeywordEnumeration(keywords.data(), keywords.length(), 0, status); + if (!result) { + status = U_MEMORY_ALLOCATION_ERROR; + } + } + } else { + status = U_INVALID_FORMAT_ERROR; + } + } + return result; +} + +StringEnumeration * +Locale::createUnicodeKeywords(UErrorCode &status) const +{ + StringEnumeration *result = nullptr; + + if (U_FAILURE(status)) { + return result; + } + + const char* variantStart = uprv_strchr(fullName, '@'); + const char* assignment = uprv_strchr(fullName, '='); + if(variantStart) { + if(assignment > variantStart) { + CharString keywords; + CharStringByteSink sink(&keywords); + ulocimp_getKeywords(variantStart+1, '@', sink, false, &status); + if (U_SUCCESS(status) && !keywords.isEmpty()) { + result = new UnicodeKeywordEnumeration(keywords.data(), keywords.length(), 0, status); + if (!result) { + status = U_MEMORY_ALLOCATION_ERROR; + } + } + } else { + status = U_INVALID_FORMAT_ERROR; + } + } + return result; +} + +int32_t +Locale::getKeywordValue(const char* keywordName, char *buffer, int32_t bufLen, UErrorCode &status) const +{ + return uloc_getKeywordValue(fullName, keywordName, buffer, bufLen, &status); +} + +void +Locale::getKeywordValue(StringPiece keywordName, ByteSink& sink, UErrorCode& status) const { + if (U_FAILURE(status)) { + return; + } + + if (fIsBogus) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + // TODO: Remove the need for a const char* to a NUL terminated buffer. + const CharString keywordName_nul(keywordName, status); + if (U_FAILURE(status)) { + return; + } + + ulocimp_getKeywordValue(fullName, keywordName_nul.data(), sink, &status); +} + +void +Locale::getUnicodeKeywordValue(StringPiece keywordName, + ByteSink& sink, + UErrorCode& status) const { + // TODO: Remove the need for a const char* to a NUL terminated buffer. + const CharString keywordName_nul(keywordName, status); + if (U_FAILURE(status)) { + return; + } + + const char* legacy_key = uloc_toLegacyKey(keywordName_nul.data()); + + if (legacy_key == nullptr) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + CharString legacy_value; + { + CharStringByteSink sink(&legacy_value); + getKeywordValue(legacy_key, sink, status); + } + + if (U_FAILURE(status)) { + return; + } + + const char* unicode_value = uloc_toUnicodeLocaleType( + keywordName_nul.data(), legacy_value.data()); + + if (unicode_value == nullptr) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + sink.Append(unicode_value, static_cast(uprv_strlen(unicode_value))); +} + +void +Locale::setKeywordValue(const char* keywordName, const char* keywordValue, UErrorCode &status) +{ + if (U_FAILURE(status)) { + return; + } + if (status == U_STRING_NOT_TERMINATED_WARNING) { + status = U_ZERO_ERROR; + } + int32_t bufferLength = uprv_max((int32_t)(uprv_strlen(fullName) + 1), ULOC_FULLNAME_CAPACITY); + int32_t newLength = uloc_setKeywordValue(keywordName, keywordValue, fullName, + bufferLength, &status) + 1; + U_ASSERT(status != U_STRING_NOT_TERMINATED_WARNING); + /* Handle the case the current buffer is not enough to hold the new id */ + if (status == U_BUFFER_OVERFLOW_ERROR) { + U_ASSERT(newLength > bufferLength); + char* newFullName = (char *)uprv_malloc(newLength); + if (newFullName == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + uprv_strcpy(newFullName, fullName); + if (fullName != fullNameBuffer) { + // if full Name is already on the heap, need to free it. + uprv_free(fullName); + if (baseName == fullName) { + baseName = newFullName; // baseName should not point to freed memory. + } + } + fullName = newFullName; + status = U_ZERO_ERROR; + uloc_setKeywordValue(keywordName, keywordValue, fullName, newLength, &status); + U_ASSERT(status != U_STRING_NOT_TERMINATED_WARNING); + } else { + U_ASSERT(newLength <= bufferLength); + } + if (U_SUCCESS(status) && baseName == fullName) { + // May have added the first keyword, meaning that the fullName is no longer also the baseName. + initBaseName(status); + } +} + +void +Locale::setKeywordValue(StringPiece keywordName, + StringPiece keywordValue, + UErrorCode& status) { + // TODO: Remove the need for a const char* to a NUL terminated buffer. + const CharString keywordName_nul(keywordName, status); + const CharString keywordValue_nul(keywordValue, status); + setKeywordValue(keywordName_nul.data(), keywordValue_nul.data(), status); +} + +void +Locale::setUnicodeKeywordValue(StringPiece keywordName, + StringPiece keywordValue, + UErrorCode& status) { + // TODO: Remove the need for a const char* to a NUL terminated buffer. + const CharString keywordName_nul(keywordName, status); + const CharString keywordValue_nul(keywordValue, status); + + if (U_FAILURE(status)) { + return; + } + + const char* legacy_key = uloc_toLegacyKey(keywordName_nul.data()); + + if (legacy_key == nullptr) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + const char* legacy_value = nullptr; + + if (!keywordValue_nul.isEmpty()) { + legacy_value = + uloc_toLegacyType(keywordName_nul.data(), keywordValue_nul.data()); + + if (legacy_value == nullptr) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + } + + setKeywordValue(legacy_key, legacy_value, status); +} + +const char * +Locale::getBaseName() const { + return baseName; +} + +Locale::Iterator::~Iterator() = default; + +//eof +U_NAMESPACE_END diff --git a/intl/icu/source/common/loclikely.cpp b/intl/icu/source/common/loclikely.cpp new file mode 100644 index 0000000000..d2a05c6364 --- /dev/null +++ b/intl/icu/source/common/loclikely.cpp @@ -0,0 +1,1415 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 1997-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: loclikely.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010feb25 +* created by: Markus W. Scherer +* +* Code for likely and minimized locale subtags, separated out from other .cpp files +* that then do not depend on resource bundle code and likely-subtags data. +*/ + +#include "unicode/bytestream.h" +#include "unicode/utypes.h" +#include "unicode/locid.h" +#include "unicode/putil.h" +#include "unicode/uchar.h" +#include "unicode/uloc.h" +#include "unicode/ures.h" +#include "unicode/uscript.h" +#include "bytesinkutil.h" +#include "charstr.h" +#include "cmemory.h" +#include "cstring.h" +#include "ulocimp.h" +#include "ustr_imp.h" + +/** + * These are the canonical strings for unknown languages, scripts and regions. + **/ +static const char* const unknownLanguage = "und"; +static const char* const unknownScript = "Zzzz"; +static const char* const unknownRegion = "ZZ"; + +/** + * This function looks for the localeID in the likelySubtags resource. + * + * @param localeID The tag to find. + * @param buffer A buffer to hold the matching entry + * @param bufferLength The length of the output buffer + * @return A pointer to "buffer" if found, or a null pointer if not. + */ +static const char* U_CALLCONV +findLikelySubtags(const char* localeID, + char* buffer, + int32_t bufferLength, + UErrorCode* err) { + const char* result = nullptr; + + if (!U_FAILURE(*err)) { + int32_t resLen = 0; + const char16_t* s = nullptr; + UErrorCode tmpErr = U_ZERO_ERROR; + icu::LocalUResourceBundlePointer subtags(ures_openDirect(nullptr, "likelySubtags", &tmpErr)); + if (U_SUCCESS(tmpErr)) { + icu::CharString und; + if (localeID != nullptr) { + if (*localeID == '\0') { + localeID = unknownLanguage; + } else if (*localeID == '_') { + und.append(unknownLanguage, *err); + und.append(localeID, *err); + if (U_FAILURE(*err)) { + return nullptr; + } + localeID = und.data(); + } + } + s = ures_getStringByKey(subtags.getAlias(), localeID, &resLen, &tmpErr); + + if (U_FAILURE(tmpErr)) { + /* + * If a resource is missing, it's not really an error, it's + * just that we don't have any data for that particular locale ID. + */ + if (tmpErr != U_MISSING_RESOURCE_ERROR) { + *err = tmpErr; + } + } + else if (resLen >= bufferLength) { + /* The buffer should never overflow. */ + *err = U_INTERNAL_PROGRAM_ERROR; + } + else { + u_UCharsToChars(s, buffer, resLen + 1); + if (resLen >= 3 && + uprv_strnicmp(buffer, unknownLanguage, 3) == 0 && + (resLen == 3 || buffer[3] == '_')) { + uprv_memmove(buffer, buffer + 3, resLen - 3 + 1); + } + result = buffer; + } + } else { + *err = tmpErr; + } + } + + return result; +} + +/** + * Append a tag to a buffer, adding the separator if necessary. The buffer + * must be large enough to contain the resulting tag plus any separator + * necessary. The tag must not be a zero-length string. + * + * @param tag The tag to add. + * @param tagLength The length of the tag. + * @param buffer The output buffer. + * @param bufferLength The length of the output buffer. This is an input/output parameter. + **/ +static void U_CALLCONV +appendTag( + const char* tag, + int32_t tagLength, + char* buffer, + int32_t* bufferLength, + UBool withSeparator) { + + if (withSeparator) { + buffer[*bufferLength] = '_'; + ++(*bufferLength); + } + + uprv_memmove( + &buffer[*bufferLength], + tag, + tagLength); + + *bufferLength += tagLength; +} + +/** + * Create a tag string from the supplied parameters. The lang, script and region + * parameters may be nullptr pointers. If they are, their corresponding length parameters + * must be less than or equal to 0. + * + * If any of the language, script or region parameters are empty, and the alternateTags + * parameter is not nullptr, it will be parsed for potential language, script and region tags + * to be used when constructing the new tag. If the alternateTags parameter is nullptr, or + * it contains no language tag, the default tag for the unknown language is used. + * + * If the length of the new string exceeds the capacity of the output buffer, + * the function copies as many bytes to the output buffer as it can, and returns + * the error U_BUFFER_OVERFLOW_ERROR. + * + * If an illegal argument is provided, the function returns the error + * U_ILLEGAL_ARGUMENT_ERROR. + * + * Note that this function can return the warning U_STRING_NOT_TERMINATED_WARNING if + * the tag string fits in the output buffer, but the null terminator doesn't. + * + * @param lang The language tag to use. + * @param langLength The length of the language tag. + * @param script The script tag to use. + * @param scriptLength The length of the script tag. + * @param region The region tag to use. + * @param regionLength The length of the region tag. + * @param trailing Any trailing data to append to the new tag. + * @param trailingLength The length of the trailing data. + * @param alternateTags A string containing any alternate tags. + * @param sink The output sink receiving the tag string. + * @param err A pointer to a UErrorCode for error reporting. + **/ +static void U_CALLCONV +createTagStringWithAlternates( + const char* lang, + int32_t langLength, + const char* script, + int32_t scriptLength, + const char* region, + int32_t regionLength, + const char* trailing, + int32_t trailingLength, + const char* alternateTags, + icu::ByteSink& sink, + UErrorCode* err) { + + if (U_FAILURE(*err)) { + goto error; + } + else if (langLength >= ULOC_LANG_CAPACITY || + scriptLength >= ULOC_SCRIPT_CAPACITY || + regionLength >= ULOC_COUNTRY_CAPACITY) { + goto error; + } + else { + /** + * ULOC_FULLNAME_CAPACITY will provide enough capacity + * that we can build a string that contains the language, + * script and region code without worrying about overrunning + * the user-supplied buffer. + **/ + char tagBuffer[ULOC_FULLNAME_CAPACITY]; + int32_t tagLength = 0; + UBool regionAppended = false; + + if (langLength > 0) { + appendTag( + lang, + langLength, + tagBuffer, + &tagLength, + /*withSeparator=*/false); + } + else if (alternateTags == nullptr) { + /* + * Use the empty string for an unknown language, if + * we found no language. + */ + } + else { + /* + * Parse the alternateTags string for the language. + */ + char alternateLang[ULOC_LANG_CAPACITY]; + int32_t alternateLangLength = sizeof(alternateLang); + + alternateLangLength = + uloc_getLanguage( + alternateTags, + alternateLang, + alternateLangLength, + err); + if(U_FAILURE(*err) || + alternateLangLength >= ULOC_LANG_CAPACITY) { + goto error; + } + else if (alternateLangLength == 0) { + /* + * Use the empty string for an unknown language, if + * we found no language. + */ + } + else { + appendTag( + alternateLang, + alternateLangLength, + tagBuffer, + &tagLength, + /*withSeparator=*/false); + } + } + + if (scriptLength > 0) { + appendTag( + script, + scriptLength, + tagBuffer, + &tagLength, + /*withSeparator=*/true); + } + else if (alternateTags != nullptr) { + /* + * Parse the alternateTags string for the script. + */ + char alternateScript[ULOC_SCRIPT_CAPACITY]; + + const int32_t alternateScriptLength = + uloc_getScript( + alternateTags, + alternateScript, + sizeof(alternateScript), + err); + + if (U_FAILURE(*err) || + alternateScriptLength >= ULOC_SCRIPT_CAPACITY) { + goto error; + } + else if (alternateScriptLength > 0) { + appendTag( + alternateScript, + alternateScriptLength, + tagBuffer, + &tagLength, + /*withSeparator=*/true); + } + } + + if (regionLength > 0) { + appendTag( + region, + regionLength, + tagBuffer, + &tagLength, + /*withSeparator=*/true); + + regionAppended = true; + } + else if (alternateTags != nullptr) { + /* + * Parse the alternateTags string for the region. + */ + char alternateRegion[ULOC_COUNTRY_CAPACITY]; + + const int32_t alternateRegionLength = + uloc_getCountry( + alternateTags, + alternateRegion, + sizeof(alternateRegion), + err); + if (U_FAILURE(*err) || + alternateRegionLength >= ULOC_COUNTRY_CAPACITY) { + goto error; + } + else if (alternateRegionLength > 0) { + appendTag( + alternateRegion, + alternateRegionLength, + tagBuffer, + &tagLength, + /*withSeparator=*/true); + + regionAppended = true; + } + } + + /** + * Copy the partial tag from our internal buffer to the supplied + * target. + **/ + sink.Append(tagBuffer, tagLength); + + if (trailingLength > 0) { + if (*trailing != '@') { + sink.Append("_", 1); + if (!regionAppended) { + /* extra separator is required */ + sink.Append("_", 1); + } + } + + /* + * Copy the trailing data into the supplied buffer. + */ + sink.Append(trailing, trailingLength); + } + + return; + } + +error: + + /** + * An overflow indicates the locale ID passed in + * is ill-formed. If we got here, and there was + * no previous error, it's an implicit overflow. + **/ + if (*err == U_BUFFER_OVERFLOW_ERROR || + U_SUCCESS(*err)) { + *err = U_ILLEGAL_ARGUMENT_ERROR; + } +} + +/** + * Create a tag string from the supplied parameters. The lang, script and region + * parameters may be nullptr pointers. If they are, their corresponding length parameters + * must be less than or equal to 0. If the lang parameter is an empty string, the + * default value for an unknown language is written to the output buffer. + * + * If the length of the new string exceeds the capacity of the output buffer, + * the function copies as many bytes to the output buffer as it can, and returns + * the error U_BUFFER_OVERFLOW_ERROR. + * + * If an illegal argument is provided, the function returns the error + * U_ILLEGAL_ARGUMENT_ERROR. + * + * @param lang The language tag to use. + * @param langLength The length of the language tag. + * @param script The script tag to use. + * @param scriptLength The length of the script tag. + * @param region The region tag to use. + * @param regionLength The length of the region tag. + * @param trailing Any trailing data to append to the new tag. + * @param trailingLength The length of the trailing data. + * @param sink The output sink receiving the tag string. + * @param err A pointer to a UErrorCode for error reporting. + **/ +static void U_CALLCONV +createTagString( + const char* lang, + int32_t langLength, + const char* script, + int32_t scriptLength, + const char* region, + int32_t regionLength, + const char* trailing, + int32_t trailingLength, + icu::ByteSink& sink, + UErrorCode* err) +{ + createTagStringWithAlternates( + lang, + langLength, + script, + scriptLength, + region, + regionLength, + trailing, + trailingLength, + nullptr, + sink, + err); +} + +/** + * Parse the language, script, and region subtags from a tag string, and copy the + * results into the corresponding output parameters. The buffers are null-terminated, + * unless overflow occurs. + * + * The langLength, scriptLength, and regionLength parameters are input/output + * parameters, and must contain the capacity of their corresponding buffers on + * input. On output, they will contain the actual length of the buffers, not + * including the null terminator. + * + * If the length of any of the output subtags exceeds the capacity of the corresponding + * buffer, the function copies as many bytes to the output buffer as it can, and returns + * the error U_BUFFER_OVERFLOW_ERROR. It will not parse any more subtags once overflow + * occurs. + * + * If an illegal argument is provided, the function returns the error + * U_ILLEGAL_ARGUMENT_ERROR. + * + * @param localeID The locale ID to parse. + * @param lang The language tag buffer. + * @param langLength The length of the language tag. + * @param script The script tag buffer. + * @param scriptLength The length of the script tag. + * @param region The region tag buffer. + * @param regionLength The length of the region tag. + * @param err A pointer to a UErrorCode for error reporting. + * @return The number of chars of the localeID parameter consumed. + **/ +static int32_t U_CALLCONV +parseTagString( + const char* localeID, + char* lang, + int32_t* langLength, + char* script, + int32_t* scriptLength, + char* region, + int32_t* regionLength, + UErrorCode* err) +{ + const char* position = localeID; + int32_t subtagLength = 0; + + if(U_FAILURE(*err) || + localeID == nullptr || + lang == nullptr || + langLength == nullptr || + script == nullptr || + scriptLength == nullptr || + region == nullptr || + regionLength == nullptr) { + goto error; + } + + subtagLength = ulocimp_getLanguage(position, &position, *err).extract(lang, *langLength, *err); + + /* + * Note that we explicit consider U_STRING_NOT_TERMINATED_WARNING + * to be an error, because it indicates the user-supplied tag is + * not well-formed. + */ + if(U_FAILURE(*err)) { + goto error; + } + + *langLength = subtagLength; + + /* + * If no language was present, use the empty string instead. + * Otherwise, move past any separator. + */ + if (_isIDSeparator(*position)) { + ++position; + } + + subtagLength = ulocimp_getScript(position, &position, *err).extract(script, *scriptLength, *err); + + if(U_FAILURE(*err)) { + goto error; + } + + *scriptLength = subtagLength; + + if (*scriptLength > 0) { + if (uprv_strnicmp(script, unknownScript, *scriptLength) == 0) { + /** + * If the script part is the "unknown" script, then don't return it. + **/ + *scriptLength = 0; + } + + /* + * Move past any separator. + */ + if (_isIDSeparator(*position)) { + ++position; + } + } + + subtagLength = ulocimp_getCountry(position, &position, *err).extract(region, *regionLength, *err); + + if(U_FAILURE(*err)) { + goto error; + } + + *regionLength = subtagLength; + + if (*regionLength > 0) { + if (uprv_strnicmp(region, unknownRegion, *regionLength) == 0) { + /** + * If the region part is the "unknown" region, then don't return it. + **/ + *regionLength = 0; + } + } else if (*position != 0 && *position != '@') { + /* back up over consumed trailing separator */ + --position; + } + +exit: + + return (int32_t)(position - localeID); + +error: + + /** + * If we get here, we have no explicit error, it's the result of an + * illegal argument. + **/ + if (!U_FAILURE(*err)) { + *err = U_ILLEGAL_ARGUMENT_ERROR; + } + + goto exit; +} + +static UBool U_CALLCONV +createLikelySubtagsString( + const char* lang, + int32_t langLength, + const char* script, + int32_t scriptLength, + const char* region, + int32_t regionLength, + const char* variants, + int32_t variantsLength, + icu::ByteSink& sink, + UErrorCode* err) { + /** + * ULOC_FULLNAME_CAPACITY will provide enough capacity + * that we can build a string that contains the language, + * script and region code without worrying about overrunning + * the user-supplied buffer. + **/ + char likelySubtagsBuffer[ULOC_FULLNAME_CAPACITY]; + + if(U_FAILURE(*err)) { + goto error; + } + + /** + * Try the language with the script and region first. + **/ + if (scriptLength > 0 && regionLength > 0) { + + const char* likelySubtags = nullptr; + + icu::CharString tagBuffer; + { + icu::CharStringByteSink sink(&tagBuffer); + createTagString( + lang, + langLength, + script, + scriptLength, + region, + regionLength, + nullptr, + 0, + sink, + err); + } + if(U_FAILURE(*err)) { + goto error; + } + + likelySubtags = + findLikelySubtags( + tagBuffer.data(), + likelySubtagsBuffer, + sizeof(likelySubtagsBuffer), + err); + if(U_FAILURE(*err)) { + goto error; + } + + if (likelySubtags != nullptr) { + /* Always use the language tag from the + maximal string, since it may be more + specific than the one provided. */ + createTagStringWithAlternates( + nullptr, + 0, + nullptr, + 0, + nullptr, + 0, + variants, + variantsLength, + likelySubtags, + sink, + err); + return true; + } + } + + /** + * Try the language with just the script. + **/ + if (scriptLength > 0) { + + const char* likelySubtags = nullptr; + + icu::CharString tagBuffer; + { + icu::CharStringByteSink sink(&tagBuffer); + createTagString( + lang, + langLength, + script, + scriptLength, + nullptr, + 0, + nullptr, + 0, + sink, + err); + } + if(U_FAILURE(*err)) { + goto error; + } + + likelySubtags = + findLikelySubtags( + tagBuffer.data(), + likelySubtagsBuffer, + sizeof(likelySubtagsBuffer), + err); + if(U_FAILURE(*err)) { + goto error; + } + + if (likelySubtags != nullptr) { + /* Always use the language tag from the + maximal string, since it may be more + specific than the one provided. */ + createTagStringWithAlternates( + nullptr, + 0, + nullptr, + 0, + region, + regionLength, + variants, + variantsLength, + likelySubtags, + sink, + err); + return true; + } + } + + /** + * Try the language with just the region. + **/ + if (regionLength > 0) { + + const char* likelySubtags = nullptr; + + icu::CharString tagBuffer; + { + icu::CharStringByteSink sink(&tagBuffer); + createTagString( + lang, + langLength, + nullptr, + 0, + region, + regionLength, + nullptr, + 0, + sink, + err); + } + if(U_FAILURE(*err)) { + goto error; + } + + likelySubtags = + findLikelySubtags( + tagBuffer.data(), + likelySubtagsBuffer, + sizeof(likelySubtagsBuffer), + err); + if(U_FAILURE(*err)) { + goto error; + } + + if (likelySubtags != nullptr) { + /* Always use the language tag from the + maximal string, since it may be more + specific than the one provided. */ + createTagStringWithAlternates( + nullptr, + 0, + script, + scriptLength, + nullptr, + 0, + variants, + variantsLength, + likelySubtags, + sink, + err); + return true; + } + } + + /** + * Finally, try just the language. + **/ + { + const char* likelySubtags = nullptr; + + icu::CharString tagBuffer; + { + icu::CharStringByteSink sink(&tagBuffer); + createTagString( + lang, + langLength, + nullptr, + 0, + nullptr, + 0, + nullptr, + 0, + sink, + err); + } + if(U_FAILURE(*err)) { + goto error; + } + + likelySubtags = + findLikelySubtags( + tagBuffer.data(), + likelySubtagsBuffer, + sizeof(likelySubtagsBuffer), + err); + if(U_FAILURE(*err)) { + goto error; + } + + if (likelySubtags != nullptr) { + /* Always use the language tag from the + maximal string, since it may be more + specific than the one provided. */ + createTagStringWithAlternates( + nullptr, + 0, + script, + scriptLength, + region, + regionLength, + variants, + variantsLength, + likelySubtags, + sink, + err); + return true; + } + } + + return false; + +error: + + if (!U_FAILURE(*err)) { + *err = U_ILLEGAL_ARGUMENT_ERROR; + } + + return false; +} + +#define CHECK_TRAILING_VARIANT_SIZE(trailing, trailingLength) UPRV_BLOCK_MACRO_BEGIN { \ + int32_t count = 0; \ + int32_t i; \ + for (i = 0; i < trailingLength; i++) { \ + if (trailing[i] == '-' || trailing[i] == '_') { \ + count = 0; \ + if (count > 8) { \ + goto error; \ + } \ + } else if (trailing[i] == '@') { \ + break; \ + } else if (count > 8) { \ + goto error; \ + } else { \ + count++; \ + } \ + } \ +} UPRV_BLOCK_MACRO_END + +static UBool +_uloc_addLikelySubtags(const char* localeID, + icu::ByteSink& sink, + UErrorCode* err) { + char lang[ULOC_LANG_CAPACITY]; + int32_t langLength = sizeof(lang); + char script[ULOC_SCRIPT_CAPACITY]; + int32_t scriptLength = sizeof(script); + char region[ULOC_COUNTRY_CAPACITY]; + int32_t regionLength = sizeof(region); + const char* trailing = ""; + int32_t trailingLength = 0; + int32_t trailingIndex = 0; + UBool success = false; + + if(U_FAILURE(*err)) { + goto error; + } + if (localeID == nullptr) { + goto error; + } + + trailingIndex = parseTagString( + localeID, + lang, + &langLength, + script, + &scriptLength, + region, + ®ionLength, + err); + if(U_FAILURE(*err)) { + /* Overflow indicates an illegal argument error */ + if (*err == U_BUFFER_OVERFLOW_ERROR) { + *err = U_ILLEGAL_ARGUMENT_ERROR; + } + + goto error; + } + + /* Find the length of the trailing portion. */ + while (_isIDSeparator(localeID[trailingIndex])) { + trailingIndex++; + } + trailing = &localeID[trailingIndex]; + trailingLength = (int32_t)uprv_strlen(trailing); + + CHECK_TRAILING_VARIANT_SIZE(trailing, trailingLength); + + success = + createLikelySubtagsString( + lang, + langLength, + script, + scriptLength, + region, + regionLength, + trailing, + trailingLength, + sink, + err); + + if (!success) { + const int32_t localIDLength = (int32_t)uprv_strlen(localeID); + + /* + * If we get here, we need to return localeID. + */ + sink.Append(localeID, localIDLength); + } + + return success; + +error: + + if (!U_FAILURE(*err)) { + *err = U_ILLEGAL_ARGUMENT_ERROR; + } + return false; +} + +// Add likely subtags to the sink +// return true if the value in the sink is produced by a match during the lookup +// return false if the value in the sink is the same as input because there are +// no match after the lookup. +static UBool _ulocimp_addLikelySubtags(const char*, icu::ByteSink&, UErrorCode*); + +static void +_uloc_minimizeSubtags(const char* localeID, + icu::ByteSink& sink, + UErrorCode* err) { + icu::CharString maximizedTagBuffer; + + char lang[ULOC_LANG_CAPACITY]; + int32_t langLength = sizeof(lang); + char script[ULOC_SCRIPT_CAPACITY]; + int32_t scriptLength = sizeof(script); + char region[ULOC_COUNTRY_CAPACITY]; + int32_t regionLength = sizeof(region); + const char* trailing = ""; + int32_t trailingLength = 0; + int32_t trailingIndex = 0; + UBool successGetMax = false; + + if(U_FAILURE(*err)) { + goto error; + } + else if (localeID == nullptr) { + goto error; + } + + trailingIndex = + parseTagString( + localeID, + lang, + &langLength, + script, + &scriptLength, + region, + ®ionLength, + err); + if(U_FAILURE(*err)) { + + /* Overflow indicates an illegal argument error */ + if (*err == U_BUFFER_OVERFLOW_ERROR) { + *err = U_ILLEGAL_ARGUMENT_ERROR; + } + + goto error; + } + + /* Find the spot where the variants or the keywords begin, if any. */ + while (_isIDSeparator(localeID[trailingIndex])) { + trailingIndex++; + } + trailing = &localeID[trailingIndex]; + trailingLength = (int32_t)uprv_strlen(trailing); + + CHECK_TRAILING_VARIANT_SIZE(trailing, trailingLength); + + { + icu::CharString base; + { + icu::CharStringByteSink baseSink(&base); + createTagString( + lang, + langLength, + script, + scriptLength, + region, + regionLength, + nullptr, + 0, + baseSink, + err); + } + + /** + * First, we need to first get the maximization + * from AddLikelySubtags. + **/ + { + icu::CharStringByteSink maxSink(&maximizedTagBuffer); + successGetMax = _ulocimp_addLikelySubtags(base.data(), maxSink, err); + } + } + + if(U_FAILURE(*err)) { + goto error; + } + + if (!successGetMax) { + /** + * If we got here, return the locale ID parameter unchanged. + **/ + const int32_t localeIDLength = (int32_t)uprv_strlen(localeID); + sink.Append(localeID, localeIDLength); + return; + } + + // In the following, the lang, script, region are referring to those in + // the maximizedTagBuffer, not the one in the localeID. + langLength = sizeof(lang); + scriptLength = sizeof(script); + regionLength = sizeof(region); + parseTagString( + maximizedTagBuffer.data(), + lang, + &langLength, + script, + &scriptLength, + region, + ®ionLength, + err); + if(U_FAILURE(*err)) { + goto error; + } + + /** + * Start first with just the language. + **/ + { + icu::CharString tagBuffer; + { + icu::CharStringByteSink tagSink(&tagBuffer); + createLikelySubtagsString( + lang, + langLength, + nullptr, + 0, + nullptr, + 0, + nullptr, + 0, + tagSink, + err); + } + + if(U_FAILURE(*err)) { + goto error; + } + else if (!tagBuffer.isEmpty() && + uprv_strnicmp( + maximizedTagBuffer.data(), + tagBuffer.data(), + tagBuffer.length()) == 0) { + + createTagString( + lang, + langLength, + nullptr, + 0, + nullptr, + 0, + trailing, + trailingLength, + sink, + err); + return; + } + } + + /** + * Next, try the language and region. + **/ + if (regionLength > 0) { + + icu::CharString tagBuffer; + { + icu::CharStringByteSink tagSink(&tagBuffer); + createLikelySubtagsString( + lang, + langLength, + nullptr, + 0, + region, + regionLength, + nullptr, + 0, + tagSink, + err); + } + + if(U_FAILURE(*err)) { + goto error; + } + else if (!tagBuffer.isEmpty() && + uprv_strnicmp( + maximizedTagBuffer.data(), + tagBuffer.data(), + tagBuffer.length()) == 0) { + + createTagString( + lang, + langLength, + nullptr, + 0, + region, + regionLength, + trailing, + trailingLength, + sink, + err); + return; + } + } + + /** + * Finally, try the language and script. This is our last chance, + * since trying with all three subtags would only yield the + * maximal version that we already have. + **/ + if (scriptLength > 0) { + icu::CharString tagBuffer; + { + icu::CharStringByteSink tagSink(&tagBuffer); + createLikelySubtagsString( + lang, + langLength, + script, + scriptLength, + nullptr, + 0, + nullptr, + 0, + tagSink, + err); + } + + if(U_FAILURE(*err)) { + goto error; + } + else if (!tagBuffer.isEmpty() && + uprv_strnicmp( + maximizedTagBuffer.data(), + tagBuffer.data(), + tagBuffer.length()) == 0) { + + createTagString( + lang, + langLength, + script, + scriptLength, + nullptr, + 0, + trailing, + trailingLength, + sink, + err); + return; + } + } + + { + /** + * If we got here, return the max + trail. + **/ + createTagString( + lang, + langLength, + script, + scriptLength, + region, + regionLength, + trailing, + trailingLength, + sink, + err); + return; + } + +error: + + if (!U_FAILURE(*err)) { + *err = U_ILLEGAL_ARGUMENT_ERROR; + } +} + +static int32_t +do_canonicalize(const char* localeID, + char* buffer, + int32_t bufferCapacity, + UErrorCode* err) +{ + int32_t canonicalizedSize = uloc_canonicalize( + localeID, + buffer, + bufferCapacity, + err); + + if (*err == U_STRING_NOT_TERMINATED_WARNING || + *err == U_BUFFER_OVERFLOW_ERROR) { + return canonicalizedSize; + } + else if (U_FAILURE(*err)) { + + return -1; + } + else { + return canonicalizedSize; + } +} + +U_CAPI int32_t U_EXPORT2 +uloc_addLikelySubtags(const char* localeID, + char* maximizedLocaleID, + int32_t maximizedLocaleIDCapacity, + UErrorCode* status) { + if (U_FAILURE(*status)) { + return 0; + } + + icu::CheckedArrayByteSink sink( + maximizedLocaleID, maximizedLocaleIDCapacity); + + ulocimp_addLikelySubtags(localeID, sink, status); + int32_t reslen = sink.NumberOfBytesAppended(); + + if (U_FAILURE(*status)) { + return sink.Overflowed() ? reslen : -1; + } + + if (sink.Overflowed()) { + *status = U_BUFFER_OVERFLOW_ERROR; + } else { + u_terminateChars( + maximizedLocaleID, maximizedLocaleIDCapacity, reslen, status); + } + + return reslen; +} + +static UBool +_ulocimp_addLikelySubtags(const char* localeID, + icu::ByteSink& sink, + UErrorCode* status) { + PreflightingLocaleIDBuffer localeBuffer; + do { + localeBuffer.requestedCapacity = do_canonicalize(localeID, localeBuffer.getBuffer(), + localeBuffer.getCapacity(), status); + } while (localeBuffer.needToTryAgain(status)); + + if (U_SUCCESS(*status)) { + return _uloc_addLikelySubtags(localeBuffer.getBuffer(), sink, status); + } else { + return false; + } +} + +U_CAPI void U_EXPORT2 +ulocimp_addLikelySubtags(const char* localeID, + icu::ByteSink& sink, + UErrorCode* status) { + _ulocimp_addLikelySubtags(localeID, sink, status); +} + +U_CAPI int32_t U_EXPORT2 +uloc_minimizeSubtags(const char* localeID, + char* minimizedLocaleID, + int32_t minimizedLocaleIDCapacity, + UErrorCode* status) { + if (U_FAILURE(*status)) { + return 0; + } + + icu::CheckedArrayByteSink sink( + minimizedLocaleID, minimizedLocaleIDCapacity); + + ulocimp_minimizeSubtags(localeID, sink, status); + int32_t reslen = sink.NumberOfBytesAppended(); + + if (U_FAILURE(*status)) { + return sink.Overflowed() ? reslen : -1; + } + + if (sink.Overflowed()) { + *status = U_BUFFER_OVERFLOW_ERROR; + } else { + u_terminateChars( + minimizedLocaleID, minimizedLocaleIDCapacity, reslen, status); + } + + return reslen; +} + +U_CAPI void U_EXPORT2 +ulocimp_minimizeSubtags(const char* localeID, + icu::ByteSink& sink, + UErrorCode* status) { + PreflightingLocaleIDBuffer localeBuffer; + do { + localeBuffer.requestedCapacity = do_canonicalize(localeID, localeBuffer.getBuffer(), + localeBuffer.getCapacity(), status); + } while (localeBuffer.needToTryAgain(status)); + + _uloc_minimizeSubtags(localeBuffer.getBuffer(), sink, status); +} + +// Pairs of (language subtag, + or -) for finding out fast if common languages +// are LTR (minus) or RTL (plus). +static const char LANG_DIR_STRING[] = + "root-en-es-pt-zh-ja-ko-de-fr-it-ar+he+fa+ru-nl-pl-th-tr-"; + +// Implemented here because this calls ulocimp_addLikelySubtags(). +U_CAPI UBool U_EXPORT2 +uloc_isRightToLeft(const char *locale) { + UErrorCode errorCode = U_ZERO_ERROR; + char script[8]; + int32_t scriptLength = uloc_getScript(locale, script, UPRV_LENGTHOF(script), &errorCode); + if (U_FAILURE(errorCode) || errorCode == U_STRING_NOT_TERMINATED_WARNING || + scriptLength == 0) { + // Fastpath: We know the likely scripts and their writing direction + // for some common languages. + errorCode = U_ZERO_ERROR; + char lang[8]; + int32_t langLength = uloc_getLanguage(locale, lang, UPRV_LENGTHOF(lang), &errorCode); + if (U_FAILURE(errorCode) || errorCode == U_STRING_NOT_TERMINATED_WARNING) { + return false; + } + if (langLength > 0) { + const char* langPtr = uprv_strstr(LANG_DIR_STRING, lang); + if (langPtr != nullptr) { + switch (langPtr[langLength]) { + case '-': return false; + case '+': return true; + default: break; // partial match of a longer code + } + } + } + // Otherwise, find the likely script. + errorCode = U_ZERO_ERROR; + icu::CharString likely; + { + icu::CharStringByteSink sink(&likely); + ulocimp_addLikelySubtags(locale, sink, &errorCode); + } + if (U_FAILURE(errorCode) || errorCode == U_STRING_NOT_TERMINATED_WARNING) { + return false; + } + scriptLength = uloc_getScript(likely.data(), script, UPRV_LENGTHOF(script), &errorCode); + if (U_FAILURE(errorCode) || errorCode == U_STRING_NOT_TERMINATED_WARNING || + scriptLength == 0) { + return false; + } + } + UScriptCode scriptCode = (UScriptCode)u_getPropertyValueEnum(UCHAR_SCRIPT, script); + return uscript_isRightToLeft(scriptCode); +} + +U_NAMESPACE_BEGIN + +UBool +Locale::isRightToLeft() const { + return uloc_isRightToLeft(getBaseName()); +} + +U_NAMESPACE_END + +// The following must at least allow for rg key value (6) plus terminator (1). +#define ULOC_RG_BUFLEN 8 + +U_CAPI int32_t U_EXPORT2 +ulocimp_getRegionForSupplementalData(const char *localeID, UBool inferRegion, + char *region, int32_t regionCapacity, UErrorCode* status) { + if (U_FAILURE(*status)) { + return 0; + } + char rgBuf[ULOC_RG_BUFLEN]; + UErrorCode rgStatus = U_ZERO_ERROR; + + // First check for rg keyword value + int32_t rgLen = uloc_getKeywordValue(localeID, "rg", rgBuf, ULOC_RG_BUFLEN, &rgStatus); + if (U_FAILURE(rgStatus) || rgLen != 6) { + rgLen = 0; + } else { + // rgBuf guaranteed to be zero terminated here, with text len 6 + char *rgPtr = rgBuf; + for (; *rgPtr!= 0; rgPtr++) { + *rgPtr = uprv_toupper(*rgPtr); + } + rgLen = (uprv_strcmp(rgBuf+2, "ZZZZ") == 0)? 2: 0; + } + + if (rgLen == 0) { + // No valid rg keyword value, try for unicode_region_subtag + rgLen = uloc_getCountry(localeID, rgBuf, ULOC_RG_BUFLEN, status); + if (U_FAILURE(*status)) { + rgLen = 0; + } else if (rgLen == 0 && inferRegion) { + // no unicode_region_subtag but inferRegion true, try likely subtags + rgStatus = U_ZERO_ERROR; + icu::CharString locBuf; + { + icu::CharStringByteSink sink(&locBuf); + ulocimp_addLikelySubtags(localeID, sink, &rgStatus); + } + if (U_SUCCESS(rgStatus)) { + rgLen = uloc_getCountry(locBuf.data(), rgBuf, ULOC_RG_BUFLEN, status); + if (U_FAILURE(*status)) { + rgLen = 0; + } + } + } + } + + rgBuf[rgLen] = 0; + uprv_strncpy(region, rgBuf, regionCapacity); + return u_terminateChars(region, regionCapacity, rgLen, status); +} + diff --git a/intl/icu/source/common/loclikelysubtags.cpp b/intl/icu/source/common/loclikelysubtags.cpp new file mode 100644 index 0000000000..e913c66a35 --- /dev/null +++ b/intl/icu/source/common/loclikelysubtags.cpp @@ -0,0 +1,683 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// loclikelysubtags.cpp +// created: 2019may08 Markus W. Scherer + +#include +#include "unicode/utypes.h" +#include "unicode/bytestrie.h" +#include "unicode/localpointer.h" +#include "unicode/locid.h" +#include "unicode/uobject.h" +#include "unicode/ures.h" +#include "charstr.h" +#include "cstring.h" +#include "loclikelysubtags.h" +#include "lsr.h" +#include "uassert.h" +#include "ucln_cmn.h" +#include "uhash.h" +#include "uinvchar.h" +#include "umutex.h" +#include "uniquecharstr.h" +#include "uresdata.h" +#include "uresimp.h" + +U_NAMESPACE_BEGIN + +namespace { + +constexpr char PSEUDO_ACCENTS_PREFIX = '\''; // -XA, -PSACCENT +constexpr char PSEUDO_BIDI_PREFIX = '+'; // -XB, -PSBIDI +constexpr char PSEUDO_CRACKED_PREFIX = ','; // -XC, -PSCRACK + +} // namespace + +LocaleDistanceData::LocaleDistanceData(LocaleDistanceData &&data) : + distanceTrieBytes(data.distanceTrieBytes), + regionToPartitions(data.regionToPartitions), + partitions(data.partitions), + paradigms(data.paradigms), paradigmsLength(data.paradigmsLength), + distances(data.distances) { + data.partitions = nullptr; + data.paradigms = nullptr; +} + +LocaleDistanceData::~LocaleDistanceData() { + uprv_free(partitions); + delete[] paradigms; +} + +// TODO(ICU-20777): Rename to just LikelySubtagsData. +struct XLikelySubtagsData { + UResourceBundle *langInfoBundle = nullptr; + UniqueCharStrings strings; + CharStringMap languageAliases; + CharStringMap regionAliases; + const uint8_t *trieBytes = nullptr; + LSR *lsrs = nullptr; + int32_t lsrsLength = 0; + + LocaleDistanceData distanceData; + + XLikelySubtagsData(UErrorCode &errorCode) : strings(errorCode) {} + + ~XLikelySubtagsData() { + ures_close(langInfoBundle); + delete[] lsrs; + } + + void load(UErrorCode &errorCode) { + langInfoBundle = ures_openDirect(nullptr, "langInfo", &errorCode); + if (U_FAILURE(errorCode)) { return; } + StackUResourceBundle stackTempBundle; + ResourceDataValue value; + ures_getValueWithFallback(langInfoBundle, "likely", stackTempBundle.getAlias(), + value, errorCode); + ResourceTable likelyTable = value.getTable(errorCode); + if (U_FAILURE(errorCode)) { return; } + + // Read all strings in the resource bundle and convert them to invariant char *. + LocalMemory languageIndexes, regionIndexes, lsrSubtagIndexes; + int32_t languagesLength = 0, regionsLength = 0, lsrSubtagsLength = 0; + if (!readStrings(likelyTable, "languageAliases", value, + languageIndexes, languagesLength, errorCode) || + !readStrings(likelyTable, "regionAliases", value, + regionIndexes, regionsLength, errorCode) || + !readStrings(likelyTable, "lsrs", value, + lsrSubtagIndexes,lsrSubtagsLength, errorCode)) { + return; + } + if ((languagesLength & 1) != 0 || + (regionsLength & 1) != 0 || + (lsrSubtagsLength % 3) != 0) { + errorCode = U_INVALID_FORMAT_ERROR; + return; + } + if (lsrSubtagsLength == 0) { + errorCode = U_MISSING_RESOURCE_ERROR; + return; + } + + if (!likelyTable.findValue("trie", value)) { + errorCode = U_MISSING_RESOURCE_ERROR; + return; + } + int32_t length; + trieBytes = value.getBinary(length, errorCode); + if (U_FAILURE(errorCode)) { return; } + + // Also read distance/matcher data if available, + // to open & keep only one resource bundle pointer + // and to use one single UniqueCharStrings. + UErrorCode matchErrorCode = U_ZERO_ERROR; + ures_getValueWithFallback(langInfoBundle, "match", stackTempBundle.getAlias(), + value, matchErrorCode); + LocalMemory partitionIndexes, paradigmSubtagIndexes; + int32_t partitionsLength = 0, paradigmSubtagsLength = 0; + if (U_SUCCESS(matchErrorCode)) { + ResourceTable matchTable = value.getTable(errorCode); + if (U_FAILURE(errorCode)) { return; } + + if (matchTable.findValue("trie", value)) { + distanceData.distanceTrieBytes = value.getBinary(length, errorCode); + if (U_FAILURE(errorCode)) { return; } + } + + if (matchTable.findValue("regionToPartitions", value)) { + distanceData.regionToPartitions = value.getBinary(length, errorCode); + if (U_FAILURE(errorCode)) { return; } + if (length < LSR::REGION_INDEX_LIMIT) { + errorCode = U_INVALID_FORMAT_ERROR; + return; + } + } + + if (!readStrings(matchTable, "partitions", value, + partitionIndexes, partitionsLength, errorCode) || + !readStrings(matchTable, "paradigms", value, + paradigmSubtagIndexes, paradigmSubtagsLength, errorCode)) { + return; + } + if ((paradigmSubtagsLength % 3) != 0) { + errorCode = U_INVALID_FORMAT_ERROR; + return; + } + + if (matchTable.findValue("distances", value)) { + distanceData.distances = value.getIntVector(length, errorCode); + if (U_FAILURE(errorCode)) { return; } + if (length < 4) { // LocaleDistance IX_LIMIT + errorCode = U_INVALID_FORMAT_ERROR; + return; + } + } + } else if (matchErrorCode == U_MISSING_RESOURCE_ERROR) { + // ok for likely subtags + } else { // error other than missing resource + errorCode = matchErrorCode; + return; + } + + // Fetch & store invariant-character versions of strings + // only after we have collected and de-duplicated all of them. + strings.freeze(); + + languageAliases = CharStringMap(languagesLength / 2, errorCode); + for (int32_t i = 0; i < languagesLength; i += 2) { + languageAliases.put(strings.get(languageIndexes[i]), + strings.get(languageIndexes[i + 1]), errorCode); + } + + regionAliases = CharStringMap(regionsLength / 2, errorCode); + for (int32_t i = 0; i < regionsLength; i += 2) { + regionAliases.put(strings.get(regionIndexes[i]), + strings.get(regionIndexes[i + 1]), errorCode); + } + if (U_FAILURE(errorCode)) { return; } + + lsrsLength = lsrSubtagsLength / 3; + lsrs = new LSR[lsrsLength]; + if (lsrs == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + for (int32_t i = 0, j = 0; i < lsrSubtagsLength; i += 3, ++j) { + lsrs[j] = LSR(strings.get(lsrSubtagIndexes[i]), + strings.get(lsrSubtagIndexes[i + 1]), + strings.get(lsrSubtagIndexes[i + 2]), + LSR::IMPLICIT_LSR); + } + + if (partitionsLength > 0) { + distanceData.partitions = static_cast( + uprv_malloc(partitionsLength * sizeof(const char *))); + if (distanceData.partitions == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + for (int32_t i = 0; i < partitionsLength; ++i) { + distanceData.partitions[i] = strings.get(partitionIndexes[i]); + } + } + + if (paradigmSubtagsLength > 0) { + distanceData.paradigmsLength = paradigmSubtagsLength / 3; + LSR *paradigms = new LSR[distanceData.paradigmsLength]; + if (paradigms == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + for (int32_t i = 0, j = 0; i < paradigmSubtagsLength; i += 3, ++j) { + paradigms[j] = LSR(strings.get(paradigmSubtagIndexes[i]), + strings.get(paradigmSubtagIndexes[i + 1]), + strings.get(paradigmSubtagIndexes[i + 2]), + LSR::DONT_CARE_FLAGS); + } + distanceData.paradigms = paradigms; + } + } + +private: + bool readStrings(const ResourceTable &table, const char *key, ResourceValue &value, + LocalMemory &indexes, int32_t &length, UErrorCode &errorCode) { + if (table.findValue(key, value)) { + ResourceArray stringArray = value.getArray(errorCode); + if (U_FAILURE(errorCode)) { return false; } + length = stringArray.getSize(); + if (length == 0) { return true; } + int32_t *rawIndexes = indexes.allocateInsteadAndCopy(length); + if (rawIndexes == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return false; + } + for (int i = 0; i < length; ++i) { + stringArray.getValue(i, value); // returns true because i < length + rawIndexes[i] = strings.add(value.getUnicodeString(errorCode), errorCode); + if (U_FAILURE(errorCode)) { return false; } + } + } + return true; + } +}; + +namespace { + +XLikelySubtags *gLikelySubtags = nullptr; +UInitOnce gInitOnce {}; + +UBool U_CALLCONV cleanup() { + delete gLikelySubtags; + gLikelySubtags = nullptr; + gInitOnce.reset(); + return true; +} + +} // namespace + +void U_CALLCONV XLikelySubtags::initLikelySubtags(UErrorCode &errorCode) { + // This function is invoked only via umtx_initOnce(). + U_ASSERT(gLikelySubtags == nullptr); + XLikelySubtagsData data(errorCode); + data.load(errorCode); + if (U_FAILURE(errorCode)) { return; } + gLikelySubtags = new XLikelySubtags(data); + if (gLikelySubtags == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + ucln_common_registerCleanup(UCLN_COMMON_LIKELY_SUBTAGS, cleanup); +} + +const XLikelySubtags *XLikelySubtags::getSingleton(UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return nullptr; } + umtx_initOnce(gInitOnce, &XLikelySubtags::initLikelySubtags, errorCode); + return gLikelySubtags; +} + +XLikelySubtags::XLikelySubtags(XLikelySubtagsData &data) : + langInfoBundle(data.langInfoBundle), + strings(data.strings.orphanCharStrings()), + languageAliases(std::move(data.languageAliases)), + regionAliases(std::move(data.regionAliases)), + trie(data.trieBytes), + lsrs(data.lsrs), +#if U_DEBUG + lsrsLength(data.lsrsLength), +#endif + distanceData(std::move(data.distanceData)) { + data.langInfoBundle = nullptr; + data.lsrs = nullptr; + + // Cache the result of looking up language="und" encoded as "*", and "und-Zzzz" ("**"). + UStringTrieResult result = trie.next(u'*'); + U_ASSERT(USTRINGTRIE_HAS_NEXT(result)); + trieUndState = trie.getState64(); + result = trie.next(u'*'); + U_ASSERT(USTRINGTRIE_HAS_NEXT(result)); + trieUndZzzzState = trie.getState64(); + result = trie.next(u'*'); + U_ASSERT(USTRINGTRIE_HAS_VALUE(result)); + defaultLsrIndex = trie.getValue(); + trie.reset(); + + for (char16_t c = u'a'; c <= u'z'; ++c) { + result = trie.next(c); + if (result == USTRINGTRIE_NO_VALUE) { + trieFirstLetterStates[c - u'a'] = trie.getState64(); + } + trie.reset(); + } +} + +XLikelySubtags::~XLikelySubtags() { + ures_close(langInfoBundle); + delete strings; + delete[] lsrs; +} + +LSR XLikelySubtags::makeMaximizedLsrFrom(const Locale &locale, UErrorCode &errorCode) const { + const char *name = locale.getName(); + if (uprv_isAtSign(name[0]) && name[1] == 'x' && name[2] == '=') { // name.startsWith("@x=") + // Private use language tag x-subtag-subtag... which CLDR changes to + // und-x-subtag-subtag... + return LSR(name, "", "", LSR::EXPLICIT_LSR); + } + return makeMaximizedLsr(locale.getLanguage(), locale.getScript(), locale.getCountry(), + locale.getVariant(), errorCode); +} + +namespace { + +const char *getCanonical(const CharStringMap &aliases, const char *alias) { + const char *canonical = aliases.get(alias); + return canonical == nullptr ? alias : canonical; +} + +} // namespace + +LSR XLikelySubtags::makeMaximizedLsr(const char *language, const char *script, const char *region, + const char *variant, UErrorCode &errorCode) const { + // Handle pseudolocales like en-XA, ar-XB, fr-PSCRACK. + // They should match only themselves, + // not other locales with what looks like the same language and script subtags. + char c1; + if (region[0] == 'X' && (c1 = region[1]) != 0 && region[2] == 0) { + switch (c1) { + case 'A': + return LSR(PSEUDO_ACCENTS_PREFIX, language, script, region, + LSR::EXPLICIT_LSR, errorCode); + case 'B': + return LSR(PSEUDO_BIDI_PREFIX, language, script, region, + LSR::EXPLICIT_LSR, errorCode); + case 'C': + return LSR(PSEUDO_CRACKED_PREFIX, language, script, region, + LSR::EXPLICIT_LSR, errorCode); + default: // normal locale + break; + } + } + + if (variant[0] == 'P' && variant[1] == 'S') { + int32_t lsrFlags = *region == 0 ? + LSR::EXPLICIT_LANGUAGE | LSR::EXPLICIT_SCRIPT : LSR::EXPLICIT_LSR; + if (uprv_strcmp(variant, "PSACCENT") == 0) { + return LSR(PSEUDO_ACCENTS_PREFIX, language, script, + *region == 0 ? "XA" : region, lsrFlags, errorCode); + } else if (uprv_strcmp(variant, "PSBIDI") == 0) { + return LSR(PSEUDO_BIDI_PREFIX, language, script, + *region == 0 ? "XB" : region, lsrFlags, errorCode); + } else if (uprv_strcmp(variant, "PSCRACK") == 0) { + return LSR(PSEUDO_CRACKED_PREFIX, language, script, + *region == 0 ? "XC" : region, lsrFlags, errorCode); + } + // else normal locale + } + + language = getCanonical(languageAliases, language); + // (We have no script mappings.) + region = getCanonical(regionAliases, region); + return maximize(language, script, region); +} + +LSR XLikelySubtags::maximize(const char *language, const char *script, const char *region) const { + if (uprv_strcmp(language, "und") == 0) { + language = ""; + } + if (uprv_strcmp(script, "Zzzz") == 0) { + script = ""; + } + if (uprv_strcmp(region, "ZZ") == 0) { + region = ""; + } + if (*script != 0 && *region != 0 && *language != 0) { + return LSR(language, script, region, LSR::EXPLICIT_LSR); // already maximized + } + + uint32_t retainOldMask = 0; + BytesTrie iter(trie); + uint64_t state; + int32_t value; + // Small optimization: Array lookup for first language letter. + int32_t c0; + if (0 <= (c0 = uprv_lowerOrdinal(language[0])) && c0 <= 25 && + language[1] != 0 && // language.length() >= 2 + (state = trieFirstLetterStates[c0]) != 0) { + value = trieNext(iter.resetToState64(state), language, 1); + } else { + value = trieNext(iter, language, 0); + } + if (value >= 0) { + if (*language != 0) { + retainOldMask |= 4; + } + state = iter.getState64(); + } else { + retainOldMask |= 4; + iter.resetToState64(trieUndState); // "und" ("*") + state = 0; + } + + if (value > 0) { + // Intermediate or final value from just language. + if (value == SKIP_SCRIPT) { + value = 0; + } + if (*script != 0) { + retainOldMask |= 2; + } + } else { + value = trieNext(iter, script, 0); + if (value >= 0) { + if (*script != 0) { + retainOldMask |= 2; + } + state = iter.getState64(); + } else { + retainOldMask |= 2; + if (state == 0) { + iter.resetToState64(trieUndZzzzState); // "und-Zzzz" ("**") + } else { + iter.resetToState64(state); + value = trieNext(iter, "", 0); + U_ASSERT(value >= 0); + state = iter.getState64(); + } + } + } + + if (value > 0) { + // Final value from just language or language+script. + if (*region != 0) { + retainOldMask |= 1; + } + } else { + value = trieNext(iter, region, 0); + if (value >= 0) { + if (*region != 0) { + retainOldMask |= 1; + } + } else { + retainOldMask |= 1; + if (state == 0) { + value = defaultLsrIndex; + } else { + iter.resetToState64(state); + value = trieNext(iter, "", 0); + U_ASSERT(value > 0); + } + } + } + U_ASSERT(value < lsrsLength); + const LSR &result = lsrs[value]; + + if (*language == 0) { + language = "und"; + } + + if (retainOldMask == 0) { + // Quickly return a copy of the lookup-result LSR + // without new allocation of the subtags. + return LSR(result.language, result.script, result.region, result.flags); + } + if ((retainOldMask & 4) == 0) { + language = result.language; + } + if ((retainOldMask & 2) == 0) { + script = result.script; + } + if ((retainOldMask & 1) == 0) { + region = result.region; + } + // retainOldMask flags = LSR explicit-subtag flags + return LSR(language, script, region, retainOldMask); +} + +int32_t XLikelySubtags::compareLikely(const LSR &lsr, const LSR &other, int32_t likelyInfo) const { + // If likelyInfo >= 0: + // likelyInfo bit 1 is set if the previous comparison with lsr + // was for equal language and script. + // Otherwise the scripts differed. + if (uprv_strcmp(lsr.language, other.language) != 0) { + return 0xfffffffc; // negative, lsr not better than other + } + if (uprv_strcmp(lsr.script, other.script) != 0) { + int32_t index; + if (likelyInfo >= 0 && (likelyInfo & 2) == 0) { + index = likelyInfo >> 2; + } else { + index = getLikelyIndex(lsr.language, ""); + likelyInfo = index << 2; + } + const LSR &likely = lsrs[index]; + if (uprv_strcmp(lsr.script, likely.script) == 0) { + return likelyInfo | 1; + } else { + return likelyInfo & ~1; + } + } + if (uprv_strcmp(lsr.region, other.region) != 0) { + int32_t index; + if (likelyInfo >= 0 && (likelyInfo & 2) != 0) { + index = likelyInfo >> 2; + } else { + index = getLikelyIndex(lsr.language, lsr.region); + likelyInfo = (index << 2) | 2; + } + const LSR &likely = lsrs[index]; + if (uprv_strcmp(lsr.region, likely.region) == 0) { + return likelyInfo | 1; + } else { + return likelyInfo & ~1; + } + } + return likelyInfo & ~1; // lsr not better than other +} + +// Subset of maximize(). +int32_t XLikelySubtags::getLikelyIndex(const char *language, const char *script) const { + if (uprv_strcmp(language, "und") == 0) { + language = ""; + } + if (uprv_strcmp(script, "Zzzz") == 0) { + script = ""; + } + + BytesTrie iter(trie); + uint64_t state; + int32_t value; + // Small optimization: Array lookup for first language letter. + int32_t c0; + if (0 <= (c0 = uprv_lowerOrdinal(language[0])) && c0 <= 25 && + language[1] != 0 && // language.length() >= 2 + (state = trieFirstLetterStates[c0]) != 0) { + value = trieNext(iter.resetToState64(state), language, 1); + } else { + value = trieNext(iter, language, 0); + } + if (value >= 0) { + state = iter.getState64(); + } else { + iter.resetToState64(trieUndState); // "und" ("*") + state = 0; + } + + if (value > 0) { + // Intermediate or final value from just language. + if (value == SKIP_SCRIPT) { + value = 0; + } + } else { + value = trieNext(iter, script, 0); + if (value >= 0) { + state = iter.getState64(); + } else { + if (state == 0) { + iter.resetToState64(trieUndZzzzState); // "und-Zzzz" ("**") + } else { + iter.resetToState64(state); + value = trieNext(iter, "", 0); + U_ASSERT(value >= 0); + state = iter.getState64(); + } + } + } + + if (value > 0) { + // Final value from just language or language+script. + } else { + value = trieNext(iter, "", 0); + U_ASSERT(value > 0); + } + U_ASSERT(value < lsrsLength); + return value; +} + +int32_t XLikelySubtags::trieNext(BytesTrie &iter, const char *s, int32_t i) { + UStringTrieResult result; + uint8_t c; + if ((c = s[i]) == 0) { + result = iter.next(u'*'); + } else { + for (;;) { + c = uprv_invCharToAscii(c); + // EBCDIC: If s[i] is not an invariant character, + // then c is now 0 and will simply not match anything, which is harmless. + uint8_t next = s[++i]; + if (next != 0) { + if (!USTRINGTRIE_HAS_NEXT(iter.next(c))) { + return -1; + } + } else { + // last character of this subtag + result = iter.next(c | 0x80); + break; + } + c = next; + } + } + switch (result) { + case USTRINGTRIE_NO_MATCH: return -1; + case USTRINGTRIE_NO_VALUE: return 0; + case USTRINGTRIE_INTERMEDIATE_VALUE: + U_ASSERT(iter.getValue() == SKIP_SCRIPT); + return SKIP_SCRIPT; + case USTRINGTRIE_FINAL_VALUE: return iter.getValue(); + default: return -1; + } +} + +// TODO(ICU-20777): Switch Locale/uloc_ likely-subtags API from the old code +// in loclikely.cpp to this new code, including activating this +// minimizeSubtags() function. The LocaleMatcher does not minimize. +#if 0 +LSR XLikelySubtags::minimizeSubtags(const char *languageIn, const char *scriptIn, + const char *regionIn, ULocale.Minimize fieldToFavor, + UErrorCode &errorCode) const { + LSR result = maximize(languageIn, scriptIn, regionIn); + + // We could try just a series of checks, like: + // LSR result2 = addLikelySubtags(languageIn, "", ""); + // if result.equals(result2) return result2; + // However, we can optimize 2 of the cases: + // (languageIn, "", "") + // (languageIn, "", regionIn) + + // value00 = lookup(result.language, "", "") + BytesTrie iter = new BytesTrie(trie); + int value = trieNext(iter, result.language, 0); + U_ASSERT(value >= 0); + if (value == 0) { + value = trieNext(iter, "", 0); + U_ASSERT(value >= 0); + if (value == 0) { + value = trieNext(iter, "", 0); + } + } + U_ASSERT(value > 0); + LSR value00 = lsrs[value]; + boolean favorRegionOk = false; + if (result.script.equals(value00.script)) { //script is default + if (result.region.equals(value00.region)) { + return new LSR(result.language, "", "", LSR.DONT_CARE_FLAGS); + } else if (fieldToFavor == ULocale.Minimize.FAVOR_REGION) { + return new LSR(result.language, "", result.region, LSR.DONT_CARE_FLAGS); + } else { + favorRegionOk = true; + } + } + + // The last case is not as easy to optimize. + // Maybe do later, but for now use the straightforward code. + LSR result2 = maximize(languageIn, scriptIn, ""); + if (result2.equals(result)) { + return new LSR(result.language, result.script, "", LSR.DONT_CARE_FLAGS); + } else if (favorRegionOk) { + return new LSR(result.language, "", result.region, LSR.DONT_CARE_FLAGS); + } + return result; +} +#endif + +U_NAMESPACE_END diff --git a/intl/icu/source/common/loclikelysubtags.h b/intl/icu/source/common/loclikelysubtags.h new file mode 100644 index 0000000000..14a01a5eac --- /dev/null +++ b/intl/icu/source/common/loclikelysubtags.h @@ -0,0 +1,121 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// loclikelysubtags.h +// created: 2019may08 Markus W. Scherer + +#ifndef __LOCLIKELYSUBTAGS_H__ +#define __LOCLIKELYSUBTAGS_H__ + +#include +#include "unicode/utypes.h" +#include "unicode/bytestrie.h" +#include "unicode/locid.h" +#include "unicode/uobject.h" +#include "unicode/ures.h" +#include "charstrmap.h" +#include "lsr.h" + +U_NAMESPACE_BEGIN + +struct XLikelySubtagsData; + +struct LocaleDistanceData { + LocaleDistanceData() = default; + LocaleDistanceData(LocaleDistanceData &&data); + ~LocaleDistanceData(); + + const uint8_t *distanceTrieBytes = nullptr; + const uint8_t *regionToPartitions = nullptr; + const char **partitions = nullptr; + const LSR *paradigms = nullptr; + int32_t paradigmsLength = 0; + const int32_t *distances = nullptr; + +private: + LocaleDistanceData &operator=(const LocaleDistanceData &) = delete; +}; + +// TODO(ICU-20777): Rename to just LikelySubtags. +class XLikelySubtags final : public UMemory { +public: + ~XLikelySubtags(); + + static constexpr int32_t SKIP_SCRIPT = 1; + + // VisibleForTesting + static const XLikelySubtags *getSingleton(UErrorCode &errorCode); + + // VisibleForTesting + LSR makeMaximizedLsrFrom(const Locale &locale, UErrorCode &errorCode) const; + + /** + * Tests whether lsr is "more likely" than other. + * For example, fr-Latn-FR is more likely than fr-Latn-CH because + * FR is the default region for fr-Latn. + * + * The likelyInfo caches lookup information between calls. + * The return value is an updated likelyInfo value, + * with bit 0 set if lsr is "more likely". + * The initial value of likelyInfo must be negative. + */ + int32_t compareLikely(const LSR &lsr, const LSR &other, int32_t likelyInfo) const; + + // TODO(ICU-20777): Switch Locale/uloc_ likely-subtags API from the old code + // in loclikely.cpp to this new code, including activating this + // minimizeSubtags() function. The LocaleMatcher does not minimize. +#if 0 + LSR minimizeSubtags(const char *languageIn, const char *scriptIn, const char *regionIn, + ULocale.Minimize fieldToFavor, UErrorCode &errorCode) const; +#endif + + // visible for LocaleDistance + const LocaleDistanceData &getDistanceData() const { return distanceData; } + +private: + XLikelySubtags(XLikelySubtagsData &data); + XLikelySubtags(const XLikelySubtags &other) = delete; + XLikelySubtags &operator=(const XLikelySubtags &other) = delete; + + static void initLikelySubtags(UErrorCode &errorCode); + + LSR makeMaximizedLsr(const char *language, const char *script, const char *region, + const char *variant, UErrorCode &errorCode) const; + + /** + * Raw access to addLikelySubtags. Input must be in canonical format, eg "en", not "eng" or "EN". + */ + LSR maximize(const char *language, const char *script, const char *region) const; + + int32_t getLikelyIndex(const char *language, const char *script) const; + + static int32_t trieNext(BytesTrie &iter, const char *s, int32_t i); + + UResourceBundle *langInfoBundle; + // We could store the strings by value, except that if there were few enough strings, + // moving the contents could copy it to a different array, + // invalidating the pointers stored in the maps. + CharString *strings; + CharStringMap languageAliases; + CharStringMap regionAliases; + + // The trie maps each lang+script+region (encoded in ASCII) to an index into lsrs. + // There is also a trie value for each intermediate lang and lang+script. + // '*' is used instead of "und", "Zzzz"/"" and "ZZ"/"". + BytesTrie trie; + uint64_t trieUndState; + uint64_t trieUndZzzzState; + int32_t defaultLsrIndex; + uint64_t trieFirstLetterStates[26]; + const LSR *lsrs; +#if U_DEBUG + int32_t lsrsLength; +#endif + + // distance/matcher data: see comment in XLikelySubtagsData::load() + LocaleDistanceData distanceData; +}; + +U_NAMESPACE_END + +#endif // __LOCLIKELYSUBTAGS_H__ diff --git a/intl/icu/source/common/locmap.cpp b/intl/icu/source/common/locmap.cpp new file mode 100644 index 0000000000..7a0e90e8bd --- /dev/null +++ b/intl/icu/source/common/locmap.cpp @@ -0,0 +1,1315 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ********************************************************************** + * Copyright (C) 1996-2016, International Business Machines + * Corporation and others. All Rights Reserved. + ********************************************************************** + * + * Provides functionality for mapping between + * LCID and Posix IDs or ICU locale to codepage + * + * Note: All classes and code in this file are + * intended for internal use only. + * + * Methods of interest: + * unsigned long convertToLCID(const char*); + * const char* convertToPosix(unsigned long); + * + * Kathleen Wilson, 4/30/96 + * + * Date Name Description + * 3/11/97 aliu Fixed off-by-one bug in assignment operator. Added + * setId() method and safety check against + * MAX_ID_LENGTH. + * 04/23/99 stephen Added C wrapper for convertToPosix. + * 09/18/00 george Removed the memory leaks. + * 08/23/01 george Convert to C + */ + +#include "locmap.h" +#include "bytesinkutil.h" +#include "charstr.h" +#include "cstring.h" +#include "cmemory.h" +#include "ulocimp.h" +#include "unicode/uloc.h" + +#if U_PLATFORM_HAS_WIN32_API && UCONFIG_USE_WINDOWS_LCID_MAPPING_API +#include +#include // LCIDToLocaleName and LocaleNameToLCID +#endif + +/* + * Note: + * The mapping from Win32 locale ID numbers to POSIX locale strings should + * be the faster one. + * + * Windows LCIDs are defined at https://msdn.microsoft.com/en-us/library/cc233965.aspx + * [MS-LCID] Windows Language Code Identifier (LCID) Reference + */ + +/* +//////////////////////////////////////////////// +// +// Internal Classes for LCID <--> POSIX Mapping +// +///////////////////////////////////////////////// +*/ + +typedef struct ILcidPosixElement +{ + const uint32_t hostID; + const char * const posixID; +} ILcidPosixElement; + +typedef struct ILcidPosixMap +{ + const uint32_t numRegions; + const struct ILcidPosixElement* const regionMaps; +} ILcidPosixMap; + + +/* +///////////////////////////////////////////////// +// +// Easy macros to make the LCID <--> POSIX Mapping +// +///////////////////////////////////////////////// +*/ + +/** + * The standard one language/one country mapping for LCID. + * The first element must be the language, and the following + * elements are the language with the country. + * @param hostID LCID in host format such as 0x044d + * @param languageID posix ID of just the language such as 'de' + * @param posixID posix ID of the language_TERRITORY such as 'de_CH' + */ +#define ILCID_POSIX_ELEMENT_ARRAY(hostID, languageID, posixID) \ +static const ILcidPosixElement locmap_ ## languageID [] = { \ + {LANGUAGE_LCID(hostID), #languageID}, /* parent locale */ \ + {hostID, #posixID}, \ +}; + +/** + * Define a subtable by ID + * @param id the POSIX ID, either a language or language_TERRITORY + */ +#define ILCID_POSIX_SUBTABLE(id) \ +static const ILcidPosixElement locmap_ ## id [] = + + +/** + * Create the map for the posixID. This macro supposes that the language string + * name is the same as the global variable name, and that the first element + * in the ILcidPosixElement is just the language. + * @param _posixID the full POSIX ID for this entry. + */ +#define ILCID_POSIX_MAP(_posixID) \ + {UPRV_LENGTHOF(locmap_ ## _posixID), locmap_ ## _posixID} + +/* +//////////////////////////////////////////// +// +// Create the table of LCID to POSIX Mapping +// None of it should be dynamically created. +// +// Keep static locale variables inside the function so that +// it can be created properly during static init. +// +// Note: This table should be updated periodically. Check the [MS-LCID] Windows Language Code Identifier +// (LCID) Reference defined at https://msdn.microsoft.com/en-us/library/cc233965.aspx +// +// Microsoft is moving away from LCID in favor of locale name as of Vista. This table needs to be +// maintained for support of older Windows version. +// Update: Windows 7 (091130) +// +// Note: Microsoft assign a different LCID if a locale has a sorting variant. POSIX IDs below may contain +// @collation=XXX, but no other keywords are allowed (at least for now). When uprv_convertToLCID() is +// called from uloc_getLCID(), keywords other than collation are already removed. If we really need +// to support other keywords in this mapping data, we must update the implementation. +//////////////////////////////////////////// +*/ + +// TODO: For Windows ideally this table would be a list of exceptions rather than a complete list as +// LocaleNameToLCID and LCIDToLocaleName provide 90% of these. + +ILCID_POSIX_ELEMENT_ARRAY(0x0436, af, af_ZA) + +ILCID_POSIX_SUBTABLE(ar) { + {0x01, "ar"}, + {0x3801, "ar_AE"}, + {0x3c01, "ar_BH"}, + {0x1401, "ar_DZ"}, + {0x0c01, "ar_EG"}, + {0x0801, "ar_IQ"}, + {0x2c01, "ar_JO"}, + {0x3401, "ar_KW"}, + {0x3001, "ar_LB"}, + {0x1001, "ar_LY"}, + {0x1801, "ar_MA"}, + {0x1801, "ar_MO"}, + {0x2001, "ar_OM"}, + {0x4001, "ar_QA"}, + {0x0401, "ar_SA"}, + {0x2801, "ar_SY"}, + {0x1c01, "ar_TN"}, + {0x2401, "ar_YE"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x044d, as, as_IN) +ILCID_POSIX_ELEMENT_ARRAY(0x045e, am, am_ET) +ILCID_POSIX_ELEMENT_ARRAY(0x047a, arn,arn_CL) + +ILCID_POSIX_SUBTABLE(az) { + {0x2c, "az"}, + {0x082c, "az_Cyrl_AZ"}, /* Cyrillic based */ + {0x742c, "az_Cyrl"}, /* Cyrillic based */ + {0x042c, "az_Latn_AZ"}, /* Latin based */ + {0x782c, "az_Latn"}, /* Latin based */ + {0x042c, "az_AZ"} /* Latin based */ +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x046d, ba, ba_RU) +ILCID_POSIX_ELEMENT_ARRAY(0x0423, be, be_BY) + +/*ILCID_POSIX_SUBTABLE(ber) { + {0x5f, "ber"}, + {0x045f, "ber_Arab_DZ"}, + {0x045f, "ber_Arab"}, + {0x085f, "ber_Latn_DZ"}, + {0x085f, "ber_Latn"} +};*/ + +ILCID_POSIX_ELEMENT_ARRAY(0x0402, bg, bg_BG) + +ILCID_POSIX_SUBTABLE(bin) { + {0x66, "bin"}, + {0x0466, "bin_NG"} +}; + +ILCID_POSIX_SUBTABLE(bn) { + {0x45, "bn"}, + {0x0845, "bn_BD"}, + {0x0445, "bn_IN"} +}; + +ILCID_POSIX_SUBTABLE(bo) { + {0x51, "bo"}, + {0x0851, "bo_BT"}, + {0x0451, "bo_CN"}, + {0x0c51, "dz_BT"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x047e, br, br_FR) + +ILCID_POSIX_SUBTABLE(ca) { + {0x03, "ca"}, + {0x0403, "ca_ES"}, + {0x0803, "ca_ES_VALENCIA"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0483, co, co_FR) + +ILCID_POSIX_SUBTABLE(chr) { + {0x05c, "chr"}, + {0x7c5c, "chr_Cher"}, + {0x045c, "chr_Cher_US"}, + {0x045c, "chr_US"} +}; + +// ICU has chosen different names for these. +ILCID_POSIX_SUBTABLE(ckb) { + {0x92, "ckb"}, + {0x7c92, "ckb_Arab"}, + {0x0492, "ckb_Arab_IQ"} +}; + +/* Declared as cs_CZ to get around compiler errors on z/OS, which defines cs as a function */ +ILCID_POSIX_ELEMENT_ARRAY(0x0405, cs, cs_CZ) + +ILCID_POSIX_ELEMENT_ARRAY(0x0452, cy, cy_GB) +ILCID_POSIX_ELEMENT_ARRAY(0x0406, da, da_DK) + +// Windows doesn't know POSIX or BCP47 Unicode phonebook sort names +ILCID_POSIX_SUBTABLE(de) { + {0x07, "de"}, + {0x0c07, "de_AT"}, + {0x0807, "de_CH"}, + {0x0407, "de_DE"}, + {0x1407, "de_LI"}, + {0x1007, "de_LU"}, + {0x10407,"de_DE@collation=phonebook"}, /*This is really de_DE_PHONEBOOK on Windows*/ + {0x10407,"de@collation=phonebook"} /*This is really de_DE_PHONEBOOK on Windows*/ +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0465, dv, dv_MV) +ILCID_POSIX_ELEMENT_ARRAY(0x0408, el, el_GR) + +// Windows uses an empty string for 'invariant' +ILCID_POSIX_SUBTABLE(en) { + {0x09, "en"}, + {0x0c09, "en_AU"}, + {0x2809, "en_BZ"}, + {0x1009, "en_CA"}, + {0x0809, "en_GB"}, + {0x3c09, "en_HK"}, + {0x3809, "en_ID"}, + {0x1809, "en_IE"}, + {0x4009, "en_IN"}, + {0x2009, "en_JM"}, + {0x4409, "en_MY"}, + {0x1409, "en_NZ"}, + {0x3409, "en_PH"}, + {0x4809, "en_SG"}, + {0x2C09, "en_TT"}, + {0x0409, "en_US"}, + {0x007f, "en_US_POSIX"}, /* duplicate for round-tripping */ + {0x2409, "en_029"}, + {0x1c09, "en_ZA"}, + {0x3009, "en_ZW"}, + {0x2409, "en_VI"}, /* Virgin Islands AKA Caribbean Islands (en_CB). On Windows8+ This is 0x1000 or dynamically assigned */ + {0x0409, "en_AS"}, /* Alias for en_US. Leave last. On Windows8+ This is 0x1000 or dynamically assigned */ + {0x0409, "en_GU"}, /* Alias for en_US. Leave last. On Windows8+ This is 0x1000 or dynamically assigned */ + {0x0409, "en_MH"}, /* Alias for en_US. Leave last. On Windows8+ This is 0x1000 or dynamically assigned */ + {0x0409, "en_MP"}, /* Alias for en_US. Leave last. On Windows8+ This is 0x1000 or dynamically assigned */ + {0x0409, "en_UM"} /* Alias for en_US. Leave last. On Windows8+ This is 0x1000 or dynamically assigned */ +}; + +ILCID_POSIX_SUBTABLE(en_US_POSIX) { + {0x007f, "en_US_POSIX"} /* duplicate for roundtripping */ +}; + +// Windows doesn't know POSIX or BCP47 Unicode traditional sort names +ILCID_POSIX_SUBTABLE(es) { + {0x0a, "es"}, + {0x2c0a, "es_AR"}, + {0x400a, "es_BO"}, + {0x340a, "es_CL"}, + {0x240a, "es_CO"}, + {0x140a, "es_CR"}, + {0x5c0a, "es_CU"}, + {0x1c0a, "es_DO"}, + {0x300a, "es_EC"}, + {0x0c0a, "es_ES"}, /*Modern sort.*/ + {0x100a, "es_GT"}, + {0x480a, "es_HN"}, + {0x080a, "es_MX"}, + {0x4c0a, "es_NI"}, + {0x180a, "es_PA"}, + {0x280a, "es_PE"}, + {0x500a, "es_PR"}, + {0x3c0a, "es_PY"}, + {0x440a, "es_SV"}, + {0x540a, "es_US"}, + {0x380a, "es_UY"}, + {0x200a, "es_VE"}, + {0x580a, "es_419"}, + {0x040a, "es_ES@collation=traditional"}, + {0x040a, "es@collation=traditional"} // Windows will treat this as es-ES@collation=traditional +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0425, et, et_EE) +ILCID_POSIX_ELEMENT_ARRAY(0x042d, eu, eu_ES) + +/* ISO-639 doesn't distinguish between Persian and Dari.*/ +ILCID_POSIX_SUBTABLE(fa) { + {0x29, "fa"}, + {0x0429, "fa_IR"}, /* Persian/Farsi (Iran) */ + {0x048c, "fa_AF"} /* Persian/Dari (Afghanistan) */ +}; + + +/* duplicate for roundtripping */ +ILCID_POSIX_SUBTABLE(fa_AF) { + {0x8c, "fa_AF"}, /* Persian/Dari (Afghanistan) */ + {0x048c, "fa_AF"} /* Persian/Dari (Afghanistan) */ +}; + +ILCID_POSIX_SUBTABLE(ff) { + {0x67, "ff"}, + {0x7c67, "ff_Latn"}, + {0x0867, "ff_Latn_SN"}, + {0x0467, "ff_NG"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x040b, fi, fi_FI) +ILCID_POSIX_ELEMENT_ARRAY(0x0464, fil,fil_PH) +ILCID_POSIX_ELEMENT_ARRAY(0x0438, fo, fo_FO) + +ILCID_POSIX_SUBTABLE(fr) { + {0x0c, "fr"}, + {0x080c, "fr_BE"}, + {0x0c0c, "fr_CA"}, + {0x240c, "fr_CD"}, + {0x240c, "fr_CG"}, + {0x100c, "fr_CH"}, + {0x300c, "fr_CI"}, + {0x2c0c, "fr_CM"}, + {0x040c, "fr_FR"}, + {0x3c0c, "fr_HT"}, + {0x140c, "fr_LU"}, + {0x380c, "fr_MA"}, + {0x180c, "fr_MC"}, + {0x340c, "fr_ML"}, + {0x200c, "fr_RE"}, + {0x280c, "fr_SN"}, + {0xe40c, "fr_015"}, + {0x1c0c, "fr_029"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0467, fuv, fuv_NG) + +ILCID_POSIX_ELEMENT_ARRAY(0x0462, fy, fy_NL) + +ILCID_POSIX_SUBTABLE(ga) { /* Gaelic (Ireland) */ + {0x3c, "ga"}, + {0x083c, "ga_IE"}, + {0x043c, "gd_GB"} +}; + +ILCID_POSIX_SUBTABLE(gd) { /* Gaelic (Scotland) */ + {0x91, "gd"}, + {0x0491, "gd_GB"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0456, gl, gl_ES) +ILCID_POSIX_ELEMENT_ARRAY(0x0447, gu, gu_IN) +ILCID_POSIX_ELEMENT_ARRAY(0x0474, gn, gn_PY) +ILCID_POSIX_ELEMENT_ARRAY(0x0484, gsw,gsw_FR) + +ILCID_POSIX_SUBTABLE(ha) { + {0x68, "ha"}, + {0x7c68, "ha_Latn"}, + {0x0468, "ha_Latn_NG"}, +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0475, haw,haw_US) +ILCID_POSIX_ELEMENT_ARRAY(0x040d, he, he_IL) +ILCID_POSIX_ELEMENT_ARRAY(0x0439, hi, hi_IN) + +/* This LCID is really four different locales.*/ +ILCID_POSIX_SUBTABLE(hr) { + {0x1a, "hr"}, + {0x141a, "bs_Latn_BA"}, /* Bosnian, Bosnia and Herzegovina */ + {0x681a, "bs_Latn"}, /* Bosnian, Bosnia and Herzegovina */ + {0x141a, "bs_BA"}, /* Bosnian, Bosnia and Herzegovina */ + {0x781a, "bs"}, /* Bosnian */ + {0x201a, "bs_Cyrl_BA"}, /* Bosnian, Bosnia and Herzegovina */ + {0x641a, "bs_Cyrl"}, /* Bosnian, Bosnia and Herzegovina */ + {0x101a, "hr_BA"}, /* Croatian in Bosnia */ + {0x041a, "hr_HR"}, /* Croatian*/ + {0x2c1a, "sr_Latn_ME"}, + {0x241a, "sr_Latn_RS"}, + {0x181a, "sr_Latn_BA"}, /* Serbo-Croatian in Bosnia */ + {0x081a, "sr_Latn_CS"}, /* Serbo-Croatian*/ + {0x701a, "sr_Latn"}, /* It's 0x1a or 0x081a, pick one to make the test program happy. */ + {0x1c1a, "sr_Cyrl_BA"}, /* Serbo-Croatian in Bosnia */ + {0x0c1a, "sr_Cyrl_CS"}, /* Serbian*/ + {0x301a, "sr_Cyrl_ME"}, + {0x281a, "sr_Cyrl_RS"}, + {0x6c1a, "sr_Cyrl"}, /* It's 0x1a or 0x0c1a, pick one to make the test program happy. */ + {0x7c1a, "sr"} /* In CLDR sr is sr_Cyrl. */ +}; + +ILCID_POSIX_SUBTABLE(hsb) { + {0x2E, "hsb"}, + {0x042E, "hsb_DE"}, + {0x082E, "dsb_DE"}, + {0x7C2E, "dsb"}, +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x040e, hu, hu_HU) +ILCID_POSIX_ELEMENT_ARRAY(0x042b, hy, hy_AM) + +ILCID_POSIX_SUBTABLE(ibb) { + {0x69, "ibb"}, + {0x0469, "ibb_NG"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0421, id, id_ID) +ILCID_POSIX_ELEMENT_ARRAY(0x0470, ig, ig_NG) +ILCID_POSIX_ELEMENT_ARRAY(0x0478, ii, ii_CN) +ILCID_POSIX_ELEMENT_ARRAY(0x040f, is, is_IS) + +ILCID_POSIX_SUBTABLE(it) { + {0x10, "it"}, + {0x0810, "it_CH"}, + {0x0410, "it_IT"} +}; + +ILCID_POSIX_SUBTABLE(iu) { + {0x5d, "iu"}, + {0x045d, "iu_Cans_CA"}, + {0x785d, "iu_Cans"}, + {0x085d, "iu_Latn_CA"}, + {0x7c5d, "iu_Latn"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x040d, iw, iw_IL) /*Left in for compatibility*/ +ILCID_POSIX_ELEMENT_ARRAY(0x0411, ja, ja_JP) +ILCID_POSIX_ELEMENT_ARRAY(0x0437, ka, ka_GE) +ILCID_POSIX_ELEMENT_ARRAY(0x043f, kk, kk_KZ) +ILCID_POSIX_ELEMENT_ARRAY(0x046f, kl, kl_GL) +ILCID_POSIX_ELEMENT_ARRAY(0x0453, km, km_KH) +ILCID_POSIX_ELEMENT_ARRAY(0x044b, kn, kn_IN) + +ILCID_POSIX_SUBTABLE(ko) { + {0x12, "ko"}, + {0x0812, "ko_KP"}, + {0x0412, "ko_KR"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0457, kok, kok_IN) +ILCID_POSIX_ELEMENT_ARRAY(0x0471, kr, kr_NG) + +ILCID_POSIX_SUBTABLE(ks) { /* We could add PK and CN too */ + {0x60, "ks"}, + {0x0460, "ks_Arab_IN"}, + {0x0860, "ks_Deva_IN"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0440, ky, ky_KG) /* Kyrgyz is spoken in Kyrgyzstan */ + +ILCID_POSIX_SUBTABLE(la) { + {0x76, "la"}, + {0x0476, "la_001"}, + {0x0476, "la_IT"} /*Left in for compatibility*/ +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x046e, lb, lb_LU) +ILCID_POSIX_ELEMENT_ARRAY(0x0454, lo, lo_LA) +ILCID_POSIX_ELEMENT_ARRAY(0x0427, lt, lt_LT) +ILCID_POSIX_ELEMENT_ARRAY(0x0426, lv, lv_LV) +ILCID_POSIX_ELEMENT_ARRAY(0x0481, mi, mi_NZ) +ILCID_POSIX_ELEMENT_ARRAY(0x042f, mk, mk_MK) +ILCID_POSIX_ELEMENT_ARRAY(0x044c, ml, ml_IN) + +ILCID_POSIX_SUBTABLE(mn) { + {0x50, "mn"}, + {0x0450, "mn_MN"}, + {0x7c50, "mn_Mong"}, + {0x0850, "mn_Mong_CN"}, + {0x0850, "mn_CN"}, + {0x7850, "mn_Cyrl"}, + {0x0c50, "mn_Mong_MN"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0458, mni,mni_IN) +ILCID_POSIX_ELEMENT_ARRAY(0x047c, moh,moh_CA) +ILCID_POSIX_ELEMENT_ARRAY(0x044e, mr, mr_IN) + +ILCID_POSIX_SUBTABLE(ms) { + {0x3e, "ms"}, + {0x083e, "ms_BN"}, /* Brunei Darussalam*/ + {0x043e, "ms_MY"} /* Malaysia*/ +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x043a, mt, mt_MT) +ILCID_POSIX_ELEMENT_ARRAY(0x0455, my, my_MM) + +ILCID_POSIX_SUBTABLE(ne) { + {0x61, "ne"}, + {0x0861, "ne_IN"}, /* India*/ + {0x0461, "ne_NP"} /* Nepal*/ +}; + +ILCID_POSIX_SUBTABLE(nl) { + {0x13, "nl"}, + {0x0813, "nl_BE"}, + {0x0413, "nl_NL"} +}; + +/* The "no" locale split into nb and nn. By default in ICU, "no" is nb.*/ +// TODO: Not all of these are needed on Windows, but I don't know how ICU treats preferred ones here. +ILCID_POSIX_SUBTABLE(no) { + {0x14, "no"}, /* really nb_NO - actually Windows differentiates between neutral (no region) and specific (with region) */ + {0x7c14, "nb"}, /* really nb */ + {0x0414, "nb_NO"}, /* really nb_NO. Keep first in the 414 list. */ + {0x0414, "no_NO"}, /* really nb_NO */ + {0x0814, "nn_NO"}, /* really nn_NO. Keep first in the 814 list. */ + {0x7814, "nn"}, /* It's 0x14 or 0x814, pick one to make the test program happy. */ + {0x0814, "no_NO_NY"}/* really nn_NO */ +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x046c, nso,nso_ZA) /* TODO: Verify the ISO-639 code */ +ILCID_POSIX_ELEMENT_ARRAY(0x0482, oc, oc_FR) + +ILCID_POSIX_SUBTABLE(om) { /* TODO: Verify the country */ + {0x72, "om"}, + {0x0472, "om_ET"}, + {0x0472, "gaz_ET"} +}; + +/* Declared as or_IN to get around compiler errors*/ +ILCID_POSIX_SUBTABLE(or_IN) { + {0x48, "or"}, + {0x0448, "or_IN"}, +}; + +ILCID_POSIX_SUBTABLE(pa) { + {0x46, "pa"}, + {0x0446, "pa_IN"}, + {0x0846, "pa_Arab_PK"}, + {0x0846, "pa_PK"} +}; + +ILCID_POSIX_SUBTABLE(pap) { + {0x79, "pap"}, + {0x0479, "pap_029"}, + {0x0479, "pap_AN"} /*Left in for compatibility*/ +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0415, pl, pl_PL) +ILCID_POSIX_ELEMENT_ARRAY(0x0463, ps, ps_AF) + +ILCID_POSIX_SUBTABLE(pt) { + {0x16, "pt"}, + {0x0416, "pt_BR"}, + {0x0816, "pt_PT"} +}; + +ILCID_POSIX_SUBTABLE(qu) { + {0x6b, "qu"}, + {0x046b, "qu_BO"}, + {0x086b, "qu_EC"}, + {0x0C6b, "qu_PE"}, + {0x046b, "quz_BO"}, + {0x086b, "quz_EC"}, + {0x0C6b, "quz_PE"} +}; + +ILCID_POSIX_SUBTABLE(quc) { + {0x93, "quc"}, + {0x0493, "quc_CO"}, + /* + "quc_Latn_GT" is an exceptional case. Language ID of "quc" + is 0x93, but LCID of "quc_Latn_GT" is 0x486, which should be + under the group of "qut". "qut" is a retired ISO 639-3 language + code for West Central Quiche, and merged to "quc". + It looks Windows previously reserved "qut" for K'iche', but, + decided to use "quc" when adding a locale for K'iche' (Guatemala). + + This data structure used here assumes language ID bits in + LCID is unique for alphabetic language code. But this is not true + for "quc_Latn_GT". If we don't have the data below, LCID look up + by alphabetic locale ID (POSIX) will fail. The same entry is found + under "qut" below, which is required for reverse look up. + */ + {0x0486, "quc_Latn_GT"} +}; + +ILCID_POSIX_SUBTABLE(qut) { + {0x86, "qut"}, + {0x0486, "qut_GT"}, + /* + See the note in "quc" above. + */ + {0x0486, "quc_Latn_GT"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0417, rm, rm_CH) + +ILCID_POSIX_SUBTABLE(ro) { + {0x18, "ro"}, + {0x0418, "ro_RO"}, + {0x0818, "ro_MD"} +}; + +// TODO: This is almost certainly 'wrong'. 0 in Windows is a synonym for LOCALE_USER_DEFAULT. +// More likely this is a similar concept to the Windows 0x7f Invariant locale "" +// (Except that it's not invariant in ICU) +ILCID_POSIX_SUBTABLE(root) { + {0x00, "root"} +}; + +ILCID_POSIX_SUBTABLE(ru) { + {0x19, "ru"}, + {0x0419, "ru_RU"}, + {0x0819, "ru_MD"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0487, rw, rw_RW) +ILCID_POSIX_ELEMENT_ARRAY(0x044f, sa, sa_IN) +ILCID_POSIX_ELEMENT_ARRAY(0x0485, sah,sah_RU) + +ILCID_POSIX_SUBTABLE(sd) { + {0x59, "sd"}, + {0x0459, "sd_Deva_IN"}, + {0x0459, "sd_IN"}, + {0x0859, "sd_Arab_PK"}, + {0x0859, "sd_PK"}, + {0x7c59, "sd_Arab"} +}; + +ILCID_POSIX_SUBTABLE(se) { + {0x3b, "se"}, + {0x0c3b, "se_FI"}, + {0x043b, "se_NO"}, + {0x083b, "se_SE"}, + {0x783b, "sma"}, + {0x183b, "sma_NO"}, + {0x1c3b, "sma_SE"}, + {0x7c3b, "smj"}, + {0x703b, "smn"}, + {0x743b, "sms"}, + {0x103b, "smj_NO"}, + {0x143b, "smj_SE"}, + {0x243b, "smn_FI"}, + {0x203b, "sms_FI"}, +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x045b, si, si_LK) +ILCID_POSIX_ELEMENT_ARRAY(0x041b, sk, sk_SK) +ILCID_POSIX_ELEMENT_ARRAY(0x0424, sl, sl_SI) + +ILCID_POSIX_SUBTABLE(so) { + {0x77, "so"}, + {0x0477, "so_SO"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x041c, sq, sq_AL) +ILCID_POSIX_ELEMENT_ARRAY(0x0430, st, st_ZA) + +ILCID_POSIX_SUBTABLE(sv) { + {0x1d, "sv"}, + {0x081d, "sv_FI"}, + {0x041d, "sv_SE"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0441, sw, sw_KE) +ILCID_POSIX_ELEMENT_ARRAY(0x045A, syr, syr_SY) + +ILCID_POSIX_SUBTABLE(ta) { + {0x49, "ta"}, + {0x0449, "ta_IN"}, + {0x0849, "ta_LK"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x044a, te, te_IN) + +/* Cyrillic based by default */ +ILCID_POSIX_SUBTABLE(tg) { + {0x28, "tg"}, + {0x7c28, "tg_Cyrl"}, + {0x0428, "tg_Cyrl_TJ"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x041e, th, th_TH) + +ILCID_POSIX_SUBTABLE(ti) { + {0x73, "ti"}, + {0x0873, "ti_ER"}, + {0x0473, "ti_ET"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0442, tk, tk_TM) + +ILCID_POSIX_SUBTABLE(tn) { + {0x32, "tn"}, + {0x0832, "tn_BW"}, + {0x0432, "tn_ZA"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x041f, tr, tr_TR) +ILCID_POSIX_ELEMENT_ARRAY(0x0431, ts, ts_ZA) +ILCID_POSIX_ELEMENT_ARRAY(0x0444, tt, tt_RU) + +ILCID_POSIX_SUBTABLE(tzm) { + {0x5f, "tzm"}, + {0x7c5f, "tzm_Latn"}, + {0x085f, "tzm_Latn_DZ"}, + {0x105f, "tzm_Tfng_MA"}, + {0x045f, "tzm_Arab_MA"}, + {0x045f, "tmz"} +}; + +ILCID_POSIX_SUBTABLE(ug) { + {0x80, "ug"}, + {0x0480, "ug_CN"}, + {0x0480, "ug_Arab_CN"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0422, uk, uk_UA) + +ILCID_POSIX_SUBTABLE(ur) { + {0x20, "ur"}, + {0x0820, "ur_IN"}, + {0x0420, "ur_PK"} +}; + +ILCID_POSIX_SUBTABLE(uz) { + {0x43, "uz"}, + {0x0843, "uz_Cyrl_UZ"}, /* Cyrillic based */ + {0x7843, "uz_Cyrl"}, /* Cyrillic based */ + {0x0843, "uz_UZ"}, /* Cyrillic based */ + {0x0443, "uz_Latn_UZ"}, /* Latin based */ + {0x7c43, "uz_Latn"} /* Latin based */ +}; + +ILCID_POSIX_SUBTABLE(ve) { /* TODO: Verify the country */ + {0x33, "ve"}, + {0x0433, "ve_ZA"}, + {0x0433, "ven_ZA"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x042a, vi, vi_VN) +ILCID_POSIX_ELEMENT_ARRAY(0x0488, wo, wo_SN) +ILCID_POSIX_ELEMENT_ARRAY(0x0434, xh, xh_ZA) + +ILCID_POSIX_SUBTABLE(yi) { + {0x003d, "yi"}, + {0x043d, "yi_001"} +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x046a, yo, yo_NG) + +// Windows & ICU tend to different names for some of these +// TODO: Windows probably does not need all of these entries, but I don't know how the precedence works. +ILCID_POSIX_SUBTABLE(zh) { + {0x0004, "zh_Hans"}, + {0x7804, "zh"}, + {0x0804, "zh_CN"}, + {0x0804, "zh_Hans_CN"}, + {0x0c04, "zh_Hant_HK"}, + {0x0c04, "zh_HK"}, + {0x1404, "zh_Hant_MO"}, + {0x1404, "zh_MO"}, + {0x1004, "zh_Hans_SG"}, + {0x1004, "zh_SG"}, + {0x0404, "zh_Hant_TW"}, + {0x7c04, "zh_Hant"}, + {0x0404, "zh_TW"}, + {0x30404,"zh_Hant_TW"}, /* Bopomofo order */ + {0x30404,"zh_TW"}, /* Bopomofo order */ + {0x20004,"zh@collation=stroke"}, + {0x20404,"zh_Hant@collation=stroke"}, + {0x20404,"zh_Hant_TW@collation=stroke"}, + {0x20404,"zh_TW@collation=stroke"}, + {0x20804,"zh_Hans@collation=stroke"}, + {0x20804,"zh_Hans_CN@collation=stroke"}, + {0x20804,"zh_CN@collation=stroke"} + // TODO: Alternate collations for other LCIDs are missing, eg: 0x50804 +}; + +ILCID_POSIX_ELEMENT_ARRAY(0x0435, zu, zu_ZA) + +/* This must be static and grouped by LCID. */ +static const ILcidPosixMap gPosixIDmap[] = { + ILCID_POSIX_MAP(af), /* af Afrikaans 0x36 */ + ILCID_POSIX_MAP(am), /* am Amharic 0x5e */ + ILCID_POSIX_MAP(ar), /* ar Arabic 0x01 */ + ILCID_POSIX_MAP(arn), /* arn Araucanian/Mapudungun 0x7a */ + ILCID_POSIX_MAP(as), /* as Assamese 0x4d */ + ILCID_POSIX_MAP(az), /* az Azerbaijani 0x2c */ + ILCID_POSIX_MAP(ba), /* ba Bashkir 0x6d */ + ILCID_POSIX_MAP(be), /* be Belarusian 0x23 */ +/* ILCID_POSIX_MAP(ber), ber Berber/Tamazight 0x5f */ + ILCID_POSIX_MAP(bg), /* bg Bulgarian 0x02 */ + ILCID_POSIX_MAP(bin), /* bin Edo 0x66 */ + ILCID_POSIX_MAP(bn), /* bn Bengali; Bangla 0x45 */ + ILCID_POSIX_MAP(bo), /* bo Tibetan 0x51 */ + ILCID_POSIX_MAP(br), /* br Breton 0x7e */ + ILCID_POSIX_MAP(ca), /* ca Catalan 0x03 */ + ILCID_POSIX_MAP(chr), /* chr Cherokee 0x5c */ + ILCID_POSIX_MAP(ckb), /* ckb Sorani (Central Kurdish) 0x92 */ + ILCID_POSIX_MAP(co), /* co Corsican 0x83 */ + ILCID_POSIX_MAP(cs), /* cs Czech 0x05 */ + ILCID_POSIX_MAP(cy), /* cy Welsh 0x52 */ + ILCID_POSIX_MAP(da), /* da Danish 0x06 */ + ILCID_POSIX_MAP(de), /* de German 0x07 */ + ILCID_POSIX_MAP(dv), /* dv Divehi 0x65 */ + ILCID_POSIX_MAP(el), /* el Greek 0x08 */ + ILCID_POSIX_MAP(en), /* en English 0x09 */ + ILCID_POSIX_MAP(en_US_POSIX), /* invariant 0x7f */ + ILCID_POSIX_MAP(es), /* es Spanish 0x0a */ + ILCID_POSIX_MAP(et), /* et Estonian 0x25 */ + ILCID_POSIX_MAP(eu), /* eu Basque 0x2d */ + ILCID_POSIX_MAP(fa), /* fa Persian/Farsi 0x29 */ + ILCID_POSIX_MAP(fa_AF), /* fa Persian/Dari 0x8c */ + ILCID_POSIX_MAP(ff), /* ff Fula 0x67 */ + ILCID_POSIX_MAP(fi), /* fi Finnish 0x0b */ + ILCID_POSIX_MAP(fil), /* fil Filipino 0x64 */ + ILCID_POSIX_MAP(fo), /* fo Faroese 0x38 */ + ILCID_POSIX_MAP(fr), /* fr French 0x0c */ + ILCID_POSIX_MAP(fuv), /* fuv Fulfulde - Nigeria 0x67 */ + ILCID_POSIX_MAP(fy), /* fy Frisian 0x62 */ + ILCID_POSIX_MAP(ga), /* * Gaelic (Ireland,Scotland) 0x3c */ + ILCID_POSIX_MAP(gd), /* gd Gaelic (United Kingdom) 0x91 */ + ILCID_POSIX_MAP(gl), /* gl Galician 0x56 */ + ILCID_POSIX_MAP(gn), /* gn Guarani 0x74 */ + ILCID_POSIX_MAP(gsw), /* gsw Alemanic/Alsatian/Swiss German 0x84 */ + ILCID_POSIX_MAP(gu), /* gu Gujarati 0x47 */ + ILCID_POSIX_MAP(ha), /* ha Hausa 0x68 */ + ILCID_POSIX_MAP(haw), /* haw Hawaiian 0x75 */ + ILCID_POSIX_MAP(he), /* he Hebrew (formerly iw) 0x0d */ + ILCID_POSIX_MAP(hi), /* hi Hindi 0x39 */ + ILCID_POSIX_MAP(hr), /* * Croatian and others 0x1a */ + ILCID_POSIX_MAP(hsb), /* hsb Upper Sorbian 0x2e */ + ILCID_POSIX_MAP(hu), /* hu Hungarian 0x0e */ + ILCID_POSIX_MAP(hy), /* hy Armenian 0x2b */ + ILCID_POSIX_MAP(ibb), /* ibb Ibibio - Nigeria 0x69 */ + ILCID_POSIX_MAP(id), /* id Indonesian (formerly in) 0x21 */ + ILCID_POSIX_MAP(ig), /* ig Igbo 0x70 */ + ILCID_POSIX_MAP(ii), /* ii Sichuan Yi 0x78 */ + ILCID_POSIX_MAP(is), /* is Icelandic 0x0f */ + ILCID_POSIX_MAP(it), /* it Italian 0x10 */ + ILCID_POSIX_MAP(iu), /* iu Inuktitut 0x5d */ + ILCID_POSIX_MAP(iw), /* iw Hebrew 0x0d */ + ILCID_POSIX_MAP(ja), /* ja Japanese 0x11 */ + ILCID_POSIX_MAP(ka), /* ka Georgian 0x37 */ + ILCID_POSIX_MAP(kk), /* kk Kazakh 0x3f */ + ILCID_POSIX_MAP(kl), /* kl Kalaallisut 0x6f */ + ILCID_POSIX_MAP(km), /* km Khmer 0x53 */ + ILCID_POSIX_MAP(kn), /* kn Kannada 0x4b */ + ILCID_POSIX_MAP(ko), /* ko Korean 0x12 */ + ILCID_POSIX_MAP(kok), /* kok Konkani 0x57 */ + ILCID_POSIX_MAP(kr), /* kr Kanuri 0x71 */ + ILCID_POSIX_MAP(ks), /* ks Kashmiri 0x60 */ + ILCID_POSIX_MAP(ky), /* ky Kyrgyz 0x40 */ + ILCID_POSIX_MAP(lb), /* lb Luxembourgish 0x6e */ + ILCID_POSIX_MAP(la), /* la Latin 0x76 */ + ILCID_POSIX_MAP(lo), /* lo Lao 0x54 */ + ILCID_POSIX_MAP(lt), /* lt Lithuanian 0x27 */ + ILCID_POSIX_MAP(lv), /* lv Latvian, Lettish 0x26 */ + ILCID_POSIX_MAP(mi), /* mi Maori 0x81 */ + ILCID_POSIX_MAP(mk), /* mk Macedonian 0x2f */ + ILCID_POSIX_MAP(ml), /* ml Malayalam 0x4c */ + ILCID_POSIX_MAP(mn), /* mn Mongolian 0x50 */ + ILCID_POSIX_MAP(mni), /* mni Manipuri 0x58 */ + ILCID_POSIX_MAP(moh), /* moh Mohawk 0x7c */ + ILCID_POSIX_MAP(mr), /* mr Marathi 0x4e */ + ILCID_POSIX_MAP(ms), /* ms Malay 0x3e */ + ILCID_POSIX_MAP(mt), /* mt Maltese 0x3a */ + ILCID_POSIX_MAP(my), /* my Burmese 0x55 */ +/* ILCID_POSIX_MAP(nb), // no Norwegian 0x14 */ + ILCID_POSIX_MAP(ne), /* ne Nepali 0x61 */ + ILCID_POSIX_MAP(nl), /* nl Dutch 0x13 */ +/* ILCID_POSIX_MAP(nn), // no Norwegian 0x14 */ + ILCID_POSIX_MAP(no), /* * Norwegian 0x14 */ + ILCID_POSIX_MAP(nso), /* nso Sotho, Northern (Sepedi dialect) 0x6c */ + ILCID_POSIX_MAP(oc), /* oc Occitan 0x82 */ + ILCID_POSIX_MAP(om), /* om Oromo 0x72 */ + ILCID_POSIX_MAP(or_IN), /* or Oriya 0x48 */ + ILCID_POSIX_MAP(pa), /* pa Punjabi 0x46 */ + ILCID_POSIX_MAP(pap), /* pap Papiamentu 0x79 */ + ILCID_POSIX_MAP(pl), /* pl Polish 0x15 */ + ILCID_POSIX_MAP(ps), /* ps Pashto 0x63 */ + ILCID_POSIX_MAP(pt), /* pt Portuguese 0x16 */ + ILCID_POSIX_MAP(qu), /* qu Quechua 0x6B */ + ILCID_POSIX_MAP(quc), /* quc K'iche 0x93 */ + ILCID_POSIX_MAP(qut), /* qut K'iche 0x86 */ + ILCID_POSIX_MAP(rm), /* rm Raeto-Romance/Romansh 0x17 */ + ILCID_POSIX_MAP(ro), /* ro Romanian 0x18 */ + ILCID_POSIX_MAP(root), /* root 0x00 */ + ILCID_POSIX_MAP(ru), /* ru Russian 0x19 */ + ILCID_POSIX_MAP(rw), /* rw Kinyarwanda 0x87 */ + ILCID_POSIX_MAP(sa), /* sa Sanskrit 0x4f */ + ILCID_POSIX_MAP(sah), /* sah Yakut 0x85 */ + ILCID_POSIX_MAP(sd), /* sd Sindhi 0x59 */ + ILCID_POSIX_MAP(se), /* se Sami 0x3b */ +/* ILCID_POSIX_MAP(sh), // sh Serbo-Croatian 0x1a */ + ILCID_POSIX_MAP(si), /* si Sinhalese 0x5b */ + ILCID_POSIX_MAP(sk), /* sk Slovak 0x1b */ + ILCID_POSIX_MAP(sl), /* sl Slovenian 0x24 */ + ILCID_POSIX_MAP(so), /* so Somali 0x77 */ + ILCID_POSIX_MAP(sq), /* sq Albanian 0x1c */ +/* ILCID_POSIX_MAP(sr), // sr Serbian 0x1a */ + ILCID_POSIX_MAP(st), /* st Sutu 0x30 */ + ILCID_POSIX_MAP(sv), /* sv Swedish 0x1d */ + ILCID_POSIX_MAP(sw), /* sw Swahili 0x41 */ + ILCID_POSIX_MAP(syr), /* syr Syriac 0x5A */ + ILCID_POSIX_MAP(ta), /* ta Tamil 0x49 */ + ILCID_POSIX_MAP(te), /* te Telugu 0x4a */ + ILCID_POSIX_MAP(tg), /* tg Tajik 0x28 */ + ILCID_POSIX_MAP(th), /* th Thai 0x1e */ + ILCID_POSIX_MAP(ti), /* ti Tigrigna 0x73 */ + ILCID_POSIX_MAP(tk), /* tk Turkmen 0x42 */ + ILCID_POSIX_MAP(tn), /* tn Tswana 0x32 */ + ILCID_POSIX_MAP(tr), /* tr Turkish 0x1f */ + ILCID_POSIX_MAP(ts), /* ts Tsonga 0x31 */ + ILCID_POSIX_MAP(tt), /* tt Tatar 0x44 */ + ILCID_POSIX_MAP(tzm), /* tzm Tamazight 0x5f */ + ILCID_POSIX_MAP(ug), /* ug Uighur 0x80 */ + ILCID_POSIX_MAP(uk), /* uk Ukrainian 0x22 */ + ILCID_POSIX_MAP(ur), /* ur Urdu 0x20 */ + ILCID_POSIX_MAP(uz), /* uz Uzbek 0x43 */ + ILCID_POSIX_MAP(ve), /* ve Venda 0x33 */ + ILCID_POSIX_MAP(vi), /* vi Vietnamese 0x2a */ + ILCID_POSIX_MAP(wo), /* wo Wolof 0x88 */ + ILCID_POSIX_MAP(xh), /* xh Xhosa 0x34 */ + ILCID_POSIX_MAP(yi), /* yi Yiddish 0x3d */ + ILCID_POSIX_MAP(yo), /* yo Yoruba 0x6a */ + ILCID_POSIX_MAP(zh), /* zh Chinese 0x04 */ + ILCID_POSIX_MAP(zu), /* zu Zulu 0x35 */ +}; + +static const uint32_t gLocaleCount = UPRV_LENGTHOF(gPosixIDmap); + +/** + * Do not call this function. It is called by hostID. + * The function is not private because this struct must stay as a C struct, + * and this is an internal class. + */ +static int32_t +idCmp(const char* id1, const char* id2) +{ + int32_t diffIdx = 0; + while (*id1 == *id2 && *id1 != 0) { + diffIdx++; + id1++; + id2++; + } + return diffIdx; +} + +/** + * Searches for a Windows LCID + * + * @param posixID the Posix style locale id. + * @param status gets set to U_ILLEGAL_ARGUMENT_ERROR when the Posix ID has + * no equivalent Windows LCID. + * @return the LCID + */ +static uint32_t +getHostID(const ILcidPosixMap *this_0, const char* posixID, UErrorCode* status) +{ + int32_t bestIdx = 0; + int32_t bestIdxDiff = 0; + int32_t posixIDlen = (int32_t)uprv_strlen(posixID); + uint32_t idx; + + for (idx = 0; idx < this_0->numRegions; idx++ ) { + int32_t sameChars = idCmp(posixID, this_0->regionMaps[idx].posixID); + if (sameChars > bestIdxDiff && this_0->regionMaps[idx].posixID[sameChars] == 0) { + if (posixIDlen == sameChars) { + /* Exact match */ + return this_0->regionMaps[idx].hostID; + } + bestIdxDiff = sameChars; + bestIdx = idx; + } + } + /* We asked for something unusual, like en_ZZ, and we try to return the number for the same language. */ + /* We also have to make sure that sid and si and similar string subsets don't match. */ + if ((posixID[bestIdxDiff] == '_' || posixID[bestIdxDiff] == '@') + && this_0->regionMaps[bestIdx].posixID[bestIdxDiff] == 0) + { + *status = U_USING_FALLBACK_WARNING; + return this_0->regionMaps[bestIdx].hostID; + } + + /*no match found */ + *status = U_ILLEGAL_ARGUMENT_ERROR; + return this_0->regionMaps->hostID; +} + +static const char* +getPosixID(const ILcidPosixMap *this_0, uint32_t hostID) +{ + uint32_t i; + for (i = 0; i < this_0->numRegions; i++) + { + if (this_0->regionMaps[i].hostID == hostID) + { + return this_0->regionMaps[i].posixID; + } + } + + /* If you get here, then no matching region was found, + so return the language id with the wild card region. */ + return this_0->regionMaps[0].posixID; +} + +/* +////////////////////////////////////// +// +// LCID --> POSIX +// +///////////////////////////////////// +*/ +#if U_PLATFORM_HAS_WIN32_API && UCONFIG_USE_WINDOWS_LCID_MAPPING_API +/* + * Various language tags needs to be changed: + * quz -> qu + * prs -> fa + */ +#define FIX_LANGUAGE_ID_TAG(buffer, len) \ + if (len >= 3) { \ + if (buffer[0] == 'q' && buffer[1] == 'u' && buffer[2] == 'z') {\ + buffer[2] = 0; \ + uprv_strcat(buffer, buffer+3); \ + } else if (buffer[0] == 'p' && buffer[1] == 'r' && buffer[2] == 's') {\ + buffer[0] = 'f'; buffer[1] = 'a'; buffer[2] = 0; \ + uprv_strcat(buffer, buffer+3); \ + } \ + } + +#endif + +U_CAPI int32_t +uprv_convertToPosix(uint32_t hostid, char *posixID, int32_t posixIDCapacity, UErrorCode* status) +{ + uint16_t langID; + uint32_t localeIndex; + UBool bLookup = true; + const char *pPosixID = nullptr; + +#if U_PLATFORM_HAS_WIN32_API && UCONFIG_USE_WINDOWS_LCID_MAPPING_API + static_assert(ULOC_FULLNAME_CAPACITY > LOCALE_NAME_MAX_LENGTH, "Windows locale names have smaller length than ICU locale names."); + + char locName[LOCALE_NAME_MAX_LENGTH] = {}; + + // Note: Windows primary lang ID 0x92 in LCID is used for Central Kurdish and + // GetLocaleInfo() maps such LCID to "ku". However, CLDR uses "ku" for + // Northern Kurdish and "ckb" for Central Kurdish. For this reason, we cannot + // use the Windows API to resolve locale ID for this specific case. + if ((hostid & 0x3FF) != 0x92) { + int32_t tmpLen = 0; + char16_t windowsLocaleName[LOCALE_NAME_MAX_LENGTH] = {}; + + // Note: LOCALE_ALLOW_NEUTRAL_NAMES was enabled in Windows7+, prior versions did not handle neutral (no-region) locale names. + tmpLen = LCIDToLocaleName(hostid, (PWSTR)windowsLocaleName, UPRV_LENGTHOF(windowsLocaleName), LOCALE_ALLOW_NEUTRAL_NAMES); + if (tmpLen > 1) { + int32_t i = 0; + // Only need to look up in table if have _, eg for de-de_phoneb type alternate sort. + bLookup = false; + for (i = 0; i < UPRV_LENGTHOF(locName); i++) + { + locName[i] = (char)(windowsLocaleName[i]); + + // Windows locale name may contain sorting variant, such as "es-ES_tradnl". + // In such cases, we need special mapping data found in the hardcoded table + // in this source file. + if (windowsLocaleName[i] == L'_') + { + // Keep the base locale, without variant + // TODO: Should these be mapped from _phoneb to @collation=phonebook, etc.? + locName[i] = '\0'; + tmpLen = i; + bLookup = true; + break; + } + else if (windowsLocaleName[i] == L'-') + { + // Windows names use -, ICU uses _ + locName[i] = '_'; + } + else if (windowsLocaleName[i] == L'\0') + { + // No point in doing more work than necessary + break; + } + } + // TODO: Need to understand this better, why isn't it an alias? + FIX_LANGUAGE_ID_TAG(locName, tmpLen); + pPosixID = locName; + } + } +#endif + + if (bLookup) { + const char *pCandidate = nullptr; + langID = LANGUAGE_LCID(hostid); + + for (localeIndex = 0; localeIndex < gLocaleCount; localeIndex++) { + if (langID == gPosixIDmap[localeIndex].regionMaps->hostID) { + pCandidate = getPosixID(&gPosixIDmap[localeIndex], hostid); + break; + } + } + + /* On Windows, when locale name has a variant, we still look up the hardcoded table. + If a match in the hardcoded table is longer than the Windows locale name without + variant, we use the one as the result */ + if (pCandidate && (pPosixID == nullptr || uprv_strlen(pCandidate) > uprv_strlen(pPosixID))) { + pPosixID = pCandidate; + } + } + + if (pPosixID) { + int32_t resLen = static_cast(uprv_strlen(pPosixID)); + int32_t copyLen = resLen <= posixIDCapacity ? resLen : posixIDCapacity; + uprv_memcpy(posixID, pPosixID, copyLen); + if (resLen < posixIDCapacity) { + posixID[resLen] = 0; + if (*status == U_STRING_NOT_TERMINATED_WARNING) { + *status = U_ZERO_ERROR; + } + } else if (resLen == posixIDCapacity) { + *status = U_STRING_NOT_TERMINATED_WARNING; + } else { + *status = U_BUFFER_OVERFLOW_ERROR; + } + return resLen; + } + + /* no match found */ + *status = U_ILLEGAL_ARGUMENT_ERROR; + return -1; +} + +/* +////////////////////////////////////// +// +// POSIX --> LCID +// This should only be called from uloc_getLCID. +// The locale ID must be in canonical form. +// +///////////////////////////////////// +*/ +U_CAPI uint32_t +uprv_convertToLCIDPlatform(const char* localeID, UErrorCode* status) +{ + if (U_FAILURE(*status)) { + return 0; + } + + // The purpose of this function is to leverage the Windows platform name->lcid + // conversion functionality when available. +#if U_PLATFORM_HAS_WIN32_API && UCONFIG_USE_WINDOWS_LCID_MAPPING_API + int32_t len; + char baseName[ULOC_FULLNAME_CAPACITY] = {}; + const char * mylocaleID = localeID; + + // Check any for keywords. + if (uprv_strchr(localeID, '@')) + { + icu::CharString collVal; + { + icu::CharStringByteSink sink(&collVal); + ulocimp_getKeywordValue(localeID, "collation", sink, status); + } + if (U_SUCCESS(*status) && !collVal.isEmpty()) + { + // If it contains the keyword collation, return 0 so that the LCID lookup table will be used. + return 0; + } + else + { + // If the locale ID contains keywords other than collation, just use the base name. + len = uloc_getBaseName(localeID, baseName, UPRV_LENGTHOF(baseName) - 1, status); + + if (U_SUCCESS(*status) && len > 0) + { + baseName[len] = 0; + mylocaleID = baseName; + } + } + } + + char asciiBCP47Tag[LOCALE_NAME_MAX_LENGTH] = {}; + // this will change it from de_DE@collation=phonebook to de-DE-u-co-phonebk form + (void)uloc_toLanguageTag(mylocaleID, asciiBCP47Tag, UPRV_LENGTHOF(asciiBCP47Tag), false, status); + + if (U_SUCCESS(*status)) + { + // Need it to be UTF-16, not 8-bit + wchar_t bcp47Tag[LOCALE_NAME_MAX_LENGTH] = {}; + int32_t i; + for (i = 0; i < UPRV_LENGTHOF(bcp47Tag); i++) + { + if (asciiBCP47Tag[i] == '\0') + { + break; + } + else + { + // Copy the character + bcp47Tag[i] = static_cast(asciiBCP47Tag[i]); + } + } + + if (i < (UPRV_LENGTHOF(bcp47Tag) - 1)) + { + // Ensure it's null terminated + bcp47Tag[i] = L'\0'; + LCID lcid = LocaleNameToLCID(bcp47Tag, LOCALE_ALLOW_NEUTRAL_NAMES); + if (lcid > 0) + { + // Found LCID from windows, return that one, unless its completely ambiguous + // LOCALE_USER_DEFAULT and transients are OK because they will round trip + // for this process. + if (lcid != LOCALE_CUSTOM_UNSPECIFIED) + { + return lcid; + } + } + } + } +#else + (void) localeID; // Suppress unused variable warning. +#endif + + // Nothing found, or not implemented. + return 0; +} + +U_CAPI uint32_t +uprv_convertToLCID(const char *langID, const char* posixID, UErrorCode* status) +{ + // This function does the table lookup when native platform name->lcid conversion isn't available, + // or for locales that don't follow patterns the platform expects. + uint32_t low = 0; + uint32_t high = gLocaleCount; + uint32_t mid; + uint32_t oldmid = 0; + int32_t compVal; + + uint32_t value = 0; + uint32_t fallbackValue = (uint32_t)-1; + UErrorCode myStatus; + uint32_t idx; + + /* Check for incomplete id. */ + if (!langID || !posixID || uprv_strlen(langID) < 2 || uprv_strlen(posixID) < 2) { + return 0; + } + + /*Binary search for the map entry for normal cases */ + + while (high > low) /*binary search*/{ + + mid = (high+low) >> 1; /*Finds median*/ + + if (mid == oldmid) + break; + + compVal = uprv_strcmp(langID, gPosixIDmap[mid].regionMaps->posixID); + if (compVal < 0){ + high = mid; + } + else if (compVal > 0){ + low = mid; + } + else /*we found it*/{ + return getHostID(&gPosixIDmap[mid], posixID, status); + } + oldmid = mid; + } + + /* + * Sometimes we can't do a binary search on posixID because some LCIDs + * go to different locales. We hit one of those special cases. + */ + for (idx = 0; idx < gLocaleCount; idx++ ) { + myStatus = U_ZERO_ERROR; + value = getHostID(&gPosixIDmap[idx], posixID, &myStatus); + if (myStatus == U_ZERO_ERROR) { + return value; + } + else if (myStatus == U_USING_FALLBACK_WARNING) { + fallbackValue = value; + } + } + + if (fallbackValue != (uint32_t)-1) { + *status = U_USING_FALLBACK_WARNING; + return fallbackValue; + } + + /* no match found */ + *status = U_ILLEGAL_ARGUMENT_ERROR; + return 0; /* return international (root) */ +} diff --git a/intl/icu/source/common/locmap.h b/intl/icu/source/common/locmap.h new file mode 100644 index 0000000000..e669873a14 --- /dev/null +++ b/intl/icu/source/common/locmap.h @@ -0,0 +1,40 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1996-2013, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* File locmap.h : Locale Mapping Classes +* +* +* Created by: Helena Shih +* +* Modification History: +* +* Date Name Description +* 3/11/97 aliu Added setId(). +* 4/20/99 Madhu Added T_convertToPosix() +* 09/18/00 george Removed the memory leaks. +* 08/23/01 george Convert to C +*============================================================================ +*/ + +#ifndef LOCMAP_H +#define LOCMAP_H + +#include "unicode/utypes.h" + +#define LANGUAGE_LCID(hostID) (uint16_t)(0x03FF & hostID) + +U_CAPI int32_t uprv_convertToPosix(uint32_t hostid, char* posixID, int32_t posixIDCapacity, UErrorCode* status); + +/* Don't call these functions directly. Use uloc_getLCID instead. */ +U_CAPI uint32_t uprv_convertToLCIDPlatform(const char* localeID, UErrorCode* status); // Leverage platform conversion if possible +U_CAPI uint32_t uprv_convertToLCID(const char* langID, const char* posixID, UErrorCode* status); + +#endif /* LOCMAP_H */ + diff --git a/intl/icu/source/common/locresdata.cpp b/intl/icu/source/common/locresdata.cpp new file mode 100644 index 0000000000..7a0969dff5 --- /dev/null +++ b/intl/icu/source/common/locresdata.cpp @@ -0,0 +1,220 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 1997-2012, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: loclikely.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010feb25 +* created by: Markus W. Scherer +* +* Code for miscellaneous locale-related resource bundle data access, +* separated out from other .cpp files +* that then do not depend on resource bundle code and this data. +*/ + +#include "unicode/utypes.h" +#include "unicode/putil.h" +#include "unicode/uloc.h" +#include "unicode/ures.h" +#include "cstring.h" +#include "ulocimp.h" +#include "uresimp.h" + +/* + * Lookup a resource bundle table item with fallback on the table level. + * Regular resource bundle lookups perform fallback to parent locale bundles + * and eventually the root bundle, but only for top-level items. + * This function takes the name of a top-level table and of an item in that table + * and performs a lookup of both, falling back until a bundle contains a table + * with this item. + * + * Note: Only the opening of entire bundles falls back through the default locale + * before root. Once a bundle is open, item lookups do not go through the + * default locale because that would result in a mix of languages that is + * unpredictable to the programmer and most likely useless. + */ +U_CAPI const char16_t * U_EXPORT2 +uloc_getTableStringWithFallback(const char *path, const char *locale, + const char *tableKey, const char *subTableKey, + const char *itemKey, + int32_t *pLength, + UErrorCode *pErrorCode) +{ +/* char localeBuffer[ULOC_FULLNAME_CAPACITY*4];*/ + const char16_t *item=nullptr; + UErrorCode errorCode; + char explicitFallbackName[ULOC_FULLNAME_CAPACITY] = {0}; + + /* + * open the bundle for the current locale + * this falls back through the locale's chain to root + */ + errorCode=U_ZERO_ERROR; + icu::LocalUResourceBundlePointer rb(ures_open(path, locale, &errorCode)); + + if(U_FAILURE(errorCode)) { + /* total failure, not even root could be opened */ + *pErrorCode=errorCode; + return nullptr; + } else if(errorCode==U_USING_DEFAULT_WARNING || + (errorCode==U_USING_FALLBACK_WARNING && *pErrorCode!=U_USING_DEFAULT_WARNING) + ) { + /* set the "strongest" error code (success->fallback->default->failure) */ + *pErrorCode=errorCode; + } + + for(;;){ + icu::StackUResourceBundle table; + icu::StackUResourceBundle subTable; + ures_getByKeyWithFallback(rb.getAlias(), tableKey, table.getAlias(), &errorCode); + + if (subTableKey != nullptr) { + /* + ures_getByKeyWithFallback(table.getAlias(), subTableKey, subTable.getAlias(), &errorCode); + item = ures_getStringByKeyWithFallback(subTable.getAlias(), itemKey, pLength, &errorCode); + if(U_FAILURE(errorCode)){ + *pErrorCode = errorCode; + } + + break;*/ + + ures_getByKeyWithFallback(table.getAlias(), subTableKey, table.getAlias(), &errorCode); + } + if(U_SUCCESS(errorCode)){ + item = ures_getStringByKeyWithFallback(table.getAlias(), itemKey, pLength, &errorCode); + if(U_FAILURE(errorCode)){ + const char* replacement = nullptr; + *pErrorCode = errorCode; /*save the errorCode*/ + errorCode = U_ZERO_ERROR; + /* may be a deprecated code */ + if(uprv_strcmp(tableKey, "Countries")==0){ + replacement = uloc_getCurrentCountryID(itemKey); + }else if(uprv_strcmp(tableKey, "Languages")==0){ + replacement = uloc_getCurrentLanguageID(itemKey); + } + /*pointer comparison is ok since uloc_getCurrentCountryID & uloc_getCurrentLanguageID return the key itself is replacement is not found*/ + if(replacement!=nullptr && itemKey != replacement){ + item = ures_getStringByKeyWithFallback(table.getAlias(), replacement, pLength, &errorCode); + if(U_SUCCESS(errorCode)){ + *pErrorCode = errorCode; + break; + } + } + }else{ + break; + } + } + + if(U_FAILURE(errorCode)){ + + /* still can't figure out ?.. try the fallback mechanism */ + int32_t len = 0; + const char16_t* fallbackLocale = nullptr; + *pErrorCode = errorCode; + errorCode = U_ZERO_ERROR; + + fallbackLocale = ures_getStringByKeyWithFallback(table.getAlias(), "Fallback", &len, &errorCode); + if(U_FAILURE(errorCode)){ + *pErrorCode = errorCode; + break; + } + + u_UCharsToChars(fallbackLocale, explicitFallbackName, len); + + /* guard against recursive fallback */ + if(uprv_strcmp(explicitFallbackName, locale)==0){ + *pErrorCode = U_INTERNAL_PROGRAM_ERROR; + break; + } + rb.adoptInstead(ures_open(path, explicitFallbackName, &errorCode)); + if(U_FAILURE(errorCode)){ + *pErrorCode = errorCode; + break; + } + /* succeeded in opening the fallback bundle .. continue and try to fetch the item */ + }else{ + break; + } + } + + return item; +} + +static ULayoutType +_uloc_getOrientationHelper(const char* localeId, + const char* key, + UErrorCode *status) +{ + ULayoutType result = ULOC_LAYOUT_UNKNOWN; + + if (!U_FAILURE(*status)) { + int32_t length = 0; + char localeBuffer[ULOC_FULLNAME_CAPACITY]; + + uloc_canonicalize(localeId, localeBuffer, sizeof(localeBuffer), status); + + if (!U_FAILURE(*status)) { + const char16_t* const value = + uloc_getTableStringWithFallback( + nullptr, + localeBuffer, + "layout", + nullptr, + key, + &length, + status); + + if (!U_FAILURE(*status) && length != 0) { + switch(value[0]) + { + case 0x0062: /* 'b' */ + result = ULOC_LAYOUT_BTT; + break; + case 0x006C: /* 'l' */ + result = ULOC_LAYOUT_LTR; + break; + case 0x0072: /* 'r' */ + result = ULOC_LAYOUT_RTL; + break; + case 0x0074: /* 't' */ + result = ULOC_LAYOUT_TTB; + break; + default: + *status = U_INTERNAL_PROGRAM_ERROR; + break; + } + } + } + } + + return result; +} + +U_CAPI ULayoutType U_EXPORT2 +uloc_getCharacterOrientation(const char* localeId, + UErrorCode *status) +{ + return _uloc_getOrientationHelper(localeId, "characters", status); +} + +/** + * Get the layout line orientation for the specified locale. + * + * @param localeID locale name + * @param status Error status + * @return an enum indicating the layout orientation for lines. + */ +U_CAPI ULayoutType U_EXPORT2 +uloc_getLineOrientation(const char* localeId, + UErrorCode *status) +{ + return _uloc_getOrientationHelper(localeId, "lines", status); +} diff --git a/intl/icu/source/common/locutil.cpp b/intl/icu/source/common/locutil.cpp new file mode 100644 index 0000000000..776d1d5963 --- /dev/null +++ b/intl/icu/source/common/locutil.cpp @@ -0,0 +1,275 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ******************************************************************************* + * Copyright (C) 2002-2014, International Business Machines Corporation and + * others. All Rights Reserved. + ******************************************************************************* + */ +#include "unicode/utypes.h" + +#if !UCONFIG_NO_SERVICE || !UCONFIG_NO_TRANSLITERATION + +#include "unicode/resbund.h" +#include "unicode/uenum.h" +#include "cmemory.h" +#include "ustrfmt.h" +#include "locutil.h" +#include "charstr.h" +#include "ucln_cmn.h" +#include "uassert.h" +#include "umutex.h" + +// see LocaleUtility::getAvailableLocaleNames +static icu::UInitOnce LocaleUtilityInitOnce {}; +static icu::Hashtable * LocaleUtility_cache = nullptr; + +#define UNDERSCORE_CHAR ((char16_t)0x005f) +#define AT_SIGN_CHAR ((char16_t)64) +#define PERIOD_CHAR ((char16_t)46) + +/* + ****************************************************************** + */ + +/** + * Release all static memory held by Locale Utility. + */ +U_CDECL_BEGIN +static UBool U_CALLCONV service_cleanup() { + if (LocaleUtility_cache) { + delete LocaleUtility_cache; + LocaleUtility_cache = nullptr; + } + return true; +} + + +static void U_CALLCONV locale_utility_init(UErrorCode &status) { + using namespace icu; + U_ASSERT(LocaleUtility_cache == nullptr); + ucln_common_registerCleanup(UCLN_COMMON_SERVICE, service_cleanup); + LocaleUtility_cache = new Hashtable(status); + if (U_FAILURE(status)) { + delete LocaleUtility_cache; + LocaleUtility_cache = nullptr; + return; + } + if (LocaleUtility_cache == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + LocaleUtility_cache->setValueDeleter(uhash_deleteHashtable); +} + +U_CDECL_END + +U_NAMESPACE_BEGIN + +UnicodeString& +LocaleUtility::canonicalLocaleString(const UnicodeString* id, UnicodeString& result) +{ + if (id == nullptr) { + result.setToBogus(); + } else { + // Fix case only (no other changes) up to the first '@' or '.' or + // end of string, whichever comes first. In 3.0 I changed this to + // stop at first '@' or '.'. It used to run out to the end of + // string. My fix makes the tests pass but is probably + // structurally incorrect. See below. [alan 3.0] + + // TODO: Doug, you might want to revise this... + result = *id; + int32_t i = 0; + int32_t end = result.indexOf(AT_SIGN_CHAR); + int32_t n = result.indexOf(PERIOD_CHAR); + if (n >= 0 && n < end) { + end = n; + } + if (end < 0) { + end = result.length(); + } + n = result.indexOf(UNDERSCORE_CHAR); + if (n < 0) { + n = end; + } + for (; i < n; ++i) { + char16_t c = result.charAt(i); + if (c >= 0x0041 && c <= 0x005a) { + c += 0x20; + result.setCharAt(i, c); + } + } + for (n = end; i < n; ++i) { + char16_t c = result.charAt(i); + if (c >= 0x0061 && c <= 0x007a) { + c -= 0x20; + result.setCharAt(i, c); + } + } + } + return result; + +#if 0 + // This code does a proper full level 2 canonicalization of id. + // It's nasty to go from char16_t to char to char to char16_t -- but + // that's what you have to do to use the uloc_canonicalize + // function on UnicodeStrings. + + // I ended up doing the alternate fix (see above) not for + // performance reasons, although performance will certainly be + // better, but because doing a full level 2 canonicalization + // causes some tests to fail. [alan 3.0] + + // TODO: Doug, you might want to revisit this... + result.setToBogus(); + if (id != 0) { + int32_t buflen = id->length() + 8; // space for NUL + char* buf = (char*) uprv_malloc(buflen); + char* canon = (buf == 0) ? 0 : (char*) uprv_malloc(buflen); + if (buf != 0 && canon != 0) { + U_ASSERT(id->extract(0, INT32_MAX, buf, buflen) < buflen); + UErrorCode ec = U_ZERO_ERROR; + uloc_canonicalize(buf, canon, buflen, &ec); + if (U_SUCCESS(ec)) { + result = UnicodeString(canon); + } + } + uprv_free(buf); + uprv_free(canon); + } + return result; +#endif +} + +Locale& +LocaleUtility::initLocaleFromName(const UnicodeString& id, Locale& result) +{ + enum { BUFLEN = 128 }; // larger than ever needed + + if (id.isBogus() || id.length() >= BUFLEN) { + result.setToBogus(); + } else { + /* + * We need to convert from a UnicodeString to char * in order to + * create a Locale. + * + * Problem: Locale ID strings may contain '@' which is a variant + * character and cannot be handled by invariant-character conversion. + * + * Hack: Since ICU code can handle locale IDs with multiple encodings + * of '@' (at least for EBCDIC; it's not known to be a problem for + * ASCII-based systems), + * we use regular invariant-character conversion for everything else + * and manually convert U+0040 into a compiler-char-constant '@'. + * While this compilation-time constant may not match the runtime + * encoding of '@', it should be one of the encodings which ICU + * recognizes. + * + * There should be only at most one '@' in a locale ID. + */ + char buffer[BUFLEN]; + int32_t prev, i; + prev = 0; + for(;;) { + i = id.indexOf((char16_t)0x40, prev); + if(i < 0) { + // no @ between prev and the rest of the string + id.extract(prev, INT32_MAX, buffer + prev, BUFLEN - prev, US_INV); + break; // done + } else { + // normal invariant-character conversion for text between @s + id.extract(prev, i - prev, buffer + prev, BUFLEN - prev, US_INV); + // manually "convert" U+0040 at id[i] into '@' at buffer[i] + buffer[i] = '@'; + prev = i + 1; + } + } + result = Locale::createFromName(buffer); + } + return result; +} + +UnicodeString& +LocaleUtility::initNameFromLocale(const Locale& locale, UnicodeString& result) +{ + if (locale.isBogus()) { + result.setToBogus(); + } else { + result.append(UnicodeString(locale.getName(), -1, US_INV)); + } + return result; +} + +const Hashtable* +LocaleUtility::getAvailableLocaleNames(const UnicodeString& bundleID) +{ + // LocaleUtility_cache is a hash-of-hashes. The top-level keys + // are path strings ('bundleID') passed to + // ures_openAvailableLocales. The top-level values are + // second-level hashes. The second-level keys are result strings + // from ures_openAvailableLocales. The second-level values are + // garbage ((void*)1 or other random pointer). + + UErrorCode status = U_ZERO_ERROR; + umtx_initOnce(LocaleUtilityInitOnce, locale_utility_init, status); + Hashtable *cache = LocaleUtility_cache; + if (cache == nullptr) { + // Catastrophic failure. + return nullptr; + } + + Hashtable* htp; + umtx_lock(nullptr); + htp = (Hashtable*) cache->get(bundleID); + umtx_unlock(nullptr); + + if (htp == nullptr) { + htp = new Hashtable(status); + if (htp && U_SUCCESS(status)) { + CharString cbundleID; + cbundleID.appendInvariantChars(bundleID, status); + const char* path = cbundleID.isEmpty() ? nullptr : cbundleID.data(); + icu::LocalUEnumerationPointer uenum(ures_openAvailableLocales(path, &status)); + for (;;) { + const char16_t* id = uenum_unext(uenum.getAlias(), nullptr, &status); + if (id == nullptr) { + break; + } + htp->put(UnicodeString(id), (void*)htp, status); + } + if (U_FAILURE(status)) { + delete htp; + return nullptr; + } + umtx_lock(nullptr); + Hashtable *t = static_cast(cache->get(bundleID)); + if (t != nullptr) { + // Another thread raced through this code, creating the cache entry first. + // Discard ours and return theirs. + umtx_unlock(nullptr); + delete htp; + htp = t; + } else { + cache->put(bundleID, (void*)htp, status); + umtx_unlock(nullptr); + } + } + } + return htp; +} + +UBool +LocaleUtility::isFallbackOf(const UnicodeString& root, const UnicodeString& child) +{ + return child.indexOf(root) == 0 && + (child.length() == root.length() || + child.charAt(root.length()) == UNDERSCORE_CHAR); +} + +U_NAMESPACE_END + +/* !UCONFIG_NO_SERVICE */ +#endif + + diff --git a/intl/icu/source/common/locutil.h b/intl/icu/source/common/locutil.h new file mode 100644 index 0000000000..31bfffd7a5 --- /dev/null +++ b/intl/icu/source/common/locutil.h @@ -0,0 +1,39 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** + ******************************************************************************* + * Copyright (C) 2002-2005, International Business Machines Corporation and * + * others. All Rights Reserved. * + ******************************************************************************* + * + ******************************************************************************* + */ +#ifndef LOCUTIL_H +#define LOCUTIL_H + +#include "unicode/utypes.h" +#include "hash.h" + +#if !UCONFIG_NO_SERVICE || !UCONFIG_NO_TRANSLITERATION + + +U_NAMESPACE_BEGIN + +// temporary utility functions, till I know where to find them +// in header so tests can also access them + +class U_COMMON_API LocaleUtility { +public: + static UnicodeString& canonicalLocaleString(const UnicodeString* id, UnicodeString& result); + static Locale& initLocaleFromName(const UnicodeString& id, Locale& result); + static UnicodeString& initNameFromLocale(const Locale& locale, UnicodeString& result); + static const Hashtable* getAvailableLocaleNames(const UnicodeString& bundleID); + static UBool isFallbackOf(const UnicodeString& root, const UnicodeString& child); +}; + +U_NAMESPACE_END + + +#endif + +#endif diff --git a/intl/icu/source/common/lsr.cpp b/intl/icu/source/common/lsr.cpp new file mode 100644 index 0000000000..39eb46df27 --- /dev/null +++ b/intl/icu/source/common/lsr.cpp @@ -0,0 +1,114 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// lsr.cpp +// created: 2019may08 Markus W. Scherer + +#include "unicode/utypes.h" +#include "charstr.h" +#include "cmemory.h" +#include "cstring.h" +#include "lsr.h" +#include "uinvchar.h" +#include "ustr_imp.h" + +U_NAMESPACE_BEGIN + +LSR::LSR(char prefix, const char *lang, const char *scr, const char *r, int32_t f, + UErrorCode &errorCode) : + language(nullptr), script(nullptr), region(r), + regionIndex(indexForRegion(region)), flags(f) { + if (U_SUCCESS(errorCode)) { + CharString langScript; + langScript.append(prefix, errorCode).append(lang, errorCode).append('\0', errorCode); + int32_t scriptOffset = langScript.length(); + langScript.append(prefix, errorCode).append(scr, errorCode); + owned = langScript.cloneData(errorCode); + if (U_SUCCESS(errorCode)) { + language = owned; + script = owned + scriptOffset; + } + } +} + +LSR::LSR(LSR &&other) noexcept : + language(other.language), script(other.script), region(other.region), owned(other.owned), + regionIndex(other.regionIndex), flags(other.flags), + hashCode(other.hashCode) { + if (owned != nullptr) { + other.language = other.script = ""; + other.owned = nullptr; + other.hashCode = 0; + } +} + +void LSR::deleteOwned() { + uprv_free(owned); +} + +LSR &LSR::operator=(LSR &&other) noexcept { + this->~LSR(); + language = other.language; + script = other.script; + region = other.region; + regionIndex = other.regionIndex; + flags = other.flags; + owned = other.owned; + hashCode = other.hashCode; + if (owned != nullptr) { + other.language = other.script = ""; + other.owned = nullptr; + other.hashCode = 0; + } + return *this; +} + +UBool LSR::isEquivalentTo(const LSR &other) const { + return + uprv_strcmp(language, other.language) == 0 && + uprv_strcmp(script, other.script) == 0 && + regionIndex == other.regionIndex && + // Compare regions if both are ill-formed (and their indexes are 0). + (regionIndex > 0 || uprv_strcmp(region, other.region) == 0); +} + +bool LSR::operator==(const LSR &other) const { + return + uprv_strcmp(language, other.language) == 0 && + uprv_strcmp(script, other.script) == 0 && + regionIndex == other.regionIndex && + // Compare regions if both are ill-formed (and their indexes are 0). + (regionIndex > 0 || uprv_strcmp(region, other.region) == 0) && + flags == other.flags; +} + +int32_t LSR::indexForRegion(const char *region) { + int32_t c = region[0]; + int32_t a = c - '0'; + if (0 <= a && a <= 9) { // digits: "419" + int32_t b = region[1] - '0'; + if (b < 0 || 9 < b) { return 0; } + c = region[2] - '0'; + if (c < 0 || 9 < c || region[3] != 0) { return 0; } + return (10 * a + b) * 10 + c + 1; + } else { // letters: "DE" + a = uprv_upperOrdinal(c); + if (a < 0 || 25 < a) { return 0; } + int32_t b = uprv_upperOrdinal(region[1]); + if (b < 0 || 25 < b || region[2] != 0) { return 0; } + return 26 * a + b + 1001; + } + return 0; +} + +LSR &LSR::setHashCode() { + if (hashCode == 0) { + uint32_t h = ustr_hashCharsN(language, static_cast(uprv_strlen(language))); + h = h * 37 + ustr_hashCharsN(script, static_cast(uprv_strlen(script))); + h = h * 37 + regionIndex; + hashCode = h * 37 + flags; + } + return *this; +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/lsr.h b/intl/icu/source/common/lsr.h new file mode 100644 index 0000000000..a2f7c8bb15 --- /dev/null +++ b/intl/icu/source/common/lsr.h @@ -0,0 +1,82 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// lsr.h +// created: 2019may08 Markus W. Scherer + +#ifndef __LSR_H__ +#define __LSR_H__ + +#include "unicode/utypes.h" +#include "unicode/uobject.h" +#include "cstring.h" + +U_NAMESPACE_BEGIN + +struct LSR final : public UMemory { + static constexpr int32_t REGION_INDEX_LIMIT = 1001 + 26 * 26; + + static constexpr int32_t EXPLICIT_LSR = 7; + static constexpr int32_t EXPLICIT_LANGUAGE = 4; + static constexpr int32_t EXPLICIT_SCRIPT = 2; + static constexpr int32_t EXPLICIT_REGION = 1; + static constexpr int32_t IMPLICIT_LSR = 0; + static constexpr int32_t DONT_CARE_FLAGS = 0; + + const char *language; + const char *script; + const char *region; + char *owned = nullptr; + /** Index for region, 0 if ill-formed. @see indexForRegion */ + int32_t regionIndex = 0; + int32_t flags = 0; + /** Only set for LSRs that will be used in a hash table. */ + int32_t hashCode = 0; + + LSR() : language("und"), script(""), region("") {} + + /** Constructor which aliases all subtag pointers. */ + LSR(const char *lang, const char *scr, const char *r, int32_t f) : + language(lang), script(scr), region(r), + regionIndex(indexForRegion(region)), flags(f) {} + /** + * Constructor which prepends the prefix to the language and script, + * copies those into owned memory, and aliases the region. + */ + LSR(char prefix, const char *lang, const char *scr, const char *r, int32_t f, + UErrorCode &errorCode); + LSR(LSR &&other) noexcept; + LSR(const LSR &other) = delete; + inline ~LSR() { + // Pure inline code for almost all instances. + if (owned != nullptr) { + deleteOwned(); + } + } + + LSR &operator=(LSR &&other) noexcept; + LSR &operator=(const LSR &other) = delete; + + /** + * Returns a positive index (>0) for a well-formed region code. + * Do not rely on a particular region->index mapping; it may change. + * Returns 0 for ill-formed strings. + */ + static int32_t indexForRegion(const char *region); + + UBool isEquivalentTo(const LSR &other) const; + bool operator==(const LSR &other) const; + + inline bool operator!=(const LSR &other) const { + return !operator==(other); + } + + LSR &setHashCode(); + +private: + void deleteOwned(); +}; + +U_NAMESPACE_END + +#endif // __LSR_H__ diff --git a/intl/icu/source/common/lstmbe.cpp b/intl/icu/source/common/lstmbe.cpp new file mode 100644 index 0000000000..fb8eb01761 --- /dev/null +++ b/intl/icu/source/common/lstmbe.cpp @@ -0,0 +1,856 @@ +// © 2021 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +#include +#include + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "brkeng.h" +#include "charstr.h" +#include "cmemory.h" +#include "lstmbe.h" +#include "putilimp.h" +#include "uassert.h" +#include "ubrkimpl.h" +#include "uresimp.h" +#include "uvectr32.h" +#include "uvector.h" + +#include "unicode/brkiter.h" +#include "unicode/resbund.h" +#include "unicode/ubrk.h" +#include "unicode/uniset.h" +#include "unicode/ustring.h" +#include "unicode/utf.h" + +U_NAMESPACE_BEGIN + +// Uncomment the following #define to debug. +// #define LSTM_DEBUG 1 +// #define LSTM_VECTORIZER_DEBUG 1 + +/** + * Interface for reading 1D array. + */ +class ReadArray1D { +public: + virtual ~ReadArray1D(); + virtual int32_t d1() const = 0; + virtual float get(int32_t i) const = 0; + +#ifdef LSTM_DEBUG + void print() const { + printf("\n["); + for (int32_t i = 0; i < d1(); i++) { + printf("%0.8e ", get(i)); + if (i % 4 == 3) printf("\n"); + } + printf("]\n"); + } +#endif +}; + +ReadArray1D::~ReadArray1D() +{ +} + +/** + * Interface for reading 2D array. + */ +class ReadArray2D { +public: + virtual ~ReadArray2D(); + virtual int32_t d1() const = 0; + virtual int32_t d2() const = 0; + virtual float get(int32_t i, int32_t j) const = 0; +}; + +ReadArray2D::~ReadArray2D() +{ +} + +/** + * A class to index a float array as a 1D Array without owning the pointer or + * copy the data. + */ +class ConstArray1D : public ReadArray1D { +public: + ConstArray1D() : data_(nullptr), d1_(0) {} + + ConstArray1D(const float* data, int32_t d1) : data_(data), d1_(d1) {} + + virtual ~ConstArray1D(); + + // Init the object, the object does not own the data nor copy. + // It is designed to directly use data from memory mapped resources. + void init(const int32_t* data, int32_t d1) { + U_ASSERT(IEEE_754 == 1); + data_ = reinterpret_cast(data); + d1_ = d1; + } + + // ReadArray1D methods. + virtual int32_t d1() const override { return d1_; } + virtual float get(int32_t i) const override { + U_ASSERT(i < d1_); + return data_[i]; + } + +private: + const float* data_; + int32_t d1_; +}; + +ConstArray1D::~ConstArray1D() +{ +} + +/** + * A class to index a float array as a 2D Array without owning the pointer or + * copy the data. + */ +class ConstArray2D : public ReadArray2D { +public: + ConstArray2D() : data_(nullptr), d1_(0), d2_(0) {} + + ConstArray2D(const float* data, int32_t d1, int32_t d2) + : data_(data), d1_(d1), d2_(d2) {} + + virtual ~ConstArray2D(); + + // Init the object, the object does not own the data nor copy. + // It is designed to directly use data from memory mapped resources. + void init(const int32_t* data, int32_t d1, int32_t d2) { + U_ASSERT(IEEE_754 == 1); + data_ = reinterpret_cast(data); + d1_ = d1; + d2_ = d2; + } + + // ReadArray2D methods. + inline int32_t d1() const override { return d1_; } + inline int32_t d2() const override { return d2_; } + float get(int32_t i, int32_t j) const override { + U_ASSERT(i < d1_); + U_ASSERT(j < d2_); + return data_[i * d2_ + j]; + } + + // Expose the ith row as a ConstArray1D + inline ConstArray1D row(int32_t i) const { + U_ASSERT(i < d1_); + return ConstArray1D(data_ + i * d2_, d2_); + } + +private: + const float* data_; + int32_t d1_; + int32_t d2_; +}; + +ConstArray2D::~ConstArray2D() +{ +} + +/** + * A class to allocate data as a writable 1D array. + * This is the main class implement matrix operation. + */ +class Array1D : public ReadArray1D { +public: + Array1D() : memory_(nullptr), data_(nullptr), d1_(0) {} + Array1D(int32_t d1, UErrorCode &status) + : memory_(uprv_malloc(d1 * sizeof(float))), + data_((float*)memory_), d1_(d1) { + if (U_SUCCESS(status)) { + if (memory_ == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + clear(); + } + } + + virtual ~Array1D(); + + // A special constructor which does not own the memory but writeable + // as a slice of an array. + Array1D(float* data, int32_t d1) + : memory_(nullptr), data_(data), d1_(d1) {} + + // ReadArray1D methods. + virtual int32_t d1() const override { return d1_; } + virtual float get(int32_t i) const override { + U_ASSERT(i < d1_); + return data_[i]; + } + + // Return the index which point to the max data in the array. + inline int32_t maxIndex() const { + int32_t index = 0; + float max = data_[0]; + for (int32_t i = 1; i < d1_; i++) { + if (data_[i] > max) { + max = data_[i]; + index = i; + } + } + return index; + } + + // Slice part of the array to a new one. + inline Array1D slice(int32_t from, int32_t size) const { + U_ASSERT(from >= 0); + U_ASSERT(from < d1_); + U_ASSERT(from + size <= d1_); + return Array1D(data_ + from, size); + } + + // Add dot product of a 1D array and a 2D array into this one. + inline Array1D& addDotProduct(const ReadArray1D& a, const ReadArray2D& b) { + U_ASSERT(a.d1() == b.d1()); + U_ASSERT(b.d2() == d1()); + for (int32_t i = 0; i < d1(); i++) { + for (int32_t j = 0; j < a.d1(); j++) { + data_[i] += a.get(j) * b.get(j, i); + } + } + return *this; + } + + // Hadamard Product the values of another array of the same size into this one. + inline Array1D& hadamardProduct(const ReadArray1D& a) { + U_ASSERT(a.d1() == d1()); + for (int32_t i = 0; i < d1(); i++) { + data_[i] *= a.get(i); + } + return *this; + } + + // Add the Hadamard Product of two arrays of the same size into this one. + inline Array1D& addHadamardProduct(const ReadArray1D& a, const ReadArray1D& b) { + U_ASSERT(a.d1() == d1()); + U_ASSERT(b.d1() == d1()); + for (int32_t i = 0; i < d1(); i++) { + data_[i] += a.get(i) * b.get(i); + } + return *this; + } + + // Add the values of another array of the same size into this one. + inline Array1D& add(const ReadArray1D& a) { + U_ASSERT(a.d1() == d1()); + for (int32_t i = 0; i < d1(); i++) { + data_[i] += a.get(i); + } + return *this; + } + + // Assign the values of another array of the same size into this one. + inline Array1D& assign(const ReadArray1D& a) { + U_ASSERT(a.d1() == d1()); + for (int32_t i = 0; i < d1(); i++) { + data_[i] = a.get(i); + } + return *this; + } + + // Apply tanh to all the elements in the array. + inline Array1D& tanh() { + return tanh(*this); + } + + // Apply tanh of a and store into this array. + inline Array1D& tanh(const Array1D& a) { + U_ASSERT(a.d1() == d1()); + for (int32_t i = 0; i < d1_; i++) { + data_[i] = std::tanh(a.get(i)); + } + return *this; + } + + // Apply sigmoid to all the elements in the array. + inline Array1D& sigmoid() { + for (int32_t i = 0; i < d1_; i++) { + data_[i] = 1.0f/(1.0f + expf(-data_[i])); + } + return *this; + } + + inline Array1D& clear() { + uprv_memset(data_, 0, d1_ * sizeof(float)); + return *this; + } + +private: + void* memory_; + float* data_; + int32_t d1_; +}; + +Array1D::~Array1D() +{ + uprv_free(memory_); +} + +class Array2D : public ReadArray2D { +public: + Array2D() : memory_(nullptr), data_(nullptr), d1_(0), d2_(0) {} + Array2D(int32_t d1, int32_t d2, UErrorCode &status) + : memory_(uprv_malloc(d1 * d2 * sizeof(float))), + data_((float*)memory_), d1_(d1), d2_(d2) { + if (U_SUCCESS(status)) { + if (memory_ == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + clear(); + } + } + virtual ~Array2D(); + + // ReadArray2D methods. + virtual int32_t d1() const override { return d1_; } + virtual int32_t d2() const override { return d2_; } + virtual float get(int32_t i, int32_t j) const override { + U_ASSERT(i < d1_); + U_ASSERT(j < d2_); + return data_[i * d2_ + j]; + } + + inline Array1D row(int32_t i) const { + U_ASSERT(i < d1_); + return Array1D(data_ + i * d2_, d2_); + } + + inline Array2D& clear() { + uprv_memset(data_, 0, d1_ * d2_ * sizeof(float)); + return *this; + } + +private: + void* memory_; + float* data_; + int32_t d1_; + int32_t d2_; +}; + +Array2D::~Array2D() +{ + uprv_free(memory_); +} + +typedef enum { + BEGIN, + INSIDE, + END, + SINGLE +} LSTMClass; + +typedef enum { + UNKNOWN, + CODE_POINTS, + GRAPHEME_CLUSTER, +} EmbeddingType; + +struct LSTMData : public UMemory { + LSTMData(UResourceBundle* rb, UErrorCode &status); + ~LSTMData(); + UHashtable* fDict; + EmbeddingType fType; + const char16_t* fName; + ConstArray2D fEmbedding; + ConstArray2D fForwardW; + ConstArray2D fForwardU; + ConstArray1D fForwardB; + ConstArray2D fBackwardW; + ConstArray2D fBackwardU; + ConstArray1D fBackwardB; + ConstArray2D fOutputW; + ConstArray1D fOutputB; + +private: + UResourceBundle* fBundle; +}; + +LSTMData::LSTMData(UResourceBundle* rb, UErrorCode &status) + : fDict(nullptr), fType(UNKNOWN), fName(nullptr), + fBundle(rb) +{ + if (U_FAILURE(status)) { + return; + } + if (IEEE_754 != 1) { + status = U_UNSUPPORTED_ERROR; + return; + } + LocalUResourceBundlePointer embeddings_res( + ures_getByKey(rb, "embeddings", nullptr, &status)); + int32_t embedding_size = ures_getInt(embeddings_res.getAlias(), &status); + LocalUResourceBundlePointer hunits_res( + ures_getByKey(rb, "hunits", nullptr, &status)); + if (U_FAILURE(status)) return; + int32_t hunits = ures_getInt(hunits_res.getAlias(), &status); + const char16_t* type = ures_getStringByKey(rb, "type", nullptr, &status); + if (U_FAILURE(status)) return; + if (u_strCompare(type, -1, u"codepoints", -1, false) == 0) { + fType = CODE_POINTS; + } else if (u_strCompare(type, -1, u"graphclust", -1, false) == 0) { + fType = GRAPHEME_CLUSTER; + } + fName = ures_getStringByKey(rb, "model", nullptr, &status); + LocalUResourceBundlePointer dataRes(ures_getByKey(rb, "data", nullptr, &status)); + if (U_FAILURE(status)) return; + int32_t data_len = 0; + const int32_t* data = ures_getIntVector(dataRes.getAlias(), &data_len, &status); + fDict = uhash_open(uhash_hashUChars, uhash_compareUChars, nullptr, &status); + + StackUResourceBundle stackTempBundle; + ResourceDataValue value; + ures_getValueWithFallback(rb, "dict", stackTempBundle.getAlias(), value, status); + ResourceArray stringArray = value.getArray(status); + int32_t num_index = stringArray.getSize(); + if (U_FAILURE(status)) { return; } + + // put dict into hash + int32_t stringLength; + for (int32_t idx = 0; idx < num_index; idx++) { + stringArray.getValue(idx, value); + const char16_t* str = value.getString(stringLength, status); + uhash_putiAllowZero(fDict, (void*)str, idx, &status); + if (U_FAILURE(status)) return; +#ifdef LSTM_VECTORIZER_DEBUG + printf("Assign ["); + while (*str != 0x0000) { + printf("U+%04x ", *str); + str++; + } + printf("] map to %d\n", idx-1); +#endif + } + int32_t mat1_size = (num_index + 1) * embedding_size; + int32_t mat2_size = embedding_size * 4 * hunits; + int32_t mat3_size = hunits * 4 * hunits; + int32_t mat4_size = 4 * hunits; + int32_t mat5_size = mat2_size; + int32_t mat6_size = mat3_size; + int32_t mat7_size = mat4_size; + int32_t mat8_size = 2 * hunits * 4; +#if U_DEBUG + int32_t mat9_size = 4; + U_ASSERT(data_len == mat1_size + mat2_size + mat3_size + mat4_size + mat5_size + + mat6_size + mat7_size + mat8_size + mat9_size); +#endif + + fEmbedding.init(data, (num_index + 1), embedding_size); + data += mat1_size; + fForwardW.init(data, embedding_size, 4 * hunits); + data += mat2_size; + fForwardU.init(data, hunits, 4 * hunits); + data += mat3_size; + fForwardB.init(data, 4 * hunits); + data += mat4_size; + fBackwardW.init(data, embedding_size, 4 * hunits); + data += mat5_size; + fBackwardU.init(data, hunits, 4 * hunits); + data += mat6_size; + fBackwardB.init(data, 4 * hunits); + data += mat7_size; + fOutputW.init(data, 2 * hunits, 4); + data += mat8_size; + fOutputB.init(data, 4); +} + +LSTMData::~LSTMData() { + uhash_close(fDict); + ures_close(fBundle); +} + +class Vectorizer : public UMemory { +public: + Vectorizer(UHashtable* dict) : fDict(dict) {} + virtual ~Vectorizer(); + virtual void vectorize(UText *text, int32_t startPos, int32_t endPos, + UVector32 &offsets, UVector32 &indices, + UErrorCode &status) const = 0; +protected: + int32_t stringToIndex(const char16_t* str) const { + UBool found = false; + int32_t ret = uhash_getiAndFound(fDict, (const void*)str, &found); + if (!found) { + ret = fDict->count; + } +#ifdef LSTM_VECTORIZER_DEBUG + printf("["); + while (*str != 0x0000) { + printf("U+%04x ", *str); + str++; + } + printf("] map to %d\n", ret); +#endif + return ret; + } + +private: + UHashtable* fDict; +}; + +Vectorizer::~Vectorizer() +{ +} + +class CodePointsVectorizer : public Vectorizer { +public: + CodePointsVectorizer(UHashtable* dict) : Vectorizer(dict) {} + virtual ~CodePointsVectorizer(); + virtual void vectorize(UText *text, int32_t startPos, int32_t endPos, + UVector32 &offsets, UVector32 &indices, + UErrorCode &status) const override; +}; + +CodePointsVectorizer::~CodePointsVectorizer() +{ +} + +void CodePointsVectorizer::vectorize( + UText *text, int32_t startPos, int32_t endPos, + UVector32 &offsets, UVector32 &indices, UErrorCode &status) const +{ + if (offsets.ensureCapacity(endPos - startPos, status) && + indices.ensureCapacity(endPos - startPos, status)) { + if (U_FAILURE(status)) return; + utext_setNativeIndex(text, startPos); + int32_t current; + char16_t str[2] = {0, 0}; + while (U_SUCCESS(status) && + (current = (int32_t)utext_getNativeIndex(text)) < endPos) { + // Since the LSTMBreakEngine is currently only accept chars in BMP, + // we can ignore the possibility of hitting supplementary code + // point. + str[0] = (char16_t) utext_next32(text); + U_ASSERT(!U_IS_SURROGATE(str[0])); + offsets.addElement(current, status); + indices.addElement(stringToIndex(str), status); + } + } +} + +class GraphemeClusterVectorizer : public Vectorizer { +public: + GraphemeClusterVectorizer(UHashtable* dict) + : Vectorizer(dict) + { + } + virtual ~GraphemeClusterVectorizer(); + virtual void vectorize(UText *text, int32_t startPos, int32_t endPos, + UVector32 &offsets, UVector32 &indices, + UErrorCode &status) const override; +}; + +GraphemeClusterVectorizer::~GraphemeClusterVectorizer() +{ +} + +constexpr int32_t MAX_GRAPHEME_CLSTER_LENGTH = 10; + +void GraphemeClusterVectorizer::vectorize( + UText *text, int32_t startPos, int32_t endPos, + UVector32 &offsets, UVector32 &indices, UErrorCode &status) const +{ + if (U_FAILURE(status)) return; + if (!offsets.ensureCapacity(endPos - startPos, status) || + !indices.ensureCapacity(endPos - startPos, status)) { + return; + } + if (U_FAILURE(status)) return; + LocalPointer graphemeIter(BreakIterator::createCharacterInstance(Locale(), status)); + if (U_FAILURE(status)) return; + graphemeIter->setText(text, status); + if (U_FAILURE(status)) return; + + if (startPos != 0) { + graphemeIter->preceding(startPos); + } + int32_t last = startPos; + int32_t current = startPos; + char16_t str[MAX_GRAPHEME_CLSTER_LENGTH]; + while ((current = graphemeIter->next()) != BreakIterator::DONE) { + if (current >= endPos) { + break; + } + if (current > startPos) { + utext_extract(text, last, current, str, MAX_GRAPHEME_CLSTER_LENGTH, &status); + if (U_FAILURE(status)) return; + offsets.addElement(last, status); + indices.addElement(stringToIndex(str), status); + if (U_FAILURE(status)) return; + } + last = current; + } + if (U_FAILURE(status) || last >= endPos) { + return; + } + utext_extract(text, last, endPos, str, MAX_GRAPHEME_CLSTER_LENGTH, &status); + if (U_SUCCESS(status)) { + offsets.addElement(last, status); + indices.addElement(stringToIndex(str), status); + } +} + +// Computing LSTM as stated in +// https://en.wikipedia.org/wiki/Long_short-term_memory#LSTM_with_a_forget_gate +// ifco is temp array allocate outside which does not need to be +// input/output value but could avoid unnecessary memory alloc/free if passing +// in. +void compute( + int32_t hunits, + const ReadArray2D& W, const ReadArray2D& U, const ReadArray1D& b, + const ReadArray1D& x, Array1D& h, Array1D& c, + Array1D& ifco) +{ + // ifco = x * W + h * U + b + ifco.assign(b) + .addDotProduct(x, W) + .addDotProduct(h, U); + + ifco.slice(0*hunits, hunits).sigmoid(); // i: sigmod + ifco.slice(1*hunits, hunits).sigmoid(); // f: sigmoid + ifco.slice(2*hunits, hunits).tanh(); // c_: tanh + ifco.slice(3*hunits, hunits).sigmoid(); // o: sigmod + + c.hadamardProduct(ifco.slice(hunits, hunits)) + .addHadamardProduct(ifco.slice(0, hunits), ifco.slice(2*hunits, hunits)); + + h.tanh(c) + .hadamardProduct(ifco.slice(3*hunits, hunits)); +} + +// Minimum word size +static const int32_t MIN_WORD = 2; + +// Minimum number of characters for two words +static const int32_t MIN_WORD_SPAN = MIN_WORD * 2; + +int32_t +LSTMBreakEngine::divideUpDictionaryRange( UText *text, + int32_t startPos, + int32_t endPos, + UVector32 &foundBreaks, + UBool /* isPhraseBreaking */, + UErrorCode& status) const { + if (U_FAILURE(status)) return 0; + int32_t beginFoundBreakSize = foundBreaks.size(); + utext_setNativeIndex(text, startPos); + utext_moveIndex32(text, MIN_WORD_SPAN); + if (utext_getNativeIndex(text) >= endPos) { + return 0; // Not enough characters for two words + } + utext_setNativeIndex(text, startPos); + + UVector32 offsets(status); + UVector32 indices(status); + if (U_FAILURE(status)) return 0; + fVectorizer->vectorize(text, startPos, endPos, offsets, indices, status); + if (U_FAILURE(status)) return 0; + int32_t* offsetsBuf = offsets.getBuffer(); + int32_t* indicesBuf = indices.getBuffer(); + + int32_t input_seq_len = indices.size(); + int32_t hunits = fData->fForwardU.d1(); + + // ----- Begin of all the Array memory allocation needed for this function + // Allocate temp array used inside compute() + Array1D ifco(4 * hunits, status); + + Array1D c(hunits, status); + Array1D logp(4, status); + + // TODO: limit size of hBackward. If input_seq_len is too big, we could + // run out of memory. + // Backward LSTM + Array2D hBackward(input_seq_len, hunits, status); + + // Allocate fbRow and slice the internal array in two. + Array1D fbRow(2 * hunits, status); + + // ----- End of all the Array memory allocation needed for this function + if (U_FAILURE(status)) return 0; + + // To save the needed memory usage, the following is different from the + // Python or ICU4X implementation. We first perform the Backward LSTM + // and then merge the iteration of the forward LSTM and the output layer + // together because we only neetdto remember the h[t-1] for Forward LSTM. + for (int32_t i = input_seq_len - 1; i >= 0; i--) { + Array1D hRow = hBackward.row(i); + if (i != input_seq_len - 1) { + hRow.assign(hBackward.row(i+1)); + } +#ifdef LSTM_DEBUG + printf("hRow %d\n", i); + hRow.print(); + printf("indicesBuf[%d] = %d\n", i, indicesBuf[i]); + printf("fData->fEmbedding.row(indicesBuf[%d]):\n", i); + fData->fEmbedding.row(indicesBuf[i]).print(); +#endif // LSTM_DEBUG + compute(hunits, + fData->fBackwardW, fData->fBackwardU, fData->fBackwardB, + fData->fEmbedding.row(indicesBuf[i]), + hRow, c, ifco); + } + + + Array1D forwardRow = fbRow.slice(0, hunits); // point to first half of data in fbRow. + Array1D backwardRow = fbRow.slice(hunits, hunits); // point to second half of data n fbRow. + + // The following iteration merge the forward LSTM and the output layer + // together. + c.clear(); // reuse c since it is the same size. + for (int32_t i = 0; i < input_seq_len; i++) { +#ifdef LSTM_DEBUG + printf("forwardRow %d\n", i); + forwardRow.print(); +#endif // LSTM_DEBUG + // Forward LSTM + // Calculate the result into forwardRow, which point to the data in the first half + // of fbRow. + compute(hunits, + fData->fForwardW, fData->fForwardU, fData->fForwardB, + fData->fEmbedding.row(indicesBuf[i]), + forwardRow, c, ifco); + + // assign the data from hBackward.row(i) to second half of fbRowa. + backwardRow.assign(hBackward.row(i)); + + logp.assign(fData->fOutputB).addDotProduct(fbRow, fData->fOutputW); +#ifdef LSTM_DEBUG + printf("backwardRow %d\n", i); + backwardRow.print(); + printf("logp %d\n", i); + logp.print(); +#endif // LSTM_DEBUG + + // current = argmax(logp) + LSTMClass current = (LSTMClass)logp.maxIndex(); + // BIES logic. + if (current == BEGIN || current == SINGLE) { + if (i != 0) { + foundBreaks.addElement(offsetsBuf[i], status); + if (U_FAILURE(status)) return 0; + } + } + } + return foundBreaks.size() - beginFoundBreakSize; +} + +Vectorizer* createVectorizer(const LSTMData* data, UErrorCode &status) { + if (U_FAILURE(status)) { + return nullptr; + } + switch (data->fType) { + case CODE_POINTS: + return new CodePointsVectorizer(data->fDict); + break; + case GRAPHEME_CLUSTER: + return new GraphemeClusterVectorizer(data->fDict); + break; + default: + break; + } + UPRV_UNREACHABLE_EXIT; +} + +LSTMBreakEngine::LSTMBreakEngine(const LSTMData* data, const UnicodeSet& set, UErrorCode &status) + : DictionaryBreakEngine(), fData(data), fVectorizer(createVectorizer(fData, status)) +{ + if (U_FAILURE(status)) { + fData = nullptr; // If failure, we should not delete fData in destructor because the caller will do so. + return; + } + setCharacters(set); +} + +LSTMBreakEngine::~LSTMBreakEngine() { + delete fData; + delete fVectorizer; +} + +const char16_t* LSTMBreakEngine::name() const { + return fData->fName; +} + +UnicodeString defaultLSTM(UScriptCode script, UErrorCode& status) { + // open root from brkitr tree. + UResourceBundle *b = ures_open(U_ICUDATA_BRKITR, "", &status); + b = ures_getByKeyWithFallback(b, "lstm", b, &status); + UnicodeString result = ures_getUnicodeStringByKey(b, uscript_getShortName(script), &status); + ures_close(b); + return result; +} + +U_CAPI const LSTMData* U_EXPORT2 CreateLSTMDataForScript(UScriptCode script, UErrorCode& status) +{ + if (script != USCRIPT_KHMER && script != USCRIPT_LAO && script != USCRIPT_MYANMAR && script != USCRIPT_THAI) { + return nullptr; + } + UnicodeString name = defaultLSTM(script, status); + if (U_FAILURE(status)) return nullptr; + CharString namebuf; + namebuf.appendInvariantChars(name, status).truncate(namebuf.lastIndexOf('.')); + + LocalUResourceBundlePointer rb( + ures_openDirect(U_ICUDATA_BRKITR, namebuf.data(), &status)); + if (U_FAILURE(status)) return nullptr; + + return CreateLSTMData(rb.orphan(), status); +} + +U_CAPI const LSTMData* U_EXPORT2 CreateLSTMData(UResourceBundle* rb, UErrorCode& status) +{ + return new LSTMData(rb, status); +} + +U_CAPI const LanguageBreakEngine* U_EXPORT2 +CreateLSTMBreakEngine(UScriptCode script, const LSTMData* data, UErrorCode& status) +{ + UnicodeString unicodeSetString; + switch(script) { + case USCRIPT_THAI: + unicodeSetString = UnicodeString(u"[[:Thai:]&[:LineBreak=SA:]]"); + break; + case USCRIPT_MYANMAR: + unicodeSetString = UnicodeString(u"[[:Mymr:]&[:LineBreak=SA:]]"); + break; + default: + delete data; + return nullptr; + } + UnicodeSet unicodeSet; + unicodeSet.applyPattern(unicodeSetString, status); + const LanguageBreakEngine* engine = new LSTMBreakEngine(data, unicodeSet, status); + if (U_FAILURE(status) || engine == nullptr) { + if (engine != nullptr) { + delete engine; + } else { + status = U_MEMORY_ALLOCATION_ERROR; + } + return nullptr; + } + return engine; +} + +U_CAPI void U_EXPORT2 DeleteLSTMData(const LSTMData* data) +{ + delete data; +} + +U_CAPI const char16_t* U_EXPORT2 LSTMDataName(const LSTMData* data) +{ + return data->fName; +} + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ diff --git a/intl/icu/source/common/lstmbe.h b/intl/icu/source/common/lstmbe.h new file mode 100644 index 0000000000..77c97d85fa --- /dev/null +++ b/intl/icu/source/common/lstmbe.h @@ -0,0 +1,88 @@ +// © 2021 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +#ifndef LSTMBE_H +#define LSTMBE_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/uniset.h" +#include "unicode/ures.h" +#include "unicode/utext.h" +#include "unicode/utypes.h" + +#include "brkeng.h" +#include "dictbe.h" +#include "uvectr32.h" + +U_NAMESPACE_BEGIN + +class Vectorizer; +struct LSTMData; + +/******************************************************************* + * LSTMBreakEngine + */ + +/** + *

LSTMBreakEngine is a kind of DictionaryBreakEngine that uses a + * LSTM to determine language-specific breaks.

+ * + *

After it is constructed a LSTMBreakEngine may be shared between + * threads without synchronization.

+ */ +class LSTMBreakEngine : public DictionaryBreakEngine { +public: + /** + *

Constructor.

+ */ + LSTMBreakEngine(const LSTMData* data, const UnicodeSet& set, UErrorCode &status); + + /** + *

Virtual destructor.

+ */ + virtual ~LSTMBreakEngine(); + + virtual const char16_t* name() const; + +protected: + /** + *

Divide up a range of known dictionary characters handled by this break engine.

+ * + * @param text A UText representing the text + * @param rangeStart The start of the range of dictionary characters + * @param rangeEnd The end of the range of dictionary characters + * @param foundBreaks Output of C array of int32_t break positions, or 0 + * @param status Information on any errors encountered. + * @return The number of breaks found + */ + virtual int32_t divideUpDictionaryRange(UText *text, + int32_t rangeStart, + int32_t rangeEnd, + UVector32 &foundBreaks, + UBool isPhraseBreaking, + UErrorCode& status) const override; +private: + const LSTMData* fData; + const Vectorizer* fVectorizer; +}; + +U_CAPI const LanguageBreakEngine* U_EXPORT2 CreateLSTMBreakEngine( + UScriptCode script, const LSTMData* data, UErrorCode& status); + +U_CAPI const LSTMData* U_EXPORT2 CreateLSTMData( + UResourceBundle* rb, UErrorCode& status); + +U_CAPI const LSTMData* U_EXPORT2 CreateLSTMDataForScript( + UScriptCode script, UErrorCode& status); + +U_CAPI void U_EXPORT2 DeleteLSTMData(const LSTMData* data); +U_CAPI const char16_t* U_EXPORT2 LSTMDataName(const LSTMData* data); + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ + +#endif /* LSTMBE_H */ diff --git a/intl/icu/source/common/messageimpl.h b/intl/icu/source/common/messageimpl.h new file mode 100644 index 0000000000..061df9189d --- /dev/null +++ b/intl/icu/source/common/messageimpl.h @@ -0,0 +1,65 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2011, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: messageimpl.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2011apr04 +* created by: Markus W. Scherer +*/ + +#ifndef __MESSAGEIMPL_H__ +#define __MESSAGEIMPL_H__ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_FORMATTING + +#include "unicode/messagepattern.h" + +U_NAMESPACE_BEGIN + +/** + * Helper functions for use of MessagePattern. + * In Java, these are package-private methods in MessagePattern itself. + * In C++, they are declared here and implemented in messagepattern.cpp. + */ +class U_COMMON_API MessageImpl { +public: + /** + * @return true if getApostropheMode()==UMSGPAT_APOS_DOUBLE_REQUIRED + */ + static UBool jdkAposMode(const MessagePattern &msgPattern) { + return msgPattern.getApostropheMode()==UMSGPAT_APOS_DOUBLE_REQUIRED; + } + + /** + * Appends the s[start, limit[ substring to sb, but with only half of the apostrophes + * according to JDK pattern behavior. + */ + static void appendReducedApostrophes(const UnicodeString &s, int32_t start, int32_t limit, + UnicodeString &sb); + + /** + * Appends the sub-message to the result string. + * Omits SKIP_SYNTAX and appends whole arguments using appendReducedApostrophes(). + */ + static UnicodeString &appendSubMessageWithoutSkipSyntax(const MessagePattern &msgPattern, + int32_t msgStart, + UnicodeString &result); + +private: + MessageImpl() = delete; // no constructor: all static methods +}; + +U_NAMESPACE_END + +#endif // !UCONFIG_NO_FORMATTING + +#endif // __MESSAGEIMPL_H__ diff --git a/intl/icu/source/common/messagepattern.cpp b/intl/icu/source/common/messagepattern.cpp new file mode 100644 index 0000000000..82cb638e4e --- /dev/null +++ b/intl/icu/source/common/messagepattern.cpp @@ -0,0 +1,1233 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2011-2012, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: messagepattern.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2011mar14 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_FORMATTING + +#include "unicode/messagepattern.h" +#include "unicode/unistr.h" +#include "unicode/utf16.h" +#include "cmemory.h" +#include "cstring.h" +#include "messageimpl.h" +#include "patternprops.h" +#include "putilimp.h" +#include "uassert.h" + +U_NAMESPACE_BEGIN + +// Unicode character/code point constants ---------------------------------- *** + +static const char16_t u_pound=0x23; +static const char16_t u_apos=0x27; +static const char16_t u_plus=0x2B; +static const char16_t u_comma=0x2C; +static const char16_t u_minus=0x2D; +static const char16_t u_dot=0x2E; +static const char16_t u_colon=0x3A; +static const char16_t u_lessThan=0x3C; +static const char16_t u_equal=0x3D; +static const char16_t u_A=0x41; +static const char16_t u_C=0x43; +static const char16_t u_D=0x44; +static const char16_t u_E=0x45; +static const char16_t u_H=0x48; +static const char16_t u_I=0x49; +static const char16_t u_L=0x4C; +static const char16_t u_N=0x4E; +static const char16_t u_O=0x4F; +static const char16_t u_P=0x50; +static const char16_t u_R=0x52; +static const char16_t u_S=0x53; +static const char16_t u_T=0x54; +static const char16_t u_U=0x55; +static const char16_t u_Z=0x5A; +static const char16_t u_a=0x61; +static const char16_t u_c=0x63; +static const char16_t u_d=0x64; +static const char16_t u_e=0x65; +static const char16_t u_f=0x66; +static const char16_t u_h=0x68; +static const char16_t u_i=0x69; +static const char16_t u_l=0x6C; +static const char16_t u_n=0x6E; +static const char16_t u_o=0x6F; +static const char16_t u_p=0x70; +static const char16_t u_r=0x72; +static const char16_t u_s=0x73; +static const char16_t u_t=0x74; +static const char16_t u_u=0x75; +static const char16_t u_z=0x7A; +static const char16_t u_leftCurlyBrace=0x7B; +static const char16_t u_pipe=0x7C; +static const char16_t u_rightCurlyBrace=0x7D; +static const char16_t u_lessOrEqual=0x2264; // U+2264 is <= + +static const char16_t kOffsetColon[]={ // "offset:" + u_o, u_f, u_f, u_s, u_e, u_t, u_colon +}; + +static const char16_t kOther[]={ // "other" + u_o, u_t, u_h, u_e, u_r +}; + +// MessagePatternList ------------------------------------------------------ *** + +template +class MessagePatternList : public UMemory { +public: + MessagePatternList() {} + void copyFrom(const MessagePatternList &other, + int32_t length, + UErrorCode &errorCode); + UBool ensureCapacityForOneMore(int32_t oldLength, UErrorCode &errorCode); + UBool equals(const MessagePatternList &other, int32_t length) const { + for(int32_t i=0; i a; +}; + +template +void +MessagePatternList::copyFrom( + const MessagePatternList &other, + int32_t length, + UErrorCode &errorCode) { + if(U_SUCCESS(errorCode) && length>0) { + if(length>a.getCapacity() && nullptr==a.resize(length)) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + uprv_memcpy(a.getAlias(), other.a.getAlias(), (size_t)length*sizeof(T)); + } +} + +template +UBool +MessagePatternList::ensureCapacityForOneMore(int32_t oldLength, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return false; + } + if(a.getCapacity()>oldLength || a.resize(2*oldLength, oldLength)!=nullptr) { + return true; + } + errorCode=U_MEMORY_ALLOCATION_ERROR; + return false; +} + +// MessagePatternList specializations -------------------------------------- *** + +class MessagePatternDoubleList : public MessagePatternList { +}; + +class MessagePatternPartsList : public MessagePatternList { +}; + +// MessagePattern constructors etc. ---------------------------------------- *** + +MessagePattern::MessagePattern(UErrorCode &errorCode) + : aposMode(UCONFIG_MSGPAT_DEFAULT_APOSTROPHE_MODE), + partsList(nullptr), parts(nullptr), partsLength(0), + numericValuesList(nullptr), numericValues(nullptr), numericValuesLength(0), + hasArgNames(false), hasArgNumbers(false), needsAutoQuoting(false) { + init(errorCode); +} + +MessagePattern::MessagePattern(UMessagePatternApostropheMode mode, UErrorCode &errorCode) + : aposMode(mode), + partsList(nullptr), parts(nullptr), partsLength(0), + numericValuesList(nullptr), numericValues(nullptr), numericValuesLength(0), + hasArgNames(false), hasArgNumbers(false), needsAutoQuoting(false) { + init(errorCode); +} + +MessagePattern::MessagePattern(const UnicodeString &pattern, UParseError *parseError, UErrorCode &errorCode) + : aposMode(UCONFIG_MSGPAT_DEFAULT_APOSTROPHE_MODE), + partsList(nullptr), parts(nullptr), partsLength(0), + numericValuesList(nullptr), numericValues(nullptr), numericValuesLength(0), + hasArgNames(false), hasArgNumbers(false), needsAutoQuoting(false) { + if(init(errorCode)) { + parse(pattern, parseError, errorCode); + } +} + +UBool +MessagePattern::init(UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return false; + } + partsList=new MessagePatternPartsList(); + if(partsList==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return false; + } + parts=partsList->a.getAlias(); + return true; +} + +MessagePattern::MessagePattern(const MessagePattern &other) + : UObject(other), aposMode(other.aposMode), msg(other.msg), + partsList(nullptr), parts(nullptr), partsLength(0), + numericValuesList(nullptr), numericValues(nullptr), numericValuesLength(0), + hasArgNames(other.hasArgNames), hasArgNumbers(other.hasArgNumbers), + needsAutoQuoting(other.needsAutoQuoting) { + UErrorCode errorCode=U_ZERO_ERROR; + if(!copyStorage(other, errorCode)) { + clear(); + } +} + +MessagePattern & +MessagePattern::operator=(const MessagePattern &other) { + if(this==&other) { + return *this; + } + aposMode=other.aposMode; + msg=other.msg; + hasArgNames=other.hasArgNames; + hasArgNumbers=other.hasArgNumbers; + needsAutoQuoting=other.needsAutoQuoting; + UErrorCode errorCode=U_ZERO_ERROR; + if(!copyStorage(other, errorCode)) { + clear(); + } + return *this; +} + +UBool +MessagePattern::copyStorage(const MessagePattern &other, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return false; + } + parts=nullptr; + partsLength=0; + numericValues=nullptr; + numericValuesLength=0; + if(partsList==nullptr) { + partsList=new MessagePatternPartsList(); + if(partsList==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return false; + } + parts=partsList->a.getAlias(); + } + if(other.partsLength>0) { + partsList->copyFrom(*other.partsList, other.partsLength, errorCode); + if(U_FAILURE(errorCode)) { + return false; + } + parts=partsList->a.getAlias(); + partsLength=other.partsLength; + } + if(other.numericValuesLength>0) { + if(numericValuesList==nullptr) { + numericValuesList=new MessagePatternDoubleList(); + if(numericValuesList==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return false; + } + numericValues=numericValuesList->a.getAlias(); + } + numericValuesList->copyFrom( + *other.numericValuesList, other.numericValuesLength, errorCode); + if(U_FAILURE(errorCode)) { + return false; + } + numericValues=numericValuesList->a.getAlias(); + numericValuesLength=other.numericValuesLength; + } + return true; +} + +MessagePattern::~MessagePattern() { + delete partsList; + delete numericValuesList; +} + +// MessagePattern API ------------------------------------------------------ *** + +MessagePattern & +MessagePattern::parse(const UnicodeString &pattern, UParseError *parseError, UErrorCode &errorCode) { + preParse(pattern, parseError, errorCode); + parseMessage(0, 0, 0, UMSGPAT_ARG_TYPE_NONE, parseError, errorCode); + postParse(); + return *this; +} + +MessagePattern & +MessagePattern::parseChoiceStyle(const UnicodeString &pattern, + UParseError *parseError, UErrorCode &errorCode) { + preParse(pattern, parseError, errorCode); + parseChoiceStyle(0, 0, parseError, errorCode); + postParse(); + return *this; +} + +MessagePattern & +MessagePattern::parsePluralStyle(const UnicodeString &pattern, + UParseError *parseError, UErrorCode &errorCode) { + preParse(pattern, parseError, errorCode); + parsePluralOrSelectStyle(UMSGPAT_ARG_TYPE_PLURAL, 0, 0, parseError, errorCode); + postParse(); + return *this; +} + +MessagePattern & +MessagePattern::parseSelectStyle(const UnicodeString &pattern, + UParseError *parseError, UErrorCode &errorCode) { + preParse(pattern, parseError, errorCode); + parsePluralOrSelectStyle(UMSGPAT_ARG_TYPE_SELECT, 0, 0, parseError, errorCode); + postParse(); + return *this; +} + +void +MessagePattern::clear() { + // Mostly the same as preParse(). + msg.remove(); + hasArgNames=hasArgNumbers=false; + needsAutoQuoting=false; + partsLength=0; + numericValuesLength=0; +} + +bool +MessagePattern::operator==(const MessagePattern &other) const { + if(this==&other) { + return true; + } + return + aposMode==other.aposMode && + msg==other.msg && + // parts.equals(o.parts) + partsLength==other.partsLength && + (partsLength==0 || partsList->equals(*other.partsList, partsLength)); + // No need to compare numericValues if msg and parts are the same. +} + +int32_t +MessagePattern::hashCode() const { + int32_t hash=(aposMode*37+msg.hashCode())*37+partsLength; + for(int32_t i=0; i0;) { + const Part &part=getPart(--i); + if(part.getType()==UMSGPAT_PART_TYPE_INSERT_CHAR) { + modified.insert(part.index, (char16_t)part.value); + } + } + return modified; +} + +double +MessagePattern::getNumericValue(const Part &part) const { + UMessagePatternPartType type=part.type; + if(type==UMSGPAT_PART_TYPE_ARG_INT) { + return part.value; + } else if(type==UMSGPAT_PART_TYPE_ARG_DOUBLE) { + return numericValues[part.value]; + } else { + return UMSGPAT_NO_NUMERIC_VALUE; + } +} + +/** + * Returns the "offset:" value of a PluralFormat argument, or 0 if none is specified. + * @param pluralStart the index of the first PluralFormat argument style part. (0..countParts()-1) + * @return the "offset:" value. + * @draft ICU 4.8 + */ +double +MessagePattern::getPluralOffset(int32_t pluralStart) const { + const Part &part=getPart(pluralStart); + if(Part::hasNumericValue(part.type)) { + return getNumericValue(part); + } else { + return 0; + } +} + +// MessagePattern::Part ---------------------------------------------------- *** + +bool +MessagePattern::Part::operator==(const Part &other) const { + if(this==&other) { + return true; + } + return + type==other.type && + index==other.index && + length==other.length && + value==other.value && + limitPartIndex==other.limitPartIndex; +} + +// MessagePattern parser --------------------------------------------------- *** + +void +MessagePattern::preParse(const UnicodeString &pattern, UParseError *parseError, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return; + } + if(parseError!=nullptr) { + parseError->line=0; + parseError->offset=0; + parseError->preContext[0]=0; + parseError->postContext[0]=0; + } + msg=pattern; + hasArgNames=hasArgNumbers=false; + needsAutoQuoting=false; + partsLength=0; + numericValuesLength=0; +} + +void +MessagePattern::postParse() { + if(partsList!=nullptr) { + parts=partsList->a.getAlias(); + } + if(numericValuesList!=nullptr) { + numericValues=numericValuesList->a.getAlias(); + } +} + +int32_t +MessagePattern::parseMessage(int32_t index, int32_t msgStartLength, + int32_t nestingLevel, UMessagePatternArgType parentType, + UParseError *parseError, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return 0; + } + if(nestingLevel>Part::MAX_VALUE) { + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + int32_t msgStart=partsLength; + addPart(UMSGPAT_PART_TYPE_MSG_START, index, msgStartLength, nestingLevel, errorCode); + index+=msgStartLength; + for(;;) { // while(index=msg.length()) { + break; + } + char16_t c=msg.charAt(index++); + if(c==u_apos) { + if(index==msg.length()) { + // The apostrophe is the last character in the pattern. + // Add a Part for auto-quoting. + addPart(UMSGPAT_PART_TYPE_INSERT_CHAR, index, 0, + u_apos, errorCode); // value=char to be inserted + needsAutoQuoting=true; + } else { + c=msg.charAt(index); + if(c==u_apos) { + // double apostrophe, skip the second one + addPart(UMSGPAT_PART_TYPE_SKIP_SYNTAX, index++, 1, 0, errorCode); + } else if( + aposMode==UMSGPAT_APOS_DOUBLE_REQUIRED || + c==u_leftCurlyBrace || c==u_rightCurlyBrace || + (parentType==UMSGPAT_ARG_TYPE_CHOICE && c==u_pipe) || + (UMSGPAT_ARG_TYPE_HAS_PLURAL_STYLE(parentType) && c==u_pound) + ) { + // skip the quote-starting apostrophe + addPart(UMSGPAT_PART_TYPE_SKIP_SYNTAX, index-1, 1, 0, errorCode); + // find the end of the quoted literal text + for(;;) { + index=msg.indexOf(u_apos, index+1); + if(index>=0) { + if(/*(index+1)0 && c==u_rightCurlyBrace) || + (parentType==UMSGPAT_ARG_TYPE_CHOICE && c==u_pipe)) { + // Finish the message before the terminator. + // In a choice style, report the "}" substring only for the following ARG_LIMIT, + // not for this MSG_LIMIT. + int32_t limitLength=(parentType==UMSGPAT_ARG_TYPE_CHOICE && c==u_rightCurlyBrace) ? 0 : 1; + addLimitPart(msgStart, UMSGPAT_PART_TYPE_MSG_LIMIT, index-1, limitLength, + nestingLevel, errorCode); + if(parentType==UMSGPAT_ARG_TYPE_CHOICE) { + // Let the choice style parser see the '}' or '|'. + return index-1; + } else { + // continue parsing after the '}' + return index; + } + } // else: c is part of literal text + } + if(nestingLevel>0 && !inTopLevelChoiceMessage(nestingLevel, parentType)) { + setParseError(parseError, 0); // Unmatched '{' braces in message. + errorCode=U_UNMATCHED_BRACES; + return 0; + } + addLimitPart(msgStart, UMSGPAT_PART_TYPE_MSG_LIMIT, index, 0, nestingLevel, errorCode); + return index; +} + +int32_t +MessagePattern::parseArg(int32_t index, int32_t argStartLength, int32_t nestingLevel, + UParseError *parseError, UErrorCode &errorCode) { + int32_t argStart=partsLength; + UMessagePatternArgType argType=UMSGPAT_ARG_TYPE_NONE; + addPart(UMSGPAT_PART_TYPE_ARG_START, index, argStartLength, argType, errorCode); + if(U_FAILURE(errorCode)) { + return 0; + } + int32_t nameIndex=index=skipWhiteSpace(index+argStartLength); + if(index==msg.length()) { + setParseError(parseError, 0); // Unmatched '{' braces in message. + errorCode=U_UNMATCHED_BRACES; + return 0; + } + // parse argument name or number + index=skipIdentifier(index); + int32_t number=parseArgNumber(nameIndex, index); + if(number>=0) { + int32_t length=index-nameIndex; + if(length>Part::MAX_LENGTH || number>Part::MAX_VALUE) { + setParseError(parseError, nameIndex); // Argument number too large. + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + hasArgNumbers=true; + addPart(UMSGPAT_PART_TYPE_ARG_NUMBER, nameIndex, length, number, errorCode); + } else if(number==UMSGPAT_ARG_NAME_NOT_NUMBER) { + int32_t length=index-nameIndex; + if(length>Part::MAX_LENGTH) { + setParseError(parseError, nameIndex); // Argument name too long. + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + hasArgNames=true; + addPart(UMSGPAT_PART_TYPE_ARG_NAME, nameIndex, length, 0, errorCode); + } else { // number<-1 (ARG_NAME_NOT_VALID) + setParseError(parseError, nameIndex); // Bad argument syntax. + errorCode=U_PATTERN_SYNTAX_ERROR; + return 0; + } + index=skipWhiteSpace(index); + if(index==msg.length()) { + setParseError(parseError, 0); // Unmatched '{' braces in message. + errorCode=U_UNMATCHED_BRACES; + return 0; + } + char16_t c=msg.charAt(index); + if(c==u_rightCurlyBrace) { + // all done + } else if(c!=u_comma) { + setParseError(parseError, nameIndex); // Bad argument syntax. + errorCode=U_PATTERN_SYNTAX_ERROR; + return 0; + } else /* ',' */ { + // parse argument type: case-sensitive a-zA-Z + int32_t typeIndex=index=skipWhiteSpace(index+1); + while(indexPart::MAX_LENGTH) { + setParseError(parseError, nameIndex); // Argument type name too long. + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + argType=UMSGPAT_ARG_TYPE_SIMPLE; + if(length==6) { + // case-insensitive comparisons for complex-type names + if(isChoice(typeIndex)) { + argType=UMSGPAT_ARG_TYPE_CHOICE; + } else if(isPlural(typeIndex)) { + argType=UMSGPAT_ARG_TYPE_PLURAL; + } else if(isSelect(typeIndex)) { + argType=UMSGPAT_ARG_TYPE_SELECT; + } + } else if(length==13) { + if(isSelect(typeIndex) && isOrdinal(typeIndex+6)) { + argType=UMSGPAT_ARG_TYPE_SELECTORDINAL; + } + } + // change the ARG_START type from NONE to argType + partsList->a[argStart].value=(int16_t)argType; + if(argType==UMSGPAT_ARG_TYPE_SIMPLE) { + addPart(UMSGPAT_PART_TYPE_ARG_TYPE, typeIndex, length, 0, errorCode); + } + // look for an argument style (pattern) + if(c==u_rightCurlyBrace) { + if(argType!=UMSGPAT_ARG_TYPE_SIMPLE) { + setParseError(parseError, nameIndex); // No style field for complex argument. + errorCode=U_PATTERN_SYNTAX_ERROR; + return 0; + } + } else /* ',' */ { + ++index; + if(argType==UMSGPAT_ARG_TYPE_SIMPLE) { + index=parseSimpleStyle(index, parseError, errorCode); + } else if(argType==UMSGPAT_ARG_TYPE_CHOICE) { + index=parseChoiceStyle(index, nestingLevel, parseError, errorCode); + } else { + index=parsePluralOrSelectStyle(argType, index, nestingLevel, parseError, errorCode); + } + } + } + // Argument parsing stopped on the '}'. + addLimitPart(argStart, UMSGPAT_PART_TYPE_ARG_LIMIT, index, 1, argType, errorCode); + return index+1; +} + +int32_t +MessagePattern::parseSimpleStyle(int32_t index, UParseError *parseError, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return 0; + } + int32_t start=index; + int32_t nestedBraces=0; + while(index0) { + --nestedBraces; + } else { + int32_t length=--index-start; + if(length>Part::MAX_LENGTH) { + setParseError(parseError, start); // Argument style text too long. + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + addPart(UMSGPAT_PART_TYPE_ARG_STYLE, start, length, 0, errorCode); + return index; + } + } // c is part of literal text + } + setParseError(parseError, 0); // Unmatched '{' braces in message. + errorCode=U_UNMATCHED_BRACES; + return 0; +} + +int32_t +MessagePattern::parseChoiceStyle(int32_t index, int32_t nestingLevel, + UParseError *parseError, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return 0; + } + int32_t start=index; + index=skipWhiteSpace(index); + if(index==msg.length() || msg.charAt(index)==u_rightCurlyBrace) { + setParseError(parseError, 0); // Missing choice argument pattern. + errorCode=U_PATTERN_SYNTAX_ERROR; + return 0; + } + for(;;) { + // The choice argument style contains |-separated (number, separator, message) triples. + // Parse the number. + int32_t numberIndex=index; + index=skipDouble(index); + int32_t length=index-numberIndex; + if(length==0) { + setParseError(parseError, start); // Bad choice pattern syntax. + errorCode=U_PATTERN_SYNTAX_ERROR; + return 0; + } + if(length>Part::MAX_LENGTH) { + setParseError(parseError, numberIndex); // Choice number too long. + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + parseDouble(numberIndex, index, true, parseError, errorCode); // adds ARG_INT or ARG_DOUBLE + if(U_FAILURE(errorCode)) { + return 0; + } + // Parse the separator. + index=skipWhiteSpace(index); + if(index==msg.length()) { + setParseError(parseError, start); // Bad choice pattern syntax. + errorCode=U_PATTERN_SYNTAX_ERROR; + return 0; + } + char16_t c=msg.charAt(index); + if(!(c==u_pound || c==u_lessThan || c==u_lessOrEqual)) { // U+2264 is <= + setParseError(parseError, start); // Expected choice separator (#<\u2264) instead of c. + errorCode=U_PATTERN_SYNTAX_ERROR; + return 0; + } + addPart(UMSGPAT_PART_TYPE_ARG_SELECTOR, index, 1, 0, errorCode); + // Parse the message fragment. + index=parseMessage(++index, 0, nestingLevel+1, UMSGPAT_ARG_TYPE_CHOICE, parseError, errorCode); + if(U_FAILURE(errorCode)) { + return 0; + } + // parseMessage(..., CHOICE) returns the index of the terminator, or msg.length(). + if(index==msg.length()) { + return index; + } + if(msg.charAt(index)==u_rightCurlyBrace) { + if(!inMessageFormatPattern(nestingLevel)) { + setParseError(parseError, start); // Bad choice pattern syntax. + errorCode=U_PATTERN_SYNTAX_ERROR; + return 0; + } + return index; + } // else the terminator is '|' + index=skipWhiteSpace(index+1); + } +} + +int32_t +MessagePattern::parsePluralOrSelectStyle(UMessagePatternArgType argType, + int32_t index, int32_t nestingLevel, + UParseError *parseError, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return 0; + } + int32_t start=index; + UBool isEmpty=true; + UBool hasOther=false; + for(;;) { + // First, collect the selector looking for a small set of terminators. + // It would be a little faster to consider the syntax of each possible + // token right here, but that makes the code too complicated. + index=skipWhiteSpace(index); + UBool eos=index==msg.length(); + if(eos || msg.charAt(index)==u_rightCurlyBrace) { + if(eos==inMessageFormatPattern(nestingLevel)) { + setParseError(parseError, start); // Bad plural/select pattern syntax. + errorCode=U_PATTERN_SYNTAX_ERROR; + return 0; + } + if(!hasOther) { + setParseError(parseError, 0); // Missing 'other' keyword in plural/select pattern. + errorCode=U_DEFAULT_KEYWORD_MISSING; + return 0; + } + return index; + } + int32_t selectorIndex=index; + if(UMSGPAT_ARG_TYPE_HAS_PLURAL_STYLE(argType) && msg.charAt(selectorIndex)==u_equal) { + // explicit-value plural selector: =double + index=skipDouble(index+1); + int32_t length=index-selectorIndex; + if(length==1) { + setParseError(parseError, start); // Bad plural/select pattern syntax. + errorCode=U_PATTERN_SYNTAX_ERROR; + return 0; + } + if(length>Part::MAX_LENGTH) { + setParseError(parseError, selectorIndex); // Argument selector too long. + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + addPart(UMSGPAT_PART_TYPE_ARG_SELECTOR, selectorIndex, length, 0, errorCode); + parseDouble(selectorIndex+1, index, false, + parseError, errorCode); // adds ARG_INT or ARG_DOUBLE + } else { + index=skipIdentifier(index); + int32_t length=index-selectorIndex; + if(length==0) { + setParseError(parseError, start); // Bad plural/select pattern syntax. + errorCode=U_PATTERN_SYNTAX_ERROR; + return 0; + } + // Note: The ':' in "offset:" is just beyond the skipIdentifier() range. + if( UMSGPAT_ARG_TYPE_HAS_PLURAL_STYLE(argType) && length==6 && indexPart::MAX_LENGTH) { + setParseError(parseError, valueIndex); // Plural offset value too long. + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + parseDouble(valueIndex, index, false, + parseError, errorCode); // adds ARG_INT or ARG_DOUBLE + if(U_FAILURE(errorCode)) { + return 0; + } + isEmpty=false; + continue; // no message fragment after the offset + } else { + // normal selector word + if(length>Part::MAX_LENGTH) { + setParseError(parseError, selectorIndex); // Argument selector too long. + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + addPart(UMSGPAT_PART_TYPE_ARG_SELECTOR, selectorIndex, length, 0, errorCode); + if(0==msg.compare(selectorIndex, length, kOther, 0, 5)) { + hasOther=true; + } + } + } + if(U_FAILURE(errorCode)) { + return 0; + } + + // parse the message fragment following the selector + index=skipWhiteSpace(index); + if(index==msg.length() || msg.charAt(index)!=u_leftCurlyBrace) { + setParseError(parseError, selectorIndex); // No message fragment after plural/select selector. + errorCode=U_PATTERN_SYNTAX_ERROR; + return 0; + } + index=parseMessage(index, 1, nestingLevel+1, argType, parseError, errorCode); + if(U_FAILURE(errorCode)) { + return 0; + } + isEmpty=false; + } +} + +int32_t +MessagePattern::parseArgNumber(const UnicodeString &s, int32_t start, int32_t limit) { + // If the identifier contains only ASCII digits, then it is an argument _number_ + // and must not have leading zeros (except "0" itself). + // Otherwise it is an argument _name_. + if(start>=limit) { + return UMSGPAT_ARG_NAME_NOT_VALID; + } + int32_t number; + // Defer numeric errors until we know there are only digits. + UBool badNumber; + char16_t c=s.charAt(start++); + if(c==0x30) { + if(start==limit) { + return 0; + } else { + number=0; + badNumber=true; // leading zero + } + } else if(0x31<=c && c<=0x39) { + number=c-0x30; + badNumber=false; + } else { + return UMSGPAT_ARG_NAME_NOT_NUMBER; + } + while(start=INT32_MAX/10) { + badNumber=true; // overflow + } + number=number*10+(c-0x30); + } else { + return UMSGPAT_ARG_NAME_NOT_NUMBER; + } + } + // There are only ASCII digits. + if(badNumber) { + return UMSGPAT_ARG_NAME_NOT_VALID; + } else { + return number; + } +} + +void +MessagePattern::parseDouble(int32_t start, int32_t limit, UBool allowInfinity, + UParseError *parseError, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return; + } + U_ASSERT(start(Part::MAX_VALUE+isNegative)) { + break; // not a small-enough integer + } + if(index==limit) { + addPart(UMSGPAT_PART_TYPE_ARG_INT, start, limit-start, + isNegative!=0 ? -value : value, errorCode); + return; + } + c=msg.charAt(index++); + } + // Let Double.parseDouble() throw a NumberFormatException. + char numberChars[128]; + int32_t capacity=(int32_t)sizeof(numberChars); + int32_t length=limit-start; + if(length>=capacity) { + break; // number too long + } + msg.extract(start, length, numberChars, capacity, US_INV); + if((int32_t)uprv_strlen(numberChars)0x39 && c!=u_e && c!=u_E && c!=0x221e)) { + break; + } + ++index; + } + return index; +} + +UBool +MessagePattern::isArgTypeChar(UChar32 c) { + return (u_a<=c && c<=u_z) || (u_A<=c && c<=u_Z); +} + +UBool +MessagePattern::isChoice(int32_t index) { + char16_t c; + return + ((c=msg.charAt(index++))==u_c || c==u_C) && + ((c=msg.charAt(index++))==u_h || c==u_H) && + ((c=msg.charAt(index++))==u_o || c==u_O) && + ((c=msg.charAt(index++))==u_i || c==u_I) && + ((c=msg.charAt(index++))==u_c || c==u_C) && + ((c=msg.charAt(index))==u_e || c==u_E); +} + +UBool +MessagePattern::isPlural(int32_t index) { + char16_t c; + return + ((c=msg.charAt(index++))==u_p || c==u_P) && + ((c=msg.charAt(index++))==u_l || c==u_L) && + ((c=msg.charAt(index++))==u_u || c==u_U) && + ((c=msg.charAt(index++))==u_r || c==u_R) && + ((c=msg.charAt(index++))==u_a || c==u_A) && + ((c=msg.charAt(index))==u_l || c==u_L); +} + +UBool +MessagePattern::isSelect(int32_t index) { + char16_t c; + return + ((c=msg.charAt(index++))==u_s || c==u_S) && + ((c=msg.charAt(index++))==u_e || c==u_E) && + ((c=msg.charAt(index++))==u_l || c==u_L) && + ((c=msg.charAt(index++))==u_e || c==u_E) && + ((c=msg.charAt(index++))==u_c || c==u_C) && + ((c=msg.charAt(index))==u_t || c==u_T); +} + +UBool +MessagePattern::isOrdinal(int32_t index) { + char16_t c; + return + ((c=msg.charAt(index++))==u_o || c==u_O) && + ((c=msg.charAt(index++))==u_r || c==u_R) && + ((c=msg.charAt(index++))==u_d || c==u_D) && + ((c=msg.charAt(index++))==u_i || c==u_I) && + ((c=msg.charAt(index++))==u_n || c==u_N) && + ((c=msg.charAt(index++))==u_a || c==u_A) && + ((c=msg.charAt(index))==u_l || c==u_L); +} + +UBool +MessagePattern::inMessageFormatPattern(int32_t nestingLevel) { + return nestingLevel>0 || partsList->a[0].type==UMSGPAT_PART_TYPE_MSG_START; +} + +UBool +MessagePattern::inTopLevelChoiceMessage(int32_t nestingLevel, UMessagePatternArgType parentType) { + return + nestingLevel==1 && + parentType==UMSGPAT_ARG_TYPE_CHOICE && + partsList->a[0].type!=UMSGPAT_PART_TYPE_MSG_START; +} + +void +MessagePattern::addPart(UMessagePatternPartType type, int32_t index, int32_t length, + int32_t value, UErrorCode &errorCode) { + if(partsList->ensureCapacityForOneMore(partsLength, errorCode)) { + Part &part=partsList->a[partsLength++]; + part.type=type; + part.index=index; + part.length=(uint16_t)length; + part.value=(int16_t)value; + part.limitPartIndex=0; + } +} + +void +MessagePattern::addLimitPart(int32_t start, + UMessagePatternPartType type, int32_t index, int32_t length, + int32_t value, UErrorCode &errorCode) { + partsList->a[start].limitPartIndex=partsLength; + addPart(type, index, length, value, errorCode); +} + +void +MessagePattern::addArgDoublePart(double numericValue, int32_t start, int32_t length, + UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return; + } + int32_t numericIndex=numericValuesLength; + if(numericValuesList==nullptr) { + numericValuesList=new MessagePatternDoubleList(); + if(numericValuesList==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + } else if(!numericValuesList->ensureCapacityForOneMore(numericValuesLength, errorCode)) { + return; + } else { + if(numericIndex>Part::MAX_VALUE) { + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return; + } + } + numericValuesList->a[numericValuesLength++]=numericValue; + addPart(UMSGPAT_PART_TYPE_ARG_DOUBLE, start, length, numericIndex, errorCode); +} + +void +MessagePattern::setParseError(UParseError *parseError, int32_t index) { + if(parseError==nullptr) { + return; + } + parseError->offset=index; + + // Set preContext to some of msg before index. + // Avoid splitting a surrogate pair. + int32_t length=index; + if(length>=U_PARSE_CONTEXT_LEN) { + length=U_PARSE_CONTEXT_LEN-1; + if(length>0 && U16_IS_TRAIL(msg[index-length])) { + --length; + } + } + msg.extract(index-length, length, parseError->preContext); + parseError->preContext[length]=0; + + // Set postContext to some of msg starting at index. + length=msg.length()-index; + if(length>=U_PARSE_CONTEXT_LEN) { + length=U_PARSE_CONTEXT_LEN-1; + if(length>0 && U16_IS_LEAD(msg[index+length-1])) { + --length; + } + } + msg.extract(index, length, parseError->postContext); + parseError->postContext[length]=0; +} + +// MessageImpl ------------------------------------------------------------- *** + +void +MessageImpl::appendReducedApostrophes(const UnicodeString &s, int32_t start, int32_t limit, + UnicodeString &sb) { + int32_t doubleApos=-1; + for(;;) { + int32_t i=s.indexOf(u_apos, start); + if(i<0 || i>=limit) { + sb.append(s, start, limit-start); + break; + } + if(i==doubleApos) { + // Double apostrophe at start-1 and start==i, append one. + sb.append(u_apos); + ++start; + doubleApos=-1; + } else { + // Append text between apostrophes and skip this one. + sb.append(s, start, i-start); + doubleApos=start=i+1; + } + } +} + +// Ported from second half of ICU4J SelectFormat.format(String). +UnicodeString & +MessageImpl::appendSubMessageWithoutSkipSyntax(const MessagePattern &msgPattern, + int32_t msgStart, + UnicodeString &result) { + const UnicodeString &msgString=msgPattern.getPatternString(); + int32_t prevIndex=msgPattern.getPart(msgStart).getLimit(); + for(int32_t i=msgStart;;) { + const MessagePattern::Part &part=msgPattern.getPart(++i); + UMessagePatternPartType type=part.getType(); + int32_t index=part.getIndex(); + if(type==UMSGPAT_PART_TYPE_MSG_LIMIT) { + return result.append(msgString, prevIndex, index-prevIndex); + } else if(type==UMSGPAT_PART_TYPE_SKIP_SYNTAX) { + result.append(msgString, prevIndex, index-prevIndex); + prevIndex=part.getLimit(); + } else if(type==UMSGPAT_PART_TYPE_ARG_START) { + result.append(msgString, prevIndex, index-prevIndex); + prevIndex=index; + i=msgPattern.getLimitPartIndex(i); + index=msgPattern.getPart(i).getLimit(); + appendReducedApostrophes(msgString, prevIndex, index, result); + prevIndex=index; + } + } +} + +U_NAMESPACE_END + +#endif // !UCONFIG_NO_FORMATTING diff --git a/intl/icu/source/common/mlbe.cpp b/intl/icu/source/common/mlbe.cpp new file mode 100644 index 0000000000..7e734f2c8a --- /dev/null +++ b/intl/icu/source/common/mlbe.cpp @@ -0,0 +1,270 @@ +// © 2022 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "cmemory.h" +#include "mlbe.h" +#include "uassert.h" +#include "ubrkimpl.h" +#include "unicode/resbund.h" +#include "unicode/udata.h" +#include "unicode/utf16.h" +#include "uresimp.h" +#include "util.h" +#include "uvectr32.h" + +U_NAMESPACE_BEGIN + +enum class ModelIndex { kUWStart = 0, kBWStart = 6, kTWStart = 9 }; + +MlBreakEngine::MlBreakEngine(const UnicodeSet &digitOrOpenPunctuationOrAlphabetSet, + const UnicodeSet &closePunctuationSet, UErrorCode &status) + : fDigitOrOpenPunctuationOrAlphabetSet(digitOrOpenPunctuationOrAlphabetSet), + fClosePunctuationSet(closePunctuationSet), + fNegativeSum(0) { + if (U_FAILURE(status)) { + return; + } + loadMLModel(status); +} + +MlBreakEngine::~MlBreakEngine() {} + +int32_t MlBreakEngine::divideUpRange(UText *inText, int32_t rangeStart, int32_t rangeEnd, + UVector32 &foundBreaks, const UnicodeString &inString, + const LocalPointer &inputMap, + UErrorCode &status) const { + if (U_FAILURE(status)) { + return 0; + } + if (rangeStart >= rangeEnd) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + UVector32 boundary(inString.countChar32() + 1, status); + if (U_FAILURE(status)) { + return 0; + } + int32_t numBreaks = 0; + int32_t codePointLength = inString.countChar32(); + // The ML algorithm groups six char and evaluates whether the 4th char is a breakpoint. + // In each iteration, it evaluates the 4th char and then moves forward one char like a sliding + // window. Initially, the first six values in the indexList are [-1, -1, 0, 1, 2, 3]. After + // moving forward, finally the last six values in the indexList are + // [length-4, length-3, length-2, length-1, -1, -1]. The "+4" here means four extra "-1". + int32_t indexSize = codePointLength + 4; + int32_t *indexList = (int32_t *)uprv_malloc(indexSize * sizeof(int32_t)); + if (indexList == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return 0; + } + int32_t numCodeUnits = initIndexList(inString, indexList, status); + + // Add a break for the start. + boundary.addElement(0, status); + numBreaks++; + if (U_FAILURE(status)) return 0; + + for (int32_t idx = 0; idx + 1 < codePointLength && U_SUCCESS(status); idx++) { + numBreaks = + evaluateBreakpoint(inString, indexList, idx, numCodeUnits, numBreaks, boundary, status); + if (idx + 4 < codePointLength) { + indexList[idx + 6] = numCodeUnits; + numCodeUnits += U16_LENGTH(inString.char32At(indexList[idx + 6])); + } + } + uprv_free(indexList); + + if (U_FAILURE(status)) return 0; + + // Add a break for the end if there is not one there already. + if (boundary.lastElementi() != inString.countChar32()) { + boundary.addElement(inString.countChar32(), status); + numBreaks++; + } + + int32_t prevCPPos = -1; + int32_t prevUTextPos = -1; + int32_t correctedNumBreaks = 0; + for (int32_t i = 0; i < numBreaks; i++) { + int32_t cpPos = boundary.elementAti(i); + int32_t utextPos = inputMap.isValid() ? inputMap->elementAti(cpPos) : cpPos + rangeStart; + U_ASSERT(cpPos > prevCPPos); + U_ASSERT(utextPos >= prevUTextPos); + + if (utextPos > prevUTextPos) { + if (utextPos != rangeStart || + (utextPos > 0 && + fClosePunctuationSet.contains(utext_char32At(inText, utextPos - 1)))) { + foundBreaks.push(utextPos, status); + correctedNumBreaks++; + } + } else { + // Normalization expanded the input text, the dictionary found a boundary + // within the expansion, giving two boundaries with the same index in the + // original text. Ignore the second. See ticket #12918. + --numBreaks; + } + prevCPPos = cpPos; + prevUTextPos = utextPos; + } + (void)prevCPPos; // suppress compiler warnings about unused variable + + UChar32 nextChar = utext_char32At(inText, rangeEnd); + if (!foundBreaks.isEmpty() && foundBreaks.peeki() == rangeEnd) { + // In phrase breaking, there has to be a breakpoint between Cj character and + // the number/open punctuation. + // E.g. る文字「そうだ、京都」->る▁文字▁「そうだ、▁京都」-> breakpoint between 字 and「 + // E.g. 乗車率９０％程度だろうか -> 乗車▁率▁９０％▁程度だろうか -> breakpoint between 率 and ９ + // E.g. しかもロゴがＵｎｉｃｏｄｅ！ -> しかも▁ロゴが▁Ｕｎｉｃｏｄｅ！-> breakpoint between が and Ｕ + if (!fDigitOrOpenPunctuationOrAlphabetSet.contains(nextChar)) { + foundBreaks.popi(); + correctedNumBreaks--; + } + } + + return correctedNumBreaks; +} + +int32_t MlBreakEngine::evaluateBreakpoint(const UnicodeString &inString, int32_t *indexList, + int32_t startIdx, int32_t numCodeUnits, int32_t numBreaks, + UVector32 &boundary, UErrorCode &status) const { + if (U_FAILURE(status)) { + return numBreaks; + } + int32_t start = 0, end = 0; + int32_t score = fNegativeSum; + + for (int i = 0; i < 6; i++) { + // UW1 ~ UW6 + start = startIdx + i; + if (indexList[start] != -1) { + end = (indexList[start + 1] != -1) ? indexList[start + 1] : numCodeUnits; + score += fModel[static_cast(ModelIndex::kUWStart) + i].geti( + inString.tempSubString(indexList[start], end - indexList[start])); + } + } + for (int i = 0; i < 3; i++) { + // BW1 ~ BW3 + start = startIdx + i + 1; + if (indexList[start] != -1 && indexList[start + 1] != -1) { + end = (indexList[start + 2] != -1) ? indexList[start + 2] : numCodeUnits; + score += fModel[static_cast(ModelIndex::kBWStart) + i].geti( + inString.tempSubString(indexList[start], end - indexList[start])); + } + } + for (int i = 0; i < 4; i++) { + // TW1 ~ TW4 + start = startIdx + i; + if (indexList[start] != -1 && indexList[start + 1] != -1 && indexList[start + 2] != -1) { + end = (indexList[start + 3] != -1) ? indexList[start + 3] : numCodeUnits; + score += fModel[static_cast(ModelIndex::kTWStart) + i].geti( + inString.tempSubString(indexList[start], end - indexList[start])); + } + } + + if (score > 0) { + boundary.addElement(startIdx + 1, status); + numBreaks++; + } + return numBreaks; +} + +int32_t MlBreakEngine::initIndexList(const UnicodeString &inString, int32_t *indexList, + UErrorCode &status) const { + if (U_FAILURE(status)) { + return 0; + } + int32_t index = 0; + int32_t length = inString.countChar32(); + // Set all (lenght+4) items inside indexLength to -1 presuming -1 is 4 bytes of 0xff. + uprv_memset(indexList, 0xff, (length + 4) * sizeof(int32_t)); + if (length > 0) { + indexList[2] = 0; + index = U16_LENGTH(inString.char32At(0)); + if (length > 1) { + indexList[3] = index; + index += U16_LENGTH(inString.char32At(index)); + if (length > 2) { + indexList[4] = index; + index += U16_LENGTH(inString.char32At(index)); + if (length > 3) { + indexList[5] = index; + index += U16_LENGTH(inString.char32At(index)); + } + } + } + } + return index; +} + +void MlBreakEngine::loadMLModel(UErrorCode &error) { + // BudouX's model consists of thirteen categories, each of which is make up of pairs of the + // feature and its score. As integrating it into jaml.txt, we define thirteen kinds of key and + // value to represent the feature and the corresponding score respectively. + + if (U_FAILURE(error)) return; + + UnicodeString key; + StackUResourceBundle stackTempBundle; + ResourceDataValue modelKey; + + LocalUResourceBundlePointer rbp(ures_openDirect(U_ICUDATA_BRKITR, "jaml", &error)); + UResourceBundle *rb = rbp.getAlias(); + if (U_FAILURE(error)) return; + + int32_t index = 0; + initKeyValue(rb, "UW1Keys", "UW1Values", fModel[index++], error); + initKeyValue(rb, "UW2Keys", "UW2Values", fModel[index++], error); + initKeyValue(rb, "UW3Keys", "UW3Values", fModel[index++], error); + initKeyValue(rb, "UW4Keys", "UW4Values", fModel[index++], error); + initKeyValue(rb, "UW5Keys", "UW5Values", fModel[index++], error); + initKeyValue(rb, "UW6Keys", "UW6Values", fModel[index++], error); + initKeyValue(rb, "BW1Keys", "BW1Values", fModel[index++], error); + initKeyValue(rb, "BW2Keys", "BW2Values", fModel[index++], error); + initKeyValue(rb, "BW3Keys", "BW3Values", fModel[index++], error); + initKeyValue(rb, "TW1Keys", "TW1Values", fModel[index++], error); + initKeyValue(rb, "TW2Keys", "TW2Values", fModel[index++], error); + initKeyValue(rb, "TW3Keys", "TW3Values", fModel[index++], error); + initKeyValue(rb, "TW4Keys", "TW4Values", fModel[index++], error); + fNegativeSum /= 2; +} + +void MlBreakEngine::initKeyValue(UResourceBundle *rb, const char *keyName, const char *valueName, + Hashtable &model, UErrorCode &error) { + int32_t keySize = 0; + int32_t valueSize = 0; + int32_t stringLength = 0; + UnicodeString key; + StackUResourceBundle stackTempBundle; + ResourceDataValue modelKey; + + // get modelValues + LocalUResourceBundlePointer modelValue(ures_getByKey(rb, valueName, nullptr, &error)); + const int32_t *value = ures_getIntVector(modelValue.getAlias(), &valueSize, &error); + if (U_FAILURE(error)) return; + + // get modelKeys + ures_getValueWithFallback(rb, keyName, stackTempBundle.getAlias(), modelKey, error); + ResourceArray stringArray = modelKey.getArray(error); + keySize = stringArray.getSize(); + if (U_FAILURE(error)) return; + + for (int32_t idx = 0; idx < keySize; idx++) { + stringArray.getValue(idx, modelKey); + key = UnicodeString(modelKey.getString(stringLength, error)); + if (U_SUCCESS(error)) { + U_ASSERT(idx < valueSize); + fNegativeSum -= value[idx]; + model.puti(key, value[idx], error); + } + } +} + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ diff --git a/intl/icu/source/common/mlbe.h b/intl/icu/source/common/mlbe.h new file mode 100644 index 0000000000..38de47e5f5 --- /dev/null +++ b/intl/icu/source/common/mlbe.h @@ -0,0 +1,116 @@ +// © 2022 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +#ifndef MLBREAKENGINE_H +#define MLBREAKENGINE_H + +#include "hash.h" +#include "unicode/resbund.h" +#include "unicode/uniset.h" +#include "unicode/utext.h" +#include "uvectr32.h" + +U_NAMESPACE_BEGIN + +#if !UCONFIG_NO_BREAK_ITERATION + +/** + * A machine learning break engine for the phrase breaking in Japanese. + */ +class MlBreakEngine : public UMemory { + public: + /** + * Constructor. + * + * @param digitOrOpenPunctuationOrAlphabetSet An UnicodeSet with the digit, open punctuation and + * alphabet. + * @param closePunctuationSet An UnicodeSet with close punctuation. + * @param status Information on any errors encountered. + */ + MlBreakEngine(const UnicodeSet &digitOrOpenPunctuationOrAlphabetSet, + const UnicodeSet &closePunctuationSet, UErrorCode &status); + + /** + * Virtual destructor. + */ + virtual ~MlBreakEngine(); + + public: + /** + * Divide up a range of characters handled by this break engine. + * + * @param inText A UText representing the text + * @param rangeStart The start of the range of the characters + * @param rangeEnd The end of the range of the characters + * @param foundBreaks Output of C array of int32_t break positions, or 0 + * @param inString The normalized string of text ranging from rangeStart to rangeEnd + * @param inputMap The vector storing the native index of inText + * @param status Information on any errors encountered. + * @return The number of breaks found + */ + int32_t divideUpRange(UText *inText, int32_t rangeStart, int32_t rangeEnd, + UVector32 &foundBreaks, const UnicodeString &inString, + const LocalPointer &inputMap, UErrorCode &status) const; + + private: + /** + * Load the machine learning's model file. + * + * @param error Information on any errors encountered. + */ + void loadMLModel(UErrorCode &error); + + /** + * In the machine learning's model file, specify the name of the key and value to load the + * corresponding feature and its score. + * + * @param rb A ResouceBundle corresponding to the model file. + * @param keyName The kay name in the model file. + * @param valueName The value name in the model file. + * @param model A hashtable to store the pairs of the feature and its score. + * @param error Information on any errors encountered. + */ + void initKeyValue(UResourceBundle *rb, const char *keyName, const char *valueName, + Hashtable &model, UErrorCode &error); + + /** + * Initialize the index list from the input string. + * + * @param inString A input string to be segmented. + * @param indexList A code unit index list of inString. + * @param status Information on any errors encountered. + * @return The number of code units of the first four characters in inString. + */ + int32_t initIndexList(const UnicodeString &inString, int32_t *indexList, + UErrorCode &status) const; + + /** + * Evaluate whether the index is a potential breakpoint. + * + * @param inString A input string to be segmented. + * @param indexList A code unit index list of the inString. + * @param startIdx The start index of the indexList. + * @param numCodeUnits The current code unit boundary of the indexList. + * @param numBreaks The accumulated number of breakpoints. + * @param boundary A vector including the index of the breakpoint. + * @param status Information on any errors encountered. + * @return The number of breakpoints + */ + int32_t evaluateBreakpoint(const UnicodeString &inString, int32_t *indexList, int32_t startIdx, + int32_t numCodeUnits, int32_t numBreaks, UVector32 &boundary, + UErrorCode &status) const; + + void printUnicodeString(const UnicodeString &s) const; + + UnicodeSet fDigitOrOpenPunctuationOrAlphabetSet; + UnicodeSet fClosePunctuationSet; + Hashtable fModel[13]; // {UW1, UW2, ... UW6, BW1, ... BW3, TW1, TW2, ... TW4} 6+3+4= 13 + int32_t fNegativeSum; +}; + +#endif + +U_NAMESPACE_END + +/* MLBREAKENGINE_H */ +#endif diff --git a/intl/icu/source/common/msvcres.h b/intl/icu/source/common/msvcres.h new file mode 100644 index 0000000000..d71b5ac922 --- /dev/null +++ b/intl/icu/source/common/msvcres.h @@ -0,0 +1,25 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +//{{NO_DEPENDENCIES}} +// Copyright (c) 2003-2010 International Business Machines +// Corporation and others. All Rights Reserved. +// +// Used by common.rc and other .rc files. +//Do not edit with Microsoft Developer Studio because it will modify this +//header the wrong way. This is here to prevent Visual Studio .NET from +//unnessarily building the resource files when it's not needed. +// + +/* +These are defined before unicode/uversion.h in order to prevent +STLPort's broken stddef.h from being used when rc.exe parses this file. +*/ +#define _STLP_OUTERMOST_HEADER_ID 0 +#define _STLP_WINCE 1 + +#include "unicode/uversion.h" + +#define ICU_WEBSITE "https://icu.unicode.org/" +#define ICU_COMPANY "The ICU Project" +#define ICU_PRODUCT_PREFIX "ICU" +#define ICU_PRODUCT "International Components for Unicode" diff --git a/intl/icu/source/common/mutex.h b/intl/icu/source/common/mutex.h new file mode 100644 index 0000000000..44b1f90ba0 --- /dev/null +++ b/intl/icu/source/common/mutex.h @@ -0,0 +1,77 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1997-2013, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +*/ +//---------------------------------------------------------------------------- +// File: mutex.h +// +// Lightweight C++ wrapper for umtx_ C mutex functions +// +// Author: Alan Liu 1/31/97 +// History: +// 06/04/97 helena Updated setImplementation as per feedback from 5/21 drop. +// 04/07/1999 srl refocused as a thin wrapper +// +//---------------------------------------------------------------------------- +#ifndef MUTEX_H +#define MUTEX_H + +#include "unicode/utypes.h" +#include "unicode/uobject.h" +#include "umutex.h" + +U_NAMESPACE_BEGIN + +/** + * Mutex is a helper class for convenient locking and unlocking of a UMutex. + * + * Creating a local scope Mutex will lock a UMutex, holding the lock until the Mutex + * goes out of scope. + * + * If no UMutex is specified, the ICU global mutex is implied. + * + * For example: + * + * static UMutex myMutex; + * + * void Function(int arg1, int arg2) + * { + * static Object* foo; // Shared read-write object + * Mutex mutex(&myMutex); // or no args for the global lock + * foo->Method(); + * // When 'mutex' goes out of scope and gets destroyed here, the lock is released + * } + * + * Note: Do NOT use the form 'Mutex mutex();' as that merely forward-declares a function + * returning a Mutex. This is a common mistake which silently slips through the + * compiler!! + */ + +class U_COMMON_API Mutex : public UMemory { +public: + Mutex(UMutex *mutex = nullptr) : fMutex(mutex) { + umtx_lock(fMutex); + } + ~Mutex() { + umtx_unlock(fMutex); + } + + Mutex(const Mutex &other) = delete; // forbid assigning of this class + Mutex &operator=(const Mutex &other) = delete; // forbid copying of this class + void *operator new(size_t s) = delete; // forbid heap allocation. Locals only. + +private: + UMutex *fMutex; +}; + + +U_NAMESPACE_END + +#endif //_MUTEX_ +//eof diff --git a/intl/icu/source/common/norm2_nfc_data.h b/intl/icu/source/common/norm2_nfc_data.h new file mode 100644 index 0000000000..ebe3e6ba90 --- /dev/null +++ b/intl/icu/source/common/norm2_nfc_data.h @@ -0,0 +1,1154 @@ +// Copyright (C) 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// Copyright (C) 1999-2016, International Business Machines +// Corporation and others. All Rights Reserved. +// +// file name: norm2_nfc_data.h +// +// machine-generated by: icu/source/tools/gennorm2/n2builder.cpp + +#ifdef INCLUDED_FROM_NORMALIZER2_CPP + +static const UVersionInfo norm2_nfc_data_formatVersion={4,0,0,0}; +static const UVersionInfo norm2_nfc_data_dataVersion={0xf,0,0,0}; + +static const int32_t norm2_nfc_data_indexes[Normalizer2Impl::IX_COUNT]={ +0x50,0x4cb8,0x8920,0x8a20,0x8a20,0x8a20,0x8a20,0x8a20,0xc0,0x300,0xae2,0x29e0,0x3c66,0xfc00,0x1288,0x3b9c, +0x3c34,0x3c66,0x300,0 +}; + +static const uint16_t norm2_nfc_data_trieIndex[1788]={ +0,0x40,0x7b,0xbb,0xfb,0x13a,0x17a,0x1b2,0x1f2,0x226,0x254,0x226,0x294,0x2d4,0x313,0x353, +0x393,0x3d2,0x40f,0x44e,0x226,0x226,0x488,0x4c8,0x4f8,0x530,0x226,0x570,0x59f,0x5de,0x226,0x5f3, +0x631,0x65f,0x687,0x6bd,0x6fd,0x73a,0x75a,0x799,0x7d8,0x815,0x834,0x871,0x75a,0x8aa,0x8d8,0x917, +0x834,0x951,0x968,0x9a8,0x9bf,0x9fe,0x226,0xa34,0xa54,0xa8f,0xa9b,0xad6,0xafe,0xb3b,0xb7b,0xbb5, +0xbd0,0x226,0xc0b,0x226,0xc4b,0xc6a,0xca0,0xcdd,0x226,0x226,0x226,0x226,0x226,0xd00,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0xd2c,0x226,0x226,0xd61, +0x226,0x226,0xd7f,0x226,0xda9,0x226,0x226,0x226,0xde5,0xe05,0xe45,0xe84,0xebf,0xeff,0xf33,0xf5f, +0x839,0x226,0x226,0xf93,0x226,0x226,0x226,0xfd3,0x1013,0x1053,0x1093,0x10d3,0x1113,0x1153,0x1193,0x11d3, +0x1213,0x226,0x226,0x1243,0x1274,0x226,0x12a4,0x12d7,0x1314,0x1353,0x1393,0x13c9,0x13f7,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x1422,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0xcee,0x226,0x143f,0x226,0x147f,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x14bf,0x14f9,0x1537,0x1577,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x15b6,0x15f4,0x1614,0x226,0x226,0x226,0x226, +0x164e,0x226,0x226,0x1676,0x16a8,0x16d6,0x83d,0x16e9,0x226,0x226,0x16f9,0x1739,0x226,0x226,0x226,0x1451, +0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781, +0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791, +0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785, +0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779, +0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789, +0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d, +0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d, +0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781, +0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791, +0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785, +0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x178d,0x1779,0x1781,0x1789,0x1791,0x177d,0x1785,0x17c5,0x226, +0x1805,0x1840,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x1880,0x18c0,0x1900,0x1940,0x1980,0x19c0,0x1a00,0x1a40,0x1a63,0x1aa3,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x1ac3,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x67e,0x68e,0x6a6,0x6c5,0x6da,0x6da,0x6da,0x6de,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0xc0b,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x54f,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x40c, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x1af6,0x226,0x226,0x1b06,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0xdf7,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x1b16,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x1607,0x226,0x226,0x226,0x226,0x66b,0x226,0x226,0x226, +0x226,0x1b20,0x54f,0x226,0x226,0x1b30,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x81c,0x226,0x226, +0x1b40,0x226,0x1b50,0x1b5d,0x1b69,0x226,0x226,0x226,0x226,0x414,0x226,0x1b74,0x1b84,0x226,0x226,0x226, +0x811,0x226,0x226,0x226,0x226,0x1b94,0x226,0x226,0x226,0x1b9f,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x1ba6,0x226,0x226,0x226,0x226,0x1bb1,0x1bc0,0x927,0x1bce,0x412,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x1bdc,0x7c9,0x226,0x226,0x226,0x226,0x226,0x1bec,0x1bfb,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x907,0x1c03,0x1c13,0x226, +0x226,0x226,0x9eb,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x1c1d,0x226,0x226,0x226,0x226,0x226, +0x226,0x817,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x1c1a, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x1c2d, +0x811,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x87e,0x226,0x226,0x226,0x81e,0x81b, +0x226,0x226,0x226,0x226,0x819,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x9eb,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0xc05,0x226,0x226,0x226, +0x226,0x81b,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0xc08,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x1c3d,0x226,0x226,0x226,0xf2c,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x1c4d,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x1c4f,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x1c5e,0x1c6e,0x1c7c,0x1c89,0x226, +0x1c95,0x1ca3,0x1cb3,0x226,0x226,0x226,0x226,0xd1b,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x1cc3,0x1ccb,0x1cd9,0x226,0x226,0x226,0x226,0x226,0x4f9,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0xf2c,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x7c9,0x226,0x226,0x226,0x4fc,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x1ce4,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x1cf4,0x226,0x226,0x226,0x226, +0x226,0x226,0x1d00,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x1d10,0x1d20, +0x1d30,0x1d40,0x1d50,0x1d60,0x1d70,0x1d80,0x1d90,0x1da0,0x1db0,0x1dc0,0x1dd0,0x1de0,0x1df0,0x1e00,0x1e10,0x1e20, +0x1e30,0x1e40,0x1e50,0x1e60,0x1e70,0x1e80,0x1e90,0x1ea0,0x1eb0,0x1ec0,0x1ed0,0x1ee0,0x1ef0,0x1f00,0x1f10,0x1f20, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226, +0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x226,0x408,0x428, +0xc4,0xc4,0xc4,0x448,0x457,0x46d,0x489,0x4a6,0x4c2,0x4df,0x4fc,0x51b,0x538,0x552,0xc4,0xc4, +0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4, +0xc4,0xc4,0xc4,0x567,0xc4,0x57b,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4, +0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4, +0xc4,0xc4,0xc4,0xc4,0x59b,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0x5a6,0x5c3,0xc4, +0xc4,0xc4,0xc4,0xc4,0xc4,0x5e3,0x5f9,0x60b,0xc4,0x61e,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4, +0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4, +0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0xc4,0x63e,0x65e +}; + +static const uint16_t norm2_nfc_data_trieData[7984]={ +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,4,8,0xc,1, +1,0x10,0x50,0x5c,0x70,0x88,0xcc,0xd0,0xec,0x108,0x144,0x148,0x15c,0x174,0x180,0x1a4, +0x1e4,1,0x1ec,0x20c,0x228,0x244,0x290,0x298,0x2b0,0x2b8,0x2dc,1,1,1,1,1, +1,0x2f4,0x334,0x340,0x354,0x36c,0x3b0,0x3b4,0x3d0,0x3f0,0x428,0x430,0x444,0x45c,0x468,0x48c, +0x4cc,1,0x4d4,0x4f4,0x510,0x530,0x57c,0x584,0x5a0,0x5a8,0x5d0,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0x5e8,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,0x128a,0x1290,0xae4,0x1296,0xafa, +0xb04,0x5f4,0xb0e,0x129c,0x12a2,0xb18,0x12a8,0x12ae,0x12b4,0x12ba,0xb2e,1,0x12c0,0x12c6,0x12cc,0xb38, +0xb4e,0xb60,1,0x5fc,0x12d2,0x12d8,0x12de,0xb6a,0x12e4,1,1,0x12ea,0x12f0,0xb80,0x12f6,0xb96, +0xba0,0x600,0xbaa,0x12fc,0x1302,0xbb4,0x1308,0x130e,0x1314,0x131a,0xbca,1,0x1320,0x1326,0x132c,0xbd4, +0xbea,0xbfc,1,0x608,0x1332,0x1338,0x133e,0xc06,0x1344,1,0x134a,0x1350,0x1356,0xc1c,0xc32,0x135d, +0x1363,0x1368,0x136e,0x1374,0x137a,0x1380,0x1386,0x138c,0x1392,0x1398,0x139e,1,1,0xc48,0xc56,0x13a4, +0x13aa,0x13b0,0x13b6,0x13bd,0x13c3,0x13c8,0x13ce,0x13d4,0x13da,0x13e0,0x13e6,0x13ec,0x13f2,0x13f9,0x13ff,0x1404, +0x140a,1,1,0x1410,0x1416,0x141c,0x1422,0x1428,0x142e,0x1435,0x143b,0x1440,1,1,1,0x1447, +0x144d,0x1453,0x1459,1,0x145e,0x1464,0x146b,0x1471,0x1476,0x147c,1,1,1,0x1482,0x1488,0x148f, +0x1495,0x149a,0x14a0,1,1,1,0xc64,0xc72,0x14a6,0x14ac,0x14b2,0x14b8,1,1,0x14be,0x14c4, +0x14cb,0x14d1,0x14d6,0x14dc,0xc80,0xc8a,0x14e2,0x14e8,0x14ef,0x14f5,0xc94,0xc9e,0x14fb,0x1501,0x1506,0x150c, +1,1,0xca8,0xcb2,0xcbc,0xcc6,0x1512,0x1518,0x151e,0x1524,0x152a,0x1530,0x1537,0x153d,0x1542,0x1548, +0x154e,0x1554,0x155a,0x1560,0x1566,0x156c,0x1572,0x1578,0x157e,0x60c,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,0xcd0,0xcea,1,1,1,1, +1,1,1,1,1,1,1,1,1,0xd04,0xd1e,1,1,1,1,1, +1,0x610,1,1,1,1,1,1,1,1,1,1,1,1,1,0x1584, +0x158a,0x1590,0x1596,0x159c,0x15a2,0x15a8,0x15ae,0x15b6,0x15c0,0x15ca,0x15d4,0x15de,0x15e8,0x15f2,0x15fc,1, +0x1606,0x1610,0x161a,0x1624,0x162d,0x1633,1,1,0x1638,0x163e,0x1644,0x164a,0xd38,0xd42,0x1653,0x165d, +0x1665,0x166b,0x1671,1,1,1,0x1676,0x167c,1,1,0x1682,0x1688,0x1690,0x169a,0x16a3,0x16a9, +0x16af,0x16b5,0x16ba,0x16c0,0x16c6,0x16cc,0x16d2,0x16d8,0x16de,0x16e4,0x16ea,0x16f0,0x16f6,0x16fc,0x1702,0x1708, +0x170e,0x1714,0x171a,0x1720,0x1726,0x172c,0x1732,0x1738,0x173e,0x1744,0x174a,0x1750,0x1756,0x175c,1,1, +0x1762,0x1768,1,1,1,1,1,1,0xd4c,0xd56,0xd60,0xd6a,0x1770,0x177a,0x1784,0x178e, +0xd74,0xd7e,0x1798,0x17a2,0x17aa,0x17b0,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,0x614,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,0xfdcc,0xfdcc,0xfdcc,0xfdcc,0xfdcc,0xffcc,0xfdcc,0xfdcc,0xfdcc,0xfdcc,0xfdcc,0xfdcc, +0xfdcc,0xffcc,0xffcc,0xfdcc,0xffcc,0xfdcc,0xffcc,0xfdcc,0xfdcc,0xffd0,0xffb8,0xffb8,0xffb8,0xffb8,0xffd0,0xfdb0, +0xffb8,0xffb8,0xffb8,0xffb8,0xffb8,0xff94,0xff94,0xfdb8,0xfdb8,0xfdb8,0xfdb8,0xfd94,0xfd94,0xffb8,0xffb8,0xffb8, +0xffb8,0xfdb8,0xfdb8,0xffb8,0xfdb8,0xfdb8,0xffb8,0xffb8,0xfe02,0xfe02,0xfe02,0xfe02,0xfc02,0xffb8,0xffb8,0xffb8, +0xffb8,0xffcc,0xffcc,0xffcc,0x3c36,0x3c3c,0xfdcc,0x3c42,0x3c48,0xfde0,0xffcc,0xffb8,0xffb8,0xffb8,0xffcc,0xffcc, +0xffcc,0xffb8,0xffb8,1,0xffcc,0xffcc,0xffcc,0xffb8,0xffb8,0xffb8,0xffb8,0xffcc,0xffd0,0xffb8,0xffb8,0xffcc, +0xffd2,0xffd4,0xffd4,0xffd2,0xffd4,0xffd4,0xffd2,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc, +0xffcc,0xffcc,0xffcc,0xffcc,1,1,1,1,0x29e1,1,1,1,1,1,1,1, +1,1,0x29e5,1,1,1,1,1,0x17b7,0x17bd,0x29e9,0x17c3,0x17c9,0x17cf,1,0x17d5, +1,0x17db,0x17e1,0x17e9,0x618,1,1,1,0x634,1,0x644,1,0x658,1,1,1, +1,1,0x674,1,0x684,1,1,1,0x688,1,1,1,0x6a0,0x17f1,0x17f7,0xd88, +0x17fd,0xd92,0x1803,0x180b,0x6b4,1,1,1,0x6d4,1,0x6e4,1,0x6fc,1,1,1, +1,1,0x71c,1,0x72c,1,1,1,0x734,1,1,1,0x754,0xd9c,0xdae,0x1813, +0x1819,0xdc0,1,1,1,0x76c,0x181f,0x1825,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0x182b,0x1831,1,0x1837,1,1,0x774,0x183d,1,1,1,1,0x1843, +0x1849,0x184f,1,0x778,1,1,0x780,1,0x784,0x790,0x798,0x79c,0x1855,0x7ac,1,1, +1,0x7b0,1,1,1,1,0x7b4,1,1,1,0x7c4,1,1,1,0x7c8,1, +0x7cc,1,1,0x7d0,1,1,0x7d8,1,0x7dc,0x7e8,0x7f0,0x7f4,0x185b,0x804,1,1, +1,0x808,1,1,1,0x80c,1,1,1,0x81c,1,1,1,0x820,1,0x824, +1,1,0x1861,0x1867,1,0x186d,1,1,0x828,0x1873,1,1,1,1,0x1879,0x187f, +0x1885,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,0x82c,0x830,0x188b,0x1891,1,1,1,1,1,1, +1,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0x1897, +0x189d,1,1,1,1,1,1,1,1,1,1,1,1,1,0x18a3,0x18a9, +0x18af,0x18b5,1,1,0x18bb,0x18c1,0x834,0x838,0x18c7,0x18cd,0x18d3,0x18d9,0x18df,0x18e5,1,1, +0x18eb,0x18f1,0x18f7,0x18fd,0x1903,0x1909,0x83c,0x840,0x190f,0x1915,0x191b,0x1921,0x1927,0x192d,0x1933,0x1939, +0x193f,0x1945,0x194b,0x1951,1,1,0x1957,0x195d,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,0xffb8,0xffcc,0xffcc,0xffcc,0xffcc,0xffb8,0xffcc, +0xffcc,0xffcc,0xffbc,0xffb8,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffb8,0xffb8,0xffb8,0xffb8,0xffb8,0xffb8, +0xffcc,0xffcc,0xffb8,0xffcc,0xffcc,0xffbc,0xffc8,0xffcc,0xfe14,0xfe16,0xfe18,0xfe1a,0xfe1c,0xfe1e,0xfe20,0xfe22, +0xfe24,0xfe26,0xfe26,0xfe28,0xfe2a,0xfe2c,1,0xfe2e,1,0xfe30,0xfe32,1,0xffcc,0xffb8,1,0xfe24, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc, +0xfe3c,0xfe3e,0xfe40,1,1,1,1,1,1,1,0x1962,0x1968,0x196f,0x1975,0x197b,0x844, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,0x850,1,0x854,0xfe36,0xfe38,0xfe3a,0xfe3c,0xfe3e, +0xfe40,0xfe42,0xfe44,0xfdcc,0xfdcc,0xfdb8,0xffb8,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffb8,0xffcc,0xffcc,0xffb8, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0xfe46,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0x1981,0x858,0x1987,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,0x85c,0x198d,1,0x860,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,1,1,0xffcc, +0xffcc,0xffcc,0xffcc,0xffb8,0xffcc,1,1,0xffcc,0xffcc,1,0xffb8,0xffcc,0xffcc,0xffb8,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0xfe48,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0xffcc, +0xffb8,0xffcc,0xffcc,0xffb8,0xffcc,0xffcc,0xffb8,0xffb8,0xffb8,0xffcc,0xffb8,0xffb8,0xffcc,0xffb8,0xffcc,0xffcc, +0xffb8,0xffcc,0xffb8,0xffcc,0xffb8,0xffcc,0xffb8,0xffcc,0xffcc,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,0xffcc,0xffcc, +0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffb8,0xffcc,1,1,1,1,1,1,1,1,1, +0xffb8,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,0xffcc,0xffcc,0xffcc,0xffcc,1,0xffcc,0xffcc,0xffcc,0xffcc, +0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,1,0xffcc,0xffcc,0xffcc,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,0xffb8,0xffb8,0xffb8,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0xffcc, +0xffb8,0xffb8,0xffb8,0xffcc,0xffcc,0xffcc,0xffcc,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffb8,0xffb8,0xffb8,0xffb8, +0xffb8,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,1, +0xffb8,0xffcc,0xffcc,0xffb8,0xffcc,0xffcc,0xffb8,0xffcc,0xffcc,0xffcc,0xffb8,0xffb8,0xffb8,0xfe36,0xfe38,0xfe3a, +0xffcc,0xffcc,0xffcc,0xffb8,0xffcc,0xffcc,0xffb8,0xffb8,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,0x864,0x1993,1,1,1,1,1,1,0x868,0x1999,1, +0x86c,0x199f,1,1,1,1,1,1,1,0xfc0e,1,1,1,1,1,1, +1,1,1,1,1,1,1,0xfe12,1,1,1,0xffcc,0xffb8,0xffcc,0xffcc,1, +1,1,0x29ec,0x29f2,0x29f8,0x29fe,0x2a04,0x2a0a,0x2a10,0x2a16,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,0xfe0e,1,0xfc00,1,1,1,1,1,1,1, +0x870,1,1,1,0x19a5,0x19ab,0xfe12,1,1,1,1,1,1,1,1,1, +0xfc00,1,1,1,1,0x2a1c,0x2a22,1,0x2a28,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,0xffcc,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,0x2a2e,1,1,0x2a34,1, +1,1,1,1,0xfe0e,1,1,1,1,1,1,1,1,1,1,1, +1,1,0xfe12,1,1,1,1,1,1,1,1,1,1,1,0x2a3a,0x2a40, +0x2a46,1,1,0x2a4c,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0xfe0e,1,1,1,1,1,1,1,1,1,1,1,1,1,0xfe12,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,0x878,0x19b1,1,1,0x19b7,0x19bd,0xfe12,1,1,1,1,1,1,1,1, +0xfc00,0xfc00,1,1,1,1,0x2a52,0x2a58,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,0x884,1,0x19c3,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,0xfc00,1,1,1,1,1,1,0x888,0x890,1, +1,0x19c9,0x19cf,0x19d5,0xfe12,1,1,1,1,1,1,1,1,1,0xfc00,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,0x894,1,0x19db,1,1,1,1,0xfe12,1, +1,1,1,1,1,1,0xfea8,0xfcb6,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,0xfe0e,1,1,0x898,0x19e1,1,0xfc00,1,1,1,0x89c,0x19e7, +0x19ed,1,0xdca,0x19f5,1,0xfe12,1,1,1,1,1,1,1,0xfc00,0xfc00,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,0xfe12,0xfe12,1,0xfc00,1,1, +1,1,1,1,0x8a8,0x8b0,1,1,0x19fd,0x1a03,0x1a09,0xfe12,1,1,1,1, +1,1,1,1,1,0xfc00,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,0xfc12,1, +1,1,1,0xfc00,1,1,1,1,1,1,1,1,1,0x8b4,0x1a0f,1, +0xdd4,0x1a17,0x1a1f,0xfc00,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,0xfece,0xfece,0xfe12,1, +1,1,1,1,1,1,1,0xfed6,0xfed6,0xfed6,0xfed6,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0xfeec,0xfeec,0xfe12,1,1,1,1,1,1,1,1,0xfef4,0xfef4, +0xfef4,0xfef4,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,0xffb8,0xffb8,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0xffb8,1,0xffb8,1,0xffb0,1,1,1,1,1,1,0x2a5f,1, +1,1,1,1,1,1,1,1,0x2a65,1,1,1,1,0x2a6b,1,1, +1,1,0x2a71,1,1,1,1,0x2a77,1,1,1,1,1,1,1,1, +1,1,1,1,0x2a7d,1,1,1,1,1,1,1,0xff02,0xff04,0x3c50,0xff08, +0x3c58,0x2a82,1,0x2a88,1,0xff04,0xff04,0xff04,0xff04,1,1,0xff04,0x3c60,0xffcc,0xffcc,0xfe12, +1,0xffcc,0xffcc,1,1,1,1,1,1,1,1,1,1,1,0x2a8f,1, +1,1,1,1,1,1,1,1,0x2a95,1,1,1,1,0x2a9b,1,1, +1,1,0x2aa1,1,1,1,1,0x2aa7,1,1,1,1,1,1,1,1, +1,1,1,1,0x2aad,1,1,1,1,1,1,0xffb8,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,0x8c0,0x1a25,1,1,1,1,1,1,1,0xfc00,1, +1,1,1,1,1,1,1,0xfe0e,1,0xfe12,0xfe12,1,1,1,1,1, +1,1,1,1,1,1,1,1,0xffb8,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00, +0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00, +0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00,0xfe00, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,0xffcc,0xffcc,0xffcc, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0xfe12,0xfe12,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0xfe12,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0xfe12,1,1,1,1,1,1,1,1,1,1,0xffcc,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,0xffc8,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,0xffbc,0xffcc,0xffb8,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,0xffcc,0xffb8,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,0xfe12,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc, +0xffcc,0xffcc,1,1,0xffb8,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffb8,0xffb8,0xffb8,0xffb8,0xffb8,0xffb8, +0xffcc,0xffcc,0xffb8,1,0xffb8,0xffcc,0xffcc,0xffb8,0xffb8,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffb8,0xffcc, +0xffcc,0xffcc,0xffcc,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,0x8c4,0x1a2b,0x8c8,0x1a31,0x8cc,0x1a37,0x8d0,0x1a3d,0x8d4,0x1a43,1,1, +0x8d8,0x1a49,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0xfe0e,0xfc00,1,1,1,1,0x8dc,0x1a4f,0x8e0,0x1a55,0x8e4,0x8e8,0x1a5b, +0x1a61,0x8ec,0x1a67,0xfe12,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,0xffcc,0xffb8,0xffcc,0xffcc,0xffcc,0xffcc, +0xffcc,0xffcc,0xffcc,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,0xfe12,0xfe12,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,0xfe0e,1,1,1,1,1,1,1,1,1,1, +1,0xfe12,0xfe12,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0xffcc,0xffcc,0xffcc,1,0xfe02,0xffb8,0xffb8,0xffb8,0xffb8,0xffb8,0xffcc,0xffcc,0xffb8, +0xffb8,0xffb8,0xffb8,0xffcc,1,0xfe02,0xfe02,0xfe02,0xfe02,0xfe02,0xfe02,0xfe02,1,1,1,1, +0xffb8,1,1,1,1,1,1,0xffcc,1,1,1,0xffcc,0xffcc,1,1,1, +1,1,1,0xffcc,0xffcc,0xffb8,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffb8,0xffcc,0xffcc, +0xffd4,0xffac,0xffb8,0xff94,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc, +0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc, +0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffd0,0xffc8,0xffc8,0xffb8,0xffb4,0xffcc,0xffd2, +0xffb8,0xffcc,0xffb8,0x1a6c,0x1a72,0x1a78,0x1a7e,0x1a85,0x1a8b,0x1a91,0x1a97,0x1a9f,0x1aa9,0x1ab0,0x1ab6,0x1abc, +0x1ac2,0x1ac8,0x1ace,0x1ad5,0x1adb,0x1ae0,0x1ae6,0x1aee,0x1af8,0x1b02,0x1b0c,0x1b14,0x1b1a,0x1b20,0x1b26,0x1b2f, +0x1b39,0x1b41,0x1b47,0x1b4c,0x1b52,0x1b58,0x1b5e,0x1b64,0x1b6a,0x1b70,0x1b76,0x1b7d,0x1b83,0x1b88,0x1b8e,0x1b94, +0x1b9a,0x1ba2,0x1bac,0x1bb4,0x1bba,0x1bc0,0x1bc6,0x1bcc,0x1bd2,0xdde,0xde8,0x1bda,0x1be4,0x1bec,0x1bf2,0x1bf8, +0x1bfe,0x1c04,0x1c0a,0x1c10,0x1c16,0x1c1d,0x1c23,0x1c28,0x1c2e,0x1c34,0x1c3a,0x1c40,0x1c46,0x1c4c,0x1c52,0x1c5a, +0x1c64,0x1c6e,0x1c78,0x1c82,0x1c8c,0x1c96,0x1ca0,0x1ca9,0x1caf,0x1cb5,0x1cbb,0x1cc0,0x1cc6,0xdf2,0xdfc,0x1cce, +0x1cd8,0x1ce0,0x1ce6,0x1cec,0x1cf2,0xe06,0xe10,0x1cfa,0x1d04,0x1d0e,0x1d18,0x1d22,0x1d2c,0x1d34,0x1d3a,0x1d40, +0x1d46,0x1d4c,0x1d52,0x1d58,0x1d5e,0x1d64,0x1d6a,0x1d70,0x1d76,0x1d7c,0x1d82,0x1d8a,0x1d94,0x1d9e,0x1da8,0x1db0, +0x1db6,0x1dbd,0x1dc3,0x1dc8,0x1dce,0x1dd4,0x1dda,0x1de0,0x1de6,0x1dec,0x1df2,0x1df9,0x1dff,0x1e05,0x1e0b,0x1e11, +0x1e17,0x1e1c,0x1e22,0x1e28,0x1e2e,0x1e35,0x1e3b,0x1e41,0x1e47,0x1e4c,0x1e52,0x1e58,0x1e5e,1,0x1e65,1, +1,1,1,0xe1a,0xe28,0x1e6a,0x1e70,0x1e78,0x1e82,0x1e8c,0x1e96,0x1ea0,0x1eaa,0x1eb4,0x1ebe,0x1ec8, +0x1ed2,0x1edc,0x1ee6,0x1ef0,0x1efa,0x1f04,0x1f0e,0x1f18,0x1f22,0x1f2c,0x1f36,0xe36,0xe40,0x1f3e,0x1f44,0x1f4a, +0x1f50,0x1f58,0x1f62,0x1f6c,0x1f76,0x1f80,0x1f8a,0x1f94,0x1f9e,0x1fa8,0x1fb2,0x1fba,0x1fc0,0x1fc6,0x1fcc,0xe4a, +0xe54,0x1fd2,0x1fd8,0x1fe0,0x1fea,0x1ff4,0x1ffe,0x2008,0x2012,0x201c,0x2026,0x2030,0x203a,0x2044,0x204e,0x2058, +0x2062,0x206c,0x2076,0x2080,0x208a,0x2094,0x209e,0x20a6,0x20ac,0x20b2,0x20b8,0x20c0,0x20ca,0x20d4,0x20de,0x20e8, +0x20f2,0x20fc,0x2106,0x2110,0x211a,0x2122,0x2128,0x212f,0x2135,0x213a,0x2140,0x2146,0x214c,1,1,1, +1,1,1,0xe5e,0xe74,0xe8c,0xe9a,0xea8,0xeb6,0xec4,0xed2,0xede,0xef4,0xf0c,0xf1a,0xf28, +0xf36,0xf44,0xf52,0xf5e,0xf6c,0x2155,0x215f,0x2169,0x2173,1,1,0xf7a,0xf88,0x217d,0x2187,0x2191, +0x219b,1,1,0xf96,0xfac,0xfc4,0xfd2,0xfe0,0xfee,0xffc,0x100a,0x1016,0x102c,0x1044,0x1052,0x1060, +0x106e,0x107c,0x108a,0x1096,0x10a8,0x21a5,0x21af,0x21b9,0x21c3,0x21cd,0x21d7,0x10ba,0x10cc,0x21e1,0x21eb,0x21f5, +0x21ff,0x2209,0x2213,0x10de,0x10ec,0x221d,0x2227,0x2231,0x223b,1,1,0x10fa,0x1108,0x2245,0x224f,0x2259, +0x2263,1,1,0x1116,0x1128,0x226d,0x2277,0x2281,0x228b,0x2295,0x229f,1,0x113a,1,0x22a9,1, +0x22b3,1,0x22bd,0x114c,0x1162,0x117a,0x1188,0x1196,0x11a4,0x11b2,0x11c0,0x11cc,0x11e2,0x11fa,0x1208,0x1216, +0x1224,0x1232,0x1240,0x124c,0x3b9e,0x22c5,0x3ba6,0x1256,0x3bae,0x22cb,0x3bb6,0x22d1,0x3bbe,0x22d7,0x3bc6,0x1260, +0x3bce,1,1,0x22de,0x22e8,0x22f7,0x2307,0x2317,0x2327,0x2337,0x2347,0x2352,0x235c,0x236b,0x237b,0x238b, +0x239b,0x23ab,0x23bb,0x23c6,0x23d0,0x23df,0x23ef,0x23ff,0x240f,0x241f,0x242f,0x243a,0x2444,0x2453,0x2463,0x2473, +0x2483,0x2493,0x24a3,0x24ae,0x24b8,0x24c7,0x24d7,0x24e7,0x24f7,0x2507,0x2517,0x2522,0x252c,0x253b,0x254b,0x255b, +0x256b,0x257b,0x258b,0x2595,0x259b,0x25a3,0x25aa,0x25b3,1,0x126a,0x25bd,0x25c5,0x25cb,0x25d1,0x3bd6,0x25d6, +1,0x2ab2,0x8f0,1,0x25dd,0x25e5,0x25ec,0x25f5,1,0x1274,0x25ff,0x2607,0x3bde,0x260d,0x3be6,0x2612, +0x2619,0x261f,0x2625,0x262b,0x2631,0x2639,0x3bf0,1,1,0x2641,0x2649,0x2651,0x2657,0x265d,0x3bfa,1, +0x2663,0x2669,0x266f,0x2675,0x267b,0x2683,0x3c04,0x268b,0x2691,0x2697,0x269f,0x26a7,0x26ad,0x26b3,0x3c0e,0x26b9, +0x26bf,0x3c16,0x2ab7,1,1,0x26c7,0x26ce,0x26d7,1,0x127e,0x26e1,0x26e9,0x3c1e,0x26ef,0x3c26,0x26f4, +0x2abb,0x8fc,1,0xfa09,0xfa09,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0xffcc,0xffcc,0xfe02,0xfe02,0xffcc,0xffcc,0xffcc,0xffcc,0xfe02,0xfe02,0xfe02,0xffcc,0xffcc, +1,1,1,1,0xffcc,1,1,1,0xfe02,0xfe02,0xffcc,0xffb8,0xffcc,0xfe02,0xfe02,0xffb8, +0xffb8,0xffb8,0xffb8,0xffcc,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,0x2abe,1,1,1,0x2ac2,0x3c2e, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,0x908,1,0x90c,1,0x910,1,1,1,1,1,0x26fb,0x2701, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,0x2707,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,0x270d,0x2713,0x2719,0x914,1,0x918,1,0x91c,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,0x920,0x271f,1,1,1,0x924,0x2725,1,0x928, +0x272b,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,0x92c,0x2731,0x930,0x2737,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0x934,1,1,1,0x273d,1,0x938,0x2743,0x93c,1,0x2749,0x940,0x274f,1,1,1, +0x944,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0x2755,0x948,0x275b,1,0x94c,0x950,1,1,1,1,1,1,1, +0x2761,0x2767,0x276d,0x2773,0x2779,0x954,0x958,0x277f,0x2785,0x95c,0x960,0x278b,0x2791,0x964,0x968,0x96c, +0x970,1,1,0x2797,0x279d,0x974,0x978,0x27a3,0x27a9,0x97c,0x980,0x27af,0x27b5,1,1,1, +1,1,1,1,0x984,0x988,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,0x98c,1,1,1,1,1,0x990,0x994,1,0x998,0x27bb, +0x27c1,0x27c7,0x27cd,1,1,0x99c,0x9a0,0x9a4,0x9a8,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,0x27d3,0x27d9,0x27df,0x27e5,1,1,1, +1,1,1,0x27eb,0x27f1,0x27f7,0x27fd,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0x2ac7,0x2acb,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,0x2acf,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,0xfe12,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0xffcc, +0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc, +0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,0xffb4,0xffc8,0xffd0,0xffbc,0xffc0,0xffc0,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0x9ac, +1,1,1,1,0x9b0,0x2803,0x9b4,0x2809,0x9b8,0x280f,0x9bc,0x2815,0x9c0,0x281b,0x9c4,0x2821, +0x9c8,0x2827,0x9cc,0x282d,0x9d0,0x2833,0x9d4,0x2839,0x9d8,0x283f,0x9dc,0x2845,1,0x9e0,0x284b,0x9e4, +0x2851,0x9e8,0x2857,1,1,1,1,1,0x9ec,0x285d,0x2863,0x9f4,0x2869,0x286f,0x9fc,0x2875, +0x287b,0xa04,0x2881,0x2887,0xa0c,0x288d,0x2893,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,0x2899,1,1,1,1, +0xfc10,0xfc10,1,1,0xa14,0x289f,1,1,1,1,1,1,1,0xa18,1,1, +1,1,0xa1c,0x28a5,0xa20,0x28ab,0xa24,0x28b1,0xa28,0x28b7,0xa2c,0x28bd,0xa30,0x28c3,0xa34,0x28c9, +0xa38,0x28cf,0xa3c,0x28d5,0xa40,0x28db,0xa44,0x28e1,0xa48,0x28e7,1,0xa4c,0x28ed,0xa50,0x28f3,0xa54, +0x28f9,1,1,1,1,1,0xa58,0x28ff,0x2905,0xa60,0x290b,0x2911,0xa68,0x2917,0x291d,0xa70, +0x2923,0x2929,0xa78,0x292f,0x2935,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,0xa80,0xa84,0xa88,0xa8c,1,0x293b,1,1,0x2941,0x2947, +0x294d,0x2953,1,1,0xa90,0x2959,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,0xffcc,1,1,1,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc, +0xffcc,0xffcc,0xffcc,0xffcc,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,0xffcc,0xffcc,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,0xffcc,0xffcc,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,0xfe12,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,0xfe12,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc, +0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0xffb8,0xffb8,0xffb8,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,0xfe12,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,0xffcc,1,0xffcc,0xffcc,0xffb8,1,1, +0xffcc,0xffcc,1,1,1,1,1,0xffcc,0xffcc,1,0xffcc,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0xfe12, +1,1,1,1,1,1,1,1,1,0xae2,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289, +0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289, +0x1289,0x1289,0x1289,0x1289,0x1289,0xae2,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289, +0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289, +0x1289,0xae2,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289, +0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0xae2,0x1289,0x1289, +0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289, +0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,0x1289,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,0x3c66,1,0x3c66,0x3c66,0x3c66,0x3c66,0x3c66,0x3c66,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0x3c66, +0x3c66,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,0x3c66,1,1,1,1,0x3c66,1,1,1,0x3c66,0x3c66,0x3c66, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,0x3b97,1, +0x2ad5,0x2ad9,0x2add,0x2ae1,0x2ae5,0x2ae9,0x2aed,0x2af1,0x2af1,0x2af5,0x2af9,0x2afd,0x2b01,0x2b05,0x2b09,0x2b0d, +0x2b11,0x2b15,0x2b19,0x2b1d,0x2b21,0x2b25,0x2b29,0x2b2d,0x2b31,0x2b35,0x2b39,0x2b3d,0x2b41,0x2b45,0x2b49,0x2b4d, +0x2b51,0x2b55,0x2b59,0x2b5d,0x2b61,0x2b65,0x2b69,0x2b6d,0x2b71,0x2b75,0x2b79,0x2b7d,0x2b81,0x2b85,0x2b89,0x2b8d, +0x2b91,0x2b95,0x2b99,0x2b9d,0x2ba1,0x2ba5,0x2ba9,0x2bad,0x2bb1,0x2bb5,0x2bb9,0x2bbd,0x2bc1,0x2bc5,0x2bc9,0x2bcd, +0x2bd1,0x2bd5,0x2bd9,0x2bdd,0x2be1,0x2be5,0x2be9,0x2bed,0x2bf1,0x2bf5,0x2bf9,0x2bfd,0x2c01,0x2c05,0x2c09,0x2c0d, +0x2c11,0x2c15,0x2c19,0x2c1d,0x2c21,0x2c25,0x2c29,0x2c2d,0x2c31,0x2c35,0x2c39,0x2c3d,0x2b21,0x2c41,0x2c45,0x2c49, +0x2c4d,0x2c51,0x2c55,0x2c59,0x2c5d,0x2c61,0x2c65,0x2c69,0x2c6d,0x2c71,0x2c75,0x2c79,0x2c7d,0x2c81,0x2c85,0x2c89, +0x2c8d,0x2c91,0x2c95,0x2c99,0x2c9d,0x2ca1,0x2ca5,0x2ca9,0x2cad,0x2cb1,0x2cb5,0x2cb9,0x2cbd,0x2cc1,0x2cc5,0x2cc9, +0x2ccd,0x2cd1,0x2cd5,0x2cd9,0x2cdd,0x2ce1,0x2ce5,0x2ce9,0x2ced,0x2cf1,0x2cf5,0x2cf9,0x2cfd,0x2d01,0x2d05,0x2d09, +0x2d0d,0x2d11,0x2d15,0x2d19,0x2d1d,0x2d21,0x2d25,0x2d29,0x2d2d,0x2d31,0x2d35,0x2d39,0x2d3d,0x2d41,0x2d45,0x2d49, +0x2d4d,0x2c89,0x2d51,0x2d55,0x2d59,0x2d5d,0x2d61,0x2d65,0x2d69,0x2d6d,0x2c49,0x2d71,0x2d75,0x2d79,0x2d7d,0x2d81, +0x2d85,0x2d89,0x2d8d,0x2d91,0x2d95,0x2d99,0x2d9d,0x2da1,0x2da5,0x2da9,0x2dad,0x2db1,0x2db5,0x2db9,0x2dbd,0x2b21, +0x2dc1,0x2dc5,0x2dc9,0x2dcd,0x2dd1,0x2dd5,0x2dd9,0x2ddd,0x2de1,0x2de5,0x2de9,0x2ded,0x2df1,0x2df5,0x2df9,0x2dfd, +0x2e01,0x2e05,0x2e09,0x2e0d,0x2e11,0x2e15,0x2e19,0x2e1d,0x2e21,0x2e25,0x2e29,0x2c51,0x2e2d,0x2e31,0x2e35,0x2e39, +0x2e3d,0x2e41,0x2e45,0x2e49,0x2e4d,0x2e51,0x2e55,0x2e59,0x2e5d,0x2e61,0x2e65,0x2e69,0x2e6d,0x2e71,0x2e75,0x2e79, +0x2e7d,0x2e81,0x2e85,0x2e89,0x2e8d,0x2e91,0x2e95,0x2e99,0x2e9d,0x2ea1,0x2ea5,0x2ea9,0x2ead,0x2eb1,0x2eb5,0x2eb9, +0x2ebd,0x2ec1,0x2ec5,0x2ec9,0x2ecd,0x2ed1,0x2ed5,0x2ed9,0x2edd,0x2ee1,0x2ee5,0x2ee9,0x2eed,0x2ef1,1,1, +0x2ef5,1,0x2ef9,1,1,0x2efd,0x2f01,0x2f05,0x2f09,0x2f0d,0x2f11,0x2f15,0x2f19,0x2f1d,0x2f21,1, +0x2f25,1,0x2f29,1,1,0x2f2d,0x2f31,1,1,1,0x2f35,0x2f39,0x2f3d,0x2f41,0x2f45,0x2f49, +0x2f4d,0x2f51,0x2f55,0x2f59,0x2f5d,0x2f61,0x2f65,0x2f69,0x2f6d,0x2f71,0x2f75,0x2f79,0x2f7d,0x2f81,0x2f85,0x2f89, +0x2f8d,0x2f91,0x2f95,0x2f99,0x2f9d,0x2fa1,0x2fa5,0x2fa9,0x2fad,0x2fb1,0x2fb5,0x2fb9,0x2fbd,0x2fc1,0x2fc5,0x2fc9, +0x2fcd,0x2fd1,0x2fd5,0x2fd9,0x2fdd,0x2fe1,0x2fe5,0x2d25,0x2fe9,0x2fed,0x2ff1,0x2ff5,0x2ff9,0x2ffd,0x2ffd,0x3001, +0x3005,0x3009,0x300d,0x3011,0x3015,0x3019,0x301d,0x2f2d,0x3021,0x3025,0x3029,0x302d,0x3031,0x3037,1,1, +0x303b,0x303f,0x3043,0x3047,0x304b,0x304f,0x3053,0x3057,0x2f65,0x305b,0x305f,0x3063,0x2ef5,0x3067,0x306b,0x306f, +0x3073,0x3077,0x307b,0x307f,0x3083,0x3087,0x308b,0x308f,0x3093,0x2f89,0x3097,0x2f8d,0x309b,0x309f,0x30a3,0x30a7, +0x30ab,0x2ef9,0x2b75,0x30af,0x30b3,0x30b7,0x2c8d,0x2de9,0x30bb,0x30bf,0x2fa9,0x30c3,0x2fad,0x30c7,0x30cb,0x30cf, +0x2f01,0x30d3,0x30d7,0x30db,0x30df,0x30e3,0x2f05,0x30e7,0x30eb,0x30ef,0x30f3,0x30f7,0x30fb,0x2fe5,0x30ff,0x3103, +0x2d25,0x3107,0x2ff5,0x310b,0x310f,0x3113,0x3117,0x311b,0x3009,0x311f,0x2f29,0x3123,0x300d,0x2c41,0x3127,0x3011, +0x312b,0x3019,0x312f,0x3133,0x3137,0x313b,0x313f,0x3021,0x2f19,0x3143,0x3025,0x3147,0x3029,0x314b,0x2af1,0x314f, +0x3155,0x315b,0x3161,0x3165,0x3169,0x316d,0x3173,0x3179,0x317f,0x3183,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0x3186,0xfe34,0x318c,1,1,1,1,1,1,1,1,1,1,0x3192,0x3198,0x31a0, +0x31aa,0x31b2,0x31b8,0x31be,0x31c4,0x31ca,0x31d0,0x31d6,0x31dc,0x31e2,1,0x31e8,0x31ee,0x31f4,0x31fa,0x3200, +1,0x3206,1,0x320c,0x3212,1,0x3218,0x321e,1,0x3224,0x322a,0x3230,0x3236,0x323c,0x3242,0x3248, +0x324e,0x3254,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffb8,0xffb8,0xffb8,0xffb8,0xffb8,0xffb8, +0xffb8,0xffcc,0xffcc,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0xffb8,1,0xffcc,1,1,1,1,1,1,1,1,0xffcc,0xfe02, +0xffb8,1,1,1,1,0xfe12,1,1,1,1,0xffcc,0xffcc,0xffcc,0xffcc,1,1, +1,1,1,1,1,1,0xffb8,0xffb8,0xffcc,0xffcc,0xffcc,0xffb8,0xffcc,0xffb8,0xffb8,0xffb8, +1,1,0xffcc,0xffb8,0xffcc,0xffb8,1,1,1,1,1,1,1,1,1,1, +0xfe12,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0xfe12, +1,1,1,1,1,1,1,1,1,0xa94,0x295f,0xa9a,0x2969,1,1,1, +1,1,0xaa0,1,1,1,1,1,0x2973,1,1,1,1,1,1,1, +1,1,0xfe12,0xfc0e,1,1,1,1,1,1,1,0xfc00,1,1,1,1, +1,1,0x297d,0x2987,1,0xaa6,0xaac,0xfe12,0xfe12,1,1,1,1,1,1,1, +1,1,1,1,0xfe12,1,1,1,1,1,1,1,1,1,0xfe0e,1, +1,1,1,1,0xfe12,0xfe0e,1,1,1,1,1,1,1,1,1,0xfe0e, +0xfe12,1,1,1,1,1,1,1,1,1,1,1,0xfe0e,0xfe0e,1,0xfc00, +1,1,1,1,1,1,1,0xab2,1,1,1,0x2991,0x299b,0xfe12,1,1, +1,1,1,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,1,1,1,0xfe12,1, +1,1,0xfe0e,1,1,1,1,1,1,1,1,1,0xfc00,1,1,1, +1,1,1,1,1,0xabe,0xfc00,0x29a5,0x29af,0xfc00,0x29b9,1,1,0xfe12,0xfe0e,1, +1,1,1,1,1,1,1,1,1,1,1,0xad0,0xad6,0x29c3,0x29cd,1, +1,1,0xfe12,0xfe0e,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0xfe12,0xfe0e,1,1,1,1,1,1,1,1,0xfc00,1,1, +1,1,0xadc,1,1,0x29d7,1,1,1,1,0xfe12,0xfe12,1,0xfe02,0xfe02,0xfe02, +0xfe02,0xfe02,1,1,1,1,1,1,1,1,1,1,1,0xfe0c,0xfe0c,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,0xfe02,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,0x325a,0x3264,0x3278,0x3290, +0x32a8,0x32c0,0x32d8,0xffb0,0xffb0,0xfe02,0xfe02,0xfe02,1,1,1,0xffc4,0xffb0,0xffb0,0xffb0,1, +1,1,1,1,1,1,1,0xffb8,0xffb8,0xffb8,0xffb8,0xffb8,1,1,0xffcc,0xffcc, +0xffcc,0xffcc,0xffcc,0xffb8,0xffb8,1,1,1,1,1,1,1,1,1,1,0xffcc, +0xffcc,0xffcc,0xffcc,1,1,1,1,1,1,1,1,1,1,1,0x32e6,0x32f0, +0x3304,0x331c,0x3334,0x334c,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc, +0xffcc,0xffcc,0xffcc,0xffcc,1,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,1,0xffcc,0xffcc,1,0xffcc, +0xffcc,0xffcc,0xffcc,0xffcc,1,1,1,1,1,1,1,1,1,1,1,1, +0xffd0,0xffd0,0xffb8,0xffcc,0xffb8,0xffb8,0xffb8,0xffb8,0xffb8,0xffb8,0xffb8,1,1,1,1,1, +1,1,1,1,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xffcc,0xfe0e,1,1,1,1,1, +0x335b,0x335f,0x3363,0x3367,0x336d,0x2f4d,0x3371,0x3375,0x3379,0x337d,0x2f51,0x3381,0x3385,0x3389,0x2f55,0x338f, +0x3393,0x3397,0x339b,0x33a1,0x33a5,0x33a9,0x33ad,0x33b3,0x33b7,0x33bb,0x33bf,0x303f,0x33c3,0x33c9,0x33cd,0x33d1, +0x33d5,0x33d9,0x33dd,0x33e1,0x33e5,0x3053,0x2f59,0x2f5d,0x3057,0x33e9,0x33ed,0x2c59,0x33f1,0x2f61,0x33f5,0x33f9, +0x33fd,0x3401,0x3401,0x3401,0x3405,0x340b,0x340f,0x3413,0x3417,0x341d,0x3421,0x3425,0x3429,0x342d,0x3431,0x3435, +0x3439,0x343d,0x3441,0x3445,0x3449,0x344d,0x344d,0x305f,0x3451,0x3455,0x3459,0x345d,0x2f69,0x3461,0x3465,0x3469, +0x2ebd,0x346d,0x3471,0x3475,0x3479,0x347d,0x3481,0x3485,0x3489,0x348d,0x3493,0x3497,0x349b,0x349f,0x34a3,0x34a7, +0x34ab,0x34b1,0x34b7,0x34bb,0x34bf,0x34c3,0x34c7,0x34cb,0x34cf,0x34d3,0x34d7,0x34d7,0x34db,0x34e1,0x34e5,0x2c49, +0x34e9,0x34ed,0x34f3,0x34f7,0x34fb,0x34ff,0x3503,0x3507,0x2f7d,0x350b,0x350f,0x3513,0x3519,0x351d,0x3523,0x3527, +0x352b,0x352f,0x3533,0x3537,0x353b,0x353f,0x3543,0x3547,0x354b,0x354f,0x3555,0x3559,0x355d,0x3561,0x2b71,0x3565, +0x356b,0x356f,0x356f,0x3575,0x3579,0x3579,0x357d,0x3581,0x3587,0x358d,0x3591,0x3595,0x3599,0x359d,0x35a1,0x35a5, +0x35a9,0x35ad,0x35b1,0x2f81,0x35b5,0x35bb,0x35bf,0x35c3,0x308f,0x35c3,0x35c7,0x2f89,0x35cb,0x35cf,0x35d3,0x35d7, +0x2f8d,0x2b05,0x35db,0x35df,0x35e3,0x35e7,0x35eb,0x35ef,0x35f3,0x35f9,0x35fd,0x3601,0x3605,0x3609,0x360d,0x3613, +0x3617,0x361b,0x361f,0x3623,0x3627,0x362b,0x362f,0x3633,0x2f91,0x3637,0x363b,0x3641,0x3645,0x3649,0x364d,0x2f99, +0x3651,0x3655,0x3659,0x365d,0x3661,0x3665,0x3669,0x366d,0x2b75,0x30af,0x3671,0x3675,0x3679,0x367d,0x3683,0x3687, +0x368b,0x368f,0x2f9d,0x3693,0x3699,0x369d,0x36a1,0x3161,0x36a5,0x36a9,0x36ad,0x36b1,0x36b5,0x36bb,0x36bf,0x36c3, +0x36c7,0x36cd,0x36d1,0x36d5,0x36d9,0x2c8d,0x36dd,0x36e1,0x36e7,0x36ed,0x36f3,0x36f7,0x36fd,0x3701,0x3705,0x3709, +0x370d,0x2fa1,0x2de9,0x3711,0x3715,0x3719,0x371d,0x3723,0x3727,0x372b,0x372f,0x30bf,0x3733,0x3737,0x373d,0x3741, +0x3745,0x374b,0x3751,0x3755,0x30c3,0x3759,0x375d,0x3761,0x3765,0x3769,0x376d,0x3771,0x3777,0x377b,0x3781,0x3785, +0x378b,0x30cb,0x378f,0x3793,0x3799,0x379d,0x37a1,0x37a7,0x37ad,0x37b1,0x37b5,0x37b9,0x37bd,0x37bd,0x37c1,0x37c5, +0x30d3,0x37c9,0x37cd,0x37d1,0x37d5,0x37d9,0x37df,0x37e3,0x2c55,0x37e9,0x37ef,0x37f3,0x37f9,0x37ff,0x3805,0x3809, +0x30eb,0x380d,0x3813,0x3819,0x381f,0x3825,0x3829,0x3829,0x30ef,0x3169,0x382d,0x3831,0x3835,0x3839,0x383f,0x2bbd, +0x30f7,0x3843,0x3847,0x2fcd,0x384d,0x3853,0x2f15,0x3859,0x385d,0x2fdd,0x3861,0x3865,0x3869,0x386f,0x386f,0x3875, +0x3879,0x387d,0x3883,0x3887,0x388b,0x388f,0x3895,0x3899,0x389d,0x38a1,0x38a5,0x38a9,0x38af,0x38b3,0x38b7,0x38bb, +0x38bf,0x38c3,0x38c7,0x38cd,0x38d3,0x38d7,0x38dd,0x38e1,0x38e7,0x38eb,0x2ff5,0x38ef,0x38f5,0x38fb,0x38ff,0x3905, +0x3909,0x390f,0x3913,0x3917,0x391b,0x391f,0x3923,0x3927,0x392d,0x3933,0x3939,0x3575,0x393f,0x3943,0x3947,0x394b, +0x394f,0x3953,0x3957,0x395b,0x395f,0x3963,0x3967,0x396b,0x2c9d,0x3971,0x3975,0x3979,0x397d,0x3981,0x3985,0x3001, +0x3989,0x398d,0x3991,0x3995,0x3999,0x399f,0x39a5,0x39ab,0x39af,0x39b3,0x39b7,0x39bb,0x39c1,0x39c5,0x39cb,0x39cf, +0x39d3,0x39d9,0x39df,0x39e3,0x2ba9,0x39e7,0x39eb,0x39ef,0x39f3,0x39f7,0x39fb,0x3113,0x39ff,0x3a03,0x3a07,0x3a0b, +0x3a0f,0x3a13,0x3a17,0x3a1b,0x3a1f,0x3a23,0x3a29,0x3a2d,0x3a31,0x3a35,0x3a39,0x3a3d,0x3a43,0x3a49,0x3a4d,0x3a51, +0x3127,0x312b,0x3a55,0x3a59,0x3a5f,0x3a63,0x3a67,0x3a6b,0x3a6f,0x3a75,0x3a7b,0x3a7f,0x3a83,0x3a87,0x3a8d,0x312f, +0x3a91,0x3a97,0x3a9d,0x3aa1,0x3aa5,0x3aa9,0x3aaf,0x3ab3,0x3ab7,0x3abb,0x3abf,0x3ac3,0x3ac7,0x3acb,0x3ad1,0x3ad5, +0x3ad9,0x3add,0x3ae3,0x3ae7,0x3aeb,0x3aef,0x3af3,0x3af9,0x3aff,0x3b03,0x3b07,0x3b0b,0x3b11,0x3b15,0x3147,0x3147, +0x3b1b,0x3b1f,0x3b25,0x3b29,0x3b2d,0x3b31,0x3b35,0x3b39,0x3b3d,0x3b41,0x314b,0x3b47,0x3b4b,0x3b4f,0x3b53,0x3b57, +0x3b5b,0x3b61,0x3b65,0x3b6b,0x3b71,0x3b77,0x3b7b,0x3b7f,0x3b83,0x3b87,0x3b8b,0x3b8f,0x3b93,0x3b97,1,1 +}; + +static const UCPTrie norm2_nfc_data_trie={ + norm2_nfc_data_trieIndex, + { norm2_nfc_data_trieData }, + 1788, 7984, + 0x2fc00, 0x30, + 0, 0, + 0, 0, + 0xc4, 0x226, + 0x1, +}; + +static const uint16_t norm2_nfc_data_extraData[7732]={ +0xffff,0xffff,0x8670,0x44dc,0x8670,0x44c0,0x8670,0x44de,0x600,0x180,0x602,0x182,0x604,0x185,0x606,0x186, +0x608,0x200,0x60c,0x205,0x60e,0x44d,0x610,0x189,0x612,0x3d44,0x614,0x18b,0x618,0x39a,0x61e,0x400, +0x622,0x404,0x646,0x3d41,0x64a,0x3c00,0x8650,0x208,0x60e,0x3c04,0x646,0x3c08,0x8662,0x3c0c,0x602,0x20c, +0x604,0x210,0x60e,0x214,0x618,0x218,0x864e,0x18f,0x60e,0x3c14,0x618,0x21c,0x646,0x3c18,0x64e,0x3c20, +0x65a,0x3c24,0x8662,0x3c1c,0x600,0x190,0x602,0x192,0x604,0x195,0x606,0x3d78,0x608,0x225,0x60c,0x228, +0x60e,0x22c,0x610,0x196,0x612,0x3d74,0x618,0x234,0x61e,0x408,0x622,0x40c,0x646,0x3d71,0x64e,0x451, +0x650,0x230,0x65a,0x3c30,0x8660,0x3c34,0x860e,0x3c3c,0x602,0x3e8,0x604,0x238,0x608,0x3c40,0x60c,0x23c, +0x60e,0x240,0x618,0x3cc,0x864e,0x244,0x604,0x248,0x60e,0x3c44,0x610,0x3c4c,0x618,0x43c,0x646,0x3c48, +0x64e,0x3c50,0x865c,0x3c54,0x600,0x198,0x602,0x19a,0x604,0x19c,0x606,0x250,0x608,0x254,0x60c,0x258, +0x60e,0x260,0x610,0x19f,0x612,0x3d90,0x618,0x39e,0x61e,0x410,0x622,0x414,0x646,0x3d94,0x650,0x25c, +0x8660,0x3c58,0x8604,0x268,0x602,0x3c60,0x618,0x3d0,0x646,0x3c64,0x64e,0x26c,0x8662,0x3c68,0x602,0x272, +0x618,0x27a,0x646,0x3c6d,0x64e,0x276,0x65a,0x3c78,0x8662,0x3c74,0x602,0x3c7c,0x60e,0x3c80,0x8646,0x3c84, +0x600,0x3f0,0x602,0x286,0x606,0x1a2,0x60e,0x3c88,0x618,0x28e,0x646,0x3c8c,0x64e,0x28a,0x65a,0x3c94, +0x8662,0x3c90,0x600,0x1a4,0x602,0x1a6,0x604,0x1a9,0x606,0x1ab,0x608,0x299,0x60c,0x29c,0x60e,0x45d, +0x610,0x1ad,0x612,0x3d9c,0x616,0x2a0,0x618,0x3a2,0x61e,0x418,0x622,0x41c,0x636,0x341,0x646,0x3d99, +0x8650,0x3d5,0x602,0x3ca8,0x860e,0x3cac,0x602,0x2a8,0x60e,0x3cb0,0x618,0x2b0,0x61e,0x420,0x622,0x424, +0x646,0x3cb5,0x64e,0x2ac,0x8662,0x3cbc,0x602,0x2b5,0x604,0x2b8,0x60e,0x3cc0,0x618,0x2c1,0x646,0x3cc5, +0x64c,0x430,0x864e,0x2bc,0x60e,0x3cd4,0x618,0x2c8,0x646,0x3cd8,0x64c,0x434,0x64e,0x2c4,0x65a,0x3ce0, +0x8662,0x3cdc,0x600,0x1b2,0x602,0x1b4,0x604,0x1b6,0x606,0x2d1,0x608,0x2d5,0x60c,0x2d8,0x610,0x1b9, +0x612,0x3dcc,0x614,0x2dc,0x616,0x2e0,0x618,0x3a6,0x61e,0x428,0x622,0x42c,0x636,0x35f,0x646,0x3dc8, +0x648,0x3ce4,0x650,0x2e4,0x65a,0x3cec,0x8660,0x3ce8,0x606,0x3cf8,0x8646,0x3cfc,0x600,0x3d00,0x602,0x3d04, +0x604,0x2e8,0x60e,0x3d0c,0x610,0x3d08,0x8646,0x3d10,0x60e,0x3d14,0x8610,0x3d18,0x600,0x3de4,0x602,0x1ba, +0x604,0x2ec,0x606,0x3df0,0x608,0x464,0x60e,0x3d1c,0x610,0x2f0,0x612,0x3dec,0x8646,0x3de8,0x602,0x2f2, +0x604,0x3d20,0x60e,0x2f6,0x618,0x2fa,0x646,0x3d24,0x8662,0x3d28,0x600,0x1c0,0x602,0x1c2,0x604,0x1c5, +0x606,0x1c6,0x608,0x202,0x60c,0x207,0x60e,0x44f,0x610,0x1c9,0x612,0x3d46,0x614,0x1cb,0x618,0x39c, +0x61e,0x402,0x622,0x406,0x646,0x3d43,0x64a,0x3c02,0x8650,0x20a,0x60e,0x3c06,0x646,0x3c0a,0x8662,0x3c0e, +0x602,0x20e,0x604,0x212,0x60e,0x216,0x618,0x21a,0x864e,0x1cf,0x60e,0x3c16,0x618,0x21e,0x646,0x3c1a, +0x64e,0x3c22,0x65a,0x3c26,0x8662,0x3c1e,0x600,0x1d0,0x602,0x1d2,0x604,0x1d5,0x606,0x3d7a,0x608,0x227, +0x60c,0x22a,0x60e,0x22e,0x610,0x1d6,0x612,0x3d76,0x618,0x236,0x61e,0x40a,0x622,0x40e,0x646,0x3d73, +0x64e,0x453,0x650,0x232,0x65a,0x3c32,0x8660,0x3c36,0x860e,0x3c3e,0x602,0x3ea,0x604,0x23a,0x608,0x3c42, +0x60c,0x23e,0x60e,0x242,0x618,0x3ce,0x864e,0x246,0x604,0x24a,0x60e,0x3c46,0x610,0x3c4e,0x618,0x43e, +0x646,0x3c4a,0x64e,0x3c52,0x65c,0x3c56,0x8662,0x3d2c,0x600,0x1d8,0x602,0x1da,0x604,0x1dc,0x606,0x252, +0x608,0x256,0x60c,0x25a,0x610,0x1df,0x612,0x3d92,0x618,0x3a0,0x61e,0x412,0x622,0x416,0x646,0x3d96, +0x650,0x25e,0x8660,0x3c5a,0x604,0x26a,0x8618,0x3e0,0x602,0x3c62,0x618,0x3d2,0x646,0x3c66,0x64e,0x26e, +0x8662,0x3c6a,0x602,0x274,0x618,0x27c,0x646,0x3c6f,0x64e,0x278,0x65a,0x3c7a,0x8662,0x3c76,0x602,0x3c7e, +0x60e,0x3c82,0x8646,0x3c86,0x600,0x3f2,0x602,0x288,0x606,0x1e2,0x60e,0x3c8a,0x618,0x290,0x646,0x3c8e, +0x64e,0x28c,0x65a,0x3c96,0x8662,0x3c92,0x600,0x1e4,0x602,0x1e6,0x604,0x1e9,0x606,0x1eb,0x608,0x29b, +0x60c,0x29e,0x60e,0x45f,0x610,0x1ed,0x612,0x3d9e,0x616,0x2a2,0x618,0x3a4,0x61e,0x41a,0x622,0x41e, +0x636,0x343,0x646,0x3d9b,0x8650,0x3d7,0x602,0x3caa,0x860e,0x3cae,0x602,0x2aa,0x60e,0x3cb2,0x618,0x2b2, +0x61e,0x422,0x622,0x426,0x646,0x3cb7,0x64e,0x2ae,0x8662,0x3cbe,0x602,0x2b7,0x604,0x2ba,0x60e,0x3cc2, +0x618,0x2c3,0x646,0x3cc7,0x64c,0x432,0x864e,0x2be,0x60e,0x3cd6,0x610,0x3d2e,0x618,0x2ca,0x646,0x3cda, +0x64c,0x436,0x64e,0x2c6,0x65a,0x3ce2,0x8662,0x3cde,0x600,0x1f2,0x602,0x1f4,0x604,0x1f6,0x606,0x2d3, +0x608,0x2d7,0x60c,0x2da,0x610,0x1f9,0x612,0x3dce,0x614,0x2de,0x616,0x2e2,0x618,0x3a8,0x61e,0x42a, +0x622,0x42e,0x636,0x361,0x646,0x3dca,0x648,0x3ce6,0x650,0x2e6,0x65a,0x3cee,0x8660,0x3cea,0x606,0x3cfa, +0x8646,0x3cfe,0x600,0x3d02,0x602,0x3d06,0x604,0x2ea,0x60e,0x3d0e,0x610,0x3d0a,0x614,0x3d30,0x8646,0x3d12, +0x60e,0x3d16,0x8610,0x3d1a,0x600,0x3de6,0x602,0x1fa,0x604,0x2ee,0x606,0x3df2,0x608,0x466,0x60e,0x3d1e, +0x610,0x1fe,0x612,0x3dee,0x614,0x3d32,0x8646,0x3dea,0x602,0x2f4,0x604,0x3d22,0x60e,0x2f8,0x618,0x2fc, +0x646,0x3d26,0x8662,0x3d2a,0x600,0x3fda,0x602,0x70a,0x8684,0x3f82,0x602,0x3f8,0x8608,0x3c4,0x8602,0x3fc, +0x602,0x3fa,0x8608,0x3c6,0x8602,0x3fe,0x860e,0x3d36,0x8618,0x3dc,0x8618,0x3de,0x600,0x3f74,0x602,0x70c, +0x608,0x3f72,0x60c,0x3f70,0x626,0x3e11,0x628,0x3e13,0x868a,0x3f78,0x600,0x3f90,0x602,0x710,0x626,0x3e31, +0x8628,0x3e33,0x600,0x3f94,0x602,0x712,0x626,0x3e51,0x628,0x3e53,0x868a,0x3f98,0x600,0x3fb4,0x602,0x714, +0x608,0x3fb2,0x60c,0x3fb0,0x610,0x754,0x626,0x3e71,0x8628,0x3e73,0x600,0x3ff0,0x602,0x718,0x626,0x3e91, +0x8628,0x3e93,0x8628,0x3fd8,0x600,0x3fd4,0x602,0x71c,0x608,0x3fd2,0x60c,0x3fd0,0x610,0x756,0x8628,0x3eb3, +0x600,0x3ff4,0x602,0x71e,0x626,0x3ed1,0x628,0x3ed3,0x868a,0x3ff8,0x600,0x3ee1,0x602,0x759,0x608,0x3f62, +0x60c,0x3f60,0x626,0x3e01,0x628,0x3e03,0x684,0x3f6d,0x868a,0x3f66,0x600,0x3ee4,0x602,0x75a,0x626,0x3e21, +0x8628,0x3e23,0x600,0x3ee9,0x602,0x75d,0x626,0x3e41,0x628,0x3e43,0x684,0x3f8d,0x868a,0x3f86,0x600,0x3eec, +0x602,0x75e,0x608,0x3fa2,0x60c,0x3fa0,0x610,0x795,0x626,0x3e61,0x628,0x3e63,0x8684,0x3fac,0x600,0x3ef0, +0x602,0x798,0x626,0x3e81,0x8628,0x3e83,0x626,0x3fc8,0x8628,0x3fca,0x600,0x3ef4,0x602,0x79a,0x608,0x3fc2, +0x60c,0x3fc0,0x610,0x797,0x626,0x3ea1,0x628,0x3ea3,0x8684,0x3fcc,0x600,0x3ef9,0x602,0x79d,0x626,0x3ec1, +0x628,0x3ec3,0x684,0x3fed,0x868a,0x3fe6,0x602,0x7a6,0x8610,0x7a8,0x8610,0x80e,0x60c,0x9a0,0x8610,0x9a4, +0x8602,0x806,0x600,0x800,0x60c,0x9ac,0x8610,0x802,0x60c,0x982,0x8610,0x9b8,0x8610,0x9bc,0x600,0x81a, +0x608,0x9c4,0x60c,0x832,0x8610,0x9c8,0x8602,0x818,0x8610,0x9cc,0x608,0x9dc,0x60c,0x81c,0x610,0x9e0, +0x8616,0x9e4,0x8610,0x9e8,0x8610,0x9f0,0x8610,0x9d8,0x60c,0x9a2,0x8610,0x9a6,0x8602,0x8a6,0x600,0x8a0, +0x60c,0x9ae,0x8610,0x8a2,0x60c,0x984,0x8610,0x9ba,0x8610,0x9be,0x600,0x8ba,0x608,0x9c6,0x60c,0x872, +0x8610,0x9ca,0x8602,0x8b8,0x8610,0x9ce,0x608,0x9de,0x60c,0x8bc,0x610,0x9e2,0x8616,0x9e6,0x8610,0x9ea, +0x8610,0x9f2,0x8610,0x9da,0x8610,0x8ae,0x861e,0x8ec,0x861e,0x8ee,0x8610,0x9b4,0x8610,0x9b6,0x8610,0x9d4, +0x8610,0x9d6,0xca6,0xc44,0xca8,0xc46,0x8caa,0xc4a,0x8ca8,0xc48,0x8ca8,0xc4c,0x8ca8,0xd84,0x8ca8,0xda6, +0x8ca8,0xd80,0x9278,0x1252,0x9278,0x1262,0x9278,0x1268,0x137c,0x1396,0x93ae,0x1398,0x167c,0x1696,0x16ac,0x1690, +0x96ae,0x1698,0x97ae,0x1728,0x177c,0x1794,0x97ae,0x1798,0x977c,0x1796,0x98ac,0x1890,0x99aa,0x1980,0x1984,0x1995, +0x19aa,0x198e,0x99ac,0x1990,0x1a7c,0x1a94,0x9aae,0x1a98,0x9a7c,0x1a96,0x1b94,0x1bb4,0x1b9e,0x1bb9,0x9bbe,0x1bbc, +0xa05c,0x204c,0xb66a,0x360c,0xb66a,0x3610,0xb66a,0x3614,0xb66a,0x3618,0xb66a,0x361c,0xb66a,0x3624,0xb66a,0x3676, +0xb66a,0x367a,0xb66a,0x3680,0xb66a,0x3682,0xb66a,0x3686,0x600,0x3f9a,0x602,0x3f9c,0x8684,0x3f9e,0x600,0x3fba, +0x602,0x3fbc,0x8684,0x3fbe,0x8670,0x4334,0x8670,0x4336,0x8670,0x435c,0x8670,0x439a,0x8670,0x439e,0x8670,0x439c, +0x8670,0x4408,0x8670,0x4412,0x8670,0x4418,0x8670,0x4448,0x8670,0x444c,0x8670,0x4482,0x8670,0x4488,0x8670,0x448e, +0x8670,0x4492,0x8670,0x44da,0x8670,0x44c4,0x8670,0x44e0,0x8670,0x44e2,0x8670,0x44e8,0x8670,0x44ea,0x8670,0x44f0, +0x8670,0x44f2,0x8670,0x4500,0x8670,0x4502,0x8670,0x45c0,0x8670,0x45c2,0x8670,0x4508,0x8670,0x450a,0x8670,0x4510, +0x8670,0x4512,0x8670,0x45c4,0x8670,0x45c6,0x8670,0x4558,0x8670,0x455a,0x8670,0x455c,0x8670,0x455e,0x8670,0x45d4, +0x8670,0x45d6,0x8670,0x45d8,0x8670,0x45da,0xe132,0x6128,0xe132,0x6098,0xe132,0x609c,0xe132,0x60a0,0xe132,0x60a4, +0xe132,0x60a8,0xe132,0x60ac,0xe132,0x60b0,0xe132,0x60b4,0xe132,0x60b8,0xe132,0x60bc,0xe132,0x60c0,0xe132,0x60c4, +0xe132,0x60ca,0xe132,0x60ce,0xe132,0x60d2,0x6132,0x60e0,0xe134,0x60e2,0x6132,0x60e6,0xe134,0x60e8,0x6132,0x60ec, +0xe134,0x60ee,0x6132,0x60f2,0xe134,0x60f4,0x6132,0x60f8,0xe134,0x60fa,0xe132,0x613c,0xe132,0x61e8,0xe132,0x6158, +0xe132,0x615c,0xe132,0x6160,0xe132,0x6164,0xe132,0x6168,0xe132,0x616c,0xe132,0x6170,0xe132,0x6174,0xe132,0x6178, +0xe132,0x617c,0xe132,0x6180,0xe132,0x6184,0xe132,0x618a,0xe132,0x618e,0xe132,0x6192,0x6132,0x61a0,0xe134,0x61a2, +0x6132,0x61a6,0xe134,0x61a8,0x6132,0x61ac,0xe134,0x61ae,0x6132,0x61b2,0xe134,0x61b4,0x6132,0x61b8,0xe134,0x61ba, +0xe132,0x61ee,0xe132,0x61f0,0xe132,0x61f2,0xe132,0x61f4,0xe132,0x61fc,0xb489,0x2e82,0x2134,0xb489,0x2e82,0x2138, +0xb489,0x2e82,0x2156,0xb489,0x49c2,0x225c,0xb489,0x49c2,0x225e,0x3489,0xcf82,0x2696,0xb489,0xd5c2,0x2698,0x348b, +0x2c02,0x2978,0x348b,0x2e82,0x2976,0xb48b,0x2f42,0x297c,0xb48b,0x6bc2,0x2b74,0xb48b,0x6bc2,0x2b76,0xb48d,0x4c02, +0x3270,2,0xe602,0x41,0x302,0x600,0x3d4c,0x602,0x3d48,0x606,0x3d54,0x8612,0x3d50,0xe602,0x41,0x308, +0x8608,0x3bc,0xe602,0x41,0x30a,0x8602,0x3f4,0xca02,0x43,0x327,0x8602,0x3c10,0xe602,0x45,0x302,0x600, +0x3d80,0x602,0x3d7c,0x606,0x3d88,0x8612,0x3d84,0xe602,0x49,0x308,0x8602,0x3c5c,0xe602,0x4f,0x302,0x600, +0x3da4,0x602,0x3da0,0x606,0x3dac,0x8612,0x3da8,0xe602,0x4f,0x303,0x602,0x3c98,0x608,0x458,0x8610,0x3c9c, +0xe602,0x4f,0x308,0x8608,0x454,0xe602,0x55,0x308,0x600,0x3b6,0x602,0x3ae,0x608,0x3aa,0x8618,0x3b2, +0xe602,0x61,0x302,0x600,0x3d4e,0x602,0x3d4a,0x606,0x3d56,0x8612,0x3d52,0xe602,0x61,0x308,0x8608,0x3be, +0xe602,0x61,0x30a,0x8602,0x3f6,0xca02,0x63,0x327,0x8602,0x3c12,0xe602,0x65,0x302,0x600,0x3d82,0x602, +0x3d7e,0x606,0x3d8a,0x8612,0x3d86,0xe602,0x69,0x308,0x8602,0x3c5e,0xe602,0x6f,0x302,0x600,0x3da6,0x602, +0x3da2,0x606,0x3dae,0x8612,0x3daa,0xe602,0x6f,0x303,0x602,0x3c9a,0x608,0x45a,0x8610,0x3c9e,0xe602,0x6f, +0x308,0x8608,0x456,0xe602,0x75,0x308,0x600,0x3b8,0x602,0x3b0,0x608,0x3ac,0x8618,0x3b4,0xe602,0x41, +0x306,0x600,0x3d60,0x602,0x3d5c,0x606,0x3d68,0x8612,0x3d64,0xe602,0x61,0x306,0x600,0x3d62,0x602,0x3d5e, +0x606,0x3d6a,0x8612,0x3d66,0xe602,0x45,0x304,0x600,0x3c28,0x8602,0x3c2c,0xe602,0x65,0x304,0x600,0x3c2a, +0x8602,0x3c2e,0xe602,0x4f,0x304,0x600,0x3ca0,0x8602,0x3ca4,0xe602,0x6f,0x304,0x600,0x3ca2,0x8602,0x3ca6, +0xe602,0x53,0x301,0x860e,0x3cc8,0xe602,0x73,0x301,0x860e,0x3cca,0xe602,0x53,0x30c,0x860e,0x3ccc,0xe602, +0x73,0x30c,0x860e,0x3cce,0xe602,0x55,0x303,0x8602,0x3cf0,0xe602,0x75,0x303,0x8602,0x3cf2,0xe602,0x55, +0x304,0x8610,0x3cf4,0xe602,0x75,0x304,0x8610,0x3cf6,0xd802,0x4f,0x31b,0x600,0x3db8,0x602,0x3db4,0x606, +0x3dc0,0x612,0x3dbc,0x8646,0x3dc4,0xd802,0x6f,0x31b,0x600,0x3dba,0x602,0x3db6,0x606,0x3dc2,0x612,0x3dbe, +0x8646,0x3dc6,0xd802,0x55,0x31b,0x600,0x3dd4,0x602,0x3dd0,0x606,0x3ddc,0x612,0x3dd8,0x8646,0x3de0,0xd802, +0x75,0x31b,0x600,0x3dd6,0x602,0x3dd2,0x606,0x3dde,0x612,0x3dda,0x8646,0x3de2,0xca02,0x4f,0x328,0x8608, +0x3d8,0xca02,0x6f,0x328,0x8608,0x3da,0xe602,0x41,0x307,0x8608,0x3c0,0xe602,0x61,0x307,0x8608,0x3c2, +0xca02,0x45,0x327,0x860c,0x3c38,0xca02,0x65,0x327,0x860c,0x3c3a,0xe602,0x4f,0x307,0x8608,0x460,0xe602, +0x6f,0x307,0x8608,0x462,0xe602,0x3b1,0x301,0x868a,0x3f68,0xe602,0x3b7,0x301,0x868a,0x3f88,0xe602,0x3b9, +0x308,0x600,0x3fa4,0x602,0x720,0x8684,0x3fae,0xe602,0x3c5,0x308,0x600,0x3fc4,0x602,0x760,0x8684,0x3fce, +0xe602,0x3c9,0x301,0x868a,0x3fe8,2,0xcc6,0xcc2,0x99aa,0x1996,2,0xdd9,0xdcf,0x9b94,0x1bba,0xdc02, +0x4c,0x323,0x8608,0x3c70,0xdc02,0x6c,0x323,0x8608,0x3c72,0xdc02,0x52,0x323,0x8608,0x3cb8,0xdc02,0x72, +0x323,0x8608,0x3cba,0xdc02,0x53,0x323,0x860e,0x3cd0,0xdc02,0x73,0x323,0x860e,0x3cd2,0xdc02,0x41,0x323, +0x604,0x3d58,0x860c,0x3d6c,0xdc02,0x61,0x323,0x604,0x3d5a,0x860c,0x3d6e,0xdc02,0x45,0x323,0x8604,0x3d8c, +0xdc02,0x65,0x323,0x8604,0x3d8e,0xdc02,0x4f,0x323,0x8604,0x3db0,0xdc02,0x6f,0x323,0x8604,0x3db2,0xe602, +0x3b1,0x313,0x600,0x3e05,0x602,0x3e09,0x684,0x3e0d,0x868a,0x3f00,0xe602,0x3b1,0x314,0x600,0x3e07,0x602, +0x3e0b,0x684,0x3e0f,0x868a,0x3f02,0x1f00,0xe643,0x3b1,0x313,0x300,0x868a,0x3f04,0x1f01,0xe643,0x3b1,0x314, +0x300,0x868a,0x3f06,0x1f00,0xe643,0x3b1,0x313,0x301,0x868a,0x3f08,0x1f01,0xe643,0x3b1,0x314,0x301,0x868a, +0x3f0a,0x1f00,0xe643,0x3b1,0x313,0x342,0x868a,0x3f0c,0x1f01,0xe643,0x3b1,0x314,0x342,0x868a,0x3f0e,0xe602, +0x391,0x313,0x600,0x3e15,0x602,0x3e19,0x684,0x3e1d,0x868a,0x3f10,0xe602,0x391,0x314,0x600,0x3e17,0x602, +0x3e1b,0x684,0x3e1f,0x868a,0x3f12,0x1f08,0xe643,0x391,0x313,0x300,0x868a,0x3f14,0x1f09,0xe643,0x391,0x314, +0x300,0x868a,0x3f16,0x1f08,0xe643,0x391,0x313,0x301,0x868a,0x3f18,0x1f09,0xe643,0x391,0x314,0x301,0x868a, +0x3f1a,0x1f08,0xe643,0x391,0x313,0x342,0x868a,0x3f1c,0x1f09,0xe643,0x391,0x314,0x342,0x868a,0x3f1e,0xe602, +0x3b5,0x313,0x600,0x3e24,0x8602,0x3e28,0xe602,0x3b5,0x314,0x600,0x3e26,0x8602,0x3e2a,0xe602,0x395,0x313, +0x600,0x3e34,0x8602,0x3e38,0xe602,0x395,0x314,0x600,0x3e36,0x8602,0x3e3a,0xe602,0x3b7,0x313,0x600,0x3e45, +0x602,0x3e49,0x684,0x3e4d,0x868a,0x3f20,0xe602,0x3b7,0x314,0x600,0x3e47,0x602,0x3e4b,0x684,0x3e4f,0x868a, +0x3f22,0x1f20,0xe643,0x3b7,0x313,0x300,0x868a,0x3f24,0x1f21,0xe643,0x3b7,0x314,0x300,0x868a,0x3f26,0x1f20, +0xe643,0x3b7,0x313,0x301,0x868a,0x3f28,0x1f21,0xe643,0x3b7,0x314,0x301,0x868a,0x3f2a,0x1f20,0xe643,0x3b7, +0x313,0x342,0x868a,0x3f2c,0x1f21,0xe643,0x3b7,0x314,0x342,0x868a,0x3f2e,0xe602,0x397,0x313,0x600,0x3e55, +0x602,0x3e59,0x684,0x3e5d,0x868a,0x3f30,0xe602,0x397,0x314,0x600,0x3e57,0x602,0x3e5b,0x684,0x3e5f,0x868a, +0x3f32,0x1f28,0xe643,0x397,0x313,0x300,0x868a,0x3f34,0x1f29,0xe643,0x397,0x314,0x300,0x868a,0x3f36,0x1f28, +0xe643,0x397,0x313,0x301,0x868a,0x3f38,0x1f29,0xe643,0x397,0x314,0x301,0x868a,0x3f3a,0x1f28,0xe643,0x397, +0x313,0x342,0x868a,0x3f3c,0x1f29,0xe643,0x397,0x314,0x342,0x868a,0x3f3e,0xe602,0x3b9,0x313,0x600,0x3e64, +0x602,0x3e68,0x8684,0x3e6c,0xe602,0x3b9,0x314,0x600,0x3e66,0x602,0x3e6a,0x8684,0x3e6e,0xe602,0x399,0x313, +0x600,0x3e74,0x602,0x3e78,0x8684,0x3e7c,0xe602,0x399,0x314,0x600,0x3e76,0x602,0x3e7a,0x8684,0x3e7e,0xe602, +0x3bf,0x313,0x600,0x3e84,0x8602,0x3e88,0xe602,0x3bf,0x314,0x600,0x3e86,0x8602,0x3e8a,0xe602,0x39f,0x313, +0x600,0x3e94,0x8602,0x3e98,0xe602,0x39f,0x314,0x600,0x3e96,0x8602,0x3e9a,0xe602,0x3c5,0x313,0x600,0x3ea4, +0x602,0x3ea8,0x8684,0x3eac,0xe602,0x3c5,0x314,0x600,0x3ea6,0x602,0x3eaa,0x8684,0x3eae,0xe602,0x3a5,0x314, +0x600,0x3eb6,0x602,0x3eba,0x8684,0x3ebe,0xe602,0x3c9,0x313,0x600,0x3ec5,0x602,0x3ec9,0x684,0x3ecd,0x868a, +0x3f40,0xe602,0x3c9,0x314,0x600,0x3ec7,0x602,0x3ecb,0x684,0x3ecf,0x868a,0x3f42,0x1f60,0xe643,0x3c9,0x313, +0x300,0x868a,0x3f44,0x1f61,0xe643,0x3c9,0x314,0x300,0x868a,0x3f46,0x1f60,0xe643,0x3c9,0x313,0x301,0x868a, +0x3f48,0x1f61,0xe643,0x3c9,0x314,0x301,0x868a,0x3f4a,0x1f60,0xe643,0x3c9,0x313,0x342,0x868a,0x3f4c,0x1f61, +0xe643,0x3c9,0x314,0x342,0x868a,0x3f4e,0xe602,0x3a9,0x313,0x600,0x3ed5,0x602,0x3ed9,0x684,0x3edd,0x868a, +0x3f50,0xe602,0x3a9,0x314,0x600,0x3ed7,0x602,0x3edb,0x684,0x3edf,0x868a,0x3f52,0x1f68,0xe643,0x3a9,0x313, +0x300,0x868a,0x3f54,0x1f69,0xe643,0x3a9,0x314,0x300,0x868a,0x3f56,0x1f68,0xe643,0x3a9,0x313,0x301,0x868a, +0x3f58,0x1f69,0xe643,0x3a9,0x314,0x301,0x868a,0x3f5a,0x1f68,0xe643,0x3a9,0x313,0x342,0x868a,0x3f5c,0x1f69, +0xe643,0x3a9,0x314,0x342,0x868a,0x3f5e,0xe602,0x3b1,0x300,0x868a,0x3f64,0xe602,0x3b7,0x300,0x868a,0x3f84, +0xe602,0x3c9,0x300,0x868a,0x3fe4,0xe602,0x3b1,0x342,0x868a,0x3f6e,0xe602,0x3b7,0x342,0x868a,0x3f8e,0xe602, +0x3c9,0x342,0x868a,0x3fee,3,0xe602,0x41,0x300,0xe602,0x41,0x301,0xe602,0x41,0x303,0xe602,0x45, +0x300,0xe602,0x45,0x301,0xe602,0x45,0x308,0xe602,0x49,0x300,0xe602,0x49,0x301,0xe602,0x49,0x302, +0xe602,0x4e,0x303,0xe602,0x4f,0x300,0xe602,0x4f,0x301,0xe602,0x55,0x300,0xe602,0x55,0x301,0xe602, +0x55,0x302,0xe602,0x59,0x301,0xe602,0x61,0x300,0xe602,0x61,0x301,0xe602,0x61,0x303,0xe602,0x65, +0x300,0xe602,0x65,0x301,0xe602,0x65,0x308,0xe602,0x69,0x300,0xe602,0x69,0x301,0xe602,0x69,0x302, +0xe602,0x6e,0x303,0xe602,0x6f,0x300,0xe602,0x6f,0x301,0xe602,0x75,0x300,0xe602,0x75,0x301,0xe602, +0x75,0x302,0xe602,0x79,0x301,0xe602,0x79,0x308,0xe602,0x41,0x304,0xe602,0x61,0x304,0xca02,0x41, +0x328,0xca02,0x61,0x328,0xe602,0x43,0x301,0xe602,0x63,0x301,0xe602,0x43,0x302,0xe602,0x63,0x302, +0xe602,0x43,0x307,0xe602,0x63,0x307,0xe602,0x43,0x30c,0xe602,0x63,0x30c,0xe602,0x44,0x30c,0xe602, +0x64,0x30c,0xe602,0x45,0x306,0xe602,0x65,0x306,0xe602,0x45,0x307,0xe602,0x65,0x307,0xca02,0x45, +0x328,0xca02,0x65,0x328,0xe602,0x45,0x30c,0xe602,0x65,0x30c,0xe602,0x47,0x302,0xe602,0x67,0x302, +0xe602,0x47,0x306,0xe602,0x67,0x306,0xe602,0x47,0x307,0xe602,0x67,0x307,0xca02,0x47,0x327,0xca02, +0x67,0x327,0xe602,0x48,0x302,0xe602,0x68,0x302,0xe602,0x49,0x303,0xe602,0x69,0x303,0xe602,0x49, +0x304,0xe602,0x69,0x304,0xe602,0x49,0x306,0xe602,0x69,0x306,0xca02,0x49,0x328,0xca02,0x69,0x328, +0xe602,0x49,0x307,0xe602,0x4a,0x302,0xe602,0x6a,0x302,0xca02,0x4b,0x327,0xca02,0x6b,0x327,0xe602, +0x4c,0x301,0xe602,0x6c,0x301,0xca02,0x4c,0x327,0xca02,0x6c,0x327,0xe602,0x4c,0x30c,0xe602,0x6c, +0x30c,0xe602,0x4e,0x301,0xe602,0x6e,0x301,0xca02,0x4e,0x327,0xca02,0x6e,0x327,0xe602,0x4e,0x30c, +0xe602,0x6e,0x30c,0xe602,0x4f,0x306,0xe602,0x6f,0x306,0xe602,0x4f,0x30b,0xe602,0x6f,0x30b,0xe602, +0x52,0x301,0xe602,0x72,0x301,0xca02,0x52,0x327,0xca02,0x72,0x327,0xe602,0x52,0x30c,0xe602,0x72, +0x30c,0xe602,0x53,0x302,0xe602,0x73,0x302,0xca02,0x53,0x327,0xca02,0x73,0x327,0xca02,0x54,0x327, +0xca02,0x74,0x327,0xe602,0x54,0x30c,0xe602,0x74,0x30c,0xe602,0x55,0x306,0xe602,0x75,0x306,0xe602, +0x55,0x30a,0xe602,0x75,0x30a,0xe602,0x55,0x30b,0xe602,0x75,0x30b,0xca02,0x55,0x328,0xca02,0x75, +0x328,0xe602,0x57,0x302,0xe602,0x77,0x302,0xe602,0x59,0x302,0xe602,0x79,0x302,0xe602,0x59,0x308, +0xe602,0x5a,0x301,0xe602,0x7a,0x301,0xe602,0x5a,0x307,0xe602,0x7a,0x307,0xe602,0x5a,0x30c,0xe602, +0x7a,0x30c,0xe602,0x41,0x30c,0xe602,0x61,0x30c,0xe602,0x49,0x30c,0xe602,0x69,0x30c,0xe602,0x4f, +0x30c,0xe602,0x6f,0x30c,0xe602,0x55,0x30c,0xe602,0x75,0x30c,0xdc,0xe643,0x55,0x308,0x304,0xfc, +0xe643,0x75,0x308,0x304,0xdc,0xe643,0x55,0x308,0x301,0xfc,0xe643,0x75,0x308,0x301,0xdc,0xe643, +0x55,0x308,0x30c,0xfc,0xe643,0x75,0x308,0x30c,0xdc,0xe643,0x55,0x308,0x300,0xfc,0xe643,0x75, +0x308,0x300,0xc4,0xe643,0x41,0x308,0x304,0xe4,0xe643,0x61,0x308,0x304,0x226,0xe643,0x41,0x307, +0x304,0x227,0xe643,0x61,0x307,0x304,0xe602,0xc6,0x304,0xe602,0xe6,0x304,0xe602,0x47,0x30c,0xe602, +0x67,0x30c,0xe602,0x4b,0x30c,0xe602,0x6b,0x30c,0x1ea,0xe643,0x4f,0x328,0x304,0x1eb,0xe643,0x6f, +0x328,0x304,0xe602,0x1b7,0x30c,0xe602,0x292,0x30c,0xe602,0x6a,0x30c,0xe602,0x47,0x301,0xe602,0x67, +0x301,0xe602,0x4e,0x300,0xe602,0x6e,0x300,0xc5,0xe643,0x41,0x30a,0x301,0xe5,0xe643,0x61,0x30a, +0x301,0xe602,0xc6,0x301,0xe602,0xe6,0x301,0xe602,0xd8,0x301,0xe602,0xf8,0x301,0xe602,0x41,0x30f, +0xe602,0x61,0x30f,0xe602,0x41,0x311,0xe602,0x61,0x311,0xe602,0x45,0x30f,0xe602,0x65,0x30f,0xe602, +0x45,0x311,0xe602,0x65,0x311,0xe602,0x49,0x30f,0xe602,0x69,0x30f,0xe602,0x49,0x311,0xe602,0x69, +0x311,0xe602,0x4f,0x30f,0xe602,0x6f,0x30f,0xe602,0x4f,0x311,0xe602,0x6f,0x311,0xe602,0x52,0x30f, +0xe602,0x72,0x30f,0xe602,0x52,0x311,0xe602,0x72,0x311,0xe602,0x55,0x30f,0xe602,0x75,0x30f,0xe602, +0x55,0x311,0xe602,0x75,0x311,0xdc02,0x53,0x326,0xdc02,0x73,0x326,0xdc02,0x54,0x326,0xdc02,0x74, +0x326,0xe602,0x48,0x30c,0xe602,0x68,0x30c,0xd6,0xe643,0x4f,0x308,0x304,0xf6,0xe643,0x6f,0x308, +0x304,0xd5,0xe643,0x4f,0x303,0x304,0xf5,0xe643,0x6f,0x303,0x304,0x22e,0xe643,0x4f,0x307,0x304, +0x22f,0xe643,0x6f,0x307,0x304,0xe602,0x59,0x304,0xe602,0x79,0x304,0xe602,0xa8,0x301,0xe602,0x391, +0x301,0xe602,0x395,0x301,0xe602,0x397,0x301,0xe602,0x399,0x301,0xe602,0x39f,0x301,0xe602,0x3a5,0x301, +0xe602,0x3a9,0x301,0x3ca,0xe643,0x3b9,0x308,0x301,0xe602,0x399,0x308,0xe602,0x3a5,0x308,0xe602,0x3b5, +0x301,0xe602,0x3b9,0x301,0x3cb,0xe643,0x3c5,0x308,0x301,0xe602,0x3bf,0x301,0xe602,0x3c5,0x301,0xe602, +0x3d2,0x301,0xe602,0x3d2,0x308,0xe602,0x415,0x300,0xe602,0x415,0x308,0xe602,0x413,0x301,0xe602,0x406, +0x308,0xe602,0x41a,0x301,0xe602,0x418,0x300,0xe602,0x423,0x306,0xe602,0x418,0x306,0xe602,0x438,0x306, +0xe602,0x435,0x300,0xe602,0x435,0x308,0xe602,0x433,0x301,0xe602,0x456,0x308,0xe602,0x43a,0x301,0xe602, +0x438,0x300,0xe602,0x443,0x306,0xe602,0x474,0x30f,0xe602,0x475,0x30f,0xe602,0x416,0x306,0xe602,0x436, +0x306,0xe602,0x410,0x306,0xe602,0x430,0x306,0xe602,0x410,0x308,0xe602,0x430,0x308,0xe602,0x415,0x306, +0xe602,0x435,0x306,0xe602,0x4d8,0x308,0xe602,0x4d9,0x308,0xe602,0x416,0x308,0xe602,0x436,0x308,0xe602, +0x417,0x308,0xe602,0x437,0x308,0xe602,0x418,0x304,0xe602,0x438,0x304,0xe602,0x418,0x308,0xe602,0x438, +0x308,0xe602,0x41e,0x308,0xe602,0x43e,0x308,0xe602,0x4e8,0x308,0xe602,0x4e9,0x308,0xe602,0x42d,0x308, +0xe602,0x44d,0x308,0xe602,0x423,0x304,0xe602,0x443,0x304,0xe602,0x423,0x308,0xe602,0x443,0x308,0xe602, +0x423,0x30b,0xe602,0x443,0x30b,0xe602,0x427,0x308,0xe602,0x447,0x308,0xe602,0x42b,0x308,0xe602,0x44b, +0x308,0xe602,0x627,0x653,0xe602,0x627,0x654,0xe602,0x648,0x654,0xdc02,0x627,0x655,0xe602,0x64a,0x654, +0xe602,0x6d5,0x654,0xe602,0x6c1,0x654,0xe602,0x6d2,0x654,0x702,0x928,0x93c,0x702,0x930,0x93c,0x702, +0x933,0x93c,2,0x9c7,0x9be,2,0x9c7,0x9d7,2,0xb47,0xb56,2,0xb47,0xb3e,2,0xb47, +0xb57,2,0xb92,0xbd7,2,0xbc6,0xbbe,2,0xbc7,0xbbe,2,0xbc6,0xbd7,0x5b02,0xc46,0xc56, +2,0xcbf,0xcd5,2,0xcc6,0xcd5,2,0xcc6,0xcd6,0xcca,0x43,0xcc6,0xcc2,0xcd5,2,0xd46, +0xd3e,2,0xd47,0xd3e,2,0xd46,0xd57,0x902,0xdd9,0xdca,0xddc,0x943,0xdd9,0xdcf,0xdca,2, +0xdd9,0xddf,2,0x1025,0x102e,2,0x1b05,0x1b35,2,0x1b07,0x1b35,2,0x1b09,0x1b35,2,0x1b0b, +0x1b35,2,0x1b0d,0x1b35,2,0x1b11,0x1b35,2,0x1b3a,0x1b35,2,0x1b3c,0x1b35,2,0x1b3e,0x1b35, +2,0x1b3f,0x1b35,2,0x1b42,0x1b35,0xdc02,0x41,0x325,0xdc02,0x61,0x325,0xe602,0x42,0x307,0xe602, +0x62,0x307,0xdc02,0x42,0x323,0xdc02,0x62,0x323,0xdc02,0x42,0x331,0xdc02,0x62,0x331,0xc7,0xe643, +0x43,0x327,0x301,0xe7,0xe643,0x63,0x327,0x301,0xe602,0x44,0x307,0xe602,0x64,0x307,0xdc02,0x44, +0x323,0xdc02,0x64,0x323,0xdc02,0x44,0x331,0xdc02,0x64,0x331,0xca02,0x44,0x327,0xca02,0x64,0x327, +0xdc02,0x44,0x32d,0xdc02,0x64,0x32d,0x112,0xe643,0x45,0x304,0x300,0x113,0xe643,0x65,0x304,0x300, +0x112,0xe643,0x45,0x304,0x301,0x113,0xe643,0x65,0x304,0x301,0xdc02,0x45,0x32d,0xdc02,0x65,0x32d, +0xdc02,0x45,0x330,0xdc02,0x65,0x330,0x228,0xe643,0x45,0x327,0x306,0x229,0xe643,0x65,0x327,0x306, +0xe602,0x46,0x307,0xe602,0x66,0x307,0xe602,0x47,0x304,0xe602,0x67,0x304,0xe602,0x48,0x307,0xe602, +0x68,0x307,0xdc02,0x48,0x323,0xdc02,0x68,0x323,0xe602,0x48,0x308,0xe602,0x68,0x308,0xca02,0x48, +0x327,0xca02,0x68,0x327,0xdc02,0x48,0x32e,0xdc02,0x68,0x32e,0xdc02,0x49,0x330,0xdc02,0x69,0x330, +0xcf,0xe643,0x49,0x308,0x301,0xef,0xe643,0x69,0x308,0x301,0xe602,0x4b,0x301,0xe602,0x6b,0x301, +0xdc02,0x4b,0x323,0xdc02,0x6b,0x323,0xdc02,0x4b,0x331,0xdc02,0x6b,0x331,0x1e36,0xe643,0x4c,0x323, +0x304,0x1e37,0xe643,0x6c,0x323,0x304,0xdc02,0x4c,0x331,0xdc02,0x6c,0x331,0xdc02,0x4c,0x32d,0xdc02, +0x6c,0x32d,0xe602,0x4d,0x301,0xe602,0x6d,0x301,0xe602,0x4d,0x307,0xe602,0x6d,0x307,0xdc02,0x4d, +0x323,0xdc02,0x6d,0x323,0xe602,0x4e,0x307,0xe602,0x6e,0x307,0xdc02,0x4e,0x323,0xdc02,0x6e,0x323, +0xdc02,0x4e,0x331,0xdc02,0x6e,0x331,0xdc02,0x4e,0x32d,0xdc02,0x6e,0x32d,0xd5,0xe643,0x4f,0x303, +0x301,0xf5,0xe643,0x6f,0x303,0x301,0xd5,0xe643,0x4f,0x303,0x308,0xf5,0xe643,0x6f,0x303,0x308, +0x14c,0xe643,0x4f,0x304,0x300,0x14d,0xe643,0x6f,0x304,0x300,0x14c,0xe643,0x4f,0x304,0x301,0x14d, +0xe643,0x6f,0x304,0x301,0xe602,0x50,0x301,0xe602,0x70,0x301,0xe602,0x50,0x307,0xe602,0x70,0x307, +0xe602,0x52,0x307,0xe602,0x72,0x307,0x1e5a,0xe643,0x52,0x323,0x304,0x1e5b,0xe643,0x72,0x323,0x304, +0xdc02,0x52,0x331,0xdc02,0x72,0x331,0xe602,0x53,0x307,0xe602,0x73,0x307,0x15a,0xe643,0x53,0x301, +0x307,0x15b,0xe643,0x73,0x301,0x307,0x160,0xe643,0x53,0x30c,0x307,0x161,0xe643,0x73,0x30c,0x307, +0x1e62,0xe643,0x53,0x323,0x307,0x1e63,0xe643,0x73,0x323,0x307,0xe602,0x54,0x307,0xe602,0x74,0x307, +0xdc02,0x54,0x323,0xdc02,0x74,0x323,0xdc02,0x54,0x331,0xdc02,0x74,0x331,0xdc02,0x54,0x32d,0xdc02, +0x74,0x32d,0xdc02,0x55,0x324,0xdc02,0x75,0x324,0xdc02,0x55,0x330,0xdc02,0x75,0x330,0xdc02,0x55, +0x32d,0xdc02,0x75,0x32d,0x168,0xe643,0x55,0x303,0x301,0x169,0xe643,0x75,0x303,0x301,0x16a,0xe643, +0x55,0x304,0x308,0x16b,0xe643,0x75,0x304,0x308,0xe602,0x56,0x303,0xe602,0x76,0x303,0xdc02,0x56, +0x323,0xdc02,0x76,0x323,0xe602,0x57,0x300,0xe602,0x77,0x300,0xe602,0x57,0x301,0xe602,0x77,0x301, +0xe602,0x57,0x308,0xe602,0x77,0x308,0xe602,0x57,0x307,0xe602,0x77,0x307,0xdc02,0x57,0x323,0xdc02, +0x77,0x323,0xe602,0x58,0x307,0xe602,0x78,0x307,0xe602,0x58,0x308,0xe602,0x78,0x308,0xe602,0x59, +0x307,0xe602,0x79,0x307,0xe602,0x5a,0x302,0xe602,0x7a,0x302,0xdc02,0x5a,0x323,0xdc02,0x7a,0x323, +0xdc02,0x5a,0x331,0xdc02,0x7a,0x331,0xdc02,0x68,0x331,0xe602,0x74,0x308,0xe602,0x77,0x30a,0xe602, +0x79,0x30a,0xe602,0x17f,0x307,0xe602,0x41,0x309,0xe602,0x61,0x309,0xc2,0xe643,0x41,0x302,0x301, +0xe2,0xe643,0x61,0x302,0x301,0xc2,0xe643,0x41,0x302,0x300,0xe2,0xe643,0x61,0x302,0x300,0xc2, +0xe643,0x41,0x302,0x309,0xe2,0xe643,0x61,0x302,0x309,0xc2,0xe643,0x41,0x302,0x303,0xe2,0xe643, +0x61,0x302,0x303,0x1ea0,0xe643,0x41,0x323,0x302,0x1ea1,0xe643,0x61,0x323,0x302,0x102,0xe643,0x41, +0x306,0x301,0x103,0xe643,0x61,0x306,0x301,0x102,0xe643,0x41,0x306,0x300,0x103,0xe643,0x61,0x306, +0x300,0x102,0xe643,0x41,0x306,0x309,0x103,0xe643,0x61,0x306,0x309,0x102,0xe643,0x41,0x306,0x303, +0x103,0xe643,0x61,0x306,0x303,0x1ea0,0xe643,0x41,0x323,0x306,0x1ea1,0xe643,0x61,0x323,0x306,0xe602, +0x45,0x309,0xe602,0x65,0x309,0xe602,0x45,0x303,0xe602,0x65,0x303,0xca,0xe643,0x45,0x302,0x301, +0xea,0xe643,0x65,0x302,0x301,0xca,0xe643,0x45,0x302,0x300,0xea,0xe643,0x65,0x302,0x300,0xca, +0xe643,0x45,0x302,0x309,0xea,0xe643,0x65,0x302,0x309,0xca,0xe643,0x45,0x302,0x303,0xea,0xe643, +0x65,0x302,0x303,0x1eb8,0xe643,0x45,0x323,0x302,0x1eb9,0xe643,0x65,0x323,0x302,0xe602,0x49,0x309, +0xe602,0x69,0x309,0xdc02,0x49,0x323,0xdc02,0x69,0x323,0xe602,0x4f,0x309,0xe602,0x6f,0x309,0xd4, +0xe643,0x4f,0x302,0x301,0xf4,0xe643,0x6f,0x302,0x301,0xd4,0xe643,0x4f,0x302,0x300,0xf4,0xe643, +0x6f,0x302,0x300,0xd4,0xe643,0x4f,0x302,0x309,0xf4,0xe643,0x6f,0x302,0x309,0xd4,0xe643,0x4f, +0x302,0x303,0xf4,0xe643,0x6f,0x302,0x303,0x1ecc,0xe643,0x4f,0x323,0x302,0x1ecd,0xe643,0x6f,0x323, +0x302,0x1a0,0xe643,0x4f,0x31b,0x301,0x1a1,0xe643,0x6f,0x31b,0x301,0x1a0,0xe643,0x4f,0x31b,0x300, +0x1a1,0xe643,0x6f,0x31b,0x300,0x1a0,0xe643,0x4f,0x31b,0x309,0x1a1,0xe643,0x6f,0x31b,0x309,0x1a0, +0xe643,0x4f,0x31b,0x303,0x1a1,0xe643,0x6f,0x31b,0x303,0x1a0,0xdc43,0x4f,0x31b,0x323,0x1a1,0xdc43, +0x6f,0x31b,0x323,0xdc02,0x55,0x323,0xdc02,0x75,0x323,0xe602,0x55,0x309,0xe602,0x75,0x309,0x1af, +0xe643,0x55,0x31b,0x301,0x1b0,0xe643,0x75,0x31b,0x301,0x1af,0xe643,0x55,0x31b,0x300,0x1b0,0xe643, +0x75,0x31b,0x300,0x1af,0xe643,0x55,0x31b,0x309,0x1b0,0xe643,0x75,0x31b,0x309,0x1af,0xe643,0x55, +0x31b,0x303,0x1b0,0xe643,0x75,0x31b,0x303,0x1af,0xdc43,0x55,0x31b,0x323,0x1b0,0xdc43,0x75,0x31b, +0x323,0xe602,0x59,0x300,0xe602,0x79,0x300,0xdc02,0x59,0x323,0xdc02,0x79,0x323,0xe602,0x59,0x309, +0xe602,0x79,0x309,0xe602,0x59,0x303,0xe602,0x79,0x303,0x1f10,0xe643,0x3b5,0x313,0x300,0x1f11,0xe643, +0x3b5,0x314,0x300,0x1f10,0xe643,0x3b5,0x313,0x301,0x1f11,0xe643,0x3b5,0x314,0x301,0x1f18,0xe643,0x395, +0x313,0x300,0x1f19,0xe643,0x395,0x314,0x300,0x1f18,0xe643,0x395,0x313,0x301,0x1f19,0xe643,0x395,0x314, +0x301,0x1f30,0xe643,0x3b9,0x313,0x300,0x1f31,0xe643,0x3b9,0x314,0x300,0x1f30,0xe643,0x3b9,0x313,0x301, +0x1f31,0xe643,0x3b9,0x314,0x301,0x1f30,0xe643,0x3b9,0x313,0x342,0x1f31,0xe643,0x3b9,0x314,0x342,0x1f38, +0xe643,0x399,0x313,0x300,0x1f39,0xe643,0x399,0x314,0x300,0x1f38,0xe643,0x399,0x313,0x301,0x1f39,0xe643, +0x399,0x314,0x301,0x1f38,0xe643,0x399,0x313,0x342,0x1f39,0xe643,0x399,0x314,0x342,0x1f40,0xe643,0x3bf, +0x313,0x300,0x1f41,0xe643,0x3bf,0x314,0x300,0x1f40,0xe643,0x3bf,0x313,0x301,0x1f41,0xe643,0x3bf,0x314, +0x301,0x1f48,0xe643,0x39f,0x313,0x300,0x1f49,0xe643,0x39f,0x314,0x300,0x1f48,0xe643,0x39f,0x313,0x301, +0x1f49,0xe643,0x39f,0x314,0x301,0x1f50,0xe643,0x3c5,0x313,0x300,0x1f51,0xe643,0x3c5,0x314,0x300,0x1f50, +0xe643,0x3c5,0x313,0x301,0x1f51,0xe643,0x3c5,0x314,0x301,0x1f50,0xe643,0x3c5,0x313,0x342,0x1f51,0xe643, +0x3c5,0x314,0x342,0x1f59,0xe643,0x3a5,0x314,0x300,0x1f59,0xe643,0x3a5,0x314,0x301,0x1f59,0xe643,0x3a5, +0x314,0x342,0xe602,0x3b5,0x300,0xe602,0x3b9,0x300,0xe602,0x3bf,0x300,0xe602,0x3c5,0x300,0x1f00,0xf043, +0x3b1,0x313,0x345,0x1f01,0xf043,0x3b1,0x314,0x345,0x1f02,0x345,2,0xf044,0x3b1,0x313,0x300,0x345, +0x1f03,0x345,2,0xf044,0x3b1,0x314,0x300,0x345,0x1f04,0x345,2,0xf044,0x3b1,0x313,0x301,0x345, +0x1f05,0x345,2,0xf044,0x3b1,0x314,0x301,0x345,0x1f06,0x345,2,0xf044,0x3b1,0x313,0x342,0x345, +0x1f07,0x345,2,0xf044,0x3b1,0x314,0x342,0x345,0x1f08,0xf043,0x391,0x313,0x345,0x1f09,0xf043,0x391, +0x314,0x345,0x1f0a,0x345,2,0xf044,0x391,0x313,0x300,0x345,0x1f0b,0x345,2,0xf044,0x391,0x314, +0x300,0x345,0x1f0c,0x345,2,0xf044,0x391,0x313,0x301,0x345,0x1f0d,0x345,2,0xf044,0x391,0x314, +0x301,0x345,0x1f0e,0x345,2,0xf044,0x391,0x313,0x342,0x345,0x1f0f,0x345,2,0xf044,0x391,0x314, +0x342,0x345,0x1f20,0xf043,0x3b7,0x313,0x345,0x1f21,0xf043,0x3b7,0x314,0x345,0x1f22,0x345,2,0xf044, +0x3b7,0x313,0x300,0x345,0x1f23,0x345,2,0xf044,0x3b7,0x314,0x300,0x345,0x1f24,0x345,2,0xf044, +0x3b7,0x313,0x301,0x345,0x1f25,0x345,2,0xf044,0x3b7,0x314,0x301,0x345,0x1f26,0x345,2,0xf044, +0x3b7,0x313,0x342,0x345,0x1f27,0x345,2,0xf044,0x3b7,0x314,0x342,0x345,0x1f28,0xf043,0x397,0x313, +0x345,0x1f29,0xf043,0x397,0x314,0x345,0x1f2a,0x345,2,0xf044,0x397,0x313,0x300,0x345,0x1f2b,0x345, +2,0xf044,0x397,0x314,0x300,0x345,0x1f2c,0x345,2,0xf044,0x397,0x313,0x301,0x345,0x1f2d,0x345, +2,0xf044,0x397,0x314,0x301,0x345,0x1f2e,0x345,2,0xf044,0x397,0x313,0x342,0x345,0x1f2f,0x345, +2,0xf044,0x397,0x314,0x342,0x345,0x1f60,0xf043,0x3c9,0x313,0x345,0x1f61,0xf043,0x3c9,0x314,0x345, +0x1f62,0x345,2,0xf044,0x3c9,0x313,0x300,0x345,0x1f63,0x345,2,0xf044,0x3c9,0x314,0x300,0x345, +0x1f64,0x345,2,0xf044,0x3c9,0x313,0x301,0x345,0x1f65,0x345,2,0xf044,0x3c9,0x314,0x301,0x345, +0x1f66,0x345,2,0xf044,0x3c9,0x313,0x342,0x345,0x1f67,0x345,2,0xf044,0x3c9,0x314,0x342,0x345, +0x1f68,0xf043,0x3a9,0x313,0x345,0x1f69,0xf043,0x3a9,0x314,0x345,0x1f6a,0x345,2,0xf044,0x3a9,0x313, +0x300,0x345,0x1f6b,0x345,2,0xf044,0x3a9,0x314,0x300,0x345,0x1f6c,0x345,2,0xf044,0x3a9,0x313, +0x301,0x345,0x1f6d,0x345,2,0xf044,0x3a9,0x314,0x301,0x345,0x1f6e,0x345,2,0xf044,0x3a9,0x313, +0x342,0x345,0x1f6f,0x345,2,0xf044,0x3a9,0x314,0x342,0x345,0xe602,0x3b1,0x306,0xe602,0x3b1,0x304, +0x1f70,0xf043,0x3b1,0x300,0x345,0xf002,0x3b1,0x345,0x3ac,0xf043,0x3b1,0x301,0x345,0x1fb6,0xf043,0x3b1, +0x342,0x345,0xe602,0x391,0x306,0xe602,0x391,0x304,0xe602,0x391,0x300,0xf002,0x391,0x345,0xe602,0xa8, +0x342,0x1f74,0xf043,0x3b7,0x300,0x345,0xf002,0x3b7,0x345,0x3ae,0xf043,0x3b7,0x301,0x345,0x1fc6,0xf043, +0x3b7,0x342,0x345,0xe602,0x395,0x300,0xe602,0x397,0x300,0xf002,0x397,0x345,0xe602,0x1fbf,0x300,0xe602, +0x1fbf,0x301,0xe602,0x1fbf,0x342,0xe602,0x3b9,0x306,0xe602,0x3b9,0x304,0x3ca,0xe643,0x3b9,0x308,0x300, +0xe602,0x3b9,0x342,0x3ca,0xe643,0x3b9,0x308,0x342,0xe602,0x399,0x306,0xe602,0x399,0x304,0xe602,0x399, +0x300,0xe602,0x1ffe,0x300,0xe602,0x1ffe,0x301,0xe602,0x1ffe,0x342,0xe602,0x3c5,0x306,0xe602,0x3c5,0x304, +0x3cb,0xe643,0x3c5,0x308,0x300,0xe602,0x3c1,0x313,0xe602,0x3c1,0x314,0xe602,0x3c5,0x342,0x3cb,0xe643, +0x3c5,0x308,0x342,0xe602,0x3a5,0x306,0xe602,0x3a5,0x304,0xe602,0x3a5,0x300,0xe602,0x3a1,0x314,0xe602, +0xa8,0x300,0x1f7c,0xf043,0x3c9,0x300,0x345,0xf002,0x3c9,0x345,0x3ce,0xf043,0x3c9,0x301,0x345,0x1ff6, +0xf043,0x3c9,0x342,0x345,0xe602,0x39f,0x300,0xe602,0x3a9,0x300,0xf002,0x3a9,0x345,0x102,0x2190,0x338, +0x102,0x2192,0x338,0x102,0x2194,0x338,0x102,0x21d0,0x338,0x102,0x21d4,0x338,0x102,0x21d2,0x338,0x102, +0x2203,0x338,0x102,0x2208,0x338,0x102,0x220b,0x338,0x102,0x2223,0x338,0x102,0x2225,0x338,0x102,0x223c, +0x338,0x102,0x2243,0x338,0x102,0x2245,0x338,0x102,0x2248,0x338,0x102,0x3d,0x338,0x102,0x2261,0x338, +0x102,0x224d,0x338,0x102,0x3c,0x338,0x102,0x3e,0x338,0x102,0x2264,0x338,0x102,0x2265,0x338,0x102, +0x2272,0x338,0x102,0x2273,0x338,0x102,0x2276,0x338,0x102,0x2277,0x338,0x102,0x227a,0x338,0x102,0x227b, +0x338,0x102,0x2282,0x338,0x102,0x2283,0x338,0x102,0x2286,0x338,0x102,0x2287,0x338,0x102,0x22a2,0x338, +0x102,0x22a8,0x338,0x102,0x22a9,0x338,0x102,0x22ab,0x338,0x102,0x227c,0x338,0x102,0x227d,0x338,0x102, +0x2291,0x338,0x102,0x2292,0x338,0x102,0x22b2,0x338,0x102,0x22b3,0x338,0x102,0x22b4,0x338,0x102,0x22b5, +0x338,0x802,0x304b,0x3099,0x802,0x304d,0x3099,0x802,0x304f,0x3099,0x802,0x3051,0x3099,0x802,0x3053,0x3099, +0x802,0x3055,0x3099,0x802,0x3057,0x3099,0x802,0x3059,0x3099,0x802,0x305b,0x3099,0x802,0x305d,0x3099,0x802, +0x305f,0x3099,0x802,0x3061,0x3099,0x802,0x3064,0x3099,0x802,0x3066,0x3099,0x802,0x3068,0x3099,0x802,0x306f, +0x3099,0x802,0x306f,0x309a,0x802,0x3072,0x3099,0x802,0x3072,0x309a,0x802,0x3075,0x3099,0x802,0x3075,0x309a, +0x802,0x3078,0x3099,0x802,0x3078,0x309a,0x802,0x307b,0x3099,0x802,0x307b,0x309a,0x802,0x3046,0x3099,0x802, +0x309d,0x3099,0x802,0x30ab,0x3099,0x802,0x30ad,0x3099,0x802,0x30af,0x3099,0x802,0x30b1,0x3099,0x802,0x30b3, +0x3099,0x802,0x30b5,0x3099,0x802,0x30b7,0x3099,0x802,0x30b9,0x3099,0x802,0x30bb,0x3099,0x802,0x30bd,0x3099, +0x802,0x30bf,0x3099,0x802,0x30c1,0x3099,0x802,0x30c4,0x3099,0x802,0x30c6,0x3099,0x802,0x30c8,0x3099,0x802, +0x30cf,0x3099,0x802,0x30cf,0x309a,0x802,0x30d2,0x3099,0x802,0x30d2,0x309a,0x802,0x30d5,0x3099,0x802,0x30d5, +0x309a,0x802,0x30d8,0x3099,0x802,0x30d8,0x309a,0x802,0x30db,0x3099,0x802,0x30db,0x309a,0x802,0x30a6,0x3099, +0x802,0x30ef,0x3099,0x802,0x30f0,0x3099,0x802,0x30f1,0x3099,0x802,0x30f2,0x3099,0x802,0x30fd,0x3099,0x704, +0xd804,0xdc99,0xd804,0xdcba,0x704,0xd804,0xdc9b,0xd804,0xdcba,0x704,0xd804,0xdca5,0xd804,0xdcba,4,0xd804, +0xdd31,0xd804,0xdd27,4,0xd804,0xdd32,0xd804,0xdd27,4,0xd804,0xdf47,0xd804,0xdf3e,4,0xd804,0xdf47, +0xd804,0xdf57,4,0xd805,0xdcb9,0xd805,0xdcba,4,0xd805,0xdcb9,0xd805,0xdcb0,4,0xd805,0xdcb9,0xd805, +0xdcbd,4,0xd805,0xddb8,0xd805,0xddaf,4,0xd805,0xddb9,0xd805,0xddaf,4,0xd806,0xdd35,0xd806,0xdd30, +1,0x2b9,1,0x3b,1,0xb7,0x702,0x915,0x93c,0x702,0x916,0x93c,0x702,0x917,0x93c,0x702, +0x91c,0x93c,0x702,0x921,0x93c,0x702,0x922,0x93c,0x702,0x92b,0x93c,0x702,0x92f,0x93c,0x702,0x9a1, +0x9bc,0x702,0x9a2,0x9bc,0x702,0x9af,0x9bc,0x702,0xa32,0xa3c,0x702,0xa38,0xa3c,0x702,0xa16,0xa3c, +0x702,0xa17,0xa3c,0x702,0xa1c,0xa3c,0x702,0xa2b,0xa3c,0x702,0xb21,0xb3c,0x702,0xb22,0xb3c,2, +0xf42,0xfb7,2,0xf4c,0xfb7,2,0xf51,0xfb7,2,0xf56,0xfb7,2,0xf5b,0xfb7,2,0xf40, +0xfb5,0x8202,0xfb2,0xf80,0x8202,0xfb3,0xf80,2,0xf92,0xfb7,2,0xf9c,0xfb7,2,0xfa1,0xfb7, +2,0xfa6,0xfb7,2,0xfab,0xfb7,2,0xf90,0xfb5,1,0x3b9,1,0x60,1,0xb4,1, +0x3a9,1,0x4b,1,0x3008,1,0x3009,0x102,0x2add,0x338,1,0x8c48,1,0x66f4,1,0x8eca, +1,0x8cc8,1,0x6ed1,1,0x4e32,1,0x53e5,1,0x9f9c,1,0x5951,1,0x91d1,1,0x5587, +1,0x5948,1,0x61f6,1,0x7669,1,0x7f85,1,0x863f,1,0x87ba,1,0x88f8,1,0x908f, +1,0x6a02,1,0x6d1b,1,0x70d9,1,0x73de,1,0x843d,1,0x916a,1,0x99f1,1,0x4e82, +1,0x5375,1,0x6b04,1,0x721b,1,0x862d,1,0x9e1e,1,0x5d50,1,0x6feb,1,0x85cd, +1,0x8964,1,0x62c9,1,0x81d8,1,0x881f,1,0x5eca,1,0x6717,1,0x6d6a,1,0x72fc, +1,0x90ce,1,0x4f86,1,0x51b7,1,0x52de,1,0x64c4,1,0x6ad3,1,0x7210,1,0x76e7, +1,0x8001,1,0x8606,1,0x865c,1,0x8def,1,0x9732,1,0x9b6f,1,0x9dfa,1,0x788c, +1,0x797f,1,0x7da0,1,0x83c9,1,0x9304,1,0x9e7f,1,0x8ad6,1,0x58df,1,0x5f04, +1,0x7c60,1,0x807e,1,0x7262,1,0x78ca,1,0x8cc2,1,0x96f7,1,0x58d8,1,0x5c62, +1,0x6a13,1,0x6dda,1,0x6f0f,1,0x7d2f,1,0x7e37,1,0x964b,1,0x52d2,1,0x808b, +1,0x51dc,1,0x51cc,1,0x7a1c,1,0x7dbe,1,0x83f1,1,0x9675,1,0x8b80,1,0x62cf, +1,0x8afe,1,0x4e39,1,0x5be7,1,0x6012,1,0x7387,1,0x7570,1,0x5317,1,0x78fb, +1,0x4fbf,1,0x5fa9,1,0x4e0d,1,0x6ccc,1,0x6578,1,0x7d22,1,0x53c3,1,0x585e, +1,0x7701,1,0x8449,1,0x8aaa,1,0x6bba,1,0x8fb0,1,0x6c88,1,0x62fe,1,0x82e5, +1,0x63a0,1,0x7565,1,0x4eae,1,0x5169,1,0x51c9,1,0x6881,1,0x7ce7,1,0x826f, +1,0x8ad2,1,0x91cf,1,0x52f5,1,0x5442,1,0x5973,1,0x5eec,1,0x65c5,1,0x6ffe, +1,0x792a,1,0x95ad,1,0x9a6a,1,0x9e97,1,0x9ece,1,0x529b,1,0x66c6,1,0x6b77, +1,0x8f62,1,0x5e74,1,0x6190,1,0x6200,1,0x649a,1,0x6f23,1,0x7149,1,0x7489, +1,0x79ca,1,0x7df4,1,0x806f,1,0x8f26,1,0x84ee,1,0x9023,1,0x934a,1,0x5217, +1,0x52a3,1,0x54bd,1,0x70c8,1,0x88c2,1,0x5ec9,1,0x5ff5,1,0x637b,1,0x6bae, +1,0x7c3e,1,0x7375,1,0x4ee4,1,0x56f9,1,0x5dba,1,0x601c,1,0x73b2,1,0x7469, +1,0x7f9a,1,0x8046,1,0x9234,1,0x96f6,1,0x9748,1,0x9818,1,0x4f8b,1,0x79ae, +1,0x91b4,1,0x96b8,1,0x60e1,1,0x4e86,1,0x50da,1,0x5bee,1,0x5c3f,1,0x6599, +1,0x71ce,1,0x7642,1,0x84fc,1,0x907c,1,0x9f8d,1,0x6688,1,0x962e,1,0x5289, +1,0x677b,1,0x67f3,1,0x6d41,1,0x6e9c,1,0x7409,1,0x7559,1,0x786b,1,0x7d10, +1,0x985e,1,0x516d,1,0x622e,1,0x9678,1,0x502b,1,0x5d19,1,0x6dea,1,0x8f2a, +1,0x5f8b,1,0x6144,1,0x6817,1,0x9686,1,0x5229,1,0x540f,1,0x5c65,1,0x6613, +1,0x674e,1,0x68a8,1,0x6ce5,1,0x7406,1,0x75e2,1,0x7f79,1,0x88cf,1,0x88e1, +1,0x91cc,1,0x96e2,1,0x533f,1,0x6eba,1,0x541d,1,0x71d0,1,0x7498,1,0x85fa, +1,0x96a3,1,0x9c57,1,0x9e9f,1,0x6797,1,0x6dcb,1,0x81e8,1,0x7acb,1,0x7b20, +1,0x7c92,1,0x72c0,1,0x7099,1,0x8b58,1,0x4ec0,1,0x8336,1,0x523a,1,0x5207, +1,0x5ea6,1,0x62d3,1,0x7cd6,1,0x5b85,1,0x6d1e,1,0x66b4,1,0x8f3b,1,0x884c, +1,0x964d,1,0x898b,1,0x5ed3,1,0x5140,1,0x55c0,1,0x585a,1,0x6674,1,0x51de, +1,0x732a,1,0x76ca,1,0x793c,1,0x795e,1,0x7965,1,0x798f,1,0x9756,1,0x7cbe, +1,0x7fbd,1,0x8612,1,0x8af8,1,0x9038,1,0x90fd,1,0x98ef,1,0x98fc,1,0x9928, +1,0x9db4,1,0x90de,1,0x96b7,1,0x4fae,1,0x50e7,1,0x514d,1,0x52c9,1,0x52e4, +1,0x5351,1,0x559d,1,0x5606,1,0x5668,1,0x5840,1,0x58a8,1,0x5c64,1,0x5c6e, +1,0x6094,1,0x6168,1,0x618e,1,0x61f2,1,0x654f,1,0x65e2,1,0x6691,1,0x6885, +1,0x6d77,1,0x6e1a,1,0x6f22,1,0x716e,1,0x722b,1,0x7422,1,0x7891,1,0x793e, +1,0x7949,1,0x7948,1,0x7950,1,0x7956,1,0x795d,1,0x798d,1,0x798e,1,0x7a40, +1,0x7a81,1,0x7bc0,1,0x7e09,1,0x7e41,1,0x7f72,1,0x8005,1,0x81ed,1,0x8279, +1,0x8457,1,0x8910,1,0x8996,1,0x8b01,1,0x8b39,1,0x8cd3,1,0x8d08,1,0x8fb6, +1,0x96e3,1,0x97ff,1,0x983b,1,0x6075,2,0xd850,0xdeee,1,0x8218,1,0x4e26,1, +0x51b5,1,0x5168,1,0x4f80,1,0x5145,1,0x5180,1,0x52c7,1,0x52fa,1,0x5555,1, +0x5599,1,0x55e2,1,0x58b3,1,0x5944,1,0x5954,1,0x5a62,1,0x5b28,1,0x5ed2,1, +0x5ed9,1,0x5f69,1,0x5fad,1,0x60d8,1,0x614e,1,0x6108,1,0x6160,1,0x6234,1, +0x63c4,1,0x641c,1,0x6452,1,0x6556,1,0x671b,1,0x6756,1,0x6b79,1,0x6edb,1, +0x6ecb,1,0x701e,1,0x77a7,1,0x7235,1,0x72af,1,0x7471,1,0x7506,1,0x753b,1, +0x761d,1,0x761f,1,0x76db,1,0x76f4,1,0x774a,1,0x7740,1,0x78cc,1,0x7ab1,1, +0x7c7b,1,0x7d5b,1,0x7f3e,1,0x8352,1,0x83ef,1,0x8779,1,0x8941,1,0x8986,1, +0x8abf,1,0x8acb,1,0x8aed,1,0x8b8a,1,0x8f38,1,0x9072,1,0x9199,1,0x9276,1, +0x967c,1,0x97db,1,0x980b,1,0x9b12,2,0xd84a,0xdc4a,2,0xd84a,0xdc44,2,0xd84c,0xdfd5, +1,0x3b9d,1,0x4018,1,0x4039,2,0xd854,0xde49,2,0xd857,0xdcd0,2,0xd85f,0xded3,1, +0x9f43,1,0x9f8e,0xe02,0x5d9,0x5b4,0x1102,0x5f2,0x5b7,0x1802,0x5e9,0x5c1,0x1902,0x5e9,0x5c2,0xfb49, +0x1843,0x5e9,0x5bc,0x5c1,0xfb49,0x1943,0x5e9,0x5bc,0x5c2,0x1102,0x5d0,0x5b7,0x1202,0x5d0,0x5b8,0x1502, +0x5d0,0x5bc,0x1502,0x5d1,0x5bc,0x1502,0x5d2,0x5bc,0x1502,0x5d3,0x5bc,0x1502,0x5d4,0x5bc,0x1502,0x5d5, +0x5bc,0x1502,0x5d6,0x5bc,0x1502,0x5d8,0x5bc,0x1502,0x5d9,0x5bc,0x1502,0x5da,0x5bc,0x1502,0x5db,0x5bc, +0x1502,0x5dc,0x5bc,0x1502,0x5de,0x5bc,0x1502,0x5e0,0x5bc,0x1502,0x5e1,0x5bc,0x1502,0x5e3,0x5bc,0x1502, +0x5e4,0x5bc,0x1502,0x5e6,0x5bc,0x1502,0x5e7,0x5bc,0x1502,0x5e8,0x5bc,0x1502,0x5e9,0x5bc,0x1502,0x5ea, +0x5bc,0x1302,0x5d5,0x5b9,0x1702,0x5d1,0x5bf,0x1702,0x5db,0x5bf,0x1702,0x5e4,0x5bf,0xd804,0xd834,0xdd57, +0xd834,0xdd65,0xd804,0xd834,0xdd58,0xd834,0xdd65,0xd834,0xdd5f,0xd834,0xdd6e,4,0xd846,0xd834,0xdd58,0xd834, +0xdd65,0xd834,0xdd6e,0xd834,0xdd5f,0xd834,0xdd6f,4,0xd846,0xd834,0xdd58,0xd834,0xdd65,0xd834,0xdd6f,0xd834, +0xdd5f,0xd834,0xdd70,4,0xd846,0xd834,0xdd58,0xd834,0xdd65,0xd834,0xdd70,0xd834,0xdd5f,0xd834,0xdd71,4, +0xd846,0xd834,0xdd58,0xd834,0xdd65,0xd834,0xdd71,0xd834,0xdd5f,0xd834,0xdd72,4,0xd846,0xd834,0xdd58,0xd834, +0xdd65,0xd834,0xdd72,0xd804,0xd834,0xddb9,0xd834,0xdd65,0xd804,0xd834,0xddba,0xd834,0xdd65,0xd834,0xddbb,0xd834, +0xdd6e,4,0xd846,0xd834,0xddb9,0xd834,0xdd65,0xd834,0xdd6e,0xd834,0xddbc,0xd834,0xdd6e,4,0xd846,0xd834, +0xddba,0xd834,0xdd65,0xd834,0xdd6e,0xd834,0xddbb,0xd834,0xdd6f,4,0xd846,0xd834,0xddb9,0xd834,0xdd65,0xd834, +0xdd6f,0xd834,0xddbc,0xd834,0xdd6f,4,0xd846,0xd834,0xddba,0xd834,0xdd65,0xd834,0xdd6f,1,0x4e3d,1, +0x4e38,1,0x4e41,2,0xd840,0xdd22,1,0x4f60,1,0x4fbb,1,0x5002,1,0x507a,1,0x5099, +1,0x50cf,1,0x349e,2,0xd841,0xde3a,1,0x5154,1,0x5164,1,0x5177,2,0xd841,0xdd1c, +1,0x34b9,1,0x5167,1,0x518d,2,0xd841,0xdd4b,1,0x5197,1,0x51a4,1,0x4ecc,1, +0x51ac,2,0xd864,0xdddf,1,0x51f5,1,0x5203,1,0x34df,1,0x523b,1,0x5246,1,0x5272, +1,0x5277,1,0x3515,1,0x5305,1,0x5306,1,0x5349,1,0x535a,1,0x5373,1,0x537d, +1,0x537f,2,0xd842,0xde2c,1,0x7070,1,0x53ca,1,0x53df,2,0xd842,0xdf63,1,0x53eb, +1,0x53f1,1,0x5406,1,0x549e,1,0x5438,1,0x5448,1,0x5468,1,0x54a2,1,0x54f6, +1,0x5510,1,0x5553,1,0x5563,1,0x5584,1,0x55ab,1,0x55b3,1,0x55c2,1,0x5716, +1,0x5717,1,0x5651,1,0x5674,1,0x58ee,1,0x57ce,1,0x57f4,1,0x580d,1,0x578b, +1,0x5832,1,0x5831,1,0x58ac,2,0xd845,0xdce4,1,0x58f2,1,0x58f7,1,0x5906,1, +0x591a,1,0x5922,1,0x5962,2,0xd845,0xdea8,2,0xd845,0xdeea,1,0x59ec,1,0x5a1b,1, +0x5a27,1,0x59d8,1,0x5a66,1,0x36ee,1,0x36fc,1,0x5b08,1,0x5b3e,2,0xd846,0xddc8, +1,0x5bc3,1,0x5bd8,1,0x5bf3,2,0xd846,0xdf18,1,0x5bff,1,0x5c06,1,0x5f53,1, +0x5c22,1,0x3781,1,0x5c60,1,0x5cc0,1,0x5c8d,2,0xd847,0xdde4,1,0x5d43,2,0xd847, +0xdde6,1,0x5d6e,1,0x5d6b,1,0x5d7c,1,0x5de1,1,0x5de2,1,0x382f,1,0x5dfd,1, +0x5e28,1,0x5e3d,1,0x5e69,1,0x3862,2,0xd848,0xdd83,1,0x387c,1,0x5eb0,1,0x5eb3, +1,0x5eb6,2,0xd868,0xdf92,1,0x5efe,2,0xd848,0xdf31,1,0x8201,1,0x5f22,1,0x38c7, +2,0xd84c,0xdeb8,2,0xd858,0xddda,1,0x5f62,1,0x5f6b,1,0x38e3,1,0x5f9a,1,0x5fcd, +1,0x5fd7,1,0x5ff9,1,0x6081,1,0x393a,1,0x391c,2,0xd849,0xded4,1,0x60c7,1, +0x6148,1,0x614c,1,0x617a,1,0x61b2,1,0x61a4,1,0x61af,1,0x61de,1,0x6210,1, +0x621b,1,0x625d,1,0x62b1,1,0x62d4,1,0x6350,2,0xd84a,0xdf0c,1,0x633d,1,0x62fc, +1,0x6368,1,0x6383,1,0x63e4,2,0xd84a,0xdff1,1,0x6422,1,0x63c5,1,0x63a9,1, +0x3a2e,1,0x6469,1,0x647e,1,0x649d,1,0x6477,1,0x3a6c,1,0x656c,2,0xd84c,0xdc0a, +1,0x65e3,1,0x66f8,1,0x6649,1,0x3b19,1,0x3b08,1,0x3ae4,1,0x5192,1,0x5195, +1,0x6700,1,0x669c,1,0x80ad,1,0x43d9,1,0x6721,1,0x675e,1,0x6753,2,0xd84c, +0xdfc3,1,0x3b49,1,0x67fa,1,0x6785,1,0x6852,2,0xd84d,0xdc6d,1,0x688e,1,0x681f, +1,0x6914,1,0x6942,1,0x69a3,1,0x69ea,1,0x6aa8,2,0xd84d,0xdea3,1,0x6adb,1, +0x3c18,1,0x6b21,2,0xd84e,0xdca7,1,0x6b54,1,0x3c4e,1,0x6b72,1,0x6b9f,1,0x6bbb, +2,0xd84e,0xde8d,2,0xd847,0xdd0b,2,0xd84e,0xdefa,1,0x6c4e,2,0xd84f,0xdcbc,1,0x6cbf, +1,0x6ccd,1,0x6c67,1,0x6d16,1,0x6d3e,1,0x6d69,1,0x6d78,1,0x6d85,2,0xd84f, +0xdd1e,1,0x6d34,1,0x6e2f,1,0x6e6e,1,0x3d33,1,0x6ec7,2,0xd84f,0xded1,1,0x6df9, +1,0x6f6e,2,0xd84f,0xdf5e,2,0xd84f,0xdf8e,1,0x6fc6,1,0x7039,1,0x701b,1,0x3d96, +1,0x704a,1,0x707d,1,0x7077,1,0x70ad,2,0xd841,0xdd25,1,0x7145,2,0xd850,0xde63, +1,0x719c,2,0xd850,0xdfab,1,0x7228,1,0x7250,2,0xd851,0xde08,1,0x7280,1,0x7295, +2,0xd851,0xdf35,2,0xd852,0xdc14,1,0x737a,1,0x738b,1,0x3eac,1,0x73a5,1,0x3eb8, +1,0x7447,1,0x745c,1,0x7485,1,0x74ca,1,0x3f1b,1,0x7524,2,0xd853,0xdc36,1, +0x753e,2,0xd853,0xdc92,2,0xd848,0xdd9f,1,0x7610,2,0xd853,0xdfa1,2,0xd853,0xdfb8,2, +0xd854,0xdc44,1,0x3ffc,1,0x4008,2,0xd854,0xdcf3,2,0xd854,0xdcf2,2,0xd854,0xdd19,2, +0xd854,0xdd33,1,0x771e,1,0x771f,1,0x778b,1,0x4046,1,0x4096,2,0xd855,0xdc1d,1, +0x784e,1,0x40e3,2,0xd855,0xde26,2,0xd855,0xde9a,2,0xd855,0xdec5,1,0x79eb,1,0x412f, +1,0x7a4a,1,0x7a4f,2,0xd856,0xdd7c,2,0xd856,0xdea7,1,0x7aee,1,0x4202,2,0xd856, +0xdfab,1,0x7bc6,1,0x7bc9,1,0x4227,2,0xd857,0xdc80,1,0x7cd2,1,0x42a0,1,0x7ce8, +1,0x7ce3,1,0x7d00,2,0xd857,0xdf86,1,0x7d63,1,0x4301,1,0x7dc7,1,0x7e02,1, +0x7e45,1,0x4334,2,0xd858,0xde28,2,0xd858,0xde47,1,0x4359,2,0xd858,0xded9,1,0x7f7a, +2,0xd858,0xdf3e,1,0x7f95,1,0x7ffa,2,0xd859,0xdcda,2,0xd859,0xdd23,1,0x8060,2, +0xd859,0xdda8,1,0x8070,2,0xd84c,0xdf5f,1,0x43d5,1,0x80b2,1,0x8103,1,0x440b,1, +0x813e,1,0x5ab5,2,0xd859,0xdfa7,2,0xd859,0xdfb5,2,0xd84c,0xdf93,2,0xd84c,0xdf9c,1, +0x8204,1,0x8f9e,1,0x446b,1,0x8291,1,0x828b,1,0x829d,1,0x52b3,1,0x82b1,1, +0x82b3,1,0x82bd,1,0x82e6,2,0xd85a,0xdf3c,1,0x831d,1,0x8363,1,0x83ad,1,0x8323, +1,0x83bd,1,0x83e7,1,0x8353,1,0x83ca,1,0x83cc,1,0x83dc,2,0xd85b,0xdc36,2, +0xd85b,0xdd6b,2,0xd85b,0xdcd5,1,0x452b,1,0x84f1,1,0x84f3,1,0x8516,2,0xd85c,0xdfca, +1,0x8564,2,0xd85b,0xdf2c,1,0x455d,1,0x4561,2,0xd85b,0xdfb1,2,0xd85c,0xdcd2,1, +0x456b,1,0x8650,1,0x8667,1,0x8669,1,0x86a9,1,0x8688,1,0x870e,1,0x86e2,1, +0x8728,1,0x876b,1,0x8786,1,0x45d7,1,0x87e1,1,0x8801,1,0x45f9,1,0x8860,1, +0x8863,2,0xd85d,0xde67,1,0x88d7,1,0x88de,1,0x4635,1,0x88fa,1,0x34bb,2,0xd85e, +0xdcae,2,0xd85e,0xdd66,1,0x46be,1,0x46c7,1,0x8aa0,1,0x8c55,2,0xd85f,0xdca8,1, +0x8cab,1,0x8cc1,1,0x8d1b,1,0x8d77,2,0xd85f,0xdf2f,2,0xd842,0xdc04,1,0x8dcb,1, +0x8dbc,1,0x8df0,2,0xd842,0xdcde,1,0x8ed4,2,0xd861,0xddd2,2,0xd861,0xdded,1,0x9094, +1,0x90f1,1,0x9111,2,0xd861,0xdf2e,1,0x911b,1,0x9238,1,0x92d7,1,0x92d8,1, +0x927c,1,0x93f9,1,0x9415,2,0xd862,0xdffa,1,0x958b,1,0x4995,1,0x95b7,2,0xd863, +0xdd77,1,0x49e6,1,0x96c3,1,0x5db2,1,0x9723,2,0xd864,0xdd45,2,0xd864,0xde1a,1, +0x4a6e,1,0x4a76,1,0x97e0,2,0xd865,0xdc0a,1,0x4ab2,2,0xd865,0xdc96,1,0x9829,2, +0xd865,0xddb6,1,0x98e2,1,0x4b33,1,0x9929,1,0x99a7,1,0x99c2,1,0x99fe,1,0x4bce, +2,0xd866,0xdf30,1,0x9c40,1,0x9cfd,1,0x4cce,1,0x4ced,1,0x9d67,2,0xd868,0xdcce, +1,0x4cf8,2,0xd868,0xdd05,2,0xd868,0xde0e,2,0xd868,0xde91,1,0x9ebb,1,0x4d56,1, +0x9ef9,1,0x9efe,1,0x9f05,1,0x9f0f,1,0x9f16,1,0x9f3b,2,0xd869,0xde00,0x3ac,0xe642, +0x3b1,0x301,0x3ad,0xe642,0x3b5,0x301,0x3ae,0xe642,0x3b7,0x301,0x3af,0xe642,0x3b9,0x301,0x3cc,0xe642, +0x3bf,0x301,0x3cd,0xe642,0x3c5,0x301,0x3ce,0xe642,0x3c9,0x301,0x386,0xe642,0x391,0x301,0x388,0xe642, +0x395,0x301,0x389,0xe642,0x397,0x301,0x390,1,0xe643,0x3b9,0x308,0x301,0x38a,0xe642,0x399,0x301, +0x3b0,1,0xe643,0x3c5,0x308,0x301,0x38e,0xe642,0x3a5,0x301,0x385,0xe642,0xa8,0x301,0x38c,0xe642, +0x39f,0x301,0x38f,0xe642,0x3a9,0x301,0xc5,0xe642,0x41,0x30a,0xe6e6,0xe681,0x300,0xe6e6,0xe681,0x301, +0xe6e6,0xe681,0x313,0xe6e6,0xe682,0x308,0x301,0x8100,0x8282,0xf71,0xf72,0x8100,0x8482,0xf71,0xf74,0x8100, +0x8282,0xf71,0xf80,0 +}; + +static const uint8_t norm2_nfc_data_smallFCD[256]={ +0xc0,0xef,3,0x7f,0xdf,0x70,0xcf,0x87,0xd7,0xe6,0x66,0x46,0x66,0x46,0x66,0x5b, +0x12,0,0,4,0,0,0,0x43,0x20,2,0x69,0xae,0xc2,0xc0,0xff,0xff, +0xc0,0x72,0xbf,0,0,0,0,0,0,0,0x40,0,0x80,0x88,0,0, +0xfe,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0x98,0,0xc3,0x66,0xe0,0x80,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,3,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,7,0,0,2,0 +}; + +#endif // INCLUDED_FROM_NORMALIZER2_CPP diff --git a/intl/icu/source/common/norm2allmodes.h b/intl/icu/source/common/norm2allmodes.h new file mode 100644 index 0000000000..6347fba9cb --- /dev/null +++ b/intl/icu/source/common/norm2allmodes.h @@ -0,0 +1,405 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2014, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* norm2allmodes.h +* +* created on: 2014sep07 +* created by: Markus W. Scherer +*/ + +#ifndef __NORM2ALLMODES_H__ +#define __NORM2ALLMODES_H__ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_NORMALIZATION + +#include "unicode/edits.h" +#include "unicode/normalizer2.h" +#include "unicode/stringoptions.h" +#include "unicode/unistr.h" +#include "cpputils.h" +#include "normalizer2impl.h" + +U_NAMESPACE_BEGIN + +// Intermediate class: +// Has Normalizer2Impl and does boilerplate argument checking and setup. +class Normalizer2WithImpl : public Normalizer2 { +public: + Normalizer2WithImpl(const Normalizer2Impl &ni) : impl(ni) {} + virtual ~Normalizer2WithImpl(); + + // normalize + virtual UnicodeString & + normalize(const UnicodeString &src, + UnicodeString &dest, + UErrorCode &errorCode) const override { + if(U_FAILURE(errorCode)) { + dest.setToBogus(); + return dest; + } + const char16_t *sArray=src.getBuffer(); + if(&dest==&src || sArray==nullptr) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + dest.setToBogus(); + return dest; + } + dest.remove(); + ReorderingBuffer buffer(impl, dest); + if(buffer.init(src.length(), errorCode)) { + normalize(sArray, sArray+src.length(), buffer, errorCode); + } + return dest; + } + virtual void + normalize(const char16_t *src, const char16_t *limit, + ReorderingBuffer &buffer, UErrorCode &errorCode) const = 0; + + // normalize and append + virtual UnicodeString & + normalizeSecondAndAppend(UnicodeString &first, + const UnicodeString &second, + UErrorCode &errorCode) const override { + return normalizeSecondAndAppend(first, second, true, errorCode); + } + virtual UnicodeString & + append(UnicodeString &first, + const UnicodeString &second, + UErrorCode &errorCode) const override { + return normalizeSecondAndAppend(first, second, false, errorCode); + } + UnicodeString & + normalizeSecondAndAppend(UnicodeString &first, + const UnicodeString &second, + UBool doNormalize, + UErrorCode &errorCode) const { + uprv_checkCanGetBuffer(first, errorCode); + if(U_FAILURE(errorCode)) { + return first; + } + const char16_t *secondArray=second.getBuffer(); + if(&first==&second || secondArray==nullptr) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + return first; + } + int32_t firstLength=first.length(); + UnicodeString safeMiddle; + { + ReorderingBuffer buffer(impl, first); + if(buffer.init(firstLength+second.length(), errorCode)) { + normalizeAndAppend(secondArray, secondArray+second.length(), doNormalize, + safeMiddle, buffer, errorCode); + } + } // The ReorderingBuffer destructor finalizes the first string. + if(U_FAILURE(errorCode)) { + // Restore the modified suffix of the first string. + first.replace(firstLength-safeMiddle.length(), 0x7fffffff, safeMiddle); + } + return first; + } + virtual void + normalizeAndAppend(const char16_t *src, const char16_t *limit, UBool doNormalize, + UnicodeString &safeMiddle, + ReorderingBuffer &buffer, UErrorCode &errorCode) const = 0; + virtual UBool + getDecomposition(UChar32 c, UnicodeString &decomposition) const override { + char16_t buffer[4]; + int32_t length; + const char16_t *d=impl.getDecomposition(c, buffer, length); + if(d==nullptr) { + return false; + } + if(d==buffer) { + decomposition.setTo(buffer, length); // copy the string (Jamos from Hangul syllable c) + } else { + decomposition.setTo(false, d, length); // read-only alias + } + return true; + } + virtual UBool + getRawDecomposition(UChar32 c, UnicodeString &decomposition) const override { + char16_t buffer[30]; + int32_t length; + const char16_t *d=impl.getRawDecomposition(c, buffer, length); + if(d==nullptr) { + return false; + } + if(d==buffer) { + decomposition.setTo(buffer, length); // copy the string (algorithmic decomposition) + } else { + decomposition.setTo(false, d, length); // read-only alias + } + return true; + } + virtual UChar32 + composePair(UChar32 a, UChar32 b) const override { + return impl.composePair(a, b); + } + + virtual uint8_t + getCombiningClass(UChar32 c) const override { + return impl.getCC(impl.getNorm16(c)); + } + + // quick checks + virtual UBool + isNormalized(const UnicodeString &s, UErrorCode &errorCode) const override { + if(U_FAILURE(errorCode)) { + return false; + } + const char16_t *sArray=s.getBuffer(); + if(sArray==nullptr) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + return false; + } + const char16_t *sLimit=sArray+s.length(); + return sLimit==spanQuickCheckYes(sArray, sLimit, errorCode); + } + virtual UNormalizationCheckResult + quickCheck(const UnicodeString &s, UErrorCode &errorCode) const override { + return Normalizer2WithImpl::isNormalized(s, errorCode) ? UNORM_YES : UNORM_NO; + } + virtual int32_t + spanQuickCheckYes(const UnicodeString &s, UErrorCode &errorCode) const override { + if(U_FAILURE(errorCode)) { + return 0; + } + const char16_t *sArray=s.getBuffer(); + if(sArray==nullptr) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + return (int32_t)(spanQuickCheckYes(sArray, sArray+s.length(), errorCode)-sArray); + } + virtual const char16_t * + spanQuickCheckYes(const char16_t *src, const char16_t *limit, UErrorCode &errorCode) const = 0; + + virtual UNormalizationCheckResult getQuickCheck(UChar32) const { + return UNORM_YES; + } + + const Normalizer2Impl &impl; +}; + +class DecomposeNormalizer2 : public Normalizer2WithImpl { +public: + DecomposeNormalizer2(const Normalizer2Impl &ni) : Normalizer2WithImpl(ni) {} + virtual ~DecomposeNormalizer2(); + +private: + virtual void + normalize(const char16_t *src, const char16_t *limit, + ReorderingBuffer &buffer, UErrorCode &errorCode) const override { + impl.decompose(src, limit, &buffer, errorCode); + } + using Normalizer2WithImpl::normalize; // Avoid warning about hiding base class function. + virtual void + normalizeAndAppend(const char16_t *src, const char16_t *limit, UBool doNormalize, + UnicodeString &safeMiddle, + ReorderingBuffer &buffer, UErrorCode &errorCode) const override { + impl.decomposeAndAppend(src, limit, doNormalize, safeMiddle, buffer, errorCode); + } + + void + normalizeUTF8(uint32_t options, StringPiece src, ByteSink &sink, + Edits *edits, UErrorCode &errorCode) const override { + if (U_FAILURE(errorCode)) { + return; + } + if (edits != nullptr && (options & U_EDITS_NO_RESET) == 0) { + edits->reset(); + } + const uint8_t *s = reinterpret_cast(src.data()); + impl.decomposeUTF8(options, s, s + src.length(), &sink, edits, errorCode); + sink.Flush(); + } + virtual UBool + isNormalizedUTF8(StringPiece sp, UErrorCode &errorCode) const override { + if(U_FAILURE(errorCode)) { + return false; + } + const uint8_t *s = reinterpret_cast(sp.data()); + const uint8_t *sLimit = s + sp.length(); + return sLimit == impl.decomposeUTF8(0, s, sLimit, nullptr, nullptr, errorCode); + } + + virtual const char16_t * + spanQuickCheckYes(const char16_t *src, const char16_t *limit, UErrorCode &errorCode) const override { + return impl.decompose(src, limit, nullptr, errorCode); + } + using Normalizer2WithImpl::spanQuickCheckYes; // Avoid warning about hiding base class function. + virtual UNormalizationCheckResult getQuickCheck(UChar32 c) const override { + return impl.isDecompYes(impl.getNorm16(c)) ? UNORM_YES : UNORM_NO; + } + virtual UBool hasBoundaryBefore(UChar32 c) const override { + return impl.hasDecompBoundaryBefore(c); + } + virtual UBool hasBoundaryAfter(UChar32 c) const override { + return impl.hasDecompBoundaryAfter(c); + } + virtual UBool isInert(UChar32 c) const override { + return impl.isDecompInert(c); + } +}; + +class ComposeNormalizer2 : public Normalizer2WithImpl { +public: + ComposeNormalizer2(const Normalizer2Impl &ni, UBool fcc) : + Normalizer2WithImpl(ni), onlyContiguous(fcc) {} + virtual ~ComposeNormalizer2(); + +private: + virtual void + normalize(const char16_t *src, const char16_t *limit, + ReorderingBuffer &buffer, UErrorCode &errorCode) const override { + impl.compose(src, limit, onlyContiguous, true, buffer, errorCode); + } + using Normalizer2WithImpl::normalize; // Avoid warning about hiding base class function. + + void + normalizeUTF8(uint32_t options, StringPiece src, ByteSink &sink, + Edits *edits, UErrorCode &errorCode) const override { + if (U_FAILURE(errorCode)) { + return; + } + if (edits != nullptr && (options & U_EDITS_NO_RESET) == 0) { + edits->reset(); + } + const uint8_t *s = reinterpret_cast(src.data()); + impl.composeUTF8(options, onlyContiguous, s, s + src.length(), + &sink, edits, errorCode); + sink.Flush(); + } + + virtual void + normalizeAndAppend(const char16_t *src, const char16_t *limit, UBool doNormalize, + UnicodeString &safeMiddle, + ReorderingBuffer &buffer, UErrorCode &errorCode) const override { + impl.composeAndAppend(src, limit, doNormalize, onlyContiguous, safeMiddle, buffer, errorCode); + } + + virtual UBool + isNormalized(const UnicodeString &s, UErrorCode &errorCode) const override { + if(U_FAILURE(errorCode)) { + return false; + } + const char16_t *sArray=s.getBuffer(); + if(sArray==nullptr) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + return false; + } + UnicodeString temp; + ReorderingBuffer buffer(impl, temp); + if(!buffer.init(5, errorCode)) { // small destCapacity for substring normalization + return false; + } + return impl.compose(sArray, sArray+s.length(), onlyContiguous, false, buffer, errorCode); + } + virtual UBool + isNormalizedUTF8(StringPiece sp, UErrorCode &errorCode) const override { + if(U_FAILURE(errorCode)) { + return false; + } + const uint8_t *s = reinterpret_cast(sp.data()); + return impl.composeUTF8(0, onlyContiguous, s, s + sp.length(), nullptr, nullptr, errorCode); + } + virtual UNormalizationCheckResult + quickCheck(const UnicodeString &s, UErrorCode &errorCode) const override { + if(U_FAILURE(errorCode)) { + return UNORM_MAYBE; + } + const char16_t *sArray=s.getBuffer(); + if(sArray==nullptr) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + return UNORM_MAYBE; + } + UNormalizationCheckResult qcResult=UNORM_YES; + impl.composeQuickCheck(sArray, sArray+s.length(), onlyContiguous, &qcResult); + return qcResult; + } + virtual const char16_t * + spanQuickCheckYes(const char16_t *src, const char16_t *limit, UErrorCode &) const override { + return impl.composeQuickCheck(src, limit, onlyContiguous, nullptr); + } + using Normalizer2WithImpl::spanQuickCheckYes; // Avoid warning about hiding base class function. + virtual UNormalizationCheckResult getQuickCheck(UChar32 c) const override { + return impl.getCompQuickCheck(impl.getNorm16(c)); + } + virtual UBool hasBoundaryBefore(UChar32 c) const override { + return impl.hasCompBoundaryBefore(c); + } + virtual UBool hasBoundaryAfter(UChar32 c) const override { + return impl.hasCompBoundaryAfter(c, onlyContiguous); + } + virtual UBool isInert(UChar32 c) const override { + return impl.isCompInert(c, onlyContiguous); + } + + const UBool onlyContiguous; +}; + +class FCDNormalizer2 : public Normalizer2WithImpl { +public: + FCDNormalizer2(const Normalizer2Impl &ni) : Normalizer2WithImpl(ni) {} + virtual ~FCDNormalizer2(); + +private: + virtual void + normalize(const char16_t *src, const char16_t *limit, + ReorderingBuffer &buffer, UErrorCode &errorCode) const override { + impl.makeFCD(src, limit, &buffer, errorCode); + } + using Normalizer2WithImpl::normalize; // Avoid warning about hiding base class function. + virtual void + normalizeAndAppend(const char16_t *src, const char16_t *limit, UBool doNormalize, + UnicodeString &safeMiddle, + ReorderingBuffer &buffer, UErrorCode &errorCode) const override { + impl.makeFCDAndAppend(src, limit, doNormalize, safeMiddle, buffer, errorCode); + } + virtual const char16_t * + spanQuickCheckYes(const char16_t *src, const char16_t *limit, UErrorCode &errorCode) const override { + return impl.makeFCD(src, limit, nullptr, errorCode); + } + using Normalizer2WithImpl::spanQuickCheckYes; // Avoid warning about hiding base class function. + virtual UBool hasBoundaryBefore(UChar32 c) const override { + return impl.hasFCDBoundaryBefore(c); + } + virtual UBool hasBoundaryAfter(UChar32 c) const override { + return impl.hasFCDBoundaryAfter(c); + } + virtual UBool isInert(UChar32 c) const override { + return impl.isFCDInert(c); + } +}; + +struct Norm2AllModes : public UMemory { + Norm2AllModes(Normalizer2Impl *i) + : impl(i), comp(*i, false), decomp(*i), fcd(*i), fcc(*i, true) {} + ~Norm2AllModes(); + + static Norm2AllModes *createInstance(Normalizer2Impl *impl, UErrorCode &errorCode); + static Norm2AllModes *createNFCInstance(UErrorCode &errorCode); + static Norm2AllModes *createInstance(const char *packageName, + const char *name, + UErrorCode &errorCode); + + static const Norm2AllModes *getNFCInstance(UErrorCode &errorCode); + static const Norm2AllModes *getNFKCInstance(UErrorCode &errorCode); + static const Norm2AllModes *getNFKC_CFInstance(UErrorCode &errorCode); + + Normalizer2Impl *impl; + ComposeNormalizer2 comp; + DecomposeNormalizer2 decomp; + FCDNormalizer2 fcd; + ComposeNormalizer2 fcc; +}; + +U_NAMESPACE_END + +#endif // !UCONFIG_NO_NORMALIZATION +#endif // __NORM2ALLMODES_H__ diff --git a/intl/icu/source/common/normalizer2.cpp b/intl/icu/source/common/normalizer2.cpp new file mode 100644 index 0000000000..6856736b88 --- /dev/null +++ b/intl/icu/source/common/normalizer2.cpp @@ -0,0 +1,572 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2009-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: normalizer2.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2009nov22 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_NORMALIZATION + +#include "unicode/edits.h" +#include "unicode/normalizer2.h" +#include "unicode/stringoptions.h" +#include "unicode/unistr.h" +#include "unicode/unorm.h" +#include "cstring.h" +#include "mutex.h" +#include "norm2allmodes.h" +#include "normalizer2impl.h" +#include "uassert.h" +#include "ucln_cmn.h" + +using icu::Normalizer2Impl; + +#if NORM2_HARDCODE_NFC_DATA +// NFC/NFD data machine-generated by gennorm2 --csource +#define INCLUDED_FROM_NORMALIZER2_CPP +#include "norm2_nfc_data.h" +#endif + +U_NAMESPACE_BEGIN + +// Public API dispatch via Normalizer2 subclasses -------------------------- *** + +Normalizer2::~Normalizer2() {} + +void +Normalizer2::normalizeUTF8(uint32_t /*options*/, StringPiece src, ByteSink &sink, + Edits *edits, UErrorCode &errorCode) const { + if (U_FAILURE(errorCode)) { + return; + } + if (edits != nullptr) { + errorCode = U_UNSUPPORTED_ERROR; + return; + } + UnicodeString src16 = UnicodeString::fromUTF8(src); + normalize(src16, errorCode).toUTF8(sink); +} + +UBool +Normalizer2::getRawDecomposition(UChar32, UnicodeString &) const { + return false; +} + +UChar32 +Normalizer2::composePair(UChar32, UChar32) const { + return U_SENTINEL; +} + +uint8_t +Normalizer2::getCombiningClass(UChar32 /*c*/) const { + return 0; +} + +UBool +Normalizer2::isNormalizedUTF8(StringPiece s, UErrorCode &errorCode) const { + return U_SUCCESS(errorCode) && isNormalized(UnicodeString::fromUTF8(s), errorCode); +} + +// Normalizer2 implementation for the old UNORM_NONE. +class NoopNormalizer2 : public Normalizer2 { + virtual ~NoopNormalizer2(); + + virtual UnicodeString & + normalize(const UnicodeString &src, + UnicodeString &dest, + UErrorCode &errorCode) const override { + if(U_SUCCESS(errorCode)) { + if(&dest!=&src) { + dest=src; + } else { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + } + } + return dest; + } + virtual void + normalizeUTF8(uint32_t options, StringPiece src, ByteSink &sink, + Edits *edits, UErrorCode &errorCode) const override { + if(U_SUCCESS(errorCode)) { + if (edits != nullptr) { + if ((options & U_EDITS_NO_RESET) == 0) { + edits->reset(); + } + edits->addUnchanged(src.length()); + } + if ((options & U_OMIT_UNCHANGED_TEXT) == 0) { + sink.Append(src.data(), src.length()); + } + sink.Flush(); + } + } + + virtual UnicodeString & + normalizeSecondAndAppend(UnicodeString &first, + const UnicodeString &second, + UErrorCode &errorCode) const override { + if(U_SUCCESS(errorCode)) { + if(&first!=&second) { + first.append(second); + } else { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + } + } + return first; + } + virtual UnicodeString & + append(UnicodeString &first, + const UnicodeString &second, + UErrorCode &errorCode) const override { + if(U_SUCCESS(errorCode)) { + if(&first!=&second) { + first.append(second); + } else { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + } + } + return first; + } + virtual UBool + getDecomposition(UChar32, UnicodeString &) const override { + return false; + } + // No need to override the default getRawDecomposition(). + virtual UBool + isNormalized(const UnicodeString &, UErrorCode &errorCode) const override { + return U_SUCCESS(errorCode); + } + virtual UBool + isNormalizedUTF8(StringPiece, UErrorCode &errorCode) const override { + return U_SUCCESS(errorCode); + } + virtual UNormalizationCheckResult + quickCheck(const UnicodeString &, UErrorCode &) const override { + return UNORM_YES; + } + virtual int32_t + spanQuickCheckYes(const UnicodeString &s, UErrorCode &) const override { + return s.length(); + } + virtual UBool hasBoundaryBefore(UChar32) const override { return true; } + virtual UBool hasBoundaryAfter(UChar32) const override { return true; } + virtual UBool isInert(UChar32) const override { return true; } +}; + +NoopNormalizer2::~NoopNormalizer2() {} + +Normalizer2WithImpl::~Normalizer2WithImpl() {} + +DecomposeNormalizer2::~DecomposeNormalizer2() {} + +ComposeNormalizer2::~ComposeNormalizer2() {} + +FCDNormalizer2::~FCDNormalizer2() {} + +// instance cache ---------------------------------------------------------- *** + +U_CDECL_BEGIN +static UBool U_CALLCONV uprv_normalizer2_cleanup(); +U_CDECL_END + +static Normalizer2 *noopSingleton; +static icu::UInitOnce noopInitOnce {}; + +static void U_CALLCONV initNoopSingleton(UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return; + } + noopSingleton=new NoopNormalizer2; + if(noopSingleton==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + ucln_common_registerCleanup(UCLN_COMMON_NORMALIZER2, uprv_normalizer2_cleanup); +} + +const Normalizer2 *Normalizer2Factory::getNoopInstance(UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { return nullptr; } + umtx_initOnce(noopInitOnce, &initNoopSingleton, errorCode); + return noopSingleton; +} + +const Normalizer2Impl * +Normalizer2Factory::getImpl(const Normalizer2 *norm2) { + return &((Normalizer2WithImpl *)norm2)->impl; +} + +Norm2AllModes::~Norm2AllModes() { + delete impl; +} + +Norm2AllModes * +Norm2AllModes::createInstance(Normalizer2Impl *impl, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + delete impl; + return nullptr; + } + Norm2AllModes *allModes=new Norm2AllModes(impl); + if(allModes==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + delete impl; + return nullptr; + } + return allModes; +} + +#if NORM2_HARDCODE_NFC_DATA +Norm2AllModes * +Norm2AllModes::createNFCInstance(UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return nullptr; + } + Normalizer2Impl *impl=new Normalizer2Impl; + if(impl==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + impl->init(norm2_nfc_data_indexes, &norm2_nfc_data_trie, + norm2_nfc_data_extraData, norm2_nfc_data_smallFCD); + return createInstance(impl, errorCode); +} + +static Norm2AllModes *nfcSingleton; + +static icu::UInitOnce nfcInitOnce {}; + +static void U_CALLCONV initNFCSingleton(UErrorCode &errorCode) { + nfcSingleton=Norm2AllModes::createNFCInstance(errorCode); + ucln_common_registerCleanup(UCLN_COMMON_NORMALIZER2, uprv_normalizer2_cleanup); +} + +const Norm2AllModes * +Norm2AllModes::getNFCInstance(UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { return nullptr; } + umtx_initOnce(nfcInitOnce, &initNFCSingleton, errorCode); + return nfcSingleton; +} + +const Normalizer2 * +Normalizer2::getNFCInstance(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFCInstance(errorCode); + return allModes!=nullptr ? &allModes->comp : nullptr; +} + +const Normalizer2 * +Normalizer2::getNFDInstance(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFCInstance(errorCode); + return allModes!=nullptr ? &allModes->decomp : nullptr; +} + +const Normalizer2 *Normalizer2Factory::getFCDInstance(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFCInstance(errorCode); + return allModes!=nullptr ? &allModes->fcd : nullptr; +} + +const Normalizer2 *Normalizer2Factory::getFCCInstance(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFCInstance(errorCode); + return allModes!=nullptr ? &allModes->fcc : nullptr; +} + +const Normalizer2Impl * +Normalizer2Factory::getNFCImpl(UErrorCode &errorCode) { + const Norm2AllModes *allModes=Norm2AllModes::getNFCInstance(errorCode); + return allModes!=nullptr ? allModes->impl : nullptr; +} +#endif // NORM2_HARDCODE_NFC_DATA + +U_CDECL_BEGIN + +static UBool U_CALLCONV uprv_normalizer2_cleanup() { + delete noopSingleton; + noopSingleton = nullptr; + noopInitOnce.reset(); +#if NORM2_HARDCODE_NFC_DATA + delete nfcSingleton; + nfcSingleton = nullptr; + nfcInitOnce.reset(); +#endif + return true; +} + +U_CDECL_END + +U_NAMESPACE_END + +// C API ------------------------------------------------------------------- *** + +U_NAMESPACE_USE + +U_CAPI const UNormalizer2 * U_EXPORT2 +unorm2_getNFCInstance(UErrorCode *pErrorCode) { + return (const UNormalizer2 *)Normalizer2::getNFCInstance(*pErrorCode); +} + +U_CAPI const UNormalizer2 * U_EXPORT2 +unorm2_getNFDInstance(UErrorCode *pErrorCode) { + return (const UNormalizer2 *)Normalizer2::getNFDInstance(*pErrorCode); +} + +U_CAPI void U_EXPORT2 +unorm2_close(UNormalizer2 *norm2) { + delete (Normalizer2 *)norm2; +} + +U_CAPI int32_t U_EXPORT2 +unorm2_normalize(const UNormalizer2 *norm2, + const char16_t *src, int32_t length, + char16_t *dest, int32_t capacity, + UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { + return 0; + } + if( (src==nullptr ? length!=0 : length<-1) || + (dest==nullptr ? capacity!=0 : capacity<0) || + (src==dest && src!=nullptr) + ) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + UnicodeString destString(dest, 0, capacity); + // length==0: Nothing to do, and n2wi->normalize(nullptr, nullptr, buffer, ...) would crash. + if(length!=0) { + const Normalizer2 *n2=(const Normalizer2 *)norm2; + const Normalizer2WithImpl *n2wi=dynamic_cast(n2); + if(n2wi!=nullptr) { + // Avoid duplicate argument checking and support NUL-terminated src. + ReorderingBuffer buffer(n2wi->impl, destString); + if(buffer.init(length, *pErrorCode)) { + n2wi->normalize(src, length>=0 ? src+length : nullptr, buffer, *pErrorCode); + } + } else { + UnicodeString srcString(length<0, src, length); + n2->normalize(srcString, destString, *pErrorCode); + } + } + return destString.extract(dest, capacity, *pErrorCode); +} + +static int32_t +normalizeSecondAndAppend(const UNormalizer2 *norm2, + char16_t *first, int32_t firstLength, int32_t firstCapacity, + const char16_t *second, int32_t secondLength, + UBool doNormalize, + UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { + return 0; + } + if( (second==nullptr ? secondLength!=0 : secondLength<-1) || + (first==nullptr ? (firstCapacity!=0 || firstLength!=0) : + (firstCapacity<0 || firstLength<-1)) || + (first==second && first!=nullptr) + ) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + UnicodeString firstString(first, firstLength, firstCapacity); + firstLength=firstString.length(); // In case it was -1. + // secondLength==0: Nothing to do, and n2wi->normalizeAndAppend(nullptr, nullptr, buffer, ...) would crash. + if(secondLength!=0) { + const Normalizer2 *n2=(const Normalizer2 *)norm2; + const Normalizer2WithImpl *n2wi=dynamic_cast(n2); + if(n2wi!=nullptr) { + // Avoid duplicate argument checking and support NUL-terminated src. + UnicodeString safeMiddle; + { + ReorderingBuffer buffer(n2wi->impl, firstString); + if(buffer.init(firstLength+secondLength+1, *pErrorCode)) { // destCapacity>=-1 + n2wi->normalizeAndAppend(second, secondLength>=0 ? second+secondLength : nullptr, + doNormalize, safeMiddle, buffer, *pErrorCode); + } + } // The ReorderingBuffer destructor finalizes firstString. + if(U_FAILURE(*pErrorCode) || firstString.length()>firstCapacity) { + // Restore the modified suffix of the first string. + // This does not restore first[] array contents between firstLength and firstCapacity. + // (That might be uninitialized memory, as far as we know.) + if(first!=nullptr) { /* don't dereference nullptr */ + safeMiddle.extract(0, 0x7fffffff, first+firstLength-safeMiddle.length()); + if(firstLengthnormalizeSecondAndAppend(firstString, secondString, *pErrorCode); + } else { + n2->append(firstString, secondString, *pErrorCode); + } + } + } + return firstString.extract(first, firstCapacity, *pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +unorm2_normalizeSecondAndAppend(const UNormalizer2 *norm2, + char16_t *first, int32_t firstLength, int32_t firstCapacity, + const char16_t *second, int32_t secondLength, + UErrorCode *pErrorCode) { + return normalizeSecondAndAppend(norm2, + first, firstLength, firstCapacity, + second, secondLength, + true, pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +unorm2_append(const UNormalizer2 *norm2, + char16_t *first, int32_t firstLength, int32_t firstCapacity, + const char16_t *second, int32_t secondLength, + UErrorCode *pErrorCode) { + return normalizeSecondAndAppend(norm2, + first, firstLength, firstCapacity, + second, secondLength, + false, pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +unorm2_getDecomposition(const UNormalizer2 *norm2, + UChar32 c, char16_t *decomposition, int32_t capacity, + UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { + return 0; + } + if(decomposition==nullptr ? capacity!=0 : capacity<0) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + UnicodeString destString(decomposition, 0, capacity); + if(reinterpret_cast(norm2)->getDecomposition(c, destString)) { + return destString.extract(decomposition, capacity, *pErrorCode); + } else { + return -1; + } +} + +U_CAPI int32_t U_EXPORT2 +unorm2_getRawDecomposition(const UNormalizer2 *norm2, + UChar32 c, char16_t *decomposition, int32_t capacity, + UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { + return 0; + } + if(decomposition==nullptr ? capacity!=0 : capacity<0) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + UnicodeString destString(decomposition, 0, capacity); + if(reinterpret_cast(norm2)->getRawDecomposition(c, destString)) { + return destString.extract(decomposition, capacity, *pErrorCode); + } else { + return -1; + } +} + +U_CAPI UChar32 U_EXPORT2 +unorm2_composePair(const UNormalizer2 *norm2, UChar32 a, UChar32 b) { + return reinterpret_cast(norm2)->composePair(a, b); +} + +U_CAPI uint8_t U_EXPORT2 +unorm2_getCombiningClass(const UNormalizer2 *norm2, UChar32 c) { + return reinterpret_cast(norm2)->getCombiningClass(c); +} + +U_CAPI UBool U_EXPORT2 +unorm2_isNormalized(const UNormalizer2 *norm2, + const char16_t *s, int32_t length, + UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { + return 0; + } + if((s==nullptr && length!=0) || length<-1) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + UnicodeString sString(length<0, s, length); + return ((const Normalizer2 *)norm2)->isNormalized(sString, *pErrorCode); +} + +U_CAPI UNormalizationCheckResult U_EXPORT2 +unorm2_quickCheck(const UNormalizer2 *norm2, + const char16_t *s, int32_t length, + UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { + return UNORM_NO; + } + if((s==nullptr && length!=0) || length<-1) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return UNORM_NO; + } + UnicodeString sString(length<0, s, length); + return ((const Normalizer2 *)norm2)->quickCheck(sString, *pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +unorm2_spanQuickCheckYes(const UNormalizer2 *norm2, + const char16_t *s, int32_t length, + UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { + return 0; + } + if((s==nullptr && length!=0) || length<-1) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + UnicodeString sString(length<0, s, length); + return ((const Normalizer2 *)norm2)->spanQuickCheckYes(sString, *pErrorCode); +} + +U_CAPI UBool U_EXPORT2 +unorm2_hasBoundaryBefore(const UNormalizer2 *norm2, UChar32 c) { + return ((const Normalizer2 *)norm2)->hasBoundaryBefore(c); +} + +U_CAPI UBool U_EXPORT2 +unorm2_hasBoundaryAfter(const UNormalizer2 *norm2, UChar32 c) { + return ((const Normalizer2 *)norm2)->hasBoundaryAfter(c); +} + +U_CAPI UBool U_EXPORT2 +unorm2_isInert(const UNormalizer2 *norm2, UChar32 c) { + return ((const Normalizer2 *)norm2)->isInert(c); +} + +// Some properties APIs ---------------------------------------------------- *** + +U_CAPI uint8_t U_EXPORT2 +u_getCombiningClass(UChar32 c) { + UErrorCode errorCode=U_ZERO_ERROR; + const Normalizer2 *nfd=Normalizer2::getNFDInstance(errorCode); + if(U_SUCCESS(errorCode)) { + return nfd->getCombiningClass(c); + } else { + return 0; + } +} + +U_CFUNC uint16_t +unorm_getFCD16(UChar32 c) { + UErrorCode errorCode=U_ZERO_ERROR; + const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode); + if(U_SUCCESS(errorCode)) { + return impl->getFCD16(c); + } else { + return 0; + } +} + +#endif // !UCONFIG_NO_NORMALIZATION diff --git a/intl/icu/source/common/normalizer2impl.cpp b/intl/icu/source/common/normalizer2impl.cpp new file mode 100644 index 0000000000..cdf570d76b --- /dev/null +++ b/intl/icu/source/common/normalizer2impl.cpp @@ -0,0 +1,2806 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2009-2014, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: normalizer2impl.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2009nov22 +* created by: Markus W. Scherer +*/ + +// #define UCPTRIE_DEBUG + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_NORMALIZATION + +#include "unicode/bytestream.h" +#include "unicode/edits.h" +#include "unicode/normalizer2.h" +#include "unicode/stringoptions.h" +#include "unicode/ucptrie.h" +#include "unicode/udata.h" +#include "unicode/umutablecptrie.h" +#include "unicode/ustring.h" +#include "unicode/utf16.h" +#include "unicode/utf8.h" +#include "bytesinkutil.h" +#include "cmemory.h" +#include "mutex.h" +#include "normalizer2impl.h" +#include "putilimp.h" +#include "uassert.h" +#include "ucptrie_impl.h" +#include "uset_imp.h" +#include "uvector.h" + +U_NAMESPACE_BEGIN + +namespace { + +/** + * UTF-8 lead byte for minNoMaybeCP. + * Can be lower than the actual lead byte for c. + * Typically U+0300 for NFC/NFD, U+00A0 for NFKC/NFKD, U+0041 for NFKC_Casefold. + */ +inline uint8_t leadByteForCP(UChar32 c) { + if (c <= 0x7f) { + return (uint8_t)c; + } else if (c <= 0x7ff) { + return (uint8_t)(0xc0+(c>>6)); + } else { + // Should not occur because ccc(U+0300)!=0. + return 0xe0; + } +} + +/** + * Returns the code point from one single well-formed UTF-8 byte sequence + * between cpStart and cpLimit. + * + * Trie UTF-8 macros do not assemble whole code points (for efficiency). + * When we do need the code point, we call this function. + * We should not need it for normalization-inert data (norm16==0). + * Illegal sequences yield the error value norm16==0 just like real normalization-inert code points. + */ +UChar32 codePointFromValidUTF8(const uint8_t *cpStart, const uint8_t *cpLimit) { + // Similar to U8_NEXT_UNSAFE(s, i, c). + U_ASSERT(cpStart < cpLimit); + uint8_t c = *cpStart; + switch(cpLimit-cpStart) { + case 1: + return c; + case 2: + return ((c&0x1f)<<6) | (cpStart[1]&0x3f); + case 3: + // no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (char16_t) + return (char16_t)((c<<12) | ((cpStart[1]&0x3f)<<6) | (cpStart[2]&0x3f)); + case 4: + return ((c&7)<<18) | ((cpStart[1]&0x3f)<<12) | ((cpStart[2]&0x3f)<<6) | (cpStart[3]&0x3f); + default: + UPRV_UNREACHABLE_EXIT; // Should not occur. + } +} + +/** + * Returns the last code point in [start, p[ if it is valid and in U+1000..U+D7FF. + * Otherwise returns a negative value. + */ +UChar32 previousHangulOrJamo(const uint8_t *start, const uint8_t *p) { + if ((p - start) >= 3) { + p -= 3; + uint8_t l = *p; + uint8_t t1, t2; + if (0xe1 <= l && l <= 0xed && + (t1 = (uint8_t)(p[1] - 0x80)) <= 0x3f && + (t2 = (uint8_t)(p[2] - 0x80)) <= 0x3f && + (l < 0xed || t1 <= 0x1f)) { + return ((l & 0xf) << 12) | (t1 << 6) | t2; + } + } + return U_SENTINEL; +} + +/** + * Returns the offset from the Jamo T base if [src, limit[ starts with a single Jamo T code point. + * Otherwise returns a negative value. + */ +int32_t getJamoTMinusBase(const uint8_t *src, const uint8_t *limit) { + // Jamo T: E1 86 A8..E1 87 82 + if ((limit - src) >= 3 && *src == 0xe1) { + if (src[1] == 0x86) { + uint8_t t = src[2]; + // The first Jamo T is U+11A8 but JAMO_T_BASE is 11A7. + // Offset 0 does not correspond to any conjoining Jamo. + if (0xa8 <= t && t <= 0xbf) { + return t - 0xa7; + } + } else if (src[1] == 0x87) { + uint8_t t = src[2]; + if ((int8_t)t <= (int8_t)0x82u) { + return t - (0xa7 - 0x40); + } + } + } + return -1; +} + +void +appendCodePointDelta(const uint8_t *cpStart, const uint8_t *cpLimit, int32_t delta, + ByteSink &sink, Edits *edits) { + char buffer[U8_MAX_LENGTH]; + int32_t length; + int32_t cpLength = (int32_t)(cpLimit - cpStart); + if (cpLength == 1) { + // The builder makes ASCII map to ASCII. + buffer[0] = (uint8_t)(*cpStart + delta); + length = 1; + } else { + int32_t trail = *(cpLimit-1) + delta; + if (0x80 <= trail && trail <= 0xbf) { + // The delta only changes the last trail byte. + --cpLimit; + length = 0; + do { buffer[length++] = *cpStart++; } while (cpStart < cpLimit); + buffer[length++] = (uint8_t)trail; + } else { + // Decode the code point, add the delta, re-encode. + UChar32 c = codePointFromValidUTF8(cpStart, cpLimit) + delta; + length = 0; + U8_APPEND_UNSAFE(buffer, length, c); + } + } + if (edits != nullptr) { + edits->addReplace(cpLength, length); + } + sink.Append(buffer, length); +} + +} // namespace + +// ReorderingBuffer -------------------------------------------------------- *** + +ReorderingBuffer::ReorderingBuffer(const Normalizer2Impl &ni, UnicodeString &dest, + UErrorCode &errorCode) : + impl(ni), str(dest), + start(str.getBuffer(8)), reorderStart(start), limit(start), + remainingCapacity(str.getCapacity()), lastCC(0) { + if (start == nullptr && U_SUCCESS(errorCode)) { + // getBuffer() already did str.setToBogus() + errorCode = U_MEMORY_ALLOCATION_ERROR; + } +} + +UBool ReorderingBuffer::init(int32_t destCapacity, UErrorCode &errorCode) { + int32_t length=str.length(); + start=str.getBuffer(destCapacity); + if(start==nullptr) { + // getBuffer() already did str.setToBogus() + errorCode=U_MEMORY_ALLOCATION_ERROR; + return false; + } + limit=start+length; + remainingCapacity=str.getCapacity()-length; + reorderStart=start; + if(start==limit) { + lastCC=0; + } else { + setIterator(); + lastCC=previousCC(); + // Set reorderStart after the last code point with cc<=1 if there is one. + if(lastCC>1) { + while(previousCC()>1) {} + } + reorderStart=codePointLimit; + } + return true; +} + +UBool ReorderingBuffer::equals(const char16_t *otherStart, const char16_t *otherLimit) const { + int32_t length=(int32_t)(limit-start); + return + length==(int32_t)(otherLimit-otherStart) && + 0==u_memcmp(start, otherStart, length); +} + +UBool ReorderingBuffer::equals(const uint8_t *otherStart, const uint8_t *otherLimit) const { + U_ASSERT((otherLimit - otherStart) <= INT32_MAX); // ensured by caller + int32_t length = (int32_t)(limit - start); + int32_t otherLength = (int32_t)(otherLimit - otherStart); + // For equal strings, UTF-8 is at least as long as UTF-16, and at most three times as long. + if (otherLength < length || (otherLength / 3) > length) { + return false; + } + // Compare valid strings from between normalization boundaries. + // (Invalid sequences are normalization-inert.) + for (int32_t i = 0, j = 0;;) { + if (i >= length) { + return j >= otherLength; + } else if (j >= otherLength) { + return false; + } + // Not at the end of either string yet. + UChar32 c, other; + U16_NEXT_UNSAFE(start, i, c); + U8_NEXT_UNSAFE(otherStart, j, other); + if (c != other) { + return false; + } + } +} + +UBool ReorderingBuffer::appendSupplementary(UChar32 c, uint8_t cc, UErrorCode &errorCode) { + if(remainingCapacity<2 && !resize(2, errorCode)) { + return false; + } + if(lastCC<=cc || cc==0) { + limit[0]=U16_LEAD(c); + limit[1]=U16_TRAIL(c); + limit+=2; + lastCC=cc; + if(cc<=1) { + reorderStart=limit; + } + } else { + insert(c, cc); + } + remainingCapacity-=2; + return true; +} + +UBool ReorderingBuffer::append(const char16_t *s, int32_t length, UBool isNFD, + uint8_t leadCC, uint8_t trailCC, + UErrorCode &errorCode) { + if(length==0) { + return true; + } + if(remainingCapacity=codePointStart) { + return 0; + } + UChar32 c=*--codePointStart; + char16_t c2; + if(U16_IS_TRAIL(c) && startcc;) {} + // insert c at codePointLimit, after the character with prevCC<=cc + char16_t *q=limit; + char16_t *r=limit+=U16_LENGTH(c); + do { + *--r=*--q; + } while(codePointLimit!=q); + writeCodePoint(q, c); + if(cc<=1) { + reorderStart=r; + } +} + +// Normalizer2Impl --------------------------------------------------------- *** + +struct CanonIterData : public UMemory { + CanonIterData(UErrorCode &errorCode); + ~CanonIterData(); + void addToStartSet(UChar32 origin, UChar32 decompLead, UErrorCode &errorCode); + UMutableCPTrie *mutableTrie; + UCPTrie *trie; + UVector canonStartSets; // contains UnicodeSet * +}; + +Normalizer2Impl::~Normalizer2Impl() { + delete fCanonIterData; +} + +void +Normalizer2Impl::init(const int32_t *inIndexes, const UCPTrie *inTrie, + const uint16_t *inExtraData, const uint8_t *inSmallFCD) { + minDecompNoCP = static_cast(inIndexes[IX_MIN_DECOMP_NO_CP]); + minCompNoMaybeCP = static_cast(inIndexes[IX_MIN_COMP_NO_MAYBE_CP]); + minLcccCP = static_cast(inIndexes[IX_MIN_LCCC_CP]); + + minYesNo = static_cast(inIndexes[IX_MIN_YES_NO]); + minYesNoMappingsOnly = static_cast(inIndexes[IX_MIN_YES_NO_MAPPINGS_ONLY]); + minNoNo = static_cast(inIndexes[IX_MIN_NO_NO]); + minNoNoCompBoundaryBefore = static_cast(inIndexes[IX_MIN_NO_NO_COMP_BOUNDARY_BEFORE]); + minNoNoCompNoMaybeCC = static_cast(inIndexes[IX_MIN_NO_NO_COMP_NO_MAYBE_CC]); + minNoNoEmpty = static_cast(inIndexes[IX_MIN_NO_NO_EMPTY]); + limitNoNo = static_cast(inIndexes[IX_LIMIT_NO_NO]); + minMaybeYes = static_cast(inIndexes[IX_MIN_MAYBE_YES]); + U_ASSERT((minMaybeYes & 7) == 0); // 8-aligned for noNoDelta bit fields + centerNoNoDelta = (minMaybeYes >> DELTA_SHIFT) - MAX_DELTA - 1; + + normTrie=inTrie; + + maybeYesCompositions=inExtraData; + extraData=maybeYesCompositions+((MIN_NORMAL_MAYBE_YES-minMaybeYes)>>OFFSET_SHIFT); + + smallFCD=inSmallFCD; +} + +U_CDECL_BEGIN + +static uint32_t U_CALLCONV +segmentStarterMapper(const void * /*context*/, uint32_t value) { + return value&CANON_NOT_SEGMENT_STARTER; +} + +U_CDECL_END + +void +Normalizer2Impl::addLcccChars(UnicodeSet &set) const { + UChar32 start = 0, end; + uint32_t norm16; + while ((end = ucptrie_getRange(normTrie, start, UCPMAP_RANGE_FIXED_LEAD_SURROGATES, INERT, + nullptr, nullptr, &norm16)) >= 0) { + if (norm16 > Normalizer2Impl::MIN_NORMAL_MAYBE_YES && + norm16 != Normalizer2Impl::JAMO_VT) { + set.add(start, end); + } else if (minNoNoCompNoMaybeCC <= norm16 && norm16 < limitNoNo) { + uint16_t fcd16 = getFCD16(start); + if (fcd16 > 0xff) { set.add(start, end); } + } + start = end + 1; + } +} + +void +Normalizer2Impl::addPropertyStarts(const USetAdder *sa, UErrorCode & /*errorCode*/) const { + // Add the start code point of each same-value range of the trie. + UChar32 start = 0, end; + uint32_t value; + while ((end = ucptrie_getRange(normTrie, start, UCPMAP_RANGE_FIXED_LEAD_SURROGATES, INERT, + nullptr, nullptr, &value)) >= 0) { + sa->add(sa->set, start); + if (start != end && isAlgorithmicNoNo((uint16_t)value) && + (value & Normalizer2Impl::DELTA_TCCC_MASK) > Normalizer2Impl::DELTA_TCCC_1) { + // Range of code points with same-norm16-value algorithmic decompositions. + // They might have different non-zero FCD16 values. + uint16_t prevFCD16 = getFCD16(start); + while (++start <= end) { + uint16_t fcd16 = getFCD16(start); + if (fcd16 != prevFCD16) { + sa->add(sa->set, start); + prevFCD16 = fcd16; + } + } + } + start = end + 1; + } + + /* add Hangul LV syllables and LV+1 because of skippables */ + for(char16_t c=Hangul::HANGUL_BASE; cadd(sa->set, c); + sa->add(sa->set, c+1); + } + sa->add(sa->set, Hangul::HANGUL_LIMIT); /* add Hangul+1 to continue with other properties */ +} + +void +Normalizer2Impl::addCanonIterPropertyStarts(const USetAdder *sa, UErrorCode &errorCode) const { + // Add the start code point of each same-value range of the canonical iterator data trie. + if (!ensureCanonIterData(errorCode)) { return; } + // Currently only used for the SEGMENT_STARTER property. + UChar32 start = 0, end; + uint32_t value; + while ((end = ucptrie_getRange(fCanonIterData->trie, start, UCPMAP_RANGE_NORMAL, 0, + segmentStarterMapper, nullptr, &value)) >= 0) { + sa->add(sa->set, start); + start = end + 1; + } +} + +const char16_t * +Normalizer2Impl::copyLowPrefixFromNulTerminated(const char16_t *src, + UChar32 minNeedDataCP, + ReorderingBuffer *buffer, + UErrorCode &errorCode) const { + // Make some effort to support NUL-terminated strings reasonably. + // Take the part of the fast quick check loop that does not look up + // data and check the first part of the string. + // After this prefix, determine the string length to simplify the rest + // of the code. + const char16_t *prevSrc=src; + char16_t c; + while((c=*src++)appendZeroCC(prevSrc, src, errorCode); + } + } + return src; +} + +UnicodeString & +Normalizer2Impl::decompose(const UnicodeString &src, UnicodeString &dest, + UErrorCode &errorCode) const { + if(U_FAILURE(errorCode)) { + dest.setToBogus(); + return dest; + } + const char16_t *sArray=src.getBuffer(); + if(&dest==&src || sArray==nullptr) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + dest.setToBogus(); + return dest; + } + decompose(sArray, sArray+src.length(), dest, src.length(), errorCode); + return dest; +} + +void +Normalizer2Impl::decompose(const char16_t *src, const char16_t *limit, + UnicodeString &dest, + int32_t destLengthEstimate, + UErrorCode &errorCode) const { + if(destLengthEstimate<0 && limit!=nullptr) { + destLengthEstimate=(int32_t)(limit-src); + } + dest.remove(); + ReorderingBuffer buffer(*this, dest); + if(buffer.init(destLengthEstimate, errorCode)) { + decompose(src, limit, &buffer, errorCode); + } +} + +// Dual functionality: +// buffer!=nullptr: normalize +// buffer==nullptr: isNormalized/spanQuickCheckYes +const char16_t * +Normalizer2Impl::decompose(const char16_t *src, const char16_t *limit, + ReorderingBuffer *buffer, + UErrorCode &errorCode) const { + UChar32 minNoCP=minDecompNoCP; + if(limit==nullptr) { + src=copyLowPrefixFromNulTerminated(src, minNoCP, buffer, errorCode); + if(U_FAILURE(errorCode)) { + return src; + } + limit=u_strchr(src, 0); + } + + const char16_t *prevSrc; + UChar32 c=0; + uint16_t norm16=0; + + // only for quick check + const char16_t *prevBoundary=src; + uint8_t prevCC=0; + + for(;;) { + // count code units below the minimum or with irrelevant data for the quick check + for(prevSrc=src; src!=limit;) { + if( (c=*src)appendZeroCC(prevSrc, src, errorCode)) { + break; + } + } else { + prevCC=0; + prevBoundary=src; + } + } + if(src==limit) { + break; + } + + // Check one above-minimum, relevant code point. + src+=U16_LENGTH(c); + if(buffer!=nullptr) { + if(!decompose(c, norm16, *buffer, errorCode)) { + break; + } + } else { + if(isDecompYes(norm16)) { + uint8_t cc=getCCFromYesOrMaybe(norm16); + if(prevCC<=cc || cc==0) { + prevCC=cc; + if(cc<=1) { + prevBoundary=src; + } + continue; + } + } + return prevBoundary; // "no" or cc out of order + } + } + return src; +} + +// Decompose a short piece of text which is likely to contain characters that +// fail the quick check loop and/or where the quick check loop's overhead +// is unlikely to be amortized. +// Called by the compose() and makeFCD() implementations. +const char16_t * +Normalizer2Impl::decomposeShort(const char16_t *src, const char16_t *limit, + UBool stopAtCompBoundary, UBool onlyContiguous, + ReorderingBuffer &buffer, UErrorCode &errorCode) const { + if (U_FAILURE(errorCode)) { + return nullptr; + } + while(src= limitNoNo) { + if (isMaybeOrNonZeroCC(norm16)) { + return buffer.append(c, getCCFromYesOrMaybe(norm16), errorCode); + } + // Maps to an isCompYesAndZeroCC. + c=mapAlgorithmic(c, norm16); + norm16=getRawNorm16(c); + } + if (norm16 < minYesNo) { + // c does not decompose + return buffer.append(c, 0, errorCode); + } else if(isHangulLV(norm16) || isHangulLVT(norm16)) { + // Hangul syllable: decompose algorithmically + char16_t jamos[3]; + return buffer.appendZeroCC(jamos, jamos+Hangul::decompose(c, jamos), errorCode); + } + // c decomposes, get everything from the variable-length extra data + const uint16_t *mapping=getMapping(norm16); + uint16_t firstUnit=*mapping; + int32_t length=firstUnit&MAPPING_LENGTH_MASK; + uint8_t leadCC, trailCC; + trailCC=(uint8_t)(firstUnit>>8); + if(firstUnit&MAPPING_HAS_CCC_LCCC_WORD) { + leadCC=(uint8_t)(*(mapping-1)>>8); + } else { + leadCC=0; + } + return buffer.append((const char16_t *)mapping+1, length, true, leadCC, trailCC, errorCode); +} + +// Dual functionality: +// sink != nullptr: normalize +// sink == nullptr: isNormalized/spanQuickCheckYes +const uint8_t * +Normalizer2Impl::decomposeUTF8(uint32_t options, + const uint8_t *src, const uint8_t *limit, + ByteSink *sink, Edits *edits, UErrorCode &errorCode) const { + U_ASSERT(limit != nullptr); + UnicodeString s16; + uint8_t minNoLead = leadByteForCP(minDecompNoCP); + + const uint8_t *prevBoundary = src; + // only for quick check + uint8_t prevCC = 0; + + for (;;) { + // Fast path: Scan over a sequence of characters below the minimum "no" code point, + // or with (decompYes && ccc==0) properties. + const uint8_t *fastStart = src; + const uint8_t *prevSrc; + uint16_t norm16 = 0; + + for (;;) { + if (src == limit) { + if (prevBoundary != limit && sink != nullptr) { + ByteSinkUtil::appendUnchanged(prevBoundary, limit, + *sink, options, edits, errorCode); + } + return src; + } + if (*src < minNoLead) { + ++src; + } else { + prevSrc = src; + UCPTRIE_FAST_U8_NEXT(normTrie, UCPTRIE_16, src, limit, norm16); + if (!isMostDecompYesAndZeroCC(norm16)) { + break; + } + } + } + // isMostDecompYesAndZeroCC(norm16) is false, that is, norm16>=minYesNo, + // and the current character at [prevSrc..src[ is not a common case with cc=0 + // (MIN_NORMAL_MAYBE_YES or JAMO_VT). + // It could still be a maybeYes with cc=0. + if (prevSrc != fastStart) { + // The fast path looped over yes/0 characters before the current one. + if (sink != nullptr && + !ByteSinkUtil::appendUnchanged(prevBoundary, prevSrc, + *sink, options, edits, errorCode)) { + break; + } + prevBoundary = prevSrc; + prevCC = 0; + } + + // Medium-fast path: Quick check. + if (isMaybeOrNonZeroCC(norm16)) { + // Does not decompose. + uint8_t cc = getCCFromYesOrMaybe(norm16); + if (prevCC <= cc || cc == 0) { + prevCC = cc; + if (cc <= 1) { + if (sink != nullptr && + !ByteSinkUtil::appendUnchanged(prevBoundary, src, + *sink, options, edits, errorCode)) { + break; + } + prevBoundary = src; + } + continue; + } + } + if (sink == nullptr) { + return prevBoundary; // quick check: "no" or cc out of order + } + + // Slow path + // Decompose up to and including the current character. + if (prevBoundary != prevSrc && norm16HasDecompBoundaryBefore(norm16)) { + if (!ByteSinkUtil::appendUnchanged(prevBoundary, prevSrc, + *sink, options, edits, errorCode)) { + break; + } + prevBoundary = prevSrc; + } + ReorderingBuffer buffer(*this, s16, errorCode); + if (U_FAILURE(errorCode)) { + break; + } + decomposeShort(prevBoundary, src, STOP_AT_LIMIT, false /* onlyContiguous */, + buffer, errorCode); + // Decompose until the next boundary. + if (buffer.getLastCC() > 1) { + src = decomposeShort(src, limit, STOP_AT_DECOMP_BOUNDARY, false /* onlyContiguous */, + buffer, errorCode); + } + if (U_FAILURE(errorCode)) { + break; + } + if ((src - prevSrc) > INT32_MAX) { // guard before buffer.equals() + errorCode = U_INDEX_OUTOFBOUNDS_ERROR; + break; + } + // We already know there was a change if the original character decomposed; + // otherwise compare. + if (isMaybeOrNonZeroCC(norm16) && buffer.equals(prevBoundary, src)) { + if (!ByteSinkUtil::appendUnchanged(prevBoundary, src, + *sink, options, edits, errorCode)) { + break; + } + } else { + if (!ByteSinkUtil::appendChange(prevBoundary, src, buffer.getStart(), buffer.length(), + *sink, edits, errorCode)) { + break; + } + } + prevBoundary = src; + prevCC = 0; + } + return src; +} + +const uint8_t * +Normalizer2Impl::decomposeShort(const uint8_t *src, const uint8_t *limit, + StopAt stopAt, UBool onlyContiguous, + ReorderingBuffer &buffer, UErrorCode &errorCode) const { + if (U_FAILURE(errorCode)) { + return nullptr; + } + while (src < limit) { + const uint8_t *prevSrc = src; + uint16_t norm16; + UCPTRIE_FAST_U8_NEXT(normTrie, UCPTRIE_16, src, limit, norm16); + // Get the decomposition and the lead and trail cc's. + UChar32 c = U_SENTINEL; + if (norm16 >= limitNoNo) { + if (isMaybeOrNonZeroCC(norm16)) { + // No comp boundaries around this character. + uint8_t cc = getCCFromYesOrMaybe(norm16); + if (cc == 0 && stopAt == STOP_AT_DECOMP_BOUNDARY) { + return prevSrc; + } + c = codePointFromValidUTF8(prevSrc, src); + if (!buffer.append(c, cc, errorCode)) { + return nullptr; + } + if (stopAt == STOP_AT_DECOMP_BOUNDARY && buffer.getLastCC() <= 1) { + return src; + } + continue; + } + // Maps to an isCompYesAndZeroCC. + if (stopAt != STOP_AT_LIMIT) { + return prevSrc; + } + c = codePointFromValidUTF8(prevSrc, src); + c = mapAlgorithmic(c, norm16); + norm16 = getRawNorm16(c); + } else if (stopAt != STOP_AT_LIMIT && norm16 < minNoNoCompNoMaybeCC) { + return prevSrc; + } + // norm16!=INERT guarantees that [prevSrc, src[ is valid UTF-8. + // We do not see invalid UTF-8 here because + // its norm16==INERT is normalization-inert, + // so it gets copied unchanged in the fast path, + // and we stop the slow path where invalid UTF-8 begins. + // c >= 0 is the result of an algorithmic mapping. + U_ASSERT(c >= 0 || norm16 != INERT); + if (norm16 < minYesNo) { + if (c < 0) { + c = codePointFromValidUTF8(prevSrc, src); + } + // does not decompose + if (!buffer.append(c, 0, errorCode)) { + return nullptr; + } + } else if (isHangulLV(norm16) || isHangulLVT(norm16)) { + // Hangul syllable: decompose algorithmically + if (c < 0) { + c = codePointFromValidUTF8(prevSrc, src); + } + char16_t jamos[3]; + if (!buffer.appendZeroCC(jamos, jamos+Hangul::decompose(c, jamos), errorCode)) { + return nullptr; + } + } else { + // The character decomposes, get everything from the variable-length extra data. + const uint16_t *mapping = getMapping(norm16); + uint16_t firstUnit = *mapping; + int32_t length = firstUnit & MAPPING_LENGTH_MASK; + uint8_t trailCC = (uint8_t)(firstUnit >> 8); + uint8_t leadCC; + if (firstUnit & MAPPING_HAS_CCC_LCCC_WORD) { + leadCC = (uint8_t)(*(mapping-1) >> 8); + } else { + leadCC = 0; + } + if (leadCC == 0 && stopAt == STOP_AT_DECOMP_BOUNDARY) { + return prevSrc; + } + if (!buffer.append((const char16_t *)mapping+1, length, true, leadCC, trailCC, errorCode)) { + return nullptr; + } + } + if ((stopAt == STOP_AT_COMP_BOUNDARY && norm16HasCompBoundaryAfter(norm16, onlyContiguous)) || + (stopAt == STOP_AT_DECOMP_BOUNDARY && buffer.getLastCC() <= 1)) { + return src; + } + } + return src; +} + +const char16_t * +Normalizer2Impl::getDecomposition(UChar32 c, char16_t buffer[4], int32_t &length) const { + uint16_t norm16; + if(c>7)&1)-1; + uint16_t rm0=*rawMapping; + if(rm0<=MAPPING_LENGTH_MASK) { + length=rm0; + return (const char16_t *)rawMapping-rm0; + } else { + // Copy the normal mapping and replace its first two code units with rm0. + buffer[0]=(char16_t)rm0; + u_memcpy(buffer+1, (const char16_t *)mapping+1+2, mLength-2); + length=mLength-1; + return buffer; + } + } else { + length=mLength; + return (const char16_t *)mapping+1; + } +} + +void Normalizer2Impl::decomposeAndAppend(const char16_t *src, const char16_t *limit, + UBool doDecompose, + UnicodeString &safeMiddle, + ReorderingBuffer &buffer, + UErrorCode &errorCode) const { + buffer.copyReorderableSuffixTo(safeMiddle); + if(doDecompose) { + decompose(src, limit, &buffer, errorCode); + return; + } + // Just merge the strings at the boundary. + bool isFirst = true; + uint8_t firstCC = 0, prevCC = 0, cc; + const char16_t *p = src; + while (p != limit) { + const char16_t *codePointStart = p; + UChar32 c; + uint16_t norm16; + UCPTRIE_FAST_U16_NEXT(normTrie, UCPTRIE_16, p, limit, c, norm16); + if ((cc = getCC(norm16)) == 0) { + p = codePointStart; + break; + } + if (isFirst) { + firstCC = cc; + isFirst = false; + } + prevCC = cc; + } + if(limit==nullptr) { // appendZeroCC() needs limit!=nullptr + limit=u_strchr(p, 0); + } + + if (buffer.append(src, (int32_t)(p - src), false, firstCC, prevCC, errorCode)) { + buffer.appendZeroCC(p, limit, errorCode); + } +} + +UBool Normalizer2Impl::hasDecompBoundaryBefore(UChar32 c) const { + return c < minLcccCP || (c <= 0xffff && !singleLeadMightHaveNonZeroFCD16(c)) || + norm16HasDecompBoundaryBefore(getNorm16(c)); +} + +UBool Normalizer2Impl::norm16HasDecompBoundaryBefore(uint16_t norm16) const { + if (norm16 < minNoNoCompNoMaybeCC) { + return true; + } + if (norm16 >= limitNoNo) { + return norm16 <= MIN_NORMAL_MAYBE_YES || norm16 == JAMO_VT; + } + // c decomposes, get everything from the variable-length extra data + const uint16_t *mapping=getMapping(norm16); + uint16_t firstUnit=*mapping; + // true if leadCC==0 (hasFCDBoundaryBefore()) + return (firstUnit&MAPPING_HAS_CCC_LCCC_WORD)==0 || (*(mapping-1)&0xff00)==0; +} + +UBool Normalizer2Impl::hasDecompBoundaryAfter(UChar32 c) const { + if (c < minDecompNoCP) { + return true; + } + if (c <= 0xffff && !singleLeadMightHaveNonZeroFCD16(c)) { + return true; + } + return norm16HasDecompBoundaryAfter(getNorm16(c)); +} + +UBool Normalizer2Impl::norm16HasDecompBoundaryAfter(uint16_t norm16) const { + if(norm16 <= minYesNo || isHangulLVT(norm16)) { + return true; + } + if (norm16 >= limitNoNo) { + if (isMaybeOrNonZeroCC(norm16)) { + return norm16 <= MIN_NORMAL_MAYBE_YES || norm16 == JAMO_VT; + } + // Maps to an isCompYesAndZeroCC. + return (norm16 & DELTA_TCCC_MASK) <= DELTA_TCCC_1; + } + // c decomposes, get everything from the variable-length extra data + const uint16_t *mapping=getMapping(norm16); + uint16_t firstUnit=*mapping; + // decomp after-boundary: same as hasFCDBoundaryAfter(), + // fcd16<=1 || trailCC==0 + if(firstUnit>0x1ff) { + return false; // trailCC>1 + } + if(firstUnit<=0xff) { + return true; // trailCC==0 + } + // if(trailCC==1) test leadCC==0, same as checking for before-boundary + // true if leadCC==0 (hasFCDBoundaryBefore()) + return (firstUnit&MAPPING_HAS_CCC_LCCC_WORD)==0 || (*(mapping-1)&0xff00)==0; +} + +/* + * Finds the recomposition result for + * a forward-combining "lead" character, + * specified with a pointer to its compositions list, + * and a backward-combining "trail" character. + * + * If the lead and trail characters combine, then this function returns + * the following "compositeAndFwd" value: + * Bits 21..1 composite character + * Bit 0 set if the composite is a forward-combining starter + * otherwise it returns -1. + * + * The compositions list has (trail, compositeAndFwd) pair entries, + * encoded as either pairs or triples of 16-bit units. + * The last entry has the high bit of its first unit set. + * + * The list is sorted by ascending trail characters (there are no duplicates). + * A linear search is used. + * + * See normalizer2impl.h for a more detailed description + * of the compositions list format. + */ +int32_t Normalizer2Impl::combine(const uint16_t *list, UChar32 trail) { + uint16_t key1, firstUnit; + if(trail(firstUnit=*list)) { + list+=2+(firstUnit&COMP_1_TRIPLE); + } + if(key1==(firstUnit&COMP_1_TRAIL_MASK)) { + if(firstUnit&COMP_1_TRIPLE) { + return ((int32_t)list[1]<<16)|list[2]; + } else { + return list[1]; + } + } + } else { + // trail character is 3400..10FFFF + // result entry has 3 units + key1=(uint16_t)(COMP_1_TRAIL_LIMIT+ + (((trail>>COMP_1_TRAIL_SHIFT))& + ~COMP_1_TRIPLE)); + uint16_t key2=(uint16_t)(trail<(firstUnit=*list)) { + list+=2+(firstUnit&COMP_1_TRIPLE); + } else if(key1==(firstUnit&COMP_1_TRAIL_MASK)) { + if(key2>(secondUnit=list[1])) { + if(firstUnit&COMP_1_LAST_TUPLE) { + break; + } else { + list+=3; + } + } else if(key2==(secondUnit&COMP_2_TRAIL_MASK)) { + return ((int32_t)(secondUnit&~COMP_2_TRAIL_MASK)<<16)|list[2]; + } else { + break; + } + } else { + break; + } + } + } + return -1; +} + +/** + * @param list some character's compositions list + * @param set recursively receives the composites from these compositions + */ +void Normalizer2Impl::addComposites(const uint16_t *list, UnicodeSet &set) const { + uint16_t firstUnit; + int32_t compositeAndFwd; + do { + firstUnit=*list; + if((firstUnit&COMP_1_TRIPLE)==0) { + compositeAndFwd=list[1]; + list+=2; + } else { + compositeAndFwd=(((int32_t)list[1]&~COMP_2_TRAIL_MASK)<<16)|list[2]; + list+=3; + } + UChar32 composite=compositeAndFwd>>1; + if((compositeAndFwd&1)!=0) { + addComposites(getCompositionsListForComposite(getRawNorm16(composite)), set); + } + set.add(composite); + } while((firstUnit&COMP_1_LAST_TUPLE)==0); +} + +/* + * Recomposes the buffer text starting at recomposeStartIndex + * (which is in NFD - decomposed and canonically ordered), + * and truncates the buffer contents. + * + * Note that recomposition never lengthens the text: + * Any character consists of either one or two code units; + * a composition may contain at most one more code unit than the original starter, + * while the combining mark that is removed has at least one code unit. + */ +void Normalizer2Impl::recompose(ReorderingBuffer &buffer, int32_t recomposeStartIndex, + UBool onlyContiguous) const { + char16_t *p=buffer.getStart()+recomposeStartIndex; + char16_t *limit=buffer.getLimit(); + if(p==limit) { + return; + } + + char16_t *starter, *pRemove, *q, *r; + const uint16_t *compositionsList; + UChar32 c, compositeAndFwd; + uint16_t norm16; + uint8_t cc, prevCC; + UBool starterIsSupplementary; + + // Some of the following variables are not used until we have a forward-combining starter + // and are only initialized now to avoid compiler warnings. + compositionsList=nullptr; // used as indicator for whether we have a forward-combining starter + starter=nullptr; + starterIsSupplementary=false; + prevCC=0; + + for(;;) { + UCPTRIE_FAST_U16_NEXT(normTrie, UCPTRIE_16, p, limit, c, norm16); + cc=getCCFromYesOrMaybe(norm16); + if( // this character combines backward and + isMaybe(norm16) && + // we have seen a starter that combines forward and + compositionsList!=nullptr && + // the backward-combining character is not blocked + (prevCC=0) { + // The starter and the combining mark (c) do combine. + UChar32 composite=compositeAndFwd>>1; + + // Replace the starter with the composite, remove the combining mark. + pRemove=p-U16_LENGTH(c); // pRemove & p: start & limit of the combining mark + if(starterIsSupplementary) { + if(U_IS_SUPPLEMENTARY(composite)) { + // both are supplementary + starter[0]=U16_LEAD(composite); + starter[1]=U16_TRAIL(composite); + } else { + *starter=(char16_t)composite; + // The composite is shorter than the starter, + // move the intermediate characters forward one. + starterIsSupplementary=false; + q=starter+1; + r=q+1; + while(rminYesNo) { // composite 'a' has both mapping & compositions list + list+= // mapping pointer + 1+ // +1 to skip the first unit with the mapping length + (*list&MAPPING_LENGTH_MASK); // + mapping length + } + } + } else if(norm16>1; +#else + int32_t compositeAndFwd=combine(list, b); + return compositeAndFwd>=0 ? compositeAndFwd>>1 : U_SENTINEL; +#endif +} + +// Very similar to composeQuickCheck(): Make the same changes in both places if relevant. +// doCompose: normalize +// !doCompose: isNormalized (buffer must be empty and initialized) +UBool +Normalizer2Impl::compose(const char16_t *src, const char16_t *limit, + UBool onlyContiguous, + UBool doCompose, + ReorderingBuffer &buffer, + UErrorCode &errorCode) const { + const char16_t *prevBoundary=src; + UChar32 minNoMaybeCP=minCompNoMaybeCP; + if(limit==nullptr) { + src=copyLowPrefixFromNulTerminated(src, minNoMaybeCP, + doCompose ? &buffer : nullptr, + errorCode); + if(U_FAILURE(errorCode)) { + return false; + } + limit=u_strchr(src, 0); + if (prevBoundary != src) { + if (hasCompBoundaryAfter(*(src-1), onlyContiguous)) { + prevBoundary = src; + } else { + buffer.removeSuffix(1); + prevBoundary = --src; + } + } + } + + for (;;) { + // Fast path: Scan over a sequence of characters below the minimum "no or maybe" code point, + // or with (compYes && ccc==0) properties. + const char16_t *prevSrc; + UChar32 c = 0; + uint16_t norm16 = 0; + for (;;) { + if (src == limit) { + if (prevBoundary != limit && doCompose) { + buffer.appendZeroCC(prevBoundary, limit, errorCode); + } + return true; + } + if( (c=*src)=minNoNo. + // The current character is either a "noNo" (has a mapping) + // or a "maybeYes" (combines backward) + // or a "yesYes" with ccc!=0. + // It is not a Hangul syllable or Jamo L because those have "yes" properties. + + // Medium-fast path: Handle cases that do not require full decomposition and recomposition. + if (!isMaybeOrNonZeroCC(norm16)) { // minNoNo <= norm16 < minMaybeYes + if (!doCompose) { + return false; + } + // Fast path for mapping a character that is immediately surrounded by boundaries. + // In this case, we need not decompose around the current character. + if (isDecompNoAlgorithmic(norm16)) { + // Maps to a single isCompYesAndZeroCC character + // which also implies hasCompBoundaryBefore. + if (norm16HasCompBoundaryAfter(norm16, onlyContiguous) || + hasCompBoundaryBefore(src, limit)) { + if (prevBoundary != prevSrc && !buffer.appendZeroCC(prevBoundary, prevSrc, errorCode)) { + break; + } + if(!buffer.append(mapAlgorithmic(c, norm16), 0, errorCode)) { + break; + } + prevBoundary = src; + continue; + } + } else if (norm16 < minNoNoCompBoundaryBefore) { + // The mapping is comp-normalized which also implies hasCompBoundaryBefore. + if (norm16HasCompBoundaryAfter(norm16, onlyContiguous) || + hasCompBoundaryBefore(src, limit)) { + if (prevBoundary != prevSrc && !buffer.appendZeroCC(prevBoundary, prevSrc, errorCode)) { + break; + } + const char16_t *mapping = reinterpret_cast(getMapping(norm16)); + int32_t length = *mapping++ & MAPPING_LENGTH_MASK; + if(!buffer.appendZeroCC(mapping, mapping + length, errorCode)) { + break; + } + prevBoundary = src; + continue; + } + } else if (norm16 >= minNoNoEmpty) { + // The current character maps to nothing. + // Simply omit it from the output if there is a boundary before _or_ after it. + // The character itself implies no boundaries. + if (hasCompBoundaryBefore(src, limit) || + hasCompBoundaryAfter(prevBoundary, prevSrc, onlyContiguous)) { + if (prevBoundary != prevSrc && !buffer.appendZeroCC(prevBoundary, prevSrc, errorCode)) { + break; + } + prevBoundary = src; + continue; + } + } + // Other "noNo" type, or need to examine more text around this character: + // Fall through to the slow path. + } else if (isJamoVT(norm16) && prevBoundary != prevSrc) { + char16_t prev=*(prevSrc-1); + if(c= 0) { + UChar32 syllable = Hangul::HANGUL_BASE + + (l*Hangul::JAMO_V_COUNT + (c-Hangul::JAMO_V_BASE)) * + Hangul::JAMO_T_COUNT + t; + --prevSrc; // Replace the Jamo L as well. + if (prevBoundary != prevSrc && !buffer.appendZeroCC(prevBoundary, prevSrc, errorCode)) { + break; + } + if(!buffer.appendBMP((char16_t)syllable, 0, errorCode)) { + break; + } + prevBoundary = src; + continue; + } + // If we see L+V+x where x!=T then we drop to the slow path, + // decompose and recompose. + // This is to deal with NFKC finding normal L and V but a + // compatibility variant of a T. + // We need to either fully compose that combination here + // (which would complicate the code and may not work with strange custom data) + // or use the slow path. + } + } else if (Hangul::isHangulLV(prev)) { + // The current character is a Jamo Trailing consonant, + // compose with previous Hangul LV that does not contain a Jamo T. + if (!doCompose) { + return false; + } + UChar32 syllable = prev + c - Hangul::JAMO_T_BASE; + --prevSrc; // Replace the Hangul LV as well. + if (prevBoundary != prevSrc && !buffer.appendZeroCC(prevBoundary, prevSrc, errorCode)) { + break; + } + if(!buffer.appendBMP((char16_t)syllable, 0, errorCode)) { + break; + } + prevBoundary = src; + continue; + } + // No matching context, or may need to decompose surrounding text first: + // Fall through to the slow path. + } else if (norm16 > JAMO_VT) { // norm16 >= MIN_YES_YES_WITH_CC + // One or more combining marks that do not combine-back: + // Check for canonical order, copy unchanged if ok and + // if followed by a character with a boundary-before. + uint8_t cc = getCCFromNormalYesOrMaybe(norm16); // cc!=0 + if (onlyContiguous /* FCC */ && getPreviousTrailCC(prevBoundary, prevSrc) > cc) { + // Fails FCD test, need to decompose and contiguously recompose. + if (!doCompose) { + return false; + } + } else { + // If !onlyContiguous (not FCC), then we ignore the tccc of + // the previous character which passed the quick check "yes && ccc==0" test. + const char16_t *nextSrc; + uint16_t n16; + for (;;) { + if (src == limit) { + if (doCompose) { + buffer.appendZeroCC(prevBoundary, limit, errorCode); + } + return true; + } + uint8_t prevCC = cc; + nextSrc = src; + UCPTRIE_FAST_U16_NEXT(normTrie, UCPTRIE_16, nextSrc, limit, c, n16); + if (n16 >= MIN_YES_YES_WITH_CC) { + cc = getCCFromNormalYesOrMaybe(n16); + if (prevCC > cc) { + if (!doCompose) { + return false; + } + break; + } + } else { + break; + } + src = nextSrc; + } + // src is after the last in-order combining mark. + // If there is a boundary here, then we continue with no change. + if (norm16HasCompBoundaryBefore(n16)) { + if (isCompYesAndZeroCC(n16)) { + src = nextSrc; + } + continue; + } + // Use the slow path. There is no boundary in [prevSrc, src[. + } + } + + // Slow path: Find the nearest boundaries around the current character, + // decompose and recompose. + if (prevBoundary != prevSrc && !norm16HasCompBoundaryBefore(norm16)) { + const char16_t *p = prevSrc; + UCPTRIE_FAST_U16_PREV(normTrie, UCPTRIE_16, prevBoundary, p, c, norm16); + if (!norm16HasCompBoundaryAfter(norm16, onlyContiguous)) { + prevSrc = p; + } + } + if (doCompose && prevBoundary != prevSrc && !buffer.appendZeroCC(prevBoundary, prevSrc, errorCode)) { + break; + } + int32_t recomposeStartIndex=buffer.length(); + // We know there is not a boundary here. + decomposeShort(prevSrc, src, false /* !stopAtCompBoundary */, onlyContiguous, + buffer, errorCode); + // Decompose until the next boundary. + src = decomposeShort(src, limit, true /* stopAtCompBoundary */, onlyContiguous, + buffer, errorCode); + if (U_FAILURE(errorCode)) { + break; + } + if ((src - prevSrc) > INT32_MAX) { // guard before buffer.equals() + errorCode = U_INDEX_OUTOFBOUNDS_ERROR; + return true; + } + recompose(buffer, recomposeStartIndex, onlyContiguous); + if(!doCompose) { + if(!buffer.equals(prevSrc, src)) { + return false; + } + buffer.remove(); + } + prevBoundary=src; + } + return true; +} + +// Very similar to compose(): Make the same changes in both places if relevant. +// pQCResult==nullptr: spanQuickCheckYes +// pQCResult!=nullptr: quickCheck (*pQCResult must be UNORM_YES) +const char16_t * +Normalizer2Impl::composeQuickCheck(const char16_t *src, const char16_t *limit, + UBool onlyContiguous, + UNormalizationCheckResult *pQCResult) const { + const char16_t *prevBoundary=src; + UChar32 minNoMaybeCP=minCompNoMaybeCP; + if(limit==nullptr) { + UErrorCode errorCode=U_ZERO_ERROR; + src=copyLowPrefixFromNulTerminated(src, minNoMaybeCP, nullptr, errorCode); + limit=u_strchr(src, 0); + if (prevBoundary != src) { + if (hasCompBoundaryAfter(*(src-1), onlyContiguous)) { + prevBoundary = src; + } else { + prevBoundary = --src; + } + } + } + + for(;;) { + // Fast path: Scan over a sequence of characters below the minimum "no or maybe" code point, + // or with (compYes && ccc==0) properties. + const char16_t *prevSrc; + UChar32 c = 0; + uint16_t norm16 = 0; + for (;;) { + if(src==limit) { + return src; + } + if( (c=*src)=minNoNo. + // The current character is either a "noNo" (has a mapping) + // or a "maybeYes" (combines backward) + // or a "yesYes" with ccc!=0. + // It is not a Hangul syllable or Jamo L because those have "yes" properties. + + uint16_t prevNorm16 = INERT; + if (prevBoundary != prevSrc) { + if (norm16HasCompBoundaryBefore(norm16)) { + prevBoundary = prevSrc; + } else { + const char16_t *p = prevSrc; + uint16_t n16; + UCPTRIE_FAST_U16_PREV(normTrie, UCPTRIE_16, prevBoundary, p, c, n16); + if (norm16HasCompBoundaryAfter(n16, onlyContiguous)) { + prevBoundary = prevSrc; + } else { + prevBoundary = p; + prevNorm16 = n16; + } + } + } + + if(isMaybeOrNonZeroCC(norm16)) { + uint8_t cc=getCCFromYesOrMaybe(norm16); + if (onlyContiguous /* FCC */ && cc != 0 && + getTrailCCFromCompYesAndZeroCC(prevNorm16) > cc) { + // The [prevBoundary..prevSrc[ character + // passed the quick check "yes && ccc==0" test + // but is out of canonical order with the current combining mark. + } else { + // If !onlyContiguous (not FCC), then we ignore the tccc of + // the previous character which passed the quick check "yes && ccc==0" test. + const char16_t *nextSrc; + for (;;) { + if (norm16 < MIN_YES_YES_WITH_CC) { + if (pQCResult != nullptr) { + *pQCResult = UNORM_MAYBE; + } else { + return prevBoundary; + } + } + if (src == limit) { + return src; + } + uint8_t prevCC = cc; + nextSrc = src; + UCPTRIE_FAST_U16_NEXT(normTrie, UCPTRIE_16, nextSrc, limit, c, norm16); + if (isMaybeOrNonZeroCC(norm16)) { + cc = getCCFromYesOrMaybe(norm16); + if (!(prevCC <= cc || cc == 0)) { + break; + } + } else { + break; + } + src = nextSrc; + } + // src is after the last in-order combining mark. + if (isCompYesAndZeroCC(norm16)) { + prevBoundary = src; + src = nextSrc; + continue; + } + } + } + if(pQCResult!=nullptr) { + *pQCResult=UNORM_NO; + } + return prevBoundary; + } +} + +void Normalizer2Impl::composeAndAppend(const char16_t *src, const char16_t *limit, + UBool doCompose, + UBool onlyContiguous, + UnicodeString &safeMiddle, + ReorderingBuffer &buffer, + UErrorCode &errorCode) const { + if(!buffer.isEmpty()) { + const char16_t *firstStarterInSrc=findNextCompBoundary(src, limit, onlyContiguous); + if(src!=firstStarterInSrc) { + const char16_t *lastStarterInDest=findPreviousCompBoundary(buffer.getStart(), + buffer.getLimit(), onlyContiguous); + int32_t destSuffixLength=(int32_t)(buffer.getLimit()-lastStarterInDest); + UnicodeString middle(lastStarterInDest, destSuffixLength); + buffer.removeSuffix(destSuffixLength); + safeMiddle=middle; + middle.append(src, (int32_t)(firstStarterInSrc-src)); + const char16_t *middleStart=middle.getBuffer(); + compose(middleStart, middleStart+middle.length(), onlyContiguous, + true, buffer, errorCode); + if(U_FAILURE(errorCode)) { + return; + } + src=firstStarterInSrc; + } + } + if(doCompose) { + compose(src, limit, onlyContiguous, true, buffer, errorCode); + } else { + if(limit==nullptr) { // appendZeroCC() needs limit!=nullptr + limit=u_strchr(src, 0); + } + buffer.appendZeroCC(src, limit, errorCode); + } +} + +UBool +Normalizer2Impl::composeUTF8(uint32_t options, UBool onlyContiguous, + const uint8_t *src, const uint8_t *limit, + ByteSink *sink, Edits *edits, UErrorCode &errorCode) const { + U_ASSERT(limit != nullptr); + UnicodeString s16; + uint8_t minNoMaybeLead = leadByteForCP(minCompNoMaybeCP); + const uint8_t *prevBoundary = src; + + for (;;) { + // Fast path: Scan over a sequence of characters below the minimum "no or maybe" code point, + // or with (compYes && ccc==0) properties. + const uint8_t *prevSrc; + uint16_t norm16 = 0; + for (;;) { + if (src == limit) { + if (prevBoundary != limit && sink != nullptr) { + ByteSinkUtil::appendUnchanged(prevBoundary, limit, + *sink, options, edits, errorCode); + } + return true; + } + if (*src < minNoMaybeLead) { + ++src; + } else { + prevSrc = src; + UCPTRIE_FAST_U8_NEXT(normTrie, UCPTRIE_16, src, limit, norm16); + if (!isCompYesAndZeroCC(norm16)) { + break; + } + } + } + // isCompYesAndZeroCC(norm16) is false, that is, norm16>=minNoNo. + // The current character is either a "noNo" (has a mapping) + // or a "maybeYes" (combines backward) + // or a "yesYes" with ccc!=0. + // It is not a Hangul syllable or Jamo L because those have "yes" properties. + + // Medium-fast path: Handle cases that do not require full decomposition and recomposition. + if (!isMaybeOrNonZeroCC(norm16)) { // minNoNo <= norm16 < minMaybeYes + if (sink == nullptr) { + return false; + } + // Fast path for mapping a character that is immediately surrounded by boundaries. + // In this case, we need not decompose around the current character. + if (isDecompNoAlgorithmic(norm16)) { + // Maps to a single isCompYesAndZeroCC character + // which also implies hasCompBoundaryBefore. + if (norm16HasCompBoundaryAfter(norm16, onlyContiguous) || + hasCompBoundaryBefore(src, limit)) { + if (prevBoundary != prevSrc && + !ByteSinkUtil::appendUnchanged(prevBoundary, prevSrc, + *sink, options, edits, errorCode)) { + break; + } + appendCodePointDelta(prevSrc, src, getAlgorithmicDelta(norm16), *sink, edits); + prevBoundary = src; + continue; + } + } else if (norm16 < minNoNoCompBoundaryBefore) { + // The mapping is comp-normalized which also implies hasCompBoundaryBefore. + if (norm16HasCompBoundaryAfter(norm16, onlyContiguous) || + hasCompBoundaryBefore(src, limit)) { + if (prevBoundary != prevSrc && + !ByteSinkUtil::appendUnchanged(prevBoundary, prevSrc, + *sink, options, edits, errorCode)) { + break; + } + const uint16_t *mapping = getMapping(norm16); + int32_t length = *mapping++ & MAPPING_LENGTH_MASK; + if (!ByteSinkUtil::appendChange(prevSrc, src, (const char16_t *)mapping, length, + *sink, edits, errorCode)) { + break; + } + prevBoundary = src; + continue; + } + } else if (norm16 >= minNoNoEmpty) { + // The current character maps to nothing. + // Simply omit it from the output if there is a boundary before _or_ after it. + // The character itself implies no boundaries. + if (hasCompBoundaryBefore(src, limit) || + hasCompBoundaryAfter(prevBoundary, prevSrc, onlyContiguous)) { + if (prevBoundary != prevSrc && + !ByteSinkUtil::appendUnchanged(prevBoundary, prevSrc, + *sink, options, edits, errorCode)) { + break; + } + if (edits != nullptr) { + edits->addReplace((int32_t)(src - prevSrc), 0); + } + prevBoundary = src; + continue; + } + } + // Other "noNo" type, or need to examine more text around this character: + // Fall through to the slow path. + } else if (isJamoVT(norm16)) { + // Jamo L: E1 84 80..92 + // Jamo V: E1 85 A1..B5 + // Jamo T: E1 86 A8..E1 87 82 + U_ASSERT((src - prevSrc) == 3 && *prevSrc == 0xe1); + UChar32 prev = previousHangulOrJamo(prevBoundary, prevSrc); + if (prevSrc[1] == 0x85) { + // The current character is a Jamo Vowel, + // compose with previous Jamo L and following Jamo T. + UChar32 l = prev - Hangul::JAMO_L_BASE; + if ((uint32_t)l < Hangul::JAMO_L_COUNT) { + if (sink == nullptr) { + return false; + } + int32_t t = getJamoTMinusBase(src, limit); + if (t >= 0) { + // The next character is a Jamo T. + src += 3; + } else if (hasCompBoundaryBefore(src, limit)) { + // No Jamo T follows, not even via decomposition. + t = 0; + } + if (t >= 0) { + UChar32 syllable = Hangul::HANGUL_BASE + + (l*Hangul::JAMO_V_COUNT + (prevSrc[2]-0xa1)) * + Hangul::JAMO_T_COUNT + t; + prevSrc -= 3; // Replace the Jamo L as well. + if (prevBoundary != prevSrc && + !ByteSinkUtil::appendUnchanged(prevBoundary, prevSrc, + *sink, options, edits, errorCode)) { + break; + } + ByteSinkUtil::appendCodePoint(prevSrc, src, syllable, *sink, edits); + prevBoundary = src; + continue; + } + // If we see L+V+x where x!=T then we drop to the slow path, + // decompose and recompose. + // This is to deal with NFKC finding normal L and V but a + // compatibility variant of a T. + // We need to either fully compose that combination here + // (which would complicate the code and may not work with strange custom data) + // or use the slow path. + } + } else if (Hangul::isHangulLV(prev)) { + // The current character is a Jamo Trailing consonant, + // compose with previous Hangul LV that does not contain a Jamo T. + if (sink == nullptr) { + return false; + } + UChar32 syllable = prev + getJamoTMinusBase(prevSrc, src); + prevSrc -= 3; // Replace the Hangul LV as well. + if (prevBoundary != prevSrc && + !ByteSinkUtil::appendUnchanged(prevBoundary, prevSrc, + *sink, options, edits, errorCode)) { + break; + } + ByteSinkUtil::appendCodePoint(prevSrc, src, syllable, *sink, edits); + prevBoundary = src; + continue; + } + // No matching context, or may need to decompose surrounding text first: + // Fall through to the slow path. + } else if (norm16 > JAMO_VT) { // norm16 >= MIN_YES_YES_WITH_CC + // One or more combining marks that do not combine-back: + // Check for canonical order, copy unchanged if ok and + // if followed by a character with a boundary-before. + uint8_t cc = getCCFromNormalYesOrMaybe(norm16); // cc!=0 + if (onlyContiguous /* FCC */ && getPreviousTrailCC(prevBoundary, prevSrc) > cc) { + // Fails FCD test, need to decompose and contiguously recompose. + if (sink == nullptr) { + return false; + } + } else { + // If !onlyContiguous (not FCC), then we ignore the tccc of + // the previous character which passed the quick check "yes && ccc==0" test. + const uint8_t *nextSrc; + uint16_t n16; + for (;;) { + if (src == limit) { + if (sink != nullptr) { + ByteSinkUtil::appendUnchanged(prevBoundary, limit, + *sink, options, edits, errorCode); + } + return true; + } + uint8_t prevCC = cc; + nextSrc = src; + UCPTRIE_FAST_U8_NEXT(normTrie, UCPTRIE_16, nextSrc, limit, n16); + if (n16 >= MIN_YES_YES_WITH_CC) { + cc = getCCFromNormalYesOrMaybe(n16); + if (prevCC > cc) { + if (sink == nullptr) { + return false; + } + break; + } + } else { + break; + } + src = nextSrc; + } + // src is after the last in-order combining mark. + // If there is a boundary here, then we continue with no change. + if (norm16HasCompBoundaryBefore(n16)) { + if (isCompYesAndZeroCC(n16)) { + src = nextSrc; + } + continue; + } + // Use the slow path. There is no boundary in [prevSrc, src[. + } + } + + // Slow path: Find the nearest boundaries around the current character, + // decompose and recompose. + if (prevBoundary != prevSrc && !norm16HasCompBoundaryBefore(norm16)) { + const uint8_t *p = prevSrc; + UCPTRIE_FAST_U8_PREV(normTrie, UCPTRIE_16, prevBoundary, p, norm16); + if (!norm16HasCompBoundaryAfter(norm16, onlyContiguous)) { + prevSrc = p; + } + } + ReorderingBuffer buffer(*this, s16, errorCode); + if (U_FAILURE(errorCode)) { + break; + } + // We know there is not a boundary here. + decomposeShort(prevSrc, src, STOP_AT_LIMIT, onlyContiguous, + buffer, errorCode); + // Decompose until the next boundary. + src = decomposeShort(src, limit, STOP_AT_COMP_BOUNDARY, onlyContiguous, + buffer, errorCode); + if (U_FAILURE(errorCode)) { + break; + } + if ((src - prevSrc) > INT32_MAX) { // guard before buffer.equals() + errorCode = U_INDEX_OUTOFBOUNDS_ERROR; + return true; + } + recompose(buffer, 0, onlyContiguous); + if (!buffer.equals(prevSrc, src)) { + if (sink == nullptr) { + return false; + } + if (prevBoundary != prevSrc && + !ByteSinkUtil::appendUnchanged(prevBoundary, prevSrc, + *sink, options, edits, errorCode)) { + break; + } + if (!ByteSinkUtil::appendChange(prevSrc, src, buffer.getStart(), buffer.length(), + *sink, edits, errorCode)) { + break; + } + prevBoundary = src; + } + } + return true; +} + +UBool Normalizer2Impl::hasCompBoundaryBefore(const char16_t *src, const char16_t *limit) const { + if (src == limit || *src < minCompNoMaybeCP) { + return true; + } + UChar32 c; + uint16_t norm16; + UCPTRIE_FAST_U16_NEXT(normTrie, UCPTRIE_16, src, limit, c, norm16); + return norm16HasCompBoundaryBefore(norm16); +} + +UBool Normalizer2Impl::hasCompBoundaryBefore(const uint8_t *src, const uint8_t *limit) const { + if (src == limit) { + return true; + } + uint16_t norm16; + UCPTRIE_FAST_U8_NEXT(normTrie, UCPTRIE_16, src, limit, norm16); + return norm16HasCompBoundaryBefore(norm16); +} + +UBool Normalizer2Impl::hasCompBoundaryAfter(const char16_t *start, const char16_t *p, + UBool onlyContiguous) const { + if (start == p) { + return true; + } + UChar32 c; + uint16_t norm16; + UCPTRIE_FAST_U16_PREV(normTrie, UCPTRIE_16, start, p, c, norm16); + return norm16HasCompBoundaryAfter(norm16, onlyContiguous); +} + +UBool Normalizer2Impl::hasCompBoundaryAfter(const uint8_t *start, const uint8_t *p, + UBool onlyContiguous) const { + if (start == p) { + return true; + } + uint16_t norm16; + UCPTRIE_FAST_U8_PREV(normTrie, UCPTRIE_16, start, p, norm16); + return norm16HasCompBoundaryAfter(norm16, onlyContiguous); +} + +const char16_t *Normalizer2Impl::findPreviousCompBoundary(const char16_t *start, const char16_t *p, + UBool onlyContiguous) const { + while (p != start) { + const char16_t *codePointLimit = p; + UChar32 c; + uint16_t norm16; + UCPTRIE_FAST_U16_PREV(normTrie, UCPTRIE_16, start, p, c, norm16); + if (norm16HasCompBoundaryAfter(norm16, onlyContiguous)) { + return codePointLimit; + } + if (hasCompBoundaryBefore(c, norm16)) { + return p; + } + } + return p; +} + +const char16_t *Normalizer2Impl::findNextCompBoundary(const char16_t *p, const char16_t *limit, + UBool onlyContiguous) const { + while (p != limit) { + const char16_t *codePointStart = p; + UChar32 c; + uint16_t norm16; + UCPTRIE_FAST_U16_NEXT(normTrie, UCPTRIE_16, p, limit, c, norm16); + if (hasCompBoundaryBefore(c, norm16)) { + return codePointStart; + } + if (norm16HasCompBoundaryAfter(norm16, onlyContiguous)) { + return p; + } + } + return p; +} + +uint8_t Normalizer2Impl::getPreviousTrailCC(const char16_t *start, const char16_t *p) const { + if (start == p) { + return 0; + } + int32_t i = (int32_t)(p - start); + UChar32 c; + U16_PREV(start, 0, i, c); + return (uint8_t)getFCD16(c); +} + +uint8_t Normalizer2Impl::getPreviousTrailCC(const uint8_t *start, const uint8_t *p) const { + if (start == p) { + return 0; + } + int32_t i = (int32_t)(p - start); + UChar32 c; + U8_PREV(start, 0, i, c); + return (uint8_t)getFCD16(c); +} + +// Note: normalizer2impl.cpp r30982 (2011-nov-27) +// still had getFCDTrie() which built and cached an FCD trie. +// That provided faster access to FCD data than getFCD16FromNormData() +// but required synchronization and consumed some 10kB of heap memory +// in any process that uses FCD (e.g., via collation). +// minDecompNoCP etc. and smallFCD[] are intended to help with any loss of performance, +// at least for ASCII & CJK. + +// Ticket 20907 - The optimizer in MSVC/Visual Studio versions below 16.4 has trouble with this +// function on Windows ARM64. As a work-around, we disable optimizations for this function. +// This work-around could/should be removed once the following versions of Visual Studio are no +// longer supported: All versions of VS2017, and versions of VS2019 below 16.4. +#if (defined(_MSC_VER) && (defined(_M_ARM64)) && (_MSC_VER < 1924)) +#pragma optimize( "", off ) +#endif +// Gets the FCD value from the regular normalization data. +uint16_t Normalizer2Impl::getFCD16FromNormData(UChar32 c) const { + uint16_t norm16=getNorm16(c); + if (norm16 >= limitNoNo) { + if(norm16>=MIN_NORMAL_MAYBE_YES) { + // combining mark + norm16=getCCFromNormalYesOrMaybe(norm16); + return norm16|(norm16<<8); + } else if(norm16>=minMaybeYes) { + return 0; + } else { // isDecompNoAlgorithmic(norm16) + uint16_t deltaTrailCC = norm16 & DELTA_TCCC_MASK; + if (deltaTrailCC <= DELTA_TCCC_1) { + return deltaTrailCC >> OFFSET_SHIFT; + } + // Maps to an isCompYesAndZeroCC. + c=mapAlgorithmic(c, norm16); + norm16=getRawNorm16(c); + } + } + if(norm16<=minYesNo || isHangulLVT(norm16)) { + // no decomposition or Hangul syllable, all zeros + return 0; + } + // c decomposes, get everything from the variable-length extra data + const uint16_t *mapping=getMapping(norm16); + uint16_t firstUnit=*mapping; + norm16=firstUnit>>8; // tccc + if(firstUnit&MAPPING_HAS_CCC_LCCC_WORD) { + norm16|=*(mapping-1)&0xff00; // lccc + } + return norm16; +} +#if (defined(_MSC_VER) && (defined(_M_ARM64)) && (_MSC_VER < 1924)) +#pragma optimize( "", on ) +#endif + +// Dual functionality: +// buffer!=nullptr: normalize +// buffer==nullptr: isNormalized/quickCheck/spanQuickCheckYes +const char16_t * +Normalizer2Impl::makeFCD(const char16_t *src, const char16_t *limit, + ReorderingBuffer *buffer, + UErrorCode &errorCode) const { + // Tracks the last FCD-safe boundary, before lccc=0 or after properly-ordered tccc<=1. + // Similar to the prevBoundary in the compose() implementation. + const char16_t *prevBoundary=src; + int32_t prevFCD16=0; + if(limit==nullptr) { + src=copyLowPrefixFromNulTerminated(src, minLcccCP, buffer, errorCode); + if(U_FAILURE(errorCode)) { + return src; + } + if(prevBoundary1) { + --prevBoundary; + } + } + limit=u_strchr(src, 0); + } + + // Note: In this function we use buffer->appendZeroCC() because we track + // the lead and trail combining classes here, rather than leaving it to + // the ReorderingBuffer. + // The exception is the call to decomposeShort() which uses the buffer + // in the normal way. + + const char16_t *prevSrc; + UChar32 c=0; + uint16_t fcd16=0; + + for(;;) { + // count code units with lccc==0 + for(prevSrc=src; src!=limit;) { + if((c=*src)appendZeroCC(prevSrc, src, errorCode)) { + break; + } + if(src==limit) { + break; + } + prevBoundary=src; + // We know that the previous character's lccc==0. + if(prevFCD16<0) { + // Fetching the fcd16 value was deferred for this below-minLcccCP code point. + UChar32 prev=~prevFCD16; + if(prev1) { + --prevBoundary; + } + } + } else { + const char16_t *p=src-1; + if(U16_IS_TRAIL(*p) && prevSrc

1) { + prevBoundary=p; + } + } + // The start of the current character (c). + prevSrc=src; + } else if(src==limit) { + break; + } + + src+=U16_LENGTH(c); + // The current character (c) at [prevSrc..src[ has a non-zero lead combining class. + // Check for proper order, and decompose locally if necessary. + if((prevFCD16&0xff)<=(fcd16>>8)) { + // proper order: prev tccc <= current lccc + if((fcd16&0xff)<=1) { + prevBoundary=src; + } + if(buffer!=nullptr && !buffer->appendZeroCC(c, errorCode)) { + break; + } + prevFCD16=fcd16; + continue; + } else if(buffer==nullptr) { + return prevBoundary; // quick check "no" + } else { + /* + * Back out the part of the source that we copied or appended + * already but is now going to be decomposed. + * prevSrc is set to after what was copied/appended. + */ + buffer->removeSuffix((int32_t)(prevSrc-prevBoundary)); + /* + * Find the part of the source that needs to be decomposed, + * up to the next safe boundary. + */ + src=findNextFCDBoundary(src, limit); + /* + * The source text does not fulfill the conditions for FCD. + * Decompose and reorder a limited piece of the text. + */ + decomposeShort(prevBoundary, src, false, false, *buffer, errorCode); + if (U_FAILURE(errorCode)) { + break; + } + prevBoundary=src; + prevFCD16=0; + } + } + return src; +} + +void Normalizer2Impl::makeFCDAndAppend(const char16_t *src, const char16_t *limit, + UBool doMakeFCD, + UnicodeString &safeMiddle, + ReorderingBuffer &buffer, + UErrorCode &errorCode) const { + if(!buffer.isEmpty()) { + const char16_t *firstBoundaryInSrc=findNextFCDBoundary(src, limit); + if(src!=firstBoundaryInSrc) { + const char16_t *lastBoundaryInDest=findPreviousFCDBoundary(buffer.getStart(), + buffer.getLimit()); + int32_t destSuffixLength=(int32_t)(buffer.getLimit()-lastBoundaryInDest); + UnicodeString middle(lastBoundaryInDest, destSuffixLength); + buffer.removeSuffix(destSuffixLength); + safeMiddle=middle; + middle.append(src, (int32_t)(firstBoundaryInSrc-src)); + const char16_t *middleStart=middle.getBuffer(); + makeFCD(middleStart, middleStart+middle.length(), &buffer, errorCode); + if(U_FAILURE(errorCode)) { + return; + } + src=firstBoundaryInSrc; + } + } + if(doMakeFCD) { + makeFCD(src, limit, &buffer, errorCode); + } else { + if(limit==nullptr) { // appendZeroCC() needs limit!=nullptr + limit=u_strchr(src, 0); + } + buffer.appendZeroCC(src, limit, errorCode); + } +} + +const char16_t *Normalizer2Impl::findPreviousFCDBoundary(const char16_t *start, const char16_t *p) const { + while(start lpSet(new UnicodeSet, errorCode); + set=lpSet.getAlias(); + if(U_FAILURE(errorCode)) { + return; + } + UChar32 firstOrigin=(UChar32)(canonValue&CANON_VALUE_MASK); + canonValue=(canonValue&~CANON_VALUE_MASK)|CANON_HAS_SET|(uint32_t)canonStartSets.size(); + umutablecptrie_set(mutableTrie, decompLead, canonValue, &errorCode); + canonStartSets.adoptElement(lpSet.orphan(), errorCode); + if (U_FAILURE(errorCode)) { + return; + } + if(firstOrigin!=0) { + set->add(firstOrigin); + } + } else { + set=(UnicodeSet *)canonStartSets[(int32_t)(canonValue&CANON_VALUE_MASK)]; + } + set->add(origin); + } +} + +// C++ class for friend access to private Normalizer2Impl members. +class InitCanonIterData { +public: + static void doInit(Normalizer2Impl *impl, UErrorCode &errorCode); +}; + +U_CDECL_BEGIN + +// UInitOnce instantiation function for CanonIterData +static void U_CALLCONV +initCanonIterData(Normalizer2Impl *impl, UErrorCode &errorCode) { + InitCanonIterData::doInit(impl, errorCode); +} + +U_CDECL_END + +void InitCanonIterData::doInit(Normalizer2Impl *impl, UErrorCode &errorCode) { + U_ASSERT(impl->fCanonIterData == nullptr); + impl->fCanonIterData = new CanonIterData(errorCode); + if (impl->fCanonIterData == nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } + if (U_SUCCESS(errorCode)) { + UChar32 start = 0, end; + uint32_t value; + while ((end = ucptrie_getRange(impl->normTrie, start, + UCPMAP_RANGE_FIXED_LEAD_SURROGATES, Normalizer2Impl::INERT, + nullptr, nullptr, &value)) >= 0) { + // Call Normalizer2Impl::makeCanonIterDataFromNorm16() for a range of same-norm16 characters. + if (value != Normalizer2Impl::INERT) { + impl->makeCanonIterDataFromNorm16(start, end, value, *impl->fCanonIterData, errorCode); + } + start = end + 1; + } +#ifdef UCPTRIE_DEBUG + umutablecptrie_setName(impl->fCanonIterData->mutableTrie, "CanonIterData"); +#endif + impl->fCanonIterData->trie = umutablecptrie_buildImmutable( + impl->fCanonIterData->mutableTrie, UCPTRIE_TYPE_SMALL, UCPTRIE_VALUE_BITS_32, &errorCode); + umutablecptrie_close(impl->fCanonIterData->mutableTrie); + impl->fCanonIterData->mutableTrie = nullptr; + } + if (U_FAILURE(errorCode)) { + delete impl->fCanonIterData; + impl->fCanonIterData = nullptr; + } +} + +void Normalizer2Impl::makeCanonIterDataFromNorm16(UChar32 start, UChar32 end, const uint16_t norm16, + CanonIterData &newData, + UErrorCode &errorCode) const { + if(isInert(norm16) || (minYesNo<=norm16 && norm16 minYesNo) { + // c decomposes, get everything from the variable-length extra data + const uint16_t *mapping=getMapping(norm16_2); + uint16_t firstUnit=*mapping; + int32_t length=firstUnit&MAPPING_LENGTH_MASK; + if((firstUnit&MAPPING_HAS_CCC_LCCC_WORD)!=0) { + if(c==c2 && (*(mapping-1)&0xff)!=0) { + newValue|=CANON_NOT_SEGMENT_STARTER; // original c has cc!=0 + } + } + // Skip empty mappings (no characters in the decomposition). + if(length!=0) { + ++mapping; // skip over the firstUnit + // add c to first code point's start set + int32_t i=0; + U16_NEXT_UNSAFE(mapping, i, c2); + newData.addToStartSet(c, c2, errorCode); + // Set CANON_NOT_SEGMENT_STARTER for each remaining code point of a + // one-way mapping. A 2-way mapping is possible here after + // intermediate algorithmic mapping. + if(norm16_2>=minNoNo) { + while(i(this); + umtx_initOnce(me->fCanonIterDataInitOnce, &initCanonIterData, me, errorCode); + return U_SUCCESS(errorCode); +} + +int32_t Normalizer2Impl::getCanonValue(UChar32 c) const { + return (int32_t)ucptrie_get(fCanonIterData->trie, c); +} + +const UnicodeSet &Normalizer2Impl::getCanonStartSet(int32_t n) const { + return *(const UnicodeSet *)fCanonIterData->canonStartSets[n]; +} + +UBool Normalizer2Impl::isCanonSegmentStarter(UChar32 c) const { + return getCanonValue(c)>=0; +} + +UBool Normalizer2Impl::getCanonStartSet(UChar32 c, UnicodeSet &set) const { + int32_t canonValue=getCanonValue(c)&~CANON_NOT_SEGMENT_STARTER; + if(canonValue==0) { + return false; + } + set.clear(); + int32_t value=canonValue&CANON_VALUE_MASK; + if((canonValue&CANON_HAS_SET)!=0) { + set.addAll(getCanonStartSet(value)); + } else if(value!=0) { + set.add(value); + } + if((canonValue&CANON_HAS_COMPOSITIONS)!=0) { + uint16_t norm16=getRawNorm16(c); + if(norm16==JAMO_L) { + UChar32 syllable= + (UChar32)(Hangul::HANGUL_BASE+(c-Hangul::JAMO_L_BASE)*Hangul::JAMO_VT_COUNT); + set.add(syllable, syllable+Hangul::JAMO_VT_COUNT-1); + } else { + addComposites(getCompositionsList(norm16), set); + } + } + return true; +} + +U_NAMESPACE_END + +// Normalizer2 data swapping ----------------------------------------------- *** + +U_NAMESPACE_USE + +U_CAPI int32_t U_EXPORT2 +unorm2_swap(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const UDataInfo *pInfo; + int32_t headerSize; + + const uint8_t *inBytes; + uint8_t *outBytes; + + const int32_t *inIndexes; + int32_t indexes[Normalizer2Impl::IX_TOTAL_SIZE+1]; + + int32_t i, offset, nextOffset, size; + + /* udata_swapDataHeader checks the arguments */ + headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode); + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + /* check data format and format version */ + pInfo=(const UDataInfo *)((const char *)inData+4); + uint8_t formatVersion0=pInfo->formatVersion[0]; + if(!( + pInfo->dataFormat[0]==0x4e && /* dataFormat="Nrm2" */ + pInfo->dataFormat[1]==0x72 && + pInfo->dataFormat[2]==0x6d && + pInfo->dataFormat[3]==0x32 && + (1<=formatVersion0 && formatVersion0<=4) + )) { + udata_printError(ds, "unorm2_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized as Normalizer2 data\n", + pInfo->dataFormat[0], pInfo->dataFormat[1], + pInfo->dataFormat[2], pInfo->dataFormat[3], + pInfo->formatVersion[0]); + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + + inBytes=(const uint8_t *)inData+headerSize; + outBytes=(outData == nullptr) ? nullptr : (uint8_t *)outData+headerSize; + + inIndexes=(const int32_t *)inBytes; + int32_t minIndexesLength; + if(formatVersion0==1) { + minIndexesLength=Normalizer2Impl::IX_MIN_MAYBE_YES+1; + } else if(formatVersion0==2) { + minIndexesLength=Normalizer2Impl::IX_MIN_YES_NO_MAPPINGS_ONLY+1; + } else { + minIndexesLength=Normalizer2Impl::IX_MIN_LCCC_CP+1; + } + + if(length>=0) { + length-=headerSize; + if(length=0) { + if(lengthswapArray32(ds, inBytes, nextOffset-offset, outBytes, pErrorCode); + offset=nextOffset; + + /* swap the trie */ + nextOffset=indexes[Normalizer2Impl::IX_EXTRA_DATA_OFFSET]; + utrie_swapAnyVersion(ds, inBytes+offset, nextOffset-offset, outBytes+offset, pErrorCode); + offset=nextOffset; + + /* swap the uint16_t extraData[] */ + nextOffset=indexes[Normalizer2Impl::IX_SMALL_FCD_OFFSET]; + ds->swapArray16(ds, inBytes+offset, nextOffset-offset, outBytes+offset, pErrorCode); + offset=nextOffset; + + /* no need to swap the uint8_t smallFCD[] (new in formatVersion 2) */ + nextOffset=indexes[Normalizer2Impl::IX_SMALL_FCD_OFFSET+1]; + offset=nextOffset; + + U_ASSERT(offset==size); + } + + return headerSize+size; +} + +#endif // !UCONFIG_NO_NORMALIZATION diff --git a/intl/icu/source/common/normalizer2impl.h b/intl/icu/source/common/normalizer2impl.h new file mode 100644 index 0000000000..2cca33d349 --- /dev/null +++ b/intl/icu/source/common/normalizer2impl.h @@ -0,0 +1,987 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2009-2014, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: normalizer2impl.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2009nov22 +* created by: Markus W. Scherer +*/ + +#ifndef __NORMALIZER2IMPL_H__ +#define __NORMALIZER2IMPL_H__ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_NORMALIZATION + +#include "unicode/normalizer2.h" +#include "unicode/ucptrie.h" +#include "unicode/unistr.h" +#include "unicode/unorm.h" +#include "unicode/utf.h" +#include "unicode/utf16.h" +#include "mutex.h" +#include "udataswp.h" +#include "uset_imp.h" + +// When the nfc.nrm data is *not* hardcoded into the common library +// (with this constant set to 0), +// then it needs to be built into the data package: +// Add nfc.nrm to icu4c/source/data/Makefile.in DAT_FILES_SHORT +#define NORM2_HARDCODE_NFC_DATA 1 + +U_NAMESPACE_BEGIN + +struct CanonIterData; + +class ByteSink; +class Edits; +class InitCanonIterData; +class LcccContext; + +class U_COMMON_API Hangul { +public: + /* Korean Hangul and Jamo constants */ + enum { + JAMO_L_BASE=0x1100, /* "lead" jamo */ + JAMO_L_END=0x1112, + JAMO_V_BASE=0x1161, /* "vowel" jamo */ + JAMO_V_END=0x1175, + JAMO_T_BASE=0x11a7, /* "trail" jamo */ + JAMO_T_END=0x11c2, + + HANGUL_BASE=0xac00, + HANGUL_END=0xd7a3, + + JAMO_L_COUNT=19, + JAMO_V_COUNT=21, + JAMO_T_COUNT=28, + + JAMO_VT_COUNT=JAMO_V_COUNT*JAMO_T_COUNT, + + HANGUL_COUNT=JAMO_L_COUNT*JAMO_V_COUNT*JAMO_T_COUNT, + HANGUL_LIMIT=HANGUL_BASE+HANGUL_COUNT + }; + + static inline UBool isHangul(UChar32 c) { + return HANGUL_BASE<=c && c(INERT) : + UCPTRIE_FAST_GET(normTrie, UCPTRIE_16, c); + } + uint16_t getRawNorm16(UChar32 c) const { return UCPTRIE_FAST_GET(normTrie, UCPTRIE_16, c); } + + UNormalizationCheckResult getCompQuickCheck(uint16_t norm16) const { + if(norm16=MIN_NORMAL_MAYBE_YES) { + return getCCFromNormalYesOrMaybe(norm16); + } + if(norm16> OFFSET_SHIFT); + } + static uint8_t getCCFromYesOrMaybe(uint16_t norm16) { + return norm16>=MIN_NORMAL_MAYBE_YES ? getCCFromNormalYesOrMaybe(norm16) : 0; + } + uint8_t getCCFromYesOrMaybeCP(UChar32 c) const { + if (c < minCompNoMaybeCP) { return 0; } + return getCCFromYesOrMaybe(getNorm16(c)); + } + + /** + * Returns the FCD data for code point c. + * @param c A Unicode code point. + * @return The lccc(c) in bits 15..8 and tccc(c) in bits 7..0. + */ + uint16_t getFCD16(UChar32 c) const { + if(c>8]; + if(bits==0) { return false; } + return (UBool)((bits>>((lead>>5)&7))&1); + } + /** Returns the FCD value from the regular normalization data. */ + uint16_t getFCD16FromNormData(UChar32 c) const; + + /** + * Gets the decomposition for one code point. + * @param c code point + * @param buffer out-only buffer for algorithmic decompositions + * @param length out-only, takes the length of the decomposition, if any + * @return pointer to the decomposition, or NULL if none + */ + const char16_t *getDecomposition(UChar32 c, char16_t buffer[4], int32_t &length) const; + + /** + * Gets the raw decomposition for one code point. + * @param c code point + * @param buffer out-only buffer for algorithmic decompositions + * @param length out-only, takes the length of the decomposition, if any + * @return pointer to the decomposition, or NULL if none + */ + const char16_t *getRawDecomposition(UChar32 c, char16_t buffer[30], int32_t &length) const; + + UChar32 composePair(UChar32 a, UChar32 b) const; + + UBool isCanonSegmentStarter(UChar32 c) const; + UBool getCanonStartSet(UChar32 c, UnicodeSet &set) const; + + enum { + // Fixed norm16 values. + MIN_YES_YES_WITH_CC=0xfe02, + JAMO_VT=0xfe00, + MIN_NORMAL_MAYBE_YES=0xfc00, + JAMO_L=2, // offset=1 hasCompBoundaryAfter=false + INERT=1, // offset=0 hasCompBoundaryAfter=true + + // norm16 bit 0 is comp-boundary-after. + HAS_COMP_BOUNDARY_AFTER=1, + OFFSET_SHIFT=1, + + // For algorithmic one-way mappings, norm16 bits 2..1 indicate the + // tccc (0, 1, >1) for quick FCC boundary-after tests. + DELTA_TCCC_0=0, + DELTA_TCCC_1=2, + DELTA_TCCC_GT_1=4, + DELTA_TCCC_MASK=6, + DELTA_SHIFT=3, + + MAX_DELTA=0x40 + }; + + enum { + // Byte offsets from the start of the data, after the generic header. + IX_NORM_TRIE_OFFSET, + IX_EXTRA_DATA_OFFSET, + IX_SMALL_FCD_OFFSET, + IX_RESERVED3_OFFSET, + IX_RESERVED4_OFFSET, + IX_RESERVED5_OFFSET, + IX_RESERVED6_OFFSET, + IX_TOTAL_SIZE, + + // Code point thresholds for quick check codes. + IX_MIN_DECOMP_NO_CP, + IX_MIN_COMP_NO_MAYBE_CP, + + // Norm16 value thresholds for quick check combinations and types of extra data. + + /** Mappings & compositions in [minYesNo..minYesNoMappingsOnly[. */ + IX_MIN_YES_NO, + /** Mappings are comp-normalized. */ + IX_MIN_NO_NO, + IX_LIMIT_NO_NO, + IX_MIN_MAYBE_YES, + + /** Mappings only in [minYesNoMappingsOnly..minNoNo[. */ + IX_MIN_YES_NO_MAPPINGS_ONLY, + /** Mappings are not comp-normalized but have a comp boundary before. */ + IX_MIN_NO_NO_COMP_BOUNDARY_BEFORE, + /** Mappings do not have a comp boundary before. */ + IX_MIN_NO_NO_COMP_NO_MAYBE_CC, + /** Mappings to the empty string. */ + IX_MIN_NO_NO_EMPTY, + + IX_MIN_LCCC_CP, + IX_RESERVED19, + IX_COUNT + }; + + enum { + MAPPING_HAS_CCC_LCCC_WORD=0x80, + MAPPING_HAS_RAW_MAPPING=0x40, + // unused bit 0x20, + MAPPING_LENGTH_MASK=0x1f + }; + + enum { + COMP_1_LAST_TUPLE=0x8000, + COMP_1_TRIPLE=1, + COMP_1_TRAIL_LIMIT=0x3400, + COMP_1_TRAIL_MASK=0x7ffe, + COMP_1_TRAIL_SHIFT=9, // 10-1 for the "triple" bit + COMP_2_TRAIL_SHIFT=6, + COMP_2_TRAIL_MASK=0xffc0 + }; + + // higher-level functionality ------------------------------------------ *** + + // NFD without an NFD Normalizer2 instance. + UnicodeString &decompose(const UnicodeString &src, UnicodeString &dest, + UErrorCode &errorCode) const; + /** + * Decomposes [src, limit[ and writes the result to dest. + * limit can be NULL if src is NUL-terminated. + * destLengthEstimate is the initial dest buffer capacity and can be -1. + */ + void decompose(const char16_t *src, const char16_t *limit, + UnicodeString &dest, int32_t destLengthEstimate, + UErrorCode &errorCode) const; + + const char16_t *decompose(const char16_t *src, const char16_t *limit, + ReorderingBuffer *buffer, UErrorCode &errorCode) const; + void decomposeAndAppend(const char16_t *src, const char16_t *limit, + UBool doDecompose, + UnicodeString &safeMiddle, + ReorderingBuffer &buffer, + UErrorCode &errorCode) const; + + /** sink==nullptr: isNormalized()/spanQuickCheckYes() */ + const uint8_t *decomposeUTF8(uint32_t options, + const uint8_t *src, const uint8_t *limit, + ByteSink *sink, Edits *edits, UErrorCode &errorCode) const; + + UBool compose(const char16_t *src, const char16_t *limit, + UBool onlyContiguous, + UBool doCompose, + ReorderingBuffer &buffer, + UErrorCode &errorCode) const; + const char16_t *composeQuickCheck(const char16_t *src, const char16_t *limit, + UBool onlyContiguous, + UNormalizationCheckResult *pQCResult) const; + void composeAndAppend(const char16_t *src, const char16_t *limit, + UBool doCompose, + UBool onlyContiguous, + UnicodeString &safeMiddle, + ReorderingBuffer &buffer, + UErrorCode &errorCode) const; + + /** sink==nullptr: isNormalized() */ + UBool composeUTF8(uint32_t options, UBool onlyContiguous, + const uint8_t *src, const uint8_t *limit, + ByteSink *sink, icu::Edits *edits, UErrorCode &errorCode) const; + + const char16_t *makeFCD(const char16_t *src, const char16_t *limit, + ReorderingBuffer *buffer, UErrorCode &errorCode) const; + void makeFCDAndAppend(const char16_t *src, const char16_t *limit, + UBool doMakeFCD, + UnicodeString &safeMiddle, + ReorderingBuffer &buffer, + UErrorCode &errorCode) const; + + UBool hasDecompBoundaryBefore(UChar32 c) const; + UBool norm16HasDecompBoundaryBefore(uint16_t norm16) const; + UBool hasDecompBoundaryAfter(UChar32 c) const; + UBool norm16HasDecompBoundaryAfter(uint16_t norm16) const; + UBool isDecompInert(UChar32 c) const { return isDecompYesAndZeroCC(getNorm16(c)); } + + UBool hasCompBoundaryBefore(UChar32 c) const { + return c=minMaybeYes; } + static UBool isInert(uint16_t norm16) { return norm16==INERT; } + static UBool isJamoL(uint16_t norm16) { return norm16==JAMO_L; } + static UBool isJamoVT(uint16_t norm16) { return norm16==JAMO_VT; } + uint16_t hangulLVT() const { return minYesNoMappingsOnly|HAS_COMP_BOUNDARY_AFTER; } + UBool isHangulLV(uint16_t norm16) const { return norm16==minYesNo; } + UBool isHangulLVT(uint16_t norm16) const { + return norm16==hangulLVT(); + } + UBool isCompYesAndZeroCC(uint16_t norm16) const { return norm16=MIN_YES_YES_WITH_CC || norm16=limitNoNo; } + + // For use with isCompYes(). + // Perhaps the compiler can combine the two tests for MIN_YES_YES_WITH_CC. + // static uint8_t getCCFromYes(uint16_t norm16) { + // return norm16>=MIN_YES_YES_WITH_CC ? getCCFromNormalYesOrMaybe(norm16) : 0; + // } + uint8_t getCCFromNoNo(uint16_t norm16) const { + const uint16_t *mapping=getMapping(norm16); + if(*mapping&MAPPING_HAS_CCC_LCCC_WORD) { + return (uint8_t)*(mapping-1); + } else { + return 0; + } + } + // requires that the [cpStart..cpLimit[ character passes isCompYesAndZeroCC() + uint8_t getTrailCCFromCompYesAndZeroCC(uint16_t norm16) const { + if(norm16<=minYesNo) { + return 0; // yesYes and Hangul LV have ccc=tccc=0 + } else { + // For Hangul LVT we harmlessly fetch a firstUnit with tccc=0 here. + return (uint8_t)(*getMapping(norm16)>>8); // tccc from yesNo + } + } + uint8_t getPreviousTrailCC(const char16_t *start, const char16_t *p) const; + uint8_t getPreviousTrailCC(const uint8_t *start, const uint8_t *p) const; + + // Requires algorithmic-NoNo. + UChar32 mapAlgorithmic(UChar32 c, uint16_t norm16) const { + return c+(norm16>>DELTA_SHIFT)-centerNoNoDelta; + } + UChar32 getAlgorithmicDelta(uint16_t norm16) const { + return (norm16>>DELTA_SHIFT)-centerNoNoDelta; + } + + // Requires minYesNo>OFFSET_SHIFT); } + const uint16_t *getCompositionsListForDecompYes(uint16_t norm16) const { + if(norm16>OFFSET_SHIFT); + } + /** + * @param c code point must have compositions + * @return compositions list pointer + */ + const uint16_t *getCompositionsList(uint16_t norm16) const { + return isDecompYes(norm16) ? + getCompositionsListForDecompYes(norm16) : + getCompositionsListForComposite(norm16); + } + + const char16_t *copyLowPrefixFromNulTerminated(const char16_t *src, + UChar32 minNeedDataCP, + ReorderingBuffer *buffer, + UErrorCode &errorCode) const; + + enum StopAt { STOP_AT_LIMIT, STOP_AT_DECOMP_BOUNDARY, STOP_AT_COMP_BOUNDARY }; + + const char16_t *decomposeShort(const char16_t *src, const char16_t *limit, + UBool stopAtCompBoundary, UBool onlyContiguous, + ReorderingBuffer &buffer, UErrorCode &errorCode) const; + UBool decompose(UChar32 c, uint16_t norm16, + ReorderingBuffer &buffer, UErrorCode &errorCode) const; + + const uint8_t *decomposeShort(const uint8_t *src, const uint8_t *limit, + StopAt stopAt, UBool onlyContiguous, + ReorderingBuffer &buffer, UErrorCode &errorCode) const; + + static int32_t combine(const uint16_t *list, UChar32 trail); + void addComposites(const uint16_t *list, UnicodeSet &set) const; + void recompose(ReorderingBuffer &buffer, int32_t recomposeStartIndex, + UBool onlyContiguous) const; + + UBool hasCompBoundaryBefore(UChar32 c, uint16_t norm16) const { + return cclone()), + currentIndex(copy.currentIndex), nextIndex(copy.nextIndex), + buffer(copy.buffer), bufferPos(copy.bufferPos) +{ + init(); +} + +void +Normalizer::init() { + UErrorCode errorCode=U_ZERO_ERROR; + fNorm2=Normalizer2Factory::getInstance(fUMode, errorCode); + if(fOptions&UNORM_UNICODE_3_2) { + delete fFilteredNorm2; + fNorm2=fFilteredNorm2= + new FilteredNormalizer2(*fNorm2, *uniset_getUnicode32Instance(errorCode)); + } + if(U_FAILURE(errorCode)) { + errorCode=U_ZERO_ERROR; + fNorm2=Normalizer2Factory::getNoopInstance(errorCode); + } +} + +Normalizer::~Normalizer() +{ + delete fFilteredNorm2; + delete text; +} + +Normalizer* +Normalizer::clone() const +{ + return new Normalizer(*this); +} + +/** + * Generates a hash code for this iterator. + */ +int32_t Normalizer::hashCode() const +{ + return text->hashCode() + fUMode + fOptions + buffer.hashCode() + bufferPos + currentIndex + nextIndex; +} + +bool Normalizer::operator==(const Normalizer& that) const +{ + return + this==&that || + (fUMode==that.fUMode && + fOptions==that.fOptions && + *text==*that.text && + buffer==that.buffer && + bufferPos==that.bufferPos && + nextIndex==that.nextIndex); +} + +//------------------------------------------------------------------------- +// Static utility methods +//------------------------------------------------------------------------- + +void U_EXPORT2 +Normalizer::normalize(const UnicodeString& source, + UNormalizationMode mode, int32_t options, + UnicodeString& result, + UErrorCode &status) { + if(source.isBogus() || U_FAILURE(status)) { + result.setToBogus(); + if(U_SUCCESS(status)) { + status=U_ILLEGAL_ARGUMENT_ERROR; + } + } else { + UnicodeString localDest; + UnicodeString *dest; + + if(&source!=&result) { + dest=&result; + } else { + // the source and result strings are the same object, use a temporary one + dest=&localDest; + } + const Normalizer2 *n2=Normalizer2Factory::getInstance(mode, status); + if(U_SUCCESS(status)) { + if(options&UNORM_UNICODE_3_2) { + FilteredNormalizer2(*n2, *uniset_getUnicode32Instance(status)). + normalize(source, *dest, status); + } else { + n2->normalize(source, *dest, status); + } + } + if(dest==&localDest && U_SUCCESS(status)) { + result=*dest; + } + } +} + +void U_EXPORT2 +Normalizer::compose(const UnicodeString& source, + UBool compat, int32_t options, + UnicodeString& result, + UErrorCode &status) { + normalize(source, compat ? UNORM_NFKC : UNORM_NFC, options, result, status); +} + +void U_EXPORT2 +Normalizer::decompose(const UnicodeString& source, + UBool compat, int32_t options, + UnicodeString& result, + UErrorCode &status) { + normalize(source, compat ? UNORM_NFKD : UNORM_NFD, options, result, status); +} + +UNormalizationCheckResult +Normalizer::quickCheck(const UnicodeString& source, + UNormalizationMode mode, int32_t options, + UErrorCode &status) { + const Normalizer2 *n2=Normalizer2Factory::getInstance(mode, status); + if(U_SUCCESS(status)) { + if(options&UNORM_UNICODE_3_2) { + return FilteredNormalizer2(*n2, *uniset_getUnicode32Instance(status)). + quickCheck(source, status); + } else { + return n2->quickCheck(source, status); + } + } else { + return UNORM_MAYBE; + } +} + +UBool +Normalizer::isNormalized(const UnicodeString& source, + UNormalizationMode mode, int32_t options, + UErrorCode &status) { + const Normalizer2 *n2=Normalizer2Factory::getInstance(mode, status); + if(U_SUCCESS(status)) { + if(options&UNORM_UNICODE_3_2) { + return FilteredNormalizer2(*n2, *uniset_getUnicode32Instance(status)). + isNormalized(source, status); + } else { + return n2->isNormalized(source, status); + } + } else { + return false; + } +} + +UnicodeString & U_EXPORT2 +Normalizer::concatenate(const UnicodeString &left, const UnicodeString &right, + UnicodeString &result, + UNormalizationMode mode, int32_t options, + UErrorCode &errorCode) { + if(left.isBogus() || right.isBogus() || U_FAILURE(errorCode)) { + result.setToBogus(); + if(U_SUCCESS(errorCode)) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + } + } else { + UnicodeString localDest; + UnicodeString *dest; + + if(&right!=&result) { + dest=&result; + } else { + // the right and result strings are the same object, use a temporary one + dest=&localDest; + } + *dest=left; + const Normalizer2 *n2=Normalizer2Factory::getInstance(mode, errorCode); + if(U_SUCCESS(errorCode)) { + if(options&UNORM_UNICODE_3_2) { + FilteredNormalizer2(*n2, *uniset_getUnicode32Instance(errorCode)). + append(*dest, right, errorCode); + } else { + n2->append(*dest, right, errorCode); + } + } + if(dest==&localDest && U_SUCCESS(errorCode)) { + result=*dest; + } + } + return result; +} + +//------------------------------------------------------------------------- +// Iteration API +//------------------------------------------------------------------------- + +/** + * Return the current character in the normalized text. + */ +UChar32 Normalizer::current() { + if(bufferPos0 || previousNormalize()) { + UChar32 c=buffer.char32At(bufferPos-1); + bufferPos-=U16_LENGTH(c); + return c; + } else { + return DONE; + } +} + +void Normalizer::reset() { + currentIndex=nextIndex=text->setToStart(); + clearBuffer(); +} + +void +Normalizer::setIndexOnly(int32_t index) { + text->setIndex(index); // pins index + currentIndex=nextIndex=text->getIndex(); + clearBuffer(); +} + +/** + * Return the first character in the normalized text. This resets + * the Normalizer's position to the beginning of the text. + */ +UChar32 Normalizer::first() { + reset(); + return next(); +} + +/** + * Return the last character in the normalized text. This resets + * the Normalizer's position to be just before the + * the input text corresponding to that normalized character. + */ +UChar32 Normalizer::last() { + currentIndex=nextIndex=text->setToEnd(); + clearBuffer(); + return previous(); +} + +/** + * Retrieve the current iteration position in the input text that is + * being normalized. This method is useful in applications such as + * searching, where you need to be able to determine the position in + * the input text that corresponds to a given normalized output character. + *

+ * Note: This method sets the position in the input, while + * {@link #next} and {@link #previous} iterate through characters in the + * output. This means that there is not necessarily a one-to-one + * correspondence between characters returned by next and + * previous and the indices passed to and returned from + * setIndex and {@link #getIndex}. + * + */ +int32_t Normalizer::getIndex() const { + if(bufferPosCharacterIterator or the start (i.e. 0) of the String + * over which this Normalizer is iterating + */ +int32_t Normalizer::startIndex() const { + return text->startIndex(); +} + +/** + * Retrieve the index of the end of the input text. This is the end index + * of the CharacterIterator or the length of the String + * over which this Normalizer is iterating + */ +int32_t Normalizer::endIndex() const { + return text->endIndex(); +} + +//------------------------------------------------------------------------- +// Property access methods +//------------------------------------------------------------------------- + +void +Normalizer::setMode(UNormalizationMode newMode) +{ + fUMode = newMode; + init(); +} + +UNormalizationMode +Normalizer::getUMode() const +{ + return fUMode; +} + +void +Normalizer::setOption(int32_t option, + UBool value) +{ + if (value) { + fOptions |= option; + } else { + fOptions &= (~option); + } + init(); +} + +UBool +Normalizer::getOption(int32_t option) const +{ + return (fOptions & option) != 0; +} + +/** + * Set the input text over which this Normalizer will iterate. + * The iteration position is set to the beginning of the input text. + */ +void +Normalizer::setText(const UnicodeString& newText, + UErrorCode &status) +{ + if (U_FAILURE(status)) { + return; + } + CharacterIterator *newIter = new StringCharacterIterator(newText); + if (newIter == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + delete text; + text = newIter; + reset(); +} + +/** + * Set the input text over which this Normalizer will iterate. + * The iteration position is set to the beginning of the string. + */ +void +Normalizer::setText(const CharacterIterator& newText, + UErrorCode &status) +{ + if (U_FAILURE(status)) { + return; + } + CharacterIterator *newIter = newText.clone(); + if (newIter == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + delete text; + text = newIter; + reset(); +} + +void +Normalizer::setText(ConstChar16Ptr newText, + int32_t length, + UErrorCode &status) +{ + if (U_FAILURE(status)) { + return; + } + CharacterIterator *newIter = new UCharCharacterIterator(newText, length); + if (newIter == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + delete text; + text = newIter; + reset(); +} + +/** + * Copies the text under iteration into the UnicodeString referred to by "result". + * @param result Receives a copy of the text under iteration. + */ +void +Normalizer::getText(UnicodeString& result) +{ + text->getText(result); +} + +//------------------------------------------------------------------------- +// Private utility methods +//------------------------------------------------------------------------- + +void Normalizer::clearBuffer() { + buffer.remove(); + bufferPos=0; +} + +UBool +Normalizer::nextNormalize() { + clearBuffer(); + currentIndex=nextIndex; + text->setIndex(nextIndex); + if(!text->hasNext()) { + return false; + } + // Skip at least one character so we make progress. + UnicodeString segment(text->next32PostInc()); + while(text->hasNext()) { + UChar32 c; + if(fNorm2->hasBoundaryBefore(c=text->next32PostInc())) { + text->move32(-1, CharacterIterator::kCurrent); + break; + } + segment.append(c); + } + nextIndex=text->getIndex(); + UErrorCode errorCode=U_ZERO_ERROR; + fNorm2->normalize(segment, buffer, errorCode); + return U_SUCCESS(errorCode) && !buffer.isEmpty(); +} + +UBool +Normalizer::previousNormalize() { + clearBuffer(); + nextIndex=currentIndex; + text->setIndex(currentIndex); + if(!text->hasPrevious()) { + return false; + } + UnicodeString segment; + while(text->hasPrevious()) { + UChar32 c=text->previous32(); + segment.insert(0, c); + if(fNorm2->hasBoundaryBefore(c)) { + break; + } + } + currentIndex=text->getIndex(); + UErrorCode errorCode=U_ZERO_ERROR; + fNorm2->normalize(segment, buffer, errorCode); + bufferPos=buffer.length(); + return U_SUCCESS(errorCode) && !buffer.isEmpty(); +} + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_NORMALIZATION */ diff --git a/intl/icu/source/common/parsepos.cpp b/intl/icu/source/common/parsepos.cpp new file mode 100644 index 0000000000..56c6c78813 --- /dev/null +++ b/intl/icu/source/common/parsepos.cpp @@ -0,0 +1,23 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2003-2003, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ + +#include "unicode/parsepos.h" + +U_NAMESPACE_BEGIN + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(ParsePosition) + +ParsePosition::~ParsePosition() {} + +ParsePosition * +ParsePosition::clone() const { + return new ParsePosition(*this); +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/patternprops.cpp b/intl/icu/source/common/patternprops.cpp new file mode 100644 index 0000000000..9922683cda --- /dev/null +++ b/intl/icu/source/common/patternprops.cpp @@ -0,0 +1,230 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2011, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: patternprops.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2011mar13 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" +#include "patternprops.h" + +U_NAMESPACE_BEGIN + +/* + * One byte per Latin-1 character. + * Bit 0 is set if either Pattern property is true, + * bit 1 if Pattern_Syntax is true, + * bit 2 if Pattern_White_Space is true. + * That is, Pattern_Syntax is encoded as 3 and Pattern_White_Space as 5. + */ +static const uint8_t latin1[256]={ + // WS: 9..D + 0, 0, 0, 0, 0, 0, 0, 0, 0, 5, 5, 5, 5, 5, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + // WS: 20 Syntax: 21..2F + 5, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, + // Syntax: 3A..40 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 3, 3, 3, 3, 3, + 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + // Syntax: 5B..5E + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 3, 3, 3, 0, + // Syntax: 60 + 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + // Syntax: 7B..7E + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 3, 3, 3, 0, + // WS: 85 + 0, 0, 0, 0, 0, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + // Syntax: A1..A7, A9, AB, AC, AE + 0, 3, 3, 3, 3, 3, 3, 3, 0, 3, 0, 3, 3, 0, 3, 0, + // Syntax: B0, B1, B6, BB, BF + 3, 3, 0, 0, 0, 0, 3, 0, 0, 0, 0, 3, 0, 0, 0, 3, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + // Syntax: D7 + 0, 0, 0, 0, 0, 0, 0, 3, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + // Syntax: F7 + 0, 0, 0, 0, 0, 0, 0, 3, 0, 0, 0, 0, 0, 0, 0, 0 +}; + +/* + * One byte per 32 characters from U+2000..U+303F indexing into + * a small table of 32-bit data words. + * The first two data words are all-zeros and all-ones. + */ +static const uint8_t index2000[130]={ + 2, 3, 4, 0, 0, 0, 0, 0, // 20xx + 0, 0, 0, 0, 5, 1, 1, 1, // 21xx + 1, 1, 1, 1, 1, 1, 1, 1, // 22xx + 1, 1, 1, 1, 1, 1, 1, 1, // 23xx + 1, 1, 1, 0, 0, 0, 0, 0, // 24xx + 1, 1, 1, 1, 1, 1, 1, 1, // 25xx + 1, 1, 1, 1, 1, 1, 1, 1, // 26xx + 1, 1, 1, 6, 7, 1, 1, 1, // 27xx + 1, 1, 1, 1, 1, 1, 1, 1, // 28xx + 1, 1, 1, 1, 1, 1, 1, 1, // 29xx + 1, 1, 1, 1, 1, 1, 1, 1, // 2Axx + 1, 1, 1, 1, 1, 1, 1, 1, // 2Bxx + 0, 0, 0, 0, 0, 0, 0, 0, // 2Cxx + 0, 0, 0, 0, 0, 0, 0, 0, // 2Dxx + 1, 1, 1, 1, 0, 0, 0, 0, // 2Exx + 0, 0, 0, 0, 0, 0, 0, 0, // 2Fxx + 8, 9 // 3000..303F +}; + +/* + * One 32-bit integer per 32 characters. Ranges of all-false and all-true + * are mapped to the first two values, other ranges map to appropriate bit patterns. + */ +static const uint32_t syntax2000[]={ + 0, + 0xffffffff, + 0xffff0000, // 2: 2010..201F + 0x7fff00ff, // 3: 2020..2027, 2030..203E + 0x7feffffe, // 4: 2041..2053, 2055..205E + 0xffff0000, // 5: 2190..219F + 0x003fffff, // 6: 2760..2775 + 0xfff00000, // 7: 2794..279F + 0xffffff0e, // 8: 3001..3003, 3008..301F + 0x00010001 // 9: 3020, 3030 +}; + +/* + * Same as syntax2000, but with additional bits set for the + * Pattern_White_Space characters 200E 200F 2028 2029. + */ +static const uint32_t syntaxOrWhiteSpace2000[]={ + 0, + 0xffffffff, + 0xffffc000, // 2: 200E..201F + 0x7fff03ff, // 3: 2020..2029, 2030..203E + 0x7feffffe, // 4: 2041..2053, 2055..205E + 0xffff0000, // 5: 2190..219F + 0x003fffff, // 6: 2760..2775 + 0xfff00000, // 7: 2794..279F + 0xffffff0e, // 8: 3001..3003, 3008..301F + 0x00010001 // 9: 3020, 3030 +}; + +UBool +PatternProps::isSyntax(UChar32 c) { + if(c<0) { + return false; + } else if(c<=0xff) { + return (UBool)(latin1[c]>>1)&1; + } else if(c<0x2010) { + return false; + } else if(c<=0x3030) { + uint32_t bits=syntax2000[index2000[(c-0x2000)>>5]]; + return (UBool)((bits>>(c&0x1f))&1); + } else if(0xfd3e<=c && c<=0xfe46) { + return c<=0xfd3f || 0xfe45<=c; + } else { + return false; + } +} + +UBool +PatternProps::isSyntaxOrWhiteSpace(UChar32 c) { + if(c<0) { + return false; + } else if(c<=0xff) { + return (UBool)(latin1[c]&1); + } else if(c<0x200e) { + return false; + } else if(c<=0x3030) { + uint32_t bits=syntaxOrWhiteSpace2000[index2000[(c-0x2000)>>5]]; + return (UBool)((bits>>(c&0x1f))&1); + } else if(0xfd3e<=c && c<=0xfe46) { + return c<=0xfd3f || 0xfe45<=c; + } else { + return false; + } +} + +UBool +PatternProps::isWhiteSpace(UChar32 c) { + if(c<0) { + return false; + } else if(c<=0xff) { + return (UBool)(latin1[c]>>2)&1; + } else if(0x200e<=c && c<=0x2029) { + return c<=0x200f || 0x2028<=c; + } else { + return false; + } +} + +const char16_t * +PatternProps::skipWhiteSpace(const char16_t *s, int32_t length) { + while(length>0 && isWhiteSpace(*s)) { + ++s; + --length; + } + return s; +} + +int32_t +PatternProps::skipWhiteSpace(const UnicodeString& s, int32_t start) { + int32_t i = start; + int32_t length = s.length(); + while(i0 && !isSyntaxOrWhiteSpace(*s)) { + ++s; + --length; + } + return s; +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/patternprops.h b/intl/icu/source/common/patternprops.h new file mode 100644 index 0000000000..23a88333f4 --- /dev/null +++ b/intl/icu/source/common/patternprops.h @@ -0,0 +1,98 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2011, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: patternprops.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2011mar13 +* created by: Markus W. Scherer +*/ + +#ifndef __PATTERNPROPS_H__ +#define __PATTERNPROPS_H__ + +#include "unicode/unistr.h" +#include "unicode/utypes.h" + +U_NAMESPACE_BEGIN + +/** + * Implements the immutable Unicode properties Pattern_Syntax and Pattern_White_Space. + * Hardcodes these properties, does not load data, does not depend on other ICU classes. + *

+ * Note: Both properties include ASCII as well as non-ASCII, non-Latin-1 code points, + * and both properties only include BMP code points (no supplementary ones). + * Pattern_Syntax includes some unassigned code points. + *

+ * [:Pattern_White_Space:] = + * [\u0009-\u000D\ \u0085\u200E\u200F\u2028\u2029] + *

+ * [:Pattern_Syntax:] = + * [!-/\:-@\[-\^`\{-~\u00A1-\u00A7\u00A9\u00AB\u00AC\u00AE + * \u00B0\u00B1\u00B6\u00BB\u00BF\u00D7\u00F7 + * \u2010-\u2027\u2030-\u203E\u2041-\u2053\u2055-\u205E + * \u2190-\u245F\u2500-\u2775\u2794-\u2BFF\u2E00-\u2E7F + * \u3001-\u3003\u3008-\u3020\u3030\uFD3E\uFD3F\uFE45\uFE46] + * @author mscherer + */ +class U_COMMON_API PatternProps { +public: + /** + * @return true if c is a Pattern_Syntax code point. + */ + static UBool isSyntax(UChar32 c); + + /** + * @return true if c is a Pattern_Syntax or Pattern_White_Space code point. + */ + static UBool isSyntaxOrWhiteSpace(UChar32 c); + + /** + * @return true if c is a Pattern_White_Space character. + */ + static UBool isWhiteSpace(UChar32 c); + + /** + * Skips over Pattern_White_Space starting at s. + * @return The smallest pointer at or after s with a non-white space character. + */ + static const char16_t *skipWhiteSpace(const char16_t *s, int32_t length); + + /** + * Skips over Pattern_White_Space starting at index start in s. + * @return The smallest index at or after start with a non-white space character. + */ + static int32_t skipWhiteSpace(const UnicodeString &s, int32_t start); + + /** + * @return s except with leading and trailing Pattern_White_Space removed and length adjusted. + */ + static const char16_t *trimWhiteSpace(const char16_t *s, int32_t &length); + + /** + * Tests whether the string contains a "pattern identifier", that is, + * whether it contains only non-Pattern_White_Space, non-Pattern_Syntax characters. + * @return true if there are no Pattern_White_Space or Pattern_Syntax characters in s. + */ + static UBool isIdentifier(const char16_t *s, int32_t length); + + /** + * Skips over a "pattern identifier" starting at index s. + * @return The smallest pointer at or after s with + * a Pattern_White_Space or Pattern_Syntax character. + */ + static const char16_t *skipIdentifier(const char16_t *s, int32_t length); + +private: + PatternProps() = delete; // no constructor: all static methods +}; + +U_NAMESPACE_END + +#endif // __PATTERNPROPS_H__ diff --git a/intl/icu/source/common/pluralmap.cpp b/intl/icu/source/common/pluralmap.cpp new file mode 100644 index 0000000000..7e6156de89 --- /dev/null +++ b/intl/icu/source/common/pluralmap.cpp @@ -0,0 +1,44 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + * Copyright (C) 2015, International Business Machines Corporation and + * others. All Rights Reserved. + */ + +#include "unicode/unistr.h" +#include "charstr.h" +#include "cstring.h" +#include "pluralmap.h" + +U_NAMESPACE_BEGIN + +static const char * const gPluralForms[] = { + "other", "zero", "one", "two", "few", "many"}; + +PluralMapBase::Category +PluralMapBase::toCategory(const char *pluralForm) { + for (int32_t i = 0; i < UPRV_LENGTHOF(gPluralForms); ++i) { + if (uprv_strcmp(pluralForm, gPluralForms[i]) == 0) { + return static_cast(i); + } + } + return NONE; +} + +PluralMapBase::Category +PluralMapBase::toCategory(const UnicodeString &pluralForm) { + CharString cCategory; + UErrorCode status = U_ZERO_ERROR; + cCategory.appendInvariantChars(pluralForm, status); + return U_FAILURE(status) ? NONE : toCategory(cCategory.data()); +} + +const char *PluralMapBase::getCategoryName(Category c) { + int32_t index = c; + return (index < 0 || index >= UPRV_LENGTHOF(gPluralForms)) ? + nullptr : gPluralForms[index]; +} + + +U_NAMESPACE_END + diff --git a/intl/icu/source/common/pluralmap.h b/intl/icu/source/common/pluralmap.h new file mode 100644 index 0000000000..11683599ed --- /dev/null +++ b/intl/icu/source/common/pluralmap.h @@ -0,0 +1,292 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 2015, International Business Machines Corporation and +* others. All Rights Reserved. +****************************************************************************** +* +* File pluralmap.h - PluralMap class that maps plural categories to values. +****************************************************************************** +*/ + +#ifndef __PLURAL_MAP_H__ +#define __PLURAL_MAP_H__ + +#include "unicode/uobject.h" +#include "cmemory.h" + +U_NAMESPACE_BEGIN + +class UnicodeString; + +class U_COMMON_API PluralMapBase : public UMemory { +public: + /** + * The names of all the plural categories. NONE is not an actual plural + * category, but rather represents the absence of a plural category. + */ + enum Category { + NONE = -1, + OTHER, + ZERO, + ONE, + TWO, + FEW, + MANY, + CATEGORY_COUNT + }; + + /** + * Converts a category name such as "zero", "one", "two", "few", "many" + * or "other" to a category enum. Returns NONE for an unrecognized + * category name. + */ + static Category toCategory(const char *categoryName); + + /** + * Converts a category name such as "zero", "one", "two", "few", "many" + * or "other" to a category enum. Returns NONE for unrecognized + * category name. + */ + static Category toCategory(const UnicodeString &categoryName); + + /** + * Converts a category to a name. + * Passing NONE or CATEGORY_COUNT for category returns nullptr. + */ + static const char *getCategoryName(Category category); +}; + +/** + * A Map of plural categories to values. It maintains ownership of the + * values. + * + * Type T is the value type. T must provide the following: + * 1) Default constructor + * 2) Copy constructor + * 3) Assignment operator + * 4) Must extend UMemory + */ +template +class PluralMap : public PluralMapBase { +public: + /** + * Other category is maps to a copy of the default value. + */ + PluralMap() : fOtherVariant() { + initializeNew(); + } + + /** + * Other category is mapped to otherVariant. + */ + PluralMap(const T &otherVariant) : fOtherVariant(otherVariant) { + initializeNew(); + } + + PluralMap(const PluralMap &other) : fOtherVariant(other.fOtherVariant) { + fVariants[0] = &fOtherVariant; + for (int32_t i = 1; i < UPRV_LENGTHOF(fVariants); ++i) { + fVariants[i] = other.fVariants[i] ? + new T(*other.fVariants[i]) : nullptr; + } + } + + PluralMap &operator=(const PluralMap &other) { + if (this == &other) { + return *this; + } + for (int32_t i = 0; i < UPRV_LENGTHOF(fVariants); ++i) { + if (fVariants[i] != nullptr && other.fVariants[i] != nullptr) { + *fVariants[i] = *other.fVariants[i]; + } else if (fVariants[i] != nullptr) { + delete fVariants[i]; + fVariants[i] = nullptr; + } else if (other.fVariants[i] != nullptr) { + fVariants[i] = new T(*other.fVariants[i]); + } else { + // do nothing + } + } + return *this; + } + + ~PluralMap() { + for (int32_t i = 1; i < UPRV_LENGTHOF(fVariants); ++i) { + delete fVariants[i]; + } + } + + /** + * Removes all mappings and makes 'other' point to the default value. + */ + void clear() { + *fVariants[0] = T(); + for (int32_t i = 1; i < UPRV_LENGTHOF(fVariants); ++i) { + delete fVariants[i]; + fVariants[i] = nullptr; + } + } + + /** + * Iterates through the mappings in this instance, set index to NONE + * prior to using. Call next repeatedly to get the values until it + * returns nullptr. Each time next returns, caller may pass index + * to getCategoryName() to get the name of the plural category. + * When this function returns nullptr, index is CATEGORY_COUNT + */ + const T *next(Category &index) const { + int32_t idx = index; + ++idx; + for (; idx < UPRV_LENGTHOF(fVariants); ++idx) { + if (fVariants[idx] != nullptr) { + index = static_cast(idx); + return fVariants[idx]; + } + } + index = static_cast(idx); + return nullptr; + } + + /** + * non const version of next. + */ + T *nextMutable(Category &index) { + const T *result = next(index); + return const_cast(result); + } + + /** + * Returns the 'other' variant. + * Same as calling get(OTHER). + */ + const T &getOther() const { + return get(OTHER); + } + + /** + * Returns the value associated with a category. + * If no value found, or v is NONE or CATEGORY_COUNT, falls + * back to returning the value for the 'other' category. + */ + const T &get(Category v) const { + int32_t index = v; + if (index < 0 || index >= UPRV_LENGTHOF(fVariants) || fVariants[index] == nullptr) { + return *fVariants[0]; + } + return *fVariants[index]; + } + + /** + * Convenience routine to get the value by category name. Otherwise + * works just like get(Category). + */ + const T &get(const char *category) const { + return get(toCategory(category)); + } + + /** + * Convenience routine to get the value by category name as a + * UnicodeString. Otherwise works just like get(category). + */ + const T &get(const UnicodeString &category) const { + return get(toCategory(category)); + } + + /** + * Returns a pointer to the value associated with a category + * that caller can safely modify. If the value was defaulting to the 'other' + * variant because no explicit value was stored, this method creates a + * new value using the default constructor at the returned pointer. + * + * @param category the category with the value to change. + * @param status error returned here if index is NONE or CATEGORY_COUNT + * or memory could not be allocated, or any other error happens. + */ + T *getMutable( + Category category, + UErrorCode &status) { + return getMutable(category, nullptr, status); + } + + /** + * Convenience routine to get a mutable pointer to a value by category name. + * Otherwise works just like getMutable(Category, UErrorCode &). + * reports an error if the category name is invalid. + */ + T *getMutable( + const char *category, + UErrorCode &status) { + return getMutable(toCategory(category), nullptr, status); + } + + /** + * Just like getMutable(Category, UErrorCode &) but copies defaultValue to + * returned pointer if it was defaulting to the 'other' variant + * because no explicit value was stored. + */ + T *getMutableWithDefault( + Category category, + const T &defaultValue, + UErrorCode &status) { + return getMutable(category, &defaultValue, status); + } + + /** + * Returns true if this object equals rhs. + */ + UBool equals( + const PluralMap &rhs, + UBool (*eqFunc)(const T &, const T &)) const { + for (int32_t i = 0; i < UPRV_LENGTHOF(fVariants); ++i) { + if (fVariants[i] == rhs.fVariants[i]) { + continue; + } + if (fVariants[i] == nullptr || rhs.fVariants[i] == nullptr) { + return false; + } + if (!eqFunc(*fVariants[i], *rhs.fVariants[i])) { + return false; + } + } + return true; + } + +private: + T fOtherVariant; + T* fVariants[6]; + + T *getMutable( + Category category, + const T *defaultValue, + UErrorCode &status) { + if (U_FAILURE(status)) { + return nullptr; + } + int32_t index = category; + if (index < 0 || index >= UPRV_LENGTHOF(fVariants)) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + if (fVariants[index] == nullptr) { + fVariants[index] = defaultValue == nullptr ? + new T() : new T(*defaultValue); + } + if (!fVariants[index]) { + status = U_MEMORY_ALLOCATION_ERROR; + } + return fVariants[index]; + } + + void initializeNew() { + fVariants[0] = &fOtherVariant; + for (int32_t i = 1; i < UPRV_LENGTHOF(fVariants); ++i) { + fVariants[i] = nullptr; + } + } +}; + +U_NAMESPACE_END + +#endif diff --git a/intl/icu/source/common/propname.cpp b/intl/icu/source/common/propname.cpp new file mode 100644 index 0000000000..45062bfbd9 --- /dev/null +++ b/intl/icu/source/common/propname.cpp @@ -0,0 +1,334 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2002-2014, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* Author: Alan Liu +* Created: October 30 2002 +* Since: ICU 2.4 +* 2010nov19 Markus Scherer Rewrite for formatVersion 2. +********************************************************************** +*/ +#include "propname.h" +#include "unicode/uchar.h" +#include "unicode/udata.h" +#include "unicode/uscript.h" +#include "umutex.h" +#include "cmemory.h" +#include "cstring.h" +#include "uarrsort.h" +#include "uinvchar.h" + +#define INCLUDED_FROM_PROPNAME_CPP +#include "propname_data.h" + +U_CDECL_BEGIN + +/** + * Get the next non-ignorable ASCII character from a property name + * and lowercases it. + * @return ((advance count for the name)<<8)|character + */ +static inline int32_t +getASCIIPropertyNameChar(const char *name) { + int32_t i; + char c; + + /* Ignore delimiters '-', '_', and ASCII White_Space */ + for(i=0; + (c=name[i++])==0x2d || c==0x5f || + c==0x20 || (0x09<=c && c<=0x0d); + ) {} + + if(c!=0) { + return (i<<8)|(uint8_t)uprv_asciitolower((char)c); + } else { + return i<<8; + } +} + +/** + * Get the next non-ignorable EBCDIC character from a property name + * and lowercases it. + * @return ((advance count for the name)<<8)|character + */ +static inline int32_t +getEBCDICPropertyNameChar(const char *name) { + int32_t i; + char c; + + /* Ignore delimiters '-', '_', and EBCDIC White_Space */ + for(i=0; + (c=name[i++])==0x60 || c==0x6d || + c==0x40 || c==0x05 || c==0x15 || c==0x25 || c==0x0b || c==0x0c || c==0x0d; + ) {} + + if(c!=0) { + return (i<<8)|(uint8_t)uprv_ebcdictolower((char)c); + } else { + return i<<8; + } +} + +/** + * Unicode property names and property value names are compared "loosely". + * + * UCD.html 4.0.1 says: + * For all property names, property value names, and for property values for + * Enumerated, Binary, or Catalog properties, use the following + * loose matching rule: + * + * LM3. Ignore case, whitespace, underscore ('_'), and hyphens. + * + * This function does just that, for (char *) name strings. + * It is almost identical to ucnv_compareNames() but also ignores + * C0 White_Space characters (U+0009..U+000d, and U+0085 on EBCDIC). + * + * @internal + */ + +U_CAPI int32_t U_EXPORT2 +uprv_compareASCIIPropertyNames(const char *name1, const char *name2) { + int32_t rc, r1, r2; + + for(;;) { + r1=getASCIIPropertyNameChar(name1); + r2=getASCIIPropertyNameChar(name2); + + /* If we reach the ends of both strings then they match */ + if(((r1|r2)&0xff)==0) { + return 0; + } + + /* Compare the lowercased characters */ + if(r1!=r2) { + rc=(r1&0xff)-(r2&0xff); + if(rc!=0) { + return rc; + } + } + + name1+=r1>>8; + name2+=r2>>8; + } +} + +U_CAPI int32_t U_EXPORT2 +uprv_compareEBCDICPropertyNames(const char *name1, const char *name2) { + int32_t rc, r1, r2; + + for(;;) { + r1=getEBCDICPropertyNameChar(name1); + r2=getEBCDICPropertyNameChar(name2); + + /* If we reach the ends of both strings then they match */ + if(((r1|r2)&0xff)==0) { + return 0; + } + + /* Compare the lowercased characters */ + if(r1!=r2) { + rc=(r1&0xff)-(r2&0xff); + if(rc!=0) { + return rc; + } + } + + name1+=r1>>8; + name2+=r2>>8; + } +} + +U_CDECL_END + +U_NAMESPACE_BEGIN + +int32_t PropNameData::findProperty(int32_t property) { + int32_t i=1; // valueMaps index, initially after numRanges + for(int32_t numRanges=valueMaps[0]; numRanges>0; --numRanges) { + // Read and skip the start and limit of this range. + int32_t start=valueMaps[i]; + int32_t limit=valueMaps[i+1]; + i+=2; + if(property0; --numRanges) { + // Read and skip the start and limit of this range. + int32_t start=valueMaps[valueMapIndex]; + int32_t limit=valueMaps[valueMapIndex+1]; + valueMapIndex+=2; + if(value0; --nameIndex) { + nameGroup=uprv_strchr(nameGroup, 0)+1; + } + if(*nameGroup==0) { + return nullptr; // no name (Property[Value]Aliases.txt has "n/a") + } + return nameGroup; +} + +UBool PropNameData::containsName(BytesTrie &trie, const char *name) { + if(name==nullptr) { + return false; + } + UStringTrieResult result=USTRINGTRIE_NO_VALUE; + char c; + while((c=*name++)!=0) { + c=uprv_invCharToLowercaseAscii(c); + // Ignore delimiters '-', '_', and ASCII White_Space. + if(c==0x2d || c==0x5f || c==0x20 || (0x09<=c && c<=0x0d)) { + continue; + } + if(!USTRINGTRIE_HAS_NEXT(result)) { + return false; + } + result=trie.next((uint8_t)c); + } + return USTRINGTRIE_HAS_VALUE(result); +} + +const char *PropNameData::getPropertyName(int32_t property, int32_t nameChoice) { + int32_t valueMapIndex=findProperty(property); + if(valueMapIndex==0) { + return nullptr; // Not a known property. + } + return getName(nameGroups+valueMaps[valueMapIndex], nameChoice); +} + +const char *PropNameData::getPropertyValueName(int32_t property, int32_t value, int32_t nameChoice) { + int32_t valueMapIndex=findProperty(property); + if(valueMapIndex==0) { + return nullptr; // Not a known property. + } + int32_t nameGroupOffset=findPropertyValueNameGroup(valueMaps[valueMapIndex+1], value); + if(nameGroupOffset==0) { + return nullptr; + } + return getName(nameGroups+nameGroupOffset, nameChoice); +} + +int32_t PropNameData::getPropertyOrValueEnum(int32_t bytesTrieOffset, const char *alias) { + BytesTrie trie(bytesTries+bytesTrieOffset); + if(containsName(trie, alias)) { + return trie.getValue(); + } else { + return UCHAR_INVALID_CODE; + } +} + +int32_t PropNameData::getPropertyEnum(const char *alias) { + return getPropertyOrValueEnum(0, alias); +} + +int32_t PropNameData::getPropertyValueEnum(int32_t property, const char *alias) { + int32_t valueMapIndex=findProperty(property); + if(valueMapIndex==0) { + return UCHAR_INVALID_CODE; // Not a known property. + } + valueMapIndex=valueMaps[valueMapIndex+1]; + if(valueMapIndex==0) { + return UCHAR_INVALID_CODE; // The property does not have named values. + } + // valueMapIndex is the start of the property's valueMap, + // where the first word is the BytesTrie offset. + return getPropertyOrValueEnum(valueMaps[valueMapIndex], alias); +} +U_NAMESPACE_END + +//---------------------------------------------------------------------- +// Public API implementation + +U_CAPI const char* U_EXPORT2 +u_getPropertyName(UProperty property, + UPropertyNameChoice nameChoice) UPRV_NO_SANITIZE_UNDEFINED { + // The nameChoice is really an integer with a couple of named constants. + // Unicode allows for names other than short and long ones. + // If present, these will be returned for U_LONG_PROPERTY_NAME + i, where i=1, 2,... + U_NAMESPACE_USE + return PropNameData::getPropertyName(property, nameChoice); +} + +U_CAPI UProperty U_EXPORT2 +u_getPropertyEnum(const char* alias) { + U_NAMESPACE_USE + return (UProperty)PropNameData::getPropertyEnum(alias); +} + +U_CAPI const char* U_EXPORT2 +u_getPropertyValueName(UProperty property, + int32_t value, + UPropertyNameChoice nameChoice) UPRV_NO_SANITIZE_UNDEFINED { + // The nameChoice is really an integer with a couple of named constants. + // Unicode allows for names other than short and long ones. + // If present, these will be returned for U_LONG_PROPERTY_NAME + i, where i=1, 2,... + U_NAMESPACE_USE + return PropNameData::getPropertyValueName(property, value, nameChoice); +} + +U_CAPI int32_t U_EXPORT2 +u_getPropertyValueEnum(UProperty property, + const char* alias) { + U_NAMESPACE_USE + return PropNameData::getPropertyValueEnum(property, alias); +} + +U_CAPI const char* U_EXPORT2 +uscript_getName(UScriptCode scriptCode){ + return u_getPropertyValueName(UCHAR_SCRIPT, scriptCode, + U_LONG_PROPERTY_NAME); +} + +U_CAPI const char* U_EXPORT2 +uscript_getShortName(UScriptCode scriptCode){ + return u_getPropertyValueName(UCHAR_SCRIPT, scriptCode, + U_SHORT_PROPERTY_NAME); +} diff --git a/intl/icu/source/common/propname.h b/intl/icu/source/common/propname.h new file mode 100644 index 0000000000..1a8ced5b87 --- /dev/null +++ b/intl/icu/source/common/propname.h @@ -0,0 +1,212 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2002-2011, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* Author: Alan Liu +* Created: October 30 2002 +* Since: ICU 2.4 +* 2010nov19 Markus Scherer Rewrite for formatVersion 2. +********************************************************************** +*/ +#ifndef PROPNAME_H +#define PROPNAME_H + +#include "unicode/utypes.h" +#include "unicode/bytestrie.h" +#include "unicode/uchar.h" +#include "udataswp.h" +#include "uprops.h" + +/* + * This header defines the in-memory layout of the property names data + * structure representing the UCD data files PropertyAliases.txt and + * PropertyValueAliases.txt. It is used by: + * propname.cpp - reads data + * genpname - creates data + */ + +/* low-level char * property name comparison -------------------------------- */ + +U_CDECL_BEGIN + +/** + * \var uprv_comparePropertyNames + * Unicode property names and property value names are compared "loosely". + * + * UCD.html 4.0.1 says: + * For all property names, property value names, and for property values for + * Enumerated, Binary, or Catalog properties, use the following + * loose matching rule: + * + * LM3. Ignore case, whitespace, underscore ('_'), and hyphens. + * + * This function does just that, for (char *) name strings. + * It is almost identical to ucnv_compareNames() but also ignores + * C0 White_Space characters (U+0009..U+000d, and U+0085 on EBCDIC). + * + * @internal + */ + +U_CAPI int32_t U_EXPORT2 +uprv_compareASCIIPropertyNames(const char *name1, const char *name2); + +U_CAPI int32_t U_EXPORT2 +uprv_compareEBCDICPropertyNames(const char *name1, const char *name2); + +#if U_CHARSET_FAMILY==U_ASCII_FAMILY +# define uprv_comparePropertyNames uprv_compareASCIIPropertyNames +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY +# define uprv_comparePropertyNames uprv_compareEBCDICPropertyNames +#else +# error U_CHARSET_FAMILY is not valid +#endif + +U_CDECL_END + +/* UDataMemory structure and signatures ------------------------------------- */ + +#define PNAME_DATA_NAME "pnames" +#define PNAME_DATA_TYPE "icu" + +/* Fields in UDataInfo: */ + +/* PNAME_SIG[] is encoded as numeric literals for compatibility with the HP compiler */ +#define PNAME_SIG_0 ((uint8_t)0x70) /* p */ +#define PNAME_SIG_1 ((uint8_t)0x6E) /* n */ +#define PNAME_SIG_2 ((uint8_t)0x61) /* a */ +#define PNAME_SIG_3 ((uint8_t)0x6D) /* m */ + +U_NAMESPACE_BEGIN + +class PropNameData { +public: + enum { + // Byte offsets from the start of the data, after the generic header. + IX_VALUE_MAPS_OFFSET, + IX_BYTE_TRIES_OFFSET, + IX_NAME_GROUPS_OFFSET, + IX_RESERVED3_OFFSET, + IX_RESERVED4_OFFSET, + IX_TOTAL_SIZE, + + // Other values. + IX_MAX_NAME_LENGTH, + IX_RESERVED7, + IX_COUNT + }; + + static const char *getPropertyName(int32_t property, int32_t nameChoice); + static const char *getPropertyValueName(int32_t property, int32_t value, int32_t nameChoice); + + static int32_t getPropertyEnum(const char *alias); + static int32_t getPropertyValueEnum(int32_t property, const char *alias); + +private: + static int32_t findProperty(int32_t property); + static int32_t findPropertyValueNameGroup(int32_t valueMapIndex, int32_t value); + static const char *getName(const char *nameGroup, int32_t nameIndex); + static UBool containsName(BytesTrie &trie, const char *name); + + static int32_t getPropertyOrValueEnum(int32_t bytesTrieOffset, const char *alias); + + static const int32_t indexes[]; + static const int32_t valueMaps[]; + static const uint8_t bytesTries[]; + static const char nameGroups[]; +}; + +/* + * pnames.icu formatVersion 2 + * + * formatVersion 2 is new in ICU 4.8. + * In ICU 4.8, the pnames.icu data file is used only in ICU4J. + * ICU4C 4.8 has the same data structures hardcoded in source/common/propname_data.h. + * + * For documentation of pnames.icu formatVersion 1 see ICU4C 4.6 (2010-dec-01) + * or earlier versions of this header file (source/common/propname.h). + * + * The pnames.icu begins with the standard ICU DataHeader/UDataInfo. + * After that: + * + * int32_t indexes[8]; + * + * (See the PropNameData::IX_... constants.) + * + * The first 6 indexes are byte offsets from the beginning of the data + * (beginning of indexes[]) to following structures. + * The length of each structure is the difference between its offset + * and the next one. + * All offsets are filled in: Where there is no data between two offsets, + * those two offsets are the same. + * The last offset (indexes[PropNameData::IX_TOTAL_SIZE]) indicates the + * total number of bytes in the file. (Not counting the standard headers.) + * + * The sixth index (indexes[PropNameData::IX_MAX_NAME_LENGTH]) has the + * maximum length of any Unicode property (or property value) alias. + * (Without normalization, that is, including underscores etc.) + * + * int32_t valueMaps[]; + * + * The valueMaps[] begins with a map from UProperty enums to properties, + * followed by the per-property value maps from property values to names, + * for those properties that have named values. + * (Binary & enumerated, plus General_Category_Mask.) + * + * valueMaps[0] contains the number of UProperty enum ranges. + * For each range: + * int32_t start, limit -- first and last+1 UProperty enum of a dense range + * Followed by (limit-start) pairs of + * int32_t nameGroupOffset; + * Offset into nameGroups[] for the property's names/aliases. + * int32_t valueMapIndex; + * Offset of the property's value map in the valueMaps[] array. + * If the valueMapIndex is 0, then the property does not have named values. + * + * For each property's value map: + * int32_t bytesTrieOffset; -- Offset into bytesTries[] for name->value mapping. + * int32_t numRanges; + * If numRanges is in the range 1..15, then that many ranges of values follow. + * Per range: + * int32_t start, limit -- first and last+1 UProperty enum of a range + * Followed by (limit-start) entries of + * int32_t nameGroupOffset; + * Offset into nameGroups[] for the property value's names/aliases. + * If the nameGroupOffset is 0, then this is not a named value for this property. + * (That is, the ranges need not be dense.) + * If numRanges is >=0x10, then (numRanges-0x10) sorted values + * and then (numRanges-0x10) corresponding nameGroupOffsets follow. + * Values are sorted as signed integers. + * In this case, the set of values is dense; no nameGroupOffset will be 0. + * + * For both properties and property values, ranges are sorted by their start/limit values. + * + * uint8_t bytesTries[]; + * + * This is a sequence of BytesTrie structures, byte-serialized tries for + * mapping from names/aliases to values. + * The first one maps from property names/aliases to UProperty enum constants. + * The following ones are indexed by property value map bytesTrieOffsets + * for mapping each property's names/aliases to their property values. + * + * char nameGroups[]; + * + * This is a sequence of property name groups. + * Each group is a list of names/aliases (invariant-character strings) for + * one property or property value, in the order of UCharNameChoice. + * The first byte of each group is the number of names in the group. + * It is followed by that many NUL-terminated strings. + * The first string is for the short name; if there is no short name, + * then the first string is empty. + * The second string is the long name. Further strings are additional aliases. + * + * The first name group is for a property rather than a property value, + * so that a nameGroupOffset of 0 can be used to indicate "no value" + * in a property's sparse value ranges. + */ + +U_NAMESPACE_END + +#endif diff --git a/intl/icu/source/common/propname_data.h b/intl/icu/source/common/propname_data.h new file mode 100644 index 0000000000..1e247874b6 --- /dev/null +++ b/intl/icu/source/common/propname_data.h @@ -0,0 +1,1984 @@ +// Copyright (C) 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// Copyright (C) 1999-2016, International Business Machines +// Corporation and others. All Rights Reserved. +// +// file name: propname_data.h +// +// machine-generated by: icu/tools/unicode/c/genprops/pnamesbuilder.cpp + +#ifdef INCLUDED_FROM_PROPNAME_CPP + +U_NAMESPACE_BEGIN + +const int32_t PropNameData::indexes[8]={0x20,0x1660,0x5294,0xacd0,0xacd0,0xacd0,0x31,0}; + +const int32_t PropNameData::valueMaps[1424]={ +6,0,0x48,0,0xf1,0x368,0xf1,0x37e,0xf1,0x393,0xf1,0x3a9,0xf1,0x3b4,0xf1,0x3d5, +0xf1,0x3e5,0xf1,0x3f4,0xf1,0x402,0xf1,0x426,0xf1,0x43d,0xf1,0x455,0xf1,0x46c,0xf1,0x47b, +0xf1,0x48a,0xf1,0x49b,0xf1,0x4a9,0xf1,0x4bb,0xf1,0x4d5,0xf1,0x4f0,0xf1,0x505,0xf1,0x522, +0xf1,0x533,0xf1,0x53e,0xf1,0x55d,0xf1,0x573,0xf1,0x584,0xf1,0x594,0xf1,0x5af,0xf1,0x5c8, +0xf1,0x5d9,0xf1,0x5f3,0xf1,0x606,0xf1,0x616,0xf1,0x630,0xf1,0x649,0xf1,0x660,0xf1,0x674, +0xf1,0x68a,0xf1,0x69e,0xf1,0x6b4,0xf1,0x6ce,0xf1,0x6e6,0xf1,0x702,0xf1,0x70a,0xf1,0x712, +0xf1,0x71a,0xf1,0x722,0xf1,0x72b,0xf1,0x738,0xf1,0x74b,0xf1,0x768,0xf1,0x785,0xf1,0x7a2, +0xf1,0x7c0,0xf1,0x7de,0xf1,0x802,0xf1,0x80f,0xf1,0x829,0xf1,0x83e,0xf1,0x859,0xf1,0x870, +0xf1,0x887,0xf1,0x8a9,0xf1,0x8c8,0xf1,0x8e1,0xf1,0x90e,0xf1,0x947,0xf1,0x978,0xf1,0x9a7, +0xf1,0x9d6,0xf1,0x1000,0x1019,0x9eb,0x16d,0xc0b,0x188,0x3279,0xf7,0x3298,0x2d4,0x33d6,0x2ea,0x3430, +0x2f4,0x368d,0x316,0x3fb8,0x382,0x4028,0x38c,0x42c2,0x3bb,0x4300,0x3c3,0x4e45,0x48f,0x4ec3,0x499,0x4ee8, +0x49f,0x4f02,0x4a5,0x4f23,0x4ac,0x4f3d,0xf7,0x4f62,0xf7,0x4f88,0x4b3,0x5032,0x4c9,0x50ab,0x4dc,0x515d, +0x4f7,0x5194,0x4fe,0x5374,0x512,0x57f4,0x53a,0x2000,0x2001,0x5853,0x542,0x3000,0x3001,0x58df,0,0x4000, +0x400e,0x58f1,0,0x58fa,0,0x5914,0,0x5925,0,0x5936,0,0x594c,0,0x5955,0,0x5972, +0,0x5990,0,0x59ae,0,0x59cc,0,0x59e2,0,0x59f6,0,0x5a0c,0,0x7000,0x7001,0x5a25, +0,0x844,0x12,0,1,0x12,0x20,0x862,0x4a,0,1,6,7,8,9,0xa, +0xb,0xc,0xd,0xe,0xf,0x10,0x11,0x12,0x13,0x14,0x15,0x16,0x17,0x18,0x19,0x1a, +0x1b,0x1c,0x1d,0x1e,0x1f,0x20,0x21,0x22,0x23,0x24,0x54,0x5b,0x67,0x6b,0x76,0x7a, +0x81,0x82,0x84,0x85,0xc8,0xca,0xd6,0xd8,0xda,0xdc,0xde,0xe0,0xe2,0xe4,0xe6,0xe8, +0xe9,0xea,0xf0,0x2e,0x40,0x4c,0x5e,0x68,0x79,0x84,0x91,0x9e,0xab,0xb8,0xc5,0xd2, +0xdf,0xec,0xf9,0x106,0x113,0x120,0x12d,0x13a,0x147,0x154,0x161,0x16e,0x17b,0x188,0x195,0x1a2, +0x1af,0x1bc,0x1c9,0x1d6,0x1e3,0x1f0,0x1fd,0x20c,0x21b,0x22a,0x239,0x248,0x257,0x266,0x275,0x28f, +0x2a3,0x2b7,0x2d2,0x2e1,0x2ea,0x2fa,0x302,0x30b,0x31a,0x323,0x333,0x344,0x355,0xa03,1,0, +0x17,0x9fa,0xa0b,0xa1c,0xa30,0xa47,0xa5f,0xa71,0xa86,0xa9d,0xab2,0xac2,0xad4,0xaf1,0xb0d,0xb1f, +0xb3c,0xb58,0xb74,0xb89,0xb9e,0xbb8,0xbd3,0xbee,0xba5,1,0,0x148,0xc16,0xc23,0xc36,0xc5e, +0xc7c,0xc9a,0xcb2,0xcdd,0xd07,0xd1f,0xd32,0xd45,0xd54,0xd63,0xd72,0xd81,0xd98,0xda9,0xdbc,0xdcf, +0xddc,0xde9,0xdf8,0xe09,0xe1e,0xe2f,0xe3a,0xe43,0xe54,0xe65,0xe78,0xe8a,0xe9d,0xeb0,0xeef,0xefc, +0xf09,0xf16,0xf2b,0xf5b,0xf75,0xf96,0xfc1,0xfe4,0x1042,0x1069,0x1084,0x1093,0x10ba,0x10e2,0x1105,0x1128, +0x1152,0x116b,0x118a,0x11ad,0x11d1,0x11e4,0x11fe,0x1228,0x1240,0x1268,0x1291,0x12a4,0x12b7,0x12ca,0x12f1,0x1300, +0x1320,0x134e,0x136c,0x139a,0x13b6,0x13d1,0x13ea,0x1403,0x1424,0x1454,0x1473,0x1495,0x14c9,0x14f6,0x153b,0x155c, +0x1586,0x15a7,0x15d0,0x15e3,0x1616,0x162d,0x163c,0x164d,0x1678,0x168f,0x16c0,0x16ee,0x1731,0x173c,0x1775,0x1786, +0x1797,0x17a4,0x17b7,0x17f1,0x1815,0x1839,0x1873,0x18ab,0x18d6,0x18ee,0x191a,0x1946,0x1953,0x1962,0x197f,0x19a1, +0x19cf,0x19ef,0x1a16,0x1a3d,0x1a5c,0x1a6f,0x1a80,0x1a91,0x1ab6,0x1adb,0x1b02,0x1b36,0x1b63,0x1b81,0x1b94,0x1bad, +0x1be6,0x1bf5,0x1c15,0x1c37,0x1c59,0x1c70,0x1c87,0x1cb4,0x1ccd,0x1ce6,0x1d17,0x1d41,0x1d5c,0x1d6f,0x1d8e,0x1d97, +0x1daa,0x1dc8,0x1de6,0x1df9,0x1e10,0x1e25,0x1e5a,0x1e7e,0x1e93,0x1ea2,0x1eb5,0x1ed9,0x1ee2,0x1f06,0x1f1d,0x1f30, +0x1f3f,0x1f4a,0x1f6b,0x1f83,0x1f92,0x1fa1,0x1fb0,0x1fc7,0x1fdc,0x1ff1,0x202a,0x203d,0x2059,0x2064,0x2071,0x209f, +0x20c3,0x20e6,0x20f9,0x211b,0x212e,0x2149,0x216c,0x218f,0x21b4,0x21c5,0x21f4,0x2221,0x2238,0x2253,0x2262,0x228d, +0x22c5,0x22ff,0x232d,0x233e,0x234b,0x236f,0x237e,0x239a,0x23b4,0x23d1,0x2409,0x241e,0x244b,0x246a,0x2498,0x24b8, +0x24ec,0x24fb,0x2525,0x2548,0x2573,0x257e,0x258f,0x25aa,0x25ce,0x25db,0x25f0,0x2617,0x2642,0x2679,0x268c,0x269d, +0x26cd,0x26de,0x26ed,0x2702,0x2720,0x2733,0x2746,0x275d,0x277a,0x2785,0x278e,0x27b0,0x27c5,0x27ea,0x2801,0x282a, +0x2845,0x285a,0x2873,0x2894,0x28c9,0x28da,0x290b,0x292f,0x2940,0x2959,0x2964,0x2991,0x29b3,0x29e1,0x2a14,0x2a23, +0x2a34,0x2a51,0x2a93,0x2aba,0x2ac7,0x2adc,0x2b00,0x2b26,0x2b5f,0x2b70,0x2b94,0x2b9f,0x2bac,0x2bbb,0x2be0,0x2c0e, +0x2c2a,0x2c47,0x2c54,0x2c65,0x2c83,0x2ca6,0x2cc3,0x2cd0,0x2cf0,0x2d0d,0x2d2e,0x2d57,0x2d68,0x2d87,0x2da0,0x2db9, +0x2dca,0x2e13,0x2e24,0x2e3d,0x2e6c,0x2e99,0x2ebe,0x2f00,0x2f1c,0x2f2b,0x2f42,0x2f70,0x2f89,0x2fb2,0x2fcc,0x3007, +0x3025,0x3034,0x3054,0x306f,0x3093,0x30af,0x30cd,0x30eb,0x3102,0x3111,0x311c,0x3159,0x316c,0x3196,0x31b6,0x31e4, +0x3208,0x3230,0x3255,0x3260,0x1fa9,1,0,0x12,0x32af,0x32bf,0x32d2,0x32e2,0x32f2,0x3301,0x3311,0x3323, +0x3336,0x3348,0x3358,0x3368,0x3377,0x3386,0x3396,0x33a3,0x33b2,0x33c6,0x2067,1,0,6,0x33eb,0x33f6, +0x3403,0x3410,0x341d,0x3428,0x20ab,1,0,0x1e,0x3445,0x3454,0x3469,0x347e,0x3493,0x34a7,0x34b8,0x34cc, +0x34df,0x34f0,0x3509,0x351b,0x352c,0x3540,0x3553,0x356b,0x357d,0x3588,0x3598,0x35a6,0x35bb,0x35d0,0x35e6,0x3600, +0x3616,0x3626,0x363a,0x364e,0x365f,0x3677,0x22d6,1,0,0x68,0x369f,0x36c2,0x36cb,0x36d8,0x36e3,0x36ec, +0x36f7,0x3700,0x3719,0x371e,0x3727,0x3744,0x374d,0x375a,0x3763,0x3787,0x378e,0x3797,0x37aa,0x37b5,0x37be,0x37c9, +0x37e2,0x37eb,0x37fa,0x3805,0x380e,0x3819,0x3822,0x3829,0x3832,0x383d,0x3846,0x385f,0x3868,0x3875,0x3880,0x3891, +0x389c,0x38b1,0x38c8,0x38d1,0x38da,0x38f3,0x38fe,0x3907,0x3910,0x3927,0x3944,0x394f,0x3960,0x396b,0x3972,0x397f, +0x398c,0x39b9,0x39ce,0x39d7,0x39f2,0x3a15,0x3a36,0x3a57,0x3a7c,0x3aa3,0x3ac4,0x3ae7,0x3b08,0x3b2f,0x3b50,0x3b75, +0x3b94,0x3bb3,0x3bd2,0x3bef,0x3c10,0x3c31,0x3c54,0x3c79,0x3c98,0x3cb7,0x3cd8,0x3cff,0x3d24,0x3d43,0x3d64,0x3d87, +0x3da2,0x3dbb,0x3dd6,0x3def,0x3e0c,0x3e27,0x3e44,0x3e63,0x3e80,0x3e9d,0x3ebc,0x3ed9,0x3ef4,0x3f11,0x3f2e,0x3f61, +0x3f88,0x3f9b,0x2639,1,0,6,0x3fc9,0x3fd8,0x3fe8,0x3ff8,0x4008,0x4019,0x2697,1,0,0x2b, +0x4037,0x4043,0x4051,0x4060,0x406f,0x407f,0x4090,0x40a4,0x40b9,0x40cf,0x40e2,0x40f6,0x4106,0x410f,0x411a,0x412a, +0x4146,0x4158,0x4166,0x4175,0x4181,0x4196,0x41aa,0x41bd,0x41cb,0x41df,0x41ed,0x41f7,0x4209,0x4215,0x4223,0x4233, +0x423a,0x4241,0x4248,0x424f,0x4256,0x426c,0x428d,0x870,0x429f,0x42aa,0x42b9,0x28f0,1,0,4,0x42d3, +0x42de,0x42ea,0x42f4,0x2916,1,0,0xc8,0x430b,0x4318,0x432d,0x433a,0x4349,0x4357,0x4366,0x4375,0x4387, +0x4396,0x43a4,0x43b5,0x43c4,0x43d3,0x43e0,0x43ec,0x43fb,0x440a,0x4414,0x4421,0x442e,0x443d,0x444b,0x445a,0x4466, +0x4470,0x447c,0x448c,0x449c,0x44aa,0x44b6,0x44c7,0x44d3,0x44df,0x44ed,0x44fa,0x4506,0x4513,0xe2f,0x4520,0x452e, +0x4548,0x4551,0x455f,0x456d,0x4579,0x4588,0x4596,0x45a4,0x45b0,0x45bf,0x45cd,0x45db,0x45e8,0x45f7,0x4612,0x4621, +0x4632,0x4643,0x4656,0x4668,0x4677,0x4689,0x4698,0x46a4,0x46af,0x1f3f,0x46bc,0x46c7,0x46d2,0x46dd,0x46e8,0x4703, +0x470e,0x4719,0x4724,0x4737,0x474b,0x4756,0x4765,0x4774,0x477f,0x478a,0x4797,0x47a6,0x47b4,0x47bf,0x47da,0x47e4, +0x47f5,0x4806,0x4815,0x4826,0x4831,0x483c,0x4847,0x4852,0x485d,0x4868,0x4873,0x487d,0x4888,0x4898,0x48a3,0x48b1, +0x48be,0x48c9,0x48d8,0x48e5,0x48f2,0x4901,0x490e,0x491f,0x4931,0x4941,0x494c,0x495f,0x4976,0x4984,0x4991,0x499c, +0x49a9,0x49ba,0x49d6,0x49ec,0x49f7,0x4a14,0x4a24,0x4a33,0x4a3e,0x4a49,0x2059,0x4a55,0x4a60,0x4a78,0x4a88,0x4a97, +0x4aa5,0x4ab3,0x4abe,0x4ac9,0x4add,0x4af4,0x4b0c,0x4b1c,0x4b2c,0x4b3c,0x4b4e,0x4b59,0x4b64,0x4b6e,0x4b7a,0x4b88, +0x4b9b,0x4ba7,0x4bb4,0x4bbf,0x4bdb,0x4be8,0x4bf6,0x4c0f,0x2959,0x4c1e,0x277a,0x4c2b,0x4c39,0x4c4b,0x4c59,0x4c65, +0x4c75,0x2b94,0x4c83,0x4c8f,0x4c9a,0x4ca5,0x4cb0,0x4cc4,0x4cd2,0x4ce9,0x4cf5,0x4d09,0x4d17,0x4d29,0x4d3f,0x4d4d, +0x4d5f,0x4d6d,0x4d8a,0x4d9c,0x4da9,0x4dba,0x4dcc,0x4de6,0x4df3,0x4e06,0x4e17,0x3111,0x4e24,0x3255,0x4e33,0x3370, +1,0,6,0x4e5f,0x4e72,0x4e82,0x4e90,0x4ea1,0x4eb1,0x33cc,0x12,0,1,0x4edb,0x4ee1,0x33d9, +0x12,0,1,0x4edb,0x4ee1,0x33e6,1,0,3,0x4edb,0x4ee1,0x4f1a,0x33fc,1,0,3, +0x4edb,0x4ee1,0x4f1a,0x3412,1,0,0x12,0x4fa4,0x4fae,0x4fba,0x4fc1,0x4fcc,0x4fd1,0x4fd8,0x4fdf,0x4fe8, +0x4fed,0x4ff2,0x5002,0x870,0x429f,0x500e,0x42aa,0x501e,0x42b9,0x34bb,1,0,0xf,0x4fa4,0x5045,0x504f, +0x5059,0x5064,0x4175,0x506e,0x507a,0x5082,0x5089,0x5093,0x4fba,0x4fc1,0x4fd1,0x509d,0x3542,1,0,0x17, +0x4fa4,0x50ba,0x5059,0x50c6,0x50d3,0x50e1,0x4175,0x50ec,0x4fba,0x50fd,0x4fd1,0x510c,0x511a,0x870,0x428d,0x5126, +0x5137,0x429f,0x500e,0x42aa,0x501e,0x42b9,0x5148,0x365f,1,0,3,0x517b,0x5183,0x518b,0x3678,1, +0,0x10,0x51b4,0x51bb,0x51ca,0x51eb,0x520e,0x5219,0x5238,0x524f,0x525c,0x5265,0x5284,0x52b7,0x52d2,0x5301, +0x531e,0x5343,0x3711,1,0,0x24,0x5392,0x539f,0x53b2,0x53bf,0x53ec,0x5411,0x5426,0x5445,0x5466,0x5493, +0x54cc,0x54ef,0x5512,0x553f,0x5574,0x559b,0x55c4,0x55fb,0x562a,0x564b,0x5670,0x567f,0x56a2,0x56b9,0x56c6,0x56d5, +0x56f2,0x570b,0x572e,0x5753,0x576c,0x5781,0x5790,0x57a1,0x57ae,0x57cf,0x38e1,1,0,4,0x580d,0x5818, +0x5830,0x5848,0x391d,0x36,1,2,4,8,0xe,0x10,0x20,0x3e,0x40,0x80,0x100,0x1c0, +0x200,0x400,0x800,0xe00,0x1000,0x2000,0x4000,0x7000,0x8000,0x10000,0x20000,0x40000,0x78001,0x80000,0x100000,0x200000, +0x400000,0x800000,0x1000000,0x2000000,0x4000000,0x8000000,0xf000000,0x10000000,0x20000000,0x30f80000,0x3445,0x3454,0x3469,0x347e,0x5881,0x3493, +0x34a7,0x5877,0x34b8,0x34cc,0x34df,0x5892,0x34f0,0x3509,0x351b,0x58a9,0x352c,0x3540,0x3553,0x58d2,0x356b,0x357d, +0x3588,0x3598,0x586e,0x35a6,0x35bb,0x35d0,0x35e6,0x3600,0x3616,0x3626,0x363a,0x364e,0x58c8,0x365f,0x3677,0x58b3 +}; + +const uint8_t PropNameData::bytesTries[15412]={ +0,0x15,0x6d,0xc3,0xc7,0x73,0xc2,0x12,0x76,0x7a,0x76,0x6a,0x77,0xa2,0x52,0x78, +1,0x64,0x50,0x69,0x10,0x64,1,0x63,0x30,0x73,0x62,0x13,0x74,0x61,0x72,0x74, +0x63,0x60,0x16,0x6f,0x6e,0x74,0x69,0x6e,0x75,0x65,0x61,0x13,0x69,0x67,0x69,0x74, +0x81,3,0x61,0x2e,0x65,0x4c,0x6f,0xc3,0x18,0x73,0x69,0x1e,0x72,0x69,0x61,0x74, +0x69,0x6f,0x6e,0x73,0x65,0x6c,0x65,0x63,0x74,0x6f,0x72,0x69,0x10,0x72,0x1f,0x74, +0x69,0x63,0x61,0x6c,0x6f,0x72,0x69,0x65,0x6e,0x74,0x61,0x74,0x69,0x6f,0x6e,0xc3, +0x18,3,0x62,0xc3,0x14,0x68,0x32,0x6f,0x42,0x73,0x13,0x70,0x61,0x63,0x65,0x5f, +0x17,0x69,0x74,0x65,0x73,0x70,0x61,0x63,0x65,0x5f,0x16,0x72,0x64,0x62,0x72,0x65, +0x61,0x6b,0xc3,0x14,0x73,0xa2,0x49,0x74,0xa4,0x3b,0x75,3,0x63,0xd9,0x40,0xc, +0x69,0x52,0x6e,0x58,0x70,0x12,0x70,0x65,0x72,0x5c,0x13,0x63,0x61,0x73,0x65,0x5c, +0x16,0x6d,0x61,0x70,0x70,0x69,0x6e,0x67,0xd9,0x40,0xc,0x12,0x64,0x65,0x6f,0x5b, +0x10,0x69,1,0x63,0x3e,0x66,0x1b,0x69,0x65,0x64,0x69,0x64,0x65,0x6f,0x67,0x72, +0x61,0x70,0x68,0x5b,0x17,0x6f,0x64,0x65,0x31,0x6e,0x61,0x6d,0x65,0xd9,0x40,0xb, +0xa,0x69,0x84,0x70,0x19,0x70,0x30,0x74,0x36,0x75,0x10,0x63,0xd9,0x40,9,0x12, +0x61,0x63,0x65,0x5f,1,0x63,0xd9,0x40,8,0x65,0x11,0x72,0x6d,0x67,0x69,0x3c, +0x6c,0xa2,0x5f,0x6f,0x17,0x66,0x74,0x64,0x6f,0x74,0x74,0x65,0x64,0x57,0x13,0x6d, +0x70,0x6c,0x65,3,0x63,0x50,0x6c,0x68,0x74,0x8a,0x75,0x1e,0x70,0x70,0x65,0x72, +0x63,0x61,0x73,0x65,0x6d,0x61,0x70,0x70,0x69,0x6e,0x67,0xd9,0x40,9,0x19,0x61, +0x73,0x65,0x66,0x6f,0x6c,0x64,0x69,0x6e,0x67,0xd9,0x40,6,0x1e,0x6f,0x77,0x65, +0x72,0x63,0x61,0x73,0x65,0x6d,0x61,0x70,0x70,0x69,0x6e,0x67,0xd9,0x40,7,0x1e, +0x69,0x74,0x6c,0x65,0x63,0x61,0x73,0x65,0x6d,0x61,0x70,0x70,0x69,0x6e,0x67,0xd9, +0x40,8,0x10,0x63,0xd9,0x40,7,0x62,0xc3,0x13,0x63,0x34,0x64,0x57,0x65,0x6e, +0x66,0x10,0x63,0xd9,0x40,6,0xc2,0xa,2,0x66,0xd9,0x40,6,0x72,0x28,0x78, +0xd9,0x70,0,0x12,0x69,0x70,0x74,0xc2,0xa,0x19,0x65,0x78,0x74,0x65,0x6e,0x73, +0x69,0x6f,0x6e,0x73,0xd9,0x70,0,1,0x67,0x6a,0x6e,1,0x73,0x54,0x74,0x13, +0x65,0x6e,0x63,0x65,1,0x62,0x34,0x74,0x16,0x65,0x72,0x6d,0x69,0x6e,0x61,0x6c, +0x67,0x13,0x72,0x65,0x61,0x6b,0xc3,0x13,0x14,0x69,0x74,0x69,0x76,0x65,0x65,1, +0x6d,0x2e,0x73,0x13,0x74,0x61,0x72,0x74,0x73,0x19,0x65,0x6e,0x74,0x73,0x74,0x61, +0x72,0x74,0x65,0x72,0x73,3,0x63,0x66,0x65,0x72,0x69,0x98,0x72,0x19,0x61,0x69, +0x6c,0x63,0x61,0x6e,0x6f,0x6e,0x69,0x63,0x1f,0x61,0x6c,0x63,0x6f,0x6d,0x62,0x69, +0x6e,0x69,0x6e,0x67,0x63,0x6c,0x61,0x73,0x73,0xc3,0x11,0xd8,0x40,0xa,0x11,0x63, +0x63,0xc3,0x11,0x11,0x72,0x6d,0x58,0x1e,0x69,0x6e,0x61,0x6c,0x70,0x75,0x6e,0x63, +0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0x59,0x1d,0x74,0x6c,0x65,0x63,0x61,0x73,0x65, +0x6d,0x61,0x70,0x70,0x69,0x6e,0x67,0xd9,0x40,0xa,0x6d,0xa2,0x76,0x6e,0xa2,0x78, +0x70,0xa4,0x3e,0x71,0xa4,0x90,0x72,3,0x61,0x2c,0x65,0x36,0x67,0x54,0x69,0x9d, +0x14,0x64,0x69,0x63,0x61,0x6c,0x55,0x1e,0x67,0x69,0x6f,0x6e,0x61,0x6c,0x69,0x6e, +0x64,0x69,0x63,0x61,0x74,0x6f,0x72,0x9d,0x15,0x69,0x65,0x6d,0x6f,0x6a,0x69,0xa2, +0x47,3,0x66,0x44,0x6d,0x5c,0x74,0x7c,0x7a,0x19,0x77,0x6a,0x73,0x65,0x71,0x75, +0x65,0x6e,0x63,0x65,0xa3,0x46,0x1a,0x6c,0x61,0x67,0x73,0x65,0x71,0x75,0x65,0x6e, +0x63,0x65,0xa3,0x44,0x1e,0x6f,0x64,0x69,0x66,0x69,0x65,0x72,0x73,0x65,0x71,0x75, +0x65,0x6e,0x63,0x65,0xa3,0x43,0x19,0x61,0x67,0x73,0x65,0x71,0x75,0x65,0x6e,0x63, +0x65,0xa3,0x45,0x12,0x61,0x74,0x68,0x4f,6,0x6f,0x39,0x6f,0x32,0x74,0xc3,9, +0x75,0x54,0x76,0xd9,0x30,0,0x12,0x6e,0x63,0x68,0x1f,0x61,0x72,0x61,0x63,0x74, +0x65,0x72,0x63,0x6f,0x64,0x65,0x70,0x6f,0x69,0x6e,0x74,0x51,0x14,0x6d,0x65,0x72, +0x69,0x63,1,0x74,0x32,0x76,0x13,0x61,0x6c,0x75,0x65,0xd9,0x30,0,0x12,0x79, +0x70,0x65,0xc3,9,0x61,0xa2,0x77,0x63,0xa2,0x82,0x66,2,0x63,0x98,0x64,0xa2, +0x53,0x6b,1,0x63,0x56,0x64,1,0x69,0x42,0x71,1,0x63,0xc3,0xd,0x75,0x17, +0x69,0x63,0x6b,0x63,0x68,0x65,0x63,0x6b,0xc3,0xd,0x13,0x6e,0x65,0x72,0x74,0x6d, +1,0x69,0x42,0x71,1,0x63,0xc3,0xf,0x75,0x17,0x69,0x63,0x6b,0x63,0x68,0x65, +0x63,0x6b,0xc3,0xf,0x13,0x6e,0x65,0x72,0x74,0x71,1,0x69,0x42,0x71,1,0x63, +0xc3,0xe,0x75,0x17,0x69,0x63,0x6b,0x63,0x68,0x65,0x63,0x6b,0xc3,0xe,0x13,0x6e, +0x65,0x72,0x74,0x6f,1,0x69,0x42,0x71,1,0x63,0xc3,0xc,0x75,0x17,0x69,0x63, +0x6b,0x63,0x68,0x65,0x63,0x6b,0xc3,0xc,0x13,0x6e,0x65,0x72,0x74,0x6b,0xd8,0x40, +5,1,0x31,0xd9,0x40,0xb,0x6d,0x10,0x65,0xd9,0x40,5,0x12,0x68,0x61,0x72, +0x51,2,0x61,0x6c,0x63,0xa2,0x4c,0x72,1,0x65,0x2a,0x69,0x11,0x6e,0x74,0x7f, +0x16,0x70,0x65,0x6e,0x64,0x65,0x64,0x63,0x1f,0x6f,0x6e,0x63,0x61,0x74,0x65,0x6e, +0x61,0x74,0x69,0x6f,0x6e,0x6d,0x61,0x72,0x6b,0x9f,0x10,0x74,2,0x73,0x2c,0x74, +0x30,0x77,0x10,0x73,0x77,0x11,0x79,0x6e,0x75,0x12,0x65,0x72,0x6e,1,0x73,0x38, +0x77,0x18,0x68,0x69,0x74,0x65,0x73,0x70,0x61,0x63,0x65,0x77,0x14,0x79,0x6e,0x74, +0x61,0x78,0x75,0x10,0x6d,0x9f,1,0x6d,0x3c,0x75,0x1a,0x6f,0x74,0x61,0x74,0x69, +0x6f,0x6e,0x6d,0x61,0x72,0x6b,0x53,0x12,0x61,0x72,0x6b,0x53,0x66,0xc1,0xf8,0x69, +0xc1,0x3c,0x69,0xa2,0x6f,0x6a,0xa4,9,0x6c,4,0x62,0xc3,8,0x63,0x8c,0x65, +0x98,0x69,0xa2,0x56,0x6f,2,0x65,0x4b,0x67,0x4c,0x77,0x11,0x65,0x72,0x4c,0x13, +0x63,0x61,0x73,0x65,0x4c,0x16,0x6d,0x61,0x70,0x70,0x69,0x6e,0x67,0xd9,0x40,4, +0x11,0x69,0x63,0x1f,0x61,0x6c,0x6f,0x72,0x64,0x65,0x72,0x65,0x78,0x63,0x65,0x70, +0x74,0x69,0x6f,0x6e,0x4b,0xd8,0x40,4,0x11,0x63,0x63,0xc3,0x10,0x18,0x61,0x64, +0x63,0x61,0x6e,0x6f,0x6e,0x69,0x63,0x1f,0x61,0x6c,0x63,0x6f,0x6d,0x62,0x69,0x6e, +0x69,0x6e,0x67,0x63,0x6c,0x61,0x73,0x73,0xc3,0x10,0x16,0x6e,0x65,0x62,0x72,0x65, +0x61,0x6b,0xc3,8,2,0x64,0x4a,0x6e,0xa2,0x5b,0x73,1,0x63,0xd9,0x40,3, +0x6f,0x16,0x63,0x6f,0x6d,0x6d,0x65,0x6e,0x74,0xd9,0x40,3,2,0x63,0x80,0x65, +0x90,0x73,0x40,1,0x62,0x52,0x74,0x46,1,0x61,0x40,0x72,0x1c,0x69,0x6e,0x61, +0x72,0x79,0x6f,0x70,0x65,0x72,0x61,0x74,0x6f,0x72,0x47,0x11,0x72,0x74,0x41,0x44, +0x1c,0x69,0x6e,0x61,0x72,0x79,0x6f,0x70,0x65,0x72,0x61,0x74,0x6f,0x72,0x45,0x3e, +0x16,0x6f,0x6e,0x74,0x69,0x6e,0x75,0x65,0x3f,0x10,0x6f,0x42,0x16,0x67,0x72,0x61, +0x70,0x68,0x69,0x63,0x43,2,0x64,0x2e,0x70,0x86,0x73,0x10,0x63,0xc3,0x17,0x11, +0x69,0x63,1,0x70,0x46,0x73,0x1e,0x79,0x6c,0x6c,0x61,0x62,0x69,0x63,0x63,0x61, +0x74,0x65,0x67,0x6f,0x72,0x79,0xc3,0x17,0x10,0x6f,0x1f,0x73,0x69,0x74,0x69,0x6f, +0x6e,0x61,0x6c,0x63,0x61,0x74,0x65,0x67,0x6f,0x72,0x79,0xc3,0x16,0x10,0x63,0xc3, +0x16,2,0x67,0xc3,6,0x6f,0x26,0x74,0xc3,7,0x11,0x69,0x6e,1,0x63,0x4a, +0x69,0x11,0x6e,0x67,1,0x67,0x2e,0x74,0x12,0x79,0x70,0x65,0xc3,7,0x13,0x72, +0x6f,0x75,0x70,0xc3,6,0x48,0x15,0x6f,0x6e,0x74,0x72,0x6f,0x6c,0x49,0x66,0x86, +0x67,0xa2,0x4a,0x68,3,0x61,0x36,0x65,0x58,0x73,0x68,0x79,0x13,0x70,0x68,0x65, +0x6e,0x3d,0x1f,0x6e,0x67,0x75,0x6c,0x73,0x79,0x6c,0x6c,0x61,0x62,0x6c,0x65,0x74, +0x79,0x70,0x65,0xc3,0xb,0x10,0x78,0x3a,0x14,0x64,0x69,0x67,0x69,0x74,0x3b,0x10, +0x74,0xc3,0xb,0x16,0x75,0x6c,0x6c,0x63,0x6f,0x6d,0x70,0x1f,0x6f,0x73,0x69,0x74, +0x69,0x6f,0x6e,0x65,0x78,0x63,0x6c,0x75,0x73,0x69,0x6f,0x6e,0x33,2,0x63,0xa2, +0x44,0x65,0xa2,0x4b,0x72,3,0x61,0x34,0x62,0x84,0x65,0x8a,0x6c,0x12,0x69,0x6e, +0x6b,0x39,0x11,0x70,0x68,0x7c,0x12,0x65,0x6d,0x65,3,0x62,0x5e,0x63,0x30,0x65, +0x48,0x6c,0x12,0x69,0x6e,0x6b,0x39,0x1a,0x6c,0x75,0x73,0x74,0x65,0x72,0x62,0x72, +0x65,0x61,0x6b,0xc3,0x12,0x14,0x78,0x74,0x65,0x6e,0x64,0x37,0x12,0x61,0x73,0x65, +0x35,0x11,0x78,0x74,0x37,0xc2,5,1,0x62,0xc3,0x12,0x6d,0xd9,0x20,0,0x1c, +0x6e,0x65,0x72,0x61,0x6c,0x63,0x61,0x74,0x65,0x67,0x6f,0x72,0x79,0xc2,5,0x13, +0x6d,0x61,0x73,0x6b,0xd9,0x20,0,0x61,0xa2,0xa2,0x62,0xa2,0xd0,0x63,0xa4,0x4f, +0x64,0xa6,0x1c,0x65,5,0x6d,0x75,0x6d,0x6e,0x70,0xa2,0x6b,0x78,0x10,0x74,0x30, +1,0x65,0x2c,0x70,0x12,0x69,0x63,0x74,0xa1,0x12,0x6e,0x64,0x65,1,0x64,0x24, +0x72,0x31,0x1b,0x70,0x69,0x63,0x74,0x6f,0x67,0x72,0x61,0x70,0x68,0x69,0x63,0xa1, +0x10,0x6f,1,0x64,0x97,0x6a,0x10,0x69,0x92,3,0x63,0x44,0x6b,0x54,0x6d,0x70, +0x70,0x1a,0x72,0x65,0x73,0x65,0x6e,0x74,0x61,0x74,0x69,0x6f,0x6e,0x95,0x17,0x6f, +0x6d,0x70,0x6f,0x6e,0x65,0x6e,0x74,0x9b,0x1c,0x65,0x79,0x63,0x61,0x70,0x73,0x65, +0x71,0x75,0x65,0x6e,0x63,0x65,0xa3,0x42,0x16,0x6f,0x64,0x69,0x66,0x69,0x65,0x72, +0x96,0x13,0x62,0x61,0x73,0x65,0x99,0x12,0x72,0x65,0x73,0x95,0x61,0x30,0x62,0x4e, +0x63,0x12,0x6f,0x6d,0x70,0x9b,0xc2,4,0x1b,0x73,0x74,0x61,0x73,0x69,0x61,0x6e, +0x77,0x69,0x64,0x74,0x68,0xc3,4,0x12,0x61,0x73,0x65,0x99,3,0x67,0x44,0x68, +0x4a,0x6c,0x4e,0x73,0x1a,0x63,0x69,0x69,0x68,0x65,0x78,0x64,0x69,0x67,0x69,0x74, +0x23,0x10,0x65,0xd9,0x40,0,0x11,0x65,0x78,0x23,1,0x6e,0x38,0x70,0x11,0x68, +0x61,0x20,0x14,0x62,0x65,0x74,0x69,0x63,0x21,0x11,0x75,0x6d,0x79,5,0x6c,0x22, +0x6c,0x36,0x6d,0x52,0x70,1,0x62,0xd9,0x40,0xd,0x74,0xc3,0x15,2,0x61,0x32, +0x6b,0xc3,1,0x6f,0x11,0x63,0x6b,0xc3,1,0x11,0x6e,0x6b,0x7b,0x10,0x67,0xd9, +0x40,1,0x61,0xa2,0x4f,0x63,0xc3,0,0x69,0x11,0x64,0x69,2,0x63,0x54,0x6d, +0x74,0x70,0x1b,0x61,0x69,0x72,0x65,0x64,0x62,0x72,0x61,0x63,0x6b,0x65,0x74,0xd8, +0x40,0xd,0x13,0x74,0x79,0x70,0x65,0xc3,0x15,0x24,1,0x6c,0x30,0x6f,0x14,0x6e, +0x74,0x72,0x6f,0x6c,0x25,0x12,0x61,0x73,0x73,0xc3,0,0x26,0x14,0x69,0x72,0x72, +0x6f,0x72,1,0x65,0x38,0x69,0x16,0x6e,0x67,0x67,0x6c,0x79,0x70,0x68,0xd9,0x40, +1,0x10,0x64,0x27,0x17,0x73,0x69,0x63,0x65,0x6d,0x6f,0x6a,0x69,0xa3,0x41,6, +0x68,0x7c,0x68,0x54,0x69,0x85,0x6f,0xa2,0x6f,0x77,4,0x63,0x30,0x6b,0x36,0x6c, +0x87,0x74,0x8b,0x75,0x89,1,0x66,0x8d,0x6d,0x8f,0x11,0x63,0x66,0x91,0x18,0x61, +0x6e,0x67,0x65,0x73,0x77,0x68,0x65,0x6e,4,0x63,0x44,0x6c,0x6c,0x6e,0x7e,0x74, +0x98,0x75,0x18,0x70,0x70,0x65,0x72,0x63,0x61,0x73,0x65,0x64,0x89,0x12,0x61,0x73, +0x65,1,0x66,0x30,0x6d,0x14,0x61,0x70,0x70,0x65,0x64,0x8f,0x14,0x6f,0x6c,0x64, +0x65,0x64,0x8d,0x18,0x6f,0x77,0x65,0x72,0x63,0x61,0x73,0x65,0x64,0x87,0x1c,0x66, +0x6b,0x63,0x63,0x61,0x73,0x65,0x66,0x6f,0x6c,0x64,0x65,0x64,0x91,0x18,0x69,0x74, +0x6c,0x65,0x63,0x61,0x73,0x65,0x64,0x8b,0x13,0x6d,0x70,0x65,0x78,0x33,0x61,0x2e, +0x63,0xa2,0x48,0x66,0xd9,0x40,2,1,0x6e,0x72,0x73,0x10,0x65,3,0x64,0x83, +0x66,0x3a,0x69,0x4a,0x73,0x17,0x65,0x6e,0x73,0x69,0x74,0x69,0x76,0x65,0x65,0x15, +0x6f,0x6c,0x64,0x69,0x6e,0x67,0xd9,0x40,2,0x17,0x67,0x6e,0x6f,0x72,0x61,0x62, +0x6c,0x65,0x85,0x13,0x6f,0x6e,0x69,0x63,0x1f,0x61,0x6c,0x63,0x6f,0x6d,0x62,0x69, +0x6e,0x69,0x6e,0x67,0x63,0x6c,0x61,0x73,0x73,0xc3,2,0x10,0x63,0xc3,2,3, +0x61,0x30,0x65,0x34,0x69,0xa2,0x41,0x74,0xc3,3,0x11,0x73,0x68,0x29,2,0x63, +0x3a,0x66,0x58,0x70,0x2c,0x16,0x72,0x65,0x63,0x61,0x74,0x65,0x64,0x2d,0x1d,0x6f, +0x6d,0x70,0x6f,0x73,0x69,0x74,0x69,0x6f,0x6e,0x74,0x79,0x70,0x65,0xc3,3,0x15, +0x61,0x75,0x6c,0x74,0x69,0x67,0x1f,0x6e,0x6f,0x72,0x61,0x62,0x6c,0x65,0x63,0x6f, +0x64,0x65,0x70,0x6f,0x69,0x6e,0x74,0x2b,0x2a,0x10,0x61,0x2e,0x15,0x63,0x72,0x69, +0x74,0x69,0x63,0x2f,3,0x66,0x34,0x6e,0x3e,0x74,0x42,0x79,0x22,0x11,0x65,0x73, +0x23,0x20,0x13,0x61,0x6c,0x73,0x65,0x21,0x20,0x10,0x6f,0x21,0x22,0x12,0x72,0x75, +0x65,0x23,0xb,0x6b,0x5b,0x6f,0x23,0x6f,0x3c,0x72,0x4c,0x76,1,0x69,0x24,0x72, +0x33,0x13,0x72,0x61,0x6d,0x61,0x33,0x10,0x76,0x22,0x14,0x65,0x72,0x6c,0x61,0x79, +0x23,0xa2,0xe2,0x13,0x69,0x67,0x68,0x74,0xa3,0xe2,0x6b,0x58,0x6c,0x74,0x6e,3, +0x6b,0x2f,0x6f,0x30,0x72,0x21,0x75,0x12,0x6b,0x74,0x61,0x2f,0x19,0x74,0x72,0x65, +0x6f,0x72,0x64,0x65,0x72,0x65,0x64,0x21,1,0x61,0x24,0x76,0x31,0x18,0x6e,0x61, +0x76,0x6f,0x69,0x63,0x69,0x6e,0x67,0x31,0xa2,0xe0,0x12,0x65,0x66,0x74,0xa3,0xe0, +0x64,0x45,0x64,0x4e,0x68,0x88,0x69,1,0x6f,0x26,0x73,0xa3,0xf0,0x1a,0x74,0x61, +0x73,0x75,0x62,0x73,0x63,0x72,0x69,0x70,0x74,0xa3,0xf0,2,0x61,0xa3,0xea,0x62, +0xa3,0xe9,0x6f,0x13,0x75,0x62,0x6c,0x65,1,0x61,0x30,0x62,0x13,0x65,0x6c,0x6f, +0x77,0xa3,0xe9,0x13,0x62,0x6f,0x76,0x65,0xa3,0xea,0x12,0x61,0x6e,0x72,0x2c,0x15, +0x65,0x61,0x64,0x69,0x6e,0x67,0x2d,0x61,0xa2,0x7b,0x62,0xa2,0xd4,0x63,0x11,0x63, +0x63,4,0x31,0x3c,0x32,0xa2,0x42,0x33,0xa2,0x56,0x38,0xa2,0x64,0x39,0x10,0x31, +0xa3,0x5b,9,0x35,0xa,0x35,0x3f,0x36,0x41,0x37,0x43,0x38,0x45,0x39,0x47,0x30, +0x30,0x31,0x3c,0x32,0x42,0x33,0x4e,0x34,0x3d,0x34,1,0x33,0xa3,0x67,0x37,0xa3, +0x6b,0x36,0x10,0x38,0xa3,0x76,0x38,1,0x32,0xa3,0x7a,0x39,0xa3,0x81,0x3a,2, +0x30,0xa3,0x82,0x32,0xa3,0x84,0x33,0xa3,0x85,9,0x35,0xa,0x35,0x53,0x36,0x55, +0x37,0x57,0x38,0x59,0x39,0x5b,0x30,0x49,0x31,0x4b,0x32,0x4d,0x33,0x4f,0x34,0x51, +6,0x33,8,0x33,0x63,0x34,0x65,0x35,0x67,0x36,0x69,0x30,0x5d,0x31,0x5f,0x32, +0x61,0x10,0x34,0xa3,0x54,0xa2,0xe6,3,0x62,0xa0,0x6c,0xa3,0xe4,0x72,0xa3,0xe8, +0x74,2,0x61,0x74,0x62,0x7c,0x74,0x14,0x61,0x63,0x68,0x65,0x64,1,0x61,0x3e, +0x62,0x13,0x65,0x6c,0x6f,0x77,0xa2,0xca,0x13,0x6c,0x65,0x66,0x74,0xa3,0xc8,0x13, +0x62,0x6f,0x76,0x65,0xa2,0xd6,0x14,0x72,0x69,0x67,0x68,0x74,0xa3,0xd8,0xa2,0xd6, +0x10,0x72,0xa3,0xd8,0xa2,0xca,0x10,0x6c,0xa3,0xc8,0x12,0x6f,0x76,0x65,0xa2,0xe6, +1,0x6c,0x30,0x72,0x13,0x69,0x67,0x68,0x74,0xa3,0xe8,0x12,0x65,0x66,0x74,0xa3, +0xe4,0xa2,0xdc,2,0x65,0x2c,0x6c,0xa3,0xda,0x72,0xa3,0xde,0x12,0x6c,0x6f,0x77, +0xa2,0xdc,1,0x6c,0x30,0x72,0x13,0x69,0x67,0x68,0x74,0xa3,0xde,0x12,0x65,0x66, +0x74,0xa3,0xda,0xb,0x6e,0xc0,0xca,0x72,0x5f,0x72,0x46,0x73,0xa2,0x48,0x77,1, +0x68,0x24,0x73,0x33,0x17,0x69,0x74,0x65,0x73,0x70,0x61,0x63,0x65,0x33,0x22,1, +0x69,0x30,0x6c,2,0x65,0x3d,0x69,0x4b,0x6f,0x3f,0x18,0x67,0x68,0x74,0x74,0x6f, +0x6c,0x65,0x66,0x74,0x22,2,0x65,0x38,0x69,0x48,0x6f,0x16,0x76,0x65,0x72,0x72, +0x69,0x64,0x65,0x3f,0x17,0x6d,0x62,0x65,0x64,0x64,0x69,0x6e,0x67,0x3d,0x15,0x73, +0x6f,0x6c,0x61,0x74,0x65,0x4b,0x30,0x1e,0x65,0x67,0x6d,0x65,0x6e,0x74,0x73,0x65, +0x70,0x61,0x72,0x61,0x74,0x6f,0x72,0x31,0x6e,0xa2,0x41,0x6f,0xa2,0x53,0x70,2, +0x61,0x66,0x64,0x86,0x6f,0x1b,0x70,0x64,0x69,0x72,0x65,0x63,0x74,0x69,0x6f,0x6e, +0x61,0x6c,1,0x66,0x32,0x69,0x15,0x73,0x6f,0x6c,0x61,0x74,0x65,0x4d,0x14,0x6f, +0x72,0x6d,0x61,0x74,0x41,0x1f,0x72,0x61,0x67,0x72,0x61,0x70,0x68,0x73,0x65,0x70, +0x61,0x72,0x61,0x74,0x6f,0x72,0x2f,1,0x66,0x41,0x69,0x4d,1,0x6f,0x28,0x73, +0x10,0x6d,0x43,0x1b,0x6e,0x73,0x70,0x61,0x63,0x69,0x6e,0x67,0x6d,0x61,0x72,0x6b, +0x43,1,0x6e,0x35,0x74,0x19,0x68,0x65,0x72,0x6e,0x65,0x75,0x74,0x72,0x61,0x6c, +0x35,0x65,0x88,0x65,0x98,0x66,0xa2,0x6a,0x6c,0x20,1,0x65,0x30,0x72,2,0x65, +0x37,0x69,0x49,0x6f,0x39,0x18,0x66,0x74,0x74,0x6f,0x72,0x69,0x67,0x68,0x74,0x20, +2,0x65,0x38,0x69,0x48,0x6f,0x16,0x76,0x65,0x72,0x72,0x69,0x64,0x65,0x39,0x17, +0x6d,0x62,0x65,0x64,0x64,0x69,0x6e,0x67,0x37,0x15,0x73,0x6f,0x6c,0x61,0x74,0x65, +0x49,3,0x6e,0x25,0x73,0x27,0x74,0x29,0x75,0x15,0x72,0x6f,0x70,0x65,0x61,0x6e, +2,0x6e,0x3c,0x73,0x46,0x74,0x18,0x65,0x72,0x6d,0x69,0x6e,0x61,0x74,0x6f,0x72, +0x29,0x14,0x75,0x6d,0x62,0x65,0x72,0x25,0x17,0x65,0x70,0x61,0x72,0x61,0x74,0x6f, +0x72,0x27,1,0x69,0x28,0x73,0x10,0x69,0x47,0x1f,0x72,0x73,0x74,0x73,0x74,0x72, +0x6f,0x6e,0x67,0x69,0x73,0x6f,0x6c,0x61,0x74,0x65,0x47,0x61,0x4e,0x62,0x84,0x63, +1,0x6f,0x24,0x73,0x2d,0x1c,0x6d,0x6d,0x6f,0x6e,0x73,0x65,0x70,0x61,0x72,0x61, +0x74,0x6f,0x72,0x2d,2,0x6c,0x3b,0x6e,0x2b,0x72,0x13,0x61,0x62,0x69,0x63,1, +0x6c,0x30,0x6e,0x14,0x75,0x6d,0x62,0x65,0x72,0x2b,0x14,0x65,0x74,0x74,0x65,0x72, +0x3b,0x2e,1,0x6e,0x45,0x6f,0x1c,0x75,0x6e,0x64,0x61,0x72,0x79,0x6e,0x65,0x75, +0x74,0x72,0x61,0x6c,0x45,0,0x16,0x6d,0xc9,0x20,0x74,0xc2,0x30,0x77,0x89,0x77, +0x86,0x79,0xa2,0x46,0x7a,1,0x61,0x58,0x6e,0x1a,0x61,0x6d,0x65,0x6e,0x6e,0x79, +0x6d,0x75,0x73,0x69,0x63,0xa4,0x40,0x19,0x61,0x6c,0x6e,0x6f,0x74,0x61,0x74,0x69, +0x6f,0x6e,0xa5,0x40,0x1c,0x6e,0x61,0x62,0x61,0x7a,0x61,0x72,0x73,0x71,0x75,0x61, +0x72,0x65,0xa5,0x18,0x10,0x61,1,0x6e,0x36,0x72,0x16,0x61,0x6e,0x67,0x63,0x69, +0x74,0x69,0xa3,0xfc,0x12,0x63,0x68,0x6f,0xa5,0x2c,1,0x65,0x88,0x69,2,0x6a, +0x3c,0x72,0x68,0x73,0x17,0x79,0x6c,0x6c,0x61,0x62,0x6c,0x65,0x73,0xa3,0x48,0x12, +0x69,0x6e,0x67,0xa2,0x74,0x1e,0x68,0x65,0x78,0x61,0x67,0x72,0x61,0x6d,0x73,0x79, +0x6d,0x62,0x6f,0x6c,0x73,0xa3,0x74,0x16,0x61,0x64,0x69,0x63,0x61,0x6c,0x73,0xa3, +0x49,0x13,0x7a,0x69,0x64,0x69,0xa5,0x34,0x74,0xa2,0x65,0x75,0xa4,0x4f,0x76,3, +0x61,0x3c,0x65,0x80,0x69,0xa2,0x50,0x73,0xa2,0x6c,0x12,0x73,0x75,0x70,0xa3,0x7d, +1,0x69,0xa3,0x9f,0x72,0x1e,0x69,0x61,0x74,0x69,0x6f,0x6e,0x73,0x65,0x6c,0x65, +0x63,0x74,0x6f,0x72,0x73,0xa2,0x6c,0x19,0x73,0x75,0x70,0x70,0x6c,0x65,0x6d,0x65, +0x6e,0x74,0xa3,0x7d,1,0x64,0x3c,0x72,0x19,0x74,0x69,0x63,0x61,0x6c,0x66,0x6f, +0x72,0x6d,0x73,0xa3,0x91,0x14,0x69,0x63,0x65,0x78,0x74,0xa2,0xaf,0x16,0x65,0x6e, +0x73,0x69,0x6f,0x6e,0x73,0xa3,0xaf,0x15,0x74,0x68,0x6b,0x75,0x71,0x69,0xa5,0x3f, +5,0x69,0x3f,0x69,0x5a,0x6f,0x8c,0x72,0x1c,0x61,0x6e,0x73,0x70,0x6f,0x72,0x74, +0x61,0x6e,0x64,0x6d,0x61,0x70,0xa2,0xcf,0x16,0x73,0x79,0x6d,0x62,0x6f,0x6c,0x73, +0xa3,0xcf,2,0x62,0x34,0x66,0x3c,0x72,0x13,0x68,0x75,0x74,0x61,0xa3,0xfb,0x13, +0x65,0x74,0x61,0x6e,0x57,0x14,0x69,0x6e,0x61,0x67,0x68,0xa3,0x90,0x11,0x74,0x6f, +0xa5,0x3d,0x61,0x3e,0x65,0xa2,0xa0,0x68,0x10,0x61,1,0x61,0x24,0x69,0x53,0x11, +0x6e,0x61,0x3d,4,0x67,0x8e,0x69,0xa2,0x49,0x6b,0xa2,0x72,0x6d,0xa2,0x74,0x6e, +0x10,0x67,1,0x73,0x68,0x75,0x10,0x74,0xa4,0x10,1,0x63,0x40,0x73,0x11,0x75, +0x70,0xa4,0x33,0x16,0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa5,0x33,0x18,0x6f,0x6d, +0x70,0x6f,0x6e,0x65,0x6e,0x74,0x73,0xa5,0x11,0x10,0x61,0xa5,0x3c,2,0x61,0x2a, +0x62,0x32,0x73,0xa3,0x60,0x12,0x6c,0x6f,0x67,0xa3,0x62,0x13,0x61,0x6e,0x77,0x61, +0xa3,0x65,3,0x6c,0x52,0x74,0x56,0x76,0x5e,0x78,0x16,0x75,0x61,0x6e,0x6a,0x69, +0x6e,0x67,0xa2,0x7c,0x16,0x73,0x79,0x6d,0x62,0x6f,0x6c,0x73,0xa3,0x7c,0x10,0x65, +0xa3,0x70,0x12,0x68,0x61,0x6d,0xa3,0xae,0x12,0x69,0x65,0x74,0xa3,0xb7,0x11,0x72, +0x69,0xa3,0xdc,0x11,0x69,0x6c,0x48,0x12,0x73,0x75,0x70,0xa4,0x2b,0x16,0x70,0x6c, +0x65,0x6d,0x65,0x6e,0x74,0xa5,0x2b,0x13,0x6c,0x75,0x67,0x75,0x4b,2,0x63,0x8c, +0x67,0xa2,0x41,0x6e,0x1f,0x69,0x66,0x69,0x65,0x64,0x63,0x61,0x6e,0x61,0x64,0x69, +0x61,0x6e,0x61,0x62,0x6f,0x1f,0x72,0x69,0x67,0x69,0x6e,0x61,0x6c,0x73,0x79,0x6c, +0x6c,0x61,0x62,0x69,0x63,0x73,0x62,0x17,0x65,0x78,0x74,0x65,0x6e,0x64,0x65,0x64, +0xa2,0xad,0x10,0x61,0xa5,0x3e,0x11,0x61,0x73,0x62,0x12,0x65,0x78,0x74,0xa2,0xad, +0x10,0x61,0xa5,0x3e,0x15,0x61,0x72,0x69,0x74,0x69,0x63,0xa3,0x78,0x70,0xc3,0x4b, +0x70,0xa6,0x61,0x72,0xa8,0x1d,0x73,7,0x6f,0xc1,0xbe,0x6f,0xa2,0x69,0x70,0xa2, +0x85,0x75,0xa2,0xa4,0x79,2,0x6c,0x50,0x6d,0x62,0x72,0x12,0x69,0x61,0x63,0x3a, +0x12,0x73,0x75,0x70,0xa4,0x17,0x16,0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa5,0x17, +0x17,0x6f,0x74,0x69,0x6e,0x61,0x67,0x72,0x69,0xa3,0x8f,0x13,0x62,0x6f,0x6c,0x73, +1,0x61,0x4c,0x66,0x10,0x6f,0x1f,0x72,0x6c,0x65,0x67,0x61,0x63,0x79,0x63,0x6f, +0x6d,0x70,0x75,0x74,0x69,0x6e,0x67,0xa5,0x32,0x1f,0x6e,0x64,0x70,0x69,0x63,0x74, +0x6f,0x67,0x72,0x61,0x70,0x68,0x73,0x65,0x78,0x74,1,0x61,0xa5,0x2a,0x65,0x14, +0x6e,0x64,0x65,0x64,0x61,0xa5,0x2a,2,0x67,0x34,0x72,0x3e,0x79,0x13,0x6f,0x6d, +0x62,0x6f,0xa5,0x16,0x13,0x64,0x69,0x61,0x6e,0xa5,0x23,0x17,0x61,0x73,0x6f,0x6d, +0x70,0x65,0x6e,0x67,0xa3,0xda,1,0x61,0x32,0x65,0x14,0x63,0x69,0x61,0x6c,0x73, +0xa3,0x56,0x12,0x63,0x69,0x6e,0x1f,0x67,0x6d,0x6f,0x64,0x69,0x66,0x69,0x65,0x72, +0x6c,0x65,0x74,0x74,0x65,0x72,0x73,0x2d,2,0x6e,0x48,0x70,0x76,0x74,0x1d,0x74, +0x6f,0x6e,0x73,0x69,0x67,0x6e,0x77,0x72,0x69,0x74,0x69,0x6e,0x67,0xa5,6,0x15, +0x64,0x61,0x6e,0x65,0x73,0x65,0xa2,0x9b,0x12,0x73,0x75,0x70,0xa2,0xdb,0x16,0x70, +0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa3,0xdb,4,0x61,0xa2,0xa8,0x65,0x5c,0x6d,0x9e, +0x70,0xa2,0x4b,0x73,0x13,0x79,0x6d,0x62,0x6f,0x1f,0x6c,0x73,0x61,0x6e,0x64,0x70, +0x69,0x63,0x74,0x6f,0x67,0x72,0x61,0x70,0x68,0x73,0xa5,5,0x10,0x72,1,0x61, +0x4e,0x73,0x12,0x63,0x72,0x69,0x1f,0x70,0x74,0x73,0x61,0x6e,0x64,0x73,0x75,0x62, +0x73,0x63,0x72,0x69,0x70,0x74,0x73,0x73,0x14,0x6e,0x64,0x73,0x75,0x62,0x73,0x1b, +0x61,0x74,0x68,0x6f,0x70,0x65,0x72,0x61,0x74,0x6f,0x72,0x73,0xa3,0x6a,1,0x6c, +0x40,0x75,1,0x61,0x6e,0x6e,0x17,0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0xa3, +0x8e,0x15,0x65,0x6d,0x65,0x6e,0x74,0x61,1,0x6c,0x50,0x72,0x1e,0x79,0x70,0x72, +0x69,0x76,0x61,0x74,0x65,0x75,0x73,0x65,0x61,0x72,0x65,0x61,1,0x61,0xa3,0x6d, +0x62,0xa3,0x6e,3,0x61,0x5c,0x6d,0x78,0x70,0xa2,0x41,0x73,0x13,0x79,0x6d,0x62, +0x6f,0x1f,0x6c,0x73,0x61,0x6e,0x64,0x70,0x69,0x63,0x74,0x6f,0x67,0x72,0x61,0x70, +0x68,0x73,0xa5,5,0x14,0x72,0x72,0x6f,0x77,0x73,2,0x61,0xa3,0x67,0x62,0xa3, +0x68,0x63,0xa3,0xfa,0x13,0x61,0x74,0x68,0x65,0x1f,0x6d,0x61,0x74,0x69,0x63,0x61, +0x6c,0x6f,0x70,0x65,0x72,0x61,0x74,0x6f,0x72,0x73,0xa3,0x6a,0x19,0x75,0x6e,0x63, +0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0xa3,0x8e,0x61,0x88,0x68,0xa2,0x48,0x69,0xa2, +0x71,0x6d,0x12,0x61,0x6c,0x6c,1,0x66,0x46,0x6b,0x15,0x61,0x6e,0x61,0x65,0x78, +0x74,0xa4,0x29,0x15,0x65,0x6e,0x73,0x69,0x6f,0x6e,0xa5,0x29,0x12,0x6f,0x72,0x6d, +1,0x73,0xa3,0x54,0x76,0x16,0x61,0x72,0x69,0x61,0x6e,0x74,0x73,0xa3,0x54,1, +0x6d,0x36,0x75,0x16,0x72,0x61,0x73,0x68,0x74,0x72,0x61,0xa3,0xa1,0x15,0x61,0x72, +0x69,0x74,0x61,0x6e,0xa3,0xac,1,0x61,0x52,0x6f,0x13,0x72,0x74,0x68,0x61,0x1f, +0x6e,0x64,0x66,0x6f,0x72,0x6d,0x61,0x74,0x63,0x6f,0x6e,0x74,0x72,0x6f,0x6c,0x73, +0xa3,0xf7,1,0x72,0x2e,0x76,0x12,0x69,0x61,0x6e,0xa3,0x79,0x12,0x61,0x64,0x61, +0xa3,0xd9,1,0x64,0x50,0x6e,0x13,0x68,0x61,0x6c,0x61,0x50,0x1d,0x61,0x72,0x63, +0x68,0x61,0x69,0x63,0x6e,0x75,0x6d,0x62,0x65,0x72,0x73,0xa3,0xf9,0x13,0x64,0x68, +0x61,0x6d,0xa3,0xf8,5,0x72,0x35,0x72,0x44,0x73,0x64,0x75,1,0x61,0xa3,0x4e, +0x6e,0x17,0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0x71,0x17,0x69,0x76,0x61,0x74, +0x65,0x75,0x73,0x65,0xa2,0x4e,0x13,0x61,0x72,0x65,0x61,0xa3,0x4e,0x1b,0x61,0x6c, +0x74,0x65,0x72,0x70,0x61,0x68,0x6c,0x61,0x76,0x69,0xa3,0xf6,0x61,0x40,0x68,0x82, +0x6c,0x19,0x61,0x79,0x69,0x6e,0x67,0x63,0x61,0x72,0x64,0x73,0xa3,0xcc,2,0x68, +0x38,0x6c,0x4a,0x75,0x15,0x63,0x69,0x6e,0x68,0x61,0x75,0xa3,0xf5,0x17,0x61,0x77, +0x68,0x68,0x6d,0x6f,0x6e,0x67,0xa3,0xf3,0x15,0x6d,0x79,0x72,0x65,0x6e,0x65,0xa3, +0xf4,1,0x61,0x8e,0x6f,1,0x65,0x74,0x6e,0x16,0x65,0x74,0x69,0x63,0x65,0x78, +0x74,0xa2,0x72,1,0x65,0x2c,0x73,0x11,0x75,0x70,0xa3,0x8d,0x15,0x6e,0x73,0x69, +0x6f,0x6e,0x73,0xa2,0x72,0x19,0x73,0x75,0x70,0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74, +0xa3,0x8d,0x15,0x6e,0x69,0x63,0x69,0x61,0x6e,0xa3,0x97,1,0x67,0x3e,0x69,0x13, +0x73,0x74,0x6f,0x73,0xa2,0xa6,0x13,0x64,0x69,0x73,0x63,0xa3,0xa6,0x12,0x73,0x70, +0x61,0xa3,0x96,1,0x65,0x5c,0x75,1,0x6d,0x2a,0x6e,0x11,0x69,0x63,0x67,0x10, +0x69,0xa2,0xc0,0x1d,0x6e,0x75,0x6d,0x65,0x72,0x61,0x6c,0x73,0x79,0x6d,0x62,0x6f, +0x6c,0x73,0xa3,0xc0,0x13,0x6a,0x61,0x6e,0x67,0xa3,0xa3,0x6d,0xa2,0xf0,0x6e,0xa8, +0x23,0x6f,6,0x70,0x63,0x70,0x56,0x72,0x8a,0x73,0xa2,0x4c,0x74,0x10,0x74,0x1f, +0x6f,0x6d,0x61,0x6e,0x73,0x69,0x79,0x61,0x71,0x6e,0x75,0x6d,0x62,0x65,0x72,0x73, +0xa5,0x28,0x18,0x74,0x69,0x63,0x61,0x6c,0x63,0x68,0x61,0x72,0x1f,0x61,0x63,0x74, +0x65,0x72,0x72,0x65,0x63,0x6f,0x67,0x6e,0x69,0x74,0x69,0x6f,0x6e,0x85,1,0x69, +0x46,0x6e,0x1e,0x61,0x6d,0x65,0x6e,0x74,0x61,0x6c,0x64,0x69,0x6e,0x67,0x62,0x61, +0x74,0x73,0xa3,0xf2,0x11,0x79,0x61,0x47,1,0x61,0x30,0x6d,0x13,0x61,0x6e,0x79, +0x61,0xa3,0x7a,0x11,0x67,0x65,0xa5,0xf,0x63,0xa2,0x7b,0x67,0xa2,0x7b,0x6c,1, +0x63,0xa2,0x6c,0x64,6,0x70,0x42,0x70,0x3a,0x73,0x5a,0x74,0x88,0x75,0x14,0x79, +0x67,0x68,0x75,0x72,0xa5,0x3b,0x11,0x65,0x72,1,0x6d,0x2e,0x73,0x12,0x69,0x61, +0x6e,0xa3,0x8c,0x11,0x69,0x63,0xa3,0xf1,0x10,0x6f,1,0x67,0x3a,0x75,0x18,0x74, +0x68,0x61,0x72,0x61,0x62,0x69,0x61,0x6e,0xa3,0xbb,0x13,0x64,0x69,0x61,0x6e,0xa5, +0x22,0x14,0x75,0x72,0x6b,0x69,0x63,0xa3,0xbf,0x68,0x42,0x69,0x54,0x6e,0x1a,0x6f, +0x72,0x74,0x68,0x61,0x72,0x61,0x62,0x69,0x61,0x6e,0xa3,0xf0,0x17,0x75,0x6e,0x67, +0x61,0x72,0x69,0x61,0x6e,0xa5,4,0x14,0x74,0x61,0x6c,0x69,0x63,0xa3,0x58,0x13, +0x68,0x69,0x6b,0x69,0xa3,0x9d,0x10,0x72,0x85,0x12,0x68,0x61,0x6d,0x65,6,0x6f, +0x86,0x6f,0x6c,0x72,0xa2,0x61,0x75,0xa2,0x62,0x79,0x14,0x61,0x6e,0x6d,0x61,0x72, +0x58,0x12,0x65,0x78,0x74,2,0x61,0xa3,0xb6,0x62,0xa3,0xee,0x65,0x13,0x6e,0x64, +0x65,0x64,1,0x61,0xa3,0xb6,0x62,0xa3,0xee,1,0x64,0x52,0x6e,0x15,0x67,0x6f, +0x6c,0x69,0x61,0x6e,0x6a,0x12,0x73,0x75,0x70,0xa4,0xd,0x16,0x70,0x6c,0x65,0x6d, +0x65,0x6e,0x74,0xa5,0xd,0x10,0x69,0xa2,0xec,0x13,0x66,0x69,0x65,0x72,1,0x6c, +0x3c,0x74,0x19,0x6f,0x6e,0x65,0x6c,0x65,0x74,0x74,0x65,0x72,0x73,0xa3,0x8a,0x15, +0x65,0x74,0x74,0x65,0x72,0x73,0x2d,0x10,0x6f,0xa3,0xed,1,0x6c,0x44,0x73,0x11, +0x69,0x63,0xa2,0x5c,0x18,0x61,0x6c,0x73,0x79,0x6d,0x62,0x6f,0x6c,0x73,0xa3,0x5c, +0x13,0x74,0x61,0x6e,0x69,0xa5,3,0x61,0xa2,0x9b,0x65,0xa4,0x4c,0x69,1,0x61, +0xa2,0x8f,0x73,0x10,0x63,5,0x70,0x18,0x70,0xa2,0x71,0x73,0x36,0x74,0x17,0x65, +0x63,0x68,0x6e,0x69,0x63,0x61,0x6c,0x81,0x15,0x79,0x6d,0x62,0x6f,0x6c,0x73,0x8f, +0x61,0xa2,0x66,0x65,0x46,0x6d,0x19,0x61,0x74,0x68,0x73,0x79,0x6d,0x62,0x6f,0x6c, +0x73,1,0x61,0xa3,0x66,0x62,0xa3,0x69,0x17,0x6c,0x6c,0x61,0x6e,0x65,0x6f,0x75, +0x73,2,0x6d,0x3a,0x73,0x6c,0x74,0x17,0x65,0x63,0x68,0x6e,0x69,0x63,0x61,0x6c, +0x81,0x11,0x61,0x74,0x1f,0x68,0x65,0x6d,0x61,0x74,0x69,0x63,0x61,0x6c,0x73,0x79, +0x6d,0x62,0x6f,0x6c,0x73,1,0x61,0xa3,0x66,0x62,0xa3,0x69,0x15,0x79,0x6d,0x62, +0x6f,0x6c,0x73,0x8e,0x12,0x61,0x6e,0x64,1,0x61,0x3c,0x70,0x19,0x69,0x63,0x74, +0x6f,0x67,0x72,0x61,0x70,0x68,0x73,0xa3,0xcd,0x14,0x72,0x72,0x6f,0x77,0x73,0xa3, +0x73,0x10,0x6f,0xa3,0xd8,7,0x72,0x6f,0x72,0x44,0x73,0x4e,0x74,0x62,0x79,0x19, +0x61,0x6e,0x6e,0x75,0x6d,0x65,0x72,0x61,0x6c,0x73,0xa5,0x20,0x13,0x63,0x68,0x65, +0x6e,0xa5,0xc,0x18,0x61,0x72,0x61,0x6d,0x67,0x6f,0x6e,0x64,0x69,0xa5,0x14,0x10, +0x68,2,0x61,0x3a,0x65,0x4a,0x6f,0x17,0x70,0x65,0x72,0x61,0x74,0x6f,0x72,0x73, +0x7f,0x16,0x6c,0x70,0x68,0x61,0x6e,0x75,0x6d,0xa3,0x5d,0x16,0x6d,0x61,0x74,0x69, +0x63,0x61,0x6c,1,0x61,0x36,0x6f,0x17,0x70,0x65,0x72,0x61,0x74,0x6f,0x72,0x73, +0x7f,0x11,0x6c,0x70,0x1f,0x68,0x61,0x6e,0x75,0x6d,0x65,0x72,0x69,0x63,0x73,0x79, +0x6d,0x62,0x6f,0x6c,0x73,0xa3,0x5d,0x68,0x50,0x6b,0x7e,0x6c,0x88,0x6e,1,0x64, +0x34,0x69,0x15,0x63,0x68,0x61,0x65,0x61,0x6e,0xa3,0xea,0x12,0x61,0x69,0x63,0xa3, +0xc6,1,0x61,0x3e,0x6a,0x12,0x6f,0x6e,0x67,0xa2,0xaa,0x14,0x74,0x69,0x6c,0x65, +0x73,0xa3,0xaa,0x13,0x6a,0x61,0x6e,0x69,0xa3,0xe9,0x13,0x61,0x73,0x61,0x72,0xa5, +0x1f,0x15,0x61,0x79,0x61,0x6c,0x61,0x6d,0x4f,3,0x64,0x6c,0x65,0x7e,0x6e,0xa2, +0x47,0x72,0x14,0x6f,0x69,0x74,0x69,0x63,1,0x63,0x3c,0x68,0x19,0x69,0x65,0x72, +0x6f,0x67,0x6c,0x79,0x70,0x68,0x73,0xa3,0xd7,0x15,0x75,0x72,0x73,0x69,0x76,0x65, +0xa3,0xd6,0x17,0x65,0x66,0x61,0x69,0x64,0x72,0x69,0x6e,0xa5,0x21,0x17,0x74,0x65, +0x69,0x6d,0x61,0x79,0x65,0x6b,0xa2,0xb8,0x12,0x65,0x78,0x74,0xa2,0xd5,0x16,0x65, +0x6e,0x73,0x69,0x6f,0x6e,0x73,0xa3,0xd5,0x18,0x64,0x65,0x6b,0x69,0x6b,0x61,0x6b, +0x75,0x69,0xa3,0xeb,6,0x6b,0x3b,0x6b,0x56,0x6f,0x5a,0x75,0x64,0x79,0x11,0x69, +0x61,0x1f,0x6b,0x65,0x6e,0x67,0x70,0x75,0x61,0x63,0x68,0x75,0x65,0x68,0x6d,0x6f, +0x6e,0x67,0xa5,0x27,0x10,0x6f,0xa3,0x92,0x14,0x62,0x6c,0x6f,0x63,0x6b,0x21,1, +0x6d,0x2c,0x73,0x11,0x68,0x75,0xa5,0x15,0x17,0x62,0x65,0x72,0x66,0x6f,0x72,0x6d, +0x73,0x7b,0x61,0x44,0x62,0x21,0x65,0x10,0x77,1,0x61,0xa5,0xe,0x74,0x14,0x61, +0x69,0x6c,0x75,0x65,0xa3,0x8b,2,0x62,0x3c,0x67,0x4a,0x6e,0x17,0x64,0x69,0x6e, +0x61,0x67,0x61,0x72,0x69,0xa5,0x26,0x15,0x61,0x74,0x61,0x65,0x61,0x6e,0xa3,0xef, +0x16,0x6d,0x75,0x6e,0x64,0x61,0x72,0x69,0xa5,0x47,0x67,0xc4,0x5d,0x6a,0xc1,0xe4, +0x6a,0xa2,0xdf,0x6b,0xa2,0xf8,0x6c,4,0x61,0x54,0x65,0xa2,0x6b,0x69,0xa2,0x82, +0x6f,0xa2,0xc1,0x79,1,0x63,0x2e,0x64,0x12,0x69,0x61,0x6e,0xa3,0xa9,0x12,0x69, +0x61,0x6e,0xa3,0xa7,1,0x6f,0x55,0x74,0x11,0x69,0x6e,1,0x31,0x96,0x65,0x11, +0x78,0x74,6,0x64,0x21,0x64,0xa3,0x95,0x65,0x2c,0x66,0xa5,0x39,0x67,0xa5,0x3a, +0xa2,0xe7,0x13,0x6e,0x64,0x65,0x64,6,0x64,0xc,0x64,0xa3,0x95,0x65,0xa3,0xe7, +0x66,0xa5,0x39,0x67,0xa5,0x3a,0x61,0x2a,0x62,0x29,0x63,0xa3,0x94,0x26,0x18,0x64, +0x64,0x69,0x74,0x69,0x6f,0x6e,0x61,0x6c,0x6d,0x24,0x12,0x73,0x75,0x70,0x24,0x16, +0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0x25,1,0x70,0x42,0x74,0x1d,0x74,0x65,0x72, +0x6c,0x69,0x6b,0x65,0x73,0x79,0x6d,0x62,0x6f,0x6c,0x73,0x79,0x12,0x63,0x68,0x61, +0xa3,0x9c,2,0x6d,0x4e,0x6e,0x54,0x73,0x10,0x75,0xa2,0xb0,0x12,0x73,0x75,0x70, +0xa4,0x31,0x16,0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa5,0x31,0x11,0x62,0x75,0xa3, +0x6f,0x12,0x65,0x61,0x72,1,0x61,0xa3,0xe8,0x62,1,0x69,0x38,0x73,0x17,0x79, +0x6c,0x6c,0x61,0x62,0x61,0x72,0x79,0xa3,0x75,0x17,0x64,0x65,0x6f,0x67,0x72,0x61, +0x6d,0x73,0xa3,0x76,0x1a,0x77,0x73,0x75,0x72,0x72,0x6f,0x67,0x61,0x74,0x65,0x73, +0xa3,0x4d,0x10,0x61,1,0x6d,0x32,0x76,0x14,0x61,0x6e,0x65,0x73,0x65,0xa3,0xb5, +0x10,0x6f,0x5c,0x12,0x65,0x78,0x74,1,0x61,0xa3,0xb4,0x62,0xa3,0xb9,1,0x61, +0xa2,0x43,0x68,4,0x61,0x40,0x69,0x50,0x6d,0x6e,0x6f,0x86,0x75,0x15,0x64,0x61, +0x77,0x61,0x64,0x69,0xa3,0xe6,0x16,0x72,0x6f,0x73,0x68,0x74,0x68,0x69,0xa3,0x89, +0x1d,0x74,0x61,0x6e,0x73,0x6d,0x61,0x6c,0x6c,0x73,0x63,0x72,0x69,0x70,0x74,0xa5, +0x30,0x11,0x65,0x72,0x68,0x16,0x73,0x79,0x6d,0x62,0x6f,0x6c,0x73,0xa3,0x71,0x12, +0x6a,0x6b,0x69,0xa3,0xe5,5,0x74,0x35,0x74,0x34,0x77,0x7a,0x79,0x13,0x61,0x68, +0x6c,0x69,0xa3,0xa2,0x14,0x61,0x6b,0x61,0x6e,0x61,0x9e,1,0x65,0x4c,0x70,0x10, +0x68,0x1f,0x6f,0x6e,0x65,0x74,0x69,0x63,0x65,0x78,0x74,0x65,0x6e,0x73,0x69,0x6f, +0x6e,0x73,0xa3,0x6b,0x11,0x78,0x74,0xa3,0x6b,0x10,0x69,0xa5,0x46,0x69,0xa2,0x4e, +0x6b,0xa2,0x51,0x6e,3,0x61,0x34,0x62,0x84,0x67,0x8a,0x6e,0x12,0x61,0x64,0x61, +0x4d,1,0x65,0x40,0x73,0x11,0x75,0x70,0xa2,0xcb,0x16,0x70,0x6c,0x65,0x6d,0x65, +0x6e,0x74,0xa3,0xcb,0x11,0x78,0x74,2,0x61,0xa5,0x13,0x62,0xa5,0x38,0x65,0x13, +0x6e,0x64,0x65,0x64,1,0x61,0xa5,0x13,0x62,0xa5,0x38,0x11,0x75,0x6e,0xa3,0x42, +0x11,0x78,0x69,0x96,0x17,0x72,0x61,0x64,0x69,0x63,0x61,0x6c,0x73,0x97,0x12,0x74, +0x68,0x69,0xa3,0xc1,0x1c,0x74,0x6f,0x76,0x69,0x6b,0x6e,0x75,0x6d,0x65,0x72,0x61, +0x6c,0x73,0xa5,0x45,0x67,0xa2,0xb5,0x68,0xa4,0x84,0x69,3,0x64,0x4c,0x6d,0xa2, +0x55,0x6e,0xa2,0x62,0x70,0x13,0x61,0x65,0x78,0x74,0x2a,0x16,0x65,0x6e,0x73,0x69, +0x6f,0x6e,0x73,0x2b,1,0x63,0x99,0x65,0x17,0x6f,0x67,0x72,0x61,0x70,0x68,0x69, +0x63,1,0x64,0x56,0x73,0x15,0x79,0x6d,0x62,0x6f,0x6c,0x73,0xa4,0xb,0x1d,0x61, +0x6e,0x64,0x70,0x75,0x6e,0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0xa5,0xb,0x13, +0x65,0x73,0x63,0x72,0x1f,0x69,0x70,0x74,0x69,0x6f,0x6e,0x63,0x68,0x61,0x72,0x61, +0x63,0x74,0x65,0x72,0x73,0x99,0x1c,0x70,0x65,0x72,0x69,0x61,0x6c,0x61,0x72,0x61, +0x6d,0x61,0x69,0x63,0xa3,0xba,1,0x64,0x62,0x73,0x1b,0x63,0x72,0x69,0x70,0x74, +0x69,0x6f,0x6e,0x61,0x6c,0x70,0x61,1,0x68,0x32,0x72,0x14,0x74,0x68,0x69,0x61, +0x6e,0xa3,0xbd,0x13,0x6c,0x61,0x76,0x69,0xa3,0xbe,0x11,0x69,0x63,1,0x6e,0x3e, +0x73,0x1a,0x69,0x79,0x61,0x71,0x6e,0x75,0x6d,0x62,0x65,0x72,0x73,0xa5,0x1e,0x19, +0x75,0x6d,0x62,0x65,0x72,0x66,0x6f,0x72,0x6d,0x73,0xa3,0xb2,4,0x65,0x74,0x6c, +0xa2,0x82,0x6f,0xa2,0x9a,0x72,0xa2,0x9e,0x75,2,0x6a,0x34,0x6e,0x3e,0x72,0x14, +0x6d,0x75,0x6b,0x68,0x69,0x43,0x14,0x61,0x72,0x61,0x74,0x69,0x45,0x18,0x6a,0x61, +0x6c,0x61,0x67,0x6f,0x6e,0x64,0x69,0xa5,0x1c,1,0x6e,0xa2,0x46,0x6f,1,0x6d, +0x6e,0x72,0x13,0x67,0x69,0x61,0x6e,0x5a,1,0x65,0x40,0x73,0x11,0x75,0x70,0xa2, +0x87,0x16,0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa3,0x87,0x11,0x78,0x74,0xa4,0x1b, +0x14,0x65,0x6e,0x64,0x65,0x64,0xa5,0x1b,0x1a,0x65,0x74,0x72,0x69,0x63,0x73,0x68, +0x61,0x70,0x65,0x73,0x8c,0x12,0x65,0x78,0x74,0xa2,0xe3,0x14,0x65,0x6e,0x64,0x65, +0x64,0xa3,0xe3,0x1e,0x65,0x72,0x61,0x6c,0x70,0x75,0x6e,0x63,0x74,0x75,0x61,0x74, +0x69,0x6f,0x6e,0x71,0x17,0x61,0x67,0x6f,0x6c,0x69,0x74,0x69,0x63,0xa2,0x88,0x12, +0x73,0x75,0x70,0xa4,0xa,0x16,0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa5,0xa,0x13, +0x74,0x68,0x69,0x63,0xa3,0x59,1,0x61,0x5c,0x65,0x11,0x65,0x6b,0x30,1,0x61, +0x38,0x65,0x11,0x78,0x74,0x6e,0x14,0x65,0x6e,0x64,0x65,0x64,0x6f,0x17,0x6e,0x64, +0x63,0x6f,0x70,0x74,0x69,0x63,0x31,0x13,0x6e,0x74,0x68,0x61,0xa3,0xe4,2,0x61, +0xa2,0x48,0x65,0xa2,0xdf,0x69,1,0x67,0x30,0x72,0x14,0x61,0x67,0x61,0x6e,0x61, +0x9d,0x10,0x68,1,0x70,0x3a,0x73,0x18,0x75,0x72,0x72,0x6f,0x67,0x61,0x74,0x65, +0x73,0xa3,0x4b,1,0x72,0x3c,0x75,0x19,0x73,0x75,0x72,0x72,0x6f,0x67,0x61,0x74, +0x65,0x73,0xa3,0x4c,0x11,0x69,0x76,0x1f,0x61,0x74,0x65,0x75,0x73,0x65,0x73,0x75, +0x72,0x72,0x6f,0x67,0x61,0x74,0x65,0x73,0xa3,0x4c,2,0x6c,0x32,0x6e,0x9a,0x74, +0x12,0x72,0x61,0x6e,0xa5,2,0x10,0x66,2,0x61,0x58,0x6d,0x70,0x77,0x14,0x69, +0x64,0x74,0x68,0x61,0x1f,0x6e,0x64,0x66,0x75,0x6c,0x6c,0x77,0x69,0x64,0x74,0x68, +0x66,0x6f,0x72,0x6d,0x73,0xa3,0x57,0x1a,0x6e,0x64,0x66,0x75,0x6c,0x6c,0x66,0x6f, +0x72,0x6d,0x73,0xa3,0x57,0x13,0x61,0x72,0x6b,0x73,0xa3,0x52,2,0x67,0x34,0x69, +0xa2,0x45,0x75,0x12,0x6e,0x6f,0x6f,0xa3,0x63,0x11,0x75,0x6c,0xa2,0x4a,2,0x63, +0x3c,0x6a,0x5e,0x73,0x17,0x79,0x6c,0x6c,0x61,0x62,0x6c,0x65,0x73,0xa3,0x4a,0x1f, +0x6f,0x6d,0x70,0x61,0x74,0x69,0x62,0x69,0x6c,0x69,0x74,0x79,0x6a,0x61,0x6d,0x6f, +0xa3,0x41,0x12,0x61,0x6d,0x6f,0x5c,0x17,0x65,0x78,0x74,0x65,0x6e,0x64,0x65,0x64, +1,0x61,0xa3,0xb4,0x62,0xa3,0xb9,0x19,0x66,0x69,0x72,0x6f,0x68,0x69,0x6e,0x67, +0x79,0x61,0xa5,0x1d,0x13,0x62,0x72,0x65,0x77,0x37,0x61,0xa4,0xc,0x62,0xa6,0x59, +0x63,0xa8,0x2e,0x64,0xac,0xe3,0x65,5,0x6d,0xa9,0x6d,0x94,0x6e,0xa2,0x41,0x74, +0x15,0x68,0x69,0x6f,0x70,0x69,0x63,0x5e,1,0x65,0x40,0x73,0x11,0x75,0x70,0xa2, +0x86,0x16,0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa3,0x86,0x11,0x78,0x74,0xa2,0x85, +2,0x61,0xa3,0xc8,0x62,0xa5,0x37,0x65,0x13,0x6e,0x64,0x65,0x64,0xa2,0x85,1, +0x61,0xa3,0xc8,0x62,0xa5,0x37,0x16,0x6f,0x74,0x69,0x63,0x6f,0x6e,0x73,0xa3,0xce, +0x15,0x63,0x6c,0x6f,0x73,0x65,0x64,2,0x61,0x5a,0x63,0x9e,0x69,0x1c,0x64,0x65, +0x6f,0x67,0x72,0x61,0x70,0x68,0x69,0x63,0x73,0x75,0x70,0xa2,0xc4,0x16,0x70,0x6c, +0x65,0x6d,0x65,0x6e,0x74,0xa3,0xc4,0x16,0x6c,0x70,0x68,0x61,0x6e,0x75,0x6d,0x86, +1,0x65,0x2c,0x73,0x11,0x75,0x70,0xa3,0xc3,0x13,0x72,0x69,0x63,0x73,0x86,0x18, +0x75,0x70,0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa3,0xc3,0x11,0x6a,0x6b,0xa2,0x44, +0x1f,0x6c,0x65,0x74,0x74,0x65,0x72,0x73,0x61,0x6e,0x64,0x6d,0x6f,0x6e,0x74,0x68, +0x73,0xa3,0x44,0x61,0x4a,0x67,0x76,0x6c,1,0x62,0x30,0x79,0x13,0x6d,0x61,0x69, +0x63,0xa5,0x25,0x13,0x61,0x73,0x61,0x6e,0xa3,0xe2,0x13,0x72,0x6c,0x79,0x64,0x1f, +0x79,0x6e,0x61,0x73,0x74,0x69,0x63,0x63,0x75,0x6e,0x65,0x69,0x66,0x6f,0x72,0x6d, +0xa5,1,0x1f,0x79,0x70,0x74,0x69,0x61,0x6e,0x68,0x69,0x65,0x72,0x6f,0x67,0x6c, +0x79,0x70,0x68,1,0x66,0x26,0x73,0xa3,0xc2,0x1c,0x6f,0x72,0x6d,0x61,0x74,0x63, +0x6f,0x6e,0x74,0x72,0x6f,0x6c,0x73,0xa5,0x24,7,0x6e,0xc0,0xf2,0x6e,0x3e,0x72, +0xa2,0x5d,0x73,0xa2,0xe5,0x76,0x14,0x65,0x73,0x74,0x61,0x6e,0xa3,0xbc,1,0x61, +0x92,0x63,0x13,0x69,0x65,0x6e,0x74,1,0x67,0x34,0x73,0x15,0x79,0x6d,0x62,0x6f, +0x6c,0x73,0xa3,0xa5,0x13,0x72,0x65,0x65,0x6b,1,0x6d,0x34,0x6e,0x15,0x75,0x6d, +0x62,0x65,0x72,0x73,0xa3,0x7f,0x13,0x75,0x73,0x69,0x63,0xa2,0x7e,0x19,0x61,0x6c, +0x6e,0x6f,0x74,0x61,0x74,0x69,0x6f,0x6e,0xa3,0x7e,0x10,0x74,0x1f,0x6f,0x6c,0x69, +0x61,0x6e,0x68,0x69,0x65,0x72,0x6f,0x67,0x6c,0x79,0x70,0x68,0x73,0xa3,0xfe,2, +0x61,0x32,0x6d,0xa2,0x7e,0x72,0x12,0x6f,0x77,0x73,0x7d,0x12,0x62,0x69,0x63,0x38, +3,0x65,0x4a,0x6d,0x80,0x70,0xa2,0x50,0x73,0x11,0x75,0x70,0xa2,0x80,0x16,0x70, +0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa3,0x80,0x11,0x78,0x74,3,0x61,0xa3,0xd2,0x62, +0xa5,0x35,0x63,0xa5,0x41,0x65,0x13,0x6e,0x64,0x65,0x64,2,0x61,0xa3,0xd2,0x62, +0xa5,0x35,0x63,0xa5,0x41,0x12,0x61,0x74,0x68,0xa2,0xd3,0x18,0x65,0x6d,0x61,0x74, +0x69,0x63,0x61,0x6c,0x61,0x1f,0x6c,0x70,0x68,0x61,0x62,0x65,0x74,0x69,0x63,0x73, +0x79,0x6d,0x62,0x6f,0x6c,0x73,0xa3,0xd3,1,0x66,0x42,0x72,0x1e,0x65,0x73,0x65, +0x6e,0x74,0x61,0x74,0x69,0x6f,0x6e,0x66,0x6f,0x72,0x6d,0x73,1,0x61,0xa3,0x51, +0x62,0xa3,0x55,0x14,0x65,0x6e,0x69,0x61,0x6e,0x35,0x12,0x63,0x69,0x69,0x23,0x64, +0x9e,0x65,0xa2,0x42,0x68,0xa2,0x4d,0x6c,1,0x63,0x62,0x70,0x17,0x68,0x61,0x62, +0x65,0x74,0x69,0x63,0x70,1,0x66,0xa3,0x50,0x72,0x1e,0x65,0x73,0x65,0x6e,0x74, +0x61,0x74,0x69,0x6f,0x6e,0x66,0x6f,0x72,0x6d,0x73,0xa3,0x50,0x16,0x68,0x65,0x6d, +0x69,0x63,0x61,0x6c,0xa2,0xd0,0x16,0x73,0x79,0x6d,0x62,0x6f,0x6c,0x73,0xa3,0xd0, +0x12,0x6c,0x61,0x6d,0xa5,7,0x1a,0x67,0x65,0x61,0x6e,0x6e,0x75,0x6d,0x62,0x65, +0x72,0x73,0xa3,0x77,0x11,0x6f,0x6d,0xa3,0xfd,7,0x6f,0x71,0x6f,0x64,0x72,0xa2, +0x41,0x75,0xa2,0x58,0x79,0x1b,0x7a,0x61,0x6e,0x74,0x69,0x6e,0x65,0x6d,0x75,0x73, +0x69,0x63,0xa2,0x5b,0x18,0x61,0x6c,0x73,0x79,0x6d,0x62,0x6f,0x6c,0x73,0xa3,0x5b, +1,0x70,0x34,0x78,0x16,0x64,0x72,0x61,0x77,0x69,0x6e,0x67,0x89,0x14,0x6f,0x6d, +0x6f,0x66,0x6f,0xa0,0x12,0x65,0x78,0x74,0xa2,0x43,0x14,0x65,0x6e,0x64,0x65,0x64, +0xa3,0x43,0x10,0x61,1,0x68,0x40,0x69,0x12,0x6c,0x6c,0x65,0x92,0x17,0x70,0x61, +0x74,0x74,0x65,0x72,0x6e,0x73,0x93,0x11,0x6d,0x69,0xa3,0xc9,1,0x67,0x2c,0x68, +0x11,0x69,0x64,0xa3,0x64,0x14,0x69,0x6e,0x65,0x73,0x65,0xa3,0x81,0x61,0x48,0x65, +0xa2,0x4e,0x68,0xa2,0x52,0x6c,0x1a,0x6f,0x63,0x6b,0x65,0x6c,0x65,0x6d,0x65,0x6e, +0x74,0x73,0x8b,3,0x6c,0x34,0x6d,0x40,0x73,0x66,0x74,0x11,0x61,0x6b,0xa3,0xc7, +0x14,0x69,0x6e,0x65,0x73,0x65,0xa3,0x93,0x11,0x75,0x6d,0xa2,0xb1,0x12,0x73,0x75, +0x70,0xa2,0xca,0x16,0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa3,0xca,1,0x69,0x30, +0x73,0x13,0x61,0x76,0x61,0x68,0xa3,0xdd,0x15,0x63,0x6c,0x61,0x74,0x69,0x6e,0x23, +0x14,0x6e,0x67,0x61,0x6c,0x69,0x41,0x16,0x61,0x69,0x6b,0x73,0x75,0x6b,0x69,0xa5, +8,5,0x6f,0xc1,0x60,0x6f,0xa2,0x69,0x75,0xa4,0x24,0x79,1,0x70,0xa2,0x44, +0x72,0x14,0x69,0x6c,0x6c,0x69,0x63,0x32,1,0x65,0x4c,0x73,0x11,0x75,0x70,0xa2, +0x61,0x16,0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa2,0x61,0x12,0x61,0x72,0x79,0xa3, +0x61,0x11,0x78,0x74,4,0x61,0xa3,0x9e,0x62,0xa3,0xa0,0x63,0xa5,9,0x64,0xa5, +0x43,0x65,0x13,0x6e,0x64,0x65,0x64,3,0x61,0xa3,0x9e,0x62,0xa3,0xa0,0x63,0xa5, +9,0x64,0xa5,0x43,0x10,0x72,1,0x69,0x34,0x6f,0x15,0x6d,0x69,0x6e,0x6f,0x61, +0x6e,0xa5,0x36,0x1a,0x6f,0x74,0x73,0x79,0x6c,0x6c,0x61,0x62,0x61,0x72,0x79,0xa3, +0x7b,3,0x6d,0x5a,0x6e,0xa2,0x95,0x70,0xa2,0xa0,0x75,0x17,0x6e,0x74,0x69,0x6e, +0x67,0x72,0x6f,0x64,0xa2,0x9a,0x17,0x6e,0x75,0x6d,0x65,0x72,0x61,0x6c,0x73,0xa3, +0x9a,2,0x62,0x3a,0x6d,0xa2,0x5f,0x70,0x15,0x61,0x74,0x6a,0x61,0x6d,0x6f,0xa3, +0x41,0x14,0x69,0x6e,0x69,0x6e,0x67,2,0x64,0x46,0x68,0x9e,0x6d,0x1d,0x61,0x72, +0x6b,0x73,0x66,0x6f,0x72,0x73,0x79,0x6d,0x62,0x6f,0x6c,0x73,0x77,0x1e,0x69,0x61, +0x63,0x72,0x69,0x74,0x69,0x63,0x61,0x6c,0x6d,0x61,0x72,0x6b,0x73,0x2e,2,0x65, +0x40,0x66,0xa6,0x4c,0x73,0x18,0x75,0x70,0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa3, +0x83,0x16,0x78,0x74,0x65,0x6e,0x64,0x65,0x64,0xa3,0xe0,0x17,0x61,0x6c,0x66,0x6d, +0x61,0x72,0x6b,0x73,0xa3,0x52,0x11,0x6f,0x6e,0x1f,0x69,0x6e,0x64,0x69,0x63,0x6e, +0x75,0x6d,0x62,0x65,0x72,0x66,0x6f,0x72,0x6d,0x73,0xa3,0xb2,0x1b,0x74,0x72,0x6f, +0x6c,0x70,0x69,0x63,0x74,0x75,0x72,0x65,0x73,0x83,0x12,0x74,0x69,0x63,0xa2,0x84, +0x1b,0x65,0x70,0x61,0x63,0x74,0x6e,0x75,0x6d,0x62,0x65,0x72,0x73,0xa3,0xdf,1, +0x6e,0x3e,0x72,0x1b,0x72,0x65,0x6e,0x63,0x79,0x73,0x79,0x6d,0x62,0x6f,0x6c,0x73, +0x75,0x15,0x65,0x69,0x66,0x6f,0x72,0x6d,0xa2,0x98,0x16,0x6e,0x75,0x6d,0x62,0x65, +0x72,0x73,0xa2,0x99,0x1d,0x61,0x6e,0x64,0x70,0x75,0x6e,0x63,0x74,0x75,0x61,0x74, +0x69,0x6f,0x6e,0xa3,0x99,0x61,0xa2,0xe4,0x68,0xa4,0xe,0x6a,0x10,0x6b,0xa2,0x47, +4,0x63,0x8c,0x65,0xa2,0x80,0x72,0xa2,0x9b,0x73,0xa2,0xad,0x75,0x1f,0x6e,0x69, +0x66,0x69,0x65,0x64,0x69,0x64,0x65,0x6f,0x67,0x72,0x61,0x70,0x68,0x73,0xa2,0x47, +0x18,0x65,0x78,0x74,0x65,0x6e,0x73,0x69,0x6f,0x6e,7,0x65,0x6b,0x65,0xa5,0, +0x66,0xa5,0x12,0x67,0xa5,0x2e,0x68,0xa5,0x42,0x14,0x6f,0x6d,0x70,0x61,0x74,0xa2, +0x45,1,0x66,0x96,0x69,1,0x62,0x44,0x64,0x17,0x65,0x6f,0x67,0x72,0x61,0x70, +0x68,0x73,0xa2,0x4f,0x12,0x73,0x75,0x70,0xa3,0x5f,0x14,0x69,0x6c,0x69,0x74,0x79, +0xa2,0x45,1,0x66,0x54,0x69,0x18,0x64,0x65,0x6f,0x67,0x72,0x61,0x70,0x68,0x73, +0xa2,0x4f,0x19,0x73,0x75,0x70,0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa3,0x5f,0x13, +0x6f,0x72,0x6d,0x73,0xa3,0x53,0x11,0x78,0x74,7,0x65,0xc,0x65,0xa5,0,0x66, +0xa5,0x12,0x67,0xa5,0x2e,0x68,0xa5,0x42,0x61,0xa3,0x46,0x62,0xa3,0x5e,0x63,0xa3, +0xc5,0x64,0xa3,0xd1,0x19,0x61,0x64,0x69,0x63,0x61,0x6c,0x73,0x73,0x75,0x70,0x94, +0x16,0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0x95,1,0x74,0x50,0x79,0x14,0x6d,0x62, +0x6f,0x6c,0x73,0x9a,0x1d,0x61,0x6e,0x64,0x70,0x75,0x6e,0x63,0x74,0x75,0x61,0x74, +0x69,0x6f,0x6e,0x9b,0x14,0x72,0x6f,0x6b,0x65,0x73,0xa3,0x82,2,0x6e,0x48,0x72, +0x64,0x75,0x1d,0x63,0x61,0x73,0x69,0x61,0x6e,0x61,0x6c,0x62,0x61,0x6e,0x69,0x61, +0x6e,0xa3,0xde,0x1d,0x61,0x64,0x69,0x61,0x6e,0x73,0x79,0x6c,0x6c,0x61,0x62,0x69, +0x63,0x73,0x63,0x12,0x69,0x61,0x6e,0xa3,0xa8,2,0x61,0x3a,0x65,0x4c,0x6f,0x16, +0x72,0x61,0x73,0x6d,0x69,0x61,0x6e,0xa5,0x2d,1,0x6b,0x26,0x6d,0xa3,0xa4,0x11, +0x6d,0x61,0xa3,0xd4,1,0x72,0x38,0x73,0x17,0x73,0x73,0x79,0x6d,0x62,0x6f,0x6c, +0x73,0xa5,0x19,0x13,0x6f,0x6b,0x65,0x65,0x60,0x12,0x73,0x75,0x70,0xa2,0xff,0x16, +0x70,0x6c,0x65,0x6d,0x65,0x6e,0x74,0xa3,0xff,3,0x65,0x3e,0x69,0x8e,0x6f,0xa2, +0x71,0x75,0x15,0x70,0x6c,0x6f,0x79,0x61,0x6e,0xa3,0xe1,1,0x73,0x60,0x76,0x16, +0x61,0x6e,0x61,0x67,0x61,0x72,0x69,0x3e,0x12,0x65,0x78,0x74,0xa2,0xb3,1,0x61, +0xa5,0x44,0x65,0x13,0x6e,0x64,0x65,0x64,0xa2,0xb3,0x10,0x61,0xa5,0x44,0x13,0x65, +0x72,0x65,0x74,0xa3,0x5a,2,0x61,0x3a,0x6e,0x82,0x76,0x16,0x65,0x73,0x61,0x6b, +0x75,0x72,0x75,0xa5,0x2f,0x18,0x63,0x72,0x69,0x74,0x69,0x63,0x61,0x6c,0x73,0x2e, +2,0x65,0x30,0x66,0x36,0x73,0x11,0x75,0x70,0xa3,0x83,0x11,0x78,0x74,0xa3,0xe0, +0x18,0x6f,0x72,0x73,0x79,0x6d,0x62,0x6f,0x6c,0x73,0x77,0x14,0x67,0x62,0x61,0x74, +0x73,0x91,1,0x67,0x3e,0x6d,0x12,0x69,0x6e,0x6f,0xa2,0xab,0x14,0x74,0x69,0x6c, +0x65,0x73,0xa3,0xab,0x11,0x72,0x61,0xa5,0x1a,8,0x6d,0x5f,0x6d,0x3a,0x6e,0x48, +0x73,0x7a,0x76,0xa2,0x4b,0x77,0x12,0x69,0x64,0x65,0x43,0x11,0x65,0x64,0x32,0x12, +0x69,0x61,0x6c,0x33,2,0x61,0x40,0x62,0x37,0x6f,1,0x62,0x28,0x6e,0x10,0x65, +0x21,0x13,0x72,0x65,0x61,0x6b,0x37,0x10,0x72,0x34,0x12,0x72,0x6f,0x77,0x35,2, +0x6d,0x38,0x71,0x46,0x75,1,0x62,0x3d,0x70,0x3e,0x11,0x65,0x72,0x3f,1,0x61, +0x24,0x6c,0x39,0x11,0x6c,0x6c,0x39,1,0x72,0x3b,0x75,0x12,0x61,0x72,0x65,0x3b, +0x12,0x65,0x72,0x74,0x40,0x13,0x69,0x63,0x61,0x6c,0x41,0x63,0x58,0x65,0x92,0x66, +0x96,0x69,1,0x6e,0x36,0x73,0x10,0x6f,0x30,0x14,0x6c,0x61,0x74,0x65,0x64,0x31, +0x11,0x69,0x74,0x2e,0x12,0x69,0x61,0x6c,0x2f,2,0x61,0x36,0x69,0x48,0x6f,0x10, +0x6d,0x24,0x12,0x70,0x61,0x74,0x25,0x10,0x6e,0x22,0x15,0x6f,0x6e,0x69,0x63,0x61, +0x6c,0x23,0x13,0x72,0x63,0x6c,0x65,0x27,0x11,0x6e,0x63,0x27,2,0x69,0x3a,0x6f, +0x44,0x72,0x10,0x61,0x2c,0x14,0x63,0x74,0x69,0x6f,0x6e,0x2d,0x10,0x6e,0x28,0x11, +0x61,0x6c,0x29,0x11,0x6e,0x74,0x2b,4,0x61,0x3a,0x66,0x4c,0x68,0x5e,0x6e,0x70, +0x77,0x2a,0x12,0x69,0x64,0x65,0x2b,0x22,0x17,0x6d,0x62,0x69,0x67,0x75,0x6f,0x75, +0x73,0x23,0x26,0x17,0x75,0x6c,0x6c,0x77,0x69,0x64,0x74,0x68,0x27,0x24,0x17,0x61, +0x6c,0x66,0x77,0x69,0x64,0x74,0x68,0x25,0x20,1,0x61,0x30,0x65,0x14,0x75,0x74, +0x72,0x61,0x6c,0x21,0x28,0x13,0x72,0x72,0x6f,0x77,0x29,0xd,0x6e,0xc0,0xfb,0x73, +0x6d,0x73,0x3a,0x74,0x98,0x75,0xa2,0x49,0x7a,2,0x6c,0x3b,0x70,0x3d,0x73,0x39, +5,0x6f,0x28,0x6f,0x57,0x70,0x34,0x75,0x16,0x72,0x72,0x6f,0x67,0x61,0x74,0x65, +0x45,0x11,0x61,0x63,1,0x65,0x32,0x69,0x15,0x6e,0x67,0x6d,0x61,0x72,0x6b,0x31, +0x18,0x73,0x65,0x70,0x61,0x72,0x61,0x74,0x6f,0x72,0x39,0x63,0x53,0x6b,0x55,0x6d, +0x51,0x1d,0x69,0x74,0x6c,0x65,0x63,0x61,0x73,0x65,0x6c,0x65,0x74,0x74,0x65,0x72, +0x27,1,0x6e,0x40,0x70,0x1c,0x70,0x65,0x72,0x63,0x61,0x73,0x65,0x6c,0x65,0x74, +0x74,0x65,0x72,0x23,0x17,0x61,0x73,0x73,0x69,0x67,0x6e,0x65,0x64,0x21,0x6e,0x8a, +0x6f,0xa2,0x47,0x70,8,0x66,0x14,0x66,0x5b,0x69,0x59,0x6f,0x4f,0x72,0x24,0x73, +0x49,0x17,0x69,0x76,0x61,0x74,0x65,0x75,0x73,0x65,0x43,0x61,0x2c,0x63,0x4d,0x64, +0x47,0x65,0x4b,0x1f,0x72,0x61,0x67,0x72,0x61,0x70,0x68,0x73,0x65,0x70,0x61,0x72, +0x61,0x74,0x6f,0x72,0x3d,2,0x64,0x33,0x6c,0x35,0x6f,0x36,0x1b,0x6e,0x73,0x70, +0x61,0x63,0x69,0x6e,0x67,0x6d,0x61,0x72,0x6b,0x2d,1,0x70,0x7c,0x74,0x12,0x68, +0x65,0x72,3,0x6c,0x38,0x6e,0x42,0x70,0x4c,0x73,0x14,0x79,0x6d,0x62,0x6f,0x6c, +0x57,0x14,0x65,0x74,0x74,0x65,0x72,0x2b,0x14,0x75,0x6d,0x62,0x65,0x72,0x37,0x19, +0x75,0x6e,0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0x4f,0x1c,0x65,0x6e,0x70,0x75, +0x6e,0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0x49,0x66,0x9e,0x66,0x88,0x69,0xa2, +0x4b,0x6c,0xa2,0x5c,0x6d,4,0x61,0x60,0x63,0x31,0x65,0x2f,0x6e,0x2d,0x6f,0x15, +0x64,0x69,0x66,0x69,0x65,0x72,1,0x6c,0x30,0x73,0x14,0x79,0x6d,0x62,0x6f,0x6c, +0x55,0x14,0x65,0x74,0x74,0x65,0x72,0x29,0x17,0x74,0x68,0x73,0x79,0x6d,0x62,0x6f, +0x6c,0x51,1,0x69,0x2e,0x6f,0x13,0x72,0x6d,0x61,0x74,0x41,0x1d,0x6e,0x61,0x6c, +0x70,0x75,0x6e,0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0x5b,0x10,0x6e,0x1f,0x69, +0x74,0x69,0x61,0x6c,0x70,0x75,0x6e,0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0x59, +6,0x6d,0x18,0x6d,0x29,0x6f,0x28,0x74,0x27,0x75,0x23,0x2a,0x1c,0x77,0x65,0x72, +0x63,0x61,0x73,0x65,0x6c,0x65,0x74,0x74,0x65,0x72,0x25,0x65,0x28,0x69,0x3c,0x6c, +0x25,0x19,0x74,0x74,0x65,0x72,0x6e,0x75,0x6d,0x62,0x65,0x72,0x35,0x1a,0x6e,0x65, +0x73,0x65,0x70,0x61,0x72,0x61,0x74,0x6f,0x72,0x3b,0x63,0x44,0x64,0xa2,0x60,0x65, +0x1b,0x6e,0x63,0x6c,0x6f,0x73,0x69,0x6e,0x67,0x6d,0x61,0x72,0x6b,0x2f,6,0x6e, +0x39,0x6e,0x46,0x6f,0x4e,0x73,0x45,0x75,0x1b,0x72,0x72,0x65,0x6e,0x63,0x79,0x73, +0x79,0x6d,0x62,0x6f,0x6c,0x53,0x20,0x12,0x74,0x72,0x6c,0x3f,0x42,0x10,0x6e,1, +0x6e,0x2c,0x74,0x12,0x72,0x6f,0x6c,0x3f,0x1f,0x65,0x63,0x74,0x6f,0x72,0x70,0x75, +0x6e,0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0x4d,0x63,0x3f,0x66,0x41,0x6c,0x1d, +0x6f,0x73,0x65,0x70,0x75,0x6e,0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0x4b,2, +0x61,0x30,0x65,0x4a,0x69,0x12,0x67,0x69,0x74,0x33,0x1c,0x73,0x68,0x70,0x75,0x6e, +0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0x47,0x1a,0x63,0x69,0x6d,0x61,0x6c,0x6e, +0x75,0x6d,0x62,0x65,0x72,0x33,0,0x13,0x6e,0xc1,0xf,0x74,0x76,0x74,0x4c,0x76, +0x9a,0x77,0xa2,0x48,0x79,0xa2,0x49,0x7a,1,0x61,0x2c,0x68,0x12,0x61,0x69,0x6e, +0x8b,0x11,0x69,0x6e,0x85,2,0x61,0x36,0x65,0x3c,0x68,0x14,0x69,0x6e,0x79,0x65, +0x68,0xa3,0x66,1,0x68,0x71,0x77,0x73,1,0x68,0x28,0x74,0x10,0x68,0x77,0x16, +0x6d,0x61,0x72,0x62,0x75,0x74,0x61,0x74,0x13,0x67,0x6f,0x61,0x6c,0x3d,0x1a,0x65, +0x72,0x74,0x69,0x63,0x61,0x6c,0x74,0x61,0x69,0x6c,0xa3,0x67,0x11,0x61,0x77,0x79, +1,0x65,0x32,0x75,0x11,0x64,0x68,0x80,0x11,0x68,0x65,0x83,0x10,0x68,0x7a,1, +0x62,0x34,0x77,0x16,0x69,0x74,0x68,0x74,0x61,0x69,0x6c,0x7f,0x14,0x61,0x72,0x72, +0x65,0x65,0x7d,0x6e,0xa2,0x4c,0x70,0xa2,0x69,0x71,0xa2,0x69,0x72,0xa2,0x6f,0x73, +5,0x74,0x22,0x74,0x38,0x77,0x4c,0x79,0x16,0x72,0x69,0x61,0x63,0x77,0x61,0x77, +0x6f,0x18,0x72,0x61,0x69,0x67,0x68,0x74,0x77,0x61,0x77,0xa3,0x55,0x15,0x61,0x73, +0x68,0x6b,0x61,0x66,0x6d,0x61,0x2e,0x65,0x38,0x68,0x11,0x69,0x6e,0x6b,0x10,0x64, +0x62,0x11,0x68,0x65,0x65,1,0x65,0x2e,0x6d,0x13,0x6b,0x61,0x74,0x68,0x69,0x10, +0x6e,0x67,2,0x6f,0x2c,0x75,0x50,0x79,0x10,0x61,0x91,1,0x6a,0x28,0x6f,0x10, +0x6e,0x55,0x1a,0x6f,0x69,0x6e,0x69,0x6e,0x67,0x67,0x72,0x6f,0x75,0x70,0x21,0x10, +0x6e,0x57,0x10,0x65,0x59,0x10,0x61,1,0x66,0x5b,0x70,0x10,0x68,0x5d,1,0x65, +0x38,0x6f,0x18,0x68,0x69,0x6e,0x67,0x79,0x61,0x79,0x65,0x68,0x93,1,0x68,0x5f, +0x76,0x16,0x65,0x72,0x73,0x65,0x64,0x70,0x65,0x61,0x67,0xc1,0xc7,0x67,0xa4,0x52, +0x68,0xa4,0x59,0x6b,0xa4,0x99,0x6c,0xa4,0xb2,0x6d,2,0x61,0x2e,0x65,0xa4,0x3e, +0x69,0x10,0x6d,0x53,1,0x6c,0xa2,0xe7,0x6e,0x16,0x69,0x63,0x68,0x61,0x65,0x61, +0x6e,0,0x12,0x6e,0x76,0x73,0x51,0x73,0x3e,0x74,0x5c,0x77,0xa0,0x79,0xa2,0x42, +0x7a,0x13,0x61,0x79,0x69,0x6e,0xa3,0x54,0x10,0x61,1,0x64,0x2e,0x6d,0x12,0x65, +0x6b,0x68,0xa3,0x4c,0x11,0x68,0x65,0xa3,0x4b,3,0x61,0x38,0x65,0x3c,0x68,0x4a, +0x77,0x13,0x65,0x6e,0x74,0x79,0xa3,0x51,0x10,0x77,0xa3,0x4d,1,0x6e,0xa3,0x4e, +0x74,0x10,0x68,0xa3,0x4f,0x14,0x61,0x6d,0x65,0x64,0x68,0xa3,0x50,0x11,0x61,0x77, +0xa3,0x52,0x12,0x6f,0x64,0x68,0xa3,0x53,0x6e,0x3a,0x6f,0x40,0x70,0x46,0x71,0x4a, +0x72,0x12,0x65,0x73,0x68,0xa3,0x4a,0x11,0x75,0x6e,0xa3,0x46,0x11,0x6e,0x65,0xa3, +0x47,0x10,0x65,0xa3,0x48,0x12,0x6f,0x70,0x68,0xa3,0x49,0x67,0x33,0x67,0x38,0x68, +0x40,0x6b,0x5e,0x6c,0x66,0x6d,0x11,0x65,0x6d,0xa3,0x45,0x13,0x69,0x6d,0x65,0x6c, +0xa1,1,0x65,0x32,0x75,0x14,0x6e,0x64,0x72,0x65,0x64,0xa3,0x42,0x11,0x74,0x68, +0xa3,0x41,0x12,0x61,0x70,0x68,0xa3,0x43,0x14,0x61,0x6d,0x65,0x64,0x68,0xa3,0x44, +0x61,0x34,0x62,0x4a,0x64,0x50,0x66,0x12,0x69,0x76,0x65,0x9f,1,0x6c,0x2a,0x79, +0x11,0x69,0x6e,0x97,0x12,0x65,0x70,0x68,0x95,0x12,0x65,0x74,0x68,0x99,1,0x61, +0x30,0x68,0x14,0x61,0x6d,0x65,0x64,0x68,0x9d,0x13,0x6c,0x65,0x74,0x68,0x9b,0x15, +0x61,0x79,0x61,0x6c,0x61,0x6d,6,0x6e,0x2c,0x6e,0x34,0x72,0x5e,0x73,0x62,0x74, +0x11,0x74,0x61,0xa3,0x63,2,0x67,0x2e,0x6e,0x32,0x79,0x10,0x61,0xa3,0x60,0x10, +0x61,0xa3,0x5d,1,0x61,0xa3,0x5e,0x6e,0x10,0x61,0xa3,0x5f,0x10,0x61,0xa3,0x61, +0x11,0x73,0x61,0xa3,0x62,0x62,0x3c,0x6a,0x42,0x6c,0x10,0x6c,1,0x61,0xa3,0x5b, +0x6c,0x10,0x61,0xa3,0x5c,0x11,0x68,0x61,0xa3,0x59,0x10,0x61,0xa3,0x5a,0x11,0x65, +0x6d,0x51,0x10,0x61,1,0x66,0x37,0x6d,0x11,0x61,0x6c,0x39,1,0x61,0x40,0x65, +0x3e,1,0x68,0x28,0x74,0x10,0x68,0x45,0x40,0x13,0x67,0x6f,0x61,0x6c,0x43,2, +0x68,0x3b,0x6d,0x5c,0x6e,0x1a,0x69,0x66,0x69,0x72,0x6f,0x68,0x69,0x6e,0x67,0x79, +0x61,1,0x6b,0x2a,0x70,0x10,0x61,0xa3,0x65,0x15,0x69,0x6e,0x6e,0x61,0x79,0x61, +0xa3,0x64,0x1a,0x7a,0x61,0x6f,0x6e,0x68,0x65,0x68,0x67,0x6f,0x61,0x6c,0x3d,2, +0x61,0x3a,0x68,0x44,0x6e,0x17,0x6f,0x74,0x74,0x65,0x64,0x68,0x65,0x68,0x4b,1, +0x66,0x47,0x70,0x10,0x68,0x49,0x12,0x61,0x70,0x68,0x89,0x11,0x61,0x6d,0x4c,0x12, +0x61,0x64,0x68,0x4f,0x61,0x6e,0x62,0xa2,0x54,0x64,0xa2,0x70,0x65,0x31,0x66,2, +0x61,0x3e,0x65,0x4a,0x69,0x19,0x6e,0x61,0x6c,0x73,0x65,0x6d,0x6b,0x61,0x74,0x68, +0x35,0x15,0x72,0x73,0x69,0x79,0x65,0x68,0x8f,0x86,0x10,0x68,0x33,2,0x66,0x3c, +0x69,0x70,0x6c,1,0x61,0x28,0x65,0x10,0x66,0x27,0x11,0x70,0x68,0x25,0x14,0x72, +0x69,0x63,0x61,0x6e,2,0x66,0x30,0x6e,0x36,0x71,0x11,0x61,0x66,0xa3,0x58,0x11, +0x65,0x68,0xa3,0x56,0x12,0x6f,0x6f,0x6e,0xa3,0x57,0x10,0x6e,0x23,1,0x65,0x4a, +0x75,0x10,0x72,0x1f,0x75,0x73,0x68,0x61,0x73,0x6b,0x69,0x79,0x65,0x68,0x62,0x61, +0x72,0x72,0x65,0x65,0x8d,1,0x68,0x29,0x74,0x10,0x68,0x2b,0x11,0x61,0x6c,0x2c, +0x16,0x61,0x74,0x68,0x72,0x69,0x73,0x68,0x2f,7,0x6e,0x2e,0x6e,0x2c,0x72,0x3e, +0x74,0x56,0x75,0x21,0x18,0x6f,0x6e,0x6a,0x6f,0x69,0x6e,0x69,0x6e,0x67,0x21,0x28, +0x1a,0x69,0x67,0x68,0x74,0x6a,0x6f,0x69,0x6e,0x69,0x6e,0x67,0x29,0x2a,0x19,0x72, +0x61,0x6e,0x73,0x70,0x61,0x72,0x65,0x6e,0x74,0x2b,0x63,0x23,0x64,0x40,0x6a,0x56, +0x6c,0x26,0x19,0x65,0x66,0x74,0x6a,0x6f,0x69,0x6e,0x69,0x6e,0x67,0x27,0x24,0x19, +0x75,0x61,0x6c,0x6a,0x6f,0x69,0x6e,0x69,0x6e,0x67,0x25,0x19,0x6f,0x69,0x6e,0x63, +0x61,0x75,0x73,0x69,0x6e,0x67,0x23,0,0x13,0x6e,0xc0,0xd0,0x73,0x49,0x73,0x48, +0x75,0x78,0x77,0x84,0x78,0x9c,0x7a,0x10,0x77,0x58,1,0x6a,0x75,0x73,0x13,0x70, +0x61,0x63,0x65,0x59,4,0x61,0x51,0x67,0x53,0x70,0x28,0x75,0x30,0x79,0x57,0x54, +0x12,0x61,0x63,0x65,0x55,0x16,0x72,0x72,0x6f,0x67,0x61,0x74,0x65,0x53,0x15,0x6e, +0x6b,0x6e,0x6f,0x77,0x6e,0x21,1,0x6a,0x5d,0x6f,0x17,0x72,0x64,0x6a,0x6f,0x69, +0x6e,0x65,0x72,0x5d,0x10,0x78,0x21,0x6e,0x60,0x6f,0xa2,0x41,0x70,0xa2,0x50,0x71, +0xa2,0x6e,0x72,1,0x65,0x24,0x69,0x6f,0x1e,0x67,0x69,0x6f,0x6e,0x61,0x6c,0x69, +0x6e,0x64,0x69,0x63,0x61,0x74,0x6f,0x72,0x6f,4,0x65,0x3e,0x6c,0x5b,0x6f,0x46, +0x73,0x45,0x75,0x46,0x14,0x6d,0x65,0x72,0x69,0x63,0x47,0x15,0x78,0x74,0x6c,0x69, +0x6e,0x65,0x5b,0x17,0x6e,0x73,0x74,0x61,0x72,0x74,0x65,0x72,0x45,0x10,0x70,0x48, +0x1c,0x65,0x6e,0x70,0x75,0x6e,0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0x49,1, +0x6f,0x3e,0x72,0x4c,0x1a,0x65,0x66,0x69,0x78,0x6e,0x75,0x6d,0x65,0x72,0x69,0x63, +0x4d,0x4a,0x1b,0x73,0x74,0x66,0x69,0x78,0x6e,0x75,0x6d,0x65,0x72,0x69,0x63,0x4b, +0x10,0x75,0x4e,0x16,0x6f,0x74,0x61,0x74,0x69,0x6f,0x6e,0x4f,0x68,0x7b,0x68,0x50, +0x69,0x86,0x6a,0xa2,0x61,0x6c,0xa2,0x65,0x6d,0x1c,0x61,0x6e,0x64,0x61,0x74,0x6f, +0x72,0x79,0x62,0x72,0x65,0x61,0x6b,0x2d,4,0x32,0x5f,0x33,0x61,0x65,0x34,0x6c, +0x6d,0x79,0x3a,0x13,0x70,0x68,0x65,0x6e,0x3b,0x19,0x62,0x72,0x65,0x77,0x6c,0x65, +0x74,0x74,0x65,0x72,0x6d,2,0x64,0x28,0x6e,0x3c,0x73,0x41,0x3c,0x18,0x65,0x6f, +0x67,0x72,0x61,0x70,0x68,0x69,0x63,0x3d,0x3e,1,0x66,0x3e,0x73,0x11,0x65,0x70, +1,0x61,0x22,0x65,0x14,0x72,0x61,0x62,0x6c,0x65,0x3f,0x18,0x69,0x78,0x6e,0x75, +0x6d,0x65,0x72,0x69,0x63,0x41,2,0x6c,0x63,0x74,0x65,0x76,0x67,1,0x66,0x43, +0x69,0x15,0x6e,0x65,0x66,0x65,0x65,0x64,0x43,0x61,0x40,0x62,0x70,0x63,0xa2,0x55, +0x65,0xa2,0xdb,0x67,0x10,0x6c,0x38,0x11,0x75,0x65,0x39,2,0x69,0x23,0x6c,0x34, +0x6d,0x16,0x62,0x69,0x67,0x75,0x6f,0x75,0x73,0x23,0x24,0x17,0x70,0x68,0x61,0x62, +0x65,0x74,0x69,0x63,0x25,4,0x32,0x27,0x61,0x29,0x62,0x2b,0x6b,0x2d,0x72,0x12, +0x65,0x61,0x6b,2,0x61,0x36,0x62,0x3e,0x73,0x15,0x79,0x6d,0x62,0x6f,0x6c,0x73, +0x57,0x13,0x66,0x74,0x65,0x72,0x29,1,0x65,0x2a,0x6f,0x11,0x74,0x68,0x27,0x13, +0x66,0x6f,0x72,0x65,0x2b,7,0x6d,0x51,0x6d,0x33,0x6f,0x28,0x70,0x69,0x72,0x35, +1,0x6d,0x76,0x6e,1,0x64,0x3c,0x74,0x1a,0x69,0x6e,0x67,0x65,0x6e,0x74,0x62, +0x72,0x65,0x61,0x6b,0x2f,0x15,0x69,0x74,0x69,0x6f,0x6e,0x61,0x1f,0x6c,0x6a,0x61, +0x70,0x61,0x6e,0x65,0x73,0x65,0x73,0x74,0x61,0x72,0x74,0x65,0x72,0x6b,1,0x62, +0x3a,0x70,0x19,0x6c,0x65,0x78,0x63,0x6f,0x6e,0x74,0x65,0x78,0x74,0x51,0x18,0x69, +0x6e,0x69,0x6e,0x67,0x6d,0x61,0x72,0x6b,0x33,0x61,0x6a,0x62,0x2f,0x6a,0x6b,0x6c, +0x30,0x13,0x6f,0x73,0x65,0x70,1,0x61,0x38,0x75,0x18,0x6e,0x63,0x74,0x75,0x61, +0x74,0x69,0x6f,0x6e,0x31,0x18,0x72,0x65,0x6e,0x74,0x68,0x65,0x73,0x69,0x73,0x69, +0x1b,0x72,0x72,0x69,0x61,0x67,0x65,0x72,0x65,0x74,0x75,0x72,0x6e,0x35,2,0x62, +0x3e,0x6d,0x46,0x78,0x36,0x18,0x63,0x6c,0x61,0x6d,0x61,0x74,0x69,0x6f,0x6e,0x37, +0x70,0x12,0x61,0x73,0x65,0x71,0x72,0x16,0x6f,0x64,0x69,0x66,0x69,0x65,0x72,0x73, +1,0x64,0x42,0x6e,1,0x6f,0x32,0x75,0x26,0x14,0x6d,0x65,0x72,0x69,0x63,0x27, +0x11,0x6e,0x65,0x21,1,0x65,0x2e,0x69,0x24,0x12,0x67,0x69,0x74,0x25,0x22,0x14, +0x63,0x69,0x6d,0x61,0x6c,0x23,0,0x18,0x6e,0xc4,0x6f,0x74,0xc1,0x91,0x77,0x96, +0x77,0xa2,0x4c,0x78,0xa2,0x70,0x79,0xa2,0x7a,0x7a,6,0x73,0x1e,0x73,0x34,0x78, +0x42,0x79,0x48,0x7a,0x11,0x7a,0x7a,0xa3,0x67,0x10,0x79,1,0x65,0xa3,0xae,0x6d, +0xa3,0x81,0x11,0x78,0x78,0xa3,0x66,0x11,0x79,0x79,0x21,0x61,0x30,0x69,0x58,0x6d, +0x11,0x74,0x68,0xa3,0x80,0x10,0x6e,1,0x61,0x26,0x62,0xa3,0xb1,0x1a,0x62,0x61, +0x7a,0x61,0x72,0x73,0x71,0x75,0x61,0x72,0x65,0xa3,0xb1,0x11,0x6e,0x68,0x23,2, +0x61,0x30,0x63,0x5a,0x6f,0x11,0x6c,0x65,0xa3,0x9b,1,0x6e,0x3c,0x72,0x10,0x61, +0xa2,0x92,0x15,0x6e,0x67,0x63,0x69,0x74,0x69,0xa3,0x92,0x12,0x63,0x68,0x6f,0xa3, +0xbc,0x11,0x68,0x6f,0xa3,0xbc,1,0x70,0x2c,0x73,0x11,0x75,0x78,0xa3,0x65,0x11, +0x65,0x6f,0x9b,1,0x65,0x2c,0x69,0x72,0x11,0x69,0x69,0x73,0x11,0x7a,0x69,0xa2, +0xc0,0x11,0x64,0x69,0xa3,0xc0,0x74,0x66,0x75,0xa2,0xde,0x76,1,0x61,0x48,0x69, +1,0x73,0x38,0x74,0x10,0x68,0xa2,0xc5,0x13,0x6b,0x75,0x71,0x69,0xa3,0xc5,0x10, +0x70,0xa3,0x64,0x10,0x69,0xa2,0x63,0x10,0x69,0xa3,0x63,7,0x68,0x3e,0x68,0x34, +0x69,0x48,0x6e,0x86,0x6f,0x11,0x74,0x6f,0xa3,0xc4,0x10,0x61,1,0x61,0x24,0x69, +0x6d,0x6a,0x11,0x6e,0x61,0x6b,2,0x62,0x3a,0x66,0x4a,0x72,0x10,0x68,0xa2,0x9e, +0x12,0x75,0x74,0x61,0xa3,0x9e,1,0x65,0x24,0x74,0x6f,0x12,0x74,0x61,0x6e,0x6f, +0x14,0x69,0x6e,0x61,0x67,0x68,0x99,0x11,0x73,0x61,0xa3,0xc3,0x61,0x36,0x65,0xa2, +0x65,0x66,0xa2,0x71,0x67,0x11,0x6c,0x67,0x75,6,0x6c,0x28,0x6c,0x32,0x6d,0x38, +0x6e,0x44,0x76,0x10,0x74,0xa3,0x7f,1,0x65,0x89,0x75,0x97,1,0x69,0x24,0x6c, +0x67,0x10,0x6c,0x67,0x10,0x67,0xa2,0x9a,1,0x73,0x2a,0x75,0x10,0x74,0xa3,0x9a, +0x10,0x61,0xa3,0xc3,0x67,0x36,0x69,0x52,0x6b,0x10,0x72,0xa2,0x99,0x10,0x69,0xa3, +0x99,1,0x61,0x30,0x62,0x7a,0x13,0x61,0x6e,0x77,0x61,0x7b,0x12,0x6c,0x6f,0x67, +0x75,2,0x6c,0x32,0x74,0x34,0x76,0x12,0x69,0x65,0x74,0xa3,0x7f,0x10,0x65,0x89, +0x12,0x68,0x61,0x6d,0xa3,0x6a,1,0x6c,0x2a,0x6e,0x10,0x67,0xa3,0x62,0x10,0x75, +0x68,0x11,0x67,0x75,0x69,0x11,0x6e,0x67,0x99,1,0x67,0x32,0x6e,0x14,0x6b,0x6e, +0x6f,0x77,0x6e,0xa3,0x67,0x11,0x61,0x72,0x8a,0x13,0x69,0x74,0x69,0x63,0x8b,0x71, +0xc1,0x13,0x71,0xa2,0xde,0x72,0xa2,0xe3,0x73,6,0x69,0x8a,0x69,0x72,0x6f,0xa2, +0x4c,0x75,0xa2,0x75,0x79,1,0x6c,0x46,0x72,4,0x63,0x65,0x65,0xa3,0x5f,0x69, +0x2c,0x6a,0xa3,0x60,0x6e,0xa3,0x61,0x11,0x61,0x63,0x65,0x10,0x6f,0x94,0x16,0x74, +0x69,0x6e,0x61,0x67,0x72,0x69,0x95,2,0x64,0x3c,0x67,0x4c,0x6e,1,0x64,0xa3, +0x91,0x68,0x62,0x12,0x61,0x6c,0x61,0x63,0x10,0x64,0xa2,0xa6,0x12,0x68,0x61,0x6d, +0xa3,0xa6,0x17,0x6e,0x77,0x72,0x69,0x74,0x69,0x6e,0x67,0xa3,0x70,2,0x67,0x3a, +0x72,0x52,0x79,0x10,0x6f,0xa2,0xb0,0x12,0x6d,0x62,0x6f,0xa3,0xb0,1,0x64,0x26, +0x6f,0xa3,0xb8,0xa2,0xb7,0x12,0x69,0x61,0x6e,0xa3,0xb7,0x10,0x61,0xa2,0x98,0x16, +0x73,0x6f,0x6d,0x70,0x65,0x6e,0x67,0xa3,0x98,0x11,0x6e,0x64,0xa2,0x71,0x14,0x61, +0x6e,0x65,0x73,0x65,0xa3,0x71,0x61,0x5c,0x67,0xa2,0x43,0x68,1,0x61,0x2a,0x72, +0x10,0x64,0xa3,0x97,2,0x72,0x28,0x76,0x30,0x77,0x87,0x12,0x61,0x64,0x61,0xa3, +0x97,0x12,0x69,0x61,0x6e,0x87,2,0x6d,0x40,0x72,0x58,0x75,0x10,0x72,0xa2,0x6f, +0x15,0x61,0x73,0x68,0x74,0x72,0x61,0xa3,0x6f,1,0x61,0x26,0x72,0xa3,0x7e,0x14, +0x72,0x69,0x74,0x61,0x6e,0xa3,0x7e,1,0x61,0xa3,0x5e,0x62,0xa3,0x85,0x11,0x6e, +0x77,0xa3,0x70,0x11,0x61,0x61,1,0x63,0x2f,0x69,0x23,3,0x65,0x3e,0x6a,0x48, +0x6f,0x4e,0x75,0x10,0x6e,1,0x69,0x24,0x72,0x61,0x10,0x63,0x61,0x13,0x6a,0x61, +0x6e,0x67,0xa3,0x6e,0x11,0x6e,0x67,0xa3,0x6e,1,0x68,0x2a,0x72,0x10,0x6f,0xa3, +0x5d,0x10,0x67,0xa3,0xb6,0x6e,0xa2,0x83,0x6f,0xa4,1,0x70,5,0x6c,0x1e,0x6c, +0x44,0x72,0x4a,0x73,0x1b,0x61,0x6c,0x74,0x65,0x72,0x70,0x61,0x68,0x6c,0x61,0x76, +0x69,0xa3,0x7b,0x11,0x72,0x64,0xa3,0x5c,0x11,0x74,0x69,0xa3,0x7d,0x61,0x7c,0x65, +0xa2,0x54,0x68,3,0x61,0x3e,0x6c,0x4e,0x6e,0x5e,0x6f,0x16,0x65,0x6e,0x69,0x63, +0x69,0x61,0x6e,0xa3,0x5b,0x10,0x67,0xa2,0x5a,0x12,0x73,0x70,0x61,0xa3,0x5a,2, +0x69,0xa3,0x7a,0x70,0xa3,0x7b,0x76,0xa3,0x7c,0x10,0x78,0xa3,0x5b,2,0x68,0x3e, +0x6c,0x50,0x75,0x10,0x63,0xa2,0xa5,0x14,0x69,0x6e,0x68,0x61,0x75,0xa3,0xa5,0x17, +0x61,0x77,0x68,0x68,0x6d,0x6f,0x6e,0x67,0xa3,0x4b,0x10,0x6d,0xa2,0x90,0x14,0x79, +0x72,0x65,0x6e,0x65,0xa3,0x90,0x11,0x72,0x6d,0xa3,0x59,6,0x6b,0x36,0x6b,0x56, +0x73,0x6e,0x75,0x74,0x79,0x11,0x69,0x61,0x1f,0x6b,0x65,0x6e,0x67,0x70,0x75,0x61, +0x63,0x68,0x75,0x65,0x68,0x6d,0x6f,0x6e,0x67,0xa3,0xba,1,0x67,0x2e,0x6f,0xa2, +0x57,0x10,0x6f,0xa3,0x57,0x10,0x62,0xa3,0x84,0x11,0x68,0x75,0xa3,0x96,0x12,0x73, +0x68,0x75,0xa3,0x96,0x61,0x42,0x62,0x9e,0x65,0x10,0x77,1,0x61,0xa3,0xaa,0x74, +0x14,0x61,0x69,0x6c,0x75,0x65,0x97,3,0x62,0x32,0x67,0x40,0x6e,0x56,0x72,0x10, +0x62,0xa3,0x8e,0x15,0x61,0x74,0x61,0x65,0x61,0x6e,0xa3,0x8f,0x10,0x6d,0xa2,0xc7, +0x15,0x75,0x6e,0x64,0x61,0x72,0x69,0xa3,0xc7,0x10,0x64,0xa2,0xbb,0x16,0x69,0x6e, +0x61,0x67,0x61,0x72,0x69,0xa3,0xbb,0x11,0x61,0x74,0xa3,0x8f,4,0x67,0x3c,0x6c, +0x4e,0x72,0xa2,0x8e,0x73,0xa2,0x9c,0x75,0x11,0x67,0x72,0xa3,0xc2,1,0x61,0x2a, +0x68,0x11,0x61,0x6d,0x5b,0x10,0x6d,0x5b,1,0x63,0xa2,0x6a,0x64,6,0x70,0x41, +0x70,0x3a,0x73,0x58,0x74,0x86,0x75,0x14,0x79,0x67,0x68,0x75,0x72,0xa3,0xc2,0x11, +0x65,0x72,1,0x6d,0x2c,0x73,0x12,0x69,0x61,0x6e,0x9b,0x11,0x69,0x63,0xa3,0x59, +0x10,0x6f,1,0x67,0x3a,0x75,0x18,0x74,0x68,0x61,0x72,0x61,0x62,0x69,0x61,0x6e, +0xa3,0x85,0x13,0x64,0x69,0x61,0x6e,0xa3,0xb8,0x14,0x75,0x72,0x6b,0x69,0x63,0xa3, +0x58,0x68,0x42,0x69,0x54,0x6e,0x1a,0x6f,0x72,0x74,0x68,0x61,0x72,0x61,0x62,0x69, +0x61,0x6e,0xa3,0x8e,0x17,0x75,0x6e,0x67,0x61,0x72,0x69,0x61,0x6e,0xa3,0x4c,0x14, +0x74,0x61,0x6c,0x69,0x63,0x5d,1,0x68,0x26,0x6b,0xa3,0x6d,0x12,0x69,0x6b,0x69, +0xa3,0x6d,2,0x69,0x2c,0x6b,0x30,0x79,0x10,0x61,0x5f,0x11,0x79,0x61,0x5f,0x10, +0x68,0xa3,0x58,2,0x61,0x36,0x67,0x3c,0x6d,0x10,0x61,0x84,0x12,0x6e,0x79,0x61, +0x85,0x11,0x67,0x65,0xa3,0xab,0x10,0x65,0xa3,0xab,0x68,0xc3,0x15,0x6b,0xc2,0x2c, +0x6b,0xa4,0x17,0x6c,0xa4,0xba,0x6d,8,0x6f,0x46,0x6f,0x48,0x72,0x74,0x74,0x80, +0x75,0x86,0x79,1,0x61,0x28,0x6d,0x10,0x72,0x59,0x13,0x6e,0x6d,0x61,0x72,0x59, +2,0x64,0x2e,0x6e,0x32,0x6f,0x10,0x6e,0xa3,0x72,0x10,0x69,0xa3,0xa3,0x10,0x67, +0x56,0x14,0x6f,0x6c,0x69,0x61,0x6e,0x57,0x10,0x6f,0xa2,0x95,0x10,0x6f,0xa3,0x95, +0x11,0x65,0x69,0xa3,0x73,0x11,0x6c,0x74,0xa2,0xa4,0x12,0x61,0x6e,0x69,0xa3,0xa4, +0x61,0x36,0x65,0xa2,0x67,0x69,0xa2,0xbd,0x6c,0x11,0x79,0x6d,0x55,6,0x6e,0x38, +0x6e,0x32,0x72,0x5c,0x73,0x6c,0x79,0x10,0x61,0xa3,0x55,1,0x64,0x38,0x69,0xa2, +0x79,0x15,0x63,0x68,0x61,0x65,0x61,0x6e,0xa3,0x79,0xa2,0x54,0x12,0x61,0x69,0x63, +0xa3,0x54,0x10,0x63,0xa2,0xa9,0x12,0x68,0x65,0x6e,0xa3,0xa9,0x18,0x61,0x72,0x61, +0x6d,0x67,0x6f,0x6e,0x64,0x69,0xa3,0xaf,0x68,0x36,0x6b,0x4c,0x6c,0x15,0x61,0x79, +0x61,0x6c,0x61,0x6d,0x55,1,0x61,0x26,0x6a,0xa3,0xa0,0x13,0x6a,0x61,0x6e,0x69, +0xa3,0xa0,0x10,0x61,0xa2,0xb4,0x12,0x73,0x61,0x72,0xa3,0xb4,3,0x64,0x78,0x65, +0x94,0x6e,0xa2,0x42,0x72,1,0x63,0xa3,0x8d,0x6f,0xa2,0x56,0x13,0x69,0x74,0x69, +0x63,1,0x63,0x3c,0x68,0x19,0x69,0x65,0x72,0x6f,0x67,0x6c,0x79,0x70,0x68,0x73, +0xa3,0x56,0x15,0x75,0x72,0x73,0x69,0x76,0x65,0xa3,0x8d,1,0x65,0x26,0x66,0xa3, +0xb5,0x16,0x66,0x61,0x69,0x64,0x72,0x69,0x6e,0xa3,0xb5,0x17,0x74,0x65,0x69,0x6d, +0x61,0x79,0x65,0x6b,0xa3,0x73,0x10,0x64,0xa2,0x8c,0x17,0x65,0x6b,0x69,0x6b,0x61, +0x6b,0x75,0x69,0xa3,0x8c,0x11,0x61,0x6f,0xa3,0x5c,6,0x6e,0x1a,0x6e,0x34,0x6f, +0x38,0x70,0x3e,0x74,0x11,0x68,0x69,0xa3,0x78,0x11,0x64,0x61,0x4b,0x11,0x72,0x65, +0xa3,0x77,0x11,0x65,0x6c,0xa3,0x8a,0x61,0x32,0x68,0xa2,0x44,0x69,0x11,0x74,0x73, +0xa3,0xbf,5,0x74,0x23,0x74,0x34,0x77,0x56,0x79,0x13,0x61,0x68,0x6c,0x69,0xa3, +0x4f,0x14,0x61,0x6b,0x61,0x6e,0x61,0x4c,0x19,0x6f,0x72,0x68,0x69,0x72,0x61,0x67, +0x61,0x6e,0x61,0x8d,0x10,0x69,0xa3,0xc6,0x69,0x38,0x6c,0x40,0x6e,1,0x61,0x4d, +0x6e,0x12,0x61,0x64,0x61,0x4b,0x12,0x74,0x68,0x69,0xa3,0x78,0x10,0x69,0xa3,0x4f, +4,0x61,0x40,0x69,0x52,0x6d,0x70,0x6f,0x7c,0x75,0x15,0x64,0x61,0x77,0x61,0x64, +0x69,0xa3,0x91,0x10,0x72,0x92,0x15,0x6f,0x73,0x68,0x74,0x68,0x69,0x93,0x1d,0x74, +0x61,0x6e,0x73,0x6d,0x61,0x6c,0x6c,0x73,0x63,0x72,0x69,0x70,0x74,0xa3,0xbf,1, +0x65,0x24,0x72,0x4f,0x10,0x72,0x4f,0x10,0x6a,0xa2,0x9d,0x11,0x6b,0x69,0xa3,0x9d, +4,0x61,0x5c,0x65,0x90,0x69,0xa0,0x6f,0xa2,0x5d,0x79,1,0x63,0x34,0x64,0x10, +0x69,0xa2,0x6c,0x11,0x61,0x6e,0xa3,0x6c,0x10,0x69,0xa2,0x6b,0x11,0x61,0x6e,0xa3, +0x6b,2,0x6e,0x42,0x6f,0x46,0x74,3,0x66,0xa3,0x50,0x67,0xa3,0x51,0x69,0x24, +0x6e,0x53,0x10,0x6e,0x53,0x10,0x61,0xa3,0x6a,0x50,0x10,0x6f,0x51,0x11,0x70,0x63, +0xa2,0x52,0x11,0x68,0x61,0xa3,0x52,2,0x6d,0x2e,0x6e,0x36,0x73,0x10,0x75,0xa3, +0x83,0x10,0x62,0x80,0x10,0x75,0x81,2,0x61,0xa3,0x53,0x62,0x83,0x65,0x11,0x61, +0x72,1,0x61,0xa3,0x53,0x62,0x83,0x11,0x6d,0x61,0xa3,0x8b,0x68,0x6e,0x69,0xa2, +0x95,0x6a,2,0x61,0x30,0x70,0x52,0x75,0x11,0x72,0x63,0xa3,0x94,1,0x6d,0x38, +0x76,0x10,0x61,0xa2,0x4e,0x13,0x6e,0x65,0x73,0x65,0xa3,0x4e,0x10,0x6f,0xa3,0xad, +0x11,0x61,0x6e,0xa3,0x69,6,0x6c,0x1e,0x6c,0x34,0x6d,0x3a,0x72,0x48,0x75,0x11, +0x6e,0x67,0xa3,0x4c,0x11,0x75,0x77,0xa3,0x9c,0x10,0x6e,1,0x67,0xa3,0x4b,0x70, +0xa3,0xba,0x11,0x6b,0x74,0x8d,0x61,0x3c,0x65,0xa2,0x43,0x69,0x11,0x72,0x61,0x48, +0x13,0x67,0x61,0x6e,0x61,0x49,1,0x6e,0x34,0x74,0x10,0x72,0xa2,0xa2,0x11,0x61, +0x6e,0xa3,0xa2,0x42,6,0x6f,0xe,0x6f,0x77,0x73,0xa3,0x49,0x74,0xa3,0x4a,0x75, +0x12,0x6e,0x6f,0x6f,0x77,0x62,0xa3,0xac,0x67,0x3e,0x69,0x42,0x19,0x66,0x69,0x72, +0x6f,0x68,0x69,0x6e,0x67,0x79,0x61,0xa3,0xb6,0x44,0x11,0x75,0x6c,0x45,0x11,0x62, +0x72,0x46,0x11,0x65,0x77,0x47,2,0x6d,0x2e,0x6e,0x4a,0x74,0x11,0x61,0x6c,0x5d, +0x1c,0x70,0x65,0x72,0x69,0x61,0x6c,0x61,0x72,0x61,0x6d,0x61,0x69,0x63,0xa3,0x74, +2,0x64,0x66,0x68,0x6a,0x73,0x1b,0x63,0x72,0x69,0x70,0x74,0x69,0x6f,0x6e,0x61, +0x6c,0x70,0x61,1,0x68,0x32,0x72,0x14,0x74,0x68,0x69,0x61,0x6e,0xa3,0x7d,0x13, +0x6c,0x61,0x76,0x69,0xa3,0x7a,0x10,0x73,0xa3,0x4d,0x15,0x65,0x72,0x69,0x74,0x65, +0x64,0x23,0x64,0xc1,0xd,0x64,0xa2,0x7a,0x65,0xa2,0xc1,0x67,4,0x65,0x82,0x6c, +0x9a,0x6f,0xa2,0x46,0x72,0xa2,0x55,0x75,2,0x6a,0x3c,0x6e,0x4e,0x72,1,0x6d, +0x24,0x75,0x41,0x13,0x75,0x6b,0x68,0x69,0x41,1,0x61,0x24,0x72,0x3f,0x13,0x72, +0x61,0x74,0x69,0x3f,0x18,0x6a,0x61,0x6c,0x61,0x67,0x6f,0x6e,0x64,0x69,0xa3,0xb3, +0x10,0x6f,1,0x6b,0xa3,0x48,0x72,0x38,0x13,0x67,0x69,0x61,0x6e,0x39,0x11,0x61, +0x67,0x90,0x15,0x6f,0x6c,0x69,0x74,0x69,0x63,0x91,1,0x6e,0x30,0x74,0x10,0x68, +0x3a,0x11,0x69,0x63,0x3b,1,0x67,0xa3,0xb3,0x6d,0xa3,0xaf,1,0x61,0x32,0x65, +1,0x65,0x24,0x6b,0x3d,0x10,0x6b,0x3d,0x10,0x6e,0xa2,0x89,0x12,0x74,0x68,0x61, +0xa3,0x89,4,0x65,0x46,0x69,0x6c,0x6f,0x8c,0x73,0x9a,0x75,0x11,0x70,0x6c,0xa2, +0x87,0x13,0x6f,0x79,0x61,0x6e,0xa3,0x87,1,0x73,0x38,0x76,0x10,0x61,0x34,0x15, +0x6e,0x61,0x67,0x61,0x72,0x69,0x35,0x13,0x65,0x72,0x65,0x74,0x33,1,0x61,0x36, +0x76,0x16,0x65,0x73,0x61,0x6b,0x75,0x72,0x75,0xa3,0xbe,0x10,0x6b,0xa3,0xbe,0x11, +0x67,0x72,0xa2,0xb2,0x10,0x61,0xa3,0xb2,0x11,0x72,0x74,0x33,2,0x67,0x3a,0x6c, +0x72,0x74,0x11,0x68,0x69,0x36,0x13,0x6f,0x70,0x69,0x63,0x37,0x10,0x79,2,0x64, +0xa3,0x45,0x68,0xa3,0x46,0x70,0xa2,0x47,0x1e,0x74,0x69,0x61,0x6e,0x68,0x69,0x65, +0x72,0x6f,0x67,0x6c,0x79,0x70,0x68,0x73,0xa3,0x47,1,0x62,0x36,0x79,0x10,0x6d, +0xa2,0xb9,0x12,0x61,0x69,0x63,0xa3,0xb9,0x10,0x61,0xa2,0x88,0x12,0x73,0x61,0x6e, +0xa3,0x88,0x61,0xa2,0xc9,0x62,0xa4,0x2e,0x63,6,0x6f,0x52,0x6f,0x76,0x70,0x92, +0x75,0xa2,0x41,0x79,1,0x70,0x3e,0x72,2,0x69,0x2a,0x6c,0x31,0x73,0xa3,0x44, +0x13,0x6c,0x6c,0x69,0x63,0x31,0x10,0x72,1,0x69,0x34,0x6f,0x15,0x6d,0x69,0x6e, +0x6f,0x61,0x6e,0xa3,0xc1,0x11,0x6f,0x74,0x7f,1,0x6d,0x30,0x70,0x10,0x74,0x2e, +0x11,0x69,0x63,0x2f,0x12,0x6d,0x6f,0x6e,0x21,1,0x6d,0x28,0x72,0x10,0x74,0x7f, +0x10,0x6e,0xa3,0xc1,0x16,0x6e,0x65,0x69,0x66,0x6f,0x72,0x6d,0xa3,0x65,0x61,0x32, +0x68,0xa2,0x41,0x69,0x11,0x72,0x74,0xa3,0x43,3,0x6b,0x4c,0x6e,0x50,0x72,0x76, +0x75,0x1d,0x63,0x61,0x73,0x69,0x61,0x6e,0x61,0x6c,0x62,0x61,0x6e,0x69,0x61,0x6e, +0xa3,0x9f,0x10,0x6d,0xa3,0x76,1,0x61,0x24,0x73,0x71,0x1d,0x64,0x69,0x61,0x6e, +0x61,0x62,0x6f,0x72,0x69,0x67,0x69,0x6e,0x61,0x6c,0x71,0x10,0x69,0xa2,0x68,0x11, +0x61,0x6e,0xa3,0x68,3,0x61,0x32,0x65,0x44,0x6f,0x52,0x72,0x10,0x73,0xa3,0xbd, +1,0x6b,0x26,0x6d,0xa3,0x42,0x11,0x6d,0x61,0xa3,0x76,0x10,0x72,0x2c,0x13,0x6f, +0x6b,0x65,0x65,0x2d,0x16,0x72,0x61,0x73,0x6d,0x69,0x61,0x6e,0xa3,0xbd,6,0x68, +0x4a,0x68,0x48,0x6e,0x4e,0x72,0x76,0x76,1,0x65,0x2a,0x73,0x10,0x74,0xa3,0x75, +0x13,0x73,0x74,0x61,0x6e,0xa3,0x75,0x11,0x6f,0x6d,0xa3,0xa1,0x11,0x61,0x74,0x1f, +0x6f,0x6c,0x69,0x61,0x6e,0x68,0x69,0x65,0x72,0x6f,0x67,0x6c,0x79,0x70,0x68,0x73, +0xa3,0x9c,1,0x61,0x3e,0x6d,2,0x65,0x2a,0x69,0xa3,0x74,0x6e,0x27,0x13,0x6e, +0x69,0x61,0x6e,0x27,0x10,0x62,0x24,0x11,0x69,0x63,0x25,0x64,0x30,0x66,0x44,0x67, +0x11,0x68,0x62,0xa3,0x9f,0x10,0x6c,1,0x61,0x26,0x6d,0xa3,0xa7,0x10,0x6d,0xa3, +0xa7,0x11,0x61,0x6b,0xa3,0x93,6,0x6c,0x3c,0x6c,0x52,0x6f,0x56,0x72,0x66,0x75, +1,0x67,0x30,0x68,1,0x64,0x79,0x69,0x10,0x64,0x79,0x10,0x69,0x8e,0x13,0x6e, +0x65,0x73,0x65,0x8f,0x11,0x69,0x73,0xa1,0x11,0x70,0x6f,0x2a,0x13,0x6d,0x6f,0x66, +0x6f,0x2b,0x10,0x61,1,0x68,0x2e,0x69,0x7c,0x12,0x6c,0x6c,0x65,0x7d,0xa2,0x41, +0x11,0x6d,0x69,0xa3,0x41,0x61,0x48,0x65,0x9c,0x68,1,0x61,0x2a,0x6b,0x10,0x73, +0xa3,0xa8,0x15,0x69,0x6b,0x73,0x75,0x6b,0x69,0xa3,0xa8,3,0x6c,0x3a,0x6d,0x48, +0x73,0x54,0x74,1,0x61,0x24,0x6b,0x9f,0x10,0x6b,0x9f,0x10,0x69,0x9c,0x13,0x6e, +0x65,0x73,0x65,0x9d,0x10,0x75,0xa2,0x82,0x10,0x6d,0xa3,0x82,0x10,0x73,0xa2,0x86, +0x13,0x61,0x76,0x61,0x68,0xa3,0x86,0x11,0x6e,0x67,0x28,0x12,0x61,0x6c,0x69,0x29, +3,0x6c,0x42,0x6e,0x90,0x74,0xa2,0x46,0x76,0x24,0x17,0x6f,0x77,0x65,0x6c,0x6a, +0x61,0x6d,0x6f,0x25,0x22,1,0x65,0x54,0x76,0x28,1,0x73,0x38,0x74,0x2a,0x17, +0x73,0x79,0x6c,0x6c,0x61,0x62,0x6c,0x65,0x2b,0x16,0x79,0x6c,0x6c,0x61,0x62,0x6c, +0x65,0x29,0x18,0x61,0x64,0x69,0x6e,0x67,0x6a,0x61,0x6d,0x6f,0x23,1,0x61,0x21, +0x6f,0x1a,0x74,0x61,0x70,0x70,0x6c,0x69,0x63,0x61,0x62,0x6c,0x65,0x21,0x26,0x1a, +0x72,0x61,0x69,0x6c,0x69,0x6e,0x67,0x6a,0x61,0x6d,0x6f,0x27,1,0x6e,0x2c,0x79, +0x22,0x11,0x65,0x73,0x23,0x20,0x10,0x6f,0x21,1,0x6e,0x2c,0x79,0x22,0x11,0x65, +0x73,0x23,0x20,0x10,0x6f,0x21,2,0x6d,0x30,0x6e,0x3a,0x79,0x22,0x11,0x65,0x73, +0x23,0x24,0x13,0x61,0x79,0x62,0x65,0x25,0x20,0x10,0x6f,0x21,2,0x6d,0x30,0x6e, +0x3a,0x79,0x22,0x11,0x65,0x73,0x23,0x24,0x13,0x61,0x79,0x62,0x65,0x25,0x20,0x10, +0x6f,0x21,0xb,0x72,0x39,0x76,0xc,0x76,0x33,0x78,0x2a,0x7a,0x11,0x77,0x6a,0x43, +0x10,0x78,0x21,0x72,0x28,0x73,0x50,0x74,0x31,1,0x65,0x24,0x69,0x39,0x1e,0x67, +0x69,0x6f,0x6e,0x61,0x6c,0x69,0x6e,0x64,0x69,0x63,0x61,0x74,0x6f,0x72,0x39,1, +0x6d,0x35,0x70,0x18,0x61,0x63,0x69,0x6e,0x67,0x6d,0x61,0x72,0x6b,0x35,0x6c,0x1f, +0x6c,0x3c,0x6f,0x4a,0x70,1,0x70,0x37,0x72,0x14,0x65,0x70,0x65,0x6e,0x64,0x37, +0x28,1,0x66,0x2b,0x76,0x2c,0x10,0x74,0x2f,0x13,0x74,0x68,0x65,0x72,0x21,0x63, +0x4c,0x65,0x64,0x67,1,0x61,0x3a,0x6c,0x19,0x75,0x65,0x61,0x66,0x74,0x65,0x72, +0x7a,0x77,0x6a,0x41,0x10,0x7a,0x41,2,0x6e,0x23,0x6f,0x24,0x72,0x25,0x14,0x6e, +0x74,0x72,0x6f,0x6c,0x23,2,0x62,0x34,0x6d,0x4e,0x78,0x26,0x13,0x74,0x65,0x6e, +0x64,0x27,0x3a,1,0x61,0x24,0x67,0x3d,0x11,0x73,0x65,0x3a,0x12,0x67,0x61,0x7a, +0x3d,0x3e,0x16,0x6f,0x64,0x69,0x66,0x69,0x65,0x72,0x3f,9,0x6e,0x4a,0x6e,0x34, +0x6f,0x44,0x73,0x60,0x75,0x94,0x78,0x10,0x78,0x21,0x10,0x75,0x2a,0x14,0x6d,0x65, +0x72,0x69,0x63,0x2b,1,0x6c,0x2c,0x74,0x12,0x68,0x65,0x72,0x21,0x14,0x65,0x74, +0x74,0x65,0x72,0x2d,3,0x63,0x36,0x65,0x46,0x70,0x31,0x74,0x32,0x12,0x65,0x72, +0x6d,0x33,0x3c,0x16,0x6f,0x6e,0x74,0x69,0x6e,0x75,0x65,0x3d,0x2e,0x10,0x70,0x2f, +0x10,0x70,0x34,0x12,0x70,0x65,0x72,0x35,0x61,0x46,0x63,0x52,0x65,0x64,0x66,0x72, +0x6c,2,0x65,0x2d,0x66,0x3b,0x6f,0x28,0x12,0x77,0x65,0x72,0x29,0x10,0x74,0x22, +0x12,0x65,0x72,0x6d,0x23,1,0x6c,0x24,0x72,0x37,0x24,0x12,0x6f,0x73,0x65,0x25, +0x10,0x78,0x38,0x13,0x74,0x65,0x6e,0x64,0x39,0x10,0x6f,0x26,0x13,0x72,0x6d,0x61, +0x74,0x27,0,0x10,0x6c,0x88,0x72,0x40,0x72,0x36,0x73,0x5e,0x77,0x7a,0x78,0x8a, +0x7a,0x11,0x77,0x6a,0x4b,1,0x65,0x24,0x69,0x3b,0x1e,0x67,0x69,0x6f,0x6e,0x61, +0x6c,0x69,0x6e,0x64,0x69,0x63,0x61,0x74,0x6f,0x72,0x3b,1,0x69,0x24,0x71,0x3f, +0x18,0x6e,0x67,0x6c,0x65,0x71,0x75,0x6f,0x74,0x65,0x3f,0x17,0x73,0x65,0x67,0x73, +0x70,0x61,0x63,0x65,0x4d,0x10,0x78,0x21,0x6c,0x36,0x6d,0x3c,0x6e,0x76,0x6f,0x13, +0x74,0x68,0x65,0x72,0x21,1,0x65,0x23,0x66,0x35,3,0x62,0x37,0x69,0x28,0x6c, +0x29,0x6e,0x2b,0x10,0x64,1,0x6c,0x34,0x6e,0x11,0x75,0x6d,0x2a,0x12,0x6c,0x65, +0x74,0x37,0x14,0x65,0x74,0x74,0x65,0x72,0x29,2,0x65,0x36,0x6c,0x39,0x75,0x2c, +0x14,0x6d,0x65,0x72,0x69,0x63,0x2d,0x14,0x77,0x6c,0x69,0x6e,0x65,0x39,0x66,0x3f, +0x66,0x40,0x67,0x4e,0x68,0x70,0x6b,0x10,0x61,0x26,0x15,0x74,0x61,0x6b,0x61,0x6e, +0x61,0x27,0x10,0x6f,0x24,0x13,0x72,0x6d,0x61,0x74,0x25,1,0x61,0x3a,0x6c,0x19, +0x75,0x65,0x61,0x66,0x74,0x65,0x72,0x7a,0x77,0x6a,0x49,0x10,0x7a,0x49,1,0x65, +0x24,0x6c,0x3d,0x19,0x62,0x72,0x65,0x77,0x6c,0x65,0x74,0x74,0x65,0x72,0x3d,0x61, +0x86,0x63,0x92,0x64,0x94,0x65,2,0x62,0x44,0x6d,0x5e,0x78,0x2e,0x13,0x74,0x65, +0x6e,0x64,0x32,0x15,0x6e,0x75,0x6d,0x6c,0x65,0x74,0x2f,0x42,1,0x61,0x24,0x67, +0x45,0x11,0x73,0x65,0x42,0x12,0x67,0x61,0x7a,0x45,0x46,0x16,0x6f,0x64,0x69,0x66, +0x69,0x65,0x72,0x47,0x15,0x6c,0x65,0x74,0x74,0x65,0x72,0x23,0x10,0x72,0x31,1, +0x6f,0x24,0x71,0x41,0x18,0x75,0x62,0x6c,0x65,0x71,0x75,0x6f,0x74,0x65,0x41,2, +0x63,0x32,0x6e,0x3c,0x6f,0x22,0x12,0x70,0x65,0x6e,0x23,0x24,0x13,0x6c,0x6f,0x73, +0x65,0x25,0x20,0x12,0x6f,0x6e,0x65,0x21,6,0x6f,0x65,0x6f,0x4a,0x72,0x5c,0x74, +0x64,0x76,0x1d,0x69,0x73,0x75,0x61,0x6c,0x6f,0x72,0x64,0x65,0x72,0x6c,0x65,0x66, +0x74,0x3d,0x18,0x76,0x65,0x72,0x73,0x74,0x72,0x75,0x63,0x6b,0x2d,0x13,0x69,0x67, +0x68,0x74,0x2f,0x11,0x6f,0x70,0x30,0x12,0x61,0x6e,0x64,2,0x62,0x32,0x6c,0x62, +0x72,0x13,0x69,0x67,0x68,0x74,0x3b,0x14,0x6f,0x74,0x74,0x6f,0x6d,0x32,0x12,0x61, +0x6e,0x64,1,0x6c,0x2e,0x72,0x13,0x69,0x67,0x68,0x74,0x35,0x12,0x65,0x66,0x74, +0x3f,0x12,0x65,0x66,0x74,0x36,0x17,0x61,0x6e,0x64,0x72,0x69,0x67,0x68,0x74,0x39, +0x62,0x2c,0x6c,0x5c,0x6e,0x10,0x61,0x21,0x14,0x6f,0x74,0x74,0x6f,0x6d,0x22,0x12, +0x61,0x6e,0x64,1,0x6c,0x2e,0x72,0x13,0x69,0x67,0x68,0x74,0x27,0x12,0x65,0x66, +0x74,0x25,0x12,0x65,0x66,0x74,0x28,0x17,0x61,0x6e,0x64,0x72,0x69,0x67,0x68,0x74, +0x2b,0xd,0x6e,0xaa,0x72,0x70,0x72,0x92,0x73,0xa2,0x46,0x74,0xa2,0x54,0x76,1, +0x69,0x60,0x6f,0x12,0x77,0x65,0x6c,0x62,1,0x64,0x3a,0x69,0x19,0x6e,0x64,0x65, +0x70,0x65,0x6e,0x64,0x65,0x6e,0x74,0x67,0x17,0x65,0x70,0x65,0x6e,0x64,0x65,0x6e, +0x74,0x65,1,0x72,0x2e,0x73,0x13,0x61,0x72,0x67,0x61,0x61,0x12,0x61,0x6d,0x61, +0x5f,0x1d,0x65,0x67,0x69,0x73,0x74,0x65,0x72,0x73,0x68,0x69,0x66,0x74,0x65,0x72, +0x57,0x1e,0x79,0x6c,0x6c,0x61,0x62,0x6c,0x65,0x6d,0x6f,0x64,0x69,0x66,0x69,0x65, +0x72,0x59,0x12,0x6f,0x6e,0x65,1,0x6c,0x2c,0x6d,0x12,0x61,0x72,0x6b,0x5d,0x14, +0x65,0x74,0x74,0x65,0x72,0x5b,0x6e,0x3c,0x6f,0x7c,0x70,0x18,0x75,0x72,0x65,0x6b, +0x69,0x6c,0x6c,0x65,0x72,0x55,1,0x6f,0x4c,0x75,1,0x6b,0x3c,0x6d,0x12,0x62, +0x65,0x72,0x50,0x15,0x6a,0x6f,0x69,0x6e,0x65,0x72,0x53,0x11,0x74,0x61,0x4f,0x16, +0x6e,0x6a,0x6f,0x69,0x6e,0x65,0x72,0x4d,0x13,0x74,0x68,0x65,0x72,0x21,0x67,0x3e, +0x67,0x4a,0x69,0x64,0x6a,0x82,0x6d,0x1d,0x6f,0x64,0x69,0x66,0x79,0x69,0x6e,0x67, +0x6c,0x65,0x74,0x74,0x65,0x72,0x4b,0x1c,0x65,0x6d,0x69,0x6e,0x61,0x74,0x69,0x6f, +0x6e,0x6d,0x61,0x72,0x6b,0x45,0x1e,0x6e,0x76,0x69,0x73,0x69,0x62,0x6c,0x65,0x73, +0x74,0x61,0x63,0x6b,0x65,0x72,0x47,0x14,0x6f,0x69,0x6e,0x65,0x72,0x49,0x61,0xa2, +0xba,0x62,0xa2,0xc0,0x63,1,0x61,0xa2,0xa2,0x6f,0x16,0x6e,0x73,0x6f,0x6e,0x61, +0x6e,0x74,0x2a,8,0x6b,0x67,0x6b,0x48,0x6d,0x52,0x70,0x5c,0x73,0xa2,0x42,0x77, +0x19,0x69,0x74,0x68,0x73,0x74,0x61,0x63,0x6b,0x65,0x72,0x43,0x14,0x69,0x6c,0x6c, +0x65,0x72,0x35,0x14,0x65,0x64,0x69,0x61,0x6c,0x37,1,0x6c,0x52,0x72,0x10,0x65, +1,0x63,0x2e,0x66,0x13,0x69,0x78,0x65,0x64,0x3d,0x19,0x65,0x64,0x69,0x6e,0x67, +0x72,0x65,0x70,0x68,0x61,0x3b,0x18,0x61,0x63,0x65,0x68,0x6f,0x6c,0x64,0x65,0x72, +0x39,0x10,0x75,1,0x62,0x3e,0x63,0x1b,0x63,0x65,0x65,0x64,0x69,0x6e,0x67,0x72, +0x65,0x70,0x68,0x61,0x41,0x15,0x6a,0x6f,0x69,0x6e,0x65,0x64,0x3f,0x64,0x4c,0x66, +0x52,0x68,0x5a,0x69,0x1e,0x6e,0x69,0x74,0x69,0x61,0x6c,0x70,0x6f,0x73,0x74,0x66, +0x69,0x78,0x65,0x64,0x33,0x12,0x65,0x61,0x64,0x2d,0x13,0x69,0x6e,0x61,0x6c,0x2f, +0x18,0x65,0x61,0x64,0x6c,0x65,0x74,0x74,0x65,0x72,0x31,0x1d,0x6e,0x74,0x69,0x6c, +0x6c,0x61,0x74,0x69,0x6f,0x6e,0x6d,0x61,0x72,0x6b,0x29,0x16,0x76,0x61,0x67,0x72, +0x61,0x68,0x61,0x23,1,0x69,0x4a,0x72,0x10,0x61,0x1f,0x68,0x6d,0x69,0x6a,0x6f, +0x69,0x6e,0x69,0x6e,0x67,0x6e,0x75,0x6d,0x62,0x65,0x72,0x27,0x12,0x6e,0x64,0x75, +0x25,2,0x72,0x38,0x74,0x46,0x75,0x26,0x15,0x70,0x72,0x69,0x67,0x68,0x74,0x27, +0x20,0x15,0x6f,0x74,0x61,0x74,0x65,0x64,0x21,1,0x72,0x24,0x75,0x25,0x22,0x18, +0x61,0x6e,0x73,0x66,0x6f,0x72,0x6d,0x65,0x64,1,0x72,0x32,0x75,0x15,0x70,0x72, +0x69,0x67,0x68,0x74,0x25,0x15,0x6f,0x74,0x61,0x74,0x65,0x64,0x23,0xd,0x6e,0xc1, +0x86,0x73,0xa8,0x73,0x4c,0x74,0xa2,0x76,0x75,0xa2,0x83,0x7a,0xd8,0x70,0,2, +0x6c,0xd9,0x20,0,0x70,0xd9,0x40,0,0x73,0xc3,0,0xfe,0xf,0,0,0, +7,0x6f,0x3c,0x6f,0xff,8,0,0,0,0x70,0x3a,0x75,0x6e,0x79,0x13,0x6d, +0x62,0x6f,0x6c,0xff,0xf,0,0,0,0x11,0x61,0x63,1,0x65,0x34,0x69,0x15, +0x6e,0x67,0x6d,0x61,0x72,0x6b,0xa5,0,0x18,0x73,0x65,0x70,0x61,0x72,0x61,0x74, +0x6f,0x72,0xc3,0,0x16,0x72,0x72,0x6f,0x67,0x61,0x74,0x65,0xe1,0,0,0x63, +0xff,2,0,0,0,0x65,0x38,0x6b,0xff,4,0,0,0,0x6d,0xff,1, +0,0,0,0x16,0x70,0x61,0x72,0x61,0x74,0x6f,0x72,0xd9,0x70,0,0x1d,0x69, +0x74,0x6c,0x65,0x63,0x61,0x73,0x65,0x6c,0x65,0x74,0x74,0x65,0x72,0x31,1,0x6e, +0x40,0x70,0x1c,0x70,0x65,0x72,0x63,0x61,0x73,0x65,0x6c,0x65,0x74,0x74,0x65,0x72, +0x25,0x17,0x61,0x73,0x73,0x69,0x67,0x6e,0x65,0x64,0x23,0x6e,0xa2,0x69,0x6f,0xa2, +0x89,0x70,0xfe,0x30,0xf8,0,0,9,0x69,0x33,0x69,0xff,0x10,0,0,0, +0x6f,0xfd,0x80,0,0,0x72,0x54,0x73,0xf9,0,0,0x75,0x12,0x6e,0x63,0x74, +0xfe,0x30,0xf8,0,0,0x15,0x75,0x61,0x74,0x69,0x6f,0x6e,0xff,0x30,0xf8,0, +0,0x17,0x69,0x76,0x61,0x74,0x65,0x75,0x73,0x65,0xdd,0,0,0x61,0x48,0x63, +0xfd,0x40,0,0,0x64,0xe9,0,0,0x65,0xfd,0x20,0,0,0x66,0xff,0x20, +0,0,0,0x1f,0x72,0x61,0x67,0x72,0x61,0x70,0x68,0x73,0x65,0x70,0x61,0x72, +0x61,0x74,0x6f,0x72,0xd9,0x40,0,0xbe,0,3,0x64,0xa7,0,0x6c,0xab,0, +0x6f,0x30,0x75,0x13,0x6d,0x62,0x65,0x72,0xbf,0,0xb2,0,0x1b,0x6e,0x73,0x70, +0x61,0x63,0x69,0x6e,0x67,0x6d,0x61,0x72,0x6b,0xa1,1,0x70,0x92,0x74,0x12,0x68, +0x65,0x72,0xe6,0x80,1,3,0x6c,0x40,0x6e,0x4a,0x70,0x56,0x73,0x14,0x79,0x6d, +0x62,0x6f,0x6c,0xff,8,0,0,0,0x14,0x65,0x74,0x74,0x65,0x72,0x61,0x14, +0x75,0x6d,0x62,0x65,0x72,0xb3,0,0x19,0x75,0x6e,0x63,0x74,0x75,0x61,0x74,0x69, +0x6f,0x6e,0xfd,0x80,0,0,0x1c,0x65,0x6e,0x70,0x75,0x6e,0x63,0x74,0x75,0x61, +0x74,0x69,0x6f,0x6e,0xf9,0,0,0x66,0xc0,0xc4,0x66,0xa2,0x47,0x69,0xa2,0x64, +0x6c,0xa2,0x79,0x6d,0xa4,0xc0,4,0x61,0x6c,0x63,0xa5,0,0x65,0xa3,0x80,0x6e, +0xa1,0x6f,0x15,0x64,0x69,0x66,0x69,0x65,0x72,1,0x6c,0x38,0x73,0x14,0x79,0x6d, +0x62,0x6f,0x6c,0xff,4,0,0,0,0x14,0x65,0x74,0x74,0x65,0x72,0x41,1, +0x72,0x3c,0x74,0x16,0x68,0x73,0x79,0x6d,0x62,0x6f,0x6c,0xff,1,0,0,0, +0x10,0x6b,0xa5,0xc0,1,0x69,0x32,0x6f,0x13,0x72,0x6d,0x61,0x74,0xdb,0,0, +0x1d,0x6e,0x61,0x6c,0x70,0x75,0x6e,0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0xff, +0x20,0,0,0,0x10,0x6e,0x1f,0x69,0x74,0x69,0x61,0x6c,0x70,0x75,0x6e,0x63, +0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0xff,0x10,0,0,0,0x9c,7,0x6d,0x18, +0x6d,0x41,0x6f,0x28,0x74,0x31,0x75,0x25,0x60,0x1c,0x77,0x65,0x72,0x63,0x61,0x73, +0x65,0x6c,0x65,0x74,0x74,0x65,0x72,0x29,0x63,0x3d,0x65,0x28,0x69,0x42,0x6c,0x29, +0x13,0x74,0x74,0x65,0x72,0x9c,0x15,0x6e,0x75,0x6d,0x62,0x65,0x72,0xab,0,0x1a, +0x6e,0x65,0x73,0x65,0x70,0x61,0x72,0x61,0x74,0x6f,0x72,0xd9,0x20,0,0x63,0x46, +0x64,0xa2,0x96,0x65,0x1b,0x6e,0x63,0x6c,0x6f,0x73,0x69,0x6e,0x67,0x6d,0x61,0x72, +0x6b,0xa3,0x80,0xe6,0x80,1,7,0x6e,0x57,0x6e,0x52,0x6f,0x5e,0x73,0xe1,0, +0,0x75,0x1b,0x72,0x72,0x65,0x6e,0x63,0x79,0x73,0x79,0x6d,0x62,0x6f,0x6c,0xff, +2,0,0,0,0x22,0x12,0x74,0x72,0x6c,0xd9,0x80,0,0xdc,0,0,1, +0x6d,0x62,0x6e,1,0x6e,0x30,0x74,0x12,0x72,0x6f,0x6c,0xd9,0x80,0,0x1f,0x65, +0x63,0x74,0x6f,0x72,0x70,0x75,0x6e,0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0xfd, +0x40,0,0,0x19,0x62,0x69,0x6e,0x69,0x6e,0x67,0x6d,0x61,0x72,0x6b,0xa5,0xc0, +0x61,0x58,0x63,0xd9,0x80,0,0x66,0xdb,0,0,0x6c,0x1d,0x6f,0x73,0x65,0x70, +0x75,0x6e,0x63,0x74,0x75,0x61,0x74,0x69,0x6f,0x6e,0xfd,0x20,0,0,0x18,0x73, +0x65,0x64,0x6c,0x65,0x74,0x74,0x65,0x72,0x3d,2,0x61,0x32,0x65,0x50,0x69,0x12, +0x67,0x69,0x74,0xa7,0,0x1c,0x73,0x68,0x70,0x75,0x6e,0x63,0x74,0x75,0x61,0x74, +0x69,0x6f,0x6e,0xe9,0,0,0x1a,0x63,0x69,0x6d,0x61,0x6c,0x6e,0x75,0x6d,0x62, +0x65,0x72,0xa7,0 +}; + +const char PropNameData::nameGroups[23100]={ +2,'A','l','p','h','a',0,'A','l','p','h','a','b','e','t','i','c',0, +4,'N',0,'N','o',0,'F',0,'F','a','l','s','e',0,4,'Y',0,'Y','e','s',0,'T',0,'T','r','u','e',0, +2,'N','R',0,'N','o','t','_','R','e','o','r','d','e','r','e','d',0, +2,'O','V',0,'O','v','e','r','l','a','y',0,2,'H','A','N','R',0,'H','a','n','_','R','e','a','d','i','n','g',0, +2,'N','K',0,'N','u','k','t','a',0,2,'K','V',0,'K','a','n','a','_','V','o','i','c','i','n','g',0, +2,'V','R',0,'V','i','r','a','m','a',0,2,'C','C','C','1','0',0,'C','C','C','1','0',0, +2,'C','C','C','1','1',0,'C','C','C','1','1',0,2,'C','C','C','1','2',0,'C','C','C','1','2',0, +2,'C','C','C','1','3',0,'C','C','C','1','3',0,2,'C','C','C','1','4',0,'C','C','C','1','4',0, +2,'C','C','C','1','5',0,'C','C','C','1','5',0,2,'C','C','C','1','6',0,'C','C','C','1','6',0, +2,'C','C','C','1','7',0,'C','C','C','1','7',0,2,'C','C','C','1','8',0,'C','C','C','1','8',0, +2,'C','C','C','1','9',0,'C','C','C','1','9',0,2,'C','C','C','2','0',0,'C','C','C','2','0',0, +2,'C','C','C','2','1',0,'C','C','C','2','1',0,2,'C','C','C','2','2',0,'C','C','C','2','2',0, +2,'C','C','C','2','3',0,'C','C','C','2','3',0,2,'C','C','C','2','4',0,'C','C','C','2','4',0, +2,'C','C','C','2','5',0,'C','C','C','2','5',0,2,'C','C','C','2','6',0,'C','C','C','2','6',0, +2,'C','C','C','2','7',0,'C','C','C','2','7',0,2,'C','C','C','2','8',0,'C','C','C','2','8',0, +2,'C','C','C','2','9',0,'C','C','C','2','9',0,2,'C','C','C','3','0',0,'C','C','C','3','0',0, +2,'C','C','C','3','1',0,'C','C','C','3','1',0,2,'C','C','C','3','2',0,'C','C','C','3','2',0, +2,'C','C','C','3','3',0,'C','C','C','3','3',0,2,'C','C','C','3','4',0,'C','C','C','3','4',0, +2,'C','C','C','3','5',0,'C','C','C','3','5',0,2,'C','C','C','3','6',0,'C','C','C','3','6',0, +2,'C','C','C','8','4',0,'C','C','C','8','4',0,2,'C','C','C','9','1',0,'C','C','C','9','1',0, +2,'C','C','C','1','0','3',0,'C','C','C','1','0','3',0,2,'C','C','C','1','0','7',0,'C','C','C','1','0','7',0, +2,'C','C','C','1','1','8',0,'C','C','C','1','1','8',0,2,'C','C','C','1','2','2',0,'C','C','C','1','2','2',0, +2,'C','C','C','1','2','9',0,'C','C','C','1','2','9',0,2,'C','C','C','1','3','0',0,'C','C','C','1','3','0',0, +2,'C','C','C','1','3','2',0,'C','C','C','1','3','2',0,2,'C','C','C','1','3','3',0,'C','C','C','1','3','3',0, +2,'A','T','B','L',0,'A','t','t','a','c','h','e','d','_','B','e','l','o','w','_','L','e','f','t',0, +2,'A','T','B',0,'A','t','t','a','c','h','e','d','_','B','e','l','o','w',0, +2,'A','T','A',0,'A','t','t','a','c','h','e','d','_','A','b','o','v','e',0, +2,'A','T','A','R',0,'A','t','t','a','c','h','e','d','_','A','b','o','v','e','_','R','i','g','h','t',0, +2,'B','L',0,'B','e','l','o','w','_','L','e','f','t',0,2,'B',0,'B','e','l','o','w',0, +2,'B','R',0,'B','e','l','o','w','_','R','i','g','h','t',0, +2,'L',0,'L','e','f','t',0,2,'R',0,'R','i','g','h','t',0, +2,'A','L',0,'A','b','o','v','e','_','L','e','f','t',0,2,'A',0,'A','b','o','v','e',0, +2,'A','R',0,'A','b','o','v','e','_','R','i','g','h','t',0, +2,'D','B',0,'D','o','u','b','l','e','_','B','e','l','o','w',0, +2,'D','A',0,'D','o','u','b','l','e','_','A','b','o','v','e',0, +2,'I','S',0,'I','o','t','a','_','S','u','b','s','c','r','i','p','t',0, +2,'A','H','e','x',0,'A','S','C','I','I','_','H','e','x','_','D','i','g','i','t',0, +2,'B','i','d','i','_','C',0,'B','i','d','i','_','C','o','n','t','r','o','l',0, +2,'B','i','d','i','_','M',0,'B','i','d','i','_','M','i','r','r','o','r','e','d',0, +2,'D','a','s','h',0,'D','a','s','h',0,2,'D','I',0,'D','e','f','a','u','l','t','_','I','g','n','o','r','a','b','l','e', +'_','C','o','d','e','_','P','o','i','n','t',0,2,'D','e','p',0,'D','e','p','r','e','c','a','t','e','d',0, +2,'D','i','a',0,'D','i','a','c','r','i','t','i','c',0,2,'E','x','t',0,'E','x','t','e','n','d','e','r',0, +2,'C','o','m','p','_','E','x',0,'F','u','l','l','_','C','o','m','p','o','s','i','t','i','o','n','_','E','x','c','l','u','s', +'i','o','n',0,2,'G','r','_','B','a','s','e',0,'G','r','a','p','h','e','m','e','_','B','a','s','e',0, +2,'G','r','_','E','x','t',0,'G','r','a','p','h','e','m','e','_','E','x','t','e','n','d',0, +2,'G','r','_','L','i','n','k',0,'G','r','a','p','h','e','m','e','_','L','i','n','k',0, +2,'H','e','x',0,'H','e','x','_','D','i','g','i','t',0,2,'H','y','p','h','e','n',0,'H','y','p','h','e','n',0, +2,'I','D','C',0,'I','D','_','C','o','n','t','i','n','u','e',0, +2,'I','D','S',0,'I','D','_','S','t','a','r','t',0,2,'I','d','e','o',0,'I','d','e','o','g','r','a','p','h','i','c',0, +2,'I','D','S','B',0,'I','D','S','_','B','i','n','a','r','y','_','O','p','e','r','a','t','o','r',0, +2,'I','D','S','T',0,'I','D','S','_','T','r','i','n','a','r','y','_','O','p','e','r','a','t','o','r',0, +2,'J','o','i','n','_','C',0,'J','o','i','n','_','C','o','n','t','r','o','l',0, +2,'L','O','E',0,'L','o','g','i','c','a','l','_','O','r','d','e','r','_','E','x','c','e','p','t','i','o','n',0, +2,'L','o','w','e','r',0,'L','o','w','e','r','c','a','s','e',0, +2,'M','a','t','h',0,'M','a','t','h',0,2,'N','C','h','a','r',0,'N','o','n','c','h','a','r','a','c','t','e','r','_','C', +'o','d','e','_','P','o','i','n','t',0,2,'Q','M','a','r','k',0,'Q','u','o','t','a','t','i','o','n','_','M','a','r','k',0, +2,'R','a','d','i','c','a','l',0,'R','a','d','i','c','a','l',0, +2,'S','D',0,'S','o','f','t','_','D','o','t','t','e','d',0, +2,'T','e','r','m',0,'T','e','r','m','i','n','a','l','_','P','u','n','c','t','u','a','t','i','o','n',0, +2,'U','I','d','e','o',0,'U','n','i','f','i','e','d','_','I','d','e','o','g','r','a','p','h',0, +2,'U','p','p','e','r',0,'U','p','p','e','r','c','a','s','e',0, +3,'W','S','p','a','c','e',0,'W','h','i','t','e','_','S','p','a','c','e',0,'s','p','a','c','e',0, +2,'X','I','D','C',0,'X','I','D','_','C','o','n','t','i','n','u','e',0, +2,'X','I','D','S',0,'X','I','D','_','S','t','a','r','t',0, +2,'S','e','n','s','i','t','i','v','e',0,'C','a','s','e','_','S','e','n','s','i','t','i','v','e',0, +2,'S','T','e','r','m',0,'S','e','n','t','e','n','c','e','_','T','e','r','m','i','n','a','l',0, +2,'V','S',0,'V','a','r','i','a','t','i','o','n','_','S','e','l','e','c','t','o','r',0, +2,'n','f','d','i','n','e','r','t',0,'N','F','D','_','I','n','e','r','t',0, +2,'n','f','k','d','i','n','e','r','t',0,'N','F','K','D','_','I','n','e','r','t',0, +2,'n','f','c','i','n','e','r','t',0,'N','F','C','_','I','n','e','r','t',0, +2,'n','f','k','c','i','n','e','r','t',0,'N','F','K','C','_','I','n','e','r','t',0, +2,'s','e','g','s','t','a','r','t',0,'S','e','g','m','e','n','t','_','S','t','a','r','t','e','r',0, +2,'P','a','t','_','S','y','n',0,'P','a','t','t','e','r','n','_','S','y','n','t','a','x',0, +2,'P','a','t','_','W','S',0,'P','a','t','t','e','r','n','_','W','h','i','t','e','_','S','p','a','c','e',0, +2,0,'a','l','n','u','m',0,2,0,'b','l','a','n','k',0, +2,0,'g','r','a','p','h',0,2,0,'p','r','i','n','t',0, +2,0,'x','d','i','g','i','t',0,2,'C','a','s','e','d',0,'C','a','s','e','d',0, +2,'C','I',0,'C','a','s','e','_','I','g','n','o','r','a','b','l','e',0, +2,'C','W','L',0,'C','h','a','n','g','e','s','_','W','h','e','n','_','L','o','w','e','r','c','a','s','e','d',0, +2,'C','W','U',0,'C','h','a','n','g','e','s','_','W','h','e','n','_','U','p','p','e','r','c','a','s','e','d',0, +2,'C','W','T',0,'C','h','a','n','g','e','s','_','W','h','e','n','_','T','i','t','l','e','c','a','s','e','d',0, +2,'C','W','C','F',0,'C','h','a','n','g','e','s','_','W','h','e','n','_','C','a','s','e','f','o','l','d','e','d',0, +2,'C','W','C','M',0,'C','h','a','n','g','e','s','_','W','h','e','n','_','C','a','s','e','m','a','p','p','e','d',0, +2,'C','W','K','C','F',0,'C','h','a','n','g','e','s','_','W','h','e','n','_','N','F','K','C','_','C','a','s','e','f','o','l', +'d','e','d',0,2,'E','m','o','j','i',0,'E','m','o','j','i',0, +2,'E','P','r','e','s',0,'E','m','o','j','i','_','P','r','e','s','e','n','t','a','t','i','o','n',0, +2,'E','M','o','d',0,'E','m','o','j','i','_','M','o','d','i','f','i','e','r',0, +2,'E','B','a','s','e',0,'E','m','o','j','i','_','M','o','d','i','f','i','e','r','_','B','a','s','e',0, +2,'E','C','o','m','p',0,'E','m','o','j','i','_','C','o','m','p','o','n','e','n','t',0, +2,'R','I',0,'R','e','g','i','o','n','a','l','_','I','n','d','i','c','a','t','o','r',0, +2,'P','C','M',0,'P','r','e','p','e','n','d','e','d','_','C','o','n','c','a','t','e','n','a','t','i','o','n','_','M','a','r', +'k',0,2,'E','x','t','P','i','c','t',0,'E','x','t','e','n','d','e','d','_','P','i','c','t','o','g','r','a','p','h','i','c', +0,2,'B','a','s','i','c','_','E','m','o','j','i',0,'B','a','s','i','c','_','E','m','o','j','i',0, +2,'E','m','o','j','i','_','K','e','y','c','a','p','_','S','e','q','u','e','n','c','e',0,'E','m','o','j','i','_','K','e','y', +'c','a','p','_','S','e','q','u','e','n','c','e',0,2,'R','G','I','_','E','m','o','j','i','_','M','o','d','i','f','i','e','r', +'_','S','e','q','u','e','n','c','e',0,'R','G','I','_','E','m','o','j','i','_','M','o','d','i','f','i','e','r','_','S','e','q', +'u','e','n','c','e',0,2,'R','G','I','_','E','m','o','j','i','_','F','l','a','g','_','S','e','q','u','e','n','c','e',0, +'R','G','I','_','E','m','o','j','i','_','F','l','a','g','_','S','e','q','u','e','n','c','e',0, +2,'R','G','I','_','E','m','o','j','i','_','T','a','g','_','S','e','q','u','e','n','c','e',0, +'R','G','I','_','E','m','o','j','i','_','T','a','g','_','S','e','q','u','e','n','c','e',0, +2,'R','G','I','_','E','m','o','j','i','_','Z','W','J','_','S','e','q','u','e','n','c','e',0, +'R','G','I','_','E','m','o','j','i','_','Z','W','J','_','S','e','q','u','e','n','c','e',0, +2,'R','G','I','_','E','m','o','j','i',0,'R','G','I','_','E','m','o','j','i',0, +2,'b','c',0,'B','i','d','i','_','C','l','a','s','s',0,2,'L',0,'L','e','f','t','_','T','o','_','R','i','g','h','t',0, +2,'R',0,'R','i','g','h','t','_','T','o','_','L','e','f','t',0, +2,'E','N',0,'E','u','r','o','p','e','a','n','_','N','u','m','b','e','r',0, +2,'E','S',0,'E','u','r','o','p','e','a','n','_','S','e','p','a','r','a','t','o','r',0, +2,'E','T',0,'E','u','r','o','p','e','a','n','_','T','e','r','m','i','n','a','t','o','r',0, +2,'A','N',0,'A','r','a','b','i','c','_','N','u','m','b','e','r',0, +2,'C','S',0,'C','o','m','m','o','n','_','S','e','p','a','r','a','t','o','r',0, +2,'B',0,'P','a','r','a','g','r','a','p','h','_','S','e','p','a','r','a','t','o','r',0, +2,'S',0,'S','e','g','m','e','n','t','_','S','e','p','a','r','a','t','o','r',0, +2,'W','S',0,'W','h','i','t','e','_','S','p','a','c','e',0, +2,'O','N',0,'O','t','h','e','r','_','N','e','u','t','r','a','l',0, +2,'L','R','E',0,'L','e','f','t','_','T','o','_','R','i','g','h','t','_','E','m','b','e','d','d','i','n','g',0, +2,'L','R','O',0,'L','e','f','t','_','T','o','_','R','i','g','h','t','_','O','v','e','r','r','i','d','e',0, +2,'A','L',0,'A','r','a','b','i','c','_','L','e','t','t','e','r',0, +2,'R','L','E',0,'R','i','g','h','t','_','T','o','_','L','e','f','t','_','E','m','b','e','d','d','i','n','g',0, +2,'R','L','O',0,'R','i','g','h','t','_','T','o','_','L','e','f','t','_','O','v','e','r','r','i','d','e',0, +2,'P','D','F',0,'P','o','p','_','D','i','r','e','c','t','i','o','n','a','l','_','F','o','r','m','a','t',0, +2,'N','S','M',0,'N','o','n','s','p','a','c','i','n','g','_','M','a','r','k',0, +2,'B','N',0,'B','o','u','n','d','a','r','y','_','N','e','u','t','r','a','l',0, +2,'F','S','I',0,'F','i','r','s','t','_','S','t','r','o','n','g','_','I','s','o','l','a','t','e',0, +2,'L','R','I',0,'L','e','f','t','_','T','o','_','R','i','g','h','t','_','I','s','o','l','a','t','e',0, +2,'R','L','I',0,'R','i','g','h','t','_','T','o','_','L','e','f','t','_','I','s','o','l','a','t','e',0, +2,'P','D','I',0,'P','o','p','_','D','i','r','e','c','t','i','o','n','a','l','_','I','s','o','l','a','t','e',0, +2,'b','l','k',0,'B','l','o','c','k',0,2,'N','B',0,'N','o','_','B','l','o','c','k',0, +2,'A','S','C','I','I',0,'B','a','s','i','c','_','L','a','t','i','n',0, +3,'L','a','t','i','n','_','1','_','S','u','p',0,'L','a','t','i','n','_','1','_','S','u','p','p','l','e','m','e','n','t',0, +'L','a','t','i','n','_','1',0,2,'L','a','t','i','n','_','E','x','t','_','A',0,'L','a','t','i','n','_','E','x','t','e','n', +'d','e','d','_','A',0,2,'L','a','t','i','n','_','E','x','t','_','B',0,'L','a','t','i','n','_','E','x','t','e','n','d','e', +'d','_','B',0,2,'I','P','A','_','E','x','t',0,'I','P','A','_','E','x','t','e','n','s','i','o','n','s',0, +2,'M','o','d','i','f','i','e','r','_','L','e','t','t','e','r','s',0,'S','p','a','c','i','n','g','_','M','o','d','i','f','i', +'e','r','_','L','e','t','t','e','r','s',0,2,'D','i','a','c','r','i','t','i','c','a','l','s',0, +'C','o','m','b','i','n','i','n','g','_','D','i','a','c','r','i','t','i','c','a','l','_','M','a','r','k','s',0, +2,'G','r','e','e','k',0,'G','r','e','e','k','_','A','n','d','_','C','o','p','t','i','c',0, +2,'C','y','r','i','l','l','i','c',0,'C','y','r','i','l','l','i','c',0, +2,'A','r','m','e','n','i','a','n',0,'A','r','m','e','n','i','a','n',0, +2,'H','e','b','r','e','w',0,'H','e','b','r','e','w',0,2,'A','r','a','b','i','c',0,'A','r','a','b','i','c',0, +2,'S','y','r','i','a','c',0,'S','y','r','i','a','c',0,2,'T','h','a','a','n','a',0,'T','h','a','a','n','a',0, +2,'D','e','v','a','n','a','g','a','r','i',0,'D','e','v','a','n','a','g','a','r','i',0, +2,'B','e','n','g','a','l','i',0,'B','e','n','g','a','l','i',0, +2,'G','u','r','m','u','k','h','i',0,'G','u','r','m','u','k','h','i',0, +2,'G','u','j','a','r','a','t','i',0,'G','u','j','a','r','a','t','i',0, +2,'O','r','i','y','a',0,'O','r','i','y','a',0,2,'T','a','m','i','l',0,'T','a','m','i','l',0, +2,'T','e','l','u','g','u',0,'T','e','l','u','g','u',0,2,'K','a','n','n','a','d','a',0, +'K','a','n','n','a','d','a',0,2,'M','a','l','a','y','a','l','a','m',0,'M','a','l','a','y','a','l','a','m',0, +2,'S','i','n','h','a','l','a',0,'S','i','n','h','a','l','a',0, +2,'T','h','a','i',0,'T','h','a','i',0,2,'L','a','o',0,'L','a','o',0, +2,'T','i','b','e','t','a','n',0,'T','i','b','e','t','a','n',0, +2,'M','y','a','n','m','a','r',0,'M','y','a','n','m','a','r',0, +2,'G','e','o','r','g','i','a','n',0,'G','e','o','r','g','i','a','n',0, +2,'J','a','m','o',0,'H','a','n','g','u','l','_','J','a','m','o',0, +2,'E','t','h','i','o','p','i','c',0,'E','t','h','i','o','p','i','c',0, +2,'C','h','e','r','o','k','e','e',0,'C','h','e','r','o','k','e','e',0, +3,'U','C','A','S',0,'U','n','i','f','i','e','d','_','C','a','n','a','d','i','a','n','_','A','b','o','r','i','g','i','n','a', +'l','_','S','y','l','l','a','b','i','c','s',0,'C','a','n','a','d','i','a','n','_','S','y','l','l','a','b','i','c','s',0, +2,'O','g','h','a','m',0,'O','g','h','a','m',0,2,'R','u','n','i','c',0,'R','u','n','i','c',0, +2,'K','h','m','e','r',0,'K','h','m','e','r',0,2,'M','o','n','g','o','l','i','a','n',0, +'M','o','n','g','o','l','i','a','n',0,2,'L','a','t','i','n','_','E','x','t','_','A','d','d','i','t','i','o','n','a','l',0, +'L','a','t','i','n','_','E','x','t','e','n','d','e','d','_','A','d','d','i','t','i','o','n','a','l',0, +2,'G','r','e','e','k','_','E','x','t',0,'G','r','e','e','k','_','E','x','t','e','n','d','e','d',0, +2,'P','u','n','c','t','u','a','t','i','o','n',0,'G','e','n','e','r','a','l','_','P','u','n','c','t','u','a','t','i','o','n', +0,2,'S','u','p','e','r','_','A','n','d','_','S','u','b',0,'S','u','p','e','r','s','c','r','i','p','t','s','_','A','n','d', +'_','S','u','b','s','c','r','i','p','t','s',0,2,'C','u','r','r','e','n','c','y','_','S','y','m','b','o','l','s',0, +'C','u','r','r','e','n','c','y','_','S','y','m','b','o','l','s',0, +3,'D','i','a','c','r','i','t','i','c','a','l','s','_','F','o','r','_','S','y','m','b','o','l','s',0, +'C','o','m','b','i','n','i','n','g','_','D','i','a','c','r','i','t','i','c','a','l','_','M','a','r','k','s','_','F','o','r','_', +'S','y','m','b','o','l','s',0,'C','o','m','b','i','n','i','n','g','_','M','a','r','k','s','_','F','o','r','_','S','y','m','b', +'o','l','s',0,2,'L','e','t','t','e','r','l','i','k','e','_','S','y','m','b','o','l','s',0, +'L','e','t','t','e','r','l','i','k','e','_','S','y','m','b','o','l','s',0, +2,'N','u','m','b','e','r','_','F','o','r','m','s',0,'N','u','m','b','e','r','_','F','o','r','m','s',0, +2,'A','r','r','o','w','s',0,'A','r','r','o','w','s',0,2,'M','a','t','h','_','O','p','e','r','a','t','o','r','s',0, +'M','a','t','h','e','m','a','t','i','c','a','l','_','O','p','e','r','a','t','o','r','s',0, +2,'M','i','s','c','_','T','e','c','h','n','i','c','a','l',0,'M','i','s','c','e','l','l','a','n','e','o','u','s','_','T','e', +'c','h','n','i','c','a','l',0,2,'C','o','n','t','r','o','l','_','P','i','c','t','u','r','e','s',0, +'C','o','n','t','r','o','l','_','P','i','c','t','u','r','e','s',0, +2,'O','C','R',0,'O','p','t','i','c','a','l','_','C','h','a','r','a','c','t','e','r','_','R','e','c','o','g','n','i','t','i', +'o','n',0,2,'E','n','c','l','o','s','e','d','_','A','l','p','h','a','n','u','m',0,'E','n','c','l','o','s','e','d','_','A', +'l','p','h','a','n','u','m','e','r','i','c','s',0,2,'B','o','x','_','D','r','a','w','i','n','g',0, +'B','o','x','_','D','r','a','w','i','n','g',0,2,'B','l','o','c','k','_','E','l','e','m','e','n','t','s',0, +'B','l','o','c','k','_','E','l','e','m','e','n','t','s',0,2,'G','e','o','m','e','t','r','i','c','_','S','h','a','p','e','s', +0,'G','e','o','m','e','t','r','i','c','_','S','h','a','p','e','s',0, +2,'M','i','s','c','_','S','y','m','b','o','l','s',0,'M','i','s','c','e','l','l','a','n','e','o','u','s','_','S','y','m','b', +'o','l','s',0,2,'D','i','n','g','b','a','t','s',0,'D','i','n','g','b','a','t','s',0, +2,'B','r','a','i','l','l','e',0,'B','r','a','i','l','l','e','_','P','a','t','t','e','r','n','s',0, +2,'C','J','K','_','R','a','d','i','c','a','l','s','_','S','u','p',0,'C','J','K','_','R','a','d','i','c','a','l','s','_','S', +'u','p','p','l','e','m','e','n','t',0,2,'K','a','n','g','x','i',0,'K','a','n','g','x','i','_','R','a','d','i','c','a','l', +'s',0,2,'I','D','C',0,'I','d','e','o','g','r','a','p','h','i','c','_','D','e','s','c','r','i','p','t','i','o','n','_','C', +'h','a','r','a','c','t','e','r','s',0,2,'C','J','K','_','S','y','m','b','o','l','s',0,'C','J','K','_','S','y','m','b','o', +'l','s','_','A','n','d','_','P','u','n','c','t','u','a','t','i','o','n',0, +2,'H','i','r','a','g','a','n','a',0,'H','i','r','a','g','a','n','a',0, +2,'K','a','t','a','k','a','n','a',0,'K','a','t','a','k','a','n','a',0, +2,'B','o','p','o','m','o','f','o',0,'B','o','p','o','m','o','f','o',0, +2,'C','o','m','p','a','t','_','J','a','m','o',0,'H','a','n','g','u','l','_','C','o','m','p','a','t','i','b','i','l','i','t', +'y','_','J','a','m','o',0,2,'K','a','n','b','u','n',0,'K','a','n','b','u','n',0, +2,'B','o','p','o','m','o','f','o','_','E','x','t',0,'B','o','p','o','m','o','f','o','_','E','x','t','e','n','d','e','d',0, +2,'E','n','c','l','o','s','e','d','_','C','J','K',0,'E','n','c','l','o','s','e','d','_','C','J','K','_','L','e','t','t','e', +'r','s','_','A','n','d','_','M','o','n','t','h','s',0,2,'C','J','K','_','C','o','m','p','a','t',0, +'C','J','K','_','C','o','m','p','a','t','i','b','i','l','i','t','y',0, +2,'C','J','K','_','E','x','t','_','A',0,'C','J','K','_','U','n','i','f','i','e','d','_','I','d','e','o','g','r','a','p','h', +'s','_','E','x','t','e','n','s','i','o','n','_','A',0,2,'C','J','K',0,'C','J','K','_','U','n','i','f','i','e','d','_','I', +'d','e','o','g','r','a','p','h','s',0,2,'Y','i','_','S','y','l','l','a','b','l','e','s',0, +'Y','i','_','S','y','l','l','a','b','l','e','s',0,2,'Y','i','_','R','a','d','i','c','a','l','s',0, +'Y','i','_','R','a','d','i','c','a','l','s',0,2,'H','a','n','g','u','l',0,'H','a','n','g','u','l','_','S','y','l','l','a', +'b','l','e','s',0,2,'H','i','g','h','_','S','u','r','r','o','g','a','t','e','s',0,'H','i','g','h','_','S','u','r','r','o', +'g','a','t','e','s',0,2,'H','i','g','h','_','P','U','_','S','u','r','r','o','g','a','t','e','s',0, +'H','i','g','h','_','P','r','i','v','a','t','e','_','U','s','e','_','S','u','r','r','o','g','a','t','e','s',0, +2,'L','o','w','_','S','u','r','r','o','g','a','t','e','s',0,'L','o','w','_','S','u','r','r','o','g','a','t','e','s',0, +3,'P','U','A',0,'P','r','i','v','a','t','e','_','U','s','e','_','A','r','e','a',0,'P','r','i','v','a','t','e','_','U','s', +'e',0,2,'C','J','K','_','C','o','m','p','a','t','_','I','d','e','o','g','r','a','p','h','s',0, +'C','J','K','_','C','o','m','p','a','t','i','b','i','l','i','t','y','_','I','d','e','o','g','r','a','p','h','s',0, +2,'A','l','p','h','a','b','e','t','i','c','_','P','F',0,'A','l','p','h','a','b','e','t','i','c','_','P','r','e','s','e','n', +'t','a','t','i','o','n','_','F','o','r','m','s',0,3,'A','r','a','b','i','c','_','P','F','_','A',0, +'A','r','a','b','i','c','_','P','r','e','s','e','n','t','a','t','i','o','n','_','F','o','r','m','s','_','A',0, +'A','r','a','b','i','c','_','P','r','e','s','e','n','t','a','t','i','o','n','_','F','o','r','m','s','-','A',0, +2,'H','a','l','f','_','M','a','r','k','s',0,'C','o','m','b','i','n','i','n','g','_','H','a','l','f','_','M','a','r','k','s', +0,2,'C','J','K','_','C','o','m','p','a','t','_','F','o','r','m','s',0,'C','J','K','_','C','o','m','p','a','t','i','b','i', +'l','i','t','y','_','F','o','r','m','s',0,2,'S','m','a','l','l','_','F','o','r','m','s',0, +'S','m','a','l','l','_','F','o','r','m','_','V','a','r','i','a','n','t','s',0, +2,'A','r','a','b','i','c','_','P','F','_','B',0,'A','r','a','b','i','c','_','P','r','e','s','e','n','t','a','t','i','o','n', +'_','F','o','r','m','s','_','B',0,2,'S','p','e','c','i','a','l','s',0,'S','p','e','c','i','a','l','s',0, +2,'H','a','l','f','_','A','n','d','_','F','u','l','l','_','F','o','r','m','s',0,'H','a','l','f','w','i','d','t','h','_','A', +'n','d','_','F','u','l','l','w','i','d','t','h','_','F','o','r','m','s',0, +2,'O','l','d','_','I','t','a','l','i','c',0,'O','l','d','_','I','t','a','l','i','c',0, +2,'G','o','t','h','i','c',0,'G','o','t','h','i','c',0,2,'D','e','s','e','r','e','t',0, +'D','e','s','e','r','e','t',0,2,'B','y','z','a','n','t','i','n','e','_','M','u','s','i','c',0, +'B','y','z','a','n','t','i','n','e','_','M','u','s','i','c','a','l','_','S','y','m','b','o','l','s',0, +2,'M','u','s','i','c',0,'M','u','s','i','c','a','l','_','S','y','m','b','o','l','s',0, +2,'M','a','t','h','_','A','l','p','h','a','n','u','m',0,'M','a','t','h','e','m','a','t','i','c','a','l','_','A','l','p','h', +'a','n','u','m','e','r','i','c','_','S','y','m','b','o','l','s',0, +2,'C','J','K','_','E','x','t','_','B',0,'C','J','K','_','U','n','i','f','i','e','d','_','I','d','e','o','g','r','a','p','h', +'s','_','E','x','t','e','n','s','i','o','n','_','B',0,2,'C','J','K','_','C','o','m','p','a','t','_','I','d','e','o','g','r', +'a','p','h','s','_','S','u','p',0,'C','J','K','_','C','o','m','p','a','t','i','b','i','l','i','t','y','_','I','d','e','o','g', +'r','a','p','h','s','_','S','u','p','p','l','e','m','e','n','t',0, +2,'T','a','g','s',0,'T','a','g','s',0,3,'C','y','r','i','l','l','i','c','_','S','u','p',0, +'C','y','r','i','l','l','i','c','_','S','u','p','p','l','e','m','e','n','t',0,'C','y','r','i','l','l','i','c','_','S','u','p', +'p','l','e','m','e','n','t','a','r','y',0,2,'T','a','g','a','l','o','g',0,'T','a','g','a','l','o','g',0, +2,'H','a','n','u','n','o','o',0,'H','a','n','u','n','o','o',0, +2,'B','u','h','i','d',0,'B','u','h','i','d',0,2,'T','a','g','b','a','n','w','a',0,'T','a','g','b','a','n','w','a',0, +2,'M','i','s','c','_','M','a','t','h','_','S','y','m','b','o','l','s','_','A',0,'M','i','s','c','e','l','l','a','n','e','o', +'u','s','_','M','a','t','h','e','m','a','t','i','c','a','l','_','S','y','m','b','o','l','s','_','A',0, +2,'S','u','p','_','A','r','r','o','w','s','_','A',0,'S','u','p','p','l','e','m','e','n','t','a','l','_','A','r','r','o','w', +'s','_','A',0,2,'S','u','p','_','A','r','r','o','w','s','_','B',0,'S','u','p','p','l','e','m','e','n','t','a','l','_','A', +'r','r','o','w','s','_','B',0,2,'M','i','s','c','_','M','a','t','h','_','S','y','m','b','o','l','s','_','B',0, +'M','i','s','c','e','l','l','a','n','e','o','u','s','_','M','a','t','h','e','m','a','t','i','c','a','l','_','S','y','m','b','o', +'l','s','_','B',0,2,'S','u','p','_','M','a','t','h','_','O','p','e','r','a','t','o','r','s',0, +'S','u','p','p','l','e','m','e','n','t','a','l','_','M','a','t','h','e','m','a','t','i','c','a','l','_','O','p','e','r','a','t', +'o','r','s',0,2,'K','a','t','a','k','a','n','a','_','E','x','t',0,'K','a','t','a','k','a','n','a','_','P','h','o','n','e', +'t','i','c','_','E','x','t','e','n','s','i','o','n','s',0,2,'V','S',0,'V','a','r','i','a','t','i','o','n','_','S','e','l', +'e','c','t','o','r','s',0,2,'S','u','p','_','P','U','A','_','A',0,'S','u','p','p','l','e','m','e','n','t','a','r','y','_', +'P','r','i','v','a','t','e','_','U','s','e','_','A','r','e','a','_','A',0, +2,'S','u','p','_','P','U','A','_','B',0,'S','u','p','p','l','e','m','e','n','t','a','r','y','_','P','r','i','v','a','t','e', +'_','U','s','e','_','A','r','e','a','_','B',0,2,'L','i','m','b','u',0,'L','i','m','b','u',0, +2,'T','a','i','_','L','e',0,'T','a','i','_','L','e',0,2,'K','h','m','e','r','_','S','y','m','b','o','l','s',0, +'K','h','m','e','r','_','S','y','m','b','o','l','s',0,2,'P','h','o','n','e','t','i','c','_','E','x','t',0, +'P','h','o','n','e','t','i','c','_','E','x','t','e','n','s','i','o','n','s',0, +2,'M','i','s','c','_','A','r','r','o','w','s',0,'M','i','s','c','e','l','l','a','n','e','o','u','s','_','S','y','m','b','o', +'l','s','_','A','n','d','_','A','r','r','o','w','s',0,2,'Y','i','j','i','n','g',0,'Y','i','j','i','n','g','_','H','e','x', +'a','g','r','a','m','_','S','y','m','b','o','l','s',0,2,'L','i','n','e','a','r','_','B','_','S','y','l','l','a','b','a','r', +'y',0,'L','i','n','e','a','r','_','B','_','S','y','l','l','a','b','a','r','y',0, +2,'L','i','n','e','a','r','_','B','_','I','d','e','o','g','r','a','m','s',0,'L','i','n','e','a','r','_','B','_','I','d','e', +'o','g','r','a','m','s',0,2,'A','e','g','e','a','n','_','N','u','m','b','e','r','s',0,'A','e','g','e','a','n','_','N','u', +'m','b','e','r','s',0,2,'U','g','a','r','i','t','i','c',0,'U','g','a','r','i','t','i','c',0, +2,'S','h','a','v','i','a','n',0,'S','h','a','v','i','a','n',0, +2,'O','s','m','a','n','y','a',0,'O','s','m','a','n','y','a',0, +2,'C','y','p','r','i','o','t','_','S','y','l','l','a','b','a','r','y',0,'C','y','p','r','i','o','t','_','S','y','l','l','a', +'b','a','r','y',0,2,'T','a','i','_','X','u','a','n','_','J','i','n','g',0,'T','a','i','_','X','u','a','n','_','J','i','n', +'g','_','S','y','m','b','o','l','s',0,2,'V','S','_','S','u','p',0,'V','a','r','i','a','t','i','o','n','_','S','e','l','e', +'c','t','o','r','s','_','S','u','p','p','l','e','m','e','n','t',0, +2,'A','n','c','i','e','n','t','_','G','r','e','e','k','_','M','u','s','i','c',0,'A','n','c','i','e','n','t','_','G','r','e', +'e','k','_','M','u','s','i','c','a','l','_','N','o','t','a','t','i','o','n',0, +2,'A','n','c','i','e','n','t','_','G','r','e','e','k','_','N','u','m','b','e','r','s',0,'A','n','c','i','e','n','t','_','G', +'r','e','e','k','_','N','u','m','b','e','r','s',0,2,'A','r','a','b','i','c','_','S','u','p',0, +'A','r','a','b','i','c','_','S','u','p','p','l','e','m','e','n','t',0, +2,'B','u','g','i','n','e','s','e',0,'B','u','g','i','n','e','s','e',0, +2,'C','J','K','_','S','t','r','o','k','e','s',0,'C','J','K','_','S','t','r','o','k','e','s',0, +2,'D','i','a','c','r','i','t','i','c','a','l','s','_','S','u','p',0,'C','o','m','b','i','n','i','n','g','_','D','i','a','c', +'r','i','t','i','c','a','l','_','M','a','r','k','s','_','S','u','p','p','l','e','m','e','n','t',0, +2,'C','o','p','t','i','c',0,'C','o','p','t','i','c',0,2,'E','t','h','i','o','p','i','c','_','E','x','t',0, +'E','t','h','i','o','p','i','c','_','E','x','t','e','n','d','e','d',0, +2,'E','t','h','i','o','p','i','c','_','S','u','p',0,'E','t','h','i','o','p','i','c','_','S','u','p','p','l','e','m','e','n', +'t',0,2,'G','e','o','r','g','i','a','n','_','S','u','p',0,'G','e','o','r','g','i','a','n','_','S','u','p','p','l','e','m', +'e','n','t',0,2,'G','l','a','g','o','l','i','t','i','c',0,'G','l','a','g','o','l','i','t','i','c',0, +2,'K','h','a','r','o','s','h','t','h','i',0,'K','h','a','r','o','s','h','t','h','i',0, +2,'M','o','d','i','f','i','e','r','_','T','o','n','e','_','L','e','t','t','e','r','s',0,'M','o','d','i','f','i','e','r','_', +'T','o','n','e','_','L','e','t','t','e','r','s',0,2,'N','e','w','_','T','a','i','_','L','u','e',0, +'N','e','w','_','T','a','i','_','L','u','e',0,2,'O','l','d','_','P','e','r','s','i','a','n',0, +'O','l','d','_','P','e','r','s','i','a','n',0,2,'P','h','o','n','e','t','i','c','_','E','x','t','_','S','u','p',0, +'P','h','o','n','e','t','i','c','_','E','x','t','e','n','s','i','o','n','s','_','S','u','p','p','l','e','m','e','n','t',0, +2,'S','u','p','_','P','u','n','c','t','u','a','t','i','o','n',0,'S','u','p','p','l','e','m','e','n','t','a','l','_','P','u', +'n','c','t','u','a','t','i','o','n',0,2,'S','y','l','o','t','i','_','N','a','g','r','i',0, +'S','y','l','o','t','i','_','N','a','g','r','i',0,2,'T','i','f','i','n','a','g','h',0,'T','i','f','i','n','a','g','h',0, +2,'V','e','r','t','i','c','a','l','_','F','o','r','m','s',0,'V','e','r','t','i','c','a','l','_','F','o','r','m','s',0, +2,'N','K','o',0,'N','K','o',0,2,'B','a','l','i','n','e','s','e',0,'B','a','l','i','n','e','s','e',0, +2,'L','a','t','i','n','_','E','x','t','_','C',0,'L','a','t','i','n','_','E','x','t','e','n','d','e','d','_','C',0, +2,'L','a','t','i','n','_','E','x','t','_','D',0,'L','a','t','i','n','_','E','x','t','e','n','d','e','d','_','D',0, +2,'P','h','a','g','s','_','P','a',0,'P','h','a','g','s','_','P','a',0, +2,'P','h','o','e','n','i','c','i','a','n',0,'P','h','o','e','n','i','c','i','a','n',0, +2,'C','u','n','e','i','f','o','r','m',0,'C','u','n','e','i','f','o','r','m',0, +2,'C','u','n','e','i','f','o','r','m','_','N','u','m','b','e','r','s',0,'C','u','n','e','i','f','o','r','m','_','N','u','m', +'b','e','r','s','_','A','n','d','_','P','u','n','c','t','u','a','t','i','o','n',0, +2,'C','o','u','n','t','i','n','g','_','R','o','d',0,'C','o','u','n','t','i','n','g','_','R','o','d','_','N','u','m','e','r', +'a','l','s',0,2,'S','u','n','d','a','n','e','s','e',0,'S','u','n','d','a','n','e','s','e',0, +2,'L','e','p','c','h','a',0,'L','e','p','c','h','a',0,2,'O','l','_','C','h','i','k','i',0, +'O','l','_','C','h','i','k','i',0,2,'C','y','r','i','l','l','i','c','_','E','x','t','_','A',0, +'C','y','r','i','l','l','i','c','_','E','x','t','e','n','d','e','d','_','A',0, +2,'V','a','i',0,'V','a','i',0,2,'C','y','r','i','l','l','i','c','_','E','x','t','_','B',0, +'C','y','r','i','l','l','i','c','_','E','x','t','e','n','d','e','d','_','B',0, +2,'S','a','u','r','a','s','h','t','r','a',0,'S','a','u','r','a','s','h','t','r','a',0, +2,'K','a','y','a','h','_','L','i',0,'K','a','y','a','h','_','L','i',0, +2,'R','e','j','a','n','g',0,'R','e','j','a','n','g',0,2,'C','h','a','m',0,'C','h','a','m',0, +2,'A','n','c','i','e','n','t','_','S','y','m','b','o','l','s',0,'A','n','c','i','e','n','t','_','S','y','m','b','o','l','s', +0,2,'P','h','a','i','s','t','o','s',0,'P','h','a','i','s','t','o','s','_','D','i','s','c',0, +2,'L','y','c','i','a','n',0,'L','y','c','i','a','n',0,2,'C','a','r','i','a','n',0,'C','a','r','i','a','n',0, +2,'L','y','d','i','a','n',0,'L','y','d','i','a','n',0,2,'M','a','h','j','o','n','g',0, +'M','a','h','j','o','n','g','_','T','i','l','e','s',0,2,'D','o','m','i','n','o',0,'D','o','m','i','n','o','_','T','i','l', +'e','s',0,2,'S','a','m','a','r','i','t','a','n',0,'S','a','m','a','r','i','t','a','n',0, +2,'U','C','A','S','_','E','x','t',0,'U','n','i','f','i','e','d','_','C','a','n','a','d','i','a','n','_','A','b','o','r','i', +'g','i','n','a','l','_','S','y','l','l','a','b','i','c','s','_','E','x','t','e','n','d','e','d',0, +2,'T','a','i','_','T','h','a','m',0,'T','a','i','_','T','h','a','m',0, +2,'V','e','d','i','c','_','E','x','t',0,'V','e','d','i','c','_','E','x','t','e','n','s','i','o','n','s',0, +2,'L','i','s','u',0,'L','i','s','u',0,2,'B','a','m','u','m',0,'B','a','m','u','m',0, +2,'I','n','d','i','c','_','N','u','m','b','e','r','_','F','o','r','m','s',0,'C','o','m','m','o','n','_','I','n','d','i','c', +'_','N','u','m','b','e','r','_','F','o','r','m','s',0,2,'D','e','v','a','n','a','g','a','r','i','_','E','x','t',0, +'D','e','v','a','n','a','g','a','r','i','_','E','x','t','e','n','d','e','d',0, +2,'J','a','m','o','_','E','x','t','_','A',0,'H','a','n','g','u','l','_','J','a','m','o','_','E','x','t','e','n','d','e','d', +'_','A',0,2,'J','a','v','a','n','e','s','e',0,'J','a','v','a','n','e','s','e',0, +2,'M','y','a','n','m','a','r','_','E','x','t','_','A',0,'M','y','a','n','m','a','r','_','E','x','t','e','n','d','e','d','_', +'A',0,2,'T','a','i','_','V','i','e','t',0,'T','a','i','_','V','i','e','t',0, +2,'M','e','e','t','e','i','_','M','a','y','e','k',0,'M','e','e','t','e','i','_','M','a','y','e','k',0, +2,'J','a','m','o','_','E','x','t','_','B',0,'H','a','n','g','u','l','_','J','a','m','o','_','E','x','t','e','n','d','e','d', +'_','B',0,2,'I','m','p','e','r','i','a','l','_','A','r','a','m','a','i','c',0,'I','m','p','e','r','i','a','l','_','A','r', +'a','m','a','i','c',0,2,'O','l','d','_','S','o','u','t','h','_','A','r','a','b','i','a','n',0, +'O','l','d','_','S','o','u','t','h','_','A','r','a','b','i','a','n',0, +2,'A','v','e','s','t','a','n',0,'A','v','e','s','t','a','n',0, +2,'I','n','s','c','r','i','p','t','i','o','n','a','l','_','P','a','r','t','h','i','a','n',0, +'I','n','s','c','r','i','p','t','i','o','n','a','l','_','P','a','r','t','h','i','a','n',0, +2,'I','n','s','c','r','i','p','t','i','o','n','a','l','_','P','a','h','l','a','v','i',0,'I','n','s','c','r','i','p','t','i', +'o','n','a','l','_','P','a','h','l','a','v','i',0,2,'O','l','d','_','T','u','r','k','i','c',0, +'O','l','d','_','T','u','r','k','i','c',0,2,'R','u','m','i',0,'R','u','m','i','_','N','u','m','e','r','a','l','_','S','y', +'m','b','o','l','s',0,2,'K','a','i','t','h','i',0,'K','a','i','t','h','i',0, +2,'E','g','y','p','t','i','a','n','_','H','i','e','r','o','g','l','y','p','h','s',0,'E','g','y','p','t','i','a','n','_','H', +'i','e','r','o','g','l','y','p','h','s',0,2,'E','n','c','l','o','s','e','d','_','A','l','p','h','a','n','u','m','_','S','u', +'p',0,'E','n','c','l','o','s','e','d','_','A','l','p','h','a','n','u','m','e','r','i','c','_','S','u','p','p','l','e','m','e', +'n','t',0,2,'E','n','c','l','o','s','e','d','_','I','d','e','o','g','r','a','p','h','i','c','_','S','u','p',0, +'E','n','c','l','o','s','e','d','_','I','d','e','o','g','r','a','p','h','i','c','_','S','u','p','p','l','e','m','e','n','t',0, +2,'C','J','K','_','E','x','t','_','C',0,'C','J','K','_','U','n','i','f','i','e','d','_','I','d','e','o','g','r','a','p','h', +'s','_','E','x','t','e','n','s','i','o','n','_','C',0,2,'M','a','n','d','a','i','c',0,'M','a','n','d','a','i','c',0, +2,'B','a','t','a','k',0,'B','a','t','a','k',0,2,'E','t','h','i','o','p','i','c','_','E','x','t','_','A',0, +'E','t','h','i','o','p','i','c','_','E','x','t','e','n','d','e','d','_','A',0, +2,'B','r','a','h','m','i',0,'B','r','a','h','m','i',0,2,'B','a','m','u','m','_','S','u','p',0, +'B','a','m','u','m','_','S','u','p','p','l','e','m','e','n','t',0, +2,'K','a','n','a','_','S','u','p',0,'K','a','n','a','_','S','u','p','p','l','e','m','e','n','t',0, +2,'P','l','a','y','i','n','g','_','C','a','r','d','s',0,'P','l','a','y','i','n','g','_','C','a','r','d','s',0, +2,'M','i','s','c','_','P','i','c','t','o','g','r','a','p','h','s',0,'M','i','s','c','e','l','l','a','n','e','o','u','s','_', +'S','y','m','b','o','l','s','_','A','n','d','_','P','i','c','t','o','g','r','a','p','h','s',0, +2,'E','m','o','t','i','c','o','n','s',0,'E','m','o','t','i','c','o','n','s',0, +2,'T','r','a','n','s','p','o','r','t','_','A','n','d','_','M','a','p',0,'T','r','a','n','s','p','o','r','t','_','A','n','d', +'_','M','a','p','_','S','y','m','b','o','l','s',0,2,'A','l','c','h','e','m','i','c','a','l',0, +'A','l','c','h','e','m','i','c','a','l','_','S','y','m','b','o','l','s',0, +2,'C','J','K','_','E','x','t','_','D',0,'C','J','K','_','U','n','i','f','i','e','d','_','I','d','e','o','g','r','a','p','h', +'s','_','E','x','t','e','n','s','i','o','n','_','D',0,2,'A','r','a','b','i','c','_','E','x','t','_','A',0, +'A','r','a','b','i','c','_','E','x','t','e','n','d','e','d','_','A',0, +2,'A','r','a','b','i','c','_','M','a','t','h',0,'A','r','a','b','i','c','_','M','a','t','h','e','m','a','t','i','c','a','l', +'_','A','l','p','h','a','b','e','t','i','c','_','S','y','m','b','o','l','s',0, +2,'C','h','a','k','m','a',0,'C','h','a','k','m','a',0,2,'M','e','e','t','e','i','_','M','a','y','e','k','_','E','x','t', +0,'M','e','e','t','e','i','_','M','a','y','e','k','_','E','x','t','e','n','s','i','o','n','s',0, +2,'M','e','r','o','i','t','i','c','_','C','u','r','s','i','v','e',0,'M','e','r','o','i','t','i','c','_','C','u','r','s','i', +'v','e',0,2,'M','e','r','o','i','t','i','c','_','H','i','e','r','o','g','l','y','p','h','s',0, +'M','e','r','o','i','t','i','c','_','H','i','e','r','o','g','l','y','p','h','s',0, +2,'M','i','a','o',0,'M','i','a','o',0,2,'S','h','a','r','a','d','a',0,'S','h','a','r','a','d','a',0, +2,'S','o','r','a','_','S','o','m','p','e','n','g',0,'S','o','r','a','_','S','o','m','p','e','n','g',0, +2,'S','u','n','d','a','n','e','s','e','_','S','u','p',0,'S','u','n','d','a','n','e','s','e','_','S','u','p','p','l','e','m', +'e','n','t',0,2,'T','a','k','r','i',0,'T','a','k','r','i',0, +2,'B','a','s','s','a','_','V','a','h',0,'B','a','s','s','a','_','V','a','h',0, +2,'C','a','u','c','a','s','i','a','n','_','A','l','b','a','n','i','a','n',0,'C','a','u','c','a','s','i','a','n','_','A','l', +'b','a','n','i','a','n',0,2,'C','o','p','t','i','c','_','E','p','a','c','t','_','N','u','m','b','e','r','s',0, +'C','o','p','t','i','c','_','E','p','a','c','t','_','N','u','m','b','e','r','s',0, +2,'D','i','a','c','r','i','t','i','c','a','l','s','_','E','x','t',0,'C','o','m','b','i','n','i','n','g','_','D','i','a','c', +'r','i','t','i','c','a','l','_','M','a','r','k','s','_','E','x','t','e','n','d','e','d',0, +2,'D','u','p','l','o','y','a','n',0,'D','u','p','l','o','y','a','n',0, +2,'E','l','b','a','s','a','n',0,'E','l','b','a','s','a','n',0, +2,'G','e','o','m','e','t','r','i','c','_','S','h','a','p','e','s','_','E','x','t',0,'G','e','o','m','e','t','r','i','c','_', +'S','h','a','p','e','s','_','E','x','t','e','n','d','e','d',0, +2,'G','r','a','n','t','h','a',0,'G','r','a','n','t','h','a',0, +2,'K','h','o','j','k','i',0,'K','h','o','j','k','i',0,2,'K','h','u','d','a','w','a','d','i',0, +'K','h','u','d','a','w','a','d','i',0,2,'L','a','t','i','n','_','E','x','t','_','E',0,'L','a','t','i','n','_','E','x','t', +'e','n','d','e','d','_','E',0,2,'L','i','n','e','a','r','_','A',0,'L','i','n','e','a','r','_','A',0, +2,'M','a','h','a','j','a','n','i',0,'M','a','h','a','j','a','n','i',0, +2,'M','a','n','i','c','h','a','e','a','n',0,'M','a','n','i','c','h','a','e','a','n',0, +2,'M','e','n','d','e','_','K','i','k','a','k','u','i',0,'M','e','n','d','e','_','K','i','k','a','k','u','i',0, +2,'M','o','d','i',0,'M','o','d','i',0,2,'M','r','o',0,'M','r','o',0, +2,'M','y','a','n','m','a','r','_','E','x','t','_','B',0,'M','y','a','n','m','a','r','_','E','x','t','e','n','d','e','d','_', +'B',0,2,'N','a','b','a','t','a','e','a','n',0,'N','a','b','a','t','a','e','a','n',0, +2,'O','l','d','_','N','o','r','t','h','_','A','r','a','b','i','a','n',0,'O','l','d','_','N','o','r','t','h','_','A','r','a', +'b','i','a','n',0,2,'O','l','d','_','P','e','r','m','i','c',0,'O','l','d','_','P','e','r','m','i','c',0, +2,'O','r','n','a','m','e','n','t','a','l','_','D','i','n','g','b','a','t','s',0,'O','r','n','a','m','e','n','t','a','l','_', +'D','i','n','g','b','a','t','s',0,2,'P','a','h','a','w','h','_','H','m','o','n','g',0,'P','a','h','a','w','h','_','H','m', +'o','n','g',0,2,'P','a','l','m','y','r','e','n','e',0,'P','a','l','m','y','r','e','n','e',0, +2,'P','a','u','_','C','i','n','_','H','a','u',0,'P','a','u','_','C','i','n','_','H','a','u',0, +2,'P','s','a','l','t','e','r','_','P','a','h','l','a','v','i',0,'P','s','a','l','t','e','r','_','P','a','h','l','a','v','i', +0,2,'S','h','o','r','t','h','a','n','d','_','F','o','r','m','a','t','_','C','o','n','t','r','o','l','s',0, +'S','h','o','r','t','h','a','n','d','_','F','o','r','m','a','t','_','C','o','n','t','r','o','l','s',0, +2,'S','i','d','d','h','a','m',0,'S','i','d','d','h','a','m',0, +2,'S','i','n','h','a','l','a','_','A','r','c','h','a','i','c','_','N','u','m','b','e','r','s',0, +'S','i','n','h','a','l','a','_','A','r','c','h','a','i','c','_','N','u','m','b','e','r','s',0, +2,'S','u','p','_','A','r','r','o','w','s','_','C',0,'S','u','p','p','l','e','m','e','n','t','a','l','_','A','r','r','o','w', +'s','_','C',0,2,'T','i','r','h','u','t','a',0,'T','i','r','h','u','t','a',0, +2,'W','a','r','a','n','g','_','C','i','t','i',0,'W','a','r','a','n','g','_','C','i','t','i',0, +2,'A','h','o','m',0,'A','h','o','m',0,2,'A','n','a','t','o','l','i','a','n','_','H','i','e','r','o','g','l','y','p','h', +'s',0,'A','n','a','t','o','l','i','a','n','_','H','i','e','r','o','g','l','y','p','h','s',0, +2,'C','h','e','r','o','k','e','e','_','S','u','p',0,'C','h','e','r','o','k','e','e','_','S','u','p','p','l','e','m','e','n', +'t',0,2,'C','J','K','_','E','x','t','_','E',0,'C','J','K','_','U','n','i','f','i','e','d','_','I','d','e','o','g','r','a', +'p','h','s','_','E','x','t','e','n','s','i','o','n','_','E',0, +2,'E','a','r','l','y','_','D','y','n','a','s','t','i','c','_','C','u','n','e','i','f','o','r','m',0, +'E','a','r','l','y','_','D','y','n','a','s','t','i','c','_','C','u','n','e','i','f','o','r','m',0, +2,'H','a','t','r','a','n',0,'H','a','t','r','a','n',0,2,'M','u','l','t','a','n','i',0, +'M','u','l','t','a','n','i',0,2,'O','l','d','_','H','u','n','g','a','r','i','a','n',0,'O','l','d','_','H','u','n','g','a', +'r','i','a','n',0,2,'S','u','p','_','S','y','m','b','o','l','s','_','A','n','d','_','P','i','c','t','o','g','r','a','p','h', +'s',0,'S','u','p','p','l','e','m','e','n','t','a','l','_','S','y','m','b','o','l','s','_','A','n','d','_','P','i','c','t','o', +'g','r','a','p','h','s',0,2,'S','u','t','t','o','n','_','S','i','g','n','W','r','i','t','i','n','g',0, +'S','u','t','t','o','n','_','S','i','g','n','W','r','i','t','i','n','g',0, +2,'A','d','l','a','m',0,'A','d','l','a','m',0,2,'B','h','a','i','k','s','u','k','i',0, +'B','h','a','i','k','s','u','k','i',0,2,'C','y','r','i','l','l','i','c','_','E','x','t','_','C',0, +'C','y','r','i','l','l','i','c','_','E','x','t','e','n','d','e','d','_','C',0, +2,'G','l','a','g','o','l','i','t','i','c','_','S','u','p',0,'G','l','a','g','o','l','i','t','i','c','_','S','u','p','p','l', +'e','m','e','n','t',0,2,'I','d','e','o','g','r','a','p','h','i','c','_','S','y','m','b','o','l','s',0, +'I','d','e','o','g','r','a','p','h','i','c','_','S','y','m','b','o','l','s','_','A','n','d','_','P','u','n','c','t','u','a','t', +'i','o','n',0,2,'M','a','r','c','h','e','n',0,'M','a','r','c','h','e','n',0, +2,'M','o','n','g','o','l','i','a','n','_','S','u','p',0,'M','o','n','g','o','l','i','a','n','_','S','u','p','p','l','e','m', +'e','n','t',0,2,'N','e','w','a',0,'N','e','w','a',0,2,'O','s','a','g','e',0,'O','s','a','g','e',0, +2,'T','a','n','g','u','t',0,'T','a','n','g','u','t',0,2,'T','a','n','g','u','t','_','C','o','m','p','o','n','e','n','t', +'s',0,'T','a','n','g','u','t','_','C','o','m','p','o','n','e','n','t','s',0, +2,'C','J','K','_','E','x','t','_','F',0,'C','J','K','_','U','n','i','f','i','e','d','_','I','d','e','o','g','r','a','p','h', +'s','_','E','x','t','e','n','s','i','o','n','_','F',0,2,'K','a','n','a','_','E','x','t','_','A',0, +'K','a','n','a','_','E','x','t','e','n','d','e','d','_','A',0, +2,'M','a','s','a','r','a','m','_','G','o','n','d','i',0,'M','a','s','a','r','a','m','_','G','o','n','d','i',0, +2,'N','u','s','h','u',0,'N','u','s','h','u',0,2,'S','o','y','o','m','b','o',0,'S','o','y','o','m','b','o',0, +2,'S','y','r','i','a','c','_','S','u','p',0,'S','y','r','i','a','c','_','S','u','p','p','l','e','m','e','n','t',0, +2,'Z','a','n','a','b','a','z','a','r','_','S','q','u','a','r','e',0,'Z','a','n','a','b','a','z','a','r','_','S','q','u','a', +'r','e',0,2,'C','h','e','s','s','_','S','y','m','b','o','l','s',0,'C','h','e','s','s','_','S','y','m','b','o','l','s',0, +2,'D','o','g','r','a',0,'D','o','g','r','a',0,2,'G','e','o','r','g','i','a','n','_','E','x','t',0, +'G','e','o','r','g','i','a','n','_','E','x','t','e','n','d','e','d',0, +2,'G','u','n','j','a','l','a','_','G','o','n','d','i',0,'G','u','n','j','a','l','a','_','G','o','n','d','i',0, +2,'H','a','n','i','f','i','_','R','o','h','i','n','g','y','a',0,'H','a','n','i','f','i','_','R','o','h','i','n','g','y','a', +0,2,'I','n','d','i','c','_','S','i','y','a','q','_','N','u','m','b','e','r','s',0,'I','n','d','i','c','_','S','i','y','a', +'q','_','N','u','m','b','e','r','s',0,2,'M','a','k','a','s','a','r',0,'M','a','k','a','s','a','r',0, +2,'M','a','y','a','n','_','N','u','m','e','r','a','l','s',0,'M','a','y','a','n','_','N','u','m','e','r','a','l','s',0, +2,'M','e','d','e','f','a','i','d','r','i','n',0,'M','e','d','e','f','a','i','d','r','i','n',0, +2,'O','l','d','_','S','o','g','d','i','a','n',0,'O','l','d','_','S','o','g','d','i','a','n',0, +2,'S','o','g','d','i','a','n',0,'S','o','g','d','i','a','n',0, +2,'E','g','y','p','t','i','a','n','_','H','i','e','r','o','g','l','y','p','h','_','F','o','r','m','a','t','_','C','o','n','t', +'r','o','l','s',0,'E','g','y','p','t','i','a','n','_','H','i','e','r','o','g','l','y','p','h','_','F','o','r','m','a','t','_', +'C','o','n','t','r','o','l','s',0,2,'E','l','y','m','a','i','c',0,'E','l','y','m','a','i','c',0, +2,'N','a','n','d','i','n','a','g','a','r','i',0,'N','a','n','d','i','n','a','g','a','r','i',0, +2,'N','y','i','a','k','e','n','g','_','P','u','a','c','h','u','e','_','H','m','o','n','g',0, +'N','y','i','a','k','e','n','g','_','P','u','a','c','h','u','e','_','H','m','o','n','g',0, +2,'O','t','t','o','m','a','n','_','S','i','y','a','q','_','N','u','m','b','e','r','s',0,'O','t','t','o','m','a','n','_','S', +'i','y','a','q','_','N','u','m','b','e','r','s',0,2,'S','m','a','l','l','_','K','a','n','a','_','E','x','t',0, +'S','m','a','l','l','_','K','a','n','a','_','E','x','t','e','n','s','i','o','n',0, +2,'S','y','m','b','o','l','s','_','A','n','d','_','P','i','c','t','o','g','r','a','p','h','s','_','E','x','t','_','A',0, +'S','y','m','b','o','l','s','_','A','n','d','_','P','i','c','t','o','g','r','a','p','h','s','_','E','x','t','e','n','d','e','d', +'_','A',0,2,'T','a','m','i','l','_','S','u','p',0,'T','a','m','i','l','_','S','u','p','p','l','e','m','e','n','t',0, +2,'W','a','n','c','h','o',0,'W','a','n','c','h','o',0,2,'C','h','o','r','a','s','m','i','a','n',0, +'C','h','o','r','a','s','m','i','a','n',0,2,'C','J','K','_','E','x','t','_','G',0,'C','J','K','_','U','n','i','f','i','e', +'d','_','I','d','e','o','g','r','a','p','h','s','_','E','x','t','e','n','s','i','o','n','_','G',0, +2,'D','i','v','e','s','_','A','k','u','r','u',0,'D','i','v','e','s','_','A','k','u','r','u',0, +2,'K','h','i','t','a','n','_','S','m','a','l','l','_','S','c','r','i','p','t',0,'K','h','i','t','a','n','_','S','m','a','l', +'l','_','S','c','r','i','p','t',0,2,'L','i','s','u','_','S','u','p',0,'L','i','s','u','_','S','u','p','p','l','e','m','e', +'n','t',0,2,'S','y','m','b','o','l','s','_','F','o','r','_','L','e','g','a','c','y','_','C','o','m','p','u','t','i','n','g', +0,'S','y','m','b','o','l','s','_','F','o','r','_','L','e','g','a','c','y','_','C','o','m','p','u','t','i','n','g',0, +2,'T','a','n','g','u','t','_','S','u','p',0,'T','a','n','g','u','t','_','S','u','p','p','l','e','m','e','n','t',0, +2,'Y','e','z','i','d','i',0,'Y','e','z','i','d','i',0,2,'A','r','a','b','i','c','_','E','x','t','_','B',0, +'A','r','a','b','i','c','_','E','x','t','e','n','d','e','d','_','B',0, +2,'C','y','p','r','o','_','M','i','n','o','a','n',0,'C','y','p','r','o','_','M','i','n','o','a','n',0, +2,'E','t','h','i','o','p','i','c','_','E','x','t','_','B',0,'E','t','h','i','o','p','i','c','_','E','x','t','e','n','d','e', +'d','_','B',0,2,'K','a','n','a','_','E','x','t','_','B',0,'K','a','n','a','_','E','x','t','e','n','d','e','d','_','B',0, +2,'L','a','t','i','n','_','E','x','t','_','F',0,'L','a','t','i','n','_','E','x','t','e','n','d','e','d','_','F',0, +2,'L','a','t','i','n','_','E','x','t','_','G',0,'L','a','t','i','n','_','E','x','t','e','n','d','e','d','_','G',0, +2,'O','l','d','_','U','y','g','h','u','r',0,'O','l','d','_','U','y','g','h','u','r',0, +2,'T','a','n','g','s','a',0,'T','a','n','g','s','a',0,2,'T','o','t','o',0,'T','o','t','o',0, +2,'U','C','A','S','_','E','x','t','_','A',0,'U','n','i','f','i','e','d','_','C','a','n','a','d','i','a','n','_','A','b','o', +'r','i','g','i','n','a','l','_','S','y','l','l','a','b','i','c','s','_','E','x','t','e','n','d','e','d','_','A',0, +2,'V','i','t','h','k','u','q','i',0,'V','i','t','h','k','u','q','i',0, +2,'Z','n','a','m','e','n','n','y','_','M','u','s','i','c',0,'Z','n','a','m','e','n','n','y','_','M','u','s','i','c','a','l', +'_','N','o','t','a','t','i','o','n',0,2,'A','r','a','b','i','c','_','E','x','t','_','C',0, +'A','r','a','b','i','c','_','E','x','t','e','n','d','e','d','_','C',0, +2,'C','J','K','_','E','x','t','_','H',0,'C','J','K','_','U','n','i','f','i','e','d','_','I','d','e','o','g','r','a','p','h', +'s','_','E','x','t','e','n','s','i','o','n','_','H',0,2,'C','y','r','i','l','l','i','c','_','E','x','t','_','D',0, +'C','y','r','i','l','l','i','c','_','E','x','t','e','n','d','e','d','_','D',0, +2,'D','e','v','a','n','a','g','a','r','i','_','E','x','t','_','A',0,'D','e','v','a','n','a','g','a','r','i','_','E','x','t', +'e','n','d','e','d','_','A',0,2,'K','a','k','t','o','v','i','k','_','N','u','m','e','r','a','l','s',0, +'K','a','k','t','o','v','i','k','_','N','u','m','e','r','a','l','s',0, +2,'K','a','w','i',0,'K','a','w','i',0,2,'N','a','g','_','M','u','n','d','a','r','i',0, +'N','a','g','_','M','u','n','d','a','r','i',0,2,'c','c','c',0,'C','a','n','o','n','i','c','a','l','_','C','o','m','b','i', +'n','i','n','g','_','C','l','a','s','s',0,2,'d','t',0,'D','e','c','o','m','p','o','s','i','t','i','o','n','_','T','y','p', +'e',0,3,'N','o','n','e',0,'N','o','n','e',0,'n','o','n','e',0, +3,'C','a','n',0,'C','a','n','o','n','i','c','a','l',0,'c','a','n',0, +3,'C','o','m',0,'C','o','m','p','a','t',0,'c','o','m',0, +3,'E','n','c',0,'C','i','r','c','l','e',0,'e','n','c',0, +3,'F','i','n',0,'F','i','n','a','l',0,'f','i','n',0,3,'F','o','n','t',0,'F','o','n','t',0, +'f','o','n','t',0,3,'F','r','a',0,'F','r','a','c','t','i','o','n',0,'f','r','a',0, +3,'I','n','i','t',0,'I','n','i','t','i','a','l',0,'i','n','i','t',0, +3,'I','s','o',0,'I','s','o','l','a','t','e','d',0,'i','s','o',0, +3,'M','e','d',0,'M','e','d','i','a','l',0,'m','e','d',0, +3,'N','a','r',0,'N','a','r','r','o','w',0,'n','a','r',0, +3,'N','b',0,'N','o','b','r','e','a','k',0,'n','b',0,3,'S','m','l',0,'S','m','a','l','l',0, +'s','m','l',0,3,'S','q','r',0,'S','q','u','a','r','e',0,'s','q','r',0, +3,'S','u','b',0,'S','u','b',0,'s','u','b',0,3,'S','u','p',0,'S','u','p','e','r',0, +'s','u','p',0,3,'V','e','r','t',0,'V','e','r','t','i','c','a','l',0,'v','e','r','t',0, +3,'W','i','d','e',0,'W','i','d','e',0,'w','i','d','e',0, +2,'e','a',0,'E','a','s','t','_','A','s','i','a','n','_','W','i','d','t','h',0, +2,'N',0,'N','e','u','t','r','a','l',0,2,'A',0,'A','m','b','i','g','u','o','u','s',0, +2,'H',0,'H','a','l','f','w','i','d','t','h',0,2,'F',0,'F','u','l','l','w','i','d','t','h',0, +2,'N','a',0,'N','a','r','r','o','w',0,2,'W',0,'W','i','d','e',0, +2,'g','c',0,'G','e','n','e','r','a','l','_','C','a','t','e','g','o','r','y',0, +2,'C','n',0,'U','n','a','s','s','i','g','n','e','d',0,2,'L','u',0,'U','p','p','e','r','c','a','s','e','_','L','e','t', +'t','e','r',0,2,'L','l',0,'L','o','w','e','r','c','a','s','e','_','L','e','t','t','e','r',0, +2,'L','t',0,'T','i','t','l','e','c','a','s','e','_','L','e','t','t','e','r',0, +2,'L','m',0,'M','o','d','i','f','i','e','r','_','L','e','t','t','e','r',0, +2,'L','o',0,'O','t','h','e','r','_','L','e','t','t','e','r',0, +2,'M','n',0,'N','o','n','s','p','a','c','i','n','g','_','M','a','r','k',0, +2,'M','e',0,'E','n','c','l','o','s','i','n','g','_','M','a','r','k',0, +2,'M','c',0,'S','p','a','c','i','n','g','_','M','a','r','k',0, +3,'N','d',0,'D','e','c','i','m','a','l','_','N','u','m','b','e','r',0,'d','i','g','i','t',0, +2,'N','l',0,'L','e','t','t','e','r','_','N','u','m','b','e','r',0, +2,'N','o',0,'O','t','h','e','r','_','N','u','m','b','e','r',0, +2,'Z','s',0,'S','p','a','c','e','_','S','e','p','a','r','a','t','o','r',0, +2,'Z','l',0,'L','i','n','e','_','S','e','p','a','r','a','t','o','r',0, +2,'Z','p',0,'P','a','r','a','g','r','a','p','h','_','S','e','p','a','r','a','t','o','r',0, +3,'C','c',0,'C','o','n','t','r','o','l',0,'c','n','t','r','l',0, +2,'C','f',0,'F','o','r','m','a','t',0,2,'C','o',0,'P','r','i','v','a','t','e','_','U','s','e',0, +2,'C','s',0,'S','u','r','r','o','g','a','t','e',0,2,'P','d',0,'D','a','s','h','_','P','u','n','c','t','u','a','t','i', +'o','n',0,2,'P','s',0,'O','p','e','n','_','P','u','n','c','t','u','a','t','i','o','n',0, +2,'P','e',0,'C','l','o','s','e','_','P','u','n','c','t','u','a','t','i','o','n',0, +2,'P','c',0,'C','o','n','n','e','c','t','o','r','_','P','u','n','c','t','u','a','t','i','o','n',0, +2,'P','o',0,'O','t','h','e','r','_','P','u','n','c','t','u','a','t','i','o','n',0, +2,'S','m',0,'M','a','t','h','_','S','y','m','b','o','l',0, +2,'S','c',0,'C','u','r','r','e','n','c','y','_','S','y','m','b','o','l',0, +2,'S','k',0,'M','o','d','i','f','i','e','r','_','S','y','m','b','o','l',0, +2,'S','o',0,'O','t','h','e','r','_','S','y','m','b','o','l',0, +2,'P','i',0,'I','n','i','t','i','a','l','_','P','u','n','c','t','u','a','t','i','o','n',0, +2,'P','f',0,'F','i','n','a','l','_','P','u','n','c','t','u','a','t','i','o','n',0, +2,'j','g',0,'J','o','i','n','i','n','g','_','G','r','o','u','p',0, +2,'N','o','_','J','o','i','n','i','n','g','_','G','r','o','u','p',0,'N','o','_','J','o','i','n','i','n','g','_','G','r','o', +'u','p',0,2,'A','i','n',0,'A','i','n',0,2,'A','l','a','p','h',0,'A','l','a','p','h',0, +2,'A','l','e','f',0,'A','l','e','f',0,2,'B','e','h',0,'B','e','h',0, +2,'B','e','t','h',0,'B','e','t','h',0,2,'D','a','l',0,'D','a','l',0, +2,'D','a','l','a','t','h','_','R','i','s','h',0,'D','a','l','a','t','h','_','R','i','s','h',0, +2,'E',0,'E',0,2,'F','e','h',0,'F','e','h',0,2,'F','i','n','a','l','_','S','e','m','k','a','t','h',0, +'F','i','n','a','l','_','S','e','m','k','a','t','h',0,2,'G','a','f',0,'G','a','f',0, +2,'G','a','m','a','l',0,'G','a','m','a','l',0,2,'H','a','h',0,'H','a','h',0, +2,'T','e','h','_','M','a','r','b','u','t','a','_','G','o','a','l',0,'H','a','m','z','a','_','O','n','_','H','e','h','_','G', +'o','a','l',0,2,'H','e',0,'H','e',0,2,'H','e','h',0,'H','e','h',0, +2,'H','e','h','_','G','o','a','l',0,'H','e','h','_','G','o','a','l',0, +2,'H','e','t','h',0,'H','e','t','h',0,2,'K','a','f',0,'K','a','f',0, +2,'K','a','p','h',0,'K','a','p','h',0,2,'K','n','o','t','t','e','d','_','H','e','h',0, +'K','n','o','t','t','e','d','_','H','e','h',0,2,'L','a','m',0,'L','a','m',0, +2,'L','a','m','a','d','h',0,'L','a','m','a','d','h',0,2,'M','e','e','m',0,'M','e','e','m',0, +2,'M','i','m',0,'M','i','m',0,2,'N','o','o','n',0,'N','o','o','n',0, +2,'N','u','n',0,'N','u','n',0,2,'P','e',0,'P','e',0, +2,'Q','a','f',0,'Q','a','f',0,2,'Q','a','p','h',0,'Q','a','p','h',0, +2,'R','e','h',0,'R','e','h',0,2,'R','e','v','e','r','s','e','d','_','P','e',0,'R','e','v','e','r','s','e','d','_','P', +'e',0,2,'S','a','d',0,'S','a','d',0,2,'S','a','d','h','e',0,'S','a','d','h','e',0, +2,'S','e','e','n',0,'S','e','e','n',0,2,'S','e','m','k','a','t','h',0,'S','e','m','k','a','t','h',0, +2,'S','h','i','n',0,'S','h','i','n',0,2,'S','w','a','s','h','_','K','a','f',0,'S','w','a','s','h','_','K','a','f',0, +2,'S','y','r','i','a','c','_','W','a','w',0,'S','y','r','i','a','c','_','W','a','w',0, +2,'T','a','h',0,'T','a','h',0,2,'T','a','w',0,'T','a','w',0, +2,'T','e','h','_','M','a','r','b','u','t','a',0,'T','e','h','_','M','a','r','b','u','t','a',0, +2,'T','e','t','h',0,'T','e','t','h',0,2,'W','a','w',0,'W','a','w',0, +2,'Y','e','h',0,'Y','e','h',0,2,'Y','e','h','_','B','a','r','r','e','e',0,'Y','e','h','_','B','a','r','r','e','e',0, +2,'Y','e','h','_','W','i','t','h','_','T','a','i','l',0,'Y','e','h','_','W','i','t','h','_','T','a','i','l',0, +2,'Y','u','d','h',0,'Y','u','d','h',0,2,'Y','u','d','h','_','H','e',0,'Y','u','d','h','_','H','e',0, +2,'Z','a','i','n',0,'Z','a','i','n',0,2,'F','e',0,'F','e',0, +2,'K','h','a','p','h',0,'K','h','a','p','h',0,2,'Z','h','a','i','n',0,'Z','h','a','i','n',0, +2,'B','u','r','u','s','h','a','s','k','i','_','Y','e','h','_','B','a','r','r','e','e',0,'B','u','r','u','s','h','a','s','k', +'i','_','Y','e','h','_','B','a','r','r','e','e',0,2,'F','a','r','s','i','_','Y','e','h',0, +'F','a','r','s','i','_','Y','e','h',0,2,'N','y','a',0,'N','y','a',0, +2,'R','o','h','i','n','g','y','a','_','Y','e','h',0,'R','o','h','i','n','g','y','a','_','Y','e','h',0, +2,'M','a','n','i','c','h','a','e','a','n','_','A','l','e','p','h',0,'M','a','n','i','c','h','a','e','a','n','_','A','l','e', +'p','h',0,2,'M','a','n','i','c','h','a','e','a','n','_','A','y','i','n',0,'M','a','n','i','c','h','a','e','a','n','_','A', +'y','i','n',0,2,'M','a','n','i','c','h','a','e','a','n','_','B','e','t','h',0,'M','a','n','i','c','h','a','e','a','n','_', +'B','e','t','h',0,2,'M','a','n','i','c','h','a','e','a','n','_','D','a','l','e','t','h',0, +'M','a','n','i','c','h','a','e','a','n','_','D','a','l','e','t','h',0, +2,'M','a','n','i','c','h','a','e','a','n','_','D','h','a','m','e','d','h',0,'M','a','n','i','c','h','a','e','a','n','_','D', +'h','a','m','e','d','h',0,2,'M','a','n','i','c','h','a','e','a','n','_','F','i','v','e',0, +'M','a','n','i','c','h','a','e','a','n','_','F','i','v','e',0, +2,'M','a','n','i','c','h','a','e','a','n','_','G','i','m','e','l',0,'M','a','n','i','c','h','a','e','a','n','_','G','i','m', +'e','l',0,2,'M','a','n','i','c','h','a','e','a','n','_','H','e','t','h',0,'M','a','n','i','c','h','a','e','a','n','_','H', +'e','t','h',0,2,'M','a','n','i','c','h','a','e','a','n','_','H','u','n','d','r','e','d',0, +'M','a','n','i','c','h','a','e','a','n','_','H','u','n','d','r','e','d',0, +2,'M','a','n','i','c','h','a','e','a','n','_','K','a','p','h',0,'M','a','n','i','c','h','a','e','a','n','_','K','a','p','h', +0,2,'M','a','n','i','c','h','a','e','a','n','_','L','a','m','e','d','h',0,'M','a','n','i','c','h','a','e','a','n','_','L', +'a','m','e','d','h',0,2,'M','a','n','i','c','h','a','e','a','n','_','M','e','m',0,'M','a','n','i','c','h','a','e','a','n', +'_','M','e','m',0,2,'M','a','n','i','c','h','a','e','a','n','_','N','u','n',0,'M','a','n','i','c','h','a','e','a','n','_', +'N','u','n',0,2,'M','a','n','i','c','h','a','e','a','n','_','O','n','e',0,'M','a','n','i','c','h','a','e','a','n','_','O', +'n','e',0,2,'M','a','n','i','c','h','a','e','a','n','_','P','e',0,'M','a','n','i','c','h','a','e','a','n','_','P','e',0, +2,'M','a','n','i','c','h','a','e','a','n','_','Q','o','p','h',0,'M','a','n','i','c','h','a','e','a','n','_','Q','o','p','h', +0,2,'M','a','n','i','c','h','a','e','a','n','_','R','e','s','h',0,'M','a','n','i','c','h','a','e','a','n','_','R','e','s', +'h',0,2,'M','a','n','i','c','h','a','e','a','n','_','S','a','d','h','e',0,'M','a','n','i','c','h','a','e','a','n','_','S', +'a','d','h','e',0,2,'M','a','n','i','c','h','a','e','a','n','_','S','a','m','e','k','h',0, +'M','a','n','i','c','h','a','e','a','n','_','S','a','m','e','k','h',0, +2,'M','a','n','i','c','h','a','e','a','n','_','T','a','w',0,'M','a','n','i','c','h','a','e','a','n','_','T','a','w',0, +2,'M','a','n','i','c','h','a','e','a','n','_','T','e','n',0,'M','a','n','i','c','h','a','e','a','n','_','T','e','n',0, +2,'M','a','n','i','c','h','a','e','a','n','_','T','e','t','h',0,'M','a','n','i','c','h','a','e','a','n','_','T','e','t','h', +0,2,'M','a','n','i','c','h','a','e','a','n','_','T','h','a','m','e','d','h',0,'M','a','n','i','c','h','a','e','a','n','_', +'T','h','a','m','e','d','h',0,2,'M','a','n','i','c','h','a','e','a','n','_','T','w','e','n','t','y',0, +'M','a','n','i','c','h','a','e','a','n','_','T','w','e','n','t','y',0, +2,'M','a','n','i','c','h','a','e','a','n','_','W','a','w',0,'M','a','n','i','c','h','a','e','a','n','_','W','a','w',0, +2,'M','a','n','i','c','h','a','e','a','n','_','Y','o','d','h',0,'M','a','n','i','c','h','a','e','a','n','_','Y','o','d','h', +0,2,'M','a','n','i','c','h','a','e','a','n','_','Z','a','y','i','n',0,'M','a','n','i','c','h','a','e','a','n','_','Z','a', +'y','i','n',0,2,'S','t','r','a','i','g','h','t','_','W','a','w',0,'S','t','r','a','i','g','h','t','_','W','a','w',0, +2,'A','f','r','i','c','a','n','_','F','e','h',0,'A','f','r','i','c','a','n','_','F','e','h',0, +2,'A','f','r','i','c','a','n','_','N','o','o','n',0,'A','f','r','i','c','a','n','_','N','o','o','n',0, +2,'A','f','r','i','c','a','n','_','Q','a','f',0,'A','f','r','i','c','a','n','_','Q','a','f',0, +2,'M','a','l','a','y','a','l','a','m','_','B','h','a',0,'M','a','l','a','y','a','l','a','m','_','B','h','a',0, +2,'M','a','l','a','y','a','l','a','m','_','J','a',0,'M','a','l','a','y','a','l','a','m','_','J','a',0, +2,'M','a','l','a','y','a','l','a','m','_','L','l','a',0,'M','a','l','a','y','a','l','a','m','_','L','l','a',0, +2,'M','a','l','a','y','a','l','a','m','_','L','l','l','a',0,'M','a','l','a','y','a','l','a','m','_','L','l','l','a',0, +2,'M','a','l','a','y','a','l','a','m','_','N','g','a',0,'M','a','l','a','y','a','l','a','m','_','N','g','a',0, +2,'M','a','l','a','y','a','l','a','m','_','N','n','a',0,'M','a','l','a','y','a','l','a','m','_','N','n','a',0, +2,'M','a','l','a','y','a','l','a','m','_','N','n','n','a',0,'M','a','l','a','y','a','l','a','m','_','N','n','n','a',0, +2,'M','a','l','a','y','a','l','a','m','_','N','y','a',0,'M','a','l','a','y','a','l','a','m','_','N','y','a',0, +2,'M','a','l','a','y','a','l','a','m','_','R','a',0,'M','a','l','a','y','a','l','a','m','_','R','a',0, +2,'M','a','l','a','y','a','l','a','m','_','S','s','a',0,'M','a','l','a','y','a','l','a','m','_','S','s','a',0, +2,'M','a','l','a','y','a','l','a','m','_','T','t','a',0,'M','a','l','a','y','a','l','a','m','_','T','t','a',0, +2,'H','a','n','i','f','i','_','R','o','h','i','n','g','y','a','_','K','i','n','n','a','_','Y','a',0, +'H','a','n','i','f','i','_','R','o','h','i','n','g','y','a','_','K','i','n','n','a','_','Y','a',0, +2,'H','a','n','i','f','i','_','R','o','h','i','n','g','y','a','_','P','a',0,'H','a','n','i','f','i','_','R','o','h','i','n', +'g','y','a','_','P','a',0,2,'T','h','i','n','_','Y','e','h',0,'T','h','i','n','_','Y','e','h',0, +2,'V','e','r','t','i','c','a','l','_','T','a','i','l',0,'V','e','r','t','i','c','a','l','_','T','a','i','l',0, +2,'j','t',0,'J','o','i','n','i','n','g','_','T','y','p','e',0, +2,'U',0,'N','o','n','_','J','o','i','n','i','n','g',0,2,'C',0,'J','o','i','n','_','C','a','u','s','i','n','g',0, +2,'D',0,'D','u','a','l','_','J','o','i','n','i','n','g',0, +2,'L',0,'L','e','f','t','_','J','o','i','n','i','n','g',0, +2,'R',0,'R','i','g','h','t','_','J','o','i','n','i','n','g',0, +2,'T',0,'T','r','a','n','s','p','a','r','e','n','t',0,2,'l','b',0,'L','i','n','e','_','B','r','e','a','k',0, +2,'X','X',0,'U','n','k','n','o','w','n',0,2,'A','I',0,'A','m','b','i','g','u','o','u','s',0, +2,'A','L',0,'A','l','p','h','a','b','e','t','i','c',0,2,'B','2',0,'B','r','e','a','k','_','B','o','t','h',0, +2,'B','A',0,'B','r','e','a','k','_','A','f','t','e','r',0, +2,'B','B',0,'B','r','e','a','k','_','B','e','f','o','r','e',0, +2,'B','K',0,'M','a','n','d','a','t','o','r','y','_','B','r','e','a','k',0, +2,'C','B',0,'C','o','n','t','i','n','g','e','n','t','_','B','r','e','a','k',0, +2,'C','L',0,'C','l','o','s','e','_','P','u','n','c','t','u','a','t','i','o','n',0, +2,'C','M',0,'C','o','m','b','i','n','i','n','g','_','M','a','r','k',0, +2,'C','R',0,'C','a','r','r','i','a','g','e','_','R','e','t','u','r','n',0, +2,'E','X',0,'E','x','c','l','a','m','a','t','i','o','n',0, +2,'G','L',0,'G','l','u','e',0,2,'H','Y',0,'H','y','p','h','e','n',0, +2,'I','D',0,'I','d','e','o','g','r','a','p','h','i','c',0, +3,'I','N',0,'I','n','s','e','p','a','r','a','b','l','e',0,'I','n','s','e','p','e','r','a','b','l','e',0, +2,'I','S',0,'I','n','f','i','x','_','N','u','m','e','r','i','c',0, +2,'L','F',0,'L','i','n','e','_','F','e','e','d',0,2,'N','S',0,'N','o','n','s','t','a','r','t','e','r',0, +2,'N','U',0,'N','u','m','e','r','i','c',0,2,'O','P',0,'O','p','e','n','_','P','u','n','c','t','u','a','t','i','o','n', +0,2,'P','O',0,'P','o','s','t','f','i','x','_','N','u','m','e','r','i','c',0, +2,'P','R',0,'P','r','e','f','i','x','_','N','u','m','e','r','i','c',0, +2,'Q','U',0,'Q','u','o','t','a','t','i','o','n',0,2,'S','A',0,'C','o','m','p','l','e','x','_','C','o','n','t','e','x', +'t',0,2,'S','G',0,'S','u','r','r','o','g','a','t','e',0, +2,'S','P',0,'S','p','a','c','e',0,2,'S','Y',0,'B','r','e','a','k','_','S','y','m','b','o','l','s',0, +2,'Z','W',0,'Z','W','S','p','a','c','e',0,2,'N','L',0,'N','e','x','t','_','L','i','n','e',0, +2,'W','J',0,'W','o','r','d','_','J','o','i','n','e','r',0, +2,'H','2',0,'H','2',0,2,'H','3',0,'H','3',0,2,'J','L',0,'J','L',0, +2,'J','T',0,'J','T',0,2,'J','V',0,'J','V',0,2,'C','P',0,'C','l','o','s','e','_','P','a','r','e','n','t','h','e', +'s','i','s',0,2,'C','J',0,'C','o','n','d','i','t','i','o','n','a','l','_','J','a','p','a','n','e','s','e','_','S','t','a', +'r','t','e','r',0,2,'H','L',0,'H','e','b','r','e','w','_','L','e','t','t','e','r',0, +2,'E','B',0,'E','_','B','a','s','e',0,2,'E','M',0,'E','_','M','o','d','i','f','i','e','r',0, +2,'Z','W','J',0,'Z','W','J',0,2,'n','t',0,'N','u','m','e','r','i','c','_','T','y','p','e',0, +2,'N','o','n','e',0,'N','o','n','e',0,2,'D','e',0,'D','e','c','i','m','a','l',0, +2,'D','i',0,'D','i','g','i','t',0,2,'N','u',0,'N','u','m','e','r','i','c',0, +2,'s','c',0,'S','c','r','i','p','t',0,2,'Z','y','y','y',0,'C','o','m','m','o','n',0, +3,'Z','i','n','h',0,'I','n','h','e','r','i','t','e','d',0,'Q','a','a','i',0, +2,'A','r','a','b',0,'A','r','a','b','i','c',0,2,'A','r','m','n',0,'A','r','m','e','n','i','a','n',0, +2,'B','e','n','g',0,'B','e','n','g','a','l','i',0,2,'B','o','p','o',0,'B','o','p','o','m','o','f','o',0, +2,'C','h','e','r',0,'C','h','e','r','o','k','e','e',0,3,'C','o','p','t',0,'C','o','p','t','i','c',0, +'Q','a','a','c',0,2,'C','y','r','l',0,'C','y','r','i','l','l','i','c',0, +2,'D','s','r','t',0,'D','e','s','e','r','e','t',0,2,'D','e','v','a',0,'D','e','v','a','n','a','g','a','r','i',0, +2,'E','t','h','i',0,'E','t','h','i','o','p','i','c',0,2,'G','e','o','r',0,'G','e','o','r','g','i','a','n',0, +2,'G','o','t','h',0,'G','o','t','h','i','c',0,2,'G','r','e','k',0,'G','r','e','e','k',0, +2,'G','u','j','r',0,'G','u','j','a','r','a','t','i',0,2,'G','u','r','u',0,'G','u','r','m','u','k','h','i',0, +2,'H','a','n','i',0,'H','a','n',0,2,'H','a','n','g',0,'H','a','n','g','u','l',0, +2,'H','e','b','r',0,'H','e','b','r','e','w',0,2,'H','i','r','a',0,'H','i','r','a','g','a','n','a',0, +2,'K','n','d','a',0,'K','a','n','n','a','d','a',0,2,'K','a','n','a',0,'K','a','t','a','k','a','n','a',0, +2,'K','h','m','r',0,'K','h','m','e','r',0,2,'L','a','o','o',0,'L','a','o',0, +2,'L','a','t','n',0,'L','a','t','i','n',0,2,'M','l','y','m',0,'M','a','l','a','y','a','l','a','m',0, +2,'M','o','n','g',0,'M','o','n','g','o','l','i','a','n',0, +2,'M','y','m','r',0,'M','y','a','n','m','a','r',0,2,'O','g','a','m',0,'O','g','h','a','m',0, +2,'I','t','a','l',0,'O','l','d','_','I','t','a','l','i','c',0, +2,'O','r','y','a',0,'O','r','i','y','a',0,2,'R','u','n','r',0,'R','u','n','i','c',0, +2,'S','i','n','h',0,'S','i','n','h','a','l','a',0,2,'S','y','r','c',0,'S','y','r','i','a','c',0, +2,'T','a','m','l',0,'T','a','m','i','l',0,2,'T','e','l','u',0,'T','e','l','u','g','u',0, +2,'T','h','a','a',0,'T','h','a','a','n','a',0,2,'T','i','b','t',0,'T','i','b','e','t','a','n',0, +2,'C','a','n','s',0,'C','a','n','a','d','i','a','n','_','A','b','o','r','i','g','i','n','a','l',0, +2,'Y','i','i','i',0,'Y','i',0,2,'T','g','l','g',0,'T','a','g','a','l','o','g',0, +2,'H','a','n','o',0,'H','a','n','u','n','o','o',0,2,'B','u','h','d',0,'B','u','h','i','d',0, +2,'T','a','g','b',0,'T','a','g','b','a','n','w','a',0,2,'B','r','a','i',0,'B','r','a','i','l','l','e',0, +2,'C','p','r','t',0,'C','y','p','r','i','o','t',0,2,'L','i','m','b',0,'L','i','m','b','u',0, +2,'L','i','n','b',0,'L','i','n','e','a','r','_','B',0,2,'O','s','m','a',0,'O','s','m','a','n','y','a',0, +2,'S','h','a','w',0,'S','h','a','v','i','a','n',0,2,'T','a','l','e',0,'T','a','i','_','L','e',0, +2,'U','g','a','r',0,'U','g','a','r','i','t','i','c',0,2,'H','r','k','t',0,'K','a','t','a','k','a','n','a','_','O','r', +'_','H','i','r','a','g','a','n','a',0,2,'B','u','g','i',0,'B','u','g','i','n','e','s','e',0, +2,'G','l','a','g',0,'G','l','a','g','o','l','i','t','i','c',0, +2,'K','h','a','r',0,'K','h','a','r','o','s','h','t','h','i',0, +2,'S','y','l','o',0,'S','y','l','o','t','i','_','N','a','g','r','i',0, +2,'T','a','l','u',0,'N','e','w','_','T','a','i','_','L','u','e',0, +2,'T','f','n','g',0,'T','i','f','i','n','a','g','h',0,2,'X','p','e','o',0,'O','l','d','_','P','e','r','s','i','a','n', +0,2,'B','a','l','i',0,'B','a','l','i','n','e','s','e',0, +2,'B','a','t','k',0,'B','a','t','a','k',0,2,'B','l','i','s',0,'B','l','i','s',0, +2,'B','r','a','h',0,'B','r','a','h','m','i',0,2,'C','i','r','t',0,'C','i','r','t',0, +2,'C','y','r','s',0,'C','y','r','s',0,2,'E','g','y','d',0,'E','g','y','d',0, +2,'E','g','y','h',0,'E','g','y','h',0,2,'E','g','y','p',0,'E','g','y','p','t','i','a','n','_','H','i','e','r','o','g', +'l','y','p','h','s',0,2,'G','e','o','k',0,'G','e','o','k',0, +2,'H','a','n','s',0,'H','a','n','s',0,2,'H','a','n','t',0,'H','a','n','t',0, +2,'H','m','n','g',0,'P','a','h','a','w','h','_','H','m','o','n','g',0, +2,'H','u','n','g',0,'O','l','d','_','H','u','n','g','a','r','i','a','n',0, +2,'I','n','d','s',0,'I','n','d','s',0,2,'J','a','v','a',0,'J','a','v','a','n','e','s','e',0, +2,'K','a','l','i',0,'K','a','y','a','h','_','L','i',0,2,'L','a','t','f',0,'L','a','t','f',0, +2,'L','a','t','g',0,'L','a','t','g',0,2,'L','e','p','c',0,'L','e','p','c','h','a',0, +2,'L','i','n','a',0,'L','i','n','e','a','r','_','A',0,2,'M','a','n','d',0,'M','a','n','d','a','i','c',0, +2,'M','a','y','a',0,'M','a','y','a',0,2,'M','e','r','o',0,'M','e','r','o','i','t','i','c','_','H','i','e','r','o','g', +'l','y','p','h','s',0,2,'N','k','o','o',0,'N','k','o',0, +2,'O','r','k','h',0,'O','l','d','_','T','u','r','k','i','c',0, +2,'P','e','r','m',0,'O','l','d','_','P','e','r','m','i','c',0, +2,'P','h','a','g',0,'P','h','a','g','s','_','P','a',0,2,'P','h','n','x',0,'P','h','o','e','n','i','c','i','a','n',0, +2,'P','l','r','d',0,'M','i','a','o',0,2,'R','o','r','o',0,'R','o','r','o',0, +2,'S','a','r','a',0,'S','a','r','a',0,2,'S','y','r','e',0,'S','y','r','e',0, +2,'S','y','r','j',0,'S','y','r','j',0,2,'S','y','r','n',0,'S','y','r','n',0, +2,'T','e','n','g',0,'T','e','n','g',0,2,'V','a','i','i',0,'V','a','i',0, +2,'V','i','s','p',0,'V','i','s','p',0,2,'X','s','u','x',0,'C','u','n','e','i','f','o','r','m',0, +2,'Z','x','x','x',0,'Z','x','x','x',0,2,'Z','z','z','z',0,'U','n','k','n','o','w','n',0, +2,'C','a','r','i',0,'C','a','r','i','a','n',0,2,'J','p','a','n',0,'J','p','a','n',0, +2,'L','a','n','a',0,'T','a','i','_','T','h','a','m',0,2,'L','y','c','i',0,'L','y','c','i','a','n',0, +2,'L','y','d','i',0,'L','y','d','i','a','n',0,2,'O','l','c','k',0,'O','l','_','C','h','i','k','i',0, +2,'R','j','n','g',0,'R','e','j','a','n','g',0,2,'S','a','u','r',0,'S','a','u','r','a','s','h','t','r','a',0, +2,'S','g','n','w',0,'S','i','g','n','W','r','i','t','i','n','g',0, +2,'S','u','n','d',0,'S','u','n','d','a','n','e','s','e',0, +2,'M','o','o','n',0,'M','o','o','n',0,2,'M','t','e','i',0,'M','e','e','t','e','i','_','M','a','y','e','k',0, +2,'A','r','m','i',0,'I','m','p','e','r','i','a','l','_','A','r','a','m','a','i','c',0, +2,'A','v','s','t',0,'A','v','e','s','t','a','n',0,2,'C','a','k','m',0,'C','h','a','k','m','a',0, +2,'K','o','r','e',0,'K','o','r','e',0,2,'K','t','h','i',0,'K','a','i','t','h','i',0, +2,'M','a','n','i',0,'M','a','n','i','c','h','a','e','a','n',0, +2,'P','h','l','i',0,'I','n','s','c','r','i','p','t','i','o','n','a','l','_','P','a','h','l','a','v','i',0, +2,'P','h','l','p',0,'P','s','a','l','t','e','r','_','P','a','h','l','a','v','i',0, +2,'P','h','l','v',0,'P','h','l','v',0,2,'P','r','t','i',0,'I','n','s','c','r','i','p','t','i','o','n','a','l','_','P', +'a','r','t','h','i','a','n',0,2,'S','a','m','r',0,'S','a','m','a','r','i','t','a','n',0, +2,'T','a','v','t',0,'T','a','i','_','V','i','e','t',0,2,'Z','m','t','h',0,'Z','m','t','h',0, +2,'Z','s','y','m',0,'Z','s','y','m',0,2,'B','a','m','u',0,'B','a','m','u','m',0, +2,'N','k','g','b',0,'N','k','g','b',0,2,'S','a','r','b',0,'O','l','d','_','S','o','u','t','h','_','A','r','a','b','i', +'a','n',0,2,'B','a','s','s',0,'B','a','s','s','a','_','V','a','h',0, +2,'D','u','p','l',0,'D','u','p','l','o','y','a','n',0,2,'E','l','b','a',0,'E','l','b','a','s','a','n',0, +2,'G','r','a','n',0,'G','r','a','n','t','h','a',0,2,'K','p','e','l',0,'K','p','e','l',0, +2,'L','o','m','a',0,'L','o','m','a',0,2,'M','e','n','d',0,'M','e','n','d','e','_','K','i','k','a','k','u','i',0, +2,'M','e','r','c',0,'M','e','r','o','i','t','i','c','_','C','u','r','s','i','v','e',0, +2,'N','a','r','b',0,'O','l','d','_','N','o','r','t','h','_','A','r','a','b','i','a','n',0, +2,'N','b','a','t',0,'N','a','b','a','t','a','e','a','n',0, +2,'P','a','l','m',0,'P','a','l','m','y','r','e','n','e',0, +2,'S','i','n','d',0,'K','h','u','d','a','w','a','d','i',0, +2,'W','a','r','a',0,'W','a','r','a','n','g','_','C','i','t','i',0, +2,'A','f','a','k',0,'A','f','a','k',0,2,'J','u','r','c',0,'J','u','r','c',0, +2,'M','r','o','o',0,'M','r','o',0,2,'N','s','h','u',0,'N','u','s','h','u',0, +2,'S','h','r','d',0,'S','h','a','r','a','d','a',0,2,'S','o','r','a',0,'S','o','r','a','_','S','o','m','p','e','n','g', +0,2,'T','a','k','r',0,'T','a','k','r','i',0,2,'T','a','n','g',0,'T','a','n','g','u','t',0, +2,'W','o','l','e',0,'W','o','l','e',0,2,'H','l','u','w',0,'A','n','a','t','o','l','i','a','n','_','H','i','e','r','o', +'g','l','y','p','h','s',0,2,'K','h','o','j',0,'K','h','o','j','k','i',0, +2,'T','i','r','h',0,'T','i','r','h','u','t','a',0,2,'A','g','h','b',0,'C','a','u','c','a','s','i','a','n','_','A','l', +'b','a','n','i','a','n',0,2,'M','a','h','j',0,'M','a','h','a','j','a','n','i',0, +2,'H','a','t','r',0,'H','a','t','r','a','n',0,2,'M','u','l','t',0,'M','u','l','t','a','n','i',0, +2,'P','a','u','c',0,'P','a','u','_','C','i','n','_','H','a','u',0, +2,'S','i','d','d',0,'S','i','d','d','h','a','m',0,2,'A','d','l','m',0,'A','d','l','a','m',0, +2,'B','h','k','s',0,'B','h','a','i','k','s','u','k','i',0, +2,'M','a','r','c',0,'M','a','r','c','h','e','n',0,2,'O','s','g','e',0,'O','s','a','g','e',0, +2,'H','a','n','b',0,'H','a','n','b',0,2,'J','a','m','o',0,'J','a','m','o',0, +2,'Z','s','y','e',0,'Z','s','y','e',0,2,'G','o','n','m',0,'M','a','s','a','r','a','m','_','G','o','n','d','i',0, +2,'S','o','y','o',0,'S','o','y','o','m','b','o',0,2,'Z','a','n','b',0,'Z','a','n','a','b','a','z','a','r','_','S','q', +'u','a','r','e',0,2,'D','o','g','r',0,'D','o','g','r','a',0, +2,'G','o','n','g',0,'G','u','n','j','a','l','a','_','G','o','n','d','i',0, +2,'M','a','k','a',0,'M','a','k','a','s','a','r',0,2,'M','e','d','f',0,'M','e','d','e','f','a','i','d','r','i','n',0, +2,'R','o','h','g',0,'H','a','n','i','f','i','_','R','o','h','i','n','g','y','a',0, +2,'S','o','g','d',0,'S','o','g','d','i','a','n',0,2,'S','o','g','o',0,'O','l','d','_','S','o','g','d','i','a','n',0, +2,'E','l','y','m',0,'E','l','y','m','a','i','c',0,2,'H','m','n','p',0,'N','y','i','a','k','e','n','g','_','P','u','a', +'c','h','u','e','_','H','m','o','n','g',0,2,'N','a','n','d',0,'N','a','n','d','i','n','a','g','a','r','i',0, +2,'W','c','h','o',0,'W','a','n','c','h','o',0,2,'C','h','r','s',0,'C','h','o','r','a','s','m','i','a','n',0, +2,'D','i','a','k',0,'D','i','v','e','s','_','A','k','u','r','u',0, +2,'K','i','t','s',0,'K','h','i','t','a','n','_','S','m','a','l','l','_','S','c','r','i','p','t',0, +2,'Y','e','z','i',0,'Y','e','z','i','d','i',0,2,'C','p','m','n',0,'C','y','p','r','o','_','M','i','n','o','a','n',0, +2,'O','u','g','r',0,'O','l','d','_','U','y','g','h','u','r',0, +2,'T','n','s','a',0,'T','a','n','g','s','a',0,2,'V','i','t','h',0,'V','i','t','h','k','u','q','i',0, +2,'N','a','g','m',0,'N','a','g','_','M','u','n','d','a','r','i',0, +2,'h','s','t',0,'H','a','n','g','u','l','_','S','y','l','l','a','b','l','e','_','T','y','p','e',0, +2,'N','A',0,'N','o','t','_','A','p','p','l','i','c','a','b','l','e',0, +2,'L',0,'L','e','a','d','i','n','g','_','J','a','m','o',0, +2,'V',0,'V','o','w','e','l','_','J','a','m','o',0,2,'T',0,'T','r','a','i','l','i','n','g','_','J','a','m','o',0, +2,'L','V',0,'L','V','_','S','y','l','l','a','b','l','e',0, +2,'L','V','T',0,'L','V','T','_','S','y','l','l','a','b','l','e',0, +2,'N','F','D','_','Q','C',0,'N','F','D','_','Q','u','i','c','k','_','C','h','e','c','k',0, +2,'N',0,'N','o',0,2,'Y',0,'Y','e','s',0,2,'N','F','K','D','_','Q','C',0,'N','F','K','D','_','Q','u','i','c','k', +'_','C','h','e','c','k',0,2,'N','F','C','_','Q','C',0,'N','F','C','_','Q','u','i','c','k','_','C','h','e','c','k',0, +2,'M',0,'M','a','y','b','e',0,2,'N','F','K','C','_','Q','C',0,'N','F','K','C','_','Q','u','i','c','k','_','C','h','e', +'c','k',0,2,'l','c','c','c',0,'L','e','a','d','_','C','a','n','o','n','i','c','a','l','_','C','o','m','b','i','n','i','n', +'g','_','C','l','a','s','s',0,2,'t','c','c','c',0,'T','r','a','i','l','_','C','a','n','o','n','i','c','a','l','_','C','o', +'m','b','i','n','i','n','g','_','C','l','a','s','s',0,2,'G','C','B',0,'G','r','a','p','h','e','m','e','_','C','l','u','s', +'t','e','r','_','B','r','e','a','k',0,2,'X','X',0,'O','t','h','e','r',0, +2,'C','N',0,'C','o','n','t','r','o','l',0,2,'C','R',0,'C','R',0, +2,'E','X',0,'E','x','t','e','n','d',0,2,'L',0,'L',0, +2,'L','F',0,'L','F',0,2,'L','V',0,'L','V',0,2,'L','V','T',0,'L','V','T',0, +2,'T',0,'T',0,2,'V',0,'V',0,2,'S','M',0,'S','p','a','c','i','n','g','M','a','r','k',0, +2,'P','P',0,'P','r','e','p','e','n','d',0,2,'E','B','G',0,'E','_','B','a','s','e','_','G','A','Z',0, +2,'G','A','Z',0,'G','l','u','e','_','A','f','t','e','r','_','Z','w','j',0, +2,'S','B',0,'S','e','n','t','e','n','c','e','_','B','r','e','a','k',0, +2,'A','T',0,'A','T','e','r','m',0,2,'C','L',0,'C','l','o','s','e',0, +2,'F','O',0,'F','o','r','m','a','t',0,2,'L','O',0,'L','o','w','e','r',0, +2,'L','E',0,'O','L','e','t','t','e','r',0,2,'S','E',0,'S','e','p',0, +2,'S','P',0,'S','p',0,2,'S','T',0,'S','T','e','r','m',0, +2,'U','P',0,'U','p','p','e','r',0,2,'S','C',0,'S','C','o','n','t','i','n','u','e',0, +2,'W','B',0,'W','o','r','d','_','B','r','e','a','k',0,2,'L','E',0,'A','L','e','t','t','e','r',0, +2,'K','A',0,'K','a','t','a','k','a','n','a',0,2,'M','L',0,'M','i','d','L','e','t','t','e','r',0, +2,'M','N',0,'M','i','d','N','u','m',0,2,'E','X',0,'E','x','t','e','n','d','N','u','m','L','e','t',0, +2,'E','x','t','e','n','d',0,'E','x','t','e','n','d',0,2,'M','B',0,'M','i','d','N','u','m','L','e','t',0, +2,'N','L',0,'N','e','w','l','i','n','e',0,2,'S','Q',0,'S','i','n','g','l','e','_','Q','u','o','t','e',0, +2,'D','Q',0,'D','o','u','b','l','e','_','Q','u','o','t','e',0, +2,'W','S','e','g','S','p','a','c','e',0,'W','S','e','g','S','p','a','c','e',0, +2,'b','p','t',0,'B','i','d','i','_','P','a','i','r','e','d','_','B','r','a','c','k','e','t','_','T','y','p','e',0, +2,'n',0,'N','o','n','e',0,2,'o',0,'O','p','e','n',0, +2,'c',0,'C','l','o','s','e',0,2,'I','n','P','C',0,'I','n','d','i','c','_','P','o','s','i','t','i','o','n','a','l','_', +'C','a','t','e','g','o','r','y',0,2,'N','A',0,'N','A',0, +2,'B','o','t','t','o','m',0,'B','o','t','t','o','m',0,2,'B','o','t','t','o','m','_','A','n','d','_','L','e','f','t',0, +'B','o','t','t','o','m','_','A','n','d','_','L','e','f','t',0, +2,'B','o','t','t','o','m','_','A','n','d','_','R','i','g','h','t',0,'B','o','t','t','o','m','_','A','n','d','_','R','i','g', +'h','t',0,2,'L','e','f','t',0,'L','e','f','t',0,2,'L','e','f','t','_','A','n','d','_','R','i','g','h','t',0, +'L','e','f','t','_','A','n','d','_','R','i','g','h','t',0,2,'O','v','e','r','s','t','r','u','c','k',0, +'O','v','e','r','s','t','r','u','c','k',0,2,'R','i','g','h','t',0,'R','i','g','h','t',0, +2,'T','o','p',0,'T','o','p',0,2,'T','o','p','_','A','n','d','_','B','o','t','t','o','m',0, +'T','o','p','_','A','n','d','_','B','o','t','t','o','m',0,2,'T','o','p','_','A','n','d','_','B','o','t','t','o','m','_','A', +'n','d','_','R','i','g','h','t',0,'T','o','p','_','A','n','d','_','B','o','t','t','o','m','_','A','n','d','_','R','i','g','h', +'t',0,2,'T','o','p','_','A','n','d','_','L','e','f','t',0,'T','o','p','_','A','n','d','_','L','e','f','t',0, +2,'T','o','p','_','A','n','d','_','L','e','f','t','_','A','n','d','_','R','i','g','h','t',0, +'T','o','p','_','A','n','d','_','L','e','f','t','_','A','n','d','_','R','i','g','h','t',0, +2,'T','o','p','_','A','n','d','_','R','i','g','h','t',0,'T','o','p','_','A','n','d','_','R','i','g','h','t',0, +2,'V','i','s','u','a','l','_','O','r','d','e','r','_','L','e','f','t',0,'V','i','s','u','a','l','_','O','r','d','e','r','_', +'L','e','f','t',0,2,'T','o','p','_','A','n','d','_','B','o','t','t','o','m','_','A','n','d','_','L','e','f','t',0, +'T','o','p','_','A','n','d','_','B','o','t','t','o','m','_','A','n','d','_','L','e','f','t',0, +2,'I','n','S','C',0,'I','n','d','i','c','_','S','y','l','l','a','b','i','c','_','C','a','t','e','g','o','r','y',0, +2,'O','t','h','e','r',0,'O','t','h','e','r',0,2,'A','v','a','g','r','a','h','a',0,'A','v','a','g','r','a','h','a',0, +2,'B','i','n','d','u',0,'B','i','n','d','u',0,2,'B','r','a','h','m','i','_','J','o','i','n','i','n','g','_','N','u','m', +'b','e','r',0,'B','r','a','h','m','i','_','J','o','i','n','i','n','g','_','N','u','m','b','e','r',0, +2,'C','a','n','t','i','l','l','a','t','i','o','n','_','M','a','r','k',0,'C','a','n','t','i','l','l','a','t','i','o','n','_', +'M','a','r','k',0,2,'C','o','n','s','o','n','a','n','t',0,'C','o','n','s','o','n','a','n','t',0, +2,'C','o','n','s','o','n','a','n','t','_','D','e','a','d',0,'C','o','n','s','o','n','a','n','t','_','D','e','a','d',0, +2,'C','o','n','s','o','n','a','n','t','_','F','i','n','a','l',0,'C','o','n','s','o','n','a','n','t','_','F','i','n','a','l', +0,2,'C','o','n','s','o','n','a','n','t','_','H','e','a','d','_','L','e','t','t','e','r',0, +'C','o','n','s','o','n','a','n','t','_','H','e','a','d','_','L','e','t','t','e','r',0, +2,'C','o','n','s','o','n','a','n','t','_','I','n','i','t','i','a','l','_','P','o','s','t','f','i','x','e','d',0, +'C','o','n','s','o','n','a','n','t','_','I','n','i','t','i','a','l','_','P','o','s','t','f','i','x','e','d',0, +2,'C','o','n','s','o','n','a','n','t','_','K','i','l','l','e','r',0,'C','o','n','s','o','n','a','n','t','_','K','i','l','l', +'e','r',0,2,'C','o','n','s','o','n','a','n','t','_','M','e','d','i','a','l',0,'C','o','n','s','o','n','a','n','t','_','M', +'e','d','i','a','l',0,2,'C','o','n','s','o','n','a','n','t','_','P','l','a','c','e','h','o','l','d','e','r',0, +'C','o','n','s','o','n','a','n','t','_','P','l','a','c','e','h','o','l','d','e','r',0, +2,'C','o','n','s','o','n','a','n','t','_','P','r','e','c','e','d','i','n','g','_','R','e','p','h','a',0, +'C','o','n','s','o','n','a','n','t','_','P','r','e','c','e','d','i','n','g','_','R','e','p','h','a',0, +2,'C','o','n','s','o','n','a','n','t','_','P','r','e','f','i','x','e','d',0,'C','o','n','s','o','n','a','n','t','_','P','r', +'e','f','i','x','e','d',0,2,'C','o','n','s','o','n','a','n','t','_','S','u','b','j','o','i','n','e','d',0, +'C','o','n','s','o','n','a','n','t','_','S','u','b','j','o','i','n','e','d',0, +2,'C','o','n','s','o','n','a','n','t','_','S','u','c','c','e','e','d','i','n','g','_','R','e','p','h','a',0, +'C','o','n','s','o','n','a','n','t','_','S','u','c','c','e','e','d','i','n','g','_','R','e','p','h','a',0, +2,'C','o','n','s','o','n','a','n','t','_','W','i','t','h','_','S','t','a','c','k','e','r',0, +'C','o','n','s','o','n','a','n','t','_','W','i','t','h','_','S','t','a','c','k','e','r',0, +2,'G','e','m','i','n','a','t','i','o','n','_','M','a','r','k',0,'G','e','m','i','n','a','t','i','o','n','_','M','a','r','k', +0,2,'I','n','v','i','s','i','b','l','e','_','S','t','a','c','k','e','r',0,'I','n','v','i','s','i','b','l','e','_','S','t', +'a','c','k','e','r',0,2,'J','o','i','n','e','r',0,'J','o','i','n','e','r',0, +2,'M','o','d','i','f','y','i','n','g','_','L','e','t','t','e','r',0,'M','o','d','i','f','y','i','n','g','_','L','e','t','t', +'e','r',0,2,'N','o','n','_','J','o','i','n','e','r',0,'N','o','n','_','J','o','i','n','e','r',0, +2,'N','u','k','t','a',0,'N','u','k','t','a',0,2,'N','u','m','b','e','r',0,'N','u','m','b','e','r',0, +2,'N','u','m','b','e','r','_','J','o','i','n','e','r',0,'N','u','m','b','e','r','_','J','o','i','n','e','r',0, +2,'P','u','r','e','_','K','i','l','l','e','r',0,'P','u','r','e','_','K','i','l','l','e','r',0, +2,'R','e','g','i','s','t','e','r','_','S','h','i','f','t','e','r',0,'R','e','g','i','s','t','e','r','_','S','h','i','f','t', +'e','r',0,2,'S','y','l','l','a','b','l','e','_','M','o','d','i','f','i','e','r',0,'S','y','l','l','a','b','l','e','_','M', +'o','d','i','f','i','e','r',0,2,'T','o','n','e','_','L','e','t','t','e','r',0,'T','o','n','e','_','L','e','t','t','e','r', +0,2,'T','o','n','e','_','M','a','r','k',0,'T','o','n','e','_','M','a','r','k',0, +2,'V','i','r','a','m','a',0,'V','i','r','a','m','a',0,2,'V','i','s','a','r','g','a',0, +'V','i','s','a','r','g','a',0,2,'V','o','w','e','l',0,'V','o','w','e','l',0, +2,'V','o','w','e','l','_','D','e','p','e','n','d','e','n','t',0,'V','o','w','e','l','_','D','e','p','e','n','d','e','n','t', +0,2,'V','o','w','e','l','_','I','n','d','e','p','e','n','d','e','n','t',0,'V','o','w','e','l','_','I','n','d','e','p','e', +'n','d','e','n','t',0,2,'v','o',0,'V','e','r','t','i','c','a','l','_','O','r','i','e','n','t','a','t','i','o','n',0, +2,'R',0,'R','o','t','a','t','e','d',0,2,'T','r',0,'T','r','a','n','s','f','o','r','m','e','d','_','R','o','t','a','t', +'e','d',0,2,'T','u',0,'T','r','a','n','s','f','o','r','m','e','d','_','U','p','r','i','g','h','t',0, +2,'U',0,'U','p','r','i','g','h','t',0,2,'g','c','m',0,'G','e','n','e','r','a','l','_','C','a','t','e','g','o','r','y', +'_','M','a','s','k',0,2,'C',0,'O','t','h','e','r',0,2,'L',0,'L','e','t','t','e','r',0, +2,'L','C',0,'C','a','s','e','d','_','L','e','t','t','e','r',0, +3,'M',0,'M','a','r','k',0,'C','o','m','b','i','n','i','n','g','_','M','a','r','k',0, +2,'N',0,'N','u','m','b','e','r',0,3,'P',0,'P','u','n','c','t','u','a','t','i','o','n',0, +'p','u','n','c','t',0,2,'S',0,'S','y','m','b','o','l',0, +2,'Z',0,'S','e','p','a','r','a','t','o','r',0,2,'n','v',0,'N','u','m','e','r','i','c','_','V','a','l','u','e',0, +2,'a','g','e',0,'A','g','e',0,2,'b','m','g',0,'B','i','d','i','_','M','i','r','r','o','r','i','n','g','_','G','l','y', +'p','h',0,2,'c','f',0,'C','a','s','e','_','F','o','l','d','i','n','g',0, +2,'i','s','c',0,'I','S','O','_','C','o','m','m','e','n','t',0, +2,'l','c',0,'L','o','w','e','r','c','a','s','e','_','M','a','p','p','i','n','g',0, +2,'n','a',0,'N','a','m','e',0,3,'s','c','f',0,'S','i','m','p','l','e','_','C','a','s','e','_','F','o','l','d','i','n', +'g',0,'s','f','c',0,2,'s','l','c',0,'S','i','m','p','l','e','_','L','o','w','e','r','c','a','s','e','_','M','a','p','p', +'i','n','g',0,2,'s','t','c',0,'S','i','m','p','l','e','_','T','i','t','l','e','c','a','s','e','_','M','a','p','p','i','n', +'g',0,2,'s','u','c',0,'S','i','m','p','l','e','_','U','p','p','e','r','c','a','s','e','_','M','a','p','p','i','n','g',0, +2,'t','c',0,'T','i','t','l','e','c','a','s','e','_','M','a','p','p','i','n','g',0, +2,'n','a','1',0,'U','n','i','c','o','d','e','_','1','_','N','a','m','e',0, +2,'u','c',0,'U','p','p','e','r','c','a','s','e','_','M','a','p','p','i','n','g',0, +2,'b','p','b',0,'B','i','d','i','_','P','a','i','r','e','d','_','B','r','a','c','k','e','t',0, +2,'s','c','x',0,'S','c','r','i','p','t','_','E','x','t','e','n','s','i','o','n','s',0 +}; + +U_NAMESPACE_END + +#endif // INCLUDED_FROM_PROPNAME_CPP diff --git a/intl/icu/source/common/propsvec.cpp b/intl/icu/source/common/propsvec.cpp new file mode 100644 index 0000000000..18cc3e8cd8 --- /dev/null +++ b/intl/icu/source/common/propsvec.cpp @@ -0,0 +1,529 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2002-2011, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: propsvec.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002feb22 +* created by: Markus W. Scherer +* +* Store bits (Unicode character properties) in bit set vectors. +*/ + +#include +#include "unicode/utypes.h" +#include "cmemory.h" +#include "utrie.h" +#include "utrie2.h" +#include "uarrsort.h" +#include "propsvec.h" +#include "uassert.h" + +struct UPropsVectors { + uint32_t *v; + int32_t columns; /* number of columns, plus two for start & limit values */ + int32_t maxRows; + int32_t rows; + int32_t prevRow; /* search optimization: remember last row seen */ + UBool isCompacted; +}; + +#define UPVEC_INITIAL_ROWS (1<<12) +#define UPVEC_MEDIUM_ROWS ((int32_t)1<<16) +#define UPVEC_MAX_ROWS (UPVEC_MAX_CP+1) + +U_CAPI UPropsVectors * U_EXPORT2 +upvec_open(int32_t columns, UErrorCode *pErrorCode) { + UPropsVectors *pv; + uint32_t *v, *row; + uint32_t cp; + + if(U_FAILURE(*pErrorCode)) { + return nullptr; + } + if(columns<1) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + columns+=2; /* count range start and limit columns */ + + pv=(UPropsVectors *)uprv_malloc(sizeof(UPropsVectors)); + v=(uint32_t *)uprv_malloc(UPVEC_INITIAL_ROWS*columns*4); + if(pv==nullptr || v==nullptr) { + uprv_free(pv); + uprv_free(v); + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memset(pv, 0, sizeof(UPropsVectors)); + pv->v=v; + pv->columns=columns; + pv->maxRows=UPVEC_INITIAL_ROWS; + pv->rows=2+(UPVEC_MAX_CP-UPVEC_FIRST_SPECIAL_CP); + + /* set the all-Unicode row and the special-value rows */ + row=pv->v; + uprv_memset(row, 0, pv->rows*columns*4); + row[0]=0; + row[1]=0x110000; + row+=columns; + for(cp=UPVEC_FIRST_SPECIAL_CP; cp<=UPVEC_MAX_CP; ++cp) { + row[0]=cp; + row[1]=cp+1; + row+=columns; + } + return pv; +} + +U_CAPI void U_EXPORT2 +upvec_close(UPropsVectors *pv) { + if(pv!=nullptr) { + uprv_free(pv->v); + uprv_free(pv); + } +} + +static uint32_t * +_findRow(UPropsVectors *pv, UChar32 rangeStart) { + uint32_t *row; + int32_t columns, i, start, limit, prevRow; + + columns=pv->columns; + limit=pv->rows; + prevRow=pv->prevRow; + + /* check the vicinity of the last-seen row (start searching with an unrolled loop) */ + row=pv->v+prevRow*columns; + if(rangeStart>=(UChar32)row[0]) { + if(rangeStart<(UChar32)row[1]) { + /* same row as last seen */ + return row; + } else if(rangeStart<(UChar32)(row+=columns)[1]) { + /* next row after the last one */ + pv->prevRow=prevRow+1; + return row; + } else if(rangeStart<(UChar32)(row+=columns)[1]) { + /* second row after the last one */ + pv->prevRow=prevRow+2; + return row; + } else if((rangeStart-(UChar32)row[1])<10) { + /* we are close, continue looping */ + prevRow+=2; + do { + ++prevRow; + row+=columns; + } while(rangeStart>=(UChar32)row[1]); + pv->prevRow=prevRow; + return row; + } + } else if(rangeStart<(UChar32)pv->v[1]) { + /* the very first row */ + pv->prevRow=0; + return pv->v; + } + + /* do a binary search for the start of the range */ + start=0; + while(startv+i*columns; + if(rangeStart<(UChar32)row[0]) { + limit=i; + } else if(rangeStart<(UChar32)row[1]) { + pv->prevRow=i; + return row; + } else { + start=i; + } + } + + /* must be found because all ranges together always cover all of Unicode */ + pv->prevRow=start; + return pv->v+start*columns; +} + +U_CAPI void U_EXPORT2 +upvec_setValue(UPropsVectors *pv, + UChar32 start, UChar32 end, + int32_t column, + uint32_t value, uint32_t mask, + UErrorCode *pErrorCode) { + uint32_t *firstRow, *lastRow; + int32_t columns; + UChar32 limit; + UBool splitFirstRow, splitLastRow; + + /* argument checking */ + if(U_FAILURE(*pErrorCode)) { + return; + } + if( pv==nullptr || + start<0 || start>end || end>UPVEC_MAX_CP || + column<0 || column>=(pv->columns-2) + ) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + if(pv->isCompacted) { + *pErrorCode=U_NO_WRITE_PERMISSION; + return; + } + limit=end+1; + + /* initialize */ + columns=pv->columns; + column+=2; /* skip range start and limit columns */ + value&=mask; + + /* find the rows whose ranges overlap with the input range */ + + /* find the first and last rows, always successful */ + firstRow=_findRow(pv, start); + lastRow=_findRow(pv, end); + + /* + * Rows need to be split if they partially overlap with the + * input range (only possible for the first and last rows) + * and if their value differs from the input value. + */ + splitFirstRow= (UBool)(start!=(UChar32)firstRow[0] && value!=(firstRow[column]&mask)); + splitLastRow= (UBool)(limit!=(UChar32)lastRow[1] && value!=(lastRow[column]&mask)); + + /* split first/last rows if necessary */ + if(splitFirstRow || splitLastRow) { + int32_t count, rows; + + rows=pv->rows; + if((rows+splitFirstRow+splitLastRow)>pv->maxRows) { + uint32_t *newVectors; + int32_t newMaxRows; + + if(pv->maxRowsmaxRowsv, (size_t)rows*columns*4); + firstRow=newVectors+(firstRow-pv->v); + lastRow=newVectors+(lastRow-pv->v); + uprv_free(pv->v); + pv->v=newVectors; + pv->maxRows=newMaxRows; + } + + /* count the number of row cells to move after the last row, and move them */ + count = (int32_t)((pv->v+rows*columns)-(lastRow+columns)); + if(count>0) { + uprv_memmove( + lastRow+(1+splitFirstRow+splitLastRow)*columns, + lastRow+columns, + count*4); + } + pv->rows=rows+splitFirstRow+splitLastRow; + + /* split the first row, and move the firstRow pointer to the second part */ + if(splitFirstRow) { + /* copy all affected rows up one and move the lastRow pointer */ + count = (int32_t)((lastRow-firstRow)+columns); + uprv_memmove(firstRow+columns, firstRow, (size_t)count*4); + lastRow+=columns; + + /* split the range and move the firstRow pointer */ + firstRow[1]=firstRow[columns]=(uint32_t)start; + firstRow+=columns; + } + + /* split the last row */ + if(splitLastRow) { + /* copy the last row data */ + uprv_memcpy(lastRow+columns, lastRow, (size_t)columns*4); + + /* split the range and move the firstRow pointer */ + lastRow[1]=lastRow[columns]=(uint32_t)limit; + } + } + + /* set the "row last seen" to the last row for the range */ + pv->prevRow=(int32_t)((lastRow-(pv->v))/columns); + + /* set the input value in all remaining rows */ + firstRow+=column; + lastRow+=column; + mask=~mask; + for(;;) { + *firstRow=(*firstRow&mask)|value; + if(firstRow==lastRow) { + break; + } + firstRow+=columns; + } +} + +U_CAPI uint32_t U_EXPORT2 +upvec_getValue(const UPropsVectors *pv, UChar32 c, int32_t column) { + uint32_t *row; + UPropsVectors *ncpv; + + if(pv->isCompacted || c<0 || c>UPVEC_MAX_CP || column<0 || column>=(pv->columns-2)) { + return 0; + } + ncpv=(UPropsVectors *)pv; + row=_findRow(ncpv, c); + return row[2+column]; +} + +U_CAPI uint32_t * U_EXPORT2 +upvec_getRow(const UPropsVectors *pv, int32_t rowIndex, + UChar32 *pRangeStart, UChar32 *pRangeEnd) { + uint32_t *row; + int32_t columns; + + if(pv->isCompacted || rowIndex<0 || rowIndex>=pv->rows) { + return nullptr; + } + + columns=pv->columns; + row=pv->v+rowIndex*columns; + if(pRangeStart!=nullptr) { + *pRangeStart=(UChar32)row[0]; + } + if(pRangeEnd!=nullptr) { + *pRangeEnd=(UChar32)row[1]-1; + } + return row+2; +} + +static int32_t U_CALLCONV +upvec_compareRows(const void *context, const void *l, const void *r) { + const uint32_t *left=(const uint32_t *)l, *right=(const uint32_t *)r; + const UPropsVectors *pv=(const UPropsVectors *)context; + int32_t i, count, columns; + + count=columns=pv->columns; /* includes start/limit columns */ + + /* start comparing after start/limit but wrap around to them */ + i=2; + do { + if(left[i]!=right[i]) { + return left[i]0); + + return 0; +} + +U_CAPI void U_EXPORT2 +upvec_compact(UPropsVectors *pv, UPVecCompactHandler *handler, void *context, UErrorCode *pErrorCode) { + uint32_t *row; + int32_t i, columns, valueColumns, rows, count; + UChar32 start, limit; + + /* argument checking */ + if(U_FAILURE(*pErrorCode)) { + return; + } + if(handler==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + if(pv->isCompacted) { + return; + } + + /* Set the flag now: Sorting and compacting destroys the builder data structure. */ + pv->isCompacted=true; + + rows=pv->rows; + columns=pv->columns; + U_ASSERT(columns>=3); /* upvec_open asserts this */ + valueColumns=columns-2; /* not counting start & limit */ + + /* sort the properties vectors to find unique vector values */ + uprv_sortArray(pv->v, rows, columns*4, + upvec_compareRows, pv, false, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + return; + } + + /* + * Find and set the special values. + * This has to do almost the same work as the compaction below, + * to find the indexes where the special-value rows will move. + */ + row=pv->v; + count=-valueColumns; + for(i=0; i=UPVEC_FIRST_SPECIAL_CP) { + handler(context, start, start, count, row+2, valueColumns, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + return; + } + } + + row+=columns; + } + + /* count is at the beginning of the last vector, add valueColumns to include that last vector */ + count+=valueColumns; + + /* Call the handler once more to signal the start of delivering real values. */ + handler(context, UPVEC_START_REAL_VALUES_CP, UPVEC_START_REAL_VALUES_CP, + count, row-valueColumns, valueColumns, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + return; + } + + /* + * Move vector contents up to a contiguous array with only unique + * vector values, and call the handler function for each vector. + * + * This destroys the Properties Vector structure and replaces it + * with an array of just vector values. + */ + row=pv->v; + count=-valueColumns; + for(i=0; iv+count, valueColumns*4)) { + count+=valueColumns; + uprv_memmove(pv->v+count, row+2, (size_t)valueColumns*4); + } + + if(startv+count, valueColumns, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + return; + } + } + + row+=columns; + } + + /* count is at the beginning of the last vector, add one to include that last vector */ + pv->rows=count/valueColumns+1; +} + +U_CAPI const uint32_t * U_EXPORT2 +upvec_getArray(const UPropsVectors *pv, int32_t *pRows, int32_t *pColumns) { + if(!pv->isCompacted) { + return nullptr; + } + if(pRows!=nullptr) { + *pRows=pv->rows; + } + if(pColumns!=nullptr) { + *pColumns=pv->columns-2; + } + return pv->v; +} + +U_CAPI uint32_t * U_EXPORT2 +upvec_cloneArray(const UPropsVectors *pv, + int32_t *pRows, int32_t *pColumns, UErrorCode *pErrorCode) { + uint32_t *clonedArray; + int32_t byteLength; + + if(U_FAILURE(*pErrorCode)) { + return nullptr; + } + if(!pv->isCompacted) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + byteLength=pv->rows*(pv->columns-2)*4; + clonedArray=(uint32_t *)uprv_malloc(byteLength); + if(clonedArray==nullptr) { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memcpy(clonedArray, pv->v, byteLength); + if(pRows!=nullptr) { + *pRows=pv->rows; + } + if(pColumns!=nullptr) { + *pColumns=pv->columns-2; + } + return clonedArray; +} + +U_CAPI UTrie2 * U_EXPORT2 +upvec_compactToUTrie2WithRowIndexes(UPropsVectors *pv, UErrorCode *pErrorCode) { + UPVecToUTrie2Context toUTrie2={ nullptr, 0, 0, 0 }; + upvec_compact(pv, upvec_compactToUTrie2Handler, &toUTrie2, pErrorCode); + utrie2_freeze(toUTrie2.trie, UTRIE2_16_VALUE_BITS, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + utrie2_close(toUTrie2.trie); + toUTrie2.trie=nullptr; + } + return toUTrie2.trie; +} + +/* + * TODO(markus): Add upvec_16BitsToUTrie2() function that enumerates all rows, extracts + * some 16-bit field and builds and returns a UTrie2. + */ + +U_CAPI void U_CALLCONV +upvec_compactToUTrie2Handler(void *context, + UChar32 start, UChar32 end, + int32_t rowIndex, uint32_t *row, int32_t columns, + UErrorCode *pErrorCode) { + (void)row; + (void)columns; + UPVecToUTrie2Context *toUTrie2=(UPVecToUTrie2Context *)context; + if(starttrie, start, end, (uint32_t)rowIndex, true, pErrorCode); + } else { + switch(start) { + case UPVEC_INITIAL_VALUE_CP: + toUTrie2->initialValue=rowIndex; + break; + case UPVEC_ERROR_VALUE_CP: + toUTrie2->errorValue=rowIndex; + break; + case UPVEC_START_REAL_VALUES_CP: + toUTrie2->maxValue=rowIndex; + if(rowIndex>0xffff) { + /* too many rows for a 16-bit trie */ + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + } else { + toUTrie2->trie=utrie2_open(toUTrie2->initialValue, + toUTrie2->errorValue, pErrorCode); + } + break; + default: + break; + } + } +} diff --git a/intl/icu/source/common/propsvec.h b/intl/icu/source/common/propsvec.h new file mode 100644 index 0000000000..39080615ea --- /dev/null +++ b/intl/icu/source/common/propsvec.h @@ -0,0 +1,178 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2002-2010, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: propsvec.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002feb22 +* created by: Markus W. Scherer +* +* Store bits (Unicode character properties) in bit set vectors. +*/ + +#ifndef __UPROPSVEC_H__ +#define __UPROPSVEC_H__ + +#include "unicode/utypes.h" +#include "utrie.h" +#include "utrie2.h" + +U_CDECL_BEGIN + +/** + * Unicode Properties Vectors associated with code point ranges. + * + * Rows of uint32_t integers in a contiguous array store + * the range limits and the properties vectors. + * + * Logically, each row has a certain number of uint32_t values, + * which is set via the upvec_open() "columns" parameter. + * + * Internally, two additional columns are stored. + * In each internal row, + * row[0] contains the start code point and + * row[1] contains the limit code point, + * which is the start of the next range. + * + * Initially, there is only one "normal" row for + * range [0..0x110000[ with values 0. + * There are additional rows for special purposes, see UPVEC_FIRST_SPECIAL_CP. + * + * It would be possible to store only one range boundary per row, + * but self-contained rows allow to later sort them by contents. + */ +struct UPropsVectors; +typedef struct UPropsVectors UPropsVectors; + +/* + * Special pseudo code points for storing the initialValue and the errorValue, + * which are used to initialize a UTrie2 or similar. + */ +#define UPVEC_FIRST_SPECIAL_CP 0x110000 +#define UPVEC_INITIAL_VALUE_CP 0x110000 +#define UPVEC_ERROR_VALUE_CP 0x110001 +#define UPVEC_MAX_CP 0x110001 + +/* + * Special pseudo code point used in upvec_compact() signalling the end of + * delivering special values and the beginning of delivering real ones. + * Stable value, unlike UPVEC_MAX_CP which might grow over time. + */ +#define UPVEC_START_REAL_VALUES_CP 0x200000 + +/* + * Open a UPropsVectors object. + * @param columns Number of value integers (uint32_t) per row. + */ +U_CAPI UPropsVectors * U_EXPORT2 +upvec_open(int32_t columns, UErrorCode *pErrorCode); + +U_CAPI void U_EXPORT2 +upvec_close(UPropsVectors *pv); + +/* + * In rows for code points [start..end], select the column, + * reset the mask bits and set the value bits (ANDed with the mask). + * + * Will set U_NO_WRITE_PERMISSION if called after upvec_compact(). + */ +U_CAPI void U_EXPORT2 +upvec_setValue(UPropsVectors *pv, + UChar32 start, UChar32 end, + int32_t column, + uint32_t value, uint32_t mask, + UErrorCode *pErrorCode); + +/* + * Logically const but must not be used on the same pv concurrently! + * Always returns 0 if called after upvec_compact(). + */ +U_CAPI uint32_t U_EXPORT2 +upvec_getValue(const UPropsVectors *pv, UChar32 c, int32_t column); + +/* + * pRangeStart and pRangeEnd can be NULL. + * @return NULL if rowIndex out of range and for illegal arguments, + * or if called after upvec_compact() + */ +U_CAPI uint32_t * U_EXPORT2 +upvec_getRow(const UPropsVectors *pv, int32_t rowIndex, + UChar32 *pRangeStart, UChar32 *pRangeEnd); + +/* + * Compact the vectors: + * - modify the memory + * - keep only unique vectors + * - store them contiguously from the beginning of the memory + * - for each (non-unique) row, call the handler function + * + * The handler's rowIndex is the index of the row in the compacted + * memory block. + * (Therefore, it starts at 0 increases in increments of the columns value.) + * + * In a first phase, only special values are delivered (each exactly once), + * with start==end both equalling a special pseudo code point. + * Then the handler is called once more with start==end==UPVEC_START_REAL_VALUES_CP + * where rowIndex is the length of the compacted array, + * and the row is arbitrary (but not NULL). + * Then, in the second phase, the handler is called for each row of real values. + */ +typedef void U_CALLCONV +UPVecCompactHandler(void *context, + UChar32 start, UChar32 end, + int32_t rowIndex, uint32_t *row, int32_t columns, + UErrorCode *pErrorCode); + +U_CAPI void U_EXPORT2 +upvec_compact(UPropsVectors *pv, UPVecCompactHandler *handler, void *context, UErrorCode *pErrorCode); + +/* + * Get the vectors array after calling upvec_compact(). + * The caller must not modify nor release the returned array. + * Returns NULL if called before upvec_compact(). + */ +U_CAPI const uint32_t * U_EXPORT2 +upvec_getArray(const UPropsVectors *pv, int32_t *pRows, int32_t *pColumns); + +/* + * Get a clone of the vectors array after calling upvec_compact(). + * The caller owns the returned array and must uprv_free() it. + * Returns NULL if called before upvec_compact(). + */ +U_CAPI uint32_t * U_EXPORT2 +upvec_cloneArray(const UPropsVectors *pv, + int32_t *pRows, int32_t *pColumns, UErrorCode *pErrorCode); + +/* + * Call upvec_compact(), create a 16-bit UTrie2 with indexes into the compacted + * vectors array, and freeze the trie. + */ +U_CAPI UTrie2 * U_EXPORT2 +upvec_compactToUTrie2WithRowIndexes(UPropsVectors *pv, UErrorCode *pErrorCode); + +struct UPVecToUTrie2Context { + UTrie2 *trie; + int32_t initialValue; + int32_t errorValue; + int32_t maxValue; +}; +typedef struct UPVecToUTrie2Context UPVecToUTrie2Context; + +/* context=UPVecToUTrie2Context, creates the trie and stores the rowIndex values */ +U_CAPI void U_CALLCONV +upvec_compactToUTrie2Handler(void *context, + UChar32 start, UChar32 end, + int32_t rowIndex, uint32_t *row, int32_t columns, + UErrorCode *pErrorCode); + +U_CDECL_END + +#endif diff --git a/intl/icu/source/common/punycode.cpp b/intl/icu/source/common/punycode.cpp new file mode 100644 index 0000000000..7ebdebc188 --- /dev/null +++ b/intl/icu/source/common/punycode.cpp @@ -0,0 +1,590 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2002-2011, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: punycode.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002jan31 +* created by: Markus W. Scherer +*/ + + +/* This ICU code derived from: */ +/* +punycode.c 0.4.0 (2001-Nov-17-Sat) +http://www.cs.berkeley.edu/~amc/idn/ +Adam M. Costello +http://www.nicemice.net/amc/ + +Disclaimer and license + + Regarding this entire document or any portion of it (including + the pseudocode and C code), the author makes no guarantees and + is not responsible for any damage resulting from its use. The + author grants irrevocable permission to anyone to use, modify, + and distribute it in any way that does not diminish the rights + of anyone else to use, modify, and distribute it, provided that + redistributed derivative works do not contain misleading author or + version information. Derivative works need not be licensed under + similar terms. +*/ +/* + * ICU modifications: + * - ICU data types and coding conventions + * - ICU string buffer handling with implicit source lengths + * and destination preflighting + * - UTF-16 handling + */ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_IDNA + +#include "unicode/ustring.h" +#include "unicode/utf.h" +#include "unicode/utf16.h" +#include "ustr_imp.h" +#include "cstring.h" +#include "cmemory.h" +#include "punycode.h" +#include "uassert.h" + + +/* Punycode ----------------------------------------------------------------- */ + +/* Punycode parameters for Bootstring */ +#define BASE 36 +#define TMIN 1 +#define TMAX 26 +#define SKEW 38 +#define DAMP 700 +#define INITIAL_BIAS 72 +#define INITIAL_N 0x80 + +/* "Basic" Unicode/ASCII code points */ +#define _HYPHEN 0X2d +#define DELIMITER _HYPHEN + +#define _ZERO_ 0X30 +#define _NINE 0x39 + +#define _SMALL_A 0X61 +#define _SMALL_Z 0X7a + +#define _CAPITAL_A 0X41 +#define _CAPITAL_Z 0X5a + +#define IS_BASIC(c) ((c)<0x80) +#define IS_BASIC_UPPERCASE(c) (_CAPITAL_A<=(c) && (c)<=_CAPITAL_Z) + +/** + * digitToBasic() returns the basic code point whose value + * (when used for representing integers) is d, which must be in the + * range 0 to BASE-1. The lowercase form is used unless the uppercase flag is + * nonzero, in which case the uppercase form is used. + */ +static inline char +digitToBasic(int32_t digit, UBool uppercase) { + /* 0..25 map to ASCII a..z or A..Z */ + /* 26..35 map to ASCII 0..9 */ + if(digit<26) { + if(uppercase) { + return (char)(_CAPITAL_A+digit); + } else { + return (char)(_SMALL_A+digit); + } + } else { + return (char)((_ZERO_-26)+digit); + } +} + +/** + * @return the numeric value of a basic code point (for use in representing integers) + * in the range 0 to BASE-1, or a negative value if cp is invalid. + */ +static int32_t decodeDigit(int32_t cp) { + if(cp<=u'Z') { + if(cp<=u'9') { + if(cp 26..35 + } + } else { + return cp-u'A'; // A-Z -> 0..25 + } + } else if(cp<=u'z') { + return cp-'a'; // a..z -> 0..25 + } else { + return -1; + } +} + +static inline char +asciiCaseMap(char b, UBool uppercase) { + if(uppercase) { + if(_SMALL_A<=b && b<=_SMALL_Z) { + b-=(_SMALL_A-_CAPITAL_A); + } + } else { + if(_CAPITAL_A<=b && b<=_CAPITAL_Z) { + b+=(_SMALL_A-_CAPITAL_A); + } + } + return b; +} + +/* Punycode-specific Bootstring code ---------------------------------------- */ + +/* + * The following code omits the {parts} of the pseudo-algorithm in the spec + * that are not used with the Punycode parameter set. + */ + +/* Bias adaptation function. */ +static int32_t +adaptBias(int32_t delta, int32_t length, UBool firstTime) { + int32_t count; + + if(firstTime) { + delta/=DAMP; + } else { + delta/=2; + } + + delta+=delta/length; + for(count=0; delta>((BASE-TMIN)*TMAX)/2; count+=BASE) { + delta/=(BASE-TMIN); + } + + return count+(((BASE-TMIN+1)*delta)/(delta+SKEW)); +} + +namespace { + +// ICU-13727: Limit input length for n^2 algorithm +// where well-formed strings are at most 59 characters long. +constexpr int32_t ENCODE_MAX_CODE_UNITS=1000; +constexpr int32_t DECODE_MAX_CHARS=2000; + +} // namespace + +// encode +U_CAPI int32_t +u_strToPunycode(const char16_t *src, int32_t srcLength, + char16_t *dest, int32_t destCapacity, + const UBool *caseFlags, + UErrorCode *pErrorCode) { + + int32_t cpBuffer[ENCODE_MAX_CODE_UNITS]; + int32_t n, delta, handledCPCount, basicLength, destLength, bias, j, m, q, k, t, srcCPCount; + char16_t c, c2; + + /* argument checking */ + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + if(src==nullptr || srcLength<-1 || (dest==nullptr && destCapacity!=0)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + if (srcLength>ENCODE_MAX_CODE_UNITS) { + *pErrorCode=U_INPUT_TOO_LONG_ERROR; + return 0; + } + + /* + * Handle the basic code points and + * convert extended ones to UTF-32 in cpBuffer (caseFlag in sign bit): + */ + srcCPCount=destLength=0; + if(srcLength==-1) { + /* NUL-terminated input */ + for(j=0; /* no condition */; ++j) { + if((c=src[j])==0) { + break; + } + if(j>=ENCODE_MAX_CODE_UNITS) { + *pErrorCode=U_INPUT_TOO_LONG_ERROR; + return 0; + } + if(IS_BASIC(c)) { + cpBuffer[srcCPCount++]=0; + if(destLength0) { + if(destLength state to , but guard against overflow: + */ + if(m-n>(0x7fffffff-handledCPCount-delta)/(handledCPCount+1)) { + *pErrorCode=U_INTERNAL_PROGRAM_ERROR; + return 0; + } + delta+=(m-n)*(handledCPCount+1); + n=m; + + /* Encode a sequence of same code points n */ + for(j=0; jTMAX) { + t=TMAX; + } + */ + + t=k-bias; + if(t=(bias+TMAX)) { + t=TMAX; + } + + if(qDECODE_MAX_CHARS) { + *pErrorCode=U_INPUT_TOO_LONG_ERROR; + return 0; + } + + /* + * Handle the basic code points: + * Let basicLength be the number of input code points + * before the last delimiter, or 0 if there is none, + * then copy the first basicLength code points to the output. + * + * The two following loops iterate backward. + */ + for(j=srcLength; j>0;) { + if(src[--j]==DELIMITER) { + break; + } + } + destLength=basicLength=destCPCount=j; + U_ASSERT(destLength>=0); + + while(j>0) { + b=src[--j]; + if(!IS_BASIC(b)) { + *pErrorCode=U_INVALID_CHAR_FOUND; + return 0; + } + + if(j0 ? basicLength+1 : 0; in=srcLength) { + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + return 0; + } + + digit=decodeDigit(src[in++]); + if(digit<0) { + *pErrorCode=U_INVALID_CHAR_FOUND; + return 0; + } + if(digit>(0x7fffffff-i)/w) { + /* integer overflow */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + return 0; + } + + i+=digit*w; + /** RAM: comment out the old code for conformance with draft-ietf-idn-punycode-03.txt + t=k-bias; + if(tTMAX) { + t=TMAX; + } + */ + t=k-bias; + if(t=(bias+TMAX)) { + t=TMAX; + } + if(digit0x7fffffff/(BASE-t)) { + /* integer overflow */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + return 0; + } + w*=BASE-t; + } + + /* + * Modification from sample code: + * Increments destCPCount here, + * where needed instead of in for() loop tail. + */ + ++destCPCount; + bias=adaptBias(i-oldi, destCPCount, (UBool)(oldi==0)); + + /* + * i was supposed to wrap around from (incremented) destCPCount to 0, + * incrementing n each time, so we'll fix that now: + */ + if(i/destCPCount>(0x7fffffff-n)) { + /* integer overflow */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + return 0; + } + + n+=i/destCPCount; + i%=destCPCount; + /* not needed for Punycode: */ + /* if (decode_digit(n) <= BASE) return punycode_invalid_input; */ + + if(n>0x10ffff || U_IS_SURROGATE(n)) { + /* Unicode code point overflow */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + return 0; + } + + /* Insert n at position i of the output: */ + cpLength=U16_LENGTH(n); + if(dest!=nullptr && ((destLength+cpLength)<=destCapacity)) { + int32_t codeUnitIndex; + + /* + * Handle indexes when supplementary code points are present. + * + * In almost all cases, there will be only BMP code points before i + * and even in the entire string. + * This is handled with the same efficiency as with UTF-32. + * + * Only the rare cases with supplementary code points are handled + * more slowly - but not too bad since this is an insertion anyway. + */ + if(i<=firstSupplementaryIndex) { + codeUnitIndex=i; + if(cpLength>1) { + firstSupplementaryIndex=codeUnitIndex; + } else { + ++firstSupplementaryIndex; + } + } else { + codeUnitIndex=firstSupplementaryIndex; + U16_FWD_N(dest, codeUnitIndex, destLength, i-codeUnitIndex); + } + + /* use the char16_t index codeUnitIndex instead of the code point index i */ + if(codeUnitIndex=0); + ++i; + } + + return u_terminateUChars(dest, destCapacity, destLength, pErrorCode); +} + +/* ### check notes on overflow handling - only necessary if not IDNA? are these Punycode functions to be public? */ + +#endif /* #if !UCONFIG_NO_IDNA */ diff --git a/intl/icu/source/common/punycode.h b/intl/icu/source/common/punycode.h new file mode 100644 index 0000000000..9e28f770c4 --- /dev/null +++ b/intl/icu/source/common/punycode.h @@ -0,0 +1,120 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2002-2003, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: punycode.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002jan31 +* created by: Markus W. Scherer +*/ + +/* This ICU code derived from: */ +/* +punycode.c 0.4.0 (2001-Nov-17-Sat) +http://www.cs.berkeley.edu/~amc/idn/ +Adam M. Costello +http://www.nicemice.net/amc/ +*/ + +#ifndef __PUNYCODE_H__ +#define __PUNYCODE_H__ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_IDNA + +/** + * u_strToPunycode() converts Unicode to Punycode. + * + * The input string must not contain single, unpaired surrogates. + * The output will be represented as an array of ASCII code points. + * + * The output string is NUL-terminated according to normal ICU + * string output rules. + * + * @param src Input Unicode string. + * This function handles a limited amount of code points + * (the limit is >=64). + * U_INDEX_OUTOFBOUNDS_ERROR is set if the limit is exceeded. + * @param srcLength Number of UChars in src, or -1 if NUL-terminated. + * @param dest Output Punycode array. + * @param destCapacity Size of dest. + * @param caseFlags Vector of boolean values, one per input UChar, + * indicating that the corresponding character is to be + * marked for the decoder optionally + * uppercasing (true) or lowercasing (false) + * the character. + * ASCII characters are output directly in the case as marked. + * Flags corresponding to trail surrogates are ignored. + * If caseFlags==NULL then input characters are not + * case-mapped. + * @param pErrorCode ICU in/out error code parameter. + * U_INVALID_CHAR_FOUND if src contains + * unmatched single surrogates. + * U_INDEX_OUTOFBOUNDS_ERROR if src contains + * too many code points. + * @return Number of ASCII characters in puny. + * + * @see u_strFromPunycode + */ +U_CAPI int32_t +u_strToPunycode(const UChar *src, int32_t srcLength, + UChar *dest, int32_t destCapacity, + const UBool *caseFlags, + UErrorCode *pErrorCode); + +/** + * u_strFromPunycode() converts Punycode to Unicode. + * The Unicode string will be at most as long (in UChars) + * than the Punycode string (in chars). + * + * @param src Input Punycode string. + * @param srcLength Length of puny, or -1 if NUL-terminated + * @param dest Output Unicode string buffer. + * @param destCapacity Size of dest in number of UChars, + * and of caseFlags in numbers of UBools. + * @param caseFlags Output array for case flags as + * defined by the Punycode string. + * The caller should uppercase (true) or lowercase (FASLE) + * the corresponding character in dest. + * For supplementary characters, only the lead surrogate + * is marked, and false is stored for the trail surrogate. + * This is redundant and not necessary for ASCII characters + * because they are already in the case indicated. + * Can be NULL if the case flags are not needed. + * @param pErrorCode ICU in/out error code parameter. + * U_INVALID_CHAR_FOUND if a non-ASCII character + * precedes the last delimiter ('-'), + * or if an invalid character (not a-zA-Z0-9) is found + * after the last delimiter. + * U_ILLEGAL_CHAR_FOUND if the delta sequence is ill-formed. + * @return Number of UChars written to dest. + * + * @see u_strToPunycode + */ +U_CAPI int32_t +u_strFromPunycode(const UChar *src, int32_t srcLength, + UChar *dest, int32_t destCapacity, + UBool *caseFlags, + UErrorCode *pErrorCode); + +#endif /* #if !UCONFIG_NO_IDNA */ + +#endif + +/* + * Hey, Emacs, please set the following: + * + * Local Variables: + * indent-tabs-mode: nil + * End: + * + */ diff --git a/intl/icu/source/common/putil.cpp b/intl/icu/source/common/putil.cpp new file mode 100644 index 0000000000..ab25f3b996 --- /dev/null +++ b/intl/icu/source/common/putil.cpp @@ -0,0 +1,2506 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1997-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* FILE NAME : putil.c (previously putil.cpp and ptypes.cpp) +* +* Date Name Description +* 04/14/97 aliu Creation. +* 04/24/97 aliu Added getDefaultDataDirectory() and +* getDefaultLocaleID(). +* 04/28/97 aliu Rewritten to assume Unix and apply general methods +* for assumed case. Non-UNIX platforms must be +* special-cased. Rewrote numeric methods dealing +* with NaN and Infinity to be platform independent +* over all IEEE 754 platforms. +* 05/13/97 aliu Restored sign of timezone +* (semantics are hours West of GMT) +* 06/16/98 erm Added IEEE_754 stuff, cleaned up isInfinite, isNan, +* nextDouble.. +* 07/22/98 stephen Added remainder, max, min, trunc +* 08/13/98 stephen Added isNegativeInfinity, isPositiveInfinity +* 08/24/98 stephen Added longBitsFromDouble +* 09/08/98 stephen Minor changes for Mac Port +* 03/02/99 stephen Removed openFile(). Added AS400 support. +* Fixed EBCDIC tables +* 04/15/99 stephen Converted to C. +* 06/28/99 stephen Removed mutex locking in u_isBigEndian(). +* 08/04/99 jeffrey R. Added OS/2 changes +* 11/15/99 helena Integrated S/390 IEEE support. +* 04/26/01 Barry N. OS/400 support for uprv_getDefaultLocaleID +* 08/15/01 Steven H. OS/400 support for uprv_getDefaultCodepage +* 01/03/08 Steven L. Fake Time Support +****************************************************************************** +*/ + +// Defines _XOPEN_SOURCE for access to POSIX functions. +// Must be before any other #includes. +#include "uposixdefs.h" + +// First, the platform type. Need this for U_PLATFORM. +#include "unicode/platform.h" + +#if U_PLATFORM == U_PF_MINGW && defined __STRICT_ANSI__ +/* tzset isn't defined in strict ANSI on MinGW. */ +#undef __STRICT_ANSI__ +#endif + +/* + * Cygwin with GCC requires inclusion of time.h after the above disabling strict asci mode statement. + */ +#include + +#if !U_PLATFORM_USES_ONLY_WIN32_API +#include +#endif + +/* include the rest of the ICU headers */ +#include "unicode/putil.h" +#include "unicode/ustring.h" +#include "putilimp.h" +#include "uassert.h" +#include "umutex.h" +#include "cmemory.h" +#include "cstring.h" +#include "locmap.h" +#include "ucln_cmn.h" +#include "charstr.h" + +/* Include standard headers. */ +#include +#include +#include +#include +#include +#include + +#ifndef U_COMMON_IMPLEMENTATION +#error U_COMMON_IMPLEMENTATION not set - must be set for all ICU source files in common/ - see https://unicode-org.github.io/icu/userguide/howtouseicu +#endif + + +/* include system headers */ +#if U_PLATFORM_USES_ONLY_WIN32_API + /* + * TODO: U_PLATFORM_USES_ONLY_WIN32_API includes MinGW. + * Should Cygwin be included as well (U_PLATFORM_HAS_WIN32_API) + * to use native APIs as much as possible? + */ +#ifndef WIN32_LEAN_AND_MEAN +# define WIN32_LEAN_AND_MEAN +#endif +# define VC_EXTRALEAN +# define NOUSER +# define NOSERVICE +# define NOIME +# define NOMCX +# include +# include "unicode/uloc.h" +# include "wintz.h" +#elif U_PLATFORM == U_PF_OS400 +# include +# include /* error code structure */ +# include +# include /* EPT_CALL macro - this include must be after all other "QSYSINCs" */ +# include /* For uprv_maximumPtr */ +#elif U_PLATFORM == U_PF_OS390 +# include "unicode/ucnv.h" /* Needed for UCNV_SWAP_LFNL_OPTION_STRING */ +#elif U_PLATFORM_IS_DARWIN_BASED || U_PLATFORM_IS_LINUX_BASED || U_PLATFORM == U_PF_BSD || U_PLATFORM == U_PF_SOLARIS +# include +# include +# if U_PLATFORM == U_PF_SOLARIS +# ifndef _XPG4_2 +# define _XPG4_2 +# endif +# elif U_PLATFORM == U_PF_ANDROID +# include +# include +# endif +#elif U_PLATFORM == U_PF_QNX +# include +#endif + + +/* + * Only include langinfo.h if we have a way to get the codeset. If we later + * depend on more feature, we can test on U_HAVE_NL_LANGINFO. + * + */ + +#if U_HAVE_NL_LANGINFO_CODESET +#include +#endif + +/** + * Simple things (presence of functions, etc) should just go in configure.in and be added to + * icucfg.h via autoheader. + */ +#if U_PLATFORM_IMPLEMENTS_POSIX +# if U_PLATFORM == U_PF_OS400 +# define HAVE_DLFCN_H 0 +# define HAVE_DLOPEN 0 +# else +# ifndef HAVE_DLFCN_H +# define HAVE_DLFCN_H 1 +# endif +# ifndef HAVE_DLOPEN +# define HAVE_DLOPEN 1 +# endif +# endif +# ifndef HAVE_GETTIMEOFDAY +# define HAVE_GETTIMEOFDAY 1 +# endif +#else +# define HAVE_DLFCN_H 0 +# define HAVE_DLOPEN 0 +# define HAVE_GETTIMEOFDAY 0 +#endif + +U_NAMESPACE_USE + +/* Define the extension for data files, again... */ +#define DATA_TYPE "dat" + +/* Leave this copyright notice here! */ +static const char copyright[] = U_COPYRIGHT_STRING; + +/* floating point implementations ------------------------------------------- */ + +/* We return QNAN rather than SNAN*/ +#define SIGN 0x80000000U + +/* Make it easy to define certain types of constants */ +typedef union { + int64_t i64; /* This must be defined first in order to allow the initialization to work. This is a C89 feature. */ + double d64; +} BitPatternConversion; +static const BitPatternConversion gNan = { (int64_t) INT64_C(0x7FF8000000000000) }; +static const BitPatternConversion gInf = { (int64_t) INT64_C(0x7FF0000000000000) }; + +/*--------------------------------------------------------------------------- + Platform utilities + Our general strategy is to assume we're on a POSIX platform. Platforms which + are non-POSIX must declare themselves so. The default POSIX implementation + will sometimes work for non-POSIX platforms as well (e.g., the NaN-related + functions). + ---------------------------------------------------------------------------*/ + +#if U_PLATFORM_USES_ONLY_WIN32_API || U_PLATFORM == U_PF_OS400 +# undef U_POSIX_LOCALE +#else +# define U_POSIX_LOCALE 1 +#endif + +/* + WARNING! u_topNBytesOfDouble and u_bottomNBytesOfDouble + can't be properly optimized by the gcc compiler sometimes (i.e. gcc 3.2). +*/ +#if !IEEE_754 +static char* +u_topNBytesOfDouble(double* d, int n) +{ +#if U_IS_BIG_ENDIAN + return (char*)d; +#else + return (char*)(d + 1) - n; +#endif +} + +static char* +u_bottomNBytesOfDouble(double* d, int n) +{ +#if U_IS_BIG_ENDIAN + return (char*)(d + 1) - n; +#else + return (char*)d; +#endif +} +#endif /* !IEEE_754 */ + +#if IEEE_754 +static UBool +u_signBit(double d) { + uint8_t hiByte; +#if U_IS_BIG_ENDIAN + hiByte = *(uint8_t *)&d; +#else + hiByte = *(((uint8_t *)&d) + sizeof(double) - 1); +#endif + return (hiByte & 0x80) != 0; +} +#endif + + + +#if defined (U_DEBUG_FAKETIME) +/* Override the clock to test things without having to move the system clock. + * Assumes POSIX gettimeofday() will function + */ +UDate fakeClock_t0 = 0; /** Time to start the clock from **/ +UDate fakeClock_dt = 0; /** Offset (fake time - real time) **/ +UBool fakeClock_set = false; /** True if fake clock has spun up **/ + +static UDate getUTCtime_real() { + struct timeval posixTime; + gettimeofday(&posixTime, nullptr); + return (UDate)(((int64_t)posixTime.tv_sec * U_MILLIS_PER_SECOND) + (posixTime.tv_usec/1000)); +} + +static UDate getUTCtime_fake() { + static UMutex fakeClockMutex; + umtx_lock(&fakeClockMutex); + if(!fakeClock_set) { + UDate real = getUTCtime_real(); + const char *fake_start = getenv("U_FAKETIME_START"); + if((fake_start!=nullptr) && (fake_start[0]!=0)) { + sscanf(fake_start,"%lf",&fakeClock_t0); + fakeClock_dt = fakeClock_t0 - real; + fprintf(stderr,"U_DEBUG_FAKETIME was set at compile time, so the ICU clock will start at a preset value\n" + "env variable U_FAKETIME_START=%.0f (%s) for an offset of %.0f ms from the current time %.0f\n", + fakeClock_t0, fake_start, fakeClock_dt, real); + } else { + fakeClock_dt = 0; + fprintf(stderr,"U_DEBUG_FAKETIME was set at compile time, but U_FAKETIME_START was not set.\n" + "Set U_FAKETIME_START to the number of milliseconds since 1/1/1970 to set the ICU clock.\n"); + } + fakeClock_set = true; + } + umtx_unlock(&fakeClockMutex); + + return getUTCtime_real() + fakeClock_dt; +} +#endif + +#if U_PLATFORM_USES_ONLY_WIN32_API +typedef union { + int64_t int64; + FILETIME fileTime; +} FileTimeConversion; /* This is like a ULARGE_INTEGER */ + +/* Number of 100 nanoseconds from 1/1/1601 to 1/1/1970 */ +#define EPOCH_BIAS INT64_C(116444736000000000) +#define HECTONANOSECOND_PER_MILLISECOND 10000 + +#endif + +/*--------------------------------------------------------------------------- + Universal Implementations + These are designed to work on all platforms. Try these, and if they + don't work on your platform, then special case your platform with new + implementations. +---------------------------------------------------------------------------*/ + +U_CAPI UDate U_EXPORT2 +uprv_getUTCtime() +{ +#if defined(U_DEBUG_FAKETIME) + return getUTCtime_fake(); /* Hook for overriding the clock */ +#else + return uprv_getRawUTCtime(); +#endif +} + +/* Return UTC (GMT) time measured in milliseconds since 0:00 on 1/1/70.*/ +U_CAPI UDate U_EXPORT2 +uprv_getRawUTCtime() +{ +#if U_PLATFORM_USES_ONLY_WIN32_API + + FileTimeConversion winTime; + GetSystemTimeAsFileTime(&winTime.fileTime); + return (UDate)((winTime.int64 - EPOCH_BIAS) / HECTONANOSECOND_PER_MILLISECOND); +#else + +#if HAVE_GETTIMEOFDAY + struct timeval posixTime; + gettimeofday(&posixTime, nullptr); + return (UDate)(((int64_t)posixTime.tv_sec * U_MILLIS_PER_SECOND) + (posixTime.tv_usec/1000)); +#else + time_t epochtime; + time(&epochtime); + return (UDate)epochtime * U_MILLIS_PER_SECOND; +#endif + +#endif +} + +/*----------------------------------------------------------------------------- + IEEE 754 + These methods detect and return NaN and infinity values for doubles + conforming to IEEE 754. Platforms which support this standard include X86, + Mac 680x0, Mac PowerPC, AIX RS/6000, and most others. + If this doesn't work on your platform, you have non-IEEE floating-point, and + will need to code your own versions. A naive implementation is to return 0.0 + for getNaN and getInfinity, and false for isNaN and isInfinite. + ---------------------------------------------------------------------------*/ + +U_CAPI UBool U_EXPORT2 +uprv_isNaN(double number) +{ +#if IEEE_754 + BitPatternConversion convertedNumber; + convertedNumber.d64 = number; + /* Infinity is 0x7FF0000000000000U. Anything greater than that is a NaN */ + return (UBool)((convertedNumber.i64 & U_INT64_MAX) > gInf.i64); + +#elif U_PLATFORM == U_PF_OS390 + uint32_t highBits = *(uint32_t*)u_topNBytesOfDouble(&number, + sizeof(uint32_t)); + uint32_t lowBits = *(uint32_t*)u_bottomNBytesOfDouble(&number, + sizeof(uint32_t)); + + return ((highBits & 0x7F080000L) == 0x7F080000L) && + (lowBits == 0x00000000L); + +#else + /* If your platform doesn't support IEEE 754 but *does* have an NaN value,*/ + /* you'll need to replace this default implementation with what's correct*/ + /* for your platform.*/ + return number != number; +#endif +} + +U_CAPI UBool U_EXPORT2 +uprv_isInfinite(double number) +{ +#if IEEE_754 + BitPatternConversion convertedNumber; + convertedNumber.d64 = number; + /* Infinity is exactly 0x7FF0000000000000U. */ + return (UBool)((convertedNumber.i64 & U_INT64_MAX) == gInf.i64); +#elif U_PLATFORM == U_PF_OS390 + uint32_t highBits = *(uint32_t*)u_topNBytesOfDouble(&number, + sizeof(uint32_t)); + uint32_t lowBits = *(uint32_t*)u_bottomNBytesOfDouble(&number, + sizeof(uint32_t)); + + return ((highBits & ~SIGN) == 0x70FF0000L) && (lowBits == 0x00000000L); + +#else + /* If your platform doesn't support IEEE 754 but *does* have an infinity*/ + /* value, you'll need to replace this default implementation with what's*/ + /* correct for your platform.*/ + return number == (2.0 * number); +#endif +} + +U_CAPI UBool U_EXPORT2 +uprv_isPositiveInfinity(double number) +{ +#if IEEE_754 || U_PLATFORM == U_PF_OS390 + return (UBool)(number > 0 && uprv_isInfinite(number)); +#else + return uprv_isInfinite(number); +#endif +} + +U_CAPI UBool U_EXPORT2 +uprv_isNegativeInfinity(double number) +{ +#if IEEE_754 || U_PLATFORM == U_PF_OS390 + return (UBool)(number < 0 && uprv_isInfinite(number)); + +#else + uint32_t highBits = *(uint32_t*)u_topNBytesOfDouble(&number, + sizeof(uint32_t)); + return((highBits & SIGN) && uprv_isInfinite(number)); + +#endif +} + +U_CAPI double U_EXPORT2 +uprv_getNaN() +{ +#if IEEE_754 || U_PLATFORM == U_PF_OS390 + return gNan.d64; +#else + /* If your platform doesn't support IEEE 754 but *does* have an NaN value,*/ + /* you'll need to replace this default implementation with what's correct*/ + /* for your platform.*/ + return 0.0; +#endif +} + +U_CAPI double U_EXPORT2 +uprv_getInfinity() +{ +#if IEEE_754 || U_PLATFORM == U_PF_OS390 + return gInf.d64; +#else + /* If your platform doesn't support IEEE 754 but *does* have an infinity*/ + /* value, you'll need to replace this default implementation with what's*/ + /* correct for your platform.*/ + return 0.0; +#endif +} + +U_CAPI double U_EXPORT2 +uprv_floor(double x) +{ + return floor(x); +} + +U_CAPI double U_EXPORT2 +uprv_ceil(double x) +{ + return ceil(x); +} + +U_CAPI double U_EXPORT2 +uprv_round(double x) +{ + return uprv_floor(x + 0.5); +} + +U_CAPI double U_EXPORT2 +uprv_fabs(double x) +{ + return fabs(x); +} + +U_CAPI double U_EXPORT2 +uprv_modf(double x, double* y) +{ + return modf(x, y); +} + +U_CAPI double U_EXPORT2 +uprv_fmod(double x, double y) +{ + return fmod(x, y); +} + +U_CAPI double U_EXPORT2 +uprv_pow(double x, double y) +{ + /* This is declared as "double pow(double x, double y)" */ + return pow(x, y); +} + +U_CAPI double U_EXPORT2 +uprv_pow10(int32_t x) +{ + return pow(10.0, (double)x); +} + +U_CAPI double U_EXPORT2 +uprv_fmax(double x, double y) +{ +#if IEEE_754 + /* first handle NaN*/ + if(uprv_isNaN(x) || uprv_isNaN(y)) + return uprv_getNaN(); + + /* check for -0 and 0*/ + if(x == 0.0 && y == 0.0 && u_signBit(x)) + return y; + +#endif + + /* this should work for all flt point w/o NaN and Inf special cases */ + return (x > y ? x : y); +} + +U_CAPI double U_EXPORT2 +uprv_fmin(double x, double y) +{ +#if IEEE_754 + /* first handle NaN*/ + if(uprv_isNaN(x) || uprv_isNaN(y)) + return uprv_getNaN(); + + /* check for -0 and 0*/ + if(x == 0.0 && y == 0.0 && u_signBit(y)) + return y; + +#endif + + /* this should work for all flt point w/o NaN and Inf special cases */ + return (x > y ? y : x); +} + +U_CAPI UBool U_EXPORT2 +uprv_add32_overflow(int32_t a, int32_t b, int32_t* res) { + // NOTE: Some compilers (GCC, Clang) have primitives available, like __builtin_add_overflow. + // This function could be optimized by calling one of those primitives. + auto a64 = static_cast(a); + auto b64 = static_cast(b); + int64_t res64 = a64 + b64; + *res = static_cast(res64); + return res64 != *res; +} + +U_CAPI UBool U_EXPORT2 +uprv_mul32_overflow(int32_t a, int32_t b, int32_t* res) { + // NOTE: Some compilers (GCC, Clang) have primitives available, like __builtin_mul_overflow. + // This function could be optimized by calling one of those primitives. + auto a64 = static_cast(a); + auto b64 = static_cast(b); + int64_t res64 = a64 * b64; + *res = static_cast(res64); + return res64 != *res; +} + +/** + * Truncates the given double. + * trunc(3.3) = 3.0, trunc (-3.3) = -3.0 + * This is different than calling floor() or ceil(): + * floor(3.3) = 3, floor(-3.3) = -4 + * ceil(3.3) = 4, ceil(-3.3) = -3 + */ +U_CAPI double U_EXPORT2 +uprv_trunc(double d) +{ +#if IEEE_754 + /* handle error cases*/ + if(uprv_isNaN(d)) + return uprv_getNaN(); + if(uprv_isInfinite(d)) + return uprv_getInfinity(); + + if(u_signBit(d)) /* Signbit() picks up -0.0; d<0 does not. */ + return ceil(d); + else + return floor(d); + +#else + return d >= 0 ? floor(d) : ceil(d); + +#endif +} + +/** + * Return the largest positive number that can be represented by an integer + * type of arbitrary bit length. + */ +U_CAPI double U_EXPORT2 +uprv_maxMantissa() +{ + return pow(2.0, DBL_MANT_DIG + 1.0) - 1.0; +} + +U_CAPI double U_EXPORT2 +uprv_log(double d) +{ + return log(d); +} + +U_CAPI void * U_EXPORT2 +uprv_maximumPtr(void * base) +{ +#if U_PLATFORM == U_PF_OS400 + /* + * With the provided function we should never be out of range of a given segment + * (a traditional/typical segment that is). Our segments have 5 bytes for the + * id and 3 bytes for the offset. The key is that the casting takes care of + * only retrieving the offset portion minus x1000. Hence, the smallest offset + * seen in a program is x001000 and when casted to an int would be 0. + * That's why we can only add 0xffefff. Otherwise, we would exceed the segment. + * + * Currently, 16MB is the current addressing limitation on i5/OS if the activation is + * non-TERASPACE. If it is TERASPACE it is 2GB - 4k(header information). + * This function determines the activation based on the pointer that is passed in and + * calculates the appropriate maximum available size for + * each pointer type (TERASPACE and non-TERASPACE) + * + * Unlike other operating systems, the pointer model isn't determined at + * compile time on i5/OS. + */ + if ((base != nullptr) && (_TESTPTR(base, _C_TERASPACE_CHECK))) { + /* if it is a TERASPACE pointer the max is 2GB - 4k */ + return ((void *)(((char *)base)-((uint32_t)(base))+((uint32_t)0x7fffefff))); + } + /* otherwise 16MB since nullptr ptr is not checkable or the ptr is not TERASPACE */ + return ((void *)(((char *)base)-((uint32_t)(base))+((uint32_t)0xffefff))); + +#else + return U_MAX_PTR(base); +#endif +} + +/*--------------------------------------------------------------------------- + Platform-specific Implementations + Try these, and if they don't work on your platform, then special case your + platform with new implementations. + ---------------------------------------------------------------------------*/ + +/* Generic time zone layer -------------------------------------------------- */ + +/* Time zone utilities */ +U_CAPI void U_EXPORT2 +uprv_tzset() +{ +#if defined(U_TZSET) + U_TZSET(); +#else + /* no initialization*/ +#endif +} + +U_CAPI int32_t U_EXPORT2 +uprv_timezone() +{ +#ifdef U_TIMEZONE + return U_TIMEZONE; +#else + time_t t, t1, t2; + struct tm tmrec; + int32_t tdiff = 0; + + time(&t); + uprv_memcpy( &tmrec, localtime(&t), sizeof(tmrec) ); +#if U_PLATFORM != U_PF_IPHONE + UBool dst_checked = (tmrec.tm_isdst != 0); /* daylight savings time is checked*/ +#endif + t1 = mktime(&tmrec); /* local time in seconds*/ + uprv_memcpy( &tmrec, gmtime(&t), sizeof(tmrec) ); + t2 = mktime(&tmrec); /* GMT (or UTC) in seconds*/ + tdiff = t2 - t1; + +#if U_PLATFORM != U_PF_IPHONE + /* imitate NT behaviour, which returns same timezone offset to GMT for + winter and summer. + This does not work on all platforms. For instance, on glibc on Linux + and on Mac OS 10.5, tdiff calculated above remains the same + regardless of whether DST is in effect or not. iOS is another + platform where this does not work. Linux + glibc and Mac OS 10.5 + have U_TIMEZONE defined so that this code is not reached. + */ + if (dst_checked) + tdiff += 3600; +#endif + return tdiff; +#endif +} + +/* Note that U_TZNAME does *not* have to be tzname, but if it is, + some platforms need to have it declared here. */ + +#if defined(U_TZNAME) && (U_PLATFORM == U_PF_IRIX || U_PLATFORM_IS_DARWIN_BASED) +/* RS6000 and others reject char **tzname. */ +extern U_IMPORT char *U_TZNAME[]; +#endif + +#if !UCONFIG_NO_FILE_IO && ((U_PLATFORM_IS_DARWIN_BASED && (U_PLATFORM != U_PF_IPHONE || defined(U_TIMEZONE))) || U_PLATFORM_IS_LINUX_BASED || U_PLATFORM == U_PF_BSD || U_PLATFORM == U_PF_SOLARIS) +/* These platforms are likely to use Olson timezone IDs. */ +/* common targets of the symbolic link at TZDEFAULT are: + * "/usr/share/zoneinfo/" default, older Linux distros, macOS to 10.12 + * "../usr/share/zoneinfo/" newer Linux distros: Red Hat Enterprise Linux 7, Ubuntu 16, SuSe Linux 12 + * "/usr/share/lib/zoneinfo/" Solaris + * "../usr/share/lib/zoneinfo/" Solaris + * "/var/db/timezone/zoneinfo/" macOS 10.13 + * To avoid checking lots of paths, just check that the target path + * before the ends with "/zoneinfo/", and the is valid. + */ + +#define CHECK_LOCALTIME_LINK 1 +#if U_PLATFORM_IS_DARWIN_BASED +#include +#define TZZONEINFO (TZDIR "/") +#elif U_PLATFORM == U_PF_SOLARIS +#define TZDEFAULT "/etc/localtime" +#define TZZONEINFO "/usr/share/lib/zoneinfo/" +#define TZ_ENV_CHECK "localtime" +#else +#define TZDEFAULT "/etc/localtime" +#define TZZONEINFO "/usr/share/zoneinfo/" +#endif +#define TZZONEINFOTAIL "/zoneinfo/" +#if U_HAVE_DIRENT_H +#define TZFILE_SKIP "posixrules" /* tz file to skip when searching. */ +/* Some Linux distributions have 'localtime' in /usr/share/zoneinfo + symlinked to /etc/localtime, which makes searchForTZFile return + 'localtime' when it's the first match. */ +#define TZFILE_SKIP2 "localtime" +#define SEARCH_TZFILE +#include /* Needed to search through system timezone files */ +#endif +static char gTimeZoneBuffer[PATH_MAX]; +static const char *gTimeZoneBufferPtr = nullptr; +#endif + +#if !U_PLATFORM_USES_ONLY_WIN32_API +#define isNonDigit(ch) (ch < '0' || '9' < ch) +#define isDigit(ch) ('0' <= ch && ch <= '9') +static UBool isValidOlsonID(const char *id) { + int32_t idx = 0; + int32_t idxMax = 0; + + /* Determine if this is something like Iceland (Olson ID) + or AST4ADT (non-Olson ID) */ + while (id[idx] && isNonDigit(id[idx]) && id[idx] != ',') { + idx++; + } + + /* Allow at maximum 2 numbers at the end of the id to support zone id's + like GMT+11. */ + idxMax = idx + 2; + while (id[idx] && isDigit(id[idx]) && idx < idxMax) { + idx++; + } + + /* If we went through the whole string, then it might be okay. + The timezone is sometimes set to "CST-7CDT", "CST6CDT5,J129,J131/19:30", + "GRNLNDST3GRNLNDDT" or similar, so we cannot use it. + The rest of the time it could be an Olson ID. George */ + return (UBool)(id[idx] == 0 + || uprv_strcmp(id, "PST8PDT") == 0 + || uprv_strcmp(id, "MST7MDT") == 0 + || uprv_strcmp(id, "CST6CDT") == 0 + || uprv_strcmp(id, "EST5EDT") == 0); +} + +/* On some Unix-like OS, 'posix' subdirectory in + /usr/share/zoneinfo replicates the top-level contents. 'right' + subdirectory has the same set of files, but individual files + are different from those in the top-level directory or 'posix' + because 'right' has files for TAI (Int'l Atomic Time) while 'posix' + has files for UTC. + When the first match for /etc/localtime is in either of them + (usually in posix because 'right' has different file contents), + or TZ environment variable points to one of them, createTimeZone + fails because, say, 'posix/America/New_York' is not an Olson + timezone id ('America/New_York' is). So, we have to skip + 'posix/' and 'right/' at the beginning. */ +static void skipZoneIDPrefix(const char** id) { + if (uprv_strncmp(*id, "posix/", 6) == 0 + || uprv_strncmp(*id, "right/", 6) == 0) + { + *id += 6; + } +} +#endif + +#if defined(U_TZNAME) && !U_PLATFORM_USES_ONLY_WIN32_API + +#define CONVERT_HOURS_TO_SECONDS(offset) (int32_t)(offset*3600) +typedef struct OffsetZoneMapping { + int32_t offsetSeconds; + int32_t daylightType; /* 0=U_DAYLIGHT_NONE, 1=daylight in June-U_DAYLIGHT_JUNE, 2=daylight in December=U_DAYLIGHT_DECEMBER*/ + const char *stdID; + const char *dstID; + const char *olsonID; +} OffsetZoneMapping; + +enum { U_DAYLIGHT_NONE=0,U_DAYLIGHT_JUNE=1,U_DAYLIGHT_DECEMBER=2 }; + +/* +This list tries to disambiguate a set of abbreviated timezone IDs and offsets +and maps it to an Olson ID. +Before adding anything to this list, take a look at +icu/source/tools/tzcode/tz.alias +Sometimes no daylight savings (0) is important to define due to aliases. +This list can be tested with icu/source/test/compat/tzone.pl +More values could be added to daylightType to increase precision. +*/ +static const struct OffsetZoneMapping OFFSET_ZONE_MAPPINGS[] = { + {-45900, 2, "CHAST", "CHADT", "Pacific/Chatham"}, + {-43200, 1, "PETT", "PETST", "Asia/Kamchatka"}, + {-43200, 2, "NZST", "NZDT", "Pacific/Auckland"}, + {-43200, 1, "ANAT", "ANAST", "Asia/Anadyr"}, + {-39600, 1, "MAGT", "MAGST", "Asia/Magadan"}, + {-37800, 2, "LHST", "LHST", "Australia/Lord_Howe"}, + {-36000, 2, "EST", "EST", "Australia/Sydney"}, + {-36000, 1, "SAKT", "SAKST", "Asia/Sakhalin"}, + {-36000, 1, "VLAT", "VLAST", "Asia/Vladivostok"}, + {-34200, 2, "CST", "CST", "Australia/South"}, + {-32400, 1, "YAKT", "YAKST", "Asia/Yakutsk"}, + {-32400, 1, "CHOT", "CHOST", "Asia/Choibalsan"}, + {-31500, 2, "CWST", "CWST", "Australia/Eucla"}, + {-28800, 1, "IRKT", "IRKST", "Asia/Irkutsk"}, + {-28800, 1, "ULAT", "ULAST", "Asia/Ulaanbaatar"}, + {-28800, 2, "WST", "WST", "Australia/West"}, + {-25200, 1, "HOVT", "HOVST", "Asia/Hovd"}, + {-25200, 1, "KRAT", "KRAST", "Asia/Krasnoyarsk"}, + {-21600, 1, "NOVT", "NOVST", "Asia/Novosibirsk"}, + {-21600, 1, "OMST", "OMSST", "Asia/Omsk"}, + {-18000, 1, "YEKT", "YEKST", "Asia/Yekaterinburg"}, + {-14400, 1, "SAMT", "SAMST", "Europe/Samara"}, + {-14400, 1, "AMT", "AMST", "Asia/Yerevan"}, + {-14400, 1, "AZT", "AZST", "Asia/Baku"}, + {-10800, 1, "AST", "ADT", "Asia/Baghdad"}, + {-10800, 1, "MSK", "MSD", "Europe/Moscow"}, + {-10800, 1, "VOLT", "VOLST", "Europe/Volgograd"}, + {-7200, 0, "EET", "CEST", "Africa/Tripoli"}, + {-7200, 1, "EET", "EEST", "Europe/Athens"}, /* Conflicts with Africa/Cairo */ + {-7200, 1, "IST", "IDT", "Asia/Jerusalem"}, + {-3600, 0, "CET", "WEST", "Africa/Algiers"}, + {-3600, 2, "WAT", "WAST", "Africa/Windhoek"}, + {0, 1, "GMT", "IST", "Europe/Dublin"}, + {0, 1, "GMT", "BST", "Europe/London"}, + {0, 0, "WET", "WEST", "Africa/Casablanca"}, + {0, 0, "WET", "WET", "Africa/El_Aaiun"}, + {3600, 1, "AZOT", "AZOST", "Atlantic/Azores"}, + {3600, 1, "EGT", "EGST", "America/Scoresbysund"}, + {10800, 1, "PMST", "PMDT", "America/Miquelon"}, + {10800, 2, "UYT", "UYST", "America/Montevideo"}, + {10800, 1, "WGT", "WGST", "America/Godthab"}, + {10800, 2, "BRT", "BRST", "Brazil/East"}, + {12600, 1, "NST", "NDT", "America/St_Johns"}, + {14400, 1, "AST", "ADT", "Canada/Atlantic"}, + {14400, 2, "AMT", "AMST", "America/Cuiaba"}, + {14400, 2, "CLT", "CLST", "Chile/Continental"}, + {14400, 2, "FKT", "FKST", "Atlantic/Stanley"}, + {14400, 2, "PYT", "PYST", "America/Asuncion"}, + {18000, 1, "CST", "CDT", "America/Havana"}, + {18000, 1, "EST", "EDT", "US/Eastern"}, /* Conflicts with America/Grand_Turk */ + {21600, 2, "EAST", "EASST", "Chile/EasterIsland"}, + {21600, 0, "CST", "MDT", "Canada/Saskatchewan"}, + {21600, 0, "CST", "CDT", "America/Guatemala"}, + {21600, 1, "CST", "CDT", "US/Central"}, /* Conflicts with Mexico/General */ + {25200, 1, "MST", "MDT", "US/Mountain"}, /* Conflicts with Mexico/BajaSur */ + {28800, 0, "PST", "PST", "Pacific/Pitcairn"}, + {28800, 1, "PST", "PDT", "US/Pacific"}, /* Conflicts with Mexico/BajaNorte */ + {32400, 1, "AKST", "AKDT", "US/Alaska"}, + {36000, 1, "HAST", "HADT", "US/Aleutian"} +}; + +/*#define DEBUG_TZNAME*/ + +static const char* remapShortTimeZone(const char *stdID, const char *dstID, int32_t daylightType, int32_t offset) +{ + int32_t idx; +#ifdef DEBUG_TZNAME + fprintf(stderr, "TZ=%s std=%s dst=%s daylight=%d offset=%d\n", getenv("TZ"), stdID, dstID, daylightType, offset); +#endif + for (idx = 0; idx < UPRV_LENGTHOF(OFFSET_ZONE_MAPPINGS); idx++) + { + if (offset == OFFSET_ZONE_MAPPINGS[idx].offsetSeconds + && daylightType == OFFSET_ZONE_MAPPINGS[idx].daylightType + && strcmp(OFFSET_ZONE_MAPPINGS[idx].stdID, stdID) == 0 + && strcmp(OFFSET_ZONE_MAPPINGS[idx].dstID, dstID) == 0) + { + return OFFSET_ZONE_MAPPINGS[idx].olsonID; + } + } + return nullptr; +} +#endif + +#ifdef SEARCH_TZFILE +#define MAX_READ_SIZE 512 + +typedef struct DefaultTZInfo { + char* defaultTZBuffer; + int64_t defaultTZFileSize; + FILE* defaultTZFilePtr; + UBool defaultTZstatus; + int32_t defaultTZPosition; +} DefaultTZInfo; + +/* + * This method compares the two files given to see if they are a match. + * It is currently use to compare two TZ files. + */ +static UBool compareBinaryFiles(const char* defaultTZFileName, const char* TZFileName, DefaultTZInfo* tzInfo) { + FILE* file; + int64_t sizeFile; + int64_t sizeFileLeft; + int32_t sizeFileRead; + int32_t sizeFileToRead; + char bufferFile[MAX_READ_SIZE]; + UBool result = true; + + if (tzInfo->defaultTZFilePtr == nullptr) { + tzInfo->defaultTZFilePtr = fopen(defaultTZFileName, "r"); + } + file = fopen(TZFileName, "r"); + + tzInfo->defaultTZPosition = 0; /* reset position to begin search */ + + if (file != nullptr && tzInfo->defaultTZFilePtr != nullptr) { + /* First check that the file size are equal. */ + if (tzInfo->defaultTZFileSize == 0) { + fseek(tzInfo->defaultTZFilePtr, 0, SEEK_END); + tzInfo->defaultTZFileSize = ftell(tzInfo->defaultTZFilePtr); + } + fseek(file, 0, SEEK_END); + sizeFile = ftell(file); + sizeFileLeft = sizeFile; + + if (sizeFile != tzInfo->defaultTZFileSize) { + result = false; + } else { + /* Store the data from the files in separate buffers and + * compare each byte to determine equality. + */ + if (tzInfo->defaultTZBuffer == nullptr) { + rewind(tzInfo->defaultTZFilePtr); + tzInfo->defaultTZBuffer = (char*)uprv_malloc(sizeof(char) * tzInfo->defaultTZFileSize); + sizeFileRead = fread(tzInfo->defaultTZBuffer, 1, tzInfo->defaultTZFileSize, tzInfo->defaultTZFilePtr); + } + rewind(file); + while(sizeFileLeft > 0) { + uprv_memset(bufferFile, 0, MAX_READ_SIZE); + sizeFileToRead = sizeFileLeft < MAX_READ_SIZE ? sizeFileLeft : MAX_READ_SIZE; + + sizeFileRead = fread(bufferFile, 1, sizeFileToRead, file); + if (memcmp(tzInfo->defaultTZBuffer + tzInfo->defaultTZPosition, bufferFile, sizeFileRead) != 0) { + result = false; + break; + } + sizeFileLeft -= sizeFileRead; + tzInfo->defaultTZPosition += sizeFileRead; + } + } + } else { + result = false; + } + + if (file != nullptr) { + fclose(file); + } + + return result; +} + + +/* dirent also lists two entries: "." and ".." that we can safely ignore. */ +#define SKIP1 "." +#define SKIP2 ".." +static UBool U_CALLCONV putil_cleanup(); +static CharString *gSearchTZFileResult = nullptr; + +/* + * This method recursively traverses the directory given for a matching TZ file and returns the first match. + * This function is not thread safe - it uses a global, gSearchTZFileResult, to hold its results. + */ +static char* searchForTZFile(const char* path, DefaultTZInfo* tzInfo) { + DIR* dirp = nullptr; + struct dirent* dirEntry = nullptr; + char* result = nullptr; + UErrorCode status = U_ZERO_ERROR; + + /* Save the current path */ + CharString curpath(path, -1, status); + if (U_FAILURE(status)) { + goto cleanupAndReturn; + } + + dirp = opendir(path); + if (dirp == nullptr) { + goto cleanupAndReturn; + } + + if (gSearchTZFileResult == nullptr) { + gSearchTZFileResult = new CharString; + if (gSearchTZFileResult == nullptr) { + goto cleanupAndReturn; + } + ucln_common_registerCleanup(UCLN_COMMON_PUTIL, putil_cleanup); + } + + /* Check each entry in the directory. */ + while((dirEntry = readdir(dirp)) != nullptr) { + const char* dirName = dirEntry->d_name; + if (uprv_strcmp(dirName, SKIP1) != 0 && uprv_strcmp(dirName, SKIP2) != 0 + && uprv_strcmp(TZFILE_SKIP, dirName) != 0 && uprv_strcmp(TZFILE_SKIP2, dirName) != 0) { + /* Create a newpath with the new entry to test each entry in the directory. */ + CharString newpath(curpath, status); + newpath.append(dirName, -1, status); + if (U_FAILURE(status)) { + break; + } + + DIR* subDirp = nullptr; + if ((subDirp = opendir(newpath.data())) != nullptr) { + /* If this new path is a directory, make a recursive call with the newpath. */ + closedir(subDirp); + newpath.append('/', status); + if (U_FAILURE(status)) { + break; + } + result = searchForTZFile(newpath.data(), tzInfo); + /* + Have to get out here. Otherwise, we'd keep looking + and return the first match in the top-level directory + if there's a match in the top-level. If not, this function + would return nullptr and set gTimeZoneBufferPtr to nullptr in initDefault(). + It worked without this in most cases because we have a fallback of calling + localtime_r to figure out the default timezone. + */ + if (result != nullptr) + break; + } else { + if(compareBinaryFiles(TZDEFAULT, newpath.data(), tzInfo)) { + int32_t amountToSkip = sizeof(TZZONEINFO) - 1; + if (amountToSkip > newpath.length()) { + amountToSkip = newpath.length(); + } + const char* zoneid = newpath.data() + amountToSkip; + skipZoneIDPrefix(&zoneid); + gSearchTZFileResult->clear(); + gSearchTZFileResult->append(zoneid, -1, status); + if (U_FAILURE(status)) { + break; + } + result = gSearchTZFileResult->data(); + /* Get out after the first one found. */ + break; + } + } + } + } + + cleanupAndReturn: + if (dirp) { + closedir(dirp); + } + return result; +} +#endif + +#if U_PLATFORM == U_PF_ANDROID +typedef int(system_property_read_callback)(const prop_info* info, + void (*callback)(void* cookie, + const char* name, + const char* value, + uint32_t serial), + void* cookie); +typedef int(system_property_get)(const char*, char*); + +static char gAndroidTimeZone[PROP_VALUE_MAX] = { '\0' }; + +static void u_property_read(void* cookie, const char* name, const char* value, + uint32_t serial) { + uprv_strcpy((char* )cookie, value); +} +#endif + +U_CAPI void U_EXPORT2 +uprv_tzname_clear_cache() +{ +#if U_PLATFORM == U_PF_ANDROID + /* Android's timezone is stored in system property. */ + gAndroidTimeZone[0] = '\0'; + void* libc = dlopen("libc.so", RTLD_NOLOAD); + if (libc) { + /* Android API 26+ has new API to get system property and old API + * (__system_property_get) is deprecated */ + system_property_read_callback* property_read_callback = + (system_property_read_callback*)dlsym( + libc, "__system_property_read_callback"); + if (property_read_callback) { + const prop_info* info = + __system_property_find("persist.sys.timezone"); + if (info) { + property_read_callback(info, &u_property_read, gAndroidTimeZone); + } + } else { + system_property_get* property_get = + (system_property_get*)dlsym(libc, "__system_property_get"); + if (property_get) { + property_get("persist.sys.timezone", gAndroidTimeZone); + } + } + dlclose(libc); + } +#endif + +#if defined(CHECK_LOCALTIME_LINK) && !defined(DEBUG_SKIP_LOCALTIME_LINK) + gTimeZoneBufferPtr = nullptr; +#endif +} + +U_CAPI const char* U_EXPORT2 +uprv_tzname(int n) +{ + (void)n; // Avoid unreferenced parameter warning. + const char *tzid = nullptr; +#if U_PLATFORM_USES_ONLY_WIN32_API + tzid = uprv_detectWindowsTimeZone(); + + if (tzid != nullptr) { + return tzid; + } + +#ifndef U_TZNAME + // The return value is free'd in timezone.cpp on Windows because + // the other code path returns a pointer to a heap location. + // If we don't have a name already, then tzname wouldn't be any + // better, so just fall back. + return uprv_strdup(""); +#endif // !U_TZNAME + +#else + +/*#if U_PLATFORM_IS_DARWIN_BASED + int ret; + + tzid = getenv("TZFILE"); + if (tzid != nullptr) { + return tzid; + } +#endif*/ + +/* This code can be temporarily disabled to test tzname resolution later on. */ +#ifndef DEBUG_TZNAME +#if U_PLATFORM == U_PF_ANDROID + tzid = gAndroidTimeZone; +#else + tzid = getenv("TZ"); +#endif + if (tzid != nullptr && isValidOlsonID(tzid) +#if U_PLATFORM == U_PF_SOLARIS + /* Don't misinterpret TZ "localtime" on Solaris as a time zone name. */ + && uprv_strcmp(tzid, TZ_ENV_CHECK) != 0 +#endif + ) { + /* The colon forces tzset() to treat the remainder as zoneinfo path */ + if (tzid[0] == ':') { + tzid++; + } + /* This might be a good Olson ID. */ + skipZoneIDPrefix(&tzid); + return tzid; + } + /* else U_TZNAME will give a better result. */ +#endif + +#if defined(CHECK_LOCALTIME_LINK) && !defined(DEBUG_SKIP_LOCALTIME_LINK) + /* Caller must handle threading issues */ + if (gTimeZoneBufferPtr == nullptr) { + /* + This is a trick to look at the name of the link to get the Olson ID + because the tzfile contents is underspecified. + This isn't guaranteed to work because it may not be a symlink. + */ + char *ret = realpath(TZDEFAULT, gTimeZoneBuffer); + if (ret != nullptr && uprv_strcmp(TZDEFAULT, gTimeZoneBuffer) != 0) { + int32_t tzZoneInfoTailLen = uprv_strlen(TZZONEINFOTAIL); + const char *tzZoneInfoTailPtr = uprv_strstr(gTimeZoneBuffer, TZZONEINFOTAIL); + // MacOS14 has the realpath as something like + // /usr/share/zoneinfo.default/Australia/Melbourne + // which will not have "/zoneinfo/" in the path. + // Therefore if we fail, we fall back to read the link which is + // /var/db/timezone/zoneinfo/Australia/Melbourne + // We also fall back to reading the link if the realpath leads to something like + // /usr/share/zoneinfo/posixrules + if (tzZoneInfoTailPtr == nullptr || + uprv_strcmp(tzZoneInfoTailPtr + tzZoneInfoTailLen, "posixrules") == 0) { + ssize_t size = readlink(TZDEFAULT, gTimeZoneBuffer, sizeof(gTimeZoneBuffer)-1); + if (size > 0) { + gTimeZoneBuffer[size] = 0; + tzZoneInfoTailPtr = uprv_strstr(gTimeZoneBuffer, TZZONEINFOTAIL); + } + } + if (tzZoneInfoTailPtr != nullptr) { + tzZoneInfoTailPtr += tzZoneInfoTailLen; + skipZoneIDPrefix(&tzZoneInfoTailPtr); + if (isValidOlsonID(tzZoneInfoTailPtr)) { + return (gTimeZoneBufferPtr = tzZoneInfoTailPtr); + } + } + } else { +#if defined(SEARCH_TZFILE) + DefaultTZInfo* tzInfo = (DefaultTZInfo*)uprv_malloc(sizeof(DefaultTZInfo)); + if (tzInfo != nullptr) { + tzInfo->defaultTZBuffer = nullptr; + tzInfo->defaultTZFileSize = 0; + tzInfo->defaultTZFilePtr = nullptr; + tzInfo->defaultTZstatus = false; + tzInfo->defaultTZPosition = 0; + + gTimeZoneBufferPtr = searchForTZFile(TZZONEINFO, tzInfo); + + /* Free previously allocated memory */ + if (tzInfo->defaultTZBuffer != nullptr) { + uprv_free(tzInfo->defaultTZBuffer); + } + if (tzInfo->defaultTZFilePtr != nullptr) { + fclose(tzInfo->defaultTZFilePtr); + } + uprv_free(tzInfo); + } + + if (gTimeZoneBufferPtr != nullptr && isValidOlsonID(gTimeZoneBufferPtr)) { + return gTimeZoneBufferPtr; + } +#endif + } + } + else { + return gTimeZoneBufferPtr; + } +#endif +#endif + +#ifdef U_TZNAME +#if U_PLATFORM_USES_ONLY_WIN32_API + /* The return value is free'd in timezone.cpp on Windows because + * the other code path returns a pointer to a heap location. */ + return uprv_strdup(U_TZNAME[n]); +#else + /* + U_TZNAME is usually a non-unique abbreviation, which isn't normally usable. + So we remap the abbreviation to an olson ID. + + Since Windows exposes a little more timezone information, + we normally don't use this code on Windows because + uprv_detectWindowsTimeZone should have already given the correct answer. + */ + { + struct tm juneSol, decemberSol; + int daylightType; + static const time_t juneSolstice=1182478260; /*2007-06-21 18:11 UT*/ + static const time_t decemberSolstice=1198332540; /*2007-12-22 06:09 UT*/ + + /* This probing will tell us when daylight savings occurs. */ + localtime_r(&juneSolstice, &juneSol); + localtime_r(&decemberSolstice, &decemberSol); + if(decemberSol.tm_isdst > 0) { + daylightType = U_DAYLIGHT_DECEMBER; + } else if(juneSol.tm_isdst > 0) { + daylightType = U_DAYLIGHT_JUNE; + } else { + daylightType = U_DAYLIGHT_NONE; + } + tzid = remapShortTimeZone(U_TZNAME[0], U_TZNAME[1], daylightType, uprv_timezone()); + if (tzid != nullptr) { + return tzid; + } + } + return U_TZNAME[n]; +#endif +#else + return ""; +#endif +} + +/* Get and set the ICU data directory --------------------------------------- */ + +static icu::UInitOnce gDataDirInitOnce {}; +static char *gDataDirectory = nullptr; + +UInitOnce gTimeZoneFilesInitOnce {}; +static CharString *gTimeZoneFilesDirectory = nullptr; + +#if U_POSIX_LOCALE || U_PLATFORM_USES_ONLY_WIN32_API + static const char *gCorrectedPOSIXLocale = nullptr; /* Sometimes heap allocated */ + static bool gCorrectedPOSIXLocaleHeapAllocated = false; +#endif + +static UBool U_CALLCONV putil_cleanup() +{ + if (gDataDirectory && *gDataDirectory) { + uprv_free(gDataDirectory); + } + gDataDirectory = nullptr; + gDataDirInitOnce.reset(); + + delete gTimeZoneFilesDirectory; + gTimeZoneFilesDirectory = nullptr; + gTimeZoneFilesInitOnce.reset(); + +#ifdef SEARCH_TZFILE + delete gSearchTZFileResult; + gSearchTZFileResult = nullptr; +#endif + +#if U_POSIX_LOCALE || U_PLATFORM_USES_ONLY_WIN32_API + if (gCorrectedPOSIXLocale && gCorrectedPOSIXLocaleHeapAllocated) { + uprv_free(const_cast(gCorrectedPOSIXLocale)); + gCorrectedPOSIXLocale = nullptr; + gCorrectedPOSIXLocaleHeapAllocated = false; + } +#endif + return true; +} + +/* + * Set the data directory. + * Make a copy of the passed string, and set the global data dir to point to it. + */ +U_CAPI void U_EXPORT2 +u_setDataDirectory(const char *directory) { + char *newDataDir; + int32_t length; + + if(directory==nullptr || *directory==0) { + /* A small optimization to prevent the malloc and copy when the + shared library is used, and this is a way to make sure that nullptr + is never returned. + */ + newDataDir = (char *)""; + } + else { + length=(int32_t)uprv_strlen(directory); + newDataDir = (char *)uprv_malloc(length + 2); + /* Exit out if newDataDir could not be created. */ + if (newDataDir == nullptr) { + return; + } + uprv_strcpy(newDataDir, directory); + +#if (U_FILE_SEP_CHAR != U_FILE_ALT_SEP_CHAR) + { + char *p; + while((p = uprv_strchr(newDataDir, U_FILE_ALT_SEP_CHAR)) != nullptr) { + *p = U_FILE_SEP_CHAR; + } + } +#endif + } + + if (gDataDirectory && *gDataDirectory) { + uprv_free(gDataDirectory); + } + gDataDirectory = newDataDir; + ucln_common_registerCleanup(UCLN_COMMON_PUTIL, putil_cleanup); +} + +U_CAPI UBool U_EXPORT2 +uprv_pathIsAbsolute(const char *path) +{ + if(!path || !*path) { + return false; + } + + if(*path == U_FILE_SEP_CHAR) { + return true; + } + +#if (U_FILE_SEP_CHAR != U_FILE_ALT_SEP_CHAR) + if(*path == U_FILE_ALT_SEP_CHAR) { + return true; + } +#endif + +#if U_PLATFORM_USES_ONLY_WIN32_API + if( (((path[0] >= 'A') && (path[0] <= 'Z')) || + ((path[0] >= 'a') && (path[0] <= 'z'))) && + path[1] == ':' ) { + return true; + } +#endif + + return false; +} + +/* Backup setting of ICU_DATA_DIR_PREFIX_ENV_VAR + (needed for some Darwin ICU build environments) */ +#if U_PLATFORM_IS_DARWIN_BASED && defined(TARGET_OS_SIMULATOR) && TARGET_OS_SIMULATOR +# if !defined(ICU_DATA_DIR_PREFIX_ENV_VAR) +# define ICU_DATA_DIR_PREFIX_ENV_VAR "IPHONE_SIMULATOR_ROOT" +# endif +#endif + +#if defined(ICU_DATA_DIR_WINDOWS) +// Helper function to get the ICU Data Directory under the Windows directory location. +static BOOL U_CALLCONV getIcuDataDirectoryUnderWindowsDirectory(char* directoryBuffer, UINT bufferLength) +{ + wchar_t windowsPath[MAX_PATH]; + char windowsPathUtf8[MAX_PATH]; + + UINT length = GetSystemWindowsDirectoryW(windowsPath, UPRV_LENGTHOF(windowsPath)); + if ((length > 0) && (length < (UPRV_LENGTHOF(windowsPath) - 1))) { + // Convert UTF-16 to a UTF-8 string. + UErrorCode status = U_ZERO_ERROR; + int32_t windowsPathUtf8Len = 0; + u_strToUTF8(windowsPathUtf8, static_cast(UPRV_LENGTHOF(windowsPathUtf8)), + &windowsPathUtf8Len, reinterpret_cast(windowsPath), -1, &status); + + if (U_SUCCESS(status) && (status != U_STRING_NOT_TERMINATED_WARNING) && + (windowsPathUtf8Len < (UPRV_LENGTHOF(windowsPathUtf8) - 1))) { + // Ensure it always has a separator, so we can append the ICU data path. + if (windowsPathUtf8[windowsPathUtf8Len - 1] != U_FILE_SEP_CHAR) { + windowsPathUtf8[windowsPathUtf8Len++] = U_FILE_SEP_CHAR; + windowsPathUtf8[windowsPathUtf8Len] = '\0'; + } + // Check if the concatenated string will fit. + if ((windowsPathUtf8Len + UPRV_LENGTHOF(ICU_DATA_DIR_WINDOWS)) < bufferLength) { + uprv_strcpy(directoryBuffer, windowsPathUtf8); + uprv_strcat(directoryBuffer, ICU_DATA_DIR_WINDOWS); + return true; + } + } + } + + return false; +} +#endif + +static void U_CALLCONV dataDirectoryInitFn() { + /* If we already have the directory, then return immediately. Will happen if user called + * u_setDataDirectory(). + */ + if (gDataDirectory) { + return; + } + + const char *path = nullptr; +#if defined(ICU_DATA_DIR_PREFIX_ENV_VAR) + char datadir_path_buffer[PATH_MAX]; +#endif + + /* + When ICU_NO_USER_DATA_OVERRIDE is defined, users aren't allowed to + override ICU's data with the ICU_DATA environment variable. This prevents + problems where multiple custom copies of ICU's specific version of data + are installed on a system. Either the application must define the data + directory with u_setDataDirectory, define ICU_DATA_DIR when compiling + ICU, set the data with udata_setCommonData or trust that all of the + required data is contained in ICU's data library that contains + the entry point defined by U_ICUDATA_ENTRY_POINT. + + There may also be some platforms where environment variables + are not allowed. + */ +# if !defined(ICU_NO_USER_DATA_OVERRIDE) && !UCONFIG_NO_FILE_IO + /* First try to get the environment variable */ +# if U_PLATFORM_HAS_WINUWP_API == 0 // Windows UWP does not support getenv + path=getenv("ICU_DATA"); +# endif +# endif + + /* ICU_DATA_DIR may be set as a compile option. + * U_ICU_DATA_DEFAULT_DIR is provided and is set by ICU at compile time + * and is used only when data is built in archive mode eliminating the need + * for ICU_DATA_DIR to be set. U_ICU_DATA_DEFAULT_DIR is set to the installation + * directory of the data dat file. Users should use ICU_DATA_DIR if they want to + * set their own path. + */ +#if defined(ICU_DATA_DIR) || defined(U_ICU_DATA_DEFAULT_DIR) + if(path==nullptr || *path==0) { +# if defined(ICU_DATA_DIR_PREFIX_ENV_VAR) + const char *prefix = getenv(ICU_DATA_DIR_PREFIX_ENV_VAR); +# endif +# ifdef ICU_DATA_DIR + path=ICU_DATA_DIR; +# else + path=U_ICU_DATA_DEFAULT_DIR; +# endif +# if defined(ICU_DATA_DIR_PREFIX_ENV_VAR) + if (prefix != nullptr) { + snprintf(datadir_path_buffer, sizeof(datadir_path_buffer), "%s%s", prefix, path); + path=datadir_path_buffer; + } +# endif + } +#endif + +#if defined(ICU_DATA_DIR_WINDOWS) + char datadir_path_buffer[MAX_PATH]; + if (getIcuDataDirectoryUnderWindowsDirectory(datadir_path_buffer, UPRV_LENGTHOF(datadir_path_buffer))) { + path = datadir_path_buffer; + } +#endif + + if(path==nullptr) { + /* It looks really bad, set it to something. */ + path = ""; + } + + u_setDataDirectory(path); + return; +} + +U_CAPI const char * U_EXPORT2 +u_getDataDirectory() { + umtx_initOnce(gDataDirInitOnce, &dataDirectoryInitFn); + return gDataDirectory; +} + +static void setTimeZoneFilesDir(const char *path, UErrorCode &status) { + if (U_FAILURE(status)) { + return; + } + gTimeZoneFilesDirectory->clear(); + gTimeZoneFilesDirectory->append(path, status); +#if (U_FILE_SEP_CHAR != U_FILE_ALT_SEP_CHAR) + char *p = gTimeZoneFilesDirectory->data(); + while ((p = uprv_strchr(p, U_FILE_ALT_SEP_CHAR)) != nullptr) { + *p = U_FILE_SEP_CHAR; + } +#endif +} + +#define TO_STRING(x) TO_STRING_2(x) +#define TO_STRING_2(x) #x + +static void U_CALLCONV TimeZoneDataDirInitFn(UErrorCode &status) { + U_ASSERT(gTimeZoneFilesDirectory == nullptr); + ucln_common_registerCleanup(UCLN_COMMON_PUTIL, putil_cleanup); + gTimeZoneFilesDirectory = new CharString(); + if (gTimeZoneFilesDirectory == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + + const char *dir = ""; + +#if defined(ICU_TIMEZONE_FILES_DIR_PREFIX_ENV_VAR) + char timezonefilesdir_path_buffer[PATH_MAX]; + const char *prefix = getenv(ICU_TIMEZONE_FILES_DIR_PREFIX_ENV_VAR); +#endif + +#if U_PLATFORM_HAS_WINUWP_API == 1 +// The UWP version does not support the environment variable setting. + +# if defined(ICU_DATA_DIR_WINDOWS) + // When using the Windows system data, we can possibly pick up time zone data from the Windows directory. + char datadir_path_buffer[MAX_PATH]; + if (getIcuDataDirectoryUnderWindowsDirectory(datadir_path_buffer, UPRV_LENGTHOF(datadir_path_buffer))) { + dir = datadir_path_buffer; + } +# endif + +#else + dir = getenv("ICU_TIMEZONE_FILES_DIR"); +#endif // U_PLATFORM_HAS_WINUWP_API + +#if defined(U_TIMEZONE_FILES_DIR) + if (dir == nullptr) { + // Build time configuration setting. + dir = TO_STRING(U_TIMEZONE_FILES_DIR); + } +#endif + + if (dir == nullptr) { + dir = ""; + } + +#if defined(ICU_TIMEZONE_FILES_DIR_PREFIX_ENV_VAR) + if (prefix != nullptr) { + snprintf(timezonefilesdir_path_buffer, sizeof(timezonefilesdir_path_buffer), "%s%s", prefix, dir); + dir = timezonefilesdir_path_buffer; + } +#endif + + setTimeZoneFilesDir(dir, status); +} + + +U_CAPI const char * U_EXPORT2 +u_getTimeZoneFilesDirectory(UErrorCode *status) { + umtx_initOnce(gTimeZoneFilesInitOnce, &TimeZoneDataDirInitFn, *status); + return U_SUCCESS(*status) ? gTimeZoneFilesDirectory->data() : ""; +} + +U_CAPI void U_EXPORT2 +u_setTimeZoneFilesDirectory(const char *path, UErrorCode *status) { + umtx_initOnce(gTimeZoneFilesInitOnce, &TimeZoneDataDirInitFn, *status); + setTimeZoneFilesDir(path, *status); + + // Note: this function does some extra churn, first setting based on the + // environment, then immediately replacing with the value passed in. + // The logic is simpler that way, and performance shouldn't be an issue. +} + + +#if U_POSIX_LOCALE +/* A helper function used by uprv_getPOSIXIDForDefaultLocale and + * uprv_getPOSIXIDForDefaultCodepage. Returns the posix locale id for + * LC_CTYPE and LC_MESSAGES. It doesn't support other locale categories. + */ +static const char *uprv_getPOSIXIDForCategory(int category) +{ + const char* posixID = nullptr; + if (category == LC_MESSAGES || category == LC_CTYPE) { + /* + * On Solaris two different calls to setlocale can result in + * different values. Only get this value once. + * + * We must check this first because an application can set this. + * + * LC_ALL can't be used because it's platform dependent. The LANG + * environment variable seems to affect LC_CTYPE variable by default. + * Here is what setlocale(LC_ALL, nullptr) can return. + * HPUX can return 'C C C C C C C' + * Solaris can return /en_US/C/C/C/C/C on the second try. + * Linux can return LC_CTYPE=C;LC_NUMERIC=C;... + * + * The default codepage detection also needs to use LC_CTYPE. + * + * Do not call setlocale(LC_*, "")! Using an empty string instead + * of nullptr, will modify the libc behavior. + */ + posixID = setlocale(category, nullptr); + if ((posixID == 0) + || (uprv_strcmp("C", posixID) == 0) + || (uprv_strcmp("POSIX", posixID) == 0)) + { + /* Maybe we got some garbage. Try something more reasonable */ + posixID = getenv("LC_ALL"); + /* Solaris speaks POSIX - See IEEE Std 1003.1-2008 + * This is needed to properly handle empty env. variables + */ +#if U_PLATFORM == U_PF_SOLARIS + if ((posixID == 0) || (posixID[0] == '\0')) { + posixID = getenv(category == LC_MESSAGES ? "LC_MESSAGES" : "LC_CTYPE"); + if ((posixID == 0) || (posixID[0] == '\0')) { +#else + if (posixID == 0) { + posixID = getenv(category == LC_MESSAGES ? "LC_MESSAGES" : "LC_CTYPE"); + if (posixID == 0) { +#endif + posixID = getenv("LANG"); + } + } + } + } + if ((posixID==0) + || (uprv_strcmp("C", posixID) == 0) + || (uprv_strcmp("POSIX", posixID) == 0)) + { + /* Nothing worked. Give it a nice POSIX default value. */ + posixID = "en_US_POSIX"; + // Note: this test will not catch 'C.UTF-8', + // that will be handled in uprv_getDefaultLocaleID(). + // Leave this mapping here for the uprv_getPOSIXIDForDefaultCodepage() + // caller which expects to see "en_US_POSIX" in many branches. + } + return posixID; +} + +/* Return just the POSIX id for the default locale, whatever happens to be in + * it. It gets the value from LC_MESSAGES and indirectly from LC_ALL and LANG. + */ +static const char *uprv_getPOSIXIDForDefaultLocale() +{ + static const char* posixID = nullptr; + if (posixID == 0) { + posixID = uprv_getPOSIXIDForCategory(LC_MESSAGES); + } + return posixID; +} + +#if !U_CHARSET_IS_UTF8 +/* Return just the POSIX id for the default codepage, whatever happens to be in + * it. It gets the value from LC_CTYPE and indirectly from LC_ALL and LANG. + */ +static const char *uprv_getPOSIXIDForDefaultCodepage() +{ + static const char* posixID = nullptr; + if (posixID == 0) { + posixID = uprv_getPOSIXIDForCategory(LC_CTYPE); + } + return posixID; +} +#endif +#endif + +/* NOTE: The caller should handle thread safety */ +U_CAPI const char* U_EXPORT2 +uprv_getDefaultLocaleID() +{ +#if U_POSIX_LOCALE +/* + Note that: (a '!' means the ID is improper somehow) + LC_ALL ----> default_loc codepage +-------------------------------------------------------- + ab.CD ab CD + ab@CD ab__CD - + ab@CD.EF ab__CD EF + + ab_CD.EF@GH ab_CD_GH EF + +Some 'improper' ways to do the same as above: + ! ab_CD@GH.EF ab_CD_GH EF + ! ab_CD.EF@GH.IJ ab_CD_GH EF + ! ab_CD@ZZ.EF@GH.IJ ab_CD_GH EF + + _CD@GH _CD_GH - + _CD.EF@GH _CD_GH EF + +The variant cannot have dots in it. +The 'rightmost' variant (@xxx) wins. +The leftmost codepage (.xxx) wins. +*/ + const char* posixID = uprv_getPOSIXIDForDefaultLocale(); + + /* Format: (no spaces) + ll [ _CC ] [ . MM ] [ @ VV] + + l = lang, C = ctry, M = charmap, V = variant + */ + + if (gCorrectedPOSIXLocale != nullptr) { + return gCorrectedPOSIXLocale; + } + + // Copy the ID into owned memory. + // Over-allocate in case we replace "C" with "en_US_POSIX" (+10), + null termination + char *correctedPOSIXLocale = static_cast(uprv_malloc(uprv_strlen(posixID) + 10 + 1)); + if (correctedPOSIXLocale == nullptr) { + return nullptr; + } + uprv_strcpy(correctedPOSIXLocale, posixID); + + char *limit; + if ((limit = uprv_strchr(correctedPOSIXLocale, '.')) != nullptr) { + *limit = 0; + } + if ((limit = uprv_strchr(correctedPOSIXLocale, '@')) != nullptr) { + *limit = 0; + } + + if ((uprv_strcmp("C", correctedPOSIXLocale) == 0) // no @ variant + || (uprv_strcmp("POSIX", correctedPOSIXLocale) == 0)) { + // Raw input was C.* or POSIX.*, Give it a nice POSIX default value. + // (The "C"/"POSIX" case is handled in uprv_getPOSIXIDForCategory()) + uprv_strcpy(correctedPOSIXLocale, "en_US_POSIX"); + } + + /* Note that we scan the *uncorrected* ID. */ + const char *p; + if ((p = uprv_strrchr(posixID, '@')) != nullptr) { + p++; + + /* Take care of any special cases here.. */ + if (!uprv_strcmp(p, "nynorsk")) { + p = "NY"; + /* Don't worry about no__NY. In practice, it won't appear. */ + } + + if (uprv_strchr(correctedPOSIXLocale,'_') == nullptr) { + uprv_strcat(correctedPOSIXLocale, "__"); /* aa@b -> aa__b (note this can make the new locale 1 char longer) */ + } + else { + uprv_strcat(correctedPOSIXLocale, "_"); /* aa_CC@b -> aa_CC_b */ + } + + const char *q; + if ((q = uprv_strchr(p, '.')) != nullptr) { + /* How big will the resulting string be? */ + int32_t len = (int32_t)(uprv_strlen(correctedPOSIXLocale) + (q-p)); + uprv_strncat(correctedPOSIXLocale, p, q-p); // do not include charset + correctedPOSIXLocale[len] = 0; + } + else { + /* Anything following the @ sign */ + uprv_strcat(correctedPOSIXLocale, p); + } + + /* Should there be a map from 'no@nynorsk' -> no_NO_NY here? + * How about 'russian' -> 'ru'? + * Many of the other locales using ISO codes will be handled by the + * canonicalization functions in uloc_getDefault. + */ + } + + if (gCorrectedPOSIXLocale == nullptr) { + gCorrectedPOSIXLocale = correctedPOSIXLocale; + gCorrectedPOSIXLocaleHeapAllocated = true; + ucln_common_registerCleanup(UCLN_COMMON_PUTIL, putil_cleanup); + correctedPOSIXLocale = nullptr; + } + posixID = gCorrectedPOSIXLocale; + + if (correctedPOSIXLocale != nullptr) { /* Was already set - clean up. */ + uprv_free(correctedPOSIXLocale); + } + + return posixID; + +#elif U_PLATFORM_USES_ONLY_WIN32_API +#define POSIX_LOCALE_CAPACITY 64 + UErrorCode status = U_ZERO_ERROR; + char *correctedPOSIXLocale = nullptr; + + // If we have already figured this out just use the cached value + if (gCorrectedPOSIXLocale != nullptr) { + return gCorrectedPOSIXLocale; + } + + // No cached value, need to determine the current value + static WCHAR windowsLocale[LOCALE_NAME_MAX_LENGTH] = {}; + int length = GetLocaleInfoEx(LOCALE_NAME_USER_DEFAULT, LOCALE_SNAME, windowsLocale, LOCALE_NAME_MAX_LENGTH); + + // Now we should have a Windows locale name that needs converted to the POSIX style. + if (length > 0) // If length is 0, then the GetLocaleInfoEx failed. + { + // First we need to go from UTF-16 to char (and also convert from _ to - while we're at it.) + char modifiedWindowsLocale[LOCALE_NAME_MAX_LENGTH] = {}; + + int32_t i; + for (i = 0; i < UPRV_LENGTHOF(modifiedWindowsLocale); i++) + { + if (windowsLocale[i] == '_') + { + modifiedWindowsLocale[i] = '-'; + } + else + { + modifiedWindowsLocale[i] = static_cast(windowsLocale[i]); + } + + if (modifiedWindowsLocale[i] == '\0') + { + break; + } + } + + if (i >= UPRV_LENGTHOF(modifiedWindowsLocale)) + { + // Ran out of room, can't really happen, maybe we'll be lucky about a matching + // locale when tags are dropped + modifiedWindowsLocale[UPRV_LENGTHOF(modifiedWindowsLocale) - 1] = '\0'; + } + + // Now normalize the resulting name + correctedPOSIXLocale = static_cast(uprv_malloc(POSIX_LOCALE_CAPACITY + 1)); + /* TODO: Should we just exit on memory allocation failure? */ + if (correctedPOSIXLocale) + { + int32_t posixLen = uloc_canonicalize(modifiedWindowsLocale, correctedPOSIXLocale, POSIX_LOCALE_CAPACITY, &status); + if (U_SUCCESS(status)) + { + *(correctedPOSIXLocale + posixLen) = 0; + gCorrectedPOSIXLocale = correctedPOSIXLocale; + gCorrectedPOSIXLocaleHeapAllocated = true; + ucln_common_registerCleanup(UCLN_COMMON_PUTIL, putil_cleanup); + } + else + { + uprv_free(correctedPOSIXLocale); + } + } + } + + // If unable to find a locale we can agree upon, use en-US by default + if (gCorrectedPOSIXLocale == nullptr) { + gCorrectedPOSIXLocale = "en_US"; + } + return gCorrectedPOSIXLocale; + +#elif U_PLATFORM == U_PF_OS400 + /* locales are process scoped and are by definition thread safe */ + static char correctedLocale[64]; + const char *localeID = getenv("LC_ALL"); + char *p; + + if (localeID == nullptr) + localeID = getenv("LANG"); + if (localeID == nullptr) + localeID = setlocale(LC_ALL, nullptr); + /* Make sure we have something... */ + if (localeID == nullptr) + return "en_US_POSIX"; + + /* Extract the locale name from the path. */ + if((p = uprv_strrchr(localeID, '/')) != nullptr) + { + /* Increment p to start of locale name. */ + p++; + localeID = p; + } + + /* Copy to work location. */ + uprv_strcpy(correctedLocale, localeID); + + /* Strip off the '.locale' extension. */ + if((p = uprv_strchr(correctedLocale, '.')) != nullptr) { + *p = 0; + } + + /* Upper case the locale name. */ + T_CString_toUpperCase(correctedLocale); + + /* See if we are using the POSIX locale. Any of the + * following are equivalent and use the same QLGPGCMA + * (POSIX) locale. + * QLGPGCMA2 means UCS2 + * QLGPGCMA_4 means UTF-32 + * QLGPGCMA_8 means UTF-8 + */ + if ((uprv_strcmp("C", correctedLocale) == 0) || + (uprv_strcmp("POSIX", correctedLocale) == 0) || + (uprv_strncmp("QLGPGCMA", correctedLocale, 8) == 0)) + { + uprv_strcpy(correctedLocale, "en_US_POSIX"); + } + else + { + int16_t LocaleLen; + + /* Lower case the lang portion. */ + for(p = correctedLocale; *p != 0 && *p != '_'; p++) + { + *p = uprv_tolower(*p); + } + + /* Adjust for Euro. After '_E' add 'URO'. */ + LocaleLen = uprv_strlen(correctedLocale); + if (correctedLocale[LocaleLen - 2] == '_' && + correctedLocale[LocaleLen - 1] == 'E') + { + uprv_strcat(correctedLocale, "URO"); + } + + /* If using Lotus-based locale then convert to + * equivalent non Lotus. + */ + else if (correctedLocale[LocaleLen - 2] == '_' && + correctedLocale[LocaleLen - 1] == 'L') + { + correctedLocale[LocaleLen - 2] = 0; + } + + /* There are separate simplified and traditional + * locales called zh_HK_S and zh_HK_T. + */ + else if (uprv_strncmp(correctedLocale, "zh_HK", 5) == 0) + { + uprv_strcpy(correctedLocale, "zh_HK"); + } + + /* A special zh_CN_GBK locale... + */ + else if (uprv_strcmp(correctedLocale, "zh_CN_GBK") == 0) + { + uprv_strcpy(correctedLocale, "zh_CN"); + } + + } + + return correctedLocale; +#endif + +} + +#if !U_CHARSET_IS_UTF8 +#if U_POSIX_LOCALE +/* +Due to various platform differences, one platform may specify a charset, +when they really mean a different charset. Remap the names so that they are +compatible with ICU. Only conflicting/ambiguous aliases should be resolved +here. Before adding anything to this function, please consider adding unique +names to the ICU alias table in the data directory. +*/ +static const char* +remapPlatformDependentCodepage(const char *locale, const char *name) { + if (locale != nullptr && *locale == 0) { + /* Make sure that an empty locale is handled the same way. */ + locale = nullptr; + } + if (name == nullptr) { + return nullptr; + } +#if U_PLATFORM == U_PF_AIX + if (uprv_strcmp(name, "IBM-943") == 0) { + /* Use the ASCII compatible ibm-943 */ + name = "Shift-JIS"; + } + else if (uprv_strcmp(name, "IBM-1252") == 0) { + /* Use the windows-1252 that contains the Euro */ + name = "IBM-5348"; + } +#elif U_PLATFORM == U_PF_SOLARIS + if (locale != nullptr && uprv_strcmp(name, "EUC") == 0) { + /* Solaris underspecifies the "EUC" name. */ + if (uprv_strcmp(locale, "zh_CN") == 0) { + name = "EUC-CN"; + } + else if (uprv_strcmp(locale, "zh_TW") == 0) { + name = "EUC-TW"; + } + else if (uprv_strcmp(locale, "ko_KR") == 0) { + name = "EUC-KR"; + } + } + else if (uprv_strcmp(name, "eucJP") == 0) { + /* + ibm-954 is the best match. + ibm-33722 is the default for eucJP (similar to Windows). + */ + name = "eucjis"; + } + else if (uprv_strcmp(name, "646") == 0) { + /* + * The default codepage given by Solaris is 646 but the C library routines treat it as if it was + * ISO-8859-1 instead of US-ASCII(646). + */ + name = "ISO-8859-1"; + } +#elif U_PLATFORM_IS_DARWIN_BASED + if (locale == nullptr && *name == 0) { + /* + No locale was specified, and an empty name was passed in. + This usually indicates that nl_langinfo didn't return valid information. + Mac OS X uses UTF-8 by default (especially the locale data and console). + */ + name = "UTF-8"; + } + else if (uprv_strcmp(name, "CP949") == 0) { + /* Remap CP949 to a similar codepage to avoid issues with backslash and won symbol. */ + name = "EUC-KR"; + } + else if (locale != nullptr && uprv_strcmp(locale, "en_US_POSIX") != 0 && uprv_strcmp(name, "US-ASCII") == 0) { + /* + * For non C/POSIX locale, default the code page to UTF-8 instead of US-ASCII. + */ + name = "UTF-8"; + } +#elif U_PLATFORM == U_PF_BSD + if (uprv_strcmp(name, "CP949") == 0) { + /* Remap CP949 to a similar codepage to avoid issues with backslash and won symbol. */ + name = "EUC-KR"; + } +#elif U_PLATFORM == U_PF_HPUX + if (locale != nullptr && uprv_strcmp(locale, "zh_HK") == 0 && uprv_strcmp(name, "big5") == 0) { + /* HP decided to extend big5 as hkbig5 even though it's not compatible :-( */ + /* zh_TW.big5 is not the same charset as zh_HK.big5! */ + name = "hkbig5"; + } + else if (uprv_strcmp(name, "eucJP") == 0) { + /* + ibm-1350 is the best match, but unavailable. + ibm-954 is mostly a superset of ibm-1350. + ibm-33722 is the default for eucJP (similar to Windows). + */ + name = "eucjis"; + } +#elif U_PLATFORM == U_PF_LINUX + if (locale != nullptr && uprv_strcmp(name, "euc") == 0) { + /* Linux underspecifies the "EUC" name. */ + if (uprv_strcmp(locale, "korean") == 0) { + name = "EUC-KR"; + } + else if (uprv_strcmp(locale, "japanese") == 0) { + /* See comment below about eucJP */ + name = "eucjis"; + } + } + else if (uprv_strcmp(name, "eucjp") == 0) { + /* + ibm-1350 is the best match, but unavailable. + ibm-954 is mostly a superset of ibm-1350. + ibm-33722 is the default for eucJP (similar to Windows). + */ + name = "eucjis"; + } + else if (locale != nullptr && uprv_strcmp(locale, "en_US_POSIX") != 0 && + (uprv_strcmp(name, "ANSI_X3.4-1968") == 0 || uprv_strcmp(name, "US-ASCII") == 0)) { + /* + * For non C/POSIX locale, default the code page to UTF-8 instead of US-ASCII. + */ + name = "UTF-8"; + } + /* + * Linux returns ANSI_X3.4-1968 for C/POSIX, but the call site takes care of + * it by falling back to 'US-ASCII' when nullptr is returned from this + * function. So, we don't have to worry about it here. + */ +#endif + /* return nullptr when "" is passed in */ + if (*name == 0) { + name = nullptr; + } + return name; +} + +static const char* +getCodepageFromPOSIXID(const char *localeName, char * buffer, int32_t buffCapacity) +{ + char localeBuf[100]; + const char *name = nullptr; + char *variant = nullptr; + + if (localeName != nullptr && (name = (uprv_strchr(localeName, '.'))) != nullptr) { + size_t localeCapacity = uprv_min(sizeof(localeBuf), (name-localeName)+1); + uprv_strncpy(localeBuf, localeName, localeCapacity); + localeBuf[localeCapacity-1] = 0; /* ensure NUL termination */ + name = uprv_strncpy(buffer, name+1, buffCapacity); + buffer[buffCapacity-1] = 0; /* ensure NUL termination */ + if ((variant = const_cast(uprv_strchr(name, '@'))) != nullptr) { + *variant = 0; + } + name = remapPlatformDependentCodepage(localeBuf, name); + } + return name; +} +#endif + +static const char* +int_getDefaultCodepage() +{ +#if U_PLATFORM == U_PF_OS400 + uint32_t ccsid = 37; /* Default to ibm-37 */ + static char codepage[64]; + Qwc_JOBI0400_t jobinfo; + Qus_EC_t error = { sizeof(Qus_EC_t) }; /* SPI error code */ + + EPT_CALL(QUSRJOBI)(&jobinfo, sizeof(jobinfo), "JOBI0400", + "* ", " ", &error); + + if (error.Bytes_Available == 0) { + if (jobinfo.Coded_Char_Set_ID != 0xFFFF) { + ccsid = (uint32_t)jobinfo.Coded_Char_Set_ID; + } + else if (jobinfo.Default_Coded_Char_Set_Id != 0xFFFF) { + ccsid = (uint32_t)jobinfo.Default_Coded_Char_Set_Id; + } + /* else use the default */ + } + snprintf(codepage, sizeof(codepage), "ibm-%d", ccsid); + return codepage; + +#elif U_PLATFORM == U_PF_OS390 + static char codepage[64]; + + strncpy(codepage, nl_langinfo(CODESET),63-strlen(UCNV_SWAP_LFNL_OPTION_STRING)); + strcat(codepage,UCNV_SWAP_LFNL_OPTION_STRING); + codepage[63] = 0; /* NUL terminate */ + + return codepage; + +#elif U_PLATFORM_USES_ONLY_WIN32_API + static char codepage[64]; + DWORD codepageNumber = 0; + +#if U_PLATFORM_HAS_WINUWP_API == 1 + // UWP doesn't have a direct API to get the default ACP as Microsoft would rather + // have folks use Unicode than a "system" code page, however this is the same + // codepage as the system default locale codepage. (FWIW, the system locale is + // ONLY used for codepage, it should never be used for anything else) + GetLocaleInfoEx(LOCALE_NAME_SYSTEM_DEFAULT, LOCALE_IDEFAULTANSICODEPAGE | LOCALE_RETURN_NUMBER, + (LPWSTR)&codepageNumber, sizeof(codepageNumber) / sizeof(WCHAR)); +#else + // Win32 apps can call GetACP + codepageNumber = GetACP(); +#endif + // Special case for UTF-8 + if (codepageNumber == 65001) + { + return "UTF-8"; + } + // Windows codepages can look like windows-1252, so format the found number + // the numbers are eclectic, however all valid system code pages, besides UTF-8 + // are between 3 and 19999 + if (codepageNumber > 0 && codepageNumber < 20000) + { + snprintf(codepage, sizeof(codepage), "windows-%ld", codepageNumber); + return codepage; + } + // If the codepage number call failed then return UTF-8 + return "UTF-8"; + +#elif U_POSIX_LOCALE + static char codesetName[100]; + const char *localeName = nullptr; + const char *name = nullptr; + + localeName = uprv_getPOSIXIDForDefaultCodepage(); + uprv_memset(codesetName, 0, sizeof(codesetName)); + /* On Solaris nl_langinfo returns C locale values unless setlocale + * was called earlier. + */ +#if (U_HAVE_NL_LANGINFO_CODESET && U_PLATFORM != U_PF_SOLARIS) + /* When available, check nl_langinfo first because it usually gives more + useful names. It depends on LC_CTYPE. + nl_langinfo may use the same buffer as setlocale. */ + { + const char *codeset = nl_langinfo(U_NL_LANGINFO_CODESET); +#if U_PLATFORM_IS_DARWIN_BASED || U_PLATFORM_IS_LINUX_BASED + /* + * On Linux and MacOSX, ensure that default codepage for non C/POSIX locale is UTF-8 + * instead of ASCII. + */ + if (uprv_strcmp(localeName, "en_US_POSIX") != 0) { + codeset = remapPlatformDependentCodepage(localeName, codeset); + } else +#endif + { + codeset = remapPlatformDependentCodepage(nullptr, codeset); + } + + if (codeset != nullptr) { + uprv_strncpy(codesetName, codeset, sizeof(codesetName)); + codesetName[sizeof(codesetName)-1] = 0; + return codesetName; + } + } +#endif + + /* Use setlocale in a nice way, and then check some environment variables. + Maybe the application used setlocale already. + */ + uprv_memset(codesetName, 0, sizeof(codesetName)); + name = getCodepageFromPOSIXID(localeName, codesetName, sizeof(codesetName)); + if (name) { + /* if we can find the codeset name from setlocale, return that. */ + return name; + } + + if (*codesetName == 0) + { + /* Everything failed. Return US ASCII (ISO 646). */ + (void)uprv_strcpy(codesetName, "US-ASCII"); + } + return codesetName; +#else + return "US-ASCII"; +#endif +} + + +U_CAPI const char* U_EXPORT2 +uprv_getDefaultCodepage() +{ + static char const *name = nullptr; + umtx_lock(nullptr); + if (name == nullptr) { + name = int_getDefaultCodepage(); + } + umtx_unlock(nullptr); + return name; +} +#endif /* !U_CHARSET_IS_UTF8 */ + + +/* end of platform-specific implementation -------------- */ + +/* version handling --------------------------------------------------------- */ + +U_CAPI void U_EXPORT2 +u_versionFromString(UVersionInfo versionArray, const char *versionString) { + char *end; + uint16_t part=0; + + if(versionArray==nullptr) { + return; + } + + if(versionString!=nullptr) { + for(;;) { + versionArray[part]=(uint8_t)uprv_strtoul(versionString, &end, 10); + if(end==versionString || ++part==U_MAX_VERSION_LENGTH || *end!=U_VERSION_DELIMITER) { + break; + } + versionString=end+1; + } + } + + while(partU_MAX_VERSION_STRING_LENGTH) { + len = U_MAX_VERSION_STRING_LENGTH; + } + u_UCharsToChars(versionString, versionChars, len); + versionChars[len]=0; + u_versionFromString(versionArray, versionChars); + } +} + +U_CAPI void U_EXPORT2 +u_versionToString(const UVersionInfo versionArray, char *versionString) { + uint16_t count, part; + uint8_t field; + + if(versionString==nullptr) { + return; + } + + if(versionArray==nullptr) { + versionString[0]=0; + return; + } + + /* count how many fields need to be written */ + for(count=4; count>0 && versionArray[count-1]==0; --count) { + } + + if(count <= 1) { + count = 2; + } + + /* write the first part */ + /* write the decimal field value */ + field=versionArray[0]; + if(field>=100) { + *versionString++=(char)('0'+field/100); + field%=100; + } + if(field>=10) { + *versionString++=(char)('0'+field/10); + field%=10; + } + *versionString++=(char)('0'+field); + + /* write the following parts */ + for(part=1; part=100) { + *versionString++=(char)('0'+field/100); + field%=100; + } + if(field>=10) { + *versionString++=(char)('0'+field/10); + field%=10; + } + *versionString++=(char)('0'+field); + } + + /* NUL-terminate */ + *versionString=0; +} + +U_CAPI void U_EXPORT2 +u_getVersion(UVersionInfo versionArray) { + (void)copyright; // Suppress unused variable warning from clang. + u_versionFromString(versionArray, U_ICU_VERSION); +} + +/** + * icucfg.h dependent code + */ + +#if U_ENABLE_DYLOAD && HAVE_DLOPEN && !U_PLATFORM_USES_ONLY_WIN32_API + +#if HAVE_DLFCN_H +#ifdef __MVS__ +#ifndef __SUSV3 +#define __SUSV3 1 +#endif +#endif +#include +#endif /* HAVE_DLFCN_H */ + +U_CAPI void * U_EXPORT2 +uprv_dl_open(const char *libName, UErrorCode *status) { + void *ret = nullptr; + if(U_FAILURE(*status)) return ret; + ret = dlopen(libName, RTLD_NOW|RTLD_GLOBAL); + if(ret==nullptr) { +#ifdef U_TRACE_DYLOAD + printf("dlerror on dlopen(%s): %s\n", libName, dlerror()); +#endif + *status = U_MISSING_RESOURCE_ERROR; + } + return ret; +} + +U_CAPI void U_EXPORT2 +uprv_dl_close(void *lib, UErrorCode *status) { + if(U_FAILURE(*status)) return; + dlclose(lib); +} + +U_CAPI UVoidFunction* U_EXPORT2 +uprv_dlsym_func(void *lib, const char* sym, UErrorCode *status) { + union { + UVoidFunction *fp; + void *vp; + } uret; + uret.fp = nullptr; + if(U_FAILURE(*status)) return uret.fp; + uret.vp = dlsym(lib, sym); + if(uret.vp == nullptr) { +#ifdef U_TRACE_DYLOAD + printf("dlerror on dlsym(%p,%s): %s\n", lib,sym, dlerror()); +#endif + *status = U_MISSING_RESOURCE_ERROR; + } + return uret.fp; +} + +#elif U_ENABLE_DYLOAD && U_PLATFORM_USES_ONLY_WIN32_API && !U_PLATFORM_HAS_WINUWP_API + +/* Windows API implementation. */ +// Note: UWP does not expose/allow these APIs, so the UWP version gets the null implementation. */ + +U_CAPI void * U_EXPORT2 +uprv_dl_open(const char *libName, UErrorCode *status) { + HMODULE lib = nullptr; + + if(U_FAILURE(*status)) return nullptr; + + lib = LoadLibraryA(libName); + + if(lib==nullptr) { + *status = U_MISSING_RESOURCE_ERROR; + } + + return (void*)lib; +} + +U_CAPI void U_EXPORT2 +uprv_dl_close(void *lib, UErrorCode *status) { + HMODULE handle = (HMODULE)lib; + if(U_FAILURE(*status)) return; + + FreeLibrary(handle); + + return; +} + +U_CAPI UVoidFunction* U_EXPORT2 +uprv_dlsym_func(void *lib, const char* sym, UErrorCode *status) { + HMODULE handle = (HMODULE)lib; + UVoidFunction* addr = nullptr; + + if(U_FAILURE(*status) || lib==nullptr) return nullptr; + + addr = (UVoidFunction*)GetProcAddress(handle, sym); + + if(addr==nullptr) { + DWORD lastError = GetLastError(); + if(lastError == ERROR_PROC_NOT_FOUND) { + *status = U_MISSING_RESOURCE_ERROR; + } else { + *status = U_UNSUPPORTED_ERROR; /* other unknown error. */ + } + } + + return addr; +} + +#else + +/* No dynamic loading, null (nonexistent) implementation. */ + +U_CAPI void * U_EXPORT2 +uprv_dl_open(const char *libName, UErrorCode *status) { + (void)libName; + if(U_FAILURE(*status)) return nullptr; + *status = U_UNSUPPORTED_ERROR; + return nullptr; +} + +U_CAPI void U_EXPORT2 +uprv_dl_close(void *lib, UErrorCode *status) { + (void)lib; + if(U_FAILURE(*status)) return; + *status = U_UNSUPPORTED_ERROR; + return; +} + +U_CAPI UVoidFunction* U_EXPORT2 +uprv_dlsym_func(void *lib, const char* sym, UErrorCode *status) { + (void)lib; + (void)sym; + if(U_SUCCESS(*status)) { + *status = U_UNSUPPORTED_ERROR; + } + return (UVoidFunction*)nullptr; +} + +#endif + +/* + * Hey, Emacs, please set the following: + * + * Local Variables: + * indent-tabs-mode: nil + * End: + * + */ diff --git a/intl/icu/source/common/putilimp.h b/intl/icu/source/common/putilimp.h new file mode 100644 index 0000000000..7097232e71 --- /dev/null +++ b/intl/icu/source/common/putilimp.h @@ -0,0 +1,621 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1997-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* FILE NAME : putilimp.h +* +* Date Name Description +* 10/17/04 grhoten Move internal functions from putil.h to this file. +****************************************************************************** +*/ + +#ifndef PUTILIMP_H +#define PUTILIMP_H + +#include "unicode/utypes.h" +#include "unicode/putil.h" + +/** + * \def U_SIGNED_RIGHT_SHIFT_IS_ARITHMETIC + * Nearly all CPUs and compilers implement a right-shift of a signed integer + * as an Arithmetic Shift Right which copies the sign bit (the Most Significant Bit (MSB)) + * into the vacated bits (sign extension). + * For example, (int32_t)0xfff5fff3>>4 becomes 0xffff5fff and -1>>1=-1. + * + * This can be useful for storing a signed value in the upper bits + * and another bit field in the lower bits. + * The signed value can be retrieved by simple right-shifting. + * + * This is consistent with the Java language. + * + * However, the C standard allows compilers to implement a right-shift of a signed integer + * as a Logical Shift Right which copies a 0 into the vacated bits. + * For example, (int32_t)0xfff5fff3>>4 becomes 0x0fff5fff and -1>>1=0x7fffffff. + * + * Code that depends on the natural behavior should be guarded with this macro, + * with an alternate path for unusual platforms. + * @internal + */ +#ifdef U_SIGNED_RIGHT_SHIFT_IS_ARITHMETIC + /* Use the predefined value. */ +#else + /* + * Nearly all CPUs & compilers implement a right-shift of a signed integer + * as an Arithmetic Shift Right (with sign extension). + */ +# define U_SIGNED_RIGHT_SHIFT_IS_ARITHMETIC 1 +#endif + +/** Define this to 1 if your platform supports IEEE 754 floating point, + to 0 if it does not. */ +#ifndef IEEE_754 +# define IEEE_754 1 +#endif + +/** + * uintptr_t is an optional part of the standard definitions in stdint.h. + * The opengroup.org documentation for stdint.h says + * "On XSI-conformant systems, the intptr_t and uintptr_t types are required; + * otherwise, they are optional." + * We assume that when uintptr_t is defined, UINTPTR_MAX is defined as well. + * + * Do not use ptrdiff_t since it is signed. size_t is unsigned. + */ +/* TODO: This check fails on some z environments. Filed a ticket #9357 for this. */ +#if !defined(__intptr_t_defined) && !defined(UINTPTR_MAX) && (U_PLATFORM != U_PF_OS390) +typedef size_t uintptr_t; +#endif + +/*===========================================================================*/ +/** @{ Information about POSIX support */ +/*===========================================================================*/ + +#ifdef U_HAVE_NL_LANGINFO_CODESET + /* Use the predefined value. */ +#elif U_PLATFORM_USES_ONLY_WIN32_API || U_PLATFORM == U_PF_ANDROID || U_PLATFORM == U_PF_QNX +# define U_HAVE_NL_LANGINFO_CODESET 0 +#else +# define U_HAVE_NL_LANGINFO_CODESET 1 +#endif + +#ifdef U_NL_LANGINFO_CODESET + /* Use the predefined value. */ +#elif !U_HAVE_NL_LANGINFO_CODESET +# define U_NL_LANGINFO_CODESET -1 +#elif U_PLATFORM == U_PF_OS400 + /* not defined */ +#else +# define U_NL_LANGINFO_CODESET CODESET +#endif + +#if defined(U_TZSET) || defined(U_HAVE_TZSET) + /* Use the predefined value. */ +#elif U_PLATFORM_USES_ONLY_WIN32_API + // UWP doesn't support tzset or environment variables for tz +#if U_PLATFORM_HAS_WINUWP_API == 0 +# define U_TZSET _tzset +#endif +#elif U_PLATFORM == U_PF_OS400 + /* not defined */ +#elif defined(__wasi__) + /* not defined */ +#else +# define U_TZSET tzset +#endif + +#if defined(U_TIMEZONE) || defined(U_HAVE_TIMEZONE) + /* Use the predefined value. */ +#elif U_PLATFORM == U_PF_ANDROID +# define U_TIMEZONE timezone +#elif defined(__UCLIBC__) + // uClibc does not have __timezone or _timezone. +#elif defined(_NEWLIB_VERSION) +# define U_TIMEZONE _timezone +#elif defined(__GLIBC__) + // glibc +# define U_TIMEZONE __timezone +#elif U_PLATFORM_IS_LINUX_BASED + // not defined +#elif U_PLATFORM_USES_ONLY_WIN32_API +# define U_TIMEZONE _timezone +#elif U_PLATFORM == U_PF_BSD && !defined(__NetBSD__) + /* not defined */ +#elif U_PLATFORM == U_PF_OS400 + /* not defined */ +#elif U_PLATFORM == U_PF_IPHONE + /* not defined */ +#elif defined(__wasi__) + /* not defined */ +#else +# define U_TIMEZONE timezone +#endif + +#if defined(U_TZNAME) || defined(U_HAVE_TZNAME) + /* Use the predefined value. */ +#elif U_PLATFORM_USES_ONLY_WIN32_API + /* not usable on all windows platforms */ +#if U_PLATFORM_HAS_WINUWP_API == 0 +# define U_TZNAME _tzname +#endif +#elif U_PLATFORM == U_PF_OS400 + /* not defined */ +#elif defined(__wasi__) + /* not defined */ +#else +# define U_TZNAME tzname +#endif + +#ifdef U_HAVE_MMAP + /* Use the predefined value. */ +#elif U_PLATFORM_USES_ONLY_WIN32_API +# define U_HAVE_MMAP 0 +#else +# define U_HAVE_MMAP 1 +#endif + +#ifdef U_HAVE_POPEN + /* Use the predefined value. */ +#elif U_PLATFORM_USES_ONLY_WIN32_API +# define U_HAVE_POPEN 0 +#elif U_PLATFORM == U_PF_OS400 +# define U_HAVE_POPEN 0 +#else +# define U_HAVE_POPEN 1 +#endif + +/** + * \def U_HAVE_DIRENT_H + * Defines whether dirent.h is available. + * @internal + */ +#ifdef U_HAVE_DIRENT_H + /* Use the predefined value. */ +#elif U_PLATFORM_USES_ONLY_WIN32_API +# define U_HAVE_DIRENT_H 0 +#else +# define U_HAVE_DIRENT_H 1 +#endif + +/** @} */ + +/*===========================================================================*/ +/** @{ Programs used by ICU code */ +/*===========================================================================*/ + +/** + * \def U_MAKE_IS_NMAKE + * Defines whether the "make" program is Windows nmake. + */ +#ifdef U_MAKE_IS_NMAKE + /* Use the predefined value. */ +#elif U_PLATFORM == U_PF_WINDOWS +# define U_MAKE_IS_NMAKE 1 +#else +# define U_MAKE_IS_NMAKE 0 +#endif + +/** @} */ + +/*==========================================================================*/ +/* Platform utilities */ +/*==========================================================================*/ + +/** + * Platform utilities isolates the platform dependencies of the + * library. For each platform which this code is ported to, these + * functions may have to be re-implemented. + */ + +/** + * Floating point utility to determine if a double is Not a Number (NaN). + * @internal + */ +U_CAPI UBool U_EXPORT2 uprv_isNaN(double d); +/** + * Floating point utility to determine if a double has an infinite value. + * @internal + */ +U_CAPI UBool U_EXPORT2 uprv_isInfinite(double d); +/** + * Floating point utility to determine if a double has a positive infinite value. + * @internal + */ +U_CAPI UBool U_EXPORT2 uprv_isPositiveInfinity(double d); +/** + * Floating point utility to determine if a double has a negative infinite value. + * @internal + */ +U_CAPI UBool U_EXPORT2 uprv_isNegativeInfinity(double d); +/** + * Floating point utility that returns a Not a Number (NaN) value. + * @internal + */ +U_CAPI double U_EXPORT2 uprv_getNaN(void); +/** + * Floating point utility that returns an infinite value. + * @internal + */ +U_CAPI double U_EXPORT2 uprv_getInfinity(void); + +/** + * Floating point utility to truncate a double. + * @internal + */ +U_CAPI double U_EXPORT2 uprv_trunc(double d); +/** + * Floating point utility to calculate the floor of a double. + * @internal + */ +U_CAPI double U_EXPORT2 uprv_floor(double d); +/** + * Floating point utility to calculate the ceiling of a double. + * @internal + */ +U_CAPI double U_EXPORT2 uprv_ceil(double d); +/** + * Floating point utility to calculate the absolute value of a double. + * @internal + */ +U_CAPI double U_EXPORT2 uprv_fabs(double d); +/** + * Floating point utility to calculate the fractional and integer parts of a double. + * @internal + */ +U_CAPI double U_EXPORT2 uprv_modf(double d, double* pinteger); +/** + * Floating point utility to calculate the remainder of a double divided by another double. + * @internal + */ +U_CAPI double U_EXPORT2 uprv_fmod(double d, double y); +/** + * Floating point utility to calculate d to the power of exponent (d^exponent). + * @internal + */ +U_CAPI double U_EXPORT2 uprv_pow(double d, double exponent); +/** + * Floating point utility to calculate 10 to the power of exponent (10^exponent). + * @internal + */ +U_CAPI double U_EXPORT2 uprv_pow10(int32_t exponent); +/** + * Floating point utility to calculate the maximum value of two doubles. + * @internal + */ +U_CAPI double U_EXPORT2 uprv_fmax(double d, double y); +/** + * Floating point utility to calculate the minimum value of two doubles. + * @internal + */ +U_CAPI double U_EXPORT2 uprv_fmin(double d, double y); +/** + * Private utility to calculate the maximum value of two integers. + * @internal + */ +U_CAPI int32_t U_EXPORT2 uprv_max(int32_t d, int32_t y); +/** + * Private utility to calculate the minimum value of two integers. + * @internal + */ +U_CAPI int32_t U_EXPORT2 uprv_min(int32_t d, int32_t y); + +#if U_IS_BIG_ENDIAN +# define uprv_isNegative(number) (*((signed char *)&(number))<0) +#else +# define uprv_isNegative(number) (*((signed char *)&(number)+sizeof(number)-1)<0) +#endif + +/** + * Return the largest positive number that can be represented by an integer + * type of arbitrary bit length. + * @internal + */ +U_CAPI double U_EXPORT2 uprv_maxMantissa(void); + +/** + * Floating point utility to calculate the logarithm of a double. + * @internal + */ +U_CAPI double U_EXPORT2 uprv_log(double d); + +/** + * Does common notion of rounding e.g. uprv_floor(x + 0.5); + * @param x the double number + * @return the rounded double + * @internal + */ +U_CAPI double U_EXPORT2 uprv_round(double x); + +/** + * Adds the signed integers a and b, storing the result in res. + * Checks for signed integer overflow. + * Similar to the GCC/Clang extension __builtin_add_overflow + * + * @param a The first operand. + * @param b The second operand. + * @param res a + b + * @return true if overflow occurred; false if no overflow occurred. + * @internal + */ +U_CAPI UBool U_EXPORT2 uprv_add32_overflow(int32_t a, int32_t b, int32_t* res); + +/** + * Multiplies the signed integers a and b, storing the result in res. + * Checks for signed integer overflow. + * Similar to the GCC/Clang extension __builtin_mul_overflow + * + * @param a The first multiplicand. + * @param b The second multiplicand. + * @param res a * b + * @return true if overflow occurred; false if no overflow occurred. + * @internal + */ +U_CAPI UBool U_EXPORT2 uprv_mul32_overflow(int32_t a, int32_t b, int32_t* res); + +#if 0 +/** + * Returns the number of digits after the decimal point in a double number x. + * + * @param x the double number + * @return the number of digits after the decimal point in a double number x. + * @internal + */ +/*U_CAPI int32_t U_EXPORT2 uprv_digitsAfterDecimal(double x);*/ +#endif + +#if !U_CHARSET_IS_UTF8 +/** + * Please use ucnv_getDefaultName() instead. + * Return the default codepage for this platform and locale. + * This function can call setlocale() on Unix platforms. Please read the + * platform documentation on setlocale() before calling this function. + * @return the default codepage for this platform + * @internal + */ +U_CAPI const char* U_EXPORT2 uprv_getDefaultCodepage(void); +#endif + +/** + * Please use uloc_getDefault() instead. + * Return the default locale ID string by querying the system, or + * zero if one cannot be found. + * This function can call setlocale() on Unix platforms. Please read the + * platform documentation on setlocale() before calling this function. + * @return the default locale ID string + * @internal + */ +U_CAPI const char* U_EXPORT2 uprv_getDefaultLocaleID(void); + +/** + * Time zone utilities + * + * Wrappers for C runtime library functions relating to timezones. + * The t_tzset() function (similar to tzset) uses the current setting + * of the environment variable TZ to assign values to three global + * variables: daylight, timezone, and tzname. These variables have the + * following meanings, and are declared in <time.h>. + * + * daylight Nonzero if daylight-saving-time zone (DST) is specified + * in TZ; otherwise, 0. Default value is 1. + * timezone Difference in seconds between coordinated universal + * time and local time. E.g., -28,800 for PST (GMT-8hrs) + * tzname(0) Three-letter time-zone name derived from TZ environment + * variable. E.g., "PST". + * tzname(1) Three-letter DST zone name derived from TZ environment + * variable. E.g., "PDT". If DST zone is omitted from TZ, + * tzname(1) is an empty string. + * + * Notes: For example, to set the TZ environment variable to correspond + * to the current time zone in Germany, you can use one of the + * following statements: + * + * set TZ=GST1GDT + * set TZ=GST+1GDT + * + * If the TZ value is not set, t_tzset() attempts to use the time zone + * information specified by the operating system. Under Windows NT + * and Windows 95, this information is specified in the Control Panel's + * Date/Time application. + * @internal + */ +U_CAPI void U_EXPORT2 uprv_tzset(void); + +/** + * Difference in seconds between coordinated universal + * time and local time. E.g., -28,800 for PST (GMT-8hrs) + * @return the difference in seconds between coordinated universal time and local time. + * @internal + */ +U_CAPI int32_t U_EXPORT2 uprv_timezone(void); + +/** + * tzname(0) Three-letter time-zone name derived from TZ environment + * variable. E.g., "PST". + * tzname(1) Three-letter DST zone name derived from TZ environment + * variable. E.g., "PDT". If DST zone is omitted from TZ, + * tzname(1) is an empty string. + * @internal + */ +U_CAPI const char* U_EXPORT2 uprv_tzname(int n); + +/** + * Reset the global tzname cache. + * @internal + */ +U_CAPI void uprv_tzname_clear_cache(void); + +/** + * Get UTC (GMT) time measured in milliseconds since 0:00 on 1/1/1970. + * This function is affected by 'faketime' and should be the bottleneck for all user-visible ICU time functions. + * @return the UTC time measured in milliseconds + * @internal + */ +U_CAPI UDate U_EXPORT2 uprv_getUTCtime(void); + +/** + * Get UTC (GMT) time measured in milliseconds since 0:00 on 1/1/1970. + * This function is not affected by 'faketime', so it should only be used by low level test functions- not by anything that + * exposes time to the end user. + * @return the UTC time measured in milliseconds + * @internal + */ +U_CAPI UDate U_EXPORT2 uprv_getRawUTCtime(void); + +/** + * Determine whether a pathname is absolute or not, as defined by the platform. + * @param path Pathname to test + * @return true if the path is absolute + * @internal (ICU 3.0) + */ +U_CAPI UBool U_EXPORT2 uprv_pathIsAbsolute(const char *path); + +/** + * Use U_MAX_PTR instead of this function. + * @param void pointer to test + * @return the largest possible pointer greater than the base + * @internal (ICU 3.8) + */ +U_CAPI void * U_EXPORT2 uprv_maximumPtr(void *base); + +/** + * Maximum value of a (void*) - use to indicate the limit of an 'infinite' buffer. + * In fact, buffer sizes must not exceed 2GB so that the difference between + * the buffer limit and the buffer start can be expressed in an int32_t. + * + * The definition of U_MAX_PTR must fulfill the following conditions: + * - return the largest possible pointer greater than base + * - return a valid pointer according to the machine architecture (AS/400, 64-bit, etc.) + * - avoid wrapping around at high addresses + * - make sure that the returned pointer is not farther from base than 0x7fffffff bytes + * + * @param base The beginning of a buffer to find the maximum offset from + * @internal + */ +#ifndef U_MAX_PTR +# if U_PLATFORM == U_PF_OS390 && !defined(_LP64) + /* We have 31-bit pointers. */ +# define U_MAX_PTR(base) ((void *)0x7fffffff) +# elif U_PLATFORM == U_PF_OS400 +# define U_MAX_PTR(base) uprv_maximumPtr((void *)base) +# elif 0 + /* + * For platforms where pointers are scalar values (which is normal, but unlike i5/OS) + * but that do not define uintptr_t. + * + * However, this does not work on modern compilers: + * The C++ standard does not define pointer overflow, and allows compilers to + * assume that p+u>p for any pointer p and any integer u>0. + * Thus, modern compilers optimize away the ">" comparison. + * (See ICU tickets #7187 and #8096.) + */ +# define U_MAX_PTR(base) \ + ((void *)(((char *)(base)+0x7fffffffu) > (char *)(base) \ + ? ((char *)(base)+0x7fffffffu) \ + : (char *)-1)) +# else + /* Default version. C++ standard compliant for scalar pointers. */ +# define U_MAX_PTR(base) \ + ((void *)(((uintptr_t)(base)+0x7fffffffu) > (uintptr_t)(base) \ + ? ((uintptr_t)(base)+0x7fffffffu) \ + : (uintptr_t)-1)) +# endif +#endif + + +#ifdef __cplusplus +/** + * Pin a buffer capacity such that doing pointer arithmetic + * on the destination pointer and capacity cannot overflow. + * + * The pinned capacity must fulfill the following conditions (for positive capacities): + * - dest + capacity is a valid pointer according to the machine architecture (AS/400, 64-bit, etc.) + * - (dest + capacity) >= dest + * - The size (in bytes) of T[capacity] does not exceed 0x7fffffff + * + * @param dest the destination buffer pointer. + * @param capacity the requested buffer capacity, in units of type T. + * @return the pinned capacity. + * @internal + */ +template +inline int32_t pinCapacity(T *dest, int32_t capacity) { + if (capacity <= 0) { return capacity; } + + uintptr_t destInt = (uintptr_t)dest; + uintptr_t maxInt; + +# if U_PLATFORM == U_PF_OS390 && !defined(_LP64) + // We have 31-bit pointers. + maxInt = 0x7fffffff; +# elif U_PLATFORM == U_PF_OS400 + maxInt = (uintptr_t)uprv_maximumPtr((void *)dest); +# else + maxInt = destInt + 0x7fffffffu; + if (maxInt < destInt) { + // Less than 2GB to the end of the address space. + // Pin to that to prevent address overflow. + maxInt = (uintptr_t)-1; + } +# endif + + uintptr_t maxBytes = maxInt - destInt; // max. 2GB + int32_t maxCapacity = (int32_t)(maxBytes / sizeof(T)); + return capacity <= maxCapacity ? capacity : maxCapacity; +} +#endif // __cplusplus + +/* Dynamic Library Functions */ + +typedef void (UVoidFunction)(void); + +#if U_ENABLE_DYLOAD +/** + * Load a library + * @internal (ICU 4.4) + */ +U_CAPI void * U_EXPORT2 uprv_dl_open(const char *libName, UErrorCode *status); + +/** + * Close a library + * @internal (ICU 4.4) + */ +U_CAPI void U_EXPORT2 uprv_dl_close( void *lib, UErrorCode *status); + +/** + * Extract a symbol from a library (function) + * @internal (ICU 4.8) + */ +U_CAPI UVoidFunction* U_EXPORT2 uprv_dlsym_func( void *lib, const char *symbolName, UErrorCode *status); + +/** + * Extract a symbol from a library (function) + * Not implemented, no clients. + * @internal + */ +/* U_CAPI void * U_EXPORT2 uprv_dlsym_data( void *lib, const char *symbolName, UErrorCode *status); */ + +#endif + +/** + * Define malloc and related functions + * @internal + */ +#if U_PLATFORM == U_PF_OS400 +# define uprv_default_malloc(x) _C_TS_malloc(x) +# define uprv_default_realloc(x,y) _C_TS_realloc(x,y) +# define uprv_default_free(x) _C_TS_free(x) +/* also _C_TS_calloc(x) */ +#else +/* C defaults */ +# define uprv_default_malloc(x) malloc(x) +# define uprv_default_realloc(x,y) realloc(x,y) +# define uprv_default_free(x) free(x) +#endif + + +#endif diff --git a/intl/icu/source/common/rbbi.cpp b/intl/icu/source/common/rbbi.cpp new file mode 100644 index 0000000000..73716ab406 --- /dev/null +++ b/intl/icu/source/common/rbbi.cpp @@ -0,0 +1,1286 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +*************************************************************************** +* Copyright (C) 1999-2016 International Business Machines Corporation +* and others. All rights reserved. +*************************************************************************** +*/ +// +// file: rbbi.cpp Contains the implementation of the rule based break iterator +// runtime engine and the API implementation for +// class RuleBasedBreakIterator +// + +#include "utypeinfo.h" // for 'typeid' to work + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include + +#include "unicode/rbbi.h" +#include "unicode/schriter.h" +#include "unicode/uchriter.h" +#include "unicode/uclean.h" +#include "unicode/udata.h" + +#include "brkeng.h" +#include "ucln_cmn.h" +#include "cmemory.h" +#include "cstring.h" +#include "localsvc.h" +#include "rbbidata.h" +#include "rbbi_cache.h" +#include "rbbirb.h" +#include "uassert.h" +#include "umutex.h" +#include "uvectr32.h" + +#ifdef RBBI_DEBUG +static UBool gTrace = false; +#endif + +U_NAMESPACE_BEGIN + +// The state number of the starting state +constexpr int32_t START_STATE = 1; + +// The state-transition value indicating "stop" +constexpr int32_t STOP_STATE = 0; + + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(RuleBasedBreakIterator) + + +//======================================================================= +// constructors +//======================================================================= + +/** + * Constructs a RuleBasedBreakIterator that uses the already-created + * tables object that is passed in as a parameter. + */ +RuleBasedBreakIterator::RuleBasedBreakIterator(RBBIDataHeader* data, UErrorCode &status) + : RuleBasedBreakIterator(&status) +{ + fData = new RBBIDataWrapper(data, status); // status checked in constructor + if (U_FAILURE(status)) {return;} + if(fData == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + if (fData->fForwardTable->fLookAheadResultsSize > 0) { + fLookAheadMatches = static_cast( + uprv_malloc(fData->fForwardTable->fLookAheadResultsSize * sizeof(int32_t))); + if (fLookAheadMatches == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + } +} + +//------------------------------------------------------------------------------- +// +// Constructor from a UDataMemory handle to precompiled break rules +// stored in an ICU data file. This construcotr is private API, +// only for internal use. +// +//------------------------------------------------------------------------------- +RuleBasedBreakIterator::RuleBasedBreakIterator(UDataMemory* udm, UBool isPhraseBreaking, + UErrorCode &status) : RuleBasedBreakIterator(udm, status) +{ + fIsPhraseBreaking = isPhraseBreaking; +} + +// +// Construct from precompiled binary rules (tables). This constructor is public API, +// taking the rules as a (const uint8_t *) to match the type produced by getBinaryRules(). +// +RuleBasedBreakIterator::RuleBasedBreakIterator(const uint8_t *compiledRules, + uint32_t ruleLength, + UErrorCode &status) + : RuleBasedBreakIterator(&status) +{ + if (U_FAILURE(status)) { + return; + } + if (compiledRules == nullptr || ruleLength < sizeof(RBBIDataHeader)) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + const RBBIDataHeader *data = (const RBBIDataHeader *)compiledRules; + if (data->fLength > ruleLength) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + fData = new RBBIDataWrapper(data, RBBIDataWrapper::kDontAdopt, status); + if (U_FAILURE(status)) {return;} + if(fData == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + if (fData->fForwardTable->fLookAheadResultsSize > 0) { + fLookAheadMatches = static_cast( + uprv_malloc(fData->fForwardTable->fLookAheadResultsSize * sizeof(int32_t))); + if (fLookAheadMatches == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + } +} + + +//------------------------------------------------------------------------------- +// +// Constructor from a UDataMemory handle to precompiled break rules +// stored in an ICU data file. +// +//------------------------------------------------------------------------------- +RuleBasedBreakIterator::RuleBasedBreakIterator(UDataMemory* udm, UErrorCode &status) + : RuleBasedBreakIterator(&status) +{ + fData = new RBBIDataWrapper(udm, status); // status checked in constructor + if (U_FAILURE(status)) {return;} + if(fData == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + if (fData->fForwardTable->fLookAheadResultsSize > 0) { + fLookAheadMatches = static_cast( + uprv_malloc(fData->fForwardTable->fLookAheadResultsSize * sizeof(int32_t))); + if (fLookAheadMatches == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + } +} + + + +//------------------------------------------------------------------------------- +// +// Constructor from a set of rules supplied as a string. +// +//------------------------------------------------------------------------------- +RuleBasedBreakIterator::RuleBasedBreakIterator( const UnicodeString &rules, + UParseError &parseError, + UErrorCode &status) + : RuleBasedBreakIterator(&status) +{ + if (U_FAILURE(status)) {return;} + RuleBasedBreakIterator *bi = (RuleBasedBreakIterator *) + RBBIRuleBuilder::createRuleBasedBreakIterator(rules, &parseError, status); + // Note: This is a bit awkward. The RBBI ruleBuilder has a factory method that + // creates and returns a complete RBBI. From here, in a constructor, we + // can't just return the object created by the builder factory, hence + // the assignment of the factory created object to "this". + if (U_SUCCESS(status)) { + *this = *bi; + delete bi; + } +} + + +//------------------------------------------------------------------------------- +// +// Default Constructor. Create an empty shell that can be set up later. +// Used when creating a RuleBasedBreakIterator from a set +// of rules. +//------------------------------------------------------------------------------- +RuleBasedBreakIterator::RuleBasedBreakIterator() + : RuleBasedBreakIterator(nullptr) +{ +} + +/** + * Simple Constructor with an error code. + * Handles common initialization for all other constructors. + */ +RuleBasedBreakIterator::RuleBasedBreakIterator(UErrorCode *status) { + UErrorCode ec = U_ZERO_ERROR; + if (status == nullptr) { + status = &ec; + } + utext_openUChars(&fText, nullptr, 0, status); + LocalPointer lpDictionaryCache(new DictionaryCache(this, *status), *status); + LocalPointer lpBreakCache(new BreakCache(this, *status), *status); + if (U_FAILURE(*status)) { + fErrorCode = *status; + return; + } + fDictionaryCache = lpDictionaryCache.orphan(); + fBreakCache = lpBreakCache.orphan(); + +#ifdef RBBI_DEBUG + static UBool debugInitDone = false; + if (debugInitDone == false) { + char *debugEnv = getenv("U_RBBIDEBUG"); + if (debugEnv && uprv_strstr(debugEnv, "trace")) { + gTrace = true; + } + debugInitDone = true; + } +#endif +} + + +//------------------------------------------------------------------------------- +// +// Copy constructor. Will produce a break iterator with the same behavior, +// and which iterates over the same text, as the one passed in. +// +//------------------------------------------------------------------------------- +RuleBasedBreakIterator::RuleBasedBreakIterator(const RuleBasedBreakIterator& other) +: RuleBasedBreakIterator() +{ + *this = other; +} + + +/** + * Destructor + */ +RuleBasedBreakIterator::~RuleBasedBreakIterator() { + if (fCharIter != &fSCharIter) { + // fCharIter was adopted from the outside. + delete fCharIter; + } + fCharIter = nullptr; + + utext_close(&fText); + + if (fData != nullptr) { + fData->removeReference(); + fData = nullptr; + } + delete fBreakCache; + fBreakCache = nullptr; + + delete fDictionaryCache; + fDictionaryCache = nullptr; + + delete fLanguageBreakEngines; + fLanguageBreakEngines = nullptr; + + delete fUnhandledBreakEngine; + fUnhandledBreakEngine = nullptr; + + uprv_free(fLookAheadMatches); + fLookAheadMatches = nullptr; +} + +/** + * Assignment operator. Sets this iterator to have the same behavior, + * and iterate over the same text, as the one passed in. + * TODO: needs better handling of memory allocation errors. + */ +RuleBasedBreakIterator& +RuleBasedBreakIterator::operator=(const RuleBasedBreakIterator& that) { + if (this == &that) { + return *this; + } + BreakIterator::operator=(that); + + if (fLanguageBreakEngines != nullptr) { + delete fLanguageBreakEngines; + fLanguageBreakEngines = nullptr; // Just rebuild for now + } + // TODO: clone fLanguageBreakEngines from "that" + UErrorCode status = U_ZERO_ERROR; + utext_clone(&fText, &that.fText, false, true, &status); + + if (fCharIter != &fSCharIter) { + delete fCharIter; + } + fCharIter = &fSCharIter; + + if (that.fCharIter != nullptr && that.fCharIter != &that.fSCharIter) { + // This is a little bit tricky - it will initially appear that + // this->fCharIter is adopted, even if that->fCharIter was + // not adopted. That's ok. + fCharIter = that.fCharIter->clone(); + } + fSCharIter = that.fSCharIter; + if (fCharIter == nullptr) { + fCharIter = &fSCharIter; + } + + if (fData != nullptr) { + fData->removeReference(); + fData = nullptr; + } + if (that.fData != nullptr) { + fData = that.fData->addReference(); + } + + uprv_free(fLookAheadMatches); + fLookAheadMatches = nullptr; + if (fData && fData->fForwardTable->fLookAheadResultsSize > 0) { + fLookAheadMatches = static_cast( + uprv_malloc(fData->fForwardTable->fLookAheadResultsSize * sizeof(int32_t))); + } + + + fPosition = that.fPosition; + fRuleStatusIndex = that.fRuleStatusIndex; + fDone = that.fDone; + + // TODO: both the dictionary and the main cache need to be copied. + // Current position could be within a dictionary range. Trying to continue + // the iteration without the caches present would go to the rules, with + // the assumption that the current position is on a rule boundary. + fBreakCache->reset(fPosition, fRuleStatusIndex); + fDictionaryCache->reset(); + + return *this; +} + +//----------------------------------------------------------------------------- +// +// clone - Returns a newly-constructed RuleBasedBreakIterator with the same +// behavior, and iterating over the same text, as this one. +// Virtual function: does the right thing with subclasses. +// +//----------------------------------------------------------------------------- +RuleBasedBreakIterator* +RuleBasedBreakIterator::clone() const { + return new RuleBasedBreakIterator(*this); +} + +/** + * Equality operator. Returns true if both BreakIterators are of the + * same class, have the same behavior, and iterate over the same text. + */ +bool +RuleBasedBreakIterator::operator==(const BreakIterator& that) const { + if (typeid(*this) != typeid(that)) { + return false; + } + if (this == &that) { + return true; + } + + // The base class BreakIterator carries no state that participates in equality, + // and does not implement an equality function that would otherwise be + // checked at this point. + + const RuleBasedBreakIterator& that2 = static_cast(that); + + if (!utext_equals(&fText, &that2.fText)) { + // The two break iterators are operating on different text, + // or have a different iteration position. + // Note that fText's position is always the same as the break iterator's position. + return false; + } + + if (!(fPosition == that2.fPosition && + fRuleStatusIndex == that2.fRuleStatusIndex && + fDone == that2.fDone)) { + return false; + } + + if (that2.fData == fData || + (fData != nullptr && that2.fData != nullptr && *that2.fData == *fData)) { + // The two break iterators are using the same rules. + return true; + } + return false; +} + +/** + * Compute a hash code for this BreakIterator + * @return A hash code + */ +int32_t +RuleBasedBreakIterator::hashCode() const { + int32_t hash = 0; + if (fData != nullptr) { + hash = fData->hashCode(); + } + return hash; +} + + +void RuleBasedBreakIterator::setText(UText *ut, UErrorCode &status) { + if (U_FAILURE(status)) { + return; + } + fBreakCache->reset(); + fDictionaryCache->reset(); + utext_clone(&fText, ut, false, true, &status); + + // Set up a dummy CharacterIterator to be returned if anyone + // calls getText(). With input from UText, there is no reasonable + // way to return a characterIterator over the actual input text. + // Return one over an empty string instead - this is the closest + // we can come to signaling a failure. + // (GetText() is obsolete, this failure is sort of OK) + fSCharIter.setText(u"", 0); + + if (fCharIter != &fSCharIter) { + // existing fCharIter was adopted from the outside. Delete it now. + delete fCharIter; + } + fCharIter = &fSCharIter; + + this->first(); +} + + +UText *RuleBasedBreakIterator::getUText(UText *fillIn, UErrorCode &status) const { + UText *result = utext_clone(fillIn, &fText, false, true, &status); + return result; +} + + +//======================================================================= +// BreakIterator overrides +//======================================================================= + +/** + * Return a CharacterIterator over the text being analyzed. + */ +CharacterIterator& +RuleBasedBreakIterator::getText() const { + return *fCharIter; +} + +/** + * Set the iterator to analyze a new piece of text. This function resets + * the current iteration position to the beginning of the text. + * @param newText An iterator over the text to analyze. + */ +void +RuleBasedBreakIterator::adoptText(CharacterIterator* newText) { + // If we are holding a CharacterIterator adopted from a + // previous call to this function, delete it now. + if (fCharIter != &fSCharIter) { + delete fCharIter; + } + + fCharIter = newText; + UErrorCode status = U_ZERO_ERROR; + fBreakCache->reset(); + fDictionaryCache->reset(); + if (newText==nullptr || newText->startIndex() != 0) { + // startIndex !=0 wants to be an error, but there's no way to report it. + // Make the iterator text be an empty string. + utext_openUChars(&fText, nullptr, 0, &status); + } else { + utext_openCharacterIterator(&fText, newText, &status); + } + this->first(); +} + +/** + * Set the iterator to analyze a new piece of text. This function resets + * the current iteration position to the beginning of the text. + * @param newText An iterator over the text to analyze. + */ +void +RuleBasedBreakIterator::setText(const UnicodeString& newText) { + UErrorCode status = U_ZERO_ERROR; + fBreakCache->reset(); + fDictionaryCache->reset(); + utext_openConstUnicodeString(&fText, &newText, &status); + + // Set up a character iterator on the string. + // Needed in case someone calls getText(). + // Can not, unfortunately, do this lazily on the (probably never) + // call to getText(), because getText is const. + fSCharIter.setText(newText.getBuffer(), newText.length()); + + if (fCharIter != &fSCharIter) { + // old fCharIter was adopted from the outside. Delete it. + delete fCharIter; + } + fCharIter = &fSCharIter; + + this->first(); +} + + +/** + * Provide a new UText for the input text. Must reference text with contents identical + * to the original. + * Intended for use with text data originating in Java (garbage collected) environments + * where the data may be moved in memory at arbitrary times. + */ +RuleBasedBreakIterator &RuleBasedBreakIterator::refreshInputText(UText *input, UErrorCode &status) { + if (U_FAILURE(status)) { + return *this; + } + if (input == nullptr) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return *this; + } + int64_t pos = utext_getNativeIndex(&fText); + // Shallow read-only clone of the new UText into the existing input UText + utext_clone(&fText, input, false, true, &status); + if (U_FAILURE(status)) { + return *this; + } + utext_setNativeIndex(&fText, pos); + if (utext_getNativeIndex(&fText) != pos) { + // Sanity check. The new input utext is supposed to have the exact same + // contents as the old. If we can't set to the same position, it doesn't. + // The contents underlying the old utext might be invalid at this point, + // so it's not safe to check directly. + status = U_ILLEGAL_ARGUMENT_ERROR; + } + return *this; +} + + +/** + * Sets the current iteration position to the beginning of the text, position zero. + * @return The new iterator position, which is zero. + */ +int32_t RuleBasedBreakIterator::first() { + UErrorCode status = U_ZERO_ERROR; + if (!fBreakCache->seek(0)) { + fBreakCache->populateNear(0, status); + } + fBreakCache->current(); + U_ASSERT(fPosition == 0); + return 0; +} + +/** + * Sets the current iteration position to the end of the text. + * @return The text's past-the-end offset. + */ +int32_t RuleBasedBreakIterator::last() { + int32_t endPos = (int32_t)utext_nativeLength(&fText); + UBool endShouldBeBoundary = isBoundary(endPos); // Has side effect of setting iterator position. + (void)endShouldBeBoundary; + U_ASSERT(endShouldBeBoundary); + U_ASSERT(fPosition == endPos); + return endPos; +} + +/** + * Advances the iterator either forward or backward the specified number of steps. + * Negative values move backward, and positive values move forward. This is + * equivalent to repeatedly calling next() or previous(). + * @param n The number of steps to move. The sign indicates the direction + * (negative is backwards, and positive is forwards). + * @return The character offset of the boundary position n boundaries away from + * the current one. + */ +int32_t RuleBasedBreakIterator::next(int32_t n) { + int32_t result = 0; + if (n > 0) { + for (; n > 0 && result != UBRK_DONE; --n) { + result = next(); + } + } else if (n < 0) { + for (; n < 0 && result != UBRK_DONE; ++n) { + result = previous(); + } + } else { + result = current(); + } + return result; +} + +/** + * Advances the iterator to the next boundary position. + * @return The position of the first boundary after this one. + */ +int32_t RuleBasedBreakIterator::next() { + fBreakCache->next(); + return fDone ? UBRK_DONE : fPosition; +} + +/** + * Move the iterator backwards, to the boundary preceding the current one. + * + * Starts from the current position within fText. + * Starting position need not be on a boundary. + * + * @return The position of the boundary position immediately preceding the starting position. + */ +int32_t RuleBasedBreakIterator::previous() { + UErrorCode status = U_ZERO_ERROR; + fBreakCache->previous(status); + return fDone ? UBRK_DONE : fPosition; +} + +/** + * Sets the iterator to refer to the first boundary position following + * the specified position. + * @param startPos The position from which to begin searching for a break position. + * @return The position of the first break after the current position. + */ +int32_t RuleBasedBreakIterator::following(int32_t startPos) { + // if the supplied position is before the beginning, return the + // text's starting offset + if (startPos < 0) { + return first(); + } + + // Move requested offset to a code point start. It might be on a trail surrogate, + // or on a trail byte if the input is UTF-8. Or it may be beyond the end of the text. + utext_setNativeIndex(&fText, startPos); + startPos = (int32_t)utext_getNativeIndex(&fText); + + UErrorCode status = U_ZERO_ERROR; + fBreakCache->following(startPos, status); + return fDone ? UBRK_DONE : fPosition; +} + +/** + * Sets the iterator to refer to the last boundary position before the + * specified position. + * @param offset The position to begin searching for a break from. + * @return The position of the last boundary before the starting position. + */ +int32_t RuleBasedBreakIterator::preceding(int32_t offset) { + if (offset > utext_nativeLength(&fText)) { + return last(); + } + + // Move requested offset to a code point start. It might be on a trail surrogate, + // or on a trail byte if the input is UTF-8. + + utext_setNativeIndex(&fText, offset); + int32_t adjustedOffset = static_cast(utext_getNativeIndex(&fText)); + + UErrorCode status = U_ZERO_ERROR; + fBreakCache->preceding(adjustedOffset, status); + return fDone ? UBRK_DONE : fPosition; +} + +/** + * Returns true if the specified position is a boundary position. As a side + * effect, leaves the iterator pointing to the first boundary position at + * or after "offset". + * + * @param offset the offset to check. + * @return True if "offset" is a boundary position. + */ +UBool RuleBasedBreakIterator::isBoundary(int32_t offset) { + // out-of-range indexes are never boundary positions + if (offset < 0) { + first(); // For side effects on current position, tag values. + return false; + } + + // Adjust offset to be on a code point boundary and not beyond the end of the text. + // Note that isBoundary() is always false for offsets that are not on code point boundaries. + // But we still need the side effect of leaving iteration at the following boundary. + + utext_setNativeIndex(&fText, offset); + int32_t adjustedOffset = static_cast(utext_getNativeIndex(&fText)); + + bool result = false; + UErrorCode status = U_ZERO_ERROR; + if (fBreakCache->seek(adjustedOffset) || fBreakCache->populateNear(adjustedOffset, status)) { + result = (fBreakCache->current() == offset); + } + + if (result && adjustedOffset < offset && utext_char32At(&fText, offset) == U_SENTINEL) { + // Original offset is beyond the end of the text. Return false, it's not a boundary, + // but the iteration position remains set to the end of the text, which is a boundary. + return false; + } + if (!result) { + // Not on a boundary. isBoundary() must leave iterator on the following boundary. + // Cache->seek(), above, left us on the preceding boundary, so advance one. + next(); + } + return result; +} + + +/** + * Returns the current iteration position. + * @return The current iteration position. + */ +int32_t RuleBasedBreakIterator::current() const { + return fPosition; +} + + +//======================================================================= +// implementation +//======================================================================= + +// +// RBBIRunMode - the state machine runs an extra iteration at the beginning and end +// of user text. A variable with this enum type keeps track of where we +// are. The state machine only fetches user input while in the RUN mode. +// +enum RBBIRunMode { + RBBI_START, // state machine processing is before first char of input + RBBI_RUN, // state machine processing is in the user text + RBBI_END // state machine processing is after end of user text. +}; + + +// Wrapper functions to select the appropriate handleNext() or handleSafePrevious() +// instantiation, based on whether an 8 or 16 bit table is required. +// +// These Trie access functions will be inlined within the handleNext()/Previous() instantions. +static inline uint16_t TrieFunc8(const UCPTrie *trie, UChar32 c) { + return UCPTRIE_FAST_GET(trie, UCPTRIE_8, c); +} + +static inline uint16_t TrieFunc16(const UCPTrie *trie, UChar32 c) { + return UCPTRIE_FAST_GET(trie, UCPTRIE_16, c); +} + +int32_t RuleBasedBreakIterator::handleNext() { + const RBBIStateTable *statetable = fData->fForwardTable; + bool use8BitsTrie = ucptrie_getValueWidth(fData->fTrie) == UCPTRIE_VALUE_BITS_8; + if (statetable->fFlags & RBBI_8BITS_ROWS) { + if (use8BitsTrie) { + return handleNext(); + } else { + return handleNext(); + } + } else { + if (use8BitsTrie) { + return handleNext(); + } else { + return handleNext(); + } + } +} + +int32_t RuleBasedBreakIterator::handleSafePrevious(int32_t fromPosition) { + const RBBIStateTable *statetable = fData->fReverseTable; + bool use8BitsTrie = ucptrie_getValueWidth(fData->fTrie) == UCPTRIE_VALUE_BITS_8; + if (statetable->fFlags & RBBI_8BITS_ROWS) { + if (use8BitsTrie) { + return handleSafePrevious(fromPosition); + } else { + return handleSafePrevious(fromPosition); + } + } else { + if (use8BitsTrie) { + return handleSafePrevious(fromPosition); + } else { + return handleSafePrevious(fromPosition); + } + } +} + + +//----------------------------------------------------------------------------------- +// +// handleNext() +// Run the state machine to find a boundary +// +//----------------------------------------------------------------------------------- +template +int32_t RuleBasedBreakIterator::handleNext() { + int32_t state; + uint16_t category = 0; + RBBIRunMode mode; + + RowType *row; + UChar32 c; + int32_t result = 0; + int32_t initialPosition = 0; + const RBBIStateTable *statetable = fData->fForwardTable; + const char *tableData = statetable->fTableData; + uint32_t tableRowLen = statetable->fRowLen; + uint32_t dictStart = statetable->fDictCategoriesStart; + #ifdef RBBI_DEBUG + if (gTrace) { + RBBIDebugPuts("Handle Next pos char state category"); + } + #endif + + // handleNext always sets the break tag value. + // Set the default for it. + fRuleStatusIndex = 0; + + fDictionaryCharCount = 0; + + // if we're already at the end of the text, return DONE. + initialPosition = fPosition; + UTEXT_SETNATIVEINDEX(&fText, initialPosition); + result = initialPosition; + c = UTEXT_NEXT32(&fText); + if (c==U_SENTINEL) { + fDone = true; + return UBRK_DONE; + } + + // Set the initial state for the state machine + state = START_STATE; + row = (RowType *) + //(statetable->fTableData + (statetable->fRowLen * state)); + (tableData + tableRowLen * state); + + + mode = RBBI_RUN; + if (statetable->fFlags & RBBI_BOF_REQUIRED) { + category = 2; + mode = RBBI_START; + } + + + // loop until we reach the end of the text or transition to state 0 + // + for (;;) { + if (c == U_SENTINEL) { + // Reached end of input string. + if (mode == RBBI_END) { + // We have already run the loop one last time with the + // character set to the psueudo {eof} value. Now it is time + // to unconditionally bail out. + break; + } + // Run the loop one last time with the fake end-of-input character category. + mode = RBBI_END; + category = 1; + } + + // + // Get the char category. An incoming category of 1 or 2 means that + // we are preset for doing the beginning or end of input, and + // that we shouldn't get a category from an actual text input character. + // + if (mode == RBBI_RUN) { + // look up the current character's character category, which tells us + // which column in the state table to look at. + category = trieFunc(fData->fTrie, c); + fDictionaryCharCount += (category >= dictStart); + } + + #ifdef RBBI_DEBUG + if (gTrace) { + RBBIDebugPrintf(" %4" PRId64 " ", utext_getNativeIndex(&fText)); + if (0x20<=c && c<0x7f) { + RBBIDebugPrintf("\"%c\" ", c); + } else { + RBBIDebugPrintf("%5x ", c); + } + RBBIDebugPrintf("%3d %3d\n", state, category); + } + #endif + + // State Transition - move machine to its next state + // + + // fNextState is a variable-length array. + U_ASSERT(categoryfHeader->fCatCount); + state = row->fNextState[category]; /*Not accessing beyond memory*/ + row = (RowType *) + // (statetable->fTableData + (statetable->fRowLen * state)); + (tableData + tableRowLen * state); + + + uint16_t accepting = row->fAccepting; + if (accepting == ACCEPTING_UNCONDITIONAL) { + // Match found, common case. + if (mode != RBBI_START) { + result = (int32_t)UTEXT_GETNATIVEINDEX(&fText); + } + fRuleStatusIndex = row->fTagsIdx; // Remember the break status (tag) values. + } else if (accepting > ACCEPTING_UNCONDITIONAL) { + // Lookahead match is completed. + U_ASSERT(accepting < fData->fForwardTable->fLookAheadResultsSize); + int32_t lookaheadResult = fLookAheadMatches[accepting]; + if (lookaheadResult >= 0) { + fRuleStatusIndex = row->fTagsIdx; + fPosition = lookaheadResult; + return lookaheadResult; + } + } + + // If we are at the position of the '/' in a look-ahead (hard break) rule; + // record the current position, to be returned later, if the full rule matches. + // TODO: Move this check before the previous check of fAccepting. + // This would enable hard-break rules with no following context. + // But there are line break test failures when trying this. Investigate. + // Issue ICU-20837 + uint16_t rule = row->fLookAhead; + U_ASSERT(rule == 0 || rule > ACCEPTING_UNCONDITIONAL); + U_ASSERT(rule == 0 || rule < fData->fForwardTable->fLookAheadResultsSize); + if (rule > ACCEPTING_UNCONDITIONAL) { + int32_t pos = (int32_t)UTEXT_GETNATIVEINDEX(&fText); + fLookAheadMatches[rule] = pos; + } + + if (state == STOP_STATE) { + // This is the normal exit from the lookup state machine. + // We have advanced through the string until it is certain that no + // longer match is possible, no matter what characters follow. + break; + } + + // Advance to the next character. + // If this is a beginning-of-input loop iteration, don't advance + // the input position. The next iteration will be processing the + // first real input character. + if (mode == RBBI_RUN) { + c = UTEXT_NEXT32(&fText); + } else { + if (mode == RBBI_START) { + mode = RBBI_RUN; + } + } + } + + // The state machine is done. Check whether it found a match... + + // If the iterator failed to advance in the match engine, force it ahead by one. + // (This really indicates a defect in the break rules. They should always match + // at least one character.) + if (result == initialPosition) { + utext_setNativeIndex(&fText, initialPosition); + utext_next32(&fText); + result = (int32_t)utext_getNativeIndex(&fText); + fRuleStatusIndex = 0; + } + + // Leave the iterator at our result position. + fPosition = result; + #ifdef RBBI_DEBUG + if (gTrace) { + RBBIDebugPrintf("result = %d\n\n", result); + } + #endif + return result; +} + + +//----------------------------------------------------------------------------------- +// +// handleSafePrevious() +// +// Iterate backwards using the safe reverse rules. +// The logic of this function is similar to handleNext(), but simpler +// because the safe table does not require as many options. +// +//----------------------------------------------------------------------------------- +template +int32_t RuleBasedBreakIterator::handleSafePrevious(int32_t fromPosition) { + + int32_t state; + uint16_t category = 0; + RowType *row; + UChar32 c; + int32_t result = 0; + + const RBBIStateTable *stateTable = fData->fReverseTable; + UTEXT_SETNATIVEINDEX(&fText, fromPosition); + #ifdef RBBI_DEBUG + if (gTrace) { + RBBIDebugPuts("Handle Previous pos char state category"); + } + #endif + + // if we're already at the start of the text, return DONE. + if (fData == nullptr || UTEXT_GETNATIVEINDEX(&fText)==0) { + return BreakIterator::DONE; + } + + // Set the initial state for the state machine + c = UTEXT_PREVIOUS32(&fText); + state = START_STATE; + row = (RowType *) + (stateTable->fTableData + (stateTable->fRowLen * state)); + + // loop until we reach the start of the text or transition to state 0 + // + for (; c != U_SENTINEL; c = UTEXT_PREVIOUS32(&fText)) { + + // look up the current character's character category, which tells us + // which column in the state table to look at. + // + // Off the dictionary flag bit. For reverse iteration it is not used. + category = trieFunc(fData->fTrie, c); + + #ifdef RBBI_DEBUG + if (gTrace) { + RBBIDebugPrintf(" %4d ", (int32_t)utext_getNativeIndex(&fText)); + if (0x20<=c && c<0x7f) { + RBBIDebugPrintf("\"%c\" ", c); + } else { + RBBIDebugPrintf("%5x ", c); + } + RBBIDebugPrintf("%3d %3d\n", state, category); + } + #endif + + // State Transition - move machine to its next state + // + // fNextState is a variable-length array. + U_ASSERT(categoryfHeader->fCatCount); + state = row->fNextState[category]; /*Not accessing beyond memory*/ + row = (RowType *) + (stateTable->fTableData + (stateTable->fRowLen * state)); + + if (state == STOP_STATE) { + // This is the normal exit from the lookup state machine. + // Transition to state zero means we have found a safe point. + break; + } + } + + // The state machine is done. Check whether it found a match... + result = (int32_t)UTEXT_GETNATIVEINDEX(&fText); + #ifdef RBBI_DEBUG + if (gTrace) { + RBBIDebugPrintf("result = %d\n\n", result); + } + #endif + return result; +} + + +//------------------------------------------------------------------------------- +// +// getRuleStatus() Return the break rule tag associated with the current +// iterator position. If the iterator arrived at its current +// position by iterating forwards, the value will have been +// cached by the handleNext() function. +// +//------------------------------------------------------------------------------- + +int32_t RuleBasedBreakIterator::getRuleStatus() const { + + // fLastRuleStatusIndex indexes to the start of the appropriate status record + // (the number of status values.) + // This function returns the last (largest) of the array of status values. + int32_t idx = fRuleStatusIndex + fData->fRuleStatusTable[fRuleStatusIndex]; + int32_t tagVal = fData->fRuleStatusTable[idx]; + + return tagVal; +} + + +int32_t RuleBasedBreakIterator::getRuleStatusVec( + int32_t *fillInVec, int32_t capacity, UErrorCode &status) { + if (U_FAILURE(status)) { + return 0; + } + + int32_t numVals = fData->fRuleStatusTable[fRuleStatusIndex]; + int32_t numValsToCopy = numVals; + if (numVals > capacity) { + status = U_BUFFER_OVERFLOW_ERROR; + numValsToCopy = capacity; + } + int i; + for (i=0; ifRuleStatusTable[fRuleStatusIndex + i + 1]; + } + return numVals; +} + + + +//------------------------------------------------------------------------------- +// +// getBinaryRules Access to the compiled form of the rules, +// for use by build system tools that save the data +// for standard iterator types. +// +//------------------------------------------------------------------------------- +const uint8_t *RuleBasedBreakIterator::getBinaryRules(uint32_t &length) { + const uint8_t *retPtr = nullptr; + length = 0; + + if (fData != nullptr) { + retPtr = (const uint8_t *)fData->fHeader; + length = fData->fHeader->fLength; + } + return retPtr; +} + + +RuleBasedBreakIterator *RuleBasedBreakIterator::createBufferClone( + void * /*stackBuffer*/, int32_t &bufferSize, UErrorCode &status) { + if (U_FAILURE(status)){ + return nullptr; + } + + if (bufferSize == 0) { + bufferSize = 1; // preflighting for deprecated functionality + return nullptr; + } + + BreakIterator *clonedBI = clone(); + if (clonedBI == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + } else { + status = U_SAFECLONE_ALLOCATED_WARNING; + } + return (RuleBasedBreakIterator *)clonedBI; +} + +U_NAMESPACE_END + + +static icu::UStack *gLanguageBreakFactories = nullptr; +static const icu::UnicodeString *gEmptyString = nullptr; +static icu::UInitOnce gLanguageBreakFactoriesInitOnce {}; +static icu::UInitOnce gRBBIInitOnce {}; + +/** + * Release all static memory held by breakiterator. + */ +U_CDECL_BEGIN +UBool U_CALLCONV rbbi_cleanup() { + delete gLanguageBreakFactories; + gLanguageBreakFactories = nullptr; + delete gEmptyString; + gEmptyString = nullptr; + gLanguageBreakFactoriesInitOnce.reset(); + gRBBIInitOnce.reset(); + return true; +} +U_CDECL_END + +U_CDECL_BEGIN +static void U_CALLCONV _deleteFactory(void *obj) { + delete (icu::LanguageBreakFactory *) obj; +} +U_CDECL_END +U_NAMESPACE_BEGIN + +static void U_CALLCONV rbbiInit() { + gEmptyString = new UnicodeString(); + ucln_common_registerCleanup(UCLN_COMMON_RBBI, rbbi_cleanup); +} + +static void U_CALLCONV initLanguageFactories() { + UErrorCode status = U_ZERO_ERROR; + U_ASSERT(gLanguageBreakFactories == nullptr); + gLanguageBreakFactories = new UStack(_deleteFactory, nullptr, status); + if (gLanguageBreakFactories != nullptr && U_SUCCESS(status)) { + ICULanguageBreakFactory *builtIn = new ICULanguageBreakFactory(status); + gLanguageBreakFactories->push(builtIn, status); +#ifdef U_LOCAL_SERVICE_HOOK + LanguageBreakFactory *extra = (LanguageBreakFactory *)uprv_svc_hook("languageBreakFactory", &status); + if (extra != nullptr) { + gLanguageBreakFactories->push(extra, status); + } +#endif + } + ucln_common_registerCleanup(UCLN_COMMON_RBBI, rbbi_cleanup); +} + + +static const LanguageBreakEngine* +getLanguageBreakEngineFromFactory(UChar32 c) +{ + umtx_initOnce(gLanguageBreakFactoriesInitOnce, &initLanguageFactories); + if (gLanguageBreakFactories == nullptr) { + return nullptr; + } + + int32_t i = gLanguageBreakFactories->size(); + const LanguageBreakEngine *lbe = nullptr; + while (--i >= 0) { + LanguageBreakFactory *factory = (LanguageBreakFactory *)(gLanguageBreakFactories->elementAt(i)); + lbe = factory->getEngineFor(c); + if (lbe != nullptr) { + break; + } + } + return lbe; +} + + +//------------------------------------------------------------------------------- +// +// getLanguageBreakEngine Find an appropriate LanguageBreakEngine for the +// the character c. +// +//------------------------------------------------------------------------------- +const LanguageBreakEngine * +RuleBasedBreakIterator::getLanguageBreakEngine(UChar32 c) { + const LanguageBreakEngine *lbe = nullptr; + UErrorCode status = U_ZERO_ERROR; + + if (fLanguageBreakEngines == nullptr) { + fLanguageBreakEngines = new UStack(status); + if (fLanguageBreakEngines == nullptr || U_FAILURE(status)) { + delete fLanguageBreakEngines; + fLanguageBreakEngines = 0; + return nullptr; + } + } + + int32_t i = fLanguageBreakEngines->size(); + while (--i >= 0) { + lbe = (const LanguageBreakEngine *)(fLanguageBreakEngines->elementAt(i)); + if (lbe->handles(c)) { + return lbe; + } + } + + // No existing dictionary took the character. See if a factory wants to + // give us a new LanguageBreakEngine for this character. + lbe = getLanguageBreakEngineFromFactory(c); + + // If we got one, use it and push it on our stack. + if (lbe != nullptr) { + fLanguageBreakEngines->push((void *)lbe, status); + // Even if we can't remember it, we can keep looking it up, so + // return it even if the push fails. + return lbe; + } + + // No engine is forthcoming for this character. Add it to the + // reject set. Create the reject break engine if needed. + if (fUnhandledBreakEngine == nullptr) { + fUnhandledBreakEngine = new UnhandledEngine(status); + if (U_SUCCESS(status) && fUnhandledBreakEngine == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + // Put it last so that scripts for which we have an engine get tried + // first. + fLanguageBreakEngines->insertElementAt(fUnhandledBreakEngine, 0, status); + // If we can't insert it, or creation failed, get rid of it + U_ASSERT(!fLanguageBreakEngines->hasDeleter()); + if (U_FAILURE(status)) { + delete fUnhandledBreakEngine; + fUnhandledBreakEngine = 0; + return nullptr; + } + } + + // Tell the reject engine about the character; at its discretion, it may + // add more than just the one character. + fUnhandledBreakEngine->handleCharacter(c); + + return fUnhandledBreakEngine; +} + +void RuleBasedBreakIterator::dumpCache() { + fBreakCache->dumpCache(); +} + +void RuleBasedBreakIterator::dumpTables() { + fData->printData(); +} + +/** + * Returns the description used to create this iterator + */ + +const UnicodeString& +RuleBasedBreakIterator::getRules() const { + if (fData != nullptr) { + return fData->getRuleSourceString(); + } else { + umtx_initOnce(gRBBIInitOnce, &rbbiInit); + return *gEmptyString; + } +} + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ diff --git a/intl/icu/source/common/rbbi_cache.cpp b/intl/icu/source/common/rbbi_cache.cpp new file mode 100644 index 0000000000..02ca555a89 --- /dev/null +++ b/intl/icu/source/common/rbbi_cache.cpp @@ -0,0 +1,701 @@ +// Copyright (C) 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// file: rbbi_cache.cpp + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/ubrk.h" +#include "unicode/rbbi.h" + +#include "rbbi_cache.h" + +#include "brkeng.h" +#include "cmemory.h" +#include "rbbidata.h" +#include "rbbirb.h" +#include "uassert.h" +#include "uvectr32.h" + +U_NAMESPACE_BEGIN + +/* + * DictionaryCache implementation + */ + +RuleBasedBreakIterator::DictionaryCache::DictionaryCache(RuleBasedBreakIterator *bi, UErrorCode &status) : + fBI(bi), fBreaks(status), fPositionInCache(-1), + fStart(0), fLimit(0), fFirstRuleStatusIndex(0), fOtherRuleStatusIndex(0) { +} + +RuleBasedBreakIterator::DictionaryCache::~DictionaryCache() { +} + +void RuleBasedBreakIterator::DictionaryCache::reset() { + fPositionInCache = -1; + fStart = 0; + fLimit = 0; + fFirstRuleStatusIndex = 0; + fOtherRuleStatusIndex = 0; + fBreaks.removeAllElements(); +} + +UBool RuleBasedBreakIterator::DictionaryCache::following(int32_t fromPos, int32_t *result, int32_t *statusIndex) { + if (fromPos >= fLimit || fromPos < fStart) { + fPositionInCache = -1; + return false; + } + + // Sequential iteration, move from previous boundary to the following + + int32_t r = 0; + if (fPositionInCache >= 0 && fPositionInCache < fBreaks.size() && fBreaks.elementAti(fPositionInCache) == fromPos) { + ++fPositionInCache; + if (fPositionInCache >= fBreaks.size()) { + fPositionInCache = -1; + return false; + } + r = fBreaks.elementAti(fPositionInCache); + U_ASSERT(r > fromPos); + *result = r; + *statusIndex = fOtherRuleStatusIndex; + return true; + } + + // Random indexing. Linear search for the boundary following the given position. + + for (fPositionInCache = 0; fPositionInCache < fBreaks.size(); ++fPositionInCache) { + r= fBreaks.elementAti(fPositionInCache); + if (r > fromPos) { + *result = r; + *statusIndex = fOtherRuleStatusIndex; + return true; + } + } + UPRV_UNREACHABLE_EXIT; +} + + +UBool RuleBasedBreakIterator::DictionaryCache::preceding(int32_t fromPos, int32_t *result, int32_t *statusIndex) { + if (fromPos <= fStart || fromPos > fLimit) { + fPositionInCache = -1; + return false; + } + + if (fromPos == fLimit) { + fPositionInCache = fBreaks.size() - 1; + if (fPositionInCache >= 0) { + U_ASSERT(fBreaks.elementAti(fPositionInCache) == fromPos); + } + } + + int32_t r; + if (fPositionInCache > 0 && fPositionInCache < fBreaks.size() && fBreaks.elementAti(fPositionInCache) == fromPos) { + --fPositionInCache; + r = fBreaks.elementAti(fPositionInCache); + U_ASSERT(r < fromPos); + *result = r; + *statusIndex = ( r== fStart) ? fFirstRuleStatusIndex : fOtherRuleStatusIndex; + return true; + } + + if (fPositionInCache == 0) { + fPositionInCache = -1; + return false; + } + + for (fPositionInCache = fBreaks.size()-1; fPositionInCache >= 0; --fPositionInCache) { + r = fBreaks.elementAti(fPositionInCache); + if (r < fromPos) { + *result = r; + *statusIndex = ( r == fStart) ? fFirstRuleStatusIndex : fOtherRuleStatusIndex; + return true; + } + } + UPRV_UNREACHABLE_EXIT; +} + +void RuleBasedBreakIterator::DictionaryCache::populateDictionary(int32_t startPos, int32_t endPos, + int32_t firstRuleStatus, int32_t otherRuleStatus) { + if ((endPos - startPos) <= 1) { + return; + } + + reset(); + fFirstRuleStatusIndex = firstRuleStatus; + fOtherRuleStatusIndex = otherRuleStatus; + + int32_t rangeStart = startPos; + int32_t rangeEnd = endPos; + + uint16_t category; + int32_t current; + UErrorCode status = U_ZERO_ERROR; + int32_t foundBreakCount = 0; + UText *text = &fBI->fText; + + // Loop through the text, looking for ranges of dictionary characters. + // For each span, find the appropriate break engine, and ask it to find + // any breaks within the span. + + utext_setNativeIndex(text, rangeStart); + UChar32 c = utext_current32(text); + category = ucptrie_get(fBI->fData->fTrie, c); + uint32_t dictStart = fBI->fData->fForwardTable->fDictCategoriesStart; + + while(U_SUCCESS(status)) { + while((current = (int32_t)UTEXT_GETNATIVEINDEX(text)) < rangeEnd + && (category < dictStart)) { + utext_next32(text); // TODO: cleaner loop structure. + c = utext_current32(text); + category = ucptrie_get(fBI->fData->fTrie, c); + } + if (current >= rangeEnd) { + break; + } + + // We now have a dictionary character. Get the appropriate language object + // to deal with it. + const LanguageBreakEngine *lbe = fBI->getLanguageBreakEngine(c); + + // Ask the language object if there are any breaks. It will add them to the cache and + // leave the text pointer on the other side of its range, ready to search for the next one. + if (lbe != nullptr) { + foundBreakCount += lbe->findBreaks(text, rangeStart, rangeEnd, fBreaks, fBI->fIsPhraseBreaking, status); + } + + // Reload the loop variables for the next go-round + c = utext_current32(text); + category = ucptrie_get(fBI->fData->fTrie, c); + } + + // If we found breaks, ensure that the first and last entries are + // the original starting and ending position. And initialize the + // cache iteration position to the first entry. + + // printf("foundBreakCount = %d\n", foundBreakCount); + if (foundBreakCount > 0) { + U_ASSERT(foundBreakCount == fBreaks.size()); + if (startPos < fBreaks.elementAti(0)) { + // The dictionary did not place a boundary at the start of the segment of text. + // Add one now. This should not commonly happen, but it would be easy for interactions + // of the rules for dictionary segments and the break engine implementations to + // inadvertently cause it. Cover it here, just in case. + fBreaks.insertElementAt(startPos, 0, status); + } + if (endPos > fBreaks.peeki()) { + fBreaks.push(endPos, status); + } + fPositionInCache = 0; + // Note: Dictionary matching may extend beyond the original limit. + fStart = fBreaks.elementAti(0); + fLimit = fBreaks.peeki(); + } else { + // there were no language-based breaks, even though the segment contained + // dictionary characters. Subsequent attempts to fetch boundaries from the dictionary cache + // for this range will fail, and the calling code will fall back to the rule based boundaries. + } +} + + +/* + * BreakCache implementation + */ + +RuleBasedBreakIterator::BreakCache::BreakCache(RuleBasedBreakIterator *bi, UErrorCode &status) : + fBI(bi), fSideBuffer(status) { + reset(); +} + + +RuleBasedBreakIterator::BreakCache::~BreakCache() { +} + + +void RuleBasedBreakIterator::BreakCache::reset(int32_t pos, int32_t ruleStatus) { + fStartBufIdx = 0; + fEndBufIdx = 0; + fTextIdx = pos; + fBufIdx = 0; + fBoundaries[0] = pos; + fStatuses[0] = (uint16_t)ruleStatus; +} + + +int32_t RuleBasedBreakIterator::BreakCache::current() { + fBI->fPosition = fTextIdx; + fBI->fRuleStatusIndex = fStatuses[fBufIdx]; + fBI->fDone = false; + return fTextIdx; +} + + +void RuleBasedBreakIterator::BreakCache::following(int32_t startPos, UErrorCode &status) { + if (U_FAILURE(status)) { + return; + } + if (startPos == fTextIdx || seek(startPos) || populateNear(startPos, status)) { + // startPos is in the cache. Do a next() from that position. + // TODO: an awkward set of interactions with bi->fDone + // seek() does not clear it; it can't because of interactions with populateNear(). + // next() does not clear it in the fast-path case, where everything matters. Maybe it should. + // So clear it here, for the case where seek() succeeded on an iterator that had previously run off the end. + fBI->fDone = false; + next(); + } + return; +} + + +void RuleBasedBreakIterator::BreakCache::preceding(int32_t startPos, UErrorCode &status) { + if (U_FAILURE(status)) { + return; + } + if (startPos == fTextIdx || seek(startPos) || populateNear(startPos, status)) { + if (startPos == fTextIdx) { + previous(status); + } else { + // seek() leaves the BreakCache positioned at the preceding boundary + // if the requested position is between two boundaries. + // current() pushes the BreakCache position out to the BreakIterator itself. + U_ASSERT(startPos > fTextIdx); + current(); + } + } + return; +} + + +/* + * Out-of-line code for BreakCache::next(). + * Cache does not already contain the boundary + */ +void RuleBasedBreakIterator::BreakCache::nextOL() { + fBI->fDone = !populateFollowing(); + fBI->fPosition = fTextIdx; + fBI->fRuleStatusIndex = fStatuses[fBufIdx]; + return; +} + + +void RuleBasedBreakIterator::BreakCache::previous(UErrorCode &status) { + if (U_FAILURE(status)) { + return; + } + int32_t initialBufIdx = fBufIdx; + if (fBufIdx == fStartBufIdx) { + // At start of cache. Prepend to it. + populatePreceding(status); + } else { + // Cache already holds the next boundary + fBufIdx = modChunkSize(fBufIdx - 1); + fTextIdx = fBoundaries[fBufIdx]; + } + fBI->fDone = (fBufIdx == initialBufIdx); + fBI->fPosition = fTextIdx; + fBI->fRuleStatusIndex = fStatuses[fBufIdx]; + return; +} + + +UBool RuleBasedBreakIterator::BreakCache::seek(int32_t pos) { + if (pos < fBoundaries[fStartBufIdx] || pos > fBoundaries[fEndBufIdx]) { + return false; + } + if (pos == fBoundaries[fStartBufIdx]) { + // Common case: seek(0), from BreakIterator::first() + fBufIdx = fStartBufIdx; + fTextIdx = fBoundaries[fBufIdx]; + return true; + } + if (pos == fBoundaries[fEndBufIdx]) { + fBufIdx = fEndBufIdx; + fTextIdx = fBoundaries[fBufIdx]; + return true; + } + + int32_t min = fStartBufIdx; + int32_t max = fEndBufIdx; + while (min != max) { + int32_t probe = (min + max + (min>max ? CACHE_SIZE : 0)) / 2; + probe = modChunkSize(probe); + if (fBoundaries[probe] > pos) { + max = probe; + } else { + min = modChunkSize(probe + 1); + } + } + U_ASSERT(fBoundaries[max] > pos); + fBufIdx = modChunkSize(max - 1); + fTextIdx = fBoundaries[fBufIdx]; + U_ASSERT(fTextIdx <= pos); + return true; +} + + +UBool RuleBasedBreakIterator::BreakCache::populateNear(int32_t position, UErrorCode &status) { + if (U_FAILURE(status)) { + return false; + } + U_ASSERT(position < fBoundaries[fStartBufIdx] || position > fBoundaries[fEndBufIdx]); + + // Add boundaries to the cache near the specified position. + // The given position need not be a boundary itself. + // The input position must be within the range of the text, and + // on a code point boundary. + // If the requested position is a break boundary, leave the iteration + // position on it. + // If the requested position is not a boundary, leave the iteration + // position on the preceding boundary and include both the + // preceding and following boundaries in the cache. + // Additional boundaries, either preceding or following, may be added + // to the cache as a side effect. + + // If the requested position is not near already cached positions, clear the existing cache, + // find a near-by boundary and begin new cache contents there. + + // Threshold for a text position to be considered near to existing cache contents. + // TODO: See issue ICU-22024 "perf tuning of Cache needed." + // This value is subject to change. See the ticket for more details. + static constexpr int32_t CACHE_NEAR = 15; + + int32_t aBoundary = -1; + int32_t ruleStatusIndex = 0; + bool retainCache = false; + if ((position > fBoundaries[fStartBufIdx] - CACHE_NEAR) && position < (fBoundaries[fEndBufIdx] + CACHE_NEAR)) { + // Requested position is near the existing cache. Retain it. + retainCache = true; + } else if (position <= CACHE_NEAR) { + // Requested position is near the start of the text. Fill cache from start, skipping + // the need to find a safe point. + retainCache = false; + aBoundary = 0; + } else { + // Requested position is not near the existing cache. + // Find a safe point to refill the cache from. + int32_t backupPos = fBI->handleSafePrevious(position); + + if (fBoundaries[fEndBufIdx] < position && fBoundaries[fEndBufIdx] >= (backupPos - CACHE_NEAR)) { + // The requested position is beyond the end of the existing cache, but the + // reverse rules produced a position near or before the cached region. + // Retain the existing cache, and fill from the end of it. + retainCache = true; + } else if (backupPos < CACHE_NEAR) { + // The safe reverse rules moved us to near the start of text. + // Take that (index 0) as the backup boundary, avoiding the complication + // (in the following block) of moving forward from the safe point to a known boundary. + // + // Retain the cache if it begins not too far from the requested position. + aBoundary = 0; + retainCache = (fBoundaries[fStartBufIdx] <= (position + CACHE_NEAR)); + } else { + // The safe reverse rules produced a position that is neither near the existing + // cache, nor near the start of text. + // Advance to the boundary following. + // There is a complication: the safe reverse rules identify pairs of code points + // that are safe. If advancing from the safe point moves forwards by less than + // two code points, we need to advance one more time to ensure that the boundary + // is good, including a correct rules status value. + retainCache = false; + fBI->fPosition = backupPos; + aBoundary = fBI->handleNext(); + if (aBoundary != UBRK_DONE && aBoundary <= backupPos + 4) { + // +4 is a quick test for possibly having advanced only one codepoint. + // Four being the length of the longest potential code point, a supplementary in UTF-8 + utext_setNativeIndex(&fBI->fText, aBoundary); + if (backupPos == utext_getPreviousNativeIndex(&fBI->fText)) { + // The initial handleNext() only advanced by a single code point. Go again. + aBoundary = fBI->handleNext(); // Safe rules identify safe pairs. + } + } + if (aBoundary == UBRK_DONE) { + // Note (Andy Heninger): I don't think this condition can occur, but it's hard + // to prove that it can't. We ran off the end of the string looking a boundary + // following a safe point; choose the end of the string as that boundary. + aBoundary = utext_nativeLength(&fBI->fText); + } + ruleStatusIndex = fBI->fRuleStatusIndex; + } + } + + if (!retainCache) { + U_ASSERT(aBoundary != -1); + reset(aBoundary, ruleStatusIndex); // Reset cache to hold aBoundary as a single starting point. + } + + // Fill in boundaries between existing cache content and the new requested position. + + if (fBoundaries[fEndBufIdx] < position) { + // The last position in the cache precedes the requested position. + // Add following position(s) to the cache. + while (fBoundaries[fEndBufIdx] < position) { + if (!populateFollowing()) { + UPRV_UNREACHABLE_EXIT; + } + } + fBufIdx = fEndBufIdx; // Set iterator position to the end of the buffer. + fTextIdx = fBoundaries[fBufIdx]; // Required because populateFollowing may add extra boundaries. + while (fTextIdx > position) { // Move backwards to a position at or preceding the requested pos. + previous(status); + } + return true; + } + + if (fBoundaries[fStartBufIdx] > position) { + // The first position in the cache is beyond the requested position. + // back up more until we get a boundary <= the requested position. + while (fBoundaries[fStartBufIdx] > position) { + populatePreceding(status); + } + fBufIdx = fStartBufIdx; // Set iterator position to the start of the buffer. + fTextIdx = fBoundaries[fBufIdx]; // Required because populatePreceding may add extra boundaries. + while (fTextIdx < position) { // Move forwards to a position at or following the requested pos. + next(); + } + if (fTextIdx > position) { + // If position is not itself a boundary, the next() loop above will overshoot. + // Back up one, leaving cache position at the boundary preceding the requested position. + previous(status); + } + return true; + } + + U_ASSERT(fTextIdx == position); + return true; +} + + + +UBool RuleBasedBreakIterator::BreakCache::populateFollowing() { + int32_t fromPosition = fBoundaries[fEndBufIdx]; + int32_t fromRuleStatusIdx = fStatuses[fEndBufIdx]; + int32_t pos = 0; + int32_t ruleStatusIdx = 0; + + if (fBI->fDictionaryCache->following(fromPosition, &pos, &ruleStatusIdx)) { + addFollowing(pos, ruleStatusIdx, UpdateCachePosition); + return true; + } + + fBI->fPosition = fromPosition; + pos = fBI->handleNext(); + if (pos == UBRK_DONE) { + return false; + } + + ruleStatusIdx = fBI->fRuleStatusIndex; + if (fBI->fDictionaryCharCount > 0) { + // The text segment obtained from the rules includes dictionary characters. + // Subdivide it, with subdivided results going into the dictionary cache. + fBI->fDictionaryCache->populateDictionary(fromPosition, pos, fromRuleStatusIdx, ruleStatusIdx); + if (fBI->fDictionaryCache->following(fromPosition, &pos, &ruleStatusIdx)) { + addFollowing(pos, ruleStatusIdx, UpdateCachePosition); + return true; + // TODO: may want to move a sizable chunk of dictionary cache to break cache at this point. + // But be careful with interactions with populateNear(). + } + } + + // Rule based segment did not include dictionary characters. + // Or, it did contain dictionary chars, but the dictionary segmenter didn't handle them, + // meaning that we didn't take the return, above. + // Add its end point to the cache. + addFollowing(pos, ruleStatusIdx, UpdateCachePosition); + + // Add several non-dictionary boundaries at this point, to optimize straight forward iteration. + // (subsequent calls to BreakIterator::next() will take the fast path, getting cached results. + // + for (int count=0; count<6; ++count) { + pos = fBI->handleNext(); + if (pos == UBRK_DONE || fBI->fDictionaryCharCount > 0) { + break; + } + addFollowing(pos, fBI->fRuleStatusIndex, RetainCachePosition); + } + + return true; +} + + +UBool RuleBasedBreakIterator::BreakCache::populatePreceding(UErrorCode &status) { + if (U_FAILURE(status)) { + return false; + } + + int32_t fromPosition = fBoundaries[fStartBufIdx]; + if (fromPosition == 0) { + return false; + } + + int32_t position = 0; + int32_t positionStatusIdx = 0; + + if (fBI->fDictionaryCache->preceding(fromPosition, &position, &positionStatusIdx)) { + addPreceding(position, positionStatusIdx, UpdateCachePosition); + return true; + } + + int32_t backupPosition = fromPosition; + + // Find a boundary somewhere preceding the first already-cached boundary + do { + backupPosition = backupPosition - 30; + if (backupPosition <= 0) { + backupPosition = 0; + } else { + backupPosition = fBI->handleSafePrevious(backupPosition); + } + if (backupPosition == UBRK_DONE || backupPosition == 0) { + position = 0; + positionStatusIdx = 0; + } else { + // Advance to the boundary following the backup position. + // There is a complication: the safe reverse rules identify pairs of code points + // that are safe. If advancing from the safe point moves forwards by less than + // two code points, we need to advance one more time to ensure that the boundary + // is good, including a correct rules status value. + // + fBI->fPosition = backupPosition; + position = fBI->handleNext(); + if (position <= backupPosition + 4) { + // +4 is a quick test for possibly having advanced only one codepoint. + // Four being the length of the longest potential code point, a supplementary in UTF-8 + utext_setNativeIndex(&fBI->fText, position); + if (backupPosition == utext_getPreviousNativeIndex(&fBI->fText)) { + // The initial handleNext() only advanced by a single code point. Go again. + position = fBI->handleNext(); // Safe rules identify safe pairs. + } + } + positionStatusIdx = fBI->fRuleStatusIndex; + } + } while (position >= fromPosition); + + // Find boundaries between the one we just located and the first already-cached boundary + // Put them in a side buffer, because we don't yet know where they will fall in the circular cache buffer.. + + fSideBuffer.removeAllElements(); + fSideBuffer.addElement(position, status); + fSideBuffer.addElement(positionStatusIdx, status); + + do { + int32_t prevPosition = fBI->fPosition = position; + int32_t prevStatusIdx = positionStatusIdx; + position = fBI->handleNext(); + positionStatusIdx = fBI->fRuleStatusIndex; + if (position == UBRK_DONE) { + break; + } + + UBool segmentHandledByDictionary = false; + if (fBI->fDictionaryCharCount != 0) { + // Segment from the rules includes dictionary characters. + // Subdivide it, with subdivided results going into the dictionary cache. + int32_t dictSegEndPosition = position; + fBI->fDictionaryCache->populateDictionary(prevPosition, dictSegEndPosition, prevStatusIdx, positionStatusIdx); + while (fBI->fDictionaryCache->following(prevPosition, &position, &positionStatusIdx)) { + segmentHandledByDictionary = true; + U_ASSERT(position > prevPosition); + if (position >= fromPosition) { + break; + } + U_ASSERT(position <= dictSegEndPosition); + fSideBuffer.addElement(position, status); + fSideBuffer.addElement(positionStatusIdx, status); + prevPosition = position; + } + U_ASSERT(position==dictSegEndPosition || position>=fromPosition); + } + + if (!segmentHandledByDictionary && position < fromPosition) { + fSideBuffer.addElement(position, status); + fSideBuffer.addElement(positionStatusIdx, status); + } + } while (position < fromPosition); + + // Move boundaries from the side buffer to the main circular buffer. + UBool success = false; + if (!fSideBuffer.isEmpty()) { + positionStatusIdx = fSideBuffer.popi(); + position = fSideBuffer.popi(); + addPreceding(position, positionStatusIdx, UpdateCachePosition); + success = true; + } + + while (!fSideBuffer.isEmpty()) { + positionStatusIdx = fSideBuffer.popi(); + position = fSideBuffer.popi(); + if (!addPreceding(position, positionStatusIdx, RetainCachePosition)) { + // No space in circular buffer to hold a new preceding result while + // also retaining the current cache (iteration) position. + // Bailing out is safe; the cache will refill again if needed. + break; + } + } + + return success; +} + + +void RuleBasedBreakIterator::BreakCache::addFollowing(int32_t position, int32_t ruleStatusIdx, UpdatePositionValues update) { + U_ASSERT(position > fBoundaries[fEndBufIdx]); + U_ASSERT(ruleStatusIdx <= UINT16_MAX); + int32_t nextIdx = modChunkSize(fEndBufIdx + 1); + if (nextIdx == fStartBufIdx) { + fStartBufIdx = modChunkSize(fStartBufIdx + 6); // TODO: experiment. Probably revert to 1. + } + fBoundaries[nextIdx] = position; + fStatuses[nextIdx] = static_cast(ruleStatusIdx); + fEndBufIdx = nextIdx; + if (update == UpdateCachePosition) { + // Set current position to the newly added boundary. + fBufIdx = nextIdx; + fTextIdx = position; + } else { + // Retaining the original cache position. + // Check if the added boundary wraps around the buffer, and would over-write the original position. + // It's the responsibility of callers of this function to not add too many. + U_ASSERT(nextIdx != fBufIdx); + } +} + +bool RuleBasedBreakIterator::BreakCache::addPreceding(int32_t position, int32_t ruleStatusIdx, UpdatePositionValues update) { + U_ASSERT(position < fBoundaries[fStartBufIdx]); + U_ASSERT(ruleStatusIdx <= UINT16_MAX); + int32_t nextIdx = modChunkSize(fStartBufIdx - 1); + if (nextIdx == fEndBufIdx) { + if (fBufIdx == fEndBufIdx && update == RetainCachePosition) { + // Failure. The insertion of the new boundary would claim the buffer position that is the + // current iteration position. And we also want to retain the current iteration position. + // (The buffer is already completely full of entries that precede the iteration position.) + return false; + } + fEndBufIdx = modChunkSize(fEndBufIdx - 1); + } + fBoundaries[nextIdx] = position; + fStatuses[nextIdx] = static_cast(ruleStatusIdx); + fStartBufIdx = nextIdx; + if (update == UpdateCachePosition) { + fBufIdx = nextIdx; + fTextIdx = position; + } + return true; +} + + +void RuleBasedBreakIterator::BreakCache::dumpCache() { +#ifdef RBBI_DEBUG + RBBIDebugPrintf("fTextIdx:%d fBufIdx:%d\n", fTextIdx, fBufIdx); + for (int32_t i=fStartBufIdx; ; i=modChunkSize(i+1)) { + RBBIDebugPrintf("%d %d\n", i, fBoundaries[i]); + if (i == fEndBufIdx) { + break; + } + } +#endif +} + +U_NAMESPACE_END + +#endif // #if !UCONFIG_NO_BREAK_ITERATION diff --git a/intl/icu/source/common/rbbi_cache.h b/intl/icu/source/common/rbbi_cache.h new file mode 100644 index 0000000000..597312e85c --- /dev/null +++ b/intl/icu/source/common/rbbi_cache.h @@ -0,0 +1,203 @@ +// Copyright (C) 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// file: rbbi_cache.h +// +#ifndef RBBI_CACHE_H +#define RBBI_CACHE_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/rbbi.h" +#include "unicode/uobject.h" + +#include "uvectr32.h" + +U_NAMESPACE_BEGIN + +/* DictionaryCache stores the boundaries obtained from a run of dictionary characters. + * Dictionary boundaries are moved first to this cache, then from here + * to the main BreakCache, where they may inter-leave with non-dictionary + * boundaries. The public BreakIterator API always fetches directly + * from the main BreakCache, not from here. + * + * In common situations, the number of boundaries in a single dictionary run + * should be quite small, it will be terminated by punctuation, spaces, + * or any other non-dictionary characters. The main BreakCache may end + * up with boundaries from multiple dictionary based runs. + * + * The boundaries are stored in a simple ArrayList (vector), with the + * assumption that they will be accessed sequentially. + */ +class RuleBasedBreakIterator::DictionaryCache: public UMemory { + public: + DictionaryCache(RuleBasedBreakIterator *bi, UErrorCode &status); + ~DictionaryCache(); + + void reset(); + + UBool following(int32_t fromPos, int32_t *pos, int32_t *statusIndex); + UBool preceding(int32_t fromPos, int32_t *pos, int32_t *statusIndex); + + /** + * Populate the cache with the dictionary based boundaries within a region of text. + * @param startPos The start position of a range of text + * @param endPos The end position of a range of text + * @param firstRuleStatus The rule status index that applies to the break at startPos + * @param otherRuleStatus The rule status index that applies to boundaries other than startPos + * @internal + */ + void populateDictionary(int32_t startPos, int32_t endPos, + int32_t firstRuleStatus, int32_t otherRuleStatus); + + + + RuleBasedBreakIterator *fBI; + + UVector32 fBreaks; // A vector containing the boundaries. + int32_t fPositionInCache; // Index in fBreaks of last boundary returned by following() + // or preceding(). Optimizes sequential access. + int32_t fStart; // Text position of first boundary in cache. + int32_t fLimit; // Last boundary in cache. Which is the limit of the + // text segment being handled by the dictionary. + int32_t fFirstRuleStatusIndex; // Rule status info for first boundary. + int32_t fOtherRuleStatusIndex; // Rule status info for 2nd through last boundaries. +}; + + +/* + * class BreakCache + * + * Cache of break boundary positions and rule status values. + * Break iterator API functions, next(), previous(), etc., will use cached results + * when possible, and otherwise cache new results as they are obtained. + * + * Uniformly caches both dictionary and rule based (non-dictionary) boundaries. + * + * The cache is implemented as a single circular buffer. + */ + +/* + * size of the circular cache buffer. + */ + +class RuleBasedBreakIterator::BreakCache: public UMemory { + public: + BreakCache(RuleBasedBreakIterator *bi, UErrorCode &status); + virtual ~BreakCache(); + void reset(int32_t pos = 0, int32_t ruleStatus = 0); + void next() { if (fBufIdx == fEndBufIdx) { + nextOL(); + } else { + fBufIdx = modChunkSize(fBufIdx + 1); + fTextIdx = fBI->fPosition = fBoundaries[fBufIdx]; + fBI->fRuleStatusIndex = fStatuses[fBufIdx]; + } + } + + + void nextOL(); + void previous(UErrorCode &status); + + // Move the iteration state to the position following the startPosition. + // Input position must be pinned to the input length. + void following(int32_t startPosition, UErrorCode &status); + + void preceding(int32_t startPosition, UErrorCode &status); + + /* + * Update the state of the public BreakIterator (fBI) to reflect the + * current state of the break iterator cache (this). + */ + int32_t current(); + + /** + * Add boundaries to the cache near the specified position. + * The given position need not be a boundary itself. + * The input position must be within the range of the text, and + * on a code point boundary. + * If the requested position is a break boundary, leave the iteration + * position on it. + * If the requested position is not a boundary, leave the iteration + * position on the preceding boundary and include both the + * preceding and following boundaries in the cache. + * Additional boundaries, either preceding or following, may be added + * to the cache as a side effect. + * + * Return false if the operation failed. + */ + UBool populateNear(int32_t position, UErrorCode &status); + + /** + * Add boundary(s) to the cache following the current last boundary. + * Return false if at the end of the text, and no more boundaries can be added. + * Leave iteration position at the first newly added boundary, or unchanged if no boundary was added. + */ + UBool populateFollowing(); + + /** + * Add one or more boundaries to the cache preceding the first currently cached boundary. + * Leave the iteration position on the first added boundary. + * Return false if no boundaries could be added (if at the start of the text.) + */ + UBool populatePreceding(UErrorCode &status); + + enum UpdatePositionValues { + RetainCachePosition = 0, + UpdateCachePosition = 1 + }; + + /* + * Add the boundary following the current position. + * The current position can be left as it was, or changed to the newly added boundary, + * as specified by the update parameter. + */ + void addFollowing(int32_t position, int32_t ruleStatusIdx, UpdatePositionValues update); + + + /* + * Add the boundary preceding the current position. + * The current position can be left as it was, or changed to the newly added boundary, + * as specified by the update parameter. + */ + bool addPreceding(int32_t position, int32_t ruleStatusIdx, UpdatePositionValues update); + + /** + * Set the cache position to the specified position, or, if the position + * falls between to cached boundaries, to the preceding boundary. + * Fails if the requested position is outside of the range of boundaries currently held by the cache. + * The startPosition must be on a code point boundary. + * + * Return true if successful, false if the specified position is after + * the last cached boundary or before the first. + */ + UBool seek(int32_t startPosition); + + void dumpCache(); + + private: + static inline int32_t modChunkSize(int index) { return index & (CACHE_SIZE - 1); } + + static constexpr int32_t CACHE_SIZE = 128; + static_assert((CACHE_SIZE & (CACHE_SIZE-1)) == 0, "CACHE_SIZE must be power of two."); + + RuleBasedBreakIterator *fBI; + int32_t fStartBufIdx; + int32_t fEndBufIdx; // inclusive + + int32_t fTextIdx; + int32_t fBufIdx; + + int32_t fBoundaries[CACHE_SIZE]; + uint16_t fStatuses[CACHE_SIZE]; + + UVector32 fSideBuffer; +}; + +U_NAMESPACE_END + +#endif // #if !UCONFIG_NO_BREAK_ITERATION + +#endif // RBBI_CACHE_H diff --git a/intl/icu/source/common/rbbicst.pl b/intl/icu/source/common/rbbicst.pl new file mode 100755 index 0000000000..65907b0f63 --- /dev/null +++ b/intl/icu/source/common/rbbicst.pl @@ -0,0 +1,459 @@ +#************************************************************************** +# Copyright (C) 2016 and later: Unicode, Inc. and others. +# License & terms of use: http://www.unicode.org/copyright.html +#************************************************************************** +#************************************************************************** +# Copyright (C) 2002-2016 International Business Machines Corporation +# and others. All rights reserved. +#************************************************************************** +# +# rbbicst Compile the RBBI rule parser state table data into initialized C data. +# Usage: +# cd icu/source/common +# perl rbbicst.pl < rbbirpt.txt > rbbirpt.h +# perl rbbicst.pl -j < rbbirpt.txt > RBBIRuleParseTable.java +# +# The output file, rbbrpt.h, is included by some of the .cpp rbbi +# implementation files. This perl script is NOT run as part +# of a normal ICU build. It is run by hand when needed, and the +# rbbirpt.h generated file is put back into cvs. +# +# See rbbirpt.txt for a description of the input format for this script. +# + +if ($ARGV[0] eq "-j") { + $javaOutput = 1; + shift @ARGV; +} + + +$num_states = 1; # Always the state number for the line being compiled. +$line_num = 0; # The line number in the input file. + +$states{"pop"} = 255; # Add the "pop" to the list of defined state names. + # This prevents any state from being labelled with "pop", + # and resolves references to "pop" in the next state field. + +line_loop: while (<>) { + chomp(); + $line = $_; + @fields = split(); + $line_num++; + + # Remove # comments, which are any fields beginning with a #, plus all + # that follow on the line. + for ($i=0; $i<@fields; $i++) { + if ($fields[$i] =~ /^#/) { + @fields = @fields[0 .. $i-1]; + last; + } + } + # ignore blank lines, and those with no fields left after stripping comments.. + if (@fields == 0) { + next; + } + + # + # State Label: handling. + # Does the first token end with a ":"? If so, it's the name of a state. + # Put in a hash, together with the current state number, + # so that we can later look up the number from the name. + # + if (@fields[0] =~ /.*:$/) { + $state_name = @fields[0]; + $state_name =~ s/://; # strip off the colon from the state name. + + if ($states{$state_name} != 0) { + print " rbbicst: at line $line-num duplicate definition of state $state_name\n"; + } + $states{$state_name} = $num_states; + $stateNames[$num_states] = $state_name; + + # if the label was the only thing on this line, go on to the next line, + # otherwise assume that a state definition is on the same line and fall through. + if (@fields == 1) { + next line_loop; + } + shift @fields; # shift off label field in preparation + # for handling the rest of the line. + } + + # + # State Transition line. + # syntax is this, + # character [n] target-state [^push-state] [function-name] + # where + # [something] is an optional something + # character is either a single quoted character e.g. '[' + # or a name of a character class, e.g. white_space + # + + $state_line_num[$num_states] = $line_num; # remember line number with each state + # so we can make better error messages later. + # + # First field, character class or literal character for this transition. + # + if ($fields[0] =~ /^'.'$/) { + # We've got a quoted literal character. + $state_literal_chars[$num_states] = $fields[0]; + $state_literal_chars[$num_states] =~ s/'//g; + } else { + # We've got the name of a character class. + $state_char_class[$num_states] = $fields[0]; + if ($fields[0] =~ /[\W]/) { + print " rbbicsts: at line $line_num, bad character literal or character class name.\n"; + print " scanning $fields[0]\n"; + exit(-1); + } + } + shift @fields; + + # + # do the 'n' flag + # + $state_flag[$num_states] = "false"; + if ($fields[0] eq "n") { + $state_flag[$num_states] = "true"; + shift @fields; + } + + # + # do the destination state. + # + $state_dest_state[$num_states] = $fields[0]; + if ($fields[0] eq "") { + print " rbbicsts: at line $line_num, destination state missing.\n"; + exit(-1); + } + shift @fields; + + # + # do the push state, if present. + # + if ($fields[0] =~ /^\^/) { + $fields[0] =~ s/^\^//; + $state_push_state[$num_states] = $fields[0]; + if ($fields[0] eq "" ) { + print " rbbicsts: at line $line_num, expected state after ^ (no spaces).\n"; + exit(-1); + } + shift @fields; + } + + # + # Lastly, do the optional action name. + # + if ($fields[0] ne "") { + $state_func_name[$num_states] = $fields[0]; + shift @fields; + } + + # + # There should be no fields left on the line at this point. + # + if (@fields > 0) { + print " rbbicsts: at line $line_num, unexpected extra stuff on input line.\n"; + print " scanning $fields[0]\n"; + } + $num_states++; +} + +# +# We've read in the whole file, now go back and output the +# C source code for the state transition table. +# +# We read all states first, before writing anything, so that the state numbers +# for the destination states are all available to be written. +# + +# +# Make hashes for the names of the character classes and +# for the names of the actions that appeared. +# +for ($state=1; $state < $num_states; $state++) { + if ($state_char_class[$state] ne "") { + if ($charClasses{$state_char_class[$state]} == 0) { + $charClasses{$state_char_class[$state]} = 1; + } + } + if ($state_func_name[$state] eq "") { + $state_func_name[$state] = "doNOP"; + } + if ($actions{$state_action_name[$state]} == 0) { + $actions{$state_func_name[$state]} = 1; + } +} + +# +# Check that all of the destination states have been defined +# +# +$states{"exit"} = 0; # Predefined state name, terminates state machine. +for ($state=1; $state<$num_states; $state++) { + if ($states{$state_dest_state[$state]} == 0 && $state_dest_state[$state] ne "exit") { + print "Error at line $state_line_num[$state]: target state \"$state_dest_state[$state]\" is not defined.\n"; + $errors++; + } + if ($state_push_state[$state] ne "" && $states{$state_push_state[$state]} == 0) { + print "Error at line $state_line_num[$state]: target state \"$state_push_state[$state]\" is not defined.\n"; + $errors++; + } +} + +die if ($errors>0); + +# +# Assign numbers to each of the character classes classes used. +# Sets are numbered from 128 - 250 +# The values 0-127 in the state table are used for matching +# individual ASCII characters (the only thing that can appear in the rules.) +# The "set" names appearing in the code below (default, etc.) need special +# handling because they do not correspond to a normal set of characters, +# but trigger special handling by code in the state machine. +# +$i = 128; +foreach $setName (sort keys %charClasses) { + if ($setName eq "default") { + $charClasses{$setName} = 255;} + elsif ($setName eq "escaped") { + $charClasses{$setName} = 254;} + elsif ($setName eq "escapedP") { + $charClasses{$setName} = 253;} + elsif ($setName eq "eof") { + $charClasses{$setName} = 252;} + else { + # Normal (single) character class. Number them. + $charClasses{$setName} = $i; + $i++; + } +} + + +my ($sec, $min, $hour, , $day, $mon, $year, $wday, $yday, $isdst) = localtime; +$year += 1900; + +if ($javaOutput) { + print "/*\n"; + print " *******************************************************************************\n"; + print " * Copyright (C) 2003-$year,\n"; + print " * International Business Machines Corporation and others. All Rights Reserved.\n"; + print " *******************************************************************************\n"; + print " */\n"; + print " \n"; + print "package com.ibm.icu.text;\n"; + print " \n"; + print "/**\n"; + print " * Generated Java File. Do not edit by hand.\n"; + print " * This file contains the state table for the ICU Rule Based Break Iterator\n"; + print " * rule parser.\n"; + print " * It is generated by the Perl script \"rbbicst.pl\" from\n"; + print " * the rule parser state definitions file \"rbbirpt.txt\".\n"; + print " * \@internal \n"; + print " *\n"; + print " */\n"; + + print "class RBBIRuleParseTable\n"; + print "{\n"; + + # + # Emit the constants for the actions to be performed. + # + $n = 1; + foreach $act (sort keys %actions) { + print " static final short $act = $n;\n"; + $n++; + } + print " \n"; + + # + # Emit constants for char class names + # + foreach $setName (sort keys %charClasses) { + print " static final short kRuleSet_$setName = $charClasses{$setName};\n"; + } + print "\n\n"; + + + print " static class RBBIRuleTableElement { \n"; + print " short fAction; \n"; + print " short fCharClass; \n"; + print " short fNextState; \n"; + print " short fPushState; \n"; + print " boolean fNextChar; \n"; + print " String fStateName; \n"; + print " RBBIRuleTableElement(short a, int cc, int ns, int ps, boolean nc, String sn) { \n"; + print " fAction = a; \n"; + print " fCharClass = (short)cc; \n"; + print " fNextState = (short)ns; \n"; + print " fPushState = (short)ps; \n"; + print " fNextChar = nc; \n"; + print " fStateName = sn; \n"; + print " } \n"; + print " }; \n"; + print " \n"; + + + print " static RBBIRuleTableElement[] gRuleParseStateTable = { \n "; + print " new RBBIRuleTableElement(doNOP, 0, 0,0, true, null ) // 0 \n"; #output the unused state 0. + for ($state=1; $state < $num_states; $state++) { + print " , new RBBIRuleTableElement($state_func_name[$state],"; + if ($state_literal_chars[$state] ne "") { + $c = $state_literal_chars[$state]; + print("'$c', "); + }else { + print " $charClasses{$state_char_class[$state]},"; + } + print " $states{$state_dest_state[$state]},"; + + # The push-state field is optional. If omitted, fill field with a zero, which flags + # the state machine that there is no push state. + if ($state_push_state[$state] eq "") { + print "0, "; + } else { + print " $states{$state_push_state[$state]},"; + } + print " $state_flag[$state], "; + + # if this is the first row of the table for this state, put out the state name. + if ($stateNames[$state] ne "") { + print " \"$stateNames[$state]\") "; + } else { + print " null ) "; + } + + # Put out a comment showing the number (index) of this state row, + print " // $state "; + print "\n"; + } + print " };\n"; + + print "}; \n"; + +} +else +{ + # + # C++ Output ... + # + + + print "//---------------------------------------------------------------------------------\n"; + print "//\n"; + print "// Generated Header File. Do not edit by hand.\n"; + print "// This file contains the state table for the ICU Rule Based Break Iterator\n"; + print "// rule parser.\n"; + print "// It is generated by the Perl script \"rbbicst.pl\" from\n"; + print "// the rule parser state definitions file \"rbbirpt.txt\".\n"; + print "//\n"; + print "// Copyright (C) 2002-$year International Business Machines Corporation \n"; + print "// and others. All rights reserved. \n"; + print "//\n"; + print "//---------------------------------------------------------------------------------\n"; + print "#ifndef RBBIRPT_H\n"; + print "#define RBBIRPT_H\n"; + print "\n"; + print "#include \"unicode/utypes.h\"\n"; + print "\n"; + print "U_NAMESPACE_BEGIN\n"; + + # + # Emit the constants for indices of Unicode Sets + # Define one constant for each of the character classes encountered. + # At the same time, store the index corresponding to the set name back into hash. + # + print "//\n"; + print "// Character classes for RBBI rule scanning.\n"; + print "//\n"; + foreach $setName (sort keys %charClasses) { + if ($charClasses{$setName} < 250) { + # Normal character class. + print " static const uint8_t kRuleSet_$setName = $charClasses{$setName};\n"; + } + } + print "\n\n"; + + # + # Emit the enum for the actions to be performed. + # + print "enum RBBI_RuleParseAction {\n"; + foreach $act (sort keys %actions) { + print " $act,\n"; + } + print " rbbiLastAction};\n\n"; + + # + # Emit the struct definition for transition table elements. + # + print "//-------------------------------------------------------------------------------\n"; + print "//\n"; + print "// RBBIRuleTableEl represents the structure of a row in the transition table\n"; + print "// for the rule parser state machine.\n"; + print "//-------------------------------------------------------------------------------\n"; + print "struct RBBIRuleTableEl {\n"; + print " RBBI_RuleParseAction fAction;\n"; + print " uint8_t fCharClass; // 0-127: an individual ASCII character\n"; + print " // 128-255: character class index\n"; + print " uint8_t fNextState; // 0-250: normal next-stat numbers\n"; + print " // 255: pop next-state from stack.\n"; + print " uint8_t fPushState;\n"; + print " UBool fNextChar;\n"; + print "};\n\n"; + + # + # emit the state transition table + # + print "static const struct RBBIRuleTableEl gRuleParseStateTable[] = {\n"; + print " {doNOP, 0, 0, 0, true}\n"; # State 0 is a dummy. Real states start with index = 1. + for ($state=1; $state < $num_states; $state++) { + print " , {$state_func_name[$state],"; + if ($state_literal_chars[$state] ne "") { + $c = $state_literal_chars[$state]; + printf(" %d /* $c */,", ord($c)); # use numeric value, so EBCDIC machines are ok. + }else { + print " $charClasses{$state_char_class[$state]},"; + } + print " $states{$state_dest_state[$state]},"; + + # The push-state field is optional. If omitted, fill field with a zero, which flags + # the state machine that there is no push state. + if ($state_push_state[$state] eq "") { + print "0, "; + } else { + print " $states{$state_push_state[$state]},"; + } + print " $state_flag[$state]} "; + + # Put out a C++ comment showing the number (index) of this state row, + # and, if this is the first row of the table for this state, the state name. + print " // $state "; + if ($stateNames[$state] ne "") { + print " $stateNames[$state]"; + } + print "\n"; + }; + print " };\n"; + + + # + # emit a mapping array from state numbers to state names. + # + # This array is used for producing debugging output from the rule parser. + # + print "#ifdef RBBI_DEBUG\n"; + print "static const char * const RBBIRuleStateNames[] = {"; + for ($state=0; $state<$num_states; $state++) { + if ($stateNames[$state] ne "") { + print " \"$stateNames[$state]\",\n"; + } else { + print " 0,\n"; + } + } + print " 0};\n"; + print "#endif\n\n"; + + print "U_NAMESPACE_END\n"; + print "#endif\n"; +} + + + diff --git a/intl/icu/source/common/rbbidata.cpp b/intl/icu/source/common/rbbidata.cpp new file mode 100644 index 0000000000..49603b6193 --- /dev/null +++ b/intl/icu/source/common/rbbidata.cpp @@ -0,0 +1,476 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +*************************************************************************** +* Copyright (C) 1999-2014 International Business Machines Corporation * +* and others. All rights reserved. * +*************************************************************************** +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/ucptrie.h" +#include "unicode/utypes.h" +#include "rbbidata.h" +#include "rbbirb.h" +#include "udatamem.h" +#include "cmemory.h" +#include "cstring.h" +#include "umutex.h" + +#include "uassert.h" + + +U_NAMESPACE_BEGIN + +//----------------------------------------------------------------------------- +// +// Constructors. +// +//----------------------------------------------------------------------------- +RBBIDataWrapper::RBBIDataWrapper(const RBBIDataHeader *data, UErrorCode &status) { + init0(); + init(data, status); +} + +RBBIDataWrapper::RBBIDataWrapper(const RBBIDataHeader *data, enum EDontAdopt, UErrorCode &status) { + init0(); + init(data, status); + fDontFreeData = true; +} + +RBBIDataWrapper::RBBIDataWrapper(UDataMemory* udm, UErrorCode &status) { + init0(); + if (U_FAILURE(status)) { + return; + } + const DataHeader *dh = udm->pHeader; + int32_t headerSize = dh->dataHeader.headerSize; + if ( !(headerSize >= 20 && + dh->info.isBigEndian == U_IS_BIG_ENDIAN && + dh->info.charsetFamily == U_CHARSET_FAMILY && + dh->info.dataFormat[0] == 0x42 && // dataFormat="Brk " + dh->info.dataFormat[1] == 0x72 && + dh->info.dataFormat[2] == 0x6b && + dh->info.dataFormat[3] == 0x20 && + isDataVersionAcceptable(dh->info.formatVersion)) + ) { + status = U_INVALID_FORMAT_ERROR; + return; + } + const char *dataAsBytes = reinterpret_cast(dh); + const RBBIDataHeader *rbbidh = reinterpret_cast(dataAsBytes + headerSize); + init(rbbidh, status); + fUDataMem = udm; +} + +UBool RBBIDataWrapper::isDataVersionAcceptable(const UVersionInfo version) { + return RBBI_DATA_FORMAT_VERSION[0] == version[0]; +} + + +//----------------------------------------------------------------------------- +// +// init(). Does most of the work of construction, shared between the +// constructors. +// +//----------------------------------------------------------------------------- +void RBBIDataWrapper::init0() { + fHeader = nullptr; + fForwardTable = nullptr; + fReverseTable = nullptr; + fRuleSource = nullptr; + fRuleStatusTable = nullptr; + fTrie = nullptr; + fUDataMem = nullptr; + fRefCount = 0; + fDontFreeData = true; +} + +void RBBIDataWrapper::init(const RBBIDataHeader *data, UErrorCode &status) { + if (U_FAILURE(status)) { + return; + } + fHeader = data; + if (fHeader->fMagic != 0xb1a0 || !isDataVersionAcceptable(fHeader->fFormatVersion)) { + status = U_INVALID_FORMAT_ERROR; + return; + } + // Note: in ICU version 3.2 and earlier, there was a formatVersion 1 + // that is no longer supported. At that time fFormatVersion was + // an int32_t field, rather than an array of 4 bytes. + + fDontFreeData = false; + if (data->fFTableLen != 0) { + fForwardTable = (RBBIStateTable *)((char *)data + fHeader->fFTable); + } + if (data->fRTableLen != 0) { + fReverseTable = (RBBIStateTable *)((char *)data + fHeader->fRTable); + } + + fTrie = ucptrie_openFromBinary(UCPTRIE_TYPE_FAST, + UCPTRIE_VALUE_BITS_ANY, + (uint8_t *)data + fHeader->fTrie, + fHeader->fTrieLen, + nullptr, // *actual length + &status); + if (U_FAILURE(status)) { + return; + } + + UCPTrieValueWidth width = ucptrie_getValueWidth(fTrie); + if (!(width == UCPTRIE_VALUE_BITS_8 || width == UCPTRIE_VALUE_BITS_16)) { + status = U_INVALID_FORMAT_ERROR; + return; + } + + fRuleSource = ((char *)data + fHeader->fRuleSource); + fRuleString = UnicodeString::fromUTF8(StringPiece(fRuleSource, fHeader->fRuleSourceLen)); + U_ASSERT(data->fRuleSourceLen > 0); + + fRuleStatusTable = (int32_t *)((char *)data + fHeader->fStatusTable); + fStatusMaxIdx = data->fStatusTableLen / sizeof(int32_t); + + fRefCount = 1; + +#ifdef RBBI_DEBUG + char *debugEnv = getenv("U_RBBIDEBUG"); + if (debugEnv && uprv_strstr(debugEnv, "data")) {this->printData();} +#endif +} + + +//----------------------------------------------------------------------------- +// +// Destructor. Don't call this - use removeReference() instead. +// +//----------------------------------------------------------------------------- +RBBIDataWrapper::~RBBIDataWrapper() { + U_ASSERT(fRefCount == 0); + ucptrie_close(fTrie); + fTrie = nullptr; + if (fUDataMem) { + udata_close(fUDataMem); + } else if (!fDontFreeData) { + uprv_free((void *)fHeader); + } +} + + + +//----------------------------------------------------------------------------- +// +// Operator == Consider two RBBIDataWrappers to be equal if they +// refer to the same underlying data. Although +// the data wrappers are normally shared between +// iterator instances, it's possible to independently +// open the same data twice, and get two instances, which +// should still be ==. +// +//----------------------------------------------------------------------------- +bool RBBIDataWrapper::operator ==(const RBBIDataWrapper &other) const { + if (fHeader == other.fHeader) { + return true; + } + if (fHeader->fLength != other.fHeader->fLength) { + return false; + } + if (uprv_memcmp(fHeader, other.fHeader, fHeader->fLength) == 0) { + return true; + } + return false; +} + +int32_t RBBIDataWrapper::hashCode() { + return fHeader->fFTableLen; +} + + + +//----------------------------------------------------------------------------- +// +// Reference Counting. A single RBBIDataWrapper object is shared among +// however many RulesBasedBreakIterator instances are +// referencing the same data. +// +//----------------------------------------------------------------------------- +void RBBIDataWrapper::removeReference() { + if (umtx_atomic_dec(&fRefCount) == 0) { + delete this; + } +} + + +RBBIDataWrapper *RBBIDataWrapper::addReference() { + umtx_atomic_inc(&fRefCount); + return this; +} + + + +//----------------------------------------------------------------------------- +// +// getRuleSourceString +// +//----------------------------------------------------------------------------- +const UnicodeString &RBBIDataWrapper::getRuleSourceString() const { + return fRuleString; +} + + +//----------------------------------------------------------------------------- +// +// print - debugging function to dump the runtime data tables. +// +//----------------------------------------------------------------------------- +#ifdef RBBI_DEBUG +void RBBIDataWrapper::printTable(const char *heading, const RBBIStateTable *table) { + uint32_t c; + uint32_t s; + + RBBIDebugPrintf("%s\n", heading); + + RBBIDebugPrintf(" fDictCategoriesStart: %d\n", table->fDictCategoriesStart); + RBBIDebugPrintf(" fLookAheadResultsSize: %d\n", table->fLookAheadResultsSize); + RBBIDebugPrintf(" Flags: %4x RBBI_LOOKAHEAD_HARD_BREAK=%s RBBI_BOF_REQUIRED=%s RBBI_8BITS_ROWS=%s\n", + table->fFlags, + table->fFlags & RBBI_LOOKAHEAD_HARD_BREAK ? "T" : "F", + table->fFlags & RBBI_BOF_REQUIRED ? "T" : "F", + table->fFlags & RBBI_8BITS_ROWS ? "T" : "F"); + RBBIDebugPrintf("\nState | Acc LA TagIx"); + for (c=0; cfCatCount; c++) {RBBIDebugPrintf("%3d ", c);} + RBBIDebugPrintf("\n------|---------------"); for (c=0;cfCatCount; c++) { + RBBIDebugPrintf("----"); + } + RBBIDebugPrintf("\n"); + + if (table == nullptr) { + RBBIDebugPrintf(" N U L L T A B L E\n\n"); + return; + } + UBool use8Bits = table->fFlags & RBBI_8BITS_ROWS; + for (s=0; sfNumStates; s++) { + RBBIStateTableRow *row = (RBBIStateTableRow *) + (table->fTableData + (table->fRowLen * s)); + if (use8Bits) { + RBBIDebugPrintf("%4d | %3d %3d %3d ", s, row->r8.fAccepting, row->r8.fLookAhead, row->r8.fTagsIdx); + for (c=0; cfCatCount; c++) { + RBBIDebugPrintf("%3d ", row->r8.fNextState[c]); + } + } else { + RBBIDebugPrintf("%4d | %3d %3d %3d ", s, row->r16.fAccepting, row->r16.fLookAhead, row->r16.fTagsIdx); + for (c=0; cfCatCount; c++) { + RBBIDebugPrintf("%3d ", row->r16.fNextState[c]); + } + } + RBBIDebugPrintf("\n"); + } + RBBIDebugPrintf("\n"); +} +#endif + + +void RBBIDataWrapper::printData() { +#ifdef RBBI_DEBUG + RBBIDebugPrintf("RBBI Data at %p\n", (void *)fHeader); + RBBIDebugPrintf(" Version = {%d %d %d %d}\n", fHeader->fFormatVersion[0], fHeader->fFormatVersion[1], + fHeader->fFormatVersion[2], fHeader->fFormatVersion[3]); + RBBIDebugPrintf(" total length of data = %d\n", fHeader->fLength); + RBBIDebugPrintf(" number of character categories = %d\n\n", fHeader->fCatCount); + + printTable("Forward State Transition Table", fForwardTable); + printTable("Reverse State Transition Table", fReverseTable); + + RBBIDebugPrintf("\nOriginal Rules source:\n"); + for (int32_t c=0; fRuleSource[c] != 0; c++) { + RBBIDebugPrintf("%c", fRuleSource[c]); + } + RBBIDebugPrintf("\n\n"); +#endif +} + + +U_NAMESPACE_END +U_NAMESPACE_USE + +//----------------------------------------------------------------------------- +// +// ubrk_swap - byte swap and char encoding swap of RBBI data +// +//----------------------------------------------------------------------------- + +U_CAPI int32_t U_EXPORT2 +ubrk_swap(const UDataSwapper *ds, const void *inData, int32_t length, void *outData, + UErrorCode *status) { + + if (status == nullptr || U_FAILURE(*status)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<-1 || (length>0 && outData==nullptr)) { + *status=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + // + // Check that the data header is for for break data. + // (Header contents are defined in genbrk.cpp) + // + const UDataInfo *pInfo = (const UDataInfo *)((const char *)inData+4); + if(!( pInfo->dataFormat[0]==0x42 && /* dataFormat="Brk " */ + pInfo->dataFormat[1]==0x72 && + pInfo->dataFormat[2]==0x6b && + pInfo->dataFormat[3]==0x20 && + RBBIDataWrapper::isDataVersionAcceptable(pInfo->formatVersion) )) { + udata_printError(ds, "ubrk_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized\n", + pInfo->dataFormat[0], pInfo->dataFormat[1], + pInfo->dataFormat[2], pInfo->dataFormat[3], + pInfo->formatVersion[0]); + *status=U_UNSUPPORTED_ERROR; + return 0; + } + + // + // Swap the data header. (This is the generic ICU Data Header, not the RBBI Specific + // RBBIDataHeader). This swap also conveniently gets us + // the size of the ICU d.h., which lets us locate the start + // of the RBBI specific data. + // + int32_t headerSize=udata_swapDataHeader(ds, inData, length, outData, status); + + + // + // Get the RRBI Data Header, and check that it appears to be OK. + // + const uint8_t *inBytes =(const uint8_t *)inData+headerSize; + RBBIDataHeader *rbbiDH = (RBBIDataHeader *)inBytes; + if (ds->readUInt32(rbbiDH->fMagic) != 0xb1a0 || + !RBBIDataWrapper::isDataVersionAcceptable(rbbiDH->fFormatVersion) || + ds->readUInt32(rbbiDH->fLength) < sizeof(RBBIDataHeader)) { + udata_printError(ds, "ubrk_swap(): RBBI Data header is invalid.\n"); + *status=U_UNSUPPORTED_ERROR; + return 0; + } + + // + // Prefight operation? Just return the size + // + int32_t breakDataLength = ds->readUInt32(rbbiDH->fLength); + int32_t totalSize = headerSize + breakDataLength; + if (length < 0) { + return totalSize; + } + + // + // Check that length passed in is consistent with length from RBBI data header. + // + if (length < totalSize) { + udata_printError(ds, "ubrk_swap(): too few bytes (%d after ICU Data header) for break data.\n", + breakDataLength); + *status=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + + + // + // Swap the Data. Do the data itself first, then the RBBI Data Header, because + // we need to reference the header to locate the data, and an + // inplace swap of the header leaves it unusable. + // + uint8_t *outBytes = (uint8_t *)outData + headerSize; + RBBIDataHeader *outputDH = (RBBIDataHeader *)outBytes; + + int32_t tableStartOffset; + int32_t tableLength; + + // + // If not swapping in place, zero out the output buffer before starting. + // Individual tables and other data items within are aligned to 8 byte boundaries + // when originally created. Any unused space between items needs to be zero. + // + if (inBytes != outBytes) { + uprv_memset(outBytes, 0, breakDataLength); + } + + // + // Each state table begins with several 32 bit fields. Calculate the size + // in bytes of these. + // + int32_t topSize = offsetof(RBBIStateTable, fTableData); + + // Forward state table. + tableStartOffset = ds->readUInt32(rbbiDH->fFTable); + tableLength = ds->readUInt32(rbbiDH->fFTableLen); + + if (tableLength > 0) { + RBBIStateTable *rbbiST = (RBBIStateTable *)(inBytes+tableStartOffset); + UBool use8Bits = ds->readUInt32(rbbiST->fFlags) & RBBI_8BITS_ROWS; + + ds->swapArray32(ds, inBytes+tableStartOffset, topSize, + outBytes+tableStartOffset, status); + + // Swap the state table if the table is in 16 bits. + if (use8Bits) { + if (outBytes != inBytes) { + uprv_memmove(outBytes+tableStartOffset+topSize, + inBytes+tableStartOffset+topSize, + tableLength-topSize); + } + } else { + ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize, + outBytes+tableStartOffset+topSize, status); + } + } + + // Reverse state table. Same layout as forward table, above. + tableStartOffset = ds->readUInt32(rbbiDH->fRTable); + tableLength = ds->readUInt32(rbbiDH->fRTableLen); + + if (tableLength > 0) { + RBBIStateTable *rbbiST = (RBBIStateTable *)(inBytes+tableStartOffset); + UBool use8Bits = ds->readUInt32(rbbiST->fFlags) & RBBI_8BITS_ROWS; + + ds->swapArray32(ds, inBytes+tableStartOffset, topSize, + outBytes+tableStartOffset, status); + + // Swap the state table if the table is in 16 bits. + if (use8Bits) { + if (outBytes != inBytes) { + uprv_memmove(outBytes+tableStartOffset+topSize, + inBytes+tableStartOffset+topSize, + tableLength-topSize); + } + } else { + ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize, + outBytes+tableStartOffset+topSize, status); + } + } + + // Trie table for character categories + ucptrie_swap(ds, inBytes+ds->readUInt32(rbbiDH->fTrie), ds->readUInt32(rbbiDH->fTrieLen), + outBytes+ds->readUInt32(rbbiDH->fTrie), status); + + // Source Rules Text. It's UTF8 data + if (outBytes != inBytes) { + uprv_memmove(outBytes+ds->readUInt32(rbbiDH->fRuleSource), + inBytes+ds->readUInt32(rbbiDH->fRuleSource), + ds->readUInt32(rbbiDH->fRuleSourceLen)); + } + + // Table of rule status values. It's all int_32 values + ds->swapArray32(ds, inBytes+ds->readUInt32(rbbiDH->fStatusTable), ds->readUInt32(rbbiDH->fStatusTableLen), + outBytes+ds->readUInt32(rbbiDH->fStatusTable), status); + + // And, last, the header. + // It is all int32_t values except for fFormataVersion, which is an array of four bytes. + // Swap the whole thing as int32_t, then re-swap the one field. + // + ds->swapArray32(ds, inBytes, sizeof(RBBIDataHeader), outBytes, status); + ds->swapArray32(ds, outputDH->fFormatVersion, 4, outputDH->fFormatVersion, status); + + return totalSize; +} + + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ diff --git a/intl/icu/source/common/rbbidata.h b/intl/icu/source/common/rbbidata.h new file mode 100644 index 0000000000..c48de3d935 --- /dev/null +++ b/intl/icu/source/common/rbbidata.h @@ -0,0 +1,212 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 1999-2014 International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: rbbidata.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* RBBI data formats Includes +* +* Structs that describes the format of the Binary RBBI data, +* as it is stored in ICU's data file. +* +* RBBIDataWrapper - Instances of this class sit between the +* raw data structs and the RulesBasedBreakIterator objects +* that are created by applications. The wrapper class +* provides reference counting for the underlying data, +* and direct pointers to data that would not otherwise +* be accessible without ugly pointer arithmetic. The +* wrapper does not attempt to provide any higher level +* abstractions for the data itself. +* +* There will be only one instance of RBBIDataWrapper for any +* set of RBBI run time data being shared by instances +* (clones) of RulesBasedBreakIterator. +*/ + +#ifndef __RBBIDATA_H__ +#define __RBBIDATA_H__ + +#include "unicode/utypes.h" +#include "unicode/udata.h" +#include "udataswp.h" + +/** + * Swap RBBI data. See udataswp.h. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +ubrk_swap(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +#ifdef __cplusplus + +#include "unicode/ucptrie.h" +#include "unicode/uobject.h" +#include "unicode/unistr.h" +#include "unicode/uversion.h" +#include "umutex.h" + + +U_NAMESPACE_BEGIN + +// The current RBBI data format version. +static const uint8_t RBBI_DATA_FORMAT_VERSION[] = {6, 0, 0, 0}; + +/* + * The following structs map exactly onto the raw data from ICU common data file. + */ +struct RBBIDataHeader { + uint32_t fMagic; /* == 0xbla0 */ + UVersionInfo fFormatVersion; /* Data Format. Same as the value in struct UDataInfo */ + /* if there is one associated with this data. */ + /* (version originates in rbbi, is copied to UDataInfo) */ + uint32_t fLength; /* Total length in bytes of this RBBI Data, */ + /* including all sections, not just the header. */ + uint32_t fCatCount; /* Number of character categories. */ + + /* */ + /* Offsets and sizes of each of the subsections within the RBBI data. */ + /* All offsets are bytes from the start of the RBBIDataHeader. */ + /* All sizes are in bytes. */ + /* */ + uint32_t fFTable; /* forward state transition table. */ + uint32_t fFTableLen; + uint32_t fRTable; /* Offset to the reverse state transition table. */ + uint32_t fRTableLen; + uint32_t fTrie; /* Offset to Trie data for character categories */ + uint32_t fTrieLen; + uint32_t fRuleSource; /* Offset to the source for for the break */ + uint32_t fRuleSourceLen; /* rules. Stored char16_t *. */ + uint32_t fStatusTable; /* Offset to the table of rule status values */ + uint32_t fStatusTableLen; + + uint32_t fReserved[6]; /* Reserved for expansion */ + +}; + + + +template +struct RBBIStateTableRowT { + T fAccepting; // Non-zero if this row is for an accepting state. + // Value 0: not an accepting state. + // 1: (ACCEPTING_UNCONDITIONAL) Unconditional Accepting state. + // >1: Look-ahead match has completed. + // Actual boundary position happened earlier. + // Value here == fLookAhead in earlier + // state, at actual boundary pos. + T fLookAhead; // Non-zero if this row is for a state that + // corresponds to a '/' in the rule source. + // Value is the same as the fAccepting + // value for the rule (which will appear + // in a different state. + T fTagsIdx; // Non-zero if this row covers a {tagged} position + // from a rule. Value is the index in the + // StatusTable of the set of matching + // tags (rule status values) + T fNextState[1]; // Next State, indexed by char category. + // Variable-length array declared with length 1 + // to disable bounds checkers. + // Array Size is actually fData->fHeader->fCatCount + // CAUTION: see RBBITableBuilder::getTableSize() + // before changing anything here. +}; + +typedef RBBIStateTableRowT RBBIStateTableRow8; +typedef RBBIStateTableRowT RBBIStateTableRow16; + +constexpr uint16_t ACCEPTING_UNCONDITIONAL = 1; // Value constant for RBBIStateTableRow::fAccepting + +union RBBIStateTableRow { + RBBIStateTableRow16 r16; + RBBIStateTableRow8 r8; +}; + +struct RBBIStateTable { + uint32_t fNumStates; // Number of states. + uint32_t fRowLen; // Length of a state table row, in bytes. + uint32_t fDictCategoriesStart; // Char category number of the first dictionary + // char class, or the the largest category number + 1 + // if there are no dictionary categories. + uint32_t fLookAheadResultsSize; // Size of run-time array required for holding + // look-ahead results. Indexed by row.fLookAhead. + uint32_t fFlags; // Option Flags for this state table. + char fTableData[1]; // First RBBIStateTableRow begins here. + // Variable-length array declared with length 1 + // to disable bounds checkers. + // (making it char[] simplifies ugly address + // arithmetic for indexing variable length rows.) +}; + +constexpr uint32_t RBBI_LOOKAHEAD_HARD_BREAK = 1; +constexpr uint32_t RBBI_BOF_REQUIRED = 2; +constexpr uint32_t RBBI_8BITS_ROWS = 4; + + +/* */ +/* The reference counting wrapper class */ +/* */ +class RBBIDataWrapper : public UMemory { +public: + enum EDontAdopt { + kDontAdopt + }; + RBBIDataWrapper(const RBBIDataHeader *data, UErrorCode &status); + RBBIDataWrapper(const RBBIDataHeader *data, enum EDontAdopt dontAdopt, UErrorCode &status); + RBBIDataWrapper(UDataMemory* udm, UErrorCode &status); + ~RBBIDataWrapper(); + + static UBool isDataVersionAcceptable(const UVersionInfo version); + + void init0(); + void init(const RBBIDataHeader *data, UErrorCode &status); + RBBIDataWrapper *addReference(); + void removeReference(); + bool operator ==(const RBBIDataWrapper &other) const; + int32_t hashCode(); + const UnicodeString &getRuleSourceString() const; + void printData(); + void printTable(const char *heading, const RBBIStateTable *table); + + /* */ + /* Pointers to items within the data */ + /* */ + const RBBIDataHeader *fHeader; + const RBBIStateTable *fForwardTable; + const RBBIStateTable *fReverseTable; + const char *fRuleSource; + const int32_t *fRuleStatusTable; + + /* number of int32_t values in the rule status table. Used to sanity check indexing */ + int32_t fStatusMaxIdx; + + UCPTrie *fTrie; + +private: + u_atomic_int32_t fRefCount; + UDataMemory *fUDataMem; + UnicodeString fRuleString; + UBool fDontFreeData; + + RBBIDataWrapper(const RBBIDataWrapper &other) = delete; /* forbid copying of this class */ + RBBIDataWrapper &operator=(const RBBIDataWrapper &other) = delete; /* forbid copying of this class */ +}; + + + +U_NAMESPACE_END + +U_CFUNC UBool rbbi_cleanup(); + +#endif /* C++ */ + +#endif diff --git a/intl/icu/source/common/rbbinode.cpp b/intl/icu/source/common/rbbinode.cpp new file mode 100644 index 0000000000..7aa75d5ffb --- /dev/null +++ b/intl/icu/source/common/rbbinode.cpp @@ -0,0 +1,373 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +*************************************************************************** +* Copyright (C) 2002-2016 International Business Machines Corporation * +* and others. All rights reserved. * +*************************************************************************** +*/ + +// +// File: rbbinode.cpp +// +// Implementation of class RBBINode, which represents a node in the +// tree generated when parsing the Rules Based Break Iterator rules. +// +// This "Class" is actually closer to a struct. +// Code using it is expected to directly access fields much of the time. +// + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/unistr.h" +#include "unicode/uniset.h" +#include "unicode/uchar.h" +#include "unicode/parsepos.h" + +#include "cstr.h" +#include "uvector.h" + +#include "rbbirb.h" +#include "rbbinode.h" + +#include "uassert.h" + + +U_NAMESPACE_BEGIN + +#ifdef RBBI_DEBUG +static int gLastSerial = 0; +#endif + + +//------------------------------------------------------------------------- +// +// Constructor. Just set the fields to reasonable default values. +// +//------------------------------------------------------------------------- +RBBINode::RBBINode(NodeType t) : UMemory() { +#ifdef RBBI_DEBUG + fSerialNum = ++gLastSerial; +#endif + fType = t; + fParent = nullptr; + fLeftChild = nullptr; + fRightChild = nullptr; + fInputSet = nullptr; + fFirstPos = 0; + fLastPos = 0; + fNullable = false; + fLookAheadEnd = false; + fRuleRoot = false; + fChainIn = false; + fVal = 0; + fPrecedence = precZero; + + UErrorCode status = U_ZERO_ERROR; + fFirstPosSet = new UVector(status); // TODO - get a real status from somewhere + fLastPosSet = new UVector(status); + fFollowPos = new UVector(status); + if (t==opCat) {fPrecedence = precOpCat;} + else if (t==opOr) {fPrecedence = precOpOr;} + else if (t==opStart) {fPrecedence = precStart;} + else if (t==opLParen) {fPrecedence = precLParen;} + +} + + +RBBINode::RBBINode(const RBBINode &other) : UMemory(other) { +#ifdef RBBI_DEBUG + fSerialNum = ++gLastSerial; +#endif + fType = other.fType; + fParent = nullptr; + fLeftChild = nullptr; + fRightChild = nullptr; + fInputSet = other.fInputSet; + fPrecedence = other.fPrecedence; + fText = other.fText; + fFirstPos = other.fFirstPos; + fLastPos = other.fLastPos; + fNullable = other.fNullable; + fVal = other.fVal; + fRuleRoot = false; + fChainIn = other.fChainIn; + UErrorCode status = U_ZERO_ERROR; + fFirstPosSet = new UVector(status); // TODO - get a real status from somewhere + fLastPosSet = new UVector(status); + fFollowPos = new UVector(status); +} + + +//------------------------------------------------------------------------- +// +// Destructor. Deletes both this node AND any child nodes, +// except in the case of variable reference nodes. For +// these, the l. child points back to the definition, which +// is common for all references to the variable, meaning +// it can't be deleted here. +// +//------------------------------------------------------------------------- +RBBINode::~RBBINode() { + // printf("deleting node %8x serial %4d\n", this, this->fSerialNum); + delete fInputSet; + fInputSet = nullptr; + + switch (this->fType) { + case varRef: + case setRef: + // for these node types, multiple instances point to the same "children" + // Storage ownership of children handled elsewhere. Don't delete here. + break; + + default: + delete fLeftChild; + fLeftChild = nullptr; + delete fRightChild; + fRightChild = nullptr; + } + + + delete fFirstPosSet; + delete fLastPosSet; + delete fFollowPos; + +} + + +//------------------------------------------------------------------------- +// +// cloneTree Make a copy of the subtree rooted at this node. +// Discard any variable references encountered along the way, +// and replace with copies of the variable's definitions. +// Used to replicate the expression underneath variable +// references in preparation for generating the DFA tables. +// +//------------------------------------------------------------------------- +RBBINode *RBBINode::cloneTree() { + RBBINode *n; + + if (fType == RBBINode::varRef) { + // If the current node is a variable reference, skip over it + // and clone the definition of the variable instead. + n = fLeftChild->cloneTree(); + } else if (fType == RBBINode::uset) { + n = this; + } else { + n = new RBBINode(*this); + // Check for null pointer. + if (n != nullptr) { + if (fLeftChild != nullptr) { + n->fLeftChild = fLeftChild->cloneTree(); + n->fLeftChild->fParent = n; + } + if (fRightChild != nullptr) { + n->fRightChild = fRightChild->cloneTree(); + n->fRightChild->fParent = n; + } + } + } + return n; +} + + + +//------------------------------------------------------------------------- +// +// flattenVariables Walk a parse tree, replacing any variable +// references with a copy of the variable's definition. +// Aside from variables, the tree is not changed. +// +// Return the root of the tree. If the root was not a variable +// reference, it remains unchanged - the root we started with +// is the root we return. If, however, the root was a variable +// reference, the root of the newly cloned replacement tree will +// be returned, and the original tree deleted. +// +// This function works by recursively walking the tree +// without doing anything until a variable reference is +// found, then calling cloneTree() at that point. Any +// nested references are handled by cloneTree(), not here. +// +//------------------------------------------------------------------------- +RBBINode *RBBINode::flattenVariables() { + if (fType == varRef) { + RBBINode *retNode = fLeftChild->cloneTree(); + if (retNode != nullptr) { + retNode->fRuleRoot = this->fRuleRoot; + retNode->fChainIn = this->fChainIn; + } + delete this; // TODO: undefined behavior. Fix. + return retNode; + } + + if (fLeftChild != nullptr) { + fLeftChild = fLeftChild->flattenVariables(); + fLeftChild->fParent = this; + } + if (fRightChild != nullptr) { + fRightChild = fRightChild->flattenVariables(); + fRightChild->fParent = this; + } + return this; +} + + +//------------------------------------------------------------------------- +// +// flattenSets Walk the parse tree, replacing any nodes of type setRef +// with a copy of the expression tree for the set. A set's +// equivalent expression tree is precomputed and saved as +// the left child of the uset node. +// +//------------------------------------------------------------------------- +void RBBINode::flattenSets() { + U_ASSERT(fType != setRef); + + if (fLeftChild != nullptr) { + if (fLeftChild->fType==setRef) { + RBBINode *setRefNode = fLeftChild; + RBBINode *usetNode = setRefNode->fLeftChild; + RBBINode *replTree = usetNode->fLeftChild; + fLeftChild = replTree->cloneTree(); + fLeftChild->fParent = this; + delete setRefNode; + } else { + fLeftChild->flattenSets(); + } + } + + if (fRightChild != nullptr) { + if (fRightChild->fType==setRef) { + RBBINode *setRefNode = fRightChild; + RBBINode *usetNode = setRefNode->fLeftChild; + RBBINode *replTree = usetNode->fLeftChild; + fRightChild = replTree->cloneTree(); + fRightChild->fParent = this; + delete setRefNode; + } else { + fRightChild->flattenSets(); + } + } +} + + + +//------------------------------------------------------------------------- +// +// findNodes() Locate all the nodes of the specified type, starting +// at the specified root. +// +//------------------------------------------------------------------------- +void RBBINode::findNodes(UVector *dest, RBBINode::NodeType kind, UErrorCode &status) { + /* test for buffer overflows */ + if (U_FAILURE(status)) { + return; + } + U_ASSERT(!dest->hasDeleter()); + if (fType == kind) { + dest->addElement(this, status); + } + if (fLeftChild != nullptr) { + fLeftChild->findNodes(dest, kind, status); + } + if (fRightChild != nullptr) { + fRightChild->findNodes(dest, kind, status); + } +} + + +//------------------------------------------------------------------------- +// +// print. Print out a single node, for debugging. +// +//------------------------------------------------------------------------- +#ifdef RBBI_DEBUG + +static int32_t serial(const RBBINode *node) { + return (node == nullptr? -1 : node->fSerialNum); +} + + +void RBBINode::printNode(const RBBINode *node) { + static const char * const nodeTypeNames[] = { + "setRef", + "uset", + "varRef", + "leafChar", + "lookAhead", + "tag", + "endMark", + "opStart", + "opCat", + "opOr", + "opStar", + "opPlus", + "opQuestion", + "opBreak", + "opReverse", + "opLParen" + }; + + if (node==nullptr) { + RBBIDebugPrintf("%10p", (void *)node); + } else { + RBBIDebugPrintf("%10p %5d %12s %c%c %5d %5d %5d %6d %d ", + (void *)node, node->fSerialNum, nodeTypeNames[node->fType], + node->fRuleRoot?'R':' ', node->fChainIn?'C':' ', + serial(node->fLeftChild), serial(node->fRightChild), serial(node->fParent), + node->fFirstPos, node->fVal); + if (node->fType == varRef) { + RBBI_DEBUG_printUnicodeString(node->fText); + } + } + RBBIDebugPrintf("\n"); +} +#endif + + +#ifdef RBBI_DEBUG +U_CFUNC void RBBI_DEBUG_printUnicodeString(const UnicodeString &s, int minWidth) { + RBBIDebugPrintf("%*s", minWidth, CStr(s)()); +} +#endif + + +//------------------------------------------------------------------------- +// +// print. Print out the tree of nodes rooted at "this" +// +//------------------------------------------------------------------------- +#ifdef RBBI_DEBUG +void RBBINode::printNodeHeader() { + RBBIDebugPrintf(" Address serial type LeftChild RightChild Parent position value\n"); +} + +void RBBINode::printTree(const RBBINode *node, UBool printHeading) { + if (printHeading) { + printNodeHeader(); + } + printNode(node); + if (node != nullptr) { + // Only dump the definition under a variable reference if asked to. + // Unconditionally dump children of all other node types. + if (node->fType != varRef) { + if (node->fLeftChild != nullptr) { + printTree(node->fLeftChild, false); + } + + if (node->fRightChild != nullptr) { + printTree(node->fRightChild, false); + } + } + } +} +#endif + + + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ diff --git a/intl/icu/source/common/rbbinode.h b/intl/icu/source/common/rbbinode.h new file mode 100644 index 0000000000..4ed84d4e07 --- /dev/null +++ b/intl/icu/source/common/rbbinode.h @@ -0,0 +1,127 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/******************************************************************** + * COPYRIGHT: + * Copyright (c) 2001-2016, International Business Machines Corporation and + * others. All Rights Reserved. + ********************************************************************/ + +#ifndef RBBINODE_H +#define RBBINODE_H + +#include "unicode/utypes.h" +#include "unicode/unistr.h" +#include "unicode/uobject.h" + +// +// class RBBINode +// +// Represents a node in the parse tree generated when reading +// a rule file. +// + +U_NAMESPACE_BEGIN + +class UnicodeSet; +class UVector; + +class RBBINode : public UMemory { + public: + enum NodeType { + setRef, + uset, + varRef, + leafChar, + lookAhead, + tag, + endMark, + opStart, + opCat, + opOr, + opStar, + opPlus, + opQuestion, + opBreak, + opReverse, + opLParen + }; + + enum OpPrecedence { + precZero, + precStart, + precLParen, + precOpOr, + precOpCat + }; + + NodeType fType; + RBBINode *fParent; + RBBINode *fLeftChild; + RBBINode *fRightChild; + UnicodeSet *fInputSet; // For uset nodes only. + OpPrecedence fPrecedence; // For binary ops only. + + UnicodeString fText; // Text corresponding to this node. + // May be lazily evaluated when (if) needed + // for some node types. + int fFirstPos; // Position in the rule source string of the + // first text associated with the node. + // If there's a left child, this will be the same + // as that child's left pos. + int fLastPos; // Last position in the rule source string + // of any text associated with this node. + // If there's a right child, this will be the same + // as that child's last position. + + UBool fNullable; // See Aho. + int32_t fVal; // For leafChar nodes, the value. + // Values are the character category, + // corresponds to columns in the final + // state transition table. + + UBool fLookAheadEnd; // For endMark nodes, set true if + // marking the end of a look-ahead rule. + + UBool fRuleRoot; // True if this node is the root of a rule. + UBool fChainIn; // True if chaining into this rule is allowed + // (no '^' present). + + UVector *fFirstPosSet; + UVector *fLastPosSet; // TODO: rename fFirstPos & fLastPos to avoid confusion. + UVector *fFollowPos; + + + RBBINode(NodeType t); + RBBINode(const RBBINode &other); + ~RBBINode(); + + RBBINode *cloneTree(); + RBBINode *flattenVariables(); + void flattenSets(); + void findNodes(UVector *dest, RBBINode::NodeType kind, UErrorCode &status); + +#ifdef RBBI_DEBUG + static void printNodeHeader(); + static void printNode(const RBBINode *n); + static void printTree(const RBBINode *n, UBool withHeading); +#endif + + private: + RBBINode &operator = (const RBBINode &other); // No defs. + bool operator == (const RBBINode &other); // Private, so these functions won't accidentally be used. + +#ifdef RBBI_DEBUG + public: + int fSerialNum; // Debugging aids. +#endif +}; + +#ifdef RBBI_DEBUG +U_CFUNC void +RBBI_DEBUG_printUnicodeString(const UnicodeString &s, int minWidth=0); +#endif + +U_NAMESPACE_END + +#endif + diff --git a/intl/icu/source/common/rbbirb.cpp b/intl/icu/source/common/rbbirb.cpp new file mode 100644 index 0000000000..7177254ec4 --- /dev/null +++ b/intl/icu/source/common/rbbirb.cpp @@ -0,0 +1,361 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// +// file: rbbirb.cpp +// +// Copyright (C) 2002-2011, International Business Machines Corporation and others. +// All Rights Reserved. +// +// This file contains the RBBIRuleBuilder class implementation. This is the main class for +// building (compiling) break rules into the tables required by the runtime +// RBBI engine. +// + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/brkiter.h" +#include "unicode/rbbi.h" +#include "unicode/ubrk.h" +#include "unicode/unistr.h" +#include "unicode/uniset.h" +#include "unicode/uchar.h" +#include "unicode/uchriter.h" +#include "unicode/ustring.h" +#include "unicode/parsepos.h" +#include "unicode/parseerr.h" + +#include "cmemory.h" +#include "cstring.h" +#include "rbbirb.h" +#include "rbbinode.h" +#include "rbbiscan.h" +#include "rbbisetb.h" +#include "rbbitblb.h" +#include "rbbidata.h" +#include "uassert.h" + + +U_NAMESPACE_BEGIN + + +//---------------------------------------------------------------------------------------- +// +// Constructor. +// +//---------------------------------------------------------------------------------------- +RBBIRuleBuilder::RBBIRuleBuilder(const UnicodeString &rules, + UParseError *parseErr, + UErrorCode &status) + : fRules(rules), fStrippedRules(rules) +{ + fStatus = &status; // status is checked below + fParseError = parseErr; + fDebugEnv = nullptr; +#ifdef RBBI_DEBUG + fDebugEnv = getenv("U_RBBIDEBUG"); +#endif + + + fForwardTree = nullptr; + fReverseTree = nullptr; + fSafeFwdTree = nullptr; + fSafeRevTree = nullptr; + fDefaultTree = &fForwardTree; + fForwardTable = nullptr; + fRuleStatusVals = nullptr; + fChainRules = false; + fLBCMNoChain = false; + fLookAheadHardBreak = false; + fUSetNodes = nullptr; + fRuleStatusVals = nullptr; + fScanner = nullptr; + fSetBuilder = nullptr; + if (parseErr) { + uprv_memset(parseErr, 0, sizeof(UParseError)); + } + + if (U_FAILURE(status)) { + return; + } + + fUSetNodes = new UVector(status); // bcos status gets overwritten here + fRuleStatusVals = new UVector(status); + fScanner = new RBBIRuleScanner(this); + fSetBuilder = new RBBISetBuilder(this); + if (U_FAILURE(status)) { + return; + } + if(fSetBuilder == 0 || fScanner == 0 || fUSetNodes == 0 || fRuleStatusVals == 0) { + status = U_MEMORY_ALLOCATION_ERROR; + } +} + + + +//---------------------------------------------------------------------------------------- +// +// Destructor +// +//---------------------------------------------------------------------------------------- +RBBIRuleBuilder::~RBBIRuleBuilder() { + + int i; + for (i=0; ; i++) { + RBBINode *n = (RBBINode *)fUSetNodes->elementAt(i); + if (n==nullptr) { + break; + } + delete n; + } + + delete fUSetNodes; + delete fSetBuilder; + delete fForwardTable; + delete fForwardTree; + delete fReverseTree; + delete fSafeFwdTree; + delete fSafeRevTree; + delete fScanner; + delete fRuleStatusVals; +} + + + + + +//---------------------------------------------------------------------------------------- +// +// flattenData() - Collect up the compiled RBBI rule data and put it into +// the format for saving in ICU data files, +// which is also the format needed by the RBBI runtime engine. +// +//---------------------------------------------------------------------------------------- +static int32_t align8(int32_t i) {return (i+7) & 0xfffffff8;} + +RBBIDataHeader *RBBIRuleBuilder::flattenData() { + int32_t i; + + if (U_FAILURE(*fStatus)) { + return nullptr; + } + + // Remove whitespace from the rules to make it smaller. + // The rule parser has already removed comments. + fStrippedRules = fScanner->stripRules(fStrippedRules); + + // Calculate the size of each section in the data. + // Sizes here are padded up to a multiple of 8 for better memory alignment. + // Sections sizes actually stored in the header are for the actual data + // without the padding. + // + int32_t headerSize = align8(sizeof(RBBIDataHeader)); + int32_t forwardTableSize = align8(fForwardTable->getTableSize()); + int32_t reverseTableSize = align8(fForwardTable->getSafeTableSize()); + int32_t trieSize = align8(fSetBuilder->getTrieSize()); + int32_t statusTableSize = align8(fRuleStatusVals->size() * sizeof(int32_t)); + + int32_t rulesLengthInUTF8 = 0; + u_strToUTF8WithSub(0, 0, &rulesLengthInUTF8, + fStrippedRules.getBuffer(), fStrippedRules.length(), + 0xfffd, nullptr, fStatus); + *fStatus = U_ZERO_ERROR; + + int32_t rulesSize = align8((rulesLengthInUTF8+1)); + + int32_t totalSize = headerSize + + forwardTableSize + + reverseTableSize + + statusTableSize + trieSize + rulesSize; + +#ifdef RBBI_DEBUG + if (fDebugEnv && uprv_strstr(fDebugEnv, "size")) { + RBBIDebugPrintf("Header Size: %8d\n", headerSize); + RBBIDebugPrintf("Forward Table Size: %8d\n", forwardTableSize); + RBBIDebugPrintf("Reverse Table Size: %8d\n", reverseTableSize); + RBBIDebugPrintf("Trie Size: %8d\n", trieSize); + RBBIDebugPrintf("Status Table Size: %8d\n", statusTableSize); + RBBIDebugPrintf("Rules Size: %8d\n", rulesSize); + RBBIDebugPrintf("-----------------------------\n"); + RBBIDebugPrintf("Total Size: %8d\n", totalSize); + } +#endif + + RBBIDataHeader *data = (RBBIDataHeader *)uprv_malloc(totalSize); + if (data == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memset(data, 0, totalSize); + + + data->fMagic = 0xb1a0; + data->fFormatVersion[0] = RBBI_DATA_FORMAT_VERSION[0]; + data->fFormatVersion[1] = RBBI_DATA_FORMAT_VERSION[1]; + data->fFormatVersion[2] = RBBI_DATA_FORMAT_VERSION[2]; + data->fFormatVersion[3] = RBBI_DATA_FORMAT_VERSION[3]; + data->fLength = totalSize; + data->fCatCount = fSetBuilder->getNumCharCategories(); + + data->fFTable = headerSize; + data->fFTableLen = forwardTableSize; + + data->fRTable = data->fFTable + data->fFTableLen; + data->fRTableLen = reverseTableSize; + + data->fTrie = data->fRTable + data->fRTableLen; + data->fTrieLen = trieSize; + data->fStatusTable = data->fTrie + data->fTrieLen; + data->fStatusTableLen= statusTableSize; + data->fRuleSource = data->fStatusTable + statusTableSize; + data->fRuleSourceLen = rulesLengthInUTF8; + + uprv_memset(data->fReserved, 0, sizeof(data->fReserved)); + + fForwardTable->exportTable((uint8_t *)data + data->fFTable); + fForwardTable->exportSafeTable((uint8_t *)data + data->fRTable); + fSetBuilder->serializeTrie ((uint8_t *)data + data->fTrie); + + int32_t *ruleStatusTable = (int32_t *)((uint8_t *)data + data->fStatusTable); + for (i=0; isize(); i++) { + ruleStatusTable[i] = fRuleStatusVals->elementAti(i); + } + + u_strToUTF8WithSub((char *)data+data->fRuleSource, rulesSize, &rulesLengthInUTF8, + fStrippedRules.getBuffer(), fStrippedRules.length(), + 0xfffd, nullptr, fStatus); + if (U_FAILURE(*fStatus)) { + return nullptr; + } + + return data; +} + + +//---------------------------------------------------------------------------------------- +// +// createRuleBasedBreakIterator construct from source rules that are passed in +// in a UnicodeString +// +//---------------------------------------------------------------------------------------- +BreakIterator * +RBBIRuleBuilder::createRuleBasedBreakIterator( const UnicodeString &rules, + UParseError *parseError, + UErrorCode &status) +{ + // + // Read the input rules, generate a parse tree, symbol table, + // and list of all Unicode Sets referenced by the rules. + // + RBBIRuleBuilder builder(rules, parseError, status); + if (U_FAILURE(status)) { // status checked here bcos build below doesn't + return nullptr; + } + + RBBIDataHeader *data = builder.build(status); + + if (U_FAILURE(status)) { + return nullptr; + } + + // + // Create a break iterator from the compiled rules. + // (Identical to creation from stored pre-compiled rules) + // + // status is checked after init in construction. + RuleBasedBreakIterator *This = new RuleBasedBreakIterator(data, status); + if (U_FAILURE(status)) { + delete This; + This = nullptr; + } + else if(This == nullptr) { // test for nullptr + status = U_MEMORY_ALLOCATION_ERROR; + } + return This; +} + +RBBIDataHeader *RBBIRuleBuilder::build(UErrorCode &status) { + if (U_FAILURE(status)) { + return nullptr; + } + + fScanner->parse(); + if (U_FAILURE(status)) { + return nullptr; + } + + // + // UnicodeSet processing. + // Munge the Unicode Sets to create an initial set of character categories. + // + fSetBuilder->buildRanges(); + + // + // Generate the DFA state transition table. + // + fForwardTable = new RBBITableBuilder(this, &fForwardTree, status); + if (fForwardTable == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + + fForwardTable->buildForwardTable(); + + // State table and character category optimization. + // Merge equivalent rows and columns. + // Note that this process alters the initial set of character categories, + // causing the representation of UnicodeSets in the parse tree to become invalid. + + optimizeTables(); + fForwardTable->buildSafeReverseTable(status); + + +#ifdef RBBI_DEBUG + if (fDebugEnv && uprv_strstr(fDebugEnv, "states")) { + fForwardTable->printStates(); + fForwardTable->printRuleStatusTable(); + fForwardTable->printReverseTable(); + } +#endif + + // Generate the mapping tables (TRIE) from input code points to + // the character categories. + // + fSetBuilder->buildTrie(); + + // + // Package up the compiled data into a memory image + // in the run-time format. + // + RBBIDataHeader *data = flattenData(); // returns nullptr if error + if (U_FAILURE(status)) { + return nullptr; + } + return data; +} + +void RBBIRuleBuilder::optimizeTables() { + bool didSomething; + do { + didSomething = false; + + // Begin looking for duplicates with char class 3. + // Classes 0, 1 and 2 are special; they are unused, {bof} and {eof} respectively, + // and should not have other categories merged into them. + IntPair duplPair = {3, 0}; + while (fForwardTable->findDuplCharClassFrom(&duplPair)) { + fSetBuilder->mergeCategories(duplPair); + fForwardTable->removeColumn(duplPair.second); + didSomething = true; + } + + while (fForwardTable->removeDuplicateStates() > 0) { + didSomething = true; + } + } while (didSomething); +} + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ diff --git a/intl/icu/source/common/rbbirb.h b/intl/icu/source/common/rbbirb.h new file mode 100644 index 0000000000..d983a184b6 --- /dev/null +++ b/intl/icu/source/common/rbbirb.h @@ -0,0 +1,237 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// +// rbbirb.h +// +// Copyright (C) 2002-2008, International Business Machines Corporation and others. +// All Rights Reserved. +// +// This file contains declarations for several classes from the +// Rule Based Break Iterator rule builder. +// + + +#ifndef RBBIRB_H +#define RBBIRB_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include + +#include "unicode/uobject.h" +#include "unicode/rbbi.h" +#include "unicode/uniset.h" +#include "unicode/parseerr.h" +#include "uhash.h" +#include "uvector.h" +#include "unicode/symtable.h"// For UnicodeSet parsing, is the interface that + // looks up references to $variables within a set. + + +U_NAMESPACE_BEGIN + +class RBBIRuleScanner; +struct RBBIRuleTableEl; +class RBBISetBuilder; +class RBBINode; +class RBBITableBuilder; + + + +//-------------------------------------------------------------------------------- +// +// RBBISymbolTable. Implements SymbolTable interface that is used by the +// UnicodeSet parser to resolve references to $variables. +// +//-------------------------------------------------------------------------------- +class RBBISymbolTableEntry : public UMemory { // The symbol table hash table contains one +public: // of these structs for each entry. + RBBISymbolTableEntry(); + UnicodeString key; + RBBINode *val; + ~RBBISymbolTableEntry(); + +private: + RBBISymbolTableEntry(const RBBISymbolTableEntry &other) = delete; // forbid copying of this class + RBBISymbolTableEntry &operator=(const RBBISymbolTableEntry &other) = delete; // forbid copying of this class +}; + + +class RBBISymbolTable : public UMemory, public SymbolTable { +private: + const UnicodeString &fRules; + UHashtable *fHashTable; + RBBIRuleScanner *fRuleScanner; + + // These next two fields are part of the mechanism for passing references to + // already-constructed UnicodeSets back to the UnicodeSet constructor + // when the pattern includes $variable references. + const UnicodeString ffffString; // = "/uffff" + UnicodeSet *fCachedSetLookup; + +public: + // API inherited from class SymbolTable + virtual const UnicodeString* lookup(const UnicodeString& s) const override; + virtual const UnicodeFunctor* lookupMatcher(UChar32 ch) const override; + virtual UnicodeString parseReference(const UnicodeString& text, + ParsePosition& pos, int32_t limit) const override; + + // Additional Functions + RBBISymbolTable(RBBIRuleScanner *, const UnicodeString &fRules, UErrorCode &status); + virtual ~RBBISymbolTable(); + + virtual RBBINode *lookupNode(const UnicodeString &key) const; + virtual void addEntry (const UnicodeString &key, RBBINode *val, UErrorCode &err); + +#ifdef RBBI_DEBUG + virtual void rbbiSymtablePrint() const; +#else + // A do-nothing inline function for non-debug builds. Member funcs can't be empty + // or the call sites won't compile. + int32_t fFakeField; + #define rbbiSymtablePrint() fFakeField=0; +#endif + +private: + RBBISymbolTable(const RBBISymbolTable &other); // forbid copying of this class + RBBISymbolTable &operator=(const RBBISymbolTable &other); // forbid copying of this class +}; + + +//-------------------------------------------------------------------------------- +// +// class RBBIRuleBuilder The top-level class handling RBBI rule compiling. +// +//-------------------------------------------------------------------------------- +class RBBIRuleBuilder : public UMemory { +public: + + // Create a rule based break iterator from a set of rules. + // This function is the main entry point into the rule builder. The + // public ICU API for creating RBBIs uses this function to do the actual work. + // + static BreakIterator * createRuleBasedBreakIterator( const UnicodeString &rules, + UParseError *parseError, + UErrorCode &status); + +public: + // The "public" functions and data members that appear below are accessed + // (and shared) by the various parts that make up the rule builder. They + // are NOT intended to be accessed by anything outside of the + // rule builder implementation. + RBBIRuleBuilder(const UnicodeString &rules, + UParseError *parseErr, + UErrorCode &status + ); + + virtual ~RBBIRuleBuilder(); + + /** + * Build the state tables and char class Trie from the source rules. + */ + RBBIDataHeader *build(UErrorCode &status); + + + /** + * Fold together redundant character classes (table columns) and + * redundant states (table rows). Done after initial table generation, + * before serializing the result. + */ + void optimizeTables(); + + char *fDebugEnv; // controls debug trace output + UErrorCode *fStatus; // Error reporting. Keeping status + UParseError *fParseError; // here avoids passing it everywhere. + const UnicodeString &fRules; // The rule string that we are compiling + UnicodeString fStrippedRules; // The rule string, with comments stripped. + + RBBIRuleScanner *fScanner; // The scanner. + RBBINode *fForwardTree; // The parse trees, generated by the scanner, + RBBINode *fReverseTree; // then manipulated by subsequent steps. + RBBINode *fSafeFwdTree; + RBBINode *fSafeRevTree; + + RBBINode **fDefaultTree; // For rules not qualified with a ! + // the tree to which they belong to. + + UBool fChainRules; // True for chained Unicode TR style rules. + // False for traditional regexp rules. + + UBool fLBCMNoChain; // True: suppress chaining of rules on + // chars with LineBreak property == CM. + + UBool fLookAheadHardBreak; // True: Look ahead matches cause an + // immediate break, no continuing for the + // longest match. + + RBBISetBuilder *fSetBuilder; // Set and Character Category builder. + UVector *fUSetNodes; // Vector of all uset nodes. + + RBBITableBuilder *fForwardTable; // State transition table, build time form. + + UVector *fRuleStatusVals; // The values that can be returned + // from getRuleStatus(). + + RBBIDataHeader *flattenData(); // Create the flattened (runtime format) + // data tables.. +private: + RBBIRuleBuilder(const RBBIRuleBuilder &other) = delete; // forbid copying of this class + RBBIRuleBuilder &operator=(const RBBIRuleBuilder &other) = delete; // forbid copying of this class +}; + + + + +//---------------------------------------------------------------------------- +// +// RBBISetTableEl is an entry in the hash table of UnicodeSets that have +// been encountered. The val Node will be of nodetype uset +// and contain pointers to the actual UnicodeSets. +// The Key is the source string for initializing the set. +// +// The hash table is used to avoid creating duplicate +// unnamed (not $var references) UnicodeSets. +// +// Memory Management: +// The Hash Table owns these RBBISetTableEl structs and +// the key strings. It does NOT own the val nodes. +// +//---------------------------------------------------------------------------- +struct RBBISetTableEl { + UnicodeString *key; + RBBINode *val; +}; + +/** + * A pair of ints, used to bundle pairs of states or pairs of character classes. + */ +typedef std::pair IntPair; + + +//---------------------------------------------------------------------------- +// +// RBBIDebugPrintf Printf equivalent, for debugging output. +// Conditional compilation of the implementation lets us +// get rid of the stdio dependency in environments where it +// is unavailable. +// +//---------------------------------------------------------------------------- +#ifdef RBBI_DEBUG +#include +#define RBBIDebugPrintf printf +#define RBBIDebugPuts puts +#else +#undef RBBIDebugPrintf +#define RBBIDebugPuts(arg) +#endif + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ + +#endif + + + diff --git a/intl/icu/source/common/rbbirpt.h b/intl/icu/source/common/rbbirpt.h new file mode 100644 index 0000000000..ca1bcf45dc --- /dev/null +++ b/intl/icu/source/common/rbbirpt.h @@ -0,0 +1,296 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +//--------------------------------------------------------------------------------- +// +// Generated Header File. Do not edit by hand. +// This file contains the state table for the ICU Rule Based Break Iterator +// rule parser. +// It is generated by the Perl script "rbbicst.pl" from +// the rule parser state definitions file "rbbirpt.txt". +// +// Copyright (C) 2002-2016 International Business Machines Corporation +// and others. All rights reserved. +// +//--------------------------------------------------------------------------------- +#ifndef RBBIRPT_H +#define RBBIRPT_H + +#include "unicode/utypes.h" + +U_NAMESPACE_BEGIN +// +// Character classes for RBBI rule scanning. +// + static const uint8_t kRuleSet_digit_char = 128; + static const uint8_t kRuleSet_name_char = 129; + static const uint8_t kRuleSet_name_start_char = 130; + static const uint8_t kRuleSet_rule_char = 131; + static const uint8_t kRuleSet_white_space = 132; + + +enum RBBI_RuleParseAction { + doCheckVarDef, + doDotAny, + doEndAssign, + doEndOfRule, + doEndVariableName, + doExit, + doExprCatOperator, + doExprFinished, + doExprOrOperator, + doExprRParen, + doExprStart, + doLParen, + doNOP, + doNoChain, + doOptionEnd, + doOptionStart, + doReverseDir, + doRuleChar, + doRuleError, + doRuleErrorAssignExpr, + doScanUnicodeSet, + doSlash, + doStartAssign, + doStartTagValue, + doStartVariableName, + doTagDigit, + doTagExpectedError, + doTagValue, + doUnaryOpPlus, + doUnaryOpQuestion, + doUnaryOpStar, + doVariableNameExpectedErr, + rbbiLastAction}; + +//------------------------------------------------------------------------------- +// +// RBBIRuleTableEl represents the structure of a row in the transition table +// for the rule parser state machine. +//------------------------------------------------------------------------------- +struct RBBIRuleTableEl { + RBBI_RuleParseAction fAction; + uint8_t fCharClass; // 0-127: an individual ASCII character + // 128-255: character class index + uint8_t fNextState; // 0-250: normal next-stat numbers + // 255: pop next-state from stack. + uint8_t fPushState; + UBool fNextChar; +}; + +static const struct RBBIRuleTableEl gRuleParseStateTable[] = { + {doNOP, 0, 0, 0, true} + , {doExprStart, 254, 29, 9, false} // 1 start + , {doNOP, 132, 1,0, true} // 2 + , {doNoChain, 94 /* ^ */, 12, 9, true} // 3 + , {doExprStart, 36 /* $ */, 88, 98, false} // 4 + , {doNOP, 33 /* ! */, 19,0, true} // 5 + , {doNOP, 59 /* ; */, 1,0, true} // 6 + , {doNOP, 252, 0,0, false} // 7 + , {doExprStart, 255, 29, 9, false} // 8 + , {doEndOfRule, 59 /* ; */, 1,0, true} // 9 break-rule-end + , {doNOP, 132, 9,0, true} // 10 + , {doRuleError, 255, 103,0, false} // 11 + , {doExprStart, 254, 29,0, false} // 12 start-after-caret + , {doNOP, 132, 12,0, true} // 13 + , {doRuleError, 94 /* ^ */, 103,0, false} // 14 + , {doExprStart, 36 /* $ */, 88, 37, false} // 15 + , {doRuleError, 59 /* ; */, 103,0, false} // 16 + , {doRuleError, 252, 103,0, false} // 17 + , {doExprStart, 255, 29,0, false} // 18 + , {doNOP, 33 /* ! */, 21,0, true} // 19 rev-option + , {doReverseDir, 255, 28, 9, false} // 20 + , {doOptionStart, 130, 23,0, true} // 21 option-scan1 + , {doRuleError, 255, 103,0, false} // 22 + , {doNOP, 129, 23,0, true} // 23 option-scan2 + , {doOptionEnd, 255, 25,0, false} // 24 + , {doNOP, 59 /* ; */, 1,0, true} // 25 option-scan3 + , {doNOP, 132, 25,0, true} // 26 + , {doRuleError, 255, 103,0, false} // 27 + , {doExprStart, 255, 29, 9, false} // 28 reverse-rule + , {doRuleChar, 254, 38,0, true} // 29 term + , {doNOP, 132, 29,0, true} // 30 + , {doRuleChar, 131, 38,0, true} // 31 + , {doNOP, 91 /* [ */, 94, 38, false} // 32 + , {doLParen, 40 /* ( */, 29, 38, true} // 33 + , {doNOP, 36 /* $ */, 88, 37, false} // 34 + , {doDotAny, 46 /* . */, 38,0, true} // 35 + , {doRuleError, 255, 103,0, false} // 36 + , {doCheckVarDef, 255, 38,0, false} // 37 term-var-ref + , {doNOP, 132, 38,0, true} // 38 expr-mod + , {doUnaryOpStar, 42 /* * */, 43,0, true} // 39 + , {doUnaryOpPlus, 43 /* + */, 43,0, true} // 40 + , {doUnaryOpQuestion, 63 /* ? */, 43,0, true} // 41 + , {doNOP, 255, 43,0, false} // 42 + , {doExprCatOperator, 254, 29,0, false} // 43 expr-cont + , {doNOP, 132, 43,0, true} // 44 + , {doExprCatOperator, 131, 29,0, false} // 45 + , {doExprCatOperator, 91 /* [ */, 29,0, false} // 46 + , {doExprCatOperator, 40 /* ( */, 29,0, false} // 47 + , {doExprCatOperator, 36 /* $ */, 29,0, false} // 48 + , {doExprCatOperator, 46 /* . */, 29,0, false} // 49 + , {doExprCatOperator, 47 /* / */, 55,0, false} // 50 + , {doExprCatOperator, 123 /* { */, 67,0, true} // 51 + , {doExprOrOperator, 124 /* | */, 29,0, true} // 52 + , {doExprRParen, 41 /* ) */, 255,0, true} // 53 + , {doExprFinished, 255, 255,0, false} // 54 + , {doSlash, 47 /* / */, 57,0, true} // 55 look-ahead + , {doNOP, 255, 103,0, false} // 56 + , {doExprCatOperator, 254, 29,0, false} // 57 expr-cont-no-slash + , {doNOP, 132, 43,0, true} // 58 + , {doExprCatOperator, 131, 29,0, false} // 59 + , {doExprCatOperator, 91 /* [ */, 29,0, false} // 60 + , {doExprCatOperator, 40 /* ( */, 29,0, false} // 61 + , {doExprCatOperator, 36 /* $ */, 29,0, false} // 62 + , {doExprCatOperator, 46 /* . */, 29,0, false} // 63 + , {doExprOrOperator, 124 /* | */, 29,0, true} // 64 + , {doExprRParen, 41 /* ) */, 255,0, true} // 65 + , {doExprFinished, 255, 255,0, false} // 66 + , {doNOP, 132, 67,0, true} // 67 tag-open + , {doStartTagValue, 128, 70,0, false} // 68 + , {doTagExpectedError, 255, 103,0, false} // 69 + , {doNOP, 132, 74,0, true} // 70 tag-value + , {doNOP, 125 /* } */, 74,0, false} // 71 + , {doTagDigit, 128, 70,0, true} // 72 + , {doTagExpectedError, 255, 103,0, false} // 73 + , {doNOP, 132, 74,0, true} // 74 tag-close + , {doTagValue, 125 /* } */, 77,0, true} // 75 + , {doTagExpectedError, 255, 103,0, false} // 76 + , {doExprCatOperator, 254, 29,0, false} // 77 expr-cont-no-tag + , {doNOP, 132, 77,0, true} // 78 + , {doExprCatOperator, 131, 29,0, false} // 79 + , {doExprCatOperator, 91 /* [ */, 29,0, false} // 80 + , {doExprCatOperator, 40 /* ( */, 29,0, false} // 81 + , {doExprCatOperator, 36 /* $ */, 29,0, false} // 82 + , {doExprCatOperator, 46 /* . */, 29,0, false} // 83 + , {doExprCatOperator, 47 /* / */, 55,0, false} // 84 + , {doExprOrOperator, 124 /* | */, 29,0, true} // 85 + , {doExprRParen, 41 /* ) */, 255,0, true} // 86 + , {doExprFinished, 255, 255,0, false} // 87 + , {doStartVariableName, 36 /* $ */, 90,0, true} // 88 scan-var-name + , {doNOP, 255, 103,0, false} // 89 + , {doNOP, 130, 92,0, true} // 90 scan-var-start + , {doVariableNameExpectedErr, 255, 103,0, false} // 91 + , {doNOP, 129, 92,0, true} // 92 scan-var-body + , {doEndVariableName, 255, 255,0, false} // 93 + , {doScanUnicodeSet, 91 /* [ */, 255,0, true} // 94 scan-unicode-set + , {doScanUnicodeSet, 112 /* p */, 255,0, true} // 95 + , {doScanUnicodeSet, 80 /* P */, 255,0, true} // 96 + , {doNOP, 255, 103,0, false} // 97 + , {doNOP, 132, 98,0, true} // 98 assign-or-rule + , {doStartAssign, 61 /* = */, 29, 101, true} // 99 + , {doNOP, 255, 37, 9, false} // 100 + , {doEndAssign, 59 /* ; */, 1,0, true} // 101 assign-end + , {doRuleErrorAssignExpr, 255, 103,0, false} // 102 + , {doExit, 255, 103,0, true} // 103 errorDeath + }; +#ifdef RBBI_DEBUG +static const char * const RBBIRuleStateNames[] = { 0, + "start", + 0, + 0, + 0, + 0, + 0, + 0, + 0, + "break-rule-end", + 0, + 0, + "start-after-caret", + 0, + 0, + 0, + 0, + 0, + 0, + "rev-option", + 0, + "option-scan1", + 0, + "option-scan2", + 0, + "option-scan3", + 0, + 0, + "reverse-rule", + "term", + 0, + 0, + 0, + 0, + 0, + 0, + 0, + "term-var-ref", + "expr-mod", + 0, + 0, + 0, + 0, + "expr-cont", + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + "look-ahead", + 0, + "expr-cont-no-slash", + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + "tag-open", + 0, + 0, + "tag-value", + 0, + 0, + 0, + "tag-close", + 0, + 0, + "expr-cont-no-tag", + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + "scan-var-name", + 0, + "scan-var-start", + 0, + "scan-var-body", + 0, + "scan-unicode-set", + 0, + 0, + 0, + "assign-or-rule", + 0, + 0, + "assign-end", + 0, + "errorDeath", + 0}; +#endif + +U_NAMESPACE_END +#endif diff --git a/intl/icu/source/common/rbbirpt.txt b/intl/icu/source/common/rbbirpt.txt new file mode 100644 index 0000000000..86b75c11e9 --- /dev/null +++ b/intl/icu/source/common/rbbirpt.txt @@ -0,0 +1,337 @@ + +#***************************************************************************** +# +# Copyright (C) 2016 and later: Unicode, Inc. and others. +# License & terms of use: http://www.unicode.org/copyright.html +# +#***************************************************************************** +#***************************************************************************** +# +# Copyright (C) 2002-2016, International Business Machines Corporation and others. +# All Rights Reserved. +# +#***************************************************************************** +# +# file: rbbirpt.txt +# ICU Break Iterator Rule Parser State Table +# +# This state table is used when reading and parsing a set of RBBI rules +# The rule parser uses a state machine; the data in this file define the +# state transitions that occur for each input character. +# +# *** This file defines the RBBI rule grammar. This is it. +# *** The determination of what is accepted is here. +# +# This file is processed by a perl script "rbbicst.pl" to produce initialized C arrays +# that are then built with the rule parser. +# +# perl rbbicst.pl < rbbirpt.txt > rbbirpt.h + +# +# Here is the syntax of the state definitions in this file: +# +# +#StateName: +# input-char n next-state ^push-state action +# input-char n next-state ^push-state action +# | | | | | +# | | | | |--- action to be performed by state machine +# | | | | See function RBBIRuleScanner::doParseActions() +# | | | | +# | | | |--- Push this named state onto the state stack. +# | | | Later, when next state is specified as "pop", +# | | | the pushed state will become the current state. +# | | | +# | | |--- Transition to this state if the current input character matches the input +# | | character or char class in the left hand column. "pop" causes the next +# | | state to be popped from the state stack. +# | | +# | |--- When making the state transition specified on this line, advance to the next +# | character from the input only if 'n' appears here. +# | +# |--- Character or named character classes to test for. If the current character being scanned +# matches, perform the actions and go to the state specified on this line. +# The input character is tested sequentally, in the order written. The characters and +# character classes tested for do not need to be mutually exclusive. The first match wins. +# + + + + +# +# start state, scan position is at the beginning of the rules file, or in between two rules. +# +start: + escaped term ^break-rule-end doExprStart + white_space n start + '^' n start-after-caret ^break-rule-end doNoChain + '$' scan-var-name ^assign-or-rule doExprStart + '!' n rev-option + ';' n start # ignore empty rules. + eof exit + default term ^break-rule-end doExprStart + +# +# break-rule-end: Returned from doing a break-rule expression. +# +break-rule-end: + ';' n start doEndOfRule + white_space n break-rule-end + default errorDeath doRuleError + +# +# start of a rule, after having seen a '^' (inhibits rule chain in). +# Similar to the main 'start' state in most respects, except +# - empty rule is an error. +# - A second '^' is an error. +# +start-after-caret: + escaped term doExprStart + white_space n start-after-caret + '^' errorDeath doRuleError # two '^'s + '$' scan-var-name ^term-var-ref doExprStart + ';' errorDeath doRuleError # ^ ; + eof errorDeath doRuleError + default term doExprStart + +# +# ! We've just scanned a '!', indicating either a !!key word flag or a +# !Reverse rule. +# +rev-option: + '!' n option-scan1 + default reverse-rule ^break-rule-end doReverseDir + +option-scan1: + name_start_char n option-scan2 doOptionStart + default errorDeath doRuleError + +option-scan2: + name_char n option-scan2 + default option-scan3 doOptionEnd + +option-scan3: + ';' n start + white_space n option-scan3 + default errorDeath doRuleError + + +reverse-rule: + default term ^break-rule-end doExprStart + + +# +# term. Eat through a single rule character, or a composite thing, which +# could be a parenthesized expression, a variable name, or a Unicode Set. +# +term: + escaped n expr-mod doRuleChar + white_space n term + rule_char n expr-mod doRuleChar + '[' scan-unicode-set ^expr-mod + '(' n term ^expr-mod doLParen + '$' scan-var-name ^term-var-ref + '.' n expr-mod doDotAny + default errorDeath doRuleError + + + +# +# term-var-ref We've just finished scanning a reference to a $variable. +# Check that the variable was defined. +# The variable name scanning is in common with assignment statements, +# so the check can't be done there. +term-var-ref: + default expr-mod doCheckVarDef + + +# +# expr-mod We've just finished scanning a term, now look for the optional +# trailing '*', '?', '+' +# +expr-mod: + white_space n expr-mod + '*' n expr-cont doUnaryOpStar + '+' n expr-cont doUnaryOpPlus + '?' n expr-cont doUnaryOpQuestion + default expr-cont + + +# +# expr-cont Expression, continuation. At a point where additional terms are +# allowed, but not required. +# +expr-cont: + escaped term doExprCatOperator + white_space n expr-cont + rule_char term doExprCatOperator + '[' term doExprCatOperator + '(' term doExprCatOperator + '$' term doExprCatOperator + '.' term doExprCatOperator + '/' look-ahead doExprCatOperator + '{' n tag-open doExprCatOperator + '|' n term doExprOrOperator + ')' n pop doExprRParen + default pop doExprFinished + + +# +# look-ahead Scanning a '/', which identifies a break point, assuming that the +# remainder of the expression matches. +# +# Generate a parse tree as if this was a special kind of input symbol +# appearing in an otherwise normal concatenation expression. +# +look-ahead: + '/' n expr-cont-no-slash doSlash + default errorDeath + + +# +# expr-cont-no-slash Expression, continuation. At a point where additional terms are +# allowed, but not required. Just like +# expr-cont, above, except that no '/' +# look-ahead symbol is permitted. +# +expr-cont-no-slash: + escaped term doExprCatOperator + white_space n expr-cont + rule_char term doExprCatOperator + '[' term doExprCatOperator + '(' term doExprCatOperator + '$' term doExprCatOperator + '.' term doExprCatOperator + '|' n term doExprOrOperator + ')' n pop doExprRParen + default pop doExprFinished + + +# +# tags scanning a '{', the opening delimiter for a tag that identifies +# the kind of match. Scan the whole {dddd} tag, where d=digit +# +tag-open: + white_space n tag-open + digit_char tag-value doStartTagValue + default errorDeath doTagExpectedError + +tag-value: + white_space n tag-close + '}' tag-close + digit_char n tag-value doTagDigit + default errorDeath doTagExpectedError + +tag-close: + white_space n tag-close + '}' n expr-cont-no-tag doTagValue + default errorDeath doTagExpectedError + + + +# +# expr-cont-no-tag Expression, continuation. At a point where additional terms are +# allowed, but not required. Just like +# expr-cont, above, except that no "{ddd}" +# tagging is permitted. +# +expr-cont-no-tag: + escaped term doExprCatOperator + white_space n expr-cont-no-tag + rule_char term doExprCatOperator + '[' term doExprCatOperator + '(' term doExprCatOperator + '$' term doExprCatOperator + '.' term doExprCatOperator + '/' look-ahead doExprCatOperator + '|' n term doExprOrOperator + ')' n pop doExprRParen + default pop doExprFinished + + + + +# +# Variable Name Scanning. +# +# The state that branched to here must have pushed a return state +# to go to after completion of the variable name scanning. +# +# The current input character must be the $ that introduces the name. +# The $ is consumed here rather than in the state that first detected it +# so that the doStartVariableName action only needs to happen in one +# place (here), and the other states don't need to worry about it. +# +scan-var-name: + '$' n scan-var-start doStartVariableName + default errorDeath + + +scan-var-start: + name_start_char n scan-var-body + default errorDeath doVariableNameExpectedErr + +scan-var-body: + name_char n scan-var-body + default pop doEndVariableName + + + +# +# scan-unicode-set Unicode Sets are parsed by the the UnicodeSet class. +# Within the RBBI parser, after finding the first character +# of a Unicode Set, we just hand the rule input at that +# point of to the Unicode Set constructor, then pick +# up parsing after the close of the set. +# +# The action for this state invokes the UnicodeSet parser. +# +scan-unicode-set: + '[' n pop doScanUnicodeSet + 'p' n pop doScanUnicodeSet + 'P' n pop doScanUnicodeSet + default errorDeath + + + + + + + +# +# assign-or-rule. A $variable was encountered at the start of something, could be +# either an assignment statement or a rule, depending on whether an '=' +# follows the variable name. We get to this state when the variable name +# scanning does a return. +# +assign-or-rule: + white_space n assign-or-rule + '=' n term ^assign-end doStartAssign # variable was target of assignment + default term-var-ref ^break-rule-end # variable was a term in a rule + + + +# +# assign-end This state is entered when the end of the expression on the +# right hand side of an assignment is found. We get here via +# a pop; this state is pushed when the '=' in an assignment is found. +# +# The only thing allowed at this point is a ';'. The RHS of an +# assignment must look like a rule expression, and we come here +# when what is being scanned no longer looks like an expression. +# +assign-end: + ';' n start doEndAssign + default errorDeath doRuleErrorAssignExpr + + + +# +# errorDeath. This state is specified as the next state whenever a syntax error +# in the source rules is detected. Barring bugs, the state machine will never +# actually get here, but will stop because of the action associated with the error. +# But, just in case, this state asks the state machine to exit. +errorDeath: + default n errorDeath doExit + + diff --git a/intl/icu/source/common/rbbiscan.cpp b/intl/icu/source/common/rbbiscan.cpp new file mode 100644 index 0000000000..455ace78b8 --- /dev/null +++ b/intl/icu/source/common/rbbiscan.cpp @@ -0,0 +1,1285 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// +// file: rbbiscan.cpp +// +// Copyright (C) 2002-2016, International Business Machines Corporation and others. +// All Rights Reserved. +// +// This file contains the Rule Based Break Iterator Rule Builder functions for +// scanning the rules and assembling a parse tree. This is the first phase +// of compiling the rules. +// +// The overall of the rules is managed by class RBBIRuleBuilder, which will +// create and use an instance of this class as part of the process. +// + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/unistr.h" +#include "unicode/uniset.h" +#include "unicode/uchar.h" +#include "unicode/uchriter.h" +#include "unicode/parsepos.h" +#include "unicode/parseerr.h" +#include "cmemory.h" +#include "cstring.h" + +#include "rbbirpt.h" // Contains state table for the rbbi rules parser. + // generated by a Perl script. +#include "rbbirb.h" +#include "rbbinode.h" +#include "rbbiscan.h" +#include "rbbitblb.h" + +#include "uassert.h" + +//------------------------------------------------------------------------------ +// +// Unicode Set init strings for each of the character classes needed for parsing a rule file. +// (Initialized with hex values for portability to EBCDIC based machines. +// Really ugly, but there's no good way to avoid it.) +// +// The sets are referred to by name in the rbbirpt.txt, which is the +// source form of the state transition table for the RBBI rule parser. +// +//------------------------------------------------------------------------------ +static const char16_t gRuleSet_rule_char_pattern[] = { + // Characters that may appear as literals in patterns without escaping or quoting. + // [ ^ [ \ p { Z } \ u 0 0 2 0 + 0x5b, 0x5e, 0x5b, 0x5c, 0x70, 0x7b, 0x5a, 0x7d, 0x5c, 0x75, 0x30, 0x30, 0x32, 0x30, + // - \ u 0 0 7 f ] - [ \ p + 0x2d, 0x5c, 0x75, 0x30, 0x30, 0x37, 0x66, 0x5d, 0x2d, 0x5b, 0x5c, 0x70, + // { L } ] - [ \ p { N } ] ] + 0x7b, 0x4c, 0x7d, 0x5d, 0x2d, 0x5b, 0x5c, 0x70, 0x7b, 0x4e, 0x7d, 0x5d, 0x5d, 0}; + +static const char16_t gRuleSet_name_char_pattern[] = { +// [ _ \ p { L } \ p { N } ] + 0x5b, 0x5f, 0x5c, 0x70, 0x7b, 0x4c, 0x7d, 0x5c, 0x70, 0x7b, 0x4e, 0x7d, 0x5d, 0}; + +static const char16_t gRuleSet_digit_char_pattern[] = { +// [ 0 - 9 ] + 0x5b, 0x30, 0x2d, 0x39, 0x5d, 0}; + +static const char16_t gRuleSet_name_start_char_pattern[] = { +// [ _ \ p { L } ] + 0x5b, 0x5f, 0x5c, 0x70, 0x7b, 0x4c, 0x7d, 0x5d, 0 }; + +static const char16_t kAny[] = {0x61, 0x6e, 0x79, 0x00}; // "any" + + +U_CDECL_BEGIN +static void U_CALLCONV RBBISetTable_deleter(void *p) { + icu::RBBISetTableEl *px = (icu::RBBISetTableEl *)p; + delete px->key; + // Note: px->val is owned by the linked list "fSetsListHead" in scanner. + // Don't delete the value nodes here. + uprv_free(px); +} +U_CDECL_END + +U_NAMESPACE_BEGIN + +//------------------------------------------------------------------------------ +// +// Constructor. +// +//------------------------------------------------------------------------------ +RBBIRuleScanner::RBBIRuleScanner(RBBIRuleBuilder *rb) +{ + fRB = rb; + fScanIndex = 0; + fNextIndex = 0; + fQuoteMode = false; + fLineNum = 1; + fCharNum = 0; + fLastChar = 0; + + fStateTable = nullptr; + fStack[0] = 0; + fStackPtr = 0; + fNodeStack[0] = nullptr; + fNodeStackPtr = 0; + + fReverseRule = false; + fLookAheadRule = false; + fNoChainInRule = false; + + fSymbolTable = nullptr; + fSetTable = nullptr; + fRuleNum = 0; + fOptionStart = 0; + + // Do not check status until after all critical fields are sufficiently initialized + // that the destructor can run cleanly. + if (U_FAILURE(*rb->fStatus)) { + return; + } + + // + // Set up the constant Unicode Sets. + // Note: These could be made static, lazily initialized, and shared among + // all instances of RBBIRuleScanners. BUT this is quite a bit simpler, + // and the time to build these few sets should be small compared to a + // full break iterator build. + fRuleSets[kRuleSet_rule_char-128] + = UnicodeSet(UnicodeString(gRuleSet_rule_char_pattern), *rb->fStatus); + // fRuleSets[kRuleSet_white_space-128] = [:Pattern_White_Space:] + fRuleSets[kRuleSet_white_space-128]. + add(9, 0xd).add(0x20).add(0x85).add(0x200e, 0x200f).add(0x2028, 0x2029); + fRuleSets[kRuleSet_name_char-128] + = UnicodeSet(UnicodeString(gRuleSet_name_char_pattern), *rb->fStatus); + fRuleSets[kRuleSet_name_start_char-128] + = UnicodeSet(UnicodeString(gRuleSet_name_start_char_pattern), *rb->fStatus); + fRuleSets[kRuleSet_digit_char-128] + = UnicodeSet(UnicodeString(gRuleSet_digit_char_pattern), *rb->fStatus); + if (*rb->fStatus == U_ILLEGAL_ARGUMENT_ERROR) { + // This case happens if ICU's data is missing. UnicodeSet tries to look up property + // names from the init string, can't find them, and claims an illegal argument. + // Change the error so that the actual problem will be clearer to users. + *rb->fStatus = U_BRK_INIT_ERROR; + } + if (U_FAILURE(*rb->fStatus)) { + return; + } + + fSymbolTable = new RBBISymbolTable(this, rb->fRules, *rb->fStatus); + if (fSymbolTable == nullptr) { + *rb->fStatus = U_MEMORY_ALLOCATION_ERROR; + return; + } + fSetTable = uhash_open(uhash_hashUnicodeString, uhash_compareUnicodeString, nullptr, rb->fStatus); + if (U_FAILURE(*rb->fStatus)) { + return; + } + uhash_setValueDeleter(fSetTable, RBBISetTable_deleter); +} + + + +//------------------------------------------------------------------------------ +// +// Destructor +// +//------------------------------------------------------------------------------ +RBBIRuleScanner::~RBBIRuleScanner() { + delete fSymbolTable; + if (fSetTable != nullptr) { + uhash_close(fSetTable); + fSetTable = nullptr; + + } + + + // Node Stack. + // Normally has one entry, which is the entire parse tree for the rules. + // If errors occurred, there may be additional subtrees left on the stack. + while (fNodeStackPtr > 0) { + delete fNodeStack[fNodeStackPtr]; + fNodeStackPtr--; + } + +} + +//------------------------------------------------------------------------------ +// +// doParseAction Do some action during rule parsing. +// Called by the parse state machine. +// Actions build the parse tree and Unicode Sets, +// and maintain the parse stack for nested expressions. +// +// TODO: unify EParseAction and RBBI_RuleParseAction enum types. +// They represent exactly the same thing. They're separate +// only to work around enum forward declaration restrictions +// in some compilers, while at the same time avoiding multiple +// definitions problems. I'm sure that there's a better way. +// +//------------------------------------------------------------------------------ +UBool RBBIRuleScanner::doParseActions(int32_t action) +{ + RBBINode *n = nullptr; + + UBool returnVal = true; + + switch (action) { + + case doExprStart: + pushNewNode(RBBINode::opStart); + fRuleNum++; + break; + + + case doNoChain: + // Scanned a '^' while on the rule start state. + fNoChainInRule = true; + break; + + + case doExprOrOperator: + { + fixOpStack(RBBINode::precOpCat); + RBBINode *operandNode = fNodeStack[fNodeStackPtr--]; + RBBINode *orNode = pushNewNode(RBBINode::opOr); + if (U_FAILURE(*fRB->fStatus)) { + break; + } + orNode->fLeftChild = operandNode; + operandNode->fParent = orNode; + } + break; + + case doExprCatOperator: + // concatenation operator. + // For the implicit concatenation of adjacent terms in an expression that are + // not separated by any other operator. Action is invoked between the + // actions for the two terms. + { + fixOpStack(RBBINode::precOpCat); + RBBINode *operandNode = fNodeStack[fNodeStackPtr--]; + RBBINode *catNode = pushNewNode(RBBINode::opCat); + if (U_FAILURE(*fRB->fStatus)) { + break; + } + catNode->fLeftChild = operandNode; + operandNode->fParent = catNode; + } + break; + + case doLParen: + // Open Paren. + // The openParen node is a dummy operation type with a low precedence, + // which has the affect of ensuring that any real binary op that + // follows within the parens binds more tightly to the operands than + // stuff outside of the parens. + pushNewNode(RBBINode::opLParen); + break; + + case doExprRParen: + fixOpStack(RBBINode::precLParen); + break; + + case doNOP: + break; + + case doStartAssign: + // We've just scanned "$variable = " + // The top of the node stack has the $variable ref node. + + // Save the start position of the RHS text in the StartExpression node + // that precedes the $variableReference node on the stack. + // This will eventually be used when saving the full $variable replacement + // text as a string. + n = fNodeStack[fNodeStackPtr-1]; + n->fFirstPos = fNextIndex; // move past the '=' + + // Push a new start-of-expression node; needed to keep parse of the + // RHS expression happy. + pushNewNode(RBBINode::opStart); + break; + + + + + case doEndAssign: + { + // We have reached the end of an assignment statement. + // Current scan char is the ';' that terminates the assignment. + + // Terminate expression, leaves expression parse tree rooted in TOS node. + fixOpStack(RBBINode::precStart); + + RBBINode *startExprNode = fNodeStack[fNodeStackPtr-2]; + RBBINode *varRefNode = fNodeStack[fNodeStackPtr-1]; + RBBINode *RHSExprNode = fNodeStack[fNodeStackPtr]; + + // Save original text of right side of assignment, excluding the terminating ';' + // in the root of the node for the right-hand-side expression. + RHSExprNode->fFirstPos = startExprNode->fFirstPos; + RHSExprNode->fLastPos = fScanIndex; + fRB->fRules.extractBetween(RHSExprNode->fFirstPos, RHSExprNode->fLastPos, RHSExprNode->fText); + + // Expression parse tree becomes l. child of the $variable reference node. + varRefNode->fLeftChild = RHSExprNode; + RHSExprNode->fParent = varRefNode; + + // Make a symbol table entry for the $variableRef node. + fSymbolTable->addEntry(varRefNode->fText, varRefNode, *fRB->fStatus); + if (U_FAILURE(*fRB->fStatus)) { + // This is a round-about way to get the parse position set + // so that duplicate symbols error messages include a line number. + UErrorCode t = *fRB->fStatus; + *fRB->fStatus = U_ZERO_ERROR; + error(t); + } + + // Clean up the stack. + delete startExprNode; + fNodeStackPtr-=3; + break; + } + + case doEndOfRule: + { + fixOpStack(RBBINode::precStart); // Terminate expression, leaves expression + if (U_FAILURE(*fRB->fStatus)) { // parse tree rooted in TOS node. + break; + } +#ifdef RBBI_DEBUG + if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "rtree")) {printNodeStack("end of rule");} +#endif + U_ASSERT(fNodeStackPtr == 1); + RBBINode *thisRule = fNodeStack[fNodeStackPtr]; + + // If this rule includes a look-ahead '/', add a endMark node to the + // expression tree. + if (fLookAheadRule) { + RBBINode *endNode = pushNewNode(RBBINode::endMark); + RBBINode *catNode = pushNewNode(RBBINode::opCat); + if (U_FAILURE(*fRB->fStatus)) { + break; + } + fNodeStackPtr -= 2; + catNode->fLeftChild = thisRule; + catNode->fRightChild = endNode; + fNodeStack[fNodeStackPtr] = catNode; + endNode->fVal = fRuleNum; + endNode->fLookAheadEnd = true; + thisRule = catNode; + + // TODO: Disable chaining out of look-ahead (hard break) rules. + // The break on rule match is forced, so there is no point in building up + // the state table to chain into another rule for a longer match. + } + + // Mark this node as being the root of a rule. + thisRule->fRuleRoot = true; + + // Flag if chaining into this rule is wanted. + // + if (fRB->fChainRules && // If rule chaining is enabled globally via !!chain + !fNoChainInRule) { // and no '^' chain-in inhibit was on this rule + thisRule->fChainIn = true; + } + + + // All rule expressions are ORed together. + // The ';' that terminates an expression really just functions as a '|' with + // a low operator prededence. + // + // Each of the four sets of rules are collected separately. + // (forward, reverse, safe_forward, safe_reverse) + // OR this rule into the appropriate group of them. + // + RBBINode **destRules = (fReverseRule? &fRB->fSafeRevTree : fRB->fDefaultTree); + + if (*destRules != nullptr) { + // This is not the first rule encountered. + // OR previous stuff (from *destRules) + // with the current rule expression (on the Node Stack) + // with the resulting OR expression going to *destRules + // + thisRule = fNodeStack[fNodeStackPtr]; + RBBINode *prevRules = *destRules; + RBBINode *orNode = pushNewNode(RBBINode::opOr); + if (U_FAILURE(*fRB->fStatus)) { + break; + } + orNode->fLeftChild = prevRules; + prevRules->fParent = orNode; + orNode->fRightChild = thisRule; + thisRule->fParent = orNode; + *destRules = orNode; + } + else + { + // This is the first rule encountered (for this direction). + // Just move its parse tree from the stack to *destRules. + *destRules = fNodeStack[fNodeStackPtr]; + } + fReverseRule = false; // in preparation for the next rule. + fLookAheadRule = false; + fNoChainInRule = false; + fNodeStackPtr = 0; + } + break; + + + case doRuleError: + error(U_BRK_RULE_SYNTAX); + returnVal = false; + break; + + + case doVariableNameExpectedErr: + error(U_BRK_RULE_SYNTAX); + break; + + + // + // Unary operands + ? * + // These all appear after the operand to which they apply. + // When we hit one, the operand (may be a whole sub expression) + // will be on the top of the stack. + // Unary Operator becomes TOS, with the old TOS as its one child. + case doUnaryOpPlus: + { + RBBINode *operandNode = fNodeStack[fNodeStackPtr--]; + RBBINode *plusNode = pushNewNode(RBBINode::opPlus); + if (U_FAILURE(*fRB->fStatus)) { + break; + } + plusNode->fLeftChild = operandNode; + operandNode->fParent = plusNode; + } + break; + + case doUnaryOpQuestion: + { + RBBINode *operandNode = fNodeStack[fNodeStackPtr--]; + RBBINode *qNode = pushNewNode(RBBINode::opQuestion); + if (U_FAILURE(*fRB->fStatus)) { + break; + } + qNode->fLeftChild = operandNode; + operandNode->fParent = qNode; + } + break; + + case doUnaryOpStar: + { + RBBINode *operandNode = fNodeStack[fNodeStackPtr--]; + RBBINode *starNode = pushNewNode(RBBINode::opStar); + if (U_FAILURE(*fRB->fStatus)) { + break; + } + starNode->fLeftChild = operandNode; + operandNode->fParent = starNode; + } + break; + + case doRuleChar: + // A "Rule Character" is any single character that is a literal part + // of the regular expression. Like a, b and c in the expression "(abc*) | [:L:]" + // These are pretty uncommon in break rules; the terms are more commonly + // sets. To keep things uniform, treat these characters like as + // sets that just happen to contain only one character. + { + n = pushNewNode(RBBINode::setRef); + if (U_FAILURE(*fRB->fStatus)) { + break; + } + findSetFor(UnicodeString(fC.fChar), n); + n->fFirstPos = fScanIndex; + n->fLastPos = fNextIndex; + fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText); + break; + } + + case doDotAny: + // scanned a ".", meaning match any single character. + { + n = pushNewNode(RBBINode::setRef); + if (U_FAILURE(*fRB->fStatus)) { + break; + } + findSetFor(UnicodeString(true, kAny, 3), n); + n->fFirstPos = fScanIndex; + n->fLastPos = fNextIndex; + fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText); + break; + } + + case doSlash: + // Scanned a '/', which identifies a look-ahead break position in a rule. + n = pushNewNode(RBBINode::lookAhead); + if (U_FAILURE(*fRB->fStatus)) { + break; + } + n->fVal = fRuleNum; + n->fFirstPos = fScanIndex; + n->fLastPos = fNextIndex; + fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText); + fLookAheadRule = true; + break; + + + case doStartTagValue: + // Scanned a '{', the opening delimiter for a tag value within a rule. + n = pushNewNode(RBBINode::tag); + if (U_FAILURE(*fRB->fStatus)) { + break; + } + n->fVal = 0; + n->fFirstPos = fScanIndex; + n->fLastPos = fNextIndex; + break; + + case doTagDigit: + // Just scanned a decimal digit that's part of a tag value + { + n = fNodeStack[fNodeStackPtr]; + uint32_t v = u_charDigitValue(fC.fChar); + U_ASSERT(v < 10); + n->fVal = n->fVal*10 + v; + break; + } + + case doTagValue: + n = fNodeStack[fNodeStackPtr]; + n->fLastPos = fNextIndex; + fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText); + break; + + case doTagExpectedError: + error(U_BRK_MALFORMED_RULE_TAG); + returnVal = false; + break; + + case doOptionStart: + // Scanning a !!option. At the start of string. + fOptionStart = fScanIndex; + break; + + case doOptionEnd: + { + UnicodeString opt(fRB->fRules, fOptionStart, fScanIndex-fOptionStart); + if (opt == UNICODE_STRING("chain", 5)) { + fRB->fChainRules = true; + } else if (opt == UNICODE_STRING("LBCMNoChain", 11)) { + fRB->fLBCMNoChain = true; + } else if (opt == UNICODE_STRING("forward", 7)) { + fRB->fDefaultTree = &fRB->fForwardTree; + } else if (opt == UNICODE_STRING("reverse", 7)) { + fRB->fDefaultTree = &fRB->fReverseTree; + } else if (opt == UNICODE_STRING("safe_forward", 12)) { + fRB->fDefaultTree = &fRB->fSafeFwdTree; + } else if (opt == UNICODE_STRING("safe_reverse", 12)) { + fRB->fDefaultTree = &fRB->fSafeRevTree; + } else if (opt == UNICODE_STRING("lookAheadHardBreak", 18)) { + fRB->fLookAheadHardBreak = true; + } else if (opt == UNICODE_STRING("quoted_literals_only", 20)) { + fRuleSets[kRuleSet_rule_char-128].clear(); + } else if (opt == UNICODE_STRING("unquoted_literals", 17)) { + fRuleSets[kRuleSet_rule_char-128].applyPattern(UnicodeString(gRuleSet_rule_char_pattern), *fRB->fStatus); + } else { + error(U_BRK_UNRECOGNIZED_OPTION); + } + } + break; + + case doReverseDir: + fReverseRule = true; + break; + + case doStartVariableName: + n = pushNewNode(RBBINode::varRef); + if (U_FAILURE(*fRB->fStatus)) { + break; + } + n->fFirstPos = fScanIndex; + break; + + case doEndVariableName: + n = fNodeStack[fNodeStackPtr]; + if (n==nullptr || n->fType != RBBINode::varRef) { + error(U_BRK_INTERNAL_ERROR); + break; + } + n->fLastPos = fScanIndex; + fRB->fRules.extractBetween(n->fFirstPos+1, n->fLastPos, n->fText); + // Look the newly scanned name up in the symbol table + // If there's an entry, set the l. child of the var ref to the replacement expression. + // (We also pass through here when scanning assignments, but no harm is done, other + // than a slight wasted effort that seems hard to avoid. Lookup will be null) + n->fLeftChild = fSymbolTable->lookupNode(n->fText); + break; + + case doCheckVarDef: + n = fNodeStack[fNodeStackPtr]; + if (n->fLeftChild == nullptr) { + error(U_BRK_UNDEFINED_VARIABLE); + returnVal = false; + } + break; + + case doExprFinished: + break; + + case doRuleErrorAssignExpr: + error(U_BRK_ASSIGN_ERROR); + returnVal = false; + break; + + case doExit: + returnVal = false; + break; + + case doScanUnicodeSet: + scanSet(); + break; + + default: + error(U_BRK_INTERNAL_ERROR); + returnVal = false; + break; + } + return returnVal && U_SUCCESS(*fRB->fStatus); +} + + + + +//------------------------------------------------------------------------------ +// +// Error Report a rule parse error. +// Only report it if no previous error has been recorded. +// +//------------------------------------------------------------------------------ +void RBBIRuleScanner::error(UErrorCode e) { + if (U_SUCCESS(*fRB->fStatus)) { + *fRB->fStatus = e; + if (fRB->fParseError) { + fRB->fParseError->line = fLineNum; + fRB->fParseError->offset = fCharNum; + fRB->fParseError->preContext[0] = 0; + fRB->fParseError->postContext[0] = 0; + } + } +} + + + + +//------------------------------------------------------------------------------ +// +// fixOpStack The parse stack holds partially assembled chunks of the parse tree. +// An entry on the stack may be as small as a single setRef node, +// or as large as the parse tree +// for an entire expression (this will be the one item left on the stack +// when the parsing of an RBBI rule completes. +// +// This function is called when a binary operator is encountered. +// It looks back up the stack for operators that are not yet associated +// with a right operand, and if the precedence of the stacked operator >= +// the precedence of the current operator, binds the operand left, +// to the previously encountered operator. +// +//------------------------------------------------------------------------------ +void RBBIRuleScanner::fixOpStack(RBBINode::OpPrecedence p) { + RBBINode *n; + // printNodeStack("entering fixOpStack()"); + for (;;) { + n = fNodeStack[fNodeStackPtr-1]; // an operator node + if (n->fPrecedence == 0) { + RBBIDebugPuts("RBBIRuleScanner::fixOpStack, bad operator node"); + error(U_BRK_INTERNAL_ERROR); + return; + } + + if (n->fPrecedence < p || n->fPrecedence <= RBBINode::precLParen) { + // The most recent operand goes with the current operator, + // not with the previously stacked one. + break; + } + // Stack operator is a binary op ( '|' or concatenation) + // TOS operand becomes right child of this operator. + // Resulting subexpression becomes the TOS operand. + n->fRightChild = fNodeStack[fNodeStackPtr]; + fNodeStack[fNodeStackPtr]->fParent = n; + fNodeStackPtr--; + // printNodeStack("looping in fixOpStack() "); + } + + if (p <= RBBINode::precLParen) { + // Scan is at a right paren or end of expression. + // The scanned item must match the stack, or else there was an error. + // Discard the left paren (or start expr) node from the stack, + // leaving the completed (sub)expression as TOS. + if (n->fPrecedence != p) { + // Right paren encountered matched start of expression node, or + // end of expression matched with a left paren node. + error(U_BRK_MISMATCHED_PAREN); + } + fNodeStack[fNodeStackPtr-1] = fNodeStack[fNodeStackPtr]; + fNodeStackPtr--; + // Delete the now-discarded LParen or Start node. + delete n; + } + // printNodeStack("leaving fixOpStack()"); +} + + + + +//------------------------------------------------------------------------------ +// +// findSetFor given a UnicodeString, +// - find the corresponding Unicode Set (uset node) +// (create one if necessary) +// - Set fLeftChild of the caller's node (should be a setRef node) +// to the uset node +// Maintain a hash table of uset nodes, so the same one is always used +// for the same string. +// If a "to adopt" set is provided and we haven't seen this key before, +// add the provided set to the hash table. +// If the string is one (32 bit) char in length, the set contains +// just one element which is the char in question. +// If the string is "any", return a set containing all chars. +// +//------------------------------------------------------------------------------ +void RBBIRuleScanner::findSetFor(const UnicodeString &s, RBBINode *node, UnicodeSet *setToAdopt) { + + RBBISetTableEl *el; + + // First check whether we've already cached a set for this string. + // If so, just use the cached set in the new node. + // delete any set provided by the caller, since we own it. + el = (RBBISetTableEl *)uhash_get(fSetTable, &s); + if (el != nullptr) { + delete setToAdopt; + node->fLeftChild = el->val; + U_ASSERT(node->fLeftChild->fType == RBBINode::uset); + return; + } + + // Haven't seen this set before. + // If the caller didn't provide us with a prebuilt set, + // create a new UnicodeSet now. + if (setToAdopt == nullptr) { + if (s.compare(kAny, -1) == 0) { + setToAdopt = new UnicodeSet(0x000000, 0x10ffff); + } else { + UChar32 c; + c = s.char32At(0); + setToAdopt = new UnicodeSet(c, c); + } + } + + // + // Make a new uset node to refer to this UnicodeSet + // This new uset node becomes the child of the caller's setReference node. + // + RBBINode *usetNode = new RBBINode(RBBINode::uset); + if (usetNode == nullptr) { + error(U_MEMORY_ALLOCATION_ERROR); + return; + } + usetNode->fInputSet = setToAdopt; + usetNode->fParent = node; + node->fLeftChild = usetNode; + usetNode->fText = s; + + + // + // Add the new uset node to the list of all uset nodes. + // + fRB->fUSetNodes->addElement(usetNode, *fRB->fStatus); + + + // + // Add the new set to the set hash table. + // + el = (RBBISetTableEl *)uprv_malloc(sizeof(RBBISetTableEl)); + UnicodeString *tkey = new UnicodeString(s); + if (tkey == nullptr || el == nullptr || setToAdopt == nullptr) { + // Delete to avoid memory leak + delete tkey; + tkey = nullptr; + uprv_free(el); + el = nullptr; + delete setToAdopt; + setToAdopt = nullptr; + + error(U_MEMORY_ALLOCATION_ERROR); + return; + } + el->key = tkey; + el->val = usetNode; + uhash_put(fSetTable, el->key, el, fRB->fStatus); + + return; +} + + + +// +// Assorted Unicode character constants. +// Numeric because there is no portable way to enter them as literals. +// (Think EBCDIC). +// +static const char16_t chCR = 0x0d; // New lines, for terminating comments. +static const char16_t chLF = 0x0a; +static const char16_t chNEL = 0x85; // NEL newline variant +static const char16_t chLS = 0x2028; // Unicode Line Separator +static const char16_t chApos = 0x27; // single quote, for quoted chars. +static const char16_t chPound = 0x23; // '#', introduces a comment. +static const char16_t chBackSlash = 0x5c; // '\' introduces a char escape +static const char16_t chLParen = 0x28; +static const char16_t chRParen = 0x29; + + +//------------------------------------------------------------------------------ +// +// stripRules Return a rules string without extra spaces. +// (Comments are removed separately, during rule parsing.) +// +//------------------------------------------------------------------------------ +UnicodeString RBBIRuleScanner::stripRules(const UnicodeString &rules) { + UnicodeString strippedRules; + int32_t rulesLength = rules.length(); + + for (int32_t idx=0; idx= fRB->fRules.length()) { + return (UChar32)-1; + } + ch = fRB->fRules.char32At(fNextIndex); + if (U_IS_SURROGATE(ch)) { + error(U_ILLEGAL_CHAR_FOUND); + return U_SENTINEL; + } + fNextIndex = fRB->fRules.moveIndex32(fNextIndex, 1); + + if (ch == chCR || + ch == chNEL || + ch == chLS || + (ch == chLF && fLastChar != chCR)) { + // Character is starting a new line. Bump up the line number, and + // reset the column to 0. + fLineNum++; + fCharNum=0; + if (fQuoteMode) { + error(U_BRK_NEW_LINE_IN_QUOTED_STRING); + fQuoteMode = false; + } + } + else { + // Character is not starting a new line. Except in the case of a + // LF following a CR, increment the column position. + if (ch != chLF) { + fCharNum++; + } + } + fLastChar = ch; + return ch; +} + + +//------------------------------------------------------------------------------ +// +// nextChar for rules scanning. At this level, we handle stripping +// out comments and processing backslash character escapes. +// The rest of the rules grammar is handled at the next level up. +// +//------------------------------------------------------------------------------ +void RBBIRuleScanner::nextChar(RBBIRuleChar &c) { + + // Unicode Character constants needed for the processing done by nextChar(), + // in hex because literals wont work on EBCDIC machines. + + fScanIndex = fNextIndex; + c.fChar = nextCharLL(); + c.fEscaped = false; + + // + // check for '' sequence. + // These are recognized in all contexts, whether in quoted text or not. + // + if (c.fChar == chApos) { + if (fRB->fRules.char32At(fNextIndex) == chApos) { + c.fChar = nextCharLL(); // get nextChar officially so character counts + c.fEscaped = true; // stay correct. + } + else + { + // Single quote, by itself. + // Toggle quoting mode. + // Return either '(' or ')', because quotes cause a grouping of the quoted text. + fQuoteMode = !fQuoteMode; + if (fQuoteMode) { + c.fChar = chLParen; + } else { + c.fChar = chRParen; + } + c.fEscaped = false; // The paren that we return is not escaped. + return; + } + } + + if (fQuoteMode) { + c.fEscaped = true; + } + else + { + // We are not in a 'quoted region' of the source. + // + if (c.fChar == chPound) { + // Start of a comment. Consume the rest of it. + // The new-line char that terminates the comment is always returned. + // It will be treated as white-space, and serves to break up anything + // that might otherwise incorrectly clump together with a comment in + // the middle (a variable name, for example.) + int32_t commentStart = fScanIndex; + for (;;) { + c.fChar = nextCharLL(); + if (c.fChar == (UChar32)-1 || // EOF + c.fChar == chCR || + c.fChar == chLF || + c.fChar == chNEL || + c.fChar == chLS) {break;} + } + for (int32_t i=commentStart; ifStrippedRules.setCharAt(i, u' '); + } + } + if (c.fChar == (UChar32)-1) { + return; + } + + // + // check for backslash escaped characters. + // Use UnicodeString::unescapeAt() to handle them. + // + if (c.fChar == chBackSlash) { + c.fEscaped = true; + int32_t startX = fNextIndex; + c.fChar = fRB->fRules.unescapeAt(fNextIndex); + if (fNextIndex == startX) { + error(U_BRK_HEX_DIGITS_EXPECTED); + } + fCharNum += fNextIndex-startX; + } + } + // putc(c.fChar, stdout); +} + +//------------------------------------------------------------------------------ +// +// Parse RBBI rules. The state machine for rules parsing is here. +// The state tables are hand-written in the file rbbirpt.txt, +// and converted to the form used here by a perl +// script rbbicst.pl +// +//------------------------------------------------------------------------------ +void RBBIRuleScanner::parse() { + uint16_t state; + const RBBIRuleTableEl *tableEl; + + if (U_FAILURE(*fRB->fStatus)) { + return; + } + + state = 1; + nextChar(fC); + // + // Main loop for the rule parsing state machine. + // Runs once per state transition. + // Each time through optionally performs, depending on the state table, + // - an advance to the the next input char + // - an action to be performed. + // - pushing or popping a state to/from the local state return stack. + // + for (;;) { + // Bail out if anything has gone wrong. + // RBBI rule file parsing stops on the first error encountered. + if (U_FAILURE(*fRB->fStatus)) { + break; + } + + // Quit if state == 0. This is the normal way to exit the state machine. + // + if (state == 0) { + break; + } + + // Find the state table element that matches the input char from the rule, or the + // class of the input character. Start with the first table row for this + // state, then linearly scan forward until we find a row that matches the + // character. The last row for each state always matches all characters, so + // the search will stop there, if not before. + // + tableEl = &gRuleParseStateTable[state]; + #ifdef RBBI_DEBUG + if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "scan")) { + RBBIDebugPrintf("char, line, col = (\'%c\', %d, %d) state=%s ", + fC.fChar, fLineNum, fCharNum, RBBIRuleStateNames[state]); + } + #endif + + for (;;) { + #ifdef RBBI_DEBUG + if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "scan")) { RBBIDebugPrintf("."); fflush(stdout);} + #endif + if (tableEl->fCharClass < 127 && fC.fEscaped == false && tableEl->fCharClass == fC.fChar) { + // Table row specified an individual character, not a set, and + // the input character is not escaped, and + // the input character matched it. + break; + } + if (tableEl->fCharClass == 255) { + // Table row specified default, match anything character class. + break; + } + if (tableEl->fCharClass == 254 && fC.fEscaped) { + // Table row specified "escaped" and the char was escaped. + break; + } + if (tableEl->fCharClass == 253 && fC.fEscaped && + (fC.fChar == 0x50 || fC.fChar == 0x70 )) { + // Table row specified "escaped P" and the char is either 'p' or 'P'. + break; + } + if (tableEl->fCharClass == 252 && fC.fChar == (UChar32)-1) { + // Table row specified eof and we hit eof on the input. + break; + } + + if (tableEl->fCharClass >= 128 && tableEl->fCharClass < 240 && // Table specs a char class && + fC.fEscaped == false && // char is not escaped && + fC.fChar != (UChar32)-1) { // char is not EOF + U_ASSERT((tableEl->fCharClass-128) < UPRV_LENGTHOF(fRuleSets)); + if (fRuleSets[tableEl->fCharClass-128].contains(fC.fChar)) { + // Table row specified a character class, or set of characters, + // and the current char matches it. + break; + } + } + + // No match on this row, advance to the next row for this state, + tableEl++; + } + if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "scan")) { RBBIDebugPuts("");} + + // + // We've found the row of the state table that matches the current input + // character from the rules string. + // Perform any action specified by this row in the state table. + if (doParseActions((int32_t)tableEl->fAction) == false) { + // Break out of the state machine loop if the + // the action signalled some kind of error, or + // the action was to exit, occurs on normal end-of-rules-input. + break; + } + + if (tableEl->fPushState != 0) { + fStackPtr++; + if (fStackPtr >= kStackSize) { + error(U_BRK_INTERNAL_ERROR); + RBBIDebugPuts("RBBIRuleScanner::parse() - state stack overflow."); + fStackPtr--; + } + fStack[fStackPtr] = tableEl->fPushState; + } + + if (tableEl->fNextChar) { + nextChar(fC); + } + + // Get the next state from the table entry, or from the + // state stack if the next state was specified as "pop". + if (tableEl->fNextState != 255) { + state = tableEl->fNextState; + } else { + state = fStack[fStackPtr]; + fStackPtr--; + if (fStackPtr < 0) { + error(U_BRK_INTERNAL_ERROR); + RBBIDebugPuts("RBBIRuleScanner::parse() - state stack underflow."); + fStackPtr++; + } + } + + } + + if (U_FAILURE(*fRB->fStatus)) { + return; + } + + // If there are no forward rules set an error. + // + if (fRB->fForwardTree == nullptr) { + error(U_BRK_RULE_SYNTAX); + return; + } + + // + // Parsing of the input RBBI rules is complete. + // We now have a parse tree for the rule expressions + // and a list of all UnicodeSets that are referenced. + // +#ifdef RBBI_DEBUG + if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "symbols")) {fSymbolTable->rbbiSymtablePrint();} + if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "ptree")) { + RBBIDebugPrintf("Completed Forward Rules Parse Tree...\n"); + RBBINode::printTree(fRB->fForwardTree, true); + RBBIDebugPrintf("\nCompleted Reverse Rules Parse Tree...\n"); + RBBINode::printTree(fRB->fReverseTree, true); + RBBIDebugPrintf("\nCompleted Safe Point Forward Rules Parse Tree...\n"); + RBBINode::printTree(fRB->fSafeFwdTree, true); + RBBIDebugPrintf("\nCompleted Safe Point Reverse Rules Parse Tree...\n"); + RBBINode::printTree(fRB->fSafeRevTree, true); + } +#endif +} + + +//------------------------------------------------------------------------------ +// +// printNodeStack for debugging... +// +//------------------------------------------------------------------------------ +#ifdef RBBI_DEBUG +void RBBIRuleScanner::printNodeStack(const char *title) { + int i; + RBBIDebugPrintf("%s. Dumping node stack...\n", title); + for (i=fNodeStackPtr; i>0; i--) {RBBINode::printTree(fNodeStack[i], true);} +} +#endif + + + + +//------------------------------------------------------------------------------ +// +// pushNewNode create a new RBBINode of the specified type and push it +// onto the stack of nodes. +// +//------------------------------------------------------------------------------ +RBBINode *RBBIRuleScanner::pushNewNode(RBBINode::NodeType t) { + if (U_FAILURE(*fRB->fStatus)) { + return nullptr; + } + if (fNodeStackPtr >= kStackSize - 1) { + error(U_BRK_RULE_SYNTAX); + RBBIDebugPuts("RBBIRuleScanner::pushNewNode - stack overflow."); + return nullptr; + } + fNodeStackPtr++; + fNodeStack[fNodeStackPtr] = new RBBINode(t); + if (fNodeStack[fNodeStackPtr] == nullptr) { + *fRB->fStatus = U_MEMORY_ALLOCATION_ERROR; + } + return fNodeStack[fNodeStackPtr]; +} + + + +//------------------------------------------------------------------------------ +// +// scanSet Construct a UnicodeSet from the text at the current scan +// position. Advance the scan position to the first character +// after the set. +// +// A new RBBI setref node referring to the set is pushed onto the node +// stack. +// +// The scan position is normally under the control of the state machine +// that controls rule parsing. UnicodeSets, however, are parsed by +// the UnicodeSet constructor, not by the RBBI rule parser. +// +//------------------------------------------------------------------------------ +void RBBIRuleScanner::scanSet() { + UnicodeSet *uset; + ParsePosition pos; + int startPos; + int i; + + if (U_FAILURE(*fRB->fStatus)) { + return; + } + + pos.setIndex(fScanIndex); + startPos = fScanIndex; + UErrorCode localStatus = U_ZERO_ERROR; + uset = new UnicodeSet(); + if (uset == nullptr) { + localStatus = U_MEMORY_ALLOCATION_ERROR; + } else { + uset->applyPatternIgnoreSpace(fRB->fRules, pos, fSymbolTable, localStatus); + } + if (U_FAILURE(localStatus)) { + // TODO: Get more accurate position of the error from UnicodeSet's return info. + // UnicodeSet appears to not be reporting correctly at this time. + #ifdef RBBI_DEBUG + RBBIDebugPrintf("UnicodeSet parse position.ErrorIndex = %d\n", pos.getIndex()); + #endif + error(localStatus); + delete uset; + return; + } + + // Verify that the set contains at least one code point. + // + U_ASSERT(uset!=nullptr); + if (uset->isEmpty()) { + // This set is empty. + // Make it an error, because it almost certainly is not what the user wanted. + // Also, avoids having to think about corner cases in the tree manipulation code + // that occurs later on. + error(U_BRK_RULE_EMPTY_SET); + delete uset; + return; + } + + + // Advance the RBBI parse position over the UnicodeSet pattern. + // Don't just set fScanIndex because the line/char positions maintained + // for error reporting would be thrown off. + i = pos.getIndex(); + for (;;) { + if (fNextIndex >= i) { + break; + } + nextCharLL(); + } + + if (U_SUCCESS(*fRB->fStatus)) { + RBBINode *n; + + n = pushNewNode(RBBINode::setRef); + if (U_FAILURE(*fRB->fStatus)) { + return; + } + n->fFirstPos = startPos; + n->fLastPos = fNextIndex; + fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText); + // findSetFor() serves several purposes here: + // - Adopts storage for the UnicodeSet, will be responsible for deleting. + // - Maintains collection of all sets in use, needed later for establishing + // character categories for run time engine. + // - Eliminates mulitiple instances of the same set. + // - Creates a new uset node if necessary (if this isn't a duplicate.) + findSetFor(n->fText, n, uset); + } + +} + +int32_t RBBIRuleScanner::numRules() { + return fRuleNum; +} + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ diff --git a/intl/icu/source/common/rbbiscan.h b/intl/icu/source/common/rbbiscan.h new file mode 100644 index 0000000000..8a419b9d76 --- /dev/null +++ b/intl/icu/source/common/rbbiscan.h @@ -0,0 +1,167 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// +// rbbiscan.h +// +// Copyright (C) 2002-2016, International Business Machines Corporation and others. +// All Rights Reserved. +// +// This file contains declarations for class RBBIRuleScanner +// + + +#ifndef RBBISCAN_H +#define RBBISCAN_H + +#include "unicode/utypes.h" +#include "unicode/uobject.h" +#include "unicode/rbbi.h" +#include "unicode/uniset.h" +#include "unicode/parseerr.h" +#include "uhash.h" +#include "uvector.h" +#include "unicode/symtable.h"// For UnicodeSet parsing, is the interface that + // looks up references to $variables within a set. +#include "rbbinode.h" +#include "rbbirpt.h" + +U_NAMESPACE_BEGIN + +class RBBIRuleBuilder; +class RBBISymbolTable; + + +//-------------------------------------------------------------------------------- +// +// class RBBIRuleScanner does the lowest level, character-at-a-time +// scanning of break iterator rules. +// +// The output of the scanner is parse trees for +// the rule expressions and a list of all Unicode Sets +// encountered. +// +//-------------------------------------------------------------------------------- + +class RBBIRuleScanner : public UMemory { +public: + + enum { + kStackSize = 100 // The size of the state stack for + }; // rules parsing. Corresponds roughly + // to the depth of parentheses nesting + // that is allowed in the rules. + + struct RBBIRuleChar { + UChar32 fChar; + UBool fEscaped; + RBBIRuleChar() : fChar(0), fEscaped(false) {} + }; + + RBBIRuleScanner(RBBIRuleBuilder *rb); + + + virtual ~RBBIRuleScanner(); + + void nextChar(RBBIRuleChar &c); // Get the next char from the input stream. + // Return false if at end. + + UBool push(const RBBIRuleChar &c); // Push (unget) one character. + // Only a single character may be pushed. + + void parse(); // Parse the rules, generating two parse + // trees, one each for the forward and + // reverse rules, + // and a list of UnicodeSets encountered. + + int32_t numRules(); // Return the number of rules that have been seen. + + /** + * Return a rules string without unnecessary + * characters. + */ + static UnicodeString stripRules(const UnicodeString &rules); +private: + + UBool doParseActions(int32_t a); + void error(UErrorCode e); // error reporting convenience function. + void fixOpStack(RBBINode::OpPrecedence p); + // a character. + void findSetFor(const UnicodeString &s, RBBINode *node, UnicodeSet *setToAdopt = nullptr); + + UChar32 nextCharLL(); +#ifdef RBBI_DEBUG + void printNodeStack(const char *title); +#endif + RBBINode *pushNewNode(RBBINode::NodeType t); + void scanSet(); + + + RBBIRuleBuilder *fRB; // The rule builder that we are part of. + + int32_t fScanIndex; // Index of current character being processed + // in the rule input string. + int32_t fNextIndex; // Index of the next character, which + // is the first character not yet scanned. + UBool fQuoteMode; // Scan is in a 'quoted region' + int32_t fLineNum; // Line number in input file. + int32_t fCharNum; // Char position within the line. + UChar32 fLastChar; // Previous char, needed to count CR-LF + // as a single line, not two. + + RBBIRuleChar fC; // Current char for parse state machine + // processing. + UnicodeString fVarName; // $variableName, valid when we've just + // scanned one. + + RBBIRuleTableEl **fStateTable; // State Transition Table for RBBI Rule + // parsing. index by p[state][char-class] + + uint16_t fStack[kStackSize]; // State stack, holds state pushes + int32_t fStackPtr; // and pops as specified in the state + // transition rules. + + RBBINode *fNodeStack[kStackSize]; // Node stack, holds nodes created + // during the parse of a rule + int32_t fNodeStackPtr; + + + UBool fReverseRule; // True if the rule currently being scanned + // is a reverse direction rule (if it + // starts with a '!') + + UBool fLookAheadRule; // True if the rule includes a '/' + // somewhere within it. + + UBool fNoChainInRule; // True if the current rule starts with a '^'. + + RBBISymbolTable *fSymbolTable; // symbol table, holds definitions of + // $variable symbols. + + UHashtable *fSetTable; // UnicocodeSet hash table, holds indexes to + // the sets created while parsing rules. + // The key is the string used for creating + // the set. + + UnicodeSet fRuleSets[10]; // Unicode Sets that are needed during + // the scanning of RBBI rules. The + // indices for these are assigned by the + // perl script that builds the state tables. + // See rbbirpt.h. + + int32_t fRuleNum; // Counts each rule as it is scanned. + + int32_t fOptionStart; // Input index of start of a !!option + // keyword, while being scanned. + + UnicodeSet *gRuleSet_rule_char; + UnicodeSet *gRuleSet_white_space; + UnicodeSet *gRuleSet_name_char; + UnicodeSet *gRuleSet_name_start_char; + + RBBIRuleScanner(const RBBIRuleScanner &other) = delete; // forbid copying of this class + RBBIRuleScanner &operator=(const RBBIRuleScanner &other) = delete; // forbid copying of this class +}; + +U_NAMESPACE_END + +#endif diff --git a/intl/icu/source/common/rbbisetb.cpp b/intl/icu/source/common/rbbisetb.cpp new file mode 100644 index 0000000000..e6c72954bc --- /dev/null +++ b/intl/icu/source/common/rbbisetb.cpp @@ -0,0 +1,694 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// +// rbbisetb.cpp +// +/* +*************************************************************************** +* Copyright (C) 2002-2008 International Business Machines Corporation * +* and others. All rights reserved. * +*************************************************************************** +*/ +// +// RBBISetBuilder Handles processing of Unicode Sets from RBBI rules +// (part of the rule building process.) +// +// Starting with the rules parse tree from the scanner, +// +// - Enumerate the set of UnicodeSets that are referenced +// by the RBBI rules. +// - compute a set of non-overlapping character ranges +// with all characters within a range belonging to the same +// set of input unicode sets. +// - Derive a set of non-overlapping UnicodeSet (like things) +// that will correspond to columns in the state table for +// the RBBI execution engine. All characters within one +// of these sets belong to the same set of the original +// UnicodeSets from the user's rules. +// - construct the trie table that maps input characters +// to the index of the matching non-overlapping set of set from +// the previous step. +// + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/uniset.h" +#include "uvector.h" +#include "uassert.h" +#include "cmemory.h" +#include "cstring.h" + +#include "rbbisetb.h" +#include "rbbinode.h" + +U_NAMESPACE_BEGIN + +const int32_t kMaxCharCategoriesFor8BitsTrie = 255; +//------------------------------------------------------------------------ +// +// Constructor +// +//------------------------------------------------------------------------ +RBBISetBuilder::RBBISetBuilder(RBBIRuleBuilder *rb) +{ + fRB = rb; + fStatus = rb->fStatus; + fRangeList = nullptr; + fMutableTrie = nullptr; + fTrie = nullptr; + fTrieSize = 0; + fGroupCount = 0; + fSawBOF = false; +} + + +//------------------------------------------------------------------------ +// +// Destructor +// +//------------------------------------------------------------------------ +RBBISetBuilder::~RBBISetBuilder() +{ + RangeDescriptor *nextRangeDesc; + + // Walk through & delete the linked list of RangeDescriptors + for (nextRangeDesc = fRangeList; nextRangeDesc!=nullptr;) { + RangeDescriptor *r = nextRangeDesc; + nextRangeDesc = r->fNext; + delete r; + } + + ucptrie_close(fTrie); + umutablecptrie_close(fMutableTrie); +} + + + + +//------------------------------------------------------------------------ +// +// build Build the list of non-overlapping character ranges +// from the Unicode Sets. +// +//------------------------------------------------------------------------ +void RBBISetBuilder::buildRanges() { + RBBINode *usetNode; + RangeDescriptor *rlRange; + + if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "usets")) {printSets();} + + // + // Initialize the process by creating a single range encompassing all characters + // that is in no sets. + // + fRangeList = new RangeDescriptor(*fStatus); // will check for status here + if (fRangeList == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + return; + } + fRangeList->fStartChar = 0; + fRangeList->fEndChar = 0x10ffff; + + if (U_FAILURE(*fStatus)) { + return; + } + + // + // Find the set of non-overlapping ranges of characters + // + int ni; + for (ni=0; ; ni++) { // Loop over each of the UnicodeSets encountered in the input rules + usetNode = (RBBINode *)this->fRB->fUSetNodes->elementAt(ni); + if (usetNode==nullptr) { + break; + } + + UnicodeSet *inputSet = usetNode->fInputSet; + int32_t inputSetRangeCount = inputSet->getRangeCount(); + int inputSetRangeIndex = 0; + rlRange = fRangeList; + + for (;;) { + if (inputSetRangeIndex >= inputSetRangeCount) { + break; + } + UChar32 inputSetRangeBegin = inputSet->getRangeStart(inputSetRangeIndex); + UChar32 inputSetRangeEnd = inputSet->getRangeEnd(inputSetRangeIndex); + + // skip over ranges from the range list that are completely + // below the current range from the input unicode set. + while (rlRange->fEndChar < inputSetRangeBegin) { + rlRange = rlRange->fNext; + } + + // If the start of the range from the range list is before with + // the start of the range from the unicode set, split the range list range + // in two, with one part being before (wholly outside of) the unicode set + // and the other containing the rest. + // Then continue the loop; the post-split current range will then be skipped + // over + if (rlRange->fStartChar < inputSetRangeBegin) { + rlRange->split(inputSetRangeBegin, *fStatus); + if (U_FAILURE(*fStatus)) { + return; + } + continue; + } + + // Same thing at the end of the ranges... + // If the end of the range from the range list doesn't coincide with + // the end of the range from the unicode set, split the range list + // range in two. The first part of the split range will be + // wholly inside the Unicode set. + if (rlRange->fEndChar > inputSetRangeEnd) { + rlRange->split(inputSetRangeEnd+1, *fStatus); + if (U_FAILURE(*fStatus)) { + return; + } + } + + // The current rlRange is now entirely within the UnicodeSet range. + // Add this unicode set to the list of sets for this rlRange + if (rlRange->fIncludesSets->indexOf(usetNode) == -1) { + rlRange->fIncludesSets->addElement(usetNode, *fStatus); + if (U_FAILURE(*fStatus)) { + return; + } + } + + // Advance over ranges that we are finished with. + if (inputSetRangeEnd == rlRange->fEndChar) { + inputSetRangeIndex++; + } + rlRange = rlRange->fNext; + } + } + + if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "range")) { printRanges();} + + // + // Group the above ranges, with each group consisting of one or more + // ranges that are in exactly the same set of original UnicodeSets. + // The groups are numbered, and these group numbers are the set of + // input symbols recognized by the run-time state machine. + // + // Numbering: # 0 (state table column 0) is unused. + // # 1 is reserved - table column 1 is for end-of-input + // # 2 is reserved - table column 2 is for beginning-of-input + // # 3 is the first range list. + // + RangeDescriptor *rlSearchRange; + int32_t dictGroupCount = 0; + + for (rlRange = fRangeList; rlRange!=nullptr; rlRange=rlRange->fNext) { + for (rlSearchRange=fRangeList; rlSearchRange != rlRange; rlSearchRange=rlSearchRange->fNext) { + if (rlRange->fIncludesSets->equals(*rlSearchRange->fIncludesSets)) { + rlRange->fNum = rlSearchRange->fNum; + rlRange->fIncludesDict = rlSearchRange->fIncludesDict; + break; + } + } + if (rlRange->fNum == 0) { + rlRange->fFirstInGroup = true; + if (rlRange->isDictionaryRange()) { + rlRange->fNum = ++dictGroupCount; + rlRange->fIncludesDict = true; + } else { + fGroupCount++; + rlRange->fNum = fGroupCount+2; + addValToSets(rlRange->fIncludesSets, rlRange->fNum); + } + } + } + + // Move the character category numbers for any dictionary ranges up, so that they + // immediately follow the non-dictionary ranges. + + fDictCategoriesStart = fGroupCount + 3; + for (rlRange = fRangeList; rlRange!=nullptr; rlRange=rlRange->fNext) { + if (rlRange->fIncludesDict) { + rlRange->fNum += fDictCategoriesStart - 1; + if (rlRange->fFirstInGroup) { + addValToSets(rlRange->fIncludesSets, rlRange->fNum); + } + } + } + fGroupCount += dictGroupCount; + + + // Handle input sets that contain the special string {eof}. + // Column 1 of the state table is reserved for EOF on input. + // Column 2 is reserved for before-the-start-input. + // (This column can be optimized away later if there are no rule + // references to {bof}.) + // Add this column value (1 or 2) to the equivalent expression + // subtree for each UnicodeSet that contains the string {eof} + // Because {bof} and {eof} are not characters in the normal sense, + // they don't affect the computation of the ranges or TRIE. + + UnicodeString eofString(u"eof"); + UnicodeString bofString(u"bof"); + for (ni=0; ; ni++) { // Loop over each of the UnicodeSets encountered in the input rules + usetNode = (RBBINode *)this->fRB->fUSetNodes->elementAt(ni); + if (usetNode==nullptr) { + break; + } + UnicodeSet *inputSet = usetNode->fInputSet; + if (inputSet->contains(eofString)) { + addValToSet(usetNode, 1); + } + if (inputSet->contains(bofString)) { + addValToSet(usetNode, 2); + fSawBOF = true; + } + } + + + if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "rgroup")) {printRangeGroups();} + if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "esets")) {printSets();} +} + + +// +// Build the Trie table for mapping UChar32 values to the corresponding +// range group number. +// +void RBBISetBuilder::buildTrie() { + fMutableTrie = umutablecptrie_open( + 0, // Initial value for all code points. + 0, // Error value for out-of-range input. + fStatus); + + for (RangeDescriptor *range = fRangeList; range!=nullptr && U_SUCCESS(*fStatus); range=range->fNext) { + umutablecptrie_setRange(fMutableTrie, + range->fStartChar, // Range start + range->fEndChar, // Range end (inclusive) + range->fNum, // value for range + fStatus); + } +} + + +void RBBISetBuilder::mergeCategories(IntPair categories) { + U_ASSERT(categories.first >= 1); + U_ASSERT(categories.second > categories.first); + U_ASSERT((categories.first < fDictCategoriesStart && categories.second < fDictCategoriesStart) || + (categories.first >= fDictCategoriesStart && categories.second >= fDictCategoriesStart)); + + for (RangeDescriptor *rd = fRangeList; rd != nullptr; rd = rd->fNext) { + int32_t rangeNum = rd->fNum; + if (rangeNum == categories.second) { + rd->fNum = categories.first; + } else if (rangeNum > categories.second) { + rd->fNum--; + } + } + --fGroupCount; + if (categories.second <= fDictCategoriesStart) { + --fDictCategoriesStart; + } +} + + +//----------------------------------------------------------------------------------- +// +// getTrieSize() Return the size that will be required to serialize the Trie. +// +//----------------------------------------------------------------------------------- +int32_t RBBISetBuilder::getTrieSize() { + if (U_FAILURE(*fStatus)) { + return 0; + } + if (fTrie == nullptr) { + bool use8Bits = getNumCharCategories() <= kMaxCharCategoriesFor8BitsTrie; + fTrie = umutablecptrie_buildImmutable( + fMutableTrie, + UCPTRIE_TYPE_FAST, + use8Bits ? UCPTRIE_VALUE_BITS_8 : UCPTRIE_VALUE_BITS_16, + fStatus); + fTrieSize = ucptrie_toBinary(fTrie, nullptr, 0, fStatus); + if (*fStatus == U_BUFFER_OVERFLOW_ERROR) { + *fStatus = U_ZERO_ERROR; + } + } + return fTrieSize; +} + + +//----------------------------------------------------------------------------------- +// +// serializeTrie() Put the serialized trie at the specified address. +// Trust the caller to have given us enough memory. +// getTrieSize() MUST be called first. +// +//----------------------------------------------------------------------------------- +void RBBISetBuilder::serializeTrie(uint8_t *where) { + ucptrie_toBinary(fTrie, + where, // Buffer + fTrieSize, // Capacity + fStatus); +} + +//------------------------------------------------------------------------ +// +// addValToSets Add a runtime-mapped input value to each uset from a +// list of uset nodes. (val corresponds to a state table column.) +// For each of the original Unicode sets - which correspond +// directly to uset nodes - a logically equivalent expression +// is constructed in terms of the remapped runtime input +// symbol set. This function adds one runtime input symbol to +// a list of sets. +// +// The "logically equivalent expression" is the tree for an +// or-ing together of all of the symbols that go into the set. +// +//------------------------------------------------------------------------ +void RBBISetBuilder::addValToSets(UVector *sets, uint32_t val) { + int32_t ix; + + for (ix=0; ixsize(); ix++) { + RBBINode *usetNode = (RBBINode *)sets->elementAt(ix); + addValToSet(usetNode, val); + } +} + +void RBBISetBuilder::addValToSet(RBBINode *usetNode, uint32_t val) { + RBBINode *leafNode = new RBBINode(RBBINode::leafChar); + if (leafNode == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + return; + } + leafNode->fVal = (unsigned short)val; + if (usetNode->fLeftChild == nullptr) { + usetNode->fLeftChild = leafNode; + leafNode->fParent = usetNode; + } else { + // There are already input symbols present for this set. + // Set up an OR node, with the previous stuff as the left child + // and the new value as the right child. + RBBINode *orNode = new RBBINode(RBBINode::opOr); + if (orNode == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + return; + } + orNode->fLeftChild = usetNode->fLeftChild; + orNode->fRightChild = leafNode; + orNode->fLeftChild->fParent = orNode; + orNode->fRightChild->fParent = orNode; + usetNode->fLeftChild = orNode; + orNode->fParent = usetNode; + } +} + + +//------------------------------------------------------------------------ +// +// getNumCharCategories +// +//------------------------------------------------------------------------ +int32_t RBBISetBuilder::getNumCharCategories() const { + return fGroupCount + 3; +} + + +//------------------------------------------------------------------------ +// +// getDictCategoriesStart +// +//------------------------------------------------------------------------ +int32_t RBBISetBuilder::getDictCategoriesStart() const { + return fDictCategoriesStart; +} + + +//------------------------------------------------------------------------ +// +// sawBOF +// +//------------------------------------------------------------------------ +UBool RBBISetBuilder::sawBOF() const { + return fSawBOF; +} + + +//------------------------------------------------------------------------ +// +// getFirstChar Given a runtime RBBI character category, find +// the first UChar32 that is in the set of chars +// in the category. +//------------------------------------------------------------------------ +UChar32 RBBISetBuilder::getFirstChar(int32_t category) const { + RangeDescriptor *rlRange; + UChar32 retVal = (UChar32)-1; + for (rlRange = fRangeList; rlRange!=nullptr; rlRange=rlRange->fNext) { + if (rlRange->fNum == category) { + retVal = rlRange->fStartChar; + break; + } + } + return retVal; +} + + +//------------------------------------------------------------------------ +// +// printRanges A debugging function. +// dump out all of the range definitions. +// +//------------------------------------------------------------------------ +#ifdef RBBI_DEBUG +void RBBISetBuilder::printRanges() { + RangeDescriptor *rlRange; + int i; + + RBBIDebugPrintf("\n\n Nonoverlapping Ranges ...\n"); + for (rlRange = fRangeList; rlRange!=nullptr; rlRange=rlRange->fNext) { + RBBIDebugPrintf("%4x-%4x ", rlRange->fStartChar, rlRange->fEndChar); + + for (i=0; ifIncludesSets->size(); i++) { + RBBINode *usetNode = (RBBINode *)rlRange->fIncludesSets->elementAt(i); + UnicodeString setName {u"anon"}; + RBBINode *setRef = usetNode->fParent; + if (setRef != nullptr) { + RBBINode *varRef = setRef->fParent; + if (varRef != nullptr && varRef->fType == RBBINode::varRef) { + setName = varRef->fText; + } + } + RBBI_DEBUG_printUnicodeString(setName); RBBIDebugPrintf(" "); + } + RBBIDebugPrintf("\n"); + } +} +#endif + + +//------------------------------------------------------------------------ +// +// printRangeGroups A debugging function. +// dump out all of the range groups. +// +//------------------------------------------------------------------------ +#ifdef RBBI_DEBUG +void RBBISetBuilder::printRangeGroups() { + int i; + + RBBIDebugPrintf("\nRanges grouped by Unicode Set Membership...\n"); + for (RangeDescriptor *rlRange = fRangeList; rlRange!=nullptr; rlRange=rlRange->fNext) { + if (rlRange->fFirstInGroup) { + int groupNum = rlRange->fNum; + RBBIDebugPrintf("%2i ", groupNum); + + if (groupNum >= fDictCategoriesStart) { RBBIDebugPrintf(" ");} + + for (i=0; ifIncludesSets->size(); i++) { + RBBINode *usetNode = (RBBINode *)rlRange->fIncludesSets->elementAt(i); + UnicodeString setName = UNICODE_STRING("anon", 4); + RBBINode *setRef = usetNode->fParent; + if (setRef != nullptr) { + RBBINode *varRef = setRef->fParent; + if (varRef != nullptr && varRef->fType == RBBINode::varRef) { + setName = varRef->fText; + } + } + RBBI_DEBUG_printUnicodeString(setName); RBBIDebugPrintf(" "); + } + + i = 0; + for (RangeDescriptor *tRange = rlRange; tRange != nullptr; tRange = tRange->fNext) { + if (tRange->fNum == rlRange->fNum) { + if (i++ % 5 == 0) { + RBBIDebugPrintf("\n "); + } + RBBIDebugPrintf(" %05x-%05x", tRange->fStartChar, tRange->fEndChar); + } + } + RBBIDebugPrintf("\n"); + } + } + RBBIDebugPrintf("\n"); +} +#endif + + +//------------------------------------------------------------------------ +// +// printSets A debugging function. +// dump out all of the set definitions. +// +//------------------------------------------------------------------------ +#ifdef RBBI_DEBUG +void RBBISetBuilder::printSets() { + int i; + + RBBIDebugPrintf("\n\nUnicode Sets List\n------------------\n"); + for (i=0; ; i++) { + RBBINode *usetNode; + RBBINode *setRef; + RBBINode *varRef; + UnicodeString setName; + + usetNode = (RBBINode *)fRB->fUSetNodes->elementAt(i); + if (usetNode == nullptr) { + break; + } + + RBBIDebugPrintf("%3d ", i); + setName = UNICODE_STRING("anonymous", 9); + setRef = usetNode->fParent; + if (setRef != nullptr) { + varRef = setRef->fParent; + if (varRef != nullptr && varRef->fType == RBBINode::varRef) { + setName = varRef->fText; + } + } + RBBI_DEBUG_printUnicodeString(setName); + RBBIDebugPrintf(" "); + RBBI_DEBUG_printUnicodeString(usetNode->fText); + RBBIDebugPrintf("\n"); + if (usetNode->fLeftChild != nullptr) { + RBBINode::printTree(usetNode->fLeftChild, true); + } + } + RBBIDebugPrintf("\n"); +} +#endif + + + +//------------------------------------------------------------------------------------- +// +// RangeDescriptor copy constructor +// +//------------------------------------------------------------------------------------- + +RangeDescriptor::RangeDescriptor(const RangeDescriptor &other, UErrorCode &status) : + fStartChar(other.fStartChar), fEndChar {other.fEndChar}, fNum {other.fNum}, + fIncludesDict{other.fIncludesDict}, fFirstInGroup{other.fFirstInGroup} { + + if (U_FAILURE(status)) { + return; + } + fIncludesSets = new UVector(status); + if (this->fIncludesSets == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + } + if (U_FAILURE(status)) { + return; + } + + for (int32_t i=0; isize(); i++) { + this->fIncludesSets->addElement(other.fIncludesSets->elementAt(i), status); + } +} + + +//------------------------------------------------------------------------------------- +// +// RangeDesriptor default constructor +// +//------------------------------------------------------------------------------------- +RangeDescriptor::RangeDescriptor(UErrorCode &status) { + if (U_FAILURE(status)) { + return; + } + fIncludesSets = new UVector(status); + if (fIncludesSets == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + } +} + + +//------------------------------------------------------------------------------------- +// +// RangeDesriptor Destructor +// +//------------------------------------------------------------------------------------- +RangeDescriptor::~RangeDescriptor() { + delete fIncludesSets; + fIncludesSets = nullptr; +} + +//------------------------------------------------------------------------------------- +// +// RangeDesriptor::split() +// +//------------------------------------------------------------------------------------- +void RangeDescriptor::split(UChar32 where, UErrorCode &status) { + U_ASSERT(where>fStartChar && where<=fEndChar); + RangeDescriptor *nr = new RangeDescriptor(*this, status); + if(nr == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + if (U_FAILURE(status)) { + delete nr; + return; + } + // RangeDescriptor copy constructor copies all fields. + // Only need to update those that are different after the split. + nr->fStartChar = where; + this->fEndChar = where-1; + nr->fNext = this->fNext; + this->fNext = nr; +} + + +//------------------------------------------------------------------------------------- +// +// RangeDescriptor::isDictionaryRange +// +// Test whether this range includes characters from +// the original Unicode Set named "dictionary". +// +// This function looks through the Unicode Sets that +// the range includes, checking for one named "dictionary" +// +// TODO: a faster way would be to find the set node for +// "dictionary" just once, rather than looking it +// up by name every time. +// +//------------------------------------------------------------------------------------- +bool RangeDescriptor::isDictionaryRange() { + static const char16_t *dictionary = u"dictionary"; + for (int32_t i=0; isize(); i++) { + RBBINode *usetNode = (RBBINode *)fIncludesSets->elementAt(i); + RBBINode *setRef = usetNode->fParent; + if (setRef != nullptr) { + RBBINode *varRef = setRef->fParent; + if (varRef && varRef->fType == RBBINode::varRef) { + const UnicodeString *setName = &varRef->fText; + if (setName->compare(dictionary, -1) == 0) { + return true; + } + } + } + } + return false; +} + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ diff --git a/intl/icu/source/common/rbbisetb.h b/intl/icu/source/common/rbbisetb.h new file mode 100644 index 0000000000..cd09d3317a --- /dev/null +++ b/intl/icu/source/common/rbbisetb.h @@ -0,0 +1,147 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// +// rbbisetb.h +/* +********************************************************************** +* Copyright (c) 2001-2005, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ + +#ifndef RBBISETB_H +#define RBBISETB_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/ucptrie.h" +#include "unicode/umutablecptrie.h" +#include "unicode/uobject.h" +#include "rbbirb.h" +#include "uvector.h" + +U_NAMESPACE_BEGIN + +// +// RBBISetBuilder Derives the character categories used by the runtime RBBI engine +// from the Unicode Sets appearing in the source RBBI rules, and +// creates the TRIE table used to map from Unicode to the +// character categories. +// + + +// +// RangeDescriptor +// +// Each of the non-overlapping character ranges gets one of these descriptors. +// All of them are strung together in a linked list, which is kept in order +// (by character) +// +class RangeDescriptor : public UMemory { +public: + UChar32 fStartChar {}; // Start of range, unicode 32 bit value. + UChar32 fEndChar {}; // End of range, unicode 32 bit value. + int32_t fNum {0}; // runtime-mapped input value for this range. + bool fIncludesDict {false}; // True if the range includes $dictionary. + bool fFirstInGroup {false}; // True if first range in a group with the same fNum. + UVector *fIncludesSets {nullptr}; // vector of the the original + // Unicode sets that include this range. + // (Contains ptrs to uset nodes) + RangeDescriptor *fNext {nullptr}; // Next RangeDescriptor in the linked list. + + RangeDescriptor(UErrorCode &status); + RangeDescriptor(const RangeDescriptor &other, UErrorCode &status); + ~RangeDescriptor(); + void split(UChar32 where, UErrorCode &status); // Spit this range in two at "where", with + // where appearing in the second (higher) part. + bool isDictionaryRange(); // Check whether this range appears as part of + // the Unicode set named "dictionary" + + RangeDescriptor(const RangeDescriptor &other) = delete; // forbid default copying of this class + RangeDescriptor &operator=(const RangeDescriptor &other) = delete; // forbid assigning of this class +}; + + +// +// RBBISetBuilder Handles processing of Unicode Sets from RBBI rules. +// +// Starting with the rules parse tree from the scanner, +// +// - Enumerate the set of UnicodeSets that are referenced +// by the RBBI rules. +// - compute a derived set of non-overlapping UnicodeSets +// that will correspond to columns in the state table for +// the RBBI execution engine. +// - construct the trie table that maps input characters +// to set numbers in the non-overlapping set of sets. +// + + +class RBBISetBuilder : public UMemory { +public: + RBBISetBuilder(RBBIRuleBuilder *rb); + ~RBBISetBuilder(); + + void buildRanges(); + void buildTrie(); + void addValToSets(UVector *sets, uint32_t val); + void addValToSet (RBBINode *usetNode, uint32_t val); + int32_t getNumCharCategories() const; // CharCategories are the same as input symbol set to the + // runtime state machine, which are the same as + // columns in the DFA state table + int32_t getDictCategoriesStart() const; // First char category that includes $dictionary, or + // last category + 1 if there are no dictionary categories. + int32_t getTrieSize() /*const*/; // Size in bytes of the serialized Trie. + void serializeTrie(uint8_t *where); // write out the serialized Trie. + UChar32 getFirstChar(int32_t val) const; + UBool sawBOF() const; // Indicate whether any references to the {bof} pseudo + // character were encountered. + /** + * Merge two character categories that have been identified as having equivalent behavior. + * The ranges belonging to the second category (table column) will be added to the first. + * @param categories the pair of categories to be merged. + */ + void mergeCategories(IntPair categories); + +#ifdef RBBI_DEBUG + void printSets(); + void printRanges(); + void printRangeGroups(); +#else + #define printSets() + #define printRanges() + #define printRangeGroups() +#endif + +private: + RBBIRuleBuilder *fRB; // The RBBI Rule Compiler that owns us. + UErrorCode *fStatus; + + RangeDescriptor *fRangeList; // Head of the linked list of RangeDescriptors + + UMutableCPTrie *fMutableTrie; // The mapping TRIE that is the end result of processing + UCPTrie *fTrie; // the Unicode Sets. + uint32_t fTrieSize; + + // Number of range groups, which are groups of ranges that are in the same original UnicodeSets. + int32_t fGroupCount; + + // The number of the first dictionary char category. + // If there are no Dictionary categories, set to the last category + 1. + int32_t fDictCategoriesStart; + + UBool fSawBOF; + + RBBISetBuilder(const RBBISetBuilder &other) = delete; // forbid copying of this class + RBBISetBuilder &operator=(const RBBISetBuilder &other) = delete; // forbid copying of this class +}; + + + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ + +#endif diff --git a/intl/icu/source/common/rbbistbl.cpp b/intl/icu/source/common/rbbistbl.cpp new file mode 100644 index 0000000000..844351bd0d --- /dev/null +++ b/intl/icu/source/common/rbbistbl.cpp @@ -0,0 +1,270 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// +// file: rbbistbl.cpp Implementation of the ICU RBBISymbolTable class +// +/* +*************************************************************************** +* Copyright (C) 2002-2014 International Business Machines Corporation +* and others. All rights reserved. +*************************************************************************** +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/unistr.h" +#include "unicode/uniset.h" +#include "unicode/uchar.h" +#include "unicode/parsepos.h" + +#include "cstr.h" +#include "rbbinode.h" +#include "rbbirb.h" +#include "umutex.h" + + +// +// RBBISymbolTableEntry_deleter Used by the UHashTable to delete the contents +// when the hash table is deleted. +// +U_CDECL_BEGIN +static void U_CALLCONV RBBISymbolTableEntry_deleter(void *p) { + icu::RBBISymbolTableEntry *px = (icu::RBBISymbolTableEntry *)p; + delete px; +} +U_CDECL_END + + + +U_NAMESPACE_BEGIN + +RBBISymbolTable::RBBISymbolTable(RBBIRuleScanner *rs, const UnicodeString &rules, UErrorCode &status) + :fRules(rules), fRuleScanner(rs), ffffString(char16_t(0xffff)) +{ + fHashTable = nullptr; + fCachedSetLookup = nullptr; + + fHashTable = uhash_open(uhash_hashUnicodeString, uhash_compareUnicodeString, nullptr, &status); + // uhash_open checks status + if (U_FAILURE(status)) { + return; + } + uhash_setValueDeleter(fHashTable, RBBISymbolTableEntry_deleter); +} + + + +RBBISymbolTable::~RBBISymbolTable() +{ + uhash_close(fHashTable); +} + + +// +// RBBISymbolTable::lookup This function from the abstract symbol table interface +// looks up a variable name and returns a UnicodeString +// containing the substitution text. +// +// The variable name does NOT include the leading $. +// +const UnicodeString *RBBISymbolTable::lookup(const UnicodeString& s) const +{ + RBBISymbolTableEntry *el; + RBBINode *varRefNode; + RBBINode *exprNode; + RBBINode *usetNode; + const UnicodeString *retString; + RBBISymbolTable *This = (RBBISymbolTable *)this; // cast off const + + el = (RBBISymbolTableEntry *)uhash_get(fHashTable, &s); + if (el == nullptr) { + return nullptr; + } + + varRefNode = el->val; + exprNode = varRefNode->fLeftChild; // Root node of expression for variable + if (exprNode->fType == RBBINode::setRef) { + // The $variable refers to a single UnicodeSet + // return the ffffString, which will subsequently be interpreted as a + // stand-in character for the set by RBBISymbolTable::lookupMatcher() + usetNode = exprNode->fLeftChild; + This->fCachedSetLookup = usetNode->fInputSet; + retString = &ffffString; + } + else + { + // The variable refers to something other than just a set. + // return the original source string for the expression + retString = &exprNode->fText; + This->fCachedSetLookup = nullptr; + } + return retString; +} + + + +// +// RBBISymbolTable::lookupMatcher This function from the abstract symbol table +// interface maps a single stand-in character to a +// pointer to a Unicode Set. The Unicode Set code uses this +// mechanism to get all references to the same $variable +// name to refer to a single common Unicode Set instance. +// +// This implementation cheats a little, and does not maintain a map of stand-in chars +// to sets. Instead, it takes advantage of the fact that the UnicodeSet +// constructor will always call this function right after calling lookup(), +// and we just need to remember what set to return between these two calls. +const UnicodeFunctor *RBBISymbolTable::lookupMatcher(UChar32 ch) const +{ + UnicodeSet *retVal = nullptr; + RBBISymbolTable *This = (RBBISymbolTable *)this; // cast off const + if (ch == 0xffff) { + retVal = fCachedSetLookup; + This->fCachedSetLookup = 0; + } + return retVal; +} + +// +// RBBISymbolTable::parseReference This function from the abstract symbol table interface +// looks for a $variable name in the source text. +// It does not look it up, only scans for it. +// It is used by the UnicodeSet parser. +// +// This implementation is lifted pretty much verbatim +// from the rules based transliterator implementation. +// I didn't see an obvious way of sharing it. +// +UnicodeString RBBISymbolTable::parseReference(const UnicodeString& text, + ParsePosition& pos, int32_t limit) const +{ + int32_t start = pos.getIndex(); + int32_t i = start; + UnicodeString result; + while (i < limit) { + char16_t c = text.charAt(i); + if ((i==start && !u_isIDStart(c)) || !u_isIDPart(c)) { + break; + } + ++i; + } + if (i == start) { // No valid name chars + return result; // Indicate failure with empty string + } + pos.setIndex(i); + text.extractBetween(start, i, result); + return result; +} + + + +// +// RBBISymbolTable::lookupNode Given a key (a variable name), return the +// corresponding RBBI Node. If there is no entry +// in the table for this name, return nullptr. +// +RBBINode *RBBISymbolTable::lookupNode(const UnicodeString &key) const{ + + RBBINode *retNode = nullptr; + RBBISymbolTableEntry *el; + + el = (RBBISymbolTableEntry *)uhash_get(fHashTable, &key); + if (el != nullptr) { + retNode = el->val; + } + return retNode; +} + + +// +// RBBISymbolTable::addEntry Add a new entry to the symbol table. +// Indicate an error if the name already exists - +// this will only occur in the case of duplicate +// variable assignments. +// +void RBBISymbolTable::addEntry (const UnicodeString &key, RBBINode *val, UErrorCode &err) { + RBBISymbolTableEntry *e; + /* test for buffer overflows */ + if (U_FAILURE(err)) { + return; + } + e = (RBBISymbolTableEntry *)uhash_get(fHashTable, &key); + if (e != nullptr) { + err = U_BRK_VARIABLE_REDFINITION; + return; + } + + e = new RBBISymbolTableEntry; + if (e == nullptr) { + err = U_MEMORY_ALLOCATION_ERROR; + return; + } + e->key = key; + e->val = val; + uhash_put( fHashTable, &e->key, e, &err); +} + + +RBBISymbolTableEntry::RBBISymbolTableEntry() : UMemory(), key(), val(nullptr) {} + +RBBISymbolTableEntry::~RBBISymbolTableEntry() { + // The "val" of a symbol table entry is a variable reference node. + // The l. child of the val is the rhs expression from the assignment. + // Unlike other node types, children of variable reference nodes are not + // automatically recursively deleted. We do it manually here. + delete val->fLeftChild; + val->fLeftChild = nullptr; + + delete val; + + // Note: the key UnicodeString is destructed by virtue of being in the object by value. +} + + +// +// RBBISymbolTable::print Debugging function, dump out the symbol table contents. +// +#ifdef RBBI_DEBUG +void RBBISymbolTable::rbbiSymtablePrint() const { + RBBIDebugPrintf("Variable Definitions Symbol Table\n" + "Name Node serial String Val\n" + "-------------------------------------------------------------------\n"); + + int32_t pos = UHASH_FIRST; + const UHashElement *e = nullptr; + for (;;) { + e = uhash_nextElement(fHashTable, &pos); + if (e == nullptr ) { + break; + } + RBBISymbolTableEntry *s = (RBBISymbolTableEntry *)e->value.pointer; + + RBBIDebugPrintf("%-19s %8p %7d ", CStr(s->key)(), (void *)s->val, s->val->fSerialNum); + RBBIDebugPrintf(" %s\n", CStr(s->val->fLeftChild->fText)()); + } + + RBBIDebugPrintf("\nParsed Variable Definitions\n"); + pos = -1; + for (;;) { + e = uhash_nextElement(fHashTable, &pos); + if (e == nullptr ) { + break; + } + RBBISymbolTableEntry *s = (RBBISymbolTableEntry *)e->value.pointer; + RBBIDebugPrintf("%s\n", CStr(s->key)()); + RBBINode::printTree(s->val, true); + RBBINode::printTree(s->val->fLeftChild, false); + RBBIDebugPrintf("\n"); + } +} +#endif + + + + + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ diff --git a/intl/icu/source/common/rbbitblb.cpp b/intl/icu/source/common/rbbitblb.cpp new file mode 100644 index 0000000000..0c2bcff4e5 --- /dev/null +++ b/intl/icu/source/common/rbbitblb.cpp @@ -0,0 +1,1810 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2002-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ +// +// rbbitblb.cpp +// + + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/unistr.h" +#include "rbbitblb.h" +#include "rbbirb.h" +#include "rbbiscan.h" +#include "rbbisetb.h" +#include "rbbidata.h" +#include "cstring.h" +#include "uassert.h" +#include "uvectr32.h" +#include "cmemory.h" + +U_NAMESPACE_BEGIN + +const int32_t kMaxStateFor8BitsTable = 255; + +RBBITableBuilder::RBBITableBuilder(RBBIRuleBuilder *rb, RBBINode **rootNode, UErrorCode &status) : + fRB(rb), + fTree(*rootNode), + fStatus(&status), + fDStates(nullptr), + fSafeTable(nullptr) { + if (U_FAILURE(status)) { + return; + } + // fDStates is UVector + fDStates = new UVector(status); + if (U_SUCCESS(status) && fDStates == nullptr ) { + status = U_MEMORY_ALLOCATION_ERROR; + } +} + + + +RBBITableBuilder::~RBBITableBuilder() { + int i; + for (i=0; isize(); i++) { + delete (RBBIStateDescriptor *)fDStates->elementAt(i); + } + delete fDStates; + delete fSafeTable; + delete fLookAheadRuleMap; +} + + +//----------------------------------------------------------------------------- +// +// RBBITableBuilder::buildForwardTable - This is the main function for building +// the DFA state transition table from the RBBI rules parse tree. +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::buildForwardTable() { + + if (U_FAILURE(*fStatus)) { + return; + } + + // If there were no rules, just return. This situation can easily arise + // for the reverse rules. + if (fTree==nullptr) { + return; + } + + // + // Walk through the tree, replacing any references to $variables with a copy of the + // parse tree for the substitution expression. + // + fTree = fTree->flattenVariables(); +#ifdef RBBI_DEBUG + if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "ftree")) { + RBBIDebugPuts("\nParse tree after flattening variable references."); + RBBINode::printTree(fTree, true); + } +#endif + + // + // If the rules contained any references to {bof} + // add a {bof} to the + // tree. Means that all matches must start out with the + // {bof} fake character. + // + if (fRB->fSetBuilder->sawBOF()) { + RBBINode *bofTop = new RBBINode(RBBINode::opCat); + RBBINode *bofLeaf = new RBBINode(RBBINode::leafChar); + // Delete and exit if memory allocation failed. + if (bofTop == nullptr || bofLeaf == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + delete bofTop; + delete bofLeaf; + return; + } + bofTop->fLeftChild = bofLeaf; + bofTop->fRightChild = fTree; + bofLeaf->fParent = bofTop; + bofLeaf->fVal = 2; // Reserved value for {bof}. + fTree = bofTop; + } + + // + // Add a unique right-end marker to the expression. + // Appears as a cat-node, left child being the original tree, + // right child being the end marker. + // + RBBINode *cn = new RBBINode(RBBINode::opCat); + // Exit if memory allocation failed. + if (cn == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + return; + } + cn->fLeftChild = fTree; + fTree->fParent = cn; + RBBINode *endMarkerNode = cn->fRightChild = new RBBINode(RBBINode::endMark); + // Delete and exit if memory allocation failed. + if (cn->fRightChild == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + delete cn; + return; + } + cn->fRightChild->fParent = cn; + fTree = cn; + + // + // Replace all references to UnicodeSets with the tree for the equivalent + // expression. + // + fTree->flattenSets(); +#ifdef RBBI_DEBUG + if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "stree")) { + RBBIDebugPuts("\nParse tree after flattening Unicode Set references."); + RBBINode::printTree(fTree, true); + } +#endif + + + // + // calculate the functions nullable, firstpos, lastpos and followpos on + // nodes in the parse tree. + // See the algorithm description in Aho. + // Understanding how this works by looking at the code alone will be + // nearly impossible. + // + calcNullable(fTree); + calcFirstPos(fTree); + calcLastPos(fTree); + calcFollowPos(fTree); + if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "pos")) { + RBBIDebugPuts("\n"); + printPosSets(fTree); + } + + // + // For "chained" rules, modify the followPos sets + // + if (fRB->fChainRules) { + calcChainedFollowPos(fTree, endMarkerNode); + } + + // + // BOF (start of input) test fixup. + // + if (fRB->fSetBuilder->sawBOF()) { + bofFixup(); + } + + // + // Build the DFA state transition tables. + // + buildStateTable(); + mapLookAheadRules(); + flagAcceptingStates(); + flagLookAheadStates(); + flagTaggedStates(); + + // + // Update the global table of rule status {tag} values + // The rule builder has a global vector of status values that are common + // for all tables. Merge the ones from this table into the global set. + // + mergeRuleStatusVals(); +} + + + +//----------------------------------------------------------------------------- +// +// calcNullable. Impossible to explain succinctly. See Aho, section 3.9 +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::calcNullable(RBBINode *n) { + if (n == nullptr) { + return; + } + if (n->fType == RBBINode::setRef || + n->fType == RBBINode::endMark ) { + // These are non-empty leaf node types. + n->fNullable = false; + return; + } + + if (n->fType == RBBINode::lookAhead || n->fType == RBBINode::tag) { + // Lookahead marker node. It's a leaf, so no recursion on children. + // It's nullable because it does not match any literal text from the input stream. + n->fNullable = true; + return; + } + + + // The node is not a leaf. + // Calculate nullable on its children. + calcNullable(n->fLeftChild); + calcNullable(n->fRightChild); + + // Apply functions from table 3.40 in Aho + if (n->fType == RBBINode::opOr) { + n->fNullable = n->fLeftChild->fNullable || n->fRightChild->fNullable; + } + else if (n->fType == RBBINode::opCat) { + n->fNullable = n->fLeftChild->fNullable && n->fRightChild->fNullable; + } + else if (n->fType == RBBINode::opStar || n->fType == RBBINode::opQuestion) { + n->fNullable = true; + } + else { + n->fNullable = false; + } +} + + + + +//----------------------------------------------------------------------------- +// +// calcFirstPos. Impossible to explain succinctly. See Aho, section 3.9 +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::calcFirstPos(RBBINode *n) { + if (n == nullptr) { + return; + } + if (n->fType == RBBINode::leafChar || + n->fType == RBBINode::endMark || + n->fType == RBBINode::lookAhead || + n->fType == RBBINode::tag) { + // These are non-empty leaf node types. + // Note: In order to maintain the sort invariant on the set, + // this function should only be called on a node whose set is + // empty to start with. + n->fFirstPosSet->addElement(n, *fStatus); + return; + } + + // The node is not a leaf. + // Calculate firstPos on its children. + calcFirstPos(n->fLeftChild); + calcFirstPos(n->fRightChild); + + // Apply functions from table 3.40 in Aho + if (n->fType == RBBINode::opOr) { + setAdd(n->fFirstPosSet, n->fLeftChild->fFirstPosSet); + setAdd(n->fFirstPosSet, n->fRightChild->fFirstPosSet); + } + else if (n->fType == RBBINode::opCat) { + setAdd(n->fFirstPosSet, n->fLeftChild->fFirstPosSet); + if (n->fLeftChild->fNullable) { + setAdd(n->fFirstPosSet, n->fRightChild->fFirstPosSet); + } + } + else if (n->fType == RBBINode::opStar || + n->fType == RBBINode::opQuestion || + n->fType == RBBINode::opPlus) { + setAdd(n->fFirstPosSet, n->fLeftChild->fFirstPosSet); + } +} + + + +//----------------------------------------------------------------------------- +// +// calcLastPos. Impossible to explain succinctly. See Aho, section 3.9 +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::calcLastPos(RBBINode *n) { + if (n == nullptr) { + return; + } + if (n->fType == RBBINode::leafChar || + n->fType == RBBINode::endMark || + n->fType == RBBINode::lookAhead || + n->fType == RBBINode::tag) { + // These are non-empty leaf node types. + // Note: In order to maintain the sort invariant on the set, + // this function should only be called on a node whose set is + // empty to start with. + n->fLastPosSet->addElement(n, *fStatus); + return; + } + + // The node is not a leaf. + // Calculate lastPos on its children. + calcLastPos(n->fLeftChild); + calcLastPos(n->fRightChild); + + // Apply functions from table 3.40 in Aho + if (n->fType == RBBINode::opOr) { + setAdd(n->fLastPosSet, n->fLeftChild->fLastPosSet); + setAdd(n->fLastPosSet, n->fRightChild->fLastPosSet); + } + else if (n->fType == RBBINode::opCat) { + setAdd(n->fLastPosSet, n->fRightChild->fLastPosSet); + if (n->fRightChild->fNullable) { + setAdd(n->fLastPosSet, n->fLeftChild->fLastPosSet); + } + } + else if (n->fType == RBBINode::opStar || + n->fType == RBBINode::opQuestion || + n->fType == RBBINode::opPlus) { + setAdd(n->fLastPosSet, n->fLeftChild->fLastPosSet); + } +} + + + +//----------------------------------------------------------------------------- +// +// calcFollowPos. Impossible to explain succinctly. See Aho, section 3.9 +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::calcFollowPos(RBBINode *n) { + if (n == nullptr || + n->fType == RBBINode::leafChar || + n->fType == RBBINode::endMark) { + return; + } + + calcFollowPos(n->fLeftChild); + calcFollowPos(n->fRightChild); + + // Aho rule #1 + if (n->fType == RBBINode::opCat) { + RBBINode *i; // is 'i' in Aho's description + uint32_t ix; + + UVector *LastPosOfLeftChild = n->fLeftChild->fLastPosSet; + + for (ix=0; ix<(uint32_t)LastPosOfLeftChild->size(); ix++) { + i = (RBBINode *)LastPosOfLeftChild->elementAt(ix); + setAdd(i->fFollowPos, n->fRightChild->fFirstPosSet); + } + } + + // Aho rule #2 + if (n->fType == RBBINode::opStar || + n->fType == RBBINode::opPlus) { + RBBINode *i; // again, n and i are the names from Aho's description. + uint32_t ix; + + for (ix=0; ix<(uint32_t)n->fLastPosSet->size(); ix++) { + i = (RBBINode *)n->fLastPosSet->elementAt(ix); + setAdd(i->fFollowPos, n->fFirstPosSet); + } + } + + + +} + +//----------------------------------------------------------------------------- +// +// addRuleRootNodes Recursively walk a parse tree, adding all nodes flagged +// as roots of a rule to a destination vector. +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::addRuleRootNodes(UVector *dest, RBBINode *node) { + if (node == nullptr || U_FAILURE(*fStatus)) { + return; + } + U_ASSERT(!dest->hasDeleter()); + if (node->fRuleRoot) { + dest->addElement(node, *fStatus); + // Note: rules cannot nest. If we found a rule start node, + // no child node can also be a start node. + return; + } + addRuleRootNodes(dest, node->fLeftChild); + addRuleRootNodes(dest, node->fRightChild); +} + +//----------------------------------------------------------------------------- +// +// calcChainedFollowPos. Modify the previously calculated followPos sets +// to implement rule chaining. NOT described by Aho +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::calcChainedFollowPos(RBBINode *tree, RBBINode *endMarkNode) { + + UVector leafNodes(*fStatus); + if (U_FAILURE(*fStatus)) { + return; + } + + // get a list all leaf nodes + tree->findNodes(&leafNodes, RBBINode::leafChar, *fStatus); + if (U_FAILURE(*fStatus)) { + return; + } + + // Collect all leaf nodes that can start matches for rules + // with inbound chaining enabled, which is the union of the + // firstPosition sets from each of the rule root nodes. + + UVector ruleRootNodes(*fStatus); + addRuleRootNodes(&ruleRootNodes, tree); + + UVector matchStartNodes(*fStatus); + for (int j=0; j(ruleRootNodes.elementAt(j)); + if (node->fChainIn) { + setAdd(&matchStartNodes, node->fFirstPosSet); + } + } + if (U_FAILURE(*fStatus)) { + return; + } + + int32_t endNodeIx; + int32_t startNodeIx; + + for (endNodeIx=0; endNodeIxfFollowPos->contains(endMarkNode)) { + continue; + } + + // We've got a node that can end a match. + + // !!LBCMNoChain implementation: If this node's val correspond to + // the Line Break $CM char class, don't chain from it. + // TODO: Remove this. !!LBCMNoChain is deprecated, and is not used + // by any of the standard ICU rules. + if (fRB->fLBCMNoChain) { + UChar32 c = this->fRB->fSetBuilder->getFirstChar(endNode->fVal); + if (c != -1) { + // c == -1 occurs with sets containing only the {eof} marker string. + ULineBreak cLBProp = (ULineBreak)u_getIntPropertyValue(c, UCHAR_LINE_BREAK); + if (cLBProp == U_LB_COMBINING_MARK) { + continue; + } + } + } + + // Now iterate over the nodes that can start a match, looking for ones + // with the same char class as our ending node. + RBBINode *startNode; + for (startNodeIx = 0; startNodeIxfType != RBBINode::leafChar) { + continue; + } + + if (endNode->fVal == startNode->fVal) { + // The end val (character class) of one possible match is the + // same as the start of another. + + // Add all nodes from the followPos of the start node to the + // followPos set of the end node, which will have the effect of + // letting matches transition from a match state at endNode + // to the second char of a match starting with startNode. + setAdd(endNode->fFollowPos, startNode->fFollowPos); + } + } + } +} + + +//----------------------------------------------------------------------------- +// +// bofFixup. Fixup for state tables that include {bof} beginning of input testing. +// Do an swizzle similar to chaining, modifying the followPos set of +// the bofNode to include the followPos nodes from other {bot} nodes +// scattered through the tree. +// +// This function has much in common with calcChainedFollowPos(). +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::bofFixup() { + + if (U_FAILURE(*fStatus)) { + return; + } + + // The parse tree looks like this ... + // fTree root ---> + // / \ . + // <#end node> + // / \ . + // rest + // of tree + // + // We will be adding things to the followPos set of the + // + RBBINode *bofNode = fTree->fLeftChild->fLeftChild; + U_ASSERT(bofNode->fType == RBBINode::leafChar); + U_ASSERT(bofNode->fVal == 2); + + // Get all nodes that can be the start a match of the user-written rules + // (excluding the fake bofNode) + // We want the nodes that can start a match in the + // part labeled "rest of tree" + // + UVector *matchStartNodes = fTree->fLeftChild->fRightChild->fFirstPosSet; + + RBBINode *startNode; + int startNodeIx; + for (startNodeIx = 0; startNodeIxsize(); startNodeIx++) { + startNode = (RBBINode *)matchStartNodes->elementAt(startNodeIx); + if (startNode->fType != RBBINode::leafChar) { + continue; + } + + if (startNode->fVal == bofNode->fVal) { + // We found a leaf node corresponding to a {bof} that was + // explicitly written into a rule. + // Add everything from the followPos set of this node to the + // followPos set of the fake bofNode at the start of the tree. + // + setAdd(bofNode->fFollowPos, startNode->fFollowPos); + } + } +} + +//----------------------------------------------------------------------------- +// +// buildStateTable() Determine the set of runtime DFA states and the +// transition tables for these states, by the algorithm +// of fig. 3.44 in Aho. +// +// Most of the comments are quotes of Aho's psuedo-code. +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::buildStateTable() { + if (U_FAILURE(*fStatus)) { + return; + } + RBBIStateDescriptor *failState; + // Set it to nullptr to avoid uninitialized warning + RBBIStateDescriptor *initialState = nullptr; + // + // Add a dummy state 0 - the stop state. Not from Aho. + int lastInputSymbol = fRB->fSetBuilder->getNumCharCategories() - 1; + failState = new RBBIStateDescriptor(lastInputSymbol, fStatus); + if (failState == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + goto ExitBuildSTdeleteall; + } + failState->fPositions = new UVector(*fStatus); + if (failState->fPositions == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + } + if (failState->fPositions == nullptr || U_FAILURE(*fStatus)) { + goto ExitBuildSTdeleteall; + } + fDStates->addElement(failState, *fStatus); + if (U_FAILURE(*fStatus)) { + goto ExitBuildSTdeleteall; + } + + // initially, the only unmarked state in Dstates is firstpos(root), + // where toot is the root of the syntax tree for (r)#; + initialState = new RBBIStateDescriptor(lastInputSymbol, fStatus); + if (initialState == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + } + if (U_FAILURE(*fStatus)) { + goto ExitBuildSTdeleteall; + } + initialState->fPositions = new UVector(*fStatus); + if (initialState->fPositions == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + } + if (U_FAILURE(*fStatus)) { + goto ExitBuildSTdeleteall; + } + setAdd(initialState->fPositions, fTree->fFirstPosSet); + fDStates->addElement(initialState, *fStatus); + if (U_FAILURE(*fStatus)) { + goto ExitBuildSTdeleteall; + } + + // while there is an unmarked state T in Dstates do begin + for (;;) { + RBBIStateDescriptor *T = nullptr; + int32_t tx; + for (tx=1; txsize(); tx++) { + RBBIStateDescriptor *temp; + temp = (RBBIStateDescriptor *)fDStates->elementAt(tx); + if (temp->fMarked == false) { + T = temp; + break; + } + } + if (T == nullptr) { + break; + } + + // mark T; + T->fMarked = true; + + // for each input symbol a do begin + int32_t a; + for (a = 1; a<=lastInputSymbol; a++) { + // let U be the set of positions that are in followpos(p) + // for some position p in T + // such that the symbol at position p is a; + UVector *U = nullptr; + RBBINode *p; + int32_t px; + for (px=0; pxfPositions->size(); px++) { + p = (RBBINode *)T->fPositions->elementAt(px); + if ((p->fType == RBBINode::leafChar) && (p->fVal == a)) { + if (U == nullptr) { + U = new UVector(*fStatus); + if (U == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + goto ExitBuildSTdeleteall; + } + } + setAdd(U, p->fFollowPos); + } + } + + // if U is not empty and not in DStates then + int32_t ux = 0; + UBool UinDstates = false; + if (U != nullptr) { + U_ASSERT(U->size() > 0); + int ix; + for (ix=0; ixsize(); ix++) { + RBBIStateDescriptor *temp2; + temp2 = (RBBIStateDescriptor *)fDStates->elementAt(ix); + if (setEquals(U, temp2->fPositions)) { + delete U; + U = temp2->fPositions; + ux = ix; + UinDstates = true; + break; + } + } + + // Add U as an unmarked state to Dstates + if (!UinDstates) + { + RBBIStateDescriptor *newState = new RBBIStateDescriptor(lastInputSymbol, fStatus); + if (newState == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + } + if (U_FAILURE(*fStatus)) { + goto ExitBuildSTdeleteall; + } + newState->fPositions = U; + fDStates->addElement(newState, *fStatus); + if (U_FAILURE(*fStatus)) { + return; + } + ux = fDStates->size()-1; + } + + // Dtran[T, a] := U; + T->fDtran->setElementAt(ux, a); + } + } + } + return; + // delete local pointers only if error occurred. +ExitBuildSTdeleteall: + delete initialState; + delete failState; +} + + +/** + * mapLookAheadRules + * + */ +void RBBITableBuilder::mapLookAheadRules() { + fLookAheadRuleMap = new UVector32(fRB->fScanner->numRules() + 1, *fStatus); + if (fLookAheadRuleMap == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + } + if (U_FAILURE(*fStatus)) { + return; + } + fLookAheadRuleMap->setSize(fRB->fScanner->numRules() + 1); + + for (int32_t n=0; nsize(); n++) { + RBBIStateDescriptor *sd = (RBBIStateDescriptor *)fDStates->elementAt(n); + int32_t laSlotForState = 0; + + // Establish the look-ahead slot for this state, if the state covers + // any look-ahead nodes - corresponding to the '/' in look-ahead rules. + + // If any of the look-ahead nodes already have a slot assigned, use it, + // otherwise assign a new one. + + bool sawLookAheadNode = false; + for (int32_t ipos=0; iposfPositions->size(); ++ipos) { + RBBINode *node = static_cast(sd->fPositions->elementAt(ipos)); + if (node->fType != RBBINode::NodeType::lookAhead) { + continue; + } + sawLookAheadNode = true; + int32_t ruleNum = node->fVal; // Set when rule was originally parsed. + U_ASSERT(ruleNum < fLookAheadRuleMap->size()); + U_ASSERT(ruleNum > 0); + int32_t laSlot = fLookAheadRuleMap->elementAti(ruleNum); + if (laSlot != 0) { + if (laSlotForState == 0) { + laSlotForState = laSlot; + } else { + // TODO: figure out if this can fail, change to setting an error code if so. + U_ASSERT(laSlot == laSlotForState); + } + } + } + if (!sawLookAheadNode) { + continue; + } + + if (laSlotForState == 0) { + laSlotForState = ++fLASlotsInUse; + } + + // For each look ahead node covered by this state, + // set the mapping from the node's rule number to the look ahead slot. + // There can be multiple nodes/rule numbers going to the same la slot. + + for (int32_t ipos=0; iposfPositions->size(); ++ipos) { + RBBINode *node = static_cast(sd->fPositions->elementAt(ipos)); + if (node->fType != RBBINode::NodeType::lookAhead) { + continue; + } + int32_t ruleNum = node->fVal; // Set when rule was originally parsed. + int32_t existingVal = fLookAheadRuleMap->elementAti(ruleNum); + (void)existingVal; + U_ASSERT(existingVal == 0 || existingVal == laSlotForState); + fLookAheadRuleMap->setElementAt(laSlotForState, ruleNum); + } + } + +} + +//----------------------------------------------------------------------------- +// +// flagAcceptingStates Identify accepting states. +// First get a list of all of the end marker nodes. +// Then, for each state s, +// if s contains one of the end marker nodes in its list of tree positions then +// s is an accepting state. +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::flagAcceptingStates() { + if (U_FAILURE(*fStatus)) { + return; + } + UVector endMarkerNodes(*fStatus); + RBBINode *endMarker; + int32_t i; + int32_t n; + + if (U_FAILURE(*fStatus)) { + return; + } + + fTree->findNodes(&endMarkerNodes, RBBINode::endMark, *fStatus); + if (U_FAILURE(*fStatus)) { + return; + } + + for (i=0; isize(); n++) { + RBBIStateDescriptor *sd = (RBBIStateDescriptor *)fDStates->elementAt(n); + if (sd->fPositions->indexOf(endMarker) >= 0) { + // Any non-zero value for fAccepting means this is an accepting node. + // The value is what will be returned to the user as the break status. + // If no other value was specified, force it to ACCEPTING_UNCONDITIONAL (1). + + if (sd->fAccepting==0) { + // State hasn't been marked as accepting yet. Do it now. + sd->fAccepting = fLookAheadRuleMap->elementAti(endMarker->fVal); + if (sd->fAccepting == 0) { + sd->fAccepting = ACCEPTING_UNCONDITIONAL; + } + } + if (sd->fAccepting==ACCEPTING_UNCONDITIONAL && endMarker->fVal != 0) { + // Both lookahead and non-lookahead accepting for this state. + // Favor the look-ahead, because a look-ahead match needs to + // immediately stop the run-time engine. First match, not longest. + sd->fAccepting = fLookAheadRuleMap->elementAti(endMarker->fVal); + } + // implicit else: + // if sd->fAccepting already had a value other than 0 or 1, leave it be. + } + } + } +} + + +//----------------------------------------------------------------------------- +// +// flagLookAheadStates Very similar to flagAcceptingStates, above. +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::flagLookAheadStates() { + if (U_FAILURE(*fStatus)) { + return; + } + UVector lookAheadNodes(*fStatus); + RBBINode *lookAheadNode; + int32_t i; + int32_t n; + + fTree->findNodes(&lookAheadNodes, RBBINode::lookAhead, *fStatus); + if (U_FAILURE(*fStatus)) { + return; + } + for (i=0; ifType == RBBINode::NodeType::lookAhead); + + for (n=0; nsize(); n++) { + RBBIStateDescriptor *sd = (RBBIStateDescriptor *)fDStates->elementAt(n); + int32_t positionsIdx = sd->fPositions->indexOf(lookAheadNode); + if (positionsIdx >= 0) { + U_ASSERT(lookAheadNode == sd->fPositions->elementAt(positionsIdx)); + uint32_t lookaheadSlot = fLookAheadRuleMap->elementAti(lookAheadNode->fVal); + U_ASSERT(sd->fLookAhead == 0 || sd->fLookAhead == lookaheadSlot); + // if (sd->fLookAhead != 0 && sd->fLookAhead != lookaheadSlot) { + // printf("%s:%d Bingo. sd->fLookAhead:%d lookaheadSlot:%d\n", + // __FILE__, __LINE__, sd->fLookAhead, lookaheadSlot); + // } + sd->fLookAhead = lookaheadSlot; + } + } + } +} + + + + +//----------------------------------------------------------------------------- +// +// flagTaggedStates +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::flagTaggedStates() { + if (U_FAILURE(*fStatus)) { + return; + } + UVector tagNodes(*fStatus); + RBBINode *tagNode; + int32_t i; + int32_t n; + + if (U_FAILURE(*fStatus)) { + return; + } + fTree->findNodes(&tagNodes, RBBINode::tag, *fStatus); + if (U_FAILURE(*fStatus)) { + return; + } + for (i=0; isize(); n++) { // For each state s (row in the state table) + RBBIStateDescriptor *sd = (RBBIStateDescriptor *)fDStates->elementAt(n); + if (sd->fPositions->indexOf(tagNode) >= 0) { // if s include the tag node t + sortedAdd(&sd->fTagVals, tagNode->fVal); + } + } + } +} + + + + +//----------------------------------------------------------------------------- +// +// mergeRuleStatusVals +// +// Update the global table of rule status {tag} values +// The rule builder has a global vector of status values that are common +// for all tables. Merge the ones from this table into the global set. +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::mergeRuleStatusVals() { + // + // The basic outline of what happens here is this... + // + // for each state in this state table + // if the status tag list for this state is in the global statuses list + // record where and + // continue with the next state + // else + // add the tag list for this state to the global list. + // + int i; + int n; + + // Pre-set a single tag of {0} into the table. + // We will need this as a default, for rule sets with no explicit tagging. + if (fRB->fRuleStatusVals->size() == 0) { + fRB->fRuleStatusVals->addElement(1, *fStatus); // Num of statuses in group + fRB->fRuleStatusVals->addElement((int32_t)0, *fStatus); // and our single status of zero + } + + // For each state + for (n=0; nsize(); n++) { + RBBIStateDescriptor *sd = (RBBIStateDescriptor *)fDStates->elementAt(n); + UVector *thisStatesTagValues = sd->fTagVals; + if (thisStatesTagValues == nullptr) { + // No tag values are explicitly associated with this state. + // Set the default tag value. + sd->fTagsIdx = 0; + continue; + } + + // There are tag(s) associated with this state. + // fTagsIdx will be the index into the global tag list for this state's tag values. + // Initial value of -1 flags that we haven't got it set yet. + sd->fTagsIdx = -1; + int32_t thisTagGroupStart = 0; // indexes into the global rule status vals list + int32_t nextTagGroupStart = 0; + + // Loop runs once per group of tags in the global list + while (nextTagGroupStart < fRB->fRuleStatusVals->size()) { + thisTagGroupStart = nextTagGroupStart; + nextTagGroupStart += fRB->fRuleStatusVals->elementAti(thisTagGroupStart) + 1; + if (thisStatesTagValues->size() != fRB->fRuleStatusVals->elementAti(thisTagGroupStart)) { + // The number of tags for this state is different from + // the number of tags in this group from the global list. + // Continue with the next group from the global list. + continue; + } + // The lengths match, go ahead and compare the actual tag values + // between this state and the group from the global list. + for (i=0; isize(); i++) { + if (thisStatesTagValues->elementAti(i) != + fRB->fRuleStatusVals->elementAti(thisTagGroupStart + 1 + i) ) { + // Mismatch. + break; + } + } + + if (i == thisStatesTagValues->size()) { + // We found a set of tag values in the global list that match + // those for this state. Use them. + sd->fTagsIdx = thisTagGroupStart; + break; + } + } + + if (sd->fTagsIdx == -1) { + // No suitable entry in the global tag list already. Add one + sd->fTagsIdx = fRB->fRuleStatusVals->size(); + fRB->fRuleStatusVals->addElement(thisStatesTagValues->size(), *fStatus); + for (i=0; isize(); i++) { + fRB->fRuleStatusVals->addElement(thisStatesTagValues->elementAti(i), *fStatus); + } + } + } +} + + + + + + + +//----------------------------------------------------------------------------- +// +// sortedAdd Add a value to a vector of sorted values (ints). +// Do not replicate entries; if the value is already there, do not +// add a second one. +// Lazily create the vector if it does not already exist. +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::sortedAdd(UVector **vector, int32_t val) { + int32_t i; + + if (*vector == nullptr) { + *vector = new UVector(*fStatus); + } + if (*vector == nullptr || U_FAILURE(*fStatus)) { + return; + } + UVector *vec = *vector; + int32_t vSize = vec->size(); + for (i=0; ielementAti(i); + if (valAtI == val) { + // The value is already in the vector. Don't add it again. + return; + } + if (valAtI > val) { + break; + } + } + vec->insertElementAt(val, i, *fStatus); +} + + + +//----------------------------------------------------------------------------- +// +// setAdd Set operation on UVector +// dest = dest union source +// Elements may only appear once and must be sorted. +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::setAdd(UVector *dest, UVector *source) { + U_ASSERT(!dest->hasDeleter()); + U_ASSERT(!source->hasDeleter()); + int32_t destOriginalSize = dest->size(); + int32_t sourceSize = source->size(); + int32_t di = 0; + MaybeStackArray destArray, sourceArray; // Handle small cases without malloc + void **destPtr, **sourcePtr; + void **destLim, **sourceLim; + + if (destOriginalSize > destArray.getCapacity()) { + if (destArray.resize(destOriginalSize) == nullptr) { + return; + } + } + destPtr = destArray.getAlias(); + destLim = destPtr + destOriginalSize; // destArray.getArrayLimit()? + + if (sourceSize > sourceArray.getCapacity()) { + if (sourceArray.resize(sourceSize) == nullptr) { + return; + } + } + sourcePtr = sourceArray.getAlias(); + sourceLim = sourcePtr + sourceSize; // sourceArray.getArrayLimit()? + + // Avoid multiple "get element" calls by getting the contents into arrays + (void) dest->toArray(destPtr); + (void) source->toArray(sourcePtr); + + dest->setSize(sourceSize+destOriginalSize, *fStatus); + if (U_FAILURE(*fStatus)) { + return; + } + + while (sourcePtr < sourceLim && destPtr < destLim) { + if (*destPtr == *sourcePtr) { + dest->setElementAt(*sourcePtr++, di++); + destPtr++; + } + // This check is required for machines with segmented memory, like i5/OS. + // Direct pointer comparison is not recommended. + else if (uprv_memcmp(destPtr, sourcePtr, sizeof(void *)) < 0) { + dest->setElementAt(*destPtr++, di++); + } + else { /* *sourcePtr < *destPtr */ + dest->setElementAt(*sourcePtr++, di++); + } + } + + // At most one of these two cleanup loops will execute + while (destPtr < destLim) { + dest->setElementAt(*destPtr++, di++); + } + while (sourcePtr < sourceLim) { + dest->setElementAt(*sourcePtr++, di++); + } + + dest->setSize(di, *fStatus); +} + + + +//----------------------------------------------------------------------------- +// +// setEqual Set operation on UVector. +// Compare for equality. +// Elements must be sorted. +// +//----------------------------------------------------------------------------- +UBool RBBITableBuilder::setEquals(UVector *a, UVector *b) { + return a->equals(*b); +} + + +//----------------------------------------------------------------------------- +// +// printPosSets Debug function. Dump Nullable, firstpos, lastpos and followpos +// for each node in the tree. +// +//----------------------------------------------------------------------------- +#ifdef RBBI_DEBUG +void RBBITableBuilder::printPosSets(RBBINode *n) { + if (n==nullptr) { + return; + } + printf("\n"); + RBBINode::printNodeHeader(); + RBBINode::printNode(n); + RBBIDebugPrintf(" Nullable: %s\n", n->fNullable?"true":"false"); + + RBBIDebugPrintf(" firstpos: "); + printSet(n->fFirstPosSet); + + RBBIDebugPrintf(" lastpos: "); + printSet(n->fLastPosSet); + + RBBIDebugPrintf(" followpos: "); + printSet(n->fFollowPos); + + printPosSets(n->fLeftChild); + printPosSets(n->fRightChild); +} +#endif + +// +// findDuplCharClassFrom() +// +bool RBBITableBuilder::findDuplCharClassFrom(IntPair *categories) { + int32_t numStates = fDStates->size(); + int32_t numCols = fRB->fSetBuilder->getNumCharCategories(); + + for (; categories->first < numCols-1; categories->first++) { + // Note: dictionary & non-dictionary columns cannot be merged. + // The limitSecond value prevents considering mixed pairs. + // Dictionary categories are >= DictCategoriesStart. + // Non dict categories are < DictCategoriesStart. + int limitSecond = categories->first < fRB->fSetBuilder->getDictCategoriesStart() ? + fRB->fSetBuilder->getDictCategoriesStart() : numCols; + for (categories->second=categories->first+1; categories->second < limitSecond; categories->second++) { + // Initialized to different values to prevent returning true if numStates = 0 (implies no duplicates). + uint16_t table_base = 0; + uint16_t table_dupl = 1; + for (int32_t state=0; stateelementAt(state); + table_base = (uint16_t)sd->fDtran->elementAti(categories->first); + table_dupl = (uint16_t)sd->fDtran->elementAti(categories->second); + if (table_base != table_dupl) { + break; + } + } + if (table_base == table_dupl) { + return true; + } + } + } + return false; +} + + +// +// removeColumn() +// +void RBBITableBuilder::removeColumn(int32_t column) { + int32_t numStates = fDStates->size(); + for (int32_t state=0; stateelementAt(state); + U_ASSERT(column < sd->fDtran->size()); + sd->fDtran->removeElementAt(column); + } +} + +/* + * findDuplicateState + */ +bool RBBITableBuilder::findDuplicateState(IntPair *states) { + int32_t numStates = fDStates->size(); + int32_t numCols = fRB->fSetBuilder->getNumCharCategories(); + + for (; states->firstfirst++) { + RBBIStateDescriptor *firstSD = (RBBIStateDescriptor *)fDStates->elementAt(states->first); + for (states->second=states->first+1; states->secondsecond++) { + RBBIStateDescriptor *duplSD = (RBBIStateDescriptor *)fDStates->elementAt(states->second); + if (firstSD->fAccepting != duplSD->fAccepting || + firstSD->fLookAhead != duplSD->fLookAhead || + firstSD->fTagsIdx != duplSD->fTagsIdx) { + continue; + } + bool rowsMatch = true; + for (int32_t col=0; col < numCols; ++col) { + int32_t firstVal = firstSD->fDtran->elementAti(col); + int32_t duplVal = duplSD->fDtran->elementAti(col); + if (!((firstVal == duplVal) || + ((firstVal == states->first || firstVal == states->second) && + (duplVal == states->first || duplVal == states->second)))) { + rowsMatch = false; + break; + } + } + if (rowsMatch) { + return true; + } + } + } + return false; +} + + +bool RBBITableBuilder::findDuplicateSafeState(IntPair *states) { + int32_t numStates = fSafeTable->size(); + + for (; states->firstfirst++) { + UnicodeString *firstRow = static_cast(fSafeTable->elementAt(states->first)); + for (states->second=states->first+1; states->secondsecond++) { + UnicodeString *duplRow = static_cast(fSafeTable->elementAt(states->second)); + bool rowsMatch = true; + int32_t numCols = firstRow->length(); + for (int32_t col=0; col < numCols; ++col) { + int32_t firstVal = firstRow->charAt(col); + int32_t duplVal = duplRow->charAt(col); + if (!((firstVal == duplVal) || + ((firstVal == states->first || firstVal == states->second) && + (duplVal == states->first || duplVal == states->second)))) { + rowsMatch = false; + break; + } + } + if (rowsMatch) { + return true; + } + } + } + return false; +} + + +void RBBITableBuilder::removeState(IntPair duplStates) { + const int32_t keepState = duplStates.first; + const int32_t duplState = duplStates.second; + U_ASSERT(keepState < duplState); + U_ASSERT(duplState < fDStates->size()); + + RBBIStateDescriptor *duplSD = (RBBIStateDescriptor *)fDStates->elementAt(duplState); + fDStates->removeElementAt(duplState); + delete duplSD; + + int32_t numStates = fDStates->size(); + int32_t numCols = fRB->fSetBuilder->getNumCharCategories(); + for (int32_t state=0; stateelementAt(state); + for (int32_t col=0; colfDtran->elementAti(col); + int32_t newVal = existingVal; + if (existingVal == duplState) { + newVal = keepState; + } else if (existingVal > duplState) { + newVal = existingVal - 1; + } + sd->fDtran->setElementAt(newVal, col); + } + } +} + +void RBBITableBuilder::removeSafeState(IntPair duplStates) { + const int32_t keepState = duplStates.first; + const int32_t duplState = duplStates.second; + U_ASSERT(keepState < duplState); + U_ASSERT(duplState < fSafeTable->size()); + + fSafeTable->removeElementAt(duplState); // Note that fSafeTable has a deleter function + // and will auto-delete the removed element. + int32_t numStates = fSafeTable->size(); + for (int32_t state=0; stateelementAt(state); + int32_t numCols = sd->length(); + for (int32_t col=0; colcharAt(col); + int32_t newVal = existingVal; + if (existingVal == duplState) { + newVal = keepState; + } else if (existingVal > duplState) { + newVal = existingVal - 1; + } + sd->setCharAt(col, static_cast(newVal)); + } + } +} + + +/* + * RemoveDuplicateStates + */ +int32_t RBBITableBuilder::removeDuplicateStates() { + IntPair dupls = {3, 0}; + int32_t numStatesRemoved = 0; + + while (findDuplicateState(&dupls)) { + // printf("Removing duplicate states (%d, %d)\n", dupls.first, dupls.second); + removeState(dupls); + ++numStatesRemoved; + } + return numStatesRemoved; +} + + +//----------------------------------------------------------------------------- +// +// getTableSize() Calculate the size of the runtime form of this +// state transition table. +// +//----------------------------------------------------------------------------- +int32_t RBBITableBuilder::getTableSize() const { + int32_t size = 0; + int32_t numRows; + int32_t numCols; + int32_t rowSize; + + if (fTree == nullptr) { + return 0; + } + + size = offsetof(RBBIStateTable, fTableData); // The header, with no rows to the table. + + numRows = fDStates->size(); + numCols = fRB->fSetBuilder->getNumCharCategories(); + + if (use8BitsForTable()) { + rowSize = offsetof(RBBIStateTableRow8, fNextState) + sizeof(int8_t)*numCols; + } else { + rowSize = offsetof(RBBIStateTableRow16, fNextState) + sizeof(int16_t)*numCols; + } + size += numRows * rowSize; + return size; +} + +bool RBBITableBuilder::use8BitsForTable() const { + return fDStates->size() <= kMaxStateFor8BitsTable; +} + +//----------------------------------------------------------------------------- +// +// exportTable() export the state transition table in the format required +// by the runtime engine. getTableSize() bytes of memory +// must be available at the output address "where". +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::exportTable(void *where) { + RBBIStateTable *table = (RBBIStateTable *)where; + uint32_t state; + int col; + + if (U_FAILURE(*fStatus) || fTree == nullptr) { + return; + } + + int32_t catCount = fRB->fSetBuilder->getNumCharCategories(); + if (catCount > 0x7fff || + fDStates->size() > 0x7fff) { + *fStatus = U_BRK_INTERNAL_ERROR; + return; + } + + table->fNumStates = fDStates->size(); + table->fDictCategoriesStart = fRB->fSetBuilder->getDictCategoriesStart(); + table->fLookAheadResultsSize = fLASlotsInUse == ACCEPTING_UNCONDITIONAL ? 0 : fLASlotsInUse + 1; + table->fFlags = 0; + if (use8BitsForTable()) { + table->fRowLen = offsetof(RBBIStateTableRow8, fNextState) + sizeof(uint8_t) * catCount; + table->fFlags |= RBBI_8BITS_ROWS; + } else { + table->fRowLen = offsetof(RBBIStateTableRow16, fNextState) + sizeof(int16_t) * catCount; + } + if (fRB->fLookAheadHardBreak) { + table->fFlags |= RBBI_LOOKAHEAD_HARD_BREAK; + } + if (fRB->fSetBuilder->sawBOF()) { + table->fFlags |= RBBI_BOF_REQUIRED; + } + + for (state=0; statefNumStates; state++) { + RBBIStateDescriptor *sd = (RBBIStateDescriptor *)fDStates->elementAt(state); + RBBIStateTableRow *row = (RBBIStateTableRow *)(table->fTableData + state*table->fRowLen); + if (use8BitsForTable()) { + U_ASSERT (sd->fAccepting <= 255); + U_ASSERT (sd->fLookAhead <= 255); + U_ASSERT (0 <= sd->fTagsIdx && sd->fTagsIdx <= 255); + RBBIStateTableRow8 *r8 = (RBBIStateTableRow8*)row; + r8->fAccepting = sd->fAccepting; + r8->fLookAhead = sd->fLookAhead; + r8->fTagsIdx = sd->fTagsIdx; + for (col=0; colfDtran->elementAti(col) <= kMaxStateFor8BitsTable); + r8->fNextState[col] = sd->fDtran->elementAti(col); + } + } else { + U_ASSERT (sd->fAccepting <= 0xffff); + U_ASSERT (sd->fLookAhead <= 0xffff); + U_ASSERT (0 <= sd->fTagsIdx && sd->fTagsIdx <= 0xffff); + row->r16.fAccepting = sd->fAccepting; + row->r16.fLookAhead = sd->fLookAhead; + row->r16.fTagsIdx = sd->fTagsIdx; + for (col=0; colr16.fNextState[col] = sd->fDtran->elementAti(col); + } + } + } +} + + +/** + * Synthesize a safe state table from the main state table. + */ +void RBBITableBuilder::buildSafeReverseTable(UErrorCode &status) { + // The safe table creation has three steps: + + // 1. Identify pairs of character classes that are "safe." Safe means that boundaries + // following the pair do not depend on context or state before the pair. To test + // whether a pair is safe, run it through the main forward state table, starting + // from each state. If the the final state is the same, no matter what the starting state, + // the pair is safe. + // + // 2. Build a state table that recognizes the safe pairs. It's similar to their + // forward table, with a column for each input character [class], and a row for + // each state. Row 1 is the start state, and row 0 is the stop state. Initially + // create an additional state for each input character category; being in + // one of these states means that the character has been seen, and is potentially + // the first of a pair. In each of these rows, the entry for the second character + // of a safe pair is set to the stop state (0), indicating that a match was found. + // All other table entries are set to the state corresponding the current input + // character, allowing that character to be the of a start following pair. + // + // Because the safe rules are to be run in reverse, moving backwards in the text, + // the first and second pair categories are swapped when building the table. + // + // 3. Compress the table. There are typically many rows (states) that are + // equivalent - that have zeroes (match completed) in the same columns - + // and can be folded together. + + // Each safe pair is stored as two UChars in the safePair string. + UnicodeString safePairs; + + int32_t numCharClasses = fRB->fSetBuilder->getNumCharCategories(); + int32_t numStates = fDStates->size(); + + for (int32_t c1=0; c1(fDStates->elementAt(startState)); + int32_t s2 = startStateD->fDtran->elementAti(c1); + RBBIStateDescriptor *s2StateD = static_cast(fDStates->elementAt(s2)); + endState = s2StateD->fDtran->elementAti(c2); + if (wantedEndState < 0) { + wantedEndState = endState; + } else { + if (wantedEndState != endState) { + break; + } + } + } + if (wantedEndState == endState) { + safePairs.append((char16_t)c1); + safePairs.append((char16_t)c2); + // printf("(%d, %d) ", c1, c2); + } + } + // printf("\n"); + } + + // Populate the initial safe table. + // The table as a whole is UVector + // Each row is represented by a UnicodeString, being used as a Vector. + // Row 0 is the stop state. + // Row 1 is the start state. + // Row 2 and beyond are other states, initially one per char class, but + // after initial construction, many of the states will be combined, compacting the table. + // The String holds the nextState data only. The four leading fields of a row, fAccepting, + // fLookAhead, etc. are not needed for the safe table, and are omitted at this stage of building. + + U_ASSERT(fSafeTable == nullptr); + LocalPointer lpSafeTable( + new UVector(uprv_deleteUObject, uhash_compareUnicodeString, numCharClasses + 2, status), status); + if (U_FAILURE(status)) { + return; + } + fSafeTable = lpSafeTable.orphan(); + for (int32_t row=0; row lpString(new UnicodeString(numCharClasses, 0, numCharClasses+4), status); + fSafeTable->adoptElement(lpString.orphan(), status); + } + if (U_FAILURE(status)) { + return; + } + + // From the start state, each input char class transitions to the state for that input. + UnicodeString &startState = *static_cast(fSafeTable->elementAt(1)); + for (int32_t charClass=0; charClass < numCharClasses; ++charClass) { + // Note: +2 for the start & stop state. + startState.setCharAt(charClass, static_cast(charClass+2)); + } + + // Initially make every other state table row look like the start state row, + for (int32_t row=2; row(fSafeTable->elementAt(row)); + rowState = startState; // UnicodeString assignment, copies contents. + } + + // Run through the safe pairs, set the next state to zero when pair has been seen. + // Zero being the stop state, meaning we found a safe point. + for (int32_t pairIdx=0; pairIdx(fSafeTable->elementAt(c2 + 2)); + rowState.setCharAt(c1, 0); + } + + // Remove duplicate or redundant rows from the table. + IntPair states = {1, 0}; + while (findDuplicateSafeState(&states)) { + // printf("Removing duplicate safe states (%d, %d)\n", states.first, states.second); + removeSafeState(states); + } +} + + +//----------------------------------------------------------------------------- +// +// getSafeTableSize() Calculate the size of the runtime form of this +// safe state table. +// +//----------------------------------------------------------------------------- +int32_t RBBITableBuilder::getSafeTableSize() const { + int32_t size = 0; + int32_t numRows; + int32_t numCols; + int32_t rowSize; + + if (fSafeTable == nullptr) { + return 0; + } + + size = offsetof(RBBIStateTable, fTableData); // The header, with no rows to the table. + + numRows = fSafeTable->size(); + numCols = fRB->fSetBuilder->getNumCharCategories(); + + if (use8BitsForSafeTable()) { + rowSize = offsetof(RBBIStateTableRow8, fNextState) + sizeof(int8_t)*numCols; + } else { + rowSize = offsetof(RBBIStateTableRow16, fNextState) + sizeof(int16_t)*numCols; + } + size += numRows * rowSize; + return size; +} + +bool RBBITableBuilder::use8BitsForSafeTable() const { + return fSafeTable->size() <= kMaxStateFor8BitsTable; +} + +//----------------------------------------------------------------------------- +// +// exportSafeTable() export the state transition table in the format required +// by the runtime engine. getTableSize() bytes of memory +// must be available at the output address "where". +// +//----------------------------------------------------------------------------- +void RBBITableBuilder::exportSafeTable(void *where) { + RBBIStateTable *table = (RBBIStateTable *)where; + uint32_t state; + int col; + + if (U_FAILURE(*fStatus) || fSafeTable == nullptr) { + return; + } + + int32_t catCount = fRB->fSetBuilder->getNumCharCategories(); + if (catCount > 0x7fff || + fSafeTable->size() > 0x7fff) { + *fStatus = U_BRK_INTERNAL_ERROR; + return; + } + + table->fNumStates = fSafeTable->size(); + table->fFlags = 0; + if (use8BitsForSafeTable()) { + table->fRowLen = offsetof(RBBIStateTableRow8, fNextState) + sizeof(uint8_t) * catCount; + table->fFlags |= RBBI_8BITS_ROWS; + } else { + table->fRowLen = offsetof(RBBIStateTableRow16, fNextState) + sizeof(int16_t) * catCount; + } + + for (state=0; statefNumStates; state++) { + UnicodeString *rowString = (UnicodeString *)fSafeTable->elementAt(state); + RBBIStateTableRow *row = (RBBIStateTableRow *)(table->fTableData + state*table->fRowLen); + if (use8BitsForSafeTable()) { + RBBIStateTableRow8 *r8 = (RBBIStateTableRow8*)row; + r8->fAccepting = 0; + r8->fLookAhead = 0; + r8->fTagsIdx = 0; + for (col=0; colcharAt(col) <= kMaxStateFor8BitsTable); + r8->fNextState[col] = static_cast(rowString->charAt(col)); + } + } else { + row->r16.fAccepting = 0; + row->r16.fLookAhead = 0; + row->r16.fTagsIdx = 0; + for (col=0; colr16.fNextState[col] = rowString->charAt(col); + } + } + } +} + + + + +//----------------------------------------------------------------------------- +// +// printSet Debug function. Print the contents of a UVector +// +//----------------------------------------------------------------------------- +#ifdef RBBI_DEBUG +void RBBITableBuilder::printSet(UVector *s) { + int32_t i; + for (i=0; isize(); i++) { + const RBBINode *v = static_cast(s->elementAt(i)); + RBBIDebugPrintf("%5d", v==nullptr? -1 : v->fSerialNum); + } + RBBIDebugPrintf("\n"); +} +#endif + + +//----------------------------------------------------------------------------- +// +// printStates Debug Function. Dump the fully constructed state transition table. +// +//----------------------------------------------------------------------------- +#ifdef RBBI_DEBUG +void RBBITableBuilder::printStates() { + int c; // input "character" + int n; // state number + + RBBIDebugPrintf("state | i n p u t s y m b o l s \n"); + RBBIDebugPrintf(" | Acc LA Tag"); + for (c=0; cfSetBuilder->getNumCharCategories(); c++) { + RBBIDebugPrintf(" %3d", c); + } + RBBIDebugPrintf("\n"); + RBBIDebugPrintf(" |---------------"); + for (c=0; cfSetBuilder->getNumCharCategories(); c++) { + RBBIDebugPrintf("----"); + } + RBBIDebugPrintf("\n"); + + for (n=0; nsize(); n++) { + RBBIStateDescriptor *sd = (RBBIStateDescriptor *)fDStates->elementAt(n); + RBBIDebugPrintf(" %3d | " , n); + RBBIDebugPrintf("%3d %3d %5d ", sd->fAccepting, sd->fLookAhead, sd->fTagsIdx); + for (c=0; cfSetBuilder->getNumCharCategories(); c++) { + RBBIDebugPrintf(" %3d", sd->fDtran->elementAti(c)); + } + RBBIDebugPrintf("\n"); + } + RBBIDebugPrintf("\n\n"); +} +#endif + + +//----------------------------------------------------------------------------- +// +// printSafeTable Debug Function. Dump the fully constructed safe table. +// +//----------------------------------------------------------------------------- +#ifdef RBBI_DEBUG +void RBBITableBuilder::printReverseTable() { + int c; // input "character" + int n; // state number + + RBBIDebugPrintf(" Safe Reverse Table \n"); + if (fSafeTable == nullptr) { + RBBIDebugPrintf(" --- nullptr ---\n"); + return; + } + RBBIDebugPrintf("state | i n p u t s y m b o l s \n"); + RBBIDebugPrintf(" | Acc LA Tag"); + for (c=0; cfSetBuilder->getNumCharCategories(); c++) { + RBBIDebugPrintf(" %2d", c); + } + RBBIDebugPrintf("\n"); + RBBIDebugPrintf(" |---------------"); + for (c=0; cfSetBuilder->getNumCharCategories(); c++) { + RBBIDebugPrintf("---"); + } + RBBIDebugPrintf("\n"); + + for (n=0; nsize(); n++) { + UnicodeString *rowString = (UnicodeString *)fSafeTable->elementAt(n); + RBBIDebugPrintf(" %3d | " , n); + RBBIDebugPrintf("%3d %3d %5d ", 0, 0, 0); // Accepting, LookAhead, Tags + for (c=0; cfSetBuilder->getNumCharCategories(); c++) { + RBBIDebugPrintf(" %2d", rowString->charAt(c)); + } + RBBIDebugPrintf("\n"); + } + RBBIDebugPrintf("\n\n"); +} +#endif + + + +//----------------------------------------------------------------------------- +// +// printRuleStatusTable Debug Function. Dump the common rule status table +// +//----------------------------------------------------------------------------- +#ifdef RBBI_DEBUG +void RBBITableBuilder::printRuleStatusTable() { + int32_t thisRecord = 0; + int32_t nextRecord = 0; + int i; + UVector *tbl = fRB->fRuleStatusVals; + + RBBIDebugPrintf("index | tags \n"); + RBBIDebugPrintf("-------------------\n"); + + while (nextRecord < tbl->size()) { + thisRecord = nextRecord; + nextRecord = thisRecord + tbl->elementAti(thisRecord) + 1; + RBBIDebugPrintf("%4d ", thisRecord); + for (i=thisRecord+1; ielementAti(i)); + } + RBBIDebugPrintf("\n"); + } + RBBIDebugPrintf("\n\n"); +} +#endif + + +//----------------------------------------------------------------------------- +// +// RBBIStateDescriptor Methods. This is a very struct-like class +// Most access is directly to the fields. +// +//----------------------------------------------------------------------------- + +RBBIStateDescriptor::RBBIStateDescriptor(int lastInputSymbol, UErrorCode *fStatus) { + fMarked = false; + fAccepting = 0; + fLookAhead = 0; + fTagsIdx = 0; + fTagVals = nullptr; + fPositions = nullptr; + fDtran = nullptr; + + fDtran = new UVector32(lastInputSymbol+1, *fStatus); + if (U_FAILURE(*fStatus)) { + return; + } + if (fDtran == nullptr) { + *fStatus = U_MEMORY_ALLOCATION_ERROR; + return; + } + fDtran->setSize(lastInputSymbol+1); // fDtran needs to be pre-sized. + // It is indexed by input symbols, and will + // hold the next state number for each + // symbol. +} + + +RBBIStateDescriptor::~RBBIStateDescriptor() { + delete fPositions; + delete fDtran; + delete fTagVals; + fPositions = nullptr; + fDtran = nullptr; + fTagVals = nullptr; +} + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ diff --git a/intl/icu/source/common/rbbitblb.h b/intl/icu/source/common/rbbitblb.h new file mode 100644 index 0000000000..01eee3da4e --- /dev/null +++ b/intl/icu/source/common/rbbitblb.h @@ -0,0 +1,232 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// +// rbbitblb.h +// + +/* +********************************************************************** +* Copyright (c) 2002-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ + +#ifndef RBBITBLB_H +#define RBBITBLB_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/uobject.h" +#include "unicode/rbbi.h" +#include "rbbidata.h" +#include "rbbirb.h" +#include "rbbinode.h" + + +U_NAMESPACE_BEGIN + +class RBBIRuleScanner; +class RBBIRuleBuilder; +class UVector32; + +// +// class RBBITableBuilder is part of the RBBI rule compiler. +// It builds the state transition table used by the RBBI runtime +// from the expression syntax tree generated by the rule scanner. +// +// This class is part of the RBBI implementation only. +// There is no user-visible public API here. +// + +class RBBITableBuilder : public UMemory { +public: + RBBITableBuilder(RBBIRuleBuilder *rb, RBBINode **rootNode, UErrorCode &status); + ~RBBITableBuilder(); + + void buildForwardTable(); + + /** Return the runtime size in bytes of the built state table. */ + int32_t getTableSize() const; + + /** Fill in the runtime state table. Sufficient memory must exist at the specified location. + */ + void exportTable(void *where); + + /** Use 8 bits to encode the forward table */ + bool use8BitsForTable() const; + + /** + * Find duplicate (redundant) character classes. Begin looking with categories.first. + * Duplicate, if found are returned in the categories parameter. + * This is an iterator-like function, used to identify character classes + * (state table columns) that can be eliminated. + * @param categories in/out parameter, specifies where to start looking for duplicates, + * and returns the first pair of duplicates found, if any. + * @return true if duplicate char classes were found, false otherwise. + */ + bool findDuplCharClassFrom(IntPair *categories); + + /** Remove a column from the state table. Used when two character categories + * have been found equivalent, and merged together, to eliminate the unneeded table column. + */ + void removeColumn(int32_t column); + + /** + * Check for, and remove duplicate states (table rows). + * @return the number of states removed. + */ + int32_t removeDuplicateStates(); + + /** Build the safe reverse table from the already-constructed forward table. */ + void buildSafeReverseTable(UErrorCode &status); + + /** Return the runtime size in bytes of the built safe reverse state table. */ + int32_t getSafeTableSize() const; + + /** Fill in the runtime safe state table. Sufficient memory must exist at the specified location. + */ + void exportSafeTable(void *where); + + /** Use 8 bits to encode the safe reverse table */ + bool use8BitsForSafeTable() const; + +private: + void calcNullable(RBBINode *n); + void calcFirstPos(RBBINode *n); + void calcLastPos(RBBINode *n); + void calcFollowPos(RBBINode *n); + void calcChainedFollowPos(RBBINode *n, RBBINode *endMarkNode); + void bofFixup(); + void buildStateTable(); + void mapLookAheadRules(); + void flagAcceptingStates(); + void flagLookAheadStates(); + void flagTaggedStates(); + void mergeRuleStatusVals(); + + /** + * Merge redundant state table columns, eliminating character classes with identical behavior. + * Done after the state tables are generated, just before converting to their run-time format. + */ + int32_t mergeColumns(); + + void addRuleRootNodes(UVector *dest, RBBINode *node); + + /** + * Find duplicate (redundant) states, beginning at the specified pair, + * within this state table. This is an iterator-like function, used to + * identify states (state table rows) that can be eliminated. + * @param states in/out parameter, specifies where to start looking for duplicates, + * and returns the first pair of duplicates found, if any. + * @return true if duplicate states were found, false otherwise. + */ + bool findDuplicateState(IntPair *states); + + /** Remove a duplicate state. + * @param duplStates The duplicate states. The first is kept, the second is removed. + * All references to the second in the state table are retargeted + * to the first. + */ + void removeState(IntPair duplStates); + + /** Find the next duplicate state in the safe reverse table. An iterator function. + * @param states in/out parameter, specifies where to start looking for duplicates, + * and returns the first pair of duplicates found, if any. + * @return true if a duplicate pair of states was found. + */ + bool findDuplicateSafeState(IntPair *states); + + /** Remove a duplicate state from the safe table. + * @param duplStates The duplicate states. The first is kept, the second is removed. + * All references to the second in the state table are retargeted + * to the first. + */ + void removeSafeState(IntPair duplStates); + + // Set functions for UVector. + // TODO: make a USet subclass of UVector + + void setAdd(UVector *dest, UVector *source); + UBool setEquals(UVector *a, UVector *b); + + void sortedAdd(UVector **dest, int32_t val); + +public: +#ifdef RBBI_DEBUG + void printSet(UVector *s); + void printPosSets(RBBINode *n /* = nullptr */); + void printStates(); + void printRuleStatusTable(); + void printReverseTable(); +#else + #define printSet(s) + #define printPosSets(n) + #define printStates() + #define printRuleStatusTable() + #define printReverseTable() +#endif + +private: + RBBIRuleBuilder *fRB; + RBBINode *&fTree; // The root node of the parse tree to build a + // table for. + UErrorCode *fStatus; + + /** State Descriptors, UVector */ + UVector *fDStates; // D states (Aho's terminology) + // Index is state number + // Contents are RBBIStateDescriptor pointers. + + /** Synthesized safe table, UVector of UnicodeString, one string per table row. */ + UVector *fSafeTable; + + /** Map from rule number (fVal in look ahead nodes) to sequential lookahead index. */ + UVector32 *fLookAheadRuleMap = nullptr; + + /* Counter used when assigning lookahead rule numbers. + * Contains the last look-ahead number already in use. + * The first look-ahead number is 2; Number 1 (ACCEPTING_UNCONDITIONAL) is reserved + * for non-lookahead accepting states. See the declarations of RBBIStateTableRowT. */ + int32_t fLASlotsInUse = ACCEPTING_UNCONDITIONAL; + + + RBBITableBuilder(const RBBITableBuilder &other) = delete; // forbid copying of this class + RBBITableBuilder &operator=(const RBBITableBuilder &other) = delete; // forbid copying of this class +}; + +// +// RBBIStateDescriptor - The DFA is constructed as a set of these descriptors, +// one for each state. +class RBBIStateDescriptor : public UMemory { +public: + UBool fMarked; + uint32_t fAccepting; + uint32_t fLookAhead; + UVector *fTagVals; + int32_t fTagsIdx; + UVector *fPositions; // Set of parse tree positions associated + // with this state. Unordered (it's a set). + // UVector contents are RBBINode * + + UVector32 *fDtran; // Transitions out of this state. + // indexed by input character + // contents is int index of dest state + // in RBBITableBuilder.fDStates + + RBBIStateDescriptor(int maxInputSymbol, UErrorCode *fStatus); + ~RBBIStateDescriptor(); + +private: + RBBIStateDescriptor(const RBBIStateDescriptor &other) = delete; // forbid copying of this class + RBBIStateDescriptor &operator=(const RBBIStateDescriptor &other) = delete; // forbid copying of this class +}; + + + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ + +#endif diff --git a/intl/icu/source/common/resbund.cpp b/intl/icu/source/common/resbund.cpp new file mode 100644 index 0000000000..54383c981d --- /dev/null +++ b/intl/icu/source/common/resbund.cpp @@ -0,0 +1,399 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1997-2013, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* +* File resbund.cpp +* +* Modification History: +* +* Date Name Description +* 02/05/97 aliu Fixed bug in chopLocale. Added scanForLocaleInFile +* based on code taken from scanForLocale. Added +* constructor which attempts to read resource bundle +* from a specific file, without searching other files. +* 02/11/97 aliu Added UErrorCode return values to constructors. Fixed +* infinite loops in scanForFile and scanForLocale. +* Modified getRawResourceData to not delete storage in +* localeData and resourceData which it doesn't own. +* Added Mac compatibility #ifdefs for tellp() and +* ios::nocreate. +* 03/04/97 aliu Modified to use ExpandingDataSink objects instead of +* the highly inefficient ostrstream objects. +* 03/13/97 aliu Rewrote to load in entire resource bundle and store +* it as a Hashtable of ResourceBundleData objects. +* Added state table to govern parsing of files. +* Modified to load locale index out of new file distinct +* from default.txt. +* 03/25/97 aliu Modified to support 2-d arrays, needed for timezone data. +* Added support for custom file suffixes. Again, needed +* to support timezone data. Improved error handling to +* detect duplicate tags and subtags. +* 04/07/97 aliu Fixed bug in getHashtableForLocale(). Fixed handling +* of failing UErrorCode values on entry to API methods. +* Fixed bugs in getArrayItem() for negative indices. +* 04/29/97 aliu Update to use new Hashtable deletion protocol. +* 05/06/97 aliu Flattened kTransitionTable for HP compiler. +* Fixed usage of CharString. +* 06/11/99 stephen Removed parsing of .txt files. +* Reworked to use new binary format. +* Cleaned up. +* 06/14/99 stephen Removed methods taking a filename suffix. +* 06/22/99 stephen Added missing T_FileStream_close in parse() +* 11/09/99 weiv Added getLocale(), rewritten constructForLocale() +* March 2000 weiv complete overhaul. +****************************************************************************** +*/ + +#include "unicode/utypes.h" +#include "unicode/resbund.h" + +#include "cmemory.h" +#include "mutex.h" +#include "uassert.h" +#include "umutex.h" + +#include "uresimp.h" + +U_NAMESPACE_BEGIN + +/*----------------------------------------------------------------------------- + * Implementation Notes + * + * Resource bundles are read in once, and thereafter cached. + * ResourceBundle statically keeps track of which files have been + * read, so we are guaranteed that each file is read at most once. + * Resource bundles can be loaded from different data directories and + * will be treated as distinct, even if they are for the same locale. + * + * Resource bundles are lightweight objects, which have pointers to + * one or more shared Hashtable objects containing all the data. + * Copying would be cheap, but there is no copy constructor, since + * there wasn't one in the original API. + * + * The ResourceBundle parsing mechanism is implemented as a transition + * network, for easy maintenance and modification. The network is + * implemented as a matrix (instead of in code) to make this even + * easier. The matrix contains Transition objects. Each Transition + * object describes a destination node and an action to take before + * moving to the destination node. The source node is encoded by the + * index of the object in the array that contains it. The pieces + * needed to understand the transition network are the enums for node + * IDs and actions, the parse() method, which walks through the + * network and implements the actions, and the network itself. The + * network guarantees certain conditions, for example, that a new + * resource will not be closed until one has been opened first; or + * that data will not be stored into a TaggedList until a TaggedList + * has been created. Nonetheless, the code in parse() does some + * consistency checks as it runs the network, and fails with an + * U_INTERNAL_PROGRAM_ERROR if one of these checks fails. If the input + * data has a bad format, an U_INVALID_FORMAT_ERROR is returned. If you + * see an U_INTERNAL_PROGRAM_ERROR the transition matrix has a bug in + * it. + * + * Old functionality of multiple locales in a single file is still + * supported. For this reason, LOCALE names override FILE names. If + * data for en_US is located in the en.txt file, once it is loaded, + * the code will not care where it came from (other than remembering + * which directory it came from). However, if there is an en_US + * resource in en_US.txt, that will take precedence. There is no + * limit to the number or type of resources that can be stored in a + * file, however, files are only searched in a specific way. If + * en_US_CA is requested, then first en_US_CA.txt is searched, then + * en_US.txt, then en.txt, then default.txt. So it only makes sense + * to put certain locales in certain files. In this example, it would + * be logical to put en_US_CA, en_US, and en into the en.txt file, + * since they would be found there if asked for. The extreme example + * is to place all locale resources into default.txt, which should + * also work. + * + * Inheritance is implemented. For example, xx_YY_zz inherits as + * follows: xx_YY_zz, xx_YY, xx, default. Inheritance is implemented + * as an array of hashtables. There will be from 1 to 4 hashtables in + * the array. + * + * Fallback files are implemented. The fallback pattern is Language + * Country Variant (LCV) -> LC -> L. Fallback is first done for the + * requested locale. Then it is done for the default locale, as + * returned by Locale::getDefault(). Then the special file + * default.txt is searched for the default locale. The overall FILE + * fallback path is LCV -> LC -> L -> dLCV -> dLC -> dL -> default. + * + * Note that although file name searching includes the default locale, + * once a ResourceBundle object is constructed, the inheritance path + * no longer includes the default locale. The path is LCV -> LC -> L + * -> default. + * + * File parsing is lazy. Nothing is parsed unless it is called for by + * someone. So when a ResourceBundle for xx_YY_zz is constructed, + * only that locale is parsed (along with anything else in the same + * file). Later, if the FooBar tag is asked for, and if it isn't + * found in xx_YY_zz, then xx_YY.txt will be parsed and checked, and + * so forth, until the chain is exhausted or the tag is found. + * + * Thread-safety is implemented around caches, both the cache that + * stores all the resource data, and the cache that stores flags + * indicating whether or not a file has been visited. These caches + * delete their storage at static cleanup time, when the process + * quits. + * + * ResourceBundle supports TableCollation as a special case. This + * involves having special ResourceBundle objects which DO own their + * data, since we don't want large collation rule strings in the + * ResourceBundle cache (these are already cached in the + * TableCollation cache). TableCollation files (.ctx files) have the + * same format as normal resource data files, with a different + * interpretation, from the standpoint of ResourceBundle. .ctx files + * are loaded into otherwise ordinary ResourceBundle objects. They + * don't inherit (that's implemented by TableCollation) and they own + * their data (as mentioned above). However, they still support + * possible multiple locales in a single .ctx file. (This is in + * practice a bad idea, since you only want the one locale you're + * looking for, and only one tag will be present + * ("CollationElements"), so you don't need an inheritance chain of + * multiple locales.) Up to 4 locale resources will be loaded from a + * .ctx file; everything after the first 4 is ignored (parsed and + * deleted). (Normal .txt files have no limit.) Instead of being + * loaded into the cache, and then looked up as needed, the locale + * resources are read straight into the ResourceBundle object. + * + * The Index, which used to reside in default.txt, has been moved to a + * new file, index.txt. This file contains a slightly modified format + * with the addition of the "InstalledLocales" tag; it looks like: + * + * Index { + * InstalledLocales { + * ar + * .. + * zh_TW + * } + * } + */ +//----------------------------------------------------------------------------- + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(ResourceBundle) + +ResourceBundle::ResourceBundle(UErrorCode &err) + :UObject(), fLocale(nullptr) +{ + fResource = ures_open(0, Locale::getDefault().getName(), &err); +} + +ResourceBundle::ResourceBundle(const ResourceBundle &other) + :UObject(other), fLocale(nullptr) +{ + UErrorCode status = U_ZERO_ERROR; + + if (other.fResource) { + fResource = ures_copyResb(0, other.fResource, &status); + } else { + /* Copying a bad resource bundle */ + fResource = nullptr; + } +} + +ResourceBundle::ResourceBundle(UResourceBundle *res, UErrorCode& err) + :UObject(), fLocale(nullptr) +{ + if (res) { + fResource = ures_copyResb(0, res, &err); + } else { + /* Copying a bad resource bundle */ + fResource = nullptr; + } +} + +ResourceBundle::ResourceBundle(const char* path, const Locale& locale, UErrorCode& err) + :UObject(), fLocale(nullptr) +{ + fResource = ures_open(path, locale.getName(), &err); +} + + +ResourceBundle& ResourceBundle::operator=(const ResourceBundle& other) +{ + if(this == &other) { + return *this; + } + if(fResource != 0) { + ures_close(fResource); + fResource = nullptr; + } + if (fLocale != nullptr) { + delete fLocale; + fLocale = nullptr; + } + UErrorCode status = U_ZERO_ERROR; + if (other.fResource) { + fResource = ures_copyResb(0, other.fResource, &status); + } else { + /* Copying a bad resource bundle */ + fResource = nullptr; + } + return *this; +} + +ResourceBundle::~ResourceBundle() +{ + if(fResource != 0) { + ures_close(fResource); + } + if(fLocale != nullptr) { + delete(fLocale); + } +} + +ResourceBundle * +ResourceBundle::clone() const { + return new ResourceBundle(*this); +} + +UnicodeString ResourceBundle::getString(UErrorCode& status) const { + int32_t len = 0; + const char16_t *r = ures_getString(fResource, &len, &status); + return UnicodeString(true, r, len); +} + +const uint8_t *ResourceBundle::getBinary(int32_t& len, UErrorCode& status) const { + return ures_getBinary(fResource, &len, &status); +} + +const int32_t *ResourceBundle::getIntVector(int32_t& len, UErrorCode& status) const { + return ures_getIntVector(fResource, &len, &status); +} + +uint32_t ResourceBundle::getUInt(UErrorCode& status) const { + return ures_getUInt(fResource, &status); +} + +int32_t ResourceBundle::getInt(UErrorCode& status) const { + return ures_getInt(fResource, &status); +} + +const char *ResourceBundle::getName() const { + return ures_getName(fResource); +} + +const char *ResourceBundle::getKey() const { + return ures_getKey(fResource); +} + +UResType ResourceBundle::getType() const { + return ures_getType(fResource); +} + +int32_t ResourceBundle::getSize() const { + return ures_getSize(fResource); +} + +UBool ResourceBundle::hasNext() const { + return ures_hasNext(fResource); +} + +void ResourceBundle::resetIterator() { + ures_resetIterator(fResource); +} + +ResourceBundle ResourceBundle::getNext(UErrorCode& status) { + UResourceBundle r; + + ures_initStackObject(&r); + ures_getNextResource(fResource, &r, &status); + ResourceBundle res(&r, status); + if (U_SUCCESS(status)) { + ures_close(&r); + } + return res; +} + +UnicodeString ResourceBundle::getNextString(UErrorCode& status) { + int32_t len = 0; + const char16_t* r = ures_getNextString(fResource, &len, 0, &status); + return UnicodeString(true, r, len); +} + +UnicodeString ResourceBundle::getNextString(const char ** key, UErrorCode& status) { + int32_t len = 0; + const char16_t* r = ures_getNextString(fResource, &len, key, &status); + return UnicodeString(true, r, len); +} + +ResourceBundle ResourceBundle::get(int32_t indexR, UErrorCode& status) const { + UResourceBundle r; + + ures_initStackObject(&r); + ures_getByIndex(fResource, indexR, &r, &status); + ResourceBundle res(&r, status); + if (U_SUCCESS(status)) { + ures_close(&r); + } + return res; +} + +UnicodeString ResourceBundle::getStringEx(int32_t indexS, UErrorCode& status) const { + int32_t len = 0; + const char16_t* r = ures_getStringByIndex(fResource, indexS, &len, &status); + return UnicodeString(true, r, len); +} + +ResourceBundle ResourceBundle::get(const char* key, UErrorCode& status) const { + UResourceBundle r; + + ures_initStackObject(&r); + ures_getByKey(fResource, key, &r, &status); + ResourceBundle res(&r, status); + if (U_SUCCESS(status)) { + ures_close(&r); + } + return res; +} + +ResourceBundle ResourceBundle::getWithFallback(const char* key, UErrorCode& status){ + UResourceBundle r; + ures_initStackObject(&r); + ures_getByKeyWithFallback(fResource, key, &r, &status); + ResourceBundle res(&r, status); + if(U_SUCCESS(status)){ + ures_close(&r); + } + return res; +} +UnicodeString ResourceBundle::getStringEx(const char* key, UErrorCode& status) const { + int32_t len = 0; + const char16_t* r = ures_getStringByKey(fResource, key, &len, &status); + return UnicodeString(true, r, len); +} + +const char* +ResourceBundle::getVersionNumber() const +{ + return ures_getVersionNumberInternal(fResource); +} + +void ResourceBundle::getVersion(UVersionInfo versionInfo) const { + ures_getVersion(fResource, versionInfo); +} + +const Locale &ResourceBundle::getLocale() const { + static UMutex gLocaleLock; + Mutex lock(&gLocaleLock); + if (fLocale != nullptr) { + return *fLocale; + } + UErrorCode status = U_ZERO_ERROR; + const char *localeName = ures_getLocaleInternal(fResource, &status); + ResourceBundle *ncThis = const_cast(this); + ncThis->fLocale = new Locale(localeName); + return ncThis->fLocale != nullptr ? *ncThis->fLocale : Locale::getDefault(); +} + +const Locale ResourceBundle::getLocale(ULocDataLocaleType type, UErrorCode &status) const +{ + return ures_getLocaleByType(fResource, type, &status); +} + +U_NAMESPACE_END +//eof diff --git a/intl/icu/source/common/resbund_cnv.cpp b/intl/icu/source/common/resbund_cnv.cpp new file mode 100644 index 0000000000..7f83f06f0e --- /dev/null +++ b/intl/icu/source/common/resbund_cnv.cpp @@ -0,0 +1,57 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 1997-2006, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: resbund_cnv.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2004aug25 +* created by: Markus W. Scherer +* +* Character conversion functions moved here from resbund.cpp +*/ + +#include "unicode/utypes.h" +#include "unicode/resbund.h" +#include "uinvchar.h" + +U_NAMESPACE_BEGIN + +ResourceBundle::ResourceBundle( const UnicodeString& path, + const Locale& locale, + UErrorCode& error) + :UObject(), fLocale(nullptr) +{ + constructForLocale(path, locale, error); +} + +ResourceBundle::ResourceBundle( const UnicodeString& path, + UErrorCode& error) + :UObject(), fLocale(nullptr) +{ + constructForLocale(path, Locale::getDefault(), error); +} + +void +ResourceBundle::constructForLocale(const UnicodeString& path, + const Locale& locale, + UErrorCode& error) +{ + if (path.isEmpty()) { + fResource = ures_open(nullptr, locale.getName(), &error); + } + else { + UnicodeString nullTerminatedPath(path); + nullTerminatedPath.append((char16_t)0); + fResource = ures_openU(nullTerminatedPath.getBuffer(), locale.getName(), &error); + } +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/resource.cpp b/intl/icu/source/common/resource.cpp new file mode 100644 index 0000000000..3d41a16029 --- /dev/null +++ b/intl/icu/source/common/resource.cpp @@ -0,0 +1,22 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2015-2016, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* resource.cpp +* +* created on: 2015nov04 +* created by: Markus W. Scherer +*/ + +#include "resource.h" + +U_NAMESPACE_BEGIN + +ResourceValue::~ResourceValue() {} + +ResourceSink::~ResourceSink() {} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/resource.h b/intl/icu/source/common/resource.h new file mode 100644 index 0000000000..79ed897bd1 --- /dev/null +++ b/intl/icu/source/common/resource.h @@ -0,0 +1,295 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2015-2016, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* resource.h +* +* created on: 2015nov04 +* created by: Markus W. Scherer +*/ + +#ifndef __URESOURCE_H__ +#define __URESOURCE_H__ + +/** + * \file + * \brief ICU resource bundle key and value types. + */ + +// Note: Ported from ICU4J class UResource and its nested classes, +// but the C++ classes are separate, not nested. + +// We use the Resource prefix for C++ classes, as usual. +// The UResource prefix would be used for C types. + +#include "unicode/utypes.h" +#include "unicode/unistr.h" +#include "unicode/ures.h" +#include "restrace.h" + +struct ResourceData; + +U_NAMESPACE_BEGIN + +class ResourceValue; + +// Note: In C++, we use const char * pointers for keys, +// rather than an abstraction like Java UResource.Key. + +/** + * Interface for iterating over a resource bundle array resource. + */ +class U_COMMON_API ResourceArray { +public: + /** Constructs an empty array object. */ + ResourceArray() : items16(nullptr), items32(nullptr), length(0) {} + + /** Only for implementation use. @internal */ + ResourceArray(const uint16_t *i16, const uint32_t *i32, int32_t len, + const ResourceTracer& traceInfo) : + items16(i16), items32(i32), length(len), + fTraceInfo(traceInfo) {} + + /** + * @return The number of items in the array resource. + */ + int32_t getSize() const { return length; } + /** + * @param i Array item index. + * @param value Output-only, receives the value of the i'th item. + * @return true if i is non-negative and less than getSize(). + */ + UBool getValue(int32_t i, ResourceValue &value) const; + + /** Only for implementation use. @internal */ + uint32_t internalGetResource(const ResourceData *pResData, int32_t i) const; + +private: + const uint16_t *items16; + const uint32_t *items32; + int32_t length; + ResourceTracer fTraceInfo; +}; + +/** + * Interface for iterating over a resource bundle table resource. + */ +class U_COMMON_API ResourceTable { +public: + /** Constructs an empty table object. */ + ResourceTable() : keys16(nullptr), keys32(nullptr), items16(nullptr), items32(nullptr), length(0) {} + + /** Only for implementation use. @internal */ + ResourceTable(const uint16_t *k16, const int32_t *k32, + const uint16_t *i16, const uint32_t *i32, int32_t len, + const ResourceTracer& traceInfo) : + keys16(k16), keys32(k32), items16(i16), items32(i32), length(len), + fTraceInfo(traceInfo) {} + + /** + * @return The number of items in the array resource. + */ + int32_t getSize() const { return length; } + /** + * @param i Table item index. + * @param key Output-only, receives the key of the i'th item. + * @param value Output-only, receives the value of the i'th item. + * @return true if i is non-negative and less than getSize(). + */ + UBool getKeyAndValue(int32_t i, const char *&key, ResourceValue &value) const; + + /** + * @param key Key string to find in the table. + * @param value Output-only, receives the value of the item with that key. + * @return true if the table contains the key. + */ + UBool findValue(const char *key, ResourceValue &value) const; + +private: + const uint16_t *keys16; + const int32_t *keys32; + const uint16_t *items16; + const uint32_t *items32; + int32_t length; + ResourceTracer fTraceInfo; +}; + +/** + * Represents a resource bundle item's value. + * Avoids object creations as much as possible. + * Mutable, not thread-safe. + */ +class U_COMMON_API ResourceValue : public UObject { +public: + virtual ~ResourceValue(); + + /** + * @return ICU resource type, for example, URES_STRING + */ + virtual UResType getType() const = 0; + + /** + * Sets U_RESOURCE_TYPE_MISMATCH if this is not a string resource. + * + * @see ures_getString() + */ + virtual const char16_t *getString(int32_t &length, UErrorCode &errorCode) const = 0; + + inline UnicodeString getUnicodeString(UErrorCode &errorCode) const { + int32_t len = 0; + const char16_t *r = getString(len, errorCode); + return UnicodeString(true, r, len); + } + + /** + * Sets U_RESOURCE_TYPE_MISMATCH if this is not an alias resource. + */ + virtual const char16_t *getAliasString(int32_t &length, UErrorCode &errorCode) const = 0; + + inline UnicodeString getAliasUnicodeString(UErrorCode &errorCode) const { + int32_t len = 0; + const char16_t *r = getAliasString(len, errorCode); + return UnicodeString(true, r, len); + } + + /** + * Sets U_RESOURCE_TYPE_MISMATCH if this is not an integer resource. + * + * @see ures_getInt() + */ + virtual int32_t getInt(UErrorCode &errorCode) const = 0; + + /** + * Sets U_RESOURCE_TYPE_MISMATCH if this is not an integer resource. + * + * @see ures_getUInt() + */ + virtual uint32_t getUInt(UErrorCode &errorCode) const = 0; + + /** + * Sets U_RESOURCE_TYPE_MISMATCH if this is not an intvector resource. + * + * @see ures_getIntVector() + */ + virtual const int32_t *getIntVector(int32_t &length, UErrorCode &errorCode) const = 0; + + /** + * Sets U_RESOURCE_TYPE_MISMATCH if this is not a binary-blob resource. + * + * @see ures_getBinary() + */ + virtual const uint8_t *getBinary(int32_t &length, UErrorCode &errorCode) const = 0; + + /** + * Sets U_RESOURCE_TYPE_MISMATCH if this is not an array resource + */ + virtual ResourceArray getArray(UErrorCode &errorCode) const = 0; + + /** + * Sets U_RESOURCE_TYPE_MISMATCH if this is not a table resource + */ + virtual ResourceTable getTable(UErrorCode &errorCode) const = 0; + + /** + * Is this a no-fallback/no-inheritance marker string? + * Such a marker is used for + * CLDR no-fallback data values of (three empty-set symbols)=={2205, 2205, 2205} + * when enumerating tables with fallback from the specific resource bundle to root. + * + * @return true if this is a no-inheritance marker string + */ + virtual UBool isNoInheritanceMarker() const = 0; + + /** + * Sets the dest strings from the string values in this array resource. + * + * @return the number of strings in this array resource. + * If greater than capacity, then an overflow error is set. + * + * Sets U_RESOURCE_TYPE_MISMATCH if this is not an array resource + * or if any of the array items is not a string + */ + virtual int32_t getStringArray(UnicodeString *dest, int32_t capacity, + UErrorCode &errorCode) const = 0; + + /** + * Same as + * 

+     * if (getType() == URES_STRING) {
+     *     return new String[] { getString(); }
+     * } else {
+     *     return getStringArray();
+     * }
+     *
+ * + * Sets U_RESOURCE_TYPE_MISMATCH if this is + * neither a string resource nor an array resource containing strings + * @see getString() + * @see getStringArray() + */ + virtual int32_t getStringArrayOrStringAsArray(UnicodeString *dest, int32_t capacity, + UErrorCode &errorCode) const = 0; + + /** + * Same as + * 
+     * if (getType() == URES_STRING) {
+     *     return getString();
+     * } else {
+     *     return getStringArray()[0];
+     * }
+     *
+ * + * Sets U_RESOURCE_TYPE_MISMATCH if this is + * neither a string resource nor an array resource containing strings + * @see getString() + * @see getStringArray() + */ + virtual UnicodeString getStringOrFirstOfArray(UErrorCode &errorCode) const = 0; + +protected: + ResourceValue() {} + +private: + ResourceValue(const ResourceValue &); // no copy constructor + ResourceValue &operator=(const ResourceValue &); // no assignment operator +}; + +/** + * Sink for ICU resource bundle contents. + */ +class U_COMMON_API ResourceSink : public UObject { +public: + ResourceSink() {} + virtual ~ResourceSink(); + + /** + * Called once for each bundle (child-parent-...-root). + * The value is normally an array or table resource, + * and implementations of this method normally iterate over the + * tree of resource items stored there. + * + * @param key The key string of the enumeration-start resource. + * Empty if the enumeration starts at the top level of the bundle. + * @param value Call getArray() or getTable() as appropriate. Then reuse for + * output values from Array and Table getters. Note: ResourceTable and + * ResourceArray instances must outlive the ResourceValue instance for + * ResourceTracer to be happy. + * @param noFallback true if the bundle has no parent; + * that is, its top-level table has the nofallback attribute, + * or it is the root bundle of a locale tree. + */ + virtual void put(const char *key, ResourceValue &value, UBool noFallback, + UErrorCode &errorCode) = 0; + +private: + ResourceSink(const ResourceSink &) = delete; // no copy constructor + ResourceSink &operator=(const ResourceSink &) = delete; // no assignment operator +}; + +U_NAMESPACE_END + +#endif diff --git a/intl/icu/source/common/restrace.cpp b/intl/icu/source/common/restrace.cpp new file mode 100644 index 0000000000..1f83372d68 --- /dev/null +++ b/intl/icu/source/common/restrace.cpp @@ -0,0 +1,133 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +#include "unicode/utypes.h" + +#if U_ENABLE_TRACING + +#include "restrace.h" +#include "charstr.h" +#include "cstring.h" +#include "utracimp.h" +#include "uresimp.h" +#include "uassert.h" +#include "util.h" + +U_NAMESPACE_BEGIN + +ResourceTracer::~ResourceTracer() = default; + +void ResourceTracer::trace(const char* resType) const { + U_ASSERT(fResB || fParent); + UTRACE_ENTRY(UTRACE_UDATA_RESOURCE); + UErrorCode status = U_ZERO_ERROR; + + CharString filePath; + getFilePath(filePath, status); + + CharString resPath; + getResPath(resPath, status); + + // The longest type ("intvector") is 9 chars + const char kSpaces[] = " "; + CharString format; + format.append(kSpaces, sizeof(kSpaces) - 1 - uprv_strlen(resType), status); + format.append("(%s) %s @ %s", status); + + UTRACE_DATA3(UTRACE_VERBOSE, + format.data(), + resType, + filePath.data(), + resPath.data()); + UTRACE_EXIT_STATUS(status); +} + +void ResourceTracer::traceOpen() const { + U_ASSERT(fResB); + UTRACE_ENTRY(UTRACE_UDATA_BUNDLE); + UErrorCode status = U_ZERO_ERROR; + + CharString filePath; + UTRACE_DATA1(UTRACE_VERBOSE, "%s", getFilePath(filePath, status).data()); + UTRACE_EXIT_STATUS(status); +} + +CharString& ResourceTracer::getFilePath(CharString& output, UErrorCode& status) const { + if (fResB) { + // Note: if you get a segfault around here, check that ResourceTable and + // ResourceArray instances outlive ResourceValue instances referring to + // their contents: + output.append(fResB->fData->fPath, status); + output.append('/', status); + output.append(fResB->fData->fName, status); + output.append(".res", status); + } else { + fParent->getFilePath(output, status); + } + return output; +} + +CharString& ResourceTracer::getResPath(CharString& output, UErrorCode& status) const { + if (fResB) { + output.append('/', status); + output.append(fResB->fResPath, status); + // removing the trailing / + U_ASSERT(output[output.length()-1] == '/'); + output.truncate(output.length()-1); + } else { + fParent->getResPath(output, status); + } + if (fKey) { + output.append('/', status); + output.append(fKey, status); + } + if (fIndex != -1) { + output.append('[', status); + UnicodeString indexString; + ICU_Utility::appendNumber(indexString, fIndex); + output.appendInvariantChars(indexString, status); + output.append(']', status); + } + return output; +} + +void FileTracer::traceOpen(const char* path, const char* type, const char* name) { + if (uprv_strcmp(type, "res") == 0) { + traceOpenResFile(path, name); + } else { + traceOpenDataFile(path, type, name); + } +} + +void FileTracer::traceOpenDataFile(const char* path, const char* type, const char* name) { + UTRACE_ENTRY(UTRACE_UDATA_DATA_FILE); + UErrorCode status = U_ZERO_ERROR; + + CharString filePath; + filePath.append(path, status); + filePath.append('/', status); + filePath.append(name, status); + filePath.append('.', status); + filePath.append(type, status); + + UTRACE_DATA1(UTRACE_VERBOSE, "%s", filePath.data()); + UTRACE_EXIT_STATUS(status); +} + +void FileTracer::traceOpenResFile(const char* path, const char* name) { + UTRACE_ENTRY(UTRACE_UDATA_RES_FILE); + UErrorCode status = U_ZERO_ERROR; + + CharString filePath; + filePath.append(path, status); + filePath.append('/', status); + filePath.append(name, status); + filePath.append(".res", status); + + UTRACE_DATA1(UTRACE_VERBOSE, "%s", filePath.data()); + UTRACE_EXIT_STATUS(status); +} + +U_NAMESPACE_END + +#endif // U_ENABLE_TRACING diff --git a/intl/icu/source/common/restrace.h b/intl/icu/source/common/restrace.h new file mode 100644 index 0000000000..ef29eaed57 --- /dev/null +++ b/intl/icu/source/common/restrace.h @@ -0,0 +1,147 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +#ifndef __RESTRACE_H__ +#define __RESTRACE_H__ + +#include "unicode/utypes.h" + +#if U_ENABLE_TRACING + +struct UResourceBundle; + +U_NAMESPACE_BEGIN + +class CharString; + +/** + * Instances of this class store information used to trace reads from resource + * bundles when ICU is built with --enable-tracing. + * + * All arguments of type const UResourceBundle*, const char*, and + * const ResourceTracer& are stored as pointers. The caller must retain + * ownership for the lifetime of this ResourceTracer. + * + * Exported as U_COMMON_API for Windows because it is a value field + * in other exported types. + */ +class U_COMMON_API ResourceTracer { +public: + ResourceTracer() : + fResB(nullptr), + fParent(nullptr), + fKey(nullptr), + fIndex(-1) {} + + ResourceTracer(const UResourceBundle* resB) : + fResB(resB), + fParent(nullptr), + fKey(nullptr), + fIndex(-1) {} + + ResourceTracer(const UResourceBundle* resB, const char* key) : + fResB(resB), + fParent(nullptr), + fKey(key), + fIndex(-1) {} + + ResourceTracer(const UResourceBundle* resB, int32_t index) : + fResB(resB), + fParent(nullptr), + fKey(nullptr), + fIndex(index) {} + + ResourceTracer(const ResourceTracer& parent, const char* key) : + fResB(nullptr), + fParent(&parent), + fKey(key), + fIndex(-1) {} + + ResourceTracer(const ResourceTracer& parent, int32_t index) : + fResB(nullptr), + fParent(&parent), + fKey(nullptr), + fIndex(index) {} + + ~ResourceTracer(); + + void trace(const char* type) const; + void traceOpen() const; + + /** + * Calls trace() if the resB or parent provided to the constructor was + * non-null; otherwise, does nothing. + */ + void maybeTrace(const char* type) const { + if (fResB || fParent) { + trace(type); + } + } + +private: + const UResourceBundle* fResB; + const ResourceTracer* fParent; + const char* fKey; + int32_t fIndex; + + CharString& getFilePath(CharString& output, UErrorCode& status) const; + + CharString& getResPath(CharString& output, UErrorCode& status) const; +}; + +/** + * This class provides methods to trace data file reads when ICU is built + * with --enable-tracing. + */ +class FileTracer { +public: + static void traceOpen(const char* path, const char* type, const char* name); + +private: + static void traceOpenDataFile(const char* path, const char* type, const char* name); + static void traceOpenResFile(const char* path, const char* name); +}; + +U_NAMESPACE_END + +#else // U_ENABLE_TRACING + +U_NAMESPACE_BEGIN + +/** + * Default trivial implementation when --enable-tracing is not used. + */ +class U_COMMON_API ResourceTracer { +public: + ResourceTracer() {} + + ResourceTracer(const void*) {} + + ResourceTracer(const void*, const char*) {} + + ResourceTracer(const void*, int32_t) {} + + ResourceTracer(const ResourceTracer&, const char*) {} + + ResourceTracer(const ResourceTracer&, int32_t) {} + + void trace(const char*) const {} + + void traceOpen() const {} + + void maybeTrace(const char*) const {} +}; + +/** + * Default trivial implementation when --enable-tracing is not used. + */ +class FileTracer { +public: + static void traceOpen(const char*, const char*, const char*) {} +}; + +U_NAMESPACE_END + +#endif // U_ENABLE_TRACING + +#endif //__RESTRACE_H__ diff --git a/intl/icu/source/common/ruleiter.cpp b/intl/icu/source/common/ruleiter.cpp new file mode 100644 index 0000000000..690635a908 --- /dev/null +++ b/intl/icu/source/common/ruleiter.cpp @@ -0,0 +1,162 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2003-2011, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* Author: Alan Liu +* Created: September 24 2003 +* Since: ICU 2.8 +********************************************************************** +*/ +#include "ruleiter.h" +#include "unicode/parsepos.h" +#include "unicode/symtable.h" +#include "unicode/unistr.h" +#include "unicode/utf16.h" +#include "patternprops.h" + +/* \U87654321 or \ud800\udc00 */ +#define MAX_U_NOTATION_LEN 12 + +U_NAMESPACE_BEGIN + +RuleCharacterIterator::RuleCharacterIterator(const UnicodeString& theText, const SymbolTable* theSym, + ParsePosition& thePos) : + text(theText), + pos(thePos), + sym(theSym), + buf(0), + bufPos(0) +{} + +UBool RuleCharacterIterator::atEnd() const { + return buf == 0 && pos.getIndex() == text.length(); +} + +UChar32 RuleCharacterIterator::next(int32_t options, UBool& isEscaped, UErrorCode& ec) { + if (U_FAILURE(ec)) return DONE; + + UChar32 c = DONE; + isEscaped = false; + + for (;;) { + c = _current(); + _advance(U16_LENGTH(c)); + + if (c == SymbolTable::SYMBOL_REF && buf == 0 && + (options & PARSE_VARIABLES) != 0 && sym != 0) { + UnicodeString name = sym->parseReference(text, pos, text.length()); + // If name is empty there was an isolated SYMBOL_REF; + // return it. Caller must be prepared for this. + if (name.length() == 0) { + break; + } + bufPos = 0; + buf = sym->lookup(name); + if (buf == 0) { + ec = U_UNDEFINED_VARIABLE; + return DONE; + } + // Handle empty variable value + if (buf->length() == 0) { + buf = 0; + } + continue; + } + + if ((options & SKIP_WHITESPACE) != 0 && PatternProps::isWhiteSpace(c)) { + continue; + } + + if (c == 0x5C /*'\\'*/ && (options & PARSE_ESCAPES) != 0) { + UnicodeString tempEscape; + int32_t offset = 0; + c = lookahead(tempEscape, MAX_U_NOTATION_LEN).unescapeAt(offset); + jumpahead(offset); + isEscaped = true; + if (c < 0) { + ec = U_MALFORMED_UNICODE_ESCAPE; + return DONE; + } + } + + break; + } + + return c; +} + +void RuleCharacterIterator::getPos(RuleCharacterIterator::Pos& p) const { + p.buf = buf; + p.pos = pos.getIndex(); + p.bufPos = bufPos; +} + +void RuleCharacterIterator::setPos(const RuleCharacterIterator::Pos& p) { + buf = p.buf; + pos.setIndex(p.pos); + bufPos = p.bufPos; +} + +void RuleCharacterIterator::skipIgnored(int32_t options) { + if ((options & SKIP_WHITESPACE) != 0) { + for (;;) { + UChar32 a = _current(); + if (!PatternProps::isWhiteSpace(a)) break; + _advance(U16_LENGTH(a)); + } + } +} + +UnicodeString& RuleCharacterIterator::lookahead(UnicodeString& result, int32_t maxLookAhead) const { + if (maxLookAhead < 0) { + maxLookAhead = 0x7FFFFFFF; + } + if (buf != 0) { + buf->extract(bufPos, maxLookAhead, result); + } else { + text.extract(pos.getIndex(), maxLookAhead, result); + } + return result; +} + +void RuleCharacterIterator::jumpahead(int32_t count) { + _advance(count); +} + +/* +UnicodeString& RuleCharacterIterator::toString(UnicodeString& result) const { + int32_t b = pos.getIndex(); + text.extract(0, b, result); + return result.append((char16_t) 0x7C).append(text, b, 0x7FFFFFFF); // Insert '|' at index +} +*/ + +UChar32 RuleCharacterIterator::_current() const { + if (buf != 0) { + return buf->char32At(bufPos); + } else { + int i = pos.getIndex(); + return (i < text.length()) ? text.char32At(i) : (UChar32)DONE; + } +} + +void RuleCharacterIterator::_advance(int32_t count) { + if (buf != 0) { + bufPos += count; + if (bufPos == buf->length()) { + buf = 0; + } + } else { + pos.setIndex(pos.getIndex() + count); + if (pos.getIndex() > text.length()) { + pos.setIndex(text.length()); + } + } +} + +U_NAMESPACE_END + +//eof diff --git a/intl/icu/source/common/ruleiter.h b/intl/icu/source/common/ruleiter.h new file mode 100644 index 0000000000..41731407da --- /dev/null +++ b/intl/icu/source/common/ruleiter.h @@ -0,0 +1,233 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2003-2011, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* Author: Alan Liu +* Created: September 24 2003 +* Since: ICU 2.8 +********************************************************************** +*/ +#ifndef _RULEITER_H_ +#define _RULEITER_H_ + +#include "unicode/uobject.h" + +U_NAMESPACE_BEGIN + +class UnicodeString; +class ParsePosition; +class SymbolTable; + +/** + * An iterator that returns 32-bit code points. This class is deliberately + * not related to any of the ICU character iterator classes + * in order to minimize complexity. + * @author Alan Liu + * @since ICU 2.8 + */ +class RuleCharacterIterator : public UMemory { + + // TODO: Ideas for later. (Do not implement if not needed, lest the + // code coverage numbers go down due to unused methods.) + // 1. Add a copy constructor, operator==() method. + // 2. Rather than return DONE, throw an exception if the end + // is reached -- this is an alternate usage model, probably not useful. + +private: + /** + * Text being iterated. + */ + const UnicodeString& text; + + /** + * Position of iterator. + */ + ParsePosition& pos; + + /** + * Symbol table used to parse and dereference variables. May be 0. + */ + const SymbolTable* sym; + + /** + * Current variable expansion, or 0 if none. + */ + const UnicodeString* buf; + + /** + * Position within buf. Meaningless if buf == 0. + */ + int32_t bufPos; + +public: + /** + * Value returned when there are no more characters to iterate. + */ + static constexpr int32_t DONE = -1; + + /** + * Bitmask option to enable parsing of variable names. If (options & + * PARSE_VARIABLES) != 0, then an embedded variable will be expanded to + * its value. Variables are parsed using the SymbolTable API. + */ + static constexpr int32_t PARSE_VARIABLES = 1; + + /** + * Bitmask option to enable parsing of escape sequences. If (options & + * PARSE_ESCAPES) != 0, then an embedded escape sequence will be expanded + * to its value. Escapes are parsed using Utility.unescapeAt(). + */ + static constexpr int32_t PARSE_ESCAPES = 2; + + /** + * Bitmask option to enable skipping of whitespace. If (options & + * SKIP_WHITESPACE) != 0, then Pattern_White_Space characters will be silently + * skipped, as if they were not present in the input. + */ + static constexpr int32_t SKIP_WHITESPACE = 4; + + /** + * Constructs an iterator over the given text, starting at the given + * position. + * @param text the text to be iterated + * @param sym the symbol table, or null if there is none. If sym is null, + * then variables will not be dereferenced, even if the PARSE_VARIABLES + * option is set. + * @param pos upon input, the index of the next character to return. If a + * variable has been dereferenced, then pos will not increment as + * characters of the variable value are iterated. + */ + RuleCharacterIterator(const UnicodeString& text, const SymbolTable* sym, + ParsePosition& pos); + + /** + * Returns true if this iterator has no more characters to return. + */ + UBool atEnd() const; + + /** + * Returns the next character using the given options, or DONE if there + * are no more characters, and advance the position to the next + * character. + * @param options one or more of the following options, bitwise-OR-ed + * together: PARSE_VARIABLES, PARSE_ESCAPES, SKIP_WHITESPACE. + * @param isEscaped output parameter set to true if the character + * was escaped + * @param ec input-output error code. An error will only be set by + * this routing if options includes PARSE_VARIABLES and an unknown + * variable name is seen, or if options includes PARSE_ESCAPES and + * an invalid escape sequence is seen. + * @return the current 32-bit code point, or DONE + */ + UChar32 next(int32_t options, UBool& isEscaped, UErrorCode& ec); + + /** + * Returns true if this iterator is currently within a variable expansion. + */ + inline UBool inVariable() const; + + /** + * An opaque object representing the position of a RuleCharacterIterator. + */ + struct Pos : public UMemory { + private: + const UnicodeString* buf; + int32_t pos; + int32_t bufPos; + friend class RuleCharacterIterator; + }; + + /** + * Sets an object which, when later passed to setPos(), will + * restore this iterator's position. Usage idiom: + * + * RuleCharacterIterator iterator = ...; + * RuleCharacterIterator::Pos pos; + * iterator.getPos(pos); + * for (;;) { + * iterator.getPos(pos); + * int c = iterator.next(...); + * ... + * } + * iterator.setPos(pos); + * + * @param p a position object to be set to this iterator's + * current position. + */ + void getPos(Pos& p) const; + + /** + * Restores this iterator to the position it had when getPos() + * set the given object. + * @param p a position object previously set by getPos() + */ + void setPos(const Pos& p); + + /** + * Skips ahead past any ignored characters, as indicated by the given + * options. This is useful in conjunction with the lookahead() method. + * + * Currently, this only has an effect for SKIP_WHITESPACE. + * @param options one or more of the following options, bitwise-OR-ed + * together: PARSE_VARIABLES, PARSE_ESCAPES, SKIP_WHITESPACE. + */ + void skipIgnored(int32_t options); + + /** + * Returns a string containing the remainder of the characters to be + * returned by this iterator, without any option processing. If the + * iterator is currently within a variable expansion, this will only + * extend to the end of the variable expansion. This method is provided + * so that iterators may interoperate with string-based APIs. The typical + * sequence of calls is to call skipIgnored(), then call lookahead(), then + * parse the string returned by lookahead(), then call jumpahead() to + * resynchronize the iterator. + * @param result a string to receive the characters to be returned + * by future calls to next() + * @param maxLookAhead The maximum to copy into the result. + * @return a reference to result + */ + UnicodeString& lookahead(UnicodeString& result, int32_t maxLookAhead = -1) const; + + /** + * Advances the position by the given number of 16-bit code units. + * This is useful in conjunction with the lookahead() method. + * @param count the number of 16-bit code units to jump over + */ + void jumpahead(int32_t count); + + /** + * Returns a string representation of this object, consisting of the + * characters being iterated, with a '|' marking the current position. + * Position within an expanded variable is not indicated. + * @param result output parameter to receive a string + * representation of this object + */ +// UnicodeString& toString(UnicodeString& result) const; + +private: + /** + * Returns the current 32-bit code point without parsing escapes, parsing + * variables, or skipping whitespace. + * @return the current 32-bit code point + */ + UChar32 _current() const; + + /** + * Advances the position by the given amount. + * @param count the number of 16-bit code units to advance past + */ + void _advance(int32_t count); +}; + +inline UBool RuleCharacterIterator::inVariable() const { + return buf != 0; +} + +U_NAMESPACE_END + +#endif // _RULEITER_H_ +//eof diff --git a/intl/icu/source/common/schriter.cpp b/intl/icu/source/common/schriter.cpp new file mode 100644 index 0000000000..d4bac8fd4f --- /dev/null +++ b/intl/icu/source/common/schriter.cpp @@ -0,0 +1,119 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 1998-2012, International Business Machines Corporation and +* others. All Rights Reserved. +****************************************************************************** +* +* File schriter.cpp +* +* Modification History: +* +* Date Name Description +* 05/05/99 stephen Cleaned up. +****************************************************************************** +*/ + +#include "utypeinfo.h" // for 'typeid' to work + +#include "unicode/chariter.h" +#include "unicode/schriter.h" + +U_NAMESPACE_BEGIN + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(StringCharacterIterator) + +StringCharacterIterator::StringCharacterIterator() + : UCharCharacterIterator(), + text() +{ + // NEVER DEFAULT CONSTRUCT! +} + +StringCharacterIterator::StringCharacterIterator(const UnicodeString& textStr) + : UCharCharacterIterator(textStr.getBuffer(), textStr.length()), + text(textStr) +{ + // we had set the input parameter's array, now we need to set our copy's array + UCharCharacterIterator::text = this->text.getBuffer(); +} + +StringCharacterIterator::StringCharacterIterator(const UnicodeString& textStr, + int32_t textPos) + : UCharCharacterIterator(textStr.getBuffer(), textStr.length(), textPos), + text(textStr) +{ + // we had set the input parameter's array, now we need to set our copy's array + UCharCharacterIterator::text = this->text.getBuffer(); +} + +StringCharacterIterator::StringCharacterIterator(const UnicodeString& textStr, + int32_t textBegin, + int32_t textEnd, + int32_t textPos) + : UCharCharacterIterator(textStr.getBuffer(), textStr.length(), textBegin, textEnd, textPos), + text(textStr) +{ + // we had set the input parameter's array, now we need to set our copy's array + UCharCharacterIterator::text = this->text.getBuffer(); +} + +StringCharacterIterator::StringCharacterIterator(const StringCharacterIterator& that) + : UCharCharacterIterator(that), + text(that.text) +{ + // we had set the input parameter's array, now we need to set our copy's array + UCharCharacterIterator::text = this->text.getBuffer(); +} + +StringCharacterIterator::~StringCharacterIterator() { +} + +StringCharacterIterator& +StringCharacterIterator::operator=(const StringCharacterIterator& that) { + UCharCharacterIterator::operator=(that); + text = that.text; + // we had set the input parameter's array, now we need to set our copy's array + UCharCharacterIterator::text = this->text.getBuffer(); + return *this; +} + +bool +StringCharacterIterator::operator==(const ForwardCharacterIterator& that) const { + if (this == &that) { + return true; + } + + // do not call UCharCharacterIterator::operator==() + // because that checks for array pointer equality + // while we compare UnicodeString objects + + if (typeid(*this) != typeid(that)) { + return false; + } + + const StringCharacterIterator& realThat = static_cast(that); + + return text == realThat.text + && pos == realThat.pos + && begin == realThat.begin + && end == realThat.end; +} + +StringCharacterIterator* +StringCharacterIterator::clone() const { + return new StringCharacterIterator(*this); +} + +void +StringCharacterIterator::setText(const UnicodeString& newText) { + text = newText; + UCharCharacterIterator::setText(text.getBuffer(), text.length()); +} + +void +StringCharacterIterator::getText(UnicodeString& result) { + result = text; +} +U_NAMESPACE_END diff --git a/intl/icu/source/common/serv.cpp b/intl/icu/source/common/serv.cpp new file mode 100644 index 0000000000..5248f7c192 --- /dev/null +++ b/intl/icu/source/common/serv.cpp @@ -0,0 +1,968 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** +******************************************************************************* +* Copyright (C) 2001-2014, International Business Machines Corporation. +* All Rights Reserved. +******************************************************************************* +*/ + +#include "unicode/utypes.h" +#include "unicode/localpointer.h" + +#if !UCONFIG_NO_SERVICE + +#include "serv.h" +#include "umutex.h" + +#undef SERVICE_REFCOUNT + +// in case we use the refcount stuff + +U_NAMESPACE_BEGIN + +/* +****************************************************************** +*/ + +const char16_t ICUServiceKey::PREFIX_DELIMITER = 0x002F; /* '/' */ + +ICUServiceKey::ICUServiceKey(const UnicodeString& id) +: _id(id) { +} + +ICUServiceKey::~ICUServiceKey() +{ +} + +const UnicodeString& +ICUServiceKey::getID() const +{ + return _id; +} + +UnicodeString& +ICUServiceKey::canonicalID(UnicodeString& result) const +{ + return result.append(_id); +} + +UnicodeString& +ICUServiceKey::currentID(UnicodeString& result) const +{ + return canonicalID(result); +} + +UnicodeString& +ICUServiceKey::currentDescriptor(UnicodeString& result) const +{ + prefix(result); + result.append(PREFIX_DELIMITER); + return currentID(result); +} + +UBool +ICUServiceKey::fallback() +{ + return false; +} + +UBool +ICUServiceKey::isFallbackOf(const UnicodeString& id) const +{ + return id == _id; +} + +UnicodeString& +ICUServiceKey::prefix(UnicodeString& result) const +{ + return result; +} + +UnicodeString& +ICUServiceKey::parsePrefix(UnicodeString& result) +{ + int32_t n = result.indexOf(PREFIX_DELIMITER); + if (n < 0) { + n = 0; + } + result.remove(n); + return result; +} + +UnicodeString& +ICUServiceKey::parseSuffix(UnicodeString& result) +{ + int32_t n = result.indexOf(PREFIX_DELIMITER); + if (n >= 0) { + result.remove(0, n+1); + } + return result; +} + +#ifdef SERVICE_DEBUG +UnicodeString& +ICUServiceKey::debug(UnicodeString& result) const +{ + debugClass(result); + result.append((UnicodeString)" id: "); + result.append(_id); + return result; +} + +UnicodeString& +ICUServiceKey::debugClass(UnicodeString& result) const +{ + return result.append((UnicodeString)"ICUServiceKey"); +} +#endif + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(ICUServiceKey) + +/* +****************************************************************** +*/ + +ICUServiceFactory::~ICUServiceFactory() {} + +SimpleFactory::SimpleFactory(UObject* instanceToAdopt, const UnicodeString& id, UBool visible) +: _instance(instanceToAdopt), _id(id), _visible(visible) +{ +} + +SimpleFactory::~SimpleFactory() +{ + delete _instance; +} + +UObject* +SimpleFactory::create(const ICUServiceKey& key, const ICUService* service, UErrorCode& status) const +{ + if (U_SUCCESS(status)) { + UnicodeString temp; + if (_id == key.currentID(temp)) { + return service->cloneInstance(_instance); + } + } + return nullptr; +} + +void +SimpleFactory::updateVisibleIDs(Hashtable& result, UErrorCode& status) const +{ + if (_visible) { + result.put(_id, (void*)this, status); // cast away const + } else { + result.remove(_id); + } +} + +UnicodeString& +SimpleFactory::getDisplayName(const UnicodeString& id, const Locale& /* locale */, UnicodeString& result) const +{ + if (_visible && _id == id) { + result = _id; + } else { + result.setToBogus(); + } + return result; +} + +#ifdef SERVICE_DEBUG +UnicodeString& +SimpleFactory::debug(UnicodeString& toAppendTo) const +{ + debugClass(toAppendTo); + toAppendTo.append((UnicodeString)" id: "); + toAppendTo.append(_id); + toAppendTo.append((UnicodeString)", visible: "); + toAppendTo.append(_visible ? (UnicodeString)"T" : (UnicodeString)"F"); + return toAppendTo; +} + +UnicodeString& +SimpleFactory::debugClass(UnicodeString& toAppendTo) const +{ + return toAppendTo.append((UnicodeString)"SimpleFactory"); +} +#endif + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(SimpleFactory) + +/* +****************************************************************** +*/ + +ServiceListener::~ServiceListener() {} + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(ServiceListener) + +/* +****************************************************************** +*/ + +// Record the actual id for this service in the cache, so we can return it +// even if we succeed later with a different id. +class CacheEntry : public UMemory { +private: + int32_t refcount; + +public: + UnicodeString actualDescriptor; + UObject* service; + + /** + * Releases a reference to the shared resource. + */ + ~CacheEntry() { + delete service; + } + + CacheEntry(const UnicodeString& _actualDescriptor, UObject* _service) + : refcount(1), actualDescriptor(_actualDescriptor), service(_service) { + } + + /** + * Instantiation creates an initial reference, so don't call this + * unless you're creating a new pointer to this. Management of + * that pointer will have to know how to deal with refcounts. + * Return true if the resource has not already been released. + */ + CacheEntry* ref() { + ++refcount; + return this; + } + + /** + * Destructions removes a reference, so don't call this unless + * you're removing pointer to this somewhere. Management of that + * pointer will have to know how to deal with refcounts. Once + * the refcount drops to zero, the resource is released. Return + * false if the resource has been released. + */ + CacheEntry* unref() { + if ((--refcount) == 0) { + delete this; + return nullptr; + } + return this; + } + + /** + * Return true if there is at least one reference to this and the + * resource has not been released. + */ + UBool isShared() const { + return refcount > 1; + } +}; + +// Deleter for serviceCache +U_CDECL_BEGIN +static void U_CALLCONV +cacheDeleter(void* obj) { + U_NAMESPACE_USE ((CacheEntry*)obj)->unref(); +} + +U_CDECL_END + +/* +****************************************************************** +*/ + +class DNCache : public UMemory { +public: + Hashtable cache; + const Locale locale; + + DNCache(const Locale& _locale) + : cache(), locale(_locale) + { + // cache.setKeyDeleter(uprv_deleteUObject); + } +}; + + +/* +****************************************************************** +*/ + +StringPair* +StringPair::create(const UnicodeString& displayName, + const UnicodeString& id, + UErrorCode& status) +{ + if (U_SUCCESS(status)) { + StringPair* sp = new StringPair(displayName, id); + if (sp == nullptr || sp->isBogus()) { + status = U_MEMORY_ALLOCATION_ERROR; + delete sp; + return nullptr; + } + return sp; + } + return nullptr; +} + +UBool +StringPair::isBogus() const { + return displayName.isBogus() || id.isBogus(); +} + +StringPair::StringPair(const UnicodeString& _displayName, + const UnicodeString& _id) +: displayName(_displayName) +, id(_id) +{ +} + +U_CDECL_BEGIN +static void U_CALLCONV +userv_deleteStringPair(void *obj) { + U_NAMESPACE_USE delete (StringPair*) obj; +} +U_CDECL_END + +/* +****************************************************************** +*/ + +static UMutex lock; + +ICUService::ICUService() +: name() +, timestamp(0) +, factories(nullptr) +, serviceCache(nullptr) +, idCache(nullptr) +, dnCache(nullptr) +{ +} + +ICUService::ICUService(const UnicodeString& newName) +: name(newName) +, timestamp(0) +, factories(nullptr) +, serviceCache(nullptr) +, idCache(nullptr) +, dnCache(nullptr) +{ +} + +ICUService::~ICUService() +{ + { + Mutex mutex(&lock); + clearCaches(); + delete factories; + factories = nullptr; + } +} + +UObject* +ICUService::get(const UnicodeString& descriptor, UErrorCode& status) const +{ + return get(descriptor, nullptr, status); +} + +UObject* +ICUService::get(const UnicodeString& descriptor, UnicodeString* actualReturn, UErrorCode& status) const +{ + UObject* result = nullptr; + ICUServiceKey* key = createKey(&descriptor, status); + if (key) { + result = getKey(*key, actualReturn, status); + delete key; + } + return result; +} + +UObject* +ICUService::getKey(ICUServiceKey& key, UErrorCode& status) const +{ + return getKey(key, nullptr, status); +} + +// this is a vector that subclasses of ICUService can override to further customize the result object +// before returning it. All other public get functions should call this one. + +UObject* +ICUService::getKey(ICUServiceKey& key, UnicodeString* actualReturn, UErrorCode& status) const +{ + return getKey(key, actualReturn, nullptr, status); +} + +// make it possible to call reentrantly on systems that don't have reentrant mutexes. +// we can use this simple approach since we know the situation where we're calling +// reentrantly even without knowing the thread. +class XMutex : public UMemory { +public: + inline XMutex(UMutex *mutex, UBool reentering) + : fMutex(mutex) + , fActive(!reentering) + { + if (fActive) umtx_lock(fMutex); + } + inline ~XMutex() { + if (fActive) umtx_unlock(fMutex); + } + +private: + UMutex *fMutex; + UBool fActive; +}; + +// called only by factories, treat as private +UObject* +ICUService::getKey(ICUServiceKey& key, UnicodeString* actualReturn, const ICUServiceFactory* factory, UErrorCode& status) const +{ + if (U_FAILURE(status)) { + return nullptr; + } + + if (isDefault()) { + return handleDefault(key, actualReturn, status); + } + + ICUService* ncthis = (ICUService*)this; // cast away semantic const + + CacheEntry* result = nullptr; + { + // The factory list can't be modified until we're done, + // otherwise we might update the cache with an invalid result. + // The cache has to stay in synch with the factory list. + // ICU doesn't have monitors so we can't use rw locks, so + // we single-thread everything using this service, for now. + + // if factory is not null, we're calling from within the mutex, + // and since some unix machines don't have reentrant mutexes we + // need to make sure not to try to lock it again. + XMutex mutex(&lock, factory != nullptr); + + if (serviceCache == nullptr) { + ncthis->serviceCache = new Hashtable(status); + if (ncthis->serviceCache == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + if (U_FAILURE(status)) { + delete serviceCache; + return nullptr; + } + serviceCache->setValueDeleter(cacheDeleter); + } + + UnicodeString currentDescriptor; + LocalPointer cacheDescriptorList; + UBool putInCache = false; + + int32_t startIndex = 0; + int32_t limit = factories->size(); + UBool cacheResult = true; + + if (factory != nullptr) { + for (int32_t i = 0; i < limit; ++i) { + if (factory == (const ICUServiceFactory*)factories->elementAt(i)) { + startIndex = i + 1; + break; + } + } + if (startIndex == 0) { + // throw new InternalError("Factory " + factory + "not registered with service: " + this); + status = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + cacheResult = false; + } + + do { + currentDescriptor.remove(); + key.currentDescriptor(currentDescriptor); + result = (CacheEntry*)serviceCache->get(currentDescriptor); + if (result != nullptr) { + break; + } + + // first test of cache failed, so we'll have to update + // the cache if we eventually succeed-- that is, if we're + // going to update the cache at all. + putInCache = true; + + int32_t index = startIndex; + while (index < limit) { + ICUServiceFactory* f = (ICUServiceFactory*)factories->elementAt(index++); + LocalPointer service(f->create(key, this, status)); + if (U_FAILURE(status)) { + return nullptr; + } + if (service.isValid()) { + result = new CacheEntry(currentDescriptor, service.getAlias()); + if (result == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + service.orphan(); // result now owns service. + + goto outerEnd; + } + } + + // prepare to load the cache with all additional ids that + // will resolve to result, assuming we'll succeed. We + // don't want to keep querying on an id that's going to + // fallback to the one that succeeded, we want to hit the + // cache the first time next goaround. + if (cacheDescriptorList.isNull()) { + cacheDescriptorList.adoptInsteadAndCheckErrorCode(new UVector(uprv_deleteUObject, nullptr, 5, status), status); + if (U_FAILURE(status)) { + return nullptr; + } + } + + LocalPointer idToCache(new UnicodeString(currentDescriptor), status); + if (U_FAILURE(status)) { + return nullptr; + } + if (idToCache->isBogus()) { + status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + cacheDescriptorList->adoptElement(idToCache.orphan(), status); + if (U_FAILURE(status)) { + return nullptr; + } + } while (key.fallback()); +outerEnd: + + if (result != nullptr) { + if (putInCache && cacheResult) { + serviceCache->put(result->actualDescriptor, result, status); + if (U_FAILURE(status)) { + return nullptr; + } + + if (cacheDescriptorList.isValid()) { + for (int32_t i = cacheDescriptorList->size(); --i >= 0;) { + UnicodeString* desc = (UnicodeString*)cacheDescriptorList->elementAt(i); + + serviceCache->put(*desc, result, status); + if (U_FAILURE(status)) { + return nullptr; + } + + result->ref(); + cacheDescriptorList->removeElementAt(i); + } + } + } + + if (actualReturn != nullptr) { + // strip null prefix + if (result->actualDescriptor.indexOf((char16_t)0x2f) == 0) { // U+002f=slash (/) + actualReturn->remove(); + actualReturn->append(result->actualDescriptor, + 1, + result->actualDescriptor.length() - 1); + } else { + *actualReturn = result->actualDescriptor; + } + + if (actualReturn->isBogus()) { + status = U_MEMORY_ALLOCATION_ERROR; + delete result; + return nullptr; + } + } + + UObject* service = cloneInstance(result->service); + if (putInCache && !cacheResult) { + delete result; + } + return service; + } + } + + return handleDefault(key, actualReturn, status); +} + +UObject* +ICUService::handleDefault(const ICUServiceKey& /* key */, UnicodeString* /* actualIDReturn */, UErrorCode& /* status */) const +{ + return nullptr; +} + +UVector& +ICUService::getVisibleIDs(UVector& result, UErrorCode& status) const { + return getVisibleIDs(result, nullptr, status); +} + +UVector& +ICUService::getVisibleIDs(UVector& result, const UnicodeString* matchID, UErrorCode& status) const +{ + result.removeAllElements(); + + if (U_FAILURE(status)) { + return result; + } + UObjectDeleter *savedDeleter = result.setDeleter(uprv_deleteUObject); + + { + Mutex mutex(&lock); + const Hashtable* map = getVisibleIDMap(status); + if (map != nullptr) { + ICUServiceKey* fallbackKey = createKey(matchID, status); + + for (int32_t pos = UHASH_FIRST; U_SUCCESS(status); ) { + const UHashElement* e = map->nextElement(pos); + if (e == nullptr) { + break; + } + + const UnicodeString* id = (const UnicodeString*)e->key.pointer; + if (fallbackKey != nullptr) { + if (!fallbackKey->isFallbackOf(*id)) { + continue; + } + } + + LocalPointer idClone(id->clone(), status); + result.adoptElement(idClone.orphan(), status); + } + delete fallbackKey; + } + } + if (U_FAILURE(status)) { + result.removeAllElements(); + } + result.setDeleter(savedDeleter); + return result; +} + +const Hashtable* +ICUService::getVisibleIDMap(UErrorCode& status) const { + if (U_FAILURE(status)) return nullptr; + + // must only be called when lock is already held + + ICUService* ncthis = (ICUService*)this; // cast away semantic const + if (idCache == nullptr) { + ncthis->idCache = new Hashtable(status); + if (idCache == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + } else if (factories != nullptr) { + for (int32_t pos = factories->size(); --pos >= 0;) { + ICUServiceFactory* f = (ICUServiceFactory*)factories->elementAt(pos); + f->updateVisibleIDs(*idCache, status); + } + if (U_FAILURE(status)) { + delete idCache; + ncthis->idCache = nullptr; + } + } + } + + return idCache; +} + + +UnicodeString& +ICUService::getDisplayName(const UnicodeString& id, UnicodeString& result) const +{ + return getDisplayName(id, result, Locale::getDefault()); +} + +UnicodeString& +ICUService::getDisplayName(const UnicodeString& id, UnicodeString& result, const Locale& locale) const +{ + { + UErrorCode status = U_ZERO_ERROR; + Mutex mutex(&lock); + const Hashtable* map = getVisibleIDMap(status); + if (map != nullptr) { + ICUServiceFactory* f = (ICUServiceFactory*)map->get(id); + if (f != nullptr) { + f->getDisplayName(id, locale, result); + return result; + } + + // fallback + status = U_ZERO_ERROR; + ICUServiceKey* fallbackKey = createKey(&id, status); + while (fallbackKey != nullptr && fallbackKey->fallback()) { + UnicodeString us; + fallbackKey->currentID(us); + f = (ICUServiceFactory*)map->get(us); + if (f != nullptr) { + f->getDisplayName(id, locale, result); + delete fallbackKey; + return result; + } + } + delete fallbackKey; + } + } + result.setToBogus(); + return result; +} + +UVector& +ICUService::getDisplayNames(UVector& result, UErrorCode& status) const +{ + return getDisplayNames(result, Locale::getDefault(), nullptr, status); +} + + +UVector& +ICUService::getDisplayNames(UVector& result, const Locale& locale, UErrorCode& status) const +{ + return getDisplayNames(result, locale, nullptr, status); +} + +UVector& +ICUService::getDisplayNames(UVector& result, + const Locale& locale, + const UnicodeString* matchID, + UErrorCode& status) const +{ + result.removeAllElements(); + result.setDeleter(userv_deleteStringPair); + if (U_SUCCESS(status)) { + ICUService* ncthis = (ICUService*)this; // cast away semantic const + Mutex mutex(&lock); + + if (dnCache != nullptr && dnCache->locale != locale) { + delete dnCache; + ncthis->dnCache = nullptr; + } + + if (dnCache == nullptr) { + const Hashtable* m = getVisibleIDMap(status); + if (U_FAILURE(status)) { + return result; + } + ncthis->dnCache = new DNCache(locale); + if (dnCache == nullptr) { + status = U_MEMORY_ALLOCATION_ERROR; + return result; + } + + int32_t pos = UHASH_FIRST; + const UHashElement* entry = nullptr; + while ((entry = m->nextElement(pos)) != nullptr) { + const UnicodeString* id = (const UnicodeString*)entry->key.pointer; + ICUServiceFactory* f = (ICUServiceFactory*)entry->value.pointer; + UnicodeString dname; + f->getDisplayName(*id, locale, dname); + if (dname.isBogus()) { + status = U_MEMORY_ALLOCATION_ERROR; + } else { + dnCache->cache.put(dname, (void*)id, status); // share pointer with visibleIDMap + if (U_SUCCESS(status)) { + continue; + } + } + delete dnCache; + ncthis->dnCache = nullptr; + return result; + } + } + } + + ICUServiceKey* matchKey = createKey(matchID, status); + /* To ensure that all elements in the hashtable are iterated, set pos to -1. + * nextElement(pos) will skip the position at pos and begin the iteration + * at the next position, which in this case will be 0. + */ + int32_t pos = UHASH_FIRST; + const UHashElement *entry = nullptr; + while ((entry = dnCache->cache.nextElement(pos)) != nullptr) { + const UnicodeString* id = (const UnicodeString*)entry->value.pointer; + if (matchKey != nullptr && !matchKey->isFallbackOf(*id)) { + continue; + } + const UnicodeString* dn = (const UnicodeString*)entry->key.pointer; + StringPair* sp = StringPair::create(*id, *dn, status); + result.adoptElement(sp, status); + if (U_FAILURE(status)) { + result.removeAllElements(); + break; + } + } + delete matchKey; + + return result; +} + +URegistryKey +ICUService::registerInstance(UObject* objToAdopt, const UnicodeString& id, UErrorCode& status) +{ + return registerInstance(objToAdopt, id, true, status); +} + +URegistryKey +ICUService::registerInstance(UObject* objToAdopt, const UnicodeString& id, UBool visible, UErrorCode& status) +{ + ICUServiceKey* key = createKey(&id, status); + if (key != nullptr) { + UnicodeString canonicalID; + key->canonicalID(canonicalID); + delete key; + + ICUServiceFactory* f = createSimpleFactory(objToAdopt, canonicalID, visible, status); + if (f != nullptr) { + return registerFactory(f, status); + } + } + delete objToAdopt; + return nullptr; +} + +ICUServiceFactory* +ICUService::createSimpleFactory(UObject* objToAdopt, const UnicodeString& id, UBool visible, UErrorCode& status) +{ + if (U_SUCCESS(status)) { + if ((objToAdopt != nullptr) && (!id.isBogus())) { + return new SimpleFactory(objToAdopt, id, visible); + } + status = U_ILLEGAL_ARGUMENT_ERROR; + } + return nullptr; +} + +URegistryKey +ICUService::registerFactory(ICUServiceFactory* factoryToAdopt, UErrorCode& status) +{ + LocalPointerlpFactoryToAdopt(factoryToAdopt); + if (U_FAILURE(status) || factoryToAdopt == nullptr) { + return nullptr; + } + { + Mutex mutex(&lock); + + if (factories == nullptr) { + LocalPointer lpFactories(new UVector(uprv_deleteUObject, nullptr, status), status); + if (U_FAILURE(status)) { + return nullptr; + } + factories = lpFactories.orphan(); + } + factories->insertElementAt(lpFactoryToAdopt.orphan(), 0, status); + if (U_SUCCESS(status)) { + clearCaches(); + } + } // Close of mutex lock block. + + if (U_SUCCESS(status)) { + notifyChanged(); + return (URegistryKey)factoryToAdopt; + } else { + return nullptr; + } +} + +UBool +ICUService::unregister(URegistryKey rkey, UErrorCode& status) +{ + ICUServiceFactory *factory = (ICUServiceFactory*)rkey; + UBool result = false; + if (factory != nullptr && factories != nullptr) { + Mutex mutex(&lock); + + if (factories->removeElement(factory)) { + clearCaches(); + result = true; + } else { + status = U_ILLEGAL_ARGUMENT_ERROR; + delete factory; + } + } + if (result) { + notifyChanged(); + } + return result; +} + +void +ICUService::reset() +{ + { + Mutex mutex(&lock); + reInitializeFactories(); + clearCaches(); + } + notifyChanged(); +} + +void +ICUService::reInitializeFactories() +{ + if (factories != nullptr) { + factories->removeAllElements(); + } +} + +UBool +ICUService::isDefault() const +{ + return countFactories() == 0; +} + +ICUServiceKey* +ICUService::createKey(const UnicodeString* id, UErrorCode& status) const +{ + return (U_FAILURE(status) || id == nullptr) ? nullptr : new ICUServiceKey(*id); +} + +void +ICUService::clearCaches() +{ + // callers synchronize before use + ++timestamp; + delete dnCache; + dnCache = nullptr; + delete idCache; + idCache = nullptr; + delete serviceCache; serviceCache = nullptr; +} + +void +ICUService::clearServiceCache() +{ + // callers synchronize before use + delete serviceCache; serviceCache = nullptr; +} + +UBool +ICUService::acceptsListener(const EventListener& l) const +{ + return dynamic_cast(&l) != nullptr; +} + +void +ICUService::notifyListener(EventListener& l) const +{ + (static_cast(l)).serviceChanged(*this); +} + +UnicodeString& +ICUService::getName(UnicodeString& result) const +{ + return result.append(name); +} + +int32_t +ICUService::countFactories() const +{ + return factories == nullptr ? 0 : factories->size(); +} + +int32_t +ICUService::getTimestamp() const +{ + return timestamp; +} + +U_NAMESPACE_END + +/* UCONFIG_NO_SERVICE */ +#endif diff --git a/intl/icu/source/common/serv.h b/intl/icu/source/common/serv.h new file mode 100644 index 0000000000..9aea548fc3 --- /dev/null +++ b/intl/icu/source/common/serv.h @@ -0,0 +1,996 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** + ******************************************************************************* + * Copyright (C) 2001-2011, International Business Machines Corporation. * + * All Rights Reserved. * + ******************************************************************************* + */ + +#ifndef ICUSERV_H +#define ICUSERV_H + +#include "unicode/utypes.h" + +#if UCONFIG_NO_SERVICE + +U_NAMESPACE_BEGIN + +/* + * Allow the declaration of APIs with pointers to ICUService + * even when service is removed from the build. + */ +class ICUService; + +U_NAMESPACE_END + +#else + +#include "unicode/unistr.h" +#include "unicode/locid.h" +#include "unicode/umisc.h" + +#include "hash.h" +#include "uvector.h" +#include "servnotf.h" + +class ICUServiceTest; + +U_NAMESPACE_BEGIN + +class ICUServiceKey; +class ICUServiceFactory; +class SimpleFactory; +class ServiceListener; +class ICUService; + +class DNCache; + +/******************************************************************* + * ICUServiceKey + */ + +/** + *

ICUServiceKeys are used to communicate with factories to + * generate an instance of the service. ICUServiceKeys define how + * ids are canonicalized, provide both a current id and a current + * descriptor to use in querying the cache and factories, and + * determine the fallback strategy.

+ * + *

ICUServiceKeys provide both a currentDescriptor and a currentID. + * The descriptor contains an optional prefix, followed by '/' + * and the currentID. Factories that handle complex keys, + * for example number format factories that generate multiple + * kinds of formatters for the same locale, use the descriptor + * to provide a fully unique identifier for the service object, + * while using the currentID (in this case, the locale string), + * as the visible IDs that can be localized.

+ * + *

The default implementation of ICUServiceKey has no fallbacks and + * has no custom descriptors.

+ */ +class U_COMMON_API ICUServiceKey : public UObject { + private: + const UnicodeString _id; + + protected: + static const char16_t PREFIX_DELIMITER; + + public: + + /** + *

Construct a key from an id.

+ * + * @param id the ID from which to construct the key. + */ + ICUServiceKey(const UnicodeString& id); + + /** + *

Virtual destructor.

+ */ + virtual ~ICUServiceKey(); + + /** + *

Return the original ID used to construct this key.

+ * + * @return the ID used to construct this key. + */ + virtual const UnicodeString& getID() const; + + /** + *

Return the canonical version of the original ID. This implementation + * appends the original ID to result. Result is returned as a convenience.

+ * + * @param result the output parameter to which the id will be appended. + * @return the modified result. + */ + virtual UnicodeString& canonicalID(UnicodeString& result) const; + + /** + *

Return the (canonical) current ID. This implementation appends + * the canonical ID to result. Result is returned as a convenience.

+ * + * @param result the output parameter to which the current id will be appended. + * @return the modified result. + */ + virtual UnicodeString& currentID(UnicodeString& result) const; + + /** + *

Return the current descriptor. This implementation appends + * the current descriptor to result. Result is returned as a convenience.

+ * + *

The current descriptor is used to fully + * identify an instance of the service in the cache. A + * factory may handle all descriptors for an ID, or just a + * particular descriptor. The factory can either parse the + * descriptor or use custom API on the key in order to + * instantiate the service.

+ * + * @param result the output parameter to which the current id will be appended. + * @return the modified result. + */ + virtual UnicodeString& currentDescriptor(UnicodeString& result) const; + + /** + *

If the key has a fallback, modify the key and return true, + * otherwise return false. The current ID will change if there + * is a fallback. No currentIDs should be repeated, and fallback + * must eventually return false. This implementation has no fallbacks + * and always returns false.

+ * + * @return true if the ICUServiceKey changed to a valid fallback value. + */ + virtual UBool fallback(); + + /** + *

Return true if a key created from id matches, or would eventually + * fallback to match, the canonical ID of this ICUServiceKey.

+ * + * @param id the id to test. + * @return true if this ICUServiceKey's canonical ID is a fallback of id. + */ + virtual UBool isFallbackOf(const UnicodeString& id) const; + + /** + *

Return the prefix. This implementation leaves result unchanged. + * Result is returned as a convenience.

+ * + * @param result the output parameter to which the prefix will be appended. + * @return the modified result. + */ + virtual UnicodeString& prefix(UnicodeString& result) const; + + /** + *

A utility to parse the prefix out of a descriptor string. Only + * the (undelimited) prefix, if any, remains in result. Result is returned as a + * convenience.

+ * + * @param result an input/output parameter that on entry is a descriptor, and + * on exit is the prefix of that descriptor. + * @return the modified result. + */ + static UnicodeString& parsePrefix(UnicodeString& result); + + /** + *

A utility to parse the suffix out of a descriptor string. Only + * the (undelimited) suffix, if any, remains in result. Result is returned as a + * convenience.

+ * + * @param result an input/output parameter that on entry is a descriptor, and + * on exit is the suffix of that descriptor. + * @return the modified result. + */ + static UnicodeString& parseSuffix(UnicodeString& result); + +public: + /** + * UObject RTTI boilerplate. + */ + static UClassID U_EXPORT2 getStaticClassID(); + + /** + * UObject RTTI boilerplate. + */ + virtual UClassID getDynamicClassID() const override; + +#ifdef SERVICE_DEBUG + public: + virtual UnicodeString& debug(UnicodeString& result) const; + virtual UnicodeString& debugClass(UnicodeString& result) const; +#endif + +}; + + /******************************************************************* + * ICUServiceFactory + */ + + /** + *

An implementing ICUServiceFactory generates the service objects maintained by the + * service. A factory generates a service object from a key, + * updates id->factory mappings, and returns the display name for + * a supported id.

+ */ +class U_COMMON_API ICUServiceFactory : public UObject { + public: + virtual ~ICUServiceFactory(); + + /** + *

Create a service object from the key, if this factory + * supports the key. Otherwise, return nullptr.

+ * + *

If the factory supports the key, then it can call + * the service's getKey(ICUServiceKey, String[], ICUServiceFactory) method + * passing itself as the factory to get the object that + * the service would have created prior to the factory's + * registration with the service. This can change the + * key, so any information required from the key should + * be extracted before making such a callback.

+ * + * @param key the service key. + * @param service the service with which this factory is registered. + * @param status the error code status. + * @return the service object, or nullptr if the factory does not support the key. + */ + virtual UObject* create(const ICUServiceKey& key, const ICUService* service, UErrorCode& status) const = 0; + + /** + *

Update result to reflect the IDs (not descriptors) that this + * factory publicly handles. Result contains mappings from ID to + * factory. On entry it will contain all (visible) mappings from + * previously-registered factories.

+ * + *

This function, together with getDisplayName, are used to + * support ICUService::getDisplayNames. The factory determines + * which IDs (of those it supports) it will make visible, and of + * those, which it will provide localized display names for. In + * most cases it will register mappings from all IDs it supports + * to itself.

+ * + * @param result the mapping table to update. + * @param status the error code status. + */ + virtual void updateVisibleIDs(Hashtable& result, UErrorCode& status) const = 0; + + /** + *

Return, in result, the display name of the id in the provided locale. + * This is an id, not a descriptor. If the id is + * not visible, sets result to bogus. If the + * incoming result is bogus, it remains bogus. Result is returned as a + * convenience. Results are not defined if id is not one supported by this + * factory.

+ * + * @param id a visible id supported by this factory. + * @param locale the locale for which to generate the corresponding localized display name. + * @param result output parameter to hold the display name. + * @return result. + */ + virtual UnicodeString& getDisplayName(const UnicodeString& id, const Locale& locale, UnicodeString& result) const = 0; +}; + +/* + ****************************************************************** + */ + + /** + *

A default implementation of factory. This provides default + * implementations for subclasses, and implements a singleton + * factory that matches a single ID and returns a single + * (possibly deferred-initialized) instance. This implements + * updateVisibleIDs to add a mapping from its ID to itself + * if visible is true, or to remove any existing mapping + * for its ID if visible is false. No localization of display + * names is performed.

+ */ +class U_COMMON_API SimpleFactory : public ICUServiceFactory { + protected: + UObject* _instance; + const UnicodeString _id; + const UBool _visible; + + public: + /** + *

Construct a SimpleFactory that maps a single ID to a single + * service instance. If visible is true, the ID will be visible. + * The instance must not be nullptr. The SimpleFactory will adopt + * the instance, which must not be changed subsequent to this call.

+ * + * @param instanceToAdopt the service instance to adopt. + * @param id the ID to assign to this service instance. + * @param visible if true, the ID will be visible. + */ + SimpleFactory(UObject* instanceToAdopt, const UnicodeString& id, UBool visible = true); + + /** + *

Destructor.

+ */ + virtual ~SimpleFactory(); + + /** + *

This implementation returns a clone of the service instance if the factory's ID is equal to + * the key's currentID. Service and prefix are ignored.

+ * + * @param key the service key. + * @param service the service with which this factory is registered. + * @param status the error code status. + * @return the service object, or nullptr if the factory does not support the key. + */ + virtual UObject* create(const ICUServiceKey& key, const ICUService* service, UErrorCode& status) const override; + + /** + *

This implementation adds a mapping from ID -> this to result if visible is true, + * otherwise it removes ID from result.

+ * + * @param result the mapping table to update. + * @param status the error code status. + */ + virtual void updateVisibleIDs(Hashtable& result, UErrorCode& status) const override; + + /** + *

This implementation returns the factory ID if it equals id and visible is true, + * otherwise it returns the empty string. (This implementation provides + * no localized id information.)

+ * + * @param id a visible id supported by this factory. + * @param locale the locale for which to generate the corresponding localized display name. + * @param result output parameter to hold the display name. + * @return result. + */ + virtual UnicodeString& getDisplayName(const UnicodeString& id, const Locale& locale, UnicodeString& result) const override; + +public: + /** + * UObject RTTI boilerplate. + */ + static UClassID U_EXPORT2 getStaticClassID(); + + /** + * UObject RTTI boilerplate. + */ + virtual UClassID getDynamicClassID() const override; + +#ifdef SERVICE_DEBUG + public: + virtual UnicodeString& debug(UnicodeString& toAppendTo) const; + virtual UnicodeString& debugClass(UnicodeString& toAppendTo) const; +#endif + +}; + +/* + ****************************************************************** + */ + +/** + *

ServiceListener is the listener that ICUService provides by default. + * ICUService will notify this listener when factories are added to + * or removed from the service. Subclasses can provide + * different listener interfaces that extend EventListener, and modify + * acceptsListener and notifyListener as appropriate.

+ */ +class U_COMMON_API ServiceListener : public EventListener { +public: + virtual ~ServiceListener(); + + /** + *

This method is called when the service changes. At the time of the + * call this listener is registered with the service. It must + * not modify the notifier in the context of this call.

+ * + * @param service the service that changed. + */ + virtual void serviceChanged(const ICUService& service) const = 0; + +public: + /** + * UObject RTTI boilerplate. + */ + static UClassID U_EXPORT2 getStaticClassID(); + + /** + * UObject RTTI boilerplate. + */ + virtual UClassID getDynamicClassID() const override; + +}; + +/* + ****************************************************************** + */ + +/** + *

A StringPair holds a displayName/ID pair. ICUService uses it + * as the array elements returned by getDisplayNames. + */ +class U_COMMON_API StringPair : public UMemory { +public: + /** + *

The display name of the pair.

+ */ + const UnicodeString displayName; + + /** + *

The ID of the pair.

+ */ + const UnicodeString id; + + /** + *

Creates a string pair from a displayName and an ID.

+ * + * @param displayName the displayName. + * @param id the ID. + * @param status the error code status. + * @return a StringPair if the creation was successful, otherwise nullptr. + */ + static StringPair* create(const UnicodeString& displayName, + const UnicodeString& id, + UErrorCode& status); + + /** + *

Return true if either string of the pair is bogus.

+ * @return true if either string of the pair is bogus. + */ + UBool isBogus() const; + +private: + StringPair(const UnicodeString& displayName, const UnicodeString& id); +}; + +/******************************************************************* + * ICUService + */ + + /** + *

A Service provides access to service objects that implement a + * particular service, e.g. transliterators. Users provide a String + * id (for example, a locale string) to the service, and get back an + * object for that id. Service objects can be any kind of object. A + * new service object is returned for each query. The caller is + * responsible for deleting it.

+ * + *

Services 'canonicalize' the query ID and use the canonical ID to + * query for the service. The service also defines a mechanism to + * 'fallback' the ID multiple times. Clients can optionally request + * the actual ID that was matched by a query when they use an ID to + * retrieve a service object.

+ * + *

Service objects are instantiated by ICUServiceFactory objects + * registered with the service. The service queries each + * ICUServiceFactory in turn, from most recently registered to + * earliest registered, until one returns a service object. If none + * responds with a service object, a fallback ID is generated, and the + * process repeats until a service object is returned or until the ID + * has no further fallbacks.

+ * + *

In ICU 2.4, UObject (the base class of service instances) does + * not define a polymorphic clone function. ICUService uses clones to + * manage ownership. Thus, for now, ICUService defines an abstract + * method, cloneInstance, that clients must implement to create clones + * of the service instances. This may change in future releases of + * ICU.

+ * + *

ICUServiceFactories can be dynamically registered and + * unregistered with the service. When registered, an + * ICUServiceFactory is installed at the head of the factory list, and + * so gets 'first crack' at any keys or fallback keys. When + * unregistered, it is removed from the service and can no longer be + * located through it. Service objects generated by this factory and + * held by the client are unaffected.

+ * + *

If a service has variants (e.g., the different variants of + * BreakIterator) an ICUServiceFactory can use the prefix of the + * ICUServiceKey to determine the variant of a service to generate. + * If it does not support all variants, it can request + * previously-registered factories to handle the ones it does not + * support.

+ * + *

ICUService uses ICUServiceKeys to query factories and perform + * fallback. The ICUServiceKey defines the canonical form of the ID, + * and implements the fallback strategy. Custom ICUServiceKeys can be + * defined that parse complex IDs into components that + * ICUServiceFactories can more easily use. The ICUServiceKey can + * cache the results of this parsing to save repeated effort. + * ICUService provides convenience APIs that take UnicodeStrings and + * generate default ICUServiceKeys for use in querying.

+ * + *

ICUService provides API to get the list of IDs publicly + * supported by the service (although queries aren't restricted to + * this list). This list contains only 'simple' IDs, and not fully + * unique IDs. ICUServiceFactories are associated with each simple ID + * and the responsible factory can also return a human-readable + * localized version of the simple ID, for use in user interfaces. + * ICUService can also provide an array of the all the localized + * visible IDs and their corresponding internal IDs.

+ * + *

ICUService implements ICUNotifier, so that clients can register + * to receive notification when factories are added or removed from + * the service. ICUService provides a default EventListener + * subinterface, ServiceListener, which can be registered with the + * service. When the service changes, the ServiceListener's + * serviceChanged method is called with the service as the + * argument.

+ * + *

The ICUService API is both rich and generic, and it is expected + * that most implementations will statically 'wrap' ICUService to + * present a more appropriate API-- for example, to declare the type + * of the objects returned from get, to limit the factories that can + * be registered with the service, or to define their own listener + * interface with a custom callback method. They might also customize + * ICUService by overriding it, for example, to customize the + * ICUServiceKey and fallback strategy. ICULocaleService is a + * subclass of ICUService that uses Locale names as IDs and uses + * ICUServiceKeys that implement the standard resource bundle fallback + * strategy. Most clients will wish to subclass it instead of + * ICUService.

+ */ +class U_COMMON_API ICUService : public ICUNotifier { + protected: + /** + * Name useful for debugging. + */ + const UnicodeString name; + + private: + + /** + * Timestamp so iterators can be fail-fast. + */ + uint32_t timestamp; + + /** + * All the factories registered with this service. + */ + UVector* factories; + + /** + * The service cache. + */ + Hashtable* serviceCache; + + /** + * The ID cache. + */ + Hashtable* idCache; + + /** + * The name cache. + */ + DNCache* dnCache; + + /** + * Constructor. + */ + public: + /** + *

Construct a new ICUService.

+ */ + ICUService(); + + /** + *

Construct with a name (useful for debugging).

+ * + * @param name a name to use in debugging. + */ + ICUService(const UnicodeString& name); + + /** + *

Destructor.

+ */ + virtual ~ICUService(); + + /** + *

Return the name of this service. This will be the empty string if none was assigned. + * Returns result as a convenience.

+ * + * @param result an output parameter to contain the name of this service. + * @return the name of this service. + */ + UnicodeString& getName(UnicodeString& result) const; + + /** + *

Convenience override for get(ICUServiceKey&, UnicodeString*). This uses + * createKey to create a key for the provided descriptor.

+ * + * @param descriptor the descriptor. + * @param status the error code status. + * @return the service instance, or nullptr. + */ + UObject* get(const UnicodeString& descriptor, UErrorCode& status) const; + + /** + *

Convenience override for get(ICUServiceKey&, UnicodeString*). This uses + * createKey to create a key from the provided descriptor.

+ * + * @param descriptor the descriptor. + * @param actualReturn a pointer to a UnicodeString to hold the matched descriptor, or nullptr. + * @param status the error code status. + * @return the service instance, or nullptr. + */ + UObject* get(const UnicodeString& descriptor, UnicodeString* actualReturn, UErrorCode& status) const; + + /** + *

Convenience override for get(ICUServiceKey&, UnicodeString*).

+ * + * @param key the key. + * @param status the error code status. + * @return the service instance, or nullptr. + */ + UObject* getKey(ICUServiceKey& key, UErrorCode& status) const; + + /** + *

Given a key, return a service object, and, if actualReturn + * is not nullptr, the descriptor with which it was found in the + * first element of actualReturn. If no service object matches + * this key, returns nullptr and leaves actualReturn unchanged.

+ * + *

This queries the cache using the key's descriptor, and if no + * object in the cache matches, tries the key on each + * registered factory, in order. If none generates a service + * object for the key, repeats the process with each fallback of + * the key, until either a factory returns a service object, or the key + * has no fallback. If no object is found, the result of handleDefault + * is returned.

+ * + *

Subclasses can override this method to further customize the + * result before returning it. + * + * @param key the key. + * @param actualReturn a pointer to a UnicodeString to hold the matched descriptor, or nullptr. + * @param status the error code status. + * @return the service instance, or nullptr. + */ + virtual UObject* getKey(ICUServiceKey& key, UnicodeString* actualReturn, UErrorCode& status) const; + + /** + *

This version of getKey is only called by ICUServiceFactories within the scope + * of a previous getKey call, to determine what previously-registered factories would + * have returned. For details, see getKey(ICUServiceKey&, UErrorCode&). Subclasses + * should not call it directly, but call through one of the other get functions.

+ * + * @param key the key. + * @param actualReturn a pointer to a UnicodeString to hold the matched descriptor, or nullptr. + * @param factory the factory making the recursive call. + * @param status the error code status. + * @return the service instance, or nullptr. + */ + UObject* getKey(ICUServiceKey& key, UnicodeString* actualReturn, const ICUServiceFactory* factory, UErrorCode& status) const; + + /** + *

Convenience override for getVisibleIDs(String) that passes null + * as the fallback, thus returning all visible IDs.

+ * + * @param result a vector to hold the returned IDs. + * @param status the error code status. + * @return the result vector. + */ + UVector& getVisibleIDs(UVector& result, UErrorCode& status) const; + + /** + *

Return a snapshot of the visible IDs for this service. This + * list will not change as ICUServiceFactories are added or removed, but the + * supported IDs will, so there is no guarantee that all and only + * the IDs in the returned list will be visible and supported by the + * service in subsequent calls.

+ * + *

The IDs are returned as pointers to UnicodeStrings. The + * caller owns the IDs. Previous contents of result are discarded before + * new elements, if any, are added.

+ * + *

matchID is passed to createKey to create a key. If the key + * is not nullptr, its isFallbackOf method is used to filter out IDs + * that don't match the key or have it as a fallback.

+ * + * @param result a vector to hold the returned IDs. + * @param matchID an ID used to filter the result, or nullptr if all IDs are desired. + * @param status the error code status. + * @return the result vector. + */ + UVector& getVisibleIDs(UVector& result, const UnicodeString* matchID, UErrorCode& status) const; + + /** + *

Convenience override for getDisplayName(const UnicodeString&, const Locale&, UnicodeString&) that + * uses the current default locale.

+ * + * @param id the ID for which to retrieve the localized displayName. + * @param result an output parameter to hold the display name. + * @return the modified result. + */ + UnicodeString& getDisplayName(const UnicodeString& id, UnicodeString& result) const; + + /** + *

Given a visible ID, return the display name in the requested locale. + * If there is no directly supported ID corresponding to this ID, result is + * set to bogus.

+ * + * @param id the ID for which to retrieve the localized displayName. + * @param result an output parameter to hold the display name. + * @param locale the locale in which to localize the ID. + * @return the modified result. + */ + UnicodeString& getDisplayName(const UnicodeString& id, UnicodeString& result, const Locale& locale) const; + + /** + *

Convenience override of getDisplayNames(const Locale&, const UnicodeString*) that + * uses the current default Locale as the locale and nullptr for + * the matchID.

+ * + * @param result a vector to hold the returned displayName/id StringPairs. + * @param status the error code status. + * @return the modified result vector. + */ + UVector& getDisplayNames(UVector& result, UErrorCode& status) const; + + /** + *

Convenience override of getDisplayNames(const Locale&, const UnicodeString*) that + * uses nullptr for the matchID.

+ * + * @param result a vector to hold the returned displayName/id StringPairs. + * @param locale the locale in which to localize the ID. + * @param status the error code status. + * @return the modified result vector. + */ + UVector& getDisplayNames(UVector& result, const Locale& locale, UErrorCode& status) const; + + /** + *

Return a snapshot of the mapping from display names to visible + * IDs for this service. This set will not change as factories + * are added or removed, but the supported IDs will, so there is + * no guarantee that all and only the IDs in the returned map will + * be visible and supported by the service in subsequent calls, + * nor is there any guarantee that the current display names match + * those in the result.

+ * + *

The names are returned as pointers to StringPairs, which + * contain both the displayName and the corresponding ID. The + * caller owns the StringPairs. Previous contents of result are + * discarded before new elements, if any, are added.

+ * + *

matchID is passed to createKey to create a key. If the key + * is not nullptr, its isFallbackOf method is used to filter out IDs + * that don't match the key or have it as a fallback.

+ * + * @param result a vector to hold the returned displayName/id StringPairs. + * @param locale the locale in which to localize the ID. + * @param matchID an ID used to filter the result, or nullptr if all IDs are desired. + * @param status the error code status. + * @return the result vector. */ + UVector& getDisplayNames(UVector& result, + const Locale& locale, + const UnicodeString* matchID, + UErrorCode& status) const; + + /** + *

A convenience override of registerInstance(UObject*, const UnicodeString&, UBool) + * that defaults visible to true.

+ * + * @param objToAdopt the object to register and adopt. + * @param id the ID to assign to this object. + * @param status the error code status. + * @return a registry key that can be passed to unregister to unregister + * (and discard) this instance. + */ + URegistryKey registerInstance(UObject* objToAdopt, const UnicodeString& id, UErrorCode& status); + + /** + *

Register a service instance with the provided ID. The ID will be + * canonicalized. The canonicalized ID will be returned by + * getVisibleIDs if visible is true. The service instance will be adopted and + * must not be modified subsequent to this call.

+ * + *

This issues a serviceChanged notification to registered listeners.

+ * + *

This implementation wraps the object using + * createSimpleFactory, and calls registerFactory.

+ * + * @param objToAdopt the object to register and adopt. + * @param id the ID to assign to this object. + * @param visible true if getVisibleIDs is to return this ID. + * @param status the error code status. + * @return a registry key that can be passed to unregister() to unregister + * (and discard) this instance. + */ + virtual URegistryKey registerInstance(UObject* objToAdopt, const UnicodeString& id, UBool visible, UErrorCode& status); + + /** + *

Register an ICUServiceFactory. Returns a registry key that + * can be used to unregister the factory. The factory + * must not be modified subsequent to this call. The service owns + * all registered factories. In case of an error, the factory is + * deleted.

+ * + *

This issues a serviceChanged notification to registered listeners.

+ * + *

The default implementation accepts all factories.

+ * + * @param factoryToAdopt the factory to register and adopt. + * @param status the error code status. + * @return a registry key that can be passed to unregister to unregister + * (and discard) this factory. + */ + virtual URegistryKey registerFactory(ICUServiceFactory* factoryToAdopt, UErrorCode& status); + + /** + *

Unregister a factory using a registry key returned by + * registerInstance or registerFactory. After a successful call, + * the factory will be removed from the service factory list and + * deleted, and the key becomes invalid.

+ * + *

This issues a serviceChanged notification to registered + * listeners.

+ * + * @param rkey the registry key. + * @param status the error code status. + * @return true if the call successfully unregistered the factory. + */ + virtual UBool unregister(URegistryKey rkey, UErrorCode& status); + + /** + *

Reset the service to the default factories. The factory + * lock is acquired and then reInitializeFactories is called.

+ * + *

This issues a serviceChanged notification to registered listeners.

+ */ + virtual void reset(); + + /** + *

Return true if the service is in its default state.

+ * + *

The default implementation returns true if there are no + * factories registered.

+ */ + virtual UBool isDefault() const; + + /** + *

Create a key from an ID. If ID is nullptr, returns nullptr.

+ * + *

The default implementation creates an ICUServiceKey instance. + * Subclasses can override to define more useful keys appropriate + * to the factories they accept.

+ * + * @param a pointer to the ID for which to create a default ICUServiceKey. + * @param status the error code status. + * @return the ICUServiceKey corresponding to ID, or nullptr. + */ + virtual ICUServiceKey* createKey(const UnicodeString* id, UErrorCode& status) const; + + /** + *

Clone object so that caller can own the copy. In ICU2.4, UObject doesn't define + * clone, so we need an instance-aware method that knows how to do this. + * This is public so factories can call it, but should really be protected.

+ * + * @param instance the service instance to clone. + * @return a clone of the passed-in instance, or nullptr if cloning was unsuccessful. + */ + virtual UObject* cloneInstance(UObject* instance) const = 0; + + + /************************************************************************ + * Subclassing API + */ + + protected: + + /** + *

Create a factory that wraps a single service object. Called by registerInstance.

+ * + *

The default implementation returns an instance of SimpleFactory.

+ * + * @param instanceToAdopt the service instance to adopt. + * @param id the ID to assign to this service instance. + * @param visible if true, the ID will be visible. + * @param status the error code status. + * @return an instance of ICUServiceFactory that maps this instance to the provided ID. + */ + virtual ICUServiceFactory* createSimpleFactory(UObject* instanceToAdopt, const UnicodeString& id, UBool visible, UErrorCode& status); + + /** + *

Reinitialize the factory list to its default state. After this call, isDefault() + * must return true.

+ * + *

This issues a serviceChanged notification to registered listeners.

+ * + *

The default implementation clears the factory list. + * Subclasses can override to provide other default initialization + * of the factory list. Subclasses must not call this method + * directly, since it must only be called while holding write + * access to the factory list.

+ */ + virtual void reInitializeFactories(); + + /** + *

Default handler for this service if no factory in the factory list + * handled the key passed to getKey.

+ * + *

The default implementation returns nullptr.

+ * + * @param key the key. + * @param actualReturn a pointer to a UnicodeString to hold the matched descriptor, or nullptr. + * @param status the error code status. + * @return the service instance, or nullptr. + */ + virtual UObject* handleDefault(const ICUServiceKey& key, UnicodeString* actualReturn, UErrorCode& status) const; + + /** + *

Clear caches maintained by this service.

+ * + *

Subclasses can override if they implement additional caches + * that need to be cleared when the service changes. Subclasses + * should generally not call this method directly, as it must only + * be called while synchronized on the factory lock.

+ */ + virtual void clearCaches(); + + /** + *

Return true if the listener is accepted.

+ * + *

The default implementation accepts the listener if it is + * a ServiceListener. Subclasses can override this to accept + * different listeners.

+ * + * @param l the listener to test. + * @return true if the service accepts the listener. + */ + virtual UBool acceptsListener(const EventListener& l) const override; + + /** + *

Notify the listener of a service change.

+ * + *

The default implementation assumes a ServiceListener. + * If acceptsListener has been overridden to accept different + * listeners, this should be overridden as well.

+ * + * @param l the listener to notify. + */ + virtual void notifyListener(EventListener& l) const override; + + /************************************************************************ + * Utilities for subclasses. + */ + + /** + *

Clear only the service cache.

+ * + *

This can be called by subclasses when a change affects the service + * cache but not the ID caches, e.g., when the default locale changes + * the resolution of IDs also changes, requiring the cache to be + * flushed, but not the visible IDs themselves.

+ */ + void clearServiceCache(); + + /** + *

Return a map from visible IDs to factories. + * This must only be called when the mutex is held.

+ * + * @param status the error code status. + * @return a Hashtable containing mappings from visible + * IDs to factories. + */ + const Hashtable* getVisibleIDMap(UErrorCode& status) const; + + /** + *

Allow subclasses to read the time stamp.

+ * + * @return the timestamp. + */ + int32_t getTimestamp() const; + + /** + *

Return the number of registered factories.

+ * + * @return the number of factories registered at the time of the call. + */ + int32_t countFactories() const; + +private: + + friend class ::ICUServiceTest; // give tests access to countFactories. +}; + +U_NAMESPACE_END + + /* UCONFIG_NO_SERVICE */ +#endif + + /* ICUSERV_H */ +#endif + diff --git a/intl/icu/source/common/servlk.cpp b/intl/icu/source/common/servlk.cpp new file mode 100644 index 0000000000..ba275ea301 --- /dev/null +++ b/intl/icu/source/common/servlk.cpp @@ -0,0 +1,188 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** + ******************************************************************************* + * Copyright (C) 2001-2014, International Business Machines Corporation and * + * others. All Rights Reserved. * + ******************************************************************************* + * + ******************************************************************************* + */ +#include "unicode/utypes.h" + +#if !UCONFIG_NO_SERVICE + +#include "unicode/resbund.h" +#include "uresimp.h" +#include "cmemory.h" +#include "servloc.h" +#include "ustrfmt.h" +#include "uhash.h" +#include "charstr.h" +#include "uassert.h" + +#define UNDERSCORE_CHAR ((char16_t)0x005f) +#define AT_SIGN_CHAR ((char16_t)64) +#define PERIOD_CHAR ((char16_t)46) + +U_NAMESPACE_BEGIN + +LocaleKey* +LocaleKey::createWithCanonicalFallback(const UnicodeString* primaryID, + const UnicodeString* canonicalFallbackID, + UErrorCode& status) +{ + return LocaleKey::createWithCanonicalFallback(primaryID, canonicalFallbackID, KIND_ANY, status); +} + +LocaleKey* +LocaleKey::createWithCanonicalFallback(const UnicodeString* primaryID, + const UnicodeString* canonicalFallbackID, + int32_t kind, + UErrorCode& status) +{ + if (primaryID == nullptr || U_FAILURE(status)) { + return nullptr; + } + UnicodeString canonicalPrimaryID; + LocaleUtility::canonicalLocaleString(primaryID, canonicalPrimaryID); + return new LocaleKey(*primaryID, canonicalPrimaryID, canonicalFallbackID, kind); +} + +LocaleKey::LocaleKey(const UnicodeString& primaryID, + const UnicodeString& canonicalPrimaryID, + const UnicodeString* canonicalFallbackID, + int32_t kind) + : ICUServiceKey(primaryID) + , _kind(kind) + , _primaryID(canonicalPrimaryID) + , _fallbackID() + , _currentID() +{ + _fallbackID.setToBogus(); + if (_primaryID.length() != 0) { + if (canonicalFallbackID != nullptr && _primaryID != *canonicalFallbackID) { + _fallbackID = *canonicalFallbackID; + } + } + + _currentID = _primaryID; +} + +LocaleKey::~LocaleKey() {} + +UnicodeString& +LocaleKey::prefix(UnicodeString& result) const { + if (_kind != KIND_ANY) { + char16_t buffer[64]; + uprv_itou(buffer, 64, _kind, 10, 0); + UnicodeString temp(buffer); + result.append(temp); + } + return result; +} + +int32_t +LocaleKey::kind() const { + return _kind; +} + +UnicodeString& +LocaleKey::canonicalID(UnicodeString& result) const { + return result.append(_primaryID); +} + +UnicodeString& +LocaleKey::currentID(UnicodeString& result) const { + if (!_currentID.isBogus()) { + result.append(_currentID); + } + return result; +} + +UnicodeString& +LocaleKey::currentDescriptor(UnicodeString& result) const { + if (!_currentID.isBogus()) { + prefix(result).append(PREFIX_DELIMITER).append(_currentID); + } else { + result.setToBogus(); + } + return result; +} + +Locale& +LocaleKey::canonicalLocale(Locale& result) const { + return LocaleUtility::initLocaleFromName(_primaryID, result); +} + +Locale& +LocaleKey::currentLocale(Locale& result) const { + return LocaleUtility::initLocaleFromName(_currentID, result); +} + +UBool +LocaleKey::fallback() { + if (!_currentID.isBogus()) { + int x = _currentID.lastIndexOf(UNDERSCORE_CHAR); + if (x != -1) { + _currentID.remove(x); // truncate current or fallback, whichever we're pointing to + return true; + } + + if (!_fallbackID.isBogus()) { + _currentID = _fallbackID; + _fallbackID.setToBogus(); + return true; + } + + if (_currentID.length() > 0) { + _currentID.remove(0); // completely truncate + return true; + } + + _currentID.setToBogus(); + } + + return false; +} + +UBool +LocaleKey::isFallbackOf(const UnicodeString& id) const { + UnicodeString temp(id); + parseSuffix(temp); + return temp.indexOf(_primaryID) == 0 && + (temp.length() == _primaryID.length() || + temp.charAt(_primaryID.length()) == UNDERSCORE_CHAR); +} + +#ifdef SERVICE_DEBUG +UnicodeString& +LocaleKey::debug(UnicodeString& result) const +{ + ICUServiceKey::debug(result); + result.append((UnicodeString)" kind: "); + result.append(_kind); + result.append((UnicodeString)" primaryID: "); + result.append(_primaryID); + result.append((UnicodeString)" fallbackID: "); + result.append(_fallbackID); + result.append((UnicodeString)" currentID: "); + result.append(_currentID); + return result; +} + +UnicodeString& +LocaleKey::debugClass(UnicodeString& result) const +{ + return result.append((UnicodeString)"LocaleKey "); +} +#endif + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(LocaleKey) + +U_NAMESPACE_END + +/* !UCONFIG_NO_SERVICE */ +#endif + + diff --git a/intl/icu/source/common/servlkf.cpp b/intl/icu/source/common/servlkf.cpp new file mode 100644 index 0000000000..f5bb0418c4 --- /dev/null +++ b/intl/icu/source/common/servlkf.cpp @@ -0,0 +1,152 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** + ******************************************************************************* + * Copyright (C) 2001-2014, International Business Machines Corporation and * + * others. All Rights Reserved. * + ******************************************************************************* + * + ******************************************************************************* + */ +#include "unicode/utypes.h" + +#if !UCONFIG_NO_SERVICE + +#include "unicode/resbund.h" +#include "uresimp.h" +#include "cmemory.h" +#include "servloc.h" +#include "ustrfmt.h" +#include "uhash.h" +#include "charstr.h" +#include "ucln_cmn.h" +#include "uassert.h" + +#define UNDERSCORE_CHAR ((char16_t)0x005f) +#define AT_SIGN_CHAR ((char16_t)64) +#define PERIOD_CHAR ((char16_t)46) + + +U_NAMESPACE_BEGIN + +LocaleKeyFactory::LocaleKeyFactory(int32_t coverage) + : _name() + , _coverage(coverage) +{ +} + +LocaleKeyFactory::LocaleKeyFactory(int32_t coverage, const UnicodeString& name) + : _name(name) + , _coverage(coverage) +{ +} + +LocaleKeyFactory::~LocaleKeyFactory() { +} + +UObject* +LocaleKeyFactory::create(const ICUServiceKey& key, const ICUService* service, UErrorCode& status) const { + if (handlesKey(key, status)) { + const LocaleKey& lkey = static_cast(key); + int32_t kind = lkey.kind(); + Locale loc; + lkey.currentLocale(loc); + + return handleCreate(loc, kind, service, status); + } + return nullptr; +} + +UBool +LocaleKeyFactory::handlesKey(const ICUServiceKey& key, UErrorCode& status) const { + const Hashtable* supported = getSupportedIDs(status); + if (supported) { + UnicodeString id; + key.currentID(id); + return supported->get(id) != nullptr; + } + return false; +} + +void +LocaleKeyFactory::updateVisibleIDs(Hashtable& result, UErrorCode& status) const { + const Hashtable* supported = getSupportedIDs(status); + if (supported) { + UBool visible = (_coverage & 0x1) == 0; + const UHashElement* elem = nullptr; + int32_t pos = UHASH_FIRST; + while ((elem = supported->nextElement(pos)) != nullptr) { + const UnicodeString& id = *((const UnicodeString*)elem->key.pointer); + if (!visible) { + result.remove(id); + } else { + result.put(id, (void*)this, status); // this is dummy non-void marker used for set semantics + if (U_FAILURE(status)) { + break; + } + } + } + } +} + +UnicodeString& +LocaleKeyFactory::getDisplayName(const UnicodeString& id, const Locale& locale, UnicodeString& result) const { + if ((_coverage & 0x1) == 0) { + //UErrorCode status = U_ZERO_ERROR; + // assume if this is called on us, we support some fallback of this id + // if (isSupportedID(id, status)) { + Locale loc; + LocaleUtility::initLocaleFromName(id, loc); + return loc.getDisplayName(locale, result); + // } + } + result.setToBogus(); + return result; +} + +UObject* +LocaleKeyFactory::handleCreate(const Locale& /* loc */, + int32_t /* kind */, + const ICUService* /* service */, + UErrorCode& /* status */) const { + return nullptr; +} + +//UBool +//LocaleKeyFactory::isSupportedID(const UnicodeString& id, UErrorCode& status) const { +// const Hashtable* ids = getSupportedIDs(status); +// return ids && ids->get(id); +//} + +const Hashtable* +LocaleKeyFactory::getSupportedIDs(UErrorCode& /* status */) const { + return nullptr; +} + +#ifdef SERVICE_DEBUG +UnicodeString& +LocaleKeyFactory::debug(UnicodeString& result) const +{ + debugClass(result); + result.append((UnicodeString)", name: "); + result.append(_name); + result.append((UnicodeString)", coverage: "); + result.append(_coverage); + return result; +} + +UnicodeString& +LocaleKeyFactory::debugClass(UnicodeString& result) const +{ + return result.append((UnicodeString)"LocaleKeyFactory"); +} +#endif + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(LocaleKeyFactory) + +U_NAMESPACE_END + +/* !UCONFIG_NO_SERVICE */ +#endif + + diff --git a/intl/icu/source/common/servloc.h b/intl/icu/source/common/servloc.h new file mode 100644 index 0000000000..49d1240905 --- /dev/null +++ b/intl/icu/source/common/servloc.h @@ -0,0 +1,551 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** + ******************************************************************************* + * Copyright (C) 2001-2011, International Business Machines Corporation and * + * others. All Rights Reserved. * + ******************************************************************************* + * + ******************************************************************************* + */ +#ifndef ICULSERV_H +#define ICULSERV_H + +#include "unicode/utypes.h" + +#if UCONFIG_NO_SERVICE + +U_NAMESPACE_BEGIN + +/* + * Allow the declaration of APIs with pointers to ICUService + * even when service is removed from the build. + */ +class ICULocaleService; + +U_NAMESPACE_END + +#else + +#include "unicode/unistr.h" +#include "unicode/locid.h" +#include "unicode/strenum.h" + +#include "hash.h" +#include "uvector.h" + +#include "serv.h" +#include "locutil.h" + +U_NAMESPACE_BEGIN + +class ICULocaleService; + +class LocaleKey; +class LocaleKeyFactory; +class SimpleLocaleKeyFactory; +class ServiceListener; + +/* + ****************************************************************** + */ + +/** + * A subclass of Key that implements a locale fallback mechanism. + * The first locale to search for is the locale provided by the + * client, and the fallback locale to search for is the current + * default locale. If a prefix is present, the currentDescriptor + * includes it before the locale proper, separated by "/". This + * is the default key instantiated by ICULocaleService.

+ * + *

Canonicalization adjusts the locale string so that the + * section before the first understore is in lower case, and the rest + * is in upper case, with no trailing underscores.

+ */ + +class U_COMMON_API LocaleKey : public ICUServiceKey { + private: + int32_t _kind; + UnicodeString _primaryID; + UnicodeString _fallbackID; + UnicodeString _currentID; + + public: + enum { + KIND_ANY = -1 + }; + + /** + * Create a LocaleKey with canonical primary and fallback IDs. + */ + static LocaleKey* createWithCanonicalFallback(const UnicodeString* primaryID, + const UnicodeString* canonicalFallbackID, + UErrorCode& status); + + /** + * Create a LocaleKey with canonical primary and fallback IDs. + */ + static LocaleKey* createWithCanonicalFallback(const UnicodeString* primaryID, + const UnicodeString* canonicalFallbackID, + int32_t kind, + UErrorCode& status); + + protected: + /** + * PrimaryID is the user's requested locale string, + * canonicalPrimaryID is this string in canonical form, + * fallbackID is the current default locale's string in + * canonical form. + */ + LocaleKey(const UnicodeString& primaryID, + const UnicodeString& canonicalPrimaryID, + const UnicodeString* canonicalFallbackID, + int32_t kind); + + public: + /** + * Append the prefix associated with the kind, or nothing if the kind is KIND_ANY. + */ + virtual UnicodeString& prefix(UnicodeString& result) const override; + + /** + * Return the kind code associated with this key. + */ + virtual int32_t kind() const; + + /** + * Return the canonicalID. + */ + virtual UnicodeString& canonicalID(UnicodeString& result) const override; + + /** + * Return the currentID. + */ + virtual UnicodeString& currentID(UnicodeString& result) const override; + + /** + * Return the (canonical) current descriptor, or null if no current id. + */ + virtual UnicodeString& currentDescriptor(UnicodeString& result) const override; + + /** + * Convenience method to return the locale corresponding to the (canonical) original ID. + */ + virtual Locale& canonicalLocale(Locale& result) const; + + /** + * Convenience method to return the locale corresponding to the (canonical) current ID. + */ + virtual Locale& currentLocale(Locale& result) const; + + /** + * If the key has a fallback, modify the key and return true, + * otherwise return false.

+ * + *

First falls back through the primary ID, then through + * the fallbackID. The final fallback is the empty string, + * unless the primary id was the empty string, in which case + * there is no fallback. + */ + virtual UBool fallback() override; + + /** + * Return true if a key created from id matches, or would eventually + * fallback to match, the canonical ID of this key. + */ + virtual UBool isFallbackOf(const UnicodeString& id) const override; + + public: + /** + * UObject boilerplate. + */ + static UClassID U_EXPORT2 getStaticClassID(); + + virtual UClassID getDynamicClassID() const override; + + /** + * Destructor. + */ + virtual ~LocaleKey(); + +#ifdef SERVICE_DEBUG + public: + virtual UnicodeString& debug(UnicodeString& result) const; + virtual UnicodeString& debugClass(UnicodeString& result) const; +#endif + +}; + +/* + ****************************************************************** + */ + +/** + * A subclass of ICUServiceFactory that uses LocaleKeys, and is able to + * 'cover' more specific locales with more general locales that it + * supports. + * + *

Coverage may be either of the values VISIBLE or INVISIBLE. + * + *

'Visible' indicates that the specific locale(s) supported by + * the factory are registered in getSupportedIDs, 'Invisible' + * indicates that they are not. + * + *

Localization of visible ids is handled + * by the handling factory, regardless of kind. + */ +class U_COMMON_API LocaleKeyFactory : public ICUServiceFactory { +protected: + const UnicodeString _name; + const int32_t _coverage; + +public: + enum { + /** + * Coverage value indicating that the factory makes + * its locales visible, and does not cover more specific + * locales. + */ + VISIBLE = 0, + + /** + * Coverage value indicating that the factory does not make + * its locales visible, and does not cover more specific + * locales. + */ + INVISIBLE = 1 + }; + + /** + * Destructor. + */ + virtual ~LocaleKeyFactory(); + +protected: + /** + * Constructor used by subclasses. + */ + LocaleKeyFactory(int32_t coverage); + + /** + * Constructor used by subclasses. + */ + LocaleKeyFactory(int32_t coverage, const UnicodeString& name); + + /** + * Implement superclass abstract method. This checks the currentID of + * the key against the supported IDs, and passes the canonicalLocale and + * kind off to handleCreate (which subclasses must implement). + */ +public: + virtual UObject* create(const ICUServiceKey& key, const ICUService* service, UErrorCode& status) const override; + +protected: + virtual UBool handlesKey(const ICUServiceKey& key, UErrorCode& status) const; + +public: + /** + * Override of superclass method. This adjusts the result based + * on the coverage rule for this factory. + */ + virtual void updateVisibleIDs(Hashtable& result, UErrorCode& status) const override; + + /** + * Return a localized name for the locale represented by id. + */ + virtual UnicodeString& getDisplayName(const UnicodeString& id, const Locale& locale, UnicodeString& result) const override; + +protected: + /** + * Utility method used by create(ICUServiceKey, ICUService). Subclasses can implement + * this instead of create. The default returns nullptr. + */ + virtual UObject* handleCreate(const Locale& loc, int32_t kind, const ICUService* service, UErrorCode& status) const; + + /** + * Return true if this id is one the factory supports (visible or + * otherwise). + */ + // virtual UBool isSupportedID(const UnicodeString& id, UErrorCode& status) const; + + /** + * Return the set of ids that this factory supports (visible or + * otherwise). This can be called often and might need to be + * cached if it is expensive to create. + */ + virtual const Hashtable* getSupportedIDs(UErrorCode& status) const; + +public: + /** + * UObject boilerplate. + */ + static UClassID U_EXPORT2 getStaticClassID(); + + virtual UClassID getDynamicClassID() const override; + +#ifdef SERVICE_DEBUG + public: + virtual UnicodeString& debug(UnicodeString& result) const; + virtual UnicodeString& debugClass(UnicodeString& result) const; +#endif + +}; + +/* + ****************************************************************** + */ + +/** + * A LocaleKeyFactory that just returns a single object for a kind/locale. + */ + +class U_COMMON_API SimpleLocaleKeyFactory : public LocaleKeyFactory { + private: + UObject* _obj; + UnicodeString _id; + const int32_t _kind; + + public: + SimpleLocaleKeyFactory(UObject* objToAdopt, + const UnicodeString& locale, + int32_t kind, + int32_t coverage); + + SimpleLocaleKeyFactory(UObject* objToAdopt, + const Locale& locale, + int32_t kind, + int32_t coverage); + + /** + * Destructor. + */ + virtual ~SimpleLocaleKeyFactory(); + + /** + * Override of superclass method. Returns the service object if kind/locale match. Service is not used. + */ + virtual UObject* create(const ICUServiceKey& key, const ICUService* service, UErrorCode& status) const override; + + /** + * Override of superclass method. This adjusts the result based + * on the coverage rule for this factory. + */ + virtual void updateVisibleIDs(Hashtable& result, UErrorCode& status) const override; + + protected: + /** + * Return true if this id is equal to the locale name. + */ + //virtual UBool isSupportedID(const UnicodeString& id, UErrorCode& status) const; + + +public: + /** + * UObject boilerplate. + */ + static UClassID U_EXPORT2 getStaticClassID(); + + virtual UClassID getDynamicClassID() const override; + +#ifdef SERVICE_DEBUG + public: + virtual UnicodeString& debug(UnicodeString& result) const; + virtual UnicodeString& debugClass(UnicodeString& result) const; +#endif + +}; + +/* + ****************************************************************** + */ + +/** + * A LocaleKeyFactory that creates a service based on the ICU locale data. + * This is a base class for most ICU factories. Subclasses instantiate it + * with a constructor that takes a bundle name, which determines the supported + * IDs. Subclasses then override handleCreate to create the actual service + * object. The default implementation returns a resource bundle. + */ +class U_COMMON_API ICUResourceBundleFactory : public LocaleKeyFactory +{ + protected: + UnicodeString _bundleName; + + public: + /** + * Convenience constructor that uses the main ICU bundle name. + */ + ICUResourceBundleFactory(); + + /** + * A service factory based on ICU resource data in resources with + * the given name. This should be a 'path' that can be passed to + * ures_openAvailableLocales, such as U_ICUDATA or U_ICUDATA_COLL. + * The empty string is equivalent to U_ICUDATA. + */ + ICUResourceBundleFactory(const UnicodeString& bundleName); + + /** + * Destructor + */ + virtual ~ICUResourceBundleFactory(); + +protected: + /** + * Return the supported IDs. This is the set of all locale names in ICULocaleData. + */ + virtual const Hashtable* getSupportedIDs(UErrorCode& status) const override; + + /** + * Create the service. The default implementation returns the resource bundle + * for the locale, ignoring kind, and service. + */ + virtual UObject* handleCreate(const Locale& loc, int32_t kind, const ICUService* service, UErrorCode& status) const override; + +public: + /** + * UObject boilerplate. + */ + static UClassID U_EXPORT2 getStaticClassID(); + virtual UClassID getDynamicClassID() const override; + + +#ifdef SERVICE_DEBUG + public: + virtual UnicodeString& debug(UnicodeString& result) const; + virtual UnicodeString& debugClass(UnicodeString& result) const; +#endif + +}; + +/* + ****************************************************************** + */ + +class U_COMMON_API ICULocaleService : public ICUService +{ + private: + Locale fallbackLocale; + UnicodeString fallbackLocaleName; + + public: + /** + * Construct an ICULocaleService. + */ + ICULocaleService(); + + /** + * Construct an ICULocaleService with a name (useful for debugging). + */ + ICULocaleService(const UnicodeString& name); + + /** + * Destructor. + */ + virtual ~ICULocaleService(); + +#if 0 + // redeclare because of overload resolution rules? + // no, causes ambiguities since both UnicodeString and Locale have constructors that take a const char* + // need some compiler flag to remove warnings + UObject* get(const UnicodeString& descriptor, UErrorCode& status) const { + return ICUService::get(descriptor, status); + } + + UObject* get(const UnicodeString& descriptor, UnicodeString* actualReturn, UErrorCode& status) const { + return ICUService::get(descriptor, actualReturn, status); + } +#endif + + /** + * Convenience override for callers using locales. This calls + * get(Locale, int, Locale[]) with KIND_ANY for kind and null for + * actualReturn. + */ + UObject* get(const Locale& locale, UErrorCode& status) const; + + /** + * Convenience override for callers using locales. This calls + * get(Locale, int, Locale[]) with a null actualReturn. + */ + UObject* get(const Locale& locale, int32_t kind, UErrorCode& status) const; + + /** + * Convenience override for callers using locales. This calls + * get(Locale, String, Locale[]) with a null kind. + */ + UObject* get(const Locale& locale, Locale* actualReturn, UErrorCode& status) const; + + /** + * Convenience override for callers using locales. This uses + * createKey(Locale.toString(), kind) to create a key, calls getKey, and then + * if actualReturn is not null, returns the actualResult from + * getKey (stripping any prefix) into a Locale. + */ + UObject* get(const Locale& locale, int32_t kind, Locale* actualReturn, UErrorCode& status) const; + + /** + * Convenience override for callers using locales. This calls + * registerObject(Object, Locale, int32_t kind, int coverage) + * passing KIND_ANY for the kind, and VISIBLE for the coverage. + */ + virtual URegistryKey registerInstance(UObject* objToAdopt, const Locale& locale, UErrorCode& status); + + /** + * Convenience function for callers using locales. This calls + * registerObject(Object, Locale, int kind, int coverage) + * passing VISIBLE for the coverage. + */ + virtual URegistryKey registerInstance(UObject* objToAdopt, const Locale& locale, int32_t kind, UErrorCode& status); + + /** + * Convenience function for callers using locales. This instantiates + * a SimpleLocaleKeyFactory, and registers the factory. + */ + virtual URegistryKey registerInstance(UObject* objToAdopt, const Locale& locale, int32_t kind, int32_t coverage, UErrorCode& status); + + + /** + * (Stop compiler from complaining about hidden overrides.) + * Since both UnicodeString and Locale have constructors that take const char*, adding a public + * method that takes UnicodeString causes ambiguity at call sites that use const char*. + * We really need a flag that is understood by all compilers that will suppress the warning about + * hidden overrides. + */ + virtual URegistryKey registerInstance(UObject* objToAdopt, const UnicodeString& locale, UBool visible, UErrorCode& status) override; + + /** + * Convenience method for callers using locales. This returns the standard + * service ID enumeration. + */ + virtual StringEnumeration* getAvailableLocales() const; + + protected: + + /** + * Return the name of the current fallback locale. If it has changed since this was + * last accessed, the service cache is cleared. + */ + const UnicodeString& validateFallbackLocale() const; + + /** + * Override superclass createKey method. + */ + virtual ICUServiceKey* createKey(const UnicodeString* id, UErrorCode& status) const override; + + /** + * Additional createKey that takes a kind. + */ + virtual ICUServiceKey* createKey(const UnicodeString* id, int32_t kind, UErrorCode& status) const; + + friend class ServiceEnumeration; +}; + +U_NAMESPACE_END + + /* UCONFIG_NO_SERVICE */ +#endif + + /* ICULSERV_H */ +#endif + diff --git a/intl/icu/source/common/servls.cpp b/intl/icu/source/common/servls.cpp new file mode 100644 index 0000000000..978a956df7 --- /dev/null +++ b/intl/icu/source/common/servls.cpp @@ -0,0 +1,296 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** + ******************************************************************************* + * Copyright (C) 2001-2014, International Business Machines Corporation and * + * others. All Rights Reserved. * + ******************************************************************************* + * + ******************************************************************************* + */ +#include "unicode/utypes.h" + +#if !UCONFIG_NO_SERVICE + +#include "unicode/resbund.h" +#include "uresimp.h" +#include "cmemory.h" +#include "servloc.h" +#include "ustrfmt.h" +#include "charstr.h" +#include "uassert.h" + +#define UNDERSCORE_CHAR ((char16_t)0x005f) +#define AT_SIGN_CHAR ((char16_t)64) +#define PERIOD_CHAR ((char16_t)46) + +U_NAMESPACE_BEGIN + +ICULocaleService::ICULocaleService() + : fallbackLocale(Locale::getDefault()) +{ +} + +ICULocaleService::ICULocaleService(const UnicodeString& dname) + : ICUService(dname) + , fallbackLocale(Locale::getDefault()) +{ +} + +ICULocaleService::~ICULocaleService() +{ +} + +UObject* +ICULocaleService::get(const Locale& locale, UErrorCode& status) const +{ + return get(locale, LocaleKey::KIND_ANY, nullptr, status); +} + +UObject* +ICULocaleService::get(const Locale& locale, int32_t kind, UErrorCode& status) const +{ + return get(locale, kind, nullptr, status); +} + +UObject* +ICULocaleService::get(const Locale& locale, Locale* actualReturn, UErrorCode& status) const +{ + return get(locale, LocaleKey::KIND_ANY, actualReturn, status); +} + +UObject* +ICULocaleService::get(const Locale& locale, int32_t kind, Locale* actualReturn, UErrorCode& status) const +{ + UObject* result = nullptr; + if (U_FAILURE(status)) { + return result; + } + + UnicodeString locName(locale.getName(), -1, US_INV); + if (locName.isBogus()) { + status = U_MEMORY_ALLOCATION_ERROR; + } else { + ICUServiceKey* key = createKey(&locName, kind, status); + if (key) { + if (actualReturn == nullptr) { + result = getKey(*key, status); + } else { + UnicodeString temp; + result = getKey(*key, &temp, status); + + if (result != nullptr) { + key->parseSuffix(temp); + LocaleUtility::initLocaleFromName(temp, *actualReturn); + } + } + delete key; + } + } + return result; +} + + +URegistryKey +ICULocaleService::registerInstance(UObject* objToAdopt, const UnicodeString& locale, + UBool visible, UErrorCode& status) +{ + Locale loc; + LocaleUtility::initLocaleFromName(locale, loc); + return registerInstance(objToAdopt, loc, LocaleKey::KIND_ANY, + visible ? LocaleKeyFactory::VISIBLE : LocaleKeyFactory::INVISIBLE, status); +} + +URegistryKey +ICULocaleService::registerInstance(UObject* objToAdopt, const Locale& locale, UErrorCode& status) +{ + return registerInstance(objToAdopt, locale, LocaleKey::KIND_ANY, LocaleKeyFactory::VISIBLE, status); +} + +URegistryKey +ICULocaleService::registerInstance(UObject* objToAdopt, const Locale& locale, int32_t kind, UErrorCode& status) +{ + return registerInstance(objToAdopt, locale, kind, LocaleKeyFactory::VISIBLE, status); +} + +URegistryKey +ICULocaleService::registerInstance(UObject* objToAdopt, const Locale& locale, int32_t kind, int32_t coverage, UErrorCode& status) +{ + ICUServiceFactory * factory = new SimpleLocaleKeyFactory(objToAdopt, locale, kind, coverage); + if (factory != nullptr) { + return registerFactory(factory, status); + } + delete objToAdopt; + return nullptr; +} + +#if 0 +URegistryKey +ICULocaleService::registerInstance(UObject* objToAdopt, const UnicodeString& locale, UErrorCode& status) +{ + return registerInstance(objToAdopt, locale, LocaleKey::KIND_ANY, LocaleKeyFactory::VISIBLE, status); +} + +URegistryKey +ICULocaleService::registerInstance(UObject* objToAdopt, const UnicodeString& locale, UBool visible, UErrorCode& status) +{ + return registerInstance(objToAdopt, locale, LocaleKey::KIND_ANY, + visible ? LocaleKeyFactory::VISIBLE : LocaleKeyFactory::INVISIBLE, + status); +} + +URegistryKey +ICULocaleService::registerInstance(UObject* objToAdopt, const UnicodeString& locale, int32_t kind, int32_t coverage, UErrorCode& status) +{ + ICUServiceFactory * factory = new SimpleLocaleKeyFactory(objToAdopt, locale, kind, coverage); + if (factory != nullptr) { + return registerFactory(factory, status); + } + delete objToAdopt; + return nullptr; +} +#endif + +class ServiceEnumeration : public StringEnumeration { +private: + const ICULocaleService* _service; + int32_t _timestamp; + UVector _ids; + int32_t _pos; + +private: + ServiceEnumeration(const ICULocaleService* service, UErrorCode &status) + : _service(service) + , _timestamp(service->getTimestamp()) + , _ids(uprv_deleteUObject, nullptr, status) + , _pos(0) + { + _service->getVisibleIDs(_ids, status); + } + + ServiceEnumeration(const ServiceEnumeration &other, UErrorCode &status) + : _service(other._service) + , _timestamp(other._timestamp) + , _ids(uprv_deleteUObject, nullptr, status) + , _pos(0) + { + if(U_SUCCESS(status)) { + int32_t i, length; + + length = other._ids.size(); + for(i = 0; i < length; ++i) { + LocalPointer clonedId(((UnicodeString *)other._ids.elementAt(i))->clone(), status); + _ids.adoptElement(clonedId.orphan(), status); + } + + if(U_SUCCESS(status)) { + _pos = other._pos; + } + } + } + +public: + static ServiceEnumeration* create(const ICULocaleService* service) { + UErrorCode status = U_ZERO_ERROR; + ServiceEnumeration* result = new ServiceEnumeration(service, status); + if (U_SUCCESS(status)) { + return result; + } + delete result; + return nullptr; + } + + virtual ~ServiceEnumeration(); + + virtual StringEnumeration *clone() const override { + UErrorCode status = U_ZERO_ERROR; + ServiceEnumeration *cl = new ServiceEnumeration(*this, status); + if(U_FAILURE(status)) { + delete cl; + cl = nullptr; + } + return cl; + } + + UBool upToDate(UErrorCode& status) const { + if (U_SUCCESS(status)) { + if (_timestamp == _service->getTimestamp()) { + return true; + } + status = U_ENUM_OUT_OF_SYNC_ERROR; + } + return false; + } + + virtual int32_t count(UErrorCode& status) const override { + return upToDate(status) ? _ids.size() : 0; + } + + virtual const UnicodeString* snext(UErrorCode& status) override { + if (upToDate(status) && (_pos < _ids.size())) { + return (const UnicodeString*)_ids[_pos++]; + } + return nullptr; + } + + virtual void reset(UErrorCode& status) override { + if (status == U_ENUM_OUT_OF_SYNC_ERROR) { + status = U_ZERO_ERROR; + } + if (U_SUCCESS(status)) { + _timestamp = _service->getTimestamp(); + _pos = 0; + _service->getVisibleIDs(_ids, status); + } + } + +public: + static UClassID U_EXPORT2 getStaticClassID(); + virtual UClassID getDynamicClassID() const override; +}; + +ServiceEnumeration::~ServiceEnumeration() {} + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(ServiceEnumeration) + +StringEnumeration* +ICULocaleService::getAvailableLocales() const +{ + return ServiceEnumeration::create(this); +} + +const UnicodeString& +ICULocaleService::validateFallbackLocale() const +{ + const Locale& loc = Locale::getDefault(); + ICULocaleService* ncThis = (ICULocaleService*)this; + static UMutex llock; + { + Mutex mutex(&llock); + if (loc != fallbackLocale) { + ncThis->fallbackLocale = loc; + LocaleUtility::initNameFromLocale(loc, ncThis->fallbackLocaleName); + ncThis->clearServiceCache(); + } + } + return fallbackLocaleName; +} + +ICUServiceKey* +ICULocaleService::createKey(const UnicodeString* id, UErrorCode& status) const +{ + return LocaleKey::createWithCanonicalFallback(id, &validateFallbackLocale(), status); +} + +ICUServiceKey* +ICULocaleService::createKey(const UnicodeString* id, int32_t kind, UErrorCode& status) const +{ + return LocaleKey::createWithCanonicalFallback(id, &validateFallbackLocale(), kind, status); +} + +U_NAMESPACE_END + +/* !UCONFIG_NO_SERVICE */ +#endif + + diff --git a/intl/icu/source/common/servnotf.cpp b/intl/icu/source/common/servnotf.cpp new file mode 100644 index 0000000000..b2ad663a48 --- /dev/null +++ b/intl/icu/source/common/servnotf.cpp @@ -0,0 +1,122 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** + ******************************************************************************* + * Copyright (C) 2001-2012, International Business Machines Corporation and * + * others. All Rights Reserved. * + ******************************************************************************* + */ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_SERVICE + +#include "servnotf.h" +#ifdef NOTIFIER_DEBUG +#include +#endif + +U_NAMESPACE_BEGIN + +EventListener::~EventListener() {} +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(EventListener) + +static UMutex notifyLock; + +ICUNotifier::ICUNotifier() +: listeners(nullptr) +{ +} + +ICUNotifier::~ICUNotifier() { + { + Mutex lmx(¬ifyLock); + delete listeners; + listeners = nullptr; + } +} + + +void +ICUNotifier::addListener(const EventListener* l, UErrorCode& status) +{ + if (U_SUCCESS(status)) { + if (l == nullptr) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + if (acceptsListener(*l)) { + Mutex lmx(¬ifyLock); + if (listeners == nullptr) { + LocalPointer lpListeners(new UVector(5, status), status); + if (U_FAILURE(status)) { + return; + } + listeners = lpListeners.orphan(); + } else { + for (int i = 0, e = listeners->size(); i < e; ++i) { + const EventListener* el = (const EventListener*)(listeners->elementAt(i)); + if (l == el) { + return; + } + } + } + + listeners->addElement((void*)l, status); // cast away const + } +#ifdef NOTIFIER_DEBUG + else { + fprintf(stderr, "Listener invalid for this notifier."); + exit(1); + } +#endif + } +} + +void +ICUNotifier::removeListener(const EventListener *l, UErrorCode& status) +{ + if (U_SUCCESS(status)) { + if (l == nullptr) { + status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + { + Mutex lmx(¬ifyLock); + if (listeners != nullptr) { + // identity equality check + for (int i = 0, e = listeners->size(); i < e; ++i) { + const EventListener* el = (const EventListener*)listeners->elementAt(i); + if (l == el) { + listeners->removeElementAt(i); + if (listeners->size() == 0) { + delete listeners; + listeners = nullptr; + } + return; + } + } + } + } + } +} + +void +ICUNotifier::notifyChanged() +{ + Mutex lmx(¬ifyLock); + if (listeners != nullptr) { + for (int i = 0, e = listeners->size(); i < e; ++i) { + EventListener* el = (EventListener*)listeners->elementAt(i); + notifyListener(*el); + } + } +} + +U_NAMESPACE_END + +/* UCONFIG_NO_SERVICE */ +#endif + diff --git a/intl/icu/source/common/servnotf.h b/intl/icu/source/common/servnotf.h new file mode 100644 index 0000000000..840fe1ae87 --- /dev/null +++ b/intl/icu/source/common/servnotf.h @@ -0,0 +1,125 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** + ******************************************************************************* + * Copyright (C) 2001-2014, International Business Machines Corporation and * + * others. All Rights Reserved. * + ******************************************************************************* + */ +#ifndef ICUNOTIF_H +#define ICUNOTIF_H + +#include "unicode/utypes.h" + +#if UCONFIG_NO_SERVICE + +U_NAMESPACE_BEGIN + +/* + * Allow the declaration of APIs with pointers to BreakIterator + * even when break iteration is removed from the build. + */ +class ICUNotifier; + +U_NAMESPACE_END + +#else + +#include "unicode/uobject.h" +#include "unicode/unistr.h" + +#include "mutex.h" +#include "uvector.h" + +U_NAMESPACE_BEGIN + +class U_COMMON_API EventListener : public UObject { +public: + virtual ~EventListener(); + +public: + static UClassID U_EXPORT2 getStaticClassID(); + + virtual UClassID getDynamicClassID() const override; + +public: +#ifdef SERVICE_DEBUG + virtual UnicodeString& debug(UnicodeString& result) const { + return debugClass(result); + } + + virtual UnicodeString& debugClass(UnicodeString& result) const { + return result.append((UnicodeString)"Key"); + } +#endif +}; + +/** + *

Abstract implementation of a notification facility. Clients add + * EventListeners with addListener and remove them with removeListener. + * Notifiers call notifyChanged when they wish to notify listeners. + * This queues the listener list on the notification thread, which + * eventually dequeues the list and calls notifyListener on each + * listener in the list.

+ * + *

Subclasses override acceptsListener and notifyListener + * to add type-safe notification. AcceptsListener should return + * true if the listener is of the appropriate type; ICUNotifier + * itself will ensure the listener is non-null and that the + * identical listener is not already registered with the Notifier. + * NotifyListener should cast the listener to the appropriate + * type and call the appropriate method on the listener. + */ + +class U_COMMON_API ICUNotifier : public UMemory { +private: UVector* listeners; + +public: + ICUNotifier(); + + virtual ~ICUNotifier(); + + /** + * Add a listener to be notified when notifyChanged is called. + * The listener must not be null. AcceptsListener must return + * true for the listener. Attempts to concurrently + * register the identical listener more than once will be + * silently ignored. + */ + virtual void addListener(const EventListener* l, UErrorCode& status); + + /** + * Stop notifying this listener. The listener must + * not be null. Attempts to remove a listener that is + * not registered will be silently ignored. + */ + virtual void removeListener(const EventListener* l, UErrorCode& status); + + /** + * ICU doesn't spawn its own threads. All listeners are notified in + * the thread of the caller. Misbehaved listeners can therefore + * indefinitely block the calling thread. Callers should beware of + * deadlock situations. + */ + virtual void notifyChanged(); + +protected: + /** + * Subclasses implement this to return true if the listener is + * of the appropriate type. + */ + virtual UBool acceptsListener(const EventListener& l) const = 0; + + /** + * Subclasses implement this to notify the listener. + */ + virtual void notifyListener(EventListener& l) const = 0; +}; + +U_NAMESPACE_END + +/* UCONFIG_NO_SERVICE */ +#endif + +/* ICUNOTIF_H */ +#endif diff --git a/intl/icu/source/common/servrbf.cpp b/intl/icu/source/common/servrbf.cpp new file mode 100644 index 0000000000..a4feca2c3b --- /dev/null +++ b/intl/icu/source/common/servrbf.cpp @@ -0,0 +1,96 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** + ******************************************************************************* + * Copyright (C) 2001-2014, International Business Machines Corporation and * + * others. All Rights Reserved. * + ******************************************************************************* + * + ******************************************************************************* + */ +#include "unicode/utypes.h" + +#if !UCONFIG_NO_SERVICE + +#include "unicode/resbund.h" +#include "uresimp.h" +#include "cmemory.h" +#include "servloc.h" +#include "ustrfmt.h" +#include "uhash.h" +#include "charstr.h" +#include "ucln_cmn.h" +#include "uassert.h" + +#define UNDERSCORE_CHAR ((char16_t)0x005f) +#define AT_SIGN_CHAR ((char16_t)64) +#define PERIOD_CHAR ((char16_t)46) + +U_NAMESPACE_BEGIN + +ICUResourceBundleFactory::ICUResourceBundleFactory() + : LocaleKeyFactory(VISIBLE) + , _bundleName() +{ +} + +ICUResourceBundleFactory::ICUResourceBundleFactory(const UnicodeString& bundleName) + : LocaleKeyFactory(VISIBLE) + , _bundleName(bundleName) +{ +} + +ICUResourceBundleFactory::~ICUResourceBundleFactory() {} + +const Hashtable* +ICUResourceBundleFactory::getSupportedIDs(UErrorCode& status) const +{ + if (U_SUCCESS(status)) { + return LocaleUtility::getAvailableLocaleNames(_bundleName); + } + return nullptr; +} + +UObject* +ICUResourceBundleFactory::handleCreate(const Locale& loc, int32_t /* kind */, const ICUService* /* service */, UErrorCode& status) const +{ + if (U_SUCCESS(status)) { + // _bundleName is a package name + // and should only contain invariant characters + // ??? is it always true that the max length of the bundle name is 19? + // who made this change? -- dlf + char pkg[20]; + int32_t length; + length=_bundleName.extract(0, INT32_MAX, pkg, (int32_t)sizeof(pkg), US_INV); + if(length>=(int32_t)sizeof(pkg)) { + return nullptr; + } + return new ResourceBundle(pkg, loc, status); + } + return nullptr; +} + +#ifdef SERVICE_DEBUG +UnicodeString& +ICUResourceBundleFactory::debug(UnicodeString& result) const +{ + LocaleKeyFactory::debug(result); + result.append((UnicodeString)", bundle: "); + return result.append(_bundleName); +} + +UnicodeString& +ICUResourceBundleFactory::debugClass(UnicodeString& result) const +{ + return result.append((UnicodeString)"ICUResourceBundleFactory"); +} +#endif + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(ICUResourceBundleFactory) + +U_NAMESPACE_END + +/* !UCONFIG_NO_SERVICE */ +#endif + + diff --git a/intl/icu/source/common/servslkf.cpp b/intl/icu/source/common/servslkf.cpp new file mode 100644 index 0000000000..b3ab1cc9ce --- /dev/null +++ b/intl/icu/source/common/servslkf.cpp @@ -0,0 +1,123 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/** + ******************************************************************************* + * Copyright (C) 2001-2014, International Business Machines Corporation and * + * others. All Rights Reserved. * + ******************************************************************************* + * + ******************************************************************************* + */ +#include "unicode/utypes.h" + +#if !UCONFIG_NO_SERVICE + +#include "unicode/resbund.h" +#include "uresimp.h" +#include "cmemory.h" +#include "servloc.h" +#include "ustrfmt.h" +#include "uhash.h" +#include "charstr.h" +#include "uassert.h" + +#define UNDERSCORE_CHAR ((char16_t)0x005f) +#define AT_SIGN_CHAR ((char16_t)64) +#define PERIOD_CHAR ((char16_t)46) + +U_NAMESPACE_BEGIN + +/* + ****************************************************************** + */ + +SimpleLocaleKeyFactory::SimpleLocaleKeyFactory(UObject* objToAdopt, + const UnicodeString& locale, + int32_t kind, + int32_t coverage) + : LocaleKeyFactory(coverage) + , _obj(objToAdopt) + , _id(locale) + , _kind(kind) +{ +} + +SimpleLocaleKeyFactory::SimpleLocaleKeyFactory(UObject* objToAdopt, + const Locale& locale, + int32_t kind, + int32_t coverage) + : LocaleKeyFactory(coverage) + , _obj(objToAdopt) + , _id() + , _kind(kind) +{ + LocaleUtility::initNameFromLocale(locale, _id); +} + +SimpleLocaleKeyFactory::~SimpleLocaleKeyFactory() +{ + delete _obj; + _obj = nullptr; +} + +UObject* +SimpleLocaleKeyFactory::create(const ICUServiceKey& key, const ICUService* service, UErrorCode& status) const +{ + if (U_SUCCESS(status)) { + const LocaleKey& lkey = static_cast(key); + if (_kind == LocaleKey::KIND_ANY || _kind == lkey.kind()) { + UnicodeString keyID; + lkey.currentID(keyID); + if (_id == keyID) { + return service->cloneInstance(_obj); + } + } + } + return nullptr; +} + +//UBool +//SimpleLocaleKeyFactory::isSupportedID(const UnicodeString& id, UErrorCode& /* status */) const +//{ +// return id == _id; +//} + +void +SimpleLocaleKeyFactory::updateVisibleIDs(Hashtable& result, UErrorCode& status) const +{ + if (U_SUCCESS(status)) { + if (_coverage & 0x1) { + result.remove(_id); + } else { + result.put(_id, (void*)this, status); + } + } +} + +#ifdef SERVICE_DEBUG +UnicodeString& +SimpleLocaleKeyFactory::debug(UnicodeString& result) const +{ + LocaleKeyFactory::debug(result); + result.append((UnicodeString)", id: "); + result.append(_id); + result.append((UnicodeString)", kind: "); + result.append(_kind); + return result; +} + +UnicodeString& +SimpleLocaleKeyFactory::debugClass(UnicodeString& result) const +{ + return result.append((UnicodeString)"SimpleLocaleKeyFactory"); +} +#endif + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(SimpleLocaleKeyFactory) + +U_NAMESPACE_END + +/* !UCONFIG_NO_SERVICE */ +#endif + + diff --git a/intl/icu/source/common/sharedobject.cpp b/intl/icu/source/common/sharedobject.cpp new file mode 100644 index 0000000000..6eeca8605f --- /dev/null +++ b/intl/icu/source/common/sharedobject.cpp @@ -0,0 +1,62 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 2015, International Business Machines +* Corporation and others. All Rights Reserved. +****************************************************************************** +* sharedobject.cpp +*/ +#include "sharedobject.h" +#include "mutex.h" +#include "uassert.h" +#include "umutex.h" +#include "unifiedcache.h" + +U_NAMESPACE_BEGIN + +SharedObject::~SharedObject() {} + +UnifiedCacheBase::~UnifiedCacheBase() {} + +void +SharedObject::addRef() const { + umtx_atomic_inc(&hardRefCount); +} + +// removeRef Decrement the reference count and delete if it is zero. +// Note that SharedObjects with a non-null cachePtr are owned by the +// unified cache, and the cache will be responsible for the actual deletion. +// The deletion could be as soon as immediately following the +// update to the reference count, if another thread is running +// a cache eviction cycle concurrently. +// NO ACCESS TO *this PERMITTED AFTER REFERENCE COUNT == 0 for cached objects. +// THE OBJECT MAY ALREADY BE GONE. +void +SharedObject::removeRef() const { + const UnifiedCacheBase *cache = this->cachePtr; + int32_t updatedRefCount = umtx_atomic_dec(&hardRefCount); + U_ASSERT(updatedRefCount >= 0); + if (updatedRefCount == 0) { + if (cache) { + cache->handleUnreferencedObject(); + } else { + delete this; + } + } +} + + +int32_t +SharedObject::getRefCount() const { + return umtx_loadAcquire(hardRefCount); +} + +void +SharedObject::deleteIfZeroRefCount() const { + if (this->cachePtr == nullptr && getRefCount() == 0) { + delete this; + } +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/sharedobject.h b/intl/icu/source/common/sharedobject.h new file mode 100644 index 0000000000..5532ec48d8 --- /dev/null +++ b/intl/icu/source/common/sharedobject.h @@ -0,0 +1,184 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 2015-2016, International Business Machines +* Corporation and others. All Rights Reserved. +****************************************************************************** +* sharedobject.h +*/ + +#ifndef __SHAREDOBJECT_H__ +#define __SHAREDOBJECT_H__ + + +#include "unicode/uobject.h" +#include "umutex.h" + +U_NAMESPACE_BEGIN + +class SharedObject; + +/** + * Base class for unified cache exposing enough methods to SharedObject + * instances to allow their addRef() and removeRef() methods to + * update cache metrics. No other part of ICU, except for SharedObject, + * should directly call the methods of this base class. + */ +class U_COMMON_API UnifiedCacheBase : public UObject { +public: + UnifiedCacheBase() { } + + /** + * Notify the cache implementation that an object was seen transitioning to + * zero hard references. The cache may use this to keep track the number of + * unreferenced SharedObjects, and to trigger evictions. + */ + virtual void handleUnreferencedObject() const = 0; + + virtual ~UnifiedCacheBase(); +private: + UnifiedCacheBase(const UnifiedCacheBase &) = delete; + UnifiedCacheBase &operator=(const UnifiedCacheBase &) = delete; +}; + +/** + * Base class for shared, reference-counted, auto-deleted objects. + * Subclasses can be immutable. + * If they are mutable, then they must implement their copy constructor + * so that copyOnWrite() works. + * + * Either stack-allocate, use LocalPointer, or use addRef()/removeRef(). + * Sharing requires reference-counting. + */ +class U_COMMON_API SharedObject : public UObject { +public: + /** Initializes totalRefCount, softRefCount to 0. */ + SharedObject() : + softRefCount(0), + hardRefCount(0), + cachePtr(nullptr) {} + + /** Initializes totalRefCount, softRefCount to 0. */ + SharedObject(const SharedObject &other) : + UObject(other), + softRefCount(0), + hardRefCount(0), + cachePtr(nullptr) {} + + virtual ~SharedObject(); + + /** + * Increments the number of hard references to this object. Thread-safe. + * Not for use from within the Unified Cache implementation. + */ + void addRef() const; + + /** + * Decrements the number of hard references to this object, and + * arrange for possible cache-eviction and/or deletion if ref + * count goes to zero. Thread-safe. + * + * Not for use from within the UnifiedCache implementation. + */ + void removeRef() const; + + /** + * Returns the number of hard references for this object. + * Uses a memory barrier. + */ + int32_t getRefCount() const; + + /** + * If noHardReferences() == true then this object has no hard references. + * Must be called only from within the internals of UnifiedCache. + */ + inline UBool noHardReferences() const { return getRefCount() == 0; } + + /** + * If hasHardReferences() == true then this object has hard references. + * Must be called only from within the internals of UnifiedCache. + */ + inline UBool hasHardReferences() const { return getRefCount() != 0; } + + /** + * Deletes this object if it has no references. + * Available for non-cached SharedObjects only. Ownership of cached objects + * is with the UnifiedCache, which is solely responsible for eviction and deletion. + */ + void deleteIfZeroRefCount() const; + + + /** + * Returns a writable version of ptr. + * If there is exactly one owner, then ptr itself is returned as a + * non-const pointer. + * If there are multiple owners, then ptr is replaced with a + * copy-constructed clone, + * and that is returned. + * Returns nullptr if cloning failed. + * + * T must be a subclass of SharedObject. + */ + template + static T *copyOnWrite(const T *&ptr) { + const T *p = ptr; + if(p->getRefCount() <= 1) { return const_cast(p); } + T *p2 = new T(*p); + if(p2 == nullptr) { return nullptr; } + p->removeRef(); + ptr = p2; + p2->addRef(); + return p2; + } + + /** + * Makes dest an owner of the object pointed to by src while adjusting + * reference counts and deleting the previous object dest pointed to + * if necessary. Before this call is made, dest must either be nullptr or + * be included in the reference count of the object it points to. + * + * T must be a subclass of SharedObject. + */ + template + static void copyPtr(const T *src, const T *&dest) { + if(src != dest) { + if(dest != nullptr) { dest->removeRef(); } + dest = src; + if(src != nullptr) { src->addRef(); } + } + } + + /** + * Equivalent to copyPtr(nullptr, dest). + */ + template + static void clearPtr(const T *&ptr) { + if (ptr != nullptr) { + ptr->removeRef(); + ptr = nullptr; + } + } + +private: + /** + * The number of references from the UnifiedCache, which is + * the number of times that the sharedObject is stored as a hash table value. + * For use by UnifiedCache implementation code only. + * All access is synchronized by UnifiedCache's gCacheMutex + */ + mutable int32_t softRefCount; + friend class UnifiedCache; + + /** + * Reference count, excluding references from within the UnifiedCache implementation. + */ + mutable u_atomic_int32_t hardRefCount; + + mutable const UnifiedCacheBase *cachePtr; + +}; + +U_NAMESPACE_END + +#endif diff --git a/intl/icu/source/common/simpleformatter.cpp b/intl/icu/source/common/simpleformatter.cpp new file mode 100644 index 0000000000..1625627072 --- /dev/null +++ b/intl/icu/source/common/simpleformatter.cpp @@ -0,0 +1,325 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 2014-2016, International Business Machines +* Corporation and others. All Rights Reserved. +****************************************************************************** +* simpleformatter.cpp +*/ + +#include "unicode/utypes.h" +#include "unicode/simpleformatter.h" +#include "unicode/unistr.h" +#include "uassert.h" + +U_NAMESPACE_BEGIN + +namespace { + +/** + * Argument numbers must be smaller than this limit. + * Text segment lengths are offset by this much. + * This is currently the only unused char value in compiled patterns, + * except it is the maximum value of the first unit (max arg +1). + */ +const int32_t ARG_NUM_LIMIT = 0x100; +/** + * Initial and maximum char/char16_t value set for a text segment. + * Segment length char values are from ARG_NUM_LIMIT+1 to this value here. + * Normally 0xffff, but can be as small as ARG_NUM_LIMIT+1 for testing. + */ +const char16_t SEGMENT_LENGTH_PLACEHOLDER_CHAR = 0xffff; +/** + * Maximum length of a text segment. Longer segments are split into shorter ones. + */ +const int32_t MAX_SEGMENT_LENGTH = SEGMENT_LENGTH_PLACEHOLDER_CHAR - ARG_NUM_LIMIT; + +enum { + APOS = 0x27, + DIGIT_ZERO = 0x30, + DIGIT_ONE = 0x31, + DIGIT_NINE = 0x39, + OPEN_BRACE = 0x7b, + CLOSE_BRACE = 0x7d +}; + +inline UBool isInvalidArray(const void *array, int32_t length) { + return (length < 0 || (array == nullptr && length != 0)); +} + +} // namespace + +SimpleFormatter &SimpleFormatter::operator=(const SimpleFormatter& other) { + if (this == &other) { + return *this; + } + compiledPattern = other.compiledPattern; + return *this; +} + +SimpleFormatter::~SimpleFormatter() {} + +UBool SimpleFormatter::applyPatternMinMaxArguments( + const UnicodeString &pattern, + int32_t min, int32_t max, + UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { + return false; + } + // Parse consistent with MessagePattern, but + // - support only simple numbered arguments + // - build a simple binary structure into the result string + const char16_t *patternBuffer = pattern.getBuffer(); + int32_t patternLength = pattern.length(); + // Reserve the first char for the number of arguments. + compiledPattern.setTo((char16_t)0); + int32_t textLength = 0; + int32_t maxArg = -1; + UBool inQuote = false; + for (int32_t i = 0; i < patternLength;) { + char16_t c = patternBuffer[i++]; + if (c == APOS) { + if (i < patternLength && (c = patternBuffer[i]) == APOS) { + // double apostrophe, skip the second one + ++i; + } else if (inQuote) { + // skip the quote-ending apostrophe + inQuote = false; + continue; + } else if (c == OPEN_BRACE || c == CLOSE_BRACE) { + // Skip the quote-starting apostrophe, find the end of the quoted literal text. + ++i; + inQuote = true; + } else { + // The apostrophe is part of literal text. + c = APOS; + } + } else if (!inQuote && c == OPEN_BRACE) { + if (textLength > 0) { + compiledPattern.setCharAt(compiledPattern.length() - textLength - 1, + (char16_t)(ARG_NUM_LIMIT + textLength)); + textLength = 0; + } + int32_t argNumber; + if ((i + 1) < patternLength && + 0 <= (argNumber = patternBuffer[i] - DIGIT_ZERO) && argNumber <= 9 && + patternBuffer[i + 1] == CLOSE_BRACE) { + i += 2; + } else { + // Multi-digit argument number (no leading zero) or syntax error. + // MessagePattern permits PatternProps.skipWhiteSpace(pattern, index) + // around the number, but this class does not. + argNumber = -1; + if (i < patternLength && DIGIT_ONE <= (c = patternBuffer[i++]) && c <= DIGIT_NINE) { + argNumber = c - DIGIT_ZERO; + while (i < patternLength && + DIGIT_ZERO <= (c = patternBuffer[i++]) && c <= DIGIT_NINE) { + argNumber = argNumber * 10 + (c - DIGIT_ZERO); + if (argNumber >= ARG_NUM_LIMIT) { + break; + } + } + } + if (argNumber < 0 || c != CLOSE_BRACE) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return false; + } + } + if (argNumber > maxArg) { + maxArg = argNumber; + } + compiledPattern.append((char16_t)argNumber); + continue; + } // else: c is part of literal text + // Append c and track the literal-text segment length. + if (textLength == 0) { + // Reserve a char for the length of a new text segment, preset the maximum length. + compiledPattern.append(SEGMENT_LENGTH_PLACEHOLDER_CHAR); + } + compiledPattern.append(c); + if (++textLength == MAX_SEGMENT_LENGTH) { + textLength = 0; + } + } + if (textLength > 0) { + compiledPattern.setCharAt(compiledPattern.length() - textLength - 1, + (char16_t)(ARG_NUM_LIMIT + textLength)); + } + int32_t argCount = maxArg + 1; + if (argCount < min || max < argCount) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return false; + } + compiledPattern.setCharAt(0, (char16_t)argCount); + return true; +} + +UnicodeString& SimpleFormatter::format( + const UnicodeString &value0, + UnicodeString &appendTo, UErrorCode &errorCode) const { + const UnicodeString *values[] = { &value0 }; + return formatAndAppend(values, 1, appendTo, nullptr, 0, errorCode); +} + +UnicodeString& SimpleFormatter::format( + const UnicodeString &value0, + const UnicodeString &value1, + UnicodeString &appendTo, UErrorCode &errorCode) const { + const UnicodeString *values[] = { &value0, &value1 }; + return formatAndAppend(values, 2, appendTo, nullptr, 0, errorCode); +} + +UnicodeString& SimpleFormatter::format( + const UnicodeString &value0, + const UnicodeString &value1, + const UnicodeString &value2, + UnicodeString &appendTo, UErrorCode &errorCode) const { + const UnicodeString *values[] = { &value0, &value1, &value2 }; + return formatAndAppend(values, 3, appendTo, nullptr, 0, errorCode); +} + +UnicodeString& SimpleFormatter::formatAndAppend( + const UnicodeString *const *values, int32_t valuesLength, + UnicodeString &appendTo, + int32_t *offsets, int32_t offsetsLength, UErrorCode &errorCode) const { + if (U_FAILURE(errorCode)) { + return appendTo; + } + if (isInvalidArray(values, valuesLength) || isInvalidArray(offsets, offsetsLength) || + valuesLength < getArgumentLimit()) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return appendTo; + } + return format(compiledPattern.getBuffer(), compiledPattern.length(), values, + appendTo, nullptr, true, + offsets, offsetsLength, errorCode); +} + +UnicodeString &SimpleFormatter::formatAndReplace( + const UnicodeString *const *values, int32_t valuesLength, + UnicodeString &result, + int32_t *offsets, int32_t offsetsLength, UErrorCode &errorCode) const { + if (U_FAILURE(errorCode)) { + return result; + } + if (isInvalidArray(values, valuesLength) || isInvalidArray(offsets, offsetsLength)) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return result; + } + const char16_t *cp = compiledPattern.getBuffer(); + int32_t cpLength = compiledPattern.length(); + if (valuesLength < getArgumentLimit(cp, cpLength)) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return result; + } + + // If the pattern starts with an argument whose value is the same object + // as the result, then we keep the result contents and append to it. + // Otherwise we replace its contents. + int32_t firstArg = -1; + // If any non-initial argument value is the same object as the result, + // then we first copy its contents and use that instead while formatting. + UnicodeString resultCopy; + if (getArgumentLimit(cp, cpLength) > 0) { + for (int32_t i = 1; i < cpLength;) { + int32_t n = cp[i++]; + if (n < ARG_NUM_LIMIT) { + if (values[n] == &result) { + if (i == 2) { + firstArg = n; + } else if (resultCopy.isEmpty() && !result.isEmpty()) { + resultCopy = result; + } + } + } else { + i += n - ARG_NUM_LIMIT; + } + } + } + if (firstArg < 0) { + result.remove(); + } + return format(cp, cpLength, values, + result, &resultCopy, false, + offsets, offsetsLength, errorCode); +} + +UnicodeString SimpleFormatter::getTextWithNoArguments( + const char16_t *compiledPattern, + int32_t compiledPatternLength, + int32_t* offsets, + int32_t offsetsLength) { + for (int32_t i = 0; i < offsetsLength; i++) { + offsets[i] = -1; + } + int32_t capacity = compiledPatternLength - 1 - + getArgumentLimit(compiledPattern, compiledPatternLength); + UnicodeString sb(capacity, 0, 0); // Java: StringBuilder + for (int32_t i = 1; i < compiledPatternLength;) { + int32_t n = compiledPattern[i++]; + if (n > ARG_NUM_LIMIT) { + n -= ARG_NUM_LIMIT; + sb.append(compiledPattern + i, n); + i += n; + } else if (n < offsetsLength) { + // TODO(ICU-20406): This does not distinguish between "{0}{1}" and "{1}{0}". + // Consider removing this function and replacing it with an iterator interface. + offsets[n] = sb.length(); + } + } + return sb; +} + +UnicodeString &SimpleFormatter::format( + const char16_t *compiledPattern, int32_t compiledPatternLength, + const UnicodeString *const *values, + UnicodeString &result, const UnicodeString *resultCopy, UBool forbidResultAsValue, + int32_t *offsets, int32_t offsetsLength, + UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { + return result; + } + for (int32_t i = 0; i < offsetsLength; i++) { + offsets[i] = -1; + } + for (int32_t i = 1; i < compiledPatternLength;) { + int32_t n = compiledPattern[i++]; + if (n < ARG_NUM_LIMIT) { + const UnicodeString *value = values[n]; + if (value == nullptr) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return result; + } + if (value == &result) { + if (forbidResultAsValue) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return result; + } + if (i == 2) { + // We are appending to result which is also the first value object. + if (n < offsetsLength) { + offsets[n] = 0; + } + } else { + if (n < offsetsLength) { + offsets[n] = result.length(); + } + result.append(*resultCopy); + } + } else { + if (n < offsetsLength) { + offsets[n] = result.length(); + } + result.append(*value); + } + } else { + int32_t length = n - ARG_NUM_LIMIT; + result.append(compiledPattern + i, length); + i += length; + } + } + return result; +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/sources.txt b/intl/icu/source/common/sources.txt new file mode 100644 index 0000000000..90171fe9bd --- /dev/null +++ b/intl/icu/source/common/sources.txt @@ -0,0 +1,199 @@ +appendable.cpp +bmpset.cpp +brkeng.cpp +brkiter.cpp +bytesinkutil.cpp +bytestream.cpp +bytestrie.cpp +bytestriebuilder.cpp +bytestrieiterator.cpp +caniter.cpp +characterproperties.cpp +chariter.cpp +charstr.cpp +cmemory.cpp +cstr.cpp +cstring.cpp +cwchar.cpp +dictbe.cpp +dictionarydata.cpp +dtintrv.cpp +edits.cpp +emojiprops.cpp +errorcode.cpp +filteredbrk.cpp +filterednormalizer2.cpp +icudataver.cpp +icuplug.cpp +loadednormalizer2impl.cpp +localebuilder.cpp +localematcher.cpp +localeprioritylist.cpp +locavailable.cpp +locbased.cpp +locdispnames.cpp +locdistance.cpp +locdspnm.cpp +locid.cpp +loclikely.cpp +loclikelysubtags.cpp +locmap.cpp +locresdata.cpp +locutil.cpp +lsr.cpp +lstmbe.cpp +messagepattern.cpp +mlbe.cpp +normalizer2.cpp +normalizer2impl.cpp +normlzr.cpp +parsepos.cpp +patternprops.cpp +pluralmap.cpp +propname.cpp +propsvec.cpp +punycode.cpp +putil.cpp +rbbi.cpp +rbbi_cache.cpp +rbbidata.cpp +rbbinode.cpp +rbbirb.cpp +rbbiscan.cpp +rbbisetb.cpp +rbbistbl.cpp +rbbitblb.cpp +resbund.cpp +resbund_cnv.cpp +resource.cpp +restrace.cpp +ruleiter.cpp +schriter.cpp +serv.cpp +servlk.cpp +servlkf.cpp +servls.cpp +servnotf.cpp +servrbf.cpp +servslkf.cpp +sharedobject.cpp +simpleformatter.cpp +static_unicode_sets.cpp +stringpiece.cpp +stringtriebuilder.cpp +uarrsort.cpp +ubidi.cpp +ubidi_props.cpp +ubidiln.cpp +ubiditransform.cpp +ubidiwrt.cpp +ubrk.cpp +ucase.cpp +ucasemap.cpp +ucasemap_titlecase_brkiter.cpp +ucat.cpp +uchar.cpp +ucharstrie.cpp +ucharstriebuilder.cpp +ucharstrieiterator.cpp +uchriter.cpp +ucln_cmn.cpp +ucmndata.cpp +ucnv.cpp +ucnv2022.cpp +ucnv_bld.cpp +ucnv_cb.cpp +ucnv_cnv.cpp +ucnv_ct.cpp +ucnv_err.cpp +ucnv_ext.cpp +ucnv_io.cpp +ucnv_lmb.cpp +ucnv_set.cpp +ucnv_u16.cpp +ucnv_u32.cpp +ucnv_u7.cpp +ucnv_u8.cpp +ucnvbocu.cpp +ucnvdisp.cpp +ucnvhz.cpp +ucnvisci.cpp +ucnvlat1.cpp +ucnvmbcs.cpp +ucnvscsu.cpp +ucnvsel.cpp +ucol_swp.cpp +ucptrie.cpp +ucurr.cpp +udata.cpp +udatamem.cpp +udataswp.cpp +uenum.cpp +uhash.cpp +uhash_us.cpp +uidna.cpp +uinit.cpp +uinvchar.cpp +uiter.cpp +ulist.cpp +uloc.cpp +uloc_keytype.cpp +uloc_tag.cpp +umapfile.cpp +umath.cpp +umutablecptrie.cpp +umutex.cpp +unames.cpp +unifiedcache.cpp +unifilt.cpp +unifunct.cpp +uniset.cpp +uniset_closure.cpp +uniset_props.cpp +unisetspan.cpp +unistr.cpp +unistr_case.cpp +unistr_case_locale.cpp +unistr_cnv.cpp +unistr_props.cpp +unistr_titlecase_brkiter.cpp +unorm.cpp +unormcmp.cpp +uobject.cpp +uprops.cpp +ures_cnv.cpp +uresbund.cpp +uresdata.cpp +usc_impl.cpp +uscript.cpp +uscript_props.cpp +uset.cpp +uset_props.cpp +usetiter.cpp +ushape.cpp +usprep.cpp +ustack.cpp +ustr_cnv.cpp +ustr_titlecase_brkiter.cpp +ustr_wcs.cpp +ustrcase.cpp +ustrcase_locale.cpp +ustrenum.cpp +ustrfmt.cpp +ustring.cpp +ustrtrns.cpp +utext.cpp +utf_impl.cpp +util.cpp +util_props.cpp +utrace.cpp +utrie.cpp +utrie2.cpp +utrie2_builder.cpp +utrie_swap.cpp +uts46.cpp +utypes.cpp +uvector.cpp +uvectr32.cpp +uvectr64.cpp +wintz.cpp diff --git a/intl/icu/source/common/sprpimpl.h b/intl/icu/source/common/sprpimpl.h new file mode 100644 index 0000000000..ca0bcdb516 --- /dev/null +++ b/intl/icu/source/common/sprpimpl.h @@ -0,0 +1,130 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ******************************************************************************* + * + * Copyright (C) 2003-2006, International Business Machines + * Corporation and others. All Rights Reserved. + * + ******************************************************************************* + * file name: sprpimpl.h + * encoding: UTF-8 + * tab size: 8 (not used) + * indentation:4 + * + * created on: 2003feb1 + * created by: Ram Viswanadha + */ + +#ifndef SPRPIMPL_H +#define SPRPIMPL_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_IDNA + +#include "unicode/ustring.h" +#include "unicode/parseerr.h" +#include "unicode/usprep.h" +#include "unicode/udata.h" +#include "utrie.h" +#include "udataswp.h" +#include "ubidi_props.h" + +#define _SPREP_DATA_TYPE "spp" + +enum UStringPrepType{ + USPREP_UNASSIGNED = 0x0000 , + USPREP_MAP = 0x0001 , + USPREP_PROHIBITED = 0x0002 , + USPREP_DELETE = 0x0003 , + USPREP_TYPE_LIMIT = 0x0004 +}; + +typedef enum UStringPrepType UStringPrepType; + +#ifdef USPREP_TYPE_NAMES_ARRAY +static const char* usprepTypeNames[] ={ + "UNASSIGNED" , + "MAP" , + "PROHIBITED" , + "DELETE", + "TYPE_LIMIT" +}; +#endif + +enum{ + _SPREP_NORMALIZATION_ON = 0x0001, + _SPREP_CHECK_BIDI_ON = 0x0002 +}; + +enum{ + _SPREP_TYPE_THRESHOLD = 0xFFF0, + _SPREP_MAX_INDEX_VALUE = 0x3FBF, /*16139*/ + _SPREP_MAX_INDEX_TOP_LENGTH = 0x0003 +}; + +/* indexes[] value names */ +enum { + _SPREP_INDEX_TRIE_SIZE = 0, /* number of bytes in StringPrep trie */ + _SPREP_INDEX_MAPPING_DATA_SIZE = 1, /* The array that contains the mapping */ + _SPREP_NORM_CORRECTNS_LAST_UNI_VERSION = 2, /* The index of Unicode version of last entry in NormalizationCorrections.txt */ + _SPREP_ONE_UCHAR_MAPPING_INDEX_START = 3, /* The starting index of 1 UChar mapping index in the mapping data array */ + _SPREP_TWO_UCHARS_MAPPING_INDEX_START = 4, /* The starting index of 2 UChars mapping index in the mapping data array */ + _SPREP_THREE_UCHARS_MAPPING_INDEX_START = 5, /* The starting index of 3 UChars mapping index in the mapping data array */ + _SPREP_FOUR_UCHARS_MAPPING_INDEX_START = 6, /* The starting index of 4 UChars mapping index in the mapping data array */ + _SPREP_OPTIONS = 7, /* Bit set of options to turn on in the profile */ + _SPREP_INDEX_TOP=16 /* changing this requires a new formatVersion */ +}; + +typedef struct UStringPrepKey UStringPrepKey; + + +struct UStringPrepKey{ + char* name; + char* path; +}; + +struct UStringPrepProfile{ + int32_t indexes[_SPREP_INDEX_TOP]; + UTrie sprepTrie; + const uint16_t* mappingData; + UDataMemory* sprepData; + int32_t refCount; + UBool isDataLoaded; + UBool doNFKC; + UBool checkBiDi; +}; + +/** + * Helper function for populating the UParseError struct + * @internal + */ +U_CAPI void U_EXPORT2 +uprv_syntaxError(const UChar* rules, + int32_t pos, + int32_t rulesLen, + UParseError* parseError); + + +/** + * Swap StringPrep .spp profile data. See udataswp.h. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +usprep_swap(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +#endif /* #if !UCONFIG_NO_IDNA */ + +#endif + +/* + * Hey, Emacs, please set the following: + * + * Local Variables: + * indent-tabs-mode: nil + * End: + * + */ diff --git a/intl/icu/source/common/static_unicode_sets.cpp b/intl/icu/source/common/static_unicode_sets.cpp new file mode 100644 index 0000000000..0db5ea000d --- /dev/null +++ b/intl/icu/source/common/static_unicode_sets.cpp @@ -0,0 +1,245 @@ +// © 2018 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_FORMATTING + +// Allow implicit conversion from char16_t* to UnicodeString for this file: +// Helpful in toString methods and elsewhere. +#define UNISTR_FROM_STRING_EXPLICIT + +#include "static_unicode_sets.h" +#include "umutex.h" +#include "ucln_cmn.h" +#include "unicode/uniset.h" +#include "uresimp.h" +#include "cstring.h" +#include "uassert.h" + +using namespace icu; +using namespace icu::unisets; + + +namespace { + +UnicodeSet* gUnicodeSets[UNISETS_KEY_COUNT] = {}; + +// Save the empty instance in static memory to have well-defined behavior if a +// regular UnicodeSet cannot be allocated. +alignas(UnicodeSet) +char gEmptyUnicodeSet[sizeof(UnicodeSet)]; + +// Whether the gEmptyUnicodeSet is initialized and ready to use. +UBool gEmptyUnicodeSetInitialized = false; + +inline UnicodeSet* getImpl(Key key) { + UnicodeSet* candidate = gUnicodeSets[key]; + if (candidate == nullptr) { + return reinterpret_cast(gEmptyUnicodeSet); + } + return candidate; +} + +UnicodeSet* computeUnion(Key k1, Key k2) { + UnicodeSet* result = new UnicodeSet(); + if (result == nullptr) { + return nullptr; + } + result->addAll(*getImpl(k1)); + result->addAll(*getImpl(k2)); + result->freeze(); + return result; +} + +UnicodeSet* computeUnion(Key k1, Key k2, Key k3) { + UnicodeSet* result = new UnicodeSet(); + if (result == nullptr) { + return nullptr; + } + result->addAll(*getImpl(k1)); + result->addAll(*getImpl(k2)); + result->addAll(*getImpl(k3)); + result->freeze(); + return result; +} + + +void saveSet(Key key, const UnicodeString& unicodeSetPattern, UErrorCode& status) { + // assert unicodeSets.get(key) == null; + gUnicodeSets[key] = new UnicodeSet(unicodeSetPattern, status); +} + +class ParseDataSink : public ResourceSink { + public: + void put(const char* key, ResourceValue& value, UBool /*noFallback*/, UErrorCode& status) override { + ResourceTable contextsTable = value.getTable(status); + if (U_FAILURE(status)) { return; } + for (int i = 0; contextsTable.getKeyAndValue(i, key, value); i++) { + if (uprv_strcmp(key, "date") == 0) { + // ignore + } else { + ResourceTable strictnessTable = value.getTable(status); + if (U_FAILURE(status)) { return; } + for (int j = 0; strictnessTable.getKeyAndValue(j, key, value); j++) { + bool isLenient = (uprv_strcmp(key, "lenient") == 0); + ResourceArray array = value.getArray(status); + if (U_FAILURE(status)) { return; } + for (int k = 0; k < array.getSize(); k++) { + array.getValue(k, value); + UnicodeString str = value.getUnicodeString(status); + if (U_FAILURE(status)) { return; } + // There is both lenient and strict data for comma/period, + // but not for any of the other symbols. + if (str.indexOf(u'.') != -1) { + saveSet(isLenient ? PERIOD : STRICT_PERIOD, str, status); + } else if (str.indexOf(u',') != -1) { + saveSet(isLenient ? COMMA : STRICT_COMMA, str, status); + } else if (str.indexOf(u'+') != -1) { + saveSet(PLUS_SIGN, str, status); + } else if (str.indexOf(u'-') != -1) { + saveSet(MINUS_SIGN, str, status); + } else if (str.indexOf(u'$') != -1) { + saveSet(DOLLAR_SIGN, str, status); + } else if (str.indexOf(u'£') != -1) { + saveSet(POUND_SIGN, str, status); + } else if (str.indexOf(u'₹') != -1) { + saveSet(RUPEE_SIGN, str, status); + } else if (str.indexOf(u'¥') != -1) { + saveSet(YEN_SIGN, str, status); + } else if (str.indexOf(u'₩') != -1) { + saveSet(WON_SIGN, str, status); + } else if (str.indexOf(u'%') != -1) { + saveSet(PERCENT_SIGN, str, status); + } else if (str.indexOf(u'‰') != -1) { + saveSet(PERMILLE_SIGN, str, status); + } else if (str.indexOf(u'’') != -1) { + saveSet(APOSTROPHE_SIGN, str, status); + } else { + // Unknown class of parse lenients + // TODO(ICU-20428): Make ICU automatically accept new classes? + U_ASSERT(false); + } + if (U_FAILURE(status)) { return; } + } + } + } + } + } +}; + + +icu::UInitOnce gNumberParseUniSetsInitOnce {}; + +UBool U_CALLCONV cleanupNumberParseUniSets() { + if (gEmptyUnicodeSetInitialized) { + reinterpret_cast(gEmptyUnicodeSet)->~UnicodeSet(); + gEmptyUnicodeSetInitialized = false; + } + for (int32_t i = 0; i < UNISETS_KEY_COUNT; i++) { + delete gUnicodeSets[i]; + gUnicodeSets[i] = nullptr; + } + gNumberParseUniSetsInitOnce.reset(); + return true; +} + +void U_CALLCONV initNumberParseUniSets(UErrorCode& status) { + ucln_common_registerCleanup(UCLN_COMMON_NUMPARSE_UNISETS, cleanupNumberParseUniSets); + + // Initialize the empty instance for well-defined fallback behavior + new(gEmptyUnicodeSet) UnicodeSet(); + reinterpret_cast(gEmptyUnicodeSet)->freeze(); + gEmptyUnicodeSetInitialized = true; + + // These sets were decided after discussion with icu-design@. See tickets #13084 and #13309. + // Zs+TAB is "horizontal whitespace" according to UTS #18 (blank property). + gUnicodeSets[DEFAULT_IGNORABLES] = new UnicodeSet( + u"[[:Zs:][\\u0009][:Bidi_Control:][:Variation_Selector:]]", status); + gUnicodeSets[STRICT_IGNORABLES] = new UnicodeSet(u"[[:Bidi_Control:]]", status); + + LocalUResourceBundlePointer rb(ures_open(nullptr, "root", &status)); + if (U_FAILURE(status)) { return; } + ParseDataSink sink; + ures_getAllItemsWithFallback(rb.getAlias(), "parse", sink, status); + if (U_FAILURE(status)) { return; } + + // NOTE: It is OK for these assertions to fail if there was a no-data build. + U_ASSERT(gUnicodeSets[COMMA] != nullptr); + U_ASSERT(gUnicodeSets[STRICT_COMMA] != nullptr); + U_ASSERT(gUnicodeSets[PERIOD] != nullptr); + U_ASSERT(gUnicodeSets[STRICT_PERIOD] != nullptr); + U_ASSERT(gUnicodeSets[APOSTROPHE_SIGN] != nullptr); + + LocalPointer otherGrouping(new UnicodeSet( + u"[٬‘\\u0020\\u00A0\\u2000-\\u200A\\u202F\\u205F\\u3000]", + status + ), status); + if (U_FAILURE(status)) { return; } + otherGrouping->addAll(*gUnicodeSets[APOSTROPHE_SIGN]); + gUnicodeSets[OTHER_GROUPING_SEPARATORS] = otherGrouping.orphan(); + gUnicodeSets[ALL_SEPARATORS] = computeUnion(COMMA, PERIOD, OTHER_GROUPING_SEPARATORS); + gUnicodeSets[STRICT_ALL_SEPARATORS] = computeUnion( + STRICT_COMMA, STRICT_PERIOD, OTHER_GROUPING_SEPARATORS); + + U_ASSERT(gUnicodeSets[MINUS_SIGN] != nullptr); + U_ASSERT(gUnicodeSets[PLUS_SIGN] != nullptr); + U_ASSERT(gUnicodeSets[PERCENT_SIGN] != nullptr); + U_ASSERT(gUnicodeSets[PERMILLE_SIGN] != nullptr); + + gUnicodeSets[INFINITY_SIGN] = new UnicodeSet(u"[∞]", status); + if (U_FAILURE(status)) { return; } + + U_ASSERT(gUnicodeSets[DOLLAR_SIGN] != nullptr); + U_ASSERT(gUnicodeSets[POUND_SIGN] != nullptr); + U_ASSERT(gUnicodeSets[RUPEE_SIGN] != nullptr); + U_ASSERT(gUnicodeSets[YEN_SIGN] != nullptr); + U_ASSERT(gUnicodeSets[WON_SIGN] != nullptr); + + gUnicodeSets[DIGITS] = new UnicodeSet(u"[:digit:]", status); + if (U_FAILURE(status)) { return; } + gUnicodeSets[DIGITS_OR_ALL_SEPARATORS] = computeUnion(DIGITS, ALL_SEPARATORS); + gUnicodeSets[DIGITS_OR_STRICT_ALL_SEPARATORS] = computeUnion(DIGITS, STRICT_ALL_SEPARATORS); + + for (auto* uniset : gUnicodeSets) { + if (uniset != nullptr) { + uniset->freeze(); + } + } +} + +} + +const UnicodeSet* unisets::get(Key key) { + UErrorCode localStatus = U_ZERO_ERROR; + umtx_initOnce(gNumberParseUniSetsInitOnce, &initNumberParseUniSets, localStatus); + if (U_FAILURE(localStatus)) { + return reinterpret_cast(gEmptyUnicodeSet); + } + return getImpl(key); +} + +Key unisets::chooseFrom(UnicodeString str, Key key1) { + return get(key1)->contains(str) ? key1 : NONE; +} + +Key unisets::chooseFrom(UnicodeString str, Key key1, Key key2) { + return get(key1)->contains(str) ? key1 : chooseFrom(str, key2); +} + +//Key unisets::chooseCurrency(UnicodeString str) { +// if (get(DOLLAR_SIGN)->contains(str)) { +// return DOLLAR_SIGN; +// } else if (get(POUND_SIGN)->contains(str)) { +// return POUND_SIGN; +// } else if (get(RUPEE_SIGN)->contains(str)) { +// return RUPEE_SIGN; +// } else if (get(YEN_SIGN)->contains(str)) { +// return YEN_SIGN; +// } else { +// return NONE; +// } +//} + + +#endif /* #if !UCONFIG_NO_FORMATTING */ diff --git a/intl/icu/source/common/static_unicode_sets.h b/intl/icu/source/common/static_unicode_sets.h new file mode 100644 index 0000000000..5d90ce5908 --- /dev/null +++ b/intl/icu/source/common/static_unicode_sets.h @@ -0,0 +1,140 @@ +// © 2018 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// This file contains utilities to deal with static-allocated UnicodeSets. +// +// Common use case: you write a "private static final" UnicodeSet in Java, and +// want something similarly easy in C++. Originally written for number +// parsing, but this header can be used for other applications. +// +// Main entrypoint: `unisets::get(unisets::MY_SET_ID_HERE)` +// +// This file is in common instead of i18n because it is needed by ucurr.cpp. +// +// Author: sffc + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_FORMATTING +#ifndef __STATIC_UNICODE_SETS_H__ +#define __STATIC_UNICODE_SETS_H__ + +#include "unicode/uniset.h" +#include "unicode/unistr.h" + +U_NAMESPACE_BEGIN +namespace unisets { + +enum Key { + // NONE is used to indicate null in chooseFrom(). + // EMPTY is used to get an empty UnicodeSet. + NONE = -1, + EMPTY = 0, + + // Ignorables + DEFAULT_IGNORABLES, + STRICT_IGNORABLES, + + // Separators + // Notes: + // - COMMA is a superset of STRICT_COMMA + // - PERIOD is a superset of SCRICT_PERIOD + // - ALL_SEPARATORS is the union of COMMA, PERIOD, and OTHER_GROUPING_SEPARATORS + // - STRICT_ALL_SEPARATORS is the union of STRICT_COMMA, STRICT_PERIOD, and OTHER_GRP_SEPARATORS + COMMA, + PERIOD, + STRICT_COMMA, + STRICT_PERIOD, + APOSTROPHE_SIGN, + OTHER_GROUPING_SEPARATORS, + ALL_SEPARATORS, + STRICT_ALL_SEPARATORS, + + // Symbols + MINUS_SIGN, + PLUS_SIGN, + PERCENT_SIGN, + PERMILLE_SIGN, + INFINITY_SIGN, + + // Currency Symbols + DOLLAR_SIGN, + POUND_SIGN, + RUPEE_SIGN, + YEN_SIGN, + WON_SIGN, + + // Other + DIGITS, + + // Combined Separators with Digits (for lead code points) + DIGITS_OR_ALL_SEPARATORS, + DIGITS_OR_STRICT_ALL_SEPARATORS, + + // The number of elements in the enum. + UNISETS_KEY_COUNT +}; + +/** + * Gets the static-allocated UnicodeSet according to the provided key. The + * pointer will be deleted during u_cleanup(); the caller should NOT delete it. + * + * Exported as U_COMMON_API for ucurr.cpp + * + * This method is always safe and OK to chain: in the case of a memory or other + * error, it returns an empty set from static memory. + * + * Example: + * + * UBool hasIgnorables = unisets::get(unisets::DEFAULT_IGNORABLES)->contains(...); + * + * @param key The desired UnicodeSet according to the enum in this file. + * @return The requested UnicodeSet. Guaranteed to be frozen and non-null, but + * may be empty if an error occurred during data loading. + */ +U_COMMON_API const UnicodeSet* get(Key key); + +/** + * Checks if the UnicodeSet given by key1 contains the given string. + * + * Exported as U_COMMON_API for numparse_decimal.cpp + * + * @param str The string to check. + * @param key1 The set to check. + * @return key1 if the set contains str, or NONE if not. + */ +U_COMMON_API Key chooseFrom(UnicodeString str, Key key1); + +/** + * Checks if the UnicodeSet given by either key1 or key2 contains the string. + * + * Exported as U_COMMON_API for numparse_decimal.cpp + * + * @param str The string to check. + * @param key1 The first set to check. + * @param key2 The second set to check. + * @return key1 if that set contains str; key2 if that set contains str; or + * NONE if neither set contains str. + */ +U_COMMON_API Key chooseFrom(UnicodeString str, Key key1, Key key2); + +// TODO: Load these from data: ICU-20108 +// Unused in C++: +// Key chooseCurrency(UnicodeString str); +// Used instead: +static const struct { + Key key; + UChar32 exemplar; +} kCurrencyEntries[] = { + {DOLLAR_SIGN, u'$'}, + {POUND_SIGN, u'£'}, + {RUPEE_SIGN, u'₹'}, + {YEN_SIGN, u'¥'}, + {WON_SIGN, u'₩'}, +}; + +} // namespace unisets +U_NAMESPACE_END + +#endif //__STATIC_UNICODE_SETS_H__ +#endif /* #if !UCONFIG_NO_FORMATTING */ diff --git a/intl/icu/source/common/stringpiece.cpp b/intl/icu/source/common/stringpiece.cpp new file mode 100644 index 0000000000..eb9766c018 --- /dev/null +++ b/intl/icu/source/common/stringpiece.cpp @@ -0,0 +1,116 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// Copyright (C) 2009-2013, International Business Machines +// Corporation and others. All Rights Reserved. +// +// Copyright 2004 and onwards Google Inc. +// +// Author: wilsonh@google.com (Wilson Hsieh) +// + +#include "unicode/utypes.h" +#include "unicode/stringpiece.h" +#include "cstring.h" +#include "cmemory.h" + +U_NAMESPACE_BEGIN + +StringPiece::StringPiece(const char* str) + : ptr_(str), length_((str == nullptr) ? 0 : static_cast(uprv_strlen(str))) { } + +StringPiece::StringPiece(const StringPiece& x, int32_t pos) { + if (pos < 0) { + pos = 0; + } else if (pos > x.length_) { + pos = x.length_; + } + ptr_ = x.ptr_ + pos; + length_ = x.length_ - pos; +} + +StringPiece::StringPiece(const StringPiece& x, int32_t pos, int32_t len) { + if (pos < 0) { + pos = 0; + } else if (pos > x.length_) { + pos = x.length_; + } + if (len < 0) { + len = 0; + } else if (len > x.length_ - pos) { + len = x.length_ - pos; + } + ptr_ = x.ptr_ + pos; + length_ = len; +} + +void StringPiece::set(const char* str) { + ptr_ = str; + if (str != nullptr) + length_ = static_cast(uprv_strlen(str)); + else + length_ = 0; +} + +int32_t StringPiece::find(StringPiece needle, int32_t offset) { + if (length() == 0 && needle.length() == 0) { + return 0; + } + // TODO: Improve to be better than O(N^2)? + for (int32_t i = offset; i < length(); i++) { + int32_t j = 0; + for (; j < needle.length(); i++, j++) { + if (data()[i] != needle.data()[j]) { + i -= j; + goto outer_end; + } + } + return i - j; + outer_end: void(); + } + return -1; +} + +int32_t StringPiece::compare(StringPiece other) { + int32_t i = 0; + for (; i < length(); i++) { + if (i == other.length()) { + // this is longer + return 1; + } + char a = data()[i]; + char b = other.data()[i]; + if (a < b) { + return -1; + } else if (a > b) { + return 1; + } + } + if (i < other.length()) { + // other is longer + return -1; + } + return 0; +} + +U_EXPORT UBool U_EXPORT2 +operator==(const StringPiece& x, const StringPiece& y) { + int32_t len = x.size(); + if (len != y.size()) { + return false; + } + if (len == 0) { + return true; + } + const char* p = x.data(); + const char* p2 = y.data(); + // Test last byte in case strings share large common prefix + --len; + if (p[len] != p2[len]) return false; + // At this point we can, but don't have to, ignore the last byte. + return uprv_memcmp(p, p2, len) == 0; +} + + +const int32_t StringPiece::npos = 0x7fffffff; + +U_NAMESPACE_END diff --git a/intl/icu/source/common/stringtriebuilder.cpp b/intl/icu/source/common/stringtriebuilder.cpp new file mode 100644 index 0000000000..2a40dc88fd --- /dev/null +++ b/intl/icu/source/common/stringtriebuilder.cpp @@ -0,0 +1,618 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2012, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: stringtriebuilder.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010dec24 +* created by: Markus W. Scherer +*/ + +#include "utypeinfo.h" // for 'typeid' to work +#include "unicode/utypes.h" +#include "unicode/stringtriebuilder.h" +#include "uassert.h" +#include "uhash.h" + +U_CDECL_BEGIN + +static int32_t U_CALLCONV +hashStringTrieNode(const UHashTok key) { + return icu::StringTrieBuilder::hashNode(key.pointer); +} + +static UBool U_CALLCONV +equalStringTrieNodes(const UHashTok key1, const UHashTok key2) { + return icu::StringTrieBuilder::equalNodes(key1.pointer, key2.pointer); +} + +U_CDECL_END + +U_NAMESPACE_BEGIN + +StringTrieBuilder::StringTrieBuilder() : nodes(nullptr) {} + +StringTrieBuilder::~StringTrieBuilder() { + deleteCompactBuilder(); +} + +void +StringTrieBuilder::createCompactBuilder(int32_t sizeGuess, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return; + } + nodes=uhash_openSize(hashStringTrieNode, equalStringTrieNodes, nullptr, + sizeGuess, &errorCode); + if(U_SUCCESS(errorCode)) { + if(nodes==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } else { + uhash_setKeyDeleter(nodes, uprv_deleteUObject); + } + } +} + +void +StringTrieBuilder::deleteCompactBuilder() { + uhash_close(nodes); + nodes=nullptr; +} + +void +StringTrieBuilder::build(UStringTrieBuildOption buildOption, int32_t elementsLength, + UErrorCode &errorCode) { + if(buildOption==USTRINGTRIE_BUILD_FAST) { + writeNode(0, elementsLength, 0); + } else /* USTRINGTRIE_BUILD_SMALL */ { + createCompactBuilder(2*elementsLength, errorCode); + Node *root=makeNode(0, elementsLength, 0, errorCode); + if(U_SUCCESS(errorCode)) { + root->markRightEdgesFirst(-1); + root->write(*this); + } + deleteCompactBuilder(); + } +} + +// Requires startmaxLinearMatchLength) { + lastUnitIndex-=maxLinearMatchLength; + length-=maxLinearMatchLength; + writeElementUnits(start, lastUnitIndex, maxLinearMatchLength); + write(getMinLinearMatch()+maxLinearMatchLength-1); + } + writeElementUnits(start, unitIndex, length); + type=getMinLinearMatch()+length-1; + } else { + // Branch node. + int32_t length=countElementUnits(start, limit, unitIndex); + // length>=2 because minUnit!=maxUnit. + writeBranchSubNode(start, limit, unitIndex, length); + if(--lengthgetMaxBranchLinearSubNodeLength()) { + // Branch on the middle unit. + // First, find the middle unit. + int32_t i=skipElementsBySomeUnits(start, unitIndex, length/2); + // Encode the less-than branch first. + middleUnits[ltLength]=getElementUnit(i, unitIndex); // middle unit + lessThan[ltLength]=writeBranchSubNode(start, i, unitIndex, length/2); + ++ltLength; + // Continue for the greater-or-equal branch. + start=i; + length=length-length/2; + } + // For each unit, find its elements array start and whether it has a final value. + int32_t starts[kMaxBranchLinearSubNodeLength]; + UBool isFinal[kMaxBranchLinearSubNodeLength-1]; + int32_t unitNumber=0; + do { + int32_t i=starts[unitNumber]=start; + char16_t unit=getElementUnit(i++, unitIndex); + i=indexOfElementWithNextUnit(i, unitIndex, unit); + isFinal[unitNumber]= start==i-1 && unitIndex+1==getElementStringLength(start); + start=i; + } while(++unitNumber0); + // The maxUnit sub-node is written as the very last one because we do + // not jump for it at all. + unitNumber=length-1; + writeNode(start, limit, unitIndex+1); + int32_t offset=write(getElementUnit(start, unitIndex)); + // Write the rest of this node's unit-value pairs. + while(--unitNumber>=0) { + start=starts[unitNumber]; + int32_t value; + if(isFinal[unitNumber]) { + // Write the final value for the one string ending with this unit. + value=getElementValue(start); + } else { + // Write the delta to the start position of the sub-node. + value=offset-jumpTargets[unitNumber]; + } + writeValueAndFinal(value, isFinal[unitNumber]); + offset=write(getElementUnit(start, unitIndex)); + } + // Write the split-branch nodes. + while(ltLength>0) { + --ltLength; + writeDeltaTo(lessThan[ltLength]); + offset=write(middleUnits[ltLength]); + } + return offset; +} + +// Requires startmaxLinearMatchLength) { + lastUnitIndex-=maxLinearMatchLength; + length-=maxLinearMatchLength; + node=createLinearMatchNode(start, lastUnitIndex, maxLinearMatchLength, nextNode); + nextNode=registerNode(node, errorCode); + } + node=createLinearMatchNode(start, unitIndex, length, nextNode); + } else { + // Branch node. + int32_t length=countElementUnits(start, limit, unitIndex); + // length>=2 because minUnit!=maxUnit. + Node *subNode=makeBranchSubNode(start, limit, unitIndex, length, errorCode); + node=new BranchHeadNode(length, subNode); + } + if(hasValue && node!=nullptr) { + if(matchNodesCanHaveValues()) { + ((ValueNode *)node)->setValue(value); + } else { + node=new IntermediateValueNode(value, registerNode(node, errorCode)); + } + } + return registerNode(node, errorCode); +} + +// startgetMaxBranchLinearSubNodeLength()) { + // Branch on the middle unit. + // First, find the middle unit. + int32_t i=skipElementsBySomeUnits(start, unitIndex, length/2); + // Create the less-than branch. + middleUnits[ltLength]=getElementUnit(i, unitIndex); // middle unit + lessThan[ltLength]=makeBranchSubNode(start, i, unitIndex, length/2, errorCode); + ++ltLength; + // Continue for the greater-or-equal branch. + start=i; + length=length-length/2; + } + if(U_FAILURE(errorCode)) { + return nullptr; + } + ListBranchNode *listNode=new ListBranchNode(); + if(listNode==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + // For each unit, find its elements array start and whether it has a final value. + int32_t unitNumber=0; + do { + int32_t i=start; + char16_t unit=getElementUnit(i++, unitIndex); + i=indexOfElementWithNextUnit(i, unitIndex, unit); + if(start==i-1 && unitIndex+1==getElementStringLength(start)) { + listNode->add(unit, getElementValue(start)); + } else { + listNode->add(unit, makeNode(start, i, unitIndex+1, errorCode)); + } + start=i; + } while(++unitNumberadd(unit, getElementValue(start)); + } else { + listNode->add(unit, makeNode(start, limit, unitIndex+1, errorCode)); + } + Node *node=registerNode(listNode, errorCode); + // Create the split-branch nodes. + while(ltLength>0) { + --ltLength; + node=registerNode( + new SplitBranchNode(middleUnits[ltLength], lessThan[ltLength], node), errorCode); + } + return node; +} + +StringTrieBuilder::Node * +StringTrieBuilder::registerNode(Node *newNode, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + delete newNode; + return nullptr; + } + if(newNode==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + const UHashElement *old=uhash_find(nodes, newNode); + if(old!=nullptr) { + delete newNode; + return (Node *)old->key.pointer; + } + // If uhash_puti() returns a non-zero value from an equivalent, previously + // registered node, then uhash_find() failed to find that and we will leak newNode. +#if U_DEBUG + int32_t oldValue= // Only in debug mode to avoid a compiler warning about unused oldValue. +#endif + uhash_puti(nodes, newNode, 1, &errorCode); + U_ASSERT(oldValue==0); + if(U_FAILURE(errorCode)) { + delete newNode; + return nullptr; + } + return newNode; +} + +StringTrieBuilder::Node * +StringTrieBuilder::registerFinalValue(int32_t value, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return nullptr; + } + FinalValueNode key(value); + const UHashElement *old=uhash_find(nodes, &key); + if(old!=nullptr) { + return (Node *)old->key.pointer; + } + Node *newNode=new FinalValueNode(value); + if(newNode==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + // If uhash_puti() returns a non-zero value from an equivalent, previously + // registered node, then uhash_find() failed to find that and we will leak newNode. +#if U_DEBUG + int32_t oldValue= // Only in debug mode to avoid a compiler warning about unused oldValue. +#endif + uhash_puti(nodes, newNode, 1, &errorCode); + U_ASSERT(oldValue==0); + if(U_FAILURE(errorCode)) { + delete newNode; + return nullptr; + } + return newNode; +} + +int32_t +StringTrieBuilder::hashNode(const void *node) { + return ((const Node *)node)->hashCode(); +} + +UBool +StringTrieBuilder::equalNodes(const void *left, const void *right) { + return *(const Node *)left==*(const Node *)right; +} + +bool +StringTrieBuilder::Node::operator==(const Node &other) const { + return this==&other || (typeid(*this)==typeid(other) && hash==other.hash); +} + +int32_t +StringTrieBuilder::Node::markRightEdgesFirst(int32_t edgeNumber) { + if(offset==0) { + offset=edgeNumber; + } + return edgeNumber; +} + +bool +StringTrieBuilder::FinalValueNode::operator==(const Node &other) const { + if(this==&other) { + return true; + } + if(!Node::operator==(other)) { + return false; + } + const FinalValueNode &o=static_cast(other); + return value==o.value; +} + +void +StringTrieBuilder::FinalValueNode::write(StringTrieBuilder &builder) { + offset=builder.writeValueAndFinal(value, true); +} + +bool +StringTrieBuilder::ValueNode::operator==(const Node &other) const { + if(this==&other) { + return true; + } + if(!Node::operator==(other)) { + return false; + } + const ValueNode &o=static_cast(other); + return hasValue==o.hasValue && (!hasValue || value==o.value); +} + +bool +StringTrieBuilder::IntermediateValueNode::operator==(const Node &other) const { + if(this==&other) { + return true; + } + if(!ValueNode::operator==(other)) { + return false; + } + const IntermediateValueNode &o=static_cast(other); + return next==o.next; +} + +int32_t +StringTrieBuilder::IntermediateValueNode::markRightEdgesFirst(int32_t edgeNumber) { + if(offset==0) { + offset=edgeNumber=next->markRightEdgesFirst(edgeNumber); + } + return edgeNumber; +} + +void +StringTrieBuilder::IntermediateValueNode::write(StringTrieBuilder &builder) { + next->write(builder); + offset=builder.writeValueAndFinal(value, false); +} + +bool +StringTrieBuilder::LinearMatchNode::operator==(const Node &other) const { + if(this==&other) { + return true; + } + if(!ValueNode::operator==(other)) { + return false; + } + const LinearMatchNode &o=static_cast(other); + return length==o.length && next==o.next; +} + +int32_t +StringTrieBuilder::LinearMatchNode::markRightEdgesFirst(int32_t edgeNumber) { + if(offset==0) { + offset=edgeNumber=next->markRightEdgesFirst(edgeNumber); + } + return edgeNumber; +} + +bool +StringTrieBuilder::ListBranchNode::operator==(const Node &other) const { + if(this==&other) { + return true; + } + if(!Node::operator==(other)) { + return false; + } + const ListBranchNode &o=static_cast(other); + for(int32_t i=0; imarkRightEdgesFirst(edgeNumber-step); + } + // For all but the rightmost edge, decrement the edge number. + step=1; + } while(i>0); + offset=edgeNumber; + } + return edgeNumber; +} + +void +StringTrieBuilder::ListBranchNode::write(StringTrieBuilder &builder) { + // Write the sub-nodes in reverse order: The jump lengths are deltas from + // after their own positions, so if we wrote the minUnit sub-node first, + // then its jump delta would be larger. + // Instead we write the minUnit sub-node last, for a shorter delta. + int32_t unitNumber=length-1; + Node *rightEdge=equal[unitNumber]; + int32_t rightEdgeNumber= rightEdge==nullptr ? firstEdgeNumber : rightEdge->getOffset(); + do { + --unitNumber; + if(equal[unitNumber]!=nullptr) { + equal[unitNumber]->writeUnlessInsideRightEdge(firstEdgeNumber, rightEdgeNumber, builder); + } + } while(unitNumber>0); + // The maxUnit sub-node is written as the very last one because we do + // not jump for it at all. + unitNumber=length-1; + if(rightEdge==nullptr) { + builder.writeValueAndFinal(values[unitNumber], true); + } else { + rightEdge->write(builder); + } + offset=builder.write(units[unitNumber]); + // Write the rest of this node's unit-value pairs. + while(--unitNumber>=0) { + int32_t value; + UBool isFinal; + if(equal[unitNumber]==nullptr) { + // Write the final value for the one string ending with this unit. + value=values[unitNumber]; + isFinal=true; + } else { + // Write the delta to the start position of the sub-node. + U_ASSERT(equal[unitNumber]->getOffset()>0); + value=offset-equal[unitNumber]->getOffset(); + isFinal=false; + } + builder.writeValueAndFinal(value, isFinal); + offset=builder.write(units[unitNumber]); + } +} + +bool +StringTrieBuilder::SplitBranchNode::operator==(const Node &other) const { + if(this==&other) { + return true; + } + if(!Node::operator==(other)) { + return false; + } + const SplitBranchNode &o=static_cast(other); + return unit==o.unit && lessThan==o.lessThan && greaterOrEqual==o.greaterOrEqual; +} + +int32_t +StringTrieBuilder::SplitBranchNode::markRightEdgesFirst(int32_t edgeNumber) { + if(offset==0) { + firstEdgeNumber=edgeNumber; + edgeNumber=greaterOrEqual->markRightEdgesFirst(edgeNumber); + offset=edgeNumber=lessThan->markRightEdgesFirst(edgeNumber-1); + } + return edgeNumber; +} + +void +StringTrieBuilder::SplitBranchNode::write(StringTrieBuilder &builder) { + // Encode the less-than branch first. + lessThan->writeUnlessInsideRightEdge(firstEdgeNumber, greaterOrEqual->getOffset(), builder); + // Encode the greater-or-equal branch last because we do not jump for it at all. + greaterOrEqual->write(builder); + // Write this node. + U_ASSERT(lessThan->getOffset()>0); + builder.writeDeltaTo(lessThan->getOffset()); // less-than + offset=builder.write(unit); +} + +bool +StringTrieBuilder::BranchHeadNode::operator==(const Node &other) const { + if(this==&other) { + return true; + } + if(!ValueNode::operator==(other)) { + return false; + } + const BranchHeadNode &o=static_cast(other); + return length==o.length && next==o.next; +} + +int32_t +StringTrieBuilder::BranchHeadNode::markRightEdgesFirst(int32_t edgeNumber) { + if(offset==0) { + offset=edgeNumber=next->markRightEdgesFirst(edgeNumber); + } + return edgeNumber; +} + +void +StringTrieBuilder::BranchHeadNode::write(StringTrieBuilder &builder) { + next->write(builder); + if(length<=builder.getMinLinearMatch()) { + offset=builder.writeValueAndType(hasValue, value, length-1); + } else { + builder.write(length-1); + offset=builder.writeValueAndType(hasValue, value, 0); + } +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/uarrsort.cpp b/intl/icu/source/common/uarrsort.cpp new file mode 100644 index 0000000000..f9daa4b30f --- /dev/null +++ b/intl/icu/source/common/uarrsort.cpp @@ -0,0 +1,274 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2003-2013, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: uarrsort.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2003aug04 +* created by: Markus W. Scherer +* +* Internal function for sorting arrays. +*/ + +#include + +#include "unicode/utypes.h" +#include "cmemory.h" +#include "uarrsort.h" + +enum { + /** + * "from Knuth" + * + * A binary search over 8 items performs 4 comparisons: + * log2(8)=3 to subdivide, +1 to check for equality. + * A linear search over 8 items on average also performs 4 comparisons. + */ + MIN_QSORT=9, + STACK_ITEM_SIZE=200 +}; + +static constexpr int32_t sizeInMaxAlignTs(int32_t sizeInBytes) { + return (sizeInBytes + sizeof(std::max_align_t) - 1) / sizeof(std::max_align_t); +} + +/* UComparator convenience implementations ---------------------------------- */ + +U_CAPI int32_t U_EXPORT2 +uprv_uint16Comparator(const void *context, const void *left, const void *right) { + (void)context; + return (int32_t)*(const uint16_t *)left - (int32_t)*(const uint16_t *)right; +} + +U_CAPI int32_t U_EXPORT2 +uprv_int32Comparator(const void *context, const void *left, const void *right) { + (void)context; + return *(const int32_t *)left - *(const int32_t *)right; +} + +U_CAPI int32_t U_EXPORT2 +uprv_uint32Comparator(const void *context, const void *left, const void *right) { + (void)context; + uint32_t l=*(const uint32_t *)left, r=*(const uint32_t *)right; + + /* compare directly because (l-r) would overflow the int32_t result */ + if(lr */ { + return 1; + } +} + +/* Insertion sort using binary search --------------------------------------- */ + +U_CAPI int32_t U_EXPORT2 +uprv_stableBinarySearch(char *array, int32_t limit, void *item, int32_t itemSize, + UComparator *cmp, const void *context) { + int32_t start=0; + UBool found=false; + + /* Binary search until we get down to a tiny sub-array. */ + while((limit-start)>=MIN_QSORT) { + int32_t i=(start+limit)/2; + int32_t diff=cmp(context, item, array+i*itemSize); + if(diff==0) { + /* + * Found the item. We look for the *last* occurrence of such + * an item, for stable sorting. + * If we knew that there will be only few equal items, + * we could break now and enter the linear search. + * However, if there are many equal items, then it should be + * faster to continue with the binary search. + * It seems likely that we either have all unique items + * (where found will never become true in the insertion sort) + * or potentially many duplicates. + */ + found=true; + start=i+1; + } else if(diff<0) { + limit=i; + } else { + start=i; + } + } + + /* Linear search over the remaining tiny sub-array. */ + while(start v; + if (sizeInMaxAlignTs(itemSize) > v.getCapacity() && + v.resize(sizeInMaxAlignTs(itemSize)) == nullptr) { + *pErrorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + + doInsertionSort(array, length, itemSize, cmp, context, v.getAlias()); +} + +/* QuickSort ---------------------------------------------------------------- */ + +/* + * This implementation is semi-recursive: + * It recurses for the smaller sub-array to shorten the recursion depth, + * and loops for the larger sub-array. + * + * Loosely after QuickSort algorithms in + * Niklaus Wirth + * Algorithmen und Datenstrukturen mit Modula-2 + * B.G. Teubner Stuttgart + * 4. Auflage 1986 + * ISBN 3-519-02260-5 + */ +static void +subQuickSort(char *array, int32_t start, int32_t limit, int32_t itemSize, + UComparator *cmp, const void *context, + void *px, void *pw) { + int32_t left, right; + + /* start and left are inclusive, limit and right are exclusive */ + do { + if((start+MIN_QSORT)>=limit) { + doInsertionSort(array+start*itemSize, limit-start, itemSize, cmp, context, px); + break; + } + + left=start; + right=limit; + + /* x=array[middle] */ + uprv_memcpy(px, array+(size_t)((start+limit)/2)*itemSize, itemSize); + + do { + while(/* array[left] xw; + if(sizeInMaxAlignTs(itemSize)*2 > xw.getCapacity() && + xw.resize(sizeInMaxAlignTs(itemSize) * 2) == nullptr) { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + + subQuickSort(array, 0, length, itemSize, cmp, context, + xw.getAlias(), xw.getAlias() + sizeInMaxAlignTs(itemSize)); +} + +/* uprv_sortArray() API ----------------------------------------------------- */ + +/* + * Check arguments, select an appropriate implementation, + * cast the array to char * so that array+i*itemSize works. + */ +U_CAPI void U_EXPORT2 +uprv_sortArray(void *array, int32_t length, int32_t itemSize, + UComparator *cmp, const void *context, + UBool sortStable, UErrorCode *pErrorCode) { + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return; + } + if((length>0 && array==nullptr) || length<0 || itemSize<=0 || cmp==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + if(length<=1) { + return; + } else if(length0 if left>right + * + * @internal + */ +typedef int32_t U_CALLCONV +UComparator(const void *context, const void *left, const void *right); +U_CDECL_END + +/** + * Array sorting function. + * Uses a UComparator for comparing array items to each other, and simple + * memory copying to move items. + * + * @param array The array to be sorted. + * @param length The number of items in the array. + * @param itemSize The size in bytes of each array item. + * @param cmp UComparator function used to compare two items each. + * @param context Application-specific pointer, passed through to the UComparator. + * @param sortStable If true, a stable sorting algorithm must be used. + * @param pErrorCode ICU in/out UErrorCode parameter. + * + * @internal + */ +U_CAPI void U_EXPORT2 +uprv_sortArray(void *array, int32_t length, int32_t itemSize, + UComparator *cmp, const void *context, + UBool sortStable, UErrorCode *pErrorCode); + +/** + * Convenience UComparator implementation for uint16_t arrays. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +uprv_uint16Comparator(const void *context, const void *left, const void *right); + +/** + * Convenience UComparator implementation for int32_t arrays. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +uprv_int32Comparator(const void *context, const void *left, const void *right); + +/** + * Convenience UComparator implementation for uint32_t arrays. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +uprv_uint32Comparator(const void *context, const void *left, const void *right); + +/** + * Much like Java Collections.binarySearch(list, key, comparator). + * + * Except: Java documents "If the list contains multiple elements equal to + * the specified object, there is no guarantee which one will be found." + * + * This version here will return the largest index of any equal item, + * for use in stable sorting. + * + * @return the index>=0 where the item was found: + * the largest such index, if multiple, for stable sorting; + * or the index<0 for inserting the item at ~index in sorted order + */ +U_CAPI int32_t U_EXPORT2 +uprv_stableBinarySearch(char *array, int32_t length, void *item, int32_t itemSize, + UComparator *cmp, const void *context); + +#endif diff --git a/intl/icu/source/common/uassert.h b/intl/icu/source/common/uassert.h new file mode 100644 index 0000000000..5218752899 --- /dev/null +++ b/intl/icu/source/common/uassert.h @@ -0,0 +1,71 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2002-2011, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* File uassert.h +* +* Contains the U_ASSERT and UPRV_UNREACHABLE_* macros +* +****************************************************************************** +*/ +#ifndef U_ASSERT_H +#define U_ASSERT_H + +/* utypes.h is included to get the proper define for uint8_t */ +#include "unicode/utypes.h" +/* for abort */ +#include + +/** + * \def U_ASSERT + * By default, U_ASSERT just wraps the C library assert macro. + * By changing the definition here, the assert behavior for ICU can be changed + * without affecting other non - ICU uses of the C library assert(). +*/ +#if U_DEBUG +# include +# define U_ASSERT(exp) assert(exp) +#elif U_CPLUSPLUS_VERSION +# define U_ASSERT(exp) (void)0 +#else +# define U_ASSERT(exp) +#endif + +/** + * \def UPRV_UNREACHABLE_ASSERT + * This macro is used in places that we had believed were unreachable, but + * experience has shown otherwise (possibly due to memory corruption, etc). + * In this case we call assert() in debug versions as with U_ASSERT, instead + * of unconditionally calling abort(). However we also allow redefinition as + * with UPRV_UNREACHABLE_EXIT. + * @internal +*/ +#if defined(UPRV_UNREACHABLE_ASSERT) + // Use the predefined value. +#elif U_DEBUG +# include +# define UPRV_UNREACHABLE_ASSERT assert(false) +#elif U_CPLUSPLUS_VERSION +# define UPRV_UNREACHABLE_ASSERT (void)0 +#else +# define UPRV_UNREACHABLE_ASSERT +#endif + +/** + * \def UPRV_UNREACHABLE_EXIT + * This macro is used to unconditionally abort if unreachable code is ever executed. + * @internal +*/ +#if defined(UPRV_UNREACHABLE_EXIT) + // Use the predefined value. +#else +# define UPRV_UNREACHABLE_EXIT abort() +#endif + +#endif diff --git a/intl/icu/source/common/ubidi.cpp b/intl/icu/source/common/ubidi.cpp new file mode 100644 index 0000000000..fcf82fa97a --- /dev/null +++ b/intl/icu/source/common/ubidi.cpp @@ -0,0 +1,3036 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2015, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: ubidi.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 1999jul27 +* created by: Markus W. Scherer, updated by Matitiahu Allouche +* +*/ + +#include "cmemory.h" +#include "unicode/utypes.h" +#include "unicode/ustring.h" +#include "unicode/uchar.h" +#include "unicode/ubidi.h" +#include "unicode/utf16.h" +#include "ubidi_props.h" +#include "ubidiimp.h" +#include "uassert.h" + +/* + * General implementation notes: + * + * Throughout the implementation, there are comments like (W2) that refer to + * rules of the BiDi algorithm, in this example to the second rule of the + * resolution of weak types. + * + * For handling surrogate pairs, where two char16_t's form one "abstract" (or UTF-32) + * character according to UTF-16, the second char16_t gets the directional property of + * the entire character assigned, while the first one gets a BN, a boundary + * neutral, type, which is ignored by most of the algorithm according to + * rule (X9) and the implementation suggestions of the BiDi algorithm. + * + * Later, adjustWSLevels() will set the level for each BN to that of the + * following character (char16_t), which results in surrogate pairs getting the + * same level on each of their surrogates. + * + * In a UTF-8 implementation, the same thing could be done: the last byte of + * a multi-byte sequence would get the "real" property, while all previous + * bytes of that sequence would get BN. + * + * It is not possible to assign all those parts of a character the same real + * property because this would fail in the resolution of weak types with rules + * that look at immediately surrounding types. + * + * As a related topic, this implementation does not remove Boundary Neutral + * types from the input, but ignores them wherever this is relevant. + * For example, the loop for the resolution of the weak types reads + * types until it finds a non-BN. + * Also, explicit embedding codes are neither changed into BN nor removed. + * They are only treated the same way real BNs are. + * As stated before, adjustWSLevels() takes care of them at the end. + * For the purpose of conformance, the levels of all these codes + * do not matter. + * + * Note that this implementation modifies the dirProps + * after the initial setup, when applying X5c (replace FSI by LRI or RLI), + * X6, N0 (replace paired brackets by L or R). + * + * In this implementation, the resolution of weak types (W1 to W6), + * neutrals (N1 and N2), and the assignment of the resolved level (In) + * are all done in one single loop, in resolveImplicitLevels(). + * Changes of dirProp values are done on the fly, without writing + * them back to the dirProps array. + * + * + * This implementation contains code that allows to bypass steps of the + * algorithm that are not needed on the specific paragraph + * in order to speed up the most common cases considerably, + * like text that is entirely LTR, or RTL text without numbers. + * + * Most of this is done by setting a bit for each directional property + * in a flags variable and later checking for whether there are + * any LTR characters or any RTL characters, or both, whether + * there are any explicit embedding codes, etc. + * + * If the (Xn) steps are performed, then the flags are re-evaluated, + * because they will then not contain the embedding codes any more + * and will be adjusted for override codes, so that subsequently + * more bypassing may be possible than what the initial flags suggested. + * + * If the text is not mixed-directional, then the + * algorithm steps for the weak type resolution are not performed, + * and all levels are set to the paragraph level. + * + * If there are no explicit embedding codes, then the (Xn) steps + * are not performed. + * + * If embedding levels are supplied as a parameter, then all + * explicit embedding codes are ignored, and the (Xn) steps + * are not performed. + * + * White Space types could get the level of the run they belong to, + * and are checked with a test of (flags&MASK_EMBEDDING) to + * consider if the paragraph direction should be considered in + * the flags variable. + * + * If there are no White Space types in the paragraph, then + * (L1) is not necessary in adjustWSLevels(). + */ + +/* to avoid some conditional statements, use tiny constant arrays */ +static const Flags flagLR[2]={ DIRPROP_FLAG(L), DIRPROP_FLAG(R) }; +static const Flags flagE[2]={ DIRPROP_FLAG(LRE), DIRPROP_FLAG(RLE) }; +static const Flags flagO[2]={ DIRPROP_FLAG(LRO), DIRPROP_FLAG(RLO) }; + +#define DIRPROP_FLAG_LR(level) flagLR[(level)&1] +#define DIRPROP_FLAG_E(level) flagE[(level)&1] +#define DIRPROP_FLAG_O(level) flagO[(level)&1] + +#define DIR_FROM_STRONG(strong) ((strong)==L ? L : R) + +#define NO_OVERRIDE(level) ((level)&~UBIDI_LEVEL_OVERRIDE) + +/* UBiDi object management -------------------------------------------------- */ + +U_CAPI UBiDi * U_EXPORT2 +ubidi_open() +{ + UErrorCode errorCode=U_ZERO_ERROR; + return ubidi_openSized(0, 0, &errorCode); +} + +U_CAPI UBiDi * U_EXPORT2 +ubidi_openSized(int32_t maxLength, int32_t maxRunCount, UErrorCode *pErrorCode) { + UBiDi *pBiDi; + + /* check the argument values */ + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return nullptr; + } else if(maxLength<0 || maxRunCount<0) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; /* invalid arguments */ + } + + /* allocate memory for the object */ + pBiDi=(UBiDi *)uprv_malloc(sizeof(UBiDi)); + if(pBiDi==nullptr) { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + + /* reset the object, all pointers nullptr, all flags false, all sizes 0 */ + uprv_memset(pBiDi, 0, sizeof(UBiDi)); + + /* allocate memory for arrays as requested */ + if(maxLength>0) { + if( !getInitialDirPropsMemory(pBiDi, maxLength) || + !getInitialLevelsMemory(pBiDi, maxLength) + ) { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + } + } else { + pBiDi->mayAllocateText=true; + } + + if(maxRunCount>0) { + if(maxRunCount==1) { + /* use simpleRuns[] */ + pBiDi->runsSize=sizeof(Run); + } else if(!getInitialRunsMemory(pBiDi, maxRunCount)) { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + } + } else { + pBiDi->mayAllocateRuns=true; + } + + if(U_SUCCESS(*pErrorCode)) { + return pBiDi; + } else { + ubidi_close(pBiDi); + return nullptr; + } +} + +/* + * We are allowed to allocate memory if memory==nullptr or + * mayAllocate==true for each array that we need. + * We also try to grow memory as needed if we + * allocate it. + * + * Assume sizeNeeded>0. + * If *pMemory!=nullptr, then assume *pSize>0. + * + * ### this realloc() may unnecessarily copy the old data, + * which we know we don't need any more; + * is this the best way to do this?? + */ +U_CFUNC UBool +ubidi_getMemory(BidiMemoryForAllocation *bidiMem, int32_t *pSize, UBool mayAllocate, int32_t sizeNeeded) { + void **pMemory = (void **)bidiMem; + /* check for existing memory */ + if(*pMemory==nullptr) { + /* we need to allocate memory */ + if(mayAllocate && (*pMemory=uprv_malloc(sizeNeeded))!=nullptr) { + *pSize=sizeNeeded; + return true; + } else { + return false; + } + } else { + if(sizeNeeded<=*pSize) { + /* there is already enough memory */ + return true; + } + else if(!mayAllocate) { + /* not enough memory, and we must not allocate */ + return false; + } else { + /* we try to grow */ + void *memory; + /* in most cases, we do not need the copy-old-data part of + * realloc, but it is needed when adding runs using getRunsMemory() + * in setParaRunsOnly() + */ + if((memory=uprv_realloc(*pMemory, sizeNeeded))!=nullptr) { + *pMemory=memory; + *pSize=sizeNeeded; + return true; + } else { + /* we failed to grow */ + return false; + } + } + } +} + +U_CAPI void U_EXPORT2 +ubidi_close(UBiDi *pBiDi) { + if(pBiDi!=nullptr) { + pBiDi->pParaBiDi=nullptr; /* in case one tries to reuse this block */ + if(pBiDi->dirPropsMemory!=nullptr) { + uprv_free(pBiDi->dirPropsMemory); + } + if(pBiDi->levelsMemory!=nullptr) { + uprv_free(pBiDi->levelsMemory); + } + if(pBiDi->openingsMemory!=nullptr) { + uprv_free(pBiDi->openingsMemory); + } + if(pBiDi->parasMemory!=nullptr) { + uprv_free(pBiDi->parasMemory); + } + if(pBiDi->runsMemory!=nullptr) { + uprv_free(pBiDi->runsMemory); + } + if(pBiDi->isolatesMemory!=nullptr) { + uprv_free(pBiDi->isolatesMemory); + } + if(pBiDi->insertPoints.points!=nullptr) { + uprv_free(pBiDi->insertPoints.points); + } + + uprv_free(pBiDi); + } +} + +/* set to approximate "inverse BiDi" ---------------------------------------- */ + +U_CAPI void U_EXPORT2 +ubidi_setInverse(UBiDi *pBiDi, UBool isInverse) { + if(pBiDi!=nullptr) { + pBiDi->isInverse=isInverse; + pBiDi->reorderingMode = isInverse ? UBIDI_REORDER_INVERSE_NUMBERS_AS_L + : UBIDI_REORDER_DEFAULT; + } +} + +U_CAPI UBool U_EXPORT2 +ubidi_isInverse(UBiDi *pBiDi) { + if(pBiDi!=nullptr) { + return pBiDi->isInverse; + } else { + return false; + } +} + +/* FOOD FOR THOUGHT: currently the reordering modes are a mixture of + * algorithm for direct BiDi, algorithm for inverse BiDi and the bizarre + * concept of RUNS_ONLY which is a double operation. + * It could be advantageous to divide this into 3 concepts: + * a) Operation: direct / inverse / RUNS_ONLY + * b) Direct algorithm: default / NUMBERS_SPECIAL / GROUP_NUMBERS_WITH_R + * c) Inverse algorithm: default / INVERSE_LIKE_DIRECT / NUMBERS_SPECIAL + * This would allow combinations not possible today like RUNS_ONLY with + * NUMBERS_SPECIAL. + * Also allow to set INSERT_MARKS for the direct step of RUNS_ONLY and + * REMOVE_CONTROLS for the inverse step. + * Not all combinations would be supported, and probably not all do make sense. + * This would need to document which ones are supported and what are the + * fallbacks for unsupported combinations. + */ +U_CAPI void U_EXPORT2 +ubidi_setReorderingMode(UBiDi *pBiDi, UBiDiReorderingMode reorderingMode) UPRV_NO_SANITIZE_UNDEFINED { + if ((pBiDi!=nullptr) && (reorderingMode >= UBIDI_REORDER_DEFAULT) + && (reorderingMode < UBIDI_REORDER_COUNT)) { + pBiDi->reorderingMode = reorderingMode; + pBiDi->isInverse = (UBool)(reorderingMode == UBIDI_REORDER_INVERSE_NUMBERS_AS_L); + } +} + +U_CAPI UBiDiReorderingMode U_EXPORT2 +ubidi_getReorderingMode(UBiDi *pBiDi) { + if (pBiDi!=nullptr) { + return pBiDi->reorderingMode; + } else { + return UBIDI_REORDER_DEFAULT; + } +} + +U_CAPI void U_EXPORT2 +ubidi_setReorderingOptions(UBiDi *pBiDi, uint32_t reorderingOptions) { + if (reorderingOptions & UBIDI_OPTION_REMOVE_CONTROLS) { + reorderingOptions&=~UBIDI_OPTION_INSERT_MARKS; + } + if (pBiDi!=nullptr) { + pBiDi->reorderingOptions=reorderingOptions; + } +} + +U_CAPI uint32_t U_EXPORT2 +ubidi_getReorderingOptions(UBiDi *pBiDi) { + if (pBiDi!=nullptr) { + return pBiDi->reorderingOptions; + } else { + return 0; + } +} + +U_CAPI UBiDiDirection U_EXPORT2 +ubidi_getBaseDirection(const char16_t *text, +int32_t length){ + + int32_t i; + UChar32 uchar; + UCharDirection dir; + + if( text==nullptr || length<-1 ){ + return UBIDI_NEUTRAL; + } + + if(length==-1) { + length=u_strlen(text); + } + + for( i = 0 ; i < length; ) { + /* i is incremented by U16_NEXT */ + U16_NEXT(text, i, length, uchar); + dir = u_charDirection(uchar); + if( dir == U_LEFT_TO_RIGHT ) + return UBIDI_LTR; + if( dir == U_RIGHT_TO_LEFT || dir ==U_RIGHT_TO_LEFT_ARABIC ) + return UBIDI_RTL; + } + return UBIDI_NEUTRAL; +} + +/* perform (P2)..(P3) ------------------------------------------------------- */ + +/** + * Returns the directionality of the first strong character + * after the last B in prologue, if any. + * Requires prologue!=null. + */ +static DirProp +firstL_R_AL(UBiDi *pBiDi) { + const char16_t *text=pBiDi->prologue; + int32_t length=pBiDi->proLength; + int32_t i; + UChar32 uchar; + DirProp dirProp, result=ON; + for(i=0; iparas + */ +static UBool +checkParaCount(UBiDi *pBiDi) { + int32_t count=pBiDi->paraCount; + if(pBiDi->paras==pBiDi->simpleParas) { + if(count<=SIMPLE_PARAS_COUNT) + return true; + if(!getInitialParasMemory(pBiDi, SIMPLE_PARAS_COUNT * 2)) + return false; + pBiDi->paras=pBiDi->parasMemory; + uprv_memcpy(pBiDi->parasMemory, pBiDi->simpleParas, SIMPLE_PARAS_COUNT * sizeof(Para)); + return true; + } + if(!getInitialParasMemory(pBiDi, count * 2)) + return false; + pBiDi->paras=pBiDi->parasMemory; + return true; +} + +/* + * Get the directional properties for the text, calculate the flags bit-set, and + * determine the paragraph level if necessary (in pBiDi->paras[i].level). + * FSI initiators are also resolved and their dirProp replaced with LRI or RLI. + * When encountering an FSI, it is initially replaced with an LRI, which is the + * default. Only if a strong R or AL is found within its scope will the LRI be + * replaced by an RLI. + */ +static UBool +getDirProps(UBiDi *pBiDi) { + const char16_t *text=pBiDi->text; + DirProp *dirProps=pBiDi->dirPropsMemory; /* pBiDi->dirProps is const */ + + int32_t i=0, originalLength=pBiDi->originalLength; + Flags flags=0; /* collect all directionalities in the text */ + UChar32 uchar; + DirProp dirProp=0, defaultParaLevel=0; /* initialize to avoid compiler warnings */ + UBool isDefaultLevel=IS_DEFAULT_LEVEL(pBiDi->paraLevel); + /* for inverse BiDi, the default para level is set to RTL if there is a + strong R or AL character at either end of the text */ + UBool isDefaultLevelInverse=isDefaultLevel && (UBool) + (pBiDi->reorderingMode==UBIDI_REORDER_INVERSE_LIKE_DIRECT || + pBiDi->reorderingMode==UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL); + int32_t lastArabicPos=-1; + int32_t controlCount=0; + UBool removeBiDiControls = (UBool)(pBiDi->reorderingOptions & + UBIDI_OPTION_REMOVE_CONTROLS); + + enum State { + NOT_SEEKING_STRONG, /* 0: not contextual paraLevel, not after FSI */ + SEEKING_STRONG_FOR_PARA, /* 1: looking for first strong char in para */ + SEEKING_STRONG_FOR_FSI, /* 2: looking for first strong after FSI */ + LOOKING_FOR_PDI /* 3: found strong after FSI, looking for PDI */ + }; + State state; + DirProp lastStrong=ON; /* for default level & inverse BiDi */ + /* The following stacks are used to manage isolate sequences. Those + sequences may be nested, but obviously never more deeply than the + maximum explicit embedding level. + lastStack is the index of the last used entry in the stack. A value of -1 + means that there is no open isolate sequence. + lastStack is reset to -1 on paragraph boundaries. */ + /* The following stack contains the position of the initiator of + each open isolate sequence */ + int32_t isolateStartStack[UBIDI_MAX_EXPLICIT_LEVEL+1]; + /* The following stack contains the last known state before + encountering the initiator of an isolate sequence */ + State previousStateStack[UBIDI_MAX_EXPLICIT_LEVEL+1]; + int32_t stackLast=-1; + + if(pBiDi->reorderingOptions & UBIDI_OPTION_STREAMING) + pBiDi->length=0; + defaultParaLevel=pBiDi->paraLevel&1; + if(isDefaultLevel) { + pBiDi->paras[0].level=defaultParaLevel; + lastStrong=defaultParaLevel; + if(pBiDi->proLength>0 && /* there is a prologue */ + (dirProp=firstL_R_AL(pBiDi))!=ON) { /* with a strong character */ + if(dirProp==L) + pBiDi->paras[0].level=0; /* set the default para level */ + else + pBiDi->paras[0].level=1; /* set the default para level */ + state=NOT_SEEKING_STRONG; + } else { + state=SEEKING_STRONG_FOR_PARA; + } + } else { + pBiDi->paras[0].level=pBiDi->paraLevel; + state=NOT_SEEKING_STRONG; + } + /* count paragraphs and determine the paragraph level (P2..P3) */ + /* + * see comment in ubidi.h: + * the UBIDI_DEFAULT_XXX values are designed so that + * their bit 0 alone yields the intended default + */ + for( /* i=0 above */ ; i0xffff) { /* set the lead surrogate's property to BN */ + flags|=DIRPROP_FLAG(BN); + dirProps[i-2]=BN; + } + if(removeBiDiControls && IS_BIDI_CONTROL_CHAR(uchar)) + controlCount++; + if(dirProp==L) { + if(state==SEEKING_STRONG_FOR_PARA) { + pBiDi->paras[pBiDi->paraCount-1].level=0; + state=NOT_SEEKING_STRONG; + } + else if(state==SEEKING_STRONG_FOR_FSI) { + if(stackLast<=UBIDI_MAX_EXPLICIT_LEVEL) { + /* no need for next statement, already set by default */ + /* dirProps[isolateStartStack[stackLast]]=LRI; */ + flags|=DIRPROP_FLAG(LRI); + } + state=LOOKING_FOR_PDI; + } + lastStrong=L; + continue; + } + if(dirProp==R || dirProp==AL) { + if(state==SEEKING_STRONG_FOR_PARA) { + pBiDi->paras[pBiDi->paraCount-1].level=1; + state=NOT_SEEKING_STRONG; + } + else if(state==SEEKING_STRONG_FOR_FSI) { + if(stackLast<=UBIDI_MAX_EXPLICIT_LEVEL) { + dirProps[isolateStartStack[stackLast]]=RLI; + flags|=DIRPROP_FLAG(RLI); + } + state=LOOKING_FOR_PDI; + } + lastStrong=R; + if(dirProp==AL) + lastArabicPos=i-1; + continue; + } + if(dirProp>=FSI && dirProp<=RLI) { /* FSI, LRI or RLI */ + stackLast++; + if(stackLast<=UBIDI_MAX_EXPLICIT_LEVEL) { + isolateStartStack[stackLast]=i-1; + previousStateStack[stackLast]=state; + } + if(dirProp==FSI) { + dirProps[i-1]=LRI; /* default if no strong char */ + state=SEEKING_STRONG_FOR_FSI; + } + else + state=LOOKING_FOR_PDI; + continue; + } + if(dirProp==PDI) { + if(state==SEEKING_STRONG_FOR_FSI) { + if(stackLast<=UBIDI_MAX_EXPLICIT_LEVEL) { + /* no need for next statement, already set by default */ + /* dirProps[isolateStartStack[stackLast]]=LRI; */ + flags|=DIRPROP_FLAG(LRI); + } + } + if(stackLast>=0) { + if(stackLast<=UBIDI_MAX_EXPLICIT_LEVEL) + state=previousStateStack[stackLast]; + stackLast--; + } + continue; + } + if(dirProp==B) { + if(iparas[pBiDi->paraCount-1].limit=i; + if(isDefaultLevelInverse && lastStrong==R) + pBiDi->paras[pBiDi->paraCount-1].level=1; + if(pBiDi->reorderingOptions & UBIDI_OPTION_STREAMING) { + /* When streaming, we only process whole paragraphs + thus some updates are only done on paragraph boundaries */ + pBiDi->length=i; /* i is index to next character */ + pBiDi->controlCount=controlCount; + } + if(iparaCount++; + if(checkParaCount(pBiDi)==false) /* not enough memory for a new para entry */ + return false; + if(isDefaultLevel) { + pBiDi->paras[pBiDi->paraCount-1].level=defaultParaLevel; + state=SEEKING_STRONG_FOR_PARA; + lastStrong=defaultParaLevel; + } else { + pBiDi->paras[pBiDi->paraCount-1].level=pBiDi->paraLevel; + state=NOT_SEEKING_STRONG; + } + stackLast=-1; + } + continue; + } + } + /* Ignore still open isolate sequences with overflow */ + if(stackLast>UBIDI_MAX_EXPLICIT_LEVEL) { + stackLast=UBIDI_MAX_EXPLICIT_LEVEL; + state=SEEKING_STRONG_FOR_FSI; /* to be on the safe side */ + } + /* Resolve direction of still unresolved open FSI sequences */ + while(stackLast>=0) { + if(state==SEEKING_STRONG_FOR_FSI) { + /* no need for next statement, already set by default */ + /* dirProps[isolateStartStack[stackLast]]=LRI; */ + flags|=DIRPROP_FLAG(LRI); + break; + } + state=previousStateStack[stackLast]; + stackLast--; + } + /* When streaming, ignore text after the last paragraph separator */ + if(pBiDi->reorderingOptions & UBIDI_OPTION_STREAMING) { + if(pBiDi->lengthparaCount--; + } else { + pBiDi->paras[pBiDi->paraCount-1].limit=originalLength; + pBiDi->controlCount=controlCount; + } + /* For inverse bidi, default para direction is RTL if there is + a strong R or AL at either end of the paragraph */ + if(isDefaultLevelInverse && lastStrong==R) { + pBiDi->paras[pBiDi->paraCount-1].level=1; + } + if(isDefaultLevel) { + pBiDi->paraLevel=static_cast(pBiDi->paras[0].level); + } + /* The following is needed to resolve the text direction for default level + paragraphs containing no strong character */ + for(i=0; iparaCount; i++) + flags|=DIRPROP_FLAG_LR(pBiDi->paras[i].level); + + if(pBiDi->orderParagraphsLTR && (flags&DIRPROP_FLAG(B))) { + flags|=DIRPROP_FLAG(L); + } + pBiDi->flags=flags; + pBiDi->lastArabicPos=lastArabicPos; + return true; +} + +/* determine the paragraph level at position index */ +U_CFUNC UBiDiLevel +ubidi_getParaLevelAtIndex(const UBiDi *pBiDi, int32_t pindex) { + int32_t i; + for(i=0; iparaCount; i++) + if(pindexparas[i].limit) + break; + if(i>=pBiDi->paraCount) + i=pBiDi->paraCount-1; + return (UBiDiLevel)(pBiDi->paras[i].level); +} + +/* Functions for handling paired brackets ----------------------------------- */ + +/* In the isoRuns array, the first entry is used for text outside of any + isolate sequence. Higher entries are used for each more deeply nested + isolate sequence. isoRunLast is the index of the last used entry. The + openings array is used to note the data of opening brackets not yet + matched by a closing bracket, or matched but still susceptible to change + level. + Each isoRun entry contains the index of the first and + one-after-last openings entries for pending opening brackets it + contains. The next openings entry to use is the one-after-last of the + most deeply nested isoRun entry. + isoRun entries also contain their current embedding level and the last + encountered strong character, since these will be needed to resolve + the level of paired brackets. */ + +static void +bracketInit(UBiDi *pBiDi, BracketData *bd) { + bd->pBiDi=pBiDi; + bd->isoRunLast=0; + bd->isoRuns[0].start=0; + bd->isoRuns[0].limit=0; + bd->isoRuns[0].level=GET_PARALEVEL(pBiDi, 0); + UBiDiLevel t = GET_PARALEVEL(pBiDi, 0) & 1; + bd->isoRuns[0].lastStrong = bd->isoRuns[0].lastBase = t; + bd->isoRuns[0].contextDir = (UBiDiDirection)t; + bd->isoRuns[0].contextPos=0; + if(pBiDi->openingsMemory) { + bd->openings=pBiDi->openingsMemory; + bd->openingsCount=pBiDi->openingsSize / sizeof(Opening); + } else { + bd->openings=bd->simpleOpenings; + bd->openingsCount=SIMPLE_OPENINGS_COUNT; + } + bd->isNumbersSpecial=bd->pBiDi->reorderingMode==UBIDI_REORDER_NUMBERS_SPECIAL || + bd->pBiDi->reorderingMode==UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL; +} + +/* paragraph boundary */ +static void +bracketProcessB(BracketData *bd, UBiDiLevel level) { + bd->isoRunLast=0; + bd->isoRuns[0].limit=0; + bd->isoRuns[0].level=level; + bd->isoRuns[0].lastStrong=bd->isoRuns[0].lastBase=level&1; + bd->isoRuns[0].contextDir=(UBiDiDirection)(level&1); + bd->isoRuns[0].contextPos=0; +} + +/* LRE, LRO, RLE, RLO, PDF */ +static void +bracketProcessBoundary(BracketData *bd, int32_t lastCcPos, + UBiDiLevel contextLevel, UBiDiLevel embeddingLevel) { + IsoRun *pLastIsoRun=&bd->isoRuns[bd->isoRunLast]; + DirProp *dirProps=bd->pBiDi->dirProps; + if(DIRPROP_FLAG(dirProps[lastCcPos])&MASK_ISO) /* after an isolate */ + return; + if(NO_OVERRIDE(embeddingLevel)>NO_OVERRIDE(contextLevel)) /* not a PDF */ + contextLevel=embeddingLevel; + pLastIsoRun->limit=pLastIsoRun->start; + pLastIsoRun->level=embeddingLevel; + pLastIsoRun->lastStrong=pLastIsoRun->lastBase=contextLevel&1; + pLastIsoRun->contextDir=(UBiDiDirection)(contextLevel&1); + pLastIsoRun->contextPos=(UBiDiDirection)lastCcPos; +} + +/* LRI or RLI */ +static void +bracketProcessLRI_RLI(BracketData *bd, UBiDiLevel level) { + IsoRun *pLastIsoRun=&bd->isoRuns[bd->isoRunLast]; + int16_t lastLimit; + pLastIsoRun->lastBase=ON; + lastLimit=pLastIsoRun->limit; + bd->isoRunLast++; + pLastIsoRun++; + pLastIsoRun->start=pLastIsoRun->limit=lastLimit; + pLastIsoRun->level=level; + pLastIsoRun->lastStrong=pLastIsoRun->lastBase=level&1; + pLastIsoRun->contextDir=(UBiDiDirection)(level&1); + pLastIsoRun->contextPos=0; +} + +/* PDI */ +static void +bracketProcessPDI(BracketData *bd) { + IsoRun *pLastIsoRun; + bd->isoRunLast--; + pLastIsoRun=&bd->isoRuns[bd->isoRunLast]; + pLastIsoRun->lastBase=ON; +} + +/* newly found opening bracket: create an openings entry */ +static UBool /* return true if success */ +bracketAddOpening(BracketData *bd, char16_t match, int32_t position) { + IsoRun *pLastIsoRun=&bd->isoRuns[bd->isoRunLast]; + Opening *pOpening; + if(pLastIsoRun->limit>=bd->openingsCount) { /* no available new entry */ + UBiDi *pBiDi=bd->pBiDi; + if(!getInitialOpeningsMemory(pBiDi, pLastIsoRun->limit * 2)) + return false; + if(bd->openings==bd->simpleOpenings) + uprv_memcpy(pBiDi->openingsMemory, bd->simpleOpenings, + SIMPLE_OPENINGS_COUNT * sizeof(Opening)); + bd->openings=pBiDi->openingsMemory; /* may have changed */ + bd->openingsCount=pBiDi->openingsSize / sizeof(Opening); + } + pOpening=&bd->openings[pLastIsoRun->limit]; + pOpening->position=position; + pOpening->match=match; + pOpening->contextDir=pLastIsoRun->contextDir; + pOpening->contextPos=pLastIsoRun->contextPos; + pOpening->flags=0; + pLastIsoRun->limit++; + return true; +} + +/* change N0c1 to N0c2 when a preceding bracket is assigned the embedding level */ +static void +fixN0c(BracketData *bd, int32_t openingIndex, int32_t newPropPosition, DirProp newProp) { + /* This function calls itself recursively */ + IsoRun *pLastIsoRun=&bd->isoRuns[bd->isoRunLast]; + Opening *qOpening; + DirProp *dirProps=bd->pBiDi->dirProps; + int32_t k, openingPosition, closingPosition; + for(k=openingIndex+1, qOpening=&bd->openings[k]; klimit; k++, qOpening++) { + if(qOpening->match>=0) /* not an N0c match */ + continue; + if(newPropPositioncontextPos) + break; + if(newPropPosition>=qOpening->position) + continue; + if(newProp==qOpening->contextDir) + break; + openingPosition=qOpening->position; + dirProps[openingPosition]=newProp; + closingPosition=-(qOpening->match); + dirProps[closingPosition]=newProp; + qOpening->match=0; /* prevent further changes */ + fixN0c(bd, k, openingPosition, newProp); + fixN0c(bd, k, closingPosition, newProp); + } +} + +/* process closing bracket */ +static DirProp /* return L or R if N0b or N0c, ON if N0d */ +bracketProcessClosing(BracketData *bd, int32_t openIdx, int32_t position) { + IsoRun *pLastIsoRun=&bd->isoRuns[bd->isoRunLast]; + Opening *pOpening, *qOpening; + UBiDiDirection direction; + UBool stable; + DirProp newProp; + pOpening=&bd->openings[openIdx]; + direction=(UBiDiDirection)(pLastIsoRun->level&1); + stable=true; /* assume stable until proved otherwise */ + + /* The stable flag is set when brackets are paired and their + level is resolved and cannot be changed by what will be + found later in the source string. + An unstable match can occur only when applying N0c, where + the resolved level depends on the preceding context, and + this context may be affected by text occurring later. + Example: RTL paragraph containing: abc[(latin) HEBREW] + When the closing parenthesis is encountered, it appears + that N0c1 must be applied since 'abc' sets an opposite + direction context and both parentheses receive level 2. + However, when the closing square bracket is processed, + N0b applies because of 'HEBREW' being included within the + brackets, thus the square brackets are treated like R and + receive level 1. However, this changes the preceding + context of the opening parenthesis, and it now appears + that N0c2 must be applied to the parentheses rather than + N0c1. */ + + if((direction==0 && pOpening->flags&FOUND_L) || + (direction==1 && pOpening->flags&FOUND_R)) { /* N0b */ + newProp=static_cast(direction); + } + else if(pOpening->flags&(FOUND_L|FOUND_R)) { /* N0c */ + /* it is stable if there is no containing pair or in + conditions too complicated and not worth checking */ + stable=(openIdx==pLastIsoRun->start); + if(direction!=pOpening->contextDir) + newProp= static_cast(pOpening->contextDir); /* N0c1 */ + else + newProp= static_cast(direction); /* N0c2 */ + } else { + /* forget this and any brackets nested within this pair */ + pLastIsoRun->limit= static_cast(openIdx); + return ON; /* N0d */ + } + bd->pBiDi->dirProps[pOpening->position]=newProp; + bd->pBiDi->dirProps[position]=newProp; + /* Update nested N0c pairs that may be affected */ + fixN0c(bd, openIdx, pOpening->position, newProp); + if(stable) { + pLastIsoRun->limit= static_cast(openIdx); /* forget any brackets nested within this pair */ + /* remove lower located synonyms if any */ + while(pLastIsoRun->limit>pLastIsoRun->start && + bd->openings[pLastIsoRun->limit-1].position==pOpening->position) + pLastIsoRun->limit--; + } else { + int32_t k; + pOpening->match=-position; + /* neutralize lower located synonyms if any */ + k=openIdx-1; + while(k>=pLastIsoRun->start && + bd->openings[k].position==pOpening->position) + bd->openings[k--].match=0; + /* neutralize any unmatched opening between the current pair; + this will also neutralize higher located synonyms if any */ + for(k=openIdx+1; klimit; k++) { + qOpening=&bd->openings[k]; + if(qOpening->position>=position) + break; + if(qOpening->match>0) + qOpening->match=0; + } + } + return newProp; +} + +/* handle strong characters, digits and candidates for closing brackets */ +static UBool /* return true if success */ +bracketProcessChar(BracketData *bd, int32_t position) { + IsoRun *pLastIsoRun=&bd->isoRuns[bd->isoRunLast]; + DirProp *dirProps, dirProp, newProp; + UBiDiLevel level; + dirProps=bd->pBiDi->dirProps; + dirProp=dirProps[position]; + if(dirProp==ON) { + char16_t c, match; + int32_t idx; + /* First see if it is a matching closing bracket. Hopefully, this is + more efficient than checking if it is a closing bracket at all */ + c=bd->pBiDi->text[position]; + for(idx=pLastIsoRun->limit-1; idx>=pLastIsoRun->start; idx--) { + if(bd->openings[idx].match!=c) + continue; + /* We have a match */ + newProp=bracketProcessClosing(bd, idx, position); + if(newProp==ON) { /* N0d */ + c=0; /* prevent handling as an opening */ + break; + } + pLastIsoRun->lastBase=ON; + pLastIsoRun->contextDir=(UBiDiDirection)newProp; + pLastIsoRun->contextPos=position; + level=bd->pBiDi->levels[position]; + if(level&UBIDI_LEVEL_OVERRIDE) { /* X4, X5 */ + uint16_t flag; + int32_t i; + newProp=level&1; + pLastIsoRun->lastStrong=newProp; + flag=DIRPROP_FLAG(newProp); + for(i=pLastIsoRun->start; iopenings[i].flags|=flag; + /* matching brackets are not overridden by LRO/RLO */ + bd->pBiDi->levels[position]&=~UBIDI_LEVEL_OVERRIDE; + } + /* matching brackets are not overridden by LRO/RLO */ + bd->pBiDi->levels[bd->openings[idx].position]&=~UBIDI_LEVEL_OVERRIDE; + return true; + } + /* We get here only if the ON character is not a matching closing + bracket or it is a case of N0d */ + /* Now see if it is an opening bracket */ + if(c) + match= static_cast(u_getBidiPairedBracket(c)); /* get the matching char */ + else + match=0; + if(match!=c && /* has a matching char */ + ubidi_getPairedBracketType(c)==U_BPT_OPEN) { /* opening bracket */ + /* special case: process synonyms + create an opening entry for each synonym */ + if(match==0x232A) { /* RIGHT-POINTING ANGLE BRACKET */ + if(!bracketAddOpening(bd, 0x3009, position)) + return false; + } + else if(match==0x3009) { /* RIGHT ANGLE BRACKET */ + if(!bracketAddOpening(bd, 0x232A, position)) + return false; + } + if(!bracketAddOpening(bd, match, position)) + return false; + } + } + level=bd->pBiDi->levels[position]; + if(level&UBIDI_LEVEL_OVERRIDE) { /* X4, X5 */ + newProp=level&1; + if(dirProp!=S && dirProp!=WS && dirProp!=ON) + dirProps[position]=newProp; + pLastIsoRun->lastBase=newProp; + pLastIsoRun->lastStrong=newProp; + pLastIsoRun->contextDir=(UBiDiDirection)newProp; + pLastIsoRun->contextPos=position; + } + else if(dirProp<=R || dirProp==AL) { + newProp= static_cast(DIR_FROM_STRONG(dirProp)); + pLastIsoRun->lastBase=dirProp; + pLastIsoRun->lastStrong=dirProp; + pLastIsoRun->contextDir=(UBiDiDirection)newProp; + pLastIsoRun->contextPos=position; + } + else if(dirProp==EN) { + pLastIsoRun->lastBase=EN; + if(pLastIsoRun->lastStrong==L) { + newProp=L; /* W7 */ + if(!bd->isNumbersSpecial) + dirProps[position]=ENL; + pLastIsoRun->contextDir=(UBiDiDirection)L; + pLastIsoRun->contextPos=position; + } + else { + newProp=R; /* N0 */ + if(pLastIsoRun->lastStrong==AL) + dirProps[position]=AN; /* W2 */ + else + dirProps[position]=ENR; + pLastIsoRun->contextDir=(UBiDiDirection)R; + pLastIsoRun->contextPos=position; + } + } + else if(dirProp==AN) { + newProp=R; /* N0 */ + pLastIsoRun->lastBase=AN; + pLastIsoRun->contextDir=(UBiDiDirection)R; + pLastIsoRun->contextPos=position; + } + else if(dirProp==NSM) { + /* if the last real char was ON, change NSM to ON so that it + will stay ON even if the last real char is a bracket which + may be changed to L or R */ + newProp=pLastIsoRun->lastBase; + if(newProp==ON) + dirProps[position]=newProp; + } + else { + newProp=dirProp; + pLastIsoRun->lastBase=dirProp; + } + if(newProp<=R || newProp==AL) { + int32_t i; + uint16_t flag=DIRPROP_FLAG(DIR_FROM_STRONG(newProp)); + for(i=pLastIsoRun->start; ilimit; i++) + if(position>bd->openings[i].position) + bd->openings[i].flags|=flag; + } + return true; +} + +/* perform (X1)..(X9) ------------------------------------------------------- */ + +/* determine if the text is mixed-directional or single-directional */ +static UBiDiDirection +directionFromFlags(UBiDi *pBiDi) { + Flags flags=pBiDi->flags; + /* if the text contains AN and neutrals, then some neutrals may become RTL */ + if(!(flags&MASK_RTL || ((flags&DIRPROP_FLAG(AN)) && (flags&MASK_POSSIBLE_N)))) { + return UBIDI_LTR; + } else if(!(flags&MASK_LTR)) { + return UBIDI_RTL; + } else { + return UBIDI_MIXED; + } +} + +/* + * Resolve the explicit levels as specified by explicit embedding codes. + * Recalculate the flags to have them reflect the real properties + * after taking the explicit embeddings into account. + * + * The BiDi algorithm is designed to result in the same behavior whether embedding + * levels are externally specified (from "styled text", supposedly the preferred + * method) or set by explicit embedding codes (LRx, RLx, PDF, FSI, PDI) in the plain text. + * That is why (X9) instructs to remove all not-isolate explicit codes (and BN). + * However, in a real implementation, the removal of these codes and their index + * positions in the plain text is undesirable since it would result in + * reallocated, reindexed text. + * Instead, this implementation leaves the codes in there and just ignores them + * in the subsequent processing. + * In order to get the same reordering behavior, positions with a BN or a not-isolate + * explicit embedding code just get the same level assigned as the last "real" + * character. + * + * Some implementations, not this one, then overwrite some of these + * directionality properties at "real" same-level-run boundaries by + * L or R codes so that the resolution of weak types can be performed on the + * entire paragraph at once instead of having to parse it once more and + * perform that resolution on same-level-runs. + * This limits the scope of the implicit rules in effectively + * the same way as the run limits. + * + * Instead, this implementation does not modify these codes, except for + * paired brackets whose properties (ON) may be replaced by L or R. + * On one hand, the paragraph has to be scanned for same-level-runs, but + * on the other hand, this saves another loop to reset these codes, + * or saves making and modifying a copy of dirProps[]. + * + * + * Note that (Pn) and (Xn) changed significantly from version 4 of the BiDi algorithm. + * + * + * Handling the stack of explicit levels (Xn): + * + * With the BiDi stack of explicit levels, as pushed with each + * LRE, RLE, LRO, RLO, LRI, RLI and FSI and popped with each PDF and PDI, + * the explicit level must never exceed UBIDI_MAX_EXPLICIT_LEVEL. + * + * In order to have a correct push-pop semantics even in the case of overflows, + * overflow counters and a valid isolate counter are used as described in UAX#9 + * section 3.3.2 "Explicit Levels and Directions". + * + * This implementation assumes that UBIDI_MAX_EXPLICIT_LEVEL is odd. + * + * Returns normally the direction; -1 if there was a memory shortage + * + */ +static UBiDiDirection +resolveExplicitLevels(UBiDi *pBiDi, UErrorCode *pErrorCode) { + DirProp *dirProps=pBiDi->dirProps; + UBiDiLevel *levels=pBiDi->levels; + const char16_t *text=pBiDi->text; + + int32_t i=0, length=pBiDi->length; + Flags flags=pBiDi->flags; /* collect all directionalities in the text */ + DirProp dirProp; + UBiDiLevel level=GET_PARALEVEL(pBiDi, 0); + UBiDiDirection direction; + pBiDi->isolateCount=0; + + if(U_FAILURE(*pErrorCode)) { return UBIDI_LTR; } + + /* determine if the text is mixed-directional or single-directional */ + direction=directionFromFlags(pBiDi); + + /* we may not need to resolve any explicit levels */ + if((direction!=UBIDI_MIXED)) { + /* not mixed directionality: levels don't matter - trailingWSStart will be 0 */ + return direction; + } + if(pBiDi->reorderingMode > UBIDI_REORDER_LAST_LOGICAL_TO_VISUAL) { + /* inverse BiDi: mixed, but all characters are at the same embedding level */ + /* set all levels to the paragraph level */ + int32_t paraIndex, start, limit; + for(paraIndex=0; paraIndexparaCount; paraIndex++) { + if(paraIndex==0) + start=0; + else + start=pBiDi->paras[paraIndex-1].limit; + limit=pBiDi->paras[paraIndex].limit; + level= static_cast(pBiDi->paras[paraIndex].level); + for(i=start; iparaCount; paraIndex++) { + if(paraIndex==0) + start=0; + else + start=pBiDi->paras[paraIndex-1].limit; + limit=pBiDi->paras[paraIndex].limit; + level= static_cast(pBiDi->paras[paraIndex].level); + for(i=start; i=UBIDI_MAX_EXPLICIT_LEVEL + but we need one more entry as base */ + uint32_t stackLast=0; + int32_t overflowIsolateCount=0; + int32_t overflowEmbeddingCount=0; + int32_t validIsolateCount=0; + BracketData bracketData; + bracketInit(pBiDi, &bracketData); + stack[0]=level; /* initialize base entry to para level, no override, no isolate */ + + /* recalculate the flags */ + flags=0; + + for(i=0; i0 && stack[stackLast]pBiDi->isolateCount) + pBiDi->isolateCount=validIsolateCount; + embeddingLevel=newLevel; + /* we can increment stackLast without checking because newLevel + will exceed UBIDI_MAX_EXPLICIT_LEVEL before stackLast overflows */ + stackLast++; + stack[stackLast]=embeddingLevel+ISOLATE; + bracketProcessLRI_RLI(&bracketData, embeddingLevel); + } else { + /* make it WS so that it is handled by adjustWSLevels() */ + dirProps[i]=WS; + overflowIsolateCount++; + } + break; + case PDI: + if(NO_OVERRIDE(embeddingLevel)!=NO_OVERRIDE(previousLevel)) { + bracketProcessBoundary(&bracketData, lastCcPos, + previousLevel, embeddingLevel); + flags|=DIRPROP_FLAG_MULTI_RUNS; + } + /* (X6a) */ + if(overflowIsolateCount) { + overflowIsolateCount--; + /* make it WS so that it is handled by adjustWSLevels() */ + dirProps[i]=WS; + } + else if(validIsolateCount) { + flags|=DIRPROP_FLAG(PDI); + lastCcPos=i; + overflowEmbeddingCount=0; + while(stack[stackLast]paraLevel); + if(pBiDi->orderParagraphsLTR && (flags&DIRPROP_FLAG(B))) + flags|=DIRPROP_FLAG(L); + /* again, determine if the text is mixed-directional or single-directional */ + pBiDi->flags=flags; + direction=directionFromFlags(pBiDi); + } + return direction; +} + +/* + * Use a pre-specified embedding levels array: + * + * Adjust the directional properties for overrides (->LEVEL_OVERRIDE), + * ignore all explicit codes (X9), + * and check all the preset levels. + * + * Recalculate the flags to have them reflect the real properties + * after taking the explicit embeddings into account. + */ +static UBiDiDirection +checkExplicitLevels(UBiDi *pBiDi, UErrorCode *pErrorCode) { + DirProp *dirProps=pBiDi->dirProps; + UBiDiLevel *levels=pBiDi->levels; + int32_t isolateCount=0; + + int32_t length=pBiDi->length; + Flags flags=0; /* collect all directionalities in the text */ + pBiDi->isolateCount=0; + + int32_t currentParaIndex = 0; + int32_t currentParaLimit = pBiDi->paras[0].limit; + int32_t currentParaLevel = pBiDi->paraLevel; + + for(int32_t i=0; ipBiDi->isolateCount) + pBiDi->isolateCount=isolateCount; + } + else if(dirProp==PDI) + isolateCount--; + else if(dirProp==B) + isolateCount=0; + + // optimized version of int32_t currentParaLevel = GET_PARALEVEL(pBiDi, i); + if (pBiDi->defaultParaLevel != 0 && + i == currentParaLimit && (currentParaIndex + 1) < pBiDi->paraCount) { + currentParaLevel = pBiDi->paras[++currentParaIndex].level; + currentParaLimit = pBiDi->paras[currentParaIndex].limit; + } + + UBiDiLevel overrideFlag = level & UBIDI_LEVEL_OVERRIDE; + level &= ~UBIDI_LEVEL_OVERRIDE; + if (level < currentParaLevel || UBIDI_MAX_EXPLICIT_LEVEL < level) { + if (level == 0) { + if (dirProp == B) { + // Paragraph separators are ok with explicit level 0. + // Prevents reordering of paragraphs. + } else { + // Treat explicit level 0 as a wildcard for the paragraph level. + // Avoid making the caller guess what the paragraph level would be. + level = (UBiDiLevel)currentParaLevel; + levels[i] = level | overrideFlag; + } + } else { + // 1 <= level < currentParaLevel or UBIDI_MAX_EXPLICIT_LEVEL < level + /* level out of bounds */ + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return UBIDI_LTR; + } + } + if (overrideFlag != 0) { + /* keep the override flag in levels[i] but adjust the flags */ + flags|=DIRPROP_FLAG_O(level); + } else { + /* set the flags */ + flags|=DIRPROP_FLAG_E(level)|DIRPROP_FLAG(dirProp); + } + } + if(flags&MASK_EMBEDDING) + flags|=DIRPROP_FLAG_LR(pBiDi->paraLevel); + /* determine if the text is mixed-directional or single-directional */ + pBiDi->flags=flags; + return directionFromFlags(pBiDi); +} + +/****************************************************************** + The Properties state machine table +******************************************************************* + + All table cells are 8 bits: + bits 0..4: next state + bits 5..7: action to perform (if > 0) + + Cells may be of format "n" where n represents the next state + (except for the rightmost column). + Cells may also be of format "s(x,y)" where x represents an action + to perform and y represents the next state. + +******************************************************************* + Definitions and type for properties state table +******************************************************************* +*/ +#define IMPTABPROPS_COLUMNS 16 +#define IMPTABPROPS_RES (IMPTABPROPS_COLUMNS - 1) +#define GET_STATEPROPS(cell) ((cell)&0x1f) +#define GET_ACTIONPROPS(cell) ((cell)>>5) +#define s(action, newState) ((uint8_t)(newState+(action<<5))) + +static const uint8_t groupProp[] = /* dirProp regrouped */ +{ +/* L R EN ES ET AN CS B S WS ON LRE LRO AL RLE RLO PDF NSM BN FSI LRI RLI PDI ENL ENR */ + 0, 1, 2, 7, 8, 3, 9, 6, 5, 4, 4, 10, 10, 12, 10, 10, 10, 11, 10, 4, 4, 4, 4, 13, 14 +}; +enum { DirProp_L=0, DirProp_R=1, DirProp_EN=2, DirProp_AN=3, DirProp_ON=4, DirProp_S=5, DirProp_B=6 }; /* reduced dirProp */ + +/****************************************************************** + + PROPERTIES STATE TABLE + + In table impTabProps, + - the ON column regroups ON and WS, FSI, RLI, LRI and PDI + - the BN column regroups BN, LRE, RLE, LRO, RLO, PDF + - the Res column is the reduced property assigned to a run + + Action 1: process current run1, init new run1 + 2: init new run2 + 3: process run1, process run2, init new run1 + 4: process run1, set run1=run2, init new run2 + + Notes: + 1) This table is used in resolveImplicitLevels(). + 2) This table triggers actions when there is a change in the Bidi + property of incoming characters (action 1). + 3) Most such property sequences are processed immediately (in + fact, passed to processPropertySeq(). + 4) However, numbers are assembled as one sequence. This means + that undefined situations (like CS following digits, until + it is known if the next char will be a digit) are held until + following chars define them. + Example: digits followed by CS, then comes another CS or ON; + the digits will be processed, then the CS assigned + as the start of an ON sequence (action 3). + 5) There are cases where more than one sequence must be + processed, for instance digits followed by CS followed by L: + the digits must be processed as one sequence, and the CS + must be processed as an ON sequence, all this before starting + assembling chars for the opening L sequence. + + +*/ +static const uint8_t impTabProps[][IMPTABPROPS_COLUMNS] = +{ +/* L , R , EN , AN , ON , S , B , ES , ET , CS , BN , NSM , AL , ENL , ENR , Res */ +/* 0 Init */ { 1 , 2 , 4 , 5 , 7 , 15 , 17 , 7 , 9 , 7 , 0 , 7 , 3 , 18 , 21 , DirProp_ON }, +/* 1 L */ { 1 , s(1,2), s(1,4), s(1,5), s(1,7),s(1,15),s(1,17), s(1,7), s(1,9), s(1,7), 1 , 1 , s(1,3),s(1,18),s(1,21), DirProp_L }, +/* 2 R */ { s(1,1), 2 , s(1,4), s(1,5), s(1,7),s(1,15),s(1,17), s(1,7), s(1,9), s(1,7), 2 , 2 , s(1,3),s(1,18),s(1,21), DirProp_R }, +/* 3 AL */ { s(1,1), s(1,2), s(1,6), s(1,6), s(1,8),s(1,16),s(1,17), s(1,8), s(1,8), s(1,8), 3 , 3 , 3 ,s(1,18),s(1,21), DirProp_R }, +/* 4 EN */ { s(1,1), s(1,2), 4 , s(1,5), s(1,7),s(1,15),s(1,17),s(2,10), 11 ,s(2,10), 4 , 4 , s(1,3), 18 , 21 , DirProp_EN }, +/* 5 AN */ { s(1,1), s(1,2), s(1,4), 5 , s(1,7),s(1,15),s(1,17), s(1,7), s(1,9),s(2,12), 5 , 5 , s(1,3),s(1,18),s(1,21), DirProp_AN }, +/* 6 AL:EN/AN */ { s(1,1), s(1,2), 6 , 6 , s(1,8),s(1,16),s(1,17), s(1,8), s(1,8),s(2,13), 6 , 6 , s(1,3), 18 , 21 , DirProp_AN }, +/* 7 ON */ { s(1,1), s(1,2), s(1,4), s(1,5), 7 ,s(1,15),s(1,17), 7 ,s(2,14), 7 , 7 , 7 , s(1,3),s(1,18),s(1,21), DirProp_ON }, +/* 8 AL:ON */ { s(1,1), s(1,2), s(1,6), s(1,6), 8 ,s(1,16),s(1,17), 8 , 8 , 8 , 8 , 8 , s(1,3),s(1,18),s(1,21), DirProp_ON }, +/* 9 ET */ { s(1,1), s(1,2), 4 , s(1,5), 7 ,s(1,15),s(1,17), 7 , 9 , 7 , 9 , 9 , s(1,3), 18 , 21 , DirProp_ON }, +/*10 EN+ES/CS */ { s(3,1), s(3,2), 4 , s(3,5), s(4,7),s(3,15),s(3,17), s(4,7),s(4,14), s(4,7), 10 , s(4,7), s(3,3), 18 , 21 , DirProp_EN }, +/*11 EN+ET */ { s(1,1), s(1,2), 4 , s(1,5), s(1,7),s(1,15),s(1,17), s(1,7), 11 , s(1,7), 11 , 11 , s(1,3), 18 , 21 , DirProp_EN }, +/*12 AN+CS */ { s(3,1), s(3,2), s(3,4), 5 , s(4,7),s(3,15),s(3,17), s(4,7),s(4,14), s(4,7), 12 , s(4,7), s(3,3),s(3,18),s(3,21), DirProp_AN }, +/*13 AL:EN/AN+CS */ { s(3,1), s(3,2), 6 , 6 , s(4,8),s(3,16),s(3,17), s(4,8), s(4,8), s(4,8), 13 , s(4,8), s(3,3), 18 , 21 , DirProp_AN }, +/*14 ON+ET */ { s(1,1), s(1,2), s(4,4), s(1,5), 7 ,s(1,15),s(1,17), 7 , 14 , 7 , 14 , 14 , s(1,3),s(4,18),s(4,21), DirProp_ON }, +/*15 S */ { s(1,1), s(1,2), s(1,4), s(1,5), s(1,7), 15 ,s(1,17), s(1,7), s(1,9), s(1,7), 15 , s(1,7), s(1,3),s(1,18),s(1,21), DirProp_S }, +/*16 AL:S */ { s(1,1), s(1,2), s(1,6), s(1,6), s(1,8), 16 ,s(1,17), s(1,8), s(1,8), s(1,8), 16 , s(1,8), s(1,3),s(1,18),s(1,21), DirProp_S }, +/*17 B */ { s(1,1), s(1,2), s(1,4), s(1,5), s(1,7),s(1,15), 17 , s(1,7), s(1,9), s(1,7), 17 , s(1,7), s(1,3),s(1,18),s(1,21), DirProp_B }, +/*18 ENL */ { s(1,1), s(1,2), 18 , s(1,5), s(1,7),s(1,15),s(1,17),s(2,19), 20 ,s(2,19), 18 , 18 , s(1,3), 18 , 21 , DirProp_L }, +/*19 ENL+ES/CS */ { s(3,1), s(3,2), 18 , s(3,5), s(4,7),s(3,15),s(3,17), s(4,7),s(4,14), s(4,7), 19 , s(4,7), s(3,3), 18 , 21 , DirProp_L }, +/*20 ENL+ET */ { s(1,1), s(1,2), 18 , s(1,5), s(1,7),s(1,15),s(1,17), s(1,7), 20 , s(1,7), 20 , 20 , s(1,3), 18 , 21 , DirProp_L }, +/*21 ENR */ { s(1,1), s(1,2), 21 , s(1,5), s(1,7),s(1,15),s(1,17),s(2,22), 23 ,s(2,22), 21 , 21 , s(1,3), 18 , 21 , DirProp_AN }, +/*22 ENR+ES/CS */ { s(3,1), s(3,2), 21 , s(3,5), s(4,7),s(3,15),s(3,17), s(4,7),s(4,14), s(4,7), 22 , s(4,7), s(3,3), 18 , 21 , DirProp_AN }, +/*23 ENR+ET */ { s(1,1), s(1,2), 21 , s(1,5), s(1,7),s(1,15),s(1,17), s(1,7), 23 , s(1,7), 23 , 23 , s(1,3), 18 , 21 , DirProp_AN } +}; + +/* we must undef macro s because the levels tables have a different + * structure (4 bits for action and 4 bits for next state. + */ +#undef s + +/****************************************************************** + The levels state machine tables +******************************************************************* + + All table cells are 8 bits: + bits 0..3: next state + bits 4..7: action to perform (if > 0) + + Cells may be of format "n" where n represents the next state + (except for the rightmost column). + Cells may also be of format "s(x,y)" where x represents an action + to perform and y represents the next state. + + This format limits each table to 16 states each and to 15 actions. + +******************************************************************* + Definitions and type for levels state tables +******************************************************************* +*/ +#define IMPTABLEVELS_COLUMNS (DirProp_B + 2) +#define IMPTABLEVELS_RES (IMPTABLEVELS_COLUMNS - 1) +#define GET_STATE(cell) ((cell)&0x0f) +#define GET_ACTION(cell) ((cell)>>4) +#define s(action, newState) ((uint8_t)(newState+(action<<4))) + +typedef uint8_t ImpTab[][IMPTABLEVELS_COLUMNS]; +typedef uint8_t ImpAct[]; + +/* FOOD FOR THOUGHT: each ImpTab should have its associated ImpAct, + * instead of having a pair of ImpTab and a pair of ImpAct. + */ +typedef struct ImpTabPair { + const void * pImpTab[2]; + const void * pImpAct[2]; +} ImpTabPair; + +/****************************************************************** + + LEVELS STATE TABLES + + In all levels state tables, + - state 0 is the initial state + - the Res column is the increment to add to the text level + for this property sequence. + + The impAct arrays for each table of a pair map the local action + numbers of the table to the total list of actions. For instance, + action 2 in a given table corresponds to the action number which + appears in entry [2] of the impAct array for that table. + The first entry of all impAct arrays must be 0. + + Action 1: init conditional sequence + 2: prepend conditional sequence to current sequence + 3: set ON sequence to new level - 1 + 4: init EN/AN/ON sequence + 5: fix EN/AN/ON sequence followed by R + 6: set previous level sequence to level 2 + + Notes: + 1) These tables are used in processPropertySeq(). The input + is property sequences as determined by resolveImplicitLevels. + 2) Most such property sequences are processed immediately + (levels are assigned). + 3) However, some sequences cannot be assigned a final level till + one or more following sequences are received. For instance, + ON following an R sequence within an even-level paragraph. + If the following sequence is R, the ON sequence will be + assigned basic run level+1, and so will the R sequence. + 4) S is generally handled like ON, since its level will be fixed + to paragraph level in adjustWSLevels(). + +*/ + +static const ImpTab impTabL_DEFAULT = /* Even paragraph level */ +/* In this table, conditional sequences receive the lower possible level + until proven otherwise. +*/ +{ +/* L , R , EN , AN , ON , S , B , Res */ +/* 0 : init */ { 0 , 1 , 0 , 2 , 0 , 0 , 0 , 0 }, +/* 1 : R */ { 0 , 1 , 3 , 3 , s(1,4), s(1,4), 0 , 1 }, +/* 2 : AN */ { 0 , 1 , 0 , 2 , s(1,5), s(1,5), 0 , 2 }, +/* 3 : R+EN/AN */ { 0 , 1 , 3 , 3 , s(1,4), s(1,4), 0 , 2 }, +/* 4 : R+ON */ { 0 , s(2,1), s(3,3), s(3,3), 4 , 4 , 0 , 0 }, +/* 5 : AN+ON */ { 0 , s(2,1), 0 , s(3,2), 5 , 5 , 0 , 0 } +}; +static const ImpTab impTabR_DEFAULT = /* Odd paragraph level */ +/* In this table, conditional sequences receive the lower possible level + until proven otherwise. +*/ +{ +/* L , R , EN , AN , ON , S , B , Res */ +/* 0 : init */ { 1 , 0 , 2 , 2 , 0 , 0 , 0 , 0 }, +/* 1 : L */ { 1 , 0 , 1 , 3 , s(1,4), s(1,4), 0 , 1 }, +/* 2 : EN/AN */ { 1 , 0 , 2 , 2 , 0 , 0 , 0 , 1 }, +/* 3 : L+AN */ { 1 , 0 , 1 , 3 , 5 , 5 , 0 , 1 }, +/* 4 : L+ON */ { s(2,1), 0 , s(2,1), 3 , 4 , 4 , 0 , 0 }, +/* 5 : L+AN+ON */ { 1 , 0 , 1 , 3 , 5 , 5 , 0 , 0 } +}; +static const ImpAct impAct0 = {0,1,2,3,4}; +static const ImpTabPair impTab_DEFAULT = {{&impTabL_DEFAULT, + &impTabR_DEFAULT}, + {&impAct0, &impAct0}}; + +static const ImpTab impTabL_NUMBERS_SPECIAL = /* Even paragraph level */ +/* In this table, conditional sequences receive the lower possible level + until proven otherwise. +*/ +{ +/* L , R , EN , AN , ON , S , B , Res */ +/* 0 : init */ { 0 , 2 , s(1,1), s(1,1), 0 , 0 , 0 , 0 }, +/* 1 : L+EN/AN */ { 0 , s(4,2), 1 , 1 , 0 , 0 , 0 , 0 }, +/* 2 : R */ { 0 , 2 , 4 , 4 , s(1,3), s(1,3), 0 , 1 }, +/* 3 : R+ON */ { 0 , s(2,2), s(3,4), s(3,4), 3 , 3 , 0 , 0 }, +/* 4 : R+EN/AN */ { 0 , 2 , 4 , 4 , s(1,3), s(1,3), 0 , 2 } +}; +static const ImpTabPair impTab_NUMBERS_SPECIAL = {{&impTabL_NUMBERS_SPECIAL, + &impTabR_DEFAULT}, + {&impAct0, &impAct0}}; + +static const ImpTab impTabL_GROUP_NUMBERS_WITH_R = +/* In this table, EN/AN+ON sequences receive levels as if associated with R + until proven that there is L or sor/eor on both sides. AN is handled like EN. +*/ +{ +/* L , R , EN , AN , ON , S , B , Res */ +/* 0 init */ { 0 , 3 , s(1,1), s(1,1), 0 , 0 , 0 , 0 }, +/* 1 EN/AN */ { s(2,0), 3 , 1 , 1 , 2 , s(2,0), s(2,0), 2 }, +/* 2 EN/AN+ON */ { s(2,0), 3 , 1 , 1 , 2 , s(2,0), s(2,0), 1 }, +/* 3 R */ { 0 , 3 , 5 , 5 , s(1,4), 0 , 0 , 1 }, +/* 4 R+ON */ { s(2,0), 3 , 5 , 5 , 4 , s(2,0), s(2,0), 1 }, +/* 5 R+EN/AN */ { 0 , 3 , 5 , 5 , s(1,4), 0 , 0 , 2 } +}; +static const ImpTab impTabR_GROUP_NUMBERS_WITH_R = +/* In this table, EN/AN+ON sequences receive levels as if associated with R + until proven that there is L on both sides. AN is handled like EN. +*/ +{ +/* L , R , EN , AN , ON , S , B , Res */ +/* 0 init */ { 2 , 0 , 1 , 1 , 0 , 0 , 0 , 0 }, +/* 1 EN/AN */ { 2 , 0 , 1 , 1 , 0 , 0 , 0 , 1 }, +/* 2 L */ { 2 , 0 , s(1,4), s(1,4), s(1,3), 0 , 0 , 1 }, +/* 3 L+ON */ { s(2,2), 0 , 4 , 4 , 3 , 0 , 0 , 0 }, +/* 4 L+EN/AN */ { s(2,2), 0 , 4 , 4 , 3 , 0 , 0 , 1 } +}; +static const ImpTabPair impTab_GROUP_NUMBERS_WITH_R = { + {&impTabL_GROUP_NUMBERS_WITH_R, + &impTabR_GROUP_NUMBERS_WITH_R}, + {&impAct0, &impAct0}}; + + +static const ImpTab impTabL_INVERSE_NUMBERS_AS_L = +/* This table is identical to the Default LTR table except that EN and AN are + handled like L. +*/ +{ +/* L , R , EN , AN , ON , S , B , Res */ +/* 0 : init */ { 0 , 1 , 0 , 0 , 0 , 0 , 0 , 0 }, +/* 1 : R */ { 0 , 1 , 0 , 0 , s(1,4), s(1,4), 0 , 1 }, +/* 2 : AN */ { 0 , 1 , 0 , 0 , s(1,5), s(1,5), 0 , 2 }, +/* 3 : R+EN/AN */ { 0 , 1 , 0 , 0 , s(1,4), s(1,4), 0 , 2 }, +/* 4 : R+ON */ { s(2,0), 1 , s(2,0), s(2,0), 4 , 4 , s(2,0), 1 }, +/* 5 : AN+ON */ { s(2,0), 1 , s(2,0), s(2,0), 5 , 5 , s(2,0), 1 } +}; +static const ImpTab impTabR_INVERSE_NUMBERS_AS_L = +/* This table is identical to the Default RTL table except that EN and AN are + handled like L. +*/ +{ +/* L , R , EN , AN , ON , S , B , Res */ +/* 0 : init */ { 1 , 0 , 1 , 1 , 0 , 0 , 0 , 0 }, +/* 1 : L */ { 1 , 0 , 1 , 1 , s(1,4), s(1,4), 0 , 1 }, +/* 2 : EN/AN */ { 1 , 0 , 1 , 1 , 0 , 0 , 0 , 1 }, +/* 3 : L+AN */ { 1 , 0 , 1 , 1 , 5 , 5 , 0 , 1 }, +/* 4 : L+ON */ { s(2,1), 0 , s(2,1), s(2,1), 4 , 4 , 0 , 0 }, +/* 5 : L+AN+ON */ { 1 , 0 , 1 , 1 , 5 , 5 , 0 , 0 } +}; +static const ImpTabPair impTab_INVERSE_NUMBERS_AS_L = { + {&impTabL_INVERSE_NUMBERS_AS_L, + &impTabR_INVERSE_NUMBERS_AS_L}, + {&impAct0, &impAct0}}; + +static const ImpTab impTabR_INVERSE_LIKE_DIRECT = /* Odd paragraph level */ +/* In this table, conditional sequences receive the lower possible level + until proven otherwise. +*/ +{ +/* L , R , EN , AN , ON , S , B , Res */ +/* 0 : init */ { 1 , 0 , 2 , 2 , 0 , 0 , 0 , 0 }, +/* 1 : L */ { 1 , 0 , 1 , 2 , s(1,3), s(1,3), 0 , 1 }, +/* 2 : EN/AN */ { 1 , 0 , 2 , 2 , 0 , 0 , 0 , 1 }, +/* 3 : L+ON */ { s(2,1), s(3,0), 6 , 4 , 3 , 3 , s(3,0), 0 }, +/* 4 : L+ON+AN */ { s(2,1), s(3,0), 6 , 4 , 5 , 5 , s(3,0), 3 }, +/* 5 : L+AN+ON */ { s(2,1), s(3,0), 6 , 4 , 5 , 5 , s(3,0), 2 }, +/* 6 : L+ON+EN */ { s(2,1), s(3,0), 6 , 4 , 3 , 3 , s(3,0), 1 } +}; +static const ImpAct impAct1 = {0,1,13,14}; +/* FOOD FOR THOUGHT: in LTR table below, check case "JKL 123abc" + */ +static const ImpTabPair impTab_INVERSE_LIKE_DIRECT = { + {&impTabL_DEFAULT, + &impTabR_INVERSE_LIKE_DIRECT}, + {&impAct0, &impAct1}}; + +static const ImpTab impTabL_INVERSE_LIKE_DIRECT_WITH_MARKS = +/* The case handled in this table is (visually): R EN L +*/ +{ +/* L , R , EN , AN , ON , S , B , Res */ +/* 0 : init */ { 0 , s(6,3), 0 , 1 , 0 , 0 , 0 , 0 }, +/* 1 : L+AN */ { 0 , s(6,3), 0 , 1 , s(1,2), s(3,0), 0 , 4 }, +/* 2 : L+AN+ON */ { s(2,0), s(6,3), s(2,0), 1 , 2 , s(3,0), s(2,0), 3 }, +/* 3 : R */ { 0 , s(6,3), s(5,5), s(5,6), s(1,4), s(3,0), 0 , 3 }, +/* 4 : R+ON */ { s(3,0), s(4,3), s(5,5), s(5,6), 4 , s(3,0), s(3,0), 3 }, +/* 5 : R+EN */ { s(3,0), s(4,3), 5 , s(5,6), s(1,4), s(3,0), s(3,0), 4 }, +/* 6 : R+AN */ { s(3,0), s(4,3), s(5,5), 6 , s(1,4), s(3,0), s(3,0), 4 } +}; +static const ImpTab impTabR_INVERSE_LIKE_DIRECT_WITH_MARKS = +/* The cases handled in this table are (visually): R EN L + R L AN L +*/ +{ +/* L , R , EN , AN , ON , S , B , Res */ +/* 0 : init */ { s(1,3), 0 , 1 , 1 , 0 , 0 , 0 , 0 }, +/* 1 : R+EN/AN */ { s(2,3), 0 , 1 , 1 , 2 , s(4,0), 0 , 1 }, +/* 2 : R+EN/AN+ON */ { s(2,3), 0 , 1 , 1 , 2 , s(4,0), 0 , 0 }, +/* 3 : L */ { 3 , 0 , 3 , s(3,6), s(1,4), s(4,0), 0 , 1 }, +/* 4 : L+ON */ { s(5,3), s(4,0), 5 , s(3,6), 4 , s(4,0), s(4,0), 0 }, +/* 5 : L+ON+EN */ { s(5,3), s(4,0), 5 , s(3,6), 4 , s(4,0), s(4,0), 1 }, +/* 6 : L+AN */ { s(5,3), s(4,0), 6 , 6 , 4 , s(4,0), s(4,0), 3 } +}; +static const ImpAct impAct2 = {0,1,2,5,6,7,8}; +static const ImpAct impAct3 = {0,1,9,10,11,12}; +static const ImpTabPair impTab_INVERSE_LIKE_DIRECT_WITH_MARKS = { + {&impTabL_INVERSE_LIKE_DIRECT_WITH_MARKS, + &impTabR_INVERSE_LIKE_DIRECT_WITH_MARKS}, + {&impAct2, &impAct3}}; + +static const ImpTabPair impTab_INVERSE_FOR_NUMBERS_SPECIAL = { + {&impTabL_NUMBERS_SPECIAL, + &impTabR_INVERSE_LIKE_DIRECT}, + {&impAct0, &impAct1}}; + +static const ImpTab impTabL_INVERSE_FOR_NUMBERS_SPECIAL_WITH_MARKS = +/* The case handled in this table is (visually): R EN L +*/ +{ +/* L , R , EN , AN , ON , S , B , Res */ +/* 0 : init */ { 0 , s(6,2), 1 , 1 , 0 , 0 , 0 , 0 }, +/* 1 : L+EN/AN */ { 0 , s(6,2), 1 , 1 , 0 , s(3,0), 0 , 4 }, +/* 2 : R */ { 0 , s(6,2), s(5,4), s(5,4), s(1,3), s(3,0), 0 , 3 }, +/* 3 : R+ON */ { s(3,0), s(4,2), s(5,4), s(5,4), 3 , s(3,0), s(3,0), 3 }, +/* 4 : R+EN/AN */ { s(3,0), s(4,2), 4 , 4 , s(1,3), s(3,0), s(3,0), 4 } +}; +static const ImpTabPair impTab_INVERSE_FOR_NUMBERS_SPECIAL_WITH_MARKS = { + {&impTabL_INVERSE_FOR_NUMBERS_SPECIAL_WITH_MARKS, + &impTabR_INVERSE_LIKE_DIRECT_WITH_MARKS}, + {&impAct2, &impAct3}}; + +#undef s + +typedef struct { + const ImpTab * pImpTab; /* level table pointer */ + const ImpAct * pImpAct; /* action map array */ + int32_t startON; /* start of ON sequence */ + int32_t startL2EN; /* start of level 2 sequence */ + int32_t lastStrongRTL; /* index of last found R or AL */ + int32_t state; /* current state */ + int32_t runStart; /* start position of the run */ + UBiDiLevel runLevel; /* run level before implicit solving */ +} LevState; + +/*------------------------------------------------------------------------*/ + +static void +addPoint(UBiDi *pBiDi, int32_t pos, int32_t flag) + /* param pos: position where to insert + param flag: one of LRM_BEFORE, LRM_AFTER, RLM_BEFORE, RLM_AFTER + */ +{ +#define FIRSTALLOC 10 + Point point; + InsertPoints * pInsertPoints=&(pBiDi->insertPoints); + + if (pInsertPoints->capacity == 0) + { + pInsertPoints->points=static_cast(uprv_malloc(sizeof(Point)*FIRSTALLOC)); + if (pInsertPoints->points == nullptr) + { + pInsertPoints->errorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + pInsertPoints->capacity=FIRSTALLOC; + } + if (pInsertPoints->size >= pInsertPoints->capacity) /* no room for new point */ + { + Point * savePoints=pInsertPoints->points; + pInsertPoints->points=static_cast(uprv_realloc(pInsertPoints->points, + pInsertPoints->capacity*2*sizeof(Point))); + if (pInsertPoints->points == nullptr) + { + pInsertPoints->points=savePoints; + pInsertPoints->errorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + else pInsertPoints->capacity*=2; + } + point.pos=pos; + point.flag=flag; + pInsertPoints->points[pInsertPoints->size]=point; + pInsertPoints->size++; +#undef FIRSTALLOC +} + +static void +setLevelsOutsideIsolates(UBiDi *pBiDi, int32_t start, int32_t limit, UBiDiLevel level) +{ + DirProp *dirProps=pBiDi->dirProps, dirProp; + UBiDiLevel *levels=pBiDi->levels; + int32_t isolateCount=0, k; + for(k=start; kpImpTab; + const ImpAct * pImpAct=pLevState->pImpAct; + UBiDiLevel * levels=pBiDi->levels; + UBiDiLevel level, addLevel; + InsertPoints * pInsertPoints; + int32_t start0, k; + + start0=start; /* save original start position */ + oldStateSeq=(uint8_t)pLevState->state; + cell=(*pImpTab)[oldStateSeq][_prop]; + pLevState->state=GET_STATE(cell); /* isolate the new state */ + actionSeq=(*pImpAct)[GET_ACTION(cell)]; /* isolate the action */ + addLevel=(*pImpTab)[pLevState->state][IMPTABLEVELS_RES]; + + if(actionSeq) { + switch(actionSeq) { + case 1: /* init ON seq */ + pLevState->startON=start0; + break; + + case 2: /* prepend ON seq to current seq */ + start=pLevState->startON; + break; + + case 3: /* EN/AN after R+ON */ + level=pLevState->runLevel+1; + setLevelsOutsideIsolates(pBiDi, pLevState->startON, start0, level); + break; + + case 4: /* EN/AN before R for NUMBERS_SPECIAL */ + level=pLevState->runLevel+2; + setLevelsOutsideIsolates(pBiDi, pLevState->startON, start0, level); + break; + + case 5: /* L or S after possible relevant EN/AN */ + /* check if we had EN after R/AL */ + if (pLevState->startL2EN >= 0) { + addPoint(pBiDi, pLevState->startL2EN, LRM_BEFORE); + } + pLevState->startL2EN=-1; /* not within previous if since could also be -2 */ + /* check if we had any relevant EN/AN after R/AL */ + pInsertPoints=&(pBiDi->insertPoints); + if ((pInsertPoints->capacity == 0) || + (pInsertPoints->size <= pInsertPoints->confirmed)) + { + /* nothing, just clean up */ + pLevState->lastStrongRTL=-1; + /* check if we have a pending conditional segment */ + level=(*pImpTab)[oldStateSeq][IMPTABLEVELS_RES]; + if ((level & 1) && (pLevState->startON > 0)) { /* after ON */ + start=pLevState->startON; /* reset to basic run level */ + } + if (_prop == DirProp_S) /* add LRM before S */ + { + addPoint(pBiDi, start0, LRM_BEFORE); + pInsertPoints->confirmed=pInsertPoints->size; + } + break; + } + /* reset previous RTL cont to level for LTR text */ + for (k=pLevState->lastStrongRTL+1; kconfirmed=pInsertPoints->size; + pLevState->lastStrongRTL=-1; + if (_prop == DirProp_S) /* add LRM before S */ + { + addPoint(pBiDi, start0, LRM_BEFORE); + pInsertPoints->confirmed=pInsertPoints->size; + } + break; + + case 6: /* R/AL after possible relevant EN/AN */ + /* just clean up */ + pInsertPoints=&(pBiDi->insertPoints); + if (pInsertPoints->capacity > 0) + /* remove all non confirmed insert points */ + pInsertPoints->size=pInsertPoints->confirmed; + pLevState->startON=-1; + pLevState->startL2EN=-1; + pLevState->lastStrongRTL=limit - 1; + break; + + case 7: /* EN/AN after R/AL + possible cont */ + /* check for real AN */ + if ((_prop == DirProp_AN) && (pBiDi->dirProps[start0] == AN) && + (pBiDi->reorderingMode!=UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL)) + { + /* real AN */ + if (pLevState->startL2EN == -1) /* if no relevant EN already found */ + { + /* just note the righmost digit as a strong RTL */ + pLevState->lastStrongRTL=limit - 1; + break; + } + if (pLevState->startL2EN >= 0) /* after EN, no AN */ + { + addPoint(pBiDi, pLevState->startL2EN, LRM_BEFORE); + pLevState->startL2EN=-2; + } + /* note AN */ + addPoint(pBiDi, start0, LRM_BEFORE); + break; + } + /* if first EN/AN after R/AL */ + if (pLevState->startL2EN == -1) { + pLevState->startL2EN=start0; + } + break; + + case 8: /* note location of latest R/AL */ + pLevState->lastStrongRTL=limit - 1; + pLevState->startON=-1; + break; + + case 9: /* L after R+ON/EN/AN */ + /* include possible adjacent number on the left */ + for (k=start0-1; k>=0 && !(levels[k]&1); k--); + if(k>=0) { + addPoint(pBiDi, k, RLM_BEFORE); /* add RLM before */ + pInsertPoints=&(pBiDi->insertPoints); + pInsertPoints->confirmed=pInsertPoints->size; /* confirm it */ + } + pLevState->startON=start0; + break; + + case 10: /* AN after L */ + /* AN numbers between L text on both sides may be trouble. */ + /* tentatively bracket with LRMs; will be confirmed if followed by L */ + addPoint(pBiDi, start0, LRM_BEFORE); /* add LRM before */ + addPoint(pBiDi, start0, LRM_AFTER); /* add LRM after */ + break; + + case 11: /* R after L+ON/EN/AN */ + /* false alert, infirm LRMs around previous AN */ + pInsertPoints=&(pBiDi->insertPoints); + pInsertPoints->size=pInsertPoints->confirmed; + if (_prop == DirProp_S) /* add RLM before S */ + { + addPoint(pBiDi, start0, RLM_BEFORE); + pInsertPoints->confirmed=pInsertPoints->size; + } + break; + + case 12: /* L after L+ON/AN */ + level=pLevState->runLevel + addLevel; + for(k=pLevState->startON; kinsertPoints); + pInsertPoints->confirmed=pInsertPoints->size; /* confirm inserts */ + pLevState->startON=start0; + break; + + case 13: /* L after L+ON+EN/AN/ON */ + level=pLevState->runLevel; + for(k=start0-1; k>=pLevState->startON; k--) { + if(levels[k]==level+3) { + while(levels[k]==level+3) { + levels[k--]-=2; + } + while(levels[k]==level) { + k--; + } + } + if(levels[k]==level+2) { + levels[k]=level; + continue; + } + levels[k]=level+1; + } + break; + + case 14: /* R after L+ON+EN/AN/ON */ + level=pLevState->runLevel+1; + for(k=start0-1; k>=pLevState->startON; k--) { + if(levels[k]>level) { + levels[k]-=2; + } + } + break; + + default: /* we should never get here */ + UPRV_UNREACHABLE_EXIT; + } + } + if((addLevel) || (start < start0)) { + level=pLevState->runLevel + addLevel; + if(start>=pLevState->runStart) { + for(k=start; kprologue; + int32_t length=pBiDi->proLength; + int32_t i; + UChar32 uchar; + DirProp dirProp; + for(i=length; i>0; ) { + /* i is decremented by U16_PREV */ + U16_PREV(text, 0, i, uchar); + dirProp=(DirProp)ubidi_getCustomizedClass(pBiDi, uchar); + if(dirProp==L) { + return DirProp_L; + } + if(dirProp==R || dirProp==AL) { + return DirProp_R; + } + if(dirProp==B) { + return DirProp_ON; + } + } + return DirProp_ON; +} + +/** + * Returns the directionality of the first strong character, or digit, in the epilogue, if any. + * Requires epilogue!=null. + */ +static DirProp +firstL_R_AL_EN_AN(UBiDi *pBiDi) { + const char16_t *text=pBiDi->epilogue; + int32_t length=pBiDi->epiLength; + int32_t i; + UChar32 uchar; + DirProp dirProp; + for(i=0; idirProps; + DirProp dirProp; + LevState levState; + int32_t i, start1, start2; + uint16_t oldStateImp, stateImp, actionImp; + uint8_t gprop, resProp, cell; + UBool inverseRTL; + DirProp nextStrongProp=R; + int32_t nextStrongPos=-1; + + /* check for RTL inverse BiDi mode */ + /* FOOD FOR THOUGHT: in case of RTL inverse BiDi, it would make sense to + * loop on the text characters from end to start. + * This would need a different properties state table (at least different + * actions) and different levels state tables (maybe very similar to the + * LTR corresponding ones. + */ + inverseRTL=(UBool) + ((startlastArabicPos) && (GET_PARALEVEL(pBiDi, start) & 1) && + (pBiDi->reorderingMode==UBIDI_REORDER_INVERSE_LIKE_DIRECT || + pBiDi->reorderingMode==UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL)); + + /* initialize for property and levels state tables */ + levState.startL2EN=-1; /* used for INVERSE_LIKE_DIRECT_WITH_MARKS */ + levState.lastStrongRTL=-1; /* used for INVERSE_LIKE_DIRECT_WITH_MARKS */ + levState.runStart=start; + levState.runLevel=pBiDi->levels[start]; + levState.pImpTab=(const ImpTab*)((pBiDi->pImpTabPair)->pImpTab)[levState.runLevel&1]; + levState.pImpAct=(const ImpAct*)((pBiDi->pImpTabPair)->pImpAct)[levState.runLevel&1]; + if(start==0 && pBiDi->proLength>0) { + DirProp lastStrong=lastL_R_AL(pBiDi); + if(lastStrong!=DirProp_ON) { + sor=lastStrong; + } + } + /* The isolates[] entries contain enough information to + resume the bidi algorithm in the same state as it was + when it was interrupted by an isolate sequence. */ + if(dirProps[start]==PDI && pBiDi->isolateCount >= 0) { + levState.startON=pBiDi->isolates[pBiDi->isolateCount].startON; + start1=pBiDi->isolates[pBiDi->isolateCount].start1; + stateImp=pBiDi->isolates[pBiDi->isolateCount].stateImp; + levState.state=pBiDi->isolates[pBiDi->isolateCount].state; + pBiDi->isolateCount--; + } else { + levState.startON=-1; + start1=start; + if(dirProps[start]==NSM) + stateImp = 1 + sor; + else + stateImp=0; + levState.state=0; + processPropertySeq(pBiDi, &levState, sor, start, start); + } + start2=start; /* to make Java compiler happy */ + + for(i=start; i<=limit; i++) { + if(i>=limit) { + int32_t k; + for(k=limit-1; k>start&&(DIRPROP_FLAG(dirProps[k])&MASK_BN_EXPLICIT); k--); + dirProp=dirProps[k]; + if(dirProp==LRI || dirProp==RLI) + break; /* no forced closing for sequence ending with LRI/RLI */ + gprop=eor; + } else { + DirProp prop, prop1; + prop=dirProps[i]; + if(prop==B) { + pBiDi->isolateCount=-1; /* current isolates stack entry == none */ + } + if(inverseRTL) { + if(prop==AL) { + /* AL before EN does not make it AN */ + prop=R; + } else if(prop==EN) { + if(nextStrongPos<=i) { + /* look for next strong char (L/R/AL) */ + int32_t j; + nextStrongProp=R; /* set default */ + nextStrongPos=limit; + for(j=i+1; jlength && pBiDi->epiLength>0) { + DirProp firstStrong=firstL_R_AL_EN_AN(pBiDi); + if(firstStrong!=DirProp_ON) { + eor=firstStrong; + } + } + + /* look for the last char not a BN or LRE/RLE/LRO/RLO/PDF */ + for(i=limit-1; i>start&&(DIRPROP_FLAG(dirProps[i])&MASK_BN_EXPLICIT); i--); + dirProp=dirProps[i]; + if((dirProp==LRI || dirProp==RLI) && limitlength) { + pBiDi->isolateCount++; + pBiDi->isolates[pBiDi->isolateCount].stateImp=stateImp; + pBiDi->isolates[pBiDi->isolateCount].state=levState.state; + pBiDi->isolates[pBiDi->isolateCount].start1=start1; + pBiDi->isolates[pBiDi->isolateCount].startON=levState.startON; + } + else + processPropertySeq(pBiDi, &levState, eor, limit, limit); +} + +/* perform (L1) and (X9) ---------------------------------------------------- */ + +/* + * Reset the embedding levels for some non-graphic characters (L1). + * This function also sets appropriate levels for BN, and + * explicit embedding types that are supposed to have been removed + * from the paragraph in (X9). + */ +static void +adjustWSLevels(UBiDi *pBiDi) { + const DirProp *dirProps=pBiDi->dirProps; + UBiDiLevel *levels=pBiDi->levels; + int32_t i; + + if(pBiDi->flags&MASK_WS) { + UBool orderParagraphsLTR=pBiDi->orderParagraphsLTR; + Flags flag; + + i=pBiDi->trailingWSStart; + while(i>0) { + /* reset a sequence of WS/BN before eop and B/S to the paragraph paraLevel */ + while(i>0 && (flag=DIRPROP_FLAG(dirProps[--i]))&MASK_WS) { + if(orderParagraphsLTR&&(flag&DIRPROP_FLAG(B))) { + levels[i]=0; + } else { + levels[i]=GET_PARALEVEL(pBiDi, i); + } + } + + /* reset BN to the next character's paraLevel until B/S, which restarts above loop */ + /* here, i+1 is guaranteed to be 0) { + flag=DIRPROP_FLAG(dirProps[--i]); + if(flag&MASK_BN_EXPLICIT) { + levels[i]=levels[i+1]; + } else if(orderParagraphsLTR&&(flag&DIRPROP_FLAG(B))) { + levels[i]=0; + break; + } else if(flag&MASK_B_S) { + levels[i]=GET_PARALEVEL(pBiDi, i); + break; + } + } + } + } +} + +U_CAPI void U_EXPORT2 +ubidi_setContext(UBiDi *pBiDi, + const char16_t *prologue, int32_t proLength, + const char16_t *epilogue, int32_t epiLength, + UErrorCode *pErrorCode) { + /* check the argument values */ + RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrorCode); + if(pBiDi==nullptr || proLength<-1 || epiLength<-1 || + (prologue==nullptr && proLength!=0) || (epilogue==nullptr && epiLength!=0)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + if(proLength==-1) { + pBiDi->proLength=u_strlen(prologue); + } else { + pBiDi->proLength=proLength; + } + if(epiLength==-1) { + pBiDi->epiLength=u_strlen(epilogue); + } else { + pBiDi->epiLength=epiLength; + } + pBiDi->prologue=prologue; + pBiDi->epilogue=epilogue; +} + +static void +setParaSuccess(UBiDi *pBiDi) { + pBiDi->proLength=0; /* forget the last context */ + pBiDi->epiLength=0; + pBiDi->pParaBiDi=pBiDi; /* mark successful setPara */ +} + +#define BIDI_MIN(x, y) ((x)<(y) ? (x) : (y)) +#define BIDI_ABS(x) ((x)>=0 ? (x) : (-(x))) + +static void +setParaRunsOnly(UBiDi *pBiDi, const char16_t *text, int32_t length, + UBiDiLevel paraLevel, UErrorCode *pErrorCode) { + int32_t *runsOnlyMemory = nullptr; + int32_t *visualMap; + char16_t *visualText; + int32_t saveLength, saveTrailingWSStart; + const UBiDiLevel *levels; + UBiDiLevel *saveLevels; + UBiDiDirection saveDirection; + UBool saveMayAllocateText; + Run *runs; + int32_t visualLength, i, j, visualStart, logicalStart, + runCount, runLength, addedRuns, insertRemove, + start, limit, step, indexOddBit, logicalPos, + index0, index1; + uint32_t saveOptions; + + pBiDi->reorderingMode=UBIDI_REORDER_DEFAULT; + if(length==0) { + ubidi_setPara(pBiDi, text, length, paraLevel, nullptr, pErrorCode); + goto cleanup3; + } + /* obtain memory for mapping table and visual text */ + runsOnlyMemory=static_cast(uprv_malloc(length*(sizeof(int32_t)+sizeof(char16_t)+sizeof(UBiDiLevel)))); + if(runsOnlyMemory==nullptr) { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + goto cleanup3; + } + visualMap=runsOnlyMemory; + visualText=(char16_t *)&visualMap[length]; + saveLevels=(UBiDiLevel *)&visualText[length]; + saveOptions=pBiDi->reorderingOptions; + if(saveOptions & UBIDI_OPTION_INSERT_MARKS) { + pBiDi->reorderingOptions&=~UBIDI_OPTION_INSERT_MARKS; + pBiDi->reorderingOptions|=UBIDI_OPTION_REMOVE_CONTROLS; + } + paraLevel&=1; /* accept only 0 or 1 */ + ubidi_setPara(pBiDi, text, length, paraLevel, nullptr, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + goto cleanup3; + } + /* we cannot access directly pBiDi->levels since it is not yet set if + * direction is not MIXED + */ + levels=ubidi_getLevels(pBiDi, pErrorCode); + uprv_memcpy(saveLevels, levels, (size_t)pBiDi->length*sizeof(UBiDiLevel)); + saveTrailingWSStart=pBiDi->trailingWSStart; + saveLength=pBiDi->length; + saveDirection=pBiDi->direction; + + /* FOOD FOR THOUGHT: instead of writing the visual text, we could use + * the visual map and the dirProps array to drive the second call + * to ubidi_setPara (but must make provision for possible removal of + * BiDi controls. Alternatively, only use the dirProps array via + * customized classifier callback. + */ + visualLength=ubidi_writeReordered(pBiDi, visualText, length, + UBIDI_DO_MIRRORING, pErrorCode); + ubidi_getVisualMap(pBiDi, visualMap, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + goto cleanup2; + } + pBiDi->reorderingOptions=saveOptions; + + pBiDi->reorderingMode=UBIDI_REORDER_INVERSE_LIKE_DIRECT; + paraLevel^=1; + /* Because what we did with reorderingOptions, visualText may be shorter + * than the original text. But we don't want the levels memory to be + * reallocated shorter than the original length, since we need to restore + * the levels as after the first call to ubidi_setpara() before returning. + * We will force mayAllocateText to false before the second call to + * ubidi_setpara(), and will restore it afterwards. + */ + saveMayAllocateText=pBiDi->mayAllocateText; + pBiDi->mayAllocateText=false; + ubidi_setPara(pBiDi, visualText, visualLength, paraLevel, nullptr, pErrorCode); + pBiDi->mayAllocateText=saveMayAllocateText; + ubidi_getRuns(pBiDi, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + goto cleanup1; + } + /* check if some runs must be split, count how many splits */ + addedRuns=0; + runCount=pBiDi->runCount; + runs=pBiDi->runs; + visualStart=0; + for(i=0; irunsMemory[0]=runs[0]; + } + runs=pBiDi->runs=pBiDi->runsMemory; + pBiDi->runCount+=addedRuns; + } else { + goto cleanup1; + } + } + /* split runs which are not consecutive in source text */ + for(i=runCount-1; i>=0; i--) { + runLength= i==0 ? runs[0].visualLimit : + runs[i].visualLimit-runs[i-1].visualLimit; + logicalStart=runs[i].logicalStart; + indexOddBit=GET_ODD_BIT(logicalStart); + logicalStart=GET_INDEX(logicalStart); + if(runLength<2) { + if(addedRuns) { + runs[i+addedRuns]=runs[i]; + } + logicalPos=visualMap[logicalStart]; + runs[i+addedRuns].logicalStart=MAKE_INDEX_ODD_PAIR(logicalPos, + saveLevels[logicalPos]^indexOddBit); + continue; + } + if(indexOddBit) { + start=logicalStart; + limit=logicalStart+runLength-1; + step=1; + } else { + start=logicalStart+runLength-1; + limit=logicalStart; + step=-1; + } + for(j=start; j!=limit; j+=step) { + index0=visualMap[j]; + index1=visualMap[j+step]; + if((BIDI_ABS(index0-index1)!=1) || (saveLevels[index0]!=saveLevels[index1])) { + logicalPos=BIDI_MIN(visualMap[start], index0); + runs[i+addedRuns].logicalStart=MAKE_INDEX_ODD_PAIR(logicalPos, + saveLevels[logicalPos]^indexOddBit); + runs[i+addedRuns].visualLimit=runs[i].visualLimit; + runs[i].visualLimit-=BIDI_ABS(j-start)+1; + insertRemove=runs[i].insertRemove&(LRM_AFTER|RLM_AFTER); + runs[i+addedRuns].insertRemove=insertRemove; + runs[i].insertRemove&=~insertRemove; + start=j+step; + addedRuns--; + } + } + if(addedRuns) { + runs[i+addedRuns]=runs[i]; + } + logicalPos=BIDI_MIN(visualMap[start], visualMap[limit]); + runs[i+addedRuns].logicalStart=MAKE_INDEX_ODD_PAIR(logicalPos, + saveLevels[logicalPos]^indexOddBit); + } + + cleanup1: + /* restore initial paraLevel */ + pBiDi->paraLevel^=1; + cleanup2: + /* restore real text */ + pBiDi->text=text; + pBiDi->length=saveLength; + pBiDi->originalLength=length; + pBiDi->direction=saveDirection; + /* the saved levels should never excess levelsSize, but we check anyway */ + if(saveLength>pBiDi->levelsSize) { + saveLength=pBiDi->levelsSize; + } + uprv_memcpy(pBiDi->levels, saveLevels, (size_t)saveLength*sizeof(UBiDiLevel)); + pBiDi->trailingWSStart=saveTrailingWSStart; + if(pBiDi->runCount>1) { + pBiDi->direction=UBIDI_MIXED; + } + cleanup3: + /* free memory for mapping table and visual text */ + uprv_free(runsOnlyMemory); + + pBiDi->reorderingMode=UBIDI_REORDER_RUNS_ONLY; +} + +/* ubidi_setPara ------------------------------------------------------------ */ + +U_CAPI void U_EXPORT2 +ubidi_setPara(UBiDi *pBiDi, const char16_t *text, int32_t length, + UBiDiLevel paraLevel, UBiDiLevel *embeddingLevels, + UErrorCode *pErrorCode) { + UBiDiDirection direction; + DirProp *dirProps; + + /* check the argument values */ + RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrorCode); + if(pBiDi==nullptr || text==nullptr || length<-1 || + (paraLevel>UBIDI_MAX_EXPLICIT_LEVEL && paraLevelreorderingMode==UBIDI_REORDER_RUNS_ONLY) { + setParaRunsOnly(pBiDi, text, length, paraLevel, pErrorCode); + return; + } + + /* initialize the UBiDi structure */ + pBiDi->pParaBiDi=nullptr; /* mark unfinished setPara */ + pBiDi->text=text; + pBiDi->length=pBiDi->originalLength=pBiDi->resultLength=length; + pBiDi->paraLevel=paraLevel; + pBiDi->direction=(UBiDiDirection)(paraLevel&1); + pBiDi->paraCount=1; + + pBiDi->dirProps=nullptr; + pBiDi->levels=nullptr; + pBiDi->runs=nullptr; + pBiDi->insertPoints.size=0; /* clean up from last call */ + pBiDi->insertPoints.confirmed=0; /* clean up from last call */ + + /* + * Save the original paraLevel if contextual; otherwise, set to 0. + */ + pBiDi->defaultParaLevel=IS_DEFAULT_LEVEL(paraLevel); + + if(length==0) { + /* + * For an empty paragraph, create a UBiDi object with the paraLevel and + * the flags and the direction set but without allocating zero-length arrays. + * There is nothing more to do. + */ + if(IS_DEFAULT_LEVEL(paraLevel)) { + pBiDi->paraLevel&=1; + pBiDi->defaultParaLevel=0; + } + pBiDi->flags=DIRPROP_FLAG_LR(paraLevel); + pBiDi->runCount=0; + pBiDi->paraCount=0; + setParaSuccess(pBiDi); /* mark successful setPara */ + return; + } + + pBiDi->runCount=-1; + + /* allocate paras memory */ + if(pBiDi->parasMemory) + pBiDi->paras=pBiDi->parasMemory; + else + pBiDi->paras=pBiDi->simpleParas; + + /* + * Get the directional properties, + * the flags bit-set, and + * determine the paragraph level if necessary. + */ + if(getDirPropsMemory(pBiDi, length)) { + pBiDi->dirProps=pBiDi->dirPropsMemory; + if(!getDirProps(pBiDi)) { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + } else { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + dirProps=pBiDi->dirProps; + /* the processed length may have changed if UBIDI_OPTION_STREAMING */ + length= pBiDi->length; + pBiDi->trailingWSStart=length; /* the levels[] will reflect the WS run */ + + /* are explicit levels specified? */ + if(embeddingLevels==nullptr) { + /* no: determine explicit levels according to the (Xn) rules */\ + if(getLevelsMemory(pBiDi, length)) { + pBiDi->levels=pBiDi->levelsMemory; + direction=resolveExplicitLevels(pBiDi, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + return; + } + } else { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + } else { + /* set BN for all explicit codes, check that all levels are 0 or paraLevel..UBIDI_MAX_EXPLICIT_LEVEL */ + pBiDi->levels=embeddingLevels; + direction=checkExplicitLevels(pBiDi, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + return; + } + } + + /* allocate isolate memory */ + if(pBiDi->isolateCount<=SIMPLE_ISOLATES_COUNT) + pBiDi->isolates=pBiDi->simpleIsolates; + else + if((int32_t)(pBiDi->isolateCount*sizeof(Isolate))<=pBiDi->isolatesSize) + pBiDi->isolates=pBiDi->isolatesMemory; + else { + if(getInitialIsolatesMemory(pBiDi, pBiDi->isolateCount)) { + pBiDi->isolates=pBiDi->isolatesMemory; + } else { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + } + pBiDi->isolateCount=-1; /* current isolates stack entry == none */ + + /* + * The steps after (X9) in the UBiDi algorithm are performed only if + * the paragraph text has mixed directionality! + */ + pBiDi->direction=direction; + switch(direction) { + case UBIDI_LTR: + /* all levels are implicitly at paraLevel (important for ubidi_getLevels()) */ + pBiDi->trailingWSStart=0; + break; + case UBIDI_RTL: + /* all levels are implicitly at paraLevel (important for ubidi_getLevels()) */ + pBiDi->trailingWSStart=0; + break; + default: + /* + * Choose the right implicit state table + */ + switch(pBiDi->reorderingMode) { + case UBIDI_REORDER_DEFAULT: + pBiDi->pImpTabPair=&impTab_DEFAULT; + break; + case UBIDI_REORDER_NUMBERS_SPECIAL: + pBiDi->pImpTabPair=&impTab_NUMBERS_SPECIAL; + break; + case UBIDI_REORDER_GROUP_NUMBERS_WITH_R: + pBiDi->pImpTabPair=&impTab_GROUP_NUMBERS_WITH_R; + break; + case UBIDI_REORDER_INVERSE_NUMBERS_AS_L: + pBiDi->pImpTabPair=&impTab_INVERSE_NUMBERS_AS_L; + break; + case UBIDI_REORDER_INVERSE_LIKE_DIRECT: + if (pBiDi->reorderingOptions & UBIDI_OPTION_INSERT_MARKS) { + pBiDi->pImpTabPair=&impTab_INVERSE_LIKE_DIRECT_WITH_MARKS; + } else { + pBiDi->pImpTabPair=&impTab_INVERSE_LIKE_DIRECT; + } + break; + case UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL: + if (pBiDi->reorderingOptions & UBIDI_OPTION_INSERT_MARKS) { + pBiDi->pImpTabPair=&impTab_INVERSE_FOR_NUMBERS_SPECIAL_WITH_MARKS; + } else { + pBiDi->pImpTabPair=&impTab_INVERSE_FOR_NUMBERS_SPECIAL; + } + break; + default: + /* we should never get here */ + UPRV_UNREACHABLE_EXIT; + } + /* + * If there are no external levels specified and there + * are no significant explicit level codes in the text, + * then we can treat the entire paragraph as one run. + * Otherwise, we need to perform the following rules on runs of + * the text with the same embedding levels. (X10) + * "Significant" explicit level codes are ones that actually + * affect non-BN characters. + * Examples for "insignificant" ones are empty embeddings + * LRE-PDF, LRE-RLE-PDF-PDF, etc. + */ + if(embeddingLevels==nullptr && pBiDi->paraCount<=1 && + !(pBiDi->flags&DIRPROP_FLAG_MULTI_RUNS)) { + resolveImplicitLevels(pBiDi, 0, length, + GET_LR_FROM_LEVEL(GET_PARALEVEL(pBiDi, 0)), + GET_LR_FROM_LEVEL(GET_PARALEVEL(pBiDi, length-1))); + } else { + /* sor, eor: start and end types of same-level-run */ + UBiDiLevel *levels=pBiDi->levels; + int32_t start, limit=0; + UBiDiLevel level, nextLevel; + DirProp sor, eor; + + /* determine the first sor and set eor to it because of the loop body (sor=eor there) */ + level=GET_PARALEVEL(pBiDi, 0); + nextLevel=levels[0]; + if(level0) && (dirProps[start-1]==B)) { + /* except if this is a new paragraph, then set sor = para level */ + sor=GET_LR_FROM_LEVEL(GET_PARALEVEL(pBiDi, start)); + } else { + sor=eor; + } + + /* search for the limit of this run */ + while((++limitinsertPoints.errorCode)) + { + *pErrorCode=pBiDi->insertPoints.errorCode; + return; + } + /* reset the embedding levels for some non-graphic characters (L1), (X9) */ + adjustWSLevels(pBiDi); + break; + } + /* add RLM for inverse Bidi with contextual orientation resolving + * to RTL which would not round-trip otherwise + */ + if((pBiDi->defaultParaLevel>0) && + (pBiDi->reorderingOptions & UBIDI_OPTION_INSERT_MARKS) && + ((pBiDi->reorderingMode==UBIDI_REORDER_INVERSE_LIKE_DIRECT) || + (pBiDi->reorderingMode==UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL))) { + int32_t i, j, start, last; + UBiDiLevel level; + DirProp dirProp; + for(i=0; iparaCount; i++) { + last=(pBiDi->paras[i].limit)-1; + level= static_cast(pBiDi->paras[i].level); + if(level==0) + continue; /* LTR paragraph */ + start= i==0 ? 0 : pBiDi->paras[i-1].limit; + for(j=last; j>=start; j--) { + dirProp=dirProps[j]; + if(dirProp==L) { + if(jreorderingOptions & UBIDI_OPTION_REMOVE_CONTROLS) { + pBiDi->resultLength -= pBiDi->controlCount; + } else { + pBiDi->resultLength += pBiDi->insertPoints.size; + } + setParaSuccess(pBiDi); /* mark successful setPara */ +} + +U_CAPI void U_EXPORT2 +ubidi_orderParagraphsLTR(UBiDi *pBiDi, UBool orderParagraphsLTR) { + if(pBiDi!=nullptr) { + pBiDi->orderParagraphsLTR=orderParagraphsLTR; + } +} + +U_CAPI UBool U_EXPORT2 +ubidi_isOrderParagraphsLTR(UBiDi *pBiDi) { + if(pBiDi!=nullptr) { + return pBiDi->orderParagraphsLTR; + } else { + return false; + } +} + +U_CAPI UBiDiDirection U_EXPORT2 +ubidi_getDirection(const UBiDi *pBiDi) { + if(IS_VALID_PARA_OR_LINE(pBiDi)) { + return pBiDi->direction; + } else { + return UBIDI_LTR; + } +} + +U_CAPI const char16_t * U_EXPORT2 +ubidi_getText(const UBiDi *pBiDi) { + if(IS_VALID_PARA_OR_LINE(pBiDi)) { + return pBiDi->text; + } else { + return nullptr; + } +} + +U_CAPI int32_t U_EXPORT2 +ubidi_getLength(const UBiDi *pBiDi) { + if(IS_VALID_PARA_OR_LINE(pBiDi)) { + return pBiDi->originalLength; + } else { + return 0; + } +} + +U_CAPI int32_t U_EXPORT2 +ubidi_getProcessedLength(const UBiDi *pBiDi) { + if(IS_VALID_PARA_OR_LINE(pBiDi)) { + return pBiDi->length; + } else { + return 0; + } +} + +U_CAPI int32_t U_EXPORT2 +ubidi_getResultLength(const UBiDi *pBiDi) { + if(IS_VALID_PARA_OR_LINE(pBiDi)) { + return pBiDi->resultLength; + } else { + return 0; + } +} + +/* paragraphs API functions ------------------------------------------------- */ + +U_CAPI UBiDiLevel U_EXPORT2 +ubidi_getParaLevel(const UBiDi *pBiDi) { + if(IS_VALID_PARA_OR_LINE(pBiDi)) { + return pBiDi->paraLevel; + } else { + return 0; + } +} + +U_CAPI int32_t U_EXPORT2 +ubidi_countParagraphs(UBiDi *pBiDi) { + if(!IS_VALID_PARA_OR_LINE(pBiDi)) { + return 0; + } else { + return pBiDi->paraCount; + } +} + +U_CAPI void U_EXPORT2 +ubidi_getParagraphByIndex(const UBiDi *pBiDi, int32_t paraIndex, + int32_t *pParaStart, int32_t *pParaLimit, + UBiDiLevel *pParaLevel, UErrorCode *pErrorCode) { + int32_t paraStart; + + /* check the argument values */ + RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrorCode); + RETURN_VOID_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode); + RETURN_VOID_IF_BAD_RANGE(paraIndex, 0, pBiDi->paraCount, *pErrorCode); + + pBiDi=pBiDi->pParaBiDi; /* get Para object if Line object */ + if(paraIndex) { + paraStart=pBiDi->paras[paraIndex-1].limit; + } else { + paraStart=0; + } + if(pParaStart!=nullptr) { + *pParaStart=paraStart; + } + if(pParaLimit!=nullptr) { + *pParaLimit=pBiDi->paras[paraIndex].limit; + } + if(pParaLevel!=nullptr) { + *pParaLevel=GET_PARALEVEL(pBiDi, paraStart); + } +} + +U_CAPI int32_t U_EXPORT2 +ubidi_getParagraph(const UBiDi *pBiDi, int32_t charIndex, + int32_t *pParaStart, int32_t *pParaLimit, + UBiDiLevel *pParaLevel, UErrorCode *pErrorCode) { + int32_t paraIndex; + + /* check the argument values */ + /* pErrorCode will be checked by the call to ubidi_getParagraphByIndex */ + RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrorCode, -1); + RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode, -1); + pBiDi=pBiDi->pParaBiDi; /* get Para object if Line object */ + RETURN_IF_BAD_RANGE(charIndex, 0, pBiDi->length, *pErrorCode, -1); + + for(paraIndex=0; charIndex>=pBiDi->paras[paraIndex].limit; paraIndex++); + ubidi_getParagraphByIndex(pBiDi, paraIndex, pParaStart, pParaLimit, pParaLevel, pErrorCode); + return paraIndex; +} + +U_CAPI void U_EXPORT2 +ubidi_setClassCallback(UBiDi *pBiDi, UBiDiClassCallback *newFn, + const void *newContext, UBiDiClassCallback **oldFn, + const void **oldContext, UErrorCode *pErrorCode) +{ + RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrorCode); + if(pBiDi==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + if( oldFn ) + { + *oldFn = pBiDi->fnClassCallback; + } + if( oldContext ) + { + *oldContext = pBiDi->coClassCallback; + } + pBiDi->fnClassCallback = newFn; + pBiDi->coClassCallback = newContext; +} + +U_CAPI void U_EXPORT2 +ubidi_getClassCallback(UBiDi *pBiDi, UBiDiClassCallback **fn, const void **context) +{ + if(pBiDi==nullptr) { + return; + } + if( fn ) + { + *fn = pBiDi->fnClassCallback; + } + if( context ) + { + *context = pBiDi->coClassCallback; + } +} + +U_CAPI UCharDirection U_EXPORT2 +ubidi_getCustomizedClass(UBiDi *pBiDi, UChar32 c) +{ + UCharDirection dir; + + if( pBiDi->fnClassCallback == nullptr || + (dir = (*pBiDi->fnClassCallback)(pBiDi->coClassCallback, c)) == U_BIDI_CLASS_DEFAULT ) + { + dir = ubidi_getClass(c); + } + if(dir >= U_CHAR_DIRECTION_COUNT) { + dir = (UCharDirection)ON; + } + return dir; +} diff --git a/intl/icu/source/common/ubidi_props.cpp b/intl/icu/source/common/ubidi_props.cpp new file mode 100644 index 0000000000..8f3f6a65c4 --- /dev/null +++ b/intl/icu/source/common/ubidi_props.cpp @@ -0,0 +1,254 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2004-2014, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: ubidi_props.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2004dec30 +* created by: Markus W. Scherer +* +* Low-level Unicode bidi/shaping properties access. +*/ + +#include "unicode/utypes.h" +#include "unicode/uset.h" +#include "unicode/udata.h" /* UDataInfo */ +#include "ucmndata.h" /* DataHeader */ +#include "udatamem.h" +#include "uassert.h" +#include "cmemory.h" +#include "utrie2.h" +#include "ubidi_props.h" +#include "ucln_cmn.h" + +struct UBiDiProps { + UDataMemory *mem; + const int32_t *indexes; + const uint32_t *mirrors; + const uint8_t *jgArray; + const uint8_t *jgArray2; + + UTrie2 trie; + uint8_t formatVersion[4]; +}; + +/* ubidi_props_data.h is machine-generated by genbidi --csource */ +#define INCLUDED_FROM_UBIDI_PROPS_C +#include "ubidi_props_data.h" + +/* set of property starts for UnicodeSet ------------------------------------ */ + +static UBool U_CALLCONV +_enumPropertyStartsRange(const void *context, UChar32 start, UChar32 end, uint32_t value) { + (void)end; + (void)value; + /* add the start code point to the USet */ + const USetAdder *sa=(const USetAdder *)context; + sa->add(sa->set, start); + return true; +} + +U_CFUNC void +ubidi_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) { + int32_t i, length; + UChar32 c, start, limit; + + const uint8_t *jgArray; + uint8_t prev, jg; + + if(U_FAILURE(*pErrorCode)) { + return; + } + + /* add the start code point of each same-value range of the trie */ + utrie2_enum(&ubidi_props_singleton.trie, nullptr, _enumPropertyStartsRange, sa); + + /* add the code points from the bidi mirroring table */ + length=ubidi_props_singleton.indexes[UBIDI_IX_MIRROR_LENGTH]; + for(i=0; iaddRange(sa->set, c, c+1); + } + + /* add the code points from the Joining_Group array where the value changes */ + start=ubidi_props_singleton.indexes[UBIDI_IX_JG_START]; + limit=ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT]; + jgArray=ubidi_props_singleton.jgArray; + for(;;) { + prev=0; + while(startadd(sa->set, start); + prev=jg; + } + ++start; + } + if(prev!=0) { + /* add the limit code point if the last value was not 0 (it is now start==limit) */ + sa->add(sa->set, limit); + } + if(limit==ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT]) { + /* switch to the second Joining_Group range */ + start=ubidi_props_singleton.indexes[UBIDI_IX_JG_START2]; + limit=ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT2]; + jgArray=ubidi_props_singleton.jgArray2; + } else { + break; + } + } + + /* add code points with hardcoded properties, plus the ones following them */ + + /* (none right now) */ +} + +/* property access functions ------------------------------------------------ */ + +U_CFUNC int32_t +ubidi_getMaxValue(UProperty which) { + int32_t max=ubidi_props_singleton.indexes[UBIDI_MAX_VALUES_INDEX]; + switch(which) { + case UCHAR_BIDI_CLASS: + return (max&UBIDI_CLASS_MASK); + case UCHAR_JOINING_GROUP: + return (max&UBIDI_MAX_JG_MASK)>>UBIDI_MAX_JG_SHIFT; + case UCHAR_JOINING_TYPE: + return (max&UBIDI_JT_MASK)>>UBIDI_JT_SHIFT; + case UCHAR_BIDI_PAIRED_BRACKET_TYPE: + return (max&UBIDI_BPT_MASK)>>UBIDI_BPT_SHIFT; + default: + return -1; /* undefined */ + } +} + +U_CAPI UCharDirection +ubidi_getClass(UChar32 c) { + uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c); + return (UCharDirection)UBIDI_GET_CLASS(props); +} + +U_CFUNC UBool +ubidi_isMirrored(UChar32 c) { + uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c); + return (UBool)UBIDI_GET_FLAG(props, UBIDI_IS_MIRRORED_SHIFT); +} + +static UChar32 +getMirror(UChar32 c, uint16_t props) { + int32_t delta=UBIDI_GET_MIRROR_DELTA(props); + if(delta!=UBIDI_ESC_MIRROR_DELTA) { + return c+delta; + } else { + /* look for mirror code point in the mirrors[] table */ + const uint32_t *mirrors; + uint32_t m; + int32_t i, length; + UChar32 c2; + + mirrors=ubidi_props_singleton.mirrors; + length=ubidi_props_singleton.indexes[UBIDI_IX_MIRROR_LENGTH]; + + /* linear search */ + for(i=0; i>UBIDI_JT_SHIFT); +} + +U_CFUNC UJoiningGroup +ubidi_getJoiningGroup(UChar32 c) { + UChar32 start, limit; + + start=ubidi_props_singleton.indexes[UBIDI_IX_JG_START]; + limit=ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT]; + if(start<=c && c>UBIDI_BPT_SHIFT); +} + +U_CFUNC UChar32 +ubidi_getPairedBracket(UChar32 c) { + uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c); + if((props&UBIDI_BPT_MASK)==0) { + return c; + } else { + return getMirror(c, props); + } +} + +/* public API (see uchar.h) ------------------------------------------------- */ + +U_CFUNC UCharDirection +u_charDirection(UChar32 c) { + return ubidi_getClass(c); +} + +U_CFUNC UBool +u_isMirrored(UChar32 c) { + return ubidi_isMirrored(c); +} + +U_CFUNC UChar32 +u_charMirror(UChar32 c) { + return ubidi_getMirror(c); +} + +U_CAPI UChar32 U_EXPORT2 +u_getBidiPairedBracket(UChar32 c) { + return ubidi_getPairedBracket(c); +} diff --git a/intl/icu/source/common/ubidi_props.h b/intl/icu/source/common/ubidi_props.h new file mode 100644 index 0000000000..698ee9c52b --- /dev/null +++ b/intl/icu/source/common/ubidi_props.h @@ -0,0 +1,148 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2004-2014, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: ubidi_props.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2004dec30 +* created by: Markus W. Scherer +* +* Low-level Unicode bidi/shaping properties access. +*/ + +#ifndef __UBIDI_PROPS_H__ +#define __UBIDI_PROPS_H__ + +#include "unicode/utypes.h" +#include "unicode/uset.h" +#include "putilimp.h" +#include "uset_imp.h" +#include "udataswp.h" + +U_CDECL_BEGIN + +/* library API -------------------------------------------------------------- */ + +U_CFUNC void +ubidi_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode); + +/* property access functions */ + +U_CFUNC int32_t +ubidi_getMaxValue(UProperty which); + +U_CAPI UCharDirection +ubidi_getClass(UChar32 c); + +U_CFUNC UBool +ubidi_isMirrored(UChar32 c); + +U_CFUNC UChar32 +ubidi_getMirror(UChar32 c); + +U_CFUNC UBool +ubidi_isBidiControl(UChar32 c); + +U_CFUNC UBool +ubidi_isJoinControl(UChar32 c); + +U_CFUNC UJoiningType +ubidi_getJoiningType(UChar32 c); + +U_CFUNC UJoiningGroup +ubidi_getJoiningGroup(UChar32 c); + +U_CFUNC UBidiPairedBracketType +ubidi_getPairedBracketType(UChar32 c); + +U_CFUNC UChar32 +ubidi_getPairedBracket(UChar32 c); + +/* file definitions --------------------------------------------------------- */ + +#define UBIDI_DATA_NAME "ubidi" +#define UBIDI_DATA_TYPE "icu" + +/* format "BiDi" */ +#define UBIDI_FMT_0 0x42 +#define UBIDI_FMT_1 0x69 +#define UBIDI_FMT_2 0x44 +#define UBIDI_FMT_3 0x69 + +/* indexes into indexes[] */ +enum { + UBIDI_IX_INDEX_TOP, + UBIDI_IX_LENGTH, + UBIDI_IX_TRIE_SIZE, + UBIDI_IX_MIRROR_LENGTH, + + UBIDI_IX_JG_START, + UBIDI_IX_JG_LIMIT, + UBIDI_IX_JG_START2, /* new in format version 2.2, ICU 54 */ + UBIDI_IX_JG_LIMIT2, + + UBIDI_MAX_VALUES_INDEX=15, + UBIDI_IX_TOP=16 +}; + +/* definitions for 16-bit bidi/shaping properties word ---------------------- */ + +enum { + /* UBIDI_CLASS_SHIFT=0, */ /* bidi class: 5 bits (4..0) */ + UBIDI_JT_SHIFT=5, /* joining type: 3 bits (7..5) */ + + UBIDI_BPT_SHIFT=8, /* Bidi_Paired_Bracket_Type(bpt): 2 bits (9..8) */ + + UBIDI_JOIN_CONTROL_SHIFT=10, + UBIDI_BIDI_CONTROL_SHIFT=11, + + UBIDI_IS_MIRRORED_SHIFT=12, /* 'is mirrored' */ + UBIDI_MIRROR_DELTA_SHIFT=13, /* bidi mirroring delta: 3 bits (15..13) */ + + UBIDI_MAX_JG_SHIFT=16 /* max JG value in indexes[UBIDI_MAX_VALUES_INDEX] bits 23..16 */ +}; + +#define UBIDI_CLASS_MASK 0x0000001f +#define UBIDI_JT_MASK 0x000000e0 +#define UBIDI_BPT_MASK 0x00000300 + +#define UBIDI_MAX_JG_MASK 0x00ff0000 + +#define UBIDI_GET_CLASS(props) ((props)&UBIDI_CLASS_MASK) +#define UBIDI_GET_FLAG(props, shift) (((props)>>(shift))&1) + +#if U_SIGNED_RIGHT_SHIFT_IS_ARITHMETIC +# define UBIDI_GET_MIRROR_DELTA(props) ((int16_t)(props)>>UBIDI_MIRROR_DELTA_SHIFT) +#else +# define UBIDI_GET_MIRROR_DELTA(props) (int16_t)(((props)&0x8000) ? (((props)>>UBIDI_MIRROR_DELTA_SHIFT)|0xe000) : ((props)>>UBIDI_MIRROR_DELTA_SHIFT)) +#endif + +enum { + UBIDI_ESC_MIRROR_DELTA=-4, + UBIDI_MIN_MIRROR_DELTA=-3, + UBIDI_MAX_MIRROR_DELTA=3 +}; + +/* definitions for 32-bit mirror table entry -------------------------------- */ + +enum { + /* the source Unicode code point takes 21 bits (20..0) */ + UBIDI_MIRROR_INDEX_SHIFT=21, + UBIDI_MAX_MIRROR_INDEX=0x7ff +}; + +#define UBIDI_GET_MIRROR_CODE_POINT(m) (UChar32)((m)&0x1fffff) + +#define UBIDI_GET_MIRROR_INDEX(m) ((m)>>UBIDI_MIRROR_INDEX_SHIFT) + +U_CDECL_END + +#endif diff --git a/intl/icu/source/common/ubidi_props_data.h b/intl/icu/source/common/ubidi_props_data.h new file mode 100644 index 0000000000..5dcd1d7c78 --- /dev/null +++ b/intl/icu/source/common/ubidi_props_data.h @@ -0,0 +1,950 @@ +// Copyright (C) 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// Copyright (C) 1999-2016, International Business Machines +// Corporation and others. All Rights Reserved. +// +// file name: ubidi_props_data.h +// +// machine-generated by: icu/tools/unicode/c/genprops/bidipropsbuilder.cpp + +#ifdef INCLUDED_FROM_UBIDI_PROPS_C + +static const UVersionInfo ubidi_props_dataVersion={0xf,0,0,0}; + +static const int32_t ubidi_props_indexes[UBIDI_IX_TOP]={0x10,0x6bc0,0x65d0,0x28,0x620,0x8cc,0x10ac0,0x10d24,0,0,0,0,0,0,0,0x6702b6}; + +static const uint16_t ubidi_props_trieIndex[13024]={ +0x387,0x38f,0x397,0x39f,0x3b7,0x3bf,0x3c7,0x3cf,0x3a7,0x3af,0x3a7,0x3af,0x3a7,0x3af,0x3a7,0x3af, +0x3a7,0x3af,0x3a7,0x3af,0x3d5,0x3dd,0x3e5,0x3ed,0x3f5,0x3fd,0x3f9,0x401,0x409,0x411,0x40c,0x414, +0x3a7,0x3af,0x3a7,0x3af,0x41c,0x424,0x3a7,0x3af,0x3a7,0x3af,0x3a7,0x3af,0x42a,0x432,0x43a,0x442, +0x44a,0x452,0x45a,0x462,0x468,0x470,0x478,0x480,0x488,0x490,0x496,0x49e,0x4a6,0x4ae,0x4b6,0x4be, +0x4ca,0x4c6,0x4d2,0x4da,0x4e2,0x4f2,0x4f9,0x4ea,0x501,0x503,0x50b,0x513,0x51b,0x51c,0x524,0x52c, +0x534,0x51c,0x53c,0x541,0x534,0x51c,0x549,0x551,0x51b,0x559,0x561,0x513,0x569,0x3a7,0x571,0x575, +0x57d,0x57f,0x587,0x58f,0x51b,0x597,0x59f,0x513,0x41e,0x5a3,0x524,0x513,0x51b,0x3a7,0x5ab,0x3a7, +0x3a7,0x5b1,0x5b9,0x3a7,0x3a7,0x5bd,0x5c5,0x3a7,0x5c9,0x5d0,0x3a7,0x5d8,0x5e0,0x5e7,0x568,0x3a7, +0x3a7,0x5ef,0x5f7,0x5ff,0x607,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x60f,0x3a7,0x617,0x3a7,0x3a7,0x3a7, +0x61f,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x627,0x3a7,0x3a7,0x3a7,0x62f,0x528,0x528,0x528,0x3a7,0x635,0x63d,0x617, +0x653,0x645,0x645,0x65b,0x662,0x64b,0x3a7,0x3a7,0x3a7,0x66a,0x672,0x3a7,0x3a7,0x3a7,0x674,0x67c, +0x684,0x3a7,0x68b,0x693,0x3a7,0x69b,0x4fe,0x3a7,0x558,0x6a3,0x569,0x6ab,0x41e,0x6b3,0x3a7,0x6ba, +0x3a7,0x6bf,0x3a7,0x3a7,0x3a7,0x3a7,0x6c5,0x6cd,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3f5,0x3f5, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x6d4,0x6dc,0x6e0, +0x6f8,0x6fe,0x6e8,0x6f0,0x706,0x70e,0x712,0x5ea,0x71a,0x722,0x72a,0x3a7,0x732,0x67c,0x67c,0x67c, +0x742,0x74a,0x752,0x75a,0x75f,0x767,0x76f,0x73a,0x777,0x77f,0x3a7,0x785,0x78c,0x67c,0x67c,0x67c, +0x67c,0x595,0x792,0x67c,0x79a,0x3a7,0x3a7,0x679,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c, +0x67c,0x67c,0x67c,0x67c,0x67c,0x7a2,0x67c,0x67c,0x67c,0x67c,0x67c,0x7a8,0x67c,0x67c,0x7b0,0x7b8, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x67c,0x67c,0x67c,0x67c,0x7c8,0x7d0,0x7d8,0x7c0, +0x7e8,0x7f0,0x7f8,0x7ff,0x806,0x80e,0x812,0x7e0,0x67c,0x67c,0x67c,0x81a,0x820,0x67c,0x67c,0x826, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x82e,0x3a7,0x3a7,0x3a7,0x836,0x3a7,0x3a7,0x3a7,0x3f5, +0x83e,0x846,0x849,0x3a7,0x851,0x67c,0x67c,0x67f,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x858,0x85e, +0x86e,0x866,0x3a7,0x3a7,0x876,0x61f,0x3a7,0x3ce,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x67c,0x835, +0x3dc,0x3a7,0x87e,0x886,0x3a7,0x88e,0x896,0x3a7,0x3a7,0x3a7,0x3a7,0x89a,0x3a7,0x3a7,0x674,0x3cd, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x67c,0x67c, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x87e,0x67c,0x595,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x8a1,0x3a7,0x3a7,0x8a6,0x8ae,0x3a7,0x3a7,0x5cb,0x67c,0x673,0x3a7,0x3a7,0x8b6,0x3a7,0x3a7,0x3a7, +0x8be,0x8c5,0x645,0x8cd,0x3a7,0x3a7,0x5a1,0x8d5,0x3a7,0x8dd,0x8e4,0x3a7,0x501,0x8e9,0x3a7,0x51a, +0x3a7,0x8f1,0x8f9,0x51c,0x3a7,0x8fd,0x51b,0x905,0x3a7,0x3a7,0x3a7,0x90b,0x3a7,0x3a7,0x3a7,0x912, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x926,0x91a,0x91e,0x4a6,0x4a6,0x4a6,0x4a6,0x4a6, +0x4a6,0x4a6,0x4a6,0x4a6,0x4a6,0x4a6,0x4a6,0x4a6,0x4a6,0x92e,0x936,0x4a6,0x4a6,0x4a6,0x93b,0x93f, +0x947,0x94f,0x953,0x95b,0x4a6,0x4a6,0x4a6,0x95f,0x967,0x397,0x96f,0x977,0x3a7,0x3a7,0x3a7,0x97f, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0xe9c,0xe9c,0xedc,0xf1c,0xe9c,0xe9c,0xe9c,0xe9c,0xe9c,0xe9c,0xf54,0xf94,0xfd4,0xfe4,0x1024,0x1030, +0xe9c,0xe9c,0x1070,0xe9c,0xe9c,0xe9c,0x10a8,0x10e8,0x1128,0x1168,0x11a0,0x11e0,0x1220,0x1258,0x1298,0x12d8, +0xa40,0xa80,0xac0,0xaff,0x1a0,0x1a0,0xb3f,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xb68,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xba8,0x1a0,0xbce,0xc09,0xc49,0xc89,0xcc9,0xd09,0xd49, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0xdc9,0xdd9,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0, +0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0x1a0,0xd89, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x987,0x3a7,0x67c,0x67c,0x98f,0x61f,0x3a7,0x514, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x997,0x3a7,0x3a7,0x3a7,0x99e,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x9a6,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c, +0x9ae,0x9b2,0x43c,0x43c,0x43c,0x43c,0x9c2,0x9ba,0x43c,0x9ca,0x43c,0x43c,0x9d2,0x9d8,0x43c,0x43c, +0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x9e8,0x9e0,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c, +0x43c,0x43c,0x43c,0x9f0,0x43c,0x9f8,0x4a6,0xa00,0x43c,0xa08,0xa0f,0xa15,0xa1d,0xa21,0xa29,0x43c, +0x51b,0xa31,0xa38,0xa3f,0x41e,0xa47,0x569,0x3a7,0x501,0xa4e,0x3a7,0xa54,0x41e,0xa59,0xa61,0x3a7, +0x3a7,0xa66,0x51b,0x3a7,0x3a7,0x3a7,0x836,0xa6e,0x41e,0x5a3,0x57e,0xa75,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0xa31,0xa7d,0x3a7,0x3a7,0xa85,0xa8d,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0xa91,0xa99,0x3a7, +0x3a7,0xaa1,0x57e,0xaa9,0x3a7,0xaaf,0x3a7,0x3a7,0x60f,0xab7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0xabc,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0xac3,0xacb,0x3a7,0x3a7,0x3a7,0xace,0x57e,0xad6, +0xada,0xae2,0x3a7,0xae9,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0xaf0,0x3a7,0x3a7,0xafe,0xaf8,0x3a7,0x3a7,0x3a7,0xb06,0xb0e,0x3a7,0xb12,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x5a5,0x41e,0x99e,0xb1a,0x3a7,0x3a7,0x3a7,0xb27,0xb22,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0xb2f,0xb37,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0xb3d, +0x3a7,0xb43,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0xa55,0x3a7,0xb49,0x3a7,0x3a7,0xb51,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x535,0xb59,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3f5,0xb61,0x500,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0xb69,0xb71,0xb77,0x3a7,0xb7d,0x67c,0x67c,0xb85,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x67c,0x67c,0xb8d,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0xb93, +0x3a7,0xb9a,0x3a7,0xb96,0x3a7,0xb9d,0x3a7,0xba5,0xba9,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3f5,0xbb1,0x3f5,0xbb8,0xbbf,0xbc7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0xbcf,0xbd7,0x3a7,0x3a7,0xa55,0x3a7,0x3a7, +0x3a7,0x3a7,0xb43,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0xa81,0x3a7, +0xbdc,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0xbe4,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0xbec, +0x43c,0xbf4,0xbf4,0xbfb,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c, +0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x91e,0x4a6,0x4a6,0x43c, +0x43c,0x4a6,0x4a6,0xc03,0x43c,0x43c,0x43c,0x43c,0x43c,0x4a6,0x4a6,0x4a6,0x4a6,0x4a6,0x4a6,0x4a6, +0xc0b,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x43c,0x67c,0xc13,0x67c,0x67c,0x67f,0xc18,0xc1c, +0x858,0xc24,0x3c9,0x3a7,0xc2a,0x3a7,0xc2f,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x783,0x3a7,0x3a7,0x3a7, +0x3a7,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c, +0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0xc37, +0x98f,0x67c,0x67c,0x67c,0xc3e,0x67c,0x67c,0xc45,0xc4d,0xc13,0x67c,0xc55,0x67c,0xc5d,0xc62,0x3a7, +0x3a7,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67c,0x67f,0xc6a,0xc73,0xc77,0xc7f, +0xc6f,0x67c,0x67c,0x67c,0x67c,0xc87,0x67c,0x792,0xc8f,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0xc96,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7, +0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0x3a7,0xc96,0xca6,0xc9e,0xc9e,0xc9e,0xca7,0xca7,0xca7, +0xca7,0x3f5,0x3f5,0x3f5,0x3f5,0x3f5,0x3f5,0x3f5,0xcaf,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7, +0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7, +0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7, +0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7, +0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0xca7,0x386,0x386,0x386,0x12,0x12,0x12,0x12, +0x12,0x12,0x12,0x12,0x12,8,7,8,9,7,0x12,0x12,0x12,0x12,0x12,0x12, +0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,7,7,7,8,9,0xa,0xa,4, +4,4,0xa,0xa,0x310a,0xf20a,0xa,3,6,3,6,6,2,2,2,2, +2,2,2,2,2,2,6,0xa,0x500a,0xa,0xd00a,0xa,0xa,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0x510a,0xa,0xd20a,0xa,0xa,0xa,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0x510a,0xa,0xd20a,0xa,0x12,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0x12,0x12,0x12,0x12, +0x12,7,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, +0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,6,0xa,4,4, +4,4,0xa,0xa,0xa,0xa,0,0x900a,0xa,0xb2,0xa,0xa,4,4,2,2, +0xa,0,0xa,0xa,0xa,2,0,0x900a,0xa,0xa,0xa,0xa,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0xa,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0xa,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0xa,0xa,0, +0,0,0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0,0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0,0,0,0,0xa,0xa,0,0,0,0,0,0, +0,0,0xa,0,0,0,0,0,0xa,0xa,0,0xa,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0xa,0,0,0,0,0,0,0,0,0, +0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0xa,0,0,0xa,0xa,4,1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,1,0xb1,1,0xb1,0xb1,1,0xb1,0xb1,1,0xb1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,5,5,5,5,5,5,0xa,0xa, +0xd,4,4,0xd,6,0xd,0xa,0xa,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xd,0x8ad,0xd,0xd,0xd,0x4d,0xd,0x8d,0x8d,0x8d,0x8d,0x4d,0x8d, +0x4d,0x8d,0x4d,0x4d,0x4d,0x4d,0x4d,0x8d,0x8d,0x8d,0x8d,0x4d,0x4d,0x4d,0x4d,0x4d, +0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x2d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d, +0x8d,0x4d,0x4d,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,5,5,5,5,5,5,5,5, +5,5,4,5,5,0xd,0x4d,0x4d,0xb1,0x8d,0x8d,0x8d,0xd,0x8d,0x8d,0x8d, +0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d, +0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d, +0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d, +0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d, +0x8d,0x4d,0x4d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x4d,0x8d,0x4d,0x8d, +0x4d,0x4d,0x8d,0x8d,0xd,0x8d,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,5,0xa,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xd,0xd,0xb1,0xb1,0xa,0xb1,0xb1,0xb1,0xb1,0x8d,0x8d, +2,2,2,2,2,2,2,2,2,2,0x4d,0x4d,0x4d,0xd,0xd,0x4d, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xad, +0x8d,0xb1,0x4d,0x4d,0x4d,0x8d,0x8d,0x8d,0x8d,0x8d,0x4d,0x4d,0x4d,0x4d,0x8d,0x4d, +0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x8d,0x4d,0x8d,0x4d,0x8d,0x4d,0x4d,0x8d, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xd,0xd,0x8d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d, +0x4d,0x8d,0x8d,0x8d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d, +0x4d,0x4d,0x4d,0x8d,0x8d,0x4d,0x4d,0x4d,0x4d,0x8d,0x4d,0x8d,0x8d,0x4d,0x4d,0x4d, +0x8d,0x8d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xd,0xd,0xd,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,1,1,1,1,1,1,1,1, +1,1,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41, +0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41, +0x41,0x41,0x41,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,1,1,0xa,0xa, +0xa,0xa,0x21,1,1,0xb1,1,1,0xb1,0xb1,0xb1,0xb1,1,0xb1,0xb1,0xb1, +1,0xb1,0xb1,0xb1,0xb1,0xb1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,0xb1,0xb1, +0xb1,0xb1,1,0xb1,0xb1,0xb1,0xb1,0xb1,0x81,0x41,0x41,0x41,0x41,0x41,0x81,0x81, +0x41,0x81,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x81,0x41,0x81,0x81, +0x81,0xb1,0xb1,0xb1,1,1,1,1,0x4d,0xd,0x4d,0x4d,0x4d,0x4d,0xd,0x8d, +0x4d,0x8d,0x8d,0xd,0xd,0xd,0xd,0xd,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d, +0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x8d,0x2d,0x2d,0x2d,0x4d,0xd, +0xd,0x4d,0x4d,0x4d,0x4d,0x4d,0x8d,0xd,5,5,0xd,0xd,0xd,0xd,0xd,0xd, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,5,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d, +0x4d,0x4d,0x8d,0x8d,0x8d,0xd,0x8d,0x4d,0x4d,0x8d,0x8d,0x4d,0x4d,0x4d,0x4d,0x4d, +0x4d,0x8d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0xd,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0xb1,0,0xb1,0,0,0,0,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0xb1,0,0,0,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0,0,0, +0,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0xb1,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,4,4,0,0,0,0,0,0,0,4,0,0,0xb1,0, +0,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0xb1,0xb1,0,0,0,0,0xb1,0xb1,0,0,0xb1,0xb1,0xb1,0,0, +0,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xb1,0xb1,0,0,0,0xb1,0,0,0,0,0,0, +0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0,0xb1,0xb1,0,0,0, +0,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0xb1,0xb1,0,0,0,0,0,0,0,0, +0,0,0,0,0,4,0,0,0,0,0,0,0,0,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0xb1,0,0,0xb1,0,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0, +0,0xb1,0,0,0,0,0,0,0,0xb1,0xb1,0,0,0,0,0, +0,0,0,0,0,0,0xb1,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xb1,0,0,0,0,0,0,0,0,0,0,0, +0,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,4,0xa,0, +0,0,0,0,0xb1,0,0,0,0xb1,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0xb1,0,0xb1,0xb1,0xb1,0,0,0, +0,0,0xb1,0xb1,0xb1,0,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0, +0,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0xb1,0,0,0xa0,0,0,0,0, +0,0,0xa0,0,0,0,0,0,0xb1,0xb1,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0xb1,0,0,0,0,0,0,0, +0,0,0,0,0,0,0xb1,0,0,0,0,0,0,0,0xb1,0xb1, +0xb1,0,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0xb1,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0, +0,0,0,4,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0xb1,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0xb1,0,0xb1,0,0xb1,0x310a,0xf20a,0x310a,0xf20a,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0, +0xb1,0xb1,0xb1,0xb1,0xb1,0,0xb1,0xb1,0,0,0,0,0,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0,0xb1,0xb1,0,0,0xb1,0xb1,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xb1,0xb1,0,0,0,0,0xb1,0xb1,0xb1,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1, +0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0, +0,0xb1,0xb1,0,0,0,0,0,0,0xb1,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0xb1,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0,0,0,0xa,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,9,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0x310a,0xf20a,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1, +0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0xb1,0xb1,0,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0,0,0,0,0,0,0,0,0xb1,0,0,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,4, +0,0xb1,0,0,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40, +0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40, +0x40,0x40,0x40,0x40,0x40,0xb1,0x40,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0x4a,0xa,0xa,0x2a,0xb1,0xb1,0xb1,0x12,0xb1,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0x40,0x40,0x40,0x40, +0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40, +0x40,0x40,0x40,0x40,0x40,0,0,0,0,0,0,0,0,0xb1,0xb1,0x40, +0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40, +0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0xb1,0xb1,0xb1,0,0,0,0,0xb1, +0xb1,0,0,0,0,0,0,0,0,0,0xb1,0,0,0,0,0, +0,0xb1,0xb1,0xb1,0,0,0,0,0xa,0,0,0,0xa,0xa,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0xb1,0,0,0xb1,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0xb1,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0xb1,0,0xb1,0, +0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0xb1,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0xb1,0,0xb1,0xb1,0xb1,0xb1,0xb1,0,0xb1,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1, +0xb1,0xb1,0,0,0xb1,0xb1,0,0xb1,0xb1,0xb1,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0, +0xb1,0xb1,0,0,0,0xb1,0,0xb1,0xb1,0xb1,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0,0,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xb1,0xb1,0xb1,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0, +0,0xb1,0,0,0,0,0,0,0xb1,0,0,0,0xb1,0xb1,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0xa,0,0xa, +0xa,0xa,0,0,0,0,0,0,0,0,0,0,0,0xa,0xa,0xa, +0,0,0,0,0,0,0,0,0,0,0,0,0,0xa,0xa,0xa, +0,0,0,0,0,0,0,0,0,0,0,0,0,0xa,0xa,0, +0xa,0xa,0xa,0xa,6,0x310a,0xf20a,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,9, +0xb2,0xb2,0xb2,0xb2,0xb2,0x12,0x814,0x815,0x813,0x816,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2, +2,0,0,0,2,2,2,2,2,2,3,3,0xa,0x310a,0xf20a,0, +9,9,9,9,9,9,9,9,9,9,9,0xb2,0x412,0x432,0x8a0,0x8a1, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +9,7,0x8ab,0x8ae,0x8b0,0x8ac,0x8af,6,4,4,4,4,4,0xa,0xa,0xa, +0xa,0x300a,0xf00a,0xa,0xa,0xa,0xa,0xa,2,2,2,2,2,2,2,2, +2,2,3,3,0xa,0x310a,0xf20a,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,4,4,4,4,4,4,4,4, +4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4, +4,4,4,4,4,4,4,4,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xa,0xa,0,0xa,0xa,0xa,0xa,0, +0xa,0xa,0,0,0,0,0,0,0,0,0,0,0xa,0,0xa,0xa, +0xa,0,0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0,0xa,0,0xa, +0,0xa,0,0,0,0,4,0,0,0,0,0,0,0,0,0, +0,0,0xa,0xa,0,0,0,0,0x100a,0xa,0xa,0xa,0xa,0,0,0, +0,0,0xa,0xa,0xa,0xa,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0,0,0,0,0, +0,0xa,0xa,0xa,0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a, +0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0xa,0xa,0x300a,0xf00a,0x900a,0x900a,0x900a,0x100a,0x900a,0x900a, +0x100a,0x100a,0x900a,0x900a,0x900a,0x900a,0x900a,0x100a,0xa,0x100a,0x100a,0x100a,0x100a,0xa,0xa,0xa, +0x700a,0x700a,0x700a,0xb00a,0xb00a,0xb00a,0xa,0xa,0xa,0x100a,3,4,0xa,0x900a,0x100a,0xa, +0xa,0xa,0x100a,0x100a,0x100a,0x100a,0xa,0x900a,0x900a,0x900a,0x900a,0xa,0x900a,0xa,0x100a,0xa, +0xa,0xa,0xa,0x100a,0x100a,0x100a,0x100a,0x100a,0x100a,0x100a,0x100a,0x100a,0xa,0xa,0xa,0xa, +0xa,0x100a,0xa,0x100a,0x300a,0xf00a,0x100a,0x100a,0x100a,0x100a,0x100a,0x900a,0x100a,0x900a,0x100a,0x100a, +0x100a,0x100a,0x100a,0x100a,0x900a,0xa,0xa,0xa,0xa,0xa,0x300a,0xf00a,0x300a,0xf00a,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0x100a,0x100a,0xa,0x100a,0xa,0x300a,0xf00a,0x300a,0xf00a, +0x300a,0xf00a,0x300a,0xf00a,0xa,0xa,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a, +0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x100a,0xa,0xa,0x300a,0xf00a,0x300a,0xf00a,0xa, +0xa,0xa,0xa,0xa,0x900a,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0x300a,0xf00a, +0xa,0xa,0x900a,0x100a,0x900a,0x900a,0x100a,0x900a,0x100a,0x100a,0x100a,0x100a,0x300a,0xf00a,0x300a,0xf00a, +0x300a,0xf00a,0x300a,0xf00a,0x900a,0xa,0xa,0xa,0xa,0xa,0x100a,0x100a,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0x300a,0xf00a,0x300a,0xf00a,0x900a,0xa,0xa,0x300a,0xf00a,0xa,0xa, +0xa,0xa,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0x310a,0xf20a,0x310a,0xf20a,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0x100a,0x100a,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0x310a,0xf20a,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a, +0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0x100a,0xa,0xa,0x300a,0xf00a,0x310a,0xf20a,0xa,0x300a,0xf00a,0xa,0x500a,0x100a,0xd00a,0xa,0xa, +0xa,0xa,0xa,0x100a,0x100a,0x300a,0xf00a,0xa,0xa,0xa,0xa,0xa,0x900a,0x300a,0xf00a,0xa, +0xa,0xa,0x300a,0xf00a,0x300a,0xf00a,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0x100a,0xa,0x100a,0x100a,0x100a,0xa,0xa,0x300a,0xf00a,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0x100a,0x900a,0x100a,0x100a,0x300a,0xf00a,0xa,0xa,0x310a,0xf20a,0xa,0xa, +0xa,0xa,0xa,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,0x710a,0x320a,0xf10a, +0xb20a,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,0xa,0xa,0x900a,0x100a,0x100a,0x100a,0x100a, +0x900a,0xa,0x100a,0x900a,0x300a,0xf00a,0x100a,0x100a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0x900a,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0x300a,0xf00a,0x100a,0x100a,0x300a,0xf00a,0xa,0xa,0xa,0x100a,0xa,0xa,0xa,0xa,0x100a,0x300a, +0xf00a,0x300a,0xf00a,0xa,0x300a,0xf00a,0xa,0xa,0x310a,0xf20a,0x310a,0xf20a,0x100a,0xa,0xa,0xa, +0xa,0xa,0x100a,0x900a,0x900a,0x900a,0x100a,0xa,0xa,0xa,0xa,0xa,0x300a,0xf00a,0x900a,0xa, +0xa,0xa,0xa,0x100a,0xa,0xa,0xa,0x300a,0xf00a,0x300a,0xf00a,0x100a,0xa,0x100a,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0x100a,0x100a,0x100a,0x100a,0x100a,0x100a, +0x100a,0x100a,0x100a,0x100a,0x100a,0x100a,0x100a,0x100a,0x100a,0x100a,0x100a,0x100a,0x100a,0xa,0x100a,0x100a, +0x100a,0x100a,0xa,0xa,0x100a,0xa,0x100a,0xa,0xa,0x100a,0xa,0x300a,0xf00a,0x300a,0xf00a,0xa, +0xa,0xa,0xa,0xa,0x300a,0xf00a,0xa,0xa,0xa,0xa,0xa,0xa,0x300a,0xf00a,0x100a,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0x100a,0x100a,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0x300a,0xf00a,0xa,0xa,0xa,0xa,0x100a,0x100a,0x100a,0x100a,0xa,0x100a,0x100a,0xa,0xa,0x100a, +0x100a,0xa,0xa,0xa,0xa,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a, +0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a, +0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x100a,0xa,0xa,0x300a,0xf00a, +0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0xa,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a, +0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0x300a,0xf00a,0xa, +0xa,0xa,0xa,0xa,0x100a,0xa,0x900a,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0x900a,0xa,0,0,0,0,0,0xa,0xa,0xa, +0xa,0xa,0xa,0,0,0,0,0xb1,0xb1,0xb1,0,0,0,0,0,0, +0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0xa,0xa,0x300a,0xf00a,0x300a,0xf00a,0xa,0xa, +0xa,0x300a,0xf00a,0xa,0x300a,0xf00a,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0x300a,0xf00a,0xa,0xa,0x300a,0xf00a,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a, +0x310a,0xf20a,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,0x310a, +0xf20a,0xa,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0,0,0,0,0xa,0,0,0,0,0,0,0, +0,0,0xb1,0xb1,0xb1,0xb1,0,0,0xa,0,0,0,0,0,0xa,0xa, +0,0,0,0,0,0xa,0xa,0xa,9,0xa,0xa,0xa,0xa,0,0,0, +0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,0xa,0xa,0x310a,0xf20a,0x310a,0xf20a, +0x310a,0xf20a,0x310a,0xf20a,0xa,0xa,0xa,0xa,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0xb1,0xb1,0xa,0xa,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xa,0xa,0xa,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0,0,0,0,0, +0,0,0,0,0xa,0xa,0xa,0xa,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xa, +0xa,0xa,0xa,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0xa,0xa,0xa,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xa,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xa,0xa,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0xb1,0xb1,0,0,0,0,0,0,0,0, +0xa,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0xb1,0,0,0,0xb1,0, +0,0,0,0xb1,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0xb1,0xb1,0,0xa,0xa,0xa,0xa, +0xb1,0,0,0,0,0,0,0,0,0,0,0,4,4,0,0, +0,0,0,0,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40,0x40, +0x40,0x40,0x40,0x40,0x40,0x40,0x60,0,0xa,0xa,0xa,0xa,0,0,0,0, +0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0, +0,0,0,0xb1,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0,0,0xb1,0xb1,0xb1,0xb1,0,0, +0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0,0,0xb1,0xb1,0,0,0xb1,0xb1,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0,0,0,0,0,0,0,0, +0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xb1,0,0xb1,0xb1,0xb1,0,0,0xb1,0xb1,0,0,0, +0,0,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0, +0xb1,0xb1,0,0,0,0,0,0,0,0,0xb1,0,0,0,0,0, +0,0,0,0,0,0,0xa,0xa,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0,0, +0xb1,0,0,0,0,0xb1,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1, +1,3,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,1,0xb1,1,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xa,0x12,0x12,0x12,0x12, +0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xa,0xa,0xa,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,6,0xa,6,0, +0xa,6,0xa,0xa,0xa,0x310a,0xf20a,0x310a,0xf20a,0x310a,0xf20a,4,0xa,0xa,3,3, +0x300a,0xf00a,0xa,0,0xa,4,4,0xa,0,0,0,0,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xb2,0,0xa,0xa,4, +4,4,0xa,0xa,0x310a,0xf20a,0xa,3,6,3,6,6,2,2,2,2, +2,2,2,2,2,2,6,0xa,0x500a,0xa,0xd00a,0xa,0xa,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0x510a,0xa,0xd20a,0xa,0x310a,0xf20a,0xa,0x310a,0xf20a, +0xa,0xa,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,4,4,0xa,0xa, +0xa,4,4,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0x12,0x12,0x12,0x12, +0x12,0x12,0x12,0x12,0x12,0xaa,0xaa,0xaa,0xa,0xa,0x12,0x12,0,0xa,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0xb1,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1, +0xb1,0xb1,0xb1,0,0,0,0,0,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,0xa,1,0xb1,0xb1,0xb1,1,0xb1,0xb1,1, +1,1,1,1,0xb1,0xb1,0xb1,0xb1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0xb1,0xb1,0xb1,1,1,1,1,0xb1,0x41,0x81,1,1,0x81,0xb1,0xb1,1, +1,1,1,0x41,0x41,0x41,0x41,0x81,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,0x41,0x41,0x41,0x41,0x41,0x81,1,0x81, +1,0x81,0x81,1,1,0x61,0x81,0x81,0x81,0x81,0x81,0x41,0x41,0x41,0x41,0x61, +0x41,0x41,0x41,0x41,0x41,0x81,0x41,0x41,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0x41,0x81,0x41,0x81,0x81,0x81,0x41,0x41, +0x41,0x81,0x41,0x41,0x81,0x41,0x81,0x81,0x41,0x81,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,0x81,0x81,0x81,0x81,0x41,0x41,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0x4d,0x4d,0x8d,0x4d,0xb1,0xb1,0xb1,0xb1,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd, +5,5,5,5,5,5,5,5,5,5,0xd,0xd,0xd,0xd,0xd,0xd, +0x6d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d, +0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,1, +1,1,1,1,1,1,1,1,1,1,1,0xb1,0xb1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xb1,0xb1,0xb1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0x4d,0x4d,0x4d,0x8d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d,0x4d, +0x4d,0xd,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0x4d,0x4d,0x4d, +0x8d,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0x41,0x41,0x41,0x41,0x81,0x81,0x41,0x41,0x41,0x41,0x41,0x41, +0x41,0x41,0x41,0x41,0x41,0x41,0xb1,0xb1,0xb1,0xb1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,0x41,1,0x41,0x41,0x81,0x81,0x81,1,0x41,0x81,0x81,0x41, +0x41,0x81,0x41,0x41,1,0x41,0x81,0x81,0x41,1,1,1,1,0x81,0x41,0x61, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0, +0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0,0,0,0,0,0,0,0,0,0,0xb1,0,0,0xb1, +0xb1,0,0,0,0,0,0,0,0,0,0,0xb1,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xb1, +0xb1,0xb1,0xb1,0,0,0xb1,0xb1,0,0,0,0,0,0,0,0,0xb1, +0xb1,0xb1,0xb1,0xb1,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0xb1,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1, +0xb1,0,0,0xb1,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0,0,0xb1,0,0xb1,0xb1, +0,0,0,0,0,0,0xb1,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0, +0,0,0,0,0,0,0xb1,0xb1,0xb1,0,0xb1,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0xb1,0, +0,0,0,0xb1,0xb1,0,0xb1,0xb1,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0, +0xb1,0xb1,0,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0, +0,0xb1,0,0xb1,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0,0xb1,0,0,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0,0xb1,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1, +0xb1,0xb1,0,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0xb1,0xb1,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0xb1,0,0xb1,0,0,0,0,0xb1, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1, +0,0,0xb1,0xb1,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xa0, +0xa0,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0,0,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0xb1, +0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0, +0,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0xb1,0xb1,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0xa0, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0,0xb1,0xb1,0,0xb1,0xb1,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0, +0,0,0xb1,0,0xb1,0xb1,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0xb1, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0xb1,0xb1,0,0,0,0xb1,0,0xb1, +0,0,0,0,0,0,0,0,0xb1,0,0xb1,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,4,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,4,4,4,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0xa0,0xa0,0xa0,0xa0, +0xa0,0xa0,0xa0,0xa0,0xa0,0xa0,0xa0,0xa0,0xa0,0xa0,0xa0,0xa0,0xb1,0,0,0, +0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0xa,0,0xb1,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xb2,0xb2,0xb2,0xb2,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0,0, +0,0,0,0,0,0,0,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0xa,0xa,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xa,0xa,0xb1,0xb1,0xb1,0xa,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0x100a,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0x100a,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0x100a,0,0,0,0,0,0,0,0, +0,0,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0, +0,0,0,0,0,0xb1,0,0,0,0,0,0,0,0,0,0, +0xb1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0,0xb1, +0xb1,0,0xb1,0xb1,0xb1,0xb1,0xb1,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,4, +0,0,0,0,0,0,0,0,0,0,0,0,0xb1,0xb1,0xb1,0xb1, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,1,1,1,1,1,1,1,1,1, +0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41, +0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41,0x41, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xa1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,0xd,0xd,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xa,0xa,0xd,0xd, +0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xd,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0, +0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +2,2,2,2,2,2,2,2,2,2,2,0xa,0xa,0xa,0xa,0xa, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0xa,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0, +0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0, +0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0,0,0,0,0xa,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0, +0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0, +0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0, +0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0,0,0xa,0xa,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0,0,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0,0,0,0,0,0,0,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0,0,0,0,0,0,0,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0,0,0,0,0,0,0,0,0xa,0xa,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0,0xa,0xa,0xa,0xa, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0, +0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0xa,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,2,2,2,2, +2,2,2,2,2,2,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0x12,0x12,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2, +0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2, +0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0xb2,0x12,0xb2,0x12,0x12,0x12,0x12,0x12,0x12, +0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, +0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0x12,0x12,0x12,0x12, +0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0,0,0,0 +}; + +static const uint32_t ubidi_props_mirrors[40]={ +0x2000ab,0xbb,0x4202215,0x4e0221f,0x3e02220,0x3a02221,0x3c02222,0x4c02224,0x2202243,0x1402245,0x120224c,0x4002298,0x44022a6,0x48022a8,0x46022a9,0x4a022ab, +0x38022b8,0x10022cd,0x2e022f2,0x30022f3,0x32022f4,0x34022f6,0x36022f7,0x24022fa,0x26022fb,0x28022fc,0x2a022fd,0x2c022fe,0x20027dc,0xa0299b,0xc029a0,0x8029a3, +0x16029b8,0x4029f5,0x1802ade,0x1c02ae3,0x1a02ae4,0x1e02ae5,0xe02aee,0x602bfe +}; + +static const uint8_t ubidi_props_jgArray[684]={ +0x2d,0,3,3,0x2c,3,0x2d,3,4,0x2a,4,4,0xd,0xd,0xd,6, +6,0x1f,0x1f,0x23,0x23,0x21,0x21,0x28,0x28,1,1,0xb,0xb,0x37,0x37,0x37, +0,9,0x1d,0x13,0x16,0x18,0x1a,0x10,0x2c,0x2d,0x2d,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,4,0x1d, +0,3,3,3,0,3,0x2c,0x2c,0x2d,4,4,4,4,4,4,4, +4,0xd,0xd,0xd,0xd,0xd,0xd,0xd,6,6,6,6,6,6,6,6, +6,0x1f,0x1f,0x1f,0x1f,0x1f,0x1f,0x1f,0x1f,0x1f,0x23,0x23,0x23,0x21,0x21,0x28, +1,9,9,9,9,9,9,0x1d,0x1d,0xb,0x26,0xb,0x13,0x13,0x13,0xb, +0xb,0xb,0xb,0xb,0xb,0x16,0x16,0x16,0x16,0x1a,0x1a,0x1a,0x1a,0x38,0x15,0xd, +0x2a,0x11,0x11,0xe,0x2c,0x2c,0x2c,0x2c,0x2c,0x2c,0x2c,0x2c,0x37,0x2f,0x37,0x2c, +0x2d,0x2d,0x2e,0x2e,0,0x2a,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,6,0x1f, +0,0,0,0,0,0,0,0,0,0,0x23,0x21,1,0,0,0x15, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +2,0,5,0xc,0xc,7,7,0xf,0x27,0x32,0x12,0x2b,0x2b,0x30,0x31,0x14, +0x17,0x19,0x1b,0x24,0xa,8,0x1c,0x20,0x22,0x1e,7,0x25,0x29,5,0xc,7, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0x35,0x34,0x33, +4,4,4,4,4,4,4,0xd,0xd,6,6,0x1f,0x23,1,1,1, +9,9,0xb,0xb,0xb,0x18,0x18,0x1a,0x1a,0x1a,0x16,0x1f,0x1f,0x23,0xd,0xd, +0x23,0x1f,0xd,3,3,0x37,0x37,0x2d,0x2c,0x2c,0x36,0x36,0xd,0x23,0x23,0x13, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0x5d,0x5a,0x60,0x63,0x5e,0x5f,0x59,0x61,0x5b,0x5c,0x62,0,0,0,0,0, +3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, +3,3,3,0,0,0,0x66,0,0,0x1a,0xd,0x28,0x28,0xb,0x67,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +4,4,0xd,0x28,9,0x1d,0x16,0x18,0x2d,0x2d,0x1f,0x2c,0x39,0,6,0x21, +0xb,0x55,0x1f,1,0x13,0x1d,4,4,4,0x1f,0x2d,0x56,0x58,0x57,4,4, +4,0xd,0xb,1,0x58,0xd,0xd,0x16,0xb,0,0,0 +}; + +static const uint8_t ubidi_props_jgArray2[612]={ +0x3a,0x3c,0x3c,0x40,0x40,0x3d,0,0x52,0,0x54,0x54,0,0,0x41,0x4f,0x53, +0x43,0x43,0x43,0x44,0x3e,0x50,0x45,0x46,0x4c,0x3b,0x3b,0x48,0x48,0x4b,0x49,0x49, +0x49,0x4a,0,0,0x4d,0,0,0,0,0,0,0x47,0x3f,0x4e,0x51,0x42, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0x65,0,0,0,0,0,0,0x65,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0x64,0,0,0x65,0,0x64,0, +0x64,0,0,0x64 +}; + +static const UBiDiProps ubidi_props_singleton={ + nullptr, + ubidi_props_indexes, + ubidi_props_mirrors, + ubidi_props_jgArray, + ubidi_props_jgArray2, + { + ubidi_props_trieIndex, + ubidi_props_trieIndex+3612, + nullptr, + 3612, + 9412, + 0x1a0, + 0xe9c, + 0x0, + 0x0, + 0x110000, + 0x32dc, + nullptr, 0, false, false, 0, nullptr + }, + { 2,2,0,0 } +}; + +#endif // INCLUDED_FROM_UBIDI_PROPS_C diff --git a/intl/icu/source/common/ubidiimp.h b/intl/icu/source/common/ubidiimp.h new file mode 100644 index 0000000000..e48fc6f941 --- /dev/null +++ b/intl/icu/source/common/ubidiimp.h @@ -0,0 +1,484 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: ubidiimp.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 1999aug06 +* created by: Markus W. Scherer, updated by Matitiahu Allouche +*/ + +#ifndef UBIDIIMP_H +#define UBIDIIMP_H + +#include "unicode/utypes.h" +#include "unicode/ubidi.h" +#include "unicode/uchar.h" +#include "ubidi_props.h" + +/* miscellaneous definitions ---------------------------------------------- */ + +// ICU-20853=ICU-20935 Solaris #defines CS and ES in sys/regset.h +#ifdef CS +# undef CS +#endif +#ifdef ES +# undef ES +#endif + +typedef uint8_t DirProp; +typedef uint32_t Flags; + +/* Comparing the description of the BiDi algorithm with this implementation + is easier with the same names for the BiDi types in the code as there. + See UCharDirection in uchar.h . +*/ +enum { + L= U_LEFT_TO_RIGHT, /* 0 */ + R= U_RIGHT_TO_LEFT, /* 1 */ + EN= U_EUROPEAN_NUMBER, /* 2 */ + ES= U_EUROPEAN_NUMBER_SEPARATOR, /* 3 */ + ET= U_EUROPEAN_NUMBER_TERMINATOR, /* 4 */ + AN= U_ARABIC_NUMBER, /* 5 */ + CS= U_COMMON_NUMBER_SEPARATOR, /* 6 */ + B= U_BLOCK_SEPARATOR, /* 7 */ + S= U_SEGMENT_SEPARATOR, /* 8 */ + WS= U_WHITE_SPACE_NEUTRAL, /* 9 */ + ON= U_OTHER_NEUTRAL, /* 10 */ + LRE=U_LEFT_TO_RIGHT_EMBEDDING, /* 11 */ + LRO=U_LEFT_TO_RIGHT_OVERRIDE, /* 12 */ + AL= U_RIGHT_TO_LEFT_ARABIC, /* 13 */ + RLE=U_RIGHT_TO_LEFT_EMBEDDING, /* 14 */ + RLO=U_RIGHT_TO_LEFT_OVERRIDE, /* 15 */ + PDF=U_POP_DIRECTIONAL_FORMAT, /* 16 */ + NSM=U_DIR_NON_SPACING_MARK, /* 17 */ + BN= U_BOUNDARY_NEUTRAL, /* 18 */ + FSI=U_FIRST_STRONG_ISOLATE, /* 19 */ + LRI=U_LEFT_TO_RIGHT_ISOLATE, /* 20 */ + RLI=U_RIGHT_TO_LEFT_ISOLATE, /* 21 */ + PDI=U_POP_DIRECTIONAL_ISOLATE, /* 22 */ + ENL, /* EN after W7 */ /* 23 */ + ENR, /* EN not subject to W7 */ /* 24 */ + dirPropCount +}; + +/* Sometimes, bit values are more appropriate + to deal with directionality properties. + Abbreviations in these macro names refer to names + used in the BiDi algorithm. +*/ +#define DIRPROP_FLAG(dir) (1UL<<(dir)) +#define PURE_DIRPROP(prop) ((prop)&~0xE0) ????????????????????????? + +/* special flag for multiple runs from explicit embedding codes */ +#define DIRPROP_FLAG_MULTI_RUNS (1UL<<31) + +/* are there any characters that are LTR or RTL? */ +#define MASK_LTR (DIRPROP_FLAG(L)|DIRPROP_FLAG(EN)|DIRPROP_FLAG(ENL)|DIRPROP_FLAG(ENR)|DIRPROP_FLAG(AN)|DIRPROP_FLAG(LRE)|DIRPROP_FLAG(LRO)|DIRPROP_FLAG(LRI)) +#define MASK_RTL (DIRPROP_FLAG(R)|DIRPROP_FLAG(AL)|DIRPROP_FLAG(RLE)|DIRPROP_FLAG(RLO)|DIRPROP_FLAG(RLI)) +#define MASK_R_AL (DIRPROP_FLAG(R)|DIRPROP_FLAG(AL)) +#define MASK_STRONG_EN_AN (DIRPROP_FLAG(L)|DIRPROP_FLAG(R)|DIRPROP_FLAG(AL)|DIRPROP_FLAG(EN)|DIRPROP_FLAG(AN)) + +/* explicit embedding codes */ +#define MASK_EXPLICIT (DIRPROP_FLAG(LRE)|DIRPROP_FLAG(LRO)|DIRPROP_FLAG(RLE)|DIRPROP_FLAG(RLO)|DIRPROP_FLAG(PDF)) + +/* explicit isolate codes */ +#define MASK_ISO (DIRPROP_FLAG(LRI)|DIRPROP_FLAG(RLI)|DIRPROP_FLAG(FSI)|DIRPROP_FLAG(PDI)) + +#define MASK_BN_EXPLICIT (DIRPROP_FLAG(BN)|MASK_EXPLICIT) + +/* paragraph and segment separators */ +#define MASK_B_S (DIRPROP_FLAG(B)|DIRPROP_FLAG(S)) + +/* all types that are counted as White Space or Neutral in some steps */ +#define MASK_WS (MASK_B_S|DIRPROP_FLAG(WS)|MASK_BN_EXPLICIT|MASK_ISO) + +/* types that are neutrals or could becomes neutrals in (Wn) */ +#define MASK_POSSIBLE_N (DIRPROP_FLAG(ON)|DIRPROP_FLAG(CS)|DIRPROP_FLAG(ES)|DIRPROP_FLAG(ET)|MASK_WS) + +/* + * These types may be changed to "e", + * the embedding type (L or R) of the run, + * in the BiDi algorithm (N2) + */ +#define MASK_EMBEDDING (DIRPROP_FLAG(NSM)|MASK_POSSIBLE_N) + +/* the dirProp's L and R are defined to 0 and 1 values in UCharDirection */ +#define GET_LR_FROM_LEVEL(level) ((DirProp)((level)&1)) + +#define IS_DEFAULT_LEVEL(level) ((level)>=0xfe) + +/* + * The following bit is used for the directional isolate status. + * Stack entries corresponding to isolate sequences are greater than ISOLATE. + */ +#define ISOLATE 0x0100 + +U_CFUNC UBiDiLevel +ubidi_getParaLevelAtIndex(const UBiDi *pBiDi, int32_t index); + +#define GET_PARALEVEL(ubidi, index) \ + ((UBiDiLevel)(!(ubidi)->defaultParaLevel || (index)<(ubidi)->paras[0].limit ? \ + (ubidi)->paraLevel : ubidi_getParaLevelAtIndex((ubidi), (index)))) + +/* number of paras entries allocated initially without malloc */ +#define SIMPLE_PARAS_COUNT 10 +/* number of isolate entries allocated initially without malloc */ +#define SIMPLE_ISOLATES_COUNT 5 +/* number of isolate run entries for paired brackets allocated initially without malloc */ +#define SIMPLE_OPENINGS_COUNT 20 + +#define CR 0x000D +#define LF 0x000A + +/* Run structure for reordering --------------------------------------------- */ +enum { + LRM_BEFORE=1, + LRM_AFTER=2, + RLM_BEFORE=4, + RLM_AFTER=8 +}; + +typedef struct Para { + int32_t limit; + int32_t level; +} Para; + +enum { /* flags for Opening.flags */ + FOUND_L=DIRPROP_FLAG(L), + FOUND_R=DIRPROP_FLAG(R) +}; + +typedef struct Opening { + int32_t position; /* position of opening bracket */ + int32_t match; /* matching char or -position of closing bracket */ + int32_t contextPos; /* position of last strong char found before opening */ + uint16_t flags; /* bits for L or R/AL found within the pair */ + UBiDiDirection contextDir; /* L or R according to last strong char before opening */ + uint8_t filler; /* to complete a nice multiple of 4 chars */ +} Opening; + +typedef struct IsoRun { + int32_t contextPos; /* position of char determining context */ + uint16_t start; /* index of first opening entry for this run */ + uint16_t limit; /* index after last opening entry for this run */ + UBiDiLevel level; /* level of this run */ + DirProp lastStrong; /* bidi class of last strong char found in this run */ + DirProp lastBase; /* bidi class of last base char found in this run */ + UBiDiDirection contextDir; /* L or R to use as context for following openings */ +} IsoRun; + +typedef struct BracketData { + UBiDi *pBiDi; + /* array of opening entries which should be enough in most cases; no malloc() */ + Opening simpleOpenings[SIMPLE_OPENINGS_COUNT]; + Opening *openings; /* pointer to current array of entries */ + int32_t openingsCount; /* number of allocated entries */ + int32_t isoRunLast; /* index of last used entry */ + /* array of nested isolated sequence entries; can never excess UBIDI_MAX_EXPLICIT_LEVEL + + 1 for index 0, + 1 for before the first isolated sequence */ + IsoRun isoRuns[UBIDI_MAX_EXPLICIT_LEVEL+2]; + UBool isNumbersSpecial; /* reordering mode for NUMBERS_SPECIAL */ +} BracketData; + +typedef struct Isolate { + int32_t startON; + int32_t start1; + int32_t state; + int16_t stateImp; +} Isolate; + +typedef struct Run { + int32_t logicalStart, /* first character of the run; b31 indicates even/odd level */ + visualLimit, /* last visual position of the run +1 */ + insertRemove; /* if >0, flags for inserting LRM/RLM before/after run, + if <0, count of bidi controls within run */ +} Run; + +/* in a Run, logicalStart will get this bit set if the run level is odd */ +#define INDEX_ODD_BIT (1UL<<31) + +#define MAKE_INDEX_ODD_PAIR(index, level) ((index)|((int32_t)((level)&1)<<31)) +#define ADD_ODD_BIT_FROM_LEVEL(x, level) ((x)|=((int32_t)((level)&1)<<31)) +#define REMOVE_ODD_BIT(x) ((x)&=~INDEX_ODD_BIT) + +#define GET_INDEX(x) ((x)&~INDEX_ODD_BIT) +#define GET_ODD_BIT(x) ((uint32_t)(x)>>31) +#define IS_ODD_RUN(x) ((UBool)(((x)&INDEX_ODD_BIT)!=0)) +#define IS_EVEN_RUN(x) ((UBool)(((x)&INDEX_ODD_BIT)==0)) + +U_CFUNC UBool +ubidi_getRuns(UBiDi *pBiDi, UErrorCode *pErrorCode); + +/** BiDi control code points */ +enum { + ZWNJ_CHAR=0x200c, + ZWJ_CHAR, + LRM_CHAR, + RLM_CHAR, + LRE_CHAR=0x202a, + RLE_CHAR, + PDF_CHAR, + LRO_CHAR, + RLO_CHAR, + LRI_CHAR=0x2066, + RLI_CHAR, + FSI_CHAR, + PDI_CHAR +}; + +#define IS_BIDI_CONTROL_CHAR(c) (((uint32_t)(c)&0xfffffffc)==ZWNJ_CHAR || (uint32_t)((c)-LRE_CHAR)<5 || (uint32_t)((c)-LRI_CHAR)<4) + +/* InsertPoints structure for noting where to put BiDi marks ---------------- */ + +typedef struct Point { + int32_t pos; /* position in text */ + int32_t flag; /* flag for LRM/RLM, before/after */ +} Point; + +typedef struct InsertPoints { + int32_t capacity; /* number of points allocated */ + int32_t size; /* number of points used */ + int32_t confirmed; /* number of points confirmed */ + UErrorCode errorCode; /* for eventual memory shortage */ + Point *points; /* pointer to array of points */ +} InsertPoints; + + +/* UBiDi structure ----------------------------------------------------------- */ + +struct UBiDi { + /* pointer to parent paragraph object (pointer to self if this object is + * a paragraph object); set to NULL in a newly opened object; set to a + * real value after a successful execution of ubidi_setPara or ubidi_setLine + */ + const UBiDi * pParaBiDi; + + /* alias pointer to the current text */ + const UChar *text; + + /* length of the current text */ + int32_t originalLength; + + /* if the UBIDI_OPTION_STREAMING option is set, this is the length + * of text actually processed by ubidi_setPara, which may be shorter than + * the original length. + * Otherwise, it is identical to the original length. + */ + int32_t length; + + /* if the UBIDI_OPTION_REMOVE_CONTROLS option is set, and/or + * marks are allowed to be inserted in one of the reordering mode, the + * length of the result string may be different from the processed length. + */ + int32_t resultLength; + + /* memory sizes in bytes */ + int32_t dirPropsSize, levelsSize, openingsSize, parasSize, runsSize, isolatesSize; + + /* allocated memory */ + DirProp *dirPropsMemory; + UBiDiLevel *levelsMemory; + Opening *openingsMemory; + Para *parasMemory; + Run *runsMemory; + Isolate *isolatesMemory; + + /* indicators for whether memory may be allocated after ubidi_open() */ + UBool mayAllocateText, mayAllocateRuns; + + /* arrays with one value per text-character */ + DirProp *dirProps; + UBiDiLevel *levels; + + /* are we performing an approximation of the "inverse BiDi" algorithm? */ + UBool isInverse; + + /* are we using the basic algorithm or its variation? */ + UBiDiReorderingMode reorderingMode; + + /* UBIDI_REORDER_xxx values must be ordered so that all the regular + * logical to visual modes come first, and all inverse BiDi modes + * come last. + */ + #define UBIDI_REORDER_LAST_LOGICAL_TO_VISUAL UBIDI_REORDER_NUMBERS_SPECIAL + + /* bitmask for reordering options */ + uint32_t reorderingOptions; + + /* must block separators receive level 0? */ + UBool orderParagraphsLTR; + + /* the paragraph level */ + UBiDiLevel paraLevel; + /* original paraLevel when contextual */ + /* must be one of UBIDI_DEFAULT_xxx or 0 if not contextual */ + UBiDiLevel defaultParaLevel; + + /* context data */ + const UChar *prologue; + int32_t proLength; + const UChar *epilogue; + int32_t epiLength; + + /* the following is set in ubidi_setPara, used in processPropertySeq */ + const struct ImpTabPair * pImpTabPair; /* pointer to levels state table pair */ + + /* the overall paragraph or line directionality - see UBiDiDirection */ + UBiDiDirection direction; + + /* flags is a bit set for which directional properties are in the text */ + Flags flags; + + /* lastArabicPos is index to the last AL in the text, -1 if none */ + int32_t lastArabicPos; + + /* characters after trailingWSStart are WS and are */ + /* implicitly at the paraLevel (rule (L1)) - levels may not reflect that */ + int32_t trailingWSStart; + + /* fields for paragraph handling */ + int32_t paraCount; /* set in getDirProps() */ + /* filled in getDirProps() */ + Para *paras; + + /* for relatively short text, we only need a tiny array of paras (no malloc()) */ + Para simpleParas[SIMPLE_PARAS_COUNT]; + + /* fields for line reordering */ + int32_t runCount; /* ==-1: runs not set up yet */ + Run *runs; + + /* for non-mixed text, we only need a tiny array of runs (no malloc()) */ + Run simpleRuns[1]; + + /* maximum or current nesting depth of isolate sequences */ + /* Within resolveExplicitLevels() and checkExplicitLevels(), this is the maximal + nesting encountered. + Within resolveImplicitLevels(), this is the index of the current isolates + stack entry. */ + int32_t isolateCount; + Isolate *isolates; + + /* for simple text, have a small stack (no malloc()) */ + Isolate simpleIsolates[SIMPLE_ISOLATES_COUNT]; + + /* for inverse Bidi with insertion of directional marks */ + InsertPoints insertPoints; + + /* for option UBIDI_OPTION_REMOVE_CONTROLS */ + int32_t controlCount; + + /* for Bidi class callback */ + UBiDiClassCallback *fnClassCallback; /* action pointer */ + const void *coClassCallback; /* context pointer */ +}; + +#define IS_VALID_PARA(x) ((x) && ((x)->pParaBiDi==(x))) +#define IS_VALID_PARA_OR_LINE(x) ((x) && ((x)->pParaBiDi==(x) || (((x)->pParaBiDi) && (x)->pParaBiDi->pParaBiDi==(x)->pParaBiDi))) + +typedef union { + DirProp *dirPropsMemory; + UBiDiLevel *levelsMemory; + Opening *openingsMemory; + Para *parasMemory; + Run *runsMemory; + Isolate *isolatesMemory; +} BidiMemoryForAllocation; + +/* Macros for initial checks at function entry */ +#define RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrcode, retvalue) UPRV_BLOCK_MACRO_BEGIN { \ + if((pErrcode)==NULL || U_FAILURE(*pErrcode)) return retvalue; \ +} UPRV_BLOCK_MACRO_END +#define RETURN_IF_NOT_VALID_PARA(bidi, errcode, retvalue) UPRV_BLOCK_MACRO_BEGIN { \ + if(!IS_VALID_PARA(bidi)) { \ + errcode=U_INVALID_STATE_ERROR; \ + return retvalue; \ + } \ +} UPRV_BLOCK_MACRO_END +#define RETURN_IF_NOT_VALID_PARA_OR_LINE(bidi, errcode, retvalue) UPRV_BLOCK_MACRO_BEGIN { \ + if(!IS_VALID_PARA_OR_LINE(bidi)) { \ + errcode=U_INVALID_STATE_ERROR; \ + return retvalue; \ + } \ +} UPRV_BLOCK_MACRO_END +#define RETURN_IF_BAD_RANGE(arg, start, limit, errcode, retvalue) UPRV_BLOCK_MACRO_BEGIN { \ + if((arg)<(start) || (arg)>=(limit)) { \ + (errcode)=U_ILLEGAL_ARGUMENT_ERROR; \ + return retvalue; \ + } \ +} UPRV_BLOCK_MACRO_END + +#define RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrcode) UPRV_BLOCK_MACRO_BEGIN { \ + if((pErrcode)==NULL || U_FAILURE(*pErrcode)) return; \ +} UPRV_BLOCK_MACRO_END +#define RETURN_VOID_IF_NOT_VALID_PARA(bidi, errcode) UPRV_BLOCK_MACRO_BEGIN { \ + if(!IS_VALID_PARA(bidi)) { \ + errcode=U_INVALID_STATE_ERROR; \ + return; \ + } \ +} UPRV_BLOCK_MACRO_END +#define RETURN_VOID_IF_NOT_VALID_PARA_OR_LINE(bidi, errcode) UPRV_BLOCK_MACRO_BEGIN { \ + if(!IS_VALID_PARA_OR_LINE(bidi)) { \ + errcode=U_INVALID_STATE_ERROR; \ + return; \ + } \ +} UPRV_BLOCK_MACRO_END +#define RETURN_VOID_IF_BAD_RANGE(arg, start, limit, errcode) UPRV_BLOCK_MACRO_BEGIN { \ + if((arg)<(start) || (arg)>=(limit)) { \ + (errcode)=U_ILLEGAL_ARGUMENT_ERROR; \ + return; \ + } \ +} UPRV_BLOCK_MACRO_END + +/* helper function to (re)allocate memory if allowed */ +U_CFUNC UBool +ubidi_getMemory(BidiMemoryForAllocation *pMemory, int32_t *pSize, UBool mayAllocate, int32_t sizeNeeded); + +/* helper macros for each allocated array in UBiDi */ +#define getDirPropsMemory(pBiDi, length) \ + ubidi_getMemory((BidiMemoryForAllocation *)&(pBiDi)->dirPropsMemory, &(pBiDi)->dirPropsSize, \ + (pBiDi)->mayAllocateText, (length)) + +#define getLevelsMemory(pBiDi, length) \ + ubidi_getMemory((BidiMemoryForAllocation *)&(pBiDi)->levelsMemory, &(pBiDi)->levelsSize, \ + (pBiDi)->mayAllocateText, (length)) + +#define getRunsMemory(pBiDi, length) \ + ubidi_getMemory((BidiMemoryForAllocation *)&(pBiDi)->runsMemory, &(pBiDi)->runsSize, \ + (pBiDi)->mayAllocateRuns, (length)*sizeof(Run)) + +/* additional macros used by ubidi_open() - always allow allocation */ +#define getInitialDirPropsMemory(pBiDi, length) \ + ubidi_getMemory((BidiMemoryForAllocation *)&(pBiDi)->dirPropsMemory, &(pBiDi)->dirPropsSize, \ + true, (length)) + +#define getInitialLevelsMemory(pBiDi, length) \ + ubidi_getMemory((BidiMemoryForAllocation *)&(pBiDi)->levelsMemory, &(pBiDi)->levelsSize, \ + true, (length)) + +#define getInitialOpeningsMemory(pBiDi, length) \ + ubidi_getMemory((BidiMemoryForAllocation *)&(pBiDi)->openingsMemory, &(pBiDi)->openingsSize, \ + true, (length)*sizeof(Opening)) + +#define getInitialParasMemory(pBiDi, length) \ + ubidi_getMemory((BidiMemoryForAllocation *)&(pBiDi)->parasMemory, &(pBiDi)->parasSize, \ + true, (length)*sizeof(Para)) + +#define getInitialRunsMemory(pBiDi, length) \ + ubidi_getMemory((BidiMemoryForAllocation *)&(pBiDi)->runsMemory, &(pBiDi)->runsSize, \ + true, (length)*sizeof(Run)) + +#define getInitialIsolatesMemory(pBiDi, length) \ + ubidi_getMemory((BidiMemoryForAllocation *)&(pBiDi)->isolatesMemory, &(pBiDi)->isolatesSize, \ + true, (length)*sizeof(Isolate)) + +#endif diff --git a/intl/icu/source/common/ubidiln.cpp b/intl/icu/source/common/ubidiln.cpp new file mode 100644 index 0000000000..63c4f9190a --- /dev/null +++ b/intl/icu/source/common/ubidiln.cpp @@ -0,0 +1,1347 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2015, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: ubidiln.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 1999aug06 +* created by: Markus W. Scherer, updated by Matitiahu Allouche +*/ + +#include "cmemory.h" +#include "unicode/utypes.h" +#include "unicode/ustring.h" +#include "unicode/uchar.h" +#include "unicode/ubidi.h" +#include "ubidiimp.h" +#include "uassert.h" + +/* + * General remarks about the functions in this file: + * + * These functions deal with the aspects of potentially mixed-directional + * text in a single paragraph or in a line of a single paragraph + * which has already been processed according to + * the Unicode 6.3 BiDi algorithm as defined in + * https://www.unicode.org/reports/tr9/ , version 28, + * also described in The Unicode Standard, Version 6.3.0 . + * + * This means that there is a UBiDi object with a levels + * and a dirProps array. + * paraLevel and direction are also set. + * Only if the length of the text is zero, then levels==dirProps==nullptr. + * + * The overall directionality of the paragraph + * or line is used to bypass the reordering steps if possible. + * Even purely RTL text does not need reordering there because + * the ubidi_getLogical/VisualIndex() functions can compute the + * index on the fly in such a case. + * + * The implementation of the access to same-level-runs and of the reordering + * do attempt to provide better performance and less memory usage compared to + * a direct implementation of especially rule (L2) with an array of + * one (32-bit) integer per text character. + * + * Here, the levels array is scanned as soon as necessary, and a vector of + * same-level-runs is created. Reordering then is done on this vector. + * For each run of text positions that were resolved to the same level, + * only 8 bytes are stored: the first text position of the run and the visual + * position behind the run after reordering. + * One sign bit is used to hold the directionality of the run. + * This is inefficient if there are many very short runs. If the average run + * length is <2, then this uses more memory. + * + * In a further attempt to save memory, the levels array is never changed + * after all the resolution rules (Xn, Wn, Nn, In). + * Many functions have to consider the field trailingWSStart: + * if it is less than length, then there is an implicit trailing run + * at the paraLevel, + * which is not reflected in the levels array. + * This allows a line UBiDi object to use the same levels array as + * its paragraph parent object. + * + * When a UBiDi object is created for a line of a paragraph, then the + * paragraph's levels and dirProps arrays are reused by way of setting + * a pointer into them, not by copying. This again saves memory and forbids to + * change the now shared levels for (L1). + */ + +/* handle trailing WS (L1) -------------------------------------------------- */ + +/* + * setTrailingWSStart() sets the start index for a trailing + * run of WS in the line. This is necessary because we do not modify + * the paragraph's levels array that we just point into. + * Using trailingWSStart is another form of performing (L1). + * + * To make subsequent operations easier, we also include the run + * before the WS if it is at the paraLevel - we merge the two here. + * + * This function is called only from ubidi_setLine(), so pBiDi->paraLevel is + * set correctly for the line even when contextual multiple paragraphs. + */ +static void +setTrailingWSStart(UBiDi *pBiDi) { + /* pBiDi->direction!=UBIDI_MIXED */ + + const DirProp *dirProps=pBiDi->dirProps; + UBiDiLevel *levels=pBiDi->levels; + int32_t start=pBiDi->length; + UBiDiLevel paraLevel=pBiDi->paraLevel; + + /* If the line is terminated by a block separator, all preceding WS etc... + are already set to paragraph level. + Setting trailingWSStart to pBidi->length will avoid changing the + level of B chars from 0 to paraLevel in ubidi_getLevels when + orderParagraphsLTR==true. + */ + if(dirProps[start-1]==B) { + pBiDi->trailingWSStart=start; /* currently == pBiDi->length */ + return; + } + /* go backwards across all WS, BN, explicit codes */ + while(start>0 && DIRPROP_FLAG(dirProps[start-1])&MASK_WS) { + --start; + } + + /* if the WS run can be merged with the previous run then do so here */ + while(start>0 && levels[start-1]==paraLevel) { + --start; + } + + pBiDi->trailingWSStart=start; +} + +/* ubidi_setLine ------------------------------------------------------------ */ + +U_CAPI void U_EXPORT2 +ubidi_setLine(const UBiDi *pParaBiDi, + int32_t start, int32_t limit, + UBiDi *pLineBiDi, + UErrorCode *pErrorCode) { + int32_t length; + + /* check the argument values */ + RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrorCode); + RETURN_VOID_IF_NOT_VALID_PARA(pParaBiDi, *pErrorCode); + RETURN_VOID_IF_BAD_RANGE(start, 0, limit, *pErrorCode); + RETURN_VOID_IF_BAD_RANGE(limit, 0, pParaBiDi->length+1, *pErrorCode); + if(pLineBiDi==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + if(ubidi_getParagraph(pParaBiDi, start, nullptr, nullptr, nullptr, pErrorCode) != + ubidi_getParagraph(pParaBiDi, limit-1, nullptr, nullptr, nullptr, pErrorCode)) { + /* the line crosses a paragraph boundary */ + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + /* set the values in pLineBiDi from its pParaBiDi parent */ + pLineBiDi->pParaBiDi=nullptr; /* mark unfinished setLine */ + pLineBiDi->text=pParaBiDi->text+start; + length=pLineBiDi->length=limit-start; + pLineBiDi->resultLength=pLineBiDi->originalLength=length; + pLineBiDi->paraLevel=GET_PARALEVEL(pParaBiDi, start); + pLineBiDi->paraCount=pParaBiDi->paraCount; + pLineBiDi->runs=nullptr; + pLineBiDi->flags=0; + pLineBiDi->reorderingMode=pParaBiDi->reorderingMode; + pLineBiDi->reorderingOptions=pParaBiDi->reorderingOptions; + pLineBiDi->controlCount=0; + if(pParaBiDi->controlCount>0) { + int32_t j; + for(j=start; jtext[j])) { + pLineBiDi->controlCount++; + } + } + pLineBiDi->resultLength-=pLineBiDi->controlCount; + } + + pLineBiDi->dirProps=pParaBiDi->dirProps+start; + pLineBiDi->levels=pParaBiDi->levels+start; + pLineBiDi->runCount=-1; + + if(pParaBiDi->direction!=UBIDI_MIXED) { + /* the parent is already trivial */ + pLineBiDi->direction=pParaBiDi->direction; + + /* + * The parent's levels are all either + * implicitly or explicitly ==paraLevel; + * do the same here. + */ + if(pParaBiDi->trailingWSStart<=start) { + pLineBiDi->trailingWSStart=0; + } else if(pParaBiDi->trailingWSStarttrailingWSStart=pParaBiDi->trailingWSStart-start; + } else { + pLineBiDi->trailingWSStart=length; + } + } else { + const UBiDiLevel *levels=pLineBiDi->levels; + int32_t i, trailingWSStart; + UBiDiLevel level; + + setTrailingWSStart(pLineBiDi); + trailingWSStart=pLineBiDi->trailingWSStart; + + /* recalculate pLineBiDi->direction */ + if(trailingWSStart==0) { + /* all levels are at paraLevel */ + pLineBiDi->direction=(UBiDiDirection)(pLineBiDi->paraLevel&1); + } else { + /* get the level of the first character */ + level=(UBiDiLevel)(levels[0]&1); + + /* if there is anything of a different level, then the line is mixed */ + if(trailingWSStartparaLevel&1)!=level) { + /* the trailing WS is at paraLevel, which differs from levels[0] */ + pLineBiDi->direction=UBIDI_MIXED; + } else { + /* see if levels[1..trailingWSStart-1] have the same direction as levels[0] and paraLevel */ + i=1; + for(;;) { + if(i==trailingWSStart) { + /* the direction values match those in level */ + pLineBiDi->direction=(UBiDiDirection)level; + break; + } else if((levels[i]&1)!=level) { + pLineBiDi->direction=UBIDI_MIXED; + break; + } + ++i; + } + } + } + + switch(pLineBiDi->direction) { + case UBIDI_LTR: + /* make sure paraLevel is even */ + pLineBiDi->paraLevel=(UBiDiLevel)((pLineBiDi->paraLevel+1)&~1); + + /* all levels are implicitly at paraLevel (important for ubidi_getLevels()) */ + pLineBiDi->trailingWSStart=0; + break; + case UBIDI_RTL: + /* make sure paraLevel is odd */ + pLineBiDi->paraLevel|=1; + + /* all levels are implicitly at paraLevel (important for ubidi_getLevels()) */ + pLineBiDi->trailingWSStart=0; + break; + default: + break; + } + } + pLineBiDi->pParaBiDi=pParaBiDi; /* mark successful setLine */ + return; +} + +U_CAPI UBiDiLevel U_EXPORT2 +ubidi_getLevelAt(const UBiDi *pBiDi, int32_t charIndex) { + /* return paraLevel if in the trailing WS run, otherwise the real level */ + if(!IS_VALID_PARA_OR_LINE(pBiDi) || charIndex<0 || pBiDi->length<=charIndex) { + return 0; + } else if(pBiDi->direction!=UBIDI_MIXED || charIndex>=pBiDi->trailingWSStart) { + return GET_PARALEVEL(pBiDi, charIndex); + } else { + return pBiDi->levels[charIndex]; + } +} + +U_CAPI const UBiDiLevel * U_EXPORT2 +ubidi_getLevels(UBiDi *pBiDi, UErrorCode *pErrorCode) { + int32_t start, length; + + RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrorCode, nullptr); + RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode, nullptr); + if((length=pBiDi->length)<=0) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + if((start=pBiDi->trailingWSStart)==length) { + /* the current levels array reflects the WS run */ + return pBiDi->levels; + } + + /* + * After the previous if(), we know that the levels array + * has an implicit trailing WS run and therefore does not fully + * reflect itself all the levels. + * This must be a UBiDi object for a line, and + * we need to create a new levels array. + */ + if(getLevelsMemory(pBiDi, length)) { + UBiDiLevel *levels=pBiDi->levelsMemory; + + if(start>0 && levels!=pBiDi->levels) { + uprv_memcpy(levels, pBiDi->levels, start); + } + /* pBiDi->paraLevel is ok even if contextual multiple paragraphs, + since pBidi is a line object */ + uprv_memset(levels+start, pBiDi->paraLevel, length-start); + + /* this new levels array is set for the line and reflects the WS run */ + pBiDi->trailingWSStart=length; + return pBiDi->levels=levels; + } else { + /* out of memory */ + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } +} + +U_CAPI void U_EXPORT2 +ubidi_getLogicalRun(const UBiDi *pBiDi, int32_t logicalPosition, + int32_t *pLogicalLimit, UBiDiLevel *pLevel) { + UErrorCode errorCode; + int32_t runCount, visualStart, logicalLimit, logicalFirst, i; + Run iRun; + + errorCode=U_ZERO_ERROR; + RETURN_VOID_IF_BAD_RANGE(logicalPosition, 0, pBiDi->length, errorCode); + /* ubidi_countRuns will check VALID_PARA_OR_LINE */ + runCount=ubidi_countRuns((UBiDi *)pBiDi, &errorCode); + if(U_FAILURE(errorCode)) { + return; + } + /* this is done based on runs rather than on levels since levels have + a special interpretation when UBIDI_REORDER_RUNS_ONLY + */ + visualStart=logicalLimit=0; + iRun=pBiDi->runs[0]; + + for(i=0; iruns[i]; + logicalFirst=GET_INDEX(iRun.logicalStart); + logicalLimit=logicalFirst+iRun.visualLimit-visualStart; + if((logicalPosition>=logicalFirst) && + (logicalPositionreorderingMode==UBIDI_REORDER_RUNS_ONLY) { + *pLevel=(UBiDiLevel)GET_ODD_BIT(iRun.logicalStart); + } + else if(pBiDi->direction!=UBIDI_MIXED || logicalPosition>=pBiDi->trailingWSStart) { + *pLevel=GET_PARALEVEL(pBiDi, logicalPosition); + } else { + *pLevel=pBiDi->levels[logicalPosition]; + } + } +} + +/* runs API functions ------------------------------------------------------- */ + +U_CAPI int32_t U_EXPORT2 +ubidi_countRuns(UBiDi *pBiDi, UErrorCode *pErrorCode) { + RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrorCode, -1); + RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode, -1); + ubidi_getRuns(pBiDi, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + return -1; + } + return pBiDi->runCount; +} + +U_CAPI UBiDiDirection U_EXPORT2 +ubidi_getVisualRun(UBiDi *pBiDi, int32_t runIndex, + int32_t *pLogicalStart, int32_t *pLength) +{ + int32_t start; + UErrorCode errorCode = U_ZERO_ERROR; + RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, errorCode, UBIDI_LTR); + ubidi_getRuns(pBiDi, &errorCode); + if(U_FAILURE(errorCode)) { + return UBIDI_LTR; + } + RETURN_IF_BAD_RANGE(runIndex, 0, pBiDi->runCount, errorCode, UBIDI_LTR); + + start=pBiDi->runs[runIndex].logicalStart; + if(pLogicalStart!=nullptr) { + *pLogicalStart=GET_INDEX(start); + } + if(pLength!=nullptr) { + if(runIndex>0) { + *pLength=pBiDi->runs[runIndex].visualLimit- + pBiDi->runs[runIndex-1].visualLimit; + } else { + *pLength=pBiDi->runs[0].visualLimit; + } + } + return (UBiDiDirection)GET_ODD_BIT(start); +} + +/* in trivial cases there is only one trivial run; called by ubidi_getRuns() */ +static void +getSingleRun(UBiDi *pBiDi, UBiDiLevel level) { + /* simple, single-run case */ + pBiDi->runs=pBiDi->simpleRuns; + pBiDi->runCount=1; + + /* fill and reorder the single run */ + pBiDi->runs[0].logicalStart=MAKE_INDEX_ODD_PAIR(0, level); + pBiDi->runs[0].visualLimit=pBiDi->length; + pBiDi->runs[0].insertRemove=0; +} + +/* reorder the runs array (L2) ---------------------------------------------- */ + +/* + * Reorder the same-level runs in the runs array. + * Here, runCount>1 and maxLevel>=minLevel>=paraLevel. + * All the visualStart fields=logical start before reordering. + * The "odd" bits are not set yet. + * + * Reordering with this data structure lends itself to some handy shortcuts: + * + * Since each run is moved but not modified, and since at the initial maxLevel + * each sequence of same-level runs consists of only one run each, we + * don't need to do anything there and can predecrement maxLevel. + * In many simple cases, the reordering is thus done entirely in the + * index mapping. + * Also, reordering occurs only down to the lowest odd level that occurs, + * which is minLevel|1. However, if the lowest level itself is odd, then + * in the last reordering the sequence of the runs at this level or higher + * will be all runs, and we don't need the elaborate loop to search for them. + * This is covered by ++minLevel instead of minLevel|=1 followed + * by an extra reorder-all after the reorder-some loop. + * About a trailing WS run: + * Such a run would need special treatment because its level is not + * reflected in levels[] if this is not a paragraph object. + * Instead, all characters from trailingWSStart on are implicitly at + * paraLevel. + * However, for all maxLevel>paraLevel, this run will never be reordered + * and does not need to be taken into account. maxLevel==paraLevel is only reordered + * if minLevel==paraLevel is odd, which is done in the extra segment. + * This means that for the main reordering loop we don't need to consider + * this run and can --runCount. If it is later part of the all-runs + * reordering, then runCount is adjusted accordingly. + */ +static void +reorderLine(UBiDi *pBiDi, UBiDiLevel minLevel, UBiDiLevel maxLevel) { + Run *runs, tempRun; + UBiDiLevel *levels; + int32_t firstRun, endRun, limitRun, runCount; + + /* nothing to do? */ + if(maxLevel<=(minLevel|1)) { + return; + } + + /* + * Reorder only down to the lowest odd level + * and reorder at an odd minLevel in a separate, simpler loop. + * See comments above for why minLevel is always incremented. + */ + ++minLevel; + + runs=pBiDi->runs; + levels=pBiDi->levels; + runCount=pBiDi->runCount; + + /* do not include the WS run at paraLevel<=old minLevel except in the simple loop */ + if(pBiDi->trailingWSStartlength) { + --runCount; + } + + while(--maxLevel>=minLevel) { + firstRun=0; + + /* loop for all sequences of runs */ + for(;;) { + /* look for a sequence of runs that are all at >=maxLevel */ + /* look for the first run of such a sequence */ + while(firstRun=runCount) { + break; /* no more such runs */ + } + + /* look for the limit run of such a sequence (the run behind it) */ + for(limitRun=firstRun; ++limitRun=maxLevel;) {} + + /* Swap the entire sequence of runs from firstRun to limitRun-1. */ + endRun=limitRun-1; + while(firstRuntrailingWSStart==pBiDi->length) { + --runCount; + } + + /* Swap the entire sequence of all runs. (endRun==runCount) */ + while(firstRunruns; + int32_t runCount=pBiDi->runCount, visualStart=0, i, length, logicalStart; + + for(i=0; i=logicalStart) && (logicalIndex<(logicalStart+length))) { + return i; + } + visualStart+=length; + } + /* we should never get here */ + UPRV_UNREACHABLE_EXIT; +} + +/* + * Compute the runs array from the levels array. + * After ubidi_getRuns() returns true, runCount is guaranteed to be >0 + * and the runs are reordered. + * Odd-level runs have visualStart on their visual right edge and + * they progress visually to the left. + * If option UBIDI_OPTION_INSERT_MARKS is set, insertRemove will contain the + * sum of appropriate LRM/RLM_BEFORE/AFTER flags. + * If option UBIDI_OPTION_REMOVE_CONTROLS is set, insertRemove will contain the + * negative number of BiDi control characters within this run. + */ +U_CFUNC UBool +ubidi_getRuns(UBiDi *pBiDi, UErrorCode*) { + /* + * This method returns immediately if the runs are already set. This + * includes the case of length==0 (handled in setPara).. + */ + if (pBiDi->runCount>=0) { + return true; + } + + if(pBiDi->direction!=UBIDI_MIXED) { + /* simple, single-run case - this covers length==0 */ + /* pBiDi->paraLevel is ok even for contextual multiple paragraphs */ + getSingleRun(pBiDi, pBiDi->paraLevel); + } else /* UBIDI_MIXED, length>0 */ { + /* mixed directionality */ + int32_t length=pBiDi->length, limit; + UBiDiLevel *levels=pBiDi->levels; + int32_t i, runCount; + UBiDiLevel level=UBIDI_DEFAULT_LTR; /* initialize with no valid level */ + /* + * If there are WS characters at the end of the line + * and the run preceding them has a level different from + * paraLevel, then they will form their own run at paraLevel (L1). + * Count them separately. + * We need some special treatment for this in order to not + * modify the levels array which a line UBiDi object shares + * with its paragraph parent and its other line siblings. + * In other words, for the trailing WS, it may be + * levels[]!=paraLevel but we have to treat it like it were so. + */ + limit=pBiDi->trailingWSStart; + /* count the runs, there is at least one non-WS run, and limit>0 */ + runCount=0; + for(i=0; i1 || limit1 */ + if(getRunsMemory(pBiDi, runCount)) { + runs=pBiDi->runsMemory; + } else { + return false; + } + + /* set the runs */ + /* FOOD FOR THOUGHT: this could be optimized, e.g.: + * 464->444, 484->444, 575->555, 595->555 + * However, that would take longer. Check also how it would + * interact with BiDi control removal and inserting Marks. + */ + runIndex=0; + + /* search for the run limits and initialize visualLimit values with the run lengths */ + i=0; + do { + /* prepare this run */ + start=i; + level=levels[i]; + if(levelmaxLevel) { + maxLevel=level; + } + + /* look for the run limit */ + while(++iparaLevel is ok even + if contextual multiple paragraphs. */ + if(pBiDi->paraLevelparaLevel; + } + } + + /* set the object fields */ + pBiDi->runs=runs; + pBiDi->runCount=runCount; + + reorderLine(pBiDi, minLevel, maxLevel); + + /* now add the direction flags and adjust the visualLimit's to be just that */ + /* this loop will also handle the trailing WS run */ + limit=0; + for(i=0; iparaLevel is ok even if + contextual multiple paragraphs. */ + if(runIndexparaLevel & 1) != 0)? 0 : runIndex; + + ADD_ODD_BIT_FROM_LEVEL(runs[trailingRun].logicalStart, pBiDi->paraLevel); + } + } + } + + /* handle insert LRM/RLM BEFORE/AFTER run */ + if(pBiDi->insertPoints.size>0) { + Point *point, *start=pBiDi->insertPoints.points, + *limit=start+pBiDi->insertPoints.size; + int32_t runIndex; + for(point=start; pointpos); + pBiDi->runs[runIndex].insertRemove|=point->flag; + } + } + + /* handle remove BiDi control characters */ + if(pBiDi->controlCount>0) { + int32_t runIndex; + const char16_t *start=pBiDi->text, *limit=start+pBiDi->length, *pu; + for(pu=start; puruns[runIndex].insertRemove--; + } + } + } + + return true; +} + +static UBool +prepareReorder(const UBiDiLevel *levels, int32_t length, + int32_t *indexMap, + UBiDiLevel *pMinLevel, UBiDiLevel *pMaxLevel) { + int32_t start; + UBiDiLevel level, minLevel, maxLevel; + + if(levels==nullptr || length<=0) { + return false; + } + + /* determine minLevel and maxLevel */ + minLevel=UBIDI_MAX_EXPLICIT_LEVEL+1; + maxLevel=0; + for(start=length; start>0;) { + level=levels[--start]; + if(level>UBIDI_MAX_EXPLICIT_LEVEL+1) { + return false; + } + if(levelmaxLevel) { + maxLevel=level; + } + } + *pMinLevel=minLevel; + *pMaxLevel=maxLevel; + + /* initialize the index map */ + for(start=length; start>0;) { + --start; + indexMap[start]=start; + } + + return true; +} + +/* reorder a line based on a levels array (L2) ------------------------------ */ + +U_CAPI void U_EXPORT2 +ubidi_reorderLogical(const UBiDiLevel *levels, int32_t length, int32_t *indexMap) { + int32_t start, limit, sumOfSosEos; + UBiDiLevel minLevel = 0, maxLevel = 0; + + if(indexMap==nullptr || !prepareReorder(levels, length, indexMap, &minLevel, &maxLevel)) { + return; + } + + /* nothing to do? */ + if(minLevel==maxLevel && (minLevel&1)==0) { + return; + } + + /* reorder only down to the lowest odd level */ + minLevel|=1; + + /* loop maxLevel..minLevel */ + do { + start=0; + + /* loop for all sequences of levels to reorder at the current maxLevel */ + for(;;) { + /* look for a sequence of levels that are all at >=maxLevel */ + /* look for the first index of such a sequence */ + while(start=length) { + break; /* no more such sequences */ + } + + /* look for the limit of such a sequence (the index behind it) */ + for(limit=start; ++limit=maxLevel;) {} + + /* + * sos=start of sequence, eos=end of sequence + * + * The closed (inclusive) interval from sos to eos includes all the logical + * and visual indexes within this sequence. They are logically and + * visually contiguous and in the same range. + * + * For each run, the new visual index=sos+eos-old visual index; + * we pre-add sos+eos into sumOfSosEos -> + * new visual index=sumOfSosEos-old visual index; + */ + sumOfSosEos=start+limit-1; + + /* reorder each index in the sequence */ + do { + indexMap[start]=sumOfSosEos-indexMap[start]; + } while(++start=minLevel); +} + +U_CAPI void U_EXPORT2 +ubidi_reorderVisual(const UBiDiLevel *levels, int32_t length, int32_t *indexMap) { + int32_t start, end, limit, temp; + UBiDiLevel minLevel = 0, maxLevel = 0; + + if(indexMap==nullptr || !prepareReorder(levels, length, indexMap, &minLevel, &maxLevel)) { + return; + } + + /* nothing to do? */ + if(minLevel==maxLevel && (minLevel&1)==0) { + return; + } + + /* reorder only down to the lowest odd level */ + minLevel|=1; + + /* loop maxLevel..minLevel */ + do { + start=0; + + /* loop for all sequences of levels to reorder at the current maxLevel */ + for(;;) { + /* look for a sequence of levels that are all at >=maxLevel */ + /* look for the first index of such a sequence */ + while(start=length) { + break; /* no more such runs */ + } + + /* look for the limit of such a sequence (the index behind it) */ + for(limit=start; ++limit=maxLevel;) {} + + /* + * Swap the entire interval of indexes from start to limit-1. + * We don't need to swap the levels for the purpose of this + * algorithm: the sequence of levels that we look at does not + * move anyway. + */ + end=limit-1; + while(start=minLevel); +} + +/* API functions for logical<->visual mapping ------------------------------- */ + +U_CAPI int32_t U_EXPORT2 +ubidi_getVisualIndex(UBiDi *pBiDi, int32_t logicalIndex, UErrorCode *pErrorCode) { + int32_t visualIndex=UBIDI_MAP_NOWHERE; + RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrorCode, -1); + RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode, -1); + RETURN_IF_BAD_RANGE(logicalIndex, 0, pBiDi->length, *pErrorCode, -1); + + /* we can do the trivial cases without the runs array */ + switch(pBiDi->direction) { + case UBIDI_LTR: + visualIndex=logicalIndex; + break; + case UBIDI_RTL: + visualIndex=pBiDi->length-logicalIndex-1; + break; + default: + if(!ubidi_getRuns(pBiDi, pErrorCode)) { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return -1; + } else { + Run *runs=pBiDi->runs; + int32_t i, visualStart=0, offset, length; + + /* linear search for the run, search on the visual runs */ + for(i=0; irunCount; ++i) { + length=runs[i].visualLimit-visualStart; + offset=logicalIndex-GET_INDEX(runs[i].logicalStart); + if(offset>=0 && offset=pBiDi->runCount) { + return UBIDI_MAP_NOWHERE; + } + } + } + + if(pBiDi->insertPoints.size>0) { + /* add the number of added marks until the calculated visual index */ + Run *runs=pBiDi->runs; + int32_t i, length, insertRemove; + int32_t visualStart=0, markFound=0; + for(i=0; ; i++, visualStart+=length) { + length=runs[i].visualLimit-visualStart; + insertRemove=runs[i].insertRemove; + if(insertRemove & (LRM_BEFORE|RLM_BEFORE)) { + markFound++; + } + /* is it the run containing the visual index? */ + if(visualIndexcontrolCount>0) { + /* subtract the number of controls until the calculated visual index */ + Run *runs=pBiDi->runs; + int32_t i, j, start, limit, length, insertRemove; + int32_t visualStart=0, controlFound=0; + char16_t uchar=pBiDi->text[logicalIndex]; + /* is the logical index pointing to a control ? */ + if(IS_BIDI_CONTROL_CHAR(uchar)) { + return UBIDI_MAP_NOWHERE; + } + /* loop on runs */ + for(i=0; ; i++, visualStart+=length) { + length=runs[i].visualLimit-visualStart; + insertRemove=runs[i].insertRemove; + /* calculated visual index is beyond this run? */ + if(visualIndex>=runs[i].visualLimit) { + controlFound-=insertRemove; + continue; + } + /* calculated visual index must be within current run */ + if(insertRemove==0) { + return visualIndex-controlFound; + } + if(IS_EVEN_RUN(runs[i].logicalStart)) { + /* LTR: check from run start to logical index */ + start=runs[i].logicalStart; + limit=logicalIndex; + } else { + /* RTL: check from logical index to run end */ + start=logicalIndex+1; + limit=GET_INDEX(runs[i].logicalStart)+length; + } + for(j=start; jtext[j]; + if(IS_BIDI_CONTROL_CHAR(uchar)) { + controlFound++; + } + } + return visualIndex-controlFound; + } + } + + return visualIndex; +} + +U_CAPI int32_t U_EXPORT2 +ubidi_getLogicalIndex(UBiDi *pBiDi, int32_t visualIndex, UErrorCode *pErrorCode) { + Run *runs; + int32_t i, runCount, start; + RETURN_IF_NULL_OR_FAILING_ERRCODE(pErrorCode, -1); + RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, *pErrorCode, -1); + RETURN_IF_BAD_RANGE(visualIndex, 0, pBiDi->resultLength, *pErrorCode, -1); + /* we can do the trivial cases without the runs array */ + if(pBiDi->insertPoints.size==0 && pBiDi->controlCount==0) { + if(pBiDi->direction==UBIDI_LTR) { + return visualIndex; + } + else if(pBiDi->direction==UBIDI_RTL) { + return pBiDi->length-visualIndex-1; + } + } + if(!ubidi_getRuns(pBiDi, pErrorCode)) { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return -1; + } + + runs=pBiDi->runs; + runCount=pBiDi->runCount; + if(pBiDi->insertPoints.size>0) { + /* handle inserted LRM/RLM */ + int32_t markFound=0, insertRemove; + int32_t visualStart=0, length; + runs=pBiDi->runs; + /* subtract number of marks until visual index */ + for(i=0; ; i++, visualStart+=length) { + length=runs[i].visualLimit-visualStart; + insertRemove=runs[i].insertRemove; + if(insertRemove&(LRM_BEFORE|RLM_BEFORE)) { + if(visualIndex<=(visualStart+markFound)) { + return UBIDI_MAP_NOWHERE; + } + markFound++; + } + /* is adjusted visual index within this run? */ + if(visualIndex<(runs[i].visualLimit+markFound)) { + visualIndex-=markFound; + break; + } + if(insertRemove&(LRM_AFTER|RLM_AFTER)) { + if(visualIndex==(visualStart+length+markFound)) { + return UBIDI_MAP_NOWHERE; + } + markFound++; + } + } + } + else if(pBiDi->controlCount>0) { + /* handle removed BiDi control characters */ + int32_t controlFound=0, insertRemove, length; + int32_t logicalStart, logicalEnd, visualStart=0, j, k; + char16_t uchar; + UBool evenRun; + /* add number of controls until visual index */ + for(i=0; ; i++, visualStart+=length) { + length=runs[i].visualLimit-visualStart; + insertRemove=runs[i].insertRemove; + /* is adjusted visual index beyond current run? */ + if(visualIndex>=(runs[i].visualLimit-controlFound+insertRemove)) { + controlFound-=insertRemove; + continue; + } + /* adjusted visual index is within current run */ + if(insertRemove==0) { + visualIndex+=controlFound; + break; + } + /* count non-control chars until visualIndex */ + logicalStart=runs[i].logicalStart; + evenRun=IS_EVEN_RUN(logicalStart); + REMOVE_ODD_BIT(logicalStart); + logicalEnd=logicalStart+length-1; + for(j=0; jtext[k]; + if(IS_BIDI_CONTROL_CHAR(uchar)) { + controlFound++; + } + if((visualIndex+controlFound)==(visualStart+j)) { + break; + } + } + visualIndex+=controlFound; + break; + } + } + /* handle all cases */ + if(runCount<=10) { + /* linear search for the run */ + for(i=0; visualIndex>=runs[i].visualLimit; ++i) {} + } else { + /* binary search for the run */ + int32_t begin=0, limit=runCount; + + /* the middle if() is guaranteed to find the run, we don't need a loop limit */ + for(;;) { + i=(begin+limit)/2; + if(visualIndex>=runs[i].visualLimit) { + begin=i+1; + } else if(i==0 || visualIndex>=runs[i-1].visualLimit) { + break; + } else { + limit=i; + } + } + } + + start=runs[i].logicalStart; + if(IS_EVEN_RUN(start)) { + /* LTR */ + /* the offset in runs[i] is visualIndex-runs[i-1].visualLimit */ + if(i>0) { + visualIndex-=runs[i-1].visualLimit; + } + return start+visualIndex; + } else { + /* RTL */ + return GET_INDEX(start)+runs[i].visualLimit-visualIndex-1; + } +} + +U_CAPI void U_EXPORT2 +ubidi_getLogicalMap(UBiDi *pBiDi, int32_t *indexMap, UErrorCode *pErrorCode) { + RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrorCode); + /* ubidi_countRuns() checks for VALID_PARA_OR_LINE */ + ubidi_countRuns(pBiDi, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + /* no op */ + } else if(indexMap==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + } else { + /* fill a logical-to-visual index map using the runs[] */ + int32_t visualStart, visualLimit, i, j, k; + int32_t logicalStart, logicalLimit; + Run *runs=pBiDi->runs; + if (pBiDi->length<=0) { + return; + } + if (pBiDi->length>pBiDi->resultLength) { + uprv_memset(indexMap, 0xFF, pBiDi->length*sizeof(int32_t)); + } + + visualStart=0; + for(j=0; jrunCount; ++j) { + logicalStart=GET_INDEX(runs[j].logicalStart); + visualLimit=runs[j].visualLimit; + if(IS_EVEN_RUN(runs[j].logicalStart)) { + do { /* LTR */ + indexMap[logicalStart++]=visualStart++; + } while(visualStartinsertPoints.size>0) { + int32_t markFound=0, runCount=pBiDi->runCount; + int32_t length, insertRemove; + visualStart=0; + /* add number of marks found until each index */ + for(i=0; i0) { + logicalStart=GET_INDEX(runs[i].logicalStart); + logicalLimit=logicalStart+length; + for(j=logicalStart; jcontrolCount>0) { + int32_t controlFound=0, runCount=pBiDi->runCount; + int32_t length, insertRemove; + UBool evenRun; + char16_t uchar; + visualStart=0; + /* subtract number of controls found until each index */ + for(i=0; itext[k]; + if(IS_BIDI_CONTROL_CHAR(uchar)) { + controlFound++; + indexMap[k]=UBIDI_MAP_NOWHERE; + continue; + } + indexMap[k]-=controlFound; + } + } + } + } +} + +U_CAPI void U_EXPORT2 +ubidi_getVisualMap(UBiDi *pBiDi, int32_t *indexMap, UErrorCode *pErrorCode) { + RETURN_VOID_IF_NULL_OR_FAILING_ERRCODE(pErrorCode); + if(indexMap==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + /* ubidi_countRuns() checks for VALID_PARA_OR_LINE */ + ubidi_countRuns(pBiDi, pErrorCode); + if(U_SUCCESS(*pErrorCode)) { + /* fill a visual-to-logical index map using the runs[] */ + Run *runs=pBiDi->runs, *runsLimit=runs+pBiDi->runCount; + int32_t logicalStart, visualStart, visualLimit, *pi=indexMap; + + if (pBiDi->resultLength<=0) { + return; + } + visualStart=0; + for(; runslogicalStart; + visualLimit=runs->visualLimit; + if(IS_EVEN_RUN(logicalStart)) { + do { /* LTR */ + *pi++ = logicalStart++; + } while(++visualStartinsertPoints.size>0) { + int32_t markFound=0, runCount=pBiDi->runCount; + int32_t insertRemove, i, j, k; + runs=pBiDi->runs; + /* count all inserted marks */ + for(i=0; iresultLength; + for(i=runCount-1; i>=0 && markFound>0; i--) { + insertRemove=runs[i].insertRemove; + if(insertRemove&(LRM_AFTER|RLM_AFTER)) { + indexMap[--k]= UBIDI_MAP_NOWHERE; + markFound--; + } + visualStart= i>0 ? runs[i-1].visualLimit : 0; + for(j=runs[i].visualLimit-1; j>=visualStart && markFound>0; j--) { + indexMap[--k]=indexMap[j]; + } + if(insertRemove&(LRM_BEFORE|RLM_BEFORE)) { + indexMap[--k]= UBIDI_MAP_NOWHERE; + markFound--; + } + } + } + else if(pBiDi->controlCount>0) { + int32_t runCount=pBiDi->runCount, logicalEnd; + int32_t insertRemove, length, i, j, k, m; + char16_t uchar; + UBool evenRun; + runs=pBiDi->runs; + visualStart=0; + /* move forward indexes by number of preceding controls */ + k=0; + for(i=0; itext[m]; + if(!IS_BIDI_CONTROL_CHAR(uchar)) { + indexMap[k++]=m; + } + } + } + } + } +} + +U_CAPI void U_EXPORT2 +ubidi_invertMap(const int32_t *srcMap, int32_t *destMap, int32_t length) { + if(srcMap!=nullptr && destMap!=nullptr && length>0) { + const int32_t *pi; + int32_t destLength=-1, count=0; + /* find highest value and count positive indexes in srcMap */ + pi=srcMap+length; + while(pi>srcMap) { + if(*--pi>destLength) { + destLength=*pi; + } + if(*pi>=0) { + count++; + } + } + destLength++; /* add 1 for origin 0 */ + if(count0) { + if(*--pi>=0) { + destMap[*pi]=--length; + } else { + --length; + } + } + } +} diff --git a/intl/icu/source/common/ubiditransform.cpp b/intl/icu/source/common/ubiditransform.cpp new file mode 100644 index 0000000000..01f5901a2c --- /dev/null +++ b/intl/icu/source/common/ubiditransform.cpp @@ -0,0 +1,530 @@ +/* +****************************************************************************** +* +* © 2016 and later: Unicode, Inc. and others. +* License & terms of use: http://www.unicode.org/copyright.html +* +****************************************************************************** +* file name: ubiditransform.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2016jul24 +* created by: Lina Kemmel +* +*/ + +#include "cmemory.h" +#include "unicode/ubidi.h" +#include "unicode/ustring.h" +#include "unicode/ushape.h" +#include "unicode/utf16.h" +#include "ustr_imp.h" +#include "unicode/ubiditransform.h" + +/* Some convenience defines */ +#define LTR UBIDI_LTR +#define RTL UBIDI_RTL +#define LOGICAL UBIDI_LOGICAL +#define VISUAL UBIDI_VISUAL +#define SHAPE_LOGICAL U_SHAPE_TEXT_DIRECTION_LOGICAL +#define SHAPE_VISUAL U_SHAPE_TEXT_DIRECTION_VISUAL_LTR + +#define CHECK_LEN(STR, LEN, ERROR) UPRV_BLOCK_MACRO_BEGIN { \ + if (LEN == 0) return 0; \ + if (LEN < -1) { *(ERROR) = U_ILLEGAL_ARGUMENT_ERROR; return 0; } \ + if (LEN == -1) LEN = u_strlen(STR); \ +} UPRV_BLOCK_MACRO_END + +#define MAX_ACTIONS 7 + +/** + * Typedef for a pointer to a function, which performs some operation (such as + * reordering, setting "inverse" mode, character mirroring, etc.). Return value + * indicates whether the text was changed in the course of this operation or + * not. + */ +typedef UBool (*UBiDiAction)(UBiDiTransform *, UErrorCode *); + +/** + * Structure that holds a predefined reordering scheme, including the following + * information: + *

    + *
  • an input base direction,
    • + *
  • an input order,
    • + *
  • an output base direction,
    • + *
  • an output order,
    • + *
  • a digit shaping direction,
    • + *
  • a letter shaping direction,
    • + *
  • a base direction that should be applied when the reordering engine is + * invoked (which can not always be derived from the caller-defined + * options),
    • + *
  • an array of pointers to functions that accomplish the bidi layout + * transformation.
    • + *
+ */ +typedef struct { + UBiDiLevel inLevel; /* input level */ + UBiDiOrder inOrder; /* input order */ + UBiDiLevel outLevel; /* output level */ + UBiDiOrder outOrder; /* output order */ + uint32_t digitsDir; /* digit shaping direction */ + uint32_t lettersDir; /* letter shaping direction */ + UBiDiLevel baseLevel; /* paragraph level to be used with setPara */ + const UBiDiAction actions[MAX_ACTIONS]; /* array of pointers to functions carrying out the transformation */ +} ReorderingScheme; + +struct UBiDiTransform { + UBiDi *pBidi; /* pointer to a UBiDi object */ + const ReorderingScheme *pActiveScheme; /* effective reordering scheme */ + char16_t *src; /* input text */ + char16_t *dest; /* output text */ + uint32_t srcLength; /* input text length - not really needed as we are zero-terminated and can u_strlen */ + uint32_t srcSize; /* input text capacity excluding the trailing zero */ + uint32_t destSize; /* output text capacity */ + uint32_t *pDestLength; /* number of UChars written to dest */ + uint32_t reorderingOptions; /* reordering options - currently only suppot DO_MIRRORING */ + uint32_t digits; /* digit option for ArabicShaping */ + uint32_t letters; /* letter option for ArabicShaping */ +}; + +U_CAPI UBiDiTransform* U_EXPORT2 +ubiditransform_open(UErrorCode *pErrorCode) +{ + UBiDiTransform *pBiDiTransform = nullptr; + if (U_SUCCESS(*pErrorCode)) { + pBiDiTransform = (UBiDiTransform*) uprv_calloc(1, sizeof(UBiDiTransform)); + if (pBiDiTransform == nullptr) { + *pErrorCode = U_MEMORY_ALLOCATION_ERROR; + } + } + return pBiDiTransform; +} + +U_CAPI void U_EXPORT2 +ubiditransform_close(UBiDiTransform *pBiDiTransform) +{ + if (pBiDiTransform != nullptr) { + if (pBiDiTransform->pBidi != nullptr) { + ubidi_close(pBiDiTransform->pBidi); + } + if (pBiDiTransform->src != nullptr) { + uprv_free(pBiDiTransform->src); + } + uprv_free(pBiDiTransform); + } +} + +/** + * Performs Bidi resolution of text. + * + * @param pTransform Pointer to the UBiDiTransform structure. + * @param pErrorCode Pointer to the error code value. + * + * @return Whether or not this function modifies the text. Besides the return + * value, the caller should also check U_SUCCESS(*pErrorCode). + */ +static UBool +action_resolve(UBiDiTransform *pTransform, UErrorCode *pErrorCode) +{ + ubidi_setPara(pTransform->pBidi, pTransform->src, pTransform->srcLength, + pTransform->pActiveScheme->baseLevel, nullptr, pErrorCode); + return false; +} + +/** + * Performs basic reordering of text (Logical -> Visual LTR). + * + * @param pTransform Pointer to the UBiDiTransform structure. + * @param pErrorCode Pointer to the error code value. + * + * @return Whether or not this function modifies the text. Besides the return + * value, the caller should also check U_SUCCESS(*pErrorCode). + */ +static UBool +action_reorder(UBiDiTransform *pTransform, UErrorCode *pErrorCode) +{ + ubidi_writeReordered(pTransform->pBidi, pTransform->dest, pTransform->destSize, + static_cast(pTransform->reorderingOptions), pErrorCode); + + *pTransform->pDestLength = pTransform->srcLength; + pTransform->reorderingOptions = UBIDI_REORDER_DEFAULT; + return true; +} + +/** + * Sets "inverse" mode on the UBiDi object. + * + * @param pTransform Pointer to the UBiDiTransform structure. + * @param pErrorCode Pointer to the error code value. + * + * @return Whether or not this function modifies the text. Besides the return + * value, the caller should also check U_SUCCESS(*pErrorCode). + */ +static UBool +action_setInverse(UBiDiTransform *pTransform, UErrorCode *pErrorCode) +{ + (void)pErrorCode; + ubidi_setInverse(pTransform->pBidi, true); + ubidi_setReorderingMode(pTransform->pBidi, UBIDI_REORDER_INVERSE_LIKE_DIRECT); + return false; +} + +/** + * Sets "runs only" reordering mode indicating a Logical LTR <-> Logical RTL + * transformation. + * + * @param pTransform Pointer to the UBiDiTransform structure. + * @param pErrorCode Pointer to the error code value. + * + * @return Whether or not this function modifies the text. Besides the return + * value, the caller should also check U_SUCCESS(*pErrorCode). + */ +static UBool +action_setRunsOnly(UBiDiTransform *pTransform, UErrorCode *pErrorCode) +{ + (void)pErrorCode; + ubidi_setReorderingMode(pTransform->pBidi, UBIDI_REORDER_RUNS_ONLY); + return false; +} + +/** + * Performs string reverse. + * + * @param pTransform Pointer to the UBiDiTransform structure. + * @param pErrorCode Pointer to the error code value. + * + * @return Whether or not this function modifies the text. Besides the return + * value, the caller should also check U_SUCCESS(*pErrorCode). + */ +static UBool +action_reverse(UBiDiTransform *pTransform, UErrorCode *pErrorCode) +{ + ubidi_writeReverse(pTransform->src, pTransform->srcLength, + pTransform->dest, pTransform->destSize, + UBIDI_REORDER_DEFAULT, pErrorCode); + *pTransform->pDestLength = pTransform->srcLength; + return true; +} + +/** + * Applies a new value to the text that serves as input at the current + * processing step. This value is identical to the original one when we begin + * the processing, but usually changes as the transformation progresses. + * + * @param pTransform A pointer to the UBiDiTransform structure. + * @param newSrc A pointer whose value is to be used as input text. + * @param newLength A length of the new text in char16_ts. + * @param newSize A new source capacity in char16_ts. + * @param pErrorCode Pointer to the error code value. + */ +static void +updateSrc(UBiDiTransform *pTransform, const char16_t *newSrc, uint32_t newLength, + uint32_t newSize, UErrorCode *pErrorCode) +{ + if (newSize < newLength) { + *pErrorCode = U_BUFFER_OVERFLOW_ERROR; + return; + } + if (newSize > pTransform->srcSize) { + newSize += 50; // allocate slightly more than needed right now + if (pTransform->src != nullptr) { + uprv_free(pTransform->src); + pTransform->src = nullptr; + } + pTransform->src = (char16_t *)uprv_malloc(newSize * sizeof(char16_t)); + if (pTransform->src == nullptr) { + *pErrorCode = U_MEMORY_ALLOCATION_ERROR; + //pTransform->srcLength = pTransform->srcSize = 0; + return; + } + pTransform->srcSize = newSize; + } + u_strncpy(pTransform->src, newSrc, newLength); + pTransform->srcLength = u_terminateUChars(pTransform->src, + pTransform->srcSize, newLength, pErrorCode); +} + +/** + * Calls a lower level shaping function. + * + * @param pTransform Pointer to the UBiDiTransform structure. + * @param options Shaping options. + * @param pErrorCode Pointer to the error code value. + */ +static void +doShape(UBiDiTransform *pTransform, uint32_t options, UErrorCode *pErrorCode) +{ + *pTransform->pDestLength = u_shapeArabic(pTransform->src, + pTransform->srcLength, pTransform->dest, pTransform->destSize, + options, pErrorCode); +} + +/** + * Performs digit and letter shaping. + * + * @param pTransform Pointer to the UBiDiTransform structure. + * @param pErrorCode Pointer to the error code value. + * + * @return Whether or not this function modifies the text. Besides the return + * value, the caller should also check U_SUCCESS(*pErrorCode). + */ +static UBool +action_shapeArabic(UBiDiTransform *pTransform, UErrorCode *pErrorCode) +{ + if ((pTransform->letters | pTransform->digits) == 0) { + return false; + } + if (pTransform->pActiveScheme->lettersDir == pTransform->pActiveScheme->digitsDir) { + doShape(pTransform, pTransform->letters | pTransform->digits | pTransform->pActiveScheme->lettersDir, + pErrorCode); + } else { + doShape(pTransform, pTransform->digits | pTransform->pActiveScheme->digitsDir, pErrorCode); + if (U_SUCCESS(*pErrorCode)) { + updateSrc(pTransform, pTransform->dest, *pTransform->pDestLength, + *pTransform->pDestLength, pErrorCode); + doShape(pTransform, pTransform->letters | pTransform->pActiveScheme->lettersDir, + pErrorCode); + } + } + return true; +} + +/** + * Performs character mirroring. + * + * @param pTransform Pointer to the UBiDiTransform structure. + * @param pErrorCode Pointer to the error code value. + * + * @return Whether or not this function modifies the text. Besides the return + * value, the caller should also check U_SUCCESS(*pErrorCode). + */ +static UBool +action_mirror(UBiDiTransform *pTransform, UErrorCode *pErrorCode) +{ + UChar32 c; + uint32_t i = 0, j = 0; + if (0 == (pTransform->reorderingOptions & UBIDI_DO_MIRRORING)) { + return false; + } + if (pTransform->destSize < pTransform->srcLength) { + *pErrorCode = U_BUFFER_OVERFLOW_ERROR; + return false; + } + do { + UBool isOdd = ubidi_getLevelAt(pTransform->pBidi, i) & 1; + U16_NEXT(pTransform->src, i, pTransform->srcLength, c); + U16_APPEND_UNSAFE(pTransform->dest, j, isOdd ? u_charMirror(c) : c); + } while (i < pTransform->srcLength); + + *pTransform->pDestLength = pTransform->srcLength; + pTransform->reorderingOptions = UBIDI_REORDER_DEFAULT; + return true; +} + +/** + * All possible reordering schemes. + * + */ +static const ReorderingScheme Schemes[] = +{ + /* 0: Logical LTR => Visual LTR */ + {LTR, LOGICAL, LTR, VISUAL, SHAPE_LOGICAL, SHAPE_LOGICAL, LTR, + {action_shapeArabic, action_resolve, action_reorder, nullptr}}, + /* 1: Logical RTL => Visual LTR */ + {RTL, LOGICAL, LTR, VISUAL, SHAPE_LOGICAL, SHAPE_VISUAL, RTL, + {action_resolve, action_reorder, action_shapeArabic, nullptr}}, + /* 2: Logical LTR => Visual RTL */ + {LTR, LOGICAL, RTL, VISUAL, SHAPE_LOGICAL, SHAPE_LOGICAL, LTR, + {action_shapeArabic, action_resolve, action_reorder, action_reverse, nullptr}}, + /* 3: Logical RTL => Visual RTL */ + {RTL, LOGICAL, RTL, VISUAL, SHAPE_LOGICAL, SHAPE_VISUAL, RTL, + {action_resolve, action_reorder, action_shapeArabic, action_reverse, nullptr}}, + /* 4: Visual LTR => Logical RTL */ + {LTR, VISUAL, RTL, LOGICAL, SHAPE_LOGICAL, SHAPE_VISUAL, RTL, + {action_shapeArabic, action_setInverse, action_resolve, action_reorder, nullptr}}, + /* 5: Visual RTL => Logical RTL */ + {RTL, VISUAL, RTL, LOGICAL, SHAPE_LOGICAL, SHAPE_VISUAL, RTL, + {action_reverse, action_shapeArabic, action_setInverse, action_resolve, action_reorder, nullptr}}, + /* 6: Visual LTR => Logical LTR */ + {LTR, VISUAL, LTR, LOGICAL, SHAPE_LOGICAL, SHAPE_LOGICAL, LTR, + {action_setInverse, action_resolve, action_reorder, action_shapeArabic, nullptr}}, + /* 7: Visual RTL => Logical LTR */ + {RTL, VISUAL, LTR, LOGICAL, SHAPE_LOGICAL, SHAPE_LOGICAL, LTR, + {action_reverse, action_setInverse, action_resolve, action_reorder, action_shapeArabic, nullptr}}, + /* 8: Logical LTR => Logical RTL */ + {LTR, LOGICAL, RTL, LOGICAL, SHAPE_LOGICAL, SHAPE_LOGICAL, LTR, + {action_shapeArabic, action_resolve, action_mirror, action_setRunsOnly, action_resolve, action_reorder, nullptr}}, + /* 9: Logical RTL => Logical LTR */ + {RTL, LOGICAL, LTR, LOGICAL, SHAPE_LOGICAL, SHAPE_LOGICAL, RTL, + {action_resolve, action_mirror, action_setRunsOnly, action_resolve, action_reorder, action_shapeArabic, nullptr}}, + /* 10: Visual LTR => Visual RTL */ + {LTR, VISUAL, RTL, VISUAL, SHAPE_LOGICAL, SHAPE_VISUAL, LTR, + {action_shapeArabic, action_setInverse, action_resolve, action_mirror, action_reverse, nullptr}}, + /* 11: Visual RTL => Visual LTR */ + {RTL, VISUAL, LTR, VISUAL, SHAPE_LOGICAL, SHAPE_VISUAL, LTR, + {action_reverse, action_shapeArabic, action_setInverse, action_resolve, action_mirror, nullptr}}, + /* 12: Logical LTR => Logical LTR */ + {LTR, LOGICAL, LTR, LOGICAL, SHAPE_LOGICAL, SHAPE_LOGICAL, LTR, + {action_resolve, action_mirror, action_shapeArabic, nullptr}}, + /* 13: Logical RTL => Logical RTL */ + {RTL, LOGICAL, RTL, LOGICAL, SHAPE_VISUAL, SHAPE_LOGICAL, RTL, + {action_resolve, action_mirror, action_shapeArabic, nullptr}}, + /* 14: Visual LTR => Visual LTR */ + {LTR, VISUAL, LTR, VISUAL, SHAPE_LOGICAL, SHAPE_VISUAL, LTR, + {action_resolve, action_mirror, action_shapeArabic, nullptr}}, + /* 15: Visual RTL => Visual RTL */ + {RTL, VISUAL, RTL, VISUAL, SHAPE_LOGICAL, SHAPE_VISUAL, LTR, + {action_reverse, action_resolve, action_mirror, action_shapeArabic, action_reverse, nullptr}} +}; + +static const uint32_t nSchemes = sizeof(Schemes) / sizeof(*Schemes); + +/** + * When the direction option is UBIDI_DEFAULT_LTR or + * UBIDI_DEFAULT_RTL, resolve the base direction according to that + * of the first strong bidi character. + */ +static void +resolveBaseDirection(const char16_t *text, uint32_t length, + UBiDiLevel *pInLevel, UBiDiLevel *pOutLevel) +{ + switch (*pInLevel) { + case UBIDI_DEFAULT_LTR: + case UBIDI_DEFAULT_RTL: { + UBiDiLevel level = static_cast(ubidi_getBaseDirection(text, length)); + *pInLevel = static_cast(level != UBIDI_NEUTRAL) ? level + : *pInLevel == UBIDI_DEFAULT_RTL ? static_cast(RTL) : static_cast(LTR); + break; + } + default: + *pInLevel &= 1; + break; + } + switch (*pOutLevel) { + case UBIDI_DEFAULT_LTR: + case UBIDI_DEFAULT_RTL: + *pOutLevel = *pInLevel; + break; + default: + *pOutLevel &= 1; + break; + } +} + +/** + * Finds a valid ReorderingScheme matching the + * caller-defined scheme. + * + * @return A valid ReorderingScheme object or nullptr + */ +static const ReorderingScheme* +findMatchingScheme(UBiDiLevel inLevel, UBiDiLevel outLevel, + UBiDiOrder inOrder, UBiDiOrder outOrder) +{ + uint32_t i; + for (i = 0; i < nSchemes; i++) { + const ReorderingScheme *pScheme = Schemes + i; + if (inLevel == pScheme->inLevel && outLevel == pScheme->outLevel + && inOrder == pScheme->inOrder && outOrder == pScheme->outOrder) { + return pScheme; + } + } + return nullptr; +} + +U_CAPI uint32_t U_EXPORT2 +ubiditransform_transform(UBiDiTransform *pBiDiTransform, + const char16_t *src, int32_t srcLength, + char16_t *dest, int32_t destSize, + UBiDiLevel inParaLevel, UBiDiOrder inOrder, + UBiDiLevel outParaLevel, UBiDiOrder outOrder, + UBiDiMirroring doMirroring, uint32_t shapingOptions, + UErrorCode *pErrorCode) +{ + uint32_t destLength = 0; + UBool textChanged = false; + const UBiDiTransform *pOrigTransform = pBiDiTransform; + const UBiDiAction *action = nullptr; + + if (U_FAILURE(*pErrorCode)) { + return 0; + } + if (src == nullptr || dest == nullptr) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + CHECK_LEN(src, srcLength, pErrorCode); + CHECK_LEN(dest, destSize, pErrorCode); + + if (pBiDiTransform == nullptr) { + pBiDiTransform = ubiditransform_open(pErrorCode); + if (U_FAILURE(*pErrorCode)) { + return 0; + } + } + /* Current limitation: in multiple paragraphs will be resolved according + to the 1st paragraph */ + resolveBaseDirection(src, srcLength, &inParaLevel, &outParaLevel); + + pBiDiTransform->pActiveScheme = findMatchingScheme(inParaLevel, outParaLevel, + inOrder, outOrder); + if (pBiDiTransform->pActiveScheme == nullptr) { + goto cleanup; + } + pBiDiTransform->reorderingOptions = doMirroring ? UBIDI_DO_MIRRORING + : UBIDI_REORDER_DEFAULT; + + /* Ignore TEXT_DIRECTION_* flags, as we apply our own depending on the text + scheme at the time shaping is invoked. */ + shapingOptions &= ~U_SHAPE_TEXT_DIRECTION_MASK; + pBiDiTransform->digits = shapingOptions & ~U_SHAPE_LETTERS_MASK; + pBiDiTransform->letters = shapingOptions & ~U_SHAPE_DIGITS_MASK; + + updateSrc(pBiDiTransform, src, srcLength, destSize > srcLength ? destSize : srcLength, pErrorCode); + if (U_FAILURE(*pErrorCode)) { + goto cleanup; + } + if (pBiDiTransform->pBidi == nullptr) { + pBiDiTransform->pBidi = ubidi_openSized(0, 0, pErrorCode); + if (U_FAILURE(*pErrorCode)) { + goto cleanup; + } + } + pBiDiTransform->dest = dest; + pBiDiTransform->destSize = destSize; + pBiDiTransform->pDestLength = &destLength; + + /* Checking for U_SUCCESS() within the loop to bail out on first failure. */ + for (action = pBiDiTransform->pActiveScheme->actions; *action && U_SUCCESS(*pErrorCode); action++) { + if ((*action)(pBiDiTransform, pErrorCode)) { + if (action + 1) { + updateSrc(pBiDiTransform, pBiDiTransform->dest, *pBiDiTransform->pDestLength, + *pBiDiTransform->pDestLength, pErrorCode); + } + textChanged = true; + } + } + ubidi_setInverse(pBiDiTransform->pBidi, false); + + if (!textChanged && U_SUCCESS(*pErrorCode)) { + /* Text was not changed - just copy src to dest */ + if (destSize < srcLength) { + *pErrorCode = U_BUFFER_OVERFLOW_ERROR; + } else { + u_strncpy(dest, src, srcLength); + destLength = srcLength; + } + } +cleanup: + if (pOrigTransform != pBiDiTransform) { + ubiditransform_close(pBiDiTransform); + } else { + pBiDiTransform->dest = nullptr; + pBiDiTransform->pDestLength = nullptr; + pBiDiTransform->srcLength = 0; + pBiDiTransform->destSize = 0; + } + return U_FAILURE(*pErrorCode) ? 0 : destLength; +} diff --git a/intl/icu/source/common/ubidiwrt.cpp b/intl/icu/source/common/ubidiwrt.cpp new file mode 100644 index 0000000000..969807c243 --- /dev/null +++ b/intl/icu/source/common/ubidiwrt.cpp @@ -0,0 +1,650 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2000-2015, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: ubidiwrt.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 1999aug06 +* created by: Markus W. Scherer, updated by Matitiahu Allouche +* +* This file contains implementations for BiDi functions that use +* the core algorithm and core API to write reordered text. +*/ + +#include "unicode/utypes.h" +#include "unicode/ustring.h" +#include "unicode/uchar.h" +#include "unicode/ubidi.h" +#include "unicode/utf16.h" +#include "cmemory.h" +#include "ustr_imp.h" +#include "ubidiimp.h" + +/* + * The function implementations in this file are designed + * for UTF-16 and UTF-32, not for UTF-8. + * + * Assumptions that are not true for UTF-8: + * - Any code point always needs the same number of code units + * ("minimum-length-problem" of UTF-8) + * - The BiDi control characters need only one code unit each + * + * Further assumptions for all UTFs: + * - u_charMirror(c) needs the same number of code units as c + */ +#if defined(UTF_SIZE) && UTF_SIZE==8 +# error reimplement ubidi_writeReordered() for UTF-8, see comment above +#endif + +#define IS_COMBINING(type) ((1UL<<(type))&(1UL<0); + return srcLength; + } + case UBIDI_DO_MIRRORING: { + /* do mirroring */ + int32_t i=0, j=0; + UChar32 c; + + if(destSize0) { + c=*src++; + if(!IS_BIDI_CONTROL_CHAR(c)) { + --remaining; + } + } + return destSize-remaining; + } + *dest++=c; + } + } while(--srcLength>0); + return destSize-remaining; + } + default: { + /* remove BiDi control characters and do mirroring */ + int32_t remaining=destSize; + int32_t i, j=0; + UChar32 c; + do { + i=0; + U16_NEXT(src, i, srcLength, c); + src+=i; + srcLength-=i; + if(!IS_BIDI_CONTROL_CHAR(c)) { + remaining-=i; + if(remaining<0) { + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + + /* preflight the length */ + while(srcLength>0) { + c=*src++; + if(!IS_BIDI_CONTROL_CHAR(c)) { + --remaining; + } + --srcLength; + } + return destSize-remaining; + } + c=u_charMirror(c); + U16_APPEND_UNSAFE(dest, j, c); + } + } while(srcLength>0); + return j; + } + } /* end of switch */ +} + +static int32_t +doWriteReverse(const char16_t *src, int32_t srcLength, + char16_t *dest, int32_t destSize, + uint16_t options, + UErrorCode *pErrorCode) { + /* + * RTL run - + * + * RTL runs need to be copied to the destination in reverse order + * of code points, not code units, to keep Unicode characters intact. + * + * The general strategy for this is to read the source text + * in backward order, collect all code units for a code point + * (and optionally following combining characters, see below), + * and copy all these code units in ascending order + * to the destination for this run. + * + * Several options request whether combining characters + * should be kept after their base characters, + * whether BiDi control characters should be removed, and + * whether characters should be replaced by their mirror-image + * equivalent Unicode characters. + */ + int32_t i, j; + UChar32 c; + + /* optimize for several combinations of options */ + switch(options&(UBIDI_REMOVE_BIDI_CONTROLS|UBIDI_DO_MIRRORING|UBIDI_KEEP_BASE_COMBINING)) { + case 0: + /* + * With none of the "complicated" options set, the destination + * run will have the same length as the source run, + * and there is no mirroring and no keeping combining characters + * with their base characters. + */ + if(destSize0); + break; + case UBIDI_KEEP_BASE_COMBINING: + /* + * Here, too, the destination + * run will have the same length as the source run, + * and there is no mirroring. + * We do need to keep combining characters with their base characters. + */ + if(destSize0 && IS_COMBINING(u_charType(c))); + + /* copy this "user character" */ + j=srcLength; + do { + *dest++=src[j++]; + } while(j0); + break; + default: + /* + * With several "complicated" options set, this is the most + * general and the slowest copying of an RTL run. + * We will do mirroring, remove BiDi controls, and + * keep combining characters with their base characters + * as requested. + */ + if(!(options&UBIDI_REMOVE_BIDI_CONTROLS)) { + i=srcLength; + } else { + /* we need to find out the destination length of the run, + which will not include the BiDi control characters */ + int32_t length=srcLength; + char16_t ch; + + i=0; + do { + ch=*src++; + if(!IS_BIDI_CONTROL_CHAR(ch)) { + ++i; + } + } while(--length>0); + src-=srcLength; + } + + if(destSize0 && IS_COMBINING(u_charType(c))) { + U16_PREV(src, 0, srcLength, c); + } + } + + if(options&UBIDI_REMOVE_BIDI_CONTROLS && IS_BIDI_CONTROL_CHAR(c)) { + /* do not copy this BiDi control character */ + continue; + } + + /* copy this "user character" */ + j=srcLength; + if(options&UBIDI_DO_MIRRORING) { + /* mirror only the base character */ + int32_t k=0; + c=u_charMirror(c); + U16_APPEND_UNSAFE(dest, k, c); + dest+=k; + j+=k; + } + while(j0); + break; + } /* end of switch */ + + return destSize; +} + +U_CAPI int32_t U_EXPORT2 +ubidi_writeReverse(const char16_t *src, int32_t srcLength, + char16_t *dest, int32_t destSize, + uint16_t options, + UErrorCode *pErrorCode) { + int32_t destLength; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + /* more error checking */ + if( src==nullptr || srcLength<-1 || + destSize<0 || (destSize>0 && dest==nullptr)) + { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* do input and output overlap? */ + if( dest!=nullptr && + ((src>=dest && src=src && dest0) { + destLength=doWriteReverse(src, srcLength, dest, destSize, options, pErrorCode); + } else { + /* nothing to do */ + destLength=0; + } + + return u_terminateUChars(dest, destSize, destLength, pErrorCode); +} + +// Ticket 20907 - The optimizer in MSVC/Visual Studio versions below 16.4 has trouble with this +// function on Windows ARM64. As a work-around, we disable optimizations for this function. +// This work-around could/should be removed once the following versions of Visual Studio are no +// longer supported: All versions of VS2017, and versions of VS2019 below 16.4. +#if (defined(_MSC_VER) && (defined(_M_ARM64)) && (_MSC_VER < 1924)) +#pragma optimize( "", off ) +#endif +U_CAPI int32_t U_EXPORT2 +ubidi_writeReordered(UBiDi *pBiDi, + char16_t *dest, int32_t destSize, + uint16_t options, + UErrorCode *pErrorCode) { + const char16_t *text; + char16_t *saveDest; + int32_t length, destCapacity; + int32_t run, runCount, logicalStart, runLength; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + /* more error checking */ + if( pBiDi==nullptr || + (text=pBiDi->text)==nullptr || (length=pBiDi->length)<0 || + destSize<0 || (destSize>0 && dest==nullptr)) + { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* do input and output overlap? */ + if( dest!=nullptr && + ((text>=dest && text=text && destoriginalLength))) + { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + if(length==0) { + /* nothing to do */ + return u_terminateUChars(dest, destSize, 0, pErrorCode); + } + + runCount=ubidi_countRuns(pBiDi, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + return 0; + } + + /* destSize shrinks, later destination length=destCapacity-destSize */ + saveDest=dest; + destCapacity=destSize; + + /* + * Option "insert marks" implies UBIDI_INSERT_LRM_FOR_NUMERIC if the + * reordering mode (checked below) is appropriate. + */ + if(pBiDi->reorderingOptions & UBIDI_OPTION_INSERT_MARKS) { + options|=UBIDI_INSERT_LRM_FOR_NUMERIC; + options&=~UBIDI_REMOVE_BIDI_CONTROLS; + } + /* + * Option "remove controls" implies UBIDI_REMOVE_BIDI_CONTROLS + * and cancels UBIDI_INSERT_LRM_FOR_NUMERIC. + */ + if(pBiDi->reorderingOptions & UBIDI_OPTION_REMOVE_CONTROLS) { + options|=UBIDI_REMOVE_BIDI_CONTROLS; + options&=~UBIDI_INSERT_LRM_FOR_NUMERIC; + } + /* + * If we do not perform the "inverse BiDi" algorithm, then we + * don't need to insert any LRMs, and don't need to test for it. + */ + if((pBiDi->reorderingMode != UBIDI_REORDER_INVERSE_NUMBERS_AS_L) && + (pBiDi->reorderingMode != UBIDI_REORDER_INVERSE_LIKE_DIRECT) && + (pBiDi->reorderingMode != UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL) && + (pBiDi->reorderingMode != UBIDI_REORDER_RUNS_ONLY)) { + options&=~UBIDI_INSERT_LRM_FOR_NUMERIC; + } + /* + * Iterate through all visual runs and copy the run text segments to + * the destination, according to the options. + * + * The tests for where to insert LRMs ignore the fact that there may be + * BN codes or non-BMP code points at the beginning and end of a run; + * they may insert LRMs unnecessarily but the tests are faster this way + * (this would have to be improved for UTF-8). + * + * Note that the only errors that are set by doWriteXY() are buffer overflow + * errors. Ignore them until the end, and continue for preflighting. + */ + if(!(options&UBIDI_OUTPUT_REVERSE)) { + /* forward output */ + if(!(options&UBIDI_INSERT_LRM_FOR_NUMERIC)) { + /* do not insert BiDi controls */ + for(run=0; rundirProps; + const char16_t *src; + char16_t uc; + UBiDiDirection dir; + int32_t markFlag; + + for(run=0; runruns[run].insertRemove; + if(markFlag<0) { /* BiDi controls count */ + markFlag=0; + } + + if(UBIDI_LTR==dir) { + if((pBiDi->isInverse) && + (/*run>0 &&*/ dirProps[logicalStart]!=L)) { + markFlag |= LRM_BEFORE; + } + if (markFlag & LRM_BEFORE) { + uc=LRM_CHAR; + } + else if (markFlag & RLM_BEFORE) { + uc=RLM_CHAR; + } + else uc=0; + if(uc) { + if(destSize>0) { + *dest++=uc; + } + --destSize; + } + + runLength=doWriteForward(src, runLength, + dest, destSize, + (uint16_t)(options&~UBIDI_DO_MIRRORING), pErrorCode); + if(dest!=nullptr) { + dest+=runLength; + } + destSize-=runLength; + + if((pBiDi->isInverse) && + (/*run0) { + *dest++=uc; + } + --destSize; + } + } else { /* RTL run */ + if((pBiDi->isInverse) && + (/*run>0 &&*/ !(MASK_R_AL&DIRPROP_FLAG(dirProps[logicalStart+runLength-1])))) { + markFlag |= RLM_BEFORE; + } + if (markFlag & LRM_BEFORE) { + uc=LRM_CHAR; + } + else if (markFlag & RLM_BEFORE) { + uc=RLM_CHAR; + } + else uc=0; + if(uc) { + if(destSize>0) { + *dest++=uc; + } + --destSize; + } + + runLength=doWriteReverse(src, runLength, + dest, destSize, + options, pErrorCode); + if(dest!=nullptr) { + dest+=runLength; + } + destSize-=runLength; + + if((pBiDi->isInverse) && + (/*run0) { + *dest++=uc; + } + --destSize; + } + } + } + } + } else { + /* reverse output */ + if(!(options&UBIDI_INSERT_LRM_FOR_NUMERIC)) { + /* do not insert BiDi controls */ + for(run=runCount; --run>=0;) { + if(UBIDI_LTR==ubidi_getVisualRun(pBiDi, run, &logicalStart, &runLength)) { + runLength=doWriteReverse(text+logicalStart, runLength, + dest, destSize, + (uint16_t)(options&~UBIDI_DO_MIRRORING), pErrorCode); + } else { + runLength=doWriteForward(text+logicalStart, runLength, + dest, destSize, + options, pErrorCode); + } + if(dest!=nullptr) { + dest+=runLength; + } + destSize-=runLength; + } + } else { + /* insert BiDi controls for "inverse BiDi" */ + const DirProp *dirProps=pBiDi->dirProps; + const char16_t *src; + UBiDiDirection dir; + + for(run=runCount; --run>=0;) { + /* reverse output */ + dir=ubidi_getVisualRun(pBiDi, run, &logicalStart, &runLength); + src=text+logicalStart; + + if(UBIDI_LTR==dir) { + if(/*run0) { + *dest++=LRM_CHAR; + } + --destSize; + } + + runLength=doWriteReverse(src, runLength, + dest, destSize, + (uint16_t)(options&~UBIDI_DO_MIRRORING), pErrorCode); + if(dest!=nullptr) { + dest+=runLength; + } + destSize-=runLength; + + if(/*run>0 &&*/ dirProps[logicalStart]!=L) { + if(destSize>0) { + *dest++=LRM_CHAR; + } + --destSize; + } + } else { + if(/*run0) { + *dest++=RLM_CHAR; + } + --destSize; + } + + runLength=doWriteForward(src, runLength, + dest, destSize, + options, pErrorCode); + if(dest!=nullptr) { + dest+=runLength; + } + destSize-=runLength; + + if(/*run>0 &&*/ !(MASK_R_AL&DIRPROP_FLAG(dirProps[logicalStart+runLength-1]))) { + if(destSize>0) { + *dest++=RLM_CHAR; + } + --destSize; + } + } + } + } + } + + return u_terminateUChars(saveDest, destCapacity, destCapacity-destSize, pErrorCode); +} +#if (defined(_MSC_VER) && (defined(_M_ARM64)) && (_MSC_VER < 1924)) +#pragma optimize( "", on ) +#endif diff --git a/intl/icu/source/common/ubrk.cpp b/intl/icu/source/common/ubrk.cpp new file mode 100644 index 0000000000..46ec0d785a --- /dev/null +++ b/intl/icu/source/common/ubrk.cpp @@ -0,0 +1,361 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************** +* Copyright (C) 1996-2015, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************** +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/ubrk.h" + +#include "unicode/brkiter.h" +#include "unicode/uloc.h" +#include "unicode/ustring.h" +#include "unicode/uchriter.h" +#include "unicode/rbbi.h" +#include "rbbirb.h" +#include "uassert.h" +#include "cmemory.h" + +U_NAMESPACE_USE + +//------------------------------------------------------------------------------ +// +// ubrk_open Create a canned type of break iterator based on type (word, line, etc.) +// and locale. +// +//------------------------------------------------------------------------------ +U_CAPI UBreakIterator* U_EXPORT2 +ubrk_open(UBreakIteratorType type, + const char *locale, + const char16_t *text, + int32_t textLength, + UErrorCode *status) +{ + + if(U_FAILURE(*status)) return 0; + + BreakIterator *result = 0; + + switch(type) { + + case UBRK_CHARACTER: + result = BreakIterator::createCharacterInstance(Locale(locale), *status); + break; + + case UBRK_WORD: + result = BreakIterator::createWordInstance(Locale(locale), *status); + break; + + case UBRK_LINE: + result = BreakIterator::createLineInstance(Locale(locale), *status); + break; + + case UBRK_SENTENCE: + result = BreakIterator::createSentenceInstance(Locale(locale), *status); + break; + + case UBRK_TITLE: + result = BreakIterator::createTitleInstance(Locale(locale), *status); + break; + + default: + *status = U_ILLEGAL_ARGUMENT_ERROR; + } + + // check for allocation error + if (U_FAILURE(*status)) { + return 0; + } + if(result == 0) { + *status = U_MEMORY_ALLOCATION_ERROR; + return 0; + } + + + UBreakIterator *uBI = (UBreakIterator *)result; + if (text != nullptr) { + ubrk_setText(uBI, text, textLength, status); + } + return uBI; +} + + + +//------------------------------------------------------------------------------ +// +// ubrk_openRules open a break iterator from a set of break rules. +// Invokes the rule builder. +// +//------------------------------------------------------------------------------ +U_CAPI UBreakIterator* U_EXPORT2 +ubrk_openRules( const char16_t *rules, + int32_t rulesLength, + const char16_t *text, + int32_t textLength, + UParseError *parseErr, + UErrorCode *status) { + + if (status == nullptr || U_FAILURE(*status)){ + return 0; + } + + BreakIterator *result = 0; + UnicodeString ruleString(rules, rulesLength); + result = RBBIRuleBuilder::createRuleBasedBreakIterator(ruleString, parseErr, *status); + if(U_FAILURE(*status)) { + return 0; + } + + UBreakIterator *uBI = (UBreakIterator *)result; + if (text != nullptr) { + ubrk_setText(uBI, text, textLength, status); + } + return uBI; +} + + +U_CAPI UBreakIterator* U_EXPORT2 +ubrk_openBinaryRules(const uint8_t *binaryRules, int32_t rulesLength, + const char16_t * text, int32_t textLength, + UErrorCode * status) +{ + if (U_FAILURE(*status)) { + return nullptr; + } + if (rulesLength < 0) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + LocalPointer lpRBBI(new RuleBasedBreakIterator(binaryRules, rulesLength, *status), *status); + if (U_FAILURE(*status)) { + return nullptr; + } + UBreakIterator *uBI = reinterpret_cast(lpRBBI.orphan()); + if (text != nullptr) { + ubrk_setText(uBI, text, textLength, status); + } + return uBI; +} + + +U_CAPI UBreakIterator * U_EXPORT2 +ubrk_safeClone( + const UBreakIterator *bi, + void * /*stackBuffer*/, + int32_t *pBufferSize, + UErrorCode *status) +{ + if (status == nullptr || U_FAILURE(*status)){ + return nullptr; + } + if (bi == nullptr) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + if (pBufferSize != nullptr) { + int32_t inputSize = *pBufferSize; + *pBufferSize = 1; + if (inputSize == 0) { + return nullptr; // preflighting for deprecated functionality + } + } + BreakIterator *newBI = ((BreakIterator *)bi)->clone(); + if (newBI == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + } else if (pBufferSize != nullptr) { + *status = U_SAFECLONE_ALLOCATED_WARNING; + } + return (UBreakIterator *)newBI; +} + +U_CAPI UBreakIterator * U_EXPORT2 +ubrk_clone(const UBreakIterator *bi, UErrorCode *status) { + return ubrk_safeClone(bi, nullptr, nullptr, status); +} + + +U_CAPI void U_EXPORT2 +ubrk_close(UBreakIterator *bi) +{ + delete (BreakIterator *)bi; +} + +U_CAPI void U_EXPORT2 +ubrk_setText(UBreakIterator* bi, + const char16_t* text, + int32_t textLength, + UErrorCode* status) +{ + UText ut = UTEXT_INITIALIZER; + utext_openUChars(&ut, text, textLength, status); + ((BreakIterator*)bi)->setText(&ut, *status); + // A stack allocated UText wrapping a char16_t * string + // can be dumped without explicitly closing it. +} + + + +U_CAPI void U_EXPORT2 +ubrk_setUText(UBreakIterator *bi, + UText *text, + UErrorCode *status) +{ + ((BreakIterator*)bi)->setText(text, *status); +} + + + + + +U_CAPI int32_t U_EXPORT2 +ubrk_current(const UBreakIterator *bi) +{ + + return ((BreakIterator*)bi)->current(); +} + +U_CAPI int32_t U_EXPORT2 +ubrk_next(UBreakIterator *bi) +{ + + return ((BreakIterator*)bi)->next(); +} + +U_CAPI int32_t U_EXPORT2 +ubrk_previous(UBreakIterator *bi) +{ + + return ((BreakIterator*)bi)->previous(); +} + +U_CAPI int32_t U_EXPORT2 +ubrk_first(UBreakIterator *bi) +{ + + return ((BreakIterator*)bi)->first(); +} + +U_CAPI int32_t U_EXPORT2 +ubrk_last(UBreakIterator *bi) +{ + + return ((BreakIterator*)bi)->last(); +} + +U_CAPI int32_t U_EXPORT2 +ubrk_preceding(UBreakIterator *bi, + int32_t offset) +{ + + return ((BreakIterator*)bi)->preceding(offset); +} + +U_CAPI int32_t U_EXPORT2 +ubrk_following(UBreakIterator *bi, + int32_t offset) +{ + + return ((BreakIterator*)bi)->following(offset); +} + +U_CAPI const char* U_EXPORT2 +ubrk_getAvailable(int32_t index) +{ + + return uloc_getAvailable(index); +} + +U_CAPI int32_t U_EXPORT2 +ubrk_countAvailable() +{ + + return uloc_countAvailable(); +} + + +U_CAPI UBool U_EXPORT2 +ubrk_isBoundary(UBreakIterator *bi, int32_t offset) +{ + return ((BreakIterator*)bi)->isBoundary(offset); +} + + +U_CAPI int32_t U_EXPORT2 +ubrk_getRuleStatus(UBreakIterator *bi) +{ + return ((BreakIterator*)bi)->getRuleStatus(); +} + +U_CAPI int32_t U_EXPORT2 +ubrk_getRuleStatusVec(UBreakIterator *bi, int32_t *fillInVec, int32_t capacity, UErrorCode *status) +{ + return ((BreakIterator*)bi)->getRuleStatusVec(fillInVec, capacity, *status); +} + + +U_CAPI const char* U_EXPORT2 +ubrk_getLocaleByType(const UBreakIterator *bi, + ULocDataLocaleType type, + UErrorCode* status) +{ + if (bi == nullptr) { + if (U_SUCCESS(*status)) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + } + return nullptr; + } + return ((BreakIterator*)bi)->getLocaleID(type, *status); +} + + +U_CAPI void U_EXPORT2 +ubrk_refreshUText(UBreakIterator *bi, + UText *text, + UErrorCode *status) +{ + BreakIterator *bii = reinterpret_cast(bi); + bii->refreshInputText(text, *status); +} + +U_CAPI int32_t U_EXPORT2 +ubrk_getBinaryRules(UBreakIterator *bi, + uint8_t * binaryRules, int32_t rulesCapacity, + UErrorCode * status) +{ + if (U_FAILURE(*status)) { + return 0; + } + if ((binaryRules == nullptr && rulesCapacity > 0) || rulesCapacity < 0) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + RuleBasedBreakIterator* rbbi; + if ((rbbi = dynamic_cast(reinterpret_cast(bi))) == nullptr) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + uint32_t rulesLength; + const uint8_t * returnedRules = rbbi->getBinaryRules(rulesLength); + if (rulesLength > INT32_MAX) { + *status = U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + if (binaryRules != nullptr) { // if not preflighting + // Here we know rulesLength <= INT32_MAX and rulesCapacity >= 0, can cast safely + if ((int32_t)rulesLength > rulesCapacity) { + *status = U_BUFFER_OVERFLOW_ERROR; + } else { + uprv_memcpy(binaryRules, returnedRules, rulesLength); + } + } + return (int32_t)rulesLength; +} + + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ diff --git a/intl/icu/source/common/ubrkimpl.h b/intl/icu/source/common/ubrkimpl.h new file mode 100644 index 0000000000..8197f66339 --- /dev/null +++ b/intl/icu/source/common/ubrkimpl.h @@ -0,0 +1,15 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2006, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ + +#ifndef UBRKIMPL_H +#define UBRKIMPL_H + +#define U_ICUDATA_BRKITR U_ICUDATA_NAME U_TREE_SEPARATOR_STRING "brkitr" + +#endif /*UBRKIMPL_H*/ diff --git a/intl/icu/source/common/ucase.cpp b/intl/icu/source/common/ucase.cpp new file mode 100644 index 0000000000..392e1266ae --- /dev/null +++ b/intl/icu/source/common/ucase.cpp @@ -0,0 +1,1739 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2004-2014, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: ucase.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2004aug30 +* created by: Markus W. Scherer +* +* Low-level Unicode character/string case mapping code. +* Much code moved here (and modified) from uchar.c. +*/ + +#include "unicode/utypes.h" +#include "unicode/unistr.h" +#include "unicode/uset.h" +#include "unicode/utf16.h" +#include "cmemory.h" +#include "uassert.h" +#include "ucase.h" +#include "umutex.h" +#include "utrie2.h" + +/* ucase_props_data.h is machine-generated by genprops/casepropsbuilder.cpp */ +#define INCLUDED_FROM_UCASE_CPP +#include "ucase_props_data.h" + +/* set of property starts for UnicodeSet ------------------------------------ */ + +static UBool U_CALLCONV +_enumPropertyStartsRange(const void *context, UChar32 start, UChar32 /*end*/, uint32_t /*value*/) { + /* add the start code point to the USet */ + const USetAdder *sa=(const USetAdder *)context; + sa->add(sa->set, start); + return true; +} + +U_CFUNC void U_EXPORT2 +ucase_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { + return; + } + + /* add the start code point of each same-value range of the trie */ + utrie2_enum(&ucase_props_singleton.trie, nullptr, _enumPropertyStartsRange, sa); + + /* add code points with hardcoded properties, plus the ones following them */ + + /* (none right now, see comment below) */ + + /* + * Omit code points with hardcoded specialcasing properties + * because we do not build property UnicodeSets for them right now. + */ +} + +/* data access primitives --------------------------------------------------- */ + +U_CAPI const struct UCaseProps * U_EXPORT2 +ucase_getSingleton(int32_t *pExceptionsLength, int32_t *pUnfoldLength) { + *pExceptionsLength = UPRV_LENGTHOF(ucase_props_exceptions); + *pUnfoldLength = UPRV_LENGTHOF(ucase_props_unfold); + return &ucase_props_singleton; +} + +U_CFUNC const UTrie2 * U_EXPORT2 +ucase_getTrie() { + return &ucase_props_singleton.trie; +} + +#define GET_EXCEPTIONS(csp, props) ((csp)->exceptions+((props)>>UCASE_EXC_SHIFT)) + +/* number of bits in an 8-bit integer value */ +static const uint8_t flagsOffset[256]={ + 0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4, + 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, + 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, + 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, + 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, + 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, + 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, + 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, + 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, + 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, + 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, + 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, + 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, + 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, + 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, + 4, 5, 5, 6, 5, 6, 6, 7, 5, 6, 6, 7, 6, 7, 7, 8 +}; + +#define HAS_SLOT(flags, idx) ((flags)&(1<<(idx))) +#define SLOT_OFFSET(flags, idx) flagsOffset[(flags)&((1<<(idx))-1)] + +/* + * Get the value of an optional-value slot where HAS_SLOT(excWord, idx). + * + * @param excWord (in) initial exceptions word + * @param idx (in) desired slot index + * @param pExc16 (in/out) const uint16_t * after excWord=*pExc16++; + * moved to the last uint16_t of the value, use +1 for beginning of next slot + * @param value (out) int32_t or uint32_t output if hasSlot, otherwise not modified + */ +#define GET_SLOT_VALUE(excWord, idx, pExc16, value) UPRV_BLOCK_MACRO_BEGIN { \ + if(((excWord)&UCASE_EXC_DOUBLE_SLOTS)==0) { \ + (pExc16)+=SLOT_OFFSET(excWord, idx); \ + (value)=*pExc16; \ + } else { \ + (pExc16)+=2*SLOT_OFFSET(excWord, idx); \ + (value)=*pExc16++; \ + (value)=((value)<<16)|*pExc16; \ + } \ +} UPRV_BLOCK_MACRO_END + +/* simple case mappings ----------------------------------------------------- */ + +U_CAPI UChar32 U_EXPORT2 +ucase_tolower(UChar32 c) { + uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); + if(!UCASE_HAS_EXCEPTION(props)) { + if(UCASE_IS_UPPER_OR_TITLE(props)) { + c+=UCASE_GET_DELTA(props); + } + } else { + const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); + uint16_t excWord=*pe++; + if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_IS_UPPER_OR_TITLE(props)) { + int32_t delta; + GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe, delta); + return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; + } + if(HAS_SLOT(excWord, UCASE_EXC_LOWER)) { + GET_SLOT_VALUE(excWord, UCASE_EXC_LOWER, pe, c); + } + } + return c; +} + +U_CAPI UChar32 U_EXPORT2 +ucase_toupper(UChar32 c) { + uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); + if(!UCASE_HAS_EXCEPTION(props)) { + if(UCASE_GET_TYPE(props)==UCASE_LOWER) { + c+=UCASE_GET_DELTA(props); + } + } else { + const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); + uint16_t excWord=*pe++; + if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_GET_TYPE(props)==UCASE_LOWER) { + int32_t delta; + GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe, delta); + return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; + } + if(HAS_SLOT(excWord, UCASE_EXC_UPPER)) { + GET_SLOT_VALUE(excWord, UCASE_EXC_UPPER, pe, c); + } + } + return c; +} + +U_CAPI UChar32 U_EXPORT2 +ucase_totitle(UChar32 c) { + uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); + if(!UCASE_HAS_EXCEPTION(props)) { + if(UCASE_GET_TYPE(props)==UCASE_LOWER) { + c+=UCASE_GET_DELTA(props); + } + } else { + const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); + uint16_t excWord=*pe++; + if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_GET_TYPE(props)==UCASE_LOWER) { + int32_t delta; + GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe, delta); + return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; + } + int32_t idx; + if(HAS_SLOT(excWord, UCASE_EXC_TITLE)) { + idx=UCASE_EXC_TITLE; + } else if(HAS_SLOT(excWord, UCASE_EXC_UPPER)) { + idx=UCASE_EXC_UPPER; + } else { + return c; + } + GET_SLOT_VALUE(excWord, idx, pe, c); + } + return c; +} + +static const char16_t iDot[2] = { 0x69, 0x307 }; +static const char16_t jDot[2] = { 0x6a, 0x307 }; +static const char16_t iOgonekDot[3] = { 0x12f, 0x307 }; +static const char16_t iDotGrave[3] = { 0x69, 0x307, 0x300 }; +static const char16_t iDotAcute[3] = { 0x69, 0x307, 0x301 }; +static const char16_t iDotTilde[3] = { 0x69, 0x307, 0x303 }; + + +U_CFUNC void U_EXPORT2 +ucase_addCaseClosure(UChar32 c, const USetAdder *sa) { + uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); + if(!UCASE_HAS_EXCEPTION(props)) { + if(UCASE_GET_TYPE(props)!=UCASE_NONE) { + /* add the one simple case mapping, no matter what type it is */ + int32_t delta=UCASE_GET_DELTA(props); + if(delta!=0) { + sa->add(sa->set, c+delta); + } + } + } else { + /* + * c has exceptions, so there may be multiple simple and/or + * full case mappings. Add them all. + */ + const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); + uint16_t excWord=*pe++; + const uint16_t *pe0=pe; + + // Hardcode the case closure of i and its relatives and ignore the + // data file data for these characters. + // The Turkic dotless i and dotted I with their case mapping conditions + // and case folding option make the related characters behave specially. + // This code matches their closure behavior to their case folding behavior. + if (excWord&UCASE_EXC_CONDITIONAL_FOLD) { + // These characters have Turkic case foldings. Hardcode their closure. + if (c == 0x49) { + // Regular i and I are in one equivalence class. + sa->add(sa->set, 0x69); + return; + } else if (c == 0x130) { + // Dotted I is in a class with <0069 0307> + // (for canonical equivalence with <0049 0307>). + sa->addString(sa->set, iDot, 2); + return; + } + } else if (c == 0x69) { + sa->add(sa->set, 0x49); + return; + } else if (c == 0x131) { + // Dotless i is in a class by itself. + return; + } + + /* add all simple case mappings */ + for(int32_t idx=UCASE_EXC_LOWER; idx<=UCASE_EXC_TITLE; ++idx) { + if(HAS_SLOT(excWord, idx)) { + pe=pe0; + UChar32 mapping; + GET_SLOT_VALUE(excWord, idx, pe, mapping); + sa->add(sa->set, mapping); + } + } + if(HAS_SLOT(excWord, UCASE_EXC_DELTA)) { + pe=pe0; + int32_t delta; + GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe, delta); + sa->add(sa->set, (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta); + } + + /* get the closure string pointer & length */ + const char16_t *closure; + int32_t closureLength; + if(HAS_SLOT(excWord, UCASE_EXC_CLOSURE)) { + pe=pe0; + GET_SLOT_VALUE(excWord, UCASE_EXC_CLOSURE, pe, closureLength); + closureLength&=UCASE_CLOSURE_MAX_LENGTH; /* higher bits are reserved */ + closure=(const char16_t *)pe+1; /* behind this slot, unless there are full case mappings */ + } else { + closureLength=0; + closure=nullptr; + } + + /* add the full case folding */ + if(HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { + pe=pe0; + int32_t fullLength; + GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, fullLength); + + /* start of full case mapping strings */ + ++pe; + + fullLength&=0xffff; /* bits 16 and higher are reserved */ + + /* skip the lowercase result string */ + pe+=fullLength&UCASE_FULL_LOWER; + fullLength>>=4; + + /* add the full case folding string */ + int32_t length=fullLength&0xf; + if(length!=0) { + sa->addString(sa->set, (const char16_t *)pe, length); + pe+=length; + } + + /* skip the uppercase and titlecase strings */ + fullLength>>=4; + pe+=fullLength&0xf; + fullLength>>=4; + pe+=fullLength; + + closure=(const char16_t *)pe; /* behind full case mappings */ + } + + /* add each code point in the closure string */ + for(int32_t idx=0; idxadd(sa->set, mapping); + } + } +} + +namespace { + +/** + * Add the simple case closure mapping, + * except if there is not actually an scf relationship between the two characters. + * TODO: Unicode should probably add the corresponding scf mappings. + * See https://crbug.com/v8/13377 and Unicode-internal PAG issue #23. + * If & when those scf mappings are added, we should be able to remove all of these exceptions. + */ +void addOneSimpleCaseClosure(UChar32 c, UChar32 t, const USetAdder *sa) { + switch (c) { + case 0x0390: + if (t == 0x1FD3) { return; } + break; + case 0x03B0: + if (t == 0x1FE3) { return; } + break; + case 0x1FD3: + if (t == 0x0390) { return; } + break; + case 0x1FE3: + if (t == 0x03B0) { return; } + break; + case 0xFB05: + if (t == 0xFB06) { return; } + break; + case 0xFB06: + if (t == 0xFB05) { return; } + break; + default: + break; + } + sa->add(sa->set, t); +} + +} // namespace + +U_CFUNC void U_EXPORT2 +ucase_addSimpleCaseClosure(UChar32 c, const USetAdder *sa) { + uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); + if(!UCASE_HAS_EXCEPTION(props)) { + if(UCASE_GET_TYPE(props)!=UCASE_NONE) { + /* add the one simple case mapping, no matter what type it is */ + int32_t delta=UCASE_GET_DELTA(props); + if(delta!=0) { + sa->add(sa->set, c+delta); + } + } + } else { + // c has exceptions. Add the mappings relevant for scf=Simple_Case_Folding. + const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); + uint16_t excWord=*pe++; + const uint16_t *pe0=pe; + + // Hardcode the case closure of i and its relatives and ignore the + // data file data for these characters, like in ucase_addCaseClosure(). + if (excWord&UCASE_EXC_CONDITIONAL_FOLD) { + // These characters have Turkic case foldings. Hardcode their closure. + if (c == 0x49) { + // Regular i and I are in one equivalence class. + sa->add(sa->set, 0x69); + return; + } else if (c == 0x130) { + // For scf=Simple_Case_Folding, dotted I is in a class by itself. + return; + } + } else if (c == 0x69) { + sa->add(sa->set, 0x49); + return; + } else if (c == 0x131) { + // Dotless i is in a class by itself. + return; + } + + // Add all simple case mappings. + for(int32_t idx=UCASE_EXC_LOWER; idx<=UCASE_EXC_TITLE; ++idx) { + if(HAS_SLOT(excWord, idx)) { + pe=pe0; + UChar32 mapping; + GET_SLOT_VALUE(excWord, idx, pe, mapping); + addOneSimpleCaseClosure(c, mapping, sa); + } + } + if(HAS_SLOT(excWord, UCASE_EXC_DELTA)) { + pe=pe0; + int32_t delta; + GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe, delta); + UChar32 mapping = (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; + addOneSimpleCaseClosure(c, mapping, sa); + } + + /* get the closure string pointer & length */ + const char16_t *closure; + int32_t closureLength; + if(HAS_SLOT(excWord, UCASE_EXC_CLOSURE)) { + pe=pe0; + GET_SLOT_VALUE(excWord, UCASE_EXC_CLOSURE, pe, closureLength); + closureLength&=UCASE_CLOSURE_MAX_LENGTH; /* higher bits are reserved */ + closure=(const char16_t *)pe+1; /* behind this slot, unless there are full case mappings */ + } else { + closureLength=0; + closure=nullptr; + } + + // Skip the full case mappings. + if(closureLength > 0 && HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { + pe=pe0; + int32_t fullLength; + GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, fullLength); + + /* start of full case mapping strings */ + ++pe; + + fullLength&=0xffff; /* bits 16 and higher are reserved */ + + // Skip all 4 full case mappings. + pe+=fullLength&UCASE_FULL_LOWER; + fullLength>>=4; + pe+=fullLength&0xf; + fullLength>>=4; + pe+=fullLength&0xf; + fullLength>>=4; + pe+=fullLength; + + closure=(const char16_t *)pe; /* behind full case mappings */ + } + + // Add each code point in the closure string whose scf maps back to c. + for(int32_t idx=0; idx0 and max>0 and length<=max + */ +static inline int32_t +strcmpMax(const char16_t *s, int32_t length, const char16_t *t, int32_t max) { + int32_t c1, c2; + + max-=length; /* we require length<=max, so no need to decrement max in the loop */ + do { + c1=*s++; + c2=*t++; + if(c2==0) { + return 1; /* reached the end of t but not of s */ + } + c1-=c2; + if(c1!=0) { + return c1; /* return difference result */ + } + } while(--length>0); + /* ends with length==0 */ + + if(max==0 || *t==0) { + return 0; /* equal to length of both strings */ + } else { + return -max; /* return length difference */ + } +} + +U_CFUNC UBool U_EXPORT2 +ucase_addStringCaseClosure(const char16_t *s, int32_t length, const USetAdder *sa) { + int32_t i, start, limit, result, unfoldRows, unfoldRowWidth, unfoldStringWidth; + + if(ucase_props_singleton.unfold==nullptr || s==nullptr) { + return false; /* no reverse case folding data, or no string */ + } + if(length<=1) { + /* the string is too short to find any match */ + /* + * more precise would be: + * if(!u_strHasMoreChar32Than(s, length, 1)) + * but this does not make much practical difference because + * a single supplementary code point would just not be found + */ + return false; + } + + const uint16_t *unfold=ucase_props_singleton.unfold; + unfoldRows=unfold[UCASE_UNFOLD_ROWS]; + unfoldRowWidth=unfold[UCASE_UNFOLD_ROW_WIDTH]; + unfoldStringWidth=unfold[UCASE_UNFOLD_STRING_WIDTH]; + unfold+=unfoldRowWidth; + + if(length>unfoldStringWidth) { + /* the string is too long to find any match */ + return false; + } + + /* do a binary search for the string */ + start=0; + limit=unfoldRows; + while(start(unfold+(i*unfoldRowWidth)); + result=strcmpMax(s, length, p, unfoldStringWidth); + + if(result==0) { + /* found the string: add each code point, and its case closure */ + UChar32 c; + + for(i=unfoldStringWidth; iadd(sa->set, c); + ucase_addCaseClosure(c, sa); + } + return true; + } else if(result<0) { + limit=i; + } else /* result>0 */ { + start=i+1; + } + } + + return false; /* string not found */ +} + +U_NAMESPACE_BEGIN + +FullCaseFoldingIterator::FullCaseFoldingIterator() + : unfold(reinterpret_cast(ucase_props_singleton.unfold)), + unfoldRows(unfold[UCASE_UNFOLD_ROWS]), + unfoldRowWidth(unfold[UCASE_UNFOLD_ROW_WIDTH]), + unfoldStringWidth(unfold[UCASE_UNFOLD_STRING_WIDTH]), + currentRow(0), + rowCpIndex(unfoldStringWidth) { + unfold+=unfoldRowWidth; +} + +UChar32 +FullCaseFoldingIterator::next(UnicodeString &full) { + // Advance past the last-delivered code point. + const char16_t *p=unfold+(currentRow*unfoldRowWidth); + if(rowCpIndex>=unfoldRowWidth || p[rowCpIndex]==0) { + ++currentRow; + p+=unfoldRowWidth; + rowCpIndex=unfoldStringWidth; + } + if(currentRow>=unfoldRows) { return U_SENTINEL; } + // Set "full" to the NUL-terminated string in the first unfold column. + int32_t length=unfoldStringWidth; + while(length>0 && p[length-1]==0) { --length; } + full.setTo(false, p, length); + // Return the code point. + UChar32 c; + U16_NEXT_UNSAFE(p, rowCpIndex, c); + return c; +} + +namespace LatinCase { + +const int8_t TO_LOWER_NORMAL[LIMIT] = { + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, + 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, EXC, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, + 32, 32, 32, 32, 32, 32, 32, 0, 32, 32, 32, 32, 32, 32, 32, EXC, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, + 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, + 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, + EXC, 0, 1, 0, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 1, + + 0, 1, 0, 1, 0, 1, 0, 1, 0, EXC, 1, 0, 1, 0, 1, 0, + 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, + 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, + 1, 0, 1, 0, 1, 0, 1, 0, -121, 1, 0, 1, 0, 1, 0, EXC +}; + +const int8_t TO_LOWER_TR_LT[LIMIT] = { + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 32, 32, 32, 32, 32, 32, 32, 32, EXC, EXC, 32, 32, 32, 32, 32, + 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, EXC, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, EXC, EXC, 32, 32, + 32, 32, 32, 32, 32, 32, 32, 0, 32, 32, 32, 32, 32, 32, 32, EXC, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, + 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, + 1, 0, 1, 0, 1, 0, 1, 0, EXC, 0, 1, 0, 1, 0, EXC, 0, + EXC, 0, 1, 0, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 1, + + 0, 1, 0, 1, 0, 1, 0, 1, 0, EXC, 1, 0, 1, 0, 1, 0, + 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, + 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, + 1, 0, 1, 0, 1, 0, 1, 0, -121, 1, 0, 1, 0, 1, 0, EXC +}; + +const int8_t TO_UPPER_NORMAL[LIMIT] = { + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, + -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, EXC, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, EXC, + -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, + -32, -32, -32, -32, -32, -32, -32, 0, -32, -32, -32, -32, -32, -32, -32, 121, + + 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, + 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, + 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, + 0, EXC, 0, -1, 0, -1, 0, -1, 0, 0, -1, 0, -1, 0, -1, 0, + + -1, 0, -1, 0, -1, 0, -1, 0, -1, EXC, 0, -1, 0, -1, 0, -1, + 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, + 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, + 0, -1, 0, -1, 0, -1, 0, -1, 0, 0, -1, 0, -1, 0, -1, EXC +}; + +const int8_t TO_UPPER_TR[LIMIT] = { + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, -32, -32, -32, -32, -32, -32, -32, -32, EXC, -32, -32, -32, -32, -32, -32, + -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, EXC, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, EXC, + -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, + -32, -32, -32, -32, -32, -32, -32, 0, -32, -32, -32, -32, -32, -32, -32, 121, + + 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, + 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, + 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, + 0, EXC, 0, -1, 0, -1, 0, -1, 0, 0, -1, 0, -1, 0, -1, 0, + + -1, 0, -1, 0, -1, 0, -1, 0, -1, EXC, 0, -1, 0, -1, 0, -1, + 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, + 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, + 0, -1, 0, -1, 0, -1, 0, -1, 0, 0, -1, 0, -1, 0, -1, EXC +}; + +} // namespace LatinCase + +U_NAMESPACE_END + +/** @return UCASE_NONE, UCASE_LOWER, UCASE_UPPER, UCASE_TITLE */ +U_CAPI int32_t U_EXPORT2 +ucase_getType(UChar32 c) { + uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); + return UCASE_GET_TYPE(props); +} + +/** @return same as ucase_getType() and set bit 2 if c is case-ignorable */ +U_CAPI int32_t U_EXPORT2 +ucase_getTypeOrIgnorable(UChar32 c) { + uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); + return UCASE_GET_TYPE_AND_IGNORABLE(props); +} + +/** @return UCASE_NO_DOT, UCASE_SOFT_DOTTED, UCASE_ABOVE, UCASE_OTHER_ACCENT */ +static inline int32_t +getDotType(UChar32 c) { + uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); + if(!UCASE_HAS_EXCEPTION(props)) { + return props&UCASE_DOT_MASK; + } else { + const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); + return (*pe>>UCASE_EXC_DOT_SHIFT)&UCASE_DOT_MASK; + } +} + +U_CAPI UBool U_EXPORT2 +ucase_isSoftDotted(UChar32 c) { + return (UBool)(getDotType(c)==UCASE_SOFT_DOTTED); +} + +U_CAPI UBool U_EXPORT2 +ucase_isCaseSensitive(UChar32 c) { + uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); + if(!UCASE_HAS_EXCEPTION(props)) { + return (UBool)((props&UCASE_SENSITIVE)!=0); + } else { + const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); + return (UBool)((*pe&UCASE_EXC_SENSITIVE)!=0); + } +} + +/* string casing ------------------------------------------------------------ */ + +/* + * These internal functions form the core of string case mappings. + * They map single code points to result code points or strings and take + * all necessary conditions (context, locale ID, options) into account. + * + * They do not iterate over the source or write to the destination + * so that the same functions are useful for non-standard string storage, + * such as in a Replaceable (for Transliterator) or UTF-8/32 strings etc. + * For the same reason, the "surrounding text" context is passed in as a + * UCaseContextIterator which does not make any assumptions about + * the underlying storage. + * + * This section contains helper functions that check for conditions + * in the input text surrounding the current code point + * according to SpecialCasing.txt. + * + * Each helper function gets the index + * - after the current code point if it looks at following text + * - before the current code point if it looks at preceding text + * + * Unicode 3.2 UAX 21 "Case Mappings" defines the conditions as follows: + * + * Final_Sigma + * C is preceded by a sequence consisting of + * a cased letter and a case-ignorable sequence, + * and C is not followed by a sequence consisting of + * an ignorable sequence and then a cased letter. + * + * More_Above + * C is followed by one or more characters of combining class 230 (ABOVE) + * in the combining character sequence. + * + * After_Soft_Dotted + * The last preceding character with combining class of zero before C + * was Soft_Dotted, + * and there is no intervening combining character class 230 (ABOVE). + * + * Before_Dot + * C is followed by combining dot above (U+0307). + * Any sequence of characters with a combining class that is neither 0 nor 230 + * may intervene between the current character and the combining dot above. + * + * The erratum from 2002-10-31 adds the condition + * + * After_I + * The last preceding base character was an uppercase I, and there is no + * intervening combining character class 230 (ABOVE). + * + * (See Jitterbug 2344 and the comments on After_I below.) + * + * Helper definitions in Unicode 3.2 UAX 21: + * + * D1. A character C is defined to be cased + * if it meets any of the following criteria: + * + * - The general category of C is Titlecase Letter (Lt) + * - In [CoreProps], C has one of the properties Uppercase, or Lowercase + * - Given D = NFD(C), then it is not the case that: + * D = UCD_lower(D) = UCD_upper(D) = UCD_title(D) + * (This third criterion does not add any characters to the list + * for Unicode 3.2. Ignored.) + * + * D2. A character C is defined to be case-ignorable + * if it meets either of the following criteria: + * + * - The general category of C is + * Nonspacing Mark (Mn), or Enclosing Mark (Me), or Format Control (Cf), or + * Letter Modifier (Lm), or Symbol Modifier (Sk) + * - C is one of the following characters + * U+0027 APOSTROPHE + * U+00AD SOFT HYPHEN (SHY) + * U+2019 RIGHT SINGLE QUOTATION MARK + * (the preferred character for apostrophe) + * + * D3. A case-ignorable sequence is a sequence of + * zero or more case-ignorable characters. + */ + +#define is_d(c) ((c)=='d' || (c)=='D') +#define is_e(c) ((c)=='e' || (c)=='E') +#define is_i(c) ((c)=='i' || (c)=='I') +#define is_l(c) ((c)=='l' || (c)=='L') +#define is_r(c) ((c)=='r' || (c)=='R') +#define is_t(c) ((c)=='t' || (c)=='T') +#define is_u(c) ((c)=='u' || (c)=='U') +#define is_y(c) ((c)=='y' || (c)=='Y') +#define is_z(c) ((c)=='z' || (c)=='Z') + +/* separator? */ +#define is_sep(c) ((c)=='_' || (c)=='-' || (c)==0) + +/** + * Requires non-nullptr locale ID but otherwise does the equivalent of + * checking for language codes as if uloc_getLanguage() were called: + * Accepts both 2- and 3-letter codes and accepts case variants. + */ +U_CFUNC int32_t +ucase_getCaseLocale(const char *locale) { + /* + * This function used to use uloc_getLanguage(), but the current code + * removes the dependency of this low-level code on uloc implementation code + * and is faster because not the whole locale ID has to be + * examined and copied/transformed. + * + * Because this code does not want to depend on uloc, the caller must + * pass in a non-nullptr locale, i.e., may need to call uloc_getDefault(). + */ + char c=*locale++; + // Fastpath for English "en" which is often used for default (=root locale) case mappings, + // and for Chinese "zh": Very common but no special case mapping behavior. + // Then check lowercase vs. uppercase to reduce the number of comparisons + // for other locales without special behavior. + if(c=='e') { + /* el or ell? */ + c=*locale++; + if(is_l(c)) { + c=*locale++; + if(is_l(c)) { + c=*locale; + } + if(is_sep(c)) { + return UCASE_LOC_GREEK; + } + } + // en, es, ... -> root + } else if(c=='z') { + return UCASE_LOC_ROOT; +#if U_CHARSET_FAMILY==U_ASCII_FAMILY + } else if(c>='a') { // ASCII a-z = 0x61..0x7a, after A-Z +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY + } else if(c<='z') { // EBCDIC a-z = 0x81..0xa9 with two gaps, before A-Z +#else +# error Unknown charset family! +#endif + // lowercase c + if(c=='t') { + /* tr or tur? */ + c=*locale++; + if(is_u(c)) { + c=*locale++; + } + if(is_r(c)) { + c=*locale; + if(is_sep(c)) { + return UCASE_LOC_TURKISH; + } + } + } else if(c=='a') { + /* az or aze? */ + c=*locale++; + if(is_z(c)) { + c=*locale++; + if(is_e(c)) { + c=*locale; + } + if(is_sep(c)) { + return UCASE_LOC_TURKISH; + } + } + } else if(c=='l') { + /* lt or lit? */ + c=*locale++; + if(is_i(c)) { + c=*locale++; + } + if(is_t(c)) { + c=*locale; + if(is_sep(c)) { + return UCASE_LOC_LITHUANIAN; + } + } + } else if(c=='n') { + /* nl or nld? */ + c=*locale++; + if(is_l(c)) { + c=*locale++; + if(is_d(c)) { + c=*locale; + } + if(is_sep(c)) { + return UCASE_LOC_DUTCH; + } + } + } else if(c=='h') { + /* hy or hye? *not* hyw */ + c=*locale++; + if(is_y(c)) { + c=*locale++; + if(is_e(c)) { + c=*locale; + } + if(is_sep(c)) { + return UCASE_LOC_ARMENIAN; + } + } + } + } else { + // uppercase c + // Same code as for lowercase c but also check for 'E'. + if(c=='T') { + /* tr or tur? */ + c=*locale++; + if(is_u(c)) { + c=*locale++; + } + if(is_r(c)) { + c=*locale; + if(is_sep(c)) { + return UCASE_LOC_TURKISH; + } + } + } else if(c=='A') { + /* az or aze? */ + c=*locale++; + if(is_z(c)) { + c=*locale++; + if(is_e(c)) { + c=*locale; + } + if(is_sep(c)) { + return UCASE_LOC_TURKISH; + } + } + } else if(c=='L') { + /* lt or lit? */ + c=*locale++; + if(is_i(c)) { + c=*locale++; + } + if(is_t(c)) { + c=*locale; + if(is_sep(c)) { + return UCASE_LOC_LITHUANIAN; + } + } + } else if(c=='E') { + /* el or ell? */ + c=*locale++; + if(is_l(c)) { + c=*locale++; + if(is_l(c)) { + c=*locale; + } + if(is_sep(c)) { + return UCASE_LOC_GREEK; + } + } + } else if(c=='N') { + /* nl or nld? */ + c=*locale++; + if(is_l(c)) { + c=*locale++; + if(is_d(c)) { + c=*locale; + } + if(is_sep(c)) { + return UCASE_LOC_DUTCH; + } + } + } else if(c=='H') { + /* hy or hye? *not* hyw */ + c=*locale++; + if(is_y(c)) { + c=*locale++; + if(is_e(c)) { + c=*locale; + } + if(is_sep(c)) { + return UCASE_LOC_ARMENIAN; + } + } + } + } + return UCASE_LOC_ROOT; +} + +/* + * Is followed by + * {case-ignorable}* cased + * ? + * (dir determines looking forward/backward) + * If a character is case-ignorable, it is skipped regardless of whether + * it is also cased or not. + */ +static UBool +isFollowedByCasedLetter(UCaseContextIterator *iter, void *context, int8_t dir) { + UChar32 c; + + if(iter==nullptr) { + return false; + } + + for(/* dir!=0 sets direction */; (c=iter(context, dir))>=0; dir=0) { + int32_t type=ucase_getTypeOrIgnorable(c); + if(type&4) { + /* case-ignorable, continue with the loop */ + } else if(type!=UCASE_NONE) { + return true; /* followed by cased letter */ + } else { + return false; /* uncased and not case-ignorable */ + } + } + + return false; /* not followed by cased letter */ +} + +/* Is preceded by Soft_Dotted character with no intervening cc=230 ? */ +static UBool +isPrecededBySoftDotted(UCaseContextIterator *iter, void *context) { + UChar32 c; + int32_t dotType; + int8_t dir; + + if(iter==nullptr) { + return false; + } + + for(dir=-1; (c=iter(context, dir))>=0; dir=0) { + dotType=getDotType(c); + if(dotType==UCASE_SOFT_DOTTED) { + return true; /* preceded by TYPE_i */ + } else if(dotType!=UCASE_OTHER_ACCENT) { + return false; /* preceded by different base character (not TYPE_i), or intervening cc==230 */ + } + } + + return false; /* not preceded by TYPE_i */ +} + +/* + * See Jitterbug 2344: + * The condition After_I for Turkic-lowercasing of U+0307 combining dot above + * is checked in ICU 2.0, 2.1, 2.6 but was not in 2.2 & 2.4 because + * we made those releases compatible with Unicode 3.2 which had not fixed + * a related bug in SpecialCasing.txt. + * + * From the Jitterbug 2344 text: + * ... this bug is listed as a Unicode erratum + * from 2002-10-31 at http://www.unicode.org/uni2errata/UnicodeErrata.html + * + * There are two errors in SpecialCasing.txt. + * 1. Missing semicolons on two lines. ... [irrelevant for ICU] + * 2. An incorrect context definition. Correct as follows: + * < 0307; ; 0307; 0307; tr After_Soft_Dotted; # COMBINING DOT ABOVE + * < 0307; ; 0307; 0307; az After_Soft_Dotted; # COMBINING DOT ABOVE + * --- + * > 0307; ; 0307; 0307; tr After_I; # COMBINING DOT ABOVE + * > 0307; ; 0307; 0307; az After_I; # COMBINING DOT ABOVE + * where the context After_I is defined as: + * The last preceding base character was an uppercase I, and there is no + * intervening combining character class 230 (ABOVE). + * + * + * Note that SpecialCasing.txt even in Unicode 3.2 described the condition as: + * + * # When lowercasing, remove dot_above in the sequence I + dot_above, which will turn into i. + * # This matches the behavior of the canonically equivalent I-dot_above + * + * See also the description in this place in older versions of uchar.c (revision 1.100). + * + * Markus W. Scherer 2003-feb-15 + */ + +/* Is preceded by base character 'I' with no intervening cc=230 ? */ +static UBool +isPrecededBy_I(UCaseContextIterator *iter, void *context) { + UChar32 c; + int32_t dotType; + int8_t dir; + + if(iter==nullptr) { + return false; + } + + for(dir=-1; (c=iter(context, dir))>=0; dir=0) { + if(c==0x49) { + return true; /* preceded by I */ + } + dotType=getDotType(c); + if(dotType!=UCASE_OTHER_ACCENT) { + return false; /* preceded by different base character (not I), or intervening cc==230 */ + } + } + + return false; /* not preceded by I */ +} + +/* Is followed by one or more cc==230 ? */ +static UBool +isFollowedByMoreAbove(UCaseContextIterator *iter, void *context) { + UChar32 c; + int32_t dotType; + int8_t dir; + + if(iter==nullptr) { + return false; + } + + for(dir=1; (c=iter(context, dir))>=0; dir=0) { + dotType=getDotType(c); + if(dotType==UCASE_ABOVE) { + return true; /* at least one cc==230 following */ + } else if(dotType!=UCASE_OTHER_ACCENT) { + return false; /* next base character, no more cc==230 following */ + } + } + + return false; /* no more cc==230 following */ +} + +/* Is followed by a dot above (without cc==230 in between) ? */ +static UBool +isFollowedByDotAbove(UCaseContextIterator *iter, void *context) { + UChar32 c; + int32_t dotType; + int8_t dir; + + if(iter==nullptr) { + return false; + } + + for(dir=1; (c=iter(context, dir))>=0; dir=0) { + if(c==0x307) { + return true; + } + dotType=getDotType(c); + if(dotType!=UCASE_OTHER_ACCENT) { + return false; /* next base character or cc==230 in between */ + } + } + + return false; /* no dot above following */ +} + +U_CAPI int32_t U_EXPORT2 +ucase_toFullLower(UChar32 c, + UCaseContextIterator *iter, void *context, + const char16_t **pString, + int32_t loc) { + // The sign of the result has meaning, input must be non-negative so that it can be returned as is. + U_ASSERT(c >= 0); + UChar32 result=c; + // Reset the output pointer in case it was uninitialized. + *pString=nullptr; + uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); + if(!UCASE_HAS_EXCEPTION(props)) { + if(UCASE_IS_UPPER_OR_TITLE(props)) { + result=c+UCASE_GET_DELTA(props); + } + } else { + const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props), *pe2; + uint16_t excWord=*pe++; + int32_t full; + + pe2=pe; + + if(excWord&UCASE_EXC_CONDITIONAL_SPECIAL) { + /* use hardcoded conditions and mappings */ + + /* + * Test for conditional mappings first + * (otherwise the unconditional default mappings are always taken), + * then test for characters that have unconditional mappings in SpecialCasing.txt, + * then get the UnicodeData.txt mappings. + */ + if( loc==UCASE_LOC_LITHUANIAN && + /* base characters, find accents above */ + (((c==0x49 || c==0x4a || c==0x12e) && + isFollowedByMoreAbove(iter, context)) || + /* precomposed with accent above, no need to find one */ + (c==0xcc || c==0xcd || c==0x128)) + ) { + /* + # Lithuanian + + # Lithuanian retains the dot in a lowercase i when followed by accents. + + # Introduce an explicit dot above when lowercasing capital I's and J's + # whenever there are more accents above. + # (of the accents used in Lithuanian: grave, acute, tilde above, and ogonek) + + 0049; 0069 0307; 0049; 0049; lt More_Above; # LATIN CAPITAL LETTER I + 004A; 006A 0307; 004A; 004A; lt More_Above; # LATIN CAPITAL LETTER J + 012E; 012F 0307; 012E; 012E; lt More_Above; # LATIN CAPITAL LETTER I WITH OGONEK + 00CC; 0069 0307 0300; 00CC; 00CC; lt; # LATIN CAPITAL LETTER I WITH GRAVE + 00CD; 0069 0307 0301; 00CD; 00CD; lt; # LATIN CAPITAL LETTER I WITH ACUTE + 0128; 0069 0307 0303; 0128; 0128; lt; # LATIN CAPITAL LETTER I WITH TILDE + */ + switch(c) { + case 0x49: /* LATIN CAPITAL LETTER I */ + *pString=iDot; + return 2; + case 0x4a: /* LATIN CAPITAL LETTER J */ + *pString=jDot; + return 2; + case 0x12e: /* LATIN CAPITAL LETTER I WITH OGONEK */ + *pString=iOgonekDot; + return 2; + case 0xcc: /* LATIN CAPITAL LETTER I WITH GRAVE */ + *pString=iDotGrave; + return 3; + case 0xcd: /* LATIN CAPITAL LETTER I WITH ACUTE */ + *pString=iDotAcute; + return 3; + case 0x128: /* LATIN CAPITAL LETTER I WITH TILDE */ + *pString=iDotTilde; + return 3; + default: + return 0; /* will not occur */ + } + /* # Turkish and Azeri */ + } else if(loc==UCASE_LOC_TURKISH && c==0x130) { + /* + # I and i-dotless; I-dot and i are case pairs in Turkish and Azeri + # The following rules handle those cases. + + 0130; 0069; 0130; 0130; tr # LATIN CAPITAL LETTER I WITH DOT ABOVE + 0130; 0069; 0130; 0130; az # LATIN CAPITAL LETTER I WITH DOT ABOVE + */ + return 0x69; + } else if(loc==UCASE_LOC_TURKISH && c==0x307 && isPrecededBy_I(iter, context)) { + /* + # When lowercasing, remove dot_above in the sequence I + dot_above, which will turn into i. + # This matches the behavior of the canonically equivalent I-dot_above + + 0307; ; 0307; 0307; tr After_I; # COMBINING DOT ABOVE + 0307; ; 0307; 0307; az After_I; # COMBINING DOT ABOVE + */ + return 0; /* remove the dot (continue without output) */ + } else if(loc==UCASE_LOC_TURKISH && c==0x49 && !isFollowedByDotAbove(iter, context)) { + /* + # When lowercasing, unless an I is before a dot_above, it turns into a dotless i. + + 0049; 0131; 0049; 0049; tr Not_Before_Dot; # LATIN CAPITAL LETTER I + 0049; 0131; 0049; 0049; az Not_Before_Dot; # LATIN CAPITAL LETTER I + */ + return 0x131; + } else if(c==0x130) { + /* + # Preserve canonical equivalence for I with dot. Turkic is handled below. + + 0130; 0069 0307; 0130; 0130; # LATIN CAPITAL LETTER I WITH DOT ABOVE + */ + *pString=iDot; + return 2; + } else if( c==0x3a3 && + !isFollowedByCasedLetter(iter, context, 1) && + isFollowedByCasedLetter(iter, context, -1) /* -1=preceded */ + ) { + /* greek capital sigma maps depending on surrounding cased letters (see SpecialCasing.txt) */ + /* + # Special case for final form of sigma + + 03A3; 03C2; 03A3; 03A3; Final_Sigma; # GREEK CAPITAL LETTER SIGMA + */ + return 0x3c2; /* greek small final sigma */ + } else { + /* no known conditional special case mapping, use a normal mapping */ + } + } else if(HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { + GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, full); + full&=UCASE_FULL_LOWER; + if(full!=0) { + /* set the output pointer to the lowercase mapping */ + *pString=reinterpret_cast(pe+1); + + /* return the string length */ + return full; + } + } + + if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_IS_UPPER_OR_TITLE(props)) { + int32_t delta; + GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe2, delta); + return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; + } + if(HAS_SLOT(excWord, UCASE_EXC_LOWER)) { + GET_SLOT_VALUE(excWord, UCASE_EXC_LOWER, pe2, result); + } + } + + return (result==c) ? ~result : result; +} + +/* internal */ +static int32_t +toUpperOrTitle(UChar32 c, + UCaseContextIterator *iter, void *context, + const char16_t **pString, + int32_t loc, + UBool upperNotTitle) { + // The sign of the result has meaning, input must be non-negative so that it can be returned as is. + U_ASSERT(c >= 0); + UChar32 result=c; + // Reset the output pointer in case it was uninitialized. + *pString=nullptr; + uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); + if(!UCASE_HAS_EXCEPTION(props)) { + if(UCASE_GET_TYPE(props)==UCASE_LOWER) { + result=c+UCASE_GET_DELTA(props); + } + } else { + const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props), *pe2; + uint16_t excWord=*pe++; + int32_t full, idx; + + pe2=pe; + + if(excWord&UCASE_EXC_CONDITIONAL_SPECIAL) { + /* use hardcoded conditions and mappings */ + if(loc==UCASE_LOC_TURKISH && c==0x69) { + /* + # Turkish and Azeri + + # I and i-dotless; I-dot and i are case pairs in Turkish and Azeri + # The following rules handle those cases. + + # When uppercasing, i turns into a dotted capital I + + 0069; 0069; 0130; 0130; tr; # LATIN SMALL LETTER I + 0069; 0069; 0130; 0130; az; # LATIN SMALL LETTER I + */ + return 0x130; + } else if(loc==UCASE_LOC_LITHUANIAN && c==0x307 && isPrecededBySoftDotted(iter, context)) { + /* + # Lithuanian + + # Lithuanian retains the dot in a lowercase i when followed by accents. + + # Remove DOT ABOVE after "i" with upper or titlecase + + 0307; 0307; ; ; lt After_Soft_Dotted; # COMBINING DOT ABOVE + */ + return 0; /* remove the dot (continue without output) */ + } else if(c==0x0587) { + // See ICU-13416: + // և ligature ech-yiwn + // uppercases to ԵՒ=ech+yiwn by default and in Western Armenian, + // but to ԵՎ=ech+vew in Eastern Armenian. + if(loc==UCASE_LOC_ARMENIAN) { + *pString=upperNotTitle ? u"ԵՎ" : u"Եվ"; + } else { + *pString=upperNotTitle ? u"ԵՒ" : u"Եւ"; + } + return 2; + } else { + /* no known conditional special case mapping, use a normal mapping */ + } + } else if(HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { + GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, full); + + /* start of full case mapping strings */ + ++pe; + + /* skip the lowercase and case-folding result strings */ + pe+=full&UCASE_FULL_LOWER; + full>>=4; + pe+=full&0xf; + full>>=4; + + if(upperNotTitle) { + full&=0xf; + } else { + /* skip the uppercase result string */ + pe+=full&0xf; + full=(full>>4)&0xf; + } + + if(full!=0) { + /* set the output pointer to the result string */ + *pString=reinterpret_cast(pe); + + /* return the string length */ + return full; + } + } + + if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_GET_TYPE(props)==UCASE_LOWER) { + int32_t delta; + GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe2, delta); + return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; + } + if(!upperNotTitle && HAS_SLOT(excWord, UCASE_EXC_TITLE)) { + idx=UCASE_EXC_TITLE; + } else if(HAS_SLOT(excWord, UCASE_EXC_UPPER)) { + /* here, titlecase is same as uppercase */ + idx=UCASE_EXC_UPPER; + } else { + return ~c; + } + GET_SLOT_VALUE(excWord, idx, pe2, result); + } + + return (result==c) ? ~result : result; +} + +U_CAPI int32_t U_EXPORT2 +ucase_toFullUpper(UChar32 c, + UCaseContextIterator *iter, void *context, + const char16_t **pString, + int32_t caseLocale) { + return toUpperOrTitle(c, iter, context, pString, caseLocale, true); +} + +U_CAPI int32_t U_EXPORT2 +ucase_toFullTitle(UChar32 c, + UCaseContextIterator *iter, void *context, + const char16_t **pString, + int32_t caseLocale) { + return toUpperOrTitle(c, iter, context, pString, caseLocale, false); +} + +/* case folding ------------------------------------------------------------- */ + +/* + * Case folding is similar to lowercasing. + * The result may be a simple mapping, i.e., a single code point, or + * a full mapping, i.e., a string. + * If the case folding for a code point is the same as its simple (1:1) lowercase mapping, + * then only the lowercase mapping is stored. + * + * Some special cases are hardcoded because their conditions cannot be + * parsed and processed from CaseFolding.txt. + * + * Unicode 3.2 CaseFolding.txt specifies for its status field: + +# C: common case folding, common mappings shared by both simple and full mappings. +# F: full case folding, mappings that cause strings to grow in length. Multiple characters are separated by spaces. +# S: simple case folding, mappings to single characters where different from F. +# T: special case for uppercase I and dotted uppercase I +# - For non-Turkic languages, this mapping is normally not used. +# - For Turkic languages (tr, az), this mapping can be used instead of the normal mapping for these characters. +# +# Usage: +# A. To do a simple case folding, use the mappings with status C + S. +# B. To do a full case folding, use the mappings with status C + F. +# +# The mappings with status T can be used or omitted depending on the desired case-folding +# behavior. (The default option is to exclude them.) + + * Unicode 3.2 has 'T' mappings as follows: + +0049; T; 0131; # LATIN CAPITAL LETTER I +0130; T; 0069; # LATIN CAPITAL LETTER I WITH DOT ABOVE + + * while the default mappings for these code points are: + +0049; C; 0069; # LATIN CAPITAL LETTER I +0130; F; 0069 0307; # LATIN CAPITAL LETTER I WITH DOT ABOVE + + * U+0130 has no simple case folding (simple-case-folds to itself). + */ + +/* return the simple case folding mapping for c */ +U_CAPI UChar32 U_EXPORT2 +ucase_fold(UChar32 c, uint32_t options) { + uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); + if(!UCASE_HAS_EXCEPTION(props)) { + if(UCASE_IS_UPPER_OR_TITLE(props)) { + c+=UCASE_GET_DELTA(props); + } + } else { + const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); + uint16_t excWord=*pe++; + int32_t idx; + if(excWord&UCASE_EXC_CONDITIONAL_FOLD) { + /* special case folding mappings, hardcoded */ + if((options&_FOLD_CASE_OPTIONS_MASK)==U_FOLD_CASE_DEFAULT) { + /* default mappings */ + if(c==0x49) { + /* 0049; C; 0069; # LATIN CAPITAL LETTER I */ + return 0x69; + } else if(c==0x130) { + /* no simple case folding for U+0130 */ + return c; + } + } else { + /* Turkic mappings */ + if(c==0x49) { + /* 0049; T; 0131; # LATIN CAPITAL LETTER I */ + return 0x131; + } else if(c==0x130) { + /* 0130; T; 0069; # LATIN CAPITAL LETTER I WITH DOT ABOVE */ + return 0x69; + } + } + } + if((excWord&UCASE_EXC_NO_SIMPLE_CASE_FOLDING)!=0) { + return c; + } + if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_IS_UPPER_OR_TITLE(props)) { + int32_t delta; + GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe, delta); + return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; + } + if(HAS_SLOT(excWord, UCASE_EXC_FOLD)) { + idx=UCASE_EXC_FOLD; + } else if(HAS_SLOT(excWord, UCASE_EXC_LOWER)) { + idx=UCASE_EXC_LOWER; + } else { + return c; + } + GET_SLOT_VALUE(excWord, idx, pe, c); + } + return c; +} + +/* + * Issue for canonical caseless match (UAX #21): + * Turkic casefolding (using "T" mappings in CaseFolding.txt) does not preserve + * canonical equivalence, unlike default-option casefolding. + * For example, I-grave and I + grave fold to strings that are not canonically + * equivalent. + * For more details, see the comment in unorm_compare() in unorm.cpp + * and the intermediate prototype changes for Jitterbug 2021. + * (For example, revision 1.104 of uchar.c and 1.4 of CaseFolding.txt.) + * + * This did not get fixed because it appears that it is not possible to fix + * it for uppercase and lowercase characters (I-grave vs. i-grave) + * together in a way that they still fold to common result strings. + */ + +U_CAPI int32_t U_EXPORT2 +ucase_toFullFolding(UChar32 c, + const char16_t **pString, + uint32_t options) { + // The sign of the result has meaning, input must be non-negative so that it can be returned as is. + U_ASSERT(c >= 0); + UChar32 result=c; + // Reset the output pointer in case it was uninitialized. + *pString=nullptr; + uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); + if(!UCASE_HAS_EXCEPTION(props)) { + if(UCASE_IS_UPPER_OR_TITLE(props)) { + result=c+UCASE_GET_DELTA(props); + } + } else { + const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props), *pe2; + uint16_t excWord=*pe++; + int32_t full, idx; + + pe2=pe; + + if(excWord&UCASE_EXC_CONDITIONAL_FOLD) { + /* use hardcoded conditions and mappings */ + if((options&_FOLD_CASE_OPTIONS_MASK)==U_FOLD_CASE_DEFAULT) { + /* default mappings */ + if(c==0x49) { + /* 0049; C; 0069; # LATIN CAPITAL LETTER I */ + return 0x69; + } else if(c==0x130) { + /* 0130; F; 0069 0307; # LATIN CAPITAL LETTER I WITH DOT ABOVE */ + *pString=iDot; + return 2; + } + } else { + /* Turkic mappings */ + if(c==0x49) { + /* 0049; T; 0131; # LATIN CAPITAL LETTER I */ + return 0x131; + } else if(c==0x130) { + /* 0130; T; 0069; # LATIN CAPITAL LETTER I WITH DOT ABOVE */ + return 0x69; + } + } + } else if(HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { + GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, full); + + /* start of full case mapping strings */ + ++pe; + + /* skip the lowercase result string */ + pe+=full&UCASE_FULL_LOWER; + full=(full>>4)&0xf; + + if(full!=0) { + /* set the output pointer to the result string */ + *pString=reinterpret_cast(pe); + + /* return the string length */ + return full; + } + } + + if((excWord&UCASE_EXC_NO_SIMPLE_CASE_FOLDING)!=0) { + return ~c; + } + if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_IS_UPPER_OR_TITLE(props)) { + int32_t delta; + GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe2, delta); + return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; + } + if(HAS_SLOT(excWord, UCASE_EXC_FOLD)) { + idx=UCASE_EXC_FOLD; + } else if(HAS_SLOT(excWord, UCASE_EXC_LOWER)) { + idx=UCASE_EXC_LOWER; + } else { + return ~c; + } + GET_SLOT_VALUE(excWord, idx, pe2, result); + } + + return (result==c) ? ~result : result; +} + +/* case mapping properties API ---------------------------------------------- */ + +/* public API (see uchar.h) */ + +U_CAPI UBool U_EXPORT2 +u_isULowercase(UChar32 c) { + return (UBool)(UCASE_LOWER==ucase_getType(c)); +} + +U_CAPI UBool U_EXPORT2 +u_isUUppercase(UChar32 c) { + return (UBool)(UCASE_UPPER==ucase_getType(c)); +} + +/* Transforms the Unicode character to its lower case equivalent.*/ +U_CAPI UChar32 U_EXPORT2 +u_tolower(UChar32 c) { + return ucase_tolower(c); +} + +/* Transforms the Unicode character to its upper case equivalent.*/ +U_CAPI UChar32 U_EXPORT2 +u_toupper(UChar32 c) { + return ucase_toupper(c); +} + +/* Transforms the Unicode character to its title case equivalent.*/ +U_CAPI UChar32 U_EXPORT2 +u_totitle(UChar32 c) { + return ucase_totitle(c); +} + +/* return the simple case folding mapping for c */ +U_CAPI UChar32 U_EXPORT2 +u_foldCase(UChar32 c, uint32_t options) { + return ucase_fold(c, options); +} + +U_CFUNC int32_t U_EXPORT2 +ucase_hasBinaryProperty(UChar32 c, UProperty which) { + /* case mapping properties */ + const char16_t *resultString; + switch(which) { + case UCHAR_LOWERCASE: + return (UBool)(UCASE_LOWER==ucase_getType(c)); + case UCHAR_UPPERCASE: + return (UBool)(UCASE_UPPER==ucase_getType(c)); + case UCHAR_SOFT_DOTTED: + return ucase_isSoftDotted(c); + case UCHAR_CASE_SENSITIVE: + return ucase_isCaseSensitive(c); + case UCHAR_CASED: + return (UBool)(UCASE_NONE!=ucase_getType(c)); + case UCHAR_CASE_IGNORABLE: + return (UBool)(ucase_getTypeOrIgnorable(c)>>2); + /* + * Note: The following Changes_When_Xyz are defined as testing whether + * the NFD form of the input changes when Xyz-case-mapped. + * However, this simpler implementation of these properties, + * ignoring NFD, passes the tests. + * The implementation needs to be changed if the tests start failing. + * When that happens, optimizations should be used to work with the + * per-single-code point ucase_toFullXyz() functions unless + * the NFD form has more than one code point, + * and the property starts set needs to be the union of the + * start sets for normalization and case mappings. + */ + case UCHAR_CHANGES_WHEN_LOWERCASED: + return (UBool)(ucase_toFullLower(c, nullptr, nullptr, &resultString, UCASE_LOC_ROOT)>=0); + case UCHAR_CHANGES_WHEN_UPPERCASED: + return (UBool)(ucase_toFullUpper(c, nullptr, nullptr, &resultString, UCASE_LOC_ROOT)>=0); + case UCHAR_CHANGES_WHEN_TITLECASED: + return (UBool)(ucase_toFullTitle(c, nullptr, nullptr, &resultString, UCASE_LOC_ROOT)>=0); + /* case UCHAR_CHANGES_WHEN_CASEFOLDED: -- in uprops.c */ + case UCHAR_CHANGES_WHEN_CASEMAPPED: + return (UBool)( + ucase_toFullLower(c, nullptr, nullptr, &resultString, UCASE_LOC_ROOT)>=0 || + ucase_toFullUpper(c, nullptr, nullptr, &resultString, UCASE_LOC_ROOT)>=0 || + ucase_toFullTitle(c, nullptr, nullptr, &resultString, UCASE_LOC_ROOT)>=0); + default: + return false; + } +} diff --git a/intl/icu/source/common/ucase.h b/intl/icu/source/common/ucase.h new file mode 100644 index 0000000000..e03b311870 --- /dev/null +++ b/intl/icu/source/common/ucase.h @@ -0,0 +1,464 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2004-2012, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: ucase.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2004aug30 +* created by: Markus W. Scherer +* +* Low-level Unicode character/string case mapping code. +*/ + +#ifndef __UCASE_H__ +#define __UCASE_H__ + +#include "unicode/utypes.h" +#include "unicode/uset.h" +#include "putilimp.h" +#include "uset_imp.h" +#include "udataswp.h" +#include "utrie2.h" + +#ifdef __cplusplus +U_NAMESPACE_BEGIN + +class UnicodeString; + +U_NAMESPACE_END +#endif + +/* library API -------------------------------------------------------------- */ + +U_CFUNC void U_EXPORT2 +ucase_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode); + +/** + * Requires non-NULL locale ID but otherwise does the equivalent of + * checking for language codes as if uloc_getLanguage() were called: + * Accepts both 2- and 3-letter codes and accepts case variants. + */ +U_CFUNC int32_t +ucase_getCaseLocale(const char *locale); + +/* Casing locale types for ucase_getCaseLocale */ +enum { + UCASE_LOC_UNKNOWN, + UCASE_LOC_ROOT, + UCASE_LOC_TURKISH, + UCASE_LOC_LITHUANIAN, + UCASE_LOC_GREEK, + UCASE_LOC_DUTCH, + UCASE_LOC_ARMENIAN +}; + +/** + * Bit mask for getting just the options from a string compare options word + * that are relevant for case-insensitive string comparison. + * See stringoptions.h. Also include _STRNCMP_STYLE and U_COMPARE_CODE_POINT_ORDER. + * @internal + */ +#define _STRCASECMP_OPTIONS_MASK 0xffff + +/** + * Bit mask for getting just the options from a string compare options word + * that are relevant for case folding (of a single string or code point). + * + * Currently only bit 0 for U_FOLD_CASE_EXCLUDE_SPECIAL_I. + * It is conceivable that at some point we might use one more bit for using uppercase sharp s. + * It is conceivable that at some point we might want the option to use only simple case foldings + * when operating on strings. + * + * See stringoptions.h. + * @internal + */ +#define _FOLD_CASE_OPTIONS_MASK 7 + +/* single-code point functions */ + +U_CAPI UChar32 U_EXPORT2 +ucase_tolower(UChar32 c); + +U_CAPI UChar32 U_EXPORT2 +ucase_toupper(UChar32 c); + +U_CAPI UChar32 U_EXPORT2 +ucase_totitle(UChar32 c); + +U_CAPI UChar32 U_EXPORT2 +ucase_fold(UChar32 c, uint32_t options); + +/** + * Adds all simple case mappings and the full case folding for c to sa, + * and also adds special case closure mappings. + * c itself is not added. + * For example, the mappings + * - for s include long s + * - for sharp s include ss + * - for k include the Kelvin sign + */ +U_CFUNC void U_EXPORT2 +ucase_addCaseClosure(UChar32 c, const USetAdder *sa); + +/** Case closure with only scf=Simple_Case_Folding. */ +U_CFUNC void U_EXPORT2 +ucase_addSimpleCaseClosure(UChar32 c, const USetAdder *sa); + +/** + * Maps the string to single code points and adds the associated case closure + * mappings. + * The string is mapped to code points if it is their full case folding string. + * In other words, this performs a reverse full case folding and then + * adds the case closure items of the resulting code points. + * If the string is found and its closure applied, then + * the string itself is added as well as part of its code points' closure. + * It must be length>=0. + * + * @return true if the string was found + */ +U_CFUNC UBool U_EXPORT2 +ucase_addStringCaseClosure(const UChar *s, int32_t length, const USetAdder *sa); + +#ifdef __cplusplus +U_NAMESPACE_BEGIN + +/** + * Iterator over characters with more than one code point in the full default Case_Folding. + */ +class U_COMMON_API FullCaseFoldingIterator { +public: + /** Constructor. */ + FullCaseFoldingIterator(); + /** + * Returns the next (cp, full) pair where "full" is cp's full default Case_Folding. + * Returns a negative cp value at the end of the iteration. + */ + UChar32 next(UnicodeString &full); +private: + FullCaseFoldingIterator(const FullCaseFoldingIterator &) = delete; // no copy + FullCaseFoldingIterator &operator=(const FullCaseFoldingIterator &) = delete; // no assignment + + const char16_t *unfold; + int32_t unfoldRows; + int32_t unfoldRowWidth; + int32_t unfoldStringWidth; + int32_t currentRow; + int32_t rowCpIndex; +}; + +/** + * Fast case mapping data for ASCII/Latin. + * Linear arrays of delta bytes: 0=no mapping; EXC=exception. + * Deltas must not cross the ASCII boundary, or else they cannot be easily used + * in simple UTF-8 code. + */ +namespace LatinCase { + +/** Case mapping/folding data for code points up to U+017F. */ +constexpr char16_t LIMIT = 0x180; +/** U+017F case-folds and uppercases crossing the ASCII boundary. */ +constexpr char16_t LONG_S = 0x17f; +/** Exception: Complex mapping, or too-large delta. */ +constexpr int8_t EXC = -0x80; + +/** Deltas for lowercasing for most locales, and default case folding. */ +extern const int8_t TO_LOWER_NORMAL[LIMIT]; +/** Deltas for lowercasing for tr/az/lt, and Turkic case folding. */ +extern const int8_t TO_LOWER_TR_LT[LIMIT]; + +/** Deltas for uppercasing for most locales. */ +extern const int8_t TO_UPPER_NORMAL[LIMIT]; +/** Deltas for uppercasing for tr/az. */ +extern const int8_t TO_UPPER_TR[LIMIT]; + +} // namespace LatinCase + +U_NAMESPACE_END +#endif + +/** @return UCASE_NONE, UCASE_LOWER, UCASE_UPPER, UCASE_TITLE */ +U_CAPI int32_t U_EXPORT2 +ucase_getType(UChar32 c); + +/** @return like ucase_getType() but also sets UCASE_IGNORABLE if c is case-ignorable */ +U_CAPI int32_t U_EXPORT2 +ucase_getTypeOrIgnorable(UChar32 c); + +U_CAPI UBool U_EXPORT2 +ucase_isSoftDotted(UChar32 c); + +U_CAPI UBool U_EXPORT2 +ucase_isCaseSensitive(UChar32 c); + +/* string case mapping functions */ + +U_CDECL_BEGIN + +/** + * Iterator function for string case mappings, which need to look at the + * context (surrounding text) of a given character for conditional mappings. + * + * The iterator only needs to go backward or forward away from the + * character in question. It does not use any indexes on this interface. + * It does not support random access or an arbitrary change of + * iteration direction. + * + * The code point being case-mapped itself is never returned by + * this iterator. + * + * @param context A pointer to the iterator's working data. + * @param dir If <0 then start iterating backward from the character; + * if >0 then start iterating forward from the character; + * if 0 then continue iterating in the current direction. + * @return Next code point, or <0 when the iteration is done. + */ +typedef UChar32 U_CALLCONV +UCaseContextIterator(void *context, int8_t dir); + +/** + * Sample struct which may be used by some implementations of + * UCaseContextIterator. + */ +struct UCaseContext { + void *p; + int32_t start, index, limit; + int32_t cpStart, cpLimit; + int8_t dir; + int8_t b1, b2, b3; +}; +typedef struct UCaseContext UCaseContext; + +U_CDECL_END + +#define UCASECONTEXT_INITIALIZER { NULL, 0, 0, 0, 0, 0, 0, 0, 0, 0 } + +enum { + /** + * For string case mappings, a single character (a code point) is mapped + * either to itself (in which case in-place mapping functions do nothing), + * or to another single code point, or to a string. + * Aside from the string contents, these are indicated with a single int32_t + * value as follows: + * + * Mapping to self: Negative values (~self instead of -self to support U+0000) + * + * Mapping to another code point: Positive values >UCASE_MAX_STRING_LENGTH + * + * Mapping to a string: The string length (0..UCASE_MAX_STRING_LENGTH) is + * returned. Note that the string result may indeed have zero length. + */ + UCASE_MAX_STRING_LENGTH=0x1f +}; + +/** + * Get the full lowercase mapping for c. + * + * @param csp Case mapping properties. + * @param c Character to be mapped. + * @param iter Character iterator, used for context-sensitive mappings. + * See UCaseContextIterator for details. + * If iter==NULL then a context-independent result is returned. + * @param context Pointer to be passed into iter. + * @param pString If the mapping result is a string, then the pointer is + * written to *pString. + * @param caseLocale Case locale value from ucase_getCaseLocale(). + * @return Output code point or string length, see UCASE_MAX_STRING_LENGTH. + * + * @see UCaseContextIterator + * @see UCASE_MAX_STRING_LENGTH + * @internal + */ +U_CAPI int32_t U_EXPORT2 +ucase_toFullLower(UChar32 c, + UCaseContextIterator *iter, void *context, + const UChar **pString, + int32_t caseLocale); + +U_CAPI int32_t U_EXPORT2 +ucase_toFullUpper(UChar32 c, + UCaseContextIterator *iter, void *context, + const UChar **pString, + int32_t caseLocale); + +U_CAPI int32_t U_EXPORT2 +ucase_toFullTitle(UChar32 c, + UCaseContextIterator *iter, void *context, + const UChar **pString, + int32_t caseLocale); + +U_CAPI int32_t U_EXPORT2 +ucase_toFullFolding(UChar32 c, + const UChar **pString, + uint32_t options); + +U_CFUNC int32_t U_EXPORT2 +ucase_hasBinaryProperty(UChar32 c, UProperty which); + + +U_CDECL_BEGIN + +/** + * @internal + */ +typedef int32_t U_CALLCONV +UCaseMapFull(UChar32 c, + UCaseContextIterator *iter, void *context, + const UChar **pString, + int32_t caseLocale); + +U_CDECL_END + +/* for icuexportdata -------------------------------------------------------- */ + +struct UCaseProps { + void *mem; // TODO: was unused, and type UDataMemory -- remove + const int32_t *indexes; + const uint16_t *exceptions; + const uint16_t *unfold; + + UTrie2 trie; + uint8_t formatVersion[4]; +}; + +U_CAPI const struct UCaseProps * U_EXPORT2 +ucase_getSingleton(int32_t *pExceptionsLength, int32_t *pUnfoldLength); + +/* file definitions --------------------------------------------------------- */ + +#define UCASE_DATA_NAME "ucase" +#define UCASE_DATA_TYPE "icu" + +/* format "cAsE" */ +#define UCASE_FMT_0 0x63 +#define UCASE_FMT_1 0x41 +#define UCASE_FMT_2 0x53 +#define UCASE_FMT_3 0x45 + +/* indexes into indexes[] */ +enum { + UCASE_IX_INDEX_TOP, + UCASE_IX_LENGTH, + UCASE_IX_TRIE_SIZE, + UCASE_IX_EXC_LENGTH, + UCASE_IX_UNFOLD_LENGTH, + + UCASE_IX_MAX_FULL_LENGTH=15, + UCASE_IX_TOP=16 +}; + +/* definitions for 16-bit case properties word ------------------------------ */ + +U_CFUNC const UTrie2 * U_EXPORT2 +ucase_getTrie(); + +/* 2-bit constants for types of cased characters */ +#define UCASE_TYPE_MASK 3 +enum { + UCASE_NONE, + UCASE_LOWER, + UCASE_UPPER, + UCASE_TITLE +}; + +#define UCASE_GET_TYPE(props) ((props)&UCASE_TYPE_MASK) +#define UCASE_GET_TYPE_AND_IGNORABLE(props) ((props)&7) + +#define UCASE_IS_UPPER_OR_TITLE(props) ((props)&2) + +#define UCASE_IGNORABLE 4 +#define UCASE_EXCEPTION 8 +#define UCASE_SENSITIVE 0x10 + +#define UCASE_HAS_EXCEPTION(props) ((props)&UCASE_EXCEPTION) + +#define UCASE_DOT_MASK 0x60 +enum { + UCASE_NO_DOT=0, /* normal characters with cc=0 */ + UCASE_SOFT_DOTTED=0x20, /* soft-dotted characters with cc=0 */ + UCASE_ABOVE=0x40, /* "above" accents with cc=230 */ + UCASE_OTHER_ACCENT=0x60 /* other accent character (0>UCASE_DELTA_SHIFT) +#else +# define UCASE_GET_DELTA(props) (int16_t)(((props)&0x8000) ? (((props)>>UCASE_DELTA_SHIFT)|0xfe00) : ((uint16_t)(props)>>UCASE_DELTA_SHIFT)) +#endif + +/* exception: bits 15..4 are an unsigned 12-bit index into the exceptions array */ +#define UCASE_EXC_SHIFT 4 +#define UCASE_EXC_MASK 0xfff0 +#define UCASE_MAX_EXCEPTIONS ((UCASE_EXC_MASK>>UCASE_EXC_SHIFT)+1) + +/* definitions for 16-bit main exceptions word ------------------------------ */ + +/* first 8 bits indicate values in optional slots */ +enum { + UCASE_EXC_LOWER, + UCASE_EXC_FOLD, + UCASE_EXC_UPPER, + UCASE_EXC_TITLE, + UCASE_EXC_DELTA, + UCASE_EXC_5, /* reserved */ + UCASE_EXC_CLOSURE, + UCASE_EXC_FULL_MAPPINGS, + UCASE_EXC_ALL_SLOTS /* one past the last slot */ +}; + +/* each slot is 2 uint16_t instead of 1 */ +#define UCASE_EXC_DOUBLE_SLOTS 0x100 + +enum { + UCASE_EXC_NO_SIMPLE_CASE_FOLDING=0x200, + UCASE_EXC_DELTA_IS_NEGATIVE=0x400, + UCASE_EXC_SENSITIVE=0x800 +}; + +/* UCASE_EXC_DOT_MASK=UCASE_DOT_MASK<locale; +} + +U_CAPI uint32_t U_EXPORT2 +ucasemap_getOptions(const UCaseMap *csm) { + return csm->options; +} + +U_CAPI void U_EXPORT2 +ucasemap_setLocale(UCaseMap *csm, const char *locale, UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { + return; + } + if (locale != nullptr && *locale == 0) { + csm->locale[0] = 0; + csm->caseLocale = UCASE_LOC_ROOT; + return; + } + + int32_t length=uloc_getName(locale, csm->locale, (int32_t)sizeof(csm->locale), pErrorCode); + if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR || length==sizeof(csm->locale)) { + *pErrorCode=U_ZERO_ERROR; + /* we only really need the language code for case mappings */ + length=uloc_getLanguage(locale, csm->locale, (int32_t)sizeof(csm->locale), pErrorCode); + } + if(length==sizeof(csm->locale)) { + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + if(U_SUCCESS(*pErrorCode)) { + csm->caseLocale = ucase_getCaseLocale(csm->locale); + } else { + csm->locale[0]=0; + csm->caseLocale = UCASE_LOC_ROOT; + } +} + +U_CAPI void U_EXPORT2 +ucasemap_setOptions(UCaseMap *csm, uint32_t options, UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { + return; + } + csm->options=options; +} + +/* UTF-8 string case mappings ----------------------------------------------- */ + +/* TODO(markus): Move to a new, separate utf8case.cpp file. */ + +namespace { + +/* append a full case mapping result, see UCASE_MAX_STRING_LENGTH */ +inline UBool +appendResult(int32_t cpLength, int32_t result, const char16_t *s, + ByteSink &sink, uint32_t options, icu::Edits *edits, UErrorCode &errorCode) { + U_ASSERT(U_SUCCESS(errorCode)); + + /* decode the result */ + if(result<0) { + /* (not) original code point */ + if(edits!=nullptr) { + edits->addUnchanged(cpLength); + } + if((options & U_OMIT_UNCHANGED_TEXT) == 0) { + ByteSinkUtil::appendCodePoint(cpLength, ~result, sink); + } + } else { + if(result<=UCASE_MAX_STRING_LENGTH) { + // string: "result" is the UTF-16 length + return ByteSinkUtil::appendChange(cpLength, s, result, sink, edits, errorCode); + } else { + ByteSinkUtil::appendCodePoint(cpLength, result, sink, edits); + } + } + return true; +} + +// See unicode/utf8.h U8_APPEND_UNSAFE(). +inline uint8_t getTwoByteLead(UChar32 c) { return (uint8_t)((c >> 6) | 0xc0); } +inline uint8_t getTwoByteTrail(UChar32 c) { return (uint8_t)((c & 0x3f) | 0x80); } + +UChar32 U_CALLCONV +utf8_caseContextIterator(void *context, int8_t dir) { + UCaseContext *csc=(UCaseContext *)context; + UChar32 c; + + if(dir<0) { + /* reset for backward iteration */ + csc->index=csc->cpStart; + csc->dir=dir; + } else if(dir>0) { + /* reset for forward iteration */ + csc->index=csc->cpLimit; + csc->dir=dir; + } else { + /* continue current iteration direction */ + dir=csc->dir; + } + + if(dir<0) { + if(csc->startindex) { + U8_PREV((const uint8_t *)csc->p, csc->start, csc->index, c); + return c; + } + } else { + if(csc->indexlimit) { + U8_NEXT((const uint8_t *)csc->p, csc->index, csc->limit, c); + return c; + } + } + return U_SENTINEL; +} + +/** + * caseLocale >= 0: Lowercases [srcStart..srcLimit[ but takes context [0..srcLength[ into account. + * caseLocale < 0: Case-folds [srcStart..srcLimit[. + */ +void toLower(int32_t caseLocale, uint32_t options, + const uint8_t *src, UCaseContext *csc, int32_t srcStart, int32_t srcLimit, + icu::ByteSink &sink, icu::Edits *edits, UErrorCode &errorCode) { + const int8_t *latinToLower; + if (caseLocale == UCASE_LOC_ROOT || + (caseLocale >= 0 ? + !(caseLocale == UCASE_LOC_TURKISH || caseLocale == UCASE_LOC_LITHUANIAN) : + (options & _FOLD_CASE_OPTIONS_MASK) == U_FOLD_CASE_DEFAULT)) { + latinToLower = LatinCase::TO_LOWER_NORMAL; + } else { + latinToLower = LatinCase::TO_LOWER_TR_LT; + } + const UTrie2 *trie = ucase_getTrie(); + int32_t prev = srcStart; + int32_t srcIndex = srcStart; + for (;;) { + // fast path for simple cases + int32_t cpStart; + UChar32 c; + for (;;) { + if (U_FAILURE(errorCode) || srcIndex >= srcLimit) { + c = U_SENTINEL; + break; + } + uint8_t lead = src[srcIndex++]; + if (lead <= 0x7f) { + int8_t d = latinToLower[lead]; + if (d == LatinCase::EXC) { + cpStart = srcIndex - 1; + c = lead; + break; + } + if (d == 0) { continue; } + ByteSinkUtil::appendUnchanged(src + prev, srcIndex - 1 - prev, + sink, options, edits, errorCode); + char ascii = (char)(lead + d); + sink.Append(&ascii, 1); + if (edits != nullptr) { + edits->addReplace(1, 1); + } + prev = srcIndex; + continue; + } else if (lead < 0xe3) { + uint8_t t; + if (0xc2 <= lead && lead <= 0xc5 && srcIndex < srcLimit && + (t = src[srcIndex] - 0x80) <= 0x3f) { + // U+0080..U+017F + ++srcIndex; + c = ((lead - 0xc0) << 6) | t; + int8_t d = latinToLower[c]; + if (d == LatinCase::EXC) { + cpStart = srcIndex - 2; + break; + } + if (d == 0) { continue; } + ByteSinkUtil::appendUnchanged(src + prev, srcIndex - 2 - prev, + sink, options, edits, errorCode); + ByteSinkUtil::appendTwoBytes(c + d, sink); + if (edits != nullptr) { + edits->addReplace(2, 2); + } + prev = srcIndex; + continue; + } + } else if ((lead <= 0xe9 || lead == 0xeb || lead == 0xec) && + (srcIndex + 2) <= srcLimit && + U8_IS_TRAIL(src[srcIndex]) && U8_IS_TRAIL(src[srcIndex + 1])) { + // most of CJK: no case mappings + srcIndex += 2; + continue; + } + cpStart = --srcIndex; + U8_NEXT(src, srcIndex, srcLimit, c); + if (c < 0) { + // ill-formed UTF-8 + continue; + } + uint16_t props = UTRIE2_GET16(trie, c); + if (UCASE_HAS_EXCEPTION(props)) { break; } + int32_t delta; + if (!UCASE_IS_UPPER_OR_TITLE(props) || (delta = UCASE_GET_DELTA(props)) == 0) { + continue; + } + ByteSinkUtil::appendUnchanged(src + prev, cpStart - prev, + sink, options, edits, errorCode); + ByteSinkUtil::appendCodePoint(srcIndex - cpStart, c + delta, sink, edits); + prev = srcIndex; + } + if (c < 0) { + break; + } + // slow path + const char16_t *s; + if (caseLocale >= 0) { + csc->cpStart = cpStart; + csc->cpLimit = srcIndex; + c = ucase_toFullLower(c, utf8_caseContextIterator, csc, &s, caseLocale); + } else { + c = ucase_toFullFolding(c, &s, options); + } + if (c >= 0) { + ByteSinkUtil::appendUnchanged(src + prev, cpStart - prev, + sink, options, edits, errorCode); + appendResult(srcIndex - cpStart, c, s, sink, options, edits, errorCode); + prev = srcIndex; + } + } + ByteSinkUtil::appendUnchanged(src + prev, srcIndex - prev, + sink, options, edits, errorCode); +} + +void toUpper(int32_t caseLocale, uint32_t options, + const uint8_t *src, UCaseContext *csc, int32_t srcLength, + icu::ByteSink &sink, icu::Edits *edits, UErrorCode &errorCode) { + const int8_t *latinToUpper; + if (caseLocale == UCASE_LOC_TURKISH) { + latinToUpper = LatinCase::TO_UPPER_TR; + } else { + latinToUpper = LatinCase::TO_UPPER_NORMAL; + } + const UTrie2 *trie = ucase_getTrie(); + int32_t prev = 0; + int32_t srcIndex = 0; + for (;;) { + // fast path for simple cases + int32_t cpStart; + UChar32 c; + for (;;) { + if (U_FAILURE(errorCode) || srcIndex >= srcLength) { + c = U_SENTINEL; + break; + } + uint8_t lead = src[srcIndex++]; + if (lead <= 0x7f) { + int8_t d = latinToUpper[lead]; + if (d == LatinCase::EXC) { + cpStart = srcIndex - 1; + c = lead; + break; + } + if (d == 0) { continue; } + ByteSinkUtil::appendUnchanged(src + prev, srcIndex - 1 - prev, + sink, options, edits, errorCode); + char ascii = (char)(lead + d); + sink.Append(&ascii, 1); + if (edits != nullptr) { + edits->addReplace(1, 1); + } + prev = srcIndex; + continue; + } else if (lead < 0xe3) { + uint8_t t; + if (0xc2 <= lead && lead <= 0xc5 && srcIndex < srcLength && + (t = src[srcIndex] - 0x80) <= 0x3f) { + // U+0080..U+017F + ++srcIndex; + c = ((lead - 0xc0) << 6) | t; + int8_t d = latinToUpper[c]; + if (d == LatinCase::EXC) { + cpStart = srcIndex - 2; + break; + } + if (d == 0) { continue; } + ByteSinkUtil::appendUnchanged(src + prev, srcIndex - 2 - prev, + sink, options, edits, errorCode); + ByteSinkUtil::appendTwoBytes(c + d, sink); + if (edits != nullptr) { + edits->addReplace(2, 2); + } + prev = srcIndex; + continue; + } + } else if ((lead <= 0xe9 || lead == 0xeb || lead == 0xec) && + (srcIndex + 2) <= srcLength && + U8_IS_TRAIL(src[srcIndex]) && U8_IS_TRAIL(src[srcIndex + 1])) { + // most of CJK: no case mappings + srcIndex += 2; + continue; + } + cpStart = --srcIndex; + U8_NEXT(src, srcIndex, srcLength, c); + if (c < 0) { + // ill-formed UTF-8 + continue; + } + uint16_t props = UTRIE2_GET16(trie, c); + if (UCASE_HAS_EXCEPTION(props)) { break; } + int32_t delta; + if (UCASE_GET_TYPE(props) != UCASE_LOWER || (delta = UCASE_GET_DELTA(props)) == 0) { + continue; + } + ByteSinkUtil::appendUnchanged(src + prev, cpStart - prev, + sink, options, edits, errorCode); + ByteSinkUtil::appendCodePoint(srcIndex - cpStart, c + delta, sink, edits); + prev = srcIndex; + } + if (c < 0) { + break; + } + // slow path + csc->cpStart = cpStart; + csc->cpLimit = srcIndex; + const char16_t *s; + c = ucase_toFullUpper(c, utf8_caseContextIterator, csc, &s, caseLocale); + if (c >= 0) { + ByteSinkUtil::appendUnchanged(src + prev, cpStart - prev, + sink, options, edits, errorCode); + appendResult(srcIndex - cpStart, c, s, sink, options, edits, errorCode); + prev = srcIndex; + } + } + ByteSinkUtil::appendUnchanged(src + prev, srcIndex - prev, + sink, options, edits, errorCode); +} + +} // namespace + +#if !UCONFIG_NO_BREAK_ITERATION + +namespace { + +constexpr uint8_t ACUTE_BYTE0 = u8"\u0301"[0]; + +constexpr uint8_t ACUTE_BYTE1 = u8"\u0301"[1]; + +/** + * Input: c is a letter I with or without acute accent. + * start is the index in src after c, and is less than segmentLimit. + * If a plain i/I is followed by a plain j/J, + * or an i/I with acute (precomposed or decomposed) is followed by a j/J with acute, + * then we output accordingly. + * + * @return the src index after the titlecased sequence, or the start index if no Dutch IJ + */ +int32_t maybeTitleDutchIJ(const uint8_t *src, UChar32 c, int32_t start, int32_t segmentLimit, + ByteSink &sink, uint32_t options, icu::Edits *edits, UErrorCode &errorCode) { + U_ASSERT(start < segmentLimit); + + int32_t index = start; + bool withAcute = false; + + // If the conditions are met, then the following variables tell us what to output. + int32_t unchanged1 = 0; // code units before the j, or the whole sequence (0..3) + bool doTitleJ = false; // true if the j needs to be titlecased + int32_t unchanged2 = 0; // after the j (0 or 1) + + // next character after the first letter + UChar32 c2; + c2 = src[index++]; + + // Is the first letter an i/I with accent? + if (c == u'I') { + if (c2 == ACUTE_BYTE0 && index < segmentLimit && src[index++] == ACUTE_BYTE1) { + withAcute = true; + unchanged1 = 2; // ACUTE is 2 code units in UTF-8 + if (index == segmentLimit) { return start; } + c2 = src[index++]; + } + } else { // Í + withAcute = true; + } + + // Is the next character a j/J? + if (c2 == u'j') { + doTitleJ = true; + } else if (c2 == u'J') { + ++unchanged1; + } else { + return start; + } + + // A plain i/I must be followed by a plain j/J. + // An i/I with acute must be followed by a j/J with acute. + if (withAcute) { + if ((index + 1) >= segmentLimit || src[index++] != ACUTE_BYTE0 || src[index++] != ACUTE_BYTE1) { + return start; + } + if (doTitleJ) { + unchanged2 = 2; // ACUTE is 2 code units in UTF-8 + } else { + unchanged1 = unchanged1 + 2; // ACUTE is 2 code units in UTF-8 + } + } + + // There must not be another combining mark. + if (index < segmentLimit) { + int32_t cp; + int32_t i = index; + U8_NEXT(src, i, segmentLimit, cp); + uint32_t typeMask = U_GET_GC_MASK(cp); + if ((typeMask & U_GC_M_MASK) != 0) { + return start; + } + } + + // Output the rest of the Dutch IJ. + ByteSinkUtil::appendUnchanged(src + start, unchanged1, sink, options, edits, errorCode); + start += unchanged1; + if (doTitleJ) { + ByteSinkUtil::appendCodePoint(1, u'J', sink, edits); + ++start; + } + ByteSinkUtil::appendUnchanged(src + start, unchanged2, sink, options, edits, errorCode); + + U_ASSERT(start + unchanged2 == index); + return index; +} + +} // namespace + +U_CFUNC void U_CALLCONV +ucasemap_internalUTF8ToTitle( + int32_t caseLocale, uint32_t options, BreakIterator *iter, + const uint8_t *src, int32_t srcLength, + ByteSink &sink, icu::Edits *edits, + UErrorCode &errorCode) { + if (!ustrcase_checkTitleAdjustmentOptions(options, errorCode)) { + return; + } + + /* set up local variables */ + UCaseContext csc=UCASECONTEXT_INITIALIZER; + csc.p=(void *)src; + csc.limit=srcLength; + int32_t prev=0; + UBool isFirstIndex=true; + + /* titlecasing loop */ + while(prevfirst(); + } else { + index=iter->next(); + } + if(index==UBRK_DONE || index>srcLength) { + index=srcLength; + } + + /* + * Segment [prev..index[ into 3 parts: + * a) skipped characters (copy as-is) [prev..titleStart[ + * b) first letter (titlecase) [titleStart..titleLimit[ + * c) subsequent characters (lowercase) [titleLimit..index[ + */ + if(prev=0) { + csc.cpStart=titleStart; + csc.cpLimit=titleLimit; + const char16_t *s; + c=ucase_toFullTitle(c, utf8_caseContextIterator, &csc, &s, caseLocale); + if (!appendResult(titleLimit-titleStart, c, s, sink, options, edits, errorCode)) { + return; + } + } else { + // Malformed UTF-8. + if (!ByteSinkUtil::appendUnchanged(src+titleStart, titleLimit-titleStart, + sink, options, edits, errorCode)) { + return; + } + } + + /* Special case Dutch IJ titlecasing */ + if (titleLimit < index && + caseLocale == UCASE_LOC_DUTCH) { + if (c < 0) { + c = ~c; + } + + if (c == u'I' || c == u'Í') { + titleLimit = maybeTitleDutchIJ(src, c, titleLimit, index, sink, options, edits, errorCode); + } + } + + /* lowercase [titleLimit..index[ */ + if(titleLimit 0) { + uint32_t upper = data & UPPER_MASK; + // Add a dialytika to this iota or ypsilon vowel + // if we removed a tonos from the previous vowel, + // and that previous vowel did not also have (or gain) a dialytika. + // Adding one only to the final vowel in a longer sequence + // (which does not occur in normal writing) would require lookahead. + // Set the same flag as for preserving an existing dialytika. + if ((data & HAS_VOWEL) != 0 && (state & AFTER_VOWEL_WITH_ACCENT) != 0 && + (upper == 0x399 || upper == 0x3A5)) { + data |= HAS_DIALYTIKA; + } + int32_t numYpogegrammeni = 0; // Map each one to a trailing, spacing, capital iota. + if ((data & HAS_YPOGEGRAMMENI) != 0) { + numYpogegrammeni = 1; + } + // Skip combining diacritics after this Greek letter. + int32_t nextNextIndex = nextIndex; + while (nextIndex < srcLength) { + UChar32 c2; + U8_NEXT(src, nextNextIndex, srcLength, c2); + uint32_t diacriticData = getDiacriticData(c2); + if (diacriticData != 0) { + data |= diacriticData; + if ((diacriticData & HAS_YPOGEGRAMMENI) != 0) { + ++numYpogegrammeni; + } + nextIndex = nextNextIndex; + } else { + break; // not a Greek diacritic + } + } + if ((data & HAS_VOWEL_AND_ACCENT_AND_DIALYTIKA) == HAS_VOWEL_AND_ACCENT) { + nextState |= AFTER_VOWEL_WITH_ACCENT; + } + // Map according to Greek rules. + UBool addTonos = false; + if (upper == 0x397 && + (data & HAS_ACCENT) != 0 && + numYpogegrammeni == 0 && + (state & AFTER_CASED) == 0 && + !isFollowedByCasedLetter(src, nextIndex, srcLength)) { + // Keep disjunctive "or" with (only) a tonos. + // We use the same "word boundary" conditions as for the Final_Sigma test. + if (i == nextIndex) { + upper = 0x389; // Preserve the precomposed form. + } else { + addTonos = true; + } + } else if ((data & HAS_DIALYTIKA) != 0) { + // Preserve a vowel with dialytika in precomposed form if it exists. + if (upper == 0x399) { + upper = 0x3AA; + data &= ~HAS_EITHER_DIALYTIKA; + } else if (upper == 0x3A5) { + upper = 0x3AB; + data &= ~HAS_EITHER_DIALYTIKA; + } + } + + UBool change; + if (edits == nullptr && (options & U_OMIT_UNCHANGED_TEXT) == 0) { + change = true; // common, simple usage + } else { + // Find out first whether we are changing the text. + U_ASSERT(0x370 <= upper && upper <= 0x3ff); // 2-byte UTF-8, main Greek block + change = (i + 2) > nextIndex || + src[i] != getTwoByteLead(upper) || src[i + 1] != getTwoByteTrail(upper) || + numYpogegrammeni > 0; + int32_t i2 = i + 2; + if ((data & HAS_EITHER_DIALYTIKA) != 0) { + change |= (i2 + 2) > nextIndex || + src[i2] != (uint8_t)u8"\u0308"[0] || + src[i2 + 1] != (uint8_t)u8"\u0308"[1]; + i2 += 2; + } + if (addTonos) { + change |= (i2 + 2) > nextIndex || + src[i2] != (uint8_t)u8"\u0301"[0] || + src[i2 + 1] != (uint8_t)u8"\u0301"[1]; + i2 += 2; + } + int32_t oldLength = nextIndex - i; + int32_t newLength = (i2 - i) + numYpogegrammeni * 2; // 2 bytes per U+0399 + change |= oldLength != newLength; + if (change) { + if (edits != nullptr) { + edits->addReplace(oldLength, newLength); + } + } else { + if (edits != nullptr) { + edits->addUnchanged(oldLength); + } + // Write unchanged text? + change = (options & U_OMIT_UNCHANGED_TEXT) == 0; + } + } + + if (change) { + ByteSinkUtil::appendTwoBytes(upper, sink); + if ((data & HAS_EITHER_DIALYTIKA) != 0) { + sink.AppendU8(u8"\u0308", 2); // restore or add a dialytika + } + if (addTonos) { + sink.AppendU8(u8"\u0301", 2); + } + while (numYpogegrammeni > 0) { + sink.AppendU8(u8"\u0399", 2); + --numYpogegrammeni; + } + } + } else if(c>=0) { + const char16_t *s; + c=ucase_toFullUpper(c, nullptr, nullptr, &s, UCASE_LOC_GREEK); + if (!appendResult(nextIndex - i, c, s, sink, options, edits, errorCode)) { + return; + } + } else { + // Malformed UTF-8. + if (!ByteSinkUtil::appendUnchanged(src+i, nextIndex-i, + sink, options, edits, errorCode)) { + return; + } + } + i = nextIndex; + state = nextState; + } +} + +} // namespace GreekUpper +U_NAMESPACE_END + +static void U_CALLCONV +ucasemap_internalUTF8ToLower(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED + const uint8_t *src, int32_t srcLength, + icu::ByteSink &sink, icu::Edits *edits, + UErrorCode &errorCode) { + UCaseContext csc=UCASECONTEXT_INITIALIZER; + csc.p=(void *)src; + csc.limit=srcLength; + toLower( + caseLocale, options, + src, &csc, 0, srcLength, + sink, edits, errorCode); +} + +static void U_CALLCONV +ucasemap_internalUTF8ToUpper(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED + const uint8_t *src, int32_t srcLength, + icu::ByteSink &sink, icu::Edits *edits, + UErrorCode &errorCode) { + if (caseLocale == UCASE_LOC_GREEK) { + GreekUpper::toUpper(options, src, srcLength, sink, edits, errorCode); + } else { + UCaseContext csc=UCASECONTEXT_INITIALIZER; + csc.p=(void *)src; + csc.limit=srcLength; + toUpper( + caseLocale, options, + src, &csc, srcLength, + sink, edits, errorCode); + } +} + +static void U_CALLCONV +ucasemap_internalUTF8Fold(int32_t /* caseLocale */, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED + const uint8_t *src, int32_t srcLength, + icu::ByteSink &sink, icu::Edits *edits, + UErrorCode &errorCode) { + toLower( + -1, options, + src, nullptr, 0, srcLength, + sink, edits, errorCode); +} + +void +ucasemap_mapUTF8(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM + const char *src, int32_t srcLength, + UTF8CaseMapper *stringCaseMapper, + icu::ByteSink &sink, icu::Edits *edits, + UErrorCode &errorCode) { + /* check argument values */ + if (U_FAILURE(errorCode)) { + return; + } + if ((src == nullptr && srcLength != 0) || srcLength < -1) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + // Get the string length. + if (srcLength == -1) { + srcLength = (int32_t)uprv_strlen((const char *)src); + } + + if (edits != nullptr && (options & U_EDITS_NO_RESET) == 0) { + edits->reset(); + } + stringCaseMapper(caseLocale, options, UCASEMAP_BREAK_ITERATOR + (const uint8_t *)src, srcLength, sink, edits, errorCode); + sink.Flush(); + if (U_SUCCESS(errorCode)) { + if (edits != nullptr) { + edits->copyErrorTo(errorCode); + } + } +} + +int32_t +ucasemap_mapUTF8(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM + char *dest, int32_t destCapacity, + const char *src, int32_t srcLength, + UTF8CaseMapper *stringCaseMapper, + icu::Edits *edits, + UErrorCode &errorCode) { + /* check argument values */ + if(U_FAILURE(errorCode)) { + return 0; + } + if( destCapacity<0 || + (dest==nullptr && destCapacity>0) || + (src==nullptr && srcLength!=0) || srcLength<-1 + ) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* get the string length */ + if(srcLength==-1) { + srcLength=(int32_t)uprv_strlen((const char *)src); + } + + /* check for overlapping source and destination */ + if( dest!=nullptr && + ((src>=dest && src<(dest+destCapacity)) || + (dest>=src && dest<(src+srcLength))) + ) { + errorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + CheckedArrayByteSink sink(dest, destCapacity); + if (edits != nullptr && (options & U_EDITS_NO_RESET) == 0) { + edits->reset(); + } + stringCaseMapper(caseLocale, options, UCASEMAP_BREAK_ITERATOR + (const uint8_t *)src, srcLength, sink, edits, errorCode); + sink.Flush(); + if (U_SUCCESS(errorCode)) { + if (sink.Overflowed()) { + errorCode = U_BUFFER_OVERFLOW_ERROR; + } else if (edits != nullptr) { + edits->copyErrorTo(errorCode); + } + } + return u_terminateChars(dest, destCapacity, sink.NumberOfBytesAppended(), &errorCode); +} + +/* public API functions */ + +U_CAPI int32_t U_EXPORT2 +ucasemap_utf8ToLower(const UCaseMap *csm, + char *dest, int32_t destCapacity, + const char *src, int32_t srcLength, + UErrorCode *pErrorCode) { + return ucasemap_mapUTF8( + csm->caseLocale, csm->options, UCASEMAP_BREAK_ITERATOR_NULL + dest, destCapacity, + src, srcLength, + ucasemap_internalUTF8ToLower, nullptr, *pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +ucasemap_utf8ToUpper(const UCaseMap *csm, + char *dest, int32_t destCapacity, + const char *src, int32_t srcLength, + UErrorCode *pErrorCode) { + return ucasemap_mapUTF8( + csm->caseLocale, csm->options, UCASEMAP_BREAK_ITERATOR_NULL + dest, destCapacity, + src, srcLength, + ucasemap_internalUTF8ToUpper, nullptr, *pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +ucasemap_utf8FoldCase(const UCaseMap *csm, + char *dest, int32_t destCapacity, + const char *src, int32_t srcLength, + UErrorCode *pErrorCode) { + return ucasemap_mapUTF8( + UCASE_LOC_ROOT, csm->options, UCASEMAP_BREAK_ITERATOR_NULL + dest, destCapacity, + src, srcLength, + ucasemap_internalUTF8Fold, nullptr, *pErrorCode); +} + +U_NAMESPACE_BEGIN + +void CaseMap::utf8ToLower( + const char *locale, uint32_t options, + StringPiece src, ByteSink &sink, Edits *edits, + UErrorCode &errorCode) { + ucasemap_mapUTF8( + ustrcase_getCaseLocale(locale), options, UCASEMAP_BREAK_ITERATOR_NULL + src.data(), src.length(), + ucasemap_internalUTF8ToLower, sink, edits, errorCode); +} + +void CaseMap::utf8ToUpper( + const char *locale, uint32_t options, + StringPiece src, ByteSink &sink, Edits *edits, + UErrorCode &errorCode) { + ucasemap_mapUTF8( + ustrcase_getCaseLocale(locale), options, UCASEMAP_BREAK_ITERATOR_NULL + src.data(), src.length(), + ucasemap_internalUTF8ToUpper, sink, edits, errorCode); +} + +void CaseMap::utf8Fold( + uint32_t options, + StringPiece src, ByteSink &sink, Edits *edits, + UErrorCode &errorCode) { + ucasemap_mapUTF8( + UCASE_LOC_ROOT, options, UCASEMAP_BREAK_ITERATOR_NULL + src.data(), src.length(), + ucasemap_internalUTF8Fold, sink, edits, errorCode); +} + +int32_t CaseMap::utf8ToLower( + const char *locale, uint32_t options, + const char *src, int32_t srcLength, + char *dest, int32_t destCapacity, Edits *edits, + UErrorCode &errorCode) { + return ucasemap_mapUTF8( + ustrcase_getCaseLocale(locale), options, UCASEMAP_BREAK_ITERATOR_NULL + dest, destCapacity, + src, srcLength, + ucasemap_internalUTF8ToLower, edits, errorCode); +} + +int32_t CaseMap::utf8ToUpper( + const char *locale, uint32_t options, + const char *src, int32_t srcLength, + char *dest, int32_t destCapacity, Edits *edits, + UErrorCode &errorCode) { + return ucasemap_mapUTF8( + ustrcase_getCaseLocale(locale), options, UCASEMAP_BREAK_ITERATOR_NULL + dest, destCapacity, + src, srcLength, + ucasemap_internalUTF8ToUpper, edits, errorCode); +} + +int32_t CaseMap::utf8Fold( + uint32_t options, + const char *src, int32_t srcLength, + char *dest, int32_t destCapacity, Edits *edits, + UErrorCode &errorCode) { + return ucasemap_mapUTF8( + UCASE_LOC_ROOT, options, UCASEMAP_BREAK_ITERATOR_NULL + dest, destCapacity, + src, srcLength, + ucasemap_internalUTF8Fold, edits, errorCode); +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/ucasemap_imp.h b/intl/icu/source/common/ucasemap_imp.h new file mode 100644 index 0000000000..71d0e9033f --- /dev/null +++ b/intl/icu/source/common/ucasemap_imp.h @@ -0,0 +1,282 @@ +// © 2017 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// ucasemap_imp.h +// created: 2017feb08 Markus W. Scherer + +#ifndef __UCASEMAP_IMP_H__ +#define __UCASEMAP_IMP_H__ + +#include "unicode/utypes.h" +#include "unicode/ucasemap.h" +#include "unicode/uchar.h" +#include "ucase.h" + +/** + * Bit mask for the titlecasing iterator options bit field. + * Currently only 3 out of 8 values are used: + * 0 (words), U_TITLECASE_WHOLE_STRING, U_TITLECASE_SENTENCES. + * See stringoptions.h. + * @internal + */ +#define U_TITLECASE_ITERATOR_MASK 0xe0 + +/** + * Bit mask for the titlecasing index adjustment options bit set. + * Currently two bits are defined: + * U_TITLECASE_NO_BREAK_ADJUSTMENT, U_TITLECASE_ADJUST_TO_CASED. + * See stringoptions.h. + * @internal + */ +#define U_TITLECASE_ADJUSTMENT_MASK 0x600 + +/** + * Internal API, used by u_strcasecmp() etc. + * Compare strings case-insensitively, + * in code point order or code unit order. + */ +U_CFUNC int32_t +u_strcmpFold(const UChar *s1, int32_t length1, + const UChar *s2, int32_t length2, + uint32_t options, + UErrorCode *pErrorCode); + +/** + * Internal API, used for detecting length of + * shared prefix case-insensitively. + * @param s1 input string 1 + * @param length1 length of string 1, or -1 (NULL terminated) + * @param s2 input string 2 + * @param length2 length of string 2, or -1 (NULL terminated) + * @param options compare options + * @param matchLen1 (output) length of partial prefix match in s1 + * @param matchLen2 (output) length of partial prefix match in s2 + * @param pErrorCode receives error status + */ +U_CAPI void +u_caseInsensitivePrefixMatch(const UChar *s1, int32_t length1, + const UChar *s2, int32_t length2, + uint32_t options, + int32_t *matchLen1, int32_t *matchLen2, + UErrorCode *pErrorCode); + +#ifdef __cplusplus + +U_NAMESPACE_BEGIN + +class BreakIterator; // unicode/brkiter.h +class ByteSink; +class Locale; // unicode/locid.h + +/** Returns true if the options are valid. Otherwise false, and sets an error. */ +inline UBool ustrcase_checkTitleAdjustmentOptions(uint32_t options, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { return false; } + if ((options & U_TITLECASE_ADJUSTMENT_MASK) == U_TITLECASE_ADJUSTMENT_MASK) { + // Both options together. + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return false; + } + return true; +} + +inline UBool ustrcase_isLNS(UChar32 c) { + // Letter, number, symbol, + // or a private use code point because those are typically used as letters or numbers. + // Consider modifier letters only if they are cased. + const uint32_t LNS = (U_GC_L_MASK|U_GC_N_MASK|U_GC_S_MASK|U_GC_CO_MASK) & ~U_GC_LM_MASK; + int gc = u_charType(c); + return (U_MASK(gc) & LNS) != 0 || (gc == U_MODIFIER_LETTER && ucase_getType(c) != UCASE_NONE); +} + +#if !UCONFIG_NO_BREAK_ITERATION + +/** Returns nullptr if error. Pass in either locale or locID, not both. */ +U_CFUNC +BreakIterator *ustrcase_getTitleBreakIterator( + const Locale *locale, const char *locID, uint32_t options, BreakIterator *iter, + LocalPointer &ownedIter, UErrorCode &errorCode); + +#endif + +U_NAMESPACE_END + +#include "unicode/unistr.h" // for UStringCaseMapper + +/* + * Internal string casing functions implementing + * ustring.h/ustrcase.cpp and UnicodeString case mapping functions. + */ + +struct UCaseMap : public icu::UMemory { + /** Implements most of ucasemap_open(). */ + UCaseMap(const char *localeID, uint32_t opts, UErrorCode *pErrorCode); + ~UCaseMap(); + +#if !UCONFIG_NO_BREAK_ITERATION + icu::BreakIterator *iter; /* We adopt the iterator, so we own it. */ +#endif + char locale[32]; + int32_t caseLocale; + uint32_t options; +}; + +#if UCONFIG_NO_BREAK_ITERATION +# define UCASEMAP_BREAK_ITERATOR_PARAM +# define UCASEMAP_BREAK_ITERATOR_UNUSED +# define UCASEMAP_BREAK_ITERATOR +# define UCASEMAP_BREAK_ITERATOR_NULL +#else +# define UCASEMAP_BREAK_ITERATOR_PARAM icu::BreakIterator *iter, +# define UCASEMAP_BREAK_ITERATOR_UNUSED icu::BreakIterator *, +# define UCASEMAP_BREAK_ITERATOR iter, +# define UCASEMAP_BREAK_ITERATOR_NULL NULL, +#endif + +U_CFUNC int32_t +ustrcase_getCaseLocale(const char *locale); + +// TODO: swap src / dest if approved for new public api +/** Implements UStringCaseMapper. */ +U_CFUNC int32_t U_CALLCONV +ustrcase_internalToLower(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM + char16_t *dest, int32_t destCapacity, + const char16_t *src, int32_t srcLength, + icu::Edits *edits, + UErrorCode &errorCode); + +/** Implements UStringCaseMapper. */ +U_CFUNC int32_t U_CALLCONV +ustrcase_internalToUpper(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM + char16_t *dest, int32_t destCapacity, + const char16_t *src, int32_t srcLength, + icu::Edits *edits, + UErrorCode &errorCode); + +#if !UCONFIG_NO_BREAK_ITERATION + +/** Implements UStringCaseMapper. */ +U_CFUNC int32_t U_CALLCONV +ustrcase_internalToTitle(int32_t caseLocale, uint32_t options, + icu::BreakIterator *iter, + char16_t *dest, int32_t destCapacity, + const char16_t *src, int32_t srcLength, + icu::Edits *edits, + UErrorCode &errorCode); + +#endif + +/** Implements UStringCaseMapper. */ +U_CFUNC int32_t U_CALLCONV +ustrcase_internalFold(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM + char16_t *dest, int32_t destCapacity, + const char16_t *src, int32_t srcLength, + icu::Edits *edits, + UErrorCode &errorCode); + +/** + * Common string case mapping implementation for ucasemap_toXyz() and UnicodeString::toXyz(). + * Implements argument checking. + */ +U_CFUNC int32_t +ustrcase_map(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM + char16_t *dest, int32_t destCapacity, + const char16_t *src, int32_t srcLength, + UStringCaseMapper *stringCaseMapper, + icu::Edits *edits, + UErrorCode &errorCode); + +/** + * Common string case mapping implementation for old-fashioned u_strToXyz() functions + * that allow the source string to overlap the destination buffer. + * Implements argument checking and internally works with an intermediate buffer if necessary. + */ +U_CFUNC int32_t +ustrcase_mapWithOverlap(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM + char16_t *dest, int32_t destCapacity, + const char16_t *src, int32_t srcLength, + UStringCaseMapper *stringCaseMapper, + UErrorCode &errorCode); + +/** + * UTF-8 string case mapping function type, used by ucasemap_mapUTF8(). + * UTF-8 version of UStringCaseMapper. + * All error checking must be done. + * The UCaseMap must be fully initialized, with locale and/or iter set as needed. + */ +typedef void U_CALLCONV +UTF8CaseMapper(int32_t caseLocale, uint32_t options, +#if !UCONFIG_NO_BREAK_ITERATION + icu::BreakIterator *iter, +#endif + const uint8_t *src, int32_t srcLength, + icu::ByteSink &sink, icu::Edits *edits, + UErrorCode &errorCode); + +#if !UCONFIG_NO_BREAK_ITERATION + +/** Implements UTF8CaseMapper. */ +U_CFUNC void U_CALLCONV +ucasemap_internalUTF8ToTitle(int32_t caseLocale, uint32_t options, + icu::BreakIterator *iter, + const uint8_t *src, int32_t srcLength, + icu::ByteSink &sink, icu::Edits *edits, + UErrorCode &errorCode); + +#endif + +void +ucasemap_mapUTF8(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM + const char *src, int32_t srcLength, + UTF8CaseMapper *stringCaseMapper, + icu::ByteSink &sink, icu::Edits *edits, + UErrorCode &errorCode); + +/** + * Implements argument checking and buffer handling + * for UTF-8 string case mapping as a common function. + */ +int32_t +ucasemap_mapUTF8(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM + char *dest, int32_t destCapacity, + const char *src, int32_t srcLength, + UTF8CaseMapper *stringCaseMapper, + icu::Edits *edits, + UErrorCode &errorCode); + +U_NAMESPACE_BEGIN +namespace GreekUpper { + +// Data bits. +static const uint32_t UPPER_MASK = 0x3ff; +static const uint32_t HAS_VOWEL = 0x1000; +static const uint32_t HAS_YPOGEGRAMMENI = 0x2000; +static const uint32_t HAS_ACCENT = 0x4000; +static const uint32_t HAS_DIALYTIKA = 0x8000; +// Further bits during data building and processing, not stored in the data map. +static const uint32_t HAS_COMBINING_DIALYTIKA = 0x10000; +static const uint32_t HAS_OTHER_GREEK_DIACRITIC = 0x20000; + +static const uint32_t HAS_VOWEL_AND_ACCENT = HAS_VOWEL | HAS_ACCENT; +static const uint32_t HAS_VOWEL_AND_ACCENT_AND_DIALYTIKA = + HAS_VOWEL_AND_ACCENT | HAS_DIALYTIKA; +static const uint32_t HAS_EITHER_DIALYTIKA = HAS_DIALYTIKA | HAS_COMBINING_DIALYTIKA; + +// State bits. +static const uint32_t AFTER_CASED = 1; +static const uint32_t AFTER_VOWEL_WITH_ACCENT = 2; + +uint32_t getLetterData(UChar32 c); + +/** + * Returns a non-zero value for each of the Greek combining diacritics + * listed in The Unicode Standard, version 8, chapter 7.2 Greek, + * plus some perispomeni look-alikes. + */ +uint32_t getDiacriticData(UChar32 c); + +} // namespace GreekUpper +U_NAMESPACE_END + +#endif // __cplusplus + +#endif // __UCASEMAP_IMP_H__ diff --git a/intl/icu/source/common/ucasemap_titlecase_brkiter.cpp b/intl/icu/source/common/ucasemap_titlecase_brkiter.cpp new file mode 100644 index 0000000000..c2b44a2c78 --- /dev/null +++ b/intl/icu/source/common/ucasemap_titlecase_brkiter.cpp @@ -0,0 +1,134 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2011, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: ucasemap_titlecase_brkiter.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2011jun02 +* created by: Markus W. Scherer +* +* Titlecasing functions that are based on BreakIterator +* were moved here to break dependency cycles among parts of the common library. +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/brkiter.h" +#include "unicode/ubrk.h" +#include "unicode/casemap.h" +#include "unicode/ucasemap.h" +#include "cmemory.h" +#include "ucase.h" +#include "ucasemap_imp.h" + +U_NAMESPACE_BEGIN + +void CaseMap::utf8ToTitle( + const char *locale, uint32_t options, BreakIterator *iter, + StringPiece src, ByteSink &sink, Edits *edits, + UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { + return; + } + UText utext = UTEXT_INITIALIZER; + utext_openUTF8(&utext, src.data(), src.length(), &errorCode); + LocalPointer ownedIter; + iter = ustrcase_getTitleBreakIterator(nullptr, locale, options, iter, ownedIter, errorCode); + if (iter == nullptr) { + utext_close(&utext); + return; + } + iter->setText(&utext, errorCode); + ucasemap_mapUTF8( + ustrcase_getCaseLocale(locale), options, iter, + src.data(), src.length(), + ucasemap_internalUTF8ToTitle, sink, edits, errorCode); + utext_close(&utext); +} + +int32_t CaseMap::utf8ToTitle( + const char *locale, uint32_t options, BreakIterator *iter, + const char *src, int32_t srcLength, + char *dest, int32_t destCapacity, Edits *edits, + UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { + return 0; + } + UText utext=UTEXT_INITIALIZER; + utext_openUTF8(&utext, src, srcLength, &errorCode); + LocalPointer ownedIter; + iter = ustrcase_getTitleBreakIterator(nullptr, locale, options, iter, ownedIter, errorCode); + if(iter==nullptr) { + utext_close(&utext); + return 0; + } + iter->setText(&utext, errorCode); + int32_t length=ucasemap_mapUTF8( + ustrcase_getCaseLocale(locale), options, iter, + dest, destCapacity, + src, srcLength, + ucasemap_internalUTF8ToTitle, edits, errorCode); + utext_close(&utext); + return length; +} + +U_NAMESPACE_END + +U_NAMESPACE_USE + +U_CAPI const UBreakIterator * U_EXPORT2 +ucasemap_getBreakIterator(const UCaseMap *csm) { + return reinterpret_cast(csm->iter); +} + +U_CAPI void U_EXPORT2 +ucasemap_setBreakIterator(UCaseMap *csm, UBreakIterator *iterToAdopt, UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { + return; + } + delete csm->iter; + csm->iter=reinterpret_cast(iterToAdopt); +} + +U_CAPI int32_t U_EXPORT2 +ucasemap_utf8ToTitle(UCaseMap *csm, + char *dest, int32_t destCapacity, + const char *src, int32_t srcLength, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return 0; + } + UText utext=UTEXT_INITIALIZER; + utext_openUTF8(&utext, (const char *)src, srcLength, pErrorCode); + if (U_FAILURE(*pErrorCode)) { + return 0; + } + if(csm->iter==nullptr) { + LocalPointer ownedIter; + BreakIterator *iter = ustrcase_getTitleBreakIterator( + nullptr, csm->locale, csm->options, nullptr, ownedIter, *pErrorCode); + if (iter == nullptr) { + utext_close(&utext); + return 0; + } + csm->iter = ownedIter.orphan(); + } + csm->iter->setText(&utext, *pErrorCode); + int32_t length=ucasemap_mapUTF8( + csm->caseLocale, csm->options, csm->iter, + dest, destCapacity, + src, srcLength, + ucasemap_internalUTF8ToTitle, nullptr, *pErrorCode); + utext_close(&utext); + return length; +} + +#endif // !UCONFIG_NO_BREAK_ITERATION diff --git a/intl/icu/source/common/ucat.cpp b/intl/icu/source/common/ucat.cpp new file mode 100644 index 0000000000..2f7fdcd980 --- /dev/null +++ b/intl/icu/source/common/ucat.cpp @@ -0,0 +1,78 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2003, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* Author: Alan Liu +* Created: March 19 2003 +* Since: ICU 2.6 +********************************************************************** +*/ +#include "unicode/ucat.h" +#include "unicode/ustring.h" +#include "cstring.h" +#include "uassert.h" + +/* Separator between set_num and msg_num */ +static const char SEPARATOR = '%'; + +/* Maximum length of a set_num/msg_num key, incl. terminating zero. + * Longest possible key is "-2147483648%-2147483648" */ +#define MAX_KEY_LEN (24) + +/** + * Fill in buffer with a set_num/msg_num key string, given the numeric + * values. Numeric values must be >= 0. Buffer must be of length + * MAX_KEY_LEN or more. + */ +static char* +_catkey(char* buffer, int32_t set_num, int32_t msg_num) { + int32_t i = 0; + i = T_CString_integerToString(buffer, set_num, 10); + buffer[i++] = SEPARATOR; + T_CString_integerToString(buffer+i, msg_num, 10); + return buffer; +} + +U_CAPI u_nl_catd U_EXPORT2 +u_catopen(const char* name, const char* locale, UErrorCode* ec) { + return (u_nl_catd) ures_open(name, locale, ec); +} + +U_CAPI void U_EXPORT2 +u_catclose(u_nl_catd catd) { + ures_close((UResourceBundle*) catd); /* may be nullptr */ +} + +U_CAPI const char16_t* U_EXPORT2 +u_catgets(u_nl_catd catd, int32_t set_num, int32_t msg_num, + const char16_t* s, + int32_t* len, UErrorCode* ec) { + + char key[MAX_KEY_LEN]; + const char16_t* result; + + if (ec == nullptr || U_FAILURE(*ec)) { + goto ERROR; + } + + result = ures_getStringByKey((const UResourceBundle*) catd, + _catkey(key, set_num, msg_num), + len, ec); + if (U_FAILURE(*ec)) { + goto ERROR; + } + + return result; + + ERROR: + /* In case of any failure, return s */ + if (len != nullptr) { + *len = u_strlen(s); + } + return s; +} + +/*eof*/ diff --git a/intl/icu/source/common/uchar.cpp b/intl/icu/source/common/uchar.cpp new file mode 100644 index 0000000000..ff12962baa --- /dev/null +++ b/intl/icu/source/common/uchar.cpp @@ -0,0 +1,708 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************** +* Copyright (C) 1996-2016, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************** +* +* File UCHAR.C +* +* Modification History: +* +* Date Name Description +* 04/02/97 aliu Creation. +* 4/15/99 Madhu Updated all the function definitions for C Implementation +* 5/20/99 Madhu Added the function u_getVersion() +* 8/19/1999 srl Upgraded scripts to Unicode3.0 +* 11/11/1999 weiv added u_isalnum(), cleaned comments +* 01/11/2000 helena Renamed u_getVersion to u_getUnicodeVersion. +* 06/20/2000 helena OS/400 port changes; mostly typecast. +****************************************************************************** +*/ + +#include "unicode/utypes.h" +#include "unicode/uchar.h" +#include "unicode/uscript.h" +#include "unicode/udata.h" +#include "uassert.h" +#include "cmemory.h" +#include "ucln_cmn.h" +#include "utrie2.h" +#include "udataswp.h" +#include "uprops.h" +#include "ustr_imp.h" + +/* uchar_props_data.h is machine-generated by genprops --csource */ +#define INCLUDED_FROM_UCHAR_C +#include "uchar_props_data.h" + +/* constants and macros for access to the data ------------------------------ */ + +/* getting a uint32_t properties word from the data */ +#define GET_PROPS(c, result) ((result)=UTRIE2_GET16(&propsTrie, c)) + +/* API functions ------------------------------------------------------------ */ + +/* Gets the Unicode character's general category.*/ +U_CAPI int8_t U_EXPORT2 +u_charType(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (int8_t)GET_CATEGORY(props); +} + +/* Enumerate all code points with their general categories. */ +struct _EnumTypeCallback { + UCharEnumTypeRange *enumRange; + const void *context; +}; + +static uint32_t U_CALLCONV +_enumTypeValue(const void *context, uint32_t value) { + (void)context; + return GET_CATEGORY(value); +} + +static UBool U_CALLCONV +_enumTypeRange(const void *context, UChar32 start, UChar32 end, uint32_t value) { + /* just cast the value to UCharCategory */ + return ((struct _EnumTypeCallback *)context)-> + enumRange(((struct _EnumTypeCallback *)context)->context, + start, end+1, (UCharCategory)value); +} + +U_CAPI void U_EXPORT2 +u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context) { + struct _EnumTypeCallback callback; + + if(enumRange==nullptr) { + return; + } + + callback.enumRange=enumRange; + callback.context=context; + utrie2_enum(&propsTrie, _enumTypeValue, _enumTypeRange, &callback); +} + +/* Checks if ch is a lower case letter.*/ +U_CAPI UBool U_EXPORT2 +u_islower(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)(GET_CATEGORY(props)==U_LOWERCASE_LETTER); +} + +/* Checks if ch is an upper case letter.*/ +U_CAPI UBool U_EXPORT2 +u_isupper(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)(GET_CATEGORY(props)==U_UPPERCASE_LETTER); +} + +/* Checks if ch is a title case letter; usually upper case letters.*/ +U_CAPI UBool U_EXPORT2 +u_istitle(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)(GET_CATEGORY(props)==U_TITLECASE_LETTER); +} + +/* Checks if ch is a decimal digit. */ +U_CAPI UBool U_EXPORT2 +u_isdigit(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)(GET_CATEGORY(props)==U_DECIMAL_DIGIT_NUMBER); +} + +U_CAPI UBool U_EXPORT2 +u_isxdigit(UChar32 c) { + uint32_t props; + + /* check ASCII and Fullwidth ASCII a-fA-F */ + if( + (c<=0x66 && c>=0x41 && (c<=0x46 || c>=0x61)) || + (c>=0xff21 && c<=0xff46 && (c<=0xff26 || c>=0xff41)) + ) { + return true; + } + + GET_PROPS(c, props); + return (UBool)(GET_CATEGORY(props)==U_DECIMAL_DIGIT_NUMBER); +} + +/* Checks if the Unicode character is a letter.*/ +U_CAPI UBool U_EXPORT2 +u_isalpha(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)((CAT_MASK(props)&U_GC_L_MASK)!=0); +} + +U_CAPI UBool U_EXPORT2 +u_isUAlphabetic(UChar32 c) { + return (u_getUnicodeProperties(c, 1)&U_MASK(UPROPS_ALPHABETIC))!=0; +} + +/* Checks if c is a letter or a decimal digit */ +U_CAPI UBool U_EXPORT2 +u_isalnum(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_ND_MASK))!=0); +} + +/** + * Checks if c is alphabetic, or a decimal digit; implements UCHAR_POSIX_ALNUM. + * @internal + */ +U_CFUNC UBool +u_isalnumPOSIX(UChar32 c) { + return (UBool)(u_isUAlphabetic(c) || u_isdigit(c)); +} + +/* Checks if ch is a unicode character with assigned character type.*/ +U_CAPI UBool U_EXPORT2 +u_isdefined(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)(GET_CATEGORY(props)!=0); +} + +/* Checks if the Unicode character is a base form character that can take a diacritic.*/ +U_CAPI UBool U_EXPORT2 +u_isbase(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_N_MASK|U_GC_MC_MASK|U_GC_ME_MASK))!=0); +} + +/* Checks if the Unicode character is a control character.*/ +U_CAPI UBool U_EXPORT2 +u_iscntrl(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)((CAT_MASK(props)&(U_GC_CC_MASK|U_GC_CF_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK))!=0); +} + +U_CAPI UBool U_EXPORT2 +u_isISOControl(UChar32 c) { + return (uint32_t)c<=0x9f && (c<=0x1f || c>=0x7f); +} + +/* Some control characters that are used as space. */ +#define IS_THAT_CONTROL_SPACE(c) \ + (c<=0x9f && ((c>=TAB && c<=CR) || (c>=0x1c && c <=0x1f) || c==0x85)) + +/* Java has decided that U+0085 New Line is not whitespace any more. */ +#define IS_THAT_ASCII_CONTROL_SPACE(c) \ + (c<=0x1f && c>=TAB && (c<=CR || c>=0x1c)) + +/* Checks if the Unicode character is a space character.*/ +U_CAPI UBool U_EXPORT2 +u_isspace(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)((CAT_MASK(props)&U_GC_Z_MASK)!=0 || IS_THAT_CONTROL_SPACE(c)); +} + +U_CAPI UBool U_EXPORT2 +u_isJavaSpaceChar(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)((CAT_MASK(props)&U_GC_Z_MASK)!=0); +} + +/* Checks if the Unicode character is a whitespace character.*/ +U_CAPI UBool U_EXPORT2 +u_isWhitespace(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)( + ((CAT_MASK(props)&U_GC_Z_MASK)!=0 && + c!=NBSP && c!=FIGURESP && c!=NNBSP) || /* exclude no-break spaces */ + IS_THAT_ASCII_CONTROL_SPACE(c) + ); +} + +U_CAPI UBool U_EXPORT2 +u_isblank(UChar32 c) { + if((uint32_t)c<=0x9f) { + return c==9 || c==0x20; /* TAB or SPACE */ + } else { + /* Zs */ + uint32_t props; + GET_PROPS(c, props); + return (UBool)(GET_CATEGORY(props)==U_SPACE_SEPARATOR); + } +} + +U_CAPI UBool U_EXPORT2 +u_isUWhiteSpace(UChar32 c) { + return (u_getUnicodeProperties(c, 1)&U_MASK(UPROPS_WHITE_SPACE))!=0; +} + +/* Checks if the Unicode character is printable.*/ +U_CAPI UBool U_EXPORT2 +u_isprint(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + /* comparing ==0 returns false for the categories mentioned */ + return (UBool)((CAT_MASK(props)&U_GC_C_MASK)==0); +} + +/** + * Checks if c is in \p{graph}\p{blank} - \p{cntrl}. + * Implements UCHAR_POSIX_PRINT. + * @internal + */ +U_CFUNC UBool +u_isprintPOSIX(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + /* + * The only cntrl character in graph+blank is TAB (in blank). + * Here we implement (blank-TAB)=Zs instead of calling u_isblank(). + */ + return (UBool)((GET_CATEGORY(props)==U_SPACE_SEPARATOR) || u_isgraphPOSIX(c)); +} + +U_CAPI UBool U_EXPORT2 +u_isgraph(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + /* comparing ==0 returns false for the categories mentioned */ + return (UBool)((CAT_MASK(props)& + (U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CS_MASK|U_GC_CN_MASK|U_GC_Z_MASK)) + ==0); +} + +/** + * Checks if c is in + * [^\p{space}\p{gc=Control}\p{gc=Surrogate}\p{gc=Unassigned}] + * with space=\p{Whitespace} and Control=Cc. + * Implements UCHAR_POSIX_GRAPH. + * @internal + */ +U_CFUNC UBool +u_isgraphPOSIX(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + /* \p{space}\p{gc=Control} == \p{gc=Z}\p{Control} */ + /* comparing ==0 returns false for the categories mentioned */ + return (UBool)((CAT_MASK(props)& + (U_GC_CC_MASK|U_GC_CS_MASK|U_GC_CN_MASK|U_GC_Z_MASK)) + ==0); +} + +U_CAPI UBool U_EXPORT2 +u_ispunct(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)((CAT_MASK(props)&U_GC_P_MASK)!=0); +} + +/*Checks if the Unicode character can be ignorable in a Java or Unicode identifier.*/ +U_CAPI UBool U_EXPORT2 +u_isIDIgnorable(UChar32 c) { + if(c<=0x9f) { + return u_isISOControl(c) && !IS_THAT_ASCII_CONTROL_SPACE(c); + } else { + uint32_t props; + GET_PROPS(c, props); + return (UBool)(GET_CATEGORY(props)==U_FORMAT_CHAR); + } +} + +/*Checks if the Unicode character can start a Java identifier.*/ +U_CAPI UBool U_EXPORT2 +u_isJavaIDStart(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_SC_MASK|U_GC_PC_MASK))!=0); +} + +/*Checks if the Unicode character can be a Java identifier part other than starting the + * identifier. + */ +U_CAPI UBool U_EXPORT2 +u_isJavaIDPart(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return (UBool)( + (CAT_MASK(props)& + (U_GC_ND_MASK|U_GC_NL_MASK| + U_GC_L_MASK| + U_GC_SC_MASK|U_GC_PC_MASK| + U_GC_MC_MASK|U_GC_MN_MASK) + )!=0 || + u_isIDIgnorable(c)); +} + +U_CAPI int32_t U_EXPORT2 +u_charDigitValue(UChar32 c) { + uint32_t props; + int32_t value; + GET_PROPS(c, props); + value=(int32_t)GET_NUMERIC_TYPE_VALUE(props)-UPROPS_NTV_DECIMAL_START; + if(value<=9) { + return value; + } else { + return -1; + } +} + +U_CAPI double U_EXPORT2 +u_getNumericValue(UChar32 c) { + uint32_t props; + int32_t ntv; + GET_PROPS(c, props); + ntv=(int32_t)GET_NUMERIC_TYPE_VALUE(props); + + if(ntv==UPROPS_NTV_NONE) { + return U_NO_NUMERIC_VALUE; + } else if(ntv>4)-12; + int32_t denominator=(ntv&0xf)+1; + return (double)numerator/denominator; + } else if(ntv>5)-14; + int32_t exp=(ntv&0x1f)+2; + numValue=mant; + + /* multiply by 10^exp without math.h */ + while(exp>=4) { + numValue*=10000.; + exp-=4; + } + switch(exp) { + case 3: + numValue*=1000.; + break; + case 2: + numValue*=100.; + break; + case 1: + numValue*=10.; + break; + case 0: + default: + break; + } + + return numValue; + } else if(ntv>2)-0xbf; + int32_t exp=(ntv&3)+1; + + switch(exp) { + case 4: + numValue*=60*60*60*60; + break; + case 3: + numValue*=60*60*60; + break; + case 2: + numValue*=60*60; + break; + case 1: + numValue*=60; + break; + case 0: + default: + break; + } + + return numValue; + } else if(ntv>2); + return (double)numerator/denominator; + } else if(ntv>2); + return (double)numerator/denominator; + } else { + /* reserved */ + return U_NO_NUMERIC_VALUE; + } +} + +U_CAPI int32_t U_EXPORT2 +u_digit(UChar32 ch, int8_t radix) { + int8_t value; + if((uint8_t)(radix-2)<=(36-2)) { + value=(int8_t)u_charDigitValue(ch); + if(value<0) { + /* ch is not a decimal digit, try latin letters */ + if(ch>=0x61 && ch<=0x7A) { + value=(int8_t)(ch-0x57); /* ch - 'a' + 10 */ + } else if(ch>=0x41 && ch<=0x5A) { + value=(int8_t)(ch-0x37); /* ch - 'A' + 10 */ + } else if(ch>=0xFF41 && ch<=0xFF5A) { + value=(int8_t)(ch-0xFF37); /* fullwidth ASCII a-z */ + } else if(ch>=0xFF21 && ch<=0xFF3A) { + value=(int8_t)(ch-0xFF17); /* fullwidth ASCII A-Z */ + } + } + } else { + value=-1; /* invalid radix */ + } + return (int8_t)((value(36-2) || (uint32_t)digit>=(uint32_t)radix) { + return 0; + } else if(digit<10) { + return (UChar32)(0x30+digit); + } else { + return (UChar32)((0x61-10)+digit); + } +} + +/* miscellaneous, and support for uprops.cpp -------------------------------- */ + +U_CAPI void U_EXPORT2 +u_getUnicodeVersion(UVersionInfo versionArray) { + if(versionArray!=nullptr) { + uprv_memcpy(versionArray, dataVersion, U_MAX_VERSION_LENGTH); + } +} + +U_CFUNC uint32_t +u_getMainProperties(UChar32 c) { + uint32_t props; + GET_PROPS(c, props); + return props; +} + +U_CFUNC uint32_t +u_getUnicodeProperties(UChar32 c, int32_t column) { + U_ASSERT(column>=0); + if(column>=propsVectorsColumns) { + return 0; + } else { + uint16_t vecIndex=UTRIE2_GET16(&propsVectorsTrie, c); + return propsVectors[vecIndex+column]; + } +} + +U_CFUNC int32_t +uprv_getMaxValues(int32_t column) { + switch(column) { + case 0: + return indexes[UPROPS_MAX_VALUES_INDEX]; + case 2: + return indexes[UPROPS_MAX_VALUES_2_INDEX]; + default: + return 0; + } +} + +U_CAPI void U_EXPORT2 +u_charAge(UChar32 c, UVersionInfo versionArray) { + if(versionArray!=nullptr) { + uint32_t version=u_getUnicodeProperties(c, 0)>>UPROPS_AGE_SHIFT; + versionArray[0]=(uint8_t)(version>>4); + versionArray[1]=(uint8_t)(version&0xf); + versionArray[2]=versionArray[3]=0; + } +} + +U_CAPI UScriptCode U_EXPORT2 +uscript_getScript(UChar32 c, UErrorCode *pErrorCode) { + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return USCRIPT_INVALID_CODE; + } + if((uint32_t)c>0x10ffff) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return USCRIPT_INVALID_CODE; + } + uint32_t scriptX=u_getUnicodeProperties(c, 0)&UPROPS_SCRIPT_X_MASK; + uint32_t codeOrIndex=uprops_mergeScriptCodeOrIndex(scriptX); + if(scriptX=UPROPS_SCRIPT_X_WITH_OTHER) { + scx=scriptExtensions+scx[1]; + } + uint32_t sc32=sc; + if(sc32>0x7fff) { + /* Guard against bogus input that would make us go past the Script_Extensions terminator. */ + return false; + } + while(sc32>*scx) { + ++scx; + } + return sc32==(*scx&0x7fff); +} + +U_CAPI int32_t U_EXPORT2 +uscript_getScriptExtensions(UChar32 c, + UScriptCode *scripts, int32_t capacity, + UErrorCode *pErrorCode) { + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + if(capacity<0 || (capacity>0 && scripts==nullptr)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + uint32_t scriptX=u_getUnicodeProperties(c, 0)&UPROPS_SCRIPT_X_MASK; + uint32_t codeOrIndex=uprops_mergeScriptCodeOrIndex(scriptX); + if(scriptX=UPROPS_SCRIPT_X_WITH_OTHER) { + scx=scriptExtensions+scx[1]; + } + int32_t length=0; + uint16_t sx; + do { + sx=*scx++; + if(lengthcapacity) { + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + return length; +} + +U_CAPI UBlockCode U_EXPORT2 +ublock_getCode(UChar32 c) { + return (UBlockCode)((u_getUnicodeProperties(c, 0)&UPROPS_BLOCK_MASK)>>UPROPS_BLOCK_SHIFT); +} + +/* property starts for UnicodeSet ------------------------------------------- */ + +static UBool U_CALLCONV +_enumPropertyStartsRange(const void *context, UChar32 start, UChar32 end, uint32_t value) { + /* add the start code point to the USet */ + const USetAdder *sa=(const USetAdder *)context; + sa->add(sa->set, start); + (void)end; + (void)value; + return true; +} + +#define USET_ADD_CP_AND_NEXT(sa, cp) sa->add(sa->set, cp); sa->add(sa->set, cp+1) + +U_CFUNC void U_EXPORT2 +uchar_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { + return; + } + + /* add the start code point of each same-value range of the main trie */ + utrie2_enum(&propsTrie, nullptr, _enumPropertyStartsRange, sa); + + /* add code points with hardcoded properties, plus the ones following them */ + + /* add for u_isblank() */ + USET_ADD_CP_AND_NEXT(sa, TAB); + + /* add for IS_THAT_CONTROL_SPACE() */ + sa->add(sa->set, CR+1); /* range TAB..CR */ + sa->add(sa->set, 0x1c); + sa->add(sa->set, 0x1f+1); + USET_ADD_CP_AND_NEXT(sa, 0x85); // NEXT LINE (NEL) + + /* add for u_isIDIgnorable() what was not added above */ + sa->add(sa->set, 0x7f); /* range DEL..NBSP-1, NBSP added below */ + sa->add(sa->set, HAIRSP); + sa->add(sa->set, RLM+1); + sa->add(sa->set, 0x206a); // INHIBIT SYMMETRIC SWAPPING + sa->add(sa->set, 0x206f+1); // NOMINAL DIGIT SHAPES + USET_ADD_CP_AND_NEXT(sa, ZWNBSP); + + /* add no-break spaces for u_isWhitespace() what was not added above */ + USET_ADD_CP_AND_NEXT(sa, NBSP); + USET_ADD_CP_AND_NEXT(sa, FIGURESP); + USET_ADD_CP_AND_NEXT(sa, NNBSP); + + /* add for u_digit() */ + sa->add(sa->set, u'a'); + sa->add(sa->set, u'z'+1); + sa->add(sa->set, u'A'); + sa->add(sa->set, u'Z'+1); + // fullwidth + sa->add(sa->set, u'ａ'); + sa->add(sa->set, u'ｚ'+1); + sa->add(sa->set, u'Ａ'); + sa->add(sa->set, u'Ｚ'+1); + + /* add for u_isxdigit() */ + sa->add(sa->set, u'f'+1); + sa->add(sa->set, u'F'+1); + // fullwidth + sa->add(sa->set, u'ｆ'+1); + sa->add(sa->set, u'Ｆ'+1); + + /* add for UCHAR_DEFAULT_IGNORABLE_CODE_POINT what was not added above */ + sa->add(sa->set, 0x2060); /* range 2060..206f */ + sa->add(sa->set, 0xfff0); + sa->add(sa->set, 0xfffb+1); + sa->add(sa->set, 0xe0000); + sa->add(sa->set, 0xe0fff+1); + + /* add for UCHAR_GRAPHEME_BASE and others */ + USET_ADD_CP_AND_NEXT(sa, CGJ); +} + +U_CFUNC void U_EXPORT2 +upropsvec_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { + return; + } + + /* add the start code point of each same-value range of the properties vectors trie */ + utrie2_enum(&propsVectorsTrie, nullptr, _enumPropertyStartsRange, sa); +} diff --git a/intl/icu/source/common/uchar_props_data.h b/intl/icu/source/common/uchar_props_data.h new file mode 100644 index 0000000000..f4d3932574 --- /dev/null +++ b/intl/icu/source/common/uchar_props_data.h @@ -0,0 +1,4011 @@ +// Copyright (C) 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// Copyright (C) 1999-2016, International Business Machines +// Corporation and others. All Rights Reserved. +// +// file name: uchar_props_data.h +// +// machine-generated by: icu/tools/unicode/c/genprops/corepropsbuilder.cpp + +#ifdef INCLUDED_FROM_UCHAR_C + +static const UVersionInfo dataVersion={0xf,0,0,0}; + +static const uint16_t propsTrie_index[23016]={ +0x495,0x49d,0x4a5,0x4ad,0x4c5,0x4cd,0x4d5,0x4dd,0x4e5,0x4ed,0x4f3,0x4fb,0x503,0x50b,0x513,0x51b, +0x521,0x529,0x531,0x539,0x53c,0x544,0x54c,0x554,0x55c,0x564,0x560,0x568,0x570,0x578,0x57d,0x585, +0x58d,0x595,0x599,0x5a1,0x5a9,0x5b1,0x5b9,0x5c1,0x5bd,0x5c5,0x5ca,0x5d2,0x5d8,0x5e0,0x5e8,0x5f0, +0x5f8,0x600,0x608,0x610,0x615,0x61d,0x620,0x628,0x630,0x638,0x63e,0x646,0x645,0x64d,0x655,0x65d, +0x66d,0x665,0x675,0x67d,0x683,0x600,0x693,0x68b,0x6a3,0x6a5,0x6ad,0x69b,0x6bd,0x6c3,0x6cb,0x6b5, +0x6db,0x6e1,0x6e9,0x6d3,0x6f9,0x6ff,0x707,0x6f1,0x717,0x71d,0x725,0x70f,0x735,0x73d,0x745,0x72d, +0x755,0x75b,0x763,0x74d,0x773,0x779,0x781,0x76b,0x791,0x796,0x79e,0x789,0x7ae,0x7b5,0x7bd,0x7a6, +0x641,0x7c5,0x7cd,0x4b5,0x7d5,0x7dc,0x7e4,0x4b5,0x7ec,0x7f4,0x7fc,0x801,0x809,0x810,0x818,0x4b5, +0x600,0x820,0x828,0x830,0x838,0x58d,0x848,0x840,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x850,0x600,0x858,0x85c,0x864,0x600,0x86a,0x600,0x870,0x878,0x880,0x58d,0x58d,0x888, +0x890,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x895,0x89d,0x600,0x600,0x8a5,0x8ad,0x8b5,0x8bd,0x8c5,0x600,0x8cd,0x8d5,0x8dd, +0x8ed,0x600,0x8f5,0x8f7,0x8ff,0x8e5,0x600,0x902,0x916,0x90a,0x912,0x91e,0x600,0x926,0x92c,0x934, +0x93c,0x600,0x94c,0x954,0x95c,0x944,0x96c,0x4b5,0x974,0x977,0x97f,0x964,0x98f,0x987,0x600,0x996, +0x600,0x9a5,0x99e,0x9ad,0x9b5,0x9b9,0x9c1,0x9c9,0x535,0x9d1,0x9d4,0x9da,0x9e1,0x9d4,0x55c,0x55c, +0x4e5,0x4e5,0x4e5,0x4e5,0x9e9,0x4e5,0x4e5,0x4e5,0x9f9,0xa01,0xa09,0xa11,0xa19,0xa1d,0xa25,0x9f1, +0xa3d,0xa45,0xa2d,0xa35,0xa4d,0xa55,0xa5d,0xa65,0xa7d,0xa6d,0xa75,0xa85,0xa8d,0xa9c,0xaa1,0xa94, +0xaa9,0xaa9,0xaa9,0xaa9,0xaa9,0xaa9,0xaa9,0xaa9,0xab1,0xab9,0x934,0xabc,0xac4,0xacb,0xad0,0xad8, +0x934,0xadf,0xade,0xaef,0xaf2,0x934,0x934,0xae7,0x934,0x934,0x934,0x934,0x934,0xb01,0xb09,0xaf9, +0x934,0x934,0x934,0xb0e,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0xb14,0xb1c,0x934,0xb24,0xb2b, +0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0xaa9,0xaa9,0xaa9,0xaa9,0xb33,0xaa9,0xb3a,0xb41, +0xaa9,0xaa9,0xaa9,0xaa9,0xaa9,0xaa9,0xaa9,0xaa9,0x934,0xb49,0xb50,0xb54,0xb5a,0x934,0x934,0x934, +0x58d,0x595,0x535,0xb62,0x4e5,0x4e5,0x4e5,0xb6a,0x535,0xb72,0x600,0xb78,0xb88,0xb80,0xb80,0x55c, +0xb90,0xb98,0xba0,0x4b5,0xba8,0x934,0x934,0xbaf,0x934,0x934,0x934,0x934,0x934,0x934,0xbb7,0xbbd, +0xbcd,0xbc5,0x641,0x600,0xbd5,0x890,0x600,0xbdd,0xbe5,0xbe9,0x600,0x600,0xbee,0x600,0x934,0xbf5, +0xad9,0xbfd,0xc03,0x934,0xbfd,0xc0b,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934, +0xc13,0x600,0x600,0x600,0xc1b,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0xc21,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xc26,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x934,0x934, +0xc2e,0x600,0xc31,0x600,0xc39,0xc3f,0xc47,0xc4f,0xc54,0x600,0x600,0xc58,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xc5f,0x600,0xc66,0xc6c,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xc74,0x600,0x600,0x600,0xc7c,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0xc7e,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xc85,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0xc8c,0x600,0x600,0x600,0xc93,0xc9b,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xca0,0x600,0x600,0xca8,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xcac,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xcaf,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xcb2,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0xcb8,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0xcc0,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0xcc5,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0xcca,0x600,0x600,0x600,0xccf,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0xcd7,0xcde,0xce2,0x600,0x600,0x600,0xce9,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0xcf7,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0xcef,0x934,0xcff,0x9ad,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0xd04,0xd0c,0x4e5,0xd1c,0xd14,0x600,0x600,0xd24,0xd2c,0xd3c,0x4e5,0xd41,0xd49,0xd4f,0xd56,0xd34, +0xd5e,0xd66,0x600,0xd6e,0xd7e,0xd81,0xd76,0xd89,0x655,0xd91,0xd98,0x8f6,0x6a3,0xda8,0xda0,0xdb0, +0x600,0xdb8,0xdc0,0xdc8,0x600,0xdd0,0xdd8,0xde0,0xde8,0xdf0,0xdf4,0xdfc,0x535,0x535,0x600,0xe04, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xe0c,0xe18,0xe10, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20, +0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0x600,0x600,0x600,0xe30,0x600,0xcea,0xe37,0xe3c, +0x600,0x600,0x600,0xe44,0x600,0x600,0x901,0x4b5,0xe5a,0xe4a,0xe52,0x600,0x600,0xe62,0xe6a,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xe6f,0x938,0x600,0xe77,0x600,0xe7d,0xe81, +0xe89,0xe91,0xe98,0xea0,0x600,0x600,0x600,0xea6,0xebe,0x4a5,0xec6,0xece,0xed3,0x916,0xeae,0xeb6, +0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20, +0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20,0xe20, +0x12d4,0x12d4,0x1314,0x1354,0x1394,0x13cc,0x140c,0x144c,0x1484,0x14c4,0x14f0,0x1530,0x1570,0x1580,0x15c0,0x15f4, +0x1634,0x1664,0x16a4,0x16e4,0x16f4,0x1728,0x1760,0x17a0,0x17e0,0x1820,0x1854,0x1880,0x18c0,0x18f8,0x1914,0x1954, +0xa80,0xac0,0xb00,0xb40,0xb80,0xbab,0xbeb,0xa40,0xc0e,0xa40,0xa40,0xa40,0xa40,0xc4e,0x1db,0x1db, +0xc8e,0xcce,0xa40,0xa40,0xa40,0xcf7,0xd37,0xd57,0xa40,0xd7d,0xdbd,0xdfd,0xe3d,0xe7d,0xebd,0xefd, +0xf3d,0xf74,0x1db,0x1db,0xf98,0xfcc,0x1db,0xff4,0x1db,0x1db,0x1db,0x1db,0x1021,0x1db,0x1db,0x1db, +0x1db,0x1db,0x1db,0x1db,0x1035,0x1db,0x106d,0x10ad,0x1db,0x10b8,0x1db,0x1db,0x1db,0x10ee,0xa40,0x112e, +0x1db,0x1db,0x116e,0x1db,0x1191,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0x11d1,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700, +0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x1211, +0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700, +0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x700,0x1211, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0xedb,0xee2,0xeea,0x4b5,0x600,0x600,0x600,0xef2,0xf02,0xefa,0xf19,0xf0a,0xf11,0xf21,0xf25,0xf29, +0x4b5,0x4b5,0x4b5,0x4b5,0x8f6,0x600,0xf31,0xf39,0x600,0xf41,0xf49,0xf4d,0xf55,0x600,0xf5d,0x4b5, +0x58d,0x597,0xf65,0x600,0xf69,0xf71,0xf81,0xf79,0x600,0xf89,0x600,0xf90,0xfa0,0xf98,0x4b5,0x4b5, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xb88,0x902,0xfa8,0xfb8,0xfb0,0x4b5,0x4b5, +0xfc8,0xfc0,0xfcb,0xfd3,0x916,0xfdb,0x4b5,0xfe3,0xfeb,0xff3,0x4b5,0x4b5,0x600,0x1003,0x100b,0xffb, +0x101b,0x1022,0x1013,0x102a,0x1032,0x4b5,0x1042,0x103a,0x600,0x1045,0x104d,0x1055,0x105d,0x1065,0x4b5,0x4b5, +0x600,0x600,0x106d,0x4b5,0x58d,0x1075,0x535,0x107d,0x600,0x1085,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x108d,0x600,0x1095,0x4b5,0x109a,0x10a2,0x10aa,0x10b1,0xfdf,0x10b9,0xfdf,0x10c1,0xb88, +0x10d1,0x636,0x10d9,0x10c9,0x98f,0x10e1,0x10e9,0x10ef,0x1107,0x10f7,0x10ff,0x110b,0x98f,0x111b,0x1113,0x1123, +0x113b,0x112b,0x1133,0x4b5,0x1142,0x114a,0x658,0x1152,0x1162,0x1168,0x1170,0x115a,0x4b5,0x4b5,0x4b5,0x4b5, +0x600,0x1178,0x1180,0x1099,0x600,0x1188,0x1190,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x600,0x1198,0x11a0,0x4b5, +0x600,0x11a8,0x11b0,0x11b8,0x600,0x11c8,0x11c0,0x4b5,0x870,0x11d0,0x11d8,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x600,0x11e0,0x4b5,0x4b5,0x4b5,0x58d,0x535,0x11e8,0x11f8,0x11fe,0x11f0,0x4b5,0x4b5,0x120e,0x1212,0x1206, +0x122a,0x121a,0x1222,0x600,0x1238,0x1232,0x600,0x8f7,0x1248,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x1256,0x125b,0x1240,0x1250,0x126b,0x1263,0x4b5,0x4b5,0x127a,0x127e,0x1272,0x128e,0x1286,0x11c0,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x1292,0x12a2,0x12a7,0x129a,0x4b5,0x4b5,0x12af,0x12bf,0x12b7, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x901,0x4b5,0x4b5,0x4b5, +0x12cf,0x12d7,0x12df,0x12c7,0x600,0x600,0x600,0x600,0x600,0x600,0x12e7,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0xfdf,0x600,0x600,0x12ef,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x12f7,0x12ff,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x11d8,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x8f7, +0x916,0xda4,0x600,0x916,0x1307,0x130c,0x600,0x131c,0x1324,0x132c,0x1314,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x58d,0x535,0x1334,0x4b5,0x4b5,0x4b5,0x600,0x600,0x133c,0x1341,0x1347,0x4b5,0x4b5,0x134f,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x1357,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x902,0x4b5,0x106d,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x135d,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x1365,0x136a,0x1371,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xe10,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x600,0x600,0x600,0x1377,0x137c,0x1384,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x55c,0x1394,0x139b,0x934,0x934,0x934,0x138c,0x4b5,0x934,0x934,0x934, +0x934,0x934,0x934,0x934,0xbb7,0x934,0x13a2,0x934,0x13a9,0x13b1,0x13b7,0x934,0xade,0x934,0x934,0x13bf, +0x4b5,0x4b5,0x4b5,0x13c7,0x13c7,0x934,0x934,0xadb,0x13cf,0x4b5,0x4b5,0x4b5,0x4b5,0x13df,0x13e6,0x13eb, +0x13f1,0x13f9,0x1401,0x1409,0x13e3,0x1411,0x1419,0x1421,0x1426,0x13f8,0x13df,0x13e6,0x13e2,0x13f1,0x142e,0x13e0, +0x1431,0x13e3,0x1439,0x1441,0x1449,0x1450,0x143c,0x1444,0x144c,0x1453,0x143f,0x145b,0x13d7,0x934,0x934,0x934, +0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x55c,0x146b,0x55c, +0x1472,0x1479,0x1463,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x1488,0x1490,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x1480,0x1498,0x9d4, +0x14a8,0x14a0,0x4b5,0x4b5,0x4b5,0x600,0x14b8,0x14b0,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0xfdf,0x14c0,0x600,0x14c8,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0xfdf,0x14d0,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x14d8,0x600,0x600,0x600, +0x600,0x600,0x600,0x14e0,0x4b5,0x58d,0x14f0,0x14e8,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x14f8,0x1508,0x1500,0x4b5,0x4b5,0x1518,0x1510,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x1528,0x1530,0x1538, +0x1540,0x1548,0x1550,0x4b5,0x1520,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x934,0x1558,0x934, +0x934,0xbaf,0x13a0,0x1560,0xbb7,0x1568,0x934,0x934,0x934,0x934,0xbb9,0x4b5,0x1570,0x1578,0x157c,0x1584, +0x158c,0x4b5,0x4b5,0x4b5,0x4b5,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x1594,0x934,0x934,0x934, +0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934, +0x934,0x934,0x934,0x157d,0x159c,0x934,0x934,0x934,0x15a4,0x934,0x934,0x15ab,0x15b3,0x1558,0x934,0x15bb, +0x934,0x15c3,0x15c8,0x4b5,0x4b5,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0x934,0xbaf, +0x15d0,0x15d9,0x15dd,0x15e5,0x15d5,0x934,0x934,0x934,0x934,0x15ed,0x934,0xade,0x11bc,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x15f5,0x600,0x600, +0x15fc,0x600,0x600,0x600,0x1604,0x600,0x160c,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xc90,0x600,0x600, +0x1614,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x161c,0x1624,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0xccf,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x162b,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x1632,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x1639,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x4b5,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x901,0x600,0x600,0x600,0x600,0x600,0x600,0xf69,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x1641,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x1649,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x600,0x600, +0x600,0x600,0x1651,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0xf69,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x67d,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x1314,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x1661,0x1659,0x1659,0x1659,0x4b5,0x4b5,0x4b5,0x4b5,0x55c,0x55c,0x55c,0x55c,0x55c,0x55c,0x55c, +0x1669,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5,0x4b5, +0x4b5,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0x1671,0x494,0x494,0x494,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf, +0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf, +0xf,0xf,0xf,0xf,0xc,0x17,0x17,0x17,0x19,0x17,0x17,0x17,0x14,0x15,0x17,0x18, +0x17,0x13,0x17,0x17,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x17,0x17, +0x18,0x18,0x18,0x17,0x17,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0x14, +0x17,0x15,0x1a,0x16,0x1a,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,0x14, +0x18,0x15,0x18,0xf,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf, +0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf, +0xf,0xf,0xf,0xf,0xc,0x17,0x19,0x19,0x19,0x19,0x1b,0x17,0x1a,0x1b,5,0x1c, +0x18,0x10,0x1b,0x1a,0x1b,0x18,0x34b,0x38b,0x1a,2,0x17,0x17,0x1a,0x30b,5,0x1d, +0x34cb,0x344b,0x3ccb,0x17,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,0x18,1,1,1,1, +1,1,1,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,0x18,2,2,2,2, +2,2,2,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,2,1,2,1, +2,1,2,1,2,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,1,2,1,2,1,2,2,2,1,1,2, +1,2,1,1,2,1,1,1,2,2,1,1,1,1,2,1, +1,2,1,1,1,2,2,2,1,1,2,1,1,2,1,2, +1,2,1,1,2,1,2,2,1,2,1,1,2,1,1,1, +2,1,2,1,1,2,2,5,1,2,2,2,5,5,5,5, +1,3,2,1,3,2,1,3,2,1,2,1,2,1,2,1, +2,1,2,1,2,1,2,1,2,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,2,1,3,2, +1,2,1,1,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,2,2,2,2,2,2,1,1, +2,1,1,2,2,1,2,1,1,1,1,2,1,2,1,2, +1,2,1,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,5,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4, +4,4,0x1a,0x1a,0x1a,0x1a,4,4,4,4,4,4,4,4,4,4, +4,4,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a, +4,4,4,4,4,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,4,0x1a,4,0x1a, +0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, +1,2,1,2,4,0x1a,1,2,0,0,4,2,2,2,0x17,1, +0,0,0,0,0x1a,0x1a,1,0x17,1,1,1,0,1,0,1,1, +2,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,0,1,1,1,1,1,1,1,1,1,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,1,2,2,1,1,1,2,2,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,2,2,2,2,1,2,0x18,1,2,1,1,2, +2,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,0x1b,6,6,6,6,6,7,7,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,1,2,1,2,1,2,1,2,1,2,1, +2,1,2,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,0,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0, +0,4,0x17,0x17,0x17,0x17,0x17,0x17,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,0x17,0x13,0,0,0x1b,0x1b,0x19, +0,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,0x13,6, +0x17,6,6,0x17,6,6,0x17,6,0,0,0,0,0,0,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,0,0,0,0,5, +5,5,5,0x17,0x17,0,0,0,0,0,0,0,0,0,0,0, +0x10,0x10,0x10,0x10,0x10,0x10,0x18,0x18,0x18,0x17,0x17,0x19,0x17,0x17,0x1b,0x1b, +6,6,6,6,6,6,6,6,6,6,6,0x17,0x10,0x17,0x17,0x17, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +4,5,5,5,5,5,5,5,5,5,5,6,6,6,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x17,0x17,0x17,0x17,5,5, +6,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0x17,5,6,6,6,6,6,6,6,0x10,0x1b,6, +6,6,6,6,6,4,4,6,6,0x1b,6,6,6,6,5,5, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,5,5,5,0x1b,0x1b,5, +0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0,0x10, +5,6,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,0,0,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,6,6,6,6,6,6, +6,6,6,6,6,5,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,6, +6,6,6,6,6,6,6,6,4,4,0x1b,0x17,0x17,0x17,4,0, +0,6,0x19,0x19,6,6,6,6,4,6,6,6,4,6,6,6, +6,6,0,0,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17, +0x17,0x17,0x17,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,6,6,6,6,4,6, +6,6,6,6,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,6,6,6, +0,0,0x17,0,5,5,5,5,5,5,5,5,5,5,5,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0x1a,5,5,5,5,5,5,0,0x10,0x10,0,0, +0,0,0,0,6,6,6,6,6,6,6,6,6,6,0x10,6, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,5,5,5,5, +5,5,5,5,5,4,6,6,6,6,6,6,6,6,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,5,5,6,6, +0x17,0x17,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x17,4,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,6,6,6,8, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,6,8, +6,5,8,8,8,6,6,6,6,6,6,6,6,8,8,8, +8,6,8,8,5,6,6,6,6,6,6,6,5,5,5,5, +5,5,5,5,5,5,6,6,0,0,0x49,0x89,0xc9,0x109,0x149,0x189, +0x1c9,0x209,0x249,0x289,5,5,0x19,0x19,0x37cb,0x35cb,0x3fcb,0x34cb,0x3ccb,0x94b,0x1b,0x19, +5,0x17,6,0,5,6,8,8,0,5,5,5,5,5,5,5, +5,0,0,5,5,0,0,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0,5,5,5,5,5,5,5,0,5,0, +0,0,5,5,5,5,0,0,6,5,8,8,8,6,6,6, +6,0,0,8,8,0,0,8,8,6,5,0,0,0,0,0, +0,0,0,8,0,0,0,0,5,5,0,5,0,0,0,0, +0,0,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,6,6,5,5, +5,6,0x17,0,0,0,0,0,0,0,0,0,0,6,6,8, +0,5,5,5,5,5,5,0,0,0,0,5,5,0,0,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,0,5,5, +5,5,5,5,5,0,5,5,0,5,5,0,5,5,0,0, +6,0,8,8,8,6,6,0,0,0,0,6,6,0,0,6, +6,6,0,0,0,6,0,0,0,0,0,0,0,5,5,5, +5,0,5,0,5,5,6,6,0,0,0x49,0x89,0xc9,0x109,0x149,0x189, +0x1c9,0x209,0x249,0x289,0x17,0x19,0,0,0,0,0,0,0,5,6,6, +6,6,6,6,0,6,6,8,0,5,5,5,5,5,5,5, +5,5,0,5,5,5,0,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0,5,5,5,5,5,5,5,0,5,5, +0,5,5,5,5,5,0,0,6,5,8,8,8,6,6,6, +6,6,0,6,6,8,0,8,8,6,0,0,5,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,5,5,6,6, +0,0,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x1b,5,0x34cb,0x344b, +0x3ccb,0x37cb,0x35cb,0x3fcb,0,0,0,0,0,0,0,0,0,6,8,8, +0,5,5,5,5,5,5,5,5,0,0,5,5,0,0,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,0,5,5, +5,5,5,5,5,0,5,5,0,5,5,5,5,5,0,0, +6,5,8,6,8,6,6,6,6,0,0,8,8,0,0,8, +8,6,0,0,0,0,0,0,0,6,6,8,0,0,0,0, +5,5,0,5,0,0,0,0,0,0,0x49,0x89,0xc9,0x109,0x149,0x189, +0x1c9,0x209,0x249,0x289,0x7cb,0x1e4b,0x784b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x19,0x1b,0, +0,0,0,0,0,0,6,5,0,5,5,5,5,5,5,0, +0,0,5,5,5,0,5,5,5,5,0,0,0,5,5,0, +5,0,5,5,0,0,0,5,5,0,0,0,5,5,5,0, +0,0,5,5,5,5,5,5,5,5,5,5,5,5,0,0, +0,0,8,8,6,8,8,0,0,0,8,8,8,0,8,8, +8,6,0,0,5,0,0,0,0,0,0,8,0,0,0,0, +0,0,0,0,5,5,6,6,0,0,0x49,0x89,0xc9,0x109,0x149,0x189, +0x1c9,0x209,0x249,0x289,0,0,0,0,0,0,0,0x17,0x54b,0x58b,0x5cb,0x60b, +0x58b,0x5cb,0x60b,0x1b,6,8,8,8,6,5,5,5,5,5,5,5, +5,0,5,5,5,0,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,0,0,6,5,6,6,6,8,8,8, +8,0,6,6,6,0,6,6,6,6,0,0,0,0,0,0, +0,6,6,0,5,5,5,0,0,5,0,0,5,5,6,6, +0,0,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,5,5,8, +0,0,0,0,0,0,0,0,0,0,0,0,5,6,8,8, +0x17,5,5,5,5,5,5,5,5,0,5,5,5,0,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,0,5,5, +5,5,5,5,5,5,5,5,0,5,5,5,5,5,0,0, +6,5,8,6,8,8,8,8,8,0,6,8,8,0,8,8, +6,6,0,0,0,0,0,0,0,8,8,0,0,0,0,0, +0,5,5,0,5,5,6,6,0,0,0x49,0x89,0xc9,0x109,0x149,0x189, +0x1c9,0x209,0x249,0x289,0x7cb,0x1e4b,0x784b,0x34cb,0x344b,0x3ccb,0x37cb,0x35cb,0x3fcb,0x1b,5,5, +5,5,5,5,6,6,8,8,5,5,5,5,5,5,5,5, +5,0,5,5,5,0,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,6,6,5,8,8,8,6,6,6,6,0,8,8, +8,0,8,8,8,6,5,0x1b,0,0,0,0,5,5,5,8, +0xcc0b,0xca0b,0xcb4b,0xc90b,0x364b,0xc94b,0x350b,5,0,0,0,0,0,0,0x49,0x89, +0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,8,8,0x17,0,0,0, +0,0,0,0,0,0,0,0,0,6,8,8,0,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,0, +0,0,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,0,5,5,5,5,5,5,5,5,5, +0,5,0,0,5,5,5,5,5,5,5,0,0,0,6,0, +0,0,0,8,8,8,6,6,6,0,6,0,8,8,8,8, +8,8,8,8,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,6,5,5,6,6,6,6,6,6,6,0, +0,0,0,0x19,5,5,5,5,5,5,4,6,6,6,6,6, +6,6,6,0x17,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x17,0x17, +0,0,0,0,0,5,5,0,5,0,5,5,5,5,5,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0,5,0,5,5,5,5,5,5,5,5,5, +5,6,5,5,6,6,6,6,6,6,6,6,6,5,0,0, +5,5,5,5,5,0,4,0,6,6,6,6,6,6,6,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,5,5,5,5, +5,0x1b,0x1b,0x1b,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17, +0x17,0x17,0x17,0x1b,0x17,0x1b,0x1b,0x1b,6,6,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x344b,0x3c4b,0x444b,0x4c4b,0x544b,0x5c4b, +0x644b,0x6c4b,0x744b,0x2c4b,0x1b,6,0x1b,6,0x1b,6,0x14,0x15,0x14,0x15,8,8, +5,5,5,5,5,5,5,5,0,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,0,0,0,0,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,8,6,6,6,6,6,0x17,6,6,5,5,5,5, +5,6,6,6,6,6,6,6,6,6,6,6,0,6,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,6,0,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,6,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0x1b,0x1b, +0x17,0x17,0x17,0x17,0x17,0x1b,0x1b,0x1b,0x1b,0x17,0x17,0,0,0,0,0, +5,5,5,5,5,5,5,5,5,5,5,8,8,6,6,6, +6,8,6,6,6,6,6,6,8,6,6,8,8,6,6,5, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x17,0x17,0x17,0x17,0x17,0x17, +5,5,5,5,5,5,8,8,6,6,5,5,5,5,6,6, +6,5,8,8,8,5,5,8,8,8,8,8,8,8,5,5, +5,6,6,6,6,5,5,5,5,5,5,5,5,5,5,5, +5,5,6,8,8,6,6,8,8,8,8,8,8,6,5,8, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,8,8,8,6,0x1b,0x1b, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,0x17,4,2,2,2, +1,1,1,1,1,1,0,1,0,0,0,0,0,1,0,0, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +5,5,5,5,5,5,5,5,5,0,5,5,5,5,0,0, +5,5,5,5,5,5,5,0,5,0,5,5,5,5,0,0, +5,5,5,5,5,5,5,5,5,0,5,5,5,5,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,0,5,5,5,5,0,0,5,5,5,5,5,5,5,0, +5,0,5,5,5,5,0,0,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,0,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,0,5,5,5,5,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,0,0,6,6,6, +0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x30b,0x34b,0x38b,0x3cb,0x40b,0x44b,0x48b, +0x4cb,0x50b,0x7cb,0xa4b,0xccb,0xf4b,0x11cb,0x144b,0x16cb,0x194b,0x1bcb,0x1e4b,0x788b,0,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0,0,0,0, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,0,0,2,2,2,2,2,2,0,0, +0x13,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,0x1b,0x17,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0xc,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,0x14, +0x15,0,0,0,5,5,5,5,5,5,5,5,5,5,5,0x17, +0x17,0x17,0x98a,0x9ca,0xa0a,5,5,5,5,5,5,5,5,0,0,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,6,6,6,8,0,0,0,0,0,0, +0,0,0,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,6,6,8,0x17,0x17,0,0,0,0,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,6,6,0,0,0,0,0,0,0,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,0,5,5,5,0,6,6,0,0,0,0,0,0,0,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,6,6,8,6,6,6,6,6, +6,6,8,8,8,8,8,8,8,8,6,8,8,6,6,6, +6,6,6,6,6,6,6,6,0x17,0x17,0x17,4,0x17,0x17,0x17,0x19, +5,6,0,0,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0, +0,0,0,0,0x54b,0x58b,0x5cb,0x60b,0x64b,0x68b,0x6cb,0x70b,0x74b,0x78b,0,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,6,5,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0x17,0x17,0x17,0x17,0x17,0x17,0x13,0x17,0x17,0x17,0x17,6, +6,6,0x10,6,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0, +0,0,0,0,5,5,5,4,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0,0,0,0,0,0,0,5,5,5,5, +5,6,6,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,0,0, +0,0,0,0,0,0,0,0,6,6,6,8,8,8,8,6, +6,8,8,8,0,0,0,0,8,8,6,8,8,8,8,8, +8,6,6,6,0,0,0,0,0x1b,0,0,0,0x17,0x17,0x49,0x89, +0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,0,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,0,0,5,5,5,5,5,0,0,0, +0,0,0,0,0,0,0,0,5,5,5,5,5,5,5,5, +5,5,5,5,0,0,0,0,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,0,0,0,0,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x30b,0,0,0,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,6,6,8,8,6,0,0,0x17,0x17, +0x17,0x17,0x17,0x17,0x17,0x17,0x17,4,0x17,0x17,0x17,0x17,0x17,0x17,0,0, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,7,6, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,8,6,8,6,6,6,6,6,6,6,0, +6,8,6,8,8,6,6,6,6,6,6,6,6,8,8,8, +8,8,8,6,6,6,6,6,6,6,6,6,6,0,0,6, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,0,0, +0x17,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,6,6,6,6,6, +6,6,6,6,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x17,0x17,0, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +6,6,6,6,8,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +6,8,6,6,6,6,6,8,6,8,8,8,8,8,6,8, +8,5,5,5,5,5,5,5,5,0,0,0,0x49,0x89,0xc9,0x109, +0x149,0x189,0x1c9,0x209,0x249,0x289,0x17,0x17,0x17,0x17,0x17,0x17,5,8,6,6, +6,6,8,8,6,6,8,6,6,6,5,5,0x49,0x89,0xc9,0x109, +0x149,0x189,0x1c9,0x209,0x249,0x289,5,5,5,5,5,5,6,6,8,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,6,8, +6,6,8,8,8,6,8,6,6,6,8,8,0,0,0,0, +0,0,0,0,0x17,0x17,0x17,0x17,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209, +0x249,0x289,0,0,0,5,5,5,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209, +0x249,0x289,5,5,5,5,5,5,8,8,8,8,8,8,8,8, +6,6,6,6,6,6,6,6,8,8,6,6,0,0,0,0x17, +0x17,0x17,0x17,0x17,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,4,4,4,4, +4,4,0x17,0x17,2,2,2,2,2,2,2,2,2,0,0,0, +0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0, +0,1,1,1,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0,0,0,0, +0,0,0,0,6,6,6,0x17,6,6,6,6,6,6,6,6, +6,6,6,6,6,8,6,6,6,6,6,6,6,5,5,5, +5,6,5,5,5,5,5,5,6,5,5,8,6,6,5,0, +0,0,0,0,2,2,2,2,2,2,2,2,2,2,2,2, +4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4, +4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4, +4,4,4,2,2,2,2,2,2,2,2,2,2,2,2,2, +4,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,4, +4,4,4,4,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,2,2,2,2,2,2, +2,2,1,2,2,2,2,2,2,2,2,2,1,1,1,1, +1,0x1a,0x1a,0x1a,0,0,2,2,2,0,2,2,1,1,1,1, +3,0x1a,0x1a,0,2,2,2,2,2,2,2,2,1,1,1,1, +1,1,1,1,2,2,2,2,2,2,0,0,1,1,1,1, +1,1,0,0,2,2,2,2,2,2,2,2,1,1,1,1, +1,1,1,1,2,2,2,2,2,2,2,2,1,1,1,1, +1,1,1,1,2,2,2,2,2,2,0,0,1,1,1,1, +1,1,0,0,2,2,2,2,2,2,2,2,0,1,0,1, +0,1,0,1,2,2,2,2,2,2,2,2,1,1,1,1, +1,1,1,1,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,0,0,2,2,2,2,2,2,2,2,3,3,3,3, +3,3,3,3,2,2,2,2,2,2,2,2,3,3,3,3, +3,3,3,3,2,2,2,2,2,0,2,2,1,1,1,1, +3,0x1a,2,0x1a,0x1a,0x1a,2,2,2,0,2,2,1,1,1,1, +3,0x1a,0x1a,0x1a,2,2,2,2,0,0,2,2,1,1,1,1, +0,0x1a,0x1a,0x1a,0x16,0x17,0x17,0x17,0x18,0x14,0x15,0x17,0x17,0x17,0x17,0x17, +0x17,0x17,0x17,0x17,0x17,0x17,0x18,0x17,0x16,0x17,0x17,0x17,0x17,0x17,0x17,0x17, +0x17,0x17,0x17,0xc,0x10,0x10,0x10,0x10,0x10,0,0x10,0x10,0x10,0x10,0x10,0x10, +0x10,0x10,0x10,0x10,0x2cb,4,0,0,0x3cb,0x40b,0x44b,0x48b,0x4cb,0x50b,0x18,0x18, +0x18,0x14,0x15,4,0xc,0xc,0xc,0xc,0xc,0xc,0xc,0xc,0xc,0xc,0xc,0x10, +0x10,0x10,0x10,0x10,0x13,0x13,0x13,0x13,0x13,0x13,0x17,0x17,0x1c,0x1d,0x14,0x1c, +0x1c,0x1d,0x14,0x1c,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0xd,0xe,0x10,0x10, +0x10,0x10,0x10,0xc,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x1c,0x1d,0x17, +0x17,0x17,0x17,0x16,0x2cb,0x30b,0x34b,0x38b,0x3cb,0x40b,0x44b,0x48b,0x4cb,0x50b,0x18,0x18, +0x18,0x14,0x15,0,4,4,4,4,4,4,4,4,4,4,4,4, +4,0,0,0,0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19, +0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19,0x19, +0x19,0x19,0x19,0x19,0x19,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,6,6,6,6,6,6,6,6,6,6,6,6, +6,7,7,7,7,6,7,7,7,6,6,6,6,6,6,6, +6,6,6,6,6,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1b,0x1b,0x1b,0x1b,1,0x1b,1,0x1b,1,0x1b,1,1, +1,1,0x1b,2,1,1,1,1,2,5,5,5,5,2,0x1b,0x1b, +2,2,1,1,0x18,0x18,0x18,0x18,0x18,1,2,2,2,2,0x1b,0x18, +0x1b,0x1b,2,0x1b,0x358b,0x360b,0x364b,0x348b,0x388b,0x350b,0x390b,0x3d0b,0x410b,0x354b,0x454b,0x35cb, +0x3dcb,0x45cb,0x4dcb,0x58b,0x1b,0x1b,1,0x1b,0x1b,0x1b,0x1b,1,0x1b,0x1b,2,1, +1,1,2,2,1,1,1,2,0x1b,1,0x1b,0x1b,0x18,1,1,1, +1,1,0x1b,0x1b,0x58a,0x5ca,0x60a,0x64a,0x68a,0x6ca,0x70a,0x74a,0x78a,0x7ca,0x80a,0x84a, +0x11ca,0x1e4a,0x980a,0x784a,0x58a,0x5ca,0x60a,0x64a,0x68a,0x6ca,0x70a,0x74a,0x78a,0x7ca,0x80a,0x84a, +0x11ca,0x1e4a,0x980a,0x784a,0x784a,0x984a,0x788a,1,2,0x6ca,0x11ca,0x988a,0x78ca,0x54b,0x1b,0x1b, +0,0,0,0,0x18,0x18,0x18,0x18,0x18,0x1b,0x1b,0x1b,0x1b,0x1b,0x18,0x18, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18, +0x18,0x1b,0x1b,0x18,0x1b,0x1b,0x18,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x18,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x18,0x18,0x1b,0x1b,0x18,0x1b,0x18,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18, +0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18, +0x18,0x18,0x18,0x18,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x14,0x15,0x14,0x15, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x18,0x18,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x14,0x15,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x18,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x18,0x18,0x18,0x18,0x18, +0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x18,0x18,0x18,0x18, +0x18,0x18,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x2cb,0x80b,0x84b,0x88b,0x8cb,0x90b,0x94b,0x98b,0x9cb,0xa0b, +0xa4b,0x30b,0x34b,0x38b,0x3cb,0x40b,0x44b,0x48b,0x4cb,0x50b,0x7cb,0x2cb,0x30b,0x34b,0x38b,0x3cb, +0x40b,0x44b,0x48b,0x4cb,0x50b,0x7cb,0x80b,0x84b,0x88b,0x8cb,0x90b,0x94b,0x98b,0x9cb,0xa0b,0xa4b, +0x30b,0x34b,0x38b,0x3cb,0x40b,0x44b,0x48b,0x4cb,0x50b,0x7cb,0x80b,0x84b,0x88b,0x8cb,0x90b,0x94b, +0x98b,0x9cb,0xa0b,0xa4b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x18,0x18,0x18,0x18, +0x18,0x18,0x18,0x18,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x18,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x18,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x18,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x14,0x15,0x14,0x15,0x14,0x15,0x14,0x15, +0x14,0x15,0x14,0x15,0x14,0x15,0x30b,0x34b,0x38b,0x3cb,0x40b,0x44b,0x48b,0x4cb,0x50b,0x7cb, +0x30b,0x34b,0x38b,0x3cb,0x40b,0x44b,0x48b,0x4cb,0x50b,0x7cb,0x30b,0x34b,0x38b,0x3cb,0x40b,0x44b, +0x48b,0x4cb,0x50b,0x7cb,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x18,0x18,0x18,0x18,0x18,0x14,0x15,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18, +0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18, +0x18,0x18,0x14,0x15,0x14,0x15,0x14,0x15,0x14,0x15,0x14,0x15,0x18,0x18,0x18,0x18, +0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x14, +0x15,0x14,0x15,0x14,0x15,0x14,0x15,0x14,0x15,0x14,0x15,0x14,0x15,0x14,0x15,0x14, +0x15,0x14,0x15,0x14,0x15,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18, +0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18, +0x14,0x15,0x14,0x15,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18, +0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18, +0x14,0x15,0x18,0x18,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x18, +0x18,0x18,0x18,0x18,0x18,0x1b,0x1b,0x18,0x18,0x18,0x18,0x18,0x18,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0,0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,1,2,1,1,1,2,2,1, +2,1,2,1,2,1,1,1,1,2,1,2,2,1,2,2, +2,2,2,2,4,4,1,1,1,2,1,2,2,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,1,2,1,2,6,6,6,1,2,0,0,0,0, +0,0x17,0x17,0x17,0x17,0x344b,0x17,0x17,2,2,2,2,2,2,0,2, +0,0,0,0,0,2,0,0,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,0,0,0,0,0,0,0,4, +0x17,0,0,0,0,0,0,0,0,0,0,0,0,0,0,6, +5,5,5,5,5,5,5,0,5,5,5,5,5,5,5,0, +5,5,5,5,5,5,5,0,5,5,5,5,5,5,5,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,0,0,0,0,0,0,0,0,0, +0x17,0x17,0x1c,0x1d,0x1c,0x1d,0x17,0x17,0x17,0x1c,0x1d,0x17,0x1c,0x1d,0x17,0x17, +0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x13,0x17,0x17,0x13,0x17,0x1c,0x1d,0x17,0x17, +0x1c,0x1d,0x14,0x15,0x14,0x15,0x14,0x15,0x14,0x15,0x17,0x17,0x17,0x17,0x17,4, +0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x13,0x13,0x17,0x17,0x17,0x17, +0x13,0x17,0x14,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17, +0x1b,0x1b,0x17,0x17,0x17,0x14,0x15,0x14,0x15,0x14,0x15,0x14,0x15,0x13,0,0, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0,0,0,0,0,0,0,0,0,0,0,0,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0,0,0,0,0x1b,0x58a,0x5ca,0x60a,0x64a,0x68a,0x6ca,0x70a,0x74a,0x78a,6,6, +6,6,8,8,0x13,4,4,4,4,4,0x1b,0x1b,0x7ca,0xa4a,0xcca,4, +5,0x17,0x1b,0x1b,0xc,0x17,0x17,0x17,0x1b,4,5,0x54a,0x14,0x15,0x14,0x15, +0x14,0x15,0x14,0x15,0x14,0x15,0x1b,0x1b,0x14,0x15,0x14,0x15,0x14,0x15,0x14,0x15, +0x13,0x14,0x15,0x15,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,0,0,6,6,0x1a, +0x1a,4,4,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,0x17, +4,4,4,5,0,0,0,0,0,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,0,0x1b,0x1b,0x58b,0x5cb,0x60b,0x64b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0,0,0,0,0,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0x58b,0x5cb,0x60b,0x64b,0x68b,0x6cb,0x70b,0x74b,0x78b,0x7cb,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x7cb,0xa4b,0xccb,0xf4b,0x11cb,0x144b,0x16cb,0x194b,0x1b,0xa8b,0xacb,0xb0b, +0xb4b,0xb8b,0xbcb,0xc0b,0xc4b,0xc8b,0xccb,0xd0b,0xd4b,0xd8b,0xdcb,0xe0b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0xe4b,0xe8b,0xecb, +0xf0b,0xf4b,0xf8b,0xfcb,0x100b,0x104b,0x108b,0x10cb,0x110b,0x114b,0x118b,0x11cb,5,5,5,5, +5,0x685,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,0x5c5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,0x685,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0x705,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,0x585,5,5,0x705,5,5,5,0x7885, +5,0x605,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,0x785,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +0x5c5,5,5,5,5,5,5,5,0x685,5,0x645,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,0x7985,0x7c5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,0x7845,5,5,5,5, +5,5,5,5,0x605,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,0x685,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +0x1e45,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +0x7985,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,0x7a85,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,0x5c5,5,0x745,5,0x6c5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,0x7c5,5,0x7845,0xa45,0xcc5,5,5,5,5,5,5,0xf45,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,0x605,0x605,0x605,0x605,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,0x645,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0x585,5,5,5,5,5,5,5,0x585,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,0x585,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,0x785,0xa45,5,5,5,5, +5,5,5,5,5,5,5,5,0x585,0x5c5,0x605,5,0x5c5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,0x7c5,5, +5,5,5,5,5,5,5,5,5,5,5,5,0x745,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,0x705,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,0x785,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,0x1e45,5, +5,5,5,5,5,5,0x645,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +0x7885,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,0x5c5,5,5,5,5,0x5c5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,0x5c5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,0x7845,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,0x6c5,5, +5,5,5,5,0x1e45,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +0x6c5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,0x545,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,0,0,0,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,4,5,5,5,5,5,5,5,5,5,5,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0,0,0,0,0,0,0,0,0,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,4,0x17,0x17,0x17, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,5,5,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,2,1,2,1,2,4,4,6,6, +1,2,1,2,1,2,1,2,1,2,1,2,1,2,5,6, +7,7,7,0x17,6,6,6,6,6,6,6,6,6,6,0x17,4, +5,5,5,5,5,5,0x58a,0x5ca,0x60a,0x64a,0x68a,0x6ca,0x70a,0x74a,0x78a,0x54a, +6,6,0x17,0x17,0x17,0x17,0x17,0x17,0,0,0,0,0,0,0,0, +0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a, +0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,4,4,4,4,4,4,4,4,4, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,4,4,4,1,2,5,4,4,2,5,5,5,5,5, +0x1a,0x1a,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +2,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2, +1,2,1,2,4,2,2,2,2,2,2,2,2,1,2,1, +2,1,1,2,1,2,1,2,1,2,1,2,4,0x1a,0x1a,1, +2,1,2,5,1,2,1,2,2,2,1,2,1,2,1,2, +1,2,1,2,1,2,1,1,1,1,1,2,1,1,1,1, +1,2,1,2,1,2,1,2,1,2,1,2,1,1,1,1, +2,1,2,0,0,0,0,0,1,2,0,2,0,2,1,2, +1,2,0,0,0,0,0,0,5,5,6,5,5,5,6,5, +5,5,5,6,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,8,8,6,6,8, +0x1b,0x1b,0x1b,0x1b,6,0,0,0,0x34cb,0x344b,0x3ccb,0x37cb,0x35cb,0x3fcb,0x1b,0x1b, +0x19,0x1b,0,0,0,0,0,0,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,0x17,0x17,0x17,0x17, +0,0,0,0,0,0,0,0,8,8,8,8,6,6,0,0, +0,0,0,0,0,0,0x17,0x17,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209, +0x249,0x289,0,0,0,0,0,0,8,8,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,8,8,8,8,8,8,8,8, +8,8,8,8,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,6,6,6,5,5,5,5,5,5,0x17,0x17,0x17,5, +0x17,5,5,6,5,5,5,5,5,5,6,6,6,6,6,6, +6,6,0x17,0x17,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,6,6,6,6,6,6,6,6,6, +6,6,8,8,0,0,0,0,0,0,0,0,0,0,0,0x17, +8,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0,4, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,0x17,0x17, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,6,8,8,6,6,6,6,8,8,6,6,8,8, +5,5,5,5,5,6,4,5,5,5,5,5,5,5,5,5, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,5,5,5,5,5,0, +5,5,5,5,5,5,5,5,5,6,6,6,6,6,6,8, +8,6,6,8,8,6,6,0,0,0,0,0,0,0,0,0, +5,5,5,6,5,5,5,5,5,5,5,5,6,8,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0x17,0x17,0x17,0x17, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +4,5,5,5,5,5,5,0x1b,0x1b,0x1b,5,8,6,8,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +6,5,6,6,6,5,5,6,6,5,5,5,5,5,6,6, +5,6,5,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,5,5,4,0x17,0x17, +5,5,5,5,5,5,5,5,5,5,5,8,6,6,8,8, +0x17,0x17,5,4,4,8,6,0,0,0,0,0,0,0,0,0, +0,5,5,5,5,5,5,0,0,5,5,5,5,5,5,0, +0,5,5,5,5,5,5,0,0,0,0,0,0,0,0,0, +5,5,5,5,5,5,5,0,5,5,5,5,5,5,5,0, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,0x1a,4,4,4,4, +2,2,2,2,2,2,2,2,2,4,0x1a,0x1a,0,0,0,0, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +5,5,5,8,8,6,8,8,6,8,8,0x17,8,6,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,0,0, +5,5,5,5,0,0,0,0,0,0,0,0,0,0,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,0,0,0,0, +5,5,5,5,5,5,5,0,0,0,0,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, +0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, +0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11, +0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11, +5,5,5,5,5,5,5,5,5,5,5,0x605,5,5,5,5, +5,5,5,0x7c5,5,5,5,5,0x5c5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,0x6c5,5,0x6c5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,0x7c5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,0x18,5,5,5,5,5,5,5,5,5,5,5,5,5,0, +5,5,5,5,5,0,5,0,5,5,0,5,5,0,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,2,2,2,2,2,2,2,0, +0,0,0,0,0,0,0,0,0,0,0,2,2,2,2,2, +0,0,0,0,0,5,6,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a, +0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,0x15,0x14,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,0,0,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,0,0,0,0, +0,0,0,0x1b,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +0x19,0x1b,0x1b,0x1b,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,6,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x14,0x15,0x17,0,0, +0,0,0,0,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,6,0x17,0x13,0x13,0x16,0x16,0x14,0x15,0x14,0x15,0x14,0x15,0x14, +0x15,0x14,0x15,0x14,0x15,0x17,0x17,0x14,0x15,0x17,0x17,0x17,0x17,0x16,0x16,0x16, +0x17,0x17,0x17,0,0x17,0x17,0x17,0x17,0x13,0x14,0x15,0x14,0x15,0x14,0x15,0x17, +0x17,0x17,0x18,0x13,0x18,0x18,0x18,0,0x17,0x19,0x17,0x17,0,0,0,0, +5,5,5,5,5,0,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0,0,0x10,0,0,5,5,5,5,5,5, +0,0,5,5,5,5,5,5,0,0,5,5,5,5,5,5, +0,0,5,5,5,0,0,0,0x19,0x19,0x18,0x1a,0x1b,0x19,0x19,0, +0x1b,0x18,0x18,0x18,0x18,0x1b,0x1b,0,0,0,0,0,0,0,0,0, +0,0x10,0x10,0x10,0x1b,0x1b,0,0,0,0x17,0x17,0x17,0x19,0x17,0x17,0x17, +0x14,0x15,0x17,0x18,0x17,0x13,0x17,0x17,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209, +0x249,0x289,0x17,0x17,0x18,0x18,0x18,0x17,0x1a,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,0x14,0x18,0x15,0x18,0x14,0x15,0x17,0x14,0x15,0x17,0x17,5,5, +5,5,5,5,5,5,5,5,4,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,4,4,5,5,5,5, +5,5,5,5,5,5,5,5,0,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,0,5,5,0,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,0,0,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,0,0,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,0,0,0,0,0,0xb00b,0xb80b,0x784b,0x804b,0x884b,0x904b,0x984b,0xa04b, +0xa84b,0xb04b,0xb84b,0x788b,0x808b,0x888b,0x908b,0x988b,0xa08b,0xa88b,0xb08b,0xb88b,0,0,0,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x17,0x17,0x17,0,0,0,0,0x58b, +0x5cb,0x60b,0x64b,0x68b,0x6cb,0x70b,0x74b,0x78b,0x7cb,0xa4b,0xccb,0xf4b,0x11cb,0x144b,0x16cb,0x194b, +0x1bcb,0x1e4b,0x800b,0x880b,0x900b,0x980b,0xa00b,0xa80b,0x7ca,0x7ca,0x7ca,0x7ca,0x7ca,0xcca,0x11ca,0x11ca, +0x11ca,0x11ca,0x1e4a,0x880a,0x980a,0x980a,0x980a,0x980a,0x980a,0x784a,0x984a,0x68a,0x11ca,0x344b,0x344b,0x388b, +0x3ccb,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x54b,0x34cb, +0x1b,0x1b,0x1b,0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0,0,0,0x34ca,0x344a,0x58a,0x68a,0x11ca,0x980a,0x984a,0x988a,0x68a,0x7ca,0x11ca,0x1e4a, +0x980a,0x784a,0x984a,0x68a,0x7ca,0x11ca,0x1e4a,0x980a,0x784a,0x788a,0x988a,0x7ca,0x58a,0x58a,0x58a,0x5ca, +0x5ca,0x5ca,0x5ca,0x68a,0x1b,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,6,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,6,0x58b,0x5cb,0x60b,0x64b,0x68b,0x6cb,0x70b,0x74b,0x78b,0x7cb,0xa4b, +0xccb,0xf4b,0x11cb,0x144b,0x16cb,0x194b,0x1bcb,0x1e4b,0x800b,0x880b,0x900b,0x980b,0xa00b,0xa80b,0xb00b,0xb80b, +0,0,0,0,0x58b,0x68b,0x7cb,0x11cb,0,0,0,0,0,0,0,0, +0,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0x1bca,5,5,5,5,5,5,5,5,0xb80a,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,6,6,6,6,6,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,0,0x17,5,5,5,5,0,0,0,0,5,5,5,5, +5,5,5,5,0x17,0x58a,0x5ca,0x7ca,0xa4a,0x1e4a,0,0,0,0,0,0, +0,0,0,0,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,0,0,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0, +0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,0,0,0,0,2,2,2,2, +2,2,2,2,5,5,5,5,5,5,5,5,0,0,0,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0,0,0,0,0,0,0,0,0,0,0,0x17, +1,1,1,1,1,1,1,1,1,1,1,0,1,1,1,1, +2,2,0,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,0,2,2,2,2,2,2,2,0,2,2,0,0,0, +1,1,1,1,1,1,1,1,1,1,1,0,1,1,1,1, +1,1,1,0,1,1,0,2,2,2,2,2,2,2,2,2, +5,5,5,5,5,5,5,5,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4, +4,0,4,4,4,4,4,4,4,4,4,0,0,0,0,0, +4,4,4,4,4,4,0,4,4,4,4,4,4,4,4,4, +4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,0,5,5,0,0,0,5,0,0,5, +5,5,5,5,5,5,0,0,5,0,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,0,0x17,0x58b,0x5cb,0x60b,0x7cb,0xa4b,0x1e4b,0x784b,0x788b,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,0x1b,0x1b,0x58b,0x5cb,0x60b,0x64b,0x68b,0x7cb,0xa4b,0,0,0,0, +0,0,0,0x58b,0x5cb,0x60b,0x64b,0x64b,0x68b,0x7cb,0xa4b,0x1e4b,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,0, +5,5,0,0,0,0,0,0x58b,0x68b,0x7cb,0xa4b,0x1e4b,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,0x58b,0x7cb,0xa4b,0x1e4b,0x5cb,0x60b,0,0,0,0x17,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,0,0,0,0,0,0x17,0xa04b,0xa84b,0xb04b,0xb84b, +0x788b,0x808b,0x888b,0x908b,0x988b,0xa08b,0xa88b,0xb08b,0xb88b,0x78cb,0x80cb,0x88cb,0x90cb,0x98cb,0xa0cb,0xa8cb, +0xb0cb,0xb8cb,0x36cb,0x354b,0x34cb,0x348b,0x46cb,0x344b,0x4ecb,0x388b,0x3ccb,0x454b,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0,0,0,0,0x5ecb,0x344b,5,5,0x58b,0x5cb,0x60b,0x64b, +0x68b,0x6cb,0x70b,0x74b,0x78b,0x7cb,0xa4b,0xccb,0xf4b,0x11cb,0x144b,0x16cb,0,0,0x1e4b,0x800b, +0x880b,0x900b,0x980b,0xa00b,0xa80b,0xb00b,0xb80b,0x784b,0x804b,0x884b,0x904b,0x984b,0x30b,0x34b,0x38b,0x3cb, +0x7cb,0xa4b,0x1e4b,0x784b,0x344b,0,0,0,0,0,0,0,0x17,0x17,0x17,0x17, +0x17,0x17,0x17,0x17,0x17,0,0,0,0,0,0,0,5,6,6,6, +0,6,6,0,0,0,0,0,6,6,6,6,5,5,5,5, +0,5,5,5,0,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,0,0, +6,6,6,0,0,0,0,6,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0x58b,0x11cb,0x17,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0x58b,0x7cb,0xa4b,5,5,5,5,5,6,6,0, +0,0,0,0x58b,0x68b,0x7cb,0xa4b,0x1e4b,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0, +0,0,0,0,0,0,0,0,5,5,5,5,5,5,5,5, +0x1b,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,0,0,0,0x17,0x17,0x17, +0x17,0x17,0x17,0x17,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,0,0,0x58b,0x5cb,0x60b,0x64b, +0x7cb,0xa4b,0x1e4b,0x784b,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,0,0,0,0,0,0x58b,0x5cb,0x60b,0x64b, +0x7cb,0xa4b,0x1e4b,0x784b,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,0,0,0,0,0,0,0,0x17,0x17,0x17, +0x17,0,0,0,0,0,0,0,0,0,0,0,0,0x58b,0x5cb,0x60b, +0x64b,0x7cb,0xa4b,0x1e4b,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0, +0,0,0,0,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,0,0,0,0,0,0,0,0x58b,0x68b, +0x7cb,0x11cb,0x1e4b,0x784b,5,5,5,5,6,6,6,6,0,0,0,0, +0,0,0,0,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0, +0,0,0,0,0x30b,0x34b,0x38b,0x3cb,0x40b,0x44b,0x48b,0x4cb,0x50b,0x7cb,0xa4b,0xccb, +0xf4b,0x11cb,0x144b,0x16cb,0x194b,0x1bcb,0x1e4b,0x800b,0x880b,0x900b,0x980b,0xa00b,0xa80b,0xb00b,0xb80b,0x344b, +0x34cb,0x348b,0x388b,0,5,5,5,5,5,5,5,5,5,5,0,6, +6,0x13,0,0,5,5,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,6,6,6,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0x58b,0x5cb,0x60b,0x64b,0x68b,0x7cb,0xa4b,0xccb,0x1e4b,0x344b,5, +0,0,0,0,0,0,0,0,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,6,6,6,6,6,6, +6,6,6,6,6,0x58b,0x7cb,0xa4b,0x1e4b,0x17,0x17,0x17,0x17,0x17,0,0, +0,0,0,0,5,5,6,6,6,6,0x17,0x17,0x17,0x17,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,5,5,5,5,5,0x58b,0x5cb,0x60b,0x64b,0x7cb,0xa4b,0x1e4b, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x144b,0x16cb,0x194b,0x1bcb,0x1e4b,0x784b,0x49,0x89,0xc9,0x109,0x149,0x189, +0x1c9,0x209,0x249,0x289,6,5,5,6,6,5,0,0,0,0,0,0, +0,0,0,6,8,6,8,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,6,6,6,6,6,6,6,0x17,0x17,0x17,0x17,0x17, +0x17,0x17,0,0,0,0,0x30b,0x34b,0x38b,0x3cb,0x40b,0x44b,0x48b,0x4cb,0x50b,0x7cb, +0xa4b,0xccb,0xf4b,0x11cb,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,8,8,8,6,6,6,6,8,8,6,6,0x17, +0x17,0x10,0x17,0x17,0x17,0x17,6,0,0,0,0,0,0,0,0,0, +0,0x10,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0,0,0,0,0,0,0,0x49,0x89,0xc9,0x109, +0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,0,0,5,5,5,5, +5,5,5,6,6,6,6,6,8,6,6,6,6,6,6,6, +6,0,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x17,0x17,0x17,0x17, +5,8,8,5,0,0,0,0,0,0,0,0,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,6,6,6,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,6, +0x17,0x17,5,0,0,0,0,0,0,0,0,0,8,5,5,5, +5,0x17,0x17,0x17,0x17,6,6,6,6,0x17,8,6,0x49,0x89,0xc9,0x109, +0x149,0x189,0x1c9,0x209,0x249,0x289,5,0x17,5,0x17,0x17,0x17,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,8, +8,8,6,6,6,6,6,6,6,6,6,8,0,0x58b,0x5cb,0x60b, +0x64b,0x68b,0x6cb,0x70b,0x74b,0x78b,0x7cb,0xa4b,0xccb,0xf4b,0x11cb,0x144b,0x16cb,0x194b,0x1bcb,0x1e4b, +0x784b,0,0,0,0,0,0,0,0,0,0,0,5,5,5,5, +5,5,5,5,5,5,5,5,8,8,8,6,6,6,8,8, +6,8,6,6,0x17,0x17,0x17,0x17,0x17,0x17,6,5,5,6,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,0,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,0, +5,0,5,5,5,5,0,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,0,5,5,5,5,5,5,5,5,5, +5,0x17,0,0,0,0,0,0,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,8,8,8,6,6,6,6,6, +6,6,6,0,0,0,0,0,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209, +0x249,0x289,0,0,0,0,0,0,5,5,8,8,0,0,6,6, +6,6,6,6,6,0,0,0,6,6,6,6,6,0,0,0, +0,0,0,0,0,0,0,0,6,6,8,8,0,5,5,5, +5,5,5,5,5,0,0,5,5,0,0,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,0,5,5,5,5,5,5, +5,0,5,5,0,5,5,5,5,5,0,6,6,5,8,8, +6,8,8,8,8,0,0,8,8,0,0,8,8,8,0,0, +5,0,0,0,0,0,0,8,0,0,0,0,0,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,8,8,8,6,6,6,6,6,6,6,6, +8,8,6,6,6,8,6,5,5,5,5,0x17,0x17,0x17,0x17,0x17, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x17,0x17,0,0x17,6,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +8,8,8,6,6,6,6,6,6,8,6,8,8,8,8,6, +6,8,6,6,5,5,0x17,5,0,0,0,0,0,0,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,8, +8,8,6,6,6,6,0,0,8,8,8,8,6,6,8,6, +6,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17, +0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,5,5,5,5,6,6,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +8,8,8,6,6,6,6,6,6,6,6,8,8,6,8,6, +6,0x17,0x17,0x17,5,0,0,0,0,0,0,0,0,0,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,0,0, +0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +5,5,5,5,5,5,5,5,5,5,5,6,8,6,8,8, +6,6,6,6,6,6,8,6,5,0x17,0,0,0,0,0,0, +8,8,6,6,6,6,8,6,6,6,6,6,0,0,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x7cb,0xa4b,0x17,0x17,0x17,0x1b, +5,5,5,5,5,5,5,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,8,8,8,6, +6,6,6,6,6,6,6,6,8,6,6,0x17,0,0,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x7cb,0xa4b,0xccb,0xf4b,0x11cb,0x144b, +0x16cb,0x194b,0x1bcb,0,0,0,0,0,0,0,0,0,0,0,0,5, +8,5,8,6,0x17,0x17,0x17,0,0,0,0,0,0,0,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,0,0, +5,5,5,5,5,5,5,0,0,5,0,0,5,5,5,5, +5,5,5,5,0,5,5,0,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,8,8,8,8,8,8,0,8, +8,0,0,6,6,8,6,5,6,5,0x17,5,8,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,5,5,5,5,5,5,5,5, +0,0,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,8,8,8,6,6,6,6, +0,0,6,6,8,8,8,8,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,6,6,6,6,6, +6,8,5,6,6,6,6,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,6, +0,0,0,0,0,0,0,0,5,6,6,6,6,6,6,8, +8,6,6,6,5,5,5,5,5,6,6,6,6,6,6,6, +6,6,6,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,0x17,0x17,0x17,0,0,0,0,0, +0,0,0,0,0,0,0,0,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,6,6,6,6,6,6, +6,6,6,6,6,6,6,8,6,6,0x17,0x17,0x17,5,0x17,0x17, +5,0x17,0x17,0x17,0x17,0x17,0,0,0,0,0,0,0,0,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x58b,0x5cb,0x60b,0x64b,0x68b,0x6cb, +0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0x70b,0x74b,0x78b,0x7cb,0xa4b,0xccb,0xf4b,0x11cb,0x144b,0x16cb,0x194b,0x1bcb,0x1e4b,0,0,0, +0x17,0x17,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,0,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,8,6,6,6,6, +6,6,6,0,6,6,6,6,6,6,8,6,6,6,6,6, +6,6,6,6,0,8,6,6,6,6,6,6,6,8,6,6, +8,6,6,0,0,0,0,0,0,0,0,0,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,0,0,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,5,6, +0,0,0,0,0,0,0,0,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209, +0x249,0x289,0,0,0,0,0,0,5,5,5,5,5,5,5,0, +5,5,0,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,6,6,6,6,6,6,0, +0,0,6,0,6,6,0,6,5,5,5,5,5,5,5,5, +5,5,8,8,8,8,8,0,6,6,0,8,8,6,8,6, +5,0,0,0,0,0,0,0,5,5,5,5,5,5,0,5, +5,0,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,6,6,8,8,0x17, +0x17,0,0,0,0,0,0,0,6,8,6,0x17,0x17,0x17,0x17,0x17, +0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x17,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209, +0x249,0x289,0,0,0,0,0,0,6,6,5,8,5,5,5,5, +5,5,5,5,5,5,5,5,5,0,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +8,8,6,6,6,6,6,0,0,0,8,8,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,5,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0x19,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0, +0,0,0,0,0,0,0,0,0,0,0,0x17,0xcd0b,0xcc0b,0xcb0b,0xd00b, +0xca0b,0xcf0b,0xcb4b,0xd04b,0xc90b,0x37cb,0x37cb,0x364b,0x35cb,0xc94b,0x3fcb,0x350b,0x34cb,0x344b,0x344b,0x3ccb, +0xcd0b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x19,0x19,0x19,0x34ca,0x354a,0x34ca,0x34ca, +0x344a,0x348a,0x388a,0xf4a,0x11ca,0x64a,0x68a,0x6ca,0x70a,0x74a,0x78a,0,0x17,0x17,0x17,0x17, +0x17,0,0,0,0,0,0,0,0,0,0,0,0x5ca,0x60a,0x64a,0x68a, +0x6ca,0x70a,0x74a,0x78a,0x60a,0x64a,0x68a,0x6ca,0x70a,0x74a,0x78a,0x64a,0x68a,0x6ca,0x70a,0x74a, +0x78a,0x58a,0x5ca,0x60a,0x64a,0x68a,0x6ca,0x70a,0x74a,0x78a,0x58a,0x5ca,0x60a,0x64a,0x68a,0x5ca, +0x60a,0x60a,0x64a,0x68a,0x6ca,0x70a,0x74a,0x78a,0x58a,0x5ca,0x60a,0x60a,0x64a,0x68a,0xc08a,0xc18a, +0x58a,0x5ca,0x60a,0x60a,0x64a,0x68a,0x60a,0x60a,0x64a,0x64a,0x64a,0x64a,0x6ca,0x70a,0x70a,0x70a, +0x74a,0x74a,0x78a,0x78a,0x78a,0x78a,0x5ca,0x60a,0x64a,0x68a,0x6ca,0x58a,0x5ca,0x60a,0x64a,0x64a, +0x68a,0x68a,0x5ca,0x60a,0x58a,0x5ca,0x348a,0x388a,0x454a,0x348a,0x388a,0x35ca,5,5,5,5, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,0x17,0x17,0, +0,0,0,0,0,0,0,0,0,0,0,0,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,0x10,0x10,0x10,0x10, +0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,6,5,5,5, +5,5,5,6,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,0,0,0,0,0,0,0,0,0,0,0x49,0x89,0xc9,0x109, +0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,0,0,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,0,0, +6,6,6,6,6,0x17,0,0,0,0,0,0,0,0,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +6,6,6,6,6,6,6,0x17,0x17,0x17,0x17,0x17,0x1b,0x1b,0x1b,0x1b, +4,4,4,4,0x17,0x1b,0,0,0,0,0,0,0,0,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0x7cb,0x1e4b,0x788b,0x790b,0x798b, +0x7a0b,0x7a8b,0,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,0,0,0,0,0,5,5,5, +0x54b,0x58b,0x5cb,0x60b,0x64b,0x68b,0x6cb,0x70b,0x74b,0x78b,0x7cb,0x80b,0x84b,0x88b,0x8cb,0x90b, +0x94b,0x98b,0x9cb,0xa0b,0x58b,0x5cb,0x60b,0x17,0x17,0x17,0x17,0,0,0,0,0, +5,5,5,5,5,5,5,5,5,5,5,0,0,0,0,6, +5,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, +8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, +8,8,8,8,0,0,0,0,0,0,0,6,6,6,6,4, +4,4,4,4,4,4,4,4,4,4,4,4,4,4,0x17,4, +6,0,0,0,0,0,0,0,0,0,0,0,8,8,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,4,4,4,4,0,4,4,4,4,4,4,4, +0,4,4,0,5,5,5,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,5,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,5,5,5,0,0,5,0,0, +0,0,0,0,0,0,0,0,5,5,5,5,0,0,0,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,0,0,0,0,0,5,5,5,5, +5,5,5,5,5,5,5,5,5,0,0,0,0,0,0,0, +5,5,5,5,5,5,5,5,5,5,0,0,0x1b,6,6,0x17, +0x10,0x10,0x10,0x10,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0x1b,0x1b,0x1b,0x1b,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,0,0, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,0,0,0,0,0,0,0,0,0,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0, +0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,8,8,6,6,6,0x1b,0x1b, +0x1b,8,8,8,8,8,8,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,6, +6,6,6,6,6,6,6,0x1b,0x1b,6,6,6,6,6,6,6, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,6,6,6,6,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,6,6, +6,0x1b,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0x54b,0x58b,0x5cb,0x60b, +0x64b,0x68b,0x6cb,0x70b,0x74b,0x78b,0x7cb,0x80b,0x84b,0x88b,0x8cb,0x90b,0x94b,0x98b,0x9cb,0xa0b, +0,0,0,0,0,0,0,0,0,0,0,0,0x58b,0x5cb,0x60b,0x64b, +0x68b,0x6cb,0x70b,0x74b,0x78b,0x7cb,0xa4b,0xccb,0xf4b,0x11cb,0x144b,0x16cb,0x194b,0x1bcb,0x58b,0x5cb, +0x60b,0x64b,0x68b,0x58b,0x68b,0,0,0,0,0,0,0,0x249,0x289,0x49,0x89, +0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209, +0x249,0x289,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,2,2,2,2,2,2, +2,0,2,2,2,2,2,2,2,2,2,2,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +1,0,1,1,0,0,1,0,0,1,1,0,0,1,1,1, +1,0,1,1,1,1,1,1,1,1,2,2,2,2,0,2, +0,2,2,2,2,2,2,2,0,2,2,2,2,2,2,2, +2,2,2,2,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,2,2,2,2,1,1,0,1,1,1,1,0, +0,1,1,1,1,1,1,1,1,0,1,1,1,1,1,1, +1,0,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,1,1,0,1, +1,1,1,0,1,1,1,1,1,0,1,0,0,0,1,1, +1,1,1,1,1,0,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,2,2,2,2,2,2,0,0,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,0x18,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,0x18, +2,2,2,2,2,2,1,1,1,1,1,1,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0x18, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,0x18,2,2,2,2,2,2,1,1,1,1, +1,1,1,1,1,1,1,1,1,1,1,1,2,2,2,0x18, +2,2,2,2,2,2,1,2,0,0,0x49,0x89,0xc9,0x109,0x149,0x189, +0x1c9,0x209,0x249,0x289,0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0,6,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,6,6,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,0x1b,0x1b,0x1b,0x1b,6,6,6,6,6,6,6,6,6, +6,6,6,6,6,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,6,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,6,0x1b,0x1b,0x17,0x17,0x17,0x17,0x17, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,6, +6,6,6,6,6,6,6,0,6,6,6,6,6,6,6,6, +6,6,6,6,6,6,6,6,6,0,0,6,6,6,6,6, +2,2,2,2,2,2,2,2,2,2,5,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,0, +0,0,0,0,0,2,2,2,2,2,2,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +6,6,0,6,6,0,6,6,6,6,6,0,0,0,0,0, +4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,6, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +4,4,4,4,4,4,4,4,4,4,4,4,4,4,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,5,0x1b, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,0,0,0, +6,6,6,6,6,6,6,4,4,4,4,4,4,4,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,6,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +5,5,5,5,5,5,5,5,5,5,5,5,6,6,6,6, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,0,0x19, +5,5,5,5,5,5,5,5,5,5,5,4,6,6,6,6, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,0,0, +5,5,5,5,5,5,5,0,5,5,5,5,0,5,5,0, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,0, +5,5,5,5,5,0,0,0x58b,0x5cb,0x60b,0x64b,0x68b,0x6cb,0x70b,0x74b,0x78b, +6,6,6,6,6,6,6,0,0,0,0,0,0,0,0,0, +2,2,2,2,6,6,6,6,6,6,6,4,0,0,0,0, +0x49,0x89,0xc9,0x109,0x149,0x189,0x1c9,0x209,0x249,0x289,0,0,0,0,0x17,0x17, +1,1,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0x58b,0x5cb,0x60b,0x64b,0x68b,0x6cb,0x70b,0x74b,0x78b,0x7cb,0xa4b,0xccb,0xf4b,0x11cb,0x144b, +0x78cb,0x794b,0x814b,0x58b,0x5cb,0x60b,0x64b,0x68b,0x6cb,0x70b,0x74b,0x78b,0x1b,0x34cb,0x344b,0x3ccb, +0x19,0x58b,0x5cb,0x788b,0x78cb,0,0,0,0,0,0,0,0,0,0,0, +0x16cb,0x194b,0x1bcb,0x1e4b,0x800b,0x880b,0x900b,0x980b,0xa00b,0xa80b,0xb00b,0xb80b,0x784b,0x804b,0x884b,0x904b, +0x984b,0xa04b,0xa84b,0xb04b,0xb84b,0x788b,0x808b,0x888b,0x908b,0x988b,0xa08b,0xa88b,0xb08b,0xb88b,0x78cb,0x80cb, +0x984b,0xa04b,0xa84b,0xb04b,0xb84b,0x788b,0x808b,0x888b,0x908b,0x988b,0xa08b,0xa88b,0xb08b,0xb88b,0x1b,0x5cb, +0x60b,0x64b,0x68b,0x6cb,0x70b,0x74b,0x78b,0x7cb,0x900b,0xa00b,0x804b,0x788b,0x344b,0x354b,0,0, +0,0x58b,0x5cb,0x60b,0x64b,0x68b,0x6cb,0x70b,0x74b,0x78b,0x7cb,0xa4b,0xccb,0xf4b,0x11cb,0x144b, +0x16cb,0x194b,0x1bcb,0x1e4b,0x800b,0x880b,0x900b,0x980b,0xa00b,0xa80b,0xb00b,0xb80b,0x784b,0x804b,0x884b,0x904b, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0x18,0x18,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +5,5,5,5,0,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +0,5,5,0,5,0,0,5,0,5,5,5,5,5,5,5, +5,5,5,0,5,5,5,5,0,5,0,5,0,0,0,0, +0,0,5,0,0,0,0,5,0,5,0,5,0,5,5,5, +0,5,5,0,5,0,0,5,0,5,0,5,0,5,0,5, +0,5,5,0,5,0,0,5,5,5,5,0,5,5,5,5, +5,5,5,0,5,5,5,5,0,5,5,5,5,0,5,0, +5,5,5,5,5,5,5,5,5,5,0,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,0,0,0,0, +0,5,5,5,0,5,5,5,5,5,0,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,0,0,0,0, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0,0, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x2cb,0x2cb,0x30b,0x34b,0x38b,0x3cb,0x40b,0x44b,0x48b,0x4cb,0x50b,0x54b,0x54b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0,0,0,0,0,0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0,0,0,0,0,0,0,0,0,0,0,0,0, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0,0, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0,0,0,0,0, +0x1b,0x1b,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1a,0x1a,0x1a,0x1a,0x1a, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0,0,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0,0,0,0,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0,0,0x1b,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0,0,0,0,0,0,0,0,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0,0,0,0,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0,0,0,0,0,0,0,0,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0, +0x1b,0x1b,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0, +0,0,0,0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0, +0,0,0,0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0,0,0,0,0, +0,0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0,0,0,0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x1b,0x1b,0x1b,5,0x705,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0x645,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,0x645,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,0x685,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,0xcc5,5,5,5,5,5,5,5,5,0xf45,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,0xf45,5,5,5, +5,5,5,5,5,5,5,5,5,5,0x6c5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,0x605,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,0x605,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0x605,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,0x605,5,5,5,5,5,5,5,5,5,5,5,5, +5,0x645,5,5,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,5,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,5,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,5,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0x785,5,5,5,5,5,5,5,5,5,5,5, +5,5,5,5,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10, +0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10, +0x10,0x10,0x10,0x10,0,0x10,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,6,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11, +0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11,0x11, +0x11,0x11,0,0,0,0,0,0 +}; + +static const UTrie2 propsTrie={ + propsTrie_index, + propsTrie_index+4692, + nullptr, + 4692, + 18324, + 0xa40, + 0x12d4, + 0x0, + 0x0, + 0x110000, + 0x59e4, + nullptr, 0, false, false, 0, nullptr +}; + +static const uint16_t propsVectorsTrie_index[32692]={ +0x539,0x541,0x549,0x551,0x569,0x571,0x579,0x581,0x589,0x591,0x599,0x5a1,0x5a9,0x5b1,0x5b9,0x5c1, +0x5c8,0x5d0,0x5d8,0x5e0,0x5e3,0x5eb,0x5f3,0x5fb,0x603,0x60b,0x613,0x61b,0x623,0x62b,0x633,0x63b, +0x643,0x64b,0x652,0x65a,0x662,0x66a,0x672,0x67a,0x682,0x68a,0x68f,0x697,0x69e,0x6a6,0x6ae,0x6b6, +0x6be,0x6c6,0x6ce,0x6d6,0x6dd,0x6e5,0x6ed,0x6f5,0x6fd,0x705,0x70d,0x715,0x71d,0x725,0x72d,0x735, +0x1b39,0xd8a,0xe56,0x118d,0x12cc,0x1d01,0x1ea0,0x1cf9,0x13e6,0x13f6,0x13de,0x13ee,0x80a,0x810,0x818,0x820, +0x828,0x82e,0x836,0x83e,0x846,0x84c,0x854,0x85c,0x864,0x86a,0x872,0x87a,0x882,0x88a,0x892,0x899, +0x8a1,0x8a7,0x8af,0x8b7,0x8bf,0x8c5,0x8cd,0x8d5,0x8dd,0x13fe,0x8e5,0x8ed,0x8f5,0x8fc,0x904,0x90c, +0x914,0x918,0x920,0x927,0x92f,0x937,0x93f,0x947,0x1719,0x1721,0x94f,0x957,0x95f,0x967,0x96f,0x976, +0x177f,0x176f,0x1777,0x1a74,0x1a7c,0x140e,0x97e,0x1406,0x1662,0x1662,0x1664,0x1422,0x1423,0x1416,0x1418,0x141a, +0x1787,0x1789,0x986,0x1789,0x98e,0x993,0x99b,0x178e,0x9a1,0x1789,0x9a7,0x9af,0xc6a,0x1796,0x1796,0x9b7, +0x17a6,0x17a7,0x17a7,0x17a7,0x17a7,0x17a7,0x17a7,0x17a7,0x17a7,0x17a7,0x17a7,0x17a7,0x17a7,0x17a7,0x17a7,0x17a7, +0x17a7,0x17a7,0x17a7,0x179e,0x9bf,0x17af,0x17af,0x9c7,0xb92,0xb9a,0xba2,0xbaa,0x17bf,0x17b7,0x9cf,0x9d7, +0x9df,0x17c9,0x17d1,0x9e7,0x17c7,0x9ef,0x1b41,0xd92,0xbb2,0xbba,0xbc2,0xbc7,0x19da,0xc91,0xc98,0x1936, +0xc42,0x1b49,0xd9a,0xda2,0xdaa,0xdb2,0xf60,0xf64,0x1a3a,0x1a3f,0xcd0,0xcd8,0x1ab0,0x1ab8,0x1c19,0xe5e, +0x1ac0,0xd1e,0xd26,0x1ac8,0x1105,0x11b5,0xf38,0xdba,0x1956,0x193e,0x194e,0x1946,0x19f2,0x19ea,0x19a6,0x1a32, +0x142b,0x142b,0x142b,0x142b,0x142e,0x142b,0x142b,0x1436,0x9f7,0x143e,0x9fb,0xa03,0x143e,0xa0b,0xa13,0xa1b, +0x144e,0x1446,0x1456,0xa23,0xa2b,0x145e,0xa33,0xa3b,0x1466,0x146e,0x1476,0x147e,0xa43,0x1486,0x148d,0x1495, +0x149d,0x14a5,0x14ad,0x14b5,0x14bd,0x14c4,0x14cc,0x14d4,0x14dc,0x14e4,0x14e7,0x14e9,0x17d9,0x18cc,0x18d2,0x1a22, +0x14f1,0xa4b,0xa53,0x1617,0x161c,0x161f,0x1625,0x14f9,0x162d,0x162d,0x1509,0x1501,0x1511,0x1519,0x1521,0x1529, +0x1531,0x1539,0x1541,0x1549,0x18da,0x192e,0x1a84,0x1be1,0x1559,0x155f,0x1567,0x156f,0x1551,0x1577,0x18e2,0x18e9, +0x17e1,0x17e1,0x17e1,0x17e1,0x17e1,0x17e1,0x17e1,0x17e1,0x18f1,0x18f1,0x18f1,0x18f1,0x18f9,0x1900,0x1902,0x1909, +0x1911,0x1915,0x1915,0x1918,0x1915,0x1915,0x191e,0x1915,0x195e,0x1a2a,0x1a8c,0xbcf,0xbd5,0x1d45,0x1d4d,0x1e2b, +0x19ca,0x19be,0x19c2,0x1a47,0x19ae,0x19ae,0x19ae,0xc52,0x19b6,0xc72,0x1a0a,0xcc0,0xc5a,0xc62,0xc62,0x1ad0, +0x19fa,0x1a94,0xca8,0xcb0,0xa5b,0x17e9,0x17e9,0xa63,0x17f1,0x17f1,0x17f1,0x17f1,0x17f1,0x17f1,0xa6b,0x73d, +0x164a,0x166c,0xa73,0x1674,0xa7b,0x167c,0x1684,0x168c,0xa83,0xa88,0x1694,0x169b,0xa8d,0x17f9,0x1a1a,0xc4a, +0xa95,0x16f6,0x16fd,0x16a3,0x1705,0x1709,0x16ab,0x16af,0x16c8,0x16c8,0x16ca,0x16b7,0x16bf,0x16bf,0x16c0,0x1711, +0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801, +0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801, +0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801, +0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801, +0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801, +0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801, +0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801, +0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801, +0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801, +0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801, +0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801, +0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801, +0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1801,0x1804,0x1966,0x1966, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2, +0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d2,0x16d9,0x1b31,0x1f0c, +0x180c,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812, +0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812,0x1812, +0x1812,0x1812,0x1812,0x1812,0xa9d,0x181a,0xaa5,0x1b51,0x1adc,0x1adc,0x1adc,0x1adc,0x1adc,0x1adc,0x1adc,0x1adc, +0x1ad8,0xd2e,0x1aec,0x1ae4,0x1aee,0x1b59,0x1b59,0xdc2,0x19d2,0x1a4f,0x1aa4,0x1aa8,0x1a9c,0x1c11,0xce0,0xce7, +0x1a02,0xcb8,0x1a57,0xcef,0x1af6,0x1af9,0xd36,0x1b61,0x1b09,0x1b01,0xd3e,0xdca,0x1b69,0x1b6d,0xdd2,0x100f, +0x1b11,0xd46,0xd4e,0x1b75,0x1b85,0x1b7d,0xdda,0xf08,0xe66,0xe6e,0x1d9b,0xfbf,0x1e48,0x1e48,0x1b8d,0xde2, +0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762, +0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764, +0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766, +0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761, +0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763, +0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765, +0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767, +0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762, +0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764, +0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766, +0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761, +0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763, +0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765, +0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767, +0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762, +0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764, +0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766, +0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761, +0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763, +0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765, +0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767, +0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0x1767,0x1761,0x1762,0x1763,0x1764,0x1765,0x1766,0xaad,0xdea,0xded, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739, +0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739,0x1739, +0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635, +0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635, +0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635, +0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635, +0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635, +0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635, +0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635, +0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635, +0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635, +0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635, +0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635, +0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635, +0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x16e1,0x16e1,0x16e1,0x16e1,0x16e1,0x16e1,0x16e1,0x16e1, +0x16e6,0x16ee,0x1926,0x13a3,0x1a12,0x1a12,0x13a7,0x13ae,0xab5,0xabd,0xac5,0x1597,0x159e,0x15a6,0xacd,0x15ae, +0x15ec,0x15ec,0x157f,0x1587,0x15b6,0x15e3,0x15e4,0x15f4,0x15be,0x15c3,0x15cb,0x15d3,0xad5,0x15db,0xadd,0x158f, +0xcc8,0x15fc,0xae5,0xaed,0x1604,0x160a,0x160f,0xaf5,0xb05,0x1652,0x165a,0x163d,0x1642,0xb0d,0xb15,0xafd, +0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729, +0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1729,0x1731,0x1731,0x1731,0x1731, +0x1564,0x1564,0x15a4,0x15e4,0x1624,0x1664,0x16a4,0x16e4,0x1720,0x1760,0x178c,0x17cc,0x180c,0x184c,0x188c,0x18cc, +0x190c,0x1948,0x1988,0x19c8,0x1a08,0x1a3c,0x1a78,0x1ab8,0x1af8,0x1b38,0x1b74,0x1bb4,0x1bf4,0x1c34,0x1c74,0x1cb4, +0xe59,0xa80,0xac0,0xb00,0xb40,0xb6b,0xf99,0xa40,0xed9,0xa40,0xa40,0xa40,0xa40,0xbab,0x13e2,0x13e2, +0xf19,0xfd9,0xa40,0xa40,0xa40,0xbeb,0xf59,0xc2b,0xa40,0xc51,0xc91,0xcd1,0xd11,0xd51,0xe99,0xdc9, +0x1322,0x1322,0x1322,0x1322,0x1322,0x1322,0x1322,0x1322,0x1322,0x1322,0x1322,0x1322,0x1322,0x1322,0x1322,0x1322, +0x1322,0x1322,0x1322,0x1322,0x1019,0x1362,0x1157,0x1197,0x13a2,0x11a2,0x1422,0x1422,0x1422,0x1059,0x1079,0x10b9, +0x1462,0x1462,0x11e2,0x14a2,0x10f9,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079, +0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1079,0x1117, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xd89, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xd89, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xd89, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xd89, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xd89, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xd89, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xd89, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xd89, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xd89, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xd89, +0xe09,0xe19,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40, +0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xa40,0xd89, +0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2, +0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x12a2,0x1222, +0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2, +0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x12e2,0x1262, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0xc27,0xc2a,0xdf5,0x1deb,0x1017,0x745,0x559,0x10b1,0xcf7,0xd76,0x559,0x559,0x1d11,0xf10,0xf18,0x1e33, +0xc7a,0xc81,0xc89,0x1b95,0x1dcb,0x559,0x1dab,0xfe7,0x1b9d,0xdfd,0xe05,0xe0d,0x103f,0x74d,0x559,0x559, +0x1ba5,0x1ba5,0x755,0x559,0x1e60,0x10c9,0x1e58,0x10d1,0x1f4c,0x11cb,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0xe15,0x1fa4,0x12c4,0x1346,0x1347,0x1f6c,0x11f3,0x11fa,0x1201,0x1303,0x1307,0x127b,0x1211, +0x1c21,0x1c23,0xe76,0xe7d,0x1bad,0x1bb5,0xe1d,0xf30,0x1d09,0xef8,0xf00,0xfdf,0x1d29,0x1d2d,0x1d35,0x105f, +0xfaf,0x1d8b,0x75d,0x559,0x10b9,0x10c1,0x1d93,0xfb7,0xf91,0xf97,0xf9f,0xfa7,0x559,0x559,0x559,0x559, +0x1ed0,0x1ec8,0x113b,0x1143,0x1e13,0x1e0b,0x1087,0x559,0x559,0x559,0x559,0x559,0x1dfb,0x1047,0x104f,0x1057, +0x1dc3,0x1dbb,0xff7,0x1133,0x1d3d,0xf40,0x765,0x559,0x1097,0x109f,0x76d,0x559,0x559,0x559,0x559,0x559, +0x1f44,0x11ad,0x775,0x559,0x559,0x1e23,0x1e1b,0x108f,0x1283,0x1289,0x1291,0x559,0x559,0x1219,0x121d,0x1225, +0x1f04,0x1efc,0x1195,0x1ef4,0x1eec,0x1185,0x1df3,0x1037,0x1357,0x135a,0x135a,0x559,0x559,0x559,0x559,0x559, +0x10e9,0x10ee,0x10f6,0x10fd,0x1125,0x112b,0x559,0x559,0x1169,0x116d,0x1175,0x11bd,0x11c3,0x77d,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x11db,0x136a,0x136f,0x1377,0x559,0x559,0x781,0x1f8c,0x126b, +0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f, +0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a5f,0x1a64,0xcff,0xd06,0xd06,0xd06, +0x1a6c,0x1a6c,0x1a6c,0xd0e,0x1e50,0x1e50,0x1e50,0x1e50,0x1e50,0x1e50,0x789,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x78d,0x1fbc,0x1fbc,0x12d4,0x1c2b,0x1c2b,0x1c2b,0x1c2b,0x1c2b, +0x1c2b,0x1c2b,0x1c2b,0x1c2b,0x1c2b,0x1c2b,0x1c2b,0x1c2b,0x1c2b,0x1c2b,0x1c2b,0x1c2b,0xe85,0xfff,0x1007,0x1fc4, +0x130f,0x1317,0xf48,0x1de3,0x1ddb,0x1027,0x102f,0x795,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x1f64,0x1f5c,0x11eb, +0x559,0x559,0x559,0x1d21,0x1d21,0xf20,0x1d19,0xf28,0x559,0x559,0x111d,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x799,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x1d73,0x1d73,0x1d73,0xf6c,0xf71, +0x7a1,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x1fd4,0x1337,0x133e,0x1fcc,0x1fcc,0x1fcc,0x7a9, +0x559,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0xb2b,0x184f,0xb33,0x1850,0x1847,0x1858,0x185e,0x1866, +0xb3b,0x198e,0x198e,0x7b1,0x559,0x559,0x559,0x1362,0x11e3,0x197e,0x197e,0xc32,0xd16,0x559,0x559,0x559, +0x559,0x1897,0x189e,0xb43,0x18a1,0xb4b,0xb53,0xb5b,0x189b,0xb63,0xb6b,0xb73,0x18a0,0x18a8,0x1897,0x189e, +0x189a,0x18a1,0x18a9,0x1898,0x189f,0x189b,0xb7a,0x186e,0x1876,0x187d,0x1884,0x1871,0x1879,0x1880,0x1887,0xb82, +0x188f,0x1e78,0x1e78,0x1e78,0x1e78,0x1e78,0x1e78,0x1e78,0x1e78,0x1e78,0x1e78,0x1e78,0x1e78,0x1e78,0x1e78,0x1e78, +0x1e78,0x1e68,0x1e6b,0x1e68,0x1e72,0x10d9,0x7b9,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x12f0,0x12f8,0x12fb,0x12fb,0x12fb,0x12fb,0x12fb, +0x12fb,0x110d,0x1115,0x1fdc,0x134f,0x7c1,0x559,0x559,0x559,0x1f84,0x122d,0x7c9,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x7cd,0x131f,0x1f94,0x1273,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x7d5,0x137f,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x12dc,0x1db3,0x1db3,0x1db3,0x1db3,0x1db3,0x1db3,0xfef,0x559,0x1ec0,0x1eb8,0x10e1,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x7dd,0x1f54,0x11d3,0x559,0x559,0x1235,0x1236,0x7e5,0x559,0x559,0x559,0x559, +0x559,0xebd,0xec5,0xecd,0xed5,0xedd,0xee5,0xeec,0xef0,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x7e9,0x1067,0x1e03,0x106d,0x1e03,0x1075,0x107a,0x107f, +0x107f,0x1e88,0x1ea8,0x1eb0,0x1f1c,0x1e90,0x1f74,0x1e98,0x1f24,0x1f7c,0x1f7c,0x119d,0x11a5,0x124d,0x1253,0x125b, +0x1263,0x1f9c,0x1f9c,0x1f9c,0x1f9c,0x12a7,0x1f9c,0x12ad,0x12b1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1, +0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1, +0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f1,0x7f2,0xb8a,0x18b1,0x18b1,0x18b1,0x7fa,0x7fa,0x7fa, +0x7fa,0x1986,0x1986,0x1986,0x1986,0x1986,0x1986,0x1986,0x802,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa, +0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa, +0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa, +0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa, +0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0x7fa,0xbdd,0xbe4,0xbec,0xbf4,0x196e,0x196e,0x196e, +0xbfc,0xc04,0xc07,0x199e,0x1996,0xc3a,0xd56,0xd5a,0xd5e,0x559,0x559,0x559,0x559,0xd66,0x1b19,0xd6e, +0xf58,0x1822,0xb1d,0xb23,0x101f,0xc0f,0x19e2,0xca0,0x559,0x1837,0x182a,0x182f,0x1976,0xc17,0xc1f,0x114b, +0x1151,0x1d7b,0xf79,0x1d6b,0xf50,0x1327,0x132f,0x559,0x559,0x1da3,0x1da3,0x1da3,0x1da3,0x1da3,0x1da3,0x1da3, +0x1da3,0x1da3,0xfc7,0xfcf,0xfd7,0x12e4,0x12e8,0x559,0x559,0x1b21,0xd7e,0x1b29,0x1b29,0xd82,0xe8d,0xe95, +0xe9d,0x1bf1,0x1bd9,0x1bf9,0x1c01,0x1be9,0xe25,0xe29,0xe30,0xe38,0xe3c,0xe44,0xe4c,0xe4e,0xe4e,0xe4e, +0xe4e,0x1c62,0x1c6a,0x1c62,0x1c70,0x1c78,0x1c43,0x1c80,0x1c88,0x1c62,0x1c90,0x1c98,0x1c9f,0x1ca7,0x1c4b,0x1c62, +0x1cac,0x1c53,0x1c5a,0x1cb4,0x1cba,0x1d5c,0x1d63,0x1d55,0x1cc1,0x1cc9,0x1cd1,0x1cd9,0x1dd3,0x1ce1,0x1ce9,0xea5, +0xead,0x1c33,0x1c33,0x1c33,0xeb5,0x1d83,0x1d83,0xf81,0xf89,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x1e3b,0x1e3b,0x1e3b,0x1e3b,0x1e3b,0x1e3b,0x1e3b, +0x1e3b,0x1e3b,0x1e3b,0x1e3b,0x1e3b,0x1e3b,0x1e3b,0x1e40,0x1e3b,0x1e3b,0x1e3b,0x10a7,0x10a9,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8, +0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8, +0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8, +0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8, +0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1159,0x1c3b,0x1f14,0x1f14,0x1f14,0x1f14,0x1f14,0x1f14, +0x1f14,0x1f34,0x1161,0x123e,0x1245,0x1f3c,0x1f3c,0x1f3c,0x1f3c,0x1f3c,0x1f3c,0x1f3c,0x1f3c,0x1f3c,0x1f3c,0x1f3c, +0x117d,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bbd, +0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bbf,0x1bbd,0x1bc7,0x1bbd,0x1bbd, +0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bca,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1bd1,0x1209,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0, +0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0,0x1ee0, +0x1ee4,0x1fb4,0x1fb4,0x1fb4,0x1fb4,0x1fb4,0x1fb4,0x1fb4,0x1fb4,0x1fb4,0x1fb4,0x1fb4,0x1fb4,0x1fb4,0x1fb4,0x1299, +0x129f,0x12b9,0x12bc,0x12bc,0x12bc,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559, +0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x559,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9, +0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9, +0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9, +0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18bc, +0x1387,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c, +0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c, +0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x138f,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387, +0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387, +0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387, +0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387, +0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x18c4,0x18c4,0x18c4,0x18c4,0x18c4,0x18c4,0x18c4, +0x18c4,0x18c4,0x18c4,0x18c4,0x18c4,0x18c4,0x18c4,0x18c4,0x18c4,0x13b6,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387, +0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387, +0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387, +0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1393,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4, +0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4, +0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x139b,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387, +0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387, +0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387, +0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1387, +0x1387,0x1387,0x1387,0x1387,0x1387,0x1387,0x1393,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09, +0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09, +0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09, +0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09, +0x13be,0x1cf1,0x1cf1,0x1cf1,0x1cf1,0x1cf1,0x1cf1,0x13c6,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80, +0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80, +0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80, +0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80, +0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x13ce,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c, +0x1f2c,0x1f2c,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac, +0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x13d6,0x1fe4,0x1fe4,0x1fe4, +0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4, +0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4, +0x1fe4,0x1fe4,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751, +0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751, +0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751, +0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751, +0x1751,0x1741,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759, +0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759, +0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759, +0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759, +0x1759,0x1749,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751, +0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751, +0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751, +0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751,0x1751, +0x1751,0x1751,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759, +0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759, +0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759, +0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759,0x1759, +0x1759,0x1759,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9, +0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9, +0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9, +0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9,0x18b9, +0x18b9,0x18b9,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09, +0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09, +0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09, +0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09,0x1c09, +0x1c09,0x1c09,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80, +0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80, +0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80, +0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80,0x1e80, +0x1e80,0x1e80,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8, +0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8, +0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8, +0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8,0x1ed8, +0x1ed8,0x1ed8,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c, +0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c, +0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c, +0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c,0x1f2c, +0x1f2c,0x1f2c,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac, +0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac, +0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac, +0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac,0x1fac, +0x1fac,0x1fac,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4, +0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4, +0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4, +0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4,0x1fe4, +0x1fe4,0x1fe4,0x538,0x538,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e5,0x2ee,0x2e8, +0x2e8,0x2eb,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2,0x2e2, +0x2e2,0x2e2,0x2e2,0x2e2,0x7da,0x7d4,0x7b9,0x79e,0x7aa,0x7a7,0x79e,0x7b6,0x7a4,0x7b0,0x79e,0x7cb, +0x7c2,0x7b3,0x7d7,0x7ad,0x79b,0x79b,0x79b,0x79b,0x79b,0x79b,0x79b,0x79b,0x79b,0x79b,0x7bf,0x7bc, +0x7c5,0x7c5,0x7c5,0x7d4,0x79e,0x7e6,0x7e6,0x7e6,0x7e6,0x7e6,0x7e6,0x7e0,0x7e0,0x7e0,0x7e0,0x7e0, +0x7e0,0x7e0,0x7e0,0x7e0,0x7e0,0x7e0,0x7e0,0x7e0,0x7e0,0x7e0,0x7e0,0x7e0,0x7e0,0x7e0,0x7e0,0x7a4, +0x7aa,0x7b0,0x7d1,0x798,0x7ce,0x7e3,0x7e3,0x7e3,0x7e3,0x7e3,0x7e3,0x7dd,0x7dd,0x7dd,0x7dd,0x7dd, +0x7dd,0x7dd,0x7dd,0x7dd,0x7dd,0x7dd,0x7dd,0x7dd,0x7dd,0x7dd,0x7dd,0x7dd,0x7dd,0x7dd,0x7dd,0x7a4, +0x7c8,0x7a1,0x7c5,0x2e2,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1,0x300,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1, +0x2f1,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1,0x2f1, +0x2f1,0x2f1,0x2f1,0x2f1,0x2f4,0x64b,0x7ef,0x7f2,0x651,0x7f2,0x7ec,0x645,0x63c,0x2fa,0x65a,0x2fd, +0x7f5,0x633,0x648,0x7e9,0x64e,0x657,0x639,0x639,0x63f,0x2f7,0x645,0x642,0x63c,0x639,0x65a,0x2fd, +0x636,0x636,0x636,0x64b,0x306,0x306,0x306,0x306,0x306,0x306,0x663,0x306,0x306,0x306,0x306,0x306, +0x306,0x306,0x306,0x306,0x663,0x306,0x306,0x306,0x306,0x306,0x306,0x654,0x663,0x306,0x306,0x306, +0x306,0x306,0x663,0x65d,0x660,0x660,0x303,0x303,0x303,0x303,0x65d,0x303,0x660,0x660,0x660,0x303, +0x660,0x660,0x303,0x303,0x65d,0x303,0x660,0x660,0x303,0x303,0x303,0x654,0x65d,0x660,0x660,0x303, +0x660,0x303,0x65d,0x303,0x312,0x669,0x312,0x309,0x312,0x309,0x312,0x309,0x312,0x309,0x312,0x309, +0x312,0x309,0x312,0x309,0x30f,0x666,0x312,0x669,0x312,0x309,0x312,0x309,0x312,0x309,0x312,0x669, +0x312,0x309,0x312,0x309,0x312,0x309,0x312,0x309,0x312,0x309,0x66f,0x666,0x312,0x309,0x312,0x669, +0x312,0x309,0x312,0x309,0x312,0x666,0x672,0x66c,0x312,0x309,0x312,0x309,0x666,0x312,0x309,0x312, +0x309,0x312,0x309,0x672,0x66c,0x66f,0x666,0x312,0x669,0x312,0x309,0x312,0x669,0x675,0x66f,0x666, +0x312,0x669,0x312,0x309,0x312,0x309,0x66f,0x666,0x312,0x309,0x312,0x309,0x312,0x309,0x312,0x309, +0x312,0x309,0x312,0x309,0x312,0x309,0x312,0x309,0x312,0x309,0x66f,0x666,0x312,0x309,0x312,0x669, +0x312,0x309,0x312,0x309,0x312,0x309,0x312,0x309,0x312,0x309,0x312,0x309,0x312,0x312,0x309,0x312, +0x309,0x312,0x309,0x30c,0x315,0x321,0x321,0x315,0x321,0x315,0x321,0x321,0x315,0x321,0x321,0x321, +0x315,0x315,0x321,0x321,0x321,0x321,0x315,0x321,0x321,0x315,0x321,0x321,0x321,0x315,0x315,0x315, +0x321,0x321,0x315,0x321,0x324,0x318,0x321,0x315,0x321,0x315,0x321,0x321,0x315,0x321,0x315,0x315, +0x321,0x315,0x321,0x324,0x318,0x321,0x321,0x321,0x315,0x321,0x315,0x321,0x321,0x315,0x315,0x31e, +0x321,0x315,0x315,0x315,0x31e,0x31e,0x31e,0x31e,0x327,0x327,0x31b,0x327,0x327,0x31b,0x327,0x327, +0x31b,0x324,0x678,0x324,0x678,0x324,0x678,0x324,0x678,0x324,0x678,0x324,0x678,0x324,0x678,0x324, +0x678,0x315,0x324,0x318,0x324,0x318,0x324,0x318,0x321,0x315,0x324,0x318,0x324,0x318,0x324,0x318, +0x324,0x318,0x324,0x318,0x318,0x327,0x327,0x31b,0x324,0x318,0x9cf,0x9cf,0x9d2,0x9cc,0x324,0x318, +0x324,0x318,0x324,0x318,0x324,0x318,0x324,0x318,0x324,0x318,0x324,0x318,0x324,0x318,0x324,0x318, +0x324,0x318,0x324,0x318,0x324,0x318,0x324,0x318,0x9d2,0x9cc,0x9d2,0x9cc,0x9cf,0x9c9,0x9d2,0x9cc, +0xb8b,0xc84,0x9cf,0x9c9,0x9cf,0x9c9,0x9d2,0x9cc,0x9d2,0x9cc,0x9d2,0x9cc,0x9d2,0x9cc,0x9d2,0x9cc, +0x9d2,0x9cc,0x9d2,0x9cc,0xc84,0xc84,0xc84,0xd7d,0xd7d,0xd7d,0xd80,0xd80,0xd7d,0xd80,0xd80,0xd7d, +0xd7d,0xd80,0xebe,0xec1,0xec1,0xec1,0xec1,0xebe,0xec1,0xebe,0xec1,0xebe,0xec1,0xebe,0xec1,0xebe, +0x32a,0x67b,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a, +0x32a,0x67b,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a, +0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a, +0x32d,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a,0x32a, +0x32a,0x32a,0x32a,0x32a,0x32a,0x9d5,0x9d5,0x9d5,0x9d5,0x9d5,0xc87,0xc87,0x342,0x342,0x342,0x342, +0x342,0x342,0x342,0x342,0x342,0x339,0x339,0x339,0x339,0x339,0x339,0x339,0x336,0x336,0x333,0x333, +0x681,0x333,0x339,0x684,0x33c,0x684,0x684,0x684,0x33c,0x684,0x339,0x339,0x687,0x33f,0x333,0x333, +0x333,0x333,0x333,0x333,0x67e,0x67e,0x67e,0x67e,0x330,0x67e,0x333,0xb01,0x342,0x342,0x342,0x342, +0x342,0x333,0x333,0x333,0x333,0x333,0x9de,0x9de,0x9db,0x9d8,0x9db,0xc8a,0xc8a,0xc8a,0xc8a,0xc8a, +0xc8a,0xc8a,0xc8a,0xc8a,0xc8a,0xc8a,0xc8a,0xc8a,0xc8a,0xc8a,0xc8a,0xc8a,0x68a,0x68a,0x68a,0x68a, +0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a, +0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a, +0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a, +0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68a,0x68d,0x68d,0x92d,0x68d, +0x68d,0x930,0xb04,0xb04,0xb04,0xb04,0xb04,0xb04,0xb04,0xb04,0xb04,0xc3c,0xd47,0xd47,0xd47,0xd47, +0xd47,0xd47,0xd47,0xd47,0xe82,0xe82,0xe82,0xe82,0xe85,0xd4a,0xd4a,0xd4a,0x690,0x690,0xb07,0xc81, +0xc81,0xc81,0xc81,0xc81,0xc81,0xc81,0xc81,0xc81,0xc81,0xc81,0xc81,0xc81,0xf6c,0xf69,0xf6c,0xf69, +0x34e,0x357,0xf6c,0xf69,9,9,0x35d,0xec4,0xec4,0xec4,0x345,0x14af,9,9,9,9, +0x35a,0x348,0x36c,0x34b,0x36c,0x36c,0x36c,9,0x36c,9,0x36c,0x36c,0x363,0x696,0x696,0x696, +0x696,0x696,0x696,0x696,0x696,0x696,0x696,0x696,0x696,0x696,0x696,0x696,0x696,0x696,9,0x696, +0x696,0x696,0x696,0x696,0x696,0x696,0x36c,0x36c,0x363,0x363,0x363,0x363,0x363,0x693,0x693,0x693, +0x693,0x693,0x693,0x693,0x693,0x693,0x693,0x693,0x693,0x693,0x693,0x693,0x693,0x693,0x360,0x693, +0x693,0x693,0x693,0x693,0x693,0x693,0x363,0x363,0x363,0x363,0x363,0xf6c,0x36f,0x36f,0x372,0x36c, +0x36c,0x36f,0x366,0x9e1,0xb94,0xb91,0x369,0x9e1,0x369,0x9e1,0x369,0x9e1,0x369,0x9e1,0x354,0x351, +0x354,0x351,0x354,0x351,0x354,0x351,0x354,0x351,0x354,0x351,0x354,0x351,0x36f,0x36f,0x366,0x360, +0xb43,0xb40,0xb8e,0xc90,0xc8d,0xc93,0xc90,0xc8d,0xd83,0xd86,0xd86,0xd86,0x9f0,0x6a2,0x37e,0x381, +0x37e,0x37e,0x37e,0x381,0x37e,0x37e,0x37e,0x37e,0x381,0x9f0,0x381,0x37e,0x69f,0x69f,0x69f,0x69f, +0x69f,0x69f,0x69f,0x69f,0x69f,0x6a2,0x69f,0x69f,0x69f,0x69f,0x69f,0x69f,0x69f,0x69f,0x69f,0x69f, +0x69f,0x69f,0x69f,0x69f,0x69f,0x69f,0x69f,0x69f,0x69f,0x69f,0x69f,0x69f,0x699,0x699,0x699,0x699, +0x699,0x699,0x699,0x699,0x699,0x69c,0x699,0x699,0x699,0x699,0x699,0x699,0x699,0x699,0x699,0x699, +0x699,0x699,0x699,0x699,0x699,0x699,0x699,0x699,0x9ea,0x69c,0x378,0x37b,0x378,0x378,0x378,0x37b, +0x378,0x378,0x378,0x378,0x37b,0x9ea,0x37b,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378, +0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x381,0x37b, +0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x375,0x939,0x93c,0x91e,0x91e,0x1116, +0x9e4,0x9e4,0xb9a,0xb97,0x9ed,0x9e7,0x9ed,0x9e7,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378, +0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378, +0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378, +0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x378,0x37e,0x381,0x37b,0x37e,0x378,0xb9a,0xb97,0x37e, +0x378,0xb9a,0xb97,0x37e,0x378,0xb9a,0xb97,0xec7,0x381,0x37b,0x381,0x37b,0x37e,0x378,0x381,0x37b, +0x37e,0x378,0x381,0x37b,0x381,0x37b,0x381,0x37b,0x37e,0x378,0x381,0x37b,0x381,0x37b,0x381,0x37b, +0x37e,0x378,0x381,0x37b,0x9f0,0x9ea,0x381,0x37b,0x381,0x37b,0x381,0x37b,0x381,0x37b,0xd8c,0xd89, +0x381,0x37b,0xeca,0xec7,0xeca,0xec7,0xeca,0xec7,0xc00,0xbfd,0xc00,0xbfd,0xc00,0xbfd,0xc00,0xbfd, +0xc00,0xbfd,0xc00,0xbfd,0xc00,0xbfd,0xc00,0xbfd,0xef7,0xef4,0xef7,0xef4,0xfe7,0xfe4,0xfe7,0xfe4, +0xfe7,0xfe4,0xfe7,0xfe4,0xfe7,0xfe4,0xfe7,0xfe4,0xfe7,0xfe4,0xfe7,0xfe4,0x114f,0x114c,0x1329,0x1326, +0x14e5,0x14e2,0x14e5,0x14e2,0x14e5,0x14e2,0x14e5,0x14e2,0xc,0x393,0x393,0x393,0x393,0x393,0x393,0x393, +0x393,0x393,0x393,0x393,0x393,0x393,0x393,0x393,0x393,0x393,0x393,0x393,0x393,0x393,0x393,0x393, +0x393,0x393,0x393,0xc,0xc,0x396,0x384,0x384,0x384,0x38a,0x384,0x387,0x18ea,0x38d,0x38d,0x38d, +0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d, +0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x38d,0x390, +0x18ea,0x399,0x9f3,0xc,0xc,0x14b2,0x14b2,0x13ce,0xf,0x960,0x960,0x960,0x960,0x960,0x960,0x960, +0x960,0x960,0x960,0x960,0x960,0x960,0x960,0x960,0x960,0x960,0xd8f,0x960,0x960,0x960,0x960,0x960, +0x960,0x960,0x960,0x960,0x960,0x960,0x960,0x960,0x39c,0x39c,0x39c,0x39c,0x39c,0x39c,0x39c,0x39c, +0x39c,0x39c,0xecd,0x39c,0x39c,0x39c,0x3a8,0x39c,0x39f,0x39c,0x39c,0x3ab,0x963,0xd92,0xd95,0xd92, +0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae, +0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae,0x3ae, +0x3ae,0x3ae,0x3ae,0xf,0xf,0xf,0xf,0x18ed,0x3ae,0x3ae,0x3ae,0x3a5,0x3a2,0xf,0xf,0xf, +0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xca8,0xca8,0xca8,0xca8,0x13d1,0x14b5,0xf75,0xf75, +0xf75,0xf72,0xf72,0xd9b,0x8a6,0xca2,0xc9f,0xc9f,0xc96,0xc96,0xc96,0xc96,0xc96,0xc96,0xf6f,0xf6f, +0xf6f,0xf6f,0xf6f,0x8a3,0x14ac,0x1afd,0xd9e,0x8a9,0x12f0,0x3c9,0x3cc,0x3cc,0x3cc,0x3cc,0x3cc,0x3c9, +0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9, +0x3c9,0x3c9,0x3c9,0xf78,0xf78,0xf78,0xf78,0xf78,0x8ac,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9, +0x3c9,0x3c9,0x3c9,0x924,0x924,0x924,0x924,0x924,0x924,0x924,0x924,0xb3a,0xb3a,0xb3a,0xc96,0xc9c, +0xc99,0xd98,0xd98,0xd98,0xd98,0xd98,0xd98,0x12ed,0x93f,0x93f,0x93f,0x93f,0x93f,0x93f,0x93f,0x93f, +0x93f,0x93f,0x3c3,0x3c0,0x3bd,0x3ba,0xb9d,0xb9d,0x921,0x3c9,0x3c9,0x3d5,0x3c9,0x3cf,0x3cf,0x3cf, +0x3cf,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9, +0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9, +0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9, +0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x9f9,0x9f9,0x3c9,0x3c9, +0x3c9,0x3c9,0x3c9,0x9f9,0x3cc,0x3c9,0x3cc,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9,0x3c9, +0x3c9,0x3c9,0x3c9,0x9f9,0x3c9,0x3c9,0x3c9,0x3cc,0x942,0x3c9,0x3b4,0x3b4,0x3b4,0x3b4,0x3b4,0x3b4, +0x3b4,0x3b1,0x3ba,0x3b7,0x3b7,0x3b4,0x3b4,0x3b4,0x3b4,0x3d2,0x3d2,0x3b4,0x3b4,0x3ba,0x3b7,0x3b7, +0x3b7,0x3b4,0xca5,0xca5,0x3c6,0x3c6,0x3c6,0x3c6,0x3c6,0x3c6,0x3c6,0x3c6,0x3c6,0x3c6,0x9f9,0x9f9, +0x9f9,0x9f6,0x9f6,0xca5,0xa0e,0xa0e,0xa0e,0xa08,0xa08,0xa08,0xa08,0xa08,0xa08,0xa08,0xa08,0xa05, +0xa08,0xa05,0x12,0xa11,0xa0b,0x9fc,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b, +0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b,0xa0b, +0xa0b,0xcab,0xcab,0xcab,0xa02,0xa02,0xa02,0xa02,0xa02,0xa02,0xa02,0xa02,0xa02,0xa02,0xa02,0xa02, +0xa02,0xa02,0xa02,0xa02,0x9ff,0x9ff,0x9ff,0x9ff,0x9ff,0x9ff,0x9ff,0x9ff,0x9ff,0x9ff,0x9ff,0x12, +0x12,0xcab,0xcab,0xcab,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb, +0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb,0xdfb, +0xdfb,0xdfb,0xff9,0xff9,0xff9,0xff9,0xff9,0xff9,0xff9,0xff9,0xff9,0xff9,0xff9,0xff9,0xff9,0xff9, +0xff9,0xff9,0xff9,0xff9,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17, +0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17, +0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa17,0xa14,0xa14,0xa14,0xa14,0xa14,0xa14, +0xa14,0xa14,0xa14,0xa14,0xa14,0xba0,0x15,0x15,0x15,0x15,0x15,0x15,0x15,0x15,0x15,0x15, +0x15,0x15,0x15,0x15,0xf0f,0xf0f,0xf0f,0xf0f,0xf0f,0xf0f,0xf0f,0xf0f,0xf0f,0xf0f,0xf12,0xf12, +0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12, +0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf12,0xf06, +0xf06,0xf06,0xf06,0xf06,0xf06,0xf06,0xf06,0xf06,0xf15,0xf15,0xf09,0xf09,0xf0c,0xf1b,0xf18,0x102, +0x102,0x1911,0x1914,0x1914,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xb13,0xb13,0xb16,0xb16,0xb13,0xb13,0xb13,0xb13,0xb13,0xb13,0xb13,0xb13, +0x6f,0x6f,0x6f,0x6f,0x1e0,0x1e0,0x1e0,0x1e0,0x1e0,0x1e0,0x1e0,0x1617,0x1617,0x1617,0x1617,0x1617, +0x1617,0x1617,0x1617,0x1617,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1ec,0x1ec,0x1ec,0x1ec,0x1ec,0x1ec,0x1ec,0x1ec,0x1ec,0x1650,0x1650,0x1650, +0x1650,0x1650,0x1650,0x1650,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x16b,0x16b,0x16b, +0x16b,0x16b,0x16b,0x16b,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1bdb,0x1bd8,0x1c2,0x1c2,0x1c2,0x1c2,0x1c2,0x1c2,0x1c2,0x1c2,0x1c2,0x1c2, +0x1c2,0x1c2,0x1c2,0x1c2,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1488,0x1488,0x1488,0x1488,0x1488,0x1488,0x1488,0x1488,0x1488,0x1488,0x1a7,0x1a7, +0x1a7,0x1a7,0x1a7,0x1a7,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1b5a,0x1b5a,0x1b5a,0x1b5a,0x1b5a,0x1b5a,0x1b5a,0x201,0x201,0x201,0x201,0x201, +0x201,0x201,0x201,0x201,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x249,0x249,0x249,0x249,0x249,0x249,0x249,0x249,0x249,0x249,0x249,0x249, +0x249,0x249,0x249,0x249,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x195f,0x195f,0x195f,0x195f,0x195f,0x195f,0x195f,0x195f,0x195f,0x195f,0x24f,0x24f, +0x24f,0x24f,0x24f,0x24f,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1ac1,0x28b,0x28b,0x28b,0x28b,0x28b,0x28b,0x28b,0x28b,0x28b,0x28b,0x28b, +0x28b,0x28b,0x28b,0x28b,0x1752,0x1752,0x1752,0x1752,0x207,0x207,0x207,0x207,0x207,0x207,0x207,0x207, +0x207,0x207,0x207,0x207,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c, +0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e, +0x163e,0x163e,0x163e,0x163e,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1bc3,0x1bc3,0x1bc3,0x1bc3,0x2a0,0x1bc3,0x1bc3,0x1bc3,0x1bc3,0x1bc3,0x1bc3,0x1bc3, +0x2a0,0x1bc3,0x1bc3,0x2a0,0x16b6,0x16b6,0x16b6,0x16b6,0x1ef,0x1ef,0x1ef,0x1ef,0x1ef,0x1ef,0x1ef,0x1ef, +0x1ef,0x1ef,0x1ef,0x1ef,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x2b5,0x2b5,0x2b5,0x2b5,0x2b5,0x2b5,0x2b5,0x2b5, +0x2b5,0x2b5,0x2b5,0x2b5,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0xdf2,0xdf2,0xdef,0xdef,0xdef,0xdf2,0xd5,0xd5,0xd5,0xd5,0xd5,0xd5, +0xd5,0xd5,0xd5,0xd5,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x213,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a, +0x176a,0x176a,0x176a,0x176a,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb, +0x2bb,0x2bb,0x2bb,0x1bf6,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1a13,0x1a13,0x1a13,0x1a13,0x1a13,0x1a13,0x1a13,0x1a13,0x1a13,0x1a13,0x270,0x270, +0x270,0x270,0x1a16,0x1a10,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1b93,0x1b93,0x1b93,0x1b93,0x1b93,0x1b93,0x1b93,0x1b93,0x1b93,0x1b93,0x1b93,0x1b93, +0x1b93,0x1b93,0x1b93,0x1b93,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x1c26,0x1c26,0x1c26,0x1c26,0x1c26,0x1c26,0x1c26,0x1c26,0x1c26,0x1c26,0x1c26,0x1c26, +0x1c26,0x1c26,0x1c26,0x1c26,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0x255,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974, +0x1974,0x1974,0x1974,0x1974,0x273,0x273,0x273,0x273,0x273,0x273,0x273,0x273,0x273,0x273,0x273,0x273, +0x273,0x273,0x273,0x273,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, +0,0,0x95d,0x95d,3,3,3,3,3,3,3,3,3,3,3,3, +3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, +3,3,3,3,3,3,0x95d,0x95d,6,6,6,6,6,6,6,6, +6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, +6,6,6,6,6,6,6,6,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50, +0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,6,6,6,6,6,6,6,6, +6,6,6,6,6,6,6,6,0x14bb,0x3f0,0x3ff,0x3ff,0x18,0x405,0x405,0x405, +0x405,0x405,0x405,0x405,0x405,0x18,0x18,0x405,0x405,0x18,0x18,0x405,0x405,0x405,0x405,0x405, +0x405,0x405,0x405,0x405,0x405,0x405,0x405,0x405,0x405,0x18,0x405,0x405,0x405,0x405,0x405,0x405, +0x405,0x18,0x405,0x18,0x18,0x18,0x405,0x405,0x405,0x405,0x18,0x18,0x3f3,0xcb1,0x3f0,0x3ff, +0x3ff,0x3f0,0x3f0,0x3f0,0x3f0,0x18,0x18,0x3ff,0x3ff,0x18,0x18,0x402,0x402,0x3f6,0xda4,0x18, +0x18,0x18,0x18,0x18,0x18,0x18,0x18,0x3f0,0x18,0x18,0x18,0x18,0x408,0x408,0x18,0x408, +0x405,0x405,0x3f0,0x3f0,0x18,0x18,0x948,0x948,0x948,0x948,0x948,0x948,0x948,0x948,0x948,0x948, +0x405,0x405,0x3fc,0x3fc,0x3f9,0x3f9,0x3f9,0x3f9,0x3f9,0x3fc,0x3f9,0x1125,0x184b,0x1848,0x18f0,0x18, +0x1b,0xcb4,0x40b,0xcb7,0x1b,0x417,0x417,0x417,0x417,0x417,0x417,0x1b,0x1b,0x1b,0x1b,0x417, +0x417,0x1b,0x1b,0x417,0x417,0x417,0x417,0x417,0x417,0x417,0x417,0x417,0x417,0x417,0x417,0x417, +0x417,0x1b,0x417,0x417,0x417,0x417,0x417,0x417,0x417,0x1b,0x417,0x41a,0x1b,0x417,0x41a,0x1b, +0x417,0x417,0x1b,0x1b,0x40e,0x1b,0x414,0x414,0x414,0x40b,0x40b,0x1b,0x1b,0x1b,0x1b,0x40b, +0x40b,0x1b,0x1b,0x40b,0x40b,0x411,0x1b,0x1b,0x1b,0xf81,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b, +0x1b,0x41a,0x41a,0x41a,0x417,0x1b,0x41a,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x94b,0x94b, +0x94b,0x94b,0x94b,0x94b,0x94b,0x94b,0x94b,0x94b,0x40b,0x40b,0x417,0x417,0x417,0xf81,0x18f3,0x1b, +0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1b,0x1e,0x41d,0x41d,0x426,0x1e,0x429,0x429,0x429, +0x429,0x429,0x429,0x429,0xcc0,0x429,0x1e,0x429,0x429,0x429,0x1e,0x429,0x429,0x429,0x429,0x429, +0x429,0x429,0x429,0x429,0x429,0x429,0x429,0x429,0x429,0x1e,0x429,0x429,0x429,0x429,0x429,0x429, +0x429,0x1e,0x429,0x429,0x1e,0x429,0x429,0x429,0x429,0x429,0x1e,0x1e,0x420,0x429,0x426,0x426, +0x426,0x41d,0x41d,0x41d,0x41d,0x41d,0x1e,0x41d,0x41d,0x426,0x1e,0x426,0x426,0x423,0x1e,0x1e, +0x429,0x1e,0x1e,0x1e,0x1e,0x1e,0x1e,0x1e,0x1e,0x1e,0x1e,0x1e,0x1e,0x1e,0x1e,0x1e, +0x429,0xcc0,0xcba,0xcba,0x1e,0x1e,0x94e,0x94e,0x94e,0x94e,0x94e,0x94e,0x94e,0x94e,0x94e,0x94e, +0x13d4,0xcbd,0x1e,0x1e,0x1e,0x1e,0x1e,0x1e,0x1e,0x16cb,0x184e,0x184e,0x184e,0x1851,0x1851,0x1851, +0x21,0x42c,0x43b,0x43b,0x21,0x441,0x441,0x441,0x441,0x441,0x441,0x441,0x441,0x21,0x21,0x441, +0x441,0x21,0x21,0x441,0x441,0x441,0x441,0x441,0x441,0x441,0x441,0x441,0x441,0x441,0x441,0x441, +0x441,0x21,0x441,0x441,0x441,0x441,0x441,0x441,0x441,0x21,0x441,0x441,0x21,0xcc3,0x441,0x441, +0x441,0x441,0x21,0x21,0x42f,0x441,0x42c,0x42c,0x43b,0x42c,0x42c,0x42c,0xf84,0x21,0x21,0x43b, +0x43e,0x21,0x21,0x43e,0x43e,0x432,0x21,0x21,0x21,0x21,0x21,0x21,0x21,0x1a5b,0x42c,0x42c, +0x21,0x21,0x21,0x21,0x444,0x444,0x21,0x441,0x441,0x441,0xf84,0xf84,0x21,0x21,0x438,0x438, +0x438,0x438,0x438,0x438,0x438,0x438,0x438,0x438,0x435,0xcc3,0x12fc,0x12fc,0x12fc,0x12fc,0x12fc,0x12fc, +0x21,0x21,0x21,0x21,0x21,0x21,0x21,0x21,0x24,0x24,0x447,0x453,0x24,0x453,0x453,0x453, +0x453,0x453,0x453,0x24,0x24,0x24,0x453,0x453,0x453,0x24,0x453,0x453,0x456,0x453,0x24,0x24, +0x24,0x453,0x453,0x24,0x453,0x24,0x453,0x453,0x24,0x24,0x24,0x453,0x453,0x24,0x24,0x24, +0x453,0x453,0x453,0x24,0x24,0x24,0x453,0x453,0x453,0x453,0x453,0x453,0x453,0x453,0xda7,0x453, +0x453,0x453,0x24,0x24,0x24,0x24,0x447,0x44d,0x447,0x44d,0x44d,0x24,0x24,0x24,0x44d,0x44d, +0x44d,0x24,0x450,0x450,0x450,0x44a,0x24,0x24,0xf87,0x24,0x24,0x24,0x24,0x24,0x24,0x447, +0x24,0x24,0x24,0x24,0x24,0x24,0x24,0x24,0x24,0x24,0xebb,0x954,0x954,0x954,0x954,0x954, +0x954,0x954,0x954,0x954,0x951,0x951,0x951,0xd77,0xcc6,0xcc6,0xcc6,0xcc6,0xcc6,0xcc9,0xcc6,0x24, +0x24,0x24,0x24,0x24,0x14be,0x465,0x465,0x465,0x18f6,0x468,0x468,0x468,0x468,0x468,0x468,0x468, +0x468,0x27,0x468,0x468,0x468,0x27,0x468,0x468,0x468,0x468,0x468,0x468,0x468,0x468,0x468,0x468, +0x468,0x468,0x468,0x468,0x468,0x27,0x468,0x468,0x468,0x468,0x468,0x468,0x468,0x468,0x468,0x468, +0x14c1,0x468,0x468,0x468,0x468,0x468,0x27,0x27,0x1b00,0xf90,0x459,0x459,0x459,0x465,0x465,0x465, +0x465,0x27,0x459,0x459,0x45c,0x27,0x459,0x459,0x459,0x45f,0x27,0x27,0x27,0x27,0x27,0x27, +0x27,0x459,0x459,0x27,0xf90,0xf90,0x16ce,0x27,0x27,0x1b03,0x27,0x27,0x468,0x468,0xf8a,0xf8a, +0x27,0x27,0x462,0x462,0x462,0x462,0x462,0x462,0x462,0x462,0x462,0x462,0x27,0x27,0x27,0x27, +0x27,0x27,0x27,0x19bf,0xf8d,0xf8d,0xf8d,0xf8d,0xf8d,0xf8d,0xf8d,0xf8d,0x178e,0x14c4,0x471,0x471, +0x18f9,0x477,0x477,0x477,0x477,0x477,0x477,0x477,0x477,0x2a,0x477,0x477,0x477,0x2a,0x477,0x477, +0x477,0x477,0x477,0x477,0x477,0x477,0x477,0x477,0x477,0x477,0x477,0x477,0x477,0x2a,0x477,0x477, +0x477,0x477,0x477,0x477,0x477,0x477,0x477,0x477,0x2a,0x477,0x477,0x477,0x477,0x477,0x2a,0x2a, +0xccc,0xccf,0x471,0x46b,0x474,0x471,0x46b,0x471,0x471,0x2a,0x46b,0x474,0x474,0x2a,0x474,0x474, +0x46b,0x46e,0x2a,0x2a,0x2a,0x2a,0x2a,0x2a,0x2a,0x46b,0x46b,0x2a,0x2a,0x2a,0x2a,0x2a, +0x2a,0x1b06,0x477,0x2a,0x477,0x477,0xed3,0xed3,0x2a,0x2a,0x957,0x957,0x957,0x957,0x957,0x957, +0x957,0x957,0x957,0x957,0x2a,0xed6,0xed6,0x1bc9,0x2a,0x2a,0x2a,0x2a,0x2a,0x2a,0x2a,0x2a, +0x2a,0x2a,0x2a,0x2a,0x1854,0x14c7,0x483,0x483,0x1a5e,0x489,0x489,0x489,0x489,0x489,0x489,0x489, +0x489,0x2d,0x489,0x489,0x489,0x2d,0x489,0x489,0x489,0x489,0x489,0x489,0x489,0x489,0x489,0x489, +0x489,0x489,0x489,0x489,0x483,0x47a,0x47a,0x47a,0xf93,0x2d,0x483,0x483,0x483,0x2d,0x486,0x486, +0x486,0x47d,0x1302,0x1791,0x2d,0x2d,0x2d,0x2d,0x1794,0x1794,0x1794,0x47a,0x1791,0x1791,0x1791,0x1791, +0x1791,0x1791,0x1791,0x16d1,0x489,0x489,0xf93,0xf93,0x2d,0x2d,0x480,0x480,0x480,0x480,0x480,0x480, +0x480,0x480,0x480,0x480,0xf96,0xf96,0xf96,0xf96,0xf96,0xf96,0x1791,0x1791,0x1791,0xf99,0xf9c,0xf9c, +0xf9c,0xf9c,0xf9c,0xf9c,0x30,0x1a61,0xa23,0xa23,0x30,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29, +0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0x30,0x30,0x30,0xa29,0xa29, +0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29, +0xa29,0xa29,0x30,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0x30,0xa29,0x30,0x30, +0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0xa29,0x30,0x30,0x30,0xa1d,0x30,0x30,0x30,0x30,0xa1a, +0xa23,0xa23,0xa1a,0xa1a,0xa1a,0x30,0xa1a,0x30,0xa23,0xa23,0xa26,0xa23,0xa26,0xa26,0xa26,0xa1a, +0x30,0x30,0x30,0x30,0x30,0x30,0x14ca,0x14ca,0x14ca,0x14ca,0x14ca,0x14ca,0x14ca,0x14ca,0x14ca,0x14ca, +0x30,0x30,0xa23,0xa23,0xa20,0x30,0x30,0x30,0x30,0x30,0x30,0x30,0x30,0x30,0x30,0x30, +0x33,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4, +0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4,0x4a4, +0x4a4,0x48f,0x4a4,0x4a1,0x48f,0x48f,0x48f,0x48f,0x48f,0x48f,0x495,0x33,0x33,0x33,0x33,0x48c, +0x4aa,0x4aa,0x4aa,0x4aa,0x4aa,0x4a4,0x4a7,0x492,0x492,0x492,0x492,0x492,0x492,0x48f,0x492,0x498, +0x49e,0x49e,0x49e,0x49e,0x49e,0x49e,0x49e,0x49e,0x49e,0x49e,0x49b,0x49b,0x33,0x33,0x33,0x33, +0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33, +0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x33,0x36,0x4b9,0x4b9,0x36, +0x4b9,0x36,0x19c5,0x4b9,0x4b9,0x19c5,0x4b9,0x36,0x19c5,0x4b9,0x19c5,0x19c5,0x19c5,0x19c5,0x19c5,0x19c5, +0x4b9,0x4b9,0x4b9,0x4b9,0x19c5,0x4b9,0x4b9,0x4b9,0x4b9,0x4b9,0x4b9,0x4b9,0x19c5,0x4b9,0x4b9,0x4b9, +0x36,0x4b9,0x36,0x4b9,0x19c5,0x19c5,0x4b9,0x4b9,0x19c5,0x4b9,0x4b9,0x4b9,0x4b9,0x4ad,0x4b9,0x4b6, +0x4ad,0x4ad,0x4ad,0x4ad,0x4ad,0x4ad,0x19c2,0x4ad,0x4ad,0x4b9,0x36,0x36,0x4c2,0x4c2,0x4c2,0x4c2, +0x4c2,0x36,0x4bf,0x36,0x4b0,0x4b0,0x4b0,0x4b0,0x4b0,0x4ad,0x1bcc,0x36,0x4b3,0x4b3,0x4b3,0x4b3, +0x4b3,0x4b3,0x4b3,0x4b3,0x4b3,0x4b3,0x36,0x36,0x4bc,0x4bc,0x13d7,0x13d7,0x36,0x36,0x36,0x36, +0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36, +0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x36,0x99c,0x99c,0x99c,0x99f, +0x99c,0x99c,0x99c,0x99c,0x39,0x99c,0x99c,0x99c,0x99c,0x99f,0x99c,0x99c,0x99c,0x99c,0x99f,0x99c, +0x99c,0x99c,0x99c,0x99f,0x99c,0x99c,0x99c,0x99c,0x99f,0x99c,0x99c,0x99c,0x99c,0x99c,0x99c,0x99c, +0x99c,0x99c,0x99c,0x99c,0x99c,0x99f,0xa38,0xfa8,0xfa8,0x39,0x39,0x39,0x39,0x966,0x966,0x969, +0x966,0x969,0x969,0x975,0x969,0x975,0x966,0x966,0x966,0x966,0x966,0x996,0x966,0x969,0x96f,0x96f, +0x972,0x97b,0x96c,0x96c,0x99c,0x99c,0x99c,0x99c,0x130b,0x1305,0x1305,0x1305,0x966,0x966,0x966,0x969, +0x966,0x966,0xa2c,0x966,0x39,0x966,0x966,0x966,0x966,0x969,0x966,0x966,0x966,0x966,0x969,0x966, +0x966,0x966,0x966,0x969,0x966,0x966,0x966,0x966,0x969,0x966,0xa2c,0xa2c,0xa2c,0x966,0x966,0x966, +0x966,0x966,0x966,0x966,0xa2c,0x969,0xa2c,0xa2c,0xa2c,0x39,0xa35,0xa35,0xa32,0xa32,0xa32,0xa32, +0xa32,0xa32,0xa2f,0xa32,0xa32,0xa32,0xa32,0xa32,0xa32,0x39,0xf9f,0xa32,0xdaa,0xdaa,0xfa2,0xfa5, +0xf9f,0x1128,0x1128,0x1128,0x1128,0x1308,0x1308,0x39,0x39,0x39,0x39,0x39,0x39,0x39,0x39,0x39, +0x39,0x39,0x39,0x39,0x39,0x39,0x39,0x39,0x39,0x39,0x39,0x39,0x39,0x39,0x39,0x39, +0x39,0x39,0x39,0x39,0x39,0x39,0x39,0x39,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x3c,0x13dd, +0x3c,0x3c,0x3c,0x3c,0x3c,0x13dd,0x3c,0x3c,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5, +0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xdb9, +0xa62,0x3f,0xa62,0xa62,0xa62,0xa62,0x3f,0x3f,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0x3f, +0xa62,0x3f,0xa62,0xa62,0xa62,0xa62,0x3f,0x3f,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xdb9, +0xa62,0x3f,0xa62,0xa62,0xa62,0xa62,0x3f,0x3f,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62, +0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xdb9,0xa62,0x3f,0xa62,0xa62, +0xa62,0xa62,0x3f,0x3f,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0x3f,0xa62,0x3f,0xa62,0xa62, +0xa62,0xa62,0x3f,0x3f,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xdb9,0xa62,0xa62,0xa62,0xa62, +0xa62,0xa62,0xa62,0x3f,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62, +0xa62,0xa62,0xa62,0xdb9,0xa62,0x3f,0xa62,0xa62,0xa62,0xa62,0x3f,0x3f,0xa62,0xa62,0xa62,0xa62, +0xa62,0xa62,0xa62,0xdb9,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62, +0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0x3f,0x3f,0x130e,0x130e,0xdb3,0xdb6,0xa5c,0xa65,0xa59, +0xa59,0xa59,0xa59,0xa65,0xa65,0xa5f,0xa5f,0xa5f,0xa5f,0xa5f,0xa5f,0xa5f,0xa5f,0xa5f,0xa56,0xa56, +0xa56,0xa56,0xa56,0xa56,0xa56,0xa56,0xa56,0xa56,0xa56,0x3f,0x3f,0x3f,0xa68,0xa68,0xa68,0xa68, +0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68, +0xa68,0x16d7,0x42,0x42,0x16d4,0x16d4,0x16d4,0x16d4,0x16d4,0x16d4,0x42,0x42,0xa7a,0xa7d,0xa7d,0xa7d, +0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa7d, +0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa7d,0xa77,0xa74,0x45,0x45,0x45,0xa83,0xa83,0xa83,0xa83, +0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa80,0xa80,0xa80,0xa83,0xa83,0xa83,0x14cd,0x14cd,0x14cd, +0x14cd,0x14cd,0x14cd,0x14cd,0x14cd,0x48,0x48,0x48,0x48,0x48,0x48,0x48,0xaa4,0xaa4,0xaa4,0xaa4, +0xaa4,0xaa4,0xa86,0xaa4,0xaa4,0xa89,0xa89,0xa89,0xa89,0xa89,0xa89,0xa89,0xa89,0xa89,0xa8c,0xa89, +0xa9b,0xa9b,0xa9e,0xaa7,0xa95,0xa92,0xa9b,0xa98,0xaa7,0xcd2,0x4b,0x4b,0xaa1,0xaa1,0xaa1,0xaa1, +0xaa1,0xaa1,0xaa1,0xaa1,0xaa1,0xaa1,0x4b,0x4b,0x4b,0x4b,0x4b,0x4b,0xcd5,0xcd5,0xcd5,0xcd5, +0xcd5,0xcd5,0xcd5,0xcd5,0xcd5,0xcd5,0x4b,0x4b,0x4b,0x4b,0x4b,0x4b,0xab6,0xab6,0xb2e,0xb31, +0xabc,0xb2b,0xab9,0xab6,0xabf,0xace,0xac2,0xad1,0xad1,0xad1,0xaad,0x1b09,0xac5,0xac5,0xac5,0xac5, +0xac5,0xac5,0xac5,0xac5,0xac5,0xac5,0x4e,0x4e,0x4e,0x4e,0x4e,0x4e,0xac8,0xac8,0xac8,0xac8, +0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8, +0xac8,0xac8,0xac8,0xac8,0x18fc,0x4e,0x4e,0x4e,0x4e,0x4e,0x4e,0x4e,0xac8,0xac8,0xac8,0xac8, +0xac8,0xac8,0xac8,0xac8,0xac8,0xab0,0xfc6,0x4e,0x4e,0x4e,0x4e,0x4e,0x117f,0x117f,0x117f,0x117f, +0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x4e6,0x4e6,0x4e6,0x4e6, +0x4e6,0x4e6,0x4e6,0x4e6,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e6,0x4e6,0x4e6,0x4e6, +0x4e6,0x4e6,0x51,0x51,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x51,0x51,0x4e6,0x4e6,0x4e6,0x4e6, +0x4e6,0x4e6,0x4e6,0x4e6,0x51,0x4e9,0x51,0x4e9,0x51,0x4e9,0x51,0x4e9,0x4e6,0x4e6,0x4e6,0x4e6, +0x4e6,0x4e6,0x4e6,0x4e6,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e6,0x4e6,0x4e6,0x4e6, +0x4e6,0x4e6,0x4e6,0x4e6,0x4e6,0x4e6,0x4e6,0x4e6,0x4e6,0x4e6,0x51,0x51,0x4e6,0x4e6,0x4e6,0x4e6, +0x4e6,0x4e6,0x4e6,0x4e6,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e6,0x4e6,0x4e6,0x4e6, +0x4e6,0x51,0x4e6,0x4e6,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e0,0x4e6,0x4e0,0x4e0,0x4dd,0x4e6,0x4e6, +0x4e6,0x51,0x4e6,0x4e6,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4dd,0x4dd,0x4dd,0x4e6,0x4e6,0x4e6,0x4e6, +0x51,0x51,0x4e6,0x4e6,0x4e9,0x4e9,0x4e9,0x4e9,0x51,0x4dd,0x4dd,0x4dd,0x4e6,0x4e6,0x4e6,0x4e6, +0x4e6,0x4e6,0x4e6,0x4e6,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4dd,0x4dd,0x4dd,0x51,0x51,0x4e6,0x4e6, +0x4e6,0x51,0x4e6,0x4e6,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e3,0x4e0,0x51,0xba6,0xba9,0xba9,0xba9, +0xfcf,0x54,0x14a9,0x14a9,0x14a9,0x14a9,0x4f2,0x4f2,0x4f2,0x4f2,0x4f2,0x4f2,0x53d,0xbbb,0x57,0x57, +0x6d8,0x53d,0x53d,0x53d,0x53d,0x53d,0x543,0x555,0x543,0x54f,0x549,0x6db,0x53a,0x6d5,0x6d5,0x6d5, +0x6d5,0x53a,0x53a,0x53a,0x53a,0x53a,0x540,0x552,0x540,0x54c,0x546,0x57,0xdc2,0xdc2,0xdc2,0xdc2, +0xdc2,0x1311,0x1311,0x1311,0x1311,0x1311,0x1311,0x1311,0x1311,0x57,0x57,0x57,0x1b0c,0x5a,0x5a,0x5a, +0x5a,0x5a,0x5a,0x5a,0x5a,0x5a,0x5a,0x5a,0x5a,0x5a,0x5a,0x5a,0x564,0x564,0x564,0x564, +0x564,0x564,0x564,0x564,0x564,0x564,0x564,0x564,0x564,0x561,0x561,0x561,0x561,0x564,0xadd,0xadd, +0xbc1,0xbc7,0xbc7,0xbc4,0xbc4,0xbc4,0xbc4,0xdc8,0xed9,0xed9,0xed9,0xed9,0x1113,0x5d,0x5d,0x5d, +0x5d,0x5d,0x5d,0x5d,0x5d,0x5d,0x5d,0x5d,0x5d,0x5d,0x5d,0x5d,0x594,0x594,0x594,0xae6, +0xee2,0xfd5,0xfd5,0xfd5,0xfd5,0x126f,0x16dd,0x16dd,0x60,0x60,0x60,0x60,0x702,0x702,0x702,0x702, +0x702,0x702,0x702,0x702,0x702,0x702,0x5a0,0x5a0,0x59d,0x59d,0x59d,0x59d,0x5c1,0x5c1,0x5c1,0x5c1, +0x5c1,0xaef,0xaef,0x63,0x63,0x63,0x63,0x63,0x63,0x63,0x63,0x63,0x63,0x63,0x63,0x63, +0x63,0x63,0x63,0x63,0x63,0x63,0x63,0x63,0x63,0x63,0x63,0x63,0x5c4,0x5c4,0x5c4,0x5c4, +0x5c4,0x5c4,0x5c4,0x5c4,0x5c4,0x5c4,0x5c4,0x66,0x66,0x66,0x66,0x66,0x66,0x66,0x66,0x66, +0x66,0x66,0x66,0x66,0x66,0x66,0x66,0x66,0x66,0x66,0x66,0x66,0xb0a,0xb0a,0xb0a,0xb0a, +0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a, +0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0x69,0xb0a,0xb0a,0xb0a,0xb0a,0xb0d,0xb0a,0xb0a,0xb0a,0xb0a, +0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0d, +0x69,0x69,0x69,0x69,0x69,0x69,0x69,0x69,0x69,0x69,0x69,0x69,0xb10,0xb10,0xb10,0xb10, +0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10, +0xb10,0xb10,0x6c,0x6c,0x6c,0x6c,0x6c,0x6c,0x6c,0x6c,0x6c,0x6c,0x72,0x822,0x81c,0x822, +0x81c,0x822,0x81c,0x822,0x81c,0x822,0x81c,0x81c,0x81f,0x81c,0x81f,0x81c,0x81f,0x81c,0x81f,0x81c, +0x81f,0x81c,0x81f,0x81c,0x81f,0x81c,0x81f,0x81c,0x81f,0x81c,0x81f,0x81c,0x81c,0x81c,0x81c,0x822, +0x81c,0x822,0x81c,0x822,0x81c,0x81c,0x81c,0x81c,0x81c,0x81c,0x822,0x81c,0x81c,0x81c,0x81c,0x81c, +0x81f,0xc60,0xc60,0x72,0x72,0x936,0x936,0x8fd,0x8fd,0x825,0x828,0xc5d,0x75,0x75,0x75,0x75, +0x75,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a, +0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x83a,0x1101,0x18c3,0x19aa, +0x78,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d, +0x83d,0x83d,0x83d,0x78,0x906,0x906,0x909,0x909,0x909,0x909,0x909,0x909,0x909,0x909,0x909,0x909, +0x909,0x909,0x909,0x909,0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846, +0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846,0x846, +0x846,0xd5c,0xd5c,0x7b,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22, +0xb22,0x7e,0x7e,0x7e,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28, +0xb28,0xb28,0xb28,0xb28,0xb28,0xc69,0xb28,0xb28,0xb28,0xc69,0xb28,0x81,0x81,0x81,0x81,0x81, +0x81,0x81,0x81,0x81,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6, +0x11a6,0x11a6,0x11a6,0x11a6,0x9c0,0x9c0,0x9c0,0x9c0,0x84,0x84,0x84,0x84,0x84,0x84,0x84,0x84, +0x84,0x84,0x84,0x84,0x121b,0x121b,0x121b,0x121b,0x121b,0x121b,0x121b,0x121b,0x121b,0x121b,0x121b,0x121b, +0x121b,0x121b,0x121b,0x121b,0x609,0x609,0x609,0x609,0x609,0x609,0x609,0x87,0x87,0x87,0x87,0x87, +0x87,0x87,0x87,0x87,0x87,0x87,0x87,0x5f7,0x5f7,0x5f7,0x5f7,0x5f7,0x87,0x87,0x87,0x87, +0x87,0xafb,0x5fa,0x600,0x606,0x606,0x606,0x606,0x606,0x606,0x606,0x606,0x606,0x5fd,0x600,0x600, +0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x600,0x87,0x600,0x600,0x600,0x600, +0x600,0x87,0x600,0x87,0x600,0x600,0x87,0x600,0x600,0x87,0x600,0x600,0x600,0x600,0x600,0x600, +0x600,0x600,0x600,0x603,0x615,0x60f,0x615,0x60f,0x612,0x618,0x615,0x60f,0x612,0x618,0x615,0x60f, +0x612,0x618,0x615,0x60f,0x1323,0x1323,0x1b0f,0x8a,0x8a,0x8a,0x8a,0x8a,0x8a,0x8a,0x8a,0x8a, +0x8a,0x8a,0x8a,0x8a,0x8a,0x8a,0x8a,0x615,0x60f,0x612,0x618,0x615,0x60f,0x615,0x60f,0x615, +0x60f,0x615,0x615,0x60f,0x60f,0x60f,0x60f,0x612,0x60f,0x60f,0x612,0x60f,0x612,0x612,0x612,0x60f, +0x612,0x612,0x612,0x612,0x8a,0x8a,0x612,0x612,0x612,0x612,0x60f,0x60f,0x612,0x60f,0x60f,0x60f, +0x60f,0x612,0x60f,0x60f,0x60f,0x60f,0x60f,0x612,0x612,0x612,0x60f,0x60f,0x8a,0x8a,0x8a,0x8a, +0x8a,0x8a,0x8a,0x1b0f,0xb46,0xb46,0xb46,0xb46,0xb46,0xb46,0xb46,0xb46,0xb46,0xb46,0xb46,0xb46, +0xb46,0xb46,0xb46,0xb46,0x85e,0x870,0x86d,0x870,0x86d,0xc7e,0xc7e,0xd68,0xd65,0x861,0x861,0x861, +0x861,0x873,0x873,0x873,0x88b,0x88e,0x89d,0x8d,0x891,0x894,0x8a0,0x8a0,0x888,0x87f,0x879,0x87f, +0x879,0x87f,0x879,0x87c,0x87c,0x897,0x897,0x89a,0x897,0x897,0x897,0x8d,0x897,0x885,0x882,0x87c, +0x8d,0x8d,0x8d,0x8d,0x621,0x62d,0x621,0xbfa,0x621,0x90,0x621,0x62d,0x621,0x62d,0x621,0x62d, +0x621,0x62d,0x621,0x62d,0x62d,0x62a,0x624,0x627,0x62d,0x62a,0x624,0x627,0x62d,0x62a,0x624,0x627, +0x62d,0x62a,0x624,0x62a,0x624,0x62a,0x624,0x627,0x62d,0x62a,0x624,0x62a,0x624,0x62a,0x624,0x62a, +0x624,0x90,0x90,0x61e,0x75f,0x762,0x777,0x77a,0x759,0x762,0x762,0x96,0x741,0x744,0x744,0x744, +0x744,0x741,0x741,0x96,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0xafe,0xafe,0xafe, +0x9c3,0x73b,0x630,0x630,0x96,0x789,0x768,0x759,0x762,0x75f,0x759,0x76b,0x75c,0x756,0x759,0x777, +0x76e,0x765,0x786,0x759,0x783,0x783,0x783,0x783,0x783,0x783,0x783,0x783,0x783,0x783,0x774,0x771, +0x777,0x777,0x777,0x789,0x74a,0x747,0x747,0x747,0x747,0x747,0x747,0x747,0x747,0x747,0x747,0x747, +0x747,0x747,0x747,0x747,0x747,0x747,0x747,0x747,0x747,0x747,0x747,0x747,0x747,0x747,0x747,0x747, +0x747,0x747,0x747,0x96,0x96,0x96,0x747,0x747,0x747,0x747,0x747,0x747,0x96,0x96,0x747,0x747, +0x747,0x747,0x747,0x747,0x96,0x96,0x747,0x747,0x747,0x747,0x747,0x747,0x96,0x96,0x747,0x747, +0x747,0x96,0x96,0x96,0xb49,0xb49,0xb49,0xb49,0x99,0x99,0x99,0x99,0x99,0x99,0x99,0x99, +0x99,0x1860,0x1860,0x1860,0xb4f,0xb4f,0xb4f,0xb4f,0xb4f,0xb4f,0xb4f,0xb4f,0xb4f,0xb4f,0xb4f,0xb4f, +0xb4f,0xb4f,0xb4f,0xb4f,0xb4f,0xb4f,0xb4f,0x9c,0x9c,0x9c,0x9c,0x9c,0x1626,0x1626,0x1626,0x1626, +0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0xb58,0xb58,0xb58,0xb58, +0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58, +0xb58,0xb58,0x9f,0x9f,0x9f,0x9f,0x9f,0x9f,0x9f,0x9f,0x9f,0x9f,0xb64,0xb64,0xb64,0xb64, +0xb64,0xb64,0xb64,0xa2,0xa2,0xfe1,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64, +0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0x16e3,0x16e3,0x16e3,0x16e3, +0x16e3,0x16e3,0x16e3,0x16e3,0x16e3,0x1b12,0x1b12,0xa2,0xa2,0xa2,0xa2,0xa2,0xa2,0xa2,0xa2,0xa2, +0xa2,0xa2,0xa2,0xa2,0xa2,0xa2,0xa2,0xa2,0xa2,0xa2,0xa2,0xa2,0xb7c,0xb7c,0xb7c,0xb7c, +0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79, +0xb79,0xa5,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb7c,0xb7c,0xb79,0xb79, +0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79, +0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb7c,0xa5,0xb7c,0xb7c,0xa5,0xa5,0xb7c,0xa5, +0xa5,0xb7c,0xb7c,0xa5,0xa5,0xb7c,0xb7c,0xb7c,0xb7c,0xa5,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c, +0xb7c,0xb7c,0xb79,0xb79,0xb79,0xb79,0xa5,0xb79,0xa5,0xb79,0xb79,0xb79,0xb79,0xcf0,0xb79,0xb79, +0xa5,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb7c,0xb7c,0xb7c,0xb7c, +0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb79,0xb79,0xb79,0xb79, +0xb7c,0xb7c,0xa5,0xb7c,0xb7c,0xb7c,0xb7c,0xa5,0xa5,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c, +0xb7c,0xa5,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xa5,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79, +0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79, +0xb79,0xb79,0xb79,0xb79,0xb7c,0xb7c,0xa5,0xb7c,0xb7c,0xb7c,0xb7c,0xa5,0xb7c,0xb7c,0xb7c,0xb7c, +0xb7c,0xa5,0xb7c,0xa5,0xa5,0xa5,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xa5,0xb79,0xb79, +0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xdda,0xdda,0xa5,0xa5, +0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c, +0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb79,0xb79,0xb79,0xb73,0xb79,0xb79,0xb79,0xb79, +0xb79,0xb79,0xef1,0xeee,0xa5,0xa5,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76, +0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xa8,0xb82,0xa8,0xa8,0xa8,0xa8,0xa8,0xa8, +0xa8,0xa8,0xa8,0xa8,0xa8,0xa8,0xa8,0xa8,0xa8,0xa8,0xa8,0xa8,0xa8,0xa8,0xa8,0xa8, +0xa8,0xa8,0xa8,0xa8,0xa8,0xa8,0xa8,0xa8,0xc09,0xc09,0xc09,0xc09,0xc09,0xc09,0xc09,0xc09, +0xc09,0xc09,0xc09,0xc09,0xc09,0x1b18,0xc09,0xc09,0xc09,0xc09,0xc03,0xc03,0xc06,0x1b15,0xab,0xab, +0xab,0xab,0xab,0xab,0xab,0xab,0xab,0x1b18,0xc12,0xc12,0xc12,0xc12,0xc12,0xc12,0xc12,0xc12, +0xc12,0xc12,0xc12,0xc12,0xc12,0xc12,0xc12,0xc12,0xc12,0xc12,0xc0c,0xc0c,0xc0f,0xc72,0xc72,0xae, +0xae,0xae,0xae,0xae,0xae,0xae,0xae,0xae,0xc18,0xc18,0xc18,0xc18,0xc18,0xc18,0xc18,0xc18, +0xc18,0xc18,0xc18,0xc18,0xc18,0xc18,0xc18,0xc18,0xc18,0xc18,0xc15,0xc15,0xb1,0xb1,0xb1,0xb1, +0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xb1,0xc1e,0xc1e,0xc1e,0xc1e,0xc1e,0xc1e,0xc1e,0xc1e, +0xc1e,0xc1e,0xc1e,0xc1e,0xc1e,0xb4,0xc1e,0xc1e,0xc1e,0xb4,0xc1b,0xc1b,0xb4,0xb4,0xb4,0xb4, +0xb4,0xb4,0xb4,0xb4,0xb4,0xb4,0xb4,0xb4,0xd02,0xd02,0xd02,0xd02,0xd02,0xd02,0xd02,0xd02, +0xd02,0xd02,0xd02,0xd02,0xd02,0xd02,0xd02,0xd02,0xd02,0xd02,0xd02,0xd02,0xd02,0xd02,0xd02,0xd02, +0xd02,0xd02,0xd02,0xd02,0xd02,0x14e8,0x14e8,0xb7,0xcf3,0xcf3,0xcf3,0xcff,0xcff,0xcff,0xcff,0xcf3, +0xcf3,0xcff,0xcff,0xcff,0xb7,0xb7,0xb7,0xb7,0xcff,0xcff,0xcf3,0xcff,0xcff,0xcff,0xcff,0xcff, +0xcff,0xcf6,0xcf6,0xcf6,0xb7,0xb7,0xb7,0xb7,0xcf9,0xb7,0xb7,0xb7,0xd05,0xd05,0xcfc,0xcfc, +0xcfc,0xcfc,0xcfc,0xcfc,0xcfc,0xcfc,0xcfc,0xcfc,0xd08,0xd08,0xd08,0xd08,0xd08,0xd08,0xd08,0xd08, +0xd08,0xd08,0xd08,0xd08,0xd08,0xd08,0xd08,0xd08,0xd08,0xd08,0xba,0xba,0xd08,0xd08,0xd08,0xd08, +0xd08,0xba,0xba,0xba,0xba,0xba,0xba,0xba,0xba,0xba,0xba,0xba,0x14eb,0x14eb,0x14eb,0x14eb, +0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb, +0xbd,0xbd,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb, +0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0xbd,0x1a64,0x14eb,0x14eb,0x14eb,0x14eb, +0x14eb,0x14eb,0x14eb,0x14eb,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c, +0xc0,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c, +0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xc0,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c, +0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xc0,0xd2c,0xd2c,0xc0,0xd2c, +0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xc0,0xc0, +0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xd2c,0xc0,0xc0, +0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0, +0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0,0xc0, +0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f, +0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xc3,0xc3,0xc3,0xc3,0xc3, +0xd6e,0xd6e,0xd74,0xc6,0xc6,0xc6,0xc6,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b, +0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b,0xd6b, +0xc6,0xc6,0xc6,0xd71,0xd71,0xd71,0xd71,0xd71,0xd71,0xd71,0xd71,0xd71,0xd35,0xd35,0xd35,0xd35, +0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35, +0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xd35,0xc9,0xd32,0xd3e,0xd3e,0xd3e,0xd3e, +0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e, +0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xd3e,0xcc,0xcc,0xd3b,0xd3b,0xd3b,0xd3b, +0xd3b,0xd3b,0xd3b,0xd3b,0xd3b,0xd3b,0xcc,0xcc,0xcc,0xcc,0xcc,0xcc,0x1824,0x1824,0x1824,0x1824, +0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0xd41,0xd41,0xd41,0xd41, +0xd41,0xd41,0xcf,0xcf,0xd41,0xcf,0xd41,0xd41,0xd41,0xd41,0xd41,0xd41,0xd41,0xd41,0xd41,0xd41, +0xd41,0xd41,0xd41,0xd41,0xd41,0xd41,0xd41,0xd41,0xd41,0xd41,0xd41,0xd41,0xd41,0xd41,0xcf,0xd41, +0xd41,0xcf,0xcf,0xcf,0xd41,0xcf,0xcf,0xd41,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44, +0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd2, +0xd2,0xd2,0xd2,0xd2,0xd2,0xd2,0xd2,0xd2,0xdf5,0xdf5,0xdf5,0xdf5,0xdf5,0xdf5,0xdf5,0xdf5, +0xdf5,0xdf5,0xdf5,0x14ee,0x14ee,0x179a,0x179a,0xd8,0x10e0,0x10e0,0x10e0,0x10e0,0x10e0,0x10e0,0x10e0,0x10e0, +0x10e0,0x10e0,0x10e0,0x10e0,0x1a73,0x129,0x129,0x129,0xe07,0xe07,0xe07,0xe07,0xe07,0xe07,0xe07,0xe07, +0xe07,0xe07,0xe07,0xe07,0xe07,0xe07,0xe07,0xe07,0xe07,0xe07,0xe07,0xe07,0xe07,0xe07,0xe07,0xdfe, +0xdfe,0xe04,0xe04,0xdfe,0xdb,0xdb,0xe01,0xe01,0x1110,0x1110,0x1110,0x1110,0xde,0xde,0xde,0xde, +0xde,0xde,0xde,0xde,0xde,0xde,0xde,0xde,0xc6f,0xc6f,0xc6f,0xc6f,0xc6f,0xc6f,0xc6f,0xc6f, +0xc6f,0xc6f,0xc6f,0xc6f,0xc6f,0xc6f,0xc6f,0xc6f,0xe19,0xe16,0xe19,0xe16,0xe16,0xe0d,0xe0d,0xe0d, +0xe0d,0xe0d,0xe0d,0x115b,0x1158,0x115b,0x1158,0x1155,0x1155,0x1155,0x13e6,0x13e3,0xe1,0xe1,0xe1,0xe1, +0xe1,0xe13,0xe10,0xe10,0xe10,0xe0d,0xe13,0xe10,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c, +0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe4, +0xe4,0xe4,0xe4,0xe4,0xe4,0xe4,0xe4,0xe4,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe4, +0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe4,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe4, +0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe1c,0xe4,0xe22,0xe22,0xe22,0xe22,0xe22,0xe22,0xe22,0xe22, +0xe22,0xe22,0xe22,0xe22,0xe22,0xe22,0xe22,0xe22,0xe1f,0xe1f,0xe1f,0xe1f,0xe1f,0xe1f,0xe1f,0xe1f, +0xe1f,0xe1f,0xe7,0xe7,0xe7,0xe7,0xe7,0xe7,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xea,0x13e9, +0xea,0xea,0xea,0xea,0xea,0x13e9,0xea,0xea,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c, +0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe3d,0xe31,0xe31,0xe31,0xed,0xe31,0xe31,0xed, +0xed,0xed,0xed,0xed,0xe31,0xe31,0xe31,0xe31,0xe3d,0xe3d,0xe3d,0xe3d,0xed,0xe3d,0xe3d,0xe3d, +0xed,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d, +0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0xe3d,0x1905,0x1905,0xed,0xed,0xe2e,0xe2e,0xe2e,0xed, +0xed,0xed,0xed,0xe34,0xe37,0xe37,0xe37,0xe37,0xe37,0xe37,0xe37,0xe37,0x1902,0xed,0xed,0xed, +0xed,0xed,0xed,0xed,0xe3a,0xe3a,0xe3a,0xe3a,0xe3a,0xe3a,0xe40,0xe40,0xe37,0xed,0xed,0xed, +0xed,0xed,0xed,0xed,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0x1161,0x1161, +0xf0,0xf0,0xf0,0xf0,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4f,0xe4f,0xe4f,0xe4c,0xe4c,0xe4f,0xe4c, +0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xf0,0xf0,0xf0,0xf0,0xf0,0xf0, +0xe49,0xe49,0xe49,0xe49,0xe49,0xe49,0xe49,0xe49,0xe49,0xe49,0x115e,0xf0,0xf0,0xf0,0xe46,0xe46, +0xe55,0xe55,0xe55,0xe55,0xf3,0xf3,0xf3,0xf3,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55, +0xe52,0xe55,0xe55,0xe55,0xe55,0xe55,0xf3,0xf3,0xf3,0xf3,0xf3,0xf3,0xf3,0xf3,0xf3,0xf3, +0x14fd,0x1503,0x1500,0x1845,0x17a0,0x1869,0x1869,0x1869,0x1869,0x1869,0x190b,0x1908,0x190e,0x1908,0x190e,0x19cb, +0x1a67,0x1a67,0x1a67,0x1b2a,0x1b2a,0x1b24,0x1b21,0x1b24,0x1b21,0x1b24,0x1b21,0x1b24,0x1b21,0x1b27,0xf6,0xf6, +0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6, +0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6,0xf6, +0xe79,0xe79,0xe79,0xe76,0xe76,0xe6d,0xe6d,0xe76,0xe73,0xe73,0xe73,0xe73,0x1a6a,0xf9,0xf9,0xf9, +0x12cc,0x12cc,0x12cc,0x12cf,0x12cf,0x12cf,0x12c6,0x12c6,0x12c9,0x12c6,0x14d,0x14d,0x14d,0x14d,0x14d,0x14d, +0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0x13f5,0x13f5,0xfc,0xfc,0xfc,0xfc,0xfc,0xfc,0xfc,0xe7f, +0x1335,0xfc,0xfc,0xfc,0xfc,0xfc,0xfc,0xfc,0xfc,0xfc,0xfc,0xfc,0xfc,0xfc,0xfc,0x1332, +0xc42,0xc42,0xc42,0xc42,0xc42,0xc42,0xc42,0xc42,0xc42,0xc42,0xc42,0xc42,0xc42,0xc42,0xc42,0xc42, +0xeac,0xe9d,0xe97,0xea9,0xea6,0xea0,0xea0,0xeaf,0xe9a,0xea3,0xff,0xff,0xff,0xff,0xff,0xff, +0xf33,0xf33,0xf1e,0xf33,0xf36,0xf39,0xf39,0xf39,0xf39,0xf39,0xf39,0xf39,0x1b2d,0x105,0x105,0x105, +0xf2d,0xf2d,0xf2d,0xf2d,0xf2d,0xf2d,0xf2d,0xf2d,0xf2d,0xf2d,0xf3f,0xf3f,0xf24,0xf2a,0xf3f,0xf3f, +0xf27,0xf24,0xf24,0xf24,0xf24,0xf24,0xf24,0xf24,0xf24,0xf24,0xf24,0xf21,0xf21,0xf21,0xf21,0xf21, +0xf21,0xf21,0xf21,0xf21,0xf24,0xf24,0xf24,0xf24,0xf24,0xf24,0xf24,0xf24,0xf24,0x1b30,0x1b30,0x105, +0x1b39,0x1b33,0x19d1,0x19ce,0x19d1,0x19d1,0x19d1,0x1a70,0x1a6d,0x1a70,0x1a6d,0x108,0x108,0x108,0x108,0x108, +0x1b39,0x1b33,0x108,0x1b33,0x108,0x1b33,0x1b39,0x1b33,0x1b39,0x1b33,0x108,0x108,0x108,0x108,0x108,0x108, +0x108,0x108,0x108,0x108,0x108,0x108,0x108,0x108,0x108,0x108,0x108,0x108,0x108,0x108,0x1b36,0x1b36, +0x1b36,0x1a70,0x1a6d,0x150c,0x13fe,0x13fe,0x1338,0x103b,0x103b,0x103b,0x103b,0x103b,0xf4e,0xf4e,0xf4e,0xf4e, +0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e, +0xf4b,0xf4b,0xf51,0xf51,0x10b,0x10b,0x10b,0x10b,0x10b,0x10b,0x10b,0x10b,0xf5a,0xf5a,0xf5a,0xf5a, +0xf5a,0xf5a,0xf5a,0xf5a,0xf5a,0xf5a,0xf5a,0xf5a,0xf5a,0xf5a,0xf5a,0xf5a,0xf5a,0xf5a,0xf5a,0xf5a, +0xf5a,0xf5a,0xf54,0xf54,0xf54,0xf54,0x116a,0x116a,0x10e,0x10e,0x10e,0xf57,0x1512,0x1512,0x1512,0x1512, +0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512, +0x1512,0x1512,0x1512,0x1512,0x1512,0x16f2,0x111,0x111,0x111,0x111,0x111,0x111,0x111,0x111,0x111,0x111, +0x111,0x111,0x111,0x111,0x111,0x111,0x111,0x111,0x111,0x111,0x111,0x111,0x111,0x111,0x111,0x111, +0x111,0x111,0x111,0x111,0x111,0x111,0x111,0x111,0xf63,0xf63,0xf63,0x1518,0x1518,0x1518,0x1518,0x1518, +0x1518,0x1518,0x1518,0x1518,0x1518,0x1518,0x1518,0x114,0xf60,0xf60,0xf60,0xf60,0x1515,0x114,0x114,0x114, +0x114,0x114,0x114,0x114,0x114,0x114,0x114,0x114,0xf66,0xf66,0xf66,0xf66,0xf66,0xf66,0xf66,0xf66, +0xf66,0xf66,0xf66,0xf66,0xf66,0xf66,0xf66,0xf66,0xf66,0xf66,0x191d,0x191d,0x191d,0x191d,0x191d,0x191d, +0x191d,0x117,0x117,0x117,0x117,0x117,0x117,0x117,0x1062,0x1062,0x1062,0x1062,0x105f,0x105f,0x105f,0x105f, +0x105f,0x105f,0x105f,0x105f,0x1050,0x1050,0x1050,0x1050,0x1050,0x1050,0x1050,0x1050,0x105f,0x105f,0x1056,0x1053, +0x11a,0x11a,0x11a,0x1065,0x1065,0x1059,0x1059,0x1059,0x105c,0x105c,0x105c,0x105c,0x105c,0x105c,0x105c,0x105c, +0x105c,0x105c,0x11a,0x11a,0x11a,0x1062,0x1062,0x1062,0x1068,0x1068,0x1068,0x1068,0x1068,0x1068,0x1068,0x1068, +0x1068,0x1068,0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x107d,0x107d,0x107d,0x107d,0x107d,0x107d,0x107d,0x107d, +0x107d,0x107d,0x1080,0x1080,0x11d,0x11d,0x11d,0x11d,0x11d,0x11d,0x11d,0x11d,0x11d,0x11d,0x11d,0x11d, +0x11d,0x11d,0x11d,0x11d,0x11d,0x11d,0x11d,0x11d,0x10a7,0x10a7,0x10a7,0x10a7,0x10a1,0x17a6,0x120,0x120, +0x120,0x120,0x120,0x120,0x120,0x120,0x10ad,0x10ad,0x10a4,0x10a4,0x10a4,0x10a4,0x10a4,0x10a4,0x10a4,0x10a4, +0x10a4,0x10a4,0x120,0x120,0x120,0x120,0x120,0x120,0x10cb,0x10cb,0x10cb,0x10cb,0x10cb,0x10cb,0x10cb,0x10bf, +0x10bf,0x10bf,0x10bf,0x10bf,0x10bf,0x10bf,0x10bf,0x10bf,0x10bf,0x10bf,0x10c5,0x10c8,0x123,0x123,0x123,0x123, +0x123,0x123,0x123,0x123,0x123,0x123,0x123,0x10c2,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da, +0x10da,0x10ce,0x10ce,0x10ce,0x10ce,0x10ce,0x10ce,0x10d7,0x10d7,0x10ce,0x10ce,0x10d7,0x10d7,0x10ce,0x10ce,0x126, +0x126,0x126,0x126,0x126,0x126,0x126,0x126,0x126,0x10da,0x10da,0x10da,0x10ce,0x10da,0x10da,0x10da,0x10da, +0x10da,0x10da,0x10da,0x10da,0x10ce,0x10d7,0x126,0x126,0x10d4,0x10d4,0x10d4,0x10d4,0x10d4,0x10d4,0x10d4,0x10d4, +0x10d4,0x10d4,0x126,0x126,0x10d1,0x10dd,0x10dd,0x10dd,0x1524,0x129,0x129,0x129,0x129,0x129,0x129,0x129, +0x129,0x129,0x129,0x129,0x129,0x129,0x129,0x129,0x129,0x129,0x129,0x129,0x129,0x129,0x129,0x129, +0x129,0x129,0x129,0x129,0x129,0x129,0x129,0x129,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3, +0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e3, +0x10e3,0x10e3,0x10e3,0x10e3,0x10e3,0x10e6,0x12c,0x12c,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9, +0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x10e9, +0x10e9,0x10e9,0x10e9,0x10e9,0x10e9,0x12f,0x12f,0x12f,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec, +0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x132,0x132,0x132,0x132,0x132,0x132,0x132, +0x132,0x132,0x132,0x132,0x132,0x132,0x132,0x132,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2, +0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2,0x10f2, +0x10f2,0x10f2,0x135,0x135,0x135,0x135,0x135,0x10ef,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5, +0x10f5,0x10f5,0x10f5,0x10f5,0x138,0x138,0x138,0x138,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8, +0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x13b,0x13b,0x13b,0x13b, +0x13b,0x13b,0x13b,0x13b,0x13b,0x13b,0x13b,0x13b,0x1170,0x1170,0x1170,0x1170,0x1179,0x1170,0x1170,0x1170, +0x1179,0x1170,0x1170,0x1170,0x1170,0x116d,0x13e,0x13e,0x1176,0x1176,0x1176,0x1176,0x1176,0x1176,0x1176,0x117c, +0x1176,0x117c,0x1176,0x1176,0x1176,0x117c,0x117c,0x13e,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f, +0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x141,0x141, +0x141,0x141,0x141,0x141,0x141,0x141,0x141,0x141,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a, +0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x1197,0x1182,0x1197, +0x1182,0x1182,0x1182,0x1182,0x1182,0x1182,0x1182,0x144,0x118b,0x1194,0x1182,0x1194,0x1194,0x1182,0x1182,0x1182, +0x1182,0x1182,0x1182,0x1182,0x1182,0x1197,0x1197,0x1197,0x1197,0x1197,0x1197,0x1182,0x1182,0x1188,0x1188,0x1188, +0x1188,0x1188,0x1188,0x1188,0x1188,0x144,0x144,0x1185,0x1191,0x1191,0x1191,0x1191,0x1191,0x1191,0x1191,0x1191, +0x1191,0x1191,0x144,0x144,0x144,0x144,0x144,0x144,0x1191,0x1191,0x1191,0x1191,0x1191,0x1191,0x1191,0x1191, +0x1191,0x1191,0x144,0x144,0x144,0x144,0x144,0x144,0x118e,0x118e,0x118e,0x118e,0x118e,0x118e,0x118e,0x119d, +0x11a0,0x11a0,0x11a0,0x11a0,0x118e,0x118e,0x144,0x144,0x1563,0x1563,0x1563,0x1563,0x1563,0x1563,0x1563,0x1563, +0x1563,0x1563,0x1563,0x1563,0x1563,0x1563,0x1560,0x1a85,0x12e1,0x12ba,0x12d8,0x12d8,0x12d8,0x12d8,0x12d8,0x12d8, +0x12d8,0x12c0,0x12bd,0x12b4,0x12b4,0x12de,0x12b4,0x12b4,0x12b4,0x12b4,0x12c3,0x149d,0x14a3,0x14a0,0x14a0,0x18e4, +0x16b9,0x16b9,0x1a52,0x147,0x147,0x147,0x147,0x147,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5, +0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11ac,0x11ac,0x11af,0x11b8,0x11b2,0x11b2,0x11b2,0x11b8, +0x14a,0x14a,0x14a,0x14a,0x14a,0x14a,0x14a,0x14a,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5, +0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x12a5, +0x12a5,0x12a5,0x12a5,0x12a5,0x12a5,0x150,0x150,0x150,0x11d6,0x11ca,0x11ca,0x11ca,0x11ca,0x11ca,0x11ca,0x11cd, +0x11dc,0x11dc,0x11ca,0x11ca,0x11ca,0x11ca,0x153,0x12d5,0x11d0,0x11d0,0x11d0,0x11d0,0x11d0,0x11d0,0x11d0,0x11d0, +0x11d0,0x11d0,0x153,0x153,0x153,0x153,0x11ca,0x11ca,0x11fa,0x11ee,0x11fa,0x156,0x156,0x156,0x156,0x156, +0x156,0x156,0x156,0x156,0x156,0x156,0x156,0x156,0x156,0x156,0x156,0x156,0x156,0x156,0x156,0x156, +0x156,0x156,0x156,0x11f7,0x11f7,0x11fd,0x11f1,0x11f4,0x1212,0x1212,0x1212,0x120c,0x120c,0x1203,0x120c,0x120c, +0x1203,0x120c,0x120c,0x1215,0x120f,0x1206,0x159,0x159,0x1209,0x1209,0x1209,0x1209,0x1209,0x1209,0x1209,0x1209, +0x1209,0x1209,0x159,0x159,0x159,0x159,0x159,0x159,0x121b,0x121b,0x121b,0x121b,0x121b,0x121b,0x121b,0x15c, +0x15c,0x15c,0x15c,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218, +0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218,0x1218, +0x15c,0x15c,0x15c,0x15c,0x1224,0x1224,0x1224,0x1224,0x1224,0x1224,0x1224,0x1224,0x1224,0x1224,0x1224,0x1224, +0x1224,0x1224,0x1224,0x1224,0x1224,0x1224,0x1224,0x1224,0x1224,0x1224,0x15f,0x1221,0x121e,0x121e,0x121e,0x121e, +0x121e,0x121e,0x121e,0x121e,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233, +0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x162,0x162,0x162,0x122d,0x1230,0x1230, +0x1230,0x1230,0x1230,0x1230,0x1239,0x1239,0x1239,0x1239,0x1239,0x1239,0x1239,0x1239,0x1239,0x1239,0x1239,0x1239, +0x1239,0x1239,0x1239,0x1239,0x1239,0x1239,0x1239,0x1239,0x1239,0x1239,0x165,0x165,0x1236,0x1236,0x1236,0x1236, +0x1236,0x1236,0x1236,0x1236,0x123f,0x123f,0x123f,0x123f,0x123f,0x123f,0x123f,0x123f,0x123f,0x123f,0x123f,0x123f, +0x123f,0x123f,0x123f,0x123f,0x123f,0x123f,0x123f,0x168,0x168,0x168,0x168,0x168,0x123c,0x123c,0x123c,0x123c, +0x123c,0x123c,0x123c,0x123c,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245, +0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245,0x1245, +0x1245,0x1245,0x1245,0x16e,0x125d,0x125d,0x1b3c,0x171,0x171,0x171,0x171,0x171,0x171,0x171,0x171,0x171, +0x171,0x1926,0x171,0x171,0x147c,0x147c,0x147c,0x147c,0x147c,0x147c,0x147c,0x147c,0x147c,0x147c,0x147c,0x147c, +0x147c,0x147c,0x147c,0x147c,0x1827,0x1827,0x1827,0x1827,0x1827,0x1827,0x1827,0x1827,0x1827,0x1827,0x1827,0x1827, +0x1827,0x1a76,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174, +0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174,0x174, +0x174,0x174,0x174,0x174,0x174,0x174,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344, +0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344,0x1344, +0x12ae,0x13a7,0x13a7,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177, +0x12ab,0x12ab,0x12ab,0x12ab,0x12ab,0x12ab,0x12ab,0x12ab,0x12ab,0x12ab,0x12ab,0x12ab,0x12ab,0x12ab,0x12ab,0x12ab, +0x12ab,0x12ab,0x13a7,0x13a7,0x13a7,0x13a7,0x13a7,0x13a7,0x13a7,0x13a7,0x13a7,0x182a,0x177,0x177,0x177,0x177, +0x12a8,0x12a8,0x12a8,0x12a8,0x12a8,0x12a8,0x12a8,0x12a8,0x12a8,0x177,0x177,0x177,0x177,0x177,0x177,0x177, +0x13cb,0x13cb,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177, +0x18c9,0x18c9,0x18c9,0x18c9,0x18c9,0x18c9,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177, +0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177, +0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x177,0x134d,0x134d,0x134d,0x134d,0x134d,0x134d,0x134d,0x134d, +0x134d,0x134d,0x134d,0x134d,0x134d,0x134d,0x134d,0x134d,0x134d,0x134d,0x134d,0x134d,0x134d,0x134d,0x134d,0x134d, +0x134d,0x1347,0x1347,0x1347,0x17a,0x17a,0x134a,0x17a,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x1350,0x1359, +0x1353,0x1353,0x1359,0x1359,0x1359,0x1353,0x1359,0x1353,0x1353,0x1353,0x135c,0x135c,0x17d,0x17d,0x17d,0x17d, +0x17d,0x17d,0x17d,0x17d,0x1356,0x1356,0x1356,0x1356,0x180,0x1362,0x1362,0x1362,0x1362,0x1362,0x1362,0x180, +0x180,0x1362,0x1362,0x1362,0x1362,0x1362,0x1362,0x180,0x180,0x1362,0x1362,0x1362,0x1362,0x1362,0x1362,0x180, +0x180,0x180,0x180,0x180,0x180,0x180,0x180,0x180,0x1362,0x1362,0x1362,0x1362,0x1362,0x1362,0x1362,0x180, +0x1362,0x1362,0x1362,0x1362,0x1362,0x1362,0x1362,0x180,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0, +0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x1365,0x1365,0x1365,0x1365,0x1365,0x1365,0x1368,0x137a, +0x137a,0x136e,0x136e,0x136e,0x136e,0x136e,0x183,0x183,0x183,0x183,0x136b,0x136b,0x136b,0x136b,0x136b,0x136b, +0x136b,0x136b,0x136b,0x136b,0x136b,0x136b,0x136b,0x136b,0x136b,0x136b,0x1371,0x1371,0x1371,0x1371,0x1371,0x1371, +0x1371,0x1371,0x1371,0x1371,0x1b42,0x1b45,0x1b45,0x1b3f,0x1b3f,0x1b45,0x183,0x183,0x183,0x183,0x183,0x183, +0x183,0x183,0x183,0x1533,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d, +0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x186,0x186,0x186, +0x186,0x186,0x186,0x186,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380, +0x1380,0x1380,0x1380,0x189,0x189,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380, +0x1380,0x1380,0x1380,0x1536,0x189,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380, +0x1380,0x1380,0x1380,0x13b0,0x189,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380,0x1380, +0x1380,0x1380,0x1380,0x1380,0x1536,0x1536,0x1536,0x1536,0x1536,0x1536,0x1536,0x1536,0x1536,0x1536,0x1536,0x1536, +0x1536,0x1536,0x1536,0x1536,0x1536,0x1536,0x1536,0x1536,0x1536,0x1536,0x189,0x189,0x189,0x189,0x189,0x189, +0x189,0x189,0x189,0x189,0x13c5,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x1542,0x1542,0x1542,0x1542,0x1542,0x1542, +0x16b0,0x1542,0x1542,0x1542,0x1782,0x1833,0x1833,0x186c,0x186c,0x1a34,0x1adf,0x1adf,0x18c,0x18c,0x18c,0x18c, +0x1c2c,0x1bae,0x1bae,0x1bae,0x1542,0x1542,0x1542,0x1542,0x1542,0x1542,0x1542,0x1542,0x1542,0x1542,0x1542,0x16ad, +0x16ad,0x18c,0x18c,0x18c,0x1542,0x1542,0x1542,0x1542,0x1833,0x1833,0x1833,0x18cf,0x18cf,0x19b0,0x1a34,0x1adf, +0x1adf,0x18c,0x18c,0x18c,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383, +0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1bd2,0x1bd2,0x1bd2,0x18f,0x18f,0x18f,0x18f,0x1bd2, +0x1bd2,0x1bd2,0x1bd2,0x1bd2,0x141f,0x141f,0x141f,0x141f,0x192,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f, +0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f, +0x141f,0x141f,0x141f,0x141f,0x192,0x141f,0x141f,0x192,0x141f,0x192,0x192,0x141f,0x192,0x141f,0x141f,0x141f, +0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x192,0x141f,0x141f,0x141f,0x141f,0x192,0x141f,0x192,0x141f, +0x192,0x192,0x192,0x192,0x192,0x192,0x141f,0x192,0x192,0x192,0x192,0x141f,0x192,0x141f,0x192,0x141f, +0x192,0x141f,0x141f,0x141f,0x192,0x141f,0x141f,0x192,0x141f,0x192,0x192,0x141f,0x192,0x141f,0x192,0x141f, +0x192,0x141f,0x192,0x141f,0x192,0x141f,0x141f,0x192,0x141f,0x192,0x192,0x141f,0x141f,0x141f,0x141f,0x192, +0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x192,0x141f,0x141f,0x141f,0x141f,0x192,0x141f,0x141f,0x141f, +0x141f,0x192,0x141f,0x192,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x192,0x141f, +0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f, +0x192,0x192,0x192,0x192,0x192,0x141f,0x141f,0x141f,0x192,0x141f,0x141f,0x141f,0x141f,0x141f,0x192,0x141f, +0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f,0x141f, +0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192, +0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192, +0x141c,0x141c,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192,0x192, +0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1422,0x1422,0x1422,0x1422,0x1422,0x1431,0x1422,0x1425,0x1425, +0x1422,0x1422,0x1422,0x1428,0x1428,0x195,0x142e,0x142e,0x142e,0x142e,0x142e,0x142e,0x142e,0x142e,0x142e,0x142e, +0x142b,0x1437,0x1437,0x1437,0x1932,0x192f,0x192f,0x1a7c,0x195,0x195,0x195,0x195,0x195,0x195,0x195,0x195, +0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2, +0x1443,0x1443,0x1443,0x1443,0x1443,0x1443,0x1443,0x1443,0x1443,0x1443,0x1443,0x1440,0x143a,0x143a,0x1440,0x1440, +0x1449,0x1449,0x1443,0x1446,0x1446,0x1440,0x143d,0x198,0x198,0x198,0x198,0x198,0x198,0x198,0x198,0x198, +0x144c,0x144c,0x144c,0x144c,0x144c,0x144c,0x144c,0x144c,0x144c,0x144c,0x144c,0x144c,0x144c,0x144c,0x144c,0x144c, +0x144c,0x144c,0x144c,0x144c,0x144c,0x144c,0x144c,0x144c,0x19b,0x19b,0x19b,0x19b,0x1707,0x1707,0x144c,0x144c, +0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707, +0x19b,0x19b,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707, +0x1458,0x1458,0x1458,0x1458,0x1458,0x19dd,0x19dd,0x19dd,0x19dd,0x19dd,0x19dd,0x19e,0x19e,0x19e,0x19e,0x19d7, +0x1458,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455, +0x19da,0x19da,0x19da,0x19da,0x19da,0x19da,0x19da,0x19da,0x19e,0x19e,0x19e,0x19e,0x19e,0x19e,0x19e,0x1452, +0x1452,0x1452,0x1452,0x145b,0x145b,0x145b,0x145b,0x145b,0x145b,0x145b,0x145b,0x145b,0x145b,0x145b,0x145b,0x145b, +0x147c,0x147c,0x147c,0x147c,0x147c,0x147c,0x147c,0x147c,0x147c,0x1a1,0x1a1,0x1a1,0x1a1,0x1a1,0x1a1,0x1a1, +0x1479,0x1479,0x1479,0x1479,0x1479,0x1479,0x1479,0x1479,0x1479,0x1479,0x1a1,0x1a1,0x1a1,0x1a1,0x1a1,0x1a1, +0x147f,0x147f,0x147f,0x147f,0x147f,0x147f,0x147f,0x147f,0x1a4,0x1a4,0x1a4,0x1a4,0x1a4,0x1a4,0x1a4,0x1a4, +0x12db,0x12d8,0x12db,0x12b7,0x12d8,0x12de,0x12de,0x12e1,0x12de,0x12e1,0x12e4,0x12d8,0x12e1,0x12e1,0x12d8,0x12d8, +0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1482,0x148b,0x1482,0x148b,0x148b, +0x1482,0x1482,0x1482,0x1482,0x1482,0x1482,0x148e,0x1485,0x19e0,0x1b51,0x1a7,0x1a7,0x1a7,0x1a7,0x1a7,0x1a7, +0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1aa,0x1aa, +0x1551,0x1551,0x1551,0x1551,0x1551,0x1557,0x1aa,0x1aa,0x1aa,0x1aa,0x1aa,0x1aa,0x1aa,0x1aa,0x1aa,0x1aa, +0x155d,0x155d,0x155d,0x155d,0x1ad,0x1ad,0x1ad,0x1ad,0x1ad,0x1ad,0x1ad,0x1ad,0x1ad,0x1ad,0x1ad,0x155a, +0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x2b2,0x1b9c,0x1b9c,0x1b9c,0x1b9c, +0x16bc,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3, +0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x16b3,0x1b0,0x1b0,0x1b0,0x1b0, +0x1a85,0x1b57,0x1b57,0x1b57,0x1b57,0x1b57,0x1b57,0x1b57,0x1b57,0x1b57,0x1b57,0x1b57,0x1b54,0x1b54,0x1b54,0x1b3, +0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3, +0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3,0x1b3, +0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x1b6,0x1b6,0x1b6,0x1b6,0x1b6, +0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x1b6,0x1b6,0x1b6, +0x1b6,0x1b6,0x1b6,0x1b6,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x1b6,0x1b6, +0x156c,0x1566,0x1569,0x1572,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1b9,0x1b9,0x1b9,0x1b9, +0x1b9,0x1b9,0x1b9,0x1b9,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d, +0x155d,0x155d,0x155d,0x155d,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578, +0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1935,0x1935,0x1935,0x1935,0x1bd5,0x1bc,0x1bc, +0x1bc,0x1bc,0x1bc,0x1bc,0x1a37,0x1a37,0x1a37,0x1a37,0x1a37,0x1a37,0x1a37,0x1a37,0x1a37,0x1a37,0x1a37,0x1a37, +0x1bc,0x1bc,0x1bc,0x1bc,0x1bb1,0x1bc,0x1bc,0x1bc,0x1bc,0x1bc,0x1bc,0x1bc,0x1bc,0x1bc,0x1bc,0x1bc, +0x1bc,0x1bc,0x1bc,0x1bc,0x171c,0x16bf,0x1581,0x16c5,0x1bf,0x158a,0x158a,0x158a,0x158a,0x158a,0x158a,0x158a, +0x158a,0x1bf,0x1bf,0x158a,0x158a,0x1bf,0x1bf,0x158a,0x158a,0x158a,0x158a,0x158a,0x158a,0x158a,0x158a,0x158a, +0x158a,0x158a,0x158a,0x158a,0x158a,0x1bf,0x158a,0x158a,0x158a,0x158a,0x158a,0x158a,0x158a,0x1bf,0x158a,0x158a, +0x1bf,0x158a,0x158a,0x158a,0x158a,0x158a,0x1bf,0x19bc,0x16c2,0x158a,0x157b,0x1581,0x157b,0x1581,0x1581,0x1581, +0x1581,0x1bf,0x1bf,0x1581,0x1581,0x1bf,0x1bf,0x1584,0x1584,0x1587,0x1bf,0x1bf,0x171f,0x1bf,0x1bf,0x1bf, +0x1bf,0x1bf,0x1bf,0x157b,0x1bf,0x1bf,0x1bf,0x1bf,0x1bf,0x158d,0x158a,0x158a,0x158a,0x158a,0x1581,0x1581, +0x1bf,0x1bf,0x157e,0x157e,0x157e,0x157e,0x157e,0x157e,0x157e,0x1bf,0x1bf,0x1bf,0x157e,0x157e,0x157e,0x157e, +0x157e,0x1bf,0x1bf,0x1bf,0x1bf,0x1bf,0x1bf,0x1bf,0x1bf,0x1bf,0x1bf,0x1bf,0x15a2,0x15a2,0x15a2,0x15a2, +0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x1c2,0x15a2, +0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15b4,0x15b4,0x15b4,0x15a8, +0x15a8,0x15a8,0x15a8,0x15a8,0x15a8,0x15ab,0x15ae,0x1c5,0x1c5,0x1c5,0x1c5,0x1c5,0x15b1,0x15b1,0x15b1,0x15b1, +0x15b1,0x15b1,0x15b1,0x15b1,0x15b1,0x15b1,0x1c5,0x1c5,0x1c5,0x1c5,0x1c5,0x1c5,0x1722,0x1722,0x1722,0x1722, +0x15c0,0x15bd,0x19e3,0x19e3,0x1a8b,0x1a8e,0x1a88,0x1a88,0x1c8,0x1c8,0x1c8,0x1c8,0x174f,0x174f,0x174f,0x174f, +0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x15c6,0x15c6,0x15c6,0x15c6, +0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6, +0x15c6,0x15c6,0x15c6,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x15c6,0x15c6,0x15c6,0x15c6, +0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6, +0x15c6,0x15c6,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x15c6,0x15c6,0x15c6,0x15c6, +0x15c6,0x15c6,0x15c6,0x15c6,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb, +0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x1cb,0x15d2,0x15d2,0x15d2,0x15d2, +0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15d2,0x15c9, +0x15cc,0x15cf,0x15d2,0x1ce,0x1ce,0x1ce,0x1ce,0x1ce,0x1ce,0x1ce,0x1ce,0x1ce,0x15e1,0x15e1,0x15e1,0x15e1, +0x15e1,0x15d5,0x15d5,0x1d1,0x1d1,0x1d1,0x1d1,0x15d8,0x15d8,0x15d8,0x15d8,0x15d8,0x15de,0x15de,0x16c8,0x15de, +0x15de,0x15de,0x15db,0x1d1,0x1d1,0x1d1,0x1d1,0x1d1,0x1d1,0x1d1,0x1d1,0x1d1,0x15ea,0x15ea,0x15ea,0x15ea, +0x15ea,0x1d4,0x1d4,0x15e7,0x15e7,0x15e7,0x15e7,0x15e7,0x15e7,0x15e7,0x15e7,0x15e7,0x15e4,0x15e4,0x15e4,0x15e4, +0x15e4,0x15e4,0x15e4,0x1d4,0x1d4,0x1d4,0x1d4,0x1d4,0x1d4,0x1d4,0x1d4,0x1d4,0x15ed,0x15ff,0x15ff,0x15f3, +0x15fc,0x1d7,0x1d7,0x1d7,0x1d7,0x1d7,0x1d7,0x1d7,0x1d7,0x1d7,0x1d7,0x1d7,0x15f6,0x15f6,0x15f6,0x15f6, +0x15f6,0x15f6,0x15f6,0x15f6,0x15f6,0x15f6,0x1d7,0x1d7,0x1d7,0x1d7,0x1d7,0x1d7,0x1605,0x1605,0x1605,0x1605, +0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605, +0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1605,0x1da,0x1602,0x1602,0x1602,0x1602, +0x1602,0x1602,0x1602,0x1602,0x1602,0x1602,0x1da,0x1da,0x1da,0x1da,0x1608,0x1608,0x1b8d,0x1b8d,0x1b8d,0x1b8d, +0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1611,0x1611,0x1611,0x1611, +0x1611,0x160b,0x1614,0x1611,0x1611,0x1611,0x1611,0x1611,0x1611,0x1611,0x1611,0x1611,0x160e,0x160e,0x160e,0x160e, +0x160e,0x160e,0x160e,0x160e,0x160e,0x160e,0x1611,0x1611,0x1611,0x1611,0x1611,0x1dd,0x161a,0x161a,0x161a,0x161a, +0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a, +0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x161a,0x1e0,0x1626,0x1626,0x1626,0x1626, +0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626,0x1626, +0x1626,0x1626,0x1623,0x1623,0x1623,0x1623,0x1623,0x1e3,0x1e3,0x1e3,0x1e3,0x1e3,0x163e,0x163e,0x1641,0x1641, +0x1644,0x1635,0x1e6,0x1e6,0x1e6,0x1e6,0x1e6,0x1e6,0x1e6,0x1e6,0x1e6,0x1e6,0x163b,0x163b,0x163b,0x163b, +0x163b,0x163b,0x163b,0x163b,0x163b,0x163b,0x1e6,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x1e6,0x163e, +0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e, +0x163e,0x163e,0x163e,0x163e,0x1e6,0x1e6,0x1e6,0x1e6,0x1e6,0x163e,0x163e,0x163e,0x164d,0x164d,0x164d,0x164d, +0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d, +0x164d,0x164d,0x164d,0x164d,0x164d,0x1e9,0x1e9,0x1e9,0x1e9,0x1e9,0x1e9,0x1e9,0x1656,0x1656,0x1656,0x1656, +0x1656,0x1656,0x1656,0x1656,0x1656,0x1656,0x1656,0x1656,0x1656,0x1656,0x1656,0x1656,0x1656,0x1656,0x1ec,0x1ec, +0x1ec,0x1ec,0x1ec,0x1ec,0x1ec,0x1653,0x1653,0x1653,0x1653,0x1ec,0x1ec,0x1ec,0x1671,0x1671,0x1671,0x1671, +0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1659,0x166b,0x166b,0x1659,0x1659, +0x1659,0x1659,0x1f2,0x1f2,0x166b,0x166b,0x166e,0x166e,0x1659,0x1659,0x166b,0x165f,0x165c,0x1662,0x1674,0x1674, +0x1665,0x1665,0x1668,0x1668,0x1668,0x1674,0x172b,0x172b,0x172b,0x172b,0x172b,0x172b,0x172b,0x172b,0x172b,0x172b, +0x172b,0x172b,0x172b,0x172b,0x1728,0x1728,0x1728,0x1728,0x1725,0x1725,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2, +0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2, +0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f2,0x1f5,0x1677,0x1677,0x1677, +0x1677,0x1677,0x1677,0x1677,0x1677,0x1677,0x1677,0x1677,0x1677,0x1677,0x1677,0x1677,0x1677,0x1677,0x1677,0x1677, +0x1677,0x1f5,0x1f5,0x1f5,0x1f5,0x1f5,0x1f5,0x1f5,0x1f5,0x1f5,0x1f5,0x1f5,0x167a,0x167a,0x167a,0x167a, +0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x1f8,0x1f8,0x1f8,0x1f8,0x167a,0x167a,0x167a,0x167a, +0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x1f8,0x1f8,0x1f8,0x1f8, +0x1f8,0x1f8,0x1f8,0x1f8,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x1f8,0x1f8, +0x1f8,0x1f8,0x1f8,0x1f8,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x1f8,0x1f8,0x1f8,0x1f8, +0x1f8,0x1f8,0x1f8,0x1f8,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a, +0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x1f8,0x1f8,0x1a91,0x1a91,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8, +0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8, +0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x1f8,0x167d,0x168c,0x1683,0x1680, +0x1692,0x1692,0x1686,0x1692,0x1fb,0x1fb,0x1fb,0x1fb,0x1fb,0x1fb,0x1fb,0x1fb,0x1689,0x1689,0x1689,0x1689, +0x1689,0x1689,0x1689,0x1689,0x1689,0x1689,0x1fb,0x1fb,0x1fb,0x1fb,0x1fb,0x1fb,0x1698,0x1698,0x1698,0x1698, +0x1698,0x1698,0x1698,0x1698,0x1698,0x1698,0x1695,0x1695,0x1695,0x1695,0x1695,0x1695,0x1695,0x1695,0x1695,0x1fe, +0x1fe,0x1fe,0x1fe,0x1fe,0x1fe,0x1fe,0x1fe,0x1fe,0x1fe,0x1fe,0x1fe,0x169e,0x1740,0x1740,0x1740,0x1740, +0x1740,0x1740,0x1740,0x1740,0x1740,0x1740,0x1740,0x1740,0x1740,0x1740,0x1740,0x1740,0x1740,0x1740,0x1740,0x1740, +0x1740,0x1740,0x1740,0x1740,0x1740,0x1740,0x1938,0x201,0x201,0x172e,0x172e,0x172e,0x173a,0x173a,0x172e,0x172e, +0x172e,0x172e,0x173d,0x172e,0x172e,0x172e,0x172e,0x1731,0x201,0x201,0x201,0x201,0x1737,0x1737,0x1737,0x1737, +0x1737,0x1737,0x1737,0x1737,0x1737,0x1737,0x1734,0x1734,0x1743,0x1743,0x1743,0x1734,0x1746,0x1746,0x1746,0x1746, +0x1746,0x1746,0x1746,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204, +0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204,0x204, +0x204,0x204,0x204,0x204,0x1758,0x1758,0x1758,0x1758,0x1758,0x1758,0x1758,0x1758,0x1758,0x1758,0x1758,0x1758, +0x1758,0x1758,0x1758,0x1758,0x1758,0x1758,0x1758,0x20a,0x1758,0x1758,0x20a,0x20a,0x20a,0x20a,0x20a,0x1755, +0x1755,0x1755,0x1755,0x1755,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x20d,0x175b,0x20d,0x175b,0x175b, +0x175b,0x175b,0x20d,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b, +0x175b,0x175b,0x20d,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x175b,0x175e,0x20d,0x20d, +0x20d,0x20d,0x20d,0x20d,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7, +0x15b7,0x15b7,0x15b7,0x15b7,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767, +0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x210,0x210,0x210,0x210,0x210,0x210,0x210,0x210,0x210, +0x210,0x210,0x210,0x210,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764, +0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x210,0x210,0x210,0x210,0x210,0x210,0x210,0x1761,0x1761, +0x1761,0x1761,0x1761,0x1761,0x176d,0x176d,0x176d,0x176d,0x176a,0x176d,0x176d,0x1770,0x1773,0x1770,0x1770,0x176d, +0x213,0x213,0x213,0x213,0x213,0x213,0x213,0x213,0x213,0x213,0x213,0x213,0x213,0x213,0x213,0x176a, +0x176a,0x176a,0x176a,0x176a,0x17ca,0x17ca,0x17ca,0x17ca,0x17c1,0x17c1,0x17c1,0x17bb,0x17be,0x17be,0x17be,0x19e6, +0x216,0x216,0x216,0x216,0x17c7,0x17c7,0x17c7,0x17c7,0x17c7,0x17c7,0x17c7,0x17c7,0x17c7,0x17c7,0x216,0x216, +0x216,0x216,0x17c4,0x17c4,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x219,0x17e5,0x17e5, +0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5, +0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e5,0x17e2,0x17d0,0x17d0,0x17d0,0x17d0,0x17d0,0x17d0,0x17d0,0x219, +0x17d0,0x17d0,0x17d0,0x17d0,0x17d0,0x17d0,0x17e2,0x17d3,0x17e5,0x17e8,0x17e8,0x17dc,0x17d9,0x17d9,0x219,0x219, +0x219,0x219,0x219,0x219,0x219,0x219,0x219,0x219,0x17df,0x17df,0x17df,0x17df,0x17df,0x17df,0x17df,0x17df, +0x17df,0x17df,0x17d6,0x17d6,0x17d6,0x17d6,0x17d6,0x17d6,0x17d6,0x17d6,0x17d6,0x17d6,0x17d6,0x17d6,0x17d6,0x17d6, +0x17d6,0x219,0x219,0x219,0x17f4,0x17f7,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd, +0x17fd,0x17fd,0x17fd,0x17fd,0x17eb,0x17eb,0x17eb,0x17eb,0x17eb,0x17eb,0x17eb,0x17eb,0x17eb,0x21c,0x21c,0x21c, +0x21c,0x21c,0x21c,0x21c,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956, +0x1956,0x1956,0x1956,0x1956,0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x21f,0x17ee,0x17ee,0x17ee,0x17ee, +0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x21f,0x21f,0x17ee, +0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x21f,0x17ee,0x17ee,0x21f,0x17ee,0x17ee,0x17ee,0x17ee,0x17ee,0x21f, +0x21f,0x21f,0x21f,0x21f,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc, +0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x183c,0x18d8,0x1a40,0x1a43,0x1aeb,0x222,0x222,0x222,0x222,0x222,0x222,0x222, +0x222,0x222,0x222,0x222,0x1ae8,0x1ae8,0x222,0x222,0x222,0x222,0x222,0x222,0x222,0x222,0x222,0x222, +0x222,0x222,0x222,0x222,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd,0x17fd, +0x17fd,0x17fd,0x17fd,0x17fd,0x225,0x225,0x17f1,0x17f1,0x17f1,0x17f1,0x17f1,0x17f1,0x17f1,0x17f1,0x17f1,0x17f1, +0x17f1,0x17f1,0x17f1,0x17f1,0x225,0x17fa,0x17f1,0x17f1,0x17f1,0x17f1,0x17f1,0x17f1,0x17f1,0x17fa,0x17f1,0x17f1, +0x17fa,0x17f1,0x17f1,0x225,0x225,0x225,0x225,0x225,0x225,0x225,0x225,0x225,0x1800,0x1800,0x1800,0x1800, +0x1800,0x1800,0x1800,0x1800,0x1800,0x1800,0x1800,0x1800,0x1800,0x228,0x228,0x228,0x228,0x228,0x228,0x228, +0x228,0x228,0x228,0x228,0x228,0x228,0x228,0x228,0x228,0x228,0x228,0x228,0x1818,0x1818,0x1809,0x1803, +0x1803,0x1818,0x1806,0x181b,0x181b,0x181b,0x181b,0x181e,0x181e,0x1812,0x180f,0x180c,0x1815,0x1815,0x1815,0x1815, +0x1815,0x1815,0x1815,0x1815,0x1815,0x1815,0x1a94,0x1812,0x22b,0x180c,0x193b,0x19e9,0x1a97,0x1a97,0x22b,0x22b, +0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b, +0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x22b,0x1824,0x1824,0x1824,0x1824, +0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824,0x1824, +0x22e,0x22e,0x22e,0x22e,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821, +0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821,0x1821, +0x22e,0x22e,0x22e,0x22e,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f, +0x183f,0x19b9,0x19b9,0x19b9,0x19b9,0x19b9,0x1a46,0x1a46,0x1a46,0x1a46,0x1a46,0x1a46,0x231,0x231,0x231,0x231, +0x231,0x231,0x231,0x231,0x1bba,0x1bba,0x1bba,0x234,0x234,0x234,0x234,0x234,0x234,0x234,0x234,0x234, +0x234,0x234,0x234,0x234,0x276,0x276,0x1c2f,0x276,0x276,0x276,0x276,0x276,0x276,0x276,0x276,0x276, +0x276,0x276,0x276,0x276,0x187e,0x187e,0x187e,0x187e,0x187e,0x187e,0x187e,0x237,0x187e,0x187e,0x237,0x187e, +0x187e,0x187e,0x187e,0x187e,0x187e,0x187e,0x187e,0x187e,0x187e,0x187e,0x187e,0x187e,0x187e,0x187e,0x187e,0x187e, +0x187e,0x187e,0x187e,0x187e,0x187e,0x1872,0x1872,0x1872,0x1872,0x1872,0x1872,0x237,0x237,0x237,0x1872,0x237, +0x1872,0x1872,0x237,0x1872,0x1872,0x1872,0x1875,0x1872,0x1878,0x1878,0x1881,0x1872,0x237,0x237,0x237,0x237, +0x237,0x237,0x237,0x237,0x187b,0x187b,0x187b,0x187b,0x187b,0x187b,0x187b,0x187b,0x187b,0x187b,0x237,0x237, +0x237,0x237,0x237,0x237,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1, +0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1, +0x23a,0x23a,0x23a,0x23a,0x1890,0x1893,0x1893,0x23d,0x23d,0x23d,0x23d,0x23d,0x23d,0x23d,0x23d,0x23d, +0x23d,0x23d,0x23d,0x23d,0x1b96,0x1b96,0x1b96,0x1b96,0x1b96,0x1b96,0x1b96,0x1b96,0x1b96,0x1b96,0x1b96,0x1b96, +0x1b96,0x1b96,0x1b96,0x1b96,0x18a2,0x18a2,0x18a2,0x18a2,0x18a2,0x18a2,0x18a2,0x18a2,0x18a2,0x18a2,0x18a2,0x240, +0x240,0x240,0x240,0x240,0x1b63,0x1b63,0x1b63,0x1b63,0x1b63,0x1b63,0x1b63,0x1b63,0x1b63,0x1b63,0x1b63,0x1b63, +0x1b63,0x1b63,0x1b63,0x1b63,0x18ae,0x18b1,0x18c0,0x18c0,0x18b1,0x18b4,0x18ae,0x18ab,0x243,0x243,0x243,0x243, +0x243,0x243,0x243,0x243,0x1899,0x1884,0x1884,0x1884,0x1884,0x1884,0x1884,0x1896,0x1896,0x1884,0x1884,0x1884, +0x1899,0x1899,0x1899,0x1899,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef, +0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x246,0x246,0x246,0x246,0x246,0x246,0x246,0x246, +0x246,0x246,0x246,0x246,0x1941,0x1941,0x1941,0x1941,0x1941,0x1941,0x1941,0x1941,0x1941,0x1941,0x1941,0x1941, +0x1941,0x1941,0x246,0x246,0x1a4f,0x1a4f,0x1a4f,0x1a4f,0x1af1,0x1c35,0x1c35,0x1c35,0x1a4f,0x1a4f,0x1a4f,0x1bbd, +0x1bbd,0x279,0x279,0x279,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953, +0x1950,0x1950,0x1950,0x1944,0x1944,0x1944,0x1944,0x1944,0x1944,0x1944,0x1944,0x1944,0x1950,0x194a,0x1947,0x194d, +0x249,0x249,0x249,0x249,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956, +0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x1956,0x24c, +0x24c,0x1956,0x1956,0x1956,0x1965,0x1965,0x1965,0x1965,0x1965,0x1965,0x24f,0x1965,0x1965,0x24f,0x1965,0x1965, +0x1965,0x1965,0x1965,0x1965,0x1965,0x1965,0x1965,0x1965,0x1965,0x1965,0x1965,0x1965,0x1965,0x1965,0x1965,0x1965, +0x1965,0x1965,0x1965,0x1965,0x1965,0x1965,0x1962,0x1962,0x1962,0x1962,0x1962,0x24f,0x1959,0x1959,0x24f,0x1962, +0x1962,0x1959,0x1962,0x195c,0x1965,0x24f,0x24f,0x24f,0x24f,0x24f,0x24f,0x24f,0x196e,0x196e,0x1971,0x1971, +0x1968,0x1968,0x1968,0x1968,0x252,0x252,0x252,0x252,0x252,0x252,0x252,0x252,0x196b,0x196b,0x196b,0x196b, +0x196b,0x196b,0x196b,0x196b,0x196b,0x196b,0x252,0x252,0x252,0x252,0x252,0x252,0x1974,0x1974,0x1974,0x1974, +0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1977,0x1974,0x1974,0x1974,0x1977,0x1974,0x1974,0x1974, +0x1974,0x255,0x255,0x255,0x255,0x255,0x255,0x255,0x255,0x255,0x255,0x255,0x1980,0x1980,0x1980,0x1980, +0x1980,0x1980,0x1980,0x1980,0x1980,0x1980,0x1980,0x1980,0x1980,0x1980,0x1980,0x1980,0x1980,0x1980,0x1980,0x197a, +0x197a,0x197d,0x197d,0x1983,0x1983,0x258,0x258,0x258,0x258,0x258,0x258,0x258,0x1986,0x1986,0x1986,0x1986, +0x1986,0x1986,0x1986,0x1986,0x1986,0x1986,0x1986,0x1986,0x1986,0x1986,0x1986,0x1986,0x1986,0x1986,0x1986,0x1986, +0x25b,0x25b,0x25b,0x25b,0x25b,0x25b,0x25b,0x25b,0x25b,0x25b,0x25b,0x25b,0x1989,0x1989,0x1989,0x1989, +0x1989,0x1989,0x1989,0x1989,0x1989,0x1989,0x1989,0x1989,0x1989,0x1989,0x1989,0x1989,0x1989,0x1989,0x1989,0x1989, +0x1989,0x1989,0x1989,0x198c,0x1995,0x1989,0x1989,0x25e,0x25e,0x25e,0x25e,0x25e,0x1998,0x1998,0x1998,0x1998, +0x1998,0x1998,0x1998,0x199b,0x261,0x261,0x261,0x261,0x261,0x261,0x261,0x261,0x19a4,0x19a4,0x19a4,0x19a4, +0x19a4,0x19a4,0x19a4,0x19a4,0x19a4,0x19a4,0x19a4,0x19a4,0x19a4,0x19a4,0x19a4,0x19a4,0x19a4,0x19a4,0x199e,0x199e, +0x199e,0x199e,0x199e,0x199e,0x199e,0x199e,0x199e,0x199e,0x199e,0x19a1,0x19a1,0x19a1,0x19a1,0x19a7,0x19a7,0x19a7, +0x19a7,0x19a7,0x264,0x264,0x264,0x264,0x264,0x264,0x264,0x264,0x264,0x264,0x264,0x264,0x264,0x264, +0x264,0x264,0x264,0x264,0x1b84,0x1b84,0x1b84,0x1b84,0x1b84,0x1b84,0x1b84,0x1b84,0x1b84,0x1b84,0x1b84,0x1b84, +0x1b84,0x1b84,0x1b84,0x1b84,0x1be7,0x1bed,0x1bed,0x1bed,0x1bed,0x1bed,0x1bed,0x1bea,0x1bea,0x1bea,0x1bea,0x1bea, +0x1bea,0x1bea,0x1bea,0x1bea,0x1bea,0x1bea,0x1bea,0x1bea,0x1bea,0x1bea,0x267,0x267,0x267,0x267,0x267,0x267, +0x267,0x267,0x267,0x267,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb, +0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x19fb,0x26a,0x26a,0x26a,0x26a,0x26a, +0x26a,0x26a,0x26a,0x26a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x26d,0x26d,0x1a0a,0x1a0a, +0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a, +0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a0a,0x1a07,0x1a07,0x1a07,0x19fe,0x19fe,0x19fe,0x19fe,0x26d,0x26d,0x19fe,0x19fe, +0x1a07,0x1a07,0x1a07,0x1a07,0x1a01,0x1a0a,0x1a04,0x1a0a,0x1a07,0x26d,0x26d,0x26d,0x26d,0x26d,0x26d,0x26d, +0x26d,0x26d,0x26d,0x26d,0x26d,0x26d,0x26d,0x26d,0x26d,0x26d,0x26d,0x26d,0x26d,0x26d,0x26d,0x26d, +0x26d,0x26d,0x26d,0x26d,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16, +0x1a16,0x270,0x270,0x270,0x1a0d,0x1a0d,0x1a0d,0x1a0d,0x1a0d,0x1a0d,0x1a0d,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16, +0x1a19,0x1a19,0x270,0x270,0x273,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c, +0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c, +0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x1a1c,0x273,0x273,0x276,0x276,0x276,0x276,0x276,0x276,0x276,0x276, +0x276,0x276,0x276,0x276,0x276,0x276,0x276,0x276,0x1a49,0x1a49,0x1a49,0x276,0x276,0x1c32,0x276,0x276, +0x276,0x276,0x276,0x276,0x276,0x276,0x276,0x276,0x1a4c,0x1a4c,0x1a4c,0x1a4c,0x276,0x276,0x276,0x276, +0x276,0x276,0x276,0x276,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1, +0x18e1,0x18e1,0x18e1,0x18e1,0x1a4f,0x1a4f,0x1a4f,0x1af1,0x1af1,0x1af1,0x1af1,0x1c35,0x1c35,0x279,0x279,0x279, +0x279,0x279,0x279,0x279,0x1a4f,0x1a4f,0x1a4f,0x1a4f,0x1a4f,0x1a4f,0x1af1,0x1af1,0x1af1,0x1af1,0x1af1,0x1af1, +0x1af1,0x1af1,0x1af1,0x1af1,0x1af1,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1c35,0x1c35,0x1c35,0x1af1,0x1af1,0x1af1,0x1af1, +0x1af1,0x1af1,0x1af1,0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1c35,0x1c35,0x1c35,0x279,0x1c35,0x1af1,0x1af1,0x1af1,0x1bc0, +0x1bc0,0x1bc0,0x279,0x279,0x279,0x279,0x279,0x279,0x279,0x279,0x1c35,0x1c35,0x1af1,0x1af1,0x1af1,0x1af1, +0x1af1,0x1af1,0x1af1,0x1bbd,0x1bbd,0x1bbd,0x1c35,0x1c35,0x279,0x279,0x279,0x279,0x1bbd,0x1bbd,0x1bbd,0x1bbd, +0x1bbd,0x1bbd,0x1bbd,0x1bbd,0x1c35,0x279,0x279,0x279,0x279,0x279,0x279,0x279,0x1bc0,0x1bc0,0x1bc0,0x1bc0, +0x1bc0,0x1bc0,0x1bc0,0x1c38,0x1c38,0x279,0x279,0x279,0x279,0x279,0x279,0x279,0x1a25,0x1a1f,0x1a1f,0x1a1f, +0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x27c,0x27c, +0x27c,0x27c,0x27c,0x27c,0x27c,0x27c,0x27c,0x27c,0x27c,0x27c,0x27c,0x1a22,0x1a31,0x1a31,0x1a31,0x1a31, +0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a28,0x1a28,0x1a28,0x1a28,0x1a2e,0x1a2e,0x1a2e,0x1a2e, +0x1a2e,0x1a2e,0x1a2e,0x1a2e,0x1a2e,0x1a2e,0x27f,0x27f,0x27f,0x27f,0x27f,0x1a2b,0x1a9d,0x1a9d,0x1a9d,0x1a9d, +0x1a9d,0x1a9a,0x1a9a,0x1a9a,0x1a9a,0x1a9a,0x1a9a,0x1a9a,0x282,0x282,0x282,0x282,0x282,0x282,0x282,0x282, +0x282,0x282,0x282,0x282,0x282,0x282,0x282,0x282,0x282,0x282,0x282,0x282,0x1ab8,0x1ab8,0x1ab8,0x1ab8, +0x1ab8,0x1ab8,0x1ab8,0x285,0x285,0x1ab8,0x285,0x285,0x1ab8,0x1ab8,0x1ab8,0x1ab8,0x1ab8,0x1ab8,0x1ab8,0x1ab8, +0x285,0x1ab8,0x1ab8,0x285,0x1ab8,0x1ab8,0x1ab8,0x1ab8,0x1ab8,0x1ab8,0x1ab8,0x1ab8,0x1ab8,0x1ab8,0x1ab8,0x1ab8, +0x1ab8,0x1ab8,0x1ab8,0x1ab8,0x1aa0,0x1aaf,0x1aaf,0x1aaf,0x1aaf,0x1aaf,0x285,0x1aaf,0x1ab2,0x285,0x285,0x1aa0, +0x1aa0,0x1ab5,0x1aa6,0x1abb,0x1aaf,0x1abb,0x1aaf,0x1aa3,0x1abe,0x1aa9,0x1abe,0x285,0x285,0x285,0x285,0x285, +0x285,0x285,0x285,0x285,0x1aac,0x1aac,0x1aac,0x1aac,0x1aac,0x1aac,0x1aac,0x1aac,0x1aac,0x1aac,0x285,0x285, +0x285,0x285,0x285,0x285,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7, +0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x288,0x288,0x288,0x288,0x288,0x288, +0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288, +0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x288,0x1ac4,0x1ac4,0x1ac4,0x1ac4, +0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x28e, +0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x28e, +0x28e,0x28e,0x28e,0x28e,0x28e,0x28e,0x28e,0x28e,0x28e,0x28e,0x28e,0x28e,0x28e,0x28e,0x28e,0x28e, +0x28e,0x28e,0x28e,0x28e,0x1ac7,0x1ac7,0x1ac7,0x1ac7,0x1ac7,0x1ac7,0x1ac7,0x1ac7,0x1ac7,0x1ac7,0x28e,0x28e, +0x28e,0x28e,0x28e,0x28e,0x1afa,0x1afa,0x1afa,0x1afa,0x1afa,0x1afa,0x1afa,0x1afa,0x1afa,0x291,0x291,0x291, +0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291, +0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291,0x291, +0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x294,0x1aca,0x1aca,0x1acd,0x294,0x294, +0x1ad0,0x1ad0,0x294,0x294,0x294,0x294,0x294,0x294,0x294,0x294,0x294,0x294,0x294,0x294,0x294,0x294, +0x1b63,0x1b63,0x1b63,0x1b63,0x1b63,0x1b63,0x1b63,0x1b63,0x1b60,0x1b63,0x1b63,0x1b63,0x1b63,0x1b63,0x1b63,0x297, +0x1b66,0x1b66,0x297,0x297,0x297,0x297,0x297,0x297,0x1b5d,0x1b5d,0x1b5d,0x1b5d,0x1b5d,0x1b5d,0x1b5d,0x1b5d, +0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c, +0x1b6c,0x1b69,0x1b69,0x29a,0x29a,0x29a,0x29a,0x29a,0x29a,0x29a,0x29a,0x29a,0x29a,0x29a,0x29a,0x29a, +0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x29d,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x29d,0x1b6f,0x1b6f,0x29d, +0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x1b6f,0x29d, +0x1b72,0x1b78,0x1b78,0x1b75,0x1b75,0x1b75,0x2a3,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75, +0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75, +0x1b75,0x2a3,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x1b75,0x2a3,0x2a3,0x2a3,0x2a3,0x2a3, +0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7e,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b, +0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x1b7b,0x2a6, +0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x1bf0,0x1bf0,0x1bf0,0x1bf0,0x1bf0,0x1bf0,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6, +0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6, +0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x2a6,0x1b84,0x1b84,0x1b81,0x1b81, +0x1b81,0x1b81,0x1b87,0x1b87,0x1b87,0x1b87,0x2a9,0x2a9,0x2a9,0x2a9,0x2a9,0x2a9,0x2a9,0x2a9,0x2a9,0x2a9, +0x2a9,0x2a9,0x2a9,0x2a9,0x2a9,0x2a9,0x2a9,0x2a9,0x2a9,0x2a9,0x2a9,0x2a9,0x1a9d,0x1a9d,0x1a9d,0x1a9d, +0x1a9d,0x1a9d,0x1a9d,0x1a9d,0x1a9d,0x1a9d,0x1a9d,0x1a9d,0x1a9d,0x1a9d,0x1a9d,0x1a9d,0x1b8d,0x1b8d,0x1b8d,0x1b8d, +0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d, +0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x2ac,0x1b8a,0x1b8a,0x1b8a,0x1b8a, +0x1b8a,0x1b8a,0x1b8a,0x1b8a,0x1b8a,0x1b8a,0x2ac,0x2ac,0x2ac,0x2ac,0x2ac,0x2ac,0x1554,0x1554,0x1554,0x1554, +0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1554,0x1b93,0x1b93,0x1b93,0x1b93, +0x1b93,0x1b93,0x1b93,0x1b93,0x1b93,0x1b93,0x1b93,0x1b93,0x1b93,0x1b93,0x1b90,0x2af,0x2af,0x2af,0x2af,0x2af, +0x2af,0x2af,0x2af,0x2af,0x2af,0x2af,0x2af,0x2af,0x2af,0x2af,0x2af,0x2af,0x1b9c,0x1b9c,0x1b9c,0x1b9c, +0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x2b2,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x1b9c,0x2b2, +0x1b9c,0x1b9c,0x2b2,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x2b2,0x1b99, +0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x2b2,0x1b99, +0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x1b99,0x2b2,0x1b99,0x1b99,0x2b2,0x2b2,0x2b2,0x1ba2,0x1ba2,0x1ba2,0x1ba2, +0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x2b5,0x2b5,0x1ba2,0x1ba2,0x1ba2,0x1ba2, +0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x2b5, +0x2b5,0x2b5,0x2b5,0x2b5,0x2b5,0x2b5,0x2b5,0x2b5,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f, +0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8, +0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8, +0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x2b8,0x1bf3,0x1bf3,0x1bf3,0x1bf9,0x1bf9,0x1bf9,0x1bf9, +0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bfc,0x1bfc,0x1bfc,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb, +0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x2bb,0x1bff,0x1bff,0x1bff,0x1bff, +0x1bff,0x1bff,0x1bff,0x1bff,0x1bff,0x1bff,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be, +0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be, +0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x2be,0x1c02,0x1c02,0x1c02,0x1c02,0x1c02,0x1c02,0x1c02,0x1c02, +0x1c02,0x1c02,0x1c02,0x1c02,0x1c02,0x1c02,0x1c02,0x1c02,0x1c02,0x1c02,0x1c02,0x1c02,0x2c1,0x2c1,0x2c1,0x2c1, +0x2c1,0x2c1,0x2c1,0x2c1,0x2c1,0x2c1,0x2c1,0x2c1,0x1c05,0x1c05,0x1c1a,0x1c11,0x1c17,0x1c17,0x1c17,0x1c17, +0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x2c4,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17, +0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17,0x1c17, +0x1c11,0x1c11,0x1c05,0x1c05,0x1c05,0x1c05,0x1c05,0x2c4,0x2c4,0x2c4,0x1c11,0x1c11,0x1c05,0x1c14,0x1c08,0x1c1d, +0x1c1d,0x1c0b,0x1c0b,0x1c0b,0x1c0b,0x1c0b,0x1c0b,0x1c0b,0x1c0b,0x1c0b,0x1c0b,0x1c0b,0x1c0e,0x1c0e,0x1c0e,0x1c0e, +0x1c0e,0x1c0e,0x1c0e,0x1c0e,0x1c0e,0x1c0e,0x2c4,0x2c4,0x2c4,0x2c4,0x2c4,0x2c4,0x1c26,0x1c26,0x1c26,0x1c26, +0x1c26,0x1c26,0x1c26,0x1c26,0x1c26,0x1c26,0x1c26,0x1c26,0x1c20,0x1c20,0x1c20,0x1c20,0x1c23,0x1c23,0x1c23,0x1c23, +0x1c23,0x1c23,0x1c23,0x1c23,0x1c23,0x1c23,0x2c7,0x2c7,0x2c7,0x2c7,0x2c7,0x2c7,0x2ca,0x2ca,0x2ca,0x2ca, +0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca, +0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x18db,0x2dc,0x2dc,0x2dc, +0x2dc,0x2dc,0x2dc,0x2dc,0x2dc,0x2dc,0x2dc,0x2dc,0x2dc,0x2dc,0x2dc,0x2dc,0x2ca,0x2ca,0x2ca,0x2ca, +0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca, +0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x95d,0x95d,0x1c3b,0x1c3b,0x1c3b,0x1c3b, +0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x2ca,0x2ca,0x2ca,0x2ca, +0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0x2ca,0xc6c,0xc6c,0xc6c,0xc6c, +0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0x129f,0x129f,0x129f,0x2cd,0x2cd,0xe94,0xe94,0xe94,0xe94, +0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94, +0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd, +0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd, +0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0x2cd,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88, +0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88, +0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0x2d0,0x2d0,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1, +0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x1bab,0x1bab,0x1bab, +0x1bab,0x1c29,0x2d3,0x2d3,0x2d3,0x2d3,0x2d3,0x2d3,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8, +0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8, +0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x2d6,0x2d6,0x1785,0x1785,0x2d9,0x2d9,0x2d9,0x2d9,0x2d9,0x2d9, +0x2d9,0x2d9,0x2d9,0x2d9,0x2d9,0x2d9,0x2d9,0x2d9,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db, +0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4, +0x1af4,0x1af4,0x1af4,0x2df,0x2df,0x2df,0x2df,0x2df,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b, +0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x3e7,0x3db,0x3db,0x3db,0x3db,0x3db,0x3db,0x3db, +0x3db,0x3e7,0x3e7,0x3e7,0x3e7,0x3e1,0x111f,0x12f6,0x3ea,0x927,0x92a,0x3d8,0x3d8,0x111c,0x12f3,0x12f3, +0x3ed,0x3ed,0x3ed,0x3ed,0x3ed,0x3ed,0x3ed,0x3ed,0x111c,0x3db,0x3db,0x3e7,0xcae,0x3ea,0x3ea,0x3ea, +0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea, +0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3db,0x3db,0x8af,0x8b2,0x945,0x945, +0x945,0x945,0x945,0x945,0x945,0x945,0x945,0x945,0x3e4,0xf7e,0xf7b,0x12f9,0x12f9,0x12f9,0x12f9,0x12f9, +0x14b8,0x1122,0x1122,0xed0,0xed0,0xda1,0xed0,0xed0,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea, +0x3ea,0x3ed,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ea,0x3ed,0x3ea,0x3ea,0x3ed,0x3ea,0x3ea,0x3ea, +0x3ea,0x3ea,0x12f3,0x12f6,0x3de,0x3ea,0x3e7,0x3e7,0x489,0x489,0x489,0x489,0x489,0x489,0x489,0x489, +0x489,0x12ff,0x489,0x489,0x489,0x489,0x489,0x489,0x489,0x489,0x489,0x489,0x489,0x489,0x489,0x489, +0x489,0x489,0x12ff,0x1857,0x1857,0xf9c,0x47a,0x483,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5, +0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0x4c5,0xba3, +0xba3,0xdb0,0xdb0,0x8b5,0xdad,0x13da,0x13da,0x13da,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8, +0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8, +0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4c8,0x4ce,0x4ce,0x4ce,0x1137,0x1137,0x1137,0x1137,0x1137, +0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb, +0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb,0x4cb, +0x4cb,0x4cb,0x1134,0x1134,0x1134,0x1134,0x1134,0x1134,0x4d1,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce, +0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce, +0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4ce,0x4da,0x4d4,0x4da,0x4d4, +0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4, +0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4d4,0x4d4, +0x4d4,0x4d4,0x4d7,0x9a2,0xfc9,0xfc9,0xfcc,0xfc9,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4, +0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4,0x4da,0x4d4, +0x4da,0x4d4,0xfcc,0xfc9,0xfcc,0xfc9,0xfcc,0xfc9,0x4e6,0x4e6,0x4e6,0x4e6,0x4e6,0x4e6,0x4e6,0x4e6, +0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e6,0x4e6,0x4e6,0x4e6,0x4e6,0x4e6,0x4e6,0x4e6, +0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x4e9,0x6a5,0x6a5,0x6a8,0x504,0x6b4,0x6b1,0x6b1,0x6ae, +0x52e,0x52e,0x4ec,0x4ec,0x4ec,0x4ec,0x4ec,0xb34,0x6b7,0x510,0x6cf,0x6d2,0x525,0x6b7,0x513,0x513, +0x504,0x51f,0x51f,0x6a5,0x52b,0x528,0x6ab,0x4fe,0x4f5,0x4f5,0x4f8,0x4f8,0x4f8,0x4f8,0x4f8,0x4fb, +0x4f8,0x4f8,0x4f8,0x4ef,0x537,0x534,0x531,0x531,0x6c3,0x519,0x516,0x6c0,0x6bd,0x6ba,0x6cc,0x507, +0x6c9,0x6c9,0x51c,0x51f,0x6c6,0x6c6,0x51c,0x51f,0x501,0x504,0x504,0x504,0x522,0x50d,0x50a,0xbb8, +0xad7,0xad7,0xad4,0xad4,0xad4,0xad4,0xbaf,0xbaf,0xbaf,0xbaf,0xbb5,0xcdb,0xcd8,0xdbc,0xdbf,0xbb2, +0xdbf,0xdbf,0xdbf,0xdbf,0xdbc,0xdbf,0xdbf,0xbac,0x55b,0x55b,0x55b,0x55b,0x55b,0x55b,0x55b,0x558, +0x55e,0x73e,0x55b,0x9a5,0x9c6,0xada,0xada,0xada,0xbbe,0xbbe,0xdc5,0xdc5,0xdc5,0xdc5,0x1140,0x1143, +0x1143,0x1314,0x14a6,0x14d0,0x14d3,0x14d3,0x16da,0x185a,0x56a,0x56a,0x582,0x6e4,0x567,0x6de,0x56a,0x57f, +0x567,0x6e4,0x579,0x582,0x582,0x582,0x579,0x579,0x582,0x582,0x582,0x6ea,0x567,0x582,0x6e7,0x567, +0x576,0x582,0x582,0x582,0x582,0x582,0x567,0x567,0x56d,0x6de,0x6e1,0x567,0x582,0x567,0x6ed,0x567, +0x582,0x570,0x588,0x6f0,0x582,0x582,0x573,0x579,0x582,0x582,0x585,0x582,0x579,0x57c,0x57c,0x57c, +0x57c,0xae3,0xae0,0xcde,0xdce,0xbd3,0xbd6,0xbd6,0xbd0,0xbcd,0xbcd,0xbcd,0xbcd,0xbd6,0xbd3,0xbd3, +0xbd3,0xbd3,0xbca,0xbcd,0xdcb,0xedc,0xedf,0xfd2,0x1146,0x1146,0x1146,0x6f6,0x6f3,0x58b,0x58e,0x58e, +0x58e,0x58e,0x58e,0x6f3,0x6f6,0x6f6,0x6f3,0x58e,0x6fc,0x6fc,0x6fc,0x6fc,0x6fc,0x6fc,0x6fc,0x6fc, +0x6fc,0x6fc,0x6fc,0x6fc,0x597,0x597,0x597,0x597,0x6f9,0x6f9,0x6f9,0x6f9,0x6f9,0x6f9,0x6f9,0x6f9, +0x6f9,0x6f9,0x591,0x591,0x591,0x591,0x591,0x591,0x59d,0x59d,0x59d,0x59d,0x59d,0x59d,0x59d,0x59d, +0x59a,0x59d,0x59d,0x59d,0x59d,0x59d,0x5a0,0x59a,0x59d,0x59d,0x59a,0x59a,0x59a,0x59a,0x59d,0x59d, +0x6ff,0x6ff,0x59a,0x59a,0x59d,0x59d,0x59d,0x59d,0x59d,0x59d,0x59d,0x59d,0x59d,0x59d,0x59d,0x59d, +0x59d,0x5a0,0x5a0,0x5a0,0x59d,0x59d,0x702,0x59d,0x702,0x59d,0x59d,0x59d,0x59d,0x59d,0x59d,0x59d, +0x59a,0x59d,0x59a,0x59a,0x59a,0x59a,0x59a,0x59a,0x59d,0x59d,0x59a,0x6ff,0x59a,0x59a,0x59a,0xae9, +0xae9,0xae9,0xae9,0xae9,0xae9,0xae9,0xae9,0xae9,0xbd9,0xbd9,0xbd9,0xbd9,0xbd9,0xbd9,0xbd9,0xbd9, +0xbd9,0xbd9,0xbd9,0xbd9,0x705,0x5a3,0x705,0x705,0x5a6,0x5a3,0x5a3,0x705,0x705,0x5a6,0x5a3,0x705, +0x5a6,0x5a3,0x5a3,0x705,0x5a3,0x705,0x5b2,0x5af,0x5a3,0x705,0x5a3,0x5a3,0x5a3,0x5a3,0x705,0x5a3, +0x5a3,0x705,0x705,0x705,0x705,0x5a3,0x5a3,0x705,0x5a6,0x705,0x5a6,0x705,0x705,0x705,0x705,0x705, +0x70b,0x5a9,0x705,0x5a9,0x5a9,0x5a3,0x5a3,0x5a3,0x705,0x705,0x705,0x705,0x5a3,0x5a3,0x5a3,0x5a3, +0x705,0x705,0x5a3,0x5a3,0x5a3,0x5a6,0x5a3,0x5a3,0x5a6,0x5a3,0x5a3,0x5a6,0x705,0x5a6,0x5a3,0x5a3, +0x705,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x705,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3, +0x5a3,0x5a3,0x5a3,0x5a3,0x708,0x705,0x5a6,0x5a3,0x705,0x705,0x705,0x705,0x5a3,0x5a3,0x705,0x705, +0x5a3,0x5a6,0x708,0x708,0x5a6,0x5a6,0x5a3,0x5a3,0x5a6,0x5a6,0x5a3,0x5a3,0x5a6,0x5a6,0x5a3,0x5a3, +0x5a3,0x5a3,0x5a3,0x5a3,0x5a6,0x5a6,0x705,0x705,0x5a6,0x5a6,0x705,0x705,0x5a6,0x5a6,0x5a3,0x5a3, +0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x705,0x5a3,0x5a3,0x5a3,0x705,0x5a3,0x5a3, +0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x705,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a6,0x5a6,0x5a6,0x5a6, +0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x705, +0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3, +0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3, +0x5a6,0x5a6,0x5a6,0x5a6,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a3,0x5a6,0x5a6,0x5a6,0x5a6,0x5a3,0x5ac, +0x5a3,0x5a3,0xbdc,0xbdc,0xbdc,0xbdc,0xbdc,0xbdc,0xbdc,0xbdc,0xbdc,0xbdc,0xbdc,0xbdc,0xbdc,0xbdc, +0x5b5,0xaec,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5be,0x5bb,0x5be,0x5bb,0x5b5,0x5b5,0x5b5,0x5b5, +0x5b5,0x5b5,0x70e,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x7fe,0x7fe,0x5b5,0x5b5,0x5b5,0x5b5, +0x5b8,0x5b8,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x804,0x801,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5, +0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5, +0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0x5b5,0xaec, +0xbe2,0xaec,0xaec,0xaec,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1, +0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1,0x5c1, +0x5c1,0x5c1,0x5c1,0x5c1,0x717,0x717,0x717,0x717,0x717,0x717,0x717,0x717,0x717,0x717,0x5c7,0xc3f, +0xc3f,0xc3f,0xc3f,0xc3f,0xc3f,0xc3f,0xc3f,0xc3f,0xc3f,0xc3f,0xc3f,0xc3f,0xc3f,0xc3f,0xc3f,0xc3f, +0xc3f,0xc3f,0xc3f,0xd4d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d, +0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x5ca,0x5cd,0x5cd,0x5cd,0x5cd,0x5cd,0x5cd,0x5cd, +0x5cd,0x5cd,0x5cd,0x5cd,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d, +0x5cd,0x5cd,0x5cd,0x5cd,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d, +0x71d,0x71d,0x71d,0x71d,0x720,0x720,0x720,0x720,0x720,0x720,0x720,0x720,0x720,0x720,0x720,0x720, +0x720,0x720,0x720,0x720,0x5d0,0x5d0,0x720,0x720,0x720,0x720,0xbe5,0xbe5,0xbe5,0xbe5,0xbe5,0xbe5, +0xbe5,0xbe5,0xbe5,0xbe5,0x726,0x726,0x5d3,0x723,0x723,0x723,0x723,0x723,0x723,0x723,0x5d3,0x5d3, +0x5d3,0x5d3,0x5d6,0x5d6,0x5d6,0x5d6,0x726,0x726,0x5d6,0x5d6,0x726,0x726,0x5d3,0x5d3,0x5d3,0x5d3, +0x726,0x726,0x5d6,0x5d6,0x726,0x726,0x5d3,0x5d3,0x5d3,0x5d3,0x726,0x726,0x723,0x5d3,0x5d6,0x726, +0x5d3,0x5d3,0x723,0x726,0x726,0x726,0x5d6,0x5d6,0x5d3,0x5d3,0x5d3,0x5d3,0x5d3,0x5d3,0x5d3,0x5d3, +0x5d3,0x5d3,0x5d3,0x5d3,0x5d3,0x5d3,0x726,0x723,0x726,0x723,0x5d3,0x5d6,0x5d6,0x5d6,0x5d6,0x5d6, +0x5d6,0x5d3,0x5d3,0x723,0xaf2,0xaf2,0xaf2,0xaf2,0xaf2,0xaf2,0xaf2,0xaf2,0xbe8,0xbe8,0xbe8,0xbe8, +0xbe8,0xc57,0xc57,0xbe8,0x5dc,0x5dc,0x5dc,0x5dc,0x5d9,0x72f,0x72f,0x5d9,0x5d9,0x729,0x5d9,0x5d9, +0x5d9,0x5d9,0x729,0x729,0x5d9,0x5d9,0x5d9,0x5d9,0xd56,0xd56,0xbeb,0xbeb,0xdd7,0xaf5,0x5dc,0x5dc, +0x72c,0x5df,0x72c,0x5dc,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9, +0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5dc,0x5dc,0x5dc, +0x5d9,0x5d9,0x5d9,0x5d9,0x72f,0x5d9,0x72f,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x807,0x807,0x807,0x807, +0x807,0x807,0x807,0x807,0x807,0x807,0x807,0x807,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9,0x5d9, +0x5d9,0x5d9,0x5d9,0x5d9,0x72f,0x72f,0x5e2,0x72f,0x729,0x729,0x5d9,0x729,0x72c,0x729,0x729,0x5d9, +0x729,0x72f,0x5e2,0x72f,0xaf5,0xaf5,0xbee,0xbee,0xbee,0xbee,0xbee,0xbee,0xbee,0xbee,0xbee,0xbee, +0xbee,0xbee,0xdd4,0xe8b,0x5e5,0x5e5,0x5e5,0x5e5,0x5e5,0x5e5,0x5e5,0x5e5,0x5e5,0x5e5,0x5e5,0x5e5, +0x5e5,0x5e5,0x5e5,0x5e5,0x5e5,0x5e5,0x5e5,0x5e5,0x5e8,0x139b,0x139b,0x139b,0x5e8,0x5e8,0x5e8,0x5e8, +0x5e8,0x5e8,0x5e8,0x5e8,0x14d9,0x5ee,0x5ee,0x5ee,0x5ee,0x139b,0x5e8,0x5e8,0x5ee,0x5ee,0x139e,0x139e, +0x5f4,0x5f4,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8, +0x5e8,0x5e8,0x5e8,0x5e8,0x139b,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8, +0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x735,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8, +0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x139b,0x5e8,0x139b,0x5e8,0x5e8,0x5e8,0x5e8,0x139b, +0x139b,0x139b,0x5e8,0x1299,0x5e8,0x5e8,0x5e8,0x5f1,0x5f1,0x5f1,0x5f1,0x1320,0x1320,0x5e8,0x5eb,0x5eb, +0x5ee,0x5e8,0x5e8,0x5e8,0xbf4,0xbf1,0xbf4,0xbf1,0xbf4,0xbf1,0xbf4,0xbf1,0xbf4,0xbf1,0xbf4,0xbf1, +0xbf4,0xbf1,0x732,0x732,0x732,0x732,0x732,0x732,0x732,0x732,0x732,0x732,0x5e8,0x5e8,0x5e8,0x5e8, +0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x139b,0x5e8,0x5e8,0x5e8, +0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x5e8,0x139b,0x615,0x615,0x615,0x615, +0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615, +0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x60c,0x60c,0x60c,0x60c,0x60c,0x60c, +0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f, +0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0xb46,0xb46,0xb46,0xb46, +0xb46,0xb46,0xb46,0xb46,0xb46,0xb46,0xb46,0xb46,0xb46,0xb46,0xb46,0xb46,0x615,0x615,0x95a,0x615, +0x615,0x615,0x615,0x615,0x615,0x615,0x60c,0x60c,0xbf7,0xd7a,0x1b0f,0x1b0f,0x612,0x618,0x615,0x60f, +0x612,0x618,0x615,0x60f,0x612,0x618,0x615,0x60f,0x612,0x618,0x615,0x60f,0x612,0x618,0x615,0x60f, +0x612,0x618,0x615,0x60f,0x612,0x618,0x615,0x60f,0x612,0x618,0x615,0x60f,0x615,0x60f,0x615,0x60f, +0x615,0x60f,0x615,0x60f,0x615,0x60f,0x615,0x60f,0x612,0x618,0x615,0x60f,0x612,0x618,0x615,0x60f, +0x612,0x618,0x615,0x60f,0x612,0x618,0x615,0x60f,0x615,0x60f,0x612,0x618,0x615,0x60f,0x615,0x60f, +0x612,0x618,0x615,0x60f,0x612,0x618,0x615,0x60f,0x615,0x60f,0x1323,0x1323,0x1323,0x1323,0x1323,0x1323, +0x1323,0x1323,0x1323,0x1323,0x1323,0x1323,0x1323,0x1323,0x615,0x60f,0x615,0x60f,0x615,0x60f,0x612,0x618, +0x612,0x618,0x615,0x60f,0x615,0x60f,0x615,0x60f,0x615,0x60f,0x615,0x60f,0x615,0x60f,0x615,0x60f, +0x612,0x615,0x60f,0x612,0x615,0x60f,0x612,0x618,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f, +0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x612, +0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615, +0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f, +0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x612,0x612,0x612,0x612,0x612,0x612,0x612, +0x618,0x618,0x618,0x618,0x618,0x618,0x618,0x618,0x60f,0x615,0x918,0x91b,0x1b0f,0x1b0f,0x1b0f,0x1b0f, +0x1b0f,0x1b0f,0x1b0f,0x1b0f,0x1b0f,0x1b0f,0x1b0f,0x1b0f,0x1b0f,0x1b0f,0x1b0f,0x1b0f,0x612,0x60f,0x612,0x612, +0x612,0x612,0x612,0x612,0x60f,0x612,0x60f,0x60f,0x612,0x612,0x60f,0x60f,0x612,0x612,0x60f,0x612, +0x60f,0x612,0x60f,0x60f,0x612,0x60f,0x60f,0x612,0x60f,0x612,0x60f,0x60f,0x612,0x60f,0x612,0x612, +0x60f,0x60f,0x60f,0x612,0x60f,0x60f,0x60f,0x60f,0x60f,0x612,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f, +0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f,0x60f, +0x612,0x612,0x60f,0x60f,0x612,0x60f,0x612,0x60f,0x60f,0x60f,0x60f,0x60f,0x612,0x612,0x612,0x612, +0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612, +0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x612,0x618, +0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615, +0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615, +0x618,0x618,0x618,0x618,0x618,0x618,0x618,0x618,0x618,0x618,0x618,0x618,0x618,0x618,0x618,0x618, +0x618,0x618,0x618,0x618,0x618,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615,0x615, +0x61b,0x61b,0x61b,0x61b,0xfde,0xfde,0xfde,0x14dc,0x14dc,0x14dc,0x14dc,0x14dc,0x14dc,0x14dc,0x16e0,0x16e0, +0x864,0x86a,0x86a,0x876,0x876,0x867,0x85e,0x867,0x85e,0x867,0x85e,0x867,0x85e,0x867,0x85e,0x867, +0x62a,0x62a,0x624,0x62a,0x624,0x62a,0x624,0x62a,0x624,0x62a,0x624,0x627,0x62d,0x62a,0x624,0x62a, +0x624,0x627,0x62d,0x62a,0x624,0x62a,0x624,0x627,0x62d,0x62a,0x624,0x627,0x62d,0x62a,0x624,0x627, +0x62d,0x62a,0x624,0x62a,0x624,0x62a,0x624,0x62a,0x624,0x62a,0x624,0x627,0x62d,0x62a,0x624,0x627, +0x62d,0x62a,0x624,0x627,0x62d,0x62a,0x624,0x627,0x62d,0x62a,0x624,0x627,0x62d,0x62a,0x624,0x627, +0x62d,0x62a,0x624,0x627,0x62d,0x62a,0x624,0x627,0x62d,0x62a,0x624,0x627,0x714,0x714,0x714,0x714, +0x714,0x714,0x714,0x714,0x714,0x714,0x714,0x714,0x714,0x714,0x714,0x714,0x714,0x714,0x714,0x714, +0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711, +0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711,0x711, +0x711,0x711,0x71a,0x71a,0x71a,0x71a,0x71a,0x71a,0x71a,0x71a,0x71a,0x71a,0x71a,0x71a,0x71a,0x71a, +0x71a,0x71a,0x71a,0x71a,0x717,0x717,0x717,0x717,0x717,0x717,0x717,0x717,0x717,0x717,0x717,0x717, +0x717,0x717,0x717,0x717,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d, +0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d,0x71d, +0x71d,0x71d,0x71d,0x71d,0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738, +0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738,0x738, +0x738,0x738,0x738,0x738,0xc45,0x8c7,0x8c1,0x8be,0x8c4,0x8bb,0x74d,0x750,0x750,0x750,0x750,0x750, +0x750,0x750,0x750,0x750,0x8cd,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d, +0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x74d, +0x74d,0x74d,0x74d,0x74d,0x74d,0x74d,0x8ca,0x8ca,0x753,0x8dc,0x8df,0x8e5,0x80a,0x816,0x8fa,0x813, +0x8d3,0x8d0,0x8d3,0x8d0,0x8d9,0x8d6,0x8d9,0x8d6,0x8d3,0x8d0,0x810,0x8e5,0x8d3,0x8d0,0x8d3,0x8d0, +0x8d3,0x8d0,0x8d3,0x8d0,0x8eb,0x8f1,0x8ee,0x8ee,0x759,0x795,0x795,0x795,0x795,0x795,0x795,0x78f, +0x78f,0x78f,0x78f,0x78f,0x78f,0x78f,0x78f,0x78f,0x78f,0x78f,0x78f,0x78f,0x78f,0x78f,0x78f,0x78f, +0x78f,0x78f,0x78f,0x75c,0x777,0x756,0x77d,0x780,0x77a,0x792,0x792,0x792,0x792,0x792,0x792,0x78c, +0x78c,0x78c,0x78c,0x78c,0x78c,0x78c,0x78c,0x78c,0x78c,0x78c,0x78c,0x78c,0x78c,0x78c,0x78c,0x78c, +0x78c,0x78c,0x78c,0x75c,0x777,0x756,0x777,0xc48,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8, +0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8, +0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x7f8,0x1293,0x1293,0x1293,0x1293,0x1293,0x7fb, +0x810,0x813,0x813,0x813,0x813,0x813,0x813,0x813,0x813,0x813,0x933,0x933,0x933,0x933,0x819,0x819, +0x8e8,0x8f7,0x8f7,0x8f7,0x8f7,0x8f4,0x80d,0x8e2,0xb19,0xb19,0xb19,0xc5a,0xc78,0xc75,0xb37,0x8b8, +0x81f,0x81c,0x81f,0x822,0x81c,0x81f,0x81c,0x81f,0x81c,0x81f,0x81c,0x81c,0x81c,0x81c,0x81c,0x81c, +0x81f,0x81f,0x81c,0x81f,0x81f,0x81c,0x81f,0x81f,0x81c,0x81f,0x81f,0x81c,0x81f,0x81f,0x81c,0x81c, +0xc7b,0x831,0x82b,0x831,0x82b,0x831,0x82b,0x831,0x82b,0x831,0x82b,0x82b,0x82e,0x82b,0x82e,0x82b, +0x82e,0x82b,0x82e,0x82b,0x82e,0x82b,0x82e,0x82b,0x82e,0x82b,0x82e,0x82b,0x82e,0x82b,0x82e,0x82b, +0x82e,0x82b,0x82e,0x831,0x82b,0x82e,0x82b,0x82e,0x82b,0x82e,0x82b,0x82b,0x82b,0x82b,0x82b,0x82b, +0x82e,0x82e,0x82b,0x82e,0x82e,0x82b,0x82e,0x82e,0x82b,0x82e,0x82e,0x82b,0x82e,0x82e,0x82b,0x82b, +0x82b,0x82b,0x82b,0x831,0x82b,0x831,0x82b,0x831,0x82b,0x82b,0x82b,0x82b,0x82b,0x82b,0x831,0x82b, +0x82b,0x82b,0x82b,0x82b,0x82e,0x831,0x831,0x82e,0x82e,0x82e,0x82e,0x900,0x903,0x834,0x837,0xc63, +0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d, +0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d, +0x840,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d, +0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x83d,0x849,0x849,0x849,0x849, +0x849,0x849,0x849,0x849,0x849,0x849,0x849,0x849,0x849,0x849,0x849,0x849,0x849,0x849,0x849,0x849, +0x849,0x849,0x849,0x849,0x849,0x849,0x849,0x849,0xd5f,0xd5f,0xe8e,0x843,0x90c,0x90c,0x90c,0x90c, +0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0xd59,0xd59,0xd59,0xd59,0x84c,0x84c,0x84c,0x84c, +0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c, +0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x84c,0x1a58,0x912,0x912,0x912,0x912, +0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x84f,0x84f,0x84f, +0x84f,0x84f,0x84f,0xd62,0xd62,0xd62,0xd62,0x915,0x915,0x915,0x915,0x915,0x84f,0x84f,0x84f,0x84f, +0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f, +0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0x84f,0xd62,0xd62, +0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852, +0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852,0x852, +0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x855,0x855,0x855,0x855,0x855,0x855,0x855,0x855, +0x855,0x855,0x855,0x855,0x855,0x855,0x855,0x855,0x855,0x855,0x855,0x855,0x855,0x855,0x855,0x855, +0x855,0x855,0x855,0x855,0x855,0x855,0x855,0x855,0x855,0x855,0xe91,0xe91,0xe91,0xe91,0xe91,0xe91, +0xe91,0xe91,0xe91,0xe91,0xe91,0xe91,0xe91,0xe91,0xe91,0xe91,0xe91,0xe91,0xe91,0xe91,0xe91,0xe91, +0x1104,0x1104,0x1104,0x1104,0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858, +0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x858, +0x858,0x858,0x858,0x858,0x858,0x858,0x85b,0x85b,0x858,0x85b,0x858,0x85b,0x85b,0x858,0x858,0x858, +0x858,0x858,0x858,0x858,0x858,0x858,0x858,0x85b,0x858,0x85b,0x858,0x85b,0x85b,0x858,0x858,0x85b, +0x85b,0x85b,0x858,0x858,0x858,0x858,0x1497,0x1497,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c, +0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c, +0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c, +0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x90c,0x12d2,0x12d2,0x12d2,0x12d2,0x127b,0x127b,0x127b,0x127b, +0x127b,0x127b,0x127b,0x127b,0xd59,0xc66,0xc66,0xc66,0xc66,0xc66,0xc66,0xc66,0xc66,0xc66,0xc66,0xc66, +0xc66,0xc66,0xc66,0xc66,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f, +0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f,0x90f, +0x90f,0x90f,0x90f,0x90f,0x90f,0xc66,0xc66,0xc66,0xc66,0xc66,0xc66,0xc66,0xc66,0xc66,0xc66,0xc66, +0xc66,0xc66,0xc66,0xc66,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912, +0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912,0x912, +0x912,0x912,0x912,0xd62,0x99c,0x97e,0x97e,0x97e,0x97e,0x978,0x97e,0x97e,0x990,0x97e,0x97e,0x97b, +0x987,0x98d,0x98d,0x98d,0x98d,0x98d,0x990,0x978,0x984,0x978,0x978,0x978,0x96c,0x96c,0x978,0x978, +0x978,0x978,0x978,0x978,0x993,0x993,0x993,0x993,0x993,0x993,0x993,0x993,0x993,0x993,0x978,0x978, +0x978,0x978,0x978,0x978,0x978,0x978,0x978,0x978,0x97b,0x96c,0x978,0x96c,0x978,0x96c,0x98a,0x981, +0x98a,0x981,0x999,0x999,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8, +0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8,0x9a8, +0x9a8,0x9a8,0x9a8,0x9a8,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab, +0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab,0x9ab, +0x9ab,0x9ab,0x9ab,0x9ab,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae, +0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae,0x9ae, +0x9ae,0x9ae,0x9ae,0x9ae,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7, +0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7, +0x9b7,0x9b7,0x9b1,0x9b1,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba, +0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba, +0x9ba,0x9ba,0x9b4,0x9b4,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7, +0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7,0x9b7, +0x9b7,0x9b7,0x9b7,0x9b7,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba, +0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba,0x9ba, +0x9ba,0x9ba,0x9ba,0x9ba,0x9bd,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0, +0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0, +0x9bd,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0, +0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0x9c0,0xa4d,0xa4d,0xfc3,0xa4d, +0xa4d,0xa4d,0xa50,0xa4d,0xfc3,0xa4d,0xa4d,0xfba,0xa47,0xa3b,0xa3b,0xa3b,0xa3b,0xa4a,0xa3b,0xfab, +0xfab,0xfab,0xa3b,0xa3e,0xa47,0xa41,0xfb1,0xfbd,0xfbd,0xfab,0xfab,0xfc3,0xb3d,0xb3d,0xb3d,0xb3d, +0xb3d,0xb3d,0xb3d,0xb3d,0xb3d,0xb3d,0xa53,0xa53,0xa44,0xa44,0xa44,0xa44,0xa4d,0xa4d,0xa4d,0xa4d, +0xa4d,0xa4d,0xa4a,0xa4a,0xa3b,0xa3b,0xfc3,0xfc3,0xfc3,0xfc3,0xfab,0xfab,0xa4d,0xa4d,0xa4d,0xa4d, +0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d, +0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa4d,0xa62,0xa62,0xa62,0xa62, +0xa62,0xa62,0xa62,0xdb9,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62, +0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62, +0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xdb9,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62, +0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa62,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68, +0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68, +0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa68,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e, +0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6b,0xa71,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0x113d, +0x113d,0x113d,0x113d,0x113d,0x113d,0x113d,0x113d,0x113d,0x113a,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e, +0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e, +0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa6e,0xa83,0xa83,0xa83,0xa83, +0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83, +0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xa83,0xaa7,0xaa7,0xaa7,0xaaa, +0xaaa,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7, +0xa8f,0xa8f,0xaa4,0xa86,0xa86,0xa86,0xa86,0xa86,0xa86,0xa86,0xaa4,0xaa4,0xaa7,0xaa7,0xaa7,0xaa7, +0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7, +0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xaa7,0xac8,0xac8,0xac8,0xac8, +0xac8,0xab3,0xab3,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8, +0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8, +0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xacb,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8, +0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8,0xac8, +0xac8,0xac8,0xac8,0xac8,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec, +0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xaec,0xbe2, +0xbe2,0xbe2,0xbe2,0xbe2,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8, +0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8,0xaf8, +0xaf8,0xaf8,0xaf8,0xaf8,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a, +0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a,0xb0a, +0xb0a,0xb0a,0xb0a,0xb0a,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10, +0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10,0xb10, +0xb10,0xb10,0xb10,0xb10,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c, +0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0xb1c,0x13a1,0x13a1,0x13a1,0x1ad3, +0x1ad3,0x1ad3,0x1ad3,0x1ad3,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f, +0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f, +0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0xb1f,0x1ad6,0x1ad6,0x1ad6,0x1ad6,0x1ad6,0x1ad6,0x1ad6,0x1ad6,0x1ad6,0x1ad6, +0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22, +0xb22,0xb22,0xb22,0xb22,0xb22,0xb25,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22, +0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22, +0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb22,0xb28,0xb28,0xc69,0xc69,0xb28,0xb28,0xb28,0xb28, +0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xc69,0xb28,0xb28,0xb28, +0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb28,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c, +0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c, +0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0xb4c,0x14df,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xced,0xced, +0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52, +0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xb52,0xcea,0xcea,0xd38,0xd38,0xd38,0xd38, +0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xb55,0xb55,0xb55,0xb55, +0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55, +0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb55,0xb58,0xb58,0xb58,0xb58, +0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58, +0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb58,0xb67,0xb67,0xb67,0xb67, +0xb67,0xb5e,0xb6a,0xb70,0xb70,0xb70,0xb64,0xb64,0xb64,0xb6d,0xb61,0xb61,0xb61,0xb61,0xb61,0xb5b, +0xb5b,0xb5b,0xb5b,0xb5b,0xb5b,0xb5b,0xb5b,0xb70,0xb70,0xb70,0xb70,0xb70,0xb64,0xb64,0xb64,0xb64, +0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64, +0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb67,0xb67, +0xb70,0xb70,0xb70,0xb64,0xb64,0xb70,0xb70,0xb70,0xb70,0xb70,0xb70,0xb70,0xb64,0xb64,0xb64,0xb64, +0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64, +0xb64,0xb64,0xb70,0xb70,0xb70,0xb70,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64, +0xb64,0xb64,0xb64,0xb67,0xb67,0xb67,0xb67,0xb67,0xb67,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64, +0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0xb64, +0xb64,0xb64,0xb64,0xb64,0xb64,0xb64,0x16e3,0x16e3,0xb7c,0xb73,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79, +0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79, +0xb79,0xb79,0xb79,0xb73,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c, +0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c, +0xb7c,0xb7c,0xb7c,0xb73,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79, +0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb73,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79, +0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c, +0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb73,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79, +0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb76,0xb76,0xb76,0xb76, +0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76, +0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb76,0xb7c,0xb7c,0xb7c,0xb7c, +0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c, +0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79, +0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb7c,0xb7c,0xb7c,0xb7c, +0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c, +0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb7c,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79, +0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79,0xb79, +0xb7c,0xb7c,0xb7c,0xb7c,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f, +0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f,0xb7f, +0xb7f,0xb7f,0xb7f,0xb7f,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85, +0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85, +0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0xb85,0x1adc,0x1adc,0x1adc,0x1adc,0x1adc,0x1adc,0x1adc,0x1ba8,0x1ba8, +0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88, +0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88,0xb88, +0xbe2,0xbe2,0xbe2,0xbe2,0xbe2,0xbe2,0xbe2,0xbe2,0xbe2,0xbe2,0xbe2,0xbe2,0xbe2,0xbe2,0xbe2,0xbe2, +0xbe2,0xbe2,0xbe2,0xbe2,0xbe2,0xbe2,0xbdf,0xbe2,0xbdf,0xbdf,0xbdf,0xbdf,0xbdf,0xbdf,0xbdf,0xbdf, +0xbdf,0xbdf,0xbdf,0xbdf,0xbdf,0xbdf,0xbdf,0xce1,0xce4,0xdd1,0xdd1,0xdd1,0xdd1,0xdd1,0xdd1,0xdd1, +0xdd1,0xdd1,0xdd1,0xdd1,0xee8,0xee8,0xee8,0xee8,0xbee,0xbee,0xbee,0xbee,0xbee,0xbee,0xbee,0xbee, +0xbee,0xbee,0xce7,0xce7,0xce7,0xce7,0xce7,0xce7,0xce7,0xce7,0xdd4,0xe88,0xdd4,0xdd4,0xdd4,0xdd4, +0xdd4,0xdd4,0xdd4,0xdd4,0xdd4,0xfd8,0x1272,0x1272,0xddd,0xddd,0xddd,0xddd,0xddd,0xde3,0xde0,0xefa, +0xefa,0xefa,0xefa,0x13e0,0xfea,0x13e0,0x132c,0x132c,0xc21,0xc21,0xc21,0xc21,0xc21,0xc21,0xc21,0xc21, +0xc21,0xc21,0xc21,0xc21,0xc21,0xc21,0xc21,0xc21,0xc21,0xc21,0xc4e,0xc4b,0xc4e,0xc4b,0xc4e,0xc4b, +0x10fe,0x10fb,0xff0,0xfed,0xc24,0xc24,0xc24,0xc24,0xc24,0xc24,0xc24,0xc24,0xc24,0xc24,0xc24,0xc24, +0xc24,0xc24,0xc24,0xc24,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27, +0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27,0xc27, +0xc27,0xc27,0xc27,0xc27,0xc2a,0xc2a,0xc2a,0xc30,0xc2d,0xc54,0xc51,0xc30,0xc2d,0xc30,0xc2d,0xc30, +0xc2d,0xc30,0xc2d,0xc30,0xc2d,0xc30,0xc2d,0xc30,0xc2d,0xc30,0xc2d,0xc30,0xc2d,0xc2a,0xc2a,0xc2a, +0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a, +0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a, +0xc30,0xc2d,0xc30,0xc2d,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a, +0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a,0xc2a, +0xc30,0xc2d,0xc2a,0xc2a,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33, +0xc39,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33, +0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33, +0xc33,0xc33,0xc33,0xc33,0xc39,0xc39,0xc39,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33, +0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33,0xc33, +0xc33,0xc33,0xc33,0xc33,0xc36,0xc33,0xc33,0xc33,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c, +0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c, +0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xc6c,0xce7,0xd53,0xdd4,0xdd4,0xdd4,0xdd4,0xdd4,0xdd4, +0xdd4,0xdd4,0xe88,0xe88,0xdd4,0xdd4,0xdd4,0xdd4,0xdd4,0xdd4,0xeeb,0xfd8,0xfd8,0xfd8,0xfd8,0xfd8, +0xfd8,0xfd8,0xfd8,0xfd8,0xfd8,0x1296,0x1296,0x1275,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b, +0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b, +0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd0b,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd11,0xd11, +0xd11,0xd11,0xd11,0xd0e,0xd23,0xd23,0xd23,0xd1d,0xd23,0xd23,0xd23,0xd23,0xd23,0xd23,0xd23,0xd23, +0xd23,0xd23,0xd23,0xd1d,0xd23,0xd23,0xd23,0xd23,0xd17,0xd17,0xd20,0xd20,0xd20,0xd20,0xd14,0xd14, +0xd14,0xd14,0xd14,0xd1a,0xde9,0xde9,0xde9,0xde9,0xde9,0xde9,0xde9,0xde9,0xde9,0xde9,0xde9,0xde9, +0xde6,0xde9,0xde9,0xde9,0xde9,0xde9,0xde9,0xde9,0xd23,0xd23,0xd23,0xd23,0xd23,0xd23,0xd23,0xd23, +0xd23,0xd23,0xd23,0xd23,0xd23,0xd23,0xd1d,0xd23,0xd23,0xd23,0xd23,0xd23,0xd23,0xd23,0xd23,0xd23, +0xd23,0xd23,0xd23,0xd23,0xd23,0xd17,0xd17,0xd17,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a, +0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a, +0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd1a,0xd26,0xd26,0xd26,0xd26,0xd26,0xd26,0xd26,0xd26, +0xd26,0xd26,0xd26,0xd26,0xd26,0xd26,0xdec,0xdec,0xdec,0xdec,0xdec,0xdec,0xefd,0xefd,0xefd,0xefd, +0xefd,0xefd,0xefd,0x1107,0x1107,0xff3,0xff3,0xff3,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29, +0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29, +0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd29,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f, +0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f, +0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd2f,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38, +0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38, +0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd38,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44, +0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44, +0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd44,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50, +0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50, +0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xd50,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2, +0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2, +0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf2,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8, +0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf5,0xdf5,0xdf5, +0xdf5,0xdf5,0xdf5,0xdf5,0xdf5,0xdf5,0xdf5,0xdf5,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8, +0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8, +0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xdf8,0xeb8,0xeb8,0xe0a,0xe0a,0xf00,0xf00,0xf00,0xf00, +0xf00,0xf00,0xf00,0xfff,0xfff,0x1002,0xfff,0xfff,0xffc,0xffc,0xffc,0xffc,0xffc,0xffc,0xffc,0xffc, +0xffc,0xffc,0xffc,0xffc,0xffc,0xffc,0xffc,0xffc,0xe19,0xe16,0xe19,0xe16,0xe19,0xe16,0xe19,0xe16, +0xe19,0xe16,0xe19,0xe16,0xe19,0xe16,0xe19,0xe16,0xe19,0xe16,0xe19,0xe16,0xe19,0xe16,0xe19,0xe16, +0xe19,0xe16,0xe19,0xe16,0xe19,0xe16,0xe19,0xe16,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25, +0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25, +0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe25,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b, +0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0x1b1e,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28, +0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0xe28,0x1b1b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b, +0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b, +0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xe2b,0xeb2,0xeb2,0xeb2,0xeb2,0xeb2,0xeb2,0xeb2,0xeb2, +0xe43,0xe43,0xe43,0xe43,0xe43,0xe43,0xe43,0xe43,0xe43,0xe43,0xe43,0xe43,0xe43,0xe43,0xe43,0xf03, +0xf03,0xf03,0xf03,0x1005,0x1005,0x1005,0x1005,0x1005,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c, +0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c, +0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe4c,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55, +0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55, +0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe55,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e, +0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e, +0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe58,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b, +0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b,0xe5b, +0xe5b,0xe5b,0xe5b,0xe5e,0xe5e,0xe5e,0xe5e,0xe5e,0xe67,0xe67,0xe67,0xe67,0xe67,0xe67,0xe67,0xe67, +0xe67,0xe67,0xe67,0xe67,0xe67,0xe67,0xe64,0xe64,0xe64,0xe64,0xe64,0xe64,0xe64,0xe64,0xe61,0xe6a, +0x1011,0x100b,0x101a,0x1008,0xe67,0xe67,0x1008,0x1008,0xe79,0xe79,0xe6d,0xe79,0xe79,0xe79,0xe70,0xe79, +0xe79,0xe79,0xe79,0xe6d,0xe79,0xe79,0xe79,0xe79,0xe79,0xe79,0xe79,0xe79,0xe79,0xe79,0xe79,0xe79, +0xe79,0xe79,0xe79,0xe79,0xe79,0xe79,0xe79,0xe79,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c, +0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c, +0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe7c,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94, +0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94, +0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xe94,0xeb5,0xeb5,0xeb5,0xeb5,0xeb5,0xeb5,0xeb5,0xeb5, +0xeb5,0xeb5,0xeb5,0xeb5,0xeb5,0xeb5,0xeb5,0xeb5,0x1110,0x1110,0x1110,0x1110,0x1110,0x1110,0x1110,0x1110, +0x1110,0x1110,0x1110,0x1110,0x1110,0x1110,0x1110,0x1110,0xee8,0xee8,0xee8,0xee5,0xee5,0xee5,0xee5,0xee5, +0x1149,0x1392,0x1392,0x1392,0x1392,0x1317,0x1317,0x1317,0x1395,0x131a,0x131a,0x1395,0x14d6,0x14d6,0x14d6,0x14d6, +0x14d6,0x14d6,0x14d6,0x1797,0x1797,0x1797,0x1797,0x185d,0xefd,0xefd,0xefd,0xefd,0xff3,0xff3,0xff3,0xff3, +0xff3,0xff3,0xff3,0xff3,0xff3,0xff3,0xff3,0xff3,0xff6,0xff6,0xff6,0xff6,0xff6,0xff6,0xff6,0xff6, +0xff6,0xff6,0xff6,0xff6,0xff6,0xff6,0xff6,0xff6,0xffc,0xffc,0xffc,0xffc,0xffc,0xffc,0xffc,0x14f1, +0x14f1,0x14f1,0x14f1,0x14f1,0x14f1,0x14f1,0x14f1,0x14f1,0x14f1,0x14f1,0x14f1,0x14f1,0x14f1,0x14f4,0x1866,0x1866, +0x18e7,0x1866,0x1bc6,0x179d,0x132f,0x1152,0xf00,0xf00,0xf1e,0xf1e,0xf1e,0xf1e,0xf30,0xf39,0xf3c,0xf39, +0xf3c,0xf39,0xf3c,0xf39,0xf3c,0xf39,0xf3c,0xf39,0xf39,0xf39,0xf3c,0xf39,0xf39,0xf39,0xf39,0xf39, +0xf39,0xf39,0xf39,0xf39,0xf39,0xf39,0xf39,0xf39,0xf39,0xf39,0xf39,0xf39,0xf39,0xf39,0xf39,0xf39, +0xf21,0xf1e,0xf1e,0xf1e,0xf1e,0xf1e,0xf1e,0xf33,0xf1e,0xf33,0xf30,0xf30,0xf45,0xf42,0xf45,0xf45, +0xf45,0xf42,0xf42,0xf45,0xf42,0xf45,0xf42,0xf45,0xf42,0x102c,0x102c,0x102c,0x1167,0x1023,0x102c,0x1023, +0xf42,0xf45,0xf42,0xf42,0x1023,0x1023,0x1023,0x1023,0x1026,0x1029,0x1167,0x1167,0xf48,0xf48,0x103e,0x1035, +0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x1035,0x1035,0x103e,0x1035, +0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0xf4e,0xf4e,0xf4e,0xf4e, +0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e, +0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf4e,0xf5d,0xf5d,0xf5d,0xf5d, +0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d, +0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0xf5d,0x1512, +0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512,0x1512, +0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63, +0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63,0xf63, +0xfab,0xfc3,0xfba,0xfc0,0xfc0,0xfc3,0xfc3,0xfba,0xfba,0xfc0,0xfc0,0xfc0,0xfc0,0xfc0,0xfc3,0xfc3, +0xfc3,0xfab,0xfab,0xfab,0xfab,0xfc3,0xfc3,0xfc3,0xfc3,0xfc3,0xfc3,0xfc3,0xfc3,0xfc3,0xfc3,0xfc3, +0xfc3,0xfc3,0xfab,0xfba,0xfbd,0xfab,0xfab,0xfc0,0xfc0,0xfc0,0xfc0,0xfc0,0xfc0,0xfae,0xfc3,0xfc0, +0xfb7,0xfb7,0xfb7,0xfb7,0xfb7,0xfb7,0xfb7,0xfb7,0xfb7,0xfb7,0x1131,0x1131,0x112e,0x112b,0xfb4,0xfb4, +0xfdb,0xfdb,0xfdb,0xfdb,0x1296,0x1296,0x1275,0x1275,0x1275,0x1272,0x1272,0x1272,0x1272,0x1275,0x1398,0x1275, +0x1275,0x1275,0x1272,0x1275,0x1296,0x1272,0x1272,0x1272,0x1275,0x1275,0x1272,0x1272,0x1275,0x1272,0x1272,0x1275, +0xff6,0xff6,0xff6,0xff6,0xff6,0xff3,0xff3,0xff6,0xff6,0xff6,0xff6,0xff6,0xff6,0x14eb,0x14eb,0x14eb, +0x1107,0xff3,0xff3,0xff3,0xff3,0x12a2,0x127e,0x127e,0x127e,0x127e,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb, +0x1017,0x1017,0x1014,0x100e,0x1014,0x100e,0x1014,0x100e,0x1014,0x100e,0x100b,0x100b,0x100b,0x100b,0x1020,0x101d, +0x100b,0x1164,0x13ec,0x13ef,0x13ef,0x13ec,0x13ec,0x13ec,0x13ec,0x13ec,0x13f2,0x13f2,0x1506,0x14fa,0x14fa,0x14f7, +0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x1032,0x102f,0x102f,0x103e,0x1035,0x133b,0x1338,0x16ec, +0x133b,0x1338,0x13fb,0x13f8,0x1509,0x1509,0x150f,0x1509,0x150f,0x1509,0x150f,0x1509,0x150f,0x1509,0x150f,0x1509, +0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035, +0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x1035, +0x1038,0x1035,0x1035,0x1035,0x1035,0x1035,0x1035,0x1035,0x1035,0x103e,0x1035,0x103e,0x1035,0x103e,0x103e,0x1035, +0x1041,0x1041,0x1047,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d, +0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d,0x104d, +0x104d,0x1047,0x1041,0x1041,0x1041,0x1041,0x1047,0x1047,0x1041,0x1041,0x104a,0x1404,0x1401,0x1401,0x104d,0x104d, +0x1044,0x1044,0x1044,0x1044,0x1044,0x1044,0x1044,0x1044,0x1044,0x1044,0x1407,0x1407,0x1407,0x1407,0x1407,0x1407, +0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062, +0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062,0x1062, +0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x106b, +0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x106b,0x106e,0x106e,0x106e,0x1071,0x106e,0x106e,0x1074,0x1074, +0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077, +0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077,0x1077, +0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1083,0x107a,0x1089,0x1086, +0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080, +0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080,0x1080, +0x1341,0x133e,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x1098,0x1119, +0x108c,0x108c,0x108c,0x1092,0x140a,0x140a,0x140a,0x140a,0x140a,0x140a,0x140a,0x140a,0x108f,0x108f,0x1092,0x109e, +0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095, +0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095,0x109b,0x1095, +0x151e,0x151b,0x151e,0x151b,0x1521,0x1521,0x16f5,0x140a,0x10a7,0x10a7,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa, +0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa, +0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10aa,0x10a7,0x10a7,0x10a7,0x10a7,0x10a7,0x10a7,0x10a7,0x10a7, +0x10a7,0x10a7,0x10a7,0x10a7,0x10b9,0x10b9,0x10b9,0x10b9,0x10b9,0x10b9,0x10b0,0x10b0,0x10b0,0x10b0,0x10b0,0x10b3, +0x10b3,0x10b3,0x110d,0x10bc,0x10cb,0x10cb,0x10cb,0x10cb,0x10cb,0x10cb,0x10cb,0x10cb,0x10cb,0x10cb,0x10cb,0x10cb, +0x10cb,0x10cb,0x10cb,0x10cb,0x10b6,0x10b6,0x10b6,0x10b6,0x10b6,0x10b6,0x10b6,0x10b6,0x10b6,0x10b6,0x10b9,0x10b9, +0x10b9,0x10b9,0x10b9,0x10b9,0x10b9,0x10b9,0x10b9,0x10b9,0x10b9,0x10b9,0x10b9,0x10b9,0x10b9,0x10b9,0x10b9,0x10b9, +0x10b9,0x10b9,0x10b9,0x10b9,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da, +0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da,0x10da, +0x10da,0x10da,0x10da,0x10da,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec, +0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec,0x10ec, +0x10ec,0x10ec,0x10ec,0x10ec,0x10f5,0x10f5,0x10f5,0x10f5,0x110a,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5, +0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5,0x10f5, +0x10f5,0x10f5,0x10f5,0x10f5,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8, +0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8,0x10f8, +0x10f8,0x10f8,0x10f8,0x10f8,0x1104,0x1104,0x1104,0x1104,0x129c,0x129c,0x129c,0x129c,0x129c,0x129c,0x129c,0x129c, +0x1494,0x1776,0x1776,0x1776,0x1776,0x1776,0x1776,0x1776,0x1776,0x1776,0x18c6,0x18c6,0x18c6,0x18c6,0x18c6,0x18c6, +0x18c6,0x18c6,0x18c6,0x18c6,0x1179,0x1179,0x1179,0x1179,0x1179,0x1179,0x1179,0x1179,0x1179,0x1179,0x1179,0x1179, +0x1179,0x1179,0x1179,0x1179,0x1179,0x1179,0x1179,0x1179,0x1179,0x1179,0x1170,0x1170,0x1173,0x1173,0x1179,0x1170, +0x1170,0x1170,0x1170,0x1170,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f, +0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f,0x117f, +0x117f,0x117f,0x117f,0x117f,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a, +0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a,0x119a, +0x119a,0x119a,0x119a,0x119a,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6, +0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6,0x11a6, +0x11a6,0x11a6,0x11a3,0x11a9,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5, +0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5,0x11b5, +0x11b5,0x11b5,0x11b5,0x11b5,0x11bb,0x11bb,0x11bb,0x11bb,0x11bb,0x11bb,0x11bb,0x11bb,0x11bb,0x11bb,0x11bb,0x11bb, +0x11bb,0x11bb,0x11bb,0x11bb,0x11bb,0x12e7,0x11c1,0x12ea,0x11c1,0x11c1,0x11c1,0x11c1,0x11be,0x11be,0x11be,0x11c1, +0x16f8,0x16fb,0x1923,0x1920,0x11c4,0x11c4,0x11c4,0x11d3,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9, +0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9, +0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11d9,0x11c7,0x11d3,0x11d3,0x11c4,0x11c4,0x11c4,0x11c4,0x11d3,0x11d3, +0x11c4,0x11c4,0x11d3,0x11d3,0x11e5,0x11e5,0x11e5,0x11e5,0x11e5,0x11e5,0x11e5,0x11e5,0x11e5,0x11e5,0x11e5,0x11e5, +0x11e5,0x11e5,0x11e5,0x11e5,0x11e8,0x11e5,0x11e5,0x11e5,0x11e5,0x11e5,0x11e5,0x11df,0x11df,0x11df,0x11e5,0x11e2, +0x1527,0x152a,0x152d,0x152d,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7, +0x11f7,0x11f7,0x11f7,0x11f7,0x11eb,0x11f7,0x11eb,0x11eb,0x11eb,0x1200,0x1200,0x11eb,0x11eb,0x1200,0x11f7,0x1200, +0x1200,0x11f7,0x11eb,0x11ee,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7, +0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7,0x11f7, +0x11f7,0x11f7,0x11f7,0x11f7,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212, +0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212,0x1212, +0x1212,0x1212,0x1212,0x1212,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a, +0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a,0x122a, +0x122a,0x1227,0x1227,0x1227,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233, +0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233,0x1233, +0x1233,0x1233,0x1233,0x1233,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242, +0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242,0x1242, +0x1242,0x1242,0x1242,0x1242,0x1248,0x1248,0x1254,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257, +0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x125a,0x1257, +0x125a,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x1257,0x125a, +0x1257,0x1257,0x1257,0x1257,0x1254,0x1254,0x1254,0x1248,0x1248,0x1248,0x1248,0x1254,0x1254,0x124e,0x124b,0x1251, +0x1251,0x1260,0x125d,0x125d,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263, +0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263, +0x1263,0x1263,0x1263,0x1263,0x1269,0x1269,0x1269,0x1266,0x1266,0x1266,0x1263,0x1263,0x1263,0x1263,0x1266,0x1263, +0x1263,0x1263,0x1269,0x1266,0x1269,0x1266,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263, +0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263, +0x1263,0x1269,0x1266,0x1266,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263,0x1263, +0x1263,0x1263,0x1263,0x1bcf,0x19f5,0x19f5,0x19f5,0x19f5,0x19f5,0x19f5,0x19f5,0x19f8,0x19f2,0x1be1,0x1be1,0x1be1, +0x1be4,0x1bde,0x1be4,0x1bde,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1287, +0x1287,0x1287,0x126c,0x1929,0x138f,0x1290,0x138f,0x138f,0x138f,0x138f,0x138f,0x138f,0x138f,0x138f,0x138f,0x138f, +0x138f,0x1290,0x138f,0x1290,0x1275,0x1275,0x131d,0x1272,0x131d,0x131d,0x131d,0x131d,0x1272,0x1272,0x1296,0x1272, +0x1272,0x1272,0x1272,0x1272,0x1272,0x1275,0x1296,0x1296,0x1275,0x1296,0x1272,0x1275,0x1275,0x1278,0x1296,0x1272, +0x1272,0x1296,0x1275,0x1275,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x1281,0x1281, +0x1281,0x1281,0x13a4,0x1386,0x128a,0x13a4,0x13a4,0x13a4,0x13a4,0x13a4,0x13a4,0x13a4,0x13a4,0x13a4,0x13a4,0x1827, +0x1827,0x1827,0x1827,0x1827,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1530, +0x1530,0x1a76,0x1a76,0x1a76,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284,0x1284, +0x1284,0x1284,0x1284,0x1284,0x138f,0x138f,0x1290,0x138f,0x138f,0x138f,0x1290,0x138f,0x138f,0x138f,0x128a,0x128a, +0x128a,0x128a,0x128a,0x1389,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x128d,0x138c,0x138c,0x138c,0x138c, +0x138c,0x138c,0x138c,0x128d,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x140d,0x140d, +0x19d4,0x1a76,0x1a76,0x1a76,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x138c,0x128d,0x138c,0x128d, +0x128d,0x138c,0x138c,0x128d,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1, +0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1,0x12b1, +0x12b1,0x12b1,0x12b1,0x12b1,0x133b,0x1338,0x133b,0x1338,0x133b,0x1338,0x133b,0x1338,0x133b,0x1338,0x13fb,0x150f, +0x150f,0x150f,0x17a3,0x1917,0x150f,0x150f,0x16ef,0x16ef,0x16ef,0x16e9,0x16ef,0x16e9,0x191a,0x1917,0x19d1,0x19ce, +0x19d1,0x19ce,0x19d1,0x19ce,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f, +0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f,0x135f, +0x135f,0x135f,0x135f,0x135f,0x1374,0x1365,0x1374,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377, +0x1377,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377,0x1377, +0x1377,0x1377,0x1377,0x1377,0x1365,0x1365,0x1365,0x1365,0x1365,0x1365,0x1365,0x1365,0x137d,0x137d,0x137d,0x137d, +0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d, +0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x137d,0x1383,0x1383,0x1383,0x1383, +0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383, +0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x1383,0x13ad,0x13aa,0x18cc,0x18cc, +0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc, +0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x13b6,0x13b6,0x13b6,0x13b6, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6, +0x13b6,0x13b3,0x13b3,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b3,0x13b6,0x13b6,0x13b6,0x13b3,0x13b6,0x13b3,0x13b6, +0x13b3,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b9,0x13b6,0x13b6,0x13b6,0x13b6,0x13b3,0x13b6,0x13b3,0x13b3,0x13b6, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b3,0x13b3,0x13b3,0x13b3, +0x13b3,0x13b3,0x13b3,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6, +0x13b6,0x13b6,0x13b6,0x13b3,0x13b3,0x13b3,0x13b3,0x13b3,0x13b3,0x13b3,0x13b3,0x13b3,0x13b3,0x13b6,0x13b6,0x13b6, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b3,0x13b3,0x13b3,0x13b3,0x13b3,0x13b3, +0x13b3,0x13b3,0x13b3,0x13b3,0x13b3,0x13b3,0x1539,0x1539,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c, +0x153c,0x153c,0x153c,0x153c,0x153c,0x1779,0x1779,0x1779,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x153c,0x13b6, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x153c,0x1779,0x1779, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b9,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6, +0x13b6,0x13b6,0x13b6,0x13b6,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x1539,0x1539,0x153c,0x153c, +0x13b6,0x13b6,0x13b9,0x13b9,0x13b9,0x16a4,0x13b6,0x13b9,0x13b6,0x13b6,0x13b9,0x153f,0x153f,0x153c,0x153c,0x1779, +0x1779,0x1779,0x1779,0x1779,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6, +0x13b6,0x153c,0x153c,0x153c,0x16a4,0x153c,0x153c,0x153c,0x1779,0x1779,0x1779,0x177c,0x177c,0x177c,0x177c,0x177c, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x153c, +0x13b6,0x153c,0x13b9,0x13b9,0x13b6,0x13b6,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9, +0x13b9,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6, +0x13b6,0x13b6,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b9, +0x13b9,0x13b9,0x13b9,0x13b9,0x13b9,0x13b6,0x13b6,0x13b6,0x13b9,0x13b6,0x13b6,0x13b6,0x13b6,0x13b9,0x13b9,0x13b9, +0x13b6,0x13b9,0x13b9,0x13b9,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b9,0x13b6,0x13b9,0x13b6,0x13b6, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x16a4,0x13b6,0x13b6,0x13b6,0x13b6,0x153c,0x153c,0x1779, +0x1410,0x1410,0x1410,0x1410,0x1539,0x1539,0x1539,0x1539,0x1539,0x1539,0x153c,0x1779,0x1779,0x1779,0x1779,0x16fe, +0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6,0x13b6, +0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153f,0x153f,0x153c,0x153c, +0x153c,0x153c,0x1830,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c, +0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x1539,0x1539,0x1539,0x1539,0x1539,0x1539,0x153c,0x13b6, +0x13b6,0x13b6,0x13b6,0x13b6,0x149a,0x13bc,0x13bc,0x13bc,0x13bc,0x13bc,0x13bc,0x13bc,0x13bc,0x13bc,0x13bc,0x13bc, +0x13bc,0x13bc,0x13bc,0x13bc,0x13bc,0x149a,0x13bc,0x13bc,0x13bc,0x149a,0x13bc,0x149a,0x13bc,0x149a,0x13bc,0x149a, +0x13bc,0x13bc,0x13bc,0x149a,0x13bc,0x13bc,0x13bc,0x13bc,0x13bc,0x13bc,0x149a,0x149a,0x13bc,0x13bc,0x13bc,0x13bc, +0x149a,0x13bc,0x149a,0x149a,0x13bc,0x13bc,0x13bc,0x13bc,0x149a,0x13bc,0x13bc,0x13bc,0x13bc,0x13bc,0x13bc,0x13bc, +0x13bc,0x13bc,0x13bc,0x13bc,0x13bc,0x16aa,0x16aa,0x177f,0x177f,0x13bf,0x13bf,0x13bf,0x13bc,0x13bc,0x13bc,0x13bf, +0x13bf,0x13bf,0x13bf,0x13bf,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629, +0x1629,0x1629,0x1629,0x1629,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2, +0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2, +0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c5,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2, +0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2,0x13c5,0x13c5,0x13c5,0x13c2,0x13c2,0x13c2,0x13c2,0x13c2, +0x13c2,0x13c2,0x13c2,0x13c2,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8, +0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8,0x13c8, +0x13c8,0x13c8,0x13c8,0x13c8,0x17ac,0x17ac,0x17a9,0x1701,0x1416,0x1416,0x1416,0x1416,0x1416,0x1416,0x1413,0x1413, +0x1413,0x1413,0x1413,0x1413,0x1416,0x1416,0x1416,0x1416,0x1416,0x1416,0x1416,0x1416,0x1416,0x1416,0x1416,0x1416, +0x1416,0x1416,0x1416,0x1545,0x1419,0x1548,0x1419,0x1419,0x1419,0x1419,0x1419,0x1419,0x1419,0x1419,0x1419,0x1419, +0x1419,0x1548,0x1548,0x1548,0x1548,0x1548,0x1548,0x1704,0x1704,0x1b4b,0x17b2,0x17b2,0x17b2,0x17b2,0x17b2,0x17b2, +0x17b2,0x17b2,0x1a79,0x1a79,0x1422,0x1422,0x1422,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434, +0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434,0x1434, +0x1434,0x1434,0x1434,0x1434,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f, +0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f,0x144f, +0x144f,0x144f,0x144f,0x144f,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455, +0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455,0x1455, +0x1455,0x1455,0x1455,0x19da,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458, +0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458,0x1458, +0x1458,0x1458,0x1458,0x1458,0x145e,0x145e,0x146a,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470, +0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470, +0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x1470,0x146a,0x146a,0x146a,0x145e,0x145e,0x145e,0x145e,0x145e,0x145e, +0x145e,0x145e,0x145e,0x146a,0x146d,0x1470,0x1473,0x1473,0x1470,0x1476,0x1476,0x1461,0x1464,0x170a,0x170d,0x170d, +0x170d,0x154e,0x1a82,0x1a7f,0x1467,0x1467,0x1467,0x1467,0x1467,0x1467,0x1467,0x1467,0x1467,0x1467,0x154b,0x1713, +0x1716,0x1710,0x1719,0x1719,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491, +0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491,0x1491, +0x1491,0x1491,0x1491,0x1491,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb, +0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x18ff,0x18ff, +0x18ff,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x19c8,0x14eb,0x14eb, +0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x14eb,0x1863,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff, +0x18ff,0x18ff,0x18ff,0x18ff,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c, +0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x1539,0x1539,0x1539,0x1539,0x1539,0x1539,0x1539,0x1539, +0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c, +0x153f,0x153c,0x153c,0x153c,0x153c,0x16a7,0x16a7,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c, +0x182d,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c, +0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x153c,0x155d,0x155d,0x155d,0x155d, +0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d, +0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x155d,0x156f,0x156f,0x156f,0x156f, +0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f, +0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x156f,0x1575,0x1575,0x1575,0x1575, +0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575, +0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1575,0x1578,0x1578,0x1578,0x1578, +0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578, +0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x1578,0x15a2,0x15a2,0x15a2,0x15a2, +0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x15a2,0x159c,0x159c,0x159c,0x1590,0x1590,0x1590,0x159c,0x159c, +0x1590,0x159f,0x1593,0x1590,0x15a5,0x15a5,0x1599,0x15a5,0x15a5,0x1596,0x17b5,0x1bdb,0x15b7,0x15b7,0x15b7,0x15b7, +0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7, +0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15b7,0x15a8,0x15c0,0x15c0,0x15c0,0x15c0, +0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0, +0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15c0,0x15ba,0x15c3,0x15c3,0x15c3,0x15c3,0x15c6,0x15c6,0x15c6,0x15c6, +0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6, +0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15c6,0x15e1,0x15e1,0x15e1,0x15e1, +0x15e1,0x15e1,0x15e1,0x15e1,0x15d8,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1, +0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15e1,0x15ea,0x15ea,0x15ea,0x15ea, +0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea, +0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15ea,0x15fc,0x15fc,0x15fc,0x15fc, +0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15f9,0x15f9,0x15f9,0x15ed, +0x15ed,0x15ed,0x15ed,0x15ed,0x15ed,0x15ed,0x15ed,0x15f9,0x15f9,0x15ed,0x15f9,0x15f0,0x15fc,0x15fc,0x15fc,0x15fc, +0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc, +0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x15fc,0x1620,0x1620,0x1620,0x1620, +0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x1620, +0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x1620,0x161d,0x161d,0x161d,0x1629,0x1629,0x1629,0x1629, +0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629,0x1629, +0x1629,0x1629,0x162f,0x162f,0x162f,0x162c,0x162c,0x162c,0x1629,0x1629,0x1629,0x1629,0x163e,0x163e,0x163e,0x163e, +0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x1632,0x1632,0x1632,0x1632, +0x1632,0x1632,0x1632,0x1644,0x1644,0x1638,0x1635,0x1635,0x1635,0x1635,0x1635,0x1635,0x163e,0x163e,0x163e,0x163e, +0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e, +0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x163e,0x164a,0x164a,0x164a,0x164a, +0x164a,0x164a,0x164a,0x164a,0x164a,0x164a,0x164a,0x164a,0x164a,0x164a,0x164a,0x164a,0x164a,0x164a,0x164a,0x164a, +0x164a,0x164a,0x164a,0x1647,0x1647,0x1647,0x1647,0x1647,0x1647,0x1647,0x1647,0x1647,0x164d,0x164d,0x164d,0x164d, +0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d, +0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x164d,0x1671,0x1671,0x1671,0x1671, +0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671, +0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x1671,0x167a,0x167a,0x167a,0x167a, +0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a, +0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x167a,0x1692,0x1692,0x1692,0x1692, +0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x167d,0x168c,0x168c,0x167d, +0x167d,0x167d,0x167d,0x167d,0x167d,0x168c,0x167d,0x168f,0x168f,0x167d,0x168f,0x167d,0x1692,0x1692,0x1692,0x1692, +0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692, +0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x1692,0x169b,0x169b,0x169b,0x169b, +0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b, +0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x169b,0x16a1,0x16a1,0x16a1,0x16a1, +0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1, +0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x16a1,0x18ff,0x18ff,0x18ff,0x18ff, +0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x16e6,0x16e6,0x16e6,0x16e6,0x18ff,0x18ff,0x18ff,0x18ff, +0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x18ff,0x19c8,0x1707,0x1707,0x1707,0x1707, +0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707, +0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1707,0x1746,0x1746,0x1746,0x1746, +0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746, +0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x174c,0x1749, +0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746,0x1746, +0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f, +0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f,0x174f, +0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752, +0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752,0x1752, +0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764, +0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764,0x1764, +0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767, +0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767,0x1767, +0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a, +0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a, +0x176a,0x176a,0x176a,0x176d,0x176d,0x176d,0x176d,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a,0x176a, +0x176a,0x176a,0x176a,0x176a,0x176a,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176a,0x176d,0x176d, +0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d, +0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d,0x176d, +0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785, +0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785,0x1785, +0x186f,0x186f,0x186f,0x186f,0x186f,0x186f,0x186f,0x186f,0x186f,0x186f,0x186f,0x186f,0x1ae5,0x1a3a,0x1a3a,0x1a3d, +0x1788,0x1788,0x1788,0x1788,0x1788,0x1788,0x1788,0x1788,0x178b,0x1839,0x1839,0x1839,0x1839,0x1839,0x1839,0x18d5, +0x1788,0x1788,0x1788,0x1788,0x1788,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836, +0x1836,0x1836,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3, +0x1788,0x19b3,0x19b3,0x1a3a,0x1a3a,0x1a3a,0x1a3a,0x1a3a,0x1a3a,0x1a3a,0x1a3a,0x1ae2,0x1bb4,0x1a3d,0x1a3d,0x1a3d, +0x18d2,0x18d5,0x18d5,0x18d5,0x18d5,0x18d5,0x18d5,0x18d5,0x18d5,0x18d5,0x18d5,0x18d5,0x18d5,0x18d5,0x18d2,0x18d2, +0x1a79,0x1a79,0x1a79,0x1a79,0x1a79,0x1a79,0x1a79,0x1a79,0x1b4b,0x1b4e,0x1b48,0x1b48,0x1b48,0x1b48,0x1b48,0x1b48, +0x1b48,0x1b48,0x1b48,0x192c,0x17af,0x17af,0x17af,0x17af,0x17af,0x17af,0x17af,0x17af,0x17af,0x17af,0x17af,0x17af, +0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1839,0x1836,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2, +0x1839,0x18d5,0x18d5,0x1839,0x1839,0x1839,0x1839,0x1839,0x1839,0x1839,0x1836,0x17b8,0x1839,0x1839,0x1839,0x1a3a, +0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x17b8,0x1836,0x1836,0x1836,0x1836,0x1836,0x18d2,0x19b3,0x19b3,0x19b3, +0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x1836,0x18d2, +0x17cd,0x17cd,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca, +0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca,0x17ca, +0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd, +0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd,0x17cd, +0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b, +0x181b,0x181b,0x181b,0x181b,0x181b,0x1818,0x1818,0x1818,0x1803,0x1803,0x1803,0x1803,0x1803,0x1803,0x1803,0x1803, +0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b, +0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b,0x181b, +0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f, +0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f,0x183f, +0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842, +0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842,0x1842, +0x1842,0x1842,0x1842,0x1aee,0x1aee,0x1aee,0x1aee,0x1aee,0x1aee,0x1aee,0x1aee,0x1aee,0x1aee,0x1aee,0x1aee,0x1aee, +0x1899,0x1899,0x1899,0x1899,0x19ec,0x19ec,0x189c,0x189c,0x189c,0x189c,0x1884,0x1884,0x1884,0x1884,0x1884,0x1884, +0x1884,0x1884,0x1884,0x1884,0x1884,0x1884,0x1884,0x1896,0x1887,0x188a,0x188d,0x189f,0x189f,0x193e,0x1890,0x1890, +0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899, +0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899,0x1899, +0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba, +0x18ba,0x18ba,0x18ba,0x18a5,0x18ab,0x18a8,0x18a8,0x18a8,0x18a8,0x18b7,0x18bd,0x18a8,0x18a8,0x18a8,0x18a8,0x18b4, +0x18ba,0x18a8,0x18a8,0x18a8,0x18a8,0x18a8,0x18a8,0x18a8,0x18a8,0x18a8,0x18a8,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba, +0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba,0x18ba, +0x18c6,0x18c6,0x18c6,0x18c6,0x18c6,0x18c6,0x18c6,0x18c6,0x18c6,0x18c6,0x18c6,0x19ad,0x19ad,0x19ad,0x19ad,0x19ad, +0x1ad9,0x1ad9,0x1ad9,0x1ad9,0x1ad9,0x1ad9,0x1ad9,0x1ad9,0x1ad9,0x1ad9,0x1ad9,0x1ad9,0x1ad9,0x1ba5,0x1ba5,0x1ba5, +0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc, +0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc,0x18cc, +0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x19b3,0x19b3,0x19b3,0x19b3, +0x19b3,0x1a3a,0x1ae2,0x19b3,0x19b3,0x19b3,0x19b3,0x1ae5,0x1ae2,0x1bb4,0x19b3,0x1a3a,0x19b3,0x19b3,0x19b3,0x19b3, +0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x18d2,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3, +0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b3, +0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db, +0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db,0x18db, +0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de, +0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x18de,0x1bb7, +0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1, +0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1,0x18e1, +0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953, +0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953,0x1953, +0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e, +0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e,0x196e, +0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974, +0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974,0x1974, +0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f, +0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f,0x198f, +0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992, +0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992,0x1992, +0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b, +0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x199b,0x1998,0x1998,0x1998, +0x19b3,0x19b3,0x19b3,0x1ae2,0x1ae2,0x1a3a,0x1a3a,0x1a3a,0x1a3a,0x1a3a,0x1a3a,0x1ae2,0x1ae2,0x1ae2,0x1a3a,0x1a3a, +0x19b3,0x19b3,0x19b3,0x19b3,0x19b3,0x19b6,0x19b6,0x19b3,0x19b6,0x19b6,0x1a3a,0x1a3d,0x1a3a,0x1a3a,0x1a3a,0x1a3a, +0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef, +0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef,0x19ef, +0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16, +0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16,0x1a16, +0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f, +0x1a55,0x1a55,0x1a1f,0x1a55,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a1f,0x1a25,0x1a25,0x1a25, +0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31, +0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31,0x1a31, +0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4, +0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4,0x1ac4, +0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0, +0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0,0x1ad0, +0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4, +0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4,0x1af4, +0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7, +0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7,0x1af7, +0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c, +0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c,0x1b6c, +0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d, +0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d,0x1b8d, +0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f, +0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f,0x1b9f, +0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2, +0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2,0x1ba2, +0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc,0x1bfc, +0x1bfc,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9,0x1bf9, +0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b, +0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b,0x1c3b, +0,0,0,0 +}; + +static const UTrie2 propsVectorsTrie={ + propsVectorsTrie_index, + propsVectorsTrie_index+5348, + nullptr, + 5348, + 27344, + 0xa40, + 0x1564, + 0x0, + 0x0, + 0x110000, + 0x7fb0, + nullptr, 0, false, false, 0, nullptr +}; + +static const uint32_t propsVectors[7230]={ +0x67,0,0,0x67,0,0xe00000,0x67,0x80000,0x20,0x867,0,0,0xa67,0,0,0xb67, +0,0,0xd67,0,0,0xe67,0,0,0x1067,0,0,0x1167,0,0,0x1267,0, +0,0x1367,0,0,0x1467,0,0,0x1567,0,0,0x1667,0,0,0x1767,0,0, +0x1867,0,0,0x1967,0,0,0x1a67,0,0,0x1b67,0,0,0x1d67,0,0,0x1f67, +0,0,0x2067,0,0,0x2267,0,0,0x2367,0,0,0x2467,0,0,0x2567,0, +0,0x2767,0,0,0x2867,0x80000,0x20,0x2967,0,0,0x2a67,0,0x1600000,0x2b67,0,0, +0x2d67,0,0,0x3167,0x20000000,0,0x3267,0x20000000,0,0x3a67,0,0,0x3b67,0,0,0x3c67, +0,0,0x3e67,0,0,0x4067,0,0,0x4167,0,0,0x4467,0,0,0x4867,0, +0,0x4967,0,0,0x4a67,0,0,0x5067,0,0,0x5167,0,0,0x5467,0,0, +0x5567,0,0,0x5667,0x80000,0x20,0x5767,0,0,0x5867,0,0,0x5967,0,0,0x5b67, +0,0,0x5c67,0,0,0x5d67,0,0,0x6067,0x80000,0x20,0x6267,0,0,0x6367,0, +0,0x6467,0,0,0x6567,0,0,0x6f67,0,0,0x7067,0,0,0x7367,0x20000000,0, +0x7567,0,0,0x7667,0,0,0x7767,0,0,0x7867,0,0,0x7a67,0,0,0x7b67, +0,0,0x7c67,0,0,0x7e67,0,0,0x7f67,0,0,0x8167,0,0,0x8267,0, +0,0x8467,0,0,0x8567,0,0,0x8667,0,0,0x8767,0,0,0x8967,0,0, +0x8b67,0,0,0x8c67,0,0,0x8e67,0x20000000,0,0x8f67,0,0,0x9067,0,0,0x9167, +0,0,0x9267,0,0,0x9367,0,0,0x9567,0,0,0x9667,0,0,0x9767,0, +0,0x9867,0,0,0x9967,0,0,0x9a67,0,0,0x9c67,0,0,0x9f67,0,0, +0xa167,0,0,0xa367,0,0,0xa467,0,0,0xa567,0,0,0xa667,0,0,0xa767, +0,0,0xa867,0,0,0xa967,0,0,0xaa67,0,0xe00000,0xab67,0,0xe00000,0xac67,0, +0,0xad67,0,0,0xae67,0,0,0xaf67,0,0,0xb167,0,0,0xb267,0,0, +0xb467,0,0,0xb567,0,0,0xb767,0,0,0xb867,0,0,0xb967,0,0,0xba67, +0,0,0xbc67,0,0,0xbd67,0,0,0xbe67,0,0,0xbf67,0,0,0xc067,0, +0,0xc167,0,0,0xc367,0,0xe00000,0xc467,0,0xe00000,0xc667,0,0,0xc767,0,0, +0xc867,0,0,0xc967,0,0,0xca67,0,0,0xcc67,0,0xe00000,0xcf67,0,0xe00000,0xd067, +0,0xe00000,0xd367,0,0,0xd467,0,0,0xd567,0,0,0xd667,0,0,0xd867,0, +0,0xda67,0,0,0xdb67,0,0,0xdc67,0,0,0xdd67,0,0,0xde67,0,0, +0xdf67,0,0,0xe067,0,0,0xe167,0,0,0xe267,0,0,0xe367,0,0xe00000,0xe467, +0,0,0xe567,0,0,0xe667,0,0,0xe767,0,0,0xe867,0,0,0xe967,0, +0,0xea67,0,0,0xeb67,0,0,0xec67,0,0,0xed67,0,0,0xee67,0,0, +0xef67,0,0,0xf167,0,0,0xf367,0,0,0xf567,0,0,0xf667,0,0,0xf767, +0,0,0xf867,0,0,0xf967,0,0,0xfa67,0,0xe00000,0xfb67,0,0,0xfc67,0, +0,0xfd67,0,0,0xfe67,0,0,0x10167,0,0,0x10267,0,0,0x10367,0,0, +0x10467,0,0,0x10667,0,0,0x10767,0,0,0x10867,0,0,0x10967,0,0,0x10a67, +0,0,0x10b67,0,0,0x10c67,0,0,0x10d67,0,0,0x10e67,0,0,0x10f67,0, +0,0x11067,0,0,0x11367,0,0,0x11467,0,0,0x11567,0,0,0x11667,0,0, +0x11767,0,0,0x11867,0,0,0x11967,0,0xe00000,0x11a67,0,0,0x11b67,0,0,0x11c67, +0,0,0x11d67,0,0,0x11e67,0,0,0x11f67,0,0,0x12067,0,0,0x12167,0, +0,0x12267,0,0,0x12367,0,0,0x12467,0,0,0x12567,0,0,0x12667,0,0, +0x12767,0,0,0x12867,0,0,0x12967,0,0,0x12a67,0,0xe00000,0x12b67,0,0,0x12c67, +0,0,0x12d67,0,0,0x12f67,0,0,0x13067,0,0,0x13167,0,0,0x13267,0, +0,0x13367,0,0,0x13467,0,0,0x13567,0,0,0x13667,0,0,0x13767,0,0, +0x13867,0,0,0x13967,0,0,0x13a67,0,0,0x13b67,0,0,0x13c67,0,0,0x13d67, +0,0,0x13f67,0,0,0x14067,0,0,0x14167,0,0,0x14367,0,0,0x14467,0, +0,0x14567,0,0,0x14667,0,0,0x14767,0,0,0xa0067,0,0xe00000,0xa4f67,0,0xe00000, +0xa5f67,0,0xe00000,0xac567,0,0xe00000,0xad167,0,0xe00000,0xb0067,0,0xe00000,0xb1267,0,0xe00000,0xb2e67, +0,0xe00000,0x11000100,0,0x900020,0x11000100,0x40000001,0x440020,0x11000100,0x40000001,0x643020,0x11000100,0x40000001,0xa5a040,0x11000100,0x40000001, +0x116a8a0,0x11000200,0,0x900020,0x11000200,0x4000001,0xc4000b,0x11000200,0x7c00100,0x220402,0x11000200,0x24000000,0x200000,0x11000200,0x24000008,0x1710000, +0x11000200,0x40000001,0x1d3b020,0x11000219,0x7c00100,0x220401,0x11000219,0x7c00100,0x250401,0x11000319,0x7c00100,0x220401,0x11000319,0x7c00100,0x220402,0x11000319, +0x7c00100,0x250400,0x11000319,0x7c00100,0x250401,0x11000419,0x7c00100,0x220400,0x11000419,0x7c00100,0x220401,0x11000419,0x7c00100,0x220402,0x11000419,0x7c00100, +0x230400,0x11000419,0x7c00100,0x250400,0x11000419,0x7c00100,0x250401,0x11000419,0x7c00100,0x250402,0x11000519,0x7c00100,0x220400,0x11000519,0x7c00100,0x230400, +0x11000600,0x4000400,0x200002,0x11000600,0x4000400,0x200400,0x11000600,0x7c00500,0x220400,0x11000600,0x7c00500,0x230400,0x11000600,0x7c00500,0x530400,0x11000600, +0x7c00d00,0x230400,0x11000619,0x7c00500,0x22040f,0x11000800,0x4000010,0x1001401,0x11000800,0x4000400,0x200001,0x11000800,0x6800010,0x201001,0x11000800,0x7c00500, +0x230401,0x11000807,0x7c00100,0x220400,0x11000807,0x7c00100,0x250400,0x1100080e,0x4000400,0x200000,0x1100080e,0x4000400,0x200002,0x1100080e,0x7000500,0x220402, +0x1100080e,0x7c00100,0x220400,0x1100080e,0x7c00100,0x220401,0x1100080e,0x7c00100,0x220402,0x1100080e,0x7c00100,0x250400,0x1100080e,0x7c00100,0x250401,0x1100080e, +0x7c00120,0x220402,0x1100080e,0x7c00120,0x250402,0x11000908,0x4000000,0x200000,0x11000908,0x7c00100,0x220400,0x11000908,0x7c00100,0x220401,0x11000908,0x7c00100, +0x250400,0x11000908,0x7c00100,0x250401,0x11000a03,0x4000000,0x200400,0x11000a03,0x4000000,0x201000,0x11000a03,0x4000000,0x270000,0x11000a03,0x7c00100,0x220400, +0x11000a03,0x7c00100,0x220402,0x11000a03,0x7c00100,0x250400,0x11000a03,0x7c00500,0x230400,0x11000a03,0xc000010,0x1049400,0x11000b13,0x2802500,0x962460,0x11000b13, +0x4000000,0x200000,0x11000b13,0x4000000,0x201000,0x11000b13,0x4000000,0x230400,0x11000b13,0x4000002,0x400000,0x11000b13,0x4000010,0x200000,0x11000b13,0x7c00100, +0x2633800,0x11000c00,0x80000000,0x218960,0x11000c02,0x2802100,0x962460,0x11000c02,0x2802400,0x962460,0x11000c02,0x4000000,0x200000,0x11000c02,0x4000000,0x1329400, +0x11000c02,0x4000000,0x1329800,0x11000c02,0x4000000,0x1500000,0x11000c02,0x6800000,0x1329800,0x11000c02,0x7c00100,0x230400,0x11000c02,0x7c00100,0x230401,0x11000c02, +0x7c00100,0x230402,0x11000c02,0x7c00500,0x230400,0x11000c02,0x7d00100,0x230400,0x11000f01,0x2802400,0x962460,0x11000f0a,0x2802100,0x962460,0x11000f0a,0x2802400, +0x962460,0x11000f0a,0x2806400,0x962460,0x11000f0a,0x4000000,0x200000,0x11000f0a,0x6800100,0x962540,0x11000f0a,0x7c00100,0x230400,0x11000f0a,0x7c00100,0x230401, +0x11001004,0x2802100,0x962460,0x11001004,0x2802400,0x962460,0x11001004,0x2806400,0x962460,0x11001004,0x4000000,0x200000,0x11001004,0x4000000,0x1500000,0x11001004, +0x6800100,0x962540,0x11001004,0x6800100,0x962541,0x11001004,0x7c00100,0x230400,0x11001004,0x7c00100,0x230401,0x11001110,0x2802100,0x962460,0x11001110,0x2802400, +0x962460,0x11001110,0x2806400,0x962460,0x11001110,0x6800100,0x962540,0x11001110,0x7c00100,0x230400,0x11001110,0x7c00100,0x230401,0x1100120f,0x2802100,0x962460, +0x1100120f,0x2802400,0x962460,0x1100120f,0x2806400,0x962460,0x1100120f,0x6800100,0x962540,0x1100120f,0x7c00100,0x230400,0x1100131f,0x2802100,0x962460,0x1100131f, +0x2802400,0x962460,0x1100131f,0x2806400,0x962460,0x1100131f,0x4000000,0x200000,0x1100131f,0x6800000,0x1329800,0x1100131f,0x6800100,0x962540,0x1100131f,0x6800100, +0x962541,0x1100131f,0x7c00100,0x230400,0x1100131f,0x7c00100,0x230401,0x11001423,0x2802100,0x962460,0x11001423,0x2806400,0x962460,0x11001423,0x6800100,0x962540, +0x11001423,0x6800100,0x962541,0x11001423,0x7c00100,0x230400,0x11001423,0x7c00100,0x230401,0x11001524,0x2802100,0x962460,0x11001524,0x2802100,0x962461,0x11001524, +0x2806400,0x962460,0x11001524,0x6800000,0x1329800,0x11001524,0x6800100,0x962540,0x11001524,0x7c00100,0x230400,0x11001615,0x2802100,0x962460,0x11001615,0x2806400, +0x962460,0x11001615,0x6800100,0x962540,0x11001615,0x6800100,0x962541,0x11001615,0x7c00100,0x230400,0x1100171a,0x2802100,0x962460,0x1100171a,0x2806400,0x962460, +0x1100171a,0x6800000,0x1329800,0x1100171a,0x6800100,0x962540,0x1100171a,0x6800100,0x962541,0x1100171a,0x7c00100,0x230400,0x11001900,0x4000000,0x1600000,0x11001926, +0x2802100,0x1862460,0x11001926,0x2802400,0x1862460,0x11001926,0x2806100,0x1862460,0x11001926,0x4000000,0x200000,0x11001926,0x4000010,0x400000,0x11001926,0x6800000, +0x1329800,0x11001926,0x7800100,0x1830142,0x11001926,0x7c00100,0x1830000,0x11001926,0x7c00900,0x1830000,0x11001926,0x7e00100,0x1830000,0x11001a18,0x2802100,0x1862460, +0x11001a18,0x2802400,0x1862460,0x11001a18,0x6800000,0x1329800,0x11001a18,0x7800100,0x1830142,0x11001a18,0x7c00100,0x1830000,0x11001a18,0x7c00100,0x1830002,0x11001a18, +0x7c00900,0x1830000,0x11001a18,0x7e00100,0x1830000,0x11001d0c,0x7c00100,0x230400,0x11001d0c,0x7c00100,0x250400,0x11001e12,0x7c00100,0x2230500,0x11001e12,0x7c00100, +0x2330520,0x11001e12,0x7c80100,0x2330520,0x11002619,0x7c00100,0x220401,0x11002619,0x7c00100,0x220402,0x11002619,0x7c00100,0x250401,0x1100270e,0x4000400,0x200001, +0x1100270e,0x4000400,0x200002,0x1100270e,0x4000400,0x500001,0x1100270e,0x7c00100,0x220401,0x1100270e,0x7c00100,0x250401,0x11002800,0x80000,0x918820,0x11002800, +0x80000,0x1c18020,0x11002800,0x180000,0x918820,0x11002800,0x4000001,0x445801,0x11002800,0x4000001,0x445802,0x11002800,0x4000001,0xc4000b,0x11002800,0x6800000, +0x201c00,0x11002800,0x6800020,0x201c00,0x11002800,0x24000000,0x200000,0x11002800,0x24000000,0x200002,0x11002800,0x24000000,0x810000,0x11002800,0x24000000,0x1410000, +0x11002800,0x24000000,0x1500000,0x11002800,0x24000000,0x1500002,0x11002800,0x24000002,0x400000,0x11002800,0x24000006,0xc0000b,0x11002800,0x24000008,0x1410000,0x11002800, +0x24000008,0x1710000,0x11002800,0x24000020,0x1001400,0x11002800,0x24000020,0x1500002,0x11002800,0x2c000010,0x1248000,0x11002800,0x2c000010,0x1248002,0x11002800,0x40000001, +0x63b020,0x11002800,0x40080000,0x918820,0x11002801,0x80000,0x2a65620,0x11002801,0x82000,0x962460,0x11002900,0x4000000,0x20000e,0x11002900,0x4000000,0x20000f, +0x11002900,0x4000020,0x20000e,0x11002900,0x4000020,0x20000f,0x11002900,0x4000020,0x81000e,0x11002900,0x4000020,0x81000f,0x11002900,0x4000020,0x141000e,0x11002900, +0x4000020,0x141000f,0x11002900,0x4000022,0x20000e,0x11002900,0x4000022,0x20000f,0x11002a00,0x4000000,0x1500000,0x11002a00,0x4000000,0x1600000,0x11002a00,0x4000000, +0x1600002,0x11002b01,0x2000,0x962460,0x11002b01,0x2802020,0x962460,0x11002c00,0x4000000,0x200000,0x11002c00,0x4000000,0x200002,0x11002c00,0x4000000,0x20000f, +0x11002c00,0x4000020,0x200000,0x11002c00,0x7c00000,0x200000,0x11002c00,0x7c00020,0x200000,0x11002c00,0x7c00120,0x220405,0x11002c00,0x7c00120,0x230402,0x11002c00, +0x7c00120,0x250402,0x11002c00,0x7c00120,0x250405,0x11002c19,0x7c00100,0x250400,0x11002c19,0x7c00100,0x250401,0x11002d00,0x4000000,0x100006,0x11002d00,0x4000000, +0x200006,0x11002d19,0x7c00100,0x220402,0x11002d19,0x7c00100,0x230400,0x11002d19,0x7c00100,0x250402,0x11002e00,0x24000000,0x200000,0x11002e00,0x24000020,0x200000, +0x11002e00,0x24000020,0x200001,0x11002f00,0x24000020,0x200000,0x11002f00,0x24000020,0x200001,0x11002f00,0x24000020,0x200002,0x11002f00,0x24000020,0xf00000,0x11002f00, +0x24000020,0x1600000,0x11002f00,0x24000022,0x1600000,0x11003000,0x24000000,0x200000,0x11003000,0x24000020,0x200000,0x11003000,0x24000020,0x810000,0x11003000,0x24000020, +0x1410000,0x11003100,0x24000000,0x200000,0x11003200,0x24000000,0x200000,0x11003300,0x4000000,0x100003,0x11003400,0x24000000,0x100000,0x11003400,0x24000000,0x200000, +0x11003500,0x24000000,0x200000,0x11003600,0x24000000,0x200000,0x11003600,0x24000020,0x200000,0x11003700,0x24000000,0x200000,0x11003700,0x24000000,0xe00000,0x11003700, +0x24000000,0x2800000,0x11003700,0x24000020,0x200000,0x11003800,0x4000000,0x100000,0x11003800,0x24000000,0x200000,0x11003800,0x24000000,0xb00000,0x11003800,0x24000000, +0xe00000,0x11003800,0x24000000,0x1710000,0x11003800,0x24000000,0x2800000,0x11005003,0x7c00100,0x220402,0x11005013,0x2802500,0x962460,0x11005013,0x4000020,0x200005, +0x11005013,0x7c00100,0x2633801,0x11005013,0x7c00100,0x2633802,0x11005013,0x7c00100,0x2633805,0x11005019,0x7c00100,0x220402,0x11005102,0x7000100,0x230408,0x11005102, +0x7c00100,0x230404,0x11005102,0x7c00100,0x230407,0x11005102,0x7c00100,0x230408,0x11005102,0x7c00100,0x230409,0x11005201,0x2802400,0x962460,0x11005500,0x80000, +0x1e18820,0x11005502,0x7000100,0x230408,0x11005502,0x7c00100,0x230404,0x11005502,0x7c00100,0x230407,0x11005502,0x7c00100,0x230408,0x11005502,0x7c00100,0x230409, +0x11005667,0x1000,0,0x11020200,0x80004,0x418820,0x11020200,0x4000000,0x100006,0x11020200,0x4000000,0x10000f,0x11020200,0x4000400,0x100002,0x11020200, +0x4000400,0x500002,0x11020200,0x6800c00,0x101000,0x11020200,0x24000000,0x100000,0x11020200,0x24000000,0x200000,0x11020200,0x24000000,0x1400000,0x11020200,0x24000000, +0x1500000,0x11020200,0x24000000,0x1600000,0x11020200,0x24000020,0x100000,0x11020200,0x24000020,0x1600000,0x11020219,0x7c00100,0x12040f,0x11020219,0x7c00100,0x220400, +0x11020219,0x7c00100,0x220401,0x11020219,0x7c00100,0x250400,0x11020319,0x7c00100,0x220400,0x11020319,0x7c00100,0x220401,0x11020319,0x7c00100,0x220402,0x11020319, +0x7c00100,0x250400,0x11020319,0x7c00100,0x250402,0x11020319,0x7d00100,0x220402,0x11020419,0x7c00100,0x220401,0x11020519,0x7c00100,0x220400,0x11020600,0x4000400, +0x100002,0x11020600,0x4000400,0x200400,0x11020600,0x7c00500,0x130400,0x11020600,0x7c00d00,0x130400,0x11020701,0x2802400,0x962460,0x11020701,0x2802400,0x962461, +0x11020701,0x2802400,0xc62460,0x1102080e,0x7c00100,0x220400,0x1102080e,0x7c00100,0x250400,0x11020908,0x7c00100,0x220400,0x11020908,0x7c00100,0x220401,0x11020908, +0x7c00100,0x250400,0x11020908,0x7c00100,0x250401,0x11022800,0x24000000,0x100000,0x11022800,0x24000000,0x200000,0x11022800,0x24000000,0x200002,0x11022800,0x24000000, +0x401000,0x11022800,0x24000000,0xf00002,0x11022800,0x24000000,0xf0ac02,0x11022800,0x24000000,0x1500000,0x11022800,0x24000002,0x100000,0x11022800,0x24000002,0x370000, +0x11022800,0x24000002,0x470000,0x11022800,0x24000006,0x400000,0x11022800,0x24000008,0x1710000,0x11022800,0x24000008,0x1712c00,0x11022800,0x24000020,0x100000,0x11022800, +0x24000020,0x1500000,0x11022800,0x24000020,0x1500002,0x11022900,0x4000000,0x10000e,0x11022900,0x4000000,0x10000f,0x11022919,0x7c00100,0x12040f,0x11022c00,0x4000000, +0x100002,0x11022c00,0x4000000,0x10000f,0x11022c00,0x4000000,0x1500002,0x11022c00,0x4000000,0x1600002,0x11022c00,0x7c00120,0x120405,0x11022c0e,0x7c00100,0x250401, +0x11022c19,0x7c00100,0x150401,0x11022d00,0x4000000,0x100006,0x11022d00,0x4000000,0x200006,0x11022d19,0x7c00100,0x120402,0x11022d19,0x7c00100,0x150402,0x11022e00, +0x24000000,0x200000,0x11022e00,0x24000020,0x100000,0x11022f00,0x24000020,0x100000,0x11022f00,0x24000020,0x100001,0x11022f00,0x24000020,0x100002,0x11023000,0x24000000, +0x100000,0x11023300,0x4000000,0x100002,0x11023300,0x4000000,0x100003,0x11023300,0x4000100,0x120403,0x11023300,0x4000100,0x150403,0x11023400,0x24000000,0x100000, +0x11023500,0x24000000,0x100000,0x11023600,0x24000000,0x100000,0x11023600,0x24000020,0x100000,0x11023700,0x24000000,0x100000,0x11023700,0x24000000,0xe00000,0x11023700, +0x24000020,0x100000,0x11023800,0x4000000,0x100000,0x11023800,0x24000000,0x200000,0x11024e67,0,0,0x11025600,0x4000000,0x100000,0x11042a00,0x4000000, +0x1600000,0x11045700,0x4000000,0x20000a,0x11045700,0x4000020,0x20000a,0x11045712,0x7c00100,0xe3040a,0x11045712,0x7c80100,0xe3040a,0x11045716,0x7c00100,0xe30c0a, +0x11045716,0x7c00100,0x2530c0a,0x11063d00,0x4000001,0x445811,0x11065700,0x4000000,0x810011,0x11065700,0x4000000,0xe00011,0x11065700,0x4000000,0x1410011,0x11065700, +0x4000000,0x1500011,0x11065700,0x4000000,0x1600011,0x11065700,0x4000006,0xe70011,0x11065700,0x4000008,0xe00011,0x11065700,0x4000008,0xe02c11,0x11065700,0x4000010, +0x871411,0x11065700,0x4000010,0x1201411,0x11065700,0x4000010,0x1271011,0x11065700,0x4000020,0xe00011,0x11065700,0x4000400,0xe00011,0x11065700,0x4000420,0xe00011, +0x11065700,0x6800000,0xe01c11,0x11065700,0x6800040,0xe29811,0x11065700,0xc000010,0x80ac11,0x11065700,0xc000010,0xb48011,0x11065719,0x7c00100,0xe20411,0x11065719, +0x7c00100,0xe50411,0x11065719,0x7c00140,0xe20411,0x11065719,0x7c00140,0xe50411,0x11080100,0x6800000,0x201c00,0x11080100,0x68000c0,0x1329800,0x11080100,0x24000000, +0x200000,0x11080100,0x24000000,0x810000,0x11080100,0x24000000,0x1410000,0x11080100,0x24000000,0x1500000,0x11080100,0x24000000,0x1600000,0x11080100,0x24000000,0x1b00000, +0x11080100,0x24000000,0x2410000,0x11080100,0x24000006,0xd70000,0x11080100,0x24000008,0x1713c00,0x11080100,0x24000008,0x1714000,0x11080100,0x24000010,0x1001400,0x11080100, +0x24000010,0x1071000,0x11080100,0x24000010,0x1071400,0x11080100,0x24000020,0x200000,0x11080100,0x24000020,0x400000,0x11080100,0x24000020,0x1600000,0x11080100,0x24000400, +0x200000,0x11080100,0x24000420,0x200000,0x11080100,0x2c000010,0xb48000,0x11080100,0x2c000010,0x100ac00,0x11080100,0x44000001,0x1a45800,0x11080119,0x7c00100,0x220400, +0x11080119,0x7c00100,0x250400,0x11080119,0x7c001c0,0x220400,0x11080119,0x7c001c0,0x250400,0x11080200,0x4000400,0x200002,0x11080200,0x24000000,0x200000,0x11080200, +0x24000000,0x1500000,0x11080200,0x24000000,0x1600000,0x11080200,0x24000020,0x200000,0x110a1e12,0x7c00100,0x2130480,0x110a1e12,0x7c80100,0x2130480,0x110a3000,0x24000000, +0xe00000,0x110a3000,0x24100000,0x810001,0x110a3000,0x24100000,0x1410001,0x110a3700,0x24000000,0x200000,0x110a3d00,0x4000000,0xe00000,0x110a3d00,0x4000000,0xe00002, +0x110a3d00,0x24000000,0xe00000,0x110a3d11,0x7c00300,0xe30000,0x110a3d11,0x7c00900,0x1230400,0x110a3d12,0x2802400,0x962460,0x110a3e14,0x7c00100,0xe30000,0x110a3e14, +0x7c00100,0xe30001,0x110a3e14,0x7c00100,0x2530000,0x110a3e14,0x7c00900,0x1230000,0x110a3e14,0x7c00900,0x1230001,0x110a3f16,0x7c00100,0xe30c00,0x110a3f16,0x7c00100, +0xe30c01,0x110a3f16,0x7c00100,0x2530c00,0x110a3f16,0x7c00900,0x1230c00,0x110a3f16,0x7c00900,0x1230c01,0x110a4005,0x7c00100,0xe30400,0x110a4112,0x7c00100,0xe30402, +0x110a4112,0x7c80100,0xe30402,0x110a4400,0x4000000,0xe00000,0x110a4412,0x4000000,0xe00002,0x110a4412,0x4000000,0xe00003,0x110a4416,0x4000000,0xe00c03,0x110a4500, +0x4000000,0xe0000d,0x110a4516,0x4000000,0xe00c0d,0x110a4711,0x7c40300,0xe30000,0x110a4f11,0x7c00300,0xe30001,0x110a4f11,0x7c40300,0xe30000,0x110a5300,0x4000000, +0x810010,0x110a5300,0x4000000,0xe00002,0x110a5300,0x4000000,0xe00010,0x110a5300,0x4000000,0x1410010,0x110a5300,0x4000002,0xe70010,0x110a5300,0x4000008,0x810010, +0x110a5300,0x4000008,0x1410010,0x110a5300,0x6800000,0xe01c02,0x110a5300,0x6800000,0xe01c10,0x110a5400,0x4000000,0x81000c,0x110a5400,0x4000000,0xe0000c,0x110a5400, +0x4000000,0x141000c,0x110a5400,0x4000000,0x150000c,0x110a5400,0x4000000,0x160000c,0x110a5400,0x4000002,0xe7000c,0x110a5400,0x4000010,0x87140c,0x110a5400,0x4000010, +0xe7000c,0x110a5400,0x4000010,0x120140c,0x110a5400,0x4000010,0x127100c,0x110a5400,0x4000020,0xe0000c,0x110a5400,0x4000026,0xe7000c,0x110a5400,0xc000010,0x80ac0c, +0x110a5400,0xc000010,0xb4800c,0x11400c0c,0x4000010,0xb00000,0x11400c0c,0x4000010,0x1071400,0x11400c17,0xc000010,0xb48000,0x11400c1e,0x7c00900,0x230400,0x11400f4b, +0xc000010,0x448000,0x11400f5f,0xc000010,0x448000,0x11401d94,0x4000000,0x200000,0x11403dca,0x4000000,0xe00000,0x114457bf,0x4000004,0x120000a,0x114457bf,0x4000008, +0x81000a,0x114457bf,0x4000008,0x141000a,0x114457bf,0x4000010,0x87000a,0x114457bf,0xc000010,0x84800a,0x114457c8,0x3802500,0x126246a,0x114457c8,0x7c00d00,0x2530c0a, +0x114a3dbf,0x24000000,0x810000,0x114a3dbf,0x24000000,0x1410000,0x114a3dbf,0x24000008,0x810000,0x114a3dbf,0x24000008,0x1410000,0x114a3dbf,0x24000010,0x870000,0x114a3dbf, +0x2c000010,0x848000,0x114a3dc5,0x4000000,0xe00000,0x114a3dc5,0x24000000,0xe00000,0x114a3dc5,0x24000002,0xe00000,0x114a3dc5,0x24000002,0x1200000,0x114a3dc5,0x24000008, +0x810000,0x114a3dc5,0x24000008,0x1410000,0x114a3dc8,0x7c00900,0x930c00,0x114a3dc8,0x7c00900,0xe30c00,0x114a3dca,0x7c00300,0xe30000,0x114a3ec8,0x7000400,0x1200c02, +0x114a3fbf,0x4000004,0x1200000,0x114a3fc8,0x7c00d00,0x2530c00,0x114a42ca,0x4000000,0xe00000,0x114a42ca,0x4000000,0xe0000f,0x114a44ca,0x4000000,0xe00002,0x114a44ca, +0x4000000,0xe00003,0x114a45ca,0x4000000,0xe00002,0x114a45ca,0x4000000,0xe0000d,0x11505103,0x24000000,0x810000,0x11505103,0x24000000,0x1410000,0x1180090a,0x2802400, +0x962460,0x11800c27,0x2802100,0x962460,0x11800c27,0x2802500,0x962460,0x11800f32,0x2802400,0x962460,0x11800f3f,0x2802400,0x962460,0x11820700,0x2802400,0x962460, +0x11820700,0x2802500,0x962460,0x118a3dcb,0x2802400,0x962460,0x118a3ec8,0x2802400,0x962460,0x11c00904,0x2802400,0x962460,0x11c00908,0x2802400,0x962460,0x11c00c2c, +0x6800000,0x1329800,0x11c00c30,0xc000010,0xb48000,0x11c00f78,0x6800000,0x1329800,0x11c0107d,0x6800000,0x1329800,0x11c01181,0x6800000,0x1329800,0x11c01285,0x6800000, +0x1329800,0x11c01489,0x4000000,0x200000,0x11c01489,0x6800000,0x1329800,0x11c0168d,0x6800000,0x1329800,0x11d05107,0x7c00100,0x230408,0x20000067,0x1000,0, +0x20000b13,0x2802400,0x962460,0x20000b13,0x2802500,0x962460,0x20001b27,0x2802100,0x962460,0x20001b27,0x2802100,0x962461,0x20001b27,0x2802400,0x962460,0x20001b27, +0x2802500,0x962460,0x20001b27,0x2806400,0x962460,0x20001b27,0x2902100,0x962462,0x20001b27,0x4000000,0x200000,0x20001b27,0x4000000,0x400000,0x20001b27,0x4000000, +0x500000,0x20001b27,0x4000000,0x810000,0x20001b27,0x4000000,0xb00000,0x20001b27,0x4000000,0xc0000b,0x20001b27,0x4000000,0x1410000,0x20001b27,0x4000010,0xb00000, +0x20001b27,0x4000010,0xc00000,0x20001b27,0x6800000,0x1329800,0x20001b27,0x6800100,0x462540,0x20001b27,0x6800400,0x962540,0x20001b27,0x7c00100,0x230400,0x20001b27, +0x7c00100,0x230401,0x20002619,0x7c00100,0x220401,0x20002a00,0x4000000,0x1600000,0x20004b67,0,0x1900000,0x20004c67,0,0x1900000,0x20004d67,0, +0x1900000,0x20006d67,0x1000,0,0x20006e67,0x1000,0,0x20026d67,0,0,0x20026e67,0,0,0x200a4a12,0x7c00100,0x1f304c1, +0x200a4a12,0x7c00100,0x20304e1,0x21005600,0x4000000,0x700000,0x21022a00,0x4000000,0x1600000,0x30000419,0x7c00100,0x220400,0x30000419,0x7c00100,0x220401,0x30000419, +0x7c00100,0x250400,0x30000419,0x7c00100,0x250401,0x30000519,0x7c00100,0x220400,0x30000600,0x4000400,0x200400,0x30000600,0x7c00500,0x230400,0x30000605,0x4000400, +0x200400,0x3000080e,0x7c00100,0x220400,0x30000908,0x2000,0x962460,0x30000908,0x7c00100,0x220400,0x30000908,0x7c00100,0x220401,0x30000908,0x7c00100,0x250400, +0x30000908,0x7c00100,0x250401,0x30000a03,0x4000006,0x400400,0x30000c02,0x4000000,0x200000,0x30000c02,0x7c00100,0x230400,0x30000d22,0x2802100,0x962460,0x30000d22, +0x2802400,0x962460,0x30000d22,0x2802500,0x962460,0x30000d22,0x4000000,0x200000,0x30000d22,0x4000010,0x200000,0x30000d22,0x7c00100,0x230400,0x30000d22,0xc000010, +0x248000,0x30000d22,0x80000000,0x218960,0x30000e25,0x2802500,0x962460,0x30000e25,0x7c00100,0x230400,0x30001821,0x2802100,0x962460,0x30001821,0x2806400,0x962460, +0x30001821,0x4000000,0x200000,0x30001821,0x6800100,0x962540,0x30001821,0x6800100,0x962541,0x30001821,0x7c00100,0x230400,0x30001b27,0x2802100,0x962460,0x30001b27, +0x2802400,0x962460,0x30001b27,0x4000000,0x200000,0x30001b27,0x4000000,0x400000,0x30001b27,0x7c00100,0x230400,0x30001c1c,0x2802100,0x1862460,0x30001c1c,0x2802400, +0x1862460,0x30001c1c,0x2806400,0x1862460,0x30001c1c,0x4000000,0x200000,0x30001c1c,0x6800100,0x1862400,0x30001c1c,0x6800100,0x1862540,0x30001c1c,0x7c00100,0x1830000, +0x30001c1c,0x7c00100,0x1830001,0x30001c1c,0xc000010,0x448000,0x30001f0b,0x4000000,0x200000,0x30001f0b,0x4000010,0x200000,0x30001f0b,0x4000010,0x400000,0x30001f0b, +0x6800000,0x200000,0x30001f0b,0x7c00100,0x230400,0x30001f0b,0xc000010,0x248000,0x30002006,0x7c00100,0x250400,0x30002128,0x4000000,0x200000,0x30002128,0x7c00100, +0x230400,0x30002128,0xc000010,0x248000,0x3000221d,0x4000000,0x810000,0x3000221d,0x4000000,0x1410000,0x3000221d,0x4000001,0x445800,0x3000221d,0x7c00100,0x230400, +0x30002300,0x4000010,0x400000,0x30002320,0x7c00100,0x230400,0x30002417,0x2802100,0x1862460,0x30002417,0x2802400,0x1862460,0x30002417,0x2806400,0x1862460,0x30002417, +0x2882000,0x1862460,0x30002417,0x4000000,0x200000,0x30002417,0x4000000,0x400000,0x30002417,0x4000000,0x1600000,0x30002417,0x4000010,0x400000,0x30002417,0x4000010, +0x1200000,0x30002417,0x6800000,0x1329800,0x30002417,0x6800100,0x1862540,0x30002417,0x7c00100,0x1830000,0x30002417,0x7d00100,0x1830000,0x3000251b,0x80000,0xc18820, +0x3000251b,0x2802100,0x962460,0x3000251b,0x3c02100,0x962460,0x3000251b,0x4000000,0x200000,0x3000251b,0x4000006,0x500000,0x3000251b,0x4000010,0x400000,0x3000251b, +0x4000010,0xb70000,0x3000251b,0x4000800,0x200000,0x3000251b,0x6800000,0x1329800,0x3000251b,0x7c00100,0x230400,0x3000251b,0x7c00900,0x230400,0x3000251b,0xc000010, +0xb48000,0x3000251b,0x12882000,0x962460,0x30002800,0x24000000,0x200000,0x30002800,0x2c000010,0x1248002,0x30002a00,0x4000000,0x1600000,0x30002b01,0x2000,0x962460, +0x30002c00,0x4000000,0x200000,0x30002c00,0x7c00100,0x220405,0x30002d19,0x7c00100,0x250400,0x30002e00,0x24000000,0x200000,0x30003000,0x24000000,0x200000,0x30003100, +0x24000000,0x200000,0x30003600,0x24000000,0x200000,0x30003700,0x24000000,0x200000,0x3000392e,0x24000000,0x200000,0x30005013,0x7c00100,0x2633801,0x30005600,0, +0x918820,0x30020600,0x4000400,0x500400,0x30020701,0x2802400,0x962460,0x30020701,0x2802400,0xc62460,0x300a3a11,0x4020000,0xe00000,0x300a3a11,0x4020000,0xe00002, +0x300a3b11,0x4020000,0xe00002,0x300a3c00,0x4008000,0xe00000,0x300a3c00,0x4010000,0xe00000,0x300a3d11,0x7c00300,0xe30002,0x300a4305,0x7c00100,0xe30400,0x300a4611, +0x7c40300,0xe30000,0x300a4829,0x7c00100,0xe30400,0x300a4829,0x7c00900,0x1230400,0x300a4929,0x4000000,0xe00000,0x3040259a,0x4000010,0x400000,0x3040259a,0x4000010, +0xb70000,0x3040259a,0xc000010,0xb48000,0x304028ba,0x4000001,0xc41c0b,0x304a3dca,0x4000000,0xe00000,0x30800c27,0x2802100,0x962460,0x30c01c92,0x6800000,0x1329800, +0x3100080e,0x7c00120,0x220402,0x3100080e,0x7c00120,0x250402,0x31005167,0x1000,0,0x3100581e,0x4000000,0x200000,0x3100581e,0x7c00100,0x230400,0x3100590d, +0x7c00100,0x230400,0x31005a09,0x7c00100,0x220400,0x31005a09,0x7c00100,0x250400,0x31005b00,0x4000000,0x200000,0x31005c00,0x80000,0x918820,0x31005c00,0x2802000, +0x962460,0x31005c00,0x2802400,0x962460,0x31005c00,0x4000000,0x200000,0x31005c00,0x4000000,0x200001,0x31005c00,0x6800000,0x962540,0x31005c00,0x6800400,0x962540, +0x31005c01,0x2802400,0x962460,0x31005d00,0x4000020,0x200005,0x31005d00,0x6800020,0x1329805,0x31005d00,0x7c00120,0x220405,0x31005d00,0x7c00120,0x250405,0x31006000, +0x82000,0x962460,0x31006000,0x180000,0x918820,0x310a5e11,0x7c40300,0xe30000,0x310a5f11,0x7c00300,0xe30001,0x32000419,0x7c00100,0x250400,0x3200080e,0x4000020, +0x200000,0x3200080e,0x7c00100,0x220400,0x3200080e,0x7c00100,0x250400,0x32000908,0x7c00100,0x220400,0x32000908,0x7c00100,0x250400,0x32000c02,0x7c00100,0x230400, +0x32000e25,0x7c00100,0x230400,0x32001d0c,0x7c00100,0x230400,0x32002800,0x80000,0x1e18820,0x32002800,0x80020,0x218820,0x32002800,0x4000001,0x445802,0x32002800, +0x24000000,0x200000,0x32002800,0x24000000,0x1500002,0x32002800,0x24000020,0x200000,0x32002800,0x2c000010,0x1248002,0x32002919,0x7c00100,0x22040f,0x32002a00,0x4000000, +0x1600000,0x32002b01,0x2000,0x962460,0x32002b01,0x2802000,0x962460,0x32002b01,0x2802020,0x962460,0x32002c00,0x4000000,0x200000,0x32002c00,0x4000020,0x200000, +0x32002c00,0x4000020,0x200005,0x32002c00,0x7c00120,0x220405,0x32002c00,0x7c00120,0x250405,0x32002e00,0x24000020,0x200000,0x32002f00,0x24000020,0x200000,0x32003000, +0x24000000,0x200000,0x32003000,0x24000020,0x200000,0x32003500,0x24000000,0x200000,0x32003600,0x24000020,0x200000,0x32003700,0x24000000,0x100000,0x32003700,0x24000000, +0x200000,0x32003800,0x24000000,0x810000,0x32003800,0x24000000,0x1410000,0x32005102,0x4000000,0x1500008,0x32005502,0x7c00100,0x230400,0x32006108,0x7c00100,0x220400, +0x32006108,0x7c00100,0x250400,0x3200622a,0x2802100,0x962460,0x3200622a,0x2806400,0x962460,0x3200622a,0x7c00100,0x230400,0x3200632b,0x2802100,0x962460,0x3200632b, +0x6804000,0x962540,0x3200632b,0x7c00100,0x230400,0x3200642c,0x2802100,0x962460,0x3200642c,0x7c00100,0x230400,0x3200652d,0x2802100,0x962460,0x3200652d,0x7c00100, +0x230400,0x32006600,0x24000020,0x200000,0x32006700,0x24000020,0x200000,0x32006800,0x24000020,0x200000,0x32006900,0x24000020,0x200000,0x32006900,0x24000020,0x810000, +0x32006900,0x24000020,0x1410000,0x32006a00,0x24000020,0x200000,0x32006a00,0x24000020,0x200001,0x32006a00,0x24000020,0x200002,0x32020701,0x2882000,0xc62460,0x32023300, +0x4000000,0x100000,0x32026c01,0x12882000,0x962460,0x32065700,0x4000000,0x810011,0x32065700,0x4000000,0x1410011,0x32086600,0x24000020,0x810000,0x32086600,0x24000020, +0x1410000,0x32086900,0x24000020,0x810000,0x32086900,0x24000020,0x1410000,0x320a3600,0x24000020,0x200000,0x320a3d11,0x7c00100,0x1230400,0x320a3e14,0x7c00100,0xe30010, +0x320a3e14,0x7c00100,0x2530000,0x320a3f16,0x7c00100,0xe30c10,0x320a4400,0x4000000,0xe00003,0x320a4929,0x4000000,0xe00000,0x320a4f11,0x7c00300,0xe30001,0x320a6b16, +0x7c00100,0x2530c00,0x32406396,0xc000010,0x448000,0x324a3dcd,0x4000000,0xe00000,0x324a3dcd,0x7c00100,0x1230400,0x324a3fc8,0x4000002,0x1200c00,0x324a53c5,0x24000000, +0xe00000,0x32820701,0x2802000,0x962460,0x40000419,0x7c00100,0x220400,0x40000519,0x7c00100,0x220400,0x40000600,0x4000400,0x200400,0x4000080e,0x7c00100,0x220400, +0x4000080e,0x7c00100,0x250400,0x4000080e,0x7c00100,0x250402,0x40000c02,0x2802100,0x962460,0x40000c02,0x2802400,0x962460,0x40000c02,0x2802500,0x962460,0x40000c02, +0x4000000,0x200000,0x40000c02,0x4000000,0x1071400,0x40000c02,0x7c00100,0x230400,0x40000c02,0x80000000,0x218960,0x40000d22,0x7c00100,0x230400,0x40000f0a,0x7c00100, +0x230400,0x40001004,0x7c00100,0x230400,0x40001110,0x2802100,0x962460,0x40001110,0x6800100,0x962540,0x4000120f,0x2802100,0x962460,0x4000120f,0x4000000,0x1600000, +0x4000120f,0x7c00100,0x230400,0x4000131f,0x7c00100,0x230400,0x40001423,0x4000000,0x200000,0x40001423,0x4000000,0x1600000,0x40001615,0x2802400,0x962460,0x40001615, +0x7c00100,0x230400,0x40002417,0x2802400,0x1862460,0x40002417,0x4000000,0x200000,0x40002800,0x6800000,0x201c00,0x40002800,0x24000002,0x200000,0x40002c00,0x4000000, +0x200002,0x40003000,0x24000000,0x200000,0x40003000,0x24000020,0x200000,0x40003700,0x24000000,0x200000,0x40005a09,0x7c00100,0x220400,0x40005a09,0x7c00100,0x250400, +0x40005d00,0x7c00120,0x220405,0x40006f30,0x2802100,0x962460,0x40006f30,0x2802400,0x962460,0x40006f30,0x4000000,0x200000,0x40006f30,0x6800000,0x1329800,0x40006f30, +0x6800100,0x962540,0x40006f30,0x7c00100,0x230400,0x40006f30,0xc000010,0xb48000,0x40007034,0x7c00100,0x1830000,0x40007117,0x4000000,0x200000,0x40007208,0x7c00100, +0x220400,0x4000720e,0x7c00100,0x220400,0x4000720e,0x7c00500,0x22040e,0x4000720e,0x7c00500,0x22040f,0x40007219,0x7c00100,0x220400,0x40007219,0x7c00500,0x220400, +0x40007219,0x7c00500,0x22040e,0x40007219,0x7c00500,0x22040f,0x40007300,0x24000000,0x200000,0x40007400,0x4000000,0x200000,0x40007531,0x7c00100,0x230400,0x40007631, +0x7c00100,0x230400,0x40007835,0x4000010,0x400000,0x40007835,0x7c00100,0x230400,0x40007933,0x7c00100,0x230400,0x40007a32,0x6800000,0x1329800,0x40007a32,0x7c00100, +0x230400,0x40007b2f,0x7c00100,0x230400,0x40007c00,0x4000000,0x200000,0x40020701,0x2802400,0x962460,0x40020701,0x2802400,0xc62460,0x40023300,0x4000000,0x200000, +0x40027d01,0x12882000,0x962460,0x400a3700,0x24000000,0x200000,0x400a3700,0x24000000,0xe00000,0x400a4400,0x4000000,0xe0000d,0x400a4412,0x4000000,0xe00002,0x400a4412, +0x4000000,0xe00003,0x400a4500,0x4000000,0xe0000d,0x400a5300,0x4000000,0x810010,0x400a5300,0x4000000,0x1410010,0x40507709,0x4000000,0x200000,0x4050770c,0x4000000, +0x400000,0x4050770f,0x4000000,0x200000,0x4050770f,0x4000000,0x400000,0x40c01489,0x4000000,0x200000,0x40d05107,0x4000000,0x200000,0x41000419,0x7c00100,0x220400, +0x41000419,0x7c00100,0x250400,0x4100080e,0x7c00100,0x220400,0x4100080e,0x7c00100,0x250400,0x41000908,0x7c00100,0x220400,0x41000908,0x7c00100,0x250400,0x41000b13, +0x2802000,0x962460,0x41000b13,0x2802100,0x962460,0x41000b13,0x4000000,0xb00000,0x41000c02,0x2802100,0x962460,0x41000c02,0x4000000,0x1500000,0x41000c02,0xc000010, +0xb48000,0x41000f0a,0x7c00100,0x230400,0x41001004,0x7c00100,0x230400,0x41001423,0x7c00100,0x230400,0x41001b27,0x4000000,0x500000,0x41001d0c,0x7c00100,0x22040f, +0x41001d0c,0x7c00100,0x230400,0x41001f0b,0x2802400,0x962460,0x41001f0b,0x4000000,0x200000,0x41001f0b,0x7c00100,0x230400,0x41002800,0x24000000,0x200000,0x41002800, +0x24000000,0x400000,0x41002919,0x7c00100,0x22040e,0x41002a00,0x4000000,0x1600000,0x41002b01,0x2802020,0x962460,0x41002c00,0x4000000,0x200000,0x41002c00,0x7c00120, +0x220405,0x41003000,0x24000000,0x200000,0x41003700,0x24000000,0x200000,0x41003700,0x24000000,0xe00000,0x41005d00,0x7c00120,0x220405,0x41006600,0x24000020,0x200000, +0x41006600,0x24000020,0x810000,0x41006600,0x24000020,0x1410000,0x41007208,0x7c00100,0x22040f,0x41007219,0x7c00100,0x220400,0x41007300,0x24000000,0x200000,0x41007e0e, +0x2802000,0x962460,0x41007e0e,0x4000000,0x200000,0x41007f0e,0x4000000,0x200000,0x41007f0e,0x7c00100,0x230400,0x41008002,0x7c00100,0x230400,0x41008137,0x2802100, +0x962460,0x41008137,0x4000000,0x200000,0x41008137,0x6800100,0x962540,0x41008137,0x7c00100,0x230400,0x41008301,0x2802000,0x962460,0x41008407,0x4000000,0x200000, +0x41008407,0x4000000,0x400000,0x41008407,0x4000000,0xb00000,0x41008407,0x7c00100,0x220400,0x41008407,0x7c00100,0x250400,0x4100850b,0x7c00100,0x230400,0x4100860b, +0x4000000,0x200000,0x4100860b,0x7c00100,0x230400,0x4100870c,0x7c00100,0x220400,0x41008838,0x7c00100,0x220400,0x41008838,0x7c00100,0x250400,0x41008939,0x2802000, +0x962460,0x41008939,0x2802100,0x962460,0x41008939,0x2806000,0x962460,0x41008939,0x4000000,0x200000,0x41008939,0x4000000,0x400000,0x41008939,0x7c00100,0x230400, +0x41008939,0xc000010,0x448000,0x41008a00,0x4000400,0x200400,0x41008b3b,0x4000000,0x1800000,0x41008b3b,0x6800000,0x1329800,0x41008b3b,0x7c00100,0x1830000,0x41008b3b, +0x7e00100,0x1830000,0x41008c3d,0x4000010,0x400000,0x41008c3d,0x7c00100,0x230400,0x41008d0e,0x7c00100,0x22040f,0x41008d19,0x7c00100,0x220400,0x41008d19,0x7c00100, +0x22040f,0x41008e00,0x24000000,0x200000,0x41008e00,0x24000000,0x400000,0x41008e00,0x24000000,0x1710000,0x41008e00,0x24000006,0x400000,0x41008f3a,0x2802100,0x962460, +0x41008f3a,0x2806000,0x962460,0x41008f3a,0x4000000,0x200000,0x41008f3a,0x6800100,0x962540,0x41008f3a,0x7c00100,0x230400,0x4100903c,0x7c00100,0x230400,0x4100903c, +0x7c00100,0x23040f,0x41020701,0x2802000,0x962460,0x41020701,0x2802000,0xc62460,0x410a3700,0x24000000,0x200000,0x410a3700,0x24000000,0xe00000,0x410a4412,0x4000000, +0xe00003,0x410a4711,0x7c40300,0xe30000,0x410a4f11,0x7c00300,0xe30001,0x410a9100,0x4000000,0x800010,0x410a9100,0x4000000,0x810010,0x410a9100,0x4000000,0x870010, +0x410a9100,0x4000000,0xb00010,0x410a9100,0x4000000,0xf00010,0x410a9100,0x4000000,0x1001410,0x410a9100,0x4000000,0x1071010,0x410a9100,0x4000000,0x1071410,0x410a9100, +0x4000000,0x1410010,0x41408ad0,0x4000400,0x200000,0x414a82ca,0x4000000,0xe00000,0x41808300,0x2802000,0x962460,0x41c01489,0x6800000,0x1329800,0x50000419,0x7c00100, +0x220400,0x50000419,0x7c00100,0x250400,0x5000080e,0x7c00100,0x220400,0x50000908,0x7c00100,0x220400,0x50000908,0x7c00100,0x250400,0x50000b13,0x2802500,0x962460, +0x50000f0a,0x7c00100,0x230400,0x50001615,0x2802100,0x962460,0x50001615,0x7c00100,0x230400,0x50002b01,0x2802020,0x962460,0x50002c00,0x4000000,0x200000,0x50002c19, +0x7c00100,0x220400,0x50002d19,0x7c00100,0x220400,0x50003000,0x24000000,0x200000,0x50003000,0x24000020,0x200000,0x50003700,0x24000000,0x200000,0x50005d00,0x7c00120, +0x220405,0x50005d00,0x7c00120,0x250405,0x50006108,0x7c00100,0x220400,0x50006108,0x7c00100,0x250400,0x50006600,0x24000020,0x200000,0x50007300,0x24000000,0x200000, +0x50008301,0x2802400,0x962460,0x50008a00,0x7c00500,0x230400,0x50009257,0x2802400,0x962460,0x50009257,0x4000000,0x200000,0x50009257,0x4000010,0x1071400,0x50009257, +0x6800000,0x1329800,0x50009257,0x7c00100,0x230400,0x50009257,0x7c00500,0x230400,0x50009257,0x7c00900,0x230400,0x50009257,0xc000010,0xb48000,0x5000933e,0x2802100, +0x962460,0x5000933e,0x2802400,0x962460,0x5000933e,0x4000000,0x200000,0x5000933e,0x4000000,0x400000,0x5000933e,0x4000010,0x400000,0x5000933e,0x6800000,0x1329800, +0x5000933e,0x6800100,0x962540,0x5000933e,0x6800100,0x962541,0x5000933e,0x6804400,0x962540,0x5000933e,0x7c00100,0x230400,0x5000933e,0x7c00100,0x230401,0x5000933e, +0xc000010,0x448000,0x50009419,0x7c00100,0x220400,0x50009419,0x7c00100,0x250400,0x50009500,0x4000400,0x200400,0x5000965a,0x4000000,0x500000,0x5000965a,0x7c00100, +0x230400,0x5000965a,0xc000010,0xb48000,0x5000975b,0x4000000,0x200000,0x5000975b,0x4000010,0x400000,0x5000975b,0x7c00100,0x230400,0x50009865,0x7c00100,0x230400, +0x50009965,0x4000010,0x400000,0x50009965,0x7c00100,0x230400,0x50409aca,0x4000000,0x200000,0x5100080e,0x7c00100,0x220400,0x5100080e,0x7c00100,0x250400,0x51000c02, +0x2802100,0x962460,0x51000c02,0x4000000,0x1500000,0x51000c02,0x4000020,0x200000,0x51000c02,0x7c00100,0x230400,0x51000f0a,0x7c00100,0x230400,0x51000f0a,0x7c00500, +0x230400,0x51001110,0x2802100,0x962460,0x5100131f,0x2802100,0x962460,0x51001423,0x7c00100,0x230400,0x51001524,0x2802100,0x962460,0x51001524,0x4000000,0x200000, +0x51001524,0x7c00100,0x230400,0x5100171a,0x2802100,0x962460,0x5100171a,0x4000000,0x200000,0x5100171a,0x4000000,0x1500000,0x5100171a,0x7c00100,0x230400,0x51001b27, +0x4000000,0x200000,0x51001b27,0x4000000,0x400000,0x51001b27,0x4000000,0x500000,0x51001b27,0x7c00100,0x230400,0x51001c1c,0x2802100,0x1862460,0x51001c1c,0x2802500, +0x1862460,0x51001c1c,0x2806400,0x1862460,0x51001c1c,0x4000000,0x1800000,0x51001c1c,0x6800000,0x1329800,0x51001c1c,0x6800100,0x1862400,0x51001c1c,0x6800100,0x1862540, +0x51001c1c,0x6800500,0x1862400,0x51001c1c,0x7c00100,0x1830000,0x5100251b,0x7c00100,0x230400,0x51002619,0x7c00100,0x220400,0x51002619,0x7c00100,0x250400,0x51002800, +0x80020,0x218820,0x51002c00,0x4000000,0x200000,0x51002d19,0x7c00100,0x230400,0x51003700,0x24000000,0x200000,0x51003700,0x24000000,0xe00000,0x51005201,0x2802400, +0x962460,0x51005c00,0x4000000,0x200000,0x51006108,0x7c00100,0x220400,0x51006108,0x7c00100,0x250400,0x51006600,0x24000020,0x200000,0x51006600,0x24000020,0x810000, +0x51006600,0x24000020,0x1410000,0x51007300,0x24000000,0x200000,0x51007300,0x24000020,0x200000,0x51008002,0x7c00100,0x230400,0x51008301,0x2802000,0x962460,0x51008301, +0x2802400,0x962460,0x51008301,0x2802400,0xc62460,0x51008a00,0x7c00500,0x230400,0x51008e00,0x24000000,0x200000,0x51008e00,0x24000000,0x400000,0x51008e00,0x24000000, +0x810000,0x51008e00,0x24000000,0x1400000,0x51008e00,0x24000000,0x1410000,0x51008e00,0x24000000,0x1710000,0x51008e00,0x24000002,0x200000,0x51008e00,0x24000500,0x230400, +0x51008e00,0x2c000010,0xb48000,0x51009419,0x7c00100,0x220400,0x51009419,0x7c00100,0x22040e,0x51009419,0x7c00100,0x22040f,0x51009419,0x7c00100,0x250400,0x51009500, +0x4000400,0x200400,0x51009500,0x7c00500,0x230400,0x51009519,0x7c00100,0x220400,0x51009519,0x7c00100,0x22040f,0x51009519,0x7c00100,0x230400,0x51009519,0x7c00100, +0x250400,0x51009b71,0x2802100,0x962460,0x51009b71,0x6800000,0x1329800,0x51009b71,0x6800100,0x962540,0x51009b71,0x6804400,0x962540,0x51009b71,0x7c00100,0x230400, +0x51009c52,0x2802100,0x962460,0x51009c52,0x2802400,0x962460,0x51009c52,0x2802d00,0x962460,0x51009c52,0x4000010,0x400000,0x51009c52,0x6800000,0x1329800,0x51009c52, +0x6800100,0x962540,0x51009c52,0x7c00100,0x230400,0x51009c52,0xc000010,0x448000,0x51009d6d,0x6800000,0x1329800,0x51009d6d,0x7c00100,0x230400,0x51009d6d,0x7c00500, +0x230400,0x51009d6d,0x7c00d00,0x230400,0x51009d6d,0xc000010,0x448000,0x51009e08,0x2802100,0x962460,0x51009f63,0x4000010,0x400000,0x51009f63,0x6800000,0x1329800, +0x51009f63,0x7c00100,0x230400,0x51009f63,0x7c00900,0x230400,0x51009f63,0xc000010,0x448000,0x51009f63,0xc000010,0xb48000,0x5100a008,0x2000,0x962460,0x5100a008, +0x2802400,0x962460,0x5100a008,0x4000000,0x200000,0x5100a008,0x7c00100,0x220400,0x5100a008,0x7c00100,0x230400,0x5100a008,0x7c00100,0x250400,0x5100a008,0x7c00500, +0x230400,0x5100a16f,0x2806400,0x962460,0x5100a16f,0x6800000,0x1329800,0x5100a16f,0x6800100,0x962540,0x5100a16f,0x7c00100,0x230400,0x5100a16f,0xc000010,0x448000, +0x5100a24f,0x2802100,0x962460,0x5100a24f,0x2802400,0x962460,0x5100a24f,0x6800000,0x1329800,0x5100a24f,0x7c00100,0x230400,0x5100a24f,0xc000010,0x448000,0x5100a36e, +0x2802100,0x962460,0x5100a36e,0x4000000,0x200000,0x5100a36e,0x6800100,0x962540,0x5100a36e,0x6804400,0x962540,0x5100a36e,0x7c00100,0x230400,0x5100a442,0x2802100, +0x962460,0x5100a442,0x4000000,0x200000,0x5100a442,0x6800000,0x1329800,0x5100a442,0x6800100,0x962540,0x5100a442,0x7c00100,0x230400,0x5100a442,0xc000010,0x448000, +0x5100a500,0x4000000,0x200000,0x5100a600,0x4000000,0x200000,0x5100a601,0x2802000,0x962460,0x5100a76b,0x7c00100,0x230400,0x5100a868,0x7c00100,0x230400,0x5100a96c, +0x4000000,0x200000,0x5100a96c,0x7c00100,0x230400,0x5100aa00,0x4000000,0xe00000,0x5100ab00,0x4000000,0xe00000,0x51086600,0x24000020,0x810000,0x51086600,0x24000020, +0x1410000,0x510a4005,0x7c00100,0xe30400,0x510a4711,0x7c40300,0xe30000,0x510a7300,0x24000000,0x200000,0x510aaa00,0x4000000,0xe00000,0x5140a2fe,0x4000400,0x400000, +0x514a82ca,0x4000000,0xe00000,0x51802bbc,0x2802000,0x962460,0x51c00908,0x2802400,0x962460,0x51c0a008,0x2802400,0x962460,0x52000f0a,0x2802100,0x962460,0x52000f0a, +0x6800100,0x962540,0x52000f0a,0x7c00100,0x230400,0x52001004,0x4000000,0x1600000,0x52001b00,0x4000000,0x200000,0x52001c1c,0x2802100,0x1862460,0x52001c1c,0x6800100, +0x1862400,0x52001c1c,0x6800500,0x1862400,0x52001e12,0x7c00100,0x2230500,0x52001e12,0x7c00100,0x2330520,0x52002128,0x4000002,0x400000,0x52002128,0x7c00100,0x230400, +0x52002a00,0x4000000,0x1500000,0x52002a00,0x4000000,0x1600000,0x52002d00,0x4000000,0x200006,0x52003000,0x24000000,0x200000,0x52006108,0x7c00100,0x220400,0x52006108, +0x7c00100,0x250400,0x52008301,0x2802400,0x962460,0x52008407,0x2802400,0x962460,0x52008407,0x7c00100,0x220400,0x52008407,0x7c00100,0x250400,0x52008b3b,0x6800000, +0x1800000,0x52008b3b,0x7c00100,0x1830000,0x52008e00,0x24000000,0x400000,0x52009419,0x7c00100,0x250400,0x5200975b,0x4000000,0x200000,0x5200ac7e,0x2802000,0x962460, +0x5200ac7e,0x2802100,0x962460,0x5200ac7e,0x2802400,0x962460,0x5200ac7e,0x4000010,0x200000,0x5200ac7e,0x7c00100,0x230400,0x5200ac7e,0xc000010,0x248000,0x5200ad28, +0x7c00100,0x230400,0x5200ae6a,0x2802100,0x1862460,0x5200ae6a,0x2802400,0x962460,0x5200ae6a,0x2802400,0x1862460,0x5200ae6a,0x2806000,0x1862460,0x5200ae6a,0x4000000, +0x1800000,0x5200ae6a,0x6800000,0x1329800,0x5200ae6a,0x6800100,0x1862400,0x5200ae6a,0x6800100,0x1862540,0x5200ae6a,0x7c00100,0x1830000,0x5200ae6a,0x7c00900,0x1830000, +0x5200ae6a,0xc000010,0x1848000,0x5200b083,0x4000010,0x400000,0x5200b083,0x7c00100,0x230400,0x5200b083,0xc000010,0x448000,0x5200b182,0x2802400,0x962460,0x5200b182, +0x4000000,0x200000,0x5200b182,0x4000010,0x400000,0x5200b182,0x7c00100,0x230400,0x5200b182,0xc000010,0x448000,0x5200b30a,0x2802400,0x962460,0x5200b30a,0x4000000, +0x200000,0x5200b30a,0x7c00100,0x230400,0x5200b54e,0x2802100,0x962460,0x5200b54e,0x2802400,0x962460,0x5200b54e,0x4000000,0x200000,0x5200b54e,0x4000010,0x400000, +0x5200b54e,0x6800000,0x1329800,0x5200b54e,0x6800100,0x962540,0x5200b54e,0x6804400,0x962540,0x5200b54e,0x7c00100,0x230400,0x5200b54e,0xc000010,0x448000,0x5200b61c, +0x4000000,0x1800000,0x5200b61c,0x6800500,0x1862400,0x5200b61c,0x7c00100,0x1830000,0x5200b61c,0x7c00900,0x1830000,0x5200b77f,0x2802100,0x1862460,0x5200b77f,0x2802400, +0x1862460,0x5200b77f,0x4000000,0x1800000,0x5200b77f,0x4000010,0x1800000,0x5200b77f,0x7c00100,0x1830000,0x5200b77f,0x7c00500,0x1830000,0x5200b77f,0x7c00900,0x1830000, +0x5200b77f,0x7e00100,0x1830000,0x5200b873,0x2802100,0x962460,0x5200b873,0x2806400,0x962460,0x5200b873,0x6800000,0x1329800,0x5200b873,0x6800100,0x962540,0x5200b873, +0x6800400,0x962540,0x5200b873,0x7c00100,0x230400,0x5200b873,0xc000010,0x448000,0x5200b912,0x7c00100,0x2230500,0x5200b912,0x7c00100,0x2330520,0x5200ba74,0x4000000, +0x200000,0x5200ba74,0x4000010,0x400000,0x5200ba74,0x7c00100,0x230400,0x5200bb85,0x4000000,0x200000,0x5200bb85,0x7c00100,0x230400,0x5200bc75,0x4000000,0x400000, +0x5200bc75,0x4000010,0x400000,0x5200bc75,0x7c00100,0x230400,0x5200bd7d,0x4000000,0x200000,0x5200bd7d,0x7c00100,0x230400,0x5200be7a,0x4000000,0x200000,0x5200be7a, +0x7c00100,0x230400,0x5200bf58,0x7c00100,0x230400,0x5200c002,0x4000000,0x200000,0x5200c178,0x2802100,0x962460,0x5200c178,0x2802400,0x962460,0x5200c178,0x2806400, +0x962460,0x5200c178,0x4000000,0x200000,0x5200c178,0x6800100,0x962540,0x5200c178,0x7c00100,0x230400,0x5200c178,0x7c00100,0x230401,0x5200c178,0xc000010,0x448000, +0x5200c178,0x80000000,0x218960,0x5200c247,0x7c00100,0x230400,0x5200c247,0x7c00100,0x830400,0x5200c247,0x7c00100,0x1430400,0x5200c300,0x4000000,0x200003,0x52022d00, +0x4000000,0x100006,0x52023700,0x24000000,0x100000,0x52023700,0x24000000,0xe00000,0x52023700,0x24000000,0x2800000,0x52024400,0x4000000,0x100000,0x52027300,0x24000000, +0x100000,0x5202c300,0x4000000,0x100000,0x5202c300,0x4000000,0x100002,0x5202c300,0x4000000,0x100003,0x5202c300,0x4000000,0x10000d,0x5202c300,0x4000100,0x150400, +0x5202c300,0x4000100,0x15040d,0x520a1e12,0x7c00100,0x2130480,0x520a3700,0x24000000,0xe00000,0x520a3800,0x24000000,0x100000,0x520a4711,0x7c40300,0xe30000,0x520a4f11, +0x7c00300,0xe30001,0x520a7300,0x24000000,0x100000,0x520ab412,0x7c00100,0x2130480,0x520ac400,0x4000000,0xe00002,0x520ac400,0x4000000,0xe0000d,0x520ac414,0x4000000, +0xe0000d,0x520ac511,0x7c40300,0xe30000,0x5240af9c,0x7c00100,0x230400,0x5240afa1,0x4000400,0x200000,0x5240afa3,0x6800400,0x962540,0x5240afa3,0x7c00100,0x230400, +0x5240afad,0x7c00100,0x230400,0x5240afaf,0x7c00100,0x230400,0x5240b2d2,0x4000000,0x200000,0x5240b2d2,0x4000000,0x1500000,0x5240b2dd,0x4000000,0x200000,0x5240b2eb, +0x4000000,0x200000,0x524a44ca,0x4000000,0xe00003,0x5250b501,0x7c00900,0x230400,0x5280af9c,0x2802400,0x962460,0x5280af9d,0x2802400,0x962460,0x5280afa3,0x2802400, +0x962460,0x5280afa5,0x2802400,0x962460,0x5280afa7,0x2802400,0x962460,0x52c0b3f8,0x2802400,0x962460,0x52c0b3fc,0x7c00100,0x230400,0x60000c02,0x2802100,0x962460, +0x60000c02,0x7c00100,0x230400,0x60000f0a,0x2802100,0x962460,0x60000f0a,0x6800100,0x962540,0x60000f0a,0x7c00100,0x230400,0x6000131f,0x4000000,0x200000,0x6000171a, +0x7c00100,0x230400,0x6000171a,0x7c00100,0x230560,0x60001b27,0x2802100,0x962460,0x60001b27,0x4000000,0xc00000,0x60001b27,0x7c00100,0x230400,0x60001f0b,0x2802400, +0x962460,0x60002919,0x7c00100,0x22040e,0x60002a00,0x4000000,0x1600000,0x60003000,0x24000000,0x200000,0x60003000,0x24000000,0xe00000,0x60003700,0x24000000,0x200000, +0x60003800,0x24000000,0x1710000,0x60005102,0x4000000,0x200000,0x60006108,0x7c00100,0x220400,0x60006108,0x7c00100,0x250400,0x60006600,0x24000020,0x200000,0x60008301, +0x2802400,0xc62460,0x6000903c,0x2806000,0x962460,0x6000903c,0x4000000,0x400000,0x60009519,0x7c00100,0x220400,0x60009519,0x7c00100,0x250400,0x6000a008,0x7c00100, +0x220400,0x6000a008,0x7c00100,0x250400,0x6000c300,0x4000000,0x2703580,0x6000c654,0x2802000,0x962460,0x6000c654,0x4000010,0x200000,0x6000c654,0x7c00100,0x230400, +0x6000c73f,0x2802000,0x962460,0x6000c73f,0x2802100,0x962460,0x6000c73f,0x4000000,0x200000,0x6000c73f,0x6800100,0x962540,0x6000c73f,0x6804000,0x962540,0x6000c73f, +0x7c00100,0x230400,0x6000c80b,0x7c00100,0x230400,0x6000c941,0x2802100,0x962460,0x6000c941,0x2806400,0x962460,0x6000c941,0x4000000,0x200000,0x6000c941,0x4000010, +0x200000,0x6000c941,0x6800000,0x1329800,0x6000c941,0x6800100,0x962540,0x6000c941,0x7c00100,0x230400,0x6000c941,0xc000010,0x448000,0x6000ca82,0x7c00100,0x230400, +0x6000cc00,0x4000000,0xe00000,0x6000d000,0x4000000,0x200000,0x6002c300,0x4000000,0x100000,0x6002c300,0x4000000,0x10000d,0x6002c300,0x4000100,0x150400,0x6002c300, +0x4000100,0x15040d,0x600a3000,0x24000000,0x200000,0x600a3000,0x24000000,0xe00000,0x600a3700,0x24000000,0x200000,0x600a3800,0x24000000,0x200000,0x600a3800,0x24000000, +0x2800000,0x600a4305,0x7c00100,0xe30400,0x600ac300,0x4000000,0x100000,0x600ac400,0x4000000,0xe0000d,0x600acb14,0x7c00100,0xe30000,0x600acb16,0x7c00100,0xe30c00, +0x600acc00,0x4000000,0xe00000,0x600acd00,0x4000000,0x200000,0x600acd00,0x4000000,0xe00000,0x600acd00,0x4000000,0x2800000,0x600ace00,0x4000000,0xe00000,0x600ace00, +0x4000000,0x2800000,0x600acf00,0x4000000,0xe00000,0x600acf00,0x4000000,0x2800000,0x600ad111,0x7c40300,0xe30000,0x604ac4ca,0x4000000,0xe00003,0x61000a03,0x4000000, +0x1600000,0x61000c02,0x80000000,0x218960,0x6100120f,0x4000000,0x200000,0x61001a18,0x7c00100,0x1830000,0x61001d0c,0x7c00100,0x230400,0x61001d0c,0x7c00100,0x250400, +0x61006600,0x24000020,0x200000,0x61008407,0x7c00100,0x220400,0x61008407,0x7c00100,0x250400,0x6100870c,0x7c00100,0x220400,0x61008e00,0x24000000,0x200000,0x61008e00, +0x24000000,0x400000,0x61008e00,0x24000002,0x300000,0x6100903c,0x7c00100,0x230400,0x61009519,0x7c00100,0x220400,0x61009519,0x7c00100,0x250400,0x61009519,0x7c00500, +0x22040f,0x61009b71,0x2802100,0x962460,0x61009b71,0x2806400,0x962460,0x61009b71,0x7c00100,0x230400,0x6100a008,0x2802100,0x962460,0x6100c300,0x4000000,0x20000f, +0x6100cd00,0x4000000,0x200000,0x6100d202,0x2802400,0x962460,0x6100d202,0x2802500,0x962460,0x6100d202,0x7c00100,0x230400,0x6100d302,0x4000020,0x200000,0x6100d302, +0x7c00120,0x230405,0x6100d476,0x2802100,0x962460,0x6100d476,0x2802100,0x962461,0x6100d476,0x2806400,0x962460,0x6100d476,0x4000000,0x400000,0x6100d476,0x6800000, +0x1329800,0x6100d476,0x6800100,0x962540,0x6100d476,0x7c00100,0x230400,0x6100d476,0xc000010,0x448000,0x6100d573,0x2802100,0x962460,0x6100d573,0x2806400,0x962460, +0x6100d573,0x6800100,0x962540,0x6100d573,0x7c00100,0x230400,0x6100d573,0x7c00900,0x230400,0x6100d573,0xc000010,0x448000,0x6100d68d,0x7c00100,0x230400,0x6100d756, +0x7c00100,0x230400,0x6100d85c,0x2802500,0x962460,0x6100d85c,0x6800100,0x962540,0x6100d85c,0x7c00100,0x230400,0x6100d85c,0x7c00500,0x230400,0x6100d997,0x2802100, +0x962460,0x6100d997,0x4000000,0x200000,0x6100d997,0x4000000,0x400000,0x6100d997,0x6800000,0x1329800,0x6100d997,0x6800100,0x962540,0x6100d997,0x6804400,0x962540, +0x6100d997,0x7c00100,0x230400,0x6100d997,0x7c00100,0x230560,0x6100d997,0xc000010,0x448000,0x6100da98,0x6800000,0x1329800,0x6100da98,0x7c00100,0x230400,0x6100db71, +0x4000000,0x200000,0x6100dc99,0x2802100,0x962460,0x6100dc99,0x2802400,0x962460,0x6100dc99,0x6800000,0x1329800,0x6100dc99,0x6800100,0x962540,0x6100dc99,0x6804400, +0x962540,0x6100dc99,0x7c00100,0x230400,0x610a4711,0x7c40300,0xe30000,0x610a4f11,0x7c00300,0xe30001,0x610ace00,0x4000000,0xe00000,0x6140afa1,0x7c00100,0x230400, +0x6140afa3,0x7c00100,0x230400,0x6180af9e,0x2802400,0x962460,0x62002a00,0x4000000,0x1600000,0x63002800,0x80000,0x918820,0x63c00c15,0x80000,0x918820,0x7000080e, +0x7c00100,0x250400,0x70000a03,0x4000000,0x200000,0x70000c00,0x80000000,0x218960,0x70000f0a,0x7c00100,0x230400,0x70001004,0x7c00100,0x230400,0x70001524,0x2802100, +0x962460,0x70001524,0x7c00100,0x230400,0x70001615,0x2802100,0x962460,0x7000171a,0x2802100,0x962460,0x70001821,0x6800000,0x1329800,0x70002320,0x7c00100,0x230400, +0x70002a00,0x4000000,0x1500000,0x70002a00,0x4000000,0x1600000,0x70003000,0x24000000,0x200000,0x70003800,0x24000000,0xe00000,0x70005201,0x2802400,0x962460,0x7000581e, +0x7c00100,0x230400,0x70006108,0x7c00100,0x220400,0x70006108,0x7c00100,0x250400,0x70006f30,0x7c00100,0x230400,0x70007300,0x24000000,0x200000,0x70007f0e,0x4000000, +0x200000,0x70008301,0x2802100,0x962460,0x70008301,0x2802400,0x962460,0x70008e00,0x24000000,0x200000,0x70008e00,0x24000000,0x400000,0x70008e00,0x24000002,0x400000, +0x70008e00,0x24000008,0x1410000,0x70008e00,0x24000010,0x400000,0x70008e00,0x2c000010,0x448000,0x70009519,0x7c00100,0x220400,0x70009519,0x7c00100,0x230400,0x70009519, +0x7c00100,0x250400,0x70009865,0x7c00100,0x230400,0x70009965,0x4000010,0x400000,0x70009965,0x7c00100,0x230400,0x7000a008,0x7c00100,0x220400,0x7000a008,0x7c00100, +0x250400,0x7000a008,0x7c00500,0x22040f,0x7000a50e,0x4000000,0x200000,0x7000b61c,0x2802500,0x1862460,0x7000b61c,0x6800500,0x1862400,0x7000b61c,0x7c00100,0x1830000, +0x7000c300,0x4000000,0x100000,0x7000c941,0x2806000,0x962460,0x7000cc00,0x4000000,0xe00000,0x7000cd00,0x4000000,0x200000,0x7000cd00,0x4000000,0xe00000,0x7000cd00, +0x4000000,0x2800000,0x7000cf00,0x4000000,0xe00000,0x7000d202,0x2802100,0x962460,0x7000d202,0x7c00100,0x230400,0x7000d997,0x7c00100,0x230400,0x7000d997,0xc000010, +0x248000,0x7000dd86,0x2802400,0x962460,0x7000dd86,0x7c00100,0x230400,0x7000dd86,0xc000010,0x448000,0x7000de9f,0x4000000,0x200000,0x7000de9f,0x7c00100,0x230400, +0x7000e001,0x2400,0x962460,0x7000e001,0x2802400,0x962460,0x7000e187,0x2802000,0x962460,0x7000e187,0x2802100,0x962460,0x7000e187,0x4000000,0x200000,0x7000e187, +0x7c00100,0x230400,0x7000e187,0xc000010,0x448000,0x7000e288,0x7c00100,0x230400,0x7000e300,0x4000000,0x200000,0x7000e489,0x2802100,0x962460,0x7000e489,0x2802400, +0x962460,0x7000e489,0x6800100,0x962540,0x7000e489,0x6800100,0x962541,0x7000e489,0x6804400,0x962540,0x7000e489,0x7c00100,0x230400,0x7000e489,0x7c00900,0x230400, +0x7000e59d,0x2802100,0x962460,0x7000e59d,0x2802400,0x962460,0x7000e59d,0x4000000,0x200000,0x7000e59d,0x4000010,0x200000,0x7000e59d,0x6800100,0x962540,0x7000e59d, +0x6804400,0x962540,0x7000e59d,0x7c00100,0x230400,0x7000e59d,0xc000010,0x448000,0x7000e691,0x2802100,0x962460,0x7000e691,0x2802400,0x962460,0x7000e691,0x2806400, +0x962460,0x7000e691,0x6800000,0x1329800,0x7000e691,0x6800100,0x962540,0x7000e691,0x7c00100,0x230400,0x7000e700,0x4000400,0x200400,0x7000e70e,0x7c00100,0x220400, +0x7000e719,0x7c00100,0x220400,0x7000e719,0x7c00500,0x22040f,0x7000e853,0x7c00100,0x230400,0x7000e9a0,0x2802400,0x962460,0x7000e9a0,0x4000000,0x200000,0x7000e9a0, +0x4000000,0x500000,0x7000e9a0,0x7c00100,0x230400,0x7000ea79,0x2802400,0x962460,0x7000ea79,0x4000000,0x200000,0x7000ea79,0x4000000,0xf00000,0x7000ea79,0x4000010, +0x400000,0x7000ea79,0x7c00100,0x230400,0x7000eb8c,0x2802400,0x962460,0x7000eb8c,0x4000000,0x200000,0x7000eb8c,0x7c00100,0x230400,0x7000eca3,0x2802100,0x962460, +0x7000eca3,0x2806400,0x962460,0x7000eca3,0x4000000,0x200000,0x7000eca3,0x6800000,0x1329800,0x7000eca3,0x6800100,0x962540,0x7000eca3,0x7c00100,0x230400,0x7000eca3, +0xc000010,0x448000,0x7000ed95,0x6800000,0x1329800,0x7000ed95,0x7c00100,0x230400,0x7000ed95,0xc000010,0x448000,0x7000ee1c,0x2802500,0x1862460,0x7000ee1c,0x6800000, +0x1329800,0x7000ee1c,0x7c00100,0x1830000,0x7000ee1c,0x7c00900,0x1830000,0x7000ef8f,0x4000000,0x200000,0x7000ef8f,0x7c00100,0x230400,0x7000f08e,0x4000000,0x200000, +0x7000f08e,0x7c00100,0x230400,0x7000f159,0x2802100,0x962460,0x7000f159,0x7c00100,0x230400,0x7000f200,0x4000000,0x200000,0x7000f200,0x4000000,0x1200000,0x7000f200, +0x4000000,0x1710000,0x7000f34b,0x2802400,0x962460,0x7000f34b,0x4000000,0x200000,0x7000f34b,0x4000010,0x400000,0x7000f34b,0x6800000,0x1329800,0x7000f34b,0x7c00100, +0x230400,0x7000f34b,0x7c00900,0x230400,0x7000f34b,0xc000010,0x448000,0x7000f490,0x4000000,0x200000,0x7000f490,0x7c00100,0x230400,0x7000f5a5,0x7c00100,0x230400, +0x7000f67b,0x4000000,0x200000,0x7000f67b,0x4000010,0x200000,0x7000f67b,0x7c00100,0x230400,0x7000f8a6,0x2802100,0x962460,0x7000f8a6,0x2802400,0x962460,0x7000f8a6, +0x2806400,0x962460,0x7000f8a6,0x4000000,0x500000,0x7000f8a6,0x4000010,0xb00000,0x7000f8a6,0x4000800,0x200000,0x7000f8a6,0x6800100,0x962540,0x7000f8a6,0x6800100, +0x962541,0x7000f8a6,0x7c00100,0x230400,0x7000f8a6,0xc000010,0x448000,0x7000f921,0x4000000,0x200000,0x7000fa00,0x4000000,0x200000,0x7000fb9e,0x2802100,0x962460, +0x7000fb9e,0x2802400,0x962460,0x7000fb9e,0x2806400,0x962460,0x7000fb9e,0x4000000,0x200000,0x7000fb9e,0x6800000,0x1329800,0x7000fb9e,0x6800100,0x962540,0x7000fb9e, +0x6800100,0x962541,0x7000fb9e,0x7c00100,0x230400,0x7000fc92,0x4000000,0x200000,0x7000fc92,0x6800000,0x1329800,0x7000fc92,0x7c00100,0x220400,0x7000fc92,0x7c00100, +0x230400,0x7000fc92,0x7c00100,0x250400,0x700acd00,0x4000000,0xe00000,0x700acd00,0x4000000,0x2800000,0x700ace00,0x4000000,0xe00000,0x700acf00,0x4000000,0xe00000, +0x700acf00,0x4000000,0x2800000,0x7050df11,0x4000000,0x200000,0x7050f719,0x80000,0x918820,0x7080afa1,0x2802400,0x962460,0x7090df11,0x2802400,0x962460,0x70d0e417, +0x2802100,0x962460,0x70d0e417,0x2802400,0x962460,0x70d0e417,0x6800100,0x962540,0x70d0ea15,0x4000010,0x400000,0x8000120f,0x7c00100,0x230400,0x80001524,0x7c00100, +0x230400,0x8000171a,0x7c00100,0x230400,0x80002006,0x7c00100,0x220400,0x80002006,0x7c00100,0x250400,0x80002a00,0x4000000,0x1500000,0x80002d00,0x4000000,0x200000, +0x80005208,0x2802400,0x962460,0x80005c00,0x4000000,0x200000,0x80007300,0x24000000,0x200000,0x80009519,0x7c00100,0x220400,0x80009519,0x7c00100,0x230400,0x80009519, +0x7c00100,0x250400,0x80009865,0x7c00100,0x230400,0x8000a008,0x2802100,0x962460,0x8000b30a,0x4000000,0x500000,0x8000b30a,0x7c00100,0x230400,0x8000cd00,0x4000000, +0xe00000,0x8000d202,0x2802500,0x962460,0x8000d202,0x7c00100,0x230400,0x8000d68d,0x4000000,0x200000,0x8000d997,0x2802000,0x962460,0x8000d997,0x2802400,0x962460, +0x8000d997,0x4000000,0x400000,0x8000d997,0x4000000,0x500000,0x8000d997,0x7c00100,0x230400,0x8000d997,0xc000010,0x448000,0x8000e489,0x2802100,0x962460,0x8000e489, +0x7c00100,0x230400,0x8000e719,0x7c00100,0x220400,0x8000f8a6,0x2802100,0x962460,0x8000f8a6,0x7c00100,0x230400,0x8000f8a6,0xc000010,0x448000,0x8000fda1,0x2802100, +0x1862460,0x8000fda1,0x2806400,0x1862460,0x8000fda1,0x4000000,0x1800000,0x8000fda1,0x6800000,0x1329800,0x8000fda1,0x6800100,0x1862400,0x8000fda1,0x6800100,0x1862540, +0x8000fda1,0x7c00100,0x1830000,0x8000fda1,0xc000010,0x448000,0x8000fe9c,0x7c00100,0x230400,0x8000fe9c,0x7c00100,0x830400,0x8000fe9c,0x7c00100,0x1430400,0x8000ff06, +0x7c00100,0x220400,0x80010165,0x7c00100,0x230400,0x800102a2,0x4000000,0x200000,0x800102a2,0x7c00100,0x230400,0x800103a4,0x7c00100,0x230400,0x800103a4,0xc000010, +0x448000,0x8001044c,0x4000000,0x200000,0x8001044c,0x7c00100,0x220400,0x8001044c,0x7c00100,0x250400,0x80010670,0x2802000,0x962460,0x80010670,0x4000000,0x200000, +0x80010670,0x4000010,0x400000,0x80010670,0xc000010,0x448000,0x800a4711,0x7c40300,0xe30000,0x800acd00,0x4000000,0xe00000,0x800acd00,0x4000000,0x2902460,0x800ace00, +0x4000000,0xe00000,0x800acf00,0x4000000,0xe00000,0x800b0011,0x7c40300,0xe30000,0x800b0500,0x4000000,0xe00000,0x800b0500,0x4000000,0x2800000,0x90001615,0x7c00100, +0x230400,0x9000171a,0x4000000,0x200000,0x9000171a,0x7c00100,0x230400,0x90003000,0x24000000,0x200000,0x90007f0e,0x4000000,0x200000,0x90008301,0x2802400,0x962460, +0x90008e00,0x24000000,0x400000,0x90009519,0x7c00100,0x250400,0x9000a16f,0x2802100,0x962460,0x9000d200,0x80000000,0x218960,0x9000d202,0x2802000,0x962460,0x9000d202, +0x2802100,0x962460,0x9000d202,0x7c00100,0x230400,0x9000e59d,0x2802100,0x962460,0x90010500,0x4000000,0xe00000,0x900107a7,0x2802100,0x962460,0x900107a7,0x2802400, +0x962460,0x900107a7,0x2802c00,0x962460,0x900107a7,0x4000000,0x1400000,0x900107a7,0x6800000,0x1329800,0x900107a7,0x7c00100,0x220400,0x900107a7,0x7c00100,0x250400, +0x900108a8,0x2802100,0x962460,0x900108a8,0x2806400,0x962460,0x900108a8,0x4000000,0x200000,0x900108a8,0x4000000,0x400000,0x900108a8,0x4000010,0x400000,0x900108a8, +0x6800000,0x1329800,0x900108a8,0x6800100,0x962540,0x900108a8,0x7c00100,0x230400,0x900108a8,0xc000010,0x448000,0x90010908,0x7c00100,0x220400,0x90010a38,0x2802100, +0x962460,0x90010ca9,0x2802100,0x962460,0x90010ca9,0x4000000,0x500000,0x90010ca9,0x4000010,0xb00000,0x90010ca9,0x6800100,0x962540,0x90010ca9,0x7c00100,0x230400, +0x90010d1b,0x4000000,0x500000,0x90010eaa,0x2802100,0x962460,0x90010eaa,0x2802400,0x962460,0x90010eaa,0x2806400,0x962460,0x90010eaa,0x4000000,0x200000,0x90010eaa, +0x4000000,0x400000,0x90010eaa,0x4000010,0x400000,0x90010eaa,0x6800000,0x1329800,0x90010eaa,0x6800100,0x962540,0x90010eaa,0x7c00100,0x230400,0x90010eaa,0xc000010, +0x448000,0x90010fab,0x7c00100,0x220400,0x90010fab,0x7c00100,0x250400,0x9002c300,0x4000000,0x100000,0x900ac400,0x4000000,0xe0000d,0x900acd00,0x4000000,0xe00000, +0x900acd00,0x4000000,0x2800000,0x900acf00,0x4000000,0xe00000,0x900b0500,0x4000000,0xe00000,0x900b0500,0x4000000,0x2800000,0x900b0b9a,0x7c00900,0x1230400,0x900b109a, +0x7c00300,0xe30000,0x900b119a,0x7c00300,0xe30000,0x90408e06,0x24000000,0x400000,0xa0001004,0x4000000,0x200000,0xa0001004,0x7c00100,0x230400,0xa000120f,0x2802100, +0x962460,0xa000120f,0x2802400,0x962460,0xa000171a,0x2802100,0x962460,0xa000171a,0x2806400,0x962460,0xa0002a00,0x4000000,0x1600000,0xa0003000,0x24000000,0x200000, +0xa000581e,0x7c00100,0x230400,0xa0007300,0x24000000,0x200000,0xa0008301,0x2802400,0x962460,0xa0008e00,0x24000000,0x400000,0xa000cf00,0x4000000,0xe00000,0xa0010500, +0x4000000,0x200000,0xa00114af,0x2802100,0x962460,0xa00114af,0x2802400,0x962460,0xa00114af,0x2806400,0x962460,0xa00114af,0x6800000,0x1329800,0xa00114af,0x7c00100, +0x230400,0xa00114af,0x7c00100,0x230560,0xa00116b0,0x2802100,0x962460,0xa00116b0,0x2802800,0x962460,0xa00116b0,0x2806400,0x962460,0xa00116b0,0x4000000,0x400000, +0xa00116b0,0x4000000,0x500000,0xa00116b0,0x4000010,0x400000,0xa00116b0,0x6800100,0x962540,0xa00116b0,0x7c00100,0x230400,0xa00116b0,0x7c00100,0x230560,0xa00116b0, +0xc000010,0x448000,0xa0011722,0x7c00100,0x230400,0xa00118b1,0x2802000,0x962460,0xa00118b1,0x2802100,0x962460,0xa00118b1,0x2806400,0x962460,0xa00118b1,0x4000000, +0x200000,0xa00118b1,0x4000000,0x400000,0xa00118b1,0x4000000,0x500000,0xa00118b1,0x6800100,0x962540,0xa00118b1,0x7c00100,0x230400,0xa00118b1,0x7c00100,0x230560, +0xa00118b1,0xc000010,0x448000,0xa00a4005,0x7c00100,0xe30400,0xa00a4711,0x7c40300,0xe30000,0xa00ac400,0x4000000,0xe00000,0xa00acb14,0x7c00100,0xe30000,0xa00acf00, +0x4000000,0xe00000,0xa00b0500,0x4000000,0xe00000,0xa00b0500,0x4000000,0x2800000,0xa00b0b96,0x7c00900,0x1230400,0xa00b1211,0x7c40300,0xe30000,0xa00b1314,0x7c00100, +0xe30000,0xa00b1596,0x7c00300,0xe30000,0xa040afb7,0x6800400,0x962540,0xa08083b8,0x2802400,0x962460,0xb0000a03,0x7c00100,0x220400,0xb0000b13,0x7c00100,0x2633800, +0xb0001004,0x2802000,0x962460,0xb0001110,0x4000000,0x200000,0xb0001524,0x2802100,0x962460,0xb0001615,0x4000000,0x500000,0xb000251b,0x7c00100,0x230400,0xb0007300, +0x24000000,0x200000,0xb0008939,0x4000000,0x200000,0xb0008939,0x7c00100,0x230400,0xb0008e00,0x24000000,0x200000,0xb0008e00,0x24000000,0x400000,0xb0008e00,0x24000010, +0x400000,0xb0009257,0x2802000,0x962460,0xb0009257,0x4000000,0x1600000,0xb0009519,0x7c00100,0x220400,0xb0009519,0x7c00100,0x250400,0xb0009a00,0x4000000,0x200000, +0xb000b30a,0x2802100,0x962460,0xb000b30a,0x7c00100,0x230400,0xb000c178,0x80000000,0x218960,0xb000c300,0x4000000,0x200000,0xb000d202,0x2802000,0x962460,0xb000d476, +0x6800100,0x962540,0xb000d476,0x7c00100,0x230400,0xb000e300,0x4000000,0xe00000,0xb000fda1,0x7c00100,0x1830000,0xb0010eaa,0x2802000,0x962460,0xb00116b0,0x7c00100, +0x230400,0xb0011900,0x4000000,0xe00000,0xb0011ab2,0x2802100,0x962460,0xb0011ab2,0x2802400,0x962460,0xb0011ab2,0x2806400,0x962460,0xb0011ab2,0x4000000,0x200000, +0xb0011ab2,0x6800100,0x962540,0xb0011ab2,0x7c00100,0x230400,0xb0011b0c,0x7c00100,0x230400,0xb0011cb3,0x2802100,0x962460,0xb0011cb3,0x2806400,0x962460,0xb0011cb3, +0x6800000,0x1329800,0xb0011cb3,0x6800100,0x962540,0xb0011cb3,0x7c00100,0x230400,0xb0011db6,0x2802500,0x962460,0xb0011db6,0x6800000,0x1329800,0xb0011db6,0x7c00100, +0x230400,0xb0011db6,0x7c00500,0x230400,0xb0011e00,0x4000000,0x200000,0xb0011e00,0x4000000,0x1500000,0xb0011fb4,0x2802100,0x962460,0xb0011fb4,0x6800100,0x962540, +0xb0011fb4,0x7c00100,0x230400,0xb0011fb4,0xc000010,0x248000,0xb0012000,0x4000000,0x200000,0xb00121b5,0x4000000,0x200000,0xb00121b5,0x4000010,0x400000,0xb00121b5, +0x7c00100,0x220400,0xb00121b5,0x7c00100,0x250400,0xb00121b5,0xc000010,0x448000,0xb00122b8,0x4000000,0x200000,0xb00122b8,0x7c00100,0x230400,0xb00123b7,0x2802400, +0x962460,0xb00123b7,0x4000000,0x200000,0xb00123b7,0x7c00100,0x230400,0xb00123b7,0xc000010,0x248000,0xb00a4005,0x7c00100,0xe30400,0xb00a4711,0x7c40300,0xe30000, +0xb00acf00,0x4000000,0xe00000,0xb00b0500,0x4000000,0xe00000,0xb00b0500,0x4000000,0x2800000,0xb00b109a,0x7c00300,0xe30000,0xb080e487,0x2802000,0x962460,0xc0001524, +0x4000000,0x500000,0xc0001a18,0x2806400,0x1862460,0xc0001a18,0x7c00100,0x1830000,0xc0007300,0x24000000,0x200000,0xc0008e00,0x24000010,0x400000,0xc0009519,0x7c00100, +0x220400,0xc0009519,0x7c00100,0x250400,0xc000c300,0x4000000,0x20000f,0xc000d85c,0x2802100,0x962460,0xc000d85c,0x6800100,0x962540,0xc000d85c,0x7c00100,0x230400, +0xc000dc99,0x7c00100,0x230400,0xc000e719,0x7c00100,0x220400,0xc00107a7,0x7c00100,0x230400,0xc0010eaa,0x7c00100,0x230400,0xc00116b0,0x7c00100,0x230560,0xc0011900, +0x4000000,0x200000,0xc0012447,0,0x818820,0xc0012447,0,0xc18820,0xc0012447,0,0x1418820,0xc00125b9,0x7c00100,0x230400,0xc00126bb,0x2802100, +0x962460,0xc00126bb,0x2806400,0x962460,0xc00126bb,0x4000000,0x500000,0xc00126bb,0x6800100,0x962540,0xc00126bb,0x7c00100,0x230400,0xc00127ba,0x2802400,0x962460, +0xc00127ba,0x4000000,0x200000,0xc00127ba,0x6800000,0x1329800,0xc00127ba,0x7c00100,0x230400,0xc00127ba,0x7c00900,0x230400,0xc0012800,0x4000000,0x200000,0xc0012b23, +0x4000000,0x200000,0xc0012b23,0x4000000,0x400000,0xc0012b23,0x4000000,0x1500000,0xc0012cbc,0x2802400,0x962460,0xc0012cbc,0x4000000,0x1600000,0xc0012cbc,0x6800000, +0x1329800,0xc0012cbc,0x7c00100,0x230400,0xc00acf00,0x4000000,0xe00000,0xc00ae300,0x4000000,0xe00000,0xc00b0500,0x4000000,0xe00000,0xc00b0500,0x4000000,0x2800000, +0xc00b0b11,0x4000000,0x1200000,0xc00b0b11,0x7c00900,0x1230400,0xc00b109a,0x7c00300,0xe30000,0xc00b2914,0x7c00100,0x2530000,0xc00b2916,0x7c00100,0x2530c00,0xc00b2a00, +0x4000000,0xe00000,0xc040af5e,0x7c00100,0x230400,0xc0c12b89,0x4000000,0x200000,0xc14a44ca,0x4000000,0xe0000d,0xd000131f,0x2802c00,0x962460,0xd000171a,0x7c00100, +0x230400,0xd0001821,0x2802100,0x962460,0xd0007300,0x24000000,0x200000,0xd0008e00,0x24000000,0x200000,0xd0008f3a,0x2806000,0x962460,0xd0009519,0x7c00100,0x220400, +0xd0009519,0x7c00100,0x250400,0xd000a500,0x4000000,0x200000,0xd000c300,0x4000000,0xe00000,0xd000d202,0x7c00100,0x230400,0xd000d476,0x7c00100,0x230400,0xd000d997, +0x2802100,0x962460,0xd000d997,0x6800100,0x962540,0xd000e001,0x2802100,0x962460,0xd000e700,0x4000400,0x200000,0xd000e719,0x7c00100,0x220400,0xd000e719,0x7c00500, +0x22040f,0xd000fa00,0x4000000,0xe00000,0xd0010eaa,0x4000010,0x400000,0xd0010eaa,0x7c00100,0x230400,0xd0012dbd,0x4000000,0x200000,0xd0012dbd,0x7c00100,0x230400, +0xd0012fbe,0x2802100,0x962460,0xd0012fbe,0x2802400,0x962460,0xd0012fbe,0x2806400,0x962460,0xd0012fbe,0x4000000,0x400000,0xd0012fbe,0x6800000,0x1329800,0xd0012fbe, +0x6800100,0x962540,0xd0012fbe,0x6800100,0x962541,0xd0012fbe,0x6804400,0x962540,0xd0012fbe,0x7c00100,0x230400,0xd0012fbe,0x7c00100,0x230560,0xd0012fbe,0xc000010, +0x448000,0xd0013183,0x7c00100,0x230400,0xd0013200,0x4000000,0x200000,0xd0013200,0x6800000,0x1329805,0xd00134c0,0x2802100,0x962460,0xd00134c0,0x4000002,0x400000, +0xd00134c0,0x7c00100,0x230400,0xd00a4305,0x7c00100,0xe30400,0xd00a4611,0x7c40300,0xe30000,0xd00a4711,0x7c40300,0xe30000,0xd00a5e11,0x7c40300,0xe30000,0xd00acf00, +0x4000000,0xe00000,0xd00b0500,0x4000000,0xe00000,0xd00b0500,0x4000000,0x2800000,0xd00b0b11,0x6800500,0x962540,0xd00b0bbf,0x2802200,0xc62460,0xd00b119a,0x7c00300, +0xe30000,0xd00b2a00,0x4000000,0xe00000,0xd00b2e11,0x7c40300,0xe30000,0xd00b30bf,0x7c00300,0x230000,0xd00b339a,0x7c00300,0xe30000,0xe0000c02,0xc000010,0xb48000, +0xe0001524,0x2802400,0x962460,0xe0001524,0x7c00100,0x230400,0xe0001615,0x7c00100,0x230400,0xe000251b,0x12882000,0x962460,0xe0002a00,0x4000000,0x1500000,0xe0005102, +0x4000000,0x200000,0xe0005c00,0x4000000,0x200000,0xe000622a,0x6804400,0x962540,0xe000622a,0x7c00100,0x230400,0xe0008838,0x7c00100,0x220400,0xe0008838,0x7c00100, +0x250400,0xe0008e00,0x24000000,0x810000,0xe0008e00,0x24000000,0x1410000,0xe0008e00,0x24000002,0x400000,0xe0008e00,0x2c000010,0xb48000,0xe000933e,0x7c00100,0x230400, +0xe000933e,0xc000010,0x448000,0xe0009519,0x7c00100,0x220400,0xe0009519,0x7c00100,0x22040f,0xe0009519,0x7c00100,0x250400,0xe000c178,0x2802100,0x962460,0xe000c941, +0x2802100,0x962460,0xe000c941,0x2806400,0x962460,0xe000c941,0x7c00100,0x230400,0xe000d202,0x2802400,0x962460,0xe000d202,0x7c00100,0x230400,0xe000d202,0x7c00500, +0x230400,0xe000dc99,0x4000000,0x200000,0xe000e001,0x2802100,0x962460,0xe000e001,0x2802400,0x962460,0xe000fda1,0x7c00100,0x1830000,0xe0013502,0x2802400,0x962460, +0xe0013502,0x4000000,0x200000,0xe0013502,0x7c00100,0x230400,0xe0013502,0x80000000,0x218960,0xe00136c1,0x4000000,0x200000,0xe00136c1,0x7c00100,0x230400,0xe001370b, +0x7c00100,0x230400,0xe0013919,0x7c00500,0x220400,0xe0013919,0x7c00500,0x22040f,0xe0013919,0x7c00d00,0x23040f,0xe0013a19,0x7c00100,0x220400,0xe0013a19,0x7c00100, +0x230400,0xe0013bc2,0x2802400,0x962460,0xe0013bc2,0x7c00100,0x230400,0xe0013bc2,0xc000010,0x248000,0xe0013cc3,0x6800000,0x1329800,0xe0013cc3,0x7c00100,0x230400, +0xe0013dc4,0x2802400,0x962460,0xe0013dc4,0x7c00100,0x230400,0xe0013e28,0x7c00100,0x230400,0xe0013fc5,0x7c00100,0x220400,0xe0013fc5,0x7c00100,0x250400,0xe0014000, +0x4000000,0x200000,0xe0014001,0x2802400,0x962460,0xe00a4711,0x7c40300,0xe30000,0xe00a5e11,0x7c40300,0xe30000,0xe00ac511,0x7c40300,0xe30000,0xe00acf00,0x4000000, +0xe00000,0xe00ae300,0x4000000,0xe00000,0xe00b0500,0x4000000,0xe00000,0xe00b1314,0x7c00100,0xe30000,0xe00b1316,0x7c00100,0xe30c00,0xe00b2a00,0x4000000,0xe00000, +0xe00b2a00,0x4000000,0x2800000,0xe00b3816,0x7c00500,0x230c00,0xe0808328,0x2802400,0x962460,0xf0001615,0x6800100,0x962540,0xf0001a18,0x2802000,0x1862460,0xf000c247, +0x7c00100,0x230400,0xf000d000,0x4000000,0xe00000,0xf000e300,0x4000000,0xe00000,0xf000e59d,0x2802100,0x962460,0xf000e59d,0x7c00100,0x230400,0xf0012447,0, +0x818820,0xf0012447,0,0xc18820,0xf0012447,0,0x1418820,0xf0012447,0x2802000,0x962460,0xf0012447,0x2802400,0x962460,0xf0012447,0x7c00100,0x230400, +0xf0013a19,0x7c00100,0x220400,0xf0014102,0x2802400,0x962460,0xf0014308,0x2802100,0x962460,0xf0014308,0x7c00500,0x22040e,0xf0014308,0x7c00500,0x22040f,0xf001440a, +0x4000000,0x500000,0xf0014500,0x4000000,0x200000,0xf00146c6,0x2802100,0x962460,0xf00146c6,0x2806000,0x962460,0xf00146c6,0x4000000,0xe00000,0xf00146c6,0x6800000, +0x1329800,0xf00146c6,0x6800100,0x962540,0xf00146c6,0x6804000,0x962540,0xf00146c6,0x7c00100,0x230400,0xf00146c6,0x7c00100,0x230560,0xf00146c6,0xc000010,0x448000, +0xf00147c7,0x2802000,0x962460,0xf00147c7,0x6800000,0x1329800,0xf00147c7,0x7c00100,0x230400,0xf00ac511,0x7c40300,0xe30000,0xf00acf00,0x4000000,0xe00000,0xf00b2914, +0x7c00100,0x2530000,0xf00b2916,0x7c00100,0x2530c00,0xf00b2a00,0x4000000,0xe00000,0xf00b2a00,0x4000000,0x2800000,0xf00b4211,0x7c40300,0xe30000}; + +static const int32_t countPropsVectors=7230; +static const int32_t propsVectorsColumns=3; +static const uint16_t scriptExtensions[282]={ +0x800e,0x8019,8,0x8059,8,2,8,0x8038,8,6,8,0x8019,2,0x22,0x25,0x57, +0xb6,0x80c0,2,0x22,0x8025,2,0x12,2,0x22,0x25,0x57,0xa7,0xb6,0x80c0,2,0x22, +0x54,0x79,0x7b,0xa7,0xb6,0xb7,0x80c2,2,0x8022,2,0x25,0x80c0,2,0x29,2,0x80b6, +2,0x2e,4,0xa,0xf,0x10,0x15,0x19,0x1a,0x1f,0x23,0x24,0x89,0x97,0x809e,4, +0xa,0xf,0x10,0x15,0x19,0x1a,0x1f,0x23,0x24,0x89,0x809e,4,0xa,0xf,0x10,0x15, +0x1a,0x1f,0x21,0x23,0x24,0x3a,0x89,0x91,0x99,0x9e,0xa0,0xaf,0xb2,0xb3,0x80bb,4, +0xa,0xf,0x10,0x15,0x1a,0x1f,0x21,0x23,0x24,0x30,0x3a,0x89,0x91,0x99,0x9e,0xa0, +0xaf,0xb2,0xb3,0x80bb,0xa,0x78,0xa0,0x80b2,0xa,0x74,4,0x3a,0x8076,4,0x7a,0x10, +0x80a4,0x10,0x7f,0xf,0x809d,0xf,0x83,0x23,0x8089,0x23,0x87,0x15,0x80bb,0x15,0x8b,0x1c, +0x34,0x8076,0x1c,0x8f,0xc,0x8019,0x2a,0x2b,0x2c,0x802d,0x1b,0x805a,0x800a,4,0xa,0x15, +0x8089,0xa,0x8089,4,0x800a,0xa,0x8097,0xa,0x15,0x1a,0x1f,0x23,0x8024,0xa,0x80bb,4, +0xa,0x15,0x1f,0x24,0x89,0x9e,0x80bb,0x8004,8,0x8022,0x19,0x801b,0xa,0x19,0x8089,5, +0x11,0x12,0x14,0x16,0x8029,5,0x11,0x12,0x14,0x8016,0x8011,5,0x8011,0x11,0x14,0x8016, +0x11,0x8019,0xa,0xf,0x10,0x78,0x91,0x99,0x9d,0x9e,0xa0,0xa3,0x80b2,0xa,0xf,0x10, +0x15,0x1a,0x78,0x91,0x99,0x9d,0x9e,0xa0,0xa3,0xb2,0x80bb,0xa,0xf,0x10,0x15,0x78, +0x91,0x99,0x9d,0x9e,0xa0,0xa3,0xb2,0x80bb,0xa,0xa3,0xa,0x8023,0xa,0xfa,0x19,0x1c, +0x804f,0x37,0x804e,2,0x8057,2,0x8025,2,0x105,0x2f,0x31,0x8053,0x2f,0x31,0x80c1,0x2f, +0x8031,2,0x8007,0x79,0x80c2,0x79,0x113,0x89,0x87,0x8087}; + +static const int32_t indexes[UPROPS_INDEX_COUNT]={0x2d08,0x2d08,0x2d08,0x2d08,0x6ce6,3,0x8924,0x89b1,0x89b1,0x89b1,0xb47c7,0x2a75a31,0,0,0,0}; + +#endif // INCLUDED_FROM_UCHAR_C diff --git a/intl/icu/source/common/ucharstrie.cpp b/intl/icu/source/common/ucharstrie.cpp new file mode 100644 index 0000000000..ba9cea7ba3 --- /dev/null +++ b/intl/icu/source/common/ucharstrie.cpp @@ -0,0 +1,414 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2011, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: ucharstrie.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010nov14 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" +#include "unicode/appendable.h" +#include "unicode/ucharstrie.h" +#include "unicode/uobject.h" +#include "unicode/utf16.h" +#include "cmemory.h" +#include "uassert.h" + +U_NAMESPACE_BEGIN + +UCharsTrie::~UCharsTrie() { + uprv_free(ownedArray_); +} + +UStringTrieResult +UCharsTrie::current() const { + const char16_t *pos=pos_; + if(pos==nullptr) { + return USTRINGTRIE_NO_MATCH; + } else { + int32_t node; + return (remainingMatchLength_<0 && (node=*pos)>=kMinValueLead) ? + valueResult(node) : USTRINGTRIE_NO_VALUE; + } +} + +UStringTrieResult +UCharsTrie::firstForCodePoint(UChar32 cp) { + return cp<=0xffff ? + first(cp) : + (USTRINGTRIE_HAS_NEXT(first(U16_LEAD(cp))) ? + next(U16_TRAIL(cp)) : + USTRINGTRIE_NO_MATCH); +} + +UStringTrieResult +UCharsTrie::nextForCodePoint(UChar32 cp) { + return cp<=0xffff ? + next(cp) : + (USTRINGTRIE_HAS_NEXT(next(U16_LEAD(cp))) ? + next(U16_TRAIL(cp)) : + USTRINGTRIE_NO_MATCH); +} + +UStringTrieResult +UCharsTrie::branchNext(const char16_t *pos, int32_t length, int32_t uchar) { + // Branch according to the current unit. + if(length==0) { + length=*pos++; + } + ++length; + // The length of the branch is the number of units to select from. + // The data structure encodes a binary search. + while(length>kMaxBranchLinearSubNodeLength) { + if(uchar<*pos++) { + length>>=1; + pos=jumpByDelta(pos); + } else { + length=length-(length>>1); + pos=skipDelta(pos); + } + } + // Drop down to linear search for the last few units. + // length>=2 because the loop body above sees length>kMaxBranchLinearSubNodeLength>=3 + // and divides length by 2. + do { + if(uchar==*pos++) { + UStringTrieResult result; + int32_t node=*pos; + if(node&kValueIsFinal) { + // Leave the final value for getValue() to read. + result=USTRINGTRIE_FINAL_VALUE; + } else { + // Use the non-final value as the jump delta. + ++pos; + // int32_t delta=readValue(pos, node); + int32_t delta; + if(node=kMinValueLead ? valueResult(node) : USTRINGTRIE_NO_VALUE; + } + pos_=pos; + return result; + } + --length; + pos=skipValue(pos); + } while(length>1); + if(uchar==*pos++) { + pos_=pos; + int32_t node=*pos; + return node>=kMinValueLead ? valueResult(node) : USTRINGTRIE_NO_VALUE; + } else { + stop(); + return USTRINGTRIE_NO_MATCH; + } +} + +UStringTrieResult +UCharsTrie::nextImpl(const char16_t *pos, int32_t uchar) { + int32_t node=*pos++; + for(;;) { + if(node=kMinValueLead) ? + valueResult(node) : USTRINGTRIE_NO_VALUE; + } else { + // No match. + break; + } + } else if(node&kValueIsFinal) { + // No further matching units. + break; + } else { + // Skip intermediate value. + pos=skipNodeValue(pos, node); + node&=kNodeTypeMask; + } + } + stop(); + return USTRINGTRIE_NO_MATCH; +} + +UStringTrieResult +UCharsTrie::next(int32_t uchar) { + const char16_t *pos=pos_; + if(pos==nullptr) { + return USTRINGTRIE_NO_MATCH; + } + int32_t length=remainingMatchLength_; // Actual remaining match length minus 1. + if(length>=0) { + // Remaining part of a linear-match node. + if(uchar==*pos++) { + remainingMatchLength_=--length; + pos_=pos; + int32_t node; + return (length<0 && (node=*pos)>=kMinValueLead) ? + valueResult(node) : USTRINGTRIE_NO_VALUE; + } else { + stop(); + return USTRINGTRIE_NO_MATCH; + } + } + return nextImpl(pos, uchar); +} + +UStringTrieResult +UCharsTrie::next(ConstChar16Ptr ptr, int32_t sLength) { + const char16_t *s=ptr; + if(sLength<0 ? *s==0 : sLength==0) { + // Empty input. + return current(); + } + const char16_t *pos=pos_; + if(pos==nullptr) { + return USTRINGTRIE_NO_MATCH; + } + int32_t length=remainingMatchLength_; // Actual remaining match length minus 1. + for(;;) { + // Fetch the next input unit, if there is one. + // Continue a linear-match node without rechecking sLength<0. + int32_t uchar; + if(sLength<0) { + for(;;) { + if((uchar=*s++)==0) { + remainingMatchLength_=length; + pos_=pos; + int32_t node; + return (length<0 && (node=*pos)>=kMinValueLead) ? + valueResult(node) : USTRINGTRIE_NO_VALUE; + } + if(length<0) { + remainingMatchLength_=length; + break; + } + if(uchar!=*pos) { + stop(); + return USTRINGTRIE_NO_MATCH; + } + ++pos; + --length; + } + } else { + for(;;) { + if(sLength==0) { + remainingMatchLength_=length; + pos_=pos; + int32_t node; + return (length<0 && (node=*pos)>=kMinValueLead) ? + valueResult(node) : USTRINGTRIE_NO_VALUE; + } + uchar=*s++; + --sLength; + if(length<0) { + remainingMatchLength_=length; + break; + } + if(uchar!=*pos) { + stop(); + return USTRINGTRIE_NO_MATCH; + } + ++pos; + --length; + } + } + int32_t node=*pos++; + for(;;) { + if(nodekMaxBranchLinearSubNodeLength) { + ++pos; // ignore the comparison unit + if(nullptr==findUniqueValueFromBranch(jumpByDelta(pos), length>>1, haveUniqueValue, uniqueValue)) { + return nullptr; + } + length=length-(length>>1); + pos=skipDelta(pos); + } + do { + ++pos; // ignore a comparison unit + // handle its value + int32_t node=*pos++; + UBool isFinal=(UBool)(node>>15); + node&=0x7fff; + int32_t value=readValue(pos, node); + pos=skipValue(pos, node); + if(isFinal) { + if(haveUniqueValue) { + if(value!=uniqueValue) { + return nullptr; + } + } else { + uniqueValue=value; + haveUniqueValue=true; + } + } else { + if(!findUniqueValue(pos+value, haveUniqueValue, uniqueValue)) { + return nullptr; + } + haveUniqueValue=true; + } + } while(--length>1); + return pos+1; // ignore the last comparison unit +} + +UBool +UCharsTrie::findUniqueValue(const char16_t *pos, UBool haveUniqueValue, int32_t &uniqueValue) { + int32_t node=*pos++; + for(;;) { + if(node>15); + int32_t value; + if(isFinal) { + value=readValue(pos, node&0x7fff); + } else { + value=readNodeValue(pos, node); + } + if(haveUniqueValue) { + if(value!=uniqueValue) { + return false; + } + } else { + uniqueValue=value; + haveUniqueValue=true; + } + if(isFinal) { + return true; + } + pos=skipNodeValue(pos, node); + node&=kNodeTypeMask; + } + } +} + +int32_t +UCharsTrie::getNextUChars(Appendable &out) const { + const char16_t *pos=pos_; + if(pos==nullptr) { + return 0; + } + if(remainingMatchLength_>=0) { + out.appendCodeUnit(*pos); // Next unit of a pending linear-match node. + return 1; + } + int32_t node=*pos++; + if(node>=kMinValueLead) { + if(node&kValueIsFinal) { + return 0; + } else { + pos=skipNodeValue(pos, node); + node&=kNodeTypeMask; + } + } + if(nodekMaxBranchLinearSubNodeLength) { + ++pos; // ignore the comparison unit + getNextBranchUChars(jumpByDelta(pos), length>>1, out); + length=length-(length>>1); + pos=skipDelta(pos); + } + do { + out.appendCodeUnit(*pos++); + pos=skipValue(pos); + } while(--length>1); + out.appendCodeUnit(*pos); +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/ucharstriebuilder.cpp b/intl/icu/source/common/ucharstriebuilder.cpp new file mode 100644 index 0000000000..95b32711a8 --- /dev/null +++ b/intl/icu/source/common/ucharstriebuilder.cpp @@ -0,0 +1,443 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2012, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: ucharstriebuilder.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010nov14 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" +#include "unicode/ucharstrie.h" +#include "unicode/ucharstriebuilder.h" +#include "unicode/unistr.h" +#include "unicode/ustring.h" +#include "cmemory.h" +#include "uarrsort.h" +#include "uassert.h" +#include "uhash.h" +#include "ustr_imp.h" + +U_NAMESPACE_BEGIN + +/* + * Note: This builder implementation stores (string, value) pairs with full copies + * of the 16-bit-unit sequences, until the UCharsTrie is built. + * It might(!) take less memory if we collected the data in a temporary, dynamic trie. + */ + +class UCharsTrieElement : public UMemory { +public: + // Use compiler's default constructor, initializes nothing. + + void setTo(const UnicodeString &s, int32_t val, UnicodeString &strings, UErrorCode &errorCode); + + UnicodeString getString(const UnicodeString &strings) const { + int32_t length=strings[stringOffset]; + return strings.tempSubString(stringOffset+1, length); + } + int32_t getStringLength(const UnicodeString &strings) const { + return strings[stringOffset]; + } + + char16_t charAt(int32_t index, const UnicodeString &strings) const { + return strings[stringOffset+1+index]; + } + + int32_t getValue() const { return value; } + + int32_t compareStringTo(const UCharsTrieElement &o, const UnicodeString &strings) const; + +private: + // The first strings unit contains the string length. + // (Compared with a stringLength field here, this saves 2 bytes per string.) + int32_t stringOffset; + int32_t value; +}; + +void +UCharsTrieElement::setTo(const UnicodeString &s, int32_t val, + UnicodeString &strings, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return; + } + int32_t length=s.length(); + if(length>0xffff) { + // Too long: We store the length in 1 unit. + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return; + } + stringOffset=strings.length(); + strings.append((char16_t)length); + value=val; + strings.append(s); +} + +int32_t +UCharsTrieElement::compareStringTo(const UCharsTrieElement &other, const UnicodeString &strings) const { + return getString(strings).compare(other.getString(strings)); +} + +UCharsTrieBuilder::UCharsTrieBuilder(UErrorCode & /*errorCode*/) + : elements(nullptr), elementsCapacity(0), elementsLength(0), + uchars(nullptr), ucharsCapacity(0), ucharsLength(0) {} + +UCharsTrieBuilder::~UCharsTrieBuilder() { + delete[] elements; + uprv_free(uchars); +} + +UCharsTrieBuilder & +UCharsTrieBuilder::add(const UnicodeString &s, int32_t value, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return *this; + } + if(ucharsLength>0) { + // Cannot add elements after building. + errorCode=U_NO_WRITE_PERMISSION; + return *this; + } + if(elementsLength==elementsCapacity) { + int32_t newCapacity; + if(elementsCapacity==0) { + newCapacity=1024; + } else { + newCapacity=4*elementsCapacity; + } + UCharsTrieElement *newElements=new UCharsTrieElement[newCapacity]; + if(newElements==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return *this; + } + if(elementsLength>0) { + uprv_memcpy(newElements, elements, (size_t)elementsLength*sizeof(UCharsTrieElement)); + } + delete[] elements; + elements=newElements; + elementsCapacity=newCapacity; + } + elements[elementsLength++].setTo(s, value, strings, errorCode); + if(U_SUCCESS(errorCode) && strings.isBogus()) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } + return *this; +} + +U_CDECL_BEGIN + +static int32_t U_CALLCONV +compareElementStrings(const void *context, const void *left, const void *right) { + const UnicodeString *strings=static_cast(context); + const UCharsTrieElement *leftElement=static_cast(left); + const UCharsTrieElement *rightElement=static_cast(right); + return leftElement->compareStringTo(*rightElement, *strings); +} + +U_CDECL_END + +UCharsTrie * +UCharsTrieBuilder::build(UStringTrieBuildOption buildOption, UErrorCode &errorCode) { + buildUChars(buildOption, errorCode); + UCharsTrie *newTrie=nullptr; + if(U_SUCCESS(errorCode)) { + newTrie=new UCharsTrie(uchars, uchars+(ucharsCapacity-ucharsLength)); + if(newTrie==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } else { + uchars=nullptr; // The new trie now owns the array. + ucharsCapacity=0; + } + } + return newTrie; +} + +UnicodeString & +UCharsTrieBuilder::buildUnicodeString(UStringTrieBuildOption buildOption, UnicodeString &result, + UErrorCode &errorCode) { + buildUChars(buildOption, errorCode); + if(U_SUCCESS(errorCode)) { + result.setTo(false, uchars+(ucharsCapacity-ucharsLength), ucharsLength); + } + return result; +} + +void +UCharsTrieBuilder::buildUChars(UStringTrieBuildOption buildOption, UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return; + } + if(uchars!=nullptr && ucharsLength>0) { + // Already built. + return; + } + if(ucharsLength==0) { + if(elementsLength==0) { + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return; + } + if(strings.isBogus()) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + uprv_sortArray(elements, elementsLength, (int32_t)sizeof(UCharsTrieElement), + compareElementStrings, &strings, + false, // need not be a stable sort + &errorCode); + if(U_FAILURE(errorCode)) { + return; + } + // Duplicate strings are not allowed. + UnicodeString prev=elements[0].getString(strings); + for(int32_t i=1; i(uprv_malloc(capacity*2)); + if(uchars==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + ucharsCapacity=0; + return; + } + ucharsCapacity=capacity; + } + StringTrieBuilder::build(buildOption, elementsLength, errorCode); + if(uchars==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } +} + +int32_t +UCharsTrieBuilder::getElementStringLength(int32_t i) const { + return elements[i].getStringLength(strings); +} + +char16_t +UCharsTrieBuilder::getElementUnit(int32_t i, int32_t unitIndex) const { + return elements[i].charAt(unitIndex, strings); +} + +int32_t +UCharsTrieBuilder::getElementValue(int32_t i) const { + return elements[i].getValue(); +} + +int32_t +UCharsTrieBuilder::getLimitOfLinearMatch(int32_t first, int32_t last, int32_t unitIndex) const { + const UCharsTrieElement &firstElement=elements[first]; + const UCharsTrieElement &lastElement=elements[last]; + int32_t minStringLength=firstElement.getStringLength(strings); + while(++unitIndex0); + return i; +} + +int32_t +UCharsTrieBuilder::indexOfElementWithNextUnit(int32_t i, int32_t unitIndex, char16_t unit) const { + while(unit==elements[i].charAt(unitIndex, strings)) { + ++i; + } + return i; +} + +UCharsTrieBuilder::UCTLinearMatchNode::UCTLinearMatchNode(const char16_t *units, int32_t len, Node *nextNode) + : LinearMatchNode(len, nextNode), s(units) { + hash=hash*37u+ustr_hashUCharsN(units, len); +} + +bool +UCharsTrieBuilder::UCTLinearMatchNode::operator==(const Node &other) const { + if(this==&other) { + return true; + } + if(!LinearMatchNode::operator==(other)) { + return false; + } + const UCTLinearMatchNode &o=static_cast(other); + return 0==u_memcmp(s, o.s, length); +} + +void +UCharsTrieBuilder::UCTLinearMatchNode::write(StringTrieBuilder &builder) { + UCharsTrieBuilder &b=(UCharsTrieBuilder &)builder; + next->write(builder); + b.write(s, length); + offset=b.writeValueAndType(hasValue, value, b.getMinLinearMatch()+length-1); +} + +StringTrieBuilder::Node * +UCharsTrieBuilder::createLinearMatchNode(int32_t i, int32_t unitIndex, int32_t length, + Node *nextNode) const { + return new UCTLinearMatchNode( + elements[i].getString(strings).getBuffer()+unitIndex, + length, + nextNode); +} + +UBool +UCharsTrieBuilder::ensureCapacity(int32_t length) { + if(uchars==nullptr) { + return false; // previous memory allocation had failed + } + if(length>ucharsCapacity) { + int32_t newCapacity=ucharsCapacity; + do { + newCapacity*=2; + } while(newCapacity<=length); + char16_t *newUChars=static_cast(uprv_malloc(newCapacity*2)); + if(newUChars==nullptr) { + // unable to allocate memory + uprv_free(uchars); + uchars=nullptr; + ucharsCapacity=0; + return false; + } + u_memcpy(newUChars+(newCapacity-ucharsLength), + uchars+(ucharsCapacity-ucharsLength), ucharsLength); + uprv_free(uchars); + uchars=newUChars; + ucharsCapacity=newCapacity; + } + return true; +} + +int32_t +UCharsTrieBuilder::write(int32_t unit) { + int32_t newLength=ucharsLength+1; + if(ensureCapacity(newLength)) { + ucharsLength=newLength; + uchars[ucharsCapacity-ucharsLength]=(char16_t)unit; + } + return ucharsLength; +} + +int32_t +UCharsTrieBuilder::write(const char16_t *s, int32_t length) { + int32_t newLength=ucharsLength+length; + if(ensureCapacity(newLength)) { + ucharsLength=newLength; + u_memcpy(uchars+(ucharsCapacity-ucharsLength), s, length); + } + return ucharsLength; +} + +int32_t +UCharsTrieBuilder::writeElementUnits(int32_t i, int32_t unitIndex, int32_t length) { + return write(elements[i].getString(strings).getBuffer()+unitIndex, length); +} + +int32_t +UCharsTrieBuilder::writeValueAndFinal(int32_t i, UBool isFinal) { + if(0<=i && i<=UCharsTrie::kMaxOneUnitValue) { + return write(i|(isFinal<<15)); + } + char16_t intUnits[3]; + int32_t length; + if(i<0 || i>UCharsTrie::kMaxTwoUnitValue) { + intUnits[0]=(char16_t)(UCharsTrie::kThreeUnitValueLead); + intUnits[1]=(char16_t)((uint32_t)i>>16); + intUnits[2]=(char16_t)i; + length=3; + // } else if(i<=UCharsTrie::kMaxOneUnitValue) { + // intUnits[0]=(char16_t)(i); + // length=1; + } else { + intUnits[0]=(char16_t)(UCharsTrie::kMinTwoUnitValueLead+(i>>16)); + intUnits[1]=(char16_t)i; + length=2; + } + intUnits[0]=(char16_t)(intUnits[0]|(isFinal<<15)); + return write(intUnits, length); +} + +int32_t +UCharsTrieBuilder::writeValueAndType(UBool hasValue, int32_t value, int32_t node) { + if(!hasValue) { + return write(node); + } + char16_t intUnits[3]; + int32_t length; + if(value<0 || value>UCharsTrie::kMaxTwoUnitNodeValue) { + intUnits[0]=(char16_t)(UCharsTrie::kThreeUnitNodeValueLead); + intUnits[1]=(char16_t)((uint32_t)value>>16); + intUnits[2]=(char16_t)value; + length=3; + } else if(value<=UCharsTrie::kMaxOneUnitNodeValue) { + intUnits[0]=(char16_t)((value+1)<<6); + length=1; + } else { + intUnits[0]=(char16_t)(UCharsTrie::kMinTwoUnitNodeValueLead+((value>>10)&0x7fc0)); + intUnits[1]=(char16_t)value; + length=2; + } + intUnits[0]|=(char16_t)node; + return write(intUnits, length); +} + +int32_t +UCharsTrieBuilder::writeDeltaTo(int32_t jumpTarget) { + int32_t i=ucharsLength-jumpTarget; + U_ASSERT(i>=0); + if(i<=UCharsTrie::kMaxOneUnitDelta) { + return write(i); + } + char16_t intUnits[3]; + int32_t length; + if(i<=UCharsTrie::kMaxTwoUnitDelta) { + intUnits[0]=(char16_t)(UCharsTrie::kMinTwoUnitDeltaLead+(i>>16)); + length=1; + } else { + intUnits[0]=(char16_t)(UCharsTrie::kThreeUnitDeltaLead); + intUnits[1]=(char16_t)(i>>16); + length=2; + } + intUnits[length++]=(char16_t)i; + return write(intUnits, length); +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/ucharstrieiterator.cpp b/intl/icu/source/common/ucharstrieiterator.cpp new file mode 100644 index 0000000000..176aed6825 --- /dev/null +++ b/intl/icu/source/common/ucharstrieiterator.cpp @@ -0,0 +1,215 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2011, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: ucharstrieiterator.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010nov15 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" +#include "unicode/ucharstrie.h" +#include "unicode/unistr.h" +#include "uvectr32.h" + +U_NAMESPACE_BEGIN + +UCharsTrie::Iterator::Iterator(ConstChar16Ptr trieUChars, int32_t maxStringLength, + UErrorCode &errorCode) + : uchars_(trieUChars), + pos_(uchars_), initialPos_(uchars_), + remainingMatchLength_(-1), initialRemainingMatchLength_(-1), + skipValue_(false), + maxLength_(maxStringLength), value_(0), stack_(nullptr) { + if(U_FAILURE(errorCode)) { + return; + } + // stack_ is a pointer so that it's easy to turn ucharstrie.h into + // a public API header for which we would want it to depend only on + // other public headers. + // Unlike UCharsTrie itself, its Iterator performs memory allocations anyway + // via the UnicodeString and UVector32 implementations, so this additional + // cost is minimal. + stack_=new UVector32(errorCode); + if(stack_==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } +} + +UCharsTrie::Iterator::Iterator(const UCharsTrie &trie, int32_t maxStringLength, + UErrorCode &errorCode) + : uchars_(trie.uchars_), pos_(trie.pos_), initialPos_(trie.pos_), + remainingMatchLength_(trie.remainingMatchLength_), + initialRemainingMatchLength_(trie.remainingMatchLength_), + skipValue_(false), + maxLength_(maxStringLength), value_(0), stack_(nullptr) { + if(U_FAILURE(errorCode)) { + return; + } + stack_=new UVector32(errorCode); + if(U_FAILURE(errorCode)) { + return; + } + if(stack_==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + int32_t length=remainingMatchLength_; // Actual remaining match length minus 1. + if(length>=0) { + // Pending linear-match node, append remaining UChars to str_. + ++length; + if(maxLength_>0 && length>maxLength_) { + length=maxLength_; // This will leave remainingMatchLength>=0 as a signal. + } + str_.append(pos_, length); + pos_+=length; + remainingMatchLength_-=length; + } +} + +UCharsTrie::Iterator::~Iterator() { + delete stack_; +} + +UCharsTrie::Iterator & +UCharsTrie::Iterator::reset() { + pos_=initialPos_; + remainingMatchLength_=initialRemainingMatchLength_; + skipValue_=false; + int32_t length=remainingMatchLength_+1; // Remaining match length. + if(maxLength_>0 && length>maxLength_) { + length=maxLength_; + } + str_.truncate(length); + pos_+=length; + remainingMatchLength_-=length; + stack_->setSize(0); + return *this; +} + +UBool +UCharsTrie::Iterator::hasNext() const { return pos_!=nullptr || !stack_->isEmpty(); } + +UBool +UCharsTrie::Iterator::next(UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { + return false; + } + const char16_t *pos=pos_; + if(pos==nullptr) { + if(stack_->isEmpty()) { + return false; + } + // Pop the state off the stack and continue with the next outbound edge of + // the branch node. + int32_t stackSize=stack_->size(); + int32_t length=stack_->elementAti(stackSize-1); + pos=uchars_+stack_->elementAti(stackSize-2); + stack_->setSize(stackSize-2); + str_.truncate(length&0xffff); + length=(int32_t)((uint32_t)length>>16); + if(length>1) { + pos=branchNext(pos, length, errorCode); + if(pos==nullptr) { + return true; // Reached a final value. + } + } else { + str_.append(*pos++); + } + } + if(remainingMatchLength_>=0) { + // We only get here if we started in a pending linear-match node + // with more than maxLength remaining units. + return truncateAndStop(); + } + for(;;) { + int32_t node=*pos++; + if(node>=kMinValueLead) { + if(skipValue_) { + pos=skipNodeValue(pos, node); + node&=kNodeTypeMask; + skipValue_=false; + } else { + // Deliver value for the string so far. + UBool isFinal=(UBool)(node>>15); + if(isFinal) { + value_=readValue(pos, node&0x7fff); + } else { + value_=readNodeValue(pos, node); + } + if(isFinal || (maxLength_>0 && str_.length()==maxLength_)) { + pos_=nullptr; + } else { + // We cannot skip the value right here because it shares its + // lead unit with a match node which we have to evaluate + // next time. + // Instead, keep pos_ on the node lead unit itself. + pos_=pos-1; + skipValue_=true; + } + return true; + } + } + if(maxLength_>0 && str_.length()==maxLength_) { + return truncateAndStop(); + } + if(node0 && str_.length()+length>maxLength_) { + str_.append(pos, maxLength_-str_.length()); + return truncateAndStop(); + } + str_.append(pos, length); + pos+=length; + } + } +} + +// Branch node, needs to take the first outbound edge and push state for the rest. +const char16_t * +UCharsTrie::Iterator::branchNext(const char16_t *pos, int32_t length, UErrorCode &errorCode) { + while(length>kMaxBranchLinearSubNodeLength) { + ++pos; // ignore the comparison unit + // Push state for the greater-or-equal edge. + stack_->addElement((int32_t)(skipDelta(pos)-uchars_), errorCode); + stack_->addElement(((length-(length>>1))<<16)|str_.length(), errorCode); + // Follow the less-than edge. + length>>=1; + pos=jumpByDelta(pos); + } + // List of key-value pairs where values are either final values or jump deltas. + // Read the first (key, value) pair. + char16_t trieUnit=*pos++; + int32_t node=*pos++; + UBool isFinal=(UBool)(node>>15); + int32_t value=readValue(pos, node&=0x7fff); + pos=skipValue(pos, node); + stack_->addElement((int32_t)(pos-uchars_), errorCode); + stack_->addElement(((length-1)<<16)|str_.length(), errorCode); + str_.append(trieUnit); + if(isFinal) { + pos_=nullptr; + value_=value; + return nullptr; + } else { + return pos+value; + } +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/uchriter.cpp b/intl/icu/source/common/uchriter.cpp new file mode 100644 index 0000000000..a7d30977e0 --- /dev/null +++ b/intl/icu/source/common/uchriter.cpp @@ -0,0 +1,367 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 1998-2012, International Business Machines Corporation and +* others. All Rights Reserved. +****************************************************************************** +*/ + +#include "utypeinfo.h" // for 'typeid' to work + +#include "unicode/uchriter.h" +#include "unicode/ustring.h" +#include "unicode/utf16.h" +#include "ustr_imp.h" + +U_NAMESPACE_BEGIN + +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(UCharCharacterIterator) + +UCharCharacterIterator::UCharCharacterIterator() + : CharacterIterator(), + text(0) +{ + // never default construct! +} + +UCharCharacterIterator::UCharCharacterIterator(ConstChar16Ptr textPtr, + int32_t length) + : CharacterIterator(textPtr != 0 ? (length>=0 ? length : u_strlen(textPtr)) : 0), + text(textPtr) +{ +} + +UCharCharacterIterator::UCharCharacterIterator(ConstChar16Ptr textPtr, + int32_t length, + int32_t position) + : CharacterIterator(textPtr != 0 ? (length>=0 ? length : u_strlen(textPtr)) : 0, position), + text(textPtr) +{ +} + +UCharCharacterIterator::UCharCharacterIterator(ConstChar16Ptr textPtr, + int32_t length, + int32_t textBegin, + int32_t textEnd, + int32_t position) + : CharacterIterator(textPtr != 0 ? (length>=0 ? length : u_strlen(textPtr)) : 0, textBegin, textEnd, position), + text(textPtr) +{ +} + +UCharCharacterIterator::UCharCharacterIterator(const UCharCharacterIterator& that) +: CharacterIterator(that), + text(that.text) +{ +} + +UCharCharacterIterator& +UCharCharacterIterator::operator=(const UCharCharacterIterator& that) { + CharacterIterator::operator=(that); + text = that.text; + return *this; +} + +UCharCharacterIterator::~UCharCharacterIterator() { +} + +bool +UCharCharacterIterator::operator==(const ForwardCharacterIterator& that) const { + if (this == &that) { + return true; + } + if (typeid(*this) != typeid(that)) { + return false; + } + + const UCharCharacterIterator& realThat = static_cast(that); + + return text == realThat.text + && textLength == realThat.textLength + && pos == realThat.pos + && begin == realThat.begin + && end == realThat.end; +} + +int32_t +UCharCharacterIterator::hashCode() const { + return ustr_hashUCharsN(text, textLength) ^ pos ^ begin ^ end; +} + +UCharCharacterIterator* +UCharCharacterIterator::clone() const { + return new UCharCharacterIterator(*this); +} + +char16_t +UCharCharacterIterator::first() { + pos = begin; + if(pos < end) { + return text[pos]; + } else { + return DONE; + } +} + +char16_t +UCharCharacterIterator::firstPostInc() { + pos = begin; + if(pos < end) { + return text[pos++]; + } else { + return DONE; + } +} + +char16_t +UCharCharacterIterator::last() { + pos = end; + if(pos > begin) { + return text[--pos]; + } else { + return DONE; + } +} + +char16_t +UCharCharacterIterator::setIndex(int32_t position) { + if(position < begin) { + pos = begin; + } else if(position > end) { + pos = end; + } else { + pos = position; + } + if(pos < end) { + return text[pos]; + } else { + return DONE; + } +} + +char16_t +UCharCharacterIterator::current() const { + if (pos >= begin && pos < end) { + return text[pos]; + } else { + return DONE; + } +} + +char16_t +UCharCharacterIterator::next() { + if (pos + 1 < end) { + return text[++pos]; + } else { + /* make current() return DONE */ + pos = end; + return DONE; + } +} + +char16_t +UCharCharacterIterator::nextPostInc() { + if (pos < end) { + return text[pos++]; + } else { + return DONE; + } +} + +UBool +UCharCharacterIterator::hasNext() { + return (UBool)(pos < end ? true : false); +} + +char16_t +UCharCharacterIterator::previous() { + if (pos > begin) { + return text[--pos]; + } else { + return DONE; + } +} + +UBool +UCharCharacterIterator::hasPrevious() { + return (UBool)(pos > begin ? true : false); +} + +UChar32 +UCharCharacterIterator::first32() { + pos = begin; + if(pos < end) { + int32_t i = pos; + UChar32 c; + U16_NEXT(text, i, end, c); + return c; + } else { + return DONE; + } +} + +UChar32 +UCharCharacterIterator::first32PostInc() { + pos = begin; + if(pos < end) { + UChar32 c; + U16_NEXT(text, pos, end, c); + return c; + } else { + return DONE; + } +} + +UChar32 +UCharCharacterIterator::last32() { + pos = end; + if(pos > begin) { + UChar32 c; + U16_PREV(text, begin, pos, c); + return c; + } else { + return DONE; + } +} + +UChar32 +UCharCharacterIterator::setIndex32(int32_t position) { + if(position < begin) { + position = begin; + } else if(position > end) { + position = end; + } + if(position < end) { + U16_SET_CP_START(text, begin, position); + int32_t i = this->pos = position; + UChar32 c; + U16_NEXT(text, i, end, c); + return c; + } else { + this->pos = position; + return DONE; + } +} + +UChar32 +UCharCharacterIterator::current32() const { + if (pos >= begin && pos < end) { + UChar32 c; + U16_GET(text, begin, pos, end, c); + return c; + } else { + return DONE; + } +} + +UChar32 +UCharCharacterIterator::next32() { + if (pos < end) { + U16_FWD_1(text, pos, end); + if(pos < end) { + int32_t i = pos; + UChar32 c; + U16_NEXT(text, i, end, c); + return c; + } + } + /* make current() return DONE */ + pos = end; + return DONE; +} + +UChar32 +UCharCharacterIterator::next32PostInc() { + if (pos < end) { + UChar32 c; + U16_NEXT(text, pos, end, c); + return c; + } else { + return DONE; + } +} + +UChar32 +UCharCharacterIterator::previous32() { + if (pos > begin) { + UChar32 c; + U16_PREV(text, begin, pos, c); + return c; + } else { + return DONE; + } +} + +int32_t +UCharCharacterIterator::move(int32_t delta, CharacterIterator::EOrigin origin) { + switch(origin) { + case kStart: + pos = begin + delta; + break; + case kCurrent: + pos += delta; + break; + case kEnd: + pos = end + delta; + break; + default: + break; + } + + if(pos < begin) { + pos = begin; + } else if(pos > end) { + pos = end; + } + + return pos; +} + +int32_t +UCharCharacterIterator::move32(int32_t delta, CharacterIterator::EOrigin origin) { + // this implementation relies on the "safe" version of the UTF macros + // (or the trustworthiness of the caller) + switch(origin) { + case kStart: + pos = begin; + if(delta > 0) { + U16_FWD_N(text, pos, end, delta); + } + break; + case kCurrent: + if(delta > 0) { + U16_FWD_N(text, pos, end, delta); + } else { + U16_BACK_N(text, begin, pos, -delta); + } + break; + case kEnd: + pos = end; + if(delta < 0) { + U16_BACK_N(text, begin, pos, -delta); + } + break; + default: + break; + } + + return pos; +} + +void UCharCharacterIterator::setText(ConstChar16Ptr newText, + int32_t newTextLength) { + text = newText; + if(newText == 0 || newTextLength < 0) { + newTextLength = 0; + } + end = textLength = newTextLength; + pos = begin = 0; +} + +void +UCharCharacterIterator::getText(UnicodeString& result) { + result = UnicodeString(text, textLength); +} + +U_NAMESPACE_END diff --git a/intl/icu/source/common/ucln.h b/intl/icu/source/common/ucln.h new file mode 100644 index 0000000000..fe6666efed --- /dev/null +++ b/intl/icu/source/common/ucln.h @@ -0,0 +1,91 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2001-2013, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: ucln.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2001July05 +* created by: George Rhoten +*/ + +#ifndef __UCLN_H__ +#define __UCLN_H__ + +#include "unicode/utypes.h" + +/** These are the functions used to register a library's memory cleanup + * functions. Each library should define a single library register function + * to call this API. In the i18n library, it is ucln_i18n_registerCleanup(). + * + * None of the cleanup functions should use a mutex to clean up an API's + * allocated memory because a cleanup function is not meant to be thread safe, + * and plenty of data cannot be reference counted in order to make sure that + * no one else needs the allocated data. + * + * In order to make a cleanup function get called when u_cleanup is called, + * You should add your function to the library specific cleanup function. + * If the cleanup function is not in the common library, the code that + * allocates the memory should call the library specific cleanup function. + * For instance, in the i18n library, any memory allocated statically must + * call ucln_i18n_registerCleanup() from the ucln_in.h header. These library + * cleanup functions are needed in order to prevent a circular dependency + * between the common library and any other library. + * + * The order of the cleanup is very important. In general, an API that + * depends on a second API should be cleaned up before the second API. + * For instance, the default converter in ustring depends upon the converter + * API. So the default converter should be closed before the converter API + * has its cache flushed. This will prevent any memory leaks due to + * reference counting. + * + * Please see common/ucln_cmn.{h,c} and i18n/ucln_in.{h,c} for examples. + */ + +/** + * Data Type for cleanup function selector. These roughly correspond to libraries. + */ +typedef enum ECleanupLibraryType { + UCLN_START = -1, + UCLN_UPLUG, /* ICU plugins */ + UCLN_CUSTOM, /* Custom is for anyone else. */ + UCLN_CTESTFW, + UCLN_TOOLUTIL, + UCLN_LAYOUTEX, + UCLN_LAYOUT, + UCLN_IO, + UCLN_I18N, + UCLN_COMMON /* This must be the last one to cleanup. */ +} ECleanupLibraryType; + +/** + * Data type for cleanup function pointer + */ +U_CDECL_BEGIN +typedef UBool U_CALLCONV cleanupFunc(void); +typedef void U_CALLCONV initFunc(UErrorCode *); +U_CDECL_END + +/** + * Register a cleanup function + * @param type which library to register for. + * @param func the function pointer + */ +U_CAPI void U_EXPORT2 ucln_registerCleanup(ECleanupLibraryType type, + cleanupFunc *func); + +/** + * Request cleanup for one specific library. + * Not thread safe. + * @param type which library to cleanup + */ +U_CAPI void U_EXPORT2 ucln_cleanupOne(ECleanupLibraryType type); + +#endif diff --git a/intl/icu/source/common/ucln_cmn.cpp b/intl/icu/source/common/ucln_cmn.cpp new file mode 100644 index 0000000000..c63bd22192 --- /dev/null +++ b/intl/icu/source/common/ucln_cmn.cpp @@ -0,0 +1,124 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 2001-2014, International Business Machines +* Corporation and others. All Rights Reserved. +****************************************************************************** +* file name: ucln_cmn.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2001July05 +* created by: George Rhoten +*/ + +#include "unicode/utypes.h" +#include "unicode/uclean.h" +#include "cmemory.h" +#include "mutex.h" +#include "uassert.h" +#include "ucln.h" +#include "ucln_cmn.h" +#include "utracimp.h" +#include "umutex.h" + +/** Auto-client for UCLN_COMMON **/ +#define UCLN_TYPE_IS_COMMON +#include "ucln_imp.h" + +static cleanupFunc *gCommonCleanupFunctions[UCLN_COMMON_COUNT]; +static cleanupFunc *gLibCleanupFunctions[UCLN_COMMON]; + + +/************************************************ + The cleanup order is important in this function. + Please be sure that you have read ucln.h + ************************************************/ +U_CAPI void U_EXPORT2 +u_cleanup() +{ + UTRACE_ENTRY_OC(UTRACE_U_CLEANUP); + icu::umtx_lock(nullptr); /* Force a memory barrier, so that we are sure to see */ + icu::umtx_unlock(nullptr); /* all state left around by any other threads. */ + + ucln_lib_cleanup(); + + cmemory_cleanup(); /* undo any heap functions set by u_setMemoryFunctions(). */ + UTRACE_EXIT(); /* Must be before utrace_cleanup(), which turns off tracing. */ +/*#if U_ENABLE_TRACING*/ + utrace_cleanup(); +/*#endif*/ +} + +U_CAPI void U_EXPORT2 ucln_cleanupOne(ECleanupLibraryType libType) +{ + if (gLibCleanupFunctions[libType]) + { + gLibCleanupFunctions[libType](); + gLibCleanupFunctions[libType] = nullptr; + } +} + +U_CFUNC void +ucln_common_registerCleanup(ECleanupCommonType type, + cleanupFunc *func) +{ + // Thread safety messiness: From ticket 10295, calls to registerCleanup() may occur + // concurrently. Although such cases should be storing the same value, they raise errors + // from the thread sanity checker. Doing the store within a mutex avoids those. + // BUT that can trigger a recursive entry into std::call_once() in umutex.cpp when this code, + // running from the call_once function, tries to grab the ICU global mutex, which + // re-enters the mutex init path. So, work-around by special casing UCLN_COMMON_MUTEX, not + // using the ICU global mutex for it. + // + // No other point in ICU uses std::call_once(). + + U_ASSERT(UCLN_COMMON_START < type && type < UCLN_COMMON_COUNT); + if (type == UCLN_COMMON_MUTEX) { + gCommonCleanupFunctions[type] = func; + } else if (UCLN_COMMON_START < type && type < UCLN_COMMON_COUNT) { + icu::Mutex m; // See ticket 10295 for discussion. + gCommonCleanupFunctions[type] = func; + } +#if !UCLN_NO_AUTO_CLEANUP && (defined(UCLN_AUTO_ATEXIT) || defined(UCLN_AUTO_LOCAL)) + ucln_registerAutomaticCleanup(); +#endif +} + +// Note: ucln_registerCleanup() is called with the ICU global mutex locked. +// Be aware if adding anything to the function. +// See ticket 10295 for discussion. + +U_CAPI void U_EXPORT2 +ucln_registerCleanup(ECleanupLibraryType type, + cleanupFunc *func) +{ + U_ASSERT(UCLN_START < type && type < UCLN_COMMON); + if (UCLN_START < type && type < UCLN_COMMON) + { + gLibCleanupFunctions[type] = func; + } +} + +U_CFUNC UBool ucln_lib_cleanup() { + int32_t libType = UCLN_START; + int32_t commonFunc = UCLN_COMMON_START; + + for (libType++; libType(libType)); + } + + for (commonFunc++; commonFunc + +/** + * Auto cleanup of ICU libraries + * There are several methods in per library cleanup of icu libraries: + * 1) Compiler/Platform based cleanup: + * a) Windows MSVC uses DllMain() + * b) GCC uses destructor function attribute + * c) Sun Studio, AIX VA, and HP-UX aCC uses a linker option to set the exit function + * 2) Using atexit() + * 3) Implementing own automatic cleanup functions + * + * For option 1, ensure that UCLN_NO_AUTO_CLEANUP is set to 0 by using --enable-auto-cleanup + * configure option or by otherwise setting UCLN_NO_AUTO_CLEANUP to 0 + * For option 2, follow option 1 and also define UCLN_AUTO_ATEXIT + * For option 3, follow option 1 and also define UCLN_AUTO_LOCAL (see below for more information) + */ + +#if !UCLN_NO_AUTO_CLEANUP + +/* + * The following declarations are for when UCLN_AUTO_LOCAL or UCLN_AUTO_ATEXIT + * are defined. They are commented out because they are static and will be defined + * later. The information is still here to provide some guidance for the developer + * who chooses to use UCLN_AUTO_LOCAL. + */ +/** + * Give the library an opportunity to register an automatic cleanup. + * This may be called more than once. + */ +/*static void ucln_registerAutomaticCleanup();*/ +/** + * Unregister an automatic cleanup, if possible. Called from cleanup. + */ +/*static void ucln_unRegisterAutomaticCleanup();*/ + +#ifdef UCLN_TYPE_IS_COMMON +# define UCLN_CLEAN_ME_UP u_cleanup() +#else +# define UCLN_CLEAN_ME_UP ucln_cleanupOne(UCLN_TYPE) +#endif + +/* ------------ automatic cleanup: registration. Choose ONE ------- */ +#if defined(UCLN_AUTO_LOCAL) +/* To use: + * 1. define UCLN_AUTO_LOCAL, + * 2. create ucln_local_hook.c containing implementations of + * static void ucln_registerAutomaticCleanup() + * static void ucln_unRegisterAutomaticCleanup() + */ +#include "ucln_local_hook.c" + +#elif defined(UCLN_AUTO_ATEXIT) +/* + * Use the ANSI C 'atexit' function. Note that this mechanism does not + * guarantee the order of cleanup relative to other users of ICU! + */ +static UBool gAutoCleanRegistered = false; + +static void ucln_atexit_handler() +{ + UCLN_CLEAN_ME_UP; +} + +static void ucln_registerAutomaticCleanup() +{ + if(!gAutoCleanRegistered) { + gAutoCleanRegistered = true; + atexit(&ucln_atexit_handler); + } +} + +static void ucln_unRegisterAutomaticCleanup () { +} +/* ------------end of automatic cleanup: registration. ------- */ + +#elif defined (UCLN_FINI) +/** + * If UCLN_FINI is defined, it is the (versioned, etc) name of a cleanup + * entrypoint. Add a stub to call ucln_cleanupOne + * Used on AIX, Solaris, and HP-UX + */ +U_CAPI void U_EXPORT2 UCLN_FINI (void); + +U_CAPI void U_EXPORT2 UCLN_FINI () +{ + /* This function must be defined, if UCLN_FINI is defined, else link error. */ + UCLN_CLEAN_ME_UP; +} + +/* Windows: DllMain */ +#elif U_PLATFORM_HAS_WIN32_API +/* + * ICU's own DllMain. + */ + +/* these are from putil.c */ +/* READ READ READ READ! Are you getting compilation errors from windows.h? + Any source file which includes this (ucln_imp.h) header MUST + be defined with language extensions ON. */ +#ifndef WIN32_LEAN_AND_MEAN +# define WIN32_LEAN_AND_MEAN +#endif +# define VC_EXTRALEAN +# define NOUSER +# define NOSERVICE +# define NOIME +# define NOMCX +# include +/* + * This is a stub DllMain function with icu specific process handling code. + */ +BOOL WINAPI DllMain(HINSTANCE hinstDLL, DWORD fdwReason, LPVOID lpvReserved) +{ + BOOL status = true; + + switch(fdwReason) { + case DLL_PROCESS_ATTACH: + /* ICU does not trap process attach, but must pass these through properly. */ + /* ICU specific process attach could go here */ + break; + + case DLL_PROCESS_DETACH: + /* Here is the one we actually care about. */ + + UCLN_CLEAN_ME_UP; + + break; + + case DLL_THREAD_ATTACH: + /* ICU does not trap thread attach, but must pass these through properly. */ + /* ICU specific thread attach could go here */ + break; + + case DLL_THREAD_DETACH: + /* ICU does not trap thread detach, but must pass these through properly. */ + /* ICU specific thread detach could go here */ + break; + + } + return status; +} + +#elif defined(__GNUC__) +/* GCC - use __attribute((destructor)) */ +static void ucln_destructor() __attribute__((destructor)) ; + +static void ucln_destructor() +{ + UCLN_CLEAN_ME_UP; +} + +#endif + +#endif /* UCLN_NO_AUTO_CLEANUP */ + +#else +#error This file can only be included once. +#endif diff --git a/intl/icu/source/common/ucmndata.cpp b/intl/icu/source/common/ucmndata.cpp new file mode 100644 index 0000000000..69575d4e83 --- /dev/null +++ b/intl/icu/source/common/ucmndata.cpp @@ -0,0 +1,393 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2011, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************/ + + +/*------------------------------------------------------------------------------ + * + * UCommonData An abstract interface for dealing with ICU Common Data Files. + * ICU Common Data Files are a grouping of a number of individual + * data items (resources, converters, tables, anything) into a + * single file or dll. The combined format includes a table of + * contents for locating the individual items by name. + * + * Two formats for the table of contents are supported, which is + * why there is an abstract interface involved. + * + */ + +#include "unicode/utypes.h" +#include "unicode/udata.h" +#include "cstring.h" +#include "ucmndata.h" +#include "udatamem.h" + +#if defined(UDATA_DEBUG) || defined(UDATA_DEBUG_DUMP) +# include +#endif + +U_CFUNC uint16_t +udata_getHeaderSize(const DataHeader *udh) { + if(udh==nullptr) { + return 0; + } else if(udh->info.isBigEndian==U_IS_BIG_ENDIAN) { + /* same endianness */ + return udh->dataHeader.headerSize; + } else { + /* opposite endianness */ + uint16_t x=udh->dataHeader.headerSize; + return (uint16_t)((x<<8)|(x>>8)); + } +} + +U_CFUNC uint16_t +udata_getInfoSize(const UDataInfo *info) { + if(info==nullptr) { + return 0; + } else if(info->isBigEndian==U_IS_BIG_ENDIAN) { + /* same endianness */ + return info->size; + } else { + /* opposite endianness */ + uint16_t x=info->size; + return (uint16_t)((x<<8)|(x>>8)); + } +} + +/*-----------------------------------------------------------------------------* + * * + * Pointer TOCs. TODO: This form of table-of-contents should be removed * + * because DLLs must be relocated on loading to correct the * + * pointer values and this operation makes shared memory * + * mapping of the data much less likely to work. * + * * + *-----------------------------------------------------------------------------*/ +typedef struct { + const char *entryName; + const DataHeader *pHeader; +} PointerTOCEntry; + + +typedef struct { + uint32_t count; + uint32_t reserved; + /** + * Variable-length array declared with length 1 to disable bounds checkers. + * The actual array length is in the count field. + */ + PointerTOCEntry entry[1]; +} PointerTOC; + + +/* definition of OffsetTOC struct types moved to ucmndata.h */ + +/*-----------------------------------------------------------------------------* + * * + * entry point lookup implementations * + * * + *-----------------------------------------------------------------------------*/ + +#ifndef MIN +#define MIN(a,b) (((a)<(b)) ? (a) : (b)) +#endif + +/** + * Compare strings where we know the shared prefix length, + * and advance the prefix length as we find that the strings share even more characters. + */ +static int32_t +strcmpAfterPrefix(const char *s1, const char *s2, int32_t *pPrefixLength) { + int32_t pl=*pPrefixLength; + int32_t cmp=0; + s1+=pl; + s2+=pl; + for(;;) { + int32_t c1=(uint8_t)*s1++; + int32_t c2=(uint8_t)*s2++; + cmp=c1-c2; + if(cmp!=0 || c1==0) { /* different or done */ + break; + } + ++pl; /* increment shared same-prefix length */ + } + *pPrefixLength=pl; + return cmp; +} + +static int32_t +offsetTOCPrefixBinarySearch(const char *s, const char *names, + const UDataOffsetTOCEntry *toc, int32_t count) { + int32_t start=0; + int32_t limit=count; + /* + * Remember the shared prefix between s, start and limit, + * and don't compare that shared prefix again. + * The shared prefix should get longer as we narrow the [start, limit[ range. + */ + int32_t startPrefixLength=0; + int32_t limitPrefixLength=0; + if(count==0) { + return -1; + } + /* + * Prime the prefix lengths so that we don't keep prefixLength at 0 until + * both the start and limit indexes have moved. + * At the same time, we find if s is one of the start and (limit-1) names, + * and if not, exclude them from the actual binary search. + */ + if(0==strcmpAfterPrefix(s, names+toc[0].nameOffset, &startPrefixLength)) { + return 0; + } + ++start; + --limit; + if(0==strcmpAfterPrefix(s, names+toc[limit].nameOffset, &limitPrefixLength)) { + return limit; + } + while(starttoc; + if (toc != nullptr) { + retVal = toc->count; + } + return retVal; +} + +static const DataHeader * U_CALLCONV +offsetTOCLookupFn(const UDataMemory *pData, + const char *tocEntryName, + int32_t *pLength, + UErrorCode *pErrorCode) { + (void)pErrorCode; + const UDataOffsetTOC *toc = (UDataOffsetTOC *)pData->toc; + if(toc!=nullptr) { + const char *base=(const char *)toc; + int32_t number, count=(int32_t)toc->count; + + /* perform a binary search for the data in the common data's table of contents */ +#if defined (UDATA_DEBUG_DUMP) + /* list the contents of the TOC each time .. not recommended */ + for(number=0; numberentry[number].nameOffset]); + } +#endif + number=offsetTOCPrefixBinarySearch(tocEntryName, base, toc->entry, count); + if(number>=0) { + /* found it */ + const UDataOffsetTOCEntry *entry=toc->entry+number; +#ifdef UDATA_DEBUG + fprintf(stderr, "%s: Found.\n", tocEntryName); +#endif + if((number+1) < count) { + *pLength = (int32_t)(entry[1].dataOffset - entry->dataOffset); + } else { + *pLength = -1; + } + return (const DataHeader *)(base+entry->dataOffset); + } else { +#ifdef UDATA_DEBUG + fprintf(stderr, "%s: Not found.\n", tocEntryName); +#endif + return nullptr; + } + } else { +#ifdef UDATA_DEBUG + fprintf(stderr, "returning header\n"); +#endif + + return pData->pHeader; + } +} + + +static uint32_t U_CALLCONV pointerTOCEntryCount(const UDataMemory *pData) { + const PointerTOC *toc = (PointerTOC *)pData->toc; + return (uint32_t)((toc != nullptr) ? (toc->count) : 0); +} + +static const DataHeader * U_CALLCONV pointerTOCLookupFn(const UDataMemory *pData, + const char *name, + int32_t *pLength, + UErrorCode *pErrorCode) { + (void)pErrorCode; + if(pData->toc!=nullptr) { + const PointerTOC *toc = (PointerTOC *)pData->toc; + int32_t number, count=(int32_t)toc->count; + +#if defined (UDATA_DEBUG_DUMP) + /* list the contents of the TOC each time .. not recommended */ + for(number=0; numberentry[number].entryName); + } +#endif + number=pointerTOCPrefixBinarySearch(name, toc->entry, count); + if(number>=0) { + /* found it */ +#ifdef UDATA_DEBUG + fprintf(stderr, "%s: Found.\n", toc->entry[number].entryName); +#endif + *pLength=-1; + return UDataMemory_normalizeDataPointer(toc->entry[number].pHeader); + } else { +#ifdef UDATA_DEBUG + fprintf(stderr, "%s: Not found.\n", name); +#endif + return nullptr; + } + } else { + return pData->pHeader; + } +} +U_CDECL_END + + +static const commonDataFuncs CmnDFuncs = {offsetTOCLookupFn, offsetTOCEntryCount}; +static const commonDataFuncs ToCPFuncs = {pointerTOCLookupFn, pointerTOCEntryCount}; + + + +/*----------------------------------------------------------------------* + * * + * checkCommonData Validate the format of a common data file. * + * Fill in the virtual function ptr based on TOC type * + * If the data is invalid, close the UDataMemory * + * and set the appropriate error code. * + * * + *----------------------------------------------------------------------*/ +U_CFUNC void udata_checkCommonData(UDataMemory *udm, UErrorCode *err) { + if (U_FAILURE(*err)) { + return; + } + + if(udm==nullptr || udm->pHeader==nullptr) { + *err=U_INVALID_FORMAT_ERROR; + } else if(!(udm->pHeader->dataHeader.magic1==0xda && + udm->pHeader->dataHeader.magic2==0x27 && + udm->pHeader->info.isBigEndian==U_IS_BIG_ENDIAN && + udm->pHeader->info.charsetFamily==U_CHARSET_FAMILY) + ) { + /* header not valid */ + *err=U_INVALID_FORMAT_ERROR; + } + else if (udm->pHeader->info.dataFormat[0]==0x43 && + udm->pHeader->info.dataFormat[1]==0x6d && + udm->pHeader->info.dataFormat[2]==0x6e && + udm->pHeader->info.dataFormat[3]==0x44 && + udm->pHeader->info.formatVersion[0]==1 + ) { + /* dataFormat="CmnD" */ + udm->vFuncs = &CmnDFuncs; + udm->toc=(const char *)udm->pHeader+udata_getHeaderSize(udm->pHeader); + } + else if(udm->pHeader->info.dataFormat[0]==0x54 && + udm->pHeader->info.dataFormat[1]==0x6f && + udm->pHeader->info.dataFormat[2]==0x43 && + udm->pHeader->info.dataFormat[3]==0x50 && + udm->pHeader->info.formatVersion[0]==1 + ) { + /* dataFormat="ToCP" */ + udm->vFuncs = &ToCPFuncs; + udm->toc=(const char *)udm->pHeader+udata_getHeaderSize(udm->pHeader); + } + else { + /* dataFormat not recognized */ + *err=U_INVALID_FORMAT_ERROR; + } + + if (U_FAILURE(*err)) { + /* If the data is no good and we memory-mapped it ourselves, + * close the memory mapping so it doesn't leak. Note that this has + * no effect on non-memory mapped data, other than clearing fields in udm. + */ + udata_close(udm); + } +} + +/* + * TODO: Add a udata_swapPackageHeader() function that swaps an ICU .dat package + * header but not its sub-items. + * This function will be needed for automatic runtime swapping. + * Sub-items should not be swapped to limit the swapping to the parts of the + * package that are actually used. + * + * Since lengths of items are implicit in the order and offsets of their + * ToC entries, and since offsets are relative to the start of the ToC, + * a swapped version may need to generate a different data structure + * with pointers to the original data items and with their lengths + * (-1 for the last one if it is not known), and maybe even pointers to the + * swapped versions of the items. + * These pointers to swapped versions would establish a cache; + * instead, each open data item could simply own the storage for its swapped + * data. This fits better with the current design. + * + * markus 2003sep18 Jitterbug 2235 + */ diff --git a/intl/icu/source/common/ucmndata.h b/intl/icu/source/common/ucmndata.h new file mode 100644 index 0000000000..486b4fd7b5 --- /dev/null +++ b/intl/icu/source/common/ucmndata.h @@ -0,0 +1,117 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2011, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************/ + + +/*---------------------------------------------------------------------------------- + * + * UCommonData An abstract interface for dealing with ICU Common Data Files. + * ICU Common Data Files are a grouping of a number of individual + * data items (resources, converters, tables, anything) into a + * single file or dll. The combined format includes a table of + * contents for locating the individual items by name. + * + * Two formats for the table of contents are supported, which is + * why there is an abstract interface involved. + * + * These functions are part of the ICU internal implementation, and + * are not intended to be used directly by applications. + */ + +#ifndef __UCMNDATA_H__ +#define __UCMNDATA_H__ + +#include "unicode/udata.h" +#include "umapfile.h" + + +#define COMMON_DATA_NAME U_ICUDATA_NAME + +typedef struct { + uint16_t headerSize; + uint8_t magic1; + uint8_t magic2; +} MappedData; + + +typedef struct { + MappedData dataHeader; + UDataInfo info; +} DataHeader; + +typedef struct { + uint32_t nameOffset; + uint32_t dataOffset; +} UDataOffsetTOCEntry; + +typedef struct { + uint32_t count; + /** + * Variable-length array declared with length 1 to disable bounds checkers. + * The actual array length is in the count field. + */ + UDataOffsetTOCEntry entry[1]; +} UDataOffsetTOC; + +/** + * Get the header size from a const DataHeader *udh. + * Handles opposite-endian data. + * + * @internal + */ +U_CFUNC uint16_t +udata_getHeaderSize(const DataHeader *udh); + +/** + * Get the UDataInfo.size from a const UDataInfo *info. + * Handles opposite-endian data. + * + * @internal + */ +U_CFUNC uint16_t +udata_getInfoSize(const UDataInfo *info); + +U_CDECL_BEGIN +/* + * "Virtual" functions for data lookup. + * To call one, given a UDataMemory *p, the code looks like this: + * p->vFuncs.Lookup(p, tocEntryName, pErrorCode); + * (I sure do wish this was written in C++, not C) + */ + +typedef const DataHeader * +(U_CALLCONV * LookupFn)(const UDataMemory *pData, + const char *tocEntryName, + int32_t *pLength, + UErrorCode *pErrorCode); + +typedef uint32_t +(U_CALLCONV * NumEntriesFn)(const UDataMemory *pData); + +U_CDECL_END + +typedef struct { + LookupFn Lookup; + NumEntriesFn NumEntries; +} commonDataFuncs; + + +/* + * Functions to check whether a UDataMemory refers to memory containing + * a recognizable header and table of contents a Common Data Format + * + * If a valid header and TOC are found, + * set the CommonDataFuncs function dispatch vector in the UDataMemory + * to point to the right functions for the TOC type. + * otherwise + * set an errorcode. + */ +U_CFUNC void udata_checkCommonData(UDataMemory *pData, UErrorCode *pErrorCode); + +#endif diff --git a/intl/icu/source/common/ucnv.cpp b/intl/icu/source/common/ucnv.cpp new file mode 100644 index 0000000000..a7a07d65d6 --- /dev/null +++ b/intl/icu/source/common/ucnv.cpp @@ -0,0 +1,2922 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1998-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* ucnv.c: +* Implements APIs for the ICU's codeset conversion library; +* mostly calls through internal functions; +* created by Bertrand A. Damiba +* +* Modification History: +* +* Date Name Description +* 04/04/99 helena Fixed internal header inclusion. +* 05/09/00 helena Added implementation to handle fallback mappings. +* 06/20/2000 helena OS/400 port changes; mostly typecast. +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include + +#include "unicode/ustring.h" +#include "unicode/ucnv.h" +#include "unicode/ucnv_err.h" +#include "unicode/uset.h" +#include "unicode/utf.h" +#include "unicode/utf16.h" +#include "putilimp.h" +#include "cmemory.h" +#include "cstring.h" +#include "uassert.h" +#include "utracimp.h" +#include "ustr_imp.h" +#include "ucnv_imp.h" +#include "ucnv_cnv.h" +#include "ucnv_bld.h" + +/* size of intermediate and preflighting buffers in ucnv_convert() */ +#define CHUNK_SIZE 1024 + +typedef struct UAmbiguousConverter { + const char *name; + const char16_t variant5c; +} UAmbiguousConverter; + +static const UAmbiguousConverter ambiguousConverters[]={ + { "ibm-897_P100-1995", 0xa5 }, + { "ibm-942_P120-1999", 0xa5 }, + { "ibm-943_P130-1999", 0xa5 }, + { "ibm-946_P100-1995", 0xa5 }, + { "ibm-33722_P120-1999", 0xa5 }, + { "ibm-1041_P100-1995", 0xa5 }, + /*{ "ibm-54191_P100-2006", 0xa5 },*/ + /*{ "ibm-62383_P100-2007", 0xa5 },*/ + /*{ "ibm-891_P100-1995", 0x20a9 },*/ + { "ibm-944_P100-1995", 0x20a9 }, + { "ibm-949_P110-1999", 0x20a9 }, + { "ibm-1363_P110-1997", 0x20a9 }, + { "ISO_2022,locale=ko,version=0", 0x20a9 }, + { "ibm-1088_P100-1995", 0x20a9 } +}; + +/*Calls through createConverter */ +U_CAPI UConverter* U_EXPORT2 +ucnv_open (const char *name, + UErrorCode * err) +{ + UConverter *r; + + if (err == nullptr || U_FAILURE (*err)) { + return nullptr; + } + + r = ucnv_createConverter(nullptr, name, err); + return r; +} + +U_CAPI UConverter* U_EXPORT2 +ucnv_openPackage (const char *packageName, const char *converterName, UErrorCode * err) +{ + return ucnv_createConverterFromPackage(packageName, converterName, err); +} + +/*Extracts the char16_t* to a char* and calls through createConverter */ +U_CAPI UConverter* U_EXPORT2 +ucnv_openU (const char16_t * name, + UErrorCode * err) +{ + char asciiName[UCNV_MAX_CONVERTER_NAME_LENGTH]; + + if (err == nullptr || U_FAILURE(*err)) + return nullptr; + if (name == nullptr) + return ucnv_open (nullptr, err); + if (u_strlen(name) >= UCNV_MAX_CONVERTER_NAME_LENGTH) + { + *err = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + return ucnv_open(u_austrcpy(asciiName, name), err); +} + +/* Copy the string that is represented by the UConverterPlatform enum + * @param platformString An output buffer + * @param platform An enum representing a platform + * @return the length of the copied string. + */ +static int32_t +ucnv_copyPlatformString(char *platformString, UConverterPlatform pltfrm) +{ + switch (pltfrm) + { + case UCNV_IBM: + uprv_strcpy(platformString, "ibm-"); + return 4; + case UCNV_UNKNOWN: + break; + } + + /* default to empty string */ + *platformString = 0; + return 0; +} + +/*Assumes a $platform-#codepage.$CONVERTER_FILE_EXTENSION scheme and calls + *through createConverter*/ +U_CAPI UConverter* U_EXPORT2 +ucnv_openCCSID (int32_t codepage, + UConverterPlatform platform, + UErrorCode * err) +{ + char myName[UCNV_MAX_CONVERTER_NAME_LENGTH]; + int32_t myNameLen; + + if (err == nullptr || U_FAILURE (*err)) + return nullptr; + + /* ucnv_copyPlatformString could return "ibm-" or "cp" */ + myNameLen = ucnv_copyPlatformString(myName, platform); + T_CString_integerToString(myName + myNameLen, codepage, 10); + + return ucnv_createConverter(nullptr, myName, err); +} + +/* Creating a temporary stack-based object that can be used in one thread, +and created from a converter that is shared across threads. +*/ + +U_CAPI UConverter* U_EXPORT2 +ucnv_safeClone(const UConverter* cnv, void *stackBuffer, int32_t *pBufferSize, UErrorCode *status) +{ + UConverter *localConverter, *allocatedConverter; + int32_t stackBufferSize; + int32_t bufferSizeNeeded; + UErrorCode cbErr; + UConverterToUnicodeArgs toUArgs = { + sizeof(UConverterToUnicodeArgs), + true, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr + }; + UConverterFromUnicodeArgs fromUArgs = { + sizeof(UConverterFromUnicodeArgs), + true, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr + }; + + UTRACE_ENTRY_OC(UTRACE_UCNV_CLONE); + + if (status == nullptr || U_FAILURE(*status)){ + UTRACE_EXIT_STATUS(status? *status: U_ILLEGAL_ARGUMENT_ERROR); + return nullptr; + } + + if (cnv == nullptr) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + UTRACE_EXIT_STATUS(*status); + return nullptr; + } + + UTRACE_DATA3(UTRACE_OPEN_CLOSE, "clone converter %s at %p into stackBuffer %p", + ucnv_getName(cnv, status), cnv, stackBuffer); + + if (cnv->sharedData->impl->safeClone != nullptr) { + /* call the custom safeClone function for sizing */ + bufferSizeNeeded = 0; + cnv->sharedData->impl->safeClone(cnv, nullptr, &bufferSizeNeeded, status); + if (U_FAILURE(*status)) { + UTRACE_EXIT_STATUS(*status); + return nullptr; + } + } + else + { + /* inherent sizing */ + bufferSizeNeeded = sizeof(UConverter); + } + + if (pBufferSize == nullptr) { + stackBufferSize = 1; + pBufferSize = &stackBufferSize; + } else { + stackBufferSize = *pBufferSize; + if (stackBufferSize <= 0){ /* 'preflighting' request - set needed size into *pBufferSize */ + *pBufferSize = bufferSizeNeeded; + UTRACE_EXIT_VALUE(bufferSizeNeeded); + return nullptr; + } + } + + /* Adjust (if necessary) the stackBuffer pointer to be aligned correctly for a UConverter. + * TODO(Jira ICU-20736) Redo this using std::align() once g++4.9 compatibility is no longer needed. + */ + if (stackBuffer) { + uintptr_t p = reinterpret_cast(stackBuffer); + uintptr_t aligned_p = (p + alignof(UConverter) - 1) & ~(alignof(UConverter) - 1); + ptrdiff_t pointerAdjustment = aligned_p - p; + if (bufferSizeNeeded + pointerAdjustment <= stackBufferSize) { + stackBuffer = reinterpret_cast(aligned_p); + stackBufferSize -= static_cast(pointerAdjustment); + } else { + /* prevent using the stack buffer but keep the size > 0 so that we do not just preflight */ + stackBufferSize = 1; + } + } + + /* Now, see if we must allocate any memory */ + if (stackBufferSize < bufferSizeNeeded || stackBuffer == nullptr) + { + /* allocate one here...*/ + localConverter = allocatedConverter = (UConverter *) uprv_malloc (bufferSizeNeeded); + + if(localConverter == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + UTRACE_EXIT_STATUS(*status); + return nullptr; + } + // If pBufferSize was nullptr as the input, pBufferSize is set to &stackBufferSize in this function. + if (pBufferSize != &stackBufferSize) { + *status = U_SAFECLONE_ALLOCATED_WARNING; + } + + /* record the fact that memory was allocated */ + *pBufferSize = bufferSizeNeeded; + } else { + /* just use the stack buffer */ + localConverter = (UConverter*) stackBuffer; + allocatedConverter = nullptr; + } + + uprv_memset(localConverter, 0, bufferSizeNeeded); + + /* Copy initial state */ + uprv_memcpy(localConverter, cnv, sizeof(UConverter)); + localConverter->isCopyLocal = localConverter->isExtraLocal = false; + + /* copy the substitution string */ + if (cnv->subChars == (uint8_t *)cnv->subUChars) { + localConverter->subChars = (uint8_t *)localConverter->subUChars; + } else { + localConverter->subChars = (uint8_t *)uprv_malloc(UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR); + if (localConverter->subChars == nullptr) { + uprv_free(allocatedConverter); + UTRACE_EXIT_STATUS(*status); + return nullptr; + } + uprv_memcpy(localConverter->subChars, cnv->subChars, UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR); + } + + /* now either call the safeclone fcn or not */ + if (cnv->sharedData->impl->safeClone != nullptr) { + /* call the custom safeClone function */ + localConverter = cnv->sharedData->impl->safeClone(cnv, localConverter, pBufferSize, status); + } + + if(localConverter==nullptr || U_FAILURE(*status)) { + if (allocatedConverter != nullptr && allocatedConverter->subChars != (uint8_t *)allocatedConverter->subUChars) { + uprv_free(allocatedConverter->subChars); + } + uprv_free(allocatedConverter); + UTRACE_EXIT_STATUS(*status); + return nullptr; + } + + /* increment refcount of shared data if needed */ + if (cnv->sharedData->isReferenceCounted) { + ucnv_incrementRefCount(cnv->sharedData); + } + + if(localConverter == (UConverter*)stackBuffer) { + /* we're using user provided data - set to not destroy */ + localConverter->isCopyLocal = true; + } + + /* allow callback functions to handle any memory allocation */ + toUArgs.converter = fromUArgs.converter = localConverter; + cbErr = U_ZERO_ERROR; + cnv->fromCharErrorBehaviour(cnv->toUContext, &toUArgs, nullptr, 0, UCNV_CLONE, &cbErr); + cbErr = U_ZERO_ERROR; + cnv->fromUCharErrorBehaviour(cnv->fromUContext, &fromUArgs, nullptr, 0, 0, UCNV_CLONE, &cbErr); + + UTRACE_EXIT_PTR_STATUS(localConverter, *status); + return localConverter; +} + +U_CAPI UConverter* U_EXPORT2 +ucnv_clone(const UConverter* cnv, UErrorCode *status) +{ + return ucnv_safeClone(cnv, nullptr, nullptr, status); +} + +/*Decreases the reference counter in the shared immutable section of the object + *and frees the mutable part*/ + +U_CAPI void U_EXPORT2 +ucnv_close (UConverter * converter) +{ + UErrorCode errorCode = U_ZERO_ERROR; + + UTRACE_ENTRY_OC(UTRACE_UCNV_CLOSE); + + if (converter == nullptr) + { + UTRACE_EXIT(); + return; + } + + UTRACE_DATA3(UTRACE_OPEN_CLOSE, "close converter %s at %p, isCopyLocal=%b", + ucnv_getName(converter, &errorCode), converter, converter->isCopyLocal); + + /* In order to speed up the close, only call the callbacks when they have been changed. + This performance check will only work when the callbacks are set within a shared library + or from user code that statically links this code. */ + /* first, notify the callback functions that the converter is closed */ + if (converter->fromCharErrorBehaviour != UCNV_TO_U_DEFAULT_CALLBACK) { + UConverterToUnicodeArgs toUArgs = { + sizeof(UConverterToUnicodeArgs), + true, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr + }; + + toUArgs.converter = converter; + errorCode = U_ZERO_ERROR; + converter->fromCharErrorBehaviour(converter->toUContext, &toUArgs, nullptr, 0, UCNV_CLOSE, &errorCode); + } + if (converter->fromUCharErrorBehaviour != UCNV_FROM_U_DEFAULT_CALLBACK) { + UConverterFromUnicodeArgs fromUArgs = { + sizeof(UConverterFromUnicodeArgs), + true, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr + }; + fromUArgs.converter = converter; + errorCode = U_ZERO_ERROR; + converter->fromUCharErrorBehaviour(converter->fromUContext, &fromUArgs, nullptr, 0, 0, UCNV_CLOSE, &errorCode); + } + + if (converter->sharedData->impl->close != nullptr) { + converter->sharedData->impl->close(converter); + } + + if (converter->subChars != (uint8_t *)converter->subUChars) { + uprv_free(converter->subChars); + } + + if (converter->sharedData->isReferenceCounted) { + ucnv_unloadSharedDataIfReady(converter->sharedData); + } + + if(!converter->isCopyLocal){ + uprv_free(converter); + } + + UTRACE_EXIT(); +} + +/*returns a single Name from the list, will return nullptr if out of bounds + */ +U_CAPI const char* U_EXPORT2 +ucnv_getAvailableName (int32_t n) +{ + if (0 <= n && n <= 0xffff) { + UErrorCode err = U_ZERO_ERROR; + const char *name = ucnv_bld_getAvailableConverter((uint16_t)n, &err); + if (U_SUCCESS(err)) { + return name; + } + } + return nullptr; +} + +U_CAPI int32_t U_EXPORT2 +ucnv_countAvailable () +{ + UErrorCode err = U_ZERO_ERROR; + return ucnv_bld_countAvailableConverters(&err); +} + +U_CAPI void U_EXPORT2 +ucnv_getSubstChars (const UConverter * converter, + char *mySubChar, + int8_t * len, + UErrorCode * err) +{ + if (U_FAILURE (*err)) + return; + + if (converter->subCharLen <= 0) { + /* Unicode string or empty string from ucnv_setSubstString(). */ + *len = 0; + return; + } + + if (*len < converter->subCharLen) /*not enough space in subChars */ + { + *err = U_INDEX_OUTOFBOUNDS_ERROR; + return; + } + + uprv_memcpy (mySubChar, converter->subChars, converter->subCharLen); /*fills in the subchars */ + *len = converter->subCharLen; /*store # of bytes copied to buffer */ +} + +U_CAPI void U_EXPORT2 +ucnv_setSubstChars (UConverter * converter, + const char *mySubChar, + int8_t len, + UErrorCode * err) +{ + if (U_FAILURE (*err)) + return; + + /*Makes sure that the subChar is within the codepages char length boundaries */ + if ((len > converter->sharedData->staticData->maxBytesPerChar) + || (len < converter->sharedData->staticData->minBytesPerChar)) + { + *err = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + uprv_memcpy (converter->subChars, mySubChar, len); /*copies the subchars */ + converter->subCharLen = len; /*sets the new len */ + + /* + * There is currently (2001Feb) no separate API to set/get subChar1. + * In order to always have subChar written after it is explicitly set, + * we set subChar1 to 0. + */ + converter->subChar1 = 0; + + return; +} + +U_CAPI void U_EXPORT2 +ucnv_setSubstString(UConverter *cnv, + const char16_t *s, + int32_t length, + UErrorCode *err) { + alignas(UConverter) char cloneBuffer[U_CNV_SAFECLONE_BUFFERSIZE]; + char chars[UCNV_ERROR_BUFFER_LENGTH]; + + UConverter *clone; + uint8_t *subChars; + int32_t cloneSize, length8; + + /* Let the following functions check all arguments. */ + cloneSize = sizeof(cloneBuffer); + clone = ucnv_safeClone(cnv, cloneBuffer, &cloneSize, err); + ucnv_setFromUCallBack(clone, UCNV_FROM_U_CALLBACK_STOP, nullptr, nullptr, nullptr, err); + length8 = ucnv_fromUChars(clone, chars, (int32_t)sizeof(chars), s, length, err); + ucnv_close(clone); + if (U_FAILURE(*err)) { + return; + } + + if (cnv->sharedData->impl->writeSub == nullptr +#if !UCONFIG_NO_LEGACY_CONVERSION + || (cnv->sharedData->staticData->conversionType == UCNV_MBCS && + ucnv_MBCSGetType(cnv) != UCNV_EBCDIC_STATEFUL) +#endif + ) { + /* The converter is not stateful. Store the charset bytes as a fixed string. */ + subChars = (uint8_t *)chars; + } else { + /* + * The converter has a non-default writeSub() function, indicating + * that it is stateful. + * Store the Unicode string for on-the-fly conversion for correct + * state handling. + */ + if (length > UCNV_ERROR_BUFFER_LENGTH) { + /* + * Should not occur. The converter should output at least one byte + * per char16_t, which means that ucnv_fromUChars() should catch all + * overflows. + */ + *err = U_BUFFER_OVERFLOW_ERROR; + return; + } + subChars = (uint8_t *)s; + if (length < 0) { + length = u_strlen(s); + } + length8 = length * U_SIZEOF_UCHAR; + } + + /* + * For storing the substitution string, select either the small buffer inside + * UConverter or allocate a subChars buffer. + */ + if (length8 > UCNV_MAX_SUBCHAR_LEN) { + /* Use a separate buffer for the string. Outside UConverter to not make it too large. */ + if (cnv->subChars == (uint8_t *)cnv->subUChars) { + /* Allocate a new buffer for the string. */ + cnv->subChars = (uint8_t *)uprv_malloc(UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR); + if (cnv->subChars == nullptr) { + cnv->subChars = (uint8_t *)cnv->subUChars; + *err = U_MEMORY_ALLOCATION_ERROR; + return; + } + uprv_memset(cnv->subChars, 0, UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR); + } + } + + /* Copy the substitution string into the UConverter or its subChars buffer. */ + if (length8 == 0) { + cnv->subCharLen = 0; + } else { + uprv_memcpy(cnv->subChars, subChars, length8); + if (subChars == (uint8_t *)chars) { + cnv->subCharLen = (int8_t)length8; + } else /* subChars == s */ { + cnv->subCharLen = (int8_t)-length; + } + } + + /* See comment in ucnv_setSubstChars(). */ + cnv->subChar1 = 0; +} + +/*resets the internal states of a converter + *goal : have the same behaviour than a freshly created converter + */ +static void _reset(UConverter *converter, UConverterResetChoice choice, + UBool callCallback) { + if(converter == nullptr) { + return; + } + + if(callCallback) { + /* first, notify the callback functions that the converter is reset */ + UErrorCode errorCode; + + if(choice<=UCNV_RESET_TO_UNICODE && converter->fromCharErrorBehaviour != UCNV_TO_U_DEFAULT_CALLBACK) { + UConverterToUnicodeArgs toUArgs = { + sizeof(UConverterToUnicodeArgs), + true, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr + }; + toUArgs.converter = converter; + errorCode = U_ZERO_ERROR; + converter->fromCharErrorBehaviour(converter->toUContext, &toUArgs, nullptr, 0, UCNV_RESET, &errorCode); + } + if(choice!=UCNV_RESET_TO_UNICODE && converter->fromUCharErrorBehaviour != UCNV_FROM_U_DEFAULT_CALLBACK) { + UConverterFromUnicodeArgs fromUArgs = { + sizeof(UConverterFromUnicodeArgs), + true, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr + }; + fromUArgs.converter = converter; + errorCode = U_ZERO_ERROR; + converter->fromUCharErrorBehaviour(converter->fromUContext, &fromUArgs, nullptr, 0, 0, UCNV_RESET, &errorCode); + } + } + + /* now reset the converter itself */ + if(choice<=UCNV_RESET_TO_UNICODE) { + converter->toUnicodeStatus = converter->sharedData->toUnicodeStatus; + converter->mode = 0; + converter->toULength = 0; + converter->invalidCharLength = converter->UCharErrorBufferLength = 0; + converter->preToULength = 0; + } + if(choice!=UCNV_RESET_TO_UNICODE) { + converter->fromUnicodeStatus = 0; + converter->fromUChar32 = 0; + converter->invalidUCharLength = converter->charErrorBufferLength = 0; + converter->preFromUFirstCP = U_SENTINEL; + converter->preFromULength = 0; + } + + if (converter->sharedData->impl->reset != nullptr) { + /* call the custom reset function */ + converter->sharedData->impl->reset(converter, choice); + } +} + +U_CAPI void U_EXPORT2 +ucnv_reset(UConverter *converter) +{ + _reset(converter, UCNV_RESET_BOTH, true); +} + +U_CAPI void U_EXPORT2 +ucnv_resetToUnicode(UConverter *converter) +{ + _reset(converter, UCNV_RESET_TO_UNICODE, true); +} + +U_CAPI void U_EXPORT2 +ucnv_resetFromUnicode(UConverter *converter) +{ + _reset(converter, UCNV_RESET_FROM_UNICODE, true); +} + +U_CAPI int8_t U_EXPORT2 +ucnv_getMaxCharSize (const UConverter * converter) +{ + return converter->maxBytesPerUChar; +} + + +U_CAPI int8_t U_EXPORT2 +ucnv_getMinCharSize (const UConverter * converter) +{ + return converter->sharedData->staticData->minBytesPerChar; +} + +U_CAPI const char* U_EXPORT2 +ucnv_getName (const UConverter * converter, UErrorCode * err) + +{ + if (U_FAILURE (*err)) + return nullptr; + if(converter->sharedData->impl->getName){ + const char* temp= converter->sharedData->impl->getName(converter); + if(temp) + return temp; + } + return converter->sharedData->staticData->name; +} + +U_CAPI int32_t U_EXPORT2 +ucnv_getCCSID(const UConverter * converter, + UErrorCode * err) +{ + int32_t ccsid; + if (U_FAILURE (*err)) + return -1; + + ccsid = converter->sharedData->staticData->codepage; + if (ccsid == 0) { + /* Rare case. This is for cases like gb18030, + which doesn't have an IBM canonical name, but does have an IBM alias. */ + const char *standardName = ucnv_getStandardName(ucnv_getName(converter, err), "IBM", err); + if (U_SUCCESS(*err) && standardName) { + const char *ccsidStr = uprv_strchr(standardName, '-'); + if (ccsidStr) { + ccsid = (int32_t)atol(ccsidStr+1); /* +1 to skip '-' */ + } + } + } + return ccsid; +} + + +U_CAPI UConverterPlatform U_EXPORT2 +ucnv_getPlatform (const UConverter * converter, + UErrorCode * err) +{ + if (U_FAILURE (*err)) + return UCNV_UNKNOWN; + + return (UConverterPlatform)converter->sharedData->staticData->platform; +} + +U_CAPI void U_EXPORT2 + ucnv_getToUCallBack (const UConverter * converter, + UConverterToUCallback *action, + const void **context) +{ + *action = converter->fromCharErrorBehaviour; + *context = converter->toUContext; +} + +U_CAPI void U_EXPORT2 + ucnv_getFromUCallBack (const UConverter * converter, + UConverterFromUCallback *action, + const void **context) +{ + *action = converter->fromUCharErrorBehaviour; + *context = converter->fromUContext; +} + +U_CAPI void U_EXPORT2 +ucnv_setToUCallBack (UConverter * converter, + UConverterToUCallback newAction, + const void* newContext, + UConverterToUCallback *oldAction, + const void** oldContext, + UErrorCode * err) +{ + if (U_FAILURE (*err)) + return; + if (oldAction) *oldAction = converter->fromCharErrorBehaviour; + converter->fromCharErrorBehaviour = newAction; + if (oldContext) *oldContext = converter->toUContext; + converter->toUContext = newContext; +} + +U_CAPI void U_EXPORT2 +ucnv_setFromUCallBack (UConverter * converter, + UConverterFromUCallback newAction, + const void* newContext, + UConverterFromUCallback *oldAction, + const void** oldContext, + UErrorCode * err) +{ + if (U_FAILURE (*err)) + return; + if (oldAction) *oldAction = converter->fromUCharErrorBehaviour; + converter->fromUCharErrorBehaviour = newAction; + if (oldContext) *oldContext = converter->fromUContext; + converter->fromUContext = newContext; +} + +static void +_updateOffsets(int32_t *offsets, int32_t length, + int32_t sourceIndex, int32_t errorInputLength) { + int32_t *limit; + int32_t delta, offset; + + if(sourceIndex>=0) { + /* + * adjust each offset by adding the previous sourceIndex + * minus the length of the input sequence that caused an + * error, if any + */ + delta=sourceIndex-errorInputLength; + } else { + /* + * set each offset to -1 because this conversion function + * does not handle offsets + */ + delta=-1; + } + + limit=offsets+length; + if(delta==0) { + /* most common case, nothing to do */ + } else if(delta>0) { + /* add the delta to each offset (but not if the offset is <0) */ + while(offsets=0) { + *offsets=offset+delta; + } + ++offsets; + } + } else /* delta<0 */ { + /* + * set each offset to -1 because this conversion function + * does not handle offsets + * or the error input sequence started in a previous buffer + */ + while(offsetsconverter; + s=pArgs->source; + t=pArgs->target; + offsets=pArgs->offsets; + + /* get the converter implementation function */ + sourceIndex=0; + if(offsets==nullptr) { + fromUnicode=cnv->sharedData->impl->fromUnicode; + } else { + fromUnicode=cnv->sharedData->impl->fromUnicodeWithOffsets; + if(fromUnicode==nullptr) { + /* there is no WithOffsets implementation */ + fromUnicode=cnv->sharedData->impl->fromUnicode; + /* we will write -1 for each offset */ + sourceIndex=-1; + } + } + + if(cnv->preFromULength>=0) { + /* normal mode */ + realSource=nullptr; + + /* avoid compiler warnings - not otherwise necessary, and the values do not matter */ + realSourceLimit=nullptr; + realFlush=false; + realSourceIndex=0; + } else { + /* + * Previous m:n conversion stored source units from a partial match + * and failed to consume all of them. + * We need to "replay" them from a temporary buffer and convert them first. + */ + realSource=pArgs->source; + realSourceLimit=pArgs->sourceLimit; + realFlush=pArgs->flush; + realSourceIndex=sourceIndex; + + uprv_memcpy(replay, cnv->preFromU, -cnv->preFromULength*U_SIZEOF_UCHAR); + pArgs->source=replay; + pArgs->sourceLimit=replay-cnv->preFromULength; + pArgs->flush=false; + sourceIndex=-1; + + cnv->preFromULength=0; + } + + /* + * loop for conversion and error handling + * + * loop { + * convert + * loop { + * update offsets + * handle end of input + * handle errors/call callback + * } + * } + */ + for(;;) { + if(U_SUCCESS(*err)) { + /* convert */ + fromUnicode(pArgs, err); + + /* + * set a flag for whether the converter + * successfully processed the end of the input + * + * need not check cnv->preFromULength==0 because a replay (<0) will cause + * sflush && pArgs->source==pArgs->sourceLimit && + cnv->fromUChar32==0); + } else { + /* handle error from ucnv_convertEx() */ + converterSawEndOfInput=false; + } + + /* no callback called yet for this iteration */ + calledCallback=false; + + /* no sourceIndex adjustment for conversion, only for callback output */ + errorInputLength=0; + + /* + * loop for offsets and error handling + * + * iterates at most 3 times: + * 1. to clean up after the conversion function + * 2. after the callback + * 3. after the callback again if there was truncated input + */ + for(;;) { + /* update offsets if we write any */ + if(offsets!=nullptr) { + int32_t length=(int32_t)(pArgs->target-t); + if(length>0) { + _updateOffsets(offsets, length, sourceIndex, errorInputLength); + + /* + * if a converter handles offsets and updates the offsets + * pointer at the end, then pArgs->offset should not change + * here; + * however, some converters do not handle offsets at all + * (sourceIndex<0) or may not update the offsets pointer + */ + pArgs->offsets=offsets+=length; + } + + if(sourceIndex>=0) { + sourceIndex+=(int32_t)(pArgs->source-s); + } + } + + if(cnv->preFromULength<0) { + /* + * switch the source to new replay units (cannot occur while replaying) + * after offset handling and before end-of-input and callback handling + */ + if(realSource==nullptr) { + realSource=pArgs->source; + realSourceLimit=pArgs->sourceLimit; + realFlush=pArgs->flush; + realSourceIndex=sourceIndex; + + uprv_memcpy(replay, cnv->preFromU, -cnv->preFromULength*U_SIZEOF_UCHAR); + pArgs->source=replay; + pArgs->sourceLimit=replay-cnv->preFromULength; + pArgs->flush=false; + if((sourceIndex+=cnv->preFromULength)<0) { + sourceIndex=-1; + } + + cnv->preFromULength=0; + } else { + /* see implementation note before _fromUnicodeWithCallback() */ + U_ASSERT(realSource==nullptr); + *err=U_INTERNAL_PROGRAM_ERROR; + } + } + + /* update pointers */ + s=pArgs->source; + t=pArgs->target; + + if(U_SUCCESS(*err)) { + if(ssourceLimit) { + /* + * continue with the conversion loop while there is still input left + * (continue converting by breaking out of only the inner loop) + */ + break; + } else if(realSource!=nullptr) { + /* switch back from replaying to the real source and continue */ + pArgs->source=realSource; + pArgs->sourceLimit=realSourceLimit; + pArgs->flush=realFlush; + sourceIndex=realSourceIndex; + + realSource=nullptr; + break; + } else if(pArgs->flush && cnv->fromUChar32!=0) { + /* + * the entire input stream is consumed + * and there is a partial, truncated input sequence left + */ + + /* inject an error and continue with callback handling */ + *err=U_TRUNCATED_CHAR_FOUND; + calledCallback=false; /* new error condition */ + } else { + /* input consumed */ + if(pArgs->flush) { + /* + * return to the conversion loop once more if the flush + * flag is set and the conversion function has not + * successfully processed the end of the input yet + * + * (continue converting by breaking out of only the inner loop) + */ + if(!converterSawEndOfInput) { + break; + } + + /* reset the converter without calling the callback function */ + _reset(cnv, UCNV_RESET_FROM_UNICODE, false); + } + + /* done successfully */ + return; + } + } + + /* U_FAILURE(*err) */ + { + UErrorCode e; + + if( calledCallback || + (e=*err)==U_BUFFER_OVERFLOW_ERROR || + (e!=U_INVALID_CHAR_FOUND && + e!=U_ILLEGAL_CHAR_FOUND && + e!=U_TRUNCATED_CHAR_FOUND) + ) { + /* + * the callback did not or cannot resolve the error: + * set output pointers and return + * + * the check for buffer overflow is redundant but it is + * a high-runner case and hopefully documents the intent + * well + * + * if we were replaying, then the replay buffer must be + * copied back into the UConverter + * and the real arguments must be restored + */ + if(realSource!=nullptr) { + int32_t length; + + U_ASSERT(cnv->preFromULength==0); + + length=(int32_t)(pArgs->sourceLimit-pArgs->source); + if(length>0) { + u_memcpy(cnv->preFromU, pArgs->source, length); + cnv->preFromULength=(int8_t)-length; + } + + pArgs->source=realSource; + pArgs->sourceLimit=realSourceLimit; + pArgs->flush=realFlush; + } + + return; + } + } + + /* callback handling */ + { + UChar32 codePoint; + + /* get and write the code point */ + codePoint=cnv->fromUChar32; + errorInputLength=0; + U16_APPEND_UNSAFE(cnv->invalidUCharBuffer, errorInputLength, codePoint); + cnv->invalidUCharLength=(int8_t)errorInputLength; + + /* set the converter state to deal with the next character */ + cnv->fromUChar32=0; + + /* call the callback function */ + cnv->fromUCharErrorBehaviour(cnv->fromUContext, pArgs, + cnv->invalidUCharBuffer, errorInputLength, codePoint, + *err==U_INVALID_CHAR_FOUND ? UCNV_UNASSIGNED : UCNV_ILLEGAL, + err); + } + + /* + * loop back to the offset handling + * + * this flag will indicate after offset handling + * that a callback was called; + * if the callback did not resolve the error, then we return + */ + calledCallback=true; + } + } +} + +/* + * Output the fromUnicode overflow buffer. + * Call this function if(cnv->charErrorBufferLength>0). + * @return true if overflow + */ +static UBool +ucnv_outputOverflowFromUnicode(UConverter *cnv, + char **target, const char *targetLimit, + int32_t **pOffsets, + UErrorCode *err) { + int32_t *offsets; + char *overflow, *t; + int32_t i, length; + + t=*target; + if(pOffsets!=nullptr) { + offsets=*pOffsets; + } else { + offsets=nullptr; + } + + overflow=(char *)cnv->charErrorBuffer; + length=cnv->charErrorBufferLength; + i=0; + while(icharErrorBufferLength=(int8_t)j; + *target=t; + if(offsets!=nullptr) { + *pOffsets=offsets; + } + *err=U_BUFFER_OVERFLOW_ERROR; + return true; + } + + /* copy the overflow contents to the target */ + *t++=overflow[i++]; + if(offsets!=nullptr) { + *offsets++=-1; /* no source index available for old output */ + } + } + + /* the overflow buffer is completely copied to the target */ + cnv->charErrorBufferLength=0; + *target=t; + if(offsets!=nullptr) { + *pOffsets=offsets; + } + return false; +} + +U_CAPI void U_EXPORT2 +ucnv_fromUnicode(UConverter *cnv, + char **target, const char *targetLimit, + const char16_t **source, const char16_t *sourceLimit, + int32_t *offsets, + UBool flush, + UErrorCode *err) { + UConverterFromUnicodeArgs args; + const char16_t *s; + char *t; + + /* check parameters */ + if(err==nullptr || U_FAILURE(*err)) { + return; + } + + if(cnv==nullptr || target==nullptr || source==nullptr) { + *err=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + s=*source; + t=*target; + + if ((const void *)U_MAX_PTR(sourceLimit) == (const void *)sourceLimit) { + /* + Prevent code from going into an infinite loop in case we do hit this + limit. The limit pointer is expected to be on a char16_t * boundary. + This also prevents the next argument check from failing. + */ + sourceLimit = (const char16_t *)(((const char *)sourceLimit) - 1); + } + + /* + * All these conditions should never happen. + * + * 1) Make sure that the limits are >= to the address source or target + * + * 2) Make sure that the buffer sizes do not exceed the number range for + * int32_t because some functions use the size (in units or bytes) + * rather than comparing pointers, and because offsets are int32_t values. + * + * size_t is guaranteed to be unsigned and large enough for the job. + * + * Return with an error instead of adjusting the limits because we would + * not be able to maintain the semantics that either the source must be + * consumed or the target filled (unless an error occurs). + * An adjustment would be targetLimit=t+0x7fffffff; for example. + * + * 3) Make sure that the user didn't incorrectly cast a char16_t * pointer + * to a char * pointer and provide an incomplete char16_t code unit. + */ + if (sourceLimit(size_t)0x3fffffff && sourceLimit>s) || + ((size_t)(targetLimit-t)>(size_t)0x7fffffff && targetLimit>t) || + (((const char *)sourceLimit-(const char *)s) & 1) != 0) + { + *err=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + /* output the target overflow buffer */ + if( cnv->charErrorBufferLength>0 && + ucnv_outputOverflowFromUnicode(cnv, target, targetLimit, &offsets, err) + ) { + /* U_BUFFER_OVERFLOW_ERROR */ + return; + } + /* *target may have moved, therefore stop using t */ + + if(!flush && s==sourceLimit && cnv->preFromULength>=0) { + /* the overflow buffer is emptied and there is no new input: we are done */ + return; + } + + /* + * Do not simply return with a buffer overflow error if + * !flush && t==targetLimit + * because it is possible that the source will not generate any output. + * For example, the skip callback may be called; + * it does not output anything. + */ + + /* prepare the converter arguments */ + args.converter=cnv; + args.flush=flush; + args.offsets=offsets; + args.source=s; + args.sourceLimit=sourceLimit; + args.target=*target; + args.targetLimit=targetLimit; + args.size=sizeof(args); + + _fromUnicodeWithCallback(&args, err); + + *source=args.source; + *target=args.target; +} + +/* ucnv_toUnicode() --------------------------------------------------------- */ + +static void +_toUnicodeWithCallback(UConverterToUnicodeArgs *pArgs, UErrorCode *err) { + UConverterToUnicode toUnicode; + UConverter *cnv; + const char *s; + char16_t *t; + int32_t *offsets; + int32_t sourceIndex; + int32_t errorInputLength; + UBool converterSawEndOfInput, calledCallback; + + /* variables for m:n conversion */ + char replay[UCNV_EXT_MAX_BYTES]; + const char *realSource, *realSourceLimit; + int32_t realSourceIndex; + UBool realFlush; + + cnv=pArgs->converter; + s=pArgs->source; + t=pArgs->target; + offsets=pArgs->offsets; + + /* get the converter implementation function */ + sourceIndex=0; + if(offsets==nullptr) { + toUnicode=cnv->sharedData->impl->toUnicode; + } else { + toUnicode=cnv->sharedData->impl->toUnicodeWithOffsets; + if(toUnicode==nullptr) { + /* there is no WithOffsets implementation */ + toUnicode=cnv->sharedData->impl->toUnicode; + /* we will write -1 for each offset */ + sourceIndex=-1; + } + } + + if(cnv->preToULength>=0) { + /* normal mode */ + realSource=nullptr; + + /* avoid compiler warnings - not otherwise necessary, and the values do not matter */ + realSourceLimit=nullptr; + realFlush=false; + realSourceIndex=0; + } else { + /* + * Previous m:n conversion stored source units from a partial match + * and failed to consume all of them. + * We need to "replay" them from a temporary buffer and convert them first. + */ + realSource=pArgs->source; + realSourceLimit=pArgs->sourceLimit; + realFlush=pArgs->flush; + realSourceIndex=sourceIndex; + + uprv_memcpy(replay, cnv->preToU, -cnv->preToULength); + pArgs->source=replay; + pArgs->sourceLimit=replay-cnv->preToULength; + pArgs->flush=false; + sourceIndex=-1; + + cnv->preToULength=0; + } + + /* + * loop for conversion and error handling + * + * loop { + * convert + * loop { + * update offsets + * handle end of input + * handle errors/call callback + * } + * } + */ + for(;;) { + if(U_SUCCESS(*err)) { + /* convert */ + toUnicode(pArgs, err); + + /* + * set a flag for whether the converter + * successfully processed the end of the input + * + * need not check cnv->preToULength==0 because a replay (<0) will cause + * sflush && pArgs->source==pArgs->sourceLimit && + cnv->toULength==0); + } else { + /* handle error from getNextUChar() or ucnv_convertEx() */ + converterSawEndOfInput=false; + } + + /* no callback called yet for this iteration */ + calledCallback=false; + + /* no sourceIndex adjustment for conversion, only for callback output */ + errorInputLength=0; + + /* + * loop for offsets and error handling + * + * iterates at most 3 times: + * 1. to clean up after the conversion function + * 2. after the callback + * 3. after the callback again if there was truncated input + */ + for(;;) { + /* update offsets if we write any */ + if(offsets!=nullptr) { + int32_t length=(int32_t)(pArgs->target-t); + if(length>0) { + _updateOffsets(offsets, length, sourceIndex, errorInputLength); + + /* + * if a converter handles offsets and updates the offsets + * pointer at the end, then pArgs->offset should not change + * here; + * however, some converters do not handle offsets at all + * (sourceIndex<0) or may not update the offsets pointer + */ + pArgs->offsets=offsets+=length; + } + + if(sourceIndex>=0) { + sourceIndex+=(int32_t)(pArgs->source-s); + } + } + + if(cnv->preToULength<0) { + /* + * switch the source to new replay units (cannot occur while replaying) + * after offset handling and before end-of-input and callback handling + */ + if(realSource==nullptr) { + realSource=pArgs->source; + realSourceLimit=pArgs->sourceLimit; + realFlush=pArgs->flush; + realSourceIndex=sourceIndex; + + uprv_memcpy(replay, cnv->preToU, -cnv->preToULength); + pArgs->source=replay; + pArgs->sourceLimit=replay-cnv->preToULength; + pArgs->flush=false; + if((sourceIndex+=cnv->preToULength)<0) { + sourceIndex=-1; + } + + cnv->preToULength=0; + } else { + /* see implementation note before _fromUnicodeWithCallback() */ + U_ASSERT(realSource==nullptr); + *err=U_INTERNAL_PROGRAM_ERROR; + } + } + + /* update pointers */ + s=pArgs->source; + t=pArgs->target; + + if(U_SUCCESS(*err)) { + if(ssourceLimit) { + /* + * continue with the conversion loop while there is still input left + * (continue converting by breaking out of only the inner loop) + */ + break; + } else if(realSource!=nullptr) { + /* switch back from replaying to the real source and continue */ + pArgs->source=realSource; + pArgs->sourceLimit=realSourceLimit; + pArgs->flush=realFlush; + sourceIndex=realSourceIndex; + + realSource=nullptr; + break; + } else if(pArgs->flush && cnv->toULength>0) { + /* + * the entire input stream is consumed + * and there is a partial, truncated input sequence left + */ + + /* inject an error and continue with callback handling */ + *err=U_TRUNCATED_CHAR_FOUND; + calledCallback=false; /* new error condition */ + } else { + /* input consumed */ + if(pArgs->flush) { + /* + * return to the conversion loop once more if the flush + * flag is set and the conversion function has not + * successfully processed the end of the input yet + * + * (continue converting by breaking out of only the inner loop) + */ + if(!converterSawEndOfInput) { + break; + } + + /* reset the converter without calling the callback function */ + _reset(cnv, UCNV_RESET_TO_UNICODE, false); + } + + /* done successfully */ + return; + } + } + + /* U_FAILURE(*err) */ + { + UErrorCode e; + + if( calledCallback || + (e=*err)==U_BUFFER_OVERFLOW_ERROR || + (e!=U_INVALID_CHAR_FOUND && + e!=U_ILLEGAL_CHAR_FOUND && + e!=U_TRUNCATED_CHAR_FOUND && + e!=U_ILLEGAL_ESCAPE_SEQUENCE && + e!=U_UNSUPPORTED_ESCAPE_SEQUENCE) + ) { + /* + * the callback did not or cannot resolve the error: + * set output pointers and return + * + * the check for buffer overflow is redundant but it is + * a high-runner case and hopefully documents the intent + * well + * + * if we were replaying, then the replay buffer must be + * copied back into the UConverter + * and the real arguments must be restored + */ + if(realSource!=nullptr) { + int32_t length; + + U_ASSERT(cnv->preToULength==0); + + length=(int32_t)(pArgs->sourceLimit-pArgs->source); + if(length>0) { + uprv_memcpy(cnv->preToU, pArgs->source, length); + cnv->preToULength=(int8_t)-length; + } + + pArgs->source=realSource; + pArgs->sourceLimit=realSourceLimit; + pArgs->flush=realFlush; + } + + return; + } + } + + /* copy toUBytes[] to invalidCharBuffer[] */ + errorInputLength=cnv->invalidCharLength=cnv->toULength; + if(errorInputLength>0) { + uprv_memcpy(cnv->invalidCharBuffer, cnv->toUBytes, errorInputLength); + } + + /* set the converter state to deal with the next character */ + cnv->toULength=0; + + /* call the callback function */ + if(cnv->toUCallbackReason==UCNV_ILLEGAL && *err==U_INVALID_CHAR_FOUND) { + cnv->toUCallbackReason = UCNV_UNASSIGNED; + } + cnv->fromCharErrorBehaviour(cnv->toUContext, pArgs, + cnv->invalidCharBuffer, errorInputLength, + cnv->toUCallbackReason, + err); + cnv->toUCallbackReason = UCNV_ILLEGAL; /* reset to default value */ + + /* + * loop back to the offset handling + * + * this flag will indicate after offset handling + * that a callback was called; + * if the callback did not resolve the error, then we return + */ + calledCallback=true; + } + } +} + +/* + * Output the toUnicode overflow buffer. + * Call this function if(cnv->UCharErrorBufferLength>0). + * @return true if overflow + */ +static UBool +ucnv_outputOverflowToUnicode(UConverter *cnv, + char16_t **target, const char16_t *targetLimit, + int32_t **pOffsets, + UErrorCode *err) { + int32_t *offsets; + char16_t *overflow, *t; + int32_t i, length; + + t=*target; + if(pOffsets!=nullptr) { + offsets=*pOffsets; + } else { + offsets=nullptr; + } + + overflow=cnv->UCharErrorBuffer; + length=cnv->UCharErrorBufferLength; + i=0; + while(iUCharErrorBufferLength=(int8_t)j; + *target=t; + if(offsets!=nullptr) { + *pOffsets=offsets; + } + *err=U_BUFFER_OVERFLOW_ERROR; + return true; + } + + /* copy the overflow contents to the target */ + *t++=overflow[i++]; + if(offsets!=nullptr) { + *offsets++=-1; /* no source index available for old output */ + } + } + + /* the overflow buffer is completely copied to the target */ + cnv->UCharErrorBufferLength=0; + *target=t; + if(offsets!=nullptr) { + *pOffsets=offsets; + } + return false; +} + +U_CAPI void U_EXPORT2 +ucnv_toUnicode(UConverter *cnv, + char16_t **target, const char16_t *targetLimit, + const char **source, const char *sourceLimit, + int32_t *offsets, + UBool flush, + UErrorCode *err) { + UConverterToUnicodeArgs args; + const char *s; + char16_t *t; + + /* check parameters */ + if(err==nullptr || U_FAILURE(*err)) { + return; + } + + if(cnv==nullptr || target==nullptr || source==nullptr) { + *err=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + s=*source; + t=*target; + + if ((const void *)U_MAX_PTR(targetLimit) == (const void *)targetLimit) { + /* + Prevent code from going into an infinite loop in case we do hit this + limit. The limit pointer is expected to be on a char16_t * boundary. + This also prevents the next argument check from failing. + */ + targetLimit = (const char16_t *)(((const char *)targetLimit) - 1); + } + + /* + * All these conditions should never happen. + * + * 1) Make sure that the limits are >= to the address source or target + * + * 2) Make sure that the buffer sizes do not exceed the number range for + * int32_t because some functions use the size (in units or bytes) + * rather than comparing pointers, and because offsets are int32_t values. + * + * size_t is guaranteed to be unsigned and large enough for the job. + * + * Return with an error instead of adjusting the limits because we would + * not be able to maintain the semantics that either the source must be + * consumed or the target filled (unless an error occurs). + * An adjustment would be sourceLimit=t+0x7fffffff; for example. + * + * 3) Make sure that the user didn't incorrectly cast a char16_t * pointer + * to a char * pointer and provide an incomplete char16_t code unit. + */ + if (sourceLimit(size_t)0x7fffffff && sourceLimit>s) || + ((size_t)(targetLimit-t)>(size_t)0x3fffffff && targetLimit>t) || + (((const char *)targetLimit-(const char *)t) & 1) != 0 + ) { + *err=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + /* output the target overflow buffer */ + if( cnv->UCharErrorBufferLength>0 && + ucnv_outputOverflowToUnicode(cnv, target, targetLimit, &offsets, err) + ) { + /* U_BUFFER_OVERFLOW_ERROR */ + return; + } + /* *target may have moved, therefore stop using t */ + + if(!flush && s==sourceLimit && cnv->preToULength>=0) { + /* the overflow buffer is emptied and there is no new input: we are done */ + return; + } + + /* + * Do not simply return with a buffer overflow error if + * !flush && t==targetLimit + * because it is possible that the source will not generate any output. + * For example, the skip callback may be called; + * it does not output anything. + */ + + /* prepare the converter arguments */ + args.converter=cnv; + args.flush=flush; + args.offsets=offsets; + args.source=s; + args.sourceLimit=sourceLimit; + args.target=*target; + args.targetLimit=targetLimit; + args.size=sizeof(args); + + _toUnicodeWithCallback(&args, err); + + *source=args.source; + *target=args.target; +} + +/* ucnv_to/fromUChars() ----------------------------------------------------- */ + +U_CAPI int32_t U_EXPORT2 +ucnv_fromUChars(UConverter *cnv, + char *dest, int32_t destCapacity, + const char16_t *src, int32_t srcLength, + UErrorCode *pErrorCode) { + const char16_t *srcLimit; + char *originalDest, *destLimit; + int32_t destLength; + + /* check arguments */ + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + if( cnv==nullptr || + destCapacity<0 || (destCapacity>0 && dest==nullptr) || + srcLength<-1 || (srcLength!=0 && src==nullptr) + ) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* initialize */ + ucnv_resetFromUnicode(cnv); + originalDest=dest; + if(srcLength==-1) { + srcLength=u_strlen(src); + } + if(srcLength>0) { + srcLimit=src+srcLength; + destCapacity=pinCapacity(dest, destCapacity); + destLimit=dest+destCapacity; + + /* perform the conversion */ + ucnv_fromUnicode(cnv, &dest, destLimit, &src, srcLimit, 0, true, pErrorCode); + destLength=(int32_t)(dest-originalDest); + + /* if an overflow occurs, then get the preflighting length */ + if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) { + char buffer[1024]; + + destLimit=buffer+sizeof(buffer); + do { + dest=buffer; + *pErrorCode=U_ZERO_ERROR; + ucnv_fromUnicode(cnv, &dest, destLimit, &src, srcLimit, 0, true, pErrorCode); + destLength+=(int32_t)(dest-buffer); + } while(*pErrorCode==U_BUFFER_OVERFLOW_ERROR); + } + } else { + destLength=0; + } + + return u_terminateChars(originalDest, destCapacity, destLength, pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +ucnv_toUChars(UConverter *cnv, + char16_t *dest, int32_t destCapacity, + const char *src, int32_t srcLength, + UErrorCode *pErrorCode) { + const char *srcLimit; + char16_t *originalDest, *destLimit; + int32_t destLength; + + /* check arguments */ + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + if( cnv==nullptr || + destCapacity<0 || (destCapacity>0 && dest==nullptr) || + srcLength<-1 || (srcLength!=0 && src==nullptr)) + { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* initialize */ + ucnv_resetToUnicode(cnv); + originalDest=dest; + if(srcLength==-1) { + srcLength=(int32_t)uprv_strlen(src); + } + if(srcLength>0) { + srcLimit=src+srcLength; + destCapacity=pinCapacity(dest, destCapacity); + destLimit=dest+destCapacity; + + /* perform the conversion */ + ucnv_toUnicode(cnv, &dest, destLimit, &src, srcLimit, 0, true, pErrorCode); + destLength=(int32_t)(dest-originalDest); + + /* if an overflow occurs, then get the preflighting length */ + if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) + { + char16_t buffer[1024]; + + destLimit=buffer+UPRV_LENGTHOF(buffer); + do { + dest=buffer; + *pErrorCode=U_ZERO_ERROR; + ucnv_toUnicode(cnv, &dest, destLimit, &src, srcLimit, 0, true, pErrorCode); + destLength+=(int32_t)(dest-buffer); + } + while(*pErrorCode==U_BUFFER_OVERFLOW_ERROR); + } + } else { + destLength=0; + } + + return u_terminateUChars(originalDest, destCapacity, destLength, pErrorCode); +} + +/* ucnv_getNextUChar() ------------------------------------------------------ */ + +U_CAPI UChar32 U_EXPORT2 +ucnv_getNextUChar(UConverter *cnv, + const char **source, const char *sourceLimit, + UErrorCode *err) { + UConverterToUnicodeArgs args; + char16_t buffer[U16_MAX_LENGTH]; + const char *s; + UChar32 c; + int32_t i, length; + + /* check parameters */ + if(err==nullptr || U_FAILURE(*err)) { + return 0xffff; + } + + if(cnv==nullptr || source==nullptr) { + *err=U_ILLEGAL_ARGUMENT_ERROR; + return 0xffff; + } + + s=*source; + if(sourceLimit(size_t)0x7fffffff && sourceLimit>s)) { + *err=U_ILLEGAL_ARGUMENT_ERROR; + return 0xffff; + } + + c=U_SENTINEL; + + /* flush the target overflow buffer */ + if(cnv->UCharErrorBufferLength>0) { + char16_t *overflow; + + overflow=cnv->UCharErrorBuffer; + i=0; + length=cnv->UCharErrorBufferLength; + U16_NEXT(overflow, i, length, c); + + /* move the remaining overflow contents up to the beginning */ + if((cnv->UCharErrorBufferLength=(int8_t)(length-i))>0) { + uprv_memmove(cnv->UCharErrorBuffer, cnv->UCharErrorBuffer+i, + cnv->UCharErrorBufferLength*U_SIZEOF_UCHAR); + } + + if(!U16_IS_LEAD(c) || itoULength==0 && cnv->sharedData->impl->getNextUChar!=nullptr) { + c=cnv->sharedData->impl->getNextUChar(&args, err); + *source=s=args.source; + if(*err==U_INDEX_OUTOFBOUNDS_ERROR) { + /* reset the converter without calling the callback function */ + _reset(cnv, UCNV_RESET_TO_UNICODE, false); + return 0xffff; /* no output */ + } else if(U_SUCCESS(*err) && c>=0) { + return c; + /* + * else fall through to use _toUnicode() because + * UCNV_GET_NEXT_UCHAR_USE_TO_U: the native function did not want to handle it after all + * U_FAILURE: call _toUnicode() for callback handling (do not output c) + */ + } + } + + /* convert to one char16_t in buffer[0], or handle getNextUChar() errors */ + _toUnicodeWithCallback(&args, err); + + if(*err==U_BUFFER_OVERFLOW_ERROR) { + *err=U_ZERO_ERROR; + } + + i=0; + length=(int32_t)(args.target-buffer); + } else { + /* write the lead surrogate from the overflow buffer */ + buffer[0]=(char16_t)c; + args.target=buffer+1; + i=0; + length=1; + } + + /* buffer contents starts at i and ends before length */ + + if(U_FAILURE(*err)) { + c=0xffff; /* no output */ + } else if(length==0) { + /* no input or only state changes */ + *err=U_INDEX_OUTOFBOUNDS_ERROR; + /* no need to reset explicitly because _toUnicodeWithCallback() did it */ + c=0xffff; /* no output */ + } else { + c=buffer[0]; + i=1; + if(!U16_IS_LEAD(c)) { + /* consume c=buffer[0], done */ + } else { + /* got a lead surrogate, see if a trail surrogate follows */ + char16_t c2; + + if(cnv->UCharErrorBufferLength>0) { + /* got overflow output from the conversion */ + if(U16_IS_TRAIL(c2=cnv->UCharErrorBuffer[0])) { + /* got a trail surrogate, too */ + c=U16_GET_SUPPLEMENTARY(c, c2); + + /* move the remaining overflow contents up to the beginning */ + if((--cnv->UCharErrorBufferLength)>0) { + uprv_memmove(cnv->UCharErrorBuffer, cnv->UCharErrorBuffer+1, + cnv->UCharErrorBufferLength*U_SIZEOF_UCHAR); + } + } else { + /* c is an unpaired lead surrogate, just return it */ + } + } else if(args.sourceUCharErrorBufferLength)>0) { + uprv_memmove(cnv->UCharErrorBuffer+delta, cnv->UCharErrorBuffer, + length*U_SIZEOF_UCHAR); + } + cnv->UCharErrorBufferLength=(int8_t)(length+delta); + + cnv->UCharErrorBuffer[0]=buffer[i++]; + if(delta>1) { + cnv->UCharErrorBuffer[1]=buffer[i]; + } + } + + *source=args.source; + return c; +} + +/* ucnv_convert() and siblings ---------------------------------------------- */ + +U_CAPI void U_EXPORT2 +ucnv_convertEx(UConverter *targetCnv, UConverter *sourceCnv, + char **target, const char *targetLimit, + const char **source, const char *sourceLimit, + char16_t *pivotStart, char16_t **pivotSource, + char16_t **pivotTarget, const char16_t *pivotLimit, + UBool reset, UBool flush, + UErrorCode *pErrorCode) { + char16_t pivotBuffer[CHUNK_SIZE]; + const char16_t *myPivotSource; + char16_t *myPivotTarget; + const char *s; + char *t; + + UConverterToUnicodeArgs toUArgs; + UConverterFromUnicodeArgs fromUArgs; + UConverterConvert convert; + + /* error checking */ + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return; + } + + if( targetCnv==nullptr || sourceCnv==nullptr || + source==nullptr || *source==nullptr || + target==nullptr || *target==nullptr || targetLimit==nullptr + ) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + s=*source; + t=*target; + if((sourceLimit!=nullptr && sourceLimit(size_t)0x7fffffff && sourceLimit>s)) || + ((size_t)(targetLimit-t)>(size_t)0x7fffffff && targetLimit>t) + ) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + if(pivotStart==nullptr) { + if(!flush) { + /* streaming conversion requires an explicit pivot buffer */ + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + /* use the stack pivot buffer */ + myPivotSource=myPivotTarget=pivotStart=pivotBuffer; + pivotSource=(char16_t **)&myPivotSource; + pivotTarget=&myPivotTarget; + pivotLimit=pivotBuffer+CHUNK_SIZE; + } else if( pivotStart>=pivotLimit || + pivotSource==nullptr || *pivotSource==nullptr || + pivotTarget==nullptr || *pivotTarget==nullptr || + pivotLimit==nullptr + ) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + if(sourceLimit==nullptr) { + /* get limit of single-byte-NUL-terminated source string */ + sourceLimit=uprv_strchr(*source, 0); + } + + if(reset) { + ucnv_resetToUnicode(sourceCnv); + ucnv_resetFromUnicode(targetCnv); + *pivotSource=*pivotTarget=pivotStart; + } else if(targetCnv->charErrorBufferLength>0) { + /* output the targetCnv overflow buffer */ + if(ucnv_outputOverflowFromUnicode(targetCnv, target, targetLimit, nullptr, pErrorCode)) { + /* U_BUFFER_OVERFLOW_ERROR */ + return; + } + /* *target has moved, therefore stop using t */ + + if( !flush && + targetCnv->preFromULength>=0 && *pivotSource==*pivotTarget && + sourceCnv->UCharErrorBufferLength==0 && sourceCnv->preToULength>=0 && s==sourceLimit + ) { + /* the fromUnicode overflow buffer is emptied and there is no new input: we are done */ + return; + } + } + + /* Is direct-UTF-8 conversion available? */ + if( sourceCnv->sharedData->staticData->conversionType==UCNV_UTF8 && + targetCnv->sharedData->impl->fromUTF8!=nullptr + ) { + convert=targetCnv->sharedData->impl->fromUTF8; + } else if( targetCnv->sharedData->staticData->conversionType==UCNV_UTF8 && + sourceCnv->sharedData->impl->toUTF8!=nullptr + ) { + convert=sourceCnv->sharedData->impl->toUTF8; + } else { + convert=nullptr; + } + + /* + * If direct-UTF-8 conversion is available, then we use a smaller + * pivot buffer for error handling and partial matches + * so that we quickly return to direct conversion. + * + * 32 is large enough for UCNV_EXT_MAX_UCHARS and UCNV_ERROR_BUFFER_LENGTH. + * + * We could reduce the pivot buffer size further, at the cost of + * buffer overflows from callbacks. + * The pivot buffer should not be smaller than the maximum number of + * fromUnicode extension table input UChars + * (for m:n conversion, see + * targetCnv->sharedData->mbcs.extIndexes[UCNV_EXT_COUNT_UCHARS]) + * or 2 for surrogate pairs. + * + * Too small a buffer can cause thrashing between pivoting and direct + * conversion, with function call overhead outweighing the benefits + * of direct conversion. + */ + if(convert!=nullptr && (pivotLimit-pivotStart)>32) { + pivotLimit=pivotStart+32; + } + + /* prepare the converter arguments */ + fromUArgs.converter=targetCnv; + fromUArgs.flush=false; + fromUArgs.offsets=nullptr; + fromUArgs.target=*target; + fromUArgs.targetLimit=targetLimit; + fromUArgs.size=sizeof(fromUArgs); + + toUArgs.converter=sourceCnv; + toUArgs.flush=flush; + toUArgs.offsets=nullptr; + toUArgs.source=s; + toUArgs.sourceLimit=sourceLimit; + toUArgs.targetLimit=pivotLimit; + toUArgs.size=sizeof(toUArgs); + + /* + * TODO: Consider separating this function into two functions, + * extracting exactly the conversion loop, + * for readability and to reduce the set of visible variables. + * + * Otherwise stop using s and t from here on. + */ + s=t=nullptr; + + /* + * conversion loop + * + * The sequence of steps in the loop may appear backward, + * but the principle is simple: + * In the chain of + * source - sourceCnv overflow - pivot - targetCnv overflow - target + * empty out later buffers before refilling them from earlier ones. + * + * The targetCnv overflow buffer is flushed out only once before the loop. + */ + for(;;) { + /* + * if(pivot not empty or error or replay or flush fromUnicode) { + * fromUnicode(pivot -> target); + * } + * + * For pivoting conversion; and for direct conversion for + * error callback handling and flushing the replay buffer. + */ + if( *pivotSource<*pivotTarget || + U_FAILURE(*pErrorCode) || + targetCnv->preFromULength<0 || + fromUArgs.flush + ) { + fromUArgs.source=*pivotSource; + fromUArgs.sourceLimit=*pivotTarget; + _fromUnicodeWithCallback(&fromUArgs, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + /* target overflow, or conversion error */ + *pivotSource=(char16_t *)fromUArgs.source; + break; + } + + /* + * _fromUnicodeWithCallback() must have consumed the pivot contents + * (*pivotSource==*pivotTarget) since it returned with U_SUCCESS() + */ + } + + /* The pivot buffer is empty; reset it so we start at pivotStart. */ + *pivotSource=*pivotTarget=pivotStart; + + /* + * if(sourceCnv overflow buffer not empty) { + * move(sourceCnv overflow buffer -> pivot); + * continue; + * } + */ + /* output the sourceCnv overflow buffer */ + if(sourceCnv->UCharErrorBufferLength>0) { + if(ucnv_outputOverflowToUnicode(sourceCnv, pivotTarget, pivotLimit, nullptr, pErrorCode)) { + /* U_BUFFER_OVERFLOW_ERROR */ + *pErrorCode=U_ZERO_ERROR; + } + continue; + } + + /* + * check for end of input and break if done + * + * Checking both flush and fromUArgs.flush ensures that the converters + * have been called with the flush flag set if the ucnv_convertEx() + * caller set it. + */ + if( toUArgs.source==sourceLimit && + sourceCnv->preToULength>=0 && sourceCnv->toULength==0 && + (!flush || fromUArgs.flush) + ) { + /* done successfully */ + break; + } + + /* + * use direct conversion if available + * but not if continuing a partial match + * or flushing the toUnicode replay buffer + */ + if(convert!=nullptr && targetCnv->preFromUFirstCP<0 && sourceCnv->preToULength==0) { + if(*pErrorCode==U_USING_DEFAULT_WARNING) { + /* remove a warning that may be set by this function */ + *pErrorCode=U_ZERO_ERROR; + } + convert(&fromUArgs, &toUArgs, pErrorCode); + if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) { + break; + } else if(U_FAILURE(*pErrorCode)) { + if(sourceCnv->toULength>0) { + /* + * Fall through to calling _toUnicodeWithCallback() + * for callback handling. + * + * The pivot buffer will be reset with + * *pivotSource=*pivotTarget=pivotStart; + * which indicates a toUnicode error to the caller + * (*pivotSource==pivotStart shows no pivot UChars consumed). + */ + } else { + /* + * Indicate a fromUnicode error to the caller + * (*pivotSource>pivotStart shows some pivot UChars consumed). + */ + *pivotSource=*pivotTarget=pivotStart+1; + /* + * Loop around to calling _fromUnicodeWithCallbacks() + * for callback handling. + */ + continue; + } + } else if(*pErrorCode==U_USING_DEFAULT_WARNING) { + /* + * No error, but the implementation requested to temporarily + * fall back to pivoting. + */ + *pErrorCode=U_ZERO_ERROR; + /* + * The following else branches are almost identical to the end-of-input + * handling in _toUnicodeWithCallback(). + * Avoid calling it just for the end of input. + */ + } else if(flush && sourceCnv->toULength>0) { /* flush==toUArgs.flush */ + /* + * the entire input stream is consumed + * and there is a partial, truncated input sequence left + */ + + /* inject an error and continue with callback handling */ + *pErrorCode=U_TRUNCATED_CHAR_FOUND; + } else { + /* input consumed */ + if(flush) { + /* reset the converters without calling the callback functions */ + _reset(sourceCnv, UCNV_RESET_TO_UNICODE, false); + _reset(targetCnv, UCNV_RESET_FROM_UNICODE, false); + } + + /* done successfully */ + break; + } + } + + /* + * toUnicode(source -> pivot); + * + * For pivoting conversion; and for direct conversion for + * error callback handling, continuing partial matches + * and flushing the replay buffer. + * + * The pivot buffer is empty and reset. + */ + toUArgs.target=pivotStart; /* ==*pivotTarget */ + /* toUArgs.targetLimit=pivotLimit; already set before the loop */ + _toUnicodeWithCallback(&toUArgs, pErrorCode); + *pivotTarget=toUArgs.target; + if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) { + /* pivot overflow: continue with the conversion loop */ + *pErrorCode=U_ZERO_ERROR; + } else if(U_FAILURE(*pErrorCode) || (!flush && *pivotTarget==pivotStart)) { + /* conversion error, or there was nothing left to convert */ + break; + } + /* + * else: + * _toUnicodeWithCallback() wrote into the pivot buffer, + * continue with fromUnicode conversion. + * + * Set the fromUnicode flush flag if we flush and if toUnicode has + * processed the end of the input. + */ + if( flush && toUArgs.source==sourceLimit && + sourceCnv->preToULength>=0 && + sourceCnv->UCharErrorBufferLength==0 + ) { + fromUArgs.flush=true; + } + } + + /* + * The conversion loop is exited when one of the following is true: + * - the entire source text has been converted successfully to the target buffer + * - a target buffer overflow occurred + * - a conversion error occurred + */ + + *source=toUArgs.source; + *target=fromUArgs.target; + + /* terminate the target buffer if possible */ + if(flush && U_SUCCESS(*pErrorCode)) { + if(*target!=targetLimit) { + **target=0; + if(*pErrorCode==U_STRING_NOT_TERMINATED_WARNING) { + *pErrorCode=U_ZERO_ERROR; + } + } else { + *pErrorCode=U_STRING_NOT_TERMINATED_WARNING; + } + } +} + +/* internal implementation of ucnv_convert() etc. with preflighting */ +static int32_t +ucnv_internalConvert(UConverter *outConverter, UConverter *inConverter, + char *target, int32_t targetCapacity, + const char *source, int32_t sourceLength, + UErrorCode *pErrorCode) { + char16_t pivotBuffer[CHUNK_SIZE]; + char16_t *pivot, *pivot2; + + char *myTarget; + const char *sourceLimit; + const char *targetLimit; + int32_t targetLength=0; + + /* set up */ + if(sourceLength<0) { + sourceLimit=uprv_strchr(source, 0); + } else { + sourceLimit=source+sourceLength; + } + + /* if there is no input data, we're done */ + if(source==sourceLimit) { + return u_terminateChars(target, targetCapacity, 0, pErrorCode); + } + + pivot=pivot2=pivotBuffer; + myTarget=target; + targetLength=0; + + if(targetCapacity>0) { + /* perform real conversion */ + targetLimit=target+targetCapacity; + ucnv_convertEx(outConverter, inConverter, + &myTarget, targetLimit, + &source, sourceLimit, + pivotBuffer, &pivot, &pivot2, pivotBuffer+CHUNK_SIZE, + false, + true, + pErrorCode); + targetLength=(int32_t)(myTarget-target); + } + + /* + * If the output buffer is exhausted (or we are only "preflighting"), we need to stop writing + * to it but continue the conversion in order to store in targetCapacity + * the number of bytes that was required. + */ + if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR || targetCapacity==0) + { + char targetBuffer[CHUNK_SIZE]; + + targetLimit=targetBuffer+CHUNK_SIZE; + do { + *pErrorCode=U_ZERO_ERROR; + myTarget=targetBuffer; + ucnv_convertEx(outConverter, inConverter, + &myTarget, targetLimit, + &source, sourceLimit, + pivotBuffer, &pivot, &pivot2, pivotBuffer+CHUNK_SIZE, + false, + true, + pErrorCode); + targetLength+=(int32_t)(myTarget-targetBuffer); + } while(*pErrorCode==U_BUFFER_OVERFLOW_ERROR); + + /* done with preflighting, set warnings and errors as appropriate */ + return u_terminateChars(target, targetCapacity, targetLength, pErrorCode); + } + + /* no need to call u_terminateChars() because ucnv_convertEx() took care of that */ + return targetLength; +} + +U_CAPI int32_t U_EXPORT2 +ucnv_convert(const char *toConverterName, const char *fromConverterName, + char *target, int32_t targetCapacity, + const char *source, int32_t sourceLength, + UErrorCode *pErrorCode) { + UConverter in, out; /* stack-allocated */ + UConverter *inConverter, *outConverter; + int32_t targetLength; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + if( source==nullptr || sourceLength<-1 || + targetCapacity<0 || (targetCapacity>0 && target==nullptr) + ) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* if there is no input data, we're done */ + if(sourceLength==0 || (sourceLength<0 && *source==0)) { + return u_terminateChars(target, targetCapacity, 0, pErrorCode); + } + + /* create the converters */ + inConverter=ucnv_createConverter(&in, fromConverterName, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + return 0; + } + + outConverter=ucnv_createConverter(&out, toConverterName, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + ucnv_close(inConverter); + return 0; + } + + targetLength=ucnv_internalConvert(outConverter, inConverter, + target, targetCapacity, + source, sourceLength, + pErrorCode); + + ucnv_close(inConverter); + ucnv_close(outConverter); + + return targetLength; +} + +/* @internal */ +static int32_t +ucnv_convertAlgorithmic(UBool convertToAlgorithmic, + UConverterType algorithmicType, + UConverter *cnv, + char *target, int32_t targetCapacity, + const char *source, int32_t sourceLength, + UErrorCode *pErrorCode) { + UConverter algoConverterStatic; /* stack-allocated */ + UConverter *algoConverter, *to, *from; + int32_t targetLength; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + if( cnv==nullptr || source==nullptr || sourceLength<-1 || + targetCapacity<0 || (targetCapacity>0 && target==nullptr) + ) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* if there is no input data, we're done */ + if(sourceLength==0 || (sourceLength<0 && *source==0)) { + return u_terminateChars(target, targetCapacity, 0, pErrorCode); + } + + /* create the algorithmic converter */ + algoConverter=ucnv_createAlgorithmicConverter(&algoConverterStatic, algorithmicType, + "", 0, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + return 0; + } + + /* reset the other converter */ + if(convertToAlgorithmic) { + /* cnv->Unicode->algo */ + ucnv_resetToUnicode(cnv); + to=algoConverter; + from=cnv; + } else { + /* algo->Unicode->cnv */ + ucnv_resetFromUnicode(cnv); + from=algoConverter; + to=cnv; + } + + targetLength=ucnv_internalConvert(to, from, + target, targetCapacity, + source, sourceLength, + pErrorCode); + + ucnv_close(algoConverter); + + return targetLength; +} + +U_CAPI int32_t U_EXPORT2 +ucnv_toAlgorithmic(UConverterType algorithmicType, + UConverter *cnv, + char *target, int32_t targetCapacity, + const char *source, int32_t sourceLength, + UErrorCode *pErrorCode) { + return ucnv_convertAlgorithmic(true, algorithmicType, cnv, + target, targetCapacity, + source, sourceLength, + pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +ucnv_fromAlgorithmic(UConverter *cnv, + UConverterType algorithmicType, + char *target, int32_t targetCapacity, + const char *source, int32_t sourceLength, + UErrorCode *pErrorCode) UPRV_NO_SANITIZE_UNDEFINED { + + if(algorithmicType<0 || UCNV_NUMBER_OF_SUPPORTED_CONVERTER_TYPES<=algorithmicType) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + return ucnv_convertAlgorithmic(false, algorithmicType, cnv, + target, targetCapacity, + source, sourceLength, + pErrorCode); +} + +U_CAPI UConverterType U_EXPORT2 +ucnv_getType(const UConverter* converter) +{ + int8_t type = converter->sharedData->staticData->conversionType; +#if !UCONFIG_NO_LEGACY_CONVERSION + if(type == UCNV_MBCS) { + return ucnv_MBCSGetType(converter); + } +#endif + return (UConverterType)type; +} + +U_CAPI void U_EXPORT2 +ucnv_getStarters(const UConverter* converter, + UBool starters[256], + UErrorCode* err) +{ + if (err == nullptr || U_FAILURE(*err)) { + return; + } + + if(converter->sharedData->impl->getStarters != nullptr) { + converter->sharedData->impl->getStarters(converter, starters, err); + } else { + *err = U_ILLEGAL_ARGUMENT_ERROR; + } +} + +static const UAmbiguousConverter *ucnv_getAmbiguous(const UConverter *cnv) +{ + UErrorCode errorCode; + const char *name; + int32_t i; + + if(cnv==nullptr) { + return nullptr; + } + + errorCode=U_ZERO_ERROR; + name=ucnv_getName(cnv, &errorCode); + if(U_FAILURE(errorCode)) { + return nullptr; + } + + for(i=0; ivariant5c; + for(i=0; iuseFallback = usesFallback; +} + +U_CAPI UBool U_EXPORT2 +ucnv_usesFallback(const UConverter *cnv) +{ + return cnv->useFallback; +} + +U_CAPI void U_EXPORT2 +ucnv_getInvalidChars (const UConverter * converter, + char *errBytes, + int8_t * len, + UErrorCode * err) +{ + if (err == nullptr || U_FAILURE(*err)) + { + return; + } + if (len == nullptr || errBytes == nullptr || converter == nullptr) + { + *err = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + if (*len < converter->invalidCharLength) + { + *err = U_INDEX_OUTOFBOUNDS_ERROR; + return; + } + if ((*len = converter->invalidCharLength) > 0) + { + uprv_memcpy (errBytes, converter->invalidCharBuffer, *len); + } +} + +U_CAPI void U_EXPORT2 +ucnv_getInvalidUChars (const UConverter * converter, + char16_t *errChars, + int8_t * len, + UErrorCode * err) +{ + if (err == nullptr || U_FAILURE(*err)) + { + return; + } + if (len == nullptr || errChars == nullptr || converter == nullptr) + { + *err = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + if (*len < converter->invalidUCharLength) + { + *err = U_INDEX_OUTOFBOUNDS_ERROR; + return; + } + if ((*len = converter->invalidUCharLength) > 0) + { + u_memcpy (errChars, converter->invalidUCharBuffer, *len); + } +} + +#define SIG_MAX_LEN 5 + +U_CAPI const char* U_EXPORT2 +ucnv_detectUnicodeSignature( const char* source, + int32_t sourceLength, + int32_t* signatureLength, + UErrorCode* pErrorCode) { + int32_t dummy; + + /* initial 0xa5 bytes: make sure that if we read preFromUFirstCP >= 0){ + return U16_LENGTH(cnv->preFromUFirstCP)+cnv->preFromULength ; + }else if(cnv->preFromULength < 0){ + return -cnv->preFromULength ; + }else if(cnv->fromUChar32 > 0){ + return 1; + } + return 0; + +} + +U_CAPI int32_t U_EXPORT2 +ucnv_toUCountPending(const UConverter* cnv, UErrorCode* status){ + + if(status == nullptr || U_FAILURE(*status)){ + return -1; + } + if(cnv == nullptr){ + *status = U_ILLEGAL_ARGUMENT_ERROR; + return -1; + } + + if(cnv->preToULength > 0){ + return cnv->preToULength ; + }else if(cnv->preToULength < 0){ + return -cnv->preToULength; + }else if(cnv->toULength > 0){ + return cnv->toULength; + } + return 0; +} + +U_CAPI UBool U_EXPORT2 +ucnv_isFixedWidth(UConverter *cnv, UErrorCode *status){ + if (U_FAILURE(*status)) { + return false; + } + + if (cnv == nullptr) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return false; + } + + switch (ucnv_getType(cnv)) { + case UCNV_SBCS: + case UCNV_DBCS: + case UCNV_UTF32_BigEndian: + case UCNV_UTF32_LittleEndian: + case UCNV_UTF32: + case UCNV_US_ASCII: + return true; + default: + return false; + } +} +#endif + +/* + * Hey, Emacs, please set the following: + * + * Local Variables: + * indent-tabs-mode: nil + * End: + * + */ diff --git a/intl/icu/source/common/ucnv2022.cpp b/intl/icu/source/common/ucnv2022.cpp new file mode 100644 index 0000000000..5989c1b405 --- /dev/null +++ b/intl/icu/source/common/ucnv2022.cpp @@ -0,0 +1,3973 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2000-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* file name: ucnv2022.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2000feb03 +* created by: Markus W. Scherer +* +* Change history: +* +* 06/29/2000 helena Major rewrite of the callback APIs. +* 08/08/2000 Ram Included support for ISO-2022-JP-2 +* Changed implementation of toUnicode +* function +* 08/21/2000 Ram Added support for ISO-2022-KR +* 08/29/2000 Ram Seperated implementation of EBCDIC to +* ucnvebdc.c +* 09/20/2000 Ram Added support for ISO-2022-CN +* Added implementations for getNextUChar() +* for specific 2022 country variants. +* 10/31/2000 Ram Implemented offsets logic functions +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION && !UCONFIG_NO_LEGACY_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/uset.h" +#include "unicode/ucnv_err.h" +#include "unicode/ucnv_cb.h" +#include "unicode/utf16.h" +#include "ucnv_imp.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" +#include "ucnvmbcs.h" +#include "cstring.h" +#include "cmemory.h" +#include "uassert.h" + +#ifdef U_ENABLE_GENERIC_ISO_2022 +/* + * I am disabling the generic ISO-2022 converter after proposing to do so on + * the icu mailing list two days ago. + * + * Reasons: + * 1. It does not fully support the ISO-2022/ECMA-35 specification with all of + * its designation sequences, single shifts with return to the previous state, + * switch-with-no-return to UTF-16BE or similar, etc. + * This is unlike the language-specific variants like ISO-2022-JP which + * require a much smaller repertoire of ISO-2022 features. + * These variants continue to be supported. + * 2. I believe that no one is really using the generic ISO-2022 converter + * but rather always one of the language-specific variants. + * Note that ICU's generic ISO-2022 converter has always output one escape + * sequence followed by UTF-8 for the whole stream. + * 3. Switching between subcharsets is extremely slow, because each time + * the previous converter is closed and a new one opened, + * without any kind of caching, least-recently-used list, etc. + * 4. The code is currently buggy, and given the above it does not seem + * reasonable to spend the time on maintenance. + * 5. ISO-2022 subcharsets should normally be used with 7-bit byte encodings. + * This means, for example, that when ISO-8859-7 is designated, the following + * ISO-2022 bytes 00..7f should be interpreted as ISO-8859-7 bytes 80..ff. + * The ICU ISO-2022 converter does not handle this - and has no information + * about which subconverter would have to be shifted vs. which is designed + * for 7-bit ISO-2022. + * + * Markus Scherer 2003-dec-03 + */ +#endif + +#if !UCONFIG_ONLY_HTML_CONVERSION +static const char SHIFT_IN_STR[] = "\x0F"; +// static const char SHIFT_OUT_STR[] = "\x0E"; +#endif + +#define CR 0x0D +#define LF 0x0A +#define H_TAB 0x09 +#define V_TAB 0x0B +#define SPACE 0x20 + +enum { + HWKANA_START=0xff61, + HWKANA_END=0xff9f +}; + +/* + * 94-character sets with native byte values A1..FE are encoded in ISO 2022 + * as bytes 21..7E. (Subtract 0x80.) + * 96-character sets with native byte values A0..FF are encoded in ISO 2022 + * as bytes 20..7F. (Subtract 0x80.) + * Do not encode C1 control codes with native bytes 80..9F + * as bytes 00..1F (C0 control codes). + */ +enum { + GR94_START=0xa1, + GR94_END=0xfe, + GR96_START=0xa0, + GR96_END=0xff +}; + +/* + * ISO 2022 control codes must not be converted from Unicode + * because they would mess up the byte stream. + * The bit mask 0x0800c000 has bits set at bit positions 0xe, 0xf, 0x1b + * corresponding to SO, SI, and ESC. + */ +#define IS_2022_CONTROL(c) (((c)<0x20) && (((uint32_t)1<<(c))&0x0800c000)!=0) + +/* for ISO-2022-JP and -CN implementations */ +typedef enum { + /* shared values */ + INVALID_STATE=-1, + ASCII = 0, + + SS2_STATE=0x10, + SS3_STATE, + + /* JP */ + ISO8859_1 = 1 , + ISO8859_7 = 2 , + JISX201 = 3, + JISX208 = 4, + JISX212 = 5, + GB2312 =6, + KSC5601 =7, + HWKANA_7BIT=8, /* Halfwidth Katakana 7 bit */ + + /* CN */ + /* the first few enum constants must keep their values because they correspond to myConverterArray[] */ + GB2312_1=1, + ISO_IR_165=2, + CNS_11643=3, + + /* + * these are used in StateEnum and ISO2022State variables, + * but CNS_11643 must be used to index into myConverterArray[] + */ + CNS_11643_0=0x20, + CNS_11643_1, + CNS_11643_2, + CNS_11643_3, + CNS_11643_4, + CNS_11643_5, + CNS_11643_6, + CNS_11643_7 +} StateEnum; + +/* is the StateEnum charset value for a DBCS charset? */ +#if UCONFIG_ONLY_HTML_CONVERSION +#define IS_JP_DBCS(cs) (JISX208==(cs)) +#else +#define IS_JP_DBCS(cs) (JISX208<=(cs) && (cs)<=KSC5601) +#endif + +#define CSM(cs) ((uint16_t)1<<(cs)) + +/* + * Each of these charset masks (with index x) contains a bit for a charset in exact correspondence + * to whether that charset is used in the corresponding version x of ISO_2022,locale=ja,version=x + * + * Note: The converter uses some leniency: + * - The escape sequence ESC ( I for half-width 7-bit Katakana is recognized in + * all versions, not just JIS7 and JIS8. + * - ICU does not distinguish between different versions of JIS X 0208. + */ +#if UCONFIG_ONLY_HTML_CONVERSION +enum { MAX_JA_VERSION=0 }; +#else +enum { MAX_JA_VERSION=4 }; +#endif +static const uint16_t jpCharsetMasks[MAX_JA_VERSION+1]={ + CSM(ASCII)|CSM(JISX201)|CSM(JISX208)|CSM(HWKANA_7BIT), +#if !UCONFIG_ONLY_HTML_CONVERSION + CSM(ASCII)|CSM(JISX201)|CSM(JISX208)|CSM(HWKANA_7BIT)|CSM(JISX212), + CSM(ASCII)|CSM(JISX201)|CSM(JISX208)|CSM(HWKANA_7BIT)|CSM(JISX212)|CSM(GB2312)|CSM(KSC5601)|CSM(ISO8859_1)|CSM(ISO8859_7), + CSM(ASCII)|CSM(JISX201)|CSM(JISX208)|CSM(HWKANA_7BIT)|CSM(JISX212)|CSM(GB2312)|CSM(KSC5601)|CSM(ISO8859_1)|CSM(ISO8859_7), + CSM(ASCII)|CSM(JISX201)|CSM(JISX208)|CSM(HWKANA_7BIT)|CSM(JISX212)|CSM(GB2312)|CSM(KSC5601)|CSM(ISO8859_1)|CSM(ISO8859_7) +#endif +}; + +typedef enum { + ASCII1=0, + LATIN1, + SBCS, + DBCS, + MBCS, + HWKANA +}Cnv2022Type; + +typedef struct ISO2022State { + int8_t cs[4]; /* charset number for SI (G0)/SO (G1)/SS2 (G2)/SS3 (G3) */ + int8_t g; /* 0..3 for G0..G3 (SI/SO/SS2/SS3) */ + int8_t prevG; /* g before single shift (SS2 or SS3) */ +} ISO2022State; + +#define UCNV_OPTIONS_VERSION_MASK 0xf +#define UCNV_2022_MAX_CONVERTERS 10 + +typedef struct{ + UConverterSharedData *myConverterArray[UCNV_2022_MAX_CONVERTERS]; + UConverter *currentConverter; + Cnv2022Type currentType; + ISO2022State toU2022State, fromU2022State; + uint32_t key; + uint32_t version; +#ifdef U_ENABLE_GENERIC_ISO_2022 + UBool isFirstBuffer; +#endif + UBool isEmptySegment; + char name[30]; + char locale[3]; +}UConverterDataISO2022; + +/* Protos */ +/* ISO-2022 ----------------------------------------------------------------- */ + +/*Forward declaration */ +U_CFUNC void U_CALLCONV +ucnv_fromUnicode_UTF8(UConverterFromUnicodeArgs * args, + UErrorCode * err); +U_CFUNC void U_CALLCONV +ucnv_fromUnicode_UTF8_OFFSETS_LOGIC(UConverterFromUnicodeArgs * args, + UErrorCode * err); + +#define ESC_2022 0x1B /*ESC*/ + +typedef enum +{ + INVALID_2022 = -1, /*Doesn't correspond to a valid iso 2022 escape sequence*/ + VALID_NON_TERMINAL_2022 = 0, /*so far corresponds to a valid iso 2022 escape sequence*/ + VALID_TERMINAL_2022 = 1, /*corresponds to a valid iso 2022 escape sequence*/ + VALID_MAYBE_TERMINAL_2022 = 2 /*so far matches one iso 2022 escape sequence, but by adding more characters might match another escape sequence*/ +} UCNV_TableStates_2022; + +/* +* The way these state transition arrays work is: +* ex : ESC$B is the sequence for JISX208 +* a) First Iteration: char is ESC +* i) Get the value of ESC from normalize_esq_chars_2022[] with int value of ESC as index +* int x = normalize_esq_chars_2022[27] which is equal to 1 +* ii) Search for this value in escSeqStateTable_Key_2022[] +* value of x is stored at escSeqStateTable_Key_2022[0] +* iii) Save this index as offset +* iv) Get state of this sequence from escSeqStateTable_Value_2022[] +* escSeqStateTable_Value_2022[offset], which is VALID_NON_TERMINAL_2022 +* b) Switch on this state and continue to next char +* i) Get the value of $ from normalize_esq_chars_2022[] with int value of $ as index +* which is normalize_esq_chars_2022[36] == 4 +* ii) x is currently 1(from above) +* x<<=5 -- x is now 32 +* x+=normalize_esq_chars_2022[36] +* now x is 36 +* iii) Search for this value in escSeqStateTable_Key_2022[] +* value of x is stored at escSeqStateTable_Key_2022[2], so offset is 2 +* iv) Get state of this sequence from escSeqStateTable_Value_2022[] +* escSeqStateTable_Value_2022[offset], which is VALID_NON_TERMINAL_2022 +* c) Switch on this state and continue to next char +* i) Get the value of B from normalize_esq_chars_2022[] with int value of B as index +* ii) x is currently 36 (from above) +* x<<=5 -- x is now 1152 +* x+=normalize_esq_chars_2022[66] +* now x is 1161 +* iii) Search for this value in escSeqStateTable_Key_2022[] +* value of x is stored at escSeqStateTable_Key_2022[21], so offset is 21 +* iv) Get state of this sequence from escSeqStateTable_Value_2022[21] +* escSeqStateTable_Value_2022[offset], which is VALID_TERMINAL_2022 +* v) Get the converter name form escSeqStateTable_Result_2022[21] which is JISX208 +*/ + + +/*Below are the 3 arrays depicting a state transition table*/ +static const int8_t normalize_esq_chars_2022[256] = { +/* 0 1 2 3 4 5 6 7 8 9 */ + + 0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,1 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,4 ,7 ,29 ,0 + ,2 ,24 ,26 ,27 ,0 ,3 ,23 ,6 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,5 ,8 ,9 ,10 ,11 ,12 + ,13 ,14 ,15 ,16 ,17 ,18 ,19 ,20 ,25 ,28 + ,0 ,0 ,21 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,22 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 + ,0 ,0 ,0 ,0 ,0 ,0 +}; + +#ifdef U_ENABLE_GENERIC_ISO_2022 +/* + * When the generic ISO-2022 converter is completely removed, not just disabled + * per #ifdef, then the following state table and the associated tables that are + * dimensioned with MAX_STATES_2022 should be trimmed. + * + * Especially, VALID_MAYBE_TERMINAL_2022 will not be used any more, and all of + * the associated escape sequences starting with ESC ( B should be removed. + * This includes the ones with key values 1097 and all of the ones above 1000000. + * + * For the latter, the tables can simply be truncated. + * For the former, since the tables must be kept parallel, it is probably best + * to simply duplicate an adjacent table cell, parallel in all tables. + * + * It may make sense to restructure the tables, especially by using small search + * tables for the variants instead of indexing them parallel to the table here. + */ +#endif + +#define MAX_STATES_2022 74 +static const int32_t escSeqStateTable_Key_2022[MAX_STATES_2022] = { +/* 0 1 2 3 4 5 6 7 8 9 */ + + 1 ,34 ,36 ,39 ,55 ,57 ,60 ,61 ,1093 ,1096 + ,1097 ,1098 ,1099 ,1100 ,1101 ,1102 ,1103 ,1104 ,1105 ,1106 + ,1109 ,1154 ,1157 ,1160 ,1161 ,1176 ,1178 ,1179 ,1254 ,1257 + ,1768 ,1773 ,1957 ,35105 ,36933 ,36936 ,36937 ,36938 ,36939 ,36940 + ,36942 ,36943 ,36944 ,36945 ,36946 ,36947 ,36948 ,37640 ,37642 ,37644 + ,37646 ,37711 ,37744 ,37745 ,37746 ,37747 ,37748 ,40133 ,40136 ,40138 + ,40139 ,40140 ,40141 ,1123363 ,35947624 ,35947625 ,35947626 ,35947627 ,35947629 ,35947630 + ,35947631 ,35947635 ,35947636 ,35947638 +}; + +#ifdef U_ENABLE_GENERIC_ISO_2022 + +static const char* const escSeqStateTable_Result_2022[MAX_STATES_2022] = { + /* 0 1 2 3 4 5 6 7 8 9 */ + + nullptr ,nullptr ,nullptr ,nullptr ,nullptr ,nullptr ,nullptr ,nullptr ,"latin1" ,"latin1" + ,"latin1" ,"ibm-865" ,"ibm-865" ,"ibm-865" ,"ibm-865" ,"ibm-865" ,"ibm-865" ,"JISX0201" ,"JISX0201" ,"latin1" + ,"latin1" ,nullptr ,"JISX-208" ,"ibm-5478" ,"JISX-208" ,nullptr ,nullptr ,nullptr ,nullptr ,"UTF8" + ,"ISO-8859-1" ,"ISO-8859-7" ,"JIS-X-208" ,nullptr ,"ibm-955" ,"ibm-367" ,"ibm-952" ,"ibm-949" ,"JISX-212" ,"ibm-1383" + ,"ibm-952" ,"ibm-964" ,"ibm-964" ,"ibm-964" ,"ibm-964" ,"ibm-964" ,"ibm-964" ,"ibm-5478" ,"ibm-949" ,"ISO-IR-165" + ,"CNS-11643-1992,1" ,"CNS-11643-1992,2" ,"CNS-11643-1992,3" ,"CNS-11643-1992,4" ,"CNS-11643-1992,5" ,"CNS-11643-1992,6" ,"CNS-11643-1992,7" ,"UTF16_PlatformEndian" ,"UTF16_PlatformEndian" ,"UTF16_PlatformEndian" + ,"UTF16_PlatformEndian" ,"UTF16_PlatformEndian" ,"UTF16_PlatformEndian" ,nullptr ,"latin1" ,"ibm-912" ,"ibm-913" ,"ibm-914" ,"ibm-813" ,"ibm-1089" + ,"ibm-920" ,"ibm-915" ,"ibm-915" ,"latin1" +}; + +#endif + +static const int8_t escSeqStateTable_Value_2022[MAX_STATES_2022] = { +/* 0 1 2 3 4 5 6 7 8 9 */ + VALID_NON_TERMINAL_2022 ,VALID_NON_TERMINAL_2022 ,VALID_NON_TERMINAL_2022 ,VALID_NON_TERMINAL_2022 ,VALID_NON_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_NON_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 + ,VALID_MAYBE_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 + ,VALID_TERMINAL_2022 ,VALID_NON_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_NON_TERMINAL_2022 ,VALID_NON_TERMINAL_2022 ,VALID_NON_TERMINAL_2022 ,VALID_NON_TERMINAL_2022 ,VALID_TERMINAL_2022 + ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_NON_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 + ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 + ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 + ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_NON_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 + ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 ,VALID_TERMINAL_2022 +}; + +/* Type def for refactoring changeState_2022 code*/ +typedef enum{ +#ifdef U_ENABLE_GENERIC_ISO_2022 + ISO_2022=0, +#endif + ISO_2022_JP=1, +#if !UCONFIG_ONLY_HTML_CONVERSION + ISO_2022_KR=2, + ISO_2022_CN=3 +#endif +} Variant2022; + +/*********** ISO 2022 Converter Protos ***********/ +static void U_CALLCONV +_ISO2022Open(UConverter *cnv, UConverterLoadArgs *pArgs, UErrorCode *errorCode); + +static void U_CALLCONV + _ISO2022Close(UConverter *converter); + +static void U_CALLCONV +_ISO2022Reset(UConverter *converter, UConverterResetChoice choice); + +U_CDECL_BEGIN +static const char * U_CALLCONV +_ISO2022getName(const UConverter* cnv); +U_CDECL_END + +static void U_CALLCONV +_ISO_2022_WriteSub(UConverterFromUnicodeArgs *args, int32_t offsetIndex, UErrorCode *err); + +U_CDECL_BEGIN +static UConverter * U_CALLCONV +_ISO_2022_SafeClone(const UConverter *cnv, void *stackBuffer, int32_t *pBufferSize, UErrorCode *status); + +U_CDECL_END + +#ifdef U_ENABLE_GENERIC_ISO_2022 +static void U_CALLCONV +T_UConverter_toUnicode_ISO_2022_OFFSETS_LOGIC(UConverterToUnicodeArgs* args, UErrorCode* err); +#endif + +namespace { + +/*const UConverterSharedData _ISO2022Data;*/ +extern const UConverterSharedData _ISO2022JPData; + +#if !UCONFIG_ONLY_HTML_CONVERSION +extern const UConverterSharedData _ISO2022KRData; +extern const UConverterSharedData _ISO2022CNData; +#endif + +} // namespace + +/*************** Converter implementations ******************/ + +/* The purpose of this function is to get around gcc compiler warnings. */ +static inline void +fromUWriteUInt8(UConverter *cnv, + const char *bytes, int32_t length, + uint8_t **target, const char *targetLimit, + int32_t **offsets, + int32_t sourceIndex, + UErrorCode *pErrorCode) +{ + char *targetChars = (char *)*target; + ucnv_fromUWriteBytes(cnv, bytes, length, &targetChars, targetLimit, + offsets, sourceIndex, pErrorCode); + *target = (uint8_t*)targetChars; + +} + +static inline void +setInitialStateToUnicodeKR(UConverter* /*converter*/, UConverterDataISO2022 *myConverterData){ + if(myConverterData->version == 1) { + UConverter *cnv = myConverterData->currentConverter; + + cnv->toUnicodeStatus=0; /* offset */ + cnv->mode=0; /* state */ + cnv->toULength=0; /* byteIndex */ + } +} + +static inline void +setInitialStateFromUnicodeKR(UConverter* converter,UConverterDataISO2022 *myConverterData){ + /* in ISO-2022-KR the designator sequence appears only once + * in a file so we append it only once + */ + if( converter->charErrorBufferLength==0){ + + converter->charErrorBufferLength = 4; + converter->charErrorBuffer[0] = 0x1b; + converter->charErrorBuffer[1] = 0x24; + converter->charErrorBuffer[2] = 0x29; + converter->charErrorBuffer[3] = 0x43; + } + if(myConverterData->version == 1) { + UConverter *cnv = myConverterData->currentConverter; + + cnv->fromUChar32=0; + cnv->fromUnicodeStatus=1; /* prevLength */ + } +} + +static void U_CALLCONV +_ISO2022Open(UConverter *cnv, UConverterLoadArgs *pArgs, UErrorCode *errorCode){ + + char myLocale[7]={' ',' ',' ',' ',' ',' ', '\0'}; + + cnv->extraInfo = uprv_malloc (sizeof (UConverterDataISO2022)); + if(cnv->extraInfo != nullptr) { + UConverterNamePieces stackPieces; + UConverterLoadArgs stackArgs=UCNV_LOAD_ARGS_INITIALIZER; + UConverterDataISO2022 *myConverterData=(UConverterDataISO2022 *) cnv->extraInfo; + uint32_t version; + + stackArgs.onlyTestIsLoadable = pArgs->onlyTestIsLoadable; + + uprv_memset(myConverterData, 0, sizeof(UConverterDataISO2022)); + myConverterData->currentType = ASCII1; + cnv->fromUnicodeStatus =false; + if(pArgs->locale){ + uprv_strncpy(myLocale, pArgs->locale, sizeof(myLocale)-1); + } + version = pArgs->options & UCNV_OPTIONS_VERSION_MASK; + myConverterData->version = version; + if(myLocale[0]=='j' && (myLocale[1]=='a'|| myLocale[1]=='p') && + (myLocale[2]=='_' || myLocale[2]=='\0')) + { + /* open the required converters and cache them */ + if(version>MAX_JA_VERSION) { + // ICU 55 fails to open a converter for an unsupported version. + // Previously, it fell back to version 0, but that would yield + // unexpected behavior. + *errorCode = U_MISSING_RESOURCE_ERROR; + return; + } + if(jpCharsetMasks[version]&CSM(ISO8859_7)) { + myConverterData->myConverterArray[ISO8859_7] = + ucnv_loadSharedData("ISO8859_7", &stackPieces, &stackArgs, errorCode); + } + myConverterData->myConverterArray[JISX208] = + ucnv_loadSharedData("Shift-JIS", &stackPieces, &stackArgs, errorCode); + if(jpCharsetMasks[version]&CSM(JISX212)) { + myConverterData->myConverterArray[JISX212] = + ucnv_loadSharedData("jisx-212", &stackPieces, &stackArgs, errorCode); + } + if(jpCharsetMasks[version]&CSM(GB2312)) { + myConverterData->myConverterArray[GB2312] = + ucnv_loadSharedData("ibm-5478", &stackPieces, &stackArgs, errorCode); /* gb_2312_80-1 */ + } + if(jpCharsetMasks[version]&CSM(KSC5601)) { + myConverterData->myConverterArray[KSC5601] = + ucnv_loadSharedData("ksc_5601", &stackPieces, &stackArgs, errorCode); + } + + /* set the function pointers to appropriate functions */ + cnv->sharedData=(UConverterSharedData*)(&_ISO2022JPData); + uprv_strcpy(myConverterData->locale,"ja"); + + (void)uprv_strcpy(myConverterData->name,"ISO_2022,locale=ja,version="); + size_t len = uprv_strlen(myConverterData->name); + myConverterData->name[len]=(char)(myConverterData->version+(int)'0'); + myConverterData->name[len+1]='\0'; + } +#if !UCONFIG_ONLY_HTML_CONVERSION + else if(myLocale[0]=='k' && (myLocale[1]=='o'|| myLocale[1]=='r') && + (myLocale[2]=='_' || myLocale[2]=='\0')) + { + if(version>1) { + // ICU 55 fails to open a converter for an unsupported version. + // Previously, it fell back to version 0, but that would yield + // unexpected behavior. + *errorCode = U_MISSING_RESOURCE_ERROR; + return; + } + const char *cnvName; + if(version==1) { + cnvName="icu-internal-25546"; + } else { + cnvName="ibm-949"; + myConverterData->version=version=0; + } + if(pArgs->onlyTestIsLoadable) { + ucnv_canCreateConverter(cnvName, errorCode); /* errorCode carries result */ + uprv_free(cnv->extraInfo); + cnv->extraInfo=nullptr; + return; + } else { + myConverterData->currentConverter=ucnv_open(cnvName, errorCode); + if (U_FAILURE(*errorCode)) { + _ISO2022Close(cnv); + return; + } + + if(version==1) { + (void)uprv_strcpy(myConverterData->name,"ISO_2022,locale=ko,version=1"); + uprv_memcpy(cnv->subChars, myConverterData->currentConverter->subChars, 4); + cnv->subCharLen = myConverterData->currentConverter->subCharLen; + }else{ + (void)uprv_strcpy(myConverterData->name,"ISO_2022,locale=ko,version=0"); + } + + /* initialize the state variables */ + setInitialStateToUnicodeKR(cnv, myConverterData); + setInitialStateFromUnicodeKR(cnv, myConverterData); + + /* set the function pointers to appropriate functions */ + cnv->sharedData=(UConverterSharedData*)&_ISO2022KRData; + uprv_strcpy(myConverterData->locale,"ko"); + } + } + else if(((myLocale[0]=='z' && myLocale[1]=='h') || (myLocale[0]=='c'&& myLocale[1]=='n'))&& + (myLocale[2]=='_' || myLocale[2]=='\0')) + { + if(version>2) { + // ICU 55 fails to open a converter for an unsupported version. + // Previously, it fell back to version 0, but that would yield + // unexpected behavior. + *errorCode = U_MISSING_RESOURCE_ERROR; + return; + } + + /* open the required converters and cache them */ + myConverterData->myConverterArray[GB2312_1] = + ucnv_loadSharedData("ibm-5478", &stackPieces, &stackArgs, errorCode); + if(version==1) { + myConverterData->myConverterArray[ISO_IR_165] = + ucnv_loadSharedData("iso-ir-165", &stackPieces, &stackArgs, errorCode); + } + myConverterData->myConverterArray[CNS_11643] = + ucnv_loadSharedData("cns-11643-1992", &stackPieces, &stackArgs, errorCode); + + + /* set the function pointers to appropriate functions */ + cnv->sharedData=(UConverterSharedData*)&_ISO2022CNData; + uprv_strcpy(myConverterData->locale,"cn"); + + if (version==0){ + myConverterData->version = 0; + (void)uprv_strcpy(myConverterData->name,"ISO_2022,locale=zh,version=0"); + }else if (version==1){ + myConverterData->version = 1; + (void)uprv_strcpy(myConverterData->name,"ISO_2022,locale=zh,version=1"); + }else { + myConverterData->version = 2; + (void)uprv_strcpy(myConverterData->name,"ISO_2022,locale=zh,version=2"); + } + } +#endif // !UCONFIG_ONLY_HTML_CONVERSION + else{ +#ifdef U_ENABLE_GENERIC_ISO_2022 + myConverterData->isFirstBuffer = true; + + /* append the UTF-8 escape sequence */ + cnv->charErrorBufferLength = 3; + cnv->charErrorBuffer[0] = 0x1b; + cnv->charErrorBuffer[1] = 0x25; + cnv->charErrorBuffer[2] = 0x42; + + cnv->sharedData=(UConverterSharedData*)&_ISO2022Data; + /* initialize the state variables */ + uprv_strcpy(myConverterData->name,"ISO_2022"); +#else + *errorCode = U_MISSING_RESOURCE_ERROR; + // Was U_UNSUPPORTED_ERROR but changed in ICU 55 to a more standard + // data loading error code. + return; +#endif + } + + cnv->maxBytesPerUChar=cnv->sharedData->staticData->maxBytesPerChar; + + if(U_FAILURE(*errorCode) || pArgs->onlyTestIsLoadable) { + _ISO2022Close(cnv); + } + } else { + *errorCode = U_MEMORY_ALLOCATION_ERROR; + } +} + + +static void U_CALLCONV +_ISO2022Close(UConverter *converter) { + UConverterDataISO2022* myData =(UConverterDataISO2022 *) (converter->extraInfo); + UConverterSharedData **array = myData->myConverterArray; + int32_t i; + + if (converter->extraInfo != nullptr) { + /*close the array of converter pointers and free the memory*/ + for (i=0; icurrentConverter); + + if(!converter->isExtraLocal){ + uprv_free (converter->extraInfo); + converter->extraInfo = nullptr; + } + } +} + +static void U_CALLCONV +_ISO2022Reset(UConverter *converter, UConverterResetChoice choice) { + UConverterDataISO2022 *myConverterData=(UConverterDataISO2022 *) (converter->extraInfo); + if(choice<=UCNV_RESET_TO_UNICODE) { + uprv_memset(&myConverterData->toU2022State, 0, sizeof(ISO2022State)); + myConverterData->key = 0; + myConverterData->isEmptySegment = false; + } + if(choice!=UCNV_RESET_TO_UNICODE) { + uprv_memset(&myConverterData->fromU2022State, 0, sizeof(ISO2022State)); + } +#ifdef U_ENABLE_GENERIC_ISO_2022 + if(myConverterData->locale[0] == 0){ + if(choice<=UCNV_RESET_TO_UNICODE) { + myConverterData->isFirstBuffer = true; + myConverterData->key = 0; + if (converter->mode == UCNV_SO){ + ucnv_close (myConverterData->currentConverter); + myConverterData->currentConverter=nullptr; + } + converter->mode = UCNV_SI; + } + if(choice!=UCNV_RESET_TO_UNICODE) { + /* re-append UTF-8 escape sequence */ + converter->charErrorBufferLength = 3; + converter->charErrorBuffer[0] = 0x1b; + converter->charErrorBuffer[1] = 0x28; + converter->charErrorBuffer[2] = 0x42; + } + } + else +#endif + { + /* reset the state variables */ + if(myConverterData->locale[0] == 'k'){ + if(choice<=UCNV_RESET_TO_UNICODE) { + setInitialStateToUnicodeKR(converter, myConverterData); + } + if(choice!=UCNV_RESET_TO_UNICODE) { + setInitialStateFromUnicodeKR(converter, myConverterData); + } + } + } +} + +U_CDECL_BEGIN + +static const char * U_CALLCONV +_ISO2022getName(const UConverter* cnv){ + if(cnv->extraInfo){ + UConverterDataISO2022* myData= (UConverterDataISO2022*)cnv->extraInfo; + return myData->name; + } + return nullptr; +} + +U_CDECL_END + + +/*************** to unicode *******************/ +/**************************************************************************** + * Recognized escape sequences are + * (B ASCII + * .A ISO-8859-1 + * .F ISO-8859-7 + * (J JISX-201 + * (I JISX-201 + * $B JISX-208 + * $@ JISX-208 + * $(D JISX-212 + * $A GB2312 + * $(C KSC5601 + */ +static const int8_t nextStateToUnicodeJP[MAX_STATES_2022]= { +/* 0 1 2 3 4 5 6 7 8 9 */ + INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,SS2_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE + ,ASCII ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,JISX201 ,HWKANA_7BIT ,JISX201 ,INVALID_STATE + ,INVALID_STATE ,INVALID_STATE ,JISX208 ,GB2312 ,JISX208 ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE + ,ISO8859_1 ,ISO8859_7 ,JISX208 ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,KSC5601 ,JISX212 ,INVALID_STATE + ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE + ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE + ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE + ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE +}; + +#if !UCONFIG_ONLY_HTML_CONVERSION +/*************** to unicode *******************/ +static const int8_t nextStateToUnicodeCN[MAX_STATES_2022]= { +/* 0 1 2 3 4 5 6 7 8 9 */ + INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,SS2_STATE ,SS3_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE + ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE + ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE + ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE + ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,GB2312_1 ,INVALID_STATE ,ISO_IR_165 + ,CNS_11643_1 ,CNS_11643_2 ,CNS_11643_3 ,CNS_11643_4 ,CNS_11643_5 ,CNS_11643_6 ,CNS_11643_7 ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE + ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE + ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE ,INVALID_STATE +}; +#endif + + +static UCNV_TableStates_2022 +getKey_2022(char c,int32_t* key,int32_t* offset){ + int32_t togo; + int32_t low = 0; + int32_t hi = MAX_STATES_2022; + int32_t oldmid=0; + + togo = normalize_esq_chars_2022[(uint8_t)c]; + if(togo == 0) { + /* not a valid character anywhere in an escape sequence */ + *key = 0; + *offset = 0; + return INVALID_2022; + } + togo = (*key << 5) + togo; + + while (hi != low) /*binary search*/{ + + int32_t mid = (hi+low) >> 1; /*Finds median*/ + + if (mid == oldmid) + break; + + if (escSeqStateTable_Key_2022[mid] > togo){ + hi = mid; + } + else if (escSeqStateTable_Key_2022[mid] < togo){ + low = mid; + } + else /*we found it*/{ + *key = togo; + *offset = mid; + return (UCNV_TableStates_2022)escSeqStateTable_Value_2022[mid]; + } + oldmid = mid; + + } + + *key = 0; + *offset = 0; + return INVALID_2022; +} + +/*runs through a state machine to determine the escape sequence - codepage correspondence + */ +static void +changeState_2022(UConverter* _this, + const char** source, + const char* sourceLimit, + Variant2022 var, + UErrorCode* err){ + UCNV_TableStates_2022 value; + UConverterDataISO2022* myData2022 = ((UConverterDataISO2022*)_this->extraInfo); + uint32_t key = myData2022->key; + int32_t offset = 0; + int8_t initialToULength = _this->toULength; + char c; + + value = VALID_NON_TERMINAL_2022; + while (*source < sourceLimit) { + c = *(*source)++; + _this->toUBytes[_this->toULength++]=(uint8_t)c; + value = getKey_2022(c,(int32_t *) &key, &offset); + + switch (value){ + + case VALID_NON_TERMINAL_2022 : + /* continue with the loop */ + break; + + case VALID_TERMINAL_2022: + key = 0; + goto DONE; + + case INVALID_2022: + goto DONE; + + case VALID_MAYBE_TERMINAL_2022: +#ifdef U_ENABLE_GENERIC_ISO_2022 + /* ESC ( B is ambiguous only for ISO_2022 itself */ + if(var == ISO_2022) { + /* discard toUBytes[] for ESC ( B because this sequence is correct and complete */ + _this->toULength = 0; + + /* TODO need to indicate that ESC ( B was seen; if failure, then need to replay from source or from MBCS-style replay */ + + /* continue with the loop */ + value = VALID_NON_TERMINAL_2022; + break; + } else +#endif + { + /* not ISO_2022 itself, finish here */ + value = VALID_TERMINAL_2022; + key = 0; + goto DONE; + } + } + } + +DONE: + myData2022->key = key; + + if (value == VALID_NON_TERMINAL_2022) { + /* indicate that the escape sequence is incomplete: key!=0 */ + return; + } else if (value == INVALID_2022 ) { + *err = U_ILLEGAL_ESCAPE_SEQUENCE; + } else /* value == VALID_TERMINAL_2022 */ { + switch(var){ +#ifdef U_ENABLE_GENERIC_ISO_2022 + case ISO_2022: + { + const char *chosenConverterName = escSeqStateTable_Result_2022[offset]; + if(chosenConverterName == nullptr) { + /* SS2 or SS3 */ + *err = U_UNSUPPORTED_ESCAPE_SEQUENCE; + _this->toUCallbackReason = UCNV_UNASSIGNED; + return; + } + + _this->mode = UCNV_SI; + ucnv_close(myData2022->currentConverter); + myData2022->currentConverter = myUConverter = ucnv_open(chosenConverterName, err); + if(U_SUCCESS(*err)) { + myUConverter->fromCharErrorBehaviour = UCNV_TO_U_CALLBACK_STOP; + _this->mode = UCNV_SO; + } + break; + } +#endif + case ISO_2022_JP: + { + StateEnum tempState=(StateEnum)nextStateToUnicodeJP[offset]; + switch(tempState) { + case INVALID_STATE: + *err = U_UNSUPPORTED_ESCAPE_SEQUENCE; + break; + case SS2_STATE: + if(myData2022->toU2022State.cs[2]!=0) { + if(myData2022->toU2022State.g<2) { + myData2022->toU2022State.prevG=myData2022->toU2022State.g; + } + myData2022->toU2022State.g=2; + } else { + /* illegal to have SS2 before a matching designator */ + *err = U_ILLEGAL_ESCAPE_SEQUENCE; + } + break; + /* case SS3_STATE: not used in ISO-2022-JP-x */ + case ISO8859_1: + case ISO8859_7: + if((jpCharsetMasks[myData2022->version] & CSM(tempState)) == 0) { + *err = U_UNSUPPORTED_ESCAPE_SEQUENCE; + } else { + /* G2 charset for SS2 */ + myData2022->toU2022State.cs[2]=(int8_t)tempState; + } + break; + default: + if((jpCharsetMasks[myData2022->version] & CSM(tempState)) == 0) { + *err = U_UNSUPPORTED_ESCAPE_SEQUENCE; + } else { + /* G0 charset */ + myData2022->toU2022State.cs[0]=(int8_t)tempState; + } + break; + } + } + break; +#if !UCONFIG_ONLY_HTML_CONVERSION + case ISO_2022_CN: + { + StateEnum tempState=(StateEnum)nextStateToUnicodeCN[offset]; + switch(tempState) { + case INVALID_STATE: + *err = U_UNSUPPORTED_ESCAPE_SEQUENCE; + break; + case SS2_STATE: + if(myData2022->toU2022State.cs[2]!=0) { + if(myData2022->toU2022State.g<2) { + myData2022->toU2022State.prevG=myData2022->toU2022State.g; + } + myData2022->toU2022State.g=2; + } else { + /* illegal to have SS2 before a matching designator */ + *err = U_ILLEGAL_ESCAPE_SEQUENCE; + } + break; + case SS3_STATE: + if(myData2022->toU2022State.cs[3]!=0) { + if(myData2022->toU2022State.g<2) { + myData2022->toU2022State.prevG=myData2022->toU2022State.g; + } + myData2022->toU2022State.g=3; + } else { + /* illegal to have SS3 before a matching designator */ + *err = U_ILLEGAL_ESCAPE_SEQUENCE; + } + break; + case ISO_IR_165: + if(myData2022->version==0) { + *err = U_UNSUPPORTED_ESCAPE_SEQUENCE; + break; + } + U_FALLTHROUGH; + case GB2312_1: + U_FALLTHROUGH; + case CNS_11643_1: + myData2022->toU2022State.cs[1]=(int8_t)tempState; + break; + case CNS_11643_2: + myData2022->toU2022State.cs[2]=(int8_t)tempState; + break; + default: + /* other CNS 11643 planes */ + if(myData2022->version==0) { + *err = U_UNSUPPORTED_ESCAPE_SEQUENCE; + } else { + myData2022->toU2022State.cs[3]=(int8_t)tempState; + } + break; + } + } + break; + case ISO_2022_KR: + if(offset==0x30){ + /* nothing to be done, just accept this one escape sequence */ + } else { + *err = U_UNSUPPORTED_ESCAPE_SEQUENCE; + } + break; +#endif // !UCONFIG_ONLY_HTML_CONVERSION + + default: + *err = U_ILLEGAL_ESCAPE_SEQUENCE; + break; + } + } + if(U_SUCCESS(*err)) { + _this->toULength = 0; + } else if(*err==U_ILLEGAL_ESCAPE_SEQUENCE) { + if(_this->toULength>1) { + /* + * Ticket 5691: consistent illegal sequences: + * - We include at least the first byte (ESC) in the illegal sequence. + * - If any of the non-initial bytes could be the start of a character, + * we stop the illegal sequence before the first one of those. + * In escape sequences, all following bytes are "printable", that is, + * unless they are completely illegal (>7f in SBCS, outside 21..7e in DBCS), + * they are valid single/lead bytes. + * For simplicity, we always only report the initial ESC byte as the + * illegal sequence and back out all other bytes we looked at. + */ + /* Back out some bytes. */ + int8_t backOutDistance=_this->toULength-1; + int8_t bytesFromThisBuffer=_this->toULength-initialToULength; + if(backOutDistance<=bytesFromThisBuffer) { + /* same as initialToULength<=1 */ + *source-=backOutDistance; + } else { + /* Back out bytes from the previous buffer: Need to replay them. */ + _this->preToULength=(int8_t)(bytesFromThisBuffer-backOutDistance); + /* same as -(initialToULength-1) */ + /* preToULength is negative! */ + uprv_memcpy(_this->preToU, _this->toUBytes+1, -_this->preToULength); + *source-=bytesFromThisBuffer; + } + _this->toULength=1; + } + } else if(*err==U_UNSUPPORTED_ESCAPE_SEQUENCE) { + _this->toUCallbackReason = UCNV_UNASSIGNED; + } +} + +#if !UCONFIG_ONLY_HTML_CONVERSION +/*Checks the characters of the buffer against valid 2022 escape sequences +*if the match we return a pointer to the initial start of the sequence otherwise +*we return sourceLimit +*/ +/*for 2022 looks ahead in the stream + *to determine the longest possible convertible + *data stream + */ +static inline const char* +getEndOfBuffer_2022(const char** source, + const char* sourceLimit, + UBool /*flush*/){ + + const char* mySource = *source; + +#ifdef U_ENABLE_GENERIC_ISO_2022 + if (*source >= sourceLimit) + return sourceLimit; + + do{ + + if (*mySource == ESC_2022){ + int8_t i; + int32_t key = 0; + int32_t offset; + UCNV_TableStates_2022 value = VALID_NON_TERMINAL_2022; + + /* Kludge: I could not + * figure out the reason for validating an escape sequence + * twice - once here and once in changeState_2022(). + * is it possible to have an ESC character in a ISO2022 + * byte stream which is valid in a code page? Is it legal? + */ + for (i=0; + (mySource+i < sourceLimit)&&(value == VALID_NON_TERMINAL_2022); + i++) { + value = getKey_2022(*(mySource+i), &key, &offset); + } + if (value > 0 || *mySource==ESC_2022) + return mySource; + + if ((value == VALID_NON_TERMINAL_2022)&&(!flush) ) + return sourceLimit; + } + }while (++mySource < sourceLimit); + + return sourceLimit; +#else + while(mySource < sourceLimit && *mySource != ESC_2022) { + ++mySource; + } + return mySource; +#endif +} +#endif + +/* This inline function replicates code in _MBCSFromUChar32() function in ucnvmbcs.c + * any future change in _MBCSFromUChar32() function should be reflected here. + * @return number of bytes in *value; negative number if fallback; 0 if no mapping + */ +static inline int32_t +MBCS_FROM_UCHAR32_ISO2022(UConverterSharedData* sharedData, + UChar32 c, + uint32_t* value, + UBool useFallback, + int outputType) +{ + const int32_t *cx; + const uint16_t *table; + uint32_t stage2Entry; + uint32_t myValue; + int32_t length; + const uint8_t *p; + /* + * TODO(markus): Use and require new, faster MBCS conversion table structures. + * Use internal version of ucnv_open() that verifies that the new structures are available, + * else U_INTERNAL_PROGRAM_ERROR. + */ + /* BMP-only codepages are stored without stage 1 entries for supplementary code points */ + if(c<0x10000 || (sharedData->mbcs.unicodeMask&UCNV_HAS_SUPPLEMENTARY)) { + table=sharedData->mbcs.fromUnicodeTable; + stage2Entry=MBCS_STAGE_2_FROM_U(table, c); + /* get the bytes and the length for the output */ + if(outputType==MBCS_OUTPUT_2){ + myValue=MBCS_VALUE_2_FROM_STAGE_2(sharedData->mbcs.fromUnicodeBytes, stage2Entry, c); + if(myValue<=0xff) { + length=1; + } else { + length=2; + } + } else /* outputType==MBCS_OUTPUT_3 */ { + p=MBCS_POINTER_3_FROM_STAGE_2(sharedData->mbcs.fromUnicodeBytes, stage2Entry, c); + myValue=((uint32_t)*p<<16)|((uint32_t)p[1]<<8)|p[2]; + if(myValue<=0xff) { + length=1; + } else if(myValue<=0xffff) { + length=2; + } else { + length=3; + } + } + /* is this code point assigned, or do we use fallbacks? */ + if((stage2Entry&(1<<(16+(c&0xf))))!=0) { + /* assigned */ + *value=myValue; + return length; + } else if(FROM_U_USE_FALLBACK(useFallback, c) && myValue!=0) { + /* + * We allow a 0 byte output if the "assigned" bit is set for this entry. + * There is no way with this data structure for fallback output + * to be a zero byte. + */ + *value=myValue; + return -length; + } + } + + cx=sharedData->mbcs.extIndexes; + if(cx!=nullptr) { + return ucnv_extSimpleMatchFromU(cx, c, value, useFallback); + } + + /* unassigned */ + return 0; +} + +/* This inline function replicates code in _MBCSSingleFromUChar32() function in ucnvmbcs.c + * any future change in _MBCSSingleFromUChar32() function should be reflected here. + * @param retval pointer to output byte + * @return 1 roundtrip byte 0 no mapping -1 fallback byte + */ +static inline int32_t +MBCS_SINGLE_FROM_UCHAR32(UConverterSharedData* sharedData, + UChar32 c, + uint32_t* retval, + UBool useFallback) +{ + const uint16_t *table; + int32_t value; + /* BMP-only codepages are stored without stage 1 entries for supplementary code points */ + if(c>=0x10000 && !(sharedData->mbcs.unicodeMask&UCNV_HAS_SUPPLEMENTARY)) { + return 0; + } + /* convert the Unicode code point in c into codepage bytes (same as in _MBCSFromUnicodeWithOffsets) */ + table=sharedData->mbcs.fromUnicodeTable; + /* get the byte for the output */ + value=MBCS_SINGLE_RESULT_FROM_U(table, (uint16_t *)sharedData->mbcs.fromUnicodeBytes, c); + /* is this code point assigned, or do we use fallbacks? */ + *retval=(uint32_t)(value&0xff); + if(value>=0xf00) { + return 1; /* roundtrip */ + } else if(useFallback ? value>=0x800 : value>=0xc00) { + return -1; /* fallback taken */ + } else { + return 0; /* no mapping */ + } +} + +/* + * Check that the result is a 2-byte value with each byte in the range A1..FE + * (strict EUC DBCS) before accepting it and subtracting 0x80 from each byte + * to move it to the ISO 2022 range 21..7E. + * Return 0 if out of range. + */ +static inline uint32_t +_2022FromGR94DBCS(uint32_t value) { + if( (uint16_t)(value - 0xa1a1) <= (0xfefe - 0xa1a1) && + (uint8_t)(value - 0xa1) <= (0xfe - 0xa1) + ) { + return value - 0x8080; /* shift down to 21..7e byte range */ + } else { + return 0; /* not valid for ISO 2022 */ + } +} + +#if 0 /* 5691: Call sites now check for validity. They can just += 0x8080 after that. */ +/* + * This method does the reverse of _2022FromGR94DBCS(). Given the 2022 code point, it returns the + * 2 byte value that is in the range A1..FE for each byte. Otherwise it returns the 2022 code point + * unchanged. + */ +static inline uint32_t +_2022ToGR94DBCS(uint32_t value) { + uint32_t returnValue = value + 0x8080; + if( (uint16_t)(returnValue - 0xa1a1) <= (0xfefe - 0xa1a1) && + (uint8_t)(returnValue - 0xa1) <= (0xfe - 0xa1)) { + return returnValue; + } else { + return value; + } +} +#endif + +#ifdef U_ENABLE_GENERIC_ISO_2022 + +/********************************************************************************** +* ISO-2022 Converter +* +* +*/ + +static void U_CALLCONV +T_UConverter_toUnicode_ISO_2022_OFFSETS_LOGIC(UConverterToUnicodeArgs* args, + UErrorCode* err){ + const char* mySourceLimit, *realSourceLimit; + const char* sourceStart; + const char16_t* myTargetStart; + UConverter* saveThis; + UConverterDataISO2022* myData; + int8_t length; + + saveThis = args->converter; + myData=((UConverterDataISO2022*)(saveThis->extraInfo)); + + realSourceLimit = args->sourceLimit; + while (args->source < realSourceLimit) { + if(myData->key == 0) { /* are we in the middle of an escape sequence? */ + /*Find the end of the buffer e.g : Next Escape Seq | end of Buffer*/ + mySourceLimit = getEndOfBuffer_2022(&(args->source), realSourceLimit, args->flush); + + if(args->source < mySourceLimit) { + if(myData->currentConverter==nullptr) { + myData->currentConverter = ucnv_open("ASCII",err); + if(U_FAILURE(*err)){ + return; + } + + myData->currentConverter->fromCharErrorBehaviour = UCNV_TO_U_CALLBACK_STOP; + saveThis->mode = UCNV_SO; + } + + /* convert to before the ESC or until the end of the buffer */ + myData->isFirstBuffer=false; + sourceStart = args->source; + myTargetStart = args->target; + args->converter = myData->currentConverter; + ucnv_toUnicode(args->converter, + &args->target, + args->targetLimit, + &args->source, + mySourceLimit, + args->offsets, + (UBool)(args->flush && mySourceLimit == realSourceLimit), + err); + args->converter = saveThis; + + if (*err == U_BUFFER_OVERFLOW_ERROR) { + /* move the overflow buffer */ + length = saveThis->UCharErrorBufferLength = myData->currentConverter->UCharErrorBufferLength; + myData->currentConverter->UCharErrorBufferLength = 0; + if(length > 0) { + uprv_memcpy(saveThis->UCharErrorBuffer, + myData->currentConverter->UCharErrorBuffer, + length*U_SIZEOF_UCHAR); + } + return; + } + + /* + * At least one of: + * -Error while converting + * -Done with entire buffer + * -Need to write offsets or update the current offset + * (leave that up to the code in ucnv.c) + * + * or else we just stopped at an ESC byte and continue with changeState_2022() + */ + if (U_FAILURE(*err) || + (args->source == realSourceLimit) || + (args->offsets != nullptr && (args->target != myTargetStart || args->source != sourceStart) || + (mySourceLimit < realSourceLimit && myData->currentConverter->toULength > 0)) + ) { + /* copy partial or error input for truncated detection and error handling */ + if(U_FAILURE(*err)) { + length = saveThis->invalidCharLength = myData->currentConverter->invalidCharLength; + if(length > 0) { + uprv_memcpy(saveThis->invalidCharBuffer, myData->currentConverter->invalidCharBuffer, length); + } + } else { + length = saveThis->toULength = myData->currentConverter->toULength; + if(length > 0) { + uprv_memcpy(saveThis->toUBytes, myData->currentConverter->toUBytes, length); + if(args->source < mySourceLimit) { + *err = U_TRUNCATED_CHAR_FOUND; /* truncated input before ESC */ + } + } + } + return; + } + } + } + + sourceStart = args->source; + changeState_2022(args->converter, + &(args->source), + realSourceLimit, + ISO_2022, + err); + if (U_FAILURE(*err) || (args->source != sourceStart && args->offsets != nullptr)) { + /* let the ucnv.c code update its current offset */ + return; + } + } +} + +#endif + +/* + * To Unicode Callback helper function + */ +static void +toUnicodeCallback(UConverter *cnv, + const uint32_t sourceChar, const uint32_t targetUniChar, + UErrorCode* err){ + if(sourceChar>0xff){ + cnv->toUBytes[0] = (uint8_t)(sourceChar>>8); + cnv->toUBytes[1] = (uint8_t)sourceChar; + cnv->toULength = 2; + } + else{ + cnv->toUBytes[0] =(char) sourceChar; + cnv->toULength = 1; + } + + if(targetUniChar == (missingCharMarker-1/*0xfffe*/)){ + *err = U_INVALID_CHAR_FOUND; + } + else{ + *err = U_ILLEGAL_CHAR_FOUND; + } +} + +/**************************************ISO-2022-JP*************************************************/ + +/************************************** IMPORTANT ************************************************** +* The UConverter_fromUnicode_ISO2022_JP converter does not use ucnv_fromUnicode() functions for SBCS,DBCS and +* MBCS; instead, the values are obtained directly by calling _MBCSFromUChar32(). +* The converter iterates over each Unicode codepoint +* to obtain the equivalent codepoints from the codepages supported. Since the source buffer is +* processed one char at a time it would make sense to reduce the extra processing a canned converter +* would do as far as possible. +* +* If the implementation of these macros or structure of sharedData struct change in the future, make +* sure that ISO-2022 is also changed. +*************************************************************************************************** +*/ + +/*************************************************************************************************** +* Rules for ISO-2022-jp encoding +* (i) Escape sequences must be fully contained within a line they should not +* span new lines or CRs +* (ii) If the last character on a line is represented by two bytes then an ASCII or +* JIS-Roman character escape sequence should follow before the line terminates +* (iii) If the first character on the line is represented by two bytes then a two +* byte character escape sequence should precede it +* (iv) If no escape sequence is encountered then the characters are ASCII +* (v) Latin(ISO-8859-1) and Greek(ISO-8859-7) characters must be designated to G2, +* and invoked with SS2 (ESC N). +* (vi) If there is any G0 designation in text, there must be a switch to +* ASCII or to JIS X 0201-Roman before a space character (but not +* necessarily before "ESC 4/14 2/0" or "ESC N ' '") or control +* characters such as tab or CRLF. +* (vi) Supported encodings: +* ASCII, JISX201, JISX208, JISX212, GB2312, KSC5601, ISO-8859-1,ISO-8859-7 +* +* source : RFC-1554 +* +* JISX201, JISX208,JISX212 : new .cnv data files created +* KSC5601 : alias to ibm-949 mapping table +* GB2312 : alias to ibm-1386 mapping table +* ISO-8859-1 : Algorithmic implemented as LATIN1 case +* ISO-8859-7 : alias to ibm-9409 mapping table +*/ + +/* preference order of JP charsets */ +static const StateEnum jpCharsetPref[]={ + ASCII, + JISX201, + ISO8859_1, + JISX208, + ISO8859_7, + JISX212, + GB2312, + KSC5601, + HWKANA_7BIT +}; + +/* + * The escape sequences must be in order of the enum constants like JISX201 = 3, + * not in order of jpCharsetPref[]! + */ +static const char escSeqChars[][6] ={ + "\x1B\x28\x42", /* (B ASCII */ + "\x1B\x2E\x41", /* .A ISO-8859-1 */ + "\x1B\x2E\x46", /* .F ISO-8859-7 */ + "\x1B\x28\x4A", /* (J JISX-201 */ + "\x1B\x24\x42", /* $B JISX-208 */ + "\x1B\x24\x28\x44", /* $(D JISX-212 */ + "\x1B\x24\x41", /* $A GB2312 */ + "\x1B\x24\x28\x43", /* $(C KSC5601 */ + "\x1B\x28\x49" /* (I HWKANA_7BIT */ + +}; +static const int8_t escSeqCharsLen[] ={ + 3, /* length of (B ASCII */ + 3, /* length of .A ISO-8859-1 */ + 3, /* length of .F ISO-8859-7 */ + 3, /* length of (J JISX-201 */ + 3, /* length of $B JISX-208 */ + 4, /* length of $(D JISX-212 */ + 3, /* length of $A GB2312 */ + 4, /* length of $(C KSC5601 */ + 3 /* length of (I HWKANA_7BIT */ +}; + +/* +* The iteration over various code pages works this way: +* i) Get the currentState from myConverterData->currentState +* ii) Check if the character is mapped to a valid character in the currentState +* Yes -> a) set the initIterState to currentState +* b) remain in this state until an invalid character is found +* No -> a) go to the next code page and find the character +* iii) Before changing the state increment the current state check if the current state +* is equal to the intitIteration state +* Yes -> A character that cannot be represented in any of the supported encodings +* break and return a U_INVALID_CHARACTER error +* No -> Continue and find the character in next code page +* +* +* TODO: Implement a priority technique where the users are allowed to set the priority of code pages +*/ + +/* Map 00..7F to Unicode according to JIS X 0201. */ +static inline uint32_t +jisx201ToU(uint32_t value) { + if(value < 0x5c) { + return value; + } else if(value == 0x5c) { + return 0xa5; + } else if(value == 0x7e) { + return 0x203e; + } else /* value <= 0x7f */ { + return value; + } +} + +/* Map Unicode to 00..7F according to JIS X 0201. Return U+FFFE if unmappable. */ +static inline uint32_t +jisx201FromU(uint32_t value) { + if(value<=0x7f) { + if(value!=0x5c && value!=0x7e) { + return value; + } + } else if(value==0xa5) { + return 0x5c; + } else if(value==0x203e) { + return 0x7e; + } + return 0xfffe; +} + +/* + * Take a valid Shift-JIS byte pair, check that it is in the range corresponding + * to JIS X 0208, and convert it to a pair of 21..7E bytes. + * Return 0 if the byte pair is out of range. + */ +static inline uint32_t +_2022FromSJIS(uint32_t value) { + uint8_t trail; + + if(value > 0xEFFC) { + return 0; /* beyond JIS X 0208 */ + } + + trail = (uint8_t)value; + + value &= 0xff00; /* lead byte */ + if(value <= 0x9f00) { + value -= 0x7000; + } else /* 0xe000 <= value <= 0xef00 */ { + value -= 0xb000; + } + value <<= 1; + + if(trail <= 0x9e) { + value -= 0x100; + if(trail <= 0x7e) { + value |= trail - 0x1f; + } else { + value |= trail - 0x20; + } + } else /* trail <= 0xfc */ { + value |= trail - 0x7e; + } + return value; +} + +/* + * Convert a pair of JIS X 0208 21..7E bytes to Shift-JIS. + * If either byte is outside 21..7E make sure that the result is not valid + * for Shift-JIS so that the converter catches it. + * Some invalid byte values already turn into equally invalid Shift-JIS + * byte values and need not be tested explicitly. + */ +static inline void +_2022ToSJIS(uint8_t c1, uint8_t c2, char bytes[2]) { + if(c1&1) { + ++c1; + if(c2 <= 0x5f) { + c2 += 0x1f; + } else if(c2 <= 0x7e) { + c2 += 0x20; + } else { + c2 = 0; /* invalid */ + } + } else { + if((uint8_t)(c2-0x21) <= ((0x7e)-0x21)) { + c2 += 0x7e; + } else { + c2 = 0; /* invalid */ + } + } + c1 >>= 1; + if(c1 <= 0x2f) { + c1 += 0x70; + } else if(c1 <= 0x3f) { + c1 += 0xb0; + } else { + c1 = 0; /* invalid */ + } + bytes[0] = (char)c1; + bytes[1] = (char)c2; +} + +/* + * JIS X 0208 has fallbacks from Unicode half-width Katakana to full-width (DBCS) + * Katakana. + * Now that we use a Shift-JIS table for JIS X 0208 we need to hardcode these fallbacks + * because Shift-JIS roundtrips half-width Katakana to single bytes. + * These were the only fallbacks in ICU's jisx-208.ucm file. + */ +static const uint16_t hwkana_fb[HWKANA_END - HWKANA_START + 1] = { + 0x2123, /* U+FF61 */ + 0x2156, + 0x2157, + 0x2122, + 0x2126, + 0x2572, + 0x2521, + 0x2523, + 0x2525, + 0x2527, + 0x2529, + 0x2563, + 0x2565, + 0x2567, + 0x2543, + 0x213C, /* U+FF70 */ + 0x2522, + 0x2524, + 0x2526, + 0x2528, + 0x252A, + 0x252B, + 0x252D, + 0x252F, + 0x2531, + 0x2533, + 0x2535, + 0x2537, + 0x2539, + 0x253B, + 0x253D, + 0x253F, /* U+FF80 */ + 0x2541, + 0x2544, + 0x2546, + 0x2548, + 0x254A, + 0x254B, + 0x254C, + 0x254D, + 0x254E, + 0x254F, + 0x2552, + 0x2555, + 0x2558, + 0x255B, + 0x255E, + 0x255F, /* U+FF90 */ + 0x2560, + 0x2561, + 0x2562, + 0x2564, + 0x2566, + 0x2568, + 0x2569, + 0x256A, + 0x256B, + 0x256C, + 0x256D, + 0x256F, + 0x2573, + 0x212B, + 0x212C /* U+FF9F */ +}; + +static void U_CALLCONV +UConverter_fromUnicode_ISO_2022_JP_OFFSETS_LOGIC(UConverterFromUnicodeArgs* args, UErrorCode* err) { + UConverter *cnv = args->converter; + UConverterDataISO2022 *converterData; + ISO2022State *pFromU2022State; + uint8_t *target = (uint8_t *) args->target; + const uint8_t *targetLimit = (const uint8_t *) args->targetLimit; + const char16_t* source = args->source; + const char16_t* sourceLimit = args->sourceLimit; + int32_t* offsets = args->offsets; + UChar32 sourceChar; + char buffer[8]; + int32_t len, outLen; + int8_t choices[10]; + int32_t choiceCount; + uint32_t targetValue = 0; + UBool useFallback; + + int32_t i; + int8_t cs, g; + + /* set up the state */ + converterData = (UConverterDataISO2022*)cnv->extraInfo; + pFromU2022State = &converterData->fromU2022State; + + choiceCount = 0; + + /* check if the last codepoint of previous buffer was a lead surrogate*/ + if((sourceChar = cnv->fromUChar32)!=0 && target< targetLimit) { + goto getTrail; + } + + while(source < sourceLimit) { + if(target < targetLimit) { + + sourceChar = *(source++); + /*check if the char is a First surrogate*/ + if(U16_IS_SURROGATE(sourceChar)) { + if(U16_IS_SURROGATE_LEAD(sourceChar)) { +getTrail: + /*look ahead to find the trail surrogate*/ + if(source < sourceLimit) { + /* test the following code unit */ + char16_t trail=(char16_t) *source; + if(U16_IS_TRAIL(trail)) { + source++; + sourceChar=U16_GET_SUPPLEMENTARY(sourceChar, trail); + cnv->fromUChar32=0x00; + /* convert this supplementary code point */ + /* exit this condition tree */ + } else { + /* this is an unmatched lead code unit (1st surrogate) */ + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + cnv->fromUChar32=sourceChar; + break; + } + } else { + /* no more input */ + cnv->fromUChar32=sourceChar; + break; + } + } else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + cnv->fromUChar32=sourceChar; + break; + } + } + + /* do not convert SO/SI/ESC */ + if(IS_2022_CONTROL(sourceChar)) { + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + cnv->fromUChar32=sourceChar; + break; + } + + /* do the conversion */ + + if(choiceCount == 0) { + uint16_t csm; + + /* + * The csm variable keeps track of which charsets are allowed + * and not used yet while building the choices[]. + */ + csm = jpCharsetMasks[converterData->version]; + choiceCount = 0; + + /* JIS7/8: try single-byte half-width Katakana before JISX208 */ + if(converterData->version == 3 || converterData->version == 4) { + choices[choiceCount++] = (int8_t)HWKANA_7BIT; + } + /* Do not try single-byte half-width Katakana for other versions. */ + csm &= ~CSM(HWKANA_7BIT); + + /* try the current G0 charset */ + choices[choiceCount++] = cs = pFromU2022State->cs[0]; + csm &= ~CSM(cs); + + /* try the current G2 charset */ + if((cs = pFromU2022State->cs[2]) != 0) { + choices[choiceCount++] = cs; + csm &= ~CSM(cs); + } + + /* try all the other possible charsets */ + for(i = 0; i < UPRV_LENGTHOF(jpCharsetPref); ++i) { + cs = (int8_t)jpCharsetPref[i]; + if(CSM(cs) & csm) { + choices[choiceCount++] = cs; + csm &= ~CSM(cs); + } + } + } + + cs = g = 0; + /* + * len==0: no mapping found yet + * len<0: found a fallback result: continue looking for a roundtrip but no further fallbacks + * len>0: found a roundtrip result, done + */ + len = 0; + /* + * We will turn off useFallback after finding a fallback, + * but we still get fallbacks from PUA code points as usual. + * Therefore, we will also need to check that we don't overwrite + * an early fallback with a later one. + */ + useFallback = cnv->useFallback; + + for(i = 0; i < choiceCount && len <= 0; ++i) { + uint32_t value; + int32_t len2; + int8_t cs0 = choices[i]; + switch(cs0) { + case ASCII: + if(sourceChar <= 0x7f) { + targetValue = (uint32_t)sourceChar; + len = 1; + cs = cs0; + g = 0; + } + break; + case ISO8859_1: + if(GR96_START <= sourceChar && sourceChar <= GR96_END) { + targetValue = (uint32_t)sourceChar - 0x80; + len = 1; + cs = cs0; + g = 2; + } + break; + case HWKANA_7BIT: + if((uint32_t)(sourceChar - HWKANA_START) <= (HWKANA_END - HWKANA_START)) { + if(converterData->version==3) { + /* JIS7: use G1 (SO) */ + /* Shift U+FF61..U+FF9F to bytes 21..5F. */ + targetValue = (uint32_t)(sourceChar - (HWKANA_START - 0x21)); + len = 1; + pFromU2022State->cs[1] = cs = cs0; /* do not output an escape sequence */ + g = 1; + } else if(converterData->version==4) { + /* JIS8: use 8-bit bytes with any single-byte charset, see escape sequence output below */ + /* Shift U+FF61..U+FF9F to bytes A1..DF. */ + targetValue = (uint32_t)(sourceChar - (HWKANA_START - 0xa1)); + len = 1; + + cs = pFromU2022State->cs[0]; + if(IS_JP_DBCS(cs)) { + /* switch from a DBCS charset to JISX201 */ + cs = (int8_t)JISX201; + } + /* else stay in the current G0 charset */ + g = 0; + } + /* else do not use HWKANA_7BIT with other versions */ + } + break; + case JISX201: + /* G0 SBCS */ + value = jisx201FromU(sourceChar); + if(value <= 0x7f) { + targetValue = value; + len = 1; + cs = cs0; + g = 0; + useFallback = false; + } + break; + case JISX208: + /* G0 DBCS from Shift-JIS table */ + len2 = MBCS_FROM_UCHAR32_ISO2022( + converterData->myConverterArray[cs0], + sourceChar, &value, + useFallback, MBCS_OUTPUT_2); + if(len2 == 2 || (len2 == -2 && len == 0)) { /* only accept DBCS: abs(len)==2 */ + value = _2022FromSJIS(value); + if(value != 0) { + targetValue = value; + len = len2; + cs = cs0; + g = 0; + useFallback = false; + } + } else if(len == 0 && useFallback && + (uint32_t)(sourceChar - HWKANA_START) <= (HWKANA_END - HWKANA_START)) { + targetValue = hwkana_fb[sourceChar - HWKANA_START]; + len = -2; + cs = cs0; + g = 0; + useFallback = false; + } + break; + case ISO8859_7: + /* G0 SBCS forced to 7-bit output */ + len2 = MBCS_SINGLE_FROM_UCHAR32( + converterData->myConverterArray[cs0], + sourceChar, &value, + useFallback); + if(len2 != 0 && !(len2 < 0 && len != 0) && GR96_START <= value && value <= GR96_END) { + targetValue = value - 0x80; + len = len2; + cs = cs0; + g = 2; + useFallback = false; + } + break; + default: + /* G0 DBCS */ + len2 = MBCS_FROM_UCHAR32_ISO2022( + converterData->myConverterArray[cs0], + sourceChar, &value, + useFallback, MBCS_OUTPUT_2); + if(len2 == 2 || (len2 == -2 && len == 0)) { /* only accept DBCS: abs(len)==2 */ + if(cs0 == KSC5601) { + /* + * Check for valid bytes for the encoding scheme. + * This is necessary because the sub-converter (windows-949) + * has a broader encoding scheme than is valid for 2022. + */ + value = _2022FromGR94DBCS(value); + if(value == 0) { + break; + } + } + targetValue = value; + len = len2; + cs = cs0; + g = 0; + useFallback = false; + } + break; + } + } + + if(len != 0) { + if(len < 0) { + len = -len; /* fallback */ + } + outLen = 0; /* count output bytes */ + + /* write SI if necessary (only for JIS7) */ + if(pFromU2022State->g == 1 && g == 0) { + buffer[outLen++] = UCNV_SI; + pFromU2022State->g = 0; + } + + /* write the designation sequence if necessary */ + if(cs != pFromU2022State->cs[g]) { + int32_t escLen = escSeqCharsLen[cs]; + uprv_memcpy(buffer + outLen, escSeqChars[cs], escLen); + outLen += escLen; + pFromU2022State->cs[g] = cs; + + /* invalidate the choices[] */ + choiceCount = 0; + } + + /* write the shift sequence if necessary */ + if(g != pFromU2022State->g) { + switch(g) { + /* case 0 handled before writing escapes */ + case 1: + buffer[outLen++] = UCNV_SO; + pFromU2022State->g = 1; + break; + default: /* case 2 */ + buffer[outLen++] = 0x1b; + buffer[outLen++] = 0x4e; + break; + /* no case 3: no SS3 in ISO-2022-JP-x */ + } + } + + /* write the output bytes */ + if(len == 1) { + buffer[outLen++] = (char)targetValue; + } else /* len == 2 */ { + buffer[outLen++] = (char)(targetValue >> 8); + buffer[outLen++] = (char)targetValue; + } + } else { + /* + * if we cannot find the character after checking all codepages + * then this is an error + */ + *err = U_INVALID_CHAR_FOUND; + cnv->fromUChar32=sourceChar; + break; + } + + if(sourceChar == CR || sourceChar == LF) { + /* reset the G2 state at the end of a line (conversion got us into ASCII or JISX201 already) */ + pFromU2022State->cs[2] = 0; + choiceCount = 0; + } + + /* output outLen>0 bytes in buffer[] */ + if(outLen == 1) { + *target++ = buffer[0]; + if(offsets) { + *offsets++ = (int32_t)(source - args->source - 1); /* -1: known to be ASCII */ + } + } else if(outLen == 2 && (target + 2) <= targetLimit) { + *target++ = buffer[0]; + *target++ = buffer[1]; + if(offsets) { + int32_t sourceIndex = (int32_t)(source - args->source - U16_LENGTH(sourceChar)); + *offsets++ = sourceIndex; + *offsets++ = sourceIndex; + } + } else { + fromUWriteUInt8( + cnv, + buffer, outLen, + &target, (const char *)targetLimit, + &offsets, (int32_t)(source - args->source - U16_LENGTH(sourceChar)), + err); + if(U_FAILURE(*err)) { + break; + } + } + } /* end if(myTargetIndexg!=0 || pFromU2022State->cs[0]!=ASCII) && + args->flush && source>=sourceLimit && cnv->fromUChar32==0 + ) { + int32_t sourceIndex; + + outLen = 0; + + if(pFromU2022State->g != 0) { + buffer[outLen++] = UCNV_SI; + pFromU2022State->g = 0; + } + + if(pFromU2022State->cs[0] != ASCII) { + int32_t escLen = escSeqCharsLen[ASCII]; + uprv_memcpy(buffer + outLen, escSeqChars[ASCII], escLen); + outLen += escLen; + pFromU2022State->cs[0] = (int8_t)ASCII; + } + + /* get the source index of the last input character */ + /* + * TODO this would be simpler and more reliable if we used a pair + * of sourceIndex/prevSourceIndex like in ucnvmbcs.c + * so that we could simply use the prevSourceIndex here; + * this code gives an incorrect result for the rare case of an unmatched + * trail surrogate that is alone in the last buffer of the text stream + */ + sourceIndex=(int32_t)(source-args->source); + if(sourceIndex>0) { + --sourceIndex; + if( U16_IS_TRAIL(args->source[sourceIndex]) && + (sourceIndex==0 || U16_IS_LEAD(args->source[sourceIndex-1])) + ) { + --sourceIndex; + } + } else { + sourceIndex=-1; + } + + fromUWriteUInt8( + cnv, + buffer, outLen, + &target, (const char *)targetLimit, + &offsets, sourceIndex, + err); + } + + /*save the state and return */ + args->source = source; + args->target = (char*)target; +} + +/*************** to unicode *******************/ + +static void U_CALLCONV +UConverter_toUnicode_ISO_2022_JP_OFFSETS_LOGIC(UConverterToUnicodeArgs *args, + UErrorCode* err){ + char tempBuf[2]; + const char *mySource = (char *) args->source; + char16_t *myTarget = args->target; + const char *mySourceLimit = args->sourceLimit; + uint32_t targetUniChar = 0x0000; + uint32_t mySourceChar = 0x0000; + uint32_t tmpSourceChar = 0x0000; + UConverterDataISO2022* myData; + ISO2022State *pToU2022State; + StateEnum cs; + + myData=(UConverterDataISO2022*)(args->converter->extraInfo); + pToU2022State = &myData->toU2022State; + + if(myData->key != 0) { + /* continue with a partial escape sequence */ + goto escape; + } else if(args->converter->toULength == 1 && mySource < mySourceLimit && myTarget < args->targetLimit) { + /* continue with a partial double-byte character */ + mySourceChar = args->converter->toUBytes[0]; + args->converter->toULength = 0; + cs = (StateEnum)pToU2022State->cs[pToU2022State->g]; + targetUniChar = missingCharMarker; + goto getTrailByte; + } + + while(mySource < mySourceLimit){ + + targetUniChar =missingCharMarker; + + if(myTarget < args->targetLimit){ + + mySourceChar= (unsigned char) *mySource++; + + switch(mySourceChar) { + case UCNV_SI: + if(myData->version==3) { + pToU2022State->g=0; + continue; + } else { + /* only JIS7 uses SI/SO, not ISO-2022-JP-x */ + myData->isEmptySegment = false; /* reset this, we have a different error */ + break; + } + + case UCNV_SO: + if(myData->version==3) { + /* JIS7: switch to G1 half-width Katakana */ + pToU2022State->cs[1] = (int8_t)HWKANA_7BIT; + pToU2022State->g=1; + continue; + } else { + /* only JIS7 uses SI/SO, not ISO-2022-JP-x */ + myData->isEmptySegment = false; /* reset this, we have a different error */ + break; + } + + case ESC_2022: + mySource--; +escape: + { + const char * mySourceBefore = mySource; + int8_t toULengthBefore = args->converter->toULength; + + changeState_2022(args->converter,&(mySource), + mySourceLimit, ISO_2022_JP,err); + + /* If in ISO-2022-JP only and we successfully completed an escape sequence, but previous segment was empty, create an error */ + if(myData->version==0 && myData->key==0 && U_SUCCESS(*err) && myData->isEmptySegment) { + *err = U_ILLEGAL_ESCAPE_SEQUENCE; + args->converter->toUCallbackReason = UCNV_IRREGULAR; + args->converter->toULength = (int8_t)(toULengthBefore + (mySource - mySourceBefore)); + } + } + + /* invalid or illegal escape sequence */ + if(U_FAILURE(*err)){ + args->target = myTarget; + args->source = mySource; + myData->isEmptySegment = false; /* Reset to avoid future spurious errors */ + return; + } + /* If we successfully completed an escape sequence, we begin a new segment, empty so far */ + if(myData->key==0) { + myData->isEmptySegment = true; + } + continue; + + /* ISO-2022-JP does not use single-byte (C1) SS2 and SS3 */ + + case CR: + case LF: + /* automatically reset to single-byte mode */ + if((StateEnum)pToU2022State->cs[0] != ASCII && (StateEnum)pToU2022State->cs[0] != JISX201) { + pToU2022State->cs[0] = (int8_t)ASCII; + } + pToU2022State->cs[2] = 0; + pToU2022State->g = 0; + U_FALLTHROUGH; + default: + /* convert one or two bytes */ + myData->isEmptySegment = false; + cs = (StateEnum)pToU2022State->cs[pToU2022State->g]; + if( (uint8_t)(mySourceChar - 0xa1) <= (0xdf - 0xa1) && myData->version==4 && + !IS_JP_DBCS(cs) + ) { + /* 8-bit halfwidth katakana in any single-byte mode for JIS8 */ + targetUniChar = mySourceChar + (HWKANA_START - 0xa1); + + /* return from a single-shift state to the previous one */ + if(pToU2022State->g >= 2) { + pToU2022State->g=pToU2022State->prevG; + } + } else switch(cs) { + case ASCII: + if(mySourceChar <= 0x7f) { + targetUniChar = mySourceChar; + } + break; + case ISO8859_1: + if(mySourceChar <= 0x7f) { + targetUniChar = mySourceChar + 0x80; + } + /* return from a single-shift state to the previous one */ + pToU2022State->g=pToU2022State->prevG; + break; + case ISO8859_7: + if(mySourceChar <= 0x7f) { + /* convert mySourceChar+0x80 to use a normal 8-bit table */ + targetUniChar = + _MBCS_SINGLE_SIMPLE_GET_NEXT_BMP( + myData->myConverterArray[cs], + mySourceChar + 0x80); + } + /* return from a single-shift state to the previous one */ + pToU2022State->g=pToU2022State->prevG; + break; + case JISX201: + if(mySourceChar <= 0x7f) { + targetUniChar = jisx201ToU(mySourceChar); + } + break; + case HWKANA_7BIT: + if((uint8_t)(mySourceChar - 0x21) <= (0x5f - 0x21)) { + /* 7-bit halfwidth Katakana */ + targetUniChar = mySourceChar + (HWKANA_START - 0x21); + } + break; + default: + /* G0 DBCS */ + if(mySource < mySourceLimit) { + int leadIsOk, trailIsOk; + uint8_t trailByte; +getTrailByte: + trailByte = (uint8_t)*mySource; + /* + * Ticket 5691: consistent illegal sequences: + * - We include at least the first byte in the illegal sequence. + * - If any of the non-initial bytes could be the start of a character, + * we stop the illegal sequence before the first one of those. + * + * In ISO-2022 DBCS, if the second byte is in the 21..7e range or is + * an ESC/SO/SI, we report only the first byte as the illegal sequence. + * Otherwise we convert or report the pair of bytes. + */ + leadIsOk = (uint8_t)(mySourceChar - 0x21) <= (0x7e - 0x21); + trailIsOk = (uint8_t)(trailByte - 0x21) <= (0x7e - 0x21); + if (leadIsOk && trailIsOk) { + ++mySource; + tmpSourceChar = (mySourceChar << 8) | trailByte; + if(cs == JISX208) { + _2022ToSJIS((uint8_t)mySourceChar, trailByte, tempBuf); + mySourceChar = tmpSourceChar; + } else { + /* Copy before we modify tmpSourceChar so toUnicodeCallback() sees the correct bytes. */ + mySourceChar = tmpSourceChar; + if (cs == KSC5601) { + tmpSourceChar += 0x8080; /* = _2022ToGR94DBCS(tmpSourceChar) */ + } + tempBuf[0] = (char)(tmpSourceChar >> 8); + tempBuf[1] = (char)(tmpSourceChar); + } + targetUniChar = ucnv_MBCSSimpleGetNextUChar(myData->myConverterArray[cs], tempBuf, 2, false); + } else if (!(trailIsOk || IS_2022_CONTROL(trailByte))) { + /* report a pair of illegal bytes if the second byte is not a DBCS starter */ + ++mySource; + /* add another bit so that the code below writes 2 bytes in case of error */ + mySourceChar = 0x10000 | (mySourceChar << 8) | trailByte; + } + } else { + args->converter->toUBytes[0] = (uint8_t)mySourceChar; + args->converter->toULength = 1; + goto endloop; + } + } /* End of inner switch */ + break; + } /* End of outer switch */ + if(targetUniChar < (missingCharMarker-1/*0xfffe*/)){ + if(args->offsets){ + args->offsets[myTarget - args->target] = (int32_t)(mySource - args->source - (mySourceChar <= 0xff ? 1 : 2)); + } + *(myTarget++)=(char16_t)targetUniChar; + } + else if(targetUniChar > missingCharMarker){ + /* disassemble the surrogate pair and write to output*/ + targetUniChar-=0x0010000; + *myTarget = (char16_t)(0xd800+(char16_t)(targetUniChar>>10)); + if(args->offsets){ + args->offsets[myTarget - args->target] = (int32_t)(mySource - args->source - (mySourceChar <= 0xff ? 1 : 2)); + } + ++myTarget; + if(myTarget< args->targetLimit){ + *myTarget = (char16_t)(0xdc00+(char16_t)(targetUniChar&0x3ff)); + if(args->offsets){ + args->offsets[myTarget - args->target] = (int32_t)(mySource - args->source - (mySourceChar <= 0xff ? 1 : 2)); + } + ++myTarget; + }else{ + args->converter->UCharErrorBuffer[args->converter->UCharErrorBufferLength++]= + (char16_t)(0xdc00+(char16_t)(targetUniChar&0x3ff)); + } + + } + else{ + /* Call the callback function*/ + toUnicodeCallback(args->converter,mySourceChar,targetUniChar,err); + break; + } + } + else{ /* goes with "if(myTarget < args->targetLimit)" way up near top of function */ + *err =U_BUFFER_OVERFLOW_ERROR; + break; + } + } +endloop: + args->target = myTarget; + args->source = mySource; +} + + +#if !UCONFIG_ONLY_HTML_CONVERSION +/*************************************************************** +* Rules for ISO-2022-KR encoding +* i) The KSC5601 designator sequence should appear only once in a file, +* at the beginning of a line before any KSC5601 characters. This usually +* means that it appears by itself on the first line of the file +* ii) There are only 2 shifting sequences SO to shift into double byte mode +* and SI to shift into single byte mode +*/ +static void U_CALLCONV +UConverter_fromUnicode_ISO_2022_KR_OFFSETS_LOGIC_IBM(UConverterFromUnicodeArgs* args, UErrorCode* err){ + + UConverter* saveConv = args->converter; + UConverterDataISO2022 *myConverterData=(UConverterDataISO2022*)saveConv->extraInfo; + args->converter=myConverterData->currentConverter; + + myConverterData->currentConverter->fromUChar32 = saveConv->fromUChar32; + ucnv_MBCSFromUnicodeWithOffsets(args,err); + saveConv->fromUChar32 = myConverterData->currentConverter->fromUChar32; + + if(*err == U_BUFFER_OVERFLOW_ERROR) { + if(myConverterData->currentConverter->charErrorBufferLength > 0) { + uprv_memcpy( + saveConv->charErrorBuffer, + myConverterData->currentConverter->charErrorBuffer, + myConverterData->currentConverter->charErrorBufferLength); + } + saveConv->charErrorBufferLength = myConverterData->currentConverter->charErrorBufferLength; + myConverterData->currentConverter->charErrorBufferLength = 0; + } + args->converter=saveConv; +} + +static void U_CALLCONV +UConverter_fromUnicode_ISO_2022_KR_OFFSETS_LOGIC(UConverterFromUnicodeArgs* args, UErrorCode* err){ + + const char16_t *source = args->source; + const char16_t *sourceLimit = args->sourceLimit; + unsigned char *target = (unsigned char *) args->target; + unsigned char *targetLimit = (unsigned char *) args->targetLimit; + int32_t* offsets = args->offsets; + uint32_t targetByteUnit = 0x0000; + UChar32 sourceChar = 0x0000; + UBool isTargetByteDBCS; + UBool oldIsTargetByteDBCS; + UConverterDataISO2022 *converterData; + UConverterSharedData* sharedData; + UBool useFallback; + int32_t length =0; + + converterData=(UConverterDataISO2022*)args->converter->extraInfo; + /* if the version is 1 then the user is requesting + * conversion with ibm-25546 pass the arguments to + * MBCS converter and return + */ + if(converterData->version==1){ + UConverter_fromUnicode_ISO_2022_KR_OFFSETS_LOGIC_IBM(args,err); + return; + } + + /* initialize data */ + sharedData = converterData->currentConverter->sharedData; + useFallback = args->converter->useFallback; + isTargetByteDBCS=(UBool)args->converter->fromUnicodeStatus; + oldIsTargetByteDBCS = isTargetByteDBCS; + + isTargetByteDBCS = (UBool) args->converter->fromUnicodeStatus; + if((sourceChar = args->converter->fromUChar32)!=0 && target targetLimit){ + sourceChar = *source++; + + /* do not convert SO/SI/ESC */ + if(IS_2022_CONTROL(sourceChar)) { + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + args->converter->fromUChar32=sourceChar; + break; + } + + length = MBCS_FROM_UCHAR32_ISO2022(sharedData,sourceChar,&targetByteUnit,useFallback,MBCS_OUTPUT_2); + if(length < 0) { + length = -length; /* fallback */ + } + /* only DBCS or SBCS characters are expected*/ + /* DB characters with high bit set to 1 are expected */ + if( length > 2 || length==0 || + (length == 1 && targetByteUnit > 0x7f) || + (length == 2 && + ((uint16_t)(targetByteUnit - 0xa1a1) > (0xfefe - 0xa1a1) || + (uint8_t)(targetByteUnit - 0xa1) > (0xfe - 0xa1))) + ) { + targetByteUnit=missingCharMarker; + } + if (targetByteUnit != missingCharMarker){ + + oldIsTargetByteDBCS = isTargetByteDBCS; + isTargetByteDBCS = (UBool)(targetByteUnit>0x00FF); + /* append the shift sequence */ + if (oldIsTargetByteDBCS != isTargetByteDBCS ){ + + if (isTargetByteDBCS) + *target++ = UCNV_SO; + else + *target++ = UCNV_SI; + if(offsets) + *(offsets++) = (int32_t)(source - args->source-1); + } + /* write the targetUniChar to target */ + if(targetByteUnit <= 0x00FF){ + if( target < targetLimit){ + *(target++) = (unsigned char) targetByteUnit; + if(offsets){ + *(offsets++) = (int32_t)(source - args->source-1); + } + + }else{ + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = (unsigned char) (targetByteUnit); + *err = U_BUFFER_OVERFLOW_ERROR; + } + }else{ + if(target < targetLimit){ + *(target++) =(unsigned char) ((targetByteUnit>>8) -0x80); + if(offsets){ + *(offsets++) = (int32_t)(source - args->source-1); + } + if(target < targetLimit){ + *(target++) =(unsigned char) (targetByteUnit -0x80); + if(offsets){ + *(offsets++) = (int32_t)(source - args->source-1); + } + }else{ + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = (unsigned char) (targetByteUnit -0x80); + *err = U_BUFFER_OVERFLOW_ERROR; + } + }else{ + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = (unsigned char) ((targetByteUnit>>8) -0x80); + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = (unsigned char) (targetByteUnit-0x80); + *err = U_BUFFER_OVERFLOW_ERROR; + } + } + + } + else{ + /* oops.. the code point is unassingned + * set the error and reason + */ + + /*check if the char is a First surrogate*/ + if(U16_IS_SURROGATE(sourceChar)) { + if(U16_IS_SURROGATE_LEAD(sourceChar)) { +getTrail: + /*look ahead to find the trail surrogate*/ + if(source < sourceLimit) { + /* test the following code unit */ + char16_t trail=(char16_t) *source; + if(U16_IS_TRAIL(trail)) { + source++; + sourceChar=U16_GET_SUPPLEMENTARY(sourceChar, trail); + *err = U_INVALID_CHAR_FOUND; + /* convert this surrogate code point */ + /* exit this condition tree */ + } else { + /* this is an unmatched lead code unit (1st surrogate) */ + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + } + } else { + /* no more input */ + *err = U_ZERO_ERROR; + } + } else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + } + } else { + /* callback(unassigned) for a BMP code point */ + *err = U_INVALID_CHAR_FOUND; + } + + args->converter->fromUChar32=sourceChar; + break; + } + } /* end if(myTargetIndexflush && source>=sourceLimit && args->converter->fromUChar32==0 + ) { + int32_t sourceIndex; + + /* we are switching to ASCII */ + isTargetByteDBCS=false; + + /* get the source index of the last input character */ + /* + * TODO this would be simpler and more reliable if we used a pair + * of sourceIndex/prevSourceIndex like in ucnvmbcs.c + * so that we could simply use the prevSourceIndex here; + * this code gives an incorrect result for the rare case of an unmatched + * trail surrogate that is alone in the last buffer of the text stream + */ + sourceIndex=(int32_t)(source-args->source); + if(sourceIndex>0) { + --sourceIndex; + if( U16_IS_TRAIL(args->source[sourceIndex]) && + (sourceIndex==0 || U16_IS_LEAD(args->source[sourceIndex-1])) + ) { + --sourceIndex; + } + } else { + sourceIndex=-1; + } + + fromUWriteUInt8( + args->converter, + SHIFT_IN_STR, 1, + &target, (const char *)targetLimit, + &offsets, sourceIndex, + err); + } + + /*save the state and return */ + args->source = source; + args->target = (char*)target; + args->converter->fromUnicodeStatus = (uint32_t)isTargetByteDBCS; +} + +/************************ To Unicode ***************************************/ + +static void U_CALLCONV +UConverter_toUnicode_ISO_2022_KR_OFFSETS_LOGIC_IBM(UConverterToUnicodeArgs *args, + UErrorCode* err){ + char const* sourceStart; + UConverterDataISO2022* myData=(UConverterDataISO2022*)(args->converter->extraInfo); + + UConverterToUnicodeArgs subArgs; + int32_t minArgsSize; + + /* set up the subconverter arguments */ + if(args->sizesize; + } else { + minArgsSize = (int32_t)sizeof(UConverterToUnicodeArgs); + } + + uprv_memcpy(&subArgs, args, minArgsSize); + subArgs.size = (uint16_t)minArgsSize; + subArgs.converter = myData->currentConverter; + + /* remember the original start of the input for offsets */ + sourceStart = args->source; + + if(myData->key != 0) { + /* continue with a partial escape sequence */ + goto escape; + } + + while(U_SUCCESS(*err) && args->source < args->sourceLimit) { + /*Find the end of the buffer e.g : Next Escape Seq | end of Buffer*/ + subArgs.source = args->source; + subArgs.sourceLimit = getEndOfBuffer_2022(&(args->source), args->sourceLimit, args->flush); + if(subArgs.source != subArgs.sourceLimit) { + /* + * get the current partial byte sequence + * + * it needs to be moved between the public and the subconverter + * so that the conversion framework, which only sees the public + * converter, can handle truncated and illegal input etc. + */ + if(args->converter->toULength > 0) { + uprv_memcpy(subArgs.converter->toUBytes, args->converter->toUBytes, args->converter->toULength); + } + subArgs.converter->toULength = args->converter->toULength; + + /* + * Convert up to the end of the input, or to before the next escape character. + * Does not handle conversion extensions because the preToU[] state etc. + * is not copied. + */ + ucnv_MBCSToUnicodeWithOffsets(&subArgs, err); + + if(args->offsets != nullptr && sourceStart != args->source) { + /* update offsets to base them on the actual start of the input */ + int32_t *offsets = args->offsets; + char16_t *target = args->target; + int32_t delta = (int32_t)(args->source - sourceStart); + while(target < subArgs.target) { + if(*offsets >= 0) { + *offsets += delta; + } + ++offsets; + ++target; + } + } + args->source = subArgs.source; + args->target = subArgs.target; + args->offsets = subArgs.offsets; + + /* copy input/error/overflow buffers */ + if(subArgs.converter->toULength > 0) { + uprv_memcpy(args->converter->toUBytes, subArgs.converter->toUBytes, subArgs.converter->toULength); + } + args->converter->toULength = subArgs.converter->toULength; + + if(*err == U_BUFFER_OVERFLOW_ERROR) { + if(subArgs.converter->UCharErrorBufferLength > 0) { + uprv_memcpy(args->converter->UCharErrorBuffer, subArgs.converter->UCharErrorBuffer, + subArgs.converter->UCharErrorBufferLength); + } + args->converter->UCharErrorBufferLength=subArgs.converter->UCharErrorBufferLength; + subArgs.converter->UCharErrorBufferLength = 0; + } + } + + if (U_FAILURE(*err) || (args->source == args->sourceLimit)) { + return; + } + +escape: + changeState_2022(args->converter, + &(args->source), + args->sourceLimit, + ISO_2022_KR, + err); + } +} + +static void U_CALLCONV +UConverter_toUnicode_ISO_2022_KR_OFFSETS_LOGIC(UConverterToUnicodeArgs *args, + UErrorCode* err){ + char tempBuf[2]; + const char *mySource = ( char *) args->source; + char16_t *myTarget = args->target; + const char *mySourceLimit = args->sourceLimit; + UChar32 targetUniChar = 0x0000; + char16_t mySourceChar = 0x0000; + UConverterDataISO2022* myData; + UConverterSharedData* sharedData ; + UBool useFallback; + + myData=(UConverterDataISO2022*)(args->converter->extraInfo); + if(myData->version==1){ + UConverter_toUnicode_ISO_2022_KR_OFFSETS_LOGIC_IBM(args,err); + return; + } + + /* initialize state */ + sharedData = myData->currentConverter->sharedData; + useFallback = args->converter->useFallback; + + if(myData->key != 0) { + /* continue with a partial escape sequence */ + goto escape; + } else if(args->converter->toULength == 1 && mySource < mySourceLimit && myTarget < args->targetLimit) { + /* continue with a partial double-byte character */ + mySourceChar = args->converter->toUBytes[0]; + args->converter->toULength = 0; + goto getTrailByte; + } + + while(mySource< mySourceLimit){ + + if(myTarget < args->targetLimit){ + + mySourceChar= (unsigned char) *mySource++; + + if(mySourceChar==UCNV_SI){ + myData->toU2022State.g = 0; + if (myData->isEmptySegment) { + myData->isEmptySegment = false; /* we are handling it, reset to avoid future spurious errors */ + *err = U_ILLEGAL_ESCAPE_SEQUENCE; + args->converter->toUCallbackReason = UCNV_IRREGULAR; + args->converter->toUBytes[0] = (uint8_t)mySourceChar; + args->converter->toULength = 1; + args->target = myTarget; + args->source = mySource; + return; + } + /*consume the source */ + continue; + }else if(mySourceChar==UCNV_SO){ + myData->toU2022State.g = 1; + myData->isEmptySegment = true; /* Begin a new segment, empty so far */ + /*consume the source */ + continue; + }else if(mySourceChar==ESC_2022){ + mySource--; +escape: + myData->isEmptySegment = false; /* Any invalid ESC sequences will be detected separately, so just reset this */ + changeState_2022(args->converter,&(mySource), + mySourceLimit, ISO_2022_KR, err); + if(U_FAILURE(*err)){ + args->target = myTarget; + args->source = mySource; + return; + } + continue; + } + + myData->isEmptySegment = false; /* Any invalid char errors will be detected separately, so just reset this */ + if(myData->toU2022State.g == 1) { + if(mySource < mySourceLimit) { + int leadIsOk, trailIsOk; + uint8_t trailByte; +getTrailByte: + targetUniChar = missingCharMarker; + trailByte = (uint8_t)*mySource; + /* + * Ticket 5691: consistent illegal sequences: + * - We include at least the first byte in the illegal sequence. + * - If any of the non-initial bytes could be the start of a character, + * we stop the illegal sequence before the first one of those. + * + * In ISO-2022 DBCS, if the second byte is in the 21..7e range or is + * an ESC/SO/SI, we report only the first byte as the illegal sequence. + * Otherwise we convert or report the pair of bytes. + */ + leadIsOk = (uint8_t)(mySourceChar - 0x21) <= (0x7e - 0x21); + trailIsOk = (uint8_t)(trailByte - 0x21) <= (0x7e - 0x21); + if (leadIsOk && trailIsOk) { + ++mySource; + tempBuf[0] = (char)(mySourceChar + 0x80); + tempBuf[1] = (char)(trailByte + 0x80); + targetUniChar = ucnv_MBCSSimpleGetNextUChar(sharedData, tempBuf, 2, useFallback); + mySourceChar = (mySourceChar << 8) | trailByte; + } else if (!(trailIsOk || IS_2022_CONTROL(trailByte))) { + /* report a pair of illegal bytes if the second byte is not a DBCS starter */ + ++mySource; + /* add another bit so that the code below writes 2 bytes in case of error */ + mySourceChar = static_cast(0x10000 | (mySourceChar << 8) | trailByte); + } + } else { + args->converter->toUBytes[0] = (uint8_t)mySourceChar; + args->converter->toULength = 1; + break; + } + } + else if(mySourceChar <= 0x7f) { + targetUniChar = ucnv_MBCSSimpleGetNextUChar(sharedData, mySource - 1, 1, useFallback); + } else { + targetUniChar = 0xffff; + } + if(targetUniChar < 0xfffe){ + if(args->offsets) { + args->offsets[myTarget - args->target] = (int32_t)(mySource - args->source - (mySourceChar <= 0xff ? 1 : 2)); + } + *(myTarget++)=(char16_t)targetUniChar; + } + else { + /* Call the callback function*/ + toUnicodeCallback(args->converter,mySourceChar,targetUniChar,err); + break; + } + } + else{ + *err =U_BUFFER_OVERFLOW_ERROR; + break; + } + } + args->target = myTarget; + args->source = mySource; +} + +/*************************** END ISO2022-KR *********************************/ + +/*************************** ISO-2022-CN ********************************* +* +* Rules for ISO-2022-CN Encoding: +* i) The designator sequence must appear once on a line before any instance +* of character set it designates. +* ii) If two lines contain characters from the same character set, both lines +* must include the designator sequence. +* iii) Once the designator sequence is known, a shifting sequence has to be found +* to invoke the shifting +* iv) All lines start in ASCII and end in ASCII. +* v) Four shifting sequences are employed for this purpose: +* +* Sequcence ASCII Eq Charsets +* ---------- ------- --------- +* SI US-ASCII +* SO CNS-11643-1992 Plane 1, GB2312, ISO-IR-165 +* SS2 N CNS-11643-1992 Plane 2 +* SS3 O CNS-11643-1992 Planes 3-7 +* +* vi) +* SOdesignator : ESC "$" ")" finalchar_for_SO +* SS2designator : ESC "$" "*" finalchar_for_SS2 +* SS3designator : ESC "$" "+" finalchar_for_SS3 +* +* ESC $ ) A Indicates the bytes following SO are Chinese +* characters as defined in GB 2312-80, until +* another SOdesignation appears +* +* +* ESC $ ) E Indicates the bytes following SO are as defined +* in ISO-IR-165 (for details, see section 2.1), +* until another SOdesignation appears +* +* ESC $ ) G Indicates the bytes following SO are as defined +* in CNS 11643-plane-1, until another +* SOdesignation appears +* +* ESC $ * H Indicates the two bytes immediately following +* SS2 is a Chinese character as defined in CNS +* 11643-plane-2, until another SS2designation +* appears +* (Meaning N must precede every 2 byte +* sequence.) +* +* ESC $ + I Indicates the immediate two bytes following SS3 +* is a Chinese character as defined in CNS +* 11643-plane-3, until another SS3designation +* appears +* (Meaning O must precede every 2 byte +* sequence.) +* +* ESC $ + J Indicates the immediate two bytes following SS3 +* is a Chinese character as defined in CNS +* 11643-plane-4, until another SS3designation +* appears +* (In English: O must precede every 2 byte +* sequence.) +* +* ESC $ + K Indicates the immediate two bytes following SS3 +* is a Chinese character as defined in CNS +* 11643-plane-5, until another SS3designation +* appears +* +* ESC $ + L Indicates the immediate two bytes following SS3 +* is a Chinese character as defined in CNS +* 11643-plane-6, until another SS3designation +* appears +* +* ESC $ + M Indicates the immediate two bytes following SS3 +* is a Chinese character as defined in CNS +* 11643-plane-7, until another SS3designation +* appears +* +* As in ISO-2022-CN, each line starts in ASCII, and ends in ASCII, and +* has its own designation information before any Chinese characters +* appear +* +*/ + +/* The following are defined this way to make the strings truly readonly */ +static const char GB_2312_80_STR[] = "\x1B\x24\x29\x41"; +static const char ISO_IR_165_STR[] = "\x1B\x24\x29\x45"; +static const char CNS_11643_1992_Plane_1_STR[] = "\x1B\x24\x29\x47"; +static const char CNS_11643_1992_Plane_2_STR[] = "\x1B\x24\x2A\x48"; +static const char CNS_11643_1992_Plane_3_STR[] = "\x1B\x24\x2B\x49"; +static const char CNS_11643_1992_Plane_4_STR[] = "\x1B\x24\x2B\x4A"; +static const char CNS_11643_1992_Plane_5_STR[] = "\x1B\x24\x2B\x4B"; +static const char CNS_11643_1992_Plane_6_STR[] = "\x1B\x24\x2B\x4C"; +static const char CNS_11643_1992_Plane_7_STR[] = "\x1B\x24\x2B\x4D"; + +/********************** ISO2022-CN Data **************************/ +static const char* const escSeqCharsCN[10] ={ + SHIFT_IN_STR, /* 0 ASCII */ + GB_2312_80_STR, /* 1 GB2312_1 */ + ISO_IR_165_STR, /* 2 ISO_IR_165 */ + CNS_11643_1992_Plane_1_STR, + CNS_11643_1992_Plane_2_STR, + CNS_11643_1992_Plane_3_STR, + CNS_11643_1992_Plane_4_STR, + CNS_11643_1992_Plane_5_STR, + CNS_11643_1992_Plane_6_STR, + CNS_11643_1992_Plane_7_STR +}; + +static void U_CALLCONV +UConverter_fromUnicode_ISO_2022_CN_OFFSETS_LOGIC(UConverterFromUnicodeArgs* args, UErrorCode* err){ + UConverter *cnv = args->converter; + UConverterDataISO2022 *converterData; + ISO2022State *pFromU2022State; + uint8_t *target = (uint8_t *) args->target; + const uint8_t *targetLimit = (const uint8_t *) args->targetLimit; + const char16_t* source = args->source; + const char16_t* sourceLimit = args->sourceLimit; + int32_t* offsets = args->offsets; + UChar32 sourceChar; + char buffer[8]; + int32_t len; + int8_t choices[3]; + int32_t choiceCount; + uint32_t targetValue = 0; + UBool useFallback; + + /* set up the state */ + converterData = (UConverterDataISO2022*)cnv->extraInfo; + pFromU2022State = &converterData->fromU2022State; + + choiceCount = 0; + + /* check if the last codepoint of previous buffer was a lead surrogate*/ + if((sourceChar = cnv->fromUChar32)!=0 && target< targetLimit) { + goto getTrail; + } + + while( source < sourceLimit){ + if(target < targetLimit){ + + sourceChar = *(source++); + /*check if the char is a First surrogate*/ + if(U16_IS_SURROGATE(sourceChar)) { + if(U16_IS_SURROGATE_LEAD(sourceChar)) { +getTrail: + /*look ahead to find the trail surrogate*/ + if(source < sourceLimit) { + /* test the following code unit */ + char16_t trail=(char16_t) *source; + if(U16_IS_TRAIL(trail)) { + source++; + sourceChar=U16_GET_SUPPLEMENTARY(sourceChar, trail); + cnv->fromUChar32=0x00; + /* convert this supplementary code point */ + /* exit this condition tree */ + } else { + /* this is an unmatched lead code unit (1st surrogate) */ + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + cnv->fromUChar32=sourceChar; + break; + } + } else { + /* no more input */ + cnv->fromUChar32=sourceChar; + break; + } + } else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + cnv->fromUChar32=sourceChar; + break; + } + } + + /* do the conversion */ + if(sourceChar <= 0x007f ){ + /* do not convert SO/SI/ESC */ + if(IS_2022_CONTROL(sourceChar)) { + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + cnv->fromUChar32=sourceChar; + break; + } + + /* US-ASCII */ + if(pFromU2022State->g == 0) { + buffer[0] = (char)sourceChar; + len = 1; + } else { + buffer[0] = UCNV_SI; + buffer[1] = (char)sourceChar; + len = 2; + pFromU2022State->g = 0; + choiceCount = 0; + } + if(sourceChar == CR || sourceChar == LF) { + /* reset the state at the end of a line */ + uprv_memset(pFromU2022State, 0, sizeof(ISO2022State)); + choiceCount = 0; + } + } + else{ + /* convert U+0080..U+10ffff */ + int32_t i; + int8_t cs, g; + + if(choiceCount == 0) { + /* try the current SO/G1 converter first */ + choices[0] = pFromU2022State->cs[1]; + + /* default to GB2312_1 if none is designated yet */ + if(choices[0] == 0) { + choices[0] = GB2312_1; + } + + if(converterData->version == 0) { + /* ISO-2022-CN */ + + /* try the other SO/G1 converter; a CNS_11643_1 lookup may result in any plane */ + if(choices[0] == GB2312_1) { + choices[1] = (int8_t)CNS_11643_1; + } else { + choices[1] = (int8_t)GB2312_1; + } + + choiceCount = 2; + } else if (converterData->version == 1) { + /* ISO-2022-CN-EXT */ + + /* try one of the other converters */ + switch(choices[0]) { + case GB2312_1: + choices[1] = (int8_t)CNS_11643_1; + choices[2] = (int8_t)ISO_IR_165; + break; + case ISO_IR_165: + choices[1] = (int8_t)GB2312_1; + choices[2] = (int8_t)CNS_11643_1; + break; + default: /* CNS_11643_x */ + choices[1] = (int8_t)GB2312_1; + choices[2] = (int8_t)ISO_IR_165; + break; + } + + choiceCount = 3; + } else { + choices[0] = (int8_t)CNS_11643_1; + choices[1] = (int8_t)GB2312_1; + } + } + + cs = g = 0; + /* + * len==0: no mapping found yet + * len<0: found a fallback result: continue looking for a roundtrip but no further fallbacks + * len>0: found a roundtrip result, done + */ + len = 0; + /* + * We will turn off useFallback after finding a fallback, + * but we still get fallbacks from PUA code points as usual. + * Therefore, we will also need to check that we don't overwrite + * an early fallback with a later one. + */ + useFallback = cnv->useFallback; + + for(i = 0; i < choiceCount && len <= 0; ++i) { + int8_t cs0 = choices[i]; + if(cs0 > 0) { + uint32_t value; + int32_t len2; + if(cs0 >= CNS_11643_0) { + len2 = MBCS_FROM_UCHAR32_ISO2022( + converterData->myConverterArray[CNS_11643], + sourceChar, + &value, + useFallback, + MBCS_OUTPUT_3); + if(len2 == 3 || (len2 == -3 && len == 0)) { + targetValue = value; + cs = (int8_t)(CNS_11643_0 + (value >> 16) - 0x80); + if(len2 >= 0) { + len = 2; + } else { + len = -2; + useFallback = false; + } + if(cs == CNS_11643_1) { + g = 1; + } else if(cs == CNS_11643_2) { + g = 2; + } else /* plane 3..7 */ if(converterData->version == 1) { + g = 3; + } else { + /* ISO-2022-CN (without -EXT) does not support plane 3..7 */ + len = 0; + } + } + } else { + /* GB2312_1 or ISO-IR-165 */ + U_ASSERT(cs0myConverterArray[cs0], + sourceChar, + &value, + useFallback, + MBCS_OUTPUT_2); + if(len2 == 2 || (len2 == -2 && len == 0)) { + targetValue = value; + len = len2; + cs = cs0; + g = 1; + useFallback = false; + } + } + } + } + + if(len != 0) { + len = 0; /* count output bytes; it must have been abs(len) == 2 */ + + /* write the designation sequence if necessary */ + if(cs != pFromU2022State->cs[g]) { + if(cs < CNS_11643) { + uprv_memcpy(buffer, escSeqCharsCN[cs], 4); + } else { + U_ASSERT(cs >= CNS_11643_1); + uprv_memcpy(buffer, escSeqCharsCN[CNS_11643 + (cs - CNS_11643_1)], 4); + } + len = 4; + pFromU2022State->cs[g] = cs; + if(g == 1) { + /* changing the SO/G1 charset invalidates the choices[] */ + choiceCount = 0; + } + } + + /* write the shift sequence if necessary */ + if(g != pFromU2022State->g) { + switch(g) { + case 1: + buffer[len++] = UCNV_SO; + + /* set the new state only if it is the locking shift SO/G1, not for SS2 or SS3 */ + pFromU2022State->g = 1; + break; + case 2: + buffer[len++] = 0x1b; + buffer[len++] = 0x4e; + break; + default: /* case 3 */ + buffer[len++] = 0x1b; + buffer[len++] = 0x4f; + break; + } + } + + /* write the two output bytes */ + buffer[len++] = (char)(targetValue >> 8); + buffer[len++] = (char)targetValue; + } else { + /* if we cannot find the character after checking all codepages + * then this is an error + */ + *err = U_INVALID_CHAR_FOUND; + cnv->fromUChar32=sourceChar; + break; + } + } + + /* output len>0 bytes in buffer[] */ + if(len == 1) { + *target++ = buffer[0]; + if(offsets) { + *offsets++ = (int32_t)(source - args->source - 1); /* -1: known to be ASCII */ + } + } else if(len == 2 && (target + 2) <= targetLimit) { + *target++ = buffer[0]; + *target++ = buffer[1]; + if(offsets) { + int32_t sourceIndex = (int32_t)(source - args->source - U16_LENGTH(sourceChar)); + *offsets++ = sourceIndex; + *offsets++ = sourceIndex; + } + } else { + fromUWriteUInt8( + cnv, + buffer, len, + &target, (const char *)targetLimit, + &offsets, (int32_t)(source - args->source - U16_LENGTH(sourceChar)), + err); + if(U_FAILURE(*err)) { + break; + } + } + } /* end if(myTargetIndexg!=0 && + args->flush && source>=sourceLimit && cnv->fromUChar32==0 + ) { + int32_t sourceIndex; + + /* we are switching to ASCII */ + pFromU2022State->g=0; + + /* get the source index of the last input character */ + /* + * TODO this would be simpler and more reliable if we used a pair + * of sourceIndex/prevSourceIndex like in ucnvmbcs.c + * so that we could simply use the prevSourceIndex here; + * this code gives an incorrect result for the rare case of an unmatched + * trail surrogate that is alone in the last buffer of the text stream + */ + sourceIndex=(int32_t)(source-args->source); + if(sourceIndex>0) { + --sourceIndex; + if( U16_IS_TRAIL(args->source[sourceIndex]) && + (sourceIndex==0 || U16_IS_LEAD(args->source[sourceIndex-1])) + ) { + --sourceIndex; + } + } else { + sourceIndex=-1; + } + + fromUWriteUInt8( + cnv, + SHIFT_IN_STR, 1, + &target, (const char *)targetLimit, + &offsets, sourceIndex, + err); + } + + /*save the state and return */ + args->source = source; + args->target = (char*)target; +} + + +static void U_CALLCONV +UConverter_toUnicode_ISO_2022_CN_OFFSETS_LOGIC(UConverterToUnicodeArgs *args, + UErrorCode* err){ + char tempBuf[3]; + const char *mySource = (char *) args->source; + char16_t *myTarget = args->target; + const char *mySourceLimit = args->sourceLimit; + uint32_t targetUniChar = 0x0000; + uint32_t mySourceChar = 0x0000; + UConverterDataISO2022* myData; + ISO2022State *pToU2022State; + + myData=(UConverterDataISO2022*)(args->converter->extraInfo); + pToU2022State = &myData->toU2022State; + + if(myData->key != 0) { + /* continue with a partial escape sequence */ + goto escape; + } else if(args->converter->toULength == 1 && mySource < mySourceLimit && myTarget < args->targetLimit) { + /* continue with a partial double-byte character */ + mySourceChar = args->converter->toUBytes[0]; + args->converter->toULength = 0; + targetUniChar = missingCharMarker; + goto getTrailByte; + } + + while(mySource < mySourceLimit){ + + targetUniChar =missingCharMarker; + + if(myTarget < args->targetLimit){ + + mySourceChar= (unsigned char) *mySource++; + + switch(mySourceChar){ + case UCNV_SI: + pToU2022State->g=0; + if (myData->isEmptySegment) { + myData->isEmptySegment = false; /* we are handling it, reset to avoid future spurious errors */ + *err = U_ILLEGAL_ESCAPE_SEQUENCE; + args->converter->toUCallbackReason = UCNV_IRREGULAR; + args->converter->toUBytes[0] = static_cast(mySourceChar); + args->converter->toULength = 1; + args->target = myTarget; + args->source = mySource; + return; + } + continue; + + case UCNV_SO: + if(pToU2022State->cs[1] != 0) { + pToU2022State->g=1; + myData->isEmptySegment = true; /* Begin a new segment, empty so far */ + continue; + } else { + /* illegal to have SO before a matching designator */ + myData->isEmptySegment = false; /* Handling a different error, reset this to avoid future spurious errs */ + break; + } + + case ESC_2022: + mySource--; +escape: + { + const char * mySourceBefore = mySource; + int8_t toULengthBefore = args->converter->toULength; + + changeState_2022(args->converter,&(mySource), + mySourceLimit, ISO_2022_CN,err); + + /* After SO there must be at least one character before a designator (designator error handled separately) */ + if(myData->key==0 && U_SUCCESS(*err) && myData->isEmptySegment) { + *err = U_ILLEGAL_ESCAPE_SEQUENCE; + args->converter->toUCallbackReason = UCNV_IRREGULAR; + args->converter->toULength = (int8_t)(toULengthBefore + (mySource - mySourceBefore)); + } + } + + /* invalid or illegal escape sequence */ + if(U_FAILURE(*err)){ + args->target = myTarget; + args->source = mySource; + myData->isEmptySegment = false; /* Reset to avoid future spurious errors */ + return; + } + continue; + + /* ISO-2022-CN does not use single-byte (C1) SS2 and SS3 */ + + case CR: + case LF: + uprv_memset(pToU2022State, 0, sizeof(ISO2022State)); + U_FALLTHROUGH; + default: + /* convert one or two bytes */ + myData->isEmptySegment = false; + if(pToU2022State->g != 0) { + if(mySource < mySourceLimit) { + UConverterSharedData *cnv; + StateEnum tempState; + int32_t tempBufLen; + int leadIsOk, trailIsOk; + uint8_t trailByte; +getTrailByte: + trailByte = (uint8_t)*mySource; + /* + * Ticket 5691: consistent illegal sequences: + * - We include at least the first byte in the illegal sequence. + * - If any of the non-initial bytes could be the start of a character, + * we stop the illegal sequence before the first one of those. + * + * In ISO-2022 DBCS, if the second byte is in the 21..7e range or is + * an ESC/SO/SI, we report only the first byte as the illegal sequence. + * Otherwise we convert or report the pair of bytes. + */ + leadIsOk = (uint8_t)(mySourceChar - 0x21) <= (0x7e - 0x21); + trailIsOk = (uint8_t)(trailByte - 0x21) <= (0x7e - 0x21); + if (leadIsOk && trailIsOk) { + ++mySource; + tempState = (StateEnum)pToU2022State->cs[pToU2022State->g]; + if(tempState >= CNS_11643_0) { + cnv = myData->myConverterArray[CNS_11643]; + tempBuf[0] = (char) (0x80+(tempState-CNS_11643_0)); + tempBuf[1] = (char) (mySourceChar); + tempBuf[2] = (char) trailByte; + tempBufLen = 3; + + }else{ + U_ASSERT(tempStatemyConverterArray[tempState]; + tempBuf[0] = (char) (mySourceChar); + tempBuf[1] = (char) trailByte; + tempBufLen = 2; + } + targetUniChar = ucnv_MBCSSimpleGetNextUChar(cnv, tempBuf, tempBufLen, false); + mySourceChar = (mySourceChar << 8) | trailByte; + } else if (!(trailIsOk || IS_2022_CONTROL(trailByte))) { + /* report a pair of illegal bytes if the second byte is not a DBCS starter */ + ++mySource; + /* add another bit so that the code below writes 2 bytes in case of error */ + mySourceChar = 0x10000 | (mySourceChar << 8) | trailByte; + } + if(pToU2022State->g>=2) { + /* return from a single-shift state to the previous one */ + pToU2022State->g=pToU2022State->prevG; + } + } else { + args->converter->toUBytes[0] = (uint8_t)mySourceChar; + args->converter->toULength = 1; + goto endloop; + } + } + else{ + if(mySourceChar <= 0x7f) { + targetUniChar = (char16_t) mySourceChar; + } + } + break; + } + if(targetUniChar < (missingCharMarker-1/*0xfffe*/)){ + if(args->offsets){ + args->offsets[myTarget - args->target] = (int32_t)(mySource - args->source - (mySourceChar <= 0xff ? 1 : 2)); + } + *(myTarget++)=(char16_t)targetUniChar; + } + else if(targetUniChar > missingCharMarker){ + /* disassemble the surrogate pair and write to output*/ + targetUniChar-=0x0010000; + *myTarget = (char16_t)(0xd800+(char16_t)(targetUniChar>>10)); + if(args->offsets){ + args->offsets[myTarget - args->target] = (int32_t)(mySource - args->source - (mySourceChar <= 0xff ? 1 : 2)); + } + ++myTarget; + if(myTarget< args->targetLimit){ + *myTarget = (char16_t)(0xdc00+(char16_t)(targetUniChar&0x3ff)); + if(args->offsets){ + args->offsets[myTarget - args->target] = (int32_t)(mySource - args->source - (mySourceChar <= 0xff ? 1 : 2)); + } + ++myTarget; + }else{ + args->converter->UCharErrorBuffer[args->converter->UCharErrorBufferLength++]= + (char16_t)(0xdc00+(char16_t)(targetUniChar&0x3ff)); + } + + } + else{ + /* Call the callback function*/ + toUnicodeCallback(args->converter,mySourceChar,targetUniChar,err); + break; + } + } + else{ + *err =U_BUFFER_OVERFLOW_ERROR; + break; + } + } +endloop: + args->target = myTarget; + args->source = mySource; +} +#endif /* #if !UCONFIG_ONLY_HTML_CONVERSION */ + +static void U_CALLCONV +_ISO_2022_WriteSub(UConverterFromUnicodeArgs *args, int32_t offsetIndex, UErrorCode *err) { + UConverter *cnv = args->converter; + UConverterDataISO2022 *myConverterData=(UConverterDataISO2022 *) cnv->extraInfo; + ISO2022State *pFromU2022State=&myConverterData->fromU2022State; + char *p, *subchar; + char buffer[8]; + int32_t length; + + subchar=(char *)cnv->subChars; + length=cnv->subCharLen; /* assume length==1 for most variants */ + + p = buffer; + switch(myConverterData->locale[0]){ + case 'j': + { + int8_t cs; + + if(pFromU2022State->g == 1) { + /* JIS7: switch from G1 to G0 */ + pFromU2022State->g = 0; + *p++ = UCNV_SI; + } + + cs = pFromU2022State->cs[0]; + if(cs != ASCII && cs != JISX201) { + /* not in ASCII or JIS X 0201: switch to ASCII */ + pFromU2022State->cs[0] = (int8_t)ASCII; + *p++ = '\x1b'; + *p++ = '\x28'; + *p++ = '\x42'; + } + + *p++ = subchar[0]; + break; + } + case 'c': + if(pFromU2022State->g != 0) { + /* not in ASCII mode: switch to ASCII */ + pFromU2022State->g = 0; + *p++ = UCNV_SI; + } + *p++ = subchar[0]; + break; + case 'k': + if(myConverterData->version == 0) { + if(length == 1) { + if(args->converter->fromUnicodeStatus) { + /* in DBCS mode: switch to SBCS */ + args->converter->fromUnicodeStatus = 0; + *p++ = UCNV_SI; + } + *p++ = subchar[0]; + } else /* length == 2*/ { + if(!args->converter->fromUnicodeStatus) { + /* in SBCS mode: switch to DBCS */ + args->converter->fromUnicodeStatus = 1; + *p++ = UCNV_SO; + } + *p++ = subchar[0]; + *p++ = subchar[1]; + } + break; + } else { + /* save the subconverter's substitution string */ + uint8_t *currentSubChars = myConverterData->currentConverter->subChars; + int8_t currentSubCharLen = myConverterData->currentConverter->subCharLen; + + /* set our substitution string into the subconverter */ + myConverterData->currentConverter->subChars = (uint8_t *)subchar; + myConverterData->currentConverter->subCharLen = (int8_t)length; + + /* let the subconverter write the subchar, set/retrieve fromUChar32 state */ + args->converter = myConverterData->currentConverter; + myConverterData->currentConverter->fromUChar32 = cnv->fromUChar32; + ucnv_cbFromUWriteSub(args, 0, err); + cnv->fromUChar32 = myConverterData->currentConverter->fromUChar32; + args->converter = cnv; + + /* restore the subconverter's substitution string */ + myConverterData->currentConverter->subChars = currentSubChars; + myConverterData->currentConverter->subCharLen = currentSubCharLen; + + if(*err == U_BUFFER_OVERFLOW_ERROR) { + if(myConverterData->currentConverter->charErrorBufferLength > 0) { + uprv_memcpy( + cnv->charErrorBuffer, + myConverterData->currentConverter->charErrorBuffer, + myConverterData->currentConverter->charErrorBufferLength); + } + cnv->charErrorBufferLength = myConverterData->currentConverter->charErrorBufferLength; + myConverterData->currentConverter->charErrorBufferLength = 0; + } + return; + } + default: + /* not expected */ + break; + } + ucnv_cbFromUWriteBytes(args, + buffer, (int32_t)(p - buffer), + offsetIndex, err); +} + +/* + * Structure for cloning an ISO 2022 converter into a single memory block. + */ +struct cloneStruct +{ + UConverter cnv; + UConverter currentConverter; + UConverterDataISO2022 mydata; +}; + + +U_CDECL_BEGIN + +static UConverter * U_CALLCONV +_ISO_2022_SafeClone( + const UConverter *cnv, + void *stackBuffer, + int32_t *pBufferSize, + UErrorCode *status) +{ + struct cloneStruct * localClone; + UConverterDataISO2022 *cnvData; + int32_t i, size; + + if (U_FAILURE(*status)){ + return nullptr; + } + + if (*pBufferSize == 0) { /* 'preflighting' request - set needed size into *pBufferSize */ + *pBufferSize = (int32_t)sizeof(struct cloneStruct); + return nullptr; + } + + cnvData = (UConverterDataISO2022 *)cnv->extraInfo; + localClone = (struct cloneStruct *)stackBuffer; + + /* ucnv.c/ucnv_safeClone() copied the main UConverter already */ + + uprv_memcpy(&localClone->mydata, cnvData, sizeof(UConverterDataISO2022)); + localClone->cnv.extraInfo = &localClone->mydata; /* set pointer to extra data */ + localClone->cnv.isExtraLocal = true; + + /* share the subconverters */ + + if(cnvData->currentConverter != nullptr) { + size = (int32_t)sizeof(UConverter); + localClone->mydata.currentConverter = + ucnv_safeClone(cnvData->currentConverter, + &localClone->currentConverter, + &size, status); + if(U_FAILURE(*status)) { + return nullptr; + } + } + + for(i=0; imyConverterArray[i] != nullptr) { + ucnv_incrementRefCount(cnvData->myConverterArray[i]); + } + } + + return &localClone->cnv; +} + +U_CDECL_END + +static void U_CALLCONV +_ISO_2022_GetUnicodeSet(const UConverter *cnv, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode) +{ + int32_t i; + UConverterDataISO2022* cnvData; + + if (U_FAILURE(*pErrorCode)) { + return; + } +#ifdef U_ENABLE_GENERIC_ISO_2022 + if (cnv->sharedData == &_ISO2022Data) { + /* We use UTF-8 in this case */ + sa->addRange(sa->set, 0, 0xd7FF); + sa->addRange(sa->set, 0xE000, 0x10FFFF); + return; + } +#endif + + cnvData = (UConverterDataISO2022*)cnv->extraInfo; + + /* open a set and initialize it with code points that are algorithmically round-tripped */ + switch(cnvData->locale[0]){ + case 'j': + /* include JIS X 0201 which is hardcoded */ + sa->add(sa->set, 0xa5); + sa->add(sa->set, 0x203e); + if(jpCharsetMasks[cnvData->version]&CSM(ISO8859_1)) { + /* include Latin-1 for some variants of JP */ + sa->addRange(sa->set, 0, 0xff); + } else { + /* include ASCII for JP */ + sa->addRange(sa->set, 0, 0x7f); + } + if(cnvData->version==3 || cnvData->version==4 || which==UCNV_ROUNDTRIP_AND_FALLBACK_SET) { + /* + * Do not test (jpCharsetMasks[cnvData->version]&CSM(HWKANA_7BIT))!=0 + * because the bit is on for all JP versions although only versions 3 & 4 (JIS7 & JIS8) + * use half-width Katakana. + * This is because all ISO-2022-JP variants are lenient in that they accept (in toUnicode) + * half-width Katakana via the ESC ( I sequence. + * However, we only emit (fromUnicode) half-width Katakana according to the + * definition of each variant. + * + * When including fallbacks, + * we need to include half-width Katakana Unicode code points for all JP variants because + * JIS X 0208 has hardcoded fallbacks for them (which map to full-width Katakana). + */ + /* include half-width Katakana for JP */ + sa->addRange(sa->set, HWKANA_START, HWKANA_END); + } + break; +#if !UCONFIG_ONLY_HTML_CONVERSION + case 'c': + case 'z': + /* include ASCII for CN */ + sa->addRange(sa->set, 0, 0x7f); + break; + case 'k': + /* there is only one converter for KR, and it is not in the myConverterArray[] */ + cnvData->currentConverter->sharedData->impl->getUnicodeSet( + cnvData->currentConverter, sa, which, pErrorCode); + /* the loop over myConverterArray[] will simply not find another converter */ + break; +#endif + default: + break; + } + +#if 0 /* Replaced by ucnv_MBCSGetFilteredUnicodeSetForUnicode() until we implement ucnv_getUnicodeSet() with reverse fallbacks. */ + if( (cnvData->locale[0]=='c' || cnvData->locale[0]=='z') && + cnvData->version==0 && i==CNS_11643 + ) { + /* special handling for non-EXT ISO-2022-CN: add only code points for CNS planes 1 and 2 */ + ucnv_MBCSGetUnicodeSetForBytes( + cnvData->myConverterArray[i], + sa, UCNV_ROUNDTRIP_SET, + 0, 0x81, 0x82, + pErrorCode); + } +#endif + + for (i=0; imyConverterArray[i]!=nullptr) { + if(cnvData->locale[0]=='j' && i==JISX208) { + /* + * Only add code points that map to Shift-JIS codes + * corresponding to JIS X 0208. + */ + filter=UCNV_SET_FILTER_SJIS; +#if !UCONFIG_ONLY_HTML_CONVERSION + } else if( (cnvData->locale[0]=='c' || cnvData->locale[0]=='z') && + cnvData->version==0 && i==CNS_11643) { + /* + * Version-specific for CN: + * CN version 0 does not map CNS planes 3..7 although + * they are all available in the CNS conversion table; + * CN version 1 (-EXT) does map them all. + * The two versions create different Unicode sets. + */ + filter=UCNV_SET_FILTER_2022_CN; + } else if(i==KSC5601) { + /* + * Some of the KSC 5601 tables (convrtrs.txt has this aliases on multiple tables) + * are broader than GR94. + */ + filter=UCNV_SET_FILTER_GR94DBCS; +#endif + } else { + filter=UCNV_SET_FILTER_NONE; + } + ucnv_MBCSGetFilteredUnicodeSetForUnicode(cnvData->myConverterArray[i], sa, which, filter, pErrorCode); + } + } + + /* + * ISO 2022 converters must not convert SO/SI/ESC despite what + * sub-converters do by themselves. + * Remove these characters from the set. + */ + sa->remove(sa->set, 0x0e); + sa->remove(sa->set, 0x0f); + sa->remove(sa->set, 0x1b); + + /* ISO 2022 converters do not convert C1 controls either */ + sa->removeRange(sa->set, 0x80, 0x9f); +} + +static const UConverterImpl _ISO2022Impl={ + UCNV_ISO_2022, + + nullptr, + nullptr, + + _ISO2022Open, + _ISO2022Close, + _ISO2022Reset, + +#ifdef U_ENABLE_GENERIC_ISO_2022 + T_UConverter_toUnicode_ISO_2022_OFFSETS_LOGIC, + T_UConverter_toUnicode_ISO_2022_OFFSETS_LOGIC, + ucnv_fromUnicode_UTF8, + ucnv_fromUnicode_UTF8_OFFSETS_LOGIC, +#else + nullptr, + nullptr, + nullptr, + nullptr, +#endif + nullptr, + + nullptr, + _ISO2022getName, + _ISO_2022_WriteSub, + _ISO_2022_SafeClone, + _ISO_2022_GetUnicodeSet, + + nullptr, + nullptr +}; +static const UConverterStaticData _ISO2022StaticData={ + sizeof(UConverterStaticData), + "ISO_2022", + 2022, + UCNV_IBM, + UCNV_ISO_2022, + 1, + 3, /* max 3 bytes per char16_t from UTF-8 (4 bytes from surrogate _pair_) */ + { 0x1a, 0, 0, 0 }, + 1, + false, + false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; +const UConverterSharedData _ISO2022Data= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_ISO2022StaticData, &_ISO2022Impl); + +/*************JP****************/ +static const UConverterImpl _ISO2022JPImpl={ + UCNV_ISO_2022, + + nullptr, + nullptr, + + _ISO2022Open, + _ISO2022Close, + _ISO2022Reset, + + UConverter_toUnicode_ISO_2022_JP_OFFSETS_LOGIC, + UConverter_toUnicode_ISO_2022_JP_OFFSETS_LOGIC, + UConverter_fromUnicode_ISO_2022_JP_OFFSETS_LOGIC, + UConverter_fromUnicode_ISO_2022_JP_OFFSETS_LOGIC, + nullptr, + + nullptr, + _ISO2022getName, + _ISO_2022_WriteSub, + _ISO_2022_SafeClone, + _ISO_2022_GetUnicodeSet, + + nullptr, + nullptr +}; +static const UConverterStaticData _ISO2022JPStaticData={ + sizeof(UConverterStaticData), + "ISO_2022_JP", + 0, + UCNV_IBM, + UCNV_ISO_2022, + 1, + 6, /* max 6 bytes per char16_t: 4-byte escape sequence + DBCS */ + { 0x1a, 0, 0, 0 }, + 1, + false, + false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + +namespace { + +const UConverterSharedData _ISO2022JPData= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_ISO2022JPStaticData, &_ISO2022JPImpl); + +} // namespace + +#if !UCONFIG_ONLY_HTML_CONVERSION +/************* KR ***************/ +static const UConverterImpl _ISO2022KRImpl={ + UCNV_ISO_2022, + + nullptr, + nullptr, + + _ISO2022Open, + _ISO2022Close, + _ISO2022Reset, + + UConverter_toUnicode_ISO_2022_KR_OFFSETS_LOGIC, + UConverter_toUnicode_ISO_2022_KR_OFFSETS_LOGIC, + UConverter_fromUnicode_ISO_2022_KR_OFFSETS_LOGIC, + UConverter_fromUnicode_ISO_2022_KR_OFFSETS_LOGIC, + nullptr, + + nullptr, + _ISO2022getName, + _ISO_2022_WriteSub, + _ISO_2022_SafeClone, + _ISO_2022_GetUnicodeSet, + + nullptr, + nullptr +}; +static const UConverterStaticData _ISO2022KRStaticData={ + sizeof(UConverterStaticData), + "ISO_2022_KR", + 0, + UCNV_IBM, + UCNV_ISO_2022, + 1, + 8, /* max 8 bytes per char16_t */ + { 0x1a, 0, 0, 0 }, + 1, + false, + false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + +namespace { + +const UConverterSharedData _ISO2022KRData= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_ISO2022KRStaticData, &_ISO2022KRImpl); + +} // namespace + +/*************** CN ***************/ +static const UConverterImpl _ISO2022CNImpl={ + + UCNV_ISO_2022, + + nullptr, + nullptr, + + _ISO2022Open, + _ISO2022Close, + _ISO2022Reset, + + UConverter_toUnicode_ISO_2022_CN_OFFSETS_LOGIC, + UConverter_toUnicode_ISO_2022_CN_OFFSETS_LOGIC, + UConverter_fromUnicode_ISO_2022_CN_OFFSETS_LOGIC, + UConverter_fromUnicode_ISO_2022_CN_OFFSETS_LOGIC, + nullptr, + + nullptr, + _ISO2022getName, + _ISO_2022_WriteSub, + _ISO_2022_SafeClone, + _ISO_2022_GetUnicodeSet, + + nullptr, + nullptr +}; +static const UConverterStaticData _ISO2022CNStaticData={ + sizeof(UConverterStaticData), + "ISO_2022_CN", + 0, + UCNV_IBM, + UCNV_ISO_2022, + 1, + 8, /* max 8 bytes per char16_t: 4-byte CNS designator + 2 bytes for SS2/SS3 + DBCS */ + { 0x1a, 0, 0, 0 }, + 1, + false, + false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + +namespace { + +const UConverterSharedData _ISO2022CNData= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_ISO2022CNStaticData, &_ISO2022CNImpl); + +} // namespace +#endif /* #if !UCONFIG_ONLY_HTML_CONVERSION */ + +#endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */ diff --git a/intl/icu/source/common/ucnv_bld.cpp b/intl/icu/source/common/ucnv_bld.cpp new file mode 100644 index 0000000000..564b645bed --- /dev/null +++ b/intl/icu/source/common/ucnv_bld.cpp @@ -0,0 +1,1689 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ******************************************************************** + * COPYRIGHT: + * Copyright (c) 1996-2016, International Business Machines Corporation and + * others. All Rights Reserved. + ******************************************************************** + * + * ucnv_bld.cpp: + * + * Defines functions that are used in the creation/initialization/deletion + * of converters and related structures. + * uses uconv_io.h routines to access disk information + * is used by ucnv.h to implement public API create/delete/flushCache routines + * Modification History: + * + * Date Name Description + * + * 06/20/2000 helena OS/400 port changes; mostly typecast. + * 06/29/2000 helena Major rewrite of the callback interface. +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/putil.h" +#include "unicode/udata.h" +#include "unicode/ucnv.h" +#include "unicode/uloc.h" +#include "mutex.h" +#include "putilimp.h" +#include "uassert.h" +#include "utracimp.h" +#include "ucnv_io.h" +#include "ucnv_bld.h" +#include "ucnvmbcs.h" +#include "ucnv_ext.h" +#include "ucnv_cnv.h" +#include "ucnv_imp.h" +#include "uhash.h" +#include "umutex.h" +#include "cstring.h" +#include "cmemory.h" +#include "ucln_cmn.h" +#include "ustr_cnv.h" + + +#if 0 +#include +extern void UCNV_DEBUG_LOG(char *what, char *who, void *p, int l); +#define UCNV_DEBUG_LOG(x,y,z) UCNV_DEBUG_LOG(x,y,z,__LINE__) +#else +# define UCNV_DEBUG_LOG(x,y,z) +#endif + +static const UConverterSharedData * const +converterData[UCNV_NUMBER_OF_SUPPORTED_CONVERTER_TYPES]={ + nullptr, nullptr, + +#if UCONFIG_NO_LEGACY_CONVERSION + nullptr, +#else + &_MBCSData, +#endif + + &_Latin1Data, + &_UTF8Data, &_UTF16BEData, &_UTF16LEData, +#if UCONFIG_ONLY_HTML_CONVERSION + nullptr, nullptr, +#else + &_UTF32BEData, &_UTF32LEData, +#endif + nullptr, + +#if UCONFIG_NO_LEGACY_CONVERSION + nullptr, +#else + &_ISO2022Data, +#endif + +#if UCONFIG_NO_LEGACY_CONVERSION || UCONFIG_ONLY_HTML_CONVERSION + nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, + nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, + nullptr, +#else + &_LMBCSData1,&_LMBCSData2, &_LMBCSData3, &_LMBCSData4, &_LMBCSData5, &_LMBCSData6, + &_LMBCSData8,&_LMBCSData11,&_LMBCSData16,&_LMBCSData17,&_LMBCSData18,&_LMBCSData19, + &_HZData, +#endif + +#if UCONFIG_ONLY_HTML_CONVERSION + nullptr, +#else + &_SCSUData, +#endif + + +#if UCONFIG_NO_LEGACY_CONVERSION || UCONFIG_ONLY_HTML_CONVERSION + nullptr, +#else + &_ISCIIData, +#endif + + &_ASCIIData, +#if UCONFIG_ONLY_HTML_CONVERSION + nullptr, nullptr, &_UTF16Data, nullptr, nullptr, nullptr, +#else + &_UTF7Data, &_Bocu1Data, &_UTF16Data, &_UTF32Data, &_CESU8Data, &_IMAPData, +#endif + +#if UCONFIG_NO_LEGACY_CONVERSION || UCONFIG_ONLY_HTML_CONVERSION + nullptr, +#else + &_CompoundTextData +#endif +}; + +/* Please keep this in binary sorted order for getAlgorithmicTypeFromName. + Also the name should be in lower case and all spaces, dashes and underscores + removed +*/ +static struct { + const char *name; + const UConverterType type; +} const cnvNameType[] = { +#if !UCONFIG_ONLY_HTML_CONVERSION + { "bocu1", UCNV_BOCU1 }, + { "cesu8", UCNV_CESU8 }, +#endif +#if !UCONFIG_NO_LEGACY_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION + { "hz",UCNV_HZ }, +#endif +#if !UCONFIG_ONLY_HTML_CONVERSION + { "imapmailboxname", UCNV_IMAP_MAILBOX }, +#endif +#if !UCONFIG_NO_LEGACY_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION + { "iscii", UCNV_ISCII }, +#endif +#if !UCONFIG_NO_LEGACY_CONVERSION + { "iso2022", UCNV_ISO_2022 }, +#endif + { "iso88591", UCNV_LATIN_1 }, +#if !UCONFIG_NO_LEGACY_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION + { "lmbcs1", UCNV_LMBCS_1 }, + { "lmbcs11",UCNV_LMBCS_11 }, + { "lmbcs16",UCNV_LMBCS_16 }, + { "lmbcs17",UCNV_LMBCS_17 }, + { "lmbcs18",UCNV_LMBCS_18 }, + { "lmbcs19",UCNV_LMBCS_19 }, + { "lmbcs2", UCNV_LMBCS_2 }, + { "lmbcs3", UCNV_LMBCS_3 }, + { "lmbcs4", UCNV_LMBCS_4 }, + { "lmbcs5", UCNV_LMBCS_5 }, + { "lmbcs6", UCNV_LMBCS_6 }, + { "lmbcs8", UCNV_LMBCS_8 }, +#endif +#if !UCONFIG_ONLY_HTML_CONVERSION + { "scsu", UCNV_SCSU }, +#endif + { "usascii", UCNV_US_ASCII }, + { "utf16", UCNV_UTF16 }, + { "utf16be", UCNV_UTF16_BigEndian }, + { "utf16le", UCNV_UTF16_LittleEndian }, +#if U_IS_BIG_ENDIAN + { "utf16oppositeendian", UCNV_UTF16_LittleEndian }, + { "utf16platformendian", UCNV_UTF16_BigEndian }, +#else + { "utf16oppositeendian", UCNV_UTF16_BigEndian}, + { "utf16platformendian", UCNV_UTF16_LittleEndian }, +#endif +#if !UCONFIG_ONLY_HTML_CONVERSION + { "utf32", UCNV_UTF32 }, + { "utf32be", UCNV_UTF32_BigEndian }, + { "utf32le", UCNV_UTF32_LittleEndian }, +#if U_IS_BIG_ENDIAN + { "utf32oppositeendian", UCNV_UTF32_LittleEndian }, + { "utf32platformendian", UCNV_UTF32_BigEndian }, +#else + { "utf32oppositeendian", UCNV_UTF32_BigEndian }, + { "utf32platformendian", UCNV_UTF32_LittleEndian }, +#endif +#endif +#if !UCONFIG_ONLY_HTML_CONVERSION + { "utf7", UCNV_UTF7 }, +#endif + { "utf8", UCNV_UTF8 }, +#if !UCONFIG_ONLY_HTML_CONVERSION + { "x11compoundtext", UCNV_COMPOUND_TEXT} +#endif +}; + + +/*initializes some global variables */ +static UHashtable *SHARED_DATA_HASHTABLE = nullptr; +static icu::UMutex cnvCacheMutex; +/* Note: the global mutex is used for */ +/* reference count updates. */ + +static const char **gAvailableConverters = nullptr; +static uint16_t gAvailableConverterCount = 0; +static icu::UInitOnce gAvailableConvertersInitOnce {}; + +#if !U_CHARSET_IS_UTF8 + +/* This contains the resolved converter name. So no further alias lookup is needed again. */ +static char gDefaultConverterNameBuffer[UCNV_MAX_CONVERTER_NAME_LENGTH + 1]; /* +1 for nullptr */ +static const char *gDefaultConverterName = nullptr; + +/* +If the default converter is an algorithmic converter, this is the cached value. +We don't cache a full UConverter and clone it because ucnv_clone doesn't have +less overhead than an algorithmic open. We don't cache non-algorithmic converters +because ucnv_flushCache must be able to unload the default converter and its table. +*/ +static const UConverterSharedData *gDefaultAlgorithmicSharedData = nullptr; + +/* Does gDefaultConverterName have a converter option and require extra parsing? */ +static UBool gDefaultConverterContainsOption; + +#endif /* !U_CHARSET_IS_UTF8 */ + +static const char DATA_TYPE[] = "cnv"; + +/* ucnv_flushAvailableConverterCache. This is only called from ucnv_cleanup(). + * If it is ever to be called from elsewhere, synchronization + * will need to be considered. + */ +static void +ucnv_flushAvailableConverterCache() { + gAvailableConverterCount = 0; + if (gAvailableConverters) { + uprv_free((char **)gAvailableConverters); + gAvailableConverters = nullptr; + } + gAvailableConvertersInitOnce.reset(); +} + +/* ucnv_cleanup - delete all storage held by the converter cache, except any */ +/* in use by open converters. */ +/* Not thread safe. */ +/* Not supported API. */ +static UBool U_CALLCONV ucnv_cleanup() { + ucnv_flushCache(); + if (SHARED_DATA_HASHTABLE != nullptr && uhash_count(SHARED_DATA_HASHTABLE) == 0) { + uhash_close(SHARED_DATA_HASHTABLE); + SHARED_DATA_HASHTABLE = nullptr; + } + + /* Isn't called from flushCache because other threads may have preexisting references to the table. */ + ucnv_flushAvailableConverterCache(); + +#if !U_CHARSET_IS_UTF8 + gDefaultConverterName = nullptr; + gDefaultConverterNameBuffer[0] = 0; + gDefaultConverterContainsOption = false; + gDefaultAlgorithmicSharedData = nullptr; +#endif + + return (SHARED_DATA_HASHTABLE == nullptr); +} + +U_CAPI void U_EXPORT2 +ucnv_enableCleanup() { + ucln_common_registerCleanup(UCLN_COMMON_UCNV, ucnv_cleanup); +} + +static UBool U_CALLCONV +isCnvAcceptable(void * /*context*/, + const char * /*type*/, const char * /*name*/, + const UDataInfo *pInfo) { + return (UBool)( + pInfo->size>=20 && + pInfo->isBigEndian==U_IS_BIG_ENDIAN && + pInfo->charsetFamily==U_CHARSET_FAMILY && + pInfo->sizeofUChar==U_SIZEOF_UCHAR && + pInfo->dataFormat[0]==0x63 && /* dataFormat="cnvt" */ + pInfo->dataFormat[1]==0x6e && + pInfo->dataFormat[2]==0x76 && + pInfo->dataFormat[3]==0x74 && + pInfo->formatVersion[0]==6); /* Everything will be version 6 */ +} + +/** + * Un flatten shared data from a UDATA.. + */ +static UConverterSharedData* +ucnv_data_unFlattenClone(UConverterLoadArgs *pArgs, UDataMemory *pData, UErrorCode *status) +{ + /* UDataInfo info; -- necessary only if some converters have different formatVersion */ + const uint8_t *raw = (const uint8_t *)udata_getMemory(pData); + const UConverterStaticData *source = (const UConverterStaticData *) raw; + UConverterSharedData *data; + UConverterType type = (UConverterType)source->conversionType; + + if(U_FAILURE(*status)) + return nullptr; + + if( (uint16_t)type >= UCNV_NUMBER_OF_SUPPORTED_CONVERTER_TYPES || + converterData[type] == nullptr || + !converterData[type]->isReferenceCounted || + converterData[type]->referenceCounter != 1 || + source->structSize != sizeof(UConverterStaticData)) + { + *status = U_INVALID_TABLE_FORMAT; + return nullptr; + } + + data = (UConverterSharedData *)uprv_malloc(sizeof(UConverterSharedData)); + if(data == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + + /* copy initial values from the static structure for this type */ + uprv_memcpy(data, converterData[type], sizeof(UConverterSharedData)); + + data->staticData = source; + + data->sharedDataCached = false; + + /* fill in fields from the loaded data */ + data->dataMemory = (void*)pData; /* for future use */ + + if(data->impl->load != nullptr) { + data->impl->load(data, pArgs, raw + source->structSize, status); + if(U_FAILURE(*status)) { + uprv_free(data); + return nullptr; + } + } + return data; +} + +/*Takes an alias name gets an actual converter file name + *goes to disk and opens it. + *allocates the memory and returns a new UConverter object + */ +static UConverterSharedData *createConverterFromFile(UConverterLoadArgs *pArgs, UErrorCode * err) +{ + UDataMemory *data; + UConverterSharedData *sharedData; + + UTRACE_ENTRY_OC(UTRACE_UCNV_LOAD); + + if (U_FAILURE (*err)) { + UTRACE_EXIT_STATUS(*err); + return nullptr; + } + + UTRACE_DATA2(UTRACE_OPEN_CLOSE, "load converter %s from package %s", pArgs->name, pArgs->pkg); + + data = udata_openChoice(pArgs->pkg, DATA_TYPE, pArgs->name, isCnvAcceptable, nullptr, err); + if(U_FAILURE(*err)) + { + UTRACE_EXIT_STATUS(*err); + return nullptr; + } + + sharedData = ucnv_data_unFlattenClone(pArgs, data, err); + if(U_FAILURE(*err)) + { + udata_close(data); + UTRACE_EXIT_STATUS(*err); + return nullptr; + } + + /* + * TODO Store pkg in a field in the shared data so that delta-only converters + * can load base converters from the same package. + * If the pkg name is longer than the field, then either do not load the converter + * in the first place, or just set the pkg field to "". + */ + + UTRACE_EXIT_PTR_STATUS(sharedData, *err); + return sharedData; +} + +/*returns a converter type from a string + */ +static const UConverterSharedData * +getAlgorithmicTypeFromName(const char *realName) +{ + uint32_t mid, start, limit; + uint32_t lastMid; + int result; + char strippedName[UCNV_MAX_CONVERTER_NAME_LENGTH]; + + /* Lower case and remove ignoreable characters. */ + ucnv_io_stripForCompare(strippedName, realName); + + /* do a binary search for the alias */ + start = 0; + limit = UPRV_LENGTHOF(cnvNameType); + mid = limit; + lastMid = UINT32_MAX; + + for (;;) { + mid = (uint32_t)((start + limit) / 2); + if (lastMid == mid) { /* Have we moved? */ + break; /* We haven't moved, and it wasn't found. */ + } + lastMid = mid; + result = uprv_strcmp(strippedName, cnvNameType[mid].name); + + if (result < 0) { + limit = mid; + } else if (result > 0) { + start = mid; + } else { + return converterData[cnvNameType[mid].type]; + } + } + + return nullptr; +} + +/* +* Based on the number of known converters, this determines how many times larger +* the shared data hash table should be. When on small platforms, or just a couple +* of converters are used, this number should be 2. When memory is plentiful, or +* when ucnv_countAvailable is ever used with a lot of available converters, +* this should be 4. +* Larger numbers reduce the number of hash collisions, but use more memory. +*/ +#define UCNV_CACHE_LOAD_FACTOR 2 + +/* Puts the shared data in the static hashtable SHARED_DATA_HASHTABLE */ +/* Will always be called with the cnvCacheMutex already being held */ +/* by the calling function. */ +/* Stores the shared data in the SHARED_DATA_HASHTABLE + * @param data The shared data + */ +static void +ucnv_shareConverterData(UConverterSharedData * data) +{ + UErrorCode err = U_ZERO_ERROR; + /*Lazy evaluates the Hashtable itself */ + /*void *sanity = nullptr;*/ + + if (SHARED_DATA_HASHTABLE == nullptr) + { + SHARED_DATA_HASHTABLE = uhash_openSize(uhash_hashChars, uhash_compareChars, nullptr, + ucnv_io_countKnownConverters(&err)*UCNV_CACHE_LOAD_FACTOR, + &err); + ucnv_enableCleanup(); + + if (U_FAILURE(err)) + return; + } + + /* ### check to see if the element is not already there! */ + + /* + sanity = ucnv_getSharedConverterData (data->staticData->name); + if(sanity != nullptr) + { + UCNV_DEBUG_LOG("put:overwrite!",data->staticData->name,sanity); + } + UCNV_DEBUG_LOG("put:chk",data->staticData->name,sanity); + */ + + /* Mark it shared */ + data->sharedDataCached = true; + + uhash_put(SHARED_DATA_HASHTABLE, + (void*) data->staticData->name, /* Okay to cast away const as long as + keyDeleter == nullptr */ + data, + &err); + UCNV_DEBUG_LOG("put", data->staticData->name,data); + +} + +/* Look up a converter name in the shared data cache. */ +/* cnvCacheMutex must be held by the caller to protect the hash table. */ +/* gets the shared data from the SHARED_DATA_HASHTABLE (might return nullptr if it isn't there) + * @param name The name of the shared data + * @return the shared data from the SHARED_DATA_HASHTABLE + */ +static UConverterSharedData * +ucnv_getSharedConverterData(const char *name) +{ + /*special case when no Table has yet been created we return nullptr */ + if (SHARED_DATA_HASHTABLE == nullptr) + { + return nullptr; + } + else + { + UConverterSharedData *rc; + + rc = (UConverterSharedData*)uhash_get(SHARED_DATA_HASHTABLE, name); + UCNV_DEBUG_LOG("get",name,rc); + return rc; + } +} + +/*frees the string of memory blocks associates with a sharedConverter + *if and only if the referenceCounter == 0 + */ +/* Deletes (frees) the Shared data it's passed. first it checks the referenceCounter to + * see if anyone is using it, if not it frees all the memory stemming from sharedConverterData and + * returns true, + * otherwise returns false + * @param sharedConverterData The shared data + * @return if not it frees all the memory stemming from sharedConverterData and + * returns true, otherwise returns false + */ +static UBool +ucnv_deleteSharedConverterData(UConverterSharedData * deadSharedData) +{ + UTRACE_ENTRY_OC(UTRACE_UCNV_UNLOAD); + UTRACE_DATA2(UTRACE_OPEN_CLOSE, "unload converter %s shared data %p", deadSharedData->staticData->name, deadSharedData); + + if (deadSharedData->referenceCounter > 0) { + UTRACE_EXIT_VALUE((int32_t)false); + return false; + } + + if (deadSharedData->impl->unload != nullptr) { + deadSharedData->impl->unload(deadSharedData); + } + + if(deadSharedData->dataMemory != nullptr) + { + UDataMemory *data = (UDataMemory*)deadSharedData->dataMemory; + udata_close(data); + } + + uprv_free(deadSharedData); + + UTRACE_EXIT_VALUE((int32_t)true); + return true; +} + +/** + * Load a non-algorithmic converter. + * If pkg==nullptr, then this function must be called inside umtx_lock(&cnvCacheMutex). + */ +UConverterSharedData * +ucnv_load(UConverterLoadArgs *pArgs, UErrorCode *err) { + UConverterSharedData *mySharedConverterData; + + if(err == nullptr || U_FAILURE(*err)) { + return nullptr; + } + + if(pArgs->pkg != nullptr && *pArgs->pkg != 0) { + /* application-provided converters are not currently cached */ + return createConverterFromFile(pArgs, err); + } + + mySharedConverterData = ucnv_getSharedConverterData(pArgs->name); + if (mySharedConverterData == nullptr) + { + /*Not cached, we need to stream it in from file */ + mySharedConverterData = createConverterFromFile(pArgs, err); + if (U_FAILURE (*err) || (mySharedConverterData == nullptr)) + { + return nullptr; + } + else if (!pArgs->onlyTestIsLoadable) + { + /* share it with other library clients */ + ucnv_shareConverterData(mySharedConverterData); + } + } + else + { + /* The data for this converter was already in the cache. */ + /* Update the reference counter on the shared data: one more client */ + mySharedConverterData->referenceCounter++; + } + + return mySharedConverterData; +} + +/** + * Unload a non-algorithmic converter. + * It must be sharedData->isReferenceCounted + * and this function must be called inside umtx_lock(&cnvCacheMutex). + */ +U_CAPI void +ucnv_unload(UConverterSharedData *sharedData) { + if(sharedData != nullptr) { + if (sharedData->referenceCounter > 0) { + sharedData->referenceCounter--; + } + + if((sharedData->referenceCounter <= 0)&&(sharedData->sharedDataCached == false)) { + ucnv_deleteSharedConverterData(sharedData); + } + } +} + +U_CFUNC void +ucnv_unloadSharedDataIfReady(UConverterSharedData *sharedData) +{ + if(sharedData != nullptr && sharedData->isReferenceCounted) { + umtx_lock(&cnvCacheMutex); + ucnv_unload(sharedData); + umtx_unlock(&cnvCacheMutex); + } +} + +U_CFUNC void +ucnv_incrementRefCount(UConverterSharedData *sharedData) +{ + if(sharedData != nullptr && sharedData->isReferenceCounted) { + umtx_lock(&cnvCacheMutex); + sharedData->referenceCounter++; + umtx_unlock(&cnvCacheMutex); + } +} + +/* + * *pPieces must be initialized. + * The name without options will be copied to pPieces->cnvName. + * The locale and options will be copied to pPieces only if present in inName, + * otherwise the existing values in pPieces remain. + * *pArgs will be set to the pPieces values. + */ +static void +parseConverterOptions(const char *inName, + UConverterNamePieces *pPieces, + UConverterLoadArgs *pArgs, + UErrorCode *err) +{ + char *cnvName = pPieces->cnvName; + char c; + int32_t len = 0; + + pArgs->name=inName; + pArgs->locale=pPieces->locale; + pArgs->options=pPieces->options; + + /* copy the converter name itself to cnvName */ + while((c=*inName)!=0 && c!=UCNV_OPTION_SEP_CHAR) { + if (++len>=UCNV_MAX_CONVERTER_NAME_LENGTH) { + *err = U_ILLEGAL_ARGUMENT_ERROR; /* bad name */ + pPieces->cnvName[0]=0; + return; + } + *cnvName++=c; + inName++; + } + *cnvName=0; + pArgs->name=pPieces->cnvName; + + /* parse options. No more name copying should occur. */ + while((c=*inName)!=0) { + if(c==UCNV_OPTION_SEP_CHAR) { + ++inName; + } + + /* inName is behind an option separator */ + if(uprv_strncmp(inName, "locale=", 7)==0) { + /* do not modify locale itself in case we have multiple locale options */ + char *dest=pPieces->locale; + + /* copy the locale option value */ + inName+=7; + len=0; + while((c=*inName)!=0 && c!=UCNV_OPTION_SEP_CHAR) { + ++inName; + + if(++len>=ULOC_FULLNAME_CAPACITY) { + *err=U_ILLEGAL_ARGUMENT_ERROR; /* bad name */ + pPieces->locale[0]=0; + return; + } + + *dest++=c; + } + *dest=0; + } else if(uprv_strncmp(inName, "version=", 8)==0) { + /* copy the version option value into bits 3..0 of pPieces->options */ + inName+=8; + c=*inName; + if(c==0) { + pArgs->options=(pPieces->options&=~UCNV_OPTION_VERSION); + return; + } else if((uint8_t)(c-'0')<10) { + pArgs->options=pPieces->options=(pPieces->options&~UCNV_OPTION_VERSION)|(uint32_t)(c-'0'); + ++inName; + } + } else if(uprv_strncmp(inName, "swaplfnl", 8)==0) { + inName+=8; + pArgs->options=(pPieces->options|=UCNV_OPTION_SWAP_LFNL); + /* add processing for new options here with another } else if(uprv_strncmp(inName, "option-name=", XX)==0) { */ + } else { + /* ignore any other options until we define some */ + while(((c = *inName++) != 0) && (c != UCNV_OPTION_SEP_CHAR)) { + } + if(c==0) { + return; + } + } + } +} + +/*Logic determines if the converter is Algorithmic AND/OR cached + *depending on that: + * -we either go to get data from disk and cache it (Data=true, Cached=false) + * -Get it from a Hashtable (Data=X, Cached=true) + * -Call dataConverter initializer (Data=true, Cached=true) + * -Call AlgorithmicConverter initializer (Data=false, Cached=true) + */ +U_CFUNC UConverterSharedData * +ucnv_loadSharedData(const char *converterName, + UConverterNamePieces *pPieces, + UConverterLoadArgs *pArgs, + UErrorCode * err) { + UConverterNamePieces stackPieces; + UConverterLoadArgs stackArgs; + UConverterSharedData *mySharedConverterData = nullptr; + UErrorCode internalErrorCode = U_ZERO_ERROR; + UBool mayContainOption = true; + UBool checkForAlgorithmic = true; + + if (U_FAILURE (*err)) { + return nullptr; + } + + if(pPieces == nullptr) { + if(pArgs != nullptr) { + /* + * Bad: We may set pArgs pointers to stackPieces fields + * which will be invalid after this function returns. + */ + *err = U_INTERNAL_PROGRAM_ERROR; + return nullptr; + } + pPieces = &stackPieces; + } + if(pArgs == nullptr) { + uprv_memset(&stackArgs, 0, sizeof(stackArgs)); + stackArgs.size = (int32_t)sizeof(stackArgs); + pArgs = &stackArgs; + } + + pPieces->cnvName[0] = 0; + pPieces->locale[0] = 0; + pPieces->options = 0; + + pArgs->name = converterName; + pArgs->locale = pPieces->locale; + pArgs->options = pPieces->options; + + /* In case "name" is nullptr we want to open the default converter. */ + if (converterName == nullptr) { +#if U_CHARSET_IS_UTF8 + pArgs->name = "UTF-8"; + return (UConverterSharedData *)converterData[UCNV_UTF8]; +#else + /* Call ucnv_getDefaultName first to query the name from the OS. */ + pArgs->name = ucnv_getDefaultName(); + if (pArgs->name == nullptr) { + *err = U_MISSING_RESOURCE_ERROR; + return nullptr; + } + mySharedConverterData = (UConverterSharedData *)gDefaultAlgorithmicSharedData; + checkForAlgorithmic = false; + mayContainOption = gDefaultConverterContainsOption; + /* the default converter name is already canonical */ +#endif + } + else if(UCNV_FAST_IS_UTF8(converterName)) { + /* fastpath for UTF-8 */ + pArgs->name = "UTF-8"; + return (UConverterSharedData *)converterData[UCNV_UTF8]; + } + else { + /* separate the converter name from the options */ + parseConverterOptions(converterName, pPieces, pArgs, err); + if (U_FAILURE(*err)) { + /* Very bad name used. */ + return nullptr; + } + + /* get the canonical converter name */ + pArgs->name = ucnv_io_getConverterName(pArgs->name, &mayContainOption, &internalErrorCode); + if (U_FAILURE(internalErrorCode) || pArgs->name == nullptr) { + /* + * set the input name in case the converter was added + * without updating the alias table, or when there is no alias table + */ + pArgs->name = pPieces->cnvName; + } else if (internalErrorCode == U_AMBIGUOUS_ALIAS_WARNING) { + *err = U_AMBIGUOUS_ALIAS_WARNING; + } + } + + /* separate the converter name from the options */ + if(mayContainOption && pArgs->name != pPieces->cnvName) { + parseConverterOptions(pArgs->name, pPieces, pArgs, err); + } + + /* get the shared data for an algorithmic converter, if it is one */ + if (checkForAlgorithmic) { + mySharedConverterData = (UConverterSharedData *)getAlgorithmicTypeFromName(pArgs->name); + } + if (mySharedConverterData == nullptr) + { + /* it is a data-based converter, get its shared data. */ + /* Hold the cnvCacheMutex through the whole process of checking the */ + /* converter data cache, and adding new entries to the cache */ + /* to prevent other threads from modifying the cache during the */ + /* process. */ + pArgs->nestedLoads=1; + pArgs->pkg=nullptr; + + umtx_lock(&cnvCacheMutex); + mySharedConverterData = ucnv_load(pArgs, err); + umtx_unlock(&cnvCacheMutex); + if (U_FAILURE (*err) || (mySharedConverterData == nullptr)) + { + return nullptr; + } + } + + return mySharedConverterData; +} + +U_CAPI UConverter * +ucnv_createConverter(UConverter *myUConverter, const char *converterName, UErrorCode * err) +{ + UConverterNamePieces stackPieces; + UConverterLoadArgs stackArgs=UCNV_LOAD_ARGS_INITIALIZER; + UConverterSharedData *mySharedConverterData; + + UTRACE_ENTRY_OC(UTRACE_UCNV_OPEN); + + if(U_SUCCESS(*err)) { + UTRACE_DATA1(UTRACE_OPEN_CLOSE, "open converter %s", converterName); + + mySharedConverterData = ucnv_loadSharedData(converterName, &stackPieces, &stackArgs, err); + + myUConverter = ucnv_createConverterFromSharedData( + myUConverter, mySharedConverterData, + &stackArgs, + err); + + if(U_SUCCESS(*err)) { + UTRACE_EXIT_PTR_STATUS(myUConverter, *err); + return myUConverter; + } + } + + /* exit with error */ + UTRACE_EXIT_STATUS(*err); + return nullptr; +} + +U_CFUNC UBool +ucnv_canCreateConverter(const char *converterName, UErrorCode *err) { + UConverter myUConverter; + UConverterNamePieces stackPieces; + UConverterLoadArgs stackArgs=UCNV_LOAD_ARGS_INITIALIZER; + UConverterSharedData *mySharedConverterData; + + UTRACE_ENTRY_OC(UTRACE_UCNV_OPEN); + + if(U_SUCCESS(*err)) { + UTRACE_DATA1(UTRACE_OPEN_CLOSE, "test if can open converter %s", converterName); + + stackArgs.onlyTestIsLoadable=true; + mySharedConverterData = ucnv_loadSharedData(converterName, &stackPieces, &stackArgs, err); + ucnv_createConverterFromSharedData( + &myUConverter, mySharedConverterData, + &stackArgs, + err); + ucnv_unloadSharedDataIfReady(mySharedConverterData); + } + + UTRACE_EXIT_STATUS(*err); + return U_SUCCESS(*err); +} + +UConverter * +ucnv_createAlgorithmicConverter(UConverter *myUConverter, + UConverterType type, + const char *locale, uint32_t options, + UErrorCode *err) { + UConverter *cnv; + const UConverterSharedData *sharedData; + UConverterLoadArgs stackArgs=UCNV_LOAD_ARGS_INITIALIZER; + + UTRACE_ENTRY_OC(UTRACE_UCNV_OPEN_ALGORITHMIC); + UTRACE_DATA1(UTRACE_OPEN_CLOSE, "open algorithmic converter type %d", (int32_t)type); + + if(type<0 || UCNV_NUMBER_OF_SUPPORTED_CONVERTER_TYPES<=type) { + *err = U_ILLEGAL_ARGUMENT_ERROR; + UTRACE_EXIT_STATUS(U_ILLEGAL_ARGUMENT_ERROR); + return nullptr; + } + + sharedData = converterData[type]; + if(sharedData == nullptr || sharedData->isReferenceCounted) { + /* not a valid type, or not an algorithmic converter */ + *err = U_ILLEGAL_ARGUMENT_ERROR; + UTRACE_EXIT_STATUS(U_ILLEGAL_ARGUMENT_ERROR); + return nullptr; + } + + stackArgs.name = ""; + stackArgs.options = options; + stackArgs.locale=locale; + cnv = ucnv_createConverterFromSharedData( + myUConverter, (UConverterSharedData *)sharedData, + &stackArgs, err); + + UTRACE_EXIT_PTR_STATUS(cnv, *err); + return cnv; +} + +U_CFUNC UConverter* +ucnv_createConverterFromPackage(const char *packageName, const char *converterName, UErrorCode * err) +{ + UConverter *myUConverter; + UConverterSharedData *mySharedConverterData; + UConverterNamePieces stackPieces; + UConverterLoadArgs stackArgs=UCNV_LOAD_ARGS_INITIALIZER; + + UTRACE_ENTRY_OC(UTRACE_UCNV_OPEN_PACKAGE); + + if(U_FAILURE(*err)) { + UTRACE_EXIT_STATUS(*err); + return nullptr; + } + + UTRACE_DATA2(UTRACE_OPEN_CLOSE, "open converter %s from package %s", converterName, packageName); + + /* first, get the options out of the converterName string */ + stackPieces.cnvName[0] = 0; + stackPieces.locale[0] = 0; + stackPieces.options = 0; + parseConverterOptions(converterName, &stackPieces, &stackArgs, err); + if (U_FAILURE(*err)) { + /* Very bad name used. */ + UTRACE_EXIT_STATUS(*err); + return nullptr; + } + stackArgs.nestedLoads=1; + stackArgs.pkg=packageName; + + /* open the data, unflatten the shared structure */ + mySharedConverterData = createConverterFromFile(&stackArgs, err); + + if (U_FAILURE(*err)) { + UTRACE_EXIT_STATUS(*err); + return nullptr; + } + + /* create the actual converter */ + myUConverter = ucnv_createConverterFromSharedData(nullptr, mySharedConverterData, &stackArgs, err); + + if (U_FAILURE(*err)) { + ucnv_close(myUConverter); + UTRACE_EXIT_STATUS(*err); + return nullptr; + } + + UTRACE_EXIT_PTR_STATUS(myUConverter, *err); + return myUConverter; +} + + +U_CFUNC UConverter* +ucnv_createConverterFromSharedData(UConverter *myUConverter, + UConverterSharedData *mySharedConverterData, + UConverterLoadArgs *pArgs, + UErrorCode *err) +{ + UBool isCopyLocal; + + if(U_FAILURE(*err)) { + ucnv_unloadSharedDataIfReady(mySharedConverterData); + return myUConverter; + } + if(myUConverter == nullptr) + { + myUConverter = (UConverter *) uprv_malloc (sizeof (UConverter)); + if(myUConverter == nullptr) + { + *err = U_MEMORY_ALLOCATION_ERROR; + ucnv_unloadSharedDataIfReady(mySharedConverterData); + return nullptr; + } + isCopyLocal = false; + } else { + isCopyLocal = true; + } + + /* initialize the converter */ + uprv_memset(myUConverter, 0, sizeof(UConverter)); + myUConverter->isCopyLocal = isCopyLocal; + /*myUConverter->isExtraLocal = false;*/ /* Set by the memset call */ + myUConverter->sharedData = mySharedConverterData; + myUConverter->options = pArgs->options; + if(!pArgs->onlyTestIsLoadable) { + myUConverter->preFromUFirstCP = U_SENTINEL; + myUConverter->fromCharErrorBehaviour = UCNV_TO_U_DEFAULT_CALLBACK; + myUConverter->fromUCharErrorBehaviour = UCNV_FROM_U_DEFAULT_CALLBACK; + myUConverter->toUnicodeStatus = mySharedConverterData->toUnicodeStatus; + myUConverter->maxBytesPerUChar = mySharedConverterData->staticData->maxBytesPerChar; + myUConverter->subChar1 = mySharedConverterData->staticData->subChar1; + myUConverter->subCharLen = mySharedConverterData->staticData->subCharLen; + myUConverter->subChars = (uint8_t *)myUConverter->subUChars; + uprv_memcpy(myUConverter->subChars, mySharedConverterData->staticData->subChar, myUConverter->subCharLen); + myUConverter->toUCallbackReason = UCNV_ILLEGAL; /* default reason to invoke (*fromCharErrorBehaviour) */ + } + + if(mySharedConverterData->impl->open != nullptr) { + mySharedConverterData->impl->open(myUConverter, pArgs, err); + if(U_FAILURE(*err) && !pArgs->onlyTestIsLoadable) { + /* don't ucnv_close() if onlyTestIsLoadable because not fully initialized */ + ucnv_close(myUConverter); + return nullptr; + } + } + + return myUConverter; +} + +/*Frees all shared immutable objects that aren't referred to (reference count = 0) + */ +U_CAPI int32_t U_EXPORT2 +ucnv_flushCache () +{ + UConverterSharedData *mySharedData = nullptr; + int32_t pos; + int32_t tableDeletedNum = 0; + const UHashElement *e; + /*UErrorCode status = U_ILLEGAL_ARGUMENT_ERROR;*/ + int32_t i, remaining; + + UTRACE_ENTRY_OC(UTRACE_UCNV_FLUSH_CACHE); + + /* Close the default converter without creating a new one so that everything will be flushed. */ + u_flushDefaultConverter(); + + /*if shared data hasn't even been lazy evaluated yet + * return 0 + */ + if (SHARED_DATA_HASHTABLE == nullptr) { + UTRACE_EXIT_VALUE((int32_t)0); + return 0; + } + + /*creates an enumeration to iterate through every element in the + * table + * + * Synchronization: holding cnvCacheMutex will prevent any other thread from + * accessing or modifying the hash table during the iteration. + * The reference count of an entry may be decremented by + * ucnv_close while the iteration is in process, but this is + * benign. It can't be incremented (in ucnv_createConverter()) + * because the sequence of looking up in the cache + incrementing + * is protected by cnvCacheMutex. + */ + umtx_lock(&cnvCacheMutex); + /* + * double loop: A delta/extension-only converter has a pointer to its base table's + * shared data; the first iteration of the outer loop may see the delta converter + * before the base converter, and unloading the delta converter may get the base + * converter's reference counter down to 0. + */ + i = 0; + do { + remaining = 0; + pos = UHASH_FIRST; + while ((e = uhash_nextElement (SHARED_DATA_HASHTABLE, &pos)) != nullptr) + { + mySharedData = (UConverterSharedData *) e->value.pointer; + /*deletes only if reference counter == 0 */ + if (mySharedData->referenceCounter == 0) + { + tableDeletedNum++; + + UCNV_DEBUG_LOG("del",mySharedData->staticData->name,mySharedData); + + uhash_removeElement(SHARED_DATA_HASHTABLE, e); + mySharedData->sharedDataCached = false; + ucnv_deleteSharedConverterData (mySharedData); + } else { + ++remaining; + } + } + } while(++i == 1 && remaining > 0); + umtx_unlock(&cnvCacheMutex); + + UTRACE_DATA1(UTRACE_INFO, "ucnv_flushCache() exits with %d converters remaining", remaining); + + UTRACE_EXIT_VALUE(tableDeletedNum); + return tableDeletedNum; +} + +/* available converters list --------------------------------------------------- */ + +static void U_CALLCONV initAvailableConvertersList(UErrorCode &errCode) { + U_ASSERT(gAvailableConverterCount == 0); + U_ASSERT(gAvailableConverters == nullptr); + + ucnv_enableCleanup(); + UEnumeration *allConvEnum = ucnv_openAllNames(&errCode); + int32_t allConverterCount = uenum_count(allConvEnum, &errCode); + if (U_FAILURE(errCode)) { + return; + } + + /* We can't have more than "*converterTable" converters to open */ + gAvailableConverters = (const char **) uprv_malloc(allConverterCount * sizeof(char*)); + if (!gAvailableConverters) { + errCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + + /* Open the default converter to make sure that it has first dibs in the hash table. */ + UErrorCode localStatus = U_ZERO_ERROR; + UConverter tempConverter; + ucnv_close(ucnv_createConverter(&tempConverter, nullptr, &localStatus)); + + gAvailableConverterCount = 0; + + for (int32_t idx = 0; idx < allConverterCount; idx++) { + localStatus = U_ZERO_ERROR; + const char *converterName = uenum_next(allConvEnum, nullptr, &localStatus); + if (ucnv_canCreateConverter(converterName, &localStatus)) { + gAvailableConverters[gAvailableConverterCount++] = converterName; + } + } + + uenum_close(allConvEnum); +} + + +static UBool haveAvailableConverterList(UErrorCode *pErrorCode) { + umtx_initOnce(gAvailableConvertersInitOnce, &initAvailableConvertersList, *pErrorCode); + return U_SUCCESS(*pErrorCode); +} + +U_CFUNC uint16_t +ucnv_bld_countAvailableConverters(UErrorCode *pErrorCode) { + if (haveAvailableConverterList(pErrorCode)) { + return gAvailableConverterCount; + } + return 0; +} + +U_CFUNC const char * +ucnv_bld_getAvailableConverter(uint16_t n, UErrorCode *pErrorCode) { + if (haveAvailableConverterList(pErrorCode)) { + if (n < gAvailableConverterCount) { + return gAvailableConverters[n]; + } + *pErrorCode = U_INDEX_OUTOFBOUNDS_ERROR; + } + return nullptr; +} + +/* default converter name --------------------------------------------------- */ + +#if !U_CHARSET_IS_UTF8 +/* +Copy the canonical converter name. +ucnv_getDefaultName must be thread safe, which can call this function. + +ucnv_setDefaultName calls this function and it doesn't have to be +thread safe because there is no reliable/safe way to reset the +converter in use in all threads. If you did reset the converter, you +would not be sure that retrieving a default converter for one string +would be the same type of default converter for a successive string. +Since the name is a returned via ucnv_getDefaultName without copying, +you shouldn't be modifying or deleting the string from a separate thread. +*/ +static inline void +internalSetName(const char *name, UErrorCode *status) { + UConverterNamePieces stackPieces; + UConverterLoadArgs stackArgs=UCNV_LOAD_ARGS_INITIALIZER; + int32_t length=(int32_t)(uprv_strlen(name)); + UBool containsOption = (UBool)(uprv_strchr(name, UCNV_OPTION_SEP_CHAR) != nullptr); + const UConverterSharedData *algorithmicSharedData; + + stackArgs.name = name; + if(containsOption) { + stackPieces.cnvName[0] = 0; + stackPieces.locale[0] = 0; + stackPieces.options = 0; + parseConverterOptions(name, &stackPieces, &stackArgs, status); + if(U_FAILURE(*status)) { + return; + } + } + algorithmicSharedData = getAlgorithmicTypeFromName(stackArgs.name); + + umtx_lock(&cnvCacheMutex); + + gDefaultAlgorithmicSharedData = algorithmicSharedData; + gDefaultConverterContainsOption = containsOption; + uprv_memcpy(gDefaultConverterNameBuffer, name, length); + gDefaultConverterNameBuffer[length]=0; + + /* gDefaultConverterName MUST be the last global var set by this function. */ + /* It is the variable checked in ucnv_getDefaultName() to see if initialization is required. */ + // But there is nothing here preventing that from being reordered, either by the compiler + // or hardware. I'm adding the mutex to ucnv_getDefaultName for now. UMTX_CHECK is not enough. + // -- Andy + gDefaultConverterName = gDefaultConverterNameBuffer; + + ucnv_enableCleanup(); + + umtx_unlock(&cnvCacheMutex); +} +#endif + +/* + * In order to be really thread-safe, the get function would have to take + * a buffer parameter and copy the current string inside a mutex block. + * This implementation only tries to be really thread-safe while + * setting the name. + * It assumes that setting a pointer is atomic. + */ + +U_CAPI const char* U_EXPORT2 +ucnv_getDefaultName() { +#if U_CHARSET_IS_UTF8 + return "UTF-8"; +#else + /* local variable to be thread-safe */ + const char *name; + + /* + Concurrent calls to ucnv_getDefaultName must be thread safe, + but ucnv_setDefaultName is not thread safe. + */ + { + icu::Mutex lock(&cnvCacheMutex); + name = gDefaultConverterName; + } + if(name==nullptr) { + UErrorCode errorCode = U_ZERO_ERROR; + UConverter *cnv = nullptr; + + name = uprv_getDefaultCodepage(); + + /* if the name is there, test it out and get the canonical name with options */ + if(name != nullptr) { + cnv = ucnv_open(name, &errorCode); + if(U_SUCCESS(errorCode) && cnv != nullptr) { + name = ucnv_getName(cnv, &errorCode); + } + } + + if(name == nullptr || name[0] == 0 + || U_FAILURE(errorCode) || cnv == nullptr + || uprv_strlen(name)>=sizeof(gDefaultConverterNameBuffer)) + { + /* Panic time, let's use a fallback. */ +#if (U_CHARSET_FAMILY == U_ASCII_FAMILY) + name = "US-ASCII"; + /* there is no 'algorithmic' converter for EBCDIC */ +#elif U_PLATFORM == U_PF_OS390 + name = "ibm-1047_P100-1995" UCNV_SWAP_LFNL_OPTION_STRING; +#else + name = "ibm-37_P100-1995"; +#endif + } + + internalSetName(name, &errorCode); + + /* The close may make the current name go away. */ + ucnv_close(cnv); + } + + return name; +#endif +} + +#if U_CHARSET_IS_UTF8 +U_CAPI void U_EXPORT2 ucnv_setDefaultName(const char *) {} +#else +/* +This function is not thread safe, and it can't be thread safe. +See internalSetName or the API reference for details. +*/ +U_CAPI void U_EXPORT2 +ucnv_setDefaultName(const char *converterName) { + if(converterName==nullptr) { + /* reset to the default codepage */ + gDefaultConverterName=nullptr; + } else { + UErrorCode errorCode = U_ZERO_ERROR; + UConverter *cnv = nullptr; + const char *name = nullptr; + + /* if the name is there, test it out and get the canonical name with options */ + cnv = ucnv_open(converterName, &errorCode); + if(U_SUCCESS(errorCode) && cnv != nullptr) { + name = ucnv_getName(cnv, &errorCode); + } + + if(U_SUCCESS(errorCode) && name!=nullptr) { + internalSetName(name, &errorCode); + } + /* else this converter is bad to use. Don't change it to a bad value. */ + + /* The close may make the current name go away. */ + ucnv_close(cnv); + + /* reset the converter cache */ + u_flushDefaultConverter(); + } +} +#endif + +/* data swapping ------------------------------------------------------------ */ + +/* most of this might belong more properly into ucnvmbcs.c, but that is so large */ + +#if !UCONFIG_NO_LEGACY_CONVERSION + +U_CAPI int32_t U_EXPORT2 +ucnv_swap(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const UDataInfo *pInfo; + int32_t headerSize; + + const uint8_t *inBytes; + uint8_t *outBytes; + + uint32_t offset, count, staticDataSize; + int32_t size; + + const UConverterStaticData *inStaticData; + UConverterStaticData *outStaticData; + + const _MBCSHeader *inMBCSHeader; + _MBCSHeader *outMBCSHeader; + _MBCSHeader mbcsHeader; + uint32_t mbcsHeaderLength; + UBool noFromU=false; + + uint8_t outputType; + + int32_t maxFastUChar, mbcsIndexLength; + + const int32_t *inExtIndexes; + int32_t extOffset; + + /* udata_swapDataHeader checks the arguments */ + headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode); + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + /* check data format and format version */ + pInfo=(const UDataInfo *)((const char *)inData+4); + if(!( + pInfo->dataFormat[0]==0x63 && /* dataFormat="cnvt" */ + pInfo->dataFormat[1]==0x6e && + pInfo->dataFormat[2]==0x76 && + pInfo->dataFormat[3]==0x74 && + pInfo->formatVersion[0]==6 && + pInfo->formatVersion[1]>=2 + )) { + udata_printError(ds, "ucnv_swap(): data format %02x.%02x.%02x.%02x (format version %02x.%02x) is not recognized as an ICU .cnv conversion table\n", + pInfo->dataFormat[0], pInfo->dataFormat[1], + pInfo->dataFormat[2], pInfo->dataFormat[3], + pInfo->formatVersion[0], pInfo->formatVersion[1]); + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + + inBytes=(const uint8_t *)inData+headerSize; + outBytes=(outData == nullptr) ? nullptr : (uint8_t *)outData+headerSize; + + /* read the initial UConverterStaticData structure after the UDataInfo header */ + inStaticData=(const UConverterStaticData *)inBytes; + outStaticData=(UConverterStaticData *)outBytes; + + if(length<0) { + staticDataSize=ds->readUInt32(inStaticData->structSize); + } else { + length-=headerSize; + if( length<(int32_t)sizeof(UConverterStaticData) || + (uint32_t)length<(staticDataSize=ds->readUInt32(inStaticData->structSize)) + ) { + udata_printError(ds, "ucnv_swap(): too few bytes (%d after header) for an ICU .cnv conversion table\n", + length); + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + } + + if(length>=0) { + /* swap the static data */ + if(inStaticData!=outStaticData) { + uprv_memcpy(outStaticData, inStaticData, staticDataSize); + } + + ds->swapArray32(ds, &inStaticData->structSize, 4, + &outStaticData->structSize, pErrorCode); + ds->swapArray32(ds, &inStaticData->codepage, 4, + &outStaticData->codepage, pErrorCode); + + ds->swapInvChars(ds, inStaticData->name, (int32_t)uprv_strlen(inStaticData->name), + outStaticData->name, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + udata_printError(ds, "ucnv_swap(): error swapping converter name\n"); + return 0; + } + } + + inBytes+=staticDataSize; + if (outBytes != nullptr) outBytes+=staticDataSize; + if(length>=0) { + length-=(int32_t)staticDataSize; + } + + /* check for supported conversionType values */ + if(inStaticData->conversionType==UCNV_MBCS) { + /* swap MBCS data */ + inMBCSHeader=(const _MBCSHeader *)inBytes; + outMBCSHeader=(_MBCSHeader *)outBytes; + + if(0<=length && length<(int32_t)sizeof(_MBCSHeader)) { + udata_printError(ds, "ucnv_swap(): too few bytes (%d after headers) for an ICU MBCS .cnv conversion table\n", + length); + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + if(inMBCSHeader->version[0]==4 && inMBCSHeader->version[1]>=1) { + mbcsHeaderLength=MBCS_HEADER_V4_LENGTH; + } else if(inMBCSHeader->version[0]==5 && inMBCSHeader->version[1]>=3 && + ((mbcsHeader.options=ds->readUInt32(inMBCSHeader->options))& + MBCS_OPT_UNKNOWN_INCOMPATIBLE_MASK)==0 + ) { + mbcsHeaderLength=mbcsHeader.options&MBCS_OPT_LENGTH_MASK; + noFromU=(UBool)((mbcsHeader.options&MBCS_OPT_NO_FROM_U)!=0); + } else { + udata_printError(ds, "ucnv_swap(): unsupported _MBCSHeader.version %d.%d\n", + inMBCSHeader->version[0], inMBCSHeader->version[1]); + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + + uprv_memcpy(mbcsHeader.version, inMBCSHeader->version, 4); + mbcsHeader.countStates= ds->readUInt32(inMBCSHeader->countStates); + mbcsHeader.countToUFallbacks= ds->readUInt32(inMBCSHeader->countToUFallbacks); + mbcsHeader.offsetToUCodeUnits= ds->readUInt32(inMBCSHeader->offsetToUCodeUnits); + mbcsHeader.offsetFromUTable= ds->readUInt32(inMBCSHeader->offsetFromUTable); + mbcsHeader.offsetFromUBytes= ds->readUInt32(inMBCSHeader->offsetFromUBytes); + mbcsHeader.flags= ds->readUInt32(inMBCSHeader->flags); + mbcsHeader.fromUBytesLength= ds->readUInt32(inMBCSHeader->fromUBytesLength); + /* mbcsHeader.options have been read above */ + + extOffset=(int32_t)(mbcsHeader.flags>>8); + outputType=(uint8_t)mbcsHeader.flags; + if(noFromU && outputType==MBCS_OUTPUT_1) { + udata_printError(ds, "ucnv_swap(): unsupported combination of makeconv --small with SBCS\n"); + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + + /* make sure that the output type is known */ + switch(outputType) { + case MBCS_OUTPUT_1: + case MBCS_OUTPUT_2: + case MBCS_OUTPUT_3: + case MBCS_OUTPUT_4: + case MBCS_OUTPUT_3_EUC: + case MBCS_OUTPUT_4_EUC: + case MBCS_OUTPUT_2_SISO: + case MBCS_OUTPUT_EXT_ONLY: + /* OK */ + break; + default: + udata_printError(ds, "ucnv_swap(): unsupported MBCS output type 0x%x\n", + outputType); + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + + /* calculate the length of the MBCS data */ + + /* + * utf8Friendly MBCS files (mbcsHeader.version 4.3) + * contain an additional mbcsIndex table: + * uint16_t[(maxFastUChar+1)>>6]; + * where maxFastUChar=((mbcsHeader.version[2]<<8)|0xff). + */ + maxFastUChar=0; + mbcsIndexLength=0; + if( outputType!=MBCS_OUTPUT_EXT_ONLY && outputType!=MBCS_OUTPUT_1 && + mbcsHeader.version[1]>=3 && (maxFastUChar=mbcsHeader.version[2])!=0 + ) { + maxFastUChar=(maxFastUChar<<8)|0xff; + mbcsIndexLength=((maxFastUChar+1)>>6)*2; /* number of bytes */ + } + + if(extOffset==0) { + size=(int32_t)(mbcsHeader.offsetFromUBytes+mbcsIndexLength); + if(!noFromU) { + size+=(int32_t)mbcsHeader.fromUBytesLength; + } + + /* avoid compiler warnings - not otherwise necessary, and the value does not matter */ + inExtIndexes=nullptr; + } else { + /* there is extension data after the base data, see ucnv_ext.h */ + if(length>=0 && length<(extOffset+UCNV_EXT_INDEXES_MIN_LENGTH*4)) { + udata_printError(ds, "ucnv_swap(): too few bytes (%d after headers) for an ICU MBCS .cnv conversion table with extension data\n", + length); + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + + inExtIndexes=(const int32_t *)(inBytes+extOffset); + size=extOffset+udata_readInt32(ds, inExtIndexes[UCNV_EXT_SIZE]); + } + + if(length>=0) { + if(lengthswapArray32(ds, &inMBCSHeader->countStates, count-4, + &outMBCSHeader->countStates, pErrorCode); + + if(outputType==MBCS_OUTPUT_EXT_ONLY) { + /* + * extension-only file, + * contains a base name instead of normal base table data + */ + + /* swap the base name, between the header and the extension data */ + const char *inBaseName=(const char *)inBytes+count; + char *outBaseName=(char *)outBytes+count; + ds->swapInvChars(ds, inBaseName, (int32_t)uprv_strlen(inBaseName), + outBaseName, pErrorCode); + } else { + /* normal file with base table data */ + + /* swap the state table, 1kB per state */ + offset=count; + count=mbcsHeader.countStates*1024; + ds->swapArray32(ds, inBytes+offset, (int32_t)count, + outBytes+offset, pErrorCode); + + /* swap the toUFallbacks[] */ + offset+=count; + count=mbcsHeader.countToUFallbacks*8; + ds->swapArray32(ds, inBytes+offset, (int32_t)count, + outBytes+offset, pErrorCode); + + /* swap the unicodeCodeUnits[] */ + offset=mbcsHeader.offsetToUCodeUnits; + count=mbcsHeader.offsetFromUTable-offset; + ds->swapArray16(ds, inBytes+offset, (int32_t)count, + outBytes+offset, pErrorCode); + + /* offset to the stage 1 table, independent of the outputType */ + offset=mbcsHeader.offsetFromUTable; + + if(outputType==MBCS_OUTPUT_1) { + /* SBCS: swap the fromU tables, all 16 bits wide */ + count=(mbcsHeader.offsetFromUBytes-offset)+mbcsHeader.fromUBytesLength; + ds->swapArray16(ds, inBytes+offset, (int32_t)count, + outBytes+offset, pErrorCode); + } else { + /* otherwise: swap the stage tables separately */ + + /* stage 1 table: uint16_t[0x440 or 0x40] */ + if(inStaticData->unicodeMask&UCNV_HAS_SUPPLEMENTARY) { + count=0x440*2; /* for all of Unicode */ + } else { + count=0x40*2; /* only BMP */ + } + ds->swapArray16(ds, inBytes+offset, (int32_t)count, + outBytes+offset, pErrorCode); + + /* stage 2 table: uint32_t[] */ + offset+=count; + count=mbcsHeader.offsetFromUBytes-offset; + ds->swapArray32(ds, inBytes+offset, (int32_t)count, + outBytes+offset, pErrorCode); + + /* stage 3/result bytes: sometimes uint16_t[] or uint32_t[] */ + offset=mbcsHeader.offsetFromUBytes; + count= noFromU ? 0 : mbcsHeader.fromUBytesLength; + switch(outputType) { + case MBCS_OUTPUT_2: + case MBCS_OUTPUT_3_EUC: + case MBCS_OUTPUT_2_SISO: + ds->swapArray16(ds, inBytes+offset, (int32_t)count, + outBytes+offset, pErrorCode); + break; + case MBCS_OUTPUT_4: + ds->swapArray32(ds, inBytes+offset, (int32_t)count, + outBytes+offset, pErrorCode); + break; + default: + /* just uint8_t[], nothing to swap */ + break; + } + + if(mbcsIndexLength!=0) { + offset+=count; + count=mbcsIndexLength; + ds->swapArray16(ds, inBytes+offset, (int32_t)count, + outBytes+offset, pErrorCode); + } + } + } + + if(extOffset!=0) { + /* swap the extension data */ + inBytes+=extOffset; + outBytes+=extOffset; + + /* swap toUTable[] */ + offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_TO_U_INDEX]); + length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_TO_U_LENGTH]); + ds->swapArray32(ds, inBytes+offset, length*4, outBytes+offset, pErrorCode); + + /* swap toUUChars[] */ + offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_TO_U_UCHARS_INDEX]); + length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_TO_U_UCHARS_LENGTH]); + ds->swapArray16(ds, inBytes+offset, length*2, outBytes+offset, pErrorCode); + + /* swap fromUTableUChars[] */ + offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_UCHARS_INDEX]); + length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_LENGTH]); + ds->swapArray16(ds, inBytes+offset, length*2, outBytes+offset, pErrorCode); + + /* swap fromUTableValues[] */ + offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_VALUES_INDEX]); + /* same length as for fromUTableUChars[] */ + ds->swapArray32(ds, inBytes+offset, length*4, outBytes+offset, pErrorCode); + + /* no need to swap fromUBytes[] */ + + /* swap fromUStage12[] */ + offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_STAGE_12_INDEX]); + length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_STAGE_12_LENGTH]); + ds->swapArray16(ds, inBytes+offset, length*2, outBytes+offset, pErrorCode); + + /* swap fromUStage3[] */ + offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_STAGE_3_INDEX]); + length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_STAGE_3_LENGTH]); + ds->swapArray16(ds, inBytes+offset, length*2, outBytes+offset, pErrorCode); + + /* swap fromUStage3b[] */ + offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_STAGE_3B_INDEX]); + length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_STAGE_3B_LENGTH]); + ds->swapArray32(ds, inBytes+offset, length*4, outBytes+offset, pErrorCode); + + /* swap indexes[] */ + length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_INDEXES_LENGTH]); + ds->swapArray32(ds, inBytes, length*4, outBytes, pErrorCode); + } + } + } else { + udata_printError(ds, "ucnv_swap(): unknown conversionType=%d!=UCNV_MBCS\n", + inStaticData->conversionType); + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + + return headerSize+(int32_t)staticDataSize+size; +} + +#endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */ + +#endif diff --git a/intl/icu/source/common/ucnv_bld.h b/intl/icu/source/common/ucnv_bld.h new file mode 100644 index 0000000000..43e6c09ac0 --- /dev/null +++ b/intl/icu/source/common/ucnv_bld.h @@ -0,0 +1,296 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1999-2015 International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* +* +* ucnv_bld.h: +* Contains internal data structure definitions +* Created by Bertrand A. Damiba +* +* Change history: +* +* 06/29/2000 helena Major rewrite of the callback APIs. +*/ + +#ifndef UCNV_BLD_H +#define UCNV_BLD_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/ucnv_err.h" +#include "unicode/utf16.h" +#include "ucnv_cnv.h" +#include "ucnvmbcs.h" +#include "ucnv_ext.h" +#include "udataswp.h" + +/* size of the overflow buffers in UConverter, enough for escaping callbacks */ +#define UCNV_ERROR_BUFFER_LENGTH 32 + +/* at most 4 bytes per substitution character (part of .cnv file format! see UConverterStaticData) */ +#define UCNV_MAX_SUBCHAR_LEN 4 + +/* at most 8 bytes per character in toUBytes[] (UTF-8 uses up to 6) */ +#define UCNV_MAX_CHAR_LEN 8 + +/* converter options bits */ +#define UCNV_OPTION_VERSION 0xf +#define UCNV_OPTION_SWAP_LFNL 0x10 + +#define UCNV_GET_VERSION(cnv) ((cnv)->options&UCNV_OPTION_VERSION) + +U_CDECL_BEGIN /* We must declare the following as 'extern "C"' so that if ucnv + itself is compiled under C++, the linkage of the funcptrs will + work. + */ + +union UConverterTable { + UConverterMBCSTable mbcs; +}; + +typedef union UConverterTable UConverterTable; + +struct UConverterImpl; +typedef struct UConverterImpl UConverterImpl; + +/** values for the unicodeMask */ +#define UCNV_HAS_SUPPLEMENTARY 1 +#define UCNV_HAS_SURROGATES 2 + +typedef struct UConverterStaticData { /* +offset: size */ + uint32_t structSize; /* +0: 4 Size of this structure */ + + char name + [UCNV_MAX_CONVERTER_NAME_LENGTH]; /* +4: 60 internal name of the converter- invariant chars */ + + int32_t codepage; /* +64: 4 codepage # (now IBM-$codepage) */ + + int8_t platform; /* +68: 1 platform of the converter (only IBM now) */ + int8_t conversionType; /* +69: 1 conversion type */ + + int8_t minBytesPerChar; /* +70: 1 Minimum # bytes per char in this codepage */ + int8_t maxBytesPerChar; /* +71: 1 Maximum # bytes output per UChar in this codepage */ + + uint8_t subChar[UCNV_MAX_SUBCHAR_LEN]; /* +72: 4 [note: 4 and 8 byte boundary] */ + int8_t subCharLen; /* +76: 1 */ + + uint8_t hasToUnicodeFallback; /* +77: 1 UBool needs to be changed to UBool to be consistent across platform */ + uint8_t hasFromUnicodeFallback; /* +78: 1 */ + uint8_t unicodeMask; /* +79: 1 bit 0: has supplementary bit 1: has single surrogates */ + uint8_t subChar1; /* +80: 1 single-byte substitution character for IBM MBCS (0 if none) */ + uint8_t reserved[19]; /* +81: 19 to round out the structure */ + /* total size: 100 */ +} UConverterStaticData; + +/* + * Defines the UConverterSharedData struct, + * the immutable, shared part of UConverter. + */ +struct UConverterSharedData { + uint32_t structSize; /* Size of this structure */ + uint32_t referenceCounter; /* used to count number of clients, unused for static/immutable SharedData */ + + const void *dataMemory; /* from udata_openChoice() - for cleanup */ + + const UConverterStaticData *staticData; /* pointer to the static (non changing) data. */ + + UBool sharedDataCached; /* true: shared data is in cache, don't destroy on ucnv_close() if 0 ref. false: shared data isn't in the cache, do attempt to clean it up if the ref is 0 */ + /** If false, then referenceCounter is not used. Must not change after initialization. */ + UBool isReferenceCounted; + + const UConverterImpl *impl; /* vtable-style struct of mostly function pointers */ + + /*initial values of some members of the mutable part of object */ + uint32_t toUnicodeStatus; + + /* + * Shared data structures currently come in two flavors: + * - readonly for built-in algorithmic converters + * - allocated for MBCS, with a pointer to an allocated UConverterTable + * which always has a UConverterMBCSTable + * + * To eliminate one allocation, I am making the UConverterMBCSTable + * a member of the shared data. + * + * markus 2003-nov-07 + */ + UConverterMBCSTable mbcs; +}; + +/** UConverterSharedData initializer for static, non-reference-counted converters. */ +#define UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(pStaticData, pImpl) \ + { \ + sizeof(UConverterSharedData), ~((uint32_t)0), \ + NULL, pStaticData, false, false, pImpl, \ + 0, UCNV_MBCS_TABLE_INITIALIZER \ + } + +/* Defines a UConverter, the lightweight mutable part the user sees */ + +struct UConverter { + /* + * Error function pointer called when conversion issues + * occur during a ucnv_fromUnicode call + */ + void (U_EXPORT2 *fromUCharErrorBehaviour) (const void *context, + UConverterFromUnicodeArgs *args, + const UChar *codeUnits, + int32_t length, + UChar32 codePoint, + UConverterCallbackReason reason, + UErrorCode *); + /* + * Error function pointer called when conversion issues + * occur during a ucnv_toUnicode call + */ + void (U_EXPORT2 *fromCharErrorBehaviour) (const void *context, + UConverterToUnicodeArgs *args, + const char *codeUnits, + int32_t length, + UConverterCallbackReason reason, + UErrorCode *); + + /* + * Pointer to additional data that depends on the converter type. + * Used by ISO 2022, SCSU, GB 18030 converters, possibly more. + */ + void *extraInfo; + + const void *fromUContext; + const void *toUContext; + + /* + * Pointer to charset bytes for substitution string if subCharLen>0, + * or pointer to Unicode string (UChar *) if subCharLen<0. + * subCharLen==0 is equivalent to using a skip callback. + * If the pointer is !=subUChars then it is allocated with + * UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR bytes. + * The subUChars field is declared as UChar[] not uint8_t[] to + * guarantee alignment for UChars. + */ + uint8_t *subChars; + + UConverterSharedData *sharedData; /* Pointer to the shared immutable part of the converter object */ + + uint32_t options; /* options flags from UConverterOpen, may contain additional bits */ + + UBool sharedDataIsCached; /* true: shared data is in cache, don't destroy on ucnv_close() if 0 ref. false: shared data isn't in the cache, do attempt to clean it up if the ref is 0 */ + UBool isCopyLocal; /* true if UConverter is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */ + UBool isExtraLocal; /* true if extraInfo is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */ + + UBool useFallback; + int8_t toULength; /* number of bytes in toUBytes */ + uint8_t toUBytes[UCNV_MAX_CHAR_LEN-1];/* more "toU status"; keeps the bytes of the current character */ + uint32_t toUnicodeStatus; /* Used to internalize stream status information */ + int32_t mode; + uint32_t fromUnicodeStatus; + + /* + * More fromUnicode() status. Serves 3 purposes: + * - keeps a lead surrogate between buffers (similar to toUBytes[]) + * - keeps a lead surrogate at the end of the stream, + * which the framework handles as truncated input + * - if the fromUnicode() implementation returns to the framework + * (ucnv.c ucnv_fromUnicode()), then the framework calls the callback + * for this code point + */ + UChar32 fromUChar32; + + /* + * value for ucnv_getMaxCharSize() + * + * usually simply copied from the static data, but ucnvmbcs.c modifies + * the value depending on the converter type and options + */ + int8_t maxBytesPerUChar; + + int8_t subCharLen; /* length of the codepage specific character sequence */ + int8_t invalidCharLength; + int8_t charErrorBufferLength; /* number of valid bytes in charErrorBuffer */ + + int8_t invalidUCharLength; + int8_t UCharErrorBufferLength; /* number of valid UChars in charErrorBuffer */ + + uint8_t subChar1; /* single-byte substitution character if different from subChar */ + UBool useSubChar1; + char invalidCharBuffer[UCNV_MAX_CHAR_LEN]; /* bytes from last error/callback situation */ + uint8_t charErrorBuffer[UCNV_ERROR_BUFFER_LENGTH]; /* codepage output from Error functions */ + UChar subUChars[UCNV_MAX_SUBCHAR_LEN/U_SIZEOF_UCHAR]; /* see subChars documentation */ + + UChar invalidUCharBuffer[U16_MAX_LENGTH]; /* UChars from last error/callback situation */ + UChar UCharErrorBuffer[UCNV_ERROR_BUFFER_LENGTH]; /* unicode output from Error functions */ + + /* fields for conversion extension */ + + /* store previous UChars/chars to continue partial matches */ + UChar32 preFromUFirstCP; /* >=0: partial match */ + UChar preFromU[UCNV_EXT_MAX_UCHARS]; + char preToU[UCNV_EXT_MAX_BYTES]; + int8_t preFromULength, preToULength; /* negative: replay */ + int8_t preToUFirstLength; /* length of first character */ + + /* new fields for ICU 4.0 */ + UConverterCallbackReason toUCallbackReason; /* (*fromCharErrorBehaviour) reason, set when error is detected */ +}; + +U_CDECL_END /* end of UConverter */ + +#define CONVERTER_FILE_EXTENSION ".cnv" + + +/** + * Return the number of all converter names. + * @param pErrorCode The error code + * @return the number of all converter names + */ +U_CFUNC uint16_t +ucnv_bld_countAvailableConverters(UErrorCode *pErrorCode); + +/** + * Return the (n)th converter name in mixed case, or NULL + * if there is none (typically, if the data cannot be loaded). + * 0<=indexisReferenceCounted + * and this function must be called inside umtx_lock(&cnvCacheMutex). + */ +U_CAPI void +ucnv_unload(UConverterSharedData *sharedData); + +/** + * Swap ICU .cnv conversion tables. See udataswp.h. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +ucnv_swap(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +U_CAPI void U_EXPORT2 +ucnv_enableCleanup(void); + +#endif + +#endif /* _UCNV_BLD */ diff --git a/intl/icu/source/common/ucnv_cb.cpp b/intl/icu/source/common/ucnv_cb.cpp new file mode 100644 index 0000000000..dd559270dd --- /dev/null +++ b/intl/icu/source/common/ucnv_cb.cpp @@ -0,0 +1,261 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2000-2006, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** + * ucnv_cb.c: + * External APIs for the ICU's codeset conversion library + * Helena Shih + * + * Modification History: + * + * Date Name Description + * 7/28/2000 srl Implementation + */ + +/** + * @name Character Conversion C API + * + */ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/ucnv_cb.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" +#include "cmemory.h" + +/* need to update the offsets when the target moves. */ +/* Note: Recursion may occur in the cb functions, be sure to update the offsets correctly +if you don't use ucnv_cbXXX functions. Make sure you don't use the same callback within +the same call stack if the complexity arises. */ +U_CAPI void U_EXPORT2 +ucnv_cbFromUWriteBytes (UConverterFromUnicodeArgs *args, + const char* source, + int32_t length, + int32_t offsetIndex, + UErrorCode * err) +{ + if(U_FAILURE(*err)) { + return; + } + + ucnv_fromUWriteBytes( + args->converter, + source, length, + &args->target, args->targetLimit, + &args->offsets, offsetIndex, + err); +} + +U_CAPI void U_EXPORT2 +ucnv_cbFromUWriteUChars(UConverterFromUnicodeArgs *args, + const char16_t** source, + const char16_t* sourceLimit, + int32_t offsetIndex, + UErrorCode * err) +{ + /* + This is a fun one. Recursion can occur - we're basically going to + just retry shoving data through the same converter. Note, if you got + here through some kind of invalid sequence, you maybe should emit a + reset sequence of some kind and/or call ucnv_reset(). Since this + IS an actual conversion, take care that you've changed the callback + or the data, or you'll get an infinite loop. + + Please set the err value to something reasonable before calling + into this. + */ + + char *oldTarget; + + if(U_FAILURE(*err)) + { + return; + } + + oldTarget = args->target; + + ucnv_fromUnicode(args->converter, + &args->target, + args->targetLimit, + source, + sourceLimit, + nullptr, /* no offsets */ + false, /* no flush */ + err); + + if(args->offsets) + { + while (args->target != oldTarget) /* if it moved at all.. */ + { + *(args->offsets)++ = offsetIndex; + oldTarget++; + } + } + + /* + Note, if you did something like used a Stop subcallback, things would get interesting. + In fact, here's where we want to return the partially consumed in-source! + */ + if(*err == U_BUFFER_OVERFLOW_ERROR) + /* && (*source < sourceLimit && args->target >= args->targetLimit) + -- S. Hrcek */ + { + /* Overflowed the target. Now, we'll write into the charErrorBuffer. + It's a fixed size. If we overflow it... Hmm */ + char *newTarget; + const char *newTargetLimit; + UErrorCode err2 = U_ZERO_ERROR; + + int8_t errBuffLen; + + errBuffLen = args->converter->charErrorBufferLength; + + /* start the new target at the first free slot in the errbuff.. */ + newTarget = (char *)(args->converter->charErrorBuffer + errBuffLen); + + newTargetLimit = (char *)(args->converter->charErrorBuffer + + sizeof(args->converter->charErrorBuffer)); + + if(newTarget >= newTargetLimit) + { + *err = U_INTERNAL_PROGRAM_ERROR; + return; + } + + /* We're going to tell the converter that the errbuff len is empty. + This prevents the existing errbuff from being 'flushed' out onto + itself. If the errbuff is needed by the converter this time, + we're hosed - we're out of space! */ + + args->converter->charErrorBufferLength = 0; + + ucnv_fromUnicode(args->converter, + &newTarget, + newTargetLimit, + source, + sourceLimit, + nullptr, + false, + &err2); + + /* We can go ahead and overwrite the length here. We know just how + to recalculate it. */ + + args->converter->charErrorBufferLength = (int8_t)( + newTarget - (char*)args->converter->charErrorBuffer); + + if((newTarget >= newTargetLimit) || (err2 == U_BUFFER_OVERFLOW_ERROR)) + { + /* now we're REALLY in trouble. + Internal program error - callback shouldn't have written this much + data! + */ + *err = U_INTERNAL_PROGRAM_ERROR; + return; + } + /*else {*/ + /* sub errs could be invalid/truncated/illegal chars or w/e. + These might want to be passed on up.. But the problem is, we already + need to pass U_BUFFER_OVERFLOW_ERROR. That has to override these + other errs.. */ + + /* + if(U_FAILURE(err2)) + ?? + */ + /*}*/ + } +} + +U_CAPI void U_EXPORT2 +ucnv_cbFromUWriteSub (UConverterFromUnicodeArgs *args, + int32_t offsetIndex, + UErrorCode * err) +{ + UConverter *converter; + int32_t length; + + if(U_FAILURE(*err)) { + return; + } + converter = args->converter; + length = converter->subCharLen; + + if(length == 0) { + return; + } + + if(length < 0) { + /* + * Write/convert the substitution string. Its real length is -length. + * Unlike the escape callback, we need not change the converter's + * callback function because ucnv_setSubstString() verified that + * the string can be converted, so we will not get a conversion error + * and will not recurse. + * At worst we should get a U_BUFFER_OVERFLOW_ERROR. + */ + const char16_t *source = (const char16_t *)converter->subChars; + ucnv_cbFromUWriteUChars(args, &source, source - length, offsetIndex, err); + return; + } + + if(converter->sharedData->impl->writeSub!=nullptr) { + converter->sharedData->impl->writeSub(args, offsetIndex, err); + } + else if(converter->subChar1!=0 && (uint16_t)converter->invalidUCharBuffer[0]<=(uint16_t)0xffu) { + /* + TODO: Is this untestable because the MBCS converter has a writeSub function to call + and the other converters don't use subChar1? + */ + ucnv_cbFromUWriteBytes(args, + (const char *)&converter->subChar1, 1, + offsetIndex, err); + } + else { + ucnv_cbFromUWriteBytes(args, + (const char *)converter->subChars, length, + offsetIndex, err); + } +} + +U_CAPI void U_EXPORT2 +ucnv_cbToUWriteUChars (UConverterToUnicodeArgs *args, + const char16_t* source, + int32_t length, + int32_t offsetIndex, + UErrorCode * err) +{ + if(U_FAILURE(*err)) { + return; + } + + ucnv_toUWriteUChars( + args->converter, + source, length, + &args->target, args->targetLimit, + &args->offsets, offsetIndex, + err); +} + +U_CAPI void U_EXPORT2 +ucnv_cbToUWriteSub (UConverterToUnicodeArgs *args, + int32_t offsetIndex, + UErrorCode * err) +{ + static const char16_t kSubstituteChar1 = 0x1A, kSubstituteChar = 0xFFFD; + + /* could optimize this case, just one uchar */ + if(args->converter->invalidCharLength == 1 && args->converter->subChar1 != 0) { + ucnv_cbToUWriteUChars(args, &kSubstituteChar1, 1, offsetIndex, err); + } else { + ucnv_cbToUWriteUChars(args, &kSubstituteChar, 1, offsetIndex, err); + } +} + +#endif diff --git a/intl/icu/source/common/ucnv_cnv.cpp b/intl/icu/source/common/ucnv_cnv.cpp new file mode 100644 index 0000000000..ffff9429ad --- /dev/null +++ b/intl/icu/source/common/ucnv_cnv.cpp @@ -0,0 +1,182 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2004, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* uconv_cnv.c: +* Implements all the low level conversion functions +* T_UnicodeConverter_{to,from}Unicode_$ConversionType +* +* Change history: +* +* 06/29/2000 helena Major rewrite of the callback APIs. +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/ucnv_err.h" +#include "unicode/ucnv.h" +#include "unicode/uset.h" +#include "ucnv_cnv.h" +#include "ucnv_bld.h" +#include "cmemory.h" + +U_CFUNC void +ucnv_getCompleteUnicodeSet(const UConverter *cnv, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode) { + (void)cnv; + (void)which; + (void)pErrorCode; + sa->addRange(sa->set, 0, 0x10ffff); +} + +U_CFUNC void +ucnv_getNonSurrogateUnicodeSet(const UConverter *cnv, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode) { + (void)cnv; + (void)which; + (void)pErrorCode; + sa->addRange(sa->set, 0, 0xd7ff); + sa->addRange(sa->set, 0xe000, 0x10ffff); +} + +U_CFUNC void +ucnv_fromUWriteBytes(UConverter *cnv, + const char *bytes, int32_t length, + char **target, const char *targetLimit, + int32_t **offsets, + int32_t sourceIndex, + UErrorCode *pErrorCode) { + char *t=*target; + int32_t *o; + + /* write bytes */ + if(offsets==nullptr || (o=*offsets)==nullptr) { + while(length>0 && t0 && t0) { + if(cnv!=nullptr) { + t=(char *)cnv->charErrorBuffer; + cnv->charErrorBufferLength=(int8_t)length; + do { + *t++=(uint8_t)*bytes++; + } while(--length>0); + } + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } +} + +U_CFUNC void +ucnv_toUWriteUChars(UConverter *cnv, + const char16_t *uchars, int32_t length, + char16_t **target, const char16_t *targetLimit, + int32_t **offsets, + int32_t sourceIndex, + UErrorCode *pErrorCode) { + char16_t *t=*target; + int32_t *o; + + /* write UChars */ + if(offsets==nullptr || (o=*offsets)==nullptr) { + while(length>0 && t0 && t0) { + if(cnv!=nullptr) { + t=cnv->UCharErrorBuffer; + cnv->UCharErrorBufferLength=(int8_t)length; + do { + *t++=*uchars++; + } while(--length>0); + } + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } +} + +U_CFUNC void +ucnv_toUWriteCodePoint(UConverter *cnv, + UChar32 c, + char16_t **target, const char16_t *targetLimit, + int32_t **offsets, + int32_t sourceIndex, + UErrorCode *pErrorCode) { + char16_t *t; + int32_t *o; + + t=*target; + + if(t=0) { + if(cnv!=nullptr) { + int8_t i=0; + U16_APPEND_UNSAFE(cnv->UCharErrorBuffer, i, c); + cnv->UCharErrorBufferLength=i; + } + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } +} + +#endif diff --git a/intl/icu/source/common/ucnv_cnv.h b/intl/icu/source/common/ucnv_cnv.h new file mode 100644 index 0000000000..e89eebe54e --- /dev/null +++ b/intl/icu/source/common/ucnv_cnv.h @@ -0,0 +1,323 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1999-2011, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* +* ucnv_cnv.h: +* Definitions for converter implementations. +* +* Modification History: +* +* Date Name Description +* 05/09/00 helena Added implementation to handle fallback mappings. +* 06/29/2000 helena Major rewrite of the callback APIs. +*/ + +#ifndef UCNV_CNV_H +#define UCNV_CNV_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/ucnv_err.h" +#include "unicode/uset.h" +#include "uset_imp.h" + +U_CDECL_BEGIN + +/* this is used in fromUnicode DBCS tables as an "unassigned" marker */ +#define missingCharMarker 0xFFFF + +/* + * #define missingUCharMarker 0xfffe + * + * commented out because there are actually two values used in toUnicode tables: + * U+fffe "unassigned" + * U+ffff "illegal" + */ + +/** Forward declaration, see ucnv_bld.h */ +struct UConverterSharedData; +typedef struct UConverterSharedData UConverterSharedData; + +/* function types for UConverterImpl ---------------------------------------- */ + +/* struct with arguments for UConverterLoad and ucnv_load() */ +typedef struct { + int32_t size; /* sizeof(UConverterLoadArgs) */ + int32_t nestedLoads; /* count nested ucnv_load() calls */ + UBool onlyTestIsLoadable; /* input: don't actually load */ + UBool reserved0; /* reserved - for good alignment of the pointers */ + int16_t reserved; /* reserved - for good alignment of the pointers */ + uint32_t options; + const char *pkg, *name, *locale; +} UConverterLoadArgs; + +#define UCNV_LOAD_ARGS_INITIALIZER \ + { (int32_t)sizeof(UConverterLoadArgs), 0, false, false, 0, 0, NULL, NULL, NULL } + +typedef void (*UConverterLoad) (UConverterSharedData *sharedData, + UConverterLoadArgs *pArgs, + const uint8_t *raw, UErrorCode *pErrorCode); +typedef void (*UConverterUnload) (UConverterSharedData *sharedData); + +typedef void (*UConverterOpen) (UConverter *cnv, UConverterLoadArgs *pArgs, UErrorCode *pErrorCode); +typedef void (*UConverterClose) (UConverter *cnv); + +typedef enum UConverterResetChoice { + UCNV_RESET_BOTH, + UCNV_RESET_TO_UNICODE, + UCNV_RESET_FROM_UNICODE +} UConverterResetChoice; + +typedef void (*UConverterReset) (UConverter *cnv, UConverterResetChoice choice); + +/* + * Converter implementation function(s) for ucnv_toUnicode(). + * If the toUnicodeWithOffsets function pointer is NULL, + * then the toUnicode function will be used and the offsets will be set to -1. + * + * Must maintain state across buffers. Use toUBytes[toULength] for partial input + * sequences; it will be checked in ucnv.c at the end of the input stream + * to detect truncated input. + * Some converters may need additional detection and may then set U_TRUNCATED_CHAR_FOUND. + * + * The toUnicodeWithOffsets must write exactly as many offset values as target + * units. Write offset values of -1 for when the source index corresponding to + * the output unit is not known (e.g., the character started in an earlier buffer). + * The pArgs->offsets pointer need not be moved forward. + * + * At function return, either one of the following conditions must be true: + * - U_BUFFER_OVERFLOW_ERROR and the target is full: target==targetLimit + * - another error code with toUBytes[toULength] set to the offending input + * - no error, and the source is consumed: source==sourceLimit + * + * The ucnv.c code will handle the end of the input (reset) + * (reset, and truncation detection) and callbacks. + */ +typedef void (*UConverterToUnicode) (UConverterToUnicodeArgs *, UErrorCode *); + +/* + * Same rules as for UConverterToUnicode. + * A lead surrogate is kept in fromUChar32 across buffers, and if an error + * occurs, then the offending input code point must be put into fromUChar32 + * as well. + */ +typedef void (*UConverterFromUnicode) (UConverterFromUnicodeArgs *, UErrorCode *); + +/* + * Converter implementation function for ucnv_convertEx(), for direct conversion + * between two charsets without pivoting through UTF-16. + * The rules are the same as for UConverterToUnicode and UConverterFromUnicode. + * In addition, + * - The toUnicode side must behave and keep state exactly like the + * UConverterToUnicode implementation for the same source charset. + * - A U_USING_DEFAULT_WARNING can be set to request to temporarily fall back + * to pivoting. When this function is called, the conversion framework makes + * sure that this warning is not set on input. + * - Continuing a partial match and flushing the toUnicode replay buffer + * are handled by pivoting, using the toUnicode and fromUnicode functions. + */ +typedef void (*UConverterConvert) (UConverterFromUnicodeArgs *pFromUArgs, + UConverterToUnicodeArgs *pToUArgs, + UErrorCode *pErrorCode); + +/* + * Converter implementation function for ucnv_getNextUChar(). + * If the function pointer is NULL, then the toUnicode function will be used. + * + * Will be called at a character boundary (toULength==0). + * May return with + * - U_INDEX_OUTOFBOUNDS_ERROR if there was no output for the input + * (the return value will be ignored) + * - U_TRUNCATED_CHAR_FOUND or another error code (never U_BUFFER_OVERFLOW_ERROR!) + * with toUBytes[toULength] set to the offending input + * (the return value will be ignored) + * - return UCNV_GET_NEXT_UCHAR_USE_TO_U, without moving the source pointer, + * to indicate that the ucnv.c code shall call the toUnicode function instead + * - return a real code point result + * + * Unless UCNV_GET_NEXT_UCHAR_USE_TO_U is returned, the source bytes must be consumed. + * + * The ucnv.c code will handle the end of the input (reset) + * (except for truncation detection!) and callbacks. + */ +typedef UChar32 (*UConverterGetNextUChar) (UConverterToUnicodeArgs *, UErrorCode *); + +typedef void (*UConverterGetStarters)(const UConverter* converter, + UBool starters[256], + UErrorCode *pErrorCode); + +/* If this function pointer is null or if the function returns null + * the name field in static data struct should be returned by + * ucnv_getName() API function + */ +typedef const char * (*UConverterGetName) (const UConverter *cnv); + +/** + * Write the codepage substitution character. + * If this function is not set, then ucnv_cbFromUWriteSub() writes + * the substitution character from UConverter. + * For stateful converters, it is typically necessary to handle this + * specifically for the converter in order to properly maintain the state. + */ +typedef void (*UConverterWriteSub) (UConverterFromUnicodeArgs *pArgs, int32_t offsetIndex, UErrorCode *pErrorCode); + +/** + * For converter-specific safeClone processing + * If this function is not set, then ucnv_safeClone assumes that the converter has no private data that changes + * after the converter is done opening. + * If this function is set, then it is called just after a memcpy() of + * converter data to the new, empty converter, and is expected to set up + * the initial state of the converter. It is not expected to increment the + * reference counts of the standard data types such as the shared data. + */ +typedef UConverter * (*UConverterSafeClone) (const UConverter *cnv, + void *stackBuffer, + int32_t *pBufferSize, + UErrorCode *status); + +/** + * Filters for some ucnv_getUnicodeSet() implementation code. + */ +typedef enum UConverterSetFilter { + UCNV_SET_FILTER_NONE, + UCNV_SET_FILTER_DBCS_ONLY, + UCNV_SET_FILTER_2022_CN, + UCNV_SET_FILTER_SJIS, + UCNV_SET_FILTER_GR94DBCS, + UCNV_SET_FILTER_HZ, + UCNV_SET_FILTER_COUNT +} UConverterSetFilter; + +/** + * Fills the set of Unicode code points that can be converted by an ICU converter. + * The API function ucnv_getUnicodeSet() clears the USet before calling + * the converter's getUnicodeSet() implementation; the converter should only + * add the appropriate code points to allow recursive use. + * For example, the ISO-2022-JP converter will call each subconverter's + * getUnicodeSet() implementation to consecutively add code points to + * the same USet, which will result in a union of the sets of all subconverters. + * + * For more documentation, see ucnv_getUnicodeSet() in ucnv.h. + */ +typedef void (*UConverterGetUnicodeSet) (const UConverter *cnv, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode); + +UBool CONVERSION_U_SUCCESS (UErrorCode err); + +/** + * UConverterImpl contains all the data and functions for a converter type. + * Its function pointers work much like a C++ vtable. + * Many converter types need to define only a subset of the functions; + * when a function pointer is NULL, then a default action will be performed. + * + * Every converter type must implement toUnicode, fromUnicode, and getNextUChar, + * otherwise the converter may crash. + * Every converter type that has variable-length codepage sequences should + * also implement toUnicodeWithOffsets and fromUnicodeWithOffsets for + * correct offset handling. + * All other functions may or may not be implemented - it depends only on + * whether the converter type needs them. + * + * When open() fails, then close() will be called, if present. + */ +struct UConverterImpl { + UConverterType type; + + UConverterLoad load; + UConverterUnload unload; + + UConverterOpen open; + UConverterClose close; + UConverterReset reset; + + UConverterToUnicode toUnicode; + UConverterToUnicode toUnicodeWithOffsets; + UConverterFromUnicode fromUnicode; + UConverterFromUnicode fromUnicodeWithOffsets; + UConverterGetNextUChar getNextUChar; + + UConverterGetStarters getStarters; + UConverterGetName getName; + UConverterWriteSub writeSub; + UConverterSafeClone safeClone; + UConverterGetUnicodeSet getUnicodeSet; + + UConverterConvert toUTF8; + UConverterConvert fromUTF8; +}; + +extern const UConverterSharedData + _MBCSData, _Latin1Data, + _UTF8Data, _UTF16BEData, _UTF16LEData, _UTF32BEData, _UTF32LEData, + _ISO2022Data, + _LMBCSData1,_LMBCSData2, _LMBCSData3, _LMBCSData4, _LMBCSData5, _LMBCSData6, + _LMBCSData8,_LMBCSData11,_LMBCSData16,_LMBCSData17,_LMBCSData18,_LMBCSData19, + _HZData,_ISCIIData, _SCSUData, _ASCIIData, + _UTF7Data, _Bocu1Data, _UTF16Data, _UTF32Data, _CESU8Data, _IMAPData, _CompoundTextData; + +U_CDECL_END + +/** Always use fallbacks from codepage to Unicode */ +#define TO_U_USE_FALLBACK(useFallback) true +#define UCNV_TO_U_USE_FALLBACK(cnv) true + +/** Use fallbacks from Unicode to codepage when cnv->useFallback or for private-use code points */ +#define IS_PRIVATE_USE(c) ((uint32_t)((c)-0xe000)<0x1900 || (uint32_t)((c)-0xf0000)<0x20000) +#define FROM_U_USE_FALLBACK(useFallback, c) ((useFallback) || IS_PRIVATE_USE(c)) +#define UCNV_FROM_U_USE_FALLBACK(cnv, c) FROM_U_USE_FALLBACK((cnv)->useFallback, c) + +/** + * Magic number for ucnv_getNextUChar(), returned by a + * getNextUChar() implementation to indicate to use the converter's toUnicode() + * instead of the native function. + * @internal + */ +#define UCNV_GET_NEXT_UCHAR_USE_TO_U -9 + +U_CFUNC void +ucnv_getCompleteUnicodeSet(const UConverter *cnv, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode); + +U_CFUNC void +ucnv_getNonSurrogateUnicodeSet(const UConverter *cnv, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode); + +U_CFUNC void +ucnv_fromUWriteBytes(UConverter *cnv, + const char *bytes, int32_t length, + char **target, const char *targetLimit, + int32_t **offsets, + int32_t sourceIndex, + UErrorCode *pErrorCode); +U_CFUNC void +ucnv_toUWriteUChars(UConverter *cnv, + const UChar *uchars, int32_t length, + UChar **target, const UChar *targetLimit, + int32_t **offsets, + int32_t sourceIndex, + UErrorCode *pErrorCode); + +U_CFUNC void +ucnv_toUWriteCodePoint(UConverter *cnv, + UChar32 c, + UChar **target, const UChar *targetLimit, + int32_t **offsets, + int32_t sourceIndex, + UErrorCode *pErrorCode); + +#endif + +#endif /* UCNV_CNV */ diff --git a/intl/icu/source/common/ucnv_ct.cpp b/intl/icu/source/common/ucnv_ct.cpp new file mode 100644 index 0000000000..46f30e5ece --- /dev/null +++ b/intl/icu/source/common/ucnv_ct.cpp @@ -0,0 +1,646 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2010-2015, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* file name: ucnv_ct.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010Dec09 +* created by: Michael Ow +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION && !UCONFIG_NO_LEGACY_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/uset.h" +#include "unicode/ucnv_err.h" +#include "unicode/ucnv_cb.h" +#include "unicode/utf16.h" +#include "ucnv_imp.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" +#include "ucnvmbcs.h" +#include "cstring.h" +#include "cmemory.h" + +typedef enum { + INVALID = -2, + DO_SEARCH = -1, + + COMPOUND_TEXT_SINGLE_0 = 0, + COMPOUND_TEXT_SINGLE_1 = 1, + COMPOUND_TEXT_SINGLE_2 = 2, + COMPOUND_TEXT_SINGLE_3 = 3, + + COMPOUND_TEXT_DOUBLE_1 = 4, + COMPOUND_TEXT_DOUBLE_2 = 5, + COMPOUND_TEXT_DOUBLE_3 = 6, + COMPOUND_TEXT_DOUBLE_4 = 7, + COMPOUND_TEXT_DOUBLE_5 = 8, + COMPOUND_TEXT_DOUBLE_6 = 9, + COMPOUND_TEXT_DOUBLE_7 = 10, + + COMPOUND_TEXT_TRIPLE_DOUBLE = 11, + + IBM_915 = 12, + IBM_916 = 13, + IBM_914 = 14, + IBM_874 = 15, + IBM_912 = 16, + IBM_913 = 17, + ISO_8859_14 = 18, + IBM_923 = 19, + NUM_OF_CONVERTERS = 20 +} COMPOUND_TEXT_CONVERTERS; + +#define SEARCH_LENGTH 12 + +static const uint8_t escSeqCompoundText[NUM_OF_CONVERTERS][5] = { + /* Single */ + { 0x1B, 0x2D, 0x41, 0, 0 }, + { 0x1B, 0x2D, 0x4D, 0, 0 }, + { 0x1B, 0x2D, 0x46, 0, 0 }, + { 0x1B, 0x2D, 0x47, 0, 0 }, + + /* Double */ + { 0x1B, 0x24, 0x29, 0x41, 0 }, + { 0x1B, 0x24, 0x29, 0x42, 0 }, + { 0x1B, 0x24, 0x29, 0x43, 0 }, + { 0x1B, 0x24, 0x29, 0x44, 0 }, + { 0x1B, 0x24, 0x29, 0x47, 0 }, + { 0x1B, 0x24, 0x29, 0x48, 0 }, + { 0x1B, 0x24, 0x29, 0x49, 0 }, + + /* Triple/Double */ + { 0x1B, 0x25, 0x47, 0, 0 }, + + /*IBM-915*/ + { 0x1B, 0x2D, 0x4C, 0, 0 }, + /*IBM-916*/ + { 0x1B, 0x2D, 0x48, 0, 0 }, + /*IBM-914*/ + { 0x1B, 0x2D, 0x44, 0, 0 }, + /*IBM-874*/ + { 0x1B, 0x2D, 0x54, 0, 0 }, + /*IBM-912*/ + { 0x1B, 0x2D, 0x42, 0, 0 }, + /* IBM-913 */ + { 0x1B, 0x2D, 0x43, 0, 0 }, + /* ISO-8859_14 */ + { 0x1B, 0x2D, 0x5F, 0, 0 }, + /* IBM-923 */ + { 0x1B, 0x2D, 0x62, 0, 0 }, +}; + +#define ESC_START 0x1B + +#define isASCIIRange(codepoint) \ + ((codepoint == 0x0000) || (codepoint == 0x0009) || (codepoint == 0x000A) || \ + (codepoint >= 0x0020 && codepoint <= 0x007f) || (codepoint >= 0x00A0 && codepoint <= 0x00FF)) + +#define isIBM915(codepoint) \ + ((codepoint >= 0x0401 && codepoint <= 0x045F) || (codepoint == 0x2116)) + +#define isIBM916(codepoint) \ + ((codepoint >= 0x05D0 && codepoint <= 0x05EA) || (codepoint == 0x2017) || (codepoint == 0x203E)) + +#define isCompoundS3(codepoint) \ + ((codepoint == 0x060C) || (codepoint == 0x061B) || (codepoint == 0x061F) || (codepoint >= 0x0621 && codepoint <= 0x063A) || \ + (codepoint >= 0x0640 && codepoint <= 0x0652) || (codepoint >= 0x0660 && codepoint <= 0x066D) || (codepoint == 0x200B) || \ + (codepoint >= 0x0FE70 && codepoint <= 0x0FE72) || (codepoint == 0x0FE74) || (codepoint >= 0x0FE76 && codepoint <= 0x0FEBE)) + +#define isCompoundS2(codepoint) \ + ((codepoint == 0x02BC) || (codepoint == 0x02BD) || (codepoint >= 0x0384 && codepoint <= 0x03CE) || (codepoint == 0x2015)) + +#define isIBM914(codepoint) \ + ((codepoint == 0x0100) || (codepoint == 0x0101) || (codepoint == 0x0112) || (codepoint == 0x0113) || (codepoint == 0x0116) || (codepoint == 0x0117) || \ + (codepoint == 0x0122) || (codepoint == 0x0123) || (codepoint >= 0x0128 && codepoint <= 0x012B) || (codepoint == 0x012E) || (codepoint == 0x012F) || \ + (codepoint >= 0x0136 && codepoint <= 0x0138) || (codepoint == 0x013B) || (codepoint == 0x013C) || (codepoint == 0x0145) || (codepoint == 0x0146) || \ + (codepoint >= 0x014A && codepoint <= 0x014D) || (codepoint == 0x0156) || (codepoint == 0x0157) || (codepoint >= 0x0166 && codepoint <= 0x016B) || \ + (codepoint == 0x0172) || (codepoint == 0x0173)) + +#define isIBM874(codepoint) \ + ((codepoint >= 0x0E01 && codepoint <= 0x0E3A) || (codepoint >= 0x0E3F && codepoint <= 0x0E5B)) + +#define isIBM912(codepoint) \ + ((codepoint >= 0x0102 && codepoint <= 0x0107) || (codepoint >= 0x010C && codepoint <= 0x0111) || (codepoint >= 0x0118 && codepoint <= 0x011B) || \ + (codepoint == 0x0139) || (codepoint == 0x013A) || (codepoint == 0x013D) || (codepoint == 0x013E) || (codepoint >= 0x0141 && codepoint <= 0x0144) || \ + (codepoint == 0x0147) || (codepoint == 0x0147) || (codepoint == 0x0150) || (codepoint == 0x0151) || (codepoint == 0x0154) || (codepoint == 0x0155) || \ + (codepoint >= 0x0158 && codepoint <= 0x015B) || (codepoint == 0x015E) || (codepoint == 0x015F) || (codepoint >= 0x0160 && codepoint <= 0x0165) || \ + (codepoint == 0x016E) || (codepoint == 0x016F) || (codepoint == 0x0170) || (codepoint == 0x0171) || (codepoint >= 0x0179 && codepoint <= 0x017E) || \ + (codepoint == 0x02C7) || (codepoint == 0x02D8) || (codepoint == 0x02D9) || (codepoint == 0x02DB) || (codepoint == 0x02DD)) + +#define isIBM913(codepoint) \ + ((codepoint >= 0x0108 && codepoint <= 0x010B) || (codepoint == 0x011C) || \ + (codepoint == 0x011D) || (codepoint == 0x0120) || (codepoint == 0x0121) || \ + (codepoint >= 0x0124 && codepoint <= 0x0127) || (codepoint == 0x0134) || (codepoint == 0x0135) || \ + (codepoint == 0x015C) || (codepoint == 0x015D) || (codepoint == 0x016C) || (codepoint == 0x016D)) + +#define isCompoundS1(codepoint) \ + ((codepoint == 0x011E) || (codepoint == 0x011F) || (codepoint == 0x0130) || \ + (codepoint == 0x0131) || (codepoint >= 0x0218 && codepoint <= 0x021B)) + +#define isISO8859_14(codepoint) \ + ((codepoint >= 0x0174 && codepoint <= 0x0177) || (codepoint == 0x1E0A) || \ + (codepoint == 0x1E0B) || (codepoint == 0x1E1E) || (codepoint == 0x1E1F) || \ + (codepoint == 0x1E40) || (codepoint == 0x1E41) || (codepoint == 0x1E56) || \ + (codepoint == 0x1E57) || (codepoint == 0x1E60) || (codepoint == 0x1E61) || \ + (codepoint == 0x1E6A) || (codepoint == 0x1E6B) || (codepoint == 0x1EF2) || \ + (codepoint == 0x1EF3) || (codepoint >= 0x1E80 && codepoint <= 0x1E85)) + +#define isIBM923(codepoint) \ + ((codepoint == 0x0152) || (codepoint == 0x0153) || (codepoint == 0x0178) || (codepoint == 0x20AC)) + + +typedef struct{ + UConverterSharedData *myConverterArray[NUM_OF_CONVERTERS]; + COMPOUND_TEXT_CONVERTERS state; +} UConverterDataCompoundText; + +/*********** Compound Text Converter Protos ***********/ +U_CDECL_BEGIN +static void U_CALLCONV +_CompoundTextOpen(UConverter *cnv, UConverterLoadArgs *pArgs, UErrorCode *errorCode); + +static void U_CALLCONV + _CompoundTextClose(UConverter *converter); + +static void U_CALLCONV +_CompoundTextReset(UConverter *converter, UConverterResetChoice choice); + +static const char* U_CALLCONV +_CompoundTextgetName(const UConverter* cnv); + + +static int32_t findNextEsc(const char *source, const char *sourceLimit) { + int32_t length = static_cast(sourceLimit - source); + int32_t i; + for (i = 1; i < length; i++) { + if (*(source + i) == 0x1B) { + return i; + } + } + + return length; +} + +static COMPOUND_TEXT_CONVERTERS getState(int codepoint) { + COMPOUND_TEXT_CONVERTERS state = DO_SEARCH; + + if (isASCIIRange(codepoint)) { + state = COMPOUND_TEXT_SINGLE_0; + } else if (isIBM912(codepoint)) { + state = IBM_912; + }else if (isIBM913(codepoint)) { + state = IBM_913; + } else if (isISO8859_14(codepoint)) { + state = ISO_8859_14; + } else if (isIBM923(codepoint)) { + state = IBM_923; + } else if (isIBM874(codepoint)) { + state = IBM_874; + } else if (isIBM914(codepoint)) { + state = IBM_914; + } else if (isCompoundS2(codepoint)) { + state = COMPOUND_TEXT_SINGLE_2; + } else if (isCompoundS3(codepoint)) { + state = COMPOUND_TEXT_SINGLE_3; + } else if (isIBM916(codepoint)) { + state = IBM_916; + } else if (isIBM915(codepoint)) { + state = IBM_915; + } else if (isCompoundS1(codepoint)) { + state = COMPOUND_TEXT_SINGLE_1; + } + + return state; +} + +static COMPOUND_TEXT_CONVERTERS findStateFromEscSeq(const char* source, const char* sourceLimit, const uint8_t* toUBytesBuffer, int32_t toUBytesBufferLength, UErrorCode *err) { + COMPOUND_TEXT_CONVERTERS state = INVALID; + UBool matchFound = false; + int32_t i, n, offset = toUBytesBufferLength; + + for (i = 0; i < NUM_OF_CONVERTERS; i++) { + matchFound = true; + for (n = 0; escSeqCompoundText[i][n] != 0; n++) { + if (n < toUBytesBufferLength) { + if (toUBytesBuffer[n] != escSeqCompoundText[i][n]) { + matchFound = false; + break; + } + } else if ((source + (n - offset)) >= sourceLimit) { + *err = U_TRUNCATED_CHAR_FOUND; + matchFound = false; + break; + } else if (*(source + (n - offset)) != escSeqCompoundText[i][n]) { + matchFound = false; + break; + } + } + + if (matchFound) { + break; + } + } + + if (matchFound) { + state = (COMPOUND_TEXT_CONVERTERS)i; + } + + return state; +} + +static void U_CALLCONV +_CompoundTextOpen(UConverter *cnv, UConverterLoadArgs *pArgs, UErrorCode *errorCode){ + cnv->extraInfo = uprv_malloc (sizeof (UConverterDataCompoundText)); + if (cnv->extraInfo != nullptr) { + UConverterDataCompoundText *myConverterData = (UConverterDataCompoundText *) cnv->extraInfo; + + UConverterNamePieces stackPieces; + UConverterLoadArgs stackArgs=UCNV_LOAD_ARGS_INITIALIZER; + + myConverterData->myConverterArray[COMPOUND_TEXT_SINGLE_0] = nullptr; + myConverterData->myConverterArray[COMPOUND_TEXT_SINGLE_1] = ucnv_loadSharedData("icu-internal-compound-s1", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[COMPOUND_TEXT_SINGLE_2] = ucnv_loadSharedData("icu-internal-compound-s2", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[COMPOUND_TEXT_SINGLE_3] = ucnv_loadSharedData("icu-internal-compound-s3", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[COMPOUND_TEXT_DOUBLE_1] = ucnv_loadSharedData("icu-internal-compound-d1", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[COMPOUND_TEXT_DOUBLE_2] = ucnv_loadSharedData("icu-internal-compound-d2", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[COMPOUND_TEXT_DOUBLE_3] = ucnv_loadSharedData("icu-internal-compound-d3", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[COMPOUND_TEXT_DOUBLE_4] = ucnv_loadSharedData("icu-internal-compound-d4", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[COMPOUND_TEXT_DOUBLE_5] = ucnv_loadSharedData("icu-internal-compound-d5", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[COMPOUND_TEXT_DOUBLE_6] = ucnv_loadSharedData("icu-internal-compound-d6", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[COMPOUND_TEXT_DOUBLE_7] = ucnv_loadSharedData("icu-internal-compound-d7", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[COMPOUND_TEXT_TRIPLE_DOUBLE] = ucnv_loadSharedData("icu-internal-compound-t", &stackPieces, &stackArgs, errorCode); + + myConverterData->myConverterArray[IBM_915] = ucnv_loadSharedData("ibm-915_P100-1995", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[IBM_916] = ucnv_loadSharedData("ibm-916_P100-1995", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[IBM_914] = ucnv_loadSharedData("ibm-914_P100-1995", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[IBM_874] = ucnv_loadSharedData("ibm-874_P100-1995", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[IBM_912] = ucnv_loadSharedData("ibm-912_P100-1995", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[IBM_913] = ucnv_loadSharedData("ibm-913_P100-2000", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[ISO_8859_14] = ucnv_loadSharedData("iso-8859_14-1998", &stackPieces, &stackArgs, errorCode); + myConverterData->myConverterArray[IBM_923] = ucnv_loadSharedData("ibm-923_P100-1998", &stackPieces, &stackArgs, errorCode); + + if (U_FAILURE(*errorCode) || pArgs->onlyTestIsLoadable) { + _CompoundTextClose(cnv); + return; + } + + myConverterData->state = (COMPOUND_TEXT_CONVERTERS)0; + } else { + *errorCode = U_MEMORY_ALLOCATION_ERROR; + } +} + + +static void U_CALLCONV +_CompoundTextClose(UConverter *converter) { + UConverterDataCompoundText* myConverterData = (UConverterDataCompoundText*)(converter->extraInfo); + int32_t i; + + if (converter->extraInfo != nullptr) { + /*close the array of converter pointers and free the memory*/ + for (i = 0; i < NUM_OF_CONVERTERS; i++) { + if (myConverterData->myConverterArray[i] != nullptr) { + ucnv_unloadSharedDataIfReady(myConverterData->myConverterArray[i]); + } + } + + uprv_free(converter->extraInfo); + converter->extraInfo = nullptr; + } +} + +static void U_CALLCONV +_CompoundTextReset(UConverter *converter, UConverterResetChoice choice) { + (void)converter; + (void)choice; +} + +static const char* U_CALLCONV +_CompoundTextgetName(const UConverter* cnv){ + (void)cnv; + return "x11-compound-text"; +} + +static void U_CALLCONV +UConverter_fromUnicode_CompoundText_OFFSETS(UConverterFromUnicodeArgs* args, UErrorCode* err){ + UConverter *cnv = args->converter; + uint8_t *target = (uint8_t *) args->target; + const uint8_t *targetLimit = (const uint8_t *) args->targetLimit; + const char16_t* source = args->source; + const char16_t* sourceLimit = args->sourceLimit; + /* int32_t* offsets = args->offsets; */ + UChar32 sourceChar; + UBool useFallback = cnv->useFallback; + uint8_t tmpTargetBuffer[7]; + int32_t tmpTargetBufferLength = 0; + COMPOUND_TEXT_CONVERTERS currentState, tmpState; + uint32_t pValue; + int32_t pValueLength = 0; + int32_t i, n, j; + + UConverterDataCompoundText *myConverterData = (UConverterDataCompoundText *) cnv->extraInfo; + + currentState = myConverterData->state; + + /* check if the last codepoint of previous buffer was a lead surrogate*/ + if((sourceChar = cnv->fromUChar32)!=0 && target< targetLimit) { + goto getTrail; + } + + while( source < sourceLimit){ + if(target < targetLimit){ + + sourceChar = *(source++); + /*check if the char is a First surrogate*/ + if(U16_IS_SURROGATE(sourceChar)) { + if(U16_IS_SURROGATE_LEAD(sourceChar)) { +getTrail: + /*look ahead to find the trail surrogate*/ + if(source < sourceLimit) { + /* test the following code unit */ + char16_t trail=(char16_t) *source; + if(U16_IS_TRAIL(trail)) { + source++; + sourceChar=U16_GET_SUPPLEMENTARY(sourceChar, trail); + cnv->fromUChar32=0x00; + /* convert this supplementary code point */ + /* exit this condition tree */ + } else { + /* this is an unmatched lead code unit (1st surrogate) */ + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + cnv->fromUChar32=sourceChar; + break; + } + } else { + /* no more input */ + cnv->fromUChar32=sourceChar; + break; + } + } else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + cnv->fromUChar32=sourceChar; + break; + } + } + + tmpTargetBufferLength = 0; + tmpState = getState(sourceChar); + + if (tmpState != DO_SEARCH && currentState != tmpState) { + /* Get escape sequence if necessary */ + currentState = tmpState; + for (i = 0; escSeqCompoundText[currentState][i] != 0; i++) { + tmpTargetBuffer[tmpTargetBufferLength++] = escSeqCompoundText[currentState][i]; + } + } + + if (tmpState == DO_SEARCH) { + /* Test all available converters */ + for (i = 1; i < SEARCH_LENGTH; i++) { + pValueLength = ucnv_MBCSFromUChar32(myConverterData->myConverterArray[i], sourceChar, &pValue, useFallback); + if (pValueLength > 0) { + tmpState = (COMPOUND_TEXT_CONVERTERS)i; + if (currentState != tmpState) { + currentState = tmpState; + for (j = 0; escSeqCompoundText[currentState][j] != 0; j++) { + tmpTargetBuffer[tmpTargetBufferLength++] = escSeqCompoundText[currentState][j]; + } + } + for (n = (pValueLength - 1); n >= 0; n--) { + tmpTargetBuffer[tmpTargetBufferLength++] = (uint8_t)(pValue >> (n * 8)); + } + break; + } + } + } else if (tmpState == COMPOUND_TEXT_SINGLE_0) { + tmpTargetBuffer[tmpTargetBufferLength++] = (uint8_t)sourceChar; + } else { + pValueLength = ucnv_MBCSFromUChar32(myConverterData->myConverterArray[currentState], sourceChar, &pValue, useFallback); + if (pValueLength > 0) { + for (n = (pValueLength - 1); n >= 0; n--) { + tmpTargetBuffer[tmpTargetBufferLength++] = (uint8_t)(pValue >> (n * 8)); + } + } + } + + for (i = 0; i < tmpTargetBufferLength; i++) { + if (target < targetLimit) { + *target++ = tmpTargetBuffer[i]; + } else { + *err = U_BUFFER_OVERFLOW_ERROR; + break; + } + } + + if (*err == U_BUFFER_OVERFLOW_ERROR) { + for (; i < tmpTargetBufferLength; i++) { + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = tmpTargetBuffer[i]; + } + } + } else { + *err = U_BUFFER_OVERFLOW_ERROR; + break; + } + } + + /*save the state and return */ + myConverterData->state = currentState; + args->source = source; + args->target = (char*)target; +} + + +static void U_CALLCONV +UConverter_toUnicode_CompoundText_OFFSETS(UConverterToUnicodeArgs *args, + UErrorCode* err){ + const char *mySource = (char *) args->source; + char16_t *myTarget = args->target; + const char *mySourceLimit = args->sourceLimit; + const char *tmpSourceLimit = mySourceLimit; + uint32_t mySourceChar = 0x0000; + COMPOUND_TEXT_CONVERTERS currentState, tmpState; + int32_t sourceOffset = 0; + UConverterDataCompoundText *myConverterData = (UConverterDataCompoundText *) args->converter->extraInfo; + UConverterSharedData* savedSharedData = nullptr; + + UConverterToUnicodeArgs subArgs; + int32_t minArgsSize; + + /* set up the subconverter arguments */ + if(args->sizesize; + } else { + minArgsSize = (int32_t)sizeof(UConverterToUnicodeArgs); + } + + uprv_memcpy(&subArgs, args, minArgsSize); + subArgs.size = (uint16_t)minArgsSize; + + currentState = tmpState = myConverterData->state; + + while(mySource < mySourceLimit){ + if(myTarget < args->targetLimit){ + if (args->converter->toULength > 0) { + mySourceChar = args->converter->toUBytes[0]; + } else { + mySourceChar = (uint8_t)*mySource; + } + + if (mySourceChar == ESC_START) { + tmpState = findStateFromEscSeq(mySource, mySourceLimit, args->converter->toUBytes, args->converter->toULength, err); + + if (*err == U_TRUNCATED_CHAR_FOUND) { + for (; mySource < mySourceLimit;) { + args->converter->toUBytes[args->converter->toULength++] = *mySource++; + } + *err = U_ZERO_ERROR; + break; + } else if (tmpState == INVALID) { + if (args->converter->toULength == 0) { + mySource++; /* skip over the 0x1b byte */ + } + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + + if (tmpState != currentState) { + currentState = tmpState; + } + + sourceOffset = static_cast(uprv_strlen((char*)escSeqCompoundText[currentState]) - args->converter->toULength); + + mySource += sourceOffset; + + args->converter->toULength = 0; + } + + if (currentState == COMPOUND_TEXT_SINGLE_0) { + while (mySource < mySourceLimit) { + if (*mySource == ESC_START) { + break; + } + if (myTarget < args->targetLimit) { + *myTarget++ = 0x00ff&(*mySource++); + } else { + *err = U_BUFFER_OVERFLOW_ERROR; + break; + } + } + } else if (mySource < mySourceLimit){ + sourceOffset = findNextEsc(mySource, mySourceLimit); + + tmpSourceLimit = mySource + sourceOffset; + + subArgs.source = mySource; + subArgs.sourceLimit = tmpSourceLimit; + subArgs.target = myTarget; + savedSharedData = subArgs.converter->sharedData; + subArgs.converter->sharedData = myConverterData->myConverterArray[currentState]; + + ucnv_MBCSToUnicodeWithOffsets(&subArgs, err); + + subArgs.converter->sharedData = savedSharedData; + + mySource = subArgs.source; + myTarget = subArgs.target; + + if (U_FAILURE(*err)) { + if(*err == U_BUFFER_OVERFLOW_ERROR) { + if(subArgs.converter->UCharErrorBufferLength > 0) { + uprv_memcpy(args->converter->UCharErrorBuffer, subArgs.converter->UCharErrorBuffer, + subArgs.converter->UCharErrorBufferLength); + } + args->converter->UCharErrorBufferLength=subArgs.converter->UCharErrorBufferLength; + subArgs.converter->UCharErrorBufferLength = 0; + } + break; + } + } + } else { + *err = U_BUFFER_OVERFLOW_ERROR; + break; + } + } + myConverterData->state = currentState; + args->target = myTarget; + args->source = mySource; +} + +static void U_CALLCONV +_CompoundText_GetUnicodeSet(const UConverter *cnv, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode) { + UConverterDataCompoundText *myConverterData = (UConverterDataCompoundText *)cnv->extraInfo; + int32_t i; + + for (i = 1; i < NUM_OF_CONVERTERS; i++) { + ucnv_MBCSGetUnicodeSetForUnicode(myConverterData->myConverterArray[i], sa, which, pErrorCode); + } + sa->add(sa->set, 0x0000); + sa->add(sa->set, 0x0009); + sa->add(sa->set, 0x000A); + sa->addRange(sa->set, 0x0020, 0x007F); + sa->addRange(sa->set, 0x00A0, 0x00FF); +} +U_CDECL_END + +static const UConverterImpl _CompoundTextImpl = { + + UCNV_COMPOUND_TEXT, + + nullptr, + nullptr, + + _CompoundTextOpen, + _CompoundTextClose, + _CompoundTextReset, + + UConverter_toUnicode_CompoundText_OFFSETS, + UConverter_toUnicode_CompoundText_OFFSETS, + UConverter_fromUnicode_CompoundText_OFFSETS, + UConverter_fromUnicode_CompoundText_OFFSETS, + nullptr, + + nullptr, + _CompoundTextgetName, + nullptr, + nullptr, + _CompoundText_GetUnicodeSet, + nullptr, + nullptr +}; + +static const UConverterStaticData _CompoundTextStaticData = { + sizeof(UConverterStaticData), + "COMPOUND_TEXT", + 0, + UCNV_IBM, + UCNV_COMPOUND_TEXT, + 1, + 6, + { 0xef, 0, 0, 0 }, + 1, + false, + false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; +const UConverterSharedData _CompoundTextData = + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_CompoundTextStaticData, &_CompoundTextImpl); + +#endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */ diff --git a/intl/icu/source/common/ucnv_err.cpp b/intl/icu/source/common/ucnv_err.cpp new file mode 100644 index 0000000000..761c442632 --- /dev/null +++ b/intl/icu/source/common/ucnv_err.cpp @@ -0,0 +1,486 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ***************************************************************************** + * + * Copyright (C) 1998-2016, International Business Machines + * Corporation and others. All Rights Reserved. + * + ***************************************************************************** + * + * ucnv_err.c + * Implements error behaviour functions called by T_UConverter_{from,to}Unicode + * + * +* Change history: +* +* 06/29/2000 helena Major rewrite of the callback APIs. +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/ucnv_err.h" +#include "unicode/ucnv_cb.h" +#include "ucnv_cnv.h" +#include "cmemory.h" +#include "unicode/ucnv.h" +#include "ustrfmt.h" + +#define VALUE_STRING_LENGTH 48 +/*Magic # 32 = 4(number of char in value string) * 8(max number of bytes per char for any converter) */ +#define UNICODE_PERCENT_SIGN_CODEPOINT 0x0025 +#define UNICODE_U_CODEPOINT 0x0055 +#define UNICODE_X_CODEPOINT 0x0058 +#define UNICODE_RS_CODEPOINT 0x005C +#define UNICODE_U_LOW_CODEPOINT 0x0075 +#define UNICODE_X_LOW_CODEPOINT 0x0078 +#define UNICODE_AMP_CODEPOINT 0x0026 +#define UNICODE_HASH_CODEPOINT 0x0023 +#define UNICODE_SEMICOLON_CODEPOINT 0x003B +#define UNICODE_PLUS_CODEPOINT 0x002B +#define UNICODE_LEFT_CURLY_CODEPOINT 0x007B +#define UNICODE_RIGHT_CURLY_CODEPOINT 0x007D +#define UNICODE_SPACE_CODEPOINT 0x0020 +#define UCNV_PRV_ESCAPE_ICU 0 +#define UCNV_PRV_ESCAPE_C 'C' +#define UCNV_PRV_ESCAPE_XML_DEC 'D' +#define UCNV_PRV_ESCAPE_XML_HEX 'X' +#define UCNV_PRV_ESCAPE_JAVA 'J' +#define UCNV_PRV_ESCAPE_UNICODE 'U' +#define UCNV_PRV_ESCAPE_CSS2 'S' +#define UCNV_PRV_STOP_ON_ILLEGAL 'i' + +/* + * IS_DEFAULT_IGNORABLE_CODE_POINT + * This is to check if a code point has the default ignorable unicode property. + * As such, this list needs to be updated if the ignorable code point list ever + * changes. + * To avoid dependency on other code, this list is hard coded here. + * When an ignorable code point is found and is unmappable, the default callbacks + * will ignore them. + * For a list of the default ignorable code points, use this link: + * https://util.unicode.org/UnicodeJsps/list-unicodeset.jsp?a=%5B%3ADI%3A%5D&abb=on&g=&i= + * + * This list should be sync with the one in CharsetCallback.java + */ +#define IS_DEFAULT_IGNORABLE_CODE_POINT(c) ( \ + (c == 0x00AD) || \ + (c == 0x034F) || \ + (c == 0x061C) || \ + (c == 0x115F) || \ + (c == 0x1160) || \ + (0x17B4 <= c && c <= 0x17B5) || \ + (0x180B <= c && c <= 0x180F) || \ + (0x200B <= c && c <= 0x200F) || \ + (0x202A <= c && c <= 0x202E) || \ + (0x2060 <= c && c <= 0x206F) || \ + (c == 0x3164) || \ + (0xFE00 <= c && c <= 0xFE0F) || \ + (c == 0xFEFF) || \ + (c == 0xFFA0) || \ + (0xFFF0 <= c && c <= 0xFFF8) || \ + (0x1BCA0 <= c && c <= 0x1BCA3) || \ + (0x1D173 <= c && c <= 0x1D17A) || \ + (0xE0000 <= c && c <= 0xE0FFF)) + + +/*Function Pointer STOPS at the ILLEGAL_SEQUENCE */ +U_CAPI void U_EXPORT2 +UCNV_FROM_U_CALLBACK_STOP ( + const void *context, + UConverterFromUnicodeArgs *fromUArgs, + const char16_t* codeUnits, + int32_t length, + UChar32 codePoint, + UConverterCallbackReason reason, + UErrorCode * err) +{ + (void)context; + (void)fromUArgs; + (void)codeUnits; + (void)length; + if (reason == UCNV_UNASSIGNED && IS_DEFAULT_IGNORABLE_CODE_POINT(codePoint)) + { + /* + * Skip if the codepoint has unicode property of default ignorable. + */ + *err = U_ZERO_ERROR; + } + /* the caller must have set the error code accordingly */ + return; +} + + +/*Function Pointer STOPS at the ILLEGAL_SEQUENCE */ +U_CAPI void U_EXPORT2 +UCNV_TO_U_CALLBACK_STOP ( + const void *context, + UConverterToUnicodeArgs *toUArgs, + const char* codePoints, + int32_t length, + UConverterCallbackReason reason, + UErrorCode * err) +{ + /* the caller must have set the error code accordingly */ + (void)context; (void)toUArgs; (void)codePoints; (void)length; (void)reason; (void)err; + return; +} + +U_CAPI void U_EXPORT2 +UCNV_FROM_U_CALLBACK_SKIP ( + const void *context, + UConverterFromUnicodeArgs *fromUArgs, + const char16_t* codeUnits, + int32_t length, + UChar32 codePoint, + UConverterCallbackReason reason, + UErrorCode * err) +{ + (void)fromUArgs; + (void)codeUnits; + (void)length; + if (reason <= UCNV_IRREGULAR) + { + if (reason == UCNV_UNASSIGNED && IS_DEFAULT_IGNORABLE_CODE_POINT(codePoint)) + { + /* + * Skip if the codepoint has unicode property of default ignorable. + */ + *err = U_ZERO_ERROR; + } + else if (context == nullptr || (*((char*)context) == UCNV_PRV_STOP_ON_ILLEGAL && reason == UCNV_UNASSIGNED)) + { + *err = U_ZERO_ERROR; + } + /* else the caller must have set the error code accordingly. */ + } + /* else ignore the reset, close and clone calls. */ +} + +U_CAPI void U_EXPORT2 +UCNV_FROM_U_CALLBACK_SUBSTITUTE ( + const void *context, + UConverterFromUnicodeArgs *fromArgs, + const char16_t* codeUnits, + int32_t length, + UChar32 codePoint, + UConverterCallbackReason reason, + UErrorCode * err) +{ + (void)codeUnits; + (void)length; + if (reason <= UCNV_IRREGULAR) + { + if (reason == UCNV_UNASSIGNED && IS_DEFAULT_IGNORABLE_CODE_POINT(codePoint)) + { + /* + * Skip if the codepoint has unicode property of default ignorable. + */ + *err = U_ZERO_ERROR; + } + else if (context == nullptr || (*((char*)context) == UCNV_PRV_STOP_ON_ILLEGAL && reason == UCNV_UNASSIGNED)) + { + *err = U_ZERO_ERROR; + ucnv_cbFromUWriteSub(fromArgs, 0, err); + } + /* else the caller must have set the error code accordingly. */ + } + /* else ignore the reset, close and clone calls. */ +} + +/*uses uprv_itou to get a unicode escape sequence of the offensive sequence, + *uses a clean copy (resetted) of the converter, to convert that unicode + *escape sequence to the target codepage (if conversion failure happens then + *we revert to substituting with subchar) + */ +U_CAPI void U_EXPORT2 +UCNV_FROM_U_CALLBACK_ESCAPE ( + const void *context, + UConverterFromUnicodeArgs *fromArgs, + const char16_t *codeUnits, + int32_t length, + UChar32 codePoint, + UConverterCallbackReason reason, + UErrorCode * err) +{ + + char16_t valueString[VALUE_STRING_LENGTH]; + int32_t valueStringLength = 0; + int32_t i = 0; + + const char16_t *myValueSource = nullptr; + UErrorCode err2 = U_ZERO_ERROR; + UConverterFromUCallback original = nullptr; + const void *originalContext; + + UConverterFromUCallback ignoredCallback = nullptr; + const void *ignoredContext; + + if (reason > UCNV_IRREGULAR) + { + return; + } + else if (reason == UCNV_UNASSIGNED && IS_DEFAULT_IGNORABLE_CODE_POINT(codePoint)) + { + /* + * Skip if the codepoint has unicode property of default ignorable. + */ + *err = U_ZERO_ERROR; + return; + } + + ucnv_setFromUCallBack (fromArgs->converter, + (UConverterFromUCallback) UCNV_FROM_U_CALLBACK_SUBSTITUTE, + nullptr, + &original, + &originalContext, + &err2); + + if (U_FAILURE (err2)) + { + *err = err2; + return; + } + if(context==nullptr) + { + while (i < length) + { + valueString[valueStringLength++] = (char16_t) UNICODE_PERCENT_SIGN_CODEPOINT; /* adding % */ + valueString[valueStringLength++] = (char16_t) UNICODE_U_CODEPOINT; /* adding U */ + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[i++], 16, 4); + } + } + else + { + switch(*((char*)context)) + { + case UCNV_PRV_ESCAPE_JAVA: + while (i < length) + { + valueString[valueStringLength++] = (char16_t) UNICODE_RS_CODEPOINT; /* adding \ */ + valueString[valueStringLength++] = (char16_t) UNICODE_U_LOW_CODEPOINT; /* adding u */ + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[i++], 16, 4); + } + break; + + case UCNV_PRV_ESCAPE_C: + valueString[valueStringLength++] = (char16_t) UNICODE_RS_CODEPOINT; /* adding \ */ + + if(length==2){ + valueString[valueStringLength++] = (char16_t) UNICODE_U_CODEPOINT; /* adding U */ + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, codePoint, 16, 8); + + } + else{ + valueString[valueStringLength++] = (char16_t) UNICODE_U_LOW_CODEPOINT; /* adding u */ + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[0], 16, 4); + } + break; + + case UCNV_PRV_ESCAPE_XML_DEC: + + valueString[valueStringLength++] = (char16_t) UNICODE_AMP_CODEPOINT; /* adding & */ + valueString[valueStringLength++] = (char16_t) UNICODE_HASH_CODEPOINT; /* adding # */ + if(length==2){ + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, codePoint, 10, 0); + } + else{ + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[0], 10, 0); + } + valueString[valueStringLength++] = (char16_t) UNICODE_SEMICOLON_CODEPOINT; /* adding ; */ + break; + + case UCNV_PRV_ESCAPE_XML_HEX: + + valueString[valueStringLength++] = (char16_t) UNICODE_AMP_CODEPOINT; /* adding & */ + valueString[valueStringLength++] = (char16_t) UNICODE_HASH_CODEPOINT; /* adding # */ + valueString[valueStringLength++] = (char16_t) UNICODE_X_LOW_CODEPOINT; /* adding x */ + if(length==2){ + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, codePoint, 16, 0); + } + else{ + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[0], 16, 0); + } + valueString[valueStringLength++] = (char16_t) UNICODE_SEMICOLON_CODEPOINT; /* adding ; */ + break; + + case UCNV_PRV_ESCAPE_UNICODE: + valueString[valueStringLength++] = (char16_t) UNICODE_LEFT_CURLY_CODEPOINT; /* adding { */ + valueString[valueStringLength++] = (char16_t) UNICODE_U_CODEPOINT; /* adding U */ + valueString[valueStringLength++] = (char16_t) UNICODE_PLUS_CODEPOINT; /* adding + */ + if (length == 2) { + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, codePoint, 16, 4); + } else { + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[0], 16, 4); + } + valueString[valueStringLength++] = (char16_t) UNICODE_RIGHT_CURLY_CODEPOINT; /* adding } */ + break; + + case UCNV_PRV_ESCAPE_CSS2: + valueString[valueStringLength++] = (char16_t) UNICODE_RS_CODEPOINT; /* adding \ */ + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, codePoint, 16, 0); + /* Always add space character, because the next character might be whitespace, + which would erroneously be considered the termination of the escape sequence. */ + valueString[valueStringLength++] = (char16_t) UNICODE_SPACE_CODEPOINT; + break; + + default: + while (i < length) + { + valueString[valueStringLength++] = (char16_t) UNICODE_PERCENT_SIGN_CODEPOINT; /* adding % */ + valueString[valueStringLength++] = (char16_t) UNICODE_U_CODEPOINT; /* adding U */ + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[i++], 16, 4); + } + } + } + myValueSource = valueString; + + /* reset the error */ + *err = U_ZERO_ERROR; + + ucnv_cbFromUWriteUChars(fromArgs, &myValueSource, myValueSource+valueStringLength, 0, err); + + ucnv_setFromUCallBack (fromArgs->converter, + original, + originalContext, + &ignoredCallback, + &ignoredContext, + &err2); + if (U_FAILURE (err2)) + { + *err = err2; + return; + } + + return; +} + + + +U_CAPI void U_EXPORT2 +UCNV_TO_U_CALLBACK_SKIP ( + const void *context, + UConverterToUnicodeArgs *toArgs, + const char* codeUnits, + int32_t length, + UConverterCallbackReason reason, + UErrorCode * err) +{ + (void)toArgs; + (void)codeUnits; + (void)length; + if (reason <= UCNV_IRREGULAR) + { + if (context == nullptr || (*((char*)context) == UCNV_PRV_STOP_ON_ILLEGAL && reason == UCNV_UNASSIGNED)) + { + *err = U_ZERO_ERROR; + } + /* else the caller must have set the error code accordingly. */ + } + /* else ignore the reset, close and clone calls. */ +} + +U_CAPI void U_EXPORT2 +UCNV_TO_U_CALLBACK_SUBSTITUTE ( + const void *context, + UConverterToUnicodeArgs *toArgs, + const char* codeUnits, + int32_t length, + UConverterCallbackReason reason, + UErrorCode * err) +{ + (void)codeUnits; + (void)length; + if (reason <= UCNV_IRREGULAR) + { + if (context == nullptr || (*((char*)context) == UCNV_PRV_STOP_ON_ILLEGAL && reason == UCNV_UNASSIGNED)) + { + *err = U_ZERO_ERROR; + ucnv_cbToUWriteSub(toArgs,0,err); + } + /* else the caller must have set the error code accordingly. */ + } + /* else ignore the reset, close and clone calls. */ +} + +/*uses uprv_itou to get a unicode escape sequence of the offensive sequence, + *and uses that as the substitution sequence + */ +U_CAPI void U_EXPORT2 +UCNV_TO_U_CALLBACK_ESCAPE ( + const void *context, + UConverterToUnicodeArgs *toArgs, + const char* codeUnits, + int32_t length, + UConverterCallbackReason reason, + UErrorCode * err) +{ + char16_t uniValueString[VALUE_STRING_LENGTH]; + int32_t valueStringLength = 0; + int32_t i = 0; + + if (reason > UCNV_IRREGULAR) + { + return; + } + + if(context==nullptr) + { + while (i < length) + { + uniValueString[valueStringLength++] = (char16_t) UNICODE_PERCENT_SIGN_CODEPOINT; /* adding % */ + uniValueString[valueStringLength++] = (char16_t) UNICODE_X_CODEPOINT; /* adding X */ + valueStringLength += uprv_itou (uniValueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint8_t) codeUnits[i++], 16, 2); + } + } + else + { + switch(*((char*)context)) + { + case UCNV_PRV_ESCAPE_XML_DEC: + while (i < length) + { + uniValueString[valueStringLength++] = (char16_t) UNICODE_AMP_CODEPOINT; /* adding & */ + uniValueString[valueStringLength++] = (char16_t) UNICODE_HASH_CODEPOINT; /* adding # */ + valueStringLength += uprv_itou (uniValueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint8_t)codeUnits[i++], 10, 0); + uniValueString[valueStringLength++] = (char16_t) UNICODE_SEMICOLON_CODEPOINT; /* adding ; */ + } + break; + + case UCNV_PRV_ESCAPE_XML_HEX: + while (i < length) + { + uniValueString[valueStringLength++] = (char16_t) UNICODE_AMP_CODEPOINT; /* adding & */ + uniValueString[valueStringLength++] = (char16_t) UNICODE_HASH_CODEPOINT; /* adding # */ + uniValueString[valueStringLength++] = (char16_t) UNICODE_X_LOW_CODEPOINT; /* adding x */ + valueStringLength += uprv_itou (uniValueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint8_t)codeUnits[i++], 16, 0); + uniValueString[valueStringLength++] = (char16_t) UNICODE_SEMICOLON_CODEPOINT; /* adding ; */ + } + break; + case UCNV_PRV_ESCAPE_C: + while (i < length) + { + uniValueString[valueStringLength++] = (char16_t) UNICODE_RS_CODEPOINT; /* adding \ */ + uniValueString[valueStringLength++] = (char16_t) UNICODE_X_LOW_CODEPOINT; /* adding x */ + valueStringLength += uprv_itou (uniValueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint8_t)codeUnits[i++], 16, 2); + } + break; + default: + while (i < length) + { + uniValueString[valueStringLength++] = (char16_t) UNICODE_PERCENT_SIGN_CODEPOINT; /* adding % */ + uniValueString[valueStringLength++] = (char16_t) UNICODE_X_CODEPOINT; /* adding X */ + uprv_itou (uniValueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint8_t) codeUnits[i++], 16, 2); + valueStringLength += 2; + } + } + } + /* reset the error */ + *err = U_ZERO_ERROR; + + ucnv_cbToUWriteUChars(toArgs, uniValueString, valueStringLength, 0, err); +} + +#endif diff --git a/intl/icu/source/common/ucnv_ext.cpp b/intl/icu/source/common/ucnv_ext.cpp new file mode 100644 index 0000000000..10d00a62c2 --- /dev/null +++ b/intl/icu/source/common/ucnv_ext.cpp @@ -0,0 +1,1143 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2003-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: ucnv_ext.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2003jun13 +* created by: Markus W. Scherer +* +* Conversion extensions +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION && !UCONFIG_NO_LEGACY_CONVERSION + +#include "unicode/uset.h" +#include "unicode/ustring.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" +#include "ucnv_ext.h" +#include "cmemory.h" +#include "uassert.h" + +/* to Unicode --------------------------------------------------------------- */ + +/* + * @return lookup value for the byte, if found; else 0 + */ +static inline uint32_t +ucnv_extFindToU(const uint32_t *toUSection, int32_t length, uint8_t byte) { + uint32_t word0, word; + int32_t i, start, limit; + + /* check the input byte against the lowest and highest section bytes */ + start=(int32_t)UCNV_EXT_TO_U_GET_BYTE(toUSection[0]); + limit=(int32_t)UCNV_EXT_TO_U_GET_BYTE(toUSection[length-1]); + if(byte1) { + return 0; /* no match of a DBCS sequence in SBCS mode */ + } else if(preLength==1) { + srcLength=0; + } else /* preLength==0 */ { + if(srcLength>1) { + srcLength=1; + } + } + flush=true; + } + + /* we must not remember fallback matches when not using fallbacks */ + + /* match input units until there is a full match or the input is consumed */ + for(;;) { + /* go to the next section */ + toUSection=toUTable+idx; + + /* read first pair of the section */ + value=*toUSection++; + length=UCNV_EXT_TO_U_GET_BYTE(value); + value=UCNV_EXT_TO_U_GET_VALUE(value); + if( value!=0 && + (UCNV_EXT_TO_U_IS_ROUNDTRIP(value) || + TO_U_USE_FALLBACK(useFallback)) && + UCNV_EXT_TO_U_VERIFY_SISO_MATCH(sisoState, i+j) + ) { + /* remember longest match so far */ + matchValue=value; + matchLength=i+j; + } + + /* match pre[] then src[] */ + if(iUCNV_EXT_MAX_BYTES) { + /* + * end of the entire input stream, stop with the longest match so far + * or: partial match must not be longer than UCNV_EXT_MAX_BYTES + * because it must fit into state buffers + */ + break; + } else { + /* continue with more input next time */ + return -length; + } + } + + /* search for the current char16_t */ + value=ucnv_extFindToU(toUSection, length, b); + if(value==0) { + /* no match here, stop with the longest match so far */ + break; + } else { + if(UCNV_EXT_TO_U_IS_PARTIAL(value)) { + /* partial match, continue */ + idx=(int32_t)UCNV_EXT_TO_U_GET_PARTIAL_INDEX(value); + } else { + if( (UCNV_EXT_TO_U_IS_ROUNDTRIP(value) || + TO_U_USE_FALLBACK(useFallback)) && + UCNV_EXT_TO_U_VERIFY_SISO_MATCH(sisoState, i+j) + ) { + /* full match, stop with result */ + matchValue=value; + matchLength=i+j; + } else { + /* full match on fallback not taken, stop with the longest match so far */ + } + break; + } + } + } + + if(matchLength==0) { + /* no match at all */ + return 0; + } + + /* return result */ + *pMatchValue=UCNV_EXT_TO_U_MASK_ROUNDTRIP(matchValue); + return matchLength; +} + +static inline void +ucnv_extWriteToU(UConverter *cnv, const int32_t *cx, + uint32_t value, + char16_t **target, const char16_t *targetLimit, + int32_t **offsets, int32_t srcIndex, + UErrorCode *pErrorCode) { + /* output the result */ + if(UCNV_EXT_TO_U_IS_CODE_POINT(value)) { + /* output a single code point */ + ucnv_toUWriteCodePoint( + cnv, UCNV_EXT_TO_U_GET_CODE_POINT(value), + target, targetLimit, + offsets, srcIndex, + pErrorCode); + } else { + /* output a string - with correct data we have resultLength>0 */ + ucnv_toUWriteUChars( + cnv, + UCNV_EXT_ARRAY(cx, UCNV_EXT_TO_U_UCHARS_INDEX, char16_t)+ + UCNV_EXT_TO_U_GET_INDEX(value), + UCNV_EXT_TO_U_GET_LENGTH(value), + target, targetLimit, + offsets, srcIndex, + pErrorCode); + } +} + +/* + * get the SI/SO toU state (state 0 is for SBCS, 1 for DBCS), + * or 1 for DBCS-only, + * or -1 if the converter is not SI/SO stateful + * + * Note: For SI/SO stateful converters getting here, + * cnv->mode==0 is equivalent to firstLength==1. + */ +#define UCNV_SISO_STATE(cnv) \ + ((cnv)->sharedData->mbcs.outputType==MBCS_OUTPUT_2_SISO ? (int8_t)(cnv)->mode : \ + (cnv)->sharedData->mbcs.outputType==MBCS_OUTPUT_DBCS_ONLY ? 1 : -1) + +/* + * targettoUBytes, firstLength, + *src, (int32_t)(srcLimit-*src), + &value, + cnv->useFallback, flush); + if(match>0) { + /* advance src pointer for the consumed input */ + *src+=match-firstLength; + + /* write result to target */ + ucnv_extWriteToU(cnv, cx, + value, + target, targetLimit, + offsets, srcIndex, + pErrorCode); + return true; + } else if(match<0) { + /* save state for partial match */ + const char *s; + int32_t j; + + /* copy the first code point */ + s=(const char *)cnv->toUBytes; + cnv->preToUFirstLength=(int8_t)firstLength; + for(j=0; jpreToU[j]=*s++; + } + + /* now copy the newly consumed input */ + s=*src; + match=-match; + for(; jpreToU[j]=*s++; + } + *src=s; /* same as *src=srcLimit; because we reached the end of input */ + cnv->preToULength=(int8_t)match; + return true; + } else /* match==0 no match */ { + return false; + } +} + +U_CFUNC UChar32 +ucnv_extSimpleMatchToU(const int32_t *cx, + const char *source, int32_t length, + UBool useFallback) { + uint32_t value = 0; /* initialize output-only param to 0 to silence gcc */ + int32_t match; + + if(length<=0) { + return 0xffff; + } + + /* try to match */ + match=ucnv_extMatchToU(cx, -1, + source, length, + nullptr, 0, + &value, + useFallback, true); + if(match==length) { + /* write result for simple, single-character conversion */ + if(UCNV_EXT_TO_U_IS_CODE_POINT(value)) { + return UCNV_EXT_TO_U_GET_CODE_POINT(value); + } + } + + /* + * return no match because + * - match>0 && value points to string: simple conversion cannot handle multiple code points + * - match>0 && match!=length: not all input consumed, forbidden for this function + * - match==0: no match found in the first place + * - match<0: partial match, not supported for simple conversion (and flush==true) + */ + return 0xfffe; +} + +/* + * continue partial match with new input + * never called for simple, single-character conversion + */ +U_CFUNC void +ucnv_extContinueMatchToU(UConverter *cnv, + UConverterToUnicodeArgs *pArgs, int32_t srcIndex, + UErrorCode *pErrorCode) { + uint32_t value = 0; /* initialize output-only param to 0 to silence gcc */ + int32_t match, length; + + match=ucnv_extMatchToU(cnv->sharedData->mbcs.extIndexes, (int8_t)UCNV_SISO_STATE(cnv), + cnv->preToU, cnv->preToULength, + pArgs->source, (int32_t)(pArgs->sourceLimit-pArgs->source), + &value, + cnv->useFallback, pArgs->flush); + if(match>0) { + if(match>=cnv->preToULength) { + /* advance src pointer for the consumed input */ + pArgs->source+=match-cnv->preToULength; + cnv->preToULength=0; + } else { + /* the match did not use all of preToU[] - keep the rest for replay */ + length=cnv->preToULength-match; + uprv_memmove(cnv->preToU, cnv->preToU+match, length); + cnv->preToULength=(int8_t)-length; + } + + /* write result */ + ucnv_extWriteToU(cnv, cnv->sharedData->mbcs.extIndexes, + value, + &pArgs->target, pArgs->targetLimit, + &pArgs->offsets, srcIndex, + pErrorCode); + } else if(match<0) { + /* save state for partial match */ + const char *s; + int32_t j; + + /* just _append_ the newly consumed input to preToU[] */ + s=pArgs->source; + match=-match; + for(j=cnv->preToULength; jpreToU[j]=*s++; + } + pArgs->source=s; /* same as *src=srcLimit; because we reached the end of input */ + cnv->preToULength=(int8_t)match; + } else /* match==0 */ { + /* + * no match + * + * We need to split the previous input into two parts: + * + * 1. The first codepage character is unmappable - that's how we got into + * trying the extension data in the first place. + * We need to move it from the preToU buffer + * to the error buffer, set an error code, + * and prepare the rest of the previous input for 2. + * + * 2. The rest of the previous input must be converted once we + * come back from the callback for the first character. + * At that time, we have to try again from scratch to convert + * these input characters. + * The replay will be handled by the ucnv.c conversion code. + */ + + /* move the first codepage character to the error field */ + uprv_memcpy(cnv->toUBytes, cnv->preToU, cnv->preToUFirstLength); + cnv->toULength=cnv->preToUFirstLength; + + /* move the rest up inside the buffer */ + length=cnv->preToULength-cnv->preToUFirstLength; + if(length>0) { + uprv_memmove(cnv->preToU, cnv->preToU+cnv->preToUFirstLength, length); + } + + /* mark preToU for replay */ + cnv->preToULength=(int8_t)-length; + + /* set the error code for unassigned */ + *pErrorCode=U_INVALID_CHAR_FOUND; + } +} + +/* from Unicode ------------------------------------------------------------- */ + +// Use roundtrips, "good one-way" mappings, and some normal fallbacks. +static inline UBool +extFromUUseMapping(UBool useFallback, uint32_t value, UChar32 firstCP) { + return + ((value&UCNV_EXT_FROM_U_STATUS_MASK)!=0 || + FROM_U_USE_FALLBACK(useFallback, firstCP)) && + (value&UCNV_EXT_FROM_U_RESERVED_MASK)==0; +} + +/* + * @return index of the char16_t, if found; else <0 + */ +static inline int32_t +ucnv_extFindFromU(const char16_t *fromUSection, int32_t length, char16_t u) { + int32_t i, start, limit; + + /* binary search */ + start=0; + limit=length; + for(;;) { + i=limit-start; + if(i<=1) { + break; /* done */ + } + /* start=0 + * @param src UChars that can be used to complete a match + * @param srcLength length of src, >=0 + * @param pMatchValue [out] output result value for the match from the data structure + * @param useFallback "use fallback" flag, usually from cnv->useFallback + * @param flush true if the end of the input stream is reached + * @return >1: matched, return value=total match length (number of input units matched) + * 1: matched, no mapping but request for + * (only for the first code point) + * 0: no match + * <0: partial match, return value=negative total match length + * (partial matches are never returned for flush==true) + * (partial matches are never returned as being longer than UCNV_EXT_MAX_UCHARS) + * the matchLength is 2 if only firstCP matched, and >2 if firstCP and + * further code units matched + */ +static int32_t +ucnv_extMatchFromU(const int32_t *cx, + UChar32 firstCP, + const char16_t *pre, int32_t preLength, + const char16_t *src, int32_t srcLength, + uint32_t *pMatchValue, + UBool useFallback, UBool flush) { + const uint16_t *stage12, *stage3; + const uint32_t *stage3b; + + const char16_t *fromUTableUChars, *fromUSectionUChars; + const uint32_t *fromUTableValues, *fromUSectionValues; + + uint32_t value, matchValue; + int32_t i, j, idx, length, matchLength; + char16_t c; + + if(cx==nullptr) { + return 0; /* no extension data, no match */ + } + + /* trie lookup of firstCP */ + idx=firstCP>>10; /* stage 1 index */ + if(idx>=cx[UCNV_EXT_FROM_U_STAGE_1_LENGTH]) { + return 0; /* the first code point is outside the trie */ + } + + stage12=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_STAGE_12_INDEX, uint16_t); + stage3=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_STAGE_3_INDEX, uint16_t); + idx=UCNV_EXT_FROM_U(stage12, stage3, idx, firstCP); + + stage3b=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_STAGE_3B_INDEX, uint32_t); + value=stage3b[idx]; + if(value==0) { + return 0; + } + + /* + * Tests for (value&UCNV_EXT_FROM_U_RESERVED_MASK)==0: + * Do not interpret values with reserved bits used, for forward compatibility, + * and do not even remember intermediate results with reserved bits used. + */ + + if(UCNV_EXT_TO_U_IS_PARTIAL(value)) { + /* partial match, enter the loop below */ + idx=(int32_t)UCNV_EXT_FROM_U_GET_PARTIAL_INDEX(value); + + /* initialize */ + fromUTableUChars=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_UCHARS_INDEX, char16_t); + fromUTableValues=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_VALUES_INDEX, uint32_t); + + matchValue=0; + i=j=matchLength=0; + + /* we must not remember fallback matches when not using fallbacks */ + + /* match input units until there is a full match or the input is consumed */ + for(;;) { + /* go to the next section */ + fromUSectionUChars=fromUTableUChars+idx; + fromUSectionValues=fromUTableValues+idx; + + /* read first pair of the section */ + length=*fromUSectionUChars++; + value=*fromUSectionValues++; + if(value!=0 && extFromUUseMapping(useFallback, value, firstCP)) { + /* remember longest match so far */ + matchValue=value; + matchLength=2+i+j; + } + + /* match pre[] then src[] */ + if(iUCNV_EXT_MAX_UCHARS) { + /* + * end of the entire input stream, stop with the longest match so far + * or: partial match must not be longer than UCNV_EXT_MAX_UCHARS + * because it must fit into state buffers + */ + break; + } else { + /* continue with more input next time */ + return -(2+length); + } + } + + /* search for the current char16_t */ + idx=ucnv_extFindFromU(fromUSectionUChars, length, c); + if(idx<0) { + /* no match here, stop with the longest match so far */ + break; + } else { + value=fromUSectionValues[idx]; + if(UCNV_EXT_FROM_U_IS_PARTIAL(value)) { + /* partial match, continue */ + idx=(int32_t)UCNV_EXT_FROM_U_GET_PARTIAL_INDEX(value); + } else { + if(extFromUUseMapping(useFallback, value, firstCP)) { + /* full match, stop with result */ + matchValue=value; + matchLength=2+i+j; + } else { + /* full match on fallback not taken, stop with the longest match so far */ + } + break; + } + } + } + + if(matchLength==0) { + /* no match at all */ + return 0; + } + } else /* result from firstCP trie lookup */ { + if(extFromUUseMapping(useFallback, value, firstCP)) { + /* full match, stop with result */ + matchValue=value; + matchLength=2; + } else { + /* fallback not taken */ + return 0; + } + } + + /* return result */ + if(matchValue==UCNV_EXT_FROM_U_SUBCHAR1) { + return 1; /* assert matchLength==2 */ + } + + *pMatchValue=matchValue; + return matchLength; +} + +/* + * @param value fromUnicode mapping table value; ignores roundtrip and reserved bits + */ +static inline void +ucnv_extWriteFromU(UConverter *cnv, const int32_t *cx, + uint32_t value, + char **target, const char *targetLimit, + int32_t **offsets, int32_t srcIndex, + UErrorCode *pErrorCode) { + uint8_t buffer[1+UCNV_EXT_MAX_BYTES]; + const uint8_t *result; + int32_t length, prevLength; + + length=UCNV_EXT_FROM_U_GET_LENGTH(value); + value=(uint32_t)UCNV_EXT_FROM_U_GET_DATA(value); + + /* output the result */ + if(length<=UCNV_EXT_FROM_U_MAX_DIRECT_LENGTH) { + /* + * Generate a byte array and then write it below. + * This is not the fastest possible way, but it should be ok for + * extension mappings, and it is much simpler. + * Offset and overflow handling are only done once this way. + */ + uint8_t *p=buffer+1; /* reserve buffer[0] for shiftByte below */ + switch(length) { + case 3: + *p++=(uint8_t)(value>>16); + U_FALLTHROUGH; + case 2: + *p++=(uint8_t)(value>>8); + U_FALLTHROUGH; + case 1: + *p++=(uint8_t)value; + U_FALLTHROUGH; + default: + break; /* will never occur */ + } + result=buffer+1; + } else { + result=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_BYTES_INDEX, uint8_t)+value; + } + + /* with correct data we have length>0 */ + + if((prevLength=cnv->fromUnicodeStatus)!=0) { + /* handle SI/SO stateful output */ + uint8_t shiftByte; + + if(prevLength>1 && length==1) { + /* change from double-byte mode to single-byte */ + shiftByte=(uint8_t)UCNV_SI; + cnv->fromUnicodeStatus=1; + } else if(prevLength==1 && length>1) { + /* change from single-byte mode to double-byte */ + shiftByte=(uint8_t)UCNV_SO; + cnv->fromUnicodeStatus=2; + } else { + shiftByte=0; + } + + if(shiftByte!=0) { + /* prepend the shift byte to the result bytes */ + buffer[0]=shiftByte; + if(result!=buffer+1) { + uprv_memcpy(buffer+1, result, length); + } + result=buffer; + ++length; + } + } + + ucnv_fromUWriteBytes(cnv, (const char *)result, length, + target, targetLimit, + offsets, srcIndex, + pErrorCode); +} + +/* + * targetuseFallback, flush); + + /* reject a match if the result is a single byte for DBCS-only */ + if( match>=2 && + !(UCNV_EXT_FROM_U_GET_LENGTH(value)==1 && + cnv->sharedData->mbcs.outputType==MBCS_OUTPUT_DBCS_ONLY) + ) { + /* advance src pointer for the consumed input */ + *src+=match-2; /* remove 2 for the initial code point */ + + /* write result to target */ + ucnv_extWriteFromU(cnv, cx, + value, + target, targetLimit, + offsets, srcIndex, + pErrorCode); + return true; + } else if(match<0) { + /* save state for partial match */ + const char16_t *s; + int32_t j; + + /* copy the first code point */ + cnv->preFromUFirstCP=cp; + + /* now copy the newly consumed input */ + s=*src; + match=-match-2; /* remove 2 for the initial code point */ + for(j=0; jpreFromU[j]=*s++; + } + *src=s; /* same as *src=srcLimit; because we reached the end of input */ + cnv->preFromULength=(int8_t)match; + return true; + } else if(match==1) { + /* matched, no mapping but request for */ + cnv->useSubChar1=true; + return false; + } else /* match==0 no match */ { + return false; + } +} + +/* + * Used by ISO 2022 implementation. + * @return number of bytes in *pValue; negative number if fallback; 0 for no mapping + */ +U_CFUNC int32_t +ucnv_extSimpleMatchFromU(const int32_t *cx, + UChar32 cp, uint32_t *pValue, + UBool useFallback) { + uint32_t value; + int32_t match; + + /* try to match */ + match=ucnv_extMatchFromU(cx, + cp, + nullptr, 0, + nullptr, 0, + &value, + useFallback, true); + if(match>=2) { + /* write result for simple, single-character conversion */ + int32_t length; + int isRoundtrip; + + isRoundtrip=UCNV_EXT_FROM_U_IS_ROUNDTRIP(value); + length=UCNV_EXT_FROM_U_GET_LENGTH(value); + value=(uint32_t)UCNV_EXT_FROM_U_GET_DATA(value); + + if(length<=UCNV_EXT_FROM_U_MAX_DIRECT_LENGTH) { + *pValue=value; + return isRoundtrip ? length : -length; +#if 0 /* not currently used */ + } else if(length==4) { + /* de-serialize a 4-byte result */ + const uint8_t *result=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_BYTES_INDEX, uint8_t)+value; + *pValue= + ((uint32_t)result[0]<<24)| + ((uint32_t)result[1]<<16)| + ((uint32_t)result[2]<<8)| + result[3]; + return isRoundtrip ? 4 : -4; +#endif + } + } + + /* + * return no match because + * - match>1 && resultLength>4: result too long for simple conversion + * - match==1: no match found, preferred + * - match==0: no match found in the first place + * - match<0: partial match, not supported for simple conversion (and flush==true) + */ + return 0; +} + +/* + * continue partial match with new input, requires cnv->preFromUFirstCP>=0 + * never called for simple, single-character conversion + */ +U_CFUNC void +ucnv_extContinueMatchFromU(UConverter *cnv, + UConverterFromUnicodeArgs *pArgs, int32_t srcIndex, + UErrorCode *pErrorCode) { + uint32_t value = 0; /* initialize output-only param to 0 to silence gcc */ + int32_t match; + + match=ucnv_extMatchFromU(cnv->sharedData->mbcs.extIndexes, + cnv->preFromUFirstCP, + cnv->preFromU, cnv->preFromULength, + pArgs->source, (int32_t)(pArgs->sourceLimit-pArgs->source), + &value, + cnv->useFallback, pArgs->flush); + if(match>=2) { + match-=2; /* remove 2 for the initial code point */ + + if(match>=cnv->preFromULength) { + /* advance src pointer for the consumed input */ + pArgs->source+=match-cnv->preFromULength; + cnv->preFromULength=0; + } else { + /* the match did not use all of preFromU[] - keep the rest for replay */ + int32_t length=cnv->preFromULength-match; + u_memmove(cnv->preFromU, cnv->preFromU+match, length); + cnv->preFromULength=(int8_t)-length; + } + + /* finish the partial match */ + cnv->preFromUFirstCP=U_SENTINEL; + + /* write result */ + ucnv_extWriteFromU(cnv, cnv->sharedData->mbcs.extIndexes, + value, + &pArgs->target, pArgs->targetLimit, + &pArgs->offsets, srcIndex, + pErrorCode); + } else if(match<0) { + /* save state for partial match */ + const char16_t *s; + int32_t j; + + /* just _append_ the newly consumed input to preFromU[] */ + s=pArgs->source; + match=-match-2; /* remove 2 for the initial code point */ + for(j=cnv->preFromULength; j=0); + cnv->preFromU[j]=*s++; + } + pArgs->source=s; /* same as *src=srcLimit; because we reached the end of input */ + cnv->preFromULength=(int8_t)match; + } else /* match==0 or 1 */ { + /* + * no match + * + * We need to split the previous input into two parts: + * + * 1. The first code point is unmappable - that's how we got into + * trying the extension data in the first place. + * We need to move it from the preFromU buffer + * to the error buffer, set an error code, + * and prepare the rest of the previous input for 2. + * + * 2. The rest of the previous input must be converted once we + * come back from the callback for the first code point. + * At that time, we have to try again from scratch to convert + * these input characters. + * The replay will be handled by the ucnv.c conversion code. + */ + + if(match==1) { + /* matched, no mapping but request for */ + cnv->useSubChar1=true; + } + + /* move the first code point to the error field */ + cnv->fromUChar32=cnv->preFromUFirstCP; + cnv->preFromUFirstCP=U_SENTINEL; + + /* mark preFromU for replay */ + cnv->preFromULength=-cnv->preFromULength; + + /* set the error code for unassigned */ + *pErrorCode=U_INVALID_CHAR_FOUND; + } +} + +static UBool +extSetUseMapping(UConverterUnicodeSet which, int32_t minLength, uint32_t value) { + if(which==UCNV_ROUNDTRIP_SET) { + // Add only code points for which the roundtrip flag is set. + // Do not add any fallbacks, even if ucnv_fromUnicode() would use them + // (fallbacks from PUA). See the API docs for ucnv_getUnicodeSet(). + // + // By analogy, also do not add "good one-way" mappings. + // + // Do not add entries with reserved bits set. + if(((value&(UCNV_EXT_FROM_U_ROUNDTRIP_FLAG|UCNV_EXT_FROM_U_RESERVED_MASK))!= + UCNV_EXT_FROM_U_ROUNDTRIP_FLAG)) { + return false; + } + } else /* UCNV_ROUNDTRIP_AND_FALLBACK_SET */ { + // Do not add entries with reserved bits set. + if((value&UCNV_EXT_FROM_U_RESERVED_MASK)!=0) { + return false; + } + } + // Do not add entries or other (future?) pseudo-entries + // with an output length of 0. + return UCNV_EXT_FROM_U_GET_LENGTH(value)>=minLength; +} + +static void +ucnv_extGetUnicodeSetString(const UConverterSharedData *sharedData, + const int32_t *cx, + const USetAdder *sa, + UConverterUnicodeSet which, + int32_t minLength, + UChar32 firstCP, + char16_t s[UCNV_EXT_MAX_UCHARS], int32_t length, + int32_t sectionIndex, + UErrorCode *pErrorCode) { + const char16_t *fromUSectionUChars; + const uint32_t *fromUSectionValues; + + uint32_t value; + int32_t i, count; + + fromUSectionUChars=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_UCHARS_INDEX, char16_t)+sectionIndex; + fromUSectionValues=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_VALUES_INDEX, uint32_t)+sectionIndex; + + /* read first pair of the section */ + count=*fromUSectionUChars++; + value=*fromUSectionValues++; + + if(extSetUseMapping(which, minLength, value)) { + if(length==U16_LENGTH(firstCP)) { + /* add the initial code point */ + sa->add(sa->set, firstCP); + } else { + /* add the string so far */ + sa->addString(sa->set, s, length); + } + } + + for(i=0; iaddString(sa->set, s, length+1); + } + } +} + +U_CFUNC void +ucnv_extGetUnicodeSet(const UConverterSharedData *sharedData, + const USetAdder *sa, + UConverterUnicodeSet which, + UConverterSetFilter filter, + UErrorCode *pErrorCode) { + const int32_t *cx; + const uint16_t *stage12, *stage3, *ps2, *ps3; + const uint32_t *stage3b; + + uint32_t value; + int32_t st1, stage1Length, st2, st3, minLength; + + char16_t s[UCNV_EXT_MAX_UCHARS]; + UChar32 c; + int32_t length; + + cx=sharedData->mbcs.extIndexes; + if(cx==nullptr) { + return; + } + + stage12=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_STAGE_12_INDEX, uint16_t); + stage3=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_STAGE_3_INDEX, uint16_t); + stage3b=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_STAGE_3B_INDEX, uint32_t); + + stage1Length=cx[UCNV_EXT_FROM_U_STAGE_1_LENGTH]; + + /* enumerate the from-Unicode trie table */ + c=0; /* keep track of the current code point while enumerating */ + + if(filter==UCNV_SET_FILTER_2022_CN) { + minLength=3; + } else if( sharedData->mbcs.outputType==MBCS_OUTPUT_DBCS_ONLY || + filter!=UCNV_SET_FILTER_NONE + ) { + /* DBCS-only, ignore single-byte results */ + minLength=2; + } else { + minLength=1; + } + + /* + * the trie enumeration is almost the same as + * in MBCSGetUnicodeSet() for MBCS_OUTPUT_1 + */ + for(st1=0; st1stage1Length) { + ps2=stage12+st2; + for(st2=0; st2<64; ++st2) { + if((st3=(int32_t)ps2[st2]<=0x8140 && value<=0xeffc)) { + continue; + } + break; + case UCNV_SET_FILTER_GR94DBCS: + if(!(UCNV_EXT_FROM_U_GET_LENGTH(value)==2 && + (uint16_t)((value=UCNV_EXT_FROM_U_GET_DATA(value))-0xa1a1)<=(0xfefe - 0xa1a1) && + (uint8_t)(value-0xa1)<=(0xfe - 0xa1))) { + continue; + } + break; + case UCNV_SET_FILTER_HZ: + if(!(UCNV_EXT_FROM_U_GET_LENGTH(value)==2 && + (uint16_t)((value=UCNV_EXT_FROM_U_GET_DATA(value))-0xa1a1)<=(0xfdfe - 0xa1a1) && + (uint8_t)(value-0xa1)<=(0xfe - 0xa1))) { + continue; + } + break; + default: + /* + * UCNV_SET_FILTER_NONE, + * or UCNV_SET_FILTER_DBCS_ONLY which is handled via minLength + */ + break; + } + sa->add(sa->set, c); + } + } while((++c&0xf)!=0); + } else { + c+=16; /* empty stage 3 block */ + } + } + } else { + c+=1024; /* empty stage 2 block */ + } + } +} + +#endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */ diff --git a/intl/icu/source/common/ucnv_ext.h b/intl/icu/source/common/ucnv_ext.h new file mode 100644 index 0000000000..dceea7ef12 --- /dev/null +++ b/intl/icu/source/common/ucnv_ext.h @@ -0,0 +1,481 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2003-2013, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: ucnv_ext.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2003jun13 +* created by: Markus W. Scherer +* +* Conversion extensions +*/ + +#ifndef __UCNV_EXT_H__ +#define __UCNV_EXT_H__ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/ucnv.h" +#include "ucnv_cnv.h" + +/* + * See icuhtml/design/conversion/conversion_extensions.html + * + * Conversion extensions serve three purposes: + * 1. They support m:n mappings. + * 2. They support extension-only conversion files that are used together + * with the regular conversion data in base files. + * 3. They support mappings with more complicated meta data, + * for example "good one-way" mappings (|4). + * + * A base file may contain an extension table (explicitly requested or + * implicitly generated for m:n mappings), but its extension table is not + * used when an extension-only file is used. + * + * It is an error if a base file contains any regular (not extension) mapping + * from the same sequence as a mapping in the extension file + * because the base mapping would hide the extension mapping. + * + * + * Data for conversion extensions: + * + * One set of data structures per conversion direction (to/from Unicode). + * The data structures are sorted by input units to allow for binary search. + * Input sequences of more than one unit are handled like contraction tables + * in collation: + * The lookup value of a unit points to another table that is to be searched + * for the next unit, recursively. + * + * For conversion from Unicode, the initial code point is looked up in + * a 3-stage trie for speed, + * with an additional table of unique results to save space. + * + * Long output strings are stored in separate arrays, with length and index + * in the lookup tables. + * Output results also include a flag distinguishing roundtrip from + * (reverse) fallback mappings. + * + * Input Unicode strings must not begin or end with unpaired surrogates + * to avoid problems with matches on parts of surrogate pairs. + * + * Mappings from multiple characters (code points or codepage state + * table sequences) must be searched preferring the longest match. + * For this to work and be efficient, the variable-width table must contain + * all mappings that contain prefixes of the multiple characters. + * If an extension table is built on top of a base table in another file + * and a base table entry is a prefix of a multi-character mapping, then + * this is an error. + * + * + * Implementation note: + * + * Currently, the parser and several checks in the code limit the number + * of UChars or bytes in a mapping to + * UCNV_EXT_MAX_UCHARS and UCNV_EXT_MAX_BYTES, respectively, + * which are output value limits in the data structure. + * + * For input, this is not strictly necessary - it is a hard limit only for the + * buffers in UConverter that are used to store partial matches. + * + * Input sequences could otherwise be arbitrarily long if partial matches + * need not be stored (i.e., if a sequence does not span several buffers with too + * many units before the last buffer), although then results would differ + * depending on whether partial matches exceed the limits or not, + * which depends on the pattern of buffer sizes. + * + * + * Data structure: + * + * int32_t indexes[>=32]; + * + * Array of indexes and lengths etc. The length of the array is at least 32. + * The actual length is stored in indexes[0] to be forward compatible. + * + * Each index to another array is the number of bytes from indexes[]. + * Each length of an array is the number of array base units in that array. + * + * Some of the structures may not be present, in which case their indexes + * and lengths are 0. + * + * Usage of indexes[i]: + * [0] length of indexes[] + * + * // to Unicode table + * [1] index of toUTable[] (array of uint32_t) + * [2] length of toUTable[] + * [3] index of toUUChars[] (array of UChar) + * [4] length of toUUChars[] + * + * // from Unicode table, not for the initial code point + * [5] index of fromUTableUChars[] (array of UChar) + * [6] index of fromUTableValues[] (array of uint32_t) + * [7] length of fromUTableUChars[] and fromUTableValues[] + * [8] index of fromUBytes[] (array of char) + * [9] length of fromUBytes[] + * + * // from Unicode trie for initial-code point lookup + * [10] index of fromUStage12[] (combined array of uint16_t for stages 1 & 2) + * [11] length of stage 1 portion of fromUStage12[] + * [12] length of fromUStage12[] + * [13] index of fromUStage3[] (array of uint16_t indexes into fromUStage3b[]) + * [14] length of fromUStage3[] + * [15] index of fromUStage3b[] (array of uint32_t like fromUTableValues[]) + * [16] length of fromUStage3b[] + * + * [17] Bit field containing numbers of bytes: + * 31..24 reserved, 0 + * 23..16 maximum input bytes + * 15.. 8 maximum output bytes + * 7.. 0 maximum bytes per UChar + * + * [18] Bit field containing numbers of UChars: + * 31..24 reserved, 0 + * 23..16 maximum input UChars + * 15.. 8 maximum output UChars + * 7.. 0 maximum UChars per byte + * + * [19] Bit field containing flags: + * (extension table unicodeMask) + * 1 UCNV_HAS_SURROGATES flag for the extension table + * 0 UCNV_HAS_SUPPLEMENTARY flag for the extension table + * + * [20]..[30] reserved, 0 + * [31] number of bytes for the entire extension structure + * [>31] reserved; there are indexes[0] indexes + * + * + * uint32_t toUTable[]; + * + * Array of byte/value pairs for lookups for toUnicode conversion. + * The array is partitioned into sections like collation contraction tables. + * Each section contains one word with the number of following words and + * a default value for when the lookup in this section yields no match. + * + * A section is sorted in ascending order of input bytes, + * allowing for fast linear or binary searches. + * The builder may store entries for a contiguous range of byte values + * (compare difference between the first and last one with count), + * which then allows for direct array access. + * The builder should always do this for the initial table section. + * + * Entries may have 0 values, see below. + * No two entries in a section have the same byte values. + * + * Each uint32_t contains an input byte value in bits 31..24 and the + * corresponding lookup value in bits 23..0. + * Interpret the value as follows: + * if(value==0) { + * no match, see below + * } else if(value<0x1f0000) { + * partial match - use value as index to the next toUTable section + * and match the next unit; (value indexes toUTable[value]) + * } else { + * if(bit 23 set) { + * roundtrip; + * } else { + * fallback; + * } + * unset value bit 23; + * if(value<=0x2fffff) { + * (value-0x1f0000) is a code point; (BMP: value<=0x1fffff) + * } else { + * bits 17..0 (value&0x3ffff) is an index to + * the result UChars in toUUChars[]; (0 indexes toUUChars[0]) + * length of the result=((value>>18)-12); (length=0..19) + * } + * } + * + * The first word in a section contains the number of following words in the + * input byte position (bits 31..24, number=1..0xff). + * The value of the initial word is used when the current byte is not found + * in this section. + * If the value is not 0, then it represents a result as above. + * If the value is 0, then the search has to return a shorter match with an + * earlier default value as the result, or result in "unmappable" even for the + * initial bytes. + * If the value is 0 for the initial toUTable entry, then the initial byte + * does not start any mapping input. + * + * + * UChar toUUChars[]; + * + * Contains toUnicode mapping results, stored as sequences of UChars. + * Indexes and lengths stored in the toUTable[]. + * + * + * UChar fromUTableUChars[]; + * uint32_t fromUTableValues[]; + * + * The fromUTable is split into two arrays, but works otherwise much like + * the toUTable. The array is partitioned into sections like collation + * contraction tables and toUTable. + * A row in the table consists of same-index entries in fromUTableUChars[] + * and fromUTableValues[]. + * + * Interpret a value as follows: + * if(value==0) { + * no match, see below + * } else if(value<=0xffffff) { (bits 31..24 are 0) + * partial match - use value as index to the next fromUTable section + * and match the next unit; (value indexes fromUTable[value]) + * } else { + * if(value==0x80000001) { + * return no mapping, but request for ; + * } + * if(bit 31 set) { + * roundtrip (|0); + * } else if(bit 30 set) { + * "good one-way" mapping (|4); -- new in ICU4C 51, _MBCSHeader.version 5.4/4.4 + * } else { + * normal fallback (|1); + * } + * // bit 29 reserved, 0 + * length=(value>>24)&0x1f; (bits 28..24) + * if(length==1..3) { + * bits 23..0 contain 1..3 bytes, padded with 00s on the left; + * } else { + * bits 23..0 (value&0xffffff) is an index to + * the result bytes in fromUBytes[]; (0 indexes fromUBytes[0]) + * } + * } + * + * The first pair in a section contains the number of following pairs in the + * UChar position (16 bits, number=1..0xffff). + * The value of the initial pair is used when the current UChar is not found + * in this section. + * If the value is not 0, then it represents a result as above. + * If the value is 0, then the search has to return a shorter match with an + * earlier default value as the result, or result in "unmappable" even for the + * initial UChars. + * + * If the from Unicode trie is present, then the from Unicode search tables + * are not used for initial code points. + * In this case, the first entries (index 0) in the tables are not used + * (reserved, set to 0) because a value of 0 is used in trie results + * to indicate no mapping. + * + * + * uint16_t fromUStage12[]; + * + * Stages 1 & 2 of a trie that maps an initial code point. + * Indexes in stage 1 are all offset by the length of stage 1 so that the + * same array pointer can be used for both stages. + * If (c>>10)>=(length of stage 1) then c does not start any mapping. + * Same bit distribution as for regular conversion tries. + * + * + * uint16_t fromUStage3[]; + * uint32_t fromUStage3b[]; + * + * Stage 3 of the trie. The first array simply contains indexes to the second, + * which contains words in the same format as fromUTableValues[]. + * Use a stage 3 granularity of 4, which allows for 256k stage 3 entries, + * and 16-bit entries in stage 3 allow for 64k stage 3b entries. + * The stage 3 granularity means that the stage 2 entry needs to be left-shifted. + * + * Two arrays are used because it is expected that more than half of the stage 3 + * entries will be zero. The 16-bit index stage 3 array saves space even + * considering storing a total of 6 bytes per non-zero entry in both arrays + * together. + * Using a stage 3 granularity of >1 diminishes the compactability in that stage + * but provides a larger effective addressing space in stage 2. + * All but the final result stage use 16-bit entries to save space. + * + * fromUStage3b[] contains a zero for "no mapping" at its index 0, + * and may contain UCNV_EXT_FROM_U_SUBCHAR1 at index 1 for " SUB mapping" + * (i.e., "no mapping" with preference for rather than ), + * and all other items are unique non-zero results. + * + * The default value of a fromUTableValues[] section that is referenced + * _directly_ from a fromUStage3b[] item may also be UCNV_EXT_FROM_U_SUBCHAR1, + * but this value must not occur anywhere else in fromUTableValues[] + * because "no mapping" is always a property of a single code point, + * never of multiple. + * + * + * char fromUBytes[]; + * + * Contains fromUnicode mapping results, stored as sequences of chars. + * Indexes and lengths stored in the fromUTableValues[]. + */ +enum { + UCNV_EXT_INDEXES_LENGTH, /* 0 */ + + UCNV_EXT_TO_U_INDEX, /* 1 */ + UCNV_EXT_TO_U_LENGTH, + UCNV_EXT_TO_U_UCHARS_INDEX, + UCNV_EXT_TO_U_UCHARS_LENGTH, + + UCNV_EXT_FROM_U_UCHARS_INDEX, /* 5 */ + UCNV_EXT_FROM_U_VALUES_INDEX, + UCNV_EXT_FROM_U_LENGTH, + UCNV_EXT_FROM_U_BYTES_INDEX, + UCNV_EXT_FROM_U_BYTES_LENGTH, + + UCNV_EXT_FROM_U_STAGE_12_INDEX, /* 10 */ + UCNV_EXT_FROM_U_STAGE_1_LENGTH, + UCNV_EXT_FROM_U_STAGE_12_LENGTH, + UCNV_EXT_FROM_U_STAGE_3_INDEX, + UCNV_EXT_FROM_U_STAGE_3_LENGTH, + UCNV_EXT_FROM_U_STAGE_3B_INDEX, + UCNV_EXT_FROM_U_STAGE_3B_LENGTH, + + UCNV_EXT_COUNT_BYTES, /* 17 */ + UCNV_EXT_COUNT_UCHARS, + UCNV_EXT_FLAGS, + + UCNV_EXT_RESERVED_INDEX, /* 20, moves with additional indexes */ + + UCNV_EXT_SIZE=31, + UCNV_EXT_INDEXES_MIN_LENGTH=32 +}; + +/* get the pointer to an extension array from indexes[index] */ +#define UCNV_EXT_ARRAY(indexes, index, itemType) \ + ((const itemType *)((const char *)(indexes)+(indexes)[index])) + +#define UCNV_GET_MAX_BYTES_PER_UCHAR(indexes) \ + ((indexes)[UCNV_EXT_COUNT_BYTES]&0xff) + +/* internal API ------------------------------------------------------------- */ + +U_CFUNC UBool +ucnv_extInitialMatchToU(UConverter *cnv, const int32_t *cx, + int32_t firstLength, + const char **src, const char *srcLimit, + UChar **target, const UChar *targetLimit, + int32_t **offsets, int32_t srcIndex, + UBool flush, + UErrorCode *pErrorCode); + +U_CFUNC UChar32 +ucnv_extSimpleMatchToU(const int32_t *cx, + const char *source, int32_t length, + UBool useFallback); + +U_CFUNC void +ucnv_extContinueMatchToU(UConverter *cnv, + UConverterToUnicodeArgs *pArgs, int32_t srcIndex, + UErrorCode *pErrorCode); + + +U_CFUNC UBool +ucnv_extInitialMatchFromU(UConverter *cnv, const int32_t *cx, + UChar32 cp, + const UChar **src, const UChar *srcLimit, + char **target, const char *targetLimit, + int32_t **offsets, int32_t srcIndex, + UBool flush, + UErrorCode *pErrorCode); + +U_CFUNC int32_t +ucnv_extSimpleMatchFromU(const int32_t *cx, + UChar32 cp, uint32_t *pValue, + UBool useFallback); + +U_CFUNC void +ucnv_extContinueMatchFromU(UConverter *cnv, + UConverterFromUnicodeArgs *pArgs, int32_t srcIndex, + UErrorCode *pErrorCode); + +/* + * Add code points and strings to the set according to the extension mappings. + * Limitation on the UConverterSetFilter: + * The filters currently assume that they are used with 1:1 mappings. + * They only apply to single input code points, and then they pass through + * only mappings with single-charset-code results. + * For example, the Shift-JIS filter only works for 2-byte results and tests + * that those 2 bytes are in the JIS X 0208 range of Shift-JIS. + */ +U_CFUNC void +ucnv_extGetUnicodeSet(const UConverterSharedData *sharedData, + const USetAdder *sa, + UConverterUnicodeSet which, + UConverterSetFilter filter, + UErrorCode *pErrorCode); + +/* toUnicode helpers -------------------------------------------------------- */ + +#define UCNV_EXT_TO_U_BYTE_SHIFT 24 +#define UCNV_EXT_TO_U_VALUE_MASK 0xffffff +#define UCNV_EXT_TO_U_MIN_CODE_POINT 0x1f0000 +#define UCNV_EXT_TO_U_MAX_CODE_POINT 0x2fffff +#define UCNV_EXT_TO_U_ROUNDTRIP_FLAG ((uint32_t)1<<23) +#define UCNV_EXT_TO_U_INDEX_MASK 0x3ffff +#define UCNV_EXT_TO_U_LENGTH_SHIFT 18 +#define UCNV_EXT_TO_U_LENGTH_OFFSET 12 + +/* maximum number of indexed UChars */ +#define UCNV_EXT_MAX_UCHARS 19 + +#define UCNV_EXT_TO_U_MAKE_WORD(byte, value) (((uint32_t)(byte)<>UCNV_EXT_TO_U_BYTE_SHIFT) +#define UCNV_EXT_TO_U_GET_VALUE(word) ((word)&UCNV_EXT_TO_U_VALUE_MASK) + +#define UCNV_EXT_TO_U_IS_PARTIAL(value) ((value)>UCNV_EXT_TO_U_LENGTH_SHIFT)-UCNV_EXT_TO_U_LENGTH_OFFSET) + +/* fromUnicode helpers ------------------------------------------------------ */ + +/* most trie constants are shared with ucnvmbcs.h */ + +/* see similar utrie.h UTRIE_INDEX_SHIFT and UTRIE_DATA_GRANULARITY */ +#define UCNV_EXT_STAGE_2_LEFT_SHIFT 2 +#define UCNV_EXT_STAGE_3_GRANULARITY 4 + +/* trie access, returns the stage 3 value=index to stage 3b; s1Index=c>>10 */ +#define UCNV_EXT_FROM_U(stage12, stage3, s1Index, c) \ + (stage3)[ ((int32_t)(stage12)[ (stage12)[s1Index] +(((c)>>4)&0x3f) ]< (impossible roundtrip to 0 bytes, value 01) */ +#define UCNV_EXT_FROM_U_SUBCHAR1 0x80000001 + +/* at most 3 bytes in the lower part of the value */ +#define UCNV_EXT_FROM_U_MAX_DIRECT_LENGTH 3 + +/* maximum number of indexed bytes */ +#define UCNV_EXT_MAX_BYTES 0x1f + +#define UCNV_EXT_FROM_U_IS_PARTIAL(value) (((value)>>UCNV_EXT_FROM_U_LENGTH_SHIFT)==0) +#define UCNV_EXT_FROM_U_GET_PARTIAL_INDEX(value) (value) + +#define UCNV_EXT_FROM_U_IS_ROUNDTRIP(value) (((value)&UCNV_EXT_FROM_U_ROUNDTRIP_FLAG)!=0) +#define UCNV_EXT_FROM_U_MASK_ROUNDTRIP(value) ((value)&~UCNV_EXT_FROM_U_ROUNDTRIP_FLAG) + +/* get length; masks away all other bits */ +#define UCNV_EXT_FROM_U_GET_LENGTH(value) (int32_t)(((value)>>UCNV_EXT_FROM_U_LENGTH_SHIFT)&UCNV_EXT_MAX_BYTES) + +/* get bytes or bytes index */ +#define UCNV_EXT_FROM_U_GET_DATA(value) ((value)&UCNV_EXT_FROM_U_DATA_MASK) + +#endif + +#endif diff --git a/intl/icu/source/common/ucnv_imp.h b/intl/icu/source/common/ucnv_imp.h new file mode 100644 index 0000000000..cb939911e4 --- /dev/null +++ b/intl/icu/source/common/ucnv_imp.h @@ -0,0 +1,139 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1999-2011, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* +* +* ucnv_imp.h: +* Contains all internal and external data structure definitions +* Created & Maintained by Bertrand A. Damiba +* +* +* +* ATTENTION: +* --------- +* Although the data structures in this file are open and stack allocatable +* we reserve the right to hide them in further releases. +*/ + +#ifndef UCNV_IMP_H +#define UCNV_IMP_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/uloc.h" +#include "ucnv_bld.h" + +/* + * Fast check for whether a charset name is "UTF-8". + * This does not recognize all of the variations that ucnv_open() + * and other functions recognize, but it covers most cases. + * @param name const char * charset name + * @return + */ +#define UCNV_FAST_IS_UTF8(name) \ + (((name[0]=='U' ? \ + ( name[1]=='T' && name[2]=='F') : \ + (name[0]=='u' && name[1]=='t' && name[2]=='f'))) \ + && (name[3]=='-' ? \ + (name[4]=='8' && name[5]==0) : \ + (name[3]=='8' && name[4]==0))) + +typedef struct { + char cnvName[UCNV_MAX_CONVERTER_NAME_LENGTH]; + char locale[ULOC_FULLNAME_CAPACITY]; + uint32_t options; +} UConverterNamePieces; + +U_CFUNC UBool +ucnv_canCreateConverter(const char *converterName, UErrorCode *err); + +/* figures out if we need to go to file to read in the data tables. + * @param converterName The name of the converter + * @param err The error code + * @return the newly created converter + */ +U_CAPI UConverter * +ucnv_createConverter(UConverter *myUConverter, const char *converterName, UErrorCode * err); + +/* + * Open a purely algorithmic converter, specified by a type constant. + * @param myUConverter NULL, or pre-allocated UConverter structure to avoid + * a memory allocation + * @param type requested converter type + * @param locale locale parameter, or "" + * @param options converter options bit set (default 0) + * @param err ICU error code, not tested for U_FAILURE on input + * because this is an internal function + * @internal + */ +U_CFUNC UConverter * +ucnv_createAlgorithmicConverter(UConverter *myUConverter, + UConverterType type, + const char *locale, uint32_t options, + UErrorCode *err); + +/* + * Creates a converter from shared data. + * Adopts mySharedConverterData: No matter what happens, the caller must not + * unload mySharedConverterData, except via ucnv_close(return value) + * if this function is successful. + */ +U_CFUNC UConverter * +ucnv_createConverterFromSharedData(UConverter *myUConverter, + UConverterSharedData *mySharedConverterData, + UConverterLoadArgs *pArgs, + UErrorCode *err); + +U_CFUNC UConverter * +ucnv_createConverterFromPackage(const char *packageName, const char *converterName, UErrorCode *err); + +/** + * Load a converter but do not create a UConverter object. + * Simply return the UConverterSharedData. + * Performs alias lookup etc. + * The UConverterNamePieces need not be initialized + * before calling this function. + * The UConverterLoadArgs must be initialized + * before calling this function. + * If the args are passed in, then the pieces must be passed in too. + * In other words, the following combinations are allowed: + * - pieces==NULL && args==NULL + * - pieces!=NULL && args==NULL + * - pieces!=NULL && args!=NULL + * @internal + */ +U_CFUNC UConverterSharedData * +ucnv_loadSharedData(const char *converterName, + UConverterNamePieces *pieces, + UConverterLoadArgs *pArgs, + UErrorCode * err); + +/** + * This may unload the shared data in a thread safe manner. + * This will only unload the data if no other converters are sharing it. + */ +U_CFUNC void +ucnv_unloadSharedDataIfReady(UConverterSharedData *sharedData); + +/** + * This is a thread safe way to increment the reference count. + */ +U_CFUNC void +ucnv_incrementRefCount(UConverterSharedData *sharedData); + +/** + * These are the default error handling callbacks for the charset conversion framework. + * For performance reasons, they are only called to handle an error (not normally called for a reset or close). + */ +#define UCNV_TO_U_DEFAULT_CALLBACK ((UConverterToUCallback) UCNV_TO_U_CALLBACK_SUBSTITUTE) +#define UCNV_FROM_U_DEFAULT_CALLBACK ((UConverterFromUCallback) UCNV_FROM_U_CALLBACK_SUBSTITUTE) + +#endif + +#endif /* _UCNV_IMP */ diff --git a/intl/icu/source/common/ucnv_io.cpp b/intl/icu/source/common/ucnv_io.cpp new file mode 100644 index 0000000000..48bb5be42b --- /dev/null +++ b/intl/icu/source/common/ucnv_io.cpp @@ -0,0 +1,1360 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2015, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* +* ucnv_io.cpp: +* initializes global variables and defines functions pertaining to converter +* name resolution aspect of the conversion code. +* +* new implementation: +* +* created on: 1999nov22 +* created by: Markus W. Scherer +* +* Use the binary cnvalias.icu (created from convrtrs.txt) to work +* with aliases for converter names. +* +* Date Name Description +* 11/22/1999 markus Created +* 06/28/2002 grhoten Major overhaul of the converter alias design. +* Now an alias can map to different converters +* depending on the specified standard. +******************************************************************************* +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/udata.h" + +#include "umutex.h" +#include "uarrsort.h" +#include "uassert.h" +#include "udataswp.h" +#include "cstring.h" +#include "cmemory.h" +#include "ucnv_io.h" +#include "uenumimp.h" +#include "ucln_cmn.h" + +/* Format of cnvalias.icu ----------------------------------------------------- + * + * cnvalias.icu is a binary, memory-mappable form of convrtrs.txt. + * This binary form contains several tables. All indexes are to uint16_t + * units, and not to the bytes (uint8_t units). Addressing everything on + * 16-bit boundaries allows us to store more information with small index + * numbers, which are also 16-bit in size. The majority of the table (except + * the string table) are 16-bit numbers. + * + * First there is the size of the Table of Contents (TOC). The TOC + * entries contain the size of each section. In order to find the offset + * you just need to sum up the previous offsets. + * The TOC length and entries are an array of uint32_t values. + * The first section after the TOC starts immediately after the TOC. + * + * 1) This section contains a list of converters. This list contains indexes + * into the string table for the converter name. The index of this list is + * also used by other sections, which are mentioned later on. + * This list is not sorted. + * + * 2) This section contains a list of tags. This list contains indexes + * into the string table for the tag name. The index of this list is + * also used by other sections, which are mentioned later on. + * This list is in priority order of standards. + * + * 3) This section contains a list of sorted unique aliases. This + * list contains indexes into the string table for the alias name. The + * index of this list is also used by other sections, like the 4th section. + * The index for the 3rd and 4th section is used to get the + * alias -> converter name mapping. Section 3 and 4 form a two column table. + * Some of the most significant bits of each index may contain other + * information (see findConverter for details). + * + * 4) This section contains a list of mapped converter names. Consider this + * as a table that maps the 3rd section to the 1st section. This list contains + * indexes into the 1st section. The index of this list is the same index in + * the 3rd section. There is also some extra information in the high bits of + * each converter index in this table. Currently it's only used to say that + * an alias mapped to this converter is ambiguous. See UCNV_CONVERTER_INDEX_MASK + * and UCNV_AMBIGUOUS_ALIAS_MAP_BIT for more information. This section is + * the predigested form of the 5th section so that an alias lookup can be fast. + * + * 5) This section contains a 2D array with indexes to the 6th section. This + * section is the full form of all alias mappings. The column index is the + * index into the converter list (column header). The row index is the index + * to tag list (row header). This 2D array is the top part a 3D array. The + * third dimension is in the 6th section. + * + * 6) This is blob of variable length arrays. Each array starts with a size, + * and is followed by indexes to alias names in the string table. This is + * the third dimension to the section 5. No other section should be referencing + * this section. + * + * 7) Starting in ICU 3.6, this can be a UConverterAliasOptions struct. Its + * presence indicates that a section 9 exists. UConverterAliasOptions specifies + * what type of string normalization is used among other potential things in the + * future. + * + * 8) This is the string table. All strings are indexed on an even address. + * There are two reasons for this. First many chip architectures locate strings + * faster on even address boundaries. Second, since all indexes are 16-bit + * numbers, this string table can be 128KB in size instead of 64KB when we + * only have strings starting on an even address. + * + * 9) When present this is a set of prenormalized strings from section 8. This + * table contains normalized strings with the dashes and spaces stripped out, + * and all strings lowercased. In the future, the options in section 7 may state + * other types of normalization. + * + * Here is the concept of section 5 and 6. It's a 3D cube. Each tag + * has a unique alias among all converters. That same alias can + * be mentioned in other standards on different converters, + * but only one alias per tag can be unique. + * + * + * Converter Names (Usually in TR22 form) + * -------------------------------------------. + * T / /| + * a / / | + * g / / | + * s / / | + * / / | + * ------------------------------------------/ | + * A | | | + * l | | | + * i | | / + * a | | / + * s | | / + * e | | / + * s | |/ + * ------------------------------------------- + * + * + * + * Here is what it really looks like. It's like swiss cheese. + * There are holes. Some converters aren't recognized by + * a standard, or they are really old converters that the + * standard doesn't recognize anymore. + * + * Converter Names (Usually in TR22 form) + * -------------------------------------------. + * T /##########################################/| + * a / # # /# + * g / # ## ## ### # ### ### ### #/ + * s / # ##### #### ## ## #/# + * / ### # # ## # # # ### # # #/## + * ------------------------------------------/# # + * A |### # # ## # # # ### # # #|# # + * l |# # # # # ## # #|# # + * i |# # # # # # #|# + * a |# #|# + * s | #|# + * e + * s + * + */ + +/** + * Used by the UEnumeration API + */ +typedef struct UAliasContext { + uint32_t listOffset; + uint32_t listIdx; +} UAliasContext; + +static const char DATA_NAME[] = "cnvalias"; +static const char DATA_TYPE[] = "icu"; + +static UDataMemory *gAliasData=nullptr; +static icu::UInitOnce gAliasDataInitOnce {}; + +enum { + tocLengthIndex=0, + converterListIndex=1, + tagListIndex=2, + aliasListIndex=3, + untaggedConvArrayIndex=4, + taggedAliasArrayIndex=5, + taggedAliasListsIndex=6, + tableOptionsIndex=7, + stringTableIndex=8, + normalizedStringTableIndex=9, + offsetsCount, /* length of the swapper's temporary offsets[] */ + minTocLength=8 /* min. tocLength in the file, does not count the tocLengthIndex! */ +}; + +static const UConverterAliasOptions defaultTableOptions = { + UCNV_IO_UNNORMALIZED, + 0 /* containsCnvOptionInfo */ +}; +static UConverterAlias gMainTable; + +#define GET_STRING(idx) (const char *)(gMainTable.stringTable + (idx)) +#define GET_NORMALIZED_STRING(idx) (const char *)(gMainTable.normalizedStringTable + (idx)) + +static UBool U_CALLCONV +isAcceptable(void * /*context*/, + const char * /*type*/, const char * /*name*/, + const UDataInfo *pInfo) { + return (UBool)( + pInfo->size>=20 && + pInfo->isBigEndian==U_IS_BIG_ENDIAN && + pInfo->charsetFamily==U_CHARSET_FAMILY && + pInfo->dataFormat[0]==0x43 && /* dataFormat="CvAl" */ + pInfo->dataFormat[1]==0x76 && + pInfo->dataFormat[2]==0x41 && + pInfo->dataFormat[3]==0x6c && + pInfo->formatVersion[0]==3); +} + +static UBool U_CALLCONV ucnv_io_cleanup() +{ + if (gAliasData) { + udata_close(gAliasData); + gAliasData = nullptr; + } + gAliasDataInitOnce.reset(); + + uprv_memset(&gMainTable, 0, sizeof(gMainTable)); + + return true; /* Everything was cleaned up */ +} + +static void U_CALLCONV initAliasData(UErrorCode &errCode) { + UDataMemory *data; + const uint16_t *table; + const uint32_t *sectionSizes; + uint32_t tableStart; + uint32_t currOffset; + + ucln_common_registerCleanup(UCLN_COMMON_UCNV_IO, ucnv_io_cleanup); + + U_ASSERT(gAliasData == nullptr); + data = udata_openChoice(nullptr, DATA_TYPE, DATA_NAME, isAcceptable, nullptr, &errCode); + if(U_FAILURE(errCode)) { + return; + } + + sectionSizes = (const uint32_t *)udata_getMemory(data); + table = (const uint16_t *)sectionSizes; + + tableStart = sectionSizes[0]; + if (tableStart < minTocLength) { + errCode = U_INVALID_FORMAT_ERROR; + udata_close(data); + return; + } + gAliasData = data; + + gMainTable.converterListSize = sectionSizes[1]; + gMainTable.tagListSize = sectionSizes[2]; + gMainTable.aliasListSize = sectionSizes[3]; + gMainTable.untaggedConvArraySize = sectionSizes[4]; + gMainTable.taggedAliasArraySize = sectionSizes[5]; + gMainTable.taggedAliasListsSize = sectionSizes[6]; + gMainTable.optionTableSize = sectionSizes[7]; + gMainTable.stringTableSize = sectionSizes[8]; + + if (tableStart > 8) { + gMainTable.normalizedStringTableSize = sectionSizes[9]; + } + + currOffset = tableStart * (sizeof(uint32_t)/sizeof(uint16_t)) + (sizeof(uint32_t)/sizeof(uint16_t)); + gMainTable.converterList = table + currOffset; + + currOffset += gMainTable.converterListSize; + gMainTable.tagList = table + currOffset; + + currOffset += gMainTable.tagListSize; + gMainTable.aliasList = table + currOffset; + + currOffset += gMainTable.aliasListSize; + gMainTable.untaggedConvArray = table + currOffset; + + currOffset += gMainTable.untaggedConvArraySize; + gMainTable.taggedAliasArray = table + currOffset; + + /* aliasLists is a 1's based array, but it has a padding character */ + currOffset += gMainTable.taggedAliasArraySize; + gMainTable.taggedAliasLists = table + currOffset; + + currOffset += gMainTable.taggedAliasListsSize; + if (gMainTable.optionTableSize > 0 + && ((const UConverterAliasOptions *)(table + currOffset))->stringNormalizationType < UCNV_IO_NORM_TYPE_COUNT) + { + /* Faster table */ + gMainTable.optionTable = (const UConverterAliasOptions *)(table + currOffset); + } + else { + /* Smaller table, or I can't handle this normalization mode! + Use the original slower table lookup. */ + gMainTable.optionTable = &defaultTableOptions; + } + + currOffset += gMainTable.optionTableSize; + gMainTable.stringTable = table + currOffset; + + currOffset += gMainTable.stringTableSize; + gMainTable.normalizedStringTable = ((gMainTable.optionTable->stringNormalizationType == UCNV_IO_UNNORMALIZED) + ? gMainTable.stringTable : (table + currOffset)); +} + + +static UBool +haveAliasData(UErrorCode *pErrorCode) { + umtx_initOnce(gAliasDataInitOnce, &initAliasData, *pErrorCode); + return U_SUCCESS(*pErrorCode); +} + +static inline UBool +isAlias(const char *alias, UErrorCode *pErrorCode) { + if(alias==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return false; + } + return (UBool)(*alias!=0); +} + +static uint32_t getTagNumber(const char *tagname) { + if (gMainTable.tagList) { + uint32_t tagNum; + for (tagNum = 0; tagNum < gMainTable.tagListSize; tagNum++) { + if (!uprv_stricmp(GET_STRING(gMainTable.tagList[tagNum]), tagname)) { + return tagNum; + } + } + } + + return UINT32_MAX; +} + +/* character types relevant for ucnv_compareNames() */ +enum { + UIGNORE, + ZERO, + NONZERO, + MINLETTER /* any values from here on are lowercase letter mappings */ +}; + +/* character types for ASCII 00..7F */ +static const uint8_t asciiTypes[128] = { + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + ZERO, NONZERO, NONZERO, NONZERO, NONZERO, NONZERO, NONZERO, NONZERO, NONZERO, NONZERO, 0, 0, 0, 0, 0, 0, + 0, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, + 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0, 0, 0, 0, 0, + 0, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, + 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0, 0, 0, 0, 0 +}; + +#define GET_ASCII_TYPE(c) ((int8_t)(c) >= 0 ? asciiTypes[(uint8_t)c] : (uint8_t)UIGNORE) + +/* character types for EBCDIC 80..FF */ +static const uint8_t ebcdicTypes[128] = { + 0, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0, 0, 0, 0, 0, 0, + 0, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 0, 0, 0, 0, 0, 0, + 0, 0, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0, 0, 0, 0, 0, 0, + 0, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 0, 0, 0, 0, 0, 0, + 0, 0, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0, 0, 0, 0, 0, 0, + ZERO, NONZERO, NONZERO, NONZERO, NONZERO, NONZERO, NONZERO, NONZERO, NONZERO, NONZERO, 0, 0, 0, 0, 0, 0 +}; + +#define GET_EBCDIC_TYPE(c) ((int8_t)(c) < 0 ? ebcdicTypes[(c)&0x7f] : (uint8_t)UIGNORE) + +#if U_CHARSET_FAMILY==U_ASCII_FAMILY +# define GET_CHAR_TYPE(c) GET_ASCII_TYPE(c) +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY +# define GET_CHAR_TYPE(c) GET_EBCDIC_TYPE(c) +#else +# error U_CHARSET_FAMILY is not valid +#endif + + +/* @see ucnv_compareNames */ +U_CAPI char * U_CALLCONV +ucnv_io_stripASCIIForCompare(char *dst, const char *name) { + char *dstItr = dst; + uint8_t type, nextType; + char c1; + UBool afterDigit = false; + + while ((c1 = *name++) != 0) { + type = GET_ASCII_TYPE(c1); + switch (type) { + case UIGNORE: + afterDigit = false; + continue; /* ignore all but letters and digits */ + case ZERO: + if (!afterDigit) { + nextType = GET_ASCII_TYPE(*name); + if (nextType == ZERO || nextType == NONZERO) { + continue; /* ignore leading zero before another digit */ + } + } + break; + case NONZERO: + afterDigit = true; + break; + default: + c1 = (char)type; /* lowercased letter */ + afterDigit = false; + break; + } + *dstItr++ = c1; + } + *dstItr = 0; + return dst; +} + +U_CAPI char * U_CALLCONV +ucnv_io_stripEBCDICForCompare(char *dst, const char *name) { + char *dstItr = dst; + uint8_t type, nextType; + char c1; + UBool afterDigit = false; + + while ((c1 = *name++) != 0) { + type = GET_EBCDIC_TYPE(c1); + switch (type) { + case UIGNORE: + afterDigit = false; + continue; /* ignore all but letters and digits */ + case ZERO: + if (!afterDigit) { + nextType = GET_EBCDIC_TYPE(*name); + if (nextType == ZERO || nextType == NONZERO) { + continue; /* ignore leading zero before another digit */ + } + } + break; + case NONZERO: + afterDigit = true; + break; + default: + c1 = (char)type; /* lowercased letter */ + afterDigit = false; + break; + } + *dstItr++ = c1; + } + *dstItr = 0; + return dst; +} + +/** + * Do a fuzzy compare of two converter/alias names. + * The comparison is case-insensitive, ignores leading zeroes if they are not + * followed by further digits, and ignores all but letters and digits. + * Thus the strings "UTF-8", "utf_8", "u*T@f08" and "Utf 8" are exactly equivalent. + * See section 1.4, Charset Alias Matching in Unicode Technical Standard #22 + * at http://www.unicode.org/reports/tr22/ + * + * This is a symmetrical (commutative) operation; order of arguments + * is insignificant. This is an important property for sorting the + * list (when the list is preprocessed into binary form) and for + * performing binary searches on it at run time. + * + * @param name1 a converter name or alias, zero-terminated + * @param name2 a converter name or alias, zero-terminated + * @return 0 if the names match, or a negative value if the name1 + * lexically precedes name2, or a positive value if the name1 + * lexically follows name2. + * + * @see ucnv_io_stripForCompare + */ +U_CAPI int U_EXPORT2 +ucnv_compareNames(const char *name1, const char *name2) { + int rc; + uint8_t type, nextType; + char c1, c2; + UBool afterDigit1 = false, afterDigit2 = false; + + for (;;) { + while ((c1 = *name1++) != 0) { + type = GET_CHAR_TYPE(c1); + switch (type) { + case UIGNORE: + afterDigit1 = false; + continue; /* ignore all but letters and digits */ + case ZERO: + if (!afterDigit1) { + nextType = GET_CHAR_TYPE(*name1); + if (nextType == ZERO || nextType == NONZERO) { + continue; /* ignore leading zero before another digit */ + } + } + break; + case NONZERO: + afterDigit1 = true; + break; + default: + c1 = (char)type; /* lowercased letter */ + afterDigit1 = false; + break; + } + break; /* deliver c1 */ + } + while ((c2 = *name2++) != 0) { + type = GET_CHAR_TYPE(c2); + switch (type) { + case UIGNORE: + afterDigit2 = false; + continue; /* ignore all but letters and digits */ + case ZERO: + if (!afterDigit2) { + nextType = GET_CHAR_TYPE(*name2); + if (nextType == ZERO || nextType == NONZERO) { + continue; /* ignore leading zero before another digit */ + } + } + break; + case NONZERO: + afterDigit2 = true; + break; + default: + c2 = (char)type; /* lowercased letter */ + afterDigit2 = false; + break; + } + break; /* deliver c2 */ + } + + /* If we reach the ends of both strings then they match */ + if ((c1|c2)==0) { + return 0; + } + + /* Case-insensitive comparison */ + rc = (int)(unsigned char)c1 - (int)(unsigned char)c2; + if (rc != 0) { + return rc; + } + } +} + +/* + * search for an alias + * return the converter number index for gConverterList + */ +static inline uint32_t +findConverter(const char *alias, UBool *containsOption, UErrorCode *pErrorCode) { + uint32_t mid, start, limit; + uint32_t lastMid; + int result; + int isUnnormalized = (gMainTable.optionTable->stringNormalizationType == UCNV_IO_UNNORMALIZED); + char strippedName[UCNV_MAX_CONVERTER_NAME_LENGTH]; + + if (!isUnnormalized) { + if (uprv_strlen(alias) >= UCNV_MAX_CONVERTER_NAME_LENGTH) { + *pErrorCode = U_BUFFER_OVERFLOW_ERROR; + return UINT32_MAX; + } + + /* Lower case and remove ignoreable characters. */ + ucnv_io_stripForCompare(strippedName, alias); + alias = strippedName; + } + + /* do a binary search for the alias */ + start = 0; + limit = gMainTable.untaggedConvArraySize; + mid = limit; + lastMid = UINT32_MAX; + + for (;;) { + mid = (uint32_t)((start + limit) / 2); + if (lastMid == mid) { /* Have we moved? */ + break; /* We haven't moved, and it wasn't found. */ + } + lastMid = mid; + if (isUnnormalized) { + result = ucnv_compareNames(alias, GET_STRING(gMainTable.aliasList[mid])); + } + else { + result = uprv_strcmp(alias, GET_NORMALIZED_STRING(gMainTable.aliasList[mid])); + } + + if (result < 0) { + limit = mid; + } else if (result > 0) { + start = mid; + } else { + /* Since the gencnval tool folds duplicates into one entry, + * this alias in gAliasList is unique, but different standards + * may map an alias to different converters. + */ + if (gMainTable.untaggedConvArray[mid] & UCNV_AMBIGUOUS_ALIAS_MAP_BIT) { + *pErrorCode = U_AMBIGUOUS_ALIAS_WARNING; + } + /* State whether the canonical converter name contains an option. + This information is contained in this list in order to maintain backward & forward compatibility. */ + if (containsOption) { + UBool containsCnvOptionInfo = (UBool)gMainTable.optionTable->containsCnvOptionInfo; + *containsOption = (UBool)((containsCnvOptionInfo + && ((gMainTable.untaggedConvArray[mid] & UCNV_CONTAINS_OPTION_BIT) != 0)) + || !containsCnvOptionInfo); + } + return gMainTable.untaggedConvArray[mid] & UCNV_CONVERTER_INDEX_MASK; + } + } + + return UINT32_MAX; +} + +/* + * Is this alias in this list? + * alias and listOffset should be non-nullptr. + */ +static inline UBool +isAliasInList(const char *alias, uint32_t listOffset) { + if (listOffset) { + uint32_t currAlias; + uint32_t listCount = gMainTable.taggedAliasLists[listOffset]; + /* +1 to skip listCount */ + const uint16_t *currList = gMainTable.taggedAliasLists + listOffset + 1; + for (currAlias = 0; currAlias < listCount; currAlias++) { + if (currList[currAlias] + && ucnv_compareNames(alias, GET_STRING(currList[currAlias]))==0) + { + return true; + } + } + } + return false; +} + +/* + * Search for an standard name of an alias (what is the default name + * that this standard uses?) + * return the listOffset for gTaggedAliasLists. If it's 0, + * the it couldn't be found, but the parameters are valid. + */ +static uint32_t +findTaggedAliasListsOffset(const char *alias, const char *standard, UErrorCode *pErrorCode) { + uint32_t idx; + uint32_t listOffset; + uint32_t convNum; + UErrorCode myErr = U_ZERO_ERROR; + uint32_t tagNum = getTagNumber(standard); + + /* Make a quick guess. Hopefully they used a TR22 canonical alias. */ + convNum = findConverter(alias, nullptr, &myErr); + if (myErr != U_ZERO_ERROR) { + *pErrorCode = myErr; + } + + if (tagNum < (gMainTable.tagListSize - UCNV_NUM_HIDDEN_TAGS) && convNum < gMainTable.converterListSize) { + listOffset = gMainTable.taggedAliasArray[tagNum*gMainTable.converterListSize + convNum]; + if (listOffset && gMainTable.taggedAliasLists[listOffset + 1]) { + return listOffset; + } + if (myErr == U_AMBIGUOUS_ALIAS_WARNING) { + /* Uh Oh! They used an ambiguous alias. + We have to search the whole swiss cheese starting + at the highest standard affinity. + This may take a while. + */ + for (idx = 0; idx < gMainTable.taggedAliasArraySize; idx++) { + listOffset = gMainTable.taggedAliasArray[idx]; + if (listOffset && isAliasInList(alias, listOffset)) { + uint32_t currTagNum = idx/gMainTable.converterListSize; + uint32_t currConvNum = (idx - currTagNum*gMainTable.converterListSize); + uint32_t tempListOffset = gMainTable.taggedAliasArray[tagNum*gMainTable.converterListSize + currConvNum]; + if (tempListOffset && gMainTable.taggedAliasLists[tempListOffset + 1]) { + return tempListOffset; + } + /* else keep on looking */ + /* We could speed this up by starting on the next row + because an alias is unique per row, right now. + This would change if alias versioning appears. */ + } + } + /* The standard doesn't know about the alias */ + } + /* else no default name */ + return 0; + } + /* else converter or tag not found */ + + return UINT32_MAX; +} + +/* Return the canonical name */ +static uint32_t +findTaggedConverterNum(const char *alias, const char *standard, UErrorCode *pErrorCode) { + uint32_t idx; + uint32_t listOffset; + uint32_t convNum; + UErrorCode myErr = U_ZERO_ERROR; + uint32_t tagNum = getTagNumber(standard); + + /* Make a quick guess. Hopefully they used a TR22 canonical alias. */ + convNum = findConverter(alias, nullptr, &myErr); + if (myErr != U_ZERO_ERROR) { + *pErrorCode = myErr; + } + + if (tagNum < (gMainTable.tagListSize - UCNV_NUM_HIDDEN_TAGS) && convNum < gMainTable.converterListSize) { + listOffset = gMainTable.taggedAliasArray[tagNum*gMainTable.converterListSize + convNum]; + if (listOffset && isAliasInList(alias, listOffset)) { + return convNum; + } + if (myErr == U_AMBIGUOUS_ALIAS_WARNING) { + /* Uh Oh! They used an ambiguous alias. + We have to search one slice of the swiss cheese. + We search only in the requested tag, not the whole thing. + This may take a while. + */ + uint32_t convStart = (tagNum)*gMainTable.converterListSize; + uint32_t convLimit = (tagNum+1)*gMainTable.converterListSize; + for (idx = convStart; idx < convLimit; idx++) { + listOffset = gMainTable.taggedAliasArray[idx]; + if (listOffset && isAliasInList(alias, listOffset)) { + return idx-convStart; + } + } + /* The standard doesn't know about the alias */ + } + /* else no canonical name */ + } + /* else converter or tag not found */ + + return UINT32_MAX; +} + +U_CAPI const char * +ucnv_io_getConverterName(const char *alias, UBool *containsOption, UErrorCode *pErrorCode) { + const char *aliasTmp = alias; + int32_t i = 0; + for (i = 0; i < 2; i++) { + if (i == 1) { + /* + * After the first unsuccess converter lookup, check to see if + * the name begins with 'x-'. If it does, strip it off and try + * again. This behaviour is similar to how ICU4J does it. + */ + if (aliasTmp[0] == 'x' && aliasTmp[1] == '-') { + aliasTmp = aliasTmp+2; + } else { + break; + } + } + if(haveAliasData(pErrorCode) && isAlias(aliasTmp, pErrorCode)) { + uint32_t convNum = findConverter(aliasTmp, containsOption, pErrorCode); + if (convNum < gMainTable.converterListSize) { + return GET_STRING(gMainTable.converterList[convNum]); + } + /* else converter not found */ + } else { + break; + } + } + + return nullptr; +} + +U_CDECL_BEGIN + + +static int32_t U_CALLCONV +ucnv_io_countStandardAliases(UEnumeration *enumerator, UErrorCode * /*pErrorCode*/) { + int32_t value = 0; + UAliasContext *myContext = (UAliasContext *)(enumerator->context); + uint32_t listOffset = myContext->listOffset; + + if (listOffset) { + value = gMainTable.taggedAliasLists[listOffset]; + } + return value; +} + +static const char * U_CALLCONV +ucnv_io_nextStandardAliases(UEnumeration *enumerator, + int32_t* resultLength, + UErrorCode * /*pErrorCode*/) +{ + UAliasContext *myContext = (UAliasContext *)(enumerator->context); + uint32_t listOffset = myContext->listOffset; + + if (listOffset) { + uint32_t listCount = gMainTable.taggedAliasLists[listOffset]; + const uint16_t *currList = gMainTable.taggedAliasLists + listOffset + 1; + + if (myContext->listIdx < listCount) { + const char *myStr = GET_STRING(currList[myContext->listIdx++]); + if (resultLength) { + *resultLength = (int32_t)uprv_strlen(myStr); + } + return myStr; + } + } + /* Either we accessed a zero length list, or we enumerated too far. */ + if (resultLength) { + *resultLength = 0; + } + return nullptr; +} + +static void U_CALLCONV +ucnv_io_resetStandardAliases(UEnumeration *enumerator, UErrorCode * /*pErrorCode*/) { + ((UAliasContext *)(enumerator->context))->listIdx = 0; +} + +static void U_CALLCONV +ucnv_io_closeUEnumeration(UEnumeration *enumerator) { + uprv_free(enumerator->context); + uprv_free(enumerator); +} + +U_CDECL_END + +/* Enumerate the aliases for the specified converter and standard tag */ +static const UEnumeration gEnumAliases = { + nullptr, + nullptr, + ucnv_io_closeUEnumeration, + ucnv_io_countStandardAliases, + uenum_unextDefault, + ucnv_io_nextStandardAliases, + ucnv_io_resetStandardAliases +}; + +U_CAPI UEnumeration * U_EXPORT2 +ucnv_openStandardNames(const char *convName, + const char *standard, + UErrorCode *pErrorCode) +{ + UEnumeration *myEnum = nullptr; + if (haveAliasData(pErrorCode) && isAlias(convName, pErrorCode)) { + uint32_t listOffset = findTaggedAliasListsOffset(convName, standard, pErrorCode); + + /* When listOffset == 0, we want to acknowledge that the + converter name and standard are okay, but there + is nothing to enumerate. */ + if (listOffset < gMainTable.taggedAliasListsSize) { + UAliasContext *myContext; + + myEnum = static_cast(uprv_malloc(sizeof(UEnumeration))); + if (myEnum == nullptr) { + *pErrorCode = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memcpy(myEnum, &gEnumAliases, sizeof(UEnumeration)); + myContext = static_cast(uprv_malloc(sizeof(UAliasContext))); + if (myContext == nullptr) { + *pErrorCode = U_MEMORY_ALLOCATION_ERROR; + uprv_free(myEnum); + return nullptr; + } + myContext->listOffset = listOffset; + myContext->listIdx = 0; + myEnum->context = myContext; + } + /* else converter or tag not found */ + } + return myEnum; +} + +static uint16_t +ucnv_io_countAliases(const char *alias, UErrorCode *pErrorCode) { + if(haveAliasData(pErrorCode) && isAlias(alias, pErrorCode)) { + uint32_t convNum = findConverter(alias, nullptr, pErrorCode); + if (convNum < gMainTable.converterListSize) { + /* tagListNum - 1 is the ALL tag */ + int32_t listOffset = gMainTable.taggedAliasArray[(gMainTable.tagListSize - 1)*gMainTable.converterListSize + convNum]; + + if (listOffset) { + return gMainTable.taggedAliasLists[listOffset]; + } + /* else this shouldn't happen. internal program error */ + } + /* else converter not found */ + } + return 0; +} + +static uint16_t +ucnv_io_getAliases(const char *alias, uint16_t start, const char **aliases, UErrorCode *pErrorCode) { + if(haveAliasData(pErrorCode) && isAlias(alias, pErrorCode)) { + uint32_t currAlias; + uint32_t convNum = findConverter(alias, nullptr, pErrorCode); + if (convNum < gMainTable.converterListSize) { + /* tagListNum - 1 is the ALL tag */ + int32_t listOffset = gMainTable.taggedAliasArray[(gMainTable.tagListSize - 1)*gMainTable.converterListSize + convNum]; + + if (listOffset) { + uint32_t listCount = gMainTable.taggedAliasLists[listOffset]; + /* +1 to skip listCount */ + const uint16_t *currList = gMainTable.taggedAliasLists + listOffset + 1; + + for (currAlias = start; currAlias < listCount; currAlias++) { + aliases[currAlias] = GET_STRING(currList[currAlias]); + } + } + /* else this shouldn't happen. internal program error */ + } + /* else converter not found */ + } + return 0; +} + +static const char * +ucnv_io_getAlias(const char *alias, uint16_t n, UErrorCode *pErrorCode) { + if(haveAliasData(pErrorCode) && isAlias(alias, pErrorCode)) { + uint32_t convNum = findConverter(alias, nullptr, pErrorCode); + if (convNum < gMainTable.converterListSize) { + /* tagListNum - 1 is the ALL tag */ + int32_t listOffset = gMainTable.taggedAliasArray[(gMainTable.tagListSize - 1)*gMainTable.converterListSize + convNum]; + + if (listOffset) { + uint32_t listCount = gMainTable.taggedAliasLists[listOffset]; + /* +1 to skip listCount */ + const uint16_t *currList = gMainTable.taggedAliasLists + listOffset + 1; + + if (n < listCount) { + return GET_STRING(currList[n]); + } + *pErrorCode = U_INDEX_OUTOFBOUNDS_ERROR; + } + /* else this shouldn't happen. internal program error */ + } + /* else converter not found */ + } + return nullptr; +} + +static uint16_t +ucnv_io_countStandards(UErrorCode *pErrorCode) { + if (haveAliasData(pErrorCode)) { + /* Don't include the empty list */ + return (uint16_t)(gMainTable.tagListSize - UCNV_NUM_HIDDEN_TAGS); + } + + return 0; +} + +U_CAPI const char * U_EXPORT2 +ucnv_getStandard(uint16_t n, UErrorCode *pErrorCode) { + if (haveAliasData(pErrorCode)) { + if (n < gMainTable.tagListSize - UCNV_NUM_HIDDEN_TAGS) { + return GET_STRING(gMainTable.tagList[n]); + } + *pErrorCode = U_INDEX_OUTOFBOUNDS_ERROR; + } + + return nullptr; +} + +U_CAPI const char * U_EXPORT2 +ucnv_getStandardName(const char *alias, const char *standard, UErrorCode *pErrorCode) { + if (haveAliasData(pErrorCode) && isAlias(alias, pErrorCode)) { + uint32_t listOffset = findTaggedAliasListsOffset(alias, standard, pErrorCode); + + if (0 < listOffset && listOffset < gMainTable.taggedAliasListsSize) { + const uint16_t *currList = gMainTable.taggedAliasLists + listOffset + 1; + + /* Get the preferred name from this list */ + if (currList[0]) { + return GET_STRING(currList[0]); + } + /* else someone screwed up the alias table. */ + /* *pErrorCode = U_INVALID_FORMAT_ERROR */ + } + } + + return nullptr; +} + +U_CAPI uint16_t U_EXPORT2 +ucnv_countAliases(const char *alias, UErrorCode *pErrorCode) +{ + return ucnv_io_countAliases(alias, pErrorCode); +} + + +U_CAPI const char* U_EXPORT2 +ucnv_getAlias(const char *alias, uint16_t n, UErrorCode *pErrorCode) +{ + return ucnv_io_getAlias(alias, n, pErrorCode); +} + +U_CAPI void U_EXPORT2 +ucnv_getAliases(const char *alias, const char **aliases, UErrorCode *pErrorCode) +{ + ucnv_io_getAliases(alias, 0, aliases, pErrorCode); +} + +U_CAPI uint16_t U_EXPORT2 +ucnv_countStandards() +{ + UErrorCode err = U_ZERO_ERROR; + return ucnv_io_countStandards(&err); +} + +U_CAPI const char * U_EXPORT2 +ucnv_getCanonicalName(const char *alias, const char *standard, UErrorCode *pErrorCode) { + if (haveAliasData(pErrorCode) && isAlias(alias, pErrorCode)) { + uint32_t convNum = findTaggedConverterNum(alias, standard, pErrorCode); + + if (convNum < gMainTable.converterListSize) { + return GET_STRING(gMainTable.converterList[convNum]); + } + } + + return nullptr; +} + +U_CDECL_BEGIN + + +static int32_t U_CALLCONV +ucnv_io_countAllConverters(UEnumeration * /*enumerator*/, UErrorCode * /*pErrorCode*/) { + return gMainTable.converterListSize; +} + +static const char * U_CALLCONV +ucnv_io_nextAllConverters(UEnumeration *enumerator, + int32_t* resultLength, + UErrorCode * /*pErrorCode*/) +{ + uint16_t *myContext = (uint16_t *)(enumerator->context); + + if (*myContext < gMainTable.converterListSize) { + const char *myStr = GET_STRING(gMainTable.converterList[(*myContext)++]); + if (resultLength) { + *resultLength = (int32_t)uprv_strlen(myStr); + } + return myStr; + } + /* Either we accessed a zero length list, or we enumerated too far. */ + if (resultLength) { + *resultLength = 0; + } + return nullptr; +} + +static void U_CALLCONV +ucnv_io_resetAllConverters(UEnumeration *enumerator, UErrorCode * /*pErrorCode*/) { + *((uint16_t *)(enumerator->context)) = 0; +} +U_CDECL_END +static const UEnumeration gEnumAllConverters = { + nullptr, + nullptr, + ucnv_io_closeUEnumeration, + ucnv_io_countAllConverters, + uenum_unextDefault, + ucnv_io_nextAllConverters, + ucnv_io_resetAllConverters +}; + +U_CAPI UEnumeration * U_EXPORT2 +ucnv_openAllNames(UErrorCode *pErrorCode) { + UEnumeration *myEnum = nullptr; + if (haveAliasData(pErrorCode)) { + uint16_t *myContext; + + myEnum = static_cast(uprv_malloc(sizeof(UEnumeration))); + if (myEnum == nullptr) { + *pErrorCode = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memcpy(myEnum, &gEnumAllConverters, sizeof(UEnumeration)); + myContext = static_cast(uprv_malloc(sizeof(uint16_t))); + if (myContext == nullptr) { + *pErrorCode = U_MEMORY_ALLOCATION_ERROR; + uprv_free(myEnum); + return nullptr; + } + *myContext = 0; + myEnum->context = myContext; + } + return myEnum; +} + +U_CAPI uint16_t +ucnv_io_countKnownConverters(UErrorCode *pErrorCode) { + if (haveAliasData(pErrorCode)) { + return (uint16_t)gMainTable.converterListSize; + } + return 0; +} + +/* alias table swapping ----------------------------------------------------- */ + +U_CDECL_BEGIN + +typedef char * U_CALLCONV StripForCompareFn(char *dst, const char *name); +U_CDECL_END + + +/* + * row of a temporary array + * + * gets platform-endian charset string indexes and sorting indexes; + * after sorting this array by strings, the actual arrays are permutated + * according to the sorting indexes + */ +typedef struct TempRow { + uint16_t strIndex, sortIndex; +} TempRow; + +typedef struct TempAliasTable { + const char *chars; + TempRow *rows; + uint16_t *resort; + StripForCompareFn *stripForCompare; +} TempAliasTable; + +enum { + STACK_ROW_CAPACITY=500 +}; + +static int32_t U_CALLCONV +io_compareRows(const void *context, const void *left, const void *right) { + char strippedLeft[UCNV_MAX_CONVERTER_NAME_LENGTH], + strippedRight[UCNV_MAX_CONVERTER_NAME_LENGTH]; + + TempAliasTable *tempTable=(TempAliasTable *)context; + const char *chars=tempTable->chars; + + return (int32_t)uprv_strcmp(tempTable->stripForCompare(strippedLeft, chars+2*((const TempRow *)left)->strIndex), + tempTable->stripForCompare(strippedRight, chars+2*((const TempRow *)right)->strIndex)); +} + +U_CAPI int32_t U_EXPORT2 +ucnv_swapAliases(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const UDataInfo *pInfo; + int32_t headerSize; + + const uint16_t *inTable; + const uint32_t *inSectionSizes; + uint32_t toc[offsetsCount]; + uint32_t offsets[offsetsCount]; /* 16-bit-addressed offsets from inTable/outTable */ + uint32_t i, count, tocLength, topOffset; + + TempRow rows[STACK_ROW_CAPACITY]; + uint16_t resort[STACK_ROW_CAPACITY]; + TempAliasTable tempTable; + + /* udata_swapDataHeader checks the arguments */ + headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode); + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + /* check data format and format version */ + pInfo=(const UDataInfo *)((const char *)inData+4); + if(!( + pInfo->dataFormat[0]==0x43 && /* dataFormat="CvAl" */ + pInfo->dataFormat[1]==0x76 && + pInfo->dataFormat[2]==0x41 && + pInfo->dataFormat[3]==0x6c && + pInfo->formatVersion[0]==3 + )) { + udata_printError(ds, "ucnv_swapAliases(): data format %02x.%02x.%02x.%02x (format version %02x) is not an alias table\n", + pInfo->dataFormat[0], pInfo->dataFormat[1], + pInfo->dataFormat[2], pInfo->dataFormat[3], + pInfo->formatVersion[0]); + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + + /* an alias table must contain at least the table of contents array */ + if(length>=0 && (length-headerSize)<4*(1+minTocLength)) { + udata_printError(ds, "ucnv_swapAliases(): too few bytes (%d after header) for an alias table\n", + length-headerSize); + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + + inSectionSizes=(const uint32_t *)((const char *)inData+headerSize); + inTable=(const uint16_t *)inSectionSizes; + uprv_memset(toc, 0, sizeof(toc)); + toc[tocLengthIndex]=tocLength=ds->readUInt32(inSectionSizes[tocLengthIndex]); + if(tocLengthreadUInt32(inSectionSizes[i]); + } + + /* compute offsets */ + uprv_memset(offsets, 0, sizeof(offsets)); + offsets[converterListIndex]=2*(1+tocLength); /* count two 16-bit units per toc entry */ + for(i=tagListIndex; i<=tocLength; ++i) { + offsets[i]=offsets[i-1]+toc[i-1]; + } + + /* compute the overall size of the after-header data, in numbers of 16-bit units */ + topOffset=offsets[i-1]+toc[i-1]; + + if(length>=0) { + uint16_t *outTable; + const uint16_t *p, *p2; + uint16_t *q, *q2; + uint16_t oldIndex; + + if((length-headerSize)<(2*(int32_t)topOffset)) { + udata_printError(ds, "ucnv_swapAliases(): too few bytes (%d after header) for an alias table\n", + length-headerSize); + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + + outTable=(uint16_t *)((char *)outData+headerSize); + + /* swap the entire table of contents */ + ds->swapArray32(ds, inTable, 4*(1+tocLength), outTable, pErrorCode); + + /* swap unormalized strings & normalized strings */ + ds->swapInvChars(ds, inTable+offsets[stringTableIndex], 2*(int32_t)(toc[stringTableIndex]+toc[normalizedStringTableIndex]), + outTable+offsets[stringTableIndex], pErrorCode); + if(U_FAILURE(*pErrorCode)) { + udata_printError(ds, "ucnv_swapAliases().swapInvChars(charset names) failed\n"); + return 0; + } + + if(ds->inCharset==ds->outCharset) { + /* no need to sort, just swap all 16-bit values together */ + ds->swapArray16(ds, + inTable+offsets[converterListIndex], + 2*(int32_t)(offsets[stringTableIndex]-offsets[converterListIndex]), + outTable+offsets[converterListIndex], + pErrorCode); + } else { + /* allocate the temporary table for sorting */ + count=toc[aliasListIndex]; + + tempTable.chars=(const char *)(outTable+offsets[stringTableIndex]); /* sort by outCharset */ + + if(count<=STACK_ROW_CAPACITY) { + tempTable.rows=rows; + tempTable.resort=resort; + } else { + tempTable.rows=(TempRow *)uprv_malloc(count*sizeof(TempRow)+count*2); + if(tempTable.rows==nullptr) { + udata_printError(ds, "ucnv_swapAliases(): unable to allocate memory for sorting tables (max length: %u)\n", + count); + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return 0; + } + tempTable.resort=(uint16_t *)(tempTable.rows+count); + } + + if(ds->outCharset==U_ASCII_FAMILY) { + tempTable.stripForCompare=ucnv_io_stripASCIIForCompare; + } else /* U_EBCDIC_FAMILY */ { + tempTable.stripForCompare=ucnv_io_stripEBCDICForCompare; + } + + /* + * Sort unique aliases+mapped names. + * + * We need to sort the list again by outCharset strings because they + * sort differently for different charset families. + * First we set up a temporary table with the string indexes and + * sorting indexes and sort that. + * Then we permutate and copy/swap the actual values. + */ + p=inTable+offsets[aliasListIndex]; + q=outTable+offsets[aliasListIndex]; + + p2=inTable+offsets[untaggedConvArrayIndex]; + q2=outTable+offsets[untaggedConvArrayIndex]; + + for(i=0; ireadUInt16(p[i]); + tempTable.rows[i].sortIndex=(uint16_t)i; + } + + uprv_sortArray(tempTable.rows, (int32_t)count, sizeof(TempRow), + io_compareRows, &tempTable, + false, pErrorCode); + + if(U_SUCCESS(*pErrorCode)) { + /* copy/swap/permutate items */ + if(p!=q) { + for(i=0; iswapArray16(ds, p+oldIndex, 2, q+i, pErrorCode); + ds->swapArray16(ds, p2+oldIndex, 2, q2+i, pErrorCode); + } + } else { + /* + * If we swap in-place, then the permutation must use another + * temporary array (tempTable.resort) + * before the results are copied to the outBundle. + */ + uint16_t *r=tempTable.resort; + + for(i=0; iswapArray16(ds, p+oldIndex, 2, r+i, pErrorCode); + } + uprv_memcpy(q, r, 2*(size_t)count); + + for(i=0; iswapArray16(ds, p2+oldIndex, 2, r+i, pErrorCode); + } + uprv_memcpy(q2, r, 2*(size_t)count); + } + } + + if(tempTable.rows!=rows) { + uprv_free(tempTable.rows); + } + + if(U_FAILURE(*pErrorCode)) { + udata_printError(ds, "ucnv_swapAliases().uprv_sortArray(%u items) failed\n", + count); + return 0; + } + + /* swap remaining 16-bit values */ + ds->swapArray16(ds, + inTable+offsets[converterListIndex], + 2*(int32_t)(offsets[aliasListIndex]-offsets[converterListIndex]), + outTable+offsets[converterListIndex], + pErrorCode); + ds->swapArray16(ds, + inTable+offsets[taggedAliasArrayIndex], + 2*(int32_t)(offsets[stringTableIndex]-offsets[taggedAliasArrayIndex]), + outTable+offsets[taggedAliasArrayIndex], + pErrorCode); + } + } + + return headerSize+2*(int32_t)topOffset; +} + +#endif + + +/* + * Hey, Emacs, please set the following: + * + * Local Variables: + * indent-tabs-mode: nil + * End: + * + */ diff --git a/intl/icu/source/common/ucnv_io.h b/intl/icu/source/common/ucnv_io.h new file mode 100644 index 0000000000..8f2d7b5a02 --- /dev/null +++ b/intl/icu/source/common/ucnv_io.h @@ -0,0 +1,127 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ********************************************************************** + * Copyright (C) 1999-2006, International Business Machines + * Corporation and others. All Rights Reserved. + ********************************************************************** + * + * + * ucnv_io.h: + * defines variables and functions pertaining to converter name resolution + * aspect of the conversion code + */ + +#ifndef UCNV_IO_H +#define UCNV_IO_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "udataswp.h" + +#define UCNV_AMBIGUOUS_ALIAS_MAP_BIT 0x8000 +#define UCNV_CONTAINS_OPTION_BIT 0x4000 +#define UCNV_CONVERTER_INDEX_MASK 0xFFF +#define UCNV_NUM_RESERVED_TAGS 2 +#define UCNV_NUM_HIDDEN_TAGS 1 + +enum { + UCNV_IO_UNNORMALIZED, + UCNV_IO_STD_NORMALIZED, + UCNV_IO_NORM_TYPE_COUNT +}; + +typedef struct { + uint16_t stringNormalizationType; + uint16_t containsCnvOptionInfo; +} UConverterAliasOptions; + +typedef struct UConverterAlias { + const uint16_t *converterList; + const uint16_t *tagList; + const uint16_t *aliasList; + const uint16_t *untaggedConvArray; + const uint16_t *taggedAliasArray; + const uint16_t *taggedAliasLists; + const UConverterAliasOptions *optionTable; + const uint16_t *stringTable; + const uint16_t *normalizedStringTable; + + uint32_t converterListSize; + uint32_t tagListSize; + uint32_t aliasListSize; + uint32_t untaggedConvArraySize; + uint32_t taggedAliasArraySize; + uint32_t taggedAliasListsSize; + uint32_t optionTableSize; + uint32_t stringTableSize; + uint32_t normalizedStringTableSize; +} UConverterAlias; + +/** + * \var ucnv_io_stripForCompare + * Remove the underscores, dashes and spaces from the name, and convert + * the name to lower case. + * @param dst The destination buffer, which is <= the buffer of name. + * @param dst The destination buffer, which is <= the buffer of name. + * @see ucnv_compareNames + * @return the destination buffer. + */ +#if U_CHARSET_FAMILY==U_ASCII_FAMILY +# define ucnv_io_stripForCompare ucnv_io_stripASCIIForCompare +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY +# define ucnv_io_stripForCompare ucnv_io_stripEBCDICForCompare +#else +# error U_CHARSET_FAMILY is not valid +#endif + +U_CAPI char * U_CALLCONV +ucnv_io_stripASCIIForCompare(char *dst, const char *name); + +U_CAPI char * U_CALLCONV +ucnv_io_stripEBCDICForCompare(char *dst, const char *name); + +/** + * Map a converter alias name to a canonical converter name. + * The alias is searched for case-insensitively, the converter name + * is returned in mixed-case. + * Returns NULL if the alias is not found. + * @param alias The alias name to be searched. + * @param containsOption A return value stating whether the returned converter name contains an option (a comma) + * @param pErrorCode The error code + * @return the converter name in mixed-case, return NULL if the alias is not found. + */ +U_CAPI const char * +ucnv_io_getConverterName(const char *alias, UBool *containsOption, UErrorCode *pErrorCode); + +/** + * Return the number of all known converter names (no aliases). + * @param pErrorCode The error code + * @return the number of all aliases + */ +U_CAPI uint16_t +ucnv_io_countKnownConverters(UErrorCode *pErrorCode); + +/** + * Swap an ICU converter alias table. See implementation for details. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +ucnv_swapAliases(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +#endif + +#endif /* _UCNV_IO */ + +/* + * Hey, Emacs, please set the following: + * + * Local Variables: + * indent-tabs-mode: nil + * End: + * + */ diff --git a/intl/icu/source/common/ucnv_lmb.cpp b/intl/icu/source/common/ucnv_lmb.cpp new file mode 100644 index 0000000000..ab14a119ef --- /dev/null +++ b/intl/icu/source/common/ucnv_lmb.cpp @@ -0,0 +1,1388 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2000-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* file name: ucnv_lmb.cpp +* encoding: UTF-8 +* tab size: 4 (not used) +* indentation:4 +* +* created on: 2000feb09 +* created by: Brendan Murray +* extensively hacked up by: Jim Snyder-Grant +* +* Modification History: +* +* Date Name Description +* +* 06/20/2000 helena OS/400 port changes; mostly typecast. +* 06/27/2000 Jim Snyder-Grant Deal with partial characters and small buffers. +* Add comments to document LMBCS format and implementation +* restructured order & breakdown of functions +* 06/28/2000 helena Major rewrite for the callback API changes. +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION && !UCONFIG_NO_LEGACY_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION + +#include "unicode/ucnv_err.h" +#include "unicode/ucnv.h" +#include "unicode/uset.h" +#include "cmemory.h" +#include "cstring.h" +#include "uassert.h" +#include "ucnv_imp.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" + +#ifdef EBCDIC_RTL + #include "ascii_a.h" +#endif + +/* + LMBCS + + (Lotus Multi-Byte Character Set) + + LMBCS was invented in the late 1980's and is primarily used in Lotus Notes + databases and in Lotus 1-2-3 files. Programmers who work with the APIs + into these products will sometimes need to deal with strings in this format. + + The code in this file provides an implementation for an ICU converter of + LMBCS to and from Unicode. + + Since the LMBCS character set is only sparsely documented in existing + printed or online material, we have added extensive annotation to this + file to serve as a guide to understanding LMBCS. + + LMBCS was originally designed with these four sometimes-competing design goals: + + -Provide encodings for the characters in 12 existing national standards + (plus a few other characters) + -Minimal memory footprint + -Maximal speed of conversion into the existing national character sets + -No need to track a changing state as you interpret a string. + + + All of the national character sets LMBCS was trying to encode are 'ANSI' + based, in that the bytes from 0x20 - 0x7F are almost exactly the + same common Latin unaccented characters and symbols in all character sets. + + So, in order to help meet the speed & memory design goals, the common ANSI + bytes from 0x20-0x7F are represented by the same single-byte values in LMBCS. + + The general LMBCS code unit is from 1-3 bytes. We can describe the 3 bytes as + follows: + + [G] D1 [D2] + + That is, a sometimes-optional 'group' byte, followed by 1 and sometimes 2 + data bytes. The maximum size of a LMBCS character is 3 bytes: +*/ +#define ULMBCS_CHARSIZE_MAX 3 +/* + The single-byte values from 0x20 to 0x7F are examples of single D1 bytes. + We often have to figure out if byte values are below or above this, so we + use the ANSI nomenclature 'C0' and 'C1' to refer to the range of control + characters just above & below the common lower-ANSI range */ +#define ULMBCS_C0END 0x1F +#define ULMBCS_C1START 0x80 +/* + Since LMBCS is always dealing in byte units. we create a local type here for + dealing with these units of LMBCS code units: + +*/ +typedef uint8_t ulmbcs_byte_t; + +/* + Most of the values less than 0x20 are reserved in LMBCS to announce + which national character standard is being used for the 'D' bytes. + In the comments we show the common name and the IBM character-set ID + for these character-set announcers: +*/ + +#define ULMBCS_GRP_L1 0x01 /* Latin-1 :ibm-850 */ +#define ULMBCS_GRP_GR 0x02 /* Greek :ibm-851 */ +#define ULMBCS_GRP_HE 0x03 /* Hebrew :ibm-1255 */ +#define ULMBCS_GRP_AR 0x04 /* Arabic :ibm-1256 */ +#define ULMBCS_GRP_RU 0x05 /* Cyrillic :ibm-1251 */ +#define ULMBCS_GRP_L2 0x06 /* Latin-2 :ibm-852 */ +#define ULMBCS_GRP_TR 0x08 /* Turkish :ibm-1254 */ +#define ULMBCS_GRP_TH 0x0B /* Thai :ibm-874 */ +#define ULMBCS_GRP_JA 0x10 /* Japanese :ibm-943 */ +#define ULMBCS_GRP_KO 0x11 /* Korean :ibm-1261 */ +#define ULMBCS_GRP_TW 0x12 /* Chinese SC :ibm-950 */ +#define ULMBCS_GRP_CN 0x13 /* Chinese TC :ibm-1386 */ + +/* + So, the beginning of understanding LMBCS is that IF the first byte of a LMBCS + character is one of those 12 values, you can interpret the remaining bytes of + that character as coming from one of those character sets. Since the lower + ANSI bytes already are represented in single bytes, using one of the character + set announcers is used to announce a character that starts with a byte of + 0x80 or greater. + + The character sets are arranged so that the single byte sets all appear + before the multi-byte character sets. When we need to tell whether a + group byte is for a single byte char set or not we use this define: */ + +#define ULMBCS_DOUBLEOPTGROUP_START 0x10 + +/* +However, to fully understand LMBCS, you must also understand a series of +exceptions & optimizations made in service of the design goals. + +First, those of you who are character set mavens may have noticed that +the 'double-byte' character sets are actually multi-byte character sets +that can have 1 or two bytes, even in the upper-ascii range. To force +each group byte to introduce a fixed-width encoding (to make it faster to +count characters), we use a convention of doubling up on the group byte +to introduce any single-byte character > 0x80 in an otherwise double-byte +character set. So, for example, the LMBCS sequence x10 x10 xAE is the +same as '0xAE' in the Japanese code page 943. + +Next, you will notice that the list of group bytes has some gaps. +These are used in various ways. + +We reserve a few special single byte values for common control +characters. These are in the same place as their ANSI equivalents for speed. +*/ + +#define ULMBCS_HT 0x09 /* Fixed control char - Horizontal Tab */ +#define ULMBCS_LF 0x0A /* Fixed control char - Line Feed */ +#define ULMBCS_CR 0x0D /* Fixed control char - Carriage Return */ + +/* Then, 1-2-3 reserved a special single-byte character to put at the +beginning of internal 'system' range names: */ + +#define ULMBCS_123SYSTEMRANGE 0x19 + +/* Then we needed a place to put all the other ansi control characters +that must be moved to different values because LMBCS reserves those +values for other purposes. To represent the control characters, we start +with a first byte of 0xF & add the control character value as the +second byte */ +#define ULMBCS_GRP_CTRL 0x0F + +/* For the C0 controls (less than 0x20), we add 0x20 to preserve the +useful doctrine that any byte less than 0x20 in a LMBCS char must be +the first byte of a character:*/ +#define ULMBCS_CTRLOFFSET 0x20 + +/* +Where to put the characters that aren't part of any of the 12 national +character sets? The first thing that was done, in the earlier years of +LMBCS, was to use up the spaces of the form + + [G] D1, + + where 'G' was one of the single-byte character groups, and + D1 was less than 0x80. These sequences are gathered together + into a Lotus-invented doublebyte character set to represent a + lot of stray values. Internally, in this implementation, we track this + as group '0', as a place to tuck this exceptions list.*/ + +#define ULMBCS_GRP_EXCEPT 0x00 +/* + Finally, as the durability and usefulness of UNICODE became clear, + LOTUS added a new group 0x14 to hold Unicode values not otherwise + represented in LMBCS: */ +#define ULMBCS_GRP_UNICODE 0x14 +/* The two bytes appearing after a 0x14 are interpreted as UFT-16 BE +(Big-Endian) characters. The exception comes when the UTF16 +representation would have a zero as the second byte. In that case, +'F6' is used in its place, and the bytes are swapped. (This prevents +LMBCS from encoding any Unicode values of the form U+F6xx, but that's OK: +0xF6xx is in the middle of the Private Use Area.)*/ +#define ULMBCS_UNICOMPATZERO 0xF6 + +/* It is also useful in our code to have a constant for the size of +a LMBCS char that holds a literal Unicode value */ +#define ULMBCS_UNICODE_SIZE 3 + +/* +To squish the LMBCS representations down even further, and to make +translations even faster,sometimes the optimization group byte can be dropped +from a LMBCS character. This is decided on a process-by-process basis. The +group byte that is dropped is called the 'optimization group'. + +For Notes, the optimzation group is always 0x1.*/ +#define ULMBCS_DEFAULTOPTGROUP 0x1 +/* For 1-2-3 files, the optimzation group is stored in the header of the 1-2-3 +file. + + In any case, when using ICU, you either pass in the +optimization group as part of the name of the converter (LMBCS-1, LMBCS-2, +etc.). Using plain 'LMBCS' as the name of the converter will give you +LMBCS-1. + + +*** Implementation strategy *** + + +Because of the extensive use of other character sets, the LMBCS converter +keeps a mapping between optimization groups and IBM character sets, so that +ICU converters can be created and used as needed. */ + +/* As you can see, even though any byte below 0x20 could be an optimization +byte, only those at 0x13 or below can map to an actual converter. To limit +some loops and searches, we define a value for that last group converter:*/ + +#define ULMBCS_GRP_LAST 0x13 /* last LMBCS group that has a converter */ + +static const char * const OptGroupByteToCPName[ULMBCS_GRP_LAST + 1] = { + /* 0x0000 */ "lmb-excp", /* internal home for the LOTUS exceptions list */ + /* 0x0001 */ "ibm-850", + /* 0x0002 */ "ibm-851", + /* 0x0003 */ "windows-1255", + /* 0x0004 */ "windows-1256", + /* 0x0005 */ "windows-1251", + /* 0x0006 */ "ibm-852", + /* 0x0007 */ nullptr, /* Unused */ + /* 0x0008 */ "windows-1254", + /* 0x0009 */ nullptr, /* Control char HT */ + /* 0x000A */ nullptr, /* Control char LF */ + /* 0x000B */ "windows-874", + /* 0x000C */ nullptr, /* Unused */ + /* 0x000D */ nullptr, /* Control char CR */ + /* 0x000E */ nullptr, /* Unused */ + /* 0x000F */ nullptr, /* Control chars: 0x0F20 + C0/C1 character: algorithmic */ + /* 0x0010 */ "windows-932", + /* 0x0011 */ "windows-949", + /* 0x0012 */ "windows-950", + /* 0x0013 */ "windows-936" + + /* The rest are null, including the 0x0014 Unicode compatibility region + and 0x0019, the 1-2-3 system range control char */ +}; + + +/* That's approximately all the data that's needed for translating + LMBCS to Unicode. + + +However, to translate Unicode to LMBCS, we need some more support. + +That's because there are often more than one possible mappings from a Unicode +code point back into LMBCS. The first thing we do is look up into a table +to figure out if there are more than one possible mappings. This table, +arranged by Unicode values (including ranges) either lists which group +to use, or says that it could go into one or more of the SBCS sets, or +into one or more of the DBCS sets. (If the character exists in both DBCS & +SBCS, the table will place it in the SBCS sets, to make the LMBCS code point +length as small as possible. Here's the two special markers we use to indicate +ambiguous mappings: */ + +#define ULMBCS_AMBIGUOUS_SBCS 0x80 /* could fit in more than one + LMBCS sbcs native encoding + (example: most accented latin) */ +#define ULMBCS_AMBIGUOUS_MBCS 0x81 /* could fit in more than one + LMBCS mbcs native encoding + (example: Unihan) */ +#define ULMBCS_AMBIGUOUS_ALL 0x82 +/* And here's a simple way to see if a group falls in an appropriate range */ +#define ULMBCS_AMBIGUOUS_MATCH(agroup, xgroup) \ + ((((agroup) == ULMBCS_AMBIGUOUS_SBCS) && \ + (xgroup) < ULMBCS_DOUBLEOPTGROUP_START) || \ + (((agroup) == ULMBCS_AMBIGUOUS_MBCS) && \ + (xgroup) >= ULMBCS_DOUBLEOPTGROUP_START)) || \ + ((agroup) == ULMBCS_AMBIGUOUS_ALL) + + +/* The table & some code to use it: */ + + +static const struct _UniLMBCSGrpMap +{ + const char16_t uniStartRange; + const char16_t uniEndRange; + const ulmbcs_byte_t GrpType; +} UniLMBCSGrpMap[] += +{ + + {0x0001, 0x001F, ULMBCS_GRP_CTRL}, + {0x0080, 0x009F, ULMBCS_GRP_CTRL}, + {0x00A0, 0x00A6, ULMBCS_AMBIGUOUS_SBCS}, + {0x00A7, 0x00A8, ULMBCS_AMBIGUOUS_ALL}, + {0x00A9, 0x00AF, ULMBCS_AMBIGUOUS_SBCS}, + {0x00B0, 0x00B1, ULMBCS_AMBIGUOUS_ALL}, + {0x00B2, 0x00B3, ULMBCS_AMBIGUOUS_SBCS}, + {0x00B4, 0x00B4, ULMBCS_AMBIGUOUS_ALL}, + {0x00B5, 0x00B5, ULMBCS_AMBIGUOUS_SBCS}, + {0x00B6, 0x00B6, ULMBCS_AMBIGUOUS_ALL}, + {0x00B7, 0x00D6, ULMBCS_AMBIGUOUS_SBCS}, + {0x00D7, 0x00D7, ULMBCS_AMBIGUOUS_ALL}, + {0x00D8, 0x00F6, ULMBCS_AMBIGUOUS_SBCS}, + {0x00F7, 0x00F7, ULMBCS_AMBIGUOUS_ALL}, + {0x00F8, 0x01CD, ULMBCS_AMBIGUOUS_SBCS}, + {0x01CE, 0x01CE, ULMBCS_GRP_TW }, + {0x01CF, 0x02B9, ULMBCS_AMBIGUOUS_SBCS}, + {0x02BA, 0x02BA, ULMBCS_GRP_CN}, + {0x02BC, 0x02C8, ULMBCS_AMBIGUOUS_SBCS}, + {0x02C9, 0x02D0, ULMBCS_AMBIGUOUS_MBCS}, + {0x02D8, 0x02DD, ULMBCS_AMBIGUOUS_SBCS}, + {0x0384, 0x0390, ULMBCS_AMBIGUOUS_SBCS}, + {0x0391, 0x03A9, ULMBCS_AMBIGUOUS_ALL}, + {0x03AA, 0x03B0, ULMBCS_AMBIGUOUS_SBCS}, + {0x03B1, 0x03C9, ULMBCS_AMBIGUOUS_ALL}, + {0x03CA, 0x03CE, ULMBCS_AMBIGUOUS_SBCS}, + {0x0400, 0x0400, ULMBCS_GRP_RU}, + {0x0401, 0x0401, ULMBCS_AMBIGUOUS_ALL}, + {0x0402, 0x040F, ULMBCS_GRP_RU}, + {0x0410, 0x0431, ULMBCS_AMBIGUOUS_ALL}, + {0x0432, 0x044E, ULMBCS_GRP_RU}, + {0x044F, 0x044F, ULMBCS_AMBIGUOUS_ALL}, + {0x0450, 0x0491, ULMBCS_GRP_RU}, + {0x05B0, 0x05F2, ULMBCS_GRP_HE}, + {0x060C, 0x06AF, ULMBCS_GRP_AR}, + {0x0E01, 0x0E5B, ULMBCS_GRP_TH}, + {0x200C, 0x200F, ULMBCS_AMBIGUOUS_SBCS}, + {0x2010, 0x2010, ULMBCS_AMBIGUOUS_MBCS}, + {0x2013, 0x2014, ULMBCS_AMBIGUOUS_SBCS}, + {0x2015, 0x2015, ULMBCS_AMBIGUOUS_MBCS}, + {0x2016, 0x2016, ULMBCS_AMBIGUOUS_MBCS}, + {0x2017, 0x2017, ULMBCS_AMBIGUOUS_SBCS}, + {0x2018, 0x2019, ULMBCS_AMBIGUOUS_ALL}, + {0x201A, 0x201B, ULMBCS_AMBIGUOUS_SBCS}, + {0x201C, 0x201D, ULMBCS_AMBIGUOUS_ALL}, + {0x201E, 0x201F, ULMBCS_AMBIGUOUS_SBCS}, + {0x2020, 0x2021, ULMBCS_AMBIGUOUS_ALL}, + {0x2022, 0x2024, ULMBCS_AMBIGUOUS_SBCS}, + {0x2025, 0x2025, ULMBCS_AMBIGUOUS_MBCS}, + {0x2026, 0x2026, ULMBCS_AMBIGUOUS_ALL}, + {0x2027, 0x2027, ULMBCS_GRP_TW}, + {0x2030, 0x2030, ULMBCS_AMBIGUOUS_ALL}, + {0x2031, 0x2031, ULMBCS_AMBIGUOUS_SBCS}, + {0x2032, 0x2033, ULMBCS_AMBIGUOUS_MBCS}, + {0x2035, 0x2035, ULMBCS_AMBIGUOUS_MBCS}, + {0x2039, 0x203A, ULMBCS_AMBIGUOUS_SBCS}, + {0x203B, 0x203B, ULMBCS_AMBIGUOUS_MBCS}, + {0x203C, 0x203C, ULMBCS_GRP_EXCEPT}, + {0x2074, 0x2074, ULMBCS_GRP_KO}, + {0x207F, 0x207F, ULMBCS_GRP_EXCEPT}, + {0x2081, 0x2084, ULMBCS_GRP_KO}, + {0x20A4, 0x20AC, ULMBCS_AMBIGUOUS_SBCS}, + {0x2103, 0x2109, ULMBCS_AMBIGUOUS_MBCS}, + {0x2111, 0x2120, ULMBCS_AMBIGUOUS_SBCS}, + /*zhujin: upgrade, for regressiont test, spr HKIA4YHTSU*/ + {0x2121, 0x2121, ULMBCS_AMBIGUOUS_MBCS}, + {0x2122, 0x2126, ULMBCS_AMBIGUOUS_SBCS}, + {0x212B, 0x212B, ULMBCS_AMBIGUOUS_MBCS}, + {0x2135, 0x2135, ULMBCS_AMBIGUOUS_SBCS}, + {0x2153, 0x2154, ULMBCS_GRP_KO}, + {0x215B, 0x215E, ULMBCS_GRP_EXCEPT}, + {0x2160, 0x2179, ULMBCS_AMBIGUOUS_MBCS}, + {0x2190, 0x2193, ULMBCS_AMBIGUOUS_ALL}, + {0x2194, 0x2195, ULMBCS_GRP_EXCEPT}, + {0x2196, 0x2199, ULMBCS_AMBIGUOUS_MBCS}, + {0x21A8, 0x21A8, ULMBCS_GRP_EXCEPT}, + {0x21B8, 0x21B9, ULMBCS_GRP_CN}, + {0x21D0, 0x21D1, ULMBCS_GRP_EXCEPT}, + {0x21D2, 0x21D2, ULMBCS_AMBIGUOUS_MBCS}, + {0x21D3, 0x21D3, ULMBCS_GRP_EXCEPT}, + {0x21D4, 0x21D4, ULMBCS_AMBIGUOUS_MBCS}, + {0x21D5, 0x21D5, ULMBCS_GRP_EXCEPT}, + {0x21E7, 0x21E7, ULMBCS_GRP_CN}, + {0x2200, 0x2200, ULMBCS_AMBIGUOUS_MBCS}, + {0x2201, 0x2201, ULMBCS_GRP_EXCEPT}, + {0x2202, 0x2202, ULMBCS_AMBIGUOUS_MBCS}, + {0x2203, 0x2203, ULMBCS_AMBIGUOUS_MBCS}, + {0x2204, 0x2206, ULMBCS_GRP_EXCEPT}, + {0x2207, 0x2208, ULMBCS_AMBIGUOUS_MBCS}, + {0x2209, 0x220A, ULMBCS_GRP_EXCEPT}, + {0x220B, 0x220B, ULMBCS_AMBIGUOUS_MBCS}, + {0x220F, 0x2215, ULMBCS_AMBIGUOUS_MBCS}, + {0x2219, 0x2219, ULMBCS_GRP_EXCEPT}, + {0x221A, 0x221A, ULMBCS_AMBIGUOUS_MBCS}, + {0x221B, 0x221C, ULMBCS_GRP_EXCEPT}, + {0x221D, 0x221E, ULMBCS_AMBIGUOUS_MBCS}, + {0x221F, 0x221F, ULMBCS_GRP_EXCEPT}, + {0x2220, 0x2220, ULMBCS_AMBIGUOUS_MBCS}, + {0x2223, 0x222A, ULMBCS_AMBIGUOUS_MBCS}, + {0x222B, 0x223D, ULMBCS_AMBIGUOUS_MBCS}, + {0x2245, 0x2248, ULMBCS_GRP_EXCEPT}, + {0x224C, 0x224C, ULMBCS_GRP_TW}, + {0x2252, 0x2252, ULMBCS_AMBIGUOUS_MBCS}, + {0x2260, 0x2261, ULMBCS_AMBIGUOUS_MBCS}, + {0x2262, 0x2265, ULMBCS_GRP_EXCEPT}, + {0x2266, 0x226F, ULMBCS_AMBIGUOUS_MBCS}, + {0x2282, 0x2283, ULMBCS_AMBIGUOUS_MBCS}, + {0x2284, 0x2285, ULMBCS_GRP_EXCEPT}, + {0x2286, 0x2287, ULMBCS_AMBIGUOUS_MBCS}, + {0x2288, 0x2297, ULMBCS_GRP_EXCEPT}, + {0x2299, 0x22BF, ULMBCS_AMBIGUOUS_MBCS}, + {0x22C0, 0x22C0, ULMBCS_GRP_EXCEPT}, + {0x2310, 0x2310, ULMBCS_GRP_EXCEPT}, + {0x2312, 0x2312, ULMBCS_AMBIGUOUS_MBCS}, + {0x2318, 0x2321, ULMBCS_GRP_EXCEPT}, + {0x2318, 0x2321, ULMBCS_GRP_CN}, + {0x2460, 0x24E9, ULMBCS_AMBIGUOUS_MBCS}, + {0x2500, 0x2500, ULMBCS_AMBIGUOUS_SBCS}, + {0x2501, 0x2501, ULMBCS_AMBIGUOUS_MBCS}, + {0x2502, 0x2502, ULMBCS_AMBIGUOUS_ALL}, + {0x2503, 0x2503, ULMBCS_AMBIGUOUS_MBCS}, + {0x2504, 0x2505, ULMBCS_GRP_TW}, + {0x2506, 0x2665, ULMBCS_AMBIGUOUS_ALL}, + {0x2666, 0x2666, ULMBCS_GRP_EXCEPT}, + {0x2667, 0x2669, ULMBCS_AMBIGUOUS_SBCS}, + {0x266A, 0x266A, ULMBCS_AMBIGUOUS_ALL}, + {0x266B, 0x266C, ULMBCS_AMBIGUOUS_SBCS}, + {0x266D, 0x266D, ULMBCS_AMBIGUOUS_MBCS}, + {0x266E, 0x266E, ULMBCS_AMBIGUOUS_SBCS}, + {0x266F, 0x266F, ULMBCS_GRP_JA}, + {0x2670, 0x2E7F, ULMBCS_AMBIGUOUS_SBCS}, + {0x2E80, 0xF861, ULMBCS_AMBIGUOUS_MBCS}, + {0xF862, 0xF8FF, ULMBCS_GRP_EXCEPT}, + {0xF900, 0xFA2D, ULMBCS_AMBIGUOUS_MBCS}, + {0xFB00, 0xFEFF, ULMBCS_AMBIGUOUS_SBCS}, + {0xFF01, 0xFFEE, ULMBCS_AMBIGUOUS_MBCS}, + {0xFFFF, 0xFFFF, ULMBCS_GRP_UNICODE} +}; + +static ulmbcs_byte_t +FindLMBCSUniRange(char16_t uniChar) +{ + const struct _UniLMBCSGrpMap * pTable = UniLMBCSGrpMap; + + while (uniChar > pTable->uniEndRange) + { + pTable++; + } + + if (uniChar >= pTable->uniStartRange) + { + return pTable->GrpType; + } + return ULMBCS_GRP_UNICODE; +} + +/* +We also ask the creator of a converter to send in a preferred locale +that we can use in resolving ambiguous mappings. They send the locale +in as a string, and we map it, if possible, to one of the +LMBCS groups. We use this table, and the associated code, to +do the lookup: */ + +/************************************************** + This table maps locale ID's to LMBCS opt groups. + The default return is group 0x01. Note that for + performance reasons, the table is sorted in + increasing alphabetic order, with the notable + exception of zhTW. This is to force the check + for Traditonal Chinese before dropping back to + Simplified. + + Note too that the Latin-1 groups have been + commented out because it's the default, and + this shortens the table, allowing a serial + search to go quickly. + *************************************************/ + +static const struct _LocaleLMBCSGrpMap +{ + const char *LocaleID; + const ulmbcs_byte_t OptGroup; +} LocaleLMBCSGrpMap[] = +{ + {"ar", ULMBCS_GRP_AR}, + {"be", ULMBCS_GRP_RU}, + {"bg", ULMBCS_GRP_L2}, + /* {"ca", ULMBCS_GRP_L1}, */ + {"cs", ULMBCS_GRP_L2}, + /* {"da", ULMBCS_GRP_L1}, */ + /* {"de", ULMBCS_GRP_L1}, */ + {"el", ULMBCS_GRP_GR}, + /* {"en", ULMBCS_GRP_L1}, */ + /* {"es", ULMBCS_GRP_L1}, */ + /* {"et", ULMBCS_GRP_L1}, */ + /* {"fi", ULMBCS_GRP_L1}, */ + /* {"fr", ULMBCS_GRP_L1}, */ + {"he", ULMBCS_GRP_HE}, + {"hu", ULMBCS_GRP_L2}, + /* {"is", ULMBCS_GRP_L1}, */ + /* {"it", ULMBCS_GRP_L1}, */ + {"iw", ULMBCS_GRP_HE}, + {"ja", ULMBCS_GRP_JA}, + {"ko", ULMBCS_GRP_KO}, + /* {"lt", ULMBCS_GRP_L1}, */ + /* {"lv", ULMBCS_GRP_L1}, */ + {"mk", ULMBCS_GRP_RU}, + /* {"nl", ULMBCS_GRP_L1}, */ + /* {"no", ULMBCS_GRP_L1}, */ + {"pl", ULMBCS_GRP_L2}, + /* {"pt", ULMBCS_GRP_L1}, */ + {"ro", ULMBCS_GRP_L2}, + {"ru", ULMBCS_GRP_RU}, + {"sh", ULMBCS_GRP_L2}, + {"sk", ULMBCS_GRP_L2}, + {"sl", ULMBCS_GRP_L2}, + {"sq", ULMBCS_GRP_L2}, + {"sr", ULMBCS_GRP_RU}, + /* {"sv", ULMBCS_GRP_L1}, */ + {"th", ULMBCS_GRP_TH}, + {"tr", ULMBCS_GRP_TR}, + {"uk", ULMBCS_GRP_RU}, + /* {"vi", ULMBCS_GRP_L1}, */ + {"zhTW", ULMBCS_GRP_TW}, + {"zh", ULMBCS_GRP_CN}, + {nullptr, ULMBCS_GRP_L1} +}; + + +static ulmbcs_byte_t +FindLMBCSLocale(const char *LocaleID) +{ + const struct _LocaleLMBCSGrpMap *pTable = LocaleLMBCSGrpMap; + + if ((!LocaleID) || (!*LocaleID)) + { + return 0; + } + + while (pTable->LocaleID) + { + if (*pTable->LocaleID == *LocaleID) /* Check only first char for speed */ + { + /* First char matches - check whole name, for entry-length */ + if (uprv_strncmp(pTable->LocaleID, LocaleID, strlen(pTable->LocaleID)) == 0) + return pTable->OptGroup; + } + else + if (*pTable->LocaleID > *LocaleID) /* Sorted alphabetically - exit */ + break; + pTable++; + } + return ULMBCS_GRP_L1; +} + + +/* + Before we get to the main body of code, here's how we hook up to the rest + of ICU. ICU converters are required to define a structure that includes + some function pointers, and some common data, in the style of a C++ + vtable. There is also room in there for converter-specific data. LMBCS + uses that converter-specific data to keep track of the 12 subconverters + we use, the optimization group, and the group (if any) that matches the + locale. We have one structure instantiated for each of the 12 possible + optimization groups. To avoid typos & to avoid boring the reader, we + put the declarations of these structures and functions into macros. To see + the definitions of these structures, see unicode\ucnv_bld.h +*/ + +typedef struct + { + UConverterSharedData *OptGrpConverter[ULMBCS_GRP_LAST+1]; /* Converter per Opt. grp. */ + uint8_t OptGroup; /* default Opt. grp. for this LMBCS session */ + uint8_t localeConverterIndex; /* reasonable locale match for index */ + } +UConverterDataLMBCS; + +U_CDECL_BEGIN +static void U_CALLCONV _LMBCSClose(UConverter * _this); +U_CDECL_END + +#define DECLARE_LMBCS_DATA(n) \ +static const UConverterImpl _LMBCSImpl##n={\ + UCNV_LMBCS_##n,\ + nullptr,nullptr,\ + _LMBCSOpen##n,\ + _LMBCSClose,\ + nullptr,\ + _LMBCSToUnicodeWithOffsets,\ + _LMBCSToUnicodeWithOffsets,\ + _LMBCSFromUnicode,\ + _LMBCSFromUnicode,\ + nullptr,\ + nullptr,\ + nullptr,\ + nullptr,\ + _LMBCSSafeClone,\ + ucnv_getCompleteUnicodeSet,\ + nullptr,\ + nullptr\ +};\ +static const UConverterStaticData _LMBCSStaticData##n={\ + sizeof(UConverterStaticData),\ + "LMBCS-" #n,\ + 0, UCNV_IBM, UCNV_LMBCS_##n, 1, 3,\ + { 0x3f, 0, 0, 0 },1,false,false,0,0,{0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0} \ +};\ +const UConverterSharedData _LMBCSData##n= \ + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_LMBCSStaticData##n, &_LMBCSImpl##n); + + /* The only function we needed to duplicate 12 times was the 'open' +function, which will do basically the same thing except set a different +optimization group. So, we put the common stuff into a worker function, +and set up another macro to stamp out the 12 open functions:*/ +#define DEFINE_LMBCS_OPEN(n) \ +static void U_CALLCONV \ + _LMBCSOpen##n(UConverter* _this, UConverterLoadArgs* pArgs, UErrorCode* err) \ +{ _LMBCSOpenWorker(_this, pArgs, err, n); } + + + +/* Here's the open worker & the common close function */ +static void +_LMBCSOpenWorker(UConverter* _this, + UConverterLoadArgs *pArgs, + UErrorCode* err, + ulmbcs_byte_t OptGroup) +{ + UConverterDataLMBCS * extraInfo = (UConverterDataLMBCS*)uprv_malloc (sizeof (UConverterDataLMBCS)); + _this->extraInfo = extraInfo; + if(extraInfo != nullptr) + { + UConverterNamePieces stackPieces; + UConverterLoadArgs stackArgs= UCNV_LOAD_ARGS_INITIALIZER; + ulmbcs_byte_t i; + + uprv_memset(extraInfo, 0, sizeof(UConverterDataLMBCS)); + + stackArgs.onlyTestIsLoadable = pArgs->onlyTestIsLoadable; + + for (i=0; i <= ULMBCS_GRP_LAST && U_SUCCESS(*err); i++) + { + if(OptGroupByteToCPName[i] != nullptr) { + extraInfo->OptGrpConverter[i] = ucnv_loadSharedData(OptGroupByteToCPName[i], &stackPieces, &stackArgs, err); + } + } + + if(U_FAILURE(*err) || pArgs->onlyTestIsLoadable) { + _LMBCSClose(_this); + return; + } + extraInfo->OptGroup = OptGroup; + extraInfo->localeConverterIndex = FindLMBCSLocale(pArgs->locale); + } + else + { + *err = U_MEMORY_ALLOCATION_ERROR; + } +} + +U_CDECL_BEGIN +static void U_CALLCONV +_LMBCSClose(UConverter * _this) +{ + if (_this->extraInfo != nullptr) + { + ulmbcs_byte_t Ix; + UConverterDataLMBCS * extraInfo = (UConverterDataLMBCS *) _this->extraInfo; + + for (Ix=0; Ix <= ULMBCS_GRP_LAST; Ix++) + { + if (extraInfo->OptGrpConverter[Ix] != nullptr) + ucnv_unloadSharedDataIfReady(extraInfo->OptGrpConverter[Ix]); + } + if (!_this->isExtraLocal) { + uprv_free (_this->extraInfo); + _this->extraInfo = nullptr; + } + } +} + +typedef struct LMBCSClone { + UConverter cnv; + UConverterDataLMBCS lmbcs; +} LMBCSClone; + +static UConverter * U_CALLCONV +_LMBCSSafeClone(const UConverter *cnv, + void *stackBuffer, + int32_t *pBufferSize, + UErrorCode *status) { + (void)status; + LMBCSClone *newLMBCS; + UConverterDataLMBCS *extraInfo; + int32_t i; + + if(*pBufferSize<=0) { + *pBufferSize=(int32_t)sizeof(LMBCSClone); + return nullptr; + } + + extraInfo=(UConverterDataLMBCS *)cnv->extraInfo; + newLMBCS=(LMBCSClone *)stackBuffer; + + /* ucnv.c/ucnv_safeClone() copied the main UConverter already */ + + uprv_memcpy(&newLMBCS->lmbcs, extraInfo, sizeof(UConverterDataLMBCS)); + + /* share the subconverters */ + for(i = 0; i <= ULMBCS_GRP_LAST; ++i) { + if(extraInfo->OptGrpConverter[i] != nullptr) { + ucnv_incrementRefCount(extraInfo->OptGrpConverter[i]); + } + } + + newLMBCS->cnv.extraInfo = &newLMBCS->lmbcs; + newLMBCS->cnv.isExtraLocal = true; + return &newLMBCS->cnv; +} + +/* + * There used to be a _LMBCSGetUnicodeSet() function here (up to svn revision 20117) + * which added all code points except for U+F6xx + * because those cannot be represented in the Unicode group. + * However, it turns out that windows-950 has roundtrips for all of U+F6xx + * which means that LMBCS can convert all Unicode code points after all. + * We now simply use ucnv_getCompleteUnicodeSet(). + * + * This may need to be looked at again as Lotus uses _LMBCSGetUnicodeSet(). (091216) + */ + +/* + Here's the basic helper function that we use when converting from + Unicode to LMBCS, and we suspect that a Unicode character will fit into + one of the 12 groups. The return value is the number of bytes written + starting at pStartLMBCS (if any). +*/ + +static size_t +LMBCSConversionWorker ( + UConverterDataLMBCS * extraInfo, /* subconverters, opt & locale groups */ + ulmbcs_byte_t group, /* The group to try */ + ulmbcs_byte_t * pStartLMBCS, /* where to put the results */ + char16_t * pUniChar, /* The input unicode character */ + ulmbcs_byte_t * lastConverterIndex, /* output: track last successful group used */ + UBool * groups_tried /* output: track any unsuccessful groups */ +) +{ + ulmbcs_byte_t * pLMBCS = pStartLMBCS; + UConverterSharedData * xcnv = extraInfo->OptGrpConverter[group]; + + int bytesConverted; + uint32_t value; + ulmbcs_byte_t firstByte; + + U_ASSERT(xcnv); + U_ASSERT(group 0) { + firstByte = (ulmbcs_byte_t)(value >> ((bytesConverted - 1) * 8)); + } else { + /* most common failure mode is an unassigned character */ + groups_tried[group] = true; + return 0; + } + + *lastConverterIndex = group; + + /* All initial byte values in lower ascii range should have been caught by now, + except with the exception group. + */ + U_ASSERT((firstByte <= ULMBCS_C0END) || (firstByte >= ULMBCS_C1START) || (group == ULMBCS_GRP_EXCEPT)); + + /* use converted data: first write 0, 1 or two group bytes */ + if (group != ULMBCS_GRP_EXCEPT && extraInfo->OptGroup != group) + { + *pLMBCS++ = group; + if (bytesConverted == 1 && group >= ULMBCS_DOUBLEOPTGROUP_START) + { + *pLMBCS++ = group; + } + } + + /* don't emit control chars */ + if ( bytesConverted == 1 && firstByte < 0x20 ) + return 0; + + + /* then move over the converted data */ + switch(bytesConverted) + { + case 4: + *pLMBCS++ = (ulmbcs_byte_t)(value >> 24); + U_FALLTHROUGH; + case 3: + *pLMBCS++ = (ulmbcs_byte_t)(value >> 16); + U_FALLTHROUGH; + case 2: + *pLMBCS++ = (ulmbcs_byte_t)(value >> 8); + U_FALLTHROUGH; + case 1: + *pLMBCS++ = (ulmbcs_byte_t)value; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + + return (pLMBCS - pStartLMBCS); +} + + +/* This is a much simpler version of above, when we +know we are writing LMBCS using the Unicode group +*/ +static size_t +LMBCSConvertUni(ulmbcs_byte_t * pLMBCS, char16_t uniChar) +{ + /* encode into LMBCS Unicode range */ + uint8_t LowCh = (uint8_t)(uniChar & 0x00FF); + uint8_t HighCh = (uint8_t)(uniChar >> 8); + + *pLMBCS++ = ULMBCS_GRP_UNICODE; + + if (LowCh == 0) + { + *pLMBCS++ = ULMBCS_UNICOMPATZERO; + *pLMBCS++ = HighCh; + } + else + { + *pLMBCS++ = HighCh; + *pLMBCS++ = LowCh; + } + return ULMBCS_UNICODE_SIZE; +} + + + +/* The main Unicode to LMBCS conversion function */ +static void U_CALLCONV +_LMBCSFromUnicode(UConverterFromUnicodeArgs* args, + UErrorCode* err) +{ + ulmbcs_byte_t lastConverterIndex = 0; + char16_t uniChar; + ulmbcs_byte_t LMBCS[ULMBCS_CHARSIZE_MAX]; + ulmbcs_byte_t * pLMBCS; + int32_t bytes_written; + UBool groups_tried[ULMBCS_GRP_LAST+1]; + UConverterDataLMBCS * extraInfo = (UConverterDataLMBCS *) args->converter->extraInfo; + int sourceIndex = 0; + + /* Basic strategy: attempt to fill in local LMBCS 1-char buffer.(LMBCS) + If that succeeds, see if it will all fit into the target & copy it over + if it does. + + We try conversions in the following order: + + 1. Single-byte ascii & special fixed control chars (&null) + 2. Look up group in table & try that (could be + A) Unicode group + B) control group, + C) national encoding, + or ambiguous SBCS or MBCS group (on to step 4...) + + 3. If its ambiguous, try this order: + A) The optimization group + B) The locale group + C) The last group that succeeded with this string. + D) every other group that's relevant (single or double) + E) If its single-byte ambiguous, try the exceptions group + + 4. And as a grand fallback: Unicode + */ + + /*Fix for SPR#DJOE66JFN3 (Lotus)*/ + ulmbcs_byte_t OldConverterIndex = 0; + + while (args->source < args->sourceLimit && !U_FAILURE(*err)) + { + /*Fix for SPR#DJOE66JFN3 (Lotus)*/ + OldConverterIndex = extraInfo->localeConverterIndex; + + if (args->target >= args->targetLimit) + { + *err = U_BUFFER_OVERFLOW_ERROR; + break; + } + uniChar = *(args->source); + bytes_written = 0; + pLMBCS = LMBCS; + + /* check cases in rough order of how common they are, for speed */ + + /* single byte matches: strategy 1 */ + /*Fix for SPR#DJOE66JFN3 (Lotus)*/ + if((uniChar>=0x80) && (uniChar<=0xff) + /*Fix for SPR#JUYA6XAERU and TSAO7GL5NK (Lotus)*/ &&(uniChar!=0xB1) &&(uniChar!=0xD7) &&(uniChar!=0xF7) + &&(uniChar!=0xB0) &&(uniChar!=0xB4) &&(uniChar!=0xB6) &&(uniChar!=0xA7) &&(uniChar!=0xA8)) + { + extraInfo->localeConverterIndex = ULMBCS_GRP_L1; + } + if (((uniChar > ULMBCS_C0END) && (uniChar < ULMBCS_C1START)) || + uniChar == 0 || uniChar == ULMBCS_HT || uniChar == ULMBCS_CR || + uniChar == ULMBCS_LF || uniChar == ULMBCS_123SYSTEMRANGE + ) + { + *pLMBCS++ = (ulmbcs_byte_t ) uniChar; + bytes_written = 1; + } + + + if (!bytes_written) + { + /* Check by UNICODE range (Strategy 2) */ + ulmbcs_byte_t group = FindLMBCSUniRange(uniChar); + + if (group == ULMBCS_GRP_UNICODE) /* (Strategy 2A) */ + { + pLMBCS += LMBCSConvertUni(pLMBCS,uniChar); + + bytes_written = (int32_t)(pLMBCS - LMBCS); + } + else if (group == ULMBCS_GRP_CTRL) /* (Strategy 2B) */ + { + /* Handle control characters here */ + if (uniChar <= ULMBCS_C0END) + { + *pLMBCS++ = ULMBCS_GRP_CTRL; + *pLMBCS++ = (ulmbcs_byte_t)(ULMBCS_CTRLOFFSET + uniChar); + } + else if (uniChar >= ULMBCS_C1START && uniChar <= ULMBCS_C1START + ULMBCS_CTRLOFFSET) + { + *pLMBCS++ = ULMBCS_GRP_CTRL; + *pLMBCS++ = (ulmbcs_byte_t ) (uniChar & 0x00FF); + } + bytes_written = (int32_t)(pLMBCS - LMBCS); + } + else if (group < ULMBCS_GRP_UNICODE) /* (Strategy 2C) */ + { + /* a specific converter has been identified - use it */ + bytes_written = (int32_t)LMBCSConversionWorker ( + extraInfo, group, pLMBCS, &uniChar, + &lastConverterIndex, groups_tried); + } + if (!bytes_written) /* the ambiguous group cases (Strategy 3) */ + { + uprv_memset(groups_tried, 0, sizeof(groups_tried)); + + /* check for non-default optimization group (Strategy 3A )*/ + if ((extraInfo->OptGroup != 1) && (ULMBCS_AMBIGUOUS_MATCH(group, extraInfo->OptGroup))) + { + /*zhujin: upgrade, merge #39299 here (Lotus) */ + /*To make R5 compatible translation, look for exceptional group first for non-DBCS*/ + + if(extraInfo->localeConverterIndex < ULMBCS_DOUBLEOPTGROUP_START) + { + bytes_written = (int32_t)LMBCSConversionWorker (extraInfo, + ULMBCS_GRP_L1, pLMBCS, &uniChar, + &lastConverterIndex, groups_tried); + + if(!bytes_written) + { + bytes_written = (int32_t)LMBCSConversionWorker (extraInfo, + ULMBCS_GRP_EXCEPT, pLMBCS, &uniChar, + &lastConverterIndex, groups_tried); + } + if(!bytes_written) + { + bytes_written = (int32_t)LMBCSConversionWorker (extraInfo, + extraInfo->localeConverterIndex, pLMBCS, &uniChar, + &lastConverterIndex, groups_tried); + } + } + else + { + bytes_written = (int32_t)LMBCSConversionWorker (extraInfo, + extraInfo->localeConverterIndex, pLMBCS, &uniChar, + &lastConverterIndex, groups_tried); + } + } + /* check for locale optimization group (Strategy 3B) */ + if (!bytes_written && (extraInfo->localeConverterIndex) && (ULMBCS_AMBIGUOUS_MATCH(group, extraInfo->localeConverterIndex))) + { + bytes_written = (int32_t)LMBCSConversionWorker (extraInfo, + extraInfo->localeConverterIndex, pLMBCS, &uniChar, &lastConverterIndex, groups_tried); + } + /* check for last optimization group used for this string (Strategy 3C) */ + if (!bytes_written && (lastConverterIndex) && (ULMBCS_AMBIGUOUS_MATCH(group, lastConverterIndex))) + { + bytes_written = (int32_t)LMBCSConversionWorker (extraInfo, + lastConverterIndex, pLMBCS, &uniChar, &lastConverterIndex, groups_tried); + } + if (!bytes_written) + { + /* just check every possible matching converter (Strategy 3D) */ + ulmbcs_byte_t grp_start; + ulmbcs_byte_t grp_end; + ulmbcs_byte_t grp_ix; + grp_start = (ulmbcs_byte_t)((group == ULMBCS_AMBIGUOUS_MBCS) + ? ULMBCS_DOUBLEOPTGROUP_START + : ULMBCS_GRP_L1); + grp_end = (ulmbcs_byte_t)((group == ULMBCS_AMBIGUOUS_MBCS) + ? ULMBCS_GRP_LAST + : ULMBCS_GRP_TH); + if(group == ULMBCS_AMBIGUOUS_ALL) + { + grp_start = ULMBCS_GRP_L1; + grp_end = ULMBCS_GRP_LAST; + } + for (grp_ix = grp_start; + grp_ix <= grp_end && !bytes_written; + grp_ix++) + { + if (extraInfo->OptGrpConverter [grp_ix] && !groups_tried [grp_ix]) + { + bytes_written = (int32_t)LMBCSConversionWorker (extraInfo, + grp_ix, pLMBCS, &uniChar, + &lastConverterIndex, groups_tried); + } + } + /* a final conversion fallback to the exceptions group if its likely + to be single byte (Strategy 3E) */ + if (!bytes_written && grp_start == ULMBCS_GRP_L1) + { + bytes_written = (int32_t)LMBCSConversionWorker (extraInfo, + ULMBCS_GRP_EXCEPT, pLMBCS, &uniChar, + &lastConverterIndex, groups_tried); + } + } + /* all of our other strategies failed. Fallback to Unicode. (Strategy 4)*/ + if (!bytes_written) + { + + pLMBCS += LMBCSConvertUni(pLMBCS, uniChar); + bytes_written = (int32_t)(pLMBCS - LMBCS); + } + } + } + + /* we have a translation. increment source and write as much as possible to target */ + args->source++; + pLMBCS = LMBCS; + while (args->target < args->targetLimit && bytes_written--) + { + *(args->target)++ = *pLMBCS++; + if (args->offsets) + { + *(args->offsets)++ = sourceIndex; + } + } + sourceIndex++; + if (bytes_written > 0) + { + /* write any bytes that didn't fit in target to the error buffer, + common code will move this to target if we get called back with + enough target room + */ + uint8_t * pErrorBuffer = args->converter->charErrorBuffer; + *err = U_BUFFER_OVERFLOW_ERROR; + args->converter->charErrorBufferLength = (int8_t)bytes_written; + while (bytes_written--) + { + *pErrorBuffer++ = *pLMBCS++; + } + } + /*Fix for SPR#DJOE66JFN3 (Lotus)*/ + extraInfo->localeConverterIndex = OldConverterIndex; + } +} + + +/* Now, the Unicode from LMBCS section */ + + +/* A function to call when we are looking at the Unicode group byte in LMBCS */ +static char16_t +GetUniFromLMBCSUni(char const ** ppLMBCSin) /* Called with LMBCS-style Unicode byte stream */ +{ + uint8_t HighCh = *(*ppLMBCSin)++; /* Big-endian Unicode in LMBCS compatibility group*/ + uint8_t LowCh = *(*ppLMBCSin)++; + + if (HighCh == ULMBCS_UNICOMPATZERO ) + { + HighCh = LowCh; + LowCh = 0; /* zero-byte in LSB special character */ + } + return (char16_t)((HighCh << 8) | LowCh); +} + + + +/* CHECK_SOURCE_LIMIT: Helper macro to verify that there are at least'index' + bytes left in source up to sourceLimit.Errors appropriately if not. + If we reach the limit, then update the source pointer to there to consume + all input as required by ICU converter semantics. +*/ + +#define CHECK_SOURCE_LIMIT(index) UPRV_BLOCK_MACRO_BEGIN { \ + if (args->source+index > args->sourceLimit) { \ + *err = U_TRUNCATED_CHAR_FOUND; \ + args->source = args->sourceLimit; \ + return 0xffff; \ + } \ +} UPRV_BLOCK_MACRO_END + +/* Return the Unicode representation for the current LMBCS character */ + +static UChar32 U_CALLCONV +_LMBCSGetNextUCharWorker(UConverterToUnicodeArgs* args, + UErrorCode* err) +{ + UChar32 uniChar = 0; /* an output UNICODE char */ + ulmbcs_byte_t CurByte; /* A byte from the input stream */ + + /* error check */ + if (args->source >= args->sourceLimit) + { + *err = U_ILLEGAL_ARGUMENT_ERROR; + return 0xffff; + } + /* Grab first byte & save address for error recovery */ + CurByte = *((ulmbcs_byte_t *) (args->source++)); + + /* + * at entry of each if clause: + * 1. 'CurByte' points at the first byte of a LMBCS character + * 2. '*source'points to the next byte of the source stream after 'CurByte' + * + * the job of each if clause is: + * 1. set '*source' to point at the beginning of next char (nop if LMBCS char is only 1 byte) + * 2. set 'uniChar' up with the right Unicode value, or set 'err' appropriately + */ + + /* First lets check the simple fixed values. */ + + if(((CurByte > ULMBCS_C0END) && (CurByte < ULMBCS_C1START)) /* ascii range */ + || (CurByte == 0) + || CurByte == ULMBCS_HT || CurByte == ULMBCS_CR + || CurByte == ULMBCS_LF || CurByte == ULMBCS_123SYSTEMRANGE) + { + uniChar = CurByte; + } + else + { + UConverterDataLMBCS * extraInfo; + ulmbcs_byte_t group; + UConverterSharedData *cnv; + + if (CurByte == ULMBCS_GRP_CTRL) /* Control character group - no opt group update */ + { + ulmbcs_byte_t C0C1byte; + CHECK_SOURCE_LIMIT(1); + C0C1byte = *(args->source)++; + uniChar = (C0C1byte < ULMBCS_C1START) ? C0C1byte - ULMBCS_CTRLOFFSET : C0C1byte; + } + else + if (CurByte == ULMBCS_GRP_UNICODE) /* Unicode compatibility group: BigEndian UTF16 */ + { + CHECK_SOURCE_LIMIT(2); + + /* don't check for error indicators fffe/ffff below */ + return GetUniFromLMBCSUni(&(args->source)); + } + else if (CurByte <= ULMBCS_CTRLOFFSET) + { + group = CurByte; /* group byte is in the source */ + extraInfo = (UConverterDataLMBCS *) args->converter->extraInfo; + if (group > ULMBCS_GRP_LAST || (cnv = extraInfo->OptGrpConverter[group]) == nullptr) + { + /* this is not a valid group byte - no converter*/ + *err = U_INVALID_CHAR_FOUND; + } + else if (group >= ULMBCS_DOUBLEOPTGROUP_START) /* double byte conversion */ + { + + CHECK_SOURCE_LIMIT(2); + + /* check for LMBCS doubled-group-byte case */ + if (*args->source == group) { + /* single byte */ + ++args->source; + uniChar = ucnv_MBCSSimpleGetNextUChar(cnv, args->source, 1, false); + ++args->source; + } else { + /* double byte */ + uniChar = ucnv_MBCSSimpleGetNextUChar(cnv, args->source, 2, false); + args->source += 2; + } + } + else { /* single byte conversion */ + CHECK_SOURCE_LIMIT(1); + CurByte = *(args->source)++; + + if (CurByte >= ULMBCS_C1START) + { + uniChar = _MBCS_SINGLE_SIMPLE_GET_NEXT_BMP(cnv, CurByte); + } + else + { + /* The non-optimizable oddballs where there is an explicit byte + * AND the second byte is not in the upper ascii range + */ + char bytes[2]; + + extraInfo = (UConverterDataLMBCS *) args->converter->extraInfo; + cnv = extraInfo->OptGrpConverter [ULMBCS_GRP_EXCEPT]; + + /* Lookup value must include opt group */ + bytes[0] = group; + bytes[1] = CurByte; + uniChar = ucnv_MBCSSimpleGetNextUChar(cnv, bytes, 2, false); + } + } + } + else if (CurByte >= ULMBCS_C1START) /* group byte is implicit */ + { + extraInfo = (UConverterDataLMBCS *) args->converter->extraInfo; + group = extraInfo->OptGroup; + cnv = extraInfo->OptGrpConverter[group]; + if (group >= ULMBCS_DOUBLEOPTGROUP_START) /* double byte conversion */ + { + if (!ucnv_MBCSIsLeadByte(cnv, CurByte)) + { + CHECK_SOURCE_LIMIT(0); + + /* let the MBCS conversion consume CurByte again */ + uniChar = ucnv_MBCSSimpleGetNextUChar(cnv, args->source - 1, 1, false); + } + else + { + CHECK_SOURCE_LIMIT(1); + /* let the MBCS conversion consume CurByte again */ + uniChar = ucnv_MBCSSimpleGetNextUChar(cnv, args->source - 1, 2, false); + ++args->source; + } + } + else /* single byte conversion */ + { + uniChar = _MBCS_SINGLE_SIMPLE_GET_NEXT_BMP(cnv, CurByte); + } + } + } + return uniChar; +} + + +/* The exported function that converts lmbcs to one or more + UChars - currently UTF-16 +*/ +static void U_CALLCONV +_LMBCSToUnicodeWithOffsets(UConverterToUnicodeArgs* args, + UErrorCode* err) +{ + char LMBCS [ULMBCS_CHARSIZE_MAX]; + char16_t uniChar; /* one output UNICODE char */ + const char * saveSource; /* beginning of current code point */ + const char * pStartLMBCS = args->source; /* beginning of whole string */ + const char * errSource = nullptr; /* pointer to actual input in case an error occurs */ + int8_t savebytes = 0; + + /* Process from source to limit, or until error */ + while (U_SUCCESS(*err) && args->sourceLimit > args->source && args->targetLimit > args->target) + { + saveSource = args->source; /* beginning of current code point */ + + if (args->converter->toULength) /* reassemble char from previous call */ + { + const char *saveSourceLimit; + size_t size_old = args->converter->toULength; + + /* limit from source is either remainder of temp buffer, or user limit on source */ + size_t size_new_maybe_1 = sizeof(LMBCS) - size_old; + size_t size_new_maybe_2 = args->sourceLimit - args->source; + size_t size_new = (size_new_maybe_1 < size_new_maybe_2) ? size_new_maybe_1 : size_new_maybe_2; + + + uprv_memcpy(LMBCS, args->converter->toUBytes, size_old); + uprv_memcpy(LMBCS + size_old, args->source, size_new); + saveSourceLimit = args->sourceLimit; + args->source = errSource = LMBCS; + args->sourceLimit = LMBCS+size_old+size_new; + savebytes = (int8_t)(size_old+size_new); + uniChar = (char16_t) _LMBCSGetNextUCharWorker(args, err); + args->source = saveSource + ((args->source - LMBCS) - size_old); + args->sourceLimit = saveSourceLimit; + + if (*err == U_TRUNCATED_CHAR_FOUND) + { + /* evil special case: source buffers so small a char spans more than 2 buffers */ + args->converter->toULength = savebytes; + uprv_memcpy(args->converter->toUBytes, LMBCS, savebytes); + args->source = args->sourceLimit; + *err = U_ZERO_ERROR; + return; + } + else + { + /* clear the partial-char marker */ + args->converter->toULength = 0; + } + } + else + { + errSource = saveSource; + uniChar = (char16_t) _LMBCSGetNextUCharWorker(args, err); + savebytes = (int8_t)(args->source - saveSource); + } + if (U_SUCCESS(*err)) + { + if (uniChar < 0xfffe) + { + *(args->target)++ = uniChar; + if(args->offsets) + { + *(args->offsets)++ = (int32_t)(saveSource - pStartLMBCS); + } + } + else if (uniChar == 0xfffe) + { + *err = U_INVALID_CHAR_FOUND; + } + else /* if (uniChar == 0xffff) */ + { + *err = U_ILLEGAL_CHAR_FOUND; + } + } + } + /* if target ran out before source, return U_BUFFER_OVERFLOW_ERROR */ + if (U_SUCCESS(*err) && args->sourceLimit > args->source && args->targetLimit <= args->target) + { + *err = U_BUFFER_OVERFLOW_ERROR; + } + else if (U_FAILURE(*err)) + { + /* If character incomplete or unmappable/illegal, store it in toUBytes[] */ + args->converter->toULength = savebytes; + if (savebytes > 0) { + uprv_memcpy(args->converter->toUBytes, errSource, savebytes); + } + if (*err == U_TRUNCATED_CHAR_FOUND) { + *err = U_ZERO_ERROR; + } + } +} + +/* And now, the macroized declarations of data & functions: */ +DEFINE_LMBCS_OPEN(1) +DEFINE_LMBCS_OPEN(2) +DEFINE_LMBCS_OPEN(3) +DEFINE_LMBCS_OPEN(4) +DEFINE_LMBCS_OPEN(5) +DEFINE_LMBCS_OPEN(6) +DEFINE_LMBCS_OPEN(8) +DEFINE_LMBCS_OPEN(11) +DEFINE_LMBCS_OPEN(16) +DEFINE_LMBCS_OPEN(17) +DEFINE_LMBCS_OPEN(18) +DEFINE_LMBCS_OPEN(19) + + +DECLARE_LMBCS_DATA(1) +DECLARE_LMBCS_DATA(2) +DECLARE_LMBCS_DATA(3) +DECLARE_LMBCS_DATA(4) +DECLARE_LMBCS_DATA(5) +DECLARE_LMBCS_DATA(6) +DECLARE_LMBCS_DATA(8) +DECLARE_LMBCS_DATA(11) +DECLARE_LMBCS_DATA(16) +DECLARE_LMBCS_DATA(17) +DECLARE_LMBCS_DATA(18) +DECLARE_LMBCS_DATA(19) + +U_CDECL_END + +#endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */ diff --git a/intl/icu/source/common/ucnv_set.cpp b/intl/icu/source/common/ucnv_set.cpp new file mode 100644 index 0000000000..872d8bc066 --- /dev/null +++ b/intl/icu/source/common/ucnv_set.cpp @@ -0,0 +1,70 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2003-2007, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: ucnv_set.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2004sep07 +* created by: Markus W. Scherer +* +* Conversion API functions using USet (ucnv_getUnicodeSet()) +* moved here from ucnv.c for removing the dependency of other ucnv_ +* implementation functions on the USet implementation. +*/ + +#include "unicode/utypes.h" +#include "unicode/uset.h" +#include "unicode/ucnv.h" +#include "ucnv_bld.h" +#include "uset_imp.h" + +#if !UCONFIG_NO_CONVERSION + +U_CAPI void U_EXPORT2 +ucnv_getUnicodeSet(const UConverter *cnv, + USet *setFillIn, + UConverterUnicodeSet whichSet, + UErrorCode *pErrorCode) { + /* argument checking */ + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return; + } + if(cnv==nullptr || setFillIn==nullptr || whichSetsharedData->impl->getUnicodeSet==nullptr) { + *pErrorCode=U_UNSUPPORTED_ERROR; + return; + } + + { + USetAdder sa={ + nullptr, + uset_add, + uset_addRange, + uset_addString, + uset_remove, + uset_removeRange + }; + sa.set=setFillIn; + + /* empty the set */ + uset_clear(setFillIn); + + /* call the converter to add the code points it supports */ + cnv->sharedData->impl->getUnicodeSet(cnv, &sa, whichSet, pErrorCode); + } +} + +#endif diff --git a/intl/icu/source/common/ucnv_u16.cpp b/intl/icu/source/common/ucnv_u16.cpp new file mode 100644 index 0000000000..c3bcfef50c --- /dev/null +++ b/intl/icu/source/common/ucnv_u16.cpp @@ -0,0 +1,1579 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2002-2015, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* file name: ucnv_u16.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002jul01 +* created by: Markus W. Scherer +* +* UTF-16 converter implementation. Used to be in ucnv_utf.c. +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/uversion.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" +#include "cmemory.h" + +enum { + UCNV_NEED_TO_WRITE_BOM=1 +}; + +U_CDECL_BEGIN +/* + * The UTF-16 toUnicode implementation is also used for the Java-specific + * "with BOM" variants of UTF-16BE and UTF-16LE. + */ +static void U_CALLCONV +_UTF16ToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode); + +/* UTF-16BE ----------------------------------------------------------------- */ + +#if U_IS_BIG_ENDIAN +# define _UTF16PEFromUnicodeWithOffsets _UTF16BEFromUnicodeWithOffsets +#else +# define _UTF16PEFromUnicodeWithOffsets _UTF16LEFromUnicodeWithOffsets +#endif + + +static void U_CALLCONV +_UTF16BEFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const char16_t *source; + char *target; + int32_t *offsets; + + uint32_t targetCapacity, length, sourceIndex; + char16_t c, trail; + char overflow[4]; + + source=pArgs->source; + length=(int32_t)(pArgs->sourceLimit-source); + if(length<=0) { + /* no input, nothing to do */ + return; + } + + cnv=pArgs->converter; + + /* write the BOM if necessary */ + if(cnv->fromUnicodeStatus==UCNV_NEED_TO_WRITE_BOM) { + static const char bom[]={ (char)0xfeu, (char)0xffu }; + ucnv_fromUWriteBytes(cnv, + bom, 2, + &pArgs->target, pArgs->targetLimit, + &pArgs->offsets, -1, + pErrorCode); + cnv->fromUnicodeStatus=0; + } + + target=pArgs->target; + if(target >= pArgs->targetLimit) { + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + return; + } + + targetCapacity=(uint32_t)(pArgs->targetLimit-target); + offsets=pArgs->offsets; + sourceIndex=0; + + /* c!=0 indicates in several places outside the main loops that a surrogate was found */ + + if((c=(char16_t)cnv->fromUChar32)!=0 && U16_IS_TRAIL(trail=*source) && targetCapacity>=4) { + /* the last buffer ended with a lead surrogate, output the surrogate pair */ + ++source; + --length; + target[0]=(uint8_t)(c>>8); + target[1]=(uint8_t)c; + target[2]=(uint8_t)(trail>>8); + target[3]=(uint8_t)trail; + target+=4; + targetCapacity-=4; + if(offsets!=nullptr) { + *offsets++=-1; + *offsets++=-1; + *offsets++=-1; + *offsets++=-1; + } + sourceIndex=1; + cnv->fromUChar32=c=0; + } + + if(c==0) { + /* copy an even number of bytes for complete UChars */ + uint32_t count=2*length; + if(count>targetCapacity) { + count=targetCapacity&~1; + } + /* count is even */ + targetCapacity-=count; + count>>=1; + length-=count; + + if(offsets==nullptr) { + while(count>0) { + c=*source++; + if(U16_IS_SINGLE(c)) { + target[0]=(uint8_t)(c>>8); + target[1]=(uint8_t)c; + target+=2; + } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && U16_IS_TRAIL(trail=*source)) { + ++source; + --count; + target[0]=(uint8_t)(c>>8); + target[1]=(uint8_t)c; + target[2]=(uint8_t)(trail>>8); + target[3]=(uint8_t)trail; + target+=4; + } else { + break; + } + --count; + } + } else { + while(count>0) { + c=*source++; + if(U16_IS_SINGLE(c)) { + target[0]=(uint8_t)(c>>8); + target[1]=(uint8_t)c; + target+=2; + *offsets++=sourceIndex; + *offsets++=sourceIndex++; + } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && U16_IS_TRAIL(trail=*source)) { + ++source; + --count; + target[0]=(uint8_t)(c>>8); + target[1]=(uint8_t)c; + target[2]=(uint8_t)(trail>>8); + target[3]=(uint8_t)trail; + target+=4; + *offsets++=sourceIndex; + *offsets++=sourceIndex; + *offsets++=sourceIndex; + *offsets++=sourceIndex; + sourceIndex+=2; + } else { + break; + } + --count; + } + } + + if(count==0) { + /* done with the loop for complete UChars */ + if(length>0 && targetCapacity>0) { + /* + * there is more input and some target capacity - + * it must be targetCapacity==1 because otherwise + * the above would have copied more; + * prepare for overflow output + */ + if(U16_IS_SINGLE(c=*source++)) { + overflow[0]=(char)(c>>8); + overflow[1]=(char)c; + length=2; /* 2 bytes to output */ + c=0; + /* } else { keep c for surrogate handling, length will be set there */ + } + } else { + length=0; + c=0; + } + } else { + /* keep c for surrogate handling, length will be set there */ + targetCapacity+=2*count; + } + } else { + length=0; /* from here on, length counts the bytes in overflow[] */ + } + + if(c!=0) { + /* + * c is a surrogate, and + * - source or target too short + * - or the surrogate is unmatched + */ + length=0; + if(U16_IS_SURROGATE_LEAD(c)) { + if(sourcesourceLimit) { + if(U16_IS_TRAIL(trail=*source)) { + /* output the surrogate pair, will overflow (see conditions comment above) */ + ++source; + overflow[0]=(char)(c>>8); + overflow[1]=(char)c; + overflow[2]=(char)(trail>>8); + overflow[3]=(char)trail; + length=4; /* 4 bytes to output */ + c=0; + } else { + /* unmatched lead surrogate */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + } else { + /* see if the trail surrogate is in the next buffer */ + } + } else { + /* unmatched trail surrogate */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + cnv->fromUChar32=c; + } + + if(length>0) { + /* output length bytes with overflow (length>targetCapacity>0) */ + ucnv_fromUWriteBytes(cnv, + overflow, length, + (char **)&target, pArgs->targetLimit, + &offsets, sourceIndex, + pErrorCode); + targetCapacity=(uint32_t)(pArgs->targetLimit-(char *)target); + } + + if(U_SUCCESS(*pErrorCode) && sourcesourceLimit && targetCapacity==0) { + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + + /* write back the updated pointers */ + pArgs->source=source; + pArgs->target=(char *)target; + pArgs->offsets=offsets; +} + +static void U_CALLCONV +_UTF16BEToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const uint8_t *source; + char16_t *target; + int32_t *offsets; + + uint32_t targetCapacity, length, count, sourceIndex; + char16_t c, trail; + + if(pArgs->converter->mode<8) { + _UTF16ToUnicodeWithOffsets(pArgs, pErrorCode); + return; + } + + cnv=pArgs->converter; + source=(const uint8_t *)pArgs->source; + length=(int32_t)((const uint8_t *)pArgs->sourceLimit-source); + if(length<=0 && cnv->toUnicodeStatus==0) { + /* no input, nothing to do */ + return; + } + + target=pArgs->target; + if(target >= pArgs->targetLimit) { + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + return; + } + + targetCapacity=(uint32_t)(pArgs->targetLimit-target); + offsets=pArgs->offsets; + sourceIndex=0; + c=0; + + /* complete a partial char16_t or pair from the last call */ + if(cnv->toUnicodeStatus!=0) { + /* + * special case: single byte from a previous buffer, + * where the byte turned out not to belong to a trail surrogate + * and the preceding, unmatched lead surrogate was put into toUBytes[] + * for error handling + */ + cnv->toUBytes[0]=(uint8_t)cnv->toUnicodeStatus; + cnv->toULength=1; + cnv->toUnicodeStatus=0; + } + if((count=cnv->toULength)!=0) { + uint8_t *p=cnv->toUBytes; + do { + p[count++]=*source++; + ++sourceIndex; + --length; + if(count==2) { + c=((char16_t)p[0]<<8)|p[1]; + if(U16_IS_SINGLE(c)) { + /* output the BMP code point */ + *target++=c; + if(offsets!=nullptr) { + *offsets++=-1; + } + --targetCapacity; + count=0; + c=0; + break; + } else if(U16_IS_SURROGATE_LEAD(c)) { + /* continue collecting bytes for the trail surrogate */ + c=0; /* avoid unnecessary surrogate handling below */ + } else { + /* fall through to error handling for an unmatched trail surrogate */ + break; + } + } else if(count==4) { + c=((char16_t)p[0]<<8)|p[1]; + trail=((char16_t)p[2]<<8)|p[3]; + if(U16_IS_TRAIL(trail)) { + /* output the surrogate pair */ + *target++=c; + if(targetCapacity>=2) { + *target++=trail; + if(offsets!=nullptr) { + *offsets++=-1; + *offsets++=-1; + } + targetCapacity-=2; + } else /* targetCapacity==1 */ { + targetCapacity=0; + cnv->UCharErrorBuffer[0]=trail; + cnv->UCharErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + count=0; + c=0; + break; + } else { + /* unmatched lead surrogate, handle here for consistent toUBytes[] */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + + /* back out reading the code unit after it */ + if(((const uint8_t *)pArgs->source-source)>=2) { + source-=2; + } else { + /* + * if the trail unit's first byte was in a previous buffer, then + * we need to put it into a special place because toUBytes[] will be + * used for the lead unit's bytes + */ + cnv->toUnicodeStatus=0x100|p[2]; + --source; + } + cnv->toULength=2; + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + pArgs->offsets=offsets; + return; + } + } + } while(length>0); + cnv->toULength=(int8_t)count; + } + + /* copy an even number of bytes for complete UChars */ + count=2*targetCapacity; + if(count>length) { + count=length&~1; + } + if(c==0 && count>0) { + length-=count; + count>>=1; + targetCapacity-=count; + if(offsets==nullptr) { + do { + c=((char16_t)source[0]<<8)|source[1]; + source+=2; + if(U16_IS_SINGLE(c)) { + *target++=c; + } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && + U16_IS_TRAIL(trail=((char16_t)source[0]<<8)|source[1]) + ) { + source+=2; + --count; + *target++=c; + *target++=trail; + } else { + break; + } + } while(--count>0); + } else { + do { + c=((char16_t)source[0]<<8)|source[1]; + source+=2; + if(U16_IS_SINGLE(c)) { + *target++=c; + *offsets++=sourceIndex; + sourceIndex+=2; + } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && + U16_IS_TRAIL(trail=((char16_t)source[0]<<8)|source[1]) + ) { + source+=2; + --count; + *target++=c; + *target++=trail; + *offsets++=sourceIndex; + *offsets++=sourceIndex; + sourceIndex+=4; + } else { + break; + } + } while(--count>0); + } + + if(count==0) { + /* done with the loop for complete UChars */ + c=0; + } else { + /* keep c for surrogate handling, trail will be set there */ + length+=2*(count-1); /* one more byte pair was consumed than count decremented */ + targetCapacity+=count; + } + } + + if(c!=0) { + /* + * c is a surrogate, and + * - source or target too short + * - or the surrogate is unmatched + */ + cnv->toUBytes[0]=(uint8_t)(c>>8); + cnv->toUBytes[1]=(uint8_t)c; + cnv->toULength=2; + + if(U16_IS_SURROGATE_LEAD(c)) { + if(length>=2) { + if(U16_IS_TRAIL(trail=((char16_t)source[0]<<8)|source[1])) { + /* output the surrogate pair, will overflow (see conditions comment above) */ + source+=2; + length-=2; + *target++=c; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + cnv->UCharErrorBuffer[0]=trail; + cnv->UCharErrorBufferLength=1; + cnv->toULength=0; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } else { + /* unmatched lead surrogate */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + } else { + /* see if the trail surrogate is in the next buffer */ + } + } else { + /* unmatched trail surrogate */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + } + + if(U_SUCCESS(*pErrorCode)) { + /* check for a remaining source byte */ + if(length>0) { + if(targetCapacity==0) { + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } else { + /* it must be length==1 because otherwise the above would have copied more */ + cnv->toUBytes[cnv->toULength++]=*source++; + } + } + } + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + pArgs->offsets=offsets; +} + +static UChar32 U_CALLCONV +_UTF16BEGetNextUChar(UConverterToUnicodeArgs *pArgs, UErrorCode *err) { + const uint8_t *s, *sourceLimit; + UChar32 c; + + if(pArgs->converter->mode<8) { + return UCNV_GET_NEXT_UCHAR_USE_TO_U; + } + + s=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + + if(s>=sourceLimit) { + /* no input */ + *err=U_INDEX_OUTOFBOUNDS_ERROR; + return 0xffff; + } + + if(s+2>sourceLimit) { + /* only one byte: truncated char16_t */ + pArgs->converter->toUBytes[0]=*s++; + pArgs->converter->toULength=1; + pArgs->source=(const char *)s; + *err = U_TRUNCATED_CHAR_FOUND; + return 0xffff; + } + + /* get one char16_t */ + c=((UChar32)*s<<8)|s[1]; + s+=2; + + /* check for a surrogate pair */ + if(U_IS_SURROGATE(c)) { + if(U16_IS_SURROGATE_LEAD(c)) { + if(s+2<=sourceLimit) { + char16_t trail; + + /* get a second char16_t and see if it is a trail surrogate */ + trail=((char16_t)*s<<8)|s[1]; + if(U16_IS_TRAIL(trail)) { + c=U16_GET_SUPPLEMENTARY(c, trail); + s+=2; + } else { + /* unmatched lead surrogate */ + c=-2; + } + } else { + /* too few (2 or 3) bytes for a surrogate pair: truncated code point */ + uint8_t *bytes=pArgs->converter->toUBytes; + s-=2; + pArgs->converter->toULength=(int8_t)(sourceLimit-s); + do { + *bytes++=*s++; + } while(sconverter->toUBytes; + pArgs->converter->toULength=2; + *bytes=*(s-2); + bytes[1]=*(s-1); + + c=0xffff; + *err=U_ILLEGAL_CHAR_FOUND; + } + } + + pArgs->source=(const char *)s; + return c; +} + +static void U_CALLCONV +_UTF16BEReset(UConverter *cnv, UConverterResetChoice choice) { + if(choice<=UCNV_RESET_TO_UNICODE) { + /* reset toUnicode state */ + if(UCNV_GET_VERSION(cnv)==0) { + cnv->mode=8; /* no BOM handling */ + } else { + cnv->mode=0; /* Java-specific "UnicodeBig" requires BE BOM or no BOM */ + } + } + if(choice!=UCNV_RESET_TO_UNICODE && UCNV_GET_VERSION(cnv)==1) { + /* reset fromUnicode for "UnicodeBig": prepare to output the UTF-16BE BOM */ + cnv->fromUnicodeStatus=UCNV_NEED_TO_WRITE_BOM; + } +} + +static void U_CALLCONV +_UTF16BEOpen(UConverter *cnv, + UConverterLoadArgs *pArgs, + UErrorCode *pErrorCode) { + (void)pArgs; + if(UCNV_GET_VERSION(cnv)<=1) { + _UTF16BEReset(cnv, UCNV_RESET_BOTH); + } else { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + } +} + +static const char * U_CALLCONV +_UTF16BEGetName(const UConverter *cnv) { + if(UCNV_GET_VERSION(cnv)==0) { + return "UTF-16BE"; + } else { + return "UTF-16BE,version=1"; + } +} +U_CDECL_END + +static const UConverterImpl _UTF16BEImpl={ + UCNV_UTF16_BigEndian, + + nullptr, + nullptr, + + _UTF16BEOpen, + nullptr, + _UTF16BEReset, + + _UTF16BEToUnicodeWithOffsets, + _UTF16BEToUnicodeWithOffsets, + _UTF16BEFromUnicodeWithOffsets, + _UTF16BEFromUnicodeWithOffsets, + _UTF16BEGetNextUChar, + + nullptr, + _UTF16BEGetName, + nullptr, + nullptr, + ucnv_getNonSurrogateUnicodeSet, + + nullptr, + nullptr +}; + +static const UConverterStaticData _UTF16BEStaticData={ + sizeof(UConverterStaticData), + "UTF-16BE", + 1200, UCNV_IBM, UCNV_UTF16_BigEndian, 2, 2, + { 0xff, 0xfd, 0, 0 },2,false,false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + + +const UConverterSharedData _UTF16BEData= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF16BEStaticData, &_UTF16BEImpl); + +/* UTF-16LE ----------------------------------------------------------------- */ +U_CDECL_BEGIN +static void U_CALLCONV +_UTF16LEFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const char16_t *source; + char *target; + int32_t *offsets; + + uint32_t targetCapacity, length, sourceIndex; + char16_t c, trail; + char overflow[4]; + + source=pArgs->source; + length=(int32_t)(pArgs->sourceLimit-source); + if(length<=0) { + /* no input, nothing to do */ + return; + } + + cnv=pArgs->converter; + + /* write the BOM if necessary */ + if(cnv->fromUnicodeStatus==UCNV_NEED_TO_WRITE_BOM) { + static const char bom[]={ (char)0xffu, (char)0xfeu }; + ucnv_fromUWriteBytes(cnv, + bom, 2, + &pArgs->target, pArgs->targetLimit, + &pArgs->offsets, -1, + pErrorCode); + cnv->fromUnicodeStatus=0; + } + + target=pArgs->target; + if(target >= pArgs->targetLimit) { + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + return; + } + + targetCapacity=(uint32_t)(pArgs->targetLimit-pArgs->target); + offsets=pArgs->offsets; + sourceIndex=0; + + /* c!=0 indicates in several places outside the main loops that a surrogate was found */ + + if((c=(char16_t)cnv->fromUChar32)!=0 && U16_IS_TRAIL(trail=*source) && targetCapacity>=4) { + /* the last buffer ended with a lead surrogate, output the surrogate pair */ + ++source; + --length; + target[0]=(uint8_t)c; + target[1]=(uint8_t)(c>>8); + target[2]=(uint8_t)trail; + target[3]=(uint8_t)(trail>>8); + target+=4; + targetCapacity-=4; + if(offsets!=nullptr) { + *offsets++=-1; + *offsets++=-1; + *offsets++=-1; + *offsets++=-1; + } + sourceIndex=1; + cnv->fromUChar32=c=0; + } + + if(c==0) { + /* copy an even number of bytes for complete UChars */ + uint32_t count=2*length; + if(count>targetCapacity) { + count=targetCapacity&~1; + } + /* count is even */ + targetCapacity-=count; + count>>=1; + length-=count; + + if(offsets==nullptr) { + while(count>0) { + c=*source++; + if(U16_IS_SINGLE(c)) { + target[0]=(uint8_t)c; + target[1]=(uint8_t)(c>>8); + target+=2; + } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && U16_IS_TRAIL(trail=*source)) { + ++source; + --count; + target[0]=(uint8_t)c; + target[1]=(uint8_t)(c>>8); + target[2]=(uint8_t)trail; + target[3]=(uint8_t)(trail>>8); + target+=4; + } else { + break; + } + --count; + } + } else { + while(count>0) { + c=*source++; + if(U16_IS_SINGLE(c)) { + target[0]=(uint8_t)c; + target[1]=(uint8_t)(c>>8); + target+=2; + *offsets++=sourceIndex; + *offsets++=sourceIndex++; + } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && U16_IS_TRAIL(trail=*source)) { + ++source; + --count; + target[0]=(uint8_t)c; + target[1]=(uint8_t)(c>>8); + target[2]=(uint8_t)trail; + target[3]=(uint8_t)(trail>>8); + target+=4; + *offsets++=sourceIndex; + *offsets++=sourceIndex; + *offsets++=sourceIndex; + *offsets++=sourceIndex; + sourceIndex+=2; + } else { + break; + } + --count; + } + } + + if(count==0) { + /* done with the loop for complete UChars */ + if(length>0 && targetCapacity>0) { + /* + * there is more input and some target capacity - + * it must be targetCapacity==1 because otherwise + * the above would have copied more; + * prepare for overflow output + */ + if(U16_IS_SINGLE(c=*source++)) { + overflow[0]=(char)c; + overflow[1]=(char)(c>>8); + length=2; /* 2 bytes to output */ + c=0; + /* } else { keep c for surrogate handling, length will be set there */ + } + } else { + length=0; + c=0; + } + } else { + /* keep c for surrogate handling, length will be set there */ + targetCapacity+=2*count; + } + } else { + length=0; /* from here on, length counts the bytes in overflow[] */ + } + + if(c!=0) { + /* + * c is a surrogate, and + * - source or target too short + * - or the surrogate is unmatched + */ + length=0; + if(U16_IS_SURROGATE_LEAD(c)) { + if(sourcesourceLimit) { + if(U16_IS_TRAIL(trail=*source)) { + /* output the surrogate pair, will overflow (see conditions comment above) */ + ++source; + overflow[0]=(char)c; + overflow[1]=(char)(c>>8); + overflow[2]=(char)trail; + overflow[3]=(char)(trail>>8); + length=4; /* 4 bytes to output */ + c=0; + } else { + /* unmatched lead surrogate */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + } else { + /* see if the trail surrogate is in the next buffer */ + } + } else { + /* unmatched trail surrogate */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + cnv->fromUChar32=c; + } + + if(length>0) { + /* output length bytes with overflow (length>targetCapacity>0) */ + ucnv_fromUWriteBytes(cnv, + overflow, length, + &target, pArgs->targetLimit, + &offsets, sourceIndex, + pErrorCode); + targetCapacity=(uint32_t)(pArgs->targetLimit-(char *)target); + } + + if(U_SUCCESS(*pErrorCode) && sourcesourceLimit && targetCapacity==0) { + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + + /* write back the updated pointers */ + pArgs->source=source; + pArgs->target=target; + pArgs->offsets=offsets; +} + +static void U_CALLCONV +_UTF16LEToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const uint8_t *source; + char16_t *target; + int32_t *offsets; + + uint32_t targetCapacity, length, count, sourceIndex; + char16_t c, trail; + + if(pArgs->converter->mode<8) { + _UTF16ToUnicodeWithOffsets(pArgs, pErrorCode); + return; + } + + cnv=pArgs->converter; + source=(const uint8_t *)pArgs->source; + length=(int32_t)((const uint8_t *)pArgs->sourceLimit-source); + if(length<=0 && cnv->toUnicodeStatus==0) { + /* no input, nothing to do */ + return; + } + + target=pArgs->target; + if(target >= pArgs->targetLimit) { + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + return; + } + + targetCapacity=(uint32_t)(pArgs->targetLimit-pArgs->target); + offsets=pArgs->offsets; + sourceIndex=0; + c=0; + + /* complete a partial char16_t or pair from the last call */ + if(cnv->toUnicodeStatus!=0) { + /* + * special case: single byte from a previous buffer, + * where the byte turned out not to belong to a trail surrogate + * and the preceding, unmatched lead surrogate was put into toUBytes[] + * for error handling + */ + cnv->toUBytes[0]=(uint8_t)cnv->toUnicodeStatus; + cnv->toULength=1; + cnv->toUnicodeStatus=0; + } + if((count=cnv->toULength)!=0) { + uint8_t *p=cnv->toUBytes; + do { + p[count++]=*source++; + ++sourceIndex; + --length; + if(count==2) { + c=((char16_t)p[1]<<8)|p[0]; + if(U16_IS_SINGLE(c)) { + /* output the BMP code point */ + *target++=c; + if(offsets!=nullptr) { + *offsets++=-1; + } + --targetCapacity; + count=0; + c=0; + break; + } else if(U16_IS_SURROGATE_LEAD(c)) { + /* continue collecting bytes for the trail surrogate */ + c=0; /* avoid unnecessary surrogate handling below */ + } else { + /* fall through to error handling for an unmatched trail surrogate */ + break; + } + } else if(count==4) { + c=((char16_t)p[1]<<8)|p[0]; + trail=((char16_t)p[3]<<8)|p[2]; + if(U16_IS_TRAIL(trail)) { + /* output the surrogate pair */ + *target++=c; + if(targetCapacity>=2) { + *target++=trail; + if(offsets!=nullptr) { + *offsets++=-1; + *offsets++=-1; + } + targetCapacity-=2; + } else /* targetCapacity==1 */ { + targetCapacity=0; + cnv->UCharErrorBuffer[0]=trail; + cnv->UCharErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + count=0; + c=0; + break; + } else { + /* unmatched lead surrogate, handle here for consistent toUBytes[] */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + + /* back out reading the code unit after it */ + if(((const uint8_t *)pArgs->source-source)>=2) { + source-=2; + } else { + /* + * if the trail unit's first byte was in a previous buffer, then + * we need to put it into a special place because toUBytes[] will be + * used for the lead unit's bytes + */ + cnv->toUnicodeStatus=0x100|p[2]; + --source; + } + cnv->toULength=2; + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + pArgs->offsets=offsets; + return; + } + } + } while(length>0); + cnv->toULength=(int8_t)count; + } + + /* copy an even number of bytes for complete UChars */ + count=2*targetCapacity; + if(count>length) { + count=length&~1; + } + if(c==0 && count>0) { + length-=count; + count>>=1; + targetCapacity-=count; + if(offsets==nullptr) { + do { + c=((char16_t)source[1]<<8)|source[0]; + source+=2; + if(U16_IS_SINGLE(c)) { + *target++=c; + } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && + U16_IS_TRAIL(trail=((char16_t)source[1]<<8)|source[0]) + ) { + source+=2; + --count; + *target++=c; + *target++=trail; + } else { + break; + } + } while(--count>0); + } else { + do { + c=((char16_t)source[1]<<8)|source[0]; + source+=2; + if(U16_IS_SINGLE(c)) { + *target++=c; + *offsets++=sourceIndex; + sourceIndex+=2; + } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && + U16_IS_TRAIL(trail=((char16_t)source[1]<<8)|source[0]) + ) { + source+=2; + --count; + *target++=c; + *target++=trail; + *offsets++=sourceIndex; + *offsets++=sourceIndex; + sourceIndex+=4; + } else { + break; + } + } while(--count>0); + } + + if(count==0) { + /* done with the loop for complete UChars */ + c=0; + } else { + /* keep c for surrogate handling, trail will be set there */ + length+=2*(count-1); /* one more byte pair was consumed than count decremented */ + targetCapacity+=count; + } + } + + if(c!=0) { + /* + * c is a surrogate, and + * - source or target too short + * - or the surrogate is unmatched + */ + cnv->toUBytes[0]=(uint8_t)c; + cnv->toUBytes[1]=(uint8_t)(c>>8); + cnv->toULength=2; + + if(U16_IS_SURROGATE_LEAD(c)) { + if(length>=2) { + if(U16_IS_TRAIL(trail=((char16_t)source[1]<<8)|source[0])) { + /* output the surrogate pair, will overflow (see conditions comment above) */ + source+=2; + length-=2; + *target++=c; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + cnv->UCharErrorBuffer[0]=trail; + cnv->UCharErrorBufferLength=1; + cnv->toULength=0; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } else { + /* unmatched lead surrogate */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + } else { + /* see if the trail surrogate is in the next buffer */ + } + } else { + /* unmatched trail surrogate */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + } + + if(U_SUCCESS(*pErrorCode)) { + /* check for a remaining source byte */ + if(length>0) { + if(targetCapacity==0) { + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } else { + /* it must be length==1 because otherwise the above would have copied more */ + cnv->toUBytes[cnv->toULength++]=*source++; + } + } + } + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + pArgs->offsets=offsets; +} + +static UChar32 U_CALLCONV +_UTF16LEGetNextUChar(UConverterToUnicodeArgs *pArgs, UErrorCode *err) { + const uint8_t *s, *sourceLimit; + UChar32 c; + + if(pArgs->converter->mode<8) { + return UCNV_GET_NEXT_UCHAR_USE_TO_U; + } + + s=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + + if(s>=sourceLimit) { + /* no input */ + *err=U_INDEX_OUTOFBOUNDS_ERROR; + return 0xffff; + } + + if(s+2>sourceLimit) { + /* only one byte: truncated char16_t */ + pArgs->converter->toUBytes[0]=*s++; + pArgs->converter->toULength=1; + pArgs->source=(const char *)s; + *err = U_TRUNCATED_CHAR_FOUND; + return 0xffff; + } + + /* get one char16_t */ + c=((UChar32)s[1]<<8)|*s; + s+=2; + + /* check for a surrogate pair */ + if(U_IS_SURROGATE(c)) { + if(U16_IS_SURROGATE_LEAD(c)) { + if(s+2<=sourceLimit) { + char16_t trail; + + /* get a second char16_t and see if it is a trail surrogate */ + trail=((char16_t)s[1]<<8)|*s; + if(U16_IS_TRAIL(trail)) { + c=U16_GET_SUPPLEMENTARY(c, trail); + s+=2; + } else { + /* unmatched lead surrogate */ + c=-2; + } + } else { + /* too few (2 or 3) bytes for a surrogate pair: truncated code point */ + uint8_t *bytes=pArgs->converter->toUBytes; + s-=2; + pArgs->converter->toULength=(int8_t)(sourceLimit-s); + do { + *bytes++=*s++; + } while(sconverter->toUBytes; + pArgs->converter->toULength=2; + *bytes=*(s-2); + bytes[1]=*(s-1); + + c=0xffff; + *err=U_ILLEGAL_CHAR_FOUND; + } + } + + pArgs->source=(const char *)s; + return c; +} + +static void U_CALLCONV +_UTF16LEReset(UConverter *cnv, UConverterResetChoice choice) { + if(choice<=UCNV_RESET_TO_UNICODE) { + /* reset toUnicode state */ + if(UCNV_GET_VERSION(cnv)==0) { + cnv->mode=8; /* no BOM handling */ + } else { + cnv->mode=0; /* Java-specific "UnicodeLittle" requires LE BOM or no BOM */ + } + } + if(choice!=UCNV_RESET_TO_UNICODE && UCNV_GET_VERSION(cnv)==1) { + /* reset fromUnicode for "UnicodeLittle": prepare to output the UTF-16LE BOM */ + cnv->fromUnicodeStatus=UCNV_NEED_TO_WRITE_BOM; + } +} + +static void U_CALLCONV +_UTF16LEOpen(UConverter *cnv, + UConverterLoadArgs *pArgs, + UErrorCode *pErrorCode) { + (void)pArgs; + if(UCNV_GET_VERSION(cnv)<=1) { + _UTF16LEReset(cnv, UCNV_RESET_BOTH); + } else { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + } +} + +static const char * U_CALLCONV +_UTF16LEGetName(const UConverter *cnv) { + if(UCNV_GET_VERSION(cnv)==0) { + return "UTF-16LE"; + } else { + return "UTF-16LE,version=1"; + } +} +U_CDECL_END + +static const UConverterImpl _UTF16LEImpl={ + UCNV_UTF16_LittleEndian, + + nullptr, + nullptr, + + _UTF16LEOpen, + nullptr, + _UTF16LEReset, + + _UTF16LEToUnicodeWithOffsets, + _UTF16LEToUnicodeWithOffsets, + _UTF16LEFromUnicodeWithOffsets, + _UTF16LEFromUnicodeWithOffsets, + _UTF16LEGetNextUChar, + + nullptr, + _UTF16LEGetName, + nullptr, + nullptr, + ucnv_getNonSurrogateUnicodeSet, + + nullptr, + nullptr +}; + + +static const UConverterStaticData _UTF16LEStaticData={ + sizeof(UConverterStaticData), + "UTF-16LE", + 1202, UCNV_IBM, UCNV_UTF16_LittleEndian, 2, 2, + { 0xfd, 0xff, 0, 0 },2,false,false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + + +const UConverterSharedData _UTF16LEData= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF16LEStaticData, &_UTF16LEImpl); + +/* UTF-16 (Detect BOM) ------------------------------------------------------ */ + +/* + * Detect a BOM at the beginning of the stream and select UTF-16BE or UTF-16LE + * accordingly. + * This is a simpler version of the UTF-32 converter, with + * fewer states for shorter BOMs. + * + * State values: + * 0 initial state + * 1 saw first byte + * 2..5 - + * 6..7 see _UTF16ToUnicodeWithOffsets() comments in state 1 + * 8 UTF-16BE mode + * 9 UTF-16LE mode + * + * During detection: state==number of initial bytes seen so far. + * + * On output, emit U+FEFF as the first code point. + * + * Variants: + * - UTF-16,version=1 (Java "Unicode" encoding) treats a missing BOM as an error. + * - UTF-16BE,version=1 (Java "UnicodeBig" encoding) and + * UTF-16LE,version=1 (Java "UnicodeLittle" encoding) treat a reverse BOM as an error. + */ +U_CDECL_BEGIN +static void U_CALLCONV +_UTF16Reset(UConverter *cnv, UConverterResetChoice choice) { + if(choice<=UCNV_RESET_TO_UNICODE) { + /* reset toUnicode: state=0 */ + cnv->mode=0; + } + if(choice!=UCNV_RESET_TO_UNICODE) { + /* reset fromUnicode: prepare to output the UTF-16PE BOM */ + cnv->fromUnicodeStatus=UCNV_NEED_TO_WRITE_BOM; + } +} +U_CDECL_END +extern const UConverterSharedData _UTF16v2Data; +U_CDECL_BEGIN +static void U_CALLCONV +_UTF16Open(UConverter *cnv, + UConverterLoadArgs *pArgs, + UErrorCode *pErrorCode) { + if(UCNV_GET_VERSION(cnv)<=2) { + if(UCNV_GET_VERSION(cnv)==2 && !pArgs->onlyTestIsLoadable) { + /* + * Switch implementation, and switch the staticData that's different + * and was copied into the UConverter. + * (See ucnv_createConverterFromSharedData() in ucnv_bld.c.) + * UTF-16,version=2 fromUnicode() always writes a big-endian byte stream. + */ + cnv->sharedData=(UConverterSharedData*)&_UTF16v2Data; + uprv_memcpy(cnv->subChars, _UTF16v2Data.staticData->subChar, UCNV_MAX_SUBCHAR_LEN); + } + _UTF16Reset(cnv, UCNV_RESET_BOTH); + } else { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + } +} + +static const char * U_CALLCONV +_UTF16GetName(const UConverter *cnv) { + if(UCNV_GET_VERSION(cnv)==0) { + return "UTF-16"; + } else if(UCNV_GET_VERSION(cnv)==1) { + return "UTF-16,version=1"; + } else { + return "UTF-16,version=2"; + } +} +U_CDECL_END +extern const UConverterSharedData _UTF16Data; + +static inline bool IS_UTF16BE(const UConverter *cnv) { + return ((cnv)->sharedData == &_UTF16BEData); +} + +static inline bool IS_UTF16LE(const UConverter *cnv) { + return ((cnv)->sharedData == &_UTF16LEData); +} + +static inline bool IS_UTF16(const UConverter *cnv) { + return ((cnv)->sharedData==&_UTF16Data) || ((cnv)->sharedData == &_UTF16v2Data); +} + +U_CDECL_BEGIN +static void U_CALLCONV +_UTF16ToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv=pArgs->converter; + const char *source=pArgs->source; + const char *sourceLimit=pArgs->sourceLimit; + int32_t *offsets=pArgs->offsets; + + int32_t state, offsetDelta; + uint8_t b; + + state=cnv->mode; + + /* + * If we detect a BOM in this buffer, then we must add the BOM size to the + * offsets because the actual converter function will not see and count the BOM. + * offsetDelta will have the number of the BOM bytes that are in the current buffer. + */ + offsetDelta=0; + + while(sourcetoUBytes[0]=(uint8_t)*source++; + cnv->toULength=1; + state=1; + break; + case 1: + /* + * Only inside this switch case can the state variable + * temporarily take two additional values: + * 6: BOM error, continue with BE + * 7: BOM error, continue with LE + */ + b=*source; + if(cnv->toUBytes[0]==0xfe && b==0xff) { + if(IS_UTF16LE(cnv)) { + state=7; /* illegal reverse BOM for Java "UnicodeLittle" */ + } else { + state=8; /* detect UTF-16BE */ + } + } else if(cnv->toUBytes[0]==0xff && b==0xfe) { + if(IS_UTF16BE(cnv)) { + state=6; /* illegal reverse BOM for Java "UnicodeBig" */ + } else { + state=9; /* detect UTF-16LE */ + } + } else if((IS_UTF16(cnv) && UCNV_GET_VERSION(cnv)==1)) { + state=6; /* illegal missing BOM for Java "Unicode" */ + } + if(state>=8) { + /* BOM detected, consume it */ + ++source; + cnv->toULength=0; + offsetDelta=(int32_t)(source-pArgs->source); + } else if(state<6) { + /* ok: no BOM, and not a reverse BOM */ + if(source!=pArgs->source) { + /* reset the source for a correct first offset */ + source=pArgs->source; + cnv->toULength=0; + } + if(IS_UTF16LE(cnv)) { + /* Make Java "UnicodeLittle" default to LE. */ + state=9; + } else { + /* Make standard UTF-16 and Java "UnicodeBig" default to BE. */ + state=8; + } + } else { + /* + * error: missing BOM, or reverse BOM + * UTF-16,version=1: Java-specific "Unicode" requires a BOM. + * UTF-16BE,version=1: Java-specific "UnicodeBig" requires a BE BOM or no BOM. + * UTF-16LE,version=1: Java-specific "UnicodeLittle" requires an LE BOM or no BOM. + */ + /* report the non-BOM or reverse BOM as an illegal sequence */ + cnv->toUBytes[1]=b; + cnv->toULength=2; + pArgs->source=source+1; + /* continue with conversion if the callback resets the error */ + /* + * Make Java "Unicode" default to BE like standard UTF-16. + * Make Java "UnicodeBig" and "UnicodeLittle" default + * to their normal endiannesses. + */ + cnv->mode=state+2; + *pErrorCode=U_ILLEGAL_ESCAPE_SEQUENCE; + return; + } + /* convert the rest of the stream */ + cnv->mode=state; + continue; + case 8: + /* call UTF-16BE */ + pArgs->source=source; + _UTF16BEToUnicodeWithOffsets(pArgs, pErrorCode); + source=pArgs->source; + break; + case 9: + /* call UTF-16LE */ + pArgs->source=source; + _UTF16LEToUnicodeWithOffsets(pArgs, pErrorCode); + source=pArgs->source; + break; + default: + break; /* does not occur */ + } + } + + /* add BOM size to offsets - see comment at offsetDelta declaration */ + if(offsets!=nullptr && offsetDelta!=0) { + int32_t *offsetsLimit=pArgs->offsets; + while(offsetssource=source; + + if(source==sourceLimit && pArgs->flush) { + /* handle truncated input */ + switch(state) { + case 0: + break; /* no input at all, nothing to do */ + case 8: + _UTF16BEToUnicodeWithOffsets(pArgs, pErrorCode); + break; + case 9: + _UTF16LEToUnicodeWithOffsets(pArgs, pErrorCode); + break; + default: + /* 0mode=state; +} + +static UChar32 U_CALLCONV +_UTF16GetNextUChar(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + switch(pArgs->converter->mode) { + case 8: + return _UTF16BEGetNextUChar(pArgs, pErrorCode); + case 9: + return _UTF16LEGetNextUChar(pArgs, pErrorCode); + default: + return UCNV_GET_NEXT_UCHAR_USE_TO_U; + } +} +U_CDECL_END + +static const UConverterImpl _UTF16Impl = { + UCNV_UTF16, + + nullptr, + nullptr, + + _UTF16Open, + nullptr, + _UTF16Reset, + + _UTF16ToUnicodeWithOffsets, + _UTF16ToUnicodeWithOffsets, + _UTF16PEFromUnicodeWithOffsets, + _UTF16PEFromUnicodeWithOffsets, + _UTF16GetNextUChar, + + nullptr, /* ### TODO implement getStarters for all Unicode encodings?! */ + _UTF16GetName, + nullptr, + nullptr, + ucnv_getNonSurrogateUnicodeSet, + + nullptr, + nullptr +}; + +static const UConverterStaticData _UTF16StaticData = { + sizeof(UConverterStaticData), + "UTF-16", + 1204, /* CCSID for BOM sensitive UTF-16 */ + UCNV_IBM, UCNV_UTF16, 2, 2, +#if U_IS_BIG_ENDIAN + { 0xff, 0xfd, 0, 0 }, 2, +#else + { 0xfd, 0xff, 0, 0 }, 2, +#endif + false, false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + +const UConverterSharedData _UTF16Data = + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF16StaticData, &_UTF16Impl); + +static const UConverterImpl _UTF16v2Impl = { + UCNV_UTF16, + + nullptr, + nullptr, + + _UTF16Open, + nullptr, + _UTF16Reset, + + _UTF16ToUnicodeWithOffsets, + _UTF16ToUnicodeWithOffsets, + _UTF16BEFromUnicodeWithOffsets, + _UTF16BEFromUnicodeWithOffsets, + _UTF16GetNextUChar, + + nullptr, /* ### TODO implement getStarters for all Unicode encodings?! */ + _UTF16GetName, + nullptr, + nullptr, + ucnv_getNonSurrogateUnicodeSet, + + nullptr, + nullptr +}; + +static const UConverterStaticData _UTF16v2StaticData = { + sizeof(UConverterStaticData), + "UTF-16,version=2", + 1204, /* CCSID for BOM sensitive UTF-16 */ + UCNV_IBM, UCNV_UTF16, 2, 2, + { 0xff, 0xfd, 0, 0 }, 2, + false, false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + +const UConverterSharedData _UTF16v2Data = + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF16v2StaticData, &_UTF16v2Impl); + +#endif diff --git a/intl/icu/source/common/ucnv_u32.cpp b/intl/icu/source/common/ucnv_u32.cpp new file mode 100644 index 0000000000..d513e8e7e0 --- /dev/null +++ b/intl/icu/source/common/ucnv_u32.cpp @@ -0,0 +1,1253 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2002-2015, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* file name: ucnv_u32.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002jul01 +* created by: Markus W. Scherer +* +* UTF-32 converter implementation. Used to be in ucnv_utf.c. +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/utf.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" +#include "cmemory.h" + +#define MAXIMUM_UCS2 0x0000FFFF +#define MAXIMUM_UTF 0x0010FFFF +#define HALF_SHIFT 10 +#define HALF_BASE 0x0010000 +#define HALF_MASK 0x3FF +#define SURROGATE_HIGH_START 0xD800 +#define SURROGATE_LOW_START 0xDC00 + +/* -SURROGATE_LOW_START + HALF_BASE */ +#define SURROGATE_LOW_BASE 9216 + +enum { + UCNV_NEED_TO_WRITE_BOM=1 +}; + +/* UTF-32BE ----------------------------------------------------------------- */ +U_CDECL_BEGIN +static void U_CALLCONV +T_UConverter_toUnicode_UTF32_BE(UConverterToUnicodeArgs * args, + UErrorCode * err) +{ + const unsigned char *mySource = (unsigned char *) args->source; + char16_t *myTarget = args->target; + const unsigned char *sourceLimit = (unsigned char *) args->sourceLimit; + const char16_t *targetLimit = args->targetLimit; + unsigned char *toUBytes = args->converter->toUBytes; + uint32_t ch, i; + + /* Restore state of current sequence */ + if (args->converter->toULength > 0 && myTarget < targetLimit) { + i = args->converter->toULength; /* restore # of bytes consumed */ + args->converter->toULength = 0; + + ch = args->converter->toUnicodeStatus - 1;/*Stores the previously calculated ch from a previous call*/ + args->converter->toUnicodeStatus = 0; + goto morebytes; + } + + while (mySource < sourceLimit && myTarget < targetLimit) { + i = 0; + ch = 0; +morebytes: + while (i < sizeof(uint32_t)) { + if (mySource < sourceLimit) { + ch = (ch << 8) | (uint8_t)(*mySource); + toUBytes[i++] = (char) *(mySource++); + } + else { + /* stores a partially calculated target*/ + /* + 1 to make 0 a valid character */ + args->converter->toUnicodeStatus = ch + 1; + args->converter->toULength = (int8_t) i; + goto donefornow; + } + } + + if (ch <= MAXIMUM_UTF && !U_IS_SURROGATE(ch)) { + /* Normal valid byte when the loop has not prematurely terminated (i < inBytes) */ + if (ch <= MAXIMUM_UCS2) + { + /* fits in 16 bits */ + *(myTarget++) = (char16_t) ch; + } + else { + /* write out the surrogates */ + *(myTarget++) = U16_LEAD(ch); + ch = U16_TRAIL(ch); + if (myTarget < targetLimit) { + *(myTarget++) = (char16_t)ch; + } + else { + /* Put in overflow buffer (not handled here) */ + args->converter->UCharErrorBuffer[0] = (char16_t) ch; + args->converter->UCharErrorBufferLength = 1; + *err = U_BUFFER_OVERFLOW_ERROR; + break; + } + } + } + else { + args->converter->toULength = (int8_t)i; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + +donefornow: + if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err)) { + /* End of target buffer */ + *err = U_BUFFER_OVERFLOW_ERROR; + } + + args->target = myTarget; + args->source = (const char *) mySource; +} + +static void U_CALLCONV +T_UConverter_toUnicode_UTF32_BE_OFFSET_LOGIC(UConverterToUnicodeArgs * args, + UErrorCode * err) +{ + const unsigned char *mySource = (unsigned char *) args->source; + char16_t *myTarget = args->target; + int32_t *myOffsets = args->offsets; + const unsigned char *sourceLimit = (unsigned char *) args->sourceLimit; + const char16_t *targetLimit = args->targetLimit; + unsigned char *toUBytes = args->converter->toUBytes; + uint32_t ch, i; + int32_t offsetNum = 0; + + /* Restore state of current sequence */ + if (args->converter->toULength > 0 && myTarget < targetLimit) { + i = args->converter->toULength; /* restore # of bytes consumed */ + args->converter->toULength = 0; + + ch = args->converter->toUnicodeStatus - 1;/*Stores the previously calculated ch from a previous call*/ + args->converter->toUnicodeStatus = 0; + goto morebytes; + } + + while (mySource < sourceLimit && myTarget < targetLimit) { + i = 0; + ch = 0; +morebytes: + while (i < sizeof(uint32_t)) { + if (mySource < sourceLimit) { + ch = (ch << 8) | (uint8_t)(*mySource); + toUBytes[i++] = (char) *(mySource++); + } + else { + /* stores a partially calculated target*/ + /* + 1 to make 0 a valid character */ + args->converter->toUnicodeStatus = ch + 1; + args->converter->toULength = (int8_t) i; + goto donefornow; + } + } + + if (ch <= MAXIMUM_UTF && !U_IS_SURROGATE(ch)) { + /* Normal valid byte when the loop has not prematurely terminated (i < inBytes) */ + if (ch <= MAXIMUM_UCS2) { + /* fits in 16 bits */ + *(myTarget++) = (char16_t) ch; + *(myOffsets++) = offsetNum; + } + else { + /* write out the surrogates */ + *(myTarget++) = U16_LEAD(ch); + *myOffsets++ = offsetNum; + ch = U16_TRAIL(ch); + if (myTarget < targetLimit) + { + *(myTarget++) = (char16_t)ch; + *(myOffsets++) = offsetNum; + } + else { + /* Put in overflow buffer (not handled here) */ + args->converter->UCharErrorBuffer[0] = (char16_t) ch; + args->converter->UCharErrorBufferLength = 1; + *err = U_BUFFER_OVERFLOW_ERROR; + break; + } + } + } + else { + args->converter->toULength = (int8_t)i; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + offsetNum += i; + } + +donefornow: + if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err)) + { + /* End of target buffer */ + *err = U_BUFFER_OVERFLOW_ERROR; + } + + args->target = myTarget; + args->source = (const char *) mySource; + args->offsets = myOffsets; +} + +static void U_CALLCONV +T_UConverter_fromUnicode_UTF32_BE(UConverterFromUnicodeArgs * args, + UErrorCode * err) +{ + const char16_t *mySource = args->source; + unsigned char *myTarget; + const char16_t *sourceLimit = args->sourceLimit; + const unsigned char *targetLimit = (unsigned char *) args->targetLimit; + UChar32 ch, ch2; + unsigned int indexToWrite; + unsigned char temp[sizeof(uint32_t)]; + + if(mySource >= sourceLimit) { + /* no input, nothing to do */ + return; + } + + /* write the BOM if necessary */ + if(args->converter->fromUnicodeStatus==UCNV_NEED_TO_WRITE_BOM) { + static const char bom[]={ 0, 0, (char)0xfeu, (char)0xffu }; + ucnv_fromUWriteBytes(args->converter, + bom, 4, + &args->target, args->targetLimit, + &args->offsets, -1, + err); + args->converter->fromUnicodeStatus=0; + } + + myTarget = (unsigned char *) args->target; + temp[0] = 0; + + if (args->converter->fromUChar32) { + ch = args->converter->fromUChar32; + args->converter->fromUChar32 = 0; + goto lowsurogate; + } + + while (mySource < sourceLimit && myTarget < targetLimit) { + ch = *(mySource++); + + if (U_IS_SURROGATE(ch)) { + if (U_IS_LEAD(ch)) { +lowsurogate: + if (mySource < sourceLimit) { + ch2 = *mySource; + if (U_IS_TRAIL(ch2)) { + ch = ((ch - SURROGATE_HIGH_START) << HALF_SHIFT) + ch2 + SURROGATE_LOW_BASE; + mySource++; + } + else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + args->converter->fromUChar32 = ch; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + else { + /* ran out of source */ + args->converter->fromUChar32 = ch; + if (args->flush) { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + *err = U_ILLEGAL_CHAR_FOUND; + } + break; + } + } + else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + args->converter->fromUChar32 = ch; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + + /* We cannot get any larger than 10FFFF because we are coming from UTF-16 */ + temp[1] = (uint8_t) (ch >> 16 & 0x1F); + temp[2] = (uint8_t) (ch >> 8); /* unsigned cast implicitly does (ch & FF) */ + temp[3] = (uint8_t) (ch); /* unsigned cast implicitly does (ch & FF) */ + + for (indexToWrite = 0; indexToWrite <= sizeof(uint32_t) - 1; indexToWrite++) { + if (myTarget < targetLimit) { + *(myTarget++) = temp[indexToWrite]; + } + else { + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = temp[indexToWrite]; + *err = U_BUFFER_OVERFLOW_ERROR; + } + } + } + + if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err)) { + *err = U_BUFFER_OVERFLOW_ERROR; + } + + args->target = (char *) myTarget; + args->source = mySource; +} + +static void U_CALLCONV +T_UConverter_fromUnicode_UTF32_BE_OFFSET_LOGIC(UConverterFromUnicodeArgs * args, + UErrorCode * err) +{ + const char16_t *mySource = args->source; + unsigned char *myTarget; + int32_t *myOffsets; + const char16_t *sourceLimit = args->sourceLimit; + const unsigned char *targetLimit = (unsigned char *) args->targetLimit; + UChar32 ch, ch2; + int32_t offsetNum = 0; + unsigned int indexToWrite; + unsigned char temp[sizeof(uint32_t)]; + + if(mySource >= sourceLimit) { + /* no input, nothing to do */ + return; + } + + /* write the BOM if necessary */ + if(args->converter->fromUnicodeStatus==UCNV_NEED_TO_WRITE_BOM) { + static const char bom[]={ 0, 0, (char)0xfeu, (char)0xffu }; + ucnv_fromUWriteBytes(args->converter, + bom, 4, + &args->target, args->targetLimit, + &args->offsets, -1, + err); + args->converter->fromUnicodeStatus=0; + } + + myTarget = (unsigned char *) args->target; + myOffsets = args->offsets; + temp[0] = 0; + + if (args->converter->fromUChar32) { + ch = args->converter->fromUChar32; + args->converter->fromUChar32 = 0; + goto lowsurogate; + } + + while (mySource < sourceLimit && myTarget < targetLimit) { + ch = *(mySource++); + + if (U_IS_SURROGATE(ch)) { + if (U_IS_LEAD(ch)) { +lowsurogate: + if (mySource < sourceLimit) { + ch2 = *mySource; + if (U_IS_TRAIL(ch2)) { + ch = ((ch - SURROGATE_HIGH_START) << HALF_SHIFT) + ch2 + SURROGATE_LOW_BASE; + mySource++; + } + else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + args->converter->fromUChar32 = ch; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + else { + /* ran out of source */ + args->converter->fromUChar32 = ch; + if (args->flush) { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + *err = U_ILLEGAL_CHAR_FOUND; + } + break; + } + } + else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + args->converter->fromUChar32 = ch; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + + /* We cannot get any larger than 10FFFF because we are coming from UTF-16 */ + temp[1] = (uint8_t) (ch >> 16 & 0x1F); + temp[2] = (uint8_t) (ch >> 8); /* unsigned cast implicitly does (ch & FF) */ + temp[3] = (uint8_t) (ch); /* unsigned cast implicitly does (ch & FF) */ + + for (indexToWrite = 0; indexToWrite <= sizeof(uint32_t) - 1; indexToWrite++) { + if (myTarget < targetLimit) { + *(myTarget++) = temp[indexToWrite]; + *(myOffsets++) = offsetNum; + } + else { + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = temp[indexToWrite]; + *err = U_BUFFER_OVERFLOW_ERROR; + } + } + offsetNum = offsetNum + 1 + (temp[1] != 0); + } + + if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err)) { + *err = U_BUFFER_OVERFLOW_ERROR; + } + + args->target = (char *) myTarget; + args->source = mySource; + args->offsets = myOffsets; +} + +static UChar32 U_CALLCONV +T_UConverter_getNextUChar_UTF32_BE(UConverterToUnicodeArgs* args, + UErrorCode* err) +{ + const uint8_t *mySource; + UChar32 myUChar; + int32_t length; + + mySource = (const uint8_t *)args->source; + if (mySource >= (const uint8_t *)args->sourceLimit) + { + /* no input */ + *err = U_INDEX_OUTOFBOUNDS_ERROR; + return 0xffff; + } + + length = (int32_t)((const uint8_t *)args->sourceLimit - mySource); + if (length < 4) + { + /* got a partial character */ + uprv_memcpy(args->converter->toUBytes, mySource, length); + args->converter->toULength = (int8_t)length; + args->source = (const char *)(mySource + length); + *err = U_TRUNCATED_CHAR_FOUND; + return 0xffff; + } + + /* Don't even try to do a direct cast because the value may be on an odd address. */ + myUChar = ((UChar32)mySource[0] << 24) + | ((UChar32)mySource[1] << 16) + | ((UChar32)mySource[2] << 8) + | ((UChar32)mySource[3]); + + args->source = (const char *)(mySource + 4); + if ((uint32_t)myUChar <= MAXIMUM_UTF && !U_IS_SURROGATE(myUChar)) { + return myUChar; + } + + uprv_memcpy(args->converter->toUBytes, mySource, 4); + args->converter->toULength = 4; + + *err = U_ILLEGAL_CHAR_FOUND; + return 0xffff; +} +U_CDECL_END +static const UConverterImpl _UTF32BEImpl = { + UCNV_UTF32_BigEndian, + + nullptr, + nullptr, + + nullptr, + nullptr, + nullptr, + + T_UConverter_toUnicode_UTF32_BE, + T_UConverter_toUnicode_UTF32_BE_OFFSET_LOGIC, + T_UConverter_fromUnicode_UTF32_BE, + T_UConverter_fromUnicode_UTF32_BE_OFFSET_LOGIC, + T_UConverter_getNextUChar_UTF32_BE, + + nullptr, + nullptr, + nullptr, + nullptr, + ucnv_getNonSurrogateUnicodeSet, + + nullptr, + nullptr +}; + +/* The 1232 CCSID refers to any version of Unicode with any endianness of UTF-32 */ +static const UConverterStaticData _UTF32BEStaticData = { + sizeof(UConverterStaticData), + "UTF-32BE", + 1232, + UCNV_IBM, UCNV_UTF32_BigEndian, 4, 4, + { 0, 0, 0xff, 0xfd }, 4, false, false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + +const UConverterSharedData _UTF32BEData = + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF32BEStaticData, &_UTF32BEImpl); + +/* UTF-32LE ---------------------------------------------------------- */ +U_CDECL_BEGIN +static void U_CALLCONV +T_UConverter_toUnicode_UTF32_LE(UConverterToUnicodeArgs * args, + UErrorCode * err) +{ + const unsigned char *mySource = (unsigned char *) args->source; + char16_t *myTarget = args->target; + const unsigned char *sourceLimit = (unsigned char *) args->sourceLimit; + const char16_t *targetLimit = args->targetLimit; + unsigned char *toUBytes = args->converter->toUBytes; + uint32_t ch, i; + + /* Restore state of current sequence */ + if (args->converter->toULength > 0 && myTarget < targetLimit) + { + i = args->converter->toULength; /* restore # of bytes consumed */ + args->converter->toULength = 0; + + /* Stores the previously calculated ch from a previous call*/ + ch = args->converter->toUnicodeStatus - 1; + args->converter->toUnicodeStatus = 0; + goto morebytes; + } + + while (mySource < sourceLimit && myTarget < targetLimit) + { + i = 0; + ch = 0; +morebytes: + while (i < sizeof(uint32_t)) + { + if (mySource < sourceLimit) + { + ch |= ((uint8_t)(*mySource)) << (i * 8); + toUBytes[i++] = (char) *(mySource++); + } + else + { + /* stores a partially calculated target*/ + /* + 1 to make 0 a valid character */ + args->converter->toUnicodeStatus = ch + 1; + args->converter->toULength = (int8_t) i; + goto donefornow; + } + } + + if (ch <= MAXIMUM_UTF && !U_IS_SURROGATE(ch)) { + /* Normal valid byte when the loop has not prematurely terminated (i < inBytes) */ + if (ch <= MAXIMUM_UCS2) { + /* fits in 16 bits */ + *(myTarget++) = (char16_t) ch; + } + else { + /* write out the surrogates */ + *(myTarget++) = U16_LEAD(ch); + ch = U16_TRAIL(ch); + if (myTarget < targetLimit) { + *(myTarget++) = (char16_t)ch; + } + else { + /* Put in overflow buffer (not handled here) */ + args->converter->UCharErrorBuffer[0] = (char16_t) ch; + args->converter->UCharErrorBufferLength = 1; + *err = U_BUFFER_OVERFLOW_ERROR; + break; + } + } + } + else { + args->converter->toULength = (int8_t)i; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + +donefornow: + if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err)) + { + /* End of target buffer */ + *err = U_BUFFER_OVERFLOW_ERROR; + } + + args->target = myTarget; + args->source = (const char *) mySource; +} + +static void U_CALLCONV +T_UConverter_toUnicode_UTF32_LE_OFFSET_LOGIC(UConverterToUnicodeArgs * args, + UErrorCode * err) +{ + const unsigned char *mySource = (unsigned char *) args->source; + char16_t *myTarget = args->target; + int32_t *myOffsets = args->offsets; + const unsigned char *sourceLimit = (unsigned char *) args->sourceLimit; + const char16_t *targetLimit = args->targetLimit; + unsigned char *toUBytes = args->converter->toUBytes; + uint32_t ch, i; + int32_t offsetNum = 0; + + /* Restore state of current sequence */ + if (args->converter->toULength > 0 && myTarget < targetLimit) + { + i = args->converter->toULength; /* restore # of bytes consumed */ + args->converter->toULength = 0; + + /* Stores the previously calculated ch from a previous call*/ + ch = args->converter->toUnicodeStatus - 1; + args->converter->toUnicodeStatus = 0; + goto morebytes; + } + + while (mySource < sourceLimit && myTarget < targetLimit) + { + i = 0; + ch = 0; +morebytes: + while (i < sizeof(uint32_t)) + { + if (mySource < sourceLimit) + { + ch |= ((uint8_t)(*mySource)) << (i * 8); + toUBytes[i++] = (char) *(mySource++); + } + else + { + /* stores a partially calculated target*/ + /* + 1 to make 0 a valid character */ + args->converter->toUnicodeStatus = ch + 1; + args->converter->toULength = (int8_t) i; + goto donefornow; + } + } + + if (ch <= MAXIMUM_UTF && !U_IS_SURROGATE(ch)) + { + /* Normal valid byte when the loop has not prematurely terminated (i < inBytes) */ + if (ch <= MAXIMUM_UCS2) + { + /* fits in 16 bits */ + *(myTarget++) = (char16_t) ch; + *(myOffsets++) = offsetNum; + } + else { + /* write out the surrogates */ + *(myTarget++) = U16_LEAD(ch); + *(myOffsets++) = offsetNum; + ch = U16_TRAIL(ch); + if (myTarget < targetLimit) + { + *(myTarget++) = (char16_t)ch; + *(myOffsets++) = offsetNum; + } + else + { + /* Put in overflow buffer (not handled here) */ + args->converter->UCharErrorBuffer[0] = (char16_t) ch; + args->converter->UCharErrorBufferLength = 1; + *err = U_BUFFER_OVERFLOW_ERROR; + break; + } + } + } + else + { + args->converter->toULength = (int8_t)i; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + offsetNum += i; + } + +donefornow: + if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err)) + { + /* End of target buffer */ + *err = U_BUFFER_OVERFLOW_ERROR; + } + + args->target = myTarget; + args->source = (const char *) mySource; + args->offsets = myOffsets; +} + +static void U_CALLCONV +T_UConverter_fromUnicode_UTF32_LE(UConverterFromUnicodeArgs * args, + UErrorCode * err) +{ + const char16_t *mySource = args->source; + unsigned char *myTarget; + const char16_t *sourceLimit = args->sourceLimit; + const unsigned char *targetLimit = (unsigned char *) args->targetLimit; + UChar32 ch, ch2; + unsigned int indexToWrite; + unsigned char temp[sizeof(uint32_t)]; + + if(mySource >= sourceLimit) { + /* no input, nothing to do */ + return; + } + + /* write the BOM if necessary */ + if(args->converter->fromUnicodeStatus==UCNV_NEED_TO_WRITE_BOM) { + static const char bom[]={ (char)0xffu, (char)0xfeu, 0, 0 }; + ucnv_fromUWriteBytes(args->converter, + bom, 4, + &args->target, args->targetLimit, + &args->offsets, -1, + err); + args->converter->fromUnicodeStatus=0; + } + + myTarget = (unsigned char *) args->target; + temp[3] = 0; + + if (args->converter->fromUChar32) + { + ch = args->converter->fromUChar32; + args->converter->fromUChar32 = 0; + goto lowsurogate; + } + + while (mySource < sourceLimit && myTarget < targetLimit) + { + ch = *(mySource++); + + if (U16_IS_SURROGATE(ch)) { + if (U16_IS_LEAD(ch)) + { +lowsurogate: + if (mySource < sourceLimit) + { + ch2 = *mySource; + if (U16_IS_TRAIL(ch2)) { + ch = ((ch - SURROGATE_HIGH_START) << HALF_SHIFT) + ch2 + SURROGATE_LOW_BASE; + mySource++; + } + else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + args->converter->fromUChar32 = ch; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + else { + /* ran out of source */ + args->converter->fromUChar32 = ch; + if (args->flush) { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + *err = U_ILLEGAL_CHAR_FOUND; + } + break; + } + } + else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + args->converter->fromUChar32 = ch; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + + /* We cannot get any larger than 10FFFF because we are coming from UTF-16 */ + temp[2] = (uint8_t) (ch >> 16 & 0x1F); + temp[1] = (uint8_t) (ch >> 8); /* unsigned cast implicitly does (ch & FF) */ + temp[0] = (uint8_t) (ch); /* unsigned cast implicitly does (ch & FF) */ + + for (indexToWrite = 0; indexToWrite <= sizeof(uint32_t) - 1; indexToWrite++) + { + if (myTarget < targetLimit) + { + *(myTarget++) = temp[indexToWrite]; + } + else + { + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = temp[indexToWrite]; + *err = U_BUFFER_OVERFLOW_ERROR; + } + } + } + + if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err)) + { + *err = U_BUFFER_OVERFLOW_ERROR; + } + + args->target = (char *) myTarget; + args->source = mySource; +} + +static void U_CALLCONV +T_UConverter_fromUnicode_UTF32_LE_OFFSET_LOGIC(UConverterFromUnicodeArgs * args, + UErrorCode * err) +{ + const char16_t *mySource = args->source; + unsigned char *myTarget; + int32_t *myOffsets; + const char16_t *sourceLimit = args->sourceLimit; + const unsigned char *targetLimit = (unsigned char *) args->targetLimit; + UChar32 ch, ch2; + unsigned int indexToWrite; + unsigned char temp[sizeof(uint32_t)]; + int32_t offsetNum = 0; + + if(mySource >= sourceLimit) { + /* no input, nothing to do */ + return; + } + + /* write the BOM if necessary */ + if(args->converter->fromUnicodeStatus==UCNV_NEED_TO_WRITE_BOM) { + static const char bom[]={ (char)0xffu, (char)0xfeu, 0, 0 }; + ucnv_fromUWriteBytes(args->converter, + bom, 4, + &args->target, args->targetLimit, + &args->offsets, -1, + err); + args->converter->fromUnicodeStatus=0; + } + + myTarget = (unsigned char *) args->target; + myOffsets = args->offsets; + temp[3] = 0; + + if (args->converter->fromUChar32) + { + ch = args->converter->fromUChar32; + args->converter->fromUChar32 = 0; + goto lowsurogate; + } + + while (mySource < sourceLimit && myTarget < targetLimit) + { + ch = *(mySource++); + + if (U16_IS_SURROGATE(ch)) { + if (U16_IS_LEAD(ch)) + { +lowsurogate: + if (mySource < sourceLimit) + { + ch2 = *mySource; + if (U16_IS_TRAIL(ch2)) + { + ch = ((ch - SURROGATE_HIGH_START) << HALF_SHIFT) + ch2 + SURROGATE_LOW_BASE; + mySource++; + } + else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + args->converter->fromUChar32 = ch; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + else { + /* ran out of source */ + args->converter->fromUChar32 = ch; + if (args->flush) { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + *err = U_ILLEGAL_CHAR_FOUND; + } + break; + } + } + else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + args->converter->fromUChar32 = ch; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + + /* We cannot get any larger than 10FFFF because we are coming from UTF-16 */ + temp[2] = (uint8_t) (ch >> 16 & 0x1F); + temp[1] = (uint8_t) (ch >> 8); /* unsigned cast implicitly does (ch & FF) */ + temp[0] = (uint8_t) (ch); /* unsigned cast implicitly does (ch & FF) */ + + for (indexToWrite = 0; indexToWrite <= sizeof(uint32_t) - 1; indexToWrite++) + { + if (myTarget < targetLimit) + { + *(myTarget++) = temp[indexToWrite]; + *(myOffsets++) = offsetNum; + } + else + { + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = temp[indexToWrite]; + *err = U_BUFFER_OVERFLOW_ERROR; + } + } + offsetNum = offsetNum + 1 + (temp[2] != 0); + } + + if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err)) + { + *err = U_BUFFER_OVERFLOW_ERROR; + } + + args->target = (char *) myTarget; + args->source = mySource; + args->offsets = myOffsets; +} + +static UChar32 U_CALLCONV +T_UConverter_getNextUChar_UTF32_LE(UConverterToUnicodeArgs* args, + UErrorCode* err) +{ + const uint8_t *mySource; + UChar32 myUChar; + int32_t length; + + mySource = (const uint8_t *)args->source; + if (mySource >= (const uint8_t *)args->sourceLimit) + { + /* no input */ + *err = U_INDEX_OUTOFBOUNDS_ERROR; + return 0xffff; + } + + length = (int32_t)((const uint8_t *)args->sourceLimit - mySource); + if (length < 4) + { + /* got a partial character */ + uprv_memcpy(args->converter->toUBytes, mySource, length); + args->converter->toULength = (int8_t)length; + args->source = (const char *)(mySource + length); + *err = U_TRUNCATED_CHAR_FOUND; + return 0xffff; + } + + /* Don't even try to do a direct cast because the value may be on an odd address. */ + myUChar = ((UChar32)mySource[3] << 24) + | ((UChar32)mySource[2] << 16) + | ((UChar32)mySource[1] << 8) + | ((UChar32)mySource[0]); + + args->source = (const char *)(mySource + 4); + if ((uint32_t)myUChar <= MAXIMUM_UTF && !U_IS_SURROGATE(myUChar)) { + return myUChar; + } + + uprv_memcpy(args->converter->toUBytes, mySource, 4); + args->converter->toULength = 4; + + *err = U_ILLEGAL_CHAR_FOUND; + return 0xffff; +} +U_CDECL_END +static const UConverterImpl _UTF32LEImpl = { + UCNV_UTF32_LittleEndian, + + nullptr, + nullptr, + + nullptr, + nullptr, + nullptr, + + T_UConverter_toUnicode_UTF32_LE, + T_UConverter_toUnicode_UTF32_LE_OFFSET_LOGIC, + T_UConverter_fromUnicode_UTF32_LE, + T_UConverter_fromUnicode_UTF32_LE_OFFSET_LOGIC, + T_UConverter_getNextUChar_UTF32_LE, + + nullptr, + nullptr, + nullptr, + nullptr, + ucnv_getNonSurrogateUnicodeSet, + + nullptr, + nullptr +}; + +/* The 1232 CCSID refers to any version of Unicode with any endianness of UTF-32 */ +static const UConverterStaticData _UTF32LEStaticData = { + sizeof(UConverterStaticData), + "UTF-32LE", + 1234, + UCNV_IBM, UCNV_UTF32_LittleEndian, 4, 4, + { 0xfd, 0xff, 0, 0 }, 4, false, false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + + +const UConverterSharedData _UTF32LEData = + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF32LEStaticData, &_UTF32LEImpl); + +/* UTF-32 (Detect BOM) ------------------------------------------------------ */ + +/* + * Detect a BOM at the beginning of the stream and select UTF-32BE or UTF-32LE + * accordingly. + * + * State values: + * 0 initial state + * 1 saw 00 + * 2 saw 00 00 + * 3 saw 00 00 FE + * 4 - + * 5 saw FF + * 6 saw FF FE + * 7 saw FF FE 00 + * 8 UTF-32BE mode + * 9 UTF-32LE mode + * + * During detection: state&3==number of matching bytes so far. + * + * On output, emit U+FEFF as the first code point. + */ +U_CDECL_BEGIN +static void U_CALLCONV +_UTF32Reset(UConverter *cnv, UConverterResetChoice choice) { + if(choice<=UCNV_RESET_TO_UNICODE) { + /* reset toUnicode: state=0 */ + cnv->mode=0; + } + if(choice!=UCNV_RESET_TO_UNICODE) { + /* reset fromUnicode: prepare to output the UTF-32PE BOM */ + cnv->fromUnicodeStatus=UCNV_NEED_TO_WRITE_BOM; + } +} + +static void U_CALLCONV +_UTF32Open(UConverter *cnv, + UConverterLoadArgs *pArgs, + UErrorCode *pErrorCode) { + (void)pArgs; + (void)pErrorCode; + _UTF32Reset(cnv, UCNV_RESET_BOTH); +} + +static const char utf32BOM[8]={ 0, 0, (char)0xfeu, (char)0xffu, (char)0xffu, (char)0xfeu, 0, 0 }; + +static void U_CALLCONV +_UTF32ToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv=pArgs->converter; + const char *source=pArgs->source; + const char *sourceLimit=pArgs->sourceLimit; + int32_t *offsets=pArgs->offsets; + + int32_t state, offsetDelta; + char b; + + state=cnv->mode; + + /* + * If we detect a BOM in this buffer, then we must add the BOM size to the + * offsets because the actual converter function will not see and count the BOM. + * offsetDelta will have the number of the BOM bytes that are in the current buffer. + */ + offsetDelta=0; + + while(sourcesource); + } else if(state==8) { + state=9; /* detect UTF-32LE */ + offsetDelta=(int32_t)(source-pArgs->source); + } + } else { + /* switch to UTF-32BE and pass the previous bytes */ + int32_t count=(int32_t)(source-pArgs->source); /* number of bytes from this buffer */ + + /* reset the source */ + source=pArgs->source; + + if(count==(state&3)) { + /* simple: all in the same buffer, just reset source */ + } else { + UBool oldFlush=pArgs->flush; + + /* some of the bytes are from a previous buffer, replay those first */ + pArgs->source=utf32BOM+(state&4); /* select the correct BOM */ + pArgs->sourceLimit=pArgs->source+((state&3)-count); /* replay previous bytes */ + pArgs->flush=false; /* this sourceLimit is not the real source stream limit */ + + /* no offsets: bytes from previous buffer, and not enough for output */ + T_UConverter_toUnicode_UTF32_BE(pArgs, pErrorCode); + + /* restore real pointers; pArgs->source will be set in case 8/9 */ + pArgs->sourceLimit=sourceLimit; + pArgs->flush=oldFlush; + } + state=8; + continue; + } + break; + case 8: + /* call UTF-32BE */ + pArgs->source=source; + if(offsets==nullptr) { + T_UConverter_toUnicode_UTF32_BE(pArgs, pErrorCode); + } else { + T_UConverter_toUnicode_UTF32_BE_OFFSET_LOGIC(pArgs, pErrorCode); + } + source=pArgs->source; + break; + case 9: + /* call UTF-32LE */ + pArgs->source=source; + if(offsets==nullptr) { + T_UConverter_toUnicode_UTF32_LE(pArgs, pErrorCode); + } else { + T_UConverter_toUnicode_UTF32_LE_OFFSET_LOGIC(pArgs, pErrorCode); + } + source=pArgs->source; + break; + default: + break; /* does not occur */ + } + } + + /* add BOM size to offsets - see comment at offsetDelta declaration */ + if(offsets!=nullptr && offsetDelta!=0) { + int32_t *offsetsLimit=pArgs->offsets; + while(offsetssource=source; + + if(source==sourceLimit && pArgs->flush) { + /* handle truncated input */ + switch(state) { + case 0: + break; /* no input at all, nothing to do */ + case 8: + T_UConverter_toUnicode_UTF32_BE(pArgs, pErrorCode); + break; + case 9: + T_UConverter_toUnicode_UTF32_LE(pArgs, pErrorCode); + break; + default: + /* handle 0source=utf32BOM+(state&4); /* select the correct BOM */ + pArgs->sourceLimit=pArgs->source+(state&3); /* replay bytes */ + + /* no offsets: not enough for output */ + T_UConverter_toUnicode_UTF32_BE(pArgs, pErrorCode); + pArgs->source=source; + pArgs->sourceLimit=sourceLimit; + state=8; + break; + } + } + + cnv->mode=state; +} + +static UChar32 U_CALLCONV +_UTF32GetNextUChar(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + switch(pArgs->converter->mode) { + case 8: + return T_UConverter_getNextUChar_UTF32_BE(pArgs, pErrorCode); + case 9: + return T_UConverter_getNextUChar_UTF32_LE(pArgs, pErrorCode); + default: + return UCNV_GET_NEXT_UCHAR_USE_TO_U; + } +} +U_CDECL_END +static const UConverterImpl _UTF32Impl = { + UCNV_UTF32, + + nullptr, + nullptr, + + _UTF32Open, + nullptr, + _UTF32Reset, + + _UTF32ToUnicodeWithOffsets, + _UTF32ToUnicodeWithOffsets, +#if U_IS_BIG_ENDIAN + T_UConverter_fromUnicode_UTF32_BE, + T_UConverter_fromUnicode_UTF32_BE_OFFSET_LOGIC, +#else + T_UConverter_fromUnicode_UTF32_LE, + T_UConverter_fromUnicode_UTF32_LE_OFFSET_LOGIC, +#endif + _UTF32GetNextUChar, + + nullptr, /* ### TODO implement getStarters for all Unicode encodings?! */ + nullptr, + nullptr, + nullptr, + ucnv_getNonSurrogateUnicodeSet, + + nullptr, + nullptr +}; + +/* The 1236 CCSID refers to any version of Unicode with a BOM sensitive endianness of UTF-32 */ +static const UConverterStaticData _UTF32StaticData = { + sizeof(UConverterStaticData), + "UTF-32", + 1236, + UCNV_IBM, UCNV_UTF32, 4, 4, +#if U_IS_BIG_ENDIAN + { 0, 0, 0xff, 0xfd }, 4, +#else + { 0xfd, 0xff, 0, 0 }, 4, +#endif + false, false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + +const UConverterSharedData _UTF32Data = + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF32StaticData, &_UTF32Impl); + +#endif diff --git a/intl/icu/source/common/ucnv_u7.cpp b/intl/icu/source/common/ucnv_u7.cpp new file mode 100644 index 0000000000..398b528e83 --- /dev/null +++ b/intl/icu/source/common/ucnv_u7.cpp @@ -0,0 +1,1491 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2002-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* file name: ucnv_u7.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002jul01 +* created by: Markus W. Scherer +* +* UTF-7 converter implementation. Used to be in ucnv_utf.c. +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION + +#include "cmemory.h" +#include "unicode/ucnv.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" +#include "uassert.h" + +/* UTF-7 -------------------------------------------------------------------- */ + +/* + * UTF-7 is a stateful encoding of Unicode. + * It is defined in RFC 2152. (http://www.ietf.org/rfc/rfc2152.txt) + * It was intended for use in Internet email systems, using in its bytewise + * encoding only a subset of 7-bit US-ASCII. + * UTF-7 is deprecated in favor of UTF-8/16/32 and SCSU, but still + * occasionally used. + * + * For converting Unicode to UTF-7, the RFC allows to encode some US-ASCII + * characters directly or in base64. Especially, the characters in set O + * as defined in the RFC (see below) may be encoded directly but are not + * allowed in, e.g., email headers. + * By default, the ICU UTF-7 converter encodes set O directly. + * By choosing the option "version=1", set O will be escaped instead. + * For example: + * utf7Converter=ucnv_open("UTF-7,version=1"); + * + * For details about email headers see RFC 2047. + */ + +/* + * Tests for US-ASCII characters belonging to character classes + * defined in UTF-7. + * + * Set D (directly encoded characters) consists of the following + * characters: the upper and lower case letters A through Z + * and a through z, the 10 digits 0-9, and the following nine special + * characters (note that "+" and "=" are omitted): + * '(),-./:? + * + * Set O (optional direct characters) consists of the following + * characters (note that "\" and "~" are omitted): + * !"#$%&*;<=>@[]^_`{|} + * + * According to the rules in RFC 2152, the byte values for the following + * US-ASCII characters are not used in UTF-7 and are therefore illegal: + * - all C0 control codes except for CR LF TAB + * - BACKSLASH + * - TILDE + * - DEL + * - all codes beyond US-ASCII, i.e. all >127 + */ +#define inSetD(c) \ + ((uint8_t)((c)-97)<26 || (uint8_t)((c)-65)<26 || /* letters */ \ + (uint8_t)((c)-48)<10 || /* digits */ \ + (uint8_t)((c)-39)<3 || /* '() */ \ + (uint8_t)((c)-44)<4 || /* ,-./ */ \ + (c)==58 || (c)==63 /* :? */ \ + ) + +#define inSetO(c) \ + ((uint8_t)((c)-33)<6 || /* !"#$%& */ \ + (uint8_t)((c)-59)<4 || /* ;<=> */ \ + (uint8_t)((c)-93)<4 || /* ]^_` */ \ + (uint8_t)((c)-123)<3 || /* {|} */ \ + (c)==42 || (c)==64 || (c)==91 /* *@[ */ \ + ) + +#define isCRLFTAB(c) ((c)==13 || (c)==10 || (c)==9) +#define isCRLFSPTAB(c) ((c)==32 || (c)==13 || (c)==10 || (c)==9) + +#define PLUS 43 +#define MINUS 45 +#define BACKSLASH 92 +#define TILDE 126 + +/* legal byte values: all US-ASCII graphic characters from space to before tilde, and CR LF TAB */ +#define isLegalUTF7(c) (((uint8_t)((c)-32)<94 && (c)!=BACKSLASH) || isCRLFTAB(c)) + +/* encode directly sets D and O and CR LF SP TAB */ +static const UBool encodeDirectlyMaximum[128]={ + /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, + + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0 +}; + +/* encode directly set D and CR LF SP TAB but not set O */ +static const UBool encodeDirectlyRestricted[128]={ + /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, + + 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, + + 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0 +}; + +static const uint8_t +toBase64[64]={ + /* A-Z */ + 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, + 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, + /* a-z */ + 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, + 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, + /* 0-9 */ + 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, + /* +/ */ + 43, 47 +}; + +static const int8_t +fromBase64[128]={ + /* C0 controls, -1 for legal ones (CR LF TAB), -3 for illegal ones */ + -3, -3, -3, -3, -3, -3, -3, -3, -3, -1, -1, -3, -3, -1, -3, -3, + -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, + + /* general punctuation with + and / and a special value (-2) for - */ + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 62, -1, -2, -1, 63, + /* digits */ + 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, -1, -1, -1, -1, -1, -1, + + /* A-Z */ + -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, + 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, -1, -3, -1, -1, -1, + + /* a-z */ + -1, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, + 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, -1, -1, -1, -3, -3 +}; + +/* + * converter status values: + * + * toUnicodeStatus: + * 24 inDirectMode (boolean) + * 23..16 base64Counter (-1..7) + * 15..0 bits (up to 14 bits incoming base64) + * + * fromUnicodeStatus: + * 31..28 version (0: set O direct 1: set O escaped) + * 24 inDirectMode (boolean) + * 23..16 base64Counter (0..2) + * 7..0 bits (6 bits outgoing base64) + * + */ + +U_CDECL_BEGIN +static void U_CALLCONV +_UTF7Reset(UConverter *cnv, UConverterResetChoice choice) { + if(choice<=UCNV_RESET_TO_UNICODE) { + /* reset toUnicode */ + cnv->toUnicodeStatus=0x1000000; /* inDirectMode=true */ + cnv->toULength=0; + } + if(choice!=UCNV_RESET_TO_UNICODE) { + /* reset fromUnicode */ + cnv->fromUnicodeStatus=(cnv->fromUnicodeStatus&0xf0000000)|0x1000000; /* keep version, inDirectMode=true */ + } +} + +static void U_CALLCONV +_UTF7Open(UConverter *cnv, + UConverterLoadArgs *pArgs, + UErrorCode *pErrorCode) { + (void)pArgs; + if(UCNV_GET_VERSION(cnv)<=1) { + /* TODO(markus): Should just use cnv->options rather than copying the version number. */ + cnv->fromUnicodeStatus=UCNV_GET_VERSION(cnv)<<28; + _UTF7Reset(cnv, UCNV_RESET_BOTH); + } else { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + } +} + +static void U_CALLCONV +_UTF7ToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const uint8_t *source, *sourceLimit; + char16_t *target; + const char16_t *targetLimit; + int32_t *offsets; + + uint8_t *bytes; + uint8_t byteIndex; + + int32_t length, targetCapacity; + + /* UTF-7 state */ + uint16_t bits; + int8_t base64Counter; + UBool inDirectMode; + + int8_t base64Value; + + int32_t sourceIndex, nextSourceIndex; + + uint8_t b; + /* set up the local pointers */ + cnv=pArgs->converter; + + source=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + target=pArgs->target; + targetLimit=pArgs->targetLimit; + offsets=pArgs->offsets; + /* get the state machine state */ + { + uint32_t status=cnv->toUnicodeStatus; + inDirectMode=(UBool)((status>>24)&1); + base64Counter=(int8_t)(status>>16); + bits=(uint16_t)status; + } + bytes=cnv->toUBytes; + byteIndex=cnv->toULength; + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex=byteIndex==0 ? 0 : -1; + nextSourceIndex=0; + + if(inDirectMode) { +directMode: + /* + * In Direct Mode, most US-ASCII characters are encoded directly, i.e., + * with their US-ASCII byte values. + * Backslash and Tilde and most control characters are not allowed in UTF-7. + * A plus sign starts Unicode (or "escape") Mode. + * + * In Direct Mode, only the sourceIndex is used. + */ + byteIndex=0; + length=(int32_t)(sourceLimit-source); + targetCapacity=(int32_t)(targetLimit-target); + if(length>targetCapacity) { + length=targetCapacity; + } + while(length>0) { + b=*source++; + if(!isLegalUTF7(b)) { + /* illegal */ + bytes[0]=b; + byteIndex=1; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } else if(b!=PLUS) { + /* write directly encoded character */ + *target++=b; + if(offsets!=nullptr) { + *offsets++=sourceIndex++; + } + } else /* PLUS */ { + /* switch to Unicode mode */ + nextSourceIndex=++sourceIndex; + inDirectMode=false; + byteIndex=0; + bits=0; + base64Counter=-1; + goto unicodeMode; + } + --length; + } + if(source=targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + } else { +unicodeMode: + /* + * In Unicode (or "escape") Mode, UTF-16BE is base64-encoded. + * The base64 sequence ends with any character that is not in the base64 alphabet. + * A terminating minus sign is consumed. + * + * In Unicode Mode, the sourceIndex has the index to the start of the current + * base64 bytes, while nextSourceIndex is precisely parallel to source, + * keeping the index to the following byte. + * Note that in 2 out of 3 cases, UChars overlap within a base64 byte. + */ + while(source=126 || (base64Value=fromBase64[b])==-3 || base64Value==-1) { + /* either + * base64Value==-1 for any legal character except base64 and minus sign, or + * base64Value==-3 for illegal characters: + * 1. In either case, leave Unicode mode. + * 2.1. If we ended with an incomplete char16_t or none after the +, then + * generate an error for the preceding erroneous sequence and deal with + * the current (possibly illegal) character next time through. + * 2.2. Else the current char comes after a complete char16_t, which was already + * pushed to the output buf, so: + * 2.2.1. If the current char is legal, just save it for processing next time. + * It may be for example, a plus which we need to deal with in direct mode. + * 2.2.2. Else if the current char is illegal, we might as well deal with it here. + */ + inDirectMode=true; + if(base64Counter==-1) { + /* illegal: + immediately followed by something other than base64 or minus sign */ + /* include the plus sign in the reported sequence, but not the subsequent char */ + --source; + bytes[0]=PLUS; + byteIndex=1; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } else if(bits!=0) { + /* bits are illegally left over, a char16_t is incomplete */ + /* don't include current char (legal or illegal) in error seq */ + --source; + --byteIndex; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } else { + /* previous char16_t was complete */ + if(base64Value==-3) { + /* current character is illegal, deal with it here */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } else { + /* un-read the current character in case it is a plus sign */ + --source; + sourceIndex=nextSourceIndex-1; + goto directMode; + } + } + } else if(base64Value>=0) { + /* collect base64 bytes into UChars */ + switch(base64Counter) { + case -1: /* -1 is immediately after the + */ + case 0: + bits=base64Value; + base64Counter=1; + break; + case 1: + case 3: + case 4: + case 6: + bits=(uint16_t)((bits<<6)|base64Value); + ++base64Counter; + break; + case 2: + *target++=(char16_t)((bits<<4)|(base64Value>>2)); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + sourceIndex=nextSourceIndex-1; + } + bytes[0]=b; /* keep this byte in case an error occurs */ + byteIndex=1; + bits=(uint16_t)(base64Value&3); + base64Counter=3; + break; + case 5: + *target++=(char16_t)((bits<<2)|(base64Value>>4)); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + sourceIndex=nextSourceIndex-1; + } + bytes[0]=b; /* keep this byte in case an error occurs */ + byteIndex=1; + bits=(uint16_t)(base64Value&15); + base64Counter=6; + break; + case 7: + *target++=(char16_t)((bits<<6)|base64Value); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + sourceIndex=nextSourceIndex; + } + byteIndex=0; + bits=0; + base64Counter=0; + break; + default: + /* will never occur */ + break; + } + } else /*base64Value==-2*/ { + /* minus sign terminates the base64 sequence */ + inDirectMode=true; + if(base64Counter==-1) { + /* +- i.e. a minus immediately following a plus */ + *target++=PLUS; + if(offsets!=nullptr) { + *offsets++=sourceIndex-1; + } + } else { + /* absorb the minus and leave the Unicode Mode */ + if(bits!=0) { + /* bits are illegally left over, a char16_t is incomplete */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } + } + sourceIndex=nextSourceIndex; + goto directMode; + } + } else { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + } + + if(U_SUCCESS(*pErrorCode) && pArgs->flush && source==sourceLimit && bits==0) { + /* + * if we are in Unicode mode, then the byteIndex might not be 0, + * but that is ok if bits==0 + * -> we set byteIndex=0 at the end of the stream to avoid a truncated error + * (not true for IMAP-mailbox-name where we must end in direct mode) + */ + byteIndex=0; + } + + /* set the converter state back into UConverter */ + cnv->toUnicodeStatus=((uint32_t)inDirectMode<<24)|((uint32_t)((uint8_t)base64Counter)<<16)|(uint32_t)bits; + cnv->toULength=byteIndex; + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + pArgs->offsets=offsets; + return; +} + +static void U_CALLCONV +_UTF7FromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const char16_t *source, *sourceLimit; + uint8_t *target, *targetLimit; + int32_t *offsets; + + int32_t length, targetCapacity, sourceIndex; + char16_t c; + + /* UTF-7 state */ + const UBool *encodeDirectly; + uint8_t bits; + int8_t base64Counter; + UBool inDirectMode; + + /* set up the local pointers */ + cnv=pArgs->converter; + + /* set up the local pointers */ + source=pArgs->source; + sourceLimit=pArgs->sourceLimit; + target=(uint8_t *)pArgs->target; + targetLimit=(uint8_t *)pArgs->targetLimit; + offsets=pArgs->offsets; + + /* get the state machine state */ + { + uint32_t status=cnv->fromUnicodeStatus; + encodeDirectly= status<0x10000000 ? encodeDirectlyMaximum : encodeDirectlyRestricted; + inDirectMode=(UBool)((status>>24)&1); + base64Counter=(int8_t)(status>>16); + bits=(uint8_t)status; + U_ASSERT(bits<=UPRV_LENGTHOF(toBase64)); + } + + /* UTF-7 always encodes UTF-16 code units, therefore we need only a simple sourceIndex */ + sourceIndex=0; + + if(inDirectMode) { +directMode: + length=(int32_t)(sourceLimit-source); + targetCapacity=(int32_t)(targetLimit-target); + if(length>targetCapacity) { + length=targetCapacity; + } + while(length>0) { + c=*source++; + /* currently always encode CR LF SP TAB directly */ + if(c<=127 && encodeDirectly[c]) { + /* encode directly */ + *target++=(uint8_t)c; + if(offsets!=nullptr) { + *offsets++=sourceIndex++; + } + } else if(c==PLUS) { + /* output +- for + */ + *target++=PLUS; + if(targetcharErrorBuffer[0]=MINUS; + cnv->charErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } else { + /* un-read this character and switch to Unicode Mode */ + --source; + *target++=PLUS; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + inDirectMode=false; + base64Counter=0; + goto unicodeMode; + } + --length; + } + if(source=targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + } else { +unicodeMode: + while(sourcecharErrorBuffer[0]=MINUS; + cnv->charErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + goto directMode; + } else { + /* + * base64 this character: + * Output 2 or 3 base64 bytes for the remaining bits of the previous character + * and the bits of this character, each implicitly in UTF-16BE. + * + * Here, bits is an 8-bit variable because only 6 bits need to be kept from one + * character to the next. The actual 2 or 4 bits are shifted to the left edge + * of the 6-bits field 5..0 to make the termination of the base64 sequence easier. + */ + switch(base64Counter) { + case 0: + *target++=toBase64[c>>10]; + if(target>4)&0x3f]; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + *offsets++=sourceIndex++; + } + } else { + if(offsets!=nullptr) { + *offsets++=sourceIndex++; + } + cnv->charErrorBuffer[0]=toBase64[(c>>4)&0x3f]; + cnv->charErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + bits=(uint8_t)((c&15)<<2); + base64Counter=1; + break; + case 1: + *target++=toBase64[bits|(c>>14)]; + if(target>8)&0x3f]; + if(target>2)&0x3f]; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + *offsets++=sourceIndex; + *offsets++=sourceIndex++; + } + } else { + if(offsets!=nullptr) { + *offsets++=sourceIndex; + *offsets++=sourceIndex++; + } + cnv->charErrorBuffer[0]=toBase64[(c>>2)&0x3f]; + cnv->charErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + } else { + if(offsets!=nullptr) { + *offsets++=sourceIndex++; + } + cnv->charErrorBuffer[0]=toBase64[(c>>8)&0x3f]; + cnv->charErrorBuffer[1]=toBase64[(c>>2)&0x3f]; + cnv->charErrorBufferLength=2; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + bits=(uint8_t)((c&3)<<4); + base64Counter=2; + break; + case 2: + *target++=toBase64[bits|(c>>12)]; + if(target>6)&0x3f]; + if(targetcharErrorBuffer[0]=toBase64[c&0x3f]; + cnv->charErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + } else { + if(offsets!=nullptr) { + *offsets++=sourceIndex++; + } + cnv->charErrorBuffer[0]=toBase64[(c>>6)&0x3f]; + cnv->charErrorBuffer[1]=toBase64[c&0x3f]; + cnv->charErrorBufferLength=2; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + bits=0; + base64Counter=0; + break; + default: + /* will never occur */ + break; + } + } + } else { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + } + + if(pArgs->flush && source>=sourceLimit) { + /* flush remaining bits to the target */ + if(!inDirectMode) { + if (base64Counter!=0) { + if(targetcharErrorBuffer[cnv->charErrorBufferLength++]=toBase64[bits]; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + } + /* Add final MINUS to terminate unicodeMode */ + if(targetcharErrorBuffer[cnv->charErrorBufferLength++]=MINUS; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + } + /* reset the state for the next conversion */ + cnv->fromUnicodeStatus=(cnv->fromUnicodeStatus&0xf0000000)|0x1000000; /* keep version, inDirectMode=true */ + } else { + /* set the converter state back into UConverter */ + cnv->fromUnicodeStatus= + (cnv->fromUnicodeStatus&0xf0000000)| /* keep version*/ + ((uint32_t)inDirectMode<<24)|((uint32_t)base64Counter<<16)|(uint32_t)bits; + } + + /* write back the updated pointers */ + pArgs->source=source; + pArgs->target=(char *)target; + pArgs->offsets=offsets; + return; +} + +static const char * U_CALLCONV +_UTF7GetName(const UConverter *cnv) { + switch(cnv->fromUnicodeStatus>>28) { + case 1: + return "UTF-7,version=1"; + default: + return "UTF-7"; + } +} +U_CDECL_END + +static const UConverterImpl _UTF7Impl={ + UCNV_UTF7, + + nullptr, + nullptr, + + _UTF7Open, + nullptr, + _UTF7Reset, + + _UTF7ToUnicodeWithOffsets, + _UTF7ToUnicodeWithOffsets, + _UTF7FromUnicodeWithOffsets, + _UTF7FromUnicodeWithOffsets, + nullptr, + + nullptr, + _UTF7GetName, + nullptr, /* we don't need writeSub() because we never call a callback at fromUnicode() */ + nullptr, + ucnv_getCompleteUnicodeSet, + + nullptr, + nullptr +}; + +static const UConverterStaticData _UTF7StaticData={ + sizeof(UConverterStaticData), + "UTF-7", + 0, /* TODO CCSID for UTF-7 */ + UCNV_IBM, UCNV_UTF7, + 1, 4, + { 0x3f, 0, 0, 0 }, 1, /* the subchar is not used */ + false, false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + +const UConverterSharedData _UTF7Data= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF7StaticData, &_UTF7Impl); + +/* IMAP mailbox name encoding ----------------------------------------------- */ + +/* + * RFC 2060: INTERNET MESSAGE ACCESS PROTOCOL - VERSION 4rev1 + * http://www.ietf.org/rfc/rfc2060.txt + * + * 5.1.3. Mailbox International Naming Convention + * + * By convention, international mailbox names are specified using a + * modified version of the UTF-7 encoding described in [UTF-7]. The + * purpose of these modifications is to correct the following problems + * with UTF-7: + * + * 1) UTF-7 uses the "+" character for shifting; this conflicts with + * the common use of "+" in mailbox names, in particular USENET + * newsgroup names. + * + * 2) UTF-7's encoding is BASE64 which uses the "/" character; this + * conflicts with the use of "/" as a popular hierarchy delimiter. + * + * 3) UTF-7 prohibits the unencoded usage of "\"; this conflicts with + * the use of "\" as a popular hierarchy delimiter. + * + * 4) UTF-7 prohibits the unencoded usage of "~"; this conflicts with + * the use of "~" in some servers as a home directory indicator. + * + * 5) UTF-7 permits multiple alternate forms to represent the same + * string; in particular, printable US-ASCII characters can be + * represented in encoded form. + * + * In modified UTF-7, printable US-ASCII characters except for "&" + * represent themselves; that is, characters with octet values 0x20-0x25 + * and 0x27-0x7e. The character "&" (0x26) is represented by the two- + * octet sequence "&-". + * + * All other characters (octet values 0x00-0x1f, 0x7f-0xff, and all + * Unicode 16-bit octets) are represented in modified BASE64, with a + * further modification from [UTF-7] that "," is used instead of "/". + * Modified BASE64 MUST NOT be used to represent any printing US-ASCII + * character which can represent itself. + * + * "&" is used to shift to modified BASE64 and "-" to shift back to US- + * ASCII. All names start in US-ASCII, and MUST end in US-ASCII (that + * is, a name that ends with a Unicode 16-bit octet MUST end with a "- + * "). + * + * For example, here is a mailbox name which mixes English, Japanese, + * and Chinese text: ~peter/mail/&ZeVnLIqe-/&U,BTFw- + */ + +/* + * Tests for US-ASCII characters belonging to character classes + * defined in UTF-7. + * + * Set D (directly encoded characters) consists of the following + * characters: the upper and lower case letters A through Z + * and a through z, the 10 digits 0-9, and the following nine special + * characters (note that "+" and "=" are omitted): + * '(),-./:? + * + * Set O (optional direct characters) consists of the following + * characters (note that "\" and "~" are omitted): + * !"#$%&*;<=>@[]^_`{|} + * + * According to the rules in RFC 2152, the byte values for the following + * US-ASCII characters are not used in UTF-7 and are therefore illegal: + * - all C0 control codes except for CR LF TAB + * - BACKSLASH + * - TILDE + * - DEL + * - all codes beyond US-ASCII, i.e. all >127 + */ + +/* uses '&' not '+' to start a base64 sequence */ +#define AMPERSAND 0x26 +#define COMMA 0x2c +#define SLASH 0x2f + +/* legal byte values: all US-ASCII graphic characters 0x20..0x7e */ +#define isLegalIMAP(c) (0x20<=(c) && (c)<=0x7e) + +/* direct-encode all of printable ASCII 0x20..0x7e except '&' 0x26 */ +#define inSetDIMAP(c) (isLegalIMAP(c) && c!=AMPERSAND) + +#define TO_BASE64_IMAP(n) ((n)<63 ? toBase64[n] : COMMA) +#define FROM_BASE64_IMAP(c) ((c)==COMMA ? 63 : (c)==SLASH ? -1 : fromBase64[c]) + +/* + * converter status values: + * + * toUnicodeStatus: + * 24 inDirectMode (boolean) + * 23..16 base64Counter (-1..7) + * 15..0 bits (up to 14 bits incoming base64) + * + * fromUnicodeStatus: + * 24 inDirectMode (boolean) + * 23..16 base64Counter (0..2) + * 7..0 bits (6 bits outgoing base64) + * + * ignore bits 31..25 + */ + +U_CDECL_BEGIN +static void U_CALLCONV +_IMAPToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const uint8_t *source, *sourceLimit; + char16_t *target; + const char16_t *targetLimit; + int32_t *offsets; + + uint8_t *bytes; + uint8_t byteIndex; + + int32_t length, targetCapacity; + + /* UTF-7 state */ + uint16_t bits; + int8_t base64Counter; + UBool inDirectMode; + + int8_t base64Value; + + int32_t sourceIndex, nextSourceIndex; + + char16_t c; + uint8_t b; + + /* set up the local pointers */ + cnv=pArgs->converter; + + source=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + target=pArgs->target; + targetLimit=pArgs->targetLimit; + offsets=pArgs->offsets; + /* get the state machine state */ + { + uint32_t status=cnv->toUnicodeStatus; + inDirectMode=(UBool)((status>>24)&1); + base64Counter=(int8_t)(status>>16); + bits=(uint16_t)status; + } + bytes=cnv->toUBytes; + byteIndex=cnv->toULength; + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex=byteIndex==0 ? 0 : -1; + nextSourceIndex=0; + + if(inDirectMode) { +directMode: + /* + * In Direct Mode, US-ASCII characters are encoded directly, i.e., + * with their US-ASCII byte values. + * An ampersand starts Unicode (or "escape") Mode. + * + * In Direct Mode, only the sourceIndex is used. + */ + byteIndex=0; + length=(int32_t)(sourceLimit-source); + targetCapacity=(int32_t)(targetLimit-target); + if(length>targetCapacity) { + length=targetCapacity; + } + while(length>0) { + b=*source++; + if(!isLegalIMAP(b)) { + /* illegal */ + bytes[0]=b; + byteIndex=1; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } else if(b!=AMPERSAND) { + /* write directly encoded character */ + *target++=b; + if(offsets!=nullptr) { + *offsets++=sourceIndex++; + } + } else /* AMPERSAND */ { + /* switch to Unicode mode */ + nextSourceIndex=++sourceIndex; + inDirectMode=false; + byteIndex=0; + bits=0; + base64Counter=-1; + goto unicodeMode; + } + --length; + } + if(source=targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + } else { +unicodeMode: + /* + * In Unicode (or "escape") Mode, UTF-16BE is base64-encoded. + * The base64 sequence ends with any character that is not in the base64 alphabet. + * A terminating minus sign is consumed. + * US-ASCII must not be base64-ed. + * + * In Unicode Mode, the sourceIndex has the index to the start of the current + * base64 bytes, while nextSourceIndex is precisely parallel to source, + * keeping the index to the following byte. + * Note that in 2 out of 3 cases, UChars overlap within a base64 byte. + */ + while(source0x7e) { + /* illegal - test other illegal US-ASCII values by base64Value==-3 */ + inDirectMode=true; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } else if((base64Value=FROM_BASE64_IMAP(b))>=0) { + /* collect base64 bytes into UChars */ + switch(base64Counter) { + case -1: /* -1 is immediately after the & */ + case 0: + bits=base64Value; + base64Counter=1; + break; + case 1: + case 3: + case 4: + case 6: + bits=(uint16_t)((bits<<6)|base64Value); + ++base64Counter; + break; + case 2: + c=(char16_t)((bits<<4)|(base64Value>>2)); + if(isLegalIMAP(c)) { + /* illegal */ + inDirectMode=true; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + goto endloop; + } + *target++=c; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + sourceIndex=nextSourceIndex-1; + } + bytes[0]=b; /* keep this byte in case an error occurs */ + byteIndex=1; + bits=(uint16_t)(base64Value&3); + base64Counter=3; + break; + case 5: + c=(char16_t)((bits<<2)|(base64Value>>4)); + if(isLegalIMAP(c)) { + /* illegal */ + inDirectMode=true; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + goto endloop; + } + *target++=c; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + sourceIndex=nextSourceIndex-1; + } + bytes[0]=b; /* keep this byte in case an error occurs */ + byteIndex=1; + bits=(uint16_t)(base64Value&15); + base64Counter=6; + break; + case 7: + c=(char16_t)((bits<<6)|base64Value); + if(isLegalIMAP(c)) { + /* illegal */ + inDirectMode=true; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + goto endloop; + } + *target++=c; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + sourceIndex=nextSourceIndex; + } + byteIndex=0; + bits=0; + base64Counter=0; + break; + default: + /* will never occur */ + break; + } + } else if(base64Value==-2) { + /* minus sign terminates the base64 sequence */ + inDirectMode=true; + if(base64Counter==-1) { + /* &- i.e. a minus immediately following an ampersand */ + *target++=AMPERSAND; + if(offsets!=nullptr) { + *offsets++=sourceIndex-1; + } + } else { + /* absorb the minus and leave the Unicode Mode */ + if(bits!=0 || (base64Counter!=0 && base64Counter!=3 && base64Counter!=6)) { + /* bits are illegally left over, a char16_t is incomplete */ + /* base64Counter other than 0, 3, 6 means non-minimal zero-padding, also illegal */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } + } + sourceIndex=nextSourceIndex; + goto directMode; + } else { + if(base64Counter==-1) { + /* illegal: & immediately followed by something other than base64 or minus sign */ + /* include the ampersand in the reported sequence */ + --sourceIndex; + bytes[0]=AMPERSAND; + bytes[1]=b; + byteIndex=2; + } + /* base64Value==-1 for characters that are illegal only in Unicode mode */ + /* base64Value==-3 for illegal characters */ + /* illegal */ + inDirectMode=true; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } + } else { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + } +endloop: + + /* + * the end of the input stream and detection of truncated input + * are handled by the framework, but here we must check if we are in Unicode + * mode and byteIndex==0 because we must end in direct mode + * + * conditions: + * successful + * in Unicode mode and byteIndex==0 + * end of input and no truncated input + */ + if( U_SUCCESS(*pErrorCode) && + !inDirectMode && byteIndex==0 && + pArgs->flush && source>=sourceLimit + ) { + if(base64Counter==-1) { + /* & at the very end of the input */ + /* make the ampersand the reported sequence */ + bytes[0]=AMPERSAND; + byteIndex=1; + } + /* else if(base64Counter!=-1) byteIndex remains 0 because there is no particular byte sequence */ + + inDirectMode=true; /* avoid looping */ + *pErrorCode=U_TRUNCATED_CHAR_FOUND; + } + + /* set the converter state back into UConverter */ + cnv->toUnicodeStatus=((uint32_t)inDirectMode<<24)|((uint32_t)((uint8_t)base64Counter)<<16)|(uint32_t)bits; + cnv->toULength=byteIndex; + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + pArgs->offsets=offsets; + return; +} + +static void U_CALLCONV +_IMAPFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const char16_t *source, *sourceLimit; + uint8_t *target, *targetLimit; + int32_t *offsets; + + int32_t length, targetCapacity, sourceIndex; + char16_t c; + uint8_t b; + + /* UTF-7 state */ + uint8_t bits; + int8_t base64Counter; + UBool inDirectMode; + + /* set up the local pointers */ + cnv=pArgs->converter; + + /* set up the local pointers */ + source=pArgs->source; + sourceLimit=pArgs->sourceLimit; + target=(uint8_t *)pArgs->target; + targetLimit=(uint8_t *)pArgs->targetLimit; + offsets=pArgs->offsets; + + /* get the state machine state */ + { + uint32_t status=cnv->fromUnicodeStatus; + inDirectMode=(UBool)((status>>24)&1); + base64Counter=(int8_t)(status>>16); + bits=(uint8_t)status; + } + + /* UTF-7 always encodes UTF-16 code units, therefore we need only a simple sourceIndex */ + sourceIndex=0; + + if(inDirectMode) { +directMode: + length=(int32_t)(sourceLimit-source); + targetCapacity=(int32_t)(targetLimit-target); + if(length>targetCapacity) { + length=targetCapacity; + } + while(length>0) { + c=*source++; + /* encode 0x20..0x7e except '&' directly */ + if(inSetDIMAP(c)) { + /* encode directly */ + *target++=(uint8_t)c; + if(offsets!=nullptr) { + *offsets++=sourceIndex++; + } + } else if(c==AMPERSAND) { + /* output &- for & */ + *target++=AMPERSAND; + if(targetcharErrorBuffer[0]=MINUS; + cnv->charErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } else { + /* un-read this character and switch to Unicode Mode */ + --source; + *target++=AMPERSAND; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + inDirectMode=false; + base64Counter=0; + goto unicodeMode; + } + --length; + } + if(source=targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + } else { +unicodeMode: + while(sourcecharErrorBuffer[0]=MINUS; + cnv->charErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + goto directMode; + } else { + /* + * base64 this character: + * Output 2 or 3 base64 bytes for the remaining bits of the previous character + * and the bits of this character, each implicitly in UTF-16BE. + * + * Here, bits is an 8-bit variable because only 6 bits need to be kept from one + * character to the next. The actual 2 or 4 bits are shifted to the left edge + * of the 6-bits field 5..0 to make the termination of the base64 sequence easier. + */ + switch(base64Counter) { + case 0: + b=(uint8_t)(c>>10); + *target++=TO_BASE64_IMAP(b); + if(target>4)&0x3f); + *target++=TO_BASE64_IMAP(b); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + *offsets++=sourceIndex++; + } + } else { + if(offsets!=nullptr) { + *offsets++=sourceIndex++; + } + b=(uint8_t)((c>>4)&0x3f); + cnv->charErrorBuffer[0]=TO_BASE64_IMAP(b); + cnv->charErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + bits=(uint8_t)((c&15)<<2); + base64Counter=1; + break; + case 1: + b=(uint8_t)(bits|(c>>14)); + *target++=TO_BASE64_IMAP(b); + if(target>8)&0x3f); + *target++=TO_BASE64_IMAP(b); + if(target>2)&0x3f); + *target++=TO_BASE64_IMAP(b); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + *offsets++=sourceIndex; + *offsets++=sourceIndex++; + } + } else { + if(offsets!=nullptr) { + *offsets++=sourceIndex; + *offsets++=sourceIndex++; + } + b=(uint8_t)((c>>2)&0x3f); + cnv->charErrorBuffer[0]=TO_BASE64_IMAP(b); + cnv->charErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + } else { + if(offsets!=nullptr) { + *offsets++=sourceIndex++; + } + b=(uint8_t)((c>>8)&0x3f); + cnv->charErrorBuffer[0]=TO_BASE64_IMAP(b); + b=(uint8_t)((c>>2)&0x3f); + cnv->charErrorBuffer[1]=TO_BASE64_IMAP(b); + cnv->charErrorBufferLength=2; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + bits=(uint8_t)((c&3)<<4); + base64Counter=2; + break; + case 2: + b=(uint8_t)(bits|(c>>12)); + *target++=TO_BASE64_IMAP(b); + if(target>6)&0x3f); + *target++=TO_BASE64_IMAP(b); + if(targetcharErrorBuffer[0]=TO_BASE64_IMAP(b); + cnv->charErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + } else { + if(offsets!=nullptr) { + *offsets++=sourceIndex++; + } + b=(uint8_t)((c>>6)&0x3f); + cnv->charErrorBuffer[0]=TO_BASE64_IMAP(b); + b=(uint8_t)(c&0x3f); + cnv->charErrorBuffer[1]=TO_BASE64_IMAP(b); + cnv->charErrorBufferLength=2; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + bits=0; + base64Counter=0; + break; + default: + /* will never occur */ + break; + } + } + } else { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + } + + if(pArgs->flush && source>=sourceLimit) { + /* flush remaining bits to the target */ + if(!inDirectMode) { + if(base64Counter!=0) { + if(targetcharErrorBuffer[cnv->charErrorBufferLength++]=TO_BASE64_IMAP(bits); + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + } + /* need to terminate with a minus */ + if(targetcharErrorBuffer[cnv->charErrorBufferLength++]=MINUS; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + } + /* reset the state for the next conversion */ + cnv->fromUnicodeStatus=(cnv->fromUnicodeStatus&0xf0000000)|0x1000000; /* keep version, inDirectMode=true */ + } else { + /* set the converter state back into UConverter */ + cnv->fromUnicodeStatus= + (cnv->fromUnicodeStatus&0xf0000000)| /* keep version*/ + ((uint32_t)inDirectMode<<24)|((uint32_t)base64Counter<<16)|(uint32_t)bits; + } + + /* write back the updated pointers */ + pArgs->source=source; + pArgs->target=(char *)target; + pArgs->offsets=offsets; + return; +} +U_CDECL_END + +static const UConverterImpl _IMAPImpl={ + UCNV_IMAP_MAILBOX, + + nullptr, + nullptr, + + _UTF7Open, + nullptr, + _UTF7Reset, + + _IMAPToUnicodeWithOffsets, + _IMAPToUnicodeWithOffsets, + _IMAPFromUnicodeWithOffsets, + _IMAPFromUnicodeWithOffsets, + nullptr, + + nullptr, + nullptr, + nullptr, /* we don't need writeSub() because we never call a callback at fromUnicode() */ + nullptr, + ucnv_getCompleteUnicodeSet, + nullptr, + nullptr +}; + +static const UConverterStaticData _IMAPStaticData={ + sizeof(UConverterStaticData), + "IMAP-mailbox-name", + 0, /* TODO CCSID for IMAP-mailbox-name */ + UCNV_IBM, UCNV_IMAP_MAILBOX, + 1, 4, + { 0x3f, 0, 0, 0 }, 1, /* the subchar is not used */ + false, false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + +const UConverterSharedData _IMAPData= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_IMAPStaticData, &_IMAPImpl); + +#endif diff --git a/intl/icu/source/common/ucnv_u8.cpp b/intl/icu/source/common/ucnv_u8.cpp new file mode 100644 index 0000000000..cf3bb22a02 --- /dev/null +++ b/intl/icu/source/common/ucnv_u8.cpp @@ -0,0 +1,944 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2002-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* file name: ucnv_u8.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002jul01 +* created by: Markus W. Scherer +* +* UTF-8 converter implementation. Used to be in ucnv_utf.c. +* +* Also, CESU-8 implementation, see UTR 26. +* The CESU-8 converter uses all the same functions as the +* UTF-8 converter, with a branch for converting supplementary code points. +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/utf.h" +#include "unicode/utf8.h" +#include "unicode/utf16.h" +#include "uassert.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" +#include "cmemory.h" +#include "ustr_imp.h" + +/* Prototypes --------------------------------------------------------------- */ + +/* Keep these here to make finicky compilers happy */ + +U_CFUNC void ucnv_fromUnicode_UTF8(UConverterFromUnicodeArgs *args, + UErrorCode *err); +U_CFUNC void ucnv_fromUnicode_UTF8_OFFSETS_LOGIC(UConverterFromUnicodeArgs *args, + UErrorCode *err); + + +/* UTF-8 -------------------------------------------------------------------- */ + +#define MAXIMUM_UCS2 0x0000FFFF + +static const uint32_t offsetsFromUTF8[5] = {0, + (uint32_t) 0x00000000, (uint32_t) 0x00003080, (uint32_t) 0x000E2080, + (uint32_t) 0x03C82080 +}; + +static UBool hasCESU8Data(const UConverter *cnv) +{ +#if UCONFIG_ONLY_HTML_CONVERSION + return false; +#else + return (UBool)(cnv->sharedData == &_CESU8Data); +#endif +} +U_CDECL_BEGIN +static void U_CALLCONV ucnv_toUnicode_UTF8 (UConverterToUnicodeArgs * args, + UErrorCode * err) +{ + UConverter *cnv = args->converter; + const unsigned char *mySource = (unsigned char *) args->source; + char16_t *myTarget = args->target; + const unsigned char *sourceLimit = (unsigned char *) args->sourceLimit; + const char16_t *targetLimit = args->targetLimit; + unsigned char *toUBytes = cnv->toUBytes; + UBool isCESU8 = hasCESU8Data(cnv); + uint32_t ch, ch2 = 0; + int32_t i, inBytes; + + /* Restore size of current sequence */ + if (cnv->toULength > 0 && myTarget < targetLimit) + { + inBytes = cnv->mode; /* restore # of bytes to consume */ + i = cnv->toULength; /* restore # of bytes consumed */ + cnv->toULength = 0; + + ch = cnv->toUnicodeStatus;/*Stores the previously calculated ch from a previous call*/ + cnv->toUnicodeStatus = 0; + goto morebytes; + } + + + while (mySource < sourceLimit && myTarget < targetLimit) + { + ch = *(mySource++); + if (U8_IS_SINGLE(ch)) /* Simple case */ + { + *(myTarget++) = (char16_t) ch; + } + else + { + /* store the first char */ + toUBytes[0] = (char)ch; + inBytes = U8_COUNT_BYTES_NON_ASCII(ch); /* lookup current sequence length */ + i = 1; + +morebytes: + while (i < inBytes) + { + if (mySource < sourceLimit) + { + toUBytes[i] = (char) (ch2 = *mySource); + if (!icu::UTF8::isValidTrail(ch, static_cast(ch2), i, inBytes) && + !(isCESU8 && i == 1 && ch == 0xed && U8_IS_TRAIL(ch2))) + { + break; /* i < inBytes */ + } + ch = (ch << 6) + ch2; + ++mySource; + i++; + } + else + { + /* stores a partially calculated target*/ + cnv->toUnicodeStatus = ch; + cnv->mode = inBytes; + cnv->toULength = (int8_t) i; + goto donefornow; + } + } + + // In CESU-8, only surrogates, not supplementary code points, are encoded directly. + if (i == inBytes && (!isCESU8 || i <= 3)) + { + /* Remove the accumulated high bits */ + ch -= offsetsFromUTF8[inBytes]; + + /* Normal valid byte when the loop has not prematurely terminated (i < inBytes) */ + if (ch <= MAXIMUM_UCS2) + { + /* fits in 16 bits */ + *(myTarget++) = (char16_t) ch; + } + else + { + /* write out the surrogates */ + *(myTarget++) = U16_LEAD(ch); + ch = U16_TRAIL(ch); + if (myTarget < targetLimit) + { + *(myTarget++) = (char16_t)ch; + } + else + { + /* Put in overflow buffer (not handled here) */ + cnv->UCharErrorBuffer[0] = (char16_t) ch; + cnv->UCharErrorBufferLength = 1; + *err = U_BUFFER_OVERFLOW_ERROR; + break; + } + } + } + else + { + cnv->toULength = (int8_t)i; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + } + +donefornow: + if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err)) + { + /* End of target buffer */ + *err = U_BUFFER_OVERFLOW_ERROR; + } + + args->target = myTarget; + args->source = (const char *) mySource; +} + +static void U_CALLCONV ucnv_toUnicode_UTF8_OFFSETS_LOGIC (UConverterToUnicodeArgs * args, + UErrorCode * err) +{ + UConverter *cnv = args->converter; + const unsigned char *mySource = (unsigned char *) args->source; + char16_t *myTarget = args->target; + int32_t *myOffsets = args->offsets; + int32_t offsetNum = 0; + const unsigned char *sourceLimit = (unsigned char *) args->sourceLimit; + const char16_t *targetLimit = args->targetLimit; + unsigned char *toUBytes = cnv->toUBytes; + UBool isCESU8 = hasCESU8Data(cnv); + uint32_t ch, ch2 = 0; + int32_t i, inBytes; + + /* Restore size of current sequence */ + if (cnv->toULength > 0 && myTarget < targetLimit) + { + inBytes = cnv->mode; /* restore # of bytes to consume */ + i = cnv->toULength; /* restore # of bytes consumed */ + cnv->toULength = 0; + + ch = cnv->toUnicodeStatus;/*Stores the previously calculated ch from a previous call*/ + cnv->toUnicodeStatus = 0; + goto morebytes; + } + + while (mySource < sourceLimit && myTarget < targetLimit) + { + ch = *(mySource++); + if (U8_IS_SINGLE(ch)) /* Simple case */ + { + *(myTarget++) = (char16_t) ch; + *(myOffsets++) = offsetNum++; + } + else + { + toUBytes[0] = (char)ch; + inBytes = U8_COUNT_BYTES_NON_ASCII(ch); + i = 1; + +morebytes: + while (i < inBytes) + { + if (mySource < sourceLimit) + { + toUBytes[i] = (char) (ch2 = *mySource); + if (!icu::UTF8::isValidTrail(ch, static_cast(ch2), i, inBytes) && + !(isCESU8 && i == 1 && ch == 0xed && U8_IS_TRAIL(ch2))) + { + break; /* i < inBytes */ + } + ch = (ch << 6) + ch2; + ++mySource; + i++; + } + else + { + cnv->toUnicodeStatus = ch; + cnv->mode = inBytes; + cnv->toULength = (int8_t)i; + goto donefornow; + } + } + + // In CESU-8, only surrogates, not supplementary code points, are encoded directly. + if (i == inBytes && (!isCESU8 || i <= 3)) + { + /* Remove the accumulated high bits */ + ch -= offsetsFromUTF8[inBytes]; + + /* Normal valid byte when the loop has not prematurely terminated (i < inBytes) */ + if (ch <= MAXIMUM_UCS2) + { + /* fits in 16 bits */ + *(myTarget++) = (char16_t) ch; + *(myOffsets++) = offsetNum; + } + else + { + /* write out the surrogates */ + *(myTarget++) = U16_LEAD(ch); + *(myOffsets++) = offsetNum; + ch = U16_TRAIL(ch); + if (myTarget < targetLimit) + { + *(myTarget++) = (char16_t)ch; + *(myOffsets++) = offsetNum; + } + else + { + cnv->UCharErrorBuffer[0] = (char16_t) ch; + cnv->UCharErrorBufferLength = 1; + *err = U_BUFFER_OVERFLOW_ERROR; + } + } + offsetNum += i; + } + else + { + cnv->toULength = (int8_t)i; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + } + +donefornow: + if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err)) + { /* End of target buffer */ + *err = U_BUFFER_OVERFLOW_ERROR; + } + + args->target = myTarget; + args->source = (const char *) mySource; + args->offsets = myOffsets; +} +U_CDECL_END + +U_CFUNC void U_CALLCONV ucnv_fromUnicode_UTF8 (UConverterFromUnicodeArgs * args, + UErrorCode * err) +{ + UConverter *cnv = args->converter; + const char16_t *mySource = args->source; + const char16_t *sourceLimit = args->sourceLimit; + uint8_t *myTarget = (uint8_t *) args->target; + const uint8_t *targetLimit = (uint8_t *) args->targetLimit; + uint8_t *tempPtr; + UChar32 ch; + uint8_t tempBuf[4]; + int32_t indexToWrite; + UBool isNotCESU8 = !hasCESU8Data(cnv); + + if (cnv->fromUChar32 && myTarget < targetLimit) + { + ch = cnv->fromUChar32; + cnv->fromUChar32 = 0; + goto lowsurrogate; + } + + while (mySource < sourceLimit && myTarget < targetLimit) + { + ch = *(mySource++); + + if (ch < 0x80) /* Single byte */ + { + *(myTarget++) = (uint8_t) ch; + } + else if (ch < 0x800) /* Double byte */ + { + *(myTarget++) = (uint8_t) ((ch >> 6) | 0xc0); + if (myTarget < targetLimit) + { + *(myTarget++) = (uint8_t) ((ch & 0x3f) | 0x80); + } + else + { + cnv->charErrorBuffer[0] = (uint8_t) ((ch & 0x3f) | 0x80); + cnv->charErrorBufferLength = 1; + *err = U_BUFFER_OVERFLOW_ERROR; + } + } + else { + /* Check for surrogates */ + if(U16_IS_SURROGATE(ch) && isNotCESU8) { +lowsurrogate: + if (mySource < sourceLimit) { + /* test both code units */ + if(U16_IS_SURROGATE_LEAD(ch) && U16_IS_TRAIL(*mySource)) { + /* convert and consume this supplementary code point */ + ch=U16_GET_SUPPLEMENTARY(ch, *mySource); + ++mySource; + /* exit this condition tree */ + } + else { + /* this is an unpaired trail or lead code unit */ + /* callback(illegal) */ + cnv->fromUChar32 = ch; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + else { + /* no more input */ + cnv->fromUChar32 = ch; + break; + } + } + + /* Do we write the buffer directly for speed, + or do we have to be careful about target buffer space? */ + tempPtr = (((targetLimit - myTarget) >= 4) ? myTarget : tempBuf); + + if (ch <= MAXIMUM_UCS2) { + indexToWrite = 2; + tempPtr[0] = (uint8_t) ((ch >> 12) | 0xe0); + } + else { + indexToWrite = 3; + tempPtr[0] = (uint8_t) ((ch >> 18) | 0xf0); + tempPtr[1] = (uint8_t) (((ch >> 12) & 0x3f) | 0x80); + } + tempPtr[indexToWrite-1] = (uint8_t) (((ch >> 6) & 0x3f) | 0x80); + tempPtr[indexToWrite] = (uint8_t) ((ch & 0x3f) | 0x80); + + if (tempPtr == myTarget) { + /* There was enough space to write the codepoint directly. */ + myTarget += (indexToWrite + 1); + } + else { + /* We might run out of room soon. Write it slowly. */ + for (; tempPtr <= (tempBuf + indexToWrite); tempPtr++) { + if (myTarget < targetLimit) { + *(myTarget++) = *tempPtr; + } + else { + cnv->charErrorBuffer[cnv->charErrorBufferLength++] = *tempPtr; + *err = U_BUFFER_OVERFLOW_ERROR; + } + } + } + } + } + + if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err)) + { + *err = U_BUFFER_OVERFLOW_ERROR; + } + + args->target = (char *) myTarget; + args->source = mySource; +} + +U_CFUNC void U_CALLCONV ucnv_fromUnicode_UTF8_OFFSETS_LOGIC (UConverterFromUnicodeArgs * args, + UErrorCode * err) +{ + UConverter *cnv = args->converter; + const char16_t *mySource = args->source; + int32_t *myOffsets = args->offsets; + const char16_t *sourceLimit = args->sourceLimit; + uint8_t *myTarget = (uint8_t *) args->target; + const uint8_t *targetLimit = (uint8_t *) args->targetLimit; + uint8_t *tempPtr; + UChar32 ch; + int32_t offsetNum, nextSourceIndex; + int32_t indexToWrite; + uint8_t tempBuf[4]; + UBool isNotCESU8 = !hasCESU8Data(cnv); + + if (cnv->fromUChar32 && myTarget < targetLimit) + { + ch = cnv->fromUChar32; + cnv->fromUChar32 = 0; + offsetNum = -1; + nextSourceIndex = 0; + goto lowsurrogate; + } else { + offsetNum = 0; + } + + while (mySource < sourceLimit && myTarget < targetLimit) + { + ch = *(mySource++); + + if (ch < 0x80) /* Single byte */ + { + *(myOffsets++) = offsetNum++; + *(myTarget++) = (char) ch; + } + else if (ch < 0x800) /* Double byte */ + { + *(myOffsets++) = offsetNum; + *(myTarget++) = (uint8_t) ((ch >> 6) | 0xc0); + if (myTarget < targetLimit) + { + *(myOffsets++) = offsetNum++; + *(myTarget++) = (uint8_t) ((ch & 0x3f) | 0x80); + } + else + { + cnv->charErrorBuffer[0] = (uint8_t) ((ch & 0x3f) | 0x80); + cnv->charErrorBufferLength = 1; + *err = U_BUFFER_OVERFLOW_ERROR; + } + } + else + /* Check for surrogates */ + { + nextSourceIndex = offsetNum + 1; + + if(U16_IS_SURROGATE(ch) && isNotCESU8) { +lowsurrogate: + if (mySource < sourceLimit) { + /* test both code units */ + if(U16_IS_SURROGATE_LEAD(ch) && U16_IS_TRAIL(*mySource)) { + /* convert and consume this supplementary code point */ + ch=U16_GET_SUPPLEMENTARY(ch, *mySource); + ++mySource; + ++nextSourceIndex; + /* exit this condition tree */ + } + else { + /* this is an unpaired trail or lead code unit */ + /* callback(illegal) */ + cnv->fromUChar32 = ch; + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + else { + /* no more input */ + cnv->fromUChar32 = ch; + break; + } + } + + /* Do we write the buffer directly for speed, + or do we have to be careful about target buffer space? */ + tempPtr = (((targetLimit - myTarget) >= 4) ? myTarget : tempBuf); + + if (ch <= MAXIMUM_UCS2) { + indexToWrite = 2; + tempPtr[0] = (uint8_t) ((ch >> 12) | 0xe0); + } + else { + indexToWrite = 3; + tempPtr[0] = (uint8_t) ((ch >> 18) | 0xf0); + tempPtr[1] = (uint8_t) (((ch >> 12) & 0x3f) | 0x80); + } + tempPtr[indexToWrite-1] = (uint8_t) (((ch >> 6) & 0x3f) | 0x80); + tempPtr[indexToWrite] = (uint8_t) ((ch & 0x3f) | 0x80); + + if (tempPtr == myTarget) { + /* There was enough space to write the codepoint directly. */ + myTarget += (indexToWrite + 1); + myOffsets[0] = offsetNum; + myOffsets[1] = offsetNum; + myOffsets[2] = offsetNum; + if (indexToWrite >= 3) { + myOffsets[3] = offsetNum; + } + myOffsets += (indexToWrite + 1); + } + else { + /* We might run out of room soon. Write it slowly. */ + for (; tempPtr <= (tempBuf + indexToWrite); tempPtr++) { + if (myTarget < targetLimit) + { + *(myOffsets++) = offsetNum; + *(myTarget++) = *tempPtr; + } + else + { + cnv->charErrorBuffer[cnv->charErrorBufferLength++] = *tempPtr; + *err = U_BUFFER_OVERFLOW_ERROR; + } + } + } + offsetNum = nextSourceIndex; + } + } + + if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err)) + { + *err = U_BUFFER_OVERFLOW_ERROR; + } + + args->target = (char *) myTarget; + args->source = mySource; + args->offsets = myOffsets; +} + +U_CDECL_BEGIN +static UChar32 U_CALLCONV ucnv_getNextUChar_UTF8(UConverterToUnicodeArgs *args, + UErrorCode *err) { + UConverter *cnv; + const uint8_t *sourceInitial; + const uint8_t *source; + uint8_t myByte; + UChar32 ch; + int8_t i; + + /* UTF-8 only here, the framework handles CESU-8 to combine surrogate pairs */ + + cnv = args->converter; + sourceInitial = source = (const uint8_t *)args->source; + if (source >= (const uint8_t *)args->sourceLimit) + { + /* no input */ + *err = U_INDEX_OUTOFBOUNDS_ERROR; + return 0xffff; + } + + myByte = (uint8_t)*(source++); + if (U8_IS_SINGLE(myByte)) + { + args->source = (const char *)source; + return (UChar32)myByte; + } + + uint16_t countTrailBytes = U8_COUNT_TRAIL_BYTES(myByte); + if (countTrailBytes == 0) { + cnv->toUBytes[0] = myByte; + cnv->toULength = 1; + *err = U_ILLEGAL_CHAR_FOUND; + args->source = (const char *)source; + return 0xffff; + } + + /*The byte sequence is longer than the buffer area passed*/ + if (((const char *)source + countTrailBytes) > args->sourceLimit) + { + /* check if all of the remaining bytes are trail bytes */ + uint16_t extraBytesToWrite = countTrailBytes + 1; + cnv->toUBytes[0] = myByte; + i = 1; + *err = U_TRUNCATED_CHAR_FOUND; + while(source < (const uint8_t *)args->sourceLimit) { + uint8_t b = *source; + if(icu::UTF8::isValidTrail(myByte, b, i, extraBytesToWrite)) { + cnv->toUBytes[i++] = b; + ++source; + } else { + /* error even before we run out of input */ + *err = U_ILLEGAL_CHAR_FOUND; + break; + } + } + cnv->toULength = i; + args->source = (const char *)source; + return 0xffff; + } + + ch = myByte << 6; + if(countTrailBytes == 2) { + uint8_t t1 = *source, t2; + if(U8_IS_VALID_LEAD3_AND_T1(myByte, t1) && U8_IS_TRAIL(t2 = *++source)) { + args->source = (const char *)(source + 1); + return (((ch + t1) << 6) + t2) - offsetsFromUTF8[3]; + } + } else if(countTrailBytes == 1) { + uint8_t t1 = *source; + if(U8_IS_TRAIL(t1)) { + args->source = (const char *)(source + 1); + return (ch + t1) - offsetsFromUTF8[2]; + } + } else { // countTrailBytes == 3 + uint8_t t1 = *source, t2, t3; + if(U8_IS_VALID_LEAD4_AND_T1(myByte, t1) && U8_IS_TRAIL(t2 = *++source) && + U8_IS_TRAIL(t3 = *++source)) { + args->source = (const char *)(source + 1); + return (((((ch + t1) << 6) + t2) << 6) + t3) - offsetsFromUTF8[4]; + } + } + args->source = (const char *)source; + + for(i = 0; sourceInitial < source; ++i) { + cnv->toUBytes[i] = *sourceInitial++; + } + cnv->toULength = i; + *err = U_ILLEGAL_CHAR_FOUND; + return 0xffff; +} +U_CDECL_END + +/* UTF-8-from-UTF-8 conversion functions ------------------------------------ */ + +U_CDECL_BEGIN +/* "Convert" UTF-8 to UTF-8: Validate and copy. Modified from ucnv_DBCSFromUTF8(). */ +static void U_CALLCONV +ucnv_UTF8FromUTF8(UConverterFromUnicodeArgs *pFromUArgs, + UConverterToUnicodeArgs *pToUArgs, + UErrorCode *pErrorCode) { + UConverter *utf8; + const uint8_t *source, *sourceLimit; + uint8_t *target; + int32_t targetCapacity; + int32_t count; + + int8_t oldToULength, toULength, toULimit; + + UChar32 c; + uint8_t b, t1, t2; + + /* set up the local pointers */ + utf8=pToUArgs->converter; + source=(uint8_t *)pToUArgs->source; + sourceLimit=(uint8_t *)pToUArgs->sourceLimit; + target=(uint8_t *)pFromUArgs->target; + targetCapacity=(int32_t)(pFromUArgs->targetLimit-pFromUArgs->target); + + /* get the converter state from the UTF-8 UConverter */ + if(utf8->toULength > 0) { + toULength=oldToULength=utf8->toULength; + toULimit=(int8_t)utf8->mode; + c=(UChar32)utf8->toUnicodeStatus; + } else { + toULength=oldToULength=toULimit=0; + c = 0; + } + + count=(int32_t)(sourceLimit-source)+oldToULength; + if(counttargetCapacity) { + count=targetCapacity; + } + + // The conversion loop checks count>0 only once per character. + // If the buffer ends with a truncated sequence, + // then we reduce the count to stop before that, + // and collect the remaining bytes after the conversion loop. + + // Do not go back into the bytes that will be read for finishing a partial + // sequence from the previous buffer. + int32_t length=count-toULength; + U8_TRUNCATE_IF_INCOMPLETE(source, 0, length); + count=toULength+length; + } + + if(c!=0) { + utf8->toUnicodeStatus=0; + utf8->toULength=0; + goto moreBytes; + /* See note in ucnv_SBCSFromUTF8() about this goto. */ + } + + /* conversion loop */ + while(count>0) { + b=*source++; + if(U8_IS_SINGLE(b)) { + /* convert ASCII */ + *target++=b; + --count; + continue; + } else { + if(b>=0xe0) { + if( /* handle U+0800..U+FFFF inline */ + b<0xf0 && + U8_IS_VALID_LEAD3_AND_T1(b, t1=source[0]) && + U8_IS_TRAIL(t2=source[1]) + ) { + source+=2; + *target++=b; + *target++=t1; + *target++=t2; + count-=3; + continue; + } + } else { + if( /* handle U+0080..U+07FF inline */ + b>=0xc2 && + U8_IS_TRAIL(t1=*source) + ) { + ++source; + *target++=b; + *target++=t1; + count-=2; + continue; + } + } + + /* handle "complicated" and error cases, and continuing partial characters */ + oldToULength=0; + toULength=1; + toULimit=U8_COUNT_BYTES_NON_ASCII(b); + c=b; +moreBytes: + while(toULengthtoUBytes[oldToULength++]=*source++; + } + utf8->toUnicodeStatus=c; + utf8->toULength=toULength; + utf8->mode=toULimit; + pToUArgs->source=(char *)source; + pFromUArgs->target=(char *)target; + return; + } + } + + if(toULength!=toULimit) { + /* error handling: illegal UTF-8 byte sequence */ + source-=(toULength-oldToULength); + while(oldToULengthtoUBytes[oldToULength++]=*source++; + } + utf8->toULength=toULength; + pToUArgs->source=(char *)source; + pFromUArgs->target=(char *)target; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + return; + } + + /* copy the legal byte sequence to the target */ + { + int8_t i; + + for(i=0; itoUBytes[i]; + } + source-=(toULength-oldToULength); + for(; i=0); + + if(U_SUCCESS(*pErrorCode) && sourcetargetLimit) { + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } else { + b=*source; + toULimit=U8_COUNT_BYTES(b); + if(toULimit>(sourceLimit-source)) { + /* collect a truncated byte sequence */ + toULength=0; + c=b; + for(;;) { + utf8->toUBytes[toULength++]=b; + if(++source==sourceLimit) { + /* partial byte sequence at end of source */ + utf8->toUnicodeStatus=c; + utf8->toULength=toULength; + utf8->mode=toULimit; + break; + } else if(!icu::UTF8::isValidTrail(c, b=*source, toULength, toULimit)) { + utf8->toULength=toULength; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } + c=(c<<6)+b; + } + } else { + /* partial-sequence target overflow: fall back to the pivoting implementation */ + *pErrorCode=U_USING_DEFAULT_WARNING; + } + } + } + + /* write back the updated pointers */ + pToUArgs->source=(char *)source; + pFromUArgs->target=(char *)target; +} + +U_CDECL_END + +/* UTF-8 converter data ----------------------------------------------------- */ + +static const UConverterImpl _UTF8Impl={ + UCNV_UTF8, + + nullptr, + nullptr, + + nullptr, + nullptr, + nullptr, + + ucnv_toUnicode_UTF8, + ucnv_toUnicode_UTF8_OFFSETS_LOGIC, + ucnv_fromUnicode_UTF8, + ucnv_fromUnicode_UTF8_OFFSETS_LOGIC, + ucnv_getNextUChar_UTF8, + + nullptr, + nullptr, + nullptr, + nullptr, + ucnv_getNonSurrogateUnicodeSet, + + ucnv_UTF8FromUTF8, + ucnv_UTF8FromUTF8 +}; + +/* The 1208 CCSID refers to any version of Unicode of UTF-8 */ +static const UConverterStaticData _UTF8StaticData={ + sizeof(UConverterStaticData), + "UTF-8", + 1208, UCNV_IBM, UCNV_UTF8, + 1, 3, /* max 3 bytes per char16_t from UTF-8 (4 bytes from surrogate _pair_) */ + { 0xef, 0xbf, 0xbd, 0 },3,false,false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + + +const UConverterSharedData _UTF8Data= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF8StaticData, &_UTF8Impl); + +/* CESU-8 converter data ---------------------------------------------------- */ + +static const UConverterImpl _CESU8Impl={ + UCNV_CESU8, + + nullptr, + nullptr, + + nullptr, + nullptr, + nullptr, + + ucnv_toUnicode_UTF8, + ucnv_toUnicode_UTF8_OFFSETS_LOGIC, + ucnv_fromUnicode_UTF8, + ucnv_fromUnicode_UTF8_OFFSETS_LOGIC, + nullptr, + + nullptr, + nullptr, + nullptr, + nullptr, + ucnv_getCompleteUnicodeSet, + + nullptr, + nullptr +}; + +static const UConverterStaticData _CESU8StaticData={ + sizeof(UConverterStaticData), + "CESU-8", + 9400, /* CCSID for CESU-8 */ + UCNV_UNKNOWN, UCNV_CESU8, 1, 3, + { 0xef, 0xbf, 0xbd, 0 },3,false,false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + + +const UConverterSharedData _CESU8Data= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_CESU8StaticData, &_CESU8Impl); + +#endif diff --git a/intl/icu/source/common/ucnvbocu.cpp b/intl/icu/source/common/ucnvbocu.cpp new file mode 100644 index 0000000000..007722e474 --- /dev/null +++ b/intl/icu/source/common/ucnvbocu.cpp @@ -0,0 +1,1413 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2002-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: ucnvbocu.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002mar27 +* created by: Markus W. Scherer +* +* This is an implementation of the Binary Ordered Compression for Unicode, +* in its MIME-friendly form as defined in http://www.unicode.org/notes/tn6/ +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/ucnv_cb.h" +#include "unicode/utf16.h" +#include "putilimp.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" +#include "uassert.h" + +/* BOCU-1 constants and macros ---------------------------------------------- */ + +/* + * BOCU-1 encodes the code points of a Unicode string as + * a sequence of byte-encoded differences (slope detection), + * preserving lexical order. + * + * Optimize the difference-taking for runs of Unicode text within + * small scripts: + * + * Most small scripts are allocated within aligned 128-blocks of Unicode + * code points. Lexical order is preserved if the "previous code point" state + * is always moved into the middle of such a block. + * + * Additionally, "prev" is moved from anywhere in the Unihan and Hangul + * areas into the middle of those areas. + * + * C0 control codes and space are encoded with their US-ASCII bytes. + * "prev" is reset for C0 controls but not for space. + */ + +/* initial value for "prev": middle of the ASCII range */ +#define BOCU1_ASCII_PREV 0x40 + +/* bounding byte values for differences */ +#define BOCU1_MIN 0x21 +#define BOCU1_MIDDLE 0x90 +#define BOCU1_MAX_LEAD 0xfe +#define BOCU1_MAX_TRAIL 0xff +#define BOCU1_RESET 0xff + +/* number of lead bytes */ +#define BOCU1_COUNT (BOCU1_MAX_LEAD-BOCU1_MIN+1) + +/* adjust trail byte counts for the use of some C0 control byte values */ +#define BOCU1_TRAIL_CONTROLS_COUNT 20 +#define BOCU1_TRAIL_BYTE_OFFSET (BOCU1_MIN-BOCU1_TRAIL_CONTROLS_COUNT) + +/* number of trail bytes */ +#define BOCU1_TRAIL_COUNT ((BOCU1_MAX_TRAIL-BOCU1_MIN+1)+BOCU1_TRAIL_CONTROLS_COUNT) + +/* + * number of positive and negative single-byte codes + * (counting 0==BOCU1_MIDDLE among the positive ones) + */ +#define BOCU1_SINGLE 64 + +/* number of lead bytes for positive and negative 2/3/4-byte sequences */ +#define BOCU1_LEAD_2 43 +#define BOCU1_LEAD_3 3 +#define BOCU1_LEAD_4 1 + +/* The difference value range for single-byters. */ +#define BOCU1_REACH_POS_1 (BOCU1_SINGLE-1) +#define BOCU1_REACH_NEG_1 (-BOCU1_SINGLE) + +/* The difference value range for double-byters. */ +#define BOCU1_REACH_POS_2 (BOCU1_REACH_POS_1+BOCU1_LEAD_2*BOCU1_TRAIL_COUNT) +#define BOCU1_REACH_NEG_2 (BOCU1_REACH_NEG_1-BOCU1_LEAD_2*BOCU1_TRAIL_COUNT) + +/* The difference value range for 3-byters. */ +#define BOCU1_REACH_POS_3 \ + (BOCU1_REACH_POS_2+BOCU1_LEAD_3*BOCU1_TRAIL_COUNT*BOCU1_TRAIL_COUNT) + +#define BOCU1_REACH_NEG_3 (BOCU1_REACH_NEG_2-BOCU1_LEAD_3*BOCU1_TRAIL_COUNT*BOCU1_TRAIL_COUNT) + +/* The lead byte start values. */ +#define BOCU1_START_POS_2 (BOCU1_MIDDLE+BOCU1_REACH_POS_1+1) +#define BOCU1_START_POS_3 (BOCU1_START_POS_2+BOCU1_LEAD_2) +#define BOCU1_START_POS_4 (BOCU1_START_POS_3+BOCU1_LEAD_3) + /* ==BOCU1_MAX_LEAD */ + +#define BOCU1_START_NEG_2 (BOCU1_MIDDLE+BOCU1_REACH_NEG_1) +#define BOCU1_START_NEG_3 (BOCU1_START_NEG_2-BOCU1_LEAD_2) +#define BOCU1_START_NEG_4 (BOCU1_START_NEG_3-BOCU1_LEAD_3) + /* ==BOCU1_MIN+1 */ + +/* The length of a byte sequence, according to the lead byte (!=BOCU1_RESET). */ +#define BOCU1_LENGTH_FROM_LEAD(lead) \ + ((BOCU1_START_NEG_2<=(lead) && (lead)>24 : 4) + +/* + * 12 commonly used C0 control codes (and space) are only used to encode + * themselves directly, + * which makes BOCU-1 MIME-usable and reasonably safe for + * ASCII-oriented software. + * + * These controls are + * 0 NUL + * + * 7 BEL + * 8 BS + * + * 9 TAB + * a LF + * b VT + * c FF + * d CR + * + * e SO + * f SI + * + * 1a SUB + * 1b ESC + * + * The other 20 C0 controls are also encoded directly (to preserve order) + * but are also used as trail bytes in difference encoding + * (for better compression). + */ +#define BOCU1_TRAIL_TO_BYTE(t) ((t)>=BOCU1_TRAIL_CONTROLS_COUNT ? (t)+BOCU1_TRAIL_BYTE_OFFSET : bocu1TrailToByte[t]) + +/* + * Byte value map for control codes, + * from external byte values 0x00..0x20 + * to trail byte values 0..19 (0..0x13) as used in the difference calculation. + * External byte values that are illegal as trail bytes are mapped to -1. + */ +static const int8_t +bocu1ByteToTrail[BOCU1_MIN]={ +/* 0 1 2 3 4 5 6 7 */ + -1, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, -1, + +/* 8 9 a b c d e f */ + -1, -1, -1, -1, -1, -1, -1, -1, + +/* 10 11 12 13 14 15 16 17 */ + 0x06, 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, + +/* 18 19 1a 1b 1c 1d 1e 1f */ + 0x0e, 0x0f, -1, -1, 0x10, 0x11, 0x12, 0x13, + +/* 20 */ + -1 +}; + +/* + * Byte value map for control codes, + * from trail byte values 0..19 (0..0x13) as used in the difference calculation + * to external byte values 0x00..0x20. + */ +static const int8_t +bocu1TrailToByte[BOCU1_TRAIL_CONTROLS_COUNT]={ +/* 0 1 2 3 4 5 6 7 */ + 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x10, 0x11, + +/* 8 9 a b c d e f */ + 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, 0x18, 0x19, + +/* 10 11 12 13 */ + 0x1c, 0x1d, 0x1e, 0x1f +}; + +/** + * Integer division and modulo with negative numerators + * yields negative modulo results and quotients that are one more than + * what we need here. + * This macro adjust the results so that the modulo-value m is always >=0. + * + * For positive n, the if() condition is always false. + * + * @param n Number to be split into quotient and rest. + * Will be modified to contain the quotient. + * @param d Divisor. + * @param m Output variable for the rest (modulo result). + */ +#define NEGDIVMOD(n, d, m) UPRV_BLOCK_MACRO_BEGIN { \ + (m)=(n)%(d); \ + (n)/=(d); \ + if((m)<0) { \ + --(n); \ + (m)+=(d); \ + } \ +} UPRV_BLOCK_MACRO_END + +/* Faster versions of packDiff() for single-byte-encoded diff values. */ + +/** Is a diff value encodable in a single byte? */ +#define DIFF_IS_SINGLE(diff) (BOCU1_REACH_NEG_1<=(diff) && (diff)<=BOCU1_REACH_POS_1) + +/** Encode a diff value in a single byte. */ +#define PACK_SINGLE_DIFF(diff) (BOCU1_MIDDLE+(diff)) + +/** Is a diff value encodable in two bytes? */ +#define DIFF_IS_DOUBLE(diff) (BOCU1_REACH_NEG_2<=(diff) && (diff)<=BOCU1_REACH_POS_2) + +/* BOCU-1 implementation functions ------------------------------------------ */ + +#define BOCU1_SIMPLE_PREV(c) (((c)&~0x7f)+BOCU1_ASCII_PREV) + +/** + * Compute the next "previous" value for differencing + * from the current code point. + * + * @param c current code point, 0x3040..0xd7a3 (rest handled by macro below) + * @return "previous code point" state value + */ +static inline int32_t +bocu1Prev(int32_t c) { + /* compute new prev */ + if(/* 0x3040<=c && */ c<=0x309f) { + /* Hiragana is not 128-aligned */ + return 0x3070; + } else if(0x4e00<=c && c<=0x9fa5) { + /* CJK Unihan */ + return 0x4e00-BOCU1_REACH_NEG_2; + } else if(0xac00<=c /* && c<=0xd7a3 */) { + /* Korean Hangul */ + return (0xd7a3+0xac00)/2; + } else { + /* mostly small scripts */ + return BOCU1_SIMPLE_PREV(c); + } +} + +/** Fast version of bocu1Prev() for most scripts. */ +#define BOCU1_PREV(c) ((c)<0x3040 || (c)>0xd7a3 ? BOCU1_SIMPLE_PREV(c) : bocu1Prev(c)) + +/* + * The BOCU-1 converter uses the standard setup code in ucnv.c/ucnv_bld.c. + * The UConverter fields are used as follows: + * + * fromUnicodeStatus encoder's prev (0 will be interpreted as BOCU1_ASCII_PREV) + * + * toUnicodeStatus decoder's prev (0 will be interpreted as BOCU1_ASCII_PREV) + * mode decoder's incomplete (diff<<2)|count (ignored when toULength==0) + */ + +/* BOCU-1-from-Unicode conversion functions --------------------------------- */ + +/** + * Encode a difference -0x10ffff..0x10ffff in 1..4 bytes + * and return a packed integer with them. + * + * The encoding favors small absolute differences with short encodings + * to compress runs of same-script characters. + * + * Optimized version with unrolled loops and fewer floating-point operations + * than the standard packDiff(). + * + * @param diff difference value -0x10ffff..0x10ffff + * @return + * 0x010000zz for 1-byte sequence zz + * 0x0200yyzz for 2-byte sequence yy zz + * 0x03xxyyzz for 3-byte sequence xx yy zz + * 0xwwxxyyzz for 4-byte sequence ww xx yy zz (ww>0x03) + */ +static int32_t +packDiff(int32_t diff) { + int32_t result, m; + + U_ASSERT(!DIFF_IS_SINGLE(diff)); /* assume we won't be called where diff==BOCU1_REACH_NEG_1=-64 */ + if(diff>=BOCU1_REACH_NEG_1) { + /* mostly positive differences, and single-byte negative ones */ +#if 0 /* single-byte case handled in macros, see below */ + if(diff<=BOCU1_REACH_POS_1) { + /* single byte */ + return 0x01000000|(BOCU1_MIDDLE+diff); + } else +#endif + if(diff<=BOCU1_REACH_POS_2) { + /* two bytes */ + diff-=BOCU1_REACH_POS_1+1; + result=0x02000000; + + m=diff%BOCU1_TRAIL_COUNT; + diff/=BOCU1_TRAIL_COUNT; + result|=BOCU1_TRAIL_TO_BYTE(m); + + result|=(BOCU1_START_POS_2+diff)<<8; + } else if(diff<=BOCU1_REACH_POS_3) { + /* three bytes */ + diff-=BOCU1_REACH_POS_2+1; + result=0x03000000; + + m=diff%BOCU1_TRAIL_COUNT; + diff/=BOCU1_TRAIL_COUNT; + result|=BOCU1_TRAIL_TO_BYTE(m); + + m=diff%BOCU1_TRAIL_COUNT; + diff/=BOCU1_TRAIL_COUNT; + result|=BOCU1_TRAIL_TO_BYTE(m)<<8; + + result|=(BOCU1_START_POS_3+diff)<<16; + } else { + /* four bytes */ + diff-=BOCU1_REACH_POS_3+1; + + m=diff%BOCU1_TRAIL_COUNT; + diff/=BOCU1_TRAIL_COUNT; + result=BOCU1_TRAIL_TO_BYTE(m); + + m=diff%BOCU1_TRAIL_COUNT; + diff/=BOCU1_TRAIL_COUNT; + result|=BOCU1_TRAIL_TO_BYTE(m)<<8; + + /* + * We know that / and % would deliver quotient 0 and rest=diff. + * Avoid division and modulo for performance. + */ + result|=BOCU1_TRAIL_TO_BYTE(diff)<<16; + + result|=((uint32_t)BOCU1_START_POS_4)<<24; + } + } else { + /* two- to four-byte negative differences */ + if(diff>=BOCU1_REACH_NEG_2) { + /* two bytes */ + diff-=BOCU1_REACH_NEG_1; + result=0x02000000; + + NEGDIVMOD(diff, BOCU1_TRAIL_COUNT, m); + result|=BOCU1_TRAIL_TO_BYTE(m); + + result|=(BOCU1_START_NEG_2+diff)<<8; + } else if(diff>=BOCU1_REACH_NEG_3) { + /* three bytes */ + diff-=BOCU1_REACH_NEG_2; + result=0x03000000; + + NEGDIVMOD(diff, BOCU1_TRAIL_COUNT, m); + result|=BOCU1_TRAIL_TO_BYTE(m); + + NEGDIVMOD(diff, BOCU1_TRAIL_COUNT, m); + result|=BOCU1_TRAIL_TO_BYTE(m)<<8; + + result|=(BOCU1_START_NEG_3+diff)<<16; + } else { + /* four bytes */ + diff-=BOCU1_REACH_NEG_3; + + NEGDIVMOD(diff, BOCU1_TRAIL_COUNT, m); + result=BOCU1_TRAIL_TO_BYTE(m); + + NEGDIVMOD(diff, BOCU1_TRAIL_COUNT, m); + result|=BOCU1_TRAIL_TO_BYTE(m)<<8; + + /* + * We know that NEGDIVMOD would deliver + * quotient -1 and rest=diff+BOCU1_TRAIL_COUNT. + * Avoid division and modulo for performance. + */ + m=diff+BOCU1_TRAIL_COUNT; + result|=BOCU1_TRAIL_TO_BYTE(m)<<16; + + result|=BOCU1_MIN<<24; + } + } + return result; +} + + +static void U_CALLCONV +_Bocu1FromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const char16_t *source, *sourceLimit; + uint8_t *target; + int32_t targetCapacity; + int32_t *offsets; + + int32_t prev, c, diff; + + int32_t sourceIndex, nextSourceIndex; + + /* set up the local pointers */ + cnv=pArgs->converter; + source=pArgs->source; + sourceLimit=pArgs->sourceLimit; + target=(uint8_t *)pArgs->target; + targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); + offsets=pArgs->offsets; + + /* get the converter state from UConverter */ + c=cnv->fromUChar32; + prev=(int32_t)cnv->fromUnicodeStatus; + if(prev==0) { + prev=BOCU1_ASCII_PREV; + } + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex= c==0 ? 0 : -1; + nextSourceIndex=0; + + /* conversion loop */ + if(c!=0 && targetCapacity>0) { + goto getTrail; + } + +fastSingle: + /* fast loop for single-byte differences */ + /* use only one loop counter variable, targetCapacity, not also source */ + diff=(int32_t)(sourceLimit-source); + if(targetCapacity>diff) { + targetCapacity=diff; + } + while(targetCapacity>0 && (c=*source)<0x3000) { + if(c<=0x20) { + if(c!=0x20) { + prev=BOCU1_ASCII_PREV; + } + *target++=(uint8_t)c; + *offsets++=nextSourceIndex++; + ++source; + --targetCapacity; + } else { + diff=c-prev; + if(DIFF_IS_SINGLE(diff)) { + prev=BOCU1_SIMPLE_PREV(c); + *target++=(uint8_t)PACK_SINGLE_DIFF(diff); + *offsets++=nextSourceIndex++; + ++source; + --targetCapacity; + } else { + break; + } + } + } + /* restore real values */ + targetCapacity=(int32_t)((const uint8_t *)pArgs->targetLimit-target); + sourceIndex=nextSourceIndex; /* wrong if offsets==nullptr but does not matter */ + + /* regular loop for all cases */ + while(source0) { + c=*source++; + ++nextSourceIndex; + + if(c<=0x20) { + /* + * ISO C0 control & space: + * Encode directly for MIME compatibility, + * and reset state except for space, to not disrupt compression. + */ + if(c!=0x20) { + prev=BOCU1_ASCII_PREV; + } + *target++=(uint8_t)c; + *offsets++=sourceIndex; + --targetCapacity; + + sourceIndex=nextSourceIndex; + continue; + } + + if(U16_IS_LEAD(c)) { +getTrail: + if(source=0) { + diff-=BOCU1_REACH_POS_1+1; + m=diff%BOCU1_TRAIL_COUNT; + diff/=BOCU1_TRAIL_COUNT; + diff+=BOCU1_START_POS_2; + } else { + diff-=BOCU1_REACH_NEG_1; + NEGDIVMOD(diff, BOCU1_TRAIL_COUNT, m); + diff+=BOCU1_START_NEG_2; + } + *target++=(uint8_t)diff; + *target++=(uint8_t)BOCU1_TRAIL_TO_BYTE(m); + *offsets++=sourceIndex; + *offsets++=sourceIndex; + targetCapacity-=2; + sourceIndex=nextSourceIndex; + } else { + int32_t length; /* will be 2..4 */ + + diff=packDiff(diff); + length=BOCU1_LENGTH_FROM_PACKED(diff); + + /* write the output character bytes from diff and length */ + /* from the first if in the loop we know that targetCapacity>0 */ + if(length<=targetCapacity) { + switch(length) { + /* each branch falls through to the next one */ + case 4: + *target++=(uint8_t)(diff>>24); + *offsets++=sourceIndex; + U_FALLTHROUGH; + case 3: + *target++=(uint8_t)(diff>>16); + *offsets++=sourceIndex; + U_FALLTHROUGH; + case 2: + *target++=(uint8_t)(diff>>8); + *offsets++=sourceIndex; + /* case 1: handled above */ + *target++=(uint8_t)diff; + *offsets++=sourceIndex; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + targetCapacity-=length; + sourceIndex=nextSourceIndex; + } else { + uint8_t *charErrorBuffer; + + /* + * We actually do this backwards here: + * In order to save an intermediate variable, we output + * first to the overflow buffer what does not fit into the + * regular target. + */ + /* we know that 1<=targetCapacitycharErrorBuffer; + switch(length) { + /* each branch falls through to the next one */ + case 3: + *charErrorBuffer++=(uint8_t)(diff>>16); + U_FALLTHROUGH; + case 2: + *charErrorBuffer++=(uint8_t)(diff>>8); + U_FALLTHROUGH; + case 1: + *charErrorBuffer=(uint8_t)diff; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + cnv->charErrorBufferLength=(int8_t)length; + + /* now output what fits into the regular target */ + diff>>=8*length; /* length was reduced by targetCapacity */ + switch(targetCapacity) { + /* each branch falls through to the next one */ + case 3: + *target++=(uint8_t)(diff>>16); + *offsets++=sourceIndex; + U_FALLTHROUGH; + case 2: + *target++=(uint8_t)(diff>>8); + *offsets++=sourceIndex; + U_FALLTHROUGH; + case 1: + *target++=(uint8_t)diff; + *offsets++=sourceIndex; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + + /* target overflow */ + targetCapacity=0; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + } else { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + + /* set the converter state back into UConverter */ + cnv->fromUChar32= c<0 ? -c : 0; + cnv->fromUnicodeStatus=(uint32_t)prev; + + /* write back the updated pointers */ + pArgs->source=source; + pArgs->target=(char *)target; + pArgs->offsets=offsets; +} + +/* + * Identical to _Bocu1FromUnicodeWithOffsets but without offset handling. + * If a change is made in the original function, then either + * change this function the same way or + * re-copy the original function and remove the variables + * offsets, sourceIndex, and nextSourceIndex. + */ +static void U_CALLCONV +_Bocu1FromUnicode(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const char16_t *source, *sourceLimit; + uint8_t *target; + int32_t targetCapacity; + + int32_t prev, c, diff; + + /* set up the local pointers */ + cnv=pArgs->converter; + source=pArgs->source; + sourceLimit=pArgs->sourceLimit; + target=(uint8_t *)pArgs->target; + targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); + + /* get the converter state from UConverter */ + c=cnv->fromUChar32; + prev=(int32_t)cnv->fromUnicodeStatus; + if(prev==0) { + prev=BOCU1_ASCII_PREV; + } + + /* conversion loop */ + if(c!=0 && targetCapacity>0) { + goto getTrail; + } + +fastSingle: + /* fast loop for single-byte differences */ + /* use only one loop counter variable, targetCapacity, not also source */ + diff=(int32_t)(sourceLimit-source); + if(targetCapacity>diff) { + targetCapacity=diff; + } + while(targetCapacity>0 && (c=*source)<0x3000) { + if(c<=0x20) { + if(c!=0x20) { + prev=BOCU1_ASCII_PREV; + } + *target++=(uint8_t)c; + } else { + diff=c-prev; + if(DIFF_IS_SINGLE(diff)) { + prev=BOCU1_SIMPLE_PREV(c); + *target++=(uint8_t)PACK_SINGLE_DIFF(diff); + } else { + break; + } + } + ++source; + --targetCapacity; + } + /* restore real values */ + targetCapacity=(int32_t)((const uint8_t *)pArgs->targetLimit-target); + + /* regular loop for all cases */ + while(source0) { + c=*source++; + + if(c<=0x20) { + /* + * ISO C0 control & space: + * Encode directly for MIME compatibility, + * and reset state except for space, to not disrupt compression. + */ + if(c!=0x20) { + prev=BOCU1_ASCII_PREV; + } + *target++=(uint8_t)c; + --targetCapacity; + continue; + } + + if(U16_IS_LEAD(c)) { +getTrail: + if(source=0) { + diff-=BOCU1_REACH_POS_1+1; + m=diff%BOCU1_TRAIL_COUNT; + diff/=BOCU1_TRAIL_COUNT; + diff+=BOCU1_START_POS_2; + } else { + diff-=BOCU1_REACH_NEG_1; + NEGDIVMOD(diff, BOCU1_TRAIL_COUNT, m); + diff+=BOCU1_START_NEG_2; + } + *target++=(uint8_t)diff; + *target++=(uint8_t)BOCU1_TRAIL_TO_BYTE(m); + targetCapacity-=2; + } else { + int32_t length; /* will be 2..4 */ + + diff=packDiff(diff); + length=BOCU1_LENGTH_FROM_PACKED(diff); + + /* write the output character bytes from diff and length */ + /* from the first if in the loop we know that targetCapacity>0 */ + if(length<=targetCapacity) { + switch(length) { + /* each branch falls through to the next one */ + case 4: + *target++=(uint8_t)(diff>>24); + U_FALLTHROUGH; + case 3: + *target++=(uint8_t)(diff>>16); + /* case 2: handled above */ + *target++=(uint8_t)(diff>>8); + /* case 1: handled above */ + *target++=(uint8_t)diff; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + targetCapacity-=length; + } else { + uint8_t *charErrorBuffer; + + /* + * We actually do this backwards here: + * In order to save an intermediate variable, we output + * first to the overflow buffer what does not fit into the + * regular target. + */ + /* we know that 1<=targetCapacitycharErrorBuffer; + switch(length) { + /* each branch falls through to the next one */ + case 3: + *charErrorBuffer++=(uint8_t)(diff>>16); + U_FALLTHROUGH; + case 2: + *charErrorBuffer++=(uint8_t)(diff>>8); + U_FALLTHROUGH; + case 1: + *charErrorBuffer=(uint8_t)diff; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + cnv->charErrorBufferLength=(int8_t)length; + + /* now output what fits into the regular target */ + diff>>=8*length; /* length was reduced by targetCapacity */ + switch(targetCapacity) { + /* each branch falls through to the next one */ + case 3: + *target++=(uint8_t)(diff>>16); + U_FALLTHROUGH; + case 2: + *target++=(uint8_t)(diff>>8); + U_FALLTHROUGH; + case 1: + *target++=(uint8_t)diff; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + + /* target overflow */ + targetCapacity=0; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + } else { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + + /* set the converter state back into UConverter */ + cnv->fromUChar32= c<0 ? -c : 0; + cnv->fromUnicodeStatus=(uint32_t)prev; + + /* write back the updated pointers */ + pArgs->source=source; + pArgs->target=(char *)target; +} + +/* BOCU-1-to-Unicode conversion functions ----------------------------------- */ + +/** + * Function for BOCU-1 decoder; handles multi-byte lead bytes. + * + * @param b lead byte; + * BOCU1_MIN<=b=BOCU1_START_NEG_2) { + /* positive difference */ + if(b=BOCU1_START_NEG_3) { + /* two bytes */ + diff=((int32_t)b-BOCU1_START_NEG_2)*BOCU1_TRAIL_COUNT+BOCU1_REACH_NEG_1; + count=1; + } else if(b>BOCU1_MIN) { + /* three bytes */ + diff=((int32_t)b-BOCU1_START_NEG_3)*BOCU1_TRAIL_COUNT*BOCU1_TRAIL_COUNT+BOCU1_REACH_NEG_2; + count=2; + } else { + /* four bytes */ + diff=-BOCU1_TRAIL_COUNT*BOCU1_TRAIL_COUNT*BOCU1_TRAIL_COUNT+BOCU1_REACH_NEG_3; + count=3; + } + } + + /* return the state for decoding the trail byte(s) */ + return ((uint32_t)diff<<2)|count; +} + +/** + * Function for BOCU-1 decoder; handles multi-byte trail bytes. + * + * @param count number of remaining trail bytes including this one + * @param b trail byte + * @return new delta for diff including b - <0 indicates an error + * + * @see decodeBocu1 + */ +static inline int32_t +decodeBocu1TrailByte(int32_t count, int32_t b) { + if(b<=0x20) { + /* skip some C0 controls and make the trail byte range contiguous */ + b=bocu1ByteToTrail[b]; + /* b<0 for an illegal trail byte value will result in return<0 below */ +#if BOCU1_MAX_TRAIL<0xff + } else if(b>BOCU1_MAX_TRAIL) { + return -99; +#endif + } else { + b-=BOCU1_TRAIL_BYTE_OFFSET; + } + + /* add trail byte into difference and decrement count */ + if(count==1) { + return b; + } else if(count==2) { + return b*BOCU1_TRAIL_COUNT; + } else /* count==3 */ { + return b*(BOCU1_TRAIL_COUNT*BOCU1_TRAIL_COUNT); + } +} + +static void U_CALLCONV +_Bocu1ToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const uint8_t *source, *sourceLimit; + char16_t *target; + const char16_t *targetLimit; + int32_t *offsets; + + int32_t prev, count, diff, c; + + int8_t byteIndex; + uint8_t *bytes; + + int32_t sourceIndex, nextSourceIndex; + + /* set up the local pointers */ + cnv=pArgs->converter; + source=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + target=pArgs->target; + targetLimit=pArgs->targetLimit; + offsets=pArgs->offsets; + + /* get the converter state from UConverter */ + prev=(int32_t)cnv->toUnicodeStatus; + if(prev==0) { + prev=BOCU1_ASCII_PREV; + } + diff=cnv->mode; /* mode may be set to UCNV_SI by ucnv_bld.c but then toULength==0 */ + count=diff&3; + diff>>=2; + + byteIndex=cnv->toULength; + bytes=cnv->toUBytes; + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex=byteIndex==0 ? 0 : -1; + nextSourceIndex=0; + + /* conversion "loop" similar to _SCSUToUnicodeWithOffsets() */ + if(count>0 && byteIndex>0 && targettargetLimit-target); + if(count>diff) { + count=diff; + } + while(count>0) { + if(BOCU1_START_NEG_2<=(c=*source) && c=targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + + ++nextSourceIndex; + c=*source++; + if(BOCU1_START_NEG_2<=c && c=BOCU1_MIDDLE) { + diff=((int32_t)c-BOCU1_START_POS_2)*BOCU1_TRAIL_COUNT+BOCU1_REACH_POS_1+1; + } else { + diff=((int32_t)c-BOCU1_START_NEG_2)*BOCU1_TRAIL_COUNT+BOCU1_REACH_NEG_1; + } + + /* trail byte */ + ++nextSourceIndex; + c=decodeBocu1TrailByte(1, *source++); + if(c<0 || (uint32_t)(c=prev+diff+c)>0x10ffff) { + bytes[0]=source[-2]; + bytes[1]=source[-1]; + byteIndex=2; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } + } else if(c==BOCU1_RESET) { + /* only reset the state, no code point */ + prev=BOCU1_ASCII_PREV; + sourceIndex=nextSourceIndex; + continue; + } else { + /* + * For multi-byte difference lead bytes, set the decoder state + * with the partial difference value from the lead byte and + * with the number of trail bytes. + */ + bytes[0]=(uint8_t)c; + byteIndex=1; + + diff=decodeBocu1LeadByte(c); + count=diff&3; + diff>>=2; +getTrail: + for(;;) { + if(source>=sourceLimit) { + goto endloop; + } + ++nextSourceIndex; + c=bytes[byteIndex++]=*source++; + + /* trail byte in any position */ + c=decodeBocu1TrailByte(count, c); + if(c<0) { + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + goto endloop; + } + + diff+=c; + if(--count==0) { + /* final trail byte, deliver a code point */ + byteIndex=0; + c=prev+diff; + if((uint32_t)c>0x10ffff) { + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + goto endloop; + } + break; + } + } + } + + /* calculate the next prev and output c */ + prev=BOCU1_PREV(c); + if(c<=0xffff) { + *target++=(char16_t)c; + *offsets++=sourceIndex; + } else { + /* output surrogate pair */ + *target++=U16_LEAD(c); + if(targetUCharErrorBuffer[0]=U16_TRAIL(c); + cnv->UCharErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + sourceIndex=nextSourceIndex; + } +endloop: + + if(*pErrorCode==U_ILLEGAL_CHAR_FOUND) { + /* set the converter state in UConverter to deal with the next character */ + cnv->toUnicodeStatus=BOCU1_ASCII_PREV; + cnv->mode=0; + } else { + /* set the converter state back into UConverter */ + cnv->toUnicodeStatus=(uint32_t)prev; + cnv->mode=(int32_t)((uint32_t)diff<<2)|count; + } + cnv->toULength=byteIndex; + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + pArgs->offsets=offsets; + return; +} + +/* + * Identical to _Bocu1ToUnicodeWithOffsets but without offset handling. + * If a change is made in the original function, then either + * change this function the same way or + * re-copy the original function and remove the variables + * offsets, sourceIndex, and nextSourceIndex. + */ +static void U_CALLCONV +_Bocu1ToUnicode(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const uint8_t *source, *sourceLimit; + char16_t *target; + const char16_t *targetLimit; + + int32_t prev, count, diff, c; + + int8_t byteIndex; + uint8_t *bytes; + + /* set up the local pointers */ + cnv=pArgs->converter; + source=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + target=pArgs->target; + targetLimit=pArgs->targetLimit; + + /* get the converter state from UConverter */ + prev=(int32_t)cnv->toUnicodeStatus; + if(prev==0) { + prev=BOCU1_ASCII_PREV; + } + diff=cnv->mode; /* mode may be set to UCNV_SI by ucnv_bld.c but then toULength==0 */ + count=diff&3; + diff>>=2; + + byteIndex=cnv->toULength; + bytes=cnv->toUBytes; + + /* conversion "loop" similar to _SCSUToUnicodeWithOffsets() */ + if(count>0 && byteIndex>0 && targettargetLimit-target); + if(count>diff) { + count=diff; + } + while(count>0) { + if(BOCU1_START_NEG_2<=(c=*source) && c=targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + + c=*source++; + if(BOCU1_START_NEG_2<=c && c=BOCU1_MIDDLE) { + diff=((int32_t)c-BOCU1_START_POS_2)*BOCU1_TRAIL_COUNT+BOCU1_REACH_POS_1+1; + } else { + diff=((int32_t)c-BOCU1_START_NEG_2)*BOCU1_TRAIL_COUNT+BOCU1_REACH_NEG_1; + } + + /* trail byte */ + c=decodeBocu1TrailByte(1, *source++); + if(c<0 || (uint32_t)(c=prev+diff+c)>0x10ffff) { + bytes[0]=source[-2]; + bytes[1]=source[-1]; + byteIndex=2; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } + } else if(c==BOCU1_RESET) { + /* only reset the state, no code point */ + prev=BOCU1_ASCII_PREV; + continue; + } else { + /* + * For multi-byte difference lead bytes, set the decoder state + * with the partial difference value from the lead byte and + * with the number of trail bytes. + */ + bytes[0]=(uint8_t)c; + byteIndex=1; + + diff=decodeBocu1LeadByte(c); + count=diff&3; + diff>>=2; +getTrail: + for(;;) { + if(source>=sourceLimit) { + goto endloop; + } + c=bytes[byteIndex++]=*source++; + + /* trail byte in any position */ + c=decodeBocu1TrailByte(count, c); + if(c<0) { + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + goto endloop; + } + + diff+=c; + if(--count==0) { + /* final trail byte, deliver a code point */ + byteIndex=0; + c=prev+diff; + if((uint32_t)c>0x10ffff) { + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + goto endloop; + } + break; + } + } + } + + /* calculate the next prev and output c */ + prev=BOCU1_PREV(c); + if(c<=0xffff) { + *target++=(char16_t)c; + } else { + /* output surrogate pair */ + *target++=U16_LEAD(c); + if(targetUCharErrorBuffer[0]=U16_TRAIL(c); + cnv->UCharErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + } +endloop: + + if(*pErrorCode==U_ILLEGAL_CHAR_FOUND) { + /* set the converter state in UConverter to deal with the next character */ + cnv->toUnicodeStatus=BOCU1_ASCII_PREV; + cnv->mode=0; + } else { + /* set the converter state back into UConverter */ + cnv->toUnicodeStatus=(uint32_t)prev; + cnv->mode=((uint32_t)diff<<2)|count; + } + cnv->toULength=byteIndex; + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + return; +} + +/* miscellaneous ------------------------------------------------------------ */ + +static const UConverterImpl _Bocu1Impl={ + UCNV_BOCU1, + + nullptr, + nullptr, + + nullptr, + nullptr, + nullptr, + + _Bocu1ToUnicode, + _Bocu1ToUnicodeWithOffsets, + _Bocu1FromUnicode, + _Bocu1FromUnicodeWithOffsets, + nullptr, + + nullptr, + nullptr, + nullptr, + nullptr, + ucnv_getCompleteUnicodeSet, + + nullptr, + nullptr +}; + +static const UConverterStaticData _Bocu1StaticData={ + sizeof(UConverterStaticData), + "BOCU-1", + 1214, /* CCSID for BOCU-1 */ + UCNV_IBM, UCNV_BOCU1, + 1, 4, /* one char16_t generates at least 1 byte and at most 4 bytes */ + { 0x1a, 0, 0, 0 }, 1, /* BOCU-1 never needs to write a subchar */ + false, false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + +const UConverterSharedData _Bocu1Data= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_Bocu1StaticData, &_Bocu1Impl); + +#endif diff --git a/intl/icu/source/common/ucnvdisp.cpp b/intl/icu/source/common/ucnvdisp.cpp new file mode 100644 index 0000000000..0830042d0b --- /dev/null +++ b/intl/icu/source/common/ucnvdisp.cpp @@ -0,0 +1,88 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1998-2004, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* ucnvdisp.c: +* Implements APIs for the ICU's codeset conversion library display names. +* +* Modification History: +* +* Date Name Description +* 04/04/99 helena Fixed internal header inclusion. +* 05/09/00 helena Added implementation to handle fallback mappings. +* 06/20/2000 helena OS/400 port changes; mostly typecast. +* 09/08/2004 grhoten split from ucnv.c +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/ustring.h" +#include "unicode/ures.h" +#include "unicode/ucnv.h" +#include "cstring.h" +#include "ustr_imp.h" +#include "ucnv_imp.h" +#include "putilimp.h" + +U_CAPI int32_t U_EXPORT2 +ucnv_getDisplayName(const UConverter *cnv, + const char *displayLocale, + char16_t *displayName, int32_t displayNameCapacity, + UErrorCode *pErrorCode) { + UResourceBundle *rb; + const char16_t *name; + int32_t length; + UErrorCode localStatus = U_ZERO_ERROR; + + /* check arguments */ + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + if(cnv==nullptr || displayNameCapacity<0 || (displayNameCapacity>0 && displayName==nullptr)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* open the resource bundle and get the display name string */ + rb=ures_open(nullptr, displayLocale, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + return 0; + } + + /* use the internal name as the key */ + name=ures_getStringByKey(rb, cnv->sharedData->staticData->name, &length, &localStatus); + ures_close(rb); + + if(U_SUCCESS(localStatus)) { + /* copy the string */ + if (*pErrorCode == U_ZERO_ERROR) { + *pErrorCode = localStatus; + } + u_memcpy(displayName, name, uprv_min(length, displayNameCapacity)*U_SIZEOF_UCHAR); + } else { + /* convert the internal name into a Unicode string */ + length=(int32_t)uprv_strlen(cnv->sharedData->staticData->name); + u_charsToUChars(cnv->sharedData->staticData->name, displayName, uprv_min(length, displayNameCapacity)); + } + return u_terminateUChars(displayName, displayNameCapacity, length, pErrorCode); +} + +#endif + +/* + * Hey, Emacs, please set the following: + * + * Local Variables: + * indent-tabs-mode: nil + * End: + * + */ diff --git a/intl/icu/source/common/ucnvhz.cpp b/intl/icu/source/common/ucnvhz.cpp new file mode 100644 index 0000000000..fa0f2b40ea --- /dev/null +++ b/intl/icu/source/common/ucnvhz.cpp @@ -0,0 +1,625 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2000-2015, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* file name: ucnvhz.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2000oct16 +* created by: Ram Viswanadha +* 10/31/2000 Ram Implemented offsets logic function +* +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION && !UCONFIG_NO_LEGACY_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION + +#include "cmemory.h" +#include "unicode/ucnv.h" +#include "unicode/ucnv_cb.h" +#include "unicode/uset.h" +#include "unicode/utf16.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" +#include "ucnv_imp.h" + +#define UCNV_TILDE 0x7E /* ~ */ +#define UCNV_OPEN_BRACE 0x7B /* { */ +#define UCNV_CLOSE_BRACE 0x7D /* } */ +#define SB_ESCAPE "\x7E\x7D" +#define DB_ESCAPE "\x7E\x7B" +#define TILDE_ESCAPE "\x7E\x7E" +#define ESC_LEN 2 + + +#define CONCAT_ESCAPE_MACRO(args, targetIndex,targetLength,strToAppend, err, len,sourceIndex) UPRV_BLOCK_MACRO_BEGIN { \ + while(len-->0){ \ + if(targetIndex < targetLength){ \ + args->target[targetIndex] = (unsigned char) *strToAppend; \ + if(args->offsets!=nullptr){ \ + *(offsets++) = sourceIndex-1; \ + } \ + targetIndex++; \ + } \ + else{ \ + args->converter->charErrorBuffer[(int)args->converter->charErrorBufferLength++] = (unsigned char) *strToAppend; \ + *err =U_BUFFER_OVERFLOW_ERROR; \ + } \ + strToAppend++; \ + } \ +} UPRV_BLOCK_MACRO_END + + +typedef struct{ + UConverter* gbConverter; + int32_t targetIndex; + int32_t sourceIndex; + UBool isEscapeAppended; + UBool isStateDBCS; + UBool isTargetUCharDBCS; + UBool isEmptySegment; +}UConverterDataHZ; + + +U_CDECL_BEGIN +static void U_CALLCONV +_HZOpen(UConverter *cnv, UConverterLoadArgs *pArgs, UErrorCode *errorCode){ + UConverter *gbConverter; + if(pArgs->onlyTestIsLoadable) { + ucnv_canCreateConverter("GBK", errorCode); /* errorCode carries result */ + return; + } + gbConverter = ucnv_open("GBK", errorCode); + if(U_FAILURE(*errorCode)) { + return; + } + cnv->toUnicodeStatus = 0; + cnv->fromUnicodeStatus= 0; + cnv->mode=0; + cnv->fromUChar32=0x0000; + cnv->extraInfo = uprv_calloc(1, sizeof(UConverterDataHZ)); + if(cnv->extraInfo != nullptr){ + ((UConverterDataHZ*)cnv->extraInfo)->gbConverter = gbConverter; + } + else { + ucnv_close(gbConverter); + *errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } +} + +static void U_CALLCONV +_HZClose(UConverter *cnv){ + if(cnv->extraInfo != nullptr) { + ucnv_close (((UConverterDataHZ *) (cnv->extraInfo))->gbConverter); + if(!cnv->isExtraLocal) { + uprv_free(cnv->extraInfo); + } + cnv->extraInfo = nullptr; + } +} + +static void U_CALLCONV +_HZReset(UConverter *cnv, UConverterResetChoice choice){ + if(choice<=UCNV_RESET_TO_UNICODE) { + cnv->toUnicodeStatus = 0; + cnv->mode=0; + if(cnv->extraInfo != nullptr){ + ((UConverterDataHZ*)cnv->extraInfo)->isStateDBCS = false; + ((UConverterDataHZ*)cnv->extraInfo)->isEmptySegment = false; + } + } + if(choice!=UCNV_RESET_TO_UNICODE) { + cnv->fromUnicodeStatus= 0; + cnv->fromUChar32=0x0000; + if(cnv->extraInfo != nullptr){ + ((UConverterDataHZ*)cnv->extraInfo)->isEscapeAppended = false; + ((UConverterDataHZ*)cnv->extraInfo)->targetIndex = 0; + ((UConverterDataHZ*)cnv->extraInfo)->sourceIndex = 0; + ((UConverterDataHZ*)cnv->extraInfo)->isTargetUCharDBCS = false; + } + } +} + +/**************************************HZ Encoding************************************************* +* Rules for HZ encoding +* +* In ASCII mode, a byte is interpreted as an ASCII character, unless a +* '~' is encountered. The character '~' is an escape character. By +* convention, it must be immediately followed ONLY by '~', '{' or '\n' +* (), with the following special meaning. + +* 1. The escape sequence '~~' is interpreted as a '~'. +* 2. The escape-to-GB sequence '~{' switches the mode from ASCII to GB. +* 3. The escape sequence '~\n' is a line-continuation marker to be +* consumed with no output produced. +* In GB mode, characters are interpreted two bytes at a time as (pure) +* GB codes until the escape-from-GB code '~}' is read. This code +* switches the mode from GB back to ASCII. (Note that the escape- +* from-GB code '~}' ($7E7D) is outside the defined GB range.) +* +* Source: RFC 1842 +* +* Note that the formal syntax in RFC 1842 is invalid. I assume that the +* intended definition of single-byte-segment is as follows (pedberg): +* single-byte-segment = single-byte-seq 1*single-byte-char +*/ + + +static void U_CALLCONV +UConverter_toUnicode_HZ_OFFSETS_LOGIC(UConverterToUnicodeArgs *args, + UErrorCode* err){ + char tempBuf[2]; + const char *mySource = ( char *) args->source; + char16_t *myTarget = args->target; + const char *mySourceLimit = args->sourceLimit; + UChar32 targetUniChar = 0x0000; + int32_t mySourceChar = 0x0000; + UConverterDataHZ* myData=(UConverterDataHZ*)(args->converter->extraInfo); + tempBuf[0]=0; + tempBuf[1]=0; + + /* Calling code already handles this situation. */ + /*if ((args->converter == nullptr) || (args->targetLimit < args->target) || (mySourceLimit < args->source)){ + *err = U_ILLEGAL_ARGUMENT_ERROR; + return; + }*/ + + while(mySource< mySourceLimit){ + + if(myTarget < args->targetLimit){ + + mySourceChar= (unsigned char) *mySource++; + + if(args->converter->mode == UCNV_TILDE) { + /* second byte after ~ */ + args->converter->mode=0; + switch(mySourceChar) { + case 0x0A: + /* no output for ~\n (line-continuation marker) */ + continue; + case UCNV_TILDE: + if(args->offsets) { + args->offsets[myTarget - args->target]=(int32_t)(mySource - args->source - 2); + } + *(myTarget++)=(char16_t)mySourceChar; + myData->isEmptySegment = false; + continue; + case UCNV_OPEN_BRACE: + case UCNV_CLOSE_BRACE: + myData->isStateDBCS = (mySourceChar == UCNV_OPEN_BRACE); + if (myData->isEmptySegment) { + myData->isEmptySegment = false; /* we are handling it, reset to avoid future spurious errors */ + *err = U_ILLEGAL_ESCAPE_SEQUENCE; + args->converter->toUCallbackReason = UCNV_IRREGULAR; + args->converter->toUBytes[0] = UCNV_TILDE; + args->converter->toUBytes[1] = static_cast(mySourceChar); + args->converter->toULength = 2; + args->target = myTarget; + args->source = mySource; + return; + } + myData->isEmptySegment = true; + continue; + default: + /* if the first byte is equal to TILDE and the trail byte + * is not a valid byte then it is an error condition + */ + /* + * Ticket 5691: consistent illegal sequences: + * - We include at least the first byte in the illegal sequence. + * - If any of the non-initial bytes could be the start of a character, + * we stop the illegal sequence before the first one of those. + */ + myData->isEmptySegment = false; /* different error here, reset this to avoid spurious future error */ + *err = U_ILLEGAL_ESCAPE_SEQUENCE; + args->converter->toUBytes[0] = UCNV_TILDE; + if( myData->isStateDBCS ? + (0x21 <= mySourceChar && mySourceChar <= 0x7e) : + mySourceChar <= 0x7f + ) { + /* The current byte could be the start of a character: Back it out. */ + args->converter->toULength = 1; + --mySource; + } else { + /* Include the current byte in the illegal sequence. */ + args->converter->toUBytes[1] = static_cast(mySourceChar); + args->converter->toULength = 2; + } + args->target = myTarget; + args->source = mySource; + return; + } + } else if(myData->isStateDBCS) { + if(args->converter->toUnicodeStatus == 0x00){ + /* lead byte */ + if(mySourceChar == UCNV_TILDE) { + args->converter->mode = UCNV_TILDE; + } else { + /* add another bit to distinguish a 0 byte from not having seen a lead byte */ + args->converter->toUnicodeStatus = (uint32_t) (mySourceChar | 0x100); + myData->isEmptySegment = false; /* the segment has something, either valid or will produce a different error, so reset this */ + } + continue; + } + else{ + /* trail byte */ + int leadIsOk, trailIsOk; + uint32_t leadByte = args->converter->toUnicodeStatus & 0xff; + targetUniChar = 0xffff; + /* + * Ticket 5691: consistent illegal sequences: + * - We include at least the first byte in the illegal sequence. + * - If any of the non-initial bytes could be the start of a character, + * we stop the illegal sequence before the first one of those. + * + * In HZ DBCS, if the second byte is in the 21..7e range, + * we report only the first byte as the illegal sequence. + * Otherwise we convert or report the pair of bytes. + */ + leadIsOk = (uint8_t)(leadByte - 0x21) <= (0x7d - 0x21); + trailIsOk = (uint8_t)(mySourceChar - 0x21) <= (0x7e - 0x21); + if (leadIsOk && trailIsOk) { + tempBuf[0] = (char) (leadByte+0x80) ; + tempBuf[1] = (char) (mySourceChar+0x80); + targetUniChar = ucnv_MBCSSimpleGetNextUChar(myData->gbConverter->sharedData, + tempBuf, 2, args->converter->useFallback); + mySourceChar= (leadByte << 8) | mySourceChar; + } else if (trailIsOk) { + /* report a single illegal byte and continue with the following DBCS starter byte */ + --mySource; + mySourceChar = (int32_t)leadByte; + } else { + /* report a pair of illegal bytes if the second byte is not a DBCS starter */ + /* add another bit so that the code below writes 2 bytes in case of error */ + mySourceChar= 0x10000 | (leadByte << 8) | mySourceChar; + } + args->converter->toUnicodeStatus =0x00; + } + } + else{ + if(mySourceChar == UCNV_TILDE) { + args->converter->mode = UCNV_TILDE; + continue; + } else if(mySourceChar <= 0x7f) { + targetUniChar = (char16_t)mySourceChar; /* ASCII */ + myData->isEmptySegment = false; /* the segment has something valid */ + } else { + targetUniChar = 0xffff; + myData->isEmptySegment = false; /* different error here, reset this to avoid spurious future error */ + } + } + if(targetUniChar < 0xfffe){ + if(args->offsets) { + args->offsets[myTarget - args->target]=(int32_t)(mySource - args->source - 1-(myData->isStateDBCS)); + } + + *(myTarget++)=(char16_t)targetUniChar; + } + else /* targetUniChar>=0xfffe */ { + if(targetUniChar == 0xfffe){ + *err = U_INVALID_CHAR_FOUND; + } + else{ + *err = U_ILLEGAL_CHAR_FOUND; + } + if(mySourceChar > 0xff){ + args->converter->toUBytes[0] = (uint8_t)(mySourceChar >> 8); + args->converter->toUBytes[1] = (uint8_t)mySourceChar; + args->converter->toULength=2; + } + else{ + args->converter->toUBytes[0] = (uint8_t)mySourceChar; + args->converter->toULength=1; + } + break; + } + } + else{ + *err =U_BUFFER_OVERFLOW_ERROR; + break; + } + } + + args->target = myTarget; + args->source = mySource; +} + + +static void U_CALLCONV +UConverter_fromUnicode_HZ_OFFSETS_LOGIC (UConverterFromUnicodeArgs * args, + UErrorCode * err){ + const char16_t *mySource = args->source; + char *myTarget = args->target; + int32_t* offsets = args->offsets; + int32_t mySourceIndex = 0; + int32_t myTargetIndex = 0; + int32_t targetLength = (int32_t)(args->targetLimit - myTarget); + int32_t mySourceLength = (int32_t)(args->sourceLimit - args->source); + uint32_t targetUniChar = 0x0000; + UChar32 mySourceChar = 0x0000; + UConverterDataHZ *myConverterData=(UConverterDataHZ*)args->converter->extraInfo; + UBool isTargetUCharDBCS = (UBool) myConverterData->isTargetUCharDBCS; + UBool oldIsTargetUCharDBCS; + int len =0; + const char* escSeq=nullptr; + + /* Calling code already handles this situation. */ + /*if ((args->converter == nullptr) || (args->targetLimit < myTarget) || (args->sourceLimit < args->source)){ + *err = U_ILLEGAL_ARGUMENT_ERROR; + return; + }*/ + if(args->converter->fromUChar32!=0 && myTargetIndex < targetLength) { + goto getTrail; + } + /*writing the char to the output stream */ + while (mySourceIndex < mySourceLength){ + targetUniChar = missingCharMarker; + if (myTargetIndex < targetLength){ + + mySourceChar = (char16_t) mySource[mySourceIndex++]; + + + oldIsTargetUCharDBCS = isTargetUCharDBCS; + if(mySourceChar ==UCNV_TILDE){ + /*concatEscape(args, &myTargetIndex, &targetLength,"\x7E\x7E",err,2,&mySourceIndex);*/ + len = ESC_LEN; + escSeq = TILDE_ESCAPE; + CONCAT_ESCAPE_MACRO(args, myTargetIndex, targetLength, escSeq,err,len,mySourceIndex); + continue; + } else if(mySourceChar <= 0x7f) { + targetUniChar = mySourceChar; + } else { + int32_t length= ucnv_MBCSFromUChar32(myConverterData->gbConverter->sharedData, + mySourceChar,&targetUniChar,args->converter->useFallback); + /* we can only use lead bytes 21..7D and trail bytes 21..7E */ + if( length == 2 && + (uint16_t)(targetUniChar - 0xa1a1) <= (0xfdfe - 0xa1a1) && + (uint8_t)(targetUniChar - 0xa1) <= (0xfe - 0xa1) + ) { + targetUniChar -= 0x8080; + } else { + targetUniChar = missingCharMarker; + } + } + if (targetUniChar != missingCharMarker){ + myConverterData->isTargetUCharDBCS = isTargetUCharDBCS = (UBool)(targetUniChar>0x00FF); + if(oldIsTargetUCharDBCS != isTargetUCharDBCS || !myConverterData->isEscapeAppended ){ + /*Shifting from a double byte to single byte mode*/ + if(!isTargetUCharDBCS){ + len =ESC_LEN; + escSeq = SB_ESCAPE; + CONCAT_ESCAPE_MACRO(args, myTargetIndex, targetLength, escSeq,err,len,mySourceIndex); + myConverterData->isEscapeAppended = true; + } + else{ /* Shifting from a single byte to double byte mode*/ + len =ESC_LEN; + escSeq = DB_ESCAPE; + CONCAT_ESCAPE_MACRO(args, myTargetIndex, targetLength, escSeq,err,len,mySourceIndex); + myConverterData->isEscapeAppended = true; + + } + } + + if(isTargetUCharDBCS){ + if( myTargetIndex > 8); + if(offsets){ + *(offsets++) = mySourceIndex-1; + } + if(myTargetIndex < targetLength){ + myTarget[myTargetIndex++] =(char) targetUniChar; + if(offsets){ + *(offsets++) = mySourceIndex-1; + } + }else{ + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = (char) targetUniChar; + *err = U_BUFFER_OVERFLOW_ERROR; + } + }else{ + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] =(char) (targetUniChar >> 8); + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = (char) targetUniChar; + *err = U_BUFFER_OVERFLOW_ERROR; + } + + }else{ + if( myTargetIndex converter->charErrorBuffer[args->converter->charErrorBufferLength++] = (char) targetUniChar; + *err = U_BUFFER_OVERFLOW_ERROR; + } + } + + } + else{ + /* oops.. the code point is unassigned */ + /*Handle surrogates */ + /*check if the char is a First surrogate*/ + if(U16_IS_SURROGATE(mySourceChar)) { + if(U16_IS_SURROGATE_LEAD(mySourceChar)) { + args->converter->fromUChar32=mySourceChar; +getTrail: + /*look ahead to find the trail surrogate*/ + if(mySourceIndex < mySourceLength) { + /* test the following code unit */ + char16_t trail=(char16_t) args->source[mySourceIndex]; + if(U16_IS_TRAIL(trail)) { + ++mySourceIndex; + mySourceChar=U16_GET_SUPPLEMENTARY(args->converter->fromUChar32, trail); + args->converter->fromUChar32=0x00; + /* there are no surrogates in GB2312*/ + *err = U_INVALID_CHAR_FOUND; + /* exit this condition tree */ + } else { + /* this is an unmatched lead code unit (1st surrogate) */ + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + } + } else { + /* no more input */ + *err = U_ZERO_ERROR; + } + } else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + } + } else { + /* callback(unassigned) for a BMP code point */ + *err = U_INVALID_CHAR_FOUND; + } + + args->converter->fromUChar32=mySourceChar; + break; + } + } + else{ + *err = U_BUFFER_OVERFLOW_ERROR; + break; + } + targetUniChar=missingCharMarker; + } + + args->target += myTargetIndex; + args->source += mySourceIndex; + myConverterData->isTargetUCharDBCS = isTargetUCharDBCS; +} + +static void U_CALLCONV +_HZ_WriteSub(UConverterFromUnicodeArgs *args, int32_t offsetIndex, UErrorCode *err) { + UConverter *cnv = args->converter; + UConverterDataHZ *convData=(UConverterDataHZ *) cnv->extraInfo; + char *p; + char buffer[4]; + p = buffer; + + if( convData->isTargetUCharDBCS){ + *p++= UCNV_TILDE; + *p++= UCNV_CLOSE_BRACE; + convData->isTargetUCharDBCS=false; + } + *p++= (char)cnv->subChars[0]; + + ucnv_cbFromUWriteBytes(args, + buffer, (int32_t)(p - buffer), + offsetIndex, err); +} + +/* + * Structure for cloning an HZ converter into a single memory block. + */ +struct cloneHZStruct +{ + UConverter cnv; + UConverter subCnv; + UConverterDataHZ mydata; +}; + + +static UConverter * U_CALLCONV +_HZ_SafeClone(const UConverter *cnv, + void *stackBuffer, + int32_t *pBufferSize, + UErrorCode *status) +{ + struct cloneHZStruct * localClone; + int32_t size, bufferSizeNeeded = sizeof(struct cloneHZStruct); + + if (U_FAILURE(*status)){ + return nullptr; + } + + if (*pBufferSize == 0){ /* 'preflighting' request - set needed size into *pBufferSize */ + *pBufferSize = bufferSizeNeeded; + return nullptr; + } + + localClone = (struct cloneHZStruct *)stackBuffer; + /* ucnv.c/ucnv_safeClone() copied the main UConverter already */ + + uprv_memcpy(&localClone->mydata, cnv->extraInfo, sizeof(UConverterDataHZ)); + localClone->cnv.extraInfo = &localClone->mydata; + localClone->cnv.isExtraLocal = true; + + /* deep-clone the sub-converter */ + size = (int32_t)sizeof(UConverter); + ((UConverterDataHZ*)localClone->cnv.extraInfo)->gbConverter = + ucnv_safeClone(((UConverterDataHZ*)cnv->extraInfo)->gbConverter, &localClone->subCnv, &size, status); + + return &localClone->cnv; +} + +static void U_CALLCONV +_HZ_GetUnicodeSet(const UConverter *cnv, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode) { + /* HZ converts all of ASCII */ + sa->addRange(sa->set, 0, 0x7f); + + /* add all of the code points that the sub-converter handles */ + ucnv_MBCSGetFilteredUnicodeSetForUnicode( + ((UConverterDataHZ*)cnv->extraInfo)->gbConverter->sharedData, + sa, which, UCNV_SET_FILTER_HZ, + pErrorCode); +} +U_CDECL_END +static const UConverterImpl _HZImpl={ + + UCNV_HZ, + + nullptr, + nullptr, + + _HZOpen, + _HZClose, + _HZReset, + + UConverter_toUnicode_HZ_OFFSETS_LOGIC, + UConverter_toUnicode_HZ_OFFSETS_LOGIC, + UConverter_fromUnicode_HZ_OFFSETS_LOGIC, + UConverter_fromUnicode_HZ_OFFSETS_LOGIC, + nullptr, + + nullptr, + nullptr, + _HZ_WriteSub, + _HZ_SafeClone, + _HZ_GetUnicodeSet, + nullptr, + nullptr +}; + +static const UConverterStaticData _HZStaticData={ + sizeof(UConverterStaticData), + "HZ", + 0, + UCNV_IBM, + UCNV_HZ, + 1, + 4, + { 0x1a, 0, 0, 0 }, + 1, + false, + false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 }, /* reserved */ + +}; + +const UConverterSharedData _HZData= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_HZStaticData, &_HZImpl); + +#endif /* #if !UCONFIG_NO_CONVERSION && !UCONFIG_NO_LEGACY_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION */ diff --git a/intl/icu/source/common/ucnvisci.cpp b/intl/icu/source/common/ucnvisci.cpp new file mode 100644 index 0000000000..507a50c892 --- /dev/null +++ b/intl/icu/source/common/ucnvisci.cpp @@ -0,0 +1,1635 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2000-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* file name: ucnvisci.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2001JUN26 +* created by: Ram Viswanadha +* +* Date Name Description +* 24/7/2001 Ram Added support for EXT character handling +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION && !UCONFIG_NO_LEGACY_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/ucnv_cb.h" +#include "unicode/utf16.h" +#include "cmemory.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" +#include "cstring.h" +#include "uassert.h" + +#define UCNV_OPTIONS_VERSION_MASK 0xf +#define NUKTA 0x093c +#define HALANT 0x094d +#define ZWNJ 0x200c /* Zero Width Non Joiner */ +#define ZWJ 0x200d /* Zero width Joiner */ +#define INVALID_CHAR 0xffff +#define ATR 0xEF /* Attribute code */ +#define EXT 0xF0 /* Extension code */ +#define DANDA 0x0964 +#define DOUBLE_DANDA 0x0965 +#define ISCII_NUKTA 0xE9 +#define ISCII_HALANT 0xE8 +#define ISCII_DANDA 0xEA +#define ISCII_INV 0xD9 +#define ISCII_VOWEL_SIGN_E 0xE0 +#define INDIC_BLOCK_BEGIN 0x0900 +#define INDIC_BLOCK_END 0x0D7F +#define INDIC_RANGE (INDIC_BLOCK_END - INDIC_BLOCK_BEGIN) +#define VOCALLIC_RR 0x0931 +#define LF 0x0A +#define ASCII_END 0xA0 +#define NO_CHAR_MARKER 0xFFFE +#define TELUGU_DELTA DELTA * TELUGU +#define DEV_ABBR_SIGN 0x0970 +#define DEV_ANUDATTA 0x0952 +#define EXT_RANGE_BEGIN 0xA1 +#define EXT_RANGE_END 0xEE + +#define PNJ_DELTA 0x0100 +#define PNJ_BINDI 0x0A02 +#define PNJ_TIPPI 0x0A70 +#define PNJ_SIGN_VIRAMA 0x0A4D +#define PNJ_ADHAK 0x0A71 +#define PNJ_HA 0x0A39 +#define PNJ_RRA 0x0A5C + +typedef enum { + DEVANAGARI =0, + BENGALI, + GURMUKHI, + GUJARATI, + ORIYA, + TAMIL, + TELUGU, + KANNADA, + MALAYALAM, + DELTA=0x80 +}UniLang; + +/** + * Enumeration for switching code pages if + + * is encountered + */ +typedef enum { + DEF = 0x40, + RMN = 0x41, + DEV = 0x42, + BNG = 0x43, + TML = 0x44, + TLG = 0x45, + ASM = 0x46, + ORI = 0x47, + KND = 0x48, + MLM = 0x49, + GJR = 0x4A, + PNJ = 0x4B, + ARB = 0x71, + PES = 0x72, + URD = 0x73, + SND = 0x74, + KSM = 0x75, + PST = 0x76 +}ISCIILang; + +typedef enum { + DEV_MASK =0x80, + PNJ_MASK =0x40, + GJR_MASK =0x20, + ORI_MASK =0x10, + BNG_MASK =0x08, + KND_MASK =0x04, + MLM_MASK =0x02, + TML_MASK =0x01, + ZERO =0x00 +}MaskEnum; + +#define ISCII_CNV_PREFIX "ISCII,version=" + +typedef struct { + char16_t contextCharToUnicode; /* previous Unicode codepoint for contextual analysis */ + char16_t contextCharFromUnicode; /* previous Unicode codepoint for contextual analysis */ + uint16_t defDeltaToUnicode; /* delta for switching to default state when DEF is encountered */ + uint16_t currentDeltaFromUnicode; /* current delta in Indic block */ + uint16_t currentDeltaToUnicode; /* current delta in Indic block */ + MaskEnum currentMaskFromUnicode; /* mask for current state in toUnicode */ + MaskEnum currentMaskToUnicode; /* mask for current state in toUnicode */ + MaskEnum defMaskToUnicode; /* mask for default state in toUnicode */ + UBool isFirstBuffer; /* boolean for fromUnicode to see if we need to announce the first script */ + UBool resetToDefaultToUnicode; /* boolean for resetting to default delta and mask when a newline is encountered*/ + char name[sizeof(ISCII_CNV_PREFIX) + 1]; + UChar32 prevToUnicodeStatus; /* Hold the previous toUnicodeStatus. This is necessary because we may need to know the last two code points. */ +} UConverterDataISCII; + +typedef struct LookupDataStruct { + UniLang uniLang; + MaskEnum maskEnum; + ISCIILang isciiLang; +} LookupDataStruct; + +static const LookupDataStruct lookupInitialData[]={ + { DEVANAGARI, DEV_MASK, DEV }, + { BENGALI, BNG_MASK, BNG }, + { GURMUKHI, PNJ_MASK, PNJ }, + { GUJARATI, GJR_MASK, GJR }, + { ORIYA, ORI_MASK, ORI }, + { TAMIL, TML_MASK, TML }, + { TELUGU, KND_MASK, TLG }, + { KANNADA, KND_MASK, KND }, + { MALAYALAM, MLM_MASK, MLM } +}; + +/* + * For special handling of certain Gurmukhi characters. + * Bit 0 (value 1): PNJ consonant + * Bit 1 (value 2): PNJ Bindi Tippi + */ +static const uint8_t pnjMap[80] = { + /* 0A00..0A0F */ + 0, 0, 0, 0, 0, 2, 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, + /* 0A10..0A1F */ + 0, 0, 0, 0, 0, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, + /* 0A20..0A2F */ + 3, 3, 3, 3, 3, 3, 3, 3, 3, 0, 3, 3, 3, 3, 3, 3, + /* 0A30..0A3F */ + 3, 0, 0, 0, 0, 3, 3, 0, 3, 3, 0, 0, 0, 0, 0, 2, + /* 0A40..0A4F */ + 0, 2, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 +}; + +static UBool +isPNJConsonant(UChar32 c) { + if (c < 0xa00 || 0xa50 <= c) { + return false; + } else { + return (UBool)(pnjMap[c - 0xa00] & 1); + } +} + +static UBool +isPNJBindiTippi(UChar32 c) { + if (c < 0xa00 || 0xa50 <= c) { + return false; + } else { + return (UBool)(pnjMap[c - 0xa00] >> 1); + } +} +U_CDECL_BEGIN +static void U_CALLCONV +_ISCIIOpen(UConverter *cnv, UConverterLoadArgs *pArgs, UErrorCode *errorCode) { + if(pArgs->onlyTestIsLoadable) { + return; + } + + cnv->extraInfo = uprv_malloc(sizeof(UConverterDataISCII)); + + if (cnv->extraInfo != nullptr) { + int32_t len=0; + UConverterDataISCII *converterData= + (UConverterDataISCII *) cnv->extraInfo; + converterData->contextCharToUnicode=NO_CHAR_MARKER; + cnv->toUnicodeStatus = missingCharMarker; + converterData->contextCharFromUnicode=0x0000; + converterData->resetToDefaultToUnicode=false; + /* check if the version requested is supported */ + if ((pArgs->options & UCNV_OPTIONS_VERSION_MASK) < 9) { + /* initialize state variables */ + converterData->currentDeltaFromUnicode + = converterData->currentDeltaToUnicode + = converterData->defDeltaToUnicode = (uint16_t)(lookupInitialData[pArgs->options & UCNV_OPTIONS_VERSION_MASK].uniLang * DELTA); + + converterData->currentMaskFromUnicode + = converterData->currentMaskToUnicode + = converterData->defMaskToUnicode = lookupInitialData[pArgs->options & UCNV_OPTIONS_VERSION_MASK].maskEnum; + + converterData->isFirstBuffer=true; + (void)uprv_strcpy(converterData->name, ISCII_CNV_PREFIX); + len = (int32_t)uprv_strlen(converterData->name); + converterData->name[len]= (char)((pArgs->options & UCNV_OPTIONS_VERSION_MASK) + '0'); + converterData->name[len+1]=0; + + converterData->prevToUnicodeStatus = 0x0000; + } else { + uprv_free(cnv->extraInfo); + cnv->extraInfo = nullptr; + *errorCode = U_ILLEGAL_ARGUMENT_ERROR; + } + + } else { + *errorCode =U_MEMORY_ALLOCATION_ERROR; + } +} + +static void U_CALLCONV +_ISCIIClose(UConverter *cnv) { + if (cnv->extraInfo!=nullptr) { + if (!cnv->isExtraLocal) { + uprv_free(cnv->extraInfo); + } + cnv->extraInfo=nullptr; + } +} + +static const char* U_CALLCONV +_ISCIIgetName(const UConverter* cnv) { + if (cnv->extraInfo) { + UConverterDataISCII* myData= (UConverterDataISCII*)cnv->extraInfo; + return myData->name; + } + return nullptr; +} + +static void U_CALLCONV +_ISCIIReset(UConverter *cnv, UConverterResetChoice choice) { + UConverterDataISCII* data =(UConverterDataISCII *) (cnv->extraInfo); + if (choice<=UCNV_RESET_TO_UNICODE) { + cnv->toUnicodeStatus = missingCharMarker; + cnv->mode=0; + data->currentDeltaToUnicode=data->defDeltaToUnicode; + data->currentMaskToUnicode = data->defMaskToUnicode; + data->contextCharToUnicode=NO_CHAR_MARKER; + data->prevToUnicodeStatus = 0x0000; + } + if (choice!=UCNV_RESET_TO_UNICODE) { + cnv->fromUChar32=0x0000; + data->contextCharFromUnicode=0x00; + data->currentMaskFromUnicode=data->defMaskToUnicode; + data->currentDeltaFromUnicode=data->defDeltaToUnicode; + data->isFirstBuffer=true; + data->resetToDefaultToUnicode=false; + } +} + +/** + * The values in validity table are indexed by the lower bits of Unicode + * range 0x0900 - 0x09ff. The values have a structure like: + * --------------------------------------------------------------- + * | DEV | PNJ | GJR | ORI | BNG | TLG | MLM | TML | + * | | | | | ASM | KND | | | + * --------------------------------------------------------------- + * If a code point is valid in a particular script + * then that bit is turned on + * + * Unicode does not distinguish between Bengali and Assamese so we use 1 bit for + * to represent these languages + * + * Telugu and Kannada have same codepoints except for Vocallic_RR which we special case + * and combine and use 1 bit to represent these languages. + * + * TODO: It is probably easier to understand and maintain to change this + * to use uint16_t and give each of the 9 Unicode/script blocks its own bit. + */ + +static const uint8_t validityTable[128] = { +/* This state table is tool generated please do not edit unless you know exactly what you are doing */ +/* Note: This table was edited to mirror the Windows XP implementation */ +/*ISCII:Valid:Unicode */ +/*0xa0 : 0x00: 0x900 */ ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0xa1 : 0xb8: 0x901 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + ZERO + ZERO + ZERO , +/*0xa2 : 0xfe: 0x902 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xa3 : 0xbf: 0x903 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0x00 : 0x00: 0x904 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0xa4 : 0xff: 0x905 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xa5 : 0xff: 0x906 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xa6 : 0xff: 0x907 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xa7 : 0xff: 0x908 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xa8 : 0xff: 0x909 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xa9 : 0xff: 0x90a */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xaa : 0xfe: 0x90b */ DEV_MASK + ZERO + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0x00 : 0x00: 0x90c */ DEV_MASK + ZERO + ZERO + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xae : 0x80: 0x90d */ DEV_MASK + ZERO + GJR_MASK + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0xab : 0x87: 0x90e */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + KND_MASK + MLM_MASK + TML_MASK , +/*0xac : 0xff: 0x90f */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xad : 0xff: 0x910 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xb2 : 0x80: 0x911 */ DEV_MASK + ZERO + GJR_MASK + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0xaf : 0x87: 0x912 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + KND_MASK + MLM_MASK + TML_MASK , +/*0xb0 : 0xff: 0x913 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xb1 : 0xff: 0x914 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xb3 : 0xff: 0x915 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xb4 : 0xfe: 0x916 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xb5 : 0xfe: 0x917 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xb6 : 0xfe: 0x918 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xb7 : 0xff: 0x919 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xb8 : 0xff: 0x91a */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xb9 : 0xfe: 0x91b */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xba : 0xff: 0x91c */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xbb : 0xfe: 0x91d */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xbc : 0xff: 0x91e */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xbd : 0xff: 0x91f */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xbe : 0xfe: 0x920 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xbf : 0xfe: 0x921 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xc0 : 0xfe: 0x922 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xc1 : 0xff: 0x923 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xc2 : 0xff: 0x924 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xc3 : 0xfe: 0x925 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xc4 : 0xfe: 0x926 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xc5 : 0xfe: 0x927 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xc6 : 0xff: 0x928 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xc7 : 0x81: 0x929 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + TML_MASK , +/*0xc8 : 0xff: 0x92a */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xc9 : 0xfe: 0x92b */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xca : 0xfe: 0x92c */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xcb : 0xfe: 0x92d */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xcc : 0xfe: 0x92e */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xcd : 0xff: 0x92f */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xcf : 0xff: 0x930 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xd0 : 0x87: 0x931 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + MLM_MASK + TML_MASK , +/*0xd1 : 0xff: 0x932 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xd2 : 0xb7: 0x933 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + ZERO + KND_MASK + MLM_MASK + TML_MASK , +/*0xd3 : 0x83: 0x934 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + MLM_MASK + TML_MASK , +/*0xd4 : 0xff: 0x935 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + ZERO + KND_MASK + MLM_MASK + TML_MASK , +/*0xd5 : 0xfe: 0x936 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0xd6 : 0xbf: 0x937 */ DEV_MASK + ZERO + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xd7 : 0xff: 0x938 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xd8 : 0xff: 0x939 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0x00 : 0x00: 0x93A */ ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x93B */ ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0xe9 : 0xda: 0x93c */ DEV_MASK + PNJ_MASK + ZERO + ORI_MASK + BNG_MASK + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x93d */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0xda : 0xff: 0x93e */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xdb : 0xff: 0x93f */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xdc : 0xff: 0x940 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xdd : 0xff: 0x941 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xde : 0xff: 0x942 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xdf : 0xbe: 0x943 */ DEV_MASK + ZERO + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0x00 : 0x00: 0x944 */ DEV_MASK + ZERO + GJR_MASK + ZERO + BNG_MASK + KND_MASK + ZERO + ZERO , +/*0xe3 : 0x80: 0x945 */ DEV_MASK + ZERO + GJR_MASK + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0xe0 : 0x87: 0x946 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + KND_MASK + MLM_MASK + TML_MASK , +/*0xe1 : 0xff: 0x947 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xe2 : 0xff: 0x948 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xe7 : 0x80: 0x949 */ DEV_MASK + ZERO + GJR_MASK + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0xe4 : 0x87: 0x94a */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + KND_MASK + MLM_MASK + TML_MASK , +/*0xe5 : 0xff: 0x94b */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xe6 : 0xff: 0x94c */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xe8 : 0xff: 0x94d */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xec : 0x00: 0x94e */ ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0xed : 0x00: 0x94f */ ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x950 */ DEV_MASK + ZERO + GJR_MASK + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x951 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x952 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x953 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x954 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x955 */ ZERO + ZERO + ZERO + ZERO + ZERO + KND_MASK + ZERO + ZERO , +/*0x00 : 0x00: 0x956 */ ZERO + ZERO + ZERO + ORI_MASK + ZERO + KND_MASK + ZERO + ZERO , +/*0x00 : 0x00: 0x957 */ ZERO + ZERO + ZERO + ORI_MASK + BNG_MASK + ZERO + MLM_MASK + ZERO , +/*0x00 : 0x00: 0x958 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x959 */ DEV_MASK + PNJ_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x95a */ DEV_MASK + PNJ_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x95b */ DEV_MASK + PNJ_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x95c */ DEV_MASK + PNJ_MASK + ZERO + ZERO + BNG_MASK + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x95d */ DEV_MASK + ZERO + ZERO + ORI_MASK + BNG_MASK + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x95e */ DEV_MASK + PNJ_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0xce : 0x98: 0x95f */ DEV_MASK + ZERO + ZERO + ORI_MASK + BNG_MASK + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x960 */ DEV_MASK + ZERO + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0x00 : 0x00: 0x961 */ DEV_MASK + ZERO + ZERO + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO , +/*0x00 : 0x00: 0x962 */ DEV_MASK + ZERO + ZERO + ZERO + BNG_MASK + ZERO + ZERO + ZERO , +/*0x00 : 0x00: 0x963 */ DEV_MASK + ZERO + ZERO + ZERO + BNG_MASK + ZERO + ZERO + ZERO , +/*0xea : 0xf8: 0x964 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0xeaea : 0x00: 0x965*/ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/*0xf1 : 0xff: 0x966 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xf2 : 0xff: 0x967 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xf3 : 0xff: 0x968 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xf4 : 0xff: 0x969 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xf5 : 0xff: 0x96a */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xf6 : 0xff: 0x96b */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xf7 : 0xff: 0x96c */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xf8 : 0xff: 0x96d */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xf9 : 0xff: 0x96e */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0xfa : 0xff: 0x96f */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK , +/*0x00 : 0x80: 0x970 */ DEV_MASK + PNJ_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO , +/* + * The length of the array is 128 to provide values for 0x900..0x97f. + * The last 15 entries for 0x971..0x97f of the validity table are all zero + * because no Indic script uses such Unicode code points. + */ +/*0x00 : 0x00: 0x9yz */ ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO +}; + +static const uint16_t fromUnicodeTable[128]={ + 0x00a0 ,/* 0x0900 */ + 0x00a1 ,/* 0x0901 */ + 0x00a2 ,/* 0x0902 */ + 0x00a3 ,/* 0x0903 */ + 0xa4e0 ,/* 0x0904 */ + 0x00a4 ,/* 0x0905 */ + 0x00a5 ,/* 0x0906 */ + 0x00a6 ,/* 0x0907 */ + 0x00a7 ,/* 0x0908 */ + 0x00a8 ,/* 0x0909 */ + 0x00a9 ,/* 0x090a */ + 0x00aa ,/* 0x090b */ + 0xA6E9 ,/* 0x090c */ + 0x00ae ,/* 0x090d */ + 0x00ab ,/* 0x090e */ + 0x00ac ,/* 0x090f */ + 0x00ad ,/* 0x0910 */ + 0x00b2 ,/* 0x0911 */ + 0x00af ,/* 0x0912 */ + 0x00b0 ,/* 0x0913 */ + 0x00b1 ,/* 0x0914 */ + 0x00b3 ,/* 0x0915 */ + 0x00b4 ,/* 0x0916 */ + 0x00b5 ,/* 0x0917 */ + 0x00b6 ,/* 0x0918 */ + 0x00b7 ,/* 0x0919 */ + 0x00b8 ,/* 0x091a */ + 0x00b9 ,/* 0x091b */ + 0x00ba ,/* 0x091c */ + 0x00bb ,/* 0x091d */ + 0x00bc ,/* 0x091e */ + 0x00bd ,/* 0x091f */ + 0x00be ,/* 0x0920 */ + 0x00bf ,/* 0x0921 */ + 0x00c0 ,/* 0x0922 */ + 0x00c1 ,/* 0x0923 */ + 0x00c2 ,/* 0x0924 */ + 0x00c3 ,/* 0x0925 */ + 0x00c4 ,/* 0x0926 */ + 0x00c5 ,/* 0x0927 */ + 0x00c6 ,/* 0x0928 */ + 0x00c7 ,/* 0x0929 */ + 0x00c8 ,/* 0x092a */ + 0x00c9 ,/* 0x092b */ + 0x00ca ,/* 0x092c */ + 0x00cb ,/* 0x092d */ + 0x00cc ,/* 0x092e */ + 0x00cd ,/* 0x092f */ + 0x00cf ,/* 0x0930 */ + 0x00d0 ,/* 0x0931 */ + 0x00d1 ,/* 0x0932 */ + 0x00d2 ,/* 0x0933 */ + 0x00d3 ,/* 0x0934 */ + 0x00d4 ,/* 0x0935 */ + 0x00d5 ,/* 0x0936 */ + 0x00d6 ,/* 0x0937 */ + 0x00d7 ,/* 0x0938 */ + 0x00d8 ,/* 0x0939 */ + 0xFFFF ,/* 0x093A */ + 0xFFFF ,/* 0x093B */ + 0x00e9 ,/* 0x093c */ + 0xEAE9 ,/* 0x093d */ + 0x00da ,/* 0x093e */ + 0x00db ,/* 0x093f */ + 0x00dc ,/* 0x0940 */ + 0x00dd ,/* 0x0941 */ + 0x00de ,/* 0x0942 */ + 0x00df ,/* 0x0943 */ + 0xDFE9 ,/* 0x0944 */ + 0x00e3 ,/* 0x0945 */ + 0x00e0 ,/* 0x0946 */ + 0x00e1 ,/* 0x0947 */ + 0x00e2 ,/* 0x0948 */ + 0x00e7 ,/* 0x0949 */ + 0x00e4 ,/* 0x094a */ + 0x00e5 ,/* 0x094b */ + 0x00e6 ,/* 0x094c */ + 0x00e8 ,/* 0x094d */ + 0x00ec ,/* 0x094e */ + 0x00ed ,/* 0x094f */ + 0xA1E9 ,/* 0x0950 */ /* OM Symbol */ + 0xFFFF ,/* 0x0951 */ + 0xF0B8 ,/* 0x0952 */ + 0xFFFF ,/* 0x0953 */ + 0xFFFF ,/* 0x0954 */ + 0xFFFF ,/* 0x0955 */ + 0xFFFF ,/* 0x0956 */ + 0xFFFF ,/* 0x0957 */ + 0xb3e9 ,/* 0x0958 */ + 0xb4e9 ,/* 0x0959 */ + 0xb5e9 ,/* 0x095a */ + 0xbae9 ,/* 0x095b */ + 0xbfe9 ,/* 0x095c */ + 0xC0E9 ,/* 0x095d */ + 0xc9e9 ,/* 0x095e */ + 0x00ce ,/* 0x095f */ + 0xAAe9 ,/* 0x0960 */ + 0xA7E9 ,/* 0x0961 */ + 0xDBE9 ,/* 0x0962 */ + 0xDCE9 ,/* 0x0963 */ + 0x00ea ,/* 0x0964 */ + 0xeaea ,/* 0x0965 */ + 0x00f1 ,/* 0x0966 */ + 0x00f2 ,/* 0x0967 */ + 0x00f3 ,/* 0x0968 */ + 0x00f4 ,/* 0x0969 */ + 0x00f5 ,/* 0x096a */ + 0x00f6 ,/* 0x096b */ + 0x00f7 ,/* 0x096c */ + 0x00f8 ,/* 0x096d */ + 0x00f9 ,/* 0x096e */ + 0x00fa ,/* 0x096f */ + 0xF0BF ,/* 0x0970 */ + 0xFFFF ,/* 0x0971 */ + 0xFFFF ,/* 0x0972 */ + 0xFFFF ,/* 0x0973 */ + 0xFFFF ,/* 0x0974 */ + 0xFFFF ,/* 0x0975 */ + 0xFFFF ,/* 0x0976 */ + 0xFFFF ,/* 0x0977 */ + 0xFFFF ,/* 0x0978 */ + 0xFFFF ,/* 0x0979 */ + 0xFFFF ,/* 0x097a */ + 0xFFFF ,/* 0x097b */ + 0xFFFF ,/* 0x097c */ + 0xFFFF ,/* 0x097d */ + 0xFFFF ,/* 0x097e */ + 0xFFFF ,/* 0x097f */ +}; +static const uint16_t toUnicodeTable[256]={ + 0x0000,/* 0x00 */ + 0x0001,/* 0x01 */ + 0x0002,/* 0x02 */ + 0x0003,/* 0x03 */ + 0x0004,/* 0x04 */ + 0x0005,/* 0x05 */ + 0x0006,/* 0x06 */ + 0x0007,/* 0x07 */ + 0x0008,/* 0x08 */ + 0x0009,/* 0x09 */ + 0x000a,/* 0x0a */ + 0x000b,/* 0x0b */ + 0x000c,/* 0x0c */ + 0x000d,/* 0x0d */ + 0x000e,/* 0x0e */ + 0x000f,/* 0x0f */ + 0x0010,/* 0x10 */ + 0x0011,/* 0x11 */ + 0x0012,/* 0x12 */ + 0x0013,/* 0x13 */ + 0x0014,/* 0x14 */ + 0x0015,/* 0x15 */ + 0x0016,/* 0x16 */ + 0x0017,/* 0x17 */ + 0x0018,/* 0x18 */ + 0x0019,/* 0x19 */ + 0x001a,/* 0x1a */ + 0x001b,/* 0x1b */ + 0x001c,/* 0x1c */ + 0x001d,/* 0x1d */ + 0x001e,/* 0x1e */ + 0x001f,/* 0x1f */ + 0x0020,/* 0x20 */ + 0x0021,/* 0x21 */ + 0x0022,/* 0x22 */ + 0x0023,/* 0x23 */ + 0x0024,/* 0x24 */ + 0x0025,/* 0x25 */ + 0x0026,/* 0x26 */ + 0x0027,/* 0x27 */ + 0x0028,/* 0x28 */ + 0x0029,/* 0x29 */ + 0x002a,/* 0x2a */ + 0x002b,/* 0x2b */ + 0x002c,/* 0x2c */ + 0x002d,/* 0x2d */ + 0x002e,/* 0x2e */ + 0x002f,/* 0x2f */ + 0x0030,/* 0x30 */ + 0x0031,/* 0x31 */ + 0x0032,/* 0x32 */ + 0x0033,/* 0x33 */ + 0x0034,/* 0x34 */ + 0x0035,/* 0x35 */ + 0x0036,/* 0x36 */ + 0x0037,/* 0x37 */ + 0x0038,/* 0x38 */ + 0x0039,/* 0x39 */ + 0x003A,/* 0x3A */ + 0x003B,/* 0x3B */ + 0x003c,/* 0x3c */ + 0x003d,/* 0x3d */ + 0x003e,/* 0x3e */ + 0x003f,/* 0x3f */ + 0x0040,/* 0x40 */ + 0x0041,/* 0x41 */ + 0x0042,/* 0x42 */ + 0x0043,/* 0x43 */ + 0x0044,/* 0x44 */ + 0x0045,/* 0x45 */ + 0x0046,/* 0x46 */ + 0x0047,/* 0x47 */ + 0x0048,/* 0x48 */ + 0x0049,/* 0x49 */ + 0x004a,/* 0x4a */ + 0x004b,/* 0x4b */ + 0x004c,/* 0x4c */ + 0x004d,/* 0x4d */ + 0x004e,/* 0x4e */ + 0x004f,/* 0x4f */ + 0x0050,/* 0x50 */ + 0x0051,/* 0x51 */ + 0x0052,/* 0x52 */ + 0x0053,/* 0x53 */ + 0x0054,/* 0x54 */ + 0x0055,/* 0x55 */ + 0x0056,/* 0x56 */ + 0x0057,/* 0x57 */ + 0x0058,/* 0x58 */ + 0x0059,/* 0x59 */ + 0x005a,/* 0x5a */ + 0x005b,/* 0x5b */ + 0x005c,/* 0x5c */ + 0x005d,/* 0x5d */ + 0x005e,/* 0x5e */ + 0x005f,/* 0x5f */ + 0x0060,/* 0x60 */ + 0x0061,/* 0x61 */ + 0x0062,/* 0x62 */ + 0x0063,/* 0x63 */ + 0x0064,/* 0x64 */ + 0x0065,/* 0x65 */ + 0x0066,/* 0x66 */ + 0x0067,/* 0x67 */ + 0x0068,/* 0x68 */ + 0x0069,/* 0x69 */ + 0x006a,/* 0x6a */ + 0x006b,/* 0x6b */ + 0x006c,/* 0x6c */ + 0x006d,/* 0x6d */ + 0x006e,/* 0x6e */ + 0x006f,/* 0x6f */ + 0x0070,/* 0x70 */ + 0x0071,/* 0x71 */ + 0x0072,/* 0x72 */ + 0x0073,/* 0x73 */ + 0x0074,/* 0x74 */ + 0x0075,/* 0x75 */ + 0x0076,/* 0x76 */ + 0x0077,/* 0x77 */ + 0x0078,/* 0x78 */ + 0x0079,/* 0x79 */ + 0x007a,/* 0x7a */ + 0x007b,/* 0x7b */ + 0x007c,/* 0x7c */ + 0x007d,/* 0x7d */ + 0x007e,/* 0x7e */ + 0x007f,/* 0x7f */ + 0x0080,/* 0x80 */ + 0x0081,/* 0x81 */ + 0x0082,/* 0x82 */ + 0x0083,/* 0x83 */ + 0x0084,/* 0x84 */ + 0x0085,/* 0x85 */ + 0x0086,/* 0x86 */ + 0x0087,/* 0x87 */ + 0x0088,/* 0x88 */ + 0x0089,/* 0x89 */ + 0x008a,/* 0x8a */ + 0x008b,/* 0x8b */ + 0x008c,/* 0x8c */ + 0x008d,/* 0x8d */ + 0x008e,/* 0x8e */ + 0x008f,/* 0x8f */ + 0x0090,/* 0x90 */ + 0x0091,/* 0x91 */ + 0x0092,/* 0x92 */ + 0x0093,/* 0x93 */ + 0x0094,/* 0x94 */ + 0x0095,/* 0x95 */ + 0x0096,/* 0x96 */ + 0x0097,/* 0x97 */ + 0x0098,/* 0x98 */ + 0x0099,/* 0x99 */ + 0x009a,/* 0x9a */ + 0x009b,/* 0x9b */ + 0x009c,/* 0x9c */ + 0x009d,/* 0x9d */ + 0x009e,/* 0x9e */ + 0x009f,/* 0x9f */ + 0x00A0,/* 0xa0 */ + 0x0901,/* 0xa1 */ + 0x0902,/* 0xa2 */ + 0x0903,/* 0xa3 */ + 0x0905,/* 0xa4 */ + 0x0906,/* 0xa5 */ + 0x0907,/* 0xa6 */ + 0x0908,/* 0xa7 */ + 0x0909,/* 0xa8 */ + 0x090a,/* 0xa9 */ + 0x090b,/* 0xaa */ + 0x090e,/* 0xab */ + 0x090f,/* 0xac */ + 0x0910,/* 0xad */ + 0x090d,/* 0xae */ + 0x0912,/* 0xaf */ + 0x0913,/* 0xb0 */ + 0x0914,/* 0xb1 */ + 0x0911,/* 0xb2 */ + 0x0915,/* 0xb3 */ + 0x0916,/* 0xb4 */ + 0x0917,/* 0xb5 */ + 0x0918,/* 0xb6 */ + 0x0919,/* 0xb7 */ + 0x091a,/* 0xb8 */ + 0x091b,/* 0xb9 */ + 0x091c,/* 0xba */ + 0x091d,/* 0xbb */ + 0x091e,/* 0xbc */ + 0x091f,/* 0xbd */ + 0x0920,/* 0xbe */ + 0x0921,/* 0xbf */ + 0x0922,/* 0xc0 */ + 0x0923,/* 0xc1 */ + 0x0924,/* 0xc2 */ + 0x0925,/* 0xc3 */ + 0x0926,/* 0xc4 */ + 0x0927,/* 0xc5 */ + 0x0928,/* 0xc6 */ + 0x0929,/* 0xc7 */ + 0x092a,/* 0xc8 */ + 0x092b,/* 0xc9 */ + 0x092c,/* 0xca */ + 0x092d,/* 0xcb */ + 0x092e,/* 0xcc */ + 0x092f,/* 0xcd */ + 0x095f,/* 0xce */ + 0x0930,/* 0xcf */ + 0x0931,/* 0xd0 */ + 0x0932,/* 0xd1 */ + 0x0933,/* 0xd2 */ + 0x0934,/* 0xd3 */ + 0x0935,/* 0xd4 */ + 0x0936,/* 0xd5 */ + 0x0937,/* 0xd6 */ + 0x0938,/* 0xd7 */ + 0x0939,/* 0xd8 */ + 0x200D,/* 0xd9 */ + 0x093e,/* 0xda */ + 0x093f,/* 0xdb */ + 0x0940,/* 0xdc */ + 0x0941,/* 0xdd */ + 0x0942,/* 0xde */ + 0x0943,/* 0xdf */ + 0x0946,/* 0xe0 */ + 0x0947,/* 0xe1 */ + 0x0948,/* 0xe2 */ + 0x0945,/* 0xe3 */ + 0x094a,/* 0xe4 */ + 0x094b,/* 0xe5 */ + 0x094c,/* 0xe6 */ + 0x0949,/* 0xe7 */ + 0x094d,/* 0xe8 */ + 0x093c,/* 0xe9 */ + 0x0964,/* 0xea */ + 0xFFFF,/* 0xeb */ + 0xFFFF,/* 0xec */ + 0xFFFF,/* 0xed */ + 0xFFFF,/* 0xee */ + 0xFFFF,/* 0xef */ + 0xFFFF,/* 0xf0 */ + 0x0966,/* 0xf1 */ + 0x0967,/* 0xf2 */ + 0x0968,/* 0xf3 */ + 0x0969,/* 0xf4 */ + 0x096a,/* 0xf5 */ + 0x096b,/* 0xf6 */ + 0x096c,/* 0xf7 */ + 0x096d,/* 0xf8 */ + 0x096e,/* 0xf9 */ + 0x096f,/* 0xfa */ + 0xFFFF,/* 0xfb */ + 0xFFFF,/* 0xfc */ + 0xFFFF,/* 0xfd */ + 0xFFFF,/* 0xfe */ + 0xFFFF /* 0xff */ +}; + +static const uint16_t vowelSignESpecialCases[][2]={ + { 2 /*length of array*/ , 0 }, + { 0xA4 , 0x0904 }, +}; + +static const uint16_t nuktaSpecialCases[][2]={ + { 16 /*length of array*/ , 0 }, + { 0xA6 , 0x090c }, + { 0xEA , 0x093D }, + { 0xDF , 0x0944 }, + { 0xA1 , 0x0950 }, + { 0xb3 , 0x0958 }, + { 0xb4 , 0x0959 }, + { 0xb5 , 0x095a }, + { 0xba , 0x095b }, + { 0xbf , 0x095c }, + { 0xC0 , 0x095d }, + { 0xc9 , 0x095e }, + { 0xAA , 0x0960 }, + { 0xA7 , 0x0961 }, + { 0xDB , 0x0962 }, + { 0xDC , 0x0963 }, +}; + + +#define WRITE_TO_TARGET_FROM_U(args,offsets,source,target,targetLimit,targetByteUnit,err) UPRV_BLOCK_MACRO_BEGIN { \ + int32_t offset = (int32_t)(source - args->source-1); \ + /* write the targetUniChar to target */ \ + if(target < targetLimit){ \ + if(targetByteUnit <= 0xFF){ \ + *(target)++ = (uint8_t)(targetByteUnit); \ + if(offsets){ \ + *(offsets++) = offset; \ + } \ + }else{ \ + if (targetByteUnit > 0xFFFF) { \ + *(target)++ = (uint8_t)(targetByteUnit>>16); \ + if (offsets) { \ + --offset; \ + *(offsets++) = offset; \ + } \ + } \ + if (!(target < targetLimit)) { \ + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = \ + (uint8_t)(targetByteUnit >> 8); \ + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = \ + (uint8_t)targetByteUnit; \ + *err = U_BUFFER_OVERFLOW_ERROR; \ + } else { \ + *(target)++ = (uint8_t)(targetByteUnit>>8); \ + if(offsets){ \ + *(offsets++) = offset; \ + } \ + if(target < targetLimit){ \ + *(target)++ = (uint8_t) targetByteUnit; \ + if(offsets){ \ + *(offsets++) = offset ; \ + } \ + }else{ \ + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] =\ + (uint8_t) (targetByteUnit); \ + *err = U_BUFFER_OVERFLOW_ERROR; \ + } \ + } \ + } \ + }else{ \ + if (targetByteUnit & 0xFF0000) { \ + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = \ + (uint8_t) (targetByteUnit >>16); \ + } \ + if(targetByteUnit & 0xFF00){ \ + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = \ + (uint8_t) (targetByteUnit >>8); \ + } \ + args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = \ + (uint8_t) (targetByteUnit); \ + *err = U_BUFFER_OVERFLOW_ERROR; \ + } \ +} UPRV_BLOCK_MACRO_END + +/* Rules: + * Explicit Halant : + * + + * Soft Halant : + * + + */ +static void U_CALLCONV +UConverter_fromUnicode_ISCII_OFFSETS_LOGIC( + UConverterFromUnicodeArgs * args, UErrorCode * err) { + const char16_t *source = args->source; + const char16_t *sourceLimit = args->sourceLimit; + unsigned char *target = (unsigned char *) args->target; + unsigned char *targetLimit = (unsigned char *) args->targetLimit; + int32_t* offsets = args->offsets; + uint32_t targetByteUnit = 0x0000; + UChar32 sourceChar = 0x0000; + UChar32 tempContextFromUnicode = 0x0000; /* For special handling of the Gurmukhi script. */ + UConverterDataISCII *converterData; + uint16_t newDelta=0; + uint16_t range = 0; + UBool deltaChanged = false; + + if ((args->converter == nullptr) || (args->targetLimit < args->target) || (args->sourceLimit < args->source)) { + *err = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + /* initialize data */ + converterData=(UConverterDataISCII*)args->converter->extraInfo; + newDelta=converterData->currentDeltaFromUnicode; + range = (uint16_t)(newDelta/DELTA); + + if ((sourceChar = args->converter->fromUChar32)!=0) { + goto getTrail; + } + + /*writing the char to the output stream */ + while (source < sourceLimit) { + /* Write the language code following LF only if LF is not the last character. */ + if (args->converter->fromUnicodeStatus == LF) { + targetByteUnit = ATR<<8; + targetByteUnit += (uint8_t) lookupInitialData[range].isciiLang; + args->converter->fromUnicodeStatus = 0x0000; + /* now append ATR and language code */ + WRITE_TO_TARGET_FROM_U(args,offsets,source,target,targetLimit,targetByteUnit,err); + if (U_FAILURE(*err)) { + break; + } + } + + sourceChar = *source++; + tempContextFromUnicode = converterData->contextCharFromUnicode; + + targetByteUnit = missingCharMarker; + + /*check if input is in ASCII and C0 control codes range*/ + if (sourceChar <= ASCII_END) { + args->converter->fromUnicodeStatus = sourceChar; + WRITE_TO_TARGET_FROM_U(args,offsets,source,target,targetLimit,sourceChar,err); + if (U_FAILURE(*err)) { + break; + } + continue; + } + switch (sourceChar) { + case ZWNJ: + /* contextChar has HALANT */ + if (converterData->contextCharFromUnicode) { + converterData->contextCharFromUnicode = 0x00; + targetByteUnit = ISCII_HALANT; + } else { + /* consume ZWNJ and continue */ + converterData->contextCharFromUnicode = 0x00; + continue; + } + break; + case ZWJ: + /* contextChar has HALANT */ + if (converterData->contextCharFromUnicode) { + targetByteUnit = ISCII_NUKTA; + } else { + targetByteUnit =ISCII_INV; + } + converterData->contextCharFromUnicode = 0x00; + break; + default: + /* is the sourceChar in the INDIC_RANGE? */ + if ((uint16_t)(INDIC_BLOCK_END-sourceChar) <= INDIC_RANGE) { + /* Danda and Double Danda are valid in Northern scripts.. since Unicode + * does not include these codepoints in all Northern scrips we need to + * filter them out + */ + if (sourceChar!= DANDA && sourceChar != DOUBLE_DANDA) { + /* find out to which block the souceChar belongs*/ + range =(uint16_t)((sourceChar-INDIC_BLOCK_BEGIN)/DELTA); + newDelta =(uint16_t)(range*DELTA); + + /* Now are we in the same block as the previous? */ + if (newDelta!= converterData->currentDeltaFromUnicode || converterData->isFirstBuffer) { + converterData->currentDeltaFromUnicode = newDelta; + converterData->currentMaskFromUnicode = lookupInitialData[range].maskEnum; + deltaChanged =true; + converterData->isFirstBuffer=false; + } + + if (converterData->currentDeltaFromUnicode == PNJ_DELTA) { + if (sourceChar == PNJ_TIPPI) { + /* Make sure Tippi is converted to Bindi. */ + sourceChar = PNJ_BINDI; + } else if (sourceChar == PNJ_ADHAK) { + /* This is for consonant cluster handling. */ + converterData->contextCharFromUnicode = PNJ_ADHAK; + } + + } + /* Normalize all Indic codepoints to Devanagari and map them to ISCII */ + /* now subtract the new delta from sourceChar*/ + sourceChar -= converterData->currentDeltaFromUnicode; + } + + /* get the target byte unit */ + targetByteUnit=fromUnicodeTable[(uint8_t)sourceChar]; + + /* is the code point valid in current script? */ + if ((validityTable[(uint8_t)sourceChar] & converterData->currentMaskFromUnicode)==0) { + /* Vocallic RR is assigned in ISCII Telugu and Unicode */ + if (converterData->currentDeltaFromUnicode!=(TELUGU_DELTA) || sourceChar!=VOCALLIC_RR) { + targetByteUnit=missingCharMarker; + } + } + + if (deltaChanged) { + /* we are in a script block which is different than + * previous sourceChar's script block write ATR and language codes + */ + uint32_t temp=0; + temp =(uint16_t)(ATR<<8); + temp += (uint16_t)((uint8_t) lookupInitialData[range].isciiLang); + /* reset */ + deltaChanged=false; + /* now append ATR and language code */ + WRITE_TO_TARGET_FROM_U(args,offsets,source,target,targetLimit,temp,err); + if (U_FAILURE(*err)) { + break; + } + } + + if (converterData->currentDeltaFromUnicode == PNJ_DELTA && (sourceChar + PNJ_DELTA) == PNJ_ADHAK) { + continue; + } + } + /* reset context char */ + converterData->contextCharFromUnicode = 0x00; + break; + } + if (converterData->currentDeltaFromUnicode == PNJ_DELTA && tempContextFromUnicode == PNJ_ADHAK && isPNJConsonant((sourceChar + PNJ_DELTA))) { + /* If the previous codepoint is Adhak and the current codepoint is a consonant, the targetByteUnit should be C + Halant + C. */ + /* reset context char */ + converterData->contextCharFromUnicode = 0x0000; + targetByteUnit = targetByteUnit << 16 | ISCII_HALANT << 8 | targetByteUnit; + /* write targetByteUnit to target */ + WRITE_TO_TARGET_FROM_U(args, offsets, source, target, targetLimit, targetByteUnit,err); + if (U_FAILURE(*err)) { + break; + } + } else if (targetByteUnit != missingCharMarker) { + if (targetByteUnit==ISCII_HALANT) { + converterData->contextCharFromUnicode = (char16_t)targetByteUnit; + } + /* write targetByteUnit to target*/ + WRITE_TO_TARGET_FROM_U(args,offsets,source,target,targetLimit,targetByteUnit,err); + if (U_FAILURE(*err)) { + break; + } + } else { + /* oops.. the code point is unassigned */ + /*check if the char is a First surrogate*/ + if (U16_IS_SURROGATE(sourceChar)) { + if (U16_IS_SURROGATE_LEAD(sourceChar)) { +getTrail: + /*look ahead to find the trail surrogate*/ + if (source < sourceLimit) { + /* test the following code unit */ + char16_t trail= (*source); + if (U16_IS_TRAIL(trail)) { + source++; + sourceChar=U16_GET_SUPPLEMENTARY(sourceChar, trail); + *err =U_INVALID_CHAR_FOUND; + /* convert this surrogate code point */ + /* exit this condition tree */ + } else { + /* this is an unmatched lead code unit (1st surrogate) */ + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + } + } else { + /* no more input */ + *err = U_ZERO_ERROR; + } + } else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + *err=U_ILLEGAL_CHAR_FOUND; + } + } else { + /* callback(unassigned) for a BMP code point */ + *err = U_INVALID_CHAR_FOUND; + } + + args->converter->fromUChar32=sourceChar; + break; + } + }/* end while(mySourceIndexsource = source; + args->target = (char*)target; +} + +static const uint16_t lookupTable[][2]={ + { ZERO, ZERO }, /*DEFAULT*/ + { ZERO, ZERO }, /*ROMAN*/ + { DEVANAGARI, DEV_MASK }, + { BENGALI, BNG_MASK }, + { TAMIL, TML_MASK }, + { TELUGU, KND_MASK }, + { BENGALI, BNG_MASK }, + { ORIYA, ORI_MASK }, + { KANNADA, KND_MASK }, + { MALAYALAM, MLM_MASK }, + { GUJARATI, GJR_MASK }, + { GURMUKHI, PNJ_MASK } +}; + +#define WRITE_TO_TARGET_TO_U(args,source,target,offsets,offset,targetUniChar,delta, err) UPRV_BLOCK_MACRO_BEGIN { \ + /* add offset to current Indic Block */ \ + if(targetUniChar>ASCII_END && \ + targetUniChar != ZWJ && \ + targetUniChar != ZWNJ && \ + targetUniChar != DANDA && \ + targetUniChar != DOUBLE_DANDA){ \ + \ + targetUniChar+=(uint16_t)(delta); \ + } \ + /* now write the targetUniChar */ \ + if(targettargetLimit){ \ + *(target)++ = (char16_t)targetUniChar; \ + if(offsets){ \ + *(offsets)++ = (int32_t)(offset); \ + } \ + }else{ \ + args->converter->UCharErrorBuffer[args->converter->UCharErrorBufferLength++] = \ + (char16_t)targetUniChar; \ + *err = U_BUFFER_OVERFLOW_ERROR; \ + } \ +} UPRV_BLOCK_MACRO_END + +#define GET_MAPPING(sourceChar,targetUniChar,data) UPRV_BLOCK_MACRO_BEGIN { \ + targetUniChar = toUnicodeTable[(sourceChar)] ; \ + /* is the code point valid in current script? */ \ + if(sourceChar> ASCII_END && \ + (validityTable[(targetUniChar & 0x7F)] & data->currentMaskToUnicode)==0){ \ + /* Vocallic RR is assigned in ISCII Telugu and Unicode */ \ + if(data->currentDeltaToUnicode!=(TELUGU_DELTA) || \ + targetUniChar!=VOCALLIC_RR){ \ + targetUniChar=missingCharMarker; \ + } \ + } \ +} UPRV_BLOCK_MACRO_END + +/*********** + * Rules for ISCII to Unicode converter + * ISCII is stateful encoding. To convert ISCII bytes to Unicode, + * which has both precomposed and decomposed forms characters + * pre-context and post-context need to be considered. + * + * Post context + * i) ATR : Attribute code is used to declare the font and script switching. + * Currently we only switch scripts and font codes consumed without generating an error + * ii) EXT : Extension code is used to declare switching to Sanskrit and for obscure, + * obsolete characters + * Pre context + * i) Halant: if preceded by a halant then it is a explicit halant + * ii) Nukta : + * a) if preceded by a halant then it is a soft halant + * b) if preceded by specific consonants and the ligatures have pre-composed + * characters in Unicode then convert to pre-composed characters + * iii) Danda: If Danda is preceded by a Danda then convert to Double Danda + * + */ + +static void U_CALLCONV +UConverter_toUnicode_ISCII_OFFSETS_LOGIC(UConverterToUnicodeArgs *args, UErrorCode* err) { + const char *source = ( char *) args->source; + char16_t *target = args->target; + const char *sourceLimit = args->sourceLimit; + const char16_t* targetLimit = args->targetLimit; + uint32_t targetUniChar = 0x0000; + uint8_t sourceChar = 0x0000; + UConverterDataISCII* data; + UChar32* toUnicodeStatus=nullptr; + UChar32 tempTargetUniChar = 0x0000; + char16_t* contextCharToUnicode= nullptr; + UBool found; + int i; + int offset = 0; + + if ((args->converter == nullptr) || (target < args->target) || (source < args->source)) { + *err = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + data = (UConverterDataISCII*)(args->converter->extraInfo); + contextCharToUnicode = &data->contextCharToUnicode; /* contains previous ISCII codepoint visited */ + toUnicodeStatus = (UChar32*)&args->converter->toUnicodeStatus;/* contains the mapping to Unicode of the above codepoint*/ + + while (U_SUCCESS(*err) && sourcecurrentDeltaToUnicode = (uint16_t)(lookupTable[sourceChar & 0x0F][0] * DELTA); + data->currentMaskToUnicode = (MaskEnum)lookupTable[sourceChar & 0x0F][1]; + } else if (sourceChar==DEF) { + /* switch back to default */ + data->currentDeltaToUnicode = data->defDeltaToUnicode; + data->currentMaskToUnicode = data->defMaskToUnicode; + } else { + if ((sourceChar >= 0x21 && sourceChar <= 0x3F)) { + /* these are display codes consume and continue */ + } else { + *err =U_ILLEGAL_CHAR_FOUND; + /* reset */ + *contextCharToUnicode=NO_CHAR_MARKER; + goto CALLBACK; + } + } + + /* reset */ + *contextCharToUnicode=NO_CHAR_MARKER; + + continue; + + } else if (*contextCharToUnicode==EXT) { + /* check if sourceChar is in 0xA1-0xEE range */ + if ((uint8_t) (EXT_RANGE_END - sourceChar) <= (EXT_RANGE_END - EXT_RANGE_BEGIN)) { + /* We currently support only Anudatta and Devanagari abbreviation sign */ + if (sourceChar==0xBF || sourceChar == 0xB8) { + targetUniChar = (sourceChar==0xBF) ? DEV_ABBR_SIGN : DEV_ANUDATTA; + + /* find out if the mapping is valid in this state */ + if (validityTable[(uint8_t)targetUniChar] & data->currentMaskToUnicode) { + *contextCharToUnicode= NO_CHAR_MARKER; + + /* Write the previous toUnicodeStatus, this was delayed to handle consonant clustering for Gurmukhi script. */ + if (data->prevToUnicodeStatus) { + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -1),data->prevToUnicodeStatus,0,err); + data->prevToUnicodeStatus = 0x0000; + } + /* write to target */ + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -2),targetUniChar,data->currentDeltaToUnicode,err); + + continue; + } + } + /* byte unit is unassigned */ + targetUniChar = missingCharMarker; + *err= U_INVALID_CHAR_FOUND; + } else { + /* only 0xA1 - 0xEE are legal after EXT char */ + *contextCharToUnicode= NO_CHAR_MARKER; + *err = U_ILLEGAL_CHAR_FOUND; + } + goto CALLBACK; + } else if (*contextCharToUnicode==ISCII_INV) { + if (sourceChar==ISCII_HALANT) { + targetUniChar = 0x0020; /* replace with space according to Indic FAQ */ + } else { + targetUniChar = ZWJ; + } + + /* Write the previous toUnicodeStatus, this was delayed to handle consonant clustering for Gurmukhi script. */ + if (data->prevToUnicodeStatus) { + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -1),data->prevToUnicodeStatus,0,err); + data->prevToUnicodeStatus = 0x0000; + } + /* write to target */ + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -2),targetUniChar,data->currentDeltaToUnicode,err); + /* reset */ + *contextCharToUnicode=NO_CHAR_MARKER; + } + + /* look at the pre-context and perform special processing */ + switch (sourceChar) { + case ISCII_INV: + case EXT: + case ATR: + *contextCharToUnicode = (char16_t)sourceChar; + + if (*toUnicodeStatus != missingCharMarker) { + /* Write the previous toUnicodeStatus, this was delayed to handle consonant clustering for Gurmukhi script. */ + if (data->prevToUnicodeStatus) { + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -1),data->prevToUnicodeStatus,0,err); + data->prevToUnicodeStatus = 0x0000; + } + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -2),*toUnicodeStatus,data->currentDeltaToUnicode,err); + *toUnicodeStatus = missingCharMarker; + } + continue; + case ISCII_DANDA: + /* handle double danda*/ + if (*contextCharToUnicode== ISCII_DANDA) { + targetUniChar = DOUBLE_DANDA; + /* clear the context */ + *contextCharToUnicode = NO_CHAR_MARKER; + *toUnicodeStatus = missingCharMarker; + } else { + GET_MAPPING(sourceChar,targetUniChar,data); + *contextCharToUnicode = sourceChar; + } + break; + case ISCII_HALANT: + /* handle explicit halant */ + if (*contextCharToUnicode == ISCII_HALANT) { + targetUniChar = ZWNJ; + /* clear the context */ + *contextCharToUnicode = NO_CHAR_MARKER; + } else { + GET_MAPPING(sourceChar,targetUniChar,data); + *contextCharToUnicode = sourceChar; + } + break; + case 0x0A: + case 0x0D: + data->resetToDefaultToUnicode = true; + GET_MAPPING(sourceChar,targetUniChar,data) + ; + *contextCharToUnicode = sourceChar; + break; + + case ISCII_VOWEL_SIGN_E: + i=1; + found=false; + for (; icurrentMaskToUnicode) { + /*targetUniChar += data->currentDeltaToUnicode ;*/ + *contextCharToUnicode= NO_CHAR_MARKER; + *toUnicodeStatus = missingCharMarker; + break; + } + } + GET_MAPPING(sourceChar,targetUniChar,data); + *contextCharToUnicode = sourceChar; + break; + + case ISCII_NUKTA: + /* handle soft halant */ + if (*contextCharToUnicode == ISCII_HALANT) { + targetUniChar = ZWJ; + /* clear the context */ + *contextCharToUnicode = NO_CHAR_MARKER; + break; + } else if (data->currentDeltaToUnicode == PNJ_DELTA && data->contextCharToUnicode == 0xc0) { + /* Write the previous toUnicodeStatus, this was delayed to handle consonant clustering for Gurmukhi script. */ + if (data->prevToUnicodeStatus) { + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -1),data->prevToUnicodeStatus,0,err); + data->prevToUnicodeStatus = 0x0000; + } + /* We got here because ISCII_NUKTA was preceded by 0xc0 and we are converting Gurmukhi. + * In that case we must convert (0xc0 0xe9) to (\u0a5c\u0a4d\u0a39). + */ + targetUniChar = PNJ_RRA; + WRITE_TO_TARGET_TO_U(args, source, target, args->offsets, (source-args->source)-2, targetUniChar, 0, err); + if (U_SUCCESS(*err)) { + targetUniChar = PNJ_SIGN_VIRAMA; + WRITE_TO_TARGET_TO_U(args, source, target, args->offsets, (source-args->source)-2, targetUniChar, 0, err); + if (U_SUCCESS(*err)) { + targetUniChar = PNJ_HA; + WRITE_TO_TARGET_TO_U(args, source, target, args->offsets, (source-args->source)-2, targetUniChar, 0, err); + } else { + args->converter->UCharErrorBuffer[args->converter->UCharErrorBufferLength++]= PNJ_HA; + } + } else { + args->converter->UCharErrorBuffer[args->converter->UCharErrorBufferLength++]= PNJ_SIGN_VIRAMA; + args->converter->UCharErrorBuffer[args->converter->UCharErrorBufferLength++]= PNJ_HA; + } + *toUnicodeStatus = missingCharMarker; + data->contextCharToUnicode = NO_CHAR_MARKER; + continue; + } else { + /* try to handle + ISCII_NUKTA special mappings */ + i=1; + found =false; + for (; icurrentMaskToUnicode) { + /*targetUniChar += data->currentDeltaToUnicode ;*/ + *contextCharToUnicode= NO_CHAR_MARKER; + *toUnicodeStatus = missingCharMarker; + if (data->currentDeltaToUnicode == PNJ_DELTA) { + /* Write the previous toUnicodeStatus, this was delayed to handle consonant clustering for Gurmukhi script. */ + if (data->prevToUnicodeStatus) { + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -1),data->prevToUnicodeStatus,0,err); + data->prevToUnicodeStatus = 0x0000; + } + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -2),targetUniChar,data->currentDeltaToUnicode,err); + continue; + } + break; + } + /* else fall through to default */ + } + /* else fall through to default */ + U_FALLTHROUGH; + } + default:GET_MAPPING(sourceChar,targetUniChar,data) + ; + *contextCharToUnicode = sourceChar; + break; + } + + if (*toUnicodeStatus != missingCharMarker) { + /* Check to make sure that consonant clusters are handled correct for Gurmukhi script. */ + if (data->currentDeltaToUnicode == PNJ_DELTA && data->prevToUnicodeStatus != 0 && isPNJConsonant(data->prevToUnicodeStatus) && + (*toUnicodeStatus + PNJ_DELTA) == PNJ_SIGN_VIRAMA && ((UChar32)(targetUniChar + PNJ_DELTA) == data->prevToUnicodeStatus)) { + /* Consonant clusters C + HALANT + C should be encoded as ADHAK + C */ + offset = (int)(source-args->source - 3); + tempTargetUniChar = PNJ_ADHAK; /* This is necessary to avoid some compiler warnings. */ + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,offset,tempTargetUniChar,0,err); + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,offset,data->prevToUnicodeStatus,0,err); + data->prevToUnicodeStatus = 0x0000; /* reset the previous unicode code point */ + *toUnicodeStatus = missingCharMarker; + continue; + } else { + /* Write the previous toUnicodeStatus, this was delayed to handle consonant clustering for Gurmukhi script. */ + if (data->prevToUnicodeStatus) { + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -1),data->prevToUnicodeStatus,0,err); + data->prevToUnicodeStatus = 0x0000; + } + /* Check to make sure that Bindi and Tippi are handled correctly for Gurmukhi script. + * If 0xA2 is preceded by a codepoint in the PNJ_BINDI_TIPPI_SET then the target codepoint should be Tippi instead of Bindi. + */ + if (data->currentDeltaToUnicode == PNJ_DELTA && (targetUniChar + PNJ_DELTA) == PNJ_BINDI && isPNJBindiTippi((*toUnicodeStatus + PNJ_DELTA))) { + targetUniChar = PNJ_TIPPI - PNJ_DELTA; + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -2),*toUnicodeStatus,PNJ_DELTA,err); + } else if (data->currentDeltaToUnicode == PNJ_DELTA && (targetUniChar + PNJ_DELTA) == PNJ_SIGN_VIRAMA && isPNJConsonant((*toUnicodeStatus + PNJ_DELTA))) { + /* Store the current toUnicodeStatus code point for later handling of consonant cluster in Gurmukhi. */ + data->prevToUnicodeStatus = *toUnicodeStatus + PNJ_DELTA; + } else { + /* write the previously mapped codepoint */ + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -2),*toUnicodeStatus,data->currentDeltaToUnicode,err); + } + } + *toUnicodeStatus = missingCharMarker; + } + + if (targetUniChar != missingCharMarker) { + /* now save the targetUniChar for delayed write */ + *toUnicodeStatus = (char16_t) targetUniChar; + if (data->resetToDefaultToUnicode) { + data->currentDeltaToUnicode = data->defDeltaToUnicode; + data->currentMaskToUnicode = data->defMaskToUnicode; + data->resetToDefaultToUnicode=false; + } + } else { + + /* we reach here only if targetUniChar == missingCharMarker + * so assign codes to reason and err + */ + *err = U_INVALID_CHAR_FOUND; +CALLBACK: + args->converter->toUBytes[0] = (uint8_t) sourceChar; + args->converter->toULength = 1; + break; + } + + } else { + *err =U_BUFFER_OVERFLOW_ERROR; + break; + } + } + + if (U_SUCCESS(*err) && args->flush && source == sourceLimit) { + /* end of the input stream */ + UConverter *cnv = args->converter; + + if (*contextCharToUnicode==ATR || *contextCharToUnicode==EXT || *contextCharToUnicode==ISCII_INV) { + /* set toUBytes[] */ + cnv->toUBytes[0] = (uint8_t)*contextCharToUnicode; + cnv->toULength = 1; + + /* avoid looping on truncated sequences */ + *contextCharToUnicode = NO_CHAR_MARKER; + } else { + cnv->toULength = 0; + } + + if (*toUnicodeStatus != missingCharMarker) { + /* output a remaining target character */ + WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source - args->source -1),*toUnicodeStatus,data->currentDeltaToUnicode,err); + *toUnicodeStatus = missingCharMarker; + } + } + + args->target = target; + args->source = source; +} + +/* structure for SafeClone calculations */ +struct cloneISCIIStruct { + UConverter cnv; + UConverterDataISCII mydata; +}; + +static UConverter * U_CALLCONV +_ISCII_SafeClone(const UConverter *cnv, + void *stackBuffer, + int32_t *pBufferSize, + UErrorCode *status) +{ + struct cloneISCIIStruct * localClone; + int32_t bufferSizeNeeded = sizeof(struct cloneISCIIStruct); + + if (U_FAILURE(*status)) { + return 0; + } + + if (*pBufferSize == 0) { /* 'preflighting' request - set needed size into *pBufferSize */ + *pBufferSize = bufferSizeNeeded; + return 0; + } + + localClone = (struct cloneISCIIStruct *)stackBuffer; + /* ucnv.c/ucnv_safeClone() copied the main UConverter already */ + + uprv_memcpy(&localClone->mydata, cnv->extraInfo, sizeof(UConverterDataISCII)); + localClone->cnv.extraInfo = &localClone->mydata; + localClone->cnv.isExtraLocal = true; + + return &localClone->cnv; +} + +static void U_CALLCONV +_ISCIIGetUnicodeSet(const UConverter *cnv, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode) +{ + (void)cnv; + (void)which; + (void)pErrorCode; + int32_t idx, script; + uint8_t mask; + + /* Since all ISCII versions allow switching to other ISCII + scripts, we add all roundtrippable characters to this set. */ + sa->addRange(sa->set, 0, ASCII_END); + for (script = DEVANAGARI; script <= MALAYALAM; script++) { + mask = (uint8_t)(lookupInitialData[script].maskEnum); + for (idx = 0; idx < DELTA; idx++) { + /* added check for TELUGU character */ + if ((validityTable[idx] & mask) || (script==TELUGU && idx==0x31)) { + sa->add(sa->set, idx + (script * DELTA) + INDIC_BLOCK_BEGIN); + } + } + } + sa->add(sa->set, DANDA); + sa->add(sa->set, DOUBLE_DANDA); + sa->add(sa->set, ZWNJ); + sa->add(sa->set, ZWJ); +} +U_CDECL_END +static const UConverterImpl _ISCIIImpl={ + + UCNV_ISCII, + + nullptr, + nullptr, + + _ISCIIOpen, + _ISCIIClose, + _ISCIIReset, + + UConverter_toUnicode_ISCII_OFFSETS_LOGIC, + UConverter_toUnicode_ISCII_OFFSETS_LOGIC, + UConverter_fromUnicode_ISCII_OFFSETS_LOGIC, + UConverter_fromUnicode_ISCII_OFFSETS_LOGIC, + nullptr, + + nullptr, + _ISCIIgetName, + nullptr, + _ISCII_SafeClone, + _ISCIIGetUnicodeSet, + nullptr, + nullptr +}; + +static const UConverterStaticData _ISCIIStaticData={ + sizeof(UConverterStaticData), + "ISCII", + 0, + UCNV_IBM, + UCNV_ISCII, + 1, + 4, + { 0x1a, 0, 0, 0 }, + 0x1, + false, + false, + 0x0, + 0x0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 }, /* reserved */ + +}; + +const UConverterSharedData _ISCIIData= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_ISCIIStaticData, &_ISCIIImpl); + +#endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */ diff --git a/intl/icu/source/common/ucnvlat1.cpp b/intl/icu/source/common/ucnvlat1.cpp new file mode 100644 index 0000000000..0920688526 --- /dev/null +++ b/intl/icu/source/common/ucnvlat1.cpp @@ -0,0 +1,756 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2000-2015, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* file name: ucnvlat1.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2000feb07 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/uset.h" +#include "unicode/utf8.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" +#include "ustr_imp.h" + +/* control optimizations according to the platform */ +#define LATIN1_UNROLL_FROM_UNICODE 1 + +/* ISO 8859-1 --------------------------------------------------------------- */ + +/* This is a table-less and callback-less version of ucnv_MBCSSingleToBMPWithOffsets(). */ +U_CDECL_BEGIN +static void U_CALLCONV +_Latin1ToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + const uint8_t *source; + char16_t *target; + int32_t targetCapacity, length; + int32_t *offsets; + + int32_t sourceIndex; + + /* set up the local pointers */ + source=(const uint8_t *)pArgs->source; + target=pArgs->target; + targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); + offsets=pArgs->offsets; + + sourceIndex=0; + + /* + * since the conversion here is 1:1 char16_t:uint8_t, we need only one counter + * for the minimum of the sourceLength and targetCapacity + */ + length=(int32_t)((const uint8_t *)pArgs->sourceLimit-source); + if(length<=targetCapacity) { + targetCapacity=length; + } else { + /* target will be full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + length=targetCapacity; + } + + if(targetCapacity>=8) { + /* This loop is unrolled for speed and improved pipelining. */ + int32_t count, loops; + + loops=count=targetCapacity>>3; + length=targetCapacity&=0x7; + do { + target[0]=source[0]; + target[1]=source[1]; + target[2]=source[2]; + target[3]=source[3]; + target[4]=source[4]; + target[5]=source[5]; + target[6]=source[6]; + target[7]=source[7]; + target+=8; + source+=8; + } while(--count>0); + + if(offsets!=nullptr) { + do { + offsets[0]=sourceIndex++; + offsets[1]=sourceIndex++; + offsets[2]=sourceIndex++; + offsets[3]=sourceIndex++; + offsets[4]=sourceIndex++; + offsets[5]=sourceIndex++; + offsets[6]=sourceIndex++; + offsets[7]=sourceIndex++; + offsets+=8; + } while(--loops>0); + } + } + + /* conversion loop */ + while(targetCapacity>0) { + *target++=*source++; + --targetCapacity; + } + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + + /* set offsets */ + if(offsets!=nullptr) { + while(length>0) { + *offsets++=sourceIndex++; + --length; + } + pArgs->offsets=offsets; + } +} + +/* This is a table-less and callback-less version of ucnv_MBCSSingleGetNextUChar(). */ +static UChar32 U_CALLCONV +_Latin1GetNextUChar(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + const uint8_t *source=(const uint8_t *)pArgs->source; + if(source<(const uint8_t *)pArgs->sourceLimit) { + pArgs->source=(const char *)(source+1); + return *source; + } + + /* no output because of empty input */ + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0xffff; +} + +/* This is a table-less version of ucnv_MBCSSingleFromBMPWithOffsets(). */ +static void U_CALLCONV +_Latin1FromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const char16_t *source, *sourceLimit; + uint8_t *target, *oldTarget; + int32_t targetCapacity, length; + int32_t *offsets; + + UChar32 cp; + char16_t c, max; + + int32_t sourceIndex; + + /* set up the local pointers */ + cnv=pArgs->converter; + source=pArgs->source; + sourceLimit=pArgs->sourceLimit; + target=oldTarget=(uint8_t *)pArgs->target; + targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); + offsets=pArgs->offsets; + + if(cnv->sharedData==&_Latin1Data) { + max=0xff; /* Latin-1 */ + } else { + max=0x7f; /* US-ASCII */ + } + + /* get the converter state from UConverter */ + cp=cnv->fromUChar32; + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex= cp==0 ? 0 : -1; + + /* + * since the conversion here is 1:1 char16_t:uint8_t, we need only one counter + * for the minimum of the sourceLength and targetCapacity + */ + length=(int32_t)(sourceLimit-source); + if(length0) { + goto getTrail; + } + +#if LATIN1_UNROLL_FROM_UNICODE + /* unroll the loop with the most common case */ + if(targetCapacity>=16) { + int32_t count, loops; + char16_t u, oredChars; + + loops=count=targetCapacity>>4; + do { + oredChars=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + oredChars|=u=*source++; + *target++=(uint8_t)u; + + /* were all 16 entries really valid? */ + if(oredChars>max) { + /* no, return to the first of these 16 */ + source-=16; + target-=16; + break; + } + } while(--count>0); + count=loops-count; + targetCapacity-=16*count; + + if(offsets!=nullptr) { + oldTarget+=16*count; + while(count>0) { + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + --count; + } + } + } +#endif + + /* conversion loop */ + c=0; + while(targetCapacity>0 && (c=*source++)<=max) { + /* convert the Unicode code point */ + *target++=(uint8_t)c; + --targetCapacity; + } + + if(c>max) { + cp=c; + if(!U_IS_SURROGATE(cp)) { + /* callback(unassigned) */ + } else if(U_IS_SURROGATE_LEAD(cp)) { +getTrail: + if(sourcefromUChar32=cp; + goto noMoreInput; + } + } else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + } + + *pErrorCode= U_IS_SURROGATE(cp) ? U_ILLEGAL_CHAR_FOUND : U_INVALID_CHAR_FOUND; + cnv->fromUChar32=cp; + } +noMoreInput: + + /* set offsets since the start */ + if(offsets!=nullptr) { + size_t count=target-oldTarget; + while(count>0) { + *offsets++=sourceIndex++; + --count; + } + } + + if(U_SUCCESS(*pErrorCode) && source=(uint8_t *)pArgs->targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + + /* write back the updated pointers */ + pArgs->source=source; + pArgs->target=(char *)target; + pArgs->offsets=offsets; +} + +/* Convert UTF-8 to Latin-1. Adapted from ucnv_SBCSFromUTF8(). */ +static void U_CALLCONV +ucnv_Latin1FromUTF8(UConverterFromUnicodeArgs *pFromUArgs, + UConverterToUnicodeArgs *pToUArgs, + UErrorCode *pErrorCode) { + UConverter *utf8; + const uint8_t *source, *sourceLimit; + uint8_t *target; + int32_t targetCapacity; + + UChar32 c; + uint8_t b, t1; + + /* set up the local pointers */ + utf8=pToUArgs->converter; + source=(uint8_t *)pToUArgs->source; + sourceLimit=(uint8_t *)pToUArgs->sourceLimit; + target=(uint8_t *)pFromUArgs->target; + targetCapacity=(int32_t)(pFromUArgs->targetLimit-pFromUArgs->target); + + /* get the converter state from the UTF-8 UConverter */ + if (utf8->toULength > 0) { + c=(UChar32)utf8->toUnicodeStatus; + } else { + c = 0; + } + if(c!=0 && source=0xc2 && c<=0xc3 && (t1=(uint8_t)(*source-0x80)) <= 0x3f) { + ++source; + *target++=(uint8_t)(((c&3)<<6)|t1); + --targetCapacity; + + utf8->toUnicodeStatus=0; + utf8->toULength=0; + } else { + /* complicated, illegal or unmappable input: fall back to the pivoting implementation */ + *pErrorCode=U_USING_DEFAULT_WARNING; + return; + } + } + + /* + * Make sure that the last byte sequence before sourceLimit is complete + * or runs into a lead byte. + * In the conversion loop compare source with sourceLimit only once + * per multi-byte character. + * For Latin-1, adjust sourceLimit only for 1 trail byte because + * the conversion loop handles at most 2-byte sequences. + */ + if(source0) { + b=*source++; + if(U8_IS_SINGLE(b)) { + /* convert ASCII */ + *target++=(uint8_t)b; + --targetCapacity; + } else if( /* handle U+0080..U+00FF inline */ + b>=0xc2 && b<=0xc3 && + (t1=(uint8_t)(*source-0x80)) <= 0x3f + ) { + ++source; + *target++=(uint8_t)(((b&3)<<6)|t1); + --targetCapacity; + } else { + /* complicated, illegal or unmappable input: fall back to the pivoting implementation */ + pToUArgs->source=(char *)(source-1); + pFromUArgs->target=(char *)target; + *pErrorCode=U_USING_DEFAULT_WARNING; + return; + } + } else { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + + /* + * The sourceLimit may have been adjusted before the conversion loop + * to stop before a truncated sequence. + * If so, then collect the truncated sequence now. + * For Latin-1, there is at most exactly one lead byte because of the + * smaller sourceLimit adjustment logic. + */ + if(U_SUCCESS(*pErrorCode) && source<(sourceLimit=(uint8_t *)pToUArgs->sourceLimit)) { + utf8->toUnicodeStatus=utf8->toUBytes[0]=b=*source++; + utf8->toULength=1; + utf8->mode=U8_COUNT_BYTES(b); + } + + /* write back the updated pointers */ + pToUArgs->source=(char *)source; + pFromUArgs->target=(char *)target; +} + +static void U_CALLCONV +_Latin1GetUnicodeSet(const UConverter *cnv, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode) { + (void)cnv; + (void)which; + (void)pErrorCode; + sa->addRange(sa->set, 0, 0xff); +} +U_CDECL_END + + +static const UConverterImpl _Latin1Impl={ + UCNV_LATIN_1, + + nullptr, + nullptr, + + nullptr, + nullptr, + nullptr, + + _Latin1ToUnicodeWithOffsets, + _Latin1ToUnicodeWithOffsets, + _Latin1FromUnicodeWithOffsets, + _Latin1FromUnicodeWithOffsets, + _Latin1GetNextUChar, + + nullptr, + nullptr, + nullptr, + nullptr, + _Latin1GetUnicodeSet, + + nullptr, + ucnv_Latin1FromUTF8 +}; + +static const UConverterStaticData _Latin1StaticData={ + sizeof(UConverterStaticData), + "ISO-8859-1", + 819, UCNV_IBM, UCNV_LATIN_1, 1, 1, + { 0x1a, 0, 0, 0 }, 1, false, false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + +const UConverterSharedData _Latin1Data= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_Latin1StaticData, &_Latin1Impl); + +/* US-ASCII ----------------------------------------------------------------- */ + +U_CDECL_BEGIN +/* This is a table-less version of ucnv_MBCSSingleToBMPWithOffsets(). */ +static void U_CALLCONV +_ASCIIToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + const uint8_t *source, *sourceLimit; + char16_t *target, *oldTarget; + int32_t targetCapacity, length; + int32_t *offsets; + + int32_t sourceIndex; + + uint8_t c; + + /* set up the local pointers */ + source=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + target=oldTarget=pArgs->target; + targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); + offsets=pArgs->offsets; + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex=0; + + /* + * since the conversion here is 1:1 char16_t:uint8_t, we need only one counter + * for the minimum of the sourceLength and targetCapacity + */ + length=(int32_t)(sourceLimit-source); + if(length=8) { + /* This loop is unrolled for speed and improved pipelining. */ + int32_t count, loops; + char16_t oredChars; + + loops=count=targetCapacity>>3; + do { + oredChars=target[0]=source[0]; + oredChars|=target[1]=source[1]; + oredChars|=target[2]=source[2]; + oredChars|=target[3]=source[3]; + oredChars|=target[4]=source[4]; + oredChars|=target[5]=source[5]; + oredChars|=target[6]=source[6]; + oredChars|=target[7]=source[7]; + + /* were all 16 entries really valid? */ + if(oredChars>0x7f) { + /* no, return to the first of these 16 */ + break; + } + source+=8; + target+=8; + } while(--count>0); + count=loops-count; + targetCapacity-=count*8; + + if(offsets!=nullptr) { + oldTarget+=count*8; + while(count>0) { + offsets[0]=sourceIndex++; + offsets[1]=sourceIndex++; + offsets[2]=sourceIndex++; + offsets[3]=sourceIndex++; + offsets[4]=sourceIndex++; + offsets[5]=sourceIndex++; + offsets[6]=sourceIndex++; + offsets[7]=sourceIndex++; + offsets+=8; + --count; + } + } + } + + /* conversion loop */ + c=0; + while(targetCapacity>0 && (c=*source++)<=0x7f) { + *target++=c; + --targetCapacity; + } + + if(c>0x7f) { + /* callback(illegal); copy the current bytes to toUBytes[] */ + UConverter *cnv=pArgs->converter; + cnv->toUBytes[0]=c; + cnv->toULength=1; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } else if(source=pArgs->targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + + /* set offsets since the start */ + if(offsets!=nullptr) { + size_t count=target-oldTarget; + while(count>0) { + *offsets++=sourceIndex++; + --count; + } + } + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + pArgs->offsets=offsets; +} + +/* This is a table-less version of ucnv_MBCSSingleGetNextUChar(). */ +static UChar32 U_CALLCONV +_ASCIIGetNextUChar(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + const uint8_t *source; + uint8_t b; + + source=(const uint8_t *)pArgs->source; + if(source<(const uint8_t *)pArgs->sourceLimit) { + b=*source++; + pArgs->source=(const char *)source; + if(b<=0x7f) { + return b; + } else { + UConverter *cnv=pArgs->converter; + cnv->toUBytes[0]=b; + cnv->toULength=1; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + return 0xffff; + } + } + + /* no output because of empty input */ + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0xffff; +} + +/* "Convert" UTF-8 to US-ASCII: Validate and copy. */ +static void U_CALLCONV +ucnv_ASCIIFromUTF8(UConverterFromUnicodeArgs *pFromUArgs, + UConverterToUnicodeArgs *pToUArgs, + UErrorCode *pErrorCode) { + const uint8_t *source, *sourceLimit; + uint8_t *target; + int32_t targetCapacity, length; + + uint8_t c; + + if(pToUArgs->converter->toULength > 0) { + /* no handling of partial UTF-8 characters here, fall back to pivoting */ + *pErrorCode=U_USING_DEFAULT_WARNING; + return; + } + + /* set up the local pointers */ + source=(const uint8_t *)pToUArgs->source; + sourceLimit=(const uint8_t *)pToUArgs->sourceLimit; + target=(uint8_t *)pFromUArgs->target; + targetCapacity=(int32_t)(pFromUArgs->targetLimit-pFromUArgs->target); + + /* + * since the conversion here is 1:1 uint8_t:uint8_t, we need only one counter + * for the minimum of the sourceLength and targetCapacity + */ + length=(int32_t)(sourceLimit-source); + if(length=16) { + int32_t count, loops; + uint8_t oredChars; + + loops=count=targetCapacity>>4; + do { + oredChars=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + oredChars|=*target++=*source++; + + /* were all 16 entries really valid? */ + if(oredChars>0x7f) { + /* no, return to the first of these 16 */ + source-=16; + target-=16; + break; + } + } while(--count>0); + count=loops-count; + targetCapacity-=16*count; + } + + /* conversion loop */ + c=0; + while(targetCapacity>0 && (c=*source)<=0x7f) { + ++source; + *target++=c; + --targetCapacity; + } + + if(c>0x7f) { + /* non-ASCII character, handle in standard converter */ + *pErrorCode=U_USING_DEFAULT_WARNING; + } else if(source=(const uint8_t *)pFromUArgs->targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + + /* write back the updated pointers */ + pToUArgs->source=(const char *)source; + pFromUArgs->target=(char *)target; +} + +static void U_CALLCONV +_ASCIIGetUnicodeSet(const UConverter *cnv, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode) { + (void)cnv; + (void)which; + (void)pErrorCode; + sa->addRange(sa->set, 0, 0x7f); +} +U_CDECL_END + +static const UConverterImpl _ASCIIImpl={ + UCNV_US_ASCII, + + nullptr, + nullptr, + + nullptr, + nullptr, + nullptr, + + _ASCIIToUnicodeWithOffsets, + _ASCIIToUnicodeWithOffsets, + _Latin1FromUnicodeWithOffsets, + _Latin1FromUnicodeWithOffsets, + _ASCIIGetNextUChar, + + nullptr, + nullptr, + nullptr, + nullptr, + _ASCIIGetUnicodeSet, + + nullptr, + ucnv_ASCIIFromUTF8 +}; + +static const UConverterStaticData _ASCIIStaticData={ + sizeof(UConverterStaticData), + "US-ASCII", + 367, UCNV_IBM, UCNV_US_ASCII, 1, 1, + { 0x1a, 0, 0, 0 }, 1, false, false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + +const UConverterSharedData _ASCIIData= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_ASCIIStaticData, &_ASCIIImpl); + +#endif diff --git a/intl/icu/source/common/ucnvmbcs.cpp b/intl/icu/source/common/ucnvmbcs.cpp new file mode 100644 index 0000000000..d760603980 --- /dev/null +++ b/intl/icu/source/common/ucnvmbcs.cpp @@ -0,0 +1,5723 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2000-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: ucnvmbcs.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2000jul03 +* created by: Markus W. Scherer +* +* The current code in this file replaces the previous implementation +* of conversion code from multi-byte codepages to Unicode and back. +* This implementation supports the following: +* - legacy variable-length codepages with up to 4 bytes per character +* - all Unicode code points (up to 0x10ffff) +* - efficient distinction of unassigned vs. illegal byte sequences +* - it is possible in fromUnicode() to directly deal with simple +* stateful encodings (used for EBCDIC_STATEFUL) +* - it is possible to convert Unicode code points +* to a single zero byte (but not as a fallback except for SBCS) +* +* Remaining limitations in fromUnicode: +* - byte sequences must not have leading zero bytes +* - except for SBCS codepages: no fallback mapping from Unicode to a zero byte +* - limitation to up to 4 bytes per character +* +* ICU 2.8 (late 2003) adds a secondary data structure which lifts some of these +* limitations and adds m:n character mappings and other features. +* See ucnv_ext.h for details. +* +* Change history: +* +* 5/6/2001 Ram Moved MBCS_SINGLE_RESULT_FROM_U,MBCS_STAGE_2_FROM_U, +* MBCS_VALUE_2_FROM_STAGE_2, MBCS_VALUE_4_FROM_STAGE_2 +* macros to ucnvmbcs.h file +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION && !UCONFIG_NO_LEGACY_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/ucnv_cb.h" +#include "unicode/udata.h" +#include "unicode/uset.h" +#include "unicode/utf8.h" +#include "unicode/utf16.h" +#include "ucnv_bld.h" +#include "ucnvmbcs.h" +#include "ucnv_ext.h" +#include "ucnv_cnv.h" +#include "cmemory.h" +#include "cstring.h" +#include "umutex.h" +#include "ustr_imp.h" + +/* control optimizations according to the platform */ +#define MBCS_UNROLL_SINGLE_TO_BMP 1 +#define MBCS_UNROLL_SINGLE_FROM_BMP 0 + +/* + * _MBCSHeader versions 5.3 & 4.3 + * (Note that the _MBCSHeader version is in addition to the converter formatVersion.) + * + * This version is optional. Version 5 is used for incompatible data format changes. + * makeconv will continue to generate version 4 files if possible. + * + * Changes from version 4: + * + * The main difference is an additional _MBCSHeader field with + * - the length (number of uint32_t) of the _MBCSHeader + * - flags for further incompatible data format changes + * - flags for further, backward compatible data format changes + * + * The MBCS_OPT_FROM_U flag indicates that most of the fromUnicode data is omitted from + * the file and needs to be reconstituted at load time. + * This requires a utf8Friendly format with an additional mbcsIndex table for fast + * (and UTF-8-friendly) fromUnicode conversion for Unicode code points up to maxFastUChar. + * (For details about these structures see below, and see ucnvmbcs.h.) + * + * utf8Friendly also implies that the fromUnicode mappings are stored in ascending order + * of the Unicode code points. (This requires that the .ucm file has the |0 etc. + * precision markers for all mappings.) + * + * All fallbacks have been moved to the extension table, leaving only roundtrips in the + * omitted data that can be reconstituted from the toUnicode data. + * + * Of the stage 2 table, the part corresponding to maxFastUChar and below is omitted. + * With only roundtrip mappings in the base fromUnicode data, this part is fully + * redundant with the mbcsIndex and will be reconstituted from that (also using the + * stage 1 table which contains the information about how stage 2 was compacted). + * + * The rest of the stage 2 table, the part for code points above maxFastUChar, + * is stored in the file and will be appended to the reconstituted part. + * + * The entire fromUBytes array is omitted from the file and will be reconstitued. + * This is done by enumerating all toUnicode roundtrip mappings, performing + * each mapping (using the stage 1 and reconstituted stage 2 tables) and + * writing instead of reading the byte values. + * + * _MBCSHeader version 4.3 + * + * Change from version 4.2: + * - Optional utf8Friendly data structures, with 64-entry stage 3 block + * allocation for parts of the BMP, and an additional mbcsIndex in non-SBCS + * files which can be used instead of stages 1 & 2. + * Faster lookups for roundtrips from most commonly used characters, + * and lookups from UTF-8 byte sequences with a natural bit distribution. + * See ucnvmbcs.h for more details. + * + * Change from version 4.1: + * - Added an optional extension table structure at the end of the .cnv file. + * It is present if the upper bits of the header flags field contains a non-zero + * byte offset to it. + * Files that contain only a conversion table and no base table + * use the special outputType MBCS_OUTPUT_EXT_ONLY. + * These contain the base table name between the MBCS header and the extension + * data. + * + * Change from version 4.0: + * - Replace header.reserved with header.fromUBytesLength so that all + * fields in the data have length. + * + * Changes from version 3 (for performance improvements): + * - new bit distribution for state table entries + * - reordered action codes + * - new data structure for single-byte fromUnicode + * + stage 2 only contains indexes + * + stage 3 stores 16 bits per character with classification bits 15..8 + * - no multiplier for stage 1 entries + * - stage 2 for non-single-byte codepages contains the index and the flags in + * one 32-bit value + * - 2-byte and 4-byte fromUnicode results are stored directly as 16/32-bit integers + * + * For more details about old versions of the MBCS data structure, see + * the corresponding versions of this file. + * + * Converting stateless codepage data ---------------------------------------*** + * (or codepage data with simple states) to Unicode. + * + * Data structure and algorithm for converting from complex legacy codepages + * to Unicode. (Designed before 2000-may-22.) + * + * The basic idea is that the structure of legacy codepages can be described + * with state tables. + * When reading a byte stream, each input byte causes a state transition. + * Some transitions result in the output of a code point, some result in + * "unassigned" or "illegal" output. + * This is used here for character conversion. + * + * The data structure begins with a state table consisting of a row + * per state, with 256 entries (columns) per row for each possible input + * byte value. + * Each entry is 32 bits wide, with two formats distinguished by + * the sign bit (bit 31): + * + * One format for transitional entries (bit 31 not set) for non-final bytes, and + * one format for final entries (bit 31 set). + * Both formats contain the number of the next state in the same bit + * positions. + * State 0 is the initial state. + * + * Most of the time, the offset values of subsequent states are added + * up to a scalar value. This value will eventually be the index of + * the Unicode code point in a table that follows the state table. + * The effect is that the code points for final state table rows + * are contiguous. The code points of final state rows follow each other + * in the order of the references to those final states by previous + * states, etc. + * + * For some terminal states, the offset is itself the output Unicode + * code point (16 bits for a BMP code point or 20 bits for a supplementary + * code point (stored as code point minus 0x10000 so that 20 bits are enough). + * For others, the code point in the Unicode table is stored with either + * one or two code units: one for BMP code points, two for a pair of + * surrogates. + * All code points for a final state entry take up the same number of code + * units, regardless of whether they all actually _use_ the same number + * of code units. This is necessary for simple array access. + * + * An additional feature comes in with what in ICU is called "fallback" + * mappings: + * + * In addition to round-trippable, precise, 1:1 mappings, there are often + * mappings defined between similar, though not the same, characters. + * Typically, such mappings occur only in fromUnicode mapping tables because + * Unicode has a superset repertoire of most other codepages. However, it + * is possible to provide such mappings in the toUnicode tables, too. + * In this case, the fallback mappings are partly integrated into the + * general state tables because the structure of the encoding includes their + * byte sequences. + * For final entries in an initial state, fallback mappings are stored in + * the entry itself like with roundtrip mappings. + * For other final entries, they are stored in the code units table if + * the entry is for a pair of code units. + * For single-unit results in the code units table, there is no space to + * alternatively hold a fallback mapping; in this case, the code unit + * is stored as U+fffe (unassigned), and the fallback mapping needs to + * be looked up by the scalar offset value in a separate table. + * + * "Unassigned" state entries really mean "structurally unassigned", + * i.e., such a byte sequence will never have a mapping result. + * + * The interpretation of the bits in each entry is as follows: + * + * Bit 31 not set, not a terminal entry ("transitional"): + * 30..24 next state + * 23..0 offset delta, to be added up + * + * Bit 31 set, terminal ("final") entry: + * 30..24 next state (regardless of action code) + * 23..20 action code: + * action codes 0 and 1 result in precise-mapping Unicode code points + * 0 valid byte sequence + * 19..16 not used, 0 + * 15..0 16-bit Unicode BMP code point + * never U+fffe or U+ffff + * 1 valid byte sequence + * 19..0 20-bit Unicode supplementary code point + * never U+fffe or U+ffff + * + * action codes 2 and 3 result in fallback (unidirectional-mapping) Unicode code points + * 2 valid byte sequence (fallback) + * 19..16 not used, 0 + * 15..0 16-bit Unicode BMP code point as fallback result + * 3 valid byte sequence (fallback) + * 19..0 20-bit Unicode supplementary code point as fallback result + * + * action codes 4 and 5 may result in roundtrip/fallback/unassigned/illegal results + * depending on the code units they result in + * 4 valid byte sequence + * 19..9 not used, 0 + * 8..0 final offset delta + * pointing to one 16-bit code unit which may be + * fffe unassigned -- look for a fallback for this offset + * ffff illegal + * 5 valid byte sequence + * 19..9 not used, 0 + * 8..0 final offset delta + * pointing to two 16-bit code units + * (typically UTF-16 surrogates) + * the result depends on the first code unit as follows: + * 0000..d7ff roundtrip BMP code point (1st alone) + * d800..dbff roundtrip surrogate pair (1st, 2nd) + * dc00..dfff fallback surrogate pair (1st-400, 2nd) + * e000 roundtrip BMP code point (2nd alone) + * e001 fallback BMP code point (2nd alone) + * fffe unassigned + * ffff illegal + * (the final offset deltas are at most 255 * 2, + * times 2 because of storing code unit pairs) + * + * 6 unassigned byte sequence + * 19..16 not used, 0 + * 15..0 16-bit Unicode BMP code point U+fffe (new with version 2) + * this does not contain a final offset delta because the main + * purpose of this action code is to save scalar offset values; + * therefore, fallback values cannot be assigned to byte + * sequences that result in this action code + * 7 illegal byte sequence + * 19..16 not used, 0 + * 15..0 16-bit Unicode BMP code point U+ffff (new with version 2) + * 8 state change only + * 19..0 not used, 0 + * useful for state changes in simple stateful encodings, + * at Shift-In/Shift-Out codes + * + * + * 9..15 reserved for future use + * current implementations will only perform a state change + * and ignore bits 19..0 + * + * An encoding with contiguous ranges of unassigned byte sequences, like + * Shift-JIS and especially EUC-TW, can be stored efficiently by having + * at least two states for the trail bytes: + * One trail byte state that results in code points, and one that only + * has "unassigned" and "illegal" terminal states. + * + * Note: partly by accident, this data structure supports simple stateful + * encodings without any additional logic. + * Currently, only simple Shift-In/Shift-Out schemes are handled with + * appropriate state tables (especially EBCDIC_STATEFUL!). + * + * MBCS version 2 added: + * unassigned and illegal action codes have U+fffe and U+ffff + * instead of unused bits; this is useful for _MBCS_SINGLE_SIMPLE_GET_NEXT_BMP() + * + * Converting from Unicode to codepage bytes --------------------------------*** + * + * The conversion data structure for fromUnicode is designed for the known + * structure of Unicode. It maps from 21-bit code points (0..0x10ffff) to + * a sequence of 1..4 bytes, in addition to a flag that indicates if there is + * a roundtrip mapping. + * + * The lookup is done with a 3-stage trie, using 11/6/4 bits for stage 1/2/3 + * like in the character properties table. + * The beginning of the trie is at offsetFromUTable, the beginning of stage 3 + * with the resulting bytes is at offsetFromUBytes. + * + * Beginning with version 4, single-byte codepages have a significantly different + * trie compared to other codepages. + * In all cases, the entry in stage 1 is directly the index of the block of + * 64 entries in stage 2. + * + * Single-byte lookup: + * + * Stage 2 only contains 16-bit indexes directly to the 16-blocks in stage 3. + * Stage 3 contains one 16-bit word per result: + * Bits 15..8 indicate the kind of result: + * f roundtrip result + * c fallback result from private-use code point + * 8 fallback result from other code points + * 0 unassigned + * Bits 7..0 contain the codepage byte. A zero byte is always possible. + * + * In version 4.3, the runtime code can build an sbcsIndex for a utf8Friendly + * file. For 2-byte UTF-8 byte sequences and some 3-byte sequences the lookup + * becomes a 2-stage (single-index) trie lookup with 6 bits for stage 3. + * ASCII code points can be looked up with a linear array access into stage 3. + * See maxFastUChar and other details in ucnvmbcs.h. + * + * Multi-byte lookup: + * + * Stage 2 contains a 32-bit word for each 16-block in stage 3: + * Bits 31..16 contain flags for which stage 3 entries contain roundtrip results + * test: MBCS_FROM_U_IS_ROUNDTRIP(stage2Entry, c) + * If this test is false, then a non-zero result will be interpreted as + * a fallback mapping. + * Bits 15..0 contain the index to stage 3, which must be multiplied by 16*(bytes per char) + * + * Stage 3 contains 2, 3, or 4 bytes per result. + * 2 or 4 bytes are stored as uint16_t/uint32_t in platform endianness, + * while 3 bytes are stored as bytes in big-endian order. + * Leading zero bytes are ignored, and the number of bytes is counted. + * A zero byte mapping result is possible as a roundtrip result. + * For some output types, the actual result is processed from this; + * see ucnv_MBCSFromUnicodeWithOffsets(). + * + * Note that stage 1 always contains 0x440=1088 entries (0x440==0x110000>>10), + * or (version 3 and up) for BMP-only codepages, it contains 64 entries. + * + * In version 4.3, a utf8Friendly file contains an mbcsIndex table. + * For 2-byte UTF-8 byte sequences and most 3-byte sequences the lookup + * becomes a 2-stage (single-index) trie lookup with 6 bits for stage 3. + * ASCII code points can be looked up with a linear array access into stage 3. + * See maxFastUChar, mbcsIndex and other details in ucnvmbcs.h. + * + * In version 3, stage 2 blocks may overlap by multiples of the multiplier + * for compaction. + * In version 4, stage 2 blocks (and for single-byte codepages, stage 3 blocks) + * may overlap by any number of entries. + * + * MBCS version 2 added: + * the converter checks for known output types, which allows + * adding new ones without crashing an unaware converter + */ + +/** + * Callback from ucnv_MBCSEnumToUnicode(), takes 32 mappings from + * consecutive sequences of bytes, starting from the one encoded in value, + * to Unicode code points. (Multiple mappings to reduce per-function call overhead.) + * Does not currently support m:n mappings or reverse fallbacks. + * This function will not be called for sequences of bytes with leading zeros. + * + * @param context an opaque pointer, as passed into ucnv_MBCSEnumToUnicode() + * @param value contains 1..4 bytes of the first byte sequence, right-aligned + * @param codePoints resulting Unicode code points, or negative if a byte sequence does + * not map to anything + * @return true to continue enumeration, false to stop + */ +typedef UBool U_CALLCONV +UConverterEnumToUCallback(const void *context, uint32_t value, UChar32 codePoints[32]); + +static void U_CALLCONV +ucnv_MBCSLoad(UConverterSharedData *sharedData, + UConverterLoadArgs *pArgs, + const uint8_t *raw, + UErrorCode *pErrorCode); + +static void U_CALLCONV +ucnv_MBCSUnload(UConverterSharedData *sharedData); + +static void U_CALLCONV +ucnv_MBCSOpen(UConverter *cnv, + UConverterLoadArgs *pArgs, + UErrorCode *pErrorCode); + +static UChar32 U_CALLCONV +ucnv_MBCSGetNextUChar(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode); + +static void U_CALLCONV +ucnv_MBCSGetStarters(const UConverter* cnv, + UBool starters[256], + UErrorCode *pErrorCode); + +U_CDECL_BEGIN +static const char* U_CALLCONV +ucnv_MBCSGetName(const UConverter *cnv); +U_CDECL_END + +static void U_CALLCONV +ucnv_MBCSWriteSub(UConverterFromUnicodeArgs *pArgs, + int32_t offsetIndex, + UErrorCode *pErrorCode); + +static UChar32 U_CALLCONV +ucnv_MBCSGetNextUChar(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode); + +static void U_CALLCONV +ucnv_SBCSFromUTF8(UConverterFromUnicodeArgs *pFromUArgs, + UConverterToUnicodeArgs *pToUArgs, + UErrorCode *pErrorCode); + +static void U_CALLCONV +ucnv_MBCSGetUnicodeSet(const UConverter *cnv, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode); + +static void U_CALLCONV +ucnv_DBCSFromUTF8(UConverterFromUnicodeArgs *pFromUArgs, + UConverterToUnicodeArgs *pToUArgs, + UErrorCode *pErrorCode); + +static const UConverterImpl _SBCSUTF8Impl={ + UCNV_MBCS, + + ucnv_MBCSLoad, + ucnv_MBCSUnload, + + ucnv_MBCSOpen, + nullptr, + nullptr, + + ucnv_MBCSToUnicodeWithOffsets, + ucnv_MBCSToUnicodeWithOffsets, + ucnv_MBCSFromUnicodeWithOffsets, + ucnv_MBCSFromUnicodeWithOffsets, + ucnv_MBCSGetNextUChar, + + ucnv_MBCSGetStarters, + ucnv_MBCSGetName, + ucnv_MBCSWriteSub, + nullptr, + ucnv_MBCSGetUnicodeSet, + + nullptr, + ucnv_SBCSFromUTF8 +}; + +static const UConverterImpl _DBCSUTF8Impl={ + UCNV_MBCS, + + ucnv_MBCSLoad, + ucnv_MBCSUnload, + + ucnv_MBCSOpen, + nullptr, + nullptr, + + ucnv_MBCSToUnicodeWithOffsets, + ucnv_MBCSToUnicodeWithOffsets, + ucnv_MBCSFromUnicodeWithOffsets, + ucnv_MBCSFromUnicodeWithOffsets, + ucnv_MBCSGetNextUChar, + + ucnv_MBCSGetStarters, + ucnv_MBCSGetName, + ucnv_MBCSWriteSub, + nullptr, + ucnv_MBCSGetUnicodeSet, + + nullptr, + ucnv_DBCSFromUTF8 +}; + +static const UConverterImpl _MBCSImpl={ + UCNV_MBCS, + + ucnv_MBCSLoad, + ucnv_MBCSUnload, + + ucnv_MBCSOpen, + nullptr, + nullptr, + + ucnv_MBCSToUnicodeWithOffsets, + ucnv_MBCSToUnicodeWithOffsets, + ucnv_MBCSFromUnicodeWithOffsets, + ucnv_MBCSFromUnicodeWithOffsets, + ucnv_MBCSGetNextUChar, + + ucnv_MBCSGetStarters, + ucnv_MBCSGetName, + ucnv_MBCSWriteSub, + nullptr, + ucnv_MBCSGetUnicodeSet, + nullptr, + nullptr +}; + +/* Static data is in tools/makeconv/ucnvstat.c for data-based + * converters. Be sure to update it as well. + */ + +const UConverterSharedData _MBCSData={ + sizeof(UConverterSharedData), 1, + nullptr, nullptr, false, true, &_MBCSImpl, + 0, UCNV_MBCS_TABLE_INITIALIZER +}; + + +/* GB 18030 data ------------------------------------------------------------ */ + +/* helper macros for linear values for GB 18030 four-byte sequences */ +#define LINEAR_18030(a, b, c, d) ((((a)*10+(b))*126L+(c))*10L+(d)) + +#define LINEAR_18030_BASE LINEAR_18030(0x81, 0x30, 0x81, 0x30) + +#define LINEAR(x) LINEAR_18030(x>>24, (x>>16)&0xff, (x>>8)&0xff, x&0xff) + +/* + * Some ranges of GB 18030 where both the Unicode code points and the + * GB four-byte sequences are contiguous and are handled algorithmically by + * the special callback functions below. + * The values are start & end of Unicode & GB codes. + * + * Note that single surrogates are not mapped by GB 18030 + * as of the re-released mapping tables from 2000-nov-30. + */ +static const uint32_t +gb18030Ranges[14][4]={ + {0x10000, 0x10FFFF, LINEAR(0x90308130), LINEAR(0xE3329A35)}, + {0x9FA6, 0xD7FF, LINEAR(0x82358F33), LINEAR(0x8336C738)}, + {0x0452, 0x1E3E, LINEAR(0x8130D330), LINEAR(0x8135F436)}, + {0x1E40, 0x200F, LINEAR(0x8135F438), LINEAR(0x8136A531)}, + {0xE865, 0xF92B, LINEAR(0x8336D030), LINEAR(0x84308534)}, + {0x2643, 0x2E80, LINEAR(0x8137A839), LINEAR(0x8138FD38)}, + {0xFA2A, 0xFE2F, LINEAR(0x84309C38), LINEAR(0x84318537)}, + {0x3CE1, 0x4055, LINEAR(0x8231D438), LINEAR(0x8232AF32)}, + {0x361B, 0x3917, LINEAR(0x8230A633), LINEAR(0x8230F237)}, + {0x49B8, 0x4C76, LINEAR(0x8234A131), LINEAR(0x8234E733)}, + {0x4160, 0x4336, LINEAR(0x8232C937), LINEAR(0x8232F837)}, + {0x478E, 0x4946, LINEAR(0x8233E838), LINEAR(0x82349638)}, + {0x44D7, 0x464B, LINEAR(0x8233A339), LINEAR(0x8233C931)}, + {0xFFE6, 0xFFFF, LINEAR(0x8431A234), LINEAR(0x8431A439)} +}; + +/* bit flag for UConverter.options indicating GB 18030 special handling */ +#define _MBCS_OPTION_GB18030 0x8000 + +/* bit flag for UConverter.options indicating KEIS,JEF,JIF special handling */ +#define _MBCS_OPTION_KEIS 0x01000 +#define _MBCS_OPTION_JEF 0x02000 +#define _MBCS_OPTION_JIPS 0x04000 + +#define KEIS_SO_CHAR_1 0x0A +#define KEIS_SO_CHAR_2 0x42 +#define KEIS_SI_CHAR_1 0x0A +#define KEIS_SI_CHAR_2 0x41 + +#define JEF_SO_CHAR 0x28 +#define JEF_SI_CHAR 0x29 + +#define JIPS_SO_CHAR_1 0x1A +#define JIPS_SO_CHAR_2 0x70 +#define JIPS_SI_CHAR_1 0x1A +#define JIPS_SI_CHAR_2 0x71 + +enum SISO_Option { + SI, + SO +}; +typedef enum SISO_Option SISO_Option; + +static int32_t getSISOBytes(SISO_Option option, uint32_t cnvOption, uint8_t *value) { + int32_t SISOLength = 0; + + switch (option) { + case SI: + if ((cnvOption&_MBCS_OPTION_KEIS)!=0) { + value[0] = KEIS_SI_CHAR_1; + value[1] = KEIS_SI_CHAR_2; + SISOLength = 2; + } else if ((cnvOption&_MBCS_OPTION_JEF)!=0) { + value[0] = JEF_SI_CHAR; + SISOLength = 1; + } else if ((cnvOption&_MBCS_OPTION_JIPS)!=0) { + value[0] = JIPS_SI_CHAR_1; + value[1] = JIPS_SI_CHAR_2; + SISOLength = 2; + } else { + value[0] = UCNV_SI; + SISOLength = 1; + } + break; + case SO: + if ((cnvOption&_MBCS_OPTION_KEIS)!=0) { + value[0] = KEIS_SO_CHAR_1; + value[1] = KEIS_SO_CHAR_2; + SISOLength = 2; + } else if ((cnvOption&_MBCS_OPTION_JEF)!=0) { + value[0] = JEF_SO_CHAR; + SISOLength = 1; + } else if ((cnvOption&_MBCS_OPTION_JIPS)!=0) { + value[0] = JIPS_SO_CHAR_1; + value[1] = JIPS_SO_CHAR_2; + SISOLength = 2; + } else { + value[0] = UCNV_SO; + SISOLength = 1; + } + break; + default: + /* Should never happen. */ + break; + } + + return SISOLength; +} + +/* Miscellaneous ------------------------------------------------------------ */ + +/* similar to ucnv_MBCSGetNextUChar() but recursive */ +static UBool +enumToU(UConverterMBCSTable *mbcsTable, int8_t stateProps[], + int32_t state, uint32_t offset, + uint32_t value, + UConverterEnumToUCallback *callback, const void *context, + UErrorCode *pErrorCode) { + UChar32 codePoints[32]; + const int32_t *row; + const uint16_t *unicodeCodeUnits; + UChar32 anyCodePoints; + int32_t b, limit; + + row=mbcsTable->stateTable[state]; + unicodeCodeUnits=mbcsTable->unicodeCodeUnits; + + value<<=8; + anyCodePoints=-1; /* becomes non-negative if there is a mapping */ + + b=(stateProps[state]&0x38)<<2; + if(b==0 && stateProps[state]>=0x40) { + /* skip byte sequences with leading zeros because they are not stored in the fromUnicode table */ + codePoints[0]=U_SENTINEL; + b=1; + } + limit=((stateProps[state]&7)+1)<<5; + while(b=0) { + /* recurse to a state with non-ignorable actions */ + if(!enumToU( + mbcsTable, stateProps, nextState, + offset+MBCS_ENTRY_TRANSITION_OFFSET(entry), + value|(uint32_t)b, + callback, context, + pErrorCode)) { + return false; + } + } + codePoints[b&0x1f]=U_SENTINEL; + } else { + UChar32 c; + int32_t action; + + /* + * An if-else-if chain provides more reliable performance for + * the most common cases compared to a switch. + */ + action=MBCS_ENTRY_FINAL_ACTION(entry); + if(action==MBCS_STATE_VALID_DIRECT_16) { + /* output BMP code point */ + c=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + } else if(action==MBCS_STATE_VALID_16) { + int32_t finalOffset=offset+MBCS_ENTRY_FINAL_VALUE_16(entry); + c=unicodeCodeUnits[finalOffset]; + if(c<0xfffe) { + /* output BMP code point */ + } else { + c=U_SENTINEL; + } + } else if(action==MBCS_STATE_VALID_16_PAIR) { + int32_t finalOffset=offset+MBCS_ENTRY_FINAL_VALUE_16(entry); + c=unicodeCodeUnits[finalOffset++]; + if(c<0xd800) { + /* output BMP code point below 0xd800 */ + } else if(c<=0xdbff) { + /* output roundtrip or fallback supplementary code point */ + c=((c&0x3ff)<<10)+unicodeCodeUnits[finalOffset]+(0x10000-0xdc00); + } else if(c==0xe000) { + /* output roundtrip BMP code point above 0xd800 or fallback BMP code point */ + c=unicodeCodeUnits[finalOffset]; + } else { + c=U_SENTINEL; + } + } else if(action==MBCS_STATE_VALID_DIRECT_20) { + /* output supplementary code point */ + c=(UChar32)(MBCS_ENTRY_FINAL_VALUE(entry)+0x10000); + } else { + c=U_SENTINEL; + } + + codePoints[b&0x1f]=c; + anyCodePoints&=c; + } + if(((++b)&0x1f)==0) { + if(anyCodePoints>=0) { + if(!callback(context, value|(uint32_t)(b-0x20), codePoints)) { + return false; + } + anyCodePoints=-1; + } + } + } + return true; +} + +/* + * Only called if stateProps[state]==-1. + * A recursive call may do stateProps[state]|=0x40 if this state is the target of an + * MBCS_STATE_CHANGE_ONLY. + */ +static int8_t +getStateProp(const int32_t (*stateTable)[256], int8_t stateProps[], int state) { + const int32_t *row; + int32_t min, max, entry, nextState; + + row=stateTable[state]; + stateProps[state]=0; + + /* find first non-ignorable state */ + for(min=0;; ++min) { + entry=row[min]; + nextState=MBCS_ENTRY_STATE(entry); + if(stateProps[nextState]==-1) { + getStateProp(stateTable, stateProps, nextState); + } + if(MBCS_ENTRY_IS_TRANSITION(entry)) { + if(stateProps[nextState]>=0) { + break; + } + } else if(MBCS_ENTRY_FINAL_ACTION(entry)>5)<<3); + + /* find last non-ignorable state */ + for(max=0xff; min=0) { + break; + } + } else if(MBCS_ENTRY_FINAL_ACTION(entry)>5); + + /* recurse further and collect direct-state information */ + while(min<=max) { + entry=row[min]; + nextState=MBCS_ENTRY_STATE(entry); + if(stateProps[nextState]==-1) { + getStateProp(stateTable, stateProps, nextState); + } + if(MBCS_ENTRY_IS_FINAL(entry)) { + stateProps[nextState]|=0x40; + if(MBCS_ENTRY_FINAL_ACTION(entry)<=MBCS_STATE_FALLBACK_DIRECT_20) { + stateProps[state]|=0x40; + } + } + ++min; + } + return stateProps[state]; +} + +/* + * Internal function enumerating the toUnicode data of an MBCS converter. + * Currently only used for reconstituting data for a MBCS_OPT_NO_FROM_U + * table, but could also be used for a future ucnv_getUnicodeSet() option + * that includes reverse fallbacks (after updating this function's implementation). + * Currently only handles roundtrip mappings. + * Does not currently handle extensions. + */ +static void +ucnv_MBCSEnumToUnicode(UConverterMBCSTable *mbcsTable, + UConverterEnumToUCallback *callback, const void *context, + UErrorCode *pErrorCode) { + /* + * Properties for each state, to speed up the enumeration. + * Ignorable actions are unassigned/illegal/state-change-only: + * They do not lead to mappings. + * + * Bits 7..6: + * 1 direct/initial state (stateful converters have multiple) + * 0 non-initial state with transitions or with non-ignorable result actions + * -1 final state with only ignorable actions + * + * Bits 5..3: + * The lowest byte value with non-ignorable actions is + * value<<5 (rounded down). + * + * Bits 2..0: + * The highest byte value with non-ignorable actions is + * (value<<5)&0x1f (rounded up). + */ + int8_t stateProps[MBCS_MAX_STATE_COUNT]; + int32_t state; + + uprv_memset(stateProps, -1, sizeof(stateProps)); + + /* recurse from state 0 and set all stateProps */ + getStateProp(mbcsTable->stateTable, stateProps, 0); + + for(state=0; statecountStates; ++state) { + /*if(stateProps[state]==-1) { + printf("unused/unreachable %d\n", state); + }*/ + if(stateProps[state]>=0x40) { + /* start from each direct state */ + enumToU( + mbcsTable, stateProps, state, 0, 0, + callback, context, + pErrorCode); + } + } +} + +U_CFUNC void +ucnv_MBCSGetFilteredUnicodeSetForUnicode(const UConverterSharedData *sharedData, + const USetAdder *sa, + UConverterUnicodeSet which, + UConverterSetFilter filter, + UErrorCode *pErrorCode) { + const UConverterMBCSTable *mbcsTable; + const uint16_t *table; + + uint32_t st3; + uint16_t st1, maxStage1, st2; + + UChar32 c; + + /* enumerate the from-Unicode trie table */ + mbcsTable=&sharedData->mbcs; + table=mbcsTable->fromUnicodeTable; + if(mbcsTable->unicodeMask&UCNV_HAS_SUPPLEMENTARY) { + maxStage1=0x440; + } else { + maxStage1=0x40; + } + + c=0; /* keep track of the current code point while enumerating */ + + if(mbcsTable->outputType==MBCS_OUTPUT_1) { + const uint16_t *stage2, *stage3, *results; + uint16_t minValue; + + results=(const uint16_t *)mbcsTable->fromUnicodeBytes; + + /* + * Set a threshold variable for selecting which mappings to use. + * See ucnv_MBCSSingleFromBMPWithOffsets() and + * MBCS_SINGLE_RESULT_FROM_U() for details. + */ + if(which==UCNV_ROUNDTRIP_SET) { + /* use only roundtrips */ + minValue=0xf00; + } else /* UCNV_ROUNDTRIP_AND_FALLBACK_SET */ { + /* use all roundtrip and fallback results */ + minValue=0x800; + } + + for(st1=0; st1maxStage1) { + stage2=table+st2; + for(st2=0; st2<64; ++st2) { + if((st3=stage2[st2])!=0) { + /* read the stage 3 block */ + stage3=results+st3; + + do { + if(*stage3++>=minValue) { + sa->add(sa->set, c); + } + } while((++c&0xf)!=0); + } else { + c+=16; /* empty stage 3 block */ + } + } + } else { + c+=1024; /* empty stage 2 block */ + } + } + } else { + const uint32_t *stage2; + const uint8_t *stage3, *bytes; + uint32_t st3Multiplier; + uint32_t value; + UBool useFallback; + + bytes=mbcsTable->fromUnicodeBytes; + + useFallback=(UBool)(which==UCNV_ROUNDTRIP_AND_FALLBACK_SET); + + switch(mbcsTable->outputType) { + case MBCS_OUTPUT_3: + case MBCS_OUTPUT_4_EUC: + st3Multiplier=3; + break; + case MBCS_OUTPUT_4: + st3Multiplier=4; + break; + default: + st3Multiplier=2; + break; + } + + for(st1=0; st1(maxStage1>>1)) { + stage2=(const uint32_t *)table+st2; + for(st2=0; st2<64; ++st2) { + if((st3=stage2[st2])!=0) { + /* read the stage 3 block */ + stage3=bytes+st3Multiplier*16*(uint32_t)(uint16_t)st3; + + /* get the roundtrip flags for the stage 3 block */ + st3>>=16; + + /* + * Add code points for which the roundtrip flag is set, + * or which map to non-zero bytes if we use fallbacks. + * See ucnv_MBCSFromUnicodeWithOffsets() for details. + */ + switch(filter) { + case UCNV_SET_FILTER_NONE: + do { + if(st3&1) { + sa->add(sa->set, c); + stage3+=st3Multiplier; + } else if(useFallback) { + uint8_t b=0; + switch(st3Multiplier) { + case 4: + b|=*stage3++; + U_FALLTHROUGH; + case 3: + b|=*stage3++; + U_FALLTHROUGH; + case 2: + b|=stage3[0]|stage3[1]; + stage3+=2; + U_FALLTHROUGH; + default: + break; + } + if(b!=0) { + sa->add(sa->set, c); + } + } + st3>>=1; + } while((++c&0xf)!=0); + break; + case UCNV_SET_FILTER_DBCS_ONLY: + /* Ignore single-byte results (<0x100). */ + do { + if(((st3&1)!=0 || useFallback) && *((const uint16_t *)stage3)>=0x100) { + sa->add(sa->set, c); + } + st3>>=1; + stage3+=2; /* +=st3Multiplier */ + } while((++c&0xf)!=0); + break; + case UCNV_SET_FILTER_2022_CN: + /* Only add code points that map to CNS 11643 planes 1 & 2 for non-EXT ISO-2022-CN. */ + do { + if(((st3&1)!=0 || useFallback) && ((value=*stage3)==0x81 || value==0x82)) { + sa->add(sa->set, c); + } + st3>>=1; + stage3+=3; /* +=st3Multiplier */ + } while((++c&0xf)!=0); + break; + case UCNV_SET_FILTER_SJIS: + /* Only add code points that map to Shift-JIS codes corresponding to JIS X 0208. */ + do { + if(((st3&1)!=0 || useFallback) && (value=*((const uint16_t *)stage3))>=0x8140 && value<=0xeffc) { + sa->add(sa->set, c); + } + st3>>=1; + stage3+=2; /* +=st3Multiplier */ + } while((++c&0xf)!=0); + break; + case UCNV_SET_FILTER_GR94DBCS: + /* Only add code points that map to ISO 2022 GR 94 DBCS codes (each byte A1..FE). */ + do { + if( ((st3&1)!=0 || useFallback) && + (uint16_t)((value=*((const uint16_t *)stage3)) - 0xa1a1)<=(0xfefe - 0xa1a1) && + (uint8_t)(value-0xa1)<=(0xfe - 0xa1) + ) { + sa->add(sa->set, c); + } + st3>>=1; + stage3+=2; /* +=st3Multiplier */ + } while((++c&0xf)!=0); + break; + case UCNV_SET_FILTER_HZ: + /* Only add code points that are suitable for HZ DBCS (lead byte A1..FD). */ + do { + if( ((st3&1)!=0 || useFallback) && + (uint16_t)((value=*((const uint16_t *)stage3))-0xa1a1)<=(0xfdfe - 0xa1a1) && + (uint8_t)(value-0xa1)<=(0xfe - 0xa1) + ) { + sa->add(sa->set, c); + } + st3>>=1; + stage3+=2; /* +=st3Multiplier */ + } while((++c&0xf)!=0); + break; + default: + *pErrorCode=U_INTERNAL_PROGRAM_ERROR; + return; + } + } else { + c+=16; /* empty stage 3 block */ + } + } + } else { + c+=1024; /* empty stage 2 block */ + } + } + } + + ucnv_extGetUnicodeSet(sharedData, sa, which, filter, pErrorCode); +} + +U_CFUNC void +ucnv_MBCSGetUnicodeSetForUnicode(const UConverterSharedData *sharedData, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode) { + ucnv_MBCSGetFilteredUnicodeSetForUnicode( + sharedData, sa, which, + sharedData->mbcs.outputType==MBCS_OUTPUT_DBCS_ONLY ? + UCNV_SET_FILTER_DBCS_ONLY : + UCNV_SET_FILTER_NONE, + pErrorCode); +} + +static void U_CALLCONV +ucnv_MBCSGetUnicodeSet(const UConverter *cnv, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode) { + if(cnv->options&_MBCS_OPTION_GB18030) { + sa->addRange(sa->set, 0, 0xd7ff); + sa->addRange(sa->set, 0xe000, 0x10ffff); + } else { + ucnv_MBCSGetUnicodeSetForUnicode(cnv->sharedData, sa, which, pErrorCode); + } +} + +/* conversion extensions for input not in the main table -------------------- */ + +/* + * Hardcoded extension handling for GB 18030. + * Definition of LINEAR macros and gb18030Ranges see near the beginning of the file. + * + * In the future, conversion extensions may handle m:n mappings and delta tables, + * see https://htmlpreview.github.io/?https://github.com/unicode-org/icu-docs/blob/main/design/conversion/conversion_extensions.html + * + * If an input character cannot be mapped, then these functions set an error + * code. The framework will then call the callback function. + */ + +/* + * @return if(U_FAILURE) return the code point for cnv->fromUChar32 + * else return 0 after output has been written to the target + */ +static UChar32 +_extFromU(UConverter *cnv, const UConverterSharedData *sharedData, + UChar32 cp, + const char16_t **source, const char16_t *sourceLimit, + uint8_t **target, const uint8_t *targetLimit, + int32_t **offsets, int32_t sourceIndex, + UBool flush, + UErrorCode *pErrorCode) { + const int32_t *cx; + + cnv->useSubChar1=false; + + if( (cx=sharedData->mbcs.extIndexes)!=nullptr && + ucnv_extInitialMatchFromU( + cnv, cx, + cp, source, sourceLimit, + (char **)target, (char *)targetLimit, + offsets, sourceIndex, + flush, + pErrorCode) + ) { + return 0; /* an extension mapping handled the input */ + } + + /* GB 18030 */ + if((cnv->options&_MBCS_OPTION_GB18030)!=0) { + const uint32_t *range; + int32_t i; + + range=gb18030Ranges[0]; + for(i=0; itoUBytes[0..length[ + * @return if(U_FAILURE) return the length (toULength, byteIndex) for the input + * else return 0 after output has been written to the target + */ +static int8_t +_extToU(UConverter *cnv, const UConverterSharedData *sharedData, + int8_t length, + const uint8_t **source, const uint8_t *sourceLimit, + char16_t **target, const char16_t *targetLimit, + int32_t **offsets, int32_t sourceIndex, + UBool flush, + UErrorCode *pErrorCode) { + const int32_t *cx; + + if( (cx=sharedData->mbcs.extIndexes)!=nullptr && + ucnv_extInitialMatchToU( + cnv, cx, + length, (const char **)source, (const char *)sourceLimit, + target, targetLimit, + offsets, sourceIndex, + flush, + pErrorCode) + ) { + return 0; /* an extension mapping handled the input */ + } + + /* GB 18030 */ + if(length==4 && (cnv->options&_MBCS_OPTION_GB18030)!=0) { + const uint32_t *range; + uint32_t linear; + int32_t i; + + linear=LINEAR_18030(cnv->toUBytes[0], cnv->toUBytes[1], cnv->toUBytes[2], cnv->toUBytes[3]); + range=gb18030Ranges[0]; + for(i=0; iNL ------------------------------------------------------ */ + +/* + * This code modifies a standard EBCDIC<->Unicode mapping table for + * OS/390 (z/OS) Unix System Services (Open Edition). + * The difference is in the mapping of Line Feed and New Line control codes: + * Standard EBCDIC maps + * + * \x25 |0 + * \x15 |0 + * + * but OS/390 USS EBCDIC swaps the control codes for LF and NL, + * mapping + * + * \x15 |0 + * \x25 |0 + * + * This code modifies a loaded standard EBCDIC<->Unicode mapping table + * by copying it into allocated memory and swapping the LF and NL values. + * It allows to support the same EBCDIC charset in both versions without + * duplicating the entire installed table. + */ + +/* standard EBCDIC codes */ +#define EBCDIC_LF 0x25 +#define EBCDIC_NL 0x15 + +/* standard EBCDIC codes with roundtrip flag as stored in Unicode-to-single-byte tables */ +#define EBCDIC_RT_LF 0xf25 +#define EBCDIC_RT_NL 0xf15 + +/* Unicode code points */ +#define U_LF 0x0a +#define U_NL 0x85 + +static UBool +_EBCDICSwapLFNL(UConverterSharedData *sharedData, UErrorCode *pErrorCode) { + UConverterMBCSTable *mbcsTable; + + const uint16_t *table, *results; + const uint8_t *bytes; + + int32_t (*newStateTable)[256]; + uint16_t *newResults; + uint8_t *p; + char *name; + + uint32_t stage2Entry; + uint32_t size, sizeofFromUBytes; + + mbcsTable=&sharedData->mbcs; + + table=mbcsTable->fromUnicodeTable; + bytes=mbcsTable->fromUnicodeBytes; + results=(const uint16_t *)bytes; + + /* + * Check that this is an EBCDIC table with SBCS portion - + * SBCS or EBCDIC_STATEFUL with standard EBCDIC LF and NL mappings. + * + * If not, ignore the option. Options are always ignored if they do not apply. + */ + if(!( + (mbcsTable->outputType==MBCS_OUTPUT_1 || mbcsTable->outputType==MBCS_OUTPUT_2_SISO) && + mbcsTable->stateTable[0][EBCDIC_LF]==MBCS_ENTRY_FINAL(0, MBCS_STATE_VALID_DIRECT_16, U_LF) && + mbcsTable->stateTable[0][EBCDIC_NL]==MBCS_ENTRY_FINAL(0, MBCS_STATE_VALID_DIRECT_16, U_NL) + )) { + return false; + } + + if(mbcsTable->outputType==MBCS_OUTPUT_1) { + if(!( + EBCDIC_RT_LF==MBCS_SINGLE_RESULT_FROM_U(table, results, U_LF) && + EBCDIC_RT_NL==MBCS_SINGLE_RESULT_FROM_U(table, results, U_NL) + )) { + return false; + } + } else /* MBCS_OUTPUT_2_SISO */ { + stage2Entry=MBCS_STAGE_2_FROM_U(table, U_LF); + if(!( + MBCS_FROM_U_IS_ROUNDTRIP(stage2Entry, U_LF)!=0 && + EBCDIC_LF==MBCS_VALUE_2_FROM_STAGE_2(bytes, stage2Entry, U_LF) + )) { + return false; + } + + stage2Entry=MBCS_STAGE_2_FROM_U(table, U_NL); + if(!( + MBCS_FROM_U_IS_ROUNDTRIP(stage2Entry, U_NL)!=0 && + EBCDIC_NL==MBCS_VALUE_2_FROM_STAGE_2(bytes, stage2Entry, U_NL) + )) { + return false; + } + } + + if(mbcsTable->fromUBytesLength>0) { + /* + * We _know_ the number of bytes in the fromUnicodeBytes array + * starting with header.version 4.1. + */ + sizeofFromUBytes=mbcsTable->fromUBytesLength; + } else { + /* + * Otherwise: + * There used to be code to enumerate the fromUnicode + * trie and find the highest entry, but it was removed in ICU 3.2 + * because it was not tested and caused a low code coverage number. + * See Jitterbug 3674. + * This affects only some .cnv file formats with a header.version + * below 4.1, and only when swaplfnl is requested. + * + * ucnvmbcs.c revision 1.99 is the last one with the + * ucnv_MBCSSizeofFromUBytes() function. + */ + *pErrorCode=U_INVALID_FORMAT_ERROR; + return false; + } + + /* + * The table has an appropriate format. + * Allocate and build + * - a modified to-Unicode state table + * - a modified from-Unicode output array + * - a converter name string with the swap option appended + */ + size= + mbcsTable->countStates*1024+ + sizeofFromUBytes+ + UCNV_MAX_CONVERTER_NAME_LENGTH+20; + p=(uint8_t *)uprv_malloc(size); + if(p==nullptr) { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return false; + } + + /* copy and modify the to-Unicode state table */ + newStateTable=(int32_t (*)[256])p; + uprv_memcpy(newStateTable, mbcsTable->stateTable, mbcsTable->countStates*1024); + + newStateTable[0][EBCDIC_LF]=MBCS_ENTRY_FINAL(0, MBCS_STATE_VALID_DIRECT_16, U_NL); + newStateTable[0][EBCDIC_NL]=MBCS_ENTRY_FINAL(0, MBCS_STATE_VALID_DIRECT_16, U_LF); + + /* copy and modify the from-Unicode result table */ + newResults=(uint16_t *)newStateTable[mbcsTable->countStates]; + uprv_memcpy(newResults, bytes, sizeofFromUBytes); + + /* conveniently, the table access macros work on the left side of expressions */ + if(mbcsTable->outputType==MBCS_OUTPUT_1) { + MBCS_SINGLE_RESULT_FROM_U(table, newResults, U_LF)=EBCDIC_RT_NL; + MBCS_SINGLE_RESULT_FROM_U(table, newResults, U_NL)=EBCDIC_RT_LF; + } else /* MBCS_OUTPUT_2_SISO */ { + stage2Entry=MBCS_STAGE_2_FROM_U(table, U_LF); + MBCS_VALUE_2_FROM_STAGE_2(newResults, stage2Entry, U_LF)=EBCDIC_NL; + + stage2Entry=MBCS_STAGE_2_FROM_U(table, U_NL); + MBCS_VALUE_2_FROM_STAGE_2(newResults, stage2Entry, U_NL)=EBCDIC_LF; + } + + /* set the canonical converter name */ + name=(char *)newResults+sizeofFromUBytes; + uprv_strcpy(name, sharedData->staticData->name); + uprv_strcat(name, UCNV_SWAP_LFNL_OPTION_STRING); + + /* set the pointers */ + icu::umtx_lock(nullptr); + if(mbcsTable->swapLFNLStateTable==nullptr) { + mbcsTable->swapLFNLStateTable=newStateTable; + mbcsTable->swapLFNLFromUnicodeBytes=(uint8_t *)newResults; + mbcsTable->swapLFNLName=name; + + newStateTable=nullptr; + } + icu::umtx_unlock(nullptr); + + /* release the allocated memory if another thread beat us to it */ + if(newStateTable!=nullptr) { + uprv_free(newStateTable); + } + return true; +} + +/* reconstitute omitted fromUnicode data ------------------------------------ */ + +/* for details, compare with genmbcs.c MBCSAddFromUnicode() and transformEUC() */ +static UBool U_CALLCONV +writeStage3Roundtrip(const void *context, uint32_t value, UChar32 codePoints[32]) { + UConverterMBCSTable *mbcsTable=(UConverterMBCSTable *)context; + const uint16_t *table; + uint32_t *stage2; + uint8_t *bytes, *p; + UChar32 c; + int32_t i, st3; + + table=mbcsTable->fromUnicodeTable; + bytes=(uint8_t *)mbcsTable->fromUnicodeBytes; + + /* for EUC outputTypes, modify the value like genmbcs.c's transformEUC() */ + switch(mbcsTable->outputType) { + case MBCS_OUTPUT_3_EUC: + if(value<=0xffff) { + /* short sequences are stored directly */ + /* code set 0 or 1 */ + } else if(value<=0x8effff) { + /* code set 2 */ + value&=0x7fff; + } else /* first byte is 0x8f */ { + /* code set 3 */ + value&=0xff7f; + } + break; + case MBCS_OUTPUT_4_EUC: + if(value<=0xffffff) { + /* short sequences are stored directly */ + /* code set 0 or 1 */ + } else if(value<=0x8effffff) { + /* code set 2 */ + value&=0x7fffff; + } else /* first byte is 0x8f */ { + /* code set 3 */ + value&=0xff7fff; + } + break; + default: + break; + } + + for(i=0; i<=0x1f; ++value, ++i) { + c=codePoints[i]; + if(c<0) { + continue; + } + + /* locate the stage 2 & 3 data */ + stage2=((uint32_t *)table)+table[c>>10]+((c>>4)&0x3f); + p=bytes; + st3=(int32_t)(uint16_t)*stage2*16+(c&0xf); + + /* write the codepage bytes into stage 3 */ + switch(mbcsTable->outputType) { + case MBCS_OUTPUT_3: + case MBCS_OUTPUT_4_EUC: + p+=st3*3; + p[0]=(uint8_t)(value>>16); + p[1]=(uint8_t)(value>>8); + p[2]=(uint8_t)value; + break; + case MBCS_OUTPUT_4: + ((uint32_t *)p)[st3]=value; + break; + default: + /* 2 bytes per character */ + ((uint16_t *)p)[st3]=(uint16_t)value; + break; + } + + /* set the roundtrip flag */ + *stage2|=(1UL<<(16+(c&0xf))); + } + return true; + } + +static void +reconstituteData(UConverterMBCSTable *mbcsTable, + uint32_t stage1Length, uint32_t stage2Length, + uint32_t fullStage2Length, /* lengths are numbers of units, not bytes */ + UErrorCode *pErrorCode) { + uint16_t *stage1; + uint32_t *stage2; + uint32_t dataLength=stage1Length*2+fullStage2Length*4+mbcsTable->fromUBytesLength; + mbcsTable->reconstitutedData=(uint8_t *)uprv_malloc(dataLength); + if(mbcsTable->reconstitutedData==nullptr) { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + uprv_memset(mbcsTable->reconstitutedData, 0, dataLength); + + /* copy existing data and reroute the pointers */ + stage1=(uint16_t *)mbcsTable->reconstitutedData; + uprv_memcpy(stage1, mbcsTable->fromUnicodeTable, stage1Length*2); + + stage2=(uint32_t *)(stage1+stage1Length); + uprv_memcpy(stage2+(fullStage2Length-stage2Length), + mbcsTable->fromUnicodeTable+stage1Length, + stage2Length*4); + + mbcsTable->fromUnicodeTable=stage1; + mbcsTable->fromUnicodeBytes=(uint8_t *)(stage2+fullStage2Length); + + /* indexes into stage 2 count from the bottom of the fromUnicodeTable */ + stage2=(uint32_t *)stage1; + + /* reconstitute the initial part of stage 2 from the mbcsIndex */ + { + int32_t stageUTF8Length=((int32_t)mbcsTable->maxFastUChar+1)>>6; + int32_t stageUTF8Index=0; + int32_t st1, st2, st3, i; + + for(st1=0; stageUTF8IndexmbcsIndex[stageUTF8Index++]; + if(st3!=0) { + /* an stage 2 entry's index is per stage 3 16-block, not per stage 3 entry */ + st3>>=4; + /* + * 4 stage 2 entries point to 4 consecutive stage 3 16-blocks which are + * allocated together as a single 64-block for access from the mbcsIndex + */ + stage2[st2++]=st3++; + stage2[st2++]=st3++; + stage2[st2++]=st3++; + stage2[st2++]=st3; + } else { + /* no stage 3 block, skip */ + st2+=4; + } + } + } else { + /* no stage 2 block, skip */ + stageUTF8Index+=16; + } + } + } + + /* reconstitute fromUnicodeBytes with roundtrips from toUnicode data */ + ucnv_MBCSEnumToUnicode(mbcsTable, writeStage3Roundtrip, mbcsTable, pErrorCode); +} + +/* MBCS setup functions ----------------------------------------------------- */ + +static void U_CALLCONV +ucnv_MBCSLoad(UConverterSharedData *sharedData, + UConverterLoadArgs *pArgs, + const uint8_t *raw, + UErrorCode *pErrorCode) { + UDataInfo info; + UConverterMBCSTable *mbcsTable=&sharedData->mbcs; + _MBCSHeader *header=(_MBCSHeader *)raw; + uint32_t offset; + uint32_t headerLength; + UBool noFromU=false; + + if(header->version[0]==4) { + headerLength=MBCS_HEADER_V4_LENGTH; + } else if(header->version[0]==5 && header->version[1]>=3 && + (header->options&MBCS_OPT_UNKNOWN_INCOMPATIBLE_MASK)==0) { + headerLength=header->options&MBCS_OPT_LENGTH_MASK; + noFromU=(UBool)((header->options&MBCS_OPT_NO_FROM_U)!=0); + } else { + *pErrorCode=U_INVALID_TABLE_FORMAT; + return; + } + + mbcsTable->outputType=(uint8_t)header->flags; + if(noFromU && mbcsTable->outputType==MBCS_OUTPUT_1) { + *pErrorCode=U_INVALID_TABLE_FORMAT; + return; + } + + /* extension data, header version 4.2 and higher */ + offset=header->flags>>8; + if(offset!=0) { + mbcsTable->extIndexes=(const int32_t *)(raw+offset); + } + + if(mbcsTable->outputType==MBCS_OUTPUT_EXT_ONLY) { + UConverterLoadArgs args=UCNV_LOAD_ARGS_INITIALIZER; + UConverterSharedData *baseSharedData; + const int32_t *extIndexes; + const char *baseName; + + /* extension-only file, load the base table and set values appropriately */ + if((extIndexes=mbcsTable->extIndexes)==nullptr) { + /* extension-only file without extension */ + *pErrorCode=U_INVALID_TABLE_FORMAT; + return; + } + + if(pArgs->nestedLoads!=1) { + /* an extension table must not be loaded as a base table */ + *pErrorCode=U_INVALID_TABLE_FILE; + return; + } + + /* load the base table */ + baseName=(const char *)header+headerLength*4; + if(0==uprv_strcmp(baseName, sharedData->staticData->name)) { + /* forbid loading this same extension-only file */ + *pErrorCode=U_INVALID_TABLE_FORMAT; + return; + } + + /* TODO parse package name out of the prefix of the base name in the extension .cnv file? */ + args.size=sizeof(UConverterLoadArgs); + args.nestedLoads=2; + args.onlyTestIsLoadable=pArgs->onlyTestIsLoadable; + args.reserved=pArgs->reserved; + args.options=pArgs->options; + args.pkg=pArgs->pkg; + args.name=baseName; + baseSharedData=ucnv_load(&args, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + return; + } + if( baseSharedData->staticData->conversionType!=UCNV_MBCS || + baseSharedData->mbcs.baseSharedData!=nullptr + ) { + ucnv_unload(baseSharedData); + *pErrorCode=U_INVALID_TABLE_FORMAT; + return; + } + if(pArgs->onlyTestIsLoadable) { + /* + * Exit as soon as we know that we can load the converter + * and the format is valid and supported. + * The worst that can happen in the following code is a memory + * allocation error. + */ + ucnv_unload(baseSharedData); + return; + } + + /* copy the base table data */ + uprv_memcpy(mbcsTable, &baseSharedData->mbcs, sizeof(UConverterMBCSTable)); + + /* overwrite values with relevant ones for the extension converter */ + mbcsTable->baseSharedData=baseSharedData; + mbcsTable->extIndexes=extIndexes; + + /* + * It would be possible to share the swapLFNL data with a base converter, + * but the generated name would have to be different, and the memory + * would have to be free'd only once. + * It is easier to just create the data for the extension converter + * separately when it is requested. + */ + mbcsTable->swapLFNLStateTable=nullptr; + mbcsTable->swapLFNLFromUnicodeBytes=nullptr; + mbcsTable->swapLFNLName=nullptr; + + /* + * The reconstitutedData must be deleted only when the base converter + * is unloaded. + */ + mbcsTable->reconstitutedData=nullptr; + + /* + * Set a special, runtime-only outputType if the extension converter + * is a DBCS version of a base converter that also maps single bytes. + */ + if( sharedData->staticData->conversionType==UCNV_DBCS || + (sharedData->staticData->conversionType==UCNV_MBCS && + sharedData->staticData->minBytesPerChar>=2) + ) { + if(baseSharedData->mbcs.outputType==MBCS_OUTPUT_2_SISO) { + /* the base converter is SI/SO-stateful */ + int32_t entry; + + /* get the dbcs state from the state table entry for SO=0x0e */ + entry=mbcsTable->stateTable[0][0xe]; + if( MBCS_ENTRY_IS_FINAL(entry) && + MBCS_ENTRY_FINAL_ACTION(entry)==MBCS_STATE_CHANGE_ONLY && + MBCS_ENTRY_FINAL_STATE(entry)!=0 + ) { + mbcsTable->dbcsOnlyState=(uint8_t)MBCS_ENTRY_FINAL_STATE(entry); + + mbcsTable->outputType=MBCS_OUTPUT_DBCS_ONLY; + } + } else if( + baseSharedData->staticData->conversionType==UCNV_MBCS && + baseSharedData->staticData->minBytesPerChar==1 && + baseSharedData->staticData->maxBytesPerChar==2 && + mbcsTable->countStates<=127 + ) { + /* non-stateful base converter, need to modify the state table */ + int32_t (*newStateTable)[256]; + int32_t *state; + int32_t i, count; + + /* allocate a new state table and copy the base state table contents */ + count=mbcsTable->countStates; + newStateTable=(int32_t (*)[256])uprv_malloc((count+1)*1024); + if(newStateTable==nullptr) { + ucnv_unload(baseSharedData); + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return; + } + + uprv_memcpy(newStateTable, mbcsTable->stateTable, count*1024); + + /* change all final single-byte entries to go to a new all-illegal state */ + state=newStateTable[0]; + for(i=0; i<256; ++i) { + if(MBCS_ENTRY_IS_FINAL(state[i])) { + state[i]=MBCS_ENTRY_TRANSITION(count, 0); + } + } + + /* build the new all-illegal state */ + state=newStateTable[count]; + for(i=0; i<256; ++i) { + state[i]=MBCS_ENTRY_FINAL(0, MBCS_STATE_ILLEGAL, 0); + } + mbcsTable->stateTable=(const int32_t (*)[256])newStateTable; + mbcsTable->countStates=(uint8_t)(count+1); + mbcsTable->stateTableOwned=true; + + mbcsTable->outputType=MBCS_OUTPUT_DBCS_ONLY; + } + } + + /* + * unlike below for files with base tables, do not get the unicodeMask + * from the sharedData; instead, use the base table's unicodeMask, + * which we copied in the memcpy above; + * this is necessary because the static data unicodeMask, especially + * the UCNV_HAS_SUPPLEMENTARY flag, is part of the base table data + */ + } else { + /* conversion file with a base table; an additional extension table is optional */ + /* make sure that the output type is known */ + switch(mbcsTable->outputType) { + case MBCS_OUTPUT_1: + case MBCS_OUTPUT_2: + case MBCS_OUTPUT_3: + case MBCS_OUTPUT_4: + case MBCS_OUTPUT_3_EUC: + case MBCS_OUTPUT_4_EUC: + case MBCS_OUTPUT_2_SISO: + /* OK */ + break; + default: + *pErrorCode=U_INVALID_TABLE_FORMAT; + return; + } + if(pArgs->onlyTestIsLoadable) { + /* + * Exit as soon as we know that we can load the converter + * and the format is valid and supported. + * The worst that can happen in the following code is a memory + * allocation error. + */ + return; + } + + mbcsTable->countStates=(uint8_t)header->countStates; + mbcsTable->countToUFallbacks=header->countToUFallbacks; + mbcsTable->stateTable=(const int32_t (*)[256])(raw+headerLength*4); + mbcsTable->toUFallbacks=(const _MBCSToUFallback *)(mbcsTable->stateTable+header->countStates); + mbcsTable->unicodeCodeUnits=(const uint16_t *)(raw+header->offsetToUCodeUnits); + + mbcsTable->fromUnicodeTable=(const uint16_t *)(raw+header->offsetFromUTable); + mbcsTable->fromUnicodeBytes=(const uint8_t *)(raw+header->offsetFromUBytes); + mbcsTable->fromUBytesLength=header->fromUBytesLength; + + /* + * converter versions 6.1 and up contain a unicodeMask that is + * used here to select the most efficient function implementations + */ + info.size=sizeof(UDataInfo); + udata_getInfo((UDataMemory *)sharedData->dataMemory, &info); + if(info.formatVersion[0]>6 || (info.formatVersion[0]==6 && info.formatVersion[1]>=1)) { + /* mask off possible future extensions to be safe */ + mbcsTable->unicodeMask=(uint8_t)(sharedData->staticData->unicodeMask&3); + } else { + /* for older versions, assume worst case: contains anything possible (prevent over-optimizations) */ + mbcsTable->unicodeMask=UCNV_HAS_SUPPLEMENTARY|UCNV_HAS_SURROGATES; + } + + /* + * _MBCSHeader.version 4.3 adds utf8Friendly data structures. + * Check for the header version, SBCS vs. MBCS, and for whether the + * data structures are optimized for code points as high as what the + * runtime code is designed for. + * The implementation does not handle mapping tables with entries for + * unpaired surrogates. + */ + if( header->version[1]>=3 && + (mbcsTable->unicodeMask&UCNV_HAS_SURROGATES)==0 && + (mbcsTable->countStates==1 ? + (header->version[2]>=(SBCS_FAST_MAX>>8)) : + (header->version[2]>=(MBCS_FAST_MAX>>8)) + ) + ) { + mbcsTable->utf8Friendly=true; + + if(mbcsTable->countStates==1) { + /* + * SBCS: Stage 3 is allocated in 64-entry blocks for U+0000..SBCS_FAST_MAX or higher. + * Build a table with indexes to each block, to be used instead of + * the regular stage 1/2 table. + */ + int32_t i; + for(i=0; i<(SBCS_FAST_LIMIT>>6); ++i) { + mbcsTable->sbcsIndex[i]=mbcsTable->fromUnicodeTable[mbcsTable->fromUnicodeTable[i>>4]+((i<<2)&0x3c)]; + } + /* set SBCS_FAST_MAX to reflect the reach of sbcsIndex[] even if header->version[2]>(SBCS_FAST_MAX>>8) */ + mbcsTable->maxFastUChar=SBCS_FAST_MAX; + } else { + /* + * MBCS: Stage 3 is allocated in 64-entry blocks for U+0000..MBCS_FAST_MAX or higher. + * The .cnv file is prebuilt with an additional stage table with indexes + * to each block. + */ + mbcsTable->mbcsIndex=(const uint16_t *) + (mbcsTable->fromUnicodeBytes+ + (noFromU ? 0 : mbcsTable->fromUBytesLength)); + mbcsTable->maxFastUChar=(((char16_t)header->version[2])<<8)|0xff; + } + } + + /* calculate a bit set of 4 ASCII characters per bit that round-trip to ASCII bytes */ + { + uint32_t asciiRoundtrips=0xffffffff; + int32_t i; + + for(i=0; i<0x80; ++i) { + if(mbcsTable->stateTable[0][i]!=MBCS_ENTRY_FINAL(0, MBCS_STATE_VALID_DIRECT_16, i)) { + asciiRoundtrips&=~((uint32_t)1<<(i>>2)); + } + } + mbcsTable->asciiRoundtrips=asciiRoundtrips; + } + + if(noFromU) { + uint32_t stage1Length= + mbcsTable->unicodeMask&UCNV_HAS_SUPPLEMENTARY ? + 0x440 : 0x40; + uint32_t stage2Length= + (header->offsetFromUBytes-header->offsetFromUTable)/4- + stage1Length/2; + reconstituteData(mbcsTable, stage1Length, stage2Length, header->fullStage2Length, pErrorCode); + } + } + + /* Set the impl pointer here so that it is set for both extension-only and base tables. */ + if(mbcsTable->utf8Friendly) { + if(mbcsTable->countStates==1) { + sharedData->impl=&_SBCSUTF8Impl; + } else { + if(mbcsTable->outputType==MBCS_OUTPUT_2) { + sharedData->impl=&_DBCSUTF8Impl; + } + } + } + + if(mbcsTable->outputType==MBCS_OUTPUT_DBCS_ONLY || mbcsTable->outputType==MBCS_OUTPUT_2_SISO) { + /* + * MBCS_OUTPUT_DBCS_ONLY: No SBCS mappings, therefore ASCII does not roundtrip. + * MBCS_OUTPUT_2_SISO: Bypass the ASCII fastpath to handle prevLength correctly. + */ + mbcsTable->asciiRoundtrips=0; + } +} + +static void U_CALLCONV +ucnv_MBCSUnload(UConverterSharedData *sharedData) { + UConverterMBCSTable *mbcsTable=&sharedData->mbcs; + + if(mbcsTable->swapLFNLStateTable!=nullptr) { + uprv_free(mbcsTable->swapLFNLStateTable); + } + if(mbcsTable->stateTableOwned) { + uprv_free((void *)mbcsTable->stateTable); + } + if(mbcsTable->baseSharedData!=nullptr) { + ucnv_unload(mbcsTable->baseSharedData); + } + if(mbcsTable->reconstitutedData!=nullptr) { + uprv_free(mbcsTable->reconstitutedData); + } +} + +static void U_CALLCONV +ucnv_MBCSOpen(UConverter *cnv, + UConverterLoadArgs *pArgs, + UErrorCode *pErrorCode) { + UConverterMBCSTable *mbcsTable; + const int32_t *extIndexes; + uint8_t outputType; + int8_t maxBytesPerUChar; + + if(pArgs->onlyTestIsLoadable) { + return; + } + + mbcsTable=&cnv->sharedData->mbcs; + outputType=mbcsTable->outputType; + + if(outputType==MBCS_OUTPUT_DBCS_ONLY) { + /* the swaplfnl option does not apply, remove it */ + cnv->options=pArgs->options&=~UCNV_OPTION_SWAP_LFNL; + } + + if((pArgs->options&UCNV_OPTION_SWAP_LFNL)!=0) { + /* do this because double-checked locking is broken */ + UBool isCached; + + icu::umtx_lock(nullptr); + isCached=mbcsTable->swapLFNLStateTable!=nullptr; + icu::umtx_unlock(nullptr); + + if(!isCached) { + if(!_EBCDICSwapLFNL(cnv->sharedData, pErrorCode)) { + if(U_FAILURE(*pErrorCode)) { + return; /* something went wrong */ + } + + /* the option does not apply, remove it */ + cnv->options=pArgs->options&=~UCNV_OPTION_SWAP_LFNL; + } + } + } + + if(uprv_strstr(pArgs->name, "18030")!=nullptr) { + if(uprv_strstr(pArgs->name, "gb18030")!=nullptr || uprv_strstr(pArgs->name, "GB18030")!=nullptr) { + /* set a flag for GB 18030 mode, which changes the callback behavior */ + cnv->options|=_MBCS_OPTION_GB18030; + } + } else if((uprv_strstr(pArgs->name, "KEIS")!=nullptr) || (uprv_strstr(pArgs->name, "keis")!=nullptr)) { + /* set a flag for KEIS converter, which changes the SI/SO character sequence */ + cnv->options|=_MBCS_OPTION_KEIS; + } else if((uprv_strstr(pArgs->name, "JEF")!=nullptr) || (uprv_strstr(pArgs->name, "jef")!=nullptr)) { + /* set a flag for JEF converter, which changes the SI/SO character sequence */ + cnv->options|=_MBCS_OPTION_JEF; + } else if((uprv_strstr(pArgs->name, "JIPS")!=nullptr) || (uprv_strstr(pArgs->name, "jips")!=nullptr)) { + /* set a flag for JIPS converter, which changes the SI/SO character sequence */ + cnv->options|=_MBCS_OPTION_JIPS; + } + + /* fix maxBytesPerUChar depending on outputType and options etc. */ + if(outputType==MBCS_OUTPUT_2_SISO) { + cnv->maxBytesPerUChar=3; /* SO+DBCS */ + } + + extIndexes=mbcsTable->extIndexes; + if(extIndexes!=nullptr) { + maxBytesPerUChar=(int8_t)UCNV_GET_MAX_BYTES_PER_UCHAR(extIndexes); + if(outputType==MBCS_OUTPUT_2_SISO) { + ++maxBytesPerUChar; /* SO + multiple DBCS */ + } + + if(maxBytesPerUChar>cnv->maxBytesPerUChar) { + cnv->maxBytesPerUChar=maxBytesPerUChar; + } + } + +#if 0 + /* + * documentation of UConverter fields used for status + * all of these fields are (re)set to 0 by ucnv_bld.c and ucnv_reset() + */ + + /* toUnicode */ + cnv->toUnicodeStatus=0; /* offset */ + cnv->mode=0; /* state */ + cnv->toULength=0; /* byteIndex */ + + /* fromUnicode */ + cnv->fromUChar32=0; + cnv->fromUnicodeStatus=1; /* prevLength */ +#endif +} + +U_CDECL_BEGIN + +static const char* U_CALLCONV +ucnv_MBCSGetName(const UConverter *cnv) { + if((cnv->options&UCNV_OPTION_SWAP_LFNL)!=0 && cnv->sharedData->mbcs.swapLFNLName!=nullptr) { + return cnv->sharedData->mbcs.swapLFNLName; + } else { + return cnv->sharedData->staticData->name; + } +} +U_CDECL_END + + +/* MBCS-to-Unicode conversion functions ------------------------------------- */ + +static UChar32 U_CALLCONV +ucnv_MBCSGetFallback(UConverterMBCSTable *mbcsTable, uint32_t offset) { + const _MBCSToUFallback *toUFallbacks; + uint32_t i, start, limit; + + limit=mbcsTable->countToUFallbacks; + if(limit>0) { + /* do a binary search for the fallback mapping */ + toUFallbacks=mbcsTable->toUFallbacks; + start=0; + while(startconverter; + source=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + target=pArgs->target; + targetLimit=pArgs->targetLimit; + offsets=pArgs->offsets; + + if((cnv->options&UCNV_OPTION_SWAP_LFNL)!=0) { + stateTable=(const int32_t (*)[256])cnv->sharedData->mbcs.swapLFNLStateTable; + } else { + stateTable=cnv->sharedData->mbcs.stateTable; + } + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex=0; + + /* conversion loop */ + while(source=targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + + entry=stateTable[0][*source++]; + /* MBCS_ENTRY_IS_FINAL(entry) */ + + /* test the most common case first */ + if(MBCS_ENTRY_FINAL_IS_VALID_DIRECT_16(entry)) { + /* output BMP code point */ + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + + /* normal end of action codes: prepare for a new character */ + ++sourceIndex; + continue; + } + + /* + * An if-else-if chain provides more reliable performance for + * the most common cases compared to a switch. + */ + action=(uint8_t)(MBCS_ENTRY_FINAL_ACTION(entry)); + if(action==MBCS_STATE_VALID_DIRECT_20 || + (action==MBCS_STATE_FALLBACK_DIRECT_20 && UCNV_TO_U_USE_FALLBACK(cnv)) + ) { + entry=MBCS_ENTRY_FINAL_VALUE(entry); + /* output surrogate pair */ + *target++=(char16_t)(0xd800|(char16_t)(entry>>10)); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + c=(char16_t)(0xdc00|(char16_t)(entry&0x3ff)); + if(targetUCharErrorBuffer[0]=c; + cnv->UCharErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + + ++sourceIndex; + continue; + } else if(action==MBCS_STATE_FALLBACK_DIRECT_16) { + if(UCNV_TO_U_USE_FALLBACK(cnv)) { + /* output BMP code point */ + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + + ++sourceIndex; + continue; + } + } else if(action==MBCS_STATE_UNASSIGNED) { + /* just fall through */ + } else if(action==MBCS_STATE_ILLEGAL) { + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } else { + /* reserved, must never occur */ + ++sourceIndex; + continue; + } + + if(U_FAILURE(*pErrorCode)) { + /* callback(illegal) */ + break; + } else /* unassigned sequences indicated with byteIndex>0 */ { + /* try an extension mapping */ + pArgs->source=(const char *)source; + cnv->toUBytes[0]=*(source-1); + cnv->toULength=_extToU(cnv, cnv->sharedData, + 1, &source, sourceLimit, + &target, targetLimit, + &offsets, sourceIndex, + pArgs->flush, + pErrorCode); + sourceIndex+=1+(int32_t)(source-(const uint8_t *)pArgs->source); + + if(U_FAILURE(*pErrorCode)) { + /* not mappable or buffer overflow */ + break; + } + } + } + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + pArgs->offsets=offsets; +} + +/* + * This version of ucnv_MBCSSingleToUnicodeWithOffsets() is optimized for single-byte, single-state codepages + * that only map to and from the BMP. + * In addition to single-byte optimizations, the offset calculations + * become much easier. + */ +static void +ucnv_MBCSSingleToBMPWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const uint8_t *source, *sourceLimit, *lastSource; + char16_t *target; + int32_t targetCapacity, length; + int32_t *offsets; + + const int32_t (*stateTable)[256]; + + int32_t sourceIndex; + + int32_t entry; + uint8_t action; + + /* set up the local pointers */ + cnv=pArgs->converter; + source=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + target=pArgs->target; + targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); + offsets=pArgs->offsets; + + if((cnv->options&UCNV_OPTION_SWAP_LFNL)!=0) { + stateTable=(const int32_t (*)[256])cnv->sharedData->mbcs.swapLFNLStateTable; + } else { + stateTable=cnv->sharedData->mbcs.stateTable; + } + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex=0; + lastSource=source; + + /* + * since the conversion here is 1:1 char16_t:uint8_t, we need only one counter + * for the minimum of the sourceLength and targetCapacity + */ + length=(int32_t)(sourceLimit-source); + if(length=16) { + int32_t count, loops, oredEntries; + + loops=count=targetCapacity>>4; + do { + oredEntries=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + oredEntries|=entry=stateTable[0][*source++]; + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + + /* were all 16 entries really valid? */ + if(!MBCS_ENTRY_FINAL_IS_VALID_DIRECT_16(oredEntries)) { + /* no, return to the first of these 16 */ + source-=16; + target-=16; + break; + } + } while(--count>0); + count=loops-count; + targetCapacity-=16*count; + + if(offsets!=nullptr) { + lastSource+=16*count; + while(count>0) { + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + --count; + } + } + } +#endif + + /* conversion loop */ + while(targetCapacity > 0 && source < sourceLimit) { + entry=stateTable[0][*source++]; + /* MBCS_ENTRY_IS_FINAL(entry) */ + + /* test the most common case first */ + if(MBCS_ENTRY_FINAL_IS_VALID_DIRECT_16(entry)) { + /* output BMP code point */ + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + --targetCapacity; + continue; + } + + /* + * An if-else-if chain provides more reliable performance for + * the most common cases compared to a switch. + */ + action=(uint8_t)(MBCS_ENTRY_FINAL_ACTION(entry)); + if(action==MBCS_STATE_FALLBACK_DIRECT_16) { + if(UCNV_TO_U_USE_FALLBACK(cnv)) { + /* output BMP code point */ + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + --targetCapacity; + continue; + } + } else if(action==MBCS_STATE_UNASSIGNED) { + /* just fall through */ + } else if(action==MBCS_STATE_ILLEGAL) { + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } else { + /* reserved, must never occur */ + continue; + } + + /* set offsets since the start or the last extension */ + if(offsets!=nullptr) { + int32_t count=(int32_t)(source-lastSource); + + /* predecrement: do not set the offset for the callback-causing character */ + while(--count>0) { + *offsets++=sourceIndex++; + } + /* offset and sourceIndex are now set for the current character */ + } + + if(U_FAILURE(*pErrorCode)) { + /* callback(illegal) */ + break; + } else /* unassigned sequences indicated with byteIndex>0 */ { + /* try an extension mapping */ + lastSource=source; + cnv->toUBytes[0]=*(source-1); + cnv->toULength=_extToU(cnv, cnv->sharedData, + 1, &source, sourceLimit, + &target, pArgs->targetLimit, + &offsets, sourceIndex, + pArgs->flush, + pErrorCode); + sourceIndex+=1+(int32_t)(source-lastSource); + + if(U_FAILURE(*pErrorCode)) { + /* not mappable or buffer overflow */ + break; + } + + /* recalculate the targetCapacity after an extension mapping */ + targetCapacity=(int32_t)(pArgs->targetLimit-target); + length=(int32_t)(sourceLimit-source); + if(length=pArgs->targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + + /* set offsets since the start or the last callback */ + if(offsets!=nullptr) { + size_t count=source-lastSource; + while(count>0) { + *offsets++=sourceIndex++; + --count; + } + } + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + pArgs->offsets=offsets; +} + +static UBool +hasValidTrailBytes(const int32_t (*stateTable)[256], uint8_t state) { + const int32_t *row=stateTable[state]; + int32_t b, entry; + /* First test for final entries in this state for some commonly valid byte values. */ + entry=row[0xa1]; + if( !MBCS_ENTRY_IS_TRANSITION(entry) && + MBCS_ENTRY_FINAL_ACTION(entry)!=MBCS_STATE_ILLEGAL + ) { + return true; + } + entry=row[0x41]; + if( !MBCS_ENTRY_IS_TRANSITION(entry) && + MBCS_ENTRY_FINAL_ACTION(entry)!=MBCS_STATE_ILLEGAL + ) { + return true; + } + /* Then test for final entries in this state. */ + for(b=0; b<=0xff; ++b) { + entry=row[b]; + if( !MBCS_ENTRY_IS_TRANSITION(entry) && + MBCS_ENTRY_FINAL_ACTION(entry)!=MBCS_STATE_ILLEGAL + ) { + return true; + } + } + /* Then recurse for transition entries. */ + for(b=0; b<=0xff; ++b) { + entry=row[b]; + if( MBCS_ENTRY_IS_TRANSITION(entry) && + hasValidTrailBytes(stateTable, (uint8_t)MBCS_ENTRY_TRANSITION_STATE(entry)) + ) { + return true; + } + } + return false; +} + +/* + * Is byte b a single/lead byte in this state? + * Recurse for transition states, because here we don't want to say that + * b is a lead byte if all byte sequences that start with b are illegal. + */ +static UBool +isSingleOrLead(const int32_t (*stateTable)[256], uint8_t state, UBool isDBCSOnly, uint8_t b) { + const int32_t *row=stateTable[state]; + int32_t entry=row[b]; + if(MBCS_ENTRY_IS_TRANSITION(entry)) { /* lead byte */ + return hasValidTrailBytes(stateTable, (uint8_t)MBCS_ENTRY_TRANSITION_STATE(entry)); + } else { + uint8_t action=(uint8_t)(MBCS_ENTRY_FINAL_ACTION(entry)); + if(action==MBCS_STATE_CHANGE_ONLY && isDBCSOnly) { + return false; /* SI/SO are illegal for DBCS-only conversion */ + } else { + return action!=MBCS_STATE_ILLEGAL; + } + } +} + +U_CFUNC void +ucnv_MBCSToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const uint8_t *source, *sourceLimit; + char16_t *target; + const char16_t *targetLimit; + int32_t *offsets; + + const int32_t (*stateTable)[256]; + const uint16_t *unicodeCodeUnits; + + uint32_t offset; + uint8_t state; + int8_t byteIndex; + uint8_t *bytes; + + int32_t sourceIndex, nextSourceIndex; + + int32_t entry; + char16_t c; + uint8_t action; + + /* use optimized function if possible */ + cnv=pArgs->converter; + + if(cnv->preToULength>0) { + /* + * pass sourceIndex=-1 because we continue from an earlier buffer + * in the future, this may change with continuous offsets + */ + ucnv_extContinueMatchToU(cnv, pArgs, -1, pErrorCode); + + if(U_FAILURE(*pErrorCode) || cnv->preToULength<0) { + return; + } + } + + if(cnv->sharedData->mbcs.countStates==1) { + if(!(cnv->sharedData->mbcs.unicodeMask&UCNV_HAS_SUPPLEMENTARY)) { + ucnv_MBCSSingleToBMPWithOffsets(pArgs, pErrorCode); + } else { + ucnv_MBCSSingleToUnicodeWithOffsets(pArgs, pErrorCode); + } + return; + } + + /* set up the local pointers */ + source=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + target=pArgs->target; + targetLimit=pArgs->targetLimit; + offsets=pArgs->offsets; + + if((cnv->options&UCNV_OPTION_SWAP_LFNL)!=0) { + stateTable=(const int32_t (*)[256])cnv->sharedData->mbcs.swapLFNLStateTable; + } else { + stateTable=cnv->sharedData->mbcs.stateTable; + } + unicodeCodeUnits=cnv->sharedData->mbcs.unicodeCodeUnits; + + /* get the converter state from UConverter */ + offset=cnv->toUnicodeStatus; + byteIndex=cnv->toULength; + bytes=cnv->toUBytes; + + /* + * if we are in the SBCS state for a DBCS-only converter, + * then load the DBCS state from the MBCS data + * (dbcsOnlyState==0 if it is not a DBCS-only converter) + */ + if((state=(uint8_t)(cnv->mode))==0) { + state=cnv->sharedData->mbcs.dbcsOnlyState; + } + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex=byteIndex==0 ? 0 : -1; + nextSourceIndex=0; + + /* conversion loop */ + while(source=targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + + if(byteIndex==0) { + /* optimized loop for 1/2-byte input and BMP output */ + if(offsets==nullptr) { + do { + entry=stateTable[state][*source]; + if(MBCS_ENTRY_IS_TRANSITION(entry)) { + state=(uint8_t)MBCS_ENTRY_TRANSITION_STATE(entry); + offset=MBCS_ENTRY_TRANSITION_OFFSET(entry); + + ++source; + if( source=sourceLimit) { + break; + } + if(target>=targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + + ++nextSourceIndex; + bytes[byteIndex++]=*source++; + } else /* byteIndex>0 */ { + ++nextSourceIndex; + entry=stateTable[state][bytes[byteIndex++]=*source++]; + } + + if(MBCS_ENTRY_IS_TRANSITION(entry)) { + state=(uint8_t)MBCS_ENTRY_TRANSITION_STATE(entry); + offset+=MBCS_ENTRY_TRANSITION_OFFSET(entry); + continue; + } + + /* save the previous state for proper extension mapping with SI/SO-stateful converters */ + cnv->mode=state; + + /* set the next state early so that we can reuse the entry variable */ + state=(uint8_t)MBCS_ENTRY_FINAL_STATE(entry); /* typically 0 */ + + /* + * An if-else-if chain provides more reliable performance for + * the most common cases compared to a switch. + */ + action=(uint8_t)(MBCS_ENTRY_FINAL_ACTION(entry)); + if(action==MBCS_STATE_VALID_16) { + offset+=MBCS_ENTRY_FINAL_VALUE_16(entry); + c=unicodeCodeUnits[offset]; + if(c<0xfffe) { + /* output BMP code point */ + *target++=c; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + byteIndex=0; + } else if(c==0xfffe) { + if(UCNV_TO_U_USE_FALLBACK(cnv) && (entry=(int32_t)ucnv_MBCSGetFallback(&cnv->sharedData->mbcs, offset))!=0xfffe) { + /* output fallback BMP code point */ + *target++=(char16_t)entry; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + byteIndex=0; + } + } else { + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + } else if(action==MBCS_STATE_VALID_DIRECT_16) { + /* output BMP code point */ + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + byteIndex=0; + } else if(action==MBCS_STATE_VALID_16_PAIR) { + offset+=MBCS_ENTRY_FINAL_VALUE_16(entry); + c=unicodeCodeUnits[offset++]; + if(c<0xd800) { + /* output BMP code point below 0xd800 */ + *target++=c; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + byteIndex=0; + } else if(UCNV_TO_U_USE_FALLBACK(cnv) ? c<=0xdfff : c<=0xdbff) { + /* output roundtrip or fallback surrogate pair */ + *target++=(char16_t)(c&0xdbff); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + byteIndex=0; + if(targetUCharErrorBuffer[0]=unicodeCodeUnits[offset]; + cnv->UCharErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + + offset=0; + break; + } + } else if(UCNV_TO_U_USE_FALLBACK(cnv) ? (c&0xfffe)==0xe000 : c==0xe000) { + /* output roundtrip BMP code point above 0xd800 or fallback BMP code point */ + *target++=unicodeCodeUnits[offset]; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + byteIndex=0; + } else if(c==0xffff) { + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + } else if(action==MBCS_STATE_VALID_DIRECT_20 || + (action==MBCS_STATE_FALLBACK_DIRECT_20 && UCNV_TO_U_USE_FALLBACK(cnv)) + ) { + entry=MBCS_ENTRY_FINAL_VALUE(entry); + /* output surrogate pair */ + *target++=(char16_t)(0xd800|(char16_t)(entry>>10)); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + byteIndex=0; + c=(char16_t)(0xdc00|(char16_t)(entry&0x3ff)); + if(targetUCharErrorBuffer[0]=c; + cnv->UCharErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + + offset=0; + break; + } + } else if(action==MBCS_STATE_CHANGE_ONLY) { + /* + * This serves as a state change without any output. + * It is useful for reading simple stateful encodings, + * for example using just Shift-In/Shift-Out codes. + * The 21 unused bits may later be used for more sophisticated + * state transitions. + */ + if(cnv->sharedData->mbcs.dbcsOnlyState==0) { + byteIndex=0; + } else { + /* SI/SO are illegal for DBCS-only conversion */ + state=(uint8_t)(cnv->mode); /* restore the previous state */ + + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + } else if(action==MBCS_STATE_FALLBACK_DIRECT_16) { + if(UCNV_TO_U_USE_FALLBACK(cnv)) { + /* output BMP code point */ + *target++=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + byteIndex=0; + } + } else if(action==MBCS_STATE_UNASSIGNED) { + /* just fall through */ + } else if(action==MBCS_STATE_ILLEGAL) { + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } else { + /* reserved, must never occur */ + byteIndex=0; + } + + /* end of action codes: prepare for a new character */ + offset=0; + + if(byteIndex==0) { + sourceIndex=nextSourceIndex; + } else if(U_FAILURE(*pErrorCode)) { + /* callback(illegal) */ + if(byteIndex>1) { + /* + * Ticket 5691: consistent illegal sequences: + * - We include at least the first byte in the illegal sequence. + * - If any of the non-initial bytes could be the start of a character, + * we stop the illegal sequence before the first one of those. + */ + UBool isDBCSOnly=(UBool)(cnv->sharedData->mbcs.dbcsOnlyState!=0); + int8_t i; + for(i=1; + isource); + byteIndex=i; /* length of reported illegal byte sequence */ + if(backOutDistance<=bytesFromThisBuffer) { + source-=backOutDistance; + } else { + /* Back out bytes from the previous buffer: Need to replay them. */ + cnv->preToULength=(int8_t)(bytesFromThisBuffer-backOutDistance); + /* preToULength is negative! */ + uprv_memcpy(cnv->preToU, bytes+i, -cnv->preToULength); + source=(const uint8_t *)pArgs->source; + } + } + } + break; + } else /* unassigned sequences indicated with byteIndex>0 */ { + /* try an extension mapping */ + pArgs->source=(const char *)source; + byteIndex=_extToU(cnv, cnv->sharedData, + byteIndex, &source, sourceLimit, + &target, targetLimit, + &offsets, sourceIndex, + pArgs->flush, + pErrorCode); + sourceIndex=nextSourceIndex+=(int32_t)(source-(const uint8_t *)pArgs->source); + + if(U_FAILURE(*pErrorCode)) { + /* not mappable or buffer overflow */ + break; + } + } + } + + /* set the converter state back into UConverter */ + cnv->toUnicodeStatus=offset; + cnv->mode=state; + cnv->toULength=byteIndex; + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + pArgs->offsets=offsets; +} + +/* + * This version of ucnv_MBCSGetNextUChar() is optimized for single-byte, single-state codepages. + * We still need a conversion loop in case we find reserved action codes, which are to be ignored. + */ +static UChar32 +ucnv_MBCSSingleGetNextUChar(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const int32_t (*stateTable)[256]; + const uint8_t *source, *sourceLimit; + + int32_t entry; + uint8_t action; + + /* set up the local pointers */ + cnv=pArgs->converter; + source=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + if((cnv->options&UCNV_OPTION_SWAP_LFNL)!=0) { + stateTable=(const int32_t (*)[256])cnv->sharedData->mbcs.swapLFNLStateTable; + } else { + stateTable=cnv->sharedData->mbcs.stateTable; + } + + /* conversion loop */ + while(sourcesource=(const char *)source; + + if(MBCS_ENTRY_FINAL_IS_VALID_DIRECT_16(entry)) { + /* output BMP code point */ + return (char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + } + + /* + * An if-else-if chain provides more reliable performance for + * the most common cases compared to a switch. + */ + action=(uint8_t)(MBCS_ENTRY_FINAL_ACTION(entry)); + if( action==MBCS_STATE_VALID_DIRECT_20 || + (action==MBCS_STATE_FALLBACK_DIRECT_20 && UCNV_TO_U_USE_FALLBACK(cnv)) + ) { + /* output supplementary code point */ + return (UChar32)(MBCS_ENTRY_FINAL_VALUE(entry)+0x10000); + } else if(action==MBCS_STATE_FALLBACK_DIRECT_16) { + if(UCNV_TO_U_USE_FALLBACK(cnv)) { + /* output BMP code point */ + return (char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + } + } else if(action==MBCS_STATE_UNASSIGNED) { + /* just fall through */ + } else if(action==MBCS_STATE_ILLEGAL) { + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } else { + /* reserved, must never occur */ + continue; + } + + if(U_FAILURE(*pErrorCode)) { + /* callback(illegal) */ + break; + } else /* unassigned sequence */ { + /* defer to the generic implementation */ + pArgs->source=(const char *)source-1; + return UCNV_GET_NEXT_UCHAR_USE_TO_U; + } + } + + /* no output because of empty input or only state changes */ + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0xffff; +} + +/* + * Version of _MBCSToUnicodeWithOffsets() optimized for single-character + * conversion without offset handling. + * + * When a character does not have a mapping to Unicode, then we return to the + * generic ucnv_getNextUChar() code for extension/GB 18030 and error/callback + * handling. + * We also defer to the generic code in other complicated cases and have them + * ultimately handled by _MBCSToUnicodeWithOffsets() itself. + * + * All normal mappings and errors are handled here. + */ +static UChar32 U_CALLCONV +ucnv_MBCSGetNextUChar(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const uint8_t *source, *sourceLimit, *lastSource; + + const int32_t (*stateTable)[256]; + const uint16_t *unicodeCodeUnits; + + uint32_t offset; + uint8_t state; + + int32_t entry; + UChar32 c; + uint8_t action; + + /* use optimized function if possible */ + cnv=pArgs->converter; + + if(cnv->preToULength>0) { + /* use the generic code in ucnv_getNextUChar() to continue with a partial match */ + return UCNV_GET_NEXT_UCHAR_USE_TO_U; + } + + if(cnv->sharedData->mbcs.unicodeMask&UCNV_HAS_SURROGATES) { + /* + * Using the generic ucnv_getNextUChar() code lets us deal correctly + * with the rare case of a codepage that maps single surrogates + * without adding the complexity to this already complicated function here. + */ + return UCNV_GET_NEXT_UCHAR_USE_TO_U; + } else if(cnv->sharedData->mbcs.countStates==1) { + return ucnv_MBCSSingleGetNextUChar(pArgs, pErrorCode); + } + + /* set up the local pointers */ + source=lastSource=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + + if((cnv->options&UCNV_OPTION_SWAP_LFNL)!=0) { + stateTable=(const int32_t (*)[256])cnv->sharedData->mbcs.swapLFNLStateTable; + } else { + stateTable=cnv->sharedData->mbcs.stateTable; + } + unicodeCodeUnits=cnv->sharedData->mbcs.unicodeCodeUnits; + + /* get the converter state from UConverter */ + offset=cnv->toUnicodeStatus; + + /* + * if we are in the SBCS state for a DBCS-only converter, + * then load the DBCS state from the MBCS data + * (dbcsOnlyState==0 if it is not a DBCS-only converter) + */ + if((state=(uint8_t)(cnv->mode))==0) { + state=cnv->sharedData->mbcs.dbcsOnlyState; + } + + /* conversion loop */ + c=U_SENTINEL; + while(sourcemode=state; + + /* set the next state early so that we can reuse the entry variable */ + state=(uint8_t)MBCS_ENTRY_FINAL_STATE(entry); /* typically 0 */ + + /* + * An if-else-if chain provides more reliable performance for + * the most common cases compared to a switch. + */ + action=(uint8_t)(MBCS_ENTRY_FINAL_ACTION(entry)); + if(action==MBCS_STATE_VALID_DIRECT_16) { + /* output BMP code point */ + c=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + break; + } else if(action==MBCS_STATE_VALID_16) { + offset+=MBCS_ENTRY_FINAL_VALUE_16(entry); + c=unicodeCodeUnits[offset]; + if(c<0xfffe) { + /* output BMP code point */ + break; + } else if(c==0xfffe) { + if(UCNV_TO_U_USE_FALLBACK(cnv) && (c=ucnv_MBCSGetFallback(&cnv->sharedData->mbcs, offset))!=0xfffe) { + break; + } + } else { + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + } else if(action==MBCS_STATE_VALID_16_PAIR) { + offset+=MBCS_ENTRY_FINAL_VALUE_16(entry); + c=unicodeCodeUnits[offset++]; + if(c<0xd800) { + /* output BMP code point below 0xd800 */ + break; + } else if(UCNV_TO_U_USE_FALLBACK(cnv) ? c<=0xdfff : c<=0xdbff) { + /* output roundtrip or fallback supplementary code point */ + c=((c&0x3ff)<<10)+unicodeCodeUnits[offset]+(0x10000-0xdc00); + break; + } else if(UCNV_TO_U_USE_FALLBACK(cnv) ? (c&0xfffe)==0xe000 : c==0xe000) { + /* output roundtrip BMP code point above 0xd800 or fallback BMP code point */ + c=unicodeCodeUnits[offset]; + break; + } else if(c==0xffff) { + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + } else if(action==MBCS_STATE_VALID_DIRECT_20 || + (action==MBCS_STATE_FALLBACK_DIRECT_20 && UCNV_TO_U_USE_FALLBACK(cnv)) + ) { + /* output supplementary code point */ + c=(UChar32)(MBCS_ENTRY_FINAL_VALUE(entry)+0x10000); + break; + } else if(action==MBCS_STATE_CHANGE_ONLY) { + /* + * This serves as a state change without any output. + * It is useful for reading simple stateful encodings, + * for example using just Shift-In/Shift-Out codes. + * The 21 unused bits may later be used for more sophisticated + * state transitions. + */ + if(cnv->sharedData->mbcs.dbcsOnlyState!=0) { + /* SI/SO are illegal for DBCS-only conversion */ + state=(uint8_t)(cnv->mode); /* restore the previous state */ + + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } + } else if(action==MBCS_STATE_FALLBACK_DIRECT_16) { + if(UCNV_TO_U_USE_FALLBACK(cnv)) { + /* output BMP code point */ + c=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + break; + } + } else if(action==MBCS_STATE_UNASSIGNED) { + /* just fall through */ + } else if(action==MBCS_STATE_ILLEGAL) { + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + } else { + /* reserved (must never occur), or only state change */ + offset=0; + lastSource=source; + continue; + } + + /* end of action codes: prepare for a new character */ + offset=0; + + if(U_FAILURE(*pErrorCode)) { + /* callback(illegal) */ + break; + } else /* unassigned sequence */ { + /* defer to the generic implementation */ + cnv->toUnicodeStatus=0; + cnv->mode=state; + pArgs->source=(const char *)lastSource; + return UCNV_GET_NEXT_UCHAR_USE_TO_U; + } + } + } + + if(c<0) { + if(U_SUCCESS(*pErrorCode) && source==sourceLimit && lastSourcetoUBytes; + cnv->toULength=(int8_t)(source-lastSource); + do { + *bytes++=*lastSource++; + } while(lastSourcesharedData->mbcs.dbcsOnlyState!=0); + uint8_t *bytes=cnv->toUBytes; + *bytes++=*lastSource++; /* first byte */ + if(lastSource==source) { + cnv->toULength=1; + } else /* lastSourcetoULength=i; + source=lastSource; + } + } else { + /* no output because of empty input or only state changes */ + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + } + c=0xffff; + } + + /* set the converter state back into UConverter, ready for a new character */ + cnv->toUnicodeStatus=0; + cnv->mode=state; + + /* write back the updated pointer */ + pArgs->source=(const char *)source; + return c; +} + +#if 0 +/* + * Code disabled 2002dec09 (ICU 2.4) because it is not currently used in ICU. markus + * Removal improves code coverage. + */ +/** + * This version of ucnv_MBCSSimpleGetNextUChar() is optimized for single-byte, single-state codepages. + * It does not handle the EBCDIC swaplfnl option (set in UConverter). + * It does not handle conversion extensions (_extToU()). + */ +U_CFUNC UChar32 +ucnv_MBCSSingleSimpleGetNextUChar(UConverterSharedData *sharedData, + uint8_t b, UBool useFallback) { + int32_t entry; + uint8_t action; + + entry=sharedData->mbcs.stateTable[0][b]; + /* MBCS_ENTRY_IS_FINAL(entry) */ + + if(MBCS_ENTRY_FINAL_IS_VALID_DIRECT_16(entry)) { + /* output BMP code point */ + return (char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + } + + /* + * An if-else-if chain provides more reliable performance for + * the most common cases compared to a switch. + */ + action=(uint8_t)(MBCS_ENTRY_FINAL_ACTION(entry)); + if(action==MBCS_STATE_VALID_DIRECT_20) { + /* output supplementary code point */ + return 0x10000+MBCS_ENTRY_FINAL_VALUE(entry); + } else if(action==MBCS_STATE_FALLBACK_DIRECT_16) { + if(!TO_U_USE_FALLBACK(useFallback)) { + return 0xfffe; + } + /* output BMP code point */ + return (char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + } else if(action==MBCS_STATE_FALLBACK_DIRECT_20) { + if(!TO_U_USE_FALLBACK(useFallback)) { + return 0xfffe; + } + /* output supplementary code point */ + return 0x10000+MBCS_ENTRY_FINAL_VALUE(entry); + } else if(action==MBCS_STATE_UNASSIGNED) { + return 0xfffe; + } else if(action==MBCS_STATE_ILLEGAL) { + return 0xffff; + } else { + /* reserved, must never occur */ + return 0xffff; + } +} +#endif + +/* + * This is a simple version of _MBCSGetNextUChar() that is used + * by other converter implementations. + * It only returns an "assigned" result if it consumes the entire input. + * It does not use state from the converter, nor error codes. + * It does not handle the EBCDIC swaplfnl option (set in UConverter). + * It handles conversion extensions but not GB 18030. + * + * Return value: + * U+fffe unassigned + * U+ffff illegal + * otherwise the Unicode code point + */ +U_CFUNC UChar32 +ucnv_MBCSSimpleGetNextUChar(UConverterSharedData *sharedData, + const char *source, int32_t length, + UBool useFallback) { + const int32_t (*stateTable)[256]; + const uint16_t *unicodeCodeUnits; + + uint32_t offset; + uint8_t state, action; + + UChar32 c; + int32_t i, entry; + + if(length<=0) { + /* no input at all: "illegal" */ + return 0xffff; + } + +#if 0 +/* + * Code disabled 2002dec09 (ICU 2.4) because it is not currently used in ICU. markus + * TODO In future releases, verify that this function is never called for SBCS + * conversions, i.e., that sharedData->mbcs.countStates==1 is still true. + * Removal improves code coverage. + */ + /* use optimized function if possible */ + if(sharedData->mbcs.countStates==1) { + if(length==1) { + return ucnv_MBCSSingleSimpleGetNextUChar(sharedData, (uint8_t)*source, useFallback); + } else { + return 0xffff; /* illegal: more than a single byte for an SBCS converter */ + } + } +#endif + + /* set up the local pointers */ + stateTable=sharedData->mbcs.stateTable; + unicodeCodeUnits=sharedData->mbcs.unicodeCodeUnits; + + /* converter state */ + offset=0; + state=sharedData->mbcs.dbcsOnlyState; + + /* conversion loop */ + for(i=0;;) { + entry=stateTable[state][(uint8_t)source[i++]]; + if(MBCS_ENTRY_IS_TRANSITION(entry)) { + state=(uint8_t)MBCS_ENTRY_TRANSITION_STATE(entry); + offset+=MBCS_ENTRY_TRANSITION_OFFSET(entry); + + if(i==length) { + return 0xffff; /* truncated character */ + } + } else { + /* + * An if-else-if chain provides more reliable performance for + * the most common cases compared to a switch. + */ + action=(uint8_t)(MBCS_ENTRY_FINAL_ACTION(entry)); + if(action==MBCS_STATE_VALID_16) { + offset+=MBCS_ENTRY_FINAL_VALUE_16(entry); + c=unicodeCodeUnits[offset]; + if(c!=0xfffe) { + /* done */ + } else if(UCNV_TO_U_USE_FALLBACK(cnv)) { + c=ucnv_MBCSGetFallback(&sharedData->mbcs, offset); + /* else done with 0xfffe */ + } + break; + } else if(action==MBCS_STATE_VALID_DIRECT_16) { + /* output BMP code point */ + c=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + break; + } else if(action==MBCS_STATE_VALID_16_PAIR) { + offset+=MBCS_ENTRY_FINAL_VALUE_16(entry); + c=unicodeCodeUnits[offset++]; + if(c<0xd800) { + /* output BMP code point below 0xd800 */ + } else if(UCNV_TO_U_USE_FALLBACK(cnv) ? c<=0xdfff : c<=0xdbff) { + /* output roundtrip or fallback supplementary code point */ + c=(UChar32)(((c&0x3ff)<<10)+unicodeCodeUnits[offset]+(0x10000-0xdc00)); + } else if(UCNV_TO_U_USE_FALLBACK(cnv) ? (c&0xfffe)==0xe000 : c==0xe000) { + /* output roundtrip BMP code point above 0xd800 or fallback BMP code point */ + c=unicodeCodeUnits[offset]; + } else if(c==0xffff) { + return 0xffff; + } else { + c=0xfffe; + } + break; + } else if(action==MBCS_STATE_VALID_DIRECT_20) { + /* output supplementary code point */ + c=0x10000+MBCS_ENTRY_FINAL_VALUE(entry); + break; + } else if(action==MBCS_STATE_FALLBACK_DIRECT_16) { + if(!TO_U_USE_FALLBACK(useFallback)) { + c=0xfffe; + break; + } + /* output BMP code point */ + c=(char16_t)MBCS_ENTRY_FINAL_VALUE_16(entry); + break; + } else if(action==MBCS_STATE_FALLBACK_DIRECT_20) { + if(!TO_U_USE_FALLBACK(useFallback)) { + c=0xfffe; + break; + } + /* output supplementary code point */ + c=0x10000+MBCS_ENTRY_FINAL_VALUE(entry); + break; + } else if(action==MBCS_STATE_UNASSIGNED) { + c=0xfffe; + break; + } + + /* + * forbid MBCS_STATE_CHANGE_ONLY for this function, + * and MBCS_STATE_ILLEGAL and reserved action codes + */ + return 0xffff; + } + } + + if(i!=length) { + /* illegal for this function: not all input consumed */ + return 0xffff; + } + + if(c==0xfffe) { + /* try an extension mapping */ + const int32_t *cx=sharedData->mbcs.extIndexes; + if(cx!=nullptr) { + return ucnv_extSimpleMatchToU(cx, source, length, useFallback); + } + } + + return c; +} + +/* MBCS-from-Unicode conversion functions ----------------------------------- */ + +/* This version of ucnv_MBCSFromUnicodeWithOffsets() is optimized for double-byte codepages. */ +static void +ucnv_MBCSDoubleFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const char16_t *source, *sourceLimit; + uint8_t *target; + int32_t targetCapacity; + int32_t *offsets; + + const uint16_t *table; + const uint16_t *mbcsIndex; + const uint8_t *bytes; + + UChar32 c; + + int32_t sourceIndex, nextSourceIndex; + + uint32_t stage2Entry; + uint32_t asciiRoundtrips; + uint32_t value; + uint8_t unicodeMask; + + /* use optimized function if possible */ + cnv=pArgs->converter; + unicodeMask=cnv->sharedData->mbcs.unicodeMask; + + /* set up the local pointers */ + source=pArgs->source; + sourceLimit=pArgs->sourceLimit; + target=(uint8_t *)pArgs->target; + targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); + offsets=pArgs->offsets; + + table=cnv->sharedData->mbcs.fromUnicodeTable; + mbcsIndex=cnv->sharedData->mbcs.mbcsIndex; + if((cnv->options&UCNV_OPTION_SWAP_LFNL)!=0) { + bytes=cnv->sharedData->mbcs.swapLFNLFromUnicodeBytes; + } else { + bytes=cnv->sharedData->mbcs.fromUnicodeBytes; + } + asciiRoundtrips=cnv->sharedData->mbcs.asciiRoundtrips; + + /* get the converter state from UConverter */ + c=cnv->fromUChar32; + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex= c==0 ? 0 : -1; + nextSourceIndex=0; + + /* conversion loop */ + if(c!=0 && targetCapacity>0) { + goto getTrail; + } + + while(source0) { + /* + * Get a correct Unicode code point: + * a single char16_t for a BMP code point or + * a matched surrogate pair for a "supplementary code point". + */ + c=*source++; + ++nextSourceIndex; + if(c<=0x7f && IS_ASCII_ROUNDTRIP(c, asciiRoundtrips)) { + *target++=(uint8_t)c; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + sourceIndex=nextSourceIndex; + } + --targetCapacity; + c=0; + continue; + } + /* + * utf8Friendly table: Test for <=0xd7ff rather than <=MBCS_FAST_MAX + * to avoid dealing with surrogates. + * MBCS_FAST_MAX must be >=0xd7ff. + */ + if(c<=0xd7ff) { + value=DBCS_RESULT_FROM_MOST_BMP(mbcsIndex, (const uint16_t *)bytes, c); + /* There are only roundtrips (!=0) and no-mapping (==0) entries. */ + if(value==0) { + goto unassigned; + } + /* output the value */ + } else { + /* + * This also tests if the codepage maps single surrogates. + * If it does, then surrogates are not paired but mapped separately. + * Note that in this case unmatched surrogates are not detected. + */ + if(U16_IS_SURROGATE(c) && !(unicodeMask&UCNV_HAS_SURROGATES)) { + if(U16_IS_SURROGATE_LEAD(c)) { +getTrail: + if(sourcesource=source; + c=_extFromU(cnv, cnv->sharedData, + c, &source, sourceLimit, + &target, target+targetCapacity, + &offsets, sourceIndex, + pArgs->flush, + pErrorCode); + nextSourceIndex+=(int32_t)(source-pArgs->source); + + if(U_FAILURE(*pErrorCode)) { + /* not mappable or buffer overflow */ + break; + } else { + /* a mapping was written to the target, continue */ + + /* recalculate the targetCapacity after an extension mapping */ + targetCapacity=(int32_t)(pArgs->targetLimit-(char *)target); + + /* normal end of conversion: prepare for a new character */ + sourceIndex=nextSourceIndex; + continue; + } + } + } + + /* write the output character bytes from value and length */ + /* from the first if in the loop we know that targetCapacity>0 */ + if(value<=0xff) { + /* this is easy because we know that there is enough space */ + *target++=(uint8_t)value; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + --targetCapacity; + } else /* length==2 */ { + *target++=(uint8_t)(value>>8); + if(2<=targetCapacity) { + *target++=(uint8_t)value; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + *offsets++=sourceIndex; + } + targetCapacity-=2; + } else { + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + cnv->charErrorBuffer[0]=(char)value; + cnv->charErrorBufferLength=1; + + /* target overflow */ + targetCapacity=0; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + c=0; + break; + } + } + + /* normal end of conversion: prepare for a new character */ + c=0; + sourceIndex=nextSourceIndex; + continue; + } else { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + + /* set the converter state back into UConverter */ + cnv->fromUChar32=c; + + /* write back the updated pointers */ + pArgs->source=source; + pArgs->target=(char *)target; + pArgs->offsets=offsets; +} + +/* This version of ucnv_MBCSFromUnicodeWithOffsets() is optimized for single-byte codepages. */ +static void +ucnv_MBCSSingleFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const char16_t *source, *sourceLimit; + uint8_t *target; + int32_t targetCapacity; + int32_t *offsets; + + const uint16_t *table; + const uint16_t *results; + + UChar32 c; + + int32_t sourceIndex, nextSourceIndex; + + uint16_t value, minValue; + UBool hasSupplementary; + + /* set up the local pointers */ + cnv=pArgs->converter; + source=pArgs->source; + sourceLimit=pArgs->sourceLimit; + target=(uint8_t *)pArgs->target; + targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); + offsets=pArgs->offsets; + + table=cnv->sharedData->mbcs.fromUnicodeTable; + if((cnv->options&UCNV_OPTION_SWAP_LFNL)!=0) { + results=(uint16_t *)cnv->sharedData->mbcs.swapLFNLFromUnicodeBytes; + } else { + results=(uint16_t *)cnv->sharedData->mbcs.fromUnicodeBytes; + } + + if(cnv->useFallback) { + /* use all roundtrip and fallback results */ + minValue=0x800; + } else { + /* use only roundtrips and fallbacks from private-use characters */ + minValue=0xc00; + } + hasSupplementary=(UBool)(cnv->sharedData->mbcs.unicodeMask&UCNV_HAS_SUPPLEMENTARY); + + /* get the converter state from UConverter */ + c=cnv->fromUChar32; + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex= c==0 ? 0 : -1; + nextSourceIndex=0; + + /* conversion loop */ + if(c!=0 && targetCapacity>0) { + goto getTrail; + } + + while(source0) { + /* + * Get a correct Unicode code point: + * a single char16_t for a BMP code point or + * a matched surrogate pair for a "supplementary code point". + */ + c=*source++; + ++nextSourceIndex; + if(U16_IS_SURROGATE(c)) { + if(U16_IS_SURROGATE_LEAD(c)) { +getTrail: + if(source=minValue) { + /* assigned, write the output character bytes from value and length */ + /* length==1 */ + /* this is easy because we know that there is enough space */ + *target++=(uint8_t)value; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + --targetCapacity; + + /* normal end of conversion: prepare for a new character */ + c=0; + sourceIndex=nextSourceIndex; + } else { /* unassigned */ +unassigned: + /* try an extension mapping */ + pArgs->source=source; + c=_extFromU(cnv, cnv->sharedData, + c, &source, sourceLimit, + &target, target+targetCapacity, + &offsets, sourceIndex, + pArgs->flush, + pErrorCode); + nextSourceIndex+=(int32_t)(source-pArgs->source); + + if(U_FAILURE(*pErrorCode)) { + /* not mappable or buffer overflow */ + break; + } else { + /* a mapping was written to the target, continue */ + + /* recalculate the targetCapacity after an extension mapping */ + targetCapacity=(int32_t)(pArgs->targetLimit-(char *)target); + + /* normal end of conversion: prepare for a new character */ + sourceIndex=nextSourceIndex; + } + } + } else { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + + /* set the converter state back into UConverter */ + cnv->fromUChar32=c; + + /* write back the updated pointers */ + pArgs->source=source; + pArgs->target=(char *)target; + pArgs->offsets=offsets; +} + +/* + * This version of ucnv_MBCSFromUnicode() is optimized for single-byte codepages + * that map only to and from the BMP. + * In addition to single-byte/state optimizations, the offset calculations + * become much easier. + * It would be possible to use the sbcsIndex for UTF-8-friendly tables, + * but measurements have shown that this diminishes performance + * in more cases than it improves it. + * See SVN revision 21013 (2007-feb-06) for the last version with #if switches + * for various MBCS and SBCS optimizations. + */ +static void +ucnv_MBCSSingleFromBMPWithOffsets(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const char16_t *source, *sourceLimit, *lastSource; + uint8_t *target; + int32_t targetCapacity, length; + int32_t *offsets; + + const uint16_t *table; + const uint16_t *results; + + UChar32 c; + + int32_t sourceIndex; + + uint32_t asciiRoundtrips; + uint16_t value, minValue; + + /* set up the local pointers */ + cnv=pArgs->converter; + source=pArgs->source; + sourceLimit=pArgs->sourceLimit; + target=(uint8_t *)pArgs->target; + targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); + offsets=pArgs->offsets; + + table=cnv->sharedData->mbcs.fromUnicodeTable; + if((cnv->options&UCNV_OPTION_SWAP_LFNL)!=0) { + results=(uint16_t *)cnv->sharedData->mbcs.swapLFNLFromUnicodeBytes; + } else { + results=(uint16_t *)cnv->sharedData->mbcs.fromUnicodeBytes; + } + asciiRoundtrips=cnv->sharedData->mbcs.asciiRoundtrips; + + if(cnv->useFallback) { + /* use all roundtrip and fallback results */ + minValue=0x800; + } else { + /* use only roundtrips and fallbacks from private-use characters */ + minValue=0xc00; + } + + /* get the converter state from UConverter */ + c=cnv->fromUChar32; + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex= c==0 ? 0 : -1; + lastSource=source; + + /* + * since the conversion here is 1:1 char16_t:uint8_t, we need only one counter + * for the minimum of the sourceLength and targetCapacity + */ + length=(int32_t)(sourceLimit-source); + if(length0) { + goto getTrail; + } + +#if MBCS_UNROLL_SINGLE_FROM_BMP + /* unrolling makes it slower on Pentium III/Windows 2000?! */ + /* unroll the loop with the most common case */ +unrolled: + if(targetCapacity>=4) { + int32_t count, loops; + uint16_t andedValues; + + loops=count=targetCapacity>>2; + do { + c=*source++; + andedValues=value=MBCS_SINGLE_RESULT_FROM_U(table, results, c); + *target++=(uint8_t)value; + c=*source++; + andedValues&=value=MBCS_SINGLE_RESULT_FROM_U(table, results, c); + *target++=(uint8_t)value; + c=*source++; + andedValues&=value=MBCS_SINGLE_RESULT_FROM_U(table, results, c); + *target++=(uint8_t)value; + c=*source++; + andedValues&=value=MBCS_SINGLE_RESULT_FROM_U(table, results, c); + *target++=(uint8_t)value; + + /* were all 4 entries really valid? */ + if(andedValues0); + count=loops-count; + targetCapacity-=4*count; + + if(offsets!=nullptr) { + lastSource+=4*count; + while(count>0) { + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + *offsets++=sourceIndex++; + --count; + } + } + + c=0; + } +#endif + + while(targetCapacity>0) { + /* + * Get a correct Unicode code point: + * a single char16_t for a BMP code point or + * a matched surrogate pair for a "supplementary code point". + */ + c=*source++; + /* + * Do not immediately check for single surrogates: + * Assume that they are unassigned and check for them in that case. + * This speeds up the conversion of assigned characters. + */ + /* convert the Unicode code point in c into codepage bytes */ + if(c<=0x7f && IS_ASCII_ROUNDTRIP(c, asciiRoundtrips)) { + *target++=(uint8_t)c; + --targetCapacity; + c=0; + continue; + } + value=MBCS_SINGLE_RESULT_FROM_U(table, results, c); + /* is this code point assigned, or do we use fallbacks? */ + if(value>=minValue) { + /* assigned, write the output character bytes from value and length */ + /* length==1 */ + /* this is easy because we know that there is enough space */ + *target++=(uint8_t)value; + --targetCapacity; + + /* normal end of conversion: prepare for a new character */ + c=0; + continue; + } else if(!U16_IS_SURROGATE(c)) { + /* normal, unassigned BMP character */ + } else if(U16_IS_SURROGATE_LEAD(c)) { +getTrail: + if(sourceflush) { + *pErrorCode=U_TRUNCATED_CHAR_FOUND; + } + break; + } + } else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } + + /* c does not have a mapping */ + + /* get the number of code units for c to correctly advance sourceIndex */ + length=U16_LENGTH(c); + + /* set offsets since the start or the last extension */ + if(offsets!=nullptr) { + int32_t count=(int32_t)(source-lastSource); + + /* do not set the offset for this character */ + count-=length; + + while(count>0) { + *offsets++=sourceIndex++; + --count; + } + /* offsets and sourceIndex are now set for the current character */ + } + + /* try an extension mapping */ + lastSource=source; + c=_extFromU(cnv, cnv->sharedData, + c, &source, sourceLimit, + &target, (const uint8_t *)(pArgs->targetLimit), + &offsets, sourceIndex, + pArgs->flush, + pErrorCode); + sourceIndex+=length+(int32_t)(source-lastSource); + lastSource=source; + + if(U_FAILURE(*pErrorCode)) { + /* not mappable or buffer overflow */ + break; + } else { + /* a mapping was written to the target, continue */ + + /* recalculate the targetCapacity after an extension mapping */ + targetCapacity=(int32_t)(pArgs->targetLimit-(char *)target); + length=(int32_t)(sourceLimit-source); + if(length=(uint8_t *)pArgs->targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + + /* set offsets since the start or the last callback */ + if(offsets!=nullptr) { + size_t count=source-lastSource; + if (count > 0 && *pErrorCode == U_TRUNCATED_CHAR_FOUND) { + /* + Caller gave us a partial supplementary character, + which this function couldn't convert in any case. + The callback will handle the offset. + */ + count--; + } + while(count>0) { + *offsets++=sourceIndex++; + --count; + } + } + + /* set the converter state back into UConverter */ + cnv->fromUChar32=c; + + /* write back the updated pointers */ + pArgs->source=source; + pArgs->target=(char *)target; + pArgs->offsets=offsets; +} + +U_CFUNC void +ucnv_MBCSFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + const char16_t *source, *sourceLimit; + uint8_t *target; + int32_t targetCapacity; + int32_t *offsets; + + const uint16_t *table; + const uint16_t *mbcsIndex; + const uint8_t *p, *bytes; + uint8_t outputType; + + UChar32 c; + + int32_t prevSourceIndex, sourceIndex, nextSourceIndex; + + uint32_t stage2Entry; + uint32_t asciiRoundtrips; + uint32_t value; + /* Shift-In and Shift-Out byte sequences differ by encoding scheme. */ + uint8_t siBytes[2] = {0, 0}; + uint8_t soBytes[2] = {0, 0}; + uint8_t siLength, soLength; + int32_t length = 0, prevLength; + uint8_t unicodeMask; + + cnv=pArgs->converter; + + if(cnv->preFromUFirstCP>=0) { + /* + * pass sourceIndex=-1 because we continue from an earlier buffer + * in the future, this may change with continuous offsets + */ + ucnv_extContinueMatchFromU(cnv, pArgs, -1, pErrorCode); + + if(U_FAILURE(*pErrorCode) || cnv->preFromULength<0) { + return; + } + } + + /* use optimized function if possible */ + outputType=cnv->sharedData->mbcs.outputType; + unicodeMask=cnv->sharedData->mbcs.unicodeMask; + if(outputType==MBCS_OUTPUT_1 && !(unicodeMask&UCNV_HAS_SURROGATES)) { + if(!(unicodeMask&UCNV_HAS_SUPPLEMENTARY)) { + ucnv_MBCSSingleFromBMPWithOffsets(pArgs, pErrorCode); + } else { + ucnv_MBCSSingleFromUnicodeWithOffsets(pArgs, pErrorCode); + } + return; + } else if(outputType==MBCS_OUTPUT_2 && cnv->sharedData->mbcs.utf8Friendly) { + ucnv_MBCSDoubleFromUnicodeWithOffsets(pArgs, pErrorCode); + return; + } + + /* set up the local pointers */ + source=pArgs->source; + sourceLimit=pArgs->sourceLimit; + target=(uint8_t *)pArgs->target; + targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); + offsets=pArgs->offsets; + + table=cnv->sharedData->mbcs.fromUnicodeTable; + if(cnv->sharedData->mbcs.utf8Friendly) { + mbcsIndex=cnv->sharedData->mbcs.mbcsIndex; + } else { + mbcsIndex=nullptr; + } + if((cnv->options&UCNV_OPTION_SWAP_LFNL)!=0) { + bytes=cnv->sharedData->mbcs.swapLFNLFromUnicodeBytes; + } else { + bytes=cnv->sharedData->mbcs.fromUnicodeBytes; + } + asciiRoundtrips=cnv->sharedData->mbcs.asciiRoundtrips; + + /* get the converter state from UConverter */ + c=cnv->fromUChar32; + + if(outputType==MBCS_OUTPUT_2_SISO) { + prevLength=cnv->fromUnicodeStatus; + if(prevLength==0) { + /* set the real value */ + prevLength=1; + } + } else { + /* prevent fromUnicodeStatus from being set to something non-0 */ + prevLength=0; + } + + /* sourceIndex=-1 if the current character began in the previous buffer */ + prevSourceIndex=-1; + sourceIndex= c==0 ? 0 : -1; + nextSourceIndex=0; + + /* Get the SI/SO character for the converter */ + siLength = static_cast(getSISOBytes(SI, cnv->options, siBytes)); + soLength = static_cast(getSISOBytes(SO, cnv->options, soBytes)); + + /* conversion loop */ + /* + * This is another piece of ugly code: + * A goto into the loop if the converter state contains a first surrogate + * from the previous function call. + * It saves me to check in each loop iteration a check of if(c==0) + * and duplicating the trail-surrogate-handling code in the else + * branch of that check. + * I could not find any other way to get around this other than + * using a function call for the conversion and callback, which would + * be even more inefficient. + * + * Markus Scherer 2000-jul-19 + */ + if(c!=0 && targetCapacity>0) { + goto getTrail; + } + + while(source0) { + /* + * Get a correct Unicode code point: + * a single char16_t for a BMP code point or + * a matched surrogate pair for a "supplementary code point". + */ + c=*source++; + ++nextSourceIndex; + if(c<=0x7f && IS_ASCII_ROUNDTRIP(c, asciiRoundtrips)) { + *target++=(uint8_t)c; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + prevSourceIndex=sourceIndex; + sourceIndex=nextSourceIndex; + } + --targetCapacity; + c=0; + continue; + } + /* + * utf8Friendly table: Test for <=0xd7ff rather than <=MBCS_FAST_MAX + * to avoid dealing with surrogates. + * MBCS_FAST_MAX must be >=0xd7ff. + */ + if(c<=0xd7ff && mbcsIndex!=nullptr) { + value=mbcsIndex[c>>6]; + + /* get the bytes and the length for the output (copied from below and adapted for utf8Friendly data) */ + /* There are only roundtrips (!=0) and no-mapping (==0) entries. */ + switch(outputType) { + case MBCS_OUTPUT_2: + value=((const uint16_t *)bytes)[value +(c&0x3f)]; + if(value<=0xff) { + if(value==0) { + goto unassigned; + } else { + length=1; + } + } else { + length=2; + } + break; + case MBCS_OUTPUT_2_SISO: + /* 1/2-byte stateful with Shift-In/Shift-Out */ + /* + * Save the old state in the converter object + * right here, then change the local prevLength state variable if necessary. + * Then, if this character turns out to be unassigned or a fallback that + * is not taken, the callback code must not save the new state in the converter + * because the new state is for a character that is not output. + * However, the callback must still restore the state from the converter + * in case the callback function changed it for its output. + */ + cnv->fromUnicodeStatus=prevLength; /* save the old state */ + value=((const uint16_t *)bytes)[value +(c&0x3f)]; + if(value<=0xff) { + if(value==0) { + goto unassigned; + } else if(prevLength<=1) { + length=1; + } else { + /* change from double-byte mode to single-byte */ + if (siLength == 1) { + value|=(uint32_t)siBytes[0]<<8; + length = 2; + } else if (siLength == 2) { + value|=(uint32_t)siBytes[1]<<8; + value|=(uint32_t)siBytes[0]<<16; + length = 3; + } + prevLength=1; + } + } else { + if(prevLength==2) { + length=2; + } else { + /* change from single-byte mode to double-byte */ + if (soLength == 1) { + value|=(uint32_t)soBytes[0]<<16; + length = 3; + } else if (soLength == 2) { + value|=(uint32_t)soBytes[1]<<16; + value|=(uint32_t)soBytes[0]<<24; + length = 4; + } + prevLength=2; + } + } + break; + case MBCS_OUTPUT_DBCS_ONLY: + /* table with single-byte results, but only DBCS mappings used */ + value=((const uint16_t *)bytes)[value +(c&0x3f)]; + if(value<=0xff) { + /* no mapping or SBCS result, not taken for DBCS-only */ + goto unassigned; + } else { + length=2; + } + break; + case MBCS_OUTPUT_3: + p=bytes+(value+(c&0x3f))*3; + value=((uint32_t)*p<<16)|((uint32_t)p[1]<<8)|p[2]; + if(value<=0xff) { + if(value==0) { + goto unassigned; + } else { + length=1; + } + } else if(value<=0xffff) { + length=2; + } else { + length=3; + } + break; + case MBCS_OUTPUT_4: + value=((const uint32_t *)bytes)[value +(c&0x3f)]; + if(value<=0xff) { + if(value==0) { + goto unassigned; + } else { + length=1; + } + } else if(value<=0xffff) { + length=2; + } else if(value<=0xffffff) { + length=3; + } else { + length=4; + } + break; + case MBCS_OUTPUT_3_EUC: + value=((const uint16_t *)bytes)[value +(c&0x3f)]; + /* EUC 16-bit fixed-length representation */ + if(value<=0xff) { + if(value==0) { + goto unassigned; + } else { + length=1; + } + } else if((value&0x8000)==0) { + value|=0x8e8000; + length=3; + } else if((value&0x80)==0) { + value|=0x8f0080; + length=3; + } else { + length=2; + } + break; + case MBCS_OUTPUT_4_EUC: + p=bytes+(value+(c&0x3f))*3; + value=((uint32_t)*p<<16)|((uint32_t)p[1]<<8)|p[2]; + /* EUC 16-bit fixed-length representation applied to the first two bytes */ + if(value<=0xff) { + if(value==0) { + goto unassigned; + } else { + length=1; + } + } else if(value<=0xffff) { + length=2; + } else if((value&0x800000)==0) { + value|=0x8e800000; + length=4; + } else if((value&0x8000)==0) { + value|=0x8f008000; + length=4; + } else { + length=3; + } + break; + default: + /* must not occur */ + /* + * To avoid compiler warnings that value & length may be + * used without having been initialized, we set them here. + * In reality, this is unreachable code. + * Not having a default branch also causes warnings with + * some compilers. + */ + value=0; + length=0; + break; + } + /* output the value */ + } else { + /* + * This also tests if the codepage maps single surrogates. + * If it does, then surrogates are not paired but mapped separately. + * Note that in this case unmatched surrogates are not detected. + */ + if(U16_IS_SURROGATE(c) && !(unicodeMask&UCNV_HAS_SURROGATES)) { + if(U16_IS_SURROGATE_LEAD(c)) { +getTrail: + if(sourcefromUnicodeStatus=prevLength; /* save the old state */ + /* callback(unassigned) */ + goto unassigned; + } + /* convert this supplementary code point */ + /* exit this condition tree */ + } else { + /* this is an unmatched lead code unit (1st surrogate) */ + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } + } else { + /* no more input */ + break; + } + } else { + /* this is an unmatched trail code unit (2nd surrogate) */ + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + break; + } + } + + /* convert the Unicode code point in c into codepage bytes */ + + /* + * The basic lookup is a triple-stage compact array (trie) lookup. + * For details see the beginning of this file. + * + * Single-byte codepages are handled with a different data structure + * by _MBCSSingle... functions. + * + * The result consists of a 32-bit value from stage 2 and + * a pointer to as many bytes as are stored per character. + * The pointer points to the character's bytes in stage 3. + * Bits 15..0 of the stage 2 entry contain the stage 3 index + * for that pointer, while bits 31..16 are flags for which of + * the 16 characters in the block are roundtrip-assigned. + * + * For 2-byte and 4-byte codepages, the bytes are stored as uint16_t + * respectively as uint32_t, in the platform encoding. + * For 3-byte codepages, the bytes are always stored in big-endian order. + * + * For EUC encodings that use only either 0x8e or 0x8f as the first + * byte of their longest byte sequences, the first two bytes in + * this third stage indicate with their 7th bits whether these bytes + * are to be written directly or actually need to be preceded by + * one of the two Single-Shift codes. With this, the third stage + * stores one byte fewer per character than the actual maximum length of + * EUC byte sequences. + * + * Other than that, leading zero bytes are removed and the other + * bytes output. A single zero byte may be output if the "assigned" + * bit in stage 2 was on. + * The data structure does not support zero byte output as a fallback, + * and also does not allow output of leading zeros. + */ + stage2Entry=MBCS_STAGE_2_FROM_U(table, c); + + /* get the bytes and the length for the output */ + switch(outputType) { + case MBCS_OUTPUT_2: + value=MBCS_VALUE_2_FROM_STAGE_2(bytes, stage2Entry, c); + if(value<=0xff) { + length=1; + } else { + length=2; + } + break; + case MBCS_OUTPUT_2_SISO: + /* 1/2-byte stateful with Shift-In/Shift-Out */ + /* + * Save the old state in the converter object + * right here, then change the local prevLength state variable if necessary. + * Then, if this character turns out to be unassigned or a fallback that + * is not taken, the callback code must not save the new state in the converter + * because the new state is for a character that is not output. + * However, the callback must still restore the state from the converter + * in case the callback function changed it for its output. + */ + cnv->fromUnicodeStatus=prevLength; /* save the old state */ + value=MBCS_VALUE_2_FROM_STAGE_2(bytes, stage2Entry, c); + if(value<=0xff) { + if(value==0 && MBCS_FROM_U_IS_ROUNDTRIP(stage2Entry, c)==0) { + /* no mapping, leave value==0 */ + length=0; + } else if(prevLength<=1) { + length=1; + } else { + /* change from double-byte mode to single-byte */ + if (siLength == 1) { + value|=(uint32_t)siBytes[0]<<8; + length = 2; + } else if (siLength == 2) { + value|=(uint32_t)siBytes[1]<<8; + value|=(uint32_t)siBytes[0]<<16; + length = 3; + } + prevLength=1; + } + } else { + if(prevLength==2) { + length=2; + } else { + /* change from single-byte mode to double-byte */ + if (soLength == 1) { + value|=(uint32_t)soBytes[0]<<16; + length = 3; + } else if (soLength == 2) { + value|=(uint32_t)soBytes[1]<<16; + value|=(uint32_t)soBytes[0]<<24; + length = 4; + } + prevLength=2; + } + } + break; + case MBCS_OUTPUT_DBCS_ONLY: + /* table with single-byte results, but only DBCS mappings used */ + value=MBCS_VALUE_2_FROM_STAGE_2(bytes, stage2Entry, c); + if(value<=0xff) { + /* no mapping or SBCS result, not taken for DBCS-only */ + value=stage2Entry=0; /* stage2Entry=0 to reset roundtrip flags */ + length=0; + } else { + length=2; + } + break; + case MBCS_OUTPUT_3: + p=MBCS_POINTER_3_FROM_STAGE_2(bytes, stage2Entry, c); + value=((uint32_t)*p<<16)|((uint32_t)p[1]<<8)|p[2]; + if(value<=0xff) { + length=1; + } else if(value<=0xffff) { + length=2; + } else { + length=3; + } + break; + case MBCS_OUTPUT_4: + value=MBCS_VALUE_4_FROM_STAGE_2(bytes, stage2Entry, c); + if(value<=0xff) { + length=1; + } else if(value<=0xffff) { + length=2; + } else if(value<=0xffffff) { + length=3; + } else { + length=4; + } + break; + case MBCS_OUTPUT_3_EUC: + value=MBCS_VALUE_2_FROM_STAGE_2(bytes, stage2Entry, c); + /* EUC 16-bit fixed-length representation */ + if(value<=0xff) { + length=1; + } else if((value&0x8000)==0) { + value|=0x8e8000; + length=3; + } else if((value&0x80)==0) { + value|=0x8f0080; + length=3; + } else { + length=2; + } + break; + case MBCS_OUTPUT_4_EUC: + p=MBCS_POINTER_3_FROM_STAGE_2(bytes, stage2Entry, c); + value=((uint32_t)*p<<16)|((uint32_t)p[1]<<8)|p[2]; + /* EUC 16-bit fixed-length representation applied to the first two bytes */ + if(value<=0xff) { + length=1; + } else if(value<=0xffff) { + length=2; + } else if((value&0x800000)==0) { + value|=0x8e800000; + length=4; + } else if((value&0x8000)==0) { + value|=0x8f008000; + length=4; + } else { + length=3; + } + break; + default: + /* must not occur */ + /* + * To avoid compiler warnings that value & length may be + * used without having been initialized, we set them here. + * In reality, this is unreachable code. + * Not having a default branch also causes warnings with + * some compilers. + */ + value=stage2Entry=0; /* stage2Entry=0 to reset roundtrip flags */ + length=0; + break; + } + + /* is this code point assigned, or do we use fallbacks? */ + if(!(MBCS_FROM_U_IS_ROUNDTRIP(stage2Entry, c)!=0 || + (UCNV_FROM_U_USE_FALLBACK(cnv, c) && value!=0)) + ) { + /* + * We allow a 0 byte output if the "assigned" bit is set for this entry. + * There is no way with this data structure for fallback output + * to be a zero byte. + */ + +unassigned: + /* try an extension mapping */ + pArgs->source=source; + c=_extFromU(cnv, cnv->sharedData, + c, &source, sourceLimit, + &target, target+targetCapacity, + &offsets, sourceIndex, + pArgs->flush, + pErrorCode); + nextSourceIndex+=(int32_t)(source-pArgs->source); + prevLength=cnv->fromUnicodeStatus; /* restore SISO state */ + + if(U_FAILURE(*pErrorCode)) { + /* not mappable or buffer overflow */ + break; + } else { + /* a mapping was written to the target, continue */ + + /* recalculate the targetCapacity after an extension mapping */ + targetCapacity=(int32_t)(pArgs->targetLimit-(char *)target); + + /* normal end of conversion: prepare for a new character */ + if(offsets!=nullptr) { + prevSourceIndex=sourceIndex; + sourceIndex=nextSourceIndex; + } + continue; + } + } + } + + /* write the output character bytes from value and length */ + /* from the first if in the loop we know that targetCapacity>0 */ + if(length<=targetCapacity) { + if(offsets==nullptr) { + switch(length) { + /* each branch falls through to the next one */ + case 4: + *target++=(uint8_t)(value>>24); + U_FALLTHROUGH; + case 3: + *target++=(uint8_t)(value>>16); + U_FALLTHROUGH; + case 2: + *target++=(uint8_t)(value>>8); + U_FALLTHROUGH; + case 1: + *target++=(uint8_t)value; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + } else { + switch(length) { + /* each branch falls through to the next one */ + case 4: + *target++=(uint8_t)(value>>24); + *offsets++=sourceIndex; + U_FALLTHROUGH; + case 3: + *target++=(uint8_t)(value>>16); + *offsets++=sourceIndex; + U_FALLTHROUGH; + case 2: + *target++=(uint8_t)(value>>8); + *offsets++=sourceIndex; + U_FALLTHROUGH; + case 1: + *target++=(uint8_t)value; + *offsets++=sourceIndex; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + } + targetCapacity-=length; + } else { + uint8_t *charErrorBuffer; + + /* + * We actually do this backwards here: + * In order to save an intermediate variable, we output + * first to the overflow buffer what does not fit into the + * regular target. + */ + /* we know that 1<=targetCapacitycharErrorBuffer; + switch(length) { + /* each branch falls through to the next one */ + case 3: + *charErrorBuffer++=(uint8_t)(value>>16); + U_FALLTHROUGH; + case 2: + *charErrorBuffer++=(uint8_t)(value>>8); + U_FALLTHROUGH; + case 1: + *charErrorBuffer=(uint8_t)value; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + cnv->charErrorBufferLength=(int8_t)length; + + /* now output what fits into the regular target */ + value>>=8*length; /* length was reduced by targetCapacity */ + switch(targetCapacity) { + /* each branch falls through to the next one */ + case 3: + *target++=(uint8_t)(value>>16); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + U_FALLTHROUGH; + case 2: + *target++=(uint8_t)(value>>8); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + U_FALLTHROUGH; + case 1: + *target++=(uint8_t)value; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + + /* target overflow */ + targetCapacity=0; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + c=0; + break; + } + + /* normal end of conversion: prepare for a new character */ + c=0; + if(offsets!=nullptr) { + prevSourceIndex=sourceIndex; + sourceIndex=nextSourceIndex; + } + continue; + } else { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + + /* + * the end of the input stream and detection of truncated input + * are handled by the framework, but for EBCDIC_STATEFUL conversion + * we need to emit an SI at the very end + * + * conditions: + * successful + * EBCDIC_STATEFUL in DBCS mode + * end of input and no truncated input + */ + if( U_SUCCESS(*pErrorCode) && + outputType==MBCS_OUTPUT_2_SISO && prevLength==2 && + pArgs->flush && source>=sourceLimit && c==0 + ) { + /* EBCDIC_STATEFUL ending with DBCS: emit an SI to return the output stream to SBCS */ + if(targetCapacity>0) { + *target++=(uint8_t)siBytes[0]; + if (siLength == 2) { + if (targetCapacity<2) { + cnv->charErrorBuffer[0]=(uint8_t)siBytes[1]; + cnv->charErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } else { + *target++=(uint8_t)siBytes[1]; + } + } + if(offsets!=nullptr) { + /* set the last source character's index (sourceIndex points at sourceLimit now) */ + *offsets++=prevSourceIndex; + } + } else { + /* target is full */ + cnv->charErrorBuffer[0]=(uint8_t)siBytes[0]; + if (siLength == 2) { + cnv->charErrorBuffer[1]=(uint8_t)siBytes[1]; + } + cnv->charErrorBufferLength=siLength; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + } + prevLength=1; /* we switched into SBCS */ + } + + /* set the converter state back into UConverter */ + cnv->fromUChar32=c; + cnv->fromUnicodeStatus=prevLength; + + /* write back the updated pointers */ + pArgs->source=source; + pArgs->target=(char *)target; + pArgs->offsets=offsets; +} + +/* + * This is another simple conversion function for internal use by other + * conversion implementations. + * It does not use the converter state nor call callbacks. + * It does not handle the EBCDIC swaplfnl option (set in UConverter). + * It handles conversion extensions but not GB 18030. + * + * It converts one single Unicode code point into codepage bytes, encoded + * as one 32-bit value. The function returns the number of bytes in *pValue: + * 1..4 the number of bytes in *pValue + * 0 unassigned (*pValue undefined) + * -1 illegal (currently not used, *pValue undefined) + * + * *pValue will contain the resulting bytes with the last byte in bits 7..0, + * the second to last byte in bits 15..8, etc. + * Currently, the function assumes but does not check that 0<=c<=0x10ffff. + */ +U_CFUNC int32_t +ucnv_MBCSFromUChar32(UConverterSharedData *sharedData, + UChar32 c, uint32_t *pValue, + UBool useFallback) { + const int32_t *cx; + const uint16_t *table; +#if 0 +/* #if 0 because this is not currently used in ICU - reduce code, increase code coverage */ + const uint8_t *p; +#endif + uint32_t stage2Entry; + uint32_t value; + int32_t length; + + /* BMP-only codepages are stored without stage 1 entries for supplementary code points */ + if(c<=0xffff || (sharedData->mbcs.unicodeMask&UCNV_HAS_SUPPLEMENTARY)) { + table=sharedData->mbcs.fromUnicodeTable; + + /* convert the Unicode code point in c into codepage bytes (same as in _MBCSFromUnicodeWithOffsets) */ + if(sharedData->mbcs.outputType==MBCS_OUTPUT_1) { + value=MBCS_SINGLE_RESULT_FROM_U(table, (uint16_t *)sharedData->mbcs.fromUnicodeBytes, c); + /* is this code point assigned, or do we use fallbacks? */ + if(useFallback ? value>=0x800 : value>=0xc00) { + *pValue=value&0xff; + return 1; + } + } else /* outputType!=MBCS_OUTPUT_1 */ { + stage2Entry=MBCS_STAGE_2_FROM_U(table, c); + + /* get the bytes and the length for the output */ + switch(sharedData->mbcs.outputType) { + case MBCS_OUTPUT_2: + value=MBCS_VALUE_2_FROM_STAGE_2(sharedData->mbcs.fromUnicodeBytes, stage2Entry, c); + if(value<=0xff) { + length=1; + } else { + length=2; + } + break; +#if 0 +/* #if 0 because this is not currently used in ICU - reduce code, increase code coverage */ + case MBCS_OUTPUT_DBCS_ONLY: + /* table with single-byte results, but only DBCS mappings used */ + value=MBCS_VALUE_2_FROM_STAGE_2(sharedData->mbcs.fromUnicodeBytes, stage2Entry, c); + if(value<=0xff) { + /* no mapping or SBCS result, not taken for DBCS-only */ + value=stage2Entry=0; /* stage2Entry=0 to reset roundtrip flags */ + length=0; + } else { + length=2; + } + break; + case MBCS_OUTPUT_3: + p=MBCS_POINTER_3_FROM_STAGE_2(sharedData->mbcs.fromUnicodeBytes, stage2Entry, c); + value=((uint32_t)*p<<16)|((uint32_t)p[1]<<8)|p[2]; + if(value<=0xff) { + length=1; + } else if(value<=0xffff) { + length=2; + } else { + length=3; + } + break; + case MBCS_OUTPUT_4: + value=MBCS_VALUE_4_FROM_STAGE_2(sharedData->mbcs.fromUnicodeBytes, stage2Entry, c); + if(value<=0xff) { + length=1; + } else if(value<=0xffff) { + length=2; + } else if(value<=0xffffff) { + length=3; + } else { + length=4; + } + break; + case MBCS_OUTPUT_3_EUC: + value=MBCS_VALUE_2_FROM_STAGE_2(sharedData->mbcs.fromUnicodeBytes, stage2Entry, c); + /* EUC 16-bit fixed-length representation */ + if(value<=0xff) { + length=1; + } else if((value&0x8000)==0) { + value|=0x8e8000; + length=3; + } else if((value&0x80)==0) { + value|=0x8f0080; + length=3; + } else { + length=2; + } + break; + case MBCS_OUTPUT_4_EUC: + p=MBCS_POINTER_3_FROM_STAGE_2(sharedData->mbcs.fromUnicodeBytes, stage2Entry, c); + value=((uint32_t)*p<<16)|((uint32_t)p[1]<<8)|p[2]; + /* EUC 16-bit fixed-length representation applied to the first two bytes */ + if(value<=0xff) { + length=1; + } else if(value<=0xffff) { + length=2; + } else if((value&0x800000)==0) { + value|=0x8e800000; + length=4; + } else if((value&0x8000)==0) { + value|=0x8f008000; + length=4; + } else { + length=3; + } + break; +#endif + default: + /* must not occur */ + return -1; + } + + /* is this code point assigned, or do we use fallbacks? */ + if( MBCS_FROM_U_IS_ROUNDTRIP(stage2Entry, c) || + (FROM_U_USE_FALLBACK(useFallback, c) && value!=0) + ) { + /* + * We allow a 0 byte output if the "assigned" bit is set for this entry. + * There is no way with this data structure for fallback output + * to be a zero byte. + */ + /* assigned */ + *pValue=value; + return length; + } + } + } + + cx=sharedData->mbcs.extIndexes; + if(cx!=nullptr) { + length=ucnv_extSimpleMatchFromU(cx, c, pValue, useFallback); + return length>=0 ? length : -length; /* return abs(length); */ + } + + /* unassigned */ + return 0; +} + + +#if 0 +/* + * This function has been moved to ucnv2022.c for inlining. + * This implementation is here only for documentation purposes + */ + +/** + * This version of ucnv_MBCSFromUChar32() is optimized for single-byte codepages. + * It does not handle the EBCDIC swaplfnl option (set in UConverter). + * It does not handle conversion extensions (_extFromU()). + * + * It returns the codepage byte for the code point, or -1 if it is unassigned. + */ +U_CFUNC int32_t +ucnv_MBCSSingleFromUChar32(UConverterSharedData *sharedData, + UChar32 c, + UBool useFallback) { + const uint16_t *table; + int32_t value; + + /* BMP-only codepages are stored without stage 1 entries for supplementary code points */ + if(c>=0x10000 && !(sharedData->mbcs.unicodeMask&UCNV_HAS_SUPPLEMENTARY)) { + return -1; + } + + /* convert the Unicode code point in c into codepage bytes (same as in _MBCSFromUnicodeWithOffsets) */ + table=sharedData->mbcs.fromUnicodeTable; + + /* get the byte for the output */ + value=MBCS_SINGLE_RESULT_FROM_U(table, (uint16_t *)sharedData->mbcs.fromUnicodeBytes, c); + /* is this code point assigned, or do we use fallbacks? */ + if(useFallback ? value>=0x800 : value>=0xc00) { + return value&0xff; + } else { + return -1; + } +} +#endif + +/* MBCS-from-UTF-8 conversion functions ------------------------------------- */ + +/* offsets for n-byte UTF-8 sequences that were calculated with ((lead<<6)+trail)<<6+trail... */ +static const UChar32 +utf8_offsets[5]={ 0, 0, 0x3080, 0xE2080, 0x3C82080 }; + +static void U_CALLCONV +ucnv_SBCSFromUTF8(UConverterFromUnicodeArgs *pFromUArgs, + UConverterToUnicodeArgs *pToUArgs, + UErrorCode *pErrorCode) { + UConverter *utf8, *cnv; + const uint8_t *source, *sourceLimit; + uint8_t *target; + int32_t targetCapacity; + + const uint16_t *table, *sbcsIndex; + const uint16_t *results; + + int8_t oldToULength, toULength, toULimit; + + UChar32 c; + uint8_t b, t1, t2; + + uint32_t asciiRoundtrips; + uint16_t value, minValue = 0; + UBool hasSupplementary; + + /* set up the local pointers */ + utf8=pToUArgs->converter; + cnv=pFromUArgs->converter; + source=(uint8_t *)pToUArgs->source; + sourceLimit=(uint8_t *)pToUArgs->sourceLimit; + target=(uint8_t *)pFromUArgs->target; + targetCapacity=(int32_t)(pFromUArgs->targetLimit-pFromUArgs->target); + + table=cnv->sharedData->mbcs.fromUnicodeTable; + sbcsIndex=cnv->sharedData->mbcs.sbcsIndex; + if((cnv->options&UCNV_OPTION_SWAP_LFNL)!=0) { + results=(uint16_t *)cnv->sharedData->mbcs.swapLFNLFromUnicodeBytes; + } else { + results=(uint16_t *)cnv->sharedData->mbcs.fromUnicodeBytes; + } + asciiRoundtrips=cnv->sharedData->mbcs.asciiRoundtrips; + + if(cnv->useFallback) { + /* use all roundtrip and fallback results */ + minValue=0x800; + } else { + /* use only roundtrips and fallbacks from private-use characters */ + minValue=0xc00; + } + hasSupplementary=(UBool)(cnv->sharedData->mbcs.unicodeMask&UCNV_HAS_SUPPLEMENTARY); + + /* get the converter state from the UTF-8 UConverter */ + if(utf8->toULength > 0) { + toULength=oldToULength=utf8->toULength; + toULimit=(int8_t)utf8->mode; + c=(UChar32)utf8->toUnicodeStatus; + } else { + toULength=oldToULength=toULimit=0; + c = 0; + } + + // The conversion loop checks source0) { + uint8_t b1=*(sourceLimit-1); + if(U8_IS_SINGLE(b1)) { + // common ASCII character + } else if(U8_IS_TRAIL(b1) && length>=2) { + uint8_t b2=*(sourceLimit-2); + if(0xe0<=b2 && b2<0xf0 && U8_IS_VALID_LEAD3_AND_T1(b2, b1)) { + // truncated 3-byte sequence + sourceLimit-=2; + } + } else if(0xc2<=b1 && b1<0xf0) { + // truncated 2- or 3-byte sequence + --sourceLimit; + } + } + } + + if(c!=0 && targetCapacity>0) { + utf8->toUnicodeStatus=0; + utf8->toULength=0; + goto moreBytes; + /* + * Note: We could avoid the goto by duplicating some of the moreBytes + * code, but only up to the point of collecting a complete UTF-8 + * sequence; then recurse for the toUBytes[toULength] + * and then continue with normal conversion. + * + * If so, move this code to just after initializing the minimum + * set of local variables for reading the UTF-8 input + * (utf8, source, target, limits but not cnv, table, minValue, etc.). + * + * Potential advantages: + * - avoid the goto + * - oldToULength could become a local variable in just those code blocks + * that deal with buffer boundaries + * - possibly faster if the goto prevents some compiler optimizations + * (this would need measuring to confirm) + * Disadvantage: + * - code duplication + */ + } + + /* conversion loop */ + while(source0) { + b=*source++; + if(U8_IS_SINGLE(b)) { + /* convert ASCII */ + if(IS_ASCII_ROUNDTRIP(b, asciiRoundtrips)) { + *target++=(uint8_t)b; + --targetCapacity; + continue; + } else { + c=b; + value=SBCS_RESULT_FROM_UTF8(sbcsIndex, results, 0, c); + } + } else { + if(b<0xe0) { + if( /* handle U+0080..U+07FF inline */ + b>=0xc2 && + (t1=(uint8_t)(*source-0x80)) <= 0x3f + ) { + c=b&0x1f; + ++source; + value=SBCS_RESULT_FROM_UTF8(sbcsIndex, results, c, t1); + if(value>=minValue) { + *target++=(uint8_t)value; + --targetCapacity; + continue; + } else { + c=(c<<6)|t1; + } + } else { + c=-1; + } + } else if(b==0xe0) { + if( /* handle U+0800..U+0FFF inline */ + (t1=(uint8_t)(source[0]-0x80)) <= 0x3f && t1 >= 0x20 && + (t2=(uint8_t)(source[1]-0x80)) <= 0x3f + ) { + c=t1; + source+=2; + value=SBCS_RESULT_FROM_UTF8(sbcsIndex, results, c, t2); + if(value>=minValue) { + *target++=(uint8_t)value; + --targetCapacity; + continue; + } else { + c=(c<<6)|t2; + } + } else { + c=-1; + } + } else { + c=-1; + } + + if(c<0) { + /* handle "complicated" and error cases, and continuing partial characters */ + oldToULength=0; + toULength=1; + toULimit=U8_COUNT_BYTES_NON_ASCII(b); + c=b; +moreBytes: + while(toULengthsourceLimit) { + b=*source; + if(icu::UTF8::isValidTrail(c, b, toULength, toULimit)) { + ++source; + ++toULength; + c=(c<<6)+b; + } else { + break; /* sequence too short, stop with toULengthtoUBytes[oldToULength++]=*source++; + } + utf8->toUnicodeStatus=c; + utf8->toULength=toULength; + utf8->mode=toULimit; + pToUArgs->source=(char *)source; + pFromUArgs->target=(char *)target; + return; + } + } + + if(toULength==toULimit) { + c-=utf8_offsets[toULength]; + if(toULength<=3) { /* BMP */ + value=MBCS_SINGLE_RESULT_FROM_U(table, results, c); + } else { + /* supplementary code point */ + if(!hasSupplementary) { + /* BMP-only codepages are stored without stage 1 entries for supplementary code points */ + value=0; + } else { + value=MBCS_SINGLE_RESULT_FROM_U(table, results, c); + } + } + } else { + /* error handling: illegal UTF-8 byte sequence */ + source-=(toULength-oldToULength); + while(oldToULengthtoUBytes[oldToULength++]=*source++; + } + utf8->toULength=toULength; + pToUArgs->source=(char *)source; + pFromUArgs->target=(char *)target; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + return; + } + } + } + + if(value>=minValue) { + /* output the mapping for c */ + *target++=(uint8_t)value; + --targetCapacity; + } else { + /* valueUTF-16->charset conversion. + */ + static const char16_t nul=0; + const char16_t *noSource=&nul; + c=_extFromU(cnv, cnv->sharedData, + c, &noSource, noSource, + &target, target+targetCapacity, + nullptr, -1, + pFromUArgs->flush, + pErrorCode); + + if(U_FAILURE(*pErrorCode)) { + /* not mappable or buffer overflow */ + cnv->fromUChar32=c; + break; + } else if(cnv->preFromUFirstCP>=0) { + /* + * Partial match, return and revert to pivoting. + * In normal from-UTF-16 conversion, we would just continue + * but then exit the loop because the extension match would + * have consumed the source. + */ + *pErrorCode=U_USING_DEFAULT_WARNING; + break; + } else { + /* a mapping was written to the target, continue */ + + /* recalculate the targetCapacity after an extension mapping */ + targetCapacity=(int32_t)(pFromUArgs->targetLimit-(char *)target); + } + } + } else { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + + /* + * The sourceLimit may have been adjusted before the conversion loop + * to stop before a truncated sequence. + * If so, then collect the truncated sequence now. + */ + if(U_SUCCESS(*pErrorCode) && + cnv->preFromUFirstCP<0 && + source<(sourceLimit=(uint8_t *)pToUArgs->sourceLimit)) { + c=utf8->toUBytes[0]=b=*source++; + toULength=1; + toULimit=U8_COUNT_BYTES(b); + while(sourcetoUBytes[toULength++]=b=*source++; + c=(c<<6)+b; + } + utf8->toUnicodeStatus=c; + utf8->toULength=toULength; + utf8->mode=toULimit; + } + + /* write back the updated pointers */ + pToUArgs->source=(char *)source; + pFromUArgs->target=(char *)target; +} + +static void U_CALLCONV +ucnv_DBCSFromUTF8(UConverterFromUnicodeArgs *pFromUArgs, + UConverterToUnicodeArgs *pToUArgs, + UErrorCode *pErrorCode) { + UConverter *utf8, *cnv; + const uint8_t *source, *sourceLimit; + uint8_t *target; + int32_t targetCapacity; + + const uint16_t *table, *mbcsIndex; + const uint16_t *results; + + int8_t oldToULength, toULength, toULimit; + + UChar32 c; + uint8_t b, t1, t2; + + uint32_t stage2Entry; + uint32_t asciiRoundtrips; + uint16_t value = 0; + UBool hasSupplementary; + + /* set up the local pointers */ + utf8=pToUArgs->converter; + cnv=pFromUArgs->converter; + source=(uint8_t *)pToUArgs->source; + sourceLimit=(uint8_t *)pToUArgs->sourceLimit; + target=(uint8_t *)pFromUArgs->target; + targetCapacity=(int32_t)(pFromUArgs->targetLimit-pFromUArgs->target); + + table=cnv->sharedData->mbcs.fromUnicodeTable; + mbcsIndex=cnv->sharedData->mbcs.mbcsIndex; + if((cnv->options&UCNV_OPTION_SWAP_LFNL)!=0) { + results=(uint16_t *)cnv->sharedData->mbcs.swapLFNLFromUnicodeBytes; + } else { + results=(uint16_t *)cnv->sharedData->mbcs.fromUnicodeBytes; + } + asciiRoundtrips=cnv->sharedData->mbcs.asciiRoundtrips; + + hasSupplementary=(UBool)(cnv->sharedData->mbcs.unicodeMask&UCNV_HAS_SUPPLEMENTARY); + + /* get the converter state from the UTF-8 UConverter */ + if(utf8->toULength > 0) { + toULength=oldToULength=utf8->toULength; + toULimit=(int8_t)utf8->mode; + c=(UChar32)utf8->toUnicodeStatus; + } else { + toULength=oldToULength=toULimit=0; + c = 0; + } + + // The conversion loop checks source0) { + uint8_t b1=*(sourceLimit-1); + if(U8_IS_SINGLE(b1)) { + // common ASCII character + } else if(U8_IS_TRAIL(b1) && length>=2) { + uint8_t b2=*(sourceLimit-2); + if(0xe0<=b2 && b2<0xf0 && U8_IS_VALID_LEAD3_AND_T1(b2, b1)) { + // truncated 3-byte sequence + sourceLimit-=2; + } + } else if(0xc2<=b1 && b1<0xf0) { + // truncated 2- or 3-byte sequence + --sourceLimit; + } + } + } + + if(c!=0 && targetCapacity>0) { + utf8->toUnicodeStatus=0; + utf8->toULength=0; + goto moreBytes; + /* See note in ucnv_SBCSFromUTF8() about this goto. */ + } + + /* conversion loop */ + while(source0) { + b=*source++; + if(U8_IS_SINGLE(b)) { + /* convert ASCII */ + if(IS_ASCII_ROUNDTRIP(b, asciiRoundtrips)) { + *target++=b; + --targetCapacity; + continue; + } else { + value=DBCS_RESULT_FROM_UTF8(mbcsIndex, results, 0, b); + if(value==0) { + c=b; + goto unassigned; + } + } + } else { + if(b>=0xe0) { + if( /* handle U+0800..U+D7FF inline */ + b<=0xed && // do not assume maxFastUChar>0xd7ff + U8_IS_VALID_LEAD3_AND_T1(b, t1=source[0]) && + (t2=(uint8_t)(source[1]-0x80)) <= 0x3f + ) { + c=((b&0xf)<<6)|(t1&0x3f); + source+=2; + value=DBCS_RESULT_FROM_UTF8(mbcsIndex, results, c, t2); + if(value==0) { + c=(c<<6)|t2; + goto unassigned; + } + } else { + c=-1; + } + } else { + if( /* handle U+0080..U+07FF inline */ + b>=0xc2 && + (t1=(uint8_t)(*source-0x80)) <= 0x3f + ) { + c=b&0x1f; + ++source; + value=DBCS_RESULT_FROM_UTF8(mbcsIndex, results, c, t1); + if(value==0) { + c=(c<<6)|t1; + goto unassigned; + } + } else { + c=-1; + } + } + + if(c<0) { + /* handle "complicated" and error cases, and continuing partial characters */ + oldToULength=0; + toULength=1; + toULimit=U8_COUNT_BYTES_NON_ASCII(b); + c=b; +moreBytes: + while(toULengthsourceLimit) { + b=*source; + if(icu::UTF8::isValidTrail(c, b, toULength, toULimit)) { + ++source; + ++toULength; + c=(c<<6)+b; + } else { + break; /* sequence too short, stop with toULengthtoUBytes[oldToULength++]=*source++; + } + utf8->toUnicodeStatus=c; + utf8->toULength=toULength; + utf8->mode=toULimit; + pToUArgs->source=(char *)source; + pFromUArgs->target=(char *)target; + return; + } + } + + if(toULength==toULimit) { + c-=utf8_offsets[toULength]; + if(toULength<=3) { /* BMP */ + stage2Entry=MBCS_STAGE_2_FROM_U(table, c); + } else { + /* supplementary code point */ + if(!hasSupplementary) { + /* BMP-only codepages are stored without stage 1 entries for supplementary code points */ + stage2Entry=0; + } else { + stage2Entry=MBCS_STAGE_2_FROM_U(table, c); + } + } + } else { + /* error handling: illegal UTF-8 byte sequence */ + source-=(toULength-oldToULength); + while(oldToULengthtoUBytes[oldToULength++]=*source++; + } + utf8->toULength=toULength; + pToUArgs->source=(char *)source; + pFromUArgs->target=(char *)target; + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + return; + } + + /* get the bytes and the length for the output */ + /* MBCS_OUTPUT_2 */ + value=MBCS_VALUE_2_FROM_STAGE_2(results, stage2Entry, c); + + /* is this code point assigned, or do we use fallbacks? */ + if(!(MBCS_FROM_U_IS_ROUNDTRIP(stage2Entry, c) || + (UCNV_FROM_U_USE_FALLBACK(cnv, c) && value!=0)) + ) { + goto unassigned; + } + } + } + + /* write the output character bytes from value and length */ + /* from the first if in the loop we know that targetCapacity>0 */ + if(value<=0xff) { + /* this is easy because we know that there is enough space */ + *target++=(uint8_t)value; + --targetCapacity; + } else /* length==2 */ { + *target++=(uint8_t)(value>>8); + if(2<=targetCapacity) { + *target++=(uint8_t)value; + targetCapacity-=2; + } else { + cnv->charErrorBuffer[0]=(char)value; + cnv->charErrorBufferLength=1; + + /* target overflow */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + continue; + +unassigned: + { + /* + * Try an extension mapping. + * Pass in no source because we don't have UTF-16 input. + * If we have a partial match on c, we will return and revert + * to UTF-8->UTF-16->charset conversion. + */ + static const char16_t nul=0; + const char16_t *noSource=&nul; + c=_extFromU(cnv, cnv->sharedData, + c, &noSource, noSource, + &target, target+targetCapacity, + nullptr, -1, + pFromUArgs->flush, + pErrorCode); + + if(U_FAILURE(*pErrorCode)) { + /* not mappable or buffer overflow */ + cnv->fromUChar32=c; + break; + } else if(cnv->preFromUFirstCP>=0) { + /* + * Partial match, return and revert to pivoting. + * In normal from-UTF-16 conversion, we would just continue + * but then exit the loop because the extension match would + * have consumed the source. + */ + *pErrorCode=U_USING_DEFAULT_WARNING; + break; + } else { + /* a mapping was written to the target, continue */ + + /* recalculate the targetCapacity after an extension mapping */ + targetCapacity=(int32_t)(pFromUArgs->targetLimit-(char *)target); + continue; + } + } + } else { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + } + + /* + * The sourceLimit may have been adjusted before the conversion loop + * to stop before a truncated sequence. + * If so, then collect the truncated sequence now. + */ + if(U_SUCCESS(*pErrorCode) && + cnv->preFromUFirstCP<0 && + source<(sourceLimit=(uint8_t *)pToUArgs->sourceLimit)) { + c=utf8->toUBytes[0]=b=*source++; + toULength=1; + toULimit=U8_COUNT_BYTES(b); + while(sourcetoUBytes[toULength++]=b=*source++; + c=(c<<6)+b; + } + utf8->toUnicodeStatus=c; + utf8->toULength=toULength; + utf8->mode=toULimit; + } + + /* write back the updated pointers */ + pToUArgs->source=(char *)source; + pFromUArgs->target=(char *)target; +} + +/* miscellaneous ------------------------------------------------------------ */ + +static void U_CALLCONV +ucnv_MBCSGetStarters(const UConverter* cnv, + UBool starters[256], + UErrorCode *) { + const int32_t *state0; + int i; + + state0=cnv->sharedData->mbcs.stateTable[cnv->sharedData->mbcs.dbcsOnlyState]; + for(i=0; i<256; ++i) { + /* all bytes that cause a state transition from state 0 are lead bytes */ + starters[i]= (UBool)MBCS_ENTRY_IS_TRANSITION(state0[i]); + } +} + +/* + * This is an internal function that allows other converter implementations + * to check whether a byte is a lead byte. + */ +U_CFUNC UBool +ucnv_MBCSIsLeadByte(UConverterSharedData *sharedData, char byte) { + return (UBool)MBCS_ENTRY_IS_TRANSITION(sharedData->mbcs.stateTable[0][(uint8_t)byte]); +} + +static void U_CALLCONV +ucnv_MBCSWriteSub(UConverterFromUnicodeArgs *pArgs, + int32_t offsetIndex, + UErrorCode *pErrorCode) { + UConverter *cnv=pArgs->converter; + char *p, *subchar; + char buffer[4]; + int32_t length; + + /* first, select between subChar and subChar1 */ + if( cnv->subChar1!=0 && + (cnv->sharedData->mbcs.extIndexes!=nullptr ? + cnv->useSubChar1 : + (cnv->invalidUCharBuffer[0]<=0xff)) + ) { + /* select subChar1 if it is set (not 0) and the unmappable Unicode code point is up to U+00ff (IBM MBCS behavior) */ + subchar=(char *)&cnv->subChar1; + length=1; + } else { + /* select subChar in all other cases */ + subchar=(char *)cnv->subChars; + length=cnv->subCharLen; + } + + /* reset the selector for the next code point */ + cnv->useSubChar1=false; + + if (cnv->sharedData->mbcs.outputType == MBCS_OUTPUT_2_SISO) { + p=buffer; + + /* fromUnicodeStatus contains prevLength */ + switch(length) { + case 1: + if(cnv->fromUnicodeStatus==2) { + /* DBCS mode and SBCS sub char: change to SBCS */ + cnv->fromUnicodeStatus=1; + *p++=UCNV_SI; + } + *p++=subchar[0]; + break; + case 2: + if(cnv->fromUnicodeStatus<=1) { + /* SBCS mode and DBCS sub char: change to DBCS */ + cnv->fromUnicodeStatus=2; + *p++=UCNV_SO; + } + *p++=subchar[0]; + *p++=subchar[1]; + break; + default: + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + subchar=buffer; + length=(int32_t)(p-buffer); + } + + ucnv_cbFromUWriteBytes(pArgs, subchar, length, offsetIndex, pErrorCode); +} + +U_CFUNC UConverterType +ucnv_MBCSGetType(const UConverter* converter) { + /* SBCS, DBCS, and EBCDIC_STATEFUL are replaced by MBCS, but here we cheat a little */ + if(converter->sharedData->mbcs.countStates==1) { + return (UConverterType)UCNV_SBCS; + } else if((converter->sharedData->mbcs.outputType&0xff)==MBCS_OUTPUT_2_SISO) { + return (UConverterType)UCNV_EBCDIC_STATEFUL; + } else if(converter->sharedData->staticData->minBytesPerChar==2 && converter->sharedData->staticData->maxBytesPerChar==2) { + return (UConverterType)UCNV_DBCS; + } + return (UConverterType)UCNV_MBCS; +} + +#endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */ diff --git a/intl/icu/source/common/ucnvmbcs.h b/intl/icu/source/common/ucnvmbcs.h new file mode 100644 index 0000000000..c8f3b89a5e --- /dev/null +++ b/intl/icu/source/common/ucnvmbcs.h @@ -0,0 +1,605 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2000-2013, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: ucnvmbcs.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2000jul07 +* created by: Markus W. Scherer +*/ + +#ifndef __UCNVMBCS_H__ +#define __UCNVMBCS_H__ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/ucnv.h" +#include "ucnv_cnv.h" +#include "ucnv_ext.h" + +/** + * ICU conversion (.cnv) data file structure, following the usual UDataInfo + * header. + * + * Format version: 6.2 + * + * struct UConverterStaticData -- struct containing the converter name, IBM CCSID, + * min/max bytes per character, etc. + * see ucnv_bld.h + * + * -------------------- + * + * The static data is followed by conversionType-specific data structures. + * At the moment, there are only variations of MBCS converters. They all have + * the same toUnicode structures, while the fromUnicode structures for SBCS + * differ from those for other MBCS-style converters. + * + * _MBCSHeader.version 5 is optional and not backward-compatible + * (as usual for changes in the major version field). + * + * Versions 5.m work like versions 4.m except: + * - The _MBCSHeader has variable length (and is always longer than in version 4). + * See the struct _MBCSHeader further description below. + * - There is a set of flags which indicate further incompatible changes. + * (Reader code must reject the file if it does not recognize them all.) + * - In particular, one of these flags indicates that most of the fromUnicode + * data is missing and must be reconstituted from the toUnicode data + * and from the utf8Friendly mbcsIndex at load time. + * (This only works with a utf8Friendly table.) + * In this case, makeconv may increase maxFastUChar automatically to U+FFFF. + * + * The first of these versions is 5.3, which is like 4.3 except for the differences above. + * + * When possible, makeconv continues to generate version 4.m files. + * + * _MBCSHeader.version 5.4/4.4 supports "good one-way" mappings (|4) + * in the extension tables (fromUTableValues bit 30). See ucnv_ext.h for details. + * + * _MBCSHeader.version 4.3 optionally modifies the fromUnicode data structures + * slightly and optionally adds a table for conversion to MBCS (non-SBCS) + * charsets. + * + * The modifications are to make the data utf8Friendly. Not every 4.3 file + * file contains utf8Friendly data. + * It is utf8Friendly if _MBCSHeader.version[2]!=0. + * In this case, the data structures are utf8Friendly up to the code point + * maxFastUChar=((_MBCSHeader.version[2]<<8)|0xff) + * + * A utf8Friendly file has fromUnicode stage 3 entries for code points up to + * maxFastUChar allocated in blocks of 64 for indexing with the 6 bits from + * a UTF-8 trail byte. ASCII is allocated linearly with 128 contiguous entries. + * + * In addition, a utf8Friendly MBCS file contains an additional + * uint16_t mbcsIndex[(maxFastUChar+1)>>6]; + * which replaces the stage 1 and 2 tables for indexing with bits from the + * UTF-8 lead byte and middle trail byte. Unlike the older MBCS stage 2 table, + * the mbcsIndex does not contain roundtrip flags. Therefore, all fallbacks + * from code points up to maxFastUChar (and roundtrips to 0x00) are moved to + * the extension data structure. This also allows for faster roundtrip + * conversion from UTF-16. + * + * SBCS files do not contain an additional sbcsIndex[] array because the + * proportional size increase would be noticeable, but the runtime + * code builds one for the code point range for which the runtime conversion + * code is optimized. + * + * For SBCS, maxFastUChar should be at least U+0FFF. The initial makeconv + * implementation sets it to U+1FFF. Because the sbcsIndex is not stored in + * the file, a larger maxFastUChar only affects stage 3 block allocation size + * and is free in empty blocks. (Larger blocks with sparse contents cause larger + * files.) U+1FFF includes almost all of the small scripts. + * U+0FFF covers UTF-8 two-byte sequences and three-byte sequences starting with + * 0xe0. This includes most scripts with legacy SBCS charsets. + * The initial runtime implementation using 4.3 files only builds an sbcsIndex + * for code points up to U+0FFF. + * + * For MBCS, maxFastUChar should be at least U+D7FF (=initial value). + * This boundary is convenient because practically all of the commonly used + * characters are below it, and because it is the boundary to surrogate + * code points, above which special handling is necessary anyway. + * (Surrogate pair assembly for UTF-16, validity checking for UTF-8.) + * + * maxFastUChar could be up to U+FFFF to cover the whole BMP, which could be + * useful especially for conversion from UTF-8 when the input can be assumed + * to be valid, because the surrogate range would then not have to be + * checked. + * (With maxFastUChar=0xffff, makeconv would have to check for mbcsIndex value + * overflow because with the all-unassigned block 0 and nearly full mappings + * from the BMP it is theoretically possible that an index into stage 3 + * exceeds 16 bits.) + * + * _MBCSHeader.version 4.2 adds an optional conversion extension data structure. + * If it is present, then an ICU version reading header versions 4.0 or 4.1 + * will be able to use the base table and ignore the extension. + * + * The unicodeMask in the static data is part of the base table data structure. + * Especially, the UCNV_HAS_SUPPLEMENTARY flag determines the length of the + * fromUnicode stage 1 array. + * The static data unicodeMask refers only to the base table's properties if + * a base table is included. + * In an extension-only file, the static data unicodeMask is 0. + * The extension data indexes have a separate field with the unicodeMask flags. + * + * MBCS-style data structure following the static data. + * Offsets are counted in bytes from the beginning of the MBCS header structure. + * Details about usage in comments in ucnvmbcs.c. + * + * struct _MBCSHeader (see the definition in this header file below) + * contains 32-bit fields as follows: + * 8 values: + * 0 uint8_t[4] MBCS version in UVersionInfo format (currently 4.3.x.0) + * 1 uint32_t countStates + * 2 uint32_t countToUFallbacks + * 3 uint32_t offsetToUCodeUnits + * 4 uint32_t offsetFromUTable + * 5 uint32_t offsetFromUBytes + * 6 uint32_t flags, bits: + * 31.. 8 offsetExtension -- _MBCSHeader.version 4.2 (ICU 2.8) and higher + * 0 for older versions and if + * there is not extension structure + * 7.. 0 outputType + * 7 uint32_t fromUBytesLength -- _MBCSHeader.version 4.1 (ICU 2.4) and higher + * counts bytes in fromUBytes[] + * + * New and required in version 5: + * 8 uint32_t options, bits: + * 31..16 reserved for flags that can be added without breaking + * backward compatibility + * 15.. 6 reserved for flags whose addition will break + * backward compatibility + * 6 MBCS_OPT_FROM_U -- if set, + * then most of the fromUnicode data is omitted; + * fullStage2Length is present and the missing + * bottom part of stage 2 must be reconstituted from + * the toUnicode data; + * stage 3 is missing completely as well; + * not used for SBCS tables + * 5.. 0 length of the _MBCSHeader (number of uint32_t) + * + * New and optional in version 5: + * 9 uint32_t fullStage2Length: used if MBCS_OPT_FROM_U is set + * specifies the full length of stage 2 + * including the omitted part + * + * if(outputType==MBCS_OUTPUT_EXT_ONLY) { + * -- base table name for extension-only table + * char baseTableName[variable]; -- with NUL plus padding for 4-alignment + * + * -- all _MBCSHeader fields except for version and flags are 0 + * } else { + * -- normal base table with optional extension + * + * int32_t stateTable[countStates][256]; + * + * struct _MBCSToUFallback { (fallbacks are sorted by offset) + * uint32_t offset; + * UChar32 codePoint; + * } toUFallbacks[countToUFallbacks]; + * + * uint16_t unicodeCodeUnits[(offsetFromUTable-offsetToUCodeUnits)/2]; + * (padded to an even number of units) + * + * -- stage 1 tables + * if(staticData.unicodeMask&UCNV_HAS_SUPPLEMENTARY) { + * -- stage 1 table for all of Unicode + * uint16_t fromUTable[0x440]; (32-bit-aligned) + * } else { + * -- BMP-only tables have a smaller stage 1 table + * uint16_t fromUTable[0x40]; (32-bit-aligned) + * } + * + * -- stage 2 tables + * length determined by top of stage 1 and bottom of stage 3 tables + * if(outputType==MBCS_OUTPUT_1) { + * -- SBCS: pure indexes + * uint16_t stage 2 indexes[?]; + * } else { + * -- DBCS, MBCS, EBCDIC_STATEFUL, ...: roundtrip flags and indexes + * uint32_t stage 2 flags and indexes[?]; + * if(options&MBCS_OPT_NO_FROM_U) { + * stage 2 really has length fullStage2Length + * and the omitted lower part must be reconstituted from + * the toUnicode data + * } + * } + * + * -- stage 3 tables with byte results + * if(outputType==MBCS_OUTPUT_1) { + * -- SBCS: each 16-bit result contains flags and the result byte, see ucnvmbcs.c + * uint16_t fromUBytes[fromUBytesLength/2]; + * } else if(!(options&MBCS_OPT_NO_FROM_U)) { + * -- DBCS, MBCS, EBCDIC_STATEFUL, ... 2/3/4 bytes result, see ucnvmbcs.c + * uint8_t fromUBytes[fromUBytesLength]; or + * uint16_t fromUBytes[fromUBytesLength/2]; or + * uint32_t fromUBytes[fromUBytesLength/4]; + * } else { + * fromUBytes[] must be reconstituted from the toUnicode data + * } + * + * -- optional utf8Friendly mbcsIndex -- _MBCSHeader.version 4.3 (ICU 3.8) and higher + * if(outputType!=MBCS_OUTPUT_1 && + * _MBCSHeader.version[1]>=3 && + * (maxFastUChar=_MBCSHeader.version[2])!=0 + * ) { + * maxFastUChar=(maxFastUChar<<8)|0xff; + * uint16_t mbcsIndex[(maxFastUChar+1)>>6]; + * } + * } + * + * -- extension table, details see ucnv_ext.h + * int32_t indexes[>=32]; ... + */ + +/* MBCS converter data and state -------------------------------------------- */ + +enum { + MBCS_MAX_STATE_COUNT=128 +}; + +/** + * MBCS action codes for conversions to Unicode. + * These values are in bits 23..20 of the state table entries. + */ +enum { + MBCS_STATE_VALID_DIRECT_16, + MBCS_STATE_VALID_DIRECT_20, + + MBCS_STATE_FALLBACK_DIRECT_16, + MBCS_STATE_FALLBACK_DIRECT_20, + + MBCS_STATE_VALID_16, + MBCS_STATE_VALID_16_PAIR, + + MBCS_STATE_UNASSIGNED, + MBCS_STATE_ILLEGAL, + + MBCS_STATE_CHANGE_ONLY +}; + +/* Macros for state table entries */ +#define MBCS_ENTRY_TRANSITION(state, offset) (int32_t)(((int32_t)(state)<<24L)|(offset)) +#define MBCS_ENTRY_TRANSITION_SET_OFFSET(entry, offset) (int32_t)(((entry)&0xff000000)|(offset)) +#define MBCS_ENTRY_TRANSITION_ADD_OFFSET(entry, offset) (int32_t)((entry)+(offset)) + +#define MBCS_ENTRY_FINAL(state, action, value) (int32_t)(0x80000000|((int32_t)(state)<<24L)|((action)<<20L)|(value)) +#define MBCS_ENTRY_SET_FINAL(entry) (int32_t)((entry)|0x80000000) +#define MBCS_ENTRY_FINAL_SET_ACTION(entry, action) (int32_t)(((entry)&0xff0fffff)|((int32_t)(action)<<20L)) +#define MBCS_ENTRY_FINAL_SET_VALUE(entry, value) (int32_t)(((entry)&0xfff00000)|(value)) +#define MBCS_ENTRY_FINAL_SET_ACTION_VALUE(entry, action, value) (int32_t)(((entry)&0xff000000)|((int32_t)(action)<<20L)|(value)) + +#define MBCS_ENTRY_SET_STATE(entry, state) (int32_t)(((entry)&0x80ffffff)|((int32_t)(state)<<24L)) + +#define MBCS_ENTRY_STATE(entry) ((((uint32_t)entry)>>24)&0x7f) + +#define MBCS_ENTRY_IS_TRANSITION(entry) ((entry)>=0) +#define MBCS_ENTRY_IS_FINAL(entry) ((entry)<0) + +#define MBCS_ENTRY_TRANSITION_STATE(entry) (((uint32_t)entry)>>24) +#define MBCS_ENTRY_TRANSITION_OFFSET(entry) ((entry)&0xffffff) + +#define MBCS_ENTRY_FINAL_STATE(entry) ((((uint32_t)entry)>>24)&0x7f) +#define MBCS_ENTRY_FINAL_IS_VALID_DIRECT_16(entry) ((entry)<(int32_t)0x80100000) +#define MBCS_ENTRY_FINAL_ACTION(entry) ((((uint32_t)entry)>>20)&0xf) +#define MBCS_ENTRY_FINAL_VALUE(entry) ((entry)&0xfffff) +#define MBCS_ENTRY_FINAL_VALUE_16(entry) (uint16_t)(entry) + +#define IS_ASCII_ROUNDTRIP(b, asciiRoundtrips) (((asciiRoundtrips) & (1<<((b)>>2)))!=0) + +/* single-byte fromUnicode: get the 16-bit result word */ +#define MBCS_SINGLE_RESULT_FROM_U(table, results, c) (results)[ (table)[ (table)[(c)>>10] +(((c)>>4)&0x3f) ] +((c)&0xf) ] + +/* single-byte fromUnicode using the sbcsIndex */ +#define SBCS_RESULT_FROM_LOW_BMP(table, results, c) (results)[ (table)[(c)>>6] +((c)&0x3f) ] + +/* single-byte fromUTF8 using the sbcsIndex; l and t must be masked externally; can be l=0 and t<=0x7f */ +#define SBCS_RESULT_FROM_UTF8(table, results, l, t) (results)[ (table)[l] +(t) ] + +/* multi-byte fromUnicode: get the 32-bit stage 2 entry */ +#define MBCS_STAGE_2_FROM_U(table, c) ((const uint32_t *)(table))[ (table)[(c)>>10] +(((c)>>4)&0x3f) ] +#define MBCS_FROM_U_IS_ROUNDTRIP(stage2Entry, c) ( ((stage2Entry) & ((uint32_t)1<< (16+((c)&0xf)) )) !=0) + +#define MBCS_VALUE_2_FROM_STAGE_2(bytes, stage2Entry, c) ((uint16_t *)(bytes))[16*(uint32_t)(uint16_t)(stage2Entry)+((c)&0xf)] +#define MBCS_VALUE_4_FROM_STAGE_2(bytes, stage2Entry, c) ((uint32_t *)(bytes))[16*(uint32_t)(uint16_t)(stage2Entry)+((c)&0xf)] + +#define MBCS_POINTER_3_FROM_STAGE_2(bytes, stage2Entry, c) ((bytes)+(16*(uint32_t)(uint16_t)(stage2Entry)+((c)&0xf))*3) + +/* double-byte fromUnicode using the mbcsIndex */ +#define DBCS_RESULT_FROM_MOST_BMP(table, results, c) (results)[ (table)[(c)>>6] +((c)&0x3f) ] + +/* double-byte fromUTF8 using the mbcsIndex; l and t1 combined into lt1; lt1 and t2 must be masked externally */ +#define DBCS_RESULT_FROM_UTF8(table, results, lt1, t2) (results)[ (table)[lt1] +(t2) ] + + +/** + * MBCS output types for conversions from Unicode. + * These per-converter types determine the storage method in stage 3 of the lookup table, + * mostly how many bytes are stored per entry. + */ +enum { + MBCS_OUTPUT_1, /* 0 */ + MBCS_OUTPUT_2, /* 1 */ + MBCS_OUTPUT_3, /* 2 */ + MBCS_OUTPUT_4, /* 3 */ + + MBCS_OUTPUT_3_EUC=8, /* 8 */ + MBCS_OUTPUT_4_EUC, /* 9 */ + + MBCS_OUTPUT_2_SISO=12, /* c */ + MBCS_OUTPUT_2_HZ, /* d */ + + MBCS_OUTPUT_EXT_ONLY, /* e */ + + MBCS_OUTPUT_COUNT, + + MBCS_OUTPUT_DBCS_ONLY=0xdb /* runtime-only type for DBCS-only handling of SISO tables */ +}; + +/** + * Fallbacks to Unicode are stored outside the normal state table and code point structures + * in a vector of items of this type. They are sorted by offset. + */ +typedef struct { + uint32_t offset; + UChar32 codePoint; +} _MBCSToUFallback; + +/** Constants for fast and UTF-8-friendly conversion. */ +enum { + SBCS_FAST_MAX=0x0fff, /* maximum code point with UTF-8-friendly SBCS runtime code, see makeconv SBCS_UTF8_MAX */ + SBCS_FAST_LIMIT=SBCS_FAST_MAX+1, /* =0x1000 */ + MBCS_FAST_MAX=0xd7ff, /* maximum code point with UTF-8-friendly MBCS runtime code, see makeconv MBCS_UTF8_MAX */ + MBCS_FAST_LIMIT=MBCS_FAST_MAX+1 /* =0xd800 */ +}; + +/** + * This is the MBCS part of the UConverterTable union (a runtime data structure). + * It keeps all the per-converter data and points into the loaded mapping tables. + * + * utf8Friendly data structures added with _MBCSHeader.version 4.3 + */ +typedef struct UConverterMBCSTable { + /* toUnicode */ + uint8_t countStates, dbcsOnlyState, stateTableOwned; + uint32_t countToUFallbacks; + + const int32_t (*stateTable)/*[countStates]*/[256]; + int32_t (*swapLFNLStateTable)/*[countStates]*/[256]; /* for swaplfnl */ + const uint16_t *unicodeCodeUnits/*[countUnicodeResults]*/; + const _MBCSToUFallback *toUFallbacks; + + /* fromUnicode */ + const uint16_t *fromUnicodeTable; + const uint16_t *mbcsIndex; /* for fast conversion from most of BMP to MBCS (utf8Friendly data) */ + uint16_t sbcsIndex[SBCS_FAST_LIMIT>>6]; /* for fast conversion from low BMP to SBCS (utf8Friendly data) */ + const uint8_t *fromUnicodeBytes; + uint8_t *swapLFNLFromUnicodeBytes; /* for swaplfnl */ + uint32_t fromUBytesLength; + uint8_t outputType, unicodeMask; + UBool utf8Friendly; /* for utf8Friendly data */ + UChar maxFastUChar; /* for utf8Friendly data */ + + /* roundtrips */ + uint32_t asciiRoundtrips; + + /* reconstituted data that was omitted from the .cnv file */ + uint8_t *reconstitutedData; + + /* converter name for swaplfnl */ + char *swapLFNLName; + + /* extension data */ + struct UConverterSharedData *baseSharedData; + const int32_t *extIndexes; +} UConverterMBCSTable; + +#define UCNV_MBCS_TABLE_INITIALIZER { \ + /* toUnicode */ \ + 0, 0, 0, \ + 0, \ + \ + NULL, \ + NULL, \ + NULL, \ + NULL, \ + \ + /* fromUnicode */ \ + NULL, \ + NULL, \ + { 0 }, \ + NULL, \ + NULL, \ + 0, \ + 0, 0, \ + false, \ + 0, \ + \ + /* roundtrips */ \ + 0, \ + \ + /* reconstituted data that was omitted from the .cnv file */ \ + NULL, \ + \ + /* converter name for swaplfnl */ \ + NULL, \ + \ + /* extension data */ \ + NULL, \ + NULL \ +} + +enum { + MBCS_OPT_LENGTH_MASK=0x3f, + MBCS_OPT_NO_FROM_U=0x40, + /* + * If any of the following options bits are set, + * then the file must be rejected. + */ + MBCS_OPT_INCOMPATIBLE_MASK=0xffc0, + /* + * Remove bits from this mask as more options are recognized + * by all implementations that use this constant. + */ + MBCS_OPT_UNKNOWN_INCOMPATIBLE_MASK=0xff80 +}; + +enum { + MBCS_HEADER_V4_LENGTH=8, + MBCS_HEADER_V5_MIN_LENGTH=9 +}; + +/** + * MBCS data header. See data format description above. + */ +typedef struct { + UVersionInfo version; + uint32_t countStates, + countToUFallbacks, + offsetToUCodeUnits, + offsetFromUTable, + offsetFromUBytes, + flags, + fromUBytesLength; + + /* new and required in version 5 */ + uint32_t options; + + /* new and optional in version 5; used if options&MBCS_OPT_NO_FROM_U */ + uint32_t fullStage2Length; /* number of 32-bit units */ +} _MBCSHeader; + +#define UCNV_MBCS_HEADER_INITIALIZER { { 0 }, 0, 0, 0, 0, 0, 0, 0, 0, 0 } + +/* + * This is a simple version of _MBCSGetNextUChar() that is used + * by other converter implementations. + * It only returns an "assigned" result if it consumes the entire input. + * It does not use state from the converter, nor error codes. + * It does not handle the EBCDIC swaplfnl option (set in UConverter). + * It handles conversion extensions but not GB 18030. + * + * Return value: + * U+fffe unassigned + * U+ffff illegal + * otherwise the Unicode code point + */ +U_CFUNC UChar32 +ucnv_MBCSSimpleGetNextUChar(UConverterSharedData *sharedData, + const char *source, int32_t length, + UBool useFallback); + +/** + * This version of _MBCSSimpleGetNextUChar() is optimized for single-byte, single-state codepages. + * It does not handle the EBCDIC swaplfnl option (set in UConverter). + * It does not handle conversion extensions (_extToU()). + */ +U_CFUNC UChar32 +ucnv_MBCSSingleSimpleGetNextUChar(UConverterSharedData *sharedData, + uint8_t b, UBool useFallback); + +/** + * This macro version of _MBCSSingleSimpleGetNextUChar() gets a code point from a byte. + * It works for single-byte, single-state codepages that only map + * to and from BMP code points, and it always + * returns fallback values. + */ +#define _MBCS_SINGLE_SIMPLE_GET_NEXT_BMP(sharedData, b) \ + (UChar)MBCS_ENTRY_FINAL_VALUE_16((sharedData)->mbcs.stateTable[0][(uint8_t)(b)]) + +/** + * This is an internal function that allows other converter implementations + * to check whether a byte is a lead byte. + */ +U_CFUNC UBool +ucnv_MBCSIsLeadByte(UConverterSharedData *sharedData, char byte); + +/** This is a macro version of _MBCSIsLeadByte(). */ +#define _MBCS_IS_LEAD_BYTE(sharedData, byte) \ + (UBool)MBCS_ENTRY_IS_TRANSITION((sharedData)->mbcs.stateTable[0][(uint8_t)(byte)]) + +/* + * This is another simple conversion function for internal use by other + * conversion implementations. + * It does not use the converter state nor call callbacks. + * It does not handle the EBCDIC swaplfnl option (set in UConverter). + * It handles conversion extensions but not GB 18030. + * + * It converts one single Unicode code point into codepage bytes, encoded + * as one 32-bit value. The function returns the number of bytes in *pValue: + * 1..4 the number of bytes in *pValue + * 0 unassigned (*pValue undefined) + * -1 illegal (currently not used, *pValue undefined) + * + * *pValue will contain the resulting bytes with the last byte in bits 7..0, + * the second to last byte in bits 15..8, etc. + * Currently, the function assumes but does not check that 0<=c<=0x10ffff. + */ +U_CFUNC int32_t +ucnv_MBCSFromUChar32(UConverterSharedData *sharedData, + UChar32 c, uint32_t *pValue, + UBool useFallback); + +/** + * This version of _MBCSFromUChar32() is optimized for single-byte codepages. + * It does not handle the EBCDIC swaplfnl option (set in UConverter). + * + * It returns the codepage byte for the code point, or -1 if it is unassigned. + */ +U_CFUNC int32_t +ucnv_MBCSSingleFromUChar32(UConverterSharedData *sharedData, + UChar32 c, + UBool useFallback); + +/** + * SBCS, DBCS, and EBCDIC_STATEFUL are replaced by MBCS, but + * we cheat a little about the type, returning the old types if appropriate. + */ +U_CFUNC UConverterType +ucnv_MBCSGetType(const UConverter* converter); + +U_CFUNC void +ucnv_MBCSFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode); +U_CFUNC void +ucnv_MBCSToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode); + +/* + * Internal function returning a UnicodeSet for toUnicode() conversion. + * Currently only used for ISO-2022-CN, and only handles roundtrip mappings. + * In the future, if we add support for fallback sets, this function + * needs to be updated. + * Handles extensions. + * Does not empty the set first. + */ +U_CFUNC void +ucnv_MBCSGetUnicodeSetForUnicode(const UConverterSharedData *sharedData, + const USetAdder *sa, + UConverterUnicodeSet which, + UErrorCode *pErrorCode); + +/* + * Same as ucnv_MBCSGetUnicodeSetForUnicode() but + * the set can be filtered by encoding scheme. + * Used by stateful converters which share regular conversion tables + * but only use a subset of their mappings. + */ +U_CFUNC void +ucnv_MBCSGetFilteredUnicodeSetForUnicode(const UConverterSharedData *sharedData, + const USetAdder *sa, + UConverterUnicodeSet which, + UConverterSetFilter filter, + UErrorCode *pErrorCode); + +#endif + +#endif diff --git a/intl/icu/source/common/ucnvscsu.cpp b/intl/icu/source/common/ucnvscsu.cpp new file mode 100644 index 0000000000..2138e289ca --- /dev/null +++ b/intl/icu/source/common/ucnvscsu.cpp @@ -0,0 +1,2045 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2000-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: ucnvscsu.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2000nov18 +* created by: Markus W. Scherer +* +* This is an implementation of the Standard Compression Scheme for Unicode +* as defined in https://www.unicode.org/reports/tr6/ . +* Reserved commands and window settings are treated as illegal sequences and +* will result in callback calls. +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/ucnv_cb.h" +#include "unicode/utf16.h" +#include "ucnv_bld.h" +#include "ucnv_cnv.h" +#include "cmemory.h" + +/* SCSU definitions --------------------------------------------------------- */ + +/* SCSU command byte values */ +enum { + SQ0=0x01, /* Quote from window pair 0 */ + SQ7=0x08, /* Quote from window pair 7 */ + SDX=0x0B, /* Define a window as extended */ + Srs=0x0C, /* reserved */ + SQU=0x0E, /* Quote a single Unicode character */ + SCU=0x0F, /* Change to Unicode mode */ + SC0=0x10, /* Select window 0 */ + SC7=0x17, /* Select window 7 */ + SD0=0x18, /* Define and select window 0 */ + SD7=0x1F, /* Define and select window 7 */ + + UC0=0xE0, /* Select window 0 */ + UC7=0xE7, /* Select window 7 */ + UD0=0xE8, /* Define and select window 0 */ + UD7=0xEF, /* Define and select window 7 */ + UQU=0xF0, /* Quote a single Unicode character */ + UDX=0xF1, /* Define a Window as extended */ + Urs=0xF2 /* reserved */ +}; + +enum { + /* + * Unicode code points from 3400 to E000 are not adressible by + * dynamic window, since in these areas no short run alphabets are + * found. Therefore add gapOffset to all values from gapThreshold. + */ + gapThreshold=0x68, + gapOffset=0xAC00, + + /* values between reservedStart and fixedThreshold are reserved */ + reservedStart=0xA8, + + /* use table of predefined fixed offsets for values from fixedThreshold */ + fixedThreshold=0xF9 +}; + +/* constant offsets for the 8 static windows */ +static const uint32_t staticOffsets[8]={ + 0x0000, /* ASCII for quoted tags */ + 0x0080, /* Latin - 1 Supplement (for access to punctuation) */ + 0x0100, /* Latin Extended-A */ + 0x0300, /* Combining Diacritical Marks */ + 0x2000, /* General Punctuation */ + 0x2080, /* Currency Symbols */ + 0x2100, /* Letterlike Symbols and Number Forms */ + 0x3000 /* CJK Symbols and punctuation */ +}; + +/* initial offsets for the 8 dynamic (sliding) windows */ +static const uint32_t initialDynamicOffsets[8]={ + 0x0080, /* Latin-1 */ + 0x00C0, /* Latin Extended A */ + 0x0400, /* Cyrillic */ + 0x0600, /* Arabic */ + 0x0900, /* Devanagari */ + 0x3040, /* Hiragana */ + 0x30A0, /* Katakana */ + 0xFF00 /* Fullwidth ASCII */ +}; + +/* Table of fixed predefined Offsets */ +static const uint32_t fixedOffsets[]={ + /* 0xF9 */ 0x00C0, /* Latin-1 Letters + half of Latin Extended A */ + /* 0xFA */ 0x0250, /* IPA extensions */ + /* 0xFB */ 0x0370, /* Greek */ + /* 0xFC */ 0x0530, /* Armenian */ + /* 0xFD */ 0x3040, /* Hiragana */ + /* 0xFE */ 0x30A0, /* Katakana */ + /* 0xFF */ 0xFF60 /* Halfwidth Katakana */ +}; + +/* state values */ +enum { + readCommand, + quotePairOne, + quotePairTwo, + quoteOne, + definePairOne, + definePairTwo, + defineOne +}; + +typedef struct SCSUData { + /* dynamic window offsets, initialize to default values from initialDynamicOffsets */ + uint32_t toUDynamicOffsets[8]; + uint32_t fromUDynamicOffsets[8]; + + /* state machine state - toUnicode */ + UBool toUIsSingleByteMode; + uint8_t toUState; + int8_t toUQuoteWindow, toUDynamicWindow; + uint8_t toUByteOne; + uint8_t toUPadding[3]; + + /* state machine state - fromUnicode */ + UBool fromUIsSingleByteMode; + int8_t fromUDynamicWindow; + + /* + * windowUse[] keeps track of the use of the dynamic windows: + * At nextWindowUseIndex there is the least recently used window, + * and the following windows (in a wrapping manner) are more and more + * recently used. + * At nextWindowUseIndex-1 there is the most recently used window. + */ + uint8_t locale; + int8_t nextWindowUseIndex; + int8_t windowUse[8]; +} SCSUData; + +static const int8_t initialWindowUse[8]={ 7, 0, 3, 2, 4, 5, 6, 1 }; +static const int8_t initialWindowUse_ja[8]={ 3, 2, 4, 1, 0, 7, 5, 6 }; + +enum { + lGeneric, l_ja +}; + +/* SCSU setup functions ----------------------------------------------------- */ +U_CDECL_BEGIN +static void U_CALLCONV +_SCSUReset(UConverter *cnv, UConverterResetChoice choice) { + SCSUData *scsu=(SCSUData *)cnv->extraInfo; + + if(choice<=UCNV_RESET_TO_UNICODE) { + /* reset toUnicode */ + uprv_memcpy(scsu->toUDynamicOffsets, initialDynamicOffsets, 32); + + scsu->toUIsSingleByteMode=true; + scsu->toUState=readCommand; + scsu->toUQuoteWindow=scsu->toUDynamicWindow=0; + scsu->toUByteOne=0; + + cnv->toULength=0; + } + if(choice!=UCNV_RESET_TO_UNICODE) { + /* reset fromUnicode */ + uprv_memcpy(scsu->fromUDynamicOffsets, initialDynamicOffsets, 32); + + scsu->fromUIsSingleByteMode=true; + scsu->fromUDynamicWindow=0; + + scsu->nextWindowUseIndex=0; + switch(scsu->locale) { + case l_ja: + uprv_memcpy(scsu->windowUse, initialWindowUse_ja, 8); + break; + default: + uprv_memcpy(scsu->windowUse, initialWindowUse, 8); + break; + } + + cnv->fromUChar32=0; + } +} + +static void U_CALLCONV +_SCSUOpen(UConverter *cnv, + UConverterLoadArgs *pArgs, + UErrorCode *pErrorCode) { + const char *locale=pArgs->locale; + if(pArgs->onlyTestIsLoadable) { + return; + } + cnv->extraInfo=uprv_malloc(sizeof(SCSUData)); + if(cnv->extraInfo!=nullptr) { + if(locale!=nullptr && locale[0]=='j' && locale[1]=='a' && (locale[2]==0 || locale[2]=='_')) { + ((SCSUData *)cnv->extraInfo)->locale=l_ja; + } else { + ((SCSUData *)cnv->extraInfo)->locale=lGeneric; + } + _SCSUReset(cnv, UCNV_RESET_BOTH); + } else { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + } + + /* Set the substitution character U+fffd as a Unicode string. */ + cnv->subUChars[0]=0xfffd; + cnv->subCharLen=-1; +} + +static void U_CALLCONV +_SCSUClose(UConverter *cnv) { + if(cnv->extraInfo!=nullptr) { + if(!cnv->isExtraLocal) { + uprv_free(cnv->extraInfo); + } + cnv->extraInfo=nullptr; + } +} + +/* SCSU-to-Unicode conversion functions ------------------------------------- */ + +static void U_CALLCONV +_SCSUToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + SCSUData *scsu; + const uint8_t *source, *sourceLimit; + char16_t *target; + const char16_t *targetLimit; + int32_t *offsets; + UBool isSingleByteMode; + uint8_t state, byteOne; + int8_t quoteWindow, dynamicWindow; + + int32_t sourceIndex, nextSourceIndex; + + uint8_t b; + + /* set up the local pointers */ + cnv=pArgs->converter; + scsu=(SCSUData *)cnv->extraInfo; + + source=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + target=pArgs->target; + targetLimit=pArgs->targetLimit; + offsets=pArgs->offsets; + + /* get the state machine state */ + isSingleByteMode=scsu->toUIsSingleByteMode; + state=scsu->toUState; + quoteWindow=scsu->toUQuoteWindow; + dynamicWindow=scsu->toUDynamicWindow; + byteOne=scsu->toUByteOne; + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex=state==readCommand ? 0 : -1; + nextSourceIndex=0; + + /* + * conversion "loop" + * + * For performance, this is not a normal C loop. + * Instead, there are two code blocks for the two SCSU modes. + * The function branches to either one, and a change of the mode is done with a goto to + * the other branch. + * + * Each branch has two conventional loops: + * - a fast-path loop for the most common codes in the mode + * - a loop for all other codes in the mode + * When the fast-path runs into a code that it cannot handle, its loop ends and it + * runs into the following loop to handle the other codes. + * The end of the input or output buffer is also handled by the slower loop. + * The slow loop jumps (goto) to the fast-path loop again as soon as possible. + * + * The callback handling is done by returning with an error code. + * The conversion framework actually calls the callback function. + */ + if(isSingleByteMode) { + /* fast path for single-byte mode */ + if(state==readCommand) { +fastSingle: + while(source=0x20) { + ++source; + ++nextSourceIndex; + if(b<=0x7f) { + /* write US-ASCII graphic character or DEL */ + *target++=(char16_t)b; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + } else { + /* write from dynamic window */ + uint32_t c=scsu->toUDynamicOffsets[dynamicWindow]+(b&0x7f); + if(c<=0xffff) { + *target++=(char16_t)c; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + } else { + /* output surrogate pair */ + *target++=(char16_t)(0xd7c0+(c>>10)); + if(targetUCharErrorBuffer[0]=(char16_t)(0xdc00|(c&0x3ff)); + cnv->UCharErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + goto endloop; + } + } + } + sourceIndex=nextSourceIndex; + } + } + + /* normal state machine for single-byte mode, minus handling for what fastSingle covers */ +singleByteMode: + while(source=targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + b=*source++; + ++nextSourceIndex; + switch(state) { + case readCommand: + /* redundant conditions are commented out */ + /* here: b<0x20 because otherwise we would be in fastSingle */ + if((1UL<toUBytes[0]=b; + cnv->toULength=1; + goto endloop; + } + + /* store the first byte of a multibyte sequence in toUBytes[] */ + cnv->toUBytes[0]=b; + cnv->toULength=1; + break; + case quotePairOne: + byteOne=b; + cnv->toUBytes[1]=b; + cnv->toULength=2; + state=quotePairTwo; + break; + case quotePairTwo: + *target++=(char16_t)((byteOne<<8)|b); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + sourceIndex=nextSourceIndex; + state=readCommand; + goto fastSingle; + case quoteOne: + if(b<0x80) { + /* all static offsets are in the BMP */ + *target++=(char16_t)(staticOffsets[quoteWindow]+b); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + } else { + /* write from dynamic window */ + uint32_t c=scsu->toUDynamicOffsets[quoteWindow]+(b&0x7f); + if(c<=0xffff) { + *target++=(char16_t)c; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + } else { + /* output surrogate pair */ + *target++=(char16_t)(0xd7c0+(c>>10)); + if(targetUCharErrorBuffer[0]=(char16_t)(0xdc00|(c&0x3ff)); + cnv->UCharErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + goto endloop; + } + } + } + sourceIndex=nextSourceIndex; + state=readCommand; + goto fastSingle; + case definePairOne: + dynamicWindow=(int8_t)((b>>5)&7); + byteOne=(uint8_t)(b&0x1f); + cnv->toUBytes[1]=b; + cnv->toULength=2; + state=definePairTwo; + break; + case definePairTwo: + scsu->toUDynamicOffsets[dynamicWindow]=0x10000+(byteOne<<15UL | b<<7UL); + sourceIndex=nextSourceIndex; + state=readCommand; + goto fastSingle; + case defineOne: + if(b==0) { + /* callback(illegal): Reserved window offset value 0 */ + cnv->toUBytes[1]=b; + cnv->toULength=2; + goto endloop; + } else if(btoUDynamicOffsets[dynamicWindow]=b<<7UL; + } else if((uint8_t)(b-gapThreshold)<(reservedStart-gapThreshold)) { + scsu->toUDynamicOffsets[dynamicWindow]=(b<<7UL)+gapOffset; + } else if(b>=fixedThreshold) { + scsu->toUDynamicOffsets[dynamicWindow]=fixedOffsets[b-fixedThreshold]; + } else { + /* callback(illegal): Reserved window offset value 0xa8..0xf8 */ + cnv->toUBytes[1]=b; + cnv->toULength=2; + goto endloop; + } + sourceIndex=nextSourceIndex; + state=readCommand; + goto fastSingle; + } + } + } else { + /* fast path for Unicode mode */ + if(state==readCommand) { +fastUnicode: + while(source+1(Urs-UC0)) { + *target++=(char16_t)((b<<8)|source[1]); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + sourceIndex=nextSourceIndex; + nextSourceIndex+=2; + source+=2; + } + } + + /* normal state machine for Unicode mode */ +/* unicodeByteMode: */ + while(source=targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + b=*source++; + ++nextSourceIndex; + switch(state) { + case readCommand: + if((uint8_t)(b-UC0)>(Urs-UC0)) { + byteOne=b; + cnv->toUBytes[0]=b; + cnv->toULength=1; + state=quotePairTwo; + } else if(/* UC0<=b && */ b<=UC7) { + dynamicWindow=(int8_t)(b-UC0); + sourceIndex=nextSourceIndex; + isSingleByteMode=true; + goto fastSingle; + } else if(/* UD0<=b && */ b<=UD7) { + dynamicWindow=(int8_t)(b-UD0); + isSingleByteMode=true; + cnv->toUBytes[0]=b; + cnv->toULength=1; + state=defineOne; + goto singleByteMode; + } else if(b==UDX) { + isSingleByteMode=true; + cnv->toUBytes[0]=b; + cnv->toULength=1; + state=definePairOne; + goto singleByteMode; + } else if(b==UQU) { + cnv->toUBytes[0]=b; + cnv->toULength=1; + state=quotePairOne; + } else /* Urs */ { + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + cnv->toUBytes[0]=b; + cnv->toULength=1; + goto endloop; + } + break; + case quotePairOne: + byteOne=b; + cnv->toUBytes[1]=b; + cnv->toULength=2; + state=quotePairTwo; + break; + case quotePairTwo: + *target++=(char16_t)((byteOne<<8)|b); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + sourceIndex=nextSourceIndex; + state=readCommand; + goto fastUnicode; + } + } + } +endloop: + + /* set the converter state back into UConverter */ + if(U_FAILURE(*pErrorCode) && *pErrorCode!=U_BUFFER_OVERFLOW_ERROR) { + /* reset to deal with the next character */ + state=readCommand; + } else if(state==readCommand) { + /* not in a multi-byte sequence, reset toULength */ + cnv->toULength=0; + } + scsu->toUIsSingleByteMode=isSingleByteMode; + scsu->toUState=state; + scsu->toUQuoteWindow=quoteWindow; + scsu->toUDynamicWindow=dynamicWindow; + scsu->toUByteOne=byteOne; + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + pArgs->offsets=offsets; + return; +} + +/* + * Identical to _SCSUToUnicodeWithOffsets but without offset handling. + * If a change is made in the original function, then either + * change this function the same way or + * re-copy the original function and remove the variables + * offsets, sourceIndex, and nextSourceIndex. + */ +static void U_CALLCONV +_SCSUToUnicode(UConverterToUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + SCSUData *scsu; + const uint8_t *source, *sourceLimit; + char16_t *target; + const char16_t *targetLimit; + UBool isSingleByteMode; + uint8_t state, byteOne; + int8_t quoteWindow, dynamicWindow; + + uint8_t b; + + /* set up the local pointers */ + cnv=pArgs->converter; + scsu=(SCSUData *)cnv->extraInfo; + + source=(const uint8_t *)pArgs->source; + sourceLimit=(const uint8_t *)pArgs->sourceLimit; + target=pArgs->target; + targetLimit=pArgs->targetLimit; + + /* get the state machine state */ + isSingleByteMode=scsu->toUIsSingleByteMode; + state=scsu->toUState; + quoteWindow=scsu->toUQuoteWindow; + dynamicWindow=scsu->toUDynamicWindow; + byteOne=scsu->toUByteOne; + + /* + * conversion "loop" + * + * For performance, this is not a normal C loop. + * Instead, there are two code blocks for the two SCSU modes. + * The function branches to either one, and a change of the mode is done with a goto to + * the other branch. + * + * Each branch has two conventional loops: + * - a fast-path loop for the most common codes in the mode + * - a loop for all other codes in the mode + * When the fast-path runs into a code that it cannot handle, its loop ends and it + * runs into the following loop to handle the other codes. + * The end of the input or output buffer is also handled by the slower loop. + * The slow loop jumps (goto) to the fast-path loop again as soon as possible. + * + * The callback handling is done by returning with an error code. + * The conversion framework actually calls the callback function. + */ + if(isSingleByteMode) { + /* fast path for single-byte mode */ + if(state==readCommand) { +fastSingle: + while(source=0x20) { + ++source; + if(b<=0x7f) { + /* write US-ASCII graphic character or DEL */ + *target++=(char16_t)b; + } else { + /* write from dynamic window */ + uint32_t c=scsu->toUDynamicOffsets[dynamicWindow]+(b&0x7f); + if(c<=0xffff) { + *target++=(char16_t)c; + } else { + /* output surrogate pair */ + *target++=(char16_t)(0xd7c0+(c>>10)); + if(targetUCharErrorBuffer[0]=(char16_t)(0xdc00|(c&0x3ff)); + cnv->UCharErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + goto endloop; + } + } + } + } + } + + /* normal state machine for single-byte mode, minus handling for what fastSingle covers */ +singleByteMode: + while(source=targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + b=*source++; + switch(state) { + case readCommand: + /* redundant conditions are commented out */ + /* here: b<0x20 because otherwise we would be in fastSingle */ + if((1UL<toUBytes[0]=b; + cnv->toULength=1; + goto endloop; + } + + /* store the first byte of a multibyte sequence in toUBytes[] */ + cnv->toUBytes[0]=b; + cnv->toULength=1; + break; + case quotePairOne: + byteOne=b; + cnv->toUBytes[1]=b; + cnv->toULength=2; + state=quotePairTwo; + break; + case quotePairTwo: + *target++=(char16_t)((byteOne<<8)|b); + state=readCommand; + goto fastSingle; + case quoteOne: + if(b<0x80) { + /* all static offsets are in the BMP */ + *target++=(char16_t)(staticOffsets[quoteWindow]+b); + } else { + /* write from dynamic window */ + uint32_t c=scsu->toUDynamicOffsets[quoteWindow]+(b&0x7f); + if(c<=0xffff) { + *target++=(char16_t)c; + } else { + /* output surrogate pair */ + *target++=(char16_t)(0xd7c0+(c>>10)); + if(targetUCharErrorBuffer[0]=(char16_t)(0xdc00|(c&0x3ff)); + cnv->UCharErrorBufferLength=1; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + goto endloop; + } + } + } + state=readCommand; + goto fastSingle; + case definePairOne: + dynamicWindow=(int8_t)((b>>5)&7); + byteOne=(uint8_t)(b&0x1f); + cnv->toUBytes[1]=b; + cnv->toULength=2; + state=definePairTwo; + break; + case definePairTwo: + scsu->toUDynamicOffsets[dynamicWindow]=0x10000+(byteOne<<15UL | b<<7UL); + state=readCommand; + goto fastSingle; + case defineOne: + if(b==0) { + /* callback(illegal): Reserved window offset value 0 */ + cnv->toUBytes[1]=b; + cnv->toULength=2; + goto endloop; + } else if(btoUDynamicOffsets[dynamicWindow]=b<<7UL; + } else if((uint8_t)(b-gapThreshold)<(reservedStart-gapThreshold)) { + scsu->toUDynamicOffsets[dynamicWindow]=(b<<7UL)+gapOffset; + } else if(b>=fixedThreshold) { + scsu->toUDynamicOffsets[dynamicWindow]=fixedOffsets[b-fixedThreshold]; + } else { + /* callback(illegal): Reserved window offset value 0xa8..0xf8 */ + cnv->toUBytes[1]=b; + cnv->toULength=2; + goto endloop; + } + state=readCommand; + goto fastSingle; + } + } + } else { + /* fast path for Unicode mode */ + if(state==readCommand) { +fastUnicode: + while(source+1(Urs-UC0)) { + *target++=(char16_t)((b<<8)|source[1]); + source+=2; + } + } + + /* normal state machine for Unicode mode */ +/* unicodeByteMode: */ + while(source=targetLimit) { + /* target is full */ + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + break; + } + b=*source++; + switch(state) { + case readCommand: + if((uint8_t)(b-UC0)>(Urs-UC0)) { + byteOne=b; + cnv->toUBytes[0]=b; + cnv->toULength=1; + state=quotePairTwo; + } else if(/* UC0<=b && */ b<=UC7) { + dynamicWindow=(int8_t)(b-UC0); + isSingleByteMode=true; + goto fastSingle; + } else if(/* UD0<=b && */ b<=UD7) { + dynamicWindow=(int8_t)(b-UD0); + isSingleByteMode=true; + cnv->toUBytes[0]=b; + cnv->toULength=1; + state=defineOne; + goto singleByteMode; + } else if(b==UDX) { + isSingleByteMode=true; + cnv->toUBytes[0]=b; + cnv->toULength=1; + state=definePairOne; + goto singleByteMode; + } else if(b==UQU) { + cnv->toUBytes[0]=b; + cnv->toULength=1; + state=quotePairOne; + } else /* Urs */ { + /* callback(illegal) */ + *pErrorCode=U_ILLEGAL_CHAR_FOUND; + cnv->toUBytes[0]=b; + cnv->toULength=1; + goto endloop; + } + break; + case quotePairOne: + byteOne=b; + cnv->toUBytes[1]=b; + cnv->toULength=2; + state=quotePairTwo; + break; + case quotePairTwo: + *target++=(char16_t)((byteOne<<8)|b); + state=readCommand; + goto fastUnicode; + } + } + } +endloop: + + /* set the converter state back into UConverter */ + if(U_FAILURE(*pErrorCode) && *pErrorCode!=U_BUFFER_OVERFLOW_ERROR) { + /* reset to deal with the next character */ + state=readCommand; + } else if(state==readCommand) { + /* not in a multi-byte sequence, reset toULength */ + cnv->toULength=0; + } + scsu->toUIsSingleByteMode=isSingleByteMode; + scsu->toUState=state; + scsu->toUQuoteWindow=quoteWindow; + scsu->toUDynamicWindow=dynamicWindow; + scsu->toUByteOne=byteOne; + + /* write back the updated pointers */ + pArgs->source=(const char *)source; + pArgs->target=target; + return; +} +U_CDECL_END +/* SCSU-from-Unicode conversion functions ----------------------------------- */ + +/* + * This SCSU Encoder is fairly simple but uses all SCSU commands to achieve + * reasonable results. The lookahead is minimal. + * Many cases are simple: + * A character fits directly into the current mode, a dynamic or static window, + * or is not compressible. These cases are tested first. + * Real compression heuristics are applied to the rest, in code branches for + * single/Unicode mode and BMP/supplementary code points. + * The heuristics used here are extremely simple. + */ + +/* get the number of the window that this character is in, or -1 */ +static int8_t +getWindow(const uint32_t offsets[8], uint32_t c) { + int i; + for(i=0; i<8; ++i) { + if((uint32_t)(c-offsets[i])<=0x7f) { + return (int8_t)(i); + } + } + return -1; +} + +/* is the character in the dynamic window starting at the offset, or in the direct-encoded range? */ +static UBool +isInOffsetWindowOrDirect(uint32_t offset, uint32_t c) { + return (UBool)(c<=offset+0x7f && + (c>=offset || (c<=0x7f && + (c>=0x20 || (1UL<windowUse[scsu->nextWindowUseIndex]; + if(++scsu->nextWindowUseIndex==8) { + scsu->nextWindowUseIndex=0; + } + return window; +} + +/* + * useDynamicWindow() adjusts + * windowUse[] and nextWindowUseIndex for the algorithm to choose + * the next dynamic window to be defined; + * a subclass may override it and provide its own algorithm. + */ +static void +useDynamicWindow(SCSUData *scsu, int8_t window) { + /* + * move the existing window, which just became the most recently used one, + * up in windowUse[] to nextWindowUseIndex-1 + */ + + /* first, find the index of the window - backwards to favor the more recently used windows */ + int i, j; + + i=scsu->nextWindowUseIndex; + do { + if(--i<0) { + i=7; + } + } while(scsu->windowUse[i]!=window); + + /* now copy each windowUse[i+1] to [i] */ + j=i+1; + if(j==8) { + j=0; + } + while(j!=scsu->nextWindowUseIndex) { + scsu->windowUse[i]=scsu->windowUse[j]; + i=j; + if(++j==8) { j=0; } + } + + /* finally, set the window into the most recently used index */ + scsu->windowUse[i]=window; +} + +/* + * calculate the offset and the code for a dynamic window that contains the character + * takes fixed offsets into account + * the offset of the window is stored in the offset variable, + * the code is returned + * + * return offset code: -1 none <=0xff code for SDn/UDn else code for SDX/UDX, subtract 0x200 to get the true code + */ +static int +getDynamicOffset(uint32_t c, uint32_t *pOffset) { + int i; + + for(i=0; i<7; ++i) { + if((uint32_t)(c-fixedOffsets[i])<=0x7f) { + *pOffset=fixedOffsets[i]; + return 0xf9+i; + } + } + + if(c<0x80) { + /* No dynamic window for US-ASCII. */ + return -1; + } else if(c<0x3400 || + (uint32_t)(c-0x10000)<(0x14000-0x10000) || + (uint32_t)(c-0x1d000)<=(0x1ffff-0x1d000) + ) { + /* This character is in a code range for a "small", i.e., reasonably windowable, script. */ + *pOffset=c&0x7fffff80; + return (int)(c>>7); + } else if(0xe000<=c && c!=0xfeff && c<0xfff0) { + /* For these characters we need to take the gapOffset into account. */ + *pOffset=c&0x7fffff80; + return (int)((c-gapOffset)>>7); + } else { + return -1; + } +} +U_CDECL_BEGIN +/* + * Idea for compression: + * - save SCSUData and other state before really starting work + * - at endloop, see if compression could be better with just unicode mode + * - don't do this if a callback has been called + * - if unicode mode would be smaller, then override the results with it - may need SCU at the beginning + * - different buffer handling! + * + * Drawback or need for corrective handling: + * it is desirable to encode U+feff as SQU fe ff for the SCSU signature, and + * it is desirable to start a document in US-ASCII/Latin-1 for as long as possible + * not only for compression but also for HTML/XML documents with following charset/encoding announcers. + * + * How to achieve both? + * - Only replace the result after an SDX or SCU? + */ + +static void U_CALLCONV +_SCSUFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + SCSUData *scsu; + const char16_t *source, *sourceLimit; + uint8_t *target; + int32_t targetCapacity; + int32_t *offsets; + + UBool isSingleByteMode; + uint8_t dynamicWindow; + uint32_t currentOffset; + + uint32_t c, delta; + + int32_t sourceIndex, nextSourceIndex; + + int32_t length; + + /* variables for compression heuristics */ + uint32_t offset; + char16_t lead, trail; + int code; + int8_t window; + + /* set up the local pointers */ + cnv=pArgs->converter; + scsu=(SCSUData *)cnv->extraInfo; + + /* set up the local pointers */ + source=pArgs->source; + sourceLimit=pArgs->sourceLimit; + target=(uint8_t *)pArgs->target; + targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); + offsets=pArgs->offsets; + + /* get the state machine state */ + isSingleByteMode=scsu->fromUIsSingleByteMode; + dynamicWindow=scsu->fromUDynamicWindow; + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]; + + c=cnv->fromUChar32; + + /* sourceIndex=-1 if the current character began in the previous buffer */ + sourceIndex= c==0 ? 0 : -1; + nextSourceIndex=0; + + /* similar conversion "loop" as in toUnicode */ +loop: + if(isSingleByteMode) { + if(c!=0 && targetCapacity>0) { + goto getTrailSingle; + } + + /* state machine for single-byte mode */ +/* singleByteMode: */ + while(sourcefromUDynamicOffsets, c))>=0) { + /* there is a dynamic window that contains this character, change to it */ + dynamicWindow=window; + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)(SC0+dynamicWindow)<<8)|(c-currentOffset)|0x80; + length=2; + goto outputBytes; + } else if((code=getDynamicOffset(c, &offset))>=0) { + /* might check if there are more characters in this window to come */ + /* define an extended window with this character */ + code-=0x200; + dynamicWindow=getNextDynamicWindow(scsu); + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)SDX<<24)|((uint32_t)dynamicWindow<<21)|((uint32_t)code<<8)|(c-currentOffset)|0x80; + length=4; + goto outputBytes; + } else { + /* change to Unicode mode and output this (lead, trail) pair */ + isSingleByteMode=false; + *target++=(uint8_t)SCU; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + --targetCapacity; + c=((uint32_t)lead<<16)|trail; + length=4; + goto outputBytes; + } + } else if(c<0xa0) { + /* quote C1 control character */ + c=(c&0x7f)|(SQ0+1)<<8; /* SQ0+1==SQ1 */ + length=2; + goto outputBytes; + } else if(c==0xfeff || c>=0xfff0) { + /* quote signature character=byte order mark and specials */ + c|=SQU<<16; + length=3; + goto outputBytes; + } else { + /* compress all other BMP characters */ + if((window=getWindow(scsu->fromUDynamicOffsets, c))>=0) { + /* there is a window defined that contains this character - switch to it or quote from it? */ + if(source>=sourceLimit || isInOffsetWindowOrDirect(scsu->fromUDynamicOffsets[window], *source)) { + /* change to dynamic window */ + dynamicWindow=window; + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)(SC0+dynamicWindow)<<8)|(c-currentOffset)|0x80; + length=2; + goto outputBytes; + } else { + /* quote from dynamic window */ + c=((uint32_t)(SQ0+window)<<8)|(c-scsu->fromUDynamicOffsets[window])|0x80; + length=2; + goto outputBytes; + } + } else if((window=getWindow(staticOffsets, c))>=0) { + /* quote from static window */ + c=((uint32_t)(SQ0+window)<<8)|(c-staticOffsets[window]); + length=2; + goto outputBytes; + } else if((code=getDynamicOffset(c, &offset))>=0) { + /* define a dynamic window with this character */ + dynamicWindow=getNextDynamicWindow(scsu); + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)(SD0+dynamicWindow)<<16)|((uint32_t)code<<8)|(c-currentOffset)|0x80; + length=3; + goto outputBytes; + } else if((uint32_t)(c-0x3400)<(0xd800-0x3400) && + (source>=sourceLimit || (uint32_t)(*source-0x3400)<(0xd800-0x3400)) + ) { + /* + * this character is not compressible (a BMP ideograph or similar); + * switch to Unicode mode if this is the last character in the block + * or there is at least one more ideograph following immediately + */ + isSingleByteMode=false; + c|=SCU<<16; + length=3; + goto outputBytes; + } else { + /* quote Unicode */ + c|=SQU<<16; + length=3; + goto outputBytes; + } + } + + /* normal end of conversion: prepare for a new character */ + c=0; + sourceIndex=nextSourceIndex; + } + } else { + if(c!=0 && targetCapacity>0) { + goto getTrailUnicode; + } + + /* state machine for Unicode mode */ +/* unicodeByteMode: */ + while(source=2) { + *target++=(uint8_t)(c>>8); + *target++=(uint8_t)c; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + *offsets++=sourceIndex; + } + targetCapacity-=2; + } else { + length=2; + goto outputBytes; + } + } else if((uint32_t)(c-0x3400)>=(0xf300-0x3400) /* c<0x3400 || c>=0xf300 */) { + /* compress BMP character if the following one is not an uncompressible ideograph */ + if(!(sourcefromUDynamicOffsets, c))>=0) { + /* there is a dynamic window that contains this character, change to it */ + isSingleByteMode=true; + dynamicWindow=window; + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)(UC0+dynamicWindow)<<8)|(c-currentOffset)|0x80; + length=2; + goto outputBytes; + } else if((code=getDynamicOffset(c, &offset))>=0) { + /* define a dynamic window with this character */ + isSingleByteMode=true; + dynamicWindow=getNextDynamicWindow(scsu); + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)(UD0+dynamicWindow)<<16)|((uint32_t)code<<8)|(c-currentOffset)|0x80; + length=3; + goto outputBytes; + } + } + + /* don't know how to compress this character, just write it directly */ + length=2; + goto outputBytes; + } else if(c<0xe000) { + /* c is a surrogate */ + if(U16_IS_SURROGATE_LEAD(c)) { +getTrailUnicode: + lead=(char16_t)c; + if(sourcefromUDynamicOffsets, c))>=0 && + !(sourcefromUDynamicOffsets[dynamicWindow]; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)(UC0+dynamicWindow)<<8)|(c-currentOffset)|0x80; + length=2; + goto outputBytes; + } else if(source=0 + ) { + /* two supplementary characters in (probably) the same window - define an extended one */ + isSingleByteMode=true; + code-=0x200; + dynamicWindow=getNextDynamicWindow(scsu); + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)UDX<<24)|((uint32_t)dynamicWindow<<21)|((uint32_t)code<<8)|(c-currentOffset)|0x80; + length=4; + goto outputBytes; + } else { + /* don't know how to compress this character, just write it directly */ + c=((uint32_t)lead<<16)|trail; + length=4; + goto outputBytes; + } + } else /* 0xe000<=c<0xf300 */ { + /* quote to avoid SCSU tags */ + c|=UQU<<16; + length=3; + goto outputBytes; + } + + /* normal end of conversion: prepare for a new character */ + c=0; + sourceIndex=nextSourceIndex; + } + } +endloop: + + /* set the converter state back into UConverter */ + scsu->fromUIsSingleByteMode=isSingleByteMode; + scsu->fromUDynamicWindow=dynamicWindow; + + cnv->fromUChar32=c; + + /* write back the updated pointers */ + pArgs->source=source; + pArgs->target=(char *)target; + pArgs->offsets=offsets; + return; + +outputBytes: + /* write the output character bytes from c and length [code copied from ucnvmbcs.c] */ + /* from the first if in the loop we know that targetCapacity>0 */ + if(length<=targetCapacity) { + if(offsets==nullptr) { + switch(length) { + /* each branch falls through to the next one */ + case 4: + *target++=(uint8_t)(c>>24); + U_FALLTHROUGH; + case 3: + *target++=(uint8_t)(c>>16); + U_FALLTHROUGH; + case 2: + *target++=(uint8_t)(c>>8); + U_FALLTHROUGH; + case 1: + *target++=(uint8_t)c; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + } else { + switch(length) { + /* each branch falls through to the next one */ + case 4: + *target++=(uint8_t)(c>>24); + *offsets++=sourceIndex; + U_FALLTHROUGH; + case 3: + *target++=(uint8_t)(c>>16); + *offsets++=sourceIndex; + U_FALLTHROUGH; + case 2: + *target++=(uint8_t)(c>>8); + *offsets++=sourceIndex; + U_FALLTHROUGH; + case 1: + *target++=(uint8_t)c; + *offsets++=sourceIndex; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + } + targetCapacity-=length; + + /* normal end of conversion: prepare for a new character */ + c=0; + sourceIndex=nextSourceIndex; + goto loop; + } else { + uint8_t *p; + + /* + * We actually do this backwards here: + * In order to save an intermediate variable, we output + * first to the overflow buffer what does not fit into the + * regular target. + */ + /* we know that 0<=targetCapacitycharErrorBuffer; + switch(length) { + /* each branch falls through to the next one */ + case 4: + *p++=(uint8_t)(c>>24); + U_FALLTHROUGH; + case 3: + *p++=(uint8_t)(c>>16); + U_FALLTHROUGH; + case 2: + *p++=(uint8_t)(c>>8); + U_FALLTHROUGH; + case 1: + *p=(uint8_t)c; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + cnv->charErrorBufferLength=(int8_t)length; + + /* now output what fits into the regular target */ + c>>=8*length; /* length was reduced by targetCapacity */ + switch(targetCapacity) { + /* each branch falls through to the next one */ + case 3: + *target++=(uint8_t)(c>>16); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + U_FALLTHROUGH; + case 2: + *target++=(uint8_t)(c>>8); + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + U_FALLTHROUGH; + case 1: + *target++=(uint8_t)c; + if(offsets!=nullptr) { + *offsets++=sourceIndex; + } + U_FALLTHROUGH; + default: + break; + } + + /* target overflow */ + targetCapacity=0; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + c=0; + goto endloop; + } +} + +/* + * Identical to _SCSUFromUnicodeWithOffsets but without offset handling. + * If a change is made in the original function, then either + * change this function the same way or + * re-copy the original function and remove the variables + * offsets, sourceIndex, and nextSourceIndex. + */ +static void U_CALLCONV +_SCSUFromUnicode(UConverterFromUnicodeArgs *pArgs, + UErrorCode *pErrorCode) { + UConverter *cnv; + SCSUData *scsu; + const char16_t *source, *sourceLimit; + uint8_t *target; + int32_t targetCapacity; + + UBool isSingleByteMode; + uint8_t dynamicWindow; + uint32_t currentOffset; + + uint32_t c, delta; + + int32_t length; + + /* variables for compression heuristics */ + uint32_t offset; + char16_t lead, trail; + int code; + int8_t window; + + /* set up the local pointers */ + cnv=pArgs->converter; + scsu=(SCSUData *)cnv->extraInfo; + + /* set up the local pointers */ + source=pArgs->source; + sourceLimit=pArgs->sourceLimit; + target=(uint8_t *)pArgs->target; + targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target); + + /* get the state machine state */ + isSingleByteMode=scsu->fromUIsSingleByteMode; + dynamicWindow=scsu->fromUDynamicWindow; + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]; + + c=cnv->fromUChar32; + + /* similar conversion "loop" as in toUnicode */ +loop: + if(isSingleByteMode) { + if(c!=0 && targetCapacity>0) { + goto getTrailSingle; + } + + /* state machine for single-byte mode */ +/* singleByteMode: */ + while(sourcefromUDynamicOffsets, c))>=0) { + /* there is a dynamic window that contains this character, change to it */ + dynamicWindow=window; + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)(SC0+dynamicWindow)<<8)|(c-currentOffset)|0x80; + length=2; + goto outputBytes; + } else if((code=getDynamicOffset(c, &offset))>=0) { + /* might check if there are more characters in this window to come */ + /* define an extended window with this character */ + code-=0x200; + dynamicWindow=getNextDynamicWindow(scsu); + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)SDX<<24)|((uint32_t)dynamicWindow<<21)|((uint32_t)code<<8)|(c-currentOffset)|0x80; + length=4; + goto outputBytes; + } else { + /* change to Unicode mode and output this (lead, trail) pair */ + isSingleByteMode=false; + *target++=(uint8_t)SCU; + --targetCapacity; + c=((uint32_t)lead<<16)|trail; + length=4; + goto outputBytes; + } + } else if(c<0xa0) { + /* quote C1 control character */ + c=(c&0x7f)|(SQ0+1)<<8; /* SQ0+1==SQ1 */ + length=2; + goto outputBytes; + } else if(c==0xfeff || c>=0xfff0) { + /* quote signature character=byte order mark and specials */ + c|=SQU<<16; + length=3; + goto outputBytes; + } else { + /* compress all other BMP characters */ + if((window=getWindow(scsu->fromUDynamicOffsets, c))>=0) { + /* there is a window defined that contains this character - switch to it or quote from it? */ + if(source>=sourceLimit || isInOffsetWindowOrDirect(scsu->fromUDynamicOffsets[window], *source)) { + /* change to dynamic window */ + dynamicWindow=window; + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)(SC0+dynamicWindow)<<8)|(c-currentOffset)|0x80; + length=2; + goto outputBytes; + } else { + /* quote from dynamic window */ + c=((uint32_t)(SQ0+window)<<8)|(c-scsu->fromUDynamicOffsets[window])|0x80; + length=2; + goto outputBytes; + } + } else if((window=getWindow(staticOffsets, c))>=0) { + /* quote from static window */ + c=((uint32_t)(SQ0+window)<<8)|(c-staticOffsets[window]); + length=2; + goto outputBytes; + } else if((code=getDynamicOffset(c, &offset))>=0) { + /* define a dynamic window with this character */ + dynamicWindow=getNextDynamicWindow(scsu); + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)(SD0+dynamicWindow)<<16)|((uint32_t)code<<8)|(c-currentOffset)|0x80; + length=3; + goto outputBytes; + } else if((uint32_t)(c-0x3400)<(0xd800-0x3400) && + (source>=sourceLimit || (uint32_t)(*source-0x3400)<(0xd800-0x3400)) + ) { + /* + * this character is not compressible (a BMP ideograph or similar); + * switch to Unicode mode if this is the last character in the block + * or there is at least one more ideograph following immediately + */ + isSingleByteMode=false; + c|=SCU<<16; + length=3; + goto outputBytes; + } else { + /* quote Unicode */ + c|=SQU<<16; + length=3; + goto outputBytes; + } + } + + /* normal end of conversion: prepare for a new character */ + c=0; + } + } else { + if(c!=0 && targetCapacity>0) { + goto getTrailUnicode; + } + + /* state machine for Unicode mode */ +/* unicodeByteMode: */ + while(source=2) { + *target++=(uint8_t)(c>>8); + *target++=(uint8_t)c; + targetCapacity-=2; + } else { + length=2; + goto outputBytes; + } + } else if((uint32_t)(c-0x3400)>=(0xf300-0x3400) /* c<0x3400 || c>=0xf300 */) { + /* compress BMP character if the following one is not an uncompressible ideograph */ + if(!(sourcefromUDynamicOffsets, c))>=0) { + /* there is a dynamic window that contains this character, change to it */ + isSingleByteMode=true; + dynamicWindow=window; + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)(UC0+dynamicWindow)<<8)|(c-currentOffset)|0x80; + length=2; + goto outputBytes; + } else if((code=getDynamicOffset(c, &offset))>=0) { + /* define a dynamic window with this character */ + isSingleByteMode=true; + dynamicWindow=getNextDynamicWindow(scsu); + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)(UD0+dynamicWindow)<<16)|((uint32_t)code<<8)|(c-currentOffset)|0x80; + length=3; + goto outputBytes; + } + } + + /* don't know how to compress this character, just write it directly */ + length=2; + goto outputBytes; + } else if(c<0xe000) { + /* c is a surrogate */ + if(U16_IS_SURROGATE_LEAD(c)) { +getTrailUnicode: + lead=(char16_t)c; + if(sourcefromUDynamicOffsets, c))>=0 && + !(sourcefromUDynamicOffsets[dynamicWindow]; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)(UC0+dynamicWindow)<<8)|(c-currentOffset)|0x80; + length=2; + goto outputBytes; + } else if(source=0 + ) { + /* two supplementary characters in (probably) the same window - define an extended one */ + isSingleByteMode=true; + code-=0x200; + dynamicWindow=getNextDynamicWindow(scsu); + currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset; + useDynamicWindow(scsu, dynamicWindow); + c=((uint32_t)UDX<<24)|((uint32_t)dynamicWindow<<21)|((uint32_t)code<<8)|(c-currentOffset)|0x80; + length=4; + goto outputBytes; + } else { + /* don't know how to compress this character, just write it directly */ + c=((uint32_t)lead<<16)|trail; + length=4; + goto outputBytes; + } + } else /* 0xe000<=c<0xf300 */ { + /* quote to avoid SCSU tags */ + c|=UQU<<16; + length=3; + goto outputBytes; + } + + /* normal end of conversion: prepare for a new character */ + c=0; + } + } +endloop: + + /* set the converter state back into UConverter */ + scsu->fromUIsSingleByteMode=isSingleByteMode; + scsu->fromUDynamicWindow=dynamicWindow; + + cnv->fromUChar32=c; + + /* write back the updated pointers */ + pArgs->source=source; + pArgs->target=(char *)target; + return; + +outputBytes: + /* write the output character bytes from c and length [code copied from ucnvmbcs.c] */ + /* from the first if in the loop we know that targetCapacity>0 */ + if(length<=targetCapacity) { + switch(length) { + /* each branch falls through to the next one */ + case 4: + *target++=(uint8_t)(c>>24); + U_FALLTHROUGH; + case 3: + *target++=(uint8_t)(c>>16); + U_FALLTHROUGH; + case 2: + *target++=(uint8_t)(c>>8); + U_FALLTHROUGH; + case 1: + *target++=(uint8_t)c; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + targetCapacity-=length; + + /* normal end of conversion: prepare for a new character */ + c=0; + goto loop; + } else { + uint8_t *p; + + /* + * We actually do this backwards here: + * In order to save an intermediate variable, we output + * first to the overflow buffer what does not fit into the + * regular target. + */ + /* we know that 0<=targetCapacitycharErrorBuffer; + switch(length) { + /* each branch falls through to the next one */ + case 4: + *p++=(uint8_t)(c>>24); + U_FALLTHROUGH; + case 3: + *p++=(uint8_t)(c>>16); + U_FALLTHROUGH; + case 2: + *p++=(uint8_t)(c>>8); + U_FALLTHROUGH; + case 1: + *p=(uint8_t)c; + U_FALLTHROUGH; + default: + /* will never occur */ + break; + } + cnv->charErrorBufferLength=(int8_t)length; + + /* now output what fits into the regular target */ + c = (length == 4) ? 0 : c >> 8*length; /* length was reduced by targetCapacity */ + switch(targetCapacity) { + /* each branch falls through to the next one */ + case 3: + *target++=(uint8_t)(c>>16); + U_FALLTHROUGH; + case 2: + *target++=(uint8_t)(c>>8); + U_FALLTHROUGH; + case 1: + *target++=(uint8_t)c; + U_FALLTHROUGH; + default: + break; + } + + /* target overflow */ + targetCapacity=0; + *pErrorCode=U_BUFFER_OVERFLOW_ERROR; + c=0; + goto endloop; + } +} + +/* miscellaneous ------------------------------------------------------------ */ + +static const char * U_CALLCONV +_SCSUGetName(const UConverter *cnv) { + SCSUData *scsu=(SCSUData *)cnv->extraInfo; + + switch(scsu->locale) { + case l_ja: + return "SCSU,locale=ja"; + default: + return "SCSU"; + } +} + +/* structure for SafeClone calculations */ +struct cloneSCSUStruct +{ + UConverter cnv; + SCSUData mydata; +}; + +static UConverter * U_CALLCONV +_SCSUSafeClone(const UConverter *cnv, + void *stackBuffer, + int32_t *pBufferSize, + UErrorCode *status) +{ + struct cloneSCSUStruct * localClone; + int32_t bufferSizeNeeded = sizeof(struct cloneSCSUStruct); + + if (U_FAILURE(*status)){ + return 0; + } + + if (*pBufferSize == 0){ /* 'preflighting' request - set needed size into *pBufferSize */ + *pBufferSize = bufferSizeNeeded; + return 0; + } + + localClone = (struct cloneSCSUStruct *)stackBuffer; + /* ucnv.c/ucnv_safeClone() copied the main UConverter already */ + + uprv_memcpy(&localClone->mydata, cnv->extraInfo, sizeof(SCSUData)); + localClone->cnv.extraInfo = &localClone->mydata; + localClone->cnv.isExtraLocal = true; + + return &localClone->cnv; +} +U_CDECL_END + +static const UConverterImpl _SCSUImpl={ + UCNV_SCSU, + + nullptr, + nullptr, + + _SCSUOpen, + _SCSUClose, + _SCSUReset, + + _SCSUToUnicode, + _SCSUToUnicodeWithOffsets, + _SCSUFromUnicode, + _SCSUFromUnicodeWithOffsets, + nullptr, + + nullptr, + _SCSUGetName, + nullptr, + _SCSUSafeClone, + ucnv_getCompleteUnicodeSet, + nullptr, + nullptr +}; + +static const UConverterStaticData _SCSUStaticData={ + sizeof(UConverterStaticData), + "SCSU", + 1212, /* CCSID for SCSU */ + UCNV_IBM, UCNV_SCSU, + 1, 3, /* one char16_t generates at least 1 byte and at most 3 bytes */ + /* + * The subchar here is ignored because _SCSUOpen() sets U+fffd as a Unicode + * substitution string. + */ + { 0x0e, 0xff, 0xfd, 0 }, 3, + false, false, + 0, + 0, + { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ +}; + +const UConverterSharedData _SCSUData= + UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_SCSUStaticData, &_SCSUImpl); + +#endif diff --git a/intl/icu/source/common/ucnvsel.cpp b/intl/icu/source/common/ucnvsel.cpp new file mode 100644 index 0000000000..a286646395 --- /dev/null +++ b/intl/icu/source/common/ucnvsel.cpp @@ -0,0 +1,823 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2008-2011, International Business Machines +* Corporation, Google and others. All Rights Reserved. +* +******************************************************************************* +*/ +// Author : eldawy@google.com (Mohamed Eldawy) +// ucnvsel.cpp +// +// Purpose: To generate a list of encodings capable of handling +// a given Unicode text +// +// Started 09-April-2008 + +/** + * \file + * + * This is an implementation of an encoding selector. + * The goal is, given a unicode string, find the encodings + * this string can be mapped to. To make processing faster + * a trie is built when you call ucnvsel_open() that + * stores all encodings a codepoint can map to + */ + +#include "unicode/ucnvsel.h" + +#if !UCONFIG_NO_CONVERSION + +#include + +#include "unicode/uchar.h" +#include "unicode/uniset.h" +#include "unicode/ucnv.h" +#include "unicode/ustring.h" +#include "unicode/uchriter.h" +#include "utrie2.h" +#include "propsvec.h" +#include "uassert.h" +#include "ucmndata.h" +#include "udataswp.h" +#include "uenumimp.h" +#include "cmemory.h" +#include "cstring.h" + +U_NAMESPACE_USE + +struct UConverterSelector { + UTrie2 *trie; // 16 bit trie containing offsets into pv + uint32_t* pv; // table of bits! + int32_t pvCount; + char** encodings; // which encodings did user ask to use? + int32_t encodingsCount; + int32_t encodingStrLength; + uint8_t* swapped; + UBool ownPv, ownEncodingStrings; +}; + +static void generateSelectorData(UConverterSelector* result, + UPropsVectors *upvec, + const USet* excludedCodePoints, + const UConverterUnicodeSet whichSet, + UErrorCode* status) { + if (U_FAILURE(*status)) { + return; + } + + int32_t columns = (result->encodingsCount+31)/32; + + // set errorValue to all-ones + for (int32_t col = 0; col < columns; col++) { + upvec_setValue(upvec, UPVEC_ERROR_VALUE_CP, UPVEC_ERROR_VALUE_CP, + col, static_cast(~0), static_cast(~0), status); + } + + for (int32_t i = 0; i < result->encodingsCount; ++i) { + uint32_t mask; + uint32_t column; + int32_t item_count; + int32_t j; + UConverter* test_converter = ucnv_open(result->encodings[i], status); + if (U_FAILURE(*status)) { + return; + } + USet* unicode_point_set; + unicode_point_set = uset_open(1, 0); // empty set + + ucnv_getUnicodeSet(test_converter, unicode_point_set, + whichSet, status); + if (U_FAILURE(*status)) { + ucnv_close(test_converter); + return; + } + + column = i / 32; + mask = 1 << (i%32); + // now iterate over intervals on set i! + item_count = uset_getItemCount(unicode_point_set); + + for (j = 0; j < item_count; ++j) { + UChar32 start_char; + UChar32 end_char; + UErrorCode smallStatus = U_ZERO_ERROR; + uset_getItem(unicode_point_set, j, &start_char, &end_char, nullptr, 0, + &smallStatus); + if (U_FAILURE(smallStatus)) { + // this will be reached for the converters that fill the set with + // strings. Those should be ignored by our system + } else { + upvec_setValue(upvec, start_char, end_char, column, static_cast(~0), mask, + status); + } + } + ucnv_close(test_converter); + uset_close(unicode_point_set); + if (U_FAILURE(*status)) { + return; + } + } + + // handle excluded encodings! Simply set their values to all 1's in the upvec + if (excludedCodePoints) { + int32_t item_count = uset_getItemCount(excludedCodePoints); + for (int32_t j = 0; j < item_count; ++j) { + UChar32 start_char; + UChar32 end_char; + + uset_getItem(excludedCodePoints, j, &start_char, &end_char, nullptr, 0, + status); + for (int32_t col = 0; col < columns; col++) { + upvec_setValue(upvec, start_char, end_char, col, static_cast(~0), static_cast(~0), + status); + } + } + } + + // alright. Now, let's put things in the same exact form you'd get when you + // unserialize things. + result->trie = upvec_compactToUTrie2WithRowIndexes(upvec, status); + result->pv = upvec_cloneArray(upvec, &result->pvCount, nullptr, status); + result->pvCount *= columns; // number of uint32_t = rows * columns + result->ownPv = true; +} + +/* open a selector. If converterListSize is 0, build for all converters. + If excludedCodePoints is nullptr, don't exclude any codepoints */ +U_CAPI UConverterSelector* U_EXPORT2 +ucnvsel_open(const char* const* converterList, int32_t converterListSize, + const USet* excludedCodePoints, + const UConverterUnicodeSet whichSet, UErrorCode* status) { + // check if already failed + if (U_FAILURE(*status)) { + return nullptr; + } + // ensure args make sense! + if (converterListSize < 0 || (converterList == nullptr && converterListSize != 0)) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + + // allocate a new converter + LocalUConverterSelectorPointer newSelector( + (UConverterSelector*)uprv_malloc(sizeof(UConverterSelector))); + if (newSelector.isNull()) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memset(newSelector.getAlias(), 0, sizeof(UConverterSelector)); + + if (converterListSize == 0) { + converterList = nullptr; + converterListSize = ucnv_countAvailable(); + } + newSelector->encodings = + (char**)uprv_malloc(converterListSize * sizeof(char*)); + if (!newSelector->encodings) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + newSelector->encodings[0] = nullptr; // now we can call ucnvsel_close() + + // make a backup copy of the list of converters + int32_t totalSize = 0; + int32_t i; + for (i = 0; i < converterListSize; i++) { + totalSize += + (int32_t)uprv_strlen(converterList != nullptr ? converterList[i] : ucnv_getAvailableName(i)) + 1; + } + // 4-align the totalSize to 4-align the size of the serialized form + int32_t encodingStrPadding = totalSize & 3; + if (encodingStrPadding != 0) { + encodingStrPadding = 4 - encodingStrPadding; + } + newSelector->encodingStrLength = totalSize += encodingStrPadding; + char* allStrings = (char*) uprv_malloc(totalSize); + if (!allStrings) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + + for (i = 0; i < converterListSize; i++) { + newSelector->encodings[i] = allStrings; + uprv_strcpy(newSelector->encodings[i], + converterList != nullptr ? converterList[i] : ucnv_getAvailableName(i)); + allStrings += uprv_strlen(newSelector->encodings[i]) + 1; + } + while (encodingStrPadding > 0) { + *allStrings++ = 0; + --encodingStrPadding; + } + + newSelector->ownEncodingStrings = true; + newSelector->encodingsCount = converterListSize; + UPropsVectors *upvec = upvec_open((converterListSize+31)/32, status); + generateSelectorData(newSelector.getAlias(), upvec, excludedCodePoints, whichSet, status); + upvec_close(upvec); + + if (U_FAILURE(*status)) { + return nullptr; + } + + return newSelector.orphan(); +} + +/* close opened selector */ +U_CAPI void U_EXPORT2 +ucnvsel_close(UConverterSelector *sel) { + if (!sel) { + return; + } + if (sel->ownEncodingStrings) { + uprv_free(sel->encodings[0]); + } + uprv_free(sel->encodings); + if (sel->ownPv) { + uprv_free(sel->pv); + } + utrie2_close(sel->trie); + uprv_free(sel->swapped); + uprv_free(sel); +} + +static const UDataInfo dataInfo = { + sizeof(UDataInfo), + 0, + + U_IS_BIG_ENDIAN, + U_CHARSET_FAMILY, + U_SIZEOF_UCHAR, + 0, + + { 0x43, 0x53, 0x65, 0x6c }, /* dataFormat="CSel" */ + { 1, 0, 0, 0 }, /* formatVersion */ + { 0, 0, 0, 0 } /* dataVersion */ +}; + +enum { + UCNVSEL_INDEX_TRIE_SIZE, // trie size in bytes + UCNVSEL_INDEX_PV_COUNT, // number of uint32_t in the bit vectors + UCNVSEL_INDEX_NAMES_COUNT, // number of encoding names + UCNVSEL_INDEX_NAMES_LENGTH, // number of encoding name bytes including padding + UCNVSEL_INDEX_SIZE = 15, // bytes following the DataHeader + UCNVSEL_INDEX_COUNT = 16 +}; + +/* + * Serialized form of a UConverterSelector, formatVersion 1: + * + * The serialized form begins with a standard ICU DataHeader with a UDataInfo + * as the template above. + * This is followed by: + * int32_t indexes[UCNVSEL_INDEX_COUNT]; // see index entry constants above + * serialized UTrie2; // indexes[UCNVSEL_INDEX_TRIE_SIZE] bytes + * uint32_t pv[indexes[UCNVSEL_INDEX_PV_COUNT]]; // bit vectors + * char* encodingNames[indexes[UCNVSEL_INDEX_NAMES_LENGTH]]; // NUL-terminated strings + padding + */ + +/* serialize a selector */ +U_CAPI int32_t U_EXPORT2 +ucnvsel_serialize(const UConverterSelector* sel, + void* buffer, int32_t bufferCapacity, UErrorCode* status) { + // check if already failed + if (U_FAILURE(*status)) { + return 0; + } + // ensure args make sense! + uint8_t *p = (uint8_t *)buffer; + if (bufferCapacity < 0 || + (bufferCapacity > 0 && (p == nullptr || (U_POINTER_MASK_LSB(p, 3) != 0))) + ) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + // add up the size of the serialized form + int32_t serializedTrieSize = utrie2_serialize(sel->trie, nullptr, 0, status); + if (*status != U_BUFFER_OVERFLOW_ERROR && U_FAILURE(*status)) { + return 0; + } + *status = U_ZERO_ERROR; + + DataHeader header; + uprv_memset(&header, 0, sizeof(header)); + header.dataHeader.headerSize = (uint16_t)((sizeof(header) + 15) & ~15); + header.dataHeader.magic1 = 0xda; + header.dataHeader.magic2 = 0x27; + uprv_memcpy(&header.info, &dataInfo, sizeof(dataInfo)); + + int32_t indexes[UCNVSEL_INDEX_COUNT] = { + serializedTrieSize, + sel->pvCount, + sel->encodingsCount, + sel->encodingStrLength + }; + + int32_t totalSize = + header.dataHeader.headerSize + + (int32_t)sizeof(indexes) + + serializedTrieSize + + sel->pvCount * 4 + + sel->encodingStrLength; + indexes[UCNVSEL_INDEX_SIZE] = totalSize - header.dataHeader.headerSize; + if (totalSize > bufferCapacity) { + *status = U_BUFFER_OVERFLOW_ERROR; + return totalSize; + } + // ok, save! + int32_t length = header.dataHeader.headerSize; + uprv_memcpy(p, &header, sizeof(header)); + uprv_memset(p + sizeof(header), 0, length - sizeof(header)); + p += length; + + length = (int32_t)sizeof(indexes); + uprv_memcpy(p, indexes, length); + p += length; + + utrie2_serialize(sel->trie, p, serializedTrieSize, status); + p += serializedTrieSize; + + length = sel->pvCount * 4; + uprv_memcpy(p, sel->pv, length); + p += length; + + uprv_memcpy(p, sel->encodings[0], sel->encodingStrLength); + p += sel->encodingStrLength; + + return totalSize; +} + +/** + * swap a selector into the desired Endianness and Asciiness of + * the system. Just as FYI, selectors are always saved in the format + * of the system that created them. They are only converted if used + * on another system. In other words, selectors created on different + * system can be different even if the params are identical (endianness + * and Asciiness differences only) + * + * @param ds pointer to data swapper containing swapping info + * @param inData pointer to incoming data + * @param length length of inData in bytes + * @param outData pointer to output data. Capacity should + * be at least equal to capacity of inData + * @param status an in/out ICU UErrorCode + * @return 0 on failure, number of bytes swapped on success + * number of bytes swapped can be smaller than length + */ +static int32_t +ucnvsel_swap(const UDataSwapper *ds, + const void *inData, int32_t length, + void *outData, UErrorCode *status) { + /* udata_swapDataHeader checks the arguments */ + int32_t headerSize = udata_swapDataHeader(ds, inData, length, outData, status); + if(U_FAILURE(*status)) { + return 0; + } + + /* check data format and format version */ + const UDataInfo *pInfo = (const UDataInfo *)((const char *)inData + 4); + if(!( + pInfo->dataFormat[0] == 0x43 && /* dataFormat="CSel" */ + pInfo->dataFormat[1] == 0x53 && + pInfo->dataFormat[2] == 0x65 && + pInfo->dataFormat[3] == 0x6c + )) { + udata_printError(ds, "ucnvsel_swap(): data format %02x.%02x.%02x.%02x is not recognized as UConverterSelector data\n", + pInfo->dataFormat[0], pInfo->dataFormat[1], + pInfo->dataFormat[2], pInfo->dataFormat[3]); + *status = U_INVALID_FORMAT_ERROR; + return 0; + } + if(pInfo->formatVersion[0] != 1) { + udata_printError(ds, "ucnvsel_swap(): format version %02x is not supported\n", + pInfo->formatVersion[0]); + *status = U_UNSUPPORTED_ERROR; + return 0; + } + + if(length >= 0) { + length -= headerSize; + if(length < 16*4) { + udata_printError(ds, "ucnvsel_swap(): too few bytes (%d after header) for UConverterSelector data\n", + length); + *status = U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + } + + const uint8_t *inBytes = (const uint8_t *)inData + headerSize; + uint8_t *outBytes = (uint8_t *)outData + headerSize; + + /* read the indexes */ + const int32_t *inIndexes = (const int32_t *)inBytes; + int32_t indexes[16]; + int32_t i; + for(i = 0; i < 16; ++i) { + indexes[i] = udata_readInt32(ds, inIndexes[i]); + } + + /* get the total length of the data */ + int32_t size = indexes[UCNVSEL_INDEX_SIZE]; + if(length >= 0) { + if(length < size) { + udata_printError(ds, "ucnvsel_swap(): too few bytes (%d after header) for all of UConverterSelector data\n", + length); + *status = U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + + /* copy the data for inaccessible bytes */ + if(inBytes != outBytes) { + uprv_memcpy(outBytes, inBytes, size); + } + + int32_t offset = 0, count; + + /* swap the int32_t indexes[] */ + count = UCNVSEL_INDEX_COUNT*4; + ds->swapArray32(ds, inBytes, count, outBytes, status); + offset += count; + + /* swap the UTrie2 */ + count = indexes[UCNVSEL_INDEX_TRIE_SIZE]; + utrie2_swap(ds, inBytes + offset, count, outBytes + offset, status); + offset += count; + + /* swap the uint32_t pv[] */ + count = indexes[UCNVSEL_INDEX_PV_COUNT]*4; + ds->swapArray32(ds, inBytes + offset, count, outBytes + offset, status); + offset += count; + + /* swap the encoding names */ + count = indexes[UCNVSEL_INDEX_NAMES_LENGTH]; + ds->swapInvChars(ds, inBytes + offset, count, outBytes + offset, status); + offset += count; + + U_ASSERT(offset == size); + } + + return headerSize + size; +} + +/* unserialize a selector */ +U_CAPI UConverterSelector* U_EXPORT2 +ucnvsel_openFromSerialized(const void* buffer, int32_t length, UErrorCode* status) { + // check if already failed + if (U_FAILURE(*status)) { + return nullptr; + } + // ensure args make sense! + const uint8_t *p = (const uint8_t *)buffer; + if (length <= 0 || + (length > 0 && (p == nullptr || (U_POINTER_MASK_LSB(p, 3) != 0))) + ) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + // header + if (length < 32) { + // not even enough space for a minimal header + *status = U_INDEX_OUTOFBOUNDS_ERROR; + return nullptr; + } + const DataHeader *pHeader = (const DataHeader *)p; + if (!( + pHeader->dataHeader.magic1==0xda && + pHeader->dataHeader.magic2==0x27 && + pHeader->info.dataFormat[0] == 0x43 && + pHeader->info.dataFormat[1] == 0x53 && + pHeader->info.dataFormat[2] == 0x65 && + pHeader->info.dataFormat[3] == 0x6c + )) { + /* header not valid or dataFormat not recognized */ + *status = U_INVALID_FORMAT_ERROR; + return nullptr; + } + if (pHeader->info.formatVersion[0] != 1) { + *status = U_UNSUPPORTED_ERROR; + return nullptr; + } + uint8_t* swapped = nullptr; + if (pHeader->info.isBigEndian != U_IS_BIG_ENDIAN || + pHeader->info.charsetFamily != U_CHARSET_FAMILY + ) { + // swap the data + UDataSwapper *ds = + udata_openSwapperForInputData(p, length, U_IS_BIG_ENDIAN, U_CHARSET_FAMILY, status); + int32_t totalSize = ucnvsel_swap(ds, p, -1, nullptr, status); + if (U_FAILURE(*status)) { + udata_closeSwapper(ds); + return nullptr; + } + if (length < totalSize) { + udata_closeSwapper(ds); + *status = U_INDEX_OUTOFBOUNDS_ERROR; + return nullptr; + } + swapped = (uint8_t*)uprv_malloc(totalSize); + if (swapped == nullptr) { + udata_closeSwapper(ds); + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + ucnvsel_swap(ds, p, length, swapped, status); + udata_closeSwapper(ds); + if (U_FAILURE(*status)) { + uprv_free(swapped); + return nullptr; + } + p = swapped; + pHeader = (const DataHeader *)p; + } + if (length < (pHeader->dataHeader.headerSize + 16 * 4)) { + // not even enough space for the header and the indexes + uprv_free(swapped); + *status = U_INDEX_OUTOFBOUNDS_ERROR; + return nullptr; + } + p += pHeader->dataHeader.headerSize; + length -= pHeader->dataHeader.headerSize; + // indexes + const int32_t *indexes = (const int32_t *)p; + if (length < indexes[UCNVSEL_INDEX_SIZE]) { + uprv_free(swapped); + *status = U_INDEX_OUTOFBOUNDS_ERROR; + return nullptr; + } + p += UCNVSEL_INDEX_COUNT * 4; + // create and populate the selector object + UConverterSelector* sel = (UConverterSelector*)uprv_malloc(sizeof(UConverterSelector)); + char **encodings = + (char **)uprv_malloc( + indexes[UCNVSEL_INDEX_NAMES_COUNT] * sizeof(char *)); + if (sel == nullptr || encodings == nullptr) { + uprv_free(swapped); + uprv_free(sel); + uprv_free(encodings); + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memset(sel, 0, sizeof(UConverterSelector)); + sel->pvCount = indexes[UCNVSEL_INDEX_PV_COUNT]; + sel->encodings = encodings; + sel->encodingsCount = indexes[UCNVSEL_INDEX_NAMES_COUNT]; + sel->encodingStrLength = indexes[UCNVSEL_INDEX_NAMES_LENGTH]; + sel->swapped = swapped; + // trie + sel->trie = utrie2_openFromSerialized(UTRIE2_16_VALUE_BITS, + p, indexes[UCNVSEL_INDEX_TRIE_SIZE], nullptr, + status); + p += indexes[UCNVSEL_INDEX_TRIE_SIZE]; + if (U_FAILURE(*status)) { + ucnvsel_close(sel); + return nullptr; + } + // bit vectors + sel->pv = (uint32_t *)p; + p += sel->pvCount * 4; + // encoding names + char* s = (char*)p; + for (int32_t i = 0; i < sel->encodingsCount; ++i) { + sel->encodings[i] = s; + s += uprv_strlen(s) + 1; + } + p += sel->encodingStrLength; + + return sel; +} + +// a bunch of functions for the enumeration thingie! Nothing fancy here. Just +// iterate over the selected encodings +struct Enumerator { + int16_t* index; + int16_t length; + int16_t cur; + const UConverterSelector* sel; +}; + +U_CDECL_BEGIN + +static void U_CALLCONV +ucnvsel_close_selector_iterator(UEnumeration *enumerator) { + uprv_free(((Enumerator*)(enumerator->context))->index); + uprv_free(enumerator->context); + uprv_free(enumerator); +} + + +static int32_t U_CALLCONV +ucnvsel_count_encodings(UEnumeration *enumerator, UErrorCode *status) { + // check if already failed + if (U_FAILURE(*status)) { + return 0; + } + return ((Enumerator*)(enumerator->context))->length; +} + + +static const char* U_CALLCONV ucnvsel_next_encoding(UEnumeration* enumerator, + int32_t* resultLength, + UErrorCode* status) { + // check if already failed + if (U_FAILURE(*status)) { + return nullptr; + } + + int16_t cur = ((Enumerator*)(enumerator->context))->cur; + const UConverterSelector* sel; + const char* result; + if (cur >= ((Enumerator*)(enumerator->context))->length) { + return nullptr; + } + sel = ((Enumerator*)(enumerator->context))->sel; + result = sel->encodings[((Enumerator*)(enumerator->context))->index[cur] ]; + ((Enumerator*)(enumerator->context))->cur++; + if (resultLength) { + *resultLength = (int32_t)uprv_strlen(result); + } + return result; +} + +static void U_CALLCONV ucnvsel_reset_iterator(UEnumeration* enumerator, + UErrorCode* status) { + // check if already failed + if (U_FAILURE(*status)) { + return ; + } + ((Enumerator*)(enumerator->context))->cur = 0; +} + +U_CDECL_END + + +static const UEnumeration defaultEncodings = { + nullptr, + nullptr, + ucnvsel_close_selector_iterator, + ucnvsel_count_encodings, + uenum_unextDefault, + ucnvsel_next_encoding, + ucnvsel_reset_iterator +}; + + +// internal fn to intersect two sets of masks +// returns whether the mask has reduced to all zeros +static UBool intersectMasks(uint32_t* dest, const uint32_t* source1, int32_t len) { + int32_t i; + uint32_t oredDest = 0; + for (i = 0 ; i < len ; ++i) { + oredDest |= (dest[i] &= source1[i]); + } + return oredDest == 0; +} + +// internal fn to count how many 1's are there in a mask +// algorithm taken from http://graphics.stanford.edu/~seander/bithacks.html +static int16_t countOnes(uint32_t* mask, int32_t len) { + int32_t i, totalOnes = 0; + for (i = 0 ; i < len ; ++i) { + uint32_t ent = mask[i]; + for (; ent; totalOnes++) + { + ent &= ent - 1; // clear the least significant bit set + } + } + return static_cast(totalOnes); +} + + +/* internal function! */ +static UEnumeration *selectForMask(const UConverterSelector* sel, + uint32_t *theMask, UErrorCode *status) { + LocalMemory mask(theMask); + // this is the context we will use. Store a table of indices to which + // encodings are legit. + LocalMemory result(static_cast(uprv_malloc(sizeof(Enumerator)))); + if (result.isNull()) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + result->index = nullptr; // this will be allocated later! + result->length = result->cur = 0; + result->sel = sel; + + LocalMemory en(static_cast(uprv_malloc(sizeof(UEnumeration)))); + if (en.isNull()) { + // TODO(markus): Combine Enumerator and UEnumeration into one struct. + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + memcpy(en.getAlias(), &defaultEncodings, sizeof(UEnumeration)); + + int32_t columns = (sel->encodingsCount+31)/32; + int16_t numOnes = countOnes(mask.getAlias(), columns); + // now, we know the exact space we need for index + if (numOnes > 0) { + result->index = static_cast(uprv_malloc(numOnes * sizeof(int16_t))); + if (result->index == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + int32_t i, j; + int16_t k = 0; + for (j = 0 ; j < columns; j++) { + uint32_t v = mask[j]; + for (i = 0 ; i < 32 && k < sel->encodingsCount; i++, k++) { + if ((v & 1) != 0) { + result->index[result->length++] = k; + } + v >>= 1; + } + } + } //otherwise, index will remain nullptr (and will never be touched by + //the enumerator code anyway) + en->context = result.orphan(); + return en.orphan(); +} + +/* check a string against the selector - UTF16 version */ +U_CAPI UEnumeration * U_EXPORT2 +ucnvsel_selectForString(const UConverterSelector* sel, + const char16_t *s, int32_t length, UErrorCode *status) { + // check if already failed + if (U_FAILURE(*status)) { + return nullptr; + } + // ensure args make sense! + if (sel == nullptr || (s == nullptr && length != 0)) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + + int32_t columns = (sel->encodingsCount+31)/32; + uint32_t* mask = (uint32_t*) uprv_malloc(columns * 4); + if (mask == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memset(mask, ~0, columns *4); + + if(s!=nullptr) { + const char16_t *limit; + if (length >= 0) { + limit = s + length; + } else { + limit = nullptr; + } + + while (limit == nullptr ? *s != 0 : s != limit) { + UChar32 c; + uint16_t pvIndex; + UTRIE2_U16_NEXT16(sel->trie, s, limit, c, pvIndex); + if (intersectMasks(mask, sel->pv+pvIndex, columns)) { + break; + } + } + } + return selectForMask(sel, mask, status); +} + +/* check a string against the selector - UTF8 version */ +U_CAPI UEnumeration * U_EXPORT2 +ucnvsel_selectForUTF8(const UConverterSelector* sel, + const char *s, int32_t length, UErrorCode *status) { + // check if already failed + if (U_FAILURE(*status)) { + return nullptr; + } + // ensure args make sense! + if (sel == nullptr || (s == nullptr && length != 0)) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + + int32_t columns = (sel->encodingsCount+31)/32; + uint32_t* mask = (uint32_t*) uprv_malloc(columns * 4); + if (mask == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memset(mask, ~0, columns *4); + + if (length < 0) { + length = (int32_t)uprv_strlen(s); + } + + if(s!=nullptr) { + const char *limit = s + length; + + while (s != limit) { + uint16_t pvIndex; + UTRIE2_U8_NEXT16(sel->trie, s, limit, pvIndex); + if (intersectMasks(mask, sel->pv+pvIndex, columns)) { + break; + } + } + } + return selectForMask(sel, mask, status); +} + +#endif // !UCONFIG_NO_CONVERSION diff --git a/intl/icu/source/common/ucol_data.h b/intl/icu/source/common/ucol_data.h new file mode 100644 index 0000000000..28cf94ffc8 --- /dev/null +++ b/intl/icu/source/common/ucol_data.h @@ -0,0 +1,89 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2000-2011, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: ucol_data.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2011jul02 +* created by: Markus Scherer +* +* Private implementation header for C/C++ collation. +* Some file data structure definitions were moved here from i18n/ucol_imp.h +* so that the common library (via ucol_swp.cpp) need not depend on the i18n library at all. +* +* We do not want to move the collation swapper to the i18n library because +* a) the resource bundle swapper depends on it and would have to move too, and +* b) we might want to eventually implement runtime data swapping, +* which might (or might not) be easier if all swappers are in the common library. +*/ + +#ifndef __UCOL_DATA_H__ +#define __UCOL_DATA_H__ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_COLLATION + +/* let us know whether reserved fields are reset to zero or junked */ +#define UCOL_HEADER_MAGIC 0x20030618 + +typedef struct { + int32_t size; + /* all the offsets are in bytes */ + /* to get the address add to the header address and cast properly */ + uint32_t options; /* these are the default options for the collator */ + uint32_t UCAConsts; /* structure which holds values for indirect positioning and implicit ranges */ + uint32_t contractionUCACombos; /* this one is needed only for UCA, to copy the appropriate contractions */ + uint32_t magic; /* magic number - lets us know whether reserved data is reset or junked */ + uint32_t mappingPosition; /* const uint8_t *mappingPosition; */ + uint32_t expansion; /* uint32_t *expansion; */ + uint32_t contractionIndex; /* char16_t *contractionIndex; */ + uint32_t contractionCEs; /* uint32_t *contractionCEs; */ + uint32_t contractionSize; /* needed for various closures */ + /*int32_t latinOneMapping;*/ /* this is now handled in the trie itself *//* fast track to latin1 chars */ + + uint32_t endExpansionCE; /* array of last collation element in + expansion */ + uint32_t expansionCESize; /* array of maximum expansion size + corresponding to the expansion + collation elements with last element + in endExpansionCE*/ + int32_t endExpansionCECount; /* size of endExpansionCE */ + uint32_t unsafeCP; /* hash table of unsafe code points */ + uint32_t contrEndCP; /* hash table of final code points */ + /* in contractions. */ + + int32_t contractionUCACombosSize; /* number of UCA contraction items. */ + /*Length is contractionUCACombosSize*contractionUCACombosWidth*sizeof(char16_t) */ + UBool jamoSpecial; /* is jamoSpecial */ + UBool isBigEndian; /* is this data big endian? from the UDataInfo header*/ + uint8_t charSetFamily; /* what is the charset family of this data from the UDataInfo header*/ + uint8_t contractionUCACombosWidth; /* width of UCA combos field */ + UVersionInfo version; + UVersionInfo UCAVersion; /* version of the UCA, read from file */ + UVersionInfo UCDVersion; /* UCD version, obtained by u_getUnicodeVersion */ + UVersionInfo formatVersion; /* format version from the UDataInfo header */ + uint32_t scriptToLeadByte; /* offset to script to lead collation byte mapping data */ + uint32_t leadByteToScript; /* offset to lead collation byte to script mapping data */ + uint8_t reserved[76]; /* for future use */ +} UCATableHeader; + +typedef struct { + uint32_t byteSize; + uint32_t tableSize; + uint32_t contsSize; + uint32_t table; + uint32_t conts; + UVersionInfo UCAVersion; /* version of the UCA, read from file */ + uint8_t padding[8]; +} InverseUCATableHeader; + +#endif /* !UCONFIG_NO_COLLATION */ + +#endif /* __UCOL_DATA_H__ */ diff --git a/intl/icu/source/common/ucol_swp.cpp b/intl/icu/source/common/ucol_swp.cpp new file mode 100644 index 0000000000..b5894a1fef --- /dev/null +++ b/intl/icu/source/common/ucol_swp.cpp @@ -0,0 +1,615 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2003-2015, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: ucol_swp.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2003sep10 +* created by: Markus W. Scherer +* +* Swap collation binaries. +*/ + +#include "unicode/udata.h" /* UDataInfo */ +#include "utrie.h" +#include "utrie2.h" +#include "udataswp.h" +#include "cmemory.h" +#include "ucol_data.h" +#include "ucol_swp.h" + +/* swapping ----------------------------------------------------------------- */ + +#if !UCONFIG_NO_COLLATION + +U_CAPI UBool U_EXPORT2 +ucol_looksLikeCollationBinary(const UDataSwapper *ds, + const void *inData, int32_t length) { + if(ds==nullptr || inData==nullptr || length<-1) { + return false; + } + + // First check for format version 4+ which has a standard data header. + UErrorCode errorCode=U_ZERO_ERROR; + (void)udata_swapDataHeader(ds, inData, -1, nullptr, &errorCode); + if(U_SUCCESS(errorCode)) { + const UDataInfo &info=*(const UDataInfo *)((const char *)inData+4); + if(info.dataFormat[0]==0x55 && // dataFormat="UCol" + info.dataFormat[1]==0x43 && + info.dataFormat[2]==0x6f && + info.dataFormat[3]==0x6c) { + return true; + } + } + + // Else check for format version 3. + const UCATableHeader *inHeader=(const UCATableHeader *)inData; + + /* + * The collation binary must contain at least the UCATableHeader, + * starting with its size field. + * sizeof(UCATableHeader)==42*4 in ICU 2.8 + * check the length against the header size before reading the size field + */ + UCATableHeader header; + uprv_memset(&header, 0, sizeof(header)); + if(length<0) { + header.size=udata_readInt32(ds, inHeader->size); + } else if((length<(42*4) || length<(header.size=udata_readInt32(ds, inHeader->size)))) { + return false; + } + + header.magic=ds->readUInt32(inHeader->magic); + if(!( + header.magic==UCOL_HEADER_MAGIC && + inHeader->formatVersion[0]==3 /*&& + inHeader->formatVersion[1]>=0*/ + )) { + return false; + } + + if(inHeader->isBigEndian!=ds->inIsBigEndian || inHeader->charSetFamily!=ds->inCharset) { + return false; + } + + return true; +} + +namespace { + +/* swap a header-less collation formatVersion=3 binary, inside a resource bundle or ucadata.icu */ +int32_t +swapFormatVersion3(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const uint8_t *inBytes; + uint8_t *outBytes; + + const UCATableHeader *inHeader; + UCATableHeader *outHeader; + UCATableHeader header; + + uint32_t count; + + /* argument checking in case we were not called from ucol_swap() */ + if(U_FAILURE(*pErrorCode)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<-1 || (length>0 && outData==nullptr)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + inBytes=(const uint8_t *)inData; + outBytes=(uint8_t *)outData; + + inHeader=(const UCATableHeader *)inData; + outHeader=(UCATableHeader *)outData; + + /* + * The collation binary must contain at least the UCATableHeader, + * starting with its size field. + * sizeof(UCATableHeader)==42*4 in ICU 2.8 + * check the length against the header size before reading the size field + */ + uprv_memset(&header, 0, sizeof(header)); + if(length<0) { + header.size=udata_readInt32(ds, inHeader->size); + } else if((length<(42*4) || length<(header.size=udata_readInt32(ds, inHeader->size)))) { + udata_printError(ds, "ucol_swap(formatVersion=3): too few bytes (%d after header) for collation data\n", + length); + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + + header.magic=ds->readUInt32(inHeader->magic); + if(!( + header.magic==UCOL_HEADER_MAGIC && + inHeader->formatVersion[0]==3 /*&& + inHeader->formatVersion[1]>=0*/ + )) { + udata_printError(ds, "ucol_swap(formatVersion=3): magic 0x%08x or format version %02x.%02x is not a collation binary\n", + header.magic, + inHeader->formatVersion[0], inHeader->formatVersion[1]); + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + + if(inHeader->isBigEndian!=ds->inIsBigEndian || inHeader->charSetFamily!=ds->inCharset) { + udata_printError(ds, "ucol_swap(formatVersion=3): endianness %d or charset %d does not match the swapper\n", + inHeader->isBigEndian, inHeader->charSetFamily); + *pErrorCode=U_INVALID_FORMAT_ERROR; + return 0; + } + + if(length>=0) { + /* copy everything, takes care of data that needs no swapping */ + if(inBytes!=outBytes) { + uprv_memcpy(outBytes, inBytes, header.size); + } + + /* swap the necessary pieces in the order of their occurrence in the data */ + + /* read more of the UCATableHeader (the size field was read above) */ + header.options= ds->readUInt32(inHeader->options); + header.UCAConsts= ds->readUInt32(inHeader->UCAConsts); + header.contractionUCACombos= ds->readUInt32(inHeader->contractionUCACombos); + header.mappingPosition= ds->readUInt32(inHeader->mappingPosition); + header.expansion= ds->readUInt32(inHeader->expansion); + header.contractionIndex= ds->readUInt32(inHeader->contractionIndex); + header.contractionCEs= ds->readUInt32(inHeader->contractionCEs); + header.contractionSize= ds->readUInt32(inHeader->contractionSize); + header.endExpansionCE= ds->readUInt32(inHeader->endExpansionCE); + header.expansionCESize= ds->readUInt32(inHeader->expansionCESize); + header.endExpansionCECount= udata_readInt32(ds, inHeader->endExpansionCECount); + header.contractionUCACombosSize=udata_readInt32(ds, inHeader->contractionUCACombosSize); + header.scriptToLeadByte= ds->readUInt32(inHeader->scriptToLeadByte); + header.leadByteToScript= ds->readUInt32(inHeader->leadByteToScript); + + /* swap the 32-bit integers in the header */ + ds->swapArray32(ds, inHeader, (int32_t)((const char *)&inHeader->jamoSpecial-(const char *)inHeader), + outHeader, pErrorCode); + ds->swapArray32(ds, &(inHeader->scriptToLeadByte), sizeof(header.scriptToLeadByte) + sizeof(header.leadByteToScript), + &(outHeader->scriptToLeadByte), pErrorCode); + /* set the output platform properties */ + outHeader->isBigEndian=ds->outIsBigEndian; + outHeader->charSetFamily=ds->outCharset; + + /* swap the options */ + if(header.options!=0) { + ds->swapArray32(ds, inBytes+header.options, header.expansion-header.options, + outBytes+header.options, pErrorCode); + } + + /* swap the expansions */ + if(header.mappingPosition!=0 && header.expansion!=0) { + if(header.contractionIndex!=0) { + /* expansions bounded by contractions */ + count=header.contractionIndex-header.expansion; + } else { + /* no contractions: expansions bounded by the main trie */ + count=header.mappingPosition-header.expansion; + } + ds->swapArray32(ds, inBytes+header.expansion, (int32_t)count, + outBytes+header.expansion, pErrorCode); + } + + /* swap the contractions */ + if(header.contractionSize!=0) { + /* contractionIndex: char16_t[] */ + ds->swapArray16(ds, inBytes+header.contractionIndex, header.contractionSize*2, + outBytes+header.contractionIndex, pErrorCode); + + /* contractionCEs: CEs[] */ + ds->swapArray32(ds, inBytes+header.contractionCEs, header.contractionSize*4, + outBytes+header.contractionCEs, pErrorCode); + } + + /* swap the main trie */ + if(header.mappingPosition!=0) { + count=header.endExpansionCE-header.mappingPosition; + utrie_swap(ds, inBytes+header.mappingPosition, (int32_t)count, + outBytes+header.mappingPosition, pErrorCode); + } + + /* swap the max expansion table */ + if(header.endExpansionCECount!=0) { + ds->swapArray32(ds, inBytes+header.endExpansionCE, header.endExpansionCECount*4, + outBytes+header.endExpansionCE, pErrorCode); + } + + /* expansionCESize, unsafeCP, contrEndCP: uint8_t[], no need to swap */ + + /* swap UCA constants */ + if(header.UCAConsts!=0) { + /* + * if UCAConsts!=0 then contractionUCACombos because we are swapping + * the UCA data file, and we know that the UCA contains contractions + */ + ds->swapArray32(ds, inBytes+header.UCAConsts, header.contractionUCACombos-header.UCAConsts, + outBytes+header.UCAConsts, pErrorCode); + } + + /* swap UCA contractions */ + if(header.contractionUCACombosSize!=0) { + count=header.contractionUCACombosSize*inHeader->contractionUCACombosWidth*U_SIZEOF_UCHAR; + ds->swapArray16(ds, inBytes+header.contractionUCACombos, (int32_t)count, + outBytes+header.contractionUCACombos, pErrorCode); + } + + /* swap the script to lead bytes */ + if(header.scriptToLeadByte!=0) { + int indexCount = ds->readUInt16(*((uint16_t*)(inBytes+header.scriptToLeadByte))); // each entry = 2 * uint16 + int dataCount = ds->readUInt16(*((uint16_t*)(inBytes+header.scriptToLeadByte + 2))); // each entry = uint16 + ds->swapArray16(ds, inBytes+header.scriptToLeadByte, + 4 + (4 * indexCount) + (2 * dataCount), + outBytes+header.scriptToLeadByte, pErrorCode); + } + + /* swap the lead byte to scripts */ + if(header.leadByteToScript!=0) { + int indexCount = ds->readUInt16(*((uint16_t*)(inBytes+header.leadByteToScript))); // each entry = uint16 + int dataCount = ds->readUInt16(*((uint16_t*)(inBytes+header.leadByteToScript + 2))); // each entry = uint16 + ds->swapArray16(ds, inBytes+header.leadByteToScript, + 4 + (2 * indexCount) + (2 * dataCount), + outBytes+header.leadByteToScript, pErrorCode); + } + } + + return header.size; +} + +// swap formatVersion 4 or 5 ----------------------------------------------- *** + +// The following are copied from CollationDataReader, trading an awkward copy of constants +// for an awkward relocation of the i18n collationdatareader.h file into the common library. +// Keep them in sync! + +enum { + IX_INDEXES_LENGTH, // 0 + IX_OPTIONS, + IX_RESERVED2, + IX_RESERVED3, + + IX_JAMO_CE32S_START, // 4 + IX_REORDER_CODES_OFFSET, + IX_REORDER_TABLE_OFFSET, + IX_TRIE_OFFSET, + + IX_RESERVED8_OFFSET, // 8 + IX_CES_OFFSET, + IX_RESERVED10_OFFSET, + IX_CE32S_OFFSET, + + IX_ROOT_ELEMENTS_OFFSET, // 12 + IX_CONTEXTS_OFFSET, + IX_UNSAFE_BWD_OFFSET, + IX_FAST_LATIN_TABLE_OFFSET, + + IX_SCRIPTS_OFFSET, // 16 + IX_COMPRESSIBLE_BYTES_OFFSET, + IX_RESERVED18_OFFSET, + IX_TOTAL_SIZE +}; + +int32_t +swapFormatVersion4(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode &errorCode) { + if(U_FAILURE(errorCode)) { return 0; } + + const uint8_t *inBytes=(const uint8_t *)inData; + uint8_t *outBytes=(uint8_t *)outData; + + const int32_t *inIndexes=(const int32_t *)inBytes; + int32_t indexes[IX_TOTAL_SIZE+1]; + + // Need at least IX_INDEXES_LENGTH and IX_OPTIONS. + if(0<=length && length<8) { + udata_printError(ds, "ucol_swap(formatVersion=4): too few bytes " + "(%d after header) for collation data\n", + length); + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + + int32_t indexesLength=indexes[0]=udata_readInt32(ds, inIndexes[0]); + if(0<=length && length<(indexesLength*4)) { + udata_printError(ds, "ucol_swap(formatVersion=4): too few bytes " + "(%d after header) for collation data\n", + length); + errorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + + for(int32_t i=1; i<=IX_TOTAL_SIZE && iIX_TOTAL_SIZE) { + size=indexes[IX_TOTAL_SIZE]; + } else if(indexesLength>IX_REORDER_CODES_OFFSET) { + size=indexes[indexesLength-1]; + } else { + size=indexesLength*4; + } + if(length<0) { return size; } + + if(lengthswapArray32(ds, inBytes, indexesLength * 4, outBytes, &errorCode); + + // The following is a modified version of CollationDataReader::read(). + // Here we use indexes[] not inIndexes[] because + // the inIndexes[] may not be in this machine's endianness. + int32_t index; // one of the indexes[] slots + int32_t offset; // byte offset for the index part + // int32_t length; // number of bytes in the index part + + index = IX_REORDER_CODES_OFFSET; + offset = indexes[index]; + length = indexes[index + 1] - offset; + if(length > 0) { + ds->swapArray32(ds, inBytes + offset, length, outBytes + offset, &errorCode); + } + + // Skip the IX_REORDER_TABLE_OFFSET byte array. + + index = IX_TRIE_OFFSET; + offset = indexes[index]; + length = indexes[index + 1] - offset; + if(length > 0) { + utrie2_swap(ds, inBytes + offset, length, outBytes + offset, &errorCode); + } + + index = IX_RESERVED8_OFFSET; + offset = indexes[index]; + length = indexes[index + 1] - offset; + if(length > 0) { + udata_printError(ds, "ucol_swap(formatVersion=4): unknown data at IX_RESERVED8_OFFSET\n", length); + errorCode = U_UNSUPPORTED_ERROR; + return 0; + } + + index = IX_CES_OFFSET; + offset = indexes[index]; + length = indexes[index + 1] - offset; + if(length > 0) { + ds->swapArray64(ds, inBytes + offset, length, outBytes + offset, &errorCode); + } + + index = IX_RESERVED10_OFFSET; + offset = indexes[index]; + length = indexes[index + 1] - offset; + if(length > 0) { + udata_printError(ds, "ucol_swap(formatVersion=4): unknown data at IX_RESERVED10_OFFSET\n", length); + errorCode = U_UNSUPPORTED_ERROR; + return 0; + } + + index = IX_CE32S_OFFSET; + offset = indexes[index]; + length = indexes[index + 1] - offset; + if(length > 0) { + ds->swapArray32(ds, inBytes + offset, length, outBytes + offset, &errorCode); + } + + index = IX_ROOT_ELEMENTS_OFFSET; + offset = indexes[index]; + length = indexes[index + 1] - offset; + if(length > 0) { + ds->swapArray32(ds, inBytes + offset, length, outBytes + offset, &errorCode); + } + + index = IX_CONTEXTS_OFFSET; + offset = indexes[index]; + length = indexes[index + 1] - offset; + if(length > 0) { + ds->swapArray16(ds, inBytes + offset, length, outBytes + offset, &errorCode); + } + + index = IX_UNSAFE_BWD_OFFSET; + offset = indexes[index]; + length = indexes[index + 1] - offset; + if(length > 0) { + ds->swapArray16(ds, inBytes + offset, length, outBytes + offset, &errorCode); + } + + index = IX_FAST_LATIN_TABLE_OFFSET; + offset = indexes[index]; + length = indexes[index + 1] - offset; + if(length > 0) { + ds->swapArray16(ds, inBytes + offset, length, outBytes + offset, &errorCode); + } + + index = IX_SCRIPTS_OFFSET; + offset = indexes[index]; + length = indexes[index + 1] - offset; + if(length > 0) { + ds->swapArray16(ds, inBytes + offset, length, outBytes + offset, &errorCode); + } + + // Skip the IX_COMPRESSIBLE_BYTES_OFFSET byte array. + + index = IX_RESERVED18_OFFSET; + offset = indexes[index]; + length = indexes[index + 1] - offset; + if(length > 0) { + udata_printError(ds, "ucol_swap(formatVersion=4): unknown data at IX_RESERVED18_OFFSET\n", length); + errorCode = U_UNSUPPORTED_ERROR; + return 0; + } + + return size; +} + +} // namespace + +/* swap ICU collation data like ucadata.icu */ +U_CAPI int32_t U_EXPORT2 +ucol_swap(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + if(U_FAILURE(*pErrorCode)) { return 0; } + + /* udata_swapDataHeader checks the arguments */ + int32_t headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + // Try to swap the old format version which did not have a standard data header. + *pErrorCode=U_ZERO_ERROR; + return swapFormatVersion3(ds, inData, length, outData, pErrorCode); + } + + /* check data format and format version */ + const UDataInfo &info=*(const UDataInfo *)((const char *)inData+4); + if(!( + info.dataFormat[0]==0x55 && // dataFormat="UCol" + info.dataFormat[1]==0x43 && + info.dataFormat[2]==0x6f && + info.dataFormat[3]==0x6c && + (3<=info.formatVersion[0] && info.formatVersion[0]<=5) + )) { + udata_printError(ds, "ucol_swap(): data format %02x.%02x.%02x.%02x " + "(format version %02x.%02x) is not recognized as collation data\n", + info.dataFormat[0], info.dataFormat[1], + info.dataFormat[2], info.dataFormat[3], + info.formatVersion[0], info.formatVersion[1]); + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + + inData=(const char *)inData+headerSize; + if(length>=0) { length-=headerSize; } + outData=(outData == nullptr) ? nullptr : (char *)outData+headerSize; + int32_t collationSize; + if(info.formatVersion[0]>=4) { + collationSize=swapFormatVersion4(ds, inData, length, outData, *pErrorCode); + } else { + collationSize=swapFormatVersion3(ds, inData, length, outData, pErrorCode); + } + if(U_SUCCESS(*pErrorCode)) { + return headerSize+collationSize; + } else { + return 0; + } +} + +/* swap inverse UCA collation data (invuca.icu) */ +U_CAPI int32_t U_EXPORT2 +ucol_swapInverseUCA(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const UDataInfo *pInfo; + int32_t headerSize; + + const uint8_t *inBytes; + uint8_t *outBytes; + + const InverseUCATableHeader *inHeader; + InverseUCATableHeader *outHeader; + InverseUCATableHeader header={ 0,0,0,0,0,{0,0,0,0},{0,0,0,0,0,0,0,0} }; + + /* udata_swapDataHeader checks the arguments */ + headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode); + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + + /* check data format and format version */ + pInfo=(const UDataInfo *)((const char *)inData+4); + if(!( + pInfo->dataFormat[0]==0x49 && /* dataFormat="InvC" */ + pInfo->dataFormat[1]==0x6e && + pInfo->dataFormat[2]==0x76 && + pInfo->dataFormat[3]==0x43 && + pInfo->formatVersion[0]==2 && + pInfo->formatVersion[1]>=1 + )) { + udata_printError(ds, "ucol_swapInverseUCA(): data format %02x.%02x.%02x.%02x (format version %02x.%02x) is not an inverse UCA collation file\n", + pInfo->dataFormat[0], pInfo->dataFormat[1], + pInfo->dataFormat[2], pInfo->dataFormat[3], + pInfo->formatVersion[0], pInfo->formatVersion[1]); + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + + inBytes=(const uint8_t *)inData+headerSize; + outBytes=(uint8_t *)outData+headerSize; + + inHeader=(const InverseUCATableHeader *)inBytes; + outHeader=(InverseUCATableHeader *)outBytes; + + /* + * The inverse UCA collation binary must contain at least the InverseUCATableHeader, + * starting with its size field. + * sizeof(UCATableHeader)==8*4 in ICU 2.8 + * check the length against the header size before reading the size field + */ + if(length<0) { + header.byteSize=udata_readInt32(ds, inHeader->byteSize); + } else if( + ((length-headerSize)<(8*4) || + (uint32_t)(length-headerSize)<(header.byteSize=udata_readInt32(ds, inHeader->byteSize))) + ) { + udata_printError(ds, "ucol_swapInverseUCA(): too few bytes (%d after header) for inverse UCA collation data\n", + length); + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + + if(length>=0) { + /* copy everything, takes care of data that needs no swapping */ + if(inBytes!=outBytes) { + uprv_memcpy(outBytes, inBytes, header.byteSize); + } + + /* swap the necessary pieces in the order of their occurrence in the data */ + + /* read more of the InverseUCATableHeader (the byteSize field was read above) */ + header.tableSize= ds->readUInt32(inHeader->tableSize); + header.contsSize= ds->readUInt32(inHeader->contsSize); + header.table= ds->readUInt32(inHeader->table); + header.conts= ds->readUInt32(inHeader->conts); + + /* swap the 32-bit integers in the header */ + ds->swapArray32(ds, inHeader, 5*4, outHeader, pErrorCode); + + /* swap the inverse table; tableSize counts uint32_t[3] rows */ + ds->swapArray32(ds, inBytes+header.table, header.tableSize*3*4, + outBytes+header.table, pErrorCode); + + /* swap the continuation table; contsSize counts UChars */ + ds->swapArray16(ds, inBytes+header.conts, header.contsSize*U_SIZEOF_UCHAR, + outBytes+header.conts, pErrorCode); + } + + return headerSize+header.byteSize; +} + +#endif /* #if !UCONFIG_NO_COLLATION */ diff --git a/intl/icu/source/common/ucol_swp.h b/intl/icu/source/common/ucol_swp.h new file mode 100644 index 0000000000..0c2990a85e --- /dev/null +++ b/intl/icu/source/common/ucol_swp.h @@ -0,0 +1,58 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2003-2014, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: ucol_swp.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2003sep10 +* created by: Markus W. Scherer +* +* Swap collation binaries. +*/ + +#ifndef __UCOL_SWP_H__ +#define __UCOL_SWP_H__ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_COLLATION + +#include "udataswp.h" + +/* + * Does the data look like a collation binary? + * @internal + */ +U_CAPI UBool U_EXPORT2 +ucol_looksLikeCollationBinary(const UDataSwapper *ds, + const void *inData, int32_t length); + +/** + * Swap ICU collation data like ucadata.icu. See udataswp.h. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +ucol_swap(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +/** + * Swap inverse UCA collation data (invuca.icu). See udataswp.h. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +ucol_swapInverseUCA(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +#endif /* #if !UCONFIG_NO_COLLATION */ + +#endif diff --git a/intl/icu/source/common/ucptrie.cpp b/intl/icu/source/common/ucptrie.cpp new file mode 100644 index 0000000000..0004160a23 --- /dev/null +++ b/intl/icu/source/common/ucptrie.cpp @@ -0,0 +1,601 @@ +// © 2017 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// ucptrie.cpp (modified from utrie2.cpp) +// created: 2017dec29 Markus W. Scherer + +// #define UCPTRIE_DEBUG +#ifdef UCPTRIE_DEBUG +# include +#endif + +#include "unicode/utypes.h" +#include "unicode/ucptrie.h" +#include "unicode/utf.h" +#include "unicode/utf8.h" +#include "unicode/utf16.h" +#include "cmemory.h" +#include "uassert.h" +#include "ucptrie_impl.h" + +U_CAPI UCPTrie * U_EXPORT2 +ucptrie_openFromBinary(UCPTrieType type, UCPTrieValueWidth valueWidth, + const void *data, int32_t length, int32_t *pActualLength, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return nullptr; + } + + if (length <= 0 || (U_POINTER_MASK_LSB(data, 3) != 0) || + type < UCPTRIE_TYPE_ANY || UCPTRIE_TYPE_SMALL < type || + valueWidth < UCPTRIE_VALUE_BITS_ANY || UCPTRIE_VALUE_BITS_8 < valueWidth) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + + // Enough data for a trie header? + if (length < (int32_t)sizeof(UCPTrieHeader)) { + *pErrorCode = U_INVALID_FORMAT_ERROR; + return nullptr; + } + + // Check the signature. + const UCPTrieHeader *header = (const UCPTrieHeader *)data; + if (header->signature != UCPTRIE_SIG) { + *pErrorCode = U_INVALID_FORMAT_ERROR; + return nullptr; + } + + int32_t options = header->options; + int32_t typeInt = (options >> 6) & 3; + int32_t valueWidthInt = options & UCPTRIE_OPTIONS_VALUE_BITS_MASK; + if (typeInt > UCPTRIE_TYPE_SMALL || valueWidthInt > UCPTRIE_VALUE_BITS_8 || + (options & UCPTRIE_OPTIONS_RESERVED_MASK) != 0) { + *pErrorCode = U_INVALID_FORMAT_ERROR; + return nullptr; + } + UCPTrieType actualType = (UCPTrieType)typeInt; + UCPTrieValueWidth actualValueWidth = (UCPTrieValueWidth)valueWidthInt; + if (type < 0) { + type = actualType; + } + if (valueWidth < 0) { + valueWidth = actualValueWidth; + } + if (type != actualType || valueWidth != actualValueWidth) { + *pErrorCode = U_INVALID_FORMAT_ERROR; + return nullptr; + } + + // Get the length values and offsets. + UCPTrie tempTrie; + uprv_memset(&tempTrie, 0, sizeof(tempTrie)); + tempTrie.indexLength = header->indexLength; + tempTrie.dataLength = + ((options & UCPTRIE_OPTIONS_DATA_LENGTH_MASK) << 4) | header->dataLength; + tempTrie.index3NullOffset = header->index3NullOffset; + tempTrie.dataNullOffset = + ((options & UCPTRIE_OPTIONS_DATA_NULL_OFFSET_MASK) << 8) | header->dataNullOffset; + + tempTrie.highStart = header->shiftedHighStart << UCPTRIE_SHIFT_2; + tempTrie.shifted12HighStart = (tempTrie.highStart + 0xfff) >> 12; + tempTrie.type = type; + tempTrie.valueWidth = valueWidth; + + // Calculate the actual length. + int32_t actualLength = (int32_t)sizeof(UCPTrieHeader) + tempTrie.indexLength * 2; + if (valueWidth == UCPTRIE_VALUE_BITS_16) { + actualLength += tempTrie.dataLength * 2; + } else if (valueWidth == UCPTRIE_VALUE_BITS_32) { + actualLength += tempTrie.dataLength * 4; + } else { + actualLength += tempTrie.dataLength; + } + if (length < actualLength) { + *pErrorCode = U_INVALID_FORMAT_ERROR; // Not enough bytes. + return nullptr; + } + + // Allocate the trie. + UCPTrie *trie = (UCPTrie *)uprv_malloc(sizeof(UCPTrie)); + if (trie == nullptr) { + *pErrorCode = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memcpy(trie, &tempTrie, sizeof(tempTrie)); +#ifdef UCPTRIE_DEBUG + trie->name = "fromSerialized"; +#endif + + // Set the pointers to its index and data arrays. + const uint16_t *p16 = (const uint16_t *)(header + 1); + trie->index = p16; + p16 += trie->indexLength; + + // Get the data. + int32_t nullValueOffset = trie->dataNullOffset; + if (nullValueOffset >= trie->dataLength) { + nullValueOffset = trie->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET; + } + switch (valueWidth) { + case UCPTRIE_VALUE_BITS_16: + trie->data.ptr16 = p16; + trie->nullValue = trie->data.ptr16[nullValueOffset]; + break; + case UCPTRIE_VALUE_BITS_32: + trie->data.ptr32 = (const uint32_t *)p16; + trie->nullValue = trie->data.ptr32[nullValueOffset]; + break; + case UCPTRIE_VALUE_BITS_8: + trie->data.ptr8 = (const uint8_t *)p16; + trie->nullValue = trie->data.ptr8[nullValueOffset]; + break; + default: + // Unreachable because valueWidth was checked above. + *pErrorCode = U_INVALID_FORMAT_ERROR; + return nullptr; + } + + if (pActualLength != nullptr) { + *pActualLength = actualLength; + } + return trie; +} + +U_CAPI void U_EXPORT2 +ucptrie_close(UCPTrie *trie) { + uprv_free(trie); +} + +U_CAPI UCPTrieType U_EXPORT2 +ucptrie_getType(const UCPTrie *trie) { + return (UCPTrieType)trie->type; +} + +U_CAPI UCPTrieValueWidth U_EXPORT2 +ucptrie_getValueWidth(const UCPTrie *trie) { + return (UCPTrieValueWidth)trie->valueWidth; +} + +U_CAPI int32_t U_EXPORT2 +ucptrie_internalSmallIndex(const UCPTrie *trie, UChar32 c) { + int32_t i1 = c >> UCPTRIE_SHIFT_1; + if (trie->type == UCPTRIE_TYPE_FAST) { + U_ASSERT(0xffff < c && c < trie->highStart); + i1 += UCPTRIE_BMP_INDEX_LENGTH - UCPTRIE_OMITTED_BMP_INDEX_1_LENGTH; + } else { + U_ASSERT((uint32_t)c < (uint32_t)trie->highStart && trie->highStart > UCPTRIE_SMALL_LIMIT); + i1 += UCPTRIE_SMALL_INDEX_LENGTH; + } + int32_t i3Block = trie->index[ + (int32_t)trie->index[i1] + ((c >> UCPTRIE_SHIFT_2) & UCPTRIE_INDEX_2_MASK)]; + int32_t i3 = (c >> UCPTRIE_SHIFT_3) & UCPTRIE_INDEX_3_MASK; + int32_t dataBlock; + if ((i3Block & 0x8000) == 0) { + // 16-bit indexes + dataBlock = trie->index[i3Block + i3]; + } else { + // 18-bit indexes stored in groups of 9 entries per 8 indexes. + i3Block = (i3Block & 0x7fff) + (i3 & ~7) + (i3 >> 3); + i3 &= 7; + dataBlock = ((int32_t)trie->index[i3Block++] << (2 + (2 * i3))) & 0x30000; + dataBlock |= trie->index[i3Block + i3]; + } + return dataBlock + (c & UCPTRIE_SMALL_DATA_MASK); +} + +U_CAPI int32_t U_EXPORT2 +ucptrie_internalSmallU8Index(const UCPTrie *trie, int32_t lt1, uint8_t t2, uint8_t t3) { + UChar32 c = (lt1 << 12) | (t2 << 6) | t3; + if (c >= trie->highStart) { + // Possible because the UTF-8 macro compares with shifted12HighStart which may be higher. + return trie->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET; + } + return ucptrie_internalSmallIndex(trie, c); +} + +U_CAPI int32_t U_EXPORT2 +ucptrie_internalU8PrevIndex(const UCPTrie *trie, UChar32 c, + const uint8_t *start, const uint8_t *src) { + int32_t i, length; + // Support 64-bit pointers by avoiding cast of arbitrary difference. + if ((src - start) <= 7) { + i = length = (int32_t)(src - start); + } else { + i = length = 7; + start = src - 7; + } + c = utf8_prevCharSafeBody(start, 0, &i, c, -1); + i = length - i; // Number of bytes read backward from src. + int32_t idx = _UCPTRIE_CP_INDEX(trie, 0xffff, c); + return (idx << 3) | i; +} + +namespace { + +inline uint32_t getValue(UCPTrieData data, UCPTrieValueWidth valueWidth, int32_t dataIndex) { + switch (valueWidth) { + case UCPTRIE_VALUE_BITS_16: + return data.ptr16[dataIndex]; + case UCPTRIE_VALUE_BITS_32: + return data.ptr32[dataIndex]; + case UCPTRIE_VALUE_BITS_8: + return data.ptr8[dataIndex]; + default: + // Unreachable if the trie is properly initialized. + return 0xffffffff; + } +} + +} // namespace + +U_CAPI uint32_t U_EXPORT2 +ucptrie_get(const UCPTrie *trie, UChar32 c) { + int32_t dataIndex; + if ((uint32_t)c <= 0x7f) { + // linear ASCII + dataIndex = c; + } else { + UChar32 fastMax = trie->type == UCPTRIE_TYPE_FAST ? 0xffff : UCPTRIE_SMALL_MAX; + dataIndex = _UCPTRIE_CP_INDEX(trie, fastMax, c); + } + return getValue(trie->data, (UCPTrieValueWidth)trie->valueWidth, dataIndex); +} + +namespace { + +constexpr int32_t MAX_UNICODE = 0x10ffff; + +inline uint32_t maybeFilterValue(uint32_t value, uint32_t trieNullValue, uint32_t nullValue, + UCPMapValueFilter *filter, const void *context) { + if (value == trieNullValue) { + value = nullValue; + } else if (filter != nullptr) { + value = filter(context, value); + } + return value; +} + +UChar32 getRange(const void *t, UChar32 start, + UCPMapValueFilter *filter, const void *context, uint32_t *pValue) { + if ((uint32_t)start > MAX_UNICODE) { + return U_SENTINEL; + } + const UCPTrie *trie = reinterpret_cast(t); + UCPTrieValueWidth valueWidth = (UCPTrieValueWidth)trie->valueWidth; + if (start >= trie->highStart) { + if (pValue != nullptr) { + int32_t di = trie->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET; + uint32_t value = getValue(trie->data, valueWidth, di); + if (filter != nullptr) { value = filter(context, value); } + *pValue = value; + } + return MAX_UNICODE; + } + + uint32_t nullValue = trie->nullValue; + if (filter != nullptr) { nullValue = filter(context, nullValue); } + const uint16_t *index = trie->index; + + int32_t prevI3Block = -1; + int32_t prevBlock = -1; + UChar32 c = start; + uint32_t trieValue, value = nullValue; + bool haveValue = false; + do { + int32_t i3Block; + int32_t i3; + int32_t i3BlockLength; + int32_t dataBlockLength; + if (c <= 0xffff && (trie->type == UCPTRIE_TYPE_FAST || c <= UCPTRIE_SMALL_MAX)) { + i3Block = 0; + i3 = c >> UCPTRIE_FAST_SHIFT; + i3BlockLength = trie->type == UCPTRIE_TYPE_FAST ? + UCPTRIE_BMP_INDEX_LENGTH : UCPTRIE_SMALL_INDEX_LENGTH; + dataBlockLength = UCPTRIE_FAST_DATA_BLOCK_LENGTH; + } else { + // Use the multi-stage index. + int32_t i1 = c >> UCPTRIE_SHIFT_1; + if (trie->type == UCPTRIE_TYPE_FAST) { + U_ASSERT(0xffff < c && c < trie->highStart); + i1 += UCPTRIE_BMP_INDEX_LENGTH - UCPTRIE_OMITTED_BMP_INDEX_1_LENGTH; + } else { + U_ASSERT(c < trie->highStart && trie->highStart > UCPTRIE_SMALL_LIMIT); + i1 += UCPTRIE_SMALL_INDEX_LENGTH; + } + i3Block = trie->index[ + (int32_t)trie->index[i1] + ((c >> UCPTRIE_SHIFT_2) & UCPTRIE_INDEX_2_MASK)]; + if (i3Block == prevI3Block && (c - start) >= UCPTRIE_CP_PER_INDEX_2_ENTRY) { + // The index-3 block is the same as the previous one, and filled with value. + U_ASSERT((c & (UCPTRIE_CP_PER_INDEX_2_ENTRY - 1)) == 0); + c += UCPTRIE_CP_PER_INDEX_2_ENTRY; + continue; + } + prevI3Block = i3Block; + if (i3Block == trie->index3NullOffset) { + // This is the index-3 null block. + if (haveValue) { + if (nullValue != value) { + return c - 1; + } + } else { + trieValue = trie->nullValue; + value = nullValue; + if (pValue != nullptr) { *pValue = nullValue; } + haveValue = true; + } + prevBlock = trie->dataNullOffset; + c = (c + UCPTRIE_CP_PER_INDEX_2_ENTRY) & ~(UCPTRIE_CP_PER_INDEX_2_ENTRY - 1); + continue; + } + i3 = (c >> UCPTRIE_SHIFT_3) & UCPTRIE_INDEX_3_MASK; + i3BlockLength = UCPTRIE_INDEX_3_BLOCK_LENGTH; + dataBlockLength = UCPTRIE_SMALL_DATA_BLOCK_LENGTH; + } + // Enumerate data blocks for one index-3 block. + do { + int32_t block; + if ((i3Block & 0x8000) == 0) { + block = index[i3Block + i3]; + } else { + // 18-bit indexes stored in groups of 9 entries per 8 indexes. + int32_t group = (i3Block & 0x7fff) + (i3 & ~7) + (i3 >> 3); + int32_t gi = i3 & 7; + block = ((int32_t)index[group++] << (2 + (2 * gi))) & 0x30000; + block |= index[group + gi]; + } + if (block == prevBlock && (c - start) >= dataBlockLength) { + // The block is the same as the previous one, and filled with value. + U_ASSERT((c & (dataBlockLength - 1)) == 0); + c += dataBlockLength; + } else { + int32_t dataMask = dataBlockLength - 1; + prevBlock = block; + if (block == trie->dataNullOffset) { + // This is the data null block. + if (haveValue) { + if (nullValue != value) { + return c - 1; + } + } else { + trieValue = trie->nullValue; + value = nullValue; + if (pValue != nullptr) { *pValue = nullValue; } + haveValue = true; + } + c = (c + dataBlockLength) & ~dataMask; + } else { + int32_t di = block + (c & dataMask); + uint32_t trieValue2 = getValue(trie->data, valueWidth, di); + if (haveValue) { + if (trieValue2 != trieValue) { + if (filter == nullptr || + maybeFilterValue(trieValue2, trie->nullValue, nullValue, + filter, context) != value) { + return c - 1; + } + trieValue = trieValue2; // may or may not help + } + } else { + trieValue = trieValue2; + value = maybeFilterValue(trieValue2, trie->nullValue, nullValue, + filter, context); + if (pValue != nullptr) { *pValue = value; } + haveValue = true; + } + while ((++c & dataMask) != 0) { + trieValue2 = getValue(trie->data, valueWidth, ++di); + if (trieValue2 != trieValue) { + if (filter == nullptr || + maybeFilterValue(trieValue2, trie->nullValue, nullValue, + filter, context) != value) { + return c - 1; + } + trieValue = trieValue2; // may or may not help + } + } + } + } + } while (++i3 < i3BlockLength); + } while (c < trie->highStart); + U_ASSERT(haveValue); + int32_t di = trie->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET; + uint32_t highValue = getValue(trie->data, valueWidth, di); + if (maybeFilterValue(highValue, trie->nullValue, nullValue, + filter, context) != value) { + return c - 1; + } else { + return MAX_UNICODE; + } +} + +} // namespace + +U_CFUNC UChar32 +ucptrie_internalGetRange(UCPTrieGetRange *getRange, + const void *trie, UChar32 start, + UCPMapRangeOption option, uint32_t surrogateValue, + UCPMapValueFilter *filter, const void *context, uint32_t *pValue) { + if (option == UCPMAP_RANGE_NORMAL) { + return getRange(trie, start, filter, context, pValue); + } + uint32_t value; + if (pValue == nullptr) { + // We need to examine the range value even if the caller does not want it. + pValue = &value; + } + UChar32 surrEnd = option == UCPMAP_RANGE_FIXED_ALL_SURROGATES ? 0xdfff : 0xdbff; + UChar32 end = getRange(trie, start, filter, context, pValue); + if (end < 0xd7ff || start > surrEnd) { + return end; + } + // The range overlaps with surrogates, or ends just before the first one. + if (*pValue == surrogateValue) { + if (end >= surrEnd) { + // Surrogates followed by a non-surrogateValue range, + // or surrogates are part of a larger surrogateValue range. + return end; + } + } else { + if (start <= 0xd7ff) { + return 0xd7ff; // Non-surrogateValue range ends before surrogateValue surrogates. + } + // Start is a surrogate with a non-surrogateValue code *unit* value. + // Return a surrogateValue code *point* range. + *pValue = surrogateValue; + if (end > surrEnd) { + return surrEnd; // Surrogate range ends before non-surrogateValue rest of range. + } + } + // See if the surrogateValue surrogate range can be merged with + // an immediately following range. + uint32_t value2; + UChar32 end2 = getRange(trie, surrEnd + 1, filter, context, &value2); + if (value2 == surrogateValue) { + return end2; + } + return surrEnd; +} + +U_CAPI UChar32 U_EXPORT2 +ucptrie_getRange(const UCPTrie *trie, UChar32 start, + UCPMapRangeOption option, uint32_t surrogateValue, + UCPMapValueFilter *filter, const void *context, uint32_t *pValue) { + return ucptrie_internalGetRange(getRange, trie, start, + option, surrogateValue, + filter, context, pValue); +} + +U_CAPI int32_t U_EXPORT2 +ucptrie_toBinary(const UCPTrie *trie, + void *data, int32_t capacity, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return 0; + } + + UCPTrieType type = (UCPTrieType)trie->type; + UCPTrieValueWidth valueWidth = (UCPTrieValueWidth)trie->valueWidth; + if (type < UCPTRIE_TYPE_FAST || UCPTRIE_TYPE_SMALL < type || + valueWidth < UCPTRIE_VALUE_BITS_16 || UCPTRIE_VALUE_BITS_8 < valueWidth || + capacity < 0 || + (capacity > 0 && (data == nullptr || (U_POINTER_MASK_LSB(data, 3) != 0)))) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + int32_t length = (int32_t)sizeof(UCPTrieHeader) + trie->indexLength * 2; + switch (valueWidth) { + case UCPTRIE_VALUE_BITS_16: + length += trie->dataLength * 2; + break; + case UCPTRIE_VALUE_BITS_32: + length += trie->dataLength * 4; + break; + case UCPTRIE_VALUE_BITS_8: + length += trie->dataLength; + break; + default: + // unreachable + break; + } + if (capacity < length) { + *pErrorCode = U_BUFFER_OVERFLOW_ERROR; + return length; + } + + char *bytes = (char *)data; + UCPTrieHeader *header = (UCPTrieHeader *)bytes; + header->signature = UCPTRIE_SIG; // "Tri3" + header->options = (uint16_t)( + ((trie->dataLength & 0xf0000) >> 4) | + ((trie->dataNullOffset & 0xf0000) >> 8) | + (trie->type << 6) | + valueWidth); + header->indexLength = (uint16_t)trie->indexLength; + header->dataLength = (uint16_t)trie->dataLength; + header->index3NullOffset = trie->index3NullOffset; + header->dataNullOffset = (uint16_t)trie->dataNullOffset; + header->shiftedHighStart = trie->highStart >> UCPTRIE_SHIFT_2; + bytes += sizeof(UCPTrieHeader); + + uprv_memcpy(bytes, trie->index, trie->indexLength * 2); + bytes += trie->indexLength * 2; + + switch (valueWidth) { + case UCPTRIE_VALUE_BITS_16: + uprv_memcpy(bytes, trie->data.ptr16, trie->dataLength * 2); + break; + case UCPTRIE_VALUE_BITS_32: + uprv_memcpy(bytes, trie->data.ptr32, trie->dataLength * 4); + break; + case UCPTRIE_VALUE_BITS_8: + uprv_memcpy(bytes, trie->data.ptr8, trie->dataLength); + break; + default: + // unreachable + break; + } + return length; +} + +namespace { + +#ifdef UCPTRIE_DEBUG +long countNull(const UCPTrie *trie) { + uint32_t nullValue=trie->nullValue; + int32_t length=trie->dataLength; + long count=0; + switch (trie->valueWidth) { + case UCPTRIE_VALUE_BITS_16: + for(int32_t i=0; idata.ptr16[i]==nullValue) { ++count; } + } + break; + case UCPTRIE_VALUE_BITS_32: + for(int32_t i=0; idata.ptr32[i]==nullValue) { ++count; } + } + break; + case UCPTRIE_VALUE_BITS_8: + for(int32_t i=0; idata.ptr8[i]==nullValue) { ++count; } + } + break; + default: + // unreachable + break; + } + return count; +} + +U_CFUNC void +ucptrie_printLengths(const UCPTrie *trie, const char *which) { + long indexLength=trie->indexLength; + long dataLength=(long)trie->dataLength; + long totalLength=(long)sizeof(UCPTrieHeader)+indexLength*2+ + dataLength*(trie->valueWidth==UCPTRIE_VALUE_BITS_16 ? 2 : + trie->valueWidth==UCPTRIE_VALUE_BITS_32 ? 4 : 1); + printf("**UCPTrieLengths(%s %s)** index:%6ld data:%6ld countNull:%6ld serialized:%6ld\n", + which, trie->name, indexLength, dataLength, countNull(trie), totalLength); +} +#endif + +} // namespace + +// UCPMap ---- +// Initially, this is the same as UCPTrie. This may well change. + +U_CAPI uint32_t U_EXPORT2 +ucpmap_get(const UCPMap *map, UChar32 c) { + return ucptrie_get(reinterpret_cast(map), c); +} + +U_CAPI UChar32 U_EXPORT2 +ucpmap_getRange(const UCPMap *map, UChar32 start, + UCPMapRangeOption option, uint32_t surrogateValue, + UCPMapValueFilter *filter, const void *context, uint32_t *pValue) { + return ucptrie_getRange(reinterpret_cast(map), start, + option, surrogateValue, + filter, context, pValue); +} diff --git a/intl/icu/source/common/ucptrie_impl.h b/intl/icu/source/common/ucptrie_impl.h new file mode 100644 index 0000000000..a7a80a8f08 --- /dev/null +++ b/intl/icu/source/common/ucptrie_impl.h @@ -0,0 +1,285 @@ +// © 2017 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// ucptrie_impl.h (modified from utrie2_impl.h) +// created: 2017dec29 Markus W. Scherer + +#ifndef __UCPTRIE_IMPL_H__ +#define __UCPTRIE_IMPL_H__ + +#include "unicode/ucptrie.h" +#ifdef UCPTRIE_DEBUG +#include "unicode/umutablecptrie.h" +#endif + +// UCPTrie signature values, in platform endianness and opposite endianness. +// The UCPTrie signature ASCII byte values spell "Tri3". +#define UCPTRIE_SIG 0x54726933 +#define UCPTRIE_OE_SIG 0x33697254 + +/** + * Header data for the binary, memory-mappable representation of a UCPTrie/CodePointTrie. + * @internal + */ +struct UCPTrieHeader { + /** "Tri3" in big-endian US-ASCII (0x54726933) */ + uint32_t signature; + + /** + * Options bit field: + * Bits 15..12: Data length bits 19..16. + * Bits 11..8: Data null block offset bits 19..16. + * Bits 7..6: UCPTrieType + * Bits 5..3: Reserved (0). + * Bits 2..0: UCPTrieValueWidth + */ + uint16_t options; + + /** Total length of the index tables. */ + uint16_t indexLength; + + /** Data length bits 15..0. */ + uint16_t dataLength; + + /** Index-3 null block offset, 0x7fff or 0xffff if none. */ + uint16_t index3NullOffset; + + /** Data null block offset bits 15..0, 0xfffff if none. */ + uint16_t dataNullOffset; + + /** + * First code point of the single-value range ending with U+10ffff, + * rounded up and then shifted right by UCPTRIE_SHIFT_2. + */ + uint16_t shiftedHighStart; +}; + +// Constants for use with UCPTrieHeader.options. +constexpr uint16_t UCPTRIE_OPTIONS_DATA_LENGTH_MASK = 0xf000; +constexpr uint16_t UCPTRIE_OPTIONS_DATA_NULL_OFFSET_MASK = 0xf00; +constexpr uint16_t UCPTRIE_OPTIONS_RESERVED_MASK = 0x38; +constexpr uint16_t UCPTRIE_OPTIONS_VALUE_BITS_MASK = 7; + +/** + * Value for index3NullOffset which indicates that there is no index-3 null block. + * Bit 15 is unused for this value because this bit is used if the index-3 contains + * 18-bit indexes. + */ +constexpr int32_t UCPTRIE_NO_INDEX3_NULL_OFFSET = 0x7fff; +constexpr int32_t UCPTRIE_NO_DATA_NULL_OFFSET = 0xfffff; + +// Internal constants. + +/** The length of the BMP index table. 1024=0x400 */ +constexpr int32_t UCPTRIE_BMP_INDEX_LENGTH = 0x10000 >> UCPTRIE_FAST_SHIFT; + +constexpr int32_t UCPTRIE_SMALL_LIMIT = 0x1000; +constexpr int32_t UCPTRIE_SMALL_INDEX_LENGTH = UCPTRIE_SMALL_LIMIT >> UCPTRIE_FAST_SHIFT; + +/** Shift size for getting the index-3 table offset. */ +constexpr int32_t UCPTRIE_SHIFT_3 = 4; + +/** Shift size for getting the index-2 table offset. */ +constexpr int32_t UCPTRIE_SHIFT_2 = 5 + UCPTRIE_SHIFT_3; + +/** Shift size for getting the index-1 table offset. */ +constexpr int32_t UCPTRIE_SHIFT_1 = 5 + UCPTRIE_SHIFT_2; + +/** + * Difference between two shift sizes, + * for getting an index-2 offset from an index-3 offset. 5=9-4 + */ +constexpr int32_t UCPTRIE_SHIFT_2_3 = UCPTRIE_SHIFT_2 - UCPTRIE_SHIFT_3; + +/** + * Difference between two shift sizes, + * for getting an index-1 offset from an index-2 offset. 5=14-9 + */ +constexpr int32_t UCPTRIE_SHIFT_1_2 = UCPTRIE_SHIFT_1 - UCPTRIE_SHIFT_2; + +/** + * Number of index-1 entries for the BMP. (4) + * This part of the index-1 table is omitted from the serialized form. + */ +constexpr int32_t UCPTRIE_OMITTED_BMP_INDEX_1_LENGTH = 0x10000 >> UCPTRIE_SHIFT_1; + +/** Number of entries in an index-2 block. 32=0x20 */ +constexpr int32_t UCPTRIE_INDEX_2_BLOCK_LENGTH = 1 << UCPTRIE_SHIFT_1_2; + +/** Mask for getting the lower bits for the in-index-2-block offset. */ +constexpr int32_t UCPTRIE_INDEX_2_MASK = UCPTRIE_INDEX_2_BLOCK_LENGTH - 1; + +/** Number of code points per index-2 table entry. 512=0x200 */ +constexpr int32_t UCPTRIE_CP_PER_INDEX_2_ENTRY = 1 << UCPTRIE_SHIFT_2; + +/** Number of entries in an index-3 block. 32=0x20 */ +constexpr int32_t UCPTRIE_INDEX_3_BLOCK_LENGTH = 1 << UCPTRIE_SHIFT_2_3; + +/** Mask for getting the lower bits for the in-index-3-block offset. */ +constexpr int32_t UCPTRIE_INDEX_3_MASK = UCPTRIE_INDEX_3_BLOCK_LENGTH - 1; + +/** Number of entries in a small data block. 16=0x10 */ +constexpr int32_t UCPTRIE_SMALL_DATA_BLOCK_LENGTH = 1 << UCPTRIE_SHIFT_3; + +/** Mask for getting the lower bits for the in-small-data-block offset. */ +constexpr int32_t UCPTRIE_SMALL_DATA_MASK = UCPTRIE_SMALL_DATA_BLOCK_LENGTH - 1; + + +typedef UChar32 +UCPTrieGetRange(const void *trie, UChar32 start, + UCPMapValueFilter *filter, const void *context, uint32_t *pValue); + +U_CFUNC UChar32 +ucptrie_internalGetRange(UCPTrieGetRange *getRange, + const void *trie, UChar32 start, + UCPMapRangeOption option, uint32_t surrogateValue, + UCPMapValueFilter *filter, const void *context, uint32_t *pValue); + +#ifdef UCPTRIE_DEBUG +U_CFUNC void +ucptrie_printLengths(const UCPTrie *trie, const char *which); + +U_CFUNC void umutablecptrie_setName(UMutableCPTrie *builder, const char *name); +#endif + +/* + * Format of the binary, memory-mappable representation of a UCPTrie/CodePointTrie. + * For overview information see https://icu.unicode.org/design/struct/utrie + * + * The binary trie data should be 32-bit-aligned. + * The overall layout is: + * + * UCPTrieHeader header; -- 16 bytes, see struct definition above + * uint16_t index[header.indexLength]; + * uintXY_t data[header.dataLength]; + * + * The trie data array is an array of uint16_t, uint32_t, or uint8_t, + * specified via the UCPTrieValueWidth when building the trie. + * The data array is 32-bit-aligned for uint32_t, otherwise 16-bit-aligned. + * The overall length of the trie data is a multiple of 4 bytes. + * (Padding is added at the end of the index array and/or near the end of the data array as needed.) + * + * The length of the data array (dataLength) is stored as an integer split across two fields + * of the header struct (high bits in header.options). + * + * The trie type can be "fast" or "small" which determines the index structure, + * specified via the UCPTrieType when building the trie. + * + * The type and valueWidth are stored in the header.options. + * There are reserved type and valueWidth values, and reserved header.options bits. + * They could be used in future format extensions. + * Code reading the trie structure must fail with an error when unknown values or options are set. + * + * Values for ASCII character (U+0000..U+007F) can always be found at the start of the data array. + * + * Values for code points below a type-specific fast-indexing limit are found via two-stage lookup. + * For a "fast" trie, the limit is the BMP/supplementary boundary at U+10000. + * For a "small" trie, the limit is UCPTRIE_SMALL_MAX+1=U+1000. + * + * All code points in the range highStart..U+10FFFF map to a single highValue + * which is stored at the second-to-last position of the data array. + * (See UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET.) + * The highStart value is header.shiftedHighStart<>UCPTRIE_SHIFT_1. + * (For 0x100000 supplementary code points U+10000..U+10ffff.) + * + * After this index-1 table follow the variable-length index-3 and index-2 tables. + * + * The supplementary index tables are omitted completely + * if there is only BMP data (highStart<=U+10000). + * + * For a "small" trie: + * + * The index array starts with a fast-index table for lookup of code points U+0000..U+0FFF. + * + * The "supplementary" index tables are always stored. + * The index-1 table starts from U+0000, its maximum length is 68=0x44=0x110000>>UCPTRIE_SHIFT_1. + * + * For both trie types: + * + * The last index-2 block may be a partial block, storing indexes only for code points + * below highStart. + * + * Lookup for ASCII code point c: + * + * Linear access from the start of the data array. + * + * value = data[c]; + * + * Lookup for fast-range code point c: + * + * Shift the code point right by UCPTRIE_FAST_SHIFT=6 bits, + * fetch the index array value at that offset, + * add the lower code point bits, index into the data array. + * + * value = data[index[c>>6] + (c&0x3f)]; + * + * (This works for ASCII as well.) + * + * Lookup for small-range code point c below highStart: + * + * Split the code point into four bit fields using several sets of shifts & masks + * to read consecutive values from the index-1, index-2, index-3 and data tables. + * + * If all of the data block offsets in an index-3 block fit within 16 bits (up to 0xffff), + * then the data block offsets are stored directly as uint16_t. + * + * Otherwise (this is very unusual but possible), the index-2 entry for the index-3 block + * has bit 15 (0x8000) set, and each set of 8 index-3 entries is preceded by + * an additional uint16_t word. Data block offsets are 18 bits wide, with the top 2 bits stored + * in the additional word. + * + * See ucptrie_internalSmallIndex() for details. + * + * (In a "small" trie, this works for ASCII and below-fast_limit code points as well.) + * + * Compaction: + * + * Multiple code point ranges ("blocks") that are aligned on certain boundaries + * (determined by the shifting/bit fields of code points) and + * map to the same data values normally share a single subsequence of the data array. + * Data blocks can also overlap partially. + * (Depending on the builder code finding duplicate and overlapping blocks.) + * + * Iteration over same-value ranges: + * + * Range iteration (ucptrie_getRange()) walks the structure from a start code point + * until some code point is found that maps to a different value; + * the end of the returned range is just before that. + * + * The header.dataNullOffset (split across two header fields, high bits in header.options) + * is the offset of a widely shared data block filled with one single value. + * It helps quickly skip over large ranges of data with that value. + * The builder must ensure that if the start of any data block (fast or small) + * matches the dataNullOffset, then the whole block must be filled with the null value. + * Special care must be taken if there is no fast null data block + * but a small one, which is shorter, and it matches the *start* of some fast data block. + * + * Similarly, the header.index3NullOffset is the index-array offset of an index-3 block + * where all index entries point to the dataNullOffset. + * If there is no such data or index-3 block, then these offsets are set to + * values that cannot be reached (data offset out of range/reserved index offset), + * normally UCPTRIE_NO_DATA_NULL_OFFSET or UCPTRIE_NO_INDEX3_NULL_OFFSET respectively. + */ + +#endif diff --git a/intl/icu/source/common/ucurr.cpp b/intl/icu/source/common/ucurr.cpp new file mode 100644 index 0000000000..ffca8aac5f --- /dev/null +++ b/intl/icu/source/common/ucurr.cpp @@ -0,0 +1,2727 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2002-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_FORMATTING + +#include "unicode/ucurr.h" +#include "unicode/locid.h" +#include "unicode/ures.h" +#include "unicode/ustring.h" +#include "unicode/parsepos.h" +#include "unicode/uniset.h" +#include "unicode/usetiter.h" +#include "unicode/utf16.h" +#include "ustr_imp.h" +#include "charstr.h" +#include "cmemory.h" +#include "cstring.h" +#include "static_unicode_sets.h" +#include "uassert.h" +#include "umutex.h" +#include "ucln_cmn.h" +#include "uenumimp.h" +#include "uhash.h" +#include "hash.h" +#include "uinvchar.h" +#include "uresimp.h" +#include "ulist.h" +#include "uresimp.h" +#include "ureslocs.h" +#include "ulocimp.h" + +using namespace icu; + +//#define UCURR_DEBUG_EQUIV 1 +#ifdef UCURR_DEBUG_EQUIV +#include "stdio.h" +#endif +//#define UCURR_DEBUG 1 +#ifdef UCURR_DEBUG +#include "stdio.h" +#endif + +typedef struct IsoCodeEntry { + const char16_t *isoCode; /* const because it's a reference to a resource bundle string. */ + UDate from; + UDate to; +} IsoCodeEntry; + +//------------------------------------------------------------ +// Constants + +// Default currency meta data of last resort. We try to use the +// defaults encoded in the meta data resource bundle. If there is a +// configuration/build error and these are not available, we use these +// hard-coded defaults (which should be identical). +static const int32_t LAST_RESORT_DATA[] = { 2, 0, 2, 0 }; + +// POW10[i] = 10^i, i=0..MAX_POW10 +static const int32_t POW10[] = { 1, 10, 100, 1000, 10000, 100000, + 1000000, 10000000, 100000000, 1000000000 }; + +static const int32_t MAX_POW10 = UPRV_LENGTHOF(POW10) - 1; + +#define ISO_CURRENCY_CODE_LENGTH 3 + +//------------------------------------------------------------ +// Resource tags +// + +static const char CURRENCY_DATA[] = "supplementalData"; +// Tag for meta-data, in root. +static const char CURRENCY_META[] = "CurrencyMeta"; + +// Tag for map from countries to currencies, in root. +static const char CURRENCY_MAP[] = "CurrencyMap"; + +// Tag for default meta-data, in CURRENCY_META +static const char DEFAULT_META[] = "DEFAULT"; + +// Variant delimiter +static const char VAR_DELIM = '_'; + +// Tag for localized display names (symbols) of currencies +static const char CURRENCIES[] = "Currencies"; +static const char CURRENCIES_NARROW[] = "Currencies%narrow"; +static const char CURRENCIES_FORMAL[] = "Currencies%formal"; +static const char CURRENCIES_VARIANT[] = "Currencies%variant"; +static const char CURRENCYPLURALS[] = "CurrencyPlurals"; + +// ISO codes mapping table +static const UHashtable* gIsoCodes = nullptr; +static icu::UInitOnce gIsoCodesInitOnce {}; + +// Currency symbol equivalances +static const icu::Hashtable* gCurrSymbolsEquiv = nullptr; +static icu::UInitOnce gCurrSymbolsEquivInitOnce {}; + +U_NAMESPACE_BEGIN + +// EquivIterator iterates over all strings that are equivalent to a given +// string, s. Note that EquivIterator will never yield s itself. +class EquivIterator : public icu::UMemory { +public: + // Constructor. hash stores the equivalence relationships; s is the string + // for which we find equivalent strings. + inline EquivIterator(const icu::Hashtable& hash, const icu::UnicodeString& s) + : _hash(hash) { + _start = _current = &s; + } + inline ~EquivIterator() { } + + // next returns the next equivalent string or nullptr if there are no more. + // If s has no equivalent strings, next returns nullptr on the first call. + const icu::UnicodeString *next(); +private: + const icu::Hashtable& _hash; + const icu::UnicodeString* _start; + const icu::UnicodeString* _current; +}; + +const icu::UnicodeString * +EquivIterator::next() { + const icu::UnicodeString* _next = (const icu::UnicodeString*) _hash.get(*_current); + if (_next == nullptr) { + U_ASSERT(_current == _start); + return nullptr; + } + if (*_next == *_start) { + return nullptr; + } + _current = _next; + return _next; +} + +U_NAMESPACE_END + +// makeEquivalent makes lhs and rhs equivalent by updating the equivalence +// relations in hash accordingly. +static void makeEquivalent( + const icu::UnicodeString &lhs, + const icu::UnicodeString &rhs, + icu::Hashtable* hash, UErrorCode &status) { + if (U_FAILURE(status)) { + return; + } + if (lhs == rhs) { + // already equivalent + return; + } + icu::EquivIterator leftIter(*hash, lhs); + icu::EquivIterator rightIter(*hash, rhs); + const icu::UnicodeString *firstLeft = leftIter.next(); + const icu::UnicodeString *firstRight = rightIter.next(); + const icu::UnicodeString *nextLeft = firstLeft; + const icu::UnicodeString *nextRight = firstRight; + while (nextLeft != nullptr && nextRight != nullptr) { + if (*nextLeft == rhs || *nextRight == lhs) { + // Already equivalent + return; + } + nextLeft = leftIter.next(); + nextRight = rightIter.next(); + } + // Not equivalent. Must join. + icu::UnicodeString *newFirstLeft; + icu::UnicodeString *newFirstRight; + if (firstRight == nullptr && firstLeft == nullptr) { + // Neither lhs or rhs belong to an equivalence circle, so we form + // a new equivalnce circle of just lhs and rhs. + newFirstLeft = new icu::UnicodeString(rhs); + newFirstRight = new icu::UnicodeString(lhs); + } else if (firstRight == nullptr) { + // lhs belongs to an equivalence circle, but rhs does not, so we link + // rhs into lhs' circle. + newFirstLeft = new icu::UnicodeString(rhs); + newFirstRight = new icu::UnicodeString(*firstLeft); + } else if (firstLeft == nullptr) { + // rhs belongs to an equivlance circle, but lhs does not, so we link + // lhs into rhs' circle. + newFirstLeft = new icu::UnicodeString(*firstRight); + newFirstRight = new icu::UnicodeString(lhs); + } else { + // Both lhs and rhs belong to different equivalnce circles. We link + // them together to form one single, larger equivalnce circle. + newFirstLeft = new icu::UnicodeString(*firstRight); + newFirstRight = new icu::UnicodeString(*firstLeft); + } + if (newFirstLeft == nullptr || newFirstRight == nullptr) { + delete newFirstLeft; + delete newFirstRight; + status = U_MEMORY_ALLOCATION_ERROR; + return; + } + hash->put(lhs, (void *) newFirstLeft, status); + hash->put(rhs, (void *) newFirstRight, status); +} + +// countEquivalent counts how many strings are equivalent to s. +// hash stores all the equivalnce relations. +// countEquivalent does not include s itself in the count. +static int32_t countEquivalent(const icu::Hashtable &hash, const icu::UnicodeString &s) { + int32_t result = 0; + icu::EquivIterator iter(hash, s); + while (iter.next() != nullptr) { + ++result; + } +#ifdef UCURR_DEBUG_EQUIV + { + char tmp[200]; + s.extract(0,s.length(),tmp, "UTF-8"); + printf("CountEquivalent('%s') = %d\n", tmp, result); + } +#endif + return result; +} + +static const icu::Hashtable* getCurrSymbolsEquiv(); + +//------------------------------------------------------------ +// Code + +/** + * Cleanup callback func + */ +static UBool U_CALLCONV +isoCodes_cleanup() +{ + if (gIsoCodes != nullptr) { + uhash_close(const_cast(gIsoCodes)); + gIsoCodes = nullptr; + } + gIsoCodesInitOnce.reset(); + return true; +} + +/** + * Cleanup callback func + */ +static UBool U_CALLCONV +currSymbolsEquiv_cleanup() +{ + delete const_cast(gCurrSymbolsEquiv); + gCurrSymbolsEquiv = nullptr; + gCurrSymbolsEquivInitOnce.reset(); + return true; +} + +/** + * Deleter for IsoCodeEntry + */ +static void U_CALLCONV +deleteIsoCodeEntry(void *obj) { + IsoCodeEntry *entry = (IsoCodeEntry*)obj; + uprv_free(entry); +} + +/** + * Deleter for gCurrSymbolsEquiv. + */ +static void U_CALLCONV +deleteUnicode(void *obj) { + icu::UnicodeString *entry = (icu::UnicodeString*)obj; + delete entry; +} + +/** + * Unfortunately, we have to convert the char16_t* currency code to char* + * to use it as a resource key. + */ +static inline char* +myUCharsToChars(char* resultOfLen4, const char16_t* currency) { + u_UCharsToChars(currency, resultOfLen4, ISO_CURRENCY_CODE_LENGTH); + resultOfLen4[ISO_CURRENCY_CODE_LENGTH] = 0; + return resultOfLen4; +} + +/** + * Internal function to look up currency data. Result is an array of + * four integers. The first is the fraction digits. The second is the + * rounding increment, or 0 if none. The rounding increment is in + * units of 10^(-fraction_digits). The third and fourth are the same + * except that they are those used in cash transactions ( cashDigits + * and cashRounding ). + */ +static const int32_t* +_findMetaData(const char16_t* currency, UErrorCode& ec) { + + if (currency == 0 || *currency == 0) { + if (U_SUCCESS(ec)) { + ec = U_ILLEGAL_ARGUMENT_ERROR; + } + return LAST_RESORT_DATA; + } + + // Get CurrencyMeta resource out of root locale file. [This may + // move out of the root locale file later; if it does, update this + // code.] + UResourceBundle* currencyData = ures_openDirect(U_ICUDATA_CURR, CURRENCY_DATA, &ec); + UResourceBundle* currencyMeta = ures_getByKey(currencyData, CURRENCY_META, currencyData, &ec); + + if (U_FAILURE(ec)) { + ures_close(currencyMeta); + // Config/build error; return hard-coded defaults + return LAST_RESORT_DATA; + } + + // Look up our currency, or if that's not available, then DEFAULT + char buf[ISO_CURRENCY_CODE_LENGTH+1]; + UErrorCode ec2 = U_ZERO_ERROR; // local error code: soft failure + UResourceBundle* rb = ures_getByKey(currencyMeta, myUCharsToChars(buf, currency), nullptr, &ec2); + if (U_FAILURE(ec2)) { + ures_close(rb); + rb = ures_getByKey(currencyMeta,DEFAULT_META, nullptr, &ec); + if (U_FAILURE(ec)) { + ures_close(currencyMeta); + ures_close(rb); + // Config/build error; return hard-coded defaults + return LAST_RESORT_DATA; + } + } + + int32_t len; + const int32_t *data = ures_getIntVector(rb, &len, &ec); + if (U_FAILURE(ec) || len != 4) { + // Config/build error; return hard-coded defaults + if (U_SUCCESS(ec)) { + ec = U_INVALID_FORMAT_ERROR; + } + ures_close(currencyMeta); + ures_close(rb); + return LAST_RESORT_DATA; + } + + ures_close(currencyMeta); + ures_close(rb); + return data; +} + +// ------------------------------------- + +static void +idForLocale(const char* locale, char* countryAndVariant, int capacity, UErrorCode* ec) +{ + ulocimp_getRegionForSupplementalData(locale, false, countryAndVariant, capacity, ec); +} + +// ------------------------------------------ +// +// Registration +// +//------------------------------------------- + +// don't use ICUService since we don't need fallback + +U_CDECL_BEGIN +static UBool U_CALLCONV currency_cleanup(); +U_CDECL_END + +#if !UCONFIG_NO_SERVICE +struct CReg; + +static UMutex gCRegLock; +static CReg* gCRegHead = 0; + +struct CReg : public icu::UMemory { + CReg *next; + char16_t iso[ISO_CURRENCY_CODE_LENGTH+1]; + char id[ULOC_FULLNAME_CAPACITY]; + + CReg(const char16_t* _iso, const char* _id) + : next(0) + { + int32_t len = (int32_t)uprv_strlen(_id); + if (len > (int32_t)(sizeof(id)-1)) { + len = (sizeof(id)-1); + } + uprv_strncpy(id, _id, len); + id[len] = 0; + u_memcpy(iso, _iso, ISO_CURRENCY_CODE_LENGTH); + iso[ISO_CURRENCY_CODE_LENGTH] = 0; + } + + static UCurrRegistryKey reg(const char16_t* _iso, const char* _id, UErrorCode* status) + { + if (status && U_SUCCESS(*status) && _iso && _id) { + CReg* n = new CReg(_iso, _id); + if (n) { + umtx_lock(&gCRegLock); + if (!gCRegHead) { + /* register for the first time */ + ucln_common_registerCleanup(UCLN_COMMON_CURRENCY, currency_cleanup); + } + n->next = gCRegHead; + gCRegHead = n; + umtx_unlock(&gCRegLock); + return n; + } + *status = U_MEMORY_ALLOCATION_ERROR; + } + return 0; + } + + static UBool unreg(UCurrRegistryKey key) { + UBool found = false; + umtx_lock(&gCRegLock); + + CReg** p = &gCRegHead; + while (*p) { + if (*p == key) { + *p = ((CReg*)key)->next; + delete (CReg*)key; + found = true; + break; + } + p = &((*p)->next); + } + + umtx_unlock(&gCRegLock); + return found; + } + + static const char16_t* get(const char* id) { + const char16_t* result = nullptr; + umtx_lock(&gCRegLock); + CReg* p = gCRegHead; + + /* register cleanup of the mutex */ + ucln_common_registerCleanup(UCLN_COMMON_CURRENCY, currency_cleanup); + while (p) { + if (uprv_strcmp(id, p->id) == 0) { + result = p->iso; + break; + } + p = p->next; + } + umtx_unlock(&gCRegLock); + return result; + } + + /* This doesn't need to be thread safe. It's for u_cleanup only. */ + static void cleanup() { + while (gCRegHead) { + CReg* n = gCRegHead; + gCRegHead = gCRegHead->next; + delete n; + } + } +}; + +// ------------------------------------- + +U_CAPI UCurrRegistryKey U_EXPORT2 +ucurr_register(const char16_t* isoCode, const char* locale, UErrorCode *status) +{ + if (status && U_SUCCESS(*status)) { + char id[ULOC_FULLNAME_CAPACITY]; + idForLocale(locale, id, sizeof(id), status); + return CReg::reg(isoCode, id, status); + } + return nullptr; +} + +// ------------------------------------- + +U_CAPI UBool U_EXPORT2 +ucurr_unregister(UCurrRegistryKey key, UErrorCode* status) +{ + if (status && U_SUCCESS(*status)) { + return CReg::unreg(key); + } + return false; +} +#endif /* UCONFIG_NO_SERVICE */ + +// ------------------------------------- + +/** + * Release all static memory held by currency. + */ +/*The declaration here is needed so currency_cleanup() + * can call this function. + */ +static UBool U_CALLCONV +currency_cache_cleanup(); + +U_CDECL_BEGIN +static UBool U_CALLCONV currency_cleanup() { +#if !UCONFIG_NO_SERVICE + CReg::cleanup(); +#endif + /* + * There might be some cached currency data or isoCodes data. + */ + currency_cache_cleanup(); + isoCodes_cleanup(); + currSymbolsEquiv_cleanup(); + + return true; +} +U_CDECL_END + +// ------------------------------------- + +U_CAPI int32_t U_EXPORT2 +ucurr_forLocale(const char* locale, + char16_t* buff, + int32_t buffCapacity, + UErrorCode* ec) { + if (U_FAILURE(*ec)) { return 0; } + if (buffCapacity < 0 || (buff == nullptr && buffCapacity > 0)) { + *ec = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + char currency[4]; // ISO currency codes are alpha3 codes. + UErrorCode localStatus = U_ZERO_ERROR; + int32_t resLen = uloc_getKeywordValue(locale, "currency", + currency, UPRV_LENGTHOF(currency), &localStatus); + if (U_SUCCESS(localStatus) && resLen == 3 && uprv_isInvariantString(currency, resLen)) { + if (resLen < buffCapacity) { + T_CString_toUpperCase(currency); + u_charsToUChars(currency, buff, resLen); + } + return u_terminateUChars(buff, buffCapacity, resLen, ec); + } + + // get country or country_variant in `id' + char id[ULOC_FULLNAME_CAPACITY]; + idForLocale(locale, id, UPRV_LENGTHOF(id), ec); + if (U_FAILURE(*ec)) { + return 0; + } + +#if !UCONFIG_NO_SERVICE + const char16_t* result = CReg::get(id); + if (result) { + if(buffCapacity > u_strlen(result)) { + u_strcpy(buff, result); + } + resLen = u_strlen(result); + return u_terminateUChars(buff, buffCapacity, resLen, ec); + } +#endif + // Remove variants, which is only needed for registration. + char *idDelim = uprv_strchr(id, VAR_DELIM); + if (idDelim) { + idDelim[0] = 0; + } + + const char16_t* s = nullptr; // Currency code from data file. + if (id[0] == 0) { + // No point looking in the data for an empty string. + // This is what we would get. + localStatus = U_MISSING_RESOURCE_ERROR; + } else { + // Look up the CurrencyMap element in the root bundle. + localStatus = U_ZERO_ERROR; + UResourceBundle *rb = ures_openDirect(U_ICUDATA_CURR, CURRENCY_DATA, &localStatus); + UResourceBundle *cm = ures_getByKey(rb, CURRENCY_MAP, rb, &localStatus); + UResourceBundle *countryArray = ures_getByKey(rb, id, cm, &localStatus); + // https://unicode-org.atlassian.net/browse/ICU-21997 + // Prefer to use currencies that are legal tender. + if (U_SUCCESS(localStatus)) { + int32_t arrayLength = ures_getSize(countryArray); + for (int32_t i = 0; i < arrayLength; ++i) { + LocalUResourceBundlePointer currencyReq( + ures_getByIndex(countryArray, i, nullptr, &localStatus)); + // The currency is legal tender if it is *not* marked with tender{"false"}. + UErrorCode tenderStatus = localStatus; + const char16_t *tender = + ures_getStringByKey(currencyReq.getAlias(), "tender", nullptr, &tenderStatus); + bool isTender = U_FAILURE(tenderStatus) || u_strcmp(tender, u"false") != 0; + if (!isTender && s != nullptr) { + // We already have a non-tender currency. Ignore all following non-tender ones. + continue; + } + // Fetch the currency code. + s = ures_getStringByKey(currencyReq.getAlias(), "id", &resLen, &localStatus); + if (isTender) { + break; + } + } + if (U_SUCCESS(localStatus) && s == nullptr) { + localStatus = U_MISSING_RESOURCE_ERROR; + } + } + ures_close(countryArray); + } + + if ((U_FAILURE(localStatus)) && strchr(id, '_') != 0) { + // We don't know about it. Check to see if we support the variant. + uloc_getParent(locale, id, UPRV_LENGTHOF(id), ec); + *ec = U_USING_FALLBACK_WARNING; + // TODO: Loop over the shortened id rather than recursing and + // looking again for a currency keyword. + return ucurr_forLocale(id, buff, buffCapacity, ec); + } + if (*ec == U_ZERO_ERROR || localStatus != U_ZERO_ERROR) { + // There is nothing to fallback to. Report the failure/warning if possible. + *ec = localStatus; + } + if (U_SUCCESS(*ec)) { + if(buffCapacity > resLen) { + u_strcpy(buff, s); + } + } + return u_terminateUChars(buff, buffCapacity, resLen, ec); +} + +// end registration + +/** + * Modify the given locale name by removing the rightmost _-delimited + * element. If there is none, empty the string ("" == root). + * NOTE: The string "root" is not recognized; do not use it. + * @return true if the fallback happened; false if locale is already + * root (""). + */ +static UBool fallback(char *loc) { + if (!*loc) { + return false; + } + UErrorCode status = U_ZERO_ERROR; + if (uprv_strcmp(loc, "en_GB") == 0) { + // HACK: See #13368. We need "en_GB" to fall back to "en_001" instead of "en" + // in order to consume the correct data strings. This hack will be removed + // when proper data sink loading is implemented here. + // NOTE: "001" adds 1 char over "GB". However, both call sites allocate + // arrays with length ULOC_FULLNAME_CAPACITY (plenty of room for en_001). + uprv_strcpy(loc + 3, "001"); + } else { + uloc_getParent(loc, loc, (int32_t)uprv_strlen(loc), &status); + } + /* + char *i = uprv_strrchr(loc, '_'); + if (i == nullptr) { + i = loc; + } + *i = 0; + */ + return true; +} + + +U_CAPI const char16_t* U_EXPORT2 +ucurr_getName(const char16_t* currency, + const char* locale, + UCurrNameStyle nameStyle, + UBool* isChoiceFormat, // fillin + int32_t* len, // fillin + UErrorCode* ec) { + + // Look up the Currencies resource for the given locale. The + // Currencies locale data looks like this: + //|en { + //| Currencies { + //| USD { "US$", "US Dollar" } + //| CHF { "Sw F", "Swiss Franc" } + //| INR { "=0#Rs|1#Re|1 4) { + *ec = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + // In the future, resource bundles may implement multi-level + // fallback. That is, if a currency is not found in the en_US + // Currencies data, then the en Currencies data will be searched. + // Currently, if a Currencies datum exists in en_US and en, the + // en_US entry hides that in en. + + // We want multi-level fallback for this resource, so we implement + // it manually. + + // Use a separate UErrorCode here that does not propagate out of + // this function. + UErrorCode ec2 = U_ZERO_ERROR; + + char loc[ULOC_FULLNAME_CAPACITY]; + uloc_getName(locale, loc, sizeof(loc), &ec2); + if (U_FAILURE(ec2) || ec2 == U_STRING_NOT_TERMINATED_WARNING) { + *ec = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + char buf[ISO_CURRENCY_CODE_LENGTH+1]; + myUCharsToChars(buf, currency); + + /* Normalize the keyword value to uppercase */ + T_CString_toUpperCase(buf); + + const char16_t* s = nullptr; + ec2 = U_ZERO_ERROR; + LocalUResourceBundlePointer rb(ures_open(U_ICUDATA_CURR, loc, &ec2)); + + if (nameStyle == UCURR_NARROW_SYMBOL_NAME || nameStyle == UCURR_FORMAL_SYMBOL_NAME || nameStyle == UCURR_VARIANT_SYMBOL_NAME) { + CharString key; + switch (nameStyle) { + case UCURR_NARROW_SYMBOL_NAME: + key.append(CURRENCIES_NARROW, ec2); + break; + case UCURR_FORMAL_SYMBOL_NAME: + key.append(CURRENCIES_FORMAL, ec2); + break; + case UCURR_VARIANT_SYMBOL_NAME: + key.append(CURRENCIES_VARIANT, ec2); + break; + default: + *ec = U_UNSUPPORTED_ERROR; + return 0; + } + key.append("/", ec2); + key.append(buf, ec2); + s = ures_getStringByKeyWithFallback(rb.getAlias(), key.data(), len, &ec2); + if (ec2 == U_MISSING_RESOURCE_ERROR) { + *ec = U_USING_FALLBACK_WARNING; + ec2 = U_ZERO_ERROR; + choice = UCURR_SYMBOL_NAME; + } + } + if (s == nullptr) { + ures_getByKey(rb.getAlias(), CURRENCIES, rb.getAlias(), &ec2); + ures_getByKeyWithFallback(rb.getAlias(), buf, rb.getAlias(), &ec2); + s = ures_getStringByIndex(rb.getAlias(), choice, len, &ec2); + } + + // If we've succeeded we're done. Otherwise, try to fallback. + // If that fails (because we are already at root) then exit. + if (U_SUCCESS(ec2)) { + if (ec2 == U_USING_DEFAULT_WARNING + || (ec2 == U_USING_FALLBACK_WARNING && *ec != U_USING_DEFAULT_WARNING)) { + *ec = ec2; + } + } + + // We no longer support choice format data in names. Data should not contain + // choice patterns. + if (isChoiceFormat != nullptr) { + *isChoiceFormat = false; + } + if (U_SUCCESS(ec2)) { + U_ASSERT(s != nullptr); + return s; + } + + // If we fail to find a match, use the ISO 4217 code + *len = u_strlen(currency); // Should == ISO_CURRENCY_CODE_LENGTH, but maybe not...? + *ec = U_USING_DEFAULT_WARNING; + return currency; +} + +U_CAPI const char16_t* U_EXPORT2 +ucurr_getPluralName(const char16_t* currency, + const char* locale, + UBool* isChoiceFormat, + const char* pluralCount, + int32_t* len, // fillin + UErrorCode* ec) { + // Look up the Currencies resource for the given locale. The + // Currencies locale data looks like this: + //|en { + //| CurrencyPlurals { + //| USD{ + //| one{"US dollar"} + //| other{"US dollars"} + //| } + //| } + //|} + + if (U_FAILURE(*ec)) { + return 0; + } + + // Use a separate UErrorCode here that does not propagate out of + // this function. + UErrorCode ec2 = U_ZERO_ERROR; + + char loc[ULOC_FULLNAME_CAPACITY]; + uloc_getName(locale, loc, sizeof(loc), &ec2); + if (U_FAILURE(ec2) || ec2 == U_STRING_NOT_TERMINATED_WARNING) { + *ec = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + char buf[ISO_CURRENCY_CODE_LENGTH+1]; + myUCharsToChars(buf, currency); + + const char16_t* s = nullptr; + ec2 = U_ZERO_ERROR; + UResourceBundle* rb = ures_open(U_ICUDATA_CURR, loc, &ec2); + + rb = ures_getByKey(rb, CURRENCYPLURALS, rb, &ec2); + + // Fetch resource with multi-level resource inheritance fallback + rb = ures_getByKeyWithFallback(rb, buf, rb, &ec2); + + s = ures_getStringByKeyWithFallback(rb, pluralCount, len, &ec2); + if (U_FAILURE(ec2)) { + // fall back to "other" + ec2 = U_ZERO_ERROR; + s = ures_getStringByKeyWithFallback(rb, "other", len, &ec2); + if (U_FAILURE(ec2)) { + ures_close(rb); + // fall back to long name in Currencies + return ucurr_getName(currency, locale, UCURR_LONG_NAME, + isChoiceFormat, len, ec); + } + } + ures_close(rb); + + // If we've succeeded we're done. Otherwise, try to fallback. + // If that fails (because we are already at root) then exit. + if (U_SUCCESS(ec2)) { + if (ec2 == U_USING_DEFAULT_WARNING + || (ec2 == U_USING_FALLBACK_WARNING && *ec != U_USING_DEFAULT_WARNING)) { + *ec = ec2; + } + U_ASSERT(s != nullptr); + return s; + } + + // If we fail to find a match, use the ISO 4217 code + *len = u_strlen(currency); // Should == ISO_CURRENCY_CODE_LENGTH, but maybe not...? + *ec = U_USING_DEFAULT_WARNING; + return currency; +} + + +//======================================================================== +// Following are structure and function for parsing currency names + +#define NEED_TO_BE_DELETED 0x1 + +// TODO: a better way to define this? +#define MAX_CURRENCY_NAME_LEN 100 + +typedef struct { + const char* IsoCode; // key + char16_t* currencyName; // value + int32_t currencyNameLen; // value length + int32_t flag; // flags +} CurrencyNameStruct; + + +#ifndef MIN +#define MIN(a,b) (((a)<(b)) ? (a) : (b)) +#endif + +#ifndef MAX +#define MAX(a,b) (((a)<(b)) ? (b) : (a)) +#endif + + +// Comparison function used in quick sort. +static int U_CALLCONV currencyNameComparator(const void* a, const void* b) { + const CurrencyNameStruct* currName_1 = (const CurrencyNameStruct*)a; + const CurrencyNameStruct* currName_2 = (const CurrencyNameStruct*)b; + for (int32_t i = 0; + i < MIN(currName_1->currencyNameLen, currName_2->currencyNameLen); + ++i) { + if (currName_1->currencyName[i] < currName_2->currencyName[i]) { + return -1; + } + if (currName_1->currencyName[i] > currName_2->currencyName[i]) { + return 1; + } + } + if (currName_1->currencyNameLen < currName_2->currencyNameLen) { + return -1; + } else if (currName_1->currencyNameLen > currName_2->currencyNameLen) { + return 1; + } + return 0; +} + + +// Give a locale, return the maximum number of currency names associated with +// this locale. +// It gets currency names from resource bundles using fallback. +// It is the maximum number because in the fallback chain, some of the +// currency names are duplicated. +// For example, given locale as "en_US", the currency names get from resource +// bundle in "en_US" and "en" are duplicated. The fallback mechanism will count +// all currency names in "en_US" and "en". +static void +getCurrencyNameCount(const char* loc, int32_t* total_currency_name_count, int32_t* total_currency_symbol_count) { + U_NAMESPACE_USE + *total_currency_name_count = 0; + *total_currency_symbol_count = 0; + const char16_t* s = nullptr; + char locale[ULOC_FULLNAME_CAPACITY] = ""; + uprv_strcpy(locale, loc); + const icu::Hashtable *currencySymbolsEquiv = getCurrSymbolsEquiv(); + for (;;) { + UErrorCode ec2 = U_ZERO_ERROR; + // TODO: ures_openDirect? + UResourceBundle* rb = ures_open(U_ICUDATA_CURR, locale, &ec2); + UResourceBundle* curr = ures_getByKey(rb, CURRENCIES, nullptr, &ec2); + int32_t n = ures_getSize(curr); + for (int32_t i=0; i(symbol->getBuffer()); + (*currencySymbols)[*total_currency_symbol_count].flag = 0; + (*currencySymbols)[(*total_currency_symbol_count)++].currencyNameLen = symbol->length(); + } + } + + // Add currency long name. + s = ures_getStringByIndex(names, UCURR_LONG_NAME, &len, &ec2); + (*currencyNames)[*total_currency_name_count].IsoCode = iso; + char16_t* upperName = toUpperCase(s, len, locale); + (*currencyNames)[*total_currency_name_count].currencyName = upperName; + (*currencyNames)[*total_currency_name_count].flag = NEED_TO_BE_DELETED; + (*currencyNames)[(*total_currency_name_count)++].currencyNameLen = len; + + // put (iso, 3, and iso) in to array + // Add currency ISO code. + (*currencySymbols)[*total_currency_symbol_count].IsoCode = iso; + (*currencySymbols)[*total_currency_symbol_count].currencyName = (char16_t*)uprv_malloc(sizeof(char16_t)*3); + // Must convert iso[] into Unicode + u_charsToUChars(iso, (*currencySymbols)[*total_currency_symbol_count].currencyName, 3); + (*currencySymbols)[*total_currency_symbol_count].flag = NEED_TO_BE_DELETED; + (*currencySymbols)[(*total_currency_symbol_count)++].currencyNameLen = 3; + + ures_close(names); + } + + // currency plurals + UErrorCode ec5 = U_ZERO_ERROR; + UResourceBundle* curr_p = ures_getByKey(rb, CURRENCYPLURALS, nullptr, &ec5); + n = ures_getSize(curr_p); + for (int32_t i=0; i= currencyNames[mid].currencyNameLen) { + first = mid + 1; + } else { + if (key > currencyNames[mid].currencyName[indexInCurrencyNames]) { + first = mid + 1; + } + else if (key < currencyNames[mid].currencyName[indexInCurrencyNames]) { + last = mid - 1; + } + else { + // Find a match, and looking for ranges + // Now do two more binary searches. First, on the left side for + // the greatest L such that CurrencyNameStruct[L] < key. + int32_t L = *begin; + int32_t R = mid; + +#ifdef UCURR_DEBUG + printf("mid = %d\n", mid); +#endif + while (L < R) { + int32_t M = (L + R) / 2; +#ifdef UCURR_DEBUG + printf("L = %d, R = %d, M = %d\n", L, R, M); +#endif + if (indexInCurrencyNames >= currencyNames[M].currencyNameLen) { + L = M + 1; + } else { + if (currencyNames[M].currencyName[indexInCurrencyNames] < key) { + L = M + 1; + } else { +#ifdef UCURR_DEBUG + U_ASSERT(currencyNames[M].currencyName[indexInCurrencyNames] == key); +#endif + R = M; + } + } + } +#ifdef UCURR_DEBUG + U_ASSERT(L == R); +#endif + *begin = L; +#ifdef UCURR_DEBUG + printf("begin = %d\n", *begin); + U_ASSERT(currencyNames[*begin].currencyName[indexInCurrencyNames] == key); +#endif + + // Now for the second search, finding the least R such that + // key < CurrencyNameStruct[R]. + L = mid; + R = *end; + while (L < R) { + int32_t M = (L + R) / 2; +#ifdef UCURR_DEBUG + printf("L = %d, R = %d, M = %d\n", L, R, M); +#endif + if (currencyNames[M].currencyNameLen < indexInCurrencyNames) { + L = M + 1; + } else { + if (currencyNames[M].currencyName[indexInCurrencyNames] > key) { + R = M; + } else { +#ifdef UCURR_DEBUG + U_ASSERT(currencyNames[M].currencyName[indexInCurrencyNames] == key); +#endif + L = M + 1; + } + } + } +#ifdef UCURR_DEBUG + U_ASSERT(L == R); +#endif + if (currencyNames[R].currencyName[indexInCurrencyNames] > key) { + *end = R - 1; + } else { + *end = R; + } +#ifdef UCURR_DEBUG + printf("end = %d\n", *end); +#endif + + // now, found the range. check whether there is exact match + if (currencyNames[*begin].currencyNameLen == indexInCurrencyNames + 1) { + return *begin; // find range and exact match. + } + return -1; // find range, but no exact match. + } + } + } + *begin = -1; + *end = -1; + return -1; // failed to find range. +} + + +// Linear search "text" in "currencyNames". +// @param begin, end: the begin and end index in currencyNames, within which +// range should the search be performed. +// @param textLen: the length of the text to be compared +// @param maxMatchLen(IN/OUT): passing in the computed max matching length +// pass out the new max matching length +// @param maxMatchIndex: the index in currencyName which has the longest +// match with input text. +static void +linearSearch(const CurrencyNameStruct* currencyNames, + int32_t begin, int32_t end, + const char16_t* text, int32_t textLen, + int32_t *partialMatchLen, + int32_t *maxMatchLen, int32_t* maxMatchIndex) { + int32_t initialPartialMatchLen = *partialMatchLen; + for (int32_t index = begin; index <= end; ++index) { + int32_t len = currencyNames[index].currencyNameLen; + if (len > *maxMatchLen && len <= textLen && + uprv_memcmp(currencyNames[index].currencyName, text, len * sizeof(char16_t)) == 0) { + *partialMatchLen = MAX(*partialMatchLen, len); + *maxMatchIndex = index; + *maxMatchLen = len; +#ifdef UCURR_DEBUG + printf("maxMatchIndex = %d, maxMatchLen = %d\n", + *maxMatchIndex, *maxMatchLen); +#endif + } else { + // Check for partial matches. + for (int32_t i=initialPartialMatchLen; icurrencyNames, entry->totalCurrencyNameCount); + deleteCurrencyNames(entry->currencySymbols, entry->totalCurrencySymbolCount); + uprv_free(entry); +} + + +// Cache clean up +static UBool U_CALLCONV +currency_cache_cleanup() { + for (int32_t i = 0; i < CURRENCY_NAME_CACHE_NUM; ++i) { + if (currCache[i]) { + deleteCacheEntry(currCache[i]); + currCache[i] = 0; + } + } + return true; +} + + +/** + * Loads the currency name data from the cache, or from resource bundles if necessary. + * The refCount is automatically incremented. It is the caller's responsibility + * to decrement it when done! + */ +static CurrencyNameCacheEntry* +getCacheEntry(const char* locale, UErrorCode& ec) { + + int32_t total_currency_name_count = 0; + CurrencyNameStruct* currencyNames = nullptr; + int32_t total_currency_symbol_count = 0; + CurrencyNameStruct* currencySymbols = nullptr; + CurrencyNameCacheEntry* cacheEntry = nullptr; + + umtx_lock(&gCurrencyCacheMutex); + // in order to handle racing correctly, + // not putting 'search' in a separate function. + int8_t found = -1; + for (int8_t i = 0; i < CURRENCY_NAME_CACHE_NUM; ++i) { + if (currCache[i]!= nullptr && + uprv_strcmp(locale, currCache[i]->locale) == 0) { + found = i; + break; + } + } + if (found != -1) { + cacheEntry = currCache[found]; + ++(cacheEntry->refCount); + } + umtx_unlock(&gCurrencyCacheMutex); + if (found == -1) { + collectCurrencyNames(locale, ¤cyNames, &total_currency_name_count, ¤cySymbols, &total_currency_symbol_count, ec); + if (U_FAILURE(ec)) { + return nullptr; + } + umtx_lock(&gCurrencyCacheMutex); + // check again. + for (int8_t i = 0; i < CURRENCY_NAME_CACHE_NUM; ++i) { + if (currCache[i]!= nullptr && + uprv_strcmp(locale, currCache[i]->locale) == 0) { + found = i; + break; + } + } + if (found == -1) { + // insert new entry to + // currentCacheEntryIndex % CURRENCY_NAME_CACHE_NUM + // and remove the existing entry + // currentCacheEntryIndex % CURRENCY_NAME_CACHE_NUM + // from cache. + cacheEntry = currCache[currentCacheEntryIndex]; + if (cacheEntry) { + --(cacheEntry->refCount); + // delete if the ref count is zero + if (cacheEntry->refCount == 0) { + deleteCacheEntry(cacheEntry); + } + } + cacheEntry = (CurrencyNameCacheEntry*)uprv_malloc(sizeof(CurrencyNameCacheEntry)); + currCache[currentCacheEntryIndex] = cacheEntry; + uprv_strcpy(cacheEntry->locale, locale); + cacheEntry->currencyNames = currencyNames; + cacheEntry->totalCurrencyNameCount = total_currency_name_count; + cacheEntry->currencySymbols = currencySymbols; + cacheEntry->totalCurrencySymbolCount = total_currency_symbol_count; + cacheEntry->refCount = 2; // one for cache, one for reference + currentCacheEntryIndex = (currentCacheEntryIndex + 1) % CURRENCY_NAME_CACHE_NUM; + ucln_common_registerCleanup(UCLN_COMMON_CURRENCY, currency_cleanup); + } else { + deleteCurrencyNames(currencyNames, total_currency_name_count); + deleteCurrencyNames(currencySymbols, total_currency_symbol_count); + cacheEntry = currCache[found]; + ++(cacheEntry->refCount); + } + umtx_unlock(&gCurrencyCacheMutex); + } + + return cacheEntry; +} + +static void releaseCacheEntry(CurrencyNameCacheEntry* cacheEntry) { + umtx_lock(&gCurrencyCacheMutex); + --(cacheEntry->refCount); + if (cacheEntry->refCount == 0) { // remove + deleteCacheEntry(cacheEntry); + } + umtx_unlock(&gCurrencyCacheMutex); +} + +U_CAPI void +uprv_parseCurrency(const char* locale, + const icu::UnicodeString& text, + icu::ParsePosition& pos, + int8_t type, + int32_t* partialMatchLen, + char16_t* result, + UErrorCode& ec) { + U_NAMESPACE_USE + if (U_FAILURE(ec)) { + return; + } + CurrencyNameCacheEntry* cacheEntry = getCacheEntry(locale, ec); + if (U_FAILURE(ec)) { + return; + } + + int32_t total_currency_name_count = cacheEntry->totalCurrencyNameCount; + CurrencyNameStruct* currencyNames = cacheEntry->currencyNames; + int32_t total_currency_symbol_count = cacheEntry->totalCurrencySymbolCount; + CurrencyNameStruct* currencySymbols = cacheEntry->currencySymbols; + + int32_t start = pos.getIndex(); + + char16_t inputText[MAX_CURRENCY_NAME_LEN]; + char16_t upperText[MAX_CURRENCY_NAME_LEN]; + int32_t textLen = MIN(MAX_CURRENCY_NAME_LEN, text.length() - start); + text.extract(start, textLen, inputText); + UErrorCode ec1 = U_ZERO_ERROR; + textLen = u_strToUpper(upperText, MAX_CURRENCY_NAME_LEN, inputText, textLen, locale, &ec1); + + // Make sure partialMatchLen is initialized + *partialMatchLen = 0; + + int32_t max = 0; + int32_t matchIndex = -1; + // case in-sensitive comparison against currency names + searchCurrencyName(currencyNames, total_currency_name_count, + upperText, textLen, partialMatchLen, &max, &matchIndex); + +#ifdef UCURR_DEBUG + printf("search in names, max = %d, matchIndex = %d\n", max, matchIndex); +#endif + + int32_t maxInSymbol = 0; + int32_t matchIndexInSymbol = -1; + if (type != UCURR_LONG_NAME) { // not name only + // case sensitive comparison against currency symbols and ISO code. + searchCurrencyName(currencySymbols, total_currency_symbol_count, + inputText, textLen, + partialMatchLen, + &maxInSymbol, &matchIndexInSymbol); + } + +#ifdef UCURR_DEBUG + printf("search in symbols, maxInSymbol = %d, matchIndexInSymbol = %d\n", maxInSymbol, matchIndexInSymbol); + if(matchIndexInSymbol != -1) { + printf("== ISO=%s\n", currencySymbols[matchIndexInSymbol].IsoCode); + } +#endif + + if (max >= maxInSymbol && matchIndex != -1) { + u_charsToUChars(currencyNames[matchIndex].IsoCode, result, 4); + pos.setIndex(start + max); + } else if (maxInSymbol >= max && matchIndexInSymbol != -1) { + u_charsToUChars(currencySymbols[matchIndexInSymbol].IsoCode, result, 4); + pos.setIndex(start + maxInSymbol); + } + + // decrease reference count + releaseCacheEntry(cacheEntry); +} + +void uprv_currencyLeads(const char* locale, icu::UnicodeSet& result, UErrorCode& ec) { + U_NAMESPACE_USE + if (U_FAILURE(ec)) { + return; + } + CurrencyNameCacheEntry* cacheEntry = getCacheEntry(locale, ec); + if (U_FAILURE(ec)) { + return; + } + + for (int32_t i=0; itotalCurrencySymbolCount; i++) { + const CurrencyNameStruct& info = cacheEntry->currencySymbols[i]; + UChar32 cp; + U16_GET(info.currencyName, 0, 0, info.currencyNameLen, cp); + result.add(cp); + } + + for (int32_t i=0; itotalCurrencyNameCount; i++) { + const CurrencyNameStruct& info = cacheEntry->currencyNames[i]; + UChar32 cp; + U16_GET(info.currencyName, 0, 0, info.currencyNameLen, cp); + result.add(cp); + } + + // decrease reference count + releaseCacheEntry(cacheEntry); +} + + +/** + * Internal method. Given a currency ISO code and a locale, return + * the "static" currency name. This is usually the same as the + * UCURR_SYMBOL_NAME, but if the latter is a choice format, then the + * format is applied to the number 2.0 (to yield the more common + * plural) to return a static name. + * + * This is used for backward compatibility with old currency logic in + * DecimalFormat and DecimalFormatSymbols. + */ +U_CAPI void +uprv_getStaticCurrencyName(const char16_t* iso, const char* loc, + icu::UnicodeString& result, UErrorCode& ec) +{ + U_NAMESPACE_USE + + int32_t len; + const char16_t* currname = ucurr_getName(iso, loc, UCURR_SYMBOL_NAME, + nullptr /* isChoiceFormat */, &len, &ec); + if (U_SUCCESS(ec)) { + result.setTo(currname, len); + } +} + +U_CAPI int32_t U_EXPORT2 +ucurr_getDefaultFractionDigits(const char16_t* currency, UErrorCode* ec) { + return ucurr_getDefaultFractionDigitsForUsage(currency,UCURR_USAGE_STANDARD,ec); +} + +U_CAPI int32_t U_EXPORT2 +ucurr_getDefaultFractionDigitsForUsage(const char16_t* currency, const UCurrencyUsage usage, UErrorCode* ec) { + int32_t fracDigits = 0; + if (U_SUCCESS(*ec)) { + switch (usage) { + case UCURR_USAGE_STANDARD: + fracDigits = (_findMetaData(currency, *ec))[0]; + break; + case UCURR_USAGE_CASH: + fracDigits = (_findMetaData(currency, *ec))[2]; + break; + default: + *ec = U_UNSUPPORTED_ERROR; + } + } + return fracDigits; +} + +U_CAPI double U_EXPORT2 +ucurr_getRoundingIncrement(const char16_t* currency, UErrorCode* ec) { + return ucurr_getRoundingIncrementForUsage(currency, UCURR_USAGE_STANDARD, ec); +} + +U_CAPI double U_EXPORT2 +ucurr_getRoundingIncrementForUsage(const char16_t* currency, const UCurrencyUsage usage, UErrorCode* ec) { + double result = 0.0; + + const int32_t *data = _findMetaData(currency, *ec); + if (U_SUCCESS(*ec)) { + int32_t fracDigits; + int32_t increment; + switch (usage) { + case UCURR_USAGE_STANDARD: + fracDigits = data[0]; + increment = data[1]; + break; + case UCURR_USAGE_CASH: + fracDigits = data[2]; + increment = data[3]; + break; + default: + *ec = U_UNSUPPORTED_ERROR; + return result; + } + + // If the meta data is invalid, return 0.0 + if (fracDigits < 0 || fracDigits > MAX_POW10) { + *ec = U_INVALID_FORMAT_ERROR; + } else { + // A rounding value of 0 or 1 indicates no rounding. + if (increment >= 2) { + // Return (increment) / 10^(fracDigits). The only actual rounding data, + // as of this writing, is CHF { 2, 5 }. + result = double(increment) / POW10[fracDigits]; + } + } + } + + return result; +} + +U_CDECL_BEGIN + +typedef struct UCurrencyContext { + uint32_t currType; /* UCurrCurrencyType */ + uint32_t listIdx; +} UCurrencyContext; + +/* +Please keep this list in alphabetical order. +You can look at the CLDR supplemental data or ISO-4217 for the meaning of some +of these items. +ISO-4217: http://www.iso.org/iso/en/prods-services/popstds/currencycodeslist.html +*/ +static const struct CurrencyList { + const char *currency; + uint32_t currType; +} gCurrencyList[] = { + {"ADP", UCURR_COMMON|UCURR_DEPRECATED}, + {"AED", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"AFA", UCURR_COMMON|UCURR_DEPRECATED}, + {"AFN", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"ALK", UCURR_COMMON|UCURR_DEPRECATED}, + {"ALL", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"AMD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"ANG", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"AOA", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"AOK", UCURR_COMMON|UCURR_DEPRECATED}, + {"AON", UCURR_COMMON|UCURR_DEPRECATED}, + {"AOR", UCURR_COMMON|UCURR_DEPRECATED}, + {"ARA", UCURR_COMMON|UCURR_DEPRECATED}, + {"ARL", UCURR_COMMON|UCURR_DEPRECATED}, + {"ARM", UCURR_COMMON|UCURR_DEPRECATED}, + {"ARP", UCURR_COMMON|UCURR_DEPRECATED}, + {"ARS", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"ATS", UCURR_COMMON|UCURR_DEPRECATED}, + {"AUD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"AWG", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"AZM", UCURR_COMMON|UCURR_DEPRECATED}, + {"AZN", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BAD", UCURR_COMMON|UCURR_DEPRECATED}, + {"BAM", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BAN", UCURR_COMMON|UCURR_DEPRECATED}, + {"BBD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BDT", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BEC", UCURR_UNCOMMON|UCURR_DEPRECATED}, + {"BEF", UCURR_COMMON|UCURR_DEPRECATED}, + {"BEL", UCURR_UNCOMMON|UCURR_DEPRECATED}, + {"BGL", UCURR_COMMON|UCURR_DEPRECATED}, + {"BGM", UCURR_COMMON|UCURR_DEPRECATED}, + {"BGN", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BGO", UCURR_COMMON|UCURR_DEPRECATED}, + {"BHD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BIF", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BMD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BND", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BOB", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BOL", UCURR_COMMON|UCURR_DEPRECATED}, + {"BOP", UCURR_COMMON|UCURR_DEPRECATED}, + {"BOV", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"BRB", UCURR_COMMON|UCURR_DEPRECATED}, + {"BRC", UCURR_COMMON|UCURR_DEPRECATED}, + {"BRE", UCURR_COMMON|UCURR_DEPRECATED}, + {"BRL", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BRN", UCURR_COMMON|UCURR_DEPRECATED}, + {"BRR", UCURR_COMMON|UCURR_DEPRECATED}, + {"BRZ", UCURR_COMMON|UCURR_DEPRECATED}, + {"BSD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BTN", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BUK", UCURR_COMMON|UCURR_DEPRECATED}, + {"BWP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BYB", UCURR_COMMON|UCURR_DEPRECATED}, + {"BYN", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"BYR", UCURR_COMMON|UCURR_DEPRECATED}, + {"BZD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"CAD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"CDF", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"CHE", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"CHF", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"CHW", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"CLE", UCURR_COMMON|UCURR_DEPRECATED}, + {"CLF", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"CLP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"CNH", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"CNX", UCURR_UNCOMMON|UCURR_DEPRECATED}, + {"CNY", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"COP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"COU", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"CRC", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"CSD", UCURR_COMMON|UCURR_DEPRECATED}, + {"CSK", UCURR_COMMON|UCURR_DEPRECATED}, + {"CUC", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"CUP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"CVE", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"CYP", UCURR_COMMON|UCURR_DEPRECATED}, + {"CZK", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"DDM", UCURR_COMMON|UCURR_DEPRECATED}, + {"DEM", UCURR_COMMON|UCURR_DEPRECATED}, + {"DJF", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"DKK", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"DOP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"DZD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"ECS", UCURR_COMMON|UCURR_DEPRECATED}, + {"ECV", UCURR_UNCOMMON|UCURR_DEPRECATED}, + {"EEK", UCURR_COMMON|UCURR_DEPRECATED}, + {"EGP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"ERN", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"ESA", UCURR_UNCOMMON|UCURR_DEPRECATED}, + {"ESB", UCURR_UNCOMMON|UCURR_DEPRECATED}, + {"ESP", UCURR_COMMON|UCURR_DEPRECATED}, + {"ETB", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"EUR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"FIM", UCURR_COMMON|UCURR_DEPRECATED}, + {"FJD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"FKP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"FRF", UCURR_COMMON|UCURR_DEPRECATED}, + {"GBP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"GEK", UCURR_COMMON|UCURR_DEPRECATED}, + {"GEL", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"GHC", UCURR_COMMON|UCURR_DEPRECATED}, + {"GHS", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"GIP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"GMD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"GNF", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"GNS", UCURR_COMMON|UCURR_DEPRECATED}, + {"GQE", UCURR_COMMON|UCURR_DEPRECATED}, + {"GRD", UCURR_COMMON|UCURR_DEPRECATED}, + {"GTQ", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"GWE", UCURR_COMMON|UCURR_DEPRECATED}, + {"GWP", UCURR_COMMON|UCURR_DEPRECATED}, + {"GYD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"HKD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"HNL", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"HRD", UCURR_COMMON|UCURR_DEPRECATED}, + {"HRK", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"HTG", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"HUF", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"IDR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"IEP", UCURR_COMMON|UCURR_DEPRECATED}, + {"ILP", UCURR_COMMON|UCURR_DEPRECATED}, + {"ILR", UCURR_COMMON|UCURR_DEPRECATED}, + {"ILS", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"INR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"IQD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"IRR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"ISJ", UCURR_COMMON|UCURR_DEPRECATED}, + {"ISK", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"ITL", UCURR_COMMON|UCURR_DEPRECATED}, + {"JMD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"JOD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"JPY", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"KES", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"KGS", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"KHR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"KMF", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"KPW", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"KRH", UCURR_COMMON|UCURR_DEPRECATED}, + {"KRO", UCURR_COMMON|UCURR_DEPRECATED}, + {"KRW", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"KWD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"KYD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"KZT", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"LAK", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"LBP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"LKR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"LRD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"LSL", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"LSM", UCURR_COMMON|UCURR_DEPRECATED}, // questionable, remove? + {"LTL", UCURR_COMMON|UCURR_DEPRECATED}, + {"LTT", UCURR_COMMON|UCURR_DEPRECATED}, + {"LUC", UCURR_UNCOMMON|UCURR_DEPRECATED}, + {"LUF", UCURR_COMMON|UCURR_DEPRECATED}, + {"LUL", UCURR_UNCOMMON|UCURR_DEPRECATED}, + {"LVL", UCURR_COMMON|UCURR_DEPRECATED}, + {"LVR", UCURR_COMMON|UCURR_DEPRECATED}, + {"LYD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MAD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MAF", UCURR_COMMON|UCURR_DEPRECATED}, + {"MCF", UCURR_COMMON|UCURR_DEPRECATED}, + {"MDC", UCURR_COMMON|UCURR_DEPRECATED}, + {"MDL", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MGA", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MGF", UCURR_COMMON|UCURR_DEPRECATED}, + {"MKD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MKN", UCURR_COMMON|UCURR_DEPRECATED}, + {"MLF", UCURR_COMMON|UCURR_DEPRECATED}, + {"MMK", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MNT", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MOP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MRO", UCURR_COMMON|UCURR_DEPRECATED}, + {"MRU", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MTL", UCURR_COMMON|UCURR_DEPRECATED}, + {"MTP", UCURR_COMMON|UCURR_DEPRECATED}, + {"MUR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MVP", UCURR_COMMON|UCURR_DEPRECATED}, // questionable, remove? + {"MVR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MWK", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MXN", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MXP", UCURR_COMMON|UCURR_DEPRECATED}, + {"MXV", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"MYR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"MZE", UCURR_COMMON|UCURR_DEPRECATED}, + {"MZM", UCURR_COMMON|UCURR_DEPRECATED}, + {"MZN", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"NAD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"NGN", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"NIC", UCURR_COMMON|UCURR_DEPRECATED}, + {"NIO", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"NLG", UCURR_COMMON|UCURR_DEPRECATED}, + {"NOK", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"NPR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"NZD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"OMR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"PAB", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"PEI", UCURR_COMMON|UCURR_DEPRECATED}, + {"PEN", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"PES", UCURR_COMMON|UCURR_DEPRECATED}, + {"PGK", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"PHP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"PKR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"PLN", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"PLZ", UCURR_COMMON|UCURR_DEPRECATED}, + {"PTE", UCURR_COMMON|UCURR_DEPRECATED}, + {"PYG", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"QAR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"RHD", UCURR_COMMON|UCURR_DEPRECATED}, + {"ROL", UCURR_COMMON|UCURR_DEPRECATED}, + {"RON", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"RSD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"RUB", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"RUR", UCURR_COMMON|UCURR_DEPRECATED}, + {"RWF", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SAR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SBD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SCR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SDD", UCURR_COMMON|UCURR_DEPRECATED}, + {"SDG", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SDP", UCURR_COMMON|UCURR_DEPRECATED}, + {"SEK", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SGD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SHP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SIT", UCURR_COMMON|UCURR_DEPRECATED}, + {"SKK", UCURR_COMMON|UCURR_DEPRECATED}, + {"SLE", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SLL", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SOS", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SRD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SRG", UCURR_COMMON|UCURR_DEPRECATED}, + {"SSP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"STD", UCURR_COMMON|UCURR_DEPRECATED}, + {"STN", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SUR", UCURR_COMMON|UCURR_DEPRECATED}, + {"SVC", UCURR_COMMON|UCURR_DEPRECATED}, + {"SYP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"SZL", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"THB", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"TJR", UCURR_COMMON|UCURR_DEPRECATED}, + {"TJS", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"TMM", UCURR_COMMON|UCURR_DEPRECATED}, + {"TMT", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"TND", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"TOP", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"TPE", UCURR_COMMON|UCURR_DEPRECATED}, + {"TRL", UCURR_COMMON|UCURR_DEPRECATED}, + {"TRY", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"TTD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"TWD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"TZS", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"UAH", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"UAK", UCURR_COMMON|UCURR_DEPRECATED}, + {"UGS", UCURR_COMMON|UCURR_DEPRECATED}, + {"UGX", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"USD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"USN", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"USS", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"UYI", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"UYP", UCURR_COMMON|UCURR_DEPRECATED}, + {"UYU", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"UYW", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"UZS", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"VEB", UCURR_COMMON|UCURR_DEPRECATED}, + {"VED", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"VEF", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"VES", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"VND", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"VNN", UCURR_COMMON|UCURR_DEPRECATED}, + {"VUV", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"WST", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"XAF", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"XAG", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XAU", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XBA", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XBB", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XBC", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XBD", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XCD", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"XDR", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XEU", UCURR_UNCOMMON|UCURR_DEPRECATED}, + {"XFO", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XFU", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XOF", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"XPD", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XPF", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"XPT", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XRE", UCURR_UNCOMMON|UCURR_DEPRECATED}, + {"XSU", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XTS", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XUA", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"XXX", UCURR_UNCOMMON|UCURR_NON_DEPRECATED}, + {"YDD", UCURR_COMMON|UCURR_DEPRECATED}, + {"YER", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"YUD", UCURR_COMMON|UCURR_DEPRECATED}, + {"YUM", UCURR_COMMON|UCURR_DEPRECATED}, + {"YUN", UCURR_COMMON|UCURR_DEPRECATED}, + {"YUR", UCURR_COMMON|UCURR_DEPRECATED}, + {"ZAL", UCURR_UNCOMMON|UCURR_DEPRECATED}, + {"ZAR", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"ZMK", UCURR_COMMON|UCURR_DEPRECATED}, + {"ZMW", UCURR_COMMON|UCURR_NON_DEPRECATED}, + {"ZRN", UCURR_COMMON|UCURR_DEPRECATED}, + {"ZRZ", UCURR_COMMON|UCURR_DEPRECATED}, + {"ZWD", UCURR_COMMON|UCURR_DEPRECATED}, + {"ZWL", UCURR_COMMON|UCURR_DEPRECATED}, + {"ZWR", UCURR_COMMON|UCURR_DEPRECATED}, + { nullptr, 0 } // Leave here to denote the end of the list. +}; + +#define UCURR_MATCHES_BITMASK(variable, typeToMatch) \ + ((typeToMatch) == UCURR_ALL || ((variable) & (typeToMatch)) == (typeToMatch)) + +static int32_t U_CALLCONV +ucurr_countCurrencyList(UEnumeration *enumerator, UErrorCode * /*pErrorCode*/) { + UCurrencyContext *myContext = (UCurrencyContext *)(enumerator->context); + uint32_t currType = myContext->currType; + int32_t count = 0; + + /* Count the number of items matching the type we are looking for. */ + for (int32_t idx = 0; gCurrencyList[idx].currency != nullptr; idx++) { + if (UCURR_MATCHES_BITMASK(gCurrencyList[idx].currType, currType)) { + count++; + } + } + return count; +} + +static const char* U_CALLCONV +ucurr_nextCurrencyList(UEnumeration *enumerator, + int32_t* resultLength, + UErrorCode * /*pErrorCode*/) +{ + UCurrencyContext *myContext = (UCurrencyContext *)(enumerator->context); + + /* Find the next in the list that matches the type we are looking for. */ + while (myContext->listIdx < UPRV_LENGTHOF(gCurrencyList)-1) { + const struct CurrencyList *currItem = &gCurrencyList[myContext->listIdx++]; + if (UCURR_MATCHES_BITMASK(currItem->currType, myContext->currType)) + { + if (resultLength) { + *resultLength = 3; /* Currency codes are only 3 chars long */ + } + return currItem->currency; + } + } + /* We enumerated too far. */ + if (resultLength) { + *resultLength = 0; + } + return nullptr; +} + +static void U_CALLCONV +ucurr_resetCurrencyList(UEnumeration *enumerator, UErrorCode * /*pErrorCode*/) { + ((UCurrencyContext *)(enumerator->context))->listIdx = 0; +} + +static void U_CALLCONV +ucurr_closeCurrencyList(UEnumeration *enumerator) { + uprv_free(enumerator->context); + uprv_free(enumerator); +} + +static void U_CALLCONV +ucurr_createCurrencyList(UHashtable *isoCodes, UErrorCode* status){ + UErrorCode localStatus = U_ZERO_ERROR; + + // Look up the CurrencyMap element in the root bundle. + UResourceBundle *rb = ures_openDirect(U_ICUDATA_CURR, CURRENCY_DATA, &localStatus); + UResourceBundle *currencyMapArray = ures_getByKey(rb, CURRENCY_MAP, rb, &localStatus); + + if (U_SUCCESS(localStatus)) { + // process each entry in currency map + for (int32_t i=0; iisoCode = isoCode; + entry->from = fromDate; + entry->to = toDate; + + localStatus = U_ZERO_ERROR; + uhash_put(isoCodes, (char16_t *)isoCode, entry, &localStatus); + } + } else { + *status = localStatus; + } + ures_close(currencyArray); + } + } else { + *status = localStatus; + } + + ures_close(currencyMapArray); +} + +static const UEnumeration gEnumCurrencyList = { + nullptr, + nullptr, + ucurr_closeCurrencyList, + ucurr_countCurrencyList, + uenum_unextDefault, + ucurr_nextCurrencyList, + ucurr_resetCurrencyList +}; +U_CDECL_END + + +static void U_CALLCONV initIsoCodes(UErrorCode &status) { + U_ASSERT(gIsoCodes == nullptr); + ucln_common_registerCleanup(UCLN_COMMON_CURRENCY, currency_cleanup); + + UHashtable *isoCodes = uhash_open(uhash_hashUChars, uhash_compareUChars, nullptr, &status); + if (U_FAILURE(status)) { + return; + } + uhash_setValueDeleter(isoCodes, deleteIsoCodeEntry); + + ucurr_createCurrencyList(isoCodes, &status); + if (U_FAILURE(status)) { + uhash_close(isoCodes); + return; + } + gIsoCodes = isoCodes; // Note: gIsoCodes is const. Once set up here it is never altered, + // and read only access is safe without synchronization. +} + +static void populateCurrSymbolsEquiv(icu::Hashtable *hash, UErrorCode &status) { + if (U_FAILURE(status)) { return; } + for (auto& entry : unisets::kCurrencyEntries) { + UnicodeString exemplar(entry.exemplar); + const UnicodeSet* set = unisets::get(entry.key); + if (set == nullptr) { return; } + UnicodeSetIterator it(*set); + while (it.next()) { + UnicodeString value = it.getString(); + if (value == exemplar) { + // No need to mark the exemplar character as an equivalent + continue; + } + makeEquivalent(exemplar, value, hash, status); + if (U_FAILURE(status)) { return; } + } + } +} + +static void U_CALLCONV initCurrSymbolsEquiv() { + U_ASSERT(gCurrSymbolsEquiv == nullptr); + UErrorCode status = U_ZERO_ERROR; + ucln_common_registerCleanup(UCLN_COMMON_CURRENCY, currency_cleanup); + icu::Hashtable *temp = new icu::Hashtable(status); + if (temp == nullptr) { + return; + } + if (U_FAILURE(status)) { + delete temp; + return; + } + temp->setValueDeleter(deleteUnicode); + populateCurrSymbolsEquiv(temp, status); + if (U_FAILURE(status)) { + delete temp; + return; + } + gCurrSymbolsEquiv = temp; +} + +U_CAPI UBool U_EXPORT2 +ucurr_isAvailable(const char16_t* isoCode, UDate from, UDate to, UErrorCode* eErrorCode) { + umtx_initOnce(gIsoCodesInitOnce, &initIsoCodes, *eErrorCode); + if (U_FAILURE(*eErrorCode)) { + return false; + } + + IsoCodeEntry* result = (IsoCodeEntry *) uhash_get(gIsoCodes, isoCode); + if (result == nullptr) { + return false; + } else if (from > to) { + *eErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return false; + } else if ((from > result->to) || (to < result->from)) { + return false; + } + return true; +} + +static const icu::Hashtable* getCurrSymbolsEquiv() { + umtx_initOnce(gCurrSymbolsEquivInitOnce, &initCurrSymbolsEquiv); + return gCurrSymbolsEquiv; +} + +U_CAPI UEnumeration * U_EXPORT2 +ucurr_openISOCurrencies(uint32_t currType, UErrorCode *pErrorCode) { + UEnumeration *myEnum = nullptr; + UCurrencyContext *myContext; + + myEnum = (UEnumeration*)uprv_malloc(sizeof(UEnumeration)); + if (myEnum == nullptr) { + *pErrorCode = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memcpy(myEnum, &gEnumCurrencyList, sizeof(UEnumeration)); + myContext = (UCurrencyContext*)uprv_malloc(sizeof(UCurrencyContext)); + if (myContext == nullptr) { + *pErrorCode = U_MEMORY_ALLOCATION_ERROR; + uprv_free(myEnum); + return nullptr; + } + myContext->currType = currType; + myContext->listIdx = 0; + myEnum->context = myContext; + return myEnum; +} + +U_CAPI int32_t U_EXPORT2 +ucurr_countCurrencies(const char* locale, + UDate date, + UErrorCode* ec) +{ + int32_t currCount = 0; + + if (ec != nullptr && U_SUCCESS(*ec)) + { + // local variables + UErrorCode localStatus = U_ZERO_ERROR; + char id[ULOC_FULLNAME_CAPACITY]; + + // get country or country_variant in `id' + idForLocale(locale, id, sizeof(id), ec); + + if (U_FAILURE(*ec)) + { + return 0; + } + + // Remove variants, which is only needed for registration. + char *idDelim = strchr(id, VAR_DELIM); + if (idDelim) + { + idDelim[0] = 0; + } + + // Look up the CurrencyMap element in the root bundle. + UResourceBundle *rb = ures_openDirect(U_ICUDATA_CURR, CURRENCY_DATA, &localStatus); + UResourceBundle *cm = ures_getByKey(rb, CURRENCY_MAP, rb, &localStatus); + + // Using the id derived from the local, get the currency data + UResourceBundle *countryArray = ures_getByKey(rb, id, cm, &localStatus); + + // process each currency to see which one is valid for the given date + if (U_SUCCESS(localStatus)) + { + for (int32_t i=0; i 2) + { + int32_t toLength = 0; + UResourceBundle *toRes = ures_getByKey(currencyRes, "to", nullptr, &localStatus); + const int32_t *toArray = ures_getIntVector(toRes, &toLength, &localStatus); + + currDate64 = (int64_t)toArray[0] << 32; + currDate64 |= ((int64_t)toArray[1] & (int64_t)INT64_C(0x00000000FFFFFFFF)); + UDate toDate = (UDate)currDate64; + + if ((fromDate <= date) && (date < toDate)) + { + currCount++; + } + + ures_close(toRes); + } + else + { + if (fromDate <= date) + { + currCount++; + } + } + + // close open resources + ures_close(currencyRes); + ures_close(fromRes); + + } // end For loop + } // end if (U_SUCCESS(localStatus)) + + ures_close(countryArray); + + // Check for errors + if (*ec == U_ZERO_ERROR || localStatus != U_ZERO_ERROR) + { + // There is nothing to fallback to. + // Report the failure/warning if possible. + *ec = localStatus; + } + + if (U_SUCCESS(*ec)) + { + // no errors + return currCount; + } + + } + + // If we got here, either error code is invalid or + // some argument passed is no good. + return 0; +} + +U_CAPI int32_t U_EXPORT2 +ucurr_forLocaleAndDate(const char* locale, + UDate date, + int32_t index, + char16_t* buff, + int32_t buffCapacity, + UErrorCode* ec) +{ + int32_t resLen = 0; + int32_t currIndex = 0; + const char16_t* s = nullptr; + + if (ec != nullptr && U_SUCCESS(*ec)) + { + // check the arguments passed + if ((buff && buffCapacity) || !buffCapacity ) + { + // local variables + UErrorCode localStatus = U_ZERO_ERROR; + char id[ULOC_FULLNAME_CAPACITY]; + + // get country or country_variant in `id' + idForLocale(locale, id, sizeof(id), ec); + if (U_FAILURE(*ec)) + { + return 0; + } + + // Remove variants, which is only needed for registration. + char *idDelim = strchr(id, VAR_DELIM); + if (idDelim) + { + idDelim[0] = 0; + } + + // Look up the CurrencyMap element in the root bundle. + UResourceBundle *rb = ures_openDirect(U_ICUDATA_CURR, CURRENCY_DATA, &localStatus); + UResourceBundle *cm = ures_getByKey(rb, CURRENCY_MAP, rb, &localStatus); + + // Using the id derived from the local, get the currency data + UResourceBundle *countryArray = ures_getByKey(rb, id, cm, &localStatus); + + // process each currency to see which one is valid for the given date + bool matchFound = false; + if (U_SUCCESS(localStatus)) + { + if ((index <= 0) || (index> ures_getSize(countryArray))) + { + // requested index is out of bounds + ures_close(countryArray); + return 0; + } + + for (int32_t i=0; i 2) + { + int32_t toLength = 0; + UResourceBundle *toRes = ures_getByKey(currencyRes, "to", nullptr, &localStatus); + const int32_t *toArray = ures_getIntVector(toRes, &toLength, &localStatus); + + currDate64 = (int64_t)toArray[0] << 32; + currDate64 |= ((int64_t)toArray[1] & (int64_t)INT64_C(0x00000000FFFFFFFF)); + UDate toDate = (UDate)currDate64; + + if ((fromDate <= date) && (date < toDate)) + { + currIndex++; + if (currIndex == index) + { + matchFound = true; + } + } + + ures_close(toRes); + } + else + { + if (fromDate <= date) + { + currIndex++; + if (currIndex == index) + { + matchFound = true; + } + } + } + + // close open resources + ures_close(currencyRes); + ures_close(fromRes); + + // check for loop exit + if (matchFound) + { + break; + } + + } // end For loop + } + + ures_close(countryArray); + + // Check for errors + if (*ec == U_ZERO_ERROR || localStatus != U_ZERO_ERROR) + { + // There is nothing to fallback to. + // Report the failure/warning if possible. + *ec = localStatus; + } + + if (U_SUCCESS(*ec)) + { + // no errors + if((buffCapacity> resLen) && matchFound) + { + // write out the currency value + u_strcpy(buff, s); + } + else + { + return 0; + } + } + + // return null terminated currency string + return u_terminateUChars(buff, buffCapacity, resLen, ec); + } + else + { + // illegal argument encountered + *ec = U_ILLEGAL_ARGUMENT_ERROR; + } + + } + + // If we got here, either error code is invalid or + // some argument passed is no good. + return resLen; +} + +static const UEnumeration defaultKeywordValues = { + nullptr, + nullptr, + ulist_close_keyword_values_iterator, + ulist_count_keyword_values, + uenum_unextDefault, + ulist_next_keyword_value, + ulist_reset_keyword_values_iterator +}; + +U_CAPI UEnumeration *U_EXPORT2 ucurr_getKeywordValuesForLocale(const char *key, const char *locale, UBool commonlyUsed, UErrorCode* status) { + // Resolve region + char prefRegion[ULOC_COUNTRY_CAPACITY]; + ulocimp_getRegionForSupplementalData(locale, true, prefRegion, sizeof(prefRegion), status); + + // Read value from supplementalData + UList *values = ulist_createEmptyList(status); + UList *otherValues = ulist_createEmptyList(status); + UEnumeration *en = (UEnumeration *)uprv_malloc(sizeof(UEnumeration)); + if (U_FAILURE(*status) || en == nullptr) { + if (en == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + } else { + uprv_free(en); + } + ulist_deleteList(values); + ulist_deleteList(otherValues); + return nullptr; + } + memcpy(en, &defaultKeywordValues, sizeof(UEnumeration)); + en->context = values; + + UResourceBundle *bundle = ures_openDirect(U_ICUDATA_CURR, "supplementalData", status); + ures_getByKey(bundle, "CurrencyMap", bundle, status); + UResourceBundle bundlekey, regbndl, curbndl, to; + ures_initStackObject(&bundlekey); + ures_initStackObject(®bndl); + ures_initStackObject(&curbndl); + ures_initStackObject(&to); + + while (U_SUCCESS(*status) && ures_hasNext(bundle)) { + ures_getNextResource(bundle, &bundlekey, status); + if (U_FAILURE(*status)) { + break; + } + const char *region = ures_getKey(&bundlekey); + UBool isPrefRegion = uprv_strcmp(region, prefRegion) == 0 ? true : false; + if (!isPrefRegion && commonlyUsed) { + // With commonlyUsed=true, we do not put + // currencies for other regions in the + // result list. + continue; + } + ures_getByKey(bundle, region, ®bndl, status); + if (U_FAILURE(*status)) { + break; + } + while (U_SUCCESS(*status) && ures_hasNext(®bndl)) { + ures_getNextResource(®bndl, &curbndl, status); + if (ures_getType(&curbndl) != URES_TABLE) { + // Currently, an empty ARRAY is mixed in. + continue; + } + char *curID = (char *)uprv_malloc(sizeof(char) * ULOC_KEYWORDS_CAPACITY); + int32_t curIDLength = ULOC_KEYWORDS_CAPACITY; + if (curID == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + break; + } + +#if U_CHARSET_FAMILY==U_ASCII_FAMILY + ures_getUTF8StringByKey(&curbndl, "id", curID, &curIDLength, true, status); + /* optimize - use the utf-8 string */ +#else + { + const char16_t* defString = ures_getStringByKey(&curbndl, "id", &curIDLength, status); + if(U_SUCCESS(*status)) { + if(curIDLength+1 > ULOC_KEYWORDS_CAPACITY) { + *status = U_BUFFER_OVERFLOW_ERROR; + } else { + u_UCharsToChars(defString, curID, curIDLength+1); + } + } + } +#endif + + if (U_FAILURE(*status)) { + break; + } + UBool hasTo = false; + ures_getByKey(&curbndl, "to", &to, status); + if (U_FAILURE(*status)) { + // Do nothing here... + *status = U_ZERO_ERROR; + } else { + hasTo = true; + } + if (isPrefRegion && !hasTo && !ulist_containsString(values, curID, (int32_t)uprv_strlen(curID))) { + // Currently active currency for the target country + ulist_addItemEndList(values, curID, true, status); + } else if (!ulist_containsString(otherValues, curID, (int32_t)uprv_strlen(curID)) && !commonlyUsed) { + ulist_addItemEndList(otherValues, curID, true, status); + } else { + uprv_free(curID); + } + } + + } + if (U_SUCCESS(*status)) { + if (commonlyUsed) { + if (ulist_getListSize(values) == 0) { + // This could happen if no valid region is supplied in the input + // locale. In this case, we use the CLDR's default. + uenum_close(en); + en = ucurr_getKeywordValuesForLocale(key, "und", true, status); + } + } else { + // Consolidate the list + char *value = nullptr; + ulist_resetList(otherValues); + while ((value = (char *)ulist_getNext(otherValues)) != nullptr) { + if (!ulist_containsString(values, value, (int32_t)uprv_strlen(value))) { + char *tmpValue = (char *)uprv_malloc(sizeof(char) * ULOC_KEYWORDS_CAPACITY); + uprv_memcpy(tmpValue, value, uprv_strlen(value) + 1); + ulist_addItemEndList(values, tmpValue, true, status); + if (U_FAILURE(*status)) { + break; + } + } + } + } + + ulist_resetList((UList *)(en->context)); + } else { + ulist_deleteList(values); + uprv_free(en); + values = nullptr; + en = nullptr; + } + ures_close(&to); + ures_close(&curbndl); + ures_close(®bndl); + ures_close(&bundlekey); + ures_close(bundle); + + ulist_deleteList(otherValues); + + return en; +} + + +U_CAPI int32_t U_EXPORT2 +ucurr_getNumericCode(const char16_t* currency) { + int32_t code = 0; + if (currency && u_strlen(currency) == ISO_CURRENCY_CODE_LENGTH) { + UErrorCode status = U_ZERO_ERROR; + + UResourceBundle *bundle = ures_openDirect(0, "currencyNumericCodes", &status); + ures_getByKey(bundle, "codeMap", bundle, &status); + if (U_SUCCESS(status)) { + char alphaCode[ISO_CURRENCY_CODE_LENGTH+1]; + myUCharsToChars(alphaCode, currency); + T_CString_toUpperCase(alphaCode); + ures_getByKey(bundle, alphaCode, bundle, &status); + int tmpCode = ures_getInt(bundle, &status); + if (U_SUCCESS(status)) { + code = tmpCode; + } + } + ures_close(bundle); + } + return code; +} +#endif /* #if !UCONFIG_NO_FORMATTING */ + +//eof diff --git a/intl/icu/source/common/ucurrimp.h b/intl/icu/source/common/ucurrimp.h new file mode 100644 index 0000000000..6d9588295d --- /dev/null +++ b/intl/icu/source/common/ucurrimp.h @@ -0,0 +1,78 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2002-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ + +#ifndef _UCURR_IMP_H_ +#define _UCURR_IMP_H_ + +#include "unicode/utypes.h" +#include "unicode/unistr.h" +#include "unicode/parsepos.h" +#include "unicode/uniset.h" + +/** + * Internal method. Given a currency ISO code and a locale, return + * the "static" currency name. This is usually the same as the + * UCURR_SYMBOL_NAME, but if the latter is a choice format, then the + * format is applied to the number 2.0 (to yield the more common + * plural) to return a static name. + * + * This is used for backward compatibility with old currency logic in + * DecimalFormat and DecimalFormatSymbols. + */ +U_CAPI void +uprv_getStaticCurrencyName(const UChar* iso, const char* loc, + icu::UnicodeString& result, UErrorCode& ec); + +/** + * Attempt to parse the given string as a currency, either as a + * display name in the given locale, or as a 3-letter ISO 4217 + * code. If multiple display names match, then the longest one is + * selected. If both a display name and a 3-letter ISO code + * match, then the display name is preferred, unless it's length + * is less than 3. + * + * The parameters must not be NULL. + * + * @param locale the locale of the display names to match + * @param text the text to parse + * @param pos input-output position; on input, the position within + * text to match; must have 0 <= pos.getIndex() < text.length(); + * on output, the position after the last matched character. If + * the parse fails, the position in unchanged upon output. + * @param type currency type to parse against, LONG_NAME only or not + * @param partialMatchLen The length of the longest matching prefix; + * this may be nonzero even if no full currency was matched. + * @return the ISO 4217 code, as a string, of the best match, or + * null if there is no match + * + * @internal + */ +U_CAPI void +uprv_parseCurrency(const char* locale, + const icu::UnicodeString& text, + icu::ParsePosition& pos, + int8_t type, + int32_t* partialMatchLen, + UChar* result, + UErrorCode& ec); + +/** + * Puts all possible first-characters of a currency into the + * specified UnicodeSet. + * + * @param locale the locale of the display names of interest + * @param result the UnicodeSet to which to add the starting characters + */ +void uprv_currencyLeads(const char* locale, icu::UnicodeSet& result, UErrorCode& ec); + + + +#endif /* #ifndef _UCURR_IMP_H_ */ + +//eof diff --git a/intl/icu/source/common/udata.cpp b/intl/icu/source/common/udata.cpp new file mode 100644 index 0000000000..76d970f3f2 --- /dev/null +++ b/intl/icu/source/common/udata.cpp @@ -0,0 +1,1460 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: udata.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 1999oct25 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" /* U_PLATFORM etc. */ + +#ifdef __GNUC__ +/* if gcc +#define ATTRIBUTE_WEAK __attribute__ ((weak)) +might have to #include some other header +*/ +#endif + +#include "unicode/putil.h" +#include "unicode/udata.h" +#include "unicode/uversion.h" +#include "charstr.h" +#include "cmemory.h" +#include "cstring.h" +#include "mutex.h" +#include "putilimp.h" +#include "restrace.h" +#include "uassert.h" +#include "ucln_cmn.h" +#include "ucmndata.h" +#include "udatamem.h" +#include "uhash.h" +#include "umapfile.h" +#include "umutex.h" + +/*********************************************************************** +* +* Notes on the organization of the ICU data implementation +* +* All of the public API is defined in udata.h +* +* The implementation is split into several files... +* +* - udata.c (this file) contains higher level code that knows about +* the search paths for locating data, caching opened data, etc. +* +* - umapfile.c contains the low level platform-specific code for actually loading +* (memory mapping, file reading, whatever) data into memory. +* +* - ucmndata.c deals with the tables of contents of ICU data items within +* an ICU common format data file. The implementation includes +* an abstract interface and support for multiple TOC formats. +* All knowledge of any specific TOC format is encapsulated here. +* +* - udatamem.c has code for managing UDataMemory structs. These are little +* descriptor objects for blocks of memory holding ICU data of +* various types. +*/ + +/* configuration ---------------------------------------------------------- */ + +/* If you are excruciatingly bored turn this on .. */ +/* #define UDATA_DEBUG 1 */ + +#if defined(UDATA_DEBUG) +# include +#endif + +U_NAMESPACE_USE + +/* + * Forward declarations + */ +static UDataMemory *udata_findCachedData(const char *path, UErrorCode &err); + +/*********************************************************************** +* +* static (Global) data +* +************************************************************************/ + +/* + * Pointers to the common ICU data. + * + * We store multiple pointers to ICU data packages and iterate through them + * when looking for a data item. + * + * It is possible to combine this with dependency inversion: + * One or more data package libraries may export + * functions that each return a pointer to their piece of the ICU data, + * and this file would import them as weak functions, without a + * strong linker dependency from the common library on the data library. + * + * Then we can have applications depend on only that part of ICU's data + * that they really need, reducing the size of binaries that take advantage + * of this. + */ +static UDataMemory *gCommonICUDataArray[10] = { nullptr }; // Access protected by icu global mutex. + +static u_atomic_int32_t gHaveTriedToLoadCommonData {0}; // See extendICUData(). + +static UHashtable *gCommonDataCache = nullptr; /* Global hash table of opened ICU data files. */ +static icu::UInitOnce gCommonDataCacheInitOnce {}; + +#if !defined(ICU_DATA_DIR_WINDOWS) +static UDataFileAccess gDataFileAccess = UDATA_DEFAULT_ACCESS; // Access not synchronized. + // Modifying is documented as thread-unsafe. +#else +// If we are using the Windows data directory, then look in one spot only. +static UDataFileAccess gDataFileAccess = UDATA_NO_FILES; +#endif + +static UBool U_CALLCONV +udata_cleanup() +{ + int32_t i; + + if (gCommonDataCache) { /* Delete the cache of user data mappings. */ + uhash_close(gCommonDataCache); /* Table owns the contents, and will delete them. */ + gCommonDataCache = nullptr; /* Cleanup is not thread safe. */ + } + gCommonDataCacheInitOnce.reset(); + + for (i = 0; i < UPRV_LENGTHOF(gCommonICUDataArray) && gCommonICUDataArray[i] != nullptr; ++i) { + udata_close(gCommonICUDataArray[i]); + gCommonICUDataArray[i] = nullptr; + } + gHaveTriedToLoadCommonData = 0; + + return true; /* Everything was cleaned up */ +} + +static UBool U_CALLCONV +findCommonICUDataByName(const char *inBasename, UErrorCode &err) +{ + UBool found = false; + int32_t i; + + UDataMemory *pData = udata_findCachedData(inBasename, err); + if (U_FAILURE(err) || pData == nullptr) + return false; + + { + Mutex lock; + for (i = 0; i < UPRV_LENGTHOF(gCommonICUDataArray); ++i) { + if ((gCommonICUDataArray[i] != nullptr) && (gCommonICUDataArray[i]->pHeader == pData->pHeader)) { + /* The data pointer is already in the array. */ + found = true; + break; + } + } + } + return found; +} + + +/* + * setCommonICUData. Set a UDataMemory to be the global ICU Data + */ +static UBool +setCommonICUData(UDataMemory *pData, /* The new common data. Belongs to caller, we copy it. */ + UBool warn, /* If true, set USING_DEFAULT warning if ICUData was */ + /* changed by another thread before we got to it. */ + UErrorCode *pErr) +{ + UDataMemory *newCommonData = UDataMemory_createNewInstance(pErr); + int32_t i; + UBool didUpdate = false; + if (U_FAILURE(*pErr)) { + return false; + } + + /* For the assignment, other threads must cleanly see either the old */ + /* or the new, not some partially initialized new. The old can not be */ + /* deleted - someone may still have a pointer to it lying around in */ + /* their locals. */ + UDatamemory_assign(newCommonData, pData); + umtx_lock(nullptr); + for (i = 0; i < UPRV_LENGTHOF(gCommonICUDataArray); ++i) { + if (gCommonICUDataArray[i] == nullptr) { + gCommonICUDataArray[i] = newCommonData; + didUpdate = true; + break; + } else if (gCommonICUDataArray[i]->pHeader == pData->pHeader) { + /* The same data pointer is already in the array. */ + break; + } + } + umtx_unlock(nullptr); + + if (i == UPRV_LENGTHOF(gCommonICUDataArray) && warn) { + *pErr = U_USING_DEFAULT_WARNING; + } + if (didUpdate) { + ucln_common_registerCleanup(UCLN_COMMON_UDATA, udata_cleanup); + } else { + uprv_free(newCommonData); + } + return didUpdate; +} + +#if !defined(ICU_DATA_DIR_WINDOWS) + +static UBool +setCommonICUDataPointer(const void *pData, UBool /*warn*/, UErrorCode *pErrorCode) { + UDataMemory tData; + UDataMemory_init(&tData); + UDataMemory_setData(&tData, pData); + udata_checkCommonData(&tData, pErrorCode); + return setCommonICUData(&tData, false, pErrorCode); +} + +#endif + +static const char * +findBasename(const char *path) { + const char *basename=uprv_strrchr(path, U_FILE_SEP_CHAR); + if(basename==nullptr) { + return path; + } else { + return basename+1; + } +} + +#ifdef UDATA_DEBUG +static const char * +packageNameFromPath(const char *path) +{ + if((path == nullptr) || (*path == 0)) { + return U_ICUDATA_NAME; + } + + path = findBasename(path); + + if((path == nullptr) || (*path == 0)) { + return U_ICUDATA_NAME; + } + + return path; +} +#endif + +/*----------------------------------------------------------------------* + * * + * Cache for common data * + * Functions for looking up or adding entries to a cache of * + * data that has been previously opened. Avoids a potentially * + * expensive operation of re-opening the data for subsequent * + * uses. * + * * + * Data remains cached for the duration of the process. * + * * + *----------------------------------------------------------------------*/ + +typedef struct DataCacheElement { + char *name; + UDataMemory *item; +} DataCacheElement; + + + +/* + * Deleter function for DataCacheElements. + * udata cleanup function closes the hash table; hash table in turn calls back to + * here for each entry. + */ +static void U_CALLCONV DataCacheElement_deleter(void *pDCEl) { + DataCacheElement *p = (DataCacheElement *)pDCEl; + udata_close(p->item); /* unmaps storage */ + uprv_free(p->name); /* delete the hash key string. */ + uprv_free(pDCEl); /* delete 'this' */ +} + +static void U_CALLCONV udata_initHashTable(UErrorCode &err) { + U_ASSERT(gCommonDataCache == nullptr); + gCommonDataCache = uhash_open(uhash_hashChars, uhash_compareChars, nullptr, &err); + if (U_FAILURE(err)) { + return; + } + U_ASSERT(gCommonDataCache != nullptr); + uhash_setValueDeleter(gCommonDataCache, DataCacheElement_deleter); + ucln_common_registerCleanup(UCLN_COMMON_UDATA, udata_cleanup); +} + + /* udata_getCacheHashTable() + * Get the hash table used to store the data cache entries. + * Lazy create it if it doesn't yet exist. + */ +static UHashtable *udata_getHashTable(UErrorCode &err) { + umtx_initOnce(gCommonDataCacheInitOnce, &udata_initHashTable, err); + return gCommonDataCache; +} + + + +static UDataMemory *udata_findCachedData(const char *path, UErrorCode &err) +{ + UHashtable *htable; + UDataMemory *retVal = nullptr; + DataCacheElement *el; + const char *baseName; + + htable = udata_getHashTable(err); + if (U_FAILURE(err)) { + return nullptr; + } + + baseName = findBasename(path); /* Cache remembers only the base name, not the full path. */ + umtx_lock(nullptr); + el = (DataCacheElement *)uhash_get(htable, baseName); + umtx_unlock(nullptr); + if (el != nullptr) { + retVal = el->item; + } +#ifdef UDATA_DEBUG + fprintf(stderr, "Cache: [%s] -> %p\n", baseName, (void*) retVal); +#endif + return retVal; +} + + +static UDataMemory *udata_cacheDataItem(const char *path, UDataMemory *item, UErrorCode *pErr) { + DataCacheElement *newElement; + const char *baseName; + int32_t nameLen; + UHashtable *htable; + DataCacheElement *oldValue = nullptr; + UErrorCode subErr = U_ZERO_ERROR; + + htable = udata_getHashTable(*pErr); + if (U_FAILURE(*pErr)) { + return nullptr; + } + + /* Create a new DataCacheElement - the thingy we store in the hash table - + * and copy the supplied path and UDataMemoryItems into it. + */ + newElement = (DataCacheElement *)uprv_malloc(sizeof(DataCacheElement)); + if (newElement == nullptr) { + *pErr = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + newElement->item = UDataMemory_createNewInstance(pErr); + if (U_FAILURE(*pErr)) { + uprv_free(newElement); + return nullptr; + } + UDatamemory_assign(newElement->item, item); + + baseName = findBasename(path); + nameLen = (int32_t)uprv_strlen(baseName); + newElement->name = (char *)uprv_malloc(nameLen+1); + if (newElement->name == nullptr) { + *pErr = U_MEMORY_ALLOCATION_ERROR; + uprv_free(newElement->item); + uprv_free(newElement); + return nullptr; + } + uprv_strcpy(newElement->name, baseName); + + /* Stick the new DataCacheElement into the hash table. + */ + umtx_lock(nullptr); + oldValue = (DataCacheElement *)uhash_get(htable, path); + if (oldValue != nullptr) { + subErr = U_USING_DEFAULT_WARNING; + } + else { + uhash_put( + htable, + newElement->name, /* Key */ + newElement, /* Value */ + &subErr); + } + umtx_unlock(nullptr); + +#ifdef UDATA_DEBUG + fprintf(stderr, "Cache: [%s] <<< %p : %s. vFunc=%p\n", newElement->name, + (void*) newElement->item, u_errorName(subErr), (void*) newElement->item->vFuncs); +#endif + + if (subErr == U_USING_DEFAULT_WARNING || U_FAILURE(subErr)) { + *pErr = subErr; /* copy sub err unto fillin ONLY if something happens. */ + uprv_free(newElement->name); + uprv_free(newElement->item); + uprv_free(newElement); + return oldValue ? oldValue->item : nullptr; + } + + return newElement->item; +} + +/*----------------------------------------------------------------------*============== + * * + * Path management. Could be shared with other tools/etc if need be * + * later on. * + * * + *----------------------------------------------------------------------*/ + +U_NAMESPACE_BEGIN + +class UDataPathIterator +{ +public: + UDataPathIterator(const char *path, const char *pkg, + const char *item, const char *suffix, UBool doCheckLastFour, + UErrorCode *pErrorCode); + const char *next(UErrorCode *pErrorCode); + +private: + const char *path; /* working path (u_icudata_Dir) */ + const char *nextPath; /* path following this one */ + const char *basename; /* item's basename (icudt22e_mt.res)*/ + + StringPiece suffix; /* item suffix (can be null) */ + + uint32_t basenameLen; /* length of basename */ + + CharString itemPath; /* path passed in with item name */ + CharString pathBuffer; /* output path for this it'ion */ + CharString packageStub; /* example: "/icudt28b". Will ignore that leaf in set paths. */ + + UBool checkLastFour; /* if true then allow paths such as '/foo/myapp.dat' + * to match, checks last 4 chars of suffix with + * last 4 of path, then previous chars. */ +}; + +/** + * @param iter The iterator to be initialized. Its current state does not matter. + * @param inPath The full pathname to be iterated over. If nullptr, defaults to U_ICUDATA_NAME + * @param pkg Package which is being searched for, ex "icudt28l". Will ignore leaf directories such as /icudt28l + * @param item Item to be searched for. Can include full path, such as /a/b/foo.dat + * @param inSuffix Optional item suffix, if not-null (ex. ".dat") then 'path' can contain 'item' explicitly. + * Ex: 'stuff.dat' would be found in '/a/foo:/tmp/stuff.dat:/bar/baz' as item #2. + * '/blarg/stuff.dat' would also be found. + * Note: inSuffix may also be the 'item' being searched for as well, (ex: "ibm-5348_P100-1997.cnv"), in which case + * the 'item' parameter is often the same as pkg. (Though sometimes might have a tree part as well, ex: "icudt62l-curr"). + */ +UDataPathIterator::UDataPathIterator(const char *inPath, const char *pkg, + const char *item, const char *inSuffix, UBool doCheckLastFour, + UErrorCode *pErrorCode) +{ +#ifdef UDATA_DEBUG + fprintf(stderr, "SUFFIX1=%s PATH=%s\n", inSuffix, inPath); +#endif + /** Path **/ + if(inPath == nullptr) { + path = u_getDataDirectory(); + } else { + path = inPath; + } + + /** Package **/ + if(pkg != nullptr) { + packageStub.append(U_FILE_SEP_CHAR, *pErrorCode).append(pkg, *pErrorCode); +#ifdef UDATA_DEBUG + fprintf(stderr, "STUB=%s [%d]\n", packageStub.data(), packageStub.length()); +#endif + } + + /** Item **/ + basename = findBasename(item); + basenameLen = (int32_t)uprv_strlen(basename); + + /** Item path **/ + if(basename == item) { + nextPath = path; + } else { + itemPath.append(item, (int32_t)(basename-item), *pErrorCode); + nextPath = itemPath.data(); + } +#ifdef UDATA_DEBUG + fprintf(stderr, "SUFFIX=%s [%p]\n", inSuffix, (void*) inSuffix); +#endif + + /** Suffix **/ + if(inSuffix != nullptr) { + suffix = inSuffix; + } else { + suffix = ""; + } + + checkLastFour = doCheckLastFour; + + /* pathBuffer will hold the output path strings returned by this iterator */ + +#ifdef UDATA_DEBUG + fprintf(stderr, "0: init %s -> [path=%s], [base=%s], [suff=%s], [itempath=%s], [nextpath=%s], [checklast4=%s]\n", + item, + path, + basename, + suffix.data(), + itemPath.data(), + nextPath, + checkLastFour?"true":"false"); +#endif +} + +/** + * Get the next path on the list. + * + * @param iter The Iter to be used + * @param len If set, pointer to the length of the returned path, for convenience. + * @return Pointer to the next path segment, or nullptr if there are no more. + */ +const char *UDataPathIterator::next(UErrorCode *pErrorCode) +{ + if(U_FAILURE(*pErrorCode)) { + return nullptr; + } + + const char *currentPath = nullptr; + int32_t pathLen = 0; + const char *pathBasename; + + do + { + if( nextPath == nullptr ) { + break; + } + currentPath = nextPath; + + if(nextPath == itemPath.data()) { /* we were processing item's path. */ + nextPath = path; /* start with regular path next tm. */ + pathLen = (int32_t)uprv_strlen(currentPath); + } else { + /* fix up next for next time */ + nextPath = uprv_strchr(currentPath, U_PATH_SEP_CHAR); + if(nextPath == nullptr) { + /* segment: entire path */ + pathLen = (int32_t)uprv_strlen(currentPath); + } else { + /* segment: until next segment */ + pathLen = (int32_t)(nextPath - currentPath); + /* skip divider */ + nextPath ++; + } + } + + if(pathLen == 0) { + continue; + } + +#ifdef UDATA_DEBUG + fprintf(stderr, "rest of path (IDD) = %s\n", currentPath); + fprintf(stderr, " "); + { + int32_t qqq; + for(qqq=0;qqq=4) && + uprv_strncmp(pathBuffer.data() +(pathLen-4), suffix.data(), 4)==0 && /* suffix matches */ + uprv_strncmp(findBasename(pathBuffer.data()), basename, basenameLen)==0 && /* base matches */ + uprv_strlen(pathBasename)==(basenameLen+4)) { /* base+suffix = full len */ + +#ifdef UDATA_DEBUG + fprintf(stderr, "Have %s file on the path: %s\n", suffix.data(), pathBuffer.data()); +#endif + /* do nothing */ + } + else + { /* regular dir path */ + if(pathBuffer[pathLen-1] != U_FILE_SEP_CHAR) { + if((pathLen>=4) && + uprv_strncmp(pathBuffer.data()+(pathLen-4), ".dat", 4) == 0) + { +#ifdef UDATA_DEBUG + fprintf(stderr, "skipping non-directory .dat file %s\n", pathBuffer.data()); +#endif + continue; + } + + /* Check if it is a directory with the same name as our package */ + if(!packageStub.isEmpty() && + (pathLen > packageStub.length()) && + !uprv_strcmp(pathBuffer.data() + pathLen - packageStub.length(), packageStub.data())) { +#ifdef UDATA_DEBUG + fprintf(stderr, "Found stub %s (will add package %s of len %d)\n", packageStub.data(), basename, basenameLen); +#endif + pathBuffer.truncate(pathLen - packageStub.length()); + } + pathBuffer.append(U_FILE_SEP_CHAR, *pErrorCode); + } + + /* + basename */ + pathBuffer.append(packageStub.data()+1, packageStub.length()-1, *pErrorCode); + + if (!suffix.empty()) /* tack on suffix */ + { + if (suffix.length() > 4) { + // If the suffix is actually an item ("ibm-5348_P100-1997.cnv") and not an extension (".res") + // then we need to ensure that the path ends with a separator. + pathBuffer.ensureEndsWithFileSeparator(*pErrorCode); + } + pathBuffer.append(suffix, *pErrorCode); + } + } + +#ifdef UDATA_DEBUG + fprintf(stderr, " --> %s\n", pathBuffer.data()); +#endif + + return pathBuffer.data(); + + } while(path); + + /* fell way off the end */ + return nullptr; +} + +U_NAMESPACE_END + +/* ==================================================================================*/ + + +/*----------------------------------------------------------------------* + * * + * Add a static reference to the common data library * + * Unless overridden by an explicit udata_setCommonData, this will be * + * our common data. * + * * + *----------------------------------------------------------------------*/ +#if !defined(ICU_DATA_DIR_WINDOWS) +// When using the Windows system data, we expect only a single data file. +extern "C" const DataHeader U_DATA_API U_ICUDATA_ENTRY_POINT; +#endif + +/* + * This would be a good place for weak-linkage declarations of + * partial-data-library access functions where each returns a pointer + * to its data package, if it is linked in. + */ +/* +extern const void *uprv_getICUData_collation() ATTRIBUTE_WEAK; +extern const void *uprv_getICUData_conversion() ATTRIBUTE_WEAK; +*/ + +/*----------------------------------------------------------------------* + * * + * openCommonData Attempt to open a common format (.dat) file * + * Map it into memory (if it's not there already) * + * and return a UDataMemory object for it. * + * * + * If the requested data is already open and cached * + * just return the cached UDataMem object. * + * * + *----------------------------------------------------------------------*/ +static UDataMemory * +openCommonData(const char *path, /* Path from OpenChoice? */ + int32_t commonDataIndex, /* ICU Data (index >= 0) if path == nullptr */ + UErrorCode *pErrorCode) +{ + UDataMemory tData; + const char *pathBuffer; + const char *inBasename; + + if (U_FAILURE(*pErrorCode)) { + return nullptr; + } + + UDataMemory_init(&tData); + + /* ??????? TODO revisit this */ + if (commonDataIndex >= 0) { + /* "mini-cache" for common ICU data */ + if(commonDataIndex >= UPRV_LENGTHOF(gCommonICUDataArray)) { + return nullptr; + } + { + Mutex lock; + if(gCommonICUDataArray[commonDataIndex] != nullptr) { + return gCommonICUDataArray[commonDataIndex]; + } +#if !defined(ICU_DATA_DIR_WINDOWS) +// When using the Windows system data, we expect only a single data file. + int32_t i; + for(i = 0; i < commonDataIndex; ++i) { + if(gCommonICUDataArray[i]->pHeader == &U_ICUDATA_ENTRY_POINT) { + /* The linked-in data is already in the list. */ + return nullptr; + } + } +#endif + } + + /* Add the linked-in data to the list. */ + /* + * This is where we would check and call weakly linked partial-data-library + * access functions. + */ + /* + if (uprv_getICUData_collation) { + setCommonICUDataPointer(uprv_getICUData_collation(), false, pErrorCode); + } + if (uprv_getICUData_conversion) { + setCommonICUDataPointer(uprv_getICUData_conversion(), false, pErrorCode); + } + */ +#if !defined(ICU_DATA_DIR_WINDOWS) +// When using the Windows system data, we expect only a single data file. + setCommonICUDataPointer(&U_ICUDATA_ENTRY_POINT, false, pErrorCode); + { + Mutex lock; + return gCommonICUDataArray[commonDataIndex]; + } +#endif + } + + + /* request is NOT for ICU Data. */ + + /* Find the base name portion of the supplied path. */ + /* inBasename will be left pointing somewhere within the original path string. */ + inBasename = findBasename(path); +#ifdef UDATA_DEBUG + fprintf(stderr, "inBasename = %s\n", inBasename); +#endif + + if(*inBasename==0) { + /* no basename. This will happen if the original path was a directory name, */ + /* like "a/b/c/". (Fallback to separate files will still work.) */ +#ifdef UDATA_DEBUG + fprintf(stderr, "ocd: no basename in %s, bailing.\n", path); +#endif + if (U_SUCCESS(*pErrorCode)) { + *pErrorCode=U_FILE_ACCESS_ERROR; + } + return nullptr; + } + + /* Is the requested common data file already open and cached? */ + /* Note that the cache is keyed by the base name only. The rest of the path, */ + /* if any, is not considered. */ + UDataMemory *dataToReturn = udata_findCachedData(inBasename, *pErrorCode); + if (dataToReturn != nullptr || U_FAILURE(*pErrorCode)) { + return dataToReturn; + } + + /* Requested item is not in the cache. + * Hunt it down, trying all the path locations + */ + + UDataPathIterator iter(u_getDataDirectory(), inBasename, path, ".dat", true, pErrorCode); + + while ((UDataMemory_isLoaded(&tData)==false) && (pathBuffer = iter.next(pErrorCode)) != nullptr) + { +#ifdef UDATA_DEBUG + fprintf(stderr, "ocd: trying path %s - ", pathBuffer); +#endif + uprv_mapFile(&tData, pathBuffer, pErrorCode); +#ifdef UDATA_DEBUG + fprintf(stderr, "%s\n", UDataMemory_isLoaded(&tData)?"LOADED":"not loaded"); +#endif + } + if (U_FAILURE(*pErrorCode)) { + return nullptr; + } + +#if defined(OS390_STUBDATA) && defined(OS390BATCH) + if (!UDataMemory_isLoaded(&tData)) { + char ourPathBuffer[1024]; + /* One more chance, for extendCommonData() */ + uprv_strncpy(ourPathBuffer, path, 1019); + ourPathBuffer[1019]=0; + uprv_strcat(ourPathBuffer, ".dat"); + uprv_mapFile(&tData, ourPathBuffer, pErrorCode); + } +#endif + + if (U_FAILURE(*pErrorCode)) { + return nullptr; + } + if (!UDataMemory_isLoaded(&tData)) { + /* no common data */ + *pErrorCode=U_FILE_ACCESS_ERROR; + return nullptr; + } + + /* we have mapped a file, check its header */ + udata_checkCommonData(&tData, pErrorCode); + + + /* Cache the UDataMemory struct for this .dat file, + * so we won't need to hunt it down and map it again next time + * something is needed from it. */ + return udata_cacheDataItem(inBasename, &tData, pErrorCode); +} + + +/*----------------------------------------------------------------------* + * * + * extendICUData If the full set of ICU data was not loaded at * + * program startup, load it now. This function will * + * be called when the lookup of an ICU data item in * + * the common ICU data fails. * + * * + * return true if new data is loaded, false otherwise.* + * * + *----------------------------------------------------------------------*/ +static UBool extendICUData(UErrorCode *pErr) +{ + UDataMemory *pData; + UDataMemory copyPData; + UBool didUpdate = false; + + /* + * There is a chance for a race condition here. + * Normally, ICU data is loaded from a DLL or via mmap() and + * setCommonICUData() will detect if the same address is set twice. + * If ICU is built with data loading via fread() then the address will + * be different each time the common data is loaded and we may add + * multiple copies of the data. + * In this case, use a mutex to prevent the race. + * Use a specific mutex to avoid nested locks of the global mutex. + */ +#if MAP_IMPLEMENTATION==MAP_STDIO + static UMutex extendICUDataMutex; + umtx_lock(&extendICUDataMutex); +#endif + if(!umtx_loadAcquire(gHaveTriedToLoadCommonData)) { + /* See if we can explicitly open a .dat file for the ICUData. */ + pData = openCommonData( + U_ICUDATA_NAME, /* "icudt20l" , for example. */ + -1, /* Pretend we're not opening ICUData */ + pErr); + + /* How about if there is no pData, eh... */ + + UDataMemory_init(©PData); + if(pData != nullptr) { + UDatamemory_assign(©PData, pData); + copyPData.map = 0; /* The mapping for this data is owned by the hash table */ + copyPData.mapAddr = 0; /* which will unmap it when ICU is shut down. */ + /* CommonICUData is also unmapped when ICU is shut down.*/ + /* To avoid unmapping the data twice, zero out the map */ + /* fields in the UDataMemory that we're assigning */ + /* to CommonICUData. */ + + didUpdate = /* no longer using this result */ + setCommonICUData(©PData,/* The new common data. */ + false, /* No warnings if write didn't happen */ + pErr); /* setCommonICUData honors errors; NOP if error set */ + } + + umtx_storeRelease(gHaveTriedToLoadCommonData, 1); + } + + didUpdate = findCommonICUDataByName(U_ICUDATA_NAME, *pErr); /* Return 'true' when a racing writes out the extended */ + /* data after another thread has failed to see it (in openCommonData), so */ + /* extended data can be examined. */ + /* Also handles a race through here before gHaveTriedToLoadCommonData is set. */ + +#if MAP_IMPLEMENTATION==MAP_STDIO + umtx_unlock(&extendICUDataMutex); +#endif + return didUpdate; /* Return true if ICUData pointer was updated. */ + /* (Could potentially have been done by another thread racing */ + /* us through here, but that's fine, we still return true */ + /* so that current thread will also examine extended data. */ +} + +/*----------------------------------------------------------------------* + * * + * udata_setCommonData * + * * + *----------------------------------------------------------------------*/ +U_CAPI void U_EXPORT2 +udata_setCommonData(const void *data, UErrorCode *pErrorCode) { + UDataMemory dataMemory; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return; + } + + if(data==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + /* set the data pointer and test for validity */ + UDataMemory_init(&dataMemory); + UDataMemory_setData(&dataMemory, data); + udata_checkCommonData(&dataMemory, pErrorCode); + if (U_FAILURE(*pErrorCode)) {return;} + + /* we have good data */ + /* Set it up as the ICU Common Data. */ + setCommonICUData(&dataMemory, true, pErrorCode); +} + +/*--------------------------------------------------------------------------- + * + * udata_setAppData + * + *---------------------------------------------------------------------------- */ +U_CAPI void U_EXPORT2 +udata_setAppData(const char *path, const void *data, UErrorCode *err) +{ + UDataMemory udm; + + if(err==nullptr || U_FAILURE(*err)) { + return; + } + if(data==nullptr) { + *err=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + UDataMemory_init(&udm); + UDataMemory_setData(&udm, data); + udata_checkCommonData(&udm, err); + udata_cacheDataItem(path, &udm, err); +} + +/*----------------------------------------------------------------------------* + * * + * checkDataItem Given a freshly located/loaded data item, either * + * an entry in a common file or a separately loaded file, * + * sanity check its header, and see if the data is * + * acceptable to the app. * + * If the data is good, create and return a UDataMemory * + * object that can be returned to the application. * + * Return nullptr on any sort of failure. * + * * + *----------------------------------------------------------------------------*/ +static UDataMemory * +checkDataItem +( + const DataHeader *pHeader, /* The data item to be checked. */ + UDataMemoryIsAcceptable *isAcceptable, /* App's call-back function */ + void *context, /* pass-thru param for above. */ + const char *type, /* pass-thru param for above. */ + const char *name, /* pass-thru param for above. */ + UErrorCode *nonFatalErr, /* Error code if this data was not acceptable */ + /* but openChoice should continue with */ + /* trying to get data from fallback path. */ + UErrorCode *fatalErr /* Bad error, caller should return immediately */ + ) +{ + UDataMemory *rDataMem = nullptr; /* the new UDataMemory, to be returned. */ + + if (U_FAILURE(*fatalErr)) { + return nullptr; + } + + if(pHeader->dataHeader.magic1==0xda && + pHeader->dataHeader.magic2==0x27 && + (isAcceptable==nullptr || isAcceptable(context, type, name, &pHeader->info)) + ) { + rDataMem=UDataMemory_createNewInstance(fatalErr); + if (U_FAILURE(*fatalErr)) { + return nullptr; + } + rDataMem->pHeader = pHeader; + } else { + /* the data is not acceptable, look further */ + /* If we eventually find something good, this errorcode will be */ + /* cleared out. */ + *nonFatalErr=U_INVALID_FORMAT_ERROR; + } + return rDataMem; +} + +/** + * @return 0 if not loaded, 1 if loaded or err + */ +static UDataMemory *doLoadFromIndividualFiles(const char *pkgName, + const char *dataPath, const char *tocEntryPathSuffix, + /* following arguments are the same as doOpenChoice itself */ + const char *path, const char *type, const char *name, + UDataMemoryIsAcceptable *isAcceptable, void *context, + UErrorCode *subErrorCode, + UErrorCode *pErrorCode) +{ + const char *pathBuffer; + UDataMemory dataMemory; + UDataMemory *pEntryData; + + /* look in ind. files: package\nam.typ ========================= */ + /* init path iterator for individual files */ + UDataPathIterator iter(dataPath, pkgName, path, tocEntryPathSuffix, false, pErrorCode); + + while ((pathBuffer = iter.next(pErrorCode)) != nullptr) + { +#ifdef UDATA_DEBUG + fprintf(stderr, "UDATA: trying individual file %s\n", pathBuffer); +#endif + if (uprv_mapFile(&dataMemory, pathBuffer, pErrorCode)) + { + pEntryData = checkDataItem(dataMemory.pHeader, isAcceptable, context, type, name, subErrorCode, pErrorCode); + if (pEntryData != nullptr) { + /* Data is good. + * Hand off ownership of the backing memory to the user's UDataMemory. + * and return it. */ + pEntryData->mapAddr = dataMemory.mapAddr; + pEntryData->map = dataMemory.map; + +#ifdef UDATA_DEBUG + fprintf(stderr, "** Mapped file: %s\n", pathBuffer); +#endif + return pEntryData; + } + + /* the data is not acceptable, or some error occurred. Either way, unmap the memory */ + udata_close(&dataMemory); + + /* If we had a nasty error, bail out completely. */ + if (U_FAILURE(*pErrorCode)) { + return nullptr; + } + + /* Otherwise remember that we found data but didn't like it for some reason */ + *subErrorCode=U_INVALID_FORMAT_ERROR; + } +#ifdef UDATA_DEBUG + fprintf(stderr, "%s\n", UDataMemory_isLoaded(&dataMemory)?"LOADED":"not loaded"); +#endif + } + return nullptr; +} + +/** + * @return 0 if not loaded, 1 if loaded or err + */ +static UDataMemory *doLoadFromCommonData(UBool isICUData, const char * /*pkgName*/, + const char * /*dataPath*/, const char * /*tocEntryPathSuffix*/, const char *tocEntryName, + /* following arguments are the same as doOpenChoice itself */ + const char *path, const char *type, const char *name, + UDataMemoryIsAcceptable *isAcceptable, void *context, + UErrorCode *subErrorCode, + UErrorCode *pErrorCode) +{ + UDataMemory *pEntryData; + const DataHeader *pHeader; + UDataMemory *pCommonData; + int32_t commonDataIndex; + UBool checkedExtendedICUData = false; + /* try to get common data. The loop is for platforms such as the 390 that do + * not initially load the full set of ICU data. If the lookup of an ICU data item + * fails, the full (but slower to load) set is loaded, the and the loop repeats, + * trying the lookup again. Once the full set of ICU data is loaded, the loop wont + * repeat because the full set will be checked the first time through. + * + * The loop also handles the fallback to a .dat file if the application linked + * to the stub data library rather than a real library. + */ + for (commonDataIndex = isICUData ? 0 : -1;;) { + pCommonData=openCommonData(path, commonDataIndex, subErrorCode); /** search for pkg **/ + + if(U_SUCCESS(*subErrorCode) && pCommonData!=nullptr) { + int32_t length; + + /* look up the data piece in the common data */ + pHeader=pCommonData->vFuncs->Lookup(pCommonData, tocEntryName, &length, subErrorCode); +#ifdef UDATA_DEBUG + fprintf(stderr, "%s: pHeader=%p - %s\n", tocEntryName, (void*) pHeader, u_errorName(*subErrorCode)); +#endif + + if(pHeader!=nullptr) { + pEntryData = checkDataItem(pHeader, isAcceptable, context, type, name, subErrorCode, pErrorCode); +#ifdef UDATA_DEBUG + fprintf(stderr, "pEntryData=%p\n", (void*) pEntryData); +#endif + if (U_FAILURE(*pErrorCode)) { + return nullptr; + } + if (pEntryData != nullptr) { + pEntryData->length = length; + return pEntryData; + } + } + } + // If we failed due to being out-of-memory, then stop early and report the error. + if (*subErrorCode == U_MEMORY_ALLOCATION_ERROR) { + *pErrorCode = *subErrorCode; + return nullptr; + } + /* Data wasn't found. If we were looking for an ICUData item and there is + * more data available, load it and try again, + * otherwise break out of this loop. */ + if (!isICUData) { + return nullptr; + } else if (pCommonData != nullptr) { + ++commonDataIndex; /* try the next data package */ + } else if ((!checkedExtendedICUData) && extendICUData(subErrorCode)) { + checkedExtendedICUData = true; + /* try this data package slot again: it changed from nullptr to non-nullptr */ + } else { + return nullptr; + } + } +} + +/* + * Identify the Time Zone resources that are subject to special override data loading. + */ +static UBool isTimeZoneFile(const char *name, const char *type) { + return ((uprv_strcmp(type, "res") == 0) && + (uprv_strcmp(name, "zoneinfo64") == 0 || + uprv_strcmp(name, "timezoneTypes") == 0 || + uprv_strcmp(name, "windowsZones") == 0 || + uprv_strcmp(name, "metaZones") == 0)); +} + +/* + * A note on the ownership of Mapped Memory + * + * For common format files, ownership resides with the UDataMemory object + * that lives in the cache of opened common data. These UDataMemorys are private + * to the udata implementation, and are never seen directly by users. + * + * The UDataMemory objects returned to users will have the address of some desired + * data within the mapped region, but they wont have the mapping info itself, and thus + * won't cause anything to be removed from memory when they are closed. + * + * For individual data files, the UDataMemory returned to the user holds the + * information necessary to unmap the data on close. If the user independently + * opens the same data file twice, two completely independent mappings will be made. + * (There is no cache of opened data items from individual files, only a cache of + * opened Common Data files, that is, files containing a collection of data items.) + * + * For common data passed in from the user via udata_setAppData() or + * udata_setCommonData(), ownership remains with the user. + * + * UDataMemory objects themselves, as opposed to the memory they describe, + * can be anywhere - heap, stack/local or global. + * They have a flag to indicate when they're heap allocated and thus + * must be deleted when closed. + */ + + +/*----------------------------------------------------------------------------* + * * + * main data loading functions * + * * + *----------------------------------------------------------------------------*/ +static UDataMemory * +doOpenChoice(const char *path, const char *type, const char *name, + UDataMemoryIsAcceptable *isAcceptable, void *context, + UErrorCode *pErrorCode) +{ + UDataMemory *retVal = nullptr; + + const char *dataPath; + + int32_t tocEntrySuffixIndex; + const char *tocEntryPathSuffix; + UErrorCode subErrorCode=U_ZERO_ERROR; + const char *treeChar; + + UBool isICUData = false; + + + FileTracer::traceOpen(path, type, name); + + + /* Is this path ICU data? */ + if(path == nullptr || + !strcmp(path, U_ICUDATA_ALIAS) || /* "ICUDATA" */ + !uprv_strncmp(path, U_ICUDATA_NAME U_TREE_SEPARATOR_STRING, /* "icudt26e-" */ + uprv_strlen(U_ICUDATA_NAME U_TREE_SEPARATOR_STRING)) || + !uprv_strncmp(path, U_ICUDATA_ALIAS U_TREE_SEPARATOR_STRING, /* "ICUDATA-" */ + uprv_strlen(U_ICUDATA_ALIAS U_TREE_SEPARATOR_STRING))) { + isICUData = true; + } + +#if (U_FILE_SEP_CHAR != U_FILE_ALT_SEP_CHAR) /* Windows: try "foo\bar" and "foo/bar" */ + /* remap from alternate path char to the main one */ + CharString altSepPath; + if(path) { + if(uprv_strchr(path,U_FILE_ALT_SEP_CHAR) != nullptr) { + altSepPath.append(path, *pErrorCode); + char *p; + while ((p = uprv_strchr(altSepPath.data(), U_FILE_ALT_SEP_CHAR)) != nullptr) { + *p = U_FILE_SEP_CHAR; + } +#if defined (UDATA_DEBUG) + fprintf(stderr, "Changed path from [%s] to [%s]\n", path, altSepPath.s); +#endif + path = altSepPath.data(); + } + } +#endif + + CharString tocEntryName; /* entry name in tree format. ex: 'icudt28b/coll/ar.res' */ + CharString tocEntryPath; /* entry name in path format. ex: 'icudt28b\\coll\\ar.res' */ + + CharString pkgName; + CharString treeName; + + /* ======= Set up strings */ + if(path==nullptr) { + pkgName.append(U_ICUDATA_NAME, *pErrorCode); + } else { + const char *pkg; + const char *first; + pkg = uprv_strrchr(path, U_FILE_SEP_CHAR); + first = uprv_strchr(path, U_FILE_SEP_CHAR); + if(uprv_pathIsAbsolute(path) || (pkg != first)) { /* more than one slash in the path- not a tree name */ + /* see if this is an /absolute/path/to/package path */ + if(pkg) { + pkgName.append(pkg+1, *pErrorCode); + } else { + pkgName.append(path, *pErrorCode); + } + } else { + treeChar = uprv_strchr(path, U_TREE_SEPARATOR); + if(treeChar) { + treeName.append(treeChar+1, *pErrorCode); /* following '-' */ + if(isICUData) { + pkgName.append(U_ICUDATA_NAME, *pErrorCode); + } else { + pkgName.append(path, (int32_t)(treeChar-path), *pErrorCode); + if (first == nullptr) { + /* + This user data has no path, but there is a tree name. + Look up the correct path from the data cache later. + */ + path = pkgName.data(); + } + } + } else { + if(isICUData) { + pkgName.append(U_ICUDATA_NAME, *pErrorCode); + } else { + pkgName.append(path, *pErrorCode); + } + } + } + } + +#ifdef UDATA_DEBUG + fprintf(stderr, " P=%s T=%s\n", pkgName.data(), treeName.data()); +#endif + + /* setting up the entry name and file name + * Make up a full name by appending the type to the supplied + * name, assuming that a type was supplied. + */ + + /* prepend the package */ + tocEntryName.append(pkgName, *pErrorCode); + tocEntryPath.append(pkgName, *pErrorCode); + tocEntrySuffixIndex = tocEntryName.length(); + + if(!treeName.isEmpty()) { + tocEntryName.append(U_TREE_ENTRY_SEP_CHAR, *pErrorCode).append(treeName, *pErrorCode); + tocEntryPath.append(U_FILE_SEP_CHAR, *pErrorCode).append(treeName, *pErrorCode); + } + + tocEntryName.append(U_TREE_ENTRY_SEP_CHAR, *pErrorCode).append(name, *pErrorCode); + tocEntryPath.append(U_FILE_SEP_CHAR, *pErrorCode).append(name, *pErrorCode); + if(type!=nullptr && *type!=0) { + tocEntryName.append(".", *pErrorCode).append(type, *pErrorCode); + tocEntryPath.append(".", *pErrorCode).append(type, *pErrorCode); + } + // The +1 is for the U_FILE_SEP_CHAR that is always appended above. + tocEntryPathSuffix = tocEntryPath.data() + tocEntrySuffixIndex + 1; /* suffix starts here */ + +#ifdef UDATA_DEBUG + fprintf(stderr, " tocEntryName = %s\n", tocEntryName.data()); + fprintf(stderr, " tocEntryPath = %s\n", tocEntryName.data()); +#endif + +#if !defined(ICU_DATA_DIR_WINDOWS) + if(path == nullptr) { + path = COMMON_DATA_NAME; /* "icudt26e" */ + } +#else + // When using the Windows system data, we expects only a single data file. + path = COMMON_DATA_NAME; /* "icudt26e" */ +#endif + + /************************ Begin loop looking for ind. files ***************/ +#ifdef UDATA_DEBUG + fprintf(stderr, "IND: inBasename = %s, pkg=%s\n", "(n/a)", packageNameFromPath(path)); +#endif + + /* End of dealing with a null basename */ + dataPath = u_getDataDirectory(); + + /**** Time zone individual files override */ + if (isICUData && isTimeZoneFile(name, type)) { + const char *tzFilesDir = u_getTimeZoneFilesDirectory(pErrorCode); + if (tzFilesDir[0] != 0) { +#ifdef UDATA_DEBUG + fprintf(stderr, "Trying Time Zone Files directory = %s\n", tzFilesDir); +#endif + retVal = doLoadFromIndividualFiles(/* pkgName.data() */ "", tzFilesDir, tocEntryPathSuffix, + /* path */ "", type, name, isAcceptable, context, &subErrorCode, pErrorCode); + if((retVal != nullptr) || U_FAILURE(*pErrorCode)) { + return retVal; + } + } + } + + /**** COMMON PACKAGE - only if packages are first. */ + if(gDataFileAccess == UDATA_PACKAGES_FIRST) { +#ifdef UDATA_DEBUG + fprintf(stderr, "Trying packages (UDATA_PACKAGES_FIRST)\n"); +#endif + /* #2 */ + retVal = doLoadFromCommonData(isICUData, + pkgName.data(), dataPath, tocEntryPathSuffix, tocEntryName.data(), + path, type, name, isAcceptable, context, &subErrorCode, pErrorCode); + if((retVal != nullptr) || U_FAILURE(*pErrorCode)) { + return retVal; + } + } + + /**** INDIVIDUAL FILES */ + if((gDataFileAccess==UDATA_PACKAGES_FIRST) || + (gDataFileAccess==UDATA_FILES_FIRST)) { +#ifdef UDATA_DEBUG + fprintf(stderr, "Trying individual files\n"); +#endif + /* Check to make sure that there is a dataPath to iterate over */ + if ((dataPath && *dataPath) || !isICUData) { + retVal = doLoadFromIndividualFiles(pkgName.data(), dataPath, tocEntryPathSuffix, + path, type, name, isAcceptable, context, &subErrorCode, pErrorCode); + if((retVal != nullptr) || U_FAILURE(*pErrorCode)) { + return retVal; + } + } + } + + /**** COMMON PACKAGE */ + if((gDataFileAccess==UDATA_ONLY_PACKAGES) || + (gDataFileAccess==UDATA_FILES_FIRST)) { +#ifdef UDATA_DEBUG + fprintf(stderr, "Trying packages (UDATA_ONLY_PACKAGES || UDATA_FILES_FIRST)\n"); +#endif + retVal = doLoadFromCommonData(isICUData, + pkgName.data(), dataPath, tocEntryPathSuffix, tocEntryName.data(), + path, type, name, isAcceptable, context, &subErrorCode, pErrorCode); + if((retVal != nullptr) || U_FAILURE(*pErrorCode)) { + return retVal; + } + } + + /* Load from DLL. If we haven't attempted package load, we also haven't had any chance to + try a DLL (static or setCommonData/etc) load. + If we ever have a "UDATA_ONLY_FILES", add it to the or list here. */ + if(gDataFileAccess==UDATA_NO_FILES) { +#ifdef UDATA_DEBUG + fprintf(stderr, "Trying common data (UDATA_NO_FILES)\n"); +#endif + retVal = doLoadFromCommonData(isICUData, + pkgName.data(), "", tocEntryPathSuffix, tocEntryName.data(), + path, type, name, isAcceptable, context, &subErrorCode, pErrorCode); + if((retVal != nullptr) || U_FAILURE(*pErrorCode)) { + return retVal; + } + } + + /* data not found */ + if(U_SUCCESS(*pErrorCode)) { + if(U_SUCCESS(subErrorCode)) { + /* file not found */ + *pErrorCode=U_FILE_ACCESS_ERROR; + } else { + /* entry point not found or rejected */ + *pErrorCode=subErrorCode; + } + } + return retVal; +} + + + +/* API ---------------------------------------------------------------------- */ + +U_CAPI UDataMemory * U_EXPORT2 +udata_open(const char *path, const char *type, const char *name, + UErrorCode *pErrorCode) { +#ifdef UDATA_DEBUG + fprintf(stderr, "udata_open(): Opening: %s : %s . %s\n", (path?path:"nullptr"), name, type); + fflush(stderr); +#endif + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return nullptr; + } else if(name==nullptr || *name==0) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } else { + return doOpenChoice(path, type, name, nullptr, nullptr, pErrorCode); + } +} + + + +U_CAPI UDataMemory * U_EXPORT2 +udata_openChoice(const char *path, const char *type, const char *name, + UDataMemoryIsAcceptable *isAcceptable, void *context, + UErrorCode *pErrorCode) { +#ifdef UDATA_DEBUG + fprintf(stderr, "udata_openChoice(): Opening: %s : %s . %s\n", (path?path:"nullptr"), name, type); +#endif + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return nullptr; + } else if(name==nullptr || *name==0 || isAcceptable==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } else { + return doOpenChoice(path, type, name, isAcceptable, context, pErrorCode); + } +} + + + +U_CAPI void U_EXPORT2 +udata_getInfo(UDataMemory *pData, UDataInfo *pInfo) { + if(pInfo!=nullptr) { + if(pData!=nullptr && pData->pHeader!=nullptr) { + const UDataInfo *info=&pData->pHeader->info; + uint16_t dataInfoSize=udata_getInfoSize(info); + if(pInfo->size>dataInfoSize) { + pInfo->size=dataInfoSize; + } + uprv_memcpy((uint16_t *)pInfo+1, (const uint16_t *)info+1, pInfo->size-2); + if(info->isBigEndian!=U_IS_BIG_ENDIAN) { + /* opposite endianness */ + uint16_t x=info->reservedWord; + pInfo->reservedWord=(uint16_t)((x<<8)|(x>>8)); + } + } else { + pInfo->size=0; + } + } +} + + +U_CAPI void U_EXPORT2 udata_setFileAccess(UDataFileAccess access, UErrorCode * /*status*/) +{ + // Note: this function is documented as not thread safe. + gDataFileAccess = access; +} diff --git a/intl/icu/source/common/udatamem.cpp b/intl/icu/source/common/udatamem.cpp new file mode 100644 index 0000000000..02be852320 --- /dev/null +++ b/intl/icu/source/common/udatamem.cpp @@ -0,0 +1,161 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2011, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************/ + + +/*---------------------------------------------------------------------------------- + * + * UDataMemory A class-like struct that serves as a handle to a piece of memory + * that contains some ICU data (resource, converters, whatever.) + * + * When an application opens ICU data (with udata_open, for example, + * a UDataMemory * is returned. + * + *----------------------------------------------------------------------------------*/ + +#include "unicode/utypes.h" +#include "cmemory.h" +#include "unicode/udata.h" + +#include "udatamem.h" + +U_CFUNC void UDataMemory_init(UDataMemory *This) { + uprv_memset(This, 0, sizeof(UDataMemory)); + This->length=-1; +} + + +U_CFUNC void UDatamemory_assign(UDataMemory *dest, UDataMemory *source) { + /* UDataMemory Assignment. Destination UDataMemory must be initialized first. */ + UBool mallocedFlag = dest->heapAllocated; + uprv_memcpy(dest, source, sizeof(UDataMemory)); + dest->heapAllocated = mallocedFlag; +} + +U_CFUNC UDataMemory *UDataMemory_createNewInstance(UErrorCode *pErr) { + UDataMemory *This; + + if (U_FAILURE(*pErr)) { + return nullptr; + } + This = (UDataMemory *)uprv_malloc(sizeof(UDataMemory)); + if (This == nullptr) { + *pErr = U_MEMORY_ALLOCATION_ERROR; } + else { + UDataMemory_init(This); + This->heapAllocated = true; + } + return This; +} + + +U_CFUNC const DataHeader * +UDataMemory_normalizeDataPointer(const void *p) { + /* allow the data to be optionally prepended with an alignment-forcing double value */ + const DataHeader *pdh = (const DataHeader *)p; + if(pdh==nullptr || (pdh->dataHeader.magic1==0xda && pdh->dataHeader.magic2==0x27)) { + return pdh; + } else { +#if U_PLATFORM == U_PF_OS400 + /* + TODO: Fix this once the compiler implements this feature. Keep in sync with genccode.c + + This is here because this platform can't currently put + const data into the read-only pages of an object or + shared library (service program). Only strings are allowed in read-only + pages, so we use char * strings to store the data. + + In order to prevent the beginning of the data from ever matching the + magic numbers we must skip the initial double. + [grhoten 4/24/2003] + */ + return (const DataHeader *)*((const void **)p+1); +#else + return (const DataHeader *)((const double *)p+1); +#endif + } +} + + +U_CFUNC void UDataMemory_setData (UDataMemory *This, const void *dataAddr) { + This->pHeader = UDataMemory_normalizeDataPointer(dataAddr); +} + + +U_CAPI void U_EXPORT2 +udata_close(UDataMemory *pData) { + if(pData!=nullptr) { + uprv_unmapFile(pData); + if(pData->heapAllocated ) { + uprv_free(pData); + } else { + UDataMemory_init(pData); + } + } +} + +U_CAPI const void * U_EXPORT2 +udata_getMemory(UDataMemory *pData) { + if(pData!=nullptr && pData->pHeader!=nullptr) { + return (char *)(pData->pHeader)+udata_getHeaderSize(pData->pHeader); + } else { + return nullptr; + } +} + +/** + * Get the length of the data item if possible. + * The length may be up to 15 bytes larger than the actual data. + * + * TODO Consider making this function public. + * It would have to return the actual length in more cases. + * For example, the length of the last item in a .dat package could be + * computed from the size of the whole .dat package minus the offset of the + * last item. + * The size of a file that was directly memory-mapped could be determined + * using some system API. + * + * In order to get perfect values for all data items, we may have to add a + * length field to UDataInfo, but that complicates data generation + * and may be overkill. + * + * @param pData The data item. + * @return the length of the data item, or -1 if not known + * @internal Currently used only in cintltst/udatatst.c + */ +U_CAPI int32_t U_EXPORT2 +udata_getLength(const UDataMemory *pData) { + if(pData!=nullptr && pData->pHeader!=nullptr && pData->length>=0) { + /* + * subtract the header size, + * return only the size of the actual data starting at udata_getMemory() + */ + return pData->length-udata_getHeaderSize(pData->pHeader); + } else { + return -1; + } +} + +/** + * Get the memory including the data header. + * Used in cintltst/udatatst.c + * @internal + */ +U_CAPI const void * U_EXPORT2 +udata_getRawMemory(const UDataMemory *pData) { + if(pData!=nullptr && pData->pHeader!=nullptr) { + return pData->pHeader; + } else { + return nullptr; + } +} + +U_CFUNC UBool UDataMemory_isLoaded(const UDataMemory *This) { + return This->pHeader != nullptr; +} diff --git a/intl/icu/source/common/udatamem.h b/intl/icu/source/common/udatamem.h new file mode 100644 index 0000000000..3db2af43aa --- /dev/null +++ b/intl/icu/source/common/udatamem.h @@ -0,0 +1,61 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2010, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************/ + + +/*---------------------------------------------------------------------------------- + * + * UDataMemory A class-like struct that serves as a handle to a piece of memory + * that contains some ICU data (resource, converters, whatever.) + * + * When an application opens ICU data (with udata_open, for example, + * a UDataMemory * is returned. + * + *----------------------------------------------------------------------------------*/ +#ifndef __UDATAMEM_H__ +#define __UDATAMEM_H__ + +#include "unicode/udata.h" +#include "ucmndata.h" + +struct UDataMemory { + const commonDataFuncs *vFuncs; /* Function Pointers for accessing TOC */ + + const DataHeader *pHeader; /* Header of the memory being described by this */ + /* UDataMemory object. */ + const void *toc; /* For common memory, table of contents for */ + /* the pieces within. */ + UBool heapAllocated; /* True if this UDataMemory Object is on the */ + /* heap and thus needs to be deleted when closed. */ + + void *mapAddr; /* For mapped or allocated memory, the start addr. */ + /* Only non-null if a close operation should unmap */ + /* the associated data. */ + void *map; /* Handle, or other data, OS dependent. */ + /* Only non-null if a close operation should unmap */ + /* the associated data, and additional info */ + /* beyond the mapAddr is needed to do that. */ + int32_t length; /* Length of the data in bytes; -1 if unknown. */ +}; + +U_CAPI UDataMemory* U_EXPORT2 UDataMemory_createNewInstance(UErrorCode *pErr); +U_CFUNC void UDatamemory_assign (UDataMemory *dest, UDataMemory *source); +U_CFUNC void UDataMemory_init (UDataMemory *This); +U_CFUNC UBool UDataMemory_isLoaded(const UDataMemory *This); +U_CFUNC void UDataMemory_setData (UDataMemory *This, const void *dataAddr); + +U_CFUNC const DataHeader *UDataMemory_normalizeDataPointer(const void *p); + +U_CAPI int32_t U_EXPORT2 +udata_getLength(const UDataMemory *pData); + +U_CAPI const void * U_EXPORT2 +udata_getRawMemory(const UDataMemory *pData); + +#endif diff --git a/intl/icu/source/common/udataswp.cpp b/intl/icu/source/common/udataswp.cpp new file mode 100644 index 0000000000..bbce1a8ecd --- /dev/null +++ b/intl/icu/source/common/udataswp.cpp @@ -0,0 +1,473 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2003-2014, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: udataswp.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2003jun05 +* created by: Markus W. Scherer +* +* Definitions for ICU data transformations for different platforms, +* changing between big- and little-endian data and/or between +* charset families (ASCII<->EBCDIC). +*/ + +#include +#include "unicode/utypes.h" +#include "unicode/udata.h" /* UDataInfo */ +#include "ucmndata.h" /* DataHeader */ +#include "cmemory.h" +#include "udataswp.h" + +/* swapping primitives ------------------------------------------------------ */ + +static int32_t U_CALLCONV +uprv_swapArray16(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const uint16_t *p; + uint16_t *q; + int32_t count; + uint16_t x; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<0 || (length&1)!=0 || outData==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* setup and swapping */ + p=(const uint16_t *)inData; + q=(uint16_t *)outData; + count=length/2; + while(count>0) { + x=*p++; + *q++=(uint16_t)((x<<8)|(x>>8)); + --count; + } + + return length; +} + +static int32_t U_CALLCONV +uprv_copyArray16(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<0 || (length&1)!=0 || outData==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + if(length>0 && inData!=outData) { + uprv_memcpy(outData, inData, length); + } + return length; +} + +static int32_t U_CALLCONV +uprv_swapArray32(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const uint32_t *p; + uint32_t *q; + int32_t count; + uint32_t x; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<0 || (length&3)!=0 || outData==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* setup and swapping */ + p=(const uint32_t *)inData; + q=(uint32_t *)outData; + count=length/4; + while(count>0) { + x=*p++; + *q++=(uint32_t)((x<<24)|((x<<8)&0xff0000)|((x>>8)&0xff00)|(x>>24)); + --count; + } + + return length; +} + +static int32_t U_CALLCONV +uprv_copyArray32(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<0 || (length&3)!=0 || outData==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + if(length>0 && inData!=outData) { + uprv_memcpy(outData, inData, length); + } + return length; +} + +static int32_t U_CALLCONV +uprv_swapArray64(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const uint64_t *p; + uint64_t *q; + int32_t count; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<0 || (length&7)!=0 || outData==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* setup and swapping */ + p=(const uint64_t *)inData; + q=(uint64_t *)outData; + count=length/8; + while(count>0) { + uint64_t x=*p++; + x=(x<<56)|((x&0xff00)<<40)|((x&0xff0000)<<24)|((x&0xff000000)<<8)| + ((x>>8)&0xff000000)|((x>>24)&0xff0000)|((x>>40)&0xff00)|(x>>56); + *q++=x; + --count; + } + + return length; +} + +static int32_t U_CALLCONV +uprv_copyArray64(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<0 || (length&7)!=0 || outData==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + if(length>0 && inData!=outData) { + uprv_memcpy(outData, inData, length); + } + return length; +} + +static uint16_t U_CALLCONV +uprv_readSwapUInt16(uint16_t x) { + return (uint16_t)((x<<8)|(x>>8)); +} + +static uint16_t U_CALLCONV +uprv_readDirectUInt16(uint16_t x) { + return x; +} + +static uint32_t U_CALLCONV +uprv_readSwapUInt32(uint32_t x) { + return (uint32_t)((x<<24)|((x<<8)&0xff0000)|((x>>8)&0xff00)|(x>>24)); +} + +static uint32_t U_CALLCONV +uprv_readDirectUInt32(uint32_t x) { + return x; +} + +static void U_CALLCONV +uprv_writeSwapUInt16(uint16_t *p, uint16_t x) { + *p=(uint16_t)((x<<8)|(x>>8)); +} + +static void U_CALLCONV +uprv_writeDirectUInt16(uint16_t *p, uint16_t x) { + *p=x; +} + +static void U_CALLCONV +uprv_writeSwapUInt32(uint32_t *p, uint32_t x) { + *p=(uint32_t)((x<<24)|((x<<8)&0xff0000)|((x>>8)&0xff00)|(x>>24)); +} + +static void U_CALLCONV +uprv_writeDirectUInt32(uint32_t *p, uint32_t x) { + *p=x; +} + +U_CAPI int16_t U_EXPORT2 +udata_readInt16(const UDataSwapper *ds, int16_t x) { + return (int16_t)ds->readUInt16((uint16_t)x); +} + +U_CAPI int32_t U_EXPORT2 +udata_readInt32(const UDataSwapper *ds, int32_t x) { + return (int32_t)ds->readUInt32((uint32_t)x); +} + +/** + * Swap a block of invariant, NUL-terminated strings, but not padding + * bytes after the last string. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +udata_swapInvStringBlock(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const char *inChars; + int32_t stringsLength; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<0 || (length>0 && outData==nullptr)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* reduce the strings length to not include bytes after the last NUL */ + inChars=(const char *)inData; + stringsLength=length; + while(stringsLength>0 && inChars[stringsLength-1]!=0) { + --stringsLength; + } + + /* swap up to the last NUL */ + ds->swapInvChars(ds, inData, stringsLength, outData, pErrorCode); + + /* copy the bytes after the last NUL */ + if(inData!=outData && length>stringsLength) { + uprv_memcpy((char *)outData+stringsLength, inChars+stringsLength, length-stringsLength); + } + + /* return the length including padding bytes */ + if(U_SUCCESS(*pErrorCode)) { + return length; + } else { + return 0; + } +} + +U_CAPI void U_EXPORT2 +udata_printError(const UDataSwapper *ds, + const char *fmt, + ...) { + va_list args; + + if(ds->printError!=nullptr) { + va_start(args, fmt); + ds->printError(ds->printErrorContext, fmt, args); + va_end(args); + } +} + +/* swap a data header ------------------------------------------------------- */ + +U_CAPI int32_t U_EXPORT2 +udata_swapDataHeader(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const DataHeader *pHeader; + uint16_t headerSize, infoSize; + + /* argument checking */ + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<-1 || (length>0 && outData==nullptr)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* check minimum length and magic bytes */ + pHeader=(const DataHeader *)inData; + if( (length>=0 && length<(int32_t)sizeof(DataHeader)) || + pHeader->dataHeader.magic1!=0xda || + pHeader->dataHeader.magic2!=0x27 || + pHeader->info.sizeofUChar!=2 + ) { + udata_printError(ds, "udata_swapDataHeader(): initial bytes do not look like ICU data\n"); + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + + headerSize=ds->readUInt16(pHeader->dataHeader.headerSize); + infoSize=ds->readUInt16(pHeader->info.size); + + if( headerSizedataHeader)+infoSize) || + (length>=0 && length0) { + DataHeader *outHeader; + const char *s; + int32_t maxLength; + + /* Most of the fields are just bytes and need no swapping. */ + if(inData!=outData) { + uprv_memcpy(outData, inData, headerSize); + } + outHeader=(DataHeader *)outData; + + outHeader->info.isBigEndian = ds->outIsBigEndian; + outHeader->info.charsetFamily = ds->outCharset; + + /* swap headerSize */ + ds->swapArray16(ds, &pHeader->dataHeader.headerSize, 2, &outHeader->dataHeader.headerSize, pErrorCode); + + /* swap UDataInfo size and reservedWord */ + ds->swapArray16(ds, &pHeader->info.size, 4, &outHeader->info.size, pErrorCode); + + /* swap copyright statement after the UDataInfo */ + infoSize+=sizeof(pHeader->dataHeader); + s=(const char *)inData+infoSize; + maxLength=headerSize-infoSize; + /* get the length of the string */ + for(length=0; lengthswapInvChars(ds, s, length, (char *)outData+infoSize, pErrorCode); + } + + return headerSize; +} + +/* API functions ------------------------------------------------------------ */ + +U_CAPI UDataSwapper * U_EXPORT2 +udata_openSwapper(UBool inIsBigEndian, uint8_t inCharset, + UBool outIsBigEndian, uint8_t outCharset, + UErrorCode *pErrorCode) { + UDataSwapper *swapper; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return nullptr; + } + if(inCharset>U_EBCDIC_FAMILY || outCharset>U_EBCDIC_FAMILY) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + + /* allocate the swapper */ + swapper=(UDataSwapper *)uprv_malloc(sizeof(UDataSwapper)); + if(swapper==nullptr) { + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memset(swapper, 0, sizeof(UDataSwapper)); + + /* set values and functions pointers according to in/out parameters */ + swapper->inIsBigEndian=inIsBigEndian; + swapper->inCharset=inCharset; + swapper->outIsBigEndian=outIsBigEndian; + swapper->outCharset=outCharset; + + swapper->readUInt16= inIsBigEndian==U_IS_BIG_ENDIAN ? uprv_readDirectUInt16 : uprv_readSwapUInt16; + swapper->readUInt32= inIsBigEndian==U_IS_BIG_ENDIAN ? uprv_readDirectUInt32 : uprv_readSwapUInt32; + + swapper->writeUInt16= outIsBigEndian==U_IS_BIG_ENDIAN ? uprv_writeDirectUInt16 : uprv_writeSwapUInt16; + swapper->writeUInt32= outIsBigEndian==U_IS_BIG_ENDIAN ? uprv_writeDirectUInt32 : uprv_writeSwapUInt32; + + swapper->compareInvChars= outCharset==U_ASCII_FAMILY ? uprv_compareInvAscii : uprv_compareInvEbcdic; + + if(inIsBigEndian==outIsBigEndian) { + swapper->swapArray16=uprv_copyArray16; + swapper->swapArray32=uprv_copyArray32; + swapper->swapArray64=uprv_copyArray64; + } else { + swapper->swapArray16=uprv_swapArray16; + swapper->swapArray32=uprv_swapArray32; + swapper->swapArray64=uprv_swapArray64; + } + + if(inCharset==U_ASCII_FAMILY) { + swapper->swapInvChars= outCharset==U_ASCII_FAMILY ? uprv_copyAscii : uprv_ebcdicFromAscii; + } else /* U_EBCDIC_FAMILY */ { + swapper->swapInvChars= outCharset==U_EBCDIC_FAMILY ? uprv_copyEbcdic : uprv_asciiFromEbcdic; + } + + return swapper; +} + +U_CAPI UDataSwapper * U_EXPORT2 +udata_openSwapperForInputData(const void *data, int32_t length, + UBool outIsBigEndian, uint8_t outCharset, + UErrorCode *pErrorCode) { + const DataHeader *pHeader; + uint16_t headerSize, infoSize; + UBool inIsBigEndian; + int8_t inCharset; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return nullptr; + } + if( data==nullptr || + (length>=0 && length<(int32_t)sizeof(DataHeader)) || + outCharset>U_EBCDIC_FAMILY + ) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + + pHeader=(const DataHeader *)data; + if( (length>=0 && length<(int32_t)sizeof(DataHeader)) || + pHeader->dataHeader.magic1!=0xda || + pHeader->dataHeader.magic2!=0x27 || + pHeader->info.sizeofUChar!=2 + ) { + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + + inIsBigEndian=(UBool)pHeader->info.isBigEndian; + inCharset=pHeader->info.charsetFamily; + + if(inIsBigEndian==U_IS_BIG_ENDIAN) { + headerSize=pHeader->dataHeader.headerSize; + infoSize=pHeader->info.size; + } else { + headerSize=uprv_readSwapUInt16(pHeader->dataHeader.headerSize); + infoSize=uprv_readSwapUInt16(pHeader->info.size); + } + + if( headerSizedataHeader)+infoSize) || + (length>=0 && lengthEBCDIC). +*/ + +#ifndef __UDATASWP_H__ +#define __UDATASWP_H__ + +#include +#include "unicode/utypes.h" + +/* forward declaration */ + +U_CDECL_BEGIN + +struct UDataSwapper; +typedef struct UDataSwapper UDataSwapper; + +/** + * Function type for data transformation. + * Transforms data, or just returns the length of the data if + * the input length is -1. + * Swap functions assume that their data pointers are aligned properly. + * + * Quick implementation outline: + * (best to copy and adapt and existing swapper implementation) + * check that the data looks like the expected format + * if(length<0) { + * preflight: + * never dereference outData + * read inData and determine the data size + * assume that inData is long enough for this + * } else { + * outData can be NULL if length==0 + * inData==outData (in-place swapping) possible but not required! + * verify that length>=(actual size) + * if there is a chance that not every byte up to size is reached + * due to padding etc.: + * if(inData!=outData) { + * memcpy(outData, inData, actual size); + * } + * swap contents + * } + * return actual size + * + * Further implementation notes: + * - read integers from inData before swapping them + * because in-place swapping can make them unreadable + * - compareInvChars compares a local Unicode string with already-swapped + * output charset strings + * + * @param ds Pointer to UDataSwapper containing global data about the + * transformation and function pointers for handling primitive + * types. + * @param inData Pointer to the input data to be transformed or examined. + * @param length Length of the data, counting bytes. May be -1 for preflighting. + * If length>=0, then transform the data. + * If length==-1, then only determine the length of the data. + * The length cannot be determined from the data itself for all + * types of data (e.g., not for simple arrays of integers). + * @param outData Pointer to the output data buffer. + * If length>=0 (transformation), then the output buffer must + * have a capacity of at least length. + * If length==-1, then outData will not be used and can be NULL. + * @param pErrorCode ICU UErrorCode parameter, must not be NULL and must + * fulfill U_SUCCESS on input. + * @return The actual length of the data. + * + * @see UDataSwapper + * @internal ICU 2.8 + */ +typedef int32_t U_CALLCONV +UDataSwapFn(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +/** + * Convert one uint16_t from input to platform endianness. + * @internal ICU 2.8 + */ +typedef uint16_t U_CALLCONV +UDataReadUInt16(uint16_t x); + +/** + * Convert one uint32_t from input to platform endianness. + * @internal ICU 2.8 + */ +typedef uint32_t U_CALLCONV +UDataReadUInt32(uint32_t x); + +/** + * Convert one uint16_t from platform to input endianness. + * @internal ICU 2.8 + */ +typedef void U_CALLCONV +UDataWriteUInt16(uint16_t *p, uint16_t x); + +/** + * Convert one uint32_t from platform to input endianness. + * @internal ICU 2.8 + */ +typedef void U_CALLCONV +UDataWriteUInt32(uint32_t *p, uint32_t x); + +/** + * Compare invariant-character strings, one in the output data and the + * other one caller-provided in Unicode. + * An output data string is compared because strings are usually swapped + * before the rest of the data, to allow for sorting of string tables + * according to the output charset. + * You can use -1 for the length parameters of NUL-terminated strings as usual. + * Returns Unicode code point order for invariant characters. + * @internal ICU 2.8 + */ +typedef int32_t U_CALLCONV +UDataCompareInvChars(const UDataSwapper *ds, + const char *outString, int32_t outLength, + const UChar *localString, int32_t localLength); + +/** + * Function for message output when an error occurs during data swapping. + * A format string and variable number of arguments are passed + * like for vprintf(). + * + * @param context A function-specific context pointer. + * @param fmt The format string. + * @param args The arguments for format string inserts. + * + * @internal ICU 2.8 + */ +typedef void U_CALLCONV +UDataPrintError(void *context, const char *fmt, va_list args); + +struct UDataSwapper { + /** Input endianness. @internal ICU 2.8 */ + UBool inIsBigEndian; + /** Input charset family. @see U_CHARSET_FAMILY @internal ICU 2.8 */ + uint8_t inCharset; + /** Output endianness. @internal ICU 2.8 */ + UBool outIsBigEndian; + /** Output charset family. @see U_CHARSET_FAMILY @internal ICU 2.8 */ + uint8_t outCharset; + + /* basic functions for reading data values */ + + /** Convert one uint16_t from input to platform endianness. @internal ICU 2.8 */ + UDataReadUInt16 *readUInt16; + /** Convert one uint32_t from input to platform endianness. @internal ICU 2.8 */ + UDataReadUInt32 *readUInt32; + /** Compare an invariant-character output string with a local one. @internal ICU 2.8 */ + UDataCompareInvChars *compareInvChars; + + /* basic functions for writing data values */ + + /** Convert one uint16_t from platform to input endianness. @internal ICU 2.8 */ + UDataWriteUInt16 *writeUInt16; + /** Convert one uint32_t from platform to input endianness. @internal ICU 2.8 */ + UDataWriteUInt32 *writeUInt32; + + /* basic functions for data transformations */ + + /** Transform an array of 16-bit integers. @internal ICU 2.8 */ + UDataSwapFn *swapArray16; + /** Transform an array of 32-bit integers. @internal ICU 2.8 */ + UDataSwapFn *swapArray32; + /** Transform an array of 64-bit integers. @internal ICU 53 */ + UDataSwapFn *swapArray64; + /** Transform an invariant-character string. @internal ICU 2.8 */ + UDataSwapFn *swapInvChars; + + /** + * Function for message output when an error occurs during data swapping. + * Can be NULL. + * @internal ICU 2.8 + */ + UDataPrintError *printError; + /** Context pointer for printError. @internal ICU 2.8 */ + void *printErrorContext; +}; + +U_CDECL_END + +U_CAPI UDataSwapper * U_EXPORT2 +udata_openSwapper(UBool inIsBigEndian, uint8_t inCharset, + UBool outIsBigEndian, uint8_t outCharset, + UErrorCode *pErrorCode); + +/** + * Open a UDataSwapper for the given input data and the specified output + * characteristics. + * Values of -1 for any of the characteristics mean the local platform's + * characteristics. + * + * @see udata_swap + * @internal ICU 2.8 + */ +U_CAPI UDataSwapper * U_EXPORT2 +udata_openSwapperForInputData(const void *data, int32_t length, + UBool outIsBigEndian, uint8_t outCharset, + UErrorCode *pErrorCode); + +U_CAPI void U_EXPORT2 +udata_closeSwapper(UDataSwapper *ds); + +/** + * Read the beginning of an ICU data piece, recognize magic bytes, + * swap the structure. + * Set a U_UNSUPPORTED_ERROR if it does not look like an ICU data piece. + * + * @return The size of the data header, in bytes. + * + * @internal ICU 2.8 + */ +U_CAPI int32_t U_EXPORT2 +udata_swapDataHeader(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +/** + * Convert one int16_t from input to platform endianness. + * @internal ICU 2.8 + */ +U_CAPI int16_t U_EXPORT2 +udata_readInt16(const UDataSwapper *ds, int16_t x); + +/** + * Convert one int32_t from input to platform endianness. + * @internal ICU 2.8 + */ +U_CAPI int32_t U_EXPORT2 +udata_readInt32(const UDataSwapper *ds, int32_t x); + +/** + * Swap a block of invariant, NUL-terminated strings, but not padding + * bytes after the last string. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +udata_swapInvStringBlock(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +U_CAPI void U_EXPORT2 +udata_printError(const UDataSwapper *ds, + const char *fmt, + ...); + +/* internal exports from putil.c -------------------------------------------- */ + +/* declared here to keep them out of the public putil.h */ + +/** + * Swap invariant char * strings ASCII->EBCDIC. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +uprv_ebcdicFromAscii(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +/** + * Copy invariant ASCII char * strings and verify they are invariant. + * @internal + */ +U_CFUNC int32_t +uprv_copyAscii(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +/** + * Swap invariant char * strings EBCDIC->ASCII. + * @internal + */ +U_CFUNC int32_t +uprv_asciiFromEbcdic(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +/** + * Copy invariant EBCDIC char * strings and verify they are invariant. + * @internal + */ +U_CFUNC int32_t +uprv_copyEbcdic(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +/** + * Compare ASCII invariant char * with Unicode invariant UChar * + * @internal + */ +U_CFUNC int32_t +uprv_compareInvAscii(const UDataSwapper *ds, + const char *outString, int32_t outLength, + const UChar *localString, int32_t localLength); + +/** + * Compare EBCDIC invariant char * with Unicode invariant UChar * + * @internal + */ +U_CFUNC int32_t +uprv_compareInvEbcdic(const UDataSwapper *ds, + const char *outString, int32_t outLength, + const UChar *localString, int32_t localLength); + +/** + * \def uprv_compareInvWithUChar + * Compare an invariant-character strings with a UChar string + * @internal + */ +#if U_CHARSET_FAMILY==U_ASCII_FAMILY +# define uprv_compareInvWithUChar uprv_compareInvAscii +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY +# define uprv_compareInvWithUChar uprv_compareInvEbcdic +#else +# error Unknown charset family! +#endif + +// utrie_swap.cpp -----------------------------------------------------------*** + +/** + * Swaps a serialized UTrie. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +utrie_swap(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +/** + * Swaps a serialized UTrie2. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +utrie2_swap(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +/** + * Swaps a serialized UCPTrie. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +ucptrie_swap(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +/** + * Swaps a serialized UTrie, UTrie2, or UCPTrie. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +utrie_swapAnyVersion(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode); + +/* material... -------------------------------------------------------------- */ + +#if 0 + +/* udata.h */ + +/** + * Public API function in udata.c + * + * Same as udata_openChoice() but automatically swaps the data. + * isAcceptable, if not NULL, may accept data with endianness and charset family + * different from the current platform's properties. + * If the data is acceptable and the platform properties do not match, then + * the swap function is called to swap an allocated version of the data. + * Preflighting may or may not be performed depending on whether the size of + * the loaded data item is known. + * + * @param isAcceptable Same as for udata_openChoice(). May be NULL. + * + * @internal ICU 2.8 + */ +U_CAPI UDataMemory * U_EXPORT2 +udata_openSwap(const char *path, const char *type, const char *name, + UDataMemoryIsAcceptable *isAcceptable, void *isAcceptableContext, + UDataSwapFn *swap, + UDataPrintError *printError, void *printErrorContext, + UErrorCode *pErrorCode); + +#endif + +#endif diff --git a/intl/icu/source/common/uelement.h b/intl/icu/source/common/uelement.h new file mode 100644 index 0000000000..2c5a2043e1 --- /dev/null +++ b/intl/icu/source/common/uelement.h @@ -0,0 +1,94 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 1997-2011, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: uelement.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2011jul04 +* created by: Markus W. Scherer +* +* Common definitions for UHashTable and UVector. +* UHashTok moved here from uhash.h and renamed UElement. +* This allows users of UVector to avoid the confusing #include of uhash.h. +* uhash.h aliases UElement to UHashTok, +* so that we need not change all of its code and its users. +*/ + +#ifndef __UELEMENT_H__ +#define __UELEMENT_H__ + +#include "unicode/utypes.h" + +U_CDECL_BEGIN + +/** + * A UVector element, or a key or value within a UHashtable. + * It may be either a 32-bit integral value or an opaque void* pointer. + * The void* pointer may be smaller than 32 bits (e.g. 24 bits) + * or may be larger (e.g. 64 bits). + * + * Because a UElement is the size of a native pointer or a 32-bit + * integer, we pass it around by value. + */ +union UElement { + void* pointer; + int32_t integer; +}; +typedef union UElement UElement; + +/** + * An element-equality (boolean) comparison function. + * @param e1 An element (object or integer) + * @param e2 An element (object or integer) + * @return true if the two elements are equal. + */ +typedef UBool U_CALLCONV UElementsAreEqual(const UElement e1, const UElement e2); + +/** + * An element sorting (three-way) comparison function. + * @param e1 An element (object or integer) + * @param e2 An element (object or integer) + * @return 32-bit signed integer comparison result: + * ==0 if the two elements are equal, + * <0 if e1 is < e2, or + * >0 if e1 is > e2. + */ +typedef int32_t U_CALLCONV UElementComparator(UElement e1, UElement e2); + +/** + * An element assignment function. It may copy an integer, copy + * a pointer, or clone a pointer, as appropriate. + * @param dst The element to be assigned to + * @param src The element to assign from + */ +typedef void U_CALLCONV UElementAssigner(UElement *dst, UElement *src); + +U_CDECL_END + +/** + * Comparator function for UnicodeString* keys. Implements UElementsAreEqual. + * @param key1 The string for comparison + * @param key2 The string for comparison + * @return true if key1 and key2 are equal, return false otherwise. + */ +U_CAPI UBool U_EXPORT2 +uhash_compareUnicodeString(const UElement key1, const UElement key2); + +/** + * Comparator function for UnicodeString* keys (case insensitive). + * Make sure to use together with uhash_hashCaselessUnicodeString. + * Implements UElementsAreEqual. + * @param key1 The string for comparison + * @param key2 The string for comparison + * @return true if key1 and key2 are equal, return false otherwise. + */ +U_CAPI UBool U_EXPORT2 +uhash_compareCaselessUnicodeString(const UElement key1, const UElement key2); + +#endif /* __UELEMENT_H__ */ diff --git a/intl/icu/source/common/uenum.cpp b/intl/icu/source/common/uenum.cpp new file mode 100644 index 0000000000..7aab58c44c --- /dev/null +++ b/intl/icu/source/common/uenum.cpp @@ -0,0 +1,189 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2002-2012, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: uenum.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:2 +* +* created on: 2002jul08 +* created by: Vladimir Weinstein +*/ + +#include "unicode/putil.h" +#include "uenumimp.h" +#include "cmemory.h" + +/* Layout of the baseContext buffer. */ +typedef struct { + int32_t len; /* number of bytes available starting at 'data' */ + char data; /* actual data starts here */ +} _UEnumBuffer; + +/* Extra bytes to allocate in the baseContext buffer. */ +static const int32_t PAD = 8; + +/* Return a pointer to the baseContext buffer, possibly allocating + or reallocating it if at least 'capacity' bytes are not available. */ +static void* _getBuffer(UEnumeration* en, int32_t capacity) { + + if (en->baseContext != nullptr) { + if (((_UEnumBuffer*) en->baseContext)->len < capacity) { + capacity += PAD; + en->baseContext = uprv_realloc(en->baseContext, + sizeof(int32_t) + capacity); + if (en->baseContext == nullptr) { + return nullptr; + } + ((_UEnumBuffer*) en->baseContext)->len = capacity; + } + } else { + capacity += PAD; + en->baseContext = uprv_malloc(sizeof(int32_t) + capacity); + if (en->baseContext == nullptr) { + return nullptr; + } + ((_UEnumBuffer*) en->baseContext)->len = capacity; + } + + return (void*) & ((_UEnumBuffer*) en->baseContext)->data; +} + +U_CAPI void U_EXPORT2 +uenum_close(UEnumeration* en) +{ + if (en) { + if (en->close != nullptr) { + if (en->baseContext) { + uprv_free(en->baseContext); + } + en->close(en); + } else { /* this seems dangerous, but we better kill the object */ + uprv_free(en); + } + } +} + +U_CAPI int32_t U_EXPORT2 +uenum_count(UEnumeration* en, UErrorCode* status) +{ + if (!en || U_FAILURE(*status)) { + return -1; + } + if (en->count != nullptr) { + return en->count(en, status); + } else { + *status = U_UNSUPPORTED_ERROR; + return -1; + } +} + +/* Don't call this directly. Only uenum_unext should be calling this. */ +U_CAPI const char16_t* U_EXPORT2 +uenum_unextDefault(UEnumeration* en, + int32_t* resultLength, + UErrorCode* status) +{ + char16_t *ustr = nullptr; + int32_t len = 0; + if (en->next != nullptr) { + const char *cstr = en->next(en, &len, status); + if (cstr != nullptr) { + ustr = (char16_t*) _getBuffer(en, (len+1) * sizeof(char16_t)); + if (ustr == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + } else { + u_charsToUChars(cstr, ustr, len+1); + } + } + } else { + *status = U_UNSUPPORTED_ERROR; + } + if (resultLength) { + *resultLength = len; + } + return ustr; +} + +/* Don't call this directly. Only uenum_next should be calling this. */ +U_CAPI const char* U_EXPORT2 +uenum_nextDefault(UEnumeration* en, + int32_t* resultLength, + UErrorCode* status) +{ + if (en->uNext != nullptr) { + char *tempCharVal; + const char16_t *tempUCharVal = en->uNext(en, resultLength, status); + if (tempUCharVal == nullptr) { + return nullptr; + } + tempCharVal = (char*) + _getBuffer(en, (*resultLength+1) * sizeof(char)); + if (!tempCharVal) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + u_UCharsToChars(tempUCharVal, tempCharVal, *resultLength + 1); + return tempCharVal; + } else { + *status = U_UNSUPPORTED_ERROR; + return nullptr; + } +} + +U_CAPI const char16_t* U_EXPORT2 +uenum_unext(UEnumeration* en, + int32_t* resultLength, + UErrorCode* status) +{ + if (!en || U_FAILURE(*status)) { + return nullptr; + } + if (en->uNext != nullptr) { + return en->uNext(en, resultLength, status); + } else { + *status = U_UNSUPPORTED_ERROR; + return nullptr; + } +} + +U_CAPI const char* U_EXPORT2 +uenum_next(UEnumeration* en, + int32_t* resultLength, + UErrorCode* status) +{ + if (!en || U_FAILURE(*status)) { + return nullptr; + } + if (en->next != nullptr) { + if (resultLength != nullptr) { + return en->next(en, resultLength, status); + } + else { + int32_t dummyLength=0; + return en->next(en, &dummyLength, status); + } + } else { + *status = U_UNSUPPORTED_ERROR; + return nullptr; + } +} + +U_CAPI void U_EXPORT2 +uenum_reset(UEnumeration* en, UErrorCode* status) +{ + if (!en || U_FAILURE(*status)) { + return; + } + if (en->reset != nullptr) { + en->reset(en, status); + } else { + *status = U_UNSUPPORTED_ERROR; + } +} diff --git a/intl/icu/source/common/uenumimp.h b/intl/icu/source/common/uenumimp.h new file mode 100644 index 0000000000..9c9df75ae0 --- /dev/null +++ b/intl/icu/source/common/uenumimp.h @@ -0,0 +1,155 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2002-2006, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: uenumimp.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:2 +* +* created on: 2002jul08 +* created by: Vladimir Weinstein +*/ + +#ifndef __UENUMIMP_H +#define __UENUMIMP_H + +#include "unicode/uenum.h" + +U_CDECL_BEGIN + +/** + * following are the type declarations for + * implementations of APIs. If any of these + * functions are NULL, U_UNSUPPORTED_ERROR + * is returned. If close is NULL, the enumeration + * object is going to be released. + * Initial error checking is done in the body + * of API function, so the implementations + * need not to check the initial error condition. + */ + +/** + * Function type declaration for uenum_close(). + * + * This function should cleanup the enumerator object + * + * @param en enumeration to be closed + */ +typedef void U_CALLCONV +UEnumClose(UEnumeration *en); + +/** + * Function type declaration for uenum_count(). + * + * This function should count the number of elements + * in this enumeration + * + * @param en enumeration to be counted + * @param status pointer to UErrorCode variable + * @return number of elements in enumeration + */ +typedef int32_t U_CALLCONV +UEnumCount(UEnumeration *en, UErrorCode *status); + +/** + * Function type declaration for uenum_unext(). + * + * This function returns the next element as a UChar *, + * or NULL after all elements haven been enumerated. + * + * @param en enumeration + * @param resultLength pointer to result length + * @param status pointer to UErrorCode variable + * @return next element as UChar *, + * or NULL after all elements haven been enumerated + */ +typedef const UChar* U_CALLCONV +UEnumUNext(UEnumeration* en, + int32_t* resultLength, + UErrorCode* status); + +/** + * Function type declaration for uenum_next(). + * + * This function returns the next element as a char *, + * or NULL after all elements haven been enumerated. + * + * @param en enumeration + * @param resultLength pointer to result length + * @param status pointer to UErrorCode variable + * @return next element as char *, + * or NULL after all elements haven been enumerated + */ +typedef const char* U_CALLCONV +UEnumNext(UEnumeration* en, + int32_t* resultLength, + UErrorCode* status); + +/** + * Function type declaration for uenum_reset(). + * + * This function should reset the enumeration + * object + * + * @param en enumeration + * @param status pointer to UErrorCode variable + */ +typedef void U_CALLCONV +UEnumReset(UEnumeration* en, + UErrorCode* status); + + +struct UEnumeration { + /* baseContext. For the base class only. Don't touch! */ + void *baseContext; + + /* context. Use it for what you need */ + void *context; + + /** + * these are functions that will + * be used for APIs + */ + /* called from uenum_close */ + UEnumClose *close; + /* called from uenum_count */ + UEnumCount *count; + /* called from uenum_unext */ + UEnumUNext *uNext; + /* called from uenum_next */ + UEnumNext *next; + /* called from uenum_reset */ + UEnumReset *reset; +}; + +U_CDECL_END + +/* This is the default implementation for uenum_unext(). + * It automatically converts the char * string to UChar *. + * Don't call this directly. This is called internally by uenum_unext + * when a UEnumeration is defined with 'uNext' pointing to this + * function. + */ +U_CAPI const UChar* U_EXPORT2 +uenum_unextDefault(UEnumeration* en, + int32_t* resultLength, + UErrorCode* status); + +/* This is the default implementation for uenum_next(). + * It automatically converts the UChar * string to char *. + * Don't call this directly. This is called internally by uenum_next + * when a UEnumeration is defined with 'next' pointing to this + * function. + */ +U_CAPI const char* U_EXPORT2 +uenum_nextDefault(UEnumeration* en, + int32_t* resultLength, + UErrorCode* status); + +#endif diff --git a/intl/icu/source/common/uhash.cpp b/intl/icu/source/common/uhash.cpp new file mode 100644 index 0000000000..4d92dfa93f --- /dev/null +++ b/intl/icu/source/common/uhash.cpp @@ -0,0 +1,1066 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 1997-2016, International Business Machines +* Corporation and others. All Rights Reserved. +****************************************************************************** +* Date Name Description +* 03/22/00 aliu Adapted from original C++ ICU Hashtable. +* 07/06/01 aliu Modified to support int32_t keys on +* platforms with sizeof(void*) < 32. +****************************************************************************** +*/ + +#include "uhash.h" +#include "unicode/ustring.h" +#include "cstring.h" +#include "cmemory.h" +#include "uassert.h" +#include "ustr_imp.h" + +/* This hashtable is implemented as a double hash. All elements are + * stored in a single array with no secondary storage for collision + * resolution (no linked list, etc.). When there is a hash collision + * (when two unequal keys have the same hashcode) we resolve this by + * using a secondary hash. The secondary hash is an increment + * computed as a hash function (a different one) of the primary + * hashcode. This increment is added to the initial hash value to + * obtain further slots assigned to the same hash code. For this to + * work, the length of the array and the increment must be relatively + * prime. The easiest way to achieve this is to have the length of + * the array be prime, and the increment be any value from + * 1..length-1. + * + * Hashcodes are 32-bit integers. We make sure all hashcodes are + * non-negative by masking off the top bit. This has two effects: (1) + * modulo arithmetic is simplified. If we allowed negative hashcodes, + * then when we computed hashcode % length, we could get a negative + * result, which we would then have to adjust back into range. It's + * simpler to just make hashcodes non-negative. (2) It makes it easy + * to check for empty vs. occupied slots in the table. We just mark + * empty or deleted slots with a negative hashcode. + * + * The central function is _uhash_find(). This function looks for a + * slot matching the given key and hashcode. If one is found, it + * returns a pointer to that slot. If the table is full, and no match + * is found, it returns nullptr -- in theory. This would make the code + * more complicated, since all callers of _uhash_find() would then + * have to check for a nullptr result. To keep this from happening, we + * don't allow the table to fill. When there is only one + * empty/deleted slot left, uhash_put() will refuse to increase the + * count, and fail. This simplifies the code. In practice, one will + * seldom encounter this using default UHashtables. However, if a + * hashtable is set to a U_FIXED resize policy, or if memory is + * exhausted, then the table may fill. + * + * High and low water ratios control rehashing. They establish levels + * of fullness (from 0 to 1) outside of which the data array is + * reallocated and repopulated. Setting the low water ratio to zero + * means the table will never shrink. Setting the high water ratio to + * one means the table will never grow. The ratios should be + * coordinated with the ratio between successive elements of the + * PRIMES table, so that when the primeIndex is incremented or + * decremented during rehashing, it brings the ratio of count / length + * back into the desired range (between low and high water ratios). + */ + +/******************************************************************** + * PRIVATE Constants, Macros + ********************************************************************/ + +/* This is a list of non-consecutive primes chosen such that + * PRIMES[i+1] ~ 2*PRIMES[i]. (Currently, the ratio ranges from 1.81 + * to 2.18; the inverse ratio ranges from 0.459 to 0.552.) If this + * ratio is changed, the low and high water ratios should also be + * adjusted to suit. + * + * These prime numbers were also chosen so that they are the largest + * prime number while being less than a power of two. + */ +static const int32_t PRIMES[] = { + 7, 13, 31, 61, 127, 251, 509, 1021, 2039, 4093, 8191, 16381, 32749, + 65521, 131071, 262139, 524287, 1048573, 2097143, 4194301, 8388593, + 16777213, 33554393, 67108859, 134217689, 268435399, 536870909, + 1073741789, 2147483647 /*, 4294967291 */ +}; + +#define PRIMES_LENGTH UPRV_LENGTHOF(PRIMES) +#define DEFAULT_PRIME_INDEX 4 + +/* These ratios are tuned to the PRIMES array such that a resize + * places the table back into the zone of non-resizing. That is, + * after a call to _uhash_rehash(), a subsequent call to + * _uhash_rehash() should do nothing (should not churn). This is only + * a potential problem with U_GROW_AND_SHRINK. + */ +static const float RESIZE_POLICY_RATIO_TABLE[6] = { + /* low, high water ratio */ + 0.0F, 0.5F, /* U_GROW: Grow on demand, do not shrink */ + 0.1F, 0.5F, /* U_GROW_AND_SHRINK: Grow and shrink on demand */ + 0.0F, 1.0F /* U_FIXED: Never change size */ +}; + +/* + Invariants for hashcode values: + + * DELETED < 0 + * EMPTY < 0 + * Real hashes >= 0 + + Hashcodes may not start out this way, but internally they are + adjusted so that they are always positive. We assume 32-bit + hashcodes; adjust these constants for other hashcode sizes. +*/ +#define HASH_DELETED ((int32_t) 0x80000000) +#define HASH_EMPTY ((int32_t) HASH_DELETED + 1) + +#define IS_EMPTY_OR_DELETED(x) ((x) < 0) + +/* This macro expects a UHashTok.pointer as its keypointer and + valuepointer parameters */ +#define HASH_DELETE_KEY_VALUE(hash, keypointer, valuepointer) UPRV_BLOCK_MACRO_BEGIN { \ + if (hash->keyDeleter != nullptr && keypointer != nullptr) { \ + (*hash->keyDeleter)(keypointer); \ + } \ + if (hash->valueDeleter != nullptr && valuepointer != nullptr) { \ + (*hash->valueDeleter)(valuepointer); \ + } \ +} UPRV_BLOCK_MACRO_END + +/* + * Constants for hinting whether a key or value is an integer + * or a pointer. If a hint bit is zero, then the associated + * token is assumed to be an integer. + */ +#define HINT_BOTH_INTEGERS (0) +#define HINT_KEY_POINTER (1) +#define HINT_VALUE_POINTER (2) +#define HINT_ALLOW_ZERO (4) + +/******************************************************************** + * PRIVATE Implementation + ********************************************************************/ + +static UHashTok +_uhash_setElement(UHashtable *hash, UHashElement* e, + int32_t hashcode, + UHashTok key, UHashTok value, int8_t hint) { + + UHashTok oldValue = e->value; + if (hash->keyDeleter != nullptr && e->key.pointer != nullptr && + e->key.pointer != key.pointer) { /* Avoid double deletion */ + (*hash->keyDeleter)(e->key.pointer); + } + if (hash->valueDeleter != nullptr) { + if (oldValue.pointer != nullptr && + oldValue.pointer != value.pointer) { /* Avoid double deletion */ + (*hash->valueDeleter)(oldValue.pointer); + } + oldValue.pointer = nullptr; + } + /* Compilers should copy the UHashTok union correctly, but even if + * they do, memory heap tools (e.g. BoundsChecker) can get + * confused when a pointer is cloaked in a union and then copied. + * TO ALLEVIATE THIS, we use hints (based on what API the user is + * calling) to copy pointers when we know the user thinks + * something is a pointer. */ + if (hint & HINT_KEY_POINTER) { + e->key.pointer = key.pointer; + } else { + e->key = key; + } + if (hint & HINT_VALUE_POINTER) { + e->value.pointer = value.pointer; + } else { + e->value = value; + } + e->hashcode = hashcode; + return oldValue; +} + +/** + * Assumes that the given element is not empty or deleted. + */ +static UHashTok +_uhash_internalRemoveElement(UHashtable *hash, UHashElement* e) { + UHashTok empty; + U_ASSERT(!IS_EMPTY_OR_DELETED(e->hashcode)); + --hash->count; + empty.pointer = nullptr; empty.integer = 0; + return _uhash_setElement(hash, e, HASH_DELETED, empty, empty, 0); +} + +static void +_uhash_internalSetResizePolicy(UHashtable *hash, enum UHashResizePolicy policy) { + U_ASSERT(hash != nullptr); + U_ASSERT(((int32_t)policy) >= 0); + U_ASSERT(((int32_t)policy) < 3); + hash->lowWaterRatio = RESIZE_POLICY_RATIO_TABLE[policy * 2]; + hash->highWaterRatio = RESIZE_POLICY_RATIO_TABLE[policy * 2 + 1]; +} + +/** + * Allocate internal data array of a size determined by the given + * prime index. If the index is out of range it is pinned into range. + * If the allocation fails the status is set to + * U_MEMORY_ALLOCATION_ERROR and all array storage is freed. In + * either case the previous array pointer is overwritten. + * + * Caller must ensure primeIndex is in range 0..PRIME_LENGTH-1. + */ +static void +_uhash_allocate(UHashtable *hash, + int32_t primeIndex, + UErrorCode *status) { + + UHashElement *p, *limit; + UHashTok emptytok; + + if (U_FAILURE(*status)) return; + + U_ASSERT(primeIndex >= 0 && primeIndex < PRIMES_LENGTH); + + hash->primeIndex = static_cast(primeIndex); + hash->length = PRIMES[primeIndex]; + + p = hash->elements = (UHashElement*) + uprv_malloc(sizeof(UHashElement) * hash->length); + + if (hash->elements == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return; + } + + emptytok.pointer = nullptr; /* Only one of these two is needed */ + emptytok.integer = 0; /* but we don't know which one. */ + + limit = p + hash->length; + while (p < limit) { + p->key = emptytok; + p->value = emptytok; + p->hashcode = HASH_EMPTY; + ++p; + } + + hash->count = 0; + hash->lowWaterMark = (int32_t)(hash->length * hash->lowWaterRatio); + hash->highWaterMark = (int32_t)(hash->length * hash->highWaterRatio); +} + +static UHashtable* +_uhash_init(UHashtable *result, + UHashFunction *keyHash, + UKeyComparator *keyComp, + UValueComparator *valueComp, + int32_t primeIndex, + UErrorCode *status) +{ + if (U_FAILURE(*status)) return nullptr; + U_ASSERT(keyHash != nullptr); + U_ASSERT(keyComp != nullptr); + + result->keyHasher = keyHash; + result->keyComparator = keyComp; + result->valueComparator = valueComp; + result->keyDeleter = nullptr; + result->valueDeleter = nullptr; + result->allocated = false; + _uhash_internalSetResizePolicy(result, U_GROW); + + _uhash_allocate(result, primeIndex, status); + + if (U_FAILURE(*status)) { + return nullptr; + } + + return result; +} + +static UHashtable* +_uhash_create(UHashFunction *keyHash, + UKeyComparator *keyComp, + UValueComparator *valueComp, + int32_t primeIndex, + UErrorCode *status) { + UHashtable *result; + + if (U_FAILURE(*status)) return nullptr; + + result = (UHashtable*) uprv_malloc(sizeof(UHashtable)); + if (result == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + + _uhash_init(result, keyHash, keyComp, valueComp, primeIndex, status); + result->allocated = true; + + if (U_FAILURE(*status)) { + uprv_free(result); + return nullptr; + } + + return result; +} + +/** + * Look for a key in the table, or if no such key exists, the first + * empty slot matching the given hashcode. Keys are compared using + * the keyComparator function. + * + * First find the start position, which is the hashcode modulo + * the length. Test it to see if it is: + * + * a. identical: First check the hash values for a quick check, + * then compare keys for equality using keyComparator. + * b. deleted + * c. empty + * + * Stop if it is identical or empty, otherwise continue by adding a + * "jump" value (moduloing by the length again to keep it within + * range) and retesting. For efficiency, there need enough empty + * values so that the searches stop within a reasonable amount of time. + * This can be changed by changing the high/low water marks. + * + * In theory, this function can return nullptr, if it is full (no empty + * or deleted slots) and if no matching key is found. In practice, we + * prevent this elsewhere (in uhash_put) by making sure the last slot + * in the table is never filled. + * + * The size of the table should be prime for this algorithm to work; + * otherwise we are not guaranteed that the jump value (the secondary + * hash) is relatively prime to the table length. + */ +static UHashElement* +_uhash_find(const UHashtable *hash, UHashTok key, + int32_t hashcode) { + + int32_t firstDeleted = -1; /* assume invalid index */ + int32_t theIndex, startIndex; + int32_t jump = 0; /* lazy evaluate */ + int32_t tableHash; + UHashElement *elements = hash->elements; + + hashcode &= 0x7FFFFFFF; /* must be positive */ + startIndex = theIndex = (hashcode ^ 0x4000000) % hash->length; + + do { + tableHash = elements[theIndex].hashcode; + if (tableHash == hashcode) { /* quick check */ + if ((*hash->keyComparator)(key, elements[theIndex].key)) { + return &(elements[theIndex]); + } + } else if (!IS_EMPTY_OR_DELETED(tableHash)) { + /* We have hit a slot which contains a key-value pair, + * but for which the hash code does not match. Keep + * looking. + */ + } else if (tableHash == HASH_EMPTY) { /* empty, end o' the line */ + break; + } else if (firstDeleted < 0) { /* remember first deleted */ + firstDeleted = theIndex; + } + if (jump == 0) { /* lazy compute jump */ + /* The jump value must be relatively prime to the table + * length. As long as the length is prime, then any value + * 1..length-1 will be relatively prime to it. + */ + jump = (hashcode % (hash->length - 1)) + 1; + } + theIndex = (theIndex + jump) % hash->length; + } while (theIndex != startIndex); + + if (firstDeleted >= 0) { + theIndex = firstDeleted; /* reset if had deleted slot */ + } else if (tableHash != HASH_EMPTY) { + /* We get to this point if the hashtable is full (no empty or + * deleted slots), and we've failed to find a match. THIS + * WILL NEVER HAPPEN as long as uhash_put() makes sure that + * count is always < length. + */ + UPRV_UNREACHABLE_EXIT; + } + return &(elements[theIndex]); +} + +/** + * Attempt to grow or shrink the data arrays in order to make the + * count fit between the high and low water marks. hash_put() and + * hash_remove() call this method when the count exceeds the high or + * low water marks. This method may do nothing, if memory allocation + * fails, or if the count is already in range, or if the length is + * already at the low or high limit. In any case, upon return the + * arrays will be valid. + */ +static void +_uhash_rehash(UHashtable *hash, UErrorCode *status) { + + UHashElement *old = hash->elements; + int32_t oldLength = hash->length; + int32_t newPrimeIndex = hash->primeIndex; + int32_t i; + + if (hash->count > hash->highWaterMark) { + if (++newPrimeIndex >= PRIMES_LENGTH) { + return; + } + } else if (hash->count < hash->lowWaterMark) { + if (--newPrimeIndex < 0) { + return; + } + } else { + return; + } + + _uhash_allocate(hash, newPrimeIndex, status); + + if (U_FAILURE(*status)) { + hash->elements = old; + hash->length = oldLength; + return; + } + + for (i = oldLength - 1; i >= 0; --i) { + if (!IS_EMPTY_OR_DELETED(old[i].hashcode)) { + UHashElement *e = _uhash_find(hash, old[i].key, old[i].hashcode); + U_ASSERT(e != nullptr); + U_ASSERT(e->hashcode == HASH_EMPTY); + e->key = old[i].key; + e->value = old[i].value; + e->hashcode = old[i].hashcode; + ++hash->count; + } + } + + uprv_free(old); +} + +static UHashTok +_uhash_remove(UHashtable *hash, + UHashTok key) { + /* First find the position of the key in the table. If the object + * has not been removed already, remove it. If the user wanted + * keys deleted, then delete it also. We have to put a special + * hashcode in that position that means that something has been + * deleted, since when we do a find, we have to continue PAST any + * deleted values. + */ + UHashTok result; + UHashElement* e = _uhash_find(hash, key, hash->keyHasher(key)); + U_ASSERT(e != nullptr); + result.pointer = nullptr; + result.integer = 0; + if (!IS_EMPTY_OR_DELETED(e->hashcode)) { + result = _uhash_internalRemoveElement(hash, e); + if (hash->count < hash->lowWaterMark) { + UErrorCode status = U_ZERO_ERROR; + _uhash_rehash(hash, &status); + } + } + return result; +} + +static UHashTok +_uhash_put(UHashtable *hash, + UHashTok key, + UHashTok value, + int8_t hint, + UErrorCode *status) { + + /* Put finds the position in the table for the new value. If the + * key is already in the table, it is deleted, if there is a + * non-nullptr keyDeleter. Then the key, the hash and the value are + * all put at the position in their respective arrays. + */ + int32_t hashcode; + UHashElement* e; + UHashTok emptytok; + + if (U_FAILURE(*status)) { + goto err; + } + U_ASSERT(hash != nullptr); + if ((hint & HINT_VALUE_POINTER) ? + value.pointer == nullptr : + value.integer == 0 && (hint & HINT_ALLOW_ZERO) == 0) { + /* Disallow storage of nullptr values, since nullptr is returned by + * get() to indicate an absent key. Storing nullptr == removing. + */ + return _uhash_remove(hash, key); + } + if (hash->count > hash->highWaterMark) { + _uhash_rehash(hash, status); + if (U_FAILURE(*status)) { + goto err; + } + } + + hashcode = (*hash->keyHasher)(key); + e = _uhash_find(hash, key, hashcode); + U_ASSERT(e != nullptr); + + if (IS_EMPTY_OR_DELETED(e->hashcode)) { + /* Important: We must never actually fill the table up. If we + * do so, then _uhash_find() will return nullptr, and we'll have + * to check for nullptr after every call to _uhash_find(). To + * avoid this we make sure there is always at least one empty + * or deleted slot in the table. This only is a problem if we + * are out of memory and rehash isn't working. + */ + ++hash->count; + if (hash->count == hash->length) { + /* Don't allow count to reach length */ + --hash->count; + *status = U_MEMORY_ALLOCATION_ERROR; + goto err; + } + } + + /* We must in all cases handle storage properly. If there was an + * old key, then it must be deleted (if the deleter != nullptr). + * Make hashcodes stored in table positive. + */ + return _uhash_setElement(hash, e, hashcode & 0x7FFFFFFF, key, value, hint); + + err: + /* If the deleters are non-nullptr, this method adopts its key and/or + * value arguments, and we must be sure to delete the key and/or + * value in all cases, even upon failure. + */ + HASH_DELETE_KEY_VALUE(hash, key.pointer, value.pointer); + emptytok.pointer = nullptr; emptytok.integer = 0; + return emptytok; +} + + +/******************************************************************** + * PUBLIC API + ********************************************************************/ + +U_CAPI UHashtable* U_EXPORT2 +uhash_open(UHashFunction *keyHash, + UKeyComparator *keyComp, + UValueComparator *valueComp, + UErrorCode *status) { + + return _uhash_create(keyHash, keyComp, valueComp, DEFAULT_PRIME_INDEX, status); +} + +U_CAPI UHashtable* U_EXPORT2 +uhash_openSize(UHashFunction *keyHash, + UKeyComparator *keyComp, + UValueComparator *valueComp, + int32_t size, + UErrorCode *status) { + + /* Find the smallest index i for which PRIMES[i] >= size. */ + int32_t i = 0; + while (i<(PRIMES_LENGTH-1) && PRIMES[i]= size. + int32_t i = 0; + while (i<(PRIMES_LENGTH-1) && PRIMES[i]elements != nullptr) { + if (hash->keyDeleter != nullptr || hash->valueDeleter != nullptr) { + int32_t pos=UHASH_FIRST; + UHashElement *e; + while ((e = (UHashElement*) uhash_nextElement(hash, &pos)) != nullptr) { + HASH_DELETE_KEY_VALUE(hash, e->key.pointer, e->value.pointer); + } + } + uprv_free(hash->elements); + hash->elements = nullptr; + } + if (hash->allocated) { + uprv_free(hash); + } +} + +U_CAPI UHashFunction *U_EXPORT2 +uhash_setKeyHasher(UHashtable *hash, UHashFunction *fn) { + UHashFunction *result = hash->keyHasher; + hash->keyHasher = fn; + return result; +} + +U_CAPI UKeyComparator *U_EXPORT2 +uhash_setKeyComparator(UHashtable *hash, UKeyComparator *fn) { + UKeyComparator *result = hash->keyComparator; + hash->keyComparator = fn; + return result; +} +U_CAPI UValueComparator *U_EXPORT2 +uhash_setValueComparator(UHashtable *hash, UValueComparator *fn){ + UValueComparator *result = hash->valueComparator; + hash->valueComparator = fn; + return result; +} + +U_CAPI UObjectDeleter *U_EXPORT2 +uhash_setKeyDeleter(UHashtable *hash, UObjectDeleter *fn) { + UObjectDeleter *result = hash->keyDeleter; + hash->keyDeleter = fn; + return result; +} + +U_CAPI UObjectDeleter *U_EXPORT2 +uhash_setValueDeleter(UHashtable *hash, UObjectDeleter *fn) { + UObjectDeleter *result = hash->valueDeleter; + hash->valueDeleter = fn; + return result; +} + +U_CAPI void U_EXPORT2 +uhash_setResizePolicy(UHashtable *hash, enum UHashResizePolicy policy) { + UErrorCode status = U_ZERO_ERROR; + _uhash_internalSetResizePolicy(hash, policy); + hash->lowWaterMark = (int32_t)(hash->length * hash->lowWaterRatio); + hash->highWaterMark = (int32_t)(hash->length * hash->highWaterRatio); + _uhash_rehash(hash, &status); +} + +U_CAPI int32_t U_EXPORT2 +uhash_count(const UHashtable *hash) { + return hash->count; +} + +U_CAPI void* U_EXPORT2 +uhash_get(const UHashtable *hash, + const void* key) { + UHashTok keyholder; + keyholder.pointer = (void*) key; + return _uhash_find(hash, keyholder, hash->keyHasher(keyholder))->value.pointer; +} + +U_CAPI void* U_EXPORT2 +uhash_iget(const UHashtable *hash, + int32_t key) { + UHashTok keyholder; + keyholder.integer = key; + return _uhash_find(hash, keyholder, hash->keyHasher(keyholder))->value.pointer; +} + +U_CAPI int32_t U_EXPORT2 +uhash_geti(const UHashtable *hash, + const void* key) { + UHashTok keyholder; + keyholder.pointer = (void*) key; + return _uhash_find(hash, keyholder, hash->keyHasher(keyholder))->value.integer; +} + +U_CAPI int32_t U_EXPORT2 +uhash_igeti(const UHashtable *hash, + int32_t key) { + UHashTok keyholder; + keyholder.integer = key; + return _uhash_find(hash, keyholder, hash->keyHasher(keyholder))->value.integer; +} + +U_CAPI int32_t U_EXPORT2 +uhash_getiAndFound(const UHashtable *hash, + const void *key, + UBool *found) { + UHashTok keyholder; + keyholder.pointer = (void *)key; + const UHashElement *e = _uhash_find(hash, keyholder, hash->keyHasher(keyholder)); + *found = !IS_EMPTY_OR_DELETED(e->hashcode); + return e->value.integer; +} + +U_CAPI int32_t U_EXPORT2 +uhash_igetiAndFound(const UHashtable *hash, + int32_t key, + UBool *found) { + UHashTok keyholder; + keyholder.integer = key; + const UHashElement *e = _uhash_find(hash, keyholder, hash->keyHasher(keyholder)); + *found = !IS_EMPTY_OR_DELETED(e->hashcode); + return e->value.integer; +} + +U_CAPI void* U_EXPORT2 +uhash_put(UHashtable *hash, + void* key, + void* value, + UErrorCode *status) { + UHashTok keyholder, valueholder; + keyholder.pointer = key; + valueholder.pointer = value; + return _uhash_put(hash, keyholder, valueholder, + HINT_KEY_POINTER | HINT_VALUE_POINTER, + status).pointer; +} + +U_CAPI void* U_EXPORT2 +uhash_iput(UHashtable *hash, + int32_t key, + void* value, + UErrorCode *status) { + UHashTok keyholder, valueholder; + keyholder.integer = key; + valueholder.pointer = value; + return _uhash_put(hash, keyholder, valueholder, + HINT_VALUE_POINTER, + status).pointer; +} + +U_CAPI int32_t U_EXPORT2 +uhash_puti(UHashtable *hash, + void* key, + int32_t value, + UErrorCode *status) { + UHashTok keyholder, valueholder; + keyholder.pointer = key; + valueholder.integer = value; + return _uhash_put(hash, keyholder, valueholder, + HINT_KEY_POINTER, + status).integer; +} + + +U_CAPI int32_t U_EXPORT2 +uhash_iputi(UHashtable *hash, + int32_t key, + int32_t value, + UErrorCode *status) { + UHashTok keyholder, valueholder; + keyholder.integer = key; + valueholder.integer = value; + return _uhash_put(hash, keyholder, valueholder, + HINT_BOTH_INTEGERS, + status).integer; +} + +U_CAPI int32_t U_EXPORT2 +uhash_putiAllowZero(UHashtable *hash, + void *key, + int32_t value, + UErrorCode *status) { + UHashTok keyholder, valueholder; + keyholder.pointer = key; + valueholder.integer = value; + return _uhash_put(hash, keyholder, valueholder, + HINT_KEY_POINTER | HINT_ALLOW_ZERO, + status).integer; +} + + +U_CAPI int32_t U_EXPORT2 +uhash_iputiAllowZero(UHashtable *hash, + int32_t key, + int32_t value, + UErrorCode *status) { + UHashTok keyholder, valueholder; + keyholder.integer = key; + valueholder.integer = value; + return _uhash_put(hash, keyholder, valueholder, + HINT_BOTH_INTEGERS | HINT_ALLOW_ZERO, + status).integer; +} + +U_CAPI void* U_EXPORT2 +uhash_remove(UHashtable *hash, + const void* key) { + UHashTok keyholder; + keyholder.pointer = (void*) key; + return _uhash_remove(hash, keyholder).pointer; +} + +U_CAPI void* U_EXPORT2 +uhash_iremove(UHashtable *hash, + int32_t key) { + UHashTok keyholder; + keyholder.integer = key; + return _uhash_remove(hash, keyholder).pointer; +} + +U_CAPI int32_t U_EXPORT2 +uhash_removei(UHashtable *hash, + const void* key) { + UHashTok keyholder; + keyholder.pointer = (void*) key; + return _uhash_remove(hash, keyholder).integer; +} + +U_CAPI int32_t U_EXPORT2 +uhash_iremovei(UHashtable *hash, + int32_t key) { + UHashTok keyholder; + keyholder.integer = key; + return _uhash_remove(hash, keyholder).integer; +} + +U_CAPI void U_EXPORT2 +uhash_removeAll(UHashtable *hash) { + int32_t pos = UHASH_FIRST; + const UHashElement *e; + U_ASSERT(hash != nullptr); + if (hash->count != 0) { + while ((e = uhash_nextElement(hash, &pos)) != nullptr) { + uhash_removeElement(hash, e); + } + } + U_ASSERT(hash->count == 0); +} + +U_CAPI UBool U_EXPORT2 +uhash_containsKey(const UHashtable *hash, const void *key) { + UHashTok keyholder; + keyholder.pointer = (void *)key; + const UHashElement *e = _uhash_find(hash, keyholder, hash->keyHasher(keyholder)); + return !IS_EMPTY_OR_DELETED(e->hashcode); +} + +/** + * Returns true if the UHashtable contains an item with this integer key. + * + * @param hash The target UHashtable. + * @param key An integer key stored in a hashtable + * @return true if the key is found. + */ +U_CAPI UBool U_EXPORT2 +uhash_icontainsKey(const UHashtable *hash, int32_t key) { + UHashTok keyholder; + keyholder.integer = key; + const UHashElement *e = _uhash_find(hash, keyholder, hash->keyHasher(keyholder)); + return !IS_EMPTY_OR_DELETED(e->hashcode); +} + +U_CAPI const UHashElement* U_EXPORT2 +uhash_find(const UHashtable *hash, const void* key) { + UHashTok keyholder; + const UHashElement *e; + keyholder.pointer = (void*) key; + e = _uhash_find(hash, keyholder, hash->keyHasher(keyholder)); + return IS_EMPTY_OR_DELETED(e->hashcode) ? nullptr : e; +} + +U_CAPI const UHashElement* U_EXPORT2 +uhash_nextElement(const UHashtable *hash, int32_t *pos) { + /* Walk through the array until we find an element that is not + * EMPTY and not DELETED. + */ + int32_t i; + U_ASSERT(hash != nullptr); + for (i = *pos + 1; i < hash->length; ++i) { + if (!IS_EMPTY_OR_DELETED(hash->elements[i].hashcode)) { + *pos = i; + return &(hash->elements[i]); + } + } + + /* No more elements */ + return nullptr; +} + +U_CAPI void* U_EXPORT2 +uhash_removeElement(UHashtable *hash, const UHashElement* e) { + U_ASSERT(hash != nullptr); + U_ASSERT(e != nullptr); + if (!IS_EMPTY_OR_DELETED(e->hashcode)) { + UHashElement *nce = (UHashElement *)e; + return _uhash_internalRemoveElement(hash, nce).pointer; + } + return nullptr; +} + +/******************************************************************** + * UHashTok convenience + ********************************************************************/ + +/** + * Return a UHashTok for an integer. + */ +/*U_CAPI UHashTok U_EXPORT2 +uhash_toki(int32_t i) { + UHashTok tok; + tok.integer = i; + return tok; +}*/ + +/** + * Return a UHashTok for a pointer. + */ +/*U_CAPI UHashTok U_EXPORT2 +uhash_tokp(void* p) { + UHashTok tok; + tok.pointer = p; + return tok; +}*/ + +/******************************************************************** + * PUBLIC Key Hash Functions + ********************************************************************/ + +U_CAPI int32_t U_EXPORT2 +uhash_hashUChars(const UHashTok key) { + const char16_t *s = (const char16_t *)key.pointer; + return s == nullptr ? 0 : ustr_hashUCharsN(s, u_strlen(s)); +} + +U_CAPI int32_t U_EXPORT2 +uhash_hashChars(const UHashTok key) { + const char *s = (const char *)key.pointer; + return s == nullptr ? 0 : static_cast(ustr_hashCharsN(s, static_cast(uprv_strlen(s)))); +} + +U_CAPI int32_t U_EXPORT2 +uhash_hashIChars(const UHashTok key) { + const char *s = (const char *)key.pointer; + return s == nullptr ? 0 : ustr_hashICharsN(s, static_cast(uprv_strlen(s))); +} + +U_CAPI UBool U_EXPORT2 +uhash_equals(const UHashtable* hash1, const UHashtable* hash2){ + int32_t count1, count2, pos, i; + + if(hash1==hash2){ + return true; + } + + /* + * Make sure that we are comparing 2 valid hashes of the same type + * with valid comparison functions. + * Without valid comparison functions, a binary comparison + * of the hash values will yield random results on machines + * with 64-bit pointers and 32-bit integer hashes. + * A valueComparator is normally optional. + */ + if (hash1==nullptr || hash2==nullptr || + hash1->keyComparator != hash2->keyComparator || + hash1->valueComparator != hash2->valueComparator || + hash1->valueComparator == nullptr) + { + /* + Normally we would return an error here about incompatible hash tables, + but we return false instead. + */ + return false; + } + + count1 = uhash_count(hash1); + count2 = uhash_count(hash2); + if(count1!=count2){ + return false; + } + + pos=UHASH_FIRST; + for(i=0; ikey; + const UHashTok val1 = elem1->value; + /* here the keys are not compared, instead the key form hash1 is used to fetch + * value from hash2. If the hashes are equal then then both hashes should + * contain equal values for the same key! + */ + const UHashElement* elem2 = _uhash_find(hash2, key1, hash2->keyHasher(key1)); + const UHashTok val2 = elem2->value; + if(hash1->valueComparator(val1, val2)==false){ + return false; + } + } + return true; +} + +/******************************************************************** + * PUBLIC Comparator Functions + ********************************************************************/ + +U_CAPI UBool U_EXPORT2 +uhash_compareUChars(const UHashTok key1, const UHashTok key2) { + const char16_t *p1 = (const char16_t*) key1.pointer; + const char16_t *p2 = (const char16_t*) key2.pointer; + if (p1 == p2) { + return true; + } + if (p1 == nullptr || p2 == nullptr) { + return false; + } + while (*p1 != 0 && *p1 == *p2) { + ++p1; + ++p2; + } + return (UBool)(*p1 == *p2); +} + +U_CAPI UBool U_EXPORT2 +uhash_compareChars(const UHashTok key1, const UHashTok key2) { + const char *p1 = (const char*) key1.pointer; + const char *p2 = (const char*) key2.pointer; + if (p1 == p2) { + return true; + } + if (p1 == nullptr || p2 == nullptr) { + return false; + } + while (*p1 != 0 && *p1 == *p2) { + ++p1; + ++p2; + } + return (UBool)(*p1 == *p2); +} + +U_CAPI UBool U_EXPORT2 +uhash_compareIChars(const UHashTok key1, const UHashTok key2) { + const char *p1 = (const char*) key1.pointer; + const char *p2 = (const char*) key2.pointer; + if (p1 == p2) { + return true; + } + if (p1 == nullptr || p2 == nullptr) { + return false; + } + while (*p1 != 0 && uprv_tolower(*p1) == uprv_tolower(*p2)) { + ++p1; + ++p2; + } + return (UBool)(*p1 == *p2); +} + +/******************************************************************** + * PUBLIC int32_t Support Functions + ********************************************************************/ + +U_CAPI int32_t U_EXPORT2 +uhash_hashLong(const UHashTok key) { + return key.integer; +} + +U_CAPI UBool U_EXPORT2 +uhash_compareLong(const UHashTok key1, const UHashTok key2) { + return (UBool)(key1.integer == key2.integer); +} diff --git a/intl/icu/source/common/uhash.h b/intl/icu/source/common/uhash.h new file mode 100644 index 0000000000..2ce296f0ec --- /dev/null +++ b/intl/icu/source/common/uhash.h @@ -0,0 +1,811 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 1997-2015, International Business Machines +* Corporation and others. All Rights Reserved. +****************************************************************************** +* Date Name Description +* 03/22/00 aliu Adapted from original C++ ICU Hashtable. +* 07/06/01 aliu Modified to support int32_t keys on +* platforms with sizeof(void*) < 32. +****************************************************************************** +*/ + +#ifndef UHASH_H +#define UHASH_H + +#include "unicode/utypes.h" +#include "cmemory.h" +#include "uelement.h" +#include "unicode/localpointer.h" + +/** + * UHashtable stores key-value pairs and does moderately fast lookup + * based on keys. It provides a good tradeoff between access time and + * storage space. As elements are added to it, it grows to accommodate + * them. By default, the table never shrinks, even if all elements + * are removed from it. + * + * Keys and values are stored as void* pointers. These void* pointers + * may be actual pointers to strings, objects, or any other structure + * in memory, or they may simply be integral values cast to void*. + * UHashtable doesn't care and manipulates them via user-supplied + * functions. These functions hash keys, compare keys, delete keys, + * and delete values. Some function pointers are optional (may be + * NULL); others must be supplied. Several prebuilt functions exist + * to handle common key types. + * + * UHashtable ownership of keys and values is flexible, and controlled + * by whether or not the key deleter and value deleter functions are + * set. If a void* key is actually a pointer to a deletable object, + * then UHashtable can be made to delete that object by setting the + * key deleter function pointer to a non-NULL value. If this is done, + * then keys passed to uhash_put() are owned by the hashtable and will + * be deleted by it at some point, either as keys are replaced, or + * when uhash_close() is finally called. The same is true of values + * and the value deleter function pointer. Keys passed to methods + * other than uhash_put() are never owned by the hashtable. + * + * NULL values are not allowed. uhash_get() returns NULL to indicate + * a key that is not in the table, and having a NULL value in the + * table would generate an ambiguous result. If a key and a NULL + * value is passed to uhash_put(), this has the effect of doing a + * uhash_remove() on that key. This keeps uhash_get(), uhash_count(), + * and uhash_nextElement() consistent with one another. + * + * Keys and values can be integers. + * Functions that work with an integer key have an "i" prefix. + * Functions that work with an integer value have an "i" suffix. + * As with putting a NULL value pointer, putting a zero value integer removes the item. + * Except, there are pairs of functions that allow setting zero values + * and fetching (value, found) pairs. + * + * To see everything in a hashtable, use uhash_nextElement() to + * iterate through its contents. Each call to this function returns a + * UHashElement pointer. A hash element contains a key, value, and + * hashcode. During iteration an element may be deleted by calling + * uhash_removeElement(); iteration may safely continue thereafter. + * The uhash_remove() function may also be safely called in + * mid-iteration. If uhash_put() is called during iteration, + * the iteration is still guaranteed to terminate reasonably, but + * there is no guarantee that every element will be returned or that + * some won't be returned more than once. + * + * Under no circumstances should the UHashElement returned by + * uhash_nextElement be modified directly. + * + * By default, the hashtable grows when necessary, but never shrinks, + * even if all items are removed. For most applications this is + * optimal. However, in a highly dynamic usage where memory is at a + * premium, the table can be set to both grow and shrink by calling + * uhash_setResizePolicy() with the policy U_GROW_AND_SHRINK. In a + * situation where memory is critical and the client wants a table + * that does not grow at all, the constant U_FIXED can be used. + */ + +/******************************************************************** + * Data Structures + ********************************************************************/ + +U_CDECL_BEGIN + +/** + * A key or value within a UHashtable. + * The hashing and comparison functions take a pointer to a + * UHashTok, but the deleter receives the void* pointer within it. + */ +typedef UElement UHashTok; + +/** + * This is a single hash element. + */ +struct UHashElement { + /* Reorder these elements to pack nicely if necessary */ + int32_t hashcode; + UHashTok value; + UHashTok key; +}; +typedef struct UHashElement UHashElement; + +/** + * A hashing function. + * @param key A key stored in a hashtable + * @return A NON-NEGATIVE hash code for parm. + */ +typedef int32_t U_CALLCONV UHashFunction(const UHashTok key); + +/** + * A key equality (boolean) comparison function. + */ +typedef UElementsAreEqual UKeyComparator; + +/** + * A value equality (boolean) comparison function. + */ +typedef UElementsAreEqual UValueComparator; + +/* see cmemory.h for UObjectDeleter and uprv_deleteUObject() */ + +/** + * This specifies whether or not, and how, the hashtable resizes itself. + * See uhash_setResizePolicy(). + */ +enum UHashResizePolicy { + U_GROW, /* Grow on demand, do not shrink */ + U_GROW_AND_SHRINK, /* Grow and shrink on demand */ + U_FIXED /* Never change size */ +}; + +/** + * The UHashtable struct. Clients should treat this as an opaque data + * type and manipulate it only through the uhash_... API. + */ +struct UHashtable { + + /* Main key-value pair storage array */ + + UHashElement *elements; + + /* Function pointers */ + + UHashFunction *keyHasher; /* Computes hash from key. + * Never null. */ + UKeyComparator *keyComparator; /* Compares keys for equality. + * Never null. */ + UValueComparator *valueComparator; /* Compares the values for equality */ + + UObjectDeleter *keyDeleter; /* Deletes keys when required. + * If NULL won't do anything */ + UObjectDeleter *valueDeleter; /* Deletes values when required. + * If NULL won't do anything */ + + /* Size parameters */ + + int32_t count; /* The number of key-value pairs in this table. + * 0 <= count <= length. In practice we + * never let count == length (see code). */ + int32_t length; /* The physical size of the arrays hashes, keys + * and values. Must be prime. */ + + /* Rehashing thresholds */ + + int32_t highWaterMark; /* If count > highWaterMark, rehash */ + int32_t lowWaterMark; /* If count < lowWaterMark, rehash */ + float highWaterRatio; /* 0..1; high water as a fraction of length */ + float lowWaterRatio; /* 0..1; low water as a fraction of length */ + + int8_t primeIndex; /* Index into our prime table for length. + * length == PRIMES[primeIndex] */ + UBool allocated; /* Was this UHashtable allocated? */ +}; +typedef struct UHashtable UHashtable; + +U_CDECL_END + +/******************************************************************** + * API + ********************************************************************/ + +/** + * Initialize a new UHashtable. + * @param keyHash A pointer to the key hashing function. Must not be + * NULL. + * @param keyComp A pointer to the function that compares keys. Must + * not be NULL. + * @param status A pointer to an UErrorCode to receive any errors. + * @return A pointer to a UHashtable, or 0 if an error occurred. + * @see uhash_openSize + */ +U_CAPI UHashtable* U_EXPORT2 +uhash_open(UHashFunction *keyHash, + UKeyComparator *keyComp, + UValueComparator *valueComp, + UErrorCode *status); + +/** + * Initialize a new UHashtable with a given initial size. + * @param keyHash A pointer to the key hashing function. Must not be + * NULL. + * @param keyComp A pointer to the function that compares keys. Must + * not be NULL. + * @param size The initial capacity of this hashtable. + * @param status A pointer to an UErrorCode to receive any errors. + * @return A pointer to a UHashtable, or 0 if an error occurred. + * @see uhash_open + */ +U_CAPI UHashtable* U_EXPORT2 +uhash_openSize(UHashFunction *keyHash, + UKeyComparator *keyComp, + UValueComparator *valueComp, + int32_t size, + UErrorCode *status); + +/** + * Initialize an existing UHashtable. + * @param keyHash A pointer to the key hashing function. Must not be + * NULL. + * @param keyComp A pointer to the function that compares keys. Must + * not be NULL. + * @param status A pointer to an UErrorCode to receive any errors. + * @return A pointer to a UHashtable, or 0 if an error occurred. + * @see uhash_openSize + */ +U_CAPI UHashtable* U_EXPORT2 +uhash_init(UHashtable *hash, + UHashFunction *keyHash, + UKeyComparator *keyComp, + UValueComparator *valueComp, + UErrorCode *status); + +/** + * Initialize an existing UHashtable. + * @param keyHash A pointer to the key hashing function. Must not be + * NULL. + * @param keyComp A pointer to the function that compares keys. Must + * not be NULL. + * @param size The initial capacity of this hashtable. + * @param status A pointer to an UErrorCode to receive any errors. + * @return A pointer to a UHashtable, or 0 if an error occurred. + * @see uhash_openSize + */ +U_CAPI UHashtable* U_EXPORT2 +uhash_initSize(UHashtable *hash, + UHashFunction *keyHash, + UKeyComparator *keyComp, + UValueComparator *valueComp, + int32_t size, + UErrorCode *status); + +/** + * Close a UHashtable, releasing the memory used. + * @param hash The UHashtable to close. If hash is NULL no operation is performed. + */ +U_CAPI void U_EXPORT2 +uhash_close(UHashtable *hash); + + + +/** + * Set the function used to hash keys. + * @param hash The UHashtable to set + * @param fn the function to be used hash keys; must not be NULL + * @return the previous key hasher; non-NULL + */ +U_CAPI UHashFunction *U_EXPORT2 +uhash_setKeyHasher(UHashtable *hash, UHashFunction *fn); + +/** + * Set the function used to compare keys. The default comparison is a + * void* pointer comparison. + * @param hash The UHashtable to set + * @param fn the function to be used compare keys; must not be NULL + * @return the previous key comparator; non-NULL + */ +U_CAPI UKeyComparator *U_EXPORT2 +uhash_setKeyComparator(UHashtable *hash, UKeyComparator *fn); + +/** + * Set the function used to compare values. The default comparison is a + * void* pointer comparison. + * @param hash The UHashtable to set + * @param fn the function to be used compare keys; must not be NULL + * @return the previous key comparator; non-NULL + */ +U_CAPI UValueComparator *U_EXPORT2 +uhash_setValueComparator(UHashtable *hash, UValueComparator *fn); + +/** + * Set the function used to delete keys. If this function pointer is + * NULL, this hashtable does not delete keys. If it is non-NULL, this + * hashtable does delete keys. This function should be set once + * before any elements are added to the hashtable and should not be + * changed thereafter. + * @param hash The UHashtable to set + * @param fn the function to be used delete keys, or NULL + * @return the previous key deleter; may be NULL + */ +U_CAPI UObjectDeleter *U_EXPORT2 +uhash_setKeyDeleter(UHashtable *hash, UObjectDeleter *fn); + +/** + * Set the function used to delete values. If this function pointer + * is NULL, this hashtable does not delete values. If it is non-NULL, + * this hashtable does delete values. This function should be set + * once before any elements are added to the hashtable and should not + * be changed thereafter. + * @param hash The UHashtable to set + * @param fn the function to be used delete values, or NULL + * @return the previous value deleter; may be NULL + */ +U_CAPI UObjectDeleter *U_EXPORT2 +uhash_setValueDeleter(UHashtable *hash, UObjectDeleter *fn); + +/** + * Specify whether or not, and how, the hashtable resizes itself. + * By default, tables grow but do not shrink (policy U_GROW). + * See enum UHashResizePolicy. + * @param hash The UHashtable to set + * @param policy The way the hashtable resizes itself, {U_GROW, U_GROW_AND_SHRINK, U_FIXED} + */ +U_CAPI void U_EXPORT2 +uhash_setResizePolicy(UHashtable *hash, enum UHashResizePolicy policy); + +/** + * Get the number of key-value pairs stored in a UHashtable. + * @param hash The UHashtable to query. + * @return The number of key-value pairs stored in hash. + */ +U_CAPI int32_t U_EXPORT2 +uhash_count(const UHashtable *hash); + +/** + * Put a (key=pointer, value=pointer) item in a UHashtable. If the + * keyDeleter is non-NULL, then the hashtable owns 'key' after this + * call. If the valueDeleter is non-NULL, then the hashtable owns + * 'value' after this call. Storing a NULL value is the same as + * calling uhash_remove(). + * @param hash The target UHashtable. + * @param key The key to store. + * @param value The value to store, may be NULL (see above). + * @param status A pointer to an UErrorCode to receive any errors. + * @return The previous value, or NULL if none. + * @see uhash_get + */ +U_CAPI void* U_EXPORT2 +uhash_put(UHashtable *hash, + void *key, + void *value, + UErrorCode *status); + +/** + * Put a (key=integer, value=pointer) item in a UHashtable. + * keyDeleter must be NULL. If the valueDeleter is non-NULL, then the + * hashtable owns 'value' after this call. Storing a NULL value is + * the same as calling uhash_remove(). + * @param hash The target UHashtable. + * @param key The integer key to store. + * @param value The value to store, may be NULL (see above). + * @param status A pointer to an UErrorCode to receive any errors. + * @return The previous value, or NULL if none. + * @see uhash_get + */ +U_CAPI void* U_EXPORT2 +uhash_iput(UHashtable *hash, + int32_t key, + void* value, + UErrorCode *status); + +/** + * Put a (key=pointer, value=integer) item in a UHashtable. If the + * keyDeleter is non-NULL, then the hashtable owns 'key' after this + * call. valueDeleter must be NULL. Storing a 0 value is the same as + * calling uhash_remove(). + * @param hash The target UHashtable. + * @param key The key to store. + * @param value The integer value to store. + * @param status A pointer to an UErrorCode to receive any errors. + * @return The previous value, or 0 if none. + * @see uhash_get + */ +U_CAPI int32_t U_EXPORT2 +uhash_puti(UHashtable *hash, + void* key, + int32_t value, + UErrorCode *status); + +/** + * Put a (key=integer, value=integer) item in a UHashtable. If the + * keyDeleter is non-NULL, then the hashtable owns 'key' after this + * call. valueDeleter must be NULL. Storing a 0 value is the same as + * calling uhash_remove(). + * @param hash The target UHashtable. + * @param key The key to store. + * @param value The integer value to store. + * @param status A pointer to an UErrorCode to receive any errors. + * @return The previous value, or 0 if none. + * @see uhash_get + */ +U_CAPI int32_t U_EXPORT2 +uhash_iputi(UHashtable *hash, + int32_t key, + int32_t value, + UErrorCode *status); + +/** + * Put a (key=pointer, value=integer) item in a UHashtable. If the + * keyDeleter is non-NULL, then the hashtable owns 'key' after this + * call. valueDeleter must be NULL. + * Storing a 0 value is possible; call uhash_igetiAndFound() to retrieve values including zero. + * + * @param hash The target UHashtable. + * @param key The key to store. + * @param value The integer value to store. + * @param status A pointer to an UErrorCode to receive any errors. + * @return The previous value, or 0 if none. + * @see uhash_getiAndFound + */ +U_CAPI int32_t U_EXPORT2 +uhash_putiAllowZero(UHashtable *hash, + void *key, + int32_t value, + UErrorCode *status); + +/** + * Put a (key=integer, value=integer) item in a UHashtable. If the + * keyDeleter is non-NULL, then the hashtable owns 'key' after this + * call. valueDeleter must be NULL. + * Storing a 0 value is possible; call uhash_igetiAndFound() to retrieve values including zero. + * + * @param hash The target UHashtable. + * @param key The key to store. + * @param value The integer value to store. + * @param status A pointer to an UErrorCode to receive any errors. + * @return The previous value, or 0 if none. + * @see uhash_igetiAndFound + */ +U_CAPI int32_t U_EXPORT2 +uhash_iputiAllowZero(UHashtable *hash, + int32_t key, + int32_t value, + UErrorCode *status); + +/** + * Retrieve a pointer value from a UHashtable using a pointer key, + * as previously stored by uhash_put(). + * @param hash The target UHashtable. + * @param key A pointer key stored in a hashtable + * @return The requested item, or NULL if not found. + */ +U_CAPI void* U_EXPORT2 +uhash_get(const UHashtable *hash, + const void *key); + +/** + * Retrieve a pointer value from a UHashtable using a integer key, + * as previously stored by uhash_iput(). + * @param hash The target UHashtable. + * @param key An integer key stored in a hashtable + * @return The requested item, or NULL if not found. + */ +U_CAPI void* U_EXPORT2 +uhash_iget(const UHashtable *hash, + int32_t key); + +/** + * Retrieve an integer value from a UHashtable using a pointer key, + * as previously stored by uhash_puti(). + * @param hash The target UHashtable. + * @param key A pointer key stored in a hashtable + * @return The requested item, or 0 if not found. + */ +U_CAPI int32_t U_EXPORT2 +uhash_geti(const UHashtable *hash, + const void* key); +/** + * Retrieve an integer value from a UHashtable using an integer key, + * as previously stored by uhash_iputi(). + * @param hash The target UHashtable. + * @param key An integer key stored in a hashtable + * @return The requested item, or 0 if not found. + */ +U_CAPI int32_t U_EXPORT2 +uhash_igeti(const UHashtable *hash, + int32_t key); + +/** + * Retrieves an integer value from a UHashtable using a pointer key, + * as previously stored by uhash_putiAllowZero() or uhash_puti(). + * + * @param hash The target UHashtable. + * @param key A pointer key stored in a hashtable + * @param found A pointer to a boolean which will be set for whether the key was found. + * @return The requested item, or 0 if not found. + */ +U_CAPI int32_t U_EXPORT2 +uhash_getiAndFound(const UHashtable *hash, + const void *key, + UBool *found); + +/** + * Retrieves an integer value from a UHashtable using an integer key, + * as previously stored by uhash_iputiAllowZero() or uhash_iputi(). + * + * @param hash The target UHashtable. + * @param key An integer key stored in a hashtable + * @param found A pointer to a boolean which will be set for whether the key was found. + * @return The requested item, or 0 if not found. + */ +U_CAPI int32_t U_EXPORT2 +uhash_igetiAndFound(const UHashtable *hash, + int32_t key, + UBool *found); + +/** + * Remove an item from a UHashtable stored by uhash_put(). + * @param hash The target UHashtable. + * @param key A key stored in a hashtable + * @return The item removed, or NULL if not found. + */ +U_CAPI void* U_EXPORT2 +uhash_remove(UHashtable *hash, + const void *key); + +/** + * Remove an item from a UHashtable stored by uhash_iput(). + * @param hash The target UHashtable. + * @param key An integer key stored in a hashtable + * @return The item removed, or NULL if not found. + */ +U_CAPI void* U_EXPORT2 +uhash_iremove(UHashtable *hash, + int32_t key); + +/** + * Remove an item from a UHashtable stored by uhash_puti(). + * @param hash The target UHashtable. + * @param key An key stored in a hashtable + * @return The item removed, or 0 if not found. + */ +U_CAPI int32_t U_EXPORT2 +uhash_removei(UHashtable *hash, + const void* key); + +/** + * Remove an item from a UHashtable stored by uhash_iputi(). + * @param hash The target UHashtable. + * @param key An integer key stored in a hashtable + * @return The item removed, or 0 if not found. + */ +U_CAPI int32_t U_EXPORT2 +uhash_iremovei(UHashtable *hash, + int32_t key); + +/** + * Remove all items from a UHashtable. + * @param hash The target UHashtable. + */ +U_CAPI void U_EXPORT2 +uhash_removeAll(UHashtable *hash); + +/** + * Returns true if the UHashtable contains an item with this pointer key. + * + * @param hash The target UHashtable. + * @param key A pointer key stored in a hashtable + * @return true if the key is found. + */ +U_CAPI UBool U_EXPORT2 +uhash_containsKey(const UHashtable *hash, const void *key); + +/** + * Returns true if the UHashtable contains an item with this integer key. + * + * @param hash The target UHashtable. + * @param key An integer key stored in a hashtable + * @return true if the key is found. + */ +U_CAPI UBool U_EXPORT2 +uhash_icontainsKey(const UHashtable *hash, int32_t key); + +/** + * Locate an element of a UHashtable. The caller must not modify the + * returned object. The primary use of this function is to obtain the + * stored key when it may not be identical to the search key. For + * example, if the compare function is a case-insensitive string + * compare, then the hash key may be desired in order to obtain the + * canonical case corresponding to a search key. + * @param hash The target UHashtable. + * @param key A key stored in a hashtable + * @return a hash element, or NULL if the key is not found. + */ +U_CAPI const UHashElement* U_EXPORT2 +uhash_find(const UHashtable *hash, const void* key); + +/** + * \def UHASH_FIRST + * Constant for use with uhash_nextElement + * @see uhash_nextElement + */ +#define UHASH_FIRST (-1) + +/** + * Iterate through the elements of a UHashtable. The caller must not + * modify the returned object. However, uhash_removeElement() may be + * called during iteration to remove an element from the table. + * Iteration may safely be resumed afterwards. If uhash_put() is + * called during iteration the iteration will then be out of sync and + * should be restarted. + * @param hash The target UHashtable. + * @param pos This should be set to UHASH_FIRST initially, and left untouched + * thereafter. + * @return a hash element, or NULL if no further key-value pairs + * exist in the table. + */ +U_CAPI const UHashElement* U_EXPORT2 +uhash_nextElement(const UHashtable *hash, + int32_t *pos); + +/** + * Remove an element, returned by uhash_nextElement(), from the table. + * Iteration may be safely continued afterwards. + * @param hash The hashtable + * @param e The element, returned by uhash_nextElement(), to remove. + * Must not be NULL. Must not be an empty or deleted element (as long + * as this was returned by uhash_nextElement() it will not be empty or + * deleted). Note: Although this parameter is const, it will be + * modified. + * @return the value that was removed. + */ +U_CAPI void* U_EXPORT2 +uhash_removeElement(UHashtable *hash, const UHashElement* e); + +/******************************************************************** + * UHashTok convenience + ********************************************************************/ + +/** + * Return a UHashTok for an integer. + * @param i The given integer + * @return a UHashTok for an integer. + */ +/*U_CAPI UHashTok U_EXPORT2 +uhash_toki(int32_t i);*/ + +/** + * Return a UHashTok for a pointer. + * @param p The given pointer + * @return a UHashTok for a pointer. + */ +/*U_CAPI UHashTok U_EXPORT2 +uhash_tokp(void* p);*/ + +/******************************************************************** + * UChar* and char* Support Functions + ********************************************************************/ + +/** + * Generate a hash code for a null-terminated UChar* string. If the + * string is not null-terminated do not use this function. Use + * together with uhash_compareUChars. + * @param key The string (const UChar*) to hash. + * @return A hash code for the key. + */ +U_CAPI int32_t U_EXPORT2 +uhash_hashUChars(const UHashTok key); + +/** + * Generate a hash code for a null-terminated char* string. If the + * string is not null-terminated do not use this function. Use + * together with uhash_compareChars. + * @param key The string (const char*) to hash. + * @return A hash code for the key. + */ +U_CAPI int32_t U_EXPORT2 +uhash_hashChars(const UHashTok key); + +/** + * Generate a case-insensitive hash code for a null-terminated char* + * string. If the string is not null-terminated do not use this + * function. Use together with uhash_compareIChars. + * @param key The string (const char*) to hash. + * @return A hash code for the key. + */ +U_CAPI int32_t U_EXPORT2 +uhash_hashIChars(const UHashTok key); + +/** + * Comparator for null-terminated UChar* strings. Use together with + * uhash_hashUChars. + * @param key1 The string for comparison + * @param key2 The string for comparison + * @return true if key1 and key2 are equal, return false otherwise. + */ +U_CAPI UBool U_EXPORT2 +uhash_compareUChars(const UHashTok key1, const UHashTok key2); + +/** + * Comparator for null-terminated char* strings. Use together with + * uhash_hashChars. + * @param key1 The string for comparison + * @param key2 The string for comparison + * @return true if key1 and key2 are equal, return false otherwise. + */ +U_CAPI UBool U_EXPORT2 +uhash_compareChars(const UHashTok key1, const UHashTok key2); + +/** + * Case-insensitive comparator for null-terminated char* strings. Use + * together with uhash_hashIChars. + * @param key1 The string for comparison + * @param key2 The string for comparison + * @return true if key1 and key2 are equal, return false otherwise. + */ +U_CAPI UBool U_EXPORT2 +uhash_compareIChars(const UHashTok key1, const UHashTok key2); + +/******************************************************************** + * UnicodeString Support Functions + ********************************************************************/ + +/** + * Hash function for UnicodeString* keys. + * @param key The string (const char*) to hash. + * @return A hash code for the key. + */ +U_CAPI int32_t U_EXPORT2 +uhash_hashUnicodeString(const UElement key); + +/** + * Hash function for UnicodeString* keys (case insensitive). + * Make sure to use together with uhash_compareCaselessUnicodeString. + * @param key The string (const char*) to hash. + * @return A hash code for the key. + */ +U_CAPI int32_t U_EXPORT2 +uhash_hashCaselessUnicodeString(const UElement key); + +/******************************************************************** + * int32_t Support Functions + ********************************************************************/ + +/** + * Hash function for 32-bit integer keys. + * @param key The string (const char*) to hash. + * @return A hash code for the key. + */ +U_CAPI int32_t U_EXPORT2 +uhash_hashLong(const UHashTok key); + +/** + * Comparator function for 32-bit integer keys. + * @param key1 The integer for comparison + * @param Key2 The integer for comparison + * @return true if key1 and key2 are equal, return false otherwise + */ +U_CAPI UBool U_EXPORT2 +uhash_compareLong(const UHashTok key1, const UHashTok key2); + +/******************************************************************** + * Other Support Functions + ********************************************************************/ + +/** + * Deleter for Hashtable objects. + * @param obj The object to be deleted + */ +U_CAPI void U_EXPORT2 +uhash_deleteHashtable(void *obj); + +/* Use uprv_free() itself as a deleter for any key or value allocated using uprv_malloc. */ + +/** + * Checks if the given hashtables are equal or not. + * @param hash1 + * @param hash2 + * @return true if the hashtables are equal and false if not. + */ +U_CAPI UBool U_EXPORT2 +uhash_equals(const UHashtable* hash1, const UHashtable* hash2); + + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUHashtablePointer + * "Smart pointer" class, closes a UHashtable via uhash_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUHashtablePointer, UHashtable, uhash_close); + +U_NAMESPACE_END + +#endif + +#endif diff --git a/intl/icu/source/common/uhash_us.cpp b/intl/icu/source/common/uhash_us.cpp new file mode 100644 index 0000000000..ef482c2746 --- /dev/null +++ b/intl/icu/source/common/uhash_us.cpp @@ -0,0 +1,26 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 1997-2011, International Business Machines +* Corporation and others. All Rights Reserved. +****************************************************************************** +* Date Name Description +* 03/22/00 aliu Creation. +* 07/06/01 aliu Modified to support int32_t keys on +* platforms with sizeof(void*) < 32. +****************************************************************************** +*/ + +#include "hash.h" + +/** + * Deleter for Hashtable objects. + */ +U_CAPI void U_EXPORT2 +uhash_deleteHashtable(void *obj) { + U_NAMESPACE_USE + delete (Hashtable*) obj; +} + +//eof diff --git a/intl/icu/source/common/uidna.cpp b/intl/icu/source/common/uidna.cpp new file mode 100644 index 0000000000..949d128f93 --- /dev/null +++ b/intl/icu/source/common/uidna.cpp @@ -0,0 +1,921 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ******************************************************************************* + * + * Copyright (C) 2003-2014, International Business Machines + * Corporation and others. All Rights Reserved. + * + ******************************************************************************* + * file name: uidna.cpp + * encoding: UTF-8 + * tab size: 8 (not used) + * indentation:4 + * + * created on: 2003feb1 + * created by: Ram Viswanadha + */ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_IDNA + +#include "unicode/uidna.h" +#include "unicode/ustring.h" +#include "unicode/usprep.h" +#include "punycode.h" +#include "ustr_imp.h" +#include "cmemory.h" +#include "uassert.h" +#include "sprpimpl.h" + +/* it is official IDNA ACE Prefix is "xn--" */ +static const char16_t ACE_PREFIX[] ={ 0x0078,0x006E,0x002d,0x002d } ; +#define ACE_PREFIX_LENGTH 4 + +#define MAX_LABEL_LENGTH 63 +/* The Max length of the labels should not be more than MAX_LABEL_LENGTH */ +#define MAX_LABEL_BUFFER_SIZE 100 + +#define MAX_DOMAIN_NAME_LENGTH 255 +/* The Max length of the domain names should not be more than MAX_DOMAIN_NAME_LENGTH */ +#define MAX_IDN_BUFFER_SIZE MAX_DOMAIN_NAME_LENGTH+1 + +#define LOWER_CASE_DELTA 0x0020 +#define HYPHEN 0x002D +#define FULL_STOP 0x002E +#define CAPITAL_A 0x0041 +#define CAPITAL_Z 0x005A + +inline static char16_t +toASCIILower(char16_t ch){ + if(CAPITAL_A <= ch && ch <= CAPITAL_Z){ + return ch + LOWER_CASE_DELTA; + } + return ch; +} + +inline static UBool +startsWithPrefix(const char16_t* src , int32_t srcLength){ + if(srcLength < ACE_PREFIX_LENGTH){ + return false; + } + + for(int8_t i=0; i< ACE_PREFIX_LENGTH; i++){ + if(toASCIILower(src[i]) != ACE_PREFIX[i]){ + return false; + } + } + return true; +} + + +inline static int32_t +compareCaseInsensitiveASCII(const char16_t* s1, int32_t s1Len, + const char16_t* s2, int32_t s2Len){ + + int32_t minLength; + int32_t lengthResult; + + // are we comparing different lengths? + if(s1Len != s2Len) { + if(s1Len < s2Len) { + minLength = s1Len; + lengthResult = -1; + } else { + minLength = s2Len; + lengthResult = 1; + } + } else { + // ok the lengths are equal + minLength = s1Len; + lengthResult = 0; + } + + char16_t c1,c2; + int32_t rc; + + for(int32_t i =0;/* no condition */;i++) { + + /* If we reach the ends of both strings then they match */ + if(i == minLength) { + return lengthResult; + } + + c1 = s1[i]; + c2 = s2[i]; + + /* Case-insensitive comparison */ + if(c1!=c2) { + rc=(int32_t)toASCIILower(c1)-(int32_t)toASCIILower(c2); + if(rc!=0) { + lengthResult=rc; + break; + } + } + } + return lengthResult; +} + + +/** + * Ascertain if the given code point is a label separator as + * defined by the IDNA RFC + * + * @param ch The code point to be ascertained + * @return true if the char is a label separator + * @stable ICU 2.8 + */ +static inline UBool isLabelSeparator(char16_t ch){ + switch(ch){ + case 0x002e: + case 0x3002: + case 0xFF0E: + case 0xFF61: + return true; + default: + return false; + } +} + +// returns the length of the label excluding the separator +// if *limit == separator then the length returned does not include +// the separtor. +static inline int32_t +getNextSeparator(char16_t *src, int32_t srcLength, + char16_t **limit, UBool *done){ + if(srcLength == -1){ + int32_t i; + for(i=0 ; ;i++){ + if(src[i] == 0){ + *limit = src + i; // point to null + *done = true; + return i; + } + if(isLabelSeparator(src[i])){ + *limit = src + (i+1); // go past the delimiter + return i; + + } + } + }else{ + int32_t i; + for(i=0;i0x007A){ + return false; + } + //[\\u002D \\u0030-\\u0039 \\u0041-\\u005A \\u0061-\\u007A] + if( (ch==0x002D) || + (0x0030 <= ch && ch <= 0x0039) || + (0x0041 <= ch && ch <= 0x005A) || + (0x0061 <= ch && ch <= 0x007A) + ){ + return true; + } + return false; +} + +static int32_t +_internal_toASCII(const char16_t* src, int32_t srcLength, + char16_t* dest, int32_t destCapacity, + int32_t options, + UStringPrepProfile* nameprep, + UParseError* parseError, + UErrorCode* status) +{ + + // TODO Revisit buffer handling. The label should not be over 63 ASCII characters. ICU4J may need to be updated too. + char16_t b1Stack[MAX_LABEL_BUFFER_SIZE], b2Stack[MAX_LABEL_BUFFER_SIZE]; + //initialize pointers to stack buffers + char16_t *b1 = b1Stack, *b2 = b2Stack; + int32_t b1Len=0, b2Len, + b1Capacity = MAX_LABEL_BUFFER_SIZE, + b2Capacity = MAX_LABEL_BUFFER_SIZE , + reqLength=0; + + int32_t namePrepOptions = ((options & UIDNA_ALLOW_UNASSIGNED) != 0) ? USPREP_ALLOW_UNASSIGNED: 0; + UBool* caseFlags = nullptr; + + // the source contains all ascii codepoints + UBool srcIsASCII = true; + // assume the source contains all LDH codepoints + UBool srcIsLDH = true; + + int32_t j=0; + + //get the options + UBool useSTD3ASCIIRules = (UBool)((options & UIDNA_USE_STD3_RULES) != 0); + + int32_t failPos = -1; + + if(srcLength == -1){ + srcLength = u_strlen(src); + } + + if(srcLength > b1Capacity){ + b1 = (char16_t*) uprv_malloc(srcLength * U_SIZEOF_UCHAR); + if(b1==nullptr){ + *status = U_MEMORY_ALLOCATION_ERROR; + goto CLEANUP; + } + b1Capacity = srcLength; + } + + // step 1 + for( j=0;j 0x7F){ + srcIsASCII = false; + } + b1[b1Len++] = src[j]; + } + + // step 2 is performed only if the source contains non ASCII + if(srcIsASCII == false){ + + // step 2 + b1Len = usprep_prepare(nameprep, src, srcLength, b1, b1Capacity, namePrepOptions, parseError, status); + + if(*status == U_BUFFER_OVERFLOW_ERROR){ + // redo processing of string + // we do not have enough room so grow the buffer + if(b1 != b1Stack){ + uprv_free(b1); + } + b1 = (char16_t*) uprv_malloc(b1Len * U_SIZEOF_UCHAR); + if(b1==nullptr){ + *status = U_MEMORY_ALLOCATION_ERROR; + goto CLEANUP; + } + + *status = U_ZERO_ERROR; // reset error + + b1Len = usprep_prepare(nameprep, src, srcLength, b1, b1Len, namePrepOptions, parseError, status); + } + } + // error bail out + if(U_FAILURE(*status)){ + goto CLEANUP; + } + if(b1Len == 0){ + *status = U_IDNA_ZERO_LENGTH_LABEL_ERROR; + goto CLEANUP; + } + + // for step 3 & 4 + srcIsASCII = true; + for( j=0;j 0x7F){ + srcIsASCII = false; + }else if(isLDHChar(b1[j])==false){ // if the char is in ASCII range verify that it is an LDH character + srcIsLDH = false; + failPos = j; + } + } + if(useSTD3ASCIIRules){ + // verify 3a and 3b + // 3(a) Verify the absence of non-LDH ASCII code points; that is, the + // absence of 0..2C, 2E..2F, 3A..40, 5B..60, and 7B..7F. + // 3(b) Verify the absence of leading and trailing hyphen-minus; that + // is, the absence of U+002D at the beginning and end of the + // sequence. + if( srcIsLDH == false /* source at this point should not contain anyLDH characters */ + || b1[0] == HYPHEN || b1[b1Len-1] == HYPHEN){ + *status = U_IDNA_STD3_ASCII_RULES_ERROR; + + /* populate the parseError struct */ + if(srcIsLDH==false){ + // failPos is always set the index of failure + uprv_syntaxError(b1,failPos, b1Len,parseError); + }else if(b1[0] == HYPHEN){ + // fail position is 0 + uprv_syntaxError(b1,0,b1Len,parseError); + }else{ + // the last index in the source is always length-1 + uprv_syntaxError(b1, (b1Len>0) ? b1Len-1 : b1Len, b1Len,parseError); + } + + goto CLEANUP; + } + } + // Step 4: if the source is ASCII then proceed to step 8 + if(srcIsASCII){ + if(b1Len <= destCapacity){ + u_memmove(dest, b1, b1Len); + reqLength = b1Len; + }else{ + reqLength = b1Len; + goto CLEANUP; + } + }else{ + // step 5 : verify the sequence does not begin with ACE prefix + if(!startsWithPrefix(b1,b1Len)){ + + //step 6: encode the sequence with punycode + + // do not preserve the case flags for now! + // TODO: Preserve the case while implementing the RFE + // caseFlags = (UBool*) uprv_malloc(b1Len * sizeof(UBool)); + // uprv_memset(caseFlags,true,b1Len); + + b2Len = u_strToPunycode(b1,b1Len,b2,b2Capacity,caseFlags, status); + + if(*status == U_BUFFER_OVERFLOW_ERROR){ + // redo processing of string + /* we do not have enough room so grow the buffer*/ + b2 = (char16_t*) uprv_malloc(b2Len * U_SIZEOF_UCHAR); + if(b2 == nullptr){ + *status = U_MEMORY_ALLOCATION_ERROR; + goto CLEANUP; + } + + *status = U_ZERO_ERROR; // reset error + + b2Len = u_strToPunycode(b1,b1Len,b2,b2Len,caseFlags, status); + } + //error bail out + if(U_FAILURE(*status)){ + goto CLEANUP; + } + // TODO : Reconsider while implementing the case preserve RFE + // convert all codepoints to lower case ASCII + // toASCIILower(b2,b2Len); + reqLength = b2Len+ACE_PREFIX_LENGTH; + + if(reqLength > destCapacity){ + *status = U_BUFFER_OVERFLOW_ERROR; + goto CLEANUP; + } + //Step 7: prepend the ACE prefix + u_memcpy(dest, ACE_PREFIX, ACE_PREFIX_LENGTH); + //Step 6: copy the contents in b2 into dest + u_memcpy(dest+ACE_PREFIX_LENGTH, b2, b2Len); + + }else{ + *status = U_IDNA_ACE_PREFIX_ERROR; + //position of failure is 0 + uprv_syntaxError(b1,0,b1Len,parseError); + goto CLEANUP; + } + } + // step 8: verify the length of label + if(reqLength > MAX_LABEL_LENGTH){ + *status = U_IDNA_LABEL_TOO_LONG_ERROR; + } + +CLEANUP: + if(b1 != b1Stack){ + uprv_free(b1); + } + if(b2 != b2Stack){ + uprv_free(b2); + } + uprv_free(caseFlags); + + return u_terminateUChars(dest, destCapacity, reqLength, status); +} + +static int32_t +_internal_toUnicode(const char16_t* src, int32_t srcLength, + char16_t* dest, int32_t destCapacity, + int32_t options, + UStringPrepProfile* nameprep, + UParseError* parseError, + UErrorCode* status) +{ + + //get the options + //UBool useSTD3ASCIIRules = (UBool)((options & UIDNA_USE_STD3_RULES) != 0); + int32_t namePrepOptions = ((options & UIDNA_ALLOW_UNASSIGNED) != 0) ? USPREP_ALLOW_UNASSIGNED: 0; + + // TODO Revisit buffer handling. The label should not be over 63 ASCII characters. ICU4J may need to be updated too. + char16_t b1Stack[MAX_LABEL_BUFFER_SIZE], b2Stack[MAX_LABEL_BUFFER_SIZE], b3Stack[MAX_LABEL_BUFFER_SIZE]; + + //initialize pointers to stack buffers + char16_t *b1 = b1Stack, *b2 = b2Stack, *b1Prime=nullptr, *b3=b3Stack; + int32_t b1Len = 0, b2Len, b1PrimeLen, b3Len, + b1Capacity = MAX_LABEL_BUFFER_SIZE, + b2Capacity = MAX_LABEL_BUFFER_SIZE, + b3Capacity = MAX_LABEL_BUFFER_SIZE, + reqLength=0; + + UBool* caseFlags = nullptr; + + UBool srcIsASCII = true; + /*UBool srcIsLDH = true; + int32_t failPos =0;*/ + + // step 1: find out if all the codepoints in src are ASCII + if(srcLength==-1){ + srcLength = 0; + for(;src[srcLength]!=0;){ + if(src[srcLength]> 0x7f){ + srcIsASCII = false; + }/*else if(isLDHChar(src[srcLength])==false){ + // here we do not assemble surrogates + // since we know that LDH code points + // are in the ASCII range only + srcIsLDH = false; + failPos = srcLength; + }*/ + srcLength++; + } + }else if(srcLength > 0){ + for(int32_t j=0; j 0x7f){ + srcIsASCII = false; + break; + }/*else if(isLDHChar(src[j])==false){ + // here we do not assemble surrogates + // since we know that LDH code points + // are in the ASCII range only + srcIsLDH = false; + failPos = j; + }*/ + } + }else{ + return 0; + } + + if(srcIsASCII == false){ + // step 2: process the string + b1Len = usprep_prepare(nameprep, src, srcLength, b1, b1Capacity, namePrepOptions, parseError, status); + if(*status == U_BUFFER_OVERFLOW_ERROR){ + // redo processing of string + /* we do not have enough room so grow the buffer*/ + b1 = (char16_t*) uprv_malloc(b1Len * U_SIZEOF_UCHAR); + if(b1==nullptr){ + *status = U_MEMORY_ALLOCATION_ERROR; + goto CLEANUP; + } + + *status = U_ZERO_ERROR; // reset error + + b1Len = usprep_prepare(nameprep, src, srcLength, b1, b1Len, namePrepOptions, parseError, status); + } + //bail out on error + if(U_FAILURE(*status)){ + goto CLEANUP; + } + }else{ + + //just point src to b1 + b1 = (char16_t*) src; + b1Len = srcLength; + } + + // The RFC states that + // + // ToUnicode never fails. If any step fails, then the original input + // is returned immediately in that step. + // + + //step 3: verify ACE Prefix + if(startsWithPrefix(b1,b1Len)){ + + //step 4: Remove the ACE Prefix + b1Prime = b1 + ACE_PREFIX_LENGTH; + b1PrimeLen = b1Len - ACE_PREFIX_LENGTH; + + //step 5: Decode using punycode + b2Len = u_strFromPunycode(b1Prime, b1PrimeLen, b2, b2Capacity, caseFlags,status); + + if(*status == U_BUFFER_OVERFLOW_ERROR){ + // redo processing of string + /* we do not have enough room so grow the buffer*/ + b2 = (char16_t*) uprv_malloc(b2Len * U_SIZEOF_UCHAR); + if(b2==nullptr){ + *status = U_MEMORY_ALLOCATION_ERROR; + goto CLEANUP; + } + + *status = U_ZERO_ERROR; // reset error + + b2Len = u_strFromPunycode(b1Prime, b1PrimeLen, b2, b2Len, caseFlags, status); + } + + + //step 6:Apply toASCII + b3Len = uidna_toASCII(b2, b2Len, b3, b3Capacity, options, parseError, status); + + if(*status == U_BUFFER_OVERFLOW_ERROR){ + // redo processing of string + /* we do not have enough room so grow the buffer*/ + b3 = (char16_t*) uprv_malloc(b3Len * U_SIZEOF_UCHAR); + if(b3==nullptr){ + *status = U_MEMORY_ALLOCATION_ERROR; + goto CLEANUP; + } + + *status = U_ZERO_ERROR; // reset error + + b3Len = uidna_toASCII(b2,b2Len,b3,b3Len,options,parseError, status); + + } + //bail out on error + if(U_FAILURE(*status)){ + goto CLEANUP; + } + + //step 7: verify + if(compareCaseInsensitiveASCII(b1, b1Len, b3, b3Len) !=0){ + // Cause the original to be returned. + *status = U_IDNA_VERIFICATION_ERROR; + goto CLEANUP; + } + + //step 8: return output of step 5 + reqLength = b2Len; + if(b2Len <= destCapacity) { + u_memmove(dest, b2, b2Len); + } + } + else{ + // See the start of this if statement for why this is commented out. + // verify that STD3 ASCII rules are satisfied + /*if(useSTD3ASCIIRules == true){ + if( srcIsLDH == false // source contains some non-LDH characters + || src[0] == HYPHEN || src[srcLength-1] == HYPHEN){ + *status = U_IDNA_STD3_ASCII_RULES_ERROR; + + // populate the parseError struct + if(srcIsLDH==false){ + // failPos is always set the index of failure + uprv_syntaxError(src,failPos, srcLength,parseError); + }else if(src[0] == HYPHEN){ + // fail position is 0 + uprv_syntaxError(src,0,srcLength,parseError); + }else{ + // the last index in the source is always length-1 + uprv_syntaxError(src, (srcLength>0) ? srcLength-1 : srcLength, srcLength,parseError); + } + + goto CLEANUP; + } + }*/ + // just return the source + //copy the source to destination + if(srcLength <= destCapacity){ + u_memmove(dest, src, srcLength); + } + reqLength = srcLength; + } + + +CLEANUP: + + if(b1 != b1Stack && b1!=src){ + uprv_free(b1); + } + if(b2 != b2Stack){ + uprv_free(b2); + } + uprv_free(caseFlags); + + // The RFC states that + // + // ToUnicode never fails. If any step fails, then the original input + // is returned immediately in that step. + // + // So if any step fails lets copy source to destination + if(U_FAILURE(*status)){ + //copy the source to destination + if(dest && srcLength <= destCapacity){ + // srcLength should have already been set earlier. + U_ASSERT(srcLength >= 0); + u_memmove(dest, src, srcLength); + } + reqLength = srcLength; + *status = U_ZERO_ERROR; + } + + return u_terminateUChars(dest, destCapacity, reqLength, status); +} + +U_CAPI int32_t U_EXPORT2 +uidna_toASCII(const char16_t* src, int32_t srcLength, + char16_t* dest, int32_t destCapacity, + int32_t options, + UParseError* parseError, + UErrorCode* status){ + + if(status == nullptr || U_FAILURE(*status)){ + return 0; + } + if((src==nullptr) || (srcLength < -1) || (destCapacity<0) || (!dest && destCapacity > 0)){ + *status = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + UStringPrepProfile* nameprep = usprep_openByType(USPREP_RFC3491_NAMEPREP, status); + + if(U_FAILURE(*status)){ + return -1; + } + + int32_t retLen = _internal_toASCII(src, srcLength, dest, destCapacity, options, nameprep, parseError, status); + + /* close the profile*/ + usprep_close(nameprep); + + return retLen; +} + +U_CAPI int32_t U_EXPORT2 +uidna_toUnicode(const char16_t* src, int32_t srcLength, + char16_t* dest, int32_t destCapacity, + int32_t options, + UParseError* parseError, + UErrorCode* status){ + + if(status == nullptr || U_FAILURE(*status)){ + return 0; + } + if( (src==nullptr) || (srcLength < -1) || (destCapacity<0) || (!dest && destCapacity > 0)){ + *status = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + UStringPrepProfile* nameprep = usprep_openByType(USPREP_RFC3491_NAMEPREP, status); + + if(U_FAILURE(*status)){ + return -1; + } + + int32_t retLen = _internal_toUnicode(src, srcLength, dest, destCapacity, options, nameprep, parseError, status); + + usprep_close(nameprep); + + return retLen; +} + + +U_CAPI int32_t U_EXPORT2 +uidna_IDNToASCII( const char16_t *src, int32_t srcLength, + char16_t* dest, int32_t destCapacity, + int32_t options, + UParseError *parseError, + UErrorCode *status){ + + if(status == nullptr || U_FAILURE(*status)){ + return 0; + } + if((src==nullptr) || (srcLength < -1) || (destCapacity<0) || (!dest && destCapacity > 0)){ + *status = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + int32_t reqLength = 0; + + UStringPrepProfile* nameprep = usprep_openByType(USPREP_RFC3491_NAMEPREP, status); + + if(U_FAILURE(*status)){ + return 0; + } + + //initialize pointers + char16_t *delimiter = (char16_t*)src; + char16_t *labelStart = (char16_t*)src; + char16_t *currentDest = (char16_t*) dest; + int32_t remainingLen = srcLength; + int32_t remainingDestCapacity = destCapacity; + int32_t labelLen = 0, labelReqLength = 0; + UBool done = false; + + + for(;;){ + + labelLen = getNextSeparator(labelStart,remainingLen, &delimiter,&done); + labelReqLength = 0; + if(!(labelLen==0 && done)){// make sure this is not a root label separator. + + labelReqLength = _internal_toASCII( labelStart, labelLen, + currentDest, remainingDestCapacity, + options, nameprep, + parseError, status); + + if(*status == U_BUFFER_OVERFLOW_ERROR){ + + *status = U_ZERO_ERROR; // reset error + remainingDestCapacity = 0; + } + } + + + if(U_FAILURE(*status)){ + break; + } + + reqLength +=labelReqLength; + // adjust the destination pointer + if(labelReqLength < remainingDestCapacity){ + currentDest = currentDest + labelReqLength; + remainingDestCapacity -= labelReqLength; + }else{ + // should never occur + remainingDestCapacity = 0; + } + + if(done){ + break; + } + + // add the label separator + if(remainingDestCapacity > 0){ + *currentDest++ = FULL_STOP; + remainingDestCapacity--; + } + reqLength++; + + labelStart = delimiter; + if(remainingLen >0 ){ + remainingLen = (int32_t)(srcLength - (delimiter - src)); + } + + } + + if(reqLength > MAX_DOMAIN_NAME_LENGTH){ + *status = U_IDNA_DOMAIN_NAME_TOO_LONG_ERROR; + } + + usprep_close(nameprep); + + return u_terminateUChars(dest, destCapacity, reqLength, status); +} + +U_CAPI int32_t U_EXPORT2 +uidna_IDNToUnicode( const char16_t* src, int32_t srcLength, + char16_t* dest, int32_t destCapacity, + int32_t options, + UParseError* parseError, + UErrorCode* status){ + + if(status == nullptr || U_FAILURE(*status)){ + return 0; + } + if((src==nullptr) || (srcLength < -1) || (destCapacity<0) || (!dest && destCapacity > 0)){ + *status = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + int32_t reqLength = 0; + + UStringPrepProfile* nameprep = usprep_openByType(USPREP_RFC3491_NAMEPREP, status); + + if(U_FAILURE(*status)){ + return 0; + } + + //initialize pointers + char16_t *delimiter = (char16_t*)src; + char16_t *labelStart = (char16_t*)src; + char16_t *currentDest = (char16_t*) dest; + int32_t remainingLen = srcLength; + int32_t remainingDestCapacity = destCapacity; + int32_t labelLen = 0, labelReqLength = 0; + UBool done = false; + + for(;;){ + + labelLen = getNextSeparator(labelStart,remainingLen, &delimiter,&done); + + // The RFC states that + // + // ToUnicode never fails. If any step fails, then the original input + // is returned immediately in that step. + // + // _internal_toUnicode will copy the label. + /*if(labelLen==0 && done==false){ + *status = U_IDNA_ZERO_LENGTH_LABEL_ERROR; + break; + }*/ + + labelReqLength = _internal_toUnicode(labelStart, labelLen, + currentDest, remainingDestCapacity, + options, nameprep, + parseError, status); + + if(*status == U_BUFFER_OVERFLOW_ERROR){ + *status = U_ZERO_ERROR; // reset error + remainingDestCapacity = 0; + } + + if(U_FAILURE(*status)){ + break; + } + + reqLength +=labelReqLength; + // adjust the destination pointer + if(labelReqLength < remainingDestCapacity){ + currentDest = currentDest + labelReqLength; + remainingDestCapacity -= labelReqLength; + }else{ + // should never occur + remainingDestCapacity = 0; + } + + if(done){ + break; + } + + // add the label separator + // Unlike the ToASCII operation we don't normalize the label separators + if(remainingDestCapacity > 0){ + *currentDest++ = *(labelStart + labelLen); + remainingDestCapacity--; + } + reqLength++; + + labelStart = delimiter; + if(remainingLen >0 ){ + remainingLen = (int32_t)(srcLength - (delimiter - src)); + } + + } + + if(reqLength > MAX_DOMAIN_NAME_LENGTH){ + *status = U_IDNA_DOMAIN_NAME_TOO_LONG_ERROR; + } + + usprep_close(nameprep); + + return u_terminateUChars(dest, destCapacity, reqLength, status); +} + +U_CAPI int32_t U_EXPORT2 +uidna_compare( const char16_t *s1, int32_t length1, + const char16_t *s2, int32_t length2, + int32_t options, + UErrorCode* status){ + + if(status == nullptr || U_FAILURE(*status)){ + return -1; + } + + char16_t b1Stack[MAX_IDN_BUFFER_SIZE], b2Stack[MAX_IDN_BUFFER_SIZE]; + char16_t *b1 = b1Stack, *b2 = b2Stack; + int32_t b1Len, b2Len, b1Capacity = MAX_IDN_BUFFER_SIZE, b2Capacity = MAX_IDN_BUFFER_SIZE; + int32_t result=-1; + + UParseError parseError; + + b1Len = uidna_IDNToASCII(s1, length1, b1, b1Capacity, options, &parseError, status); + if(*status == U_BUFFER_OVERFLOW_ERROR){ + // redo processing of string + b1 = (char16_t*) uprv_malloc(b1Len * U_SIZEOF_UCHAR); + if(b1==nullptr){ + *status = U_MEMORY_ALLOCATION_ERROR; + goto CLEANUP; + } + + *status = U_ZERO_ERROR; // reset error + + b1Len = uidna_IDNToASCII(s1,length1,b1,b1Len, options, &parseError, status); + + } + + b2Len = uidna_IDNToASCII(s2,length2, b2,b2Capacity, options, &parseError, status); + if(*status == U_BUFFER_OVERFLOW_ERROR){ + // redo processing of string + b2 = (char16_t*) uprv_malloc(b2Len * U_SIZEOF_UCHAR); + if(b2==nullptr){ + *status = U_MEMORY_ALLOCATION_ERROR; + goto CLEANUP; + } + + *status = U_ZERO_ERROR; // reset error + + b2Len = uidna_IDNToASCII(s2, length2, b2, b2Len, options, &parseError, status); + + } + // when toASCII is applied all label separators are replaced with FULL_STOP + result = compareCaseInsensitiveASCII(b1,b1Len,b2,b2Len); + +CLEANUP: + if(b1 != b1Stack){ + uprv_free(b1); + } + + if(b2 != b2Stack){ + uprv_free(b2); + } + + return result; +} + +#endif /* #if !UCONFIG_NO_IDNA */ diff --git a/intl/icu/source/common/uinit.cpp b/intl/icu/source/common/uinit.cpp new file mode 100644 index 0000000000..dc3867b17e --- /dev/null +++ b/intl/icu/source/common/uinit.cpp @@ -0,0 +1,74 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 2001-2015, International Business Machines +* Corporation and others. All Rights Reserved. +****************************************************************************** +* file name: uinit.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2001July05 +* created by: George Rhoten +*/ + +#include "unicode/utypes.h" +#include "unicode/icuplug.h" +#include "unicode/uclean.h" +#include "cmemory.h" +#include "icuplugimp.h" +#include "ucln_cmn.h" +#include "ucnv_io.h" +#include "umutex.h" +#include "utracimp.h" + +U_NAMESPACE_BEGIN + +static UInitOnce gICUInitOnce {}; + +static UBool U_CALLCONV uinit_cleanup() { + gICUInitOnce.reset(); + return true; +} + +static void U_CALLCONV +initData(UErrorCode &status) +{ +#if UCONFIG_ENABLE_PLUGINS + /* initialize plugins */ + uplug_init(&status); +#endif + +#if !UCONFIG_NO_CONVERSION + /* + * 2005-may-02 + * + * ICU4C 3.4 (jitterbug 4497) hardcodes the data for Unicode character + * properties for APIs that want to be fast. + * Therefore, we need not load them here nor check for errors. + * Instead, we load the converter alias table to see if any ICU data + * is available. + * Users should really open the service objects they need and check + * for errors there, to make sure that the actual items they need are + * available. + */ + ucnv_io_countKnownConverters(&status); +#endif + ucln_common_registerCleanup(UCLN_COMMON_UINIT, uinit_cleanup); +} + +U_NAMESPACE_END + +U_NAMESPACE_USE + +/* + * ICU Initialization Function. Need not be called. + */ +U_CAPI void U_EXPORT2 +u_init(UErrorCode *status) { + UTRACE_ENTRY_OC(UTRACE_U_INIT); + umtx_initOnce(gICUInitOnce, &initData, *status); + UTRACE_EXIT_STATUS(*status); +} diff --git a/intl/icu/source/common/uinvchar.cpp b/intl/icu/source/common/uinvchar.cpp new file mode 100644 index 0000000000..096a8e28d1 --- /dev/null +++ b/intl/icu/source/common/uinvchar.cpp @@ -0,0 +1,627 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 1999-2010, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: uinvchar.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:2 +* +* created on: 2004sep14 +* created by: Markus W. Scherer +* +* Functions for handling invariant characters, moved here from putil.c +* for better modularization. +*/ + +#include "unicode/utypes.h" +#include "unicode/ustring.h" +#include "udataswp.h" +#include "cstring.h" +#include "cmemory.h" +#include "uassert.h" +#include "uinvchar.h" + +/* invariant-character handling --------------------------------------------- */ + +/* + * These maps for ASCII to/from EBCDIC map invariant characters (see utypes.h) + * appropriately for most EBCDIC codepages. + * + * They currently also map most other ASCII graphic characters, + * appropriately for codepages 37 and 1047. + * Exceptions: The characters for []^ have different codes in 37 & 1047. + * Both versions are mapped to ASCII. + * + * ASCII 37 1047 + * [ 5B BA AD + * ] 5D BB BD + * ^ 5E B0 5F + * + * There are no mappings for variant characters from Unicode to EBCDIC. + * + * Currently, C0 control codes are also included in these maps. + * Exceptions: S/390 Open Edition swaps LF and NEL codes compared with other + * EBCDIC platforms; both codes (15 and 25) are mapped to ASCII LF (0A), + * but there is no mapping for ASCII LF back to EBCDIC. + * + * ASCII EBCDIC S/390-OE + * LF 0A 25 15 + * NEL 85 15 25 + * + * The maps below explicitly exclude the variant + * control and graphical characters that are in ASCII-based + * codepages at 0x80 and above. + * "No mapping" is expressed by mapping to a 00 byte. + * + * These tables do not establish a converter or a codepage. + */ + +static const uint8_t asciiFromEbcdic[256]={ + 0x00, 0x01, 0x02, 0x03, 0x00, 0x09, 0x00, 0x7f, 0x00, 0x00, 0x00, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, + 0x10, 0x11, 0x12, 0x13, 0x00, 0x0a, 0x08, 0x00, 0x18, 0x19, 0x00, 0x00, 0x1c, 0x1d, 0x1e, 0x1f, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x0a, 0x17, 0x1b, 0x00, 0x00, 0x00, 0x00, 0x00, 0x05, 0x06, 0x07, + 0x00, 0x00, 0x16, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x00, 0x14, 0x15, 0x00, 0x1a, + + 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x2e, 0x3c, 0x28, 0x2b, 0x7c, + 0x26, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x21, 0x24, 0x2a, 0x29, 0x3b, 0x5e, + 0x2d, 0x2f, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x2c, 0x25, 0x5f, 0x3e, 0x3f, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x60, 0x3a, 0x23, 0x40, 0x27, 0x3d, 0x22, + + 0x00, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x7e, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0x00, 0x00, 0x00, 0x5b, 0x00, 0x00, + 0x5e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x5b, 0x5d, 0x00, 0x5d, 0x00, 0x00, + + 0x7b, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x7d, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f, 0x50, 0x51, 0x52, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x5c, 0x00, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59, 0x5a, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 +}; + +static const uint8_t ebcdicFromAscii[256]={ + 0x00, 0x01, 0x02, 0x03, 0x37, 0x2d, 0x2e, 0x2f, 0x16, 0x05, 0x00, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, + 0x10, 0x11, 0x12, 0x13, 0x3c, 0x3d, 0x32, 0x26, 0x18, 0x19, 0x3f, 0x27, 0x1c, 0x1d, 0x1e, 0x1f, + 0x40, 0x00, 0x7f, 0x00, 0x00, 0x6c, 0x50, 0x7d, 0x4d, 0x5d, 0x5c, 0x4e, 0x6b, 0x60, 0x4b, 0x61, + 0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8, 0xf9, 0x7a, 0x5e, 0x4c, 0x7e, 0x6e, 0x6f, + + 0x00, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, 0xc7, 0xc8, 0xc9, 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, + 0xd7, 0xd8, 0xd9, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9, 0x00, 0x00, 0x00, 0x00, 0x6d, + 0x00, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, + 0x97, 0x98, 0x99, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0x00, 0x00, 0x00, 0x00, 0x07, + + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 +}; + +/* Same as asciiFromEbcdic[] except maps all letters to lowercase. */ +static const uint8_t lowercaseAsciiFromEbcdic[256]={ + 0x00, 0x01, 0x02, 0x03, 0x00, 0x09, 0x00, 0x7f, 0x00, 0x00, 0x00, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, + 0x10, 0x11, 0x12, 0x13, 0x00, 0x0a, 0x08, 0x00, 0x18, 0x19, 0x00, 0x00, 0x1c, 0x1d, 0x1e, 0x1f, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x0a, 0x17, 0x1b, 0x00, 0x00, 0x00, 0x00, 0x00, 0x05, 0x06, 0x07, + 0x00, 0x00, 0x16, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x00, 0x14, 0x15, 0x00, 0x1a, + + 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x2e, 0x3c, 0x28, 0x2b, 0x7c, + 0x26, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x21, 0x24, 0x2a, 0x29, 0x3b, 0x5e, + 0x2d, 0x2f, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x2c, 0x25, 0x5f, 0x3e, 0x3f, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x60, 0x3a, 0x23, 0x40, 0x27, 0x3d, 0x22, + + 0x00, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x7e, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0x00, 0x00, 0x00, 0x5b, 0x00, 0x00, + 0x5e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x5b, 0x5d, 0x00, 0x5d, 0x00, 0x00, + + 0x7b, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x7d, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x7c, 0x00, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 +}; + +/* + * Bit sets indicating which characters of the ASCII repertoire + * (by ASCII/Unicode code) are "invariant". + * See utypes.h for more details. + * + * As invariant are considered the characters of the ASCII repertoire except + * for the following: + * 21 '!' + * 23 '#' + * 24 '$' + * + * 40 '@' + * + * 5b '[' + * 5c '\' + * 5d ']' + * 5e '^' + * + * 60 '`' + * + * 7b '{' + * 7c '|' + * 7d '}' + * 7e '~' + */ +static const uint32_t invariantChars[4]={ + 0xfffffbff, /* 00..1f but not 0a */ + 0xffffffe5, /* 20..3f but not 21 23 24 */ + 0x87fffffe, /* 40..5f but not 40 5b..5e */ + 0x87fffffe /* 60..7f but not 60 7b..7e */ +}; + +/* + * test unsigned types (or values known to be non-negative) for invariant characters, + * tests ASCII-family character values + */ +#define UCHAR_IS_INVARIANT(c) (((c)<=0x7f) && (invariantChars[(c)>>5]&((uint32_t)1<<((c)&0x1f)))!=0) + +/* test signed types for invariant characters, adds test for positive values */ +#define SCHAR_IS_INVARIANT(c) ((0<=(c)) && UCHAR_IS_INVARIANT(c)) + +#if U_CHARSET_FAMILY==U_ASCII_FAMILY +#define CHAR_TO_UCHAR(c) c +#define UCHAR_TO_CHAR(c) c +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY +#define CHAR_TO_UCHAR(u) asciiFromEbcdic[u] +#define UCHAR_TO_CHAR(u) ebcdicFromAscii[u] +#else +# error U_CHARSET_FAMILY is not valid +#endif + + +U_CAPI void U_EXPORT2 +u_charsToUChars(const char *cs, char16_t *us, int32_t length) { + char16_t u; + uint8_t c; + + /* + * Allow the entire ASCII repertoire to be mapped _to_ Unicode. + * For EBCDIC systems, this works for characters with codes from + * codepages 37 and 1047 or compatible. + */ + while(length>0) { + c=(uint8_t)(*cs++); + u=(char16_t)CHAR_TO_UCHAR(c); + U_ASSERT((u!=0 || c==0)); /* only invariant chars converted? */ + *us++=u; + --length; + } +} + +U_CAPI void U_EXPORT2 +u_UCharsToChars(const char16_t *us, char *cs, int32_t length) { + char16_t u; + + while(length>0) { + u=*us++; + if(!UCHAR_IS_INVARIANT(u)) { + U_ASSERT(false); /* Variant characters were used. These are not portable in ICU. */ + u=0; + } + *cs++=(char)UCHAR_TO_CHAR(u); + --length; + } +} + +U_CAPI UBool U_EXPORT2 +uprv_isInvariantString(const char *s, int32_t length) { + uint8_t c; + + for(;;) { + if(length<0) { + /* NUL-terminated */ + c=(uint8_t)*s++; + if(c==0) { + break; + } + } else { + /* count length */ + if(length==0) { + break; + } + --length; + c=(uint8_t)*s++; + if(c==0) { + continue; /* NUL is invariant */ + } + } + /* c!=0 now, one branch below checks c==0 for variant characters */ + + /* + * no assertions here because these functions are legitimately called + * for strings with variant characters + */ +#if U_CHARSET_FAMILY==U_ASCII_FAMILY + if(!UCHAR_IS_INVARIANT(c)) { + return false; /* found a variant char */ + } +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY + c=CHAR_TO_UCHAR(c); + if(c==0 || !UCHAR_IS_INVARIANT(c)) { + return false; /* found a variant char */ + } +#else +# error U_CHARSET_FAMILY is not valid +#endif + } + return true; +} + +U_CAPI UBool U_EXPORT2 +uprv_isInvariantUString(const char16_t *s, int32_t length) { + char16_t c; + + for(;;) { + if(length<0) { + /* NUL-terminated */ + c=*s++; + if(c==0) { + break; + } + } else { + /* count length */ + if(length==0) { + break; + } + --length; + c=*s++; + } + + /* + * no assertions here because these functions are legitimately called + * for strings with variant characters + */ + if(!UCHAR_IS_INVARIANT(c)) { + return false; /* found a variant char */ + } + } + return true; +} + +/* UDataSwapFn implementations used in udataswp.c ------- */ + +/* convert ASCII to EBCDIC and verify that all characters are invariant */ +U_CAPI int32_t U_EXPORT2 +uprv_ebcdicFromAscii(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const uint8_t *s; + uint8_t *t; + uint8_t c; + + int32_t count; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<0 || (length>0 && outData==nullptr)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* setup and swapping */ + s=(const uint8_t *)inData; + t=(uint8_t *)outData; + count=length; + while(count>0) { + c=*s++; + if(!UCHAR_IS_INVARIANT(c)) { + udata_printError(ds, "uprv_ebcdicFromAscii() string[%d] contains a variant character in position %d\n", + length, length-count); + *pErrorCode=U_INVALID_CHAR_FOUND; + return 0; + } + *t++=ebcdicFromAscii[c]; + --count; + } + + return length; +} + +/* this function only checks and copies ASCII strings without conversion */ +U_CFUNC int32_t +uprv_copyAscii(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const uint8_t *s; + uint8_t c; + + int32_t count; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<0 || (length>0 && outData==nullptr)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* setup and checking */ + s=(const uint8_t *)inData; + count=length; + while(count>0) { + c=*s++; + if(!UCHAR_IS_INVARIANT(c)) { + udata_printError(ds, "uprv_copyFromAscii() string[%d] contains a variant character in position %d\n", + length, length-count); + *pErrorCode=U_INVALID_CHAR_FOUND; + return 0; + } + --count; + } + + if(length>0 && inData!=outData) { + uprv_memcpy(outData, inData, length); + } + + return length; +} + +/* convert EBCDIC to ASCII and verify that all characters are invariant */ +U_CFUNC int32_t +uprv_asciiFromEbcdic(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const uint8_t *s; + uint8_t *t; + uint8_t c; + + int32_t count; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<0 || (length>0 && outData==nullptr)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* setup and swapping */ + s=(const uint8_t *)inData; + t=(uint8_t *)outData; + count=length; + while(count>0) { + c=*s++; + if(c!=0 && ((c=asciiFromEbcdic[c])==0 || !UCHAR_IS_INVARIANT(c))) { + udata_printError(ds, "uprv_asciiFromEbcdic() string[%d] contains a variant character in position %d\n", + length, length-count); + *pErrorCode=U_INVALID_CHAR_FOUND; + return 0; + } + *t++=c; + --count; + } + + return length; +} + +/* this function only checks and copies EBCDIC strings without conversion */ +U_CFUNC int32_t +uprv_copyEbcdic(const UDataSwapper *ds, + const void *inData, int32_t length, void *outData, + UErrorCode *pErrorCode) { + const uint8_t *s; + uint8_t c; + + int32_t count; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } + if(ds==nullptr || inData==nullptr || length<0 || (length>0 && outData==nullptr)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + /* setup and checking */ + s=(const uint8_t *)inData; + count=length; + while(count>0) { + c=*s++; + if(c!=0 && ((c=asciiFromEbcdic[c])==0 || !UCHAR_IS_INVARIANT(c))) { + udata_printError(ds, "uprv_copyEbcdic() string[%] contains a variant character in position %d\n", + length, length-count); + *pErrorCode=U_INVALID_CHAR_FOUND; + return 0; + } + --count; + } + + if(length>0 && inData!=outData) { + uprv_memcpy(outData, inData, length); + } + + return length; +} + +U_CFUNC UBool +uprv_isEbcdicAtSign(char c) { + static const uint8_t ebcdicAtSigns[] = { + 0x7C, 0x44, 0x66, 0x80, 0xAC, 0xAE, 0xAF, 0xB5, 0xEC, 0xEF, 0x00 }; + return c != 0 && uprv_strchr((const char *)ebcdicAtSigns, c) != nullptr; +} + +/* compare invariant strings; variant characters compare less than others and unlike each other */ +U_CFUNC int32_t +uprv_compareInvAscii(const UDataSwapper *ds, + const char *outString, int32_t outLength, + const char16_t *localString, int32_t localLength) { + (void)ds; + int32_t minLength; + UChar32 c1, c2; + uint8_t c; + + if(outString==nullptr || outLength<-1 || localString==nullptr || localLength<-1) { + return 0; + } + + if(outLength<0) { + outLength=(int32_t)uprv_strlen(outString); + } + if(localLength<0) { + localLength=u_strlen(localString); + } + + minLength= outLength0) { + c=(uint8_t)*outString++; + if(UCHAR_IS_INVARIANT(c)) { + c1=c; + } else { + c1=-1; + } + + c2=*localString++; + if(!UCHAR_IS_INVARIANT(c2)) { + c2=-2; + } + + if((c1-=c2)!=0) { + return c1; + } + + --minLength; + } + + /* strings start with same prefix, compare lengths */ + return outLength-localLength; +} + +U_CFUNC int32_t +uprv_compareInvEbcdic(const UDataSwapper *ds, + const char *outString, int32_t outLength, + const char16_t *localString, int32_t localLength) { + (void)ds; + int32_t minLength; + UChar32 c1, c2; + uint8_t c; + + if(outString==nullptr || outLength<-1 || localString==nullptr || localLength<-1) { + return 0; + } + + if(outLength<0) { + outLength=(int32_t)uprv_strlen(outString); + } + if(localLength<0) { + localLength=u_strlen(localString); + } + + minLength= outLength0) { + c=(uint8_t)*outString++; + if(c==0) { + c1=0; + } else if((c1=asciiFromEbcdic[c])!=0 && UCHAR_IS_INVARIANT(c1)) { + /* c1 is set */ + } else { + c1=-1; + } + + c2=*localString++; + if(!UCHAR_IS_INVARIANT(c2)) { + c2=-2; + } + + if((c1-=c2)!=0) { + return c1; + } + + --minLength; + } + + /* strings start with same prefix, compare lengths */ + return outLength-localLength; +} + +U_CAPI int32_t U_EXPORT2 +uprv_compareInvEbcdicAsAscii(const char *s1, const char *s2) { + int32_t c1, c2; + + for(;; ++s1, ++s2) { + c1=(uint8_t)*s1; + c2=(uint8_t)*s2; + if(c1!=c2) { + if(c1!=0 && ((c1=asciiFromEbcdic[c1])==0 || !UCHAR_IS_INVARIANT(c1))) { + c1=-(int32_t)(uint8_t)*s1; + } + if(c2!=0 && ((c2=asciiFromEbcdic[c2])==0 || !UCHAR_IS_INVARIANT(c2))) { + c2=-(int32_t)(uint8_t)*s2; + } + return c1-c2; + } else if(c1==0) { + return 0; + } + } +} + +U_CAPI char U_EXPORT2 +uprv_ebcdicToAscii(char c) { + return (char)asciiFromEbcdic[(uint8_t)c]; +} + +U_CAPI char U_EXPORT2 +uprv_ebcdicToLowercaseAscii(char c) { + return (char)lowercaseAsciiFromEbcdic[(uint8_t)c]; +} + +U_CAPI uint8_t* U_EXPORT2 +uprv_aestrncpy(uint8_t *dst, const uint8_t *src, int32_t n) +{ + uint8_t *orig_dst = dst; + + if(n==-1) { + n = static_cast(uprv_strlen((const char*)src)+1); /* copy NUL */ + } + /* copy non-null */ + while(*src && n>0) { + *(dst++) = asciiFromEbcdic[*(src++)]; + n--; + } + /* pad */ + while(n>0) { + *(dst++) = 0; + n--; + } + return orig_dst; +} + +U_CAPI uint8_t* U_EXPORT2 +uprv_eastrncpy(uint8_t *dst, const uint8_t *src, int32_t n) +{ + uint8_t *orig_dst = dst; + + if(n==-1) { + n = static_cast(uprv_strlen((const char*)src)+1); /* copy NUL */ + } + /* copy non-null */ + while(*src && n>0) { + char ch = ebcdicFromAscii[*(src++)]; + if(ch == 0) { + ch = ebcdicFromAscii[0x3f]; /* questionmark (subchar) */ + } + *(dst++) = ch; + n--; + } + /* pad */ + while(n>0) { + *(dst++) = 0; + n--; + } + return orig_dst; +} + diff --git a/intl/icu/source/common/uinvchar.h b/intl/icu/source/common/uinvchar.h new file mode 100644 index 0000000000..9b7a9bd114 --- /dev/null +++ b/intl/icu/source/common/uinvchar.h @@ -0,0 +1,219 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 1999-2015, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: uinvchar.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:2 +* +* created on: 2004sep14 +* created by: Markus W. Scherer +* +* Definitions for handling invariant characters, moved here from putil.c +* for better modularization. +*/ + +#ifndef __UINVCHAR_H__ +#define __UINVCHAR_H__ + +#include "unicode/utypes.h" +#ifdef __cplusplus +#include "unicode/unistr.h" +#endif + +/** + * Check if a char string only contains invariant characters. + * See utypes.h for details. + * + * @param s Input string pointer. + * @param length Length of the string, can be -1 if NUL-terminated. + * @return true if s contains only invariant characters. + * + * @internal (ICU 2.8) + */ +U_CAPI UBool U_EXPORT2 +uprv_isInvariantString(const char *s, int32_t length); + +/** + * Check if a Unicode string only contains invariant characters. + * See utypes.h for details. + * + * @param s Input string pointer. + * @param length Length of the string, can be -1 if NUL-terminated. + * @return true if s contains only invariant characters. + * + * @internal (ICU 2.8) + */ +U_CAPI UBool U_EXPORT2 +uprv_isInvariantUString(const UChar *s, int32_t length); + +/** + * \def U_UPPER_ORDINAL + * Get the ordinal number of an uppercase invariant character + * @internal + */ +#if U_CHARSET_FAMILY==U_ASCII_FAMILY +# define U_UPPER_ORDINAL(x) ((x)-'A') +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY +# define U_UPPER_ORDINAL(x) (((x) < 'J') ? ((x)-'A') : \ + (((x) < 'S') ? ((x)-'J'+9) : \ + ((x)-'S'+18))) +#else +# error Unknown charset family! +#endif + +#ifdef __cplusplus + +U_NAMESPACE_BEGIN + +/** + * Like U_UPPER_ORDINAL(x) but with validation. + * Returns 0..25 for A..Z else a value outside 0..25. + */ +inline int32_t uprv_upperOrdinal(int32_t c) { +#if U_CHARSET_FAMILY==U_ASCII_FAMILY + return c - 'A'; +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY + // EBCDIC: A-Z (26 letters) is split into three ranges A-I (9 letters), J-R (9), S-Z (8). + // https://en.wikipedia.org/wiki/EBCDIC_037#Codepage_layout + if (c <= 'I') { return c - 'A'; } // A-I --> 0-8 + if (c < 'J') { return -1; } + if (c <= 'R') { return c - 'J' + 9; } // J-R --> 9..17 + if (c < 'S') { return -1; } + return c - 'S' + 18; // S-Z --> 18..25 +#else +# error Unknown charset family! +#endif +} + +// Like U_UPPER_ORDINAL(x) but for lowercase and with validation. +// Returns 0..25 for a..z else a value outside 0..25. +inline int32_t uprv_lowerOrdinal(int32_t c) { +#if U_CHARSET_FAMILY==U_ASCII_FAMILY + return c - 'a'; +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY + // EBCDIC: a-z (26 letters) is split into three ranges a-i (9 letters), j-r (9), s-z (8). + // https://en.wikipedia.org/wiki/EBCDIC_037#Codepage_layout + if (c <= 'i') { return c - 'a'; } // a-i --> 0-8 + if (c < 'j') { return -1; } + if (c <= 'r') { return c - 'j' + 9; } // j-r --> 9..17 + if (c < 's') { return -1; } + return c - 's' + 18; // s-z --> 18..25 +#else +# error Unknown charset family! +#endif +} + +U_NAMESPACE_END + +#endif + +/** + * Returns true if c == '@' is possible. + * The @ sign is variant, and the @ sign used on one + * EBCDIC machine won't be compiled the same way on other EBCDIC based machines. + * @internal + */ +U_CFUNC UBool +uprv_isEbcdicAtSign(char c); + +/** + * \def uprv_isAtSign + * Returns true if c == '@' is possible. + * For ASCII, checks for exactly '@'. For EBCDIC, calls uprv_isEbcdicAtSign(). + * @internal + */ +#if U_CHARSET_FAMILY==U_ASCII_FAMILY +# define uprv_isAtSign(c) ((c)=='@') +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY +# define uprv_isAtSign(c) uprv_isEbcdicAtSign(c) +#else +# error Unknown charset family! +#endif + +/** + * Compare two EBCDIC invariant-character strings in ASCII order. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +uprv_compareInvEbcdicAsAscii(const char *s1, const char *s2); + +/** + * \def uprv_compareInvCharsAsAscii + * Compare two invariant-character strings in ASCII order. + * @internal + */ +#if U_CHARSET_FAMILY==U_ASCII_FAMILY +# define uprv_compareInvCharsAsAscii(s1, s2) uprv_strcmp(s1, s2) +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY +# define uprv_compareInvCharsAsAscii(s1, s2) uprv_compareInvEbcdicAsAscii(s1, s2) +#else +# error Unknown charset family! +#endif + +/** + * Converts an EBCDIC invariant character to ASCII. + * @internal + */ +U_CAPI char U_EXPORT2 +uprv_ebcdicToAscii(char c); + +/** + * \def uprv_invCharToAscii + * Converts an invariant character to ASCII. + * @internal + */ +#if U_CHARSET_FAMILY==U_ASCII_FAMILY +# define uprv_invCharToAscii(c) (c) +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY +# define uprv_invCharToAscii(c) uprv_ebcdicToAscii(c) +#else +# error Unknown charset family! +#endif + +/** + * Converts an EBCDIC invariant character to lowercase ASCII. + * @internal + */ +U_CAPI char U_EXPORT2 +uprv_ebcdicToLowercaseAscii(char c); + +/** + * \def uprv_invCharToLowercaseAscii + * Converts an invariant character to lowercase ASCII. + * @internal + */ +#if U_CHARSET_FAMILY==U_ASCII_FAMILY +# define uprv_invCharToLowercaseAscii uprv_asciitolower +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY +# define uprv_invCharToLowercaseAscii uprv_ebcdicToLowercaseAscii +#else +# error Unknown charset family! +#endif + +/** + * Copy EBCDIC to ASCII + * @internal + * @see uprv_strncpy + */ +U_CAPI uint8_t* U_EXPORT2 +uprv_aestrncpy(uint8_t *dst, const uint8_t *src, int32_t n); + + +/** + * Copy ASCII to EBCDIC + * @internal + * @see uprv_strncpy + */ +U_CAPI uint8_t* U_EXPORT2 +uprv_eastrncpy(uint8_t *dst, const uint8_t *src, int32_t n); + + + +#endif diff --git a/intl/icu/source/common/uiter.cpp b/intl/icu/source/common/uiter.cpp new file mode 100644 index 0000000000..be59eab2ee --- /dev/null +++ b/intl/icu/source/common/uiter.cpp @@ -0,0 +1,1108 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2002-2012, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: uiter.cpp +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002jan18 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" +#include "unicode/ustring.h" +#include "unicode/chariter.h" +#include "unicode/rep.h" +#include "unicode/uiter.h" +#include "unicode/utf.h" +#include "unicode/utf8.h" +#include "unicode/utf16.h" +#include "cstring.h" + +U_NAMESPACE_USE + +#define IS_EVEN(n) (((n)&1)==0) +#define IS_POINTER_EVEN(p) IS_EVEN((size_t)p) + +U_CDECL_BEGIN + +/* No-Op UCharIterator implementation for illegal input --------------------- */ + +static int32_t U_CALLCONV +noopGetIndex(UCharIterator * /*iter*/, UCharIteratorOrigin /*origin*/) { + return 0; +} + +static int32_t U_CALLCONV +noopMove(UCharIterator * /*iter*/, int32_t /*delta*/, UCharIteratorOrigin /*origin*/) { + return 0; +} + +static UBool U_CALLCONV +noopHasNext(UCharIterator * /*iter*/) { + return false; +} + +static UChar32 U_CALLCONV +noopCurrent(UCharIterator * /*iter*/) { + return U_SENTINEL; +} + +static uint32_t U_CALLCONV +noopGetState(const UCharIterator * /*iter*/) { + return UITER_NO_STATE; +} + +static void U_CALLCONV +noopSetState(UCharIterator * /*iter*/, uint32_t /*state*/, UErrorCode *pErrorCode) { + *pErrorCode=U_UNSUPPORTED_ERROR; +} + +static const UCharIterator noopIterator={ + 0, 0, 0, 0, 0, 0, + noopGetIndex, + noopMove, + noopHasNext, + noopHasNext, + noopCurrent, + noopCurrent, + noopCurrent, + nullptr, + noopGetState, + noopSetState +}; + +/* UCharIterator implementation for simple strings -------------------------- */ + +/* + * This is an implementation of a code unit (char16_t) iterator + * for char16_t * strings. + * + * The UCharIterator.context field holds a pointer to the string. + */ + +static int32_t U_CALLCONV +stringIteratorGetIndex(UCharIterator *iter, UCharIteratorOrigin origin) UPRV_NO_SANITIZE_UNDEFINED { + switch(origin) { + case UITER_ZERO: + return 0; + case UITER_START: + return iter->start; + case UITER_CURRENT: + return iter->index; + case UITER_LIMIT: + return iter->limit; + case UITER_LENGTH: + return iter->length; + default: + /* not a valid origin */ + /* Should never get here! */ + return -1; + } +} + +static int32_t U_CALLCONV +stringIteratorMove(UCharIterator *iter, int32_t delta, UCharIteratorOrigin origin) UPRV_NO_SANITIZE_UNDEFINED { + int32_t pos; + + switch(origin) { + case UITER_ZERO: + pos=delta; + break; + case UITER_START: + pos=iter->start+delta; + break; + case UITER_CURRENT: + pos=iter->index+delta; + break; + case UITER_LIMIT: + pos=iter->limit+delta; + break; + case UITER_LENGTH: + pos=iter->length+delta; + break; + default: + return -1; /* Error */ + } + + if(posstart) { + pos=iter->start; + } else if(pos>iter->limit) { + pos=iter->limit; + } + + return iter->index=pos; +} + +static UBool U_CALLCONV +stringIteratorHasNext(UCharIterator *iter) { + return iter->indexlimit; +} + +static UBool U_CALLCONV +stringIteratorHasPrevious(UCharIterator *iter) { + return iter->index>iter->start; +} + +static UChar32 U_CALLCONV +stringIteratorCurrent(UCharIterator *iter) { + if(iter->indexlimit) { + return ((const char16_t *)(iter->context))[iter->index]; + } else { + return U_SENTINEL; + } +} + +static UChar32 U_CALLCONV +stringIteratorNext(UCharIterator *iter) { + if(iter->indexlimit) { + return ((const char16_t *)(iter->context))[iter->index++]; + } else { + return U_SENTINEL; + } +} + +static UChar32 U_CALLCONV +stringIteratorPrevious(UCharIterator *iter) { + if(iter->index>iter->start) { + return ((const char16_t *)(iter->context))[--iter->index]; + } else { + return U_SENTINEL; + } +} + +static uint32_t U_CALLCONV +stringIteratorGetState(const UCharIterator *iter) { + return (uint32_t)iter->index; +} + +static void U_CALLCONV +stringIteratorSetState(UCharIterator *iter, uint32_t state, UErrorCode *pErrorCode) { + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + /* do nothing */ + } else if(iter==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + } else if((int32_t)statestart || iter->limit<(int32_t)state) { + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + } else { + iter->index=(int32_t)state; + } +} + +static const UCharIterator stringIterator={ + 0, 0, 0, 0, 0, 0, + stringIteratorGetIndex, + stringIteratorMove, + stringIteratorHasNext, + stringIteratorHasPrevious, + stringIteratorCurrent, + stringIteratorNext, + stringIteratorPrevious, + nullptr, + stringIteratorGetState, + stringIteratorSetState +}; + +U_CAPI void U_EXPORT2 +uiter_setString(UCharIterator *iter, const char16_t *s, int32_t length) { + if(iter!=0) { + if(s!=0 && length>=-1) { + *iter=stringIterator; + iter->context=s; + if(length>=0) { + iter->length=length; + } else { + iter->length=u_strlen(s); + } + iter->limit=iter->length; + } else { + *iter=noopIterator; + } + } +} + +/* UCharIterator implementation for UTF-16BE strings ------------------------ */ + +/* + * This is an implementation of a code unit (char16_t) iterator + * for UTF-16BE strings, i.e., strings in byte-vectors where + * each char16_t is stored as a big-endian pair of bytes. + * + * The UCharIterator.context field holds a pointer to the string. + * Everything works just like with a normal char16_t iterator (uiter_setString), + * except that UChars are assembled from byte pairs. + */ + +/* internal helper function */ +static inline UChar32 +utf16BEIteratorGet(UCharIterator *iter, int32_t index) { + const uint8_t *p=(const uint8_t *)iter->context; + return ((char16_t)p[2*index]<<8)|(char16_t)p[2*index+1]; +} + +static UChar32 U_CALLCONV +utf16BEIteratorCurrent(UCharIterator *iter) { + int32_t index; + + if((index=iter->index)limit) { + return utf16BEIteratorGet(iter, index); + } else { + return U_SENTINEL; + } +} + +static UChar32 U_CALLCONV +utf16BEIteratorNext(UCharIterator *iter) { + int32_t index; + + if((index=iter->index)limit) { + iter->index=index+1; + return utf16BEIteratorGet(iter, index); + } else { + return U_SENTINEL; + } +} + +static UChar32 U_CALLCONV +utf16BEIteratorPrevious(UCharIterator *iter) { + int32_t index; + + if((index=iter->index)>iter->start) { + iter->index=--index; + return utf16BEIteratorGet(iter, index); + } else { + return U_SENTINEL; + } +} + +static const UCharIterator utf16BEIterator={ + 0, 0, 0, 0, 0, 0, + stringIteratorGetIndex, + stringIteratorMove, + stringIteratorHasNext, + stringIteratorHasPrevious, + utf16BEIteratorCurrent, + utf16BEIteratorNext, + utf16BEIteratorPrevious, + nullptr, + stringIteratorGetState, + stringIteratorSetState +}; + +/* + * Count the number of UChars in a UTF-16BE string before a terminating char16_t NUL, + * i.e., before a pair of 0 bytes where the first 0 byte is at an even + * offset from s. + */ +static int32_t +utf16BE_strlen(const char *s) { + if(IS_POINTER_EVEN(s)) { + /* + * even-aligned, call u_strlen(s) + * we are probably on a little-endian machine, but searching for char16_t NUL + * does not care about endianness + */ + return u_strlen((const char16_t *)s); + } else { + /* odd-aligned, search for pair of 0 bytes */ + const char *p=s; + + while(!(*p==0 && p[1]==0)) { + p+=2; + } + return (int32_t)((p-s)/2); + } +} + +U_CAPI void U_EXPORT2 +uiter_setUTF16BE(UCharIterator *iter, const char *s, int32_t length) { + if(iter!=nullptr) { + /* allow only even-length strings (the input length counts bytes) */ + if(s!=nullptr && (length==-1 || (length>=0 && IS_EVEN(length)))) { + /* length/=2, except that >>=1 also works for -1 (-1/2==0, -1>>1==-1) */ + length>>=1; + + if(U_IS_BIG_ENDIAN && IS_POINTER_EVEN(s)) { + /* big-endian machine and 2-aligned UTF-16BE string: use normal char16_t iterator */ + uiter_setString(iter, (const char16_t *)s, length); + return; + } + + *iter=utf16BEIterator; + iter->context=s; + if(length>=0) { + iter->length=length; + } else { + iter->length=utf16BE_strlen(s); + } + iter->limit=iter->length; + } else { + *iter=noopIterator; + } + } +} + +/* UCharIterator wrapper around CharacterIterator --------------------------- */ + +/* + * This is wrapper code around a C++ CharacterIterator to + * look like a C UCharIterator. + * + * The UCharIterator.context field holds a pointer to the CharacterIterator. + */ + +static int32_t U_CALLCONV +characterIteratorGetIndex(UCharIterator *iter, UCharIteratorOrigin origin) UPRV_NO_SANITIZE_UNDEFINED { + switch(origin) { + case UITER_ZERO: + return 0; + case UITER_START: + return ((CharacterIterator *)(iter->context))->startIndex(); + case UITER_CURRENT: + return ((CharacterIterator *)(iter->context))->getIndex(); + case UITER_LIMIT: + return ((CharacterIterator *)(iter->context))->endIndex(); + case UITER_LENGTH: + return ((CharacterIterator *)(iter->context))->getLength(); + default: + /* not a valid origin */ + /* Should never get here! */ + return -1; + } +} + +static int32_t U_CALLCONV +characterIteratorMove(UCharIterator *iter, int32_t delta, UCharIteratorOrigin origin) UPRV_NO_SANITIZE_UNDEFINED { + switch(origin) { + case UITER_ZERO: + ((CharacterIterator *)(iter->context))->setIndex(delta); + return ((CharacterIterator *)(iter->context))->getIndex(); + case UITER_START: + case UITER_CURRENT: + case UITER_LIMIT: + return ((CharacterIterator *)(iter->context))->move(delta, (CharacterIterator::EOrigin)origin); + case UITER_LENGTH: + ((CharacterIterator *)(iter->context))->setIndex(((CharacterIterator *)(iter->context))->getLength()+delta); + return ((CharacterIterator *)(iter->context))->getIndex(); + default: + /* not a valid origin */ + /* Should never get here! */ + return -1; + } +} + +static UBool U_CALLCONV +characterIteratorHasNext(UCharIterator *iter) { + return ((CharacterIterator *)(iter->context))->hasNext(); +} + +static UBool U_CALLCONV +characterIteratorHasPrevious(UCharIterator *iter) { + return ((CharacterIterator *)(iter->context))->hasPrevious(); +} + +static UChar32 U_CALLCONV +characterIteratorCurrent(UCharIterator *iter) { + UChar32 c; + + c=((CharacterIterator *)(iter->context))->current(); + if(c!=0xffff || ((CharacterIterator *)(iter->context))->hasNext()) { + return c; + } else { + return U_SENTINEL; + } +} + +static UChar32 U_CALLCONV +characterIteratorNext(UCharIterator *iter) { + if(((CharacterIterator *)(iter->context))->hasNext()) { + return ((CharacterIterator *)(iter->context))->nextPostInc(); + } else { + return U_SENTINEL; + } +} + +static UChar32 U_CALLCONV +characterIteratorPrevious(UCharIterator *iter) { + if(((CharacterIterator *)(iter->context))->hasPrevious()) { + return ((CharacterIterator *)(iter->context))->previous(); + } else { + return U_SENTINEL; + } +} + +static uint32_t U_CALLCONV +characterIteratorGetState(const UCharIterator *iter) { + return ((CharacterIterator *)(iter->context))->getIndex(); +} + +static void U_CALLCONV +characterIteratorSetState(UCharIterator *iter, uint32_t state, UErrorCode *pErrorCode) { + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + /* do nothing */ + } else if(iter==nullptr || iter->context==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + } else if((int32_t)state<((CharacterIterator *)(iter->context))->startIndex() || ((CharacterIterator *)(iter->context))->endIndex()<(int32_t)state) { + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + } else { + ((CharacterIterator *)(iter->context))->setIndex((int32_t)state); + } +} + +static const UCharIterator characterIteratorWrapper={ + 0, 0, 0, 0, 0, 0, + characterIteratorGetIndex, + characterIteratorMove, + characterIteratorHasNext, + characterIteratorHasPrevious, + characterIteratorCurrent, + characterIteratorNext, + characterIteratorPrevious, + nullptr, + characterIteratorGetState, + characterIteratorSetState +}; + +U_CAPI void U_EXPORT2 +uiter_setCharacterIterator(UCharIterator *iter, CharacterIterator *charIter) { + if(iter!=0) { + if(charIter!=0) { + *iter=characterIteratorWrapper; + iter->context=charIter; + } else { + *iter=noopIterator; + } + } +} + +/* UCharIterator wrapper around Replaceable --------------------------------- */ + +/* + * This is an implementation of a code unit (char16_t) iterator + * based on a Replaceable object. + * + * The UCharIterator.context field holds a pointer to the Replaceable. + * UCharIterator.length and UCharIterator.index hold Replaceable.length() + * and the iteration index. + */ + +static UChar32 U_CALLCONV +replaceableIteratorCurrent(UCharIterator *iter) { + if(iter->indexlimit) { + return ((Replaceable *)(iter->context))->charAt(iter->index); + } else { + return U_SENTINEL; + } +} + +static UChar32 U_CALLCONV +replaceableIteratorNext(UCharIterator *iter) { + if(iter->indexlimit) { + return ((Replaceable *)(iter->context))->charAt(iter->index++); + } else { + return U_SENTINEL; + } +} + +static UChar32 U_CALLCONV +replaceableIteratorPrevious(UCharIterator *iter) { + if(iter->index>iter->start) { + return ((Replaceable *)(iter->context))->charAt(--iter->index); + } else { + return U_SENTINEL; + } +} + +static const UCharIterator replaceableIterator={ + 0, 0, 0, 0, 0, 0, + stringIteratorGetIndex, + stringIteratorMove, + stringIteratorHasNext, + stringIteratorHasPrevious, + replaceableIteratorCurrent, + replaceableIteratorNext, + replaceableIteratorPrevious, + nullptr, + stringIteratorGetState, + stringIteratorSetState +}; + +U_CAPI void U_EXPORT2 +uiter_setReplaceable(UCharIterator *iter, const Replaceable *rep) { + if(iter!=0) { + if(rep!=0) { + *iter=replaceableIterator; + iter->context=rep; + iter->limit=iter->length=rep->length(); + } else { + *iter=noopIterator; + } + } +} + +/* UCharIterator implementation for UTF-8 strings --------------------------- */ + +/* + * Possible, probably necessary only for an implementation for arbitrary + * converters: + * Maintain a buffer (ring buffer?) for a piece of converted 16-bit text. + * This would require to turn reservedFn into a close function and + * to introduce a uiter_close(iter). + */ + +#define UITER_CNV_CAPACITY 16 + +/* + * Minimal implementation: + * Maintain a single-char16_t buffer for an additional surrogate. + * The caller must not modify start and limit because they are used internally. + * + * Use UCharIterator fields as follows: + * context pointer to UTF-8 string + * length UTF-16 length of the string; -1 until lazy evaluation + * start current UTF-8 index + * index current UTF-16 index; may be -1="unknown" after setState() + * limit UTF-8 length of the string + * reservedField supplementary code point + * + * Since UCharIterator delivers 16-bit code units, the iteration can be + * currently in the middle of the byte sequence for a supplementary code point. + * In this case, reservedField will contain that code point and start will + * point to after the corresponding byte sequence. The UTF-16 index will be + * one less than what it would otherwise be corresponding to the UTF-8 index. + * Otherwise, reservedField will be 0. + */ + +/* + * Possible optimization for NUL-terminated UTF-8 and UTF-16 strings: + * Add implementations that do not call strlen() for iteration but check for NUL. + */ + +static int32_t U_CALLCONV +utf8IteratorGetIndex(UCharIterator *iter, UCharIteratorOrigin origin) UPRV_NO_SANITIZE_UNDEFINED { + switch(origin) { + case UITER_ZERO: + case UITER_START: + return 0; + case UITER_CURRENT: + if(iter->index<0) { + /* the current UTF-16 index is unknown after setState(), count from the beginning */ + const uint8_t *s; + UChar32 c; + int32_t i, limit, index; + + s=(const uint8_t *)iter->context; + i=index=0; + limit=iter->start; /* count up to the UTF-8 index */ + while(istart=i; /* just in case setState() did not get us to a code point boundary */ + if(i==iter->limit) { + iter->length=index; /* in case it was <0 or wrong */ + } + if(iter->reservedField!=0) { + --index; /* we are in the middle of a supplementary code point */ + } + iter->index=index; + } + return iter->index; + case UITER_LIMIT: + case UITER_LENGTH: + if(iter->length<0) { + const uint8_t *s; + UChar32 c; + int32_t i, limit, length; + + s=(const uint8_t *)iter->context; + if(iter->index<0) { + /* + * the current UTF-16 index is unknown after setState(), + * we must first count from the beginning to here + */ + i=length=0; + limit=iter->start; + + /* count from the beginning to the current index */ + while(istart, set the UTF-16 index */ + iter->start=i; /* just in case setState() did not get us to a code point boundary */ + iter->index= iter->reservedField!=0 ? length-1 : length; + } else { + i=iter->start; + length=iter->index; + if(iter->reservedField!=0) { + ++length; + } + } + + /* count from the current index to the end */ + limit=iter->limit; + while(ilength=length; + } + return iter->length; + default: + /* not a valid origin */ + /* Should never get here! */ + return -1; + } +} + +static int32_t U_CALLCONV +utf8IteratorMove(UCharIterator *iter, int32_t delta, UCharIteratorOrigin origin) UPRV_NO_SANITIZE_UNDEFINED { + const uint8_t *s; + UChar32 c; + int32_t pos; /* requested UTF-16 index */ + int32_t i; /* UTF-8 index */ + UBool havePos; + + /* calculate the requested UTF-16 index */ + switch(origin) { + case UITER_ZERO: + case UITER_START: + pos=delta; + havePos=true; + /* iter->index<0 (unknown) is possible */ + break; + case UITER_CURRENT: + if(iter->index>=0) { + pos=iter->index+delta; + havePos=true; + } else { + /* the current UTF-16 index is unknown after setState(), use only delta */ + pos=0; + havePos=false; + } + break; + case UITER_LIMIT: + case UITER_LENGTH: + if(iter->length>=0) { + pos=iter->length+delta; + havePos=true; + } else { + /* pin to the end, avoid counting the length */ + iter->index=-1; + iter->start=iter->limit; + iter->reservedField=0; + if(delta>=0) { + return UITER_UNKNOWN_INDEX; + } else { + /* the current UTF-16 index is unknown, use only delta */ + pos=0; + havePos=false; + } + } + break; + default: + return -1; /* Error */ + } + + if(havePos) { + /* shortcuts: pinning to the edges of the string */ + if(pos<=0) { + iter->index=iter->start=iter->reservedField=0; + return 0; + } else if(iter->length>=0 && pos>=iter->length) { + iter->index=iter->length; + iter->start=iter->limit; + iter->reservedField=0; + return iter->index; + } + + /* minimize the number of U8_NEXT/PREV operations */ + if(iter->index<0 || posindex/2) { + /* go forward from the start instead of backward from the current index */ + iter->index=iter->start=iter->reservedField=0; + } else if(iter->length>=0 && (iter->length-pos)<(pos-iter->index)) { + /* + * if we have the UTF-16 index and length and the new position is + * closer to the end than the current index, + * then go backward from the end instead of forward from the current index + */ + iter->index=iter->length; + iter->start=iter->limit; + iter->reservedField=0; + } + + delta=pos-iter->index; + if(delta==0) { + return iter->index; /* nothing to do */ + } + } else { + /* move relative to unknown UTF-16 index */ + if(delta==0) { + return UITER_UNKNOWN_INDEX; /* nothing to do */ + } else if(-delta>=iter->start) { + /* moving backwards by more UChars than there are UTF-8 bytes, pin to 0 */ + iter->index=iter->start=iter->reservedField=0; + return 0; + } else if(delta>=(iter->limit-iter->start)) { + /* moving forward by more UChars than the remaining UTF-8 bytes, pin to the end */ + iter->index=iter->length; /* may or may not be <0 (unknown) */ + iter->start=iter->limit; + iter->reservedField=0; + return iter->index>=0 ? iter->index : (int32_t)UITER_UNKNOWN_INDEX; + } + } + + /* delta!=0 */ + + /* move towards the requested position, pin to the edges of the string */ + s=(const uint8_t *)iter->context; + pos=iter->index; /* could be <0 (unknown) */ + i=iter->start; + if(delta>0) { + /* go forward */ + int32_t limit=iter->limit; + if(iter->reservedField!=0) { + iter->reservedField=0; + ++pos; + --delta; + } + while(delta>0 && i=2) { + pos+=2; + delta-=2; + } else /* delta==1 */ { + /* stop in the middle of a supplementary code point */ + iter->reservedField=c; + ++pos; + break; /* delta=0; */ + } + } + if(i==limit) { + if(iter->length<0 && iter->index>=0) { + iter->length= iter->reservedField==0 ? pos : pos+1; + } else if(iter->index<0 && iter->length>=0) { + iter->index= iter->reservedField==0 ? iter->length : iter->length-1; + } + } + } else /* delta<0 */ { + /* go backward */ + if(iter->reservedField!=0) { + iter->reservedField=0; + i-=4; /* we stayed behind the supplementary code point; go before it now */ + --pos; + ++delta; + } + while(delta<0 && i>0) { + U8_PREV_OR_FFFD(s, 0, i, c); + if(c<=0xffff) { + --pos; + ++delta; + } else if(delta<=-2) { + pos-=2; + delta+=2; + } else /* delta==-1 */ { + /* stop in the middle of a supplementary code point */ + i+=4; /* back to behind this supplementary code point for consistent state */ + iter->reservedField=c; + --pos; + break; /* delta=0; */ + } + } + } + + iter->start=i; + if(iter->index>=0) { + return iter->index=pos; + } else { + /* we started with index<0 (unknown) so pos is bogus */ + if(i<=1) { + return iter->index=i; /* reached the beginning */ + } else { + /* we still don't know the UTF-16 index */ + return UITER_UNKNOWN_INDEX; + } + } +} + +static UBool U_CALLCONV +utf8IteratorHasNext(UCharIterator *iter) { + return iter->startlimit || iter->reservedField!=0; +} + +static UBool U_CALLCONV +utf8IteratorHasPrevious(UCharIterator *iter) { + return iter->start>0; +} + +static UChar32 U_CALLCONV +utf8IteratorCurrent(UCharIterator *iter) { + if(iter->reservedField!=0) { + return U16_TRAIL(iter->reservedField); + } else if(iter->startlimit) { + const uint8_t *s=(const uint8_t *)iter->context; + UChar32 c; + int32_t i=iter->start; + + U8_NEXT_OR_FFFD(s, i, iter->limit, c); + if(c<=0xffff) { + return c; + } else { + return U16_LEAD(c); + } + } else { + return U_SENTINEL; + } +} + +static UChar32 U_CALLCONV +utf8IteratorNext(UCharIterator *iter) { + int32_t index; + + if(iter->reservedField!=0) { + char16_t trail=U16_TRAIL(iter->reservedField); + iter->reservedField=0; + if((index=iter->index)>=0) { + iter->index=index+1; + } + return trail; + } else if(iter->startlimit) { + const uint8_t *s=(const uint8_t *)iter->context; + UChar32 c; + + U8_NEXT_OR_FFFD(s, iter->start, iter->limit, c); + if((index=iter->index)>=0) { + iter->index=++index; + if(iter->length<0 && iter->start==iter->limit) { + iter->length= c<=0xffff ? index : index+1; + } + } else if(iter->start==iter->limit && iter->length>=0) { + iter->index= c<=0xffff ? iter->length : iter->length-1; + } + if(c<=0xffff) { + return c; + } else { + iter->reservedField=c; + return U16_LEAD(c); + } + } else { + return U_SENTINEL; + } +} + +static UChar32 U_CALLCONV +utf8IteratorPrevious(UCharIterator *iter) { + int32_t index; + + if(iter->reservedField!=0) { + char16_t lead=U16_LEAD(iter->reservedField); + iter->reservedField=0; + iter->start-=4; /* we stayed behind the supplementary code point; go before it now */ + if((index=iter->index)>0) { + iter->index=index-1; + } + return lead; + } else if(iter->start>0) { + const uint8_t *s=(const uint8_t *)iter->context; + UChar32 c; + + U8_PREV_OR_FFFD(s, 0, iter->start, c); + if((index=iter->index)>0) { + iter->index=index-1; + } else if(iter->start<=1) { + iter->index= c<=0xffff ? iter->start : iter->start+1; + } + if(c<=0xffff) { + return c; + } else { + iter->start+=4; /* back to behind this supplementary code point for consistent state */ + iter->reservedField=c; + return U16_TRAIL(c); + } + } else { + return U_SENTINEL; + } +} + +static uint32_t U_CALLCONV +utf8IteratorGetState(const UCharIterator *iter) { + uint32_t state=(uint32_t)(iter->start<<1); + if(iter->reservedField!=0) { + state|=1; + } + return state; +} + +static void U_CALLCONV +utf8IteratorSetState(UCharIterator *iter, + uint32_t state, + UErrorCode *pErrorCode) +{ + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + /* do nothing */ + } else if(iter==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + } else if(state==utf8IteratorGetState(iter)) { + /* setting to the current state: no-op */ + } else { + int32_t index=(int32_t)(state>>1); /* UTF-8 index */ + state&=1; /* 1 if in surrogate pair, must be index>=4 */ + + if((state==0 ? index<0 : index<4) || iter->limitstart=index; /* restore UTF-8 byte index */ + if(index<=1) { + iter->index=index; + } else { + iter->index=-1; /* unknown UTF-16 index */ + } + if(state==0) { + iter->reservedField=0; + } else { + /* verified index>=4 above */ + UChar32 c; + U8_PREV_OR_FFFD((const uint8_t *)iter->context, 0, index, c); + if(c<=0xffff) { + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + } else { + iter->reservedField=c; + } + } + } + } +} + +static const UCharIterator utf8Iterator={ + 0, 0, 0, 0, 0, 0, + utf8IteratorGetIndex, + utf8IteratorMove, + utf8IteratorHasNext, + utf8IteratorHasPrevious, + utf8IteratorCurrent, + utf8IteratorNext, + utf8IteratorPrevious, + nullptr, + utf8IteratorGetState, + utf8IteratorSetState +}; + +U_CAPI void U_EXPORT2 +uiter_setUTF8(UCharIterator *iter, const char *s, int32_t length) { + if(iter!=0) { + if(s!=0 && length>=-1) { + *iter=utf8Iterator; + iter->context=s; + if(length>=0) { + iter->limit=length; + } else { + iter->limit=(int32_t)uprv_strlen(s); + } + iter->length= iter->limit<=1 ? iter->limit : -1; + } else { + *iter=noopIterator; + } + } +} + +/* Helper functions --------------------------------------------------------- */ + +U_CAPI UChar32 U_EXPORT2 +uiter_current32(UCharIterator *iter) { + UChar32 c, c2; + + c=iter->current(iter); + if(U16_IS_SURROGATE(c)) { + if(U16_IS_SURROGATE_LEAD(c)) { + /* + * go to the next code unit + * we know that we are not at the limit because c!=U_SENTINEL + */ + iter->move(iter, 1, UITER_CURRENT); + if(U16_IS_TRAIL(c2=iter->current(iter))) { + c=U16_GET_SUPPLEMENTARY(c, c2); + } + + /* undo index movement */ + iter->move(iter, -1, UITER_CURRENT); + } else { + if(U16_IS_LEAD(c2=iter->previous(iter))) { + c=U16_GET_SUPPLEMENTARY(c2, c); + } + if(c2>=0) { + /* undo index movement */ + iter->move(iter, 1, UITER_CURRENT); + } + } + } + return c; +} + +U_CAPI UChar32 U_EXPORT2 +uiter_next32(UCharIterator *iter) { + UChar32 c, c2; + + c=iter->next(iter); + if(U16_IS_LEAD(c)) { + if(U16_IS_TRAIL(c2=iter->next(iter))) { + c=U16_GET_SUPPLEMENTARY(c, c2); + } else if(c2>=0) { + /* unmatched first surrogate, undo index movement */ + iter->move(iter, -1, UITER_CURRENT); + } + } + return c; +} + +U_CAPI UChar32 U_EXPORT2 +uiter_previous32(UCharIterator *iter) { + UChar32 c, c2; + + c=iter->previous(iter); + if(U16_IS_TRAIL(c)) { + if(U16_IS_LEAD(c2=iter->previous(iter))) { + c=U16_GET_SUPPLEMENTARY(c2, c); + } else if(c2>=0) { + /* unmatched second surrogate, undo index movement */ + iter->move(iter, 1, UITER_CURRENT); + } + } + return c; +} + +U_CAPI uint32_t U_EXPORT2 +uiter_getState(const UCharIterator *iter) { + if(iter==nullptr || iter->getState==nullptr) { + return UITER_NO_STATE; + } else { + return iter->getState(iter); + } +} + +U_CAPI void U_EXPORT2 +uiter_setState(UCharIterator *iter, uint32_t state, UErrorCode *pErrorCode) { + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + /* do nothing */ + } else if(iter==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + } else if(iter->setState==nullptr) { + *pErrorCode=U_UNSUPPORTED_ERROR; + } else { + iter->setState(iter, state, pErrorCode); + } +} + +U_CDECL_END diff --git a/intl/icu/source/common/ulayout_props.h b/intl/icu/source/common/ulayout_props.h new file mode 100644 index 0000000000..c0f028c713 --- /dev/null +++ b/intl/icu/source/common/ulayout_props.h @@ -0,0 +1,46 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// ulayout_props.h +// created: 2019feb12 Markus W. Scherer + +#ifndef __ULAYOUT_PROPS_H__ +#define __ULAYOUT_PROPS_H__ + +#include "unicode/utypes.h" + +// file definitions ------------------------------------------------------------ + +#define ULAYOUT_DATA_NAME "ulayout" +#define ULAYOUT_DATA_TYPE "icu" + +// data format "Layo" +#define ULAYOUT_FMT_0 0x4c +#define ULAYOUT_FMT_1 0x61 +#define ULAYOUT_FMT_2 0x79 +#define ULAYOUT_FMT_3 0x6f + +// indexes into indexes[] +enum { + // Element 0 stores the length of the indexes[] array. + ULAYOUT_IX_INDEXES_LENGTH, + // Elements 1..7 store the tops of consecutive code point tries. + // No trie is stored if the difference between two of these is less than 16. + ULAYOUT_IX_INPC_TRIE_TOP, + ULAYOUT_IX_INSC_TRIE_TOP, + ULAYOUT_IX_VO_TRIE_TOP, + ULAYOUT_IX_RESERVED_TOP, + + ULAYOUT_IX_TRIES_TOP = 7, + + ULAYOUT_IX_MAX_VALUES = 9, + + // Length of indexes[]. Multiple of 4 to 16-align the tries. + ULAYOUT_IX_COUNT = 12 +}; + +constexpr int32_t ULAYOUT_MAX_INPC_SHIFT = 24; +constexpr int32_t ULAYOUT_MAX_INSC_SHIFT = 16; +constexpr int32_t ULAYOUT_MAX_VO_SHIFT = 8; + +#endif // __ULAYOUT_PROPS_H__ diff --git a/intl/icu/source/common/ulist.cpp b/intl/icu/source/common/ulist.cpp new file mode 100644 index 0000000000..5bdf534c84 --- /dev/null +++ b/intl/icu/source/common/ulist.cpp @@ -0,0 +1,270 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 2009-2016, International Business Machines +* Corporation and others. All Rights Reserved. +****************************************************************************** +*/ + +#include "ulist.h" +#include "cmemory.h" +#include "cstring.h" +#include "uenumimp.h" + +typedef struct UListNode UListNode; +struct UListNode { + void *data; + + UListNode *next; + UListNode *previous; + + /* When data is created with uprv_malloc, needs to be freed during deleteList function. */ + UBool forceDelete; +}; + +struct UList { + UListNode *curr; + UListNode *head; + UListNode *tail; + + int32_t size; +}; + +static void ulist_addFirstItem(UList *list, UListNode *newItem); + +U_CAPI UList *U_EXPORT2 ulist_createEmptyList(UErrorCode *status) { + UList *newList = nullptr; + + if (U_FAILURE(*status)) { + return nullptr; + } + + newList = (UList *)uprv_malloc(sizeof(UList)); + if (newList == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + + newList->curr = nullptr; + newList->head = nullptr; + newList->tail = nullptr; + newList->size = 0; + + return newList; +} + +/* + * Function called by addItemEndList or addItemBeginList when the first item is added to the list. + * This function properly sets the pointers for the first item added. + */ +static void ulist_addFirstItem(UList *list, UListNode *newItem) { + newItem->next = nullptr; + newItem->previous = nullptr; + list->head = newItem; + list->tail = newItem; +} + +static void ulist_removeItem(UList *list, UListNode *p) { + if (p->previous == nullptr) { + // p is the list head. + list->head = p->next; + } else { + p->previous->next = p->next; + } + if (p->next == nullptr) { + // p is the list tail. + list->tail = p->previous; + } else { + p->next->previous = p->previous; + } + if (p == list->curr) { + list->curr = p->next; + } + --list->size; + if (p->forceDelete) { + uprv_free(p->data); + } + uprv_free(p); +} + +U_CAPI void U_EXPORT2 ulist_addItemEndList(UList *list, const void *data, UBool forceDelete, UErrorCode *status) { + UListNode *newItem = nullptr; + + if (U_FAILURE(*status) || list == nullptr || data == nullptr) { + if (forceDelete) { + uprv_free((void *)data); + } + return; + } + + newItem = (UListNode *)uprv_malloc(sizeof(UListNode)); + if (newItem == nullptr) { + if (forceDelete) { + uprv_free((void *)data); + } + *status = U_MEMORY_ALLOCATION_ERROR; + return; + } + newItem->data = (void *)(data); + newItem->forceDelete = forceDelete; + + if (list->size == 0) { + ulist_addFirstItem(list, newItem); + } else { + newItem->next = nullptr; + newItem->previous = list->tail; + list->tail->next = newItem; + list->tail = newItem; + } + + list->size++; +} + +U_CAPI void U_EXPORT2 ulist_addItemBeginList(UList *list, const void *data, UBool forceDelete, UErrorCode *status) { + UListNode *newItem = nullptr; + + if (U_FAILURE(*status) || list == nullptr || data == nullptr) { + if (forceDelete) { + uprv_free((void *)data); + } + return; + } + + newItem = (UListNode *)uprv_malloc(sizeof(UListNode)); + if (newItem == nullptr) { + if (forceDelete) { + uprv_free((void *)data); + } + *status = U_MEMORY_ALLOCATION_ERROR; + return; + } + newItem->data = (void *)(data); + newItem->forceDelete = forceDelete; + + if (list->size == 0) { + ulist_addFirstItem(list, newItem); + } else { + newItem->previous = nullptr; + newItem->next = list->head; + list->head->previous = newItem; + list->head = newItem; + } + + list->size++; +} + +U_CAPI UBool U_EXPORT2 ulist_containsString(const UList *list, const char *data, int32_t length) { + if (list != nullptr) { + const UListNode *pointer; + for (pointer = list->head; pointer != nullptr; pointer = pointer->next) { + if (length == (int32_t)uprv_strlen((const char *)pointer->data)) { + if (uprv_memcmp(data, pointer->data, length) == 0) { + return true; + } + } + } + } + return false; +} + +U_CAPI UBool U_EXPORT2 ulist_removeString(UList *list, const char *data) { + if (list != nullptr) { + UListNode *pointer; + for (pointer = list->head; pointer != nullptr; pointer = pointer->next) { + if (uprv_strcmp(data, (const char *)pointer->data) == 0) { + ulist_removeItem(list, pointer); + // Remove only the first occurrence, like Java LinkedList.remove(Object). + return true; + } + } + } + return false; +} + +U_CAPI void *U_EXPORT2 ulist_getNext(UList *list) { + UListNode *curr = nullptr; + + if (list == nullptr || list->curr == nullptr) { + return nullptr; + } + + curr = list->curr; + list->curr = curr->next; + + return curr->data; +} + +U_CAPI int32_t U_EXPORT2 ulist_getListSize(const UList *list) { + if (list != nullptr) { + return list->size; + } + + return -1; +} + +U_CAPI void U_EXPORT2 ulist_resetList(UList *list) { + if (list != nullptr) { + list->curr = list->head; + } +} + +U_CAPI void U_EXPORT2 ulist_deleteList(UList *list) { + UListNode *listHead = nullptr; + + if (list != nullptr) { + listHead = list->head; + while (listHead != nullptr) { + UListNode *listPointer = listHead->next; + + if (listHead->forceDelete) { + uprv_free(listHead->data); + } + + uprv_free(listHead); + listHead = listPointer; + } + uprv_free(list); + list = nullptr; + } +} + +U_CAPI void U_EXPORT2 ulist_close_keyword_values_iterator(UEnumeration *en) { + if (en != nullptr) { + ulist_deleteList((UList *)(en->context)); + uprv_free(en); + } +} + +U_CAPI int32_t U_EXPORT2 ulist_count_keyword_values(UEnumeration *en, UErrorCode *status) { + if (U_FAILURE(*status)) { + return -1; + } + + return ulist_getListSize((UList *)(en->context)); +} + +U_CAPI const char * U_EXPORT2 ulist_next_keyword_value(UEnumeration *en, int32_t *resultLength, UErrorCode *status) { + const char *s; + if (U_FAILURE(*status)) { + return nullptr; + } + + s = (const char *)ulist_getNext((UList *)(en->context)); + if (s != nullptr && resultLength != nullptr) { + *resultLength = static_cast(uprv_strlen(s)); + } + return s; +} + +U_CAPI void U_EXPORT2 ulist_reset_keyword_values_iterator(UEnumeration *en, UErrorCode *status) { + if (U_FAILURE(*status)) { + return ; + } + + ulist_resetList((UList *)(en->context)); +} + +U_CAPI UList * U_EXPORT2 ulist_getListFromEnum(UEnumeration *en) { + return (UList *)(en->context); +} diff --git a/intl/icu/source/common/ulist.h b/intl/icu/source/common/ulist.h new file mode 100644 index 0000000000..de58a4ad02 --- /dev/null +++ b/intl/icu/source/common/ulist.h @@ -0,0 +1,50 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 2009-2016, International Business Machines +* Corporation and others. All Rights Reserved. +****************************************************************************** +*/ + +#ifndef ULIST_H +#define ULIST_H + +#include "unicode/utypes.h" +#include "unicode/uenum.h" + +struct UList; +typedef struct UList UList; + +U_CAPI UList * U_EXPORT2 ulist_createEmptyList(UErrorCode *status); + +U_CAPI void U_EXPORT2 ulist_addItemEndList(UList *list, const void *data, UBool forceDelete, UErrorCode *status); + +U_CAPI void U_EXPORT2 ulist_addItemBeginList(UList *list, const void *data, UBool forceDelete, UErrorCode *status); + +U_CAPI UBool U_EXPORT2 ulist_containsString(const UList *list, const char *data, int32_t length); + +U_CAPI UBool U_EXPORT2 ulist_removeString(UList *list, const char *data); + +U_CAPI void *U_EXPORT2 ulist_getNext(UList *list); + +U_CAPI int32_t U_EXPORT2 ulist_getListSize(const UList *list); + +U_CAPI void U_EXPORT2 ulist_resetList(UList *list); + +U_CAPI void U_EXPORT2 ulist_deleteList(UList *list); + +/* + * The following are for use when creating UEnumeration object backed by UList. + */ +U_CAPI void U_EXPORT2 ulist_close_keyword_values_iterator(UEnumeration *en); + +U_CAPI int32_t U_EXPORT2 ulist_count_keyword_values(UEnumeration *en, UErrorCode *status); + +U_CAPI const char * U_EXPORT2 ulist_next_keyword_value(UEnumeration* en, int32_t *resultLength, UErrorCode* status); + +U_CAPI void U_EXPORT2 ulist_reset_keyword_values_iterator(UEnumeration* en, UErrorCode* status); + +U_CAPI UList * U_EXPORT2 ulist_getListFromEnum(UEnumeration *en); + +#endif diff --git a/intl/icu/source/common/uloc.cpp b/intl/icu/source/common/uloc.cpp new file mode 100644 index 0000000000..395df1466e --- /dev/null +++ b/intl/icu/source/common/uloc.cpp @@ -0,0 +1,2228 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1997-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* +* File ULOC.CPP +* +* Modification History: +* +* Date Name Description +* 04/01/97 aliu Creation. +* 08/21/98 stephen JDK 1.2 sync +* 12/08/98 rtg New Locale implementation and C API +* 03/15/99 damiba overhaul. +* 04/06/99 stephen changed setDefault() to realloc and copy +* 06/14/99 stephen Changed calls to ures_open for new params +* 07/21/99 stephen Modified setDefault() to propagate to C++ +* 05/14/04 alan 7 years later: refactored, cleaned up, fixed bugs, +* brought canonicalization code into line with spec +*****************************************************************************/ + +/* + POSIX's locale format, from putil.c: [no spaces] + + ll [ _CC ] [ . MM ] [ @ VV] + + l = lang, C = ctry, M = charmap, V = variant +*/ + +#include "unicode/bytestream.h" +#include "unicode/errorcode.h" +#include "unicode/stringpiece.h" +#include "unicode/utypes.h" +#include "unicode/ustring.h" +#include "unicode/uloc.h" + +#include "bytesinkutil.h" +#include "putilimp.h" +#include "ustr_imp.h" +#include "ulocimp.h" +#include "umutex.h" +#include "cstring.h" +#include "cmemory.h" +#include "locmap.h" +#include "uarrsort.h" +#include "uenumimp.h" +#include "uassert.h" +#include "charstr.h" + +U_NAMESPACE_USE + +/* ### Declarations **************************************************/ + +/* Locale stuff from locid.cpp */ +U_CFUNC void locale_set_default(const char *id); +U_CFUNC const char *locale_get_default(); + +/* ### Data tables **************************************************/ + +/** + * Table of language codes, both 2- and 3-letter, with preference + * given to 2-letter codes where possible. Includes 3-letter codes + * that lack a 2-letter equivalent. + * + * This list must be in sorted order. This list is returned directly + * to the user by some API. + * + * This list must be kept in sync with LANGUAGES_3, with corresponding + * entries matched. + * + * This table should be terminated with a nullptr entry, followed by a + * second list, and another nullptr entry. The first list is visible to + * user code when this array is returned by API. The second list + * contains codes we support, but do not expose through user API. + * + * Notes + * + * Tables updated per http://lcweb.loc.gov/standards/iso639-2/ to + * include the revisions up to 2001/7/27 *CWB* + * + * The 3 character codes are the terminology codes like RFC 3066. This + * is compatible with prior ICU codes + * + * "in" "iw" "ji" "jw" & "sh" have been withdrawn but are still in the + * table but now at the end of the table because 3 character codes are + * duplicates. This avoids bad searches going from 3 to 2 character + * codes. + * + * The range qaa-qtz is reserved for local use + */ +/* Generated using org.unicode.cldr.icu.GenerateISO639LanguageTables */ +/* ISO639 table version is 20150505 */ +/* Subsequent hand addition of selected languages */ +static const char * const LANGUAGES[] = { + "aa", "ab", "ace", "ach", "ada", "ady", "ae", "aeb", + "af", "afh", "agq", "ain", "ak", "akk", "akz", "ale", + "aln", "alt", "am", "an", "ang", "anp", "ar", "arc", + "arn", "aro", "arp", "arq", "ars", "arw", "ary", "arz", "as", + "asa", "ase", "ast", "av", "avk", "awa", "ay", "az", + "ba", "bal", "ban", "bar", "bas", "bax", "bbc", "bbj", + "be", "bej", "bem", "bew", "bez", "bfd", "bfq", "bg", + "bgc", "bgn", "bho", "bi", "bik", "bin", "bjn", "bkm", "bla", + "bm", "bn", "bo", "bpy", "bqi", "br", "bra", "brh", + "brx", "bs", "bss", "bua", "bug", "bum", "byn", "byv", + "ca", "cad", "car", "cay", "cch", "ccp", "ce", "ceb", "cgg", + "ch", "chb", "chg", "chk", "chm", "chn", "cho", "chp", + "chr", "chy", "ckb", "co", "cop", "cps", "cr", "crh", + "cs", "csb", "cu", "cv", "cy", + "da", "dak", "dar", "dav", "de", "del", "den", "dgr", + "din", "dje", "doi", "dsb", "dtp", "dua", "dum", "dv", + "dyo", "dyu", "dz", "dzg", + "ebu", "ee", "efi", "egl", "egy", "eka", "el", "elx", + "en", "enm", "eo", "es", "esu", "et", "eu", "ewo", + "ext", + "fa", "fan", "fat", "ff", "fi", "fil", "fit", "fj", + "fo", "fon", "fr", "frc", "frm", "fro", "frp", "frr", + "frs", "fur", "fy", + "ga", "gaa", "gag", "gan", "gay", "gba", "gbz", "gd", + "gez", "gil", "gl", "glk", "gmh", "gn", "goh", "gom", + "gon", "gor", "got", "grb", "grc", "gsw", "gu", "guc", + "gur", "guz", "gv", "gwi", + "ha", "hai", "hak", "haw", "he", "hi", "hif", "hil", + "hit", "hmn", "ho", "hr", "hsb", "hsn", "ht", "hu", + "hup", "hy", "hz", + "ia", "iba", "ibb", "id", "ie", "ig", "ii", "ik", + "ilo", "inh", "io", "is", "it", "iu", "izh", + "ja", "jam", "jbo", "jgo", "jmc", "jpr", "jrb", "jut", + "jv", + "ka", "kaa", "kab", "kac", "kaj", "kam", "kaw", "kbd", + "kbl", "kcg", "kde", "kea", "ken", "kfo", "kg", "kgp", + "kha", "kho", "khq", "khw", "ki", "kiu", "kj", "kk", + "kkj", "kl", "kln", "km", "kmb", "kn", "ko", "koi", + "kok", "kos", "kpe", "kr", "krc", "kri", "krj", "krl", + "kru", "ks", "ksb", "ksf", "ksh", "ku", "kum", "kut", + "kv", "kw", "ky", + "la", "lad", "lag", "lah", "lam", "lb", "lez", "lfn", + "lg", "li", "lij", "liv", "lkt", "lmo", "ln", "lo", + "lol", "loz", "lrc", "lt", "ltg", "lu", "lua", "lui", + "lun", "luo", "lus", "luy", "lv", "lzh", "lzz", + "mad", "maf", "mag", "mai", "mak", "man", "mas", "mde", + "mdf", "mdh", "mdr", "men", "mer", "mfe", "mg", "mga", + "mgh", "mgo", "mh", "mi", "mic", "min", "mis", "mk", + "ml", "mn", "mnc", "mni", + "moh", "mos", "mr", "mrj", + "ms", "mt", "mua", "mul", "mus", "mwl", "mwr", "mwv", + "my", "mye", "myv", "mzn", + "na", "nan", "nap", "naq", "nb", "nd", "nds", "ne", + "new", "ng", "nia", "niu", "njo", "nl", "nmg", "nn", + "nnh", "no", "nog", "non", "nov", "nqo", "nr", "nso", + "nus", "nv", "nwc", "ny", "nym", "nyn", "nyo", "nzi", + "oc", "oj", "om", "or", "os", "osa", "ota", + "pa", "pag", "pal", "pam", "pap", "pau", "pcd", "pcm", "pdc", + "pdt", "peo", "pfl", "phn", "pi", "pl", "pms", "pnt", + "pon", "prg", "pro", "ps", "pt", + "qu", "quc", "qug", + "raj", "rap", "rar", "rgn", "rif", "rm", "rn", "ro", + "rof", "rom", "rtm", "ru", "rue", "rug", "rup", + "rw", "rwk", + "sa", "sad", "sah", "sam", "saq", "sas", "sat", "saz", + "sba", "sbp", "sc", "scn", "sco", "sd", "sdc", "sdh", + "se", "see", "seh", "sei", "sel", "ses", "sg", "sga", + "sgs", "shi", "shn", "shu", "si", "sid", "sk", + "sl", "sli", "sly", "sm", "sma", "smj", "smn", "sms", + "sn", "snk", "so", "sog", "sq", "sr", "srn", "srr", + "ss", "ssy", "st", "stq", "su", "suk", "sus", "sux", + "sv", "sw", "swb", "syc", "syr", "szl", + "ta", "tcy", "te", "tem", "teo", "ter", "tet", "tg", + "th", "ti", "tig", "tiv", "tk", "tkl", "tkr", + "tlh", "tli", "tly", "tmh", "tn", "to", "tog", "tpi", + "tr", "tru", "trv", "ts", "tsd", "tsi", "tt", "ttt", + "tum", "tvl", "tw", "twq", "ty", "tyv", "tzm", + "udm", "ug", "uga", "uk", "umb", "und", "ur", "uz", + "vai", "ve", "vec", "vep", "vi", "vls", "vmf", "vo", + "vot", "vro", "vun", + "wa", "wae", "wal", "war", "was", "wbp", "wo", "wuu", + "xal", "xh", "xmf", "xog", + "yao", "yap", "yav", "ybb", "yi", "yo", "yrl", "yue", + "za", "zap", "zbl", "zea", "zen", "zgh", "zh", "zu", + "zun", "zxx", "zza", +nullptr, + "in", "iw", "ji", "jw", "mo", "sh", "swc", "tl", /* obsolete language codes */ +nullptr +}; + +static const char* const DEPRECATED_LANGUAGES[]={ + "in", "iw", "ji", "jw", "mo", nullptr, nullptr +}; +static const char* const REPLACEMENT_LANGUAGES[]={ + "id", "he", "yi", "jv", "ro", nullptr, nullptr +}; + +/** + * Table of 3-letter language codes. + * + * This is a lookup table used to convert 3-letter language codes to + * their 2-letter equivalent, where possible. It must be kept in sync + * with LANGUAGES. For all valid i, LANGUAGES[i] must refer to the + * same language as LANGUAGES_3[i]. The commented-out lines are + * copied from LANGUAGES to make eyeballing this baby easier. + * + * Where a 3-letter language code has no 2-letter equivalent, the + * 3-letter code occupies both LANGUAGES[i] and LANGUAGES_3[i]. + * + * This table should be terminated with a nullptr entry, followed by a + * second list, and another nullptr entry. The two lists correspond to + * the two lists in LANGUAGES. + */ +/* Generated using org.unicode.cldr.icu.GenerateISO639LanguageTables */ +/* ISO639 table version is 20150505 */ +/* Subsequent hand addition of selected languages */ +static const char * const LANGUAGES_3[] = { + "aar", "abk", "ace", "ach", "ada", "ady", "ave", "aeb", + "afr", "afh", "agq", "ain", "aka", "akk", "akz", "ale", + "aln", "alt", "amh", "arg", "ang", "anp", "ara", "arc", + "arn", "aro", "arp", "arq", "ars", "arw", "ary", "arz", "asm", + "asa", "ase", "ast", "ava", "avk", "awa", "aym", "aze", + "bak", "bal", "ban", "bar", "bas", "bax", "bbc", "bbj", + "bel", "bej", "bem", "bew", "bez", "bfd", "bfq", "bul", + "bgc", "bgn", "bho", "bis", "bik", "bin", "bjn", "bkm", "bla", + "bam", "ben", "bod", "bpy", "bqi", "bre", "bra", "brh", + "brx", "bos", "bss", "bua", "bug", "bum", "byn", "byv", + "cat", "cad", "car", "cay", "cch", "ccp", "che", "ceb", "cgg", + "cha", "chb", "chg", "chk", "chm", "chn", "cho", "chp", + "chr", "chy", "ckb", "cos", "cop", "cps", "cre", "crh", + "ces", "csb", "chu", "chv", "cym", + "dan", "dak", "dar", "dav", "deu", "del", "den", "dgr", + "din", "dje", "doi", "dsb", "dtp", "dua", "dum", "div", + "dyo", "dyu", "dzo", "dzg", + "ebu", "ewe", "efi", "egl", "egy", "eka", "ell", "elx", + "eng", "enm", "epo", "spa", "esu", "est", "eus", "ewo", + "ext", + "fas", "fan", "fat", "ful", "fin", "fil", "fit", "fij", + "fao", "fon", "fra", "frc", "frm", "fro", "frp", "frr", + "frs", "fur", "fry", + "gle", "gaa", "gag", "gan", "gay", "gba", "gbz", "gla", + "gez", "gil", "glg", "glk", "gmh", "grn", "goh", "gom", + "gon", "gor", "got", "grb", "grc", "gsw", "guj", "guc", + "gur", "guz", "glv", "gwi", + "hau", "hai", "hak", "haw", "heb", "hin", "hif", "hil", + "hit", "hmn", "hmo", "hrv", "hsb", "hsn", "hat", "hun", + "hup", "hye", "her", + "ina", "iba", "ibb", "ind", "ile", "ibo", "iii", "ipk", + "ilo", "inh", "ido", "isl", "ita", "iku", "izh", + "jpn", "jam", "jbo", "jgo", "jmc", "jpr", "jrb", "jut", + "jav", + "kat", "kaa", "kab", "kac", "kaj", "kam", "kaw", "kbd", + "kbl", "kcg", "kde", "kea", "ken", "kfo", "kon", "kgp", + "kha", "kho", "khq", "khw", "kik", "kiu", "kua", "kaz", + "kkj", "kal", "kln", "khm", "kmb", "kan", "kor", "koi", + "kok", "kos", "kpe", "kau", "krc", "kri", "krj", "krl", + "kru", "kas", "ksb", "ksf", "ksh", "kur", "kum", "kut", + "kom", "cor", "kir", + "lat", "lad", "lag", "lah", "lam", "ltz", "lez", "lfn", + "lug", "lim", "lij", "liv", "lkt", "lmo", "lin", "lao", + "lol", "loz", "lrc", "lit", "ltg", "lub", "lua", "lui", + "lun", "luo", "lus", "luy", "lav", "lzh", "lzz", + "mad", "maf", "mag", "mai", "mak", "man", "mas", "mde", + "mdf", "mdh", "mdr", "men", "mer", "mfe", "mlg", "mga", + "mgh", "mgo", "mah", "mri", "mic", "min", "mis", "mkd", + "mal", "mon", "mnc", "mni", + "moh", "mos", "mar", "mrj", + "msa", "mlt", "mua", "mul", "mus", "mwl", "mwr", "mwv", + "mya", "mye", "myv", "mzn", + "nau", "nan", "nap", "naq", "nob", "nde", "nds", "nep", + "new", "ndo", "nia", "niu", "njo", "nld", "nmg", "nno", + "nnh", "nor", "nog", "non", "nov", "nqo", "nbl", "nso", + "nus", "nav", "nwc", "nya", "nym", "nyn", "nyo", "nzi", + "oci", "oji", "orm", "ori", "oss", "osa", "ota", + "pan", "pag", "pal", "pam", "pap", "pau", "pcd", "pcm", "pdc", + "pdt", "peo", "pfl", "phn", "pli", "pol", "pms", "pnt", + "pon", "prg", "pro", "pus", "por", + "que", "quc", "qug", + "raj", "rap", "rar", "rgn", "rif", "roh", "run", "ron", + "rof", "rom", "rtm", "rus", "rue", "rug", "rup", + "kin", "rwk", + "san", "sad", "sah", "sam", "saq", "sas", "sat", "saz", + "sba", "sbp", "srd", "scn", "sco", "snd", "sdc", "sdh", + "sme", "see", "seh", "sei", "sel", "ses", "sag", "sga", + "sgs", "shi", "shn", "shu", "sin", "sid", "slk", + "slv", "sli", "sly", "smo", "sma", "smj", "smn", "sms", + "sna", "snk", "som", "sog", "sqi", "srp", "srn", "srr", + "ssw", "ssy", "sot", "stq", "sun", "suk", "sus", "sux", + "swe", "swa", "swb", "syc", "syr", "szl", + "tam", "tcy", "tel", "tem", "teo", "ter", "tet", "tgk", + "tha", "tir", "tig", "tiv", "tuk", "tkl", "tkr", + "tlh", "tli", "tly", "tmh", "tsn", "ton", "tog", "tpi", + "tur", "tru", "trv", "tso", "tsd", "tsi", "tat", "ttt", + "tum", "tvl", "twi", "twq", "tah", "tyv", "tzm", + "udm", "uig", "uga", "ukr", "umb", "und", "urd", "uzb", + "vai", "ven", "vec", "vep", "vie", "vls", "vmf", "vol", + "vot", "vro", "vun", + "wln", "wae", "wal", "war", "was", "wbp", "wol", "wuu", + "xal", "xho", "xmf", "xog", + "yao", "yap", "yav", "ybb", "yid", "yor", "yrl", "yue", + "zha", "zap", "zbl", "zea", "zen", "zgh", "zho", "zul", + "zun", "zxx", "zza", +nullptr, +/* "in", "iw", "ji", "jw", "mo", "sh", "swc", "tl", */ + "ind", "heb", "yid", "jaw", "mol", "srp", "swc", "tgl", +nullptr +}; + +/** + * Table of 2-letter country codes. + * + * This list must be in sorted order. This list is returned directly + * to the user by some API. + * + * This list must be kept in sync with COUNTRIES_3, with corresponding + * entries matched. + * + * This table should be terminated with a nullptr entry, followed by a + * second list, and another nullptr entry. The first list is visible to + * user code when this array is returned by API. The second list + * contains codes we support, but do not expose through user API. + * + * Notes: + * + * ZR(ZAR) is now CD(COD) and FX(FXX) is PS(PSE) as per + * http://www.evertype.com/standards/iso3166/iso3166-1-en.html added + * new codes keeping the old ones for compatibility updated to include + * 1999/12/03 revisions *CWB* + * + * RO(ROM) is now RO(ROU) according to + * http://www.iso.org/iso/en/prods-services/iso3166ma/03updates-on-iso-3166/nlv3e-rou.html + */ +static const char * const COUNTRIES[] = { + "AD", "AE", "AF", "AG", "AI", "AL", "AM", + "AO", "AQ", "AR", "AS", "AT", "AU", "AW", "AX", "AZ", + "BA", "BB", "BD", "BE", "BF", "BG", "BH", "BI", + "BJ", "BL", "BM", "BN", "BO", "BQ", "BR", "BS", "BT", "BV", + "BW", "BY", "BZ", "CA", "CC", "CD", "CF", "CG", + "CH", "CI", "CK", "CL", "CM", "CN", "CO", "CQ", "CR", + "CU", "CV", "CW", "CX", "CY", "CZ", "DE", "DG", "DJ", "DK", + "DM", "DO", "DZ", "EA", "EC", "EE", "EG", "EH", "ER", + "ES", "ET", "FI", "FJ", "FK", "FM", "FO", "FR", + "GA", "GB", "GD", "GE", "GF", "GG", "GH", "GI", "GL", + "GM", "GN", "GP", "GQ", "GR", "GS", "GT", "GU", + "GW", "GY", "HK", "HM", "HN", "HR", "HT", "HU", + "IC", "ID", "IE", "IL", "IM", "IN", "IO", "IQ", "IR", "IS", + "IT", "JE", "JM", "JO", "JP", "KE", "KG", "KH", "KI", + "KM", "KN", "KP", "KR", "KW", "KY", "KZ", "LA", + "LB", "LC", "LI", "LK", "LR", "LS", "LT", "LU", + "LV", "LY", "MA", "MC", "MD", "ME", "MF", "MG", "MH", "MK", + "ML", "MM", "MN", "MO", "MP", "MQ", "MR", "MS", + "MT", "MU", "MV", "MW", "MX", "MY", "MZ", "NA", + "NC", "NE", "NF", "NG", "NI", "NL", "NO", "NP", + "NR", "NU", "NZ", "OM", "PA", "PE", "PF", "PG", + "PH", "PK", "PL", "PM", "PN", "PR", "PS", "PT", + "PW", "PY", "QA", "RE", "RO", "RS", "RU", "RW", "SA", + "SB", "SC", "SD", "SE", "SG", "SH", "SI", "SJ", + "SK", "SL", "SM", "SN", "SO", "SR", "SS", "ST", "SV", + "SX", "SY", "SZ", "TC", "TD", "TF", "TG", "TH", "TJ", + "TK", "TL", "TM", "TN", "TO", "TR", "TT", "TV", + "TW", "TZ", "UA", "UG", "UM", "US", "UY", "UZ", + "VA", "VC", "VE", "VG", "VI", "VN", "VU", "WF", + "WS", "XK", "YE", "YT", "ZA", "ZM", "ZW", +nullptr, + "AN", "BU", "CS", "FX", "RO", "SU", "TP", "YD", "YU", "ZR", /* obsolete country codes */ +nullptr +}; + +static const char* const DEPRECATED_COUNTRIES[] = { + "AN", "BU", "CS", "DD", "DY", "FX", "HV", "NH", "RH", "SU", "TP", "UK", "VD", "YD", "YU", "ZR", nullptr, nullptr /* deprecated country list */ +}; +static const char* const REPLACEMENT_COUNTRIES[] = { +/* "AN", "BU", "CS", "DD", "DY", "FX", "HV", "NH", "RH", "SU", "TP", "UK", "VD", "YD", "YU", "ZR" */ + "CW", "MM", "RS", "DE", "BJ", "FR", "BF", "VU", "ZW", "RU", "TL", "GB", "VN", "YE", "RS", "CD", nullptr, nullptr /* replacement country codes */ +}; + +/** + * Table of 3-letter country codes. + * + * This is a lookup table used to convert 3-letter country codes to + * their 2-letter equivalent. It must be kept in sync with COUNTRIES. + * For all valid i, COUNTRIES[i] must refer to the same country as + * COUNTRIES_3[i]. The commented-out lines are copied from COUNTRIES + * to make eyeballing this baby easier. + * + * This table should be terminated with a nullptr entry, followed by a + * second list, and another nullptr entry. The two lists correspond to + * the two lists in COUNTRIES. + */ +static const char * const COUNTRIES_3[] = { +/* "AD", "AE", "AF", "AG", "AI", "AL", "AM", */ + "AND", "ARE", "AFG", "ATG", "AIA", "ALB", "ARM", +/* "AO", "AQ", "AR", "AS", "AT", "AU", "AW", "AX", "AZ", */ + "AGO", "ATA", "ARG", "ASM", "AUT", "AUS", "ABW", "ALA", "AZE", +/* "BA", "BB", "BD", "BE", "BF", "BG", "BH", "BI", */ + "BIH", "BRB", "BGD", "BEL", "BFA", "BGR", "BHR", "BDI", +/* "BJ", "BL", "BM", "BN", "BO", "BQ", "BR", "BS", "BT", "BV", */ + "BEN", "BLM", "BMU", "BRN", "BOL", "BES", "BRA", "BHS", "BTN", "BVT", +/* "BW", "BY", "BZ", "CA", "CC", "CD", "CF", "CG", */ + "BWA", "BLR", "BLZ", "CAN", "CCK", "COD", "CAF", "COG", +/* "CH", "CI", "CK", "CL", "CM", "CN", "CO", "CQ", "CR", */ + "CHE", "CIV", "COK", "CHL", "CMR", "CHN", "COL", "CRQ", "CRI", +/* "CU", "CV", "CW", "CX", "CY", "CZ", "DE", "DG", "DJ", "DK", */ + "CUB", "CPV", "CUW", "CXR", "CYP", "CZE", "DEU", "DGA", "DJI", "DNK", +/* "DM", "DO", "DZ", "EA", "EC", "EE", "EG", "EH", "ER", */ + "DMA", "DOM", "DZA", "XEA", "ECU", "EST", "EGY", "ESH", "ERI", +/* "ES", "ET", "FI", "FJ", "FK", "FM", "FO", "FR", */ + "ESP", "ETH", "FIN", "FJI", "FLK", "FSM", "FRO", "FRA", +/* "GA", "GB", "GD", "GE", "GF", "GG", "GH", "GI", "GL", */ + "GAB", "GBR", "GRD", "GEO", "GUF", "GGY", "GHA", "GIB", "GRL", +/* "GM", "GN", "GP", "GQ", "GR", "GS", "GT", "GU", */ + "GMB", "GIN", "GLP", "GNQ", "GRC", "SGS", "GTM", "GUM", +/* "GW", "GY", "HK", "HM", "HN", "HR", "HT", "HU", */ + "GNB", "GUY", "HKG", "HMD", "HND", "HRV", "HTI", "HUN", +/* "IC", "ID", "IE", "IL", "IM", "IN", "IO", "IQ", "IR", "IS" */ + "XIC", "IDN", "IRL", "ISR", "IMN", "IND", "IOT", "IRQ", "IRN", "ISL", +/* "IT", "JE", "JM", "JO", "JP", "KE", "KG", "KH", "KI", */ + "ITA", "JEY", "JAM", "JOR", "JPN", "KEN", "KGZ", "KHM", "KIR", +/* "KM", "KN", "KP", "KR", "KW", "KY", "KZ", "LA", */ + "COM", "KNA", "PRK", "KOR", "KWT", "CYM", "KAZ", "LAO", +/* "LB", "LC", "LI", "LK", "LR", "LS", "LT", "LU", */ + "LBN", "LCA", "LIE", "LKA", "LBR", "LSO", "LTU", "LUX", +/* "LV", "LY", "MA", "MC", "MD", "ME", "MF", "MG", "MH", "MK", */ + "LVA", "LBY", "MAR", "MCO", "MDA", "MNE", "MAF", "MDG", "MHL", "MKD", +/* "ML", "MM", "MN", "MO", "MP", "MQ", "MR", "MS", */ + "MLI", "MMR", "MNG", "MAC", "MNP", "MTQ", "MRT", "MSR", +/* "MT", "MU", "MV", "MW", "MX", "MY", "MZ", "NA", */ + "MLT", "MUS", "MDV", "MWI", "MEX", "MYS", "MOZ", "NAM", +/* "NC", "NE", "NF", "NG", "NI", "NL", "NO", "NP", */ + "NCL", "NER", "NFK", "NGA", "NIC", "NLD", "NOR", "NPL", +/* "NR", "NU", "NZ", "OM", "PA", "PE", "PF", "PG", */ + "NRU", "NIU", "NZL", "OMN", "PAN", "PER", "PYF", "PNG", +/* "PH", "PK", "PL", "PM", "PN", "PR", "PS", "PT", */ + "PHL", "PAK", "POL", "SPM", "PCN", "PRI", "PSE", "PRT", +/* "PW", "PY", "QA", "RE", "RO", "RS", "RU", "RW", "SA", */ + "PLW", "PRY", "QAT", "REU", "ROU", "SRB", "RUS", "RWA", "SAU", +/* "SB", "SC", "SD", "SE", "SG", "SH", "SI", "SJ", */ + "SLB", "SYC", "SDN", "SWE", "SGP", "SHN", "SVN", "SJM", +/* "SK", "SL", "SM", "SN", "SO", "SR", "SS", "ST", "SV", */ + "SVK", "SLE", "SMR", "SEN", "SOM", "SUR", "SSD", "STP", "SLV", +/* "SX", "SY", "SZ", "TC", "TD", "TF", "TG", "TH", "TJ", */ + "SXM", "SYR", "SWZ", "TCA", "TCD", "ATF", "TGO", "THA", "TJK", +/* "TK", "TL", "TM", "TN", "TO", "TR", "TT", "TV", */ + "TKL", "TLS", "TKM", "TUN", "TON", "TUR", "TTO", "TUV", +/* "TW", "TZ", "UA", "UG", "UM", "US", "UY", "UZ", */ + "TWN", "TZA", "UKR", "UGA", "UMI", "USA", "URY", "UZB", +/* "VA", "VC", "VE", "VG", "VI", "VN", "VU", "WF", */ + "VAT", "VCT", "VEN", "VGB", "VIR", "VNM", "VUT", "WLF", +/* "WS", "XK", "YE", "YT", "ZA", "ZM", "ZW", */ + "WSM", "XKK", "YEM", "MYT", "ZAF", "ZMB", "ZWE", +nullptr, +/* "AN", "BU", "CS", "FX", "RO", "SU", "TP", "YD", "YU", "ZR" */ + "ANT", "BUR", "SCG", "FXX", "ROM", "SUN", "TMP", "YMD", "YUG", "ZAR", +nullptr +}; + +typedef struct CanonicalizationMap { + const char *id; /* input ID */ + const char *canonicalID; /* canonicalized output ID */ +} CanonicalizationMap; + +/** + * A map to canonicalize locale IDs. This handles a variety of + * different semantic kinds of transformations. + */ +static const CanonicalizationMap CANONICALIZE_MAP[] = { + { "art__LOJBAN", "jbo" }, /* registered name */ + { "hy__AREVELA", "hy" }, /* Registered IANA variant */ + { "hy__AREVMDA", "hyw" }, /* Registered IANA variant */ + { "zh__GUOYU", "zh" }, /* registered name */ + { "zh__HAKKA", "hak" }, /* registered name */ + { "zh__XIANG", "hsn" }, /* registered name */ + // subtags with 3 chars won't be treated as variants. + { "zh_GAN", "gan" }, /* registered name */ + { "zh_MIN_NAN", "nan" }, /* registered name */ + { "zh_WUU", "wuu" }, /* registered name */ + { "zh_YUE", "yue" }, /* registered name */ +}; + +/* ### BCP47 Conversion *******************************************/ +/* Test if the locale id has BCP47 u extension and does not have '@' */ +#define _hasBCP47Extension(id) (id && uprv_strstr(id, "@") == nullptr && getShortestSubtagLength(localeID) == 1) +/* Converts the BCP47 id to Unicode id. Does nothing to id if conversion fails */ +static const char* _ConvertBCP47( + const char* id, char* buffer, int32_t length, + UErrorCode* err, int32_t* pLocaleIdSize) { + const char* finalID; + int32_t localeIDSize = uloc_forLanguageTag(id, buffer, length, nullptr, err); + if (localeIDSize <= 0 || U_FAILURE(*err) || *err == U_STRING_NOT_TERMINATED_WARNING) { + finalID=id; + if (*err == U_STRING_NOT_TERMINATED_WARNING) { + *err = U_BUFFER_OVERFLOW_ERROR; + } + } else { + finalID=buffer; + } + if (pLocaleIdSize != nullptr) { + *pLocaleIdSize = localeIDSize; + } + return finalID; +} +/* Gets the size of the shortest subtag in the given localeID. */ +static int32_t getShortestSubtagLength(const char *localeID) { + int32_t localeIDLength = static_cast(uprv_strlen(localeID)); + int32_t length = localeIDLength; + int32_t tmpLength = 0; + int32_t i; + UBool reset = true; + + for (i = 0; i < localeIDLength; i++) { + if (localeID[i] != '_' && localeID[i] != '-') { + if (reset) { + tmpLength = 0; + reset = false; + } + tmpLength++; + } else { + if (tmpLength != 0 && tmpLength < length) { + length = tmpLength; + } + reset = true; + } + } + + return length; +} + +/* ### Keywords **************************************************/ +#define UPRV_ISDIGIT(c) (((c) >= '0') && ((c) <= '9')) +#define UPRV_ISALPHANUM(c) (uprv_isASCIILetter(c) || UPRV_ISDIGIT(c) ) +/* Punctuation/symbols allowed in legacy key values */ +#define UPRV_OK_VALUE_PUNCTUATION(c) ((c) == '_' || (c) == '-' || (c) == '+' || (c) == '/') + +#define ULOC_KEYWORD_BUFFER_LEN 25 +#define ULOC_MAX_NO_KEYWORDS 25 + +U_CAPI const char * U_EXPORT2 +locale_getKeywordsStart(const char *localeID) { + const char *result = nullptr; + if((result = uprv_strchr(localeID, '@')) != nullptr) { + return result; + } +#if (U_CHARSET_FAMILY == U_EBCDIC_FAMILY) + else { + /* We do this because the @ sign is variant, and the @ sign used on one + EBCDIC machine won't be compiled the same way on other EBCDIC based + machines. */ + static const uint8_t ebcdicSigns[] = { 0x7C, 0x44, 0x66, 0x80, 0xAC, 0xAE, 0xAF, 0xB5, 0xEC, 0xEF, 0x00 }; + const uint8_t *charToFind = ebcdicSigns; + while(*charToFind) { + if((result = uprv_strchr(localeID, *charToFind)) != nullptr) { + return result; + } + charToFind++; + } + } +#endif + return nullptr; +} + +/** + * @param buf buffer of size [ULOC_KEYWORD_BUFFER_LEN] + * @param keywordName incoming name to be canonicalized + * @param status return status (keyword too long) + * @return length of the keyword name + */ +static int32_t locale_canonKeywordName(char *buf, const char *keywordName, UErrorCode *status) +{ + int32_t keywordNameLen = 0; + + for (; *keywordName != 0; keywordName++) { + if (!UPRV_ISALPHANUM(*keywordName)) { + *status = U_ILLEGAL_ARGUMENT_ERROR; /* malformed keyword name */ + return 0; + } + if (keywordNameLen < ULOC_KEYWORD_BUFFER_LEN - 1) { + buf[keywordNameLen++] = uprv_tolower(*keywordName); + } else { + /* keyword name too long for internal buffer */ + *status = U_INTERNAL_PROGRAM_ERROR; + return 0; + } + } + if (keywordNameLen == 0) { + *status = U_ILLEGAL_ARGUMENT_ERROR; /* empty keyword name */ + return 0; + } + buf[keywordNameLen] = 0; /* terminate */ + + return keywordNameLen; +} + +typedef struct { + char keyword[ULOC_KEYWORD_BUFFER_LEN]; + int32_t keywordLen; + const char *valueStart; + int32_t valueLen; +} KeywordStruct; + +static int32_t U_CALLCONV +compareKeywordStructs(const void * /*context*/, const void *left, const void *right) { + const char* leftString = ((const KeywordStruct *)left)->keyword; + const char* rightString = ((const KeywordStruct *)right)->keyword; + return uprv_strcmp(leftString, rightString); +} + +U_CFUNC void +ulocimp_getKeywords(const char *localeID, + char prev, + ByteSink& sink, + UBool valuesToo, + UErrorCode *status) +{ + KeywordStruct keywordList[ULOC_MAX_NO_KEYWORDS]; + + int32_t maxKeywords = ULOC_MAX_NO_KEYWORDS; + int32_t numKeywords = 0; + const char* pos = localeID; + const char* equalSign = nullptr; + const char* semicolon = nullptr; + int32_t i = 0, j, n; + + if(prev == '@') { /* start of keyword definition */ + /* we will grab pairs, trim spaces, lowercase keywords, sort and return */ + do { + UBool duplicate = false; + /* skip leading spaces */ + while(*pos == ' ') { + pos++; + } + if (!*pos) { /* handle trailing "; " */ + break; + } + if(numKeywords == maxKeywords) { + *status = U_INTERNAL_PROGRAM_ERROR; + return; + } + equalSign = uprv_strchr(pos, '='); + semicolon = uprv_strchr(pos, ';'); + /* lack of '=' [foo@currency] is illegal */ + /* ';' before '=' [foo@currency;collation=pinyin] is illegal */ + if(!equalSign || (semicolon && semicolon= ULOC_KEYWORD_BUFFER_LEN) { + /* keyword name too long for internal buffer */ + *status = U_INTERNAL_PROGRAM_ERROR; + return; + } + for(i = 0, n = 0; i < equalSign - pos; ++i) { + if (pos[i] != ' ') { + keywordList[numKeywords].keyword[n++] = uprv_tolower(pos[i]); + } + } + + /* zero-length keyword is an error. */ + if (n == 0) { + *status = U_INVALID_FORMAT_ERROR; + return; + } + + keywordList[numKeywords].keyword[n] = 0; + keywordList[numKeywords].keywordLen = n; + /* now grab the value part. First we skip the '=' */ + equalSign++; + /* then we leading spaces */ + while(*equalSign == ' ') { + equalSign++; + } + + /* Premature end or zero-length value */ + if (!*equalSign || equalSign == semicolon) { + *status = U_INVALID_FORMAT_ERROR; + return; + } + + keywordList[numKeywords].valueStart = equalSign; + + pos = semicolon; + i = 0; + if(pos) { + while(*(pos - i - 1) == ' ') { + i++; + } + keywordList[numKeywords].valueLen = (int32_t)(pos - equalSign - i); + pos++; + } else { + i = (int32_t)uprv_strlen(equalSign); + while(i && equalSign[i-1] == ' ') { + i--; + } + keywordList[numKeywords].valueLen = i; + } + /* If this is a duplicate keyword, then ignore it */ + for (j=0; j startSearchHere && *(keyValueTail-1) == ' ') { + keyValueTail--; + } + /* now keyValueTail points to first char after the keyName */ + /* copy & normalize keyName from locale */ + if (startSearchHere == keyValueTail) { + *status = U_ILLEGAL_ARGUMENT_ERROR; /* empty keyword name in passed-in locale */ + return; + } + keyValueLen = 0; + while (startSearchHere < keyValueTail) { + if (!UPRV_ISALPHANUM(*startSearchHere)) { + *status = U_ILLEGAL_ARGUMENT_ERROR; /* malformed keyword name */ + return; + } + if (keyValueLen < ULOC_KEYWORD_BUFFER_LEN - 1) { + localeKeywordNameBuffer[keyValueLen++] = uprv_tolower(*startSearchHere++); + } else { + /* keyword name too long for internal buffer */ + *status = U_INTERNAL_PROGRAM_ERROR; + return; + } + } + localeKeywordNameBuffer[keyValueLen] = 0; /* terminate */ + + startSearchHere = uprv_strchr(nextSeparator, ';'); + + if(uprv_strcmp(keywordNameBuffer, localeKeywordNameBuffer) == 0) { + /* current entry matches the keyword. */ + nextSeparator++; /* skip '=' */ + /* First strip leading & trailing spaces (TC decided to tolerate these) */ + while(*nextSeparator == ' ') { + nextSeparator++; + } + keyValueTail = (startSearchHere)? startSearchHere: nextSeparator + uprv_strlen(nextSeparator); + while(keyValueTail > nextSeparator && *(keyValueTail-1) == ' ') { + keyValueTail--; + } + /* Now copy the value, but check well-formedness */ + if (nextSeparator == keyValueTail) { + *status = U_ILLEGAL_ARGUMENT_ERROR; /* empty key value name in passed-in locale */ + return; + } + while (nextSeparator < keyValueTail) { + if (!UPRV_ISALPHANUM(*nextSeparator) && !UPRV_OK_VALUE_PUNCTUATION(*nextSeparator)) { + *status = U_ILLEGAL_ARGUMENT_ERROR; /* malformed key value */ + return; + } + /* Should we lowercase value to return here? Tests expect as-is. */ + sink.Append(nextSeparator++, 1); + } + return; + } + } + } +} + +U_CAPI int32_t U_EXPORT2 +uloc_setKeywordValue(const char* keywordName, + const char* keywordValue, + char* buffer, int32_t bufferCapacity, + UErrorCode* status) +{ + /* TODO: sorting. removal. */ + int32_t keywordNameLen; + int32_t keywordValueLen; + int32_t bufLen; + int32_t needLen = 0; + char keywordNameBuffer[ULOC_KEYWORD_BUFFER_LEN]; + char keywordValueBuffer[ULOC_KEYWORDS_CAPACITY+1]; + char localeKeywordNameBuffer[ULOC_KEYWORD_BUFFER_LEN]; + int32_t rc; + char* nextSeparator = nullptr; + char* nextEqualsign = nullptr; + char* startSearchHere = nullptr; + char* keywordStart = nullptr; + CharString updatedKeysAndValues; + UBool handledInputKeyAndValue = false; + char keyValuePrefix = '@'; + + if(U_FAILURE(*status)) { + return -1; + } + if (*status == U_STRING_NOT_TERMINATED_WARNING) { + *status = U_ZERO_ERROR; + } + if (keywordName == nullptr || keywordName[0] == 0 || bufferCapacity <= 1) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + bufLen = (int32_t)uprv_strlen(buffer); + if(bufferCapacity= bufferCapacity) { + *status = U_BUFFER_OVERFLOW_ERROR; + return needLen; /* no change */ + } + *startSearchHere++ = '@'; + uprv_strcpy(startSearchHere, keywordNameBuffer); + startSearchHere += keywordNameLen; + *startSearchHere++ = '='; + uprv_strcpy(startSearchHere, keywordValueBuffer); + U_ASSERT(*status != U_STRING_NOT_TERMINATED_WARNING); + return needLen; + } /* end shortcut - no @ */ + + keywordStart = startSearchHere; + /* search for keyword */ + while(keywordStart) { + const char* keyValueTail; + int32_t keyValueLen; + + keywordStart++; /* skip @ or ; */ + nextEqualsign = uprv_strchr(keywordStart, '='); + if (!nextEqualsign) { + *status = U_ILLEGAL_ARGUMENT_ERROR; /* key must have =value */ + return 0; + } + /* strip leading & trailing spaces (TC decided to tolerate these) */ + while(*keywordStart == ' ') { + keywordStart++; + } + keyValueTail = nextEqualsign; + while (keyValueTail > keywordStart && *(keyValueTail-1) == ' ') { + keyValueTail--; + } + /* now keyValueTail points to first char after the keyName */ + /* copy & normalize keyName from locale */ + if (keywordStart == keyValueTail) { + *status = U_ILLEGAL_ARGUMENT_ERROR; /* empty keyword name in passed-in locale */ + return 0; + } + keyValueLen = 0; + while (keywordStart < keyValueTail) { + if (!UPRV_ISALPHANUM(*keywordStart)) { + *status = U_ILLEGAL_ARGUMENT_ERROR; /* malformed keyword name */ + return 0; + } + if (keyValueLen < ULOC_KEYWORD_BUFFER_LEN - 1) { + localeKeywordNameBuffer[keyValueLen++] = uprv_tolower(*keywordStart++); + } else { + /* keyword name too long for internal buffer */ + *status = U_INTERNAL_PROGRAM_ERROR; + return 0; + } + } + localeKeywordNameBuffer[keyValueLen] = 0; /* terminate */ + + nextSeparator = uprv_strchr(nextEqualsign, ';'); + + /* start processing the value part */ + nextEqualsign++; /* skip '=' */ + /* First strip leading & trailing spaces (TC decided to tolerate these) */ + while(*nextEqualsign == ' ') { + nextEqualsign++; + } + keyValueTail = (nextSeparator)? nextSeparator: nextEqualsign + uprv_strlen(nextEqualsign); + while(keyValueTail > nextEqualsign && *(keyValueTail-1) == ' ') { + keyValueTail--; + } + if (nextEqualsign == keyValueTail) { + *status = U_ILLEGAL_ARGUMENT_ERROR; /* empty key value in passed-in locale */ + return 0; + } + + rc = uprv_strcmp(keywordNameBuffer, localeKeywordNameBuffer); + if(rc == 0) { + /* Current entry matches the input keyword. Update the entry */ + if(keywordValueLen > 0) { /* updating a value */ + updatedKeysAndValues.append(keyValuePrefix, *status); + keyValuePrefix = ';'; /* for any subsequent key-value pair */ + updatedKeysAndValues.append(keywordNameBuffer, keywordNameLen, *status); + updatedKeysAndValues.append('=', *status); + updatedKeysAndValues.append(keywordValueBuffer, keywordValueLen, *status); + } /* else removing this entry, don't emit anything */ + handledInputKeyAndValue = true; + } else { + /* input keyword sorts earlier than current entry, add before current entry */ + if (rc < 0 && keywordValueLen > 0 && !handledInputKeyAndValue) { + /* insert new entry at this location */ + updatedKeysAndValues.append(keyValuePrefix, *status); + keyValuePrefix = ';'; /* for any subsequent key-value pair */ + updatedKeysAndValues.append(keywordNameBuffer, keywordNameLen, *status); + updatedKeysAndValues.append('=', *status); + updatedKeysAndValues.append(keywordValueBuffer, keywordValueLen, *status); + handledInputKeyAndValue = true; + } + /* copy the current entry */ + updatedKeysAndValues.append(keyValuePrefix, *status); + keyValuePrefix = ';'; /* for any subsequent key-value pair */ + updatedKeysAndValues.append(localeKeywordNameBuffer, keyValueLen, *status); + updatedKeysAndValues.append('=', *status); + updatedKeysAndValues.append(nextEqualsign, static_cast(keyValueTail-nextEqualsign), *status); + } + if (!nextSeparator && keywordValueLen > 0 && !handledInputKeyAndValue) { + /* append new entry at the end, it sorts later than existing entries */ + updatedKeysAndValues.append(keyValuePrefix, *status); + /* skip keyValuePrefix update, no subsequent key-value pair */ + updatedKeysAndValues.append(keywordNameBuffer, keywordNameLen, *status); + updatedKeysAndValues.append('=', *status); + updatedKeysAndValues.append(keywordValueBuffer, keywordValueLen, *status); + handledInputKeyAndValue = true; + } + keywordStart = nextSeparator; + } /* end loop searching */ + + /* Any error from updatedKeysAndValues.append above would be internal and not due to + * problems with the passed-in locale. So if we did encounter problems with the + * passed-in locale above, those errors took precedence and overrode any error + * status from updatedKeysAndValues.append, and also caused a return of 0. If there + * are errors here they are from updatedKeysAndValues.append; they do cause an + * error return but the passed-in locale is unmodified and the original bufLen is + * returned. + */ + if (!handledInputKeyAndValue || U_FAILURE(*status)) { + /* if input key/value specified removal of a keyword not present in locale, or + * there was an error in CharString.append, leave original locale alone. */ + U_ASSERT(*status != U_STRING_NOT_TERMINATED_WARNING); + return bufLen; + } + + // needLen = length of the part before '@' + needLen = (int32_t)(startSearchHere - buffer); + // Check to see can we fit the startSearchHere, if not, return + // U_BUFFER_OVERFLOW_ERROR without copy updatedKeysAndValues into it. + // We do this because this API function does not behave like most others: + // It promises never to set a U_STRING_NOT_TERMINATED_WARNING. + // When the contents fits but without the terminating NUL, in this case we need to not change + // the buffer contents and return with a buffer overflow error. + int32_t appendLength = updatedKeysAndValues.length(); + if (appendLength >= bufferCapacity - needLen) { + *status = U_BUFFER_OVERFLOW_ERROR; + return needLen + appendLength; + } + needLen += updatedKeysAndValues.extract( + startSearchHere, bufferCapacity - needLen, *status); + U_ASSERT(*status != U_STRING_NOT_TERMINATED_WARNING); + return needLen; +} + +/* ### ID parsing implementation **************************************************/ + +#define _isPrefixLetter(a) ((a=='x')||(a=='X')||(a=='i')||(a=='I')) + +/*returns true if one of the special prefixes is here (s=string) + 'x-' or 'i-' */ +#define _isIDPrefix(s) (_isPrefixLetter(s[0])&&_isIDSeparator(s[1])) + +/* Dot terminates it because of POSIX form where dot precedes the codepage + * except for variant + */ +#define _isTerminator(a) ((a==0)||(a=='.')||(a=='@')) + +/** + * Lookup 'key' in the array 'list'. The array 'list' should contain + * a nullptr entry, followed by more entries, and a second nullptr entry. + * + * The 'list' param should be LANGUAGES, LANGUAGES_3, COUNTRIES, or + * COUNTRIES_3. + */ +static int16_t _findIndex(const char* const* list, const char* key) +{ + const char* const* anchor = list; + int32_t pass = 0; + + /* Make two passes through two nullptr-terminated arrays at 'list' */ + while (pass++ < 2) { + while (*list) { + if (uprv_strcmp(key, *list) == 0) { + return (int16_t)(list - anchor); + } + list++; + } + ++list; /* skip final nullptr *CWB*/ + } + return -1; +} + +U_CFUNC const char* +uloc_getCurrentCountryID(const char* oldID){ + int32_t offset = _findIndex(DEPRECATED_COUNTRIES, oldID); + if (offset >= 0) { + return REPLACEMENT_COUNTRIES[offset]; + } + return oldID; +} +U_CFUNC const char* +uloc_getCurrentLanguageID(const char* oldID){ + int32_t offset = _findIndex(DEPRECATED_LANGUAGES, oldID); + if (offset >= 0) { + return REPLACEMENT_LANGUAGES[offset]; + } + return oldID; +} +/* + * the internal functions _getLanguage(), _getCountry(), _getVariant() + * avoid duplicating code to handle the earlier locale ID pieces + * in the functions for the later ones by + * setting the *pEnd pointer to where they stopped parsing + * + * TODO try to use this in Locale + */ +CharString U_EXPORT2 +ulocimp_getLanguage(const char *localeID, + const char **pEnd, + UErrorCode &status) { + CharString result; + + if (uprv_stricmp(localeID, "root") == 0) { + localeID += 4; + } else if (uprv_strnicmp(localeID, "und", 3) == 0 && + (localeID[3] == '\0' || + localeID[3] == '-' || + localeID[3] == '_' || + localeID[3] == '@')) { + localeID += 3; + } + + /* if it starts with i- or x- then copy that prefix */ + if(_isIDPrefix(localeID)) { + result.append((char)uprv_tolower(*localeID), status); + result.append('-', status); + localeID+=2; + } + + /* copy the language as far as possible and count its length */ + while(!_isTerminator(*localeID) && !_isIDSeparator(*localeID)) { + result.append((char)uprv_tolower(*localeID), status); + localeID++; + } + + if(result.length()==3) { + /* convert 3 character code to 2 character code if possible *CWB*/ + int32_t offset = _findIndex(LANGUAGES_3, result.data()); + if(offset>=0) { + result.clear(); + result.append(LANGUAGES[offset], status); + } + } + + if(pEnd!=nullptr) { + *pEnd=localeID; + } + + return result; +} + +CharString U_EXPORT2 +ulocimp_getScript(const char *localeID, + const char **pEnd, + UErrorCode &status) { + CharString result; + int32_t idLen = 0; + + if (pEnd != nullptr) { + *pEnd = localeID; + } + + /* copy the second item as far as possible and count its length */ + while(!_isTerminator(localeID[idLen]) && !_isIDSeparator(localeID[idLen]) + && uprv_isASCIILetter(localeID[idLen])) { + idLen++; + } + + /* If it's exactly 4 characters long, then it's a script and not a country. */ + if (idLen == 4) { + int32_t i; + if (pEnd != nullptr) { + *pEnd = localeID+idLen; + } + if (idLen >= 1) { + result.append((char)uprv_toupper(*(localeID++)), status); + } + for (i = 1; i < idLen; i++) { + result.append((char)uprv_tolower(*(localeID++)), status); + } + } + + return result; +} + +CharString U_EXPORT2 +ulocimp_getCountry(const char *localeID, + const char **pEnd, + UErrorCode &status) { + CharString result; + int32_t idLen=0; + + /* copy the country as far as possible and count its length */ + while(!_isTerminator(localeID[idLen]) && !_isIDSeparator(localeID[idLen])) { + result.append((char)uprv_toupper(localeID[idLen]), status); + idLen++; + } + + /* the country should be either length 2 or 3 */ + if (idLen == 2 || idLen == 3) { + /* convert 3 character code to 2 character code if possible *CWB*/ + if(idLen==3) { + int32_t offset = _findIndex(COUNTRIES_3, result.data()); + if(offset>=0) { + result.clear(); + result.append(COUNTRIES[offset], status); + } + } + localeID+=idLen; + } else { + result.clear(); + } + + if(pEnd!=nullptr) { + *pEnd=localeID; + } + + return result; +} + +/** + * @param needSeparator if true, then add leading '_' if any variants + * are added to 'variant' + */ +static void +_getVariant(const char *localeID, + char prev, + ByteSink& sink, + UBool needSeparator) { + UBool hasVariant = false; + + /* get one or more variant tags and separate them with '_' */ + if(_isIDSeparator(prev)) { + /* get a variant string after a '-' or '_' */ + while(!_isTerminator(*localeID)) { + if (needSeparator) { + sink.Append("_", 1); + needSeparator = false; + } + char c = (char)uprv_toupper(*localeID); + if (c == '-') c = '_'; + sink.Append(&c, 1); + hasVariant = true; + localeID++; + } + } + + /* if there is no variant tag after a '-' or '_' then look for '@' */ + if(!hasVariant) { + if(prev=='@') { + /* keep localeID */ + } else if((localeID=locale_getKeywordsStart(localeID))!=nullptr) { + ++localeID; /* point after the '@' */ + } else { + return; + } + while(!_isTerminator(*localeID)) { + if (needSeparator) { + sink.Append("_", 1); + needSeparator = false; + } + char c = (char)uprv_toupper(*localeID); + if (c == '-' || c == ',') c = '_'; + sink.Append(&c, 1); + localeID++; + } + } +} + +/* Keyword enumeration */ + +typedef struct UKeywordsContext { + char* keywords; + char* current; +} UKeywordsContext; + +U_CDECL_BEGIN + +static void U_CALLCONV +uloc_kw_closeKeywords(UEnumeration *enumerator) { + uprv_free(((UKeywordsContext *)enumerator->context)->keywords); + uprv_free(enumerator->context); + uprv_free(enumerator); +} + +static int32_t U_CALLCONV +uloc_kw_countKeywords(UEnumeration *en, UErrorCode * /*status*/) { + char *kw = ((UKeywordsContext *)en->context)->keywords; + int32_t result = 0; + while(*kw) { + result++; + kw += uprv_strlen(kw)+1; + } + return result; +} + +static const char * U_CALLCONV +uloc_kw_nextKeyword(UEnumeration* en, + int32_t* resultLength, + UErrorCode* /*status*/) { + const char* result = ((UKeywordsContext *)en->context)->current; + int32_t len = 0; + if(*result) { + len = (int32_t)uprv_strlen(((UKeywordsContext *)en->context)->current); + ((UKeywordsContext *)en->context)->current += len+1; + } else { + result = nullptr; + } + if (resultLength) { + *resultLength = len; + } + return result; +} + +static void U_CALLCONV +uloc_kw_resetKeywords(UEnumeration* en, + UErrorCode* /*status*/) { + ((UKeywordsContext *)en->context)->current = ((UKeywordsContext *)en->context)->keywords; +} + +U_CDECL_END + + +static const UEnumeration gKeywordsEnum = { + nullptr, + nullptr, + uloc_kw_closeKeywords, + uloc_kw_countKeywords, + uenum_unextDefault, + uloc_kw_nextKeyword, + uloc_kw_resetKeywords +}; + +U_CAPI UEnumeration* U_EXPORT2 +uloc_openKeywordList(const char *keywordList, int32_t keywordListSize, UErrorCode* status) +{ + LocalMemory myContext; + LocalMemory result; + + if (U_FAILURE(*status)) { + return nullptr; + } + myContext.adoptInstead(static_cast(uprv_malloc(sizeof(UKeywordsContext)))); + result.adoptInstead(static_cast(uprv_malloc(sizeof(UEnumeration)))); + if (myContext.isNull() || result.isNull()) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memcpy(result.getAlias(), &gKeywordsEnum, sizeof(UEnumeration)); + myContext->keywords = static_cast(uprv_malloc(keywordListSize+1)); + if (myContext->keywords == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + uprv_memcpy(myContext->keywords, keywordList, keywordListSize); + myContext->keywords[keywordListSize] = 0; + myContext->current = myContext->keywords; + result->context = myContext.orphan(); + return result.orphan(); +} + +U_CAPI UEnumeration* U_EXPORT2 +uloc_openKeywords(const char* localeID, + UErrorCode* status) +{ + char tempBuffer[ULOC_FULLNAME_CAPACITY]; + const char* tmpLocaleID; + + if(status==nullptr || U_FAILURE(*status)) { + return 0; + } + + if (_hasBCP47Extension(localeID)) { + tmpLocaleID = _ConvertBCP47(localeID, tempBuffer, + sizeof(tempBuffer), status, nullptr); + } else { + if (localeID==nullptr) { + localeID=uloc_getDefault(); + } + tmpLocaleID=localeID; + } + + /* Skip the language */ + ulocimp_getLanguage(tmpLocaleID, &tmpLocaleID, *status); + if (U_FAILURE(*status)) { + return 0; + } + + if(_isIDSeparator(*tmpLocaleID)) { + const char *scriptID; + /* Skip the script if available */ + ulocimp_getScript(tmpLocaleID+1, &scriptID, *status); + if (U_FAILURE(*status)) { + return 0; + } + if(scriptID != tmpLocaleID+1) { + /* Found optional script */ + tmpLocaleID = scriptID; + } + /* Skip the Country */ + if (_isIDSeparator(*tmpLocaleID)) { + ulocimp_getCountry(tmpLocaleID+1, &tmpLocaleID, *status); + if (U_FAILURE(*status)) { + return 0; + } + } + } + + /* keywords are located after '@' */ + if((tmpLocaleID = locale_getKeywordsStart(tmpLocaleID)) != nullptr) { + CharString keywords; + CharStringByteSink sink(&keywords); + ulocimp_getKeywords(tmpLocaleID+1, '@', sink, false, status); + if (U_FAILURE(*status)) { + return nullptr; + } + return uloc_openKeywordList(keywords.data(), keywords.length(), status); + } + return nullptr; +} + + +/* bit-flags for 'options' parameter of _canonicalize */ +#define _ULOC_STRIP_KEYWORDS 0x2 +#define _ULOC_CANONICALIZE 0x1 + +#define OPTION_SET(options, mask) ((options & mask) != 0) + +static const char i_default[] = {'i', '-', 'd', 'e', 'f', 'a', 'u', 'l', 't'}; +#define I_DEFAULT_LENGTH UPRV_LENGTHOF(i_default) + +/** + * Canonicalize the given localeID, to level 1 or to level 2, + * depending on the options. To specify level 1, pass in options=0. + * To specify level 2, pass in options=_ULOC_CANONICALIZE. + * + * This is the code underlying uloc_getName and uloc_canonicalize. + */ +static void +_canonicalize(const char* localeID, + ByteSink& sink, + uint32_t options, + UErrorCode* err) { + if (U_FAILURE(*err)) { + return; + } + + int32_t j, fieldCount=0, scriptSize=0, variantSize=0; + PreflightingLocaleIDBuffer tempBuffer; // if localeID has a BCP47 extension, tmpLocaleID points to this + CharString localeIDWithHyphens; // if localeID has a BPC47 extension and have _, tmpLocaleID points to this + const char* origLocaleID; + const char* tmpLocaleID; + const char* keywordAssign = nullptr; + const char* separatorIndicator = nullptr; + + if (_hasBCP47Extension(localeID)) { + const char* localeIDPtr = localeID; + + // convert all underbars to hyphens, unless the "BCP47 extension" comes at the beginning of the string + if (uprv_strchr(localeID, '_') != nullptr && localeID[1] != '-' && localeID[1] != '_') { + localeIDWithHyphens.append(localeID, -1, *err); + if (U_SUCCESS(*err)) { + for (char* p = localeIDWithHyphens.data(); *p != '\0'; ++p) { + if (*p == '_') { + *p = '-'; + } + } + localeIDPtr = localeIDWithHyphens.data(); + } + } + + do { + // After this call tmpLocaleID may point to localeIDPtr which may + // point to either localeID or localeIDWithHyphens.data(). + tmpLocaleID = _ConvertBCP47(localeIDPtr, tempBuffer.getBuffer(), + tempBuffer.getCapacity(), err, + &(tempBuffer.requestedCapacity)); + } while (tempBuffer.needToTryAgain(err)); + } else { + if (localeID==nullptr) { + localeID=uloc_getDefault(); + } + tmpLocaleID=localeID; + } + + origLocaleID=tmpLocaleID; + + /* get all pieces, one after another, and separate with '_' */ + CharString tag = ulocimp_getLanguage(tmpLocaleID, &tmpLocaleID, *err); + + if (tag.length() == I_DEFAULT_LENGTH && + uprv_strncmp(origLocaleID, i_default, I_DEFAULT_LENGTH) == 0) { + tag.clear(); + tag.append(uloc_getDefault(), *err); + } else if(_isIDSeparator(*tmpLocaleID)) { + const char *scriptID; + + ++fieldCount; + tag.append('_', *err); + + CharString script = ulocimp_getScript(tmpLocaleID+1, &scriptID, *err); + tag.append(script, *err); + scriptSize = script.length(); + if(scriptSize > 0) { + /* Found optional script */ + tmpLocaleID = scriptID; + ++fieldCount; + if (_isIDSeparator(*tmpLocaleID)) { + /* If there is something else, then we add the _ */ + tag.append('_', *err); + } + } + + if (_isIDSeparator(*tmpLocaleID)) { + const char *cntryID; + + CharString country = ulocimp_getCountry(tmpLocaleID+1, &cntryID, *err); + tag.append(country, *err); + if (!country.isEmpty()) { + /* Found optional country */ + tmpLocaleID = cntryID; + } + if(_isIDSeparator(*tmpLocaleID)) { + /* If there is something else, then we add the _ if we found country before. */ + if (!_isIDSeparator(*(tmpLocaleID+1))) { + ++fieldCount; + tag.append('_', *err); + } + + variantSize = -tag.length(); + { + CharStringByteSink s(&tag); + _getVariant(tmpLocaleID+1, *tmpLocaleID, s, false); + } + variantSize += tag.length(); + if (variantSize > 0) { + tmpLocaleID += variantSize + 1; /* skip '_' and variant */ + } + } + } + } + + /* Copy POSIX-style charset specifier, if any [mr.utf8] */ + if (!OPTION_SET(options, _ULOC_CANONICALIZE) && *tmpLocaleID == '.') { + UBool done = false; + do { + char c = *tmpLocaleID; + switch (c) { + case 0: + case '@': + done = true; + break; + default: + tag.append(c, *err); + ++tmpLocaleID; + break; + } + } while (!done); + } + + /* Scan ahead to next '@' and determine if it is followed by '=' and/or ';' + After this, tmpLocaleID either points to '@' or is nullptr */ + if ((tmpLocaleID=locale_getKeywordsStart(tmpLocaleID))!=nullptr) { + keywordAssign = uprv_strchr(tmpLocaleID, '='); + separatorIndicator = uprv_strchr(tmpLocaleID, ';'); + } + + /* Copy POSIX-style variant, if any [mr@FOO] */ + if (!OPTION_SET(options, _ULOC_CANONICALIZE) && + tmpLocaleID != nullptr && keywordAssign == nullptr) { + for (;;) { + char c = *tmpLocaleID; + if (c == 0) { + break; + } + tag.append(c, *err); + ++tmpLocaleID; + } + } + + if (OPTION_SET(options, _ULOC_CANONICALIZE)) { + /* Handle @FOO variant if @ is present and not followed by = */ + if (tmpLocaleID!=nullptr && keywordAssign==nullptr) { + /* Add missing '_' if needed */ + if (fieldCount < 2 || (fieldCount < 3 && scriptSize > 0)) { + do { + tag.append('_', *err); + ++fieldCount; + } while(fieldCount<2); + } + + int32_t posixVariantSize = -tag.length(); + { + CharStringByteSink s(&tag); + _getVariant(tmpLocaleID+1, '@', s, (UBool)(variantSize > 0)); + } + posixVariantSize += tag.length(); + if (posixVariantSize > 0) { + variantSize += posixVariantSize; + } + } + + /* Look up the ID in the canonicalization map */ + for (j=0; j keywordAssign)) { + sink.Append("@", 1); + ++fieldCount; + ulocimp_getKeywords(tmpLocaleID+1, '@', sink, true, err); + } + } +} + +/* ### ID parsing API **************************************************/ + +U_CAPI int32_t U_EXPORT2 +uloc_getParent(const char* localeID, + char* parent, + int32_t parentCapacity, + UErrorCode* err) +{ + const char *lastUnderscore; + int32_t i; + + if (U_FAILURE(*err)) + return 0; + + if (localeID == nullptr) + localeID = uloc_getDefault(); + + lastUnderscore=uprv_strrchr(localeID, '_'); + if(lastUnderscore!=nullptr) { + i=(int32_t)(lastUnderscore-localeID); + } else { + i=0; + } + + if (i > 0) { + if (uprv_strnicmp(localeID, "und_", 4) == 0) { + localeID += 3; + i -= 3; + uprv_memmove(parent, localeID, uprv_min(i, parentCapacity)); + } else if (parent != localeID) { + uprv_memcpy(parent, localeID, uprv_min(i, parentCapacity)); + } + } + + return u_terminateChars(parent, parentCapacity, i, err); +} + +U_CAPI int32_t U_EXPORT2 +uloc_getLanguage(const char* localeID, + char* language, + int32_t languageCapacity, + UErrorCode* err) +{ + /* uloc_getLanguage will return a 2 character iso-639 code if one exists. *CWB*/ + + if (err==nullptr || U_FAILURE(*err)) { + return 0; + } + + if(localeID==nullptr) { + localeID=uloc_getDefault(); + } + + return ulocimp_getLanguage(localeID, nullptr, *err).extract(language, languageCapacity, *err); +} + +U_CAPI int32_t U_EXPORT2 +uloc_getScript(const char* localeID, + char* script, + int32_t scriptCapacity, + UErrorCode* err) +{ + if(err==nullptr || U_FAILURE(*err)) { + return 0; + } + + if(localeID==nullptr) { + localeID=uloc_getDefault(); + } + + /* skip the language */ + ulocimp_getLanguage(localeID, &localeID, *err); + if (U_FAILURE(*err)) { + return 0; + } + + if(_isIDSeparator(*localeID)) { + return ulocimp_getScript(localeID+1, nullptr, *err).extract(script, scriptCapacity, *err); + } + return u_terminateChars(script, scriptCapacity, 0, err); +} + +U_CAPI int32_t U_EXPORT2 +uloc_getCountry(const char* localeID, + char* country, + int32_t countryCapacity, + UErrorCode* err) +{ + if(err==nullptr || U_FAILURE(*err)) { + return 0; + } + + if(localeID==nullptr) { + localeID=uloc_getDefault(); + } + + /* Skip the language */ + ulocimp_getLanguage(localeID, &localeID, *err); + if (U_FAILURE(*err)) { + return 0; + } + + if(_isIDSeparator(*localeID)) { + const char *scriptID; + /* Skip the script if available */ + ulocimp_getScript(localeID+1, &scriptID, *err); + if (U_FAILURE(*err)) { + return 0; + } + if(scriptID != localeID+1) { + /* Found optional script */ + localeID = scriptID; + } + if(_isIDSeparator(*localeID)) { + return ulocimp_getCountry(localeID+1, nullptr, *err).extract(country, countryCapacity, *err); + } + } + return u_terminateChars(country, countryCapacity, 0, err); +} + +U_CAPI int32_t U_EXPORT2 +uloc_getVariant(const char* localeID, + char* variant, + int32_t variantCapacity, + UErrorCode* err) +{ + char tempBuffer[ULOC_FULLNAME_CAPACITY]; + const char* tmpLocaleID; + int32_t i=0; + + if(err==nullptr || U_FAILURE(*err)) { + return 0; + } + + if (_hasBCP47Extension(localeID)) { + tmpLocaleID =_ConvertBCP47(localeID, tempBuffer, sizeof(tempBuffer), err, nullptr); + } else { + if (localeID==nullptr) { + localeID=uloc_getDefault(); + } + tmpLocaleID=localeID; + } + + /* Skip the language */ + ulocimp_getLanguage(tmpLocaleID, &tmpLocaleID, *err); + if (U_FAILURE(*err)) { + return 0; + } + + if(_isIDSeparator(*tmpLocaleID)) { + const char *scriptID; + /* Skip the script if available */ + ulocimp_getScript(tmpLocaleID+1, &scriptID, *err); + if (U_FAILURE(*err)) { + return 0; + } + if(scriptID != tmpLocaleID+1) { + /* Found optional script */ + tmpLocaleID = scriptID; + } + /* Skip the Country */ + if (_isIDSeparator(*tmpLocaleID)) { + const char *cntryID; + ulocimp_getCountry(tmpLocaleID+1, &cntryID, *err); + if (U_FAILURE(*err)) { + return 0; + } + if (cntryID != tmpLocaleID+1) { + /* Found optional country */ + tmpLocaleID = cntryID; + } + if(_isIDSeparator(*tmpLocaleID)) { + /* If there was no country ID, skip a possible extra IDSeparator */ + if (tmpLocaleID != cntryID && _isIDSeparator(tmpLocaleID[1])) { + tmpLocaleID++; + } + + CheckedArrayByteSink sink(variant, variantCapacity); + _getVariant(tmpLocaleID+1, *tmpLocaleID, sink, false); + + i = sink.NumberOfBytesAppended(); + + if (U_FAILURE(*err)) { + return i; + } + + if (sink.Overflowed()) { + *err = U_BUFFER_OVERFLOW_ERROR; + return i; + } + } + } + } + + return u_terminateChars(variant, variantCapacity, i, err); +} + +U_CAPI int32_t U_EXPORT2 +uloc_getName(const char* localeID, + char* name, + int32_t nameCapacity, + UErrorCode* err) +{ + if (U_FAILURE(*err)) { + return 0; + } + + CheckedArrayByteSink sink(name, nameCapacity); + ulocimp_getName(localeID, sink, err); + + int32_t reslen = sink.NumberOfBytesAppended(); + + if (U_FAILURE(*err)) { + return reslen; + } + + if (sink.Overflowed()) { + *err = U_BUFFER_OVERFLOW_ERROR; + } else { + u_terminateChars(name, nameCapacity, reslen, err); + } + + return reslen; +} + +U_CAPI void U_EXPORT2 +ulocimp_getName(const char* localeID, + ByteSink& sink, + UErrorCode* err) +{ + _canonicalize(localeID, sink, 0, err); +} + +U_CAPI int32_t U_EXPORT2 +uloc_getBaseName(const char* localeID, + char* name, + int32_t nameCapacity, + UErrorCode* err) +{ + if (U_FAILURE(*err)) { + return 0; + } + + CheckedArrayByteSink sink(name, nameCapacity); + ulocimp_getBaseName(localeID, sink, err); + + int32_t reslen = sink.NumberOfBytesAppended(); + + if (U_FAILURE(*err)) { + return reslen; + } + + if (sink.Overflowed()) { + *err = U_BUFFER_OVERFLOW_ERROR; + } else { + u_terminateChars(name, nameCapacity, reslen, err); + } + + return reslen; +} + +U_CAPI void U_EXPORT2 +ulocimp_getBaseName(const char* localeID, + ByteSink& sink, + UErrorCode* err) +{ + _canonicalize(localeID, sink, _ULOC_STRIP_KEYWORDS, err); +} + +U_CAPI int32_t U_EXPORT2 +uloc_canonicalize(const char* localeID, + char* name, + int32_t nameCapacity, + UErrorCode* err) +{ + if (U_FAILURE(*err)) { + return 0; + } + + CheckedArrayByteSink sink(name, nameCapacity); + ulocimp_canonicalize(localeID, sink, err); + + int32_t reslen = sink.NumberOfBytesAppended(); + + if (U_FAILURE(*err)) { + return reslen; + } + + if (sink.Overflowed()) { + *err = U_BUFFER_OVERFLOW_ERROR; + } else { + u_terminateChars(name, nameCapacity, reslen, err); + } + + return reslen; +} + +U_CAPI void U_EXPORT2 +ulocimp_canonicalize(const char* localeID, + ByteSink& sink, + UErrorCode* err) +{ + _canonicalize(localeID, sink, _ULOC_CANONICALIZE, err); +} + +U_CAPI const char* U_EXPORT2 +uloc_getISO3Language(const char* localeID) +{ + int16_t offset; + char lang[ULOC_LANG_CAPACITY]; + UErrorCode err = U_ZERO_ERROR; + + if (localeID == nullptr) + { + localeID = uloc_getDefault(); + } + uloc_getLanguage(localeID, lang, ULOC_LANG_CAPACITY, &err); + if (U_FAILURE(err)) + return ""; + offset = _findIndex(LANGUAGES, lang); + if (offset < 0) + return ""; + return LANGUAGES_3[offset]; +} + +U_CAPI const char* U_EXPORT2 +uloc_getISO3Country(const char* localeID) +{ + int16_t offset; + char cntry[ULOC_LANG_CAPACITY]; + UErrorCode err = U_ZERO_ERROR; + + if (localeID == nullptr) + { + localeID = uloc_getDefault(); + } + uloc_getCountry(localeID, cntry, ULOC_LANG_CAPACITY, &err); + if (U_FAILURE(err)) + return ""; + offset = _findIndex(COUNTRIES, cntry); + if (offset < 0) + return ""; + + return COUNTRIES_3[offset]; +} + +U_CAPI uint32_t U_EXPORT2 +uloc_getLCID(const char* localeID) +{ + UErrorCode status = U_ZERO_ERROR; + char langID[ULOC_FULLNAME_CAPACITY]; + uint32_t lcid = 0; + + /* Check for incomplete id. */ + if (!localeID || uprv_strlen(localeID) < 2) { + return 0; + } + + // First, attempt Windows platform lookup if available, but fall + // through to catch any special cases (ICU vs Windows name differences). + lcid = uprv_convertToLCIDPlatform(localeID, &status); + if (U_FAILURE(status)) { + return 0; + } + if (lcid > 0) { + // Windows found an LCID, return that + return lcid; + } + + uloc_getLanguage(localeID, langID, sizeof(langID), &status); + if (U_FAILURE(status) || status == U_STRING_NOT_TERMINATED_WARNING) { + return 0; + } + + if (uprv_strchr(localeID, '@')) { + // uprv_convertToLCID does not support keywords other than collation. + // Remove all keywords except collation. + int32_t len; + char tmpLocaleID[ULOC_FULLNAME_CAPACITY]; + + CharString collVal; + { + CharStringByteSink sink(&collVal); + ulocimp_getKeywordValue(localeID, "collation", sink, &status); + } + + if (U_SUCCESS(status) && !collVal.isEmpty()) { + len = uloc_getBaseName(localeID, tmpLocaleID, + UPRV_LENGTHOF(tmpLocaleID) - 1, &status); + + if (U_SUCCESS(status) && len > 0) { + tmpLocaleID[len] = 0; + + len = uloc_setKeywordValue("collation", collVal.data(), tmpLocaleID, + UPRV_LENGTHOF(tmpLocaleID) - len - 1, &status); + + if (U_SUCCESS(status) && len > 0) { + tmpLocaleID[len] = 0; + return uprv_convertToLCID(langID, tmpLocaleID, &status); + } + } + } + + // fall through - all keywords are simply ignored + status = U_ZERO_ERROR; + } + + return uprv_convertToLCID(langID, localeID, &status); +} + +U_CAPI int32_t U_EXPORT2 +uloc_getLocaleForLCID(uint32_t hostid, char *locale, int32_t localeCapacity, + UErrorCode *status) +{ + return uprv_convertToPosix(hostid, locale, localeCapacity, status); +} + +/* ### Default locale **************************************************/ + +U_CAPI const char* U_EXPORT2 +uloc_getDefault() +{ + return locale_get_default(); +} + +U_CAPI void U_EXPORT2 +uloc_setDefault(const char* newDefaultLocale, + UErrorCode* err) +{ + if (U_FAILURE(*err)) + return; + /* the error code isn't currently used for anything by this function*/ + + /* propagate change to C++ */ + locale_set_default(newDefaultLocale); +} + +/** + * Returns a list of all 2-letter language codes defined in ISO 639. This is a pointer + * to an array of pointers to arrays of char. All of these pointers are owned + * by ICU-- do not delete them, and do not write through them. The array is + * terminated with a null pointer. + */ +U_CAPI const char* const* U_EXPORT2 +uloc_getISOLanguages() +{ + return LANGUAGES; +} + +/** + * Returns a list of all 2-letter country codes defined in ISO 639. This is a + * pointer to an array of pointers to arrays of char. All of these pointers are + * owned by ICU-- do not delete them, and do not write through them. The array is + * terminated with a null pointer. + */ +U_CAPI const char* const* U_EXPORT2 +uloc_getISOCountries() +{ + return COUNTRIES; +} + +U_CAPI const char* U_EXPORT2 +uloc_toUnicodeLocaleKey(const char* keyword) +{ + const char* bcpKey = ulocimp_toBcpKey(keyword); + if (bcpKey == nullptr && ultag_isUnicodeLocaleKey(keyword, -1)) { + // unknown keyword, but syntax is fine.. + return keyword; + } + return bcpKey; +} + +U_CAPI const char* U_EXPORT2 +uloc_toUnicodeLocaleType(const char* keyword, const char* value) +{ + const char* bcpType = ulocimp_toBcpType(keyword, value, nullptr, nullptr); + if (bcpType == nullptr && ultag_isUnicodeLocaleType(value, -1)) { + // unknown keyword, but syntax is fine.. + return value; + } + return bcpType; +} + +static UBool +isWellFormedLegacyKey(const char* legacyKey) +{ + const char* p = legacyKey; + while (*p) { + if (!UPRV_ISALPHANUM(*p)) { + return false; + } + p++; + } + return true; +} + +static UBool +isWellFormedLegacyType(const char* legacyType) +{ + const char* p = legacyType; + int32_t alphaNumLen = 0; + while (*p) { + if (*p == '_' || *p == '/' || *p == '-') { + if (alphaNumLen == 0) { + return false; + } + alphaNumLen = 0; + } else if (UPRV_ISALPHANUM(*p)) { + alphaNumLen++; + } else { + return false; + } + p++; + } + return (alphaNumLen != 0); +} + +U_CAPI const char* U_EXPORT2 +uloc_toLegacyKey(const char* keyword) +{ + const char* legacyKey = ulocimp_toLegacyKey(keyword); + if (legacyKey == nullptr) { + // Checks if the specified locale key is well-formed with the legacy locale syntax. + // + // Note: + // LDML/CLDR provides some definition of keyword syntax in + // * http://www.unicode.org/reports/tr35/#Unicode_locale_identifier and + // * http://www.unicode.org/reports/tr35/#Old_Locale_Extension_Syntax + // Keys can only consist of [0-9a-zA-Z]. + if (isWellFormedLegacyKey(keyword)) { + return keyword; + } + } + return legacyKey; +} + +U_CAPI const char* U_EXPORT2 +uloc_toLegacyType(const char* keyword, const char* value) +{ + const char* legacyType = ulocimp_toLegacyType(keyword, value, nullptr, nullptr); + if (legacyType == nullptr) { + // Checks if the specified locale type is well-formed with the legacy locale syntax. + // + // Note: + // LDML/CLDR provides some definition of keyword syntax in + // * http://www.unicode.org/reports/tr35/#Unicode_locale_identifier and + // * http://www.unicode.org/reports/tr35/#Old_Locale_Extension_Syntax + // Values (types) can only consist of [0-9a-zA-Z], plus for legacy values + // we allow [/_-+] in the middle (e.g. "Etc/GMT+1", "Asia/Tel_Aviv") + if (isWellFormedLegacyType(value)) { + return value; + } + } + return legacyType; +} + +/*eof*/ diff --git a/intl/icu/source/common/uloc_keytype.cpp b/intl/icu/source/common/uloc_keytype.cpp new file mode 100644 index 0000000000..a84b860907 --- /dev/null +++ b/intl/icu/source/common/uloc_keytype.cpp @@ -0,0 +1,536 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2014-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ +#include + +#include "unicode/utypes.h" +#include "unicode/unistr.h" +#include "unicode/uobject.h" + +#include "charstr.h" +#include "cmemory.h" +#include "cstring.h" +#include "uassert.h" +#include "ucln_cmn.h" +#include "uhash.h" +#include "umutex.h" +#include "uresimp.h" +#include "uvector.h" +#include "udataswp.h" /* for InvChar functions */ + +static UHashtable* gLocExtKeyMap = nullptr; +static icu::UInitOnce gLocExtKeyMapInitOnce {}; + +// bit flags for special types +typedef enum { + SPECIALTYPE_NONE = 0, + SPECIALTYPE_CODEPOINTS = 1, + SPECIALTYPE_REORDER_CODE = 2, + SPECIALTYPE_RG_KEY_VALUE = 4 +} SpecialType; + +struct LocExtKeyData : public icu::UMemory { + const char* legacyId; + const char* bcpId; + icu::LocalUHashtablePointer typeMap; + uint32_t specialTypes; +}; + +struct LocExtType : public icu::UMemory { + const char* legacyId; + const char* bcpId; +}; + +static icu::MemoryPool* gKeyTypeStringPool = nullptr; +static icu::MemoryPool* gLocExtKeyDataEntries = nullptr; +static icu::MemoryPool* gLocExtTypeEntries = nullptr; + +U_CDECL_BEGIN + +static UBool U_CALLCONV +uloc_key_type_cleanup() { + if (gLocExtKeyMap != nullptr) { + uhash_close(gLocExtKeyMap); + gLocExtKeyMap = nullptr; + } + + delete gLocExtKeyDataEntries; + gLocExtKeyDataEntries = nullptr; + + delete gLocExtTypeEntries; + gLocExtTypeEntries = nullptr; + + delete gKeyTypeStringPool; + gKeyTypeStringPool = nullptr; + + gLocExtKeyMapInitOnce.reset(); + return true; +} + +U_CDECL_END + + +static void U_CALLCONV +initFromResourceBundle(UErrorCode& sts) { + U_NAMESPACE_USE + ucln_common_registerCleanup(UCLN_COMMON_LOCALE_KEY_TYPE, uloc_key_type_cleanup); + + gLocExtKeyMap = uhash_open(uhash_hashIChars, uhash_compareIChars, nullptr, &sts); + + LocalUResourceBundlePointer keyTypeDataRes(ures_openDirect(nullptr, "keyTypeData", &sts)); + LocalUResourceBundlePointer keyMapRes(ures_getByKey(keyTypeDataRes.getAlias(), "keyMap", nullptr, &sts)); + LocalUResourceBundlePointer typeMapRes(ures_getByKey(keyTypeDataRes.getAlias(), "typeMap", nullptr, &sts)); + + if (U_FAILURE(sts)) { + return; + } + + UErrorCode tmpSts = U_ZERO_ERROR; + LocalUResourceBundlePointer typeAliasRes(ures_getByKey(keyTypeDataRes.getAlias(), "typeAlias", nullptr, &tmpSts)); + tmpSts = U_ZERO_ERROR; + LocalUResourceBundlePointer bcpTypeAliasRes(ures_getByKey(keyTypeDataRes.getAlias(), "bcpTypeAlias", nullptr, &tmpSts)); + + // initialize pools storing dynamically allocated objects + gKeyTypeStringPool = new icu::MemoryPool; + if (gKeyTypeStringPool == nullptr) { + sts = U_MEMORY_ALLOCATION_ERROR; + return; + } + gLocExtKeyDataEntries = new icu::MemoryPool; + if (gLocExtKeyDataEntries == nullptr) { + sts = U_MEMORY_ALLOCATION_ERROR; + return; + } + gLocExtTypeEntries = new icu::MemoryPool; + if (gLocExtTypeEntries == nullptr) { + sts = U_MEMORY_ALLOCATION_ERROR; + return; + } + + // iterate through keyMap resource + LocalUResourceBundlePointer keyMapEntry; + + while (ures_hasNext(keyMapRes.getAlias())) { + keyMapEntry.adoptInstead(ures_getNextResource(keyMapRes.getAlias(), keyMapEntry.orphan(), &sts)); + if (U_FAILURE(sts)) { + break; + } + const char* legacyKeyId = ures_getKey(keyMapEntry.getAlias()); + UnicodeString uBcpKeyId = ures_getUnicodeString(keyMapEntry.getAlias(), &sts); + if (U_FAILURE(sts)) { + break; + } + + // empty value indicates that BCP key is same with the legacy key. + const char* bcpKeyId = legacyKeyId; + if (!uBcpKeyId.isEmpty()) { + icu::CharString* bcpKeyIdBuf = gKeyTypeStringPool->create(); + if (bcpKeyIdBuf == nullptr) { + sts = U_MEMORY_ALLOCATION_ERROR; + break; + } + bcpKeyIdBuf->appendInvariantChars(uBcpKeyId, sts); + if (U_FAILURE(sts)) { + break; + } + bcpKeyId = bcpKeyIdBuf->data(); + } + + UBool isTZ = uprv_strcmp(legacyKeyId, "timezone") == 0; + + UHashtable* typeDataMap = uhash_open(uhash_hashIChars, uhash_compareIChars, nullptr, &sts); + if (U_FAILURE(sts)) { + break; + } + uint32_t specialTypes = SPECIALTYPE_NONE; + + LocalUResourceBundlePointer typeAliasResByKey; + LocalUResourceBundlePointer bcpTypeAliasResByKey; + + if (typeAliasRes.isValid()) { + tmpSts = U_ZERO_ERROR; + typeAliasResByKey.adoptInstead(ures_getByKey(typeAliasRes.getAlias(), legacyKeyId, nullptr, &tmpSts)); + if (U_FAILURE(tmpSts)) { + typeAliasResByKey.orphan(); + } + } + if (bcpTypeAliasRes.isValid()) { + tmpSts = U_ZERO_ERROR; + bcpTypeAliasResByKey.adoptInstead(ures_getByKey(bcpTypeAliasRes.getAlias(), bcpKeyId, nullptr, &tmpSts)); + if (U_FAILURE(tmpSts)) { + bcpTypeAliasResByKey.orphan(); + } + } + + // look up type map for the key, and walk through the mapping data + LocalUResourceBundlePointer typeMapResByKey(ures_getByKey(typeMapRes.getAlias(), legacyKeyId, nullptr, &sts)); + if (U_FAILURE(sts)) { + // We fail here if typeMap does not have an entry corresponding to every entry in keyMap (should + // not happen for valid keyTypeData), or if ures_getByKeyfails fails for some other reason + // (e.g. data file cannot be loaded, using stubdata, over-aggressive data filtering has removed + // something like timezoneTypes.res, etc.). Error code is already set. See ICU-21669. + UPRV_UNREACHABLE_ASSERT; + } else { + LocalUResourceBundlePointer typeMapEntry; + + while (ures_hasNext(typeMapResByKey.getAlias())) { + typeMapEntry.adoptInstead(ures_getNextResource(typeMapResByKey.getAlias(), typeMapEntry.orphan(), &sts)); + if (U_FAILURE(sts)) { + break; + } + const char* legacyTypeId = ures_getKey(typeMapEntry.getAlias()); + + // special types + if (uprv_strcmp(legacyTypeId, "CODEPOINTS") == 0) { + specialTypes |= SPECIALTYPE_CODEPOINTS; + continue; + } + if (uprv_strcmp(legacyTypeId, "REORDER_CODE") == 0) { + specialTypes |= SPECIALTYPE_REORDER_CODE; + continue; + } + if (uprv_strcmp(legacyTypeId, "RG_KEY_VALUE") == 0) { + specialTypes |= SPECIALTYPE_RG_KEY_VALUE; + continue; + } + + if (isTZ) { + // a timezone key uses a colon instead of a slash in the resource. + // e.g. America:Los_Angeles + if (uprv_strchr(legacyTypeId, ':') != nullptr) { + icu::CharString* legacyTypeIdBuf = + gKeyTypeStringPool->create(legacyTypeId, sts); + if (legacyTypeIdBuf == nullptr) { + sts = U_MEMORY_ALLOCATION_ERROR; + break; + } + if (U_FAILURE(sts)) { + break; + } + std::replace( + legacyTypeIdBuf->data(), + legacyTypeIdBuf->data() + legacyTypeIdBuf->length(), + ':', '/'); + legacyTypeId = legacyTypeIdBuf->data(); + } + } + + UnicodeString uBcpTypeId = ures_getUnicodeString(typeMapEntry.getAlias(), &sts); + if (U_FAILURE(sts)) { + break; + } + + // empty value indicates that BCP type is same with the legacy type. + const char* bcpTypeId = legacyTypeId; + if (!uBcpTypeId.isEmpty()) { + icu::CharString* bcpTypeIdBuf = gKeyTypeStringPool->create(); + if (bcpTypeIdBuf == nullptr) { + sts = U_MEMORY_ALLOCATION_ERROR; + break; + } + bcpTypeIdBuf->appendInvariantChars(uBcpTypeId, sts); + if (U_FAILURE(sts)) { + break; + } + bcpTypeId = bcpTypeIdBuf->data(); + } + + // Note: legacy type value should never be + // equivalent to bcp type value of a different + // type under the same key. So we use a single + // map for lookup. + LocExtType* t = gLocExtTypeEntries->create(); + if (t == nullptr) { + sts = U_MEMORY_ALLOCATION_ERROR; + break; + } + t->bcpId = bcpTypeId; + t->legacyId = legacyTypeId; + + uhash_put(typeDataMap, (void*)legacyTypeId, t, &sts); + if (bcpTypeId != legacyTypeId) { + // different type value + uhash_put(typeDataMap, (void*)bcpTypeId, t, &sts); + } + if (U_FAILURE(sts)) { + break; + } + + // also put aliases in the map + if (typeAliasResByKey.isValid()) { + LocalUResourceBundlePointer typeAliasDataEntry; + + ures_resetIterator(typeAliasResByKey.getAlias()); + while (ures_hasNext(typeAliasResByKey.getAlias()) && U_SUCCESS(sts)) { + int32_t toLen; + typeAliasDataEntry.adoptInstead(ures_getNextResource(typeAliasResByKey.getAlias(), typeAliasDataEntry.orphan(), &sts)); + const char16_t* to = ures_getString(typeAliasDataEntry.getAlias(), &toLen, &sts); + if (U_FAILURE(sts)) { + break; + } + // check if this is an alias of canonical legacy type + if (uprv_compareInvWithUChar(nullptr, legacyTypeId, -1, to, toLen) == 0) { + const char* from = ures_getKey(typeAliasDataEntry.getAlias()); + if (isTZ) { + // replace colon with slash if necessary + if (uprv_strchr(from, ':') != nullptr) { + icu::CharString* fromBuf = + gKeyTypeStringPool->create(from, sts); + if (fromBuf == nullptr) { + sts = U_MEMORY_ALLOCATION_ERROR; + break; + } + if (U_FAILURE(sts)) { + break; + } + std::replace( + fromBuf->data(), + fromBuf->data() + fromBuf->length(), + ':', '/'); + from = fromBuf->data(); + } + } + uhash_put(typeDataMap, (void*)from, t, &sts); + } + } + if (U_FAILURE(sts)) { + break; + } + } + + if (bcpTypeAliasResByKey.isValid()) { + LocalUResourceBundlePointer bcpTypeAliasDataEntry; + + ures_resetIterator(bcpTypeAliasResByKey.getAlias()); + while (ures_hasNext(bcpTypeAliasResByKey.getAlias()) && U_SUCCESS(sts)) { + int32_t toLen; + bcpTypeAliasDataEntry.adoptInstead(ures_getNextResource(bcpTypeAliasResByKey.getAlias(), bcpTypeAliasDataEntry.orphan(), &sts)); + const char16_t* to = ures_getString(bcpTypeAliasDataEntry.getAlias(), &toLen, &sts); + if (U_FAILURE(sts)) { + break; + } + // check if this is an alias of bcp type + if (uprv_compareInvWithUChar(nullptr, bcpTypeId, -1, to, toLen) == 0) { + const char* from = ures_getKey(bcpTypeAliasDataEntry.getAlias()); + uhash_put(typeDataMap, (void*)from, t, &sts); + } + } + if (U_FAILURE(sts)) { + break; + } + } + } + } + if (U_FAILURE(sts)) { + break; + } + + LocExtKeyData* keyData = gLocExtKeyDataEntries->create(); + if (keyData == nullptr) { + sts = U_MEMORY_ALLOCATION_ERROR; + break; + } + keyData->bcpId = bcpKeyId; + keyData->legacyId = legacyKeyId; + keyData->specialTypes = specialTypes; + keyData->typeMap.adoptInstead(typeDataMap); + + uhash_put(gLocExtKeyMap, (void*)legacyKeyId, keyData, &sts); + if (legacyKeyId != bcpKeyId) { + // different key value + uhash_put(gLocExtKeyMap, (void*)bcpKeyId, keyData, &sts); + } + if (U_FAILURE(sts)) { + break; + } + } +} + +static UBool +init() { + UErrorCode sts = U_ZERO_ERROR; + umtx_initOnce(gLocExtKeyMapInitOnce, &initFromResourceBundle, sts); + if (U_FAILURE(sts)) { + return false; + } + return true; +} + +static UBool +isSpecialTypeCodepoints(const char* val) { + int32_t subtagLen = 0; + const char* p = val; + while (*p) { + if (*p == '-') { + if (subtagLen < 4 || subtagLen > 6) { + return false; + } + subtagLen = 0; + } else if ((*p >= '0' && *p <= '9') || + (*p >= 'A' && *p <= 'F') || // A-F/a-f are contiguous + (*p >= 'a' && *p <= 'f')) { // also in EBCDIC + subtagLen++; + } else { + return false; + } + p++; + } + return (subtagLen >= 4 && subtagLen <= 6); +} + +static UBool +isSpecialTypeReorderCode(const char* val) { + int32_t subtagLen = 0; + const char* p = val; + while (*p) { + if (*p == '-') { + if (subtagLen < 3 || subtagLen > 8) { + return false; + } + subtagLen = 0; + } else if (uprv_isASCIILetter(*p)) { + subtagLen++; + } else { + return false; + } + p++; + } + return (subtagLen >=3 && subtagLen <=8); +} + +static UBool +isSpecialTypeRgKeyValue(const char* val) { + int32_t subtagLen = 0; + const char* p = val; + while (*p) { + if ( (subtagLen < 2 && uprv_isASCIILetter(*p)) || + (subtagLen >= 2 && (*p == 'Z' || *p == 'z')) ) { + subtagLen++; + } else { + return false; + } + p++; + } + return (subtagLen == 6); +} + +U_CFUNC const char* +ulocimp_toBcpKey(const char* key) { + if (!init()) { + return nullptr; + } + + LocExtKeyData* keyData = (LocExtKeyData*)uhash_get(gLocExtKeyMap, key); + if (keyData != nullptr) { + return keyData->bcpId; + } + return nullptr; +} + +U_CFUNC const char* +ulocimp_toLegacyKey(const char* key) { + if (!init()) { + return nullptr; + } + + LocExtKeyData* keyData = (LocExtKeyData*)uhash_get(gLocExtKeyMap, key); + if (keyData != nullptr) { + return keyData->legacyId; + } + return nullptr; +} + +U_CFUNC const char* +ulocimp_toBcpType(const char* key, const char* type, UBool* isKnownKey, UBool* isSpecialType) { + if (isKnownKey != nullptr) { + *isKnownKey = false; + } + if (isSpecialType != nullptr) { + *isSpecialType = false; + } + + if (!init()) { + return nullptr; + } + + LocExtKeyData* keyData = (LocExtKeyData*)uhash_get(gLocExtKeyMap, key); + if (keyData != nullptr) { + if (isKnownKey != nullptr) { + *isKnownKey = true; + } + LocExtType* t = (LocExtType*)uhash_get(keyData->typeMap.getAlias(), type); + if (t != nullptr) { + return t->bcpId; + } + if (keyData->specialTypes != SPECIALTYPE_NONE) { + UBool matched = false; + if (keyData->specialTypes & SPECIALTYPE_CODEPOINTS) { + matched = isSpecialTypeCodepoints(type); + } + if (!matched && keyData->specialTypes & SPECIALTYPE_REORDER_CODE) { + matched = isSpecialTypeReorderCode(type); + } + if (!matched && keyData->specialTypes & SPECIALTYPE_RG_KEY_VALUE) { + matched = isSpecialTypeRgKeyValue(type); + } + if (matched) { + if (isSpecialType != nullptr) { + *isSpecialType = true; + } + return type; + } + } + } + return nullptr; +} + + +U_CFUNC const char* +ulocimp_toLegacyType(const char* key, const char* type, UBool* isKnownKey, UBool* isSpecialType) { + if (isKnownKey != nullptr) { + *isKnownKey = false; + } + if (isSpecialType != nullptr) { + *isSpecialType = false; + } + + if (!init()) { + return nullptr; + } + + LocExtKeyData* keyData = (LocExtKeyData*)uhash_get(gLocExtKeyMap, key); + if (keyData != nullptr) { + if (isKnownKey != nullptr) { + *isKnownKey = true; + } + LocExtType* t = (LocExtType*)uhash_get(keyData->typeMap.getAlias(), type); + if (t != nullptr) { + return t->legacyId; + } + if (keyData->specialTypes != SPECIALTYPE_NONE) { + UBool matched = false; + if (keyData->specialTypes & SPECIALTYPE_CODEPOINTS) { + matched = isSpecialTypeCodepoints(type); + } + if (!matched && keyData->specialTypes & SPECIALTYPE_REORDER_CODE) { + matched = isSpecialTypeReorderCode(type); + } + if (!matched && keyData->specialTypes & SPECIALTYPE_RG_KEY_VALUE) { + matched = isSpecialTypeRgKeyValue(type); + } + if (matched) { + if (isSpecialType != nullptr) { + *isSpecialType = true; + } + return type; + } + } + } + return nullptr; +} + diff --git a/intl/icu/source/common/uloc_tag.cpp b/intl/icu/source/common/uloc_tag.cpp new file mode 100644 index 0000000000..43d597549f --- /dev/null +++ b/intl/icu/source/common/uloc_tag.cpp @@ -0,0 +1,2869 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2009-2015, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ + +#include "unicode/bytestream.h" +#include "unicode/utypes.h" +#include "unicode/ures.h" +#include "unicode/localpointer.h" +#include "unicode/putil.h" +#include "unicode/uenum.h" +#include "unicode/uloc.h" +#include "ustr_imp.h" +#include "bytesinkutil.h" +#include "charstr.h" +#include "cmemory.h" +#include "cstring.h" +#include "putilimp.h" +#include "uinvchar.h" +#include "ulocimp.h" +#include "uassert.h" + + +/* struct holding a single variant */ +typedef struct VariantListEntry { + const char *variant; + struct VariantListEntry *next; +} VariantListEntry; + +/* struct holding a single attribute value */ +struct AttributeListEntry : public icu::UMemory { + const char *attribute; + struct AttributeListEntry *next; +}; + +/* struct holding a single extension */ +struct ExtensionListEntry : public icu::UMemory { + const char *key; + const char *value; + struct ExtensionListEntry *next; +}; + +#define MAXEXTLANG 3 +typedef struct ULanguageTag { + char *buf; /* holding parsed subtags */ + const char *language; + const char *extlang[MAXEXTLANG]; + const char *script; + const char *region; + VariantListEntry *variants; + ExtensionListEntry *extensions; + const char *privateuse; + const char *legacy; +} ULanguageTag; + +#define MINLEN 2 +#define SEP '-' +#define PRIVATEUSE 'x' +#define LDMLEXT 'u' + +#define LOCALE_SEP '_' +#define LOCALE_EXT_SEP '@' +#define LOCALE_KEYWORD_SEP ';' +#define LOCALE_KEY_TYPE_SEP '=' + +#define ISALPHA(c) uprv_isASCIILetter(c) +#define ISNUMERIC(c) ((c)>='0' && (c)<='9') + +static const char EMPTY[] = ""; +static const char LANG_UND[] = "und"; +static const char PRIVATEUSE_KEY[] = "x"; +static const char _POSIX[] = "_POSIX"; +static const char POSIX_KEY[] = "va"; +static const char POSIX_VALUE[] = "posix"; +static const char LOCALE_ATTRIBUTE_KEY[] = "attribute"; +static const char PRIVUSE_VARIANT_PREFIX[] = "lvariant"; +static const char LOCALE_TYPE_YES[] = "yes"; + +#define LANG_UND_LEN 3 + +/* + Updated on 2018-09-12 from + https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry . + + This table has 2 parts. The part for + legacy language tags (marked as “Type: grandfathered” in BCP 47) + is generated by the following scripts from the IANA language tag registry. + + curl https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry |\ + egrep -A 7 'Type: grandfathered' | \ + egrep 'Tag|Prefe' | grep -B1 'Preferred' | grep -v '^--' | \ + awk -n '/Tag/ {printf(" \"%s\", ", $2);} /Preferred/ {printf("\"%s\",\n", $2);}' |\ + tr 'A-Z' 'a-z' + + + The 2nd part is made of five ICU-specific entries. They're kept for + the backward compatibility for now, even though there are no preferred + values. They may have to be removed for the strict BCP 47 compliance. + +*/ +static const char* const LEGACY[] = { +/* legacy preferred */ + "art-lojban", "jbo", + "en-gb-oed", "en-gb-oxendict", + "i-ami", "ami", + "i-bnn", "bnn", + "i-hak", "hak", + "i-klingon", "tlh", + "i-lux", "lb", + "i-navajo", "nv", + "i-pwn", "pwn", + "i-tao", "tao", + "i-tay", "tay", + "i-tsu", "tsu", + "no-bok", "nb", + "no-nyn", "nn", + "sgn-be-fr", "sfb", + "sgn-be-nl", "vgt", + "sgn-ch-de", "sgg", + "zh-guoyu", "cmn", + "zh-hakka", "hak", + "zh-min-nan", "nan", + "zh-xiang", "hsn", + + // Legacy tags with no preferred value in the IANA + // registry. Kept for now for the backward compatibility + // because ICU has mapped them this way. + "i-default", "en-x-i-default", + "i-enochian", "und-x-i-enochian", + "i-mingo", "see-x-i-mingo", + "zh-min", "nan-x-zh-min", +}; + +/* + Updated on 2018-09-12 from + https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry . + + The table lists redundant tags with preferred value in the IANA language tag registry. + It's generated with the following command: + + curl https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry |\ + grep 'Type: redundant' -A 5 | egrep '^(Tag:|Prefer)' | grep -B1 'Preferred' | \ + awk -n '/Tag/ {printf(" \"%s\", ", $2);} /Preferred/ {printf("\"%s\",\n", $2);}' | \ + tr 'A-Z' 'a-z' + + In addition, ja-latn-hepburn-heploc is mapped to ja-latn-alalc97 because + a variant tag 'hepburn-heploc' has the preferred subtag, 'alaic97'. +*/ + +static const char* const REDUNDANT[] = { +// redundant preferred + "sgn-br", "bzs", + "sgn-co", "csn", + "sgn-de", "gsg", + "sgn-dk", "dsl", + "sgn-es", "ssp", + "sgn-fr", "fsl", + "sgn-gb", "bfi", + "sgn-gr", "gss", + "sgn-ie", "isg", + "sgn-it", "ise", + "sgn-jp", "jsl", + "sgn-mx", "mfs", + "sgn-ni", "ncs", + "sgn-nl", "dse", + "sgn-no", "nsl", + "sgn-pt", "psr", + "sgn-se", "swl", + "sgn-us", "ase", + "sgn-za", "sfs", + "zh-cmn", "cmn", + "zh-cmn-hans", "cmn-hans", + "zh-cmn-hant", "cmn-hant", + "zh-gan", "gan", + "zh-wuu", "wuu", + "zh-yue", "yue", + + // variant tag with preferred value + "ja-latn-hepburn-heploc", "ja-latn-alalc97", +}; + +/* + Updated on 2018-09-12 from + https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry . + + grep 'Type: language' -A 7 language-subtag-registry | egrep 'Subtag|Prefe' | \ + grep -B1 'Preferred' | grep -v '^--' | \ + awk -n '/Subtag/ {printf(" \"%s\", ", $2);} /Preferred/ {printf("\"%s\",\n", $2);}' + + Make sure that 2-letter language subtags come before 3-letter subtags. +*/ +static const char DEPRECATEDLANGS[][4] = { +/* deprecated new */ + "in", "id", + "iw", "he", + "ji", "yi", + "jw", "jv", + "mo", "ro", + "aam", "aas", + "adp", "dz", + "aue", "ktz", + "ayx", "nun", + "bgm", "bcg", + "bjd", "drl", + "ccq", "rki", + "cjr", "mom", + "cka", "cmr", + "cmk", "xch", + "coy", "pij", + "cqu", "quh", + "drh", "khk", + "drw", "prs", + "gav", "dev", + "gfx", "vaj", + "ggn", "gvr", + "gti", "nyc", + "guv", "duz", + "hrr", "jal", + "ibi", "opa", + "ilw", "gal", + "jeg", "oyb", + "kgc", "tdf", + "kgh", "kml", + "koj", "kwv", + "krm", "bmf", + "ktr", "dtp", + "kvs", "gdj", + "kwq", "yam", + "kxe", "tvd", + "kzj", "dtp", + "kzt", "dtp", + "lii", "raq", + "lmm", "rmx", + "meg", "cir", + "mst", "mry", + "mwj", "vaj", + "myt", "mry", + "nad", "xny", + "ncp", "kdz", + "nnx", "ngv", + "nts", "pij", + "oun", "vaj", + "pcr", "adx", + "pmc", "huw", + "pmu", "phr", + "ppa", "bfy", + "ppr", "lcq", + "pry", "prt", + "puz", "pub", + "sca", "hle", + "skk", "oyb", + "tdu", "dtp", + "thc", "tpo", + "thx", "oyb", + "tie", "ras", + "tkk", "twm", + "tlw", "weo", + "tmp", "tyj", + "tne", "kak", + "tnf", "prs", + "tsf", "taj", + "uok", "ema", + "xba", "cax", + "xia", "acn", + "xkh", "waw", + "xsj", "suj", + "ybd", "rki", + "yma", "lrr", + "ymt", "mtm", + "yos", "zom", + "yuu", "yug", +}; + +/* + Updated on 2018-04-24 from + + curl https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry | \ + grep 'Type: region' -A 7 | egrep 'Subtag|Prefe' | \ + grep -B1 'Preferred' | \ + awk -n '/Subtag/ {printf(" \"%s\", ", $2);} /Preferred/ {printf("\"%s\",\n", $2);}' +*/ +static const char DEPRECATEDREGIONS[][3] = { +/* deprecated new */ + "BU", "MM", + "DD", "DE", + "FX", "FR", + "TP", "TL", + "YD", "YE", + "ZR", "CD", +}; + +/* +* ------------------------------------------------- +* +* These ultag_ functions may be exposed as APIs later +* +* ------------------------------------------------- +*/ + +static ULanguageTag* +ultag_parse(const char* tag, int32_t tagLen, int32_t* parsedLen, UErrorCode* status); + +static void +ultag_close(ULanguageTag* langtag); + +static const char* +ultag_getLanguage(const ULanguageTag* langtag); + +#if 0 +static const char* +ultag_getJDKLanguage(const ULanguageTag* langtag); +#endif + +static const char* +ultag_getExtlang(const ULanguageTag* langtag, int32_t idx); + +static int32_t +ultag_getExtlangSize(const ULanguageTag* langtag); + +static const char* +ultag_getScript(const ULanguageTag* langtag); + +static const char* +ultag_getRegion(const ULanguageTag* langtag); + +static const char* +ultag_getVariant(const ULanguageTag* langtag, int32_t idx); + +static int32_t +ultag_getVariantsSize(const ULanguageTag* langtag); + +static const char* +ultag_getExtensionKey(const ULanguageTag* langtag, int32_t idx); + +static const char* +ultag_getExtensionValue(const ULanguageTag* langtag, int32_t idx); + +static int32_t +ultag_getExtensionsSize(const ULanguageTag* langtag); + +static const char* +ultag_getPrivateUse(const ULanguageTag* langtag); + +#if 0 +static const char* +ultag_getLegacy(const ULanguageTag* langtag); +#endif + +U_NAMESPACE_BEGIN + +/** + * \class LocalULanguageTagPointer + * "Smart pointer" class, closes a ULanguageTag via ultag_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @internal + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalULanguageTagPointer, ULanguageTag, ultag_close); + +U_NAMESPACE_END + +/* +* ------------------------------------------------- +* +* Language subtag syntax validation functions +* +* ------------------------------------------------- +*/ + +static UBool +_isAlphaString(const char* s, int32_t len) { + int32_t i; + for (i = 0; i < len; i++) { + if (!ISALPHA(*(s + i))) { + return false; + } + } + return true; +} + +static UBool +_isNumericString(const char* s, int32_t len) { + int32_t i; + for (i = 0; i < len; i++) { + if (!ISNUMERIC(*(s + i))) { + return false; + } + } + return true; +} + +static UBool +_isAlphaNumericString(const char* s, int32_t len) { + int32_t i; + for (i = 0; i < len; i++) { + if (!ISALPHA(*(s + i)) && !ISNUMERIC(*(s + i))) { + return false; + } + } + return true; +} + +static UBool +_isAlphaNumericStringLimitedLength(const char* s, int32_t len, int32_t min, int32_t max) { + if (len < 0) { + len = (int32_t)uprv_strlen(s); + } + if (len >= min && len <= max && _isAlphaNumericString(s, len)) { + return true; + } + return false; +} + +U_CFUNC UBool +ultag_isLanguageSubtag(const char* s, int32_t len) { + /* + * unicode_language_subtag = alpha{2,3} | alpha{5,8}; + * NOTE: Per ICUTC 2019/01/23- accepting alpha 4 + * See ICU-20372 + */ + if (len < 0) { + len = (int32_t)uprv_strlen(s); + } + if (len >= 2 && len <= 8 && _isAlphaString(s, len)) { + return true; + } + return false; +} + +static UBool +_isExtlangSubtag(const char* s, int32_t len) { + /* + * extlang = 3ALPHA ; selected ISO 639 codes + * *2("-" 3ALPHA) ; permanently reserved + */ + if (len < 0) { + len = (int32_t)uprv_strlen(s); + } + if (len == 3 && _isAlphaString(s, len)) { + return true; + } + return false; +} + +U_CFUNC UBool +ultag_isScriptSubtag(const char* s, int32_t len) { + /* + * script = 4ALPHA ; ISO 15924 code + */ + if (len < 0) { + len = (int32_t)uprv_strlen(s); + } + if (len == 4 && _isAlphaString(s, len)) { + return true; + } + return false; +} + +U_CFUNC UBool +ultag_isRegionSubtag(const char* s, int32_t len) { + /* + * region = 2ALPHA ; ISO 3166-1 code + * / 3DIGIT ; UN M.49 code + */ + if (len < 0) { + len = (int32_t)uprv_strlen(s); + } + if (len == 2 && _isAlphaString(s, len)) { + return true; + } + if (len == 3 && _isNumericString(s, len)) { + return true; + } + return false; +} + +static UBool +_isVariantSubtag(const char* s, int32_t len) { + /* + * variant = 5*8alphanum ; registered variants + * / (DIGIT 3alphanum) + */ + if (len < 0) { + len = (int32_t)uprv_strlen(s); + } + if (_isAlphaNumericStringLimitedLength(s, len, 5, 8)) { + return true; + } + if (len == 4 && ISNUMERIC(*s) && _isAlphaNumericString(s + 1, 3)) { + return true; + } + return false; +} + +static UBool +_isSepListOf(UBool (*test)(const char*, int32_t), const char* s, int32_t len) { + const char *p = s; + const char *pSubtag = nullptr; + + if (len < 0) { + len = (int32_t)uprv_strlen(s); + } + + while ((p - s) < len) { + if (*p == SEP) { + if (pSubtag == nullptr) { + return false; + } + if (!test(pSubtag, (int32_t)(p - pSubtag))) { + return false; + } + pSubtag = nullptr; + } else if (pSubtag == nullptr) { + pSubtag = p; + } + p++; + } + if (pSubtag == nullptr) { + return false; + } + return test(pSubtag, (int32_t)(p - pSubtag)); +} + +U_CFUNC UBool +ultag_isVariantSubtags(const char* s, int32_t len) { + return _isSepListOf(&_isVariantSubtag, s, len); +} + +// This is for the ICU-specific "lvariant" handling. +static UBool +_isPrivateuseVariantSubtag(const char* s, int32_t len) { + /* + * variant = 1*8alphanum ; registered variants + * / (DIGIT 3alphanum) + */ + return _isAlphaNumericStringLimitedLength(s, len , 1, 8); +} + +static UBool +_isExtensionSingleton(const char* s, int32_t len) { + /* + * extension = singleton 1*("-" (2*8alphanum)) + * + * singleton = DIGIT ; 0 - 9 + * / %x41-57 ; A - W + * / %x59-5A ; Y - Z + * / %x61-77 ; a - w + * / %x79-7A ; y - z + */ + if (len < 0) { + len = (int32_t)uprv_strlen(s); + } + if (len == 1 && (ISALPHA(*s) || ISNUMERIC(*s)) && (uprv_tolower(*s) != PRIVATEUSE)) { + return true; + } + return false; +} + +static UBool +_isExtensionSubtag(const char* s, int32_t len) { + /* + * extension = singleton 1*("-" (2*8alphanum)) + */ + return _isAlphaNumericStringLimitedLength(s, len, 2, 8); +} + +U_CFUNC UBool +ultag_isExtensionSubtags(const char* s, int32_t len) { + return _isSepListOf(&_isExtensionSubtag, s, len); +} + +static UBool +_isPrivateuseValueSubtag(const char* s, int32_t len) { + /* + * privateuse = "x" 1*("-" (1*8alphanum)) + */ + return _isAlphaNumericStringLimitedLength(s, len, 1, 8); +} + +U_CFUNC UBool +ultag_isPrivateuseValueSubtags(const char* s, int32_t len) { + return _isSepListOf(&_isPrivateuseValueSubtag, s, len); +} + +U_CFUNC UBool +ultag_isUnicodeLocaleAttribute(const char* s, int32_t len) { + /* + * attribute = alphanum{3,8} ; + */ + return _isAlphaNumericStringLimitedLength(s, len , 3, 8); +} + +U_CFUNC UBool +ultag_isUnicodeLocaleAttributes(const char* s, int32_t len) { + return _isSepListOf(&ultag_isUnicodeLocaleAttribute, s, len); +} + +U_CFUNC UBool +ultag_isUnicodeLocaleKey(const char* s, int32_t len) { + /* + * key = alphanum alpha ; + */ + if (len < 0) { + len = (int32_t)uprv_strlen(s); + } + if (len == 2 && (ISALPHA(*s) || ISNUMERIC(*s)) && ISALPHA(s[1])) { + return true; + } + return false; +} + +U_CFUNC UBool +_isUnicodeLocaleTypeSubtag(const char*s, int32_t len) { + /* + * alphanum{3,8} + */ + return _isAlphaNumericStringLimitedLength(s, len , 3, 8); +} + +U_CFUNC UBool +ultag_isUnicodeLocaleType(const char*s, int32_t len) { + /* + * type = alphanum{3,8} (sep alphanum{3,8})* ; + */ + return _isSepListOf(&_isUnicodeLocaleTypeSubtag, s, len); +} + +static UBool +_isTKey(const char* s, int32_t len) +{ + /* + * tkey = alpha digit ; + */ + if (len < 0) { + len = (int32_t)uprv_strlen(s); + } + if (len == 2 && ISALPHA(*s) && ISNUMERIC(*(s + 1))) { + return true; + } + return false; +} + +U_CAPI const char * U_EXPORT2 +ultag_getTKeyStart(const char *localeID) { + const char *result = localeID; + const char *sep; + while((sep = uprv_strchr(result, SEP)) != nullptr) { + if (_isTKey(result, static_cast(sep - result))) { + return result; + } + result = ++sep; + } + if (_isTKey(result, -1)) { + return result; + } + return nullptr; +} + +static UBool +_isTValue(const char* s, int32_t len) +{ + /* + * tvalue = (sep alphanum{3,8})+ ; + */ + return _isAlphaNumericStringLimitedLength(s, len , 3, 8); +} + +static UBool +_isTransformedExtensionSubtag(int32_t& state, const char* s, int32_t len) +{ + const int32_t kStart = 0; // Start, wait for unicode_language_subtag, tkey or end + const int32_t kGotLanguage = 1; // Got unicode_language_subtag, wait for unicode_script_subtag, + // unicode_region_subtag, unicode_variant_subtag, tkey or end + const int32_t kGotScript = 2; // Got unicode_script_subtag, wait for unicode_region_subtag, + // unicode_variant_subtag, tkey, or end + const int32_t kGotRegion = 3; // Got unicode_region_subtag, wait for unicode_variant_subtag, + // tkey, or end. + const int32_t kGotVariant = 4; // Got unicode_variant_subtag, wait for unicode_variant_subtag + // tkey or end. + const int32_t kGotTKey = -1; // Got tkey, wait for tvalue. ERROR if stop here. + const int32_t kGotTValue = 6; // Got tvalue, wait for tkey, tvalue or end + + + if (len < 0) { + len = (int32_t)uprv_strlen(s); + } + switch (state) { + case kStart: + if (ultag_isLanguageSubtag(s, len) && len != 4) { + state = kGotLanguage; + return true; + } + if (_isTKey(s, len)) { + state = kGotTKey; + return true; + } + return false; + case kGotLanguage: + if (ultag_isScriptSubtag(s, len)) { + state = kGotScript; + return true; + } + U_FALLTHROUGH; + case kGotScript: + if (ultag_isRegionSubtag(s, len)) { + state = kGotRegion; + return true; + } + U_FALLTHROUGH; + case kGotRegion: + U_FALLTHROUGH; + case kGotVariant: + if (_isVariantSubtag(s, len)) { + state = kGotVariant; + return true; + } + if (_isTKey(s, len)) { + state = kGotTKey; + return true; + } + return false; + case kGotTKey: + if (_isTValue(s, len)) { + state = kGotTValue; + return true; + } + return false; + case kGotTValue: + if (_isTKey(s, len)) { + state = kGotTKey; + return true; + } + if (_isTValue(s, len)) { + return true; + } + return false; + } + return false; +} + +static UBool +_isUnicodeExtensionSubtag(int32_t& state, const char* s, int32_t len) +{ + const int32_t kStart = 0; // Start, wait for a key or attribute or end + const int32_t kGotKey = 1; // Got a key, wait for type or key or end + const int32_t kGotType = 2; // Got a type, wait for key or end + + switch (state) { + case kStart: + if (ultag_isUnicodeLocaleKey(s, len)) { + state = kGotKey; + return true; + } + if (ultag_isUnicodeLocaleAttribute(s, len)) { + return true; + } + return false; + case kGotKey: + if (ultag_isUnicodeLocaleKey(s, len)) { + return true; + } + if (_isUnicodeLocaleTypeSubtag(s, len)) { + state = kGotType; + return true; + } + return false; + case kGotType: + if (ultag_isUnicodeLocaleKey(s, len)) { + state = kGotKey; + return true; + } + if (_isUnicodeLocaleTypeSubtag(s, len)) { + return true; + } + return false; + } + return false; +} + +static UBool +_isStatefulSepListOf(UBool (*test)(int32_t&, const char*, int32_t), const char* s, int32_t len) +{ + int32_t state = 0; + const char* p; + const char* start = s; + int32_t subtagLen = 0; + + if (len < 0) { + len = (int32_t)uprv_strlen(s); + } + + for (p = s; len > 0; p++, len--) { + if (*p == SEP) { + if (!test(state, start, subtagLen)) { + return false; + } + subtagLen = 0; + start = p + 1; + } else { + subtagLen++; + } + } + + if (test(state, start, subtagLen) && state >= 0) { + return true; + } + return false; +} + +U_CFUNC UBool +ultag_isTransformedExtensionSubtags(const char* s, int32_t len) +{ + return _isStatefulSepListOf(&_isTransformedExtensionSubtag, s, len); +} + +U_CFUNC UBool +ultag_isUnicodeExtensionSubtags(const char* s, int32_t len) { + return _isStatefulSepListOf(&_isUnicodeExtensionSubtag, s, len); +} + + +/* +* ------------------------------------------------- +* +* Helper functions +* +* ------------------------------------------------- +*/ + +static UBool +_addVariantToList(VariantListEntry **first, VariantListEntry *var) { + UBool bAdded = true; + + if (*first == nullptr) { + var->next = nullptr; + *first = var; + } else { + VariantListEntry *prev, *cur; + int32_t cmp; + + /* variants order should be preserved */ + prev = nullptr; + cur = *first; + while (true) { + if (cur == nullptr) { + prev->next = var; + var->next = nullptr; + break; + } + + /* Checking for duplicate variant */ + cmp = uprv_compareInvCharsAsAscii(var->variant, cur->variant); + if (cmp == 0) { + /* duplicated variant */ + bAdded = false; + break; + } + prev = cur; + cur = cur->next; + } + } + + return bAdded; +} + +static UBool +_addAttributeToList(AttributeListEntry **first, AttributeListEntry *attr) { + UBool bAdded = true; + + if (*first == nullptr) { + attr->next = nullptr; + *first = attr; + } else { + AttributeListEntry *prev, *cur; + int32_t cmp; + + /* reorder variants in alphabetical order */ + prev = nullptr; + cur = *first; + while (true) { + if (cur == nullptr) { + prev->next = attr; + attr->next = nullptr; + break; + } + cmp = uprv_compareInvCharsAsAscii(attr->attribute, cur->attribute); + if (cmp < 0) { + if (prev == nullptr) { + *first = attr; + } else { + prev->next = attr; + } + attr->next = cur; + break; + } + if (cmp == 0) { + /* duplicated variant */ + bAdded = false; + break; + } + prev = cur; + cur = cur->next; + } + } + + return bAdded; +} + + +static UBool +_addExtensionToList(ExtensionListEntry **first, ExtensionListEntry *ext, UBool localeToBCP) { + UBool bAdded = true; + + if (*first == nullptr) { + ext->next = nullptr; + *first = ext; + } else { + ExtensionListEntry *prev, *cur; + int32_t cmp; + + /* reorder variants in alphabetical order */ + prev = nullptr; + cur = *first; + while (true) { + if (cur == nullptr) { + prev->next = ext; + ext->next = nullptr; + break; + } + if (localeToBCP) { + /* special handling for locale to bcp conversion */ + int32_t len, curlen; + + len = (int32_t)uprv_strlen(ext->key); + curlen = (int32_t)uprv_strlen(cur->key); + + if (len == 1 && curlen == 1) { + if (*(ext->key) == *(cur->key)) { + cmp = 0; + } else if (*(ext->key) == PRIVATEUSE) { + cmp = 1; + } else if (*(cur->key) == PRIVATEUSE) { + cmp = -1; + } else { + cmp = *(ext->key) - *(cur->key); + } + } else if (len == 1) { + cmp = *(ext->key) - LDMLEXT; + } else if (curlen == 1) { + cmp = LDMLEXT - *(cur->key); + } else { + cmp = uprv_compareInvCharsAsAscii(ext->key, cur->key); + /* Both are u extension keys - we need special handling for 'attribute' */ + if (cmp != 0) { + if (uprv_strcmp(cur->key, LOCALE_ATTRIBUTE_KEY) == 0) { + cmp = 1; + } else if (uprv_strcmp(ext->key, LOCALE_ATTRIBUTE_KEY) == 0) { + cmp = -1; + } + } + } + } else { + cmp = uprv_compareInvCharsAsAscii(ext->key, cur->key); + } + if (cmp < 0) { + if (prev == nullptr) { + *first = ext; + } else { + prev->next = ext; + } + ext->next = cur; + break; + } + if (cmp == 0) { + /* duplicated extension key */ + bAdded = false; + break; + } + prev = cur; + cur = cur->next; + } + } + + return bAdded; +} + +static void +_initializeULanguageTag(ULanguageTag* langtag) { + int32_t i; + + langtag->buf = nullptr; + + langtag->language = EMPTY; + for (i = 0; i < MAXEXTLANG; i++) { + langtag->extlang[i] = nullptr; + } + + langtag->script = EMPTY; + langtag->region = EMPTY; + + langtag->variants = nullptr; + langtag->extensions = nullptr; + + langtag->legacy = EMPTY; + langtag->privateuse = EMPTY; +} + +static void +_appendLanguageToLanguageTag(const char* localeID, icu::ByteSink& sink, UBool strict, UErrorCode* status) { + char buf[ULOC_LANG_CAPACITY]; + UErrorCode tmpStatus = U_ZERO_ERROR; + int32_t len, i; + + if (U_FAILURE(*status)) { + return; + } + + len = uloc_getLanguage(localeID, buf, sizeof(buf), &tmpStatus); + if (U_FAILURE(tmpStatus) || tmpStatus == U_STRING_NOT_TERMINATED_WARNING) { + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + len = 0; + } + + /* Note: returned language code is in lower case letters */ + + if (len == 0) { + sink.Append(LANG_UND, LANG_UND_LEN); + } else if (!ultag_isLanguageSubtag(buf, len)) { + /* invalid language code */ + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + sink.Append(LANG_UND, LANG_UND_LEN); + } else { + /* resolve deprecated */ + for (i = 0; i < UPRV_LENGTHOF(DEPRECATEDLANGS); i += 2) { + // 2-letter deprecated subtags are listede before 3-letter + // ones in DEPRECATEDLANGS[]. Get out of loop on coming + // across the 1st 3-letter subtag, if the input is a 2-letter code. + // to avoid continuing to try when there's no match. + if (uprv_strlen(buf) < uprv_strlen(DEPRECATEDLANGS[i])) break; + if (uprv_compareInvCharsAsAscii(buf, DEPRECATEDLANGS[i]) == 0) { + uprv_strcpy(buf, DEPRECATEDLANGS[i + 1]); + len = (int32_t)uprv_strlen(buf); + break; + } + } + sink.Append(buf, len); + } +} + +static void +_appendScriptToLanguageTag(const char* localeID, icu::ByteSink& sink, UBool strict, UErrorCode* status) { + char buf[ULOC_SCRIPT_CAPACITY]; + UErrorCode tmpStatus = U_ZERO_ERROR; + int32_t len; + + if (U_FAILURE(*status)) { + return; + } + + len = uloc_getScript(localeID, buf, sizeof(buf), &tmpStatus); + if (U_FAILURE(tmpStatus) || tmpStatus == U_STRING_NOT_TERMINATED_WARNING) { + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + } + return; + } + + if (len > 0) { + if (!ultag_isScriptSubtag(buf, len)) { + /* invalid script code */ + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + } + return; + } else { + sink.Append("-", 1); + sink.Append(buf, len); + } + } +} + +static void +_appendRegionToLanguageTag(const char* localeID, icu::ByteSink& sink, UBool strict, UErrorCode* status) { + char buf[ULOC_COUNTRY_CAPACITY]; + UErrorCode tmpStatus = U_ZERO_ERROR; + int32_t len; + + if (U_FAILURE(*status)) { + return; + } + + len = uloc_getCountry(localeID, buf, sizeof(buf), &tmpStatus); + if (U_FAILURE(tmpStatus) || tmpStatus == U_STRING_NOT_TERMINATED_WARNING) { + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + } + return; + } + + if (len > 0) { + if (!ultag_isRegionSubtag(buf, len)) { + /* invalid region code */ + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + } + return; + } else { + sink.Append("-", 1); + /* resolve deprecated */ + for (int i = 0; i < UPRV_LENGTHOF(DEPRECATEDREGIONS); i += 2) { + if (uprv_compareInvCharsAsAscii(buf, DEPRECATEDREGIONS[i]) == 0) { + uprv_strcpy(buf, DEPRECATEDREGIONS[i + 1]); + len = (int32_t)uprv_strlen(buf); + break; + } + } + sink.Append(buf, len); + } + } +} + +static void _sortVariants(VariantListEntry* first) { + for (VariantListEntry* var1 = first; var1 != nullptr; var1 = var1->next) { + for (VariantListEntry* var2 = var1->next; var2 != nullptr; var2 = var2->next) { + // Swap var1->variant and var2->variant. + if (uprv_compareInvCharsAsAscii(var1->variant, var2->variant) > 0) { + const char* temp = var1->variant; + var1->variant = var2->variant; + var2->variant = temp; + } + } + } +} + +static void +_appendVariantsToLanguageTag(const char* localeID, icu::ByteSink& sink, UBool strict, UBool *hadPosix, UErrorCode* status) { + char buf[ULOC_FULLNAME_CAPACITY]; + UErrorCode tmpStatus = U_ZERO_ERROR; + int32_t len, i; + + if (U_FAILURE(*status)) { + return; + } + + len = uloc_getVariant(localeID, buf, sizeof(buf), &tmpStatus); + if (U_FAILURE(tmpStatus) || tmpStatus == U_STRING_NOT_TERMINATED_WARNING) { + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + } + return; + } + + if (len > 0) { + char *p, *pVar; + UBool bNext = true; + VariantListEntry *var; + VariantListEntry *varFirst = nullptr; + + pVar = nullptr; + p = buf; + while (bNext) { + if (*p == SEP || *p == LOCALE_SEP || *p == 0) { + if (*p == 0) { + bNext = false; + } else { + *p = 0; /* terminate */ + } + if (pVar == nullptr) { + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + break; + } + /* ignore empty variant */ + } else { + /* ICU uses upper case letters for variants, but + the canonical format is lowercase in BCP47 */ + for (i = 0; *(pVar + i) != 0; i++) { + *(pVar + i) = uprv_tolower(*(pVar + i)); + } + + /* validate */ + if (_isVariantSubtag(pVar, -1)) { + if (uprv_strcmp(pVar,POSIX_VALUE) || len != (int32_t)uprv_strlen(POSIX_VALUE)) { + /* emit the variant to the list */ + var = (VariantListEntry*)uprv_malloc(sizeof(VariantListEntry)); + if (var == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + break; + } + var->variant = pVar; + if (!_addVariantToList(&varFirst, var)) { + /* duplicated variant */ + uprv_free(var); + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + break; + } + } + } else { + /* Special handling for POSIX variant, need to remember that we had it and then */ + /* treat it like an extension later. */ + *hadPosix = true; + } + } else if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + break; + } else if (_isPrivateuseValueSubtag(pVar, -1)) { + /* Handle private use subtags separately */ + break; + } + } + /* reset variant starting position */ + pVar = nullptr; + } else if (pVar == nullptr) { + pVar = p; + } + p++; + } + + if (U_SUCCESS(*status)) { + if (varFirst != nullptr) { + int32_t varLen; + + /* per UTS35, we should sort the variants */ + _sortVariants(varFirst); + + /* write out validated/normalized variants to the target */ + var = varFirst; + while (var != nullptr) { + sink.Append("-", 1); + varLen = (int32_t)uprv_strlen(var->variant); + sink.Append(var->variant, varLen); + var = var->next; + } + } + } + + /* clean up */ + var = varFirst; + while (var != nullptr) { + VariantListEntry *tmpVar = var->next; + uprv_free(var); + var = tmpVar; + } + + if (U_FAILURE(*status)) { + return; + } + } +} + +static void +_appendKeywordsToLanguageTag(const char* localeID, icu::ByteSink& sink, UBool strict, UBool hadPosix, UErrorCode* status) { + char attrBuf[ULOC_KEYWORD_AND_VALUES_CAPACITY] = { 0 }; + int32_t attrBufLength = 0; + + icu::MemoryPool attrPool; + icu::MemoryPool extPool; + icu::MemoryPool strPool; + + icu::LocalUEnumerationPointer keywordEnum(uloc_openKeywords(localeID, status)); + if (U_FAILURE(*status) && !hadPosix) { + return; + } + if (keywordEnum.isValid() || hadPosix) { + /* reorder extensions */ + int32_t len; + const char *key; + ExtensionListEntry *firstExt = nullptr; + ExtensionListEntry *ext; + AttributeListEntry *firstAttr = nullptr; + AttributeListEntry *attr; + icu::MemoryPool extBufPool; + const char *bcpKey=nullptr, *bcpValue=nullptr; + UErrorCode tmpStatus = U_ZERO_ERROR; + int32_t keylen; + UBool isBcpUExt; + + while (true) { + key = uenum_next(keywordEnum.getAlias(), nullptr, status); + if (key == nullptr) { + break; + } + + icu::CharString buf; + { + icu::CharStringByteSink sink(&buf); + ulocimp_getKeywordValue(localeID, key, sink, &tmpStatus); + } + len = buf.length(); + + if (U_FAILURE(tmpStatus)) { + if (tmpStatus == U_MEMORY_ALLOCATION_ERROR) { + *status = U_MEMORY_ALLOCATION_ERROR; + break; + } + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + break; + } + /* ignore this keyword */ + tmpStatus = U_ZERO_ERROR; + continue; + } + + keylen = (int32_t)uprv_strlen(key); + isBcpUExt = (keylen > 1); + + /* special keyword used for representing Unicode locale attributes */ + if (uprv_strcmp(key, LOCALE_ATTRIBUTE_KEY) == 0) { + if (len > 0) { + int32_t i = 0; + while (true) { + attrBufLength = 0; + for (; i < len; i++) { + if (buf[i] != '-') { + attrBuf[attrBufLength++] = buf[i]; + } else { + i++; + break; + } + } + if (attrBufLength > 0) { + attrBuf[attrBufLength] = 0; + + } else if (i >= len){ + break; + } + + /* create AttributeListEntry */ + attr = attrPool.create(); + if (attr == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + break; + } + icu::CharString* attrValue = + strPool.create(attrBuf, attrBufLength, *status); + if (attrValue == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + break; + } + if (U_FAILURE(*status)) { + break; + } + attr->attribute = attrValue->data(); + + if (!_addAttributeToList(&firstAttr, attr)) { + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + break; + } + } + } + /* for a place holder ExtensionListEntry */ + bcpKey = LOCALE_ATTRIBUTE_KEY; + bcpValue = nullptr; + } + } else if (isBcpUExt) { + bcpKey = uloc_toUnicodeLocaleKey(key); + if (bcpKey == nullptr) { + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + break; + } + continue; + } + + /* we've checked buf is null-terminated above */ + bcpValue = uloc_toUnicodeLocaleType(key, buf.data()); + if (bcpValue == nullptr) { + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + break; + } + continue; + } + if (bcpValue == buf.data()) { + /* + When uloc_toUnicodeLocaleType(key, buf) returns the + input value as is, the value is well-formed, but has + no known mapping. This implementation normalizes the + value to lower case + */ + icu::CharString* extBuf = extBufPool.create(buf, tmpStatus); + + if (extBuf == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + break; + } + if (U_FAILURE(tmpStatus)) { + *status = tmpStatus; + break; + } + + T_CString_toLowerCase(extBuf->data()); + bcpValue = extBuf->data(); + } + } else { + if (*key == PRIVATEUSE) { + if (!ultag_isPrivateuseValueSubtags(buf.data(), len)) { + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + break; + } + continue; + } + } else { + if (!_isExtensionSingleton(key, keylen) || !ultag_isExtensionSubtags(buf.data(), len)) { + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + break; + } + continue; + } + } + bcpKey = key; + icu::CharString* extBuf = + extBufPool.create(buf.data(), len, tmpStatus); + if (extBuf == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + break; + } + if (U_FAILURE(tmpStatus)) { + *status = tmpStatus; + break; + } + bcpValue = extBuf->data(); + } + + /* create ExtensionListEntry */ + ext = extPool.create(); + if (ext == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + break; + } + ext->key = bcpKey; + ext->value = bcpValue; + + if (!_addExtensionToList(&firstExt, ext, true)) { + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + break; + } + } + } + + /* Special handling for POSIX variant - add the keywords for POSIX */ + if (hadPosix) { + /* create ExtensionListEntry for POSIX */ + ext = extPool.create(); + if (ext == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return; + } + ext->key = POSIX_KEY; + ext->value = POSIX_VALUE; + + if (!_addExtensionToList(&firstExt, ext, true)) { + // Silently ignore errors. + } + } + + if (U_SUCCESS(*status) && (firstExt != nullptr || firstAttr != nullptr)) { + UBool startLDMLExtension = false; + for (ext = firstExt; ext; ext = ext->next) { + if (!startLDMLExtension && uprv_strlen(ext->key) > 1) { + /* first LDML u singlton extension */ + sink.Append("-u", 2); + startLDMLExtension = true; + } + + /* write out the sorted BCP47 attributes, extensions and private use */ + if (uprv_strcmp(ext->key, LOCALE_ATTRIBUTE_KEY) == 0) { + /* write the value for the attributes */ + for (attr = firstAttr; attr; attr = attr->next) { + sink.Append("-", 1); + sink.Append( + attr->attribute, static_cast(uprv_strlen(attr->attribute))); + } + } else { + sink.Append("-", 1); + sink.Append(ext->key, static_cast(uprv_strlen(ext->key))); + if (uprv_strcmp(ext->value, "true") != 0 && + uprv_strcmp(ext->value, "yes") != 0) { + sink.Append("-", 1); + sink.Append(ext->value, static_cast(uprv_strlen(ext->value))); + } + } + } + } + } +} + +/** + * Append keywords parsed from LDML extension value + * e.g. "u-ca-gregory-co-trad" -> {calendar = gregorian} {collation = traditional} + * Note: char* buf is used for storing keywords + */ +static void +_appendLDMLExtensionAsKeywords(const char* ldmlext, ExtensionListEntry** appendTo, icu::MemoryPool& extPool, icu::MemoryPool& kwdBuf, UBool *posixVariant, UErrorCode *status) { + const char *pTag; /* beginning of current subtag */ + const char *pKwds; /* beginning of key-type pairs */ + UBool variantExists = *posixVariant; + + ExtensionListEntry *kwdFirst = nullptr; /* first LDML keyword */ + ExtensionListEntry *kwd, *nextKwd; + + int32_t len; + + /* Reset the posixVariant value */ + *posixVariant = false; + + pTag = ldmlext; + pKwds = nullptr; + + { + AttributeListEntry *attrFirst = nullptr; /* first attribute */ + AttributeListEntry *attr, *nextAttr; + + char attrBuf[ULOC_KEYWORD_AND_VALUES_CAPACITY]; + int32_t attrBufIdx = 0; + + icu::MemoryPool attrPool; + + /* Iterate through u extension attributes */ + while (*pTag) { + /* locate next separator char */ + for (len = 0; *(pTag + len) && *(pTag + len) != SEP; len++); + + if (ultag_isUnicodeLocaleKey(pTag, len)) { + pKwds = pTag; + break; + } + + /* add this attribute to the list */ + attr = attrPool.create(); + if (attr == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return; + } + + if (len < (int32_t)sizeof(attrBuf) - attrBufIdx) { + uprv_memcpy(&attrBuf[attrBufIdx], pTag, len); + attrBuf[attrBufIdx + len] = 0; + attr->attribute = &attrBuf[attrBufIdx]; + attrBufIdx += (len + 1); + } else { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + // duplicate attribute is ignored, causes no error. + _addAttributeToList(&attrFirst, attr); + + /* next tag */ + pTag += len; + if (*pTag) { + /* next to the separator */ + pTag++; + } + } + + if (attrFirst) { + /* emit attributes as an LDML keyword, e.g. attribute=attr1-attr2 */ + + kwd = extPool.create(); + if (kwd == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return; + } + + icu::CharString* value = kwdBuf.create(); + if (value == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return; + } + + /* attribute subtags sorted in alphabetical order as type */ + attr = attrFirst; + while (attr != nullptr) { + nextAttr = attr->next; + if (attr != attrFirst) { + value->append('-', *status); + } + value->append(attr->attribute, *status); + attr = nextAttr; + } + if (U_FAILURE(*status)) { + return; + } + + kwd->key = LOCALE_ATTRIBUTE_KEY; + kwd->value = value->data(); + + if (!_addExtensionToList(&kwdFirst, kwd, false)) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + } + } + + if (pKwds) { + const char *pBcpKey = nullptr; /* u extension key subtag */ + const char *pBcpType = nullptr; /* beginning of u extension type subtag(s) */ + int32_t bcpKeyLen = 0; + int32_t bcpTypeLen = 0; + UBool isDone = false; + + pTag = pKwds; + /* BCP47 representation of LDML key/type pairs */ + while (!isDone) { + const char *pNextBcpKey = nullptr; + int32_t nextBcpKeyLen = 0; + UBool emitKeyword = false; + + if (*pTag) { + /* locate next separator char */ + for (len = 0; *(pTag + len) && *(pTag + len) != SEP; len++); + + if (ultag_isUnicodeLocaleKey(pTag, len)) { + if (pBcpKey) { + emitKeyword = true; + pNextBcpKey = pTag; + nextBcpKeyLen = len; + } else { + pBcpKey = pTag; + bcpKeyLen = len; + } + } else { + U_ASSERT(pBcpKey != nullptr); + /* within LDML type subtags */ + if (pBcpType) { + bcpTypeLen += (len + 1); + } else { + pBcpType = pTag; + bcpTypeLen = len; + } + } + + /* next tag */ + pTag += len; + if (*pTag) { + /* next to the separator */ + pTag++; + } + } else { + /* processing last one */ + emitKeyword = true; + isDone = true; + } + + if (emitKeyword) { + const char *pKey = nullptr; /* LDML key */ + const char *pType = nullptr; /* LDML type */ + + char bcpKeyBuf[3]; /* BCP key length is always 2 for now */ + + U_ASSERT(pBcpKey != nullptr); + + if (bcpKeyLen >= (int32_t)sizeof(bcpKeyBuf)) { + /* the BCP key is invalid */ + *status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + U_ASSERT(bcpKeyLen <= 2); + + uprv_strncpy(bcpKeyBuf, pBcpKey, bcpKeyLen); + bcpKeyBuf[bcpKeyLen] = 0; + + /* u extension key to LDML key */ + pKey = uloc_toLegacyKey(bcpKeyBuf); + if (pKey == nullptr) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + if (pKey == bcpKeyBuf) { + /* + The key returned by toLegacyKey points to the input buffer. + We normalize the result key to lower case. + */ + T_CString_toLowerCase(bcpKeyBuf); + icu::CharString* key = kwdBuf.create(bcpKeyBuf, bcpKeyLen, *status); + if (key == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return; + } + if (U_FAILURE(*status)) { + return; + } + pKey = key->data(); + } + + if (pBcpType) { + char bcpTypeBuf[128]; /* practically long enough even considering multiple subtag type */ + if (bcpTypeLen >= (int32_t)sizeof(bcpTypeBuf)) { + /* the BCP type is too long */ + *status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + uprv_strncpy(bcpTypeBuf, pBcpType, bcpTypeLen); + bcpTypeBuf[bcpTypeLen] = 0; + + /* BCP type to locale type */ + pType = uloc_toLegacyType(pKey, bcpTypeBuf); + if (pType == nullptr) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + if (pType == bcpTypeBuf) { + /* + The type returned by toLegacyType points to the input buffer. + We normalize the result type to lower case. + */ + /* normalize to lower case */ + T_CString_toLowerCase(bcpTypeBuf); + icu::CharString* type = kwdBuf.create(bcpTypeBuf, bcpTypeLen, *status); + if (type == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return; + } + if (U_FAILURE(*status)) { + return; + } + pType = type->data(); + } + } else { + /* typeless - default type value is "yes" */ + pType = LOCALE_TYPE_YES; + } + + /* Special handling for u-va-posix, since we want to treat this as a variant, + not as a keyword */ + if (!variantExists && !uprv_strcmp(pKey, POSIX_KEY) && !uprv_strcmp(pType, POSIX_VALUE) ) { + *posixVariant = true; + } else { + /* create an ExtensionListEntry for this keyword */ + kwd = extPool.create(); + if (kwd == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return; + } + + kwd->key = pKey; + kwd->value = pType; + + if (!_addExtensionToList(&kwdFirst, kwd, false)) { + // duplicate keyword is allowed, Only the first + // is honored. + } + } + + pBcpKey = pNextBcpKey; + bcpKeyLen = pNextBcpKey != nullptr ? nextBcpKeyLen : 0; + pBcpType = nullptr; + bcpTypeLen = 0; + } + } + } + + kwd = kwdFirst; + while (kwd != nullptr) { + nextKwd = kwd->next; + _addExtensionToList(appendTo, kwd, false); + kwd = nextKwd; + } +} + + +static void +_appendKeywords(ULanguageTag* langtag, icu::ByteSink& sink, UErrorCode* status) { + int32_t i, n; + int32_t len; + ExtensionListEntry *kwdFirst = nullptr; + ExtensionListEntry *kwd; + const char *key, *type; + icu::MemoryPool extPool; + icu::MemoryPool kwdBuf; + UBool posixVariant = false; + + if (U_FAILURE(*status)) { + return; + } + + n = ultag_getExtensionsSize(langtag); + + /* resolve locale keywords and reordering keys */ + for (i = 0; i < n; i++) { + key = ultag_getExtensionKey(langtag, i); + type = ultag_getExtensionValue(langtag, i); + if (*key == LDMLEXT) { + /* Determine if variants already exists */ + if (ultag_getVariantsSize(langtag)) { + posixVariant = true; + } + + _appendLDMLExtensionAsKeywords(type, &kwdFirst, extPool, kwdBuf, &posixVariant, status); + if (U_FAILURE(*status)) { + break; + } + } else { + kwd = extPool.create(); + if (kwd == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + break; + } + kwd->key = key; + kwd->value = type; + if (!_addExtensionToList(&kwdFirst, kwd, false)) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + break; + } + } + } + + if (U_SUCCESS(*status)) { + type = ultag_getPrivateUse(langtag); + if ((int32_t)uprv_strlen(type) > 0) { + /* add private use as a keyword */ + kwd = extPool.create(); + if (kwd == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + } else { + kwd->key = PRIVATEUSE_KEY; + kwd->value = type; + if (!_addExtensionToList(&kwdFirst, kwd, false)) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + } + } + } + } + + /* If a POSIX variant was in the extensions, write it out before writing the keywords. */ + + if (U_SUCCESS(*status) && posixVariant) { + len = (int32_t) uprv_strlen(_POSIX); + sink.Append(_POSIX, len); + } + + if (U_SUCCESS(*status) && kwdFirst != nullptr) { + /* write out the sorted keywords */ + UBool firstValue = true; + kwd = kwdFirst; + do { + if (firstValue) { + sink.Append("@", 1); + firstValue = false; + } else { + sink.Append(";", 1); + } + + /* key */ + len = (int32_t)uprv_strlen(kwd->key); + sink.Append(kwd->key, len); + sink.Append("=", 1); + + /* type */ + len = (int32_t)uprv_strlen(kwd->value); + sink.Append(kwd->value, len); + + kwd = kwd->next; + } while (kwd); + } +} + +static void +_appendPrivateuseToLanguageTag(const char* localeID, icu::ByteSink& sink, UBool strict, UBool hadPosix, UErrorCode* status) { + (void)hadPosix; + char buf[ULOC_FULLNAME_CAPACITY]; + char tmpAppend[ULOC_FULLNAME_CAPACITY]; + UErrorCode tmpStatus = U_ZERO_ERROR; + int32_t len, i; + int32_t reslen = 0; + int32_t capacity = sizeof tmpAppend; + + if (U_FAILURE(*status)) { + return; + } + + len = uloc_getVariant(localeID, buf, sizeof(buf), &tmpStatus); + if (U_FAILURE(tmpStatus) || tmpStatus == U_STRING_NOT_TERMINATED_WARNING) { + if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + } + return; + } + + if (len > 0) { + char *p, *pPriv; + UBool bNext = true; + UBool firstValue = true; + UBool writeValue; + + pPriv = nullptr; + p = buf; + while (bNext) { + writeValue = false; + if (*p == SEP || *p == LOCALE_SEP || *p == 0) { + if (*p == 0) { + bNext = false; + } else { + *p = 0; /* terminate */ + } + if (pPriv != nullptr) { + /* Private use in the canonical format is lowercase in BCP47 */ + for (i = 0; *(pPriv + i) != 0; i++) { + *(pPriv + i) = uprv_tolower(*(pPriv + i)); + } + + /* validate */ + if (_isPrivateuseValueSubtag(pPriv, -1)) { + if (firstValue) { + if (!_isVariantSubtag(pPriv, -1)) { + writeValue = true; + } + } else { + writeValue = true; + } + } else if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + break; + } else { + break; + } + + if (writeValue) { + if (reslen < capacity) { + tmpAppend[reslen++] = SEP; + } + + if (firstValue) { + if (reslen < capacity) { + tmpAppend[reslen++] = *PRIVATEUSE_KEY; + } + + if (reslen < capacity) { + tmpAppend[reslen++] = SEP; + } + + len = (int32_t)uprv_strlen(PRIVUSE_VARIANT_PREFIX); + if (reslen < capacity) { + uprv_memcpy(tmpAppend + reslen, PRIVUSE_VARIANT_PREFIX, uprv_min(len, capacity - reslen)); + } + reslen += len; + + if (reslen < capacity) { + tmpAppend[reslen++] = SEP; + } + + firstValue = false; + } + + len = (int32_t)uprv_strlen(pPriv); + if (reslen < capacity) { + uprv_memcpy(tmpAppend + reslen, pPriv, uprv_min(len, capacity - reslen)); + } + reslen += len; + } + } + /* reset private use starting position */ + pPriv = nullptr; + } else if (pPriv == nullptr) { + pPriv = p; + } + p++; + } + + if (U_FAILURE(*status)) { + return; + } + } + + if (U_SUCCESS(*status)) { + len = reslen; + sink.Append(tmpAppend, len); + } +} + +/* +* ------------------------------------------------- +* +* ultag_ functions +* +* ------------------------------------------------- +*/ + +/* Bit flags used by the parser */ +#define LANG 0x0001 +#define EXTL 0x0002 +#define SCRT 0x0004 +#define REGN 0x0008 +#define VART 0x0010 +#define EXTS 0x0020 +#define EXTV 0x0040 +#define PRIV 0x0080 + +/** + * Ticket #12705 - The optimizer in Visual Studio 2015 Update 3 has problems optimizing this function. + * As a work-around, optimization is disabled for this function on VS2015 and VS2017. + * This work-around should be removed once the following versions of Visual Studio are no + * longer supported: All versions of VS2015/VS2017, and versions of VS2019 below 16.4. + */ +#if defined(_MSC_VER) && (_MSC_VER >= 1900) && (_MSC_VER < 1924) +#pragma optimize( "", off ) +#endif + +static ULanguageTag* +ultag_parse(const char* tag, int32_t tagLen, int32_t* parsedLen, UErrorCode* status) { + char *tagBuf; + int16_t next; + char *pSubtag, *pNext, *pLastGoodPosition; + int32_t subtagLen; + int32_t extlangIdx; + ExtensionListEntry *pExtension; + char *pExtValueSubtag, *pExtValueSubtagEnd; + int32_t i; + UBool privateuseVar = false; + int32_t legacyLen = 0; + + if (parsedLen != nullptr) { + *parsedLen = 0; + } + + if (U_FAILURE(*status)) { + return nullptr; + } + + if (tagLen < 0) { + tagLen = (int32_t)uprv_strlen(tag); + } + + /* copy the entire string */ + tagBuf = (char*)uprv_malloc(tagLen + 1); + if (tagBuf == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + + if (tagLen > 0) { + uprv_memcpy(tagBuf, tag, tagLen); + } + *(tagBuf + tagLen) = 0; + + /* create a ULanguageTag */ + icu::LocalULanguageTagPointer t( + (ULanguageTag*)uprv_malloc(sizeof(ULanguageTag))); + if (t.isNull()) { + uprv_free(tagBuf); + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + _initializeULanguageTag(t.getAlias()); + t->buf = tagBuf; + + if (tagLen < MINLEN) { + /* the input tag is too short - return empty ULanguageTag */ + return t.orphan(); + } + + size_t parsedLenDelta = 0; + // Legacy tag will be consider together. Legacy tag with intervening + // script and region such as art-DE-lojban or art-Latn-lojban won't be + // matched. + /* check if the tag is legacy */ + for (i = 0; i < UPRV_LENGTHOF(LEGACY); i += 2) { + int32_t checkLegacyLen = static_cast(uprv_strlen(LEGACY[i])); + if (tagLen < checkLegacyLen) { + continue; + } + if (tagLen > checkLegacyLen && tagBuf[checkLegacyLen] != '-') { + // make sure next char is '-'. + continue; + } + if (uprv_strnicmp(LEGACY[i], tagBuf, checkLegacyLen) == 0) { + int32_t newTagLength; + + legacyLen = checkLegacyLen; /* back up for output parsedLen */ + int32_t replacementLen = static_cast(uprv_strlen(LEGACY[i+1])); + newTagLength = replacementLen + tagLen - checkLegacyLen; + int32_t oldTagLength = tagLen; + if (tagLen < newTagLength) { + uprv_free(tagBuf); + tagBuf = (char*)uprv_malloc(newTagLength + 1); + if (tagBuf == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + t->buf = tagBuf; + tagLen = newTagLength; + } + parsedLenDelta = checkLegacyLen - replacementLen; + uprv_strcpy(t->buf, LEGACY[i + 1]); + if (checkLegacyLen != tagLen) { + uprv_memcpy(t->buf + replacementLen, tag + checkLegacyLen, + oldTagLength - checkLegacyLen); + // NUL-terminate after memcpy(). + t->buf[replacementLen + oldTagLength - checkLegacyLen] = 0; + } + break; + } + } + + if (legacyLen == 0) { + for (i = 0; i < UPRV_LENGTHOF(REDUNDANT); i += 2) { + const char* redundantTag = REDUNDANT[i]; + size_t redundantTagLen = uprv_strlen(redundantTag); + // The preferred tag for a redundant tag is always shorter than redundant + // tag. A redundant tag may or may not be followed by other subtags. + // (i.e. "zh-yue" or "zh-yue-u-co-pinyin"). + if (uprv_strnicmp(redundantTag, tagBuf, static_cast(redundantTagLen)) == 0) { + const char* redundantTagEnd = tagBuf + redundantTagLen; + if (*redundantTagEnd == '\0' || *redundantTagEnd == SEP) { + const char* preferredTag = REDUNDANT[i + 1]; + size_t preferredTagLen = uprv_strlen(preferredTag); + uprv_memcpy(t->buf, preferredTag, preferredTagLen); + if (*redundantTagEnd == SEP) { + uprv_memmove(tagBuf + preferredTagLen, + redundantTagEnd, + tagLen - redundantTagLen + 1); + } else { + tagBuf[preferredTagLen] = '\0'; + } + // parsedLen should be the length of the input + // before redundantTag is replaced by preferredTag. + // Save the delta to add it back later. + parsedLenDelta = redundantTagLen - preferredTagLen; + break; + } + } + } + } + + /* + * langtag = language + * ["-" script] + * ["-" region] + * *("-" variant) + * *("-" extension) + * ["-" privateuse] + */ + + next = LANG | PRIV; + pNext = pLastGoodPosition = tagBuf; + extlangIdx = 0; + pExtension = nullptr; + pExtValueSubtag = nullptr; + pExtValueSubtagEnd = nullptr; + + while (pNext) { + char *pSep; + + pSubtag = pNext; + + /* locate next separator char */ + pSep = pSubtag; + while (*pSep) { + if (*pSep == SEP) { + break; + } + pSep++; + } + if (*pSep == 0) { + /* last subtag */ + pNext = nullptr; + } else { + pNext = pSep + 1; + } + subtagLen = (int32_t)(pSep - pSubtag); + + if (next & LANG) { + if (ultag_isLanguageSubtag(pSubtag, subtagLen)) { + *pSep = 0; /* terminate */ + // TODO: move deprecated language code handling here. + t->language = T_CString_toLowerCase(pSubtag); + + pLastGoodPosition = pSep; + next = SCRT | REGN | VART | EXTS | PRIV; + if (subtagLen <= 3) + next |= EXTL; + continue; + } + } + if (next & EXTL) { + if (_isExtlangSubtag(pSubtag, subtagLen)) { + *pSep = 0; + t->extlang[extlangIdx++] = T_CString_toLowerCase(pSubtag); + + pLastGoodPosition = pSep; + if (extlangIdx < 3) { + next = EXTL | SCRT | REGN | VART | EXTS | PRIV; + } else { + next = SCRT | REGN | VART | EXTS | PRIV; + } + continue; + } + } + if (next & SCRT) { + if (ultag_isScriptSubtag(pSubtag, subtagLen)) { + char *p = pSubtag; + + *pSep = 0; + + /* to title case */ + *p = uprv_toupper(*p); + p++; + for (; *p; p++) { + *p = uprv_tolower(*p); + } + + t->script = pSubtag; + + pLastGoodPosition = pSep; + next = REGN | VART | EXTS | PRIV; + continue; + } + } + if (next & REGN) { + if (ultag_isRegionSubtag(pSubtag, subtagLen)) { + *pSep = 0; + // TODO: move deprecated region code handling here. + t->region = T_CString_toUpperCase(pSubtag); + + pLastGoodPosition = pSep; + next = VART | EXTS | PRIV; + continue; + } + } + if (next & VART) { + if (_isVariantSubtag(pSubtag, subtagLen) || + (privateuseVar && _isPrivateuseVariantSubtag(pSubtag, subtagLen))) { + VariantListEntry *var; + UBool isAdded; + + var = (VariantListEntry*)uprv_malloc(sizeof(VariantListEntry)); + if (var == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + *pSep = 0; + var->variant = T_CString_toUpperCase(pSubtag); + isAdded = _addVariantToList(&(t->variants), var); + if (!isAdded) { + /* duplicated variant entry */ + uprv_free(var); + break; + } + pLastGoodPosition = pSep; + next = VART | EXTS | PRIV; + continue; + } + } + if (next & EXTS) { + if (_isExtensionSingleton(pSubtag, subtagLen)) { + if (pExtension != nullptr) { + if (pExtValueSubtag == nullptr || pExtValueSubtagEnd == nullptr) { + /* the previous extension is incomplete */ + uprv_free(pExtension); + pExtension = nullptr; + break; + } + + /* terminate the previous extension value */ + *pExtValueSubtagEnd = 0; + pExtension->value = T_CString_toLowerCase(pExtValueSubtag); + + /* insert the extension to the list */ + if (_addExtensionToList(&(t->extensions), pExtension, false)) { + pLastGoodPosition = pExtValueSubtagEnd; + } else { + /* stop parsing here */ + uprv_free(pExtension); + pExtension = nullptr; + break; + } + } + + /* create a new extension */ + pExtension = (ExtensionListEntry*)uprv_malloc(sizeof(ExtensionListEntry)); + if (pExtension == nullptr) { + *status = U_MEMORY_ALLOCATION_ERROR; + return nullptr; + } + *pSep = 0; + pExtension->key = T_CString_toLowerCase(pSubtag); + pExtension->value = nullptr; /* will be set later */ + + /* + * reset the start and the end location of extension value + * subtags for this extension + */ + pExtValueSubtag = nullptr; + pExtValueSubtagEnd = nullptr; + + next = EXTV; + continue; + } + } + if (next & EXTV) { + if (_isExtensionSubtag(pSubtag, subtagLen)) { + if (pExtValueSubtag == nullptr) { + /* if the start position of this extension's value is not yet, + this one is the first value subtag */ + pExtValueSubtag = pSubtag; + } + + /* Mark the end of this subtag */ + pExtValueSubtagEnd = pSep; + next = EXTS | EXTV | PRIV; + + continue; + } + } + if (next & PRIV) { + if (uprv_tolower(*pSubtag) == PRIVATEUSE && subtagLen == 1) { + char *pPrivuseVal; + + if (pExtension != nullptr) { + /* Process the last extension */ + if (pExtValueSubtag == nullptr || pExtValueSubtagEnd == nullptr) { + /* the previous extension is incomplete */ + uprv_free(pExtension); + pExtension = nullptr; + break; + } else { + /* terminate the previous extension value */ + *pExtValueSubtagEnd = 0; + pExtension->value = T_CString_toLowerCase(pExtValueSubtag); + + /* insert the extension to the list */ + if (_addExtensionToList(&(t->extensions), pExtension, false)) { + pLastGoodPosition = pExtValueSubtagEnd; + pExtension = nullptr; + } else { + /* stop parsing here */ + uprv_free(pExtension); + pExtension = nullptr; + break; + } + } + } + + /* The rest of part will be private use value subtags */ + if (pNext == nullptr) { + /* empty private use subtag */ + break; + } + /* back up the private use value start position */ + pPrivuseVal = pNext; + + /* validate private use value subtags */ + while (pNext) { + pSubtag = pNext; + pSep = pSubtag; + while (*pSep) { + if (*pSep == SEP) { + break; + } + pSep++; + } + if (*pSep == 0) { + /* last subtag */ + pNext = nullptr; + } else { + pNext = pSep + 1; + } + subtagLen = (int32_t)(pSep - pSubtag); + + if (uprv_strncmp(pSubtag, PRIVUSE_VARIANT_PREFIX, uprv_strlen(PRIVUSE_VARIANT_PREFIX)) == 0) { + *pSep = 0; + next = VART; + privateuseVar = true; + break; + } else if (_isPrivateuseValueSubtag(pSubtag, subtagLen)) { + pLastGoodPosition = pSep; + } else { + break; + } + } + + if (next == VART) { + continue; + } + + if (pLastGoodPosition - pPrivuseVal > 0) { + *pLastGoodPosition = 0; + t->privateuse = T_CString_toLowerCase(pPrivuseVal); + } + /* No more subtags, exiting the parse loop */ + break; + } + break; + } + + /* If we fell through here, it means this subtag is illegal - quit parsing */ + break; + } + + if (pExtension != nullptr) { + /* Process the last extension */ + if (pExtValueSubtag == nullptr || pExtValueSubtagEnd == nullptr) { + /* the previous extension is incomplete */ + uprv_free(pExtension); + } else { + /* terminate the previous extension value */ + *pExtValueSubtagEnd = 0; + pExtension->value = T_CString_toLowerCase(pExtValueSubtag); + /* insert the extension to the list */ + if (_addExtensionToList(&(t->extensions), pExtension, false)) { + pLastGoodPosition = pExtValueSubtagEnd; + } else { + uprv_free(pExtension); + } + } + } + + if (parsedLen != nullptr) { + *parsedLen = (int32_t)(pLastGoodPosition - t->buf + parsedLenDelta); + } + + return t.orphan(); +} + +// Ticket #12705 - Turn optimization back on. +#if defined(_MSC_VER) && (_MSC_VER >= 1900) && (_MSC_VER < 1924) +#pragma optimize( "", on ) +#endif + +static void +ultag_close(ULanguageTag* langtag) { + + if (langtag == nullptr) { + return; + } + + uprv_free(langtag->buf); + + if (langtag->variants) { + VariantListEntry *curVar = langtag->variants; + while (curVar) { + VariantListEntry *nextVar = curVar->next; + uprv_free(curVar); + curVar = nextVar; + } + } + + if (langtag->extensions) { + ExtensionListEntry *curExt = langtag->extensions; + while (curExt) { + ExtensionListEntry *nextExt = curExt->next; + uprv_free(curExt); + curExt = nextExt; + } + } + + uprv_free(langtag); +} + +static const char* +ultag_getLanguage(const ULanguageTag* langtag) { + return langtag->language; +} + +#if 0 +static const char* +ultag_getJDKLanguage(const ULanguageTag* langtag) { + int32_t i; + for (i = 0; DEPRECATEDLANGS[i] != nullptr; i += 2) { + if (uprv_compareInvCharsAsAscii(DEPRECATEDLANGS[i], langtag->language) == 0) { + return DEPRECATEDLANGS[i + 1]; + } + } + return langtag->language; +} +#endif + +static const char* +ultag_getExtlang(const ULanguageTag* langtag, int32_t idx) { + if (idx >= 0 && idx < MAXEXTLANG) { + return langtag->extlang[idx]; + } + return nullptr; +} + +static int32_t +ultag_getExtlangSize(const ULanguageTag* langtag) { + int32_t size = 0; + int32_t i; + for (i = 0; i < MAXEXTLANG; i++) { + if (langtag->extlang[i]) { + size++; + } + } + return size; +} + +static const char* +ultag_getScript(const ULanguageTag* langtag) { + return langtag->script; +} + +static const char* +ultag_getRegion(const ULanguageTag* langtag) { + return langtag->region; +} + +static const char* +ultag_getVariant(const ULanguageTag* langtag, int32_t idx) { + const char *var = nullptr; + VariantListEntry *cur = langtag->variants; + int32_t i = 0; + while (cur) { + if (i == idx) { + var = cur->variant; + break; + } + cur = cur->next; + i++; + } + return var; +} + +static int32_t +ultag_getVariantsSize(const ULanguageTag* langtag) { + int32_t size = 0; + VariantListEntry *cur = langtag->variants; + while (true) { + if (cur == nullptr) { + break; + } + size++; + cur = cur->next; + } + return size; +} + +static const char* +ultag_getExtensionKey(const ULanguageTag* langtag, int32_t idx) { + const char *key = nullptr; + ExtensionListEntry *cur = langtag->extensions; + int32_t i = 0; + while (cur) { + if (i == idx) { + key = cur->key; + break; + } + cur = cur->next; + i++; + } + return key; +} + +static const char* +ultag_getExtensionValue(const ULanguageTag* langtag, int32_t idx) { + const char *val = nullptr; + ExtensionListEntry *cur = langtag->extensions; + int32_t i = 0; + while (cur) { + if (i == idx) { + val = cur->value; + break; + } + cur = cur->next; + i++; + } + return val; +} + +static int32_t +ultag_getExtensionsSize(const ULanguageTag* langtag) { + int32_t size = 0; + ExtensionListEntry *cur = langtag->extensions; + while (true) { + if (cur == nullptr) { + break; + } + size++; + cur = cur->next; + } + return size; +} + +static const char* +ultag_getPrivateUse(const ULanguageTag* langtag) { + return langtag->privateuse; +} + +#if 0 +static const char* +ultag_getLegacy(const ULanguageTag* langtag) { + return langtag->legacy; +} +#endif + + +/* +* ------------------------------------------------- +* +* Locale/BCP47 conversion APIs, exposed as uloc_* +* +* ------------------------------------------------- +*/ +U_CAPI int32_t U_EXPORT2 +uloc_toLanguageTag(const char* localeID, + char* langtag, + int32_t langtagCapacity, + UBool strict, + UErrorCode* status) { + if (U_FAILURE(*status)) { + return 0; + } + + icu::CheckedArrayByteSink sink(langtag, langtagCapacity); + ulocimp_toLanguageTag(localeID, sink, strict, status); + + int32_t reslen = sink.NumberOfBytesAppended(); + + if (U_FAILURE(*status)) { + return reslen; + } + + if (sink.Overflowed()) { + *status = U_BUFFER_OVERFLOW_ERROR; + } else { + u_terminateChars(langtag, langtagCapacity, reslen, status); + } + + return reslen; +} + + +U_CAPI void U_EXPORT2 +ulocimp_toLanguageTag(const char* localeID, + icu::ByteSink& sink, + UBool strict, + UErrorCode* status) { + icu::CharString canonical; + int32_t reslen; + UErrorCode tmpStatus = U_ZERO_ERROR; + UBool hadPosix = false; + const char* pKeywordStart; + + /* Note: uloc_canonicalize returns "en_US_POSIX" for input locale ID "". See #6835 */ + int32_t resultCapacity = static_cast(uprv_strlen(localeID)); + if (resultCapacity > 0) { + char* buffer; + + for (;;) { + buffer = canonical.getAppendBuffer( + /*minCapacity=*/resultCapacity, + /*desiredCapacityHint=*/resultCapacity, + resultCapacity, + tmpStatus); + + if (U_FAILURE(tmpStatus)) { + *status = tmpStatus; + return; + } + + reslen = + uloc_canonicalize(localeID, buffer, resultCapacity, &tmpStatus); + + if (tmpStatus != U_BUFFER_OVERFLOW_ERROR) { + break; + } + + resultCapacity = reslen; + tmpStatus = U_ZERO_ERROR; + } + + if (U_FAILURE(tmpStatus)) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + canonical.append(buffer, reslen, tmpStatus); + if (tmpStatus == U_STRING_NOT_TERMINATED_WARNING) { + tmpStatus = U_ZERO_ERROR; // Terminators provided by CharString. + } + + if (U_FAILURE(tmpStatus)) { + *status = tmpStatus; + return; + } + } + + /* For handling special case - private use only tag */ + pKeywordStart = locale_getKeywordsStart(canonical.data()); + if (pKeywordStart == canonical.data()) { + int kwdCnt = 0; + UBool done = false; + + icu::LocalUEnumerationPointer kwdEnum(uloc_openKeywords(canonical.data(), &tmpStatus)); + if (U_SUCCESS(tmpStatus)) { + kwdCnt = uenum_count(kwdEnum.getAlias(), &tmpStatus); + if (kwdCnt == 1) { + const char *key; + int32_t len = 0; + + key = uenum_next(kwdEnum.getAlias(), &len, &tmpStatus); + if (len == 1 && *key == PRIVATEUSE) { + icu::CharString buf; + { + icu::CharStringByteSink sink(&buf); + ulocimp_getKeywordValue(localeID, key, sink, &tmpStatus); + } + if (U_SUCCESS(tmpStatus)) { + if (ultag_isPrivateuseValueSubtags(buf.data(), buf.length())) { + /* return private use only tag */ + sink.Append("und-x-", 6); + sink.Append(buf.data(), buf.length()); + done = true; + } else if (strict) { + *status = U_ILLEGAL_ARGUMENT_ERROR; + done = true; + } + /* if not strict mode, then "und" will be returned */ + } else { + *status = U_ILLEGAL_ARGUMENT_ERROR; + done = true; + } + } + } + if (done) { + return; + } + } + } + + _appendLanguageToLanguageTag(canonical.data(), sink, strict, status); + _appendScriptToLanguageTag(canonical.data(), sink, strict, status); + _appendRegionToLanguageTag(canonical.data(), sink, strict, status); + _appendVariantsToLanguageTag(canonical.data(), sink, strict, &hadPosix, status); + _appendKeywordsToLanguageTag(canonical.data(), sink, strict, hadPosix, status); + _appendPrivateuseToLanguageTag(canonical.data(), sink, strict, hadPosix, status); +} + + +U_CAPI int32_t U_EXPORT2 +uloc_forLanguageTag(const char* langtag, + char* localeID, + int32_t localeIDCapacity, + int32_t* parsedLength, + UErrorCode* status) { + if (U_FAILURE(*status)) { + return 0; + } + + icu::CheckedArrayByteSink sink(localeID, localeIDCapacity); + ulocimp_forLanguageTag(langtag, -1, sink, parsedLength, status); + + int32_t reslen = sink.NumberOfBytesAppended(); + + if (U_FAILURE(*status)) { + return reslen; + } + + if (sink.Overflowed()) { + *status = U_BUFFER_OVERFLOW_ERROR; + } else { + u_terminateChars(localeID, localeIDCapacity, reslen, status); + } + + return reslen; +} + + +U_CAPI void U_EXPORT2 +ulocimp_forLanguageTag(const char* langtag, + int32_t tagLen, + icu::ByteSink& sink, + int32_t* parsedLength, + UErrorCode* status) { + UBool isEmpty = true; + const char *subtag, *p; + int32_t len; + int32_t i, n; + UBool noRegion = true; + + icu::LocalULanguageTagPointer lt(ultag_parse(langtag, tagLen, parsedLength, status)); + if (U_FAILURE(*status)) { + return; + } + + /* language */ + subtag = ultag_getExtlangSize(lt.getAlias()) > 0 ? ultag_getExtlang(lt.getAlias(), 0) : ultag_getLanguage(lt.getAlias()); + if (uprv_compareInvCharsAsAscii(subtag, LANG_UND) != 0) { + len = (int32_t)uprv_strlen(subtag); + if (len > 0) { + sink.Append(subtag, len); + isEmpty = false; + } + } + + /* script */ + subtag = ultag_getScript(lt.getAlias()); + len = (int32_t)uprv_strlen(subtag); + if (len > 0) { + sink.Append("_", 1); + isEmpty = false; + + /* write out the script in title case */ + char c = uprv_toupper(*subtag); + sink.Append(&c, 1); + sink.Append(subtag + 1, len - 1); + } + + /* region */ + subtag = ultag_getRegion(lt.getAlias()); + len = (int32_t)uprv_strlen(subtag); + if (len > 0) { + sink.Append("_", 1); + isEmpty = false; + + /* write out the region in upper case */ + p = subtag; + while (*p) { + char c = uprv_toupper(*p); + sink.Append(&c, 1); + p++; + } + noRegion = false; + } + + /* variants */ + _sortVariants(lt.getAlias()->variants); + n = ultag_getVariantsSize(lt.getAlias()); + if (n > 0) { + if (noRegion) { + sink.Append("_", 1); + isEmpty = false; + } + + for (i = 0; i < n; i++) { + subtag = ultag_getVariant(lt.getAlias(), i); + sink.Append("_", 1); + + /* write out the variant in upper case */ + p = subtag; + while (*p) { + char c = uprv_toupper(*p); + sink.Append(&c, 1); + p++; + } + } + } + + /* keywords */ + n = ultag_getExtensionsSize(lt.getAlias()); + subtag = ultag_getPrivateUse(lt.getAlias()); + if (n > 0 || uprv_strlen(subtag) > 0) { + if (isEmpty && n > 0) { + /* need a language */ + sink.Append(LANG_UND, LANG_UND_LEN); + } + _appendKeywords(lt.getAlias(), sink, status); + } +} diff --git a/intl/icu/source/common/ulocimp.h b/intl/icu/source/common/ulocimp.h new file mode 100644 index 0000000000..48341054e3 --- /dev/null +++ b/intl/icu/source/common/ulocimp.h @@ -0,0 +1,378 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2004-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ + +#ifndef ULOCIMP_H +#define ULOCIMP_H + +#include "unicode/bytestream.h" +#include "unicode/uloc.h" + +#include "charstr.h" + +/** + * Create an iterator over the specified keywords list + * @param keywordList double-null terminated list. Will be copied. + * @param keywordListSize size in bytes of keywordList + * @param status err code + * @return enumeration (owned by caller) of the keyword list. + * @internal ICU 3.0 + */ +U_CAPI UEnumeration* U_EXPORT2 +uloc_openKeywordList(const char *keywordList, int32_t keywordListSize, UErrorCode* status); + +/** + * Look up a resource bundle table item with fallback on the table level. + * This is accessible so it can be called by C++ code. + */ +U_CAPI const UChar * U_EXPORT2 +uloc_getTableStringWithFallback( + const char *path, + const char *locale, + const char *tableKey, + const char *subTableKey, + const char *itemKey, + int32_t *pLength, + UErrorCode *pErrorCode); + +/*returns true if a is an ID separator false otherwise*/ +#define _isIDSeparator(a) (a == '_' || a == '-') + +U_CFUNC const char* +uloc_getCurrentCountryID(const char* oldID); + +U_CFUNC const char* +uloc_getCurrentLanguageID(const char* oldID); + +U_CFUNC void +ulocimp_getKeywords(const char *localeID, + char prev, + icu::ByteSink& sink, + UBool valuesToo, + UErrorCode *status); + +icu::CharString U_EXPORT2 +ulocimp_getLanguage(const char *localeID, + const char **pEnd, + UErrorCode &status); + +icu::CharString U_EXPORT2 +ulocimp_getScript(const char *localeID, + const char **pEnd, + UErrorCode &status); + +icu::CharString U_EXPORT2 +ulocimp_getCountry(const char *localeID, + const char **pEnd, + UErrorCode &status); + +U_CAPI void U_EXPORT2 +ulocimp_getName(const char* localeID, + icu::ByteSink& sink, + UErrorCode* err); + +U_CAPI void U_EXPORT2 +ulocimp_getBaseName(const char* localeID, + icu::ByteSink& sink, + UErrorCode* err); + +U_CAPI void U_EXPORT2 +ulocimp_canonicalize(const char* localeID, + icu::ByteSink& sink, + UErrorCode* err); + +U_CAPI void U_EXPORT2 +ulocimp_getKeywordValue(const char* localeID, + const char* keywordName, + icu::ByteSink& sink, + UErrorCode* status); + +/** + * Writes a well-formed language tag for this locale ID. + * + * **Note**: When `strict` is false, any locale fields which do not satisfy the + * BCP47 syntax requirement will be omitted from the result. When `strict` is + * true, this function sets U_ILLEGAL_ARGUMENT_ERROR to the `err` if any locale + * fields do not satisfy the BCP47 syntax requirement. + * + * @param localeID the input locale ID + * @param sink the output sink receiving the BCP47 language + * tag for this Locale. + * @param strict boolean value indicating if the function returns + * an error for an ill-formed input locale ID. + * @param err error information if receiving the language + * tag failed. + * @return The length of the BCP47 language tag. + * + * @internal ICU 64 + */ +U_CAPI void U_EXPORT2 +ulocimp_toLanguageTag(const char* localeID, + icu::ByteSink& sink, + UBool strict, + UErrorCode* err); + +/** + * Returns a locale ID for the specified BCP47 language tag string. + * If the specified language tag contains any ill-formed subtags, + * the first such subtag and all following subtags are ignored. + *

+ * This implements the 'Language-Tag' production of BCP 47, and so + * supports legacy language tags (marked as “Type: grandfathered” in BCP 47) + * (regular and irregular) as well as private use language tags. + * + * Private use tags are represented as 'x-whatever', + * and legacy tags are converted to their canonical replacements where they exist. + * + * Note that a few legacy tags have no modern replacement; + * these will be converted using the fallback described in + * the first paragraph, so some information might be lost. + * + * @param langtag the input BCP47 language tag. + * @param tagLen the length of langtag, or -1 to call uprv_strlen(). + * @param sink the output sink receiving a locale ID for the + * specified BCP47 language tag. + * @param parsedLength if not NULL, successfully parsed length + * for the input language tag is set. + * @param err error information if receiving the locald ID + * failed. + * @internal ICU 63 + */ +U_CAPI void U_EXPORT2 +ulocimp_forLanguageTag(const char* langtag, + int32_t tagLen, + icu::ByteSink& sink, + int32_t* parsedLength, + UErrorCode* err); + +/** + * Get the region to use for supplemental data lookup. Uses + * (1) any region specified by locale tag "rg"; if none then + * (2) any unicode_region_tag in the locale ID; if none then + * (3) if inferRegion is true, the region suggested by + * getLikelySubtags on the localeID. + * If no region is found, returns length 0. + * + * @param localeID + * The complete locale ID (with keywords) from which + * to get the region to use for supplemental data. + * @param inferRegion + * If true, will try to infer region from localeID if + * no other region is found. + * @param region + * Buffer in which to put the region ID found; should + * have a capacity at least ULOC_COUNTRY_CAPACITY. + * @param regionCapacity + * The actual capacity of the region buffer. + * @param status + * Pointer to in/out UErrorCode value for latest status. + * @return + * The length of any region code found, or 0 if none. + * @internal ICU 57 + */ +U_CAPI int32_t U_EXPORT2 +ulocimp_getRegionForSupplementalData(const char *localeID, UBool inferRegion, + char *region, int32_t regionCapacity, UErrorCode* status); + +/** + * Add the likely subtags for a provided locale ID, per the algorithm described + * in the following CLDR technical report: + * + * http://www.unicode.org/reports/tr35/#Likely_Subtags + * + * If localeID is already in the maximal form, or there is no data available + * for maximization, it will be copied to the output buffer. For example, + * "und-Zzzz" cannot be maximized, since there is no reasonable maximization. + * + * Examples: + * + * "en" maximizes to "en_Latn_US" + * + * "de" maximizes to "de_Latn_US" + * + * "sr" maximizes to "sr_Cyrl_RS" + * + * "sh" maximizes to "sr_Latn_RS" (Note this will not reverse.) + * + * "zh_Hani" maximizes to "zh_Hans_CN" (Note this will not reverse.) + * + * @param localeID The locale to maximize + * @param sink The output sink receiving the maximized locale + * @param err Error information if maximizing the locale failed. If the length + * of the localeID and the null-terminator is greater than the maximum allowed size, + * or the localeId is not well-formed, the error code is U_ILLEGAL_ARGUMENT_ERROR. + * @internal ICU 64 + */ +U_CAPI void U_EXPORT2 +ulocimp_addLikelySubtags(const char* localeID, + icu::ByteSink& sink, + UErrorCode* err); + +/** + * Minimize the subtags for a provided locale ID, per the algorithm described + * in the following CLDR technical report: + * + * http://www.unicode.org/reports/tr35/#Likely_Subtags + * + * If localeID is already in the minimal form, or there is no data available + * for minimization, it will be copied to the output buffer. Since the + * minimization algorithm relies on proper maximization, see the comments + * for ulocimp_addLikelySubtags for reasons why there might not be any data. + * + * Examples: + * + * "en_Latn_US" minimizes to "en" + * + * "de_Latn_US" minimizes to "de" + * + * "sr_Cyrl_RS" minimizes to "sr" + * + * "zh_Hant_TW" minimizes to "zh_TW" (The region is preferred to the + * script, and minimizing to "zh" would imply "zh_Hans_CN".) + * + * @param localeID The locale to minimize + * @param sink The output sink receiving the maximized locale + * @param err Error information if minimizing the locale failed. If the length + * of the localeID and the null-terminator is greater than the maximum allowed size, + * or the localeId is not well-formed, the error code is U_ILLEGAL_ARGUMENT_ERROR. + * @internal ICU 64 + */ +U_CAPI void U_EXPORT2 +ulocimp_minimizeSubtags(const char* localeID, + icu::ByteSink& sink, + UErrorCode* err); + +U_CAPI const char * U_EXPORT2 +locale_getKeywordsStart(const char *localeID); + +U_CFUNC UBool +ultag_isExtensionSubtags(const char* s, int32_t len); + +U_CFUNC UBool +ultag_isLanguageSubtag(const char* s, int32_t len); + +U_CFUNC UBool +ultag_isPrivateuseValueSubtags(const char* s, int32_t len); + +U_CFUNC UBool +ultag_isRegionSubtag(const char* s, int32_t len); + +U_CFUNC UBool +ultag_isScriptSubtag(const char* s, int32_t len); + +U_CFUNC UBool +ultag_isTransformedExtensionSubtags(const char* s, int32_t len); + +U_CFUNC UBool +ultag_isUnicodeExtensionSubtags(const char* s, int32_t len); + +U_CFUNC UBool +ultag_isUnicodeLocaleAttribute(const char* s, int32_t len); + +U_CFUNC UBool +ultag_isUnicodeLocaleAttributes(const char* s, int32_t len); + +U_CFUNC UBool +ultag_isUnicodeLocaleKey(const char* s, int32_t len); + +U_CFUNC UBool +ultag_isUnicodeLocaleType(const char* s, int32_t len); + +U_CFUNC UBool +ultag_isVariantSubtags(const char* s, int32_t len); + +U_CAPI const char * U_EXPORT2 +ultag_getTKeyStart(const char *localeID); + +U_CFUNC const char* +ulocimp_toBcpKey(const char* key); + +U_CFUNC const char* +ulocimp_toLegacyKey(const char* key); + +U_CFUNC const char* +ulocimp_toBcpType(const char* key, const char* type, UBool* isKnownKey, UBool* isSpecialType); + +U_CFUNC const char* +ulocimp_toLegacyType(const char* key, const char* type, UBool* isKnownKey, UBool* isSpecialType); + +/* Function for testing purpose */ +U_CAPI const char* const* ulocimp_getKnownCanonicalizedLocaleForTest(int32_t* length); + +// Return true if the value is already canonicalized. +U_CAPI bool ulocimp_isCanonicalizedLocaleForTest(const char* localeName); + +/** + * A utility class for handling locale IDs that may be longer than ULOC_FULLNAME_CAPACITY. + * This encompasses all of the logic to allocate a temporary locale ID buffer on the stack, + * and then, if it's not big enough, reallocate it on the heap and try again. + * + * You use it like this: + * UErrorCode err = U_ZERO_ERROR; + * + * PreflightingLocaleIDBuffer tempBuffer; + * do { + * tempBuffer.requestedCapacity = uloc_doSomething(localeID, tempBuffer.getBuffer(), tempBuffer.getCapacity(), &err); + * } while (tempBuffer.needToTryAgain(&err)); + * if (U_SUCCESS(err)) { + * uloc_doSomethingWithTheResult(tempBuffer.getBuffer()); + * } + */ +class PreflightingLocaleIDBuffer { +private: + char stackBuffer[ULOC_FULLNAME_CAPACITY]; + char* heapBuffer = nullptr; + int32_t capacity = ULOC_FULLNAME_CAPACITY; + +public: + int32_t requestedCapacity = ULOC_FULLNAME_CAPACITY; + + // No heap allocation. Use only on the stack. + static void* U_EXPORT2 operator new(size_t) noexcept = delete; + static void* U_EXPORT2 operator new[](size_t) noexcept = delete; +#if U_HAVE_PLACEMENT_NEW + static void* U_EXPORT2 operator new(size_t, void*) noexcept = delete; +#endif + + PreflightingLocaleIDBuffer() {} + + ~PreflightingLocaleIDBuffer() { uprv_free(heapBuffer); } + + char* getBuffer() { + if (heapBuffer == nullptr) { + return stackBuffer; + } else { + return heapBuffer; + } + } + + int32_t getCapacity() { + return capacity; + } + + bool needToTryAgain(UErrorCode* err) { + if (heapBuffer != nullptr) { + return false; + } + + if (*err == U_BUFFER_OVERFLOW_ERROR || *err == U_STRING_NOT_TERMINATED_WARNING) { + int32_t newCapacity = requestedCapacity + 2; // one for the terminating null, one just for paranoia + heapBuffer = static_cast(uprv_malloc(newCapacity)); + if (heapBuffer == nullptr) { + *err = U_MEMORY_ALLOCATION_ERROR; + } else { + *err = U_ZERO_ERROR; + capacity = newCapacity; + } + return U_SUCCESS(*err); + } + return false; + } +}; + +#endif diff --git a/intl/icu/source/common/umapfile.cpp b/intl/icu/source/common/umapfile.cpp new file mode 100644 index 0000000000..faa6d80786 --- /dev/null +++ b/intl/icu/source/common/umapfile.cpp @@ -0,0 +1,530 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2013, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************/ + + +/*---------------------------------------------------------------------------- + * + * Memory mapped file wrappers for use by the ICU Data Implementation + * All of the platform-specific implementation for mapping data files + * is here. The rest of the ICU Data implementation uses only the + * wrapper functions. + * + *----------------------------------------------------------------------------*/ +/* Defines _XOPEN_SOURCE for access to POSIX functions. + * Must be before any other #includes. */ +#include "uposixdefs.h" + +#include "unicode/putil.h" +#include "unicode/ustring.h" +#include "udatamem.h" +#include "umapfile.h" + +/* memory-mapping base definitions ------------------------------------------ */ + +#if MAP_IMPLEMENTATION==MAP_WIN32 +#ifndef WIN32_LEAN_AND_MEAN +# define WIN32_LEAN_AND_MEAN +#endif +# define VC_EXTRALEAN +# define NOUSER +# define NOSERVICE +# define NOIME +# define NOMCX + +# if U_PLATFORM_HAS_WINUWP_API == 1 + // Some previous versions of the Windows 10 SDK don't expose various APIs for UWP applications + // to use, even though UWP apps are allowed to call and use them. Temporarily change the + // WINAPI family partition below to Desktop, so that function declarations are visible for UWP. +# include +# if !(WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP | WINAPI_PARTITION_SYSTEM)) +# pragma push_macro("WINAPI_PARTITION_DESKTOP") +# undef WINAPI_PARTITION_DESKTOP +# define WINAPI_PARTITION_DESKTOP 1 +# define CHANGED_WINAPI_PARTITION_DESKTOP_VALUE +# endif +# endif + +# include + +# if U_PLATFORM_HAS_WINUWP_API == 1 && defined(CHANGED_WINAPI_PARTITION_DESKTOP_VALUE) +# pragma pop_macro("WINAPI_PARTITION_DESKTOP") +# endif + +# include "cmemory.h" + +typedef HANDLE MemoryMap; + +# define IS_MAP(map) ((map)!=nullptr) + +#elif MAP_IMPLEMENTATION==MAP_POSIX || MAP_IMPLEMENTATION==MAP_390DLL + typedef size_t MemoryMap; + +# define IS_MAP(map) ((map)!=0) + +# include +# include +# include +# include + +# ifndef MAP_FAILED +# define MAP_FAILED ((void*)-1) +# endif + +# if MAP_IMPLEMENTATION==MAP_390DLL + /* No memory mapping for 390 batch mode. Fake it using dll loading. */ +# include +# include "cstring.h" +# include "cmemory.h" +# include "unicode/udata.h" +# define LIB_PREFIX "lib" +# define LIB_SUFFIX ".dll" + /* This is inconvenient until we figure out what to do with U_ICUDATA_NAME in utypes.h */ +# define U_ICUDATA_ENTRY_NAME "icudt" U_ICU_VERSION_SHORT U_LIB_SUFFIX_C_NAME_STRING "_dat" +# endif +#elif MAP_IMPLEMENTATION==MAP_STDIO +# include +# include "cmemory.h" + + typedef void *MemoryMap; + +# define IS_MAP(map) ((map)!=nullptr) +#endif + +/*----------------------------------------------------------------------------* + * * + * Memory Mapped File support. Platform dependent implementation of * + * functions used by the rest of the implementation.* + * * + *----------------------------------------------------------------------------*/ +#if MAP_IMPLEMENTATION==MAP_NONE + U_CFUNC UBool + uprv_mapFile(UDataMemory *pData, const char *path, UErrorCode *status) { + if (U_FAILURE(*status)) { + return false; + } + UDataMemory_init(pData); /* Clear the output struct. */ + return false; /* no file access */ + } + + U_CFUNC void uprv_unmapFile(UDataMemory *pData) { + /* nothing to do */ + } +#elif MAP_IMPLEMENTATION==MAP_WIN32 + U_CFUNC UBool + uprv_mapFile( + UDataMemory *pData, /* Fill in with info on the result doing the mapping. */ + /* Output only; any original contents are cleared. */ + const char *path, /* File path to be opened/mapped. */ + UErrorCode *status /* Error status, used to report out-of-memory errors. */ + ) + { + if (U_FAILURE(*status)) { + return false; + } + + HANDLE map = nullptr; + HANDLE file = INVALID_HANDLE_VALUE; + + UDataMemory_init(pData); /* Clear the output struct. */ + + /* open the input file */ +#if U_PLATFORM_HAS_WINUWP_API == 0 + // Note: In the non-UWP code-path (ie: Win32), the value of the path variable might have come from + // the CRT 'getenv' function, and would be therefore be encoded in the default ANSI code page. + // This means that we can't call the *W version of API below, whereas in the UWP code-path + // there is no 'getenv' call, and thus the string will be only UTF-8/Invariant characters. + file=CreateFileA(path, GENERIC_READ, FILE_SHARE_READ, nullptr, + OPEN_EXISTING, + FILE_ATTRIBUTE_NORMAL|FILE_FLAG_RANDOM_ACCESS, nullptr); +#else + // Convert from UTF-8 string to UTF-16 string. + wchar_t utf16Path[MAX_PATH]; + int32_t pathUtf16Len = 0; + u_strFromUTF8(reinterpret_cast(utf16Path), static_cast(UPRV_LENGTHOF(utf16Path)), &pathUtf16Len, path, -1, status); + + if (U_FAILURE(*status)) { + return false; + } + if (*status == U_STRING_NOT_TERMINATED_WARNING) { + // Report back an error instead of a warning. + *status = U_BUFFER_OVERFLOW_ERROR; + return false; + } + + file = CreateFileW(utf16Path, GENERIC_READ, FILE_SHARE_READ, nullptr, + OPEN_EXISTING, + FILE_ATTRIBUTE_NORMAL | FILE_FLAG_RANDOM_ACCESS, nullptr); +#endif + if (file == INVALID_HANDLE_VALUE) { + // If we failed to open the file due to an out-of-memory error, then we want + // to report that error back to the caller. + if (HRESULT_FROM_WIN32(GetLastError()) == E_OUTOFMEMORY) { + *status = U_MEMORY_ALLOCATION_ERROR; + } + return false; + } + + // Note: We use nullptr/nullptr for lpAttributes parameter below. + // This means our handle cannot be inherited and we will get the default security descriptor. + /* create an unnamed Windows file-mapping object for the specified file */ + map = CreateFileMappingW(file, nullptr, PAGE_READONLY, 0, 0, nullptr); + + CloseHandle(file); + if (map == nullptr) { + // If we failed to create the mapping due to an out-of-memory error, then + // we want to report that error back to the caller. + if (HRESULT_FROM_WIN32(GetLastError()) == E_OUTOFMEMORY) { + *status = U_MEMORY_ALLOCATION_ERROR; + } + return false; + } + + /* map a view of the file into our address space */ + pData->pHeader = reinterpret_cast(MapViewOfFile(map, FILE_MAP_READ, 0, 0, 0)); + if (pData->pHeader == nullptr) { + CloseHandle(map); + return false; + } + pData->map = map; + return true; + } + + U_CFUNC void + uprv_unmapFile(UDataMemory *pData) { + if (pData != nullptr && pData->map != nullptr) { + UnmapViewOfFile(pData->pHeader); + CloseHandle(pData->map); + pData->pHeader = nullptr; + pData->map = nullptr; + } + } + + + +#elif MAP_IMPLEMENTATION==MAP_POSIX + U_CFUNC UBool + uprv_mapFile(UDataMemory *pData, const char *path, UErrorCode *status) { + int fd; + int length; + struct stat mystat; + void *data; + + if (U_FAILURE(*status)) { + return false; + } + + UDataMemory_init(pData); /* Clear the output struct. */ + + /* determine the length of the file */ + if(stat(path, &mystat)!=0 || mystat.st_size<=0) { + return false; + } + length=mystat.st_size; + + /* open the file */ + fd=open(path, O_RDONLY); + if(fd==-1) { + return false; + } + + /* get a view of the mapping */ +#if U_PLATFORM != U_PF_HPUX + data=mmap(0, length, PROT_READ, MAP_SHARED, fd, 0); +#else + data=mmap(0, length, PROT_READ, MAP_PRIVATE, fd, 0); +#endif + close(fd); /* no longer needed */ + if(data==MAP_FAILED) { + // Possibly check the errno value for ENOMEM, and report U_MEMORY_ALLOCATION_ERROR? + return false; + } + + pData->map = (char *)data + length; + pData->pHeader=(const DataHeader *)data; + pData->mapAddr = data; +#if U_PLATFORM == U_PF_IPHONE + posix_madvise(data, length, POSIX_MADV_RANDOM); +#endif + return true; + } + + U_CFUNC void + uprv_unmapFile(UDataMemory *pData) { + if(pData!=nullptr && pData->map!=nullptr) { + size_t dataLen = (char *)pData->map - (char *)pData->mapAddr; + if(munmap(pData->mapAddr, dataLen)==-1) { + } + pData->pHeader=nullptr; + pData->map=0; + pData->mapAddr=nullptr; + } + } + + + +#elif MAP_IMPLEMENTATION==MAP_STDIO + /* copy of the filestrm.c/T_FileStream_size() implementation */ + static int32_t + umap_fsize(FILE *f) { + int32_t savedPos = ftell(f); + int32_t size = 0; + + /*Changes by Bertrand A. D. doesn't affect the current position + goes to the end of the file before ftell*/ + fseek(f, 0, SEEK_END); + size = (int32_t)ftell(f); + fseek(f, savedPos, SEEK_SET); + return size; + } + + U_CFUNC UBool + uprv_mapFile(UDataMemory *pData, const char *path, UErrorCode *status) { + FILE *file; + int32_t fileLength; + void *p; + + if (U_FAILURE(*status)) { + return false; + } + + UDataMemory_init(pData); /* Clear the output struct. */ + /* open the input file */ + file=fopen(path, "rb"); + if(file==nullptr) { + return false; + } + + /* get the file length */ + fileLength=umap_fsize(file); + if(ferror(file) || fileLength<=20) { + fclose(file); + return false; + } + + /* allocate the memory to hold the file data */ + p=uprv_malloc(fileLength); + if(p==nullptr) { + fclose(file); + *status = U_MEMORY_ALLOCATION_ERROR; + return false; + } + + /* read the file */ + if(fileLength!=fread(p, 1, fileLength, file)) { + uprv_free(p); + fclose(file); + return false; + } + + fclose(file); + pData->map=p; + pData->pHeader=(const DataHeader *)p; + pData->mapAddr=p; + return true; + } + + U_CFUNC void + uprv_unmapFile(UDataMemory *pData) { + if(pData!=nullptr && pData->map!=nullptr) { + uprv_free(pData->map); + pData->map = nullptr; + pData->mapAddr = nullptr; + pData->pHeader = nullptr; + } + } + + +#elif MAP_IMPLEMENTATION==MAP_390DLL + /* 390 specific Library Loading. + * This is the only platform left that dynamically loads an ICU Data Library. + * All other platforms use .data files when dynamic loading is required, but + * this turn out to be awkward to support in 390 batch mode. + * + * The idea here is to hide the fact that 390 is using dll loading from the + * rest of ICU, and make it look like there is file loading happening. + * + */ + + static char *strcpy_returnEnd(char *dest, const char *src) + { + while((*dest=*src)!=0) { + ++dest; + ++src; + } + return dest; + } + + /*------------------------------------------------------------------------------ + * + * computeDirPath given a user-supplied path of an item to be opened, + * compute and return + * - the full directory path to be used + * when opening the file. + * - Pointer to null at end of above returned path + * + * Parameters: + * path: input path. Buffer is not altered. + * pathBuffer: Output buffer. Any contents are overwritten. + * + * Returns: + * Pointer to null termination in returned pathBuffer. + * + * TODO: This works the way ICU historically has, but the + * whole data fallback search path is so complicated that + * probably almost no one will ever really understand it, + * the potential for confusion is large. (It's not just + * this one function, but the whole scheme.) + * + *------------------------------------------------------------------------------*/ + static char *uprv_computeDirPath(const char *path, char *pathBuffer) + { + char *finalSlash; /* Ptr to last dir separator in input path, or null if none. */ + int32_t pathLen; /* Length of the returned directory path */ + + finalSlash = 0; + if (path != 0) { + finalSlash = uprv_strrchr(path, U_FILE_SEP_CHAR); + } + + *pathBuffer = 0; + if (finalSlash == 0) { + /* No user-supplied path. + * Copy the ICU_DATA path to the path buffer and return that*/ + const char *icuDataDir; + icuDataDir=u_getDataDirectory(); + if(icuDataDir!=nullptr && *icuDataDir!=0) { + return strcpy_returnEnd(pathBuffer, icuDataDir); + } else { + /* there is no icuDataDir either. Just return the empty pathBuffer. */ + return pathBuffer; + } + } + + /* User supplied path did contain a directory portion. + * Copy it to the output path buffer */ + pathLen = (int32_t)(finalSlash - path + 1); + uprv_memcpy(pathBuffer, path, pathLen); + *(pathBuffer+pathLen) = 0; + return pathBuffer+pathLen; + } + + +# define DATA_TYPE "dat" + + U_CFUNC UBool uprv_mapFile(UDataMemory *pData, const char *path, UErrorCode *status) { + const char *inBasename; + char *basename; + char pathBuffer[1024]; + const DataHeader *pHeader; + dllhandle *handle; + void *val=0; + + if (U_FAILURE(*status)) { + return false; + } + + inBasename=uprv_strrchr(path, U_FILE_SEP_CHAR); + if(inBasename==nullptr) { + inBasename = path; + } else { + inBasename++; + } + basename=uprv_computeDirPath(path, pathBuffer); + if(uprv_strcmp(inBasename, U_ICUDATA_NAME".dat") != 0) { + /* must mmap file... for build */ + int fd; + int length; + struct stat mystat; + void *data; + UDataMemory_init(pData); /* Clear the output struct. */ + + /* determine the length of the file */ + if(stat(path, &mystat)!=0 || mystat.st_size<=0) { + return false; + } + length=mystat.st_size; + + /* open the file */ + fd=open(path, O_RDONLY); + if(fd==-1) { + return false; + } + + /* get a view of the mapping */ + data=mmap(0, length, PROT_READ, MAP_PRIVATE, fd, 0); + close(fd); /* no longer needed */ + if(data==MAP_FAILED) { + // Possibly check the errorno value for ENOMEM, and report U_MEMORY_ALLOCATION_ERROR? + return false; + } + pData->map = (char *)data + length; + pData->pHeader=(const DataHeader *)data; + pData->mapAddr = data; + return true; + } + +# ifdef OS390BATCH + /* ### hack: we still need to get u_getDataDirectory() fixed + for OS/390 (batch mode - always return "//"? ) + and this here straightened out with LIB_PREFIX and LIB_SUFFIX (both empty?!) + This is probably due to the strange file system on OS/390. It's more like + a database with short entry names than a typical file system. */ + /* U_ICUDATA_NAME should always have the correct name */ + /* BUT FOR BATCH MODE IT IS AN EXCEPTION BECAUSE */ + /* THE FIRST THREE LETTERS ARE PREASSIGNED TO THE */ + /* PROJECT!!!!! */ + uprv_strcpy(pathBuffer, "IXMI" U_ICU_VERSION_SHORT "DA"); +# else + /* set up the library name */ + uprv_strcpy(basename, LIB_PREFIX U_LIBICUDATA_NAME U_ICU_VERSION_SHORT LIB_SUFFIX); +# endif + +# ifdef UDATA_DEBUG + fprintf(stderr, "dllload: %s ", pathBuffer); +# endif + + handle=dllload(pathBuffer); + +# ifdef UDATA_DEBUG + fprintf(stderr, " -> %08X\n", handle ); +# endif + + if(handle != nullptr) { + /* we have a data DLL - what kind of lookup do we need here? */ + /* try to find the Table of Contents */ + UDataMemory_init(pData); /* Clear the output struct. */ + val=dllqueryvar((dllhandle*)handle, U_ICUDATA_ENTRY_NAME); + if(val == 0) { + /* failed... so keep looking */ + return false; + } +# ifdef UDATA_DEBUG + fprintf(stderr, "dllqueryvar(%08X, %s) -> %08X\n", handle, U_ICUDATA_ENTRY_NAME, val); +# endif + + pData->pHeader=(const DataHeader *)val; + return true; + } else { + return false; /* no handle */ + } + } + + U_CFUNC void uprv_unmapFile(UDataMemory *pData) { + if(pData!=nullptr && pData->map!=nullptr) { + uprv_free(pData->map); + pData->map = nullptr; + pData->mapAddr = nullptr; + pData->pHeader = nullptr; + } + } + +#else +# error MAP_IMPLEMENTATION is set incorrectly +#endif diff --git a/intl/icu/source/common/umapfile.h b/intl/icu/source/common/umapfile.h new file mode 100644 index 0000000000..004f279fe6 --- /dev/null +++ b/intl/icu/source/common/umapfile.h @@ -0,0 +1,59 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2011, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************/ + +/*---------------------------------------------------------------------------------- + * + * Memory mapped file wrappers for use by the ICU Data Implementation + * + * Porting note: The implementation of these functions is very platform specific. + * Not all platforms can do real memory mapping. Those that can't + * still must implement these functions, getting the data into memory using + * whatever means are available. + * + * These functions are part of the ICU internal implementation, and + * are not intended to be used directly by applications. + * + *----------------------------------------------------------------------------------*/ + +#ifndef __UMAPFILE_H__ +#define __UMAPFILE_H__ + +#include "unicode/putil.h" +#include "unicode/udata.h" +#include "putilimp.h" + +U_CAPI UBool U_EXPORT2 uprv_mapFile(UDataMemory *pdm, const char *path, UErrorCode *status); +U_CFUNC void uprv_unmapFile(UDataMemory *pData); + +/* MAP_NONE: no memory mapping, no file access at all */ +#define MAP_NONE 0 +#define MAP_WIN32 1 +#define MAP_POSIX 2 +#define MAP_STDIO 3 +#define MAP_390DLL 4 + +#if UCONFIG_NO_FILE_IO +# define MAP_IMPLEMENTATION MAP_NONE +#elif defined(__wasi__) +# define MAP_IMPLEMENTATION MAP_STDIO +#elif U_PLATFORM_USES_ONLY_WIN32_API +# define MAP_IMPLEMENTATION MAP_WIN32 +#elif U_HAVE_MMAP || U_PLATFORM == U_PF_OS390 +# if U_PLATFORM == U_PF_OS390 && defined (OS390_STUBDATA) + /* No memory mapping for 390 batch mode. Fake it using dll loading. */ +# define MAP_IMPLEMENTATION MAP_390DLL +# else +# define MAP_IMPLEMENTATION MAP_POSIX +# endif +#else /* unknown platform, no memory map implementation: use stdio.h and uprv_malloc() instead */ +# define MAP_IMPLEMENTATION MAP_STDIO +#endif + +#endif diff --git a/intl/icu/source/common/umath.cpp b/intl/icu/source/common/umath.cpp new file mode 100644 index 0000000000..7cf4b31749 --- /dev/null +++ b/intl/icu/source/common/umath.cpp @@ -0,0 +1,26 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1997-2006, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* This file contains platform independent math. +*/ + +#include "putilimp.h" + +U_CAPI int32_t U_EXPORT2 +uprv_max(int32_t x, int32_t y) +{ + return (x > y ? x : y); +} + +U_CAPI int32_t U_EXPORT2 +uprv_min(int32_t x, int32_t y) +{ + return (x > y ? y : x); +} + diff --git a/intl/icu/source/common/umutablecptrie.cpp b/intl/icu/source/common/umutablecptrie.cpp new file mode 100644 index 0000000000..cdbe27080b --- /dev/null +++ b/intl/icu/source/common/umutablecptrie.cpp @@ -0,0 +1,1852 @@ +// © 2017 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// umutablecptrie.cpp (inspired by utrie2_builder.cpp) +// created: 2017dec29 Markus W. Scherer + +// #define UCPTRIE_DEBUG +#ifdef UCPTRIE_DEBUG +# include +#endif + +#include "unicode/utypes.h" +#include "unicode/ucptrie.h" +#include "unicode/umutablecptrie.h" +#include "unicode/uobject.h" +#include "unicode/utf16.h" +#include "cmemory.h" +#include "uassert.h" +#include "ucptrie_impl.h" + +// ICU-20235 In case Microsoft math.h has defined this, undefine it. +#ifdef OVERFLOW +#undef OVERFLOW +#endif + +U_NAMESPACE_BEGIN + +namespace { + +constexpr int32_t MAX_UNICODE = 0x10ffff; + +constexpr int32_t UNICODE_LIMIT = 0x110000; +constexpr int32_t BMP_LIMIT = 0x10000; +constexpr int32_t ASCII_LIMIT = 0x80; + +constexpr int32_t I_LIMIT = UNICODE_LIMIT >> UCPTRIE_SHIFT_3; +constexpr int32_t BMP_I_LIMIT = BMP_LIMIT >> UCPTRIE_SHIFT_3; +constexpr int32_t ASCII_I_LIMIT = ASCII_LIMIT >> UCPTRIE_SHIFT_3; + +constexpr int32_t SMALL_DATA_BLOCKS_PER_BMP_BLOCK = (1 << (UCPTRIE_FAST_SHIFT - UCPTRIE_SHIFT_3)); + +// Flag values for data blocks. +constexpr uint8_t ALL_SAME = 0; +constexpr uint8_t MIXED = 1; +constexpr uint8_t SAME_AS = 2; + +/** Start with allocation of 16k data entries. */ +constexpr int32_t INITIAL_DATA_LENGTH = ((int32_t)1 << 14); + +/** Grow about 8x each time. */ +constexpr int32_t MEDIUM_DATA_LENGTH = ((int32_t)1 << 17); + +/** + * Maximum length of the build-time data array. + * One entry per 0x110000 code points. + */ +constexpr int32_t MAX_DATA_LENGTH = UNICODE_LIMIT; + +// Flag values for index-3 blocks while compacting/building. +constexpr uint8_t I3_NULL = 0; +constexpr uint8_t I3_BMP = 1; +constexpr uint8_t I3_16 = 2; +constexpr uint8_t I3_18 = 3; + +constexpr int32_t INDEX_3_18BIT_BLOCK_LENGTH = UCPTRIE_INDEX_3_BLOCK_LENGTH + UCPTRIE_INDEX_3_BLOCK_LENGTH / 8; + +class AllSameBlocks; +class MixedBlocks; + +class MutableCodePointTrie : public UMemory { +public: + MutableCodePointTrie(uint32_t initialValue, uint32_t errorValue, UErrorCode &errorCode); + MutableCodePointTrie(const MutableCodePointTrie &other, UErrorCode &errorCode); + MutableCodePointTrie(const MutableCodePointTrie &other) = delete; + ~MutableCodePointTrie(); + + MutableCodePointTrie &operator=(const MutableCodePointTrie &other) = delete; + + static MutableCodePointTrie *fromUCPMap(const UCPMap *map, UErrorCode &errorCode); + static MutableCodePointTrie *fromUCPTrie(const UCPTrie *trie, UErrorCode &errorCode); + + uint32_t get(UChar32 c) const; + int32_t getRange(UChar32 start, UCPMapValueFilter *filter, const void *context, + uint32_t *pValue) const; + + void set(UChar32 c, uint32_t value, UErrorCode &errorCode); + void setRange(UChar32 start, UChar32 end, uint32_t value, UErrorCode &errorCode); + + UCPTrie *build(UCPTrieType type, UCPTrieValueWidth valueWidth, UErrorCode &errorCode); + +private: + void clear(); + + bool ensureHighStart(UChar32 c); + int32_t allocDataBlock(int32_t blockLength); + int32_t getDataBlock(int32_t i); + + void maskValues(uint32_t mask); + UChar32 findHighStart() const; + int32_t compactWholeDataBlocks(int32_t fastILimit, AllSameBlocks &allSameBlocks); + int32_t compactData( + int32_t fastILimit, uint32_t *newData, int32_t newDataCapacity, + int32_t dataNullIndex, MixedBlocks &mixedBlocks, UErrorCode &errorCode); + int32_t compactIndex(int32_t fastILimit, MixedBlocks &mixedBlocks, UErrorCode &errorCode); + int32_t compactTrie(int32_t fastILimit, UErrorCode &errorCode); + + uint32_t *index = nullptr; + int32_t indexCapacity = 0; + int32_t index3NullOffset = -1; + uint32_t *data = nullptr; + int32_t dataCapacity = 0; + int32_t dataLength = 0; + int32_t dataNullOffset = -1; + + uint32_t origInitialValue; + uint32_t initialValue; + uint32_t errorValue; + UChar32 highStart; + uint32_t highValue; +#ifdef UCPTRIE_DEBUG +public: + const char *name; +#endif +private: + /** Temporary array while building the final data. */ + uint16_t *index16 = nullptr; + uint8_t flags[UNICODE_LIMIT >> UCPTRIE_SHIFT_3]; +}; + +MutableCodePointTrie::MutableCodePointTrie(uint32_t iniValue, uint32_t errValue, UErrorCode &errorCode) : + origInitialValue(iniValue), initialValue(iniValue), errorValue(errValue), + highStart(0), highValue(initialValue) +#ifdef UCPTRIE_DEBUG + , name("open") +#endif + { + if (U_FAILURE(errorCode)) { return; } + index = (uint32_t *)uprv_malloc(BMP_I_LIMIT * 4); + data = (uint32_t *)uprv_malloc(INITIAL_DATA_LENGTH * 4); + if (index == nullptr || data == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + indexCapacity = BMP_I_LIMIT; + dataCapacity = INITIAL_DATA_LENGTH; +} + +MutableCodePointTrie::MutableCodePointTrie(const MutableCodePointTrie &other, UErrorCode &errorCode) : + index3NullOffset(other.index3NullOffset), + dataNullOffset(other.dataNullOffset), + origInitialValue(other.origInitialValue), initialValue(other.initialValue), + errorValue(other.errorValue), + highStart(other.highStart), highValue(other.highValue) +#ifdef UCPTRIE_DEBUG + , name("mutable clone") +#endif + { + if (U_FAILURE(errorCode)) { return; } + int32_t iCapacity = highStart <= BMP_LIMIT ? BMP_I_LIMIT : I_LIMIT; + index = (uint32_t *)uprv_malloc(iCapacity * 4); + data = (uint32_t *)uprv_malloc(other.dataCapacity * 4); + if (index == nullptr || data == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + indexCapacity = iCapacity; + dataCapacity = other.dataCapacity; + + int32_t iLimit = highStart >> UCPTRIE_SHIFT_3; + uprv_memcpy(flags, other.flags, iLimit); + uprv_memcpy(index, other.index, iLimit * 4); + uprv_memcpy(data, other.data, (size_t)other.dataLength * 4); + dataLength = other.dataLength; + U_ASSERT(other.index16 == nullptr); +} + +MutableCodePointTrie::~MutableCodePointTrie() { + uprv_free(index); + uprv_free(data); + uprv_free(index16); +} + +MutableCodePointTrie *MutableCodePointTrie::fromUCPMap(const UCPMap *map, UErrorCode &errorCode) { + // Use the highValue as the initialValue to reduce the highStart. + uint32_t errorValue = ucpmap_get(map, -1); + uint32_t initialValue = ucpmap_get(map, 0x10ffff); + LocalPointer mutableTrie( + new MutableCodePointTrie(initialValue, errorValue, errorCode), + errorCode); + if (U_FAILURE(errorCode)) { + return nullptr; + } + UChar32 start = 0, end; + uint32_t value; + while ((end = ucpmap_getRange(map, start, UCPMAP_RANGE_NORMAL, 0, + nullptr, nullptr, &value)) >= 0) { + if (value != initialValue) { + if (start == end) { + mutableTrie->set(start, value, errorCode); + } else { + mutableTrie->setRange(start, end, value, errorCode); + } + } + start = end + 1; + } + if (U_SUCCESS(errorCode)) { + return mutableTrie.orphan(); + } else { + return nullptr; + } +} + +MutableCodePointTrie *MutableCodePointTrie::fromUCPTrie(const UCPTrie *trie, UErrorCode &errorCode) { + // Use the highValue as the initialValue to reduce the highStart. + uint32_t errorValue; + uint32_t initialValue; + switch (trie->valueWidth) { + case UCPTRIE_VALUE_BITS_16: + errorValue = trie->data.ptr16[trie->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET]; + initialValue = trie->data.ptr16[trie->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET]; + break; + case UCPTRIE_VALUE_BITS_32: + errorValue = trie->data.ptr32[trie->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET]; + initialValue = trie->data.ptr32[trie->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET]; + break; + case UCPTRIE_VALUE_BITS_8: + errorValue = trie->data.ptr8[trie->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET]; + initialValue = trie->data.ptr8[trie->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET]; + break; + default: + // Unreachable if the trie is properly initialized. + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + LocalPointer mutableTrie( + new MutableCodePointTrie(initialValue, errorValue, errorCode), + errorCode); + if (U_FAILURE(errorCode)) { + return nullptr; + } + UChar32 start = 0, end; + uint32_t value; + while ((end = ucptrie_getRange(trie, start, UCPMAP_RANGE_NORMAL, 0, + nullptr, nullptr, &value)) >= 0) { + if (value != initialValue) { + if (start == end) { + mutableTrie->set(start, value, errorCode); + } else { + mutableTrie->setRange(start, end, value, errorCode); + } + } + start = end + 1; + } + if (U_SUCCESS(errorCode)) { + return mutableTrie.orphan(); + } else { + return nullptr; + } +} + +void MutableCodePointTrie::clear() { + index3NullOffset = dataNullOffset = -1; + dataLength = 0; + highValue = initialValue = origInitialValue; + highStart = 0; + uprv_free(index16); + index16 = nullptr; +} + +uint32_t MutableCodePointTrie::get(UChar32 c) const { + if ((uint32_t)c > MAX_UNICODE) { + return errorValue; + } + if (c >= highStart) { + return highValue; + } + int32_t i = c >> UCPTRIE_SHIFT_3; + if (flags[i] == ALL_SAME) { + return index[i]; + } else { + return data[index[i] + (c & UCPTRIE_SMALL_DATA_MASK)]; + } +} + +inline uint32_t maybeFilterValue(uint32_t value, uint32_t initialValue, uint32_t nullValue, + UCPMapValueFilter *filter, const void *context) { + if (value == initialValue) { + value = nullValue; + } else if (filter != nullptr) { + value = filter(context, value); + } + return value; +} + +UChar32 MutableCodePointTrie::getRange( + UChar32 start, UCPMapValueFilter *filter, const void *context, + uint32_t *pValue) const { + if ((uint32_t)start > MAX_UNICODE) { + return U_SENTINEL; + } + if (start >= highStart) { + if (pValue != nullptr) { + uint32_t value = highValue; + if (filter != nullptr) { value = filter(context, value); } + *pValue = value; + } + return MAX_UNICODE; + } + uint32_t nullValue = initialValue; + if (filter != nullptr) { nullValue = filter(context, nullValue); } + UChar32 c = start; + uint32_t trieValue, value; + bool haveValue = false; + int32_t i = c >> UCPTRIE_SHIFT_3; + do { + if (flags[i] == ALL_SAME) { + uint32_t trieValue2 = index[i]; + if (haveValue) { + if (trieValue2 != trieValue) { + if (filter == nullptr || + maybeFilterValue(trieValue2, initialValue, nullValue, + filter, context) != value) { + return c - 1; + } + trieValue = trieValue2; // may or may not help + } + } else { + trieValue = trieValue2; + value = maybeFilterValue(trieValue2, initialValue, nullValue, filter, context); + if (pValue != nullptr) { *pValue = value; } + haveValue = true; + } + c = (c + UCPTRIE_SMALL_DATA_BLOCK_LENGTH) & ~UCPTRIE_SMALL_DATA_MASK; + } else /* MIXED */ { + int32_t di = index[i] + (c & UCPTRIE_SMALL_DATA_MASK); + uint32_t trieValue2 = data[di]; + if (haveValue) { + if (trieValue2 != trieValue) { + if (filter == nullptr || + maybeFilterValue(trieValue2, initialValue, nullValue, + filter, context) != value) { + return c - 1; + } + trieValue = trieValue2; // may or may not help + } + } else { + trieValue = trieValue2; + value = maybeFilterValue(trieValue2, initialValue, nullValue, filter, context); + if (pValue != nullptr) { *pValue = value; } + haveValue = true; + } + while ((++c & UCPTRIE_SMALL_DATA_MASK) != 0) { + trieValue2 = data[++di]; + if (trieValue2 != trieValue) { + if (filter == nullptr || + maybeFilterValue(trieValue2, initialValue, nullValue, + filter, context) != value) { + return c - 1; + } + } + trieValue = trieValue2; // may or may not help + } + } + ++i; + } while (c < highStart); + U_ASSERT(haveValue); + if (maybeFilterValue(highValue, initialValue, nullValue, + filter, context) != value) { + return c - 1; + } else { + return MAX_UNICODE; + } +} + +void +writeBlock(uint32_t *block, uint32_t value) { + uint32_t *limit = block + UCPTRIE_SMALL_DATA_BLOCK_LENGTH; + while (block < limit) { + *block++ = value; + } +} + +bool MutableCodePointTrie::ensureHighStart(UChar32 c) { + if (c >= highStart) { + // Round up to a UCPTRIE_CP_PER_INDEX_2_ENTRY boundary to simplify compaction. + c = (c + UCPTRIE_CP_PER_INDEX_2_ENTRY) & ~(UCPTRIE_CP_PER_INDEX_2_ENTRY - 1); + int32_t i = highStart >> UCPTRIE_SHIFT_3; + int32_t iLimit = c >> UCPTRIE_SHIFT_3; + if (iLimit > indexCapacity) { + uint32_t *newIndex = (uint32_t *)uprv_malloc(I_LIMIT * 4); + if (newIndex == nullptr) { return false; } + uprv_memcpy(newIndex, index, i * 4); + uprv_free(index); + index = newIndex; + indexCapacity = I_LIMIT; + } + do { + flags[i] = ALL_SAME; + index[i] = initialValue; + } while(++i < iLimit); + highStart = c; + } + return true; +} + +int32_t MutableCodePointTrie::allocDataBlock(int32_t blockLength) { + int32_t newBlock = dataLength; + int32_t newTop = newBlock + blockLength; + if (newTop > dataCapacity) { + int32_t capacity; + if (dataCapacity < MEDIUM_DATA_LENGTH) { + capacity = MEDIUM_DATA_LENGTH; + } else if (dataCapacity < MAX_DATA_LENGTH) { + capacity = MAX_DATA_LENGTH; + } else { + // Should never occur. + // Either MAX_DATA_LENGTH is incorrect, + // or the code writes more values than should be possible. + return -1; + } + uint32_t *newData = (uint32_t *)uprv_malloc(capacity * 4); + if (newData == nullptr) { + return -1; + } + uprv_memcpy(newData, data, (size_t)dataLength * 4); + uprv_free(data); + data = newData; + dataCapacity = capacity; + } + dataLength = newTop; + return newBlock; +} + +/** + * No error checking for illegal arguments. + * + * @return -1 if no new data block available (out of memory in data array) + * @internal + */ +int32_t MutableCodePointTrie::getDataBlock(int32_t i) { + if (flags[i] == MIXED) { + return index[i]; + } + if (i < BMP_I_LIMIT) { + int32_t newBlock = allocDataBlock(UCPTRIE_FAST_DATA_BLOCK_LENGTH); + if (newBlock < 0) { return newBlock; } + int32_t iStart = i & ~(SMALL_DATA_BLOCKS_PER_BMP_BLOCK -1); + int32_t iLimit = iStart + SMALL_DATA_BLOCKS_PER_BMP_BLOCK; + do { + U_ASSERT(flags[iStart] == ALL_SAME); + writeBlock(data + newBlock, index[iStart]); + flags[iStart] = MIXED; + index[iStart++] = newBlock; + newBlock += UCPTRIE_SMALL_DATA_BLOCK_LENGTH; + } while (iStart < iLimit); + return index[i]; + } else { + int32_t newBlock = allocDataBlock(UCPTRIE_SMALL_DATA_BLOCK_LENGTH); + if (newBlock < 0) { return newBlock; } + writeBlock(data + newBlock, index[i]); + flags[i] = MIXED; + index[i] = newBlock; + return newBlock; + } +} + +void MutableCodePointTrie::set(UChar32 c, uint32_t value, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { + return; + } + if ((uint32_t)c > MAX_UNICODE) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + int32_t block; + if (!ensureHighStart(c) || (block = getDataBlock(c >> UCPTRIE_SHIFT_3)) < 0) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + + data[block + (c & UCPTRIE_SMALL_DATA_MASK)] = value; +} + +void +fillBlock(uint32_t *block, UChar32 start, UChar32 limit, uint32_t value) { + uint32_t *pLimit = block + limit; + block += start; + while (block < pLimit) { + *block++ = value; + } +} + +void MutableCodePointTrie::setRange(UChar32 start, UChar32 end, uint32_t value, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { + return; + } + if ((uint32_t)start > MAX_UNICODE || (uint32_t)end > MAX_UNICODE || start > end) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return; + } + if (!ensureHighStart(end)) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + + UChar32 limit = end + 1; + if (start & UCPTRIE_SMALL_DATA_MASK) { + // Set partial block at [start..following block boundary[. + int32_t block = getDataBlock(start >> UCPTRIE_SHIFT_3); + if (block < 0) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + + UChar32 nextStart = (start + UCPTRIE_SMALL_DATA_MASK) & ~UCPTRIE_SMALL_DATA_MASK; + if (nextStart <= limit) { + fillBlock(data + block, start & UCPTRIE_SMALL_DATA_MASK, UCPTRIE_SMALL_DATA_BLOCK_LENGTH, + value); + start = nextStart; + } else { + fillBlock(data + block, start & UCPTRIE_SMALL_DATA_MASK, limit & UCPTRIE_SMALL_DATA_MASK, + value); + return; + } + } + + // Number of positions in the last, partial block. + int32_t rest = limit & UCPTRIE_SMALL_DATA_MASK; + + // Round down limit to a block boundary. + limit &= ~UCPTRIE_SMALL_DATA_MASK; + + // Iterate over all-value blocks. + while (start < limit) { + int32_t i = start >> UCPTRIE_SHIFT_3; + if (flags[i] == ALL_SAME) { + index[i] = value; + } else /* MIXED */ { + fillBlock(data + index[i], 0, UCPTRIE_SMALL_DATA_BLOCK_LENGTH, value); + } + start += UCPTRIE_SMALL_DATA_BLOCK_LENGTH; + } + + if (rest > 0) { + // Set partial block at [last block boundary..limit[. + int32_t block = getDataBlock(start >> UCPTRIE_SHIFT_3); + if (block < 0) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return; + } + + fillBlock(data + block, 0, rest, value); + } +} + +/* compaction --------------------------------------------------------------- */ + +void MutableCodePointTrie::maskValues(uint32_t mask) { + initialValue &= mask; + errorValue &= mask; + highValue &= mask; + int32_t iLimit = highStart >> UCPTRIE_SHIFT_3; + for (int32_t i = 0; i < iLimit; ++i) { + if (flags[i] == ALL_SAME) { + index[i] &= mask; + } + } + for (int32_t i = 0; i < dataLength; ++i) { + data[i] &= mask; + } +} + +template +bool equalBlocks(const UIntA *s, const UIntB *t, int32_t length) { + while (length > 0 && *s == *t) { + ++s; + ++t; + --length; + } + return length == 0; +} + +bool allValuesSameAs(const uint32_t *p, int32_t length, uint32_t value) { + const uint32_t *pLimit = p + length; + while (p < pLimit && *p == value) { ++p; } + return p == pLimit; +} + +/** Search for an identical block. */ +int32_t findSameBlock(const uint16_t *p, int32_t pStart, int32_t length, + const uint16_t *q, int32_t qStart, int32_t blockLength) { + // Ensure that we do not even partially get past length. + length -= blockLength; + + q += qStart; + while (pStart <= length) { + if (equalBlocks(p + pStart, q, blockLength)) { + return pStart; + } + ++pStart; + } + return -1; +} + +int32_t findAllSameBlock(const uint32_t *p, int32_t start, int32_t limit, + uint32_t value, int32_t blockLength) { + // Ensure that we do not even partially get past limit. + limit -= blockLength; + + for (int32_t block = start; block <= limit; ++block) { + if (p[block] == value) { + for (int32_t i = 1;; ++i) { + if (i == blockLength) { + return block; + } + if (p[block + i] != value) { + block += i; + break; + } + } + } + } + return -1; +} + +/** + * Look for maximum overlap of the beginning of the other block + * with the previous, adjacent block. + */ +template +int32_t getOverlap(const UIntA *p, int32_t length, + const UIntB *q, int32_t qStart, int32_t blockLength) { + int32_t overlap = blockLength - 1; + U_ASSERT(overlap <= length); + q += qStart; + while (overlap > 0 && !equalBlocks(p + (length - overlap), q, overlap)) { + --overlap; + } + return overlap; +} + +int32_t getAllSameOverlap(const uint32_t *p, int32_t length, uint32_t value, + int32_t blockLength) { + int32_t min = length - (blockLength - 1); + int32_t i = length; + while (min < i && p[i - 1] == value) { --i; } + return length - i; +} + +bool isStartOfSomeFastBlock(uint32_t dataOffset, const uint32_t index[], int32_t fastILimit) { + for (int32_t i = 0; i < fastILimit; i += SMALL_DATA_BLOCKS_PER_BMP_BLOCK) { + if (index[i] == dataOffset) { + return true; + } + } + return false; +} + +/** + * Finds the start of the last range in the trie by enumerating backward. + * Indexes for code points higher than this will be omitted. + */ +UChar32 MutableCodePointTrie::findHighStart() const { + int32_t i = highStart >> UCPTRIE_SHIFT_3; + while (i > 0) { + bool match; + if (flags[--i] == ALL_SAME) { + match = index[i] == highValue; + } else /* MIXED */ { + const uint32_t *p = data + index[i]; + for (int32_t j = 0;; ++j) { + if (j == UCPTRIE_SMALL_DATA_BLOCK_LENGTH) { + match = true; + break; + } + if (p[j] != highValue) { + match = false; + break; + } + } + } + if (!match) { + return (i + 1) << UCPTRIE_SHIFT_3; + } + } + return 0; +} + +class AllSameBlocks { +public: + static constexpr int32_t NEW_UNIQUE = -1; + static constexpr int32_t OVERFLOW = -2; + + AllSameBlocks() : length(0), mostRecent(-1) {} + + int32_t findOrAdd(int32_t index, int32_t count, uint32_t value) { + if (mostRecent >= 0 && values[mostRecent] == value) { + refCounts[mostRecent] += count; + return indexes[mostRecent]; + } + for (int32_t i = 0; i < length; ++i) { + if (values[i] == value) { + mostRecent = i; + refCounts[i] += count; + return indexes[i]; + } + } + if (length == CAPACITY) { + return OVERFLOW; + } + mostRecent = length; + indexes[length] = index; + values[length] = value; + refCounts[length++] = count; + return NEW_UNIQUE; + } + + /** Replaces the block which has the lowest reference count. */ + void add(int32_t index, int32_t count, uint32_t value) { + U_ASSERT(length == CAPACITY); + int32_t least = -1; + int32_t leastCount = I_LIMIT; + for (int32_t i = 0; i < length; ++i) { + U_ASSERT(values[i] != value); + if (refCounts[i] < leastCount) { + least = i; + leastCount = refCounts[i]; + } + } + U_ASSERT(least >= 0); + mostRecent = least; + indexes[least] = index; + values[least] = value; + refCounts[least] = count; + } + + int32_t findMostUsed() const { + if (length == 0) { return -1; } + int32_t max = -1; + int32_t maxCount = 0; + for (int32_t i = 0; i < length; ++i) { + if (refCounts[i] > maxCount) { + max = i; + maxCount = refCounts[i]; + } + } + return indexes[max]; + } + +private: + static constexpr int32_t CAPACITY = 32; + + int32_t length; + int32_t mostRecent; + + int32_t indexes[CAPACITY]; + uint32_t values[CAPACITY]; + int32_t refCounts[CAPACITY]; +}; + +// Custom hash table for mixed-value blocks to be found anywhere in the +// compacted data or index so far. +class MixedBlocks { +public: + MixedBlocks() {} + ~MixedBlocks() { + uprv_free(table); + } + + bool init(int32_t maxLength, int32_t newBlockLength) { + // We store actual data indexes + 1 to reserve 0 for empty entries. + int32_t maxDataIndex = maxLength - newBlockLength + 1; + int32_t newLength; + if (maxDataIndex <= 0xfff) { // 4k + newLength = 6007; + shift = 12; + mask = 0xfff; + } else if (maxDataIndex <= 0x7fff) { // 32k + newLength = 50021; + shift = 15; + mask = 0x7fff; + } else if (maxDataIndex <= 0x1ffff) { // 128k + newLength = 200003; + shift = 17; + mask = 0x1ffff; + } else { + // maxDataIndex up to around MAX_DATA_LENGTH, ca. 1.1M + newLength = 1500007; + shift = 21; + mask = 0x1fffff; + } + if (newLength > capacity) { + uprv_free(table); + table = (uint32_t *)uprv_malloc(newLength * 4); + if (table == nullptr) { + return false; + } + capacity = newLength; + } + length = newLength; + uprv_memset(table, 0, length * 4); + + blockLength = newBlockLength; + return true; + } + + template + void extend(const UInt *data, int32_t minStart, int32_t prevDataLength, int32_t newDataLength) { + int32_t start = prevDataLength - blockLength; + if (start >= minStart) { + ++start; // Skip the last block that we added last time. + } else { + start = minStart; // Begin with the first full block. + } + for (int32_t end = newDataLength - blockLength; start <= end; ++start) { + uint32_t hashCode = makeHashCode(data, start); + addEntry(data, start, hashCode, start); + } + } + + template + int32_t findBlock(const UIntA *data, const UIntB *blockData, int32_t blockStart) const { + uint32_t hashCode = makeHashCode(blockData, blockStart); + int32_t entryIndex = findEntry(data, blockData, blockStart, hashCode); + if (entryIndex >= 0) { + return (table[entryIndex] & mask) - 1; + } else { + return -1; + } + } + + int32_t findAllSameBlock(const uint32_t *data, uint32_t blockValue) const { + uint32_t hashCode = makeHashCode(blockValue); + int32_t entryIndex = findEntry(data, blockValue, hashCode); + if (entryIndex >= 0) { + return (table[entryIndex] & mask) - 1; + } else { + return -1; + } + } + +private: + template + uint32_t makeHashCode(const UInt *blockData, int32_t blockStart) const { + int32_t blockLimit = blockStart + blockLength; + uint32_t hashCode = blockData[blockStart++]; + do { + hashCode = 37 * hashCode + blockData[blockStart++]; + } while (blockStart < blockLimit); + return hashCode; + } + + uint32_t makeHashCode(uint32_t blockValue) const { + uint32_t hashCode = blockValue; + for (int32_t i = 1; i < blockLength; ++i) { + hashCode = 37 * hashCode + blockValue; + } + return hashCode; + } + + template + void addEntry(const UInt *data, int32_t blockStart, uint32_t hashCode, int32_t dataIndex) { + U_ASSERT(0 <= dataIndex && dataIndex < (int32_t)mask); + int32_t entryIndex = findEntry(data, data, blockStart, hashCode); + if (entryIndex < 0) { + table[~entryIndex] = (hashCode << shift) | (dataIndex + 1); + } + } + + template + int32_t findEntry(const UIntA *data, const UIntB *blockData, int32_t blockStart, + uint32_t hashCode) const { + uint32_t shiftedHashCode = hashCode << shift; + int32_t initialEntryIndex = (hashCode % (length - 1)) + 1; // 1..length-1 + for (int32_t entryIndex = initialEntryIndex;;) { + uint32_t entry = table[entryIndex]; + if (entry == 0) { + return ~entryIndex; + } + if ((entry & ~mask) == shiftedHashCode) { + int32_t dataIndex = (entry & mask) - 1; + if (equalBlocks(data + dataIndex, blockData + blockStart, blockLength)) { + return entryIndex; + } + } + entryIndex = nextIndex(initialEntryIndex, entryIndex); + } + } + + int32_t findEntry(const uint32_t *data, uint32_t blockValue, uint32_t hashCode) const { + uint32_t shiftedHashCode = hashCode << shift; + int32_t initialEntryIndex = (hashCode % (length - 1)) + 1; // 1..length-1 + for (int32_t entryIndex = initialEntryIndex;;) { + uint32_t entry = table[entryIndex]; + if (entry == 0) { + return ~entryIndex; + } + if ((entry & ~mask) == shiftedHashCode) { + int32_t dataIndex = (entry & mask) - 1; + if (allValuesSameAs(data + dataIndex, blockLength, blockValue)) { + return entryIndex; + } + } + entryIndex = nextIndex(initialEntryIndex, entryIndex); + } + } + + inline int32_t nextIndex(int32_t initialEntryIndex, int32_t entryIndex) const { + // U_ASSERT(0 < initialEntryIndex && initialEntryIndex < length); + return (entryIndex + initialEntryIndex) % length; + } + + // Hash table. + // The length is a prime number, larger than the maximum data length. + // The "shift" lower bits store a data index + 1. + // The remaining upper bits store a partial hashCode of the block data values. + uint32_t *table = nullptr; + int32_t capacity = 0; + int32_t length = 0; + int32_t shift = 0; + uint32_t mask = 0; + + int32_t blockLength = 0; +}; + +int32_t MutableCodePointTrie::compactWholeDataBlocks(int32_t fastILimit, AllSameBlocks &allSameBlocks) { +#ifdef UCPTRIE_DEBUG + bool overflow = false; +#endif + + // ASCII data will be stored as a linear table, even if the following code + // does not yet count it that way. + int32_t newDataCapacity = ASCII_LIMIT; + // Add room for a small data null block in case it would match the start of + // a fast data block where dataNullOffset must not be set in that case. + newDataCapacity += UCPTRIE_SMALL_DATA_BLOCK_LENGTH; + // Add room for special values (errorValue, highValue) and padding. + newDataCapacity += 4; + int32_t iLimit = highStart >> UCPTRIE_SHIFT_3; + int32_t blockLength = UCPTRIE_FAST_DATA_BLOCK_LENGTH; + int32_t inc = SMALL_DATA_BLOCKS_PER_BMP_BLOCK; + for (int32_t i = 0; i < iLimit; i += inc) { + if (i == fastILimit) { + blockLength = UCPTRIE_SMALL_DATA_BLOCK_LENGTH; + inc = 1; + } + uint32_t value = index[i]; + if (flags[i] == MIXED) { + // Really mixed? + const uint32_t *p = data + value; + value = *p; + if (allValuesSameAs(p + 1, blockLength - 1, value)) { + flags[i] = ALL_SAME; + index[i] = value; + // Fall through to ALL_SAME handling. + } else { + newDataCapacity += blockLength; + continue; + } + } else { + U_ASSERT(flags[i] == ALL_SAME); + if (inc > 1) { + // Do all of the fast-range data block's ALL_SAME parts have the same value? + bool allSame = true; + int32_t next_i = i + inc; + for (int32_t j = i + 1; j < next_i; ++j) { + U_ASSERT(flags[j] == ALL_SAME); + if (index[j] != value) { + allSame = false; + break; + } + } + if (!allSame) { + // Turn it into a MIXED block. + if (getDataBlock(i) < 0) { + return -1; + } + newDataCapacity += blockLength; + continue; + } + } + } + // Is there another ALL_SAME block with the same value? + int32_t other = allSameBlocks.findOrAdd(i, inc, value); + if (other == AllSameBlocks::OVERFLOW) { + // The fixed-size array overflowed. Slow check for a duplicate block. +#ifdef UCPTRIE_DEBUG + if (!overflow) { + puts("UCPTrie AllSameBlocks overflow"); + overflow = true; + } +#endif + int32_t jInc = SMALL_DATA_BLOCKS_PER_BMP_BLOCK; + for (int32_t j = 0;; j += jInc) { + if (j == i) { + allSameBlocks.add(i, inc, value); + break; + } + if (j == fastILimit) { + jInc = 1; + } + if (flags[j] == ALL_SAME && index[j] == value) { + allSameBlocks.add(j, jInc + inc, value); + other = j; + break; + // We could keep counting blocks with the same value + // before we add the first one, which may improve compaction in rare cases, + // but it would make it slower. + } + } + } + if (other >= 0) { + flags[i] = SAME_AS; + index[i] = other; + } else { + // New unique same-value block. + newDataCapacity += blockLength; + } + } + return newDataCapacity; +} + +#ifdef UCPTRIE_DEBUG +# define DEBUG_DO(expr) expr +#else +# define DEBUG_DO(expr) +#endif + +#ifdef UCPTRIE_DEBUG +// Braille symbols: U+28xx = UTF-8 E2 A0 80..E2 A3 BF +int32_t appendValue(char s[], int32_t length, uint32_t value) { + value ^= value >> 16; + value ^= value >> 8; + s[length] = 0xE2; + s[length + 1] = (char)(0xA0 + ((value >> 6) & 3)); + s[length + 2] = (char)(0x80 + (value & 0x3F)); + return length + 3; +} + +void printBlock(const uint32_t *block, int32_t blockLength, uint32_t value, + UChar32 start, int32_t overlap, uint32_t initialValue) { + char s[UCPTRIE_FAST_DATA_BLOCK_LENGTH * 3 + 3]; + int32_t length = 0; + int32_t i; + for (i = 0; i < overlap; ++i) { + length = appendValue(s, length, 0); // Braille blank + } + s[length++] = '|'; + for (; i < blockLength; ++i) { + if (block != nullptr) { + value = block[i]; + } + if (value == initialValue) { + value = 0x40; // Braille lower left dot + } + length = appendValue(s, length, value); + } + s[length] = 0; + start += overlap; + if (start <= 0xffff) { + printf(" %04lX %s|\n", (long)start, s); + } else if (start <= 0xfffff) { + printf(" %5lX %s|\n", (long)start, s); + } else { + printf(" %6lX %s|\n", (long)start, s); + } +} +#endif + +/** + * Compacts a build-time trie. + * + * The compaction + * - removes blocks that are identical with earlier ones + * - overlaps each new non-duplicate block as much as possible with the previously-written one + * - works with fast-range data blocks whose length is a multiple of that of + * higher-code-point data blocks + * + * It does not try to find an optimal order of writing, deduplicating, and overlapping blocks. + */ +int32_t MutableCodePointTrie::compactData( + int32_t fastILimit, uint32_t *newData, int32_t newDataCapacity, + int32_t dataNullIndex, MixedBlocks &mixedBlocks, UErrorCode &errorCode) { +#ifdef UCPTRIE_DEBUG + int32_t countSame=0, sumOverlaps=0; + bool printData = dataLength == 29088 /* line.brk */ || + // dataLength == 30048 /* CanonIterData */ || + dataLength == 50400 /* zh.txt~stroke */; +#endif + + // The linear ASCII data has been copied into newData already. + int32_t newDataLength = 0; + for (int32_t i = 0; newDataLength < ASCII_LIMIT; + newDataLength += UCPTRIE_FAST_DATA_BLOCK_LENGTH, i += SMALL_DATA_BLOCKS_PER_BMP_BLOCK) { + index[i] = newDataLength; +#ifdef UCPTRIE_DEBUG + if (printData) { + printBlock(newData + newDataLength, UCPTRIE_FAST_DATA_BLOCK_LENGTH, 0, newDataLength, 0, initialValue); + } +#endif + } + + int32_t blockLength = UCPTRIE_FAST_DATA_BLOCK_LENGTH; + if (!mixedBlocks.init(newDataCapacity, blockLength)) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return 0; + } + mixedBlocks.extend(newData, 0, 0, newDataLength); + + int32_t iLimit = highStart >> UCPTRIE_SHIFT_3; + int32_t inc = SMALL_DATA_BLOCKS_PER_BMP_BLOCK; + int32_t fastLength = 0; + for (int32_t i = ASCII_I_LIMIT; i < iLimit; i += inc) { + if (i == fastILimit) { + blockLength = UCPTRIE_SMALL_DATA_BLOCK_LENGTH; + inc = 1; + fastLength = newDataLength; + if (!mixedBlocks.init(newDataCapacity, blockLength)) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return 0; + } + mixedBlocks.extend(newData, 0, 0, newDataLength); + } + if (flags[i] == ALL_SAME) { + uint32_t value = index[i]; + // Find an earlier part of the data array of length blockLength + // that is filled with this value. + int32_t n = mixedBlocks.findAllSameBlock(newData, value); + // If we find a match, and the current block is the data null block, + // and it is not a fast block but matches the start of a fast block, + // then we need to continue looking. + // This is because this small block is shorter than the fast block, + // and not all of the rest of the fast block is filled with this value. + // Otherwise trie.getRange() would detect that the fast block starts at + // dataNullOffset and assume incorrectly that it is filled with the null value. + while (n >= 0 && i == dataNullIndex && i >= fastILimit && n < fastLength && + isStartOfSomeFastBlock(n, index, fastILimit)) { + n = findAllSameBlock(newData, n + 1, newDataLength, value, blockLength); + } + if (n >= 0) { + DEBUG_DO(++countSame); + index[i] = n; + } else { + n = getAllSameOverlap(newData, newDataLength, value, blockLength); + DEBUG_DO(sumOverlaps += n); +#ifdef UCPTRIE_DEBUG + if (printData) { + printBlock(nullptr, blockLength, value, i << UCPTRIE_SHIFT_3, n, initialValue); + } +#endif + index[i] = newDataLength - n; + int32_t prevDataLength = newDataLength; + while (n < blockLength) { + newData[newDataLength++] = value; + ++n; + } + mixedBlocks.extend(newData, 0, prevDataLength, newDataLength); + } + } else if (flags[i] == MIXED) { + const uint32_t *block = data + index[i]; + int32_t n = mixedBlocks.findBlock(newData, block, 0); + if (n >= 0) { + DEBUG_DO(++countSame); + index[i] = n; + } else { + n = getOverlap(newData, newDataLength, block, 0, blockLength); + DEBUG_DO(sumOverlaps += n); +#ifdef UCPTRIE_DEBUG + if (printData) { + printBlock(block, blockLength, 0, i << UCPTRIE_SHIFT_3, n, initialValue); + } +#endif + index[i] = newDataLength - n; + int32_t prevDataLength = newDataLength; + while (n < blockLength) { + newData[newDataLength++] = block[n++]; + } + mixedBlocks.extend(newData, 0, prevDataLength, newDataLength); + } + } else /* SAME_AS */ { + uint32_t j = index[i]; + index[i] = index[j]; + } + } + +#ifdef UCPTRIE_DEBUG + /* we saved some space */ + printf("compacting UCPTrie: count of 32-bit data words %lu->%lu countSame=%ld sumOverlaps=%ld\n", + (long)dataLength, (long)newDataLength, (long)countSame, (long)sumOverlaps); +#endif + return newDataLength; +} + +int32_t MutableCodePointTrie::compactIndex(int32_t fastILimit, MixedBlocks &mixedBlocks, + UErrorCode &errorCode) { + int32_t fastIndexLength = fastILimit >> (UCPTRIE_FAST_SHIFT - UCPTRIE_SHIFT_3); + if ((highStart >> UCPTRIE_FAST_SHIFT) <= fastIndexLength) { + // Only the linear fast index, no multi-stage index tables. + index3NullOffset = UCPTRIE_NO_INDEX3_NULL_OFFSET; + return fastIndexLength; + } + + // Condense the fast index table. + // Also, does it contain an index-3 block with all dataNullOffset? + uint16_t fastIndex[UCPTRIE_BMP_INDEX_LENGTH]; // fastIndexLength + int32_t i3FirstNull = -1; + for (int32_t i = 0, j = 0; i < fastILimit; ++j) { + uint32_t i3 = index[i]; + fastIndex[j] = (uint16_t)i3; + if (i3 == (uint32_t)dataNullOffset) { + if (i3FirstNull < 0) { + i3FirstNull = j; + } else if (index3NullOffset < 0 && + (j - i3FirstNull + 1) == UCPTRIE_INDEX_3_BLOCK_LENGTH) { + index3NullOffset = i3FirstNull; + } + } else { + i3FirstNull = -1; + } + // Set the index entries that compactData() skipped. + // Needed when the multi-stage index covers the fast index range as well. + int32_t iNext = i + SMALL_DATA_BLOCKS_PER_BMP_BLOCK; + while (++i < iNext) { + i3 += UCPTRIE_SMALL_DATA_BLOCK_LENGTH; + index[i] = i3; + } + } + + if (!mixedBlocks.init(fastIndexLength, UCPTRIE_INDEX_3_BLOCK_LENGTH)) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return 0; + } + mixedBlocks.extend(fastIndex, 0, 0, fastIndexLength); + + // Examine index-3 blocks. For each determine one of: + // - same as the index-3 null block + // - same as a fast-index block + // - 16-bit indexes + // - 18-bit indexes + // We store this in the first flags entry for the index-3 block. + // + // Also determine an upper limit for the index-3 table length. + int32_t index3Capacity = 0; + i3FirstNull = index3NullOffset; + bool hasLongI3Blocks = false; + // If the fast index covers the whole BMP, then + // the multi-stage index is only for supplementary code points. + // Otherwise, the multi-stage index covers all of Unicode. + int32_t iStart = fastILimit < BMP_I_LIMIT ? 0 : BMP_I_LIMIT; + int32_t iLimit = highStart >> UCPTRIE_SHIFT_3; + for (int32_t i = iStart; i < iLimit;) { + int32_t j = i; + int32_t jLimit = i + UCPTRIE_INDEX_3_BLOCK_LENGTH; + uint32_t oredI3 = 0; + bool isNull = true; + do { + uint32_t i3 = index[j]; + oredI3 |= i3; + if (i3 != (uint32_t)dataNullOffset) { + isNull = false; + } + } while (++j < jLimit); + if (isNull) { + flags[i] = I3_NULL; + if (i3FirstNull < 0) { + if (oredI3 <= 0xffff) { + index3Capacity += UCPTRIE_INDEX_3_BLOCK_LENGTH; + } else { + index3Capacity += INDEX_3_18BIT_BLOCK_LENGTH; + hasLongI3Blocks = true; + } + i3FirstNull = 0; + } + } else { + if (oredI3 <= 0xffff) { + int32_t n = mixedBlocks.findBlock(fastIndex, index, i); + if (n >= 0) { + flags[i] = I3_BMP; + index[i] = n; + } else { + flags[i] = I3_16; + index3Capacity += UCPTRIE_INDEX_3_BLOCK_LENGTH; + } + } else { + flags[i] = I3_18; + index3Capacity += INDEX_3_18BIT_BLOCK_LENGTH; + hasLongI3Blocks = true; + } + } + i = j; + } + + int32_t index2Capacity = (iLimit - iStart) >> UCPTRIE_SHIFT_2_3; + + // Length of the index-1 table, rounded up. + int32_t index1Length = (index2Capacity + UCPTRIE_INDEX_2_MASK) >> UCPTRIE_SHIFT_1_2; + + // Index table: Fast index, index-1, index-3, index-2. + // +1 for possible index table padding. + int32_t index16Capacity = fastIndexLength + index1Length + index3Capacity + index2Capacity + 1; + index16 = (uint16_t *)uprv_malloc(index16Capacity * 2); + if (index16 == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return 0; + } + uprv_memcpy(index16, fastIndex, fastIndexLength * 2); + + if (!mixedBlocks.init(index16Capacity, UCPTRIE_INDEX_3_BLOCK_LENGTH)) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return 0; + } + MixedBlocks longI3Blocks; + if (hasLongI3Blocks) { + if (!longI3Blocks.init(index16Capacity, INDEX_3_18BIT_BLOCK_LENGTH)) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return 0; + } + } + + // Compact the index-3 table and write an uncompacted version of the index-2 table. + uint16_t index2[UNICODE_LIMIT >> UCPTRIE_SHIFT_2]; // index2Capacity + int32_t i2Length = 0; + i3FirstNull = index3NullOffset; + int32_t index3Start = fastIndexLength + index1Length; + int32_t indexLength = index3Start; + for (int32_t i = iStart; i < iLimit; i += UCPTRIE_INDEX_3_BLOCK_LENGTH) { + int32_t i3; + uint8_t f = flags[i]; + if (f == I3_NULL && i3FirstNull < 0) { + // First index-3 null block. Write & overlap it like a normal block, then remember it. + f = dataNullOffset <= 0xffff ? I3_16 : I3_18; + i3FirstNull = 0; + } + if (f == I3_NULL) { + i3 = index3NullOffset; + } else if (f == I3_BMP) { + i3 = index[i]; + } else if (f == I3_16) { + int32_t n = mixedBlocks.findBlock(index16, index, i); + if (n >= 0) { + i3 = n; + } else { + if (indexLength == index3Start) { + // No overlap at the boundary between the index-1 and index-3 tables. + n = 0; + } else { + n = getOverlap(index16, indexLength, + index, i, UCPTRIE_INDEX_3_BLOCK_LENGTH); + } + i3 = indexLength - n; + int32_t prevIndexLength = indexLength; + while (n < UCPTRIE_INDEX_3_BLOCK_LENGTH) { + index16[indexLength++] = index[i + n++]; + } + mixedBlocks.extend(index16, index3Start, prevIndexLength, indexLength); + if (hasLongI3Blocks) { + longI3Blocks.extend(index16, index3Start, prevIndexLength, indexLength); + } + } + } else { + U_ASSERT(f == I3_18); + U_ASSERT(hasLongI3Blocks); + // Encode an index-3 block that contains one or more data indexes exceeding 16 bits. + int32_t j = i; + int32_t jLimit = i + UCPTRIE_INDEX_3_BLOCK_LENGTH; + int32_t k = indexLength; + do { + ++k; + uint32_t v = index[j++]; + uint32_t upperBits = (v & 0x30000) >> 2; + index16[k++] = v; + v = index[j++]; + upperBits |= (v & 0x30000) >> 4; + index16[k++] = v; + v = index[j++]; + upperBits |= (v & 0x30000) >> 6; + index16[k++] = v; + v = index[j++]; + upperBits |= (v & 0x30000) >> 8; + index16[k++] = v; + v = index[j++]; + upperBits |= (v & 0x30000) >> 10; + index16[k++] = v; + v = index[j++]; + upperBits |= (v & 0x30000) >> 12; + index16[k++] = v; + v = index[j++]; + upperBits |= (v & 0x30000) >> 14; + index16[k++] = v; + v = index[j++]; + upperBits |= (v & 0x30000) >> 16; + index16[k++] = v; + index16[k - 9] = upperBits; + } while (j < jLimit); + int32_t n = longI3Blocks.findBlock(index16, index16, indexLength); + if (n >= 0) { + i3 = n | 0x8000; + } else { + if (indexLength == index3Start) { + // No overlap at the boundary between the index-1 and index-3 tables. + n = 0; + } else { + n = getOverlap(index16, indexLength, + index16, indexLength, INDEX_3_18BIT_BLOCK_LENGTH); + } + i3 = (indexLength - n) | 0x8000; + int32_t prevIndexLength = indexLength; + if (n > 0) { + int32_t start = indexLength; + while (n < INDEX_3_18BIT_BLOCK_LENGTH) { + index16[indexLength++] = index16[start + n++]; + } + } else { + indexLength += INDEX_3_18BIT_BLOCK_LENGTH; + } + mixedBlocks.extend(index16, index3Start, prevIndexLength, indexLength); + if (hasLongI3Blocks) { + longI3Blocks.extend(index16, index3Start, prevIndexLength, indexLength); + } + } + } + if (index3NullOffset < 0 && i3FirstNull >= 0) { + index3NullOffset = i3; + } + // Set the index-2 table entry. + index2[i2Length++] = i3; + } + U_ASSERT(i2Length == index2Capacity); + U_ASSERT(indexLength <= index3Start + index3Capacity); + + if (index3NullOffset < 0) { + index3NullOffset = UCPTRIE_NO_INDEX3_NULL_OFFSET; + } + if (indexLength >= (UCPTRIE_NO_INDEX3_NULL_OFFSET + UCPTRIE_INDEX_3_BLOCK_LENGTH)) { + // The index-3 offsets exceed 15 bits, or + // the last one cannot be distinguished from the no-null-block value. + errorCode = U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + + // Compact the index-2 table and write the index-1 table. + static_assert(UCPTRIE_INDEX_2_BLOCK_LENGTH == UCPTRIE_INDEX_3_BLOCK_LENGTH, + "must re-init mixedBlocks"); + int32_t blockLength = UCPTRIE_INDEX_2_BLOCK_LENGTH; + int32_t i1 = fastIndexLength; + for (int32_t i = 0; i < i2Length; i += blockLength) { + int32_t n; + if ((i2Length - i) >= blockLength) { + // normal block + U_ASSERT(blockLength == UCPTRIE_INDEX_2_BLOCK_LENGTH); + n = mixedBlocks.findBlock(index16, index2, i); + } else { + // highStart is inside the last index-2 block. Shorten it. + blockLength = i2Length - i; + n = findSameBlock(index16, index3Start, indexLength, + index2, i, blockLength); + } + int32_t i2; + if (n >= 0) { + i2 = n; + } else { + if (indexLength == index3Start) { + // No overlap at the boundary between the index-1 and index-3/2 tables. + n = 0; + } else { + n = getOverlap(index16, indexLength, index2, i, blockLength); + } + i2 = indexLength - n; + int32_t prevIndexLength = indexLength; + while (n < blockLength) { + index16[indexLength++] = index2[i + n++]; + } + mixedBlocks.extend(index16, index3Start, prevIndexLength, indexLength); + } + // Set the index-1 table entry. + index16[i1++] = i2; + } + U_ASSERT(i1 == index3Start); + U_ASSERT(indexLength <= index16Capacity); + +#ifdef UCPTRIE_DEBUG + /* we saved some space */ + printf("compacting UCPTrie: count of 16-bit index words %lu->%lu\n", + (long)iLimit, (long)indexLength); +#endif + + return indexLength; +} + +int32_t MutableCodePointTrie::compactTrie(int32_t fastILimit, UErrorCode &errorCode) { + // Find the real highStart and round it up. + U_ASSERT((highStart & (UCPTRIE_CP_PER_INDEX_2_ENTRY - 1)) == 0); + highValue = get(MAX_UNICODE); + int32_t realHighStart = findHighStart(); + realHighStart = (realHighStart + (UCPTRIE_CP_PER_INDEX_2_ENTRY - 1)) & + ~(UCPTRIE_CP_PER_INDEX_2_ENTRY - 1); + if (realHighStart == UNICODE_LIMIT) { + highValue = initialValue; + } + +#ifdef UCPTRIE_DEBUG + printf("UCPTrie: highStart U+%06lx highValue 0x%lx initialValue 0x%lx\n", + (long)realHighStart, (long)highValue, (long)initialValue); +#endif + + // We always store indexes and data values for the fast range. + // Pin highStart to the top of that range while building. + UChar32 fastLimit = fastILimit << UCPTRIE_SHIFT_3; + if (realHighStart < fastLimit) { + for (int32_t i = (realHighStart >> UCPTRIE_SHIFT_3); i < fastILimit; ++i) { + flags[i] = ALL_SAME; + index[i] = highValue; + } + highStart = fastLimit; + } else { + highStart = realHighStart; + } + + uint32_t asciiData[ASCII_LIMIT]; + for (int32_t i = 0; i < ASCII_LIMIT; ++i) { + asciiData[i] = get(i); + } + + // First we look for which data blocks have the same value repeated over the whole block, + // deduplicate such blocks, find a good null data block (for faster enumeration), + // and get an upper bound for the necessary data array length. + AllSameBlocks allSameBlocks; + int32_t newDataCapacity = compactWholeDataBlocks(fastILimit, allSameBlocks); + if (newDataCapacity < 0) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return 0; + } + uint32_t *newData = (uint32_t *)uprv_malloc(newDataCapacity * 4); + if (newData == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + return 0; + } + uprv_memcpy(newData, asciiData, sizeof(asciiData)); + + int32_t dataNullIndex = allSameBlocks.findMostUsed(); + + MixedBlocks mixedBlocks; + int32_t newDataLength = compactData(fastILimit, newData, newDataCapacity, + dataNullIndex, mixedBlocks, errorCode); + if (U_FAILURE(errorCode)) { return 0; } + U_ASSERT(newDataLength <= newDataCapacity); + uprv_free(data); + data = newData; + dataCapacity = newDataCapacity; + dataLength = newDataLength; + if (dataLength > (0x3ffff + UCPTRIE_SMALL_DATA_BLOCK_LENGTH)) { + // The offset of the last data block is too high to be stored in the index table. + errorCode = U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + + if (dataNullIndex >= 0) { + dataNullOffset = index[dataNullIndex]; +#ifdef UCPTRIE_DEBUG + if (data[dataNullOffset] != initialValue) { + printf("UCPTrie initialValue %lx -> more common nullValue %lx\n", + (long)initialValue, (long)data[dataNullOffset]); + } +#endif + initialValue = data[dataNullOffset]; + } else { + dataNullOffset = UCPTRIE_NO_DATA_NULL_OFFSET; + } + + int32_t indexLength = compactIndex(fastILimit, mixedBlocks, errorCode); + highStart = realHighStart; + return indexLength; +} + +UCPTrie *MutableCodePointTrie::build(UCPTrieType type, UCPTrieValueWidth valueWidth, UErrorCode &errorCode) { + if (U_FAILURE(errorCode)) { + return nullptr; + } + if (type < UCPTRIE_TYPE_FAST || UCPTRIE_TYPE_SMALL < type || + valueWidth < UCPTRIE_VALUE_BITS_16 || UCPTRIE_VALUE_BITS_8 < valueWidth) { + errorCode = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + + // The mutable trie always stores 32-bit values. + // When we build a UCPTrie for a smaller value width, we first mask off unused bits + // before compacting the data. + switch (valueWidth) { + case UCPTRIE_VALUE_BITS_32: + break; + case UCPTRIE_VALUE_BITS_16: + maskValues(0xffff); + break; + case UCPTRIE_VALUE_BITS_8: + maskValues(0xff); + break; + default: + break; + } + + UChar32 fastLimit = type == UCPTRIE_TYPE_FAST ? BMP_LIMIT : UCPTRIE_SMALL_LIMIT; + int32_t indexLength = compactTrie(fastLimit >> UCPTRIE_SHIFT_3, errorCode); + if (U_FAILURE(errorCode)) { + clear(); + return nullptr; + } + + // Ensure data table alignment: The index length must be even for uint32_t data. + if (valueWidth == UCPTRIE_VALUE_BITS_32 && (indexLength & 1) != 0) { + index16[indexLength++] = 0xffee; // arbitrary value + } + + // Make the total trie structure length a multiple of 4 bytes by padding the data table, + // and store special values as the last two data values. + int32_t length = indexLength * 2; + if (valueWidth == UCPTRIE_VALUE_BITS_16) { + if (((indexLength ^ dataLength) & 1) != 0) { + // padding + data[dataLength++] = errorValue; + } + if (data[dataLength - 1] != errorValue || data[dataLength - 2] != highValue) { + data[dataLength++] = highValue; + data[dataLength++] = errorValue; + } + length += dataLength * 2; + } else if (valueWidth == UCPTRIE_VALUE_BITS_32) { + // 32-bit data words never need padding to a multiple of 4 bytes. + if (data[dataLength - 1] != errorValue || data[dataLength - 2] != highValue) { + if (data[dataLength - 1] != highValue) { + data[dataLength++] = highValue; + } + data[dataLength++] = errorValue; + } + length += dataLength * 4; + } else { + int32_t and3 = (length + dataLength) & 3; + if (and3 == 0 && data[dataLength - 1] == errorValue && data[dataLength - 2] == highValue) { + // all set + } else if(and3 == 3 && data[dataLength - 1] == highValue) { + data[dataLength++] = errorValue; + } else { + while (and3 != 2) { + data[dataLength++] = highValue; + and3 = (and3 + 1) & 3; + } + data[dataLength++] = highValue; + data[dataLength++] = errorValue; + } + length += dataLength; + } + + // Calculate the total length of the UCPTrie as a single memory block. + length += sizeof(UCPTrie); + U_ASSERT((length & 3) == 0); + + uint8_t *bytes = (uint8_t *)uprv_malloc(length); + if (bytes == nullptr) { + errorCode = U_MEMORY_ALLOCATION_ERROR; + clear(); + return nullptr; + } + UCPTrie *trie = reinterpret_cast(bytes); + uprv_memset(trie, 0, sizeof(UCPTrie)); + trie->indexLength = indexLength; + trie->dataLength = dataLength; + + trie->highStart = highStart; + // Round up shifted12HighStart to a multiple of 0x1000 for easy testing from UTF-8 lead bytes. + // Runtime code needs to then test for the real highStart as well. + trie->shifted12HighStart = (highStart + 0xfff) >> 12; + trie->type = type; + trie->valueWidth = valueWidth; + + trie->index3NullOffset = index3NullOffset; + trie->dataNullOffset = dataNullOffset; + trie->nullValue = initialValue; + + bytes += sizeof(UCPTrie); + + // Fill the index and data arrays. + uint16_t *dest16 = (uint16_t *)bytes; + trie->index = dest16; + + if (highStart <= fastLimit) { + // Condense only the fast index from the mutable-trie index. + for (int32_t i = 0, j = 0; j < indexLength; i += SMALL_DATA_BLOCKS_PER_BMP_BLOCK, ++j) { + *dest16++ = (uint16_t)index[i]; // dest16[j] + } + } else { + uprv_memcpy(dest16, index16, indexLength * 2); + dest16 += indexLength; + } + bytes += indexLength * 2; + + // Write the data array. + const uint32_t *p = data; + switch (valueWidth) { + case UCPTRIE_VALUE_BITS_16: + // Write 16-bit data values. + trie->data.ptr16 = dest16; + for (int32_t i = dataLength; i > 0; --i) { + *dest16++ = (uint16_t)*p++; + } + break; + case UCPTRIE_VALUE_BITS_32: + // Write 32-bit data values. + trie->data.ptr32 = (uint32_t *)bytes; + uprv_memcpy(bytes, p, (size_t)dataLength * 4); + break; + case UCPTRIE_VALUE_BITS_8: + // Write 8-bit data values. + trie->data.ptr8 = bytes; + for (int32_t i = dataLength; i > 0; --i) { + *bytes++ = (uint8_t)*p++; + } + break; + default: + // Will not occur, valueWidth checked at the beginning. + break; + } + +#ifdef UCPTRIE_DEBUG + trie->name = name; + + ucptrie_printLengths(trie, ""); +#endif + + clear(); + return trie; +} + +} // namespace + +U_NAMESPACE_END + +U_NAMESPACE_USE + +U_CAPI UMutableCPTrie * U_EXPORT2 +umutablecptrie_open(uint32_t initialValue, uint32_t errorValue, UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return nullptr; + } + LocalPointer trie( + new MutableCodePointTrie(initialValue, errorValue, *pErrorCode), *pErrorCode); + if (U_FAILURE(*pErrorCode)) { + return nullptr; + } + return reinterpret_cast(trie.orphan()); +} + +U_CAPI UMutableCPTrie * U_EXPORT2 +umutablecptrie_clone(const UMutableCPTrie *other, UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return nullptr; + } + if (other == nullptr) { + return nullptr; + } + LocalPointer clone( + new MutableCodePointTrie(*reinterpret_cast(other), *pErrorCode), *pErrorCode); + if (U_FAILURE(*pErrorCode)) { + return nullptr; + } + return reinterpret_cast(clone.orphan()); +} + +U_CAPI void U_EXPORT2 +umutablecptrie_close(UMutableCPTrie *trie) { + delete reinterpret_cast(trie); +} + +U_CAPI UMutableCPTrie * U_EXPORT2 +umutablecptrie_fromUCPMap(const UCPMap *map, UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return nullptr; + } + if (map == nullptr) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + return reinterpret_cast(MutableCodePointTrie::fromUCPMap(map, *pErrorCode)); +} + +U_CAPI UMutableCPTrie * U_EXPORT2 +umutablecptrie_fromUCPTrie(const UCPTrie *trie, UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return nullptr; + } + if (trie == nullptr) { + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; + return nullptr; + } + return reinterpret_cast(MutableCodePointTrie::fromUCPTrie(trie, *pErrorCode)); +} + +U_CAPI uint32_t U_EXPORT2 +umutablecptrie_get(const UMutableCPTrie *trie, UChar32 c) { + return reinterpret_cast(trie)->get(c); +} + +namespace { + +UChar32 getRange(const void *trie, UChar32 start, + UCPMapValueFilter *filter, const void *context, uint32_t *pValue) { + return reinterpret_cast(trie)-> + getRange(start, filter, context, pValue); +} + +} // namespace + +U_CAPI UChar32 U_EXPORT2 +umutablecptrie_getRange(const UMutableCPTrie *trie, UChar32 start, + UCPMapRangeOption option, uint32_t surrogateValue, + UCPMapValueFilter *filter, const void *context, uint32_t *pValue) { + return ucptrie_internalGetRange(getRange, trie, start, + option, surrogateValue, + filter, context, pValue); +} + +U_CAPI void U_EXPORT2 +umutablecptrie_set(UMutableCPTrie *trie, UChar32 c, uint32_t value, UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return; + } + reinterpret_cast(trie)->set(c, value, *pErrorCode); +} + +U_CAPI void U_EXPORT2 +umutablecptrie_setRange(UMutableCPTrie *trie, UChar32 start, UChar32 end, + uint32_t value, UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return; + } + reinterpret_cast(trie)->setRange(start, end, value, *pErrorCode); +} + +/* Compact and internally serialize the trie. */ +U_CAPI UCPTrie * U_EXPORT2 +umutablecptrie_buildImmutable(UMutableCPTrie *trie, UCPTrieType type, UCPTrieValueWidth valueWidth, + UErrorCode *pErrorCode) { + if (U_FAILURE(*pErrorCode)) { + return nullptr; + } + return reinterpret_cast(trie)->build(type, valueWidth, *pErrorCode); +} + +#ifdef UCPTRIE_DEBUG +U_CFUNC void umutablecptrie_setName(UMutableCPTrie *trie, const char *name) { + reinterpret_cast(trie)->name = name; +} +#endif diff --git a/intl/icu/source/common/umutex.cpp b/intl/icu/source/common/umutex.cpp new file mode 100644 index 0000000000..6c3452ca1b --- /dev/null +++ b/intl/icu/source/common/umutex.cpp @@ -0,0 +1,224 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1997-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* File umutex.cpp +* +* Modification History: +* +* Date Name Description +* 04/02/97 aliu Creation. +* 04/07/99 srl updated +* 05/13/99 stephen Changed to umutex (from cmutex). +* 11/22/99 aliu Make non-global mutex autoinitialize [j151] +****************************************************************************** +*/ + +#include "umutex.h" + +#include "unicode/utypes.h" +#include "uassert.h" +#include "ucln_cmn.h" +#include "cmemory.h" + +U_NAMESPACE_BEGIN + + +#if defined(U_USER_MUTEX_CPP) +// Support for including an alternate implementation of mutexes has been withdrawn. +// See issue ICU-20185. +#error U_USER_MUTEX_CPP not supported +#endif + + +/************************************************************************************************* + * + * ICU Mutex wrappers. + * + *************************************************************************************************/ + +#ifndef __wasi__ +namespace { +std::mutex *initMutex; +std::condition_variable *initCondition; + +// The ICU global mutex. +// Used when ICU implementation code passes nullptr for the mutex pointer. +UMutex globalMutex; + +std::once_flag initFlag; +std::once_flag *pInitFlag = &initFlag; + +} // Anonymous namespace +#endif + +U_CDECL_BEGIN +static UBool U_CALLCONV umtx_cleanup() { +#ifndef __wasi__ + initMutex->~mutex(); + initCondition->~condition_variable(); + UMutex::cleanup(); + + // Reset the once_flag, by destructing it and creating a fresh one in its place. + // Do not use this trick anywhere else in ICU; use umtx_initOnce, not std::call_once(). + pInitFlag->~once_flag(); + pInitFlag = new(&initFlag) std::once_flag(); +#endif + return true; +} + +static void U_CALLCONV umtx_init() { +#ifndef __wasi__ + initMutex = STATIC_NEW(std::mutex); + initCondition = STATIC_NEW(std::condition_variable); + ucln_common_registerCleanup(UCLN_COMMON_MUTEX, umtx_cleanup); +#endif +} +U_CDECL_END + + +#ifndef __wasi__ +std::mutex *UMutex::getMutex() { + std::mutex *retPtr = fMutex.load(std::memory_order_acquire); + if (retPtr == nullptr) { + std::call_once(*pInitFlag, umtx_init); + std::lock_guard guard(*initMutex); + retPtr = fMutex.load(std::memory_order_acquire); + if (retPtr == nullptr) { + fMutex = new(fStorage) std::mutex(); + retPtr = fMutex; + fListLink = gListHead; + gListHead = this; + } + } + U_ASSERT(retPtr != nullptr); + return retPtr; +} +#endif + +UMutex *UMutex::gListHead = nullptr; + +void UMutex::cleanup() { + UMutex *next = nullptr; + for (UMutex *m = gListHead; m != nullptr; m = next) { +#ifndef __wasi__ + (*m->fMutex).~mutex(); + m->fMutex = nullptr; +#endif + next = m->fListLink; + m->fListLink = nullptr; + } + gListHead = nullptr; +} + + +U_CAPI void U_EXPORT2 +umtx_lock(UMutex *mutex) { +#ifndef __wasi__ + if (mutex == nullptr) { + mutex = &globalMutex; + } + mutex->lock(); +#endif +} + + +U_CAPI void U_EXPORT2 +umtx_unlock(UMutex* mutex) +{ +#ifndef __wasi__ + if (mutex == nullptr) { + mutex = &globalMutex; + } + mutex->unlock(); +#endif +} + + +/************************************************************************************************* + * + * UInitOnce Implementation + * + *************************************************************************************************/ + +// This function is called when a test of a UInitOnce::fState reveals that +// initialization has not completed, that we either need to call the init +// function on this thread, or wait for some other thread to complete. +// +// The actual call to the init function is made inline by template code +// that knows the C++ types involved. This function returns true if +// the caller needs to call the Init function. +// +U_COMMON_API UBool U_EXPORT2 +umtx_initImplPreInit(UInitOnce &uio) { +#ifndef __wasi__ + std::call_once(*pInitFlag, umtx_init); + std::unique_lock lock(*initMutex); +#endif + if (umtx_loadAcquire(uio.fState) == 0) { + umtx_storeRelease(uio.fState, 1); + return true; // Caller will next call the init function. + } else { +#ifndef __wasi__ + while (umtx_loadAcquire(uio.fState) == 1) { + // Another thread is currently running the initialization. + // Wait until it completes. + initCondition->wait(lock); + } + U_ASSERT(uio.fState == 2); +#endif + return false; + } +} + + +// This function is called by the thread that ran an initialization function, +// just after completing the function. +// Some threads may be waiting on the condition, requiring the broadcast wakeup. +// Some threads may be racing to test the fState variable outside of the mutex, +// requiring the use of store/release when changing its value. + +U_COMMON_API void U_EXPORT2 +umtx_initImplPostInit(UInitOnce &uio) { +#ifndef __wasi__ + { + std::unique_lock lock(*initMutex); + umtx_storeRelease(uio.fState, 2); + } + initCondition->notify_all(); +#endif +} + +U_NAMESPACE_END + +/************************************************************************************************* + * + * Deprecated functions for setting user mutexes. + * + *************************************************************************************************/ + +U_DEPRECATED void U_EXPORT2 +u_setMutexFunctions(const void * /*context */, UMtxInitFn *, UMtxFn *, + UMtxFn *, UMtxFn *, UErrorCode *status) { + if (U_SUCCESS(*status)) { + *status = U_UNSUPPORTED_ERROR; + } + return; +} + + + +U_DEPRECATED void U_EXPORT2 +u_setAtomicIncDecFunctions(const void * /*context */, UMtxAtomicFn *, UMtxAtomicFn *, + UErrorCode *status) { + if (U_SUCCESS(*status)) { + *status = U_UNSUPPORTED_ERROR; + } + return; +} diff --git a/intl/icu/source/common/umutex.h b/intl/icu/source/common/umutex.h new file mode 100644 index 0000000000..f084a20bf8 --- /dev/null +++ b/intl/icu/source/common/umutex.h @@ -0,0 +1,313 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1997-2015, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* +* File UMUTEX.H +* +* Modification History: +* +* Date Name Description +* 04/02/97 aliu Creation. +* 04/07/99 srl rewrite - C interface, multiple mutices +* 05/13/99 stephen Changed to umutex (from cmutex) +****************************************************************************** +*/ + +#ifndef UMUTEX_H +#define UMUTEX_H + +#ifndef __wasi__ +#include +#include +#include +#endif + +#include + +#include "unicode/utypes.h" +#include "unicode/uclean.h" +#include "unicode/uobject.h" + +#include "putilimp.h" + +#if defined(U_USER_ATOMICS_H) || defined(U_USER_MUTEX_H) +// Support for including an alternate implementation of atomic & mutex operations has been withdrawn. +// See issue ICU-20185. +#error U_USER_ATOMICS and U_USER_MUTEX_H are not supported +#endif + +#ifndef __wasi__ + +// Export an explicit template instantiation of std::atomic. +// When building DLLs for Windows this is required as it is used as a data member of the exported SharedObject class. +// See digitlst.h, pluralaffix.h, datefmt.h, and others for similar examples. +// +// Similar story for std::atomic, and the exported UMutex class. +#if U_PF_WINDOWS <= U_PLATFORM && U_PLATFORM <= U_PF_CYGWIN && !defined(U_IN_DOXYGEN) +#if defined(__clang__) || defined(_MSC_VER) + #if defined(__clang__) + // Suppress the warning that the explicit instantiation after explicit specialization has no effect. + #pragma clang diagnostic push + #pragma clang diagnostic ignored "-Winstantiation-after-specialization" + #endif +template struct U_COMMON_API std::atomic; +template struct U_COMMON_API std::atomic; + #if defined(__clang__) + #pragma clang diagnostic pop + #endif +#elif defined(__GNUC__) +// For GCC this class is already exported/visible, so no need for U_COMMON_API. +template struct std::atomic; +template struct std::atomic; +#endif +#endif + +#endif + +U_NAMESPACE_BEGIN + +/**************************************************************************** + * + * Low Level Atomic Operations, ICU wrappers for. + * + ****************************************************************************/ + +#ifndef __wasi__ + +typedef std::atomic u_atomic_int32_t; + +inline int32_t umtx_loadAcquire(u_atomic_int32_t &var) { + return var.load(std::memory_order_acquire); +} + +inline void umtx_storeRelease(u_atomic_int32_t &var, int32_t val) { + var.store(val, std::memory_order_release); +} + +inline int32_t umtx_atomic_inc(u_atomic_int32_t *var) { + return var->fetch_add(1) + 1; +} + +inline int32_t umtx_atomic_dec(u_atomic_int32_t *var) { + return var->fetch_sub(1) - 1; +} + +#else + +typedef int32_t u_atomic_int32_t; +#define ATOMIC_INT32_T_INITIALIZER(val) val + +inline int32_t umtx_loadAcquire(u_atomic_int32_t &var) { + return var; +} + +inline void umtx_storeRelease(u_atomic_int32_t &var, int32_t val) { + var = val; +} + +inline int32_t umtx_atomic_inc(u_atomic_int32_t *var) { + return ++(*var); +} + +inline int32_t umtx_atomic_dec(u_atomic_int32_t *var) { + return --(*var); +} + +#endif + +/************************************************************************************************* + * + * UInitOnce Definitions. + * + *************************************************************************************************/ + +struct U_COMMON_API UInitOnce { + u_atomic_int32_t fState {0}; + UErrorCode fErrCode {U_ZERO_ERROR}; + void reset() {fState = 0;} + UBool isReset() {return umtx_loadAcquire(fState) == 0;} +// Note: isReset() is used by service registration code. +// Thread safety of this usage needs review. +}; + +U_COMMON_API UBool U_EXPORT2 umtx_initImplPreInit(UInitOnce &); +U_COMMON_API void U_EXPORT2 umtx_initImplPostInit(UInitOnce &); + +template void umtx_initOnce(UInitOnce &uio, T *obj, void (U_CALLCONV T::*fp)()) { + if (umtx_loadAcquire(uio.fState) == 2) { + return; + } + if (umtx_initImplPreInit(uio)) { + (obj->*fp)(); + umtx_initImplPostInit(uio); + } +} + + +// umtx_initOnce variant for plain functions, or static class functions. +// No context parameter. +inline void umtx_initOnce(UInitOnce &uio, void (U_CALLCONV *fp)()) { + if (umtx_loadAcquire(uio.fState) == 2) { + return; + } + if (umtx_initImplPreInit(uio)) { + (*fp)(); + umtx_initImplPostInit(uio); + } +} + +// umtx_initOnce variant for plain functions, or static class functions. +// With ErrorCode, No context parameter. +inline void umtx_initOnce(UInitOnce &uio, void (U_CALLCONV *fp)(UErrorCode &), UErrorCode &errCode) { + if (U_FAILURE(errCode)) { + return; + } + if (umtx_loadAcquire(uio.fState) != 2 && umtx_initImplPreInit(uio)) { + // We run the initialization. + (*fp)(errCode); + uio.fErrCode = errCode; + umtx_initImplPostInit(uio); + } else { + // Someone else already ran the initialization. + if (U_FAILURE(uio.fErrCode)) { + errCode = uio.fErrCode; + } + } +} + +// umtx_initOnce variant for plain functions, or static class functions, +// with a context parameter. +template void umtx_initOnce(UInitOnce &uio, void (U_CALLCONV *fp)(T), T context) { + if (umtx_loadAcquire(uio.fState) == 2) { + return; + } + if (umtx_initImplPreInit(uio)) { + (*fp)(context); + umtx_initImplPostInit(uio); + } +} + +// umtx_initOnce variant for plain functions, or static class functions, +// with a context parameter and an error code. +template void umtx_initOnce(UInitOnce &uio, void (U_CALLCONV *fp)(T, UErrorCode &), T context, UErrorCode &errCode) { + if (U_FAILURE(errCode)) { + return; + } + if (umtx_loadAcquire(uio.fState) != 2 && umtx_initImplPreInit(uio)) { + // We run the initialization. + (*fp)(context, errCode); + uio.fErrCode = errCode; + umtx_initImplPostInit(uio); + } else { + // Someone else already ran the initialization. + if (U_FAILURE(uio.fErrCode)) { + errCode = uio.fErrCode; + } + } +} + +// UMutex should be constexpr-constructible, so that no initialization code +// is run during startup. +// This works on all C++ libraries except MS VS before VS2019. +#if (defined(_CPPLIB_VER) && !defined(_MSVC_STL_VERSION)) || \ + (defined(_MSVC_STL_VERSION) && _MSVC_STL_VERSION < 142) + // (VS std lib older than VS2017) || (VS std lib version < VS2019) +# define UMUTEX_CONSTEXPR +#else +# define UMUTEX_CONSTEXPR constexpr +#endif + +/** + * UMutex - ICU Mutex class. + * + * This is the preferred Mutex class for use within ICU implementation code. + * It is a thin wrapper over C++ std::mutex, with these additions: + * - Static instances are safe, not triggering static construction or destruction, + * and the associated order of construction or destruction issues. + * - Plumbed into u_cleanup() for destructing the underlying std::mutex, + * which frees any OS level resources they may be holding. + * + * Limitations: + * - Static or global instances only. Cannot be heap allocated. Cannot appear as a + * member of another class. + * - No condition variables or other advanced features. If needed, you will need to use + * std::mutex and std::condition_variable directly. For an example, see unifiedcache.cpp + * + * Typical Usage: + * static UMutex myMutex; + * + * { + * Mutex lock(myMutex); + * ... // Do stuff that is protected by myMutex; + * } // myMutex is released when lock goes out of scope. + */ + +class U_COMMON_API UMutex { +public: + UMUTEX_CONSTEXPR UMutex() {} + ~UMutex() = default; + + UMutex(const UMutex &other) = delete; + UMutex &operator =(const UMutex &other) = delete; + void *operator new(size_t) = delete; + + // requirements for C++ BasicLockable, allows UMutex to work with std::lock_guard + void lock() { +#ifndef __wasi__ + std::mutex *m = fMutex.load(std::memory_order_acquire); + if (m == nullptr) { m = getMutex(); } + m->lock(); +#endif + } + void unlock() { +#ifndef __wasi__ + fMutex.load(std::memory_order_relaxed)->unlock(); +#endif + } + + static void cleanup(); + +private: +#ifndef __wasi__ + alignas(std::mutex) char fStorage[sizeof(std::mutex)] {}; + std::atomic fMutex { nullptr }; +#endif + + /** All initialized UMutexes are kept in a linked list, so that they can be found, + * and the underlying std::mutex destructed, by u_cleanup(). + */ + UMutex *fListLink { nullptr }; + static UMutex *gListHead; + + /** Out-of-line function to lazily initialize a UMutex on first use. + * Initial fast check is inline, in lock(). The returned value may never + * be nullptr. + */ +#ifndef __wasi__ + std::mutex *getMutex(); +#endif +}; + + +/* Lock a mutex. + * @param mutex The given mutex to be locked. Pass NULL to specify + * the global ICU mutex. Recursive locks are an error + * and may cause a deadlock on some platforms. + */ +U_CAPI void U_EXPORT2 umtx_lock(UMutex* mutex); + +/* Unlock a mutex. + * @param mutex The given mutex to be unlocked. Pass NULL to specify + * the global ICU mutex. + */ +U_CAPI void U_EXPORT2 umtx_unlock (UMutex* mutex); + + +U_NAMESPACE_END + +#endif /* UMUTEX_H */ +/*eof*/ diff --git a/intl/icu/source/common/unames.cpp b/intl/icu/source/common/unames.cpp new file mode 100644 index 0000000000..1b3192bf25 --- /dev/null +++ b/intl/icu/source/common/unames.cpp @@ -0,0 +1,2108 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2014, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: unames.c +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 1999oct04 +* created by: Markus W. Scherer +*/ + +#include "unicode/utypes.h" +#include "unicode/putil.h" +#include "unicode/uchar.h" +#include "unicode/udata.h" +#include "unicode/utf.h" +#include "unicode/utf16.h" +#include "uassert.h" +#include "ustr_imp.h" +#include "umutex.h" +#include "cmemory.h" +#include "cstring.h" +#include "ucln_cmn.h" +#include "udataswp.h" +#include "uprops.h" + +U_NAMESPACE_BEGIN + +/* prototypes ------------------------------------------------------------- */ + +static const char DATA_NAME[] = "unames"; +static const char DATA_TYPE[] = "icu"; + +#define GROUP_SHIFT 5 +#define LINES_PER_GROUP (1L<groupsOffset) + +typedef struct { + const char *otherName; + UChar32 code; +} FindName; + +#define DO_FIND_NAME nullptr + +static UDataMemory *uCharNamesData=nullptr; +static UCharNames *uCharNames=nullptr; +static icu::UInitOnce gCharNamesInitOnce {}; + +/* + * Maximum length of character names (regular & 1.0). + */ +static int32_t gMaxNameLength=0; + +/* + * Set of chars used in character names (regular & 1.0). + * Chars are platform-dependent (can be EBCDIC). + */ +static uint32_t gNameSet[8]={ 0 }; + +#define U_NONCHARACTER_CODE_POINT U_CHAR_CATEGORY_COUNT +#define U_LEAD_SURROGATE U_CHAR_CATEGORY_COUNT + 1 +#define U_TRAIL_SURROGATE U_CHAR_CATEGORY_COUNT + 2 + +#define U_CHAR_EXTENDED_CATEGORY_COUNT (U_CHAR_CATEGORY_COUNT + 3) + +static const char * const charCatNames[U_CHAR_EXTENDED_CATEGORY_COUNT] = { + "unassigned", + "uppercase letter", + "lowercase letter", + "titlecase letter", + "modifier letter", + "other letter", + "non spacing mark", + "enclosing mark", + "combining spacing mark", + "decimal digit number", + "letter number", + "other number", + "space separator", + "line separator", + "paragraph separator", + "control", + "format", + "private use area", + "surrogate", + "dash punctuation", + "start punctuation", + "end punctuation", + "connector punctuation", + "other punctuation", + "math symbol", + "currency symbol", + "modifier symbol", + "other symbol", + "initial punctuation", + "final punctuation", + "noncharacter", + "lead surrogate", + "trail surrogate" +}; + +/* implementation ----------------------------------------------------------- */ + +static UBool U_CALLCONV unames_cleanup() +{ + if(uCharNamesData) { + udata_close(uCharNamesData); + uCharNamesData = nullptr; + } + if(uCharNames) { + uCharNames = nullptr; + } + gCharNamesInitOnce.reset(); + gMaxNameLength=0; + return true; +} + +static UBool U_CALLCONV +isAcceptable(void * /*context*/, + const char * /*type*/, const char * /*name*/, + const UDataInfo *pInfo) { + return (UBool)( + pInfo->size>=20 && + pInfo->isBigEndian==U_IS_BIG_ENDIAN && + pInfo->charsetFamily==U_CHARSET_FAMILY && + pInfo->dataFormat[0]==0x75 && /* dataFormat="unam" */ + pInfo->dataFormat[1]==0x6e && + pInfo->dataFormat[2]==0x61 && + pInfo->dataFormat[3]==0x6d && + pInfo->formatVersion[0]==1); +} + +static void U_CALLCONV +loadCharNames(UErrorCode &status) { + U_ASSERT(uCharNamesData == nullptr); + U_ASSERT(uCharNames == nullptr); + + uCharNamesData = udata_openChoice(nullptr, DATA_TYPE, DATA_NAME, isAcceptable, nullptr, &status); + if(U_FAILURE(status)) { + uCharNamesData = nullptr; + } else { + uCharNames = (UCharNames *)udata_getMemory(uCharNamesData); + } + ucln_common_registerCleanup(UCLN_COMMON_UNAMES, unames_cleanup); +} + + +static UBool +isDataLoaded(UErrorCode *pErrorCode) { + umtx_initOnce(gCharNamesInitOnce, &loadCharNames, *pErrorCode); + return U_SUCCESS(*pErrorCode); +} + +#define WRITE_CHAR(buffer, bufferLength, bufferPos, c) UPRV_BLOCK_MACRO_BEGIN { \ + if((bufferLength)>0) { \ + *(buffer)++=c; \ + --(bufferLength); \ + } \ + ++(bufferPos); \ +} UPRV_BLOCK_MACRO_END + +#define U_ISO_COMMENT U_CHAR_NAME_CHOICE_COUNT + +/* + * Important: expandName() and compareName() are almost the same - + * apply fixes to both. + * + * UnicodeData.txt uses ';' as a field separator, so no + * field can contain ';' as part of its contents. + * In unames.dat, it is marked as token[';']==-1 only if the + * semicolon is used in the data file - which is iff we + * have Unicode 1.0 names or ISO comments or aliases. + * So, it will be token[';']==-1 if we store U1.0 names/ISO comments/aliases + * although we know that it will never be part of a name. + */ +static uint16_t +expandName(UCharNames *names, + const uint8_t *name, uint16_t nameLength, UCharNameChoice nameChoice, + char *buffer, uint16_t bufferLength) { + uint16_t *tokens=(uint16_t *)names+8; + uint16_t token, tokenCount=*tokens++, bufferPos=0; + uint8_t *tokenStrings=(uint8_t *)names+names->tokenStringOffset; + uint8_t c; + + if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) { + /* + * skip the modern name if it is not requested _and_ + * if the semicolon byte value is a character, not a token number + */ + if((uint8_t)';'>=tokenCount || tokens[(uint8_t)';']==(uint16_t)(-1)) { + int fieldIndex= nameChoice==U_ISO_COMMENT ? 2 : nameChoice; + do { + while(nameLength>0) { + --nameLength; + if(*name++==';') { + break; + } + } + } while(--fieldIndex>0); + } else { + /* + * the semicolon byte value is a token number, therefore + * only modern names are stored in unames.dat and there is no + * such requested alternate name here + */ + nameLength=0; + } + } + + /* write each letter directly, and write a token word per token */ + while(nameLength>0) { + --nameLength; + c=*name++; + + if(c>=tokenCount) { + if(c!=';') { + /* implicit letter */ + WRITE_CHAR(buffer, bufferLength, bufferPos, c); + } else { + /* finished */ + break; + } + } else { + token=tokens[c]; + if(token==(uint16_t)(-2)) { + /* this is a lead byte for a double-byte token */ + token=tokens[c<<8|*name++]; + --nameLength; + } + if(token==(uint16_t)(-1)) { + if(c!=';') { + /* explicit letter */ + WRITE_CHAR(buffer, bufferLength, bufferPos, c); + } else { + /* stop, but skip the semicolon if we are seeking + extended names and there was no 2.0 name but there + is a 1.0 name. */ + if(!bufferPos && nameChoice == U_EXTENDED_CHAR_NAME) { + if ((uint8_t)';'>=tokenCount || tokens[(uint8_t)';']==(uint16_t)(-1)) { + continue; + } + } + /* finished */ + break; + } + } else { + /* write token word */ + uint8_t *tokenString=tokenStrings+token; + while((c=*tokenString++)!=0) { + WRITE_CHAR(buffer, bufferLength, bufferPos, c); + } + } + } + } + + /* zero-terminate */ + if(bufferLength>0) { + *buffer=0; + } + + return bufferPos; +} + +/* + * compareName() is almost the same as expandName() except that it compares + * the currently expanded name to an input name. + * It returns the match/no match result as soon as possible. + */ +static UBool +compareName(UCharNames *names, + const uint8_t *name, uint16_t nameLength, UCharNameChoice nameChoice, + const char *otherName) { + uint16_t *tokens=(uint16_t *)names+8; + uint16_t token, tokenCount=*tokens++; + uint8_t *tokenStrings=(uint8_t *)names+names->tokenStringOffset; + uint8_t c; + const char *origOtherName = otherName; + + if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) { + /* + * skip the modern name if it is not requested _and_ + * if the semicolon byte value is a character, not a token number + */ + if((uint8_t)';'>=tokenCount || tokens[(uint8_t)';']==(uint16_t)(-1)) { + int fieldIndex= nameChoice==U_ISO_COMMENT ? 2 : nameChoice; + do { + while(nameLength>0) { + --nameLength; + if(*name++==';') { + break; + } + } + } while(--fieldIndex>0); + } else { + /* + * the semicolon byte value is a token number, therefore + * only modern names are stored in unames.dat and there is no + * such requested alternate name here + */ + nameLength=0; + } + } + + /* compare each letter directly, and compare a token word per token */ + while(nameLength>0) { + --nameLength; + c=*name++; + + if(c>=tokenCount) { + if(c!=';') { + /* implicit letter */ + if((char)c!=*otherName++) { + return false; + } + } else { + /* finished */ + break; + } + } else { + token=tokens[c]; + if(token==(uint16_t)(-2)) { + /* this is a lead byte for a double-byte token */ + token=tokens[c<<8|*name++]; + --nameLength; + } + if(token==(uint16_t)(-1)) { + if(c!=';') { + /* explicit letter */ + if((char)c!=*otherName++) { + return false; + } + } else { + /* stop, but skip the semicolon if we are seeking + extended names and there was no 2.0 name but there + is a 1.0 name. */ + if(otherName == origOtherName && nameChoice == U_EXTENDED_CHAR_NAME) { + if ((uint8_t)';'>=tokenCount || tokens[(uint8_t)';']==(uint16_t)(-1)) { + continue; + } + } + /* finished */ + break; + } + } else { + /* write token word */ + uint8_t *tokenString=tokenStrings+token; + while((c=*tokenString++)!=0) { + if((char)c!=*otherName++) { + return false; + } + } + } + } + } + + /* complete match? */ + return (UBool)(*otherName==0); +} + +static uint8_t getCharCat(UChar32 cp) { + uint8_t cat; + + if (U_IS_UNICODE_NONCHAR(cp)) { + return U_NONCHARACTER_CODE_POINT; + } + + if ((cat = u_charType(cp)) == U_SURROGATE) { + cat = U_IS_LEAD(cp) ? U_LEAD_SURROGATE : U_TRAIL_SURROGATE; + } + + return cat; +} + +static const char *getCharCatName(UChar32 cp) { + uint8_t cat = getCharCat(cp); + + /* Return unknown if the table of names above is not up to + date. */ + + if (cat >= UPRV_LENGTHOF(charCatNames)) { + return "unknown"; + } else { + return charCatNames[cat]; + } +} + +static uint16_t getExtName(uint32_t code, char *buffer, uint16_t bufferLength) { + const char *catname = getCharCatName(code); + uint16_t length = 0; + + UChar32 cp; + int ndigits, i; + + WRITE_CHAR(buffer, bufferLength, length, '<'); + while (catname[length - 1]) { + WRITE_CHAR(buffer, bufferLength, length, catname[length - 1]); + } + WRITE_CHAR(buffer, bufferLength, length, '-'); + for (cp = code, ndigits = 0; cp; ++ndigits, cp >>= 4) + ; + if (ndigits < 4) + ndigits = 4; + for (cp = code, i = ndigits; (cp || i > 0) && bufferLength; cp >>= 4, bufferLength--) { + uint8_t v = (uint8_t)(cp & 0xf); + buffer[--i] = (v < 10 ? '0' + v : 'A' + v - 10); + } + buffer += ndigits; + length += static_cast(ndigits); + WRITE_CHAR(buffer, bufferLength, length, '>'); + + return length; +} + +/* + * getGroup() does a binary search for the group that contains the + * Unicode code point "code". + * The return value is always a valid Group* that may contain "code" + * or else is the highest group before "code". + * If the lowest group is after "code", then that one is returned. + */ +static const uint16_t * +getGroup(UCharNames *names, uint32_t code) { + const uint16_t *groups=GET_GROUPS(names); + uint16_t groupMSB=(uint16_t)(code>>GROUP_SHIFT), + start=0, + limit=*groups++, + number; + + /* binary search for the group of names that contains the one for code */ + while(start=0xc, then it forms a length value with the following nibble. + * Calculation see below. + * The offsets and lengths arrays must be at least 33 (one more) long because + * there is no check here at the end if the last nibble is still used. + */ +static const uint8_t * +expandGroupLengths(const uint8_t *s, + uint16_t offsets[LINES_PER_GROUP+1], uint16_t lengths[LINES_PER_GROUP+1]) { + /* read the lengths of the 32 strings in this group and get each string's offset */ + uint16_t i=0, offset=0, length=0; + uint8_t lengthByte; + + /* all 32 lengths must be read to get the offset of the first group string */ + while(i=12) { + /* double-nibble length spread across two bytes */ + length=(uint16_t)(((length&0x3)<<4|lengthByte>>4)+12); + lengthByte&=0xf; + } else if((lengthByte /* &0xf0 */)>=0xc0) { + /* double-nibble length spread across this one byte */ + length=(uint16_t)((lengthByte&0x3f)+12); + } else { + /* single-nibble length in MSBs */ + length=(uint16_t)(lengthByte>>4); + lengthByte&=0xf; + } + + *offsets++=offset; + *lengths++=length; + + offset+=length; + ++i; + + /* read odd nibble - LSBs of lengthByte */ + if((lengthByte&0xf0)==0) { + /* this nibble was not consumed for a double-nibble length above */ + length=lengthByte; + if(length<12) { + /* single-nibble length in LSBs */ + *offsets++=offset; + *lengths++=length; + + offset+=length; + ++i; + } + } else { + length=0; /* prevent double-nibble detection in the next iteration */ + } + } + + /* now, s is at the first group string */ + return s; +} + +static uint16_t +expandGroupName(UCharNames *names, const uint16_t *group, + uint16_t lineNumber, UCharNameChoice nameChoice, + char *buffer, uint16_t bufferLength) { + uint16_t offsets[LINES_PER_GROUP+2], lengths[LINES_PER_GROUP+2]; + const uint8_t *s=(uint8_t *)names+names->groupStringOffset+GET_GROUP_OFFSET(group); + s=expandGroupLengths(s, offsets, lengths); + return expandName(names, s+offsets[lineNumber], lengths[lineNumber], nameChoice, + buffer, bufferLength); +} + +static uint16_t +getName(UCharNames *names, uint32_t code, UCharNameChoice nameChoice, + char *buffer, uint16_t bufferLength) { + const uint16_t *group=getGroup(names, code); + if((uint16_t)(code>>GROUP_SHIFT)==group[GROUP_MSB]) { + return expandGroupName(names, group, (uint16_t)(code&GROUP_MASK), nameChoice, + buffer, bufferLength); + } else { + /* group not found */ + /* zero-terminate */ + if(bufferLength>0) { + *buffer=0; + } + return 0; + } +} + +/* + * enumGroupNames() enumerates all the names in a 32-group + * and either calls the enumerator function or finds a given input name. + */ +static UBool +enumGroupNames(UCharNames *names, const uint16_t *group, + UChar32 start, UChar32 end, + UEnumCharNamesFn *fn, void *context, + UCharNameChoice nameChoice) { + uint16_t offsets[LINES_PER_GROUP+2], lengths[LINES_PER_GROUP+2]; + const uint8_t *s=(uint8_t *)names+names->groupStringOffset+GET_GROUP_OFFSET(group); + + s=expandGroupLengths(s, offsets, lengths); + if(fn!=DO_FIND_NAME) { + char buffer[200]; + uint16_t length; + + while(start<=end) { + length=expandName(names, s+offsets[start&GROUP_MASK], lengths[start&GROUP_MASK], nameChoice, buffer, sizeof(buffer)); + if (!length && nameChoice == U_EXTENDED_CHAR_NAME) { + buffer[length = getExtName(start, buffer, sizeof(buffer))] = 0; + } + /* here, we assume that the buffer is large enough */ + if(length>0) { + if(!fn(context, start, nameChoice, buffer, length)) { + return false; + } + } + ++start; + } + } else { + const char *otherName=((FindName *)context)->otherName; + while(start<=end) { + if(compareName(names, s+offsets[start&GROUP_MASK], lengths[start&GROUP_MASK], nameChoice, otherName)) { + ((FindName *)context)->code=start; + return false; + } + ++start; + } + } + return true; +} + +/* + * enumExtNames enumerate extended names. + * It only needs to do it if it is called with a real function and not + * with the dummy DO_FIND_NAME, because u_charFromName() does a check + * for extended names by itself. + */ +static UBool +enumExtNames(UChar32 start, UChar32 end, + UEnumCharNamesFn *fn, void *context) +{ + if(fn!=DO_FIND_NAME) { + char buffer[200]; + uint16_t length; + + while(start<=end) { + buffer[length = getExtName(start, buffer, sizeof(buffer))] = 0; + /* here, we assume that the buffer is large enough */ + if(length>0) { + if(!fn(context, start, U_EXTENDED_CHAR_NAME, buffer, length)) { + return false; + } + } + ++start; + } + } + + return true; +} + +static UBool +enumNames(UCharNames *names, + UChar32 start, UChar32 limit, + UEnumCharNamesFn *fn, void *context, + UCharNameChoice nameChoice) { + uint16_t startGroupMSB, endGroupMSB, groupCount; + const uint16_t *group, *groupLimit; + + startGroupMSB=(uint16_t)(start>>GROUP_SHIFT); + endGroupMSB=(uint16_t)((limit-1)>>GROUP_SHIFT); + + /* find the group that contains start, or the highest before it */ + group=getGroup(names, start); + + if(startGroupMSBlimit) { + extLimit=limit; + } + if(!enumExtNames(start, extLimit-1, fn, context)) { + return false; + } + start=extLimit; + } + + if(startGroupMSB==endGroupMSB) { + if(startGroupMSB==group[GROUP_MSB]) { + /* if start and limit-1 are in the same group, then enumerate only in that one */ + return enumGroupNames(names, group, start, limit-1, fn, context, nameChoice); + } + } else { + const uint16_t *groups=GET_GROUPS(names); + groupCount=*groups++; + groupLimit=groups+groupCount*GROUP_LENGTH; + + if(startGroupMSB==group[GROUP_MSB]) { + /* enumerate characters in the partial start group */ + if((start&GROUP_MASK)!=0) { + if(!enumGroupNames(names, group, + start, ((UChar32)startGroupMSB<group[GROUP_MSB]) { + /* make sure that we start enumerating with the first group after start */ + const uint16_t *nextGroup=NEXT_GROUP(group); + if (nextGroup < groupLimit && nextGroup[GROUP_MSB] > startGroupMSB && nameChoice == U_EXTENDED_CHAR_NAME) { + UChar32 end = nextGroup[GROUP_MSB] << GROUP_SHIFT; + if (end > limit) { + end = limit; + } + if (!enumExtNames(start, end - 1, fn, context)) { + return false; + } + } + group=nextGroup; + } + + /* enumerate entire groups between the start- and end-groups */ + while(group group[GROUP_MSB] + 1 && nameChoice == U_EXTENDED_CHAR_NAME) { + UChar32 end = nextGroup[GROUP_MSB] << GROUP_SHIFT; + if (end > limit) { + end = limit; + } + if (!enumExtNames((group[GROUP_MSB] + 1) << GROUP_SHIFT, end - 1, fn, context)) { + return false; + } + } + group=nextGroup; + } + + /* enumerate within the end group (group[GROUP_MSB]==endGroupMSB) */ + if(group start) { + start = next; + } + } else { + return true; + } + } + + /* we have not found a group, which means everything is made of + extended names. */ + if (nameChoice == U_EXTENDED_CHAR_NAME) { + if (limit > UCHAR_MAX_VALUE + 1) { + limit = UCHAR_MAX_VALUE + 1; + } + return enumExtNames(start, limit - 1, fn, context); + } + + return true; +} + +static uint16_t +writeFactorSuffix(const uint16_t *factors, uint16_t count, + const char *s, /* suffix elements */ + uint32_t code, + uint16_t indexes[8], /* output fields from here */ + const char *elementBases[8], const char *elements[8], + char *buffer, uint16_t bufferLength) { + uint16_t i, factor, bufferPos=0; + char c; + + /* write elements according to the factors */ + + /* + * the factorized elements are determined by modulo arithmetic + * with the factors of this algorithm + * + * note that for fewer operations, count is decremented here + */ + --count; + for(i=count; i>0; --i) { + factor=factors[i]; + indexes[i]=(uint16_t)(code%factor); + code/=factor; + } + /* + * we don't need to calculate the last modulus because start<=code<=end + * guarantees here that code<=factors[0] + */ + indexes[0]=(uint16_t)code; + + /* write each element */ + for(;;) { + if(elementBases!=nullptr) { + *elementBases++=s; + } + + /* skip indexes[i] strings */ + factor=indexes[i]; + while(factor>0) { + while(*s++!=0) {} + --factor; + } + if(elements!=nullptr) { + *elements++=s; + } + + /* write element */ + while((c=*s++)!=0) { + WRITE_CHAR(buffer, bufferLength, bufferPos, c); + } + + /* we do not need to perform the rest of this loop for i==count - break here */ + if(i>=count) { + break; + } + + /* skip the rest of the strings for this factors[i] */ + factor=(uint16_t)(factors[i]-indexes[i]-1); + while(factor>0) { + while(*s++!=0) {} + --factor; + } + + ++i; + } + + /* zero-terminate */ + if(bufferLength>0) { + *buffer=0; + } + + return bufferPos; +} + +/* + * Important: + * Parts of findAlgName() are almost the same as some of getAlgName(). + * Fixes must be applied to both. + */ +static uint16_t +getAlgName(AlgorithmicRange *range, uint32_t code, UCharNameChoice nameChoice, + char *buffer, uint16_t bufferLength) { + uint16_t bufferPos=0; + + /* Only the normative character name can be algorithmic. */ + if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) { + /* zero-terminate */ + if(bufferLength>0) { + *buffer=0; + } + return 0; + } + + switch(range->type) { + case 0: { + /* name = prefix hex-digits */ + const char *s=(const char *)(range+1); + char c; + + uint16_t i, count; + + /* copy prefix */ + while((c=*s++)!=0) { + WRITE_CHAR(buffer, bufferLength, bufferPos, c); + } + + /* write hexadecimal code point value */ + count=range->variant; + + /* zero-terminate */ + if(count0;) { + if(--i>=4; + } + + bufferPos+=count; + break; + } + case 1: { + /* name = prefix factorized-elements */ + uint16_t indexes[8]; + const uint16_t *factors=(const uint16_t *)(range+1); + uint16_t count=range->variant; + const char *s=(const char *)(factors+count); + char c; + + /* copy prefix */ + while((c=*s++)!=0) { + WRITE_CHAR(buffer, bufferLength, bufferPos, c); + } + + bufferPos+=writeFactorSuffix(factors, count, + s, code-range->start, indexes, nullptr, nullptr, buffer, bufferLength); + break; + } + default: + /* undefined type */ + /* zero-terminate */ + if(bufferLength>0) { + *buffer=0; + } + break; + } + + return bufferPos; +} + +/* + * Important: enumAlgNames() and findAlgName() are almost the same. + * Any fix must be applied to both. + */ +static UBool +enumAlgNames(AlgorithmicRange *range, + UChar32 start, UChar32 limit, + UEnumCharNamesFn *fn, void *context, + UCharNameChoice nameChoice) { + char buffer[200]; + uint16_t length; + + if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) { + return true; + } + + switch(range->type) { + case 0: { + char *s, *end; + char c; + + /* get the full name of the start character */ + length=getAlgName(range, (uint32_t)start, nameChoice, buffer, sizeof(buffer)); + if(length<=0) { + return true; + } + + /* call the enumerator function with this first character */ + if(!fn(context, start, nameChoice, buffer, length)) { + return false; + } + + /* go to the end of the name; all these names have the same length */ + end=buffer; + while(*end!=0) { + ++end; + } + + /* enumerate the rest of the names */ + while(++startvariant; + const char *s=(const char *)(factors+count); + char *suffix, *t; + uint16_t prefixLength, i, idx; + + char c; + + /* name = prefix factorized-elements */ + + /* copy prefix */ + suffix=buffer; + prefixLength=0; + while((c=*s++)!=0) { + *suffix++=c; + ++prefixLength; + } + + /* append the suffix of the start character */ + length=(uint16_t)(prefixLength+writeFactorSuffix(factors, count, + s, (uint32_t)start-range->start, + indexes, elementBases, elements, + suffix, (uint16_t)(sizeof(buffer)-prefixLength))); + + /* call the enumerator function with this first character */ + if(!fn(context, start, nameChoice, buffer, length)) { + return false; + } + + /* enumerate the rest of the names */ + while(++starttype) { + case 0: { + /* name = prefix hex-digits */ + const char *s=(const char *)(range+1); + char c; + + uint16_t i, count; + + /* compare prefix */ + while((c=*s++)!=0) { + if((char)c!=*otherName++) { + return 0xffff; + } + } + + /* read hexadecimal code point value */ + count=range->variant; + code=0; + for(i=0; istart<=(uint32_t)code && (uint32_t)code<=range->end) { + return code; + } + break; + } + case 1: { + char buffer[64]; + uint16_t indexes[8]; + const char *elementBases[8], *elements[8]; + const uint16_t *factors=(const uint16_t *)(range+1); + uint16_t count=range->variant; + const char *s=(const char *)(factors+count), *t; + UChar32 start, limit; + uint16_t i, idx; + + char c; + + /* name = prefix factorized-elements */ + + /* compare prefix */ + while((c=*s++)!=0) { + if((char)c!=*otherName++) { + return 0xffff; + } + } + + start=(UChar32)range->start; + limit=(UChar32)(range->end+1); + + /* initialize the suffix elements for enumeration; indexes should all be set to 0 */ + writeFactorSuffix(factors, count, s, 0, + indexes, elementBases, elements, buffer, sizeof(buffer)); + + /* compare the first suffix */ + if(0==uprv_strcmp(otherName, buffer)) { + return start; + } + + /* enumerate and compare the rest of the suffixes */ + while(++start>5]|=((uint32_t)1<<((uint8_t)c&0x1f))) +#define SET_CONTAINS(set, c) (((set)[(uint8_t)c>>5]&((uint32_t)1<<((uint8_t)c&0x1f)))!=0) + +static int32_t +calcStringSetLength(uint32_t set[8], const char *s) { + int32_t length=0; + char c; + + while((c=*s++)!=0) { + SET_ADD(set, c); + ++length; + } + return length; +} + +static int32_t +calcAlgNameSetsLengths(int32_t maxNameLength) { + AlgorithmicRange *range; + uint32_t *p; + uint32_t rangeCount; + int32_t length; + + /* enumerate algorithmic ranges */ + p=(uint32_t *)((uint8_t *)uCharNames+uCharNames->algNamesOffset); + rangeCount=*p; + range=(AlgorithmicRange *)(p+1); + while(rangeCount>0) { + switch(range->type) { + case 0: + /* name = prefix + (range->variant times) hex-digits */ + /* prefix */ + length=calcStringSetLength(gNameSet, (const char *)(range+1))+range->variant; + if(length>maxNameLength) { + maxNameLength=length; + } + break; + case 1: { + /* name = prefix factorized-elements */ + const uint16_t *factors=(const uint16_t *)(range+1); + const char *s; + int32_t i, count=range->variant, factor, factorLength, maxFactorLength; + + /* prefix length */ + s=(const char *)(factors+count); + length=calcStringSetLength(gNameSet, s); + s+=length+1; /* start of factor suffixes */ + + /* get the set and maximum factor suffix length for each factor */ + for(i=0; i0; --factor) { + factorLength=calcStringSetLength(gNameSet, s); + s+=factorLength+1; + if(factorLength>maxFactorLength) { + maxFactorLength=factorLength; + } + } + length+=maxFactorLength; + } + + if(length>maxNameLength) { + maxNameLength=length; + } + break; + } + default: + /* unknown type */ + break; + } + + range=(AlgorithmicRange *)((uint8_t *)range+range->size); + --rangeCount; + } + return maxNameLength; +} + +static int32_t +calcExtNameSetsLengths(int32_t maxNameLength) { + int32_t i, length; + + for(i=0; i + * 1 for - + * 6 for most hex digits per code point + */ + length=9+calcStringSetLength(gNameSet, charCatNames[i]); + if(length>maxNameLength) { + maxNameLength=length; + } + } + return maxNameLength; +} + +static int32_t +calcNameSetLength(const uint16_t *tokens, uint16_t tokenCount, const uint8_t *tokenStrings, int8_t *tokenLengths, + uint32_t set[8], + const uint8_t **pLine, const uint8_t *lineLimit) { + const uint8_t *line=*pLine; + int32_t length=0, tokenLength; + uint16_t c, token; + + while(line!=lineLimit && (c=*line++)!=(uint8_t)';') { + if(c>=tokenCount) { + /* implicit letter */ + SET_ADD(set, c); + ++length; + } else { + token=tokens[c]; + if(token==(uint16_t)(-2)) { + /* this is a lead byte for a double-byte token */ + c=c<<8|*line++; + token=tokens[c]; + } + if(token==(uint16_t)(-1)) { + /* explicit letter */ + SET_ADD(set, c); + ++length; + } else { + /* count token word */ + if(tokenLengths!=nullptr) { + /* use cached token length */ + tokenLength=tokenLengths[c]; + if(tokenLength==0) { + tokenLength=calcStringSetLength(set, (const char *)tokenStrings+token); + tokenLengths[c]=(int8_t)tokenLength; + } + } else { + tokenLength=calcStringSetLength(set, (const char *)tokenStrings+token); + } + length+=tokenLength; + } + } + } + + *pLine=line; + return length; +} + +static void +calcGroupNameSetsLengths(int32_t maxNameLength) { + uint16_t offsets[LINES_PER_GROUP+2], lengths[LINES_PER_GROUP+2]; + + uint16_t *tokens=(uint16_t *)uCharNames+8; + uint16_t tokenCount=*tokens++; + uint8_t *tokenStrings=(uint8_t *)uCharNames+uCharNames->tokenStringOffset; + + int8_t *tokenLengths; + + const uint16_t *group; + const uint8_t *s, *line, *lineLimit; + + int32_t groupCount, lineNumber, length; + + tokenLengths=(int8_t *)uprv_malloc(tokenCount); + if(tokenLengths!=nullptr) { + uprv_memset(tokenLengths, 0, tokenCount); + } + + group=GET_GROUPS(uCharNames); + groupCount=*group++; + + /* enumerate all groups */ + while(groupCount>0) { + s=(uint8_t *)uCharNames+uCharNames->groupStringOffset+GET_GROUP_OFFSET(group); + s=expandGroupLengths(s, offsets, lengths); + + /* enumerate all lines in each group */ + for(lineNumber=0; lineNumbermaxNameLength) { + maxNameLength=length; + } + if(line==lineLimit) { + continue; + } + + /* read Unicode 1.0 name */ + length=calcNameSetLength(tokens, tokenCount, tokenStrings, tokenLengths, gNameSet, &line, lineLimit); + if(length>maxNameLength) { + maxNameLength=length; + } + if(line==lineLimit) { + continue; + } + + /* read ISO comment */ + /*length=calcNameSetLength(tokens, tokenCount, tokenStrings, tokenLengths, gISOCommentSet, &line, lineLimit);*/ + } + + group=NEXT_GROUP(group); + --groupCount; + } + + if(tokenLengths!=nullptr) { + uprv_free(tokenLengths); + } + + /* set gMax... - name length last for threading */ + gMaxNameLength=maxNameLength; +} + +static UBool +calcNameSetsLengths(UErrorCode *pErrorCode) { + static const char extChars[]="0123456789ABCDEF<>-"; + int32_t i, maxNameLength; + + if(gMaxNameLength!=0) { + return true; + } + + if(!isDataLoaded(pErrorCode)) { + return false; + } + + /* set hex digits, used in various names, and <>-, used in extended names */ + for(i=0; i<(int32_t)sizeof(extChars)-1; ++i) { + SET_ADD(gNameSet, extChars[i]); + } + + /* set sets and lengths from algorithmic names */ + maxNameLength=calcAlgNameSetsLengths(0); + + /* set sets and lengths from extended names */ + maxNameLength=calcExtNameSetsLengths(maxNameLength); + + /* set sets and lengths from group names, set global maximum values */ + calcGroupNameSetsLengths(maxNameLength); + + return true; +} + +U_NAMESPACE_END + +/* public API --------------------------------------------------------------- */ + +U_NAMESPACE_USE + +U_CAPI int32_t U_EXPORT2 +u_charName(UChar32 code, UCharNameChoice nameChoice, + char *buffer, int32_t bufferLength, + UErrorCode *pErrorCode) { + AlgorithmicRange *algRange; + uint32_t *p; + uint32_t i; + int32_t length; + + /* check the argument values */ + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } else if(nameChoice>=U_CHAR_NAME_CHOICE_COUNT || + bufferLength<0 || (bufferLength>0 && buffer==nullptr) + ) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + if((uint32_t)code>UCHAR_MAX_VALUE || !isDataLoaded(pErrorCode)) { + return u_terminateChars(buffer, bufferLength, 0, pErrorCode); + } + + length=0; + + /* try algorithmic names first */ + p=(uint32_t *)((uint8_t *)uCharNames+uCharNames->algNamesOffset); + i=*p; + algRange=(AlgorithmicRange *)(p+1); + while(i>0) { + if(algRange->start<=(uint32_t)code && (uint32_t)code<=algRange->end) { + length=getAlgName(algRange, (uint32_t)code, nameChoice, buffer, (uint16_t)bufferLength); + break; + } + algRange=(AlgorithmicRange *)((uint8_t *)algRange+algRange->size); + --i; + } + + if(i==0) { + if (nameChoice == U_EXTENDED_CHAR_NAME) { + length = getName(uCharNames, (uint32_t )code, U_EXTENDED_CHAR_NAME, buffer, (uint16_t) bufferLength); + if (!length) { + /* extended character name */ + length = getExtName((uint32_t) code, buffer, (uint16_t) bufferLength); + } + } else { + /* normal character name */ + length=getName(uCharNames, (uint32_t)code, nameChoice, buffer, (uint16_t)bufferLength); + } + } + + return u_terminateChars(buffer, bufferLength, length, pErrorCode); +} + +U_CAPI int32_t U_EXPORT2 +u_getISOComment(UChar32 /*c*/, + char *dest, int32_t destCapacity, + UErrorCode *pErrorCode) { + /* check the argument values */ + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return 0; + } else if(destCapacity<0 || (destCapacity>0 && dest==nullptr)) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return 0; + } + + return u_terminateChars(dest, destCapacity, 0, pErrorCode); +} + +U_CAPI UChar32 U_EXPORT2 +u_charFromName(UCharNameChoice nameChoice, + const char *name, + UErrorCode *pErrorCode) { + char upper[120] = {0}; + char lower[120] = {0}; + FindName findName; + AlgorithmicRange *algRange; + uint32_t *p; + uint32_t i; + UChar32 cp = 0; + char c0; + static constexpr UChar32 error = 0xffff; /* Undefined, but use this for backwards compatibility. */ + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return error; + } + + if(nameChoice>=U_CHAR_NAME_CHOICE_COUNT || name==nullptr || *name==0) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return error; + } + + if(!isDataLoaded(pErrorCode)) { + return error; + } + + /* construct the uppercase and lowercase of the name first */ + for(i=0; i') { + // Parse a string like "" where HHHH is a hex code point. + uint32_t limit = i; + while (i >= 3 && lower[--i] != '-') {} + + // There should be 1 to 8 hex digits. + int32_t hexLength = limit - (i + 1); + if (i >= 2 && lower[i] == '-' && 1 <= hexLength && hexLength <= 8) { + uint32_t cIdx; + + lower[i] = 0; + + for (++i; i < limit; ++i) { + if (lower[i] >= '0' && lower[i] <= '9') { + cp = (cp << 4) + lower[i] - '0'; + } else if (lower[i] >= 'a' && lower[i] <= 'f') { + cp = (cp << 4) + lower[i] - 'a' + 10; + } else { + *pErrorCode = U_ILLEGAL_CHAR_FOUND; + return error; + } + // Prevent signed-integer overflow and out-of-range code points. + if (cp > UCHAR_MAX_VALUE) { + *pErrorCode = U_ILLEGAL_CHAR_FOUND; + return error; + } + } + + /* Now validate the category name. + We could use a binary search, or a trie, if + we really wanted to. */ + uint8_t cat = getCharCat(cp); + for (lower[i] = 0, cIdx = 0; cIdx < UPRV_LENGTHOF(charCatNames); ++cIdx) { + + if (!uprv_strcmp(lower + 1, charCatNames[cIdx])) { + if (cat == cIdx) { + return cp; + } + break; + } + } + } + } + + *pErrorCode = U_ILLEGAL_CHAR_FOUND; + return error; + } + + /* try algorithmic names now */ + p=(uint32_t *)((uint8_t *)uCharNames+uCharNames->algNamesOffset); + i=*p; + algRange=(AlgorithmicRange *)(p+1); + while(i>0) { + if((cp=findAlgName(algRange, nameChoice, upper))!=0xffff) { + return cp; + } + algRange=(AlgorithmicRange *)((uint8_t *)algRange+algRange->size); + --i; + } + + /* normal character name */ + findName.otherName=upper; + findName.code=error; + enumNames(uCharNames, 0, UCHAR_MAX_VALUE + 1, DO_FIND_NAME, &findName, nameChoice); + if (findName.code == error) { + *pErrorCode = U_ILLEGAL_CHAR_FOUND; + } + return findName.code; +} + +U_CAPI void U_EXPORT2 +u_enumCharNames(UChar32 start, UChar32 limit, + UEnumCharNamesFn *fn, + void *context, + UCharNameChoice nameChoice, + UErrorCode *pErrorCode) { + AlgorithmicRange *algRange; + uint32_t *p; + uint32_t i; + + if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { + return; + } + + if(nameChoice>=U_CHAR_NAME_CHOICE_COUNT || fn==nullptr) { + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; + return; + } + + if((uint32_t) limit > UCHAR_MAX_VALUE + 1) { + limit = UCHAR_MAX_VALUE + 1; + } + if((uint32_t)start>=(uint32_t)limit) { + return; + } + + if(!isDataLoaded(pErrorCode)) { + return; + } + + /* interleave the data-driven ones with the algorithmic ones */ + /* iterate over all algorithmic ranges; assume that they are in ascending order */ + p=(uint32_t *)((uint8_t *)uCharNames+uCharNames->algNamesOffset); + i=*p; + algRange=(AlgorithmicRange *)(p+1); + while(i>0) { + /* enumerate the character names before the current algorithmic range */ + /* here: startstart) { + if((uint32_t)limit<=algRange->start) { + enumNames(uCharNames, start, limit, fn, context, nameChoice); + return; + } + if(!enumNames(uCharNames, start, (UChar32)algRange->start, fn, context, nameChoice)) { + return; + } + start=(UChar32)algRange->start; + } + /* enumerate the character names in the current algorithmic range */ + /* here: algRange->start<=startend) { + if((uint32_t)limit<=(algRange->end+1)) { + enumAlgNames(algRange, start, limit, fn, context, nameChoice); + return; + } + if(!enumAlgNames(algRange, start, (UChar32)algRange->end+1, fn, context, nameChoice)) { + return; + } + start=(UChar32)algRange->end+1; + } + /* continue to the next algorithmic range (here: startsize); + --i; + } + /* enumerate the character names after the last algorithmic range */ + enumNames(uCharNames, start, limit, fn, context, nameChoice); +} + +U_CAPI int32_t U_EXPORT2 +uprv_getMaxCharNameLength() { + UErrorCode errorCode=U_ZERO_ERROR; + if(calcNameSetsLengths(&errorCode)) { + return gMaxNameLength; + } else { + return 0; + } +} + +/** + * Converts the char set cset into a Unicode set uset. + * @param cset Set of 256 bit flags corresponding to a set of chars. + * @param uset USet to receive characters. Existing contents are deleted. + */ +static void +charSetToUSet(uint32_t cset[8], const USetAdder *sa) { + char16_t us[256]; + char cs[256]; + + int32_t i, length; + UErrorCode errorCode; + + errorCode=U_ZERO_ERROR; + + if(!calcNameSetsLengths(&errorCode)) { + return; + } + + /* build a char string with all chars that are used in character names */ + length=0; + for(i=0; i<256; ++i) { + if(SET_CONTAINS(cset, i)) { + cs[length++]=(char)i; + } + } + + /* convert the char string to a char16_t string */ + u_charsToUChars(cs, us, length); + + /* add each char16_t to the USet */ + for(i=0; iadd(sa->set, us[i]); + } + } +} + +/** + * Fills set with characters that are used in Unicode character names. + * @param set USet to receive characters. + */ +U_CAPI void U_EXPORT2 +uprv_getCharNameCharacters(const USetAdder *sa) { + charSetToUSet(gNameSet, sa); +} + +/* data swapping ------------------------------------------------------------ */ + +/* + * The token table contains non-negative entries for token bytes, + * and -1 for bytes that represent themselves in the data file's charset. + * -2 entries are used for lead bytes. + * + * Direct bytes (-1 entries) must be translated from the input charset family + * to the output charset family. + * makeTokenMap() writes a permutation mapping for this. + * Use it once for single-/lead-byte tokens and once more for all trail byte + * tokens. (';' is an unused trail byte marked with -1.) + */ +static void +makeTokenMap(const UDataSwapper *ds, + int16_t tokens[], uint16_t tokenCount, + uint8_t map[256], + UErrorCode *pErrorCode) { + UBool usedOutChar[256]; + uint16_t i, j; + uint8_t c1, c2; + + if(U_FAILURE(*pErrorCode)) { + return; + } + + if(ds->inCharset==ds->outCharset) { + /* Same charset family: identity permutation */ + for(i=0; i<256; ++i) { + map[i]=(uint8_t)i; + } + } else { + uprv_memset(map, 0, 256); + uprv_memset(usedOutChar, 0, 256); + + if(tokenCount>256) { + tokenCount=256; + } + + /* set the direct bytes (byte 0 always maps to itself) */ + for(i=1; iswapInvChars(ds, &c1, 1, &c2, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + udata_printError(ds, "unames/makeTokenMap() finds variant character 0x%02x used (input charset family %d)\n", + i, ds->inCharset); + return; + } + + /* enter the converted character into the map and mark it used */ + map[c1]=c2; + usedOutChar[c2]=true; + } + } + + /* set the mappings for the rest of the permutation */ + for(i=j=1; idataFormat[0]==0x75 && /* dataFormat="unam" */ + pInfo->dataFormat[1]==0x6e && + pInfo->dataFormat[2]==0x61 && + pInfo->dataFormat[3]==0x6d && + pInfo->formatVersion[0]==1 + )) { + udata_printError(ds, "uchar_swapNames(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized as unames.icu\n", + pInfo->dataFormat[0], pInfo->dataFormat[1], + pInfo->dataFormat[2], pInfo->dataFormat[3], + pInfo->formatVersion[0]); + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + + inBytes=(const uint8_t *)inData+headerSize; + outBytes=(outData == nullptr) ? nullptr : (uint8_t *)outData+headerSize; + if(length<0) { + algNamesOffset=ds->readUInt32(((const uint32_t *)inBytes)[3]); + } else { + length-=headerSize; + if( length<20 || + (uint32_t)length<(algNamesOffset=ds->readUInt32(((const uint32_t *)inBytes)[3])) + ) { + udata_printError(ds, "uchar_swapNames(): too few bytes (%d after header) for unames.icu\n", + length); + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + } + + if(length<0) { + /* preflighting: iterate through algorithmic ranges */ + offset=algNamesOffset; + count=ds->readUInt32(*((const uint32_t *)(inBytes+offset))); + offset+=4; + + for(i=0; ireadUInt16(inRange->size); + } + } else { + /* swap data */ + const uint16_t *p; + uint16_t *q, *temp; + + int16_t tokens[512]; + uint16_t tokenCount; + + uint8_t map[256], trailMap[256]; + + /* copy the data for inaccessible bytes */ + if(inBytes!=outBytes) { + uprv_memcpy(outBytes, inBytes, length); + } + + /* the initial 4 offsets first */ + tokenStringOffset=ds->readUInt32(((const uint32_t *)inBytes)[0]); + groupsOffset=ds->readUInt32(((const uint32_t *)inBytes)[1]); + groupStringOffset=ds->readUInt32(((const uint32_t *)inBytes)[2]); + ds->swapArray32(ds, inBytes, 16, outBytes, pErrorCode); + + /* + * now the tokens table + * it needs to be permutated along with the compressed name strings + */ + p=(const uint16_t *)(inBytes+16); + q=(uint16_t *)(outBytes+16); + + /* read and swap the tokenCount */ + tokenCount=ds->readUInt16(*p); + ds->swapArray16(ds, p, 2, q, pErrorCode); + ++p; + ++q; + + /* read the first 512 tokens and make the token maps */ + if(tokenCount<=512) { + count=tokenCount; + } else { + count=512; + } + for(i=0; i256 ? tokenCount-256 : 0), trailMap, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + return 0; + } + + /* + * swap and permutate the tokens + * go through a temporary array to support in-place swapping + */ + temp=(uint16_t *)uprv_malloc(tokenCount*2); + if(temp==nullptr) { + udata_printError(ds, "out of memory swapping %u unames.icu tokens\n", + tokenCount); + *pErrorCode=U_MEMORY_ALLOCATION_ERROR; + return 0; + } + + /* swap and permutate single-/lead-byte tokens */ + for(i=0; iswapArray16(ds, p+i, 2, temp+map[i], pErrorCode); + } + + /* swap and permutate trail-byte tokens */ + for(; iswapArray16(ds, p+i, 2, temp+(i&0xffffff00)+trailMap[i&0xff], pErrorCode); + } + + /* copy the result into the output and free the temporary array */ + uprv_memcpy(q, temp, tokenCount*2); + uprv_free(temp); + + /* + * swap the token strings but not a possible padding byte after + * the terminating NUL of the last string + */ + udata_swapInvStringBlock(ds, inBytes+tokenStringOffset, (int32_t)(groupsOffset-tokenStringOffset), + outBytes+tokenStringOffset, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + udata_printError(ds, "uchar_swapNames(token strings) failed\n"); + return 0; + } + + /* swap the group table */ + count=ds->readUInt16(*((const uint16_t *)(inBytes+groupsOffset))); + ds->swapArray16(ds, inBytes+groupsOffset, (int32_t)((1+count*3)*2), + outBytes+groupsOffset, pErrorCode); + + /* + * swap the group strings + * swap the string bytes but not the nibble-encoded string lengths + */ + if(ds->inCharset!=ds->outCharset) { + uint16_t offsets[LINES_PER_GROUP+1], lengths[LINES_PER_GROUP+1]; + + const uint8_t *inStrings, *nextInStrings; + uint8_t *outStrings; + + uint8_t c; + + inStrings=inBytes+groupStringOffset; + outStrings=outBytes+groupStringOffset; + + stringsCount=algNamesOffset-groupStringOffset; + + /* iterate through string groups until only a few padding bytes are left */ + while(stringsCount>32) { + nextInStrings=expandGroupLengths(inStrings, offsets, lengths); + + /* move past the length bytes */ + stringsCount-=(uint32_t)(nextInStrings-inStrings); + outStrings+=nextInStrings-inStrings; + inStrings=nextInStrings; + + count=offsets[31]+lengths[31]; /* total number of string bytes in this group */ + stringsCount-=count; + + /* swap the string bytes using map[] and trailMap[] */ + while(count>0) { + c=*inStrings++; + *outStrings++=map[c]; + if(tokens[c]!=-2) { + --count; + } else { + /* token lead byte: swap the trail byte, too */ + *outStrings++=trailMap[*inStrings++]; + count-=2; + } + } + } + } + + /* swap the algorithmic ranges */ + offset=algNamesOffset; + count=ds->readUInt32(*((const uint32_t *)(inBytes+offset))); + ds->swapArray32(ds, inBytes+offset, 4, outBytes+offset, pErrorCode); + offset+=4; + + for(i=0; i(uint32_t)length) { + udata_printError(ds, "uchar_swapNames(): too few bytes (%d after header) for unames.icu algorithmic range %u\n", + length, i); + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; + return 0; + } + + inRange=(const AlgorithmicRange *)(inBytes+offset); + outRange=(AlgorithmicRange *)(outBytes+offset); + offset+=ds->readUInt16(inRange->size); + + ds->swapArray32(ds, inRange, 8, outRange, pErrorCode); + ds->swapArray16(ds, &inRange->size, 2, &outRange->size, pErrorCode); + switch(inRange->type) { + case 0: + /* swap prefix string */ + ds->swapInvChars(ds, inRange+1, (int32_t)uprv_strlen((const char *)(inRange+1)), + outRange+1, pErrorCode); + if(U_FAILURE(*pErrorCode)) { + udata_printError(ds, "uchar_swapNames(prefix string of algorithmic range %u) failed\n", + i); + return 0; + } + break; + case 1: + { + /* swap factors and the prefix and factor strings */ + uint32_t factorsCount; + + factorsCount=inRange->variant; + p=(const uint16_t *)(inRange+1); + q=(uint16_t *)(outRange+1); + ds->swapArray16(ds, p, (int32_t)(factorsCount*2), q, pErrorCode); + + /* swap the strings, up to the last terminating NUL */ + p+=factorsCount; + q+=factorsCount; + stringsCount=(uint32_t)((inBytes+offset)-(const uint8_t *)p); + while(stringsCount>0 && ((const uint8_t *)p)[stringsCount-1]!=0) { + --stringsCount; + } + ds->swapInvChars(ds, p, (int32_t)stringsCount, q, pErrorCode); + } + break; + default: + udata_printError(ds, "uchar_swapNames(): unknown type %u of algorithmic range %u\n", + inRange->type, i); + *pErrorCode=U_UNSUPPORTED_ERROR; + return 0; + } + } + } + + return headerSize+(int32_t)offset; +} + +/* + * Hey, Emacs, please set the following: + * + * Local Variables: + * indent-tabs-mode: nil + * End: + * + */ diff --git a/intl/icu/source/common/unicode/appendable.h b/intl/icu/source/common/unicode/appendable.h new file mode 100644 index 0000000000..0e37f4562a --- /dev/null +++ b/intl/icu/source/common/unicode/appendable.h @@ -0,0 +1,239 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2011-2012, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: appendable.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010dec07 +* created by: Markus W. Scherer +*/ + +#ifndef __APPENDABLE_H__ +#define __APPENDABLE_H__ + +/** + * \file + * \brief C++ API: Appendable class: Sink for Unicode code points and 16-bit code units (char16_ts). + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/uobject.h" + +U_NAMESPACE_BEGIN + +class UnicodeString; + +/** + * Base class for objects to which Unicode characters and strings can be appended. + * Combines elements of Java Appendable and ICU4C ByteSink. + * + * This class can be used in APIs where it does not matter whether the actual destination is + * a UnicodeString, a char16_t[] array, a UnicodeSet, or any other object + * that receives and processes characters and/or strings. + * + * Implementation classes must implement at least appendCodeUnit(char16_t). + * The base class provides default implementations for the other methods. + * + * The methods do not take UErrorCode parameters. + * If an error occurs (e.g., out-of-memory), + * in addition to returning false from failing operations, + * the implementation must prevent unexpected behavior (e.g., crashes) + * from further calls and should make the error condition available separately + * (e.g., store a UErrorCode, make/keep a UnicodeString bogus). + * @stable ICU 4.8 + */ +class U_COMMON_API Appendable : public UObject { +public: + /** + * Destructor. + * @stable ICU 4.8 + */ + ~Appendable(); + + /** + * Appends a 16-bit code unit. + * @param c code unit + * @return true if the operation succeeded + * @stable ICU 4.8 + */ + virtual UBool appendCodeUnit(char16_t c) = 0; + + /** + * Appends a code point. + * The default implementation calls appendCodeUnit(char16_t) once or twice. + * @param c code point 0..0x10ffff + * @return true if the operation succeeded + * @stable ICU 4.8 + */ + virtual UBool appendCodePoint(UChar32 c); + + /** + * Appends a string. + * The default implementation calls appendCodeUnit(char16_t) for each code unit. + * @param s string, must not be nullptr if length!=0 + * @param length string length, or -1 if NUL-terminated + * @return true if the operation succeeded + * @stable ICU 4.8 + */ + virtual UBool appendString(const char16_t *s, int32_t length); + + /** + * Tells the object that the caller is going to append roughly + * appendCapacity char16_ts. A subclass might use this to pre-allocate + * a larger buffer if necessary. + * The default implementation does nothing. (It always returns true.) + * @param appendCapacity estimated number of char16_ts that will be appended + * @return true if the operation succeeded + * @stable ICU 4.8 + */ + virtual UBool reserveAppendCapacity(int32_t appendCapacity); + + /** + * Returns a writable buffer for appending and writes the buffer's capacity to + * *resultCapacity. Guarantees *resultCapacity>=minCapacity. + * May return a pointer to the caller-owned scratch buffer which must have + * scratchCapacity>=minCapacity. + * The returned buffer is only valid until the next operation + * on this Appendable. + * + * After writing at most *resultCapacity char16_ts, call appendString() with the + * pointer returned from this function and the number of char16_ts written. + * Many appendString() implementations will avoid copying char16_ts if this function + * returned an internal buffer. + * + * Partial usage example: + * \code + * int32_t capacity; + * char16_t* buffer = app.getAppendBuffer(..., &capacity); + * ... Write n char16_ts into buffer, with n <= capacity. + * app.appendString(buffer, n); + * \endcode + * In many implementations, that call to append will avoid copying char16_ts. + * + * If the Appendable allocates or reallocates an internal buffer, it should use + * the desiredCapacityHint if appropriate. + * If a caller cannot provide a reasonable guess at the desired capacity, + * it should pass desiredCapacityHint=0. + * + * If a non-scratch buffer is returned, the caller may only pass + * a prefix to it to appendString(). + * That is, it is not correct to pass an interior pointer to appendString(). + * + * The default implementation always returns the scratch buffer. + * + * @param minCapacity required minimum capacity of the returned buffer; + * must be non-negative + * @param desiredCapacityHint desired capacity of the returned buffer; + * must be non-negative + * @param scratch default caller-owned buffer + * @param scratchCapacity capacity of the scratch buffer + * @param resultCapacity pointer to an integer which will be set to the + * capacity of the returned buffer + * @return a buffer with *resultCapacity>=minCapacity + * @stable ICU 4.8 + */ + virtual char16_t *getAppendBuffer(int32_t minCapacity, + int32_t desiredCapacityHint, + char16_t *scratch, int32_t scratchCapacity, + int32_t *resultCapacity); +}; + +/** + * An Appendable implementation which writes to a UnicodeString. + * + * This class is not intended for public subclassing. + * @stable ICU 4.8 + */ +class U_COMMON_API UnicodeStringAppendable : public Appendable { +public: + /** + * Aliases the UnicodeString (keeps its reference) for writing. + * @param s The UnicodeString to which this Appendable will write. + * @stable ICU 4.8 + */ + explicit UnicodeStringAppendable(UnicodeString &s) : str(s) {} + + /** + * Destructor. + * @stable ICU 4.8 + */ + ~UnicodeStringAppendable(); + + /** + * Appends a 16-bit code unit to the string. + * @param c code unit + * @return true if the operation succeeded + * @stable ICU 4.8 + */ + virtual UBool appendCodeUnit(char16_t c) override; + + /** + * Appends a code point to the string. + * @param c code point 0..0x10ffff + * @return true if the operation succeeded + * @stable ICU 4.8 + */ + virtual UBool appendCodePoint(UChar32 c) override; + + /** + * Appends a string to the UnicodeString. + * @param s string, must not be nullptr if length!=0 + * @param length string length, or -1 if NUL-terminated + * @return true if the operation succeeded + * @stable ICU 4.8 + */ + virtual UBool appendString(const char16_t *s, int32_t length) override; + + /** + * Tells the UnicodeString that the caller is going to append roughly + * appendCapacity char16_ts. + * @param appendCapacity estimated number of char16_ts that will be appended + * @return true if the operation succeeded + * @stable ICU 4.8 + */ + virtual UBool reserveAppendCapacity(int32_t appendCapacity) override; + + /** + * Returns a writable buffer for appending and writes the buffer's capacity to + * *resultCapacity. Guarantees *resultCapacity>=minCapacity. + * May return a pointer to the caller-owned scratch buffer which must have + * scratchCapacity>=minCapacity. + * The returned buffer is only valid until the next write operation + * on the UnicodeString. + * + * For details see Appendable::getAppendBuffer(). + * + * @param minCapacity required minimum capacity of the returned buffer; + * must be non-negative + * @param desiredCapacityHint desired capacity of the returned buffer; + * must be non-negative + * @param scratch default caller-owned buffer + * @param scratchCapacity capacity of the scratch buffer + * @param resultCapacity pointer to an integer which will be set to the + * capacity of the returned buffer + * @return a buffer with *resultCapacity>=minCapacity + * @stable ICU 4.8 + */ + virtual char16_t *getAppendBuffer(int32_t minCapacity, + int32_t desiredCapacityHint, + char16_t *scratch, int32_t scratchCapacity, + int32_t *resultCapacity) override; + +private: + UnicodeString &str; +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __APPENDABLE_H__ diff --git a/intl/icu/source/common/unicode/brkiter.h b/intl/icu/source/common/unicode/brkiter.h new file mode 100644 index 0000000000..108652799e --- /dev/null +++ b/intl/icu/source/common/unicode/brkiter.h @@ -0,0 +1,670 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************** +* Copyright (C) 1997-2016, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************** +* +* File brkiter.h +* +* Modification History: +* +* Date Name Description +* 02/18/97 aliu Added typedef for TextCount. Made DONE const. +* 05/07/97 aliu Fixed DLL declaration. +* 07/09/97 jfitz Renamed BreakIterator and interface synced with JDK +* 08/11/98 helena Sync-up JDK1.2. +* 01/13/2000 helena Added UErrorCode parameter to createXXXInstance methods. +******************************************************************************** +*/ + +#ifndef BRKITER_H +#define BRKITER_H + +#include "unicode/utypes.h" + +/** + * \file + * \brief C++ API: Break Iterator. + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#if UCONFIG_NO_BREAK_ITERATION + +U_NAMESPACE_BEGIN + +/* + * Allow the declaration of APIs with pointers to BreakIterator + * even when break iteration is removed from the build. + */ +class BreakIterator; + +U_NAMESPACE_END + +#else + +#include "unicode/uobject.h" +#include "unicode/unistr.h" +#include "unicode/chariter.h" +#include "unicode/locid.h" +#include "unicode/ubrk.h" +#include "unicode/strenum.h" +#include "unicode/utext.h" +#include "unicode/umisc.h" + +U_NAMESPACE_BEGIN + +/** + * The BreakIterator class implements methods for finding the location + * of boundaries in text. BreakIterator is an abstract base class. + * Instances of BreakIterator maintain a current position and scan over + * text returning the index of characters where boundaries occur. + *

+ * Line boundary analysis determines where a text string can be broken + * when line-wrapping. The mechanism correctly handles punctuation and + * hyphenated words. + *

+ * Sentence boundary analysis allows selection with correct + * interpretation of periods within numbers and abbreviations, and + * trailing punctuation marks such as quotation marks and parentheses. + *

+ * Word boundary analysis is used by search and replace functions, as + * well as within text editing applications that allow the user to + * select words with a double click. Word selection provides correct + * interpretation of punctuation marks within and following + * words. Characters that are not part of a word, such as symbols or + * punctuation marks, have word-breaks on both sides. + *

+ * Character boundary analysis allows users to interact with + * characters as they expect to, for example, when moving the cursor + * through a text string. Character boundary analysis provides correct + * navigation of through character strings, regardless of how the + * character is stored. For example, an accented character might be + * stored as a base character and a diacritical mark. What users + * consider to be a character can differ between languages. + *

+ * The text boundary positions are found according to the rules + * described in Unicode Standard Annex #29, Text Boundaries, and + * Unicode Standard Annex #14, Line Breaking Properties. These + * are available at http://www.unicode.org/reports/tr14/ and + * http://www.unicode.org/reports/tr29/. + *

+ * In addition to the C++ API defined in this header file, a + * plain C API with equivalent functionality is defined in the + * file ubrk.h + *

+ * Code snippets illustrating the use of the Break Iterator APIs + * are available in the ICU User Guide, + * https://unicode-org.github.io/icu/userguide/boundaryanalysis/ + * and in the sample program icu/source/samples/break/break.cpp + * + */ +class U_COMMON_API BreakIterator : public UObject { +public: + /** + * destructor + * @stable ICU 2.0 + */ + virtual ~BreakIterator(); + + /** + * Return true if another object is semantically equal to this + * one. The other object should be an instance of the same subclass of + * BreakIterator. Objects of different subclasses are considered + * unequal. + *

+ * Return true if this BreakIterator is at the same position in the + * same text, and is the same class and type (word, line, etc.) of + * BreakIterator, as the argument. Text is considered the same if + * it contains the same characters, it need not be the same + * object, and styles are not considered. + * @stable ICU 2.0 + */ + virtual bool operator==(const BreakIterator&) const = 0; + + /** + * Returns the complement of the result of operator== + * @param rhs The BreakIterator to be compared for inequality + * @return the complement of the result of operator== + * @stable ICU 2.0 + */ + bool operator!=(const BreakIterator& rhs) const { return !operator==(rhs); } + + /** + * Return a polymorphic copy of this object. This is an abstract + * method which subclasses implement. + * @stable ICU 2.0 + */ + virtual BreakIterator* clone() const = 0; + + /** + * Return a polymorphic class ID for this object. Different subclasses + * will return distinct unequal values. + * @stable ICU 2.0 + */ + virtual UClassID getDynamicClassID(void) const override = 0; + + /** + * Return a CharacterIterator over the text being analyzed. + * @stable ICU 2.0 + */ + virtual CharacterIterator& getText(void) const = 0; + + + /** + * Get a UText for the text being analyzed. + * The returned UText is a shallow clone of the UText used internally + * by the break iterator implementation. It can safely be used to + * access the text without impacting any break iterator operations, + * but the underlying text itself must not be altered. + * + * @param fillIn A UText to be filled in. If nullptr, a new UText will be + * allocated to hold the result. + * @param status receives any error codes. + * @return The current UText for this break iterator. If an input + * UText was provided, it will always be returned. + * @stable ICU 3.4 + */ + virtual UText *getUText(UText *fillIn, UErrorCode &status) const = 0; + + /** + * Change the text over which this operates. The text boundary is + * reset to the start. + * + * The BreakIterator will retain a reference to the supplied string. + * The caller must not modify or delete the text while the BreakIterator + * retains the reference. + * + * @param text The UnicodeString used to change the text. + * @stable ICU 2.0 + */ + virtual void setText(const UnicodeString &text) = 0; + + /** + * Reset the break iterator to operate over the text represented by + * the UText. The iterator position is reset to the start. + * + * This function makes a shallow clone of the supplied UText. This means + * that the caller is free to immediately close or otherwise reuse the + * Utext that was passed as a parameter, but that the underlying text itself + * must not be altered while being referenced by the break iterator. + * + * All index positions returned by break iterator functions are + * native indices from the UText. For example, when breaking UTF-8 + * encoded text, the break positions returned by next(), previous(), etc. + * will be UTF-8 string indices, not UTF-16 positions. + * + * @param text The UText used to change the text. + * @param status receives any error codes. + * @stable ICU 3.4 + */ + virtual void setText(UText *text, UErrorCode &status) = 0; + + /** + * Change the text over which this operates. The text boundary is + * reset to the start. + * Note that setText(UText *) provides similar functionality to this function, + * and is more efficient. + * @param it The CharacterIterator used to change the text. + * @stable ICU 2.0 + */ + virtual void adoptText(CharacterIterator* it) = 0; + + enum { + /** + * DONE is returned by previous() and next() after all valid + * boundaries have been returned. + * @stable ICU 2.0 + */ + DONE = (int32_t)-1 + }; + + /** + * Sets the current iteration position to the beginning of the text, position zero. + * @return The offset of the beginning of the text, zero. + * @stable ICU 2.0 + */ + virtual int32_t first(void) = 0; + + /** + * Set the iterator position to the index immediately BEYOND the last character in the text being scanned. + * @return The index immediately BEYOND the last character in the text being scanned. + * @stable ICU 2.0 + */ + virtual int32_t last(void) = 0; + + /** + * Set the iterator position to the boundary preceding the current boundary. + * @return The character index of the previous text boundary or DONE if all + * boundaries have been returned. + * @stable ICU 2.0 + */ + virtual int32_t previous(void) = 0; + + /** + * Advance the iterator to the boundary following the current boundary. + * @return The character index of the next text boundary or DONE if all + * boundaries have been returned. + * @stable ICU 2.0 + */ + virtual int32_t next(void) = 0; + + /** + * Return character index of the current iterator position within the text. + * @return The boundary most recently returned. + * @stable ICU 2.0 + */ + virtual int32_t current(void) const = 0; + + /** + * Advance the iterator to the first boundary following the specified offset. + * The value returned is always greater than the offset or + * the value BreakIterator.DONE + * @param offset the offset to begin scanning. + * @return The first boundary after the specified offset. + * @stable ICU 2.0 + */ + virtual int32_t following(int32_t offset) = 0; + + /** + * Set the iterator position to the first boundary preceding the specified offset. + * The value returned is always smaller than the offset or + * the value BreakIterator.DONE + * @param offset the offset to begin scanning. + * @return The first boundary before the specified offset. + * @stable ICU 2.0 + */ + virtual int32_t preceding(int32_t offset) = 0; + + /** + * Return true if the specified position is a boundary position. + * As a side effect, the current position of the iterator is set + * to the first boundary position at or following the specified offset. + * @param offset the offset to check. + * @return True if "offset" is a boundary position. + * @stable ICU 2.0 + */ + virtual UBool isBoundary(int32_t offset) = 0; + + /** + * Set the iterator position to the nth boundary from the current boundary + * @param n the number of boundaries to move by. A value of 0 + * does nothing. Negative values move to previous boundaries + * and positive values move to later boundaries. + * @return The new iterator position, or + * DONE if there are fewer than |n| boundaries in the specified direction. + * @stable ICU 2.0 + */ + virtual int32_t next(int32_t n) = 0; + + /** + * For RuleBasedBreakIterators, return the status tag from the break rule + * that determined the boundary at the current iteration position. + *

+ * For break iterator types that do not support a rule status, + * a default value of 0 is returned. + *

+ * @return the status from the break rule that determined the boundary at + * the current iteration position. + * @see RuleBaseBreakIterator::getRuleStatus() + * @see UWordBreak + * @stable ICU 52 + */ + virtual int32_t getRuleStatus() const; + + /** + * For RuleBasedBreakIterators, get the status (tag) values from the break rule(s) + * that determined the boundary at the current iteration position. + *

+ * For break iterator types that do not support rule status, + * no values are returned. + *

+ * The returned status value(s) are stored into an array provided by the caller. + * The values are stored in sorted (ascending) order. + * If the capacity of the output array is insufficient to hold the data, + * the output will be truncated to the available length, and a + * U_BUFFER_OVERFLOW_ERROR will be signaled. + *

+ * @see RuleBaseBreakIterator::getRuleStatusVec + * + * @param fillInVec an array to be filled in with the status values. + * @param capacity the length of the supplied vector. A length of zero causes + * the function to return the number of status values, in the + * normal way, without attempting to store any values. + * @param status receives error codes. + * @return The number of rule status values from rules that determined + * the boundary at the current iteration position. + * In the event of a U_BUFFER_OVERFLOW_ERROR, the return value + * is the total number of status values that were available, + * not the reduced number that were actually returned. + * @see getRuleStatus + * @stable ICU 52 + */ + virtual int32_t getRuleStatusVec(int32_t *fillInVec, int32_t capacity, UErrorCode &status); + + /** + * Create BreakIterator for word-breaks using the given locale. + * Returns an instance of a BreakIterator implementing word breaks. + * WordBreak is useful for word selection (ex. double click) + * @param where the locale. + * @param status the error code + * @return A BreakIterator for word-breaks. The UErrorCode& status + * parameter is used to return status information to the user. + * To check whether the construction succeeded or not, you should check + * the value of U_SUCCESS(err). If you wish more detailed information, you + * can check for informational error results which still indicate success. + * U_USING_FALLBACK_WARNING indicates that a fall back locale was used. For + * example, 'de_CH' was requested, but nothing was found there, so 'de' was + * used. U_USING_DEFAULT_WARNING indicates that the default locale data was + * used; neither the requested locale nor any of its fall back locales + * could be found. + * The caller owns the returned object and is responsible for deleting it. + * @stable ICU 2.0 + */ + static BreakIterator* U_EXPORT2 + createWordInstance(const Locale& where, UErrorCode& status); + + /** + * Create BreakIterator for line-breaks using specified locale. + * Returns an instance of a BreakIterator implementing line breaks. Line + * breaks are logically possible line breaks, actual line breaks are + * usually determined based on display width. + * LineBreak is useful for word wrapping text. + * @param where the locale. + * @param status The error code. + * @return A BreakIterator for line-breaks. The UErrorCode& status + * parameter is used to return status information to the user. + * To check whether the construction succeeded or not, you should check + * the value of U_SUCCESS(err). If you wish more detailed information, you + * can check for informational error results which still indicate success. + * U_USING_FALLBACK_WARNING indicates that a fall back locale was used. For + * example, 'de_CH' was requested, but nothing was found there, so 'de' was + * used. U_USING_DEFAULT_WARNING indicates that the default locale data was + * used; neither the requested locale nor any of its fall back locales + * could be found. + * The caller owns the returned object and is responsible for deleting it. + * @stable ICU 2.0 + */ + static BreakIterator* U_EXPORT2 + createLineInstance(const Locale& where, UErrorCode& status); + + /** + * Create BreakIterator for character-breaks using specified locale + * Returns an instance of a BreakIterator implementing character breaks. + * Character breaks are boundaries of combining character sequences. + * @param where the locale. + * @param status The error code. + * @return A BreakIterator for character-breaks. The UErrorCode& status + * parameter is used to return status information to the user. + * To check whether the construction succeeded or not, you should check + * the value of U_SUCCESS(err). If you wish more detailed information, you + * can check for informational error results which still indicate success. + * U_USING_FALLBACK_WARNING indicates that a fall back locale was used. For + * example, 'de_CH' was requested, but nothing was found there, so 'de' was + * used. U_USING_DEFAULT_WARNING indicates that the default locale data was + * used; neither the requested locale nor any of its fall back locales + * could be found. + * The caller owns the returned object and is responsible for deleting it. + * @stable ICU 2.0 + */ + static BreakIterator* U_EXPORT2 + createCharacterInstance(const Locale& where, UErrorCode& status); + + /** + * Create BreakIterator for sentence-breaks using specified locale + * Returns an instance of a BreakIterator implementing sentence breaks. + * @param where the locale. + * @param status The error code. + * @return A BreakIterator for sentence-breaks. The UErrorCode& status + * parameter is used to return status information to the user. + * To check whether the construction succeeded or not, you should check + * the value of U_SUCCESS(err). If you wish more detailed information, you + * can check for informational error results which still indicate success. + * U_USING_FALLBACK_WARNING indicates that a fall back locale was used. For + * example, 'de_CH' was requested, but nothing was found there, so 'de' was + * used. U_USING_DEFAULT_WARNING indicates that the default locale data was + * used; neither the requested locale nor any of its fall back locales + * could be found. + * The caller owns the returned object and is responsible for deleting it. + * @stable ICU 2.0 + */ + static BreakIterator* U_EXPORT2 + createSentenceInstance(const Locale& where, UErrorCode& status); + +#ifndef U_HIDE_DEPRECATED_API + /** + * Create BreakIterator for title-casing breaks using the specified locale + * Returns an instance of a BreakIterator implementing title breaks. + * The iterator returned locates title boundaries as described for + * Unicode 3.2 only. For Unicode 4.0 and above title boundary iteration, + * please use a word boundary iterator. See {@link #createWordInstance }. + * + * @param where the locale. + * @param status The error code. + * @return A BreakIterator for title-breaks. The UErrorCode& status + * parameter is used to return status information to the user. + * To check whether the construction succeeded or not, you should check + * the value of U_SUCCESS(err). If you wish more detailed information, you + * can check for informational error results which still indicate success. + * U_USING_FALLBACK_WARNING indicates that a fall back locale was used. For + * example, 'de_CH' was requested, but nothing was found there, so 'de' was + * used. U_USING_DEFAULT_WARNING indicates that the default locale data was + * used; neither the requested locale nor any of its fall back locales + * could be found. + * The caller owns the returned object and is responsible for deleting it. + * @deprecated ICU 64 Use createWordInstance instead. + */ + static BreakIterator* U_EXPORT2 + createTitleInstance(const Locale& where, UErrorCode& status); +#endif /* U_HIDE_DEPRECATED_API */ + + /** + * Get the set of Locales for which TextBoundaries are installed. + *

Note: this will not return locales added through the register + * call. To see the registered locales too, use the getAvailableLocales + * function that returns a StringEnumeration object

+ * @param count the output parameter of number of elements in the locale list + * @return available locales + * @stable ICU 2.0 + */ + static const Locale* U_EXPORT2 getAvailableLocales(int32_t& count); + + /** + * Get name of the object for the desired Locale, in the desired language. + * @param objectLocale must be from getAvailableLocales. + * @param displayLocale specifies the desired locale for output. + * @param name the fill-in parameter of the return value + * Uses best match. + * @return user-displayable name + * @stable ICU 2.0 + */ + static UnicodeString& U_EXPORT2 getDisplayName(const Locale& objectLocale, + const Locale& displayLocale, + UnicodeString& name); + + /** + * Get name of the object for the desired Locale, in the language of the + * default locale. + * @param objectLocale must be from getMatchingLocales + * @param name the fill-in parameter of the return value + * @return user-displayable name + * @stable ICU 2.0 + */ + static UnicodeString& U_EXPORT2 getDisplayName(const Locale& objectLocale, + UnicodeString& name); + +#ifndef U_FORCE_HIDE_DEPRECATED_API + /** + * Deprecated functionality. Use clone() instead. + * + * Thread safe client-buffer-based cloning operation + * Do NOT call delete on a safeclone, since 'new' is not used to create it. + * @param stackBuffer user allocated space for the new clone. If nullptr new memory will be allocated. + * If buffer is not large enough, new memory will be allocated. + * @param BufferSize reference to size of allocated space. + * If BufferSize == 0, a sufficient size for use in cloning will + * be returned ('pre-flighting') + * If BufferSize is not enough for a stack-based safe clone, + * new memory will be allocated. + * @param status to indicate whether the operation went on smoothly or there were errors + * An informational status value, U_SAFECLONE_ALLOCATED_ERROR, is used if any allocations were + * necessary. + * @return pointer to the new clone + * + * @deprecated ICU 52. Use clone() instead. + */ + virtual BreakIterator * createBufferClone(void *stackBuffer, + int32_t &BufferSize, + UErrorCode &status) = 0; +#endif // U_FORCE_HIDE_DEPRECATED_API + +#ifndef U_HIDE_DEPRECATED_API + + /** + * Determine whether the BreakIterator was created in user memory by + * createBufferClone(), and thus should not be deleted. Such objects + * must be closed by an explicit call to the destructor (not delete). + * @deprecated ICU 52. Always delete the BreakIterator. + */ + inline UBool isBufferClone(void); + +#endif /* U_HIDE_DEPRECATED_API */ + +#if !UCONFIG_NO_SERVICE + /** + * Register a new break iterator of the indicated kind, to use in the given locale. + * The break iterator will be adopted. Clones of the iterator will be returned + * if a request for a break iterator of the given kind matches or falls back to + * this locale. + * Because ICU may choose to cache BreakIterators internally, this must + * be called at application startup, prior to any calls to + * BreakIterator::createXXXInstance to avoid undefined behavior. + * @param toAdopt the BreakIterator instance to be adopted + * @param locale the Locale for which this instance is to be registered + * @param kind the type of iterator for which this instance is to be registered + * @param status the in/out status code, no special meanings are assigned + * @return a registry key that can be used to unregister this instance + * @stable ICU 2.4 + */ + static URegistryKey U_EXPORT2 registerInstance(BreakIterator* toAdopt, + const Locale& locale, + UBreakIteratorType kind, + UErrorCode& status); + + /** + * Unregister a previously-registered BreakIterator using the key returned from the + * register call. Key becomes invalid after a successful call and should not be used again. + * The BreakIterator corresponding to the key will be deleted. + * Because ICU may choose to cache BreakIterators internally, this should + * be called during application shutdown, after all calls to + * BreakIterator::createXXXInstance to avoid undefined behavior. + * @param key the registry key returned by a previous call to registerInstance + * @param status the in/out status code, no special meanings are assigned + * @return true if the iterator for the key was successfully unregistered + * @stable ICU 2.4 + */ + static UBool U_EXPORT2 unregister(URegistryKey key, UErrorCode& status); + + /** + * Return a StringEnumeration over the locales available at the time of the call, + * including registered locales. + * @return a StringEnumeration over the locales available at the time of the call + * @stable ICU 2.4 + */ + static StringEnumeration* U_EXPORT2 getAvailableLocales(void); +#endif + + /** + * Returns the locale for this break iterator. Two flavors are available: valid and + * actual locale. + * @stable ICU 2.8 + */ + Locale getLocale(ULocDataLocaleType type, UErrorCode& status) const; + +#ifndef U_HIDE_INTERNAL_API + /** Get the locale for this break iterator object. You can choose between valid and actual locale. + * @param type type of the locale we're looking for (valid or actual) + * @param status error code for the operation + * @return the locale + * @internal + */ + const char *getLocaleID(ULocDataLocaleType type, UErrorCode& status) const; +#endif /* U_HIDE_INTERNAL_API */ + + /** + * Set the subject text string upon which the break iterator is operating + * without changing any other aspect of the matching state. + * The new and previous text strings must have the same content. + * + * This function is intended for use in environments where ICU is operating on + * strings that may move around in memory. It provides a mechanism for notifying + * ICU that the string has been relocated, and providing a new UText to access the + * string in its new position. + * + * Note that the break iterator implementation never copies the underlying text + * of a string being processed, but always operates directly on the original text + * provided by the user. Refreshing simply drops the references to the old text + * and replaces them with references to the new. + * + * Caution: this function is normally used only by very specialized, + * system-level code. One example use case is with garbage collection that moves + * the text in memory. + * + * @param input The new (moved) text string. + * @param status Receives errors detected by this function. + * @return *this + * + * @stable ICU 49 + */ + virtual BreakIterator &refreshInputText(UText *input, UErrorCode &status) = 0; + + private: + static BreakIterator* buildInstance(const Locale& loc, const char *type, UErrorCode& status); + static BreakIterator* createInstance(const Locale& loc, int32_t kind, UErrorCode& status); + static BreakIterator* makeInstance(const Locale& loc, int32_t kind, UErrorCode& status); + + friend class ICUBreakIteratorFactory; + friend class ICUBreakIteratorService; + +protected: + // Do not enclose protected default/copy constructors with #ifndef U_HIDE_INTERNAL_API + // or else the compiler will create a public ones. + /** @internal */ + BreakIterator(); + /** @internal */ + BreakIterator (const BreakIterator &other); +#ifndef U_HIDE_INTERNAL_API + /** @internal */ + BreakIterator (const Locale& valid, const Locale &actual); + /** @internal. Assignment Operator, used by RuleBasedBreakIterator. */ + BreakIterator &operator = (const BreakIterator &other); +#endif /* U_HIDE_INTERNAL_API */ + +private: + + /** @internal (private) */ + char actualLocale[ULOC_FULLNAME_CAPACITY]; + char validLocale[ULOC_FULLNAME_CAPACITY]; +}; + +#ifndef U_HIDE_DEPRECATED_API + +inline UBool BreakIterator::isBufferClone() +{ + return false; +} + +#endif /* U_HIDE_DEPRECATED_API */ + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // BRKITER_H +//eof diff --git a/intl/icu/source/common/unicode/bytestream.h b/intl/icu/source/common/unicode/bytestream.h new file mode 100644 index 0000000000..997746e428 --- /dev/null +++ b/intl/icu/source/common/unicode/bytestream.h @@ -0,0 +1,307 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// Copyright (C) 2009-2012, International Business Machines +// Corporation and others. All Rights Reserved. +// +// Copyright 2007 Google Inc. All Rights Reserved. +// Author: sanjay@google.com (Sanjay Ghemawat) +// +// Abstract interface that consumes a sequence of bytes (ByteSink). +// +// Used so that we can write a single piece of code that can operate +// on a variety of output string types. +// +// Various implementations of this interface are provided: +// ByteSink: +// CheckedArrayByteSink Write to a flat array, with bounds checking +// StringByteSink Write to an STL string + +// This code is a contribution of Google code, and the style used here is +// a compromise between the original Google code and the ICU coding guidelines. +// For example, data types are ICU-ified (size_t,int->int32_t), +// and API comments doxygen-ified, but function names and behavior are +// as in the original, if possible. +// Assertion-style error handling, not available in ICU, was changed to +// parameter "pinning" similar to UnicodeString. +// +// In addition, this is only a partial port of the original Google code, +// limited to what was needed so far. The (nearly) complete original code +// is in the ICU svn repository at icuhtml/trunk/design/strings/contrib +// (see ICU ticket 6765, r25517). + +#ifndef __BYTESTREAM_H__ +#define __BYTESTREAM_H__ + +/** + * \file + * \brief C++ API: Interface for writing bytes, and implementation classes. + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/uobject.h" +#include "unicode/std_string.h" + +U_NAMESPACE_BEGIN + +/** + * A ByteSink can be filled with bytes. + * @stable ICU 4.2 + */ +class U_COMMON_API ByteSink : public UMemory { +public: + /** + * Default constructor. + * @stable ICU 4.2 + */ + ByteSink() { } + /** + * Virtual destructor. + * @stable ICU 4.2 + */ + virtual ~ByteSink(); + + /** + * Append "bytes[0,n-1]" to this. + * @param bytes the pointer to the bytes + * @param n the number of bytes; must be non-negative + * @stable ICU 4.2 + */ + virtual void Append(const char* bytes, int32_t n) = 0; + + /** + * Appends n bytes to this. Same as Append(). + * Call AppendU8() with u8"string literals" which are const char * in C++11 + * but const char8_t * in C++20. + * If the compiler does support char8_t as a distinct type, + * then an AppendU8() overload for that is defined and will be chosen. + * + * @param bytes the pointer to the bytes + * @param n the number of bytes; must be non-negative + * @stable ICU 67 + */ + inline void AppendU8(const char* bytes, int32_t n) { + Append(bytes, n); + } + +#if defined(__cpp_char8_t) || defined(U_IN_DOXYGEN) + /** + * Appends n bytes to this. Same as Append() but for a const char8_t * pointer. + * Call AppendU8() with u8"string literals" which are const char * in C++11 + * but const char8_t * in C++20. + * If the compiler does support char8_t as a distinct type, + * then this AppendU8() overload for that is defined and will be chosen. + * + * @param bytes the pointer to the bytes + * @param n the number of bytes; must be non-negative + * @stable ICU 67 + */ + inline void AppendU8(const char8_t* bytes, int32_t n) { + Append(reinterpret_cast(bytes), n); + } +#endif + + /** + * Returns a writable buffer for appending and writes the buffer's capacity to + * *result_capacity. Guarantees *result_capacity>=min_capacity. + * May return a pointer to the caller-owned scratch buffer which must have + * scratch_capacity>=min_capacity. + * The returned buffer is only valid until the next operation + * on this ByteSink. + * + * After writing at most *result_capacity bytes, call Append() with the + * pointer returned from this function and the number of bytes written. + * Many Append() implementations will avoid copying bytes if this function + * returned an internal buffer. + * + * Partial usage example: + * int32_t capacity; + * char* buffer = sink->GetAppendBuffer(..., &capacity); + * ... Write n bytes into buffer, with n <= capacity. + * sink->Append(buffer, n); + * In many implementations, that call to Append will avoid copying bytes. + * + * If the ByteSink allocates or reallocates an internal buffer, it should use + * the desired_capacity_hint if appropriate. + * If a caller cannot provide a reasonable guess at the desired capacity, + * it should pass desired_capacity_hint=0. + * + * If a non-scratch buffer is returned, the caller may only pass + * a prefix to it to Append(). + * That is, it is not correct to pass an interior pointer to Append(). + * + * The default implementation always returns the scratch buffer. + * + * @param min_capacity required minimum capacity of the returned buffer; + * must be non-negative + * @param desired_capacity_hint desired capacity of the returned buffer; + * must be non-negative + * @param scratch default caller-owned buffer + * @param scratch_capacity capacity of the scratch buffer + * @param result_capacity pointer to an integer which will be set to the + * capacity of the returned buffer + * @return a buffer with *result_capacity>=min_capacity + * @stable ICU 4.2 + */ + virtual char* GetAppendBuffer(int32_t min_capacity, + int32_t desired_capacity_hint, + char* scratch, int32_t scratch_capacity, + int32_t* result_capacity); + + /** + * Flush internal buffers. + * Some byte sinks use internal buffers or provide buffering + * and require calling Flush() at the end of the stream. + * The ByteSink should be ready for further Append() calls after Flush(). + * The default implementation of Flush() does nothing. + * @stable ICU 4.2 + */ + virtual void Flush(); + +private: + ByteSink(const ByteSink &) = delete; + ByteSink &operator=(const ByteSink &) = delete; +}; + +// ------------------------------------------------------------- +// Some standard implementations + +/** + * Implementation of ByteSink that writes to a flat byte array, + * with bounds-checking: + * This sink will not write more than capacity bytes to outbuf. + * If more than capacity bytes are Append()ed, then excess bytes are ignored, + * and Overflowed() will return true. + * Overflow does not cause a runtime error. + * @stable ICU 4.2 + */ +class U_COMMON_API CheckedArrayByteSink : public ByteSink { +public: + /** + * Constructs a ByteSink that will write to outbuf[0..capacity-1]. + * @param outbuf buffer to write to + * @param capacity size of the buffer + * @stable ICU 4.2 + */ + CheckedArrayByteSink(char* outbuf, int32_t capacity); + /** + * Destructor. + * @stable ICU 4.2 + */ + virtual ~CheckedArrayByteSink(); + /** + * Returns the sink to its original state, without modifying the buffer. + * Useful for reusing both the buffer and the sink for multiple streams. + * Resets the state to NumberOfBytesWritten()=NumberOfBytesAppended()=0 + * and Overflowed()=false. + * @return *this + * @stable ICU 4.6 + */ + virtual CheckedArrayByteSink& Reset(); + /** + * Append "bytes[0,n-1]" to this. + * @param bytes the pointer to the bytes + * @param n the number of bytes; must be non-negative + * @stable ICU 4.2 + */ + virtual void Append(const char* bytes, int32_t n) override; + /** + * Returns a writable buffer for appending and writes the buffer's capacity to + * *result_capacity. For details see the base class documentation. + * @param min_capacity required minimum capacity of the returned buffer; + * must be non-negative + * @param desired_capacity_hint desired capacity of the returned buffer; + * must be non-negative + * @param scratch default caller-owned buffer + * @param scratch_capacity capacity of the scratch buffer + * @param result_capacity pointer to an integer which will be set to the + * capacity of the returned buffer + * @return a buffer with *result_capacity>=min_capacity + * @stable ICU 4.2 + */ + virtual char* GetAppendBuffer(int32_t min_capacity, + int32_t desired_capacity_hint, + char* scratch, int32_t scratch_capacity, + int32_t* result_capacity) override; + /** + * Returns the number of bytes actually written to the sink. + * @return number of bytes written to the buffer + * @stable ICU 4.2 + */ + int32_t NumberOfBytesWritten() const { return size_; } + /** + * Returns true if any bytes were discarded, i.e., if there was an + * attempt to write more than 'capacity' bytes. + * @return true if more than 'capacity' bytes were Append()ed + * @stable ICU 4.2 + */ + UBool Overflowed() const { return overflowed_; } + /** + * Returns the number of bytes appended to the sink. + * If Overflowed() then NumberOfBytesAppended()>NumberOfBytesWritten() + * else they return the same number. + * @return number of bytes written to the buffer + * @stable ICU 4.6 + */ + int32_t NumberOfBytesAppended() const { return appended_; } +private: + char* outbuf_; + const int32_t capacity_; + int32_t size_; + int32_t appended_; + UBool overflowed_; + + CheckedArrayByteSink() = delete; + CheckedArrayByteSink(const CheckedArrayByteSink &) = delete; + CheckedArrayByteSink &operator=(const CheckedArrayByteSink &) = delete; +}; + +/** + * Implementation of ByteSink that writes to a "string". + * The StringClass is usually instantiated with a std::string. + * @stable ICU 4.2 + */ +template +class StringByteSink : public ByteSink { + public: + /** + * Constructs a ByteSink that will append bytes to the dest string. + * @param dest pointer to string object to append to + * @stable ICU 4.2 + */ + StringByteSink(StringClass* dest) : dest_(dest) { } + /** + * Constructs a ByteSink that reserves append capacity and will append bytes to the dest string. + * + * @param dest pointer to string object to append to + * @param initialAppendCapacity capacity beyond dest->length() to be reserve()d + * @stable ICU 60 + */ + StringByteSink(StringClass* dest, int32_t initialAppendCapacity) : dest_(dest) { + if (initialAppendCapacity > 0 && + (uint32_t)initialAppendCapacity > (dest->capacity() - dest->length())) { + dest->reserve(dest->length() + initialAppendCapacity); + } + } + /** + * Append "bytes[0,n-1]" to this. + * @param data the pointer to the bytes + * @param n the number of bytes; must be non-negative + * @stable ICU 4.2 + */ + virtual void Append(const char* data, int32_t n) override { dest_->append(data, n); } + private: + StringClass* dest_; + + StringByteSink() = delete; + StringByteSink(const StringByteSink &) = delete; + StringByteSink &operator=(const StringByteSink &) = delete; +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __BYTESTREAM_H__ diff --git a/intl/icu/source/common/unicode/bytestrie.h b/intl/icu/source/common/unicode/bytestrie.h new file mode 100644 index 0000000000..1719a6bb83 --- /dev/null +++ b/intl/icu/source/common/unicode/bytestrie.h @@ -0,0 +1,568 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2012, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: bytestrie.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010sep25 +* created by: Markus W. Scherer +*/ + +#ifndef __BYTESTRIE_H__ +#define __BYTESTRIE_H__ + +/** + * \file + * \brief C++ API: Trie for mapping byte sequences to integer values. + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/stringpiece.h" +#include "unicode/uobject.h" +#include "unicode/ustringtrie.h" + +class BytesTrieTest; + +U_NAMESPACE_BEGIN + +class ByteSink; +class BytesTrieBuilder; +class CharString; +class UVector32; + +/** + * Light-weight, non-const reader class for a BytesTrie. + * Traverses a byte-serialized data structure with minimal state, + * for mapping byte sequences to non-negative integer values. + * + * This class owns the serialized trie data only if it was constructed by + * the builder's build() method. + * The public constructor and the copy constructor only alias the data (only copy the pointer). + * There is no assignment operator. + * + * This class is not intended for public subclassing. + * @stable ICU 4.8 + */ +class U_COMMON_API BytesTrie : public UMemory { +public: + /** + * Constructs a BytesTrie reader instance. + * + * The trieBytes must contain a copy of a byte sequence from the BytesTrieBuilder, + * starting with the first byte of that sequence. + * The BytesTrie object will not read more bytes than + * the BytesTrieBuilder generated in the corresponding build() call. + * + * The array is not copied/cloned and must not be modified while + * the BytesTrie object is in use. + * + * @param trieBytes The byte array that contains the serialized trie. + * @stable ICU 4.8 + */ + BytesTrie(const void *trieBytes) + : ownedArray_(nullptr), bytes_(static_cast(trieBytes)), + pos_(bytes_), remainingMatchLength_(-1) {} + + /** + * Destructor. + * @stable ICU 4.8 + */ + ~BytesTrie(); + + /** + * Copy constructor, copies the other trie reader object and its state, + * but not the byte array which will be shared. (Shallow copy.) + * @param other Another BytesTrie object. + * @stable ICU 4.8 + */ + BytesTrie(const BytesTrie &other) + : ownedArray_(nullptr), bytes_(other.bytes_), + pos_(other.pos_), remainingMatchLength_(other.remainingMatchLength_) {} + + /** + * Resets this trie to its initial state. + * @return *this + * @stable ICU 4.8 + */ + BytesTrie &reset() { + pos_=bytes_; + remainingMatchLength_=-1; + return *this; + } + + /** + * Returns the state of this trie as a 64-bit integer. + * The state value is never 0. + * + * @return opaque state value + * @see resetToState64 + * @stable ICU 65 + */ + uint64_t getState64() const { + return (static_cast(remainingMatchLength_ + 2) << kState64RemainingShift) | + (uint64_t)(pos_ - bytes_); + } + + /** + * Resets this trie to the saved state. + * Unlike resetToState(State), the 64-bit state value + * must be from getState64() from the same trie object or + * from one initialized the exact same way. + * Because of no validation, this method is faster. + * + * @param state The opaque trie state value from getState64(). + * @return *this + * @see getState64 + * @see resetToState + * @see reset + * @stable ICU 65 + */ + BytesTrie &resetToState64(uint64_t state) { + remainingMatchLength_ = static_cast(state >> kState64RemainingShift) - 2; + pos_ = bytes_ + (state & kState64PosMask); + return *this; + } + + /** + * BytesTrie state object, for saving a trie's current state + * and resetting the trie back to this state later. + * @stable ICU 4.8 + */ + class State : public UMemory { + public: + /** + * Constructs an empty State. + * @stable ICU 4.8 + */ + State() { bytes=nullptr; } + private: + friend class BytesTrie; + + const uint8_t *bytes; + const uint8_t *pos; + int32_t remainingMatchLength; + }; + + /** + * Saves the state of this trie. + * @param state The State object to hold the trie's state. + * @return *this + * @see resetToState + * @stable ICU 4.8 + */ + const BytesTrie &saveState(State &state) const { + state.bytes=bytes_; + state.pos=pos_; + state.remainingMatchLength=remainingMatchLength_; + return *this; + } + + /** + * Resets this trie to the saved state. + * If the state object contains no state, or the state of a different trie, + * then this trie remains unchanged. + * @param state The State object which holds a saved trie state. + * @return *this + * @see saveState + * @see reset + * @stable ICU 4.8 + */ + BytesTrie &resetToState(const State &state) { + if(bytes_==state.bytes && bytes_!=nullptr) { + pos_=state.pos; + remainingMatchLength_=state.remainingMatchLength; + } + return *this; + } + + /** + * Determines whether the byte sequence so far matches, whether it has a value, + * and whether another input byte can continue a matching byte sequence. + * @return The match/value Result. + * @stable ICU 4.8 + */ + UStringTrieResult current() const; + + /** + * Traverses the trie from the initial state for this input byte. + * Equivalent to reset().next(inByte). + * @param inByte Input byte value. Values -0x100..-1 are treated like 0..0xff. + * Values below -0x100 and above 0xff will never match. + * @return The match/value Result. + * @stable ICU 4.8 + */ + inline UStringTrieResult first(int32_t inByte) { + remainingMatchLength_=-1; + if(inByte<0) { + inByte+=0x100; + } + return nextImpl(bytes_, inByte); + } + + /** + * Traverses the trie from the current state for this input byte. + * @param inByte Input byte value. Values -0x100..-1 are treated like 0..0xff. + * Values below -0x100 and above 0xff will never match. + * @return The match/value Result. + * @stable ICU 4.8 + */ + UStringTrieResult next(int32_t inByte); + + /** + * Traverses the trie from the current state for this byte sequence. + * Equivalent to + * \code + * Result result=current(); + * for(each c in s) + * if(!USTRINGTRIE_HAS_NEXT(result)) return USTRINGTRIE_NO_MATCH; + * result=next(c); + * return result; + * \endcode + * @param s A string or byte sequence. Can be nullptr if length is 0. + * @param length The length of the byte sequence. Can be -1 if NUL-terminated. + * @return The match/value Result. + * @stable ICU 4.8 + */ + UStringTrieResult next(const char *s, int32_t length); + + /** + * Returns a matching byte sequence's value if called immediately after + * current()/first()/next() returned USTRINGTRIE_INTERMEDIATE_VALUE or USTRINGTRIE_FINAL_VALUE. + * getValue() can be called multiple times. + * + * Do not call getValue() after USTRINGTRIE_NO_MATCH or USTRINGTRIE_NO_VALUE! + * @return The value for the byte sequence so far. + * @stable ICU 4.8 + */ + inline int32_t getValue() const { + const uint8_t *pos=pos_; + int32_t leadByte=*pos++; + // U_ASSERT(leadByte>=kMinValueLead); + return readValue(pos, leadByte>>1); + } + + /** + * Determines whether all byte sequences reachable from the current state + * map to the same value. + * @param uniqueValue Receives the unique value, if this function returns true. + * (output-only) + * @return true if all byte sequences reachable from the current state + * map to the same value. + * @stable ICU 4.8 + */ + inline UBool hasUniqueValue(int32_t &uniqueValue) const { + const uint8_t *pos=pos_; + // Skip the rest of a pending linear-match node. + return pos!=nullptr && findUniqueValue(pos+remainingMatchLength_+1, false, uniqueValue); + } + + /** + * Finds each byte which continues the byte sequence from the current state. + * That is, each byte b for which it would be next(b)!=USTRINGTRIE_NO_MATCH now. + * @param out Each next byte is appended to this object. + * (Only uses the out.Append(s, length) method.) + * @return the number of bytes which continue the byte sequence from here + * @stable ICU 4.8 + */ + int32_t getNextBytes(ByteSink &out) const; + + /** + * Iterator for all of the (byte sequence, value) pairs in a BytesTrie. + * @stable ICU 4.8 + */ + class U_COMMON_API Iterator : public UMemory { + public: + /** + * Iterates from the root of a byte-serialized BytesTrie. + * @param trieBytes The trie bytes. + * @param maxStringLength If 0, the iterator returns full strings/byte sequences. + * Otherwise, the iterator returns strings with this maximum length. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @stable ICU 4.8 + */ + Iterator(const void *trieBytes, int32_t maxStringLength, UErrorCode &errorCode); + + /** + * Iterates from the current state of the specified BytesTrie. + * @param trie The trie whose state will be copied for iteration. + * @param maxStringLength If 0, the iterator returns full strings/byte sequences. + * Otherwise, the iterator returns strings with this maximum length. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @stable ICU 4.8 + */ + Iterator(const BytesTrie &trie, int32_t maxStringLength, UErrorCode &errorCode); + + /** + * Destructor. + * @stable ICU 4.8 + */ + ~Iterator(); + + /** + * Resets this iterator to its initial state. + * @return *this + * @stable ICU 4.8 + */ + Iterator &reset(); + + /** + * @return true if there are more elements. + * @stable ICU 4.8 + */ + UBool hasNext() const; + + /** + * Finds the next (byte sequence, value) pair if there is one. + * + * If the byte sequence is truncated to the maximum length and does not + * have a real value, then the value is set to -1. + * In this case, this "not a real value" is indistinguishable from + * a real value of -1. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return true if there is another element. + * @stable ICU 4.8 + */ + UBool next(UErrorCode &errorCode); + + /** + * @return The NUL-terminated byte sequence for the last successful next(). + * @stable ICU 4.8 + */ + StringPiece getString() const; + /** + * @return The value for the last successful next(). + * @stable ICU 4.8 + */ + int32_t getValue() const { return value_; } + + private: + UBool truncateAndStop(); + + const uint8_t *branchNext(const uint8_t *pos, int32_t length, UErrorCode &errorCode); + + const uint8_t *bytes_; + const uint8_t *pos_; + const uint8_t *initialPos_; + int32_t remainingMatchLength_; + int32_t initialRemainingMatchLength_; + + CharString *str_; + int32_t maxLength_; + int32_t value_; + + // The stack stores pairs of integers for backtracking to another + // outbound edge of a branch node. + // The first integer is an offset from bytes_. + // The second integer has the str_->length() from before the node in bits 15..0, + // and the remaining branch length in bits 24..16. (Bits 31..25 are unused.) + // (We could store the remaining branch length minus 1 in bits 23..16 and not use bits 31..24, + // but the code looks more confusing that way.) + UVector32 *stack_; + }; + +private: + friend class BytesTrieBuilder; + friend class ::BytesTrieTest; + + /** + * Constructs a BytesTrie reader instance. + * Unlike the public constructor which just aliases an array, + * this constructor adopts the builder's array. + * This constructor is only called by the builder. + */ + BytesTrie(void *adoptBytes, const void *trieBytes) + : ownedArray_(static_cast(adoptBytes)), + bytes_(static_cast(trieBytes)), + pos_(bytes_), remainingMatchLength_(-1) {} + + // No assignment operator. + BytesTrie &operator=(const BytesTrie &other) = delete; + + inline void stop() { + pos_=nullptr; + } + + // Reads a compact 32-bit integer. + // pos is already after the leadByte, and the lead byte is already shifted right by 1. + static int32_t readValue(const uint8_t *pos, int32_t leadByte); + static inline const uint8_t *skipValue(const uint8_t *pos, int32_t leadByte) { + // U_ASSERT(leadByte>=kMinValueLead); + if(leadByte>=(kMinTwoByteValueLead<<1)) { + if(leadByte<(kMinThreeByteValueLead<<1)) { + ++pos; + } else if(leadByte<(kFourByteValueLead<<1)) { + pos+=2; + } else { + pos+=3+((leadByte>>1)&1); + } + } + return pos; + } + static inline const uint8_t *skipValue(const uint8_t *pos) { + int32_t leadByte=*pos++; + return skipValue(pos, leadByte); + } + + // Reads a jump delta and jumps. + static const uint8_t *jumpByDelta(const uint8_t *pos); + + static inline const uint8_t *skipDelta(const uint8_t *pos) { + int32_t delta=*pos++; + if(delta>=kMinTwoByteDeltaLead) { + if(delta>8)+1; // 0x6c + static const int32_t kFourByteValueLead=0x7e; + + // A little more than Unicode code points. (0x11ffff) + static const int32_t kMaxThreeByteValue=((kFourByteValueLead-kMinThreeByteValueLead)<<16)-1; + + static const int32_t kFiveByteValueLead=0x7f; + + // Compact delta integers. + static const int32_t kMaxOneByteDelta=0xbf; + static const int32_t kMinTwoByteDeltaLead=kMaxOneByteDelta+1; // 0xc0 + static const int32_t kMinThreeByteDeltaLead=0xf0; + static const int32_t kFourByteDeltaLead=0xfe; + static const int32_t kFiveByteDeltaLead=0xff; + + static const int32_t kMaxTwoByteDelta=((kMinThreeByteDeltaLead-kMinTwoByteDeltaLead)<<8)-1; // 0x2fff + static const int32_t kMaxThreeByteDelta=((kFourByteDeltaLead-kMinThreeByteDeltaLead)<<16)-1; // 0xdffff + + // For getState64(): + // The remainingMatchLength_ is -1..14=(kMaxLinearMatchLength=0x10)-2 + // so we need at least 5 bits for that. + // We add 2 to store it as a positive value 1..16=kMaxLinearMatchLength. + static constexpr int32_t kState64RemainingShift = 59; + static constexpr uint64_t kState64PosMask = (UINT64_C(1) << kState64RemainingShift) - 1; + + uint8_t *ownedArray_; + + // Fixed value referencing the BytesTrie bytes. + const uint8_t *bytes_; + + // Iterator variables. + + // Pointer to next trie byte to read. nullptr if no more matches. + const uint8_t *pos_; + // Remaining length of a linear-match node, minus 1. Negative if not in such a node. + int32_t remainingMatchLength_; +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __BYTESTRIE_H__ diff --git a/intl/icu/source/common/unicode/bytestriebuilder.h b/intl/icu/source/common/unicode/bytestriebuilder.h new file mode 100644 index 0000000000..ec9c625473 --- /dev/null +++ b/intl/icu/source/common/unicode/bytestriebuilder.h @@ -0,0 +1,193 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2016, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: bytestriebuilder.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010sep25 +* created by: Markus W. Scherer +*/ + +/** + * \file + * \brief C++ API: Builder for icu::BytesTrie + */ + +#ifndef __BYTESTRIEBUILDER_H__ +#define __BYTESTRIEBUILDER_H__ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/bytestrie.h" +#include "unicode/stringpiece.h" +#include "unicode/stringtriebuilder.h" + +class BytesTrieTest; + +U_NAMESPACE_BEGIN + +class BytesTrieElement; +class CharString; +/** + * Builder class for BytesTrie. + * + * This class is not intended for public subclassing. + * @stable ICU 4.8 + */ +class U_COMMON_API BytesTrieBuilder : public StringTrieBuilder { +public: + /** + * Constructs an empty builder. + * @param errorCode Standard ICU error code. + * @stable ICU 4.8 + */ + BytesTrieBuilder(UErrorCode &errorCode); + + /** + * Destructor. + * @stable ICU 4.8 + */ + virtual ~BytesTrieBuilder(); + + /** + * Adds a (byte sequence, value) pair. + * The byte sequence must be unique. + * The bytes will be copied; the builder does not keep + * a reference to the input StringPiece or its data(). + * @param s The input byte sequence. + * @param value The value associated with this byte sequence. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return *this + * @stable ICU 4.8 + */ + BytesTrieBuilder &add(StringPiece s, int32_t value, UErrorCode &errorCode); + + /** + * Builds a BytesTrie for the add()ed data. + * Once built, no further data can be add()ed until clear() is called. + * + * A BytesTrie cannot be empty. At least one (byte sequence, value) pair + * must have been add()ed. + * + * This method passes ownership of the builder's internal result array to the new trie object. + * Another call to any build() variant will re-serialize the trie. + * After clear() has been called, a new array will be used as well. + * @param buildOption Build option, see UStringTrieBuildOption. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return A new BytesTrie for the add()ed data. + * @stable ICU 4.8 + */ + BytesTrie *build(UStringTrieBuildOption buildOption, UErrorCode &errorCode); + + /** + * Builds a BytesTrie for the add()ed data and byte-serializes it. + * Once built, no further data can be add()ed until clear() is called. + * + * A BytesTrie cannot be empty. At least one (byte sequence, value) pair + * must have been add()ed. + * + * Multiple calls to buildStringPiece() return StringPieces referring to the + * builder's same byte array, without rebuilding. + * If buildStringPiece() is called after build(), the trie will be + * re-serialized into a new array (because build() passes on ownership). + * If build() is called after buildStringPiece(), the trie object returned + * by build() will become the owner of the underlying string for the + * previously returned StringPiece. + * After clear() has been called, a new array will be used as well. + * @param buildOption Build option, see UStringTrieBuildOption. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return A StringPiece which refers to the byte-serialized BytesTrie for the add()ed data. + * @stable ICU 4.8 + */ + StringPiece buildStringPiece(UStringTrieBuildOption buildOption, UErrorCode &errorCode); + + /** + * Removes all (byte sequence, value) pairs. + * New data can then be add()ed and a new trie can be built. + * @return *this + * @stable ICU 4.8 + */ + BytesTrieBuilder &clear(); + +private: + friend class ::BytesTrieTest; + + BytesTrieBuilder(const BytesTrieBuilder &other) = delete; // no copy constructor + BytesTrieBuilder &operator=(const BytesTrieBuilder &other) = delete; // no assignment operator + + void buildBytes(UStringTrieBuildOption buildOption, UErrorCode &errorCode); + + virtual int32_t getElementStringLength(int32_t i) const override; + virtual char16_t getElementUnit(int32_t i, int32_t byteIndex) const override; + virtual int32_t getElementValue(int32_t i) const override; + + virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t byteIndex) const override; + + virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t byteIndex) const override; + virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t byteIndex, int32_t count) const override; + virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t byteIndex, char16_t byte) const override; + + virtual UBool matchNodesCanHaveValues() const override { return false; } + + virtual int32_t getMaxBranchLinearSubNodeLength() const override { return BytesTrie::kMaxBranchLinearSubNodeLength; } + virtual int32_t getMinLinearMatch() const override { return BytesTrie::kMinLinearMatch; } + virtual int32_t getMaxLinearMatchLength() const override { return BytesTrie::kMaxLinearMatchLength; } + + /** + * @internal (private) + */ + class BTLinearMatchNode : public LinearMatchNode { + public: + BTLinearMatchNode(const char *units, int32_t len, Node *nextNode); + virtual bool operator==(const Node &other) const override; + virtual void write(StringTrieBuilder &builder) override; + private: + const char *s; + }; + + virtual Node *createLinearMatchNode(int32_t i, int32_t byteIndex, int32_t length, + Node *nextNode) const override; + + UBool ensureCapacity(int32_t length); + virtual int32_t write(int32_t byte) override; + int32_t write(const char *b, int32_t length); + virtual int32_t writeElementUnits(int32_t i, int32_t byteIndex, int32_t length) override; + virtual int32_t writeValueAndFinal(int32_t i, UBool isFinal) override; + virtual int32_t writeValueAndType(UBool hasValue, int32_t value, int32_t node) override; + virtual int32_t writeDeltaTo(int32_t jumpTarget) override; + static int32_t internalEncodeDelta(int32_t i, char intBytes[]); + + CharString *strings; // Pointer not object so we need not #include internal charstr.h. + BytesTrieElement *elements; + int32_t elementsCapacity; + int32_t elementsLength; + + // Byte serialization of the trie. + // Grows from the back: bytesLength measures from the end of the buffer! + char *bytes; + int32_t bytesCapacity; + int32_t bytesLength; +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __BYTESTRIEBUILDER_H__ diff --git a/intl/icu/source/common/unicode/caniter.h b/intl/icu/source/common/unicode/caniter.h new file mode 100644 index 0000000000..035bd0e64e --- /dev/null +++ b/intl/icu/source/common/unicode/caniter.h @@ -0,0 +1,214 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ******************************************************************************* + * Copyright (C) 1996-2014, International Business Machines Corporation and + * others. All Rights Reserved. + ******************************************************************************* + */ + +#ifndef CANITER_H +#define CANITER_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#if !UCONFIG_NO_NORMALIZATION + +#include "unicode/uobject.h" +#include "unicode/unistr.h" + +/** + * \file + * \brief C++ API: Canonical Iterator + */ + +/** Should permutation skip characters with combining class zero + * Should be either true or false. This is a compile time option + * @stable ICU 2.4 + */ +#ifndef CANITER_SKIP_ZEROES +#define CANITER_SKIP_ZEROES true +#endif + +U_NAMESPACE_BEGIN + +class Hashtable; +class Normalizer2; +class Normalizer2Impl; + +/** + * This class allows one to iterate through all the strings that are canonically equivalent to a given + * string. For example, here are some sample results: +Results for: {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D}{COMBINING DOT ABOVE}{COMBINING CEDILLA} +1: \\u0041\\u030A\\u0064\\u0307\\u0327 + = {LATIN CAPITAL LETTER A}{COMBINING RING ABOVE}{LATIN SMALL LETTER D}{COMBINING DOT ABOVE}{COMBINING CEDILLA} +2: \\u0041\\u030A\\u0064\\u0327\\u0307 + = {LATIN CAPITAL LETTER A}{COMBINING RING ABOVE}{LATIN SMALL LETTER D}{COMBINING CEDILLA}{COMBINING DOT ABOVE} +3: \\u0041\\u030A\\u1E0B\\u0327 + = {LATIN CAPITAL LETTER A}{COMBINING RING ABOVE}{LATIN SMALL LETTER D WITH DOT ABOVE}{COMBINING CEDILLA} +4: \\u0041\\u030A\\u1E11\\u0307 + = {LATIN CAPITAL LETTER A}{COMBINING RING ABOVE}{LATIN SMALL LETTER D WITH CEDILLA}{COMBINING DOT ABOVE} +5: \\u00C5\\u0064\\u0307\\u0327 + = {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D}{COMBINING DOT ABOVE}{COMBINING CEDILLA} +6: \\u00C5\\u0064\\u0327\\u0307 + = {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D}{COMBINING CEDILLA}{COMBINING DOT ABOVE} +7: \\u00C5\\u1E0B\\u0327 + = {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D WITH DOT ABOVE}{COMBINING CEDILLA} +8: \\u00C5\\u1E11\\u0307 + = {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D WITH CEDILLA}{COMBINING DOT ABOVE} +9: \\u212B\\u0064\\u0307\\u0327 + = {ANGSTROM SIGN}{LATIN SMALL LETTER D}{COMBINING DOT ABOVE}{COMBINING CEDILLA} +10: \\u212B\\u0064\\u0327\\u0307 + = {ANGSTROM SIGN}{LATIN SMALL LETTER D}{COMBINING CEDILLA}{COMBINING DOT ABOVE} +11: \\u212B\\u1E0B\\u0327 + = {ANGSTROM SIGN}{LATIN SMALL LETTER D WITH DOT ABOVE}{COMBINING CEDILLA} +12: \\u212B\\u1E11\\u0307 + = {ANGSTROM SIGN}{LATIN SMALL LETTER D WITH CEDILLA}{COMBINING DOT ABOVE} + *
Note: the code is intended for use with small strings, and is not suitable for larger ones, + * since it has not been optimized for that situation. + * Note, CanonicalIterator is not intended to be subclassed. + * @author M. Davis + * @author C++ port by V. Weinstein + * @stable ICU 2.4 + */ +class U_COMMON_API CanonicalIterator final : public UObject { +public: + /** + * Construct a CanonicalIterator object + * @param source string to get results for + * @param status Fill-in parameter which receives the status of this operation. + * @stable ICU 2.4 + */ + CanonicalIterator(const UnicodeString &source, UErrorCode &status); + + /** Destructor + * Cleans pieces + * @stable ICU 2.4 + */ + virtual ~CanonicalIterator(); + + /** + * Gets the NFD form of the current source we are iterating over. + * @return gets the source: NOTE: it is the NFD form of source + * @stable ICU 2.4 + */ + UnicodeString getSource(); + + /** + * Resets the iterator so that one can start again from the beginning. + * @stable ICU 2.4 + */ + void reset(); + + /** + * Get the next canonically equivalent string. + *
Warning: The strings are not guaranteed to be in any particular order. + * @return the next string that is canonically equivalent. A bogus string is returned when + * the iteration is done. + * @stable ICU 2.4 + */ + UnicodeString next(); + + /** + * Set a new source for this iterator. Allows object reuse. + * @param newSource the source string to iterate against. This allows the same iterator to be used + * while changing the source string, saving object creation. + * @param status Fill-in parameter which receives the status of this operation. + * @stable ICU 2.4 + */ + void setSource(const UnicodeString &newSource, UErrorCode &status); + +#ifndef U_HIDE_INTERNAL_API + /** + * Dumb recursive implementation of permutation. + * TODO: optimize + * @param source the string to find permutations for + * @param skipZeros determine if skip zeros + * @param result the results in a set. + * @param status Fill-in parameter which receives the status of this operation. + * @internal + */ + static void U_EXPORT2 permute(UnicodeString &source, UBool skipZeros, Hashtable *result, UErrorCode &status); +#endif /* U_HIDE_INTERNAL_API */ + + /** + * ICU "poor man's RTTI", returns a UClassID for this class. + * + * @stable ICU 2.2 + */ + static UClassID U_EXPORT2 getStaticClassID(); + + /** + * ICU "poor man's RTTI", returns a UClassID for the actual class. + * + * @stable ICU 2.2 + */ + virtual UClassID getDynamicClassID() const override; + +private: + // ===================== PRIVATES ============================== + // private default constructor + CanonicalIterator() = delete; + + + /** + * Copy constructor. Private for now. + * @internal (private) + */ + CanonicalIterator(const CanonicalIterator& other) = delete; + + /** + * Assignment operator. Private for now. + * @internal (private) + */ + CanonicalIterator& operator=(const CanonicalIterator& other) = delete; + + // fields + UnicodeString source; + UBool done; + + // 2 dimensional array holds the pieces of the string with + // their different canonically equivalent representations + UnicodeString **pieces; + int32_t pieces_length; + int32_t *pieces_lengths; + + // current is used in iterating to combine pieces + int32_t *current; + int32_t current_length; + + // transient fields + UnicodeString buffer; + + const Normalizer2 &nfd; + const Normalizer2Impl &nfcImpl; + + // we have a segment, in NFD. Find all the strings that are canonically equivalent to it. + UnicodeString *getEquivalents(const UnicodeString &segment, int32_t &result_len, UErrorCode &status); //private String[] getEquivalents(String segment) + + //Set getEquivalents2(String segment); + Hashtable *getEquivalents2(Hashtable *fillinResult, const char16_t *segment, int32_t segLen, UErrorCode &status); + //Hashtable *getEquivalents2(const UnicodeString &segment, int32_t segLen, UErrorCode &status); + + /** + * See if the decomposition of cp2 is at segment starting at segmentPos + * (with canonical rearrangement!) + * If so, take the remainder, and return the equivalents + */ + //Set extract(int comp, String segment, int segmentPos, StringBuffer buffer); + Hashtable *extract(Hashtable *fillinResult, UChar32 comp, const char16_t *segment, int32_t segLen, int32_t segmentPos, UErrorCode &status); + //Hashtable *extract(UChar32 comp, const UnicodeString &segment, int32_t segLen, int32_t segmentPos, UErrorCode &status); + + void cleanPieces(); + +}; + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_NORMALIZATION */ + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/casemap.h b/intl/icu/source/common/unicode/casemap.h new file mode 100644 index 0000000000..eca7cbf80a --- /dev/null +++ b/intl/icu/source/common/unicode/casemap.h @@ -0,0 +1,497 @@ +// © 2017 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// casemap.h +// created: 2017jan12 Markus W. Scherer + +#ifndef __CASEMAP_H__ +#define __CASEMAP_H__ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/stringpiece.h" +#include "unicode/uobject.h" + +/** + * \file + * \brief C++ API: Low-level C++ case mapping functions. + */ + +U_NAMESPACE_BEGIN + +class BreakIterator; +class ByteSink; +class Edits; + +/** + * Low-level C++ case mapping functions. + * + * @stable ICU 59 + */ +class U_COMMON_API CaseMap final : public UMemory { +public: + /** + * Lowercases a UTF-16 string and optionally records edits. + * Casing is locale-dependent and context-sensitive. + * The result may be longer or shorter than the original. + * The source string and the destination buffer must not overlap. + * + * @param locale The locale ID. ("" = root locale, nullptr = default locale.) + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT and U_EDITS_NO_RESET. + * @param src The original string. + * @param srcLength The length of the original string. If -1, then src must be NUL-terminated. + * @param dest A buffer for the result string. The result will be NUL-terminated if + * the buffer is large enough. + * The contents is undefined in case of failure. + * @param destCapacity The size of the buffer (number of char16_ts). If it is 0, then + * dest may be nullptr and the function will only return the length of the result + * without writing any of the result string. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Reference to an in/out error code value + * which must not indicate a failure before the function call. + * @return The length of the result string, if successful. + * When the result would be longer than destCapacity, + * the full length is returned and a U_BUFFER_OVERFLOW_ERROR is set. + * + * @see u_strToLower + * @stable ICU 59 + */ + static int32_t toLower( + const char *locale, uint32_t options, + const char16_t *src, int32_t srcLength, + char16_t *dest, int32_t destCapacity, Edits *edits, + UErrorCode &errorCode); + + /** + * Uppercases a UTF-16 string and optionally records edits. + * Casing is locale-dependent and context-sensitive. + * The result may be longer or shorter than the original. + * The source string and the destination buffer must not overlap. + * + * @param locale The locale ID. ("" = root locale, nullptr = default locale.) + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT and U_EDITS_NO_RESET. + * @param src The original string. + * @param srcLength The length of the original string. If -1, then src must be NUL-terminated. + * @param dest A buffer for the result string. The result will be NUL-terminated if + * the buffer is large enough. + * The contents is undefined in case of failure. + * @param destCapacity The size of the buffer (number of char16_ts). If it is 0, then + * dest may be nullptr and the function will only return the length of the result + * without writing any of the result string. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Reference to an in/out error code value + * which must not indicate a failure before the function call. + * @return The length of the result string, if successful. + * When the result would be longer than destCapacity, + * the full length is returned and a U_BUFFER_OVERFLOW_ERROR is set. + * + * @see u_strToUpper + * @stable ICU 59 + */ + static int32_t toUpper( + const char *locale, uint32_t options, + const char16_t *src, int32_t srcLength, + char16_t *dest, int32_t destCapacity, Edits *edits, + UErrorCode &errorCode); + +#if !UCONFIG_NO_BREAK_ITERATION + + /** + * Titlecases a UTF-16 string and optionally records edits. + * Casing is locale-dependent and context-sensitive. + * The result may be longer or shorter than the original. + * The source string and the destination buffer must not overlap. + * + * Titlecasing uses a break iterator to find the first characters of words + * that are to be titlecased. It titlecases those characters and lowercases + * all others. (This can be modified with options bits.) + * + * @param locale The locale ID. ("" = root locale, nullptr = default locale.) + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT, U_EDITS_NO_RESET, + * U_TITLECASE_NO_LOWERCASE, + * U_TITLECASE_NO_BREAK_ADJUSTMENT, U_TITLECASE_ADJUST_TO_CASED, + * U_TITLECASE_WHOLE_STRING, U_TITLECASE_SENTENCES. + * @param iter A break iterator to find the first characters of words that are to be titlecased. + * It is set to the source string (setText()) + * and used one or more times for iteration (first() and next()). + * If nullptr, then a word break iterator for the locale is used + * (or something equivalent). + * @param src The original string. + * @param srcLength The length of the original string. If -1, then src must be NUL-terminated. + * @param dest A buffer for the result string. The result will be NUL-terminated if + * the buffer is large enough. + * The contents is undefined in case of failure. + * @param destCapacity The size of the buffer (number of char16_ts). If it is 0, then + * dest may be nullptr and the function will only return the length of the result + * without writing any of the result string. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Reference to an in/out error code value + * which must not indicate a failure before the function call. + * @return The length of the result string, if successful. + * When the result would be longer than destCapacity, + * the full length is returned and a U_BUFFER_OVERFLOW_ERROR is set. + * + * @see u_strToTitle + * @see ucasemap_toTitle + * @stable ICU 59 + */ + static int32_t toTitle( + const char *locale, uint32_t options, BreakIterator *iter, + const char16_t *src, int32_t srcLength, + char16_t *dest, int32_t destCapacity, Edits *edits, + UErrorCode &errorCode); + +#endif // UCONFIG_NO_BREAK_ITERATION + + /** + * Case-folds a UTF-16 string and optionally records edits. + * + * Case folding is locale-independent and not context-sensitive, + * but there is an option for whether to include or exclude mappings for dotted I + * and dotless i that are marked with 'T' in CaseFolding.txt. + * + * The result may be longer or shorter than the original. + * The source string and the destination buffer must not overlap. + * + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT, U_EDITS_NO_RESET, + * U_FOLD_CASE_DEFAULT, U_FOLD_CASE_EXCLUDE_SPECIAL_I. + * @param src The original string. + * @param srcLength The length of the original string. If -1, then src must be NUL-terminated. + * @param dest A buffer for the result string. The result will be NUL-terminated if + * the buffer is large enough. + * The contents is undefined in case of failure. + * @param destCapacity The size of the buffer (number of char16_ts). If it is 0, then + * dest may be nullptr and the function will only return the length of the result + * without writing any of the result string. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Reference to an in/out error code value + * which must not indicate a failure before the function call. + * @return The length of the result string, if successful. + * When the result would be longer than destCapacity, + * the full length is returned and a U_BUFFER_OVERFLOW_ERROR is set. + * + * @see u_strFoldCase + * @stable ICU 59 + */ + static int32_t fold( + uint32_t options, + const char16_t *src, int32_t srcLength, + char16_t *dest, int32_t destCapacity, Edits *edits, + UErrorCode &errorCode); + + /** + * Lowercases a UTF-8 string and optionally records edits. + * Casing is locale-dependent and context-sensitive. + * The result may be longer or shorter than the original. + * + * @param locale The locale ID. ("" = root locale, nullptr = default locale.) + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT and U_EDITS_NO_RESET. + * @param src The original string. + * @param sink A ByteSink to which the result string is written. + * sink.Flush() is called at the end. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Reference to an in/out error code value + * which must not indicate a failure before the function call. + * + * @see ucasemap_utf8ToLower + * @stable ICU 60 + */ + static void utf8ToLower( + const char *locale, uint32_t options, + StringPiece src, ByteSink &sink, Edits *edits, + UErrorCode &errorCode); + + /** + * Uppercases a UTF-8 string and optionally records edits. + * Casing is locale-dependent and context-sensitive. + * The result may be longer or shorter than the original. + * + * @param locale The locale ID. ("" = root locale, nullptr = default locale.) + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT and U_EDITS_NO_RESET. + * @param src The original string. + * @param sink A ByteSink to which the result string is written. + * sink.Flush() is called at the end. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Reference to an in/out error code value + * which must not indicate a failure before the function call. + * + * @see ucasemap_utf8ToUpper + * @stable ICU 60 + */ + static void utf8ToUpper( + const char *locale, uint32_t options, + StringPiece src, ByteSink &sink, Edits *edits, + UErrorCode &errorCode); + +#if !UCONFIG_NO_BREAK_ITERATION + + /** + * Titlecases a UTF-8 string and optionally records edits. + * Casing is locale-dependent and context-sensitive. + * The result may be longer or shorter than the original. + * + * Titlecasing uses a break iterator to find the first characters of words + * that are to be titlecased. It titlecases those characters and lowercases + * all others. (This can be modified with options bits.) + * + * @param locale The locale ID. ("" = root locale, nullptr = default locale.) + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT, U_EDITS_NO_RESET, + * U_TITLECASE_NO_LOWERCASE, + * U_TITLECASE_NO_BREAK_ADJUSTMENT, U_TITLECASE_ADJUST_TO_CASED, + * U_TITLECASE_WHOLE_STRING, U_TITLECASE_SENTENCES. + * @param iter A break iterator to find the first characters of words that are to be titlecased. + * It is set to the source string (setUText()) + * and used one or more times for iteration (first() and next()). + * If nullptr, then a word break iterator for the locale is used + * (or something equivalent). + * @param src The original string. + * @param sink A ByteSink to which the result string is written. + * sink.Flush() is called at the end. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Reference to an in/out error code value + * which must not indicate a failure before the function call. + * + * @see ucasemap_utf8ToTitle + * @stable ICU 60 + */ + static void utf8ToTitle( + const char *locale, uint32_t options, BreakIterator *iter, + StringPiece src, ByteSink &sink, Edits *edits, + UErrorCode &errorCode); + +#endif // UCONFIG_NO_BREAK_ITERATION + + /** + * Case-folds a UTF-8 string and optionally records edits. + * + * Case folding is locale-independent and not context-sensitive, + * but there is an option for whether to include or exclude mappings for dotted I + * and dotless i that are marked with 'T' in CaseFolding.txt. + * + * The result may be longer or shorter than the original. + * + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT and U_EDITS_NO_RESET. + * @param src The original string. + * @param sink A ByteSink to which the result string is written. + * sink.Flush() is called at the end. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Reference to an in/out error code value + * which must not indicate a failure before the function call. + * + * @see ucasemap_utf8FoldCase + * @stable ICU 60 + */ + static void utf8Fold( + uint32_t options, + StringPiece src, ByteSink &sink, Edits *edits, + UErrorCode &errorCode); + + /** + * Lowercases a UTF-8 string and optionally records edits. + * Casing is locale-dependent and context-sensitive. + * The result may be longer or shorter than the original. + * The source string and the destination buffer must not overlap. + * + * @param locale The locale ID. ("" = root locale, nullptr = default locale.) + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT and U_EDITS_NO_RESET. + * @param src The original string. + * @param srcLength The length of the original string. If -1, then src must be NUL-terminated. + * @param dest A buffer for the result string. The result will be NUL-terminated if + * the buffer is large enough. + * The contents is undefined in case of failure. + * @param destCapacity The size of the buffer (number of bytes). If it is 0, then + * dest may be nullptr and the function will only return the length of the result + * without writing any of the result string. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Reference to an in/out error code value + * which must not indicate a failure before the function call. + * @return The length of the result string, if successful. + * When the result would be longer than destCapacity, + * the full length is returned and a U_BUFFER_OVERFLOW_ERROR is set. + * + * @see ucasemap_utf8ToLower + * @stable ICU 59 + */ + static int32_t utf8ToLower( + const char *locale, uint32_t options, + const char *src, int32_t srcLength, + char *dest, int32_t destCapacity, Edits *edits, + UErrorCode &errorCode); + + /** + * Uppercases a UTF-8 string and optionally records edits. + * Casing is locale-dependent and context-sensitive. + * The result may be longer or shorter than the original. + * The source string and the destination buffer must not overlap. + * + * @param locale The locale ID. ("" = root locale, nullptr = default locale.) + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT and U_EDITS_NO_RESET. + * @param src The original string. + * @param srcLength The length of the original string. If -1, then src must be NUL-terminated. + * @param dest A buffer for the result string. The result will be NUL-terminated if + * the buffer is large enough. + * The contents is undefined in case of failure. + * @param destCapacity The size of the buffer (number of bytes). If it is 0, then + * dest may be nullptr and the function will only return the length of the result + * without writing any of the result string. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Reference to an in/out error code value + * which must not indicate a failure before the function call. + * @return The length of the result string, if successful. + * When the result would be longer than destCapacity, + * the full length is returned and a U_BUFFER_OVERFLOW_ERROR is set. + * + * @see ucasemap_utf8ToUpper + * @stable ICU 59 + */ + static int32_t utf8ToUpper( + const char *locale, uint32_t options, + const char *src, int32_t srcLength, + char *dest, int32_t destCapacity, Edits *edits, + UErrorCode &errorCode); + +#if !UCONFIG_NO_BREAK_ITERATION + + /** + * Titlecases a UTF-8 string and optionally records edits. + * Casing is locale-dependent and context-sensitive. + * The result may be longer or shorter than the original. + * The source string and the destination buffer must not overlap. + * + * Titlecasing uses a break iterator to find the first characters of words + * that are to be titlecased. It titlecases those characters and lowercases + * all others. (This can be modified with options bits.) + * + * @param locale The locale ID. ("" = root locale, nullptr = default locale.) + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT, U_EDITS_NO_RESET, + * U_TITLECASE_NO_LOWERCASE, + * U_TITLECASE_NO_BREAK_ADJUSTMENT, U_TITLECASE_ADJUST_TO_CASED, + * U_TITLECASE_WHOLE_STRING, U_TITLECASE_SENTENCES. + * @param iter A break iterator to find the first characters of words that are to be titlecased. + * It is set to the source string (setUText()) + * and used one or more times for iteration (first() and next()). + * If nullptr, then a word break iterator for the locale is used + * (or something equivalent). + * @param src The original string. + * @param srcLength The length of the original string. If -1, then src must be NUL-terminated. + * @param dest A buffer for the result string. The result will be NUL-terminated if + * the buffer is large enough. + * The contents is undefined in case of failure. + * @param destCapacity The size of the buffer (number of bytes). If it is 0, then + * dest may be nullptr and the function will only return the length of the result + * without writing any of the result string. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Reference to an in/out error code value + * which must not indicate a failure before the function call. + * @return The length of the result string, if successful. + * When the result would be longer than destCapacity, + * the full length is returned and a U_BUFFER_OVERFLOW_ERROR is set. + * + * @see ucasemap_utf8ToTitle + * @stable ICU 59 + */ + static int32_t utf8ToTitle( + const char *locale, uint32_t options, BreakIterator *iter, + const char *src, int32_t srcLength, + char *dest, int32_t destCapacity, Edits *edits, + UErrorCode &errorCode); + +#endif // UCONFIG_NO_BREAK_ITERATION + + /** + * Case-folds a UTF-8 string and optionally records edits. + * + * Case folding is locale-independent and not context-sensitive, + * but there is an option for whether to include or exclude mappings for dotted I + * and dotless i that are marked with 'T' in CaseFolding.txt. + * + * The result may be longer or shorter than the original. + * The source string and the destination buffer must not overlap. + * + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT, U_EDITS_NO_RESET, + * U_FOLD_CASE_DEFAULT, U_FOLD_CASE_EXCLUDE_SPECIAL_I. + * @param src The original string. + * @param srcLength The length of the original string. If -1, then src must be NUL-terminated. + * @param dest A buffer for the result string. The result will be NUL-terminated if + * the buffer is large enough. + * The contents is undefined in case of failure. + * @param destCapacity The size of the buffer (number of bytes). If it is 0, then + * dest may be nullptr and the function will only return the length of the result + * without writing any of the result string. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Reference to an in/out error code value + * which must not indicate a failure before the function call. + * @return The length of the result string, if successful. + * When the result would be longer than destCapacity, + * the full length is returned and a U_BUFFER_OVERFLOW_ERROR is set. + * + * @see ucasemap_utf8FoldCase + * @stable ICU 59 + */ + static int32_t utf8Fold( + uint32_t options, + const char *src, int32_t srcLength, + char *dest, int32_t destCapacity, Edits *edits, + UErrorCode &errorCode); + +private: + CaseMap() = delete; + CaseMap(const CaseMap &other) = delete; + CaseMap &operator=(const CaseMap &other) = delete; +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __CASEMAP_H__ diff --git a/intl/icu/source/common/unicode/char16ptr.h b/intl/icu/source/common/unicode/char16ptr.h new file mode 100644 index 0000000000..de8182c7ad --- /dev/null +++ b/intl/icu/source/common/unicode/char16ptr.h @@ -0,0 +1,313 @@ +// © 2017 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// char16ptr.h +// created: 2017feb28 Markus W. Scherer + +#ifndef __CHAR16PTR_H__ +#define __CHAR16PTR_H__ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include + +/** + * \file + * \brief C++ API: char16_t pointer wrappers with + * implicit conversion from bit-compatible raw pointer types. + * Also conversion functions from char16_t * to UChar * and OldUChar *. + */ + +U_NAMESPACE_BEGIN + +/** + * \def U_ALIASING_BARRIER + * Barrier for pointer anti-aliasing optimizations even across function boundaries. + * @internal + */ +#ifdef U_ALIASING_BARRIER + // Use the predefined value. +#elif (defined(__clang__) || defined(__GNUC__)) && U_PLATFORM != U_PF_BROWSER_NATIVE_CLIENT +# define U_ALIASING_BARRIER(ptr) asm volatile("" : : "rm"(ptr) : "memory") +#elif defined(U_IN_DOXYGEN) +# define U_ALIASING_BARRIER(ptr) +#endif + +/** + * char16_t * wrapper with implicit conversion from distinct but bit-compatible pointer types. + * @stable ICU 59 + */ +class U_COMMON_API Char16Ptr final { +public: + /** + * Copies the pointer. + * @param p pointer + * @stable ICU 59 + */ + inline Char16Ptr(char16_t *p); +#if !U_CHAR16_IS_TYPEDEF + /** + * Converts the pointer to char16_t *. + * @param p pointer to be converted + * @stable ICU 59 + */ + inline Char16Ptr(uint16_t *p); +#endif +#if U_SIZEOF_WCHAR_T==2 || defined(U_IN_DOXYGEN) + /** + * Converts the pointer to char16_t *. + * (Only defined if U_SIZEOF_WCHAR_T==2.) + * @param p pointer to be converted + * @stable ICU 59 + */ + inline Char16Ptr(wchar_t *p); +#endif + /** + * nullptr constructor. + * @param p nullptr + * @stable ICU 59 + */ + inline Char16Ptr(std::nullptr_t p); + /** + * Destructor. + * @stable ICU 59 + */ + inline ~Char16Ptr(); + + /** + * Pointer access. + * @return the wrapped pointer + * @stable ICU 59 + */ + inline char16_t *get() const; + /** + * char16_t pointer access via type conversion (e.g., static_cast). + * @return the wrapped pointer + * @stable ICU 59 + */ + inline operator char16_t *() const { return get(); } + +private: + Char16Ptr() = delete; + +#ifdef U_ALIASING_BARRIER + template static char16_t *cast(T *t) { + U_ALIASING_BARRIER(t); + return reinterpret_cast(t); + } + + char16_t *p_; +#else + union { + char16_t *cp; + uint16_t *up; + wchar_t *wp; + } u_; +#endif +}; + +/// \cond +#ifdef U_ALIASING_BARRIER + +Char16Ptr::Char16Ptr(char16_t *p) : p_(p) {} +#if !U_CHAR16_IS_TYPEDEF +Char16Ptr::Char16Ptr(uint16_t *p) : p_(cast(p)) {} +#endif +#if U_SIZEOF_WCHAR_T==2 +Char16Ptr::Char16Ptr(wchar_t *p) : p_(cast(p)) {} +#endif +Char16Ptr::Char16Ptr(std::nullptr_t p) : p_(p) {} +Char16Ptr::~Char16Ptr() { + U_ALIASING_BARRIER(p_); +} + +char16_t *Char16Ptr::get() const { return p_; } + +#else + +Char16Ptr::Char16Ptr(char16_t *p) { u_.cp = p; } +#if !U_CHAR16_IS_TYPEDEF +Char16Ptr::Char16Ptr(uint16_t *p) { u_.up = p; } +#endif +#if U_SIZEOF_WCHAR_T==2 +Char16Ptr::Char16Ptr(wchar_t *p) { u_.wp = p; } +#endif +Char16Ptr::Char16Ptr(std::nullptr_t p) { u_.cp = p; } +Char16Ptr::~Char16Ptr() {} + +char16_t *Char16Ptr::get() const { return u_.cp; } + +#endif +/// \endcond + +/** + * const char16_t * wrapper with implicit conversion from distinct but bit-compatible pointer types. + * @stable ICU 59 + */ +class U_COMMON_API ConstChar16Ptr final { +public: + /** + * Copies the pointer. + * @param p pointer + * @stable ICU 59 + */ + inline ConstChar16Ptr(const char16_t *p); +#if !U_CHAR16_IS_TYPEDEF + /** + * Converts the pointer to char16_t *. + * @param p pointer to be converted + * @stable ICU 59 + */ + inline ConstChar16Ptr(const uint16_t *p); +#endif +#if U_SIZEOF_WCHAR_T==2 || defined(U_IN_DOXYGEN) + /** + * Converts the pointer to char16_t *. + * (Only defined if U_SIZEOF_WCHAR_T==2.) + * @param p pointer to be converted + * @stable ICU 59 + */ + inline ConstChar16Ptr(const wchar_t *p); +#endif + /** + * nullptr constructor. + * @param p nullptr + * @stable ICU 59 + */ + inline ConstChar16Ptr(const std::nullptr_t p); + + /** + * Destructor. + * @stable ICU 59 + */ + inline ~ConstChar16Ptr(); + + /** + * Pointer access. + * @return the wrapped pointer + * @stable ICU 59 + */ + inline const char16_t *get() const; + /** + * char16_t pointer access via type conversion (e.g., static_cast). + * @return the wrapped pointer + * @stable ICU 59 + */ + inline operator const char16_t *() const { return get(); } + +private: + ConstChar16Ptr() = delete; + +#ifdef U_ALIASING_BARRIER + template static const char16_t *cast(const T *t) { + U_ALIASING_BARRIER(t); + return reinterpret_cast(t); + } + + const char16_t *p_; +#else + union { + const char16_t *cp; + const uint16_t *up; + const wchar_t *wp; + } u_; +#endif +}; + +/// \cond +#ifdef U_ALIASING_BARRIER + +ConstChar16Ptr::ConstChar16Ptr(const char16_t *p) : p_(p) {} +#if !U_CHAR16_IS_TYPEDEF +ConstChar16Ptr::ConstChar16Ptr(const uint16_t *p) : p_(cast(p)) {} +#endif +#if U_SIZEOF_WCHAR_T==2 +ConstChar16Ptr::ConstChar16Ptr(const wchar_t *p) : p_(cast(p)) {} +#endif +ConstChar16Ptr::ConstChar16Ptr(const std::nullptr_t p) : p_(p) {} +ConstChar16Ptr::~ConstChar16Ptr() { + U_ALIASING_BARRIER(p_); +} + +const char16_t *ConstChar16Ptr::get() const { return p_; } + +#else + +ConstChar16Ptr::ConstChar16Ptr(const char16_t *p) { u_.cp = p; } +#if !U_CHAR16_IS_TYPEDEF +ConstChar16Ptr::ConstChar16Ptr(const uint16_t *p) { u_.up = p; } +#endif +#if U_SIZEOF_WCHAR_T==2 +ConstChar16Ptr::ConstChar16Ptr(const wchar_t *p) { u_.wp = p; } +#endif +ConstChar16Ptr::ConstChar16Ptr(const std::nullptr_t p) { u_.cp = p; } +ConstChar16Ptr::~ConstChar16Ptr() {} + +const char16_t *ConstChar16Ptr::get() const { return u_.cp; } + +#endif +/// \endcond + +/** + * Converts from const char16_t * to const UChar *. + * Includes an aliasing barrier if available. + * @param p pointer + * @return p as const UChar * + * @stable ICU 59 + */ +inline const UChar *toUCharPtr(const char16_t *p) { +#ifdef U_ALIASING_BARRIER + U_ALIASING_BARRIER(p); +#endif + return reinterpret_cast(p); +} + +/** + * Converts from char16_t * to UChar *. + * Includes an aliasing barrier if available. + * @param p pointer + * @return p as UChar * + * @stable ICU 59 + */ +inline UChar *toUCharPtr(char16_t *p) { +#ifdef U_ALIASING_BARRIER + U_ALIASING_BARRIER(p); +#endif + return reinterpret_cast(p); +} + +/** + * Converts from const char16_t * to const OldUChar *. + * Includes an aliasing barrier if available. + * @param p pointer + * @return p as const OldUChar * + * @stable ICU 59 + */ +inline const OldUChar *toOldUCharPtr(const char16_t *p) { +#ifdef U_ALIASING_BARRIER + U_ALIASING_BARRIER(p); +#endif + return reinterpret_cast(p); +} + +/** + * Converts from char16_t * to OldUChar *. + * Includes an aliasing barrier if available. + * @param p pointer + * @return p as OldUChar * + * @stable ICU 59 + */ +inline OldUChar *toOldUCharPtr(char16_t *p) { +#ifdef U_ALIASING_BARRIER + U_ALIASING_BARRIER(p); +#endif + return reinterpret_cast(p); +} + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __CHAR16PTR_H__ diff --git a/intl/icu/source/common/unicode/chariter.h b/intl/icu/source/common/unicode/chariter.h new file mode 100644 index 0000000000..45f4d984c7 --- /dev/null +++ b/intl/icu/source/common/unicode/chariter.h @@ -0,0 +1,734 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************** +* +* Copyright (C) 1997-2011, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************** +*/ + +#ifndef CHARITER_H +#define CHARITER_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/uobject.h" +#include "unicode/unistr.h" +/** + * \file + * \brief C++ API: Character Iterator + */ + +U_NAMESPACE_BEGIN +/** + * Abstract class that defines an API for forward-only iteration + * on text objects. + * This is a minimal interface for iteration without random access + * or backwards iteration. It is especially useful for wrapping + * streams with converters into an object for collation or + * normalization. + * + *

Characters can be accessed in two ways: as code units or as + * code points. + * Unicode code points are 21-bit integers and are the scalar values + * of Unicode characters. ICU uses the type UChar32 for them. + * Unicode code units are the storage units of a given + * Unicode/UCS Transformation Format (a character encoding scheme). + * With UTF-16, all code points can be represented with either one + * or two code units ("surrogates"). + * String storage is typically based on code units, while properties + * of characters are typically determined using code point values. + * Some processes may be designed to work with sequences of code units, + * or it may be known that all characters that are important to an + * algorithm can be represented with single code units. + * Other processes will need to use the code point access functions.

+ * + *

ForwardCharacterIterator provides nextPostInc() to access + * a code unit and advance an internal position into the text object, + * similar to a return text[position++].
+ * It provides next32PostInc() to access a code point and advance an internal + * position.

+ * + *

next32PostInc() assumes that the current position is that of + * the beginning of a code point, i.e., of its first code unit. + * After next32PostInc(), this will be true again. + * In general, access to code units and code points in the same + * iteration loop should not be mixed. In UTF-16, if the current position + * is on a second code unit (Low Surrogate), then only that code unit + * is returned even by next32PostInc().

+ * + *

For iteration with either function, there are two ways to + * check for the end of the iteration. When there are no more + * characters in the text object: + *

    + *
  • The hasNext() function returns false.
    • + *
  • nextPostInc() and next32PostInc() return DONE + * when one attempts to read beyond the end of the text object.
    • + *
+ * + * Example: + * \code + * void function1(ForwardCharacterIterator &it) { + * UChar32 c; + * while(it.hasNext()) { + * c=it.next32PostInc(); + * // use c + * } + * } + * + * void function1(ForwardCharacterIterator &it) { + * char16_t c; + * while((c=it.nextPostInc())!=ForwardCharacterIterator::DONE) { + * // use c + * } + * } + * \endcode + *

+ * + * @stable ICU 2.0 + */ +class U_COMMON_API ForwardCharacterIterator : public UObject { +public: + /** + * Value returned by most of ForwardCharacterIterator's functions + * when the iterator has reached the limits of its iteration. + * @stable ICU 2.0 + */ + enum { DONE = 0xffff }; + + /** + * Destructor. + * @stable ICU 2.0 + */ + virtual ~ForwardCharacterIterator(); + + /** + * Returns true when both iterators refer to the same + * character in the same character-storage object. + * @param that The ForwardCharacterIterator to be compared for equality + * @return true when both iterators refer to the same + * character in the same character-storage object + * @stable ICU 2.0 + */ + virtual bool operator==(const ForwardCharacterIterator& that) const = 0; + + /** + * Returns true when the iterators refer to different + * text-storage objects, or to different characters in the + * same text-storage object. + * @param that The ForwardCharacterIterator to be compared for inequality + * @return true when the iterators refer to different + * text-storage objects, or to different characters in the + * same text-storage object + * @stable ICU 2.0 + */ + inline bool operator!=(const ForwardCharacterIterator& that) const; + + /** + * Generates a hash code for this iterator. + * @return the hash code. + * @stable ICU 2.0 + */ + virtual int32_t hashCode(void) const = 0; + + /** + * Returns a UClassID for this ForwardCharacterIterator ("poor man's + * RTTI").

Despite the fact that this function is public, + * DO NOT CONSIDER IT PART OF CHARACTERITERATOR'S API! + * @return a UClassID for this ForwardCharacterIterator + * @stable ICU 2.0 + */ + virtual UClassID getDynamicClassID(void) const override = 0; + + /** + * Gets the current code unit for returning and advances to the next code unit + * in the iteration range + * (toward endIndex()). If there are + * no more code units to return, returns DONE. + * @return the current code unit. + * @stable ICU 2.0 + */ + virtual char16_t nextPostInc(void) = 0; + + /** + * Gets the current code point for returning and advances to the next code point + * in the iteration range + * (toward endIndex()). If there are + * no more code points to return, returns DONE. + * @return the current code point. + * @stable ICU 2.0 + */ + virtual UChar32 next32PostInc(void) = 0; + + /** + * Returns false if there are no more code units or code points + * at or after the current position in the iteration range. + * This is used with nextPostInc() or next32PostInc() in forward + * iteration. + * @returns false if there are no more code units or code points + * at or after the current position in the iteration range. + * @stable ICU 2.0 + */ + virtual UBool hasNext() = 0; + +protected: + /** Default constructor to be overridden in the implementing class. @stable ICU 2.0*/ + ForwardCharacterIterator(); + + /** Copy constructor to be overridden in the implementing class. @stable ICU 2.0*/ + ForwardCharacterIterator(const ForwardCharacterIterator &other); + + /** + * Assignment operator to be overridden in the implementing class. + * @stable ICU 2.0 + */ + ForwardCharacterIterator &operator=(const ForwardCharacterIterator&) { return *this; } +}; + +/** + * Abstract class that defines an API for iteration + * on text objects. + * This is an interface for forward and backward iteration + * and random access into a text object. + * + *

The API provides backward compatibility to the Java and older ICU + * CharacterIterator classes but extends them significantly: + *

    + *
  1. CharacterIterator is now a subclass of ForwardCharacterIterator.
    2. + *
  2. While the old API functions provided forward iteration with + * "pre-increment" semantics, the new one also provides functions + * with "post-increment" semantics. They are more efficient and should + * be the preferred iterator functions for new implementations. + * The backward iteration always had "pre-decrement" semantics, which + * are efficient.
    3. + *
  3. Just like ForwardCharacterIterator, it provides access to + * both code units and code points. Code point access versions are available + * for the old and the new iteration semantics.
    4. + *
  4. There are new functions for setting and moving the current position + * without returning a character, for efficiency.
    5. + *
+ * + * See ForwardCharacterIterator for examples for using the new forward iteration + * functions. For backward iteration, there is also a hasPrevious() function + * that can be used analogously to hasNext(). + * The old functions work as before and are shown below.

+ * + *

Examples for some of the new functions:

+ * + * Forward iteration with hasNext(): + * \code + * void forward1(CharacterIterator &it) { + * UChar32 c; + * for(it.setToStart(); it.hasNext();) { + * c=it.next32PostInc(); + * // use c + * } + * } + * \endcode + * Forward iteration more similar to loops with the old forward iteration, + * showing a way to convert simple for() loops: + * \code + * void forward2(CharacterIterator &it) { + * char16_t c; + * for(c=it.firstPostInc(); c!=CharacterIterator::DONE; c=it.nextPostInc()) { + * // use c + * } + * } + * \endcode + * Backward iteration with setToEnd() and hasPrevious(): + * \code + * void backward1(CharacterIterator &it) { + * UChar32 c; + * for(it.setToEnd(); it.hasPrevious();) { + * c=it.previous32(); + * // use c + * } + * } + * \endcode + * Backward iteration with a more traditional for() loop: + * \code + * void backward2(CharacterIterator &it) { + * char16_t c; + * for(c=it.last(); c!=CharacterIterator::DONE; c=it.previous()) { + * // use c + * } + * } + * \endcode + * + * Example for random access: + * \code + * void random(CharacterIterator &it) { + * // set to the third code point from the beginning + * it.move32(3, CharacterIterator::kStart); + * // get a code point from here without moving the position + * UChar32 c=it.current32(); + * // get the position + * int32_t pos=it.getIndex(); + * // get the previous code unit + * char16_t u=it.previous(); + * // move back one more code unit + * it.move(-1, CharacterIterator::kCurrent); + * // set the position back to where it was + * // and read the same code point c and move beyond it + * it.setIndex(pos); + * if(c!=it.next32PostInc()) { + * exit(1); // CharacterIterator inconsistent + * } + * } + * \endcode + * + *

Examples, especially for the old API:

+ * + * Function processing characters, in this example simple output + * 
+ * \code
+ *  void processChar( char16_t c )
+ *  {
+ *      cout << " " << c;
+ *  }
+ * \endcode
+ *
+ * Traverse the text from start to finish + * 
 
+ * \code
+ *  void traverseForward(CharacterIterator& iter)
+ *  {
+ *      for(char16_t c = iter.first(); c != CharacterIterator::DONE; c = iter.next()) {
+ *          processChar(c);
+ *      }
+ *  }
+ * \endcode
+ *
+ * Traverse the text backwards, from end to start + * 
+ * \code
+ *  void traverseBackward(CharacterIterator& iter)
+ *  {
+ *      for(char16_t c = iter.last(); c != CharacterIterator::DONE; c = iter.previous()) {
+ *          processChar(c);
+ *      }
+ *  }
+ * \endcode
+ *
+ * Traverse both forward and backward from a given position in the text. + * Calls to notBoundary() in this example represents some additional stopping criteria. + * 
+ * \code
+ * void traverseOut(CharacterIterator& iter, int32_t pos)
+ * {
+ *      char16_t c;
+ *      for (c = iter.setIndex(pos);
+ *      c != CharacterIterator::DONE && (Unicode::isLetter(c) || Unicode::isDigit(c));
+ *          c = iter.next()) {}
+ *      int32_t end = iter.getIndex();
+ *      for (c = iter.setIndex(pos);
+ *          c != CharacterIterator::DONE && (Unicode::isLetter(c) || Unicode::isDigit(c));
+ *          c = iter.previous()) {}
+ *      int32_t start = iter.getIndex() + 1;
+ *  
+ *      cout << "start: " << start << " end: " << end << endl;
+ *      for (c = iter.setIndex(start); iter.getIndex() < end; c = iter.next() ) {
+ *          processChar(c);
+ *     }
+ *  }
+ * \endcode
+ *
+ * Creating a StringCharacterIterator and calling the test functions + * 
+ * \code
+ *  void CharacterIterator_Example( void )
+ *   {
+ *       cout << endl << "===== CharacterIterator_Example: =====" << endl;
+ *       UnicodeString text("Ein kleiner Satz.");
+ *       StringCharacterIterator iterator(text);
+ *       cout << "----- traverseForward: -----------" << endl;
+ *       traverseForward( iterator );
+ *       cout << endl << endl << "----- traverseBackward: ----------" << endl;
+ *       traverseBackward( iterator );
+ *       cout << endl << endl << "----- traverseOut: ---------------" << endl;
+ *       traverseOut( iterator, 7 );
+ *       cout << endl << endl << "-----" << endl;
+ *   }
+ * \endcode
+ *
+ * + * @stable ICU 2.0 + */ +class U_COMMON_API CharacterIterator : public ForwardCharacterIterator { +public: + /** + * Origin enumeration for the move() and move32() functions. + * @stable ICU 2.0 + */ + enum EOrigin { kStart, kCurrent, kEnd }; + + /** + * Destructor. + * @stable ICU 2.0 + */ + virtual ~CharacterIterator(); + + /** + * Returns a pointer to a new CharacterIterator of the same + * concrete class as this one, and referring to the same + * character in the same text-storage object as this one. The + * caller is responsible for deleting the new clone. + * @return a pointer to a new CharacterIterator + * @stable ICU 2.0 + */ + virtual CharacterIterator* clone() const = 0; + + /** + * Sets the iterator to refer to the first code unit in its + * iteration range, and returns that code unit. + * This can be used to begin an iteration with next(). + * @return the first code unit in its iteration range. + * @stable ICU 2.0 + */ + virtual char16_t first(void) = 0; + + /** + * Sets the iterator to refer to the first code unit in its + * iteration range, returns that code unit, and moves the position + * to the second code unit. This is an alternative to setToStart() + * for forward iteration with nextPostInc(). + * @return the first code unit in its iteration range. + * @stable ICU 2.0 + */ + virtual char16_t firstPostInc(void); + + /** + * Sets the iterator to refer to the first code point in its + * iteration range, and returns that code unit, + * This can be used to begin an iteration with next32(). + * Note that an iteration with next32PostInc(), beginning with, + * e.g., setToStart() or firstPostInc(), is more efficient. + * @return the first code point in its iteration range. + * @stable ICU 2.0 + */ + virtual UChar32 first32(void) = 0; + + /** + * Sets the iterator to refer to the first code point in its + * iteration range, returns that code point, and moves the position + * to the second code point. This is an alternative to setToStart() + * for forward iteration with next32PostInc(). + * @return the first code point in its iteration range. + * @stable ICU 2.0 + */ + virtual UChar32 first32PostInc(void); + + /** + * Sets the iterator to refer to the first code unit or code point in its + * iteration range. This can be used to begin a forward + * iteration with nextPostInc() or next32PostInc(). + * @return the start position of the iteration range + * @stable ICU 2.0 + */ + inline int32_t setToStart(); + + /** + * Sets the iterator to refer to the last code unit in its + * iteration range, and returns that code unit. + * This can be used to begin an iteration with previous(). + * @return the last code unit. + * @stable ICU 2.0 + */ + virtual char16_t last(void) = 0; + + /** + * Sets the iterator to refer to the last code point in its + * iteration range, and returns that code unit. + * This can be used to begin an iteration with previous32(). + * @return the last code point. + * @stable ICU 2.0 + */ + virtual UChar32 last32(void) = 0; + + /** + * Sets the iterator to the end of its iteration range, just behind + * the last code unit or code point. This can be used to begin a backward + * iteration with previous() or previous32(). + * @return the end position of the iteration range + * @stable ICU 2.0 + */ + inline int32_t setToEnd(); + + /** + * Sets the iterator to refer to the "position"-th code unit + * in the text-storage object the iterator refers to, and + * returns that code unit. + * @param position the "position"-th code unit in the text-storage object + * @return the "position"-th code unit. + * @stable ICU 2.0 + */ + virtual char16_t setIndex(int32_t position) = 0; + + /** + * Sets the iterator to refer to the beginning of the code point + * that contains the "position"-th code unit + * in the text-storage object the iterator refers to, and + * returns that code point. + * The current position is adjusted to the beginning of the code point + * (its first code unit). + * @param position the "position"-th code unit in the text-storage object + * @return the "position"-th code point. + * @stable ICU 2.0 + */ + virtual UChar32 setIndex32(int32_t position) = 0; + + /** + * Returns the code unit the iterator currently refers to. + * @return the current code unit. + * @stable ICU 2.0 + */ + virtual char16_t current(void) const = 0; + + /** + * Returns the code point the iterator currently refers to. + * @return the current code point. + * @stable ICU 2.0 + */ + virtual UChar32 current32(void) const = 0; + + /** + * Advances to the next code unit in the iteration range + * (toward endIndex()), and returns that code unit. If there are + * no more code units to return, returns DONE. + * @return the next code unit. + * @stable ICU 2.0 + */ + virtual char16_t next(void) = 0; + + /** + * Advances to the next code point in the iteration range + * (toward endIndex()), and returns that code point. If there are + * no more code points to return, returns DONE. + * Note that iteration with "pre-increment" semantics is less + * efficient than iteration with "post-increment" semantics + * that is provided by next32PostInc(). + * @return the next code point. + * @stable ICU 2.0 + */ + virtual UChar32 next32(void) = 0; + + /** + * Advances to the previous code unit in the iteration range + * (toward startIndex()), and returns that code unit. If there are + * no more code units to return, returns DONE. + * @return the previous code unit. + * @stable ICU 2.0 + */ + virtual char16_t previous(void) = 0; + + /** + * Advances to the previous code point in the iteration range + * (toward startIndex()), and returns that code point. If there are + * no more code points to return, returns DONE. + * @return the previous code point. + * @stable ICU 2.0 + */ + virtual UChar32 previous32(void) = 0; + + /** + * Returns false if there are no more code units or code points + * before the current position in the iteration range. + * This is used with previous() or previous32() in backward + * iteration. + * @return false if there are no more code units or code points + * before the current position in the iteration range, return true otherwise. + * @stable ICU 2.0 + */ + virtual UBool hasPrevious() = 0; + + /** + * Returns the numeric index in the underlying text-storage + * object of the character returned by first(). Since it's + * possible to create an iterator that iterates across only + * part of a text-storage object, this number isn't + * necessarily 0. + * @returns the numeric index in the underlying text-storage + * object of the character returned by first(). + * @stable ICU 2.0 + */ + inline int32_t startIndex(void) const; + + /** + * Returns the numeric index in the underlying text-storage + * object of the position immediately BEYOND the character + * returned by last(). + * @return the numeric index in the underlying text-storage + * object of the position immediately BEYOND the character + * returned by last(). + * @stable ICU 2.0 + */ + inline int32_t endIndex(void) const; + + /** + * Returns the numeric index in the underlying text-storage + * object of the character the iterator currently refers to + * (i.e., the character returned by current()). + * @return the numeric index in the text-storage object of + * the character the iterator currently refers to + * @stable ICU 2.0 + */ + inline int32_t getIndex(void) const; + + /** + * Returns the length of the entire text in the underlying + * text-storage object. + * @return the length of the entire text in the text-storage object + * @stable ICU 2.0 + */ + inline int32_t getLength() const; + + /** + * Moves the current position relative to the start or end of the + * iteration range, or relative to the current position itself. + * The movement is expressed in numbers of code units forward + * or backward by specifying a positive or negative delta. + * @param delta the position relative to origin. A positive delta means forward; + * a negative delta means backward. + * @param origin Origin enumeration {kStart, kCurrent, kEnd} + * @return the new position + * @stable ICU 2.0 + */ + virtual int32_t move(int32_t delta, EOrigin origin) = 0; + + /** + * Moves the current position relative to the start or end of the + * iteration range, or relative to the current position itself. + * The movement is expressed in numbers of code points forward + * or backward by specifying a positive or negative delta. + * @param delta the position relative to origin. A positive delta means forward; + * a negative delta means backward. + * @param origin Origin enumeration {kStart, kCurrent, kEnd} + * @return the new position + * @stable ICU 2.0 + */ +#ifdef move32 + // One of the system headers right now is sometimes defining a conflicting macro we don't use +#undef move32 +#endif + virtual int32_t move32(int32_t delta, EOrigin origin) = 0; + + /** + * Copies the text under iteration into the UnicodeString + * referred to by "result". + * @param result Receives a copy of the text under iteration. + * @stable ICU 2.0 + */ + virtual void getText(UnicodeString& result) = 0; + +protected: + /** + * Empty constructor. + * @stable ICU 2.0 + */ + CharacterIterator(); + + /** + * Constructor, just setting the length field in this base class. + * @stable ICU 2.0 + */ + CharacterIterator(int32_t length); + + /** + * Constructor, just setting the length and position fields in this base class. + * @stable ICU 2.0 + */ + CharacterIterator(int32_t length, int32_t position); + + /** + * Constructor, just setting the length, start, end, and position fields in this base class. + * @stable ICU 2.0 + */ + CharacterIterator(int32_t length, int32_t textBegin, int32_t textEnd, int32_t position); + + /** + * Copy constructor. + * + * @param that The CharacterIterator to be copied + * @stable ICU 2.0 + */ + CharacterIterator(const CharacterIterator &that); + + /** + * Assignment operator. Sets this CharacterIterator to have the same behavior, + * as the one passed in. + * @param that The CharacterIterator passed in. + * @return the newly set CharacterIterator. + * @stable ICU 2.0 + */ + CharacterIterator &operator=(const CharacterIterator &that); + + /** + * Base class text length field. + * Necessary this for correct getText() and hashCode(). + * @stable ICU 2.0 + */ + int32_t textLength; + + /** + * Base class field for the current position. + * @stable ICU 2.0 + */ + int32_t pos; + + /** + * Base class field for the start of the iteration range. + * @stable ICU 2.0 + */ + int32_t begin; + + /** + * Base class field for the end of the iteration range. + * @stable ICU 2.0 + */ + int32_t end; +}; + +inline bool +ForwardCharacterIterator::operator!=(const ForwardCharacterIterator& that) const { + return !operator==(that); +} + +inline int32_t +CharacterIterator::setToStart() { + return move(0, kStart); +} + +inline int32_t +CharacterIterator::setToEnd() { + return move(0, kEnd); +} + +inline int32_t +CharacterIterator::startIndex(void) const { + return begin; +} + +inline int32_t +CharacterIterator::endIndex(void) const { + return end; +} + +inline int32_t +CharacterIterator::getIndex(void) const { + return pos; +} + +inline int32_t +CharacterIterator::getLength(void) const { + return textLength; +} + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/dbbi.h b/intl/icu/source/common/unicode/dbbi.h new file mode 100644 index 0000000000..3de9cc3814 --- /dev/null +++ b/intl/icu/source/common/unicode/dbbi.h @@ -0,0 +1,48 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1999-2006,2013 IBM Corp. All rights reserved. +********************************************************************** +* Date Name Description +* 12/1/99 rgillam Complete port from Java. +* 01/13/2000 helena Added UErrorCode to ctors. +********************************************************************** +*/ + +#ifndef DBBI_H +#define DBBI_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/rbbi.h" + +#if !UCONFIG_NO_BREAK_ITERATION + +/** + * \file + * \brief C++ API: Dictionary Based Break Iterator + */ + +U_NAMESPACE_BEGIN + +#ifndef U_HIDE_DEPRECATED_API +/** + * An obsolete subclass of RuleBasedBreakIterator. Handling of dictionary- + * based break iteration has been folded into the base class. This class + * is deprecated as of ICU 3.6. + * @deprecated ICU 3.6 + */ +typedef RuleBasedBreakIterator DictionaryBasedBreakIterator; + +#endif /* U_HIDE_DEPRECATED_API */ + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/docmain.h b/intl/icu/source/common/unicode/docmain.h new file mode 100644 index 0000000000..1349f49703 --- /dev/null +++ b/intl/icu/source/common/unicode/docmain.h @@ -0,0 +1,237 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/******************************************************************** + * COPYRIGHT: + * Copyright (c) 1997-2012, International Business Machines Corporation and + * others. All Rights Reserved. + * + * FILE NAME: DOCMAIN.h + * + * Date Name Description + * 12/11/2000 Ram Creation. + */ + +/** + * \file + * \brief (Non API- contains Doxygen definitions) + * + * This file contains documentation for Doxygen and does not have + * any significance with respect to C or C++ API + */ + +/*! \mainpage + * + * \section API API Reference Usage + * + *

C++ Programmers:

+ *

Use Class Hierarchy or Alphabetical List + * or Compound List + * to find the class you are interested in. For example, to find BreakIterator, + * you can go to the Alphabetical List, then click on + * "BreakIterator". Once you are at the class, you will find an inheritance + * chart, a list of the public members, a detailed description of the class, + * then detailed member descriptions.

+ * + *

C Programmers:

+ *

Use Module List or File Members + * to find a list of all the functions and constants. + * For example, to find BreakIterator functions you would click on + * File List, + * then find "ubrk.h" and click on it. You will find descriptions of Defines, + * Typedefs, Enumerations, and Functions, with detailed descriptions below. + * If you want to find a specific function, such as ubrk_next(), then click + * first on File Members, then use your browser + * Find dialog to search for "ubrk_next()".

+ * + * + *

API References for Previous Releases

+ *

The API References for each release of ICU are also available as + * a zip file from the ICU + * download page.

+ * + *
+ * + *

Architecture (User's Guide)

+ * + * + *
+ *\htmlonly

Module List

\endhtmlonly + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + *
Module NameCC++
Basic Types and Constantsutypes.hutypes.h
Strings and Character Iterationustring.h, utf8.h, utf16.h, icu::StringPiece, UText, UCharIterator, icu::ByteSinkicu::UnicodeString, icu::CharacterIterator, icu::Appendable, icu::StringPiece,icu::ByteSink
Unicode Character
Properties and Names		uchar.h, uscript.hC API
Sets of Unicode Code Points and Stringsuset.hicu::UnicodeSet
Maps from Unicode Code Points to Integer Valuesucptrie.h, umutablecptrie.hC API
Maps from Strings to Integer Values(no C API)icu::BytesTrie, icu::UCharsTrie
Codepage Conversionucnv.h, ucnvsel.hC API
Codepage Detectionucsdet.hC API
Unicode Text Compressionucnv.h
(encoding name "SCSU" or "BOCU-1")		C API
Locales uloc.hicu::Locale, icu::LocaleBuilder, icu::LocaleMatcher
Resource Bundlesures.hicu::ResourceBundle
Normalizationunorm2.hicu::Normalizer2
Calendars and Time Zonesucal.hicu::Calendar, icu::TimeZone
Date and Time Formattingudat.hicu::DateFormat
Message Formattingumsg.hicu::MessageFormat
List Formattingulistformatter.hicu::ListFormatter
Number Formatting
(includes currency and unit formatting)		unumberformatter.h, unum.h, usimplenumberformatter.hicu::number::NumberFormatter (ICU 60+) or icu::NumberFormat (older versions)
icu::number::SimpleNumberFormatter (ICU 73+)
Number Range Formatting
(includes currency and unit ranges)		unumberrangeformatter.hicu::number::NumberRangeFormatter
Number Spellout
(Rule Based Number Formatting)		unum.h
(use UNUM_SPELLOUT)		icu::RuleBasedNumberFormat
Text Transformation
(Transliteration)		utrans.hicu::Transliterator
Bidirectional Algorithmubidi.h, ubiditransform.hC API
Arabic Shapingushape.hC API
Collationucol.hicu::Collator
String Searchingusearch.hicu::StringSearch
Index Characters/
Bucketing for Sorted Lists		(no C API)icu::AlphabeticIndex
Text Boundary Analysis
(Break Iteration)		ubrk.hicu::BreakIterator
Regular Expressionsuregex.hicu::RegexPattern, icu::RegexMatcher
StringPrepusprep.hC API
International Domain Names in Applications:
+ * UTS #46 in C/C++, IDNA2003 only via C API		uidna.hidna.h
Identifier Spoofing & Confusabilityuspoof.hC API
Universal Time Scaleutmscale.hC API
Paragraph Layout / Complex Text Layoutplayout.hicu::ParagraphLayout
ICU I/Oustdio.hustream.h
+ * This main page is generated from docmain.h + */ diff --git a/intl/icu/source/common/unicode/dtintrv.h b/intl/icu/source/common/unicode/dtintrv.h new file mode 100644 index 0000000000..8c172eb7a5 --- /dev/null +++ b/intl/icu/source/common/unicode/dtintrv.h @@ -0,0 +1,164 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2008-2009, International Business Machines Corporation and +* others. All Rights Reserved. +******************************************************************************* +* +* File DTINTRV.H +* +******************************************************************************* +*/ + +#ifndef __DTINTRV_H__ +#define __DTINTRV_H__ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/uobject.h" + +/** + * \file + * \brief C++ API: Date Interval data type + */ + +U_NAMESPACE_BEGIN + + +/** + * This class represents a date interval. + * It is a pair of UDate representing from UDate 1 to UDate 2. + * @stable ICU 4.0 +**/ +class U_COMMON_API DateInterval : public UObject { +public: + + /** + * Construct a DateInterval given a from date and a to date. + * @param fromDate The from date in date interval. + * @param toDate The to date in date interval. + * @stable ICU 4.0 + */ + DateInterval(UDate fromDate, UDate toDate); + + /** + * destructor + * @stable ICU 4.0 + */ + virtual ~DateInterval(); + + /** + * Get the from date. + * @return the from date in dateInterval. + * @stable ICU 4.0 + */ + inline UDate getFromDate() const; + + /** + * Get the to date. + * @return the to date in dateInterval. + * @stable ICU 4.0 + */ + inline UDate getToDate() const; + + + /** + * Return the class ID for this class. This is useful only for comparing to + * a return value from getDynamicClassID(). For example: + * 
+     * .   Base* polymorphic_pointer = createPolymorphicObject();
+     * .   if (polymorphic_pointer->getDynamicClassID() ==
+     * .       derived::getStaticClassID()) ...
+     *
+ * @return The class ID for all objects of this class. + * @stable ICU 4.0 + */ + static UClassID U_EXPORT2 getStaticClassID(void); + + /** + * Returns a unique class ID POLYMORPHICALLY. Pure virtual override. This + * method is to implement a simple version of RTTI, since not all C++ + * compilers support genuine RTTI. Polymorphic operator==() and clone() + * methods call this method. + * + * @return The class ID for this object. All objects of a + * given class have the same class ID. Objects of + * other classes have different class IDs. + * @stable ICU 4.0 + */ + virtual UClassID getDynamicClassID(void) const override; + + + /** + * Copy constructor. + * @stable ICU 4.0 + */ + DateInterval(const DateInterval& other); + + /** + * Default assignment operator + * @stable ICU 4.0 + */ + DateInterval& operator=(const DateInterval&); + + /** + * Equality operator. + * @return true if the two DateIntervals are the same + * @stable ICU 4.0 + */ + virtual bool operator==(const DateInterval& other) const; + + /** + * Non-equality operator + * @return true if the two DateIntervals are not the same + * @stable ICU 4.0 + */ + inline bool operator!=(const DateInterval& other) const; + + + /** + * clone this object. + * The caller owns the result and should delete it when done. + * @return a cloned DateInterval + * @stable ICU 4.0 + */ + virtual DateInterval* clone() const; + +private: + /** + * Default constructor, not implemented. + */ + DateInterval() = delete; + + UDate fromDate; + UDate toDate; + +} ;// end class DateInterval + + +inline UDate +DateInterval::getFromDate() const { + return fromDate; +} + + +inline UDate +DateInterval::getToDate() const { + return toDate; +} + + +inline bool +DateInterval::operator!=(const DateInterval& other) const { + return ( !operator==(other) ); +} + + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/edits.h b/intl/icu/source/common/unicode/edits.h new file mode 100644 index 0000000000..dda9d3ca75 --- /dev/null +++ b/intl/icu/source/common/unicode/edits.h @@ -0,0 +1,531 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// edits.h +// created: 2016dec30 Markus W. Scherer + +#ifndef __EDITS_H__ +#define __EDITS_H__ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/uobject.h" + +/** + * \file + * \brief C++ API: C++ class Edits for low-level string transformations on styled text. + */ + +U_NAMESPACE_BEGIN + +class UnicodeString; + +/** + * Records lengths of string edits but not replacement text. Supports replacements, insertions, deletions + * in linear progression. Does not support moving/reordering of text. + * + * There are two types of edits: change edits and no-change edits. Add edits to + * instances of this class using {@link #addReplace(int32_t, int32_t)} (for change edits) and + * {@link #addUnchanged(int32_t)} (for no-change edits). Change edits are retained with full granularity, + * whereas adjacent no-change edits are always merged together. In no-change edits, there is a one-to-one + * mapping between code points in the source and destination strings. + * + * After all edits have been added, instances of this class should be considered immutable, and an + * {@link Edits::Iterator} can be used for queries. + * + * There are four flavors of Edits::Iterator: + * + *
    + *
  • {@link #getFineIterator()} retains full granularity of change edits. + *
  • {@link #getFineChangesIterator()} retains full granularity of change edits, and when calling + * next() on the iterator, skips over no-change edits (unchanged regions). + *
  • {@link #getCoarseIterator()} treats adjacent change edits as a single edit. (Adjacent no-change + * edits are automatically merged during the construction phase.) + *
  • {@link #getCoarseChangesIterator()} treats adjacent change edits as a single edit, and when + * calling next() on the iterator, skips over no-change edits (unchanged regions). + *
+ * + * For example, consider the string "abcßDeF", which case-folds to "abcssdef". This string has the + * following fine edits: + *
    + *
  • abc ⇨ abc (no-change) + *
  • ß ⇨ ss (change) + *
  • D ⇨ d (change) + *
  • e ⇨ e (no-change) + *
  • F ⇨ f (change) + *
+ * and the following coarse edits (note how adjacent change edits get merged together): + *
    + *
  • abc ⇨ abc (no-change) + *
  • ßD ⇨ ssd (change) + *
  • e ⇨ e (no-change) + *
  • F ⇨ f (change) + *
+ * + * The "fine changes" and "coarse changes" iterators will step through only the change edits when their + * `Edits::Iterator::next()` methods are called. They are identical to the non-change iterators when + * their `Edits::Iterator::findSourceIndex()` or `Edits::Iterator::findDestinationIndex()` + * methods are used to walk through the string. + * + * For examples of how to use this class, see the test `TestCaseMapEditsIteratorDocs` in + * UCharacterCaseTest.java. + * + * An Edits object tracks a separate UErrorCode, but ICU string transformation functions + * (e.g., case mapping functions) merge any such errors into their API's UErrorCode. + * + * @stable ICU 59 + */ +class U_COMMON_API Edits final : public UMemory { +public: + /** + * Constructs an empty object. + * @stable ICU 59 + */ + Edits() : + array(stackArray), capacity(STACK_CAPACITY), length(0), delta(0), numChanges(0), + errorCode_(U_ZERO_ERROR) {} + /** + * Copy constructor. + * @param other source edits + * @stable ICU 60 + */ + Edits(const Edits &other) : + array(stackArray), capacity(STACK_CAPACITY), length(other.length), + delta(other.delta), numChanges(other.numChanges), + errorCode_(other.errorCode_) { + copyArray(other); + } + /** + * Move constructor, might leave src empty. + * This object will have the same contents that the source object had. + * @param src source edits + * @stable ICU 60 + */ + Edits(Edits &&src) noexcept : + array(stackArray), capacity(STACK_CAPACITY), length(src.length), + delta(src.delta), numChanges(src.numChanges), + errorCode_(src.errorCode_) { + moveArray(src); + } + + /** + * Destructor. + * @stable ICU 59 + */ + ~Edits(); + + /** + * Assignment operator. + * @param other source edits + * @return *this + * @stable ICU 60 + */ + Edits &operator=(const Edits &other); + + /** + * Move assignment operator, might leave src empty. + * This object will have the same contents that the source object had. + * The behavior is undefined if *this and src are the same object. + * @param src source edits + * @return *this + * @stable ICU 60 + */ + Edits &operator=(Edits &&src) noexcept; + + /** + * Resets the data but may not release memory. + * @stable ICU 59 + */ + void reset() noexcept; + + /** + * Adds a no-change edit: a record for an unchanged segment of text. + * Normally called from inside ICU string transformation functions, not user code. + * @stable ICU 59 + */ + void addUnchanged(int32_t unchangedLength); + /** + * Adds a change edit: a record for a text replacement/insertion/deletion. + * Normally called from inside ICU string transformation functions, not user code. + * @stable ICU 59 + */ + void addReplace(int32_t oldLength, int32_t newLength); + /** + * Sets the UErrorCode if an error occurred while recording edits. + * Preserves older error codes in the outErrorCode. + * Normally called from inside ICU string transformation functions, not user code. + * @param outErrorCode Set to an error code if it does not contain one already + * and an error occurred while recording edits. + * Otherwise unchanged. + * @return true if U_FAILURE(outErrorCode) + * @stable ICU 59 + */ + UBool copyErrorTo(UErrorCode &outErrorCode) const; + + /** + * How much longer is the new text compared with the old text? + * @return new length minus old length + * @stable ICU 59 + */ + int32_t lengthDelta() const { return delta; } + /** + * @return true if there are any change edits + * @stable ICU 59 + */ + UBool hasChanges() const { return numChanges != 0; } + + /** + * @return the number of change edits + * @stable ICU 60 + */ + int32_t numberOfChanges() const { return numChanges; } + + /** + * Access to the list of edits. + * + * At any moment in time, an instance of this class points to a single edit: a "window" into a span + * of the source string and the corresponding span of the destination string. The source string span + * starts at {@link #sourceIndex()} and runs for {@link #oldLength()} chars; the destination string + * span starts at {@link #destinationIndex()} and runs for {@link #newLength()} chars. + * + * The iterator can be moved between edits using the `next()`, `findSourceIndex(int32_t, UErrorCode &)`, + * and `findDestinationIndex(int32_t, UErrorCode &)` methods. + * Calling any of these methods mutates the iterator to make it point to the corresponding edit. + * + * For more information, see the documentation for {@link Edits}. + * + * @see getCoarseIterator + * @see getFineIterator + * @stable ICU 59 + */ + struct U_COMMON_API Iterator final : public UMemory { + /** + * Default constructor, empty iterator. + * @stable ICU 60 + */ + Iterator() : + array(nullptr), index(0), length(0), + remaining(0), onlyChanges_(false), coarse(false), + dir(0), changed(false), oldLength_(0), newLength_(0), + srcIndex(0), replIndex(0), destIndex(0) {} + /** + * Copy constructor. + * @stable ICU 59 + */ + Iterator(const Iterator &other) = default; + /** + * Assignment operator. + * @stable ICU 59 + */ + Iterator &operator=(const Iterator &other) = default; + + /** + * Advances the iterator to the next edit. + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return true if there is another edit + * @stable ICU 59 + */ + UBool next(UErrorCode &errorCode) { return next(onlyChanges_, errorCode); } + + /** + * Moves the iterator to the edit that contains the source index. + * The source index may be found in a no-change edit + * even if normal iteration would skip no-change edits. + * Normal iteration can continue from a found edit. + * + * The iterator state before this search logically does not matter. + * (It may affect the performance of the search.) + * + * The iterator state after this search is undefined + * if the source index is out of bounds for the source string. + * + * @param i source index + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return true if the edit for the source index was found + * @stable ICU 59 + */ + UBool findSourceIndex(int32_t i, UErrorCode &errorCode) { + return findIndex(i, true, errorCode) == 0; + } + + /** + * Moves the iterator to the edit that contains the destination index. + * The destination index may be found in a no-change edit + * even if normal iteration would skip no-change edits. + * Normal iteration can continue from a found edit. + * + * The iterator state before this search logically does not matter. + * (It may affect the performance of the search.) + * + * The iterator state after this search is undefined + * if the source index is out of bounds for the source string. + * + * @param i destination index + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return true if the edit for the destination index was found + * @stable ICU 60 + */ + UBool findDestinationIndex(int32_t i, UErrorCode &errorCode) { + return findIndex(i, false, errorCode) == 0; + } + + /** + * Computes the destination index corresponding to the given source index. + * If the source index is inside a change edit (not at its start), + * then the destination index at the end of that edit is returned, + * since there is no information about index mapping inside a change edit. + * + * (This means that indexes to the start and middle of an edit, + * for example around a grapheme cluster, are mapped to indexes + * encompassing the entire edit. + * The alternative, mapping an interior index to the start, + * would map such an interval to an empty one.) + * + * This operation will usually but not always modify this object. + * The iterator state after this search is undefined. + * + * @param i source index + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return destination index; undefined if i is not 0..string length + * @stable ICU 60 + */ + int32_t destinationIndexFromSourceIndex(int32_t i, UErrorCode &errorCode); + + /** + * Computes the source index corresponding to the given destination index. + * If the destination index is inside a change edit (not at its start), + * then the source index at the end of that edit is returned, + * since there is no information about index mapping inside a change edit. + * + * (This means that indexes to the start and middle of an edit, + * for example around a grapheme cluster, are mapped to indexes + * encompassing the entire edit. + * The alternative, mapping an interior index to the start, + * would map such an interval to an empty one.) + * + * This operation will usually but not always modify this object. + * The iterator state after this search is undefined. + * + * @param i destination index + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return source index; undefined if i is not 0..string length + * @stable ICU 60 + */ + int32_t sourceIndexFromDestinationIndex(int32_t i, UErrorCode &errorCode); + + /** + * Returns whether the edit currently represented by the iterator is a change edit. + * + * @return true if this edit replaces oldLength() units with newLength() different ones. + * false if oldLength units remain unchanged. + * @stable ICU 59 + */ + UBool hasChange() const { return changed; } + + /** + * The length of the current span in the source string, which starts at {@link #sourceIndex}. + * + * @return the number of units in the original string which are replaced or remain unchanged. + * @stable ICU 59 + */ + int32_t oldLength() const { return oldLength_; } + + /** + * The length of the current span in the destination string, which starts at + * {@link #destinationIndex}, or in the replacement string, which starts at + * {@link #replacementIndex}. + * + * @return the number of units in the modified string, if hasChange() is true. + * Same as oldLength if hasChange() is false. + * @stable ICU 59 + */ + int32_t newLength() const { return newLength_; } + + /** + * The start index of the current span in the source string; the span has length + * {@link #oldLength}. + * + * @return the current index into the source string + * @stable ICU 59 + */ + int32_t sourceIndex() const { return srcIndex; } + + /** + * The start index of the current span in the replacement string; the span has length + * {@link #newLength}. Well-defined only if the current edit is a change edit. + * + * The *replacement string* is the concatenation of all substrings of the destination + * string corresponding to change edits. + * + * This method is intended to be used together with operations that write only replacement + * characters (e.g. operations specifying the \ref U_OMIT_UNCHANGED_TEXT option). + * The source string can then be modified in-place. + * + * @return the current index into the replacement-characters-only string, + * not counting unchanged spans + * @stable ICU 59 + */ + int32_t replacementIndex() const { + // TODO: Throw an exception if we aren't in a change edit? + return replIndex; + } + + /** + * The start index of the current span in the destination string; the span has length + * {@link #newLength}. + * + * @return the current index into the full destination string + * @stable ICU 59 + */ + int32_t destinationIndex() const { return destIndex; } + +#ifndef U_HIDE_INTERNAL_API + /** + * A string representation of the current edit represented by the iterator for debugging. You + * should not depend on the contents of the return string. + * @internal + */ + UnicodeString& toString(UnicodeString& appendTo) const; +#endif // U_HIDE_INTERNAL_API + + private: + friend class Edits; + + Iterator(const uint16_t *a, int32_t len, UBool oc, UBool crs); + + int32_t readLength(int32_t head); + void updateNextIndexes(); + void updatePreviousIndexes(); + UBool noNext(); + UBool next(UBool onlyChanges, UErrorCode &errorCode); + UBool previous(UErrorCode &errorCode); + /** @return -1: error or i<0; 0: found; 1: i>=string length */ + int32_t findIndex(int32_t i, UBool findSource, UErrorCode &errorCode); + + const uint16_t *array; + int32_t index, length; + // 0 if we are not within compressed equal-length changes. + // Otherwise the number of remaining changes, including the current one. + int32_t remaining; + UBool onlyChanges_, coarse; + + int8_t dir; // iteration direction: back(<0), initial(0), forward(>0) + UBool changed; + int32_t oldLength_, newLength_; + int32_t srcIndex, replIndex, destIndex; + }; + + /** + * Returns an Iterator for coarse-grained change edits + * (adjacent change edits are treated as one). + * Can be used to perform simple string updates. + * Skips no-change edits. + * @return an Iterator that merges adjacent changes. + * @stable ICU 59 + */ + Iterator getCoarseChangesIterator() const { + return Iterator(array, length, true, true); + } + + /** + * Returns an Iterator for coarse-grained change and no-change edits + * (adjacent change edits are treated as one). + * Can be used to perform simple string updates. + * Adjacent change edits are treated as one edit. + * @return an Iterator that merges adjacent changes. + * @stable ICU 59 + */ + Iterator getCoarseIterator() const { + return Iterator(array, length, false, true); + } + + /** + * Returns an Iterator for fine-grained change edits + * (full granularity of change edits is retained). + * Can be used for modifying styled text. + * Skips no-change edits. + * @return an Iterator that separates adjacent changes. + * @stable ICU 59 + */ + Iterator getFineChangesIterator() const { + return Iterator(array, length, true, false); + } + + /** + * Returns an Iterator for fine-grained change and no-change edits + * (full granularity of change edits is retained). + * Can be used for modifying styled text. + * @return an Iterator that separates adjacent changes. + * @stable ICU 59 + */ + Iterator getFineIterator() const { + return Iterator(array, length, false, false); + } + + /** + * Merges the two input Edits and appends the result to this object. + * + * Consider two string transformations (for example, normalization and case mapping) + * where each records Edits in addition to writing an output string.
+ * Edits ab reflect how substrings of input string a + * map to substrings of intermediate string b.
+ * Edits bc reflect how substrings of intermediate string b + * map to substrings of output string c.
+ * This function merges ab and bc such that the additional edits + * recorded in this object reflect how substrings of input string a + * map to substrings of output string c. + * + * If unrelated Edits are passed in where the output string of the first + * has a different length than the input string of the second, + * then a U_ILLEGAL_ARGUMENT_ERROR is reported. + * + * @param ab reflects how substrings of input string a + * map to substrings of intermediate string b. + * @param bc reflects how substrings of intermediate string b + * map to substrings of output string c. + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return *this, with the merged edits appended + * @stable ICU 60 + */ + Edits &mergeAndAppend(const Edits &ab, const Edits &bc, UErrorCode &errorCode); + +private: + void releaseArray() noexcept; + Edits ©Array(const Edits &other); + Edits &moveArray(Edits &src) noexcept; + + void setLastUnit(int32_t last) { array[length - 1] = (uint16_t)last; } + int32_t lastUnit() const { return length > 0 ? array[length - 1] : 0xffff; } + + void append(int32_t r); + UBool growArray(); + + static const int32_t STACK_CAPACITY = 100; + uint16_t *array; + int32_t capacity; + int32_t length; + int32_t delta; + int32_t numChanges; + UErrorCode errorCode_; + uint16_t stackArray[STACK_CAPACITY]; +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __EDITS_H__ diff --git a/intl/icu/source/common/unicode/enumset.h b/intl/icu/source/common/unicode/enumset.h new file mode 100644 index 0000000000..bde8c455c0 --- /dev/null +++ b/intl/icu/source/common/unicode/enumset.h @@ -0,0 +1,69 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2012,2014 International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +*/ + +/** + * \file + * \brief C++: internal template EnumSet<> + */ + +#ifndef ENUMSET_H +#define ENUMSET_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/* Can't use #ifndef U_HIDE_INTERNAL_API for the entire EnumSet class, needed in .h file declarations */ +/** + * enum bitset for boolean fields. Similar to Java EnumSet<>. + * Needs to range check. Used for private instance variables. + * @internal + * \cond + */ +template +class EnumSet { +public: + inline EnumSet() : fBools(0) {} + inline EnumSet(const EnumSet& other) : fBools(other.fBools) {} + inline ~EnumSet() {} +#ifndef U_HIDE_INTERNAL_API + inline void clear() { fBools=0; } + inline void add(T toAdd) { set(toAdd, 1); } + inline void remove(T toRemove) { set(toRemove, 0); } + inline int32_t contains(T toCheck) const { return get(toCheck); } + inline void set(T toSet, int32_t v) { fBools=(fBools&(~flag(toSet)))|(v?(flag(toSet)):0); } + inline int32_t get(T toCheck) const { return (fBools & flag(toCheck))?1:0; } + inline UBool isValidEnum(T toCheck) const { return (toCheck>=minValue&&toCheck& operator=(const EnumSet& other) { + fBools = other.fBools; + return *this; + } + + inline uint32_t getAll() const { + return fBools; + } +#endif /* U_HIDE_INTERNAL_API */ + +private: + inline uint32_t flag(T toCheck) const { return (1<<(toCheck-minValue)); } +private: + uint32_t fBools; +}; + +/** \endcond */ + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ +#endif /* ENUMSET_H */ diff --git a/intl/icu/source/common/unicode/errorcode.h b/intl/icu/source/common/unicode/errorcode.h new file mode 100644 index 0000000000..fe7b518323 --- /dev/null +++ b/intl/icu/source/common/unicode/errorcode.h @@ -0,0 +1,144 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2009-2011, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: errorcode.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2009mar10 +* created by: Markus W. Scherer +*/ + +#ifndef __ERRORCODE_H__ +#define __ERRORCODE_H__ + +/** + * \file + * \brief C++ API: ErrorCode class intended to make it easier to use + * ICU C and C++ APIs from C++ user code. + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/uobject.h" + +U_NAMESPACE_BEGIN + +/** + * Wrapper class for UErrorCode, with conversion operators for direct use + * in ICU C and C++ APIs. + * Intended to be used as a base class, where a subclass overrides + * the handleFailure() function so that it throws an exception, + * does an assert(), logs an error, etc. + * This is not an abstract base class. This class can be used and instantiated + * by itself, although it will be more useful when subclassed. + * + * Features: + * - The constructor initializes the internal UErrorCode to U_ZERO_ERROR, + * removing one common source of errors. + * - Same use in C APIs taking a UErrorCode * (pointer) + * and C++ taking UErrorCode & (reference) via conversion operators. + * - Possible automatic checking for success when it goes out of scope. + * + * Note: For automatic checking for success in the destructor, a subclass + * must implement such logic in its own destructor because the base class + * destructor cannot call a subclass function (like handleFailure()). + * The ErrorCode base class destructor does nothing. + * + * Note also: While it is possible for a destructor to throw an exception, + * it is generally unsafe to do so. This means that in a subclass the destructor + * and the handleFailure() function may need to take different actions. + * + * Sample code: + * \code + * class IcuErrorCode: public icu::ErrorCode { + * public: + * virtual ~IcuErrorCode() { // should be defined in .cpp as "key function" + * // Safe because our handleFailure() does not throw exceptions. + * if(isFailure()) { handleFailure(); } + * } + * protected: + * virtual void handleFailure() const { + * log_failure(u_errorName(errorCode)); + * exit(errorCode); + * } + * }; + * IcuErrorCode error_code; + * UConverter *cnv = ucnv_open("Shift-JIS", error_code); + * length = ucnv_fromUChars(dest, capacity, src, length, error_code); + * ucnv_close(cnv); + * // IcuErrorCode destructor checks for success. + * \endcode + * + * @stable ICU 4.2 + */ +class U_COMMON_API ErrorCode: public UMemory { +public: + /** + * Default constructor. Initializes its UErrorCode to U_ZERO_ERROR. + * @stable ICU 4.2 + */ + ErrorCode() : errorCode(U_ZERO_ERROR) {} + /** Destructor, does nothing. See class documentation for details. @stable ICU 4.2 */ + virtual ~ErrorCode(); + /** Conversion operator, returns a reference. @stable ICU 4.2 */ + operator UErrorCode & () { return errorCode; } + /** Conversion operator, returns a pointer. @stable ICU 4.2 */ + operator UErrorCode * () { return &errorCode; } + /** Tests for U_SUCCESS(). @stable ICU 4.2 */ + UBool isSuccess() const { return U_SUCCESS(errorCode); } + /** Tests for U_FAILURE(). @stable ICU 4.2 */ + UBool isFailure() const { return U_FAILURE(errorCode); } + /** Returns the UErrorCode value. @stable ICU 4.2 */ + UErrorCode get() const { return errorCode; } + /** Sets the UErrorCode value. @stable ICU 4.2 */ + void set(UErrorCode value) { errorCode=value; } + /** Returns the UErrorCode value and resets it to U_ZERO_ERROR. @stable ICU 4.2 */ + UErrorCode reset(); + /** + * Asserts isSuccess(). + * In other words, this method checks for a failure code, + * and the base class handles it like this: + * \code + * if(isFailure()) { handleFailure(); } + * \endcode + * @stable ICU 4.4 + */ + void assertSuccess() const; + /** + * Return a string for the UErrorCode value. + * The string will be the same as the name of the error code constant + * in the UErrorCode enum. + * @stable ICU 4.4 + */ + const char* errorName() const; + +protected: + /** + * Internal UErrorCode, accessible to subclasses. + * @stable ICU 4.2 + */ + UErrorCode errorCode; + /** + * Called by assertSuccess() if isFailure() is true. + * A subclass should override this function to deal with a failure code: + * Throw an exception, log an error, terminate the program, or similar. + * @stable ICU 4.2 + */ + virtual void handleFailure() const {} +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __ERRORCODE_H__ diff --git a/intl/icu/source/common/unicode/filteredbrk.h b/intl/icu/source/common/unicode/filteredbrk.h new file mode 100644 index 0000000000..8b07e39ae1 --- /dev/null +++ b/intl/icu/source/common/unicode/filteredbrk.h @@ -0,0 +1,152 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************** +* Copyright (C) 1997-2015, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************** +*/ + +#ifndef FILTEREDBRK_H +#define FILTEREDBRK_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/brkiter.h" + +#if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION + +U_NAMESPACE_BEGIN + +/** + * \file + * \brief C++ API: FilteredBreakIteratorBuilder + */ + +/** + * The BreakIteratorFilter is used to modify the behavior of a BreakIterator + * by constructing a new BreakIterator which suppresses certain segment boundaries. + * See http://www.unicode.org/reports/tr35/tr35-general.html#Segmentation_Exceptions . + * For example, a typical English Sentence Break Iterator would break on the space + * in the string "Mr. Smith" (resulting in two segments), + * but with "Mr." as an exception, a filtered break iterator + * would consider the string "Mr. Smith" to be a single segment. + * + * @stable ICU 56 + */ +class U_COMMON_API FilteredBreakIteratorBuilder : public UObject { + public: + /** + * destructor. + * @stable ICU 56 + */ + virtual ~FilteredBreakIteratorBuilder(); + + /** + * Construct a FilteredBreakIteratorBuilder based on rules in a locale. + * The rules are taken from CLDR exception data for the locale, + * see http://www.unicode.org/reports/tr35/tr35-general.html#Segmentation_Exceptions + * This is the equivalent of calling createInstance(UErrorCode&) + * and then repeatedly calling addNoBreakAfter(...) with the contents + * of the CLDR exception data. + * @param where the locale. + * @param status The error code. + * @return the new builder + * @stable ICU 56 + */ + static FilteredBreakIteratorBuilder *createInstance(const Locale& where, UErrorCode& status); + +#ifndef U_HIDE_DEPRECATED_API + /** + * This function has been deprecated in favor of createEmptyInstance, which has + * identical behavior. + * @param status The error code. + * @return the new builder + * @deprecated ICU 60 use createEmptyInstance instead + * @see createEmptyInstance() + */ + static FilteredBreakIteratorBuilder *createInstance(UErrorCode &status); +#endif /* U_HIDE_DEPRECATED_API */ + + /** + * Construct an empty FilteredBreakIteratorBuilder. + * In this state, it will not suppress any segment boundaries. + * @param status The error code. + * @return the new builder + * @stable ICU 60 + */ + static FilteredBreakIteratorBuilder *createEmptyInstance(UErrorCode &status); + + /** + * Suppress a certain string from being the end of a segment. + * For example, suppressing "Mr.", then segments ending in "Mr." will not be returned + * by the iterator. + * @param string the string to suppress, such as "Mr." + * @param status error code + * @return returns true if the string was not present and now added, + * false if the call was a no-op because the string was already being suppressed. + * @stable ICU 56 + */ + virtual UBool suppressBreakAfter(const UnicodeString& string, UErrorCode& status) = 0; + + /** + * Stop suppressing a certain string from being the end of the segment. + * This function does not create any new segment boundaries, but only serves to un-do + * the effect of earlier calls to suppressBreakAfter, or to un-do the effect of + * locale data which may be suppressing certain strings. + * @param string the exception to remove + * @param status error code + * @return returns true if the string was present and now removed, + * false if the call was a no-op because the string was not being suppressed. + * @stable ICU 56 + */ + virtual UBool unsuppressBreakAfter(const UnicodeString& string, UErrorCode& status) = 0; + +#ifndef U_FORCE_HIDE_DEPRECATED_API + /** + * This function has been deprecated in favor of wrapIteratorWithFilter() + * The behavior is identical. + * @param adoptBreakIterator the break iterator to adopt + * @param status error code + * @return the new BreakIterator, owned by the caller. + * @deprecated ICU 60 use wrapIteratorWithFilter() instead + * @see wrapBreakIteratorWithFilter() + */ + virtual BreakIterator *build(BreakIterator* adoptBreakIterator, UErrorCode& status) = 0; +#endif // U_FORCE_HIDE_DEPRECATED_API + + /** + * Wrap (adopt) an existing break iterator in a new filtered instance. + * The resulting BreakIterator is owned by the caller. + * The BreakIteratorFilter may be destroyed before the BreakIterator is destroyed. + * Note that the adoptBreakIterator is adopted by the new BreakIterator + * and should no longer be used by the caller. + * The FilteredBreakIteratorBuilder may be reused. + * This function is an alias for build() + * @param adoptBreakIterator the break iterator to adopt + * @param status error code + * @return the new BreakIterator, owned by the caller. + * @stable ICU 60 + */ + inline BreakIterator *wrapIteratorWithFilter(BreakIterator* adoptBreakIterator, UErrorCode& status) { + return build(adoptBreakIterator, status); + } + + protected: + /** + * For subclass use + * @stable ICU 56 + */ + FilteredBreakIteratorBuilder(); +}; + + +U_NAMESPACE_END + +#endif // #if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // #ifndef FILTEREDBRK_H diff --git a/intl/icu/source/common/unicode/icudataver.h b/intl/icu/source/common/unicode/icudataver.h new file mode 100644 index 0000000000..f218ed8ebc --- /dev/null +++ b/intl/icu/source/common/unicode/icudataver.h @@ -0,0 +1,43 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2009-2013, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +*/ + + +/** + * \file + * \brief C API: access to ICU Data Version number + */ + +#ifndef __ICU_DATA_VER_H__ +#define __ICU_DATA_VER_H__ + +#include "unicode/utypes.h" + +/** + * @stable ICU 49 + */ +#define U_ICU_VERSION_BUNDLE "icuver" + +/** + * @stable ICU 49 + */ +#define U_ICU_DATA_KEY "DataVersion" + +/** + * Retrieves the data version from icuver and stores it in dataVersionFillin. + * + * @param dataVersionFillin icuver data version information to be filled in if not-null + * @param status stores the error code from the calls to resource bundle + * + * @stable ICU 49 + */ +U_CAPI void U_EXPORT2 u_getDataVersion(UVersionInfo dataVersionFillin, UErrorCode *status); + +#endif diff --git a/intl/icu/source/common/unicode/icuplug.h b/intl/icu/source/common/unicode/icuplug.h new file mode 100644 index 0000000000..c787fcd426 --- /dev/null +++ b/intl/icu/source/common/unicode/icuplug.h @@ -0,0 +1,391 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2009-2015, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* FILE NAME : icuplug.h +* +* Date Name Description +* 10/29/2009 sl New. +****************************************************************************** +*/ + +/** + * \file + * \brief C API: ICU Plugin API + * + *

C API: ICU Plugin API

+ * + *

C API allowing run-time loadable modules that extend or modify ICU functionality.

+ * + *

Loading and Configuration

+ * + *

At ICU startup time, the environment variable "ICU_PLUGINS" will be + * queried for a directory name. If it is not set, the preprocessor symbol + * "DEFAULT_ICU_PLUGINS" will be checked for a default value.

+ * + *

Within the above-named directory, the file "icuplugins##.txt" will be + * opened, if present, where ## is the major+minor number of the currently + * running ICU (such as, 44 for ICU 4.4, thus icuplugins44.txt)

+ * + *

The configuration file has this format:

+ * + *
    + *
  • Hash (#) begins a comment line
    • + * + *
  • Non-comment lines have two or three components: + * LIBRARYNAME ENTRYPOINT [ CONFIGURATION .. ]
    • + * + *
  • Tabs or spaces separate the three items.
    • + * + *
  • LIBRARYNAME is the name of a shared library, either a short name if + * it is on the loader path, or a full pathname.
    • + * + *
  • ENTRYPOINT is the short (undecorated) symbol name of the plugin's + * entrypoint, as above.
    • + * + *
  • CONFIGURATION is the entire rest of the line . It's passed as-is to + * the plugin.
    • + *
+ * + *

An example configuration file is, in its entirety:

+ * + * \code + * # this is icuplugins44.txt + * testplug.dll myPlugin hello=world + * \endcode + *

Plugins are categorized as "high" or "low" level. Low level are those + * which must be run BEFORE high level plugins, and before any operations + * which cause ICU to be 'initialized'. If a plugin is low level but + * causes ICU to allocate memory or become initialized, that plugin is said + * to cause a 'level change'.

+ * + *

At load time, ICU first queries all plugins to determine their level, + * then loads all 'low' plugins first, and then loads all 'high' plugins. + * Plugins are otherwise loaded in the order listed in the configuration file.

+ * + *

Implementing a Plugin

+ * \code + * U_CAPI UPlugTokenReturn U_EXPORT2 + * myPlugin (UPlugData *plug, UPlugReason reason, UErrorCode *status) { + * if(reason==UPLUG_REASON_QUERY) { + * uplug_setPlugName(plug, "Simple Plugin"); + * uplug_setPlugLevel(plug, UPLUG_LEVEL_HIGH); + * } else if(reason==UPLUG_REASON_LOAD) { + * ... Set up some ICU things here.... + * } else if(reason==UPLUG_REASON_UNLOAD) { + * ... unload, clean up ... + * } + * return UPLUG_TOKEN; + * } + * \endcode + * + *

The UPlugData* is an opaque pointer to the plugin-specific data, and is + * used in all other API calls.

+ * + *

The API contract is:

+ *
  1. The plugin MUST always return UPLUG_TOKEN as a return value- to + * indicate that it is a valid plugin.
    2. + * + *
  2. When the 'reason' parameter is set to UPLUG_REASON_QUERY, the + * plugin MUST call uplug_setPlugLevel() to indicate whether it is a high + * level or low level plugin.
    3. + * + *
  3. When the 'reason' parameter is UPLUG_REASON_QUERY, the plugin + * SHOULD call uplug_setPlugName to indicate a human readable plugin name.
+ * + * + * \internal ICU 4.4 Technology Preview + */ + + +#ifndef ICUPLUG_H +#define ICUPLUG_H + +#include "unicode/utypes.h" + + +#if UCONFIG_ENABLE_PLUGINS || defined(U_IN_DOXYGEN) + + + +/* === Basic types === */ + +#ifndef U_HIDE_INTERNAL_API +struct UPlugData; +/** + * @{ + * Typedef for opaque structure passed to/from a plugin. + * Use the APIs to access it. + * @internal ICU 4.4 Technology Preview + */ +typedef struct UPlugData UPlugData; + +/** @} */ + +/** + * Random Token to identify a valid ICU plugin. Plugins must return this + * from the entrypoint. + * @internal ICU 4.4 Technology Preview + */ +#define UPLUG_TOKEN 0x54762486 + +/** + * Max width of names, symbols, and configuration strings + * @internal ICU 4.4 Technology Preview + */ +#define UPLUG_NAME_MAX 100 + + +/** + * Return value from a plugin entrypoint. + * Must always be set to UPLUG_TOKEN + * @see UPLUG_TOKEN + * @internal ICU 4.4 Technology Preview + */ +typedef uint32_t UPlugTokenReturn; + +/** + * Reason code for the entrypoint's call + * @internal ICU 4.4 Technology Preview + */ +typedef enum { + UPLUG_REASON_QUERY = 0, /**< The plugin is being queried for info. **/ + UPLUG_REASON_LOAD = 1, /**< The plugin is being loaded. **/ + UPLUG_REASON_UNLOAD = 2, /**< The plugin is being unloaded. **/ + /** + * Number of known reasons. + * @internal The numeric value may change over time, see ICU ticket #12420. + */ + UPLUG_REASON_COUNT +} UPlugReason; + + +/** + * Level of plugin loading + * INITIAL: UNKNOWN + * QUERY: INVALID -> { LOW | HIGH } + * ERR -> INVALID + * @internal ICU 4.4 Technology Preview + */ +typedef enum { + UPLUG_LEVEL_INVALID = 0, /**< The plugin is invalid, hasn't called uplug_setLevel, or can't load. **/ + UPLUG_LEVEL_UNKNOWN = 1, /**< The plugin is waiting to be installed. **/ + UPLUG_LEVEL_LOW = 2, /**< The plugin must be called before u_init completes **/ + UPLUG_LEVEL_HIGH = 3, /**< The plugin can run at any time. **/ + /** + * Number of known levels. + * @internal The numeric value may change over time, see ICU ticket #12420. + */ + UPLUG_LEVEL_COUNT +} UPlugLevel; + +/** + * Entrypoint for an ICU plugin. + * @param plug the UPlugData handle. + * @param reason the reason code for the entrypoint's call. + * @param status Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return A valid plugin must return UPLUG_TOKEN + * @internal ICU 4.4 Technology Preview + */ +typedef UPlugTokenReturn (U_EXPORT2 UPlugEntrypoint) ( + UPlugData *plug, + UPlugReason reason, + UErrorCode *status); + +/* === Needed for Implementing === */ + +/** + * Request that this plugin not be unloaded at cleanup time. + * This is appropriate for plugins which cannot be cleaned up. + * @see u_cleanup() + * @param plug plugin + * @param dontUnload set true if this plugin can't be unloaded + * @internal ICU 4.4 Technology Preview + */ +U_CAPI void U_EXPORT2 +uplug_setPlugNoUnload(UPlugData *plug, UBool dontUnload); + +/** + * Set the level of this plugin. + * @param plug plugin data handle + * @param level the level of this plugin + * @internal ICU 4.4 Technology Preview + */ +U_CAPI void U_EXPORT2 +uplug_setPlugLevel(UPlugData *plug, UPlugLevel level); + +/** + * Get the level of this plugin. + * @param plug plugin data handle + * @return the level of this plugin + * @internal ICU 4.4 Technology Preview + */ +U_CAPI UPlugLevel U_EXPORT2 +uplug_getPlugLevel(UPlugData *plug); + +/** + * Get the lowest level of plug which can currently load. + * For example, if UPLUG_LEVEL_LOW is returned, then low level plugins may load + * if UPLUG_LEVEL_HIGH is returned, then only high level plugins may load. + * @return the lowest level of plug which can currently load + * @internal ICU 4.4 Technology Preview + */ +U_CAPI UPlugLevel U_EXPORT2 +uplug_getCurrentLevel(void); + + +/** + * Get plug load status + * @return The error code of this plugin's load attempt. + * @internal ICU 4.4 Technology Preview + */ +U_CAPI UErrorCode U_EXPORT2 +uplug_getPlugLoadStatus(UPlugData *plug); + +/** + * Set the human-readable name of this plugin. + * @param plug plugin data handle + * @param name the name of this plugin. The first UPLUG_NAME_MAX characters willi be copied into a new buffer. + * @internal ICU 4.4 Technology Preview + */ +U_CAPI void U_EXPORT2 +uplug_setPlugName(UPlugData *plug, const char *name); + +/** + * Get the human-readable name of this plugin. + * @param plug plugin data handle + * @return the name of this plugin + * @internal ICU 4.4 Technology Preview + */ +U_CAPI const char * U_EXPORT2 +uplug_getPlugName(UPlugData *plug); + +/** + * Return the symbol name for this plugin, if known. + * @param plug plugin data handle + * @return the symbol name, or NULL + * @internal ICU 4.4 Technology Preview + */ +U_CAPI const char * U_EXPORT2 +uplug_getSymbolName(UPlugData *plug); + +/** + * Return the library name for this plugin, if known. + * @param plug plugin data handle + * @param status error code + * @return the library name, or NULL + * @internal ICU 4.4 Technology Preview + */ +U_CAPI const char * U_EXPORT2 +uplug_getLibraryName(UPlugData *plug, UErrorCode *status); + +/** + * Return the library used for this plugin, if known. + * Plugins could use this to load data out of their + * @param plug plugin data handle + * @return the library, or NULL + * @internal ICU 4.4 Technology Preview + */ +U_CAPI void * U_EXPORT2 +uplug_getLibrary(UPlugData *plug); + +/** + * Return the plugin-specific context data. + * @param plug plugin data handle + * @return the context, or NULL if not set + * @internal ICU 4.4 Technology Preview + */ +U_CAPI void * U_EXPORT2 +uplug_getContext(UPlugData *plug); + +/** + * Set the plugin-specific context data. + * @param plug plugin data handle + * @param context new context to set + * @internal ICU 4.4 Technology Preview + */ +U_CAPI void U_EXPORT2 +uplug_setContext(UPlugData *plug, void *context); + + +/** + * Get the configuration string, if available. + * The string is in the platform default codepage. + * @param plug plugin data handle + * @return configuration string, or else null. + * @internal ICU 4.4 Technology Preview + */ +U_CAPI const char * U_EXPORT2 +uplug_getConfiguration(UPlugData *plug); + +/** + * Return all currently installed plugins, from newest to oldest + * Usage Example: + * \code + * UPlugData *plug = NULL; + * while(plug=uplug_nextPlug(plug)) { + * ... do something with 'plug' ... + * } + * \endcode + * Not thread safe- do not call while plugs are added or removed. + * @param prior pass in 'NULL' to get the first (most recent) plug, + * otherwise pass the value returned on a prior call to uplug_nextPlug + * @return the next oldest plugin, or NULL if no more. + * @internal ICU 4.4 Technology Preview + */ +U_CAPI UPlugData* U_EXPORT2 +uplug_nextPlug(UPlugData *prior); + +/** + * Inject a plugin as if it were loaded from a library. + * This is useful for testing plugins. + * Note that it will have a 'NULL' library pointer associated + * with it, and therefore no llibrary will be closed at cleanup time. + * Low level plugins may not be able to load, as ordering can't be enforced. + * @param entrypoint entrypoint to install + * @param config user specified configuration string, if available, or NULL. + * @param status error result + * @return the new UPlugData associated with this plugin, or NULL if error. + * @internal ICU 4.4 Technology Preview + */ +U_CAPI UPlugData* U_EXPORT2 +uplug_loadPlugFromEntrypoint(UPlugEntrypoint *entrypoint, const char *config, UErrorCode *status); + + +/** + * Inject a plugin from a library, as if the information came from a config file. + * Low level plugins may not be able to load, and ordering can't be enforced. + * @param libName DLL name to load + * @param sym symbol of plugin (UPlugEntrypoint function) + * @param config configuration string, or NULL + * @param status error result + * @return the new UPlugData associated with this plugin, or NULL if error. + * @internal ICU 4.4 Technology Preview + */ +U_CAPI UPlugData* U_EXPORT2 +uplug_loadPlugFromLibrary(const char *libName, const char *sym, const char *config, UErrorCode *status); + +/** + * Remove a plugin. + * Will request the plugin to be unloaded, and close the library if needed + * @param plug plugin handle to close + * @param status error result + * @internal ICU 4.4 Technology Preview + */ +U_CAPI void U_EXPORT2 +uplug_removePlug(UPlugData *plug, UErrorCode *status); +#endif /* U_HIDE_INTERNAL_API */ + +#endif /* UCONFIG_ENABLE_PLUGINS */ + +#endif /* _ICUPLUG */ + diff --git a/intl/icu/source/common/unicode/idna.h b/intl/icu/source/common/unicode/idna.h new file mode 100644 index 0000000000..1c57205bae --- /dev/null +++ b/intl/icu/source/common/unicode/idna.h @@ -0,0 +1,330 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2012, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: idna.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010mar05 +* created by: Markus W. Scherer +*/ + +#ifndef __IDNA_H__ +#define __IDNA_H__ + +/** + * \file + * \brief C++ API: Internationalizing Domain Names in Applications (IDNA) + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#if !UCONFIG_NO_IDNA + +#include "unicode/bytestream.h" +#include "unicode/stringpiece.h" +#include "unicode/uidna.h" +#include "unicode/unistr.h" + +U_NAMESPACE_BEGIN + +class IDNAInfo; + +/** + * Abstract base class for IDNA processing. + * See http://www.unicode.org/reports/tr46/ + * and http://www.ietf.org/rfc/rfc3490.txt + * + * The IDNA class is not intended for public subclassing. + * + * This C++ API currently only implements UTS #46. + * The uidna.h C API implements both UTS #46 (functions using UIDNA service object) + * and IDNA2003 (functions that do not use a service object). + * @stable ICU 4.6 + */ +class U_COMMON_API IDNA : public UObject { +public: + /** + * Destructor. + * @stable ICU 4.6 + */ + ~IDNA(); + + /** + * Returns an IDNA instance which implements UTS #46. + * Returns an unmodifiable instance, owned by the caller. + * Cache it for multiple operations, and delete it when done. + * The instance is thread-safe, that is, it can be used concurrently. + * + * UTS #46 defines Unicode IDNA Compatibility Processing, + * updated to the latest version of Unicode and compatible with both + * IDNA2003 and IDNA2008. + * + * The worker functions use transitional processing, including deviation mappings, + * unless UIDNA_NONTRANSITIONAL_TO_ASCII or UIDNA_NONTRANSITIONAL_TO_UNICODE + * is used in which case the deviation characters are passed through without change. + * + * Disallowed characters are mapped to U+FFFD. + * + * For available options see the uidna.h header. + * Operations with the UTS #46 instance do not support the + * UIDNA_ALLOW_UNASSIGNED option. + * + * By default, the UTS #46 implementation allows all ASCII characters (as valid or mapped). + * When the UIDNA_USE_STD3_RULES option is used, ASCII characters other than + * letters, digits, hyphen (LDH) and dot/full stop are disallowed and mapped to U+FFFD. + * + * @param options Bit set to modify the processing and error checking. + * See option bit set values in uidna.h. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the UTS #46 IDNA instance, if successful + * @stable ICU 4.6 + */ + static IDNA * + createUTS46Instance(uint32_t options, UErrorCode &errorCode); + + /** + * Converts a single domain name label into its ASCII form for DNS lookup. + * If any processing step fails, then info.hasErrors() will be true and + * the result might not be an ASCII string. + * The label might be modified according to the types of errors. + * Labels with severe errors will be left in (or turned into) their Unicode form. + * + * The UErrorCode indicates an error only in exceptional cases, + * such as a U_MEMORY_ALLOCATION_ERROR. + * + * @param label Input domain name label + * @param dest Destination string object + * @param info Output container of IDNA processing details. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return dest + * @stable ICU 4.6 + */ + virtual UnicodeString & + labelToASCII(const UnicodeString &label, UnicodeString &dest, + IDNAInfo &info, UErrorCode &errorCode) const = 0; + + /** + * Converts a single domain name label into its Unicode form for human-readable display. + * If any processing step fails, then info.hasErrors() will be true. + * The label might be modified according to the types of errors. + * + * The UErrorCode indicates an error only in exceptional cases, + * such as a U_MEMORY_ALLOCATION_ERROR. + * + * @param label Input domain name label + * @param dest Destination string object + * @param info Output container of IDNA processing details. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return dest + * @stable ICU 4.6 + */ + virtual UnicodeString & + labelToUnicode(const UnicodeString &label, UnicodeString &dest, + IDNAInfo &info, UErrorCode &errorCode) const = 0; + + /** + * Converts a whole domain name into its ASCII form for DNS lookup. + * If any processing step fails, then info.hasErrors() will be true and + * the result might not be an ASCII string. + * The domain name might be modified according to the types of errors. + * Labels with severe errors will be left in (or turned into) their Unicode form. + * + * The UErrorCode indicates an error only in exceptional cases, + * such as a U_MEMORY_ALLOCATION_ERROR. + * + * @param name Input domain name + * @param dest Destination string object + * @param info Output container of IDNA processing details. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return dest + * @stable ICU 4.6 + */ + virtual UnicodeString & + nameToASCII(const UnicodeString &name, UnicodeString &dest, + IDNAInfo &info, UErrorCode &errorCode) const = 0; + + /** + * Converts a whole domain name into its Unicode form for human-readable display. + * If any processing step fails, then info.hasErrors() will be true. + * The domain name might be modified according to the types of errors. + * + * The UErrorCode indicates an error only in exceptional cases, + * such as a U_MEMORY_ALLOCATION_ERROR. + * + * @param name Input domain name + * @param dest Destination string object + * @param info Output container of IDNA processing details. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return dest + * @stable ICU 4.6 + */ + virtual UnicodeString & + nameToUnicode(const UnicodeString &name, UnicodeString &dest, + IDNAInfo &info, UErrorCode &errorCode) const = 0; + + // UTF-8 versions of the processing methods ---------------------------- *** + + /** + * Converts a single domain name label into its ASCII form for DNS lookup. + * UTF-8 version of labelToASCII(), same behavior. + * + * @param label Input domain name label + * @param dest Destination byte sink; Flush()ed if successful + * @param info Output container of IDNA processing details. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return dest + * @stable ICU 4.6 + */ + virtual void + labelToASCII_UTF8(StringPiece label, ByteSink &dest, + IDNAInfo &info, UErrorCode &errorCode) const; + + /** + * Converts a single domain name label into its Unicode form for human-readable display. + * UTF-8 version of labelToUnicode(), same behavior. + * + * @param label Input domain name label + * @param dest Destination byte sink; Flush()ed if successful + * @param info Output container of IDNA processing details. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return dest + * @stable ICU 4.6 + */ + virtual void + labelToUnicodeUTF8(StringPiece label, ByteSink &dest, + IDNAInfo &info, UErrorCode &errorCode) const; + + /** + * Converts a whole domain name into its ASCII form for DNS lookup. + * UTF-8 version of nameToASCII(), same behavior. + * + * @param name Input domain name + * @param dest Destination byte sink; Flush()ed if successful + * @param info Output container of IDNA processing details. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return dest + * @stable ICU 4.6 + */ + virtual void + nameToASCII_UTF8(StringPiece name, ByteSink &dest, + IDNAInfo &info, UErrorCode &errorCode) const; + + /** + * Converts a whole domain name into its Unicode form for human-readable display. + * UTF-8 version of nameToUnicode(), same behavior. + * + * @param name Input domain name + * @param dest Destination byte sink; Flush()ed if successful + * @param info Output container of IDNA processing details. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return dest + * @stable ICU 4.6 + */ + virtual void + nameToUnicodeUTF8(StringPiece name, ByteSink &dest, + IDNAInfo &info, UErrorCode &errorCode) const; +}; + +class UTS46; + +/** + * Output container for IDNA processing errors. + * The IDNAInfo class is not suitable for subclassing. + * @stable ICU 4.6 + */ +class U_COMMON_API IDNAInfo : public UMemory { +public: + /** + * Constructor for stack allocation. + * @stable ICU 4.6 + */ + IDNAInfo() : errors(0), labelErrors(0), isTransDiff(false), isBiDi(false), isOkBiDi(true) {} + /** + * Were there IDNA processing errors? + * @return true if there were processing errors + * @stable ICU 4.6 + */ + UBool hasErrors() const { return errors!=0; } + /** + * Returns a bit set indicating IDNA processing errors. + * See UIDNA_ERROR_... constants in uidna.h. + * @return bit set of processing errors + * @stable ICU 4.6 + */ + uint32_t getErrors() const { return errors; } + /** + * Returns true if transitional and nontransitional processing produce different results. + * This is the case when the input label or domain name contains + * one or more deviation characters outside a Punycode label (see UTS #46). + *
    + *
  • With nontransitional processing, such characters are + * copied to the destination string. + *
  • With transitional processing, such characters are + * mapped (sharp s/sigma) or removed (joiner/nonjoiner). + *
+ * @return true if transitional and nontransitional processing produce different results + * @stable ICU 4.6 + */ + UBool isTransitionalDifferent() const { return isTransDiff; } + +private: + friend class UTS46; + + IDNAInfo(const IDNAInfo &other) = delete; // no copying + IDNAInfo &operator=(const IDNAInfo &other) = delete; // no copying + + void reset() { + errors=labelErrors=0; + isTransDiff=false; + isBiDi=false; + isOkBiDi=true; + } + + uint32_t errors, labelErrors; + UBool isTransDiff; + UBool isBiDi; + UBool isOkBiDi; +}; + +U_NAMESPACE_END + +#endif // UCONFIG_NO_IDNA + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __IDNA_H__ diff --git a/intl/icu/source/common/unicode/localebuilder.h b/intl/icu/source/common/unicode/localebuilder.h new file mode 100644 index 0000000000..f708a7ed7c --- /dev/null +++ b/intl/icu/source/common/unicode/localebuilder.h @@ -0,0 +1,309 @@ +// © 2018 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +#ifndef __LOCALEBUILDER_H__ +#define __LOCALEBUILDER_H__ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/locid.h" +#include "unicode/localematcher.h" +#include "unicode/stringpiece.h" +#include "unicode/uobject.h" + +/** + * \file + * \brief C++ API: Builder API for Locale + */ + +U_NAMESPACE_BEGIN +class CharString; + +/** + * LocaleBuilder is used to build instances of Locale + * from values configured by the setters. Unlike the Locale + * constructors, the LocaleBuilder checks if a value configured by a + * setter satisfies the syntax requirements defined by the Locale + * class. A Locale object created by a LocaleBuilder is + * well-formed and can be transformed to a well-formed IETF BCP 47 language tag + * without losing information. + * + *

The following example shows how to create a Locale object + * with the LocaleBuilder. + *

+ * 
+ *     UErrorCode status = U_ZERO_ERROR;
+ *     Locale aLocale = LocaleBuilder()
+ *                          .setLanguage("sr")
+ *                          .setScript("Latn")
+ *                          .setRegion("RS")
+ *                          .build(status);
+ *     if (U_SUCCESS(status)) {
+ *       // ...
+ *     }
+ *
+ *
+ * + *

LocaleBuilders can be reused; clear() resets all + * fields to their default values. + * + *

LocaleBuilder tracks errors in an internal UErrorCode. For all setters, + * except setLanguageTag and setLocale, LocaleBuilder will return immediately + * if the internal UErrorCode is in error state. + * To reset internal state and error code, call clear method. + * The setLanguageTag and setLocale method will first clear the internal + * UErrorCode, then track the error of the validation of the input parameter + * into the internal UErrorCode. + * + * @stable ICU 64 + */ +class U_COMMON_API LocaleBuilder : public UObject { +public: + /** + * Constructs an empty LocaleBuilder. The default value of all + * fields, extensions, and private use information is the + * empty string. + * + * @stable ICU 64 + */ + LocaleBuilder(); + + /** + * Destructor + * @stable ICU 64 + */ + virtual ~LocaleBuilder(); + + /** + * Resets the LocaleBuilder to match the provided + * locale. Existing state is discarded. + * + *

All fields of the locale must be well-formed. + *

This method clears the internal UErrorCode. + * + * @param locale the locale + * @return This builder. + * + * @stable ICU 64 + */ + LocaleBuilder& setLocale(const Locale& locale); + + /** + * Resets the LocaleBuilder to match the provided IETF BCP 47 language tag. + * Discards the existing state. + * The empty string causes the builder to be reset, like {@link #clear}. + * Legacy language tags (marked as “Type: grandfathered” in BCP 47) + * are converted to their canonical form before being processed. + * Otherwise, the language tag must be well-formed, + * or else the build() method will later report an U_ILLEGAL_ARGUMENT_ERROR. + * + *

This method clears the internal UErrorCode. + * + * @param tag the language tag, defined as IETF BCP 47 language tag. + * @return This builder. + * @stable ICU 64 + */ + LocaleBuilder& setLanguageTag(StringPiece tag); + + /** + * Sets the language. If language is the empty string, the + * language in this LocaleBuilder is removed. Otherwise, the + * language must be well-formed, or else the build() method will + * later report an U_ILLEGAL_ARGUMENT_ERROR. + * + *

The syntax of language value is defined as + * [unicode_language_subtag](http://www.unicode.org/reports/tr35/tr35.html#unicode_language_subtag). + * + * @param language the language + * @return This builder. + * @stable ICU 64 + */ + LocaleBuilder& setLanguage(StringPiece language); + + /** + * Sets the script. If script is the empty string, the script in + * this LocaleBuilder is removed. + * Otherwise, the script must be well-formed, or else the build() + * method will later report an U_ILLEGAL_ARGUMENT_ERROR. + * + *

The script value is a four-letter script code as + * [unicode_script_subtag](http://www.unicode.org/reports/tr35/tr35.html#unicode_script_subtag) + * defined by ISO 15924 + * + * @param script the script + * @return This builder. + * @stable ICU 64 + */ + LocaleBuilder& setScript(StringPiece script); + + /** + * Sets the region. If region is the empty string, the region in this + * LocaleBuilder is removed. Otherwise, the region + * must be well-formed, or else the build() method will later report an + * U_ILLEGAL_ARGUMENT_ERROR. + * + *

The region value is defined by + * [unicode_region_subtag](http://www.unicode.org/reports/tr35/tr35.html#unicode_region_subtag) + * as a two-letter ISO 3166 code or a three-digit UN M.49 area code. + * + *

The region value in the Locale created by the + * LocaleBuilder is always normalized to upper case. + * + * @param region the region + * @return This builder. + * @stable ICU 64 + */ + LocaleBuilder& setRegion(StringPiece region); + + /** + * Sets the variant. If variant is the empty string, the variant in this + * LocaleBuilder is removed. Otherwise, the variant + * must be well-formed, or else the build() method will later report an + * U_ILLEGAL_ARGUMENT_ERROR. + * + *

Note: This method checks if variant + * satisfies the + * [unicode_variant_subtag](http://www.unicode.org/reports/tr35/tr35.html#unicode_variant_subtag) + * syntax requirements, and normalizes the value to lowercase letters. However, + * the Locale class does not impose any syntactic + * restriction on variant. To set an ill-formed variant, use a Locale constructor. + * If there are multiple unicode_variant_subtag, the caller must concatenate + * them with '-' as separator (ex: "foobar-fibar"). + * + * @param variant the variant + * @return This builder. + * @stable ICU 64 + */ + LocaleBuilder& setVariant(StringPiece variant); + + /** + * Sets the extension for the given key. If the value is the empty string, + * the extension is removed. Otherwise, the key and + * value must be well-formed, or else the build() method will + * later report an U_ILLEGAL_ARGUMENT_ERROR. + * + *

Note: The key ('u') is used for the Unicode locale extension. + * Setting a value for this key replaces any existing Unicode locale key/type + * pairs with those defined in the extension. + * + *

Note: The key ('x') is used for the private use code. To be + * well-formed, the value for this key needs only to have subtags of one to + * eight alphanumeric characters, not two to eight as in the general case. + * + * @param key the extension key + * @param value the extension value + * @return This builder. + * @stable ICU 64 + */ + LocaleBuilder& setExtension(char key, StringPiece value); + + /** + * Sets the Unicode locale keyword type for the given key. If the type + * StringPiece is constructed with a nullptr, the keyword is removed. + * If the type is the empty string, the keyword is set without type subtags. + * Otherwise, the key and type must be well-formed, or else the build() + * method will later report an U_ILLEGAL_ARGUMENT_ERROR. + * + *

Keys and types are converted to lower case. + * + *

Note:Setting the 'u' extension via {@link #setExtension} + * replaces all Unicode locale keywords with those defined in the + * extension. + * + * @param key the Unicode locale key + * @param type the Unicode locale type + * @return This builder. + * @stable ICU 64 + */ + LocaleBuilder& setUnicodeLocaleKeyword( + StringPiece key, StringPiece type); + + /** + * Adds a unicode locale attribute, if not already present, otherwise + * has no effect. The attribute must not be empty string and must be + * well-formed or U_ILLEGAL_ARGUMENT_ERROR will be set to status + * during the build() call. + * + * @param attribute the attribute + * @return This builder. + * @stable ICU 64 + */ + LocaleBuilder& addUnicodeLocaleAttribute(StringPiece attribute); + + /** + * Removes a unicode locale attribute, if present, otherwise has no + * effect. The attribute must not be empty string and must be well-formed + * or U_ILLEGAL_ARGUMENT_ERROR will be set to status during the build() call. + * + *

Attribute comparison for removal is case-insensitive. + * + * @param attribute the attribute + * @return This builder. + * @stable ICU 64 + */ + LocaleBuilder& removeUnicodeLocaleAttribute(StringPiece attribute); + + /** + * Resets the builder to its initial, empty state. + *

This method clears the internal UErrorCode. + * + * @return this builder + * @stable ICU 64 + */ + LocaleBuilder& clear(); + + /** + * Resets the extensions to their initial, empty state. + * Language, script, region and variant are unchanged. + * + * @return this builder + * @stable ICU 64 + */ + LocaleBuilder& clearExtensions(); + + /** + * Returns an instance of Locale created from the fields set + * on this builder. + * If any set methods or during the build() call require memory allocation + * but fail U_MEMORY_ALLOCATION_ERROR will be set to status. + * If any of the fields set by the setters are not well-formed, the status + * will be set to U_ILLEGAL_ARGUMENT_ERROR. The state of the builder will + * not change after the build() call and the caller is free to keep using + * the same builder to build more locales. + * + * @return a new Locale + * @stable ICU 64 + */ + Locale build(UErrorCode& status); + + /** + * Sets the UErrorCode if an error occurred while recording sets. + * Preserves older error codes in the outErrorCode. + * @param outErrorCode Set to an error code that occurred while setting subtags. + * Unchanged if there is no such error or if outErrorCode + * already contained an error. + * @return true if U_FAILURE(outErrorCode) + * @stable ICU 65 + */ + UBool copyErrorTo(UErrorCode &outErrorCode) const; + +private: + friend class LocaleMatcher::Result; + + void copyExtensionsFrom(const Locale& src, UErrorCode& errorCode); + + UErrorCode status_; + char language_[9]; + char script_[5]; + char region_[4]; + CharString *variant_; // Pointer not object so we need not #include internal charstr.h. + icu::Locale *extensions_; // Pointer not object. Storage for all other fields. + +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __LOCALEBUILDER_H__ diff --git a/intl/icu/source/common/unicode/localematcher.h b/intl/icu/source/common/unicode/localematcher.h new file mode 100644 index 0000000000..603daf7231 --- /dev/null +++ b/intl/icu/source/common/unicode/localematcher.h @@ -0,0 +1,708 @@ +// © 2019 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// localematcher.h +// created: 2019may08 Markus W. Scherer + +#ifndef __LOCALEMATCHER_H__ +#define __LOCALEMATCHER_H__ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/locid.h" +#include "unicode/stringpiece.h" +#include "unicode/uobject.h" + +/** + * \file + * \brief C++ API: Locale matcher: User's desired locales vs. application's supported locales. + */ + +/** + * Builder option for whether the language subtag or the script subtag is most important. + * + * @see LocaleMatcher::Builder#setFavorSubtag(ULocMatchFavorSubtag) + * @stable ICU 65 + */ +enum ULocMatchFavorSubtag { + /** + * Language differences are most important, then script differences, then region differences. + * (This is the default behavior.) + * + * @stable ICU 65 + */ + ULOCMATCH_FAVOR_LANGUAGE, + /** + * Makes script differences matter relatively more than language differences. + * + * @stable ICU 65 + */ + ULOCMATCH_FAVOR_SCRIPT +}; +#ifndef U_IN_DOXYGEN +typedef enum ULocMatchFavorSubtag ULocMatchFavorSubtag; +#endif + +/** + * Builder option for whether all desired locales are treated equally or + * earlier ones are preferred. + * + * @see LocaleMatcher::Builder#setDemotionPerDesiredLocale(ULocMatchDemotion) + * @stable ICU 65 + */ +enum ULocMatchDemotion { + /** + * All desired locales are treated equally. + * + * @stable ICU 65 + */ + ULOCMATCH_DEMOTION_NONE, + /** + * Earlier desired locales are preferred. + * + *

From each desired locale to the next, + * the distance to any supported locale is increased by an additional amount + * which is at least as large as most region mismatches. + * A later desired locale has to have a better match with some supported locale + * due to more than merely having the same region subtag. + * + *

For example: Supported={en, sv} desired=[en-GB, sv] + * yields Result(en-GB, en) because + * with the demotion of sv its perfect match is no better than + * the region distance between the earlier desired locale en-GB and en=en-US. + * + *

Notes: + *

    + *
  • In some cases, language and/or script differences can be as small as + * the typical region difference. (Example: sr-Latn vs. sr-Cyrl) + *
  • It is possible for certain region differences to be larger than usual, + * and larger than the demotion. + * (As of CLDR 35 there is no such case, but + * this is possible in future versions of the data.) + *
+ * + * @stable ICU 65 + */ + ULOCMATCH_DEMOTION_REGION +}; +#ifndef U_IN_DOXYGEN +typedef enum ULocMatchDemotion ULocMatchDemotion; +#endif + +/** + * Builder option for whether to include or ignore one-way (fallback) match data. + * The LocaleMatcher uses CLDR languageMatch data which includes fallback (oneway=true) entries. + * Sometimes it is desirable to ignore those. + * + *

For example, consider a web application with the UI in a given language, + * with a link to another, related web app. + * The link should include the UI language, and the target server may also use + * the client’s Accept-Language header data. + * The target server has its own list of supported languages. + * One may want to favor UI language consistency, that is, + * if there is a decent match for the original UI language, we want to use it, + * but not if it is merely a fallback. + * + * @see LocaleMatcher::Builder#setDirection(ULocMatchDirection) + * @stable ICU 67 + */ +enum ULocMatchDirection { + /** + * Locale matching includes one-way matches such as Breton→French. (default) + * + * @stable ICU 67 + */ + ULOCMATCH_DIRECTION_WITH_ONE_WAY, + /** + * Locale matching limited to two-way matches including e.g. Danish↔Norwegian + * but ignoring one-way matches. + * + * @stable ICU 67 + */ + ULOCMATCH_DIRECTION_ONLY_TWO_WAY +}; +#ifndef U_IN_DOXYGEN +typedef enum ULocMatchDirection ULocMatchDirection; +#endif + +struct UHashtable; + +U_NAMESPACE_BEGIN + +struct LSR; + +class LocaleDistance; +class LocaleLsrIterator; +class UVector; +class XLikelySubtags; + +/** + * Immutable class that picks the best match between a user's desired locales and + * an application's supported locales. + * Movable but not copyable. + * + *

Example: + * 

+ * UErrorCode errorCode = U_ZERO_ERROR;
+ * LocaleMatcher matcher = LocaleMatcher::Builder().setSupportedLocales("fr, en-GB, en").build(errorCode);
+ * Locale *bestSupported = matcher.getBestLocale(Locale.US, errorCode);  // "en"
+ *
+ * + *

A matcher takes into account when languages are close to one another, + * such as Danish and Norwegian, + * and when regional variants are close, like en-GB and en-AU as opposed to en-US. + * + *

If there are multiple supported locales with the same (language, script, region) + * likely subtags, then the current implementation returns the first of those locales. + * It ignores variant subtags (except for pseudolocale variants) and extensions. + * This may change in future versions. + * + *

For example, the current implementation does not distinguish between + * de, de-DE, de-Latn, de-1901, de-u-co-phonebk. + * + *

If you prefer one equivalent locale over another, then provide only the preferred one, + * or place it earlier in the list of supported locales. + * + *

Otherwise, the order of supported locales may have no effect on the best-match results. + * The current implementation compares each desired locale with supported locales + * in the following order: + * 1. Default locale, if supported; + * 2. CLDR "paradigm locales" like en-GB and es-419; + * 3. other supported locales. + * This may change in future versions. + * + *

Often a product will just need one matcher instance, built with the languages + * that it supports. However, it may want multiple instances with different + * default languages based on additional information, such as the domain. + * + *

This class is not intended for public subclassing. + * + * @stable ICU 65 + */ +class U_COMMON_API LocaleMatcher : public UMemory { +public: + /** + * Data for the best-matching pair of a desired and a supported locale. + * Movable but not copyable. + * + * @stable ICU 65 + */ + class U_COMMON_API Result : public UMemory { + public: + /** + * Move constructor; might modify the source. + * This object will have the same contents that the source object had. + * + * @param src Result to move contents from. + * @stable ICU 65 + */ + Result(Result &&src) noexcept; + + /** + * Destructor. + * + * @stable ICU 65 + */ + ~Result(); + + /** + * Move assignment; might modify the source. + * This object will have the same contents that the source object had. + * + * @param src Result to move contents from. + * @stable ICU 65 + */ + Result &operator=(Result &&src) noexcept; + + /** + * Returns the best-matching desired locale. + * nullptr if the list of desired locales is empty or if none matched well enough. + * + * @return the best-matching desired locale, or nullptr. + * @stable ICU 65 + */ + inline const Locale *getDesiredLocale() const { return desiredLocale; } + + /** + * Returns the best-matching supported locale. + * If none matched well enough, this is the default locale. + * The default locale is nullptr if Builder::setNoDefaultLocale() was called, + * or if the list of supported locales is empty and no explicit default locale is set. + * + * @return the best-matching supported locale, or nullptr. + * @stable ICU 65 + */ + inline const Locale *getSupportedLocale() const { return supportedLocale; } + + /** + * Returns the index of the best-matching desired locale in the input Iterable order. + * -1 if the list of desired locales is empty or if none matched well enough. + * + * @return the index of the best-matching desired locale, or -1. + * @stable ICU 65 + */ + inline int32_t getDesiredIndex() const { return desiredIndex; } + + /** + * Returns the index of the best-matching supported locale in the + * constructor’s or builder’s input order (“set” Collection plus “added” locales). + * If the matcher was built from a locale list string, then the iteration order is that + * of a LocalePriorityList built from the same string. + * -1 if the list of supported locales is empty or if none matched well enough. + * + * @return the index of the best-matching supported locale, or -1. + * @stable ICU 65 + */ + inline int32_t getSupportedIndex() const { return supportedIndex; } + + /** + * Takes the best-matching supported locale and adds relevant fields of the + * best-matching desired locale, such as the -t- and -u- extensions. + * May replace some fields of the supported locale. + * The result is the locale that should be used for date and number formatting, collation, etc. + * Returns the root locale if getSupportedLocale() returns nullptr. + * + *

Example: desired=ar-SA-u-nu-latn, supported=ar-EG, resolved locale=ar-SA-u-nu-latn + * + * @return a locale combining the best-matching desired and supported locales. + * @stable ICU 65 + */ + Locale makeResolvedLocale(UErrorCode &errorCode) const; + + private: + Result(const Locale *desired, const Locale *supported, + int32_t desIndex, int32_t suppIndex, UBool owned) : + desiredLocale(desired), supportedLocale(supported), + desiredIndex(desIndex), supportedIndex(suppIndex), + desiredIsOwned(owned) {} + + Result(const Result &other) = delete; + Result &operator=(const Result &other) = delete; + + const Locale *desiredLocale; + const Locale *supportedLocale; + int32_t desiredIndex; + int32_t supportedIndex; + UBool desiredIsOwned; + + friend class LocaleMatcher; + }; + + /** + * LocaleMatcher builder. + * Movable but not copyable. + * + * @stable ICU 65 + */ + class U_COMMON_API Builder : public UMemory { + public: + /** + * Constructs a builder used in chaining parameters for building a LocaleMatcher. + * + * @return a new Builder object + * @stable ICU 65 + */ + Builder() {} + + /** + * Move constructor; might modify the source. + * This builder will have the same contents that the source builder had. + * + * @param src Builder to move contents from. + * @stable ICU 65 + */ + Builder(Builder &&src) noexcept; + + /** + * Destructor. + * + * @stable ICU 65 + */ + ~Builder(); + + /** + * Move assignment; might modify the source. + * This builder will have the same contents that the source builder had. + * + * @param src Builder to move contents from. + * @stable ICU 65 + */ + Builder &operator=(Builder &&src) noexcept; + + /** + * Parses an Accept-Language string + * (RFC 2616 Section 14.4), + * such as "af, en, fr;q=0.9", and sets the supported locales accordingly. + * Allows whitespace in more places but does not allow "*". + * Clears any previously set/added supported locales first. + * + * @param locales the Accept-Language string of locales to set + * @return this Builder object + * @stable ICU 65 + */ + Builder &setSupportedLocalesFromListString(StringPiece locales); + + /** + * Copies the supported locales, preserving iteration order. + * Clears any previously set/added supported locales first. + * Duplicates are allowed, and are not removed. + * + * @param locales the list of locale + * @return this Builder object + * @stable ICU 65 + */ + Builder &setSupportedLocales(Locale::Iterator &locales); + + /** + * Copies the supported locales from the begin/end range, preserving iteration order. + * Clears any previously set/added supported locales first. + * Duplicates are allowed, and are not removed. + * + * Each of the iterator parameter values must be an + * input iterator whose value is convertible to const Locale &. + * + * @param begin Start of range. + * @param end Exclusive end of range. + * @return this Builder object + * @stable ICU 65 + */ + template + Builder &setSupportedLocales(Iter begin, Iter end) { + if (U_FAILURE(errorCode_)) { return *this; } + clearSupportedLocales(); + while (begin != end) { + addSupportedLocale(*begin++); + } + return *this; + } + + /** + * Copies the supported locales from the begin/end range, preserving iteration order. + * Calls the converter to convert each *begin to a Locale or const Locale &. + * Clears any previously set/added supported locales first. + * Duplicates are allowed, and are not removed. + * + * Each of the iterator parameter values must be an + * input iterator whose value is convertible to const Locale &. + * + * @param begin Start of range. + * @param end Exclusive end of range. + * @param converter Converter from *begin to const Locale & or compatible. + * @return this Builder object + * @stable ICU 65 + */ + template + Builder &setSupportedLocalesViaConverter(Iter begin, Iter end, Conv converter) { + if (U_FAILURE(errorCode_)) { return *this; } + clearSupportedLocales(); + while (begin != end) { + addSupportedLocale(converter(*begin++)); + } + return *this; + } + + /** + * Adds another supported locale. + * Duplicates are allowed, and are not removed. + * + * @param locale another locale + * @return this Builder object + * @stable ICU 65 + */ + Builder &addSupportedLocale(const Locale &locale); + + /** + * Sets no default locale. + * There will be no explicit or implicit default locale. + * If there is no good match, then the matcher will return nullptr for the + * best supported locale. + * + * @stable ICU 68 + */ + Builder &setNoDefaultLocale(); + + /** + * Sets the default locale; if nullptr, or if it is not set explicitly, + * then the first supported locale is used as the default locale. + * There is no default locale at all (nullptr will be returned instead) + * if setNoDefaultLocale() is called. + * + * @param defaultLocale the default locale (will be copied) + * @return this Builder object + * @stable ICU 65 + */ + Builder &setDefaultLocale(const Locale *defaultLocale); + + /** + * If ULOCMATCH_FAVOR_SCRIPT, then the language differences are smaller than script + * differences. + * This is used in situations (such as maps) where + * it is better to fall back to the same script than a similar language. + * + * @param subtag the subtag to favor + * @return this Builder object + * @stable ICU 65 + */ + Builder &setFavorSubtag(ULocMatchFavorSubtag subtag); + + /** + * Option for whether all desired locales are treated equally or + * earlier ones are preferred (this is the default). + * + * @param demotion the demotion per desired locale to set. + * @return this Builder object + * @stable ICU 65 + */ + Builder &setDemotionPerDesiredLocale(ULocMatchDemotion demotion); + + /** + * Option for whether to include or ignore one-way (fallback) match data. + * By default, they are included. + * + * @param matchDirection the match direction to set. + * @return this Builder object + * @stable ICU 67 + */ + Builder &setDirection(ULocMatchDirection matchDirection) { + if (U_SUCCESS(errorCode_)) { + direction_ = matchDirection; + } + return *this; + } + + /** + * Sets the maximum distance for an acceptable match. + * The matcher will return a match for a pair of locales only if + * they match at least as well as the pair given here. + * + * For example, setMaxDistance(en-US, en-GB) limits matches to ones where the + * (desired, support) locales have a distance no greater than a region subtag difference. + * This is much stricter than the CLDR default. + * + * The details of locale matching are subject to changes in + * CLDR data and in the algorithm. + * Specifying a maximum distance in relative terms via a sample pair of locales + * insulates from changes that affect all distance metrics similarly, + * but some changes will necessarily affect relative distances between + * different pairs of locales. + * + * @param desired the desired locale for distance comparison. + * @param supported the supported locale for distance comparison. + * @return this Builder object + * @stable ICU 68 + */ + Builder &setMaxDistance(const Locale &desired, const Locale &supported); + + /** + * Sets the UErrorCode if an error occurred while setting parameters. + * Preserves older error codes in the outErrorCode. + * + * @param outErrorCode Set to an error code if it does not contain one already + * and an error occurred while setting parameters. + * Otherwise unchanged. + * @return true if U_FAILURE(outErrorCode) + * @stable ICU 65 + */ + UBool copyErrorTo(UErrorCode &outErrorCode) const; + + /** + * Builds and returns a new locale matcher. + * This builder can continue to be used. + * + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return LocaleMatcher + * @stable ICU 65 + */ + LocaleMatcher build(UErrorCode &errorCode) const; + + private: + friend class LocaleMatcher; + + Builder(const Builder &other) = delete; + Builder &operator=(const Builder &other) = delete; + + void clearSupportedLocales(); + bool ensureSupportedLocaleVector(); + + UErrorCode errorCode_ = U_ZERO_ERROR; + UVector *supportedLocales_ = nullptr; + int32_t thresholdDistance_ = -1; + ULocMatchDemotion demotion_ = ULOCMATCH_DEMOTION_REGION; + Locale *defaultLocale_ = nullptr; + bool withDefault_ = true; + ULocMatchFavorSubtag favor_ = ULOCMATCH_FAVOR_LANGUAGE; + ULocMatchDirection direction_ = ULOCMATCH_DIRECTION_WITH_ONE_WAY; + Locale *maxDistanceDesired_ = nullptr; + Locale *maxDistanceSupported_ = nullptr; + }; + + // FYI No public LocaleMatcher constructors in C++; use the Builder. + + /** + * Move copy constructor; might modify the source. + * This matcher will have the same settings that the source matcher had. + * @param src source matcher + * @stable ICU 65 + */ + LocaleMatcher(LocaleMatcher &&src) noexcept; + + /** + * Destructor. + * @stable ICU 65 + */ + ~LocaleMatcher(); + + /** + * Move assignment operator; might modify the source. + * This matcher will have the same settings that the source matcher had. + * The behavior is undefined if *this and src are the same object. + * @param src source matcher + * @return *this + * @stable ICU 65 + */ + LocaleMatcher &operator=(LocaleMatcher &&src) noexcept; + + /** + * Returns the supported locale which best matches the desired locale. + * + * @param desiredLocale Typically a user's language. + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return the best-matching supported locale. + * @stable ICU 65 + */ + const Locale *getBestMatch(const Locale &desiredLocale, UErrorCode &errorCode) const; + + /** + * Returns the supported locale which best matches one of the desired locales. + * + * @param desiredLocales Typically a user's languages, in order of preference (descending). + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return the best-matching supported locale. + * @stable ICU 65 + */ + const Locale *getBestMatch(Locale::Iterator &desiredLocales, UErrorCode &errorCode) const; + + /** + * Parses an Accept-Language string + * (RFC 2616 Section 14.4), + * such as "af, en, fr;q=0.9", + * and returns the supported locale which best matches one of the desired locales. + * Allows whitespace in more places but does not allow "*". + * + * @param desiredLocaleList Typically a user's languages, as an Accept-Language string. + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return the best-matching supported locale. + * @stable ICU 65 + */ + const Locale *getBestMatchForListString(StringPiece desiredLocaleList, UErrorCode &errorCode) const; + + /** + * Returns the best match between the desired locale and the supported locales. + * If the result's desired locale is not nullptr, then it is the address of the input locale. + * It has not been cloned. + * + * @param desiredLocale Typically a user's language. + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return the best-matching pair of the desired and a supported locale. + * @stable ICU 65 + */ + Result getBestMatchResult(const Locale &desiredLocale, UErrorCode &errorCode) const; + + /** + * Returns the best match between the desired and supported locales. + * If the result's desired locale is not nullptr, then it is a clone of + * the best-matching desired locale. The Result object owns the clone. + * + * @param desiredLocales Typically a user's languages, in order of preference (descending). + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return the best-matching pair of a desired and a supported locale. + * @stable ICU 65 + */ + Result getBestMatchResult(Locale::Iterator &desiredLocales, UErrorCode &errorCode) const; + + /** + * Returns true if the pair of locales matches acceptably. + * This is influenced by Builder options such as setDirection(), setFavorSubtag(), + * and setMaxDistance(). + * + * @param desired The desired locale. + * @param supported The supported locale. + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return true if the pair of locales matches acceptably. + * @stable ICU 68 + */ + UBool isMatch(const Locale &desired, const Locale &supported, UErrorCode &errorCode) const; + +#ifndef U_HIDE_INTERNAL_API + /** + * Returns a fraction between 0 and 1, where 1 means that the languages are a + * perfect match, and 0 means that they are completely different. + * + *

This is mostly an implementation detail, and the precise values may change over time. + * The implementation may use either the maximized forms or the others ones, or both. + * The implementation may or may not rely on the forms to be consistent with each other. + * + *

Callers should construct and use a matcher rather than match pairs of locales directly. + * + * @param desired Desired locale. + * @param supported Supported locale. + * @param errorCode ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return value between 0 and 1, inclusive. + * @internal (has a known user) + */ + double internalMatch(const Locale &desired, const Locale &supported, UErrorCode &errorCode) const; +#endif // U_HIDE_INTERNAL_API + +private: + LocaleMatcher(const Builder &builder, UErrorCode &errorCode); + LocaleMatcher(const LocaleMatcher &other) = delete; + LocaleMatcher &operator=(const LocaleMatcher &other) = delete; + + int32_t putIfAbsent(const LSR &lsr, int32_t i, int32_t suppLength, UErrorCode &errorCode); + + int32_t getBestSuppIndex(LSR desiredLSR, LocaleLsrIterator *remainingIter, UErrorCode &errorCode) const; + + const XLikelySubtags &likelySubtags; + const LocaleDistance &localeDistance; + int32_t thresholdDistance; + int32_t demotionPerDesiredLocale; + ULocMatchFavorSubtag favorSubtag; + ULocMatchDirection direction; + + // These are in input order. + const Locale ** supportedLocales; + LSR *lsrs; + int32_t supportedLocalesLength; + // These are in preference order: 1. Default locale 2. paradigm locales 3. others. + UHashtable *supportedLsrToIndex; // Map + // Array versions of the supportedLsrToIndex keys and values. + // The distance lookup loops over the supportedLSRs and returns the index of the best match. + const LSR **supportedLSRs; + int32_t *supportedIndexes; + int32_t supportedLSRsLength; + Locale *ownedDefaultLocale; + const Locale *defaultLocale; +}; + +U_NAMESPACE_END + +#endif // U_SHOW_CPLUSPLUS_API +#endif // __LOCALEMATCHER_H__ diff --git a/intl/icu/source/common/unicode/localpointer.h b/intl/icu/source/common/unicode/localpointer.h new file mode 100644 index 0000000000..b8be3d7942 --- /dev/null +++ b/intl/icu/source/common/unicode/localpointer.h @@ -0,0 +1,595 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2009-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: localpointer.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2009nov13 +* created by: Markus W. Scherer +*/ + +#ifndef __LOCALPOINTER_H__ +#define __LOCALPOINTER_H__ + +/** + * \file + * \brief C++ API: "Smart pointers" for use with and in ICU4C C++ code. + * + * These classes are inspired by + * - std::auto_ptr + * - boost::scoped_ptr & boost::scoped_array + * - Taligent Safe Pointers (TOnlyPointerTo) + * + * but none of those provide for all of the goals for ICU smart pointers: + * - Smart pointer owns the object and releases it when it goes out of scope. + * - No transfer of ownership via copy/assignment to reduce misuse. Simpler & more robust. + * - ICU-compatible: No exceptions. + * - Need to be able to orphan/release the pointer and its ownership. + * - Need variants for normal C++ object pointers, C++ arrays, and ICU C service objects. + * + * For details see https://icu.unicode.org/design/cpp/scoped_ptr + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include + +U_NAMESPACE_BEGIN + +/** + * "Smart pointer" base class; do not use directly: use LocalPointer etc. + * + * Base class for smart pointer classes that do not throw exceptions. + * + * Do not use this base class directly, since it does not delete its pointer. + * A subclass must implement methods that delete the pointer: + * Destructor and adoptInstead(). + * + * There is no operator T *() provided because the programmer must decide + * whether to use getAlias() (without transfer of ownership) or orphan() + * (with transfer of ownership and NULLing of the pointer). + * + * @see LocalPointer + * @see LocalArray + * @see U_DEFINE_LOCAL_OPEN_POINTER + * @stable ICU 4.4 + */ +template +class LocalPointerBase { +public: + // No heap allocation. Use only on the stack. + static void* U_EXPORT2 operator new(size_t) = delete; + static void* U_EXPORT2 operator new[](size_t) = delete; +#if U_HAVE_PLACEMENT_NEW + static void* U_EXPORT2 operator new(size_t, void*) = delete; +#endif + + /** + * Constructor takes ownership. + * @param p simple pointer to an object that is adopted + * @stable ICU 4.4 + */ + explicit LocalPointerBase(T *p=nullptr) : ptr(p) {} + /** + * Destructor deletes the object it owns. + * Subclass must override: Base class does nothing. + * @stable ICU 4.4 + */ + ~LocalPointerBase() { /* delete ptr; */ } + /** + * nullptr check. + * @return true if ==nullptr + * @stable ICU 4.4 + */ + UBool isNull() const { return ptr==nullptr; } + /** + * nullptr check. + * @return true if !=nullptr + * @stable ICU 4.4 + */ + UBool isValid() const { return ptr!=nullptr; } + /** + * Comparison with a simple pointer, so that existing code + * with ==nullptr need not be changed. + * @param other simple pointer for comparison + * @return true if this pointer value equals other + * @stable ICU 4.4 + */ + bool operator==(const T *other) const { return ptr==other; } + /** + * Comparison with a simple pointer, so that existing code + * with !=nullptr need not be changed. + * @param other simple pointer for comparison + * @return true if this pointer value differs from other + * @stable ICU 4.4 + */ + bool operator!=(const T *other) const { return ptr!=other; } + /** + * Access without ownership change. + * @return the pointer value + * @stable ICU 4.4 + */ + T *getAlias() const { return ptr; } + /** + * Access without ownership change. + * @return the pointer value as a reference + * @stable ICU 4.4 + */ + T &operator*() const { return *ptr; } + /** + * Access without ownership change. + * @return the pointer value + * @stable ICU 4.4 + */ + T *operator->() const { return ptr; } + /** + * Gives up ownership; the internal pointer becomes nullptr. + * @return the pointer value; + * caller becomes responsible for deleting the object + * @stable ICU 4.4 + */ + T *orphan() { + T *p=ptr; + ptr=nullptr; + return p; + } + /** + * Deletes the object it owns, + * and adopts (takes ownership of) the one passed in. + * Subclass must override: Base class does not delete the object. + * @param p simple pointer to an object that is adopted + * @stable ICU 4.4 + */ + void adoptInstead(T *p) { + // delete ptr; + ptr=p; + } +protected: + /** + * Actual pointer. + * @internal + */ + T *ptr; +private: + // No comparison operators with other LocalPointerBases. + bool operator==(const LocalPointerBase &other); + bool operator!=(const LocalPointerBase &other); + // No ownership sharing: No copy constructor, no assignment operator. + LocalPointerBase(const LocalPointerBase &other); + void operator=(const LocalPointerBase &other); +}; + +/** + * "Smart pointer" class, deletes objects via the standard C++ delete operator. + * For most methods see the LocalPointerBase base class. + * + * Usage example: + * \code + * LocalPointer s(new UnicodeString((UChar32)0x50005)); + * int32_t length=s->length(); // 2 + * char16_t lead=s->charAt(0); // 0xd900 + * if(some condition) { return; } // no need to explicitly delete the pointer + * s.adoptInstead(new UnicodeString((char16_t)0xfffc)); + * length=s->length(); // 1 + * // no need to explicitly delete the pointer + * \endcode + * + * @see LocalPointerBase + * @stable ICU 4.4 + */ +template +class LocalPointer : public LocalPointerBase { +public: + using LocalPointerBase::operator*; + using LocalPointerBase::operator->; + /** + * Constructor takes ownership. + * @param p simple pointer to an object that is adopted + * @stable ICU 4.4 + */ + explicit LocalPointer(T *p=nullptr) : LocalPointerBase(p) {} + /** + * Constructor takes ownership and reports an error if nullptr. + * + * This constructor is intended to be used with other-class constructors + * that may report a failure UErrorCode, + * so that callers need to check only for U_FAILURE(errorCode) + * and not also separately for isNull(). + * + * @param p simple pointer to an object that is adopted + * @param errorCode in/out UErrorCode, set to U_MEMORY_ALLOCATION_ERROR + * if p==nullptr and no other failure code had been set + * @stable ICU 55 + */ + LocalPointer(T *p, UErrorCode &errorCode) : LocalPointerBase(p) { + if(p==nullptr && U_SUCCESS(errorCode)) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } + } + /** + * Move constructor, leaves src with isNull(). + * @param src source smart pointer + * @stable ICU 56 + */ + LocalPointer(LocalPointer &&src) noexcept : LocalPointerBase(src.ptr) { + src.ptr=nullptr; + } + + /** + * Constructs a LocalPointer from a C++11 std::unique_ptr. + * The LocalPointer steals the object owned by the std::unique_ptr. + * + * This constructor works via move semantics. If your std::unique_ptr is + * in a local variable, you must use std::move. + * + * @param p The std::unique_ptr from which the pointer will be stolen. + * @stable ICU 64 + */ + explicit LocalPointer(std::unique_ptr &&p) + : LocalPointerBase(p.release()) {} + + /** + * Destructor deletes the object it owns. + * @stable ICU 4.4 + */ + ~LocalPointer() { + delete LocalPointerBase::ptr; + } + /** + * Move assignment operator, leaves src with isNull(). + * The behavior is undefined if *this and src are the same object. + * @param src source smart pointer + * @return *this + * @stable ICU 56 + */ + LocalPointer &operator=(LocalPointer &&src) noexcept { + delete LocalPointerBase::ptr; + LocalPointerBase::ptr=src.ptr; + src.ptr=nullptr; + return *this; + } + + /** + * Move-assign from an std::unique_ptr to this LocalPointer. + * Steals the pointer from the std::unique_ptr. + * + * @param p The std::unique_ptr from which the pointer will be stolen. + * @return *this + * @stable ICU 64 + */ + LocalPointer &operator=(std::unique_ptr &&p) noexcept { + adoptInstead(p.release()); + return *this; + } + + /** + * Swap pointers. + * @param other other smart pointer + * @stable ICU 56 + */ + void swap(LocalPointer &other) noexcept { + T *temp=LocalPointerBase::ptr; + LocalPointerBase::ptr=other.ptr; + other.ptr=temp; + } + /** + * Non-member LocalPointer swap function. + * @param p1 will get p2's pointer + * @param p2 will get p1's pointer + * @stable ICU 56 + */ + friend inline void swap(LocalPointer &p1, LocalPointer &p2) noexcept { + p1.swap(p2); + } + /** + * Deletes the object it owns, + * and adopts (takes ownership of) the one passed in. + * @param p simple pointer to an object that is adopted + * @stable ICU 4.4 + */ + void adoptInstead(T *p) { + delete LocalPointerBase::ptr; + LocalPointerBase::ptr=p; + } + /** + * Deletes the object it owns, + * and adopts (takes ownership of) the one passed in. + * + * If U_FAILURE(errorCode), then the current object is retained and the new one deleted. + * + * If U_SUCCESS(errorCode) but the input pointer is nullptr, + * then U_MEMORY_ALLOCATION_ERROR is set, + * the current object is deleted, and nullptr is set. + * + * @param p simple pointer to an object that is adopted + * @param errorCode in/out UErrorCode, set to U_MEMORY_ALLOCATION_ERROR + * if p==nullptr and no other failure code had been set + * @stable ICU 55 + */ + void adoptInsteadAndCheckErrorCode(T *p, UErrorCode &errorCode) { + if(U_SUCCESS(errorCode)) { + delete LocalPointerBase::ptr; + LocalPointerBase::ptr=p; + if(p==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } + } else { + delete p; + } + } + + /** + * Conversion operator to a C++11 std::unique_ptr. + * Disowns the object and gives it to the returned std::unique_ptr. + * + * This operator works via move semantics. If your LocalPointer is + * in a local variable, you must use std::move. + * + * @return An std::unique_ptr owning the pointer previously owned by this + * icu::LocalPointer. + * @stable ICU 64 + */ + operator std::unique_ptr () && { + return std::unique_ptr(LocalPointerBase::orphan()); + } +}; + +/** + * "Smart pointer" class, deletes objects via the C++ array delete[] operator. + * For most methods see the LocalPointerBase base class. + * Adds operator[] for array item access. + * + * Usage example: + * \code + * LocalArray a(new UnicodeString[2]); + * a[0].append((char16_t)0x61); + * if(some condition) { return; } // no need to explicitly delete the array + * a.adoptInstead(new UnicodeString[4]); + * a[3].append((char16_t)0x62).append((char16_t)0x63).reverse(); + * // no need to explicitly delete the array + * \endcode + * + * @see LocalPointerBase + * @stable ICU 4.4 + */ +template +class LocalArray : public LocalPointerBase { +public: + using LocalPointerBase::operator*; + using LocalPointerBase::operator->; + /** + * Constructor takes ownership. + * @param p simple pointer to an array of T objects that is adopted + * @stable ICU 4.4 + */ + explicit LocalArray(T *p=nullptr) : LocalPointerBase(p) {} + /** + * Constructor takes ownership and reports an error if nullptr. + * + * This constructor is intended to be used with other-class constructors + * that may report a failure UErrorCode, + * so that callers need to check only for U_FAILURE(errorCode) + * and not also separately for isNull(). + * + * @param p simple pointer to an array of T objects that is adopted + * @param errorCode in/out UErrorCode, set to U_MEMORY_ALLOCATION_ERROR + * if p==nullptr and no other failure code had been set + * @stable ICU 56 + */ + LocalArray(T *p, UErrorCode &errorCode) : LocalPointerBase(p) { + if(p==nullptr && U_SUCCESS(errorCode)) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } + } + /** + * Move constructor, leaves src with isNull(). + * @param src source smart pointer + * @stable ICU 56 + */ + LocalArray(LocalArray &&src) noexcept : LocalPointerBase(src.ptr) { + src.ptr=nullptr; + } + + /** + * Constructs a LocalArray from a C++11 std::unique_ptr of an array type. + * The LocalPointer steals the array owned by the std::unique_ptr. + * + * This constructor works via move semantics. If your std::unique_ptr is + * in a local variable, you must use std::move. + * + * @param p The std::unique_ptr from which the array will be stolen. + * @stable ICU 64 + */ + explicit LocalArray(std::unique_ptr &&p) + : LocalPointerBase(p.release()) {} + + /** + * Destructor deletes the array it owns. + * @stable ICU 4.4 + */ + ~LocalArray() { + delete[] LocalPointerBase::ptr; + } + /** + * Move assignment operator, leaves src with isNull(). + * The behavior is undefined if *this and src are the same object. + * @param src source smart pointer + * @return *this + * @stable ICU 56 + */ + LocalArray &operator=(LocalArray &&src) noexcept { + delete[] LocalPointerBase::ptr; + LocalPointerBase::ptr=src.ptr; + src.ptr=nullptr; + return *this; + } + + /** + * Move-assign from an std::unique_ptr to this LocalPointer. + * Steals the array from the std::unique_ptr. + * + * @param p The std::unique_ptr from which the array will be stolen. + * @return *this + * @stable ICU 64 + */ + LocalArray &operator=(std::unique_ptr &&p) noexcept { + adoptInstead(p.release()); + return *this; + } + + /** + * Swap pointers. + * @param other other smart pointer + * @stable ICU 56 + */ + void swap(LocalArray &other) noexcept { + T *temp=LocalPointerBase::ptr; + LocalPointerBase::ptr=other.ptr; + other.ptr=temp; + } + /** + * Non-member LocalArray swap function. + * @param p1 will get p2's pointer + * @param p2 will get p1's pointer + * @stable ICU 56 + */ + friend inline void swap(LocalArray &p1, LocalArray &p2) noexcept { + p1.swap(p2); + } + /** + * Deletes the array it owns, + * and adopts (takes ownership of) the one passed in. + * @param p simple pointer to an array of T objects that is adopted + * @stable ICU 4.4 + */ + void adoptInstead(T *p) { + delete[] LocalPointerBase::ptr; + LocalPointerBase::ptr=p; + } + /** + * Deletes the array it owns, + * and adopts (takes ownership of) the one passed in. + * + * If U_FAILURE(errorCode), then the current array is retained and the new one deleted. + * + * If U_SUCCESS(errorCode) but the input pointer is nullptr, + * then U_MEMORY_ALLOCATION_ERROR is set, + * the current array is deleted, and nullptr is set. + * + * @param p simple pointer to an array of T objects that is adopted + * @param errorCode in/out UErrorCode, set to U_MEMORY_ALLOCATION_ERROR + * if p==nullptr and no other failure code had been set + * @stable ICU 56 + */ + void adoptInsteadAndCheckErrorCode(T *p, UErrorCode &errorCode) { + if(U_SUCCESS(errorCode)) { + delete[] LocalPointerBase::ptr; + LocalPointerBase::ptr=p; + if(p==nullptr) { + errorCode=U_MEMORY_ALLOCATION_ERROR; + } + } else { + delete[] p; + } + } + /** + * Array item access (writable). + * No index bounds check. + * @param i array index + * @return reference to the array item + * @stable ICU 4.4 + */ + T &operator[](ptrdiff_t i) const { return LocalPointerBase::ptr[i]; } + + /** + * Conversion operator to a C++11 std::unique_ptr. + * Disowns the object and gives it to the returned std::unique_ptr. + * + * This operator works via move semantics. If your LocalPointer is + * in a local variable, you must use std::move. + * + * @return An std::unique_ptr owning the pointer previously owned by this + * icu::LocalPointer. + * @stable ICU 64 + */ + operator std::unique_ptr () && { + return std::unique_ptr(LocalPointerBase::orphan()); + } +}; + +/** + * \def U_DEFINE_LOCAL_OPEN_POINTER + * "Smart pointer" definition macro, deletes objects via the closeFunction. + * Defines a subclass of LocalPointerBase which works just + * like LocalPointer except that this subclass will use the closeFunction + * rather than the C++ delete operator. + * + * Usage example: + * \code + * LocalUCaseMapPointer csm(ucasemap_open(localeID, options, &errorCode)); + * utf8OutLength=ucasemap_utf8ToLower(csm.getAlias(), + * utf8Out, (int32_t)sizeof(utf8Out), + * utf8In, utf8InLength, &errorCode); + * if(U_FAILURE(errorCode)) { return; } // no need to explicitly delete the UCaseMap + * \endcode + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +#define U_DEFINE_LOCAL_OPEN_POINTER(LocalPointerClassName, Type, closeFunction) \ + class LocalPointerClassName : public LocalPointerBase { \ + public: \ + using LocalPointerBase::operator*; \ + using LocalPointerBase::operator->; \ + explicit LocalPointerClassName(Type *p=nullptr) : LocalPointerBase(p) {} \ + LocalPointerClassName(LocalPointerClassName &&src) noexcept \ + : LocalPointerBase(src.ptr) { \ + src.ptr=nullptr; \ + } \ + /* TODO: Be agnostic of the deleter function signature from the user-provided std::unique_ptr? */ \ + explicit LocalPointerClassName(std::unique_ptr &&p) \ + : LocalPointerBase(p.release()) {} \ + ~LocalPointerClassName() { if (ptr != nullptr) { closeFunction(ptr); } } \ + LocalPointerClassName &operator=(LocalPointerClassName &&src) noexcept { \ + if (ptr != nullptr) { closeFunction(ptr); } \ + LocalPointerBase::ptr=src.ptr; \ + src.ptr=nullptr; \ + return *this; \ + } \ + /* TODO: Be agnostic of the deleter function signature from the user-provided std::unique_ptr? */ \ + LocalPointerClassName &operator=(std::unique_ptr &&p) { \ + adoptInstead(p.release()); \ + return *this; \ + } \ + void swap(LocalPointerClassName &other) noexcept { \ + Type *temp=LocalPointerBase::ptr; \ + LocalPointerBase::ptr=other.ptr; \ + other.ptr=temp; \ + } \ + friend inline void swap(LocalPointerClassName &p1, LocalPointerClassName &p2) noexcept { \ + p1.swap(p2); \ + } \ + void adoptInstead(Type *p) { \ + if (ptr != nullptr) { closeFunction(ptr); } \ + ptr=p; \ + } \ + operator std::unique_ptr () && { \ + return std::unique_ptr(LocalPointerBase::orphan(), closeFunction); \ + } \ + } + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ +#endif /* __LOCALPOINTER_H__ */ diff --git a/intl/icu/source/common/unicode/locdspnm.h b/intl/icu/source/common/unicode/locdspnm.h new file mode 100644 index 0000000000..4f06f85704 --- /dev/null +++ b/intl/icu/source/common/unicode/locdspnm.h @@ -0,0 +1,211 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 2010-2016, International Business Machines Corporation and +* others. All Rights Reserved. +****************************************************************************** +*/ + +#ifndef LOCDSPNM_H +#define LOCDSPNM_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +/** + * \file + * \brief C++ API: Provides display names of Locale and its components. + */ + +#if !UCONFIG_NO_FORMATTING + +#include "unicode/locid.h" +#include "unicode/strenum.h" +#include "unicode/uscript.h" +#include "unicode/uldnames.h" +#include "unicode/udisplaycontext.h" + +U_NAMESPACE_BEGIN + +/** + * Returns display names of Locales and components of Locales. For + * more information on language, script, region, variant, key, and + * values, see Locale. + * @stable ICU 4.4 + */ +class U_COMMON_API LocaleDisplayNames : public UObject { +public: + /** + * Destructor. + * @stable ICU 4.4 + */ + virtual ~LocaleDisplayNames(); + + /** + * Convenience overload of + * {@link #createInstance(const Locale& locale, UDialectHandling dialectHandling)} + * that specifies STANDARD dialect handling. + * @param locale the display locale + * @return a LocaleDisplayNames instance + * @stable ICU 4.4 + */ + inline static LocaleDisplayNames* U_EXPORT2 createInstance(const Locale& locale); + + /** + * Returns an instance of LocaleDisplayNames that returns names + * formatted for the provided locale, using the provided + * dialectHandling. + * + * @param locale the display locale + * @param dialectHandling how to select names for locales + * @return a LocaleDisplayNames instance + * @stable ICU 4.4 + */ + static LocaleDisplayNames* U_EXPORT2 createInstance(const Locale& locale, + UDialectHandling dialectHandling); + + /** + * Returns an instance of LocaleDisplayNames that returns names formatted + * for the provided locale, using the provided UDisplayContext settings. + * + * @param locale the display locale + * @param contexts List of one or more context settings (e.g. for dialect + * handling, capitalization, etc. + * @param length Number of items in the contexts list + * @return a LocaleDisplayNames instance + * @stable ICU 51 + */ + static LocaleDisplayNames* U_EXPORT2 createInstance(const Locale& locale, + UDisplayContext *contexts, int32_t length); + + // getters for state + /** + * Returns the locale used to determine the display names. This is + * not necessarily the same locale passed to {@link #createInstance}. + * @return the display locale + * @stable ICU 4.4 + */ + virtual const Locale& getLocale() const = 0; + + /** + * Returns the dialect handling used in the display names. + * @return the dialect handling enum + * @stable ICU 4.4 + */ + virtual UDialectHandling getDialectHandling() const = 0; + + /** + * Returns the UDisplayContext value for the specified UDisplayContextType. + * @param type the UDisplayContextType whose value to return + * @return the UDisplayContext for the specified type. + * @stable ICU 51 + */ + virtual UDisplayContext getContext(UDisplayContextType type) const = 0; + + // names for entire locales + /** + * Returns the display name of the provided locale. + * @param locale the locale whose display name to return + * @param result receives the locale's display name + * @return the display name of the provided locale + * @stable ICU 4.4 + */ + virtual UnicodeString& localeDisplayName(const Locale& locale, + UnicodeString& result) const = 0; + + /** + * Returns the display name of the provided locale id. + * @param localeId the id of the locale whose display name to return + * @param result receives the locale's display name + * @return the display name of the provided locale + * @stable ICU 4.4 + */ + virtual UnicodeString& localeDisplayName(const char* localeId, + UnicodeString& result) const = 0; + + // names for components of a locale id + /** + * Returns the display name of the provided language code. + * @param lang the language code + * @param result receives the language code's display name + * @return the display name of the provided language code + * @stable ICU 4.4 + */ + virtual UnicodeString& languageDisplayName(const char* lang, + UnicodeString& result) const = 0; + + /** + * Returns the display name of the provided script code. + * @param script the script code + * @param result receives the script code's display name + * @return the display name of the provided script code + * @stable ICU 4.4 + */ + virtual UnicodeString& scriptDisplayName(const char* script, + UnicodeString& result) const = 0; + + /** + * Returns the display name of the provided script code. + * @param scriptCode the script code number + * @param result receives the script code's display name + * @return the display name of the provided script code + * @stable ICU 4.4 + */ + virtual UnicodeString& scriptDisplayName(UScriptCode scriptCode, + UnicodeString& result) const = 0; + + /** + * Returns the display name of the provided region code. + * @param region the region code + * @param result receives the region code's display name + * @return the display name of the provided region code + * @stable ICU 4.4 + */ + virtual UnicodeString& regionDisplayName(const char* region, + UnicodeString& result) const = 0; + + /** + * Returns the display name of the provided variant. + * @param variant the variant string + * @param result receives the variant's display name + * @return the display name of the provided variant + * @stable ICU 4.4 + */ + virtual UnicodeString& variantDisplayName(const char* variant, + UnicodeString& result) const = 0; + + /** + * Returns the display name of the provided locale key. + * @param key the locale key name + * @param result receives the locale key's display name + * @return the display name of the provided locale key + * @stable ICU 4.4 + */ + virtual UnicodeString& keyDisplayName(const char* key, + UnicodeString& result) const = 0; + + /** + * Returns the display name of the provided value (used with the provided key). + * @param key the locale key name + * @param value the locale key's value + * @param result receives the value's display name + * @return the display name of the provided value + * @stable ICU 4.4 + */ + virtual UnicodeString& keyValueDisplayName(const char* key, const char* value, + UnicodeString& result) const = 0; +}; + +inline LocaleDisplayNames* LocaleDisplayNames::createInstance(const Locale& locale) { + return LocaleDisplayNames::createInstance(locale, ULDN_STANDARD_NAMES); +} + +U_NAMESPACE_END + +#endif + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/locid.h b/intl/icu/source/common/unicode/locid.h new file mode 100644 index 0000000000..a6fbbb7041 --- /dev/null +++ b/intl/icu/source/common/unicode/locid.h @@ -0,0 +1,1272 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1996-2015, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* File locid.h +* +* Created by: Helena Shih +* +* Modification History: +* +* Date Name Description +* 02/11/97 aliu Changed gLocPath to fgLocPath and added methods to +* get and set it. +* 04/02/97 aliu Made operator!= inline; fixed return value of getName(). +* 04/15/97 aliu Cleanup for AIX/Win32. +* 04/24/97 aliu Numerous changes per code review. +* 08/18/98 stephen Added tokenizeString(),changed getDisplayName() +* 09/08/98 stephen Moved definition of kEmptyString for Mac Port +* 11/09/99 weiv Added const char * getName() const; +* 04/12/00 srl removing unicodestring api's and cached hash code +* 08/10/01 grhoten Change the static Locales to accessor functions +****************************************************************************** +*/ + +#ifndef LOCID_H +#define LOCID_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/bytestream.h" +#include "unicode/localpointer.h" +#include "unicode/strenum.h" +#include "unicode/stringpiece.h" +#include "unicode/uobject.h" +#include "unicode/putil.h" +#include "unicode/uloc.h" + +/** + * \file + * \brief C++ API: Locale ID object. + */ + +U_NAMESPACE_BEGIN + +// Forward Declarations +void U_CALLCONV locale_available_init(); /**< @internal */ + +class StringEnumeration; +class UnicodeString; + +/** + * A Locale object represents a specific geographical, political, + * or cultural region. An operation that requires a Locale to perform + * its task is called locale-sensitive and uses the Locale + * to tailor information for the user. For example, displaying a number + * is a locale-sensitive operation--the number should be formatted + * according to the customs/conventions of the user's native country, + * region, or culture. + * + * The Locale class is not suitable for subclassing. + * + *

+ * You can create a Locale object using the constructor in + * this class: + * \htmlonly

\endhtmlonly + * 
+ *       Locale( const   char*  language,
+ *               const   char*  country,
+ *               const   char*  variant);
+ *
+ * \htmlonly
\endhtmlonly + * The first argument to the constructors is a valid ISO + * Language Code. These codes are the lower-case two-letter + * codes as defined by ISO-639. + * You can find a full list of these codes at: + *
+ * http://www.loc.gov/standards/iso639-2/ + * + *

+ * The second argument to the constructors is a valid ISO Country + * Code. These codes are the upper-case two-letter codes + * as defined by ISO-3166. + * You can find a full list of these codes at a number of sites, such as: + *
+ * http://www.iso.org/iso/en/prods-services/iso3166ma/index.html + * + *

+ * The third constructor requires a third argument--the Variant. + * The Variant codes are vendor and browser-specific. + * For example, use REVISED for a language's revised script orthography, and POSIX for POSIX. + * Where there are two variants, separate them with an underscore, and + * put the most important one first. For + * example, a Traditional Spanish collation might be referenced, with + * "ES", "ES", "Traditional_POSIX". + * + *

+ * Because a Locale object is just an identifier for a region, + * no validity check is performed when you construct a Locale. + * If you want to see whether particular resources are available for the + * Locale you construct, you must query those resources. For + * example, ask the NumberFormat for the locales it supports + * using its getAvailableLocales method. + *
Note: When you ask for a resource for a particular + * locale, you get back the best available match, not necessarily + * precisely what you asked for. For more information, look at + * ResourceBundle. + * + *

+ * The Locale class provides a number of convenient constants + * that you can use to create Locale objects for commonly used + * locales. For example, the following refers to a Locale object + * for the United States: + * \htmlonly

\endhtmlonly + * 
+ *       Locale::getUS()
+ *
+ * \htmlonly
\endhtmlonly + * + *

+ * Once you've created a Locale you can query it for information about + * itself. Use getCountry to get the ISO Country Code and + * getLanguage to get the ISO Language Code. You can + * use getDisplayCountry to get the + * name of the country suitable for displaying to the user. Similarly, + * you can use getDisplayLanguage to get the name of + * the language suitable for displaying to the user. Interestingly, + * the getDisplayXXX methods are themselves locale-sensitive + * and have two versions: one that uses the default locale and one + * that takes a locale as an argument and displays the name or country in + * a language appropriate to that locale. + * + *

+ * ICU provides a number of classes that perform locale-sensitive + * operations. For example, the NumberFormat class formats + * numbers, currency, or percentages in a locale-sensitive manner. Classes + * such as NumberFormat have a number of convenience methods + * for creating a default object of that type. For example, the + * NumberFormat class provides these three convenience methods + * for creating a default NumberFormat object: + * \htmlonly

\endhtmlonly + * 
+ *     UErrorCode success = U_ZERO_ERROR;
+ *     Locale myLocale;
+ *     NumberFormat *nf;
+ *
+ *     nf = NumberFormat::createInstance( success );          delete nf;
+ *     nf = NumberFormat::createCurrencyInstance( success );  delete nf;
+ *     nf = NumberFormat::createPercentInstance( success );   delete nf;
+ *
+ * \htmlonly
\endhtmlonly + * Each of these methods has two variants; one with an explicit locale + * and one without; the latter using the default locale. + * \htmlonly
\endhtmlonly + * 
+ *     nf = NumberFormat::createInstance( myLocale, success );          delete nf;
+ *     nf = NumberFormat::createCurrencyInstance( myLocale, success );  delete nf;
+ *     nf = NumberFormat::createPercentInstance( myLocale, success );   delete nf;
+ *
+ * \htmlonly
\endhtmlonly + * A Locale is the mechanism for identifying the kind of object + * (NumberFormat) that you would like to get. The locale is + * just a mechanism for identifying objects, + * not a container for the objects themselves. + * + *

+ * Each class that performs locale-sensitive operations allows you + * to get all the available objects of that type. You can sift + * through these objects by language, country, or variant, + * and use the display names to present a menu to the user. + * For example, you can create a menu of all the collation objects + * suitable for a given language. Such classes implement these + * three class methods: + * \htmlonly

\endhtmlonly + * 
+ *       static Locale* getAvailableLocales(int32_t& numLocales)
+ *       static UnicodeString& getDisplayName(const Locale&  objectLocale,
+ *                                            const Locale&  displayLocale,
+ *                                            UnicodeString& displayName)
+ *       static UnicodeString& getDisplayName(const Locale&  objectLocale,
+ *                                            UnicodeString& displayName)
+ *
+ * \htmlonly
\endhtmlonly + * + * @stable ICU 2.0 + * @see ResourceBundle + */ +class U_COMMON_API Locale : public UObject { +public: + /** Useful constant for the Root locale. @stable ICU 4.4 */ + static const Locale &U_EXPORT2 getRoot(void); + /** Useful constant for this language. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getEnglish(void); + /** Useful constant for this language. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getFrench(void); + /** Useful constant for this language. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getGerman(void); + /** Useful constant for this language. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getItalian(void); + /** Useful constant for this language. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getJapanese(void); + /** Useful constant for this language. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getKorean(void); + /** Useful constant for this language. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getChinese(void); + /** Useful constant for this language. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getSimplifiedChinese(void); + /** Useful constant for this language. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getTraditionalChinese(void); + + /** Useful constant for this country/region. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getFrance(void); + /** Useful constant for this country/region. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getGermany(void); + /** Useful constant for this country/region. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getItaly(void); + /** Useful constant for this country/region. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getJapan(void); + /** Useful constant for this country/region. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getKorea(void); + /** Useful constant for this country/region. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getChina(void); + /** Useful constant for this country/region. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getPRC(void); + /** Useful constant for this country/region. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getTaiwan(void); + /** Useful constant for this country/region. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getUK(void); + /** Useful constant for this country/region. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getUS(void); + /** Useful constant for this country/region. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getCanada(void); + /** Useful constant for this country/region. @stable ICU 2.0 */ + static const Locale &U_EXPORT2 getCanadaFrench(void); + + + /** + * Construct a default locale object, a Locale for the default locale ID. + * + * @see getDefault + * @see uloc_getDefault + * @stable ICU 2.0 + */ + Locale(); + + /** + * Construct a locale from language, country, variant. + * If an error occurs, then the constructed object will be "bogus" + * (isBogus() will return true). + * + * @param language Lowercase two-letter or three-letter ISO-639 code. + * This parameter can instead be an ICU style C locale (e.g. "en_US"), + * but the other parameters must not be used. + * This parameter can be nullptr; if so, + * the locale is initialized to match the current default locale. + * (This is the same as using the default constructor.) + * Please note: The Java Locale class does NOT accept the form + * 'new Locale("en_US")' but only 'new Locale("en","US")' + * + * @param country Uppercase two-letter ISO-3166 code. (optional) + * @param variant Uppercase vendor and browser specific code. See class + * description. (optional) + * @param keywordsAndValues A string consisting of keyword/values pairs, such as + * "collation=phonebook;currency=euro" + * + * @see getDefault + * @see uloc_getDefault + * @stable ICU 2.0 + */ + Locale( const char * language, + const char * country = 0, + const char * variant = 0, + const char * keywordsAndValues = 0); + + /** + * Initializes a Locale object from another Locale object. + * + * @param other The Locale object being copied in. + * @stable ICU 2.0 + */ + Locale(const Locale& other); + + /** + * Move constructor; might leave source in bogus state. + * This locale will have the same contents that the source locale had. + * + * @param other The Locale object being moved in. + * @stable ICU 63 + */ + Locale(Locale&& other) noexcept; + + /** + * Destructor + * @stable ICU 2.0 + */ + virtual ~Locale() ; + + /** + * Replaces the entire contents of *this with the specified value. + * + * @param other The Locale object being copied in. + * @return *this + * @stable ICU 2.0 + */ + Locale& operator=(const Locale& other); + + /** + * Move assignment operator; might leave source in bogus state. + * This locale will have the same contents that the source locale had. + * The behavior is undefined if *this and the source are the same object. + * + * @param other The Locale object being moved in. + * @return *this + * @stable ICU 63 + */ + Locale& operator=(Locale&& other) noexcept; + + /** + * Checks if two locale keys are the same. + * + * @param other The locale key object to be compared with this. + * @return true if the two locale keys are the same, false otherwise. + * @stable ICU 2.0 + */ + bool operator==(const Locale& other) const; + + /** + * Checks if two locale keys are not the same. + * + * @param other The locale key object to be compared with this. + * @return true if the two locale keys are not the same, false + * otherwise. + * @stable ICU 2.0 + */ + inline bool operator!=(const Locale& other) const; + + /** + * Clone this object. + * Clones can be used concurrently in multiple threads. + * If an error occurs, then nullptr is returned. + * The caller must delete the clone. + * + * @return a clone of this object + * + * @see getDynamicClassID + * @stable ICU 2.8 + */ + Locale *clone() const; + +#ifndef U_HIDE_SYSTEM_API + /** + * Common methods of getting the current default Locale. Used for the + * presentation: menus, dialogs, etc. Generally set once when your applet or + * application is initialized, then never reset. (If you do reset the + * default locale, you probably want to reload your GUI, so that the change + * is reflected in your interface.) + * + * More advanced programs will allow users to use different locales for + * different fields, e.g. in a spreadsheet. + * + * Note that the initial setting will match the host system. + * @return a reference to the Locale object for the default locale ID + * @system + * @stable ICU 2.0 + */ + static const Locale& U_EXPORT2 getDefault(void); + + /** + * Sets the default. Normally set once at the beginning of a process, + * then never reset. + * setDefault() only changes ICU's default locale ID, not + * the default locale ID of the runtime environment. + * + * @param newLocale Locale to set to. If nullptr, set to the value obtained + * from the runtime environment. + * @param success The error code. + * @system + * @stable ICU 2.0 + */ + static void U_EXPORT2 setDefault(const Locale& newLocale, + UErrorCode& success); +#endif /* U_HIDE_SYSTEM_API */ + + /** + * Returns a Locale for the specified BCP47 language tag string. + * If the specified language tag contains any ill-formed subtags, + * the first such subtag and all following subtags are ignored. + *

+ * This implements the 'Language-Tag' production of BCP 47, and so + * supports legacy language tags (marked as “Type: grandfathered” in BCP 47) + * (regular and irregular) as well as private use language tags. + * + * Private use tags are represented as 'x-whatever', + * and legacy tags are converted to their canonical replacements where they exist. + * + * Note that a few legacy tags have no modern replacement; + * these will be converted using the fallback described in + * the first paragraph, so some information might be lost. + * + * @param tag the input BCP47 language tag. + * @param status error information if creating the Locale failed. + * @return the Locale for the specified BCP47 language tag. + * @stable ICU 63 + */ + static Locale U_EXPORT2 forLanguageTag(StringPiece tag, UErrorCode& status); + + /** + * Returns a well-formed language tag for this Locale. + *

+ * Note: Any locale fields which do not satisfy the BCP47 syntax + * requirement will be silently omitted from the result. + * + * If this function fails, partial output may have been written to the sink. + * + * @param sink the output sink receiving the BCP47 language + * tag for this Locale. + * @param status error information if creating the language tag failed. + * @stable ICU 63 + */ + void toLanguageTag(ByteSink& sink, UErrorCode& status) const; + + /** + * Returns a well-formed language tag for this Locale. + *

+ * Note: Any locale fields which do not satisfy the BCP47 syntax + * requirement will be silently omitted from the result. + * + * @param status error information if creating the language tag failed. + * @return the BCP47 language tag for this Locale. + * @stable ICU 63 + */ + template + inline StringClass toLanguageTag(UErrorCode& status) const; + + /** + * Creates a locale which has had minimal canonicalization + * as per uloc_getName(). + * @param name The name to create from. If name is null, + * the default Locale is used. + * @return new locale object + * @stable ICU 2.0 + * @see uloc_getName + */ + static Locale U_EXPORT2 createFromName(const char *name); + + /** + * Creates a locale from the given string after canonicalizing + * the string according to CLDR by calling uloc_canonicalize(). + * @param name the locale ID to create from. Must not be nullptr. + * @return a new locale object corresponding to the given name + * @stable ICU 3.0 + * @see uloc_canonicalize + */ + static Locale U_EXPORT2 createCanonical(const char* name); + + /** + * Returns the locale's ISO-639 language code. + * @return An alias to the code + * @stable ICU 2.0 + */ + inline const char * getLanguage( ) const; + + /** + * Returns the locale's ISO-15924 abbreviation script code. + * @return An alias to the code + * @see uscript_getShortName + * @see uscript_getCode + * @stable ICU 2.8 + */ + inline const char * getScript( ) const; + + /** + * Returns the locale's ISO-3166 country code. + * @return An alias to the code + * @stable ICU 2.0 + */ + inline const char * getCountry( ) const; + + /** + * Returns the locale's variant code. + * @return An alias to the code + * @stable ICU 2.0 + */ + inline const char * getVariant( ) const; + + /** + * Returns the programmatic name of the entire locale, with the language, + * country and variant separated by underbars. If a field is missing, up + * to two leading underbars will occur. Example: "en", "de_DE", "en_US_WIN", + * "de__POSIX", "fr__MAC", "__MAC", "_MT", "_FR_EURO" + * @return A pointer to "name". + * @stable ICU 2.0 + */ + inline const char * getName() const; + + /** + * Returns the programmatic name of the entire locale as getName() would return, + * but without keywords. + * @return A pointer to "name". + * @see getName + * @stable ICU 2.8 + */ + const char * getBaseName() const; + + /** + * Add the likely subtags for this Locale, per the algorithm described + * in the following CLDR technical report: + * + * http://www.unicode.org/reports/tr35/#Likely_Subtags + * + * If this Locale is already in the maximal form, or not valid, or there is + * no data available for maximization, the Locale will be unchanged. + * + * For example, "und-Zzzz" cannot be maximized, since there is no + * reasonable maximization. + * + * Examples: + * + * "en" maximizes to "en_Latn_US" + * + * "de" maximizes to "de_Latn_US" + * + * "sr" maximizes to "sr_Cyrl_RS" + * + * "sh" maximizes to "sr_Latn_RS" (Note this will not reverse.) + * + * "zh_Hani" maximizes to "zh_Hans_CN" (Note this will not reverse.) + * + * @param status error information if maximizing this Locale failed. + * If this Locale is not well-formed, the error code is + * U_ILLEGAL_ARGUMENT_ERROR. + * @stable ICU 63 + */ + void addLikelySubtags(UErrorCode& status); + + /** + * Minimize the subtags for this Locale, per the algorithm described + * in the following CLDR technical report: + * + * http://www.unicode.org/reports/tr35/#Likely_Subtags + * + * If this Locale is already in the minimal form, or not valid, or there is + * no data available for minimization, the Locale will be unchanged. + * + * Since the minimization algorithm relies on proper maximization, see the + * comments for addLikelySubtags for reasons why there might not be any + * data. + * + * Examples: + * + * "en_Latn_US" minimizes to "en" + * + * "de_Latn_US" minimizes to "de" + * + * "sr_Cyrl_RS" minimizes to "sr" + * + * "zh_Hant_TW" minimizes to "zh_TW" (The region is preferred to the + * script, and minimizing to "zh" would imply "zh_Hans_CN".) + * + * @param status error information if maximizing this Locale failed. + * If this Locale is not well-formed, the error code is + * U_ILLEGAL_ARGUMENT_ERROR. + * @stable ICU 63 + */ + void minimizeSubtags(UErrorCode& status); + + /** + * Canonicalize the locale ID of this object according to CLDR. + * @param status the status code + * @stable ICU 67 + * @see createCanonical + */ + void canonicalize(UErrorCode& status); + + /** + * Gets the list of keywords for the specified locale. + * + * @param status the status code + * @return pointer to StringEnumeration class, or nullptr if there are no keywords. + * Client must dispose of it by calling delete. + * @see getKeywords + * @stable ICU 2.8 + */ + StringEnumeration * createKeywords(UErrorCode &status) const; + + /** + * Gets the list of Unicode keywords for the specified locale. + * + * @param status the status code + * @return pointer to StringEnumeration class, or nullptr if there are no keywords. + * Client must dispose of it by calling delete. + * @see getUnicodeKeywords + * @stable ICU 63 + */ + StringEnumeration * createUnicodeKeywords(UErrorCode &status) const; + + /** + * Gets the set of keywords for this Locale. + * + * A wrapper to call createKeywords() and write the resulting + * keywords as standard strings (or compatible objects) into any kind of + * container that can be written to by an STL style output iterator. + * + * @param iterator an STL style output iterator to write the keywords to. + * @param status error information if creating set of keywords failed. + * @stable ICU 63 + */ + template + inline void getKeywords(OutputIterator iterator, UErrorCode& status) const; + + /** + * Gets the set of Unicode keywords for this Locale. + * + * A wrapper to call createUnicodeKeywords() and write the resulting + * keywords as standard strings (or compatible objects) into any kind of + * container that can be written to by an STL style output iterator. + * + * @param iterator an STL style output iterator to write the keywords to. + * @param status error information if creating set of keywords failed. + * @stable ICU 63 + */ + template + inline void getUnicodeKeywords(OutputIterator iterator, UErrorCode& status) const; + + /** + * Gets the value for a keyword. + * + * This uses legacy keyword=value pairs, like "collation=phonebook". + * + * ICU4C doesn't do automatic conversion between legacy and Unicode + * keywords and values in getters and setters (as opposed to ICU4J). + * + * @param keywordName name of the keyword for which we want the value. Case insensitive. + * @param buffer The buffer to receive the keyword value. + * @param bufferCapacity The capacity of receiving buffer + * @param status Returns any error information while performing this operation. + * @return the length of the keyword value + * + * @stable ICU 2.8 + */ + int32_t getKeywordValue(const char* keywordName, char *buffer, int32_t bufferCapacity, UErrorCode &status) const; + + /** + * Gets the value for a keyword. + * + * This uses legacy keyword=value pairs, like "collation=phonebook". + * + * ICU4C doesn't do automatic conversion between legacy and Unicode + * keywords and values in getters and setters (as opposed to ICU4J). + * + * @param keywordName name of the keyword for which we want the value. + * @param sink the sink to receive the keyword value. + * @param status error information if getting the value failed. + * @stable ICU 63 + */ + void getKeywordValue(StringPiece keywordName, ByteSink& sink, UErrorCode& status) const; + + /** + * Gets the value for a keyword. + * + * This uses legacy keyword=value pairs, like "collation=phonebook". + * + * ICU4C doesn't do automatic conversion between legacy and Unicode + * keywords and values in getters and setters (as opposed to ICU4J). + * + * @param keywordName name of the keyword for which we want the value. + * @param status error information if getting the value failed. + * @return the keyword value. + * @stable ICU 63 + */ + template + inline StringClass getKeywordValue(StringPiece keywordName, UErrorCode& status) const; + + /** + * Gets the Unicode value for a Unicode keyword. + * + * This uses Unicode key-value pairs, like "co-phonebk". + * + * ICU4C doesn't do automatic conversion between legacy and Unicode + * keywords and values in getters and setters (as opposed to ICU4J). + * + * @param keywordName name of the keyword for which we want the value. + * @param sink the sink to receive the keyword value. + * @param status error information if getting the value failed. + * @stable ICU 63 + */ + void getUnicodeKeywordValue(StringPiece keywordName, ByteSink& sink, UErrorCode& status) const; + + /** + * Gets the Unicode value for a Unicode keyword. + * + * This uses Unicode key-value pairs, like "co-phonebk". + * + * ICU4C doesn't do automatic conversion between legacy and Unicode + * keywords and values in getters and setters (as opposed to ICU4J). + * + * @param keywordName name of the keyword for which we want the value. + * @param status error information if getting the value failed. + * @return the keyword value. + * @stable ICU 63 + */ + template + inline StringClass getUnicodeKeywordValue(StringPiece keywordName, UErrorCode& status) const; + + /** + * Sets or removes the value for a keyword. + * + * For removing all keywords, use getBaseName(), + * and construct a new Locale if it differs from getName(). + * + * This uses legacy keyword=value pairs, like "collation=phonebook". + * + * ICU4C doesn't do automatic conversion between legacy and Unicode + * keywords and values in getters and setters (as opposed to ICU4J). + * + * @param keywordName name of the keyword to be set. Case insensitive. + * @param keywordValue value of the keyword to be set. If 0-length or + * nullptr, will result in the keyword being removed. No error is given if + * that keyword does not exist. + * @param status Returns any error information while performing this operation. + * + * @stable ICU 49 + */ + void setKeywordValue(const char* keywordName, const char* keywordValue, UErrorCode &status); + + /** + * Sets or removes the value for a keyword. + * + * For removing all keywords, use getBaseName(), + * and construct a new Locale if it differs from getName(). + * + * This uses legacy keyword=value pairs, like "collation=phonebook". + * + * ICU4C doesn't do automatic conversion between legacy and Unicode + * keywords and values in getters and setters (as opposed to ICU4J). + * + * @param keywordName name of the keyword to be set. + * @param keywordValue value of the keyword to be set. If 0-length or + * nullptr, will result in the keyword being removed. No error is given if + * that keyword does not exist. + * @param status Returns any error information while performing this operation. + * @stable ICU 63 + */ + void setKeywordValue(StringPiece keywordName, StringPiece keywordValue, UErrorCode& status); + + /** + * Sets or removes the Unicode value for a Unicode keyword. + * + * For removing all keywords, use getBaseName(), + * and construct a new Locale if it differs from getName(). + * + * This uses Unicode key-value pairs, like "co-phonebk". + * + * ICU4C doesn't do automatic conversion between legacy and Unicode + * keywords and values in getters and setters (as opposed to ICU4J). + * + * @param keywordName name of the keyword to be set. + * @param keywordValue value of the keyword to be set. If 0-length or + * nullptr, will result in the keyword being removed. No error is given if + * that keyword does not exist. + * @param status Returns any error information while performing this operation. + * @stable ICU 63 + */ + void setUnicodeKeywordValue(StringPiece keywordName, StringPiece keywordValue, UErrorCode& status); + + /** + * returns the locale's three-letter language code, as specified + * in ISO draft standard ISO-639-2. + * @return An alias to the code, or an empty string + * @stable ICU 2.0 + */ + const char * getISO3Language() const; + + /** + * Fills in "name" with the locale's three-letter ISO-3166 country code. + * @return An alias to the code, or an empty string + * @stable ICU 2.0 + */ + const char * getISO3Country() const; + + /** + * Returns the Windows LCID value corresponding to this locale. + * This value is stored in the resource data for the locale as a one-to-four-digit + * hexadecimal number. If the resource is missing, in the wrong format, or + * there is no Windows LCID value that corresponds to this locale, returns 0. + * @stable ICU 2.0 + */ + uint32_t getLCID(void) const; + + /** + * Returns whether this locale's script is written right-to-left. + * If there is no script subtag, then the likely script is used, see uloc_addLikelySubtags(). + * If no likely script is known, then false is returned. + * + * A script is right-to-left according to the CLDR script metadata + * which corresponds to whether the script's letters have Bidi_Class=R or AL. + * + * Returns true for "ar" and "en-Hebr", false for "zh" and "fa-Cyrl". + * + * @return true if the locale's script is written right-to-left + * @stable ICU 54 + */ + UBool isRightToLeft() const; + + /** + * Fills in "dispLang" with the name of this locale's language in a format suitable for + * user display in the default locale. For example, if the locale's language code is + * "fr" and the default locale's language code is "en", this function would set + * dispLang to "French". + * @param dispLang Receives the language's display name. + * @return A reference to "dispLang". + * @stable ICU 2.0 + */ + UnicodeString& getDisplayLanguage(UnicodeString& dispLang) const; + + /** + * Fills in "dispLang" with the name of this locale's language in a format suitable for + * user display in the locale specified by "displayLocale". For example, if the locale's + * language code is "en" and displayLocale's language code is "fr", this function would set + * dispLang to "Anglais". + * @param displayLocale Specifies the locale to be used to display the name. In other words, + * if the locale's language code is "en", passing Locale::getFrench() for + * displayLocale would result in "Anglais", while passing Locale::getGerman() + * for displayLocale would result in "Englisch". + * @param dispLang Receives the language's display name. + * @return A reference to "dispLang". + * @stable ICU 2.0 + */ + UnicodeString& getDisplayLanguage( const Locale& displayLocale, + UnicodeString& dispLang) const; + + /** + * Fills in "dispScript" with the name of this locale's script in a format suitable + * for user display in the default locale. For example, if the locale's script code + * is "LATN" and the default locale's language code is "en", this function would set + * dispScript to "Latin". + * @param dispScript Receives the scripts's display name. + * @return A reference to "dispScript". + * @stable ICU 2.8 + */ + UnicodeString& getDisplayScript( UnicodeString& dispScript) const; + + /** + * Fills in "dispScript" with the name of this locale's country in a format suitable + * for user display in the locale specified by "displayLocale". For example, if the locale's + * script code is "LATN" and displayLocale's language code is "en", this function would set + * dispScript to "Latin". + * @param displayLocale Specifies the locale to be used to display the name. In other + * words, if the locale's script code is "LATN", passing + * Locale::getFrench() for displayLocale would result in "", while + * passing Locale::getGerman() for displayLocale would result in + * "". + * @param dispScript Receives the scripts's display name. + * @return A reference to "dispScript". + * @stable ICU 2.8 + */ + UnicodeString& getDisplayScript( const Locale& displayLocale, + UnicodeString& dispScript) const; + + /** + * Fills in "dispCountry" with the name of this locale's country in a format suitable + * for user display in the default locale. For example, if the locale's country code + * is "FR" and the default locale's language code is "en", this function would set + * dispCountry to "France". + * @param dispCountry Receives the country's display name. + * @return A reference to "dispCountry". + * @stable ICU 2.0 + */ + UnicodeString& getDisplayCountry( UnicodeString& dispCountry) const; + + /** + * Fills in "dispCountry" with the name of this locale's country in a format suitable + * for user display in the locale specified by "displayLocale". For example, if the locale's + * country code is "US" and displayLocale's language code is "fr", this function would set + * dispCountry to "États-Unis". + * @param displayLocale Specifies the locale to be used to display the name. In other + * words, if the locale's country code is "US", passing + * Locale::getFrench() for displayLocale would result in "États-Unis", while + * passing Locale::getGerman() for displayLocale would result in + * "Vereinigte Staaten". + * @param dispCountry Receives the country's display name. + * @return A reference to "dispCountry". + * @stable ICU 2.0 + */ + UnicodeString& getDisplayCountry( const Locale& displayLocale, + UnicodeString& dispCountry) const; + + /** + * Fills in "dispVar" with the name of this locale's variant code in a format suitable + * for user display in the default locale. + * @param dispVar Receives the variant's name. + * @return A reference to "dispVar". + * @stable ICU 2.0 + */ + UnicodeString& getDisplayVariant( UnicodeString& dispVar) const; + + /** + * Fills in "dispVar" with the name of this locale's variant code in a format + * suitable for user display in the locale specified by "displayLocale". + * @param displayLocale Specifies the locale to be used to display the name. + * @param dispVar Receives the variant's display name. + * @return A reference to "dispVar". + * @stable ICU 2.0 + */ + UnicodeString& getDisplayVariant( const Locale& displayLocale, + UnicodeString& dispVar) const; + + /** + * Fills in "name" with the name of this locale in a format suitable for user display + * in the default locale. This function uses getDisplayLanguage(), getDisplayCountry(), + * and getDisplayVariant() to do its work, and outputs the display name in the format + * "language (country[,variant])". For example, if the default locale is en_US, then + * fr_FR's display name would be "French (France)", and es_MX_Traditional's display name + * would be "Spanish (Mexico,Traditional)". + * @param name Receives the locale's display name. + * @return A reference to "name". + * @stable ICU 2.0 + */ + UnicodeString& getDisplayName( UnicodeString& name) const; + + /** + * Fills in "name" with the name of this locale in a format suitable for user display + * in the locale specified by "displayLocale". This function uses getDisplayLanguage(), + * getDisplayCountry(), and getDisplayVariant() to do its work, and outputs the display + * name in the format "language (country[,variant])". For example, if displayLocale is + * fr_FR, then en_US's display name would be "Anglais (États-Unis)", and no_NO_NY's + * display name would be "norvégien (Norvège,NY)". + * @param displayLocale Specifies the locale to be used to display the name. + * @param name Receives the locale's display name. + * @return A reference to "name". + * @stable ICU 2.0 + */ + UnicodeString& getDisplayName( const Locale& displayLocale, + UnicodeString& name) const; + + /** + * Generates a hash code for the locale. + * @stable ICU 2.0 + */ + int32_t hashCode(void) const; + + /** + * Sets the locale to bogus + * A bogus locale represents a non-existing locale associated + * with services that can be instantiated from non-locale data + * in addition to locale (for example, collation can be + * instantiated from a locale and from a rule set). + * @stable ICU 2.1 + */ + void setToBogus(); + + /** + * Gets the bogus state. Locale object can be bogus if it doesn't exist + * @return false if it is a real locale, true if it is a bogus locale + * @stable ICU 2.1 + */ + inline UBool isBogus(void) const; + + /** + * Returns a list of all installed locales. + * @param count Receives the number of locales in the list. + * @return A pointer to an array of Locale objects. This array is the list + * of all locales with installed resource files. The called does NOT + * get ownership of this list, and must NOT delete it. + * @stable ICU 2.0 + */ + static const Locale* U_EXPORT2 getAvailableLocales(int32_t& count); + + /** + * Gets a list of all available 2-letter country codes defined in ISO 3166. This is a + * pointer to an array of pointers to arrays of char. All of these pointers are + * owned by ICU-- do not delete them, and do not write through them. The array is + * terminated with a null pointer. + * @return a list of all available country codes + * @stable ICU 2.0 + */ + static const char* const* U_EXPORT2 getISOCountries(); + + /** + * Gets a list of all available language codes defined in ISO 639. This is a pointer + * to an array of pointers to arrays of char. All of these pointers are owned + * by ICU-- do not delete them, and do not write through them. The array is + * terminated with a null pointer. + * @return a list of all available language codes + * @stable ICU 2.0 + */ + static const char* const* U_EXPORT2 getISOLanguages(); + + /** + * ICU "poor man's RTTI", returns a UClassID for this class. + * + * @stable ICU 2.2 + */ + static UClassID U_EXPORT2 getStaticClassID(); + + /** + * ICU "poor man's RTTI", returns a UClassID for the actual class. + * + * @stable ICU 2.2 + */ + virtual UClassID getDynamicClassID() const override; + + /** + * A Locale iterator interface similar to a Java Iterator. + * @stable ICU 65 + */ + class U_COMMON_API Iterator /* not : public UObject because this is an interface/mixin class */ { + public: + /** @stable ICU 65 */ + virtual ~Iterator(); + + /** + * @return true if next() can be called again. + * @stable ICU 65 + */ + virtual UBool hasNext() const = 0; + + /** + * @return the next locale. + * @stable ICU 65 + */ + virtual const Locale &next() = 0; + }; + + /** + * A generic Locale iterator implementation over Locale input iterators. + * @stable ICU 65 + */ + template + class RangeIterator : public Iterator, public UMemory { + public: + /** + * Constructs an iterator from a begin/end range. + * Each of the iterator parameter values must be an + * input iterator whose value is convertible to const Locale &. + * + * @param begin Start of range. + * @param end Exclusive end of range. + * @stable ICU 65 + */ + RangeIterator(Iter begin, Iter end) : it_(begin), end_(end) {} + + /** + * @return true if next() can be called again. + * @stable ICU 65 + */ + UBool hasNext() const override { return it_ != end_; } + + /** + * @return the next locale. + * @stable ICU 65 + */ + const Locale &next() override { return *it_++; } + + private: + Iter it_; + const Iter end_; + }; + + /** + * A generic Locale iterator implementation over Locale input iterators. + * Calls the converter to convert each *begin to a const Locale &. + * @stable ICU 65 + */ + template + class ConvertingIterator : public Iterator, public UMemory { + public: + /** + * Constructs an iterator from a begin/end range. + * Each of the iterator parameter values must be an + * input iterator whose value the converter converts to const Locale &. + * + * @param begin Start of range. + * @param end Exclusive end of range. + * @param converter Converter from *begin to const Locale & or compatible. + * @stable ICU 65 + */ + ConvertingIterator(Iter begin, Iter end, Conv converter) : + it_(begin), end_(end), converter_(converter) {} + + /** + * @return true if next() can be called again. + * @stable ICU 65 + */ + UBool hasNext() const override { return it_ != end_; } + + /** + * @return the next locale. + * @stable ICU 65 + */ + const Locale &next() override { return converter_(*it_++); } + + private: + Iter it_; + const Iter end_; + Conv converter_; + }; + +protected: /* only protected for testing purposes. DO NOT USE. */ +#ifndef U_HIDE_INTERNAL_API + /** + * Set this from a single POSIX style locale string. + * @internal + */ + void setFromPOSIXID(const char *posixID); +#endif /* U_HIDE_INTERNAL_API */ + +private: + /** + * Initialize the locale object with a new name. + * Was deprecated - used in implementation - moved internal + * + * @param cLocaleID The new locale name. + * @param canonicalize whether to call uloc_canonicalize on cLocaleID + */ + Locale& init(const char* cLocaleID, UBool canonicalize); + + /* + * Internal constructor to allow construction of a locale object with + * NO side effects. (Default constructor tries to get + * the default locale.) + */ + enum ELocaleType { + eBOGUS + }; + Locale(ELocaleType); + + /** + * Initialize the locale cache for commonly used locales + */ + static Locale *getLocaleCache(void); + + char language[ULOC_LANG_CAPACITY]; + char script[ULOC_SCRIPT_CAPACITY]; + char country[ULOC_COUNTRY_CAPACITY]; + int32_t variantBegin; + char* fullName; + char fullNameBuffer[ULOC_FULLNAME_CAPACITY]; + // name without keywords + char* baseName; + void initBaseName(UErrorCode& status); + + UBool fIsBogus; + + static const Locale &getLocale(int locid); + + /** + * A friend to allow the default locale to be set by either the C or C++ API. + * @internal (private) + */ + friend Locale *locale_set_default_internal(const char *, UErrorCode& status); + + /** + * @internal (private) + */ + friend void U_CALLCONV locale_available_init(); +}; + +inline bool +Locale::operator!=(const Locale& other) const +{ + return !operator==(other); +} + +template inline StringClass +Locale::toLanguageTag(UErrorCode& status) const +{ + StringClass result; + StringByteSink sink(&result); + toLanguageTag(sink, status); + return result; +} + +inline const char * +Locale::getCountry() const +{ + return country; +} + +inline const char * +Locale::getLanguage() const +{ + return language; +} + +inline const char * +Locale::getScript() const +{ + return script; +} + +inline const char * +Locale::getVariant() const +{ + return &baseName[variantBegin]; +} + +inline const char * +Locale::getName() const +{ + return fullName; +} + +template inline void +Locale::getKeywords(OutputIterator iterator, UErrorCode& status) const +{ + LocalPointer keys(createKeywords(status)); + if (U_FAILURE(status) || keys.isNull()) { + return; + } + for (;;) { + int32_t resultLength; + const char* buffer = keys->next(&resultLength, status); + if (U_FAILURE(status) || buffer == nullptr) { + return; + } + *iterator++ = StringClass(buffer, resultLength); + } +} + +template inline void +Locale::getUnicodeKeywords(OutputIterator iterator, UErrorCode& status) const +{ + LocalPointer keys(createUnicodeKeywords(status)); + if (U_FAILURE(status) || keys.isNull()) { + return; + } + for (;;) { + int32_t resultLength; + const char* buffer = keys->next(&resultLength, status); + if (U_FAILURE(status) || buffer == nullptr) { + return; + } + *iterator++ = StringClass(buffer, resultLength); + } +} + +template inline StringClass +Locale::getKeywordValue(StringPiece keywordName, UErrorCode& status) const +{ + StringClass result; + StringByteSink sink(&result); + getKeywordValue(keywordName, sink, status); + return result; +} + +template inline StringClass +Locale::getUnicodeKeywordValue(StringPiece keywordName, UErrorCode& status) const +{ + StringClass result; + StringByteSink sink(&result); + getUnicodeKeywordValue(keywordName, sink, status); + return result; +} + +inline UBool +Locale::isBogus(void) const { + return fIsBogus; +} + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/messagepattern.h b/intl/icu/source/common/unicode/messagepattern.h new file mode 100644 index 0000000000..55b09bfbd4 --- /dev/null +++ b/intl/icu/source/common/unicode/messagepattern.h @@ -0,0 +1,949 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2011-2013, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: messagepattern.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2011mar14 +* created by: Markus W. Scherer +*/ + +#ifndef __MESSAGEPATTERN_H__ +#define __MESSAGEPATTERN_H__ + +/** + * \file + * \brief C++ API: MessagePattern class: Parses and represents ICU MessageFormat patterns. + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#if !UCONFIG_NO_FORMATTING + +#include "unicode/parseerr.h" +#include "unicode/unistr.h" + +/** + * Mode for when an apostrophe starts quoted literal text for MessageFormat output. + * The default is DOUBLE_OPTIONAL unless overridden via uconfig.h + * (UCONFIG_MSGPAT_DEFAULT_APOSTROPHE_MODE). + *

+ * A pair of adjacent apostrophes always results in a single apostrophe in the output, + * even when the pair is between two single, text-quoting apostrophes. + *

+ * The following table shows examples of desired MessageFormat.format() output + * with the pattern strings that yield that output. + *

+ * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + *
Desired outputDOUBLE_OPTIONALDOUBLE_REQUIRED
I see {many}I see '{many}'(same)
I said {'Wow!'}I said '{''Wow!''}'(same)
I don't knowI don't know OR
I don''t know		I don''t know
+ * @stable ICU 4.8 + * @see UCONFIG_MSGPAT_DEFAULT_APOSTROPHE_MODE + */ +enum UMessagePatternApostropheMode { + /** + * A literal apostrophe is represented by + * either a single or a double apostrophe pattern character. + * Within a MessageFormat pattern, a single apostrophe only starts quoted literal text + * if it immediately precedes a curly brace {}, + * or a pipe symbol | if inside a choice format, + * or a pound symbol # if inside a plural format. + *

+ * This is the default behavior starting with ICU 4.8. + * @stable ICU 4.8 + */ + UMSGPAT_APOS_DOUBLE_OPTIONAL, + /** + * A literal apostrophe must be represented by + * a double apostrophe pattern character. + * A single apostrophe always starts quoted literal text. + *

+ * This is the behavior of ICU 4.6 and earlier, and of the JDK. + * @stable ICU 4.8 + */ + UMSGPAT_APOS_DOUBLE_REQUIRED +}; +/** + * @stable ICU 4.8 + */ +typedef enum UMessagePatternApostropheMode UMessagePatternApostropheMode; + +/** + * MessagePattern::Part type constants. + * @stable ICU 4.8 + */ +enum UMessagePatternPartType { + /** + * Start of a message pattern (main or nested). + * The length is 0 for the top-level message + * and for a choice argument sub-message, otherwise 1 for the '{'. + * The value indicates the nesting level, starting with 0 for the main message. + *

+ * There is always a later MSG_LIMIT part. + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_MSG_START, + /** + * End of a message pattern (main or nested). + * The length is 0 for the top-level message and + * the last sub-message of a choice argument, + * otherwise 1 for the '}' or (in a choice argument style) the '|'. + * The value indicates the nesting level, starting with 0 for the main message. + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_MSG_LIMIT, + /** + * Indicates a substring of the pattern string which is to be skipped when formatting. + * For example, an apostrophe that begins or ends quoted text + * would be indicated with such a part. + * The value is undefined and currently always 0. + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_SKIP_SYNTAX, + /** + * Indicates that a syntax character needs to be inserted for auto-quoting. + * The length is 0. + * The value is the character code of the insertion character. (U+0027=APOSTROPHE) + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_INSERT_CHAR, + /** + * Indicates a syntactic (non-escaped) # symbol in a plural variant. + * When formatting, replace this part's substring with the + * (value-offset) for the plural argument value. + * The value is undefined and currently always 0. + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_REPLACE_NUMBER, + /** + * Start of an argument. + * The length is 1 for the '{'. + * The value is the ordinal value of the ArgType. Use getArgType(). + *

+ * This part is followed by either an ARG_NUMBER or ARG_NAME, + * followed by optional argument sub-parts (see UMessagePatternArgType constants) + * and finally an ARG_LIMIT part. + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_ARG_START, + /** + * End of an argument. + * The length is 1 for the '}'. + * The value is the ordinal value of the ArgType. Use getArgType(). + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_ARG_LIMIT, + /** + * The argument number, provided by the value. + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_ARG_NUMBER, + /** + * The argument name. + * The value is undefined and currently always 0. + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_ARG_NAME, + /** + * The argument type. + * The value is undefined and currently always 0. + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_ARG_TYPE, + /** + * The argument style text. + * The value is undefined and currently always 0. + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_ARG_STYLE, + /** + * A selector substring in a "complex" argument style. + * The value is undefined and currently always 0. + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_ARG_SELECTOR, + /** + * An integer value, for example the offset or an explicit selector value + * in a PluralFormat style. + * The part value is the integer value. + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_ARG_INT, + /** + * A numeric value, for example the offset or an explicit selector value + * in a PluralFormat style. + * The part value is an index into an internal array of numeric values; + * use getNumericValue(). + * @stable ICU 4.8 + */ + UMSGPAT_PART_TYPE_ARG_DOUBLE +}; +/** + * @stable ICU 4.8 + */ +typedef enum UMessagePatternPartType UMessagePatternPartType; + +/** + * Argument type constants. + * Returned by Part.getArgType() for ARG_START and ARG_LIMIT parts. + * + * Messages nested inside an argument are each delimited by MSG_START and MSG_LIMIT, + * with a nesting level one greater than the surrounding message. + * @stable ICU 4.8 + */ +enum UMessagePatternArgType { + /** + * The argument has no specified type. + * @stable ICU 4.8 + */ + UMSGPAT_ARG_TYPE_NONE, + /** + * The argument has a "simple" type which is provided by the ARG_TYPE part. + * An ARG_STYLE part might follow that. + * @stable ICU 4.8 + */ + UMSGPAT_ARG_TYPE_SIMPLE, + /** + * The argument is a ChoiceFormat with one or more + * ((ARG_INT | ARG_DOUBLE), ARG_SELECTOR, message) tuples. + * @stable ICU 4.8 + */ + UMSGPAT_ARG_TYPE_CHOICE, + /** + * The argument is a cardinal-number PluralFormat with an optional ARG_INT or ARG_DOUBLE offset + * (e.g., offset:1) + * and one or more (ARG_SELECTOR [explicit-value] message) tuples. + * If the selector has an explicit value (e.g., =2), then + * that value is provided by the ARG_INT or ARG_DOUBLE part preceding the message. + * Otherwise the message immediately follows the ARG_SELECTOR. + * @stable ICU 4.8 + */ + UMSGPAT_ARG_TYPE_PLURAL, + /** + * The argument is a SelectFormat with one or more (ARG_SELECTOR, message) pairs. + * @stable ICU 4.8 + */ + UMSGPAT_ARG_TYPE_SELECT, + /** + * The argument is an ordinal-number PluralFormat + * with the same style parts sequence and semantics as UMSGPAT_ARG_TYPE_PLURAL. + * @stable ICU 50 + */ + UMSGPAT_ARG_TYPE_SELECTORDINAL +}; +/** + * @stable ICU 4.8 + */ +typedef enum UMessagePatternArgType UMessagePatternArgType; + +/** + * \def UMSGPAT_ARG_TYPE_HAS_PLURAL_STYLE + * Returns true if the argument type has a plural style part sequence and semantics, + * for example UMSGPAT_ARG_TYPE_PLURAL and UMSGPAT_ARG_TYPE_SELECTORDINAL. + * @stable ICU 50 + */ +#define UMSGPAT_ARG_TYPE_HAS_PLURAL_STYLE(argType) \ + ((argType)==UMSGPAT_ARG_TYPE_PLURAL || (argType)==UMSGPAT_ARG_TYPE_SELECTORDINAL) + +enum { + /** + * Return value from MessagePattern.validateArgumentName() for when + * the string is a valid "pattern identifier" but not a number. + * @stable ICU 4.8 + */ + UMSGPAT_ARG_NAME_NOT_NUMBER=-1, + + /** + * Return value from MessagePattern.validateArgumentName() for when + * the string is invalid. + * It might not be a valid "pattern identifier", + * or it have only ASCII digits but there is a leading zero or the number is too large. + * @stable ICU 4.8 + */ + UMSGPAT_ARG_NAME_NOT_VALID=-2 +}; + +/** + * Special value that is returned by getNumericValue(Part) when no + * numeric value is defined for a part. + * @see MessagePattern.getNumericValue() + * @stable ICU 4.8 + */ +#define UMSGPAT_NO_NUMERIC_VALUE ((double)(-123456789)) + +U_NAMESPACE_BEGIN + +class MessagePatternDoubleList; +class MessagePatternPartsList; + +/** + * Parses and represents ICU MessageFormat patterns. + * Also handles patterns for ChoiceFormat, PluralFormat and SelectFormat. + * Used in the implementations of those classes as well as in tools + * for message validation, translation and format conversion. + *

+ * The parser handles all syntax relevant for identifying message arguments. + * This includes "complex" arguments whose style strings contain + * nested MessageFormat pattern substrings. + * For "simple" arguments (with no nested MessageFormat pattern substrings), + * the argument style is not parsed any further. + *

+ * The parser handles named and numbered message arguments and allows both in one message. + *

+ * Once a pattern has been parsed successfully, iterate through the parsed data + * with countParts(), getPart() and related methods. + *

+ * The data logically represents a parse tree, but is stored and accessed + * as a list of "parts" for fast and simple parsing and to minimize object allocations. + * Arguments and nested messages are best handled via recursion. + * For every _START "part", MessagePattern.getLimitPartIndex() efficiently returns + * the index of the corresponding _LIMIT "part". + *

+ * List of "parts": + * 

+ * message = MSG_START (SKIP_SYNTAX | INSERT_CHAR | REPLACE_NUMBER | argument)* MSG_LIMIT
+ * argument = noneArg | simpleArg | complexArg
+ * complexArg = choiceArg | pluralArg | selectArg
+ *
+ * noneArg = ARG_START.NONE (ARG_NAME | ARG_NUMBER) ARG_LIMIT.NONE
+ * simpleArg = ARG_START.SIMPLE (ARG_NAME | ARG_NUMBER) ARG_TYPE [ARG_STYLE] ARG_LIMIT.SIMPLE
+ * choiceArg = ARG_START.CHOICE (ARG_NAME | ARG_NUMBER) choiceStyle ARG_LIMIT.CHOICE
+ * pluralArg = ARG_START.PLURAL (ARG_NAME | ARG_NUMBER) pluralStyle ARG_LIMIT.PLURAL
+ * selectArg = ARG_START.SELECT (ARG_NAME | ARG_NUMBER) selectStyle ARG_LIMIT.SELECT
+ *
+ * choiceStyle = ((ARG_INT | ARG_DOUBLE) ARG_SELECTOR message)+
+ * pluralStyle = [ARG_INT | ARG_DOUBLE] (ARG_SELECTOR [ARG_INT | ARG_DOUBLE] message)+
+ * selectStyle = (ARG_SELECTOR message)+
+ *
+ *
    + *
  • Literal output text is not represented directly by "parts" but accessed + * between parts of a message, from one part's getLimit() to the next part's getIndex(). + *
  • ARG_START.CHOICE stands for an ARG_START Part with ArgType CHOICE. + *
  • In the choiceStyle, the ARG_SELECTOR has the '<', the '#' or + * the less-than-or-equal-to sign (U+2264). + *
  • In the pluralStyle, the first, optional numeric Part has the "offset:" value. + * The optional numeric Part between each (ARG_SELECTOR, message) pair + * is the value of an explicit-number selector like "=2", + * otherwise the selector is a non-numeric identifier. + *
  • The REPLACE_NUMBER Part can occur only in an immediate sub-message of the pluralStyle. + *
+ *

+ * This class is not intended for public subclassing. + * + * @stable ICU 4.8 + */ +class U_COMMON_API MessagePattern : public UObject { +public: + /** + * Constructs an empty MessagePattern with default UMessagePatternApostropheMode. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @stable ICU 4.8 + */ + MessagePattern(UErrorCode &errorCode); + + /** + * Constructs an empty MessagePattern. + * @param mode Explicit UMessagePatternApostropheMode. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @stable ICU 4.8 + */ + MessagePattern(UMessagePatternApostropheMode mode, UErrorCode &errorCode); + + /** + * Constructs a MessagePattern with default UMessagePatternApostropheMode and + * parses the MessageFormat pattern string. + * @param pattern a MessageFormat pattern string + * @param parseError Struct to receive information on the position + * of an error within the pattern. + * Can be nullptr. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * TODO: turn @throws into UErrorCode specifics? + * @throws IllegalArgumentException for syntax errors in the pattern string + * @throws IndexOutOfBoundsException if certain limits are exceeded + * (e.g., argument number too high, argument name too long, etc.) + * @throws NumberFormatException if a number could not be parsed + * @stable ICU 4.8 + */ + MessagePattern(const UnicodeString &pattern, UParseError *parseError, UErrorCode &errorCode); + + /** + * Copy constructor. + * @param other Object to copy. + * @stable ICU 4.8 + */ + MessagePattern(const MessagePattern &other); + + /** + * Assignment operator. + * @param other Object to copy. + * @return *this=other + * @stable ICU 4.8 + */ + MessagePattern &operator=(const MessagePattern &other); + + /** + * Destructor. + * @stable ICU 4.8 + */ + virtual ~MessagePattern(); + + /** + * Parses a MessageFormat pattern string. + * @param pattern a MessageFormat pattern string + * @param parseError Struct to receive information on the position + * of an error within the pattern. + * Can be nullptr. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return *this + * @throws IllegalArgumentException for syntax errors in the pattern string + * @throws IndexOutOfBoundsException if certain limits are exceeded + * (e.g., argument number too high, argument name too long, etc.) + * @throws NumberFormatException if a number could not be parsed + * @stable ICU 4.8 + */ + MessagePattern &parse(const UnicodeString &pattern, + UParseError *parseError, UErrorCode &errorCode); + + /** + * Parses a ChoiceFormat pattern string. + * @param pattern a ChoiceFormat pattern string + * @param parseError Struct to receive information on the position + * of an error within the pattern. + * Can be nullptr. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return *this + * @throws IllegalArgumentException for syntax errors in the pattern string + * @throws IndexOutOfBoundsException if certain limits are exceeded + * (e.g., argument number too high, argument name too long, etc.) + * @throws NumberFormatException if a number could not be parsed + * @stable ICU 4.8 + */ + MessagePattern &parseChoiceStyle(const UnicodeString &pattern, + UParseError *parseError, UErrorCode &errorCode); + + /** + * Parses a PluralFormat pattern string. + * @param pattern a PluralFormat pattern string + * @param parseError Struct to receive information on the position + * of an error within the pattern. + * Can be nullptr. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return *this + * @throws IllegalArgumentException for syntax errors in the pattern string + * @throws IndexOutOfBoundsException if certain limits are exceeded + * (e.g., argument number too high, argument name too long, etc.) + * @throws NumberFormatException if a number could not be parsed + * @stable ICU 4.8 + */ + MessagePattern &parsePluralStyle(const UnicodeString &pattern, + UParseError *parseError, UErrorCode &errorCode); + + /** + * Parses a SelectFormat pattern string. + * @param pattern a SelectFormat pattern string + * @param parseError Struct to receive information on the position + * of an error within the pattern. + * Can be nullptr. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return *this + * @throws IllegalArgumentException for syntax errors in the pattern string + * @throws IndexOutOfBoundsException if certain limits are exceeded + * (e.g., argument number too high, argument name too long, etc.) + * @throws NumberFormatException if a number could not be parsed + * @stable ICU 4.8 + */ + MessagePattern &parseSelectStyle(const UnicodeString &pattern, + UParseError *parseError, UErrorCode &errorCode); + + /** + * Clears this MessagePattern. + * countParts() will return 0. + * @stable ICU 4.8 + */ + void clear(); + + /** + * Clears this MessagePattern and sets the UMessagePatternApostropheMode. + * countParts() will return 0. + * @param mode The new UMessagePatternApostropheMode. + * @stable ICU 4.8 + */ + void clearPatternAndSetApostropheMode(UMessagePatternApostropheMode mode) { + clear(); + aposMode=mode; + } + + /** + * @param other another object to compare with. + * @return true if this object is equivalent to the other one. + * @stable ICU 4.8 + */ + bool operator==(const MessagePattern &other) const; + + /** + * @param other another object to compare with. + * @return false if this object is equivalent to the other one. + * @stable ICU 4.8 + */ + inline bool operator!=(const MessagePattern &other) const { + return !operator==(other); + } + + /** + * @return A hash code for this object. + * @stable ICU 4.8 + */ + int32_t hashCode() const; + + /** + * @return this instance's UMessagePatternApostropheMode. + * @stable ICU 4.8 + */ + UMessagePatternApostropheMode getApostropheMode() const { + return aposMode; + } + + // Java has package-private jdkAposMode() here. + // In C++, this is declared in the MessageImpl class. + + /** + * @return the parsed pattern string (null if none was parsed). + * @stable ICU 4.8 + */ + const UnicodeString &getPatternString() const { + return msg; + } + + /** + * Does the parsed pattern have named arguments like {first_name}? + * @return true if the parsed pattern has at least one named argument. + * @stable ICU 4.8 + */ + UBool hasNamedArguments() const { + return hasArgNames; + } + + /** + * Does the parsed pattern have numbered arguments like {2}? + * @return true if the parsed pattern has at least one numbered argument. + * @stable ICU 4.8 + */ + UBool hasNumberedArguments() const { + return hasArgNumbers; + } + + /** + * Validates and parses an argument name or argument number string. + * An argument name must be a "pattern identifier", that is, it must contain + * no Unicode Pattern_Syntax or Pattern_White_Space characters. + * If it only contains ASCII digits, then it must be a small integer with no leading zero. + * @param name Input string. + * @return >=0 if the name is a valid number, + * ARG_NAME_NOT_NUMBER (-1) if it is a "pattern identifier" but not all ASCII digits, + * ARG_NAME_NOT_VALID (-2) if it is neither. + * @stable ICU 4.8 + */ + static int32_t validateArgumentName(const UnicodeString &name); + + /** + * Returns a version of the parsed pattern string where each ASCII apostrophe + * is doubled (escaped) if it is not already, and if it is not interpreted as quoting syntax. + *

+ * For example, this turns "I don't '{know}' {gender,select,female{h''er}other{h'im}}." + * into "I don''t '{know}' {gender,select,female{h''er}other{h''im}}." + * @return the deep-auto-quoted version of the parsed pattern string. + * @see MessageFormat.autoQuoteApostrophe() + * @stable ICU 4.8 + */ + UnicodeString autoQuoteApostropheDeep() const; + + class Part; + + /** + * Returns the number of "parts" created by parsing the pattern string. + * Returns 0 if no pattern has been parsed or clear() was called. + * @return the number of pattern parts. + * @stable ICU 4.8 + */ + int32_t countParts() const { + return partsLength; + } + + /** + * Gets the i-th pattern "part". + * @param i The index of the Part data. (0..countParts()-1) + * @return the i-th pattern "part". + * @stable ICU 4.8 + */ + const Part &getPart(int32_t i) const { + return parts[i]; + } + + /** + * Returns the UMessagePatternPartType of the i-th pattern "part". + * Convenience method for getPart(i).getType(). + * @param i The index of the Part data. (0..countParts()-1) + * @return The UMessagePatternPartType of the i-th Part. + * @stable ICU 4.8 + */ + UMessagePatternPartType getPartType(int32_t i) const { + return getPart(i).type; + } + + /** + * Returns the pattern index of the specified pattern "part". + * Convenience method for getPart(partIndex).getIndex(). + * @param partIndex The index of the Part data. (0..countParts()-1) + * @return The pattern index of this Part. + * @stable ICU 4.8 + */ + int32_t getPatternIndex(int32_t partIndex) const { + return getPart(partIndex).index; + } + + /** + * Returns the substring of the pattern string indicated by the Part. + * Convenience method for getPatternString().substring(part.getIndex(), part.getLimit()). + * @param part a part of this MessagePattern. + * @return the substring associated with part. + * @stable ICU 4.8 + */ + UnicodeString getSubstring(const Part &part) const { + return msg.tempSubString(part.index, part.length); + } + + /** + * Compares the part's substring with the input string s. + * @param part a part of this MessagePattern. + * @param s a string. + * @return true if getSubstring(part).equals(s). + * @stable ICU 4.8 + */ + UBool partSubstringMatches(const Part &part, const UnicodeString &s) const { + return 0==msg.compare(part.index, part.length, s); + } + + /** + * Returns the numeric value associated with an ARG_INT or ARG_DOUBLE. + * @param part a part of this MessagePattern. + * @return the part's numeric value, or UMSGPAT_NO_NUMERIC_VALUE if this is not a numeric part. + * @stable ICU 4.8 + */ + double getNumericValue(const Part &part) const; + + /** + * Returns the "offset:" value of a PluralFormat argument, or 0 if none is specified. + * @param pluralStart the index of the first PluralFormat argument style part. (0..countParts()-1) + * @return the "offset:" value. + * @stable ICU 4.8 + */ + double getPluralOffset(int32_t pluralStart) const; + + /** + * Returns the index of the ARG|MSG_LIMIT part corresponding to the ARG|MSG_START at start. + * @param start The index of some Part data (0..countParts()-1); + * this Part should be of Type ARG_START or MSG_START. + * @return The first i>start where getPart(i).getType()==ARG|MSG_LIMIT at the same nesting level, + * or start itself if getPartType(msgStart)!=ARG|MSG_START. + * @stable ICU 4.8 + */ + int32_t getLimitPartIndex(int32_t start) const { + int32_t limit=getPart(start).limitPartIndex; + if(limit parts=new ArrayList(); + MessagePatternPartsList *partsList; + Part *parts; + int32_t partsLength; + // ArrayList numericValues; + MessagePatternDoubleList *numericValuesList; + double *numericValues; + int32_t numericValuesLength; + UBool hasArgNames; + UBool hasArgNumbers; + UBool needsAutoQuoting; +}; + +U_NAMESPACE_END + +#endif // !UCONFIG_NO_FORMATTING + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __MESSAGEPATTERN_H__ diff --git a/intl/icu/source/common/unicode/normalizer2.h b/intl/icu/source/common/unicode/normalizer2.h new file mode 100644 index 0000000000..972894ec42 --- /dev/null +++ b/intl/icu/source/common/unicode/normalizer2.h @@ -0,0 +1,771 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2009-2013, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: normalizer2.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2009nov22 +* created by: Markus W. Scherer +*/ + +#ifndef __NORMALIZER2_H__ +#define __NORMALIZER2_H__ + +/** + * \file + * \brief C++ API: New API for Unicode Normalization. + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#if !UCONFIG_NO_NORMALIZATION + +#include "unicode/stringpiece.h" +#include "unicode/uniset.h" +#include "unicode/unistr.h" +#include "unicode/unorm2.h" + +U_NAMESPACE_BEGIN + +class ByteSink; + +/** + * Unicode normalization functionality for standard Unicode normalization or + * for using custom mapping tables. + * All instances of this class are unmodifiable/immutable. + * Instances returned by getInstance() are singletons that must not be deleted by the caller. + * The Normalizer2 class is not intended for public subclassing. + * + * The primary functions are to produce a normalized string and to detect whether + * a string is already normalized. + * The most commonly used normalization forms are those defined in + * http://www.unicode.org/unicode/reports/tr15/ + * However, this API supports additional normalization forms for specialized purposes. + * For example, NFKC_Casefold is provided via getInstance("nfkc_cf", COMPOSE) + * and can be used in implementations of UTS #46. + * + * Not only are the standard compose and decompose modes supplied, + * but additional modes are provided as documented in the Mode enum. + * + * Some of the functions in this class identify normalization boundaries. + * At a normalization boundary, the portions of the string + * before it and starting from it do not interact and can be handled independently. + * + * The spanQuickCheckYes() stops at a normalization boundary. + * When the goal is a normalized string, then the text before the boundary + * can be copied, and the remainder can be processed with normalizeSecondAndAppend(). + * + * The hasBoundaryBefore(), hasBoundaryAfter() and isInert() functions test whether + * a character is guaranteed to be at a normalization boundary, + * regardless of context. + * This is used for moving from one normalization boundary to the next + * or preceding boundary, and for performing iterative normalization. + * + * Iterative normalization is useful when only a small portion of a + * longer string needs to be processed. + * For example, in ICU, iterative normalization is used by the NormalizationTransliterator + * (to avoid replacing already-normalized text) and ucol_nextSortKeyPart() + * (to process only the substring for which sort key bytes are computed). + * + * The set of normalization boundaries returned by these functions may not be + * complete: There may be more boundaries that could be returned. + * Different functions may return different boundaries. + * @stable ICU 4.4 + */ +class U_COMMON_API Normalizer2 : public UObject { +public: + /** + * Destructor. + * @stable ICU 4.4 + */ + ~Normalizer2(); + + /** + * Returns a Normalizer2 instance for Unicode NFC normalization. + * Same as getInstance(nullptr, "nfc", UNORM2_COMPOSE, errorCode). + * Returns an unmodifiable singleton instance. Do not delete it. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the requested Normalizer2, if successful + * @stable ICU 49 + */ + static const Normalizer2 * + getNFCInstance(UErrorCode &errorCode); + + /** + * Returns a Normalizer2 instance for Unicode NFD normalization. + * Same as getInstance(nullptr, "nfc", UNORM2_DECOMPOSE, errorCode). + * Returns an unmodifiable singleton instance. Do not delete it. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the requested Normalizer2, if successful + * @stable ICU 49 + */ + static const Normalizer2 * + getNFDInstance(UErrorCode &errorCode); + + /** + * Returns a Normalizer2 instance for Unicode NFKC normalization. + * Same as getInstance(nullptr, "nfkc", UNORM2_COMPOSE, errorCode). + * Returns an unmodifiable singleton instance. Do not delete it. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the requested Normalizer2, if successful + * @stable ICU 49 + */ + static const Normalizer2 * + getNFKCInstance(UErrorCode &errorCode); + + /** + * Returns a Normalizer2 instance for Unicode NFKD normalization. + * Same as getInstance(nullptr, "nfkc", UNORM2_DECOMPOSE, errorCode). + * Returns an unmodifiable singleton instance. Do not delete it. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the requested Normalizer2, if successful + * @stable ICU 49 + */ + static const Normalizer2 * + getNFKDInstance(UErrorCode &errorCode); + + /** + * Returns a Normalizer2 instance for Unicode NFKC_Casefold normalization. + * Same as getInstance(nullptr, "nfkc_cf", UNORM2_COMPOSE, errorCode). + * Returns an unmodifiable singleton instance. Do not delete it. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the requested Normalizer2, if successful + * @stable ICU 49 + */ + static const Normalizer2 * + getNFKCCasefoldInstance(UErrorCode &errorCode); + + /** + * Returns a Normalizer2 instance which uses the specified data file + * (packageName/name similar to ucnv_openPackage() and ures_open()/ResourceBundle) + * and which composes or decomposes text according to the specified mode. + * Returns an unmodifiable singleton instance. Do not delete it. + * + * Use packageName=nullptr for data files that are part of ICU's own data. + * Use name="nfc" and UNORM2_COMPOSE/UNORM2_DECOMPOSE for Unicode standard NFC/NFD. + * Use name="nfkc" and UNORM2_COMPOSE/UNORM2_DECOMPOSE for Unicode standard NFKC/NFKD. + * Use name="nfkc_cf" and UNORM2_COMPOSE for Unicode standard NFKC_CF=NFKC_Casefold. + * + * @param packageName nullptr for ICU built-in data, otherwise application data package name + * @param name "nfc" or "nfkc" or "nfkc_cf" or name of custom data file + * @param mode normalization mode (compose or decompose etc.) + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the requested Normalizer2, if successful + * @stable ICU 4.4 + */ + static const Normalizer2 * + getInstance(const char *packageName, + const char *name, + UNormalization2Mode mode, + UErrorCode &errorCode); + + /** + * Returns the normalized form of the source string. + * @param src source string + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return normalized src + * @stable ICU 4.4 + */ + UnicodeString + normalize(const UnicodeString &src, UErrorCode &errorCode) const { + UnicodeString result; + normalize(src, result, errorCode); + return result; + } + /** + * Writes the normalized form of the source string to the destination string + * (replacing its contents) and returns the destination string. + * The source and destination strings must be different objects. + * @param src source string + * @param dest destination string; its contents is replaced with normalized src + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return dest + * @stable ICU 4.4 + */ + virtual UnicodeString & + normalize(const UnicodeString &src, + UnicodeString &dest, + UErrorCode &errorCode) const = 0; + + /** + * Normalizes a UTF-8 string and optionally records how source substrings + * relate to changed and unchanged result substrings. + * + * Implemented completely for all built-in modes except for FCD. + * The base class implementation converts to & from UTF-16 and does not support edits. + * + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT and U_EDITS_NO_RESET. + * @param src Source UTF-8 string. + * @param sink A ByteSink to which the normalized UTF-8 result string is written. + * sink.Flush() is called at the end. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @stable ICU 60 + */ + virtual void + normalizeUTF8(uint32_t options, StringPiece src, ByteSink &sink, + Edits *edits, UErrorCode &errorCode) const; + + /** + * Appends the normalized form of the second string to the first string + * (merging them at the boundary) and returns the first string. + * The result is normalized if the first string was normalized. + * The first and second strings must be different objects. + * @param first string, should be normalized + * @param second string, will be normalized + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return first + * @stable ICU 4.4 + */ + virtual UnicodeString & + normalizeSecondAndAppend(UnicodeString &first, + const UnicodeString &second, + UErrorCode &errorCode) const = 0; + /** + * Appends the second string to the first string + * (merging them at the boundary) and returns the first string. + * The result is normalized if both the strings were normalized. + * The first and second strings must be different objects. + * @param first string, should be normalized + * @param second string, should be normalized + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return first + * @stable ICU 4.4 + */ + virtual UnicodeString & + append(UnicodeString &first, + const UnicodeString &second, + UErrorCode &errorCode) const = 0; + + /** + * Gets the decomposition mapping of c. + * Roughly equivalent to normalizing the String form of c + * on a UNORM2_DECOMPOSE Normalizer2 instance, but much faster, and except that this function + * returns false and does not write a string + * if c does not have a decomposition mapping in this instance's data. + * This function is independent of the mode of the Normalizer2. + * @param c code point + * @param decomposition String object which will be set to c's + * decomposition mapping, if there is one. + * @return true if c has a decomposition, otherwise false + * @stable ICU 4.6 + */ + virtual UBool + getDecomposition(UChar32 c, UnicodeString &decomposition) const = 0; + + /** + * Gets the raw decomposition mapping of c. + * + * This is similar to the getDecomposition() method but returns the + * raw decomposition mapping as specified in UnicodeData.txt or + * (for custom data) in the mapping files processed by the gennorm2 tool. + * By contrast, getDecomposition() returns the processed, + * recursively-decomposed version of this mapping. + * + * When used on a standard NFKC Normalizer2 instance, + * getRawDecomposition() returns the Unicode Decomposition_Mapping (dm) property. + * + * When used on a standard NFC Normalizer2 instance, + * it returns the Decomposition_Mapping only if the Decomposition_Type (dt) is Canonical (Can); + * in this case, the result contains either one or two code points (=1..4 char16_ts). + * + * This function is independent of the mode of the Normalizer2. + * The default implementation returns false. + * @param c code point + * @param decomposition String object which will be set to c's + * raw decomposition mapping, if there is one. + * @return true if c has a decomposition, otherwise false + * @stable ICU 49 + */ + virtual UBool + getRawDecomposition(UChar32 c, UnicodeString &decomposition) const; + + /** + * Performs pairwise composition of a & b and returns the composite if there is one. + * + * Returns a composite code point c only if c has a two-way mapping to a+b. + * In standard Unicode normalization, this means that + * c has a canonical decomposition to a+b + * and c does not have the Full_Composition_Exclusion property. + * + * This function is independent of the mode of the Normalizer2. + * The default implementation returns a negative value. + * @param a A (normalization starter) code point. + * @param b Another code point. + * @return The non-negative composite code point if there is one; otherwise a negative value. + * @stable ICU 49 + */ + virtual UChar32 + composePair(UChar32 a, UChar32 b) const; + + /** + * Gets the combining class of c. + * The default implementation returns 0 + * but all standard implementations return the Unicode Canonical_Combining_Class value. + * @param c code point + * @return c's combining class + * @stable ICU 49 + */ + virtual uint8_t + getCombiningClass(UChar32 c) const; + + /** + * Tests if the string is normalized. + * Internally, in cases where the quickCheck() method would return "maybe" + * (which is only possible for the two COMPOSE modes) this method + * resolves to "yes" or "no" to provide a definitive result, + * at the cost of doing more work in those cases. + * @param s input string + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return true if s is normalized + * @stable ICU 4.4 + */ + virtual UBool + isNormalized(const UnicodeString &s, UErrorCode &errorCode) const = 0; + /** + * Tests if the UTF-8 string is normalized. + * Internally, in cases where the quickCheck() method would return "maybe" + * (which is only possible for the two COMPOSE modes) this method + * resolves to "yes" or "no" to provide a definitive result, + * at the cost of doing more work in those cases. + * + * This works for all normalization modes. + * It is optimized for UTF-8 for all built-in modes except for FCD. + * The base class implementation converts to UTF-16 and calls isNormalized(). + * + * @param s UTF-8 input string + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return true if s is normalized + * @stable ICU 60 + */ + virtual UBool + isNormalizedUTF8(StringPiece s, UErrorCode &errorCode) const; + + + /** + * Tests if the string is normalized. + * For the two COMPOSE modes, the result could be "maybe" in cases that + * would take a little more work to resolve definitively. + * Use spanQuickCheckYes() and normalizeSecondAndAppend() for a faster + * combination of quick check + normalization, to avoid + * re-checking the "yes" prefix. + * @param s input string + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return UNormalizationCheckResult + * @stable ICU 4.4 + */ + virtual UNormalizationCheckResult + quickCheck(const UnicodeString &s, UErrorCode &errorCode) const = 0; + + /** + * Returns the end of the normalized substring of the input string. + * In other words, with end=spanQuickCheckYes(s, ec); + * the substring UnicodeString(s, 0, end) + * will pass the quick check with a "yes" result. + * + * The returned end index is usually one or more characters before the + * "no" or "maybe" character: The end index is at a normalization boundary. + * (See the class documentation for more about normalization boundaries.) + * + * When the goal is a normalized string and most input strings are expected + * to be normalized already, then call this method, + * and if it returns a prefix shorter than the input string, + * copy that prefix and use normalizeSecondAndAppend() for the remainder. + * @param s input string + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return "yes" span end index + * @stable ICU 4.4 + */ + virtual int32_t + spanQuickCheckYes(const UnicodeString &s, UErrorCode &errorCode) const = 0; + + /** + * Tests if the character always has a normalization boundary before it, + * regardless of context. + * If true, then the character does not normalization-interact with + * preceding characters. + * In other words, a string containing this character can be normalized + * by processing portions before this character and starting from this + * character independently. + * This is used for iterative normalization. See the class documentation for details. + * @param c character to test + * @return true if c has a normalization boundary before it + * @stable ICU 4.4 + */ + virtual UBool hasBoundaryBefore(UChar32 c) const = 0; + + /** + * Tests if the character always has a normalization boundary after it, + * regardless of context. + * If true, then the character does not normalization-interact with + * following characters. + * In other words, a string containing this character can be normalized + * by processing portions up to this character and after this + * character independently. + * This is used for iterative normalization. See the class documentation for details. + * Note that this operation may be significantly slower than hasBoundaryBefore(). + * @param c character to test + * @return true if c has a normalization boundary after it + * @stable ICU 4.4 + */ + virtual UBool hasBoundaryAfter(UChar32 c) const = 0; + + /** + * Tests if the character is normalization-inert. + * If true, then the character does not change, nor normalization-interact with + * preceding or following characters. + * In other words, a string containing this character can be normalized + * by processing portions before this character and after this + * character independently. + * This is used for iterative normalization. See the class documentation for details. + * Note that this operation may be significantly slower than hasBoundaryBefore(). + * @param c character to test + * @return true if c is normalization-inert + * @stable ICU 4.4 + */ + virtual UBool isInert(UChar32 c) const = 0; +}; + +/** + * Normalization filtered by a UnicodeSet. + * Normalizes portions of the text contained in the filter set and leaves + * portions not contained in the filter set unchanged. + * Filtering is done via UnicodeSet::span(..., USET_SPAN_SIMPLE). + * Not-in-the-filter text is treated as "is normalized" and "quick check yes". + * This class implements all of (and only) the Normalizer2 API. + * An instance of this class is unmodifiable/immutable but is constructed and + * must be destructed by the owner. + * @stable ICU 4.4 + */ +class U_COMMON_API FilteredNormalizer2 : public Normalizer2 { +public: + /** + * Constructs a filtered normalizer wrapping any Normalizer2 instance + * and a filter set. + * Both are aliased and must not be modified or deleted while this object + * is used. + * The filter set should be frozen; otherwise the performance will suffer greatly. + * @param n2 wrapped Normalizer2 instance + * @param filterSet UnicodeSet which determines the characters to be normalized + * @stable ICU 4.4 + */ + FilteredNormalizer2(const Normalizer2 &n2, const UnicodeSet &filterSet) : + norm2(n2), set(filterSet) {} + + /** + * Destructor. + * @stable ICU 4.4 + */ + ~FilteredNormalizer2(); + + /** + * Writes the normalized form of the source string to the destination string + * (replacing its contents) and returns the destination string. + * The source and destination strings must be different objects. + * @param src source string + * @param dest destination string; its contents is replaced with normalized src + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return dest + * @stable ICU 4.4 + */ + virtual UnicodeString & + normalize(const UnicodeString &src, + UnicodeString &dest, + UErrorCode &errorCode) const override; + + /** + * Normalizes a UTF-8 string and optionally records how source substrings + * relate to changed and unchanged result substrings. + * + * Implemented completely for most built-in modes except for FCD. + * The base class implementation converts to & from UTF-16 and does not support edits. + * + * @param options Options bit set, usually 0. See U_OMIT_UNCHANGED_TEXT and U_EDITS_NO_RESET. + * @param src Source UTF-8 string. + * @param sink A ByteSink to which the normalized UTF-8 result string is written. + * sink.Flush() is called at the end. + * @param edits Records edits for index mapping, working with styled text, + * and getting only changes (if any). + * The Edits contents is undefined if any error occurs. + * This function calls edits->reset() first unless + * options includes U_EDITS_NO_RESET. edits can be nullptr. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @stable ICU 60 + */ + virtual void + normalizeUTF8(uint32_t options, StringPiece src, ByteSink &sink, + Edits *edits, UErrorCode &errorCode) const override; + + /** + * Appends the normalized form of the second string to the first string + * (merging them at the boundary) and returns the first string. + * The result is normalized if the first string was normalized. + * The first and second strings must be different objects. + * @param first string, should be normalized + * @param second string, will be normalized + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return first + * @stable ICU 4.4 + */ + virtual UnicodeString & + normalizeSecondAndAppend(UnicodeString &first, + const UnicodeString &second, + UErrorCode &errorCode) const override; + /** + * Appends the second string to the first string + * (merging them at the boundary) and returns the first string. + * The result is normalized if both the strings were normalized. + * The first and second strings must be different objects. + * @param first string, should be normalized + * @param second string, should be normalized + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return first + * @stable ICU 4.4 + */ + virtual UnicodeString & + append(UnicodeString &first, + const UnicodeString &second, + UErrorCode &errorCode) const override; + + /** + * Gets the decomposition mapping of c. + * For details see the base class documentation. + * + * This function is independent of the mode of the Normalizer2. + * @param c code point + * @param decomposition String object which will be set to c's + * decomposition mapping, if there is one. + * @return true if c has a decomposition, otherwise false + * @stable ICU 4.6 + */ + virtual UBool + getDecomposition(UChar32 c, UnicodeString &decomposition) const override; + + /** + * Gets the raw decomposition mapping of c. + * For details see the base class documentation. + * + * This function is independent of the mode of the Normalizer2. + * @param c code point + * @param decomposition String object which will be set to c's + * raw decomposition mapping, if there is one. + * @return true if c has a decomposition, otherwise false + * @stable ICU 49 + */ + virtual UBool + getRawDecomposition(UChar32 c, UnicodeString &decomposition) const override; + + /** + * Performs pairwise composition of a & b and returns the composite if there is one. + * For details see the base class documentation. + * + * This function is independent of the mode of the Normalizer2. + * @param a A (normalization starter) code point. + * @param b Another code point. + * @return The non-negative composite code point if there is one; otherwise a negative value. + * @stable ICU 49 + */ + virtual UChar32 + composePair(UChar32 a, UChar32 b) const override; + + /** + * Gets the combining class of c. + * The default implementation returns 0 + * but all standard implementations return the Unicode Canonical_Combining_Class value. + * @param c code point + * @return c's combining class + * @stable ICU 49 + */ + virtual uint8_t + getCombiningClass(UChar32 c) const override; + + /** + * Tests if the string is normalized. + * For details see the Normalizer2 base class documentation. + * @param s input string + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return true if s is normalized + * @stable ICU 4.4 + */ + virtual UBool + isNormalized(const UnicodeString &s, UErrorCode &errorCode) const override; + /** + * Tests if the UTF-8 string is normalized. + * Internally, in cases where the quickCheck() method would return "maybe" + * (which is only possible for the two COMPOSE modes) this method + * resolves to "yes" or "no" to provide a definitive result, + * at the cost of doing more work in those cases. + * + * This works for all normalization modes. + * It is optimized for UTF-8 for all built-in modes except for FCD. + * The base class implementation converts to UTF-16 and calls isNormalized(). + * + * @param s UTF-8 input string + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return true if s is normalized + * @stable ICU 60 + */ + virtual UBool + isNormalizedUTF8(StringPiece s, UErrorCode &errorCode) const override; + /** + * Tests if the string is normalized. + * For details see the Normalizer2 base class documentation. + * @param s input string + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return UNormalizationCheckResult + * @stable ICU 4.4 + */ + virtual UNormalizationCheckResult + quickCheck(const UnicodeString &s, UErrorCode &errorCode) const override; + /** + * Returns the end of the normalized substring of the input string. + * For details see the Normalizer2 base class documentation. + * @param s input string + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return "yes" span end index + * @stable ICU 4.4 + */ + virtual int32_t + spanQuickCheckYes(const UnicodeString &s, UErrorCode &errorCode) const override; + + /** + * Tests if the character always has a normalization boundary before it, + * regardless of context. + * For details see the Normalizer2 base class documentation. + * @param c character to test + * @return true if c has a normalization boundary before it + * @stable ICU 4.4 + */ + virtual UBool hasBoundaryBefore(UChar32 c) const override; + + /** + * Tests if the character always has a normalization boundary after it, + * regardless of context. + * For details see the Normalizer2 base class documentation. + * @param c character to test + * @return true if c has a normalization boundary after it + * @stable ICU 4.4 + */ + virtual UBool hasBoundaryAfter(UChar32 c) const override; + + /** + * Tests if the character is normalization-inert. + * For details see the Normalizer2 base class documentation. + * @param c character to test + * @return true if c is normalization-inert + * @stable ICU 4.4 + */ + virtual UBool isInert(UChar32 c) const override; +private: + UnicodeString & + normalize(const UnicodeString &src, + UnicodeString &dest, + USetSpanCondition spanCondition, + UErrorCode &errorCode) const; + + void + normalizeUTF8(uint32_t options, const char *src, int32_t length, + ByteSink &sink, Edits *edits, + USetSpanCondition spanCondition, + UErrorCode &errorCode) const; + + UnicodeString & + normalizeSecondAndAppend(UnicodeString &first, + const UnicodeString &second, + UBool doNormalize, + UErrorCode &errorCode) const; + + const Normalizer2 &norm2; + const UnicodeSet &set; +}; + +U_NAMESPACE_END + +#endif // !UCONFIG_NO_NORMALIZATION + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __NORMALIZER2_H__ diff --git a/intl/icu/source/common/unicode/normlzr.h b/intl/icu/source/common/unicode/normlzr.h new file mode 100644 index 0000000000..03a7aa080d --- /dev/null +++ b/intl/icu/source/common/unicode/normlzr.h @@ -0,0 +1,816 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ******************************************************************** + * COPYRIGHT: + * Copyright (c) 1996-2015, International Business Machines Corporation and + * others. All Rights Reserved. + ******************************************************************** + */ + +#ifndef NORMLZR_H +#define NORMLZR_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +/** + * \file + * \brief C++ API: Unicode Normalization + */ + +#if !UCONFIG_NO_NORMALIZATION + +#include "unicode/chariter.h" +#include "unicode/normalizer2.h" +#include "unicode/unistr.h" +#include "unicode/unorm.h" +#include "unicode/uobject.h" + +U_NAMESPACE_BEGIN +/** + * Old Unicode normalization API. + * + * This API has been replaced by the Normalizer2 class and is only available + * for backward compatibility. This class simply delegates to the Normalizer2 class. + * There is one exception: The new API does not provide a replacement for Normalizer::compare(). + * + * The Normalizer class supports the standard normalization forms described in + * + * Unicode Standard Annex #15: Unicode Normalization Forms. + * + * The Normalizer class consists of two parts: + * - static functions that normalize strings or test if strings are normalized + * - a Normalizer object is an iterator that takes any kind of text and + * provides iteration over its normalized form + * + * The Normalizer class is not suitable for subclassing. + * + * For basic information about normalization forms and details about the C API + * please see the documentation in unorm.h. + * + * The iterator API with the Normalizer constructors and the non-static functions + * use a CharacterIterator as input. It is possible to pass a string which + * is then internally wrapped in a CharacterIterator. + * The input text is not normalized all at once, but incrementally where needed + * (providing efficient random access). + * This allows to pass in a large text but spend only a small amount of time + * normalizing a small part of that text. + * However, if the entire text is normalized, then the iterator will be + * slower than normalizing the entire text at once and iterating over the result. + * A possible use of the Normalizer iterator is also to report an index into the + * original text that is close to where the normalized characters come from. + * + * Important: The iterator API was cleaned up significantly for ICU 2.0. + * The earlier implementation reported the getIndex() inconsistently, + * and previous() could not be used after setIndex(), next(), first(), and current(). + * + * Normalizer allows to start normalizing from anywhere in the input text by + * calling setIndexOnly(), first(), or last(). + * Without calling any of these, the iterator will start at the beginning of the text. + * + * At any time, next() returns the next normalized code point (UChar32), + * with post-increment semantics (like CharacterIterator::next32PostInc()). + * previous() returns the previous normalized code point (UChar32), + * with pre-decrement semantics (like CharacterIterator::previous32()). + * + * current() returns the current code point + * (respectively the one at the newly set index) without moving + * the getIndex(). Note that if the text at the current position + * needs to be normalized, then these functions will do that. + * (This is why current() is not const.) + * It is more efficient to call setIndexOnly() instead, which does not + * normalize. + * + * getIndex() always refers to the position in the input text where the normalized + * code points are returned from. It does not always change with each returned + * code point. + * The code point that is returned from any of the functions + * corresponds to text at or after getIndex(), according to the + * function's iteration semantics (post-increment or pre-decrement). + * + * next() returns a code point from at or after the getIndex() + * from before the next() call. After the next() call, the getIndex() + * might have moved to where the next code point will be returned from + * (from a next() or current() call). + * This is semantically equivalent to array access with array[index++] + * (post-increment semantics). + * + * previous() returns a code point from at or after the getIndex() + * from after the previous() call. + * This is semantically equivalent to array access with array[--index] + * (pre-decrement semantics). + * + * Internally, the Normalizer iterator normalizes a small piece of text + * starting at the getIndex() and ending at a following "safe" index. + * The normalized results is stored in an internal string buffer, and + * the code points are iterated from there. + * With multiple iteration calls, this is repeated until the next piece + * of text needs to be normalized, and the getIndex() needs to be moved. + * + * The following "safe" index, the internal buffer, and the secondary + * iteration index into that buffer are not exposed on the API. + * This also means that it is currently not practical to return to + * a particular, arbitrary position in the text because one would need to + * know, and be able to set, in addition to the getIndex(), at least also the + * current index into the internal buffer. + * It is currently only possible to observe when getIndex() changes + * (with careful consideration of the iteration semantics), + * at which time the internal index will be 0. + * For example, if getIndex() is different after next() than before it, + * then the internal index is 0 and one can return to this getIndex() + * later with setIndexOnly(). + * + * Note: While the setIndex() and getIndex() refer to indices in the + * underlying Unicode input text, the next() and previous() methods + * iterate through characters in the normalized output. + * This means that there is not necessarily a one-to-one correspondence + * between characters returned by next() and previous() and the indices + * passed to and returned from setIndex() and getIndex(). + * It is for this reason that Normalizer does not implement the CharacterIterator interface. + * + * @author Laura Werner, Mark Davis, Markus Scherer + * @stable ICU 2.0 + */ +class U_COMMON_API Normalizer : public UObject { +public: +#ifndef U_HIDE_DEPRECATED_API + /** + * If DONE is returned from an iteration function that returns a code point, + * then there are no more normalization results available. + * @deprecated ICU 56 Use Normalizer2 instead. + */ + enum { + DONE=0xffff + }; + + // Constructors + + /** + * Creates a new Normalizer object for iterating over the + * normalized form of a given string. + *

+ * @param str The string to be normalized. The normalization + * will start at the beginning of the string. + * + * @param mode The normalization mode. + * @deprecated ICU 56 Use Normalizer2 instead. + */ + Normalizer(const UnicodeString& str, UNormalizationMode mode); + + /** + * Creates a new Normalizer object for iterating over the + * normalized form of a given string. + *

+ * @param str The string to be normalized. The normalization + * will start at the beginning of the string. + * + * @param length Length of the string, or -1 if NUL-terminated. + * @param mode The normalization mode. + * @deprecated ICU 56 Use Normalizer2 instead. + */ + Normalizer(ConstChar16Ptr str, int32_t length, UNormalizationMode mode); + + /** + * Creates a new Normalizer object for iterating over the + * normalized form of the given text. + *

+ * @param iter The input text to be normalized. The normalization + * will start at the beginning of the string. + * + * @param mode The normalization mode. + * @deprecated ICU 56 Use Normalizer2 instead. + */ + Normalizer(const CharacterIterator& iter, UNormalizationMode mode); +#endif /* U_HIDE_DEPRECATED_API */ + +#ifndef U_FORCE_HIDE_DEPRECATED_API + /** + * Copy constructor. + * @param copy The object to be copied. + * @deprecated ICU 56 Use Normalizer2 instead. + */ + Normalizer(const Normalizer& copy); + + /** + * Destructor + * @deprecated ICU 56 Use Normalizer2 instead. + */ + virtual ~Normalizer(); +#endif // U_FORCE_HIDE_DEPRECATED_API + + //------------------------------------------------------------------------- + // Static utility methods + //------------------------------------------------------------------------- + +#ifndef U_HIDE_DEPRECATED_API + /** + * Normalizes a UnicodeString according to the specified normalization mode. + * This is a wrapper for unorm_normalize(), using UnicodeString's. + * + * The options parameter specifies which optional + * Normalizer features are to be enabled for this operation. + * + * @param source the input string to be normalized. + * @param mode the normalization mode + * @param options the optional features to be enabled (0 for no options) + * @param result The normalized string (on output). + * @param status The error code. + * @deprecated ICU 56 Use Normalizer2 instead. + */ + static void U_EXPORT2 normalize(const UnicodeString& source, + UNormalizationMode mode, int32_t options, + UnicodeString& result, + UErrorCode &status); + + /** + * Compose a UnicodeString. + * This is equivalent to normalize() with mode UNORM_NFC or UNORM_NFKC. + * This is a wrapper for unorm_normalize(), using UnicodeString's. + * + * The options parameter specifies which optional + * Normalizer features are to be enabled for this operation. + * + * @param source the string to be composed. + * @param compat Perform compatibility decomposition before composition. + * If this argument is false, only canonical + * decomposition will be performed. + * @param options the optional features to be enabled (0 for no options) + * @param result The composed string (on output). + * @param status The error code. + * @deprecated ICU 56 Use Normalizer2 instead. + */ + static void U_EXPORT2 compose(const UnicodeString& source, + UBool compat, int32_t options, + UnicodeString& result, + UErrorCode &status); + + /** + * Static method to decompose a UnicodeString. + * This is equivalent to normalize() with mode UNORM_NFD or UNORM_NFKD. + * This is a wrapper for unorm_normalize(), using UnicodeString's. + * + * The options parameter specifies which optional + * Normalizer features are to be enabled for this operation. + * + * @param source the string to be decomposed. + * @param compat Perform compatibility decomposition. + * If this argument is false, only canonical + * decomposition will be performed. + * @param options the optional features to be enabled (0 for no options) + * @param result The decomposed string (on output). + * @param status The error code. + * @deprecated ICU 56 Use Normalizer2 instead. + */ + static void U_EXPORT2 decompose(const UnicodeString& source, + UBool compat, int32_t options, + UnicodeString& result, + UErrorCode &status); + + /** + * Performing quick check on a string, to quickly determine if the string is + * in a particular normalization format. + * This is a wrapper for unorm_quickCheck(), using a UnicodeString. + * + * Three types of result can be returned UNORM_YES, UNORM_NO or + * UNORM_MAYBE. Result UNORM_YES indicates that the argument + * string is in the desired normalized format, UNORM_NO determines that + * argument string is not in the desired normalized format. A + * UNORM_MAYBE result indicates that a more thorough check is required, + * the user may have to put the string in its normalized form and compare the + * results. + * @param source string for determining if it is in a normalized format + * @param mode normalization format + * @param status A reference to a UErrorCode to receive any errors + * @return UNORM_YES, UNORM_NO or UNORM_MAYBE + * + * @see isNormalized + * @deprecated ICU 56 Use Normalizer2 instead. + */ + static inline UNormalizationCheckResult + quickCheck(const UnicodeString &source, UNormalizationMode mode, UErrorCode &status); + + /** + * Performing quick check on a string; same as the other version of quickCheck + * but takes an extra options parameter like most normalization functions. + * + * @param source string for determining if it is in a normalized format + * @param mode normalization format + * @param options the optional features to be enabled (0 for no options) + * @param status A reference to a UErrorCode to receive any errors + * @return UNORM_YES, UNORM_NO or UNORM_MAYBE + * + * @see isNormalized + * @deprecated ICU 56 Use Normalizer2 instead. + */ + static UNormalizationCheckResult + quickCheck(const UnicodeString &source, UNormalizationMode mode, int32_t options, UErrorCode &status); + + /** + * Test if a string is in a given normalization form. + * This is semantically equivalent to source.equals(normalize(source, mode)) . + * + * Unlike unorm_quickCheck(), this function returns a definitive result, + * never a "maybe". + * For NFD, NFKD, and FCD, both functions work exactly the same. + * For NFC and NFKC where quickCheck may return "maybe", this function will + * perform further tests to arrive at a true/false result. + * + * @param src String that is to be tested if it is in a normalization format. + * @param mode Which normalization form to test for. + * @param errorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return Boolean value indicating whether the source string is in the + * "mode" normalization form. + * + * @see quickCheck + * @deprecated ICU 56 Use Normalizer2 instead. + */ + static inline UBool + isNormalized(const UnicodeString &src, UNormalizationMode mode, UErrorCode &errorCode); + + /** + * Test if a string is in a given normalization form; same as the other version of isNormalized + * but takes an extra options parameter like most normalization functions. + * + * @param src String that is to be tested if it is in a normalization format. + * @param mode Which normalization form to test for. + * @param options the optional features to be enabled (0 for no options) + * @param errorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return Boolean value indicating whether the source string is in the + * "mode" normalization form. + * + * @see quickCheck + * @deprecated ICU 56 Use Normalizer2 instead. + */ + static UBool + isNormalized(const UnicodeString &src, UNormalizationMode mode, int32_t options, UErrorCode &errorCode); + + /** + * Concatenate normalized strings, making sure that the result is normalized as well. + * + * If both the left and the right strings are in + * the normalization form according to "mode/options", + * then the result will be + * + * \code + * dest=normalize(left+right, mode, options) + * \endcode + * + * For details see unorm_concatenate in unorm.h. + * + * @param left Left source string. + * @param right Right source string. + * @param result The output string. + * @param mode The normalization mode. + * @param options A bit set of normalization options. + * @param errorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return result + * + * @see unorm_concatenate + * @see normalize + * @see unorm_next + * @see unorm_previous + * + * @deprecated ICU 56 Use Normalizer2 instead. + */ + static UnicodeString & + U_EXPORT2 concatenate(const UnicodeString &left, const UnicodeString &right, + UnicodeString &result, + UNormalizationMode mode, int32_t options, + UErrorCode &errorCode); +#endif /* U_HIDE_DEPRECATED_API */ + + /** + * Compare two strings for canonical equivalence. + * Further options include case-insensitive comparison and + * code point order (as opposed to code unit order). + * + * Canonical equivalence between two strings is defined as their normalized + * forms (NFD or NFC) being identical. + * This function compares strings incrementally instead of normalizing + * (and optionally case-folding) both strings entirely, + * improving performance significantly. + * + * Bulk normalization is only necessary if the strings do not fulfill the FCD + * conditions. Only in this case, and only if the strings are relatively long, + * is memory allocated temporarily. + * For FCD strings and short non-FCD strings there is no memory allocation. + * + * Semantically, this is equivalent to + * strcmp[CodePointOrder](NFD(foldCase(s1)), NFD(foldCase(s2))) + * where code point order and foldCase are all optional. + * + * UAX 21 2.5 Caseless Matching specifies that for a canonical caseless match + * the case folding must be performed first, then the normalization. + * + * @param s1 First source string. + * @param s2 Second source string. + * + * @param options A bit set of options: + * - U_FOLD_CASE_DEFAULT or 0 is used for default options: + * Case-sensitive comparison in code unit order, and the input strings + * are quick-checked for FCD. + * + * - UNORM_INPUT_IS_FCD + * Set if the caller knows that both s1 and s2 fulfill the FCD conditions. + * If not set, the function will quickCheck for FCD + * and normalize if necessary. + * + * - U_COMPARE_CODE_POINT_ORDER + * Set to choose code point order instead of code unit order + * (see u_strCompare for details). + * + * - U_COMPARE_IGNORE_CASE + * Set to compare strings case-insensitively using case folding, + * instead of case-sensitively. + * If set, then the following case folding options are used. + * + * - Options as used with case-insensitive comparisons, currently: + * + * - U_FOLD_CASE_EXCLUDE_SPECIAL_I + * (see u_strCaseCompare for details) + * + * - regular normalization options shifted left by UNORM_COMPARE_NORM_OPTIONS_SHIFT + * + * @param errorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return <0 or 0 or >0 as usual for string comparisons + * + * @see unorm_compare + * @see normalize + * @see UNORM_FCD + * @see u_strCompare + * @see u_strCaseCompare + * + * @stable ICU 2.2 + */ + static inline int32_t + compare(const UnicodeString &s1, const UnicodeString &s2, + uint32_t options, + UErrorCode &errorCode); + +#ifndef U_HIDE_DEPRECATED_API + //------------------------------------------------------------------------- + // Iteration API + //------------------------------------------------------------------------- + + /** + * Return the current character in the normalized text. + * current() may need to normalize some text at getIndex(). + * The getIndex() is not changed. + * + * @return the current normalized code point + * @deprecated ICU 56 Use Normalizer2 instead. + */ + UChar32 current(void); + + /** + * Return the first character in the normalized text. + * This is equivalent to setIndexOnly(startIndex()) followed by next(). + * (Post-increment semantics.) + * + * @return the first normalized code point + * @deprecated ICU 56 Use Normalizer2 instead. + */ + UChar32 first(void); + + /** + * Return the last character in the normalized text. + * This is equivalent to setIndexOnly(endIndex()) followed by previous(). + * (Pre-decrement semantics.) + * + * @return the last normalized code point + * @deprecated ICU 56 Use Normalizer2 instead. + */ + UChar32 last(void); + + /** + * Return the next character in the normalized text. + * (Post-increment semantics.) + * If the end of the text has already been reached, DONE is returned. + * The DONE value could be confused with a U+FFFF non-character code point + * in the text. If this is possible, you can test getIndex()startIndex() || first()!=DONE). (Calling first() will change + * the iterator state!) + * + * The C API unorm_previous() is more efficient and does not have this ambiguity. + * + * @return the previous normalized code point + * @deprecated ICU 56 Use Normalizer2 instead. + */ + UChar32 previous(void); + + /** + * Set the iteration position in the input text that is being normalized, + * without any immediate normalization. + * After setIndexOnly(), getIndex() will return the same index that is + * specified here. + * + * @param index the desired index in the input text. + * @deprecated ICU 56 Use Normalizer2 instead. + */ + void setIndexOnly(int32_t index); + + /** + * Reset the index to the beginning of the text. + * This is equivalent to setIndexOnly(startIndex)). + * @deprecated ICU 56 Use Normalizer2 instead. + */ + void reset(void); + + /** + * Retrieve the current iteration position in the input text that is + * being normalized. + * + * A following call to next() will return a normalized code point from + * the input text at or after this index. + * + * After a call to previous(), getIndex() will point at or before the + * position in the input text where the normalized code point + * was returned from with previous(). + * + * @return the current index in the input text + * @deprecated ICU 56 Use Normalizer2 instead. + */ + int32_t getIndex(void) const; + + /** + * Retrieve the index of the start of the input text. This is the begin index + * of the CharacterIterator or the start (i.e. index 0) of the string + * over which this Normalizer is iterating. + * + * @return the smallest index in the input text where the Normalizer operates + * @deprecated ICU 56 Use Normalizer2 instead. + */ + int32_t startIndex(void) const; + + /** + * Retrieve the index of the end of the input text. This is the end index + * of the CharacterIterator or the length of the string + * over which this Normalizer is iterating. + * This end index is exclusive, i.e., the Normalizer operates only on characters + * before this index. + * + * @return the first index in the input text where the Normalizer does not operate + * @deprecated ICU 56 Use Normalizer2 instead. + */ + int32_t endIndex(void) const; + + /** + * Returns true when both iterators refer to the same character in the same + * input text. + * + * @param that a Normalizer object to compare this one to + * @return comparison result + * @deprecated ICU 56 Use Normalizer2 instead. + */ + bool operator==(const Normalizer& that) const; + + /** + * Returns false when both iterators refer to the same character in the same + * input text. + * + * @param that a Normalizer object to compare this one to + * @return comparison result + * @deprecated ICU 56 Use Normalizer2 instead. + */ + inline bool operator!=(const Normalizer& that) const; + + /** + * Returns a pointer to a new Normalizer that is a clone of this one. + * The caller is responsible for deleting the new clone. + * @return a pointer to a new Normalizer + * @deprecated ICU 56 Use Normalizer2 instead. + */ + Normalizer* clone() const; + + /** + * Generates a hash code for this iterator. + * + * @return the hash code + * @deprecated ICU 56 Use Normalizer2 instead. + */ + int32_t hashCode(void) const; + + //------------------------------------------------------------------------- + // Property access methods + //------------------------------------------------------------------------- + + /** + * Set the normalization mode for this object. + *

+ * Note:If the normalization mode is changed while iterating + * over a string, calls to {@link #next() } and {@link #previous() } may + * return previously buffers characters in the old normalization mode + * until the iteration is able to re-sync at the next base character. + * It is safest to call {@link #setIndexOnly }, {@link #reset() }, + * {@link #setText }, {@link #first() }, + * {@link #last() }, etc. after calling setMode. + *

+ * @param newMode the new mode for this Normalizer. + * @see #getUMode + * @deprecated ICU 56 Use Normalizer2 instead. + */ + void setMode(UNormalizationMode newMode); + + /** + * Return the normalization mode for this object. + * + * This is an unusual name because there used to be a getMode() that + * returned a different type. + * + * @return the mode for this Normalizer + * @see #setMode + * @deprecated ICU 56 Use Normalizer2 instead. + */ + UNormalizationMode getUMode(void) const; + + /** + * Set options that affect this Normalizer's operation. + * Options do not change the basic composition or decomposition operation + * that is being performed, but they control whether + * certain optional portions of the operation are done. + * Currently the only available option is obsolete. + * + * It is possible to specify multiple options that are all turned on or off. + * + * @param option the option(s) whose value is/are to be set. + * @param value the new setting for the option. Use true to + * turn the option(s) on and false to turn it/them off. + * + * @see #getOption + * @deprecated ICU 56 Use Normalizer2 instead. + */ + void setOption(int32_t option, + UBool value); + + /** + * Determine whether an option is turned on or off. + * If multiple options are specified, then the result is true if any + * of them are set. + *

+ * @param option the option(s) that are to be checked + * @return true if any of the option(s) are set + * @see #setOption + * @deprecated ICU 56 Use Normalizer2 instead. + */ + UBool getOption(int32_t option) const; + + /** + * Set the input text over which this Normalizer will iterate. + * The iteration position is set to the beginning. + * + * @param newText a string that replaces the current input text + * @param status a UErrorCode + * @deprecated ICU 56 Use Normalizer2 instead. + */ + void setText(const UnicodeString& newText, + UErrorCode &status); + + /** + * Set the input text over which this Normalizer will iterate. + * The iteration position is set to the beginning. + * + * @param newText a CharacterIterator object that replaces the current input text + * @param status a UErrorCode + * @deprecated ICU 56 Use Normalizer2 instead. + */ + void setText(const CharacterIterator& newText, + UErrorCode &status); + + /** + * Set the input text over which this Normalizer will iterate. + * The iteration position is set to the beginning. + * + * @param newText a string that replaces the current input text + * @param length the length of the string, or -1 if NUL-terminated + * @param status a UErrorCode + * @deprecated ICU 56 Use Normalizer2 instead. + */ + void setText(ConstChar16Ptr newText, + int32_t length, + UErrorCode &status); + /** + * Copies the input text into the UnicodeString argument. + * + * @param result Receives a copy of the text under iteration. + * @deprecated ICU 56 Use Normalizer2 instead. + */ + void getText(UnicodeString& result); + + /** + * ICU "poor man's RTTI", returns a UClassID for this class. + * @returns a UClassID for this class. + * @deprecated ICU 56 Use Normalizer2 instead. + */ + static UClassID U_EXPORT2 getStaticClassID(); +#endif /* U_HIDE_DEPRECATED_API */ + +#ifndef U_FORCE_HIDE_DEPRECATED_API + /** + * ICU "poor man's RTTI", returns a UClassID for the actual class. + * @return a UClassID for the actual class. + * @deprecated ICU 56 Use Normalizer2 instead. + */ + virtual UClassID getDynamicClassID() const override; +#endif // U_FORCE_HIDE_DEPRECATED_API + +private: + //------------------------------------------------------------------------- + // Private functions + //------------------------------------------------------------------------- + + Normalizer() = delete; // default constructor not implemented + Normalizer &operator=(const Normalizer &that) = delete; // assignment operator not implemented + + // Private utility methods for iteration + // For documentation, see the source code + UBool nextNormalize(); + UBool previousNormalize(); + + void init(); + void clearBuffer(void); + + //------------------------------------------------------------------------- + // Private data + //------------------------------------------------------------------------- + + FilteredNormalizer2*fFilteredNorm2; // owned if not nullptr + const Normalizer2 *fNorm2; // not owned; may be equal to fFilteredNorm2 + UNormalizationMode fUMode; // deprecated + int32_t fOptions; + + // The input text and our position in it + CharacterIterator *text; + + // The normalization buffer is the result of normalization + // of the source in [currentIndex..nextIndex[ . + int32_t currentIndex, nextIndex; + + // A buffer for holding intermediate results + UnicodeString buffer; + int32_t bufferPos; +}; + +//------------------------------------------------------------------------- +// Inline implementations +//------------------------------------------------------------------------- + +#ifndef U_HIDE_DEPRECATED_API +inline bool +Normalizer::operator!= (const Normalizer& other) const +{ return ! operator==(other); } + +inline UNormalizationCheckResult +Normalizer::quickCheck(const UnicodeString& source, + UNormalizationMode mode, + UErrorCode &status) { + return quickCheck(source, mode, 0, status); +} + +inline UBool +Normalizer::isNormalized(const UnicodeString& source, + UNormalizationMode mode, + UErrorCode &status) { + return isNormalized(source, mode, 0, status); +} +#endif /* U_HIDE_DEPRECATED_API */ + +inline int32_t +Normalizer::compare(const UnicodeString &s1, const UnicodeString &s2, + uint32_t options, + UErrorCode &errorCode) { + // all argument checking is done in unorm_compare + return unorm_compare(toUCharPtr(s1.getBuffer()), s1.length(), + toUCharPtr(s2.getBuffer()), s2.length(), + options, + &errorCode); +} + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_NORMALIZATION */ + +#endif // NORMLZR_H + +#endif /* U_SHOW_CPLUSPLUS_API */ diff --git a/intl/icu/source/common/unicode/parseerr.h b/intl/icu/source/common/unicode/parseerr.h new file mode 100644 index 0000000000..c23cc273b8 --- /dev/null +++ b/intl/icu/source/common/unicode/parseerr.h @@ -0,0 +1,94 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1999-2005, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* Date Name Description +* 03/14/00 aliu Creation. +* 06/27/00 aliu Change from C++ class to C struct +********************************************************************** +*/ +#ifndef PARSEERR_H +#define PARSEERR_H + +#include "unicode/utypes.h" + + +/** + * \file + * \brief C API: Parse Error Information + */ +/** + * The capacity of the context strings in UParseError. + * @stable ICU 2.0 + */ +enum { U_PARSE_CONTEXT_LEN = 16 }; + +/** + * A UParseError struct is used to returned detailed information about + * parsing errors. It is used by ICU parsing engines that parse long + * rules, patterns, or programs, where the text being parsed is long + * enough that more information than a UErrorCode is needed to + * localize the error. + * + *

The line, offset, and context fields are optional; parsing + * engines may choose not to use to use them. + * + *

The preContext and postContext strings include some part of the + * context surrounding the error. If the source text is "let for=7" + * and "for" is the error (e.g., because it is a reserved word), then + * some examples of what a parser might produce are the following: + * + * 

+ * preContext   postContext
+ * ""           ""            The parser does not support context
+ * "let "       "=7"          Pre- and post-context only
+ * "let "       "for=7"       Pre- and post-context and error text
+ * ""           "for"         Error text only
+ *
+ * + *

Examples of engines which use UParseError (or may use it in the + * future) are Transliterator, RuleBasedBreakIterator, and + * RegexPattern. + * + * @stable ICU 2.0 + */ +typedef struct UParseError { + + /** + * The line on which the error occurred. If the parser uses this + * field, it sets it to the line number of the source text line on + * which the error appears, which will be a value >= 1. If the + * parse does not support line numbers, the value will be <= 0. + * @stable ICU 2.0 + */ + int32_t line; + + /** + * The character offset to the error. If the line field is >= 1, + * then this is the offset from the start of the line. Otherwise, + * this is the offset from the start of the text. If the parser + * does not support this field, it will have a value < 0. + * @stable ICU 2.0 + */ + int32_t offset; + + /** + * Textual context before the error. Null-terminated. The empty + * string if not supported by parser. + * @stable ICU 2.0 + */ + UChar preContext[U_PARSE_CONTEXT_LEN]; + + /** + * The error itself and/or textual context after the error. + * Null-terminated. The empty string if not supported by parser. + * @stable ICU 2.0 + */ + UChar postContext[U_PARSE_CONTEXT_LEN]; + +} UParseError; + +#endif diff --git a/intl/icu/source/common/unicode/parsepos.h b/intl/icu/source/common/unicode/parsepos.h new file mode 100644 index 0000000000..d33a812ad0 --- /dev/null +++ b/intl/icu/source/common/unicode/parsepos.h @@ -0,0 +1,237 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +* Copyright (C) 1997-2005, International Business Machines Corporation and others. All Rights Reserved. +******************************************************************************* +* +* File PARSEPOS.H +* +* Modification History: +* +* Date Name Description +* 07/09/97 helena Converted from java. +* 07/17/98 stephen Added errorIndex support. +* 05/11/99 stephen Cleaned up. +******************************************************************************* +*/ + +#ifndef PARSEPOS_H +#define PARSEPOS_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/uobject.h" + + +U_NAMESPACE_BEGIN + +/** + * \file + * \brief C++ API: Canonical Iterator + */ +/** + * ParsePosition is a simple class used by Format + * and its subclasses to keep track of the current position during parsing. + * The parseObject method in the various Format + * classes requires a ParsePosition object as an argument. + * + *

+ * By design, as you parse through a string with different formats, + * you can use the same ParsePosition, since the index parameter + * records the current position. + * + * The ParsePosition class is not suitable for subclassing. + * + * @version 1.3 10/30/97 + * @author Mark Davis, Helena Shih + * @see java.text.Format + */ + +class U_COMMON_API ParsePosition : public UObject { +public: + /** + * Default constructor, the index starts with 0 as default. + * @stable ICU 2.0 + */ + ParsePosition() + : UObject(), + index(0), + errorIndex(-1) + {} + + /** + * Create a new ParsePosition with the given initial index. + * @param newIndex the new text offset. + * @stable ICU 2.0 + */ + ParsePosition(int32_t newIndex) + : UObject(), + index(newIndex), + errorIndex(-1) + {} + + /** + * Copy constructor + * @param copy the object to be copied from. + * @stable ICU 2.0 + */ + ParsePosition(const ParsePosition& copy) + : UObject(copy), + index(copy.index), + errorIndex(copy.errorIndex) + {} + + /** + * Destructor + * @stable ICU 2.0 + */ + virtual ~ParsePosition(); + + /** + * Assignment operator + * @stable ICU 2.0 + */ + inline ParsePosition& operator=(const ParsePosition& copy); + + /** + * Equality operator. + * @return true if the two parse positions are equal, false otherwise. + * @stable ICU 2.0 + */ + inline bool operator==(const ParsePosition& that) const; + + /** + * Equality operator. + * @return true if the two parse positions are not equal, false otherwise. + * @stable ICU 2.0 + */ + inline bool operator!=(const ParsePosition& that) const; + + /** + * Clone this object. + * Clones can be used concurrently in multiple threads. + * If an error occurs, then nullptr is returned. + * The caller must delete the clone. + * + * @return a clone of this object + * + * @see getDynamicClassID + * @stable ICU 2.8 + */ + ParsePosition *clone() const; + + /** + * Retrieve the current parse position. On input to a parse method, this + * is the index of the character at which parsing will begin; on output, it + * is the index of the character following the last character parsed. + * @return the current index. + * @stable ICU 2.0 + */ + inline int32_t getIndex(void) const; + + /** + * Set the current parse position. + * @param index the new index. + * @stable ICU 2.0 + */ + inline void setIndex(int32_t index); + + /** + * Set the index at which a parse error occurred. Formatters + * should set this before returning an error code from their + * parseObject method. The default value is -1 if this is not + * set. + * @stable ICU 2.0 + */ + inline void setErrorIndex(int32_t ei); + + /** + * Retrieve the index at which an error occurred, or -1 if the + * error index has not been set. + * @stable ICU 2.0 + */ + inline int32_t getErrorIndex(void) const; + + /** + * ICU "poor man's RTTI", returns a UClassID for this class. + * + * @stable ICU 2.2 + */ + static UClassID U_EXPORT2 getStaticClassID(); + + /** + * ICU "poor man's RTTI", returns a UClassID for the actual class. + * + * @stable ICU 2.2 + */ + virtual UClassID getDynamicClassID() const override; + +private: + /** + * Input: the place you start parsing. + *
Output: position where the parse stopped. + * This is designed to be used serially, + * with each call setting index up for the next one. + */ + int32_t index; + + /** + * The index at which a parse error occurred. + */ + int32_t errorIndex; + +}; + +inline ParsePosition& +ParsePosition::operator=(const ParsePosition& copy) +{ + index = copy.index; + errorIndex = copy.errorIndex; + return *this; +} + +inline bool +ParsePosition::operator==(const ParsePosition& copy) const +{ + if(index != copy.index || errorIndex != copy.errorIndex) + return false; + else + return true; +} + +inline bool +ParsePosition::operator!=(const ParsePosition& copy) const +{ + return !operator==(copy); +} + +inline int32_t +ParsePosition::getIndex() const +{ + return index; +} + +inline void +ParsePosition::setIndex(int32_t offset) +{ + this->index = offset; +} + +inline int32_t +ParsePosition::getErrorIndex() const +{ + return errorIndex; +} + +inline void +ParsePosition::setErrorIndex(int32_t ei) +{ + this->errorIndex = ei; +} +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/platform.h b/intl/icu/source/common/unicode/platform.h new file mode 100644 index 0000000000..a997843660 --- /dev/null +++ b/intl/icu/source/common/unicode/platform.h @@ -0,0 +1,881 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1997-2016, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* FILE NAME : platform.h +* +* Date Name Description +* 05/13/98 nos Creation (content moved here from ptypes.h). +* 03/02/99 stephen Added AS400 support. +* 03/30/99 stephen Added Linux support. +* 04/13/99 stephen Reworked for autoconf. +****************************************************************************** +*/ + +#ifndef _PLATFORM_H +#define _PLATFORM_H + +#include "unicode/uconfig.h" +#include "unicode/uvernum.h" + +/** + * \file + * \brief Basic types for the platform. + * + * This file used to be generated by autoconf/configure. + * Starting with ICU 49, platform.h is a normal source file, + * to simplify cross-compiling and working with non-autoconf/make build systems. + * + * When a value in this file does not work on a platform, then please + * try to derive it from the U_PLATFORM value + * (for which we might need a new value constant in rare cases) + * and/or from other macros that are predefined by the compiler + * or defined in standard (POSIX or platform or compiler) headers. + * + * As a temporary workaround, you can add an explicit \#define for some macros + * before it is first tested, or add an equivalent -D macro definition + * to the compiler's command line. + * + * Note: Some compilers provide ways to show the predefined macros. + * For example, with gcc you can compile an empty .c file and have the compiler + * print the predefined macros with + * \code + * gcc -E -dM -x c /dev/null | sort + * \endcode + * (You can provide an actual empty .c file rather than /dev/null. + * -x c++ is for C++.) + */ + +/** + * Define some things so that they can be documented. + * @internal + */ +#ifdef U_IN_DOXYGEN +/* + * Problem: "platform.h:335: warning: documentation for unknown define U_HAVE_STD_STRING found." means that U_HAVE_STD_STRING is not documented. + * Solution: #define any defines for non @internal API here, so that they are visible in the docs. If you just set PREDEFINED in Doxyfile.in, they won't be documented. + */ + +/* None for now. */ +#endif + +/** + * \def U_PLATFORM + * The U_PLATFORM macro defines the platform we're on. + * + * We used to define one different, value-less macro per platform. + * That made it hard to know the set of relevant platforms and macros, + * and hard to deal with variants of platforms. + * + * Starting with ICU 49, we define platforms as numeric macros, + * with ranges of values for related platforms and their variants. + * The U_PLATFORM macro is set to one of these values. + * + * Historical note from the Solaris Wikipedia article: + * AT&T and Sun collaborated on a project to merge the most popular Unix variants + * on the market at that time: BSD, System V, and Xenix. + * This became Unix System V Release 4 (SVR4). + * + * @internal + */ + +/** Unknown platform. @internal */ +#define U_PF_UNKNOWN 0 +/** Windows @internal */ +#define U_PF_WINDOWS 1000 +/** MinGW. Windows, calls to Win32 API, but using GNU gcc and binutils. @internal */ +#define U_PF_MINGW 1800 +/** + * Cygwin. Windows, calls to cygwin1.dll for Posix functions, + * using MSVC or GNU gcc and binutils. + * @internal + */ +#define U_PF_CYGWIN 1900 +/* Reserve 2000 for U_PF_UNIX? */ +/** HP-UX is based on UNIX System V. @internal */ +#define U_PF_HPUX 2100 +/** Solaris is a Unix operating system based on SVR4. @internal */ +#define U_PF_SOLARIS 2600 +/** BSD is a UNIX operating system derivative. @internal */ +#define U_PF_BSD 3000 +/** AIX is based on UNIX System V Releases and 4.3 BSD. @internal */ +#define U_PF_AIX 3100 +/** IRIX is based on UNIX System V with BSD extensions. @internal */ +#define U_PF_IRIX 3200 +/** + * Darwin is a POSIX-compliant operating system, composed of code developed by Apple, + * as well as code derived from NeXTSTEP, BSD, and other projects, + * built around the Mach kernel. + * Darwin forms the core set of components upon which Mac OS X, Apple TV, and iOS are based. + * (Original description modified from WikiPedia.) + * @internal + */ +#define U_PF_DARWIN 3500 +/** iPhone OS (iOS) is a derivative of Mac OS X. @internal */ +#define U_PF_IPHONE 3550 +/** QNX is a commercial Unix-like real-time operating system related to BSD. @internal */ +#define U_PF_QNX 3700 +/** Linux is a Unix-like operating system. @internal */ +#define U_PF_LINUX 4000 +/** + * Native Client is pretty close to Linux. + * See https://developer.chrome.com/native-client and + * http://www.chromium.org/nativeclient + * @internal + */ +#define U_PF_BROWSER_NATIVE_CLIENT 4020 +/** Android is based on Linux. @internal */ +#define U_PF_ANDROID 4050 +/** Fuchsia is a POSIX-ish platform. @internal */ +#define U_PF_FUCHSIA 4100 +/* Maximum value for Linux-based platform is 4499 */ +/** + * Emscripten is a C++ transpiler for the Web that can target asm.js or + * WebAssembly. It provides some POSIX-compatible wrappers and stubs and + * some Linux-like functionality, but is not fully compatible with + * either. + * @internal + */ +#define U_PF_EMSCRIPTEN 5010 +/** z/OS is the successor to OS/390 which was the successor to MVS. @internal */ +#define U_PF_OS390 9000 +/** "IBM i" is the current name of what used to be i5/OS and earlier OS/400. @internal */ +#define U_PF_OS400 9400 + +#ifdef U_PLATFORM + /* Use the predefined value. */ +#elif defined(__MINGW32__) +# define U_PLATFORM U_PF_MINGW +#elif defined(__CYGWIN__) +# define U_PLATFORM U_PF_CYGWIN +#elif defined(WIN32) || defined(_WIN32) || defined(WIN64) || defined(_WIN64) +# define U_PLATFORM U_PF_WINDOWS +#elif defined(__ANDROID__) +# define U_PLATFORM U_PF_ANDROID + /* Android wchar_t support depends on the API level. */ +# include +#elif defined(__pnacl__) || defined(__native_client__) +# define U_PLATFORM U_PF_BROWSER_NATIVE_CLIENT +#elif defined(__Fuchsia__) +# define U_PLATFORM U_PF_FUCHSIA +#elif defined(linux) || defined(__linux__) || defined(__linux) +# define U_PLATFORM U_PF_LINUX +#elif defined(__APPLE__) && defined(__MACH__) +# include +# if (defined(TARGET_OS_IPHONE) && TARGET_OS_IPHONE) && (defined(TARGET_OS_MACCATALYST) && !TARGET_OS_MACCATALYST) /* variant of TARGET_OS_MAC */ +# define U_PLATFORM U_PF_IPHONE +# else +# define U_PLATFORM U_PF_DARWIN +# endif +#elif defined(BSD) || defined(__FreeBSD__) || defined(__FreeBSD_kernel__) || defined(__NetBSD__) || defined(__OpenBSD__) || defined(__MirBSD__) +# if defined(__FreeBSD__) +# include +# endif +# define U_PLATFORM U_PF_BSD +#elif defined(sun) || defined(__sun) + /* Check defined(__SVR4) || defined(__svr4__) to distinguish Solaris from SunOS? */ +# define U_PLATFORM U_PF_SOLARIS +# if defined(__GNUC__) + /* Solaris/GCC needs this header file to get the proper endianness. Normally, this + * header file is included with stddef.h but on Solairs/GCC, the GCC version of stddef.h + * is included which does not include this header file. + */ +# include +# endif +#elif defined(_AIX) || defined(__TOS_AIX__) +# define U_PLATFORM U_PF_AIX +#elif defined(_hpux) || defined(hpux) || defined(__hpux) +# define U_PLATFORM U_PF_HPUX +#elif defined(sgi) || defined(__sgi) +# define U_PLATFORM U_PF_IRIX +#elif defined(__QNX__) || defined(__QNXNTO__) +# define U_PLATFORM U_PF_QNX +#elif defined(__TOS_MVS__) +# define U_PLATFORM U_PF_OS390 +#elif defined(__OS400__) || defined(__TOS_OS400__) +# define U_PLATFORM U_PF_OS400 +#elif defined(__EMSCRIPTEN__) +# define U_PLATFORM U_PF_EMSCRIPTEN +#else +# define U_PLATFORM U_PF_UNKNOWN +#endif + +/** + * \def CYGWINMSVC + * Defined if this is Windows with Cygwin, but using MSVC rather than gcc. + * Otherwise undefined. + * @internal + */ +/* Commented out because this is already set in mh-cygwin-msvc +#if U_PLATFORM == U_PF_CYGWIN && defined(_MSC_VER) +# define CYGWINMSVC +#endif +*/ +#ifdef U_IN_DOXYGEN +# define CYGWINMSVC +#endif + +/** + * \def U_PLATFORM_USES_ONLY_WIN32_API + * Defines whether the platform uses only the Win32 API. + * Set to 1 for Windows/MSVC and MinGW but not Cygwin. + * @internal + */ +#ifdef U_PLATFORM_USES_ONLY_WIN32_API + /* Use the predefined value. */ +#elif (U_PF_WINDOWS <= U_PLATFORM && U_PLATFORM <= U_PF_MINGW) || defined(CYGWINMSVC) +# define U_PLATFORM_USES_ONLY_WIN32_API 1 +#else + /* Cygwin implements POSIX. */ +# define U_PLATFORM_USES_ONLY_WIN32_API 0 +#endif + +/** + * \def U_PLATFORM_HAS_WIN32_API + * Defines whether the Win32 API is available on the platform. + * Set to 1 for Windows/MSVC, MinGW and Cygwin. + * @internal + */ +#ifdef U_PLATFORM_HAS_WIN32_API + /* Use the predefined value. */ +#elif U_PF_WINDOWS <= U_PLATFORM && U_PLATFORM <= U_PF_CYGWIN +# define U_PLATFORM_HAS_WIN32_API 1 +#else +# define U_PLATFORM_HAS_WIN32_API 0 +#endif + +/** + * \def U_PLATFORM_HAS_WINUWP_API + * Defines whether target is intended for Universal Windows Platform API + * Set to 1 for Windows10 Release Solution Configuration + * @internal + */ +#ifdef U_PLATFORM_HAS_WINUWP_API + /* Use the predefined value. */ +#else +# define U_PLATFORM_HAS_WINUWP_API 0 +#endif + +/** + * \def U_PLATFORM_IMPLEMENTS_POSIX + * Defines whether the platform implements (most of) the POSIX API. + * Set to 1 for Cygwin and most other platforms. + * @internal + */ +#ifdef U_PLATFORM_IMPLEMENTS_POSIX + /* Use the predefined value. */ +#elif U_PLATFORM_USES_ONLY_WIN32_API +# define U_PLATFORM_IMPLEMENTS_POSIX 0 +#else +# define U_PLATFORM_IMPLEMENTS_POSIX 1 +#endif + +/** + * \def U_PLATFORM_IS_LINUX_BASED + * Defines whether the platform is Linux or one of its derivatives. + * @internal + */ +#ifdef U_PLATFORM_IS_LINUX_BASED + /* Use the predefined value. */ +#elif U_PF_LINUX <= U_PLATFORM && U_PLATFORM <= 4499 +# define U_PLATFORM_IS_LINUX_BASED 1 +#else +# define U_PLATFORM_IS_LINUX_BASED 0 +#endif + +/** + * \def U_PLATFORM_IS_DARWIN_BASED + * Defines whether the platform is Darwin or one of its derivatives. + * @internal + */ +#ifdef U_PLATFORM_IS_DARWIN_BASED + /* Use the predefined value. */ +#elif U_PF_DARWIN <= U_PLATFORM && U_PLATFORM <= U_PF_IPHONE +# define U_PLATFORM_IS_DARWIN_BASED 1 +#else +# define U_PLATFORM_IS_DARWIN_BASED 0 +#endif + +/** + * \def U_HAVE_STDINT_H + * Defines whether stdint.h is available. It is a C99 standard header. + * We used to include inttypes.h which includes stdint.h but we usually do not need + * the additional definitions from inttypes.h. + * @internal + */ +#ifdef U_HAVE_STDINT_H + /* Use the predefined value. */ +#elif U_PLATFORM_USES_ONLY_WIN32_API +# if defined(__BORLANDC__) || U_PLATFORM == U_PF_MINGW || (defined(_MSC_VER) && _MSC_VER>=1600) + /* Windows Visual Studio 9 and below do not have stdint.h & inttypes.h, but VS 2010 adds them. */ +# define U_HAVE_STDINT_H 1 +# else +# define U_HAVE_STDINT_H 0 +# endif +#elif U_PLATFORM == U_PF_SOLARIS + /* Solaris has inttypes.h but not stdint.h. */ +# define U_HAVE_STDINT_H 0 +#elif U_PLATFORM == U_PF_AIX && !defined(_AIX51) && defined(_POWER) + /* PPC AIX <= 4.3 has inttypes.h but not stdint.h. */ +# define U_HAVE_STDINT_H 0 +#else +# define U_HAVE_STDINT_H 1 +#endif + +/** + * \def U_HAVE_INTTYPES_H + * Defines whether inttypes.h is available. It is a C99 standard header. + * We include inttypes.h where it is available but stdint.h is not. + * @internal + */ +#ifdef U_HAVE_INTTYPES_H + /* Use the predefined value. */ +#elif U_PLATFORM == U_PF_SOLARIS + /* Solaris has inttypes.h but not stdint.h. */ +# define U_HAVE_INTTYPES_H 1 +#elif U_PLATFORM == U_PF_AIX && !defined(_AIX51) && defined(_POWER) + /* PPC AIX <= 4.3 has inttypes.h but not stdint.h. */ +# define U_HAVE_INTTYPES_H 1 +#else + /* Most platforms have both inttypes.h and stdint.h, or neither. */ +# define U_HAVE_INTTYPES_H U_HAVE_STDINT_H +#endif + +/*===========================================================================*/ +/** @{ Compiler and environment features */ +/*===========================================================================*/ + +/** + * \def U_GCC_MAJOR_MINOR + * Indicates whether the compiler is gcc (test for != 0), + * and if so, contains its major (times 100) and minor version numbers. + * If the compiler is not gcc, then U_GCC_MAJOR_MINOR == 0. + * + * For example, for testing for whether we have gcc, and whether it's 4.6 or higher, + * use "#if U_GCC_MAJOR_MINOR >= 406". + * @internal + */ +#ifdef __GNUC__ +# define U_GCC_MAJOR_MINOR (__GNUC__ * 100 + __GNUC_MINOR__) +#else +# define U_GCC_MAJOR_MINOR 0 +#endif + +/** + * \def U_IS_BIG_ENDIAN + * Determines the endianness of the platform. + * @internal + */ +#ifdef U_IS_BIG_ENDIAN + /* Use the predefined value. */ +#elif defined(BYTE_ORDER) && defined(BIG_ENDIAN) +# define U_IS_BIG_ENDIAN (BYTE_ORDER == BIG_ENDIAN) +#elif defined(__BYTE_ORDER__) && defined(__ORDER_BIG_ENDIAN__) + /* gcc */ +# define U_IS_BIG_ENDIAN (__BYTE_ORDER__ == __ORDER_BIG_ENDIAN__) +#elif defined(__BIG_ENDIAN__) || defined(_BIG_ENDIAN) +# define U_IS_BIG_ENDIAN 1 +#elif defined(__LITTLE_ENDIAN__) || defined(_LITTLE_ENDIAN) +# define U_IS_BIG_ENDIAN 0 +#elif U_PLATFORM == U_PF_OS390 || U_PLATFORM == U_PF_OS400 || defined(__s390__) || defined(__s390x__) + /* These platforms do not appear to predefine any endianness macros. */ +# define U_IS_BIG_ENDIAN 1 +#elif defined(_PA_RISC1_0) || defined(_PA_RISC1_1) || defined(_PA_RISC2_0) + /* HPPA do not appear to predefine any endianness macros. */ +# define U_IS_BIG_ENDIAN 1 +#elif defined(sparc) || defined(__sparc) || defined(__sparc__) + /* Some sparc based systems (e.g. Linux) do not predefine any endianness macros. */ +# define U_IS_BIG_ENDIAN 1 +#else +# define U_IS_BIG_ENDIAN 0 +#endif + +/** + * \def U_HAVE_PLACEMENT_NEW + * Determines whether to override placement new and delete for STL. + * @stable ICU 2.6 + */ +#ifdef U_HAVE_PLACEMENT_NEW + /* Use the predefined value. */ +#elif defined(__BORLANDC__) +# define U_HAVE_PLACEMENT_NEW 0 +#else +# define U_HAVE_PLACEMENT_NEW 1 +#endif + +/** + * \def U_HAVE_DEBUG_LOCATION_NEW + * Define this to define the MFC debug version of the operator new. + * + * @stable ICU 3.4 + */ +#ifdef U_HAVE_DEBUG_LOCATION_NEW + /* Use the predefined value. */ +#elif defined(_MSC_VER) +# define U_HAVE_DEBUG_LOCATION_NEW 1 +#else +# define U_HAVE_DEBUG_LOCATION_NEW 0 +#endif + +/* Compatibility with compilers other than clang: http://clang.llvm.org/docs/LanguageExtensions.html */ +#ifdef __has_attribute +# define UPRV_HAS_ATTRIBUTE(x) __has_attribute(x) +#else +# define UPRV_HAS_ATTRIBUTE(x) 0 +#endif +#ifdef __has_cpp_attribute +# define UPRV_HAS_CPP_ATTRIBUTE(x) __has_cpp_attribute(x) +#else +# define UPRV_HAS_CPP_ATTRIBUTE(x) 0 +#endif +#ifdef __has_declspec_attribute +# define UPRV_HAS_DECLSPEC_ATTRIBUTE(x) __has_declspec_attribute(x) +#else +# define UPRV_HAS_DECLSPEC_ATTRIBUTE(x) 0 +#endif +#ifdef __has_builtin +# define UPRV_HAS_BUILTIN(x) __has_builtin(x) +#else +# define UPRV_HAS_BUILTIN(x) 0 +#endif +#ifdef __has_feature +# define UPRV_HAS_FEATURE(x) __has_feature(x) +#else +# define UPRV_HAS_FEATURE(x) 0 +#endif +#ifdef __has_extension +# define UPRV_HAS_EXTENSION(x) __has_extension(x) +#else +# define UPRV_HAS_EXTENSION(x) 0 +#endif +#ifdef __has_warning +# define UPRV_HAS_WARNING(x) __has_warning(x) +#else +# define UPRV_HAS_WARNING(x) 0 +#endif + + +#if defined(__clang__) +#define UPRV_NO_SANITIZE_UNDEFINED __attribute__((no_sanitize("undefined"))) +#else +#define UPRV_NO_SANITIZE_UNDEFINED +#endif + +/** + * \def U_MALLOC_ATTR + * Attribute to mark functions as malloc-like + * @internal + */ +#if defined(__GNUC__) && __GNUC__>=3 +# define U_MALLOC_ATTR __attribute__ ((__malloc__)) +#else +# define U_MALLOC_ATTR +#endif + +/** + * \def U_ALLOC_SIZE_ATTR + * Attribute to specify the size of the allocated buffer for malloc-like functions + * @internal + */ +#if (defined(__GNUC__) && \ + (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3))) || \ + UPRV_HAS_ATTRIBUTE(alloc_size) +# define U_ALLOC_SIZE_ATTR(X) __attribute__ ((alloc_size(X))) +# define U_ALLOC_SIZE_ATTR2(X,Y) __attribute__ ((alloc_size(X,Y))) +#else +# define U_ALLOC_SIZE_ATTR(X) +# define U_ALLOC_SIZE_ATTR2(X,Y) +#endif + +/** + * \def U_CPLUSPLUS_VERSION + * 0 if no C++; 1, 11, 14, ... if C++. + * Support for specific features cannot always be determined by the C++ version alone. + * @internal + */ +#ifdef U_CPLUSPLUS_VERSION +# if U_CPLUSPLUS_VERSION != 0 && !defined(__cplusplus) +# undef U_CPLUSPLUS_VERSION +# define U_CPLUSPLUS_VERSION 0 +# endif + /* Otherwise use the predefined value. */ +#elif !defined(__cplusplus) +# define U_CPLUSPLUS_VERSION 0 +#elif __cplusplus >= 201402L || (defined(_MSVC_LANG) && _MSVC_LANG >= 201402L) +# define U_CPLUSPLUS_VERSION 14 +#elif __cplusplus >= 201103L || (defined(_MSVC_LANG) && _MSVC_LANG >= 201103L) +# define U_CPLUSPLUS_VERSION 11 +#else + // C++98 or C++03 +# define U_CPLUSPLUS_VERSION 1 +#endif + +/** + * \def U_FALLTHROUGH + * Annotate intentional fall-through between switch labels. + * http://clang.llvm.org/docs/AttributeReference.html#fallthrough-clang-fallthrough + * @internal + */ +#ifndef __cplusplus + // Not for C. +#elif defined(U_FALLTHROUGH) + // Use the predefined value. +#elif defined(__clang__) + // Test for compiler vs. feature separately. + // Other compilers might choke on the feature test. +# if UPRV_HAS_CPP_ATTRIBUTE(clang::fallthrough) || \ + (UPRV_HAS_FEATURE(cxx_attributes) && \ + UPRV_HAS_WARNING("-Wimplicit-fallthrough")) +# define U_FALLTHROUGH [[clang::fallthrough]] +# endif +#elif defined(__GNUC__) && (__GNUC__ >= 7) +# define U_FALLTHROUGH __attribute__((fallthrough)) +#endif + +#ifndef U_FALLTHROUGH +# define U_FALLTHROUGH +#endif + +/** @} */ + +/*===========================================================================*/ +/** @{ Character data types */ +/*===========================================================================*/ + +/** + * U_CHARSET_FAMILY is equal to this value when the platform is an ASCII based platform. + * @stable ICU 2.0 + */ +#define U_ASCII_FAMILY 0 + +/** + * U_CHARSET_FAMILY is equal to this value when the platform is an EBCDIC based platform. + * @stable ICU 2.0 + */ +#define U_EBCDIC_FAMILY 1 + +/** + * \def U_CHARSET_FAMILY + * + *

These definitions allow to specify the encoding of text + * in the char data type as defined by the platform and the compiler. + * It is enough to determine the code point values of "invariant characters", + * which are the ones shared by all encodings that are in use + * on a given platform.

+ * + *

Those "invariant characters" should be all the uppercase and lowercase + * latin letters, the digits, the space, and "basic punctuation". + * Also, '\\n', '\\r', '\\t' should be available.

+ * + *

The list of "invariant characters" is:
+ * \code + * A-Z a-z 0-9 SPACE " % & ' ( ) * + , - . / : ; < = > ? _ + * \endcode + *
+ * (52 letters + 10 numbers + 20 punc/sym/space = 82 total)

+ * + *

This matches the IBM Syntactic Character Set (CS 640).

+ * + *

In other words, all the graphic characters in 7-bit ASCII should + * be safely accessible except the following:

+ * + * \code + * '\' + * '[' + * ']' + * '{' + * '}' + * '^' + * '~' + * '!' + * '#' + * '|' + * '$' + * '@' + * '`' + * \endcode + * @stable ICU 2.0 + */ +#ifdef U_CHARSET_FAMILY + /* Use the predefined value. */ +#elif U_PLATFORM == U_PF_OS390 && (!defined(__CHARSET_LIB) || !__CHARSET_LIB) +# define U_CHARSET_FAMILY U_EBCDIC_FAMILY +#elif U_PLATFORM == U_PF_OS400 && !defined(__UTF32__) +# define U_CHARSET_FAMILY U_EBCDIC_FAMILY +#else +# define U_CHARSET_FAMILY U_ASCII_FAMILY +#endif + +/** + * \def U_CHARSET_IS_UTF8 + * + * Hardcode the default charset to UTF-8. + * + * If this is set to 1, then + * - ICU will assume that all non-invariant char*, StringPiece, std::string etc. + * contain UTF-8 text, regardless of what the system API uses + * - some ICU code will use fast functions like u_strFromUTF8() + * rather than the more general and more heavy-weight conversion API (ucnv.h) + * - ucnv_getDefaultName() always returns "UTF-8" + * - ucnv_setDefaultName() is disabled and will not change the default charset + * - static builds of ICU are smaller + * - more functionality is available with the UCONFIG_NO_CONVERSION build-time + * configuration option (see unicode/uconfig.h) + * - the UCONFIG_NO_CONVERSION build option in uconfig.h is more usable + * + * @stable ICU 4.2 + * @see UCONFIG_NO_CONVERSION + */ +#ifdef U_CHARSET_IS_UTF8 + /* Use the predefined value. */ +#elif U_PLATFORM_IS_LINUX_BASED || U_PLATFORM_IS_DARWIN_BASED || \ + U_PLATFORM == U_PF_EMSCRIPTEN +# define U_CHARSET_IS_UTF8 1 +#else +# define U_CHARSET_IS_UTF8 0 +#endif + +/** @} */ + +/*===========================================================================*/ +/** @{ Information about wchar support */ +/*===========================================================================*/ + +/** + * \def U_HAVE_WCHAR_H + * Indicates whether is available (1) or not (0). Set to 1 by default. + * + * @stable ICU 2.0 + */ +#ifdef U_HAVE_WCHAR_H + /* Use the predefined value. */ +#elif U_PLATFORM == U_PF_ANDROID && __ANDROID_API__ < 9 + /* + * Android before Gingerbread (Android 2.3, API level 9) did not support wchar_t. + * The type and header existed, but the library functions did not work as expected. + * The size of wchar_t was 1 but L"xyz" string literals had 32-bit units anyway. + */ +# define U_HAVE_WCHAR_H 0 +#else +# define U_HAVE_WCHAR_H 1 +#endif + +/** + * \def U_SIZEOF_WCHAR_T + * U_SIZEOF_WCHAR_T==sizeof(wchar_t) + * + * @stable ICU 2.0 + */ +#ifdef U_SIZEOF_WCHAR_T + /* Use the predefined value. */ +#elif (U_PLATFORM == U_PF_ANDROID && __ANDROID_API__ < 9) + /* + * Classic Mac OS and Mac OS X before 10.3 (Panther) did not support wchar_t or wstring. + * Newer Mac OS X has size 4. + */ +# define U_SIZEOF_WCHAR_T 1 +#elif U_PLATFORM_HAS_WIN32_API || U_PLATFORM == U_PF_CYGWIN +# define U_SIZEOF_WCHAR_T 2 +#elif U_PLATFORM == U_PF_AIX + /* + * AIX 6.1 information, section "Wide character data representation": + * "... the wchar_t datatype is 32-bit in the 64-bit environment and + * 16-bit in the 32-bit environment." + * and + * "All locales use Unicode for their wide character code values (process code), + * except the IBM-eucTW codeset." + */ +# ifdef __64BIT__ +# define U_SIZEOF_WCHAR_T 4 +# else +# define U_SIZEOF_WCHAR_T 2 +# endif +#elif U_PLATFORM == U_PF_OS390 + /* + * z/OS V1R11 information center, section "LP64 | ILP32": + * "In 31-bit mode, the size of long and pointers is 4 bytes and the size of wchar_t is 2 bytes. + * Under LP64, the size of long and pointer is 8 bytes and the size of wchar_t is 4 bytes." + */ +# ifdef _LP64 +# define U_SIZEOF_WCHAR_T 4 +# else +# define U_SIZEOF_WCHAR_T 2 +# endif +#elif U_PLATFORM == U_PF_OS400 +# if defined(__UTF32__) + /* + * LOCALETYPE(*LOCALEUTF) is specified. + * Wide-character strings are in UTF-32, + * narrow-character strings are in UTF-8. + */ +# define U_SIZEOF_WCHAR_T 4 +# elif defined(__UCS2__) + /* + * LOCALETYPE(*LOCALEUCS2) is specified. + * Wide-character strings are in UCS-2, + * narrow-character strings are in EBCDIC. + */ +# define U_SIZEOF_WCHAR_T 2 +# else + /* + * LOCALETYPE(*CLD) or LOCALETYPE(*LOCALE) is specified. + * Wide-character strings are in 16-bit EBCDIC, + * narrow-character strings are in EBCDIC. + */ +# define U_SIZEOF_WCHAR_T 2 +# endif +#else +# define U_SIZEOF_WCHAR_T 4 +#endif + +#ifndef U_HAVE_WCSCPY +#define U_HAVE_WCSCPY U_HAVE_WCHAR_H +#endif + +/** @} */ + +/** + * \def U_HAVE_CHAR16_T + * Defines whether the char16_t type is available for UTF-16 + * and u"abc" UTF-16 string literals are supported. + * This is a new standard type and standard string literal syntax in C++11 + * but has been available in some compilers before. + * @internal + */ +#ifdef U_HAVE_CHAR16_T + /* Use the predefined value. */ +#else + /* + * Notes: + * C++11 and C11 require support for UTF-16 literals + * TODO: Fix for plain C. Doesn't work on Mac. + */ +# if U_CPLUSPLUS_VERSION >= 11 || (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L) +# define U_HAVE_CHAR16_T 1 +# else +# define U_HAVE_CHAR16_T 0 +# endif +#endif + +/** + * @{ + * \def U_DECLARE_UTF16 + * Do not use this macro because it is not defined on all platforms. + * Use the UNICODE_STRING or U_STRING_DECL macros instead. + * @internal + */ +#ifdef U_DECLARE_UTF16 + /* Use the predefined value. */ +#elif U_HAVE_CHAR16_T \ + || (defined(__xlC__) && defined(__IBM_UTF_LITERAL) && U_SIZEOF_WCHAR_T != 2) \ + || (defined(__HP_aCC) && __HP_aCC >= 035000) \ + || (defined(__HP_cc) && __HP_cc >= 111106) \ + || (defined(U_IN_DOXYGEN)) +# define U_DECLARE_UTF16(string) u ## string +#elif U_SIZEOF_WCHAR_T == 2 \ + && (U_CHARSET_FAMILY == 0 || (U_PF_OS390 <= U_PLATFORM && U_PLATFORM <= U_PF_OS400 && defined(__UCS2__))) +# define U_DECLARE_UTF16(string) L ## string +#else + /* Leave U_DECLARE_UTF16 undefined. See unistr.h. */ +#endif + +/** @} */ + +/*===========================================================================*/ +/** @{ Symbol import-export control */ +/*===========================================================================*/ + +#ifdef U_EXPORT + /* Use the predefined value. */ +#elif defined(U_STATIC_IMPLEMENTATION) +# define U_EXPORT +#elif defined(_MSC_VER) || (UPRV_HAS_DECLSPEC_ATTRIBUTE(__dllexport__) && \ + UPRV_HAS_DECLSPEC_ATTRIBUTE(__dllimport__)) +# define U_EXPORT __declspec(dllexport) +#elif defined(__GNUC__) +# define U_EXPORT __attribute__((visibility("default"))) +#elif (defined(__SUNPRO_CC) && __SUNPRO_CC >= 0x550) \ + || (defined(__SUNPRO_C) && __SUNPRO_C >= 0x550) +# define U_EXPORT __global +/*#elif defined(__HP_aCC) || defined(__HP_cc) +# define U_EXPORT __declspec(dllexport)*/ +#else +# define U_EXPORT +#endif + +/* U_CALLCONV is related to U_EXPORT2 */ +#ifdef U_EXPORT2 + /* Use the predefined value. */ +#elif defined(_MSC_VER) +# define U_EXPORT2 __cdecl +#else +# define U_EXPORT2 +#endif + +#ifdef U_IMPORT + /* Use the predefined value. */ +#elif defined(_MSC_VER) || (UPRV_HAS_DECLSPEC_ATTRIBUTE(__dllexport__) && \ + UPRV_HAS_DECLSPEC_ATTRIBUTE(__dllimport__)) + /* Windows needs to export/import data. */ +# define U_IMPORT __declspec(dllimport) +#else +# define U_IMPORT +#endif + +/** + * \def U_HIDDEN + * This is used to mark internal structs declared within external classes, + * to prevent the internal structs from having the same visibility as the + * class within which they are declared. + * @internal + */ +#ifdef U_HIDDEN + /* Use the predefined value. */ +#elif defined(__GNUC__) +# define U_HIDDEN __attribute__((visibility("hidden"))) +#else +# define U_HIDDEN +#endif + +/** + * \def U_CALLCONV + * Similar to U_CDECL_BEGIN/U_CDECL_END, this qualifier is necessary + * in callback function typedefs to make sure that the calling convention + * is compatible. + * + * This is only used for non-ICU-API functions. + * When a function is a public ICU API, + * you must use the U_CAPI and U_EXPORT2 qualifiers. + * + * Please note, you need to use U_CALLCONV after the *. + * + * NO : "static const char U_CALLCONV *func( . . . )" + * YES: "static const char* U_CALLCONV func( . . . )" + * + * @stable ICU 2.0 + */ +#if U_PLATFORM == U_PF_OS390 && defined(__cplusplus) +# define U_CALLCONV __cdecl +#else +# define U_CALLCONV U_EXPORT2 +#endif + +/** + * \def U_CALLCONV_FPTR + * Similar to U_CALLCONV, but only used on function pointers. + * @internal + */ +#if U_PLATFORM == U_PF_OS390 && defined(__cplusplus) +# define U_CALLCONV_FPTR U_CALLCONV +#else +# define U_CALLCONV_FPTR +#endif +/** @} */ + +#endif // _PLATFORM_H diff --git a/intl/icu/source/common/unicode/ptypes.h b/intl/icu/source/common/unicode/ptypes.h new file mode 100644 index 0000000000..70324ffee3 --- /dev/null +++ b/intl/icu/source/common/unicode/ptypes.h @@ -0,0 +1,130 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1997-2012, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* FILE NAME : ptypes.h +* +* Date Name Description +* 05/13/98 nos Creation (content moved here from ptypes.h). +* 03/02/99 stephen Added AS400 support. +* 03/30/99 stephen Added Linux support. +* 04/13/99 stephen Reworked for autoconf. +* 09/18/08 srl Moved basic types back to ptypes.h from platform.h +****************************************************************************** +*/ + +/** + * \file + * \brief C API: Definitions of integer types of various widths + */ + +#ifndef _PTYPES_H +#define _PTYPES_H + +/** + * \def __STDC_LIMIT_MACROS + * According to the Linux stdint.h, the ISO C99 standard specifies that in C++ implementations + * macros like INT32_MIN and UINTPTR_MAX should only be defined if explicitly requested. + * We need to define __STDC_LIMIT_MACROS before including stdint.h in C++ code + * that uses such limit macros. + * @internal + */ +#ifndef __STDC_LIMIT_MACROS +#define __STDC_LIMIT_MACROS +#endif + +/* NULL, size_t, wchar_t */ +#include + +/* + * If all compilers provided all of the C99 headers and types, + * we would just unconditionally #include here + * and not need any of the stuff after including platform.h. + */ + +/* Find out if we have stdint.h etc. */ +#include "unicode/platform.h" + +/*===========================================================================*/ +/* Generic data types */ +/*===========================================================================*/ + +/* If your platform does not have the header, you may + need to edit the typedefs in the #else section below. + Use #if...#else...#endif with predefined compiler macros if possible. */ +#if U_HAVE_STDINT_H + +/* + * We mostly need (which defines the standard integer types) but not . + * includes and adds the printf/scanf helpers PRId32, SCNx16 etc. + * which we almost never use, plus stuff like imaxabs() which we never use. + */ +#include + +#if U_PLATFORM == U_PF_OS390 +/* The features header is needed to get (u)int64_t sometimes. */ +#include +/* z/OS has , but some versions are missing uint8_t (APAR PK62248). */ +#if !defined(__uint8_t) +#define __uint8_t 1 +typedef unsigned char uint8_t; +#endif +#endif /* U_PLATFORM == U_PF_OS390 */ + +#elif U_HAVE_INTTYPES_H + +# include + +#else /* neither U_HAVE_STDINT_H nor U_HAVE_INTTYPES_H */ + +/// \cond +#if ! U_HAVE_INT8_T +typedef signed char int8_t; +#endif + +#if ! U_HAVE_UINT8_T +typedef unsigned char uint8_t; +#endif + +#if ! U_HAVE_INT16_T +typedef signed short int16_t; +#endif + +#if ! U_HAVE_UINT16_T +typedef unsigned short uint16_t; +#endif + +#if ! U_HAVE_INT32_T +typedef signed int int32_t; +#endif + +#if ! U_HAVE_UINT32_T +typedef unsigned int uint32_t; +#endif + +#if ! U_HAVE_INT64_T +#ifdef _MSC_VER + typedef signed __int64 int64_t; +#else + typedef signed long long int64_t; +#endif +#endif + +#if ! U_HAVE_UINT64_T +#ifdef _MSC_VER + typedef unsigned __int64 uint64_t; +#else + typedef unsigned long long uint64_t; +#endif +#endif +/// \endcond + +#endif /* U_HAVE_STDINT_H / U_HAVE_INTTYPES_H */ + +#endif /* _PTYPES_H */ diff --git a/intl/icu/source/common/unicode/putil.h b/intl/icu/source/common/unicode/putil.h new file mode 100644 index 0000000000..500c21252f --- /dev/null +++ b/intl/icu/source/common/unicode/putil.h @@ -0,0 +1,183 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1997-2014, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* FILE NAME : putil.h +* +* Date Name Description +* 05/14/98 nos Creation (content moved here from utypes.h). +* 06/17/99 erm Added IEEE_754 +* 07/22/98 stephen Added IEEEremainder, max, min, trunc +* 08/13/98 stephen Added isNegativeInfinity, isPositiveInfinity +* 08/24/98 stephen Added longBitsFromDouble +* 03/02/99 stephen Removed openFile(). Added AS400 support. +* 04/15/99 stephen Converted to C +* 11/15/99 helena Integrated S/390 changes for IEEE support. +* 01/11/00 helena Added u_getVersion. +****************************************************************************** +*/ + +#ifndef PUTIL_H +#define PUTIL_H + +#include "unicode/utypes.h" + /** + * \file + * \brief C API: Platform Utilities + */ + +/*==========================================================================*/ +/* Platform utilities */ +/*==========================================================================*/ + +/** + * Platform utilities isolates the platform dependencies of the + * library. For each platform which this code is ported to, these + * functions may have to be re-implemented. + */ + +/** + * Return the ICU data directory. + * The data directory is where common format ICU data files (.dat files) + * are loaded from. Note that normal use of the built-in ICU + * facilities does not require loading of an external data file; + * unless you are adding custom data to ICU, the data directory + * does not need to be set. + * + * The data directory is determined as follows: + * If u_setDataDirectory() has been called, that is it, otherwise + * if the ICU_DATA environment variable is set, use that, otherwise + * If a data directory was specified at ICU build time + * + * \code + * #define ICU_DATA_DIR "path" + * \endcode + * use that, + * otherwise no data directory is available. + * + * @return the data directory, or an empty string ("") if no data directory has + * been specified. + * + * @stable ICU 2.0 + */ +U_CAPI const char* U_EXPORT2 u_getDataDirectory(void); + + +/** + * Set the ICU data directory. + * The data directory is where common format ICU data files (.dat files) + * are loaded from. Note that normal use of the built-in ICU + * facilities does not require loading of an external data file; + * unless you are adding custom data to ICU, the data directory + * does not need to be set. + * + * This function should be called at most once in a process, before the + * first ICU operation (e.g., u_init()) that will require the loading of an + * ICU data file. + * This function is not thread-safe. Use it before calling ICU APIs from + * multiple threads. + * + * @param directory The directory to be set. + * + * @see u_init + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 u_setDataDirectory(const char *directory); + +#ifndef U_HIDE_INTERNAL_API +/** + * Return the time zone files override directory, or an empty string if + * no directory was specified. Certain time zone resources will be preferentially + * loaded from individual files in this directory. + * + * @return the time zone data override directory. + * @internal + */ +U_CAPI const char * U_EXPORT2 u_getTimeZoneFilesDirectory(UErrorCode *status); + +/** + * Set the time zone files override directory. + * This function is not thread safe; it must not be called concurrently with + * u_getTimeZoneFilesDirectory() or any other use of ICU time zone functions. + * This function should only be called before using any ICU service that + * will access the time zone data. + * @internal + */ +U_CAPI void U_EXPORT2 u_setTimeZoneFilesDirectory(const char *path, UErrorCode *status); +#endif /* U_HIDE_INTERNAL_API */ + + +/** + * @{ + * Filesystem file and path separator characters. + * Example: '/' and ':' on Unix, '\\' and ';' on Windows. + * @stable ICU 2.0 + */ +#if U_PLATFORM_USES_ONLY_WIN32_API +# define U_FILE_SEP_CHAR '\\' +# define U_FILE_ALT_SEP_CHAR '/' +# define U_PATH_SEP_CHAR ';' +# define U_FILE_SEP_STRING "\\" +# define U_FILE_ALT_SEP_STRING "/" +# define U_PATH_SEP_STRING ";" +#else +# define U_FILE_SEP_CHAR '/' +# define U_FILE_ALT_SEP_CHAR '/' +# define U_PATH_SEP_CHAR ':' +# define U_FILE_SEP_STRING "/" +# define U_FILE_ALT_SEP_STRING "/" +# define U_PATH_SEP_STRING ":" +#endif + +/** @} */ + +/** + * Convert char characters to UChar characters. + * This utility function is useful only for "invariant characters" + * that are encoded in the platform default encoding. + * They are a small, constant subset of the encoding and include + * just the latin letters, digits, and some punctuation. + * For details, see U_CHARSET_FAMILY. + * + * @param cs Input string, points to length + * character bytes from a subset of the platform encoding. + * @param us Output string, points to memory for length + * Unicode characters. + * @param length The number of characters to convert; this may + * include the terminating NUL. + * + * @see U_CHARSET_FAMILY + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +u_charsToUChars(const char *cs, UChar *us, int32_t length); + +/** + * Convert UChar characters to char characters. + * This utility function is useful only for "invariant characters" + * that can be encoded in the platform default encoding. + * They are a small, constant subset of the encoding and include + * just the latin letters, digits, and some punctuation. + * For details, see U_CHARSET_FAMILY. + * + * @param us Input string, points to length + * Unicode characters that can be encoded with the + * codepage-invariant subset of the platform encoding. + * @param cs Output string, points to memory for length + * character bytes. + * @param length The number of characters to convert; this may + * include the terminating NUL. + * + * @see U_CHARSET_FAMILY + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +u_UCharsToChars(const UChar *us, char *cs, int32_t length); + +#endif diff --git a/intl/icu/source/common/unicode/rbbi.h b/intl/icu/source/common/unicode/rbbi.h new file mode 100644 index 0000000000..418b52e41f --- /dev/null +++ b/intl/icu/source/common/unicode/rbbi.h @@ -0,0 +1,745 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +*************************************************************************** +* Copyright (C) 1999-2016 International Business Machines Corporation * +* and others. All rights reserved. * +*************************************************************************** + +********************************************************************** +* Date Name Description +* 10/22/99 alan Creation. +* 11/11/99 rgillam Complete port from Java. +********************************************************************** +*/ + +#ifndef RBBI_H +#define RBBI_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +/** + * \file + * \brief C++ API: Rule Based Break Iterator + */ + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/brkiter.h" +#include "unicode/udata.h" +#include "unicode/parseerr.h" +#include "unicode/schriter.h" + +struct UCPTrie; + +U_NAMESPACE_BEGIN + +/** @internal */ +class LanguageBreakEngine; +struct RBBIDataHeader; +class RBBIDataWrapper; +class UnhandledEngine; +class UStack; + +/** + * + * A subclass of BreakIterator whose behavior is specified using a list of rules. + *

Instances of this class are most commonly created by the factory methods of + * BreakIterator::createWordInstance(), BreakIterator::createLineInstance(), etc., + * and then used via the abstract API in class BreakIterator

+ * + *

See the ICU User Guide for information on Break Iterator Rules.

+ * + *

This class is not intended to be subclassed.

+ */ +class U_COMMON_API RuleBasedBreakIterator /*final*/ : public BreakIterator { + +private: + /** + * The UText through which this BreakIterator accesses the text + * @internal (private) + */ + UText fText = UTEXT_INITIALIZER; + +#ifndef U_HIDE_INTERNAL_API +public: +#endif /* U_HIDE_INTERNAL_API */ + /** + * The rule data for this BreakIterator instance. + * Not for general use; Public only for testing purposes. + * @internal + */ + RBBIDataWrapper *fData = nullptr; + +private: + /** + * The saved error code associated with this break iterator. + * This is the value to be returned by copyErrorTo(). + */ + UErrorCode fErrorCode = U_ZERO_ERROR; + + /** + * The current position of the iterator. Pinned, 0 < fPosition <= text.length. + * Never has the value UBRK_DONE (-1). + */ + int32_t fPosition = 0; + + /** + * TODO: + */ + int32_t fRuleStatusIndex = 0; + + /** + * Cache of previously determined boundary positions. + */ + class BreakCache; + BreakCache *fBreakCache = nullptr; + + /** + * Cache of boundary positions within a region of text that has been + * sub-divided by dictionary based breaking. + */ + class DictionaryCache; + DictionaryCache *fDictionaryCache = nullptr; + + /** + * + * If present, UStack of LanguageBreakEngine objects that might handle + * dictionary characters. Searched from top to bottom to find an object to + * handle a given character. + * @internal (private) + */ + UStack *fLanguageBreakEngines = nullptr; + + /** + * + * If present, the special LanguageBreakEngine used for handling + * characters that are in the dictionary set, but not handled by any + * LanguageBreakEngine. + * @internal (private) + */ + UnhandledEngine *fUnhandledBreakEngine = nullptr; + + /** + * Counter for the number of characters encountered with the "dictionary" + * flag set. + * @internal (private) + */ + uint32_t fDictionaryCharCount = 0; + + /** + * A character iterator that refers to the same text as the UText, above. + * Only included for compatibility with old API, which was based on CharacterIterators. + * Value may be adopted from outside, or one of fSCharIter or fDCharIter, below. + */ + CharacterIterator *fCharIter = &fSCharIter; + + /** + * When the input text is provided by a UnicodeString, this will point to + * a characterIterator that wraps that data. Needed only for the + * implementation of getText(), a backwards compatibility issue. + */ + UCharCharacterIterator fSCharIter {u"", 0}; + + /** + * True when iteration has run off the end, and iterator functions should return UBRK_DONE. + */ + bool fDone = false; + + /** + * Array of look-ahead tentative results. + */ + int32_t *fLookAheadMatches = nullptr; + + /** + * A flag to indicate if phrase based breaking is enabled. + */ + UBool fIsPhraseBreaking = false; + + //======================================================================= + // constructors + //======================================================================= + + /** + * Constructor from a flattened set of RBBI data in malloced memory. + * RulesBasedBreakIterators built from a custom set of rules + * are created via this constructor; the rules are compiled + * into memory, then the break iterator is constructed here. + * + * The break iterator adopts the memory, and will + * free it when done. + * @internal (private) + */ + RuleBasedBreakIterator(RBBIDataHeader* data, UErrorCode &status); + + /** + * This constructor uses the udata interface to create a BreakIterator + * whose internal tables live in a memory-mapped file. "image" is an + * ICU UDataMemory handle for the pre-compiled break iterator tables. + * @param image handle to the memory image for the break iterator data. + * Ownership of the UDataMemory handle passes to the Break Iterator, + * which will be responsible for closing it when it is no longer needed. + * @param status Information on any errors encountered. + * @param isPhraseBreaking true if phrase based breaking is required, otherwise false. + * @see udata_open + * @see #getBinaryRules + * @internal (private) + */ + RuleBasedBreakIterator(UDataMemory* image, UBool isPhraseBreaking, UErrorCode &status); + + /** @internal */ + friend class RBBIRuleBuilder; + /** @internal */ + friend class BreakIterator; + + /** + * Default constructor with an error code parameter. + * Aside from error handling, otherwise identical to the default constructor. + * Internally, handles common initialization for other constructors. + * @internal (private) + */ + RuleBasedBreakIterator(UErrorCode *status); + +public: + + /** Default constructor. Creates an empty shell of an iterator, with no + * rules or text to iterate over. Object can subsequently be assigned to, + * but is otherwise unusable. + * @stable ICU 2.2 + */ + RuleBasedBreakIterator(); + + /** + * Copy constructor. Will produce a break iterator with the same behavior, + * and which iterates over the same text, as the one passed in. + * @param that The RuleBasedBreakIterator passed to be copied + * @stable ICU 2.0 + */ + RuleBasedBreakIterator(const RuleBasedBreakIterator& that); + + /** + * Construct a RuleBasedBreakIterator from a set of rules supplied as a string. + * @param rules The break rules to be used. + * @param parseError In the event of a syntax error in the rules, provides the location + * within the rules of the problem. + * @param status Information on any errors encountered. + * @stable ICU 2.2 + */ + RuleBasedBreakIterator( const UnicodeString &rules, + UParseError &parseError, + UErrorCode &status); + + /** + * Construct a RuleBasedBreakIterator from a set of precompiled binary rules. + * Binary rules are obtained from RulesBasedBreakIterator::getBinaryRules(). + * Construction of a break iterator in this way is substantially faster than + * construction from source rules. + * + * Ownership of the storage containing the compiled rules remains with the + * caller of this function. The compiled rules must not be modified or + * deleted during the life of the break iterator. + * + * The compiled rules are not compatible across different major versions of ICU. + * The compiled rules are compatible only between machines with the same + * byte ordering (little or big endian) and the same base character set family + * (ASCII or EBCDIC). + * + * @see #getBinaryRules + * @param compiledRules A pointer to the compiled break rules to be used. + * @param ruleLength The length of the compiled break rules, in bytes. This + * corresponds to the length value produced by getBinaryRules(). + * @param status Information on any errors encountered, including invalid + * binary rules. + * @stable ICU 4.8 + */ + RuleBasedBreakIterator(const uint8_t *compiledRules, + uint32_t ruleLength, + UErrorCode &status); + + /** + * This constructor uses the udata interface to create a BreakIterator + * whose internal tables live in a memory-mapped file. "image" is an + * ICU UDataMemory handle for the pre-compiled break iterator tables. + * @param image handle to the memory image for the break iterator data. + * Ownership of the UDataMemory handle passes to the Break Iterator, + * which will be responsible for closing it when it is no longer needed. + * @param status Information on any errors encountered. + * @see udata_open + * @see #getBinaryRules + * @stable ICU 2.8 + */ + RuleBasedBreakIterator(UDataMemory* image, UErrorCode &status); + + /** + * Destructor + * @stable ICU 2.0 + */ + virtual ~RuleBasedBreakIterator(); + + /** + * Assignment operator. Sets this iterator to have the same behavior, + * and iterate over the same text, as the one passed in. + * @param that The RuleBasedBreakItertor passed in + * @return the newly created RuleBasedBreakIterator + * @stable ICU 2.0 + */ + RuleBasedBreakIterator& operator=(const RuleBasedBreakIterator& that); + + /** + * Equality operator. Returns true if both BreakIterators are of the + * same class, have the same behavior, and iterate over the same text. + * @param that The BreakIterator to be compared for equality + * @return true if both BreakIterators are of the + * same class, have the same behavior, and iterate over the same text. + * @stable ICU 2.0 + */ + virtual bool operator==(const BreakIterator& that) const override; + + /** + * Not-equal operator. If operator== returns true, this returns false, + * and vice versa. + * @param that The BreakIterator to be compared for inequality + * @return true if both BreakIterators are not same. + * @stable ICU 2.0 + */ + inline bool operator!=(const BreakIterator& that) const { + return !operator==(that); + } + + /** + * Returns a newly-constructed RuleBasedBreakIterator with the same + * behavior, and iterating over the same text, as this one. + * Differs from the copy constructor in that it is polymorphic, and + * will correctly clone (copy) a derived class. + * clone() is thread safe. Multiple threads may simultaneously + * clone the same source break iterator. + * @return a newly-constructed RuleBasedBreakIterator + * @stable ICU 2.0 + */ + virtual RuleBasedBreakIterator* clone() const override; + + /** + * Compute a hash code for this BreakIterator + * @return A hash code + * @stable ICU 2.0 + */ + virtual int32_t hashCode(void) const; + + /** + * Returns the description used to create this iterator + * @return the description used to create this iterator + * @stable ICU 2.0 + */ + virtual const UnicodeString& getRules(void) const; + + //======================================================================= + // BreakIterator overrides + //======================================================================= + + /** + *

+ * Return a CharacterIterator over the text being analyzed. + * The returned character iterator is owned by the break iterator, and must + * not be deleted by the caller. Repeated calls to this function may + * return the same CharacterIterator. + *

+ *

+ * The returned character iterator must not be used concurrently with + * the break iterator. If concurrent operation is needed, clone the + * returned character iterator first and operate on the clone. + *

+ *

+ * When the break iterator is operating on text supplied via a UText, + * this function will fail, returning a CharacterIterator containing no text. + * The function getUText() provides similar functionality, + * is reliable, and is more efficient. + *

+ * + * TODO: deprecate this function? + * + * @return An iterator over the text being analyzed. + * @stable ICU 2.0 + */ + virtual CharacterIterator& getText(void) const override; + + + /** + * Get a UText for the text being analyzed. + * The returned UText is a shallow clone of the UText used internally + * by the break iterator implementation. It can safely be used to + * access the text without impacting any break iterator operations, + * but the underlying text itself must not be altered. + * + * @param fillIn A UText to be filled in. If nullptr, a new UText will be + * allocated to hold the result. + * @param status receives any error codes. + * @return The current UText for this break iterator. If an input + * UText was provided, it will always be returned. + * @stable ICU 3.4 + */ + virtual UText *getUText(UText *fillIn, UErrorCode &status) const override; + + /** + * Set the iterator to analyze a new piece of text. This function resets + * the current iteration position to the beginning of the text. + * @param newText An iterator over the text to analyze. The BreakIterator + * takes ownership of the character iterator. The caller MUST NOT delete it! + * @stable ICU 2.0 + */ + virtual void adoptText(CharacterIterator* newText) override; + + /** + * Set the iterator to analyze a new piece of text. This function resets + * the current iteration position to the beginning of the text. + * + * The BreakIterator will retain a reference to the supplied string. + * The caller must not modify or delete the text while the BreakIterator + * retains the reference. + * + * @param newText The text to analyze. + * @stable ICU 2.0 + */ + virtual void setText(const UnicodeString& newText) override; + + /** + * Reset the break iterator to operate over the text represented by + * the UText. The iterator position is reset to the start. + * + * This function makes a shallow clone of the supplied UText. This means + * that the caller is free to immediately close or otherwise reuse the + * Utext that was passed as a parameter, but that the underlying text itself + * must not be altered while being referenced by the break iterator. + * + * @param text The UText used to change the text. + * @param status Receives any error codes. + * @stable ICU 3.4 + */ + virtual void setText(UText *text, UErrorCode &status) override; + + /** + * Sets the current iteration position to the beginning of the text, position zero. + * @return The offset of the beginning of the text, zero. + * @stable ICU 2.0 + */ + virtual int32_t first(void) override; + + /** + * Sets the current iteration position to the end of the text. + * @return The text's past-the-end offset. + * @stable ICU 2.0 + */ + virtual int32_t last(void) override; + + /** + * Advances the iterator either forward or backward the specified number of steps. + * Negative values move backward, and positive values move forward. This is + * equivalent to repeatedly calling next() or previous(). + * @param n The number of steps to move. The sign indicates the direction + * (negative is backwards, and positive is forwards). + * @return The character offset of the boundary position n boundaries away from + * the current one. + * @stable ICU 2.0 + */ + virtual int32_t next(int32_t n) override; + + /** + * Advances the iterator to the next boundary position. + * @return The position of the first boundary after this one. + * @stable ICU 2.0 + */ + virtual int32_t next(void) override; + + /** + * Moves the iterator backwards, to the last boundary preceding this one. + * @return The position of the last boundary position preceding this one. + * @stable ICU 2.0 + */ + virtual int32_t previous(void) override; + + /** + * Sets the iterator to refer to the first boundary position following + * the specified position. + * @param offset The position from which to begin searching for a break position. + * @return The position of the first break after the current position. + * @stable ICU 2.0 + */ + virtual int32_t following(int32_t offset) override; + + /** + * Sets the iterator to refer to the last boundary position before the + * specified position. + * @param offset The position to begin searching for a break from. + * @return The position of the last boundary before the starting position. + * @stable ICU 2.0 + */ + virtual int32_t preceding(int32_t offset) override; + + /** + * Returns true if the specified position is a boundary position. As a side + * effect, leaves the iterator pointing to the first boundary position at + * or after "offset". + * @param offset the offset to check. + * @return True if "offset" is a boundary position. + * @stable ICU 2.0 + */ + virtual UBool isBoundary(int32_t offset) override; + + /** + * Returns the current iteration position. Note that UBRK_DONE is never + * returned from this function; if iteration has run to the end of a + * string, current() will return the length of the string while + * next() will return UBRK_DONE). + * @return The current iteration position. + * @stable ICU 2.0 + */ + virtual int32_t current(void) const override; + + + /** + * Return the status tag from the break rule that determined the boundary at + * the current iteration position. For break rules that do not specify a + * status, a default value of 0 is returned. If more than one break rule + * would cause a boundary to be located at some position in the text, + * the numerically largest of the applicable status values is returned. + *

+ * Of the standard types of ICU break iterators, only word break and + * line break provide status values. The values are defined in + * the header file ubrk.h. For Word breaks, the status allows distinguishing between words + * that contain alphabetic letters, "words" that appear to be numbers, + * punctuation and spaces, words containing ideographic characters, and + * more. For Line Break, the status distinguishes between hard (mandatory) breaks + * and soft (potential) break positions. + *

+ * getRuleStatus() can be called after obtaining a boundary + * position from next(), previous(), or + * any other break iterator functions that returns a boundary position. + *

+ * Note that getRuleStatus() returns the value corresponding to + * current() index even after next() has returned DONE. + *

+ * When creating custom break rules, one is free to define whatever + * status values may be convenient for the application. + *

+ * @return the status from the break rule that determined the boundary + * at the current iteration position. + * + * @see UWordBreak + * @stable ICU 2.2 + */ + virtual int32_t getRuleStatus() const override; + + /** + * Get the status (tag) values from the break rule(s) that determined the boundary + * at the current iteration position. + *

+ * The returned status value(s) are stored into an array provided by the caller. + * The values are stored in sorted (ascending) order. + * If the capacity of the output array is insufficient to hold the data, + * the output will be truncated to the available length, and a + * U_BUFFER_OVERFLOW_ERROR will be signaled. + * + * @param fillInVec an array to be filled in with the status values. + * @param capacity the length of the supplied vector. A length of zero causes + * the function to return the number of status values, in the + * normal way, without attempting to store any values. + * @param status receives error codes. + * @return The number of rule status values from the rules that determined + * the boundary at the current iteration position. + * In the event of a U_BUFFER_OVERFLOW_ERROR, the return value + * is the total number of status values that were available, + * not the reduced number that were actually returned. + * @see getRuleStatus + * @stable ICU 3.0 + */ + virtual int32_t getRuleStatusVec(int32_t *fillInVec, int32_t capacity, UErrorCode &status) override; + + /** + * Returns a unique class ID POLYMORPHICALLY. Pure virtual override. + * This method is to implement a simple version of RTTI, since not all + * C++ compilers support genuine RTTI. Polymorphic operator==() and + * clone() methods call this method. + * + * @return The class ID for this object. All objects of a + * given class have the same class ID. Objects of + * other classes have different class IDs. + * @stable ICU 2.0 + */ + virtual UClassID getDynamicClassID(void) const override; + + /** + * Returns the class ID for this class. This is useful only for + * comparing to a return value from getDynamicClassID(). For example: + * + * Base* polymorphic_pointer = createPolymorphicObject(); + * if (polymorphic_pointer->getDynamicClassID() == + * Derived::getStaticClassID()) ... + * + * @return The class ID for all objects of this class. + * @stable ICU 2.0 + */ + static UClassID U_EXPORT2 getStaticClassID(void); + +#ifndef U_FORCE_HIDE_DEPRECATED_API + /** + * Deprecated functionality. Use clone() instead. + * + * Create a clone (copy) of this break iterator in memory provided + * by the caller. The idea is to increase performance by avoiding + * a storage allocation. Use of this function is NOT RECOMMENDED. + * Performance gains are minimal, and correct buffer management is + * tricky. Use clone() instead. + * + * @param stackBuffer The pointer to the memory into which the cloned object + * should be placed. If nullptr, allocate heap memory + * for the cloned object. + * @param BufferSize The size of the buffer. If zero, return the required + * buffer size, but do not clone the object. If the + * size was too small (but not zero), allocate heap + * storage for the cloned object. + * + * @param status Error status. U_SAFECLONE_ALLOCATED_WARNING will be + * returned if the provided buffer was too small, and + * the clone was therefore put on the heap. + * + * @return Pointer to the clone object. This may differ from the stackBuffer + * address if the byte alignment of the stack buffer was not suitable + * or if the stackBuffer was too small to hold the clone. + * @deprecated ICU 52. Use clone() instead. + */ + virtual RuleBasedBreakIterator *createBufferClone(void *stackBuffer, + int32_t &BufferSize, + UErrorCode &status) override; +#endif // U_FORCE_HIDE_DEPRECATED_API + + /** + * Return the binary form of compiled break rules, + * which can then be used to create a new break iterator at some + * time in the future. Creating a break iterator from pre-compiled rules + * is much faster than building one from the source form of the + * break rules. + * + * The binary data can only be used with the same version of ICU + * and on the same platform type (processor endian-ness) + * + * @param length Returns the length of the binary data. (Out parameter.) + * + * @return A pointer to the binary (compiled) rule data. The storage + * belongs to the RulesBasedBreakIterator object, not the + * caller, and must not be modified or deleted. + * @stable ICU 4.8 + */ + virtual const uint8_t *getBinaryRules(uint32_t &length); + + /** + * Set the subject text string upon which the break iterator is operating + * without changing any other aspect of the matching state. + * The new and previous text strings must have the same content. + * + * This function is intended for use in environments where ICU is operating on + * strings that may move around in memory. It provides a mechanism for notifying + * ICU that the string has been relocated, and providing a new UText to access the + * string in its new position. + * + * Note that the break iterator implementation never copies the underlying text + * of a string being processed, but always operates directly on the original text + * provided by the user. Refreshing simply drops the references to the old text + * and replaces them with references to the new. + * + * Caution: this function is normally used only by very specialized, + * system-level code. One example use case is with garbage collection that moves + * the text in memory. + * + * @param input The new (moved) text string. + * @param status Receives errors detected by this function. + * @return *this + * + * @stable ICU 49 + */ + virtual RuleBasedBreakIterator &refreshInputText(UText *input, UErrorCode &status) override; + + +private: + //======================================================================= + // implementation + //======================================================================= + /** + * Iterate backwards from an arbitrary position in the input text using the + * synthesized Safe Reverse rules. + * This locates a "Safe Position" from which the forward break rules + * will operate correctly. A Safe Position is not necessarily a boundary itself. + * + * @param fromPosition the position in the input text to begin the iteration. + * @internal (private) + */ + int32_t handleSafePrevious(int32_t fromPosition); + + /** + * Find a rule-based boundary by running the state machine. + * Input + * fPosition, the position in the text to begin from. + * Output + * fPosition: the boundary following the starting position. + * fDictionaryCharCount the number of dictionary characters encountered. + * If > 0, the segment will be further subdivided + * fRuleStatusIndex Info from the state table indicating which rules caused the boundary. + * + * @internal (private) + */ + int32_t handleNext(); + + /* + * Templatized version of handleNext() and handleSafePrevious(). + * + * There will be exactly four instantiations, two each for 8 and 16 bit tables, + * two each for 8 and 16 bit trie. + * Having separate instantiations for the table types keeps conditional tests of + * the table type out of the inner loops, at the expense of replicated code. + * + * The template parameter for the Trie access function is a value, not a type. + * Doing it this way, the compiler will inline the Trie function in the + * expanded functions. (Both the 8 and 16 bit access functions have the same type + * signature) + */ + + typedef uint16_t (*PTrieFunc)(const UCPTrie *, UChar32); + + template + int32_t handleSafePrevious(int32_t fromPosition); + + template + int32_t handleNext(); + + + /** + * This function returns the appropriate LanguageBreakEngine for a + * given character c. + * @param c A character in the dictionary set + * @internal (private) + */ + const LanguageBreakEngine *getLanguageBreakEngine(UChar32 c); + + public: +#ifndef U_HIDE_INTERNAL_API + /** + * Debugging function only. + * @internal + */ + void dumpCache(); + + /** + * Debugging function only. + * @internal + */ + void dumpTables(); +#endif /* U_HIDE_INTERNAL_API */ +}; + +U_NAMESPACE_END + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/rep.h b/intl/icu/source/common/unicode/rep.h new file mode 100644 index 0000000000..7115c97b82 --- /dev/null +++ b/intl/icu/source/common/unicode/rep.h @@ -0,0 +1,266 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +************************************************************************** +* Copyright (C) 1999-2012, International Business Machines Corporation and +* others. All Rights Reserved. +************************************************************************** +* Date Name Description +* 11/17/99 aliu Creation. Ported from java. Modified to +* match current UnicodeString API. Forced +* to use name "handleReplaceBetween" because +* of existing methods in UnicodeString. +************************************************************************** +*/ + +#ifndef REP_H +#define REP_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/uobject.h" + +/** + * \file + * \brief C++ API: Replaceable String + */ + +U_NAMESPACE_BEGIN + +class UnicodeString; + +/** + * Replaceable is an abstract base class representing a + * string of characters that supports the replacement of a range of + * itself with a new string of characters. It is used by APIs that + * change a piece of text while retaining metadata. Metadata is data + * other than the Unicode characters returned by char32At(). One + * example of metadata is style attributes; another is an edit + * history, marking each character with an author and revision number. + * + *

An implicit aspect of the Replaceable API is that + * during a replace operation, new characters take on the metadata of + * the old characters. For example, if the string "the bold + * font" has range (4, 8) replaced with "strong", then it becomes "the + * strong font". + * + *

Replaceable specifies ranges using a start + * offset and a limit offset. The range of characters thus specified + * includes the characters at offset start..limit-1. That is, the + * start offset is inclusive, and the limit offset is exclusive. + * + *

Replaceable also includes API to access characters + * in the string: length(), charAt(), + * char32At(), and extractBetween(). + * + *

For a subclass to support metadata, typical behavior of + * replace() is the following: + *

    + *
  • Set the metadata of the new text to the metadata of the first + * character replaced
    • + *
  • If no characters are replaced, use the metadata of the + * previous character
    • + *
  • If there is no previous character (i.e. start == 0), use the + * following character
    • + *
  • If there is no following character (i.e. the replaceable was + * empty), use default metadata.
    + *
  • If the code point U+FFFF is seen, it should be interpreted as + * a special marker having no metadata
  • + *
    • + *
+ * If this is not the behavior, the subclass should document any differences. + * @author Alan Liu + * @stable ICU 2.0 + */ +class U_COMMON_API Replaceable : public UObject { + +public: + /** + * Destructor. + * @stable ICU 2.0 + */ + virtual ~Replaceable(); + + /** + * Returns the number of 16-bit code units in the text. + * @return number of 16-bit code units in text + * @stable ICU 1.8 + */ + inline int32_t length() const; + + /** + * Returns the 16-bit code unit at the given offset into the text. + * @param offset an integer between 0 and length()-1 + * inclusive + * @return 16-bit code unit of text at given offset + * @stable ICU 1.8 + */ + inline char16_t charAt(int32_t offset) const; + + /** + * Returns the 32-bit code point at the given 16-bit offset into + * the text. This assumes the text is stored as 16-bit code units + * with surrogate pairs intermixed. If the offset of a leading or + * trailing code unit of a surrogate pair is given, return the + * code point of the surrogate pair. + * + * @param offset an integer between 0 and length()-1 + * inclusive + * @return 32-bit code point of text at given offset + * @stable ICU 1.8 + */ + inline UChar32 char32At(int32_t offset) const; + + /** + * Copies characters in the range [start, limit) + * into the UnicodeString target. + * @param start offset of first character which will be copied + * @param limit offset immediately following the last character to + * be copied + * @param target UnicodeString into which to copy characters. + * @return A reference to target + * @stable ICU 2.1 + */ + virtual void extractBetween(int32_t start, + int32_t limit, + UnicodeString& target) const = 0; + + /** + * Replaces a substring of this object with the given text. If the + * characters being replaced have metadata, the new characters + * that replace them should be given the same metadata. + * + *

Subclasses must ensure that if the text between start and + * limit is equal to the replacement text, that replace has no + * effect. That is, any metadata + * should be unaffected. In addition, subclasses are encouraged to + * check for initial and trailing identical characters, and make a + * smaller replacement if possible. This will preserve as much + * metadata as possible. + * @param start the beginning index, inclusive; 0 <= start + * <= limit. + * @param limit the ending index, exclusive; start <= limit + * <= length(). + * @param text the text to replace characters start + * to limit - 1 + * @stable ICU 2.0 + */ + virtual void handleReplaceBetween(int32_t start, + int32_t limit, + const UnicodeString& text) = 0; + // Note: All other methods in this class take the names of + // existing UnicodeString methods. This method is the exception. + // It is named differently because all replace methods of + // UnicodeString return a UnicodeString&. The 'between' is + // required in order to conform to the UnicodeString naming + // convention; API taking start/length are named , and + // those taking start/limit are named . The + // 'handle' is added because 'replaceBetween' and + // 'doReplaceBetween' are already taken. + + /** + * Copies a substring of this object, retaining metadata. + * This method is used to duplicate or reorder substrings. + * The destination index must not overlap the source range. + * + * @param start the beginning index, inclusive; 0 <= start <= + * limit. + * @param limit the ending index, exclusive; start <= limit <= + * length(). + * @param dest the destination index. The characters from + * start..limit-1 will be copied to dest. + * Implementations of this method may assume that dest <= start || + * dest >= limit. + * @stable ICU 2.0 + */ + virtual void copy(int32_t start, int32_t limit, int32_t dest) = 0; + + /** + * Returns true if this object contains metadata. If a + * Replaceable object has metadata, calls to the Replaceable API + * must be made so as to preserve metadata. If it does not, calls + * to the Replaceable API may be optimized to improve performance. + * The default implementation returns true. + * @return true if this object contains metadata + * @stable ICU 2.2 + */ + virtual UBool hasMetaData() const; + + /** + * Clone this object, an instance of a subclass of Replaceable. + * Clones can be used concurrently in multiple threads. + * If a subclass does not implement clone(), or if an error occurs, + * then nullptr is returned. + * The caller must delete the clone. + * + * @return a clone of this object + * + * @see getDynamicClassID + * @stable ICU 2.6 + */ + virtual Replaceable *clone() const; + +protected: + + /** + * Default constructor. + * @stable ICU 2.4 + */ + inline Replaceable(); + + /* + * Assignment operator not declared. The compiler will provide one + * which does nothing since this class does not contain any data members. + * API/code coverage may show the assignment operator as present and + * untested - ignore. + * Subclasses need this assignment operator if they use compiler-provided + * assignment operators of their own. An alternative to not declaring one + * here would be to declare and empty-implement a protected or public one. + Replaceable &Replaceable::operator=(const Replaceable &); + */ + + /** + * Virtual version of length(). + * @stable ICU 2.4 + */ + virtual int32_t getLength() const = 0; + + /** + * Virtual version of charAt(). + * @stable ICU 2.4 + */ + virtual char16_t getCharAt(int32_t offset) const = 0; + + /** + * Virtual version of char32At(). + * @stable ICU 2.4 + */ + virtual UChar32 getChar32At(int32_t offset) const = 0; +}; + +inline Replaceable::Replaceable() {} + +inline int32_t +Replaceable::length() const { + return getLength(); +} + +inline char16_t +Replaceable::charAt(int32_t offset) const { + return getCharAt(offset); +} + +inline UChar32 +Replaceable::char32At(int32_t offset) const { + return getChar32At(offset); +} + +// There is no rep.cpp, see unistr.cpp for Replaceable function implementations. + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/resbund.h b/intl/icu/source/common/unicode/resbund.h new file mode 100644 index 0000000000..30fc2ac0ab --- /dev/null +++ b/intl/icu/source/common/unicode/resbund.h @@ -0,0 +1,498 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1996-2013, International Business Machines Corporation +* and others. All Rights Reserved. +* +****************************************************************************** +* +* File resbund.h +* +* CREATED BY +* Richard Gillam +* +* Modification History: +* +* Date Name Description +* 2/5/97 aliu Added scanForLocaleInFile. Added +* constructor which attempts to read resource bundle +* from a specific file, without searching other files. +* 2/11/97 aliu Added UErrorCode return values to constructors. Fixed +* infinite loops in scanForFile and scanForLocale. +* Modified getRawResourceData to not delete storage +* in localeData and resourceData which it doesn't own. +* Added Mac compatibility #ifdefs for tellp() and +* ios::nocreate. +* 2/18/97 helena Updated with 100% documentation coverage. +* 3/13/97 aliu Rewrote to load in entire resource bundle and store +* it as a Hashtable of ResourceBundleData objects. +* Added state table to govern parsing of files. +* Modified to load locale index out of new file +* distinct from default.txt. +* 3/25/97 aliu Modified to support 2-d arrays, needed for timezone +* data. Added support for custom file suffixes. Again, +* needed to support timezone data. +* 4/7/97 aliu Cleaned up. +* 03/02/99 stephen Removed dependency on FILE*. +* 03/29/99 helena Merged Bertrand and Stephen's changes. +* 06/11/99 stephen Removed parsing of .txt files. +* Reworked to use new binary format. +* Cleaned up. +* 06/14/99 stephen Removed methods taking a filename suffix. +* 11/09/99 weiv Added getLocale(), fRealLocale, removed fRealLocaleID +****************************************************************************** +*/ + +#ifndef RESBUND_H +#define RESBUND_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/uobject.h" +#include "unicode/ures.h" +#include "unicode/unistr.h" +#include "unicode/locid.h" + +/** + * \file + * \brief C++ API: Resource Bundle + */ + +U_NAMESPACE_BEGIN + +/** + * A class representing a collection of resource information pertaining to a given + * locale. A resource bundle provides a way of accessing locale- specific information in + * a data file. You create a resource bundle that manages the resources for a given + * locale and then ask it for individual resources. + *

+ * Resource bundles in ICU4C are currently defined using text files which conform to the following + * BNF definition. + * More on resource bundle concepts and syntax can be found in the + * Users Guide. + *

+ * + * The ResourceBundle class is not suitable for subclassing. + * + * @stable ICU 2.0 + */ +class U_COMMON_API ResourceBundle : public UObject { +public: + /** + * Constructor + * + * @param packageName The packageName and locale together point to an ICU udata object, + * as defined by udata_open( packageName, "res", locale, err) + * or equivalent. Typically, packageName will refer to a (.dat) file, or to + * a package registered with udata_setAppData(). Using a full file or directory + * pathname for packageName is deprecated. + * @param locale This is the locale this resource bundle is for. To get resources + * for the French locale, for example, you would create a + * ResourceBundle passing Locale::FRENCH for the "locale" parameter, + * and all subsequent calls to that resource bundle will return + * resources that pertain to the French locale. If the caller doesn't + * pass a locale parameter, the default locale for the system (as + * returned by Locale::getDefault()) will be used. + * @param err The Error Code. + * The UErrorCode& err parameter is used to return status information to the user. To + * check whether the construction succeeded or not, you should check the value of + * U_SUCCESS(err). If you wish more detailed information, you can check for + * informational error results which still indicate success. U_USING_FALLBACK_WARNING + * indicates that a fall back locale was used. For example, 'de_CH' was requested, + * but nothing was found there, so 'de' was used. U_USING_DEFAULT_WARNING indicates that + * the default locale data was used; neither the requested locale nor any of its + * fall back locales could be found. + * @stable ICU 2.0 + */ + ResourceBundle(const UnicodeString& packageName, + const Locale& locale, + UErrorCode& err); + + /** + * Construct a resource bundle for the default bundle in the specified package. + * + * @param packageName The packageName and locale together point to an ICU udata object, + * as defined by udata_open( packageName, "res", locale, err) + * or equivalent. Typically, packageName will refer to a (.dat) file, or to + * a package registered with udata_setAppData(). Using a full file or directory + * pathname for packageName is deprecated. + * @param err A UErrorCode value + * @stable ICU 2.0 + */ + ResourceBundle(const UnicodeString& packageName, + UErrorCode& err); + + /** + * Construct a resource bundle for the ICU default bundle. + * + * @param err A UErrorCode value + * @stable ICU 2.0 + */ + ResourceBundle(UErrorCode &err); + + /** + * Standard constructor, constructs a resource bundle for the locale-specific + * bundle in the specified package. + * + * @param packageName The packageName and locale together point to an ICU udata object, + * as defined by udata_open( packageName, "res", locale, err) + * or equivalent. Typically, packageName will refer to a (.dat) file, or to + * a package registered with udata_setAppData(). Using a full file or directory + * pathname for packageName is deprecated. + * nullptr is used to refer to ICU data. + * @param locale The locale for which to open a resource bundle. + * @param err A UErrorCode value + * @stable ICU 2.0 + */ + ResourceBundle(const char* packageName, + const Locale& locale, + UErrorCode& err); + + /** + * Copy constructor. + * + * @param original The resource bundle to copy. + * @stable ICU 2.0 + */ + ResourceBundle(const ResourceBundle &original); + + /** + * Constructor from a C UResourceBundle. The resource bundle is + * copied and not adopted. ures_close will still need to be used on the + * original resource bundle. + * + * @param res A pointer to the C resource bundle. + * @param status A UErrorCode value. + * @stable ICU 2.0 + */ + ResourceBundle(UResourceBundle *res, + UErrorCode &status); + + /** + * Assignment operator. + * + * @param other The resource bundle to copy. + * @stable ICU 2.0 + */ + ResourceBundle& + operator=(const ResourceBundle& other); + + /** Destructor. + * @stable ICU 2.0 + */ + virtual ~ResourceBundle(); + + /** + * Clone this object. + * Clones can be used concurrently in multiple threads. + * If an error occurs, then nullptr is returned. + * The caller must delete the clone. + * + * @return a clone of this object + * + * @see getDynamicClassID + * @stable ICU 2.8 + */ + ResourceBundle *clone() const; + + /** + * Returns the size of a resource. Size for scalar types is always 1, and for vector/table types is + * the number of child resources. + * @warning Integer array is treated as a scalar type. There are no + * APIs to access individual members of an integer array. It + * is always returned as a whole. + * + * @return number of resources in a given resource. + * @stable ICU 2.0 + */ + int32_t + getSize(void) const; + + /** + * returns a string from a string resource type + * + * @param status fills in the outgoing error code + * could be U_MISSING_RESOURCE_ERROR if the key is not found + * could be a warning + * e.g.: U_USING_FALLBACK_WARNING,U_USING_DEFAULT_WARNING + * @return a pointer to a zero-terminated char16_t array which lives in a memory mapped/DLL file. + * @stable ICU 2.0 + */ + UnicodeString + getString(UErrorCode& status) const; + + /** + * returns a binary data from a resource. Can be used at most primitive resource types (binaries, + * strings, ints) + * + * @param len fills in the length of resulting byte chunk + * @param status fills in the outgoing error code + * could be U_MISSING_RESOURCE_ERROR if the key is not found + * could be a warning + * e.g.: U_USING_FALLBACK_WARNING,U_USING_DEFAULT_WARNING + * @return a pointer to a chunk of unsigned bytes which live in a memory mapped/DLL file. + * @stable ICU 2.0 + */ + const uint8_t* + getBinary(int32_t& len, UErrorCode& status) const; + + + /** + * returns an integer vector from a resource. + * + * @param len fills in the length of resulting integer vector + * @param status fills in the outgoing error code + * could be U_MISSING_RESOURCE_ERROR if the key is not found + * could be a warning + * e.g.: U_USING_FALLBACK_WARNING,U_USING_DEFAULT_WARNING + * @return a pointer to a vector of integers that lives in a memory mapped/DLL file. + * @stable ICU 2.0 + */ + const int32_t* + getIntVector(int32_t& len, UErrorCode& status) const; + + /** + * returns an unsigned integer from a resource. + * This integer is originally 28 bits. + * + * @param status fills in the outgoing error code + * could be U_MISSING_RESOURCE_ERROR if the key is not found + * could be a warning + * e.g.: U_USING_FALLBACK_WARNING,U_USING_DEFAULT_WARNING + * @return an unsigned integer value + * @stable ICU 2.0 + */ + uint32_t + getUInt(UErrorCode& status) const; + + /** + * returns a signed integer from a resource. + * This integer is originally 28 bit and the sign gets propagated. + * + * @param status fills in the outgoing error code + * could be U_MISSING_RESOURCE_ERROR if the key is not found + * could be a warning + * e.g.: U_USING_FALLBACK_WARNING,U_USING_DEFAULT_WARNING + * @return a signed integer value + * @stable ICU 2.0 + */ + int32_t + getInt(UErrorCode& status) const; + + /** + * Checks whether the resource has another element to iterate over. + * + * @return true if there are more elements, false if there is no more elements + * @stable ICU 2.0 + */ + UBool + hasNext(void) const; + + /** + * Resets the internal context of a resource so that iteration starts from the first element. + * + * @stable ICU 2.0 + */ + void + resetIterator(void); + + /** + * Returns the key associated with this resource. Not all the resources have a key - only + * those that are members of a table. + * + * @return a key associated to this resource, or nullptr if it doesn't have a key + * @stable ICU 2.0 + */ + const char* + getKey(void) const; + + /** + * Gets the locale ID of the resource bundle as a string. + * Same as getLocale().getName() . + * + * @return the locale ID of the resource bundle as a string + * @stable ICU 2.0 + */ + const char* + getName(void) const; + + + /** + * Returns the type of a resource. Available types are defined in enum UResType + * + * @return type of the given resource. + * @stable ICU 2.0 + */ + UResType + getType(void) const; + + /** + * Returns the next resource in a given resource or nullptr if there are no more resources + * + * @param status fills in the outgoing error code + * @return ResourceBundle object. + * @stable ICU 2.0 + */ + ResourceBundle + getNext(UErrorCode& status); + + /** + * Returns the next string in a resource or nullptr if there are no more resources + * to iterate over. + * + * @param status fills in the outgoing error code + * @return an UnicodeString object. + * @stable ICU 2.0 + */ + UnicodeString + getNextString(UErrorCode& status); + + /** + * Returns the next string in a resource or nullptr if there are no more resources + * to iterate over. + * + * @param key fill in for key associated with this string + * @param status fills in the outgoing error code + * @return an UnicodeString object. + * @stable ICU 2.0 + */ + UnicodeString + getNextString(const char ** key, + UErrorCode& status); + + /** + * Returns the resource in a resource at the specified index. + * + * @param index an index to the wanted resource. + * @param status fills in the outgoing error code + * @return ResourceBundle object. If there is an error, resource is invalid. + * @stable ICU 2.0 + */ + ResourceBundle + get(int32_t index, + UErrorCode& status) const; + + /** + * Returns the string in a given resource at the specified index. + * + * @param index an index to the wanted string. + * @param status fills in the outgoing error code + * @return an UnicodeString object. If there is an error, string is bogus + * @stable ICU 2.0 + */ + UnicodeString + getStringEx(int32_t index, + UErrorCode& status) const; + + /** + * Returns a resource in a resource that has a given key. This procedure works only with table + * resources. + * + * @param key a key associated with the wanted resource + * @param status fills in the outgoing error code. + * @return ResourceBundle object. If there is an error, resource is invalid. + * @stable ICU 2.0 + */ + ResourceBundle + get(const char* key, + UErrorCode& status) const; + + /** + * Returns a string in a resource that has a given key. This procedure works only with table + * resources. + * + * @param key a key associated with the wanted string + * @param status fills in the outgoing error code + * @return an UnicodeString object. If there is an error, string is bogus + * @stable ICU 2.0 + */ + UnicodeString + getStringEx(const char* key, + UErrorCode& status) const; + +#ifndef U_HIDE_DEPRECATED_API + /** + * Return the version number associated with this ResourceBundle as a string. Please + * use getVersion, as this method is going to be deprecated. + * + * @return A version number string as specified in the resource bundle or its parent. + * The caller does not own this string. + * @see getVersion + * @deprecated ICU 2.8 Use getVersion instead. + */ + const char* + getVersionNumber(void) const; +#endif /* U_HIDE_DEPRECATED_API */ + + /** + * Return the version number associated with this ResourceBundle as a UVersionInfo array. + * + * @param versionInfo A UVersionInfo array that is filled with the version number + * as specified in the resource bundle or its parent. + * @stable ICU 2.0 + */ + void + getVersion(UVersionInfo versionInfo) const; + +#ifndef U_HIDE_DEPRECATED_API + /** + * Return the Locale associated with this ResourceBundle. + * + * @return a Locale object + * @deprecated ICU 2.8 Use getLocale(ULocDataLocaleType type, UErrorCode &status) overload instead. + */ + const Locale& + getLocale(void) const; +#endif /* U_HIDE_DEPRECATED_API */ + + /** + * Return the Locale associated with this ResourceBundle. + * @param type You can choose between requested, valid and actual + * locale. For description see the definition of + * ULocDataLocaleType in uloc.h + * @param status just for catching illegal arguments + * + * @return a Locale object + * @stable ICU 2.8 + */ + const Locale + getLocale(ULocDataLocaleType type, UErrorCode &status) const; +#ifndef U_HIDE_INTERNAL_API + /** + * This API implements multilevel fallback + * @internal + */ + ResourceBundle + getWithFallback(const char* key, UErrorCode& status); +#endif /* U_HIDE_INTERNAL_API */ + /** + * ICU "poor man's RTTI", returns a UClassID for the actual class. + * + * @stable ICU 2.2 + */ + virtual UClassID getDynamicClassID() const override; + + /** + * ICU "poor man's RTTI", returns a UClassID for this class. + * + * @stable ICU 2.2 + */ + static UClassID U_EXPORT2 getStaticClassID(); + +private: + ResourceBundle() = delete; // default constructor not implemented + + UResourceBundle *fResource; + void constructForLocale(const UnicodeString& path, const Locale& locale, UErrorCode& error); + Locale *fLocale; +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/schriter.h b/intl/icu/source/common/unicode/schriter.h new file mode 100644 index 0000000000..a2ab17982d --- /dev/null +++ b/intl/icu/source/common/unicode/schriter.h @@ -0,0 +1,187 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1998-2005, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* +* File schriter.h +* +* Modification History: +* +* Date Name Description +* 05/05/99 stephen Cleaned up. +****************************************************************************** +*/ + +#ifndef SCHRITER_H +#define SCHRITER_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/chariter.h" +#include "unicode/uchriter.h" + +/** + * \file + * \brief C++ API: String Character Iterator + */ + +U_NAMESPACE_BEGIN +/** + * A concrete subclass of CharacterIterator that iterates over the + * characters (code units or code points) in a UnicodeString. + * It's possible not only to create an + * iterator that iterates over an entire UnicodeString, but also to + * create one that iterates over only a subrange of a UnicodeString + * (iterators over different subranges of the same UnicodeString don't + * compare equal). + * @see CharacterIterator + * @see ForwardCharacterIterator + * @stable ICU 2.0 + */ +class U_COMMON_API StringCharacterIterator : public UCharCharacterIterator { +public: + /** + * Create an iterator over the UnicodeString referred to by "textStr". + * The UnicodeString object is copied. + * The iteration range is the whole string, and the starting position is 0. + * @param textStr The unicode string used to create an iterator + * @stable ICU 2.0 + */ + StringCharacterIterator(const UnicodeString& textStr); + + /** + * Create an iterator over the UnicodeString referred to by "textStr". + * The iteration range is the whole string, and the starting + * position is specified by "textPos". If "textPos" is outside the valid + * iteration range, the behavior of this object is undefined. + * @param textStr The unicode string used to create an iterator + * @param textPos The starting position of the iteration + * @stable ICU 2.0 + */ + StringCharacterIterator(const UnicodeString& textStr, + int32_t textPos); + + /** + * Create an iterator over the UnicodeString referred to by "textStr". + * The UnicodeString object is copied. + * The iteration range begins with the code unit specified by + * "textBegin" and ends with the code unit BEFORE the code unit specified + * by "textEnd". The starting position is specified by "textPos". If + * "textBegin" and "textEnd" don't form a valid range on "text" (i.e., + * textBegin >= textEnd or either is negative or greater than text.size()), + * or "textPos" is outside the range defined by "textBegin" and "textEnd", + * the behavior of this iterator is undefined. + * @param textStr The unicode string used to create the StringCharacterIterator + * @param textBegin The begin position of the iteration range + * @param textEnd The end position of the iteration range + * @param textPos The starting position of the iteration + * @stable ICU 2.0 + */ + StringCharacterIterator(const UnicodeString& textStr, + int32_t textBegin, + int32_t textEnd, + int32_t textPos); + + /** + * Copy constructor. The new iterator iterates over the same range + * of the same string as "that", and its initial position is the + * same as "that"'s current position. + * The UnicodeString object in "that" is copied. + * @param that The StringCharacterIterator to be copied + * @stable ICU 2.0 + */ + StringCharacterIterator(const StringCharacterIterator& that); + + /** + * Destructor. + * @stable ICU 2.0 + */ + virtual ~StringCharacterIterator(); + + /** + * Assignment operator. *this is altered to iterate over the same + * range of the same string as "that", and refers to the same + * character within that string as "that" does. + * @param that The object to be copied. + * @return the newly created object. + * @stable ICU 2.0 + */ + StringCharacterIterator& + operator=(const StringCharacterIterator& that); + + /** + * Returns true if the iterators iterate over the same range of the + * same string and are pointing at the same character. + * @param that The ForwardCharacterIterator to be compared for equality + * @return true if the iterators iterate over the same range of the + * same string and are pointing at the same character. + * @stable ICU 2.0 + */ + virtual bool operator==(const ForwardCharacterIterator& that) const override; + + /** + * Returns a new StringCharacterIterator referring to the same + * character in the same range of the same string as this one. The + * caller must delete the new iterator. + * @return the newly cloned object. + * @stable ICU 2.0 + */ + virtual StringCharacterIterator* clone() const override; + + /** + * Sets the iterator to iterate over the provided string. + * @param newText The string to be iterated over + * @stable ICU 2.0 + */ + void setText(const UnicodeString& newText); + + /** + * Copies the UnicodeString under iteration into the UnicodeString + * referred to by "result". Even if this iterator iterates across + * only a part of this string, the whole string is copied. + * @param result Receives a copy of the text under iteration. + * @stable ICU 2.0 + */ + virtual void getText(UnicodeString& result) override; + + /** + * Return a class ID for this object (not really public) + * @return a class ID for this object. + * @stable ICU 2.0 + */ + virtual UClassID getDynamicClassID(void) const override; + + /** + * Return a class ID for this class (not really public) + * @return a class ID for this class + * @stable ICU 2.0 + */ + static UClassID U_EXPORT2 getStaticClassID(void); + +protected: + /** + * Default constructor, iteration over empty string. + * @stable ICU 2.0 + */ + StringCharacterIterator(); + + /** + * Copy of the iterated string object. + * @stable ICU 2.0 + */ + UnicodeString text; + +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/simpleformatter.h b/intl/icu/source/common/unicode/simpleformatter.h new file mode 100644 index 0000000000..7f58106fad --- /dev/null +++ b/intl/icu/source/common/unicode/simpleformatter.h @@ -0,0 +1,341 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 2014-2016, International Business Machines +* Corporation and others. All Rights Reserved. +****************************************************************************** +* simpleformatter.h +*/ + +#ifndef __SIMPLEFORMATTER_H__ +#define __SIMPLEFORMATTER_H__ + +/** + * \file + * \brief C++ API: Simple formatter, minimal subset of MessageFormat. + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/unistr.h" + +U_NAMESPACE_BEGIN + +// Forward declaration: +namespace number { +namespace impl { +class SimpleModifier; +} +} + +/** + * Formats simple patterns like "{1} was born in {0}". + * Minimal subset of MessageFormat; fast, simple, minimal dependencies. + * Supports only numbered arguments with no type nor style parameters, + * and formats only string values. + * Quoting via ASCII apostrophe compatible with ICU MessageFormat default behavior. + * + * Factory methods set error codes for syntax errors + * and for too few or too many arguments/placeholders. + * + * SimpleFormatter objects are thread-safe except for assignment and applying new patterns. + * + * Example: + * 

+ * UErrorCode errorCode = U_ZERO_ERROR;
+ * SimpleFormatter fmt("{1} '{born}' in {0}", errorCode);
+ * UnicodeString result;
+ *
+ * // Output: "paul {born} in england"
+ * fmt.format("england", "paul", result, errorCode);
+ *
+ * + * This class is not intended for public subclassing. + * + * @see MessageFormat + * @see UMessagePatternApostropheMode + * @stable ICU 57 + */ +class U_COMMON_API SimpleFormatter final : public UMemory { +public: + /** + * Default constructor. + * @stable ICU 57 + */ + SimpleFormatter() : compiledPattern((char16_t)0) {} + + /** + * Constructs a formatter from the pattern string. + * + * @param pattern The pattern string. + * @param errorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * Set to U_ILLEGAL_ARGUMENT_ERROR for bad argument syntax. + * @stable ICU 57 + */ + SimpleFormatter(const UnicodeString& pattern, UErrorCode &errorCode) { + applyPattern(pattern, errorCode); + } + + /** + * Constructs a formatter from the pattern string. + * The number of arguments checked against the given limits is the + * highest argument number plus one, not the number of occurrences of arguments. + * + * @param pattern The pattern string. + * @param min The pattern must have at least this many arguments. + * @param max The pattern must have at most this many arguments. + * @param errorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * Set to U_ILLEGAL_ARGUMENT_ERROR for bad argument syntax and + * too few or too many arguments. + * @stable ICU 57 + */ + SimpleFormatter(const UnicodeString& pattern, int32_t min, int32_t max, + UErrorCode &errorCode) { + applyPatternMinMaxArguments(pattern, min, max, errorCode); + } + + /** + * Copy constructor. + * @stable ICU 57 + */ + SimpleFormatter(const SimpleFormatter& other) + : compiledPattern(other.compiledPattern) {} + + /** + * Assignment operator. + * @stable ICU 57 + */ + SimpleFormatter &operator=(const SimpleFormatter& other); + + /** + * Destructor. + * @stable ICU 57 + */ + ~SimpleFormatter(); + + /** + * Changes this object according to the new pattern. + * + * @param pattern The pattern string. + * @param errorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * Set to U_ILLEGAL_ARGUMENT_ERROR for bad argument syntax. + * @return true if U_SUCCESS(errorCode). + * @stable ICU 57 + */ + UBool applyPattern(const UnicodeString &pattern, UErrorCode &errorCode) { + return applyPatternMinMaxArguments(pattern, 0, INT32_MAX, errorCode); + } + + /** + * Changes this object according to the new pattern. + * The number of arguments checked against the given limits is the + * highest argument number plus one, not the number of occurrences of arguments. + * + * @param pattern The pattern string. + * @param min The pattern must have at least this many arguments. + * @param max The pattern must have at most this many arguments. + * @param errorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * Set to U_ILLEGAL_ARGUMENT_ERROR for bad argument syntax and + * too few or too many arguments. + * @return true if U_SUCCESS(errorCode). + * @stable ICU 57 + */ + UBool applyPatternMinMaxArguments(const UnicodeString &pattern, + int32_t min, int32_t max, UErrorCode &errorCode); + + /** + * @return The max argument number + 1. + * @stable ICU 57 + */ + int32_t getArgumentLimit() const { + return getArgumentLimit(compiledPattern.getBuffer(), compiledPattern.length()); + } + + /** + * Formats the given value, appending to the appendTo builder. + * The argument value must not be the same object as appendTo. + * getArgumentLimit() must be at most 1. + * + * @param value0 Value for argument {0}. + * @param appendTo Gets the formatted pattern and value appended. + * @param errorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return appendTo + * @stable ICU 57 + */ + UnicodeString &format( + const UnicodeString &value0, + UnicodeString &appendTo, UErrorCode &errorCode) const; + + /** + * Formats the given values, appending to the appendTo builder. + * An argument value must not be the same object as appendTo. + * getArgumentLimit() must be at most 2. + * + * @param value0 Value for argument {0}. + * @param value1 Value for argument {1}. + * @param appendTo Gets the formatted pattern and values appended. + * @param errorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return appendTo + * @stable ICU 57 + */ + UnicodeString &format( + const UnicodeString &value0, + const UnicodeString &value1, + UnicodeString &appendTo, UErrorCode &errorCode) const; + + /** + * Formats the given values, appending to the appendTo builder. + * An argument value must not be the same object as appendTo. + * getArgumentLimit() must be at most 3. + * + * @param value0 Value for argument {0}. + * @param value1 Value for argument {1}. + * @param value2 Value for argument {2}. + * @param appendTo Gets the formatted pattern and values appended. + * @param errorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return appendTo + * @stable ICU 57 + */ + UnicodeString &format( + const UnicodeString &value0, + const UnicodeString &value1, + const UnicodeString &value2, + UnicodeString &appendTo, UErrorCode &errorCode) const; + + /** + * Formats the given values, appending to the appendTo string. + * + * @param values The argument values. + * An argument value must not be the same object as appendTo. + * Can be nullptr if valuesLength==getArgumentLimit()==0. + * @param valuesLength The length of the values array. + * Must be at least getArgumentLimit(). + * @param appendTo Gets the formatted pattern and values appended. + * @param offsets offsets[i] receives the offset of where + * values[i] replaced pattern argument {i}. + * Can be shorter or longer than values. Can be nullptr if offsetsLength==0. + * If there is no {i} in the pattern, then offsets[i] is set to -1. + * @param offsetsLength The length of the offsets array. + * @param errorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return appendTo + * @stable ICU 57 + */ + UnicodeString &formatAndAppend( + const UnicodeString *const *values, int32_t valuesLength, + UnicodeString &appendTo, + int32_t *offsets, int32_t offsetsLength, UErrorCode &errorCode) const; + + /** + * Formats the given values, replacing the contents of the result string. + * May optimize by actually appending to the result if it is the same object + * as the value corresponding to the initial argument in the pattern. + * + * @param values The argument values. + * An argument value may be the same object as result. + * Can be nullptr if valuesLength==getArgumentLimit()==0. + * @param valuesLength The length of the values array. + * Must be at least getArgumentLimit(). + * @param result Gets its contents replaced by the formatted pattern and values. + * @param offsets offsets[i] receives the offset of where + * values[i] replaced pattern argument {i}. + * Can be shorter or longer than values. Can be nullptr if offsetsLength==0. + * If there is no {i} in the pattern, then offsets[i] is set to -1. + * @param offsetsLength The length of the offsets array. + * @param errorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return result + * @stable ICU 57 + */ + UnicodeString &formatAndReplace( + const UnicodeString *const *values, int32_t valuesLength, + UnicodeString &result, + int32_t *offsets, int32_t offsetsLength, UErrorCode &errorCode) const; + + /** + * Returns the pattern text with none of the arguments. + * Like formatting with all-empty string values. + * @stable ICU 57 + */ + UnicodeString getTextWithNoArguments() const { + return getTextWithNoArguments( + compiledPattern.getBuffer(), + compiledPattern.length(), + nullptr, + 0); + } + +#ifndef U_HIDE_INTERNAL_API + /** + * Returns the pattern text with none of the arguments. + * Like formatting with all-empty string values. + * + * TODO(ICU-20406): Replace this with an Iterator interface. + * + * @param offsets offsets[i] receives the offset of where {i} was located + * before it was replaced by an empty string. + * For example, "a{0}b{1}" produces offset 1 for i=0 and 2 for i=1. + * Can be nullptr if offsetsLength==0. + * If there is no {i} in the pattern, then offsets[i] is set to -1. + * @param offsetsLength The length of the offsets array. + * + * @internal + */ + UnicodeString getTextWithNoArguments(int32_t *offsets, int32_t offsetsLength) const { + return getTextWithNoArguments( + compiledPattern.getBuffer(), + compiledPattern.length(), + offsets, + offsetsLength); + } +#endif // U_HIDE_INTERNAL_API + +private: + /** + * Binary representation of the compiled pattern. + * Index 0: One more than the highest argument number. + * Followed by zero or more arguments or literal-text segments. + * + * An argument is stored as its number, less than ARG_NUM_LIMIT. + * A literal-text segment is stored as its length (at least 1) offset by ARG_NUM_LIMIT, + * followed by that many chars. + */ + UnicodeString compiledPattern; + + static inline int32_t getArgumentLimit(const char16_t *compiledPattern, + int32_t compiledPatternLength) { + return compiledPatternLength == 0 ? 0 : compiledPattern[0]; + } + + static UnicodeString getTextWithNoArguments( + const char16_t *compiledPattern, + int32_t compiledPatternLength, + int32_t *offsets, + int32_t offsetsLength); + + static UnicodeString &format( + const char16_t *compiledPattern, int32_t compiledPatternLength, + const UnicodeString *const *values, + UnicodeString &result, const UnicodeString *resultCopy, UBool forbidResultAsValue, + int32_t *offsets, int32_t offsetsLength, + UErrorCode &errorCode); + + // Give access to internals to SimpleModifier for number formatting + friend class number::impl::SimpleModifier; +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __SIMPLEFORMATTER_H__ diff --git a/intl/icu/source/common/unicode/std_string.h b/intl/icu/source/common/unicode/std_string.h new file mode 100644 index 0000000000..bf87230167 --- /dev/null +++ b/intl/icu/source/common/unicode/std_string.h @@ -0,0 +1,41 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2009-2014, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: std_string.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2009feb19 +* created by: Markus W. Scherer +*/ + +#ifndef __STD_STRING_H__ +#define __STD_STRING_H__ + +/** + * \file + * \brief C++ API: Central ICU header for including the C++ standard <string> + * header and for related definitions. + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +// Workaround for a libstdc++ bug before libstdc++4.6 (2011). +// https://bugs.llvm.org/show_bug.cgi?id=13364 +#if defined(__GLIBCXX__) +namespace std { class type_info; } +#endif +#include + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __STD_STRING_H__ diff --git a/intl/icu/source/common/unicode/strenum.h b/intl/icu/source/common/unicode/strenum.h new file mode 100644 index 0000000000..fba5c9b814 --- /dev/null +++ b/intl/icu/source/common/unicode/strenum.h @@ -0,0 +1,281 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2002-2012, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +*/ + +#ifndef STRENUM_H +#define STRENUM_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/uobject.h" +#include "unicode/unistr.h" + +/** + * \file + * \brief C++ API: String Enumeration + */ + +U_NAMESPACE_BEGIN + +/** + * Base class for 'pure' C++ implementations of uenum api. Adds a + * method that returns the next UnicodeString since in C++ this can + * be a common storage format for strings. + * + *

The model is that the enumeration is over strings maintained by + * a 'service.' At any point, the service might change, invalidating + * the enumerator (though this is expected to be rare). The iterator + * returns an error if this has occurred. Lack of the error is no + * guarantee that the service didn't change immediately after the + * call, so the returned string still might not be 'valid' on + * subsequent use.

+ * + *

Strings may take the form of const char*, const char16_t*, or const + * UnicodeString*. The type you get is determine by the variant of + * 'next' that you call. In general the StringEnumeration is + * optimized for one of these types, but all StringEnumerations can + * return all types. Returned strings are each terminated with a NUL. + * Depending on the service data, they might also include embedded NUL + * characters, so API is provided to optionally return the true + * length, counting the embedded NULs but not counting the terminating + * NUL.

+ * + *

The pointers returned by next, unext, and snext become invalid + * upon any subsequent call to the enumeration's destructor, next, + * unext, snext, or reset.

+ * + * ICU 2.8 adds some default implementations and helper functions + * for subclasses. + * + * @stable ICU 2.4 + */ +class U_COMMON_API StringEnumeration : public UObject { +public: + /** + * Destructor. + * @stable ICU 2.4 + */ + virtual ~StringEnumeration(); + + /** + * Clone this object, an instance of a subclass of StringEnumeration. + * Clones can be used concurrently in multiple threads. + * If a subclass does not implement clone(), or if an error occurs, + * then nullptr is returned. + * The caller must delete the clone. + * + * @return a clone of this object + * + * @see getDynamicClassID + * @stable ICU 2.8 + */ + virtual StringEnumeration *clone() const; + + /** + *

Return the number of elements that the iterator traverses. If + * the iterator is out of sync with its service, status is set to + * U_ENUM_OUT_OF_SYNC_ERROR, and the return value is zero.

+ * + *

The return value will not change except possibly as a result of + * a subsequent call to reset, or if the iterator becomes out of sync.

+ * + *

This is a convenience function. It can end up being very + * expensive as all the items might have to be pre-fetched + * (depending on the storage format of the data being + * traversed).

+ * + * @param status the error code. + * @return number of elements in the iterator. + * + * @stable ICU 2.4 */ + virtual int32_t count(UErrorCode& status) const = 0; + + /** + *

Returns the next element as a NUL-terminated char*. If there + * are no more elements, returns nullptr. If the resultLength pointer + * is not nullptr, the length of the string (not counting the + * terminating NUL) is returned at that address. If an error + * status is returned, the value at resultLength is undefined.

+ * + *

The returned pointer is owned by this iterator and must not be + * deleted by the caller. The pointer is valid until the next call + * to next, unext, snext, reset, or the enumerator's destructor.

+ * + *

If the iterator is out of sync with its service, status is set + * to U_ENUM_OUT_OF_SYNC_ERROR and nullptr is returned.

+ * + *

If the native service string is a char16_t* string, it is + * converted to char* with the invariant converter. If the + * conversion fails (because a character cannot be converted) then + * status is set to U_INVARIANT_CONVERSION_ERROR and the return + * value is undefined (though not nullptr).

+ * + * Starting with ICU 2.8, the default implementation calls snext() + * and handles the conversion. + * Either next() or snext() must be implemented differently by a subclass. + * + * @param status the error code. + * @param resultLength a pointer to receive the length, can be nullptr. + * @return a pointer to the string, or nullptr. + * + * @stable ICU 2.4 + */ + virtual const char* next(int32_t *resultLength, UErrorCode& status); + + /** + *

Returns the next element as a NUL-terminated char16_t*. If there + * are no more elements, returns nullptr. If the resultLength pointer + * is not nullptr, the length of the string (not counting the + * terminating NUL) is returned at that address. If an error + * status is returned, the value at resultLength is undefined.

+ * + *

The returned pointer is owned by this iterator and must not be + * deleted by the caller. The pointer is valid until the next call + * to next, unext, snext, reset, or the enumerator's destructor.

+ * + *

If the iterator is out of sync with its service, status is set + * to U_ENUM_OUT_OF_SYNC_ERROR and nullptr is returned.

+ * + * Starting with ICU 2.8, the default implementation calls snext() + * and handles the conversion. + * + * @param status the error code. + * @param resultLength a pointer to receive the length, can be nullptr. + * @return a pointer to the string, or nullptr. + * + * @stable ICU 2.4 + */ + virtual const char16_t* unext(int32_t *resultLength, UErrorCode& status); + + /** + *

Returns the next element a UnicodeString*. If there are no + * more elements, returns nullptr.

+ * + *

The returned pointer is owned by this iterator and must not be + * deleted by the caller. The pointer is valid until the next call + * to next, unext, snext, reset, or the enumerator's destructor.

+ * + *

If the iterator is out of sync with its service, status is set + * to U_ENUM_OUT_OF_SYNC_ERROR and nullptr is returned.

+ * + * Starting with ICU 2.8, the default implementation calls next() + * and handles the conversion. + * Either next() or snext() must be implemented differently by a subclass. + * + * @param status the error code. + * @return a pointer to the string, or nullptr. + * + * @stable ICU 2.4 + */ + virtual const UnicodeString* snext(UErrorCode& status); + + /** + *

Resets the iterator. This re-establishes sync with the + * service and rewinds the iterator to start at the first + * element.

+ * + *

Previous pointers returned by next, unext, or snext become + * invalid, and the value returned by count might change.

+ * + * @param status the error code. + * + * @stable ICU 2.4 + */ + virtual void reset(UErrorCode& status) = 0; + + /** + * Compares this enumeration to other to check if both are equal + * + * @param that The other string enumeration to compare this object to + * @return true if the enumerations are equal. false if not. + * @stable ICU 3.6 + */ + virtual bool operator==(const StringEnumeration& that)const; + /** + * Compares this enumeration to other to check if both are not equal + * + * @param that The other string enumeration to compare this object to + * @return true if the enumerations are equal. false if not. + * @stable ICU 3.6 + */ + virtual bool operator!=(const StringEnumeration& that)const; + +protected: + /** + * UnicodeString field for use with default implementations and subclasses. + * @stable ICU 2.8 + */ + UnicodeString unistr; + /** + * char * default buffer for use with default implementations and subclasses. + * @stable ICU 2.8 + */ + char charsBuffer[32]; + /** + * char * buffer for use with default implementations and subclasses. + * Allocated in constructor and in ensureCharsCapacity(). + * @stable ICU 2.8 + */ + char *chars; + /** + * Capacity of chars, for use with default implementations and subclasses. + * @stable ICU 2.8 + */ + int32_t charsCapacity; + + /** + * Default constructor for use with default implementations and subclasses. + * @stable ICU 2.8 + */ + StringEnumeration(); + + /** + * Ensures that chars is at least as large as the requested capacity. + * For use with default implementations and subclasses. + * + * @param capacity Requested capacity. + * @param status ICU in/out error code. + * @stable ICU 2.8 + */ + void ensureCharsCapacity(int32_t capacity, UErrorCode &status); + + /** + * Converts s to Unicode and sets unistr to the result. + * For use with default implementations and subclasses, + * especially for implementations of snext() in terms of next(). + * This is provided with a helper function instead of a default implementation + * of snext() to avoid potential infinite loops between next() and snext(). + * + * For example: + * \code + * const UnicodeString* snext(UErrorCode& status) { + * int32_t resultLength=0; + * const char *s=next(&resultLength, status); + * return setChars(s, resultLength, status); + * } + * \endcode + * + * @param s String to be converted to Unicode. + * @param length Length of the string. + * @param status ICU in/out error code. + * @return A pointer to unistr. + * @stable ICU 2.8 + */ + UnicodeString *setChars(const char *s, int32_t length, UErrorCode &status); +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +/* STRENUM_H */ +#endif diff --git a/intl/icu/source/common/unicode/stringoptions.h b/intl/icu/source/common/unicode/stringoptions.h new file mode 100644 index 0000000000..7b9f70944f --- /dev/null +++ b/intl/icu/source/common/unicode/stringoptions.h @@ -0,0 +1,190 @@ +// © 2017 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// stringoptions.h +// created: 2017jun08 Markus W. Scherer + +#ifndef __STRINGOPTIONS_H__ +#define __STRINGOPTIONS_H__ + +#include "unicode/utypes.h" + +/** + * \file + * \brief C API: Bit set option bit constants for various string and character processing functions. + */ + +/** + * Option value for case folding: Use default mappings defined in CaseFolding.txt. + * + * @stable ICU 2.0 + */ +#define U_FOLD_CASE_DEFAULT 0 + +/** + * Option value for case folding: + * + * Use the modified set of mappings provided in CaseFolding.txt to handle dotted I + * and dotless i appropriately for Turkic languages (tr, az). + * + * Before Unicode 3.2, CaseFolding.txt contains mappings marked with 'I' that + * are to be included for default mappings and + * excluded for the Turkic-specific mappings. + * + * Unicode 3.2 CaseFolding.txt instead contains mappings marked with 'T' that + * are to be excluded for default mappings and + * included for the Turkic-specific mappings. + * + * @stable ICU 2.0 + */ +#define U_FOLD_CASE_EXCLUDE_SPECIAL_I 1 + +/** + * Titlecase the string as a whole rather than each word. + * (Titlecase only the character at index 0, possibly adjusted.) + * Option bits value for titlecasing APIs that take an options bit set. + * + * It is an error to specify multiple titlecasing iterator options together, + * including both an options bit and an explicit BreakIterator. + * + * @see U_TITLECASE_ADJUST_TO_CASED + * @stable ICU 60 + */ +#define U_TITLECASE_WHOLE_STRING 0x20 + +/** + * Titlecase sentences rather than words. + * (Titlecase only the first character of each sentence, possibly adjusted.) + * Option bits value for titlecasing APIs that take an options bit set. + * + * It is an error to specify multiple titlecasing iterator options together, + * including both an options bit and an explicit BreakIterator. + * + * @see U_TITLECASE_ADJUST_TO_CASED + * @stable ICU 60 + */ +#define U_TITLECASE_SENTENCES 0x40 + +/** + * Do not lowercase non-initial parts of words when titlecasing. + * Option bit for titlecasing APIs that take an options bit set. + * + * By default, titlecasing will titlecase the character at each + * (possibly adjusted) BreakIterator index and + * lowercase all other characters up to the next iterator index. + * With this option, the other characters will not be modified. + * + * @see U_TITLECASE_ADJUST_TO_CASED + * @see UnicodeString::toTitle + * @see CaseMap::toTitle + * @see ucasemap_setOptions + * @see ucasemap_toTitle + * @see ucasemap_utf8ToTitle + * @stable ICU 3.8 + */ +#define U_TITLECASE_NO_LOWERCASE 0x100 + +/** + * Do not adjust the titlecasing BreakIterator indexes; + * titlecase exactly the characters at breaks from the iterator. + * Option bit for titlecasing APIs that take an options bit set. + * + * By default, titlecasing will take each break iterator index, + * adjust it to the next relevant character (see U_TITLECASE_ADJUST_TO_CASED), + * and titlecase that one. + * + * Other characters are lowercased. + * + * It is an error to specify multiple titlecasing adjustment options together. + * + * @see U_TITLECASE_ADJUST_TO_CASED + * @see U_TITLECASE_NO_LOWERCASE + * @see UnicodeString::toTitle + * @see CaseMap::toTitle + * @see ucasemap_setOptions + * @see ucasemap_toTitle + * @see ucasemap_utf8ToTitle + * @stable ICU 3.8 + */ +#define U_TITLECASE_NO_BREAK_ADJUSTMENT 0x200 + +/** + * Adjust each titlecasing BreakIterator index to the next cased character. + * (See the Unicode Standard, chapter 3, Default Case Conversion, R3 toTitlecase(X).) + * Option bit for titlecasing APIs that take an options bit set. + * + * This used to be the default index adjustment in ICU. + * Since ICU 60, the default index adjustment is to the next character that is + * a letter, number, symbol, or private use code point. + * (Uncased modifier letters are skipped.) + * The difference in behavior is small for word titlecasing, + * but the new adjustment is much better for whole-string and sentence titlecasing: + * It yields "49ers" and "«丰(abc)»" instead of "49Ers" and "«丰(Abc)»". + * + * It is an error to specify multiple titlecasing adjustment options together. + * + * @see U_TITLECASE_NO_BREAK_ADJUSTMENT + * @stable ICU 60 + */ +#define U_TITLECASE_ADJUST_TO_CASED 0x400 + +/** + * Option for string transformation functions to not first reset the Edits object. + * Used for example in some case-mapping and normalization functions. + * + * @see CaseMap + * @see Edits + * @see Normalizer2 + * @stable ICU 60 + */ +#define U_EDITS_NO_RESET 0x2000 + +/** + * Omit unchanged text when recording how source substrings + * relate to changed and unchanged result substrings. + * Used for example in some case-mapping and normalization functions. + * + * @see CaseMap + * @see Edits + * @see Normalizer2 + * @stable ICU 60 + */ +#define U_OMIT_UNCHANGED_TEXT 0x4000 + +/** + * Option bit for u_strCaseCompare, u_strcasecmp, unorm_compare, etc: + * Compare strings in code point order instead of code unit order. + * @stable ICU 2.2 + */ +#define U_COMPARE_CODE_POINT_ORDER 0x8000 + +/** + * Option bit for unorm_compare: + * Perform case-insensitive comparison. + * @stable ICU 2.2 + */ +#define U_COMPARE_IGNORE_CASE 0x10000 + +/** + * Option bit for unorm_compare: + * Both input strings are assumed to fulfill FCD conditions. + * @stable ICU 2.2 + */ +#define UNORM_INPUT_IS_FCD 0x20000 + +// Related definitions elsewhere. +// Options that are not meaningful in the same functions +// can share the same bits. +// +// Public: +// unicode/unorm.h #define UNORM_COMPARE_NORM_OPTIONS_SHIFT 20 +// +// Internal: (may change or be removed) +// ucase.h #define _STRCASECMP_OPTIONS_MASK 0xffff +// ucase.h #define _FOLD_CASE_OPTIONS_MASK 7 +// ucasemap_imp.h #define U_TITLECASE_ITERATOR_MASK 0xe0 +// ucasemap_imp.h #define U_TITLECASE_ADJUSTMENT_MASK 0x600 +// ustr_imp.h #define _STRNCMP_STYLE 0x1000 +// unormcmp.cpp #define _COMPARE_EQUIV 0x80000 + +#endif // __STRINGOPTIONS_H__ diff --git a/intl/icu/source/common/unicode/stringpiece.h b/intl/icu/source/common/unicode/stringpiece.h new file mode 100644 index 0000000000..df7f36089d --- /dev/null +++ b/intl/icu/source/common/unicode/stringpiece.h @@ -0,0 +1,343 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +// Copyright (C) 2009-2013, International Business Machines +// Corporation and others. All Rights Reserved. +// +// Copyright 2001 and onwards Google Inc. +// Author: Sanjay Ghemawat + +// This code is a contribution of Google code, and the style used here is +// a compromise between the original Google code and the ICU coding guidelines. +// For example, data types are ICU-ified (size_t,int->int32_t), +// and API comments doxygen-ified, but function names and behavior are +// as in the original, if possible. +// Assertion-style error handling, not available in ICU, was changed to +// parameter "pinning" similar to UnicodeString. +// +// In addition, this is only a partial port of the original Google code, +// limited to what was needed so far. The (nearly) complete original code +// is in the ICU svn repository at icuhtml/trunk/design/strings/contrib +// (see ICU ticket 6765, r25517). + +#ifndef __STRINGPIECE_H__ +#define __STRINGPIECE_H__ + +/** + * \file + * \brief C++ API: StringPiece: Read-only byte string wrapper class. + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include +#include + +#include "unicode/uobject.h" +#include "unicode/std_string.h" + +// Arghh! I wish C++ literals were "string". + +U_NAMESPACE_BEGIN + +/** + * A string-like object that points to a sized piece of memory. + * + * We provide non-explicit singleton constructors so users can pass + * in a "const char*" or a "string" wherever a "StringPiece" is + * expected. + * + * Functions or methods may use StringPiece parameters to accept either a + * "const char*" or a "string" value that will be implicitly converted to a + * StringPiece. + * + * Systematic usage of StringPiece is encouraged as it will reduce unnecessary + * conversions from "const char*" to "string" and back again. + * + * @stable ICU 4.2 + */ +class U_COMMON_API StringPiece : public UMemory { + private: + const char* ptr_; + int32_t length_; + + public: + /** + * Default constructor, creates an empty StringPiece. + * @stable ICU 4.2 + */ + StringPiece() : ptr_(nullptr), length_(0) { } + + /** + * Constructs from a NUL-terminated const char * pointer. + * @param str a NUL-terminated const char * pointer + * @stable ICU 4.2 + */ + StringPiece(const char* str); +#if defined(__cpp_char8_t) || defined(U_IN_DOXYGEN) + /** + * Constructs from a NUL-terminated const char8_t * pointer. + * @param str a NUL-terminated const char8_t * pointer + * @stable ICU 67 + */ + StringPiece(const char8_t* str) : StringPiece(reinterpret_cast(str)) {} +#endif + /** + * Constructs an empty StringPiece. + * Needed for type disambiguation from multiple other overloads. + * @param p nullptr + * @stable ICU 67 + */ + StringPiece(std::nullptr_t p) : ptr_(p), length_(0) {} + + /** + * Constructs from a std::string. + * @stable ICU 4.2 + */ + StringPiece(const std::string& str) + : ptr_(str.data()), length_(static_cast(str.size())) { } +#if defined(__cpp_lib_char8_t) || defined(U_IN_DOXYGEN) + /** + * Constructs from a std::u8string. + * @stable ICU 67 + */ + StringPiece(const std::u8string& str) + : ptr_(reinterpret_cast(str.data())), + length_(static_cast(str.size())) { } +#endif + + /** + * Constructs from some other implementation of a string piece class, from any + * C++ record type that has these two methods: + * + * \code{.cpp} + * + * struct OtherStringPieceClass { + * const char* data(); // or const char8_t* + * size_t size(); + * }; + * + * \endcode + * + * The other string piece class will typically be std::string_view from C++17 + * or absl::string_view from Abseil. + * + * Starting with C++20, data() may also return a const char8_t* pointer, + * as from std::u8string_view. + * + * @param str the other string piece + * @stable ICU 65 + */ + template ::value +#if defined(__cpp_char8_t) + || std::is_same::value +#endif + ) && + std::is_same::value>::type> + StringPiece(T str) + : ptr_(reinterpret_cast(str.data())), + length_(static_cast(str.size())) {} + + /** + * Constructs from a const char * pointer and a specified length. + * @param offset a const char * pointer (need not be terminated) + * @param len the length of the string; must be non-negative + * @stable ICU 4.2 + */ + StringPiece(const char* offset, int32_t len) : ptr_(offset), length_(len) { } +#if defined(__cpp_char8_t) || defined(U_IN_DOXYGEN) + /** + * Constructs from a const char8_t * pointer and a specified length. + * @param str a const char8_t * pointer (need not be terminated) + * @param len the length of the string; must be non-negative + * @stable ICU 67 + */ + StringPiece(const char8_t* str, int32_t len) : + StringPiece(reinterpret_cast(str), len) {} +#endif + + /** + * Substring of another StringPiece. + * @param x the other StringPiece + * @param pos start position in x; must be non-negative and <= x.length(). + * @stable ICU 4.2 + */ + StringPiece(const StringPiece& x, int32_t pos); + /** + * Substring of another StringPiece. + * @param x the other StringPiece + * @param pos start position in x; must be non-negative and <= x.length(). + * @param len length of the substring; + * must be non-negative and will be pinned to at most x.length() - pos. + * @stable ICU 4.2 + */ + StringPiece(const StringPiece& x, int32_t pos, int32_t len); + + /** + * Returns the string pointer. May be nullptr if it is empty. + * + * data() may return a pointer to a buffer with embedded NULs, and the + * returned buffer may or may not be null terminated. Therefore it is + * typically a mistake to pass data() to a routine that expects a NUL + * terminated string. + * @return the string pointer + * @stable ICU 4.2 + */ + const char* data() const { return ptr_; } + /** + * Returns the string length. Same as length(). + * @return the string length + * @stable ICU 4.2 + */ + int32_t size() const { return length_; } + /** + * Returns the string length. Same as size(). + * @return the string length + * @stable ICU 4.2 + */ + int32_t length() const { return length_; } + /** + * Returns whether the string is empty. + * @return true if the string is empty + * @stable ICU 4.2 + */ + UBool empty() const { return length_ == 0; } + + /** + * Sets to an empty string. + * @stable ICU 4.2 + */ + void clear() { ptr_ = nullptr; length_ = 0; } + + /** + * Reset the stringpiece to refer to new data. + * @param xdata pointer the new string data. Need not be nul terminated. + * @param len the length of the new data + * @stable ICU 4.8 + */ + void set(const char* xdata, int32_t len) { ptr_ = xdata; length_ = len; } + + /** + * Reset the stringpiece to refer to new data. + * @param str a pointer to a NUL-terminated string. + * @stable ICU 4.8 + */ + void set(const char* str); + +#if defined(__cpp_char8_t) || defined(U_IN_DOXYGEN) + /** + * Resets the stringpiece to refer to new data. + * @param xdata pointer the new string data. Need not be NUL-terminated. + * @param len the length of the new data + * @stable ICU 67 + */ + inline void set(const char8_t* xdata, int32_t len) { + set(reinterpret_cast(xdata), len); + } + + /** + * Resets the stringpiece to refer to new data. + * @param str a pointer to a NUL-terminated string. + * @stable ICU 67 + */ + inline void set(const char8_t* str) { + set(reinterpret_cast(str)); + } +#endif + + /** + * Removes the first n string units. + * @param n prefix length, must be non-negative and <=length() + * @stable ICU 4.2 + */ + void remove_prefix(int32_t n) { + if (n >= 0) { + if (n > length_) { + n = length_; + } + ptr_ += n; + length_ -= n; + } + } + + /** + * Removes the last n string units. + * @param n suffix length, must be non-negative and <=length() + * @stable ICU 4.2 + */ + void remove_suffix(int32_t n) { + if (n >= 0) { + if (n <= length_) { + length_ -= n; + } else { + length_ = 0; + } + } + } + + /** + * Searches the StringPiece for the given search string (needle); + * @param needle The string for which to search. + * @param offset Where to start searching within this string (haystack). + * @return The offset of needle in haystack, or -1 if not found. + * @stable ICU 67 + */ + int32_t find(StringPiece needle, int32_t offset); + + /** + * Compares this StringPiece with the other StringPiece, with semantics + * similar to std::string::compare(). + * @param other The string to compare to. + * @return below zero if this < other; above zero if this > other; 0 if this == other. + * @stable ICU 67 + */ + int32_t compare(StringPiece other); + + /** + * Maximum integer, used as a default value for substring methods. + * @stable ICU 4.2 + */ + static const int32_t npos; // = 0x7fffffff; + + /** + * Returns a substring of this StringPiece. + * @param pos start position; must be non-negative and <= length(). + * @param len length of the substring; + * must be non-negative and will be pinned to at most length() - pos. + * @return the substring StringPiece + * @stable ICU 4.2 + */ + StringPiece substr(int32_t pos, int32_t len = npos) const { + return StringPiece(*this, pos, len); + } +}; + +/** + * Global operator == for StringPiece + * @param x The first StringPiece to compare. + * @param y The second StringPiece to compare. + * @return true if the string data is equal + * @stable ICU 4.8 + */ +U_EXPORT UBool U_EXPORT2 +operator==(const StringPiece& x, const StringPiece& y); + +/** + * Global operator != for StringPiece + * @param x The first StringPiece to compare. + * @param y The second StringPiece to compare. + * @return true if the string data is not equal + * @stable ICU 4.8 + */ +inline bool operator!=(const StringPiece& x, const StringPiece& y) { + return !(x == y); +} + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __STRINGPIECE_H__ diff --git a/intl/icu/source/common/unicode/stringtriebuilder.h b/intl/icu/source/common/unicode/stringtriebuilder.h new file mode 100644 index 0000000000..429d7883f1 --- /dev/null +++ b/intl/icu/source/common/unicode/stringtriebuilder.h @@ -0,0 +1,426 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2012,2014, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: stringtriebuilder.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010dec24 +* created by: Markus W. Scherer +*/ + +#ifndef __STRINGTRIEBUILDER_H__ +#define __STRINGTRIEBUILDER_H__ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/uobject.h" + +/** + * \file + * \brief C++ API: Builder API for trie builders + */ + +// Forward declaration. +/// \cond +struct UHashtable; +typedef struct UHashtable UHashtable; +/// \endcond + +/** + * Build options for BytesTrieBuilder and CharsTrieBuilder. + * @stable ICU 4.8 + */ +enum UStringTrieBuildOption { + /** + * Builds a trie quickly. + * @stable ICU 4.8 + */ + USTRINGTRIE_BUILD_FAST, + /** + * Builds a trie more slowly, attempting to generate + * a shorter but equivalent serialization. + * This build option also uses more memory. + * + * This option can be effective when many integer values are the same + * and string/byte sequence suffixes can be shared. + * Runtime speed is not expected to improve. + * @stable ICU 4.8 + */ + USTRINGTRIE_BUILD_SMALL +}; + +U_NAMESPACE_BEGIN + +/** + * Base class for string trie builder classes. + * + * This class is not intended for public subclassing. + * @stable ICU 4.8 + */ +class U_COMMON_API StringTrieBuilder : public UObject { +public: +#ifndef U_HIDE_INTERNAL_API + /** @internal */ + static int32_t hashNode(const void *node); + /** @internal */ + static UBool equalNodes(const void *left, const void *right); +#endif /* U_HIDE_INTERNAL_API */ + +protected: + // Do not enclose the protected default constructor with #ifndef U_HIDE_INTERNAL_API + // or else the compiler will create a public default constructor. + /** @internal */ + StringTrieBuilder(); + /** @internal */ + virtual ~StringTrieBuilder(); + +#ifndef U_HIDE_INTERNAL_API + /** @internal */ + void createCompactBuilder(int32_t sizeGuess, UErrorCode &errorCode); + /** @internal */ + void deleteCompactBuilder(); + + /** @internal */ + void build(UStringTrieBuildOption buildOption, int32_t elementsLength, UErrorCode &errorCode); + + /** @internal */ + int32_t writeNode(int32_t start, int32_t limit, int32_t unitIndex); + /** @internal */ + int32_t writeBranchSubNode(int32_t start, int32_t limit, int32_t unitIndex, int32_t length); +#endif /* U_HIDE_INTERNAL_API */ + + class Node; + +#ifndef U_HIDE_INTERNAL_API + /** @internal */ + Node *makeNode(int32_t start, int32_t limit, int32_t unitIndex, UErrorCode &errorCode); + /** @internal */ + Node *makeBranchSubNode(int32_t start, int32_t limit, int32_t unitIndex, + int32_t length, UErrorCode &errorCode); +#endif /* U_HIDE_INTERNAL_API */ + + /** @internal */ + virtual int32_t getElementStringLength(int32_t i) const = 0; + /** @internal */ + virtual char16_t getElementUnit(int32_t i, int32_t unitIndex) const = 0; + /** @internal */ + virtual int32_t getElementValue(int32_t i) const = 0; + + // Finds the first unit index after this one where + // the first and last element have different units again. + /** @internal */ + virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t unitIndex) const = 0; + + // Number of different units at unitIndex. + /** @internal */ + virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t unitIndex) const = 0; + /** @internal */ + virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t unitIndex, int32_t count) const = 0; + /** @internal */ + virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t unitIndex, char16_t unit) const = 0; + + /** @internal */ + virtual UBool matchNodesCanHaveValues() const = 0; + + /** @internal */ + virtual int32_t getMaxBranchLinearSubNodeLength() const = 0; + /** @internal */ + virtual int32_t getMinLinearMatch() const = 0; + /** @internal */ + virtual int32_t getMaxLinearMatchLength() const = 0; + +#ifndef U_HIDE_INTERNAL_API + // max(BytesTrie::kMaxBranchLinearSubNodeLength, UCharsTrie::kMaxBranchLinearSubNodeLength). + /** @internal */ + static const int32_t kMaxBranchLinearSubNodeLength=5; + + // Maximum number of nested split-branch levels for a branch on all 2^16 possible char16_t units. + // log2(2^16/kMaxBranchLinearSubNodeLength) rounded up. + /** @internal */ + static const int32_t kMaxSplitBranchLevels=14; + + /** + * Makes sure that there is only one unique node registered that is + * equivalent to newNode. + * @param newNode Input node. The builder takes ownership. + * @param errorCode ICU in/out UErrorCode. + Set to U_MEMORY_ALLOCATION_ERROR if it was success but newNode==nullptr. + * @return newNode if it is the first of its kind, or + * an equivalent node if newNode is a duplicate. + * @internal + */ + Node *registerNode(Node *newNode, UErrorCode &errorCode); + /** + * Makes sure that there is only one unique FinalValueNode registered + * with this value. + * Avoids creating a node if the value is a duplicate. + * @param value A final value. + * @param errorCode ICU in/out UErrorCode. + Set to U_MEMORY_ALLOCATION_ERROR if it was success but newNode==nullptr. + * @return A FinalValueNode with the given value. + * @internal + */ + Node *registerFinalValue(int32_t value, UErrorCode &errorCode); +#endif /* U_HIDE_INTERNAL_API */ + + /* + * C++ note: + * registerNode() and registerFinalValue() take ownership of their input nodes, + * and only return owned nodes. + * If they see a failure UErrorCode, they will delete the input node. + * If they get a nullptr pointer, they will record a U_MEMORY_ALLOCATION_ERROR. + * If there is a failure, they return nullptr. + * + * nullptr Node pointers can be safely passed into other Nodes because + * they call the static Node::hashCode() which checks for a nullptr pointer first. + * + * Therefore, as long as builder functions register a new node, + * they need to check for failures only before explicitly dereferencing + * a Node pointer, or before setting a new UErrorCode. + */ + + // Hash set of nodes, maps from nodes to integer 1. + /** @internal */ + UHashtable *nodes; + + // Do not conditionalize the following with #ifndef U_HIDE_INTERNAL_API, + // it is needed for layout of other objects. + /** + * @internal + * \cond + */ + class Node : public UObject { + public: + Node(int32_t initialHash) : hash(initialHash), offset(0) {} + inline int32_t hashCode() const { return hash; } + // Handles node==nullptr. + static inline int32_t hashCode(const Node *node) { return node==nullptr ? 0 : node->hashCode(); } + // Base class operator==() compares the actual class types. + virtual bool operator==(const Node &other) const; + inline bool operator!=(const Node &other) const { return !operator==(other); } + /** + * Traverses the Node graph and numbers branch edges, with rightmost edges first. + * This is to avoid writing a duplicate node twice. + * + * Branch nodes in this trie data structure are not symmetric. + * Most branch edges "jump" to other nodes but the rightmost branch edges + * just continue without a jump. + * Therefore, write() must write the rightmost branch edge last + * (trie units are written backwards), and must write it at that point even if + * it is a duplicate of a node previously written elsewhere. + * + * This function visits and marks right branch edges first. + * Edges are numbered with increasingly negative values because we share the + * offset field which gets positive values when nodes are written. + * A branch edge also remembers the first number for any of its edges. + * + * When a further-left branch edge has a number in the range of the rightmost + * edge's numbers, then it will be written as part of the required right edge + * and we can avoid writing it first. + * + * After root.markRightEdgesFirst(-1) the offsets of all nodes are negative + * edge numbers. + * + * @param edgeNumber The first edge number for this node and its sub-nodes. + * @return An edge number that is at least the maximum-negative + * of the input edge number and the numbers of this node and all of its sub-nodes. + */ + virtual int32_t markRightEdgesFirst(int32_t edgeNumber); + // write() must set the offset to a positive value. + virtual void write(StringTrieBuilder &builder) = 0; + // See markRightEdgesFirst. + inline void writeUnlessInsideRightEdge(int32_t firstRight, int32_t lastRight, + StringTrieBuilder &builder) { + // Note: Edge numbers are negative, lastRight<=firstRight. + // If offset>0 then this node and its sub-nodes have been written already + // and we need not write them again. + // If this node is part of the unwritten right branch edge, + // then we wait until that is written. + if(offset<0 && (offsetA symbol table maintains two kinds of mappings. The first is + * between symbolic names and their values. For example, if the + * variable with the name "start" is set to the value "alpha" + * (perhaps, though not necessarily, through an expression such as + * "$start=alpha"), then the call lookup("start") will return the + * char[] array ['a', 'l', 'p', 'h', 'a']. + * + *

The second kind of mapping is between character values and + * UnicodeMatcher objects. This is used by RuleBasedTransliterator, + * which uses characters in the private use area to represent objects + * such as UnicodeSets. If U+E015 is mapped to the UnicodeSet [a-z], + * then lookupMatcher(0xE015) will return the UnicodeSet [a-z]. + * + *

Finally, a symbol table defines parsing behavior for symbolic + * names. All symbolic names start with the SYMBOL_REF character. + * When a parser encounters this character, it calls parseReference() + * with the position immediately following the SYMBOL_REF. The symbol + * table parses the name, if there is one, and returns it. + * + * @stable ICU 2.8 + */ +class U_COMMON_API SymbolTable /* not : public UObject because this is an interface/mixin class */ { +public: + + /** + * The character preceding a symbol reference name. + * @stable ICU 2.8 + */ + enum { SYMBOL_REF = 0x0024 /*$*/ }; + + /** + * Destructor. + * @stable ICU 2.8 + */ + virtual ~SymbolTable(); + + /** + * Lookup the characters associated with this string and return it. + * Return nullptr if no such name exists. The resultant + * string may have length zero. + * @param s the symbolic name to lookup + * @return a string containing the name's value, or nullptr if + * there is no mapping for s. + * @stable ICU 2.8 + */ + virtual const UnicodeString* lookup(const UnicodeString& s) const = 0; + + /** + * Lookup the UnicodeMatcher associated with the given character, and + * return it. Return nullptr if not found. + * @param ch a 32-bit code point from 0 to 0x10FFFF inclusive. + * @return the UnicodeMatcher object represented by the given + * character, or nullptr if there is no mapping for ch. + * @stable ICU 2.8 + */ + virtual const UnicodeFunctor* lookupMatcher(UChar32 ch) const = 0; + + /** + * Parse a symbol reference name from the given string, starting + * at the given position. If no valid symbol reference name is + * found, return the empty string and leave pos unchanged. That is, if the + * character at pos cannot start a name, or if pos is at or after + * text.length(), then return an empty string. This indicates an + * isolated SYMBOL_REF character. + * @param text the text to parse for the name + * @param pos on entry, the index of the first character to parse. + * This is the character following the SYMBOL_REF character. On + * exit, the index after the last parsed character. If the parse + * failed, pos is unchanged on exit. + * @param limit the index after the last character to be parsed. + * @return the parsed name, or an empty string if there is no + * valid symbolic name at the given position. + * @stable ICU 2.8 + */ + virtual UnicodeString parseReference(const UnicodeString& text, + ParsePosition& pos, int32_t limit) const = 0; +}; +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/ubidi.h b/intl/icu/source/common/unicode/ubidi.h new file mode 100644 index 0000000000..536f4172bc --- /dev/null +++ b/intl/icu/source/common/unicode/ubidi.h @@ -0,0 +1,2211 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2013, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: ubidi.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 1999jul27 +* created by: Markus W. Scherer, updated by Matitiahu Allouche +*/ + +#ifndef UBIDI_H +#define UBIDI_H + +#include "unicode/utypes.h" +#include "unicode/uchar.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +/** + *\file + * \brief C API: Bidi algorithm + * + *

Bidi algorithm for ICU

+ * + * This is an implementation of the Unicode Bidirectional Algorithm. + * The algorithm is defined in the + * Unicode Standard Annex #9.

+ * + * Note: Libraries that perform a bidirectional algorithm and + * reorder strings accordingly are sometimes called "Storage Layout Engines". + * ICU's Bidi and shaping (u_shapeArabic()) APIs can be used at the core of such + * "Storage Layout Engines". + * + *

General remarks about the API:

+ * + * In functions with an error code parameter, + * the pErrorCode pointer must be valid + * and the value that it points to must not indicate a failure before + * the function call. Otherwise, the function returns immediately. + * After the function call, the value indicates success or failure.

+ * + * The "limit" of a sequence of characters is the position just after their + * last character, i.e., one more than that position.

+ * + * Some of the API functions provide access to "runs". + * Such a "run" is defined as a sequence of characters + * that are at the same embedding level + * after performing the Bidi algorithm.

+ * + * @author Markus W. Scherer + * @version 1.0 + * + * + *

Sample code for the ICU Bidi API

+ * + *
Rendering a paragraph with the ICU Bidi API
+ * + * This is (hypothetical) sample code that illustrates + * how the ICU Bidi API could be used to render a paragraph of text. + * Rendering code depends highly on the graphics system, + * therefore this sample code must make a lot of assumptions, + * which may or may not match any existing graphics system's properties. + * + *

The basic assumptions are:

+ *
    + *
  • Rendering is done from left to right on a horizontal line.
    • + *
  • A run of single-style, unidirectional text can be rendered at once.
    • + *
  • Such a run of text is passed to the graphics system with + * characters (code units) in logical order.
    • + *
  • The line-breaking algorithm is very complicated + * and Locale-dependent - + * and therefore its implementation omitted from this sample code.
    • + *
+ * + * 
+ * \code
+ *#include 
+ *
+ *typedef enum {
+ *     styleNormal=0, styleSelected=1,
+ *     styleBold=2, styleItalics=4,
+ *     styleSuper=8, styleSub=16
+ *} Style;
+ *
+ *typedef struct { int32_t limit; Style style; } StyleRun;
+ *
+ *int getTextWidth(const UChar *text, int32_t start, int32_t limit,
+ *                  const StyleRun *styleRuns, int styleRunCount);
+ *
+ * // set *pLimit and *pStyleRunLimit for a line
+ * // from text[start] and from styleRuns[styleRunStart]
+ * // using ubidi_getLogicalRun(para, ...)
+ *void getLineBreak(const UChar *text, int32_t start, int32_t *pLimit,
+ *                  UBiDi *para,
+ *                  const StyleRun *styleRuns, int styleRunStart, int *pStyleRunLimit,
+ *                  int *pLineWidth);
+ *
+ * // render runs on a line sequentially, always from left to right
+ *
+ * // prepare rendering a new line
+ * void startLine(UBiDiDirection textDirection, int lineWidth);
+ *
+ * // render a run of text and advance to the right by the run width
+ * // the text[start..limit-1] is always in logical order
+ * void renderRun(const UChar *text, int32_t start, int32_t limit,
+ *               UBiDiDirection textDirection, Style style);
+ *
+ * // We could compute a cross-product
+ * // from the style runs with the directional runs
+ * // and then reorder it.
+ * // Instead, here we iterate over each run type
+ * // and render the intersections -
+ * // with shortcuts in simple (and common) cases.
+ * // renderParagraph() is the main function.
+ *
+ * // render a directional run with
+ * // (possibly) multiple style runs intersecting with it
+ * void renderDirectionalRun(const UChar *text,
+ *                           int32_t start, int32_t limit,
+ *                           UBiDiDirection direction,
+ *                           const StyleRun *styleRuns, int styleRunCount) {
+ *     int i;
+ *
+ *     // iterate over style runs
+ *     if(direction==UBIDI_LTR) {
+ *         int styleLimit;
+ *
+ *         for(i=0; ilimit) { styleLimit=limit; }
+ *                 renderRun(text, start, styleLimit,
+ *                           direction, styleRuns[i].style);
+ *                 if(styleLimit==limit) { break; }
+ *                 start=styleLimit;
+ *             }
+ *         }
+ *     } else {
+ *         int styleStart;
+ *
+ *         for(i=styleRunCount-1; i>=0; --i) {
+ *             if(i>0) {
+ *                 styleStart=styleRuns[i-1].limit;
+ *             } else {
+ *                 styleStart=0;
+ *             }
+ *             if(limit>=styleStart) {
+ *                 if(styleStart=length
+ *
+ *         width=getTextWidth(text, 0, length, styleRuns, styleRunCount);
+ *         if(width<=lineWidth) {
+ *             // everything fits onto one line
+ *
+ *            // prepare rendering a new line from either left or right
+ *             startLine(paraLevel, width);
+ *
+ *             renderLine(para, text, 0, length,
+ *                        styleRuns, styleRunCount, pErrorCode);
+ *         } else {
+ *             UBiDi *line;
+ *
+ *             // we need to render several lines
+ *             line=ubidi_openSized(length, 0, pErrorCode);
+ *             if(line!=NULL) {
+ *                 int32_t start=0, limit;
+ *                 int styleRunStart=0, styleRunLimit;
+ *
+ *                 for(;;) {
+ *                     limit=length;
+ *                     styleRunLimit=styleRunCount;
+ *                     getLineBreak(text, start, &limit, para,
+ *                                  styleRuns, styleRunStart, &styleRunLimit,
+ *                                 &width);
+ *                     ubidi_setLine(para, start, limit, line, pErrorCode);
+ *                     if(U_SUCCESS(*pErrorCode)) {
+ *                         // prepare rendering a new line
+ *                         // from either left or right
+ *                         startLine(paraLevel, width);
+ *
+ *                         renderLine(line, text, start, limit,
+ *                                    styleRuns+styleRunStart,
+ *                                    styleRunLimit-styleRunStart, pErrorCode);
+ *                     }
+ *                     if(limit==length) { break; }
+ *                     start=limit;
+ *                     styleRunStart=styleRunLimit-1;
+ *                     if(start>=styleRuns[styleRunStart].limit) {
+ *                         ++styleRunStart;
+ *                     }
+ *                 }
+ *
+ *                 ubidi_close(line);
+ *             }
+ *        }
+ *    }
+ *
+ *     ubidi_close(para);
+ *}
+ *\endcode
+ *
+ */ + +/*DOCXX_TAG*/ +/*@{*/ + +/** + * UBiDiLevel is the type of the level values in this + * Bidi implementation. + * It holds an embedding level and indicates the visual direction + * by its bit 0 (even/odd value).

+ * + * It can also hold non-level values for the + * paraLevel and embeddingLevels + * arguments of ubidi_setPara(); there: + *

    + *
  • bit 7 of an embeddingLevels[] + * value indicates whether the using application is + * specifying the level of a character to override whatever the + * Bidi implementation would resolve it to.
    • + *
  • paraLevel can be set to the + * pseudo-level values UBIDI_DEFAULT_LTR + * and UBIDI_DEFAULT_RTL.
    • + *
+ * + * @see ubidi_setPara + * + *

The related constants are not real, valid level values. + * UBIDI_DEFAULT_XXX can be used to specify + * a default for the paragraph level for + * when the ubidi_setPara() function + * shall determine it but there is no + * strongly typed character in the input.

+ * + * Note that the value for UBIDI_DEFAULT_LTR is even + * and the one for UBIDI_DEFAULT_RTL is odd, + * just like with normal LTR and RTL level values - + * these special values are designed that way. Also, the implementation + * assumes that UBIDI_MAX_EXPLICIT_LEVEL is odd. + * + * Note: The numeric values of the related constants will not change: + * They are tied to the use of 7-bit byte values (plus the override bit) + * and of the UBiDiLevel=uint8_t data type in this API. + * + * @see UBIDI_DEFAULT_LTR + * @see UBIDI_DEFAULT_RTL + * @see UBIDI_LEVEL_OVERRIDE + * @see UBIDI_MAX_EXPLICIT_LEVEL + * @stable ICU 2.0 + */ +typedef uint8_t UBiDiLevel; + +/** Paragraph level setting.

+ * + * Constant indicating that the base direction depends on the first strong + * directional character in the text according to the Unicode Bidirectional + * Algorithm. If no strong directional character is present, + * then set the paragraph level to 0 (left-to-right).

+ * + * If this value is used in conjunction with reordering modes + * UBIDI_REORDER_INVERSE_LIKE_DIRECT or + * UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL, the text to reorder + * is assumed to be visual LTR, and the text after reordering is required + * to be the corresponding logical string with appropriate contextual + * direction. The direction of the result string will be RTL if either + * the righmost or leftmost strong character of the source text is RTL + * or Arabic Letter, the direction will be LTR otherwise.

+ * + * If reordering option UBIDI_OPTION_INSERT_MARKS is set, an RLM may + * be added at the beginning of the result string to ensure round trip + * (that the result string, when reordered back to visual, will produce + * the original source text). + * @see UBIDI_REORDER_INVERSE_LIKE_DIRECT + * @see UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL + * @stable ICU 2.0 + */ +#define UBIDI_DEFAULT_LTR 0xfe + +/** Paragraph level setting.

+ * + * Constant indicating that the base direction depends on the first strong + * directional character in the text according to the Unicode Bidirectional + * Algorithm. If no strong directional character is present, + * then set the paragraph level to 1 (right-to-left).

+ * + * If this value is used in conjunction with reordering modes + * UBIDI_REORDER_INVERSE_LIKE_DIRECT or + * UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL, the text to reorder + * is assumed to be visual LTR, and the text after reordering is required + * to be the corresponding logical string with appropriate contextual + * direction. The direction of the result string will be RTL if either + * the righmost or leftmost strong character of the source text is RTL + * or Arabic Letter, or if the text contains no strong character; + * the direction will be LTR otherwise.

+ * + * If reordering option UBIDI_OPTION_INSERT_MARKS is set, an RLM may + * be added at the beginning of the result string to ensure round trip + * (that the result string, when reordered back to visual, will produce + * the original source text). + * @see UBIDI_REORDER_INVERSE_LIKE_DIRECT + * @see UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL + * @stable ICU 2.0 + */ +#define UBIDI_DEFAULT_RTL 0xff + +/** + * Maximum explicit embedding level. + * Same as the max_depth value in the + * Unicode Bidirectional Algorithm. + * (The maximum resolved level can be up to UBIDI_MAX_EXPLICIT_LEVEL+1). + * @stable ICU 2.0 + */ +#define UBIDI_MAX_EXPLICIT_LEVEL 125 + +/** Bit flag for level input. + * Overrides directional properties. + * @stable ICU 2.0 + */ +#define UBIDI_LEVEL_OVERRIDE 0x80 + +/** + * Special value which can be returned by the mapping functions when a logical + * index has no corresponding visual index or vice-versa. This may happen + * for the logical-to-visual mapping of a Bidi control when option + * #UBIDI_OPTION_REMOVE_CONTROLS is specified. This can also happen + * for the visual-to-logical mapping of a Bidi mark (LRM or RLM) inserted + * by option #UBIDI_OPTION_INSERT_MARKS. + * @see ubidi_getVisualIndex + * @see ubidi_getVisualMap + * @see ubidi_getLogicalIndex + * @see ubidi_getLogicalMap + * @stable ICU 3.6 + */ +#define UBIDI_MAP_NOWHERE (-1) + +/** + * UBiDiDirection values indicate the text direction. + * @stable ICU 2.0 + */ +enum UBiDiDirection { + /** Left-to-right text. This is a 0 value. + *

    + *
  • As return value for ubidi_getDirection(), it means + * that the source string contains no right-to-left characters, or + * that the source string is empty and the paragraph level is even. + *
  • As return value for ubidi_getBaseDirection(), it + * means that the first strong character of the source string has + * a left-to-right direction. + *
+ * @stable ICU 2.0 + */ + UBIDI_LTR, + /** Right-to-left text. This is a 1 value. + *
    + *
  • As return value for ubidi_getDirection(), it means + * that the source string contains no left-to-right characters, or + * that the source string is empty and the paragraph level is odd. + *
  • As return value for ubidi_getBaseDirection(), it + * means that the first strong character of the source string has + * a right-to-left direction. + *
+ * @stable ICU 2.0 + */ + UBIDI_RTL, + /** Mixed-directional text. + *

As return value for ubidi_getDirection(), it means + * that the source string contains both left-to-right and + * right-to-left characters. + * @stable ICU 2.0 + */ + UBIDI_MIXED, + /** No strongly directional text. + *

As return value for ubidi_getBaseDirection(), it means + * that the source string is missing or empty, or contains neither left-to-right + * nor right-to-left characters. + * @stable ICU 4.6 + */ + UBIDI_NEUTRAL +}; + +/** @stable ICU 2.0 */ +typedef enum UBiDiDirection UBiDiDirection; + +/** + * Forward declaration of the UBiDi structure for the declaration of + * the API functions. Its fields are implementation-specific.

+ * This structure holds information about a paragraph (or multiple paragraphs) + * of text with Bidi-algorithm-related details, or about one line of + * such a paragraph.

+ * Reordering can be done on a line, or on one or more paragraphs which are + * then interpreted each as one single line. + * @stable ICU 2.0 + */ +struct UBiDi; + +/** @stable ICU 2.0 */ +typedef struct UBiDi UBiDi; + +/** + * Allocate a UBiDi structure. + * Such an object is initially empty. It is assigned + * the Bidi properties of a piece of text containing one or more paragraphs + * by ubidi_setPara() + * or the Bidi properties of a line within a paragraph by + * ubidi_setLine().

+ * This object can be reused for as long as it is not deallocated + * by calling ubidi_close().

+ * ubidi_setPara() and ubidi_setLine() will allocate + * additional memory for internal structures as necessary. + * + * @return An empty UBiDi object. + * @stable ICU 2.0 + */ +U_CAPI UBiDi * U_EXPORT2 +ubidi_open(void); + +/** + * Allocate a UBiDi structure with preallocated memory + * for internal structures. + * This function provides a UBiDi object like ubidi_open() + * with no arguments, but it also preallocates memory for internal structures + * according to the sizings supplied by the caller.

+ * Subsequent functions will not allocate any more memory, and are thus + * guaranteed not to fail because of lack of memory.

+ * The preallocation can be limited to some of the internal memory + * by setting some values to 0 here. That means that if, e.g., + * maxRunCount cannot be reasonably predetermined and should not + * be set to maxLength (the only failproof value) to avoid + * wasting memory, then maxRunCount could be set to 0 here + * and the internal structures that are associated with it will be allocated + * on demand, just like with ubidi_open(). + * + * @param maxLength is the maximum text or line length that internal memory + * will be preallocated for. An attempt to associate this object with a + * longer text will fail, unless this value is 0, which leaves the allocation + * up to the implementation. + * + * @param maxRunCount is the maximum anticipated number of same-level runs + * that internal memory will be preallocated for. An attempt to access + * visual runs on an object that was not preallocated for as many runs + * as the text was actually resolved to will fail, + * unless this value is 0, which leaves the allocation up to the implementation.

+ * The number of runs depends on the actual text and maybe anywhere between + * 1 and maxLength. It is typically small. + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @return An empty UBiDi object with preallocated memory. + * @stable ICU 2.0 + */ +U_CAPI UBiDi * U_EXPORT2 +ubidi_openSized(int32_t maxLength, int32_t maxRunCount, UErrorCode *pErrorCode); + +/** + * ubidi_close() must be called to free the memory + * associated with a UBiDi object.

+ * + * Important: + * A parent UBiDi object must not be destroyed or reused if + * it still has children. + * If a UBiDi object has become the child + * of another one (its parent) by calling + * ubidi_setLine(), then the child object must + * be destroyed (closed) or reused (by calling + * ubidi_setPara() or ubidi_setLine()) + * before the parent object. + * + * @param pBiDi is a UBiDi object. + * + * @see ubidi_setPara + * @see ubidi_setLine + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ubidi_close(UBiDi *pBiDi); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUBiDiPointer + * "Smart pointer" class, closes a UBiDi via ubidi_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUBiDiPointer, UBiDi, ubidi_close); + +U_NAMESPACE_END + +#endif + +/** + * Modify the operation of the Bidi algorithm such that it + * approximates an "inverse Bidi" algorithm. This function + * must be called before ubidi_setPara(). + * + *

The normal operation of the Bidi algorithm as described + * in the Unicode Technical Report is to take text stored in logical + * (keyboard, typing) order and to determine the reordering of it for visual + * rendering. + * Some legacy systems store text in visual order, and for operations + * with standard, Unicode-based algorithms, the text needs to be transformed + * to logical order. This is effectively the inverse algorithm of the + * described Bidi algorithm. Note that there is no standard algorithm for + * this "inverse Bidi" and that the current implementation provides only an + * approximation of "inverse Bidi".

+ * + *

With isInverse set to true, + * this function changes the behavior of some of the subsequent functions + * in a way that they can be used for the inverse Bidi algorithm. + * Specifically, runs of text with numeric characters will be treated in a + * special way and may need to be surrounded with LRM characters when they are + * written in reordered sequence.

+ * + *

Output runs should be retrieved using ubidi_getVisualRun(). + * Since the actual input for "inverse Bidi" is visually ordered text and + * ubidi_getVisualRun() gets the reordered runs, these are actually + * the runs of the logically ordered output.

+ * + *

Calling this function with argument isInverse set to + * true is equivalent to calling + * ubidi_setReorderingMode with argument + * reorderingMode + * set to #UBIDI_REORDER_INVERSE_NUMBERS_AS_L.
+ * Calling this function with argument isInverse set to + * false is equivalent to calling + * ubidi_setReorderingMode with argument + * reorderingMode + * set to #UBIDI_REORDER_DEFAULT. + * + * @param pBiDi is a UBiDi object. + * + * @param isInverse specifies "forward" or "inverse" Bidi operation. + * + * @see ubidi_setPara + * @see ubidi_writeReordered + * @see ubidi_setReorderingMode + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ubidi_setInverse(UBiDi *pBiDi, UBool isInverse); + +/** + * Is this Bidi object set to perform the inverse Bidi algorithm? + *

Note: calling this function after setting the reordering mode with + * ubidi_setReorderingMode will return true if the + * reordering mode was set to #UBIDI_REORDER_INVERSE_NUMBERS_AS_L, + * false for all other values.

+ * + * @param pBiDi is a UBiDi object. + * @return true if the Bidi object is set to perform the inverse Bidi algorithm + * by handling numbers as L. + * + * @see ubidi_setInverse + * @see ubidi_setReorderingMode + * @stable ICU 2.0 + */ + +U_CAPI UBool U_EXPORT2 +ubidi_isInverse(UBiDi *pBiDi); + +/** + * Specify whether block separators must be allocated level zero, + * so that successive paragraphs will progress from left to right. + * This function must be called before ubidi_setPara(). + * Paragraph separators (B) may appear in the text. Setting them to level zero + * means that all paragraph separators (including one possibly appearing + * in the last text position) are kept in the reordered text after the text + * that they follow in the source text. + * When this feature is not enabled, a paragraph separator at the last + * position of the text before reordering will go to the first position + * of the reordered text when the paragraph level is odd. + * + * @param pBiDi is a UBiDi object. + * + * @param orderParagraphsLTR specifies whether paragraph separators (B) must + * receive level 0, so that successive paragraphs progress from left to right. + * + * @see ubidi_setPara + * @stable ICU 3.4 + */ +U_CAPI void U_EXPORT2 +ubidi_orderParagraphsLTR(UBiDi *pBiDi, UBool orderParagraphsLTR); + +/** + * Is this Bidi object set to allocate level 0 to block separators so that + * successive paragraphs progress from left to right? + * + * @param pBiDi is a UBiDi object. + * @return true if the Bidi object is set to allocate level 0 to block + * separators. + * + * @see ubidi_orderParagraphsLTR + * @stable ICU 3.4 + */ +U_CAPI UBool U_EXPORT2 +ubidi_isOrderParagraphsLTR(UBiDi *pBiDi); + +/** + * UBiDiReorderingMode values indicate which variant of the Bidi + * algorithm to use. + * + * @see ubidi_setReorderingMode + * @stable ICU 3.6 + */ +typedef enum UBiDiReorderingMode { + /** Regular Logical to Visual Bidi algorithm according to Unicode. + * This is a 0 value. + * @stable ICU 3.6 */ + UBIDI_REORDER_DEFAULT = 0, + /** Logical to Visual algorithm which handles numbers in a way which + * mimics the behavior of Windows XP. + * @stable ICU 3.6 */ + UBIDI_REORDER_NUMBERS_SPECIAL, + /** Logical to Visual algorithm grouping numbers with adjacent R characters + * (reversible algorithm). + * @stable ICU 3.6 */ + UBIDI_REORDER_GROUP_NUMBERS_WITH_R, + /** Reorder runs only to transform a Logical LTR string to the Logical RTL + * string with the same display, or vice-versa.
+ * If this mode is set together with option + * #UBIDI_OPTION_INSERT_MARKS, some Bidi controls in the source + * text may be removed and other controls may be added to produce the + * minimum combination which has the required display. + * @stable ICU 3.6 */ + UBIDI_REORDER_RUNS_ONLY, + /** Visual to Logical algorithm which handles numbers like L + * (same algorithm as selected by ubidi_setInverse(true). + * @see ubidi_setInverse + * @stable ICU 3.6 */ + UBIDI_REORDER_INVERSE_NUMBERS_AS_L, + /** Visual to Logical algorithm equivalent to the regular Logical to Visual + * algorithm. + * @stable ICU 3.6 */ + UBIDI_REORDER_INVERSE_LIKE_DIRECT, + /** Inverse Bidi (Visual to Logical) algorithm for the + * UBIDI_REORDER_NUMBERS_SPECIAL Bidi algorithm. + * @stable ICU 3.6 */ + UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL, +#ifndef U_HIDE_DEPRECATED_API + /** + * Number of values for reordering mode. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + UBIDI_REORDER_COUNT +#endif // U_HIDE_DEPRECATED_API +} UBiDiReorderingMode; + +/** + * Modify the operation of the Bidi algorithm such that it implements some + * variant to the basic Bidi algorithm or approximates an "inverse Bidi" + * algorithm, depending on different values of the "reordering mode". + * This function must be called before ubidi_setPara(), and stays + * in effect until called again with a different argument. + * + *

The normal operation of the Bidi algorithm as described + * in the Unicode Standard Annex #9 is to take text stored in logical + * (keyboard, typing) order and to determine how to reorder it for visual + * rendering.

+ * + *

With the reordering mode set to a value other than + * #UBIDI_REORDER_DEFAULT, this function changes the behavior of + * some of the subsequent functions in a way such that they implement an + * inverse Bidi algorithm or some other algorithm variants.

+ * + *

Some legacy systems store text in visual order, and for operations + * with standard, Unicode-based algorithms, the text needs to be transformed + * into logical order. This is effectively the inverse algorithm of the + * described Bidi algorithm. Note that there is no standard algorithm for + * this "inverse Bidi", so a number of variants are implemented here.

+ * + *

In other cases, it may be desirable to emulate some variant of the + * Logical to Visual algorithm (e.g. one used in MS Windows), or perform a + * Logical to Logical transformation.

+ * + *
    + *
  • When the reordering mode is set to #UBIDI_REORDER_DEFAULT, + * the standard Bidi Logical to Visual algorithm is applied.
    • + * + *
  • When the reordering mode is set to + * #UBIDI_REORDER_NUMBERS_SPECIAL, + * the algorithm used to perform Bidi transformations when calling + * ubidi_setPara should approximate the algorithm used in + * Microsoft Windows XP rather than strictly conform to the Unicode Bidi + * algorithm. + *
    + * The differences between the basic algorithm and the algorithm addressed + * by this option are as follows: + *
      + *
    • Within text at an even embedding level, the sequence "123AB" + * (where AB represent R or AL letters) is transformed to "123BA" by the + * Unicode algorithm and to "BA123" by the Windows algorithm.
      • + *
    • Arabic-Indic numbers (AN) are handled by the Windows algorithm just + * like regular numbers (EN).
      • + *
    • + * + *
  • When the reordering mode is set to + * #UBIDI_REORDER_GROUP_NUMBERS_WITH_R, + * numbers located between LTR text and RTL text are associated with the RTL + * text. For instance, an LTR paragraph with content "abc 123 DEF" (where + * upper case letters represent RTL characters) will be transformed to + * "abc FED 123" (and not "abc 123 FED"), "DEF 123 abc" will be transformed + * to "123 FED abc" and "123 FED abc" will be transformed to "DEF 123 abc". + * This makes the algorithm reversible and makes it useful when round trip + * (from visual to logical and back to visual) must be achieved without + * adding LRM characters. However, this is a variation from the standard + * Unicode Bidi algorithm.
    + * The source text should not contain Bidi control characters other than LRM + * or RLM.
    • + * + *
  • When the reordering mode is set to + * #UBIDI_REORDER_RUNS_ONLY, + * a "Logical to Logical" transformation must be performed: + *
      + *
    • If the default text level of the source text (argument paraLevel + * in ubidi_setPara) is even, the source text will be handled as + * LTR logical text and will be transformed to the RTL logical text which has + * the same LTR visual display.
      • + *
    • If the default level of the source text is odd, the source text + * will be handled as RTL logical text and will be transformed to the + * LTR logical text which has the same LTR visual display.
      • + *
    + * This mode may be needed when logical text which is basically Arabic or + * Hebrew, with possible included numbers or phrases in English, has to be + * displayed as if it had an even embedding level (this can happen if the + * displaying application treats all text as if it was basically LTR). + *
    + * This mode may also be needed in the reverse case, when logical text which is + * basically English, with possible included phrases in Arabic or Hebrew, has to + * be displayed as if it had an odd embedding level. + *
    + * Both cases could be handled by adding LRE or RLE at the head of the text, + * if the display subsystem supports these formatting controls. If it does not, + * the problem may be handled by transforming the source text in this mode + * before displaying it, so that it will be displayed properly.
    + * The source text should not contain Bidi control characters other than LRM + * or RLM.
    • + * + *
  • When the reordering mode is set to + * #UBIDI_REORDER_INVERSE_NUMBERS_AS_L, an "inverse Bidi" algorithm + * is applied. + * Runs of text with numeric characters will be treated like LTR letters and + * may need to be surrounded with LRM characters when they are written in + * reordered sequence (the option #UBIDI_INSERT_LRM_FOR_NUMERIC can + * be used with function ubidi_writeReordered to this end. This + * mode is equivalent to calling ubidi_setInverse() with + * argument isInverse set to true.
    • + * + *
  • When the reordering mode is set to + * #UBIDI_REORDER_INVERSE_LIKE_DIRECT, the "direct" Logical to Visual + * Bidi algorithm is used as an approximation of an "inverse Bidi" algorithm. + * This mode is similar to mode #UBIDI_REORDER_INVERSE_NUMBERS_AS_L + * but is closer to the regular Bidi algorithm. + *
    + * For example, an LTR paragraph with the content "FED 123 456 CBA" (where + * upper case represents RTL characters) will be transformed to + * "ABC 456 123 DEF", as opposed to "DEF 123 456 ABC" + * with mode UBIDI_REORDER_INVERSE_NUMBERS_AS_L.
    + * When used in conjunction with option + * #UBIDI_OPTION_INSERT_MARKS, this mode generally + * adds Bidi marks to the output significantly more sparingly than mode + * #UBIDI_REORDER_INVERSE_NUMBERS_AS_L with option + * #UBIDI_INSERT_LRM_FOR_NUMERIC in calls to + * ubidi_writeReordered.
    • + * + *
  • When the reordering mode is set to + * #UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL, the Logical to Visual + * Bidi algorithm used in Windows XP is used as an approximation of an "inverse Bidi" algorithm. + *
    + * For example, an LTR paragraph with the content "abc FED123" (where + * upper case represents RTL characters) will be transformed to "abc 123DEF."
    • + *
+ * + *

In all the reordering modes specifying an "inverse Bidi" algorithm + * (i.e. those with a name starting with UBIDI_REORDER_INVERSE), + * output runs should be retrieved using + * ubidi_getVisualRun(), and the output text with + * ubidi_writeReordered(). The caller should keep in mind that in + * "inverse Bidi" modes the input is actually visually ordered text and + * reordered output returned by ubidi_getVisualRun() or + * ubidi_writeReordered() are actually runs or character string + * of logically ordered output.
+ * For all the "inverse Bidi" modes, the source text should not contain + * Bidi control characters other than LRM or RLM.

+ * + *

Note that option #UBIDI_OUTPUT_REVERSE of + * ubidi_writeReordered has no useful meaning and should not be + * used in conjunction with any value of the reordering mode specifying + * "inverse Bidi" or with value UBIDI_REORDER_RUNS_ONLY. + * + * @param pBiDi is a UBiDi object. + * @param reorderingMode specifies the required variant of the Bidi algorithm. + * + * @see UBiDiReorderingMode + * @see ubidi_setInverse + * @see ubidi_setPara + * @see ubidi_writeReordered + * @stable ICU 3.6 + */ +U_CAPI void U_EXPORT2 +ubidi_setReorderingMode(UBiDi *pBiDi, UBiDiReorderingMode reorderingMode); + +/** + * What is the requested reordering mode for a given Bidi object? + * + * @param pBiDi is a UBiDi object. + * @return the current reordering mode of the Bidi object + * @see ubidi_setReorderingMode + * @stable ICU 3.6 + */ +U_CAPI UBiDiReorderingMode U_EXPORT2 +ubidi_getReorderingMode(UBiDi *pBiDi); + +/** + * UBiDiReorderingOption values indicate which options are + * specified to affect the Bidi algorithm. + * + * @see ubidi_setReorderingOptions + * @stable ICU 3.6 + */ +typedef enum UBiDiReorderingOption { + /** + * option value for ubidi_setReorderingOptions: + * disable all the options which can be set with this function + * @see ubidi_setReorderingOptions + * @stable ICU 3.6 + */ + UBIDI_OPTION_DEFAULT = 0, + + /** + * option bit for ubidi_setReorderingOptions: + * insert Bidi marks (LRM or RLM) when needed to ensure correct result of + * a reordering to a Logical order + * + *

This option must be set or reset before calling + * ubidi_setPara.

+ * + *

This option is significant only with reordering modes which generate + * a result with Logical order, specifically:

+ *
    + *
  • #UBIDI_REORDER_RUNS_ONLY
    • + *
  • #UBIDI_REORDER_INVERSE_NUMBERS_AS_L
    • + *
  • #UBIDI_REORDER_INVERSE_LIKE_DIRECT
    • + *
  • #UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL
    • + *
+ * + *

If this option is set in conjunction with reordering mode + * #UBIDI_REORDER_INVERSE_NUMBERS_AS_L or with calling + * ubidi_setInverse(true), it implies + * option #UBIDI_INSERT_LRM_FOR_NUMERIC + * in calls to function ubidi_writeReordered().

+ * + *

For other reordering modes, a minimum number of LRM or RLM characters + * will be added to the source text after reordering it so as to ensure + * round trip, i.e. when applying the inverse reordering mode on the + * resulting logical text with removal of Bidi marks + * (option #UBIDI_OPTION_REMOVE_CONTROLS set before calling + * ubidi_setPara() or option #UBIDI_REMOVE_BIDI_CONTROLS + * in ubidi_writeReordered), the result will be identical to the + * source text in the first transformation. + * + *

This option will be ignored if specified together with option + * #UBIDI_OPTION_REMOVE_CONTROLS. It inhibits option + * UBIDI_REMOVE_BIDI_CONTROLS in calls to function + * ubidi_writeReordered() and it implies option + * #UBIDI_INSERT_LRM_FOR_NUMERIC in calls to function + * ubidi_writeReordered() if the reordering mode is + * #UBIDI_REORDER_INVERSE_NUMBERS_AS_L.

+ * + * @see ubidi_setReorderingMode + * @see ubidi_setReorderingOptions + * @stable ICU 3.6 + */ + UBIDI_OPTION_INSERT_MARKS = 1, + + /** + * option bit for ubidi_setReorderingOptions: + * remove Bidi control characters + * + *

This option must be set or reset before calling + * ubidi_setPara.

+ * + *

This option nullifies option #UBIDI_OPTION_INSERT_MARKS. + * It inhibits option #UBIDI_INSERT_LRM_FOR_NUMERIC in calls + * to function ubidi_writeReordered() and it implies option + * #UBIDI_REMOVE_BIDI_CONTROLS in calls to that function.

+ * + * @see ubidi_setReorderingMode + * @see ubidi_setReorderingOptions + * @stable ICU 3.6 + */ + UBIDI_OPTION_REMOVE_CONTROLS = 2, + + /** + * option bit for ubidi_setReorderingOptions: + * process the output as part of a stream to be continued + * + *

This option must be set or reset before calling + * ubidi_setPara.

+ * + *

This option specifies that the caller is interested in processing large + * text object in parts. + * The results of the successive calls are expected to be concatenated by the + * caller. Only the call for the last part will have this option bit off.

+ * + *

When this option bit is on, ubidi_setPara() may process + * less than the full source text in order to truncate the text at a meaningful + * boundary. The caller should call ubidi_getProcessedLength() + * immediately after calling ubidi_setPara() in order to + * determine how much of the source text has been processed. + * Source text beyond that length should be resubmitted in following calls to + * ubidi_setPara. The processed length may be less than + * the length of the source text if a character preceding the last character of + * the source text constitutes a reasonable boundary (like a block separator) + * for text to be continued.
+ * If the last character of the source text constitutes a reasonable + * boundary, the whole text will be processed at once.
+ * If nowhere in the source text there exists + * such a reasonable boundary, the processed length will be zero.
+ * The caller should check for such an occurrence and do one of the following: + *

  • submit a larger amount of text with a better chance to include + * a reasonable boundary.
    • + *
  • resubmit the same text after turning off option + * UBIDI_OPTION_STREAMING.
+ * In all cases, this option should be turned off before processing the last + * part of the text.

+ * + *

When the UBIDI_OPTION_STREAMING option is used, + * it is recommended to call ubidi_orderParagraphsLTR() with + * argument orderParagraphsLTR set to true before + * calling ubidi_setPara so that later paragraphs may be + * concatenated to previous paragraphs on the right.

+ * + * @see ubidi_setReorderingMode + * @see ubidi_setReorderingOptions + * @see ubidi_getProcessedLength + * @see ubidi_orderParagraphsLTR + * @stable ICU 3.6 + */ + UBIDI_OPTION_STREAMING = 4 +} UBiDiReorderingOption; + +/** + * Specify which of the reordering options + * should be applied during Bidi transformations. + * + * @param pBiDi is a UBiDi object. + * @param reorderingOptions is a combination of zero or more of the following + * options: + * #UBIDI_OPTION_DEFAULT, #UBIDI_OPTION_INSERT_MARKS, + * #UBIDI_OPTION_REMOVE_CONTROLS, #UBIDI_OPTION_STREAMING. + * + * @see ubidi_getReorderingOptions + * @stable ICU 3.6 + */ +U_CAPI void U_EXPORT2 +ubidi_setReorderingOptions(UBiDi *pBiDi, uint32_t reorderingOptions); + +/** + * What are the reordering options applied to a given Bidi object? + * + * @param pBiDi is a UBiDi object. + * @return the current reordering options of the Bidi object + * @see ubidi_setReorderingOptions + * @stable ICU 3.6 + */ +U_CAPI uint32_t U_EXPORT2 +ubidi_getReorderingOptions(UBiDi *pBiDi); + +/** + * Set the context before a call to ubidi_setPara().

+ * + * ubidi_setPara() computes the left-right directionality for a given piece + * of text which is supplied as one of its arguments. Sometimes this piece + * of text (the "main text") should be considered in context, because text + * appearing before ("prologue") and/or after ("epilogue") the main text + * may affect the result of this computation.

+ * + * This function specifies the prologue and/or the epilogue for the next + * call to ubidi_setPara(). The characters specified as prologue and + * epilogue should not be modified by the calling program until the call + * to ubidi_setPara() has returned. If successive calls to ubidi_setPara() + * all need specification of a context, ubidi_setContext() must be called + * before each call to ubidi_setPara(). In other words, a context is not + * "remembered" after the following successful call to ubidi_setPara().

+ * + * If a call to ubidi_setPara() specifies UBIDI_DEFAULT_LTR or + * UBIDI_DEFAULT_RTL as paraLevel and is preceded by a call to + * ubidi_setContext() which specifies a prologue, the paragraph level will + * be computed taking in consideration the text in the prologue.

+ * + * When ubidi_setPara() is called without a previous call to + * ubidi_setContext, the main text is handled as if preceded and followed + * by strong directional characters at the current paragraph level. + * Calling ubidi_setContext() with specification of a prologue will change + * this behavior by handling the main text as if preceded by the last + * strong character appearing in the prologue, if any. + * Calling ubidi_setContext() with specification of an epilogue will change + * the behavior of ubidi_setPara() by handling the main text as if followed + * by the first strong character or digit appearing in the epilogue, if any.

+ * + * Note 1: if ubidi_setContext is called repeatedly without + * calling ubidi_setPara, the earlier calls have no effect, + * only the last call will be remembered for the next call to + * ubidi_setPara.

+ * + * Note 2: calling ubidi_setContext(pBiDi, NULL, 0, NULL, 0, &errorCode) + * cancels any previous setting of non-empty prologue or epilogue. + * The next call to ubidi_setPara() will process no + * prologue or epilogue.

+ * + * Note 3: users must be aware that even after setting the context + * before a call to ubidi_setPara() to perform e.g. a logical to visual + * transformation, the resulting string may not be identical to what it + * would have been if all the text, including prologue and epilogue, had + * been processed together.
+ * Example (upper case letters represent RTL characters):
+ *   prologue = "abc DE"
+ *   epilogue = none
+ *   main text = "FGH xyz"
+ *   paraLevel = UBIDI_LTR
+ *   display without prologue = "HGF xyz" + * ("HGF" is adjacent to "xyz")
+ *   display with prologue = "abc HGFED xyz" + * ("HGF" is not adjacent to "xyz")
+ * + * @param pBiDi is a paragraph UBiDi object. + * + * @param prologue is a pointer to the text which precedes the text that + * will be specified in a coming call to ubidi_setPara(). + * If there is no prologue to consider, then proLength + * must be zero and this pointer can be NULL. + * + * @param proLength is the length of the prologue; if proLength==-1 + * then the prologue must be zero-terminated. + * Otherwise proLength must be >= 0. If proLength==0, it means + * that there is no prologue to consider. + * + * @param epilogue is a pointer to the text which follows the text that + * will be specified in a coming call to ubidi_setPara(). + * If there is no epilogue to consider, then epiLength + * must be zero and this pointer can be NULL. + * + * @param epiLength is the length of the epilogue; if epiLength==-1 + * then the epilogue must be zero-terminated. + * Otherwise epiLength must be >= 0. If epiLength==0, it means + * that there is no epilogue to consider. + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @see ubidi_setPara + * @stable ICU 4.8 + */ +U_CAPI void U_EXPORT2 +ubidi_setContext(UBiDi *pBiDi, + const UChar *prologue, int32_t proLength, + const UChar *epilogue, int32_t epiLength, + UErrorCode *pErrorCode); + +/** + * Perform the Unicode Bidi algorithm. It is defined in the + * Unicode Standard Annex #9, + * version 13, + * also described in The Unicode Standard, Version 4.0 .

+ * + * This function takes a piece of plain text containing one or more paragraphs, + * with or without externally specified embedding levels from styled + * text and computes the left-right-directionality of each character.

+ * + * If the entire text is all of the same directionality, then + * the function may not perform all the steps described by the algorithm, + * i.e., some levels may not be the same as if all steps were performed. + * This is not relevant for unidirectional text.
+ * For example, in pure LTR text with numbers the numbers would get + * a resolved level of 2 higher than the surrounding text according to + * the algorithm. This implementation may set all resolved levels to + * the same value in such a case.

+ * + * The text can be composed of multiple paragraphs. Occurrence of a block + * separator in the text terminates a paragraph, and whatever comes next starts + * a new paragraph. The exception to this rule is when a Carriage Return (CR) + * is followed by a Line Feed (LF). Both CR and LF are block separators, but + * in that case, the pair of characters is considered as terminating the + * preceding paragraph, and a new paragraph will be started by a character + * coming after the LF. + * + * @param pBiDi A UBiDi object allocated with ubidi_open() + * which will be set to contain the reordering information, + * especially the resolved levels for all the characters in text. + * + * @param text is a pointer to the text that the Bidi algorithm will be performed on. + * This pointer is stored in the UBiDi object and can be retrieved + * with ubidi_getText().
+ * Note: the text must be (at least) length long. + * + * @param length is the length of the text; if length==-1 then + * the text must be zero-terminated. + * + * @param paraLevel specifies the default level for the text; + * it is typically 0 (LTR) or 1 (RTL). + * If the function shall determine the paragraph level from the text, + * then paraLevel can be set to + * either #UBIDI_DEFAULT_LTR + * or #UBIDI_DEFAULT_RTL; if the text contains multiple + * paragraphs, the paragraph level shall be determined separately for + * each paragraph; if a paragraph does not include any strongly typed + * character, then the desired default is used (0 for LTR or 1 for RTL). + * Any other value between 0 and #UBIDI_MAX_EXPLICIT_LEVEL + * is also valid, with odd levels indicating RTL. + * + * @param embeddingLevels (in) may be used to preset the embedding and override levels, + * ignoring characters like LRE and PDF in the text. + * A level overrides the directional property of its corresponding + * (same index) character if the level has the + * #UBIDI_LEVEL_OVERRIDE bit set.

+ * Aside from that bit, it must be + * paraLevel<=embeddingLevels[]<=UBIDI_MAX_EXPLICIT_LEVEL, + * except that level 0 is always allowed. + * Level 0 for a paragraph separator prevents reordering of paragraphs; + * this only works reliably if #UBIDI_LEVEL_OVERRIDE + * is also set for paragraph separators. + * Level 0 for other characters is treated as a wildcard + * and is lifted up to the resolved level of the surrounding paragraph.

+ * Caution: A copy of this pointer, not of the levels, + * will be stored in the UBiDi object; + * the embeddingLevels array must not be + * deallocated before the UBiDi structure is destroyed or reused, + * and the embeddingLevels + * should not be modified to avoid unexpected results on subsequent Bidi operations. + * However, the ubidi_setPara() and + * ubidi_setLine() functions may modify some or all of the levels.

+ * After the UBiDi object is reused or destroyed, the caller + * must take care of the deallocation of the embeddingLevels array.

+ * Note: the embeddingLevels array must be + * at least length long. + * This pointer can be NULL if this + * value is not necessary. + * + * @param pErrorCode must be a valid pointer to an error code value. + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ubidi_setPara(UBiDi *pBiDi, const UChar *text, int32_t length, + UBiDiLevel paraLevel, UBiDiLevel *embeddingLevels, + UErrorCode *pErrorCode); + +/** + * ubidi_setLine() sets a UBiDi to + * contain the reordering information, especially the resolved levels, + * for all the characters in a line of text. This line of text is + * specified by referring to a UBiDi object representing + * this information for a piece of text containing one or more paragraphs, + * and by specifying a range of indexes in this text.

+ * In the new line object, the indexes will range from 0 to limit-start-1.

+ * + * This is used after calling ubidi_setPara() + * for a piece of text, and after line-breaking on that text. + * It is not necessary if each paragraph is treated as a single line.

+ * + * After line-breaking, rules (L1) and (L2) for the treatment of + * trailing WS and for reordering are performed on + * a UBiDi object that represents a line.

+ * + * Important: pLineBiDi shares data with + * pParaBiDi. + * You must destroy or reuse pLineBiDi before pParaBiDi. + * In other words, you must destroy or reuse the UBiDi object for a line + * before the object for its parent paragraph.

+ * + * The text pointer that was stored in pParaBiDi is also copied, + * and start is added to it so that it points to the beginning of the + * line for this object. + * + * @param pParaBiDi is the parent paragraph object. It must have been set + * by a successful call to ubidi_setPara. + * + * @param start is the line's first index into the text. + * + * @param limit is just behind the line's last index into the text + * (its last index +1).
+ * It must be 0<=startcontaining paragraph limit. + * If the specified line crosses a paragraph boundary, the function + * will terminate with error code U_ILLEGAL_ARGUMENT_ERROR. + * + * @param pLineBiDi is the object that will now represent a line of the text. + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @see ubidi_setPara + * @see ubidi_getProcessedLength + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ubidi_setLine(const UBiDi *pParaBiDi, + int32_t start, int32_t limit, + UBiDi *pLineBiDi, + UErrorCode *pErrorCode); + +/** + * Get the directionality of the text. + * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @return a value of UBIDI_LTR, UBIDI_RTL + * or UBIDI_MIXED + * that indicates if the entire text + * represented by this object is unidirectional, + * and which direction, or if it is mixed-directional. + * Note - The value UBIDI_NEUTRAL is never returned from this method. + * + * @see UBiDiDirection + * @stable ICU 2.0 + */ +U_CAPI UBiDiDirection U_EXPORT2 +ubidi_getDirection(const UBiDi *pBiDi); + +/** + * Gets the base direction of the text provided according + * to the Unicode Bidirectional Algorithm. The base direction + * is derived from the first character in the string with bidirectional + * character type L, R, or AL. If the first such character has type L, + * UBIDI_LTR is returned. If the first such character has + * type R or AL, UBIDI_RTL is returned. If the string does + * not contain any character of these types, then + * UBIDI_NEUTRAL is returned. + * + * This is a lightweight function for use when only the base direction + * is needed and no further bidi processing of the text is needed. + * + * @param text is a pointer to the text whose base + * direction is needed. + * Note: the text must be (at least) @c length long. + * + * @param length is the length of the text; + * if length==-1 then the text + * must be zero-terminated. + * + * @return UBIDI_LTR, UBIDI_RTL, + * UBIDI_NEUTRAL + * + * @see UBiDiDirection + * @stable ICU 4.6 + */ +U_CAPI UBiDiDirection U_EXPORT2 +ubidi_getBaseDirection(const UChar *text, int32_t length ); + +/** + * Get the pointer to the text. + * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @return The pointer to the text that the UBiDi object was created for. + * + * @see ubidi_setPara + * @see ubidi_setLine + * @stable ICU 2.0 + */ +U_CAPI const UChar * U_EXPORT2 +ubidi_getText(const UBiDi *pBiDi); + +/** + * Get the length of the text. + * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @return The length of the text that the UBiDi object was created for. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ubidi_getLength(const UBiDi *pBiDi); + +/** + * Get the paragraph level of the text. + * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @return The paragraph level. If there are multiple paragraphs, their + * level may vary if the required paraLevel is UBIDI_DEFAULT_LTR or + * UBIDI_DEFAULT_RTL. In that case, the level of the first paragraph + * is returned. + * + * @see UBiDiLevel + * @see ubidi_getParagraph + * @see ubidi_getParagraphByIndex + * @stable ICU 2.0 + */ +U_CAPI UBiDiLevel U_EXPORT2 +ubidi_getParaLevel(const UBiDi *pBiDi); + +/** + * Get the number of paragraphs. + * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @return The number of paragraphs. + * @stable ICU 3.4 + */ +U_CAPI int32_t U_EXPORT2 +ubidi_countParagraphs(UBiDi *pBiDi); + +/** + * Get a paragraph, given a position within the text. + * This function returns information about a paragraph.
+ * Note: if the paragraph index is known, it is more efficient to + * retrieve the paragraph information using ubidi_getParagraphByIndex().

+ * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @param charIndex is the index of a character within the text, in the + * range [0..ubidi_getProcessedLength(pBiDi)-1]. + * + * @param pParaStart will receive the index of the first character of the + * paragraph in the text. + * This pointer can be NULL if this + * value is not necessary. + * + * @param pParaLimit will receive the limit of the paragraph. + * The l-value that you point to here may be the + * same expression (variable) as the one for + * charIndex. + * This pointer can be NULL if this + * value is not necessary. + * + * @param pParaLevel will receive the level of the paragraph. + * This pointer can be NULL if this + * value is not necessary. + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @return The index of the paragraph containing the specified position. + * + * @see ubidi_getProcessedLength + * @stable ICU 3.4 + */ +U_CAPI int32_t U_EXPORT2 +ubidi_getParagraph(const UBiDi *pBiDi, int32_t charIndex, int32_t *pParaStart, + int32_t *pParaLimit, UBiDiLevel *pParaLevel, + UErrorCode *pErrorCode); + +/** + * Get a paragraph, given the index of this paragraph. + * + * This function returns information about a paragraph.

+ * + * @param pBiDi is the paragraph UBiDi object. + * + * @param paraIndex is the number of the paragraph, in the + * range [0..ubidi_countParagraphs(pBiDi)-1]. + * + * @param pParaStart will receive the index of the first character of the + * paragraph in the text. + * This pointer can be NULL if this + * value is not necessary. + * + * @param pParaLimit will receive the limit of the paragraph. + * This pointer can be NULL if this + * value is not necessary. + * + * @param pParaLevel will receive the level of the paragraph. + * This pointer can be NULL if this + * value is not necessary. + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @stable ICU 3.4 + */ +U_CAPI void U_EXPORT2 +ubidi_getParagraphByIndex(const UBiDi *pBiDi, int32_t paraIndex, + int32_t *pParaStart, int32_t *pParaLimit, + UBiDiLevel *pParaLevel, UErrorCode *pErrorCode); + +/** + * Get the level for one character. + * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @param charIndex the index of a character. It must be in the range + * [0..ubidi_getProcessedLength(pBiDi)]. + * + * @return The level for the character at charIndex (0 if charIndex is not + * in the valid range). + * + * @see UBiDiLevel + * @see ubidi_getProcessedLength + * @stable ICU 2.0 + */ +U_CAPI UBiDiLevel U_EXPORT2 +ubidi_getLevelAt(const UBiDi *pBiDi, int32_t charIndex); + +/** + * Get an array of levels for each character.

+ * + * Note that this function may allocate memory under some + * circumstances, unlike ubidi_getLevelAt(). + * + * @param pBiDi is the paragraph or line UBiDi object, whose + * text length must be strictly positive. + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @return The levels array for the text, + * or NULL if an error occurs. + * + * @see UBiDiLevel + * @see ubidi_getProcessedLength + * @stable ICU 2.0 + */ +U_CAPI const UBiDiLevel * U_EXPORT2 +ubidi_getLevels(UBiDi *pBiDi, UErrorCode *pErrorCode); + +/** + * Get a logical run. + * This function returns information about a run and is used + * to retrieve runs in logical order.

+ * This is especially useful for line-breaking on a paragraph. + * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @param logicalPosition is a logical position within the source text. + * + * @param pLogicalLimit will receive the limit of the corresponding run. + * The l-value that you point to here may be the + * same expression (variable) as the one for + * logicalPosition. + * This pointer can be NULL if this + * value is not necessary. + * + * @param pLevel will receive the level of the corresponding run. + * This pointer can be NULL if this + * value is not necessary. + * + * @see ubidi_getProcessedLength + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ubidi_getLogicalRun(const UBiDi *pBiDi, int32_t logicalPosition, + int32_t *pLogicalLimit, UBiDiLevel *pLevel); + +/** + * Get the number of runs. + * This function may invoke the actual reordering on the + * UBiDi object, after ubidi_setPara() + * may have resolved only the levels of the text. Therefore, + * ubidi_countRuns() may have to allocate memory, + * and may fail doing so. + * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @return The number of runs. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ubidi_countRuns(UBiDi *pBiDi, UErrorCode *pErrorCode); + +/** + * Get one run's logical start, length, and directionality, + * which can be 0 for LTR or 1 for RTL. + * In an RTL run, the character at the logical start is + * visually on the right of the displayed run. + * The length is the number of characters in the run.

+ * ubidi_countRuns() should be called + * before the runs are retrieved. + * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @param runIndex is the number of the run in visual order, in the + * range [0..ubidi_countRuns(pBiDi)-1]. + * + * @param pLogicalStart is the first logical character index in the text. + * The pointer may be NULL if this index is not needed. + * + * @param pLength is the number of characters (at least one) in the run. + * The pointer may be NULL if this is not needed. + * + * @return the directionality of the run, + * UBIDI_LTR==0 or UBIDI_RTL==1, + * never UBIDI_MIXED, + * never UBIDI_NEUTRAL. + * + * @see ubidi_countRuns + * + * Example: + * 

+ * \code
+ * int32_t i, count=ubidi_countRuns(pBiDi),
+ *         logicalStart, visualIndex=0, length;
+ * for(i=0; i0);
+ *     } else {
+ *         logicalStart+=length;  // logicalLimit
+ *         do { // RTL
+ *             show_char(text[--logicalStart], visualIndex++);
+ *         } while(--length>0);
+ *     }
+ * }
+ *\endcode
+ *
+ * + * Note that in right-to-left runs, code like this places + * second surrogates before first ones (which is generally a bad idea) + * and combining characters before base characters. + *

+ * Use of ubidi_writeReordered(), optionally with the + * #UBIDI_KEEP_BASE_COMBINING option, can be considered in order + * to avoid these issues. + * @stable ICU 2.0 + */ +U_CAPI UBiDiDirection U_EXPORT2 +ubidi_getVisualRun(UBiDi *pBiDi, int32_t runIndex, + int32_t *pLogicalStart, int32_t *pLength); + +/** + * Get the visual position from a logical text position. + * If such a mapping is used many times on the same + * UBiDi object, then calling + * ubidi_getLogicalMap() is more efficient.

+ * + * The value returned may be #UBIDI_MAP_NOWHERE if there is no + * visual position because the corresponding text character is a Bidi control + * removed from output by the option #UBIDI_OPTION_REMOVE_CONTROLS. + *

+ * When the visual output is altered by using options of + * ubidi_writeReordered() such as UBIDI_INSERT_LRM_FOR_NUMERIC, + * UBIDI_KEEP_BASE_COMBINING, UBIDI_OUTPUT_REVERSE, + * UBIDI_REMOVE_BIDI_CONTROLS, the visual position returned may not + * be correct. It is advised to use, when possible, reordering options + * such as UBIDI_OPTION_INSERT_MARKS and UBIDI_OPTION_REMOVE_CONTROLS. + *

+ * Note that in right-to-left runs, this mapping places + * second surrogates before first ones (which is generally a bad idea) + * and combining characters before base characters. + * Use of ubidi_writeReordered(), optionally with the + * #UBIDI_KEEP_BASE_COMBINING option can be considered instead + * of using the mapping, in order to avoid these issues. + * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @param logicalIndex is the index of a character in the text. + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @return The visual position of this character. + * + * @see ubidi_getLogicalMap + * @see ubidi_getLogicalIndex + * @see ubidi_getProcessedLength + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ubidi_getVisualIndex(UBiDi *pBiDi, int32_t logicalIndex, UErrorCode *pErrorCode); + +/** + * Get the logical text position from a visual position. + * If such a mapping is used many times on the same + * UBiDi object, then calling + * ubidi_getVisualMap() is more efficient.

+ * + * The value returned may be #UBIDI_MAP_NOWHERE if there is no + * logical position because the corresponding text character is a Bidi mark + * inserted in the output by option #UBIDI_OPTION_INSERT_MARKS. + *

+ * This is the inverse function to ubidi_getVisualIndex(). + *

+ * When the visual output is altered by using options of + * ubidi_writeReordered() such as UBIDI_INSERT_LRM_FOR_NUMERIC, + * UBIDI_KEEP_BASE_COMBINING, UBIDI_OUTPUT_REVERSE, + * UBIDI_REMOVE_BIDI_CONTROLS, the logical position returned may not + * be correct. It is advised to use, when possible, reordering options + * such as UBIDI_OPTION_INSERT_MARKS and UBIDI_OPTION_REMOVE_CONTROLS. + * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @param visualIndex is the visual position of a character. + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @return The index of this character in the text. + * + * @see ubidi_getVisualMap + * @see ubidi_getVisualIndex + * @see ubidi_getResultLength + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ubidi_getLogicalIndex(UBiDi *pBiDi, int32_t visualIndex, UErrorCode *pErrorCode); + +/** + * Get a logical-to-visual index map (array) for the characters in the UBiDi + * (paragraph or line) object. + *

+ * Some values in the map may be #UBIDI_MAP_NOWHERE if the + * corresponding text characters are Bidi controls removed from the visual + * output by the option #UBIDI_OPTION_REMOVE_CONTROLS. + *

+ * When the visual output is altered by using options of + * ubidi_writeReordered() such as UBIDI_INSERT_LRM_FOR_NUMERIC, + * UBIDI_KEEP_BASE_COMBINING, UBIDI_OUTPUT_REVERSE, + * UBIDI_REMOVE_BIDI_CONTROLS, the visual positions returned may not + * be correct. It is advised to use, when possible, reordering options + * such as UBIDI_OPTION_INSERT_MARKS and UBIDI_OPTION_REMOVE_CONTROLS. + *

+ * Note that in right-to-left runs, this mapping places + * second surrogates before first ones (which is generally a bad idea) + * and combining characters before base characters. + * Use of ubidi_writeReordered(), optionally with the + * #UBIDI_KEEP_BASE_COMBINING option can be considered instead + * of using the mapping, in order to avoid these issues. + * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @param indexMap is a pointer to an array of ubidi_getProcessedLength() + * indexes which will reflect the reordering of the characters. + * If option #UBIDI_OPTION_INSERT_MARKS is set, the number + * of elements allocated in indexMap must be no less than + * ubidi_getResultLength(). + * The array does not need to be initialized.

+ * The index map will result in indexMap[logicalIndex]==visualIndex. + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @see ubidi_getVisualMap + * @see ubidi_getVisualIndex + * @see ubidi_getProcessedLength + * @see ubidi_getResultLength + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ubidi_getLogicalMap(UBiDi *pBiDi, int32_t *indexMap, UErrorCode *pErrorCode); + +/** + * Get a visual-to-logical index map (array) for the characters in the UBiDi + * (paragraph or line) object. + *

+ * Some values in the map may be #UBIDI_MAP_NOWHERE if the + * corresponding text characters are Bidi marks inserted in the visual output + * by the option #UBIDI_OPTION_INSERT_MARKS. + *

+ * When the visual output is altered by using options of + * ubidi_writeReordered() such as UBIDI_INSERT_LRM_FOR_NUMERIC, + * UBIDI_KEEP_BASE_COMBINING, UBIDI_OUTPUT_REVERSE, + * UBIDI_REMOVE_BIDI_CONTROLS, the logical positions returned may not + * be correct. It is advised to use, when possible, reordering options + * such as UBIDI_OPTION_INSERT_MARKS and UBIDI_OPTION_REMOVE_CONTROLS. + * + * @param pBiDi is the paragraph or line UBiDi object. + * + * @param indexMap is a pointer to an array of ubidi_getResultLength() + * indexes which will reflect the reordering of the characters. + * If option #UBIDI_OPTION_REMOVE_CONTROLS is set, the number + * of elements allocated in indexMap must be no less than + * ubidi_getProcessedLength(). + * The array does not need to be initialized.

+ * The index map will result in indexMap[visualIndex]==logicalIndex. + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @see ubidi_getLogicalMap + * @see ubidi_getLogicalIndex + * @see ubidi_getProcessedLength + * @see ubidi_getResultLength + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ubidi_getVisualMap(UBiDi *pBiDi, int32_t *indexMap, UErrorCode *pErrorCode); + +/** + * This is a convenience function that does not use a UBiDi object. + * It is intended to be used for when an application has determined the levels + * of objects (character sequences) and just needs to have them reordered (L2). + * This is equivalent to using ubidi_getLogicalMap() on a + * UBiDi object. + * + * @param levels is an array with length levels that have been determined by + * the application. + * + * @param length is the number of levels in the array, or, semantically, + * the number of objects to be reordered. + * It must be length>0. + * + * @param indexMap is a pointer to an array of length + * indexes which will reflect the reordering of the characters. + * The array does not need to be initialized.

+ * The index map will result in indexMap[logicalIndex]==visualIndex. + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ubidi_reorderLogical(const UBiDiLevel *levels, int32_t length, int32_t *indexMap); + +/** + * This is a convenience function that does not use a UBiDi object. + * It is intended to be used for when an application has determined the levels + * of objects (character sequences) and just needs to have them reordered (L2). + * This is equivalent to using ubidi_getVisualMap() on a + * UBiDi object. + * + * @param levels is an array with length levels that have been determined by + * the application. + * + * @param length is the number of levels in the array, or, semantically, + * the number of objects to be reordered. + * It must be length>0. + * + * @param indexMap is a pointer to an array of length + * indexes which will reflect the reordering of the characters. + * The array does not need to be initialized.

+ * The index map will result in indexMap[visualIndex]==logicalIndex. + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ubidi_reorderVisual(const UBiDiLevel *levels, int32_t length, int32_t *indexMap); + +/** + * Invert an index map. + * The index mapping of the first map is inverted and written to + * the second one. + * + * @param srcMap is an array with length elements + * which defines the original mapping from a source array containing + * length elements to a destination array. + * Some elements of the source array may have no mapping in the + * destination array. In that case, their value will be + * the special value UBIDI_MAP_NOWHERE. + * All elements must be >=0 or equal to UBIDI_MAP_NOWHERE. + * Some elements may have a value >= length, if the + * destination array has more elements than the source array. + * There must be no duplicate indexes (two or more elements with the + * same value except UBIDI_MAP_NOWHERE). + * + * @param destMap is an array with a number of elements equal to 1 + the highest + * value in srcMap. + * destMap will be filled with the inverse mapping. + * If element with index i in srcMap has a value k different + * from UBIDI_MAP_NOWHERE, this means that element i of + * the source array maps to element k in the destination array. + * The inverse map will have value i in its k-th element. + * For all elements of the destination array which do not map to + * an element in the source array, the corresponding element in the + * inverse map will have a value equal to UBIDI_MAP_NOWHERE. + * + * @param length is the length of each array. + * @see UBIDI_MAP_NOWHERE + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ubidi_invertMap(const int32_t *srcMap, int32_t *destMap, int32_t length); + +/** option flags for ubidi_writeReordered() */ + +/** + * option bit for ubidi_writeReordered(): + * keep combining characters after their base characters in RTL runs + * + * @see ubidi_writeReordered + * @stable ICU 2.0 + */ +#define UBIDI_KEEP_BASE_COMBINING 1 + +/** + * option bit for ubidi_writeReordered(): + * replace characters with the "mirrored" property in RTL runs + * by their mirror-image mappings + * + * @see ubidi_writeReordered + * @stable ICU 2.0 + */ +#define UBIDI_DO_MIRRORING 2 + +/** + * option bit for ubidi_writeReordered(): + * surround the run with LRMs if necessary; + * this is part of the approximate "inverse Bidi" algorithm + * + *

This option does not imply corresponding adjustment of the index + * mappings.

+ * + * @see ubidi_setInverse + * @see ubidi_writeReordered + * @stable ICU 2.0 + */ +#define UBIDI_INSERT_LRM_FOR_NUMERIC 4 + +/** + * option bit for ubidi_writeReordered(): + * remove Bidi control characters + * (this does not affect #UBIDI_INSERT_LRM_FOR_NUMERIC) + * + *

This option does not imply corresponding adjustment of the index + * mappings.

+ * + * @see ubidi_writeReordered + * @stable ICU 2.0 + */ +#define UBIDI_REMOVE_BIDI_CONTROLS 8 + +/** + * option bit for ubidi_writeReordered(): + * write the output in reverse order + * + *

This has the same effect as calling ubidi_writeReordered() + * first without this option, and then calling + * ubidi_writeReverse() without mirroring. + * Doing this in the same step is faster and avoids a temporary buffer. + * An example for using this option is output to a character terminal that + * is designed for RTL scripts and stores text in reverse order.

+ * + * @see ubidi_writeReordered + * @stable ICU 2.0 + */ +#define UBIDI_OUTPUT_REVERSE 16 + +/** + * Get the length of the source text processed by the last call to + * ubidi_setPara(). This length may be different from the length + * of the source text if option #UBIDI_OPTION_STREAMING + * has been set. + *
+ * Note that whenever the length of the text affects the execution or the + * result of a function, it is the processed length which must be considered, + * except for ubidi_setPara (which receives unprocessed source + * text) and ubidi_getLength (which returns the original length + * of the source text).
+ * In particular, the processed length is the one to consider in the following + * cases: + *
    + *
  • maximum value of the limit argument of + * ubidi_setLine
    • + *
  • maximum value of the charIndex argument of + * ubidi_getParagraph
    • + *
  • maximum value of the charIndex argument of + * ubidi_getLevelAt
    • + *
  • number of elements in the array returned by ubidi_getLevels
    • + *
  • maximum value of the logicalStart argument of + * ubidi_getLogicalRun
    • + *
  • maximum value of the logicalIndex argument of + * ubidi_getVisualIndex
    • + *
  • number of elements filled in the *indexMap argument of + * ubidi_getLogicalMap
    • + *
  • length of text processed by ubidi_writeReordered
    • + *
+ * + * @param pBiDi is the paragraph UBiDi object. + * + * @return The length of the part of the source text processed by + * the last call to ubidi_setPara. + * @see ubidi_setPara + * @see UBIDI_OPTION_STREAMING + * @stable ICU 3.6 + */ +U_CAPI int32_t U_EXPORT2 +ubidi_getProcessedLength(const UBiDi *pBiDi); + +/** + * Get the length of the reordered text resulting from the last call to + * ubidi_setPara(). This length may be different from the length + * of the source text if option #UBIDI_OPTION_INSERT_MARKS + * or option #UBIDI_OPTION_REMOVE_CONTROLS has been set. + *
+ * This resulting length is the one to consider in the following cases: + *
    + *
  • maximum value of the visualIndex argument of + * ubidi_getLogicalIndex
    • + *
  • number of elements of the *indexMap argument of + * ubidi_getVisualMap
    • + *
+ * Note that this length stays identical to the source text length if + * Bidi marks are inserted or removed using option bits of + * ubidi_writeReordered, or if option + * #UBIDI_REORDER_INVERSE_NUMBERS_AS_L has been set. + * + * @param pBiDi is the paragraph UBiDi object. + * + * @return The length of the reordered text resulting from + * the last call to ubidi_setPara. + * @see ubidi_setPara + * @see UBIDI_OPTION_INSERT_MARKS + * @see UBIDI_OPTION_REMOVE_CONTROLS + * @stable ICU 3.6 + */ +U_CAPI int32_t U_EXPORT2 +ubidi_getResultLength(const UBiDi *pBiDi); + +U_CDECL_BEGIN + +#ifndef U_HIDE_DEPRECATED_API +/** + * Value returned by UBiDiClassCallback callbacks when + * there is no need to override the standard Bidi class for a given code point. + * + * This constant is deprecated; use u_getIntPropertyMaxValue(UCHAR_BIDI_CLASS)+1 instead. + * + * @see UBiDiClassCallback + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ +#define U_BIDI_CLASS_DEFAULT U_CHAR_DIRECTION_COUNT +#endif // U_HIDE_DEPRECATED_API + +/** + * Callback type declaration for overriding default Bidi class values with + * custom ones. + *

Usually, the function pointer will be propagated to a UBiDi + * object by calling the ubidi_setClassCallback() function; + * then the callback will be invoked by the UBA implementation any time the + * class of a character is to be determined.

+ * + * @param context is a pointer to the callback private data. + * + * @param c is the code point to get a Bidi class for. + * + * @return The directional property / Bidi class for the given code point + * c if the default class has been overridden, or + * u_getIntPropertyMaxValue(UCHAR_BIDI_CLASS)+1 + * if the standard Bidi class value for c is to be used. + * @see ubidi_setClassCallback + * @see ubidi_getClassCallback + * @stable ICU 3.6 + */ +typedef UCharDirection U_CALLCONV +UBiDiClassCallback(const void *context, UChar32 c); + +U_CDECL_END + +/** + * Retrieve the Bidi class for a given code point. + *

If a #UBiDiClassCallback callback is defined and returns a + * value other than u_getIntPropertyMaxValue(UCHAR_BIDI_CLASS)+1, + * that value is used; otherwise the default class determination mechanism is invoked.

+ * + * @param pBiDi is the paragraph UBiDi object. + * + * @param c is the code point whose Bidi class must be retrieved. + * + * @return The Bidi class for character c based + * on the given pBiDi instance. + * @see UBiDiClassCallback + * @stable ICU 3.6 + */ +U_CAPI UCharDirection U_EXPORT2 +ubidi_getCustomizedClass(UBiDi *pBiDi, UChar32 c); + +/** + * Set the callback function and callback data used by the UBA + * implementation for Bidi class determination. + *

This may be useful for assigning Bidi classes to PUA characters, or + * for special application needs. For instance, an application may want to + * handle all spaces like L or R characters (according to the base direction) + * when creating the visual ordering of logical lines which are part of a report + * organized in columns: there should not be interaction between adjacent + * cells.

+ * + * @param pBiDi is the paragraph UBiDi object. + * + * @param newFn is the new callback function pointer. + * + * @param newContext is the new callback context pointer. This can be NULL. + * + * @param oldFn fillin: Returns the old callback function pointer. This can be + * NULL. + * + * @param oldContext fillin: Returns the old callback's context. This can be + * NULL. + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @see ubidi_getClassCallback + * @stable ICU 3.6 + */ +U_CAPI void U_EXPORT2 +ubidi_setClassCallback(UBiDi *pBiDi, UBiDiClassCallback *newFn, + const void *newContext, UBiDiClassCallback **oldFn, + const void **oldContext, UErrorCode *pErrorCode); + +/** + * Get the current callback function used for Bidi class determination. + * + * @param pBiDi is the paragraph UBiDi object. + * + * @param fn fillin: Returns the callback function pointer. + * + * @param context fillin: Returns the callback's private context. + * + * @see ubidi_setClassCallback + * @stable ICU 3.6 + */ +U_CAPI void U_EXPORT2 +ubidi_getClassCallback(UBiDi *pBiDi, UBiDiClassCallback **fn, const void **context); + +/** + * Take a UBiDi object containing the reordering + * information for a piece of text (one or more paragraphs) set by + * ubidi_setPara() or for a line of text set by + * ubidi_setLine() and write a reordered string to the + * destination buffer. + * + * This function preserves the integrity of characters with multiple + * code units and (optionally) combining characters. + * Characters in RTL runs can be replaced by mirror-image characters + * in the destination buffer. Note that "real" mirroring has + * to be done in a rendering engine by glyph selection + * and that for many "mirrored" characters there are no + * Unicode characters as mirror-image equivalents. + * There are also options to insert or remove Bidi control + * characters; see the description of the destSize + * and options parameters and of the option bit flags. + * + * @param pBiDi A pointer to a UBiDi object that + * is set by ubidi_setPara() or + * ubidi_setLine() and contains the reordering + * information for the text that it was defined for, + * as well as a pointer to that text.

+ * The text was aliased (only the pointer was stored + * without copying the contents) and must not have been modified + * since the ubidi_setPara() call. + * + * @param dest A pointer to where the reordered text is to be copied. + * The source text and dest[destSize] + * must not overlap. + * + * @param destSize The size of the dest buffer, + * in number of UChars. + * If the UBIDI_INSERT_LRM_FOR_NUMERIC + * option is set, then the destination length could be + * as large as + * ubidi_getLength(pBiDi)+2*ubidi_countRuns(pBiDi). + * If the UBIDI_REMOVE_BIDI_CONTROLS option + * is set, then the destination length may be less than + * ubidi_getLength(pBiDi). + * If none of these options is set, then the destination length + * will be exactly ubidi_getProcessedLength(pBiDi). + * + * @param options A bit set of options for the reordering that control + * how the reordered text is written. + * The options include mirroring the characters on a code + * point basis and inserting LRM characters, which is used + * especially for transforming visually stored text + * to logically stored text (although this is still an + * imperfect implementation of an "inverse Bidi" algorithm + * because it uses the "forward Bidi" algorithm at its core). + * The available options are: + * #UBIDI_DO_MIRRORING, + * #UBIDI_INSERT_LRM_FOR_NUMERIC, + * #UBIDI_KEEP_BASE_COMBINING, + * #UBIDI_OUTPUT_REVERSE, + * #UBIDI_REMOVE_BIDI_CONTROLS + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @return The length of the output string. + * + * @see ubidi_getProcessedLength + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ubidi_writeReordered(UBiDi *pBiDi, + UChar *dest, int32_t destSize, + uint16_t options, + UErrorCode *pErrorCode); + +/** + * Reverse a Right-To-Left run of Unicode text. + * + * This function preserves the integrity of characters with multiple + * code units and (optionally) combining characters. + * Characters can be replaced by mirror-image characters + * in the destination buffer. Note that "real" mirroring has + * to be done in a rendering engine by glyph selection + * and that for many "mirrored" characters there are no + * Unicode characters as mirror-image equivalents. + * There are also options to insert or remove Bidi control + * characters. + * + * This function is the implementation for reversing RTL runs as part + * of ubidi_writeReordered(). For detailed descriptions + * of the parameters, see there. + * Since no Bidi controls are inserted here, the output string length + * will never exceed srcLength. + * + * @see ubidi_writeReordered + * + * @param src A pointer to the RTL run text. + * + * @param srcLength The length of the RTL run. + * + * @param dest A pointer to where the reordered text is to be copied. + * src[srcLength] and dest[destSize] + * must not overlap. + * + * @param destSize The size of the dest buffer, + * in number of UChars. + * If the UBIDI_REMOVE_BIDI_CONTROLS option + * is set, then the destination length may be less than + * srcLength. + * If this option is not set, then the destination length + * will be exactly srcLength. + * + * @param options A bit set of options for the reordering that control + * how the reordered text is written. + * See the options parameter in ubidi_writeReordered(). + * + * @param pErrorCode must be a valid pointer to an error code value. + * + * @return The length of the output string. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ubidi_writeReverse(const UChar *src, int32_t srcLength, + UChar *dest, int32_t destSize, + uint16_t options, + UErrorCode *pErrorCode); + +/*#define BIDI_SAMPLE_CODE*/ +/*@}*/ + +#endif diff --git a/intl/icu/source/common/unicode/ubiditransform.h b/intl/icu/source/common/unicode/ubiditransform.h new file mode 100644 index 0000000000..24433aa8ac --- /dev/null +++ b/intl/icu/source/common/unicode/ubiditransform.h @@ -0,0 +1,326 @@ +/* +****************************************************************************** +* +* © 2016 and later: Unicode, Inc. and others. +* License & terms of use: http://www.unicode.org/copyright.html +* +****************************************************************************** +* file name: ubiditransform.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2016jul24 +* created by: Lina Kemmel +* +*/ + +#ifndef UBIDITRANSFORM_H +#define UBIDITRANSFORM_H + +#include "unicode/utypes.h" +#include "unicode/ubidi.h" +#include "unicode/uchar.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +/** + * \file + * \brief C API: Bidi Transformations + */ + +/** + * `UBiDiOrder` indicates the order of text. + * + * This bidi transformation engine supports all possible combinations (4 in + * total) of input and output text order: + * + * - : unless the output direction is RTL, this + * corresponds to a normal operation of the Bidi algorithm as described in the + * Unicode Technical Report and implemented by `UBiDi` when the + * reordering mode is set to `UBIDI_REORDER_DEFAULT`. Visual RTL + * mode is not supported by `UBiDi` and is accomplished through + * reversing a visual LTR string, + * + * - : unless the input direction is RTL, this + * corresponds to an "inverse bidi algorithm" in `UBiDi` with the + * reordering mode set to `UBIDI_REORDER_INVERSE_LIKE_DIRECT`. + * Visual RTL mode is not not supported by `UBiDi` and is + * accomplished through reversing a visual LTR string, + * + * - : if the input and output base directions + * mismatch, this corresponds to the `UBiDi` implementation with the + * reordering mode set to `UBIDI_REORDER_RUNS_ONLY`; and if the + * input and output base directions are identical, the transformation engine + * will only handle character mirroring and Arabic shaping operations without + * reordering, + * + * - : this reordering mode is not supported by + * the `UBiDi` engine; it implies character mirroring, Arabic + * shaping, and - if the input/output base directions mismatch - string + * reverse operations. + * @see ubidi_setInverse + * @see ubidi_setReorderingMode + * @see UBIDI_REORDER_DEFAULT + * @see UBIDI_REORDER_INVERSE_LIKE_DIRECT + * @see UBIDI_REORDER_RUNS_ONLY + * @stable ICU 58 + */ +typedef enum { + /** 0: Constant indicating a logical order. + * This is the default for input text. + * @stable ICU 58 + */ + UBIDI_LOGICAL = 0, + /** 1: Constant indicating a visual order. + * This is a default for output text. + * @stable ICU 58 + */ + UBIDI_VISUAL +} UBiDiOrder; + +/** + * UBiDiMirroring indicates whether or not characters with the + * "mirrored" property in RTL runs should be replaced with their mirror-image + * counterparts. + * @see UBIDI_DO_MIRRORING + * @see ubidi_setReorderingOptions + * @see ubidi_writeReordered + * @see ubidi_writeReverse + * @stable ICU 58 + */ +typedef enum { + /** 0: Constant indicating that character mirroring should not be + * performed. + * This is the default. + * @stable ICU 58 + */ + UBIDI_MIRRORING_OFF = 0, + /** 1: Constant indicating that character mirroring should be performed. + * This corresponds to calling ubidi_writeReordered or + * ubidi_writeReverse with the + * UBIDI_DO_MIRRORING option bit set. + * @stable ICU 58 + */ + UBIDI_MIRRORING_ON +} UBiDiMirroring; + +/** + * Forward declaration of the UBiDiTransform structure that stores + * information used by the layout transformation engine. + * @stable ICU 58 + */ +typedef struct UBiDiTransform UBiDiTransform; + +/** + * Performs transformation of text from the bidi layout defined by the input + * ordering scheme to the bidi layout defined by the output ordering scheme, + * and applies character mirroring and Arabic shaping operations.

+ * In terms of UBiDi, such a transformation implies: + *

    + *
  • calling ubidi_setReorderingMode as needed (when the + * reordering mode is other than normal),
    • + *
  • calling ubidi_setInverse as needed (when text should be + * transformed from a visual to a logical form),
    • + *
  • resolving embedding levels of each character in the input text by + * calling ubidi_setPara,
    • + *
  • reordering the characters based on the computed embedding levels, also + * performing character mirroring as needed, and streaming the result to the + * output, by calling ubidi_writeReordered,
    • + *
  • performing Arabic digit and letter shaping on the output text by calling + * u_shapeArabic.
    • + *
+ * An "ordering scheme" encompasses the base direction and the order of text, + * and these characteristics must be defined by the caller for both input and + * output explicitly .

+ * There are 36 possible combinations of ordering schemes, + * which are partially supported by UBiDi already. Examples of the + * currently supported combinations: + *

    + *
  • : this is equivalent to calling + * ubidi_setPara with paraLevel == UBIDI_LTR,
    • + *
  • : this is equivalent to calling + * ubidi_setPara with paraLevel == UBIDI_RTL,
    • + *
  • : this is equivalent to + * calling ubidi_setPara with + * paraLevel == UBIDI_DEFAULT_LTR,
    • + *
  • : this is equivalent to + * calling ubidi_setPara with + * paraLevel == UBIDI_DEFAULT_RTL,
    • + *
  • : this is equivalent to + * calling ubidi_setInverse(UBiDi*, true) and then + * ubidi_setPara with paraLevel == UBIDI_LTR,
    • + *
  • : this is equivalent to + * calling ubidi_setInverse(UBiDi*, true) and then + * ubidi_setPara with paraLevel == UBIDI_RTL.
    • + *
+ * All combinations that involve the Visual RTL scheme are unsupported by + * UBiDi, for instance: + *
    + *
  • ,
    • + *
  • .
    • + *
+ *

Example of usage of the transformation engine:
+ * 

+ * \code
+ * UChar text1[] = {'a', 'b', 'c', 0x0625, '1', 0};
+ * UChar text2[] = {'a', 'b', 'c', 0x0625, '1', 0};
+ * UErrorCode errorCode = U_ZERO_ERROR;
+ * // Run a transformation.
+ * ubiditransform_transform(pBidiTransform,
+ *          text1, -1, text2, -1,
+ *          UBIDI_LTR, UBIDI_VISUAL,
+ *          UBIDI_RTL, UBIDI_LOGICAL,
+ *          UBIDI_MIRRORING_OFF,
+ *          U_SHAPE_DIGITS_AN2EN | U_SHAPE_DIGIT_TYPE_AN_EXTENDED,
+ *          &errorCode);
+ * // Do something with text2.
+ *  text2[4] = '2';
+ * // Run a reverse transformation.
+ * ubiditransform_transform(pBidiTransform,
+ *          text2, -1, text1, -1,
+ *          UBIDI_RTL, UBIDI_LOGICAL,
+ *          UBIDI_LTR, UBIDI_VISUAL,
+ *          UBIDI_MIRRORING_OFF,
+ *          U_SHAPE_DIGITS_EN2AN | U_SHAPE_DIGIT_TYPE_AN_EXTENDED,
+ *          &errorCode);
+ *\endcode
+ *
+ *

+ * + * @param pBiDiTransform A pointer to a UBiDiTransform object + * allocated with ubiditransform_open() or + * NULL.

+ * This object serves for one-time setup to amortize initialization + * overheads. Use of this object is not thread-safe. All other threads + * should allocate a new UBiDiTransform object by calling + * ubiditransform_open() before using it. Alternatively, + * a caller can set this parameter to NULL, in which case + * the object will be allocated by the engine on the fly.

+ * @param src A pointer to the text that the Bidi layout transformations will + * be performed on. + *

Note: the text must be (at least) + * srcLength long.

+ * @param srcLength The length of the text, in number of UChars. If + * length == -1 then the text must be zero-terminated. + * @param dest A pointer to where the processed text is to be copied. + * @param destSize The size of the dest buffer, in number of + * UChars. If the U_SHAPE_LETTERS_UNSHAPE option is set, + * then the destination length could be as large as + * srcLength * 2. Otherwise, the destination length will + * not exceed srcLength. If the caller reserves the last + * position for zero-termination, it should be excluded from + * destSize. + *

destSize == -1 is allowed and makes sense when + * dest was holds some meaningful value, e.g. that of + * src. In this case dest must be + * zero-terminated.

+ * @param inParaLevel A base embedding level of the input as defined in + * ubidi_setPara documentation for the + * paraLevel parameter. + * @param inOrder An order of the input, which can be one of the + * UBiDiOrder values. + * @param outParaLevel A base embedding level of the output as defined in + * ubidi_setPara documentation for the + * paraLevel parameter. + * @param outOrder An order of the output, which can be one of the + * UBiDiOrder values. + * @param doMirroring Indicates whether or not to perform character mirroring, + * and can accept one of the UBiDiMirroring values. + * @param shapingOptions Arabic digit and letter shaping options defined in the + * ushape.h documentation. + *

Note: Direction indicator options are computed by + * the transformation engine based on the effective ordering schemes, so + * user-defined direction indicators will be ignored.

+ * @param pErrorCode A pointer to an error code value. + * + * @return The destination length, i.e. the number of UChars written to + * dest. If the transformation fails, the return value + * will be 0 (and the error code will be written to + * pErrorCode). + * + * @see UBiDiLevel + * @see UBiDiOrder + * @see UBiDiMirroring + * @see ubidi_setPara + * @see u_shapeArabic + * @stable ICU 58 + */ +U_CAPI uint32_t U_EXPORT2 +ubiditransform_transform(UBiDiTransform *pBiDiTransform, + const UChar *src, int32_t srcLength, + UChar *dest, int32_t destSize, + UBiDiLevel inParaLevel, UBiDiOrder inOrder, + UBiDiLevel outParaLevel, UBiDiOrder outOrder, + UBiDiMirroring doMirroring, uint32_t shapingOptions, + UErrorCode *pErrorCode); + +/** + * Allocates a UBiDiTransform object. This object can be reused, + * e.g. with different ordering schemes, mirroring or shaping options.

+ * Note:The object can only be reused in the same thread. + * All other threads should allocate a new UBiDiTransform object + * before using it.

+ * Example of usage:

+ * 

+ * \code
+ * UErrorCode errorCode = U_ZERO_ERROR;
+ * // Open a new UBiDiTransform.
+ * UBiDiTransform* transform = ubiditransform_open(&errorCode);
+ * // Run a transformation.
+ * ubiditransform_transform(transform,
+ *          text1, -1, text2, -1,
+ *          UBIDI_RTL, UBIDI_LOGICAL,
+ *          UBIDI_LTR, UBIDI_VISUAL,
+ *          UBIDI_MIRRORING_ON,
+ *          U_SHAPE_DIGITS_EN2AN,
+ *          &errorCode);
+ * // Do something with the output text and invoke another transformation using
+ * //   that text as input.
+ * ubiditransform_transform(transform,
+ *          text2, -1, text3, -1,
+ *          UBIDI_LTR, UBIDI_VISUAL,
+ *          UBIDI_RTL, UBIDI_VISUAL,
+ *          UBIDI_MIRRORING_ON,
+ *          0, &errorCode);
+ *\endcode
+ *
+ *

+ * The UBiDiTransform object must be deallocated by calling + * ubiditransform_close(). + * + * @return An empty UBiDiTransform object. + * @stable ICU 58 + */ +U_CAPI UBiDiTransform* U_EXPORT2 +ubiditransform_open(UErrorCode *pErrorCode); + +/** + * Deallocates the given UBiDiTransform object. + * @stable ICU 58 + */ +U_CAPI void U_EXPORT2 +ubiditransform_close(UBiDiTransform *pBidiTransform); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUBiDiTransformPointer + * "Smart pointer" class, closes a UBiDiTransform via ubiditransform_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 58 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUBiDiTransformPointer, UBiDiTransform, ubiditransform_close); + +U_NAMESPACE_END + +#endif + +#endif diff --git a/intl/icu/source/common/unicode/ubrk.h b/intl/icu/source/common/unicode/ubrk.h new file mode 100644 index 0000000000..2b3dc7aa57 --- /dev/null +++ b/intl/icu/source/common/unicode/ubrk.h @@ -0,0 +1,647 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 1996-2015, International Business Machines Corporation and others. +* All Rights Reserved. +****************************************************************************** +*/ + +#ifndef UBRK_H +#define UBRK_H + +#include "unicode/utypes.h" +#include "unicode/uloc.h" +#include "unicode/utext.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +/** + * A text-break iterator. + * For usage in C programs. + */ +#ifndef UBRK_TYPEDEF_UBREAK_ITERATOR +# define UBRK_TYPEDEF_UBREAK_ITERATOR + /** + * Opaque type representing an ICU Break iterator object. + * @stable ICU 2.0 + */ + typedef struct UBreakIterator UBreakIterator; +#endif + +#if !UCONFIG_NO_BREAK_ITERATION + +#include "unicode/parseerr.h" + +/** + * \file + * \brief C API: BreakIterator + * + *

BreakIterator C API

+ * + * The BreakIterator C API defines methods for finding the location + * of boundaries in text. Pointer to a UBreakIterator maintain a + * current position and scan over text returning the index of characters + * where boundaries occur. + *

+ * Line boundary analysis determines where a text string can be broken + * when line-wrapping. The mechanism correctly handles punctuation and + * hyphenated words. + *

+ * Note: The locale keyword "lb" can be used to modify line break + * behavior according to the CSS level 3 line-break options, see + * . For example: + * "ja@lb=strict", "zh@lb=loose". + *

+ * Sentence boundary analysis allows selection with correct + * interpretation of periods within numbers and abbreviations, and + * trailing punctuation marks such as quotation marks and parentheses. + *

+ * Note: The locale keyword "ss" can be used to enable use of + * segmentation suppression data (preventing breaks in English after + * abbreviations such as "Mr." or "Est.", for example), as follows: + * "en@ss=standard". + *

+ * Word boundary analysis is used by search and replace functions, as + * well as within text editing applications that allow the user to + * select words with a double click. Word selection provides correct + * interpretation of punctuation marks within and following + * words. Characters that are not part of a word, such as symbols or + * punctuation marks, have word-breaks on both sides. + *

+ * Character boundary analysis identifies the boundaries of + * "Extended Grapheme Clusters", which are groupings of codepoints + * that should be treated as character-like units for many text operations. + * Please see Unicode Standard Annex #29, Unicode Text Segmentation, + * http://www.unicode.org/reports/tr29/ for additional information + * on grapheme clusters and guidelines on their use. + *

+ * Title boundary analysis locates all positions, + * typically starts of words, that should be set to Title Case + * when title casing the text. + *

+ * The text boundary positions are found according to the rules + * described in Unicode Standard Annex #29, Text Boundaries, and + * Unicode Standard Annex #14, Line Breaking Properties. These + * are available at http://www.unicode.org/reports/tr14/ and + * http://www.unicode.org/reports/tr29/. + *

+ * In addition to the plain C API defined in this header file, an + * object oriented C++ API with equivalent functionality is defined in the + * file brkiter.h. + *

+ * Code snippets illustrating the use of the Break Iterator APIs + * are available in the ICU User Guide, + * https://unicode-org.github.io/icu/userguide/boundaryanalysis/ + * and in the sample program icu/source/samples/break/break.cpp + */ + +/** The possible types of text boundaries. @stable ICU 2.0 */ +typedef enum UBreakIteratorType { + /** Character breaks @stable ICU 2.0 */ + UBRK_CHARACTER = 0, + /** Word breaks @stable ICU 2.0 */ + UBRK_WORD = 1, + /** Line breaks @stable ICU 2.0 */ + UBRK_LINE = 2, + /** Sentence breaks @stable ICU 2.0 */ + UBRK_SENTENCE = 3, + +#ifndef U_HIDE_DEPRECATED_API + /** + * Title Case breaks + * The iterator created using this type locates title boundaries as described for + * Unicode 3.2 only. For Unicode 4.0 and above title boundary iteration, + * please use Word Boundary iterator. + * + * @deprecated ICU 2.8 Use the word break iterator for titlecasing for Unicode 4 and later. + */ + UBRK_TITLE = 4, + /** + * One more than the highest normal UBreakIteratorType value. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + UBRK_COUNT = 5 +#endif // U_HIDE_DEPRECATED_API +} UBreakIteratorType; + +/** Value indicating all text boundaries have been returned. + * @stable ICU 2.0 + */ +#define UBRK_DONE ((int32_t) -1) + + +/** + * Enum constants for the word break tags returned by + * getRuleStatus(). A range of values is defined for each category of + * word, to allow for further subdivisions of a category in future releases. + * Applications should check for tag values falling within the range, rather + * than for single individual values. + * + * The numeric values of all of these constants are stable (will not change). + * + * @stable ICU 2.2 +*/ +typedef enum UWordBreak { + /** Tag value for "words" that do not fit into any of other categories. + * Includes spaces and most punctuation. */ + UBRK_WORD_NONE = 0, + /** Upper bound for tags for uncategorized words. */ + UBRK_WORD_NONE_LIMIT = 100, + /** Tag value for words that appear to be numbers, lower limit. */ + UBRK_WORD_NUMBER = 100, + /** Tag value for words that appear to be numbers, upper limit. */ + UBRK_WORD_NUMBER_LIMIT = 200, + /** Tag value for words that contain letters, excluding + * hiragana, katakana or ideographic characters, lower limit. */ + UBRK_WORD_LETTER = 200, + /** Tag value for words containing letters, upper limit */ + UBRK_WORD_LETTER_LIMIT = 300, + /** Tag value for words containing kana characters, lower limit */ + UBRK_WORD_KANA = 300, + /** Tag value for words containing kana characters, upper limit */ + UBRK_WORD_KANA_LIMIT = 400, + /** Tag value for words containing ideographic characters, lower limit */ + UBRK_WORD_IDEO = 400, + /** Tag value for words containing ideographic characters, upper limit */ + UBRK_WORD_IDEO_LIMIT = 500 +} UWordBreak; + +/** + * Enum constants for the line break tags returned by getRuleStatus(). + * A range of values is defined for each category of + * word, to allow for further subdivisions of a category in future releases. + * Applications should check for tag values falling within the range, rather + * than for single individual values. + * + * The numeric values of all of these constants are stable (will not change). + * + * @stable ICU 2.8 +*/ +typedef enum ULineBreakTag { + /** Tag value for soft line breaks, positions at which a line break + * is acceptable but not required */ + UBRK_LINE_SOFT = 0, + /** Upper bound for soft line breaks. */ + UBRK_LINE_SOFT_LIMIT = 100, + /** Tag value for a hard, or mandatory line break */ + UBRK_LINE_HARD = 100, + /** Upper bound for hard line breaks. */ + UBRK_LINE_HARD_LIMIT = 200 +} ULineBreakTag; + + + +/** + * Enum constants for the sentence break tags returned by getRuleStatus(). + * A range of values is defined for each category of + * sentence, to allow for further subdivisions of a category in future releases. + * Applications should check for tag values falling within the range, rather + * than for single individual values. + * + * The numeric values of all of these constants are stable (will not change). + * + * @stable ICU 2.8 +*/ +typedef enum USentenceBreakTag { + /** Tag value for for sentences ending with a sentence terminator + * ('.', '?', '!', etc.) character, possibly followed by a + * hard separator (CR, LF, PS, etc.) + */ + UBRK_SENTENCE_TERM = 0, + /** Upper bound for tags for sentences ended by sentence terminators. */ + UBRK_SENTENCE_TERM_LIMIT = 100, + /** Tag value for for sentences that do not contain an ending + * sentence terminator ('.', '?', '!', etc.) character, but + * are ended only by a hard separator (CR, LF, PS, etc.) or end of input. + */ + UBRK_SENTENCE_SEP = 100, + /** Upper bound for tags for sentences ended by a separator. */ + UBRK_SENTENCE_SEP_LIMIT = 200 + /** Tag value for a hard, or mandatory line break */ +} USentenceBreakTag; + + +/** + * Open a new UBreakIterator for locating text boundaries for a specified locale. + * A UBreakIterator may be used for detecting character, line, word, + * and sentence breaks in text. + * @param type The type of UBreakIterator to open: one of UBRK_CHARACTER, UBRK_WORD, + * UBRK_LINE, UBRK_SENTENCE + * @param locale The locale specifying the text-breaking conventions. Note that + * locale keys such as "lb" and "ss" may be used to modify text break behavior, + * see general discussion of BreakIterator C API. + * @param text The text to be iterated over. May be null, in which case ubrk_setText() is + * used to specify the text to be iterated. + * @param textLength The number of characters in text, or -1 if null-terminated. + * @param status A UErrorCode to receive any errors. + * @return A UBreakIterator for the specified locale. + * @see ubrk_openRules + * @stable ICU 2.0 + */ +U_CAPI UBreakIterator* U_EXPORT2 +ubrk_open(UBreakIteratorType type, + const char *locale, + const UChar *text, + int32_t textLength, + UErrorCode *status); + +/** + * Open a new UBreakIterator for locating text boundaries using specified breaking rules. + * The rule syntax is ... (TBD) + * @param rules A set of rules specifying the text breaking conventions. + * @param rulesLength The number of characters in rules, or -1 if null-terminated. + * @param text The text to be iterated over. May be null, in which case ubrk_setText() is + * used to specify the text to be iterated. + * @param textLength The number of characters in text, or -1 if null-terminated. + * @param parseErr Receives position and context information for any syntax errors + * detected while parsing the rules. + * @param status A UErrorCode to receive any errors. + * @return A UBreakIterator for the specified rules. + * @see ubrk_open + * @stable ICU 2.2 + */ +U_CAPI UBreakIterator* U_EXPORT2 +ubrk_openRules(const UChar *rules, + int32_t rulesLength, + const UChar *text, + int32_t textLength, + UParseError *parseErr, + UErrorCode *status); + +/** + * Open a new UBreakIterator for locating text boundaries using precompiled binary rules. + * Opening a UBreakIterator this way is substantially faster than using ubrk_openRules. + * Binary rules may be obtained using ubrk_getBinaryRules. The compiled rules are not + * compatible across different major versions of ICU, nor across platforms of different + * endianness or different base character set family (ASCII vs EBCDIC). + * @param binaryRules A set of compiled binary rules specifying the text breaking + * conventions. Ownership of the storage containing the compiled + * rules remains with the caller of this function. The compiled + * rules must not be modified or deleted during the life of the + * break iterator. + * @param rulesLength The length of binaryRules in bytes; must be >= 0. + * @param text The text to be iterated over. May be null, in which case + * ubrk_setText() is used to specify the text to be iterated. + * @param textLength The number of characters in text, or -1 if null-terminated. + * @param status Pointer to UErrorCode to receive any errors. + * @return UBreakIterator for the specified rules. + * @see ubrk_getBinaryRules + * @stable ICU 59 + */ +U_CAPI UBreakIterator* U_EXPORT2 +ubrk_openBinaryRules(const uint8_t *binaryRules, int32_t rulesLength, + const UChar * text, int32_t textLength, + UErrorCode * status); + +#ifndef U_HIDE_DEPRECATED_API + +/** + * Thread safe cloning operation + * @param bi iterator to be cloned + * @param stackBuffer Deprecated functionality as of ICU 52, use NULL.
+ * user allocated space for the new clone. If NULL new memory will be allocated. + * If buffer is not large enough, new memory will be allocated. + * Clients can use the U_BRK_SAFECLONE_BUFFERSIZE. + * @param pBufferSize Deprecated functionality as of ICU 52, use NULL or 1.
+ * pointer to size of allocated space. + * If *pBufferSize == 0, a sufficient size for use in cloning will + * be returned ('pre-flighting') + * If *pBufferSize is not enough for a stack-based safe clone, + * new memory will be allocated. + * @param status to indicate whether the operation went on smoothly or there were errors + * An informational status value, U_SAFECLONE_ALLOCATED_ERROR, is used + * if pBufferSize != NULL and any allocations were necessary + * @return pointer to the new clone + * @deprecated ICU 69 Use ubrk_clone() instead. + */ +U_DEPRECATED UBreakIterator * U_EXPORT2 +ubrk_safeClone( + const UBreakIterator *bi, + void *stackBuffer, + int32_t *pBufferSize, + UErrorCode *status); + +#endif /* U_HIDE_DEPRECATED_API */ + +/** + * Thread safe cloning operation. + * @param bi iterator to be cloned + * @param status to indicate whether the operation went on smoothly or there were errors + * @return pointer to the new clone + * @stable ICU 69 + */ +U_CAPI UBreakIterator * U_EXPORT2 +ubrk_clone(const UBreakIterator *bi, + UErrorCode *status); + +#ifndef U_HIDE_DEPRECATED_API + +/** + * A recommended size (in bytes) for the memory buffer to be passed to ubrk_saveClone(). + * @deprecated ICU 52. Do not rely on ubrk_safeClone() cloning into any provided buffer. + */ +#define U_BRK_SAFECLONE_BUFFERSIZE 1 + +#endif /* U_HIDE_DEPRECATED_API */ + +/** +* Close a UBreakIterator. +* Once closed, a UBreakIterator may no longer be used. +* @param bi The break iterator to close. + * @stable ICU 2.0 +*/ +U_CAPI void U_EXPORT2 +ubrk_close(UBreakIterator *bi); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUBreakIteratorPointer + * "Smart pointer" class, closes a UBreakIterator via ubrk_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUBreakIteratorPointer, UBreakIterator, ubrk_close); + +U_NAMESPACE_END + +#endif + +/** + * Sets an existing iterator to point to a new piece of text. + * The break iterator retains a pointer to the supplied text. + * The caller must not modify or delete the text while the BreakIterator + * retains the reference. + * + * @param bi The iterator to use + * @param text The text to be set + * @param textLength The length of the text + * @param status The error code + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ubrk_setText(UBreakIterator* bi, + const UChar* text, + int32_t textLength, + UErrorCode* status); + + +/** + * Sets an existing iterator to point to a new piece of text. + * + * All index positions returned by break iterator functions are + * native indices from the UText. For example, when breaking UTF-8 + * encoded text, the break positions returned by \ref ubrk_next, \ref ubrk_previous, etc. + * will be UTF-8 string indices, not UTF-16 positions. + * + * @param bi The iterator to use + * @param text The text to be set. + * This function makes a shallow clone of the supplied UText. This means + * that the caller is free to immediately close or otherwise reuse the + * UText that was passed as a parameter, but that the underlying text itself + * must not be altered while being referenced by the break iterator. + * @param status The error code + * @stable ICU 3.4 + */ +U_CAPI void U_EXPORT2 +ubrk_setUText(UBreakIterator* bi, + UText* text, + UErrorCode* status); + + + +/** + * Determine the most recently-returned text boundary. + * + * @param bi The break iterator to use. + * @return The character index most recently returned by \ref ubrk_next, \ref ubrk_previous, + * \ref ubrk_first, or \ref ubrk_last. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ubrk_current(const UBreakIterator *bi); + +/** + * Advance the iterator to the boundary following the current boundary. + * + * @param bi The break iterator to use. + * @return The character index of the next text boundary, or UBRK_DONE + * if all text boundaries have been returned. + * @see ubrk_previous + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ubrk_next(UBreakIterator *bi); + +/** + * Set the iterator position to the boundary preceding the current boundary. + * + * @param bi The break iterator to use. + * @return The character index of the preceding text boundary, or UBRK_DONE + * if all text boundaries have been returned. + * @see ubrk_next + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ubrk_previous(UBreakIterator *bi); + +/** + * Set the iterator position to zero, the start of the text being scanned. + * @param bi The break iterator to use. + * @return The new iterator position (zero). + * @see ubrk_last + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ubrk_first(UBreakIterator *bi); + +/** + * Set the iterator position to the index immediately beyond the last character in the text being scanned. + * This is not the same as the last character. + * @param bi The break iterator to use. + * @return The character offset immediately beyond the last character in the + * text being scanned. + * @see ubrk_first + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ubrk_last(UBreakIterator *bi); + +/** + * Set the iterator position to the first boundary preceding the specified offset. + * The new position is always smaller than offset, or UBRK_DONE. + * @param bi The break iterator to use. + * @param offset The offset to begin scanning. + * @return The text boundary preceding offset, or UBRK_DONE. + * @see ubrk_following + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ubrk_preceding(UBreakIterator *bi, + int32_t offset); + +/** + * Advance the iterator to the first boundary following the specified offset. + * The value returned is always greater than offset, or UBRK_DONE. + * @param bi The break iterator to use. + * @param offset The offset to begin scanning. + * @return The text boundary following offset, or UBRK_DONE. + * @see ubrk_preceding + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ubrk_following(UBreakIterator *bi, + int32_t offset); + +/** +* Get a locale for which text breaking information is available. +* A UBreakIterator in a locale returned by this function will perform the correct +* text breaking for the locale. +* @param index The index of the desired locale. +* @return A locale for which number text breaking information is available, or 0 if none. +* @see ubrk_countAvailable +* @stable ICU 2.0 +*/ +U_CAPI const char* U_EXPORT2 +ubrk_getAvailable(int32_t index); + +/** +* Determine how many locales have text breaking information available. +* This function is most useful as determining the loop ending condition for +* calls to \ref ubrk_getAvailable. +* @return The number of locales for which text breaking information is available. +* @see ubrk_getAvailable +* @stable ICU 2.0 +*/ +U_CAPI int32_t U_EXPORT2 +ubrk_countAvailable(void); + + +/** +* Returns true if the specified position is a boundary position. As a side +* effect, leaves the iterator pointing to the first boundary position at +* or after "offset". +* @param bi The break iterator to use. +* @param offset the offset to check. +* @return True if "offset" is a boundary position. +* @stable ICU 2.0 +*/ +U_CAPI UBool U_EXPORT2 +ubrk_isBoundary(UBreakIterator *bi, int32_t offset); + +/** + * Return the status from the break rule that determined the most recently + * returned break position. The values appear in the rule source + * within brackets, {123}, for example. For rules that do not specify a + * status, a default value of 0 is returned. + *

+ * For word break iterators, the possible values are defined in enum UWordBreak. + * @stable ICU 2.2 + */ +U_CAPI int32_t U_EXPORT2 +ubrk_getRuleStatus(UBreakIterator *bi); + +/** + * Get the statuses from the break rules that determined the most recently + * returned break position. The values appear in the rule source + * within brackets, {123}, for example. The default status value for rules + * that do not explicitly provide one is zero. + *

+ * For word break iterators, the possible values are defined in enum UWordBreak. + * @param bi The break iterator to use + * @param fillInVec an array to be filled in with the status values. + * @param capacity the length of the supplied vector. A length of zero causes + * the function to return the number of status values, in the + * normal way, without attempting to store any values. + * @param status receives error codes. + * @return The number of rule status values from rules that determined + * the most recent boundary returned by the break iterator. + * @stable ICU 3.0 + */ +U_CAPI int32_t U_EXPORT2 +ubrk_getRuleStatusVec(UBreakIterator *bi, int32_t *fillInVec, int32_t capacity, UErrorCode *status); + +/** + * Return the locale of the break iterator. You can choose between the valid and + * the actual locale. + * @param bi break iterator + * @param type locale type (valid or actual) + * @param status error code + * @return locale string + * @stable ICU 2.8 + */ +U_CAPI const char* U_EXPORT2 +ubrk_getLocaleByType(const UBreakIterator *bi, ULocDataLocaleType type, UErrorCode* status); + +/** + * Set the subject text string upon which the break iterator is operating + * without changing any other aspect of the state. + * The new and previous text strings must have the same content. + * + * This function is intended for use in environments where ICU is operating on + * strings that may move around in memory. It provides a mechanism for notifying + * ICU that the string has been relocated, and providing a new UText to access the + * string in its new position. + * + * Note that the break iterator never copies the underlying text + * of a string being processed, but always operates directly on the original text + * provided by the user. Refreshing simply drops the references to the old text + * and replaces them with references to the new. + * + * Caution: this function is normally used only by very specialized + * system-level code. One example use case is with garbage collection + * that moves the text in memory. + * + * @param bi The break iterator. + * @param text The new (moved) text string. + * @param status Receives errors detected by this function. + * + * @stable ICU 49 + */ +U_CAPI void U_EXPORT2 +ubrk_refreshUText(UBreakIterator *bi, + UText *text, + UErrorCode *status); + + +/** + * Get a compiled binary version of the rules specifying the behavior of a UBreakIterator. + * The binary rules may be used with ubrk_openBinaryRules to open a new UBreakIterator + * more quickly than using ubrk_openRules. The compiled rules are not compatible across + * different major versions of ICU, nor across platforms of different endianness or + * different base character set family (ASCII vs EBCDIC). Supports preflighting (with + * binaryRules=NULL and rulesCapacity=0) to get the rules length without copying them to + * the binaryRules buffer. However, whether preflighting or not, if the actual length + * is greater than INT32_MAX, then the function returns 0 and sets *status to + * U_INDEX_OUTOFBOUNDS_ERROR. + + * @param bi The break iterator to use. + * @param binaryRules Buffer to receive the compiled binary rules; set to NULL for + * preflighting. + * @param rulesCapacity Capacity (in bytes) of the binaryRules buffer; set to 0 for + * preflighting. Must be >= 0. + * @param status Pointer to UErrorCode to receive any errors, such as + * U_BUFFER_OVERFLOW_ERROR, U_INDEX_OUTOFBOUNDS_ERROR, or + * U_ILLEGAL_ARGUMENT_ERROR. + * @return The actual byte length of the binary rules, if <= INT32_MAX; + * otherwise 0. If not preflighting and this is larger than + * rulesCapacity, *status will be set to an error. + * @see ubrk_openBinaryRules + * @stable ICU 59 + */ +U_CAPI int32_t U_EXPORT2 +ubrk_getBinaryRules(UBreakIterator *bi, + uint8_t * binaryRules, int32_t rulesCapacity, + UErrorCode * status); + +#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ + +#endif diff --git a/intl/icu/source/common/unicode/ucasemap.h b/intl/icu/source/common/unicode/ucasemap.h new file mode 100644 index 0000000000..d1c1b483ab --- /dev/null +++ b/intl/icu/source/common/unicode/ucasemap.h @@ -0,0 +1,388 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2005-2012, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: ucasemap.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2005may06 +* created by: Markus W. Scherer +* +* Case mapping service object and functions using it. +*/ + +#ifndef __UCASEMAP_H__ +#define __UCASEMAP_H__ + +#include "unicode/utypes.h" +#include "unicode/stringoptions.h" +#include "unicode/ustring.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +/** + * \file + * \brief C API: Unicode case mapping functions using a UCaseMap service object. + * + * The service object takes care of memory allocations, data loading, and setup + * for the attributes, as usual. + * + * Currently, the functionality provided here does not overlap with uchar.h + * and ustring.h, except for ucasemap_toTitle(). + * + * ucasemap_utf8XYZ() functions operate directly on UTF-8 strings. + */ + +/** + * UCaseMap is an opaque service object for newer ICU case mapping functions. + * Older functions did not use a service object. + * @stable ICU 3.4 + */ +struct UCaseMap; +typedef struct UCaseMap UCaseMap; /**< C typedef for struct UCaseMap. @stable ICU 3.4 */ + +/** + * Open a UCaseMap service object for a locale and a set of options. + * The locale ID and options are preprocessed so that functions using the + * service object need not process them in each call. + * + * @param locale ICU locale ID, used for language-dependent + * upper-/lower-/title-casing according to the Unicode standard. + * Usual semantics: ""=root, NULL=default locale, etc. + * @param options Options bit set, used for case folding and string comparisons. + * Same flags as for u_foldCase(), u_strFoldCase(), + * u_strCaseCompare(), etc. + * Use 0 or U_FOLD_CASE_DEFAULT for default behavior. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return Pointer to a UCaseMap service object, if successful. + * + * @see U_FOLD_CASE_DEFAULT + * @see U_FOLD_CASE_EXCLUDE_SPECIAL_I + * @see U_TITLECASE_NO_LOWERCASE + * @see U_TITLECASE_NO_BREAK_ADJUSTMENT + * @stable ICU 3.4 + */ +U_CAPI UCaseMap * U_EXPORT2 +ucasemap_open(const char *locale, uint32_t options, UErrorCode *pErrorCode); + +/** + * Close a UCaseMap service object. + * @param csm Object to be closed. + * @stable ICU 3.4 + */ +U_CAPI void U_EXPORT2 +ucasemap_close(UCaseMap *csm); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUCaseMapPointer + * "Smart pointer" class, closes a UCaseMap via ucasemap_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUCaseMapPointer, UCaseMap, ucasemap_close); + +U_NAMESPACE_END + +#endif + +/** + * Get the locale ID that is used for language-dependent case mappings. + * @param csm UCaseMap service object. + * @return locale ID + * @stable ICU 3.4 + */ +U_CAPI const char * U_EXPORT2 +ucasemap_getLocale(const UCaseMap *csm); + +/** + * Get the options bit set that is used for case folding and string comparisons. + * @param csm UCaseMap service object. + * @return options bit set + * @stable ICU 3.4 + */ +U_CAPI uint32_t U_EXPORT2 +ucasemap_getOptions(const UCaseMap *csm); + +/** + * Set the locale ID that is used for language-dependent case mappings. + * + * @param csm UCaseMap service object. + * @param locale Locale ID, see ucasemap_open(). + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * + * @see ucasemap_open + * @stable ICU 3.4 + */ +U_CAPI void U_EXPORT2 +ucasemap_setLocale(UCaseMap *csm, const char *locale, UErrorCode *pErrorCode); + +/** + * Set the options bit set that is used for case folding and string comparisons. + * + * @param csm UCaseMap service object. + * @param options Options bit set, see ucasemap_open(). + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * + * @see ucasemap_open + * @stable ICU 3.4 + */ +U_CAPI void U_EXPORT2 +ucasemap_setOptions(UCaseMap *csm, uint32_t options, UErrorCode *pErrorCode); + +#if !UCONFIG_NO_BREAK_ITERATION + +/** + * Get the break iterator that is used for titlecasing. + * Do not modify the returned break iterator. + * @param csm UCaseMap service object. + * @return titlecasing break iterator + * @stable ICU 3.8 + */ +U_CAPI const UBreakIterator * U_EXPORT2 +ucasemap_getBreakIterator(const UCaseMap *csm); + +/** + * Set the break iterator that is used for titlecasing. + * The UCaseMap service object releases a previously set break iterator + * and "adopts" this new one, taking ownership of it. + * It will be released in a subsequent call to ucasemap_setBreakIterator() + * or ucasemap_close(). + * + * Break iterator operations are not thread-safe. Therefore, titlecasing + * functions use non-const UCaseMap objects. It is not possible to titlecase + * strings concurrently using the same UCaseMap. + * + * @param csm UCaseMap service object. + * @param iterToAdopt Break iterator to be adopted for titlecasing. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * + * @see ucasemap_toTitle + * @see ucasemap_utf8ToTitle + * @stable ICU 3.8 + */ +U_CAPI void U_EXPORT2 +ucasemap_setBreakIterator(UCaseMap *csm, UBreakIterator *iterToAdopt, UErrorCode *pErrorCode); + +/** + * Titlecase a UTF-16 string. This function is almost a duplicate of u_strToTitle(), + * except that it takes ucasemap_setOptions() into account and has performance + * advantages from being able to use a UCaseMap object for multiple case mapping + * operations, saving setup time. + * + * Casing is locale-dependent and context-sensitive. + * Titlecasing uses a break iterator to find the first characters of words + * that are to be titlecased. It titlecases those characters and lowercases + * all others. (This can be modified with ucasemap_setOptions().) + * + * Note: This function takes a non-const UCaseMap pointer because it will + * open a default break iterator if no break iterator was set yet, + * and effectively call ucasemap_setBreakIterator(); + * also because the break iterator is stateful and will be modified during + * the iteration. + * + * The titlecase break iterator can be provided to customize for arbitrary + * styles, using rules and dictionaries beyond the standard iterators. + * The standard titlecase iterator for the root locale implements the + * algorithm of Unicode TR 21. + * + * This function uses only the setText(), first() and next() methods of the + * provided break iterator. + * + * The result may be longer or shorter than the original. + * The source string and the destination buffer must not overlap. + * + * @param csm UCaseMap service object. This pointer is non-const! + * See the note above for details. + * @param dest A buffer for the result string. The result will be NUL-terminated if + * the buffer is large enough. + * The contents is undefined in case of failure. + * @param destCapacity The size of the buffer (number of UChars). If it is 0, then + * dest may be NULL and the function will only return the length of the result + * without writing any of the result string. + * @param src The original string. + * @param srcLength The length of the original string. If -1, then src must be NUL-terminated. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return The length of the result string, if successful - or in case of a buffer overflow, + * in which case it will be greater than destCapacity. + * + * @see u_strToTitle + * @stable ICU 3.8 + */ +U_CAPI int32_t U_EXPORT2 +ucasemap_toTitle(UCaseMap *csm, + UChar *dest, int32_t destCapacity, + const UChar *src, int32_t srcLength, + UErrorCode *pErrorCode); + +#endif // UCONFIG_NO_BREAK_ITERATION + +/** + * Lowercase the characters in a UTF-8 string. + * Casing is locale-dependent and context-sensitive. + * The result may be longer or shorter than the original. + * The source string and the destination buffer must not overlap. + * + * @param csm UCaseMap service object. + * @param dest A buffer for the result string. The result will be NUL-terminated if + * the buffer is large enough. + * The contents is undefined in case of failure. + * @param destCapacity The size of the buffer (number of bytes). If it is 0, then + * dest may be NULL and the function will only return the length of the result + * without writing any of the result string. + * @param src The original string. + * @param srcLength The length of the original string. If -1, then src must be NUL-terminated. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return The length of the result string, if successful - or in case of a buffer overflow, + * in which case it will be greater than destCapacity. + * + * @see u_strToLower + * @stable ICU 3.4 + */ +U_CAPI int32_t U_EXPORT2 +ucasemap_utf8ToLower(const UCaseMap *csm, + char *dest, int32_t destCapacity, + const char *src, int32_t srcLength, + UErrorCode *pErrorCode); + +/** + * Uppercase the characters in a UTF-8 string. + * Casing is locale-dependent and context-sensitive. + * The result may be longer or shorter than the original. + * The source string and the destination buffer must not overlap. + * + * @param csm UCaseMap service object. + * @param dest A buffer for the result string. The result will be NUL-terminated if + * the buffer is large enough. + * The contents is undefined in case of failure. + * @param destCapacity The size of the buffer (number of bytes). If it is 0, then + * dest may be NULL and the function will only return the length of the result + * without writing any of the result string. + * @param src The original string. + * @param srcLength The length of the original string. If -1, then src must be NUL-terminated. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return The length of the result string, if successful - or in case of a buffer overflow, + * in which case it will be greater than destCapacity. + * + * @see u_strToUpper + * @stable ICU 3.4 + */ +U_CAPI int32_t U_EXPORT2 +ucasemap_utf8ToUpper(const UCaseMap *csm, + char *dest, int32_t destCapacity, + const char *src, int32_t srcLength, + UErrorCode *pErrorCode); + +#if !UCONFIG_NO_BREAK_ITERATION + +/** + * Titlecase a UTF-8 string. + * Casing is locale-dependent and context-sensitive. + * Titlecasing uses a break iterator to find the first characters of words + * that are to be titlecased. It titlecases those characters and lowercases + * all others. (This can be modified with ucasemap_setOptions().) + * + * Note: This function takes a non-const UCaseMap pointer because it will + * open a default break iterator if no break iterator was set yet, + * and effectively call ucasemap_setBreakIterator(); + * also because the break iterator is stateful and will be modified during + * the iteration. + * + * The titlecase break iterator can be provided to customize for arbitrary + * styles, using rules and dictionaries beyond the standard iterators. + * The standard titlecase iterator for the root locale implements the + * algorithm of Unicode TR 21. + * + * This function uses only the setUText(), first(), next() and close() methods of the + * provided break iterator. + * + * The result may be longer or shorter than the original. + * The source string and the destination buffer must not overlap. + * + * @param csm UCaseMap service object. This pointer is non-const! + * See the note above for details. + * @param dest A buffer for the result string. The result will be NUL-terminated if + * the buffer is large enough. + * The contents is undefined in case of failure. + * @param destCapacity The size of the buffer (number of bytes). If it is 0, then + * dest may be NULL and the function will only return the length of the result + * without writing any of the result string. + * @param src The original string. + * @param srcLength The length of the original string. If -1, then src must be NUL-terminated. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return The length of the result string, if successful - or in case of a buffer overflow, + * in which case it will be greater than destCapacity. + * + * @see u_strToTitle + * @see U_TITLECASE_NO_LOWERCASE + * @see U_TITLECASE_NO_BREAK_ADJUSTMENT + * @stable ICU 3.8 + */ +U_CAPI int32_t U_EXPORT2 +ucasemap_utf8ToTitle(UCaseMap *csm, + char *dest, int32_t destCapacity, + const char *src, int32_t srcLength, + UErrorCode *pErrorCode); + +#endif + +/** + * Case-folds the characters in a UTF-8 string. + * + * Case-folding is locale-independent and not context-sensitive, + * but there is an option for whether to include or exclude mappings for dotted I + * and dotless i that are marked with 'T' in CaseFolding.txt. + * + * The result may be longer or shorter than the original. + * The source string and the destination buffer must not overlap. + * + * @param csm UCaseMap service object. + * @param dest A buffer for the result string. The result will be NUL-terminated if + * the buffer is large enough. + * The contents is undefined in case of failure. + * @param destCapacity The size of the buffer (number of bytes). If it is 0, then + * dest may be NULL and the function will only return the length of the result + * without writing any of the result string. + * @param src The original string. + * @param srcLength The length of the original string. If -1, then src must be NUL-terminated. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return The length of the result string, if successful - or in case of a buffer overflow, + * in which case it will be greater than destCapacity. + * + * @see u_strFoldCase + * @see ucasemap_setOptions + * @see U_FOLD_CASE_DEFAULT + * @see U_FOLD_CASE_EXCLUDE_SPECIAL_I + * @stable ICU 3.8 + */ +U_CAPI int32_t U_EXPORT2 +ucasemap_utf8FoldCase(const UCaseMap *csm, + char *dest, int32_t destCapacity, + const char *src, int32_t srcLength, + UErrorCode *pErrorCode); + +#endif diff --git a/intl/icu/source/common/unicode/ucat.h b/intl/icu/source/common/unicode/ucat.h new file mode 100644 index 0000000000..93850348ff --- /dev/null +++ b/intl/icu/source/common/unicode/ucat.h @@ -0,0 +1,160 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2003-2004, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* Author: Alan Liu +* Created: March 19 2003 +* Since: ICU 2.6 +********************************************************************** +*/ +#ifndef UCAT_H +#define UCAT_H + +#include "unicode/utypes.h" +#include "unicode/ures.h" + +/** + * \file + * \brief C API: Message Catalog Wrappers + * + * This C API provides look-alike functions that deliberately resemble + * the POSIX catopen, catclose, and catgets functions. The underlying + * implementation is in terms of ICU resource bundles, rather than + * POSIX message catalogs. + * + * The ICU resource bundles obey standard ICU inheritance policies. + * To facilitate this, sets and messages are flattened into one tier. + * This is done by creating resource bundle keys of the form + * <set_num>%<msg_num> where set_num is the set number and msg_num is + * the message number, formatted as decimal strings. + * + * Example: Consider a message catalog containing two sets: + * + * Set 1: Message 4 = "Good morning." + * Message 5 = "Good afternoon." + * Message 7 = "Good evening." + * Message 8 = "Good night." + * Set 4: Message 14 = "Please " + * Message 19 = "Thank you." + * Message 20 = "Sincerely," + * + * The ICU resource bundle source file would, assuming it is named + * "greet.txt", would look like this: + * + * greet + * { + * 1%4 { "Good morning." } + * 1%5 { "Good afternoon." } + * 1%7 { "Good evening." } + * 1%8 { "Good night." } + * + * 4%14 { "Please " } + * 4%19 { "Thank you." } + * 4%20 { "Sincerely," } + * } + * + * The catgets function is commonly used in combination with functions + * like printf and strftime. ICU components like message format can + * be used instead, although they use a different format syntax. + * There is an ICU package, icuio, that provides some of + * the POSIX-style formatting API. + */ + +U_CDECL_BEGIN + +/** + * An ICU message catalog descriptor, analogous to nl_catd. + * + * @stable ICU 2.6 + */ +typedef UResourceBundle* u_nl_catd; + +/** + * Open and return an ICU message catalog descriptor. The descriptor + * may be passed to u_catgets() to retrieve localized strings. + * + * @param name string containing the full path pointing to the + * directory where the resources reside followed by the package name + * e.g. "/usr/resource/my_app/resources/guimessages" on a Unix system. + * If NULL, ICU default data files will be used. + * + * Unlike POSIX, environment variables are not interpolated within the + * name. + * + * @param locale the locale for which we want to open the resource. If + * NULL, the default ICU locale will be used (see uloc_getDefault). If + * strlen(locale) == 0, the root locale will be used. + * + * @param ec input/output error code. Upon output, + * U_USING_FALLBACK_WARNING indicates that a fallback locale was + * used. For example, 'de_CH' was requested, but nothing was found + * there, so 'de' was used. U_USING_DEFAULT_WARNING indicates that the + * default locale data or root locale data was used; neither the + * requested locale nor any of its fallback locales were found. + * + * @return a message catalog descriptor that may be passed to + * u_catgets(). If the ec parameter indicates success, then the caller + * is responsible for calling u_catclose() to close the message + * catalog. If the ec parameter indicates failure, then NULL will be + * returned. + * + * @stable ICU 2.6 + */ +U_CAPI u_nl_catd U_EXPORT2 +u_catopen(const char* name, const char* locale, UErrorCode* ec); + +/** + * Close an ICU message catalog, given its descriptor. + * + * @param catd a message catalog descriptor to be closed. May be NULL, + * in which case no action is taken. + * + * @stable ICU 2.6 + */ +U_CAPI void U_EXPORT2 +u_catclose(u_nl_catd catd); + +/** + * Retrieve a localized string from an ICU message catalog. + * + * @param catd a message catalog descriptor returned by u_catopen. + * + * @param set_num the message catalog set number. Sets need not be + * numbered consecutively. + * + * @param msg_num the message catalog message number within the + * set. Messages need not be numbered consecutively. + * + * @param s the default string. This is returned if the string + * specified by the set_num and msg_num is not found. It must be + * zero-terminated. + * + * @param len fill-in parameter to receive the length of the result. + * May be NULL, in which case it is ignored. + * + * @param ec input/output error code. May be U_USING_FALLBACK_WARNING + * or U_USING_DEFAULT_WARNING. U_MISSING_RESOURCE_ERROR indicates that + * the set_num/msg_num tuple does not specify a valid message string + * in this catalog. + * + * @return a pointer to a zero-terminated UChar array which lives in + * an internal buffer area, typically a memory mapped/DLL file. The + * caller must NOT delete this pointer. If the call is unsuccessful + * for any reason, then s is returned. This includes the situation in + * which ec indicates a failing error code upon entry to this + * function. + * + * @stable ICU 2.6 + */ +U_CAPI const UChar* U_EXPORT2 +u_catgets(u_nl_catd catd, int32_t set_num, int32_t msg_num, + const UChar* s, + int32_t* len, UErrorCode* ec); + +U_CDECL_END + +#endif /*UCAT_H*/ +/*eof*/ diff --git a/intl/icu/source/common/unicode/uchar.h b/intl/icu/source/common/unicode/uchar.h new file mode 100644 index 0000000000..4f82a9fb58 --- /dev/null +++ b/intl/icu/source/common/unicode/uchar.h @@ -0,0 +1,4169 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1997-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* +* File UCHAR.H +* +* Modification History: +* +* Date Name Description +* 04/02/97 aliu Creation. +* 03/29/99 helena Updated for C APIs. +* 4/15/99 Madhu Updated for C Implementation and Javadoc +* 5/20/99 Madhu Added the function u_getVersion() +* 8/19/1999 srl Upgraded scripts to Unicode 3.0 +* 8/27/1999 schererm UCharDirection constants: U_... +* 11/11/1999 weiv added u_isalnum(), cleaned comments +* 01/11/2000 helena Renamed u_getVersion to u_getUnicodeVersion(). +****************************************************************************** +*/ + +#ifndef UCHAR_H +#define UCHAR_H + +#include "unicode/utypes.h" +#include "unicode/stringoptions.h" +#include "unicode/ucpmap.h" + +#if !defined(USET_DEFINED) && !defined(U_IN_DOXYGEN) + +#define USET_DEFINED + +/** + * USet is the C API type corresponding to C++ class UnicodeSet. + * It is forward-declared here to avoid including unicode/uset.h file if related + * APIs are not used. + * + * @see ucnv_getUnicodeSet + * @stable ICU 2.4 + */ +typedef struct USet USet; + +#endif + + +U_CDECL_BEGIN + +/*==========================================================================*/ +/* Unicode version number */ +/*==========================================================================*/ +/** + * Unicode version number, default for the current ICU version. + * The actual Unicode Character Database (UCD) data is stored in uprops.dat + * and may be generated from UCD files from a different Unicode version. + * Call u_getUnicodeVersion to get the actual Unicode version of the data. + * + * @see u_getUnicodeVersion + * @stable ICU 2.0 + */ +#define U_UNICODE_VERSION "15.0" + +/** + * \file + * \brief C API: Unicode Properties + * + * This C API provides low-level access to the Unicode Character Database. + * In addition to raw property values, some convenience functions calculate + * derived properties, for example for Java-style programming. + * + * Unicode assigns each code point (not just assigned character) values for + * many properties. + * Most of them are simple boolean flags, or constants from a small enumerated list. + * For some properties, values are strings or other relatively more complex types. + * + * For more information see + * "About the Unicode Character Database" (http://www.unicode.org/ucd/) + * and the ICU User Guide chapter on Properties (https://unicode-org.github.io/icu/userguide/strings/properties). + * + * Many properties are accessible via generic functions that take a UProperty selector. + * - u_hasBinaryProperty() returns a binary value (true/false) per property and code point. + * - u_getIntPropertyValue() returns an integer value per property and code point. + * For each supported enumerated or catalog property, there is + * an enum type for all of the property's values, and + * u_getIntPropertyValue() returns the numeric values of those constants. + * - u_getBinaryPropertySet() returns a set for each ICU-supported binary property with + * all code points for which the property is true. + * - u_getIntPropertyMap() returns a map for each + * ICU-supported enumerated/catalog/int-valued property which + * maps all Unicode code points to their values for that property. + * + * Many functions are designed to match java.lang.Character functions. + * See the individual function documentation, + * and see the JDK 1.4 java.lang.Character documentation + * at http://java.sun.com/j2se/1.4/docs/api/java/lang/Character.html + * + * There are also functions that provide easy migration from C/POSIX functions + * like isblank(). Their use is generally discouraged because the C/POSIX + * standards do not define their semantics beyond the ASCII range, which means + * that different implementations exhibit very different behavior. + * Instead, Unicode properties should be used directly. + * + * There are also only a few, broad C/POSIX character classes, and they tend + * to be used for conflicting purposes. For example, the "isalpha()" class + * is sometimes used to determine word boundaries, while a more sophisticated + * approach would at least distinguish initial letters from continuation + * characters (the latter including combining marks). + * (In ICU, BreakIterator is the most sophisticated API for word boundaries.) + * Another example: There is no "istitle()" class for titlecase characters. + * + * ICU 3.4 and later provides API access for all twelve C/POSIX character classes. + * ICU implements them according to the Standard Recommendations in + * Annex C: Compatibility Properties of UTS #18 Unicode Regular Expressions + * (http://www.unicode.org/reports/tr18/#Compatibility_Properties). + * + * API access for C/POSIX character classes is as follows: + * - alpha: u_isUAlphabetic(c) or u_hasBinaryProperty(c, UCHAR_ALPHABETIC) + * - lower: u_isULowercase(c) or u_hasBinaryProperty(c, UCHAR_LOWERCASE) + * - upper: u_isUUppercase(c) or u_hasBinaryProperty(c, UCHAR_UPPERCASE) + * - punct: u_ispunct(c) + * - digit: u_isdigit(c) or u_charType(c)==U_DECIMAL_DIGIT_NUMBER + * - xdigit: u_isxdigit(c) or u_hasBinaryProperty(c, UCHAR_POSIX_XDIGIT) + * - alnum: u_hasBinaryProperty(c, UCHAR_POSIX_ALNUM) + * - space: u_isUWhiteSpace(c) or u_hasBinaryProperty(c, UCHAR_WHITE_SPACE) + * - blank: u_isblank(c) or u_hasBinaryProperty(c, UCHAR_POSIX_BLANK) + * - cntrl: u_charType(c)==U_CONTROL_CHAR + * - graph: u_hasBinaryProperty(c, UCHAR_POSIX_GRAPH) + * - print: u_hasBinaryProperty(c, UCHAR_POSIX_PRINT) + * + * Note: Some of the u_isxyz() functions in uchar.h predate, and do not match, + * the Standard Recommendations in UTS #18. Instead, they match Java + * functions according to their API documentation. + * + * \htmlonly + * The C/POSIX character classes are also available in UnicodeSet patterns, + * using patterns like [:graph:] or \p{graph}. + * \endhtmlonly + * + * Note: There are several ICU whitespace functions. + * Comparison: + * - u_isUWhiteSpace=UCHAR_WHITE_SPACE: Unicode White_Space property; + * most of general categories "Z" (separators) + most whitespace ISO controls + * (including no-break spaces, but excluding IS1..IS4) + * - u_isWhitespace: Java isWhitespace; Z + whitespace ISO controls but excluding no-break spaces + * - u_isJavaSpaceChar: Java isSpaceChar; just Z (including no-break spaces) + * - u_isspace: Z + whitespace ISO controls (including no-break spaces) + * - u_isblank: "horizontal spaces" = TAB + Zs + */ + +/** + * Constants. + */ + +/** The lowest Unicode code point value. Code points are non-negative. @stable ICU 2.0 */ +#define UCHAR_MIN_VALUE 0 + +/** + * The highest Unicode code point value (scalar value) according to + * The Unicode Standard. This is a 21-bit value (20.1 bits, rounded up). + * For a single character, UChar32 is a simple type that can hold any code point value. + * + * @see UChar32 + * @stable ICU 2.0 + */ +#define UCHAR_MAX_VALUE 0x10ffff + +/** + * Get a single-bit bit set (a flag) from a bit number 0..31. + * @stable ICU 2.1 + */ +#define U_MASK(x) ((uint32_t)1<<(x)) + +/** + * Selection constants for Unicode properties. + * These constants are used in functions like u_hasBinaryProperty to select + * one of the Unicode properties. + * + * The properties APIs are intended to reflect Unicode properties as defined + * in the Unicode Character Database (UCD) and Unicode Technical Reports (UTR). + * + * For details about the properties see + * UAX #44: Unicode Character Database (http://www.unicode.org/reports/tr44/). + * + * Important: If ICU is built with UCD files from Unicode versions below, e.g., 3.2, + * then properties marked with "new in Unicode 3.2" are not or not fully available. + * Check u_getUnicodeVersion to be sure. + * + * @see u_hasBinaryProperty + * @see u_getIntPropertyValue + * @see u_getUnicodeVersion + * @stable ICU 2.1 + */ +typedef enum UProperty { + /* + * Note: UProperty constants are parsed by preparseucd.py. + * It matches lines like + * UCHAR_=, + */ + + /* Note: Place UCHAR_ALPHABETIC before UCHAR_BINARY_START so that + debuggers display UCHAR_ALPHABETIC as the symbolic name for 0, + rather than UCHAR_BINARY_START. Likewise for other *_START + identifiers. */ + + /** Binary property Alphabetic. Same as u_isUAlphabetic, different from u_isalpha. + Lu+Ll+Lt+Lm+Lo+Nl+Other_Alphabetic @stable ICU 2.1 */ + UCHAR_ALPHABETIC=0, + /** First constant for binary Unicode properties. @stable ICU 2.1 */ + UCHAR_BINARY_START=UCHAR_ALPHABETIC, + /** Binary property ASCII_Hex_Digit. 0-9 A-F a-f @stable ICU 2.1 */ + UCHAR_ASCII_HEX_DIGIT=1, + /** Binary property Bidi_Control. + Format controls which have specific functions + in the Bidi Algorithm. @stable ICU 2.1 */ + UCHAR_BIDI_CONTROL=2, + /** Binary property Bidi_Mirrored. + Characters that may change display in RTL text. + Same as u_isMirrored. + See Bidi Algorithm, UTR 9. @stable ICU 2.1 */ + UCHAR_BIDI_MIRRORED=3, + /** Binary property Dash. Variations of dashes. @stable ICU 2.1 */ + UCHAR_DASH=4, + /** Binary property Default_Ignorable_Code_Point (new in Unicode 3.2). + Ignorable in most processing. + <2060..206F, FFF0..FFFB, E0000..E0FFF>+Other_Default_Ignorable_Code_Point+(Cf+Cc+Cs-White_Space) @stable ICU 2.1 */ + UCHAR_DEFAULT_IGNORABLE_CODE_POINT=5, + /** Binary property Deprecated (new in Unicode 3.2). + The usage of deprecated characters is strongly discouraged. @stable ICU 2.1 */ + UCHAR_DEPRECATED=6, + /** Binary property Diacritic. Characters that linguistically modify + the meaning of another character to which they apply. @stable ICU 2.1 */ + UCHAR_DIACRITIC=7, + /** Binary property Extender. + Extend the value or shape of a preceding alphabetic character, + e.g., length and iteration marks. @stable ICU 2.1 */ + UCHAR_EXTENDER=8, + /** Binary property Full_Composition_Exclusion. + CompositionExclusions.txt+Singleton Decompositions+ + Non-Starter Decompositions. @stable ICU 2.1 */ + UCHAR_FULL_COMPOSITION_EXCLUSION=9, + /** Binary property Grapheme_Base (new in Unicode 3.2). + For programmatic determination of grapheme cluster boundaries. + [0..10FFFF]-Cc-Cf-Cs-Co-Cn-Zl-Zp-Grapheme_Link-Grapheme_Extend-CGJ @stable ICU 2.1 */ + UCHAR_GRAPHEME_BASE=10, + /** Binary property Grapheme_Extend (new in Unicode 3.2). + For programmatic determination of grapheme cluster boundaries. + Me+Mn+Mc+Other_Grapheme_Extend-Grapheme_Link-CGJ @stable ICU 2.1 */ + UCHAR_GRAPHEME_EXTEND=11, + /** Binary property Grapheme_Link (new in Unicode 3.2). + For programmatic determination of grapheme cluster boundaries. @stable ICU 2.1 */ + UCHAR_GRAPHEME_LINK=12, + /** Binary property Hex_Digit. + Characters commonly used for hexadecimal numbers. @stable ICU 2.1 */ + UCHAR_HEX_DIGIT=13, + /** Binary property Hyphen. Dashes used to mark connections + between pieces of words, plus the Katakana middle dot. @stable ICU 2.1 */ + UCHAR_HYPHEN=14, + /** Binary property ID_Continue. + Characters that can continue an identifier. + DerivedCoreProperties.txt also says "NOTE: Cf characters should be filtered out." + ID_Start+Mn+Mc+Nd+Pc @stable ICU 2.1 */ + UCHAR_ID_CONTINUE=15, + /** Binary property ID_Start. + Characters that can start an identifier. + Lu+Ll+Lt+Lm+Lo+Nl @stable ICU 2.1 */ + UCHAR_ID_START=16, + /** Binary property Ideographic. + CJKV ideographs. @stable ICU 2.1 */ + UCHAR_IDEOGRAPHIC=17, + /** Binary property IDS_Binary_Operator (new in Unicode 3.2). + For programmatic determination of + Ideographic Description Sequences. @stable ICU 2.1 */ + UCHAR_IDS_BINARY_OPERATOR=18, + /** Binary property IDS_Trinary_Operator (new in Unicode 3.2). + For programmatic determination of + Ideographic Description Sequences. @stable ICU 2.1 */ + UCHAR_IDS_TRINARY_OPERATOR=19, + /** Binary property Join_Control. + Format controls for cursive joining and ligation. @stable ICU 2.1 */ + UCHAR_JOIN_CONTROL=20, + /** Binary property Logical_Order_Exception (new in Unicode 3.2). + Characters that do not use logical order and + require special handling in most processing. @stable ICU 2.1 */ + UCHAR_LOGICAL_ORDER_EXCEPTION=21, + /** Binary property Lowercase. Same as u_isULowercase, different from u_islower. + Ll+Other_Lowercase @stable ICU 2.1 */ + UCHAR_LOWERCASE=22, + /** Binary property Math. Sm+Other_Math @stable ICU 2.1 */ + UCHAR_MATH=23, + /** Binary property Noncharacter_Code_Point. + Code points that are explicitly defined as illegal + for the encoding of characters. @stable ICU 2.1 */ + UCHAR_NONCHARACTER_CODE_POINT=24, + /** Binary property Quotation_Mark. @stable ICU 2.1 */ + UCHAR_QUOTATION_MARK=25, + /** Binary property Radical (new in Unicode 3.2). + For programmatic determination of + Ideographic Description Sequences. @stable ICU 2.1 */ + UCHAR_RADICAL=26, + /** Binary property Soft_Dotted (new in Unicode 3.2). + Characters with a "soft dot", like i or j. + An accent placed on these characters causes + the dot to disappear. @stable ICU 2.1 */ + UCHAR_SOFT_DOTTED=27, + /** Binary property Terminal_Punctuation. + Punctuation characters that generally mark + the end of textual units. @stable ICU 2.1 */ + UCHAR_TERMINAL_PUNCTUATION=28, + /** Binary property Unified_Ideograph (new in Unicode 3.2). + For programmatic determination of + Ideographic Description Sequences. @stable ICU 2.1 */ + UCHAR_UNIFIED_IDEOGRAPH=29, + /** Binary property Uppercase. Same as u_isUUppercase, different from u_isupper. + Lu+Other_Uppercase @stable ICU 2.1 */ + UCHAR_UPPERCASE=30, + /** Binary property White_Space. + Same as u_isUWhiteSpace, different from u_isspace and u_isWhitespace. + Space characters+TAB+CR+LF-ZWSP-ZWNBSP @stable ICU 2.1 */ + UCHAR_WHITE_SPACE=31, + /** Binary property XID_Continue. + ID_Continue modified to allow closure under + normalization forms NFKC and NFKD. @stable ICU 2.1 */ + UCHAR_XID_CONTINUE=32, + /** Binary property XID_Start. ID_Start modified to allow + closure under normalization forms NFKC and NFKD. @stable ICU 2.1 */ + UCHAR_XID_START=33, + /** Binary property Case_Sensitive. Either the source of a case + mapping or _in_ the target of a case mapping. Not the same as + the general category Cased_Letter. @stable ICU 2.6 */ + UCHAR_CASE_SENSITIVE=34, + /** Binary property STerm (new in Unicode 4.0.1). + Sentence Terminal. Used in UAX #29: Text Boundaries + (http://www.unicode.org/reports/tr29/) + @stable ICU 3.0 */ + UCHAR_S_TERM=35, + /** Binary property Variation_Selector (new in Unicode 4.0.1). + Indicates all those characters that qualify as Variation Selectors. + For details on the behavior of these characters, + see StandardizedVariants.html and 15.6 Variation Selectors. + @stable ICU 3.0 */ + UCHAR_VARIATION_SELECTOR=36, + /** Binary property NFD_Inert. + ICU-specific property for characters that are inert under NFD, + i.e., they do not interact with adjacent characters. + See the documentation for the Normalizer2 class and the + Normalizer2::isInert() method. + @stable ICU 3.0 */ + UCHAR_NFD_INERT=37, + /** Binary property NFKD_Inert. + ICU-specific property for characters that are inert under NFKD, + i.e., they do not interact with adjacent characters. + See the documentation for the Normalizer2 class and the + Normalizer2::isInert() method. + @stable ICU 3.0 */ + UCHAR_NFKD_INERT=38, + /** Binary property NFC_Inert. + ICU-specific property for characters that are inert under NFC, + i.e., they do not interact with adjacent characters. + See the documentation for the Normalizer2 class and the + Normalizer2::isInert() method. + @stable ICU 3.0 */ + UCHAR_NFC_INERT=39, + /** Binary property NFKC_Inert. + ICU-specific property for characters that are inert under NFKC, + i.e., they do not interact with adjacent characters. + See the documentation for the Normalizer2 class and the + Normalizer2::isInert() method. + @stable ICU 3.0 */ + UCHAR_NFKC_INERT=40, + /** Binary Property Segment_Starter. + ICU-specific property for characters that are starters in terms of + Unicode normalization and combining character sequences. + They have ccc=0 and do not occur in non-initial position of the + canonical decomposition of any character + (like a-umlaut in NFD and a Jamo T in an NFD(Hangul LVT)). + ICU uses this property for segmenting a string for generating a set of + canonically equivalent strings, e.g. for canonical closure while + processing collation tailoring rules. + @stable ICU 3.0 */ + UCHAR_SEGMENT_STARTER=41, + /** Binary property Pattern_Syntax (new in Unicode 4.1). + See UAX #31 Identifier and Pattern Syntax + (http://www.unicode.org/reports/tr31/) + @stable ICU 3.4 */ + UCHAR_PATTERN_SYNTAX=42, + /** Binary property Pattern_White_Space (new in Unicode 4.1). + See UAX #31 Identifier and Pattern Syntax + (http://www.unicode.org/reports/tr31/) + @stable ICU 3.4 */ + UCHAR_PATTERN_WHITE_SPACE=43, + /** Binary property alnum (a C/POSIX character class). + Implemented according to the UTS #18 Annex C Standard Recommendation. + See the uchar.h file documentation. + @stable ICU 3.4 */ + UCHAR_POSIX_ALNUM=44, + /** Binary property blank (a C/POSIX character class). + Implemented according to the UTS #18 Annex C Standard Recommendation. + See the uchar.h file documentation. + @stable ICU 3.4 */ + UCHAR_POSIX_BLANK=45, + /** Binary property graph (a C/POSIX character class). + Implemented according to the UTS #18 Annex C Standard Recommendation. + See the uchar.h file documentation. + @stable ICU 3.4 */ + UCHAR_POSIX_GRAPH=46, + /** Binary property print (a C/POSIX character class). + Implemented according to the UTS #18 Annex C Standard Recommendation. + See the uchar.h file documentation. + @stable ICU 3.4 */ + UCHAR_POSIX_PRINT=47, + /** Binary property xdigit (a C/POSIX character class). + Implemented according to the UTS #18 Annex C Standard Recommendation. + See the uchar.h file documentation. + @stable ICU 3.4 */ + UCHAR_POSIX_XDIGIT=48, + /** Binary property Cased. For Lowercase, Uppercase and Titlecase characters. @stable ICU 4.4 */ + UCHAR_CASED=49, + /** Binary property Case_Ignorable. Used in context-sensitive case mappings. @stable ICU 4.4 */ + UCHAR_CASE_IGNORABLE=50, + /** Binary property Changes_When_Lowercased. @stable ICU 4.4 */ + UCHAR_CHANGES_WHEN_LOWERCASED=51, + /** Binary property Changes_When_Uppercased. @stable ICU 4.4 */ + UCHAR_CHANGES_WHEN_UPPERCASED=52, + /** Binary property Changes_When_Titlecased. @stable ICU 4.4 */ + UCHAR_CHANGES_WHEN_TITLECASED=53, + /** Binary property Changes_When_Casefolded. @stable ICU 4.4 */ + UCHAR_CHANGES_WHEN_CASEFOLDED=54, + /** Binary property Changes_When_Casemapped. @stable ICU 4.4 */ + UCHAR_CHANGES_WHEN_CASEMAPPED=55, + /** Binary property Changes_When_NFKC_Casefolded. @stable ICU 4.4 */ + UCHAR_CHANGES_WHEN_NFKC_CASEFOLDED=56, + /** + * Binary property Emoji. + * See http://www.unicode.org/reports/tr51/#Emoji_Properties + * + * @stable ICU 57 + */ + UCHAR_EMOJI=57, + /** + * Binary property Emoji_Presentation. + * See http://www.unicode.org/reports/tr51/#Emoji_Properties + * + * @stable ICU 57 + */ + UCHAR_EMOJI_PRESENTATION=58, + /** + * Binary property Emoji_Modifier. + * See http://www.unicode.org/reports/tr51/#Emoji_Properties + * + * @stable ICU 57 + */ + UCHAR_EMOJI_MODIFIER=59, + /** + * Binary property Emoji_Modifier_Base. + * See http://www.unicode.org/reports/tr51/#Emoji_Properties + * + * @stable ICU 57 + */ + UCHAR_EMOJI_MODIFIER_BASE=60, + /** + * Binary property Emoji_Component. + * See http://www.unicode.org/reports/tr51/#Emoji_Properties + * + * @stable ICU 60 + */ + UCHAR_EMOJI_COMPONENT=61, + /** + * Binary property Regional_Indicator. + * @stable ICU 60 + */ + UCHAR_REGIONAL_INDICATOR=62, + /** + * Binary property Prepended_Concatenation_Mark. + * @stable ICU 60 + */ + UCHAR_PREPENDED_CONCATENATION_MARK=63, + /** + * Binary property Extended_Pictographic. + * See http://www.unicode.org/reports/tr51/#Emoji_Properties + * + * @stable ICU 62 + */ + UCHAR_EXTENDED_PICTOGRAPHIC=64, + /** + * Binary property of strings Basic_Emoji. + * See https://www.unicode.org/reports/tr51/#Emoji_Sets + * + * @stable ICU 70 + */ + UCHAR_BASIC_EMOJI=65, + /** + * Binary property of strings Emoji_Keycap_Sequence. + * See https://www.unicode.org/reports/tr51/#Emoji_Sets + * + * @stable ICU 70 + */ + UCHAR_EMOJI_KEYCAP_SEQUENCE=66, + /** + * Binary property of strings RGI_Emoji_Modifier_Sequence. + * See https://www.unicode.org/reports/tr51/#Emoji_Sets + * + * @stable ICU 70 + */ + UCHAR_RGI_EMOJI_MODIFIER_SEQUENCE=67, + /** + * Binary property of strings RGI_Emoji_Flag_Sequence. + * See https://www.unicode.org/reports/tr51/#Emoji_Sets + * + * @stable ICU 70 + */ + UCHAR_RGI_EMOJI_FLAG_SEQUENCE=68, + /** + * Binary property of strings RGI_Emoji_Tag_Sequence. + * See https://www.unicode.org/reports/tr51/#Emoji_Sets + * + * @stable ICU 70 + */ + UCHAR_RGI_EMOJI_TAG_SEQUENCE=69, + /** + * Binary property of strings RGI_Emoji_ZWJ_Sequence. + * See https://www.unicode.org/reports/tr51/#Emoji_Sets + * + * @stable ICU 70 + */ + UCHAR_RGI_EMOJI_ZWJ_SEQUENCE=70, + /** + * Binary property of strings RGI_Emoji. + * See https://www.unicode.org/reports/tr51/#Emoji_Sets + * + * @stable ICU 70 + */ + UCHAR_RGI_EMOJI=71, +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the last constant for binary Unicode properties. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + UCHAR_BINARY_LIMIT=72, +#endif // U_HIDE_DEPRECATED_API + + /** Enumerated property Bidi_Class. + Same as u_charDirection, returns UCharDirection values. @stable ICU 2.2 */ + UCHAR_BIDI_CLASS=0x1000, + /** First constant for enumerated/integer Unicode properties. @stable ICU 2.2 */ + UCHAR_INT_START=UCHAR_BIDI_CLASS, + /** Enumerated property Block. + Same as ublock_getCode, returns UBlockCode values. @stable ICU 2.2 */ + UCHAR_BLOCK=0x1001, + /** Enumerated property Canonical_Combining_Class. + Same as u_getCombiningClass, returns 8-bit numeric values. @stable ICU 2.2 */ + UCHAR_CANONICAL_COMBINING_CLASS=0x1002, + /** Enumerated property Decomposition_Type. + Returns UDecompositionType values. @stable ICU 2.2 */ + UCHAR_DECOMPOSITION_TYPE=0x1003, + /** Enumerated property East_Asian_Width. + See http://www.unicode.org/reports/tr11/ + Returns UEastAsianWidth values. @stable ICU 2.2 */ + UCHAR_EAST_ASIAN_WIDTH=0x1004, + /** Enumerated property General_Category. + Same as u_charType, returns UCharCategory values. @stable ICU 2.2 */ + UCHAR_GENERAL_CATEGORY=0x1005, + /** Enumerated property Joining_Group. + Returns UJoiningGroup values. @stable ICU 2.2 */ + UCHAR_JOINING_GROUP=0x1006, + /** Enumerated property Joining_Type. + Returns UJoiningType values. @stable ICU 2.2 */ + UCHAR_JOINING_TYPE=0x1007, + /** Enumerated property Line_Break. + Returns ULineBreak values. @stable ICU 2.2 */ + UCHAR_LINE_BREAK=0x1008, + /** Enumerated property Numeric_Type. + Returns UNumericType values. @stable ICU 2.2 */ + UCHAR_NUMERIC_TYPE=0x1009, + /** Enumerated property Script. + Same as uscript_getScript, returns UScriptCode values. @stable ICU 2.2 */ + UCHAR_SCRIPT=0x100A, + /** Enumerated property Hangul_Syllable_Type, new in Unicode 4. + Returns UHangulSyllableType values. @stable ICU 2.6 */ + UCHAR_HANGUL_SYLLABLE_TYPE=0x100B, + /** Enumerated property NFD_Quick_Check. + Returns UNormalizationCheckResult values. @stable ICU 3.0 */ + UCHAR_NFD_QUICK_CHECK=0x100C, + /** Enumerated property NFKD_Quick_Check. + Returns UNormalizationCheckResult values. @stable ICU 3.0 */ + UCHAR_NFKD_QUICK_CHECK=0x100D, + /** Enumerated property NFC_Quick_Check. + Returns UNormalizationCheckResult values. @stable ICU 3.0 */ + UCHAR_NFC_QUICK_CHECK=0x100E, + /** Enumerated property NFKC_Quick_Check. + Returns UNormalizationCheckResult values. @stable ICU 3.0 */ + UCHAR_NFKC_QUICK_CHECK=0x100F, + /** Enumerated property Lead_Canonical_Combining_Class. + ICU-specific property for the ccc of the first code point + of the decomposition, or lccc(c)=ccc(NFD(c)[0]). + Useful for checking for canonically ordered text; + see UNORM_FCD and http://www.unicode.org/notes/tn5/#FCD . + Returns 8-bit numeric values like UCHAR_CANONICAL_COMBINING_CLASS. @stable ICU 3.0 */ + UCHAR_LEAD_CANONICAL_COMBINING_CLASS=0x1010, + /** Enumerated property Trail_Canonical_Combining_Class. + ICU-specific property for the ccc of the last code point + of the decomposition, or tccc(c)=ccc(NFD(c)[last]). + Useful for checking for canonically ordered text; + see UNORM_FCD and http://www.unicode.org/notes/tn5/#FCD . + Returns 8-bit numeric values like UCHAR_CANONICAL_COMBINING_CLASS. @stable ICU 3.0 */ + UCHAR_TRAIL_CANONICAL_COMBINING_CLASS=0x1011, + /** Enumerated property Grapheme_Cluster_Break (new in Unicode 4.1). + Used in UAX #29: Text Boundaries + (http://www.unicode.org/reports/tr29/) + Returns UGraphemeClusterBreak values. @stable ICU 3.4 */ + UCHAR_GRAPHEME_CLUSTER_BREAK=0x1012, + /** Enumerated property Sentence_Break (new in Unicode 4.1). + Used in UAX #29: Text Boundaries + (http://www.unicode.org/reports/tr29/) + Returns USentenceBreak values. @stable ICU 3.4 */ + UCHAR_SENTENCE_BREAK=0x1013, + /** Enumerated property Word_Break (new in Unicode 4.1). + Used in UAX #29: Text Boundaries + (http://www.unicode.org/reports/tr29/) + Returns UWordBreakValues values. @stable ICU 3.4 */ + UCHAR_WORD_BREAK=0x1014, + /** Enumerated property Bidi_Paired_Bracket_Type (new in Unicode 6.3). + Used in UAX #9: Unicode Bidirectional Algorithm + (http://www.unicode.org/reports/tr9/) + Returns UBidiPairedBracketType values. @stable ICU 52 */ + UCHAR_BIDI_PAIRED_BRACKET_TYPE=0x1015, + /** + * Enumerated property Indic_Positional_Category. + * New in Unicode 6.0 as provisional property Indic_Matra_Category; + * renamed and changed to informative in Unicode 8.0. + * See http://www.unicode.org/reports/tr44/#IndicPositionalCategory.txt + * @stable ICU 63 + */ + UCHAR_INDIC_POSITIONAL_CATEGORY=0x1016, + /** + * Enumerated property Indic_Syllabic_Category. + * New in Unicode 6.0 as provisional; informative since Unicode 8.0. + * See http://www.unicode.org/reports/tr44/#IndicSyllabicCategory.txt + * @stable ICU 63 + */ + UCHAR_INDIC_SYLLABIC_CATEGORY=0x1017, + /** + * Enumerated property Vertical_Orientation. + * Used for UAX #50 Unicode Vertical Text Layout (https://www.unicode.org/reports/tr50/). + * New as a UCD property in Unicode 10.0. + * @stable ICU 63 + */ + UCHAR_VERTICAL_ORIENTATION=0x1018, +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the last constant for enumerated/integer Unicode properties. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + UCHAR_INT_LIMIT=0x1019, +#endif // U_HIDE_DEPRECATED_API + + /** Bitmask property General_Category_Mask. + This is the General_Category property returned as a bit mask. + When used in u_getIntPropertyValue(c), same as U_MASK(u_charType(c)), + returns bit masks for UCharCategory values where exactly one bit is set. + When used with u_getPropertyValueName() and u_getPropertyValueEnum(), + a multi-bit mask is used for sets of categories like "Letters". + Mask values should be cast to uint32_t. + @stable ICU 2.4 */ + UCHAR_GENERAL_CATEGORY_MASK=0x2000, + /** First constant for bit-mask Unicode properties. @stable ICU 2.4 */ + UCHAR_MASK_START=UCHAR_GENERAL_CATEGORY_MASK, +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the last constant for bit-mask Unicode properties. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + UCHAR_MASK_LIMIT=0x2001, +#endif // U_HIDE_DEPRECATED_API + + /** Double property Numeric_Value. + Corresponds to u_getNumericValue. @stable ICU 2.4 */ + UCHAR_NUMERIC_VALUE=0x3000, + /** First constant for double Unicode properties. @stable ICU 2.4 */ + UCHAR_DOUBLE_START=UCHAR_NUMERIC_VALUE, +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the last constant for double Unicode properties. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + UCHAR_DOUBLE_LIMIT=0x3001, +#endif // U_HIDE_DEPRECATED_API + + /** String property Age. + Corresponds to u_charAge. @stable ICU 2.4 */ + UCHAR_AGE=0x4000, + /** First constant for string Unicode properties. @stable ICU 2.4 */ + UCHAR_STRING_START=UCHAR_AGE, + /** String property Bidi_Mirroring_Glyph. + Corresponds to u_charMirror. @stable ICU 2.4 */ + UCHAR_BIDI_MIRRORING_GLYPH=0x4001, + /** String property Case_Folding. + Corresponds to u_strFoldCase in ustring.h. @stable ICU 2.4 */ + UCHAR_CASE_FOLDING=0x4002, +#ifndef U_HIDE_DEPRECATED_API + /** Deprecated string property ISO_Comment. + Corresponds to u_getISOComment. @deprecated ICU 49 */ + UCHAR_ISO_COMMENT=0x4003, +#endif /* U_HIDE_DEPRECATED_API */ + /** String property Lowercase_Mapping. + Corresponds to u_strToLower in ustring.h. @stable ICU 2.4 */ + UCHAR_LOWERCASE_MAPPING=0x4004, + /** String property Name. + Corresponds to u_charName. @stable ICU 2.4 */ + UCHAR_NAME=0x4005, + /** String property Simple_Case_Folding. + Corresponds to u_foldCase. @stable ICU 2.4 */ + UCHAR_SIMPLE_CASE_FOLDING=0x4006, + /** String property Simple_Lowercase_Mapping. + Corresponds to u_tolower. @stable ICU 2.4 */ + UCHAR_SIMPLE_LOWERCASE_MAPPING=0x4007, + /** String property Simple_Titlecase_Mapping. + Corresponds to u_totitle. @stable ICU 2.4 */ + UCHAR_SIMPLE_TITLECASE_MAPPING=0x4008, + /** String property Simple_Uppercase_Mapping. + Corresponds to u_toupper. @stable ICU 2.4 */ + UCHAR_SIMPLE_UPPERCASE_MAPPING=0x4009, + /** String property Titlecase_Mapping. + Corresponds to u_strToTitle in ustring.h. @stable ICU 2.4 */ + UCHAR_TITLECASE_MAPPING=0x400A, +#ifndef U_HIDE_DEPRECATED_API + /** String property Unicode_1_Name. + This property is of little practical value. + Beginning with ICU 49, ICU APIs return an empty string for this property. + Corresponds to u_charName(U_UNICODE_10_CHAR_NAME). @deprecated ICU 49 */ + UCHAR_UNICODE_1_NAME=0x400B, +#endif /* U_HIDE_DEPRECATED_API */ + /** String property Uppercase_Mapping. + Corresponds to u_strToUpper in ustring.h. @stable ICU 2.4 */ + UCHAR_UPPERCASE_MAPPING=0x400C, + /** String property Bidi_Paired_Bracket (new in Unicode 6.3). + Corresponds to u_getBidiPairedBracket. @stable ICU 52 */ + UCHAR_BIDI_PAIRED_BRACKET=0x400D, +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the last constant for string Unicode properties. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + UCHAR_STRING_LIMIT=0x400E, +#endif // U_HIDE_DEPRECATED_API + + /** Miscellaneous property Script_Extensions (new in Unicode 6.0). + Some characters are commonly used in multiple scripts. + For more information, see UAX #24: http://www.unicode.org/reports/tr24/. + Corresponds to uscript_hasScript and uscript_getScriptExtensions in uscript.h. + @stable ICU 4.6 */ + UCHAR_SCRIPT_EXTENSIONS=0x7000, + /** First constant for Unicode properties with unusual value types. @stable ICU 4.6 */ + UCHAR_OTHER_PROPERTY_START=UCHAR_SCRIPT_EXTENSIONS, +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the last constant for Unicode properties with unusual value types. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + UCHAR_OTHER_PROPERTY_LIMIT=0x7001, +#endif // U_HIDE_DEPRECATED_API + + /** Represents a nonexistent or invalid property or property value. @stable ICU 2.4 */ + UCHAR_INVALID_CODE = -1 +} UProperty; + +/** + * Data for enumerated Unicode general category types. + * See http://www.unicode.org/Public/UNIDATA/UnicodeData.html . + * @stable ICU 2.0 + */ +typedef enum UCharCategory +{ + /* + * Note: UCharCategory constants and their API comments are parsed by preparseucd.py. + * It matches pairs of lines like + * / ** comment... * / + * U_<[A-Z_]+> = , + */ + + /** Non-category for unassigned and non-character code points. @stable ICU 2.0 */ + U_UNASSIGNED = 0, + /** Cn "Other, Not Assigned (no characters in [UnicodeData.txt] have this property)" (same as U_UNASSIGNED!) @stable ICU 2.0 */ + U_GENERAL_OTHER_TYPES = 0, + /** Lu @stable ICU 2.0 */ + U_UPPERCASE_LETTER = 1, + /** Ll @stable ICU 2.0 */ + U_LOWERCASE_LETTER = 2, + /** Lt @stable ICU 2.0 */ + U_TITLECASE_LETTER = 3, + /** Lm @stable ICU 2.0 */ + U_MODIFIER_LETTER = 4, + /** Lo @stable ICU 2.0 */ + U_OTHER_LETTER = 5, + /** Mn @stable ICU 2.0 */ + U_NON_SPACING_MARK = 6, + /** Me @stable ICU 2.0 */ + U_ENCLOSING_MARK = 7, + /** Mc @stable ICU 2.0 */ + U_COMBINING_SPACING_MARK = 8, + /** Nd @stable ICU 2.0 */ + U_DECIMAL_DIGIT_NUMBER = 9, + /** Nl @stable ICU 2.0 */ + U_LETTER_NUMBER = 10, + /** No @stable ICU 2.0 */ + U_OTHER_NUMBER = 11, + /** Zs @stable ICU 2.0 */ + U_SPACE_SEPARATOR = 12, + /** Zl @stable ICU 2.0 */ + U_LINE_SEPARATOR = 13, + /** Zp @stable ICU 2.0 */ + U_PARAGRAPH_SEPARATOR = 14, + /** Cc @stable ICU 2.0 */ + U_CONTROL_CHAR = 15, + /** Cf @stable ICU 2.0 */ + U_FORMAT_CHAR = 16, + /** Co @stable ICU 2.0 */ + U_PRIVATE_USE_CHAR = 17, + /** Cs @stable ICU 2.0 */ + U_SURROGATE = 18, + /** Pd @stable ICU 2.0 */ + U_DASH_PUNCTUATION = 19, + /** Ps @stable ICU 2.0 */ + U_START_PUNCTUATION = 20, + /** Pe @stable ICU 2.0 */ + U_END_PUNCTUATION = 21, + /** Pc @stable ICU 2.0 */ + U_CONNECTOR_PUNCTUATION = 22, + /** Po @stable ICU 2.0 */ + U_OTHER_PUNCTUATION = 23, + /** Sm @stable ICU 2.0 */ + U_MATH_SYMBOL = 24, + /** Sc @stable ICU 2.0 */ + U_CURRENCY_SYMBOL = 25, + /** Sk @stable ICU 2.0 */ + U_MODIFIER_SYMBOL = 26, + /** So @stable ICU 2.0 */ + U_OTHER_SYMBOL = 27, + /** Pi @stable ICU 2.0 */ + U_INITIAL_PUNCTUATION = 28, + /** Pf @stable ICU 2.0 */ + U_FINAL_PUNCTUATION = 29, + /** + * One higher than the last enum UCharCategory constant. + * This numeric value is stable (will not change), see + * http://www.unicode.org/policies/stability_policy.html#Property_Value + * + * @stable ICU 2.0 + */ + U_CHAR_CATEGORY_COUNT +} UCharCategory; + +/** + * U_GC_XX_MASK constants are bit flags corresponding to Unicode + * general category values. + * For each category, the nth bit is set if the numeric value of the + * corresponding UCharCategory constant is n. + * + * There are also some U_GC_Y_MASK constants for groups of general categories + * like L for all letter categories. + * + * @see u_charType + * @see U_GET_GC_MASK + * @see UCharCategory + * @stable ICU 2.1 + */ +#define U_GC_CN_MASK U_MASK(U_GENERAL_OTHER_TYPES) + +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_LU_MASK U_MASK(U_UPPERCASE_LETTER) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_LL_MASK U_MASK(U_LOWERCASE_LETTER) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_LT_MASK U_MASK(U_TITLECASE_LETTER) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_LM_MASK U_MASK(U_MODIFIER_LETTER) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_LO_MASK U_MASK(U_OTHER_LETTER) + +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_MN_MASK U_MASK(U_NON_SPACING_MARK) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_ME_MASK U_MASK(U_ENCLOSING_MARK) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_MC_MASK U_MASK(U_COMBINING_SPACING_MARK) + +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_ND_MASK U_MASK(U_DECIMAL_DIGIT_NUMBER) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_NL_MASK U_MASK(U_LETTER_NUMBER) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_NO_MASK U_MASK(U_OTHER_NUMBER) + +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_ZS_MASK U_MASK(U_SPACE_SEPARATOR) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_ZL_MASK U_MASK(U_LINE_SEPARATOR) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_ZP_MASK U_MASK(U_PARAGRAPH_SEPARATOR) + +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_CC_MASK U_MASK(U_CONTROL_CHAR) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_CF_MASK U_MASK(U_FORMAT_CHAR) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_CO_MASK U_MASK(U_PRIVATE_USE_CHAR) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_CS_MASK U_MASK(U_SURROGATE) + +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_PD_MASK U_MASK(U_DASH_PUNCTUATION) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_PS_MASK U_MASK(U_START_PUNCTUATION) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_PE_MASK U_MASK(U_END_PUNCTUATION) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_PC_MASK U_MASK(U_CONNECTOR_PUNCTUATION) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_PO_MASK U_MASK(U_OTHER_PUNCTUATION) + +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_SM_MASK U_MASK(U_MATH_SYMBOL) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_SC_MASK U_MASK(U_CURRENCY_SYMBOL) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_SK_MASK U_MASK(U_MODIFIER_SYMBOL) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_SO_MASK U_MASK(U_OTHER_SYMBOL) + +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_PI_MASK U_MASK(U_INITIAL_PUNCTUATION) +/** Mask constant for a UCharCategory. @stable ICU 2.1 */ +#define U_GC_PF_MASK U_MASK(U_FINAL_PUNCTUATION) + + +/** Mask constant for multiple UCharCategory bits (L Letters). @stable ICU 2.1 */ +#define U_GC_L_MASK \ + (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK|U_GC_LM_MASK|U_GC_LO_MASK) + +/** Mask constant for multiple UCharCategory bits (LC Cased Letters). @stable ICU 2.1 */ +#define U_GC_LC_MASK \ + (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK) + +/** Mask constant for multiple UCharCategory bits (M Marks). @stable ICU 2.1 */ +#define U_GC_M_MASK (U_GC_MN_MASK|U_GC_ME_MASK|U_GC_MC_MASK) + +/** Mask constant for multiple UCharCategory bits (N Numbers). @stable ICU 2.1 */ +#define U_GC_N_MASK (U_GC_ND_MASK|U_GC_NL_MASK|U_GC_NO_MASK) + +/** Mask constant for multiple UCharCategory bits (Z Separators). @stable ICU 2.1 */ +#define U_GC_Z_MASK (U_GC_ZS_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK) + +/** Mask constant for multiple UCharCategory bits (C Others). @stable ICU 2.1 */ +#define U_GC_C_MASK \ + (U_GC_CN_MASK|U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CO_MASK|U_GC_CS_MASK) + +/** Mask constant for multiple UCharCategory bits (P Punctuation). @stable ICU 2.1 */ +#define U_GC_P_MASK \ + (U_GC_PD_MASK|U_GC_PS_MASK|U_GC_PE_MASK|U_GC_PC_MASK|U_GC_PO_MASK| \ + U_GC_PI_MASK|U_GC_PF_MASK) + +/** Mask constant for multiple UCharCategory bits (S Symbols). @stable ICU 2.1 */ +#define U_GC_S_MASK (U_GC_SM_MASK|U_GC_SC_MASK|U_GC_SK_MASK|U_GC_SO_MASK) + +/** + * This specifies the language directional property of a character set. + * @stable ICU 2.0 + */ +typedef enum UCharDirection { + /* + * Note: UCharDirection constants and their API comments are parsed by preparseucd.py. + * It matches pairs of lines like + * / ** comment... * / + * U_<[A-Z_]+> = , + */ + + /** L @stable ICU 2.0 */ + U_LEFT_TO_RIGHT = 0, + /** R @stable ICU 2.0 */ + U_RIGHT_TO_LEFT = 1, + /** EN @stable ICU 2.0 */ + U_EUROPEAN_NUMBER = 2, + /** ES @stable ICU 2.0 */ + U_EUROPEAN_NUMBER_SEPARATOR = 3, + /** ET @stable ICU 2.0 */ + U_EUROPEAN_NUMBER_TERMINATOR = 4, + /** AN @stable ICU 2.0 */ + U_ARABIC_NUMBER = 5, + /** CS @stable ICU 2.0 */ + U_COMMON_NUMBER_SEPARATOR = 6, + /** B @stable ICU 2.0 */ + U_BLOCK_SEPARATOR = 7, + /** S @stable ICU 2.0 */ + U_SEGMENT_SEPARATOR = 8, + /** WS @stable ICU 2.0 */ + U_WHITE_SPACE_NEUTRAL = 9, + /** ON @stable ICU 2.0 */ + U_OTHER_NEUTRAL = 10, + /** LRE @stable ICU 2.0 */ + U_LEFT_TO_RIGHT_EMBEDDING = 11, + /** LRO @stable ICU 2.0 */ + U_LEFT_TO_RIGHT_OVERRIDE = 12, + /** AL @stable ICU 2.0 */ + U_RIGHT_TO_LEFT_ARABIC = 13, + /** RLE @stable ICU 2.0 */ + U_RIGHT_TO_LEFT_EMBEDDING = 14, + /** RLO @stable ICU 2.0 */ + U_RIGHT_TO_LEFT_OVERRIDE = 15, + /** PDF @stable ICU 2.0 */ + U_POP_DIRECTIONAL_FORMAT = 16, + /** NSM @stable ICU 2.0 */ + U_DIR_NON_SPACING_MARK = 17, + /** BN @stable ICU 2.0 */ + U_BOUNDARY_NEUTRAL = 18, + /** FSI @stable ICU 52 */ + U_FIRST_STRONG_ISOLATE = 19, + /** LRI @stable ICU 52 */ + U_LEFT_TO_RIGHT_ISOLATE = 20, + /** RLI @stable ICU 52 */ + U_RIGHT_TO_LEFT_ISOLATE = 21, + /** PDI @stable ICU 52 */ + U_POP_DIRECTIONAL_ISOLATE = 22, +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest UCharDirection value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_BIDI_CLASS). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_CHAR_DIRECTION_COUNT +#endif // U_HIDE_DEPRECATED_API +} UCharDirection; + +/** + * Bidi Paired Bracket Type constants. + * + * @see UCHAR_BIDI_PAIRED_BRACKET_TYPE + * @stable ICU 52 + */ +typedef enum UBidiPairedBracketType { + /* + * Note: UBidiPairedBracketType constants are parsed by preparseucd.py. + * It matches lines like + * U_BPT_ + */ + + /** Not a paired bracket. @stable ICU 52 */ + U_BPT_NONE, + /** Open paired bracket. @stable ICU 52 */ + U_BPT_OPEN, + /** Close paired bracket. @stable ICU 52 */ + U_BPT_CLOSE, +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal UBidiPairedBracketType value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_BIDI_PAIRED_BRACKET_TYPE). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_BPT_COUNT /* 3 */ +#endif // U_HIDE_DEPRECATED_API +} UBidiPairedBracketType; + +/** + * Constants for Unicode blocks, see the Unicode Data file Blocks.txt + * @stable ICU 2.0 + */ +enum UBlockCode { + /* + * Note: UBlockCode constants are parsed by preparseucd.py. + * It matches lines like + * UBLOCK_ = , + */ + + /** New No_Block value in Unicode 4. @stable ICU 2.6 */ + UBLOCK_NO_BLOCK = 0, /*[none]*/ /* Special range indicating No_Block */ + + /** @stable ICU 2.0 */ + UBLOCK_BASIC_LATIN = 1, /*[0000]*/ + + /** @stable ICU 2.0 */ + UBLOCK_LATIN_1_SUPPLEMENT=2, /*[0080]*/ + + /** @stable ICU 2.0 */ + UBLOCK_LATIN_EXTENDED_A =3, /*[0100]*/ + + /** @stable ICU 2.0 */ + UBLOCK_LATIN_EXTENDED_B =4, /*[0180]*/ + + /** @stable ICU 2.0 */ + UBLOCK_IPA_EXTENSIONS =5, /*[0250]*/ + + /** @stable ICU 2.0 */ + UBLOCK_SPACING_MODIFIER_LETTERS =6, /*[02B0]*/ + + /** @stable ICU 2.0 */ + UBLOCK_COMBINING_DIACRITICAL_MARKS =7, /*[0300]*/ + + /** + * Unicode 3.2 renames this block to "Greek and Coptic". + * @stable ICU 2.0 + */ + UBLOCK_GREEK =8, /*[0370]*/ + + /** @stable ICU 2.0 */ + UBLOCK_CYRILLIC =9, /*[0400]*/ + + /** @stable ICU 2.0 */ + UBLOCK_ARMENIAN =10, /*[0530]*/ + + /** @stable ICU 2.0 */ + UBLOCK_HEBREW =11, /*[0590]*/ + + /** @stable ICU 2.0 */ + UBLOCK_ARABIC =12, /*[0600]*/ + + /** @stable ICU 2.0 */ + UBLOCK_SYRIAC =13, /*[0700]*/ + + /** @stable ICU 2.0 */ + UBLOCK_THAANA =14, /*[0780]*/ + + /** @stable ICU 2.0 */ + UBLOCK_DEVANAGARI =15, /*[0900]*/ + + /** @stable ICU 2.0 */ + UBLOCK_BENGALI =16, /*[0980]*/ + + /** @stable ICU 2.0 */ + UBLOCK_GURMUKHI =17, /*[0A00]*/ + + /** @stable ICU 2.0 */ + UBLOCK_GUJARATI =18, /*[0A80]*/ + + /** @stable ICU 2.0 */ + UBLOCK_ORIYA =19, /*[0B00]*/ + + /** @stable ICU 2.0 */ + UBLOCK_TAMIL =20, /*[0B80]*/ + + /** @stable ICU 2.0 */ + UBLOCK_TELUGU =21, /*[0C00]*/ + + /** @stable ICU 2.0 */ + UBLOCK_KANNADA =22, /*[0C80]*/ + + /** @stable ICU 2.0 */ + UBLOCK_MALAYALAM =23, /*[0D00]*/ + + /** @stable ICU 2.0 */ + UBLOCK_SINHALA =24, /*[0D80]*/ + + /** @stable ICU 2.0 */ + UBLOCK_THAI =25, /*[0E00]*/ + + /** @stable ICU 2.0 */ + UBLOCK_LAO =26, /*[0E80]*/ + + /** @stable ICU 2.0 */ + UBLOCK_TIBETAN =27, /*[0F00]*/ + + /** @stable ICU 2.0 */ + UBLOCK_MYANMAR =28, /*[1000]*/ + + /** @stable ICU 2.0 */ + UBLOCK_GEORGIAN =29, /*[10A0]*/ + + /** @stable ICU 2.0 */ + UBLOCK_HANGUL_JAMO =30, /*[1100]*/ + + /** @stable ICU 2.0 */ + UBLOCK_ETHIOPIC =31, /*[1200]*/ + + /** @stable ICU 2.0 */ + UBLOCK_CHEROKEE =32, /*[13A0]*/ + + /** @stable ICU 2.0 */ + UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS =33, /*[1400]*/ + + /** @stable ICU 2.0 */ + UBLOCK_OGHAM =34, /*[1680]*/ + + /** @stable ICU 2.0 */ + UBLOCK_RUNIC =35, /*[16A0]*/ + + /** @stable ICU 2.0 */ + UBLOCK_KHMER =36, /*[1780]*/ + + /** @stable ICU 2.0 */ + UBLOCK_MONGOLIAN =37, /*[1800]*/ + + /** @stable ICU 2.0 */ + UBLOCK_LATIN_EXTENDED_ADDITIONAL =38, /*[1E00]*/ + + /** @stable ICU 2.0 */ + UBLOCK_GREEK_EXTENDED =39, /*[1F00]*/ + + /** @stable ICU 2.0 */ + UBLOCK_GENERAL_PUNCTUATION =40, /*[2000]*/ + + /** @stable ICU 2.0 */ + UBLOCK_SUPERSCRIPTS_AND_SUBSCRIPTS =41, /*[2070]*/ + + /** @stable ICU 2.0 */ + UBLOCK_CURRENCY_SYMBOLS =42, /*[20A0]*/ + + /** + * Unicode 3.2 renames this block to "Combining Diacritical Marks for Symbols". + * @stable ICU 2.0 + */ + UBLOCK_COMBINING_MARKS_FOR_SYMBOLS =43, /*[20D0]*/ + + /** @stable ICU 2.0 */ + UBLOCK_LETTERLIKE_SYMBOLS =44, /*[2100]*/ + + /** @stable ICU 2.0 */ + UBLOCK_NUMBER_FORMS =45, /*[2150]*/ + + /** @stable ICU 2.0 */ + UBLOCK_ARROWS =46, /*[2190]*/ + + /** @stable ICU 2.0 */ + UBLOCK_MATHEMATICAL_OPERATORS =47, /*[2200]*/ + + /** @stable ICU 2.0 */ + UBLOCK_MISCELLANEOUS_TECHNICAL =48, /*[2300]*/ + + /** @stable ICU 2.0 */ + UBLOCK_CONTROL_PICTURES =49, /*[2400]*/ + + /** @stable ICU 2.0 */ + UBLOCK_OPTICAL_CHARACTER_RECOGNITION =50, /*[2440]*/ + + /** @stable ICU 2.0 */ + UBLOCK_ENCLOSED_ALPHANUMERICS =51, /*[2460]*/ + + /** @stable ICU 2.0 */ + UBLOCK_BOX_DRAWING =52, /*[2500]*/ + + /** @stable ICU 2.0 */ + UBLOCK_BLOCK_ELEMENTS =53, /*[2580]*/ + + /** @stable ICU 2.0 */ + UBLOCK_GEOMETRIC_SHAPES =54, /*[25A0]*/ + + /** @stable ICU 2.0 */ + UBLOCK_MISCELLANEOUS_SYMBOLS =55, /*[2600]*/ + + /** @stable ICU 2.0 */ + UBLOCK_DINGBATS =56, /*[2700]*/ + + /** @stable ICU 2.0 */ + UBLOCK_BRAILLE_PATTERNS =57, /*[2800]*/ + + /** @stable ICU 2.0 */ + UBLOCK_CJK_RADICALS_SUPPLEMENT =58, /*[2E80]*/ + + /** @stable ICU 2.0 */ + UBLOCK_KANGXI_RADICALS =59, /*[2F00]*/ + + /** @stable ICU 2.0 */ + UBLOCK_IDEOGRAPHIC_DESCRIPTION_CHARACTERS =60, /*[2FF0]*/ + + /** @stable ICU 2.0 */ + UBLOCK_CJK_SYMBOLS_AND_PUNCTUATION =61, /*[3000]*/ + + /** @stable ICU 2.0 */ + UBLOCK_HIRAGANA =62, /*[3040]*/ + + /** @stable ICU 2.0 */ + UBLOCK_KATAKANA =63, /*[30A0]*/ + + /** @stable ICU 2.0 */ + UBLOCK_BOPOMOFO =64, /*[3100]*/ + + /** @stable ICU 2.0 */ + UBLOCK_HANGUL_COMPATIBILITY_JAMO =65, /*[3130]*/ + + /** @stable ICU 2.0 */ + UBLOCK_KANBUN =66, /*[3190]*/ + + /** @stable ICU 2.0 */ + UBLOCK_BOPOMOFO_EXTENDED =67, /*[31A0]*/ + + /** @stable ICU 2.0 */ + UBLOCK_ENCLOSED_CJK_LETTERS_AND_MONTHS =68, /*[3200]*/ + + /** @stable ICU 2.0 */ + UBLOCK_CJK_COMPATIBILITY =69, /*[3300]*/ + + /** @stable ICU 2.0 */ + UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A =70, /*[3400]*/ + + /** @stable ICU 2.0 */ + UBLOCK_CJK_UNIFIED_IDEOGRAPHS =71, /*[4E00]*/ + + /** @stable ICU 2.0 */ + UBLOCK_YI_SYLLABLES =72, /*[A000]*/ + + /** @stable ICU 2.0 */ + UBLOCK_YI_RADICALS =73, /*[A490]*/ + + /** @stable ICU 2.0 */ + UBLOCK_HANGUL_SYLLABLES =74, /*[AC00]*/ + + /** @stable ICU 2.0 */ + UBLOCK_HIGH_SURROGATES =75, /*[D800]*/ + + /** @stable ICU 2.0 */ + UBLOCK_HIGH_PRIVATE_USE_SURROGATES =76, /*[DB80]*/ + + /** @stable ICU 2.0 */ + UBLOCK_LOW_SURROGATES =77, /*[DC00]*/ + + /** + * Same as UBLOCK_PRIVATE_USE. + * Until Unicode 3.1.1, the corresponding block name was "Private Use", + * and multiple code point ranges had this block. + * Unicode 3.2 renames the block for the BMP PUA to "Private Use Area" and + * adds separate blocks for the supplementary PUAs. + * + * @stable ICU 2.0 + */ + UBLOCK_PRIVATE_USE_AREA =78, /*[E000]*/ + /** + * Same as UBLOCK_PRIVATE_USE_AREA. + * Until Unicode 3.1.1, the corresponding block name was "Private Use", + * and multiple code point ranges had this block. + * Unicode 3.2 renames the block for the BMP PUA to "Private Use Area" and + * adds separate blocks for the supplementary PUAs. + * + * @stable ICU 2.0 + */ + UBLOCK_PRIVATE_USE = UBLOCK_PRIVATE_USE_AREA, + + /** @stable ICU 2.0 */ + UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS =79, /*[F900]*/ + + /** @stable ICU 2.0 */ + UBLOCK_ALPHABETIC_PRESENTATION_FORMS =80, /*[FB00]*/ + + /** @stable ICU 2.0 */ + UBLOCK_ARABIC_PRESENTATION_FORMS_A =81, /*[FB50]*/ + + /** @stable ICU 2.0 */ + UBLOCK_COMBINING_HALF_MARKS =82, /*[FE20]*/ + + /** @stable ICU 2.0 */ + UBLOCK_CJK_COMPATIBILITY_FORMS =83, /*[FE30]*/ + + /** @stable ICU 2.0 */ + UBLOCK_SMALL_FORM_VARIANTS =84, /*[FE50]*/ + + /** @stable ICU 2.0 */ + UBLOCK_ARABIC_PRESENTATION_FORMS_B =85, /*[FE70]*/ + + /** @stable ICU 2.0 */ + UBLOCK_SPECIALS =86, /*[FFF0]*/ + + /** @stable ICU 2.0 */ + UBLOCK_HALFWIDTH_AND_FULLWIDTH_FORMS =87, /*[FF00]*/ + + /* New blocks in Unicode 3.1 */ + + /** @stable ICU 2.0 */ + UBLOCK_OLD_ITALIC = 88, /*[10300]*/ + /** @stable ICU 2.0 */ + UBLOCK_GOTHIC = 89, /*[10330]*/ + /** @stable ICU 2.0 */ + UBLOCK_DESERET = 90, /*[10400]*/ + /** @stable ICU 2.0 */ + UBLOCK_BYZANTINE_MUSICAL_SYMBOLS = 91, /*[1D000]*/ + /** @stable ICU 2.0 */ + UBLOCK_MUSICAL_SYMBOLS = 92, /*[1D100]*/ + /** @stable ICU 2.0 */ + UBLOCK_MATHEMATICAL_ALPHANUMERIC_SYMBOLS = 93, /*[1D400]*/ + /** @stable ICU 2.0 */ + UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B = 94, /*[20000]*/ + /** @stable ICU 2.0 */ + UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT = 95, /*[2F800]*/ + /** @stable ICU 2.0 */ + UBLOCK_TAGS = 96, /*[E0000]*/ + + /* New blocks in Unicode 3.2 */ + + /** @stable ICU 3.0 */ + UBLOCK_CYRILLIC_SUPPLEMENT = 97, /*[0500]*/ + /** + * Unicode 4.0.1 renames the "Cyrillic Supplementary" block to "Cyrillic Supplement". + * @stable ICU 2.2 + */ + UBLOCK_CYRILLIC_SUPPLEMENTARY = UBLOCK_CYRILLIC_SUPPLEMENT, + /** @stable ICU 2.2 */ + UBLOCK_TAGALOG = 98, /*[1700]*/ + /** @stable ICU 2.2 */ + UBLOCK_HANUNOO = 99, /*[1720]*/ + /** @stable ICU 2.2 */ + UBLOCK_BUHID = 100, /*[1740]*/ + /** @stable ICU 2.2 */ + UBLOCK_TAGBANWA = 101, /*[1760]*/ + /** @stable ICU 2.2 */ + UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A = 102, /*[27C0]*/ + /** @stable ICU 2.2 */ + UBLOCK_SUPPLEMENTAL_ARROWS_A = 103, /*[27F0]*/ + /** @stable ICU 2.2 */ + UBLOCK_SUPPLEMENTAL_ARROWS_B = 104, /*[2900]*/ + /** @stable ICU 2.2 */ + UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B = 105, /*[2980]*/ + /** @stable ICU 2.2 */ + UBLOCK_SUPPLEMENTAL_MATHEMATICAL_OPERATORS = 106, /*[2A00]*/ + /** @stable ICU 2.2 */ + UBLOCK_KATAKANA_PHONETIC_EXTENSIONS = 107, /*[31F0]*/ + /** @stable ICU 2.2 */ + UBLOCK_VARIATION_SELECTORS = 108, /*[FE00]*/ + /** @stable ICU 2.2 */ + UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_A = 109, /*[F0000]*/ + /** @stable ICU 2.2 */ + UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_B = 110, /*[100000]*/ + + /* New blocks in Unicode 4 */ + + /** @stable ICU 2.6 */ + UBLOCK_LIMBU = 111, /*[1900]*/ + /** @stable ICU 2.6 */ + UBLOCK_TAI_LE = 112, /*[1950]*/ + /** @stable ICU 2.6 */ + UBLOCK_KHMER_SYMBOLS = 113, /*[19E0]*/ + /** @stable ICU 2.6 */ + UBLOCK_PHONETIC_EXTENSIONS = 114, /*[1D00]*/ + /** @stable ICU 2.6 */ + UBLOCK_MISCELLANEOUS_SYMBOLS_AND_ARROWS = 115, /*[2B00]*/ + /** @stable ICU 2.6 */ + UBLOCK_YIJING_HEXAGRAM_SYMBOLS = 116, /*[4DC0]*/ + /** @stable ICU 2.6 */ + UBLOCK_LINEAR_B_SYLLABARY = 117, /*[10000]*/ + /** @stable ICU 2.6 */ + UBLOCK_LINEAR_B_IDEOGRAMS = 118, /*[10080]*/ + /** @stable ICU 2.6 */ + UBLOCK_AEGEAN_NUMBERS = 119, /*[10100]*/ + /** @stable ICU 2.6 */ + UBLOCK_UGARITIC = 120, /*[10380]*/ + /** @stable ICU 2.6 */ + UBLOCK_SHAVIAN = 121, /*[10450]*/ + /** @stable ICU 2.6 */ + UBLOCK_OSMANYA = 122, /*[10480]*/ + /** @stable ICU 2.6 */ + UBLOCK_CYPRIOT_SYLLABARY = 123, /*[10800]*/ + /** @stable ICU 2.6 */ + UBLOCK_TAI_XUAN_JING_SYMBOLS = 124, /*[1D300]*/ + /** @stable ICU 2.6 */ + UBLOCK_VARIATION_SELECTORS_SUPPLEMENT = 125, /*[E0100]*/ + + /* New blocks in Unicode 4.1 */ + + /** @stable ICU 3.4 */ + UBLOCK_ANCIENT_GREEK_MUSICAL_NOTATION = 126, /*[1D200]*/ + /** @stable ICU 3.4 */ + UBLOCK_ANCIENT_GREEK_NUMBERS = 127, /*[10140]*/ + /** @stable ICU 3.4 */ + UBLOCK_ARABIC_SUPPLEMENT = 128, /*[0750]*/ + /** @stable ICU 3.4 */ + UBLOCK_BUGINESE = 129, /*[1A00]*/ + /** @stable ICU 3.4 */ + UBLOCK_CJK_STROKES = 130, /*[31C0]*/ + /** @stable ICU 3.4 */ + UBLOCK_COMBINING_DIACRITICAL_MARKS_SUPPLEMENT = 131, /*[1DC0]*/ + /** @stable ICU 3.4 */ + UBLOCK_COPTIC = 132, /*[2C80]*/ + /** @stable ICU 3.4 */ + UBLOCK_ETHIOPIC_EXTENDED = 133, /*[2D80]*/ + /** @stable ICU 3.4 */ + UBLOCK_ETHIOPIC_SUPPLEMENT = 134, /*[1380]*/ + /** @stable ICU 3.4 */ + UBLOCK_GEORGIAN_SUPPLEMENT = 135, /*[2D00]*/ + /** @stable ICU 3.4 */ + UBLOCK_GLAGOLITIC = 136, /*[2C00]*/ + /** @stable ICU 3.4 */ + UBLOCK_KHAROSHTHI = 137, /*[10A00]*/ + /** @stable ICU 3.4 */ + UBLOCK_MODIFIER_TONE_LETTERS = 138, /*[A700]*/ + /** @stable ICU 3.4 */ + UBLOCK_NEW_TAI_LUE = 139, /*[1980]*/ + /** @stable ICU 3.4 */ + UBLOCK_OLD_PERSIAN = 140, /*[103A0]*/ + /** @stable ICU 3.4 */ + UBLOCK_PHONETIC_EXTENSIONS_SUPPLEMENT = 141, /*[1D80]*/ + /** @stable ICU 3.4 */ + UBLOCK_SUPPLEMENTAL_PUNCTUATION = 142, /*[2E00]*/ + /** @stable ICU 3.4 */ + UBLOCK_SYLOTI_NAGRI = 143, /*[A800]*/ + /** @stable ICU 3.4 */ + UBLOCK_TIFINAGH = 144, /*[2D30]*/ + /** @stable ICU 3.4 */ + UBLOCK_VERTICAL_FORMS = 145, /*[FE10]*/ + + /* New blocks in Unicode 5.0 */ + + /** @stable ICU 3.6 */ + UBLOCK_NKO = 146, /*[07C0]*/ + /** @stable ICU 3.6 */ + UBLOCK_BALINESE = 147, /*[1B00]*/ + /** @stable ICU 3.6 */ + UBLOCK_LATIN_EXTENDED_C = 148, /*[2C60]*/ + /** @stable ICU 3.6 */ + UBLOCK_LATIN_EXTENDED_D = 149, /*[A720]*/ + /** @stable ICU 3.6 */ + UBLOCK_PHAGS_PA = 150, /*[A840]*/ + /** @stable ICU 3.6 */ + UBLOCK_PHOENICIAN = 151, /*[10900]*/ + /** @stable ICU 3.6 */ + UBLOCK_CUNEIFORM = 152, /*[12000]*/ + /** @stable ICU 3.6 */ + UBLOCK_CUNEIFORM_NUMBERS_AND_PUNCTUATION = 153, /*[12400]*/ + /** @stable ICU 3.6 */ + UBLOCK_COUNTING_ROD_NUMERALS = 154, /*[1D360]*/ + + /* New blocks in Unicode 5.1 */ + + /** @stable ICU 4.0 */ + UBLOCK_SUNDANESE = 155, /*[1B80]*/ + /** @stable ICU 4.0 */ + UBLOCK_LEPCHA = 156, /*[1C00]*/ + /** @stable ICU 4.0 */ + UBLOCK_OL_CHIKI = 157, /*[1C50]*/ + /** @stable ICU 4.0 */ + UBLOCK_CYRILLIC_EXTENDED_A = 158, /*[2DE0]*/ + /** @stable ICU 4.0 */ + UBLOCK_VAI = 159, /*[A500]*/ + /** @stable ICU 4.0 */ + UBLOCK_CYRILLIC_EXTENDED_B = 160, /*[A640]*/ + /** @stable ICU 4.0 */ + UBLOCK_SAURASHTRA = 161, /*[A880]*/ + /** @stable ICU 4.0 */ + UBLOCK_KAYAH_LI = 162, /*[A900]*/ + /** @stable ICU 4.0 */ + UBLOCK_REJANG = 163, /*[A930]*/ + /** @stable ICU 4.0 */ + UBLOCK_CHAM = 164, /*[AA00]*/ + /** @stable ICU 4.0 */ + UBLOCK_ANCIENT_SYMBOLS = 165, /*[10190]*/ + /** @stable ICU 4.0 */ + UBLOCK_PHAISTOS_DISC = 166, /*[101D0]*/ + /** @stable ICU 4.0 */ + UBLOCK_LYCIAN = 167, /*[10280]*/ + /** @stable ICU 4.0 */ + UBLOCK_CARIAN = 168, /*[102A0]*/ + /** @stable ICU 4.0 */ + UBLOCK_LYDIAN = 169, /*[10920]*/ + /** @stable ICU 4.0 */ + UBLOCK_MAHJONG_TILES = 170, /*[1F000]*/ + /** @stable ICU 4.0 */ + UBLOCK_DOMINO_TILES = 171, /*[1F030]*/ + + /* New blocks in Unicode 5.2 */ + + /** @stable ICU 4.4 */ + UBLOCK_SAMARITAN = 172, /*[0800]*/ + /** @stable ICU 4.4 */ + UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED = 173, /*[18B0]*/ + /** @stable ICU 4.4 */ + UBLOCK_TAI_THAM = 174, /*[1A20]*/ + /** @stable ICU 4.4 */ + UBLOCK_VEDIC_EXTENSIONS = 175, /*[1CD0]*/ + /** @stable ICU 4.4 */ + UBLOCK_LISU = 176, /*[A4D0]*/ + /** @stable ICU 4.4 */ + UBLOCK_BAMUM = 177, /*[A6A0]*/ + /** @stable ICU 4.4 */ + UBLOCK_COMMON_INDIC_NUMBER_FORMS = 178, /*[A830]*/ + /** @stable ICU 4.4 */ + UBLOCK_DEVANAGARI_EXTENDED = 179, /*[A8E0]*/ + /** @stable ICU 4.4 */ + UBLOCK_HANGUL_JAMO_EXTENDED_A = 180, /*[A960]*/ + /** @stable ICU 4.4 */ + UBLOCK_JAVANESE = 181, /*[A980]*/ + /** @stable ICU 4.4 */ + UBLOCK_MYANMAR_EXTENDED_A = 182, /*[AA60]*/ + /** @stable ICU 4.4 */ + UBLOCK_TAI_VIET = 183, /*[AA80]*/ + /** @stable ICU 4.4 */ + UBLOCK_MEETEI_MAYEK = 184, /*[ABC0]*/ + /** @stable ICU 4.4 */ + UBLOCK_HANGUL_JAMO_EXTENDED_B = 185, /*[D7B0]*/ + /** @stable ICU 4.4 */ + UBLOCK_IMPERIAL_ARAMAIC = 186, /*[10840]*/ + /** @stable ICU 4.4 */ + UBLOCK_OLD_SOUTH_ARABIAN = 187, /*[10A60]*/ + /** @stable ICU 4.4 */ + UBLOCK_AVESTAN = 188, /*[10B00]*/ + /** @stable ICU 4.4 */ + UBLOCK_INSCRIPTIONAL_PARTHIAN = 189, /*[10B40]*/ + /** @stable ICU 4.4 */ + UBLOCK_INSCRIPTIONAL_PAHLAVI = 190, /*[10B60]*/ + /** @stable ICU 4.4 */ + UBLOCK_OLD_TURKIC = 191, /*[10C00]*/ + /** @stable ICU 4.4 */ + UBLOCK_RUMI_NUMERAL_SYMBOLS = 192, /*[10E60]*/ + /** @stable ICU 4.4 */ + UBLOCK_KAITHI = 193, /*[11080]*/ + /** @stable ICU 4.4 */ + UBLOCK_EGYPTIAN_HIEROGLYPHS = 194, /*[13000]*/ + /** @stable ICU 4.4 */ + UBLOCK_ENCLOSED_ALPHANUMERIC_SUPPLEMENT = 195, /*[1F100]*/ + /** @stable ICU 4.4 */ + UBLOCK_ENCLOSED_IDEOGRAPHIC_SUPPLEMENT = 196, /*[1F200]*/ + /** @stable ICU 4.4 */ + UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C = 197, /*[2A700]*/ + + /* New blocks in Unicode 6.0 */ + + /** @stable ICU 4.6 */ + UBLOCK_MANDAIC = 198, /*[0840]*/ + /** @stable ICU 4.6 */ + UBLOCK_BATAK = 199, /*[1BC0]*/ + /** @stable ICU 4.6 */ + UBLOCK_ETHIOPIC_EXTENDED_A = 200, /*[AB00]*/ + /** @stable ICU 4.6 */ + UBLOCK_BRAHMI = 201, /*[11000]*/ + /** @stable ICU 4.6 */ + UBLOCK_BAMUM_SUPPLEMENT = 202, /*[16800]*/ + /** @stable ICU 4.6 */ + UBLOCK_KANA_SUPPLEMENT = 203, /*[1B000]*/ + /** @stable ICU 4.6 */ + UBLOCK_PLAYING_CARDS = 204, /*[1F0A0]*/ + /** @stable ICU 4.6 */ + UBLOCK_MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS = 205, /*[1F300]*/ + /** @stable ICU 4.6 */ + UBLOCK_EMOTICONS = 206, /*[1F600]*/ + /** @stable ICU 4.6 */ + UBLOCK_TRANSPORT_AND_MAP_SYMBOLS = 207, /*[1F680]*/ + /** @stable ICU 4.6 */ + UBLOCK_ALCHEMICAL_SYMBOLS = 208, /*[1F700]*/ + /** @stable ICU 4.6 */ + UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D = 209, /*[2B740]*/ + + /* New blocks in Unicode 6.1 */ + + /** @stable ICU 49 */ + UBLOCK_ARABIC_EXTENDED_A = 210, /*[08A0]*/ + /** @stable ICU 49 */ + UBLOCK_ARABIC_MATHEMATICAL_ALPHABETIC_SYMBOLS = 211, /*[1EE00]*/ + /** @stable ICU 49 */ + UBLOCK_CHAKMA = 212, /*[11100]*/ + /** @stable ICU 49 */ + UBLOCK_MEETEI_MAYEK_EXTENSIONS = 213, /*[AAE0]*/ + /** @stable ICU 49 */ + UBLOCK_MEROITIC_CURSIVE = 214, /*[109A0]*/ + /** @stable ICU 49 */ + UBLOCK_MEROITIC_HIEROGLYPHS = 215, /*[10980]*/ + /** @stable ICU 49 */ + UBLOCK_MIAO = 216, /*[16F00]*/ + /** @stable ICU 49 */ + UBLOCK_SHARADA = 217, /*[11180]*/ + /** @stable ICU 49 */ + UBLOCK_SORA_SOMPENG = 218, /*[110D0]*/ + /** @stable ICU 49 */ + UBLOCK_SUNDANESE_SUPPLEMENT = 219, /*[1CC0]*/ + /** @stable ICU 49 */ + UBLOCK_TAKRI = 220, /*[11680]*/ + + /* New blocks in Unicode 7.0 */ + + /** @stable ICU 54 */ + UBLOCK_BASSA_VAH = 221, /*[16AD0]*/ + /** @stable ICU 54 */ + UBLOCK_CAUCASIAN_ALBANIAN = 222, /*[10530]*/ + /** @stable ICU 54 */ + UBLOCK_COPTIC_EPACT_NUMBERS = 223, /*[102E0]*/ + /** @stable ICU 54 */ + UBLOCK_COMBINING_DIACRITICAL_MARKS_EXTENDED = 224, /*[1AB0]*/ + /** @stable ICU 54 */ + UBLOCK_DUPLOYAN = 225, /*[1BC00]*/ + /** @stable ICU 54 */ + UBLOCK_ELBASAN = 226, /*[10500]*/ + /** @stable ICU 54 */ + UBLOCK_GEOMETRIC_SHAPES_EXTENDED = 227, /*[1F780]*/ + /** @stable ICU 54 */ + UBLOCK_GRANTHA = 228, /*[11300]*/ + /** @stable ICU 54 */ + UBLOCK_KHOJKI = 229, /*[11200]*/ + /** @stable ICU 54 */ + UBLOCK_KHUDAWADI = 230, /*[112B0]*/ + /** @stable ICU 54 */ + UBLOCK_LATIN_EXTENDED_E = 231, /*[AB30]*/ + /** @stable ICU 54 */ + UBLOCK_LINEAR_A = 232, /*[10600]*/ + /** @stable ICU 54 */ + UBLOCK_MAHAJANI = 233, /*[11150]*/ + /** @stable ICU 54 */ + UBLOCK_MANICHAEAN = 234, /*[10AC0]*/ + /** @stable ICU 54 */ + UBLOCK_MENDE_KIKAKUI = 235, /*[1E800]*/ + /** @stable ICU 54 */ + UBLOCK_MODI = 236, /*[11600]*/ + /** @stable ICU 54 */ + UBLOCK_MRO = 237, /*[16A40]*/ + /** @stable ICU 54 */ + UBLOCK_MYANMAR_EXTENDED_B = 238, /*[A9E0]*/ + /** @stable ICU 54 */ + UBLOCK_NABATAEAN = 239, /*[10880]*/ + /** @stable ICU 54 */ + UBLOCK_OLD_NORTH_ARABIAN = 240, /*[10A80]*/ + /** @stable ICU 54 */ + UBLOCK_OLD_PERMIC = 241, /*[10350]*/ + /** @stable ICU 54 */ + UBLOCK_ORNAMENTAL_DINGBATS = 242, /*[1F650]*/ + /** @stable ICU 54 */ + UBLOCK_PAHAWH_HMONG = 243, /*[16B00]*/ + /** @stable ICU 54 */ + UBLOCK_PALMYRENE = 244, /*[10860]*/ + /** @stable ICU 54 */ + UBLOCK_PAU_CIN_HAU = 245, /*[11AC0]*/ + /** @stable ICU 54 */ + UBLOCK_PSALTER_PAHLAVI = 246, /*[10B80]*/ + /** @stable ICU 54 */ + UBLOCK_SHORTHAND_FORMAT_CONTROLS = 247, /*[1BCA0]*/ + /** @stable ICU 54 */ + UBLOCK_SIDDHAM = 248, /*[11580]*/ + /** @stable ICU 54 */ + UBLOCK_SINHALA_ARCHAIC_NUMBERS = 249, /*[111E0]*/ + /** @stable ICU 54 */ + UBLOCK_SUPPLEMENTAL_ARROWS_C = 250, /*[1F800]*/ + /** @stable ICU 54 */ + UBLOCK_TIRHUTA = 251, /*[11480]*/ + /** @stable ICU 54 */ + UBLOCK_WARANG_CITI = 252, /*[118A0]*/ + + /* New blocks in Unicode 8.0 */ + + /** @stable ICU 56 */ + UBLOCK_AHOM = 253, /*[11700]*/ + /** @stable ICU 56 */ + UBLOCK_ANATOLIAN_HIEROGLYPHS = 254, /*[14400]*/ + /** @stable ICU 56 */ + UBLOCK_CHEROKEE_SUPPLEMENT = 255, /*[AB70]*/ + /** @stable ICU 56 */ + UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_E = 256, /*[2B820]*/ + /** @stable ICU 56 */ + UBLOCK_EARLY_DYNASTIC_CUNEIFORM = 257, /*[12480]*/ + /** @stable ICU 56 */ + UBLOCK_HATRAN = 258, /*[108E0]*/ + /** @stable ICU 56 */ + UBLOCK_MULTANI = 259, /*[11280]*/ + /** @stable ICU 56 */ + UBLOCK_OLD_HUNGARIAN = 260, /*[10C80]*/ + /** @stable ICU 56 */ + UBLOCK_SUPPLEMENTAL_SYMBOLS_AND_PICTOGRAPHS = 261, /*[1F900]*/ + /** @stable ICU 56 */ + UBLOCK_SUTTON_SIGNWRITING = 262, /*[1D800]*/ + + /* New blocks in Unicode 9.0 */ + + /** @stable ICU 58 */ + UBLOCK_ADLAM = 263, /*[1E900]*/ + /** @stable ICU 58 */ + UBLOCK_BHAIKSUKI = 264, /*[11C00]*/ + /** @stable ICU 58 */ + UBLOCK_CYRILLIC_EXTENDED_C = 265, /*[1C80]*/ + /** @stable ICU 58 */ + UBLOCK_GLAGOLITIC_SUPPLEMENT = 266, /*[1E000]*/ + /** @stable ICU 58 */ + UBLOCK_IDEOGRAPHIC_SYMBOLS_AND_PUNCTUATION = 267, /*[16FE0]*/ + /** @stable ICU 58 */ + UBLOCK_MARCHEN = 268, /*[11C70]*/ + /** @stable ICU 58 */ + UBLOCK_MONGOLIAN_SUPPLEMENT = 269, /*[11660]*/ + /** @stable ICU 58 */ + UBLOCK_NEWA = 270, /*[11400]*/ + /** @stable ICU 58 */ + UBLOCK_OSAGE = 271, /*[104B0]*/ + /** @stable ICU 58 */ + UBLOCK_TANGUT = 272, /*[17000]*/ + /** @stable ICU 58 */ + UBLOCK_TANGUT_COMPONENTS = 273, /*[18800]*/ + + // New blocks in Unicode 10.0 + + /** @stable ICU 60 */ + UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_F = 274, /*[2CEB0]*/ + /** @stable ICU 60 */ + UBLOCK_KANA_EXTENDED_A = 275, /*[1B100]*/ + /** @stable ICU 60 */ + UBLOCK_MASARAM_GONDI = 276, /*[11D00]*/ + /** @stable ICU 60 */ + UBLOCK_NUSHU = 277, /*[1B170]*/ + /** @stable ICU 60 */ + UBLOCK_SOYOMBO = 278, /*[11A50]*/ + /** @stable ICU 60 */ + UBLOCK_SYRIAC_SUPPLEMENT = 279, /*[0860]*/ + /** @stable ICU 60 */ + UBLOCK_ZANABAZAR_SQUARE = 280, /*[11A00]*/ + + // New blocks in Unicode 11.0 + + /** @stable ICU 62 */ + UBLOCK_CHESS_SYMBOLS = 281, /*[1FA00]*/ + /** @stable ICU 62 */ + UBLOCK_DOGRA = 282, /*[11800]*/ + /** @stable ICU 62 */ + UBLOCK_GEORGIAN_EXTENDED = 283, /*[1C90]*/ + /** @stable ICU 62 */ + UBLOCK_GUNJALA_GONDI = 284, /*[11D60]*/ + /** @stable ICU 62 */ + UBLOCK_HANIFI_ROHINGYA = 285, /*[10D00]*/ + /** @stable ICU 62 */ + UBLOCK_INDIC_SIYAQ_NUMBERS = 286, /*[1EC70]*/ + /** @stable ICU 62 */ + UBLOCK_MAKASAR = 287, /*[11EE0]*/ + /** @stable ICU 62 */ + UBLOCK_MAYAN_NUMERALS = 288, /*[1D2E0]*/ + /** @stable ICU 62 */ + UBLOCK_MEDEFAIDRIN = 289, /*[16E40]*/ + /** @stable ICU 62 */ + UBLOCK_OLD_SOGDIAN = 290, /*[10F00]*/ + /** @stable ICU 62 */ + UBLOCK_SOGDIAN = 291, /*[10F30]*/ + + // New blocks in Unicode 12.0 + + /** @stable ICU 64 */ + UBLOCK_EGYPTIAN_HIEROGLYPH_FORMAT_CONTROLS = 292, /*[13430]*/ + /** @stable ICU 64 */ + UBLOCK_ELYMAIC = 293, /*[10FE0]*/ + /** @stable ICU 64 */ + UBLOCK_NANDINAGARI = 294, /*[119A0]*/ + /** @stable ICU 64 */ + UBLOCK_NYIAKENG_PUACHUE_HMONG = 295, /*[1E100]*/ + /** @stable ICU 64 */ + UBLOCK_OTTOMAN_SIYAQ_NUMBERS = 296, /*[1ED00]*/ + /** @stable ICU 64 */ + UBLOCK_SMALL_KANA_EXTENSION = 297, /*[1B130]*/ + /** @stable ICU 64 */ + UBLOCK_SYMBOLS_AND_PICTOGRAPHS_EXTENDED_A = 298, /*[1FA70]*/ + /** @stable ICU 64 */ + UBLOCK_TAMIL_SUPPLEMENT = 299, /*[11FC0]*/ + /** @stable ICU 64 */ + UBLOCK_WANCHO = 300, /*[1E2C0]*/ + + // New blocks in Unicode 13.0 + + /** @stable ICU 66 */ + UBLOCK_CHORASMIAN = 301, /*[10FB0]*/ + /** @stable ICU 66 */ + UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_G = 302, /*[30000]*/ + /** @stable ICU 66 */ + UBLOCK_DIVES_AKURU = 303, /*[11900]*/ + /** @stable ICU 66 */ + UBLOCK_KHITAN_SMALL_SCRIPT = 304, /*[18B00]*/ + /** @stable ICU 66 */ + UBLOCK_LISU_SUPPLEMENT = 305, /*[11FB0]*/ + /** @stable ICU 66 */ + UBLOCK_SYMBOLS_FOR_LEGACY_COMPUTING = 306, /*[1FB00]*/ + /** @stable ICU 66 */ + UBLOCK_TANGUT_SUPPLEMENT = 307, /*[18D00]*/ + /** @stable ICU 66 */ + UBLOCK_YEZIDI = 308, /*[10E80]*/ + + // New blocks in Unicode 14.0 + + /** @stable ICU 70 */ + UBLOCK_ARABIC_EXTENDED_B = 309, /*[0870]*/ + /** @stable ICU 70 */ + UBLOCK_CYPRO_MINOAN = 310, /*[12F90]*/ + /** @stable ICU 70 */ + UBLOCK_ETHIOPIC_EXTENDED_B = 311, /*[1E7E0]*/ + /** @stable ICU 70 */ + UBLOCK_KANA_EXTENDED_B = 312, /*[1AFF0]*/ + /** @stable ICU 70 */ + UBLOCK_LATIN_EXTENDED_F = 313, /*[10780]*/ + /** @stable ICU 70 */ + UBLOCK_LATIN_EXTENDED_G = 314, /*[1DF00]*/ + /** @stable ICU 70 */ + UBLOCK_OLD_UYGHUR = 315, /*[10F70]*/ + /** @stable ICU 70 */ + UBLOCK_TANGSA = 316, /*[16A70]*/ + /** @stable ICU 70 */ + UBLOCK_TOTO = 317, /*[1E290]*/ + /** @stable ICU 70 */ + UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED_A = 318, /*[11AB0]*/ + /** @stable ICU 70 */ + UBLOCK_VITHKUQI = 319, /*[10570]*/ + /** @stable ICU 70 */ + UBLOCK_ZNAMENNY_MUSICAL_NOTATION = 320, /*[1CF00]*/ + + // New blocks in Unicode 15.0 + + /** @stable ICU 72 */ + UBLOCK_ARABIC_EXTENDED_C = 321, /*[10EC0]*/ + /** @stable ICU 72 */ + UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_H = 322, /*[31350]*/ + /** @stable ICU 72 */ + UBLOCK_CYRILLIC_EXTENDED_D = 323, /*[1E030]*/ + /** @stable ICU 72 */ + UBLOCK_DEVANAGARI_EXTENDED_A = 324, /*[11B00]*/ + /** @stable ICU 72 */ + UBLOCK_KAKTOVIK_NUMERALS = 325, /*[1D2C0]*/ + /** @stable ICU 72 */ + UBLOCK_KAWI = 326, /*[11F00]*/ + /** @stable ICU 72 */ + UBLOCK_NAG_MUNDARI = 327, /*[1E4D0]*/ + +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal UBlockCode value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_BLOCK). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + UBLOCK_COUNT = 328, +#endif // U_HIDE_DEPRECATED_API + + /** @stable ICU 2.0 */ + UBLOCK_INVALID_CODE=-1 +}; + +/** @stable ICU 2.0 */ +typedef enum UBlockCode UBlockCode; + +/** + * East Asian Width constants. + * + * @see UCHAR_EAST_ASIAN_WIDTH + * @see u_getIntPropertyValue + * @stable ICU 2.2 + */ +typedef enum UEastAsianWidth { + /* + * Note: UEastAsianWidth constants are parsed by preparseucd.py. + * It matches lines like + * U_EA_ + */ + + U_EA_NEUTRAL, /*[N]*/ + U_EA_AMBIGUOUS, /*[A]*/ + U_EA_HALFWIDTH, /*[H]*/ + U_EA_FULLWIDTH, /*[F]*/ + U_EA_NARROW, /*[Na]*/ + U_EA_WIDE, /*[W]*/ +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal UEastAsianWidth value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_EAST_ASIAN_WIDTH). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_EA_COUNT +#endif // U_HIDE_DEPRECATED_API +} UEastAsianWidth; + +/** + * Selector constants for u_charName(). + * u_charName() returns the "modern" name of a + * Unicode character; or the name that was defined in + * Unicode version 1.0, before the Unicode standard merged + * with ISO-10646; or an "extended" name that gives each + * Unicode code point a unique name. + * + * @see u_charName + * @stable ICU 2.0 + */ +typedef enum UCharNameChoice { + /** Unicode character name (Name property). @stable ICU 2.0 */ + U_UNICODE_CHAR_NAME, +#ifndef U_HIDE_DEPRECATED_API + /** + * The Unicode_1_Name property value which is of little practical value. + * Beginning with ICU 49, ICU APIs return an empty string for this name choice. + * @deprecated ICU 49 + */ + U_UNICODE_10_CHAR_NAME, +#endif /* U_HIDE_DEPRECATED_API */ + /** Standard or synthetic character name. @stable ICU 2.0 */ + U_EXTENDED_CHAR_NAME = U_UNICODE_CHAR_NAME+2, + /** Corrected name from NameAliases.txt. @stable ICU 4.4 */ + U_CHAR_NAME_ALIAS, +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal UCharNameChoice value. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_CHAR_NAME_CHOICE_COUNT +#endif // U_HIDE_DEPRECATED_API +} UCharNameChoice; + +/** + * Selector constants for u_getPropertyName() and + * u_getPropertyValueName(). These selectors are used to choose which + * name is returned for a given property or value. All properties and + * values have a long name. Most have a short name, but some do not. + * Unicode allows for additional names, beyond the long and short + * name, which would be indicated by U_LONG_PROPERTY_NAME + i, where + * i=1, 2,... + * + * @see u_getPropertyName() + * @see u_getPropertyValueName() + * @stable ICU 2.4 + */ +typedef enum UPropertyNameChoice { + U_SHORT_PROPERTY_NAME, + U_LONG_PROPERTY_NAME, +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal UPropertyNameChoice value. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_PROPERTY_NAME_CHOICE_COUNT +#endif // U_HIDE_DEPRECATED_API +} UPropertyNameChoice; + +/** + * Decomposition Type constants. + * + * @see UCHAR_DECOMPOSITION_TYPE + * @stable ICU 2.2 + */ +typedef enum UDecompositionType { + /* + * Note: UDecompositionType constants are parsed by preparseucd.py. + * It matches lines like + * U_DT_ + */ + + U_DT_NONE, /*[none]*/ + U_DT_CANONICAL, /*[can]*/ + U_DT_COMPAT, /*[com]*/ + U_DT_CIRCLE, /*[enc]*/ + U_DT_FINAL, /*[fin]*/ + U_DT_FONT, /*[font]*/ + U_DT_FRACTION, /*[fra]*/ + U_DT_INITIAL, /*[init]*/ + U_DT_ISOLATED, /*[iso]*/ + U_DT_MEDIAL, /*[med]*/ + U_DT_NARROW, /*[nar]*/ + U_DT_NOBREAK, /*[nb]*/ + U_DT_SMALL, /*[sml]*/ + U_DT_SQUARE, /*[sqr]*/ + U_DT_SUB, /*[sub]*/ + U_DT_SUPER, /*[sup]*/ + U_DT_VERTICAL, /*[vert]*/ + U_DT_WIDE, /*[wide]*/ +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal UDecompositionType value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_DECOMPOSITION_TYPE). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_DT_COUNT /* 18 */ +#endif // U_HIDE_DEPRECATED_API +} UDecompositionType; + +/** + * Joining Type constants. + * + * @see UCHAR_JOINING_TYPE + * @stable ICU 2.2 + */ +typedef enum UJoiningType { + /* + * Note: UJoiningType constants are parsed by preparseucd.py. + * It matches lines like + * U_JT_ + */ + + U_JT_NON_JOINING, /*[U]*/ + U_JT_JOIN_CAUSING, /*[C]*/ + U_JT_DUAL_JOINING, /*[D]*/ + U_JT_LEFT_JOINING, /*[L]*/ + U_JT_RIGHT_JOINING, /*[R]*/ + U_JT_TRANSPARENT, /*[T]*/ +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal UJoiningType value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_JOINING_TYPE). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_JT_COUNT /* 6 */ +#endif // U_HIDE_DEPRECATED_API +} UJoiningType; + +/** + * Joining Group constants. + * + * @see UCHAR_JOINING_GROUP + * @stable ICU 2.2 + */ +typedef enum UJoiningGroup { + /* + * Note: UJoiningGroup constants are parsed by preparseucd.py. + * It matches lines like + * U_JG_ + */ + + U_JG_NO_JOINING_GROUP, + U_JG_AIN, + U_JG_ALAPH, + U_JG_ALEF, + U_JG_BEH, + U_JG_BETH, + U_JG_DAL, + U_JG_DALATH_RISH, + U_JG_E, + U_JG_FEH, + U_JG_FINAL_SEMKATH, + U_JG_GAF, + U_JG_GAMAL, + U_JG_HAH, + U_JG_TEH_MARBUTA_GOAL, /**< @stable ICU 4.6 */ + U_JG_HAMZA_ON_HEH_GOAL=U_JG_TEH_MARBUTA_GOAL, + U_JG_HE, + U_JG_HEH, + U_JG_HEH_GOAL, + U_JG_HETH, + U_JG_KAF, + U_JG_KAPH, + U_JG_KNOTTED_HEH, + U_JG_LAM, + U_JG_LAMADH, + U_JG_MEEM, + U_JG_MIM, + U_JG_NOON, + U_JG_NUN, + U_JG_PE, + U_JG_QAF, + U_JG_QAPH, + U_JG_REH, + U_JG_REVERSED_PE, + U_JG_SAD, + U_JG_SADHE, + U_JG_SEEN, + U_JG_SEMKATH, + U_JG_SHIN, + U_JG_SWASH_KAF, + U_JG_SYRIAC_WAW, + U_JG_TAH, + U_JG_TAW, + U_JG_TEH_MARBUTA, + U_JG_TETH, + U_JG_WAW, + U_JG_YEH, + U_JG_YEH_BARREE, + U_JG_YEH_WITH_TAIL, + U_JG_YUDH, + U_JG_YUDH_HE, + U_JG_ZAIN, + U_JG_FE, /**< @stable ICU 2.6 */ + U_JG_KHAPH, /**< @stable ICU 2.6 */ + U_JG_ZHAIN, /**< @stable ICU 2.6 */ + U_JG_BURUSHASKI_YEH_BARREE, /**< @stable ICU 4.0 */ + U_JG_FARSI_YEH, /**< @stable ICU 4.4 */ + U_JG_NYA, /**< @stable ICU 4.4 */ + U_JG_ROHINGYA_YEH, /**< @stable ICU 49 */ + U_JG_MANICHAEAN_ALEPH, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_AYIN, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_BETH, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_DALETH, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_DHAMEDH, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_FIVE, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_GIMEL, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_HETH, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_HUNDRED, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_KAPH, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_LAMEDH, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_MEM, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_NUN, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_ONE, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_PE, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_QOPH, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_RESH, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_SADHE, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_SAMEKH, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_TAW, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_TEN, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_TETH, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_THAMEDH, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_TWENTY, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_WAW, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_YODH, /**< @stable ICU 54 */ + U_JG_MANICHAEAN_ZAYIN, /**< @stable ICU 54 */ + U_JG_STRAIGHT_WAW, /**< @stable ICU 54 */ + U_JG_AFRICAN_FEH, /**< @stable ICU 58 */ + U_JG_AFRICAN_NOON, /**< @stable ICU 58 */ + U_JG_AFRICAN_QAF, /**< @stable ICU 58 */ + + U_JG_MALAYALAM_BHA, /**< @stable ICU 60 */ + U_JG_MALAYALAM_JA, /**< @stable ICU 60 */ + U_JG_MALAYALAM_LLA, /**< @stable ICU 60 */ + U_JG_MALAYALAM_LLLA, /**< @stable ICU 60 */ + U_JG_MALAYALAM_NGA, /**< @stable ICU 60 */ + U_JG_MALAYALAM_NNA, /**< @stable ICU 60 */ + U_JG_MALAYALAM_NNNA, /**< @stable ICU 60 */ + U_JG_MALAYALAM_NYA, /**< @stable ICU 60 */ + U_JG_MALAYALAM_RA, /**< @stable ICU 60 */ + U_JG_MALAYALAM_SSA, /**< @stable ICU 60 */ + U_JG_MALAYALAM_TTA, /**< @stable ICU 60 */ + + U_JG_HANIFI_ROHINGYA_KINNA_YA, /**< @stable ICU 62 */ + U_JG_HANIFI_ROHINGYA_PA, /**< @stable ICU 62 */ + + U_JG_THIN_YEH, /**< @stable ICU 70 */ + U_JG_VERTICAL_TAIL, /**< @stable ICU 70 */ + +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal UJoiningGroup value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_JOINING_GROUP). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_JG_COUNT +#endif // U_HIDE_DEPRECATED_API +} UJoiningGroup; + +/** + * Grapheme Cluster Break constants. + * + * @see UCHAR_GRAPHEME_CLUSTER_BREAK + * @stable ICU 3.4 + */ +typedef enum UGraphemeClusterBreak { + /* + * Note: UGraphemeClusterBreak constants are parsed by preparseucd.py. + * It matches lines like + * U_GCB_ + */ + + U_GCB_OTHER = 0, /*[XX]*/ + U_GCB_CONTROL = 1, /*[CN]*/ + U_GCB_CR = 2, /*[CR]*/ + U_GCB_EXTEND = 3, /*[EX]*/ + U_GCB_L = 4, /*[L]*/ + U_GCB_LF = 5, /*[LF]*/ + U_GCB_LV = 6, /*[LV]*/ + U_GCB_LVT = 7, /*[LVT]*/ + U_GCB_T = 8, /*[T]*/ + U_GCB_V = 9, /*[V]*/ + /** @stable ICU 4.0 */ + U_GCB_SPACING_MARK = 10, /*[SM]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */ + /** @stable ICU 4.0 */ + U_GCB_PREPEND = 11, /*[PP]*/ + /** @stable ICU 50 */ + U_GCB_REGIONAL_INDICATOR = 12, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */ + /** @stable ICU 58 */ + U_GCB_E_BASE = 13, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */ + /** @stable ICU 58 */ + U_GCB_E_BASE_GAZ = 14, /*[EBG]*/ + /** @stable ICU 58 */ + U_GCB_E_MODIFIER = 15, /*[EM]*/ + /** @stable ICU 58 */ + U_GCB_GLUE_AFTER_ZWJ = 16, /*[GAZ]*/ + /** @stable ICU 58 */ + U_GCB_ZWJ = 17, /*[ZWJ]*/ + +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal UGraphemeClusterBreak value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_GRAPHEME_CLUSTER_BREAK). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_GCB_COUNT = 18 +#endif // U_HIDE_DEPRECATED_API +} UGraphemeClusterBreak; + +/** + * Word Break constants. + * (UWordBreak is a pre-existing enum type in ubrk.h for word break status tags.) + * + * @see UCHAR_WORD_BREAK + * @stable ICU 3.4 + */ +typedef enum UWordBreakValues { + /* + * Note: UWordBreakValues constants are parsed by preparseucd.py. + * It matches lines like + * U_WB_ + */ + + U_WB_OTHER = 0, /*[XX]*/ + U_WB_ALETTER = 1, /*[LE]*/ + U_WB_FORMAT = 2, /*[FO]*/ + U_WB_KATAKANA = 3, /*[KA]*/ + U_WB_MIDLETTER = 4, /*[ML]*/ + U_WB_MIDNUM = 5, /*[MN]*/ + U_WB_NUMERIC = 6, /*[NU]*/ + U_WB_EXTENDNUMLET = 7, /*[EX]*/ + /** @stable ICU 4.0 */ + U_WB_CR = 8, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */ + /** @stable ICU 4.0 */ + U_WB_EXTEND = 9, /*[Extend]*/ + /** @stable ICU 4.0 */ + U_WB_LF = 10, /*[LF]*/ + /** @stable ICU 4.0 */ + U_WB_MIDNUMLET =11, /*[MB]*/ + /** @stable ICU 4.0 */ + U_WB_NEWLINE =12, /*[NL]*/ + /** @stable ICU 50 */ + U_WB_REGIONAL_INDICATOR = 13, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */ + /** @stable ICU 52 */ + U_WB_HEBREW_LETTER = 14, /*[HL]*/ /* from here on: new in Unicode 6.3/ICU 52 */ + /** @stable ICU 52 */ + U_WB_SINGLE_QUOTE = 15, /*[SQ]*/ + /** @stable ICU 52 */ + U_WB_DOUBLE_QUOTE = 16, /*[DQ]*/ + /** @stable ICU 58 */ + U_WB_E_BASE = 17, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */ + /** @stable ICU 58 */ + U_WB_E_BASE_GAZ = 18, /*[EBG]*/ + /** @stable ICU 58 */ + U_WB_E_MODIFIER = 19, /*[EM]*/ + /** @stable ICU 58 */ + U_WB_GLUE_AFTER_ZWJ = 20, /*[GAZ]*/ + /** @stable ICU 58 */ + U_WB_ZWJ = 21, /*[ZWJ]*/ + /** @stable ICU 62 */ + U_WB_WSEGSPACE = 22, /*[WSEGSPACE]*/ + +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal UWordBreakValues value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_WORD_BREAK). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_WB_COUNT = 23 +#endif // U_HIDE_DEPRECATED_API +} UWordBreakValues; + +/** + * Sentence Break constants. + * + * @see UCHAR_SENTENCE_BREAK + * @stable ICU 3.4 + */ +typedef enum USentenceBreak { + /* + * Note: USentenceBreak constants are parsed by preparseucd.py. + * It matches lines like + * U_SB_ + */ + + U_SB_OTHER = 0, /*[XX]*/ + U_SB_ATERM = 1, /*[AT]*/ + U_SB_CLOSE = 2, /*[CL]*/ + U_SB_FORMAT = 3, /*[FO]*/ + U_SB_LOWER = 4, /*[LO]*/ + U_SB_NUMERIC = 5, /*[NU]*/ + U_SB_OLETTER = 6, /*[LE]*/ + U_SB_SEP = 7, /*[SE]*/ + U_SB_SP = 8, /*[SP]*/ + U_SB_STERM = 9, /*[ST]*/ + U_SB_UPPER = 10, /*[UP]*/ + U_SB_CR = 11, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */ + U_SB_EXTEND = 12, /*[EX]*/ + U_SB_LF = 13, /*[LF]*/ + U_SB_SCONTINUE = 14, /*[SC]*/ +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal USentenceBreak value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_SENTENCE_BREAK). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_SB_COUNT = 15 +#endif // U_HIDE_DEPRECATED_API +} USentenceBreak; + +/** + * Line Break constants. + * + * @see UCHAR_LINE_BREAK + * @stable ICU 2.2 + */ +typedef enum ULineBreak { + /* + * Note: ULineBreak constants are parsed by preparseucd.py. + * It matches lines like + * U_LB_ + */ + + U_LB_UNKNOWN = 0, /*[XX]*/ + U_LB_AMBIGUOUS = 1, /*[AI]*/ + U_LB_ALPHABETIC = 2, /*[AL]*/ + U_LB_BREAK_BOTH = 3, /*[B2]*/ + U_LB_BREAK_AFTER = 4, /*[BA]*/ + U_LB_BREAK_BEFORE = 5, /*[BB]*/ + U_LB_MANDATORY_BREAK = 6, /*[BK]*/ + U_LB_CONTINGENT_BREAK = 7, /*[CB]*/ + U_LB_CLOSE_PUNCTUATION = 8, /*[CL]*/ + U_LB_COMBINING_MARK = 9, /*[CM]*/ + U_LB_CARRIAGE_RETURN = 10, /*[CR]*/ + U_LB_EXCLAMATION = 11, /*[EX]*/ + U_LB_GLUE = 12, /*[GL]*/ + U_LB_HYPHEN = 13, /*[HY]*/ + U_LB_IDEOGRAPHIC = 14, /*[ID]*/ + /** Renamed from the misspelled "inseperable" in Unicode 4.0.1/ICU 3.0 @stable ICU 3.0 */ + U_LB_INSEPARABLE = 15, /*[IN]*/ + U_LB_INSEPERABLE = U_LB_INSEPARABLE, + U_LB_INFIX_NUMERIC = 16, /*[IS]*/ + U_LB_LINE_FEED = 17, /*[LF]*/ + U_LB_NONSTARTER = 18, /*[NS]*/ + U_LB_NUMERIC = 19, /*[NU]*/ + U_LB_OPEN_PUNCTUATION = 20, /*[OP]*/ + U_LB_POSTFIX_NUMERIC = 21, /*[PO]*/ + U_LB_PREFIX_NUMERIC = 22, /*[PR]*/ + U_LB_QUOTATION = 23, /*[QU]*/ + U_LB_COMPLEX_CONTEXT = 24, /*[SA]*/ + U_LB_SURROGATE = 25, /*[SG]*/ + U_LB_SPACE = 26, /*[SP]*/ + U_LB_BREAK_SYMBOLS = 27, /*[SY]*/ + U_LB_ZWSPACE = 28, /*[ZW]*/ + /** @stable ICU 2.6 */ + U_LB_NEXT_LINE = 29, /*[NL]*/ /* from here on: new in Unicode 4/ICU 2.6 */ + /** @stable ICU 2.6 */ + U_LB_WORD_JOINER = 30, /*[WJ]*/ + /** @stable ICU 3.4 */ + U_LB_H2 = 31, /*[H2]*/ /* from here on: new in Unicode 4.1/ICU 3.4 */ + /** @stable ICU 3.4 */ + U_LB_H3 = 32, /*[H3]*/ + /** @stable ICU 3.4 */ + U_LB_JL = 33, /*[JL]*/ + /** @stable ICU 3.4 */ + U_LB_JT = 34, /*[JT]*/ + /** @stable ICU 3.4 */ + U_LB_JV = 35, /*[JV]*/ + /** @stable ICU 4.4 */ + U_LB_CLOSE_PARENTHESIS = 36, /*[CP]*/ /* new in Unicode 5.2/ICU 4.4 */ + /** @stable ICU 49 */ + U_LB_CONDITIONAL_JAPANESE_STARTER = 37,/*[CJ]*/ /* new in Unicode 6.1/ICU 49 */ + /** @stable ICU 49 */ + U_LB_HEBREW_LETTER = 38, /*[HL]*/ /* new in Unicode 6.1/ICU 49 */ + /** @stable ICU 50 */ + U_LB_REGIONAL_INDICATOR = 39,/*[RI]*/ /* new in Unicode 6.2/ICU 50 */ + /** @stable ICU 58 */ + U_LB_E_BASE = 40, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */ + /** @stable ICU 58 */ + U_LB_E_MODIFIER = 41, /*[EM]*/ + /** @stable ICU 58 */ + U_LB_ZWJ = 42, /*[ZWJ]*/ +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal ULineBreak value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_LINE_BREAK). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_LB_COUNT = 43 +#endif // U_HIDE_DEPRECATED_API +} ULineBreak; + +/** + * Numeric Type constants. + * + * @see UCHAR_NUMERIC_TYPE + * @stable ICU 2.2 + */ +typedef enum UNumericType { + /* + * Note: UNumericType constants are parsed by preparseucd.py. + * It matches lines like + * U_NT_ + */ + + U_NT_NONE, /*[None]*/ + U_NT_DECIMAL, /*[de]*/ + U_NT_DIGIT, /*[di]*/ + U_NT_NUMERIC, /*[nu]*/ +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal UNumericType value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_NUMERIC_TYPE). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_NT_COUNT +#endif // U_HIDE_DEPRECATED_API +} UNumericType; + +/** + * Hangul Syllable Type constants. + * + * @see UCHAR_HANGUL_SYLLABLE_TYPE + * @stable ICU 2.6 + */ +typedef enum UHangulSyllableType { + /* + * Note: UHangulSyllableType constants are parsed by preparseucd.py. + * It matches lines like + * U_HST_ + */ + + U_HST_NOT_APPLICABLE, /*[NA]*/ + U_HST_LEADING_JAMO, /*[L]*/ + U_HST_VOWEL_JAMO, /*[V]*/ + U_HST_TRAILING_JAMO, /*[T]*/ + U_HST_LV_SYLLABLE, /*[LV]*/ + U_HST_LVT_SYLLABLE, /*[LVT]*/ +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal UHangulSyllableType value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_HANGUL_SYLLABLE_TYPE). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + U_HST_COUNT +#endif // U_HIDE_DEPRECATED_API +} UHangulSyllableType; + +/** + * Indic Positional Category constants. + * + * @see UCHAR_INDIC_POSITIONAL_CATEGORY + * @stable ICU 63 + */ +typedef enum UIndicPositionalCategory { + /* + * Note: UIndicPositionalCategory constants are parsed by preparseucd.py. + * It matches lines like + * U_INPC_ + */ + + /** @stable ICU 63 */ + U_INPC_NA, + /** @stable ICU 63 */ + U_INPC_BOTTOM, + /** @stable ICU 63 */ + U_INPC_BOTTOM_AND_LEFT, + /** @stable ICU 63 */ + U_INPC_BOTTOM_AND_RIGHT, + /** @stable ICU 63 */ + U_INPC_LEFT, + /** @stable ICU 63 */ + U_INPC_LEFT_AND_RIGHT, + /** @stable ICU 63 */ + U_INPC_OVERSTRUCK, + /** @stable ICU 63 */ + U_INPC_RIGHT, + /** @stable ICU 63 */ + U_INPC_TOP, + /** @stable ICU 63 */ + U_INPC_TOP_AND_BOTTOM, + /** @stable ICU 63 */ + U_INPC_TOP_AND_BOTTOM_AND_RIGHT, + /** @stable ICU 63 */ + U_INPC_TOP_AND_LEFT, + /** @stable ICU 63 */ + U_INPC_TOP_AND_LEFT_AND_RIGHT, + /** @stable ICU 63 */ + U_INPC_TOP_AND_RIGHT, + /** @stable ICU 63 */ + U_INPC_VISUAL_ORDER_LEFT, + /** @stable ICU 66 */ + U_INPC_TOP_AND_BOTTOM_AND_LEFT, +} UIndicPositionalCategory; + +/** + * Indic Syllabic Category constants. + * + * @see UCHAR_INDIC_SYLLABIC_CATEGORY + * @stable ICU 63 + */ +typedef enum UIndicSyllabicCategory { + /* + * Note: UIndicSyllabicCategory constants are parsed by preparseucd.py. + * It matches lines like + * U_INSC_ + */ + + /** @stable ICU 63 */ + U_INSC_OTHER, + /** @stable ICU 63 */ + U_INSC_AVAGRAHA, + /** @stable ICU 63 */ + U_INSC_BINDU, + /** @stable ICU 63 */ + U_INSC_BRAHMI_JOINING_NUMBER, + /** @stable ICU 63 */ + U_INSC_CANTILLATION_MARK, + /** @stable ICU 63 */ + U_INSC_CONSONANT, + /** @stable ICU 63 */ + U_INSC_CONSONANT_DEAD, + /** @stable ICU 63 */ + U_INSC_CONSONANT_FINAL, + /** @stable ICU 63 */ + U_INSC_CONSONANT_HEAD_LETTER, + /** @stable ICU 63 */ + U_INSC_CONSONANT_INITIAL_POSTFIXED, + /** @stable ICU 63 */ + U_INSC_CONSONANT_KILLER, + /** @stable ICU 63 */ + U_INSC_CONSONANT_MEDIAL, + /** @stable ICU 63 */ + U_INSC_CONSONANT_PLACEHOLDER, + /** @stable ICU 63 */ + U_INSC_CONSONANT_PRECEDING_REPHA, + /** @stable ICU 63 */ + U_INSC_CONSONANT_PREFIXED, + /** @stable ICU 63 */ + U_INSC_CONSONANT_SUBJOINED, + /** @stable ICU 63 */ + U_INSC_CONSONANT_SUCCEEDING_REPHA, + /** @stable ICU 63 */ + U_INSC_CONSONANT_WITH_STACKER, + /** @stable ICU 63 */ + U_INSC_GEMINATION_MARK, + /** @stable ICU 63 */ + U_INSC_INVISIBLE_STACKER, + /** @stable ICU 63 */ + U_INSC_JOINER, + /** @stable ICU 63 */ + U_INSC_MODIFYING_LETTER, + /** @stable ICU 63 */ + U_INSC_NON_JOINER, + /** @stable ICU 63 */ + U_INSC_NUKTA, + /** @stable ICU 63 */ + U_INSC_NUMBER, + /** @stable ICU 63 */ + U_INSC_NUMBER_JOINER, + /** @stable ICU 63 */ + U_INSC_PURE_KILLER, + /** @stable ICU 63 */ + U_INSC_REGISTER_SHIFTER, + /** @stable ICU 63 */ + U_INSC_SYLLABLE_MODIFIER, + /** @stable ICU 63 */ + U_INSC_TONE_LETTER, + /** @stable ICU 63 */ + U_INSC_TONE_MARK, + /** @stable ICU 63 */ + U_INSC_VIRAMA, + /** @stable ICU 63 */ + U_INSC_VISARGA, + /** @stable ICU 63 */ + U_INSC_VOWEL, + /** @stable ICU 63 */ + U_INSC_VOWEL_DEPENDENT, + /** @stable ICU 63 */ + U_INSC_VOWEL_INDEPENDENT, +} UIndicSyllabicCategory; + +/** + * Vertical Orientation constants. + * + * @see UCHAR_VERTICAL_ORIENTATION + * @stable ICU 63 + */ +typedef enum UVerticalOrientation { + /* + * Note: UVerticalOrientation constants are parsed by preparseucd.py. + * It matches lines like + * U_VO_ + */ + + /** @stable ICU 63 */ + U_VO_ROTATED, + /** @stable ICU 63 */ + U_VO_TRANSFORMED_ROTATED, + /** @stable ICU 63 */ + U_VO_TRANSFORMED_UPRIGHT, + /** @stable ICU 63 */ + U_VO_UPRIGHT, +} UVerticalOrientation; + +/** + * Check a binary Unicode property for a code point. + * + * Unicode, especially in version 3.2, defines many more properties than the + * original set in UnicodeData.txt. + * + * The properties APIs are intended to reflect Unicode properties as defined + * in the Unicode Character Database (UCD) and Unicode Technical Reports (UTR). + * For details about the properties see http://www.unicode.org/ucd/ . + * For names of Unicode properties see the UCD file PropertyAliases.txt. + * + * Important: If ICU is built with UCD files from Unicode versions below 3.2, + * then properties marked with "new in Unicode 3.2" are not or not fully available. + * + * @param c Code point to test. + * @param which UProperty selector constant, identifies which binary property to check. + * Must be UCHAR_BINARY_START<=which<UCHAR_BINARY_LIMIT. + * @return true or false according to the binary Unicode property value for c. + * Also false if 'which' is out of bounds or if the Unicode version + * does not have data for the property at all. + * + * @see UProperty + * @see u_getBinaryPropertySet + * @see u_getIntPropertyValue + * @see u_getUnicodeVersion + * @stable ICU 2.1 + */ +U_CAPI UBool U_EXPORT2 +u_hasBinaryProperty(UChar32 c, UProperty which); + +/** + * Returns true if the property is true for the string. + * Same as u_hasBinaryProperty(single code point, which) + * if the string contains exactly one code point. + * + * Most properties apply only to single code points. + * UTS #51 Unicode Emoji + * defines several properties of strings. + * + * @param s String to test. + * @param length Length of the string, or negative if NUL-terminated. + * @param which UProperty selector constant, identifies which binary property to check. + * Must be UCHAR_BINARY_START<=which<UCHAR_BINARY_LIMIT. + * @return true or false according to the binary Unicode property value for the string. + * Also false if 'which' is out of bounds or if the Unicode version + * does not have data for the property at all. + * + * @see UProperty + * @see u_hasBinaryProperty + * @see u_getBinaryPropertySet + * @see u_getIntPropertyValue + * @see u_getUnicodeVersion + * @stable ICU 70 + */ +U_CAPI UBool U_EXPORT2 +u_stringHasBinaryProperty(const UChar *s, int32_t length, UProperty which); + +/** + * Returns a frozen USet for a binary property. + * The library retains ownership over the returned object. + * Sets an error code if the property number is not one for a binary property. + * + * The returned set contains all code points for which the property is true. + * + * @param property UCHAR_BINARY_START..UCHAR_BINARY_LIMIT-1 + * @param pErrorCode an in/out ICU UErrorCode + * @return the property as a set + * @see UProperty + * @see u_hasBinaryProperty + * @see Unicode::fromUSet + * @stable ICU 63 + */ +U_CAPI const USet * U_EXPORT2 +u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode); + +/** + * Check if a code point has the Alphabetic Unicode property. + * Same as u_hasBinaryProperty(c, UCHAR_ALPHABETIC). + * This is different from u_isalpha! + * @param c Code point to test + * @return true if the code point has the Alphabetic Unicode property, false otherwise + * + * @see UCHAR_ALPHABETIC + * @see u_isalpha + * @see u_hasBinaryProperty + * @stable ICU 2.1 + */ +U_CAPI UBool U_EXPORT2 +u_isUAlphabetic(UChar32 c); + +/** + * Check if a code point has the Lowercase Unicode property. + * Same as u_hasBinaryProperty(c, UCHAR_LOWERCASE). + * This is different from u_islower! + * @param c Code point to test + * @return true if the code point has the Lowercase Unicode property, false otherwise + * + * @see UCHAR_LOWERCASE + * @see u_islower + * @see u_hasBinaryProperty + * @stable ICU 2.1 + */ +U_CAPI UBool U_EXPORT2 +u_isULowercase(UChar32 c); + +/** + * Check if a code point has the Uppercase Unicode property. + * Same as u_hasBinaryProperty(c, UCHAR_UPPERCASE). + * This is different from u_isupper! + * @param c Code point to test + * @return true if the code point has the Uppercase Unicode property, false otherwise + * + * @see UCHAR_UPPERCASE + * @see u_isupper + * @see u_hasBinaryProperty + * @stable ICU 2.1 + */ +U_CAPI UBool U_EXPORT2 +u_isUUppercase(UChar32 c); + +/** + * Check if a code point has the White_Space Unicode property. + * Same as u_hasBinaryProperty(c, UCHAR_WHITE_SPACE). + * This is different from both u_isspace and u_isWhitespace! + * + * Note: There are several ICU whitespace functions; please see the uchar.h + * file documentation for a detailed comparison. + * + * @param c Code point to test + * @return true if the code point has the White_Space Unicode property, false otherwise. + * + * @see UCHAR_WHITE_SPACE + * @see u_isWhitespace + * @see u_isspace + * @see u_isJavaSpaceChar + * @see u_hasBinaryProperty + * @stable ICU 2.1 + */ +U_CAPI UBool U_EXPORT2 +u_isUWhiteSpace(UChar32 c); + +/** + * Get the property value for an enumerated or integer Unicode property for a code point. + * Also returns binary and mask property values. + * + * Unicode, especially in version 3.2, defines many more properties than the + * original set in UnicodeData.txt. + * + * The properties APIs are intended to reflect Unicode properties as defined + * in the Unicode Character Database (UCD) and Unicode Technical Reports (UTR). + * For details about the properties see http://www.unicode.org/ . + * For names of Unicode properties see the UCD file PropertyAliases.txt. + * + * Sample usage: + * UEastAsianWidth ea=(UEastAsianWidth)u_getIntPropertyValue(c, UCHAR_EAST_ASIAN_WIDTH); + * UBool b=(UBool)u_getIntPropertyValue(c, UCHAR_IDEOGRAPHIC); + * + * @param c Code point to test. + * @param which UProperty selector constant, identifies which property to check. + * Must be UCHAR_BINARY_START<=which=0. + * True for characters with general category "Nd" (decimal digit numbers) + * as well as Latin letters a-f and A-F in both ASCII and Fullwidth ASCII. + * (That is, for letters with code points + * 0041..0046, 0061..0066, FF21..FF26, FF41..FF46.) + * + * In order to narrow the definition of hexadecimal digits to only ASCII + * characters, use (c<=0x7f && u_isxdigit(c)). + * + * This is a C/POSIX migration function. + * See the comments about C/POSIX character classification functions in the + * documentation at the top of this header file. + * + * @param c the code point to be tested + * @return true if the code point is a hexadecimal digit + * + * @stable ICU 2.6 + */ +U_CAPI UBool U_EXPORT2 +u_isxdigit(UChar32 c); + +/** + * Determines whether the specified code point is a punctuation character. + * True for characters with general categories "P" (punctuation). + * + * This is a C/POSIX migration function. + * See the comments about C/POSIX character classification functions in the + * documentation at the top of this header file. + * + * @param c the code point to be tested + * @return true if the code point is a punctuation character + * + * @stable ICU 2.6 + */ +U_CAPI UBool U_EXPORT2 +u_ispunct(UChar32 c); + +/** + * Determines whether the specified code point is a "graphic" character + * (printable, excluding spaces). + * true for all characters except those with general categories + * "Cc" (control codes), "Cf" (format controls), "Cs" (surrogates), + * "Cn" (unassigned), and "Z" (separators). + * + * This is a C/POSIX migration function. + * See the comments about C/POSIX character classification functions in the + * documentation at the top of this header file. + * + * @param c the code point to be tested + * @return true if the code point is a "graphic" character + * + * @stable ICU 2.6 + */ +U_CAPI UBool U_EXPORT2 +u_isgraph(UChar32 c); + +/** + * Determines whether the specified code point is a "blank" or "horizontal space", + * a character that visibly separates words on a line. + * The following are equivalent definitions: + * + * true for Unicode White_Space characters except for "vertical space controls" + * where "vertical space controls" are the following characters: + * U+000A (LF) U+000B (VT) U+000C (FF) U+000D (CR) U+0085 (NEL) U+2028 (LS) U+2029 (PS) + * + * same as + * + * true for U+0009 (TAB) and characters with general category "Zs" (space separators). + * + * Note: There are several ICU whitespace functions; please see the uchar.h + * file documentation for a detailed comparison. + * + * This is a C/POSIX migration function. + * See the comments about C/POSIX character classification functions in the + * documentation at the top of this header file. + * + * @param c the code point to be tested + * @return true if the code point is a "blank" + * + * @stable ICU 2.6 + */ +U_CAPI UBool U_EXPORT2 +u_isblank(UChar32 c); + +/** + * Determines whether the specified code point is "defined", + * which usually means that it is assigned a character. + * True for general categories other than "Cn" (other, not assigned), + * i.e., true for all code points mentioned in UnicodeData.txt. + * + * Note that non-character code points (e.g., U+FDD0) are not "defined" + * (they are Cn), but surrogate code points are "defined" (Cs). + * + * Same as java.lang.Character.isDefined(). + * + * @param c the code point to be tested + * @return true if the code point is assigned a character + * + * @see u_isdigit + * @see u_isalpha + * @see u_isalnum + * @see u_isupper + * @see u_islower + * @see u_istitle + * @stable ICU 2.0 + */ +U_CAPI UBool U_EXPORT2 +u_isdefined(UChar32 c); + +/** + * Determines if the specified character is a space character or not. + * + * Note: There are several ICU whitespace functions; please see the uchar.h + * file documentation for a detailed comparison. + * + * This is a C/POSIX migration function. + * See the comments about C/POSIX character classification functions in the + * documentation at the top of this header file. + * + * @param c the character to be tested + * @return true if the character is a space character; false otherwise. + * + * @see u_isJavaSpaceChar + * @see u_isWhitespace + * @see u_isUWhiteSpace + * @stable ICU 2.0 + */ +U_CAPI UBool U_EXPORT2 +u_isspace(UChar32 c); + +/** + * Determine if the specified code point is a space character according to Java. + * True for characters with general categories "Z" (separators), + * which does not include control codes (e.g., TAB or Line Feed). + * + * Same as java.lang.Character.isSpaceChar(). + * + * Note: There are several ICU whitespace functions; please see the uchar.h + * file documentation for a detailed comparison. + * + * @param c the code point to be tested + * @return true if the code point is a space character according to Character.isSpaceChar() + * + * @see u_isspace + * @see u_isWhitespace + * @see u_isUWhiteSpace + * @stable ICU 2.6 + */ +U_CAPI UBool U_EXPORT2 +u_isJavaSpaceChar(UChar32 c); + +/** + * Determines if the specified code point is a whitespace character according to Java/ICU. + * A character is considered to be a Java whitespace character if and only + * if it satisfies one of the following criteria: + * + * - It is a Unicode Separator character (categories "Z" = "Zs" or "Zl" or "Zp"), but is not + * also a non-breaking space (U+00A0 NBSP or U+2007 Figure Space or U+202F Narrow NBSP). + * - It is U+0009 HORIZONTAL TABULATION. + * - It is U+000A LINE FEED. + * - It is U+000B VERTICAL TABULATION. + * - It is U+000C FORM FEED. + * - It is U+000D CARRIAGE RETURN. + * - It is U+001C FILE SEPARATOR. + * - It is U+001D GROUP SEPARATOR. + * - It is U+001E RECORD SEPARATOR. + * - It is U+001F UNIT SEPARATOR. + * + * This API tries to sync with the semantics of Java's + * java.lang.Character.isWhitespace(), but it may not return + * the exact same results because of the Unicode version + * difference. + * + * Note: Unicode 4.0.1 changed U+200B ZERO WIDTH SPACE from a Space Separator (Zs) + * to a Format Control (Cf). Since then, isWhitespace(0x200b) returns false. + * See http://www.unicode.org/versions/Unicode4.0.1/ + * + * Note: There are several ICU whitespace functions; please see the uchar.h + * file documentation for a detailed comparison. + * + * @param c the code point to be tested + * @return true if the code point is a whitespace character according to Java/ICU + * + * @see u_isspace + * @see u_isJavaSpaceChar + * @see u_isUWhiteSpace + * @stable ICU 2.0 + */ +U_CAPI UBool U_EXPORT2 +u_isWhitespace(UChar32 c); + +/** + * Determines whether the specified code point is a control character + * (as defined by this function). + * A control character is one of the following: + * - ISO 8-bit control character (U+0000..U+001f and U+007f..U+009f) + * - U_CONTROL_CHAR (Cc) + * - U_FORMAT_CHAR (Cf) + * - U_LINE_SEPARATOR (Zl) + * - U_PARAGRAPH_SEPARATOR (Zp) + * + * This is a C/POSIX migration function. + * See the comments about C/POSIX character classification functions in the + * documentation at the top of this header file. + * + * @param c the code point to be tested + * @return true if the code point is a control character + * + * @see UCHAR_DEFAULT_IGNORABLE_CODE_POINT + * @see u_isprint + * @stable ICU 2.0 + */ +U_CAPI UBool U_EXPORT2 +u_iscntrl(UChar32 c); + +/** + * Determines whether the specified code point is an ISO control code. + * True for U+0000..U+001f and U+007f..U+009f (general category "Cc"). + * + * Same as java.lang.Character.isISOControl(). + * + * @param c the code point to be tested + * @return true if the code point is an ISO control code + * + * @see u_iscntrl + * @stable ICU 2.6 + */ +U_CAPI UBool U_EXPORT2 +u_isISOControl(UChar32 c); + +/** + * Determines whether the specified code point is a printable character. + * True for general categories other than "C" (controls). + * + * This is a C/POSIX migration function. + * See the comments about C/POSIX character classification functions in the + * documentation at the top of this header file. + * + * @param c the code point to be tested + * @return true if the code point is a printable character + * + * @see UCHAR_DEFAULT_IGNORABLE_CODE_POINT + * @see u_iscntrl + * @stable ICU 2.0 + */ +U_CAPI UBool U_EXPORT2 +u_isprint(UChar32 c); + +/** + * Non-standard: Determines whether the specified code point is a base character. + * True for general categories "L" (letters), "N" (numbers), + * "Mc" (spacing combining marks), and "Me" (enclosing marks). + * + * Note that this is different from the Unicode Standard definition in + * chapter 3.6, conformance clause D51 “Base character”, + * which defines base characters as the code points with general categories + * Letter (L), Number (N), Punctuation (P), Symbol (S), or Space Separator (Zs). + * + * @param c the code point to be tested + * @return true if the code point is a base character according to this function + * + * @see u_isalpha + * @see u_isdigit + * @stable ICU 2.0 + */ +U_CAPI UBool U_EXPORT2 +u_isbase(UChar32 c); + +/** + * Returns the bidirectional category value for the code point, + * which is used in the Unicode bidirectional algorithm + * (UAX #9 http://www.unicode.org/reports/tr9/). + * Note that some unassigned code points have bidi values + * of R or AL because they are in blocks that are reserved + * for Right-To-Left scripts. + * + * Same as java.lang.Character.getDirectionality() + * + * @param c the code point to be tested + * @return the bidirectional category (UCharDirection) value + * + * @see UCharDirection + * @stable ICU 2.0 + */ +U_CAPI UCharDirection U_EXPORT2 +u_charDirection(UChar32 c); + +/** + * Determines whether the code point has the Bidi_Mirrored property. + * This property is set for characters that are commonly used in + * Right-To-Left contexts and need to be displayed with a "mirrored" + * glyph. + * + * Same as java.lang.Character.isMirrored(). + * Same as UCHAR_BIDI_MIRRORED + * + * @param c the code point to be tested + * @return true if the character has the Bidi_Mirrored property + * + * @see UCHAR_BIDI_MIRRORED + * @stable ICU 2.0 + */ +U_CAPI UBool U_EXPORT2 +u_isMirrored(UChar32 c); + +/** + * Maps the specified character to a "mirror-image" character. + * For characters with the Bidi_Mirrored property, implementations + * sometimes need a "poor man's" mapping to another Unicode + * character (code point) such that the default glyph may serve + * as the mirror-image of the default glyph of the specified + * character. This is useful for text conversion to and from + * codepages with visual order, and for displays without glyph + * selection capabilities. + * + * @param c the code point to be mapped + * @return another Unicode code point that may serve as a mirror-image + * substitute, or c itself if there is no such mapping or c + * does not have the Bidi_Mirrored property + * + * @see UCHAR_BIDI_MIRRORED + * @see u_isMirrored + * @stable ICU 2.0 + */ +U_CAPI UChar32 U_EXPORT2 +u_charMirror(UChar32 c); + +/** + * Maps the specified character to its paired bracket character. + * For Bidi_Paired_Bracket_Type!=None, this is the same as u_charMirror(). + * Otherwise c itself is returned. + * See http://www.unicode.org/reports/tr9/ + * + * @param c the code point to be mapped + * @return the paired bracket code point, + * or c itself if there is no such mapping + * (Bidi_Paired_Bracket_Type=None) + * + * @see UCHAR_BIDI_PAIRED_BRACKET + * @see UCHAR_BIDI_PAIRED_BRACKET_TYPE + * @see u_charMirror + * @stable ICU 52 + */ +U_CAPI UChar32 U_EXPORT2 +u_getBidiPairedBracket(UChar32 c); + +/** + * Returns the general category value for the code point. + * + * Same as java.lang.Character.getType(). + * + * @param c the code point to be tested + * @return the general category (UCharCategory) value + * + * @see UCharCategory + * @stable ICU 2.0 + */ +U_CAPI int8_t U_EXPORT2 +u_charType(UChar32 c); + +/** + * Get a single-bit bit set for the general category of a character. + * This bit set can be compared bitwise with U_GC_SM_MASK, U_GC_L_MASK, etc. + * Same as U_MASK(u_charType(c)). + * + * @param c the code point to be tested + * @return a single-bit mask corresponding to the general category (UCharCategory) value + * + * @see u_charType + * @see UCharCategory + * @see U_GC_CN_MASK + * @stable ICU 2.1 + */ +#define U_GET_GC_MASK(c) U_MASK(u_charType(c)) + +/** + * Callback from u_enumCharTypes(), is called for each contiguous range + * of code points c (where start<=cnameChoice, the character name written + * into the buffer is the "modern" name or the name that was defined + * in Unicode version 1.0. + * The name contains only "invariant" characters + * like A-Z, 0-9, space, and '-'. + * Unicode 1.0 names are only retrieved if they are different from the modern + * names and if the data file contains the data for them. gennames may or may + * not be called with a command line option to include 1.0 names in unames.dat. + * + * @param code The character (code point) for which to get the name. + * It must be 0<=code<=0x10ffff. + * @param nameChoice Selector for which name to get. + * @param buffer Destination address for copying the name. + * The name will always be zero-terminated. + * If there is no name, then the buffer will be set to the empty string. + * @param bufferLength ==sizeof(buffer) + * @param pErrorCode Pointer to a UErrorCode variable; + * check for U_SUCCESS() after u_charName() + * returns. + * @return The length of the name, or 0 if there is no name for this character. + * If the bufferLength is less than or equal to the length, then the buffer + * contains the truncated name and the returned length indicates the full + * length of the name. + * The length does not include the zero-termination. + * + * @see UCharNameChoice + * @see u_charFromName + * @see u_enumCharNames + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_charName(UChar32 code, UCharNameChoice nameChoice, + char *buffer, int32_t bufferLength, + UErrorCode *pErrorCode); + +#ifndef U_HIDE_DEPRECATED_API +/** + * Returns an empty string. + * Used to return the ISO 10646 comment for a character. + * The Unicode ISO_Comment property is deprecated and has no values. + * + * @param c The character (code point) for which to get the ISO comment. + * It must be 0<=c<=0x10ffff. + * @param dest Destination address for copying the comment. + * The comment will be zero-terminated if possible. + * If there is no comment, then the buffer will be set to the empty string. + * @param destCapacity ==sizeof(dest) + * @param pErrorCode Pointer to a UErrorCode variable; + * check for U_SUCCESS() after u_getISOComment() + * returns. + * @return 0 + * + * @deprecated ICU 49 + */ +U_DEPRECATED int32_t U_EXPORT2 +u_getISOComment(UChar32 c, + char *dest, int32_t destCapacity, + UErrorCode *pErrorCode); +#endif /* U_HIDE_DEPRECATED_API */ + +/** + * Find a Unicode character by its name and return its code point value. + * The name is matched exactly and completely. + * If the name does not correspond to a code point, pErrorCode + * is set to U_INVALID_CHAR_FOUND. + * A Unicode 1.0 name is matched only if it differs from the modern name. + * Unicode names are all uppercase. Extended names are lowercase followed + * by an uppercase hexadecimal number, and within angle brackets. + * + * @param nameChoice Selector for which name to match. + * @param name The name to match. + * @param pErrorCode Pointer to a UErrorCode variable + * @return The Unicode value of the code point with the given name, + * or an undefined value if there is no such code point. + * + * @see UCharNameChoice + * @see u_charName + * @see u_enumCharNames + * @stable ICU 1.7 + */ +U_CAPI UChar32 U_EXPORT2 +u_charFromName(UCharNameChoice nameChoice, + const char *name, + UErrorCode *pErrorCode); + +/** + * Type of a callback function for u_enumCharNames() that gets called + * for each Unicode character with the code point value and + * the character name. + * If such a function returns false, then the enumeration is stopped. + * + * @param context The context pointer that was passed to u_enumCharNames(). + * @param code The Unicode code point for the character with this name. + * @param nameChoice Selector for which kind of names is enumerated. + * @param name The character's name, zero-terminated. + * @param length The length of the name. + * @return true if the enumeration should continue, false to stop it. + * + * @see UCharNameChoice + * @see u_enumCharNames + * @stable ICU 1.7 + */ +typedef UBool U_CALLCONV UEnumCharNamesFn(void *context, + UChar32 code, + UCharNameChoice nameChoice, + const char *name, + int32_t length); + +/** + * Enumerate all assigned Unicode characters between the start and limit + * code points (start inclusive, limit exclusive) and call a function + * for each, passing the code point value and the character name. + * For Unicode 1.0 names, only those are enumerated that differ from the + * modern names. + * + * @param start The first code point in the enumeration range. + * @param limit One more than the last code point in the enumeration range + * (the first one after the range). + * @param fn The function that is to be called for each character name. + * @param context An arbitrary pointer that is passed to the function. + * @param nameChoice Selector for which kind of names to enumerate. + * @param pErrorCode Pointer to a UErrorCode variable + * + * @see UCharNameChoice + * @see UEnumCharNamesFn + * @see u_charName + * @see u_charFromName + * @stable ICU 1.7 + */ +U_CAPI void U_EXPORT2 +u_enumCharNames(UChar32 start, UChar32 limit, + UEnumCharNamesFn *fn, + void *context, + UCharNameChoice nameChoice, + UErrorCode *pErrorCode); + +/** + * Return the Unicode name for a given property, as given in the + * Unicode database file PropertyAliases.txt. + * + * In addition, this function maps the property + * UCHAR_GENERAL_CATEGORY_MASK to the synthetic names "gcm" / + * "General_Category_Mask". These names are not in + * PropertyAliases.txt. + * + * @param property UProperty selector other than UCHAR_INVALID_CODE. + * If out of range, NULL is returned. + * + * @param nameChoice selector for which name to get. If out of range, + * NULL is returned. All properties have a long name. Most + * have a short name, but some do not. Unicode allows for + * additional names; if present these will be returned by + * U_LONG_PROPERTY_NAME + i, where i=1, 2,... + * + * @return a pointer to the name, or NULL if either the + * property or the nameChoice is out of range. If a given + * nameChoice returns NULL, then all larger values of + * nameChoice will return NULL, with one exception: if NULL is + * returned for U_SHORT_PROPERTY_NAME, then + * U_LONG_PROPERTY_NAME (and higher) may still return a + * non-NULL value. The returned pointer is valid until + * u_cleanup() is called. + * + * @see UProperty + * @see UPropertyNameChoice + * @stable ICU 2.4 + */ +U_CAPI const char* U_EXPORT2 +u_getPropertyName(UProperty property, + UPropertyNameChoice nameChoice); + +/** + * Return the UProperty enum for a given property name, as specified + * in the Unicode database file PropertyAliases.txt. Short, long, and + * any other variants are recognized. + * + * In addition, this function maps the synthetic names "gcm" / + * "General_Category_Mask" to the property + * UCHAR_GENERAL_CATEGORY_MASK. These names are not in + * PropertyAliases.txt. + * + * @param alias the property name to be matched. The name is compared + * using "loose matching" as described in PropertyAliases.txt. + * + * @return a UProperty enum, or UCHAR_INVALID_CODE if the given name + * does not match any property. + * + * @see UProperty + * @stable ICU 2.4 + */ +U_CAPI UProperty U_EXPORT2 +u_getPropertyEnum(const char* alias); + +/** + * Return the Unicode name for a given property value, as given in the + * Unicode database file PropertyValueAliases.txt. + * + * Note: Some of the names in PropertyValueAliases.txt can only be + * retrieved using UCHAR_GENERAL_CATEGORY_MASK, not + * UCHAR_GENERAL_CATEGORY. These include: "C" / "Other", "L" / + * "Letter", "LC" / "Cased_Letter", "M" / "Mark", "N" / "Number", "P" + * / "Punctuation", "S" / "Symbol", and "Z" / "Separator". + * + * @param property UProperty selector constant. + * Must be UCHAR_BINARY_START<=which2<=radix<=36 or if the + * value of c is not a valid digit in the specified + * radix, -1 is returned. A character is a valid digit + * if at least one of the following is true: + *

    + *
  • The character has a decimal digit value. + * Such characters have the general category "Nd" (decimal digit numbers) + * and a Numeric_Type of Decimal. + * In this case the value is the character's decimal digit value.
    • + *
  • The character is one of the uppercase Latin letters + * 'A' through 'Z'. + * In this case the value is c-'A'+10.
    • + *
  • The character is one of the lowercase Latin letters + * 'a' through 'z'. + * In this case the value is ch-'a'+10.
    • + *
  • Latin letters from both the ASCII range (0061..007A, 0041..005A) + * as well as from the Fullwidth ASCII range (FF41..FF5A, FF21..FF3A) + * are recognized.
    • + *
+ * + * Same as java.lang.Character.digit(). + * + * @param ch the code point to be tested. + * @param radix the radix. + * @return the numeric value represented by the character in the + * specified radix, + * or -1 if there is no value or if the value exceeds the radix. + * + * @see UCHAR_NUMERIC_TYPE + * @see u_forDigit + * @see u_charDigitValue + * @see u_isdigit + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_digit(UChar32 ch, int8_t radix); + +/** + * Determines the character representation for a specific digit in + * the specified radix. If the value of radix is not a + * valid radix, or the value of digit is not a valid + * digit in the specified radix, the null character + * (U+0000) is returned. + *

+ * The radix argument is valid if it is greater than or + * equal to 2 and less than or equal to 36. + * The digit argument is valid if + * 0 <= digit < radix. + *

+ * If the digit is less than 10, then + * '0' + digit is returned. Otherwise, the value + * 'a' + digit - 10 is returned. + * + * Same as java.lang.Character.forDigit(). + * + * @param digit the number to convert to a character. + * @param radix the radix. + * @return the char representation of the specified digit + * in the specified radix. + * + * @see u_digit + * @see u_charDigitValue + * @see u_isdigit + * @stable ICU 2.0 + */ +U_CAPI UChar32 U_EXPORT2 +u_forDigit(int32_t digit, int8_t radix); + +/** + * Get the "age" of the code point. + * The "age" is the Unicode version when the code point was first + * designated (as a non-character or for Private Use) + * or assigned a character. + * This can be useful to avoid emitting code points to receiving + * processes that do not accept newer characters. + * The data is from the UCD file DerivedAge.txt. + * + * @param c The code point. + * @param versionArray The Unicode version number array, to be filled in. + * + * @stable ICU 2.1 + */ +U_CAPI void U_EXPORT2 +u_charAge(UChar32 c, UVersionInfo versionArray); + +/** + * Gets the Unicode version information. + * The version array is filled in with the version information + * for the Unicode standard that is currently used by ICU. + * For example, Unicode version 3.1.1 is represented as an array with + * the values { 3, 1, 1, 0 }. + * + * @param versionArray an output array that will be filled in with + * the Unicode version number + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +u_getUnicodeVersion(UVersionInfo versionArray); + +#if !UCONFIG_NO_NORMALIZATION +/** + * Get the FC_NFKC_Closure property string for a character. + * See Unicode Standard Annex #15 for details, search for "FC_NFKC_Closure" + * or for "FNC": http://www.unicode.org/reports/tr15/ + * + * @param c The character (code point) for which to get the FC_NFKC_Closure string. + * It must be 0<=c<=0x10ffff. + * @param dest Destination address for copying the string. + * The string will be zero-terminated if possible. + * If there is no FC_NFKC_Closure string, + * then the buffer will be set to the empty string. + * @param destCapacity ==sizeof(dest) + * @param pErrorCode Pointer to a UErrorCode variable. + * @return The length of the string, or 0 if there is no FC_NFKC_Closure string for this character. + * If the destCapacity is less than or equal to the length, then the buffer + * contains the truncated name and the returned length indicates the full + * length of the name. + * The length does not include the zero-termination. + * + * @stable ICU 2.2 + */ +U_CAPI int32_t U_EXPORT2 +u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode); + +#endif + + +U_CDECL_END + +#endif /*_UCHAR*/ +/*eof*/ diff --git a/intl/icu/source/common/unicode/ucharstrie.h b/intl/icu/source/common/unicode/ucharstrie.h new file mode 100644 index 0000000000..fa1b55616c --- /dev/null +++ b/intl/icu/source/common/unicode/ucharstrie.h @@ -0,0 +1,623 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2012, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: ucharstrie.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010nov14 +* created by: Markus W. Scherer +*/ + +#ifndef __UCHARSTRIE_H__ +#define __UCHARSTRIE_H__ + +/** + * \file + * \brief C++ API: Trie for mapping Unicode strings (or 16-bit-unit sequences) + * to integer values. + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/unistr.h" +#include "unicode/uobject.h" +#include "unicode/ustringtrie.h" + +U_NAMESPACE_BEGIN + +class Appendable; +class UCharsTrieBuilder; +class UVector32; + +/** + * Light-weight, non-const reader class for a UCharsTrie. + * Traverses a char16_t-serialized data structure with minimal state, + * for mapping strings (16-bit-unit sequences) to non-negative integer values. + * + * This class owns the serialized trie data only if it was constructed by + * the builder's build() method. + * The public constructor and the copy constructor only alias the data (only copy the pointer). + * There is no assignment operator. + * + * This class is not intended for public subclassing. + * @stable ICU 4.8 + */ +class U_COMMON_API UCharsTrie : public UMemory { +public: + /** + * Constructs a UCharsTrie reader instance. + * + * The trieUChars must contain a copy of a char16_t sequence from the UCharsTrieBuilder, + * starting with the first char16_t of that sequence. + * The UCharsTrie object will not read more char16_ts than + * the UCharsTrieBuilder generated in the corresponding build() call. + * + * The array is not copied/cloned and must not be modified while + * the UCharsTrie object is in use. + * + * @param trieUChars The char16_t array that contains the serialized trie. + * @stable ICU 4.8 + */ + UCharsTrie(ConstChar16Ptr trieUChars) + : ownedArray_(nullptr), uchars_(trieUChars), + pos_(uchars_), remainingMatchLength_(-1) {} + + /** + * Destructor. + * @stable ICU 4.8 + */ + ~UCharsTrie(); + + /** + * Copy constructor, copies the other trie reader object and its state, + * but not the char16_t array which will be shared. (Shallow copy.) + * @param other Another UCharsTrie object. + * @stable ICU 4.8 + */ + UCharsTrie(const UCharsTrie &other) + : ownedArray_(nullptr), uchars_(other.uchars_), + pos_(other.pos_), remainingMatchLength_(other.remainingMatchLength_) {} + + /** + * Resets this trie to its initial state. + * @return *this + * @stable ICU 4.8 + */ + UCharsTrie &reset() { + pos_=uchars_; + remainingMatchLength_=-1; + return *this; + } + + /** + * Returns the state of this trie as a 64-bit integer. + * The state value is never 0. + * + * @return opaque state value + * @see resetToState64 + * @stable ICU 65 + */ + uint64_t getState64() const { + return (static_cast(remainingMatchLength_ + 2) << kState64RemainingShift) | + (uint64_t)(pos_ - uchars_); + } + + /** + * Resets this trie to the saved state. + * Unlike resetToState(State), the 64-bit state value + * must be from getState64() from the same trie object or + * from one initialized the exact same way. + * Because of no validation, this method is faster. + * + * @param state The opaque trie state value from getState64(). + * @return *this + * @see getState64 + * @see resetToState + * @see reset + * @stable ICU 65 + */ + UCharsTrie &resetToState64(uint64_t state) { + remainingMatchLength_ = static_cast(state >> kState64RemainingShift) - 2; + pos_ = uchars_ + (state & kState64PosMask); + return *this; + } + + /** + * UCharsTrie state object, for saving a trie's current state + * and resetting the trie back to this state later. + * @stable ICU 4.8 + */ + class State : public UMemory { + public: + /** + * Constructs an empty State. + * @stable ICU 4.8 + */ + State() { uchars=nullptr; } + private: + friend class UCharsTrie; + + const char16_t *uchars; + const char16_t *pos; + int32_t remainingMatchLength; + }; + + /** + * Saves the state of this trie. + * @param state The State object to hold the trie's state. + * @return *this + * @see resetToState + * @stable ICU 4.8 + */ + const UCharsTrie &saveState(State &state) const { + state.uchars=uchars_; + state.pos=pos_; + state.remainingMatchLength=remainingMatchLength_; + return *this; + } + + /** + * Resets this trie to the saved state. + * If the state object contains no state, or the state of a different trie, + * then this trie remains unchanged. + * @param state The State object which holds a saved trie state. + * @return *this + * @see saveState + * @see reset + * @stable ICU 4.8 + */ + UCharsTrie &resetToState(const State &state) { + if(uchars_==state.uchars && uchars_!=nullptr) { + pos_=state.pos; + remainingMatchLength_=state.remainingMatchLength; + } + return *this; + } + + /** + * Determines whether the string so far matches, whether it has a value, + * and whether another input char16_t can continue a matching string. + * @return The match/value Result. + * @stable ICU 4.8 + */ + UStringTrieResult current() const; + + /** + * Traverses the trie from the initial state for this input char16_t. + * Equivalent to reset().next(uchar). + * @param uchar Input char value. Values below 0 and above 0xffff will never match. + * @return The match/value Result. + * @stable ICU 4.8 + */ + inline UStringTrieResult first(int32_t uchar) { + remainingMatchLength_=-1; + return nextImpl(uchars_, uchar); + } + + /** + * Traverses the trie from the initial state for the + * one or two UTF-16 code units for this input code point. + * Equivalent to reset().nextForCodePoint(cp). + * @param cp A Unicode code point 0..0x10ffff. + * @return The match/value Result. + * @stable ICU 4.8 + */ + UStringTrieResult firstForCodePoint(UChar32 cp); + + /** + * Traverses the trie from the current state for this input char16_t. + * @param uchar Input char value. Values below 0 and above 0xffff will never match. + * @return The match/value Result. + * @stable ICU 4.8 + */ + UStringTrieResult next(int32_t uchar); + + /** + * Traverses the trie from the current state for the + * one or two UTF-16 code units for this input code point. + * @param cp A Unicode code point 0..0x10ffff. + * @return The match/value Result. + * @stable ICU 4.8 + */ + UStringTrieResult nextForCodePoint(UChar32 cp); + + /** + * Traverses the trie from the current state for this string. + * Equivalent to + * \code + * Result result=current(); + * for(each c in s) + * if(!USTRINGTRIE_HAS_NEXT(result)) return USTRINGTRIE_NO_MATCH; + * result=next(c); + * return result; + * \endcode + * @param s A string. Can be nullptr if length is 0. + * @param length The length of the string. Can be -1 if NUL-terminated. + * @return The match/value Result. + * @stable ICU 4.8 + */ + UStringTrieResult next(ConstChar16Ptr s, int32_t length); + + /** + * Returns a matching string's value if called immediately after + * current()/first()/next() returned USTRINGTRIE_INTERMEDIATE_VALUE or USTRINGTRIE_FINAL_VALUE. + * getValue() can be called multiple times. + * + * Do not call getValue() after USTRINGTRIE_NO_MATCH or USTRINGTRIE_NO_VALUE! + * @return The value for the string so far. + * @stable ICU 4.8 + */ + inline int32_t getValue() const { + const char16_t *pos=pos_; + int32_t leadUnit=*pos++; + // U_ASSERT(leadUnit>=kMinValueLead); + return leadUnit&kValueIsFinal ? + readValue(pos, leadUnit&0x7fff) : readNodeValue(pos, leadUnit); + } + + /** + * Determines whether all strings reachable from the current state + * map to the same value. + * @param uniqueValue Receives the unique value, if this function returns true. + * (output-only) + * @return true if all strings reachable from the current state + * map to the same value. + * @stable ICU 4.8 + */ + inline UBool hasUniqueValue(int32_t &uniqueValue) const { + const char16_t *pos=pos_; + // Skip the rest of a pending linear-match node. + return pos!=nullptr && findUniqueValue(pos+remainingMatchLength_+1, false, uniqueValue); + } + + /** + * Finds each char16_t which continues the string from the current state. + * That is, each char16_t c for which it would be next(c)!=USTRINGTRIE_NO_MATCH now. + * @param out Each next char16_t is appended to this object. + * @return the number of char16_ts which continue the string from here + * @stable ICU 4.8 + */ + int32_t getNextUChars(Appendable &out) const; + + /** + * Iterator for all of the (string, value) pairs in a UCharsTrie. + * @stable ICU 4.8 + */ + class U_COMMON_API Iterator : public UMemory { + public: + /** + * Iterates from the root of a char16_t-serialized UCharsTrie. + * @param trieUChars The trie char16_ts. + * @param maxStringLength If 0, the iterator returns full strings. + * Otherwise, the iterator returns strings with this maximum length. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @stable ICU 4.8 + */ + Iterator(ConstChar16Ptr trieUChars, int32_t maxStringLength, UErrorCode &errorCode); + + /** + * Iterates from the current state of the specified UCharsTrie. + * @param trie The trie whose state will be copied for iteration. + * @param maxStringLength If 0, the iterator returns full strings. + * Otherwise, the iterator returns strings with this maximum length. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @stable ICU 4.8 + */ + Iterator(const UCharsTrie &trie, int32_t maxStringLength, UErrorCode &errorCode); + + /** + * Destructor. + * @stable ICU 4.8 + */ + ~Iterator(); + + /** + * Resets this iterator to its initial state. + * @return *this + * @stable ICU 4.8 + */ + Iterator &reset(); + + /** + * @return true if there are more elements. + * @stable ICU 4.8 + */ + UBool hasNext() const; + + /** + * Finds the next (string, value) pair if there is one. + * + * If the string is truncated to the maximum length and does not + * have a real value, then the value is set to -1. + * In this case, this "not a real value" is indistinguishable from + * a real value of -1. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return true if there is another element. + * @stable ICU 4.8 + */ + UBool next(UErrorCode &errorCode); + + /** + * @return The string for the last successful next(). + * @stable ICU 4.8 + */ + const UnicodeString &getString() const { return str_; } + /** + * @return The value for the last successful next(). + * @stable ICU 4.8 + */ + int32_t getValue() const { return value_; } + + private: + UBool truncateAndStop() { + pos_=nullptr; + value_=-1; // no real value for str + return true; + } + + const char16_t *branchNext(const char16_t *pos, int32_t length, UErrorCode &errorCode); + + const char16_t *uchars_; + const char16_t *pos_; + const char16_t *initialPos_; + int32_t remainingMatchLength_; + int32_t initialRemainingMatchLength_; + UBool skipValue_; // Skip intermediate value which was already delivered. + + UnicodeString str_; + int32_t maxLength_; + int32_t value_; + + // The stack stores pairs of integers for backtracking to another + // outbound edge of a branch node. + // The first integer is an offset from uchars_. + // The second integer has the str_.length() from before the node in bits 15..0, + // and the remaining branch length in bits 31..16. + // (We could store the remaining branch length minus 1 in bits 30..16 and not use the sign bit, + // but the code looks more confusing that way.) + UVector32 *stack_; + }; + +private: + friend class UCharsTrieBuilder; + + /** + * Constructs a UCharsTrie reader instance. + * Unlike the public constructor which just aliases an array, + * this constructor adopts the builder's array. + * This constructor is only called by the builder. + */ + UCharsTrie(char16_t *adoptUChars, const char16_t *trieUChars) + : ownedArray_(adoptUChars), uchars_(trieUChars), + pos_(uchars_), remainingMatchLength_(-1) {} + + // No assignment operator. + UCharsTrie &operator=(const UCharsTrie &other) = delete; + + inline void stop() { + pos_=nullptr; + } + + // Reads a compact 32-bit integer. + // pos is already after the leadUnit, and the lead unit has bit 15 reset. + static inline int32_t readValue(const char16_t *pos, int32_t leadUnit) { + int32_t value; + if(leadUnit=kMinTwoUnitValueLead) { + if(leadUnit>6)-1; + } else if(leadUnit=kMinTwoUnitNodeValueLead) { + if(leadUnit=kMinTwoUnitDeltaLead) { + if(delta==kThreeUnitDeltaLead) { + delta=(pos[0]<<16)|pos[1]; + pos+=2; + } else { + delta=((delta-kMinTwoUnitDeltaLead)<<16)|*pos++; + } + } + return pos+delta; + } + + static const char16_t *skipDelta(const char16_t *pos) { + int32_t delta=*pos++; + if(delta>=kMinTwoUnitDeltaLead) { + if(delta==kThreeUnitDeltaLead) { + pos+=2; + } else { + ++pos; + } + } + return pos; + } + + static inline UStringTrieResult valueResult(int32_t node) { + return (UStringTrieResult)(USTRINGTRIE_INTERMEDIATE_VALUE-(node>>15)); + } + + // Handles a branch node for both next(uchar) and next(string). + UStringTrieResult branchNext(const char16_t *pos, int32_t length, int32_t uchar); + + // Requires remainingLength_<0. + UStringTrieResult nextImpl(const char16_t *pos, int32_t uchar); + + // Helper functions for hasUniqueValue(). + // Recursively finds a unique value (or whether there is not a unique one) + // from a branch. + static const char16_t *findUniqueValueFromBranch(const char16_t *pos, int32_t length, + UBool haveUniqueValue, int32_t &uniqueValue); + // Recursively finds a unique value (or whether there is not a unique one) + // starting from a position on a node lead unit. + static UBool findUniqueValue(const char16_t *pos, UBool haveUniqueValue, int32_t &uniqueValue); + + // Helper functions for getNextUChars(). + // getNextUChars() when pos is on a branch node. + static void getNextBranchUChars(const char16_t *pos, int32_t length, Appendable &out); + + // UCharsTrie data structure + // + // The trie consists of a series of char16_t-serialized nodes for incremental + // Unicode string/char16_t sequence matching. (char16_t=16-bit unsigned integer) + // The root node is at the beginning of the trie data. + // + // Types of nodes are distinguished by their node lead unit ranges. + // After each node, except a final-value node, another node follows to + // encode match values or continue matching further units. + // + // Node types: + // - Final-value node: Stores a 32-bit integer in a compact, variable-length format. + // The value is for the string/char16_t sequence so far. + // - Match node, optionally with an intermediate value in a different compact format. + // The value, if present, is for the string/char16_t sequence so far. + // + // Aside from the value, which uses the node lead unit's high bits: + // + // - Linear-match node: Matches a number of units. + // - Branch node: Branches to other nodes according to the current input unit. + // The node unit is the length of the branch (number of units to select from) + // minus 1. It is followed by a sub-node: + // - If the length is at most kMaxBranchLinearSubNodeLength, then + // there are length-1 (key, value) pairs and then one more comparison unit. + // If one of the key units matches, then the value is either a final value for + // the string so far, or a "jump" delta to the next node. + // If the last unit matches, then matching continues with the next node. + // (Values have the same encoding as final-value nodes.) + // - If the length is greater than kMaxBranchLinearSubNodeLength, then + // there is one unit and one "jump" delta. + // If the input unit is less than the sub-node unit, then "jump" by delta to + // the next sub-node which will have a length of length/2. + // (The delta has its own compact encoding.) + // Otherwise, skip the "jump" delta to the next sub-node + // which will have a length of length-length/2. + + // Match-node lead unit values, after masking off intermediate-value bits: + + // 0000..002f: Branch node. If node!=0 then the length is node+1, otherwise + // the length is one more than the next unit. + + // For a branch sub-node with at most this many entries, we drop down + // to a linear search. + static const int32_t kMaxBranchLinearSubNodeLength=5; + + // 0030..003f: Linear-match node, match 1..16 units and continue reading the next node. + static const int32_t kMinLinearMatch=0x30; + static const int32_t kMaxLinearMatchLength=0x10; + + // Match-node lead unit bits 14..6 for the optional intermediate value. + // If these bits are 0, then there is no intermediate value. + // Otherwise, see the *NodeValue* constants below. + static const int32_t kMinValueLead=kMinLinearMatch+kMaxLinearMatchLength; // 0x0040 + static const int32_t kNodeTypeMask=kMinValueLead-1; // 0x003f + + // A final-value node has bit 15 set. + static const int32_t kValueIsFinal=0x8000; + + // Compact value: After testing and masking off bit 15, use the following thresholds. + static const int32_t kMaxOneUnitValue=0x3fff; + + static const int32_t kMinTwoUnitValueLead=kMaxOneUnitValue+1; // 0x4000 + static const int32_t kThreeUnitValueLead=0x7fff; + + static const int32_t kMaxTwoUnitValue=((kThreeUnitValueLead-kMinTwoUnitValueLead)<<16)-1; // 0x3ffeffff + + // Compact intermediate-value integer, lead unit shared with a branch or linear-match node. + static const int32_t kMaxOneUnitNodeValue=0xff; + static const int32_t kMinTwoUnitNodeValueLead=kMinValueLead+((kMaxOneUnitNodeValue+1)<<6); // 0x4040 + static const int32_t kThreeUnitNodeValueLead=0x7fc0; + + static const int32_t kMaxTwoUnitNodeValue= + ((kThreeUnitNodeValueLead-kMinTwoUnitNodeValueLead)<<10)-1; // 0xfdffff + + // Compact delta integers. + static const int32_t kMaxOneUnitDelta=0xfbff; + static const int32_t kMinTwoUnitDeltaLead=kMaxOneUnitDelta+1; // 0xfc00 + static const int32_t kThreeUnitDeltaLead=0xffff; + + static const int32_t kMaxTwoUnitDelta=((kThreeUnitDeltaLead-kMinTwoUnitDeltaLead)<<16)-1; // 0x03feffff + + // For getState64(): + // The remainingMatchLength_ is -1..14=(kMaxLinearMatchLength=0x10)-2 + // so we need at least 5 bits for that. + // We add 2 to store it as a positive value 1..16=kMaxLinearMatchLength. + static constexpr int32_t kState64RemainingShift = 59; + static constexpr uint64_t kState64PosMask = (UINT64_C(1) << kState64RemainingShift) - 1; + + char16_t *ownedArray_; + + // Fixed value referencing the UCharsTrie words. + const char16_t *uchars_; + + // Iterator variables. + + // Pointer to next trie unit to read. nullptr if no more matches. + const char16_t *pos_; + // Remaining length of a linear-match node, minus 1. Negative if not in such a node. + int32_t remainingMatchLength_; +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __UCHARSTRIE_H__ diff --git a/intl/icu/source/common/unicode/ucharstriebuilder.h b/intl/icu/source/common/unicode/ucharstriebuilder.h new file mode 100644 index 0000000000..5c8aa33ffb --- /dev/null +++ b/intl/icu/source/common/unicode/ucharstriebuilder.h @@ -0,0 +1,193 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2016, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: ucharstriebuilder.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010nov14 +* created by: Markus W. Scherer +*/ + +#ifndef __UCHARSTRIEBUILDER_H__ +#define __UCHARSTRIEBUILDER_H__ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/stringtriebuilder.h" +#include "unicode/ucharstrie.h" +#include "unicode/unistr.h" + +/** + * \file + * \brief C++ API: Builder for icu::UCharsTrie + */ + +U_NAMESPACE_BEGIN + +class UCharsTrieElement; + +/** + * Builder class for UCharsTrie. + * + * This class is not intended for public subclassing. + * @stable ICU 4.8 + */ +class U_COMMON_API UCharsTrieBuilder : public StringTrieBuilder { +public: + /** + * Constructs an empty builder. + * @param errorCode Standard ICU error code. + * @stable ICU 4.8 + */ + UCharsTrieBuilder(UErrorCode &errorCode); + + /** + * Destructor. + * @stable ICU 4.8 + */ + virtual ~UCharsTrieBuilder(); + + /** + * Adds a (string, value) pair. + * The string must be unique. + * The string contents will be copied; the builder does not keep + * a reference to the input UnicodeString or its buffer. + * @param s The input string. + * @param value The value associated with this string. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return *this + * @stable ICU 4.8 + */ + UCharsTrieBuilder &add(const UnicodeString &s, int32_t value, UErrorCode &errorCode); + + /** + * Builds a UCharsTrie for the add()ed data. + * Once built, no further data can be add()ed until clear() is called. + * + * A UCharsTrie cannot be empty. At least one (string, value) pair + * must have been add()ed. + * + * This method passes ownership of the builder's internal result array to the new trie object. + * Another call to any build() variant will re-serialize the trie. + * After clear() has been called, a new array will be used as well. + * @param buildOption Build option, see UStringTrieBuildOption. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return A new UCharsTrie for the add()ed data. + * @stable ICU 4.8 + */ + UCharsTrie *build(UStringTrieBuildOption buildOption, UErrorCode &errorCode); + + /** + * Builds a UCharsTrie for the add()ed data and char16_t-serializes it. + * Once built, no further data can be add()ed until clear() is called. + * + * A UCharsTrie cannot be empty. At least one (string, value) pair + * must have been add()ed. + * + * Multiple calls to buildUnicodeString() set the UnicodeStrings to the + * builder's same char16_t array, without rebuilding. + * If buildUnicodeString() is called after build(), the trie will be + * re-serialized into a new array (because build() passes on ownership). + * If build() is called after buildUnicodeString(), the trie object returned + * by build() will become the owner of the underlying data for the + * previously returned UnicodeString. + * After clear() has been called, a new array will be used as well. + * @param buildOption Build option, see UStringTrieBuildOption. + * @param result A UnicodeString which will be set to the char16_t-serialized + * UCharsTrie for the add()ed data. + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return result + * @stable ICU 4.8 + */ + UnicodeString &buildUnicodeString(UStringTrieBuildOption buildOption, UnicodeString &result, + UErrorCode &errorCode); + + /** + * Removes all (string, value) pairs. + * New data can then be add()ed and a new trie can be built. + * @return *this + * @stable ICU 4.8 + */ + UCharsTrieBuilder &clear() { + strings.remove(); + elementsLength=0; + ucharsLength=0; + return *this; + } + +private: + UCharsTrieBuilder(const UCharsTrieBuilder &other) = delete; // no copy constructor + UCharsTrieBuilder &operator=(const UCharsTrieBuilder &other) = delete; // no assignment operator + + void buildUChars(UStringTrieBuildOption buildOption, UErrorCode &errorCode); + + virtual int32_t getElementStringLength(int32_t i) const override; + virtual char16_t getElementUnit(int32_t i, int32_t unitIndex) const override; + virtual int32_t getElementValue(int32_t i) const override; + + virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t unitIndex) const override; + + virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t unitIndex) const override; + virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t unitIndex, int32_t count) const override; + virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t unitIndex, char16_t unit) const override; + + virtual UBool matchNodesCanHaveValues() const override { return true; } + + virtual int32_t getMaxBranchLinearSubNodeLength() const override { return UCharsTrie::kMaxBranchLinearSubNodeLength; } + virtual int32_t getMinLinearMatch() const override { return UCharsTrie::kMinLinearMatch; } + virtual int32_t getMaxLinearMatchLength() const override { return UCharsTrie::kMaxLinearMatchLength; } + + class UCTLinearMatchNode : public LinearMatchNode { + public: + UCTLinearMatchNode(const char16_t *units, int32_t len, Node *nextNode); + virtual bool operator==(const Node &other) const override; + virtual void write(StringTrieBuilder &builder) override; + private: + const char16_t *s; + }; + + virtual Node *createLinearMatchNode(int32_t i, int32_t unitIndex, int32_t length, + Node *nextNode) const override; + + UBool ensureCapacity(int32_t length); + virtual int32_t write(int32_t unit) override; + int32_t write(const char16_t *s, int32_t length); + virtual int32_t writeElementUnits(int32_t i, int32_t unitIndex, int32_t length) override; + virtual int32_t writeValueAndFinal(int32_t i, UBool isFinal) override; + virtual int32_t writeValueAndType(UBool hasValue, int32_t value, int32_t node) override; + virtual int32_t writeDeltaTo(int32_t jumpTarget) override; + + UnicodeString strings; + UCharsTrieElement *elements; + int32_t elementsCapacity; + int32_t elementsLength; + + // char16_t serialization of the trie. + // Grows from the back: ucharsLength measures from the end of the buffer! + char16_t *uchars; + int32_t ucharsCapacity; + int32_t ucharsLength; +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif // __UCHARSTRIEBUILDER_H__ diff --git a/intl/icu/source/common/unicode/uchriter.h b/intl/icu/source/common/unicode/uchriter.h new file mode 100644 index 0000000000..9fae5e7de0 --- /dev/null +++ b/intl/icu/source/common/unicode/uchriter.h @@ -0,0 +1,393 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1998-2005, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ + +#ifndef UCHRITER_H +#define UCHRITER_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/chariter.h" + +/** + * \file + * \brief C++ API: char16_t Character Iterator + */ + +U_NAMESPACE_BEGIN + +/** + * A concrete subclass of CharacterIterator that iterates over the + * characters (code units or code points) in a char16_t array. + * It's possible not only to create an + * iterator that iterates over an entire char16_t array, but also to + * create one that iterates over only a subrange of a char16_t array + * (iterators over different subranges of the same char16_t array don't + * compare equal). + * @see CharacterIterator + * @see ForwardCharacterIterator + * @stable ICU 2.0 + */ +class U_COMMON_API UCharCharacterIterator : public CharacterIterator { +public: + /** + * Create an iterator over the char16_t array referred to by "textPtr". + * The iteration range is 0 to length-1. + * text is only aliased, not adopted (the + * destructor will not delete it). + * @param textPtr The char16_t array to be iterated over + * @param length The length of the char16_t array + * @stable ICU 2.0 + */ + UCharCharacterIterator(ConstChar16Ptr textPtr, int32_t length); + + /** + * Create an iterator over the char16_t array referred to by "textPtr". + * The iteration range is 0 to length-1. + * text is only aliased, not adopted (the + * destructor will not delete it). + * The starting + * position is specified by "position". If "position" is outside the valid + * iteration range, the behavior of this object is undefined. + * @param textPtr The char16_t array to be iterated over + * @param length The length of the char16_t array + * @param position The starting position of the iteration + * @stable ICU 2.0 + */ + UCharCharacterIterator(ConstChar16Ptr textPtr, int32_t length, + int32_t position); + + /** + * Create an iterator over the char16_t array referred to by "textPtr". + * The iteration range is 0 to end-1. + * text is only aliased, not adopted (the + * destructor will not delete it). + * The starting + * position is specified by "position". If begin and end do not + * form a valid iteration range or "position" is outside the valid + * iteration range, the behavior of this object is undefined. + * @param textPtr The char16_t array to be iterated over + * @param length The length of the char16_t array + * @param textBegin The begin position of the iteration range + * @param textEnd The end position of the iteration range + * @param position The starting position of the iteration + * @stable ICU 2.0 + */ + UCharCharacterIterator(ConstChar16Ptr textPtr, int32_t length, + int32_t textBegin, + int32_t textEnd, + int32_t position); + + /** + * Copy constructor. The new iterator iterates over the same range + * of the same string as "that", and its initial position is the + * same as "that"'s current position. + * @param that The UCharCharacterIterator to be copied + * @stable ICU 2.0 + */ + UCharCharacterIterator(const UCharCharacterIterator& that); + + /** + * Destructor. + * @stable ICU 2.0 + */ + virtual ~UCharCharacterIterator(); + + /** + * Assignment operator. *this is altered to iterate over the sane + * range of the same string as "that", and refers to the same + * character within that string as "that" does. + * @param that The object to be copied + * @return the newly created object + * @stable ICU 2.0 + */ + UCharCharacterIterator& + operator=(const UCharCharacterIterator& that); + + /** + * Returns true if the iterators iterate over the same range of the + * same string and are pointing at the same character. + * @param that The ForwardCharacterIterator used to be compared for equality + * @return true if the iterators iterate over the same range of the + * same string and are pointing at the same character. + * @stable ICU 2.0 + */ + virtual bool operator==(const ForwardCharacterIterator& that) const override; + + /** + * Generates a hash code for this iterator. + * @return the hash code. + * @stable ICU 2.0 + */ + virtual int32_t hashCode(void) const override; + + /** + * Returns a new UCharCharacterIterator referring to the same + * character in the same range of the same string as this one. The + * caller must delete the new iterator. + * @return the CharacterIterator newly created + * @stable ICU 2.0 + */ + virtual UCharCharacterIterator* clone() const override; + + /** + * Sets the iterator to refer to the first code unit in its + * iteration range, and returns that code unit. + * This can be used to begin an iteration with next(). + * @return the first code unit in its iteration range. + * @stable ICU 2.0 + */ + virtual char16_t first(void) override; + + /** + * Sets the iterator to refer to the first code unit in its + * iteration range, returns that code unit, and moves the position + * to the second code unit. This is an alternative to setToStart() + * for forward iteration with nextPostInc(). + * @return the first code unit in its iteration range + * @stable ICU 2.0 + */ + virtual char16_t firstPostInc(void) override; + + /** + * Sets the iterator to refer to the first code point in its + * iteration range, and returns that code unit, + * This can be used to begin an iteration with next32(). + * Note that an iteration with next32PostInc(), beginning with, + * e.g., setToStart() or firstPostInc(), is more efficient. + * @return the first code point in its iteration range + * @stable ICU 2.0 + */ + virtual UChar32 first32(void) override; + + /** + * Sets the iterator to refer to the first code point in its + * iteration range, returns that code point, and moves the position + * to the second code point. This is an alternative to setToStart() + * for forward iteration with next32PostInc(). + * @return the first code point in its iteration range. + * @stable ICU 2.0 + */ + virtual UChar32 first32PostInc(void) override; + + /** + * Sets the iterator to refer to the last code unit in its + * iteration range, and returns that code unit. + * This can be used to begin an iteration with previous(). + * @return the last code unit in its iteration range. + * @stable ICU 2.0 + */ + virtual char16_t last(void) override; + + /** + * Sets the iterator to refer to the last code point in its + * iteration range, and returns that code unit. + * This can be used to begin an iteration with previous32(). + * @return the last code point in its iteration range. + * @stable ICU 2.0 + */ + virtual UChar32 last32(void) override; + + /** + * Sets the iterator to refer to the "position"-th code unit + * in the text-storage object the iterator refers to, and + * returns that code unit. + * @param position the position within the text-storage object + * @return the code unit + * @stable ICU 2.0 + */ + virtual char16_t setIndex(int32_t position) override; + + /** + * Sets the iterator to refer to the beginning of the code point + * that contains the "position"-th code unit + * in the text-storage object the iterator refers to, and + * returns that code point. + * The current position is adjusted to the beginning of the code point + * (its first code unit). + * @param position the position within the text-storage object + * @return the code unit + * @stable ICU 2.0 + */ + virtual UChar32 setIndex32(int32_t position) override; + + /** + * Returns the code unit the iterator currently refers to. + * @return the code unit the iterator currently refers to. + * @stable ICU 2.0 + */ + virtual char16_t current(void) const override; + + /** + * Returns the code point the iterator currently refers to. + * @return the code point the iterator currently refers to. + * @stable ICU 2.0 + */ + virtual UChar32 current32(void) const override; + + /** + * Advances to the next code unit in the iteration range (toward + * endIndex()), and returns that code unit. If there are no more + * code units to return, returns DONE. + * @return the next code unit in the iteration range. + * @stable ICU 2.0 + */ + virtual char16_t next(void) override; + + /** + * Gets the current code unit for returning and advances to the next code unit + * in the iteration range + * (toward endIndex()). If there are + * no more code units to return, returns DONE. + * @return the current code unit. + * @stable ICU 2.0 + */ + virtual char16_t nextPostInc(void) override; + + /** + * Advances to the next code point in the iteration range (toward + * endIndex()), and returns that code point. If there are no more + * code points to return, returns DONE. + * Note that iteration with "pre-increment" semantics is less + * efficient than iteration with "post-increment" semantics + * that is provided by next32PostInc(). + * @return the next code point in the iteration range. + * @stable ICU 2.0 + */ + virtual UChar32 next32(void) override; + + /** + * Gets the current code point for returning and advances to the next code point + * in the iteration range + * (toward endIndex()). If there are + * no more code points to return, returns DONE. + * @return the current point. + * @stable ICU 2.0 + */ + virtual UChar32 next32PostInc(void) override; + + /** + * Returns false if there are no more code units or code points + * at or after the current position in the iteration range. + * This is used with nextPostInc() or next32PostInc() in forward + * iteration. + * @return false if there are no more code units or code points + * at or after the current position in the iteration range. + * @stable ICU 2.0 + */ + virtual UBool hasNext() override; + + /** + * Advances to the previous code unit in the iteration range (toward + * startIndex()), and returns that code unit. If there are no more + * code units to return, returns DONE. + * @return the previous code unit in the iteration range. + * @stable ICU 2.0 + */ + virtual char16_t previous(void) override; + + /** + * Advances to the previous code point in the iteration range (toward + * startIndex()), and returns that code point. If there are no more + * code points to return, returns DONE. + * @return the previous code point in the iteration range. + * @stable ICU 2.0 + */ + virtual UChar32 previous32(void) override; + + /** + * Returns false if there are no more code units or code points + * before the current position in the iteration range. + * This is used with previous() or previous32() in backward + * iteration. + * @return false if there are no more code units or code points + * before the current position in the iteration range. + * @stable ICU 2.0 + */ + virtual UBool hasPrevious() override; + + /** + * Moves the current position relative to the start or end of the + * iteration range, or relative to the current position itself. + * The movement is expressed in numbers of code units forward + * or backward by specifying a positive or negative delta. + * @param delta the position relative to origin. A positive delta means forward; + * a negative delta means backward. + * @param origin Origin enumeration {kStart, kCurrent, kEnd} + * @return the new position + * @stable ICU 2.0 + */ + virtual int32_t move(int32_t delta, EOrigin origin) override; + + /** + * Moves the current position relative to the start or end of the + * iteration range, or relative to the current position itself. + * The movement is expressed in numbers of code points forward + * or backward by specifying a positive or negative delta. + * @param delta the position relative to origin. A positive delta means forward; + * a negative delta means backward. + * @param origin Origin enumeration {kStart, kCurrent, kEnd} + * @return the new position + * @stable ICU 2.0 + */ +#ifdef move32 + // One of the system headers right now is sometimes defining a conflicting macro we don't use +#undef move32 +#endif + virtual int32_t move32(int32_t delta, EOrigin origin) override; + + /** + * Sets the iterator to iterate over a new range of text + * @stable ICU 2.0 + */ + void setText(ConstChar16Ptr newText, int32_t newTextLength); + + /** + * Copies the char16_t array under iteration into the UnicodeString + * referred to by "result". Even if this iterator iterates across + * only a part of this string, the whole string is copied. + * @param result Receives a copy of the text under iteration. + * @stable ICU 2.0 + */ + virtual void getText(UnicodeString& result) override; + + /** + * Return a class ID for this class (not really public) + * @return a class ID for this class + * @stable ICU 2.0 + */ + static UClassID U_EXPORT2 getStaticClassID(void); + + /** + * Return a class ID for this object (not really public) + * @return a class ID for this object. + * @stable ICU 2.0 + */ + virtual UClassID getDynamicClassID(void) const override; + +protected: + /** + * Protected constructor + * @stable ICU 2.0 + */ + UCharCharacterIterator(); + /** + * Protected member text + * @stable ICU 2.0 + */ + const char16_t* text; + +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/uclean.h b/intl/icu/source/common/unicode/uclean.h new file mode 100644 index 0000000000..f5b0aa088a --- /dev/null +++ b/intl/icu/source/common/unicode/uclean.h @@ -0,0 +1,262 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 2001-2014, International Business Machines +* Corporation and others. All Rights Reserved. +****************************************************************************** +* file name: uclean.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2001July05 +* created by: George Rhoten +*/ + +#ifndef __UCLEAN_H__ +#define __UCLEAN_H__ + +#include "unicode/utypes.h" +/** + * \file + * \brief C API: Initialize and clean up ICU + */ + +/** + * Initialize ICU. + * + * Use of this function is optional. It is OK to simply use ICU + * services and functions without first having initialized + * ICU by calling u_init(). + * + * u_init() will attempt to load some part of ICU's data, and is + * useful as a test for configuration or installation problems that + * leave the ICU data inaccessible. A successful invocation of u_init() + * does not, however, guarantee that all ICU data is accessible. + * + * Multiple calls to u_init() cause no harm, aside from the small amount + * of time required. + * + * In old versions of ICU, u_init() was required in multi-threaded applications + * to ensure the thread safety of ICU. u_init() is no longer needed for this purpose. + * + * @param status An ICU UErrorCode parameter. It must not be NULL. + * An Error will be returned if some required part of ICU data can not + * be loaded or initialized. + * The function returns immediately if the input error code indicates a + * failure, as usual. + * + * @stable ICU 2.6 + */ +U_CAPI void U_EXPORT2 +u_init(UErrorCode *status); + +#ifndef U_HIDE_SYSTEM_API +/** + * Clean up the system resources, such as allocated memory or open files, + * used in all ICU libraries. This will free/delete all memory owned by the + * ICU libraries, and return them to their original load state. All open ICU + * items (collators, resource bundles, converters, etc.) must be closed before + * calling this function, otherwise ICU may not free its allocated memory + * (e.g. close your converters and resource bundles before calling this + * function). Generally, this function should be called once just before + * an application exits. For applications that dynamically load and unload + * the ICU libraries (relatively uncommon), u_cleanup() should be called + * just before the library unload. + *

+ * u_cleanup() also clears any ICU heap functions, mutex functions or + * trace functions that may have been set for the process. + * This has the effect of restoring ICU to its initial condition, before + * any of these override functions were installed. Refer to + * u_setMemoryFunctions(), u_setMutexFunctions and + * utrace_setFunctions(). If ICU is to be reinitialized after + * calling u_cleanup(), these runtime override functions will need to + * be set up again if they are still required. + *

+ * u_cleanup() is not thread safe. All other threads should stop using ICU + * before calling this function. + *

+ * Any open ICU items will be left in an undefined state by u_cleanup(), + * and any subsequent attempt to use such an item will give unpredictable + * results. + *

+ * After calling u_cleanup(), an application may continue to use ICU by + * calling u_init(). An application must invoke u_init() first from one single + * thread before allowing other threads call u_init(). All threads existing + * at the time of the first thread's call to u_init() must also call + * u_init() themselves before continuing with other ICU operations. + *

+ * The use of u_cleanup() just before an application terminates is optional, + * but it should be called only once for performance reasons. The primary + * benefit is to eliminate reports of memory or resource leaks originating + * in ICU code from the results generated by heap analysis tools. + *

+ * Use this function with great care! + *

+ * + * @stable ICU 2.0 + * @system + */ +U_CAPI void U_EXPORT2 +u_cleanup(void); + +U_CDECL_BEGIN +/** + * Pointer type for a user supplied memory allocation function. + * @param context user supplied value, obtained from u_setMemoryFunctions(). + * @param size The number of bytes to be allocated + * @return Pointer to the newly allocated memory, or NULL if the allocation failed. + * @stable ICU 2.8 + * @system + */ +typedef void *U_CALLCONV UMemAllocFn(const void *context, size_t size); +/** + * Pointer type for a user supplied memory re-allocation function. + * @param context user supplied value, obtained from u_setMemoryFunctions(). + * @param mem Pointer to the memory block to be resized. + * @param size The new size for the block. + * @return Pointer to the newly allocated memory, or NULL if the allocation failed. + * @stable ICU 2.8 + * @system + */ +typedef void *U_CALLCONV UMemReallocFn(const void *context, void *mem, size_t size); +/** + * Pointer type for a user supplied memory free function. Behavior should be + * similar the standard C library free(). + * @param context user supplied value, obtained from u_setMemoryFunctions(). + * @param mem Pointer to the memory block to be freed. + * @return Pointer to the resized memory block, or NULL if the resizing failed. + * @stable ICU 2.8 + * @system + */ +typedef void U_CALLCONV UMemFreeFn (const void *context, void *mem); + +/** + * Set the functions that ICU will use for memory allocation. + * Use of this function is optional; by default (without this function), ICU will + * use the standard C library malloc() and free() functions. + * This function can only be used when ICU is in an initial, unused state, before + * u_init() has been called. + * @param context This pointer value will be saved, and then (later) passed as + * a parameter to the memory functions each time they + * are called. + * @param a Pointer to a user-supplied malloc function. + * @param r Pointer to a user-supplied realloc function. + * @param f Pointer to a user-supplied free function. + * @param status Receives error values. + * @stable ICU 2.8 + * @system + */ +U_CAPI void U_EXPORT2 +u_setMemoryFunctions(const void *context, UMemAllocFn * U_CALLCONV_FPTR a, UMemReallocFn * U_CALLCONV_FPTR r, UMemFreeFn * U_CALLCONV_FPTR f, + UErrorCode *status); + +U_CDECL_END + +#ifndef U_HIDE_DEPRECATED_API +/********************************************************************************* + * + * Deprecated Functions + * + * The following functions for user supplied mutexes are no longer supported. + * Any attempt to use them will return a U_UNSUPPORTED_ERROR. + * + **********************************************************************************/ + +/** + * An opaque pointer type that represents an ICU mutex. + * For user-implemented mutexes, the value will typically point to a + * struct or object that implements the mutex. + * @deprecated ICU 52. This type is no longer supported. + * @system + */ +typedef void *UMTX; + +U_CDECL_BEGIN +/** + * Function Pointer type for a user supplied mutex initialization function. + * The user-supplied function will be called by ICU whenever ICU needs to create a + * new mutex. The function implementation should create a mutex, and store a pointer + * to something that uniquely identifies the mutex into the UMTX that is supplied + * as a parameter. + * @param context user supplied value, obtained from u_setMutexFunctions(). + * @param mutex Receives a pointer that identifies the new mutex. + * The mutex init function must set the UMTX to a non-null value. + * Subsequent calls by ICU to lock, unlock, or destroy a mutex will + * identify the mutex by the UMTX value. + * @param status Error status. Report errors back to ICU by setting this variable + * with an error code. + * @deprecated ICU 52. This function is no longer supported. + * @system + */ +typedef void U_CALLCONV UMtxInitFn (const void *context, UMTX *mutex, UErrorCode* status); + + +/** + * Function Pointer type for a user supplied mutex functions. + * One of the user-supplied functions with this signature will be called by ICU + * whenever ICU needs to lock, unlock, or destroy a mutex. + * @param context user supplied value, obtained from u_setMutexFunctions(). + * @param mutex specify the mutex on which to operate. + * @deprecated ICU 52. This function is no longer supported. + * @system + */ +typedef void U_CALLCONV UMtxFn (const void *context, UMTX *mutex); +U_CDECL_END + +/** + * Set the functions that ICU will use for mutex operations + * Use of this function is optional; by default (without this function), ICU will + * directly access system functions for mutex operations + * This function can only be used when ICU is in an initial, unused state, before + * u_init() has been called. + * @param context This pointer value will be saved, and then (later) passed as + * a parameter to the user-supplied mutex functions each time they + * are called. + * @param init Pointer to a mutex initialization function. Must be non-null. + * @param destroy Pointer to the mutex destroy function. Must be non-null. + * @param lock pointer to the mutex lock function. Must be non-null. + * @param unlock Pointer to the mutex unlock function. Must be non-null. + * @param status Receives error values. + * @deprecated ICU 52. This function is no longer supported. + * @system + */ +U_DEPRECATED void U_EXPORT2 +u_setMutexFunctions(const void *context, UMtxInitFn *init, UMtxFn *destroy, UMtxFn *lock, UMtxFn *unlock, + UErrorCode *status); + + +/** + * Pointer type for a user supplied atomic increment or decrement function. + * @param context user supplied value, obtained from u_setAtomicIncDecFunctions(). + * @param p Pointer to a 32 bit int to be incremented or decremented + * @return The value of the variable after the inc or dec operation. + * @deprecated ICU 52. This function is no longer supported. + * @system + */ +typedef int32_t U_CALLCONV UMtxAtomicFn(const void *context, int32_t *p); + +/** + * Set the functions that ICU will use for atomic increment and decrement of int32_t values. + * Use of this function is optional; by default (without this function), ICU will + * use its own internal implementation of atomic increment/decrement. + * This function can only be used when ICU is in an initial, unused state, before + * u_init() has been called. + * @param context This pointer value will be saved, and then (later) passed as + * a parameter to the increment and decrement functions each time they + * are called. This function can only be called + * @param inc Pointer to a function to do an atomic increment operation. Must be non-null. + * @param dec Pointer to a function to do an atomic decrement operation. Must be non-null. + * @param status Receives error values. + * @deprecated ICU 52. This function is no longer supported. + * @system + */ +U_DEPRECATED void U_EXPORT2 +u_setAtomicIncDecFunctions(const void *context, UMtxAtomicFn *inc, UMtxAtomicFn *dec, + UErrorCode *status); + +#endif /* U_HIDE_DEPRECATED_API */ +#endif /* U_HIDE_SYSTEM_API */ + +#endif diff --git a/intl/icu/source/common/unicode/ucnv.h b/intl/icu/source/common/unicode/ucnv.h new file mode 100644 index 0000000000..20c173b662 --- /dev/null +++ b/intl/icu/source/common/unicode/ucnv.h @@ -0,0 +1,2056 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1999-2014, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** + * ucnv.h: + * External APIs for the ICU's codeset conversion library + * Bertrand A. Damiba + * + * Modification History: + * + * Date Name Description + * 04/04/99 helena Fixed internal header inclusion. + * 05/11/00 helena Added setFallback and usesFallback APIs. + * 06/29/2000 helena Major rewrite of the callback APIs. + * 12/07/2000 srl Update of documentation + */ + +/** + * \file + * \brief C API: Character conversion + * + *

Character Conversion C API

+ * + *

This API is used to convert codepage or character encoded data to and + * from UTF-16. You can open a converter with {@link ucnv_open() }. With that + * converter, you can get its properties, set options, convert your data and + * close the converter.

+ * + *

Since many software programs recognize different converter names for + * different types of converters, there are other functions in this API to + * iterate over the converter aliases. The functions {@link ucnv_getAvailableName() }, + * {@link ucnv_getAlias() } and {@link ucnv_getStandardName() } are some of the + * more frequently used alias functions to get this information.

+ * + *

When a converter encounters an illegal, irregular, invalid or unmappable character + * its default behavior is to use a substitution character to replace the + * bad byte sequence. This behavior can be changed by using {@link ucnv_setFromUCallBack() } + * or {@link ucnv_setToUCallBack() } on the converter. The header ucnv_err.h defines + * many other callback actions that can be used instead of a character substitution.

+ * + *

More information about this API can be found in our + * User Guide.

+ */ + +#ifndef UCNV_H +#define UCNV_H + +#include "unicode/ucnv_err.h" +#include "unicode/uenum.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +#if !defined(USET_DEFINED) && !defined(U_IN_DOXYGEN) + +#define USET_DEFINED + +/** + * USet is the C API type corresponding to C++ class UnicodeSet. + * It is forward-declared here to avoid including unicode/uset.h file if related + * conversion APIs are not used. + * + * @see ucnv_getUnicodeSet + * @stable ICU 2.4 + */ +typedef struct USet USet; + +#endif + +#if !UCONFIG_NO_CONVERSION + +U_CDECL_BEGIN + +/** Maximum length of a converter name including the terminating NULL @stable ICU 2.0 */ +#define UCNV_MAX_CONVERTER_NAME_LENGTH 60 +/** Maximum length of a converter name including path and terminating NULL @stable ICU 2.0 */ +#define UCNV_MAX_FULL_FILE_NAME_LENGTH (600+UCNV_MAX_CONVERTER_NAME_LENGTH) + +/** Shift in for EBDCDIC_STATEFUL and iso2022 states @stable ICU 2.0 */ +#define UCNV_SI 0x0F +/** Shift out for EBDCDIC_STATEFUL and iso2022 states @stable ICU 2.0 */ +#define UCNV_SO 0x0E + +/** + * Enum for specifying basic types of converters + * @see ucnv_getType + * @stable ICU 2.0 + */ +typedef enum { + /** @stable ICU 2.0 */ + UCNV_UNSUPPORTED_CONVERTER = -1, + /** @stable ICU 2.0 */ + UCNV_SBCS = 0, + /** @stable ICU 2.0 */ + UCNV_DBCS = 1, + /** @stable ICU 2.0 */ + UCNV_MBCS = 2, + /** @stable ICU 2.0 */ + UCNV_LATIN_1 = 3, + /** @stable ICU 2.0 */ + UCNV_UTF8 = 4, + /** @stable ICU 2.0 */ + UCNV_UTF16_BigEndian = 5, + /** @stable ICU 2.0 */ + UCNV_UTF16_LittleEndian = 6, + /** @stable ICU 2.0 */ + UCNV_UTF32_BigEndian = 7, + /** @stable ICU 2.0 */ + UCNV_UTF32_LittleEndian = 8, + /** @stable ICU 2.0 */ + UCNV_EBCDIC_STATEFUL = 9, + /** @stable ICU 2.0 */ + UCNV_ISO_2022 = 10, + + /** @stable ICU 2.0 */ + UCNV_LMBCS_1 = 11, + /** @stable ICU 2.0 */ + UCNV_LMBCS_2, + /** @stable ICU 2.0 */ + UCNV_LMBCS_3, + /** @stable ICU 2.0 */ + UCNV_LMBCS_4, + /** @stable ICU 2.0 */ + UCNV_LMBCS_5, + /** @stable ICU 2.0 */ + UCNV_LMBCS_6, + /** @stable ICU 2.0 */ + UCNV_LMBCS_8, + /** @stable ICU 2.0 */ + UCNV_LMBCS_11, + /** @stable ICU 2.0 */ + UCNV_LMBCS_16, + /** @stable ICU 2.0 */ + UCNV_LMBCS_17, + /** @stable ICU 2.0 */ + UCNV_LMBCS_18, + /** @stable ICU 2.0 */ + UCNV_LMBCS_19, + /** @stable ICU 2.0 */ + UCNV_LMBCS_LAST = UCNV_LMBCS_19, + /** @stable ICU 2.0 */ + UCNV_HZ, + /** @stable ICU 2.0 */ + UCNV_SCSU, + /** @stable ICU 2.0 */ + UCNV_ISCII, + /** @stable ICU 2.0 */ + UCNV_US_ASCII, + /** @stable ICU 2.0 */ + UCNV_UTF7, + /** @stable ICU 2.2 */ + UCNV_BOCU1, + /** @stable ICU 2.2 */ + UCNV_UTF16, + /** @stable ICU 2.2 */ + UCNV_UTF32, + /** @stable ICU 2.2 */ + UCNV_CESU8, + /** @stable ICU 2.4 */ + UCNV_IMAP_MAILBOX, + /** @stable ICU 4.8 */ + UCNV_COMPOUND_TEXT, + + /* Number of converter types for which we have conversion routines. */ + UCNV_NUMBER_OF_SUPPORTED_CONVERTER_TYPES +} UConverterType; + +/** + * Enum for specifying which platform a converter ID refers to. + * The use of platform/CCSID is not recommended. See ucnv_openCCSID(). + * + * @see ucnv_getPlatform + * @see ucnv_openCCSID + * @see ucnv_getCCSID + * @stable ICU 2.0 + */ +typedef enum { + UCNV_UNKNOWN = -1, + UCNV_IBM = 0 +} UConverterPlatform; + +/** + * Function pointer for error callback in the codepage to unicode direction. + * Called when an error has occurred in conversion to unicode, or on open/close of the callback (see reason). + * @param context Pointer to the callback's private data + * @param args Information about the conversion in progress + * @param codeUnits Points to 'length' bytes of the concerned codepage sequence + * @param length Size (in bytes) of the concerned codepage sequence + * @param reason Defines the reason the callback was invoked + * @param pErrorCode ICU error code in/out parameter. + * For converter callback functions, set to a conversion error + * before the call, and the callback may reset it to U_ZERO_ERROR. + * @see ucnv_setToUCallBack + * @see UConverterToUnicodeArgs + * @stable ICU 2.0 + */ +typedef void (U_EXPORT2 *UConverterToUCallback) ( + const void* context, + UConverterToUnicodeArgs *args, + const char *codeUnits, + int32_t length, + UConverterCallbackReason reason, + UErrorCode *pErrorCode); + +/** + * Function pointer for error callback in the unicode to codepage direction. + * Called when an error has occurred in conversion from unicode, or on open/close of the callback (see reason). + * @param context Pointer to the callback's private data + * @param args Information about the conversion in progress + * @param codeUnits Points to 'length' UChars of the concerned Unicode sequence + * @param length Size (in bytes) of the concerned codepage sequence + * @param codePoint Single UChar32 (UTF-32) containing the concerend Unicode codepoint. + * @param reason Defines the reason the callback was invoked + * @param pErrorCode ICU error code in/out parameter. + * For converter callback functions, set to a conversion error + * before the call, and the callback may reset it to U_ZERO_ERROR. + * @see ucnv_setFromUCallBack + * @stable ICU 2.0 + */ +typedef void (U_EXPORT2 *UConverterFromUCallback) ( + const void* context, + UConverterFromUnicodeArgs *args, + const UChar* codeUnits, + int32_t length, + UChar32 codePoint, + UConverterCallbackReason reason, + UErrorCode *pErrorCode); + +U_CDECL_END + +/** + * Character that separates converter names from options and options from each other. + * @see ucnv_open + * @stable ICU 2.0 + */ +#define UCNV_OPTION_SEP_CHAR ',' + +/** + * String version of UCNV_OPTION_SEP_CHAR. + * @see ucnv_open + * @stable ICU 2.0 + */ +#define UCNV_OPTION_SEP_STRING "," + +/** + * Character that separates a converter option from its value. + * @see ucnv_open + * @stable ICU 2.0 + */ +#define UCNV_VALUE_SEP_CHAR '=' + +/** + * String version of UCNV_VALUE_SEP_CHAR. + * @see ucnv_open + * @stable ICU 2.0 + */ +#define UCNV_VALUE_SEP_STRING "=" + +/** + * Converter option for specifying a locale. + * For example, ucnv_open("SCSU,locale=ja", &errorCode); + * See convrtrs.txt. + * + * @see ucnv_open + * @stable ICU 2.0 + */ +#define UCNV_LOCALE_OPTION_STRING ",locale=" + +/** + * Converter option for specifying a version selector (0..9) for some converters. + * For example, + * \code + * ucnv_open("UTF-7,version=1", &errorCode); + * \endcode + * See convrtrs.txt. + * + * @see ucnv_open + * @stable ICU 2.4 + */ +#define UCNV_VERSION_OPTION_STRING ",version=" + +/** + * Converter option for EBCDIC SBCS or mixed-SBCS/DBCS (stateful) codepages. + * Swaps Unicode mappings for EBCDIC LF and NL codes, as used on + * S/390 (z/OS) Unix System Services (Open Edition). + * For example, ucnv_open("ibm-1047,swaplfnl", &errorCode); + * See convrtrs.txt. + * + * @see ucnv_open + * @stable ICU 2.4 + */ +#define UCNV_SWAP_LFNL_OPTION_STRING ",swaplfnl" + +/** + * Do a fuzzy compare of two converter/alias names. + * The comparison is case-insensitive, ignores leading zeroes if they are not + * followed by further digits, and ignores all but letters and digits. + * Thus the strings "UTF-8", "utf_8", "u*T@f08" and "Utf 8" are exactly equivalent. + * See section 1.4, Charset Alias Matching in Unicode Technical Standard #22 + * at http://www.unicode.org/reports/tr22/ + * + * @param name1 a converter name or alias, zero-terminated + * @param name2 a converter name or alias, zero-terminated + * @return 0 if the names match, or a negative value if the name1 + * lexically precedes name2, or a positive value if the name1 + * lexically follows name2. + * @stable ICU 2.0 + */ +U_CAPI int U_EXPORT2 +ucnv_compareNames(const char *name1, const char *name2); + + +/** + * Creates a UConverter object with the name of a coded character set specified as a C string. + * The actual name will be resolved with the alias file + * using a case-insensitive string comparison that ignores + * leading zeroes and all non-alphanumeric characters. + * E.g., the names "UTF8", "utf-8", "u*T@f08" and "Utf 8" are all equivalent. + * (See also ucnv_compareNames().) + * If NULL is passed for the converter name, it will create one with the + * getDefaultName return value. + * + *

A converter name for ICU 1.5 and above may contain options + * like a locale specification to control the specific behavior of + * the newly instantiated converter. + * The meaning of the options depends on the particular converter. + * If an option is not defined for or recognized by a given converter, then it is ignored.

+ * + *

Options are appended to the converter name string, with a + * UCNV_OPTION_SEP_CHAR between the name and the first option and + * also between adjacent options.

+ * + *

If the alias is ambiguous, then the preferred converter is used + * and the status is set to U_AMBIGUOUS_ALIAS_WARNING.

+ * + *

The conversion behavior and names can vary between platforms. ICU may + * convert some characters differently from other platforms. Details on this topic + * are in the User + * Guide. Aliases starting with a "cp" prefix have no specific meaning + * other than its an alias starting with the letters "cp". Please do not + * associate any meaning to these aliases.

+ * + * \snippet samples/ucnv/convsamp.cpp ucnv_open + * + * @param converterName Name of the coded character set table. + * This may have options appended to the string. + * IANA alias character set names, IBM CCSIDs starting with "ibm-", + * Windows codepage numbers starting with "windows-" are frequently + * used for this parameter. See ucnv_getAvailableName and + * ucnv_getAlias for a complete list that is available. + * If this parameter is NULL, the default converter will be used. + * @param err outgoing error status U_MEMORY_ALLOCATION_ERROR, U_FILE_ACCESS_ERROR + * @return the created Unicode converter object, or NULL if an error occurred + * @see ucnv_openU + * @see ucnv_openCCSID + * @see ucnv_getAvailableName + * @see ucnv_getAlias + * @see ucnv_getDefaultName + * @see ucnv_close + * @see ucnv_compareNames + * @stable ICU 2.0 + */ +U_CAPI UConverter* U_EXPORT2 +ucnv_open(const char *converterName, UErrorCode *err); + + +/** + * Creates a Unicode converter with the names specified as unicode string. + * The name should be limited to the ASCII-7 alphanumerics range. + * The actual name will be resolved with the alias file + * using a case-insensitive string comparison that ignores + * leading zeroes and all non-alphanumeric characters. + * E.g., the names "UTF8", "utf-8", "u*T@f08" and "Utf 8" are all equivalent. + * (See also ucnv_compareNames().) + * If NULL is passed for the converter name, it will create + * one with the ucnv_getDefaultName() return value. + * If the alias is ambiguous, then the preferred converter is used + * and the status is set to U_AMBIGUOUS_ALIAS_WARNING. + * + *

See ucnv_open for the complete details

+ * @param name Name of the UConverter table in a zero terminated + * Unicode string + * @param err outgoing error status U_MEMORY_ALLOCATION_ERROR, + * U_FILE_ACCESS_ERROR + * @return the created Unicode converter object, or NULL if an + * error occurred + * @see ucnv_open + * @see ucnv_openCCSID + * @see ucnv_close + * @see ucnv_compareNames + * @stable ICU 2.0 + */ +U_CAPI UConverter* U_EXPORT2 +ucnv_openU(const UChar *name, + UErrorCode *err); + +/** + * Creates a UConverter object from a CCSID number and platform pair. + * Note that the usefulness of this function is limited to platforms with numeric + * encoding IDs. Only IBM and Microsoft platforms use numeric (16-bit) identifiers for + * encodings. + * + * In addition, IBM CCSIDs and Unicode conversion tables are not 1:1 related. + * For many IBM CCSIDs there are multiple (up to six) Unicode conversion tables, and + * for some Unicode conversion tables there are multiple CCSIDs. + * Some "alternate" Unicode conversion tables are provided by the + * IBM CDRA conversion table registry. + * The most prominent example of a systematic modification of conversion tables that is + * not provided in the form of conversion table files in the repository is + * that S/390 Unix System Services swaps the codes for Line Feed and New Line in all + * EBCDIC codepages, which requires such a swap in the Unicode conversion tables as well. + * + * Only IBM default conversion tables are accessible with ucnv_openCCSID(). + * ucnv_getCCSID() will return the same CCSID for all conversion tables that are associated + * with that CCSID. + * + * Currently, the only "platform" supported in the ICU converter API is UCNV_IBM. + * + * In summary, the use of CCSIDs and the associated API functions is not recommended. + * + * In order to open a converter with the default IBM CDRA Unicode conversion table, + * you can use this function or use the prefix "ibm-": + * \code + * char name[20]; + * sprintf(name, "ibm-%hu", ccsid); + * cnv=ucnv_open(name, &errorCode); + * \endcode + * + * In order to open a converter with the IBM S/390 Unix System Services variant + * of a Unicode/EBCDIC conversion table, + * you can use the prefix "ibm-" together with the option string UCNV_SWAP_LFNL_OPTION_STRING: + * \code + * char name[20]; + * sprintf(name, "ibm-%hu" UCNV_SWAP_LFNL_OPTION_STRING, ccsid); + * cnv=ucnv_open(name, &errorCode); + * \endcode + * + * In order to open a converter from a Microsoft codepage number, use the prefix "cp": + * \code + * char name[20]; + * sprintf(name, "cp%hu", codepageID); + * cnv=ucnv_open(name, &errorCode); + * \endcode + * + * If the alias is ambiguous, then the preferred converter is used + * and the status is set to U_AMBIGUOUS_ALIAS_WARNING. + * + * @param codepage codepage number to create + * @param platform the platform in which the codepage number exists + * @param err error status U_MEMORY_ALLOCATION_ERROR, U_FILE_ACCESS_ERROR + * @return the created Unicode converter object, or NULL if an error + * occurred. + * @see ucnv_open + * @see ucnv_openU + * @see ucnv_close + * @see ucnv_getCCSID + * @see ucnv_getPlatform + * @see UConverterPlatform + * @stable ICU 2.0 + */ +U_CAPI UConverter* U_EXPORT2 +ucnv_openCCSID(int32_t codepage, + UConverterPlatform platform, + UErrorCode * err); + +/** + *

Creates a UConverter object specified from a packageName and a converterName.

+ * + *

The packageName and converterName must point to an ICU udata object, as defined by + * udata_open( packageName, "cnv", converterName, err) or equivalent. + * Typically, packageName will refer to a (.dat) file, or to a package registered with + * udata_setAppData(). Using a full file or directory pathname for packageName is deprecated.

+ * + *

The name will NOT be looked up in the alias mechanism, nor will the converter be + * stored in the converter cache or the alias table. The only way to open further converters + * is call this function multiple times, or use the ucnv_clone() function to clone a + * 'primary' converter.

+ * + *

A future version of ICU may add alias table lookups and/or caching + * to this function.

+ * + *

Example Use: + * cnv = ucnv_openPackage("myapp", "myconverter", &err); + *

+ * + * @param packageName name of the package (equivalent to 'path' in udata_open() call) + * @param converterName name of the data item to be used, without suffix. + * @param err outgoing error status U_MEMORY_ALLOCATION_ERROR, U_FILE_ACCESS_ERROR + * @return the created Unicode converter object, or NULL if an error occurred + * @see udata_open + * @see ucnv_open + * @see ucnv_clone + * @see ucnv_close + * @stable ICU 2.2 + */ +U_CAPI UConverter* U_EXPORT2 +ucnv_openPackage(const char *packageName, const char *converterName, UErrorCode *err); + +/** + * Thread safe converter cloning operation. + * + * You must ucnv_close() the clone. + * + * @param cnv converter to be cloned + * @param status to indicate whether the operation went on smoothly or there were errors + * @return pointer to the new clone + * @stable ICU 71 + */ +U_CAPI UConverter* U_EXPORT2 ucnv_clone(const UConverter *cnv, UErrorCode *status); + +#ifndef U_HIDE_DEPRECATED_API + +/** + * Thread safe converter cloning operation. + * For most efficient operation, pass in a stackBuffer (and a *pBufferSize) + * with at least U_CNV_SAFECLONE_BUFFERSIZE bytes of space. + * If the buffer size is sufficient, then the clone will use the stack buffer; + * otherwise, it will be allocated, and *pBufferSize will indicate + * the actual size. (This should not occur with U_CNV_SAFECLONE_BUFFERSIZE.) + * + * You must ucnv_close() the clone in any case. + * + * If *pBufferSize==0, (regardless of whether stackBuffer==NULL or not) + * then *pBufferSize will be changed to a sufficient size + * for cloning this converter, + * without actually cloning the converter ("pure pre-flighting"). + * + * If *pBufferSize is greater than zero but not large enough for a stack-based + * clone, then the converter is cloned using newly allocated memory + * and *pBufferSize is changed to the necessary size. + * + * If the converter clone fits into the stack buffer but the stack buffer is not + * sufficiently aligned for the clone, then the clone will use an + * adjusted pointer and use an accordingly smaller buffer size. + * + * @param cnv converter to be cloned + * @param stackBuffer Deprecated functionality as of ICU 52, use NULL.
+ * user allocated space for the new clone. If NULL new memory will be allocated. + * If buffer is not large enough, new memory will be allocated. + * Clients can use the U_CNV_SAFECLONE_BUFFERSIZE. This will probably be enough to avoid memory allocations. + * @param pBufferSize Deprecated functionality as of ICU 52, use NULL or 1.
+ * pointer to size of allocated space. + * @param status to indicate whether the operation went on smoothly or there were errors + * An informational status value, U_SAFECLONE_ALLOCATED_WARNING, + * is used if pBufferSize != NULL and any allocations were necessary + * However, it is better to check if *pBufferSize grew for checking for + * allocations because warning codes can be overridden by subsequent + * function calls. + * @return pointer to the new clone + * @deprecated ICU 71 Use ucnv_clone() instead. + */ +U_DEPRECATED UConverter * U_EXPORT2 +ucnv_safeClone(const UConverter *cnv, + void *stackBuffer, + int32_t *pBufferSize, + UErrorCode *status); + +/** + * \def U_CNV_SAFECLONE_BUFFERSIZE + * Definition of a buffer size that is designed to be large enough for + * converters to be cloned with ucnv_safeClone(). + * @deprecated ICU 52. Do not rely on ucnv_safeClone() cloning into any provided buffer. + */ +#define U_CNV_SAFECLONE_BUFFERSIZE 1024 + +#endif /* U_HIDE_DEPRECATED_API */ + +/** + * Deletes the unicode converter and releases resources associated + * with just this instance. + * Does not free up shared converter tables. + * + * @param converter the converter object to be deleted + * @see ucnv_open + * @see ucnv_openU + * @see ucnv_openCCSID + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_close(UConverter * converter); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUConverterPointer + * "Smart pointer" class, closes a UConverter via ucnv_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUConverterPointer, UConverter, ucnv_close); + +U_NAMESPACE_END + +#endif + +/** + * Fills in the output parameter, subChars, with the substitution characters + * as multiple bytes. + * If ucnv_setSubstString() set a Unicode string because the converter is + * stateful, then subChars will be an empty string. + * + * @param converter the Unicode converter + * @param subChars the substitution characters + * @param len on input the capacity of subChars, on output the number + * of bytes copied to it + * @param err the outgoing error status code. + * If the substitution character array is too small, an + * U_INDEX_OUTOFBOUNDS_ERROR will be returned. + * @see ucnv_setSubstString + * @see ucnv_setSubstChars + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_getSubstChars(const UConverter *converter, + char *subChars, + int8_t *len, + UErrorCode *err); + +/** + * Sets the substitution chars when converting from unicode to a codepage. The + * substitution is specified as a string of 1-4 bytes, and may contain + * NULL bytes. + * The subChars must represent a single character. The caller needs to know the + * byte sequence of a valid character in the converter's charset. + * For some converters, for example some ISO 2022 variants, only single-byte + * substitution characters may be supported. + * The newer ucnv_setSubstString() function relaxes these limitations. + * + * @param converter the Unicode converter + * @param subChars the substitution character byte sequence we want set + * @param len the number of bytes in subChars + * @param err the error status code. U_INDEX_OUTOFBOUNDS_ERROR if + * len is bigger than the maximum number of bytes allowed in subchars + * @see ucnv_setSubstString + * @see ucnv_getSubstChars + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_setSubstChars(UConverter *converter, + const char *subChars, + int8_t len, + UErrorCode *err); + +/** + * Set a substitution string for converting from Unicode to a charset. + * The caller need not know the charset byte sequence for each charset. + * + * Unlike ucnv_setSubstChars() which is designed to set a charset byte sequence + * for a single character, this function takes a Unicode string with + * zero, one or more characters, and immediately verifies that the string can be + * converted to the charset. + * If not, or if the result is too long (more than 32 bytes as of ICU 3.6), + * then the function returns with an error accordingly. + * + * Also unlike ucnv_setSubstChars(), this function works for stateful charsets + * by converting on the fly at the point of substitution rather than setting + * a fixed byte sequence. + * + * @param cnv The UConverter object. + * @param s The Unicode string. + * @param length The number of UChars in s, or -1 for a NUL-terminated string. + * @param err Pointer to a standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * + * @see ucnv_setSubstChars + * @see ucnv_getSubstChars + * @stable ICU 3.6 + */ +U_CAPI void U_EXPORT2 +ucnv_setSubstString(UConverter *cnv, + const UChar *s, + int32_t length, + UErrorCode *err); + +/** + * Fills in the output parameter, errBytes, with the error characters from the + * last failing conversion. + * + * @param converter the Unicode converter + * @param errBytes the codepage bytes which were in error + * @param len on input the capacity of errBytes, on output the number of + * bytes which were copied to it + * @param err the error status code. + * If the substitution character array is too small, an + * U_INDEX_OUTOFBOUNDS_ERROR will be returned. + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_getInvalidChars(const UConverter *converter, + char *errBytes, + int8_t *len, + UErrorCode *err); + +/** + * Fills in the output parameter, errChars, with the error characters from the + * last failing conversion. + * + * @param converter the Unicode converter + * @param errUChars the UChars which were in error + * @param len on input the capacity of errUChars, on output the number of + * UChars which were copied to it + * @param err the error status code. + * If the substitution character array is too small, an + * U_INDEX_OUTOFBOUNDS_ERROR will be returned. + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_getInvalidUChars(const UConverter *converter, + UChar *errUChars, + int8_t *len, + UErrorCode *err); + +/** + * Resets the state of a converter to the default state. This is used + * in the case of an error, to restart a conversion from a known default state. + * It will also empty the internal output buffers. + * @param converter the Unicode converter + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_reset(UConverter *converter); + +/** + * Resets the to-Unicode part of a converter state to the default state. + * This is used in the case of an error to restart a conversion to + * Unicode to a known default state. It will also empty the internal + * output buffers used for the conversion to Unicode codepoints. + * @param converter the Unicode converter + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_resetToUnicode(UConverter *converter); + +/** + * Resets the from-Unicode part of a converter state to the default state. + * This is used in the case of an error to restart a conversion from + * Unicode to a known default state. It will also empty the internal output + * buffers used for the conversion from Unicode codepoints. + * @param converter the Unicode converter + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_resetFromUnicode(UConverter *converter); + +/** + * Returns the maximum number of bytes that are output per UChar in conversion + * from Unicode using this converter. + * The returned number can be used with UCNV_GET_MAX_BYTES_FOR_STRING + * to calculate the size of a target buffer for conversion from Unicode. + * + * Note: Before ICU 2.8, this function did not return reliable numbers for + * some stateful converters (EBCDIC_STATEFUL, ISO-2022) and LMBCS. + * + * This number may not be the same as the maximum number of bytes per + * "conversion unit". In other words, it may not be the intuitively expected + * number of bytes per character that would be published for a charset, + * and may not fulfill any other purpose than the allocation of an output + * buffer of guaranteed sufficient size for a given input length and converter. + * + * Examples for special cases that are taken into account: + * - Supplementary code points may convert to more bytes than BMP code points. + * This function returns bytes per UChar (UTF-16 code unit), not per + * Unicode code point, for efficient buffer allocation. + * - State-shifting output (SI/SO, escapes, etc.) from stateful converters. + * - When m input UChars are converted to n output bytes, then the maximum m/n + * is taken into account. + * + * The number returned here does not take into account + * (see UCNV_GET_MAX_BYTES_FOR_STRING): + * - callbacks which output more than one charset character sequence per call, + * like escape callbacks + * - initial and final non-character bytes that are output by some converters + * (automatic BOMs, initial escape sequence, final SI, etc.) + * + * Examples for returned values: + * - SBCS charsets: 1 + * - Shift-JIS: 2 + * - UTF-16: 2 (2 per BMP, 4 per surrogate _pair_, BOM not counted) + * - UTF-8: 3 (3 per BMP, 4 per surrogate _pair_) + * - EBCDIC_STATEFUL (EBCDIC mixed SBCS/DBCS): 3 (SO + DBCS) + * - ISO-2022: 3 (always outputs UTF-8) + * - ISO-2022-JP: 6 (4-byte escape sequences + DBCS) + * - ISO-2022-CN: 8 (4-byte designator sequences + 2-byte SS2/SS3 + DBCS) + * + * @param converter The Unicode converter. + * @return The maximum number of bytes per UChar (16 bit code unit) + * that are output by ucnv_fromUnicode(), + * to be used together with UCNV_GET_MAX_BYTES_FOR_STRING + * for buffer allocation. + * + * @see UCNV_GET_MAX_BYTES_FOR_STRING + * @see ucnv_getMinCharSize + * @stable ICU 2.0 + */ +U_CAPI int8_t U_EXPORT2 +ucnv_getMaxCharSize(const UConverter *converter); + +/** + * Calculates the size of a buffer for conversion from Unicode to a charset. + * The calculated size is guaranteed to be sufficient for this conversion. + * + * It takes into account initial and final non-character bytes that are output + * by some converters. + * It does not take into account callbacks which output more than one charset + * character sequence per call, like escape callbacks. + * The default (substitution) callback only outputs one charset character sequence. + * + * @param length Number of UChars to be converted. + * @param maxCharSize Return value from ucnv_getMaxCharSize() for the converter + * that will be used. + * @return Size of a buffer that will be large enough to hold the output bytes of + * converting length UChars with the converter that returned the maxCharSize. + * + * @see ucnv_getMaxCharSize + * @stable ICU 2.8 + */ +#define UCNV_GET_MAX_BYTES_FOR_STRING(length, maxCharSize) \ + (((int32_t)(length)+10)*(int32_t)(maxCharSize)) + +/** + * Returns the minimum byte length (per codepoint) for characters in this codepage. + * This is usually either 1 or 2. + * @param converter the Unicode converter + * @return the minimum number of bytes per codepoint allowed by this particular converter + * @see ucnv_getMaxCharSize + * @stable ICU 2.0 + */ +U_CAPI int8_t U_EXPORT2 +ucnv_getMinCharSize(const UConverter *converter); + +/** + * Returns the display name of the converter passed in based on the Locale + * passed in. If the locale contains no display name, the internal ASCII + * name will be filled in. + * + * @param converter the Unicode converter. + * @param displayLocale is the specific Locale we want to localized for + * @param displayName user provided buffer to be filled in + * @param displayNameCapacity size of displayName Buffer + * @param err error status code + * @return displayNameLength number of UChar needed in displayName + * @see ucnv_getName + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ucnv_getDisplayName(const UConverter *converter, + const char *displayLocale, + UChar *displayName, + int32_t displayNameCapacity, + UErrorCode *err); + +/** + * Gets the internal, canonical name of the converter (zero-terminated). + * The lifetime of the returned string will be that of the converter + * passed to this function. + * @param converter the Unicode converter + * @param err UErrorCode status + * @return the internal name of the converter + * @see ucnv_getDisplayName + * @stable ICU 2.0 + */ +U_CAPI const char * U_EXPORT2 +ucnv_getName(const UConverter *converter, UErrorCode *err); + +/** + * Gets a codepage number associated with the converter. This is not guaranteed + * to be the one used to create the converter. Some converters do not represent + * platform registered codepages and return zero for the codepage number. + * The error code fill-in parameter indicates if the codepage number + * is available. + * Does not check if the converter is NULL or if converter's data + * table is NULL. + * + * Important: The use of CCSIDs is not recommended because it is limited + * to only two platforms in principle and only one (UCNV_IBM) in the current + * ICU converter API. + * Also, CCSIDs are insufficient to identify IBM Unicode conversion tables precisely. + * For more details see ucnv_openCCSID(). + * + * @param converter the Unicode converter + * @param err the error status code. + * @return If any error occurs, -1 will be returned otherwise, the codepage number + * will be returned + * @see ucnv_openCCSID + * @see ucnv_getPlatform + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ucnv_getCCSID(const UConverter *converter, + UErrorCode *err); + +/** + * Gets a codepage platform associated with the converter. Currently, + * only UCNV_IBM will be returned. + * Does not test if the converter is NULL or if converter's data + * table is NULL. + * @param converter the Unicode converter + * @param err the error status code. + * @return The codepage platform + * @stable ICU 2.0 + */ +U_CAPI UConverterPlatform U_EXPORT2 +ucnv_getPlatform(const UConverter *converter, + UErrorCode *err); + +/** + * Gets the type of the converter + * e.g. SBCS, MBCS, DBCS, UTF8, UTF16_BE, UTF16_LE, ISO_2022, + * EBCDIC_STATEFUL, LATIN_1 + * @param converter a valid, opened converter + * @return the type of the converter + * @stable ICU 2.0 + */ +U_CAPI UConverterType U_EXPORT2 +ucnv_getType(const UConverter * converter); + +/** + * Gets the "starter" (lead) bytes for converters of type MBCS. + * Will fill in an U_ILLEGAL_ARGUMENT_ERROR if converter passed in + * is not MBCS. Fills in an array of type UBool, with the value of the byte + * as offset to the array. For example, if (starters[0x20] == true) at return, + * it means that the byte 0x20 is a starter byte in this converter. + * Context pointers are always owned by the caller. + * + * @param converter a valid, opened converter of type MBCS + * @param starters an array of size 256 to be filled in + * @param err error status, U_ILLEGAL_ARGUMENT_ERROR if the + * converter is not a type which can return starters. + * @see ucnv_getType + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_getStarters(const UConverter* converter, + UBool starters[256], + UErrorCode* err); + + +/** + * Selectors for Unicode sets that can be returned by ucnv_getUnicodeSet(). + * @see ucnv_getUnicodeSet + * @stable ICU 2.6 + */ +typedef enum UConverterUnicodeSet { + /** Select the set of roundtrippable Unicode code points. @stable ICU 2.6 */ + UCNV_ROUNDTRIP_SET, + /** Select the set of Unicode code points with roundtrip or fallback mappings. @stable ICU 4.0 */ + UCNV_ROUNDTRIP_AND_FALLBACK_SET, +#ifndef U_HIDE_DEPRECATED_API + /** + * Number of UConverterUnicodeSet selectors. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + UCNV_SET_COUNT +#endif // U_HIDE_DEPRECATED_API +} UConverterUnicodeSet; + + +/** + * Returns the set of Unicode code points that can be converted by an ICU converter. + * + * Returns one of several kinds of set: + * + * 1. UCNV_ROUNDTRIP_SET + * + * The set of all Unicode code points that can be roundtrip-converted + * (converted without any data loss) with the converter (ucnv_fromUnicode()). + * This set will not include code points that have fallback mappings + * or are only the result of reverse fallback mappings. + * This set will also not include PUA code points with fallbacks, although + * ucnv_fromUnicode() will always uses those mappings despite ucnv_setFallback(). + * See UTR #22 "Character Mapping Markup Language" + * at http://www.unicode.org/reports/tr22/ + * + * This is useful for example for + * - checking that a string or document can be roundtrip-converted with a converter, + * without/before actually performing the conversion + * - testing if a converter can be used for text for typical text for a certain locale, + * by comparing its roundtrip set with the set of ExemplarCharacters from + * ICU's locale data or other sources + * + * 2. UCNV_ROUNDTRIP_AND_FALLBACK_SET + * + * The set of all Unicode code points that can be converted with the converter (ucnv_fromUnicode()) + * when fallbacks are turned on (see ucnv_setFallback()). + * This set includes all code points with roundtrips and fallbacks (but not reverse fallbacks). + * + * In the future, there may be more UConverterUnicodeSet choices to select + * sets with different properties. + * + * @param cnv The converter for which a set is requested. + * @param setFillIn A valid USet *. It will be cleared by this function before + * the converter's specific set is filled into the USet. + * @param whichSet A UConverterUnicodeSet selector; + * currently UCNV_ROUNDTRIP_SET is the only supported value. + * @param pErrorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * + * @see UConverterUnicodeSet + * @see uset_open + * @see uset_close + * @stable ICU 2.6 + */ +U_CAPI void U_EXPORT2 +ucnv_getUnicodeSet(const UConverter *cnv, + USet *setFillIn, + UConverterUnicodeSet whichSet, + UErrorCode *pErrorCode); + +/** + * Gets the current callback function used by the converter when an illegal + * or invalid codepage sequence is found. + * Context pointers are always owned by the caller. + * + * @param converter the unicode converter + * @param action fillin: returns the callback function pointer + * @param context fillin: returns the callback's private void* context + * @see ucnv_setToUCallBack + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_getToUCallBack (const UConverter * converter, + UConverterToUCallback *action, + const void **context); + +/** + * Gets the current callback function used by the converter when illegal + * or invalid Unicode sequence is found. + * Context pointers are always owned by the caller. + * + * @param converter the unicode converter + * @param action fillin: returns the callback function pointer + * @param context fillin: returns the callback's private void* context + * @see ucnv_setFromUCallBack + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_getFromUCallBack (const UConverter * converter, + UConverterFromUCallback *action, + const void **context); + +/** + * Changes the callback function used by the converter when + * an illegal or invalid sequence is found. + * Context pointers are always owned by the caller. + * Predefined actions and contexts can be found in the ucnv_err.h header. + * + * @param converter the unicode converter + * @param newAction the new callback function + * @param newContext the new toUnicode callback context pointer. This can be NULL. + * @param oldAction fillin: returns the old callback function pointer. This can be NULL. + * @param oldContext fillin: returns the old callback's private void* context. This can be NULL. + * @param err The error code status + * @see ucnv_getToUCallBack + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_setToUCallBack (UConverter * converter, + UConverterToUCallback newAction, + const void* newContext, + UConverterToUCallback *oldAction, + const void** oldContext, + UErrorCode * err); + +/** + * Changes the current callback function used by the converter when + * an illegal or invalid sequence is found. + * Context pointers are always owned by the caller. + * Predefined actions and contexts can be found in the ucnv_err.h header. + * + * @param converter the unicode converter + * @param newAction the new callback function + * @param newContext the new fromUnicode callback context pointer. This can be NULL. + * @param oldAction fillin: returns the old callback function pointer. This can be NULL. + * @param oldContext fillin: returns the old callback's private void* context. This can be NULL. + * @param err The error code status + * @see ucnv_getFromUCallBack + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_setFromUCallBack (UConverter * converter, + UConverterFromUCallback newAction, + const void *newContext, + UConverterFromUCallback *oldAction, + const void **oldContext, + UErrorCode * err); + +/** + * Converts an array of unicode characters to an array of codepage + * characters. This function is optimized for converting a continuous + * stream of data in buffer-sized chunks, where the entire source and + * target does not fit in available buffers. + * + * The source pointer is an in/out parameter. It starts out pointing where the + * conversion is to begin, and ends up pointing after the last UChar consumed. + * + * Target similarly starts out pointer at the first available byte in the output + * buffer, and ends up pointing after the last byte written to the output. + * + * The converter always attempts to consume the entire source buffer, unless + * (1.) the target buffer is full, or (2.) a failing error is returned from the + * current callback function. When a successful error status has been + * returned, it means that all of the source buffer has been + * consumed. At that point, the caller should reset the source and + * sourceLimit pointers to point to the next chunk. + * + * At the end of the stream (flush==true), the input is completely consumed + * when *source==sourceLimit and no error code is set. + * The converter object is then automatically reset by this function. + * (This means that a converter need not be reset explicitly between data + * streams if it finishes the previous stream without errors.) + * + * This is a stateful conversion. Additionally, even when all source data has + * been consumed, some data may be in the converters' internal state. + * Call this function repeatedly, updating the target pointers with + * the next empty chunk of target in case of a + * U_BUFFER_OVERFLOW_ERROR, and updating the source pointers + * with the next chunk of source when a successful error status is + * returned, until there are no more chunks of source data. + * @param converter the Unicode converter + * @param target I/O parameter. Input : Points to the beginning of the buffer to copy + * codepage characters to. Output : points to after the last codepage character copied + * to target. + * @param targetLimit the pointer just after last of the target buffer + * @param source I/O parameter, pointer to pointer to the source Unicode character buffer. + * @param sourceLimit the pointer just after the last of the source buffer + * @param offsets if NULL is passed, nothing will happen to it, otherwise it needs to have the same number + * of allocated cells as target. Will fill in offsets from target to source pointer + * e.g: offsets[3] is equal to 6, it means that the target[3] was a result of transcoding source[6] + * For output data carried across calls, and other data without a specific source character + * (such as from escape sequences or callbacks) -1 will be placed for offsets. + * @param flush set to true if the current source buffer is the last available + * chunk of the source, false otherwise. Note that if a failing status is returned, + * this function may have to be called multiple times with flush set to true until + * the source buffer is consumed. + * @param err the error status. U_ILLEGAL_ARGUMENT_ERROR will be set if the + * converter is NULL. + * U_BUFFER_OVERFLOW_ERROR will be set if the target is full and there is + * still data to be written to the target. + * @see ucnv_fromUChars + * @see ucnv_convert + * @see ucnv_getMinCharSize + * @see ucnv_setToUCallBack + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_fromUnicode (UConverter * converter, + char **target, + const char *targetLimit, + const UChar ** source, + const UChar * sourceLimit, + int32_t* offsets, + UBool flush, + UErrorCode * err); + +/** + * Converts a buffer of codepage bytes into an array of unicode UChars + * characters. This function is optimized for converting a continuous + * stream of data in buffer-sized chunks, where the entire source and + * target does not fit in available buffers. + * + * The source pointer is an in/out parameter. It starts out pointing where the + * conversion is to begin, and ends up pointing after the last byte of source consumed. + * + * Target similarly starts out pointer at the first available UChar in the output + * buffer, and ends up pointing after the last UChar written to the output. + * It does NOT necessarily keep UChar sequences together. + * + * The converter always attempts to consume the entire source buffer, unless + * (1.) the target buffer is full, or (2.) a failing error is returned from the + * current callback function. When a successful error status has been + * returned, it means that all of the source buffer has been + * consumed. At that point, the caller should reset the source and + * sourceLimit pointers to point to the next chunk. + * + * At the end of the stream (flush==true), the input is completely consumed + * when *source==sourceLimit and no error code is set + * The converter object is then automatically reset by this function. + * (This means that a converter need not be reset explicitly between data + * streams if it finishes the previous stream without errors.) + * + * This is a stateful conversion. Additionally, even when all source data has + * been consumed, some data may be in the converters' internal state. + * Call this function repeatedly, updating the target pointers with + * the next empty chunk of target in case of a + * U_BUFFER_OVERFLOW_ERROR, and updating the source pointers + * with the next chunk of source when a successful error status is + * returned, until there are no more chunks of source data. + * @param converter the Unicode converter + * @param target I/O parameter. Input : Points to the beginning of the buffer to copy + * UChars into. Output : points to after the last UChar copied. + * @param targetLimit the pointer just after the end of the target buffer + * @param source I/O parameter, pointer to pointer to the source codepage buffer. + * @param sourceLimit the pointer to the byte after the end of the source buffer + * @param offsets if NULL is passed, nothing will happen to it, otherwise it needs to have the same number + * of allocated cells as target. Will fill in offsets from target to source pointer + * e.g: offsets[3] is equal to 6, it means that the target[3] was a result of transcoding source[6] + * For output data carried across calls, and other data without a specific source character + * (such as from escape sequences or callbacks) -1 will be placed for offsets. + * @param flush set to true if the current source buffer is the last available + * chunk of the source, false otherwise. Note that if a failing status is returned, + * this function may have to be called multiple times with flush set to true until + * the source buffer is consumed. + * @param err the error status. U_ILLEGAL_ARGUMENT_ERROR will be set if the + * converter is NULL. + * U_BUFFER_OVERFLOW_ERROR will be set if the target is full and there is + * still data to be written to the target. + * @see ucnv_fromUChars + * @see ucnv_convert + * @see ucnv_getMinCharSize + * @see ucnv_setFromUCallBack + * @see ucnv_getNextUChar + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_toUnicode(UConverter *converter, + UChar **target, + const UChar *targetLimit, + const char **source, + const char *sourceLimit, + int32_t *offsets, + UBool flush, + UErrorCode *err); + +/** + * Convert the Unicode string into a codepage string using an existing UConverter. + * The output string is NUL-terminated if possible. + * + * This function is a more convenient but less powerful version of ucnv_fromUnicode(). + * It is only useful for whole strings, not for streaming conversion. + * + * The maximum output buffer capacity required (barring output from callbacks) will be + * UCNV_GET_MAX_BYTES_FOR_STRING(srcLength, ucnv_getMaxCharSize(cnv)). + * + * @param cnv the converter object to be used (ucnv_resetFromUnicode() will be called) + * @param src the input Unicode string + * @param srcLength the input string length, or -1 if NUL-terminated + * @param dest destination string buffer, can be NULL if destCapacity==0 + * @param destCapacity the number of chars available at dest + * @param pErrorCode normal ICU error code; + * common error codes that may be set by this function include + * U_BUFFER_OVERFLOW_ERROR, U_STRING_NOT_TERMINATED_WARNING, + * U_ILLEGAL_ARGUMENT_ERROR, and conversion errors + * @return the length of the output string, not counting the terminating NUL; + * if the length is greater than destCapacity, then the string will not fit + * and a buffer of the indicated length would need to be passed in + * @see ucnv_fromUnicode + * @see ucnv_convert + * @see UCNV_GET_MAX_BYTES_FOR_STRING + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ucnv_fromUChars(UConverter *cnv, + char *dest, int32_t destCapacity, + const UChar *src, int32_t srcLength, + UErrorCode *pErrorCode); + +/** + * Convert the codepage string into a Unicode string using an existing UConverter. + * The output string is NUL-terminated if possible. + * + * This function is a more convenient but less powerful version of ucnv_toUnicode(). + * It is only useful for whole strings, not for streaming conversion. + * + * The maximum output buffer capacity required (barring output from callbacks) will be + * 2*srcLength (each char may be converted into a surrogate pair). + * + * @param cnv the converter object to be used (ucnv_resetToUnicode() will be called) + * @param src the input codepage string + * @param srcLength the input string length, or -1 if NUL-terminated + * @param dest destination string buffer, can be NULL if destCapacity==0 + * @param destCapacity the number of UChars available at dest + * @param pErrorCode normal ICU error code; + * common error codes that may be set by this function include + * U_BUFFER_OVERFLOW_ERROR, U_STRING_NOT_TERMINATED_WARNING, + * U_ILLEGAL_ARGUMENT_ERROR, and conversion errors + * @return the length of the output string, not counting the terminating NUL; + * if the length is greater than destCapacity, then the string will not fit + * and a buffer of the indicated length would need to be passed in + * @see ucnv_toUnicode + * @see ucnv_convert + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ucnv_toUChars(UConverter *cnv, + UChar *dest, int32_t destCapacity, + const char *src, int32_t srcLength, + UErrorCode *pErrorCode); + +/** + * Convert a codepage buffer into Unicode one character at a time. + * The input is completely consumed when the U_INDEX_OUTOFBOUNDS_ERROR is set. + * + * Advantage compared to ucnv_toUnicode() or ucnv_toUChars(): + * - Faster for small amounts of data, for most converters, e.g., + * US-ASCII, ISO-8859-1, UTF-8/16/32, and most "normal" charsets. + * (For complex converters, e.g., SCSU, UTF-7 and ISO 2022 variants, + * it uses ucnv_toUnicode() internally.) + * - Convenient. + * + * Limitations compared to ucnv_toUnicode(): + * - Always assumes flush=true. + * This makes ucnv_getNextUChar() unsuitable for "streaming" conversion, + * that is, for where the input is supplied in multiple buffers, + * because ucnv_getNextUChar() will assume the end of the input at the end + * of the first buffer. + * - Does not provide offset output. + * + * It is possible to "mix" ucnv_getNextUChar() and ucnv_toUnicode() because + * ucnv_getNextUChar() uses the current state of the converter + * (unlike ucnv_toUChars() which always resets first). + * However, if ucnv_getNextUChar() is called after ucnv_toUnicode() + * stopped in the middle of a character sequence (with flush=false), + * then ucnv_getNextUChar() will always use the slower ucnv_toUnicode() + * internally until the next character boundary. + * (This is new in ICU 2.6. In earlier releases, ucnv_getNextUChar() had to + * start at a character boundary.) + * + * Instead of using ucnv_getNextUChar(), it is recommended + * to convert using ucnv_toUnicode() or ucnv_toUChars() + * and then iterate over the text using U16_NEXT() or a UCharIterator (uiter.h) + * or a C++ CharacterIterator or similar. + * This allows streaming conversion and offset output, for example. + * + *

Handling of surrogate pairs and supplementary-plane code points:
+ * There are two different kinds of codepages that provide mappings for surrogate characters: + *

    + *
  • Codepages like UTF-8, UTF-32, and GB 18030 provide direct representations for Unicode + * code points U+10000-U+10ffff as well as for single surrogates U+d800-U+dfff. + * Each valid sequence will result in exactly one returned code point. + * If a sequence results in a single surrogate, then that will be returned + * by itself, even if a neighboring sequence encodes the matching surrogate.
    • + *
  • Codepages like SCSU and LMBCS (and UTF-16) provide direct representations only for BMP code points + * including surrogates. Code points in supplementary planes are represented with + * two sequences, each encoding a surrogate. + * For these codepages, matching pairs of surrogates will be combined into single + * code points for returning from this function. + * (Note that SCSU is actually a mix of these codepage types.)
    • + *

+ * + * @param converter an open UConverter + * @param source the address of a pointer to the codepage buffer, will be + * updated to point after the bytes consumed in the conversion call. + * @param sourceLimit points to the end of the input buffer + * @param err fills in error status (see ucnv_toUnicode) + * U_INDEX_OUTOFBOUNDS_ERROR will be set if the input + * is empty or does not convert to any output (e.g.: pure state-change + * codes SI/SO, escape sequences for ISO 2022, + * or if the callback did not output anything, ...). + * This function will not set a U_BUFFER_OVERFLOW_ERROR because + * the "buffer" is the return code. However, there might be subsequent output + * stored in the converter object + * that will be returned in following calls to this function. + * @return a UChar32 resulting from the partial conversion of source + * @see ucnv_toUnicode + * @see ucnv_toUChars + * @see ucnv_convert + * @stable ICU 2.0 + */ +U_CAPI UChar32 U_EXPORT2 +ucnv_getNextUChar(UConverter * converter, + const char **source, + const char * sourceLimit, + UErrorCode * err); + +/** + * Convert from one external charset to another using two existing UConverters. + * Internally, two conversions - ucnv_toUnicode() and ucnv_fromUnicode() - + * are used, "pivoting" through 16-bit Unicode. + * + * Important: For streaming conversion (multiple function calls for successive + * parts of a text stream), the caller must provide a pivot buffer explicitly, + * and must preserve the pivot buffer and associated pointers from one + * call to another. (The buffer may be moved if its contents and the relative + * pointer positions are preserved.) + * + * There is a similar function, ucnv_convert(), + * which has the following limitations: + * - it takes charset names, not converter objects, so that + * - two converters are opened for each call + * - only single-string conversion is possible, not streaming operation + * - it does not provide enough information to find out, + * in case of failure, whether the toUnicode or + * the fromUnicode conversion failed + * + * By contrast, ucnv_convertEx() + * - takes UConverter parameters instead of charset names + * - fully exposes the pivot buffer for streaming conversion and complete error handling + * + * ucnv_convertEx() also provides further convenience: + * - an option to reset the converters at the beginning + * (if reset==true, see parameters; + * also sets *pivotTarget=*pivotSource=pivotStart) + * - allow NUL-terminated input + * (only a single NUL byte, will not work for charsets with multi-byte NULs) + * (if sourceLimit==NULL, see parameters) + * - terminate with a NUL on output + * (only a single NUL byte, not useful for charsets with multi-byte NULs), + * or set U_STRING_NOT_TERMINATED_WARNING if the output exactly fills + * the target buffer + * - the pivot buffer can be provided internally; + * possible only for whole-string conversion, not streaming conversion; + * in this case, the caller will not be able to get details about where an + * error occurred + * (if pivotStart==NULL, see below) + * + * The function returns when one of the following is true: + * - the entire source text has been converted successfully to the target buffer + * - a target buffer overflow occurred (U_BUFFER_OVERFLOW_ERROR) + * - a conversion error occurred + * (other U_FAILURE(), see description of pErrorCode) + * + * Limitation compared to the direct use of + * ucnv_fromUnicode() and ucnv_toUnicode(): + * ucnv_convertEx() does not provide offset information. + * + * Limitation compared to ucnv_fromUChars() and ucnv_toUChars(): + * ucnv_convertEx() does not support preflighting directly. + * + * Sample code for converting a single string from + * one external charset to UTF-8, ignoring the location of errors: + * + * \code + * int32_t + * myToUTF8(UConverter *cnv, + * const char *s, int32_t length, + * char *u8, int32_t capacity, + * UErrorCode *pErrorCode) { + * UConverter *utf8Cnv; + * char *target; + * + * if(U_FAILURE(*pErrorCode)) { + * return 0; + * } + * + * utf8Cnv=myGetCachedUTF8Converter(pErrorCode); + * if(U_FAILURE(*pErrorCode)) { + * return 0; + * } + * + * if(length<0) { + * length=strlen(s); + * } + * target=u8; + * ucnv_convertEx(utf8Cnv, cnv, + * &target, u8+capacity, + * &s, s+length, + * NULL, NULL, NULL, NULL, + * true, true, + * pErrorCode); + * + * myReleaseCachedUTF8Converter(utf8Cnv); + * + * // return the output string length, but without preflighting + * return (int32_t)(target-u8); + * } + * \endcode + * + * @param targetCnv Output converter, used to convert from the UTF-16 pivot + * to the target using ucnv_fromUnicode(). + * @param sourceCnv Input converter, used to convert from the source to + * the UTF-16 pivot using ucnv_toUnicode(). + * @param target I/O parameter, same as for ucnv_fromUChars(). + * Input: *target points to the beginning of the target buffer. + * Output: *target points to the first unit after the last char written. + * @param targetLimit Pointer to the first unit after the target buffer. + * @param source I/O parameter, same as for ucnv_toUChars(). + * Input: *source points to the beginning of the source buffer. + * Output: *source points to the first unit after the last char read. + * @param sourceLimit Pointer to the first unit after the source buffer. + * @param pivotStart Pointer to the UTF-16 pivot buffer. If pivotStart==NULL, + * then an internal buffer is used and the other pivot + * arguments are ignored and can be NULL as well. + * @param pivotSource I/O parameter, same as source in ucnv_fromUChars() for + * conversion from the pivot buffer to the target buffer. + * @param pivotTarget I/O parameter, same as target in ucnv_toUChars() for + * conversion from the source buffer to the pivot buffer. + * It must be pivotStart<=*pivotSource<=*pivotTarget<=pivotLimit + * and pivotStart[0..ucnv_countAvailable()]) + * @return a pointer a string (library owned), or NULL if the index is out of bounds. + * @see ucnv_countAvailable + * @stable ICU 2.0 + */ +U_CAPI const char* U_EXPORT2 +ucnv_getAvailableName(int32_t n); + +/** + * Returns a UEnumeration to enumerate all of the canonical converter + * names, as per the alias file, regardless of the ability to open each + * converter. + * + * @return A UEnumeration object for getting all the recognized canonical + * converter names. + * @see ucnv_getAvailableName + * @see uenum_close + * @see uenum_next + * @stable ICU 2.4 + */ +U_CAPI UEnumeration * U_EXPORT2 +ucnv_openAllNames(UErrorCode *pErrorCode); + +/** + * Gives the number of aliases for a given converter or alias name. + * If the alias is ambiguous, then the preferred converter is used + * and the status is set to U_AMBIGUOUS_ALIAS_WARNING. + * This method only enumerates the listed entries in the alias file. + * @param alias alias name + * @param pErrorCode error status + * @return number of names on alias list for given alias + * @stable ICU 2.0 + */ +U_CAPI uint16_t U_EXPORT2 +ucnv_countAliases(const char *alias, UErrorCode *pErrorCode); + +/** + * Gives the name of the alias at given index of alias list. + * This method only enumerates the listed entries in the alias file. + * If the alias is ambiguous, then the preferred converter is used + * and the status is set to U_AMBIGUOUS_ALIAS_WARNING. + * @param alias alias name + * @param n index in alias list + * @param pErrorCode result of operation + * @return returns the name of the alias at given index + * @see ucnv_countAliases + * @stable ICU 2.0 + */ +U_CAPI const char * U_EXPORT2 +ucnv_getAlias(const char *alias, uint16_t n, UErrorCode *pErrorCode); + +/** + * Fill-up the list of alias names for the given alias. + * This method only enumerates the listed entries in the alias file. + * If the alias is ambiguous, then the preferred converter is used + * and the status is set to U_AMBIGUOUS_ALIAS_WARNING. + * @param alias alias name + * @param aliases fill-in list, aliases is a pointer to an array of + * ucnv_countAliases() string-pointers + * (const char *) that will be filled in. + * The strings themselves are owned by the library. + * @param pErrorCode result of operation + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_getAliases(const char *alias, const char **aliases, UErrorCode *pErrorCode); + +/** + * Return a new UEnumeration object for enumerating all the + * alias names for a given converter that are recognized by a standard. + * This method only enumerates the listed entries in the alias file. + * The convrtrs.txt file can be modified to change the results of + * this function. + * The first result in this list is the same result given by + * ucnv_getStandardName, which is the default alias for + * the specified standard name. The returned object must be closed with + * uenum_close when you are done with the object. + * + * @param convName original converter name + * @param standard name of the standard governing the names; MIME and IANA + * are such standards + * @param pErrorCode The error code + * @return A UEnumeration object for getting all aliases that are recognized + * by a standard. If any of the parameters are invalid, NULL + * is returned. + * @see ucnv_getStandardName + * @see uenum_close + * @see uenum_next + * @stable ICU 2.2 + */ +U_CAPI UEnumeration * U_EXPORT2 +ucnv_openStandardNames(const char *convName, + const char *standard, + UErrorCode *pErrorCode); + +/** + * Gives the number of standards associated to converter names. + * @return number of standards + * @stable ICU 2.0 + */ +U_CAPI uint16_t U_EXPORT2 +ucnv_countStandards(void); + +/** + * Gives the name of the standard at given index of standard list. + * @param n index in standard list + * @param pErrorCode result of operation + * @return returns the name of the standard at given index. Owned by the library. + * @stable ICU 2.0 + */ +U_CAPI const char * U_EXPORT2 +ucnv_getStandard(uint16_t n, UErrorCode *pErrorCode); + +/** + * Returns a standard name for a given converter name. + *

+ * Example alias table:
+ * conv alias1 { STANDARD1 } alias2 { STANDARD1* } + *

+ * Result of ucnv_getStandardName("conv", "STANDARD1") from example + * alias table:
+ * "alias2" + * + * @param name original converter name + * @param standard name of the standard governing the names; MIME and IANA + * are such standards + * @param pErrorCode result of operation + * @return returns the standard converter name; + * if a standard converter name cannot be determined, + * then NULL is returned. Owned by the library. + * @stable ICU 2.0 + */ +U_CAPI const char * U_EXPORT2 +ucnv_getStandardName(const char *name, const char *standard, UErrorCode *pErrorCode); + +/** + * This function will return the internal canonical converter name of the + * tagged alias. This is the opposite of ucnv_openStandardNames, which + * returns the tagged alias given the canonical name. + *

+ * Example alias table:
+ * conv alias1 { STANDARD1 } alias2 { STANDARD1* } + *

+ * Result of ucnv_getStandardName("alias1", "STANDARD1") from example + * alias table:
+ * "conv" + * + * @return returns the canonical converter name; + * if a standard or alias name cannot be determined, + * then NULL is returned. The returned string is + * owned by the library. + * @see ucnv_getStandardName + * @stable ICU 2.4 + */ +U_CAPI const char * U_EXPORT2 +ucnv_getCanonicalName(const char *alias, const char *standard, UErrorCode *pErrorCode); + +/** + * Returns the current default converter name. If you want to open + * a default converter, you do not need to use this function. + * It is faster if you pass a NULL argument to ucnv_open the + * default converter. + * + * If U_CHARSET_IS_UTF8 is defined to 1 in utypes.h then this function + * always returns "UTF-8". + * + * @return returns the current default converter name. + * Storage owned by the library + * @see ucnv_setDefaultName + * @stable ICU 2.0 + */ +U_CAPI const char * U_EXPORT2 +ucnv_getDefaultName(void); + +#ifndef U_HIDE_SYSTEM_API +/** + * This function is not thread safe. DO NOT call this function when ANY ICU + * function is being used from more than one thread! This function sets the + * current default converter name. If this function needs to be called, it + * should be called during application initialization. Most of the time, the + * results from ucnv_getDefaultName() or ucnv_open with a NULL string argument + * is sufficient for your application. + * + * If U_CHARSET_IS_UTF8 is defined to 1 in utypes.h then this function + * does nothing. + * + * @param name the converter name to be the default (must be known by ICU). + * @see ucnv_getDefaultName + * @system + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_setDefaultName(const char *name); +#endif /* U_HIDE_SYSTEM_API */ + +/** + * Fixes the backslash character mismapping. For example, in SJIS, the backslash + * character in the ASCII portion is also used to represent the yen currency sign. + * When mapping from Unicode character 0x005C, it's unclear whether to map the + * character back to yen or backslash in SJIS. This function will take the input + * buffer and replace all the yen sign characters with backslash. This is necessary + * when the user tries to open a file with the input buffer on Windows. + * This function will test the converter to see whether such mapping is + * required. You can sometimes avoid using this function by using the correct version + * of Shift-JIS. + * + * @param cnv The converter representing the target codepage. + * @param source the input buffer to be fixed + * @param sourceLen the length of the input buffer + * @see ucnv_isAmbiguous + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_fixFileSeparator(const UConverter *cnv, UChar *source, int32_t sourceLen); + +/** + * Determines if the converter contains ambiguous mappings of the same + * character or not. + * @param cnv the converter to be tested + * @return true if the converter contains ambiguous mapping of the same + * character, false otherwise. + * @stable ICU 2.0 + */ +U_CAPI UBool U_EXPORT2 +ucnv_isAmbiguous(const UConverter *cnv); + +/** + * Sets the converter to use fallback mappings or not. + * Regardless of this flag, the converter will always use + * fallbacks from Unicode Private Use code points, as well as + * reverse fallbacks (to Unicode). + * For details see ".ucm File Format" + * in the Conversion Data chapter of the ICU User Guide: + * https://unicode-org.github.io/icu/userguide/conversion/data.html#ucm-file-format + * + * @param cnv The converter to set the fallback mapping usage on. + * @param usesFallback true if the user wants the converter to take advantage of the fallback + * mapping, false otherwise. + * @stable ICU 2.0 + * @see ucnv_usesFallback + */ +U_CAPI void U_EXPORT2 +ucnv_setFallback(UConverter *cnv, UBool usesFallback); + +/** + * Determines if the converter uses fallback mappings or not. + * This flag has restrictions, see ucnv_setFallback(). + * + * @param cnv The converter to be tested + * @return true if the converter uses fallback, false otherwise. + * @stable ICU 2.0 + * @see ucnv_setFallback + */ +U_CAPI UBool U_EXPORT2 +ucnv_usesFallback(const UConverter *cnv); + +/** + * Detects Unicode signature byte sequences at the start of the byte stream + * and returns the charset name of the indicated Unicode charset. + * NULL is returned when no Unicode signature is recognized. + * The number of bytes in the signature is output as well. + * + * The caller can ucnv_open() a converter using the charset name. + * The first code unit (UChar) from the start of the stream will be U+FEFF + * (the Unicode BOM/signature character) and can usually be ignored. + * + * For most Unicode charsets it is also possible to ignore the indicated + * number of initial stream bytes and start converting after them. + * However, there are stateful Unicode charsets (UTF-7 and BOCU-1) for which + * this will not work. Therefore, it is best to ignore the first output UChar + * instead of the input signature bytes. + *

+ * Usage: + * \snippet samples/ucnv/convsamp.cpp ucnv_detectUnicodeSignature + * + * @param source The source string in which the signature should be detected. + * @param sourceLength Length of the input string, or -1 if terminated with a NUL byte. + * @param signatureLength A pointer to int32_t to receive the number of bytes that make up the signature + * of the detected UTF. 0 if not detected. + * Can be a NULL pointer. + * @param pErrorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return The name of the encoding detected. NULL if encoding is not detected. + * @stable ICU 2.4 + */ +U_CAPI const char* U_EXPORT2 +ucnv_detectUnicodeSignature(const char* source, + int32_t sourceLength, + int32_t *signatureLength, + UErrorCode *pErrorCode); + +/** + * Returns the number of UChars held in the converter's internal state + * because more input is needed for completing the conversion. This function is + * useful for mapping semantics of ICU's converter interface to those of iconv, + * and this information is not needed for normal conversion. + * @param cnv The converter in which the input is held + * @param status ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return The number of UChars in the state. -1 if an error is encountered. + * @stable ICU 3.4 + */ +U_CAPI int32_t U_EXPORT2 +ucnv_fromUCountPending(const UConverter* cnv, UErrorCode* status); + +/** + * Returns the number of chars held in the converter's internal state + * because more input is needed for completing the conversion. This function is + * useful for mapping semantics of ICU's converter interface to those of iconv, + * and this information is not needed for normal conversion. + * @param cnv The converter in which the input is held as internal state + * @param status ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return The number of chars in the state. -1 if an error is encountered. + * @stable ICU 3.4 + */ +U_CAPI int32_t U_EXPORT2 +ucnv_toUCountPending(const UConverter* cnv, UErrorCode* status); + +/** + * Returns whether or not the charset of the converter has a fixed number of bytes + * per charset character. + * An example of this are converters that are of the type UCNV_SBCS or UCNV_DBCS. + * Another example is UTF-32 which is always 4 bytes per character. + * A Unicode code point may be represented by more than one UTF-8 or UTF-16 code unit + * but a UTF-32 converter encodes each code point with 4 bytes. + * Note: This method is not intended to be used to determine whether the charset has a + * fixed ratio of bytes to Unicode codes units for any particular Unicode encoding form. + * false is returned with the UErrorCode if error occurs or cnv is NULL. + * @param cnv The converter to be tested + * @param status ICU error code in/out parameter + * @return true if the converter is fixed-width + * @stable ICU 4.8 + */ +U_CAPI UBool U_EXPORT2 +ucnv_isFixedWidth(UConverter *cnv, UErrorCode *status); + +#endif + +#endif +/*_UCNV*/ diff --git a/intl/icu/source/common/unicode/ucnv_cb.h b/intl/icu/source/common/unicode/ucnv_cb.h new file mode 100644 index 0000000000..b4ef99208b --- /dev/null +++ b/intl/icu/source/common/unicode/ucnv_cb.h @@ -0,0 +1,164 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2000-2004, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** + * ucnv_cb.h: + * External APIs for the ICU's codeset conversion library + * Helena Shih + * + * Modification History: + * + * Date Name Description + */ + +/** + * \file + * \brief C API: UConverter functions to aid the writers of callbacks + * + *

Callback API for UConverter

+ * + * These functions are provided here for the convenience of the callback + * writer. If you are just looking for callback functions to use, please + * see ucnv_err.h. DO NOT call these functions directly when you are + * working with converters, unless your code has been called as a callback + * via ucnv_setFromUCallback or ucnv_setToUCallback !! + * + * A note about error codes and overflow. Unlike other ICU functions, + * these functions do not expect the error status to be U_ZERO_ERROR. + * Callbacks must be much more careful about their error codes. + * The error codes used here are in/out parameters, which should be passed + * back in the callback's error parameter. + * + * For example, if you call ucnv_cbfromUWriteBytes to write data out + * to the output codepage, it may return U_BUFFER_OVERFLOW_ERROR if + * the data did not fit in the target. But this isn't a failing error, + * in fact, ucnv_cbfromUWriteBytes may be called AGAIN with the error + * status still U_BUFFER_OVERFLOW_ERROR to attempt to write further bytes, + * which will also go into the internal overflow buffers. + * + * Concerning offsets, the 'offset' parameters here are relative to the start + * of SOURCE. For example, Suppose the string "ABCD" was being converted + * from Unicode into a codepage which doesn't have a mapping for 'B'. + * 'A' will be written out correctly, but + * The FromU Callback will be called on an unassigned character for 'B'. + * At this point, this is the state of the world: + * Target: A [..] [points after A] + * Source: A B [C] D [points to C - B has been consumed] + * 0 1 2 3 + * codePoint = "B" [the unassigned codepoint] + * + * Now, suppose a callback wants to write the substitution character '?' to + * the target. It calls ucnv_cbFromUWriteBytes() to write the ?. + * It should pass ZERO as the offset, because the offset as far as the + * callback is concerned is relative to the SOURCE pointer [which points + * before 'C'.] If the callback goes into the args and consumes 'C' also, + * it would call FromUWriteBytes with an offset of 1 (and advance the source + * pointer). + * + */ + +#ifndef UCNV_CB_H +#define UCNV_CB_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/ucnv.h" +#include "unicode/ucnv_err.h" + +/** + * ONLY used by FromU callback functions. + * Writes out the specified byte output bytes to the target byte buffer or to converter internal buffers. + * + * @param args callback fromUnicode arguments + * @param source source bytes to write + * @param length length of bytes to write + * @param offsetIndex the relative offset index from callback. + * @param err error status. If U_BUFFER_OVERFLOW is returned, then U_BUFFER_OVERFLOW must + * be returned to the user, because it means that not all data could be written into the target buffer, and some is + * in the converter error buffer. + * @see ucnv_cbFromUWriteSub + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_cbFromUWriteBytes (UConverterFromUnicodeArgs *args, + const char* source, + int32_t length, + int32_t offsetIndex, + UErrorCode * err); + +/** + * ONLY used by FromU callback functions. + * This function will write out the correct substitution character sequence + * to the target. + * + * @param args callback fromUnicode arguments + * @param offsetIndex the relative offset index from the current source pointer to be used + * @param err error status. If U_BUFFER_OVERFLOW is returned, then U_BUFFER_OVERFLOW must + * be returned to the user, because it means that not all data could be written into the target buffer, and some is + * in the converter error buffer. + * @see ucnv_cbFromUWriteBytes + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ucnv_cbFromUWriteSub (UConverterFromUnicodeArgs *args, + int32_t offsetIndex, + UErrorCode * err); + +/** + * ONLY used by fromU callback functions. + * This function will write out the error character(s) to the target UChar buffer. + * + * @param args callback fromUnicode arguments + * @param source pointer to pointer to first UChar to write [on exit: 1 after last UChar processed] + * @param sourceLimit pointer after last UChar to write + * @param offsetIndex the relative offset index from callback which will be set + * @param err error status U_BUFFER_OVERFLOW + * @see ucnv_cbToUWriteSub + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 ucnv_cbFromUWriteUChars(UConverterFromUnicodeArgs *args, + const UChar** source, + const UChar* sourceLimit, + int32_t offsetIndex, + UErrorCode * err); + +/** + * ONLY used by ToU callback functions. + * This function will write out the specified characters to the target + * UChar buffer. + * + * @param args callback toUnicode arguments + * @param source source string to write + * @param length the length of source string + * @param offsetIndex the relative offset index which will be written. + * @param err error status U_BUFFER_OVERFLOW + * @see ucnv_cbToUWriteSub + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 ucnv_cbToUWriteUChars (UConverterToUnicodeArgs *args, + const UChar* source, + int32_t length, + int32_t offsetIndex, + UErrorCode * err); + +/** + * ONLY used by ToU callback functions. + * This function will write out the Unicode substitution character (U+FFFD). + * + * @param args callback fromUnicode arguments + * @param offsetIndex the relative offset index from callback. + * @param err error status U_BUFFER_OVERFLOW + * @see ucnv_cbToUWriteUChars + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 ucnv_cbToUWriteSub (UConverterToUnicodeArgs *args, + int32_t offsetIndex, + UErrorCode * err); +#endif + +#endif diff --git a/intl/icu/source/common/unicode/ucnv_err.h b/intl/icu/source/common/unicode/ucnv_err.h new file mode 100644 index 0000000000..c743e5614f --- /dev/null +++ b/intl/icu/source/common/unicode/ucnv_err.h @@ -0,0 +1,465 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1999-2009, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** + * + * + * ucnv_err.h: + */ + +/** + * \file + * \brief C API: UConverter predefined error callbacks + * + *

Error Behaviour Functions

+ * Defines some error behaviour functions called by ucnv_{from,to}Unicode + * These are provided as part of ICU and many are stable, but they + * can also be considered only as an example of what can be done with + * callbacks. You may of course write your own. + * + * If you want to write your own, you may also find the functions from + * ucnv_cb.h useful when writing your own callbacks. + * + * These functions, although public, should NEVER be called directly. + * They should be used as parameters to the ucnv_setFromUCallback + * and ucnv_setToUCallback functions, to set the behaviour of a converter + * when it encounters ILLEGAL/UNMAPPED/INVALID sequences. + * + * usage example: 'STOP' doesn't need any context, but newContext + * could be set to something other than 'NULL' if needed. The available + * contexts in this header can modify the default behavior of the callback. + * + * \code + * UErrorCode err = U_ZERO_ERROR; + * UConverter *myConverter = ucnv_open("ibm-949", &err); + * const void *oldContext; + * UConverterFromUCallback oldAction; + * + * + * if (U_SUCCESS(err)) + * { + * ucnv_setFromUCallBack(myConverter, + * UCNV_FROM_U_CALLBACK_STOP, + * NULL, + * &oldAction, + * &oldContext, + * &status); + * } + * \endcode + * + * The code above tells "myConverter" to stop when it encounters an + * ILLEGAL/TRUNCATED/INVALID sequences when it is used to convert from + * Unicode -> Codepage. The behavior from Codepage to Unicode is not changed, + * and ucnv_setToUCallBack would need to be called in order to change + * that behavior too. + * + * Here is an example with a context: + * + * \code + * UErrorCode err = U_ZERO_ERROR; + * UConverter *myConverter = ucnv_open("ibm-949", &err); + * const void *oldContext; + * UConverterFromUCallback oldAction; + * + * + * if (U_SUCCESS(err)) + * { + * ucnv_setToUCallBack(myConverter, + * UCNV_TO_U_CALLBACK_SUBSTITUTE, + * UCNV_SUB_STOP_ON_ILLEGAL, + * &oldAction, + * &oldContext, + * &status); + * } + * \endcode + * + * The code above tells "myConverter" to stop when it encounters an + * ILLEGAL/TRUNCATED/INVALID sequences when it is used to convert from + * Codepage -> Unicode. Any unmapped and legal characters will be + * substituted to be the default substitution character. + */ + +#ifndef UCNV_ERR_H +#define UCNV_ERR_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +/** Forward declaring the UConverter structure. @stable ICU 2.0 */ +struct UConverter; + +/** @stable ICU 2.0 */ +typedef struct UConverter UConverter; + +/** + * FROM_U, TO_U context options for sub callback + * @stable ICU 2.0 + */ +#define UCNV_SUB_STOP_ON_ILLEGAL "i" + +/** + * FROM_U, TO_U context options for skip callback + * @stable ICU 2.0 + */ +#define UCNV_SKIP_STOP_ON_ILLEGAL "i" + +/** + * FROM_U_CALLBACK_ESCAPE context option to escape the code unit according to ICU (%UXXXX) + * @stable ICU 2.0 + */ +#define UCNV_ESCAPE_ICU NULL +/** + * FROM_U_CALLBACK_ESCAPE context option to escape the code unit according to JAVA (\\uXXXX) + * @stable ICU 2.0 + */ +#define UCNV_ESCAPE_JAVA "J" +/** + * FROM_U_CALLBACK_ESCAPE context option to escape the code unit according to C (\\uXXXX \\UXXXXXXXX) + * TO_U_CALLBACK_ESCAPE option to escape the character value according to C (\\xXXXX) + * @stable ICU 2.0 + */ +#define UCNV_ESCAPE_C "C" +/** + * FROM_U_CALLBACK_ESCAPE context option to escape the code unit according to XML Decimal escape \htmlonly(&#DDDD;)\endhtmlonly + * TO_U_CALLBACK_ESCAPE context option to escape the character value according to XML Decimal escape \htmlonly(&#DDDD;)\endhtmlonly + * @stable ICU 2.0 + */ +#define UCNV_ESCAPE_XML_DEC "D" +/** + * FROM_U_CALLBACK_ESCAPE context option to escape the code unit according to XML Hex escape \htmlonly(&#xXXXX;)\endhtmlonly + * TO_U_CALLBACK_ESCAPE context option to escape the character value according to XML Hex escape \htmlonly(&#xXXXX;)\endhtmlonly + * @stable ICU 2.0 + */ +#define UCNV_ESCAPE_XML_HEX "X" +/** + * FROM_U_CALLBACK_ESCAPE context option to escape the code unit according to Unicode (U+XXXXX) + * @stable ICU 2.0 + */ +#define UCNV_ESCAPE_UNICODE "U" + +/** + * FROM_U_CALLBACK_ESCAPE context option to escape the code unit according to CSS2 conventions (\\HH..H, that is, + * a backslash, 1..6 hex digits, and a space) + * @stable ICU 4.0 + */ +#define UCNV_ESCAPE_CSS2 "S" + +/** + * The process condition code to be used with the callbacks. + * Codes which are greater than UCNV_IRREGULAR should be + * passed on to any chained callbacks. + * @stable ICU 2.0 + */ +typedef enum { + UCNV_UNASSIGNED = 0, /**< The code point is unassigned. + The error code U_INVALID_CHAR_FOUND will be set. */ + UCNV_ILLEGAL = 1, /**< The code point is illegal. For example, + \\x81\\x2E is illegal in SJIS because \\x2E + is not a valid trail byte for the \\x81 + lead byte. + Also, starting with Unicode 3.0.1, non-shortest byte sequences + in UTF-8 (like \\xC1\\xA1 instead of \\x61 for U+0061) + are also illegal, not just irregular. + The error code U_ILLEGAL_CHAR_FOUND will be set. */ + UCNV_IRREGULAR = 2, /**< The codepoint is not a regular sequence in + the encoding. For example, \\xED\\xA0\\x80..\\xED\\xBF\\xBF + are irregular UTF-8 byte sequences for single surrogate + code points. + The error code U_INVALID_CHAR_FOUND will be set. */ + UCNV_RESET = 3, /**< The callback is called with this reason when a + 'reset' has occurred. Callback should reset all + state. */ + UCNV_CLOSE = 4, /**< Called when the converter is closed. The + callback should release any allocated memory.*/ + UCNV_CLONE = 5 /**< Called when ucnv_safeClone() is called on the + converter. the pointer available as the + 'context' is an alias to the original converters' + context pointer. If the context must be owned + by the new converter, the callback must clone + the data and call ucnv_setFromUCallback + (or setToUCallback) with the correct pointer. + @stable ICU 2.2 + */ +} UConverterCallbackReason; + + +/** + * The structure for the fromUnicode callback function parameter. + * @stable ICU 2.0 + */ +typedef struct { + uint16_t size; /**< The size of this struct. @stable ICU 2.0 */ + UBool flush; /**< The internal state of converter will be reset and data flushed if set to true. @stable ICU 2.0 */ + UConverter *converter; /**< Pointer to the converter that is opened and to which this struct is passed as an argument. @stable ICU 2.0 */ + const UChar *source; /**< Pointer to the source source buffer. @stable ICU 2.0 */ + const UChar *sourceLimit; /**< Pointer to the limit (end + 1) of source buffer. @stable ICU 2.0 */ + char *target; /**< Pointer to the target buffer. @stable ICU 2.0 */ + const char *targetLimit; /**< Pointer to the limit (end + 1) of target buffer. @stable ICU 2.0 */ + int32_t *offsets; /**< Pointer to the buffer that receives the offsets. *offset = blah ; offset++;. @stable ICU 2.0 */ +} UConverterFromUnicodeArgs; + + +/** + * The structure for the toUnicode callback function parameter. + * @stable ICU 2.0 + */ +typedef struct { + uint16_t size; /**< The size of this struct @stable ICU 2.0 */ + UBool flush; /**< The internal state of converter will be reset and data flushed if set to true. @stable ICU 2.0 */ + UConverter *converter; /**< Pointer to the converter that is opened and to which this struct is passed as an argument. @stable ICU 2.0 */ + const char *source; /**< Pointer to the source source buffer. @stable ICU 2.0 */ + const char *sourceLimit; /**< Pointer to the limit (end + 1) of source buffer. @stable ICU 2.0 */ + UChar *target; /**< Pointer to the target buffer. @stable ICU 2.0 */ + const UChar *targetLimit; /**< Pointer to the limit (end + 1) of target buffer. @stable ICU 2.0 */ + int32_t *offsets; /**< Pointer to the buffer that receives the offsets. *offset = blah ; offset++;. @stable ICU 2.0 */ +} UConverterToUnicodeArgs; + + +/** + * DO NOT CALL THIS FUNCTION DIRECTLY! + * This From Unicode callback STOPS at the ILLEGAL_SEQUENCE, + * returning the error code back to the caller immediately. + * + * @param context Pointer to the callback's private data + * @param fromUArgs Information about the conversion in progress + * @param codeUnits Points to 'length' UChars of the concerned Unicode sequence + * @param length Size (in bytes) of the concerned codepage sequence + * @param codePoint Single UChar32 (UTF-32) containing the concerend Unicode codepoint. + * @param reason Defines the reason the callback was invoked + * @param err This should always be set to a failure status prior to calling. + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 UCNV_FROM_U_CALLBACK_STOP ( + const void *context, + UConverterFromUnicodeArgs *fromUArgs, + const UChar* codeUnits, + int32_t length, + UChar32 codePoint, + UConverterCallbackReason reason, + UErrorCode * err); + + + +/** + * DO NOT CALL THIS FUNCTION DIRECTLY! + * This To Unicode callback STOPS at the ILLEGAL_SEQUENCE, + * returning the error code back to the caller immediately. + * + * @param context Pointer to the callback's private data + * @param toUArgs Information about the conversion in progress + * @param codeUnits Points to 'length' bytes of the concerned codepage sequence + * @param length Size (in bytes) of the concerned codepage sequence + * @param reason Defines the reason the callback was invoked + * @param err This should always be set to a failure status prior to calling. + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 UCNV_TO_U_CALLBACK_STOP ( + const void *context, + UConverterToUnicodeArgs *toUArgs, + const char* codeUnits, + int32_t length, + UConverterCallbackReason reason, + UErrorCode * err); + +/** + * DO NOT CALL THIS FUNCTION DIRECTLY! + * This From Unicode callback skips any ILLEGAL_SEQUENCE, or + * skips only UNASSIGNED_SEQUENCE depending on the context parameter + * simply ignoring those characters. + * + * @param context The function currently recognizes the callback options: + * UCNV_SKIP_STOP_ON_ILLEGAL: STOPS at the ILLEGAL_SEQUENCE, + * returning the error code back to the caller immediately. + * NULL: Skips any ILLEGAL_SEQUENCE + * @param fromUArgs Information about the conversion in progress + * @param codeUnits Points to 'length' UChars of the concerned Unicode sequence + * @param length Size (in bytes) of the concerned codepage sequence + * @param codePoint Single UChar32 (UTF-32) containing the concerend Unicode codepoint. + * @param reason Defines the reason the callback was invoked + * @param err Return value will be set to success if the callback was handled, + * otherwise this value will be set to a failure status. + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 UCNV_FROM_U_CALLBACK_SKIP ( + const void *context, + UConverterFromUnicodeArgs *fromUArgs, + const UChar* codeUnits, + int32_t length, + UChar32 codePoint, + UConverterCallbackReason reason, + UErrorCode * err); + +/** + * DO NOT CALL THIS FUNCTION DIRECTLY! + * This From Unicode callback will Substitute the ILLEGAL SEQUENCE, or + * UNASSIGNED_SEQUENCE depending on context parameter, with the + * current substitution string for the converter. This is the default + * callback. + * + * @param context The function currently recognizes the callback options: + * UCNV_SUB_STOP_ON_ILLEGAL: STOPS at the ILLEGAL_SEQUENCE, + * returning the error code back to the caller immediately. + * NULL: Substitutes any ILLEGAL_SEQUENCE + * @param fromUArgs Information about the conversion in progress + * @param codeUnits Points to 'length' UChars of the concerned Unicode sequence + * @param length Size (in bytes) of the concerned codepage sequence + * @param codePoint Single UChar32 (UTF-32) containing the concerend Unicode codepoint. + * @param reason Defines the reason the callback was invoked + * @param err Return value will be set to success if the callback was handled, + * otherwise this value will be set to a failure status. + * @see ucnv_setSubstChars + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 UCNV_FROM_U_CALLBACK_SUBSTITUTE ( + const void *context, + UConverterFromUnicodeArgs *fromUArgs, + const UChar* codeUnits, + int32_t length, + UChar32 codePoint, + UConverterCallbackReason reason, + UErrorCode * err); + +/** + * DO NOT CALL THIS FUNCTION DIRECTLY! + * This From Unicode callback will Substitute the ILLEGAL SEQUENCE with the + * hexadecimal representation of the illegal codepoints + * + * @param context The function currently recognizes the callback options: + *
    + *
  • UCNV_ESCAPE_ICU: Substitutes the ILLEGAL SEQUENCE with the hexadecimal + * representation in the format %UXXXX, e.g. "%uFFFE%u00AC%uC8FE"). + * In the Event the converter doesn't support the characters {%,U}[A-F][0-9], + * it will substitute the illegal sequence with the substitution characters. + * Note that codeUnit(32bit int eg: unit of a surrogate pair) is represented as + * %UD84D%UDC56
    • + *
  • UCNV_ESCAPE_JAVA: Substitutes the ILLEGAL SEQUENCE with the hexadecimal + * representation in the format \\uXXXX, e.g. "\\uFFFE\\u00AC\\uC8FE"). + * In the Event the converter doesn't support the characters {\,u}[A-F][0-9], + * it will substitute the illegal sequence with the substitution characters. + * Note that codeUnit(32bit int eg: unit of a surrogate pair) is represented as + * \\uD84D\\uDC56
    • + *
  • UCNV_ESCAPE_C: Substitutes the ILLEGAL SEQUENCE with the hexadecimal + * representation in the format \\uXXXX, e.g. "\\uFFFE\\u00AC\\uC8FE"). + * In the Event the converter doesn't support the characters {\,u,U}[A-F][0-9], + * it will substitute the illegal sequence with the substitution characters. + * Note that codeUnit(32bit int eg: unit of a surrogate pair) is represented as + * \\U00023456
    • + *
  • UCNV_ESCAPE_XML_DEC: Substitutes the ILLEGAL SEQUENCE with the decimal + * representation in the format \htmlonly&#DDDDDDDD;, e.g. "&#65534;&#172;&#51454;")\endhtmlonly. + * In the Event the converter doesn't support the characters {&,#}[0-9], + * it will substitute the illegal sequence with the substitution characters. + * Note that codeUnit(32bit int eg: unit of a surrogate pair) is represented as + * &#144470; and Zero padding is ignored.
    • + *
  • UCNV_ESCAPE_XML_HEX:Substitutes the ILLEGAL SEQUENCE with the decimal + * representation in the format \htmlonly&#xXXXX; e.g. "&#xFFFE;&#x00AC;&#xC8FE;")\endhtmlonly. + * In the Event the converter doesn't support the characters {&,#,x}[0-9], + * it will substitute the illegal sequence with the substitution characters. + * Note that codeUnit(32bit int eg: unit of a surrogate pair) is represented as + * \htmlonly&#x23456;\endhtmlonly
    • + *
+ * @param fromUArgs Information about the conversion in progress + * @param codeUnits Points to 'length' UChars of the concerned Unicode sequence + * @param length Size (in bytes) of the concerned codepage sequence + * @param codePoint Single UChar32 (UTF-32) containing the concerend Unicode codepoint. + * @param reason Defines the reason the callback was invoked + * @param err Return value will be set to success if the callback was handled, + * otherwise this value will be set to a failure status. + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 UCNV_FROM_U_CALLBACK_ESCAPE ( + const void *context, + UConverterFromUnicodeArgs *fromUArgs, + const UChar* codeUnits, + int32_t length, + UChar32 codePoint, + UConverterCallbackReason reason, + UErrorCode * err); + + +/** + * DO NOT CALL THIS FUNCTION DIRECTLY! + * This To Unicode callback skips any ILLEGAL_SEQUENCE, or + * skips only UNASSIGNED_SEQUENCE depending on the context parameter + * simply ignoring those characters. + * + * @param context The function currently recognizes the callback options: + * UCNV_SKIP_STOP_ON_ILLEGAL: STOPS at the ILLEGAL_SEQUENCE, + * returning the error code back to the caller immediately. + * NULL: Skips any ILLEGAL_SEQUENCE + * @param toUArgs Information about the conversion in progress + * @param codeUnits Points to 'length' bytes of the concerned codepage sequence + * @param length Size (in bytes) of the concerned codepage sequence + * @param reason Defines the reason the callback was invoked + * @param err Return value will be set to success if the callback was handled, + * otherwise this value will be set to a failure status. + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 UCNV_TO_U_CALLBACK_SKIP ( + const void *context, + UConverterToUnicodeArgs *toUArgs, + const char* codeUnits, + int32_t length, + UConverterCallbackReason reason, + UErrorCode * err); + +/** + * DO NOT CALL THIS FUNCTION DIRECTLY! + * This To Unicode callback will Substitute the ILLEGAL SEQUENCE,or + * UNASSIGNED_SEQUENCE depending on context parameter, with the + * Unicode substitution character, U+FFFD. + * + * @param context The function currently recognizes the callback options: + * UCNV_SUB_STOP_ON_ILLEGAL: STOPS at the ILLEGAL_SEQUENCE, + * returning the error code back to the caller immediately. + * NULL: Substitutes any ILLEGAL_SEQUENCE + * @param toUArgs Information about the conversion in progress + * @param codeUnits Points to 'length' bytes of the concerned codepage sequence + * @param length Size (in bytes) of the concerned codepage sequence + * @param reason Defines the reason the callback was invoked + * @param err Return value will be set to success if the callback was handled, + * otherwise this value will be set to a failure status. + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 UCNV_TO_U_CALLBACK_SUBSTITUTE ( + const void *context, + UConverterToUnicodeArgs *toUArgs, + const char* codeUnits, + int32_t length, + UConverterCallbackReason reason, + UErrorCode * err); + +/** + * DO NOT CALL THIS FUNCTION DIRECTLY! + * This To Unicode callback will Substitute the ILLEGAL SEQUENCE with the + * hexadecimal representation of the illegal bytes + * (in the format %XNN, e.g. "%XFF%X0A%XC8%X03"). + * + * @param context This function currently recognizes the callback options: + * UCNV_ESCAPE_ICU, UCNV_ESCAPE_JAVA, UCNV_ESCAPE_C, UCNV_ESCAPE_XML_DEC, + * UCNV_ESCAPE_XML_HEX and UCNV_ESCAPE_UNICODE. + * @param toUArgs Information about the conversion in progress + * @param codeUnits Points to 'length' bytes of the concerned codepage sequence + * @param length Size (in bytes) of the concerned codepage sequence + * @param reason Defines the reason the callback was invoked + * @param err Return value will be set to success if the callback was handled, + * otherwise this value will be set to a failure status. + * @stable ICU 2.0 + */ + +U_CAPI void U_EXPORT2 UCNV_TO_U_CALLBACK_ESCAPE ( + const void *context, + UConverterToUnicodeArgs *toUArgs, + const char* codeUnits, + int32_t length, + UConverterCallbackReason reason, + UErrorCode * err); + +#endif + +#endif + +/*UCNV_ERR_H*/ diff --git a/intl/icu/source/common/unicode/ucnvsel.h b/intl/icu/source/common/unicode/ucnvsel.h new file mode 100644 index 0000000000..9373ec951b --- /dev/null +++ b/intl/icu/source/common/unicode/ucnvsel.h @@ -0,0 +1,193 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2008-2011, International Business Machines +* Corporation, Google and others. All Rights Reserved. +* +******************************************************************************* +*/ +/* + * Author : eldawy@google.com (Mohamed Eldawy) + * ucnvsel.h + * + * Purpose: To generate a list of encodings capable of handling + * a given Unicode text + * + * Started 09-April-2008 + */ + +#ifndef __ICU_UCNV_SEL_H__ +#define __ICU_UCNV_SEL_H__ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_CONVERSION + +#include "unicode/uset.h" +#include "unicode/utf16.h" +#include "unicode/uenum.h" +#include "unicode/ucnv.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +/** + * \file + * \brief C API: Encoding/charset encoding selector + * + * A converter selector is built with a set of encoding/charset names + * and given an input string returns the set of names of the + * corresponding converters which can convert the string. + * + * A converter selector can be serialized into a buffer and reopened + * from the serialized form. + */ + +struct UConverterSelector; +/** + * @{ + * Typedef for selector data structure. + */ +typedef struct UConverterSelector UConverterSelector; +/** @} */ + +/** + * Open a selector. + * If converterListSize is 0, build for all available converters. + * If excludedCodePoints is NULL, don't exclude any code points. + * + * @param converterList a pointer to encoding names needed to be involved. + * Can be NULL if converterListSize==0. + * The list and the names will be cloned, and the caller + * retains ownership of the original. + * @param converterListSize number of encodings in above list. + * If 0, builds a selector for all available converters. + * @param excludedCodePoints a set of code points to be excluded from consideration. + * That is, excluded code points in a string do not change + * the selection result. (They might be handled by a callback.) + * Use NULL to exclude nothing. + * @param whichSet what converter set to use? Use this to determine whether + * to consider only roundtrip mappings or also fallbacks. + * @param status an in/out ICU UErrorCode + * @return the new selector + * + * @stable ICU 4.2 + */ +U_CAPI UConverterSelector* U_EXPORT2 +ucnvsel_open(const char* const* converterList, int32_t converterListSize, + const USet* excludedCodePoints, + const UConverterUnicodeSet whichSet, UErrorCode* status); + +/** + * Closes a selector. + * If any Enumerations were returned by ucnv_select*, they become invalid. + * They can be closed before or after calling ucnv_closeSelector, + * but should never be used after the selector is closed. + * + * @see ucnv_selectForString + * @see ucnv_selectForUTF8 + * + * @param sel selector to close + * + * @stable ICU 4.2 + */ +U_CAPI void U_EXPORT2 +ucnvsel_close(UConverterSelector *sel); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUConverterSelectorPointer + * "Smart pointer" class, closes a UConverterSelector via ucnvsel_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUConverterSelectorPointer, UConverterSelector, ucnvsel_close); + +U_NAMESPACE_END + +#endif + +/** + * Open a selector from its serialized form. + * The buffer must remain valid and unchanged for the lifetime of the selector. + * This is much faster than creating a selector from scratch. + * Using a serialized form from a different machine (endianness/charset) is supported. + * + * @param buffer pointer to the serialized form of a converter selector; + * must be 32-bit-aligned + * @param length the capacity of this buffer (can be equal to or larger than + * the actual data length) + * @param status an in/out ICU UErrorCode + * @return the new selector + * + * @stable ICU 4.2 + */ +U_CAPI UConverterSelector* U_EXPORT2 +ucnvsel_openFromSerialized(const void* buffer, int32_t length, UErrorCode* status); + +/** + * Serialize a selector into a linear buffer. + * The serialized form is portable to different machines. + * + * @param sel selector to consider + * @param buffer pointer to 32-bit-aligned memory to be filled with the + * serialized form of this converter selector + * @param bufferCapacity the capacity of this buffer + * @param status an in/out ICU UErrorCode + * @return the required buffer capacity to hold serialize data (even if the call fails + * with a U_BUFFER_OVERFLOW_ERROR, it will return the required capacity) + * + * @stable ICU 4.2 + */ +U_CAPI int32_t U_EXPORT2 +ucnvsel_serialize(const UConverterSelector* sel, + void* buffer, int32_t bufferCapacity, UErrorCode* status); + +/** + * Select converters that can map all characters in a UTF-16 string, + * ignoring the excluded code points. + * + * @param sel a selector + * @param s UTF-16 string + * @param length length of the string, or -1 if NUL-terminated + * @param status an in/out ICU UErrorCode + * @return an enumeration containing encoding names. + * The returned encoding names and their order will be the same as + * supplied when building the selector. + * + * @stable ICU 4.2 + */ +U_CAPI UEnumeration * U_EXPORT2 +ucnvsel_selectForString(const UConverterSelector* sel, + const UChar *s, int32_t length, UErrorCode *status); + +/** + * Select converters that can map all characters in a UTF-8 string, + * ignoring the excluded code points. + * + * @param sel a selector + * @param s UTF-8 string + * @param length length of the string, or -1 if NUL-terminated + * @param status an in/out ICU UErrorCode + * @return an enumeration containing encoding names. + * The returned encoding names and their order will be the same as + * supplied when building the selector. + * + * @stable ICU 4.2 + */ +U_CAPI UEnumeration * U_EXPORT2 +ucnvsel_selectForUTF8(const UConverterSelector* sel, + const char *s, int32_t length, UErrorCode *status); + +#endif /* !UCONFIG_NO_CONVERSION */ + +#endif /* __ICU_UCNV_SEL_H__ */ diff --git a/intl/icu/source/common/unicode/uconfig.h b/intl/icu/source/common/unicode/uconfig.h new file mode 100644 index 0000000000..3818ca02ef --- /dev/null +++ b/intl/icu/source/common/unicode/uconfig.h @@ -0,0 +1,466 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 2002-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* file name: uconfig.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002sep19 +* created by: Markus W. Scherer +*/ + +#ifndef __UCONFIG_H__ +#define __UCONFIG_H__ + + +/*! + * \file + * \brief User-configurable settings + * + * Miscellaneous switches: + * + * A number of macros affect a variety of minor aspects of ICU. + * Most of them used to be defined elsewhere (e.g., in utypes.h or platform.h) + * and moved here to make them easier to find. + * + * Switches for excluding parts of ICU library code modules: + * + * Changing these macros allows building partial, smaller libraries for special purposes. + * By default, all modules are built. + * The switches are fairly coarse, controlling large modules. + * Basic services cannot be turned off. + * + * Building with any of these options does not guarantee that the + * ICU build process will completely work. It is recommended that + * the ICU libraries and data be built using the normal build. + * At that time you should remove the data used by those services. + * After building the ICU data library, you should rebuild the ICU + * libraries with these switches customized to your needs. + * + * @stable ICU 2.4 + */ + +/** + * If this switch is defined, ICU will attempt to load a header file named "uconfig_local.h" + * prior to determining default settings for uconfig variables. + * + * @internal ICU 4.0 + */ +#if defined(UCONFIG_USE_LOCAL) +#include "uconfig_local.h" +#endif + +/** + * \def U_DEBUG + * Determines whether to include debugging code. + * Automatically set on Windows, but most compilers do not have + * related predefined macros. + * @internal + */ +#ifdef U_DEBUG + /* Use the predefined value. */ +#elif defined(_DEBUG) + /* + * _DEBUG is defined by Visual Studio debug compilation. + * Do *not* test for its NDEBUG macro: It is an orthogonal macro + * which disables assert(). + */ +# define U_DEBUG 1 +# else +# define U_DEBUG 0 +#endif + +/** + * Determines whether to enable auto cleanup of libraries. + * @internal + */ +#ifndef UCLN_NO_AUTO_CLEANUP +#define UCLN_NO_AUTO_CLEANUP 1 +#endif + +/** + * \def U_DISABLE_RENAMING + * Determines whether to disable renaming or not. + * @internal + */ +#ifndef U_DISABLE_RENAMING +#define U_DISABLE_RENAMING 0 +#endif + +/** + * \def U_NO_DEFAULT_INCLUDE_UTF_HEADERS + * Determines whether utypes.h includes utf.h, utf8.h, utf16.h and utf_old.h. + * utypes.h includes those headers if this macro is defined to 0. + * Otherwise, each those headers must be included explicitly when using one of their macros. + * Defaults to 0 for backward compatibility, except inside ICU. + * @stable ICU 49 + */ +#ifdef U_NO_DEFAULT_INCLUDE_UTF_HEADERS + /* Use the predefined value. */ +#elif defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || defined(U_I18N_IMPLEMENTATION) || \ + defined(U_IO_IMPLEMENTATION) || defined(U_LAYOUT_IMPLEMENTATION) || defined(U_LAYOUTEX_IMPLEMENTATION) || \ + defined(U_TOOLUTIL_IMPLEMENTATION) +# define U_NO_DEFAULT_INCLUDE_UTF_HEADERS 1 +#else +# define U_NO_DEFAULT_INCLUDE_UTF_HEADERS 0 +#endif + +/** + * \def U_OVERRIDE_CXX_ALLOCATION + * Determines whether to override new and delete. + * ICU is normally built such that all of its C++ classes, via their UMemory base, + * override operators new and delete to use its internal, customizable, + * non-exception-throwing memory allocation functions. (Default value 1 for this macro.) + * + * This is especially important when the application and its libraries use multiple heaps. + * For example, on Windows, this allows the ICU DLL to be used by + * applications that statically link the C Runtime library. + * + * @stable ICU 2.2 + */ +#ifndef U_OVERRIDE_CXX_ALLOCATION +#define U_OVERRIDE_CXX_ALLOCATION 1 +#endif + +/** + * \def U_ENABLE_TRACING + * Determines whether to enable tracing. + * @internal + */ +#ifndef U_ENABLE_TRACING +#define U_ENABLE_TRACING 0 +#endif + +/** + * \def UCONFIG_ENABLE_PLUGINS + * Determines whether to enable ICU plugins. + * @internal + */ +#ifndef UCONFIG_ENABLE_PLUGINS +#define UCONFIG_ENABLE_PLUGINS 0 +#endif + +/** + * \def U_ENABLE_DYLOAD + * Whether to enable Dynamic loading in ICU. + * @internal + */ +#ifndef U_ENABLE_DYLOAD +#define U_ENABLE_DYLOAD 1 +#endif + +/** + * \def U_CHECK_DYLOAD + * Whether to test Dynamic loading as an OS capability. + * @internal + */ +#ifndef U_CHECK_DYLOAD +#define U_CHECK_DYLOAD 1 +#endif + +/** + * \def U_DEFAULT_SHOW_DRAFT + * Do we allow ICU users to use the draft APIs by default? + * @internal + */ +#ifndef U_DEFAULT_SHOW_DRAFT +#define U_DEFAULT_SHOW_DRAFT 1 +#endif + +/*===========================================================================*/ +/* Custom icu entry point renaming */ +/*===========================================================================*/ + +/** + * \def U_HAVE_LIB_SUFFIX + * 1 if a custom library suffix is set. + * @internal + */ +#ifdef U_HAVE_LIB_SUFFIX + /* Use the predefined value. */ +#elif defined(U_LIB_SUFFIX_C_NAME) || defined(U_IN_DOXYGEN) +# define U_HAVE_LIB_SUFFIX 1 +#endif + +/** + * \def U_LIB_SUFFIX_C_NAME_STRING + * Defines the library suffix as a string with C syntax. + * @internal + */ +#ifdef U_LIB_SUFFIX_C_NAME_STRING + /* Use the predefined value. */ +#elif defined(U_LIB_SUFFIX_C_NAME) +# define CONVERT_TO_STRING(s) #s +# define U_LIB_SUFFIX_C_NAME_STRING CONVERT_TO_STRING(U_LIB_SUFFIX_C_NAME) +#else +# define U_LIB_SUFFIX_C_NAME_STRING "" +#endif + +/* common/i18n library switches --------------------------------------------- */ + +/** + * \def UCONFIG_ONLY_COLLATION + * This switch turns off modules that are not needed for collation. + * + * It does not turn off legacy conversion because that is necessary + * for ICU to work on EBCDIC platforms (for the default converter). + * If you want "only collation" and do not build for EBCDIC, + * then you can define UCONFIG_NO_CONVERSION or UCONFIG_NO_LEGACY_CONVERSION to 1 as well. + * + * @stable ICU 2.4 + */ +#ifndef UCONFIG_ONLY_COLLATION +# define UCONFIG_ONLY_COLLATION 0 +#endif + +#if UCONFIG_ONLY_COLLATION + /* common library */ +# define UCONFIG_NO_BREAK_ITERATION 1 +# define UCONFIG_NO_IDNA 1 + + /* i18n library */ +# if UCONFIG_NO_COLLATION +# error Contradictory collation switches in uconfig.h. +# endif +# define UCONFIG_NO_FORMATTING 1 +# define UCONFIG_NO_TRANSLITERATION 1 +# define UCONFIG_NO_REGULAR_EXPRESSIONS 1 +#endif + +/* common library switches -------------------------------------------------- */ + +/** + * \def UCONFIG_NO_FILE_IO + * This switch turns off all file access in the common library + * where file access is only used for data loading. + * ICU data must then be provided in the form of a data DLL (or with an + * equivalent way to link to the data residing in an executable, + * as in building a combined library with both the common library's code and + * the data), or via udata_setCommonData(). + * Application data must be provided via udata_setAppData() or by using + * "open" functions that take pointers to data, for example ucol_openBinary(). + * + * File access is not used at all in the i18n library. + * + * File access cannot be turned off for the icuio library or for the ICU + * test suites and ICU tools. + * + * @stable ICU 3.6 + */ +#ifndef UCONFIG_NO_FILE_IO +# define UCONFIG_NO_FILE_IO 0 +#endif + +#if UCONFIG_NO_FILE_IO && defined(U_TIMEZONE_FILES_DIR) +# error Contradictory file io switches in uconfig.h. +#endif + +/** + * \def UCONFIG_NO_CONVERSION + * ICU will not completely build (compiling the tools fails) with this + * switch turned on. + * This switch turns off all converters. + * + * You may want to use this together with U_CHARSET_IS_UTF8 defined to 1 + * in utypes.h if char* strings in your environment are always in UTF-8. + * + * @stable ICU 3.2 + * @see U_CHARSET_IS_UTF8 + */ +#ifndef UCONFIG_NO_CONVERSION +# define UCONFIG_NO_CONVERSION 0 +#endif + +#if UCONFIG_NO_CONVERSION +# define UCONFIG_NO_LEGACY_CONVERSION 1 +#endif + +/** + * \def UCONFIG_ONLY_HTML_CONVERSION + * This switch turns off all of the converters NOT listed in + * the HTML encoding standard: + * http://www.w3.org/TR/encoding/#names-and-labels + * + * This is not possible on EBCDIC platforms + * because they need ibm-37 or ibm-1047 default converters. + * + * @stable ICU 55 + */ +#ifndef UCONFIG_ONLY_HTML_CONVERSION +# define UCONFIG_ONLY_HTML_CONVERSION 0 +#endif + +/** + * \def UCONFIG_NO_LEGACY_CONVERSION + * This switch turns off all converters except for + * - Unicode charsets (UTF-7/8/16/32, CESU-8, SCSU, BOCU-1) + * - US-ASCII + * - ISO-8859-1 + * + * Turning off legacy conversion is not possible on EBCDIC platforms + * because they need ibm-37 or ibm-1047 default converters. + * + * @stable ICU 2.4 + */ +#ifndef UCONFIG_NO_LEGACY_CONVERSION +# define UCONFIG_NO_LEGACY_CONVERSION 0 +#endif + +/** + * \def UCONFIG_NO_NORMALIZATION + * This switch turns off normalization. + * It implies turning off several other services as well, for example + * collation and IDNA. + * + * @stable ICU 2.6 + */ +#ifndef UCONFIG_NO_NORMALIZATION +# define UCONFIG_NO_NORMALIZATION 0 +#endif + +/** + * \def UCONFIG_USE_ML_PHRASE_BREAKING + * This switch turns on BudouX ML phrase-based line breaking, rather than using the dictionary. + * + * @internal + */ +#ifndef UCONFIG_USE_ML_PHRASE_BREAKING +# define UCONFIG_USE_ML_PHRASE_BREAKING 0 +#endif + +#if UCONFIG_NO_NORMALIZATION + /* common library */ + /* ICU 50 CJK dictionary BreakIterator uses normalization */ +# define UCONFIG_NO_BREAK_ITERATION 1 + /* IDNA (UTS #46) is implemented via normalization */ +# define UCONFIG_NO_IDNA 1 + + /* i18n library */ +# if UCONFIG_ONLY_COLLATION +# error Contradictory collation switches in uconfig.h. +# endif +# define UCONFIG_NO_COLLATION 1 +# define UCONFIG_NO_TRANSLITERATION 1 +#endif + +/** + * \def UCONFIG_NO_BREAK_ITERATION + * This switch turns off break iteration. + * + * @stable ICU 2.4 + */ +#ifndef UCONFIG_NO_BREAK_ITERATION +# define UCONFIG_NO_BREAK_ITERATION 0 +#endif + +/** + * \def UCONFIG_NO_IDNA + * This switch turns off IDNA. + * + * @stable ICU 2.6 + */ +#ifndef UCONFIG_NO_IDNA +# define UCONFIG_NO_IDNA 0 +#endif + +/** + * \def UCONFIG_MSGPAT_DEFAULT_APOSTROPHE_MODE + * Determines the default UMessagePatternApostropheMode. + * See the documentation for that enum. + * + * @stable ICU 4.8 + */ +#ifndef UCONFIG_MSGPAT_DEFAULT_APOSTROPHE_MODE +# define UCONFIG_MSGPAT_DEFAULT_APOSTROPHE_MODE UMSGPAT_APOS_DOUBLE_OPTIONAL +#endif + +/** + * \def UCONFIG_USE_WINDOWS_LCID_MAPPING_API + * On platforms where U_PLATFORM_HAS_WIN32_API is true, this switch determines + * if the Windows platform APIs are used for LCID<->Locale Name conversions. + * Otherwise, only the built-in ICU tables are used. + * + * @internal ICU 64 + */ +#ifndef UCONFIG_USE_WINDOWS_LCID_MAPPING_API +# define UCONFIG_USE_WINDOWS_LCID_MAPPING_API 1 +#endif + +/* i18n library switches ---------------------------------------------------- */ + +/** + * \def UCONFIG_NO_COLLATION + * This switch turns off collation and collation-based string search. + * + * @stable ICU 2.4 + */ +#ifndef UCONFIG_NO_COLLATION +# define UCONFIG_NO_COLLATION 0 +#endif + +/** + * \def UCONFIG_NO_FORMATTING + * This switch turns off formatting and calendar/timezone services. + * + * @stable ICU 2.4 + */ +#ifndef UCONFIG_NO_FORMATTING +# define UCONFIG_NO_FORMATTING 0 +#endif + +/** + * \def UCONFIG_NO_TRANSLITERATION + * This switch turns off transliteration. + * + * @stable ICU 2.4 + */ +#ifndef UCONFIG_NO_TRANSLITERATION +# define UCONFIG_NO_TRANSLITERATION 0 +#endif + +/** + * \def UCONFIG_NO_REGULAR_EXPRESSIONS + * This switch turns off regular expressions. + * + * @stable ICU 2.4 + */ +#ifndef UCONFIG_NO_REGULAR_EXPRESSIONS +# define UCONFIG_NO_REGULAR_EXPRESSIONS 0 +#endif + +/** + * \def UCONFIG_NO_SERVICE + * This switch turns off service registration. + * + * @stable ICU 3.2 + */ +#ifndef UCONFIG_NO_SERVICE +# define UCONFIG_NO_SERVICE 0 +#endif + +/** + * \def UCONFIG_HAVE_PARSEALLINPUT + * This switch turns on the "parse all input" attribute. Binary incompatible. + * + * @internal + */ +#ifndef UCONFIG_HAVE_PARSEALLINPUT +# define UCONFIG_HAVE_PARSEALLINPUT 1 +#endif + +/** + * \def UCONFIG_NO_FILTERED_BREAK_ITERATION + * This switch turns off filtered break iteration code. + * + * @internal + */ +#ifndef UCONFIG_NO_FILTERED_BREAK_ITERATION +# define UCONFIG_NO_FILTERED_BREAK_ITERATION 0 +#endif + +#endif // __UCONFIG_H__ diff --git a/intl/icu/source/common/unicode/ucpmap.h b/intl/icu/source/common/unicode/ucpmap.h new file mode 100644 index 0000000000..a740bd160f --- /dev/null +++ b/intl/icu/source/common/unicode/ucpmap.h @@ -0,0 +1,158 @@ +// © 2018 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// ucpmap.h +// created: 2018sep03 Markus W. Scherer + +#ifndef __UCPMAP_H__ +#define __UCPMAP_H__ + +#include "unicode/utypes.h" + +U_CDECL_BEGIN + +/** + * \file + * \brief C API: This file defines an abstract map from Unicode code points to integer values. + * + * @see UCPMap + * @see UCPTrie + * @see UMutableCPTrie + */ + +/** + * Abstract map from Unicode code points (U+0000..U+10FFFF) to integer values. + * + * @see UCPTrie + * @see UMutableCPTrie + * @stable ICU 63 + */ +typedef struct UCPMap UCPMap; + +/** + * Selectors for how ucpmap_getRange() etc. should report value ranges overlapping with surrogates. + * Most users should use UCPMAP_RANGE_NORMAL. + * + * @see ucpmap_getRange + * @see ucptrie_getRange + * @see umutablecptrie_getRange + * @stable ICU 63 + */ +enum UCPMapRangeOption { + /** + * ucpmap_getRange() enumerates all same-value ranges as stored in the map. + * Most users should use this option. + * @stable ICU 63 + */ + UCPMAP_RANGE_NORMAL, + /** + * ucpmap_getRange() enumerates all same-value ranges as stored in the map, + * except that lead surrogates (U+D800..U+DBFF) are treated as having the + * surrogateValue, which is passed to getRange() as a separate parameter. + * The surrogateValue is not transformed via filter(). + * See U_IS_LEAD(c). + * + * Most users should use UCPMAP_RANGE_NORMAL instead. + * + * This option is useful for maps that map surrogate code *units* to + * special values optimized for UTF-16 string processing + * or for special error behavior for unpaired surrogates, + * but those values are not to be associated with the lead surrogate code *points*. + * @stable ICU 63 + */ + UCPMAP_RANGE_FIXED_LEAD_SURROGATES, + /** + * ucpmap_getRange() enumerates all same-value ranges as stored in the map, + * except that all surrogates (U+D800..U+DFFF) are treated as having the + * surrogateValue, which is passed to getRange() as a separate parameter. + * The surrogateValue is not transformed via filter(). + * See U_IS_SURROGATE(c). + * + * Most users should use UCPMAP_RANGE_NORMAL instead. + * + * This option is useful for maps that map surrogate code *units* to + * special values optimized for UTF-16 string processing + * or for special error behavior for unpaired surrogates, + * but those values are not to be associated with the lead surrogate code *points*. + * @stable ICU 63 + */ + UCPMAP_RANGE_FIXED_ALL_SURROGATES +}; +#ifndef U_IN_DOXYGEN +typedef enum UCPMapRangeOption UCPMapRangeOption; +#endif + +/** + * Returns the value for a code point as stored in the map, with range checking. + * Returns an implementation-defined error value if c is not in the range 0..U+10FFFF. + * + * @param map the map + * @param c the code point + * @return the map value, + * or an implementation-defined error value if the code point is not in the range 0..U+10FFFF + * @stable ICU 63 + */ +U_CAPI uint32_t U_EXPORT2 +ucpmap_get(const UCPMap *map, UChar32 c); + +/** + * Callback function type: Modifies a map value. + * Optionally called by ucpmap_getRange()/ucptrie_getRange()/umutablecptrie_getRange(). + * The modified value will be returned by the getRange function. + * + * Can be used to ignore some of the value bits, + * make a filter for one of several values, + * return a value index computed from the map value, etc. + * + * @param context an opaque pointer, as passed into the getRange function + * @param value a value from the map + * @return the modified value + * @stable ICU 63 + */ +typedef uint32_t U_CALLCONV +UCPMapValueFilter(const void *context, uint32_t value); + +/** + * Returns the last code point such that all those from start to there have the same value. + * Can be used to efficiently iterate over all same-value ranges in a map. + * (This is normally faster than iterating over code points and get()ting each value, + * but much slower than a data structure that stores ranges directly.) + * + * If the UCPMapValueFilter function pointer is not NULL, then + * the value to be delivered is passed through that function, and the return value is the end + * of the range where all values are modified to the same actual value. + * The value is unchanged if that function pointer is NULL. + * + * Example: + * \code + * UChar32 start = 0, end; + * uint32_t value; + * while ((end = ucpmap_getRange(map, start, UCPMAP_RANGE_NORMAL, 0, + * NULL, NULL, &value)) >= 0) { + * // Work with the range start..end and its value. + * start = end + 1; + * } + * \endcode + * + * @param map the map + * @param start range start + * @param option defines whether surrogates are treated normally, + * or as having the surrogateValue; usually UCPMAP_RANGE_NORMAL + * @param surrogateValue value for surrogates; ignored if option==UCPMAP_RANGE_NORMAL + * @param filter a pointer to a function that may modify the map data value, + * or NULL if the values from the map are to be used unmodified + * @param context an opaque pointer that is passed on to the filter function + * @param pValue if not NULL, receives the value that every code point start..end has; + * may have been modified by filter(context, map value) + * if that function pointer is not NULL + * @return the range end code point, or -1 if start is not a valid code point + * @stable ICU 63 + */ +U_CAPI UChar32 U_EXPORT2 +ucpmap_getRange(const UCPMap *map, UChar32 start, + UCPMapRangeOption option, uint32_t surrogateValue, + UCPMapValueFilter *filter, const void *context, uint32_t *pValue); + +U_CDECL_END + +#endif diff --git a/intl/icu/source/common/unicode/ucptrie.h b/intl/icu/source/common/unicode/ucptrie.h new file mode 100644 index 0000000000..dadef79c51 --- /dev/null +++ b/intl/icu/source/common/unicode/ucptrie.h @@ -0,0 +1,645 @@ +// © 2017 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// ucptrie.h (modified from utrie2.h) +// created: 2017dec29 Markus W. Scherer + +#ifndef __UCPTRIE_H__ +#define __UCPTRIE_H__ + +#include "unicode/utypes.h" +#include "unicode/ucpmap.h" +#include "unicode/utf8.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +U_CDECL_BEGIN + +/** + * \file + * \brief C API: This file defines an immutable Unicode code point trie. + * + * @see UCPTrie + * @see UMutableCPTrie + */ + +#ifndef U_IN_DOXYGEN +/** @internal */ +typedef union UCPTrieData { + /** @internal */ + const void *ptr0; + /** @internal */ + const uint16_t *ptr16; + /** @internal */ + const uint32_t *ptr32; + /** @internal */ + const uint8_t *ptr8; +} UCPTrieData; +#endif + +/** + * Immutable Unicode code point trie structure. + * Fast, reasonably compact, map from Unicode code points (U+0000..U+10FFFF) to integer values. + * For details see https://icu.unicode.org/design/struct/utrie + * + * Do not access UCPTrie fields directly; use public functions and macros. + * Functions are easy to use: They support all trie types and value widths. + * + * When performance is really important, macros provide faster access. + * Most macros are specific to either "fast" or "small" tries, see UCPTrieType. + * There are "fast" macros for special optimized use cases. + * + * The macros will return bogus values, or may crash, if used on the wrong type or value width. + * + * @see UMutableCPTrie + * @stable ICU 63 + */ +struct UCPTrie { +#ifndef U_IN_DOXYGEN + /** @internal */ + const uint16_t *index; + /** @internal */ + UCPTrieData data; + + /** @internal */ + int32_t indexLength; + /** @internal */ + int32_t dataLength; + /** Start of the last range which ends at U+10FFFF. @internal */ + UChar32 highStart; + /** highStart>>12 @internal */ + uint16_t shifted12HighStart; + + /** @internal */ + int8_t type; // UCPTrieType + /** @internal */ + int8_t valueWidth; // UCPTrieValueWidth + + /** padding/reserved @internal */ + uint32_t reserved32; + /** padding/reserved @internal */ + uint16_t reserved16; + + /** + * Internal index-3 null block offset. + * Set to an impossibly high value (e.g., 0xffff) if there is no dedicated index-3 null block. + * @internal + */ + uint16_t index3NullOffset; + /** + * Internal data null block offset, not shifted. + * Set to an impossibly high value (e.g., 0xfffff) if there is no dedicated data null block. + * @internal + */ + int32_t dataNullOffset; + /** @internal */ + uint32_t nullValue; + +#ifdef UCPTRIE_DEBUG + /** @internal */ + const char *name; +#endif +#endif +}; +#ifndef U_IN_DOXYGEN +typedef struct UCPTrie UCPTrie; +#endif + +/** + * Selectors for the type of a UCPTrie. + * Different trade-offs for size vs. speed. + * + * @see umutablecptrie_buildImmutable + * @see ucptrie_openFromBinary + * @see ucptrie_getType + * @stable ICU 63 + */ +enum UCPTrieType { + /** + * For ucptrie_openFromBinary() to accept any type. + * ucptrie_getType() will return the actual type. + * @stable ICU 63 + */ + UCPTRIE_TYPE_ANY = -1, + /** + * Fast/simple/larger BMP data structure. Use functions and "fast" macros. + * @stable ICU 63 + */ + UCPTRIE_TYPE_FAST, + /** + * Small/slower BMP data structure. Use functions and "small" macros. + * @stable ICU 63 + */ + UCPTRIE_TYPE_SMALL +}; +#ifndef U_IN_DOXYGEN +typedef enum UCPTrieType UCPTrieType; +#endif + +/** + * Selectors for the number of bits in a UCPTrie data value. + * + * @see umutablecptrie_buildImmutable + * @see ucptrie_openFromBinary + * @see ucptrie_getValueWidth + * @stable ICU 63 + */ +enum UCPTrieValueWidth { + /** + * For ucptrie_openFromBinary() to accept any data value width. + * ucptrie_getValueWidth() will return the actual data value width. + * @stable ICU 63 + */ + UCPTRIE_VALUE_BITS_ANY = -1, + /** + * The trie stores 16 bits per data value. + * It returns them as unsigned values 0..0xffff=65535. + * @stable ICU 63 + */ + UCPTRIE_VALUE_BITS_16, + /** + * The trie stores 32 bits per data value. + * @stable ICU 63 + */ + UCPTRIE_VALUE_BITS_32, + /** + * The trie stores 8 bits per data value. + * It returns them as unsigned values 0..0xff=255. + * @stable ICU 63 + */ + UCPTRIE_VALUE_BITS_8 +}; +#ifndef U_IN_DOXYGEN +typedef enum UCPTrieValueWidth UCPTrieValueWidth; +#endif + +/** + * Opens a trie from its binary form, stored in 32-bit-aligned memory. + * Inverse of ucptrie_toBinary(). + * + * The memory must remain valid and unchanged as long as the trie is used. + * You must ucptrie_close() the trie once you are done using it. + * + * @param type selects the trie type; results in an + * U_INVALID_FORMAT_ERROR if it does not match the binary data; + * use UCPTRIE_TYPE_ANY to accept any type + * @param valueWidth selects the number of bits in a data value; results in an + * U_INVALID_FORMAT_ERROR if it does not match the binary data; + * use UCPTRIE_VALUE_BITS_ANY to accept any data value width + * @param data a pointer to 32-bit-aligned memory containing the binary data of a UCPTrie + * @param length the number of bytes available at data; + * can be more than necessary + * @param pActualLength receives the actual number of bytes at data taken up by the trie data; + * can be NULL + * @param pErrorCode an in/out ICU UErrorCode + * @return the trie + * + * @see umutablecptrie_open + * @see umutablecptrie_buildImmutable + * @see ucptrie_toBinary + * @stable ICU 63 + */ +U_CAPI UCPTrie * U_EXPORT2 +ucptrie_openFromBinary(UCPTrieType type, UCPTrieValueWidth valueWidth, + const void *data, int32_t length, int32_t *pActualLength, + UErrorCode *pErrorCode); + +/** + * Closes a trie and releases associated memory. + * + * @param trie the trie + * @stable ICU 63 + */ +U_CAPI void U_EXPORT2 +ucptrie_close(UCPTrie *trie); + +/** + * Returns the trie type. + * + * @param trie the trie + * @return the trie type + * @see ucptrie_openFromBinary + * @see UCPTRIE_TYPE_ANY + * @stable ICU 63 + */ +U_CAPI UCPTrieType U_EXPORT2 +ucptrie_getType(const UCPTrie *trie); + +/** + * Returns the number of bits in a trie data value. + * + * @param trie the trie + * @return the number of bits in a trie data value + * @see ucptrie_openFromBinary + * @see UCPTRIE_VALUE_BITS_ANY + * @stable ICU 63 + */ +U_CAPI UCPTrieValueWidth U_EXPORT2 +ucptrie_getValueWidth(const UCPTrie *trie); + +/** + * Returns the value for a code point as stored in the trie, with range checking. + * Returns the trie error value if c is not in the range 0..U+10FFFF. + * + * Easier to use than UCPTRIE_FAST_GET() and similar macros but slower. + * Easier to use because, unlike the macros, this function works on all UCPTrie + * objects, for all types and value widths. + * + * @param trie the trie + * @param c the code point + * @return the trie value, + * or the trie error value if the code point is not in the range 0..U+10FFFF + * @stable ICU 63 + */ +U_CAPI uint32_t U_EXPORT2 +ucptrie_get(const UCPTrie *trie, UChar32 c); + +/** + * Returns the last code point such that all those from start to there have the same value. + * Can be used to efficiently iterate over all same-value ranges in a trie. + * (This is normally faster than iterating over code points and get()ting each value, + * but much slower than a data structure that stores ranges directly.) + * + * If the UCPMapValueFilter function pointer is not NULL, then + * the value to be delivered is passed through that function, and the return value is the end + * of the range where all values are modified to the same actual value. + * The value is unchanged if that function pointer is NULL. + * + * Example: + * \code + * UChar32 start = 0, end; + * uint32_t value; + * while ((end = ucptrie_getRange(trie, start, UCPMAP_RANGE_NORMAL, 0, + * NULL, NULL, &value)) >= 0) { + * // Work with the range start..end and its value. + * start = end + 1; + * } + * \endcode + * + * @param trie the trie + * @param start range start + * @param option defines whether surrogates are treated normally, + * or as having the surrogateValue; usually UCPMAP_RANGE_NORMAL + * @param surrogateValue value for surrogates; ignored if option==UCPMAP_RANGE_NORMAL + * @param filter a pointer to a function that may modify the trie data value, + * or NULL if the values from the trie are to be used unmodified + * @param context an opaque pointer that is passed on to the filter function + * @param pValue if not NULL, receives the value that every code point start..end has; + * may have been modified by filter(context, trie value) + * if that function pointer is not NULL + * @return the range end code point, or -1 if start is not a valid code point + * @stable ICU 63 + */ +U_CAPI UChar32 U_EXPORT2 +ucptrie_getRange(const UCPTrie *trie, UChar32 start, + UCPMapRangeOption option, uint32_t surrogateValue, + UCPMapValueFilter *filter, const void *context, uint32_t *pValue); + +/** + * Writes a memory-mappable form of the trie into 32-bit aligned memory. + * Inverse of ucptrie_openFromBinary(). + * + * @param trie the trie + * @param data a pointer to 32-bit-aligned memory to be filled with the trie data; + * can be NULL if capacity==0 + * @param capacity the number of bytes available at data, or 0 for pure preflighting + * @param pErrorCode an in/out ICU UErrorCode; + * U_BUFFER_OVERFLOW_ERROR if the capacity is too small + * @return the number of bytes written or (if buffer overflow) needed for the trie + * + * @see ucptrie_openFromBinary() + * @stable ICU 63 + */ +U_CAPI int32_t U_EXPORT2 +ucptrie_toBinary(const UCPTrie *trie, void *data, int32_t capacity, UErrorCode *pErrorCode); + +/** + * Macro parameter value for a trie with 16-bit data values. + * Use the name of this macro as a "dataAccess" parameter in other macros. + * Do not use this macro in any other way. + * + * @see UCPTRIE_VALUE_BITS_16 + * @stable ICU 63 + */ +#define UCPTRIE_16(trie, i) ((trie)->data.ptr16[i]) + +/** + * Macro parameter value for a trie with 32-bit data values. + * Use the name of this macro as a "dataAccess" parameter in other macros. + * Do not use this macro in any other way. + * + * @see UCPTRIE_VALUE_BITS_32 + * @stable ICU 63 + */ +#define UCPTRIE_32(trie, i) ((trie)->data.ptr32[i]) + +/** + * Macro parameter value for a trie with 8-bit data values. + * Use the name of this macro as a "dataAccess" parameter in other macros. + * Do not use this macro in any other way. + * + * @see UCPTRIE_VALUE_BITS_8 + * @stable ICU 63 + */ +#define UCPTRIE_8(trie, i) ((trie)->data.ptr8[i]) + +/** + * Returns a trie value for a code point, with range checking. + * Returns the trie error value if c is not in the range 0..U+10FFFF. + * + * @param trie (const UCPTrie *, in) the trie; must have type UCPTRIE_TYPE_FAST + * @param dataAccess UCPTRIE_16, UCPTRIE_32, or UCPTRIE_8 according to the trie’s value width + * @param c (UChar32, in) the input code point + * @return The code point's trie value. + * @stable ICU 63 + */ +#define UCPTRIE_FAST_GET(trie, dataAccess, c) dataAccess(trie, _UCPTRIE_CP_INDEX(trie, 0xffff, c)) + +/** + * Returns a 16-bit trie value for a code point, with range checking. + * Returns the trie error value if c is not in the range U+0000..U+10FFFF. + * + * @param trie (const UCPTrie *, in) the trie; must have type UCPTRIE_TYPE_SMALL + * @param dataAccess UCPTRIE_16, UCPTRIE_32, or UCPTRIE_8 according to the trie’s value width + * @param c (UChar32, in) the input code point + * @return The code point's trie value. + * @stable ICU 63 + */ +#define UCPTRIE_SMALL_GET(trie, dataAccess, c) \ + dataAccess(trie, _UCPTRIE_CP_INDEX(trie, UCPTRIE_SMALL_MAX, c)) + +/** + * UTF-16: Reads the next code point (UChar32 c, out), post-increments src, + * and gets a value from the trie. + * Sets the trie error value if c is an unpaired surrogate. + * + * @param trie (const UCPTrie *, in) the trie; must have type UCPTRIE_TYPE_FAST + * @param dataAccess UCPTRIE_16, UCPTRIE_32, or UCPTRIE_8 according to the trie’s value width + * @param src (const UChar *, in/out) the source text pointer + * @param limit (const UChar *, in) the limit pointer for the text, or NULL if NUL-terminated + * @param c (UChar32, out) variable for the code point + * @param result (out) variable for the trie lookup result + * @stable ICU 63 + */ +#define UCPTRIE_FAST_U16_NEXT(trie, dataAccess, src, limit, c, result) UPRV_BLOCK_MACRO_BEGIN { \ + (c) = *(src)++; \ + int32_t __index; \ + if (!U16_IS_SURROGATE(c)) { \ + __index = _UCPTRIE_FAST_INDEX(trie, c); \ + } else { \ + uint16_t __c2; \ + if (U16_IS_SURROGATE_LEAD(c) && (src) != (limit) && U16_IS_TRAIL(__c2 = *(src))) { \ + ++(src); \ + (c) = U16_GET_SUPPLEMENTARY((c), __c2); \ + __index = _UCPTRIE_SMALL_INDEX(trie, c); \ + } else { \ + __index = (trie)->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET; \ + } \ + } \ + (result) = dataAccess(trie, __index); \ +} UPRV_BLOCK_MACRO_END + +/** + * UTF-16: Reads the previous code point (UChar32 c, out), pre-decrements src, + * and gets a value from the trie. + * Sets the trie error value if c is an unpaired surrogate. + * + * @param trie (const UCPTrie *, in) the trie; must have type UCPTRIE_TYPE_FAST + * @param dataAccess UCPTRIE_16, UCPTRIE_32, or UCPTRIE_8 according to the trie’s value width + * @param start (const UChar *, in) the start pointer for the text + * @param src (const UChar *, in/out) the source text pointer + * @param c (UChar32, out) variable for the code point + * @param result (out) variable for the trie lookup result + * @stable ICU 63 + */ +#define UCPTRIE_FAST_U16_PREV(trie, dataAccess, start, src, c, result) UPRV_BLOCK_MACRO_BEGIN { \ + (c) = *--(src); \ + int32_t __index; \ + if (!U16_IS_SURROGATE(c)) { \ + __index = _UCPTRIE_FAST_INDEX(trie, c); \ + } else { \ + uint16_t __c2; \ + if (U16_IS_SURROGATE_TRAIL(c) && (src) != (start) && U16_IS_LEAD(__c2 = *((src) - 1))) { \ + --(src); \ + (c) = U16_GET_SUPPLEMENTARY(__c2, (c)); \ + __index = _UCPTRIE_SMALL_INDEX(trie, c); \ + } else { \ + __index = (trie)->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET; \ + } \ + } \ + (result) = dataAccess(trie, __index); \ +} UPRV_BLOCK_MACRO_END + +/** + * UTF-8: Post-increments src and gets a value from the trie. + * Sets the trie error value for an ill-formed byte sequence. + * + * Unlike UCPTRIE_FAST_U16_NEXT() this UTF-8 macro does not provide the code point + * because it would be more work to do so and is often not needed. + * If the trie value differs from the error value, then the byte sequence is well-formed, + * and the code point can be assembled without revalidation. + * + * @param trie (const UCPTrie *, in) the trie; must have type UCPTRIE_TYPE_FAST + * @param dataAccess UCPTRIE_16, UCPTRIE_32, or UCPTRIE_8 according to the trie’s value width + * @param src (const char *, in/out) the source text pointer + * @param limit (const char *, in) the limit pointer for the text (must not be NULL) + * @param result (out) variable for the trie lookup result + * @stable ICU 63 + */ +#define UCPTRIE_FAST_U8_NEXT(trie, dataAccess, src, limit, result) UPRV_BLOCK_MACRO_BEGIN { \ + int32_t __lead = (uint8_t)*(src)++; \ + if (!U8_IS_SINGLE(__lead)) { \ + uint8_t __t1, __t2, __t3; \ + if ((src) != (limit) && \ + (__lead >= 0xe0 ? \ + __lead < 0xf0 ? /* U+0800..U+FFFF except surrogates */ \ + U8_LEAD3_T1_BITS[__lead &= 0xf] & (1 << ((__t1 = *(src)) >> 5)) && \ + ++(src) != (limit) && (__t2 = *(src) - 0x80) <= 0x3f && \ + (__lead = ((int32_t)(trie)->index[(__lead << 6) + (__t1 & 0x3f)]) + __t2, 1) \ + : /* U+10000..U+10FFFF */ \ + (__lead -= 0xf0) <= 4 && \ + U8_LEAD4_T1_BITS[(__t1 = *(src)) >> 4] & (1 << __lead) && \ + (__lead = (__lead << 6) | (__t1 & 0x3f), ++(src) != (limit)) && \ + (__t2 = *(src) - 0x80) <= 0x3f && \ + ++(src) != (limit) && (__t3 = *(src) - 0x80) <= 0x3f && \ + (__lead = __lead >= (trie)->shifted12HighStart ? \ + (trie)->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET : \ + ucptrie_internalSmallU8Index((trie), __lead, __t2, __t3), 1) \ + : /* U+0080..U+07FF */ \ + __lead >= 0xc2 && (__t1 = *(src) - 0x80) <= 0x3f && \ + (__lead = (int32_t)(trie)->index[__lead & 0x1f] + __t1, 1))) { \ + ++(src); \ + } else { \ + __lead = (trie)->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET; /* ill-formed*/ \ + } \ + } \ + (result) = dataAccess(trie, __lead); \ +} UPRV_BLOCK_MACRO_END + +/** + * UTF-8: Pre-decrements src and gets a value from the trie. + * Sets the trie error value for an ill-formed byte sequence. + * + * Unlike UCPTRIE_FAST_U16_PREV() this UTF-8 macro does not provide the code point + * because it would be more work to do so and is often not needed. + * If the trie value differs from the error value, then the byte sequence is well-formed, + * and the code point can be assembled without revalidation. + * + * @param trie (const UCPTrie *, in) the trie; must have type UCPTRIE_TYPE_FAST + * @param dataAccess UCPTRIE_16, UCPTRIE_32, or UCPTRIE_8 according to the trie’s value width + * @param start (const char *, in) the start pointer for the text + * @param src (const char *, in/out) the source text pointer + * @param result (out) variable for the trie lookup result + * @stable ICU 63 + */ +#define UCPTRIE_FAST_U8_PREV(trie, dataAccess, start, src, result) UPRV_BLOCK_MACRO_BEGIN { \ + int32_t __index = (uint8_t)*--(src); \ + if (!U8_IS_SINGLE(__index)) { \ + __index = ucptrie_internalU8PrevIndex((trie), __index, (const uint8_t *)(start), \ + (const uint8_t *)(src)); \ + (src) -= __index & 7; \ + __index >>= 3; \ + } \ + (result) = dataAccess(trie, __index); \ +} UPRV_BLOCK_MACRO_END + +/** + * Returns a trie value for an ASCII code point, without range checking. + * + * @param trie (const UCPTrie *, in) the trie (of either fast or small type) + * @param dataAccess UCPTRIE_16, UCPTRIE_32, or UCPTRIE_8 according to the trie’s value width + * @param c (UChar32, in) the input code point; must be U+0000..U+007F + * @return The ASCII code point's trie value. + * @stable ICU 63 + */ +#define UCPTRIE_ASCII_GET(trie, dataAccess, c) dataAccess(trie, c) + +/** + * Returns a trie value for a BMP code point (U+0000..U+FFFF), without range checking. + * Can be used to look up a value for a UTF-16 code unit if other parts of + * the string processing check for surrogates. + * + * @param trie (const UCPTrie *, in) the trie; must have type UCPTRIE_TYPE_FAST + * @param dataAccess UCPTRIE_16, UCPTRIE_32, or UCPTRIE_8 according to the trie’s value width + * @param c (UChar32, in) the input code point, must be U+0000..U+FFFF + * @return The BMP code point's trie value. + * @stable ICU 63 + */ +#define UCPTRIE_FAST_BMP_GET(trie, dataAccess, c) dataAccess(trie, _UCPTRIE_FAST_INDEX(trie, c)) + +/** + * Returns a trie value for a supplementary code point (U+10000..U+10FFFF), + * without range checking. + * + * @param trie (const UCPTrie *, in) the trie; must have type UCPTRIE_TYPE_FAST + * @param dataAccess UCPTRIE_16, UCPTRIE_32, or UCPTRIE_8 according to the trie’s value width + * @param c (UChar32, in) the input code point, must be U+10000..U+10FFFF + * @return The supplementary code point's trie value. + * @stable ICU 63 + */ +#define UCPTRIE_FAST_SUPP_GET(trie, dataAccess, c) dataAccess(trie, _UCPTRIE_SMALL_INDEX(trie, c)) + +/* Internal definitions ----------------------------------------------------- */ + +#ifndef U_IN_DOXYGEN + +/** + * Internal implementation constants. + * These are needed for the API macros, but users should not use these directly. + * @internal + */ +enum { + /** @internal */ + UCPTRIE_FAST_SHIFT = 6, + + /** Number of entries in a data block for code points below the fast limit. 64=0x40 @internal */ + UCPTRIE_FAST_DATA_BLOCK_LENGTH = 1 << UCPTRIE_FAST_SHIFT, + + /** Mask for getting the lower bits for the in-fast-data-block offset. @internal */ + UCPTRIE_FAST_DATA_MASK = UCPTRIE_FAST_DATA_BLOCK_LENGTH - 1, + + /** @internal */ + UCPTRIE_SMALL_MAX = 0xfff, + + /** + * Offset from dataLength (to be subtracted) for fetching the + * value returned for out-of-range code points and ill-formed UTF-8/16. + * @internal + */ + UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET = 1, + /** + * Offset from dataLength (to be subtracted) for fetching the + * value returned for code points highStart..U+10FFFF. + * @internal + */ + UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET = 2 +}; + +/* Internal functions and macros -------------------------------------------- */ +// Do not conditionalize with #ifndef U_HIDE_INTERNAL_API, needed for public API + +/** @internal */ +U_CAPI int32_t U_EXPORT2 +ucptrie_internalSmallIndex(const UCPTrie *trie, UChar32 c); + +/** @internal */ +U_CAPI int32_t U_EXPORT2 +ucptrie_internalSmallU8Index(const UCPTrie *trie, int32_t lt1, uint8_t t2, uint8_t t3); + +/** + * Internal function for part of the UCPTRIE_FAST_U8_PREVxx() macro implementations. + * Do not call directly. + * @internal + */ +U_CAPI int32_t U_EXPORT2 +ucptrie_internalU8PrevIndex(const UCPTrie *trie, UChar32 c, + const uint8_t *start, const uint8_t *src); + +/** Internal trie getter for a code point below the fast limit. Returns the data index. @internal */ +#define _UCPTRIE_FAST_INDEX(trie, c) \ + ((int32_t)(trie)->index[(c) >> UCPTRIE_FAST_SHIFT] + ((c) & UCPTRIE_FAST_DATA_MASK)) + +/** Internal trie getter for a code point at or above the fast limit. Returns the data index. @internal */ +#define _UCPTRIE_SMALL_INDEX(trie, c) \ + ((c) >= (trie)->highStart ? \ + (trie)->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET : \ + ucptrie_internalSmallIndex(trie, c)) + +/** + * Internal trie getter for a code point, with checking that c is in U+0000..10FFFF. + * Returns the data index. + * @internal + */ +#define _UCPTRIE_CP_INDEX(trie, fastMax, c) \ + ((uint32_t)(c) <= (uint32_t)(fastMax) ? \ + _UCPTRIE_FAST_INDEX(trie, c) : \ + (uint32_t)(c) <= 0x10ffff ? \ + _UCPTRIE_SMALL_INDEX(trie, c) : \ + (trie)->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET) + +U_CDECL_END + +#endif // U_IN_DOXYGEN + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUCPTriePointer + * "Smart pointer" class, closes a UCPTrie via ucptrie_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 63 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUCPTriePointer, UCPTrie, ucptrie_close); + +U_NAMESPACE_END + +#endif // U_SHOW_CPLUSPLUS_API + +#endif diff --git a/intl/icu/source/common/unicode/ucurr.h b/intl/icu/source/common/unicode/ucurr.h new file mode 100644 index 0000000000..5589e79990 --- /dev/null +++ b/intl/icu/source/common/unicode/ucurr.h @@ -0,0 +1,466 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2002-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ +#ifndef _UCURR_H_ +#define _UCURR_H_ + +#include "unicode/utypes.h" +#include "unicode/uenum.h" + +/** + * \file + * \brief C API: Encapsulates information about a currency. + * + * The ucurr API encapsulates information about a currency, as defined by + * ISO 4217. A currency is represented by a 3-character string + * containing its ISO 4217 code. This API can return various data + * necessary the proper display of a currency: + * + *
  • A display symbol, for a specific locale + *
  • The number of fraction digits to display + *
  • A rounding increment + *
+ * + * The DecimalFormat class uses these data to display + * currencies. + * @author Alan Liu + * @since ICU 2.2 + */ + +#if !UCONFIG_NO_FORMATTING + +/** + * Currency Usage used for Decimal Format + * @stable ICU 54 + */ +enum UCurrencyUsage { + /** + * a setting to specify currency usage which determines currency digit + * and rounding for standard usage, for example: "50.00 NT$" + * used as DEFAULT value + * @stable ICU 54 + */ + UCURR_USAGE_STANDARD=0, + /** + * a setting to specify currency usage which determines currency digit + * and rounding for cash usage, for example: "50 NT$" + * @stable ICU 54 + */ + UCURR_USAGE_CASH=1, +#ifndef U_HIDE_DEPRECATED_API + /** + * One higher than the last enum UCurrencyUsage constant. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + UCURR_USAGE_COUNT=2 +#endif // U_HIDE_DEPRECATED_API +}; +/** Currency Usage used for Decimal Format */ +typedef enum UCurrencyUsage UCurrencyUsage; + +/** + * Finds a currency code for the given locale. + * @param locale the locale for which to retrieve a currency code. + * Currency can be specified by the "currency" keyword + * in which case it overrides the default currency code + * @param buff fill in buffer. Can be NULL for preflighting. + * @param buffCapacity capacity of the fill in buffer. Can be 0 for + * preflighting. If it is non-zero, the buff parameter + * must not be NULL. + * @param ec error code + * @return length of the currency string. It should always be 3. If 0, + * currency couldn't be found or the input values are + * invalid. + * @stable ICU 2.8 + */ +U_CAPI int32_t U_EXPORT2 +ucurr_forLocale(const char* locale, + UChar* buff, + int32_t buffCapacity, + UErrorCode* ec); + +/** + * Selector constants for ucurr_getName(). + * + * @see ucurr_getName + * @stable ICU 2.6 + */ +typedef enum UCurrNameStyle { + /** + * Selector for ucurr_getName indicating a symbolic name for a + * currency, such as "$" for USD. + * @stable ICU 2.6 + */ + UCURR_SYMBOL_NAME, + + /** + * Selector for ucurr_getName indicating the long name for a + * currency, such as "US Dollar" for USD. + * @stable ICU 2.6 + */ + UCURR_LONG_NAME, + + /** + * Selector for getName() indicating the narrow currency symbol. + * The narrow currency symbol is similar to the regular currency + * symbol, but it always takes the shortest form: for example, + * "$" instead of "US$" for USD in en-CA. + * + * @stable ICU 61 + */ + UCURR_NARROW_SYMBOL_NAME, + + /** + * Selector for getName() indicating the formal currency symbol. + * The formal currency symbol is similar to the regular currency + * symbol, but it always takes the form used in formal settings + * such as banking; for example, "NT$" instead of "$" for TWD in zh-TW. + * + * @stable ICU 68 + */ + UCURR_FORMAL_SYMBOL_NAME, + + /** + * Selector for getName() indicating the variant currency symbol. + * The variant symbol for a currency is an alternative symbol + * that is not necessarily as widely used as the regular symbol. + * + * @stable ICU 68 + */ + UCURR_VARIANT_SYMBOL_NAME + +} UCurrNameStyle; + +#if !UCONFIG_NO_SERVICE +/** + * @stable ICU 2.6 + */ +typedef const void* UCurrRegistryKey; + +/** + * Register an (existing) ISO 4217 currency code for the given locale. + * Only the country code and the two variants EURO and PRE_EURO are + * recognized. + * @param isoCode the three-letter ISO 4217 currency code + * @param locale the locale for which to register this currency code + * @param status the in/out status code + * @return a registry key that can be used to unregister this currency code, or NULL + * if there was an error. + * @stable ICU 2.6 + */ +U_CAPI UCurrRegistryKey U_EXPORT2 +ucurr_register(const UChar* isoCode, + const char* locale, + UErrorCode* status); +/** + * Unregister the previously-registered currency definitions using the + * URegistryKey returned from ucurr_register. Key becomes invalid after + * a successful call and should not be used again. Any currency + * that might have been hidden by the original ucurr_register call is + * restored. + * @param key the registry key returned by a previous call to ucurr_register + * @param status the in/out status code, no special meanings are assigned + * @return true if the currency for this key was successfully unregistered + * @stable ICU 2.6 + */ +U_CAPI UBool U_EXPORT2 +ucurr_unregister(UCurrRegistryKey key, UErrorCode* status); +#endif /* UCONFIG_NO_SERVICE */ + +/** + * Returns the display name for the given currency in the + * given locale. For example, the display name for the USD + * currency object in the en_US locale is "$". + * @param currency null-terminated 3-letter ISO 4217 code + * @param locale locale in which to display currency + * @param nameStyle selector for which kind of name to return + * @param isChoiceFormat always set to false, or can be NULL; + * display names are static strings; + * since ICU 4.4, ChoiceFormat patterns are no longer supported + * @param len fill-in parameter to receive length of result + * @param ec error code + * @return pointer to display string of 'len' UChars. If the resource + * data contains no entry for 'currency', then 'currency' itself is + * returned. + * @stable ICU 2.6 + */ +U_CAPI const UChar* U_EXPORT2 +ucurr_getName(const UChar* currency, + const char* locale, + UCurrNameStyle nameStyle, + UBool* isChoiceFormat, + int32_t* len, + UErrorCode* ec); + +/** + * Returns the plural name for the given currency in the + * given locale. For example, the plural name for the USD + * currency object in the en_US locale is "US dollar" or "US dollars". + * @param currency null-terminated 3-letter ISO 4217 code + * @param locale locale in which to display currency + * @param isChoiceFormat always set to false, or can be NULL; + * display names are static strings; + * since ICU 4.4, ChoiceFormat patterns are no longer supported + * @param pluralCount plural count + * @param len fill-in parameter to receive length of result + * @param ec error code + * @return pointer to display string of 'len' UChars. If the resource + * data contains no entry for 'currency', then 'currency' itself is + * returned. + * @stable ICU 4.2 + */ +U_CAPI const UChar* U_EXPORT2 +ucurr_getPluralName(const UChar* currency, + const char* locale, + UBool* isChoiceFormat, + const char* pluralCount, + int32_t* len, + UErrorCode* ec); + +/** + * Returns the number of the number of fraction digits that should + * be displayed for the given currency. + * This is equivalent to ucurr_getDefaultFractionDigitsForUsage(currency,UCURR_USAGE_STANDARD,ec); + * + * Important: The number of fraction digits for a given currency is NOT + * guaranteed to be constant across versions of ICU or CLDR. For example, + * do NOT use this value as a mechanism for deciding the magnitude used + * to store currency values in a database. You should use this value for + * display purposes only. + * + * @param currency null-terminated 3-letter ISO 4217 code + * @param ec input-output error code + * @return a non-negative number of fraction digits to be + * displayed, or 0 if there is an error + * @stable ICU 3.0 + */ +U_CAPI int32_t U_EXPORT2 +ucurr_getDefaultFractionDigits(const UChar* currency, + UErrorCode* ec); + +/** + * Returns the number of the number of fraction digits that should + * be displayed for the given currency with usage. + * + * Important: The number of fraction digits for a given currency is NOT + * guaranteed to be constant across versions of ICU or CLDR. For example, + * do NOT use this value as a mechanism for deciding the magnitude used + * to store currency values in a database. You should use this value for + * display purposes only. + * + * @param currency null-terminated 3-letter ISO 4217 code + * @param usage enum usage for the currency + * @param ec input-output error code + * @return a non-negative number of fraction digits to be + * displayed, or 0 if there is an error + * @stable ICU 54 + */ +U_CAPI int32_t U_EXPORT2 +ucurr_getDefaultFractionDigitsForUsage(const UChar* currency, + const UCurrencyUsage usage, + UErrorCode* ec); + +/** + * Returns the rounding increment for the given currency, or 0.0 if no + * rounding is done by the currency. + * This is equivalent to ucurr_getRoundingIncrementForUsage(currency,UCURR_USAGE_STANDARD,ec); + * @param currency null-terminated 3-letter ISO 4217 code + * @param ec input-output error code + * @return the non-negative rounding increment, or 0.0 if none, + * or 0.0 if there is an error + * @stable ICU 3.0 + */ +U_CAPI double U_EXPORT2 +ucurr_getRoundingIncrement(const UChar* currency, + UErrorCode* ec); + +/** + * Returns the rounding increment for the given currency, or 0.0 if no + * rounding is done by the currency given usage. + * @param currency null-terminated 3-letter ISO 4217 code + * @param usage enum usage for the currency + * @param ec input-output error code + * @return the non-negative rounding increment, or 0.0 if none, + * or 0.0 if there is an error + * @stable ICU 54 + */ +U_CAPI double U_EXPORT2 +ucurr_getRoundingIncrementForUsage(const UChar* currency, + const UCurrencyUsage usage, + UErrorCode* ec); + +/** + * Selector constants for ucurr_openCurrencies(). + * + * @see ucurr_openCurrencies + * @stable ICU 3.2 + */ +typedef enum UCurrCurrencyType { + /** + * Select all ISO-4217 currency codes. + * @stable ICU 3.2 + */ + UCURR_ALL = INT32_MAX, + /** + * Select only ISO-4217 commonly used currency codes. + * These currencies can be found in common use, and they usually have + * bank notes or coins associated with the currency code. + * This does not include fund codes, precious metals and other + * various ISO-4217 codes limited to special financial products. + * @stable ICU 3.2 + */ + UCURR_COMMON = 1, + /** + * Select ISO-4217 uncommon currency codes. + * These codes respresent fund codes, precious metals and other + * various ISO-4217 codes limited to special financial products. + * A fund code is a monetary resource associated with a currency. + * @stable ICU 3.2 + */ + UCURR_UNCOMMON = 2, + /** + * Select only deprecated ISO-4217 codes. + * These codes are no longer in general public use. + * @stable ICU 3.2 + */ + UCURR_DEPRECATED = 4, + /** + * Select only non-deprecated ISO-4217 codes. + * These codes are in general public use. + * @stable ICU 3.2 + */ + UCURR_NON_DEPRECATED = 8 +} UCurrCurrencyType; + +/** + * Provides a UEnumeration object for listing ISO-4217 codes. + * @param currType You can use one of several UCurrCurrencyType values for this + * variable. You can also | (or) them together to get a specific list of + * currencies. Most people will want to use the (UCURR_COMMON|UCURR_NON_DEPRECATED) value to + * get a list of current currencies. + * @param pErrorCode Error code + * @stable ICU 3.2 + */ +U_CAPI UEnumeration * U_EXPORT2 +ucurr_openISOCurrencies(uint32_t currType, UErrorCode *pErrorCode); + +/** + * Queries if the given ISO 4217 3-letter code is available on the specified date range. + * + * Note: For checking availability of a currency on a specific date, specify the date on both 'from' and 'to' + * + * When 'from' is U_DATE_MIN and 'to' is U_DATE_MAX, this method checks if the specified currency is available any time. + * If 'from' and 'to' are same UDate value, this method checks if the specified currency is available on that date. + * + * @param isoCode + * The ISO 4217 3-letter code. + * + * @param from + * The lower bound of the date range, inclusive. When 'from' is U_DATE_MIN, check the availability + * of the currency any date before 'to' + * + * @param to + * The upper bound of the date range, inclusive. When 'to' is U_DATE_MAX, check the availability of + * the currency any date after 'from' + * + * @param errorCode + * ICU error code + * + * @return true if the given ISO 4217 3-letter code is supported on the specified date range. + * + * @stable ICU 4.8 + */ +U_CAPI UBool U_EXPORT2 +ucurr_isAvailable(const UChar* isoCode, + UDate from, + UDate to, + UErrorCode* errorCode); + +/** + * Finds the number of valid currency codes for the + * given locale and date. + * @param locale the locale for which to retrieve the + * currency count. + * @param date the date for which to retrieve the + * currency count for the given locale. + * @param ec error code + * @return the number of currency codes for the + * given locale and date. If 0, currency + * codes couldn't be found for the input + * values are invalid. + * @stable ICU 4.0 + */ +U_CAPI int32_t U_EXPORT2 +ucurr_countCurrencies(const char* locale, + UDate date, + UErrorCode* ec); + +/** + * Finds a currency code for the given locale and date + * @param locale the locale for which to retrieve a currency code. + * Currency can be specified by the "currency" keyword + * in which case it overrides the default currency code + * @param date the date for which to retrieve a currency code for + * the given locale. + * @param index the index within the available list of currency codes + * for the given locale on the given date. + * @param buff fill in buffer. Can be NULL for preflighting. + * @param buffCapacity capacity of the fill in buffer. Can be 0 for + * preflighting. If it is non-zero, the buff parameter + * must not be NULL. + * @param ec error code + * @return length of the currency string. It should always be 3. + * If 0, currency couldn't be found or the input values are + * invalid. + * @stable ICU 4.0 + */ +U_CAPI int32_t U_EXPORT2 +ucurr_forLocaleAndDate(const char* locale, + UDate date, + int32_t index, + UChar* buff, + int32_t buffCapacity, + UErrorCode* ec); + +/** + * Given a key and a locale, returns an array of string values in a preferred + * order that would make a difference. These are all and only those values where + * the open (creation) of the service with the locale formed from the input locale + * plus input keyword and that value has different behavior than creation with the + * input locale alone. + * @param key one of the keys supported by this service. For now, only + * "currency" is supported. + * @param locale the locale + * @param commonlyUsed if set to true it will return only commonly used values + * with the given locale in preferred order. Otherwise, + * it will return all the available values for the locale. + * @param status error status + * @return a string enumeration over keyword values for the given key and the locale. + * @stable ICU 4.2 + */ +U_CAPI UEnumeration* U_EXPORT2 +ucurr_getKeywordValuesForLocale(const char* key, + const char* locale, + UBool commonlyUsed, + UErrorCode* status); + +/** + * Returns the ISO 4217 numeric code for the currency. + *

Note: If the ISO 4217 numeric code is not assigned for the currency or + * the currency is unknown, this function returns 0. + * + * @param currency null-terminated 3-letter ISO 4217 code + * @return The ISO 4217 numeric code of the currency + * @stable ICU 49 + */ +U_CAPI int32_t U_EXPORT2 +ucurr_getNumericCode(const UChar* currency); + +#endif /* #if !UCONFIG_NO_FORMATTING */ + +#endif diff --git a/intl/icu/source/common/unicode/udata.h b/intl/icu/source/common/unicode/udata.h new file mode 100644 index 0000000000..4cda255010 --- /dev/null +++ b/intl/icu/source/common/unicode/udata.h @@ -0,0 +1,440 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2014, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: udata.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 1999oct25 +* created by: Markus W. Scherer +*/ + +#ifndef __UDATA_H__ +#define __UDATA_H__ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +U_CDECL_BEGIN + +/** + * \file + * \brief C API: Data loading interface + * + *

Information about data loading interface

+ * + * This API is used to find and efficiently load data for ICU and applications + * using ICU. It provides an abstract interface that specifies a data type and + * name to find and load the data. Normally this API is used by other ICU APIs + * to load required data out of the ICU data library, but it can be used to + * load data out of other places. + * + * See the User Guide Data Management chapter. + */ + +#ifndef U_HIDE_INTERNAL_API +/** + * Character used to separate package names from tree names + * @internal ICU 3.0 + */ +#define U_TREE_SEPARATOR '-' + +/** + * String used to separate package names from tree names + * @internal ICU 3.0 + */ +#define U_TREE_SEPARATOR_STRING "-" + +/** + * Character used to separate parts of entry names + * @internal ICU 3.0 + */ +#define U_TREE_ENTRY_SEP_CHAR '/' + +/** + * String used to separate parts of entry names + * @internal ICU 3.0 + */ +#define U_TREE_ENTRY_SEP_STRING "/" + +/** + * Alias for standard ICU data + * @internal ICU 3.0 + */ +#define U_ICUDATA_ALIAS "ICUDATA" + +#endif /* U_HIDE_INTERNAL_API */ + +/** + * UDataInfo contains the properties about the requested data. + * This is meta data. + * + *

This structure may grow in the future, indicated by the + * size field.

+ * + *

ICU data must be at least 8-aligned, and should be 16-aligned. + * The UDataInfo struct begins 4 bytes after the start of the data item, + * so it is 4-aligned. + * + *

The platform data property fields help determine if a data + * file can be efficiently used on a given machine. + * The particular fields are of importance only if the data + * is affected by the properties - if there is integer data + * with word sizes > 1 byte, char* text, or UChar* text.

+ * + *

The implementation for the udata_open[Choice]() + * functions may reject data based on the value in isBigEndian. + * No other field is used by the udata API implementation.

+ * + *

The dataFormat may be used to identify + * the kind of data, e.g. a converter table.

+ * + *

The formatVersion field should be used to + * make sure that the format can be interpreted. + * It may be a good idea to check only for the one or two highest + * of the version elements to allow the data memory to + * get more or somewhat rearranged contents, for as long + * as the using code can still interpret the older contents.

+ * + *

The dataVersion field is intended to be a + * common place to store the source version of the data; + * for data from the Unicode character database, this could + * reflect the Unicode version.

+ * + * @stable ICU 2.0 + */ +typedef struct { + /** sizeof(UDataInfo) + * @stable ICU 2.0 */ + uint16_t size; + + /** unused, set to 0 + * @stable ICU 2.0*/ + uint16_t reservedWord; + + /* platform data properties */ + /** 0 for little-endian machine, 1 for big-endian + * @stable ICU 2.0 */ + uint8_t isBigEndian; + + /** see U_CHARSET_FAMILY values in utypes.h + * @stable ICU 2.0*/ + uint8_t charsetFamily; + + /** sizeof(UChar), one of { 1, 2, 4 } + * @stable ICU 2.0*/ + uint8_t sizeofUChar; + + /** unused, set to 0 + * @stable ICU 2.0*/ + uint8_t reservedByte; + + /** data format identifier + * @stable ICU 2.0*/ + uint8_t dataFormat[4]; + + /** versions: [0] major [1] minor [2] milli [3] micro + * @stable ICU 2.0*/ + uint8_t formatVersion[4]; + + /** versions: [0] major [1] minor [2] milli [3] micro + * @stable ICU 2.0*/ + uint8_t dataVersion[4]; +} UDataInfo; + +/* API for reading data -----------------------------------------------------*/ + +/** + * Forward declaration of the data memory type. + * @stable ICU 2.0 + */ +typedef struct UDataMemory UDataMemory; + +/** + * Callback function for udata_openChoice(). + * @param context parameter passed into udata_openChoice(). + * @param type The type of the data as passed into udata_openChoice(). + * It may be NULL. + * @param name The name of the data as passed into udata_openChoice(). + * @param pInfo A pointer to the UDataInfo structure + * of data that has been loaded and will be returned + * by udata_openChoice() if this function + * returns true. + * @return true if the current data memory is acceptable + * @stable ICU 2.0 + */ +typedef UBool U_CALLCONV +UDataMemoryIsAcceptable(void *context, + const char *type, const char *name, + const UDataInfo *pInfo); + + +/** + * Convenience function. + * This function works the same as udata_openChoice + * except that any data that matches the type and name + * is assumed to be acceptable. + * @param path Specifies an absolute path and/or a basename for the + * finding of the data in the file system. + * NULL for ICU data. + * @param type A string that specifies the type of data to be loaded. + * For example, resource bundles are loaded with type "res", + * conversion tables with type "cnv". + * This may be NULL or empty. + * @param name A string that specifies the name of the data. + * @param pErrorCode An ICU UErrorCode parameter. It must not be NULL. + * @return A pointer (handle) to a data memory object, or NULL + * if an error occurs. Call udata_getMemory() + * to get a pointer to the actual data. + * + * @see udata_openChoice + * @stable ICU 2.0 + */ +U_CAPI UDataMemory * U_EXPORT2 +udata_open(const char *path, const char *type, const char *name, + UErrorCode *pErrorCode); + +/** + * Data loading function. + * This function is used to find and load efficiently data for + * ICU and applications using ICU. + * It provides an abstract interface that allows to specify a data + * type and name to find and load the data. + * + *

The implementation depends on platform properties and user preferences + * and may involve loading shared libraries (DLLs), mapping + * files into memory, or fopen()/fread() files. + * It may also involve using static memory or database queries etc. + * Several or all data items may be combined into one entity + * (DLL, memory-mappable file).

+ * + *

The data is always preceded by a header that includes + * a UDataInfo structure. + * The caller's isAcceptable() function is called to make + * sure that the data is useful. It may be called several times if it + * rejects the data and there is more than one location with data + * matching the type and name.

+ * + *

If path==NULL, then ICU data is loaded. + * Otherwise, it is separated into a basename and a basename-less directory string. + * The basename is used as the data package name, and the directory is + * logically prepended to the ICU data directory string.

+ * + *

For details about ICU data loading see the User Guide + * Data Management chapter. (https://unicode-org.github.io/icu/userguide/icu_data/)

+ * + * @param path Specifies an absolute path and/or a basename for the + * finding of the data in the file system. + * NULL for ICU data. + * @param type A string that specifies the type of data to be loaded. + * For example, resource bundles are loaded with type "res", + * conversion tables with type "cnv". + * This may be NULL or empty. + * @param name A string that specifies the name of the data. + * @param isAcceptable This function is called to verify that loaded data + * is useful for the client code. If it returns false + * for all data items, then udata_openChoice() + * will return with an error. + * @param context Arbitrary parameter to be passed into isAcceptable. + * @param pErrorCode An ICU UErrorCode parameter. It must not be NULL. + * @return A pointer (handle) to a data memory object, or NULL + * if an error occurs. Call udata_getMemory() + * to get a pointer to the actual data. + * @stable ICU 2.0 + */ +U_CAPI UDataMemory * U_EXPORT2 +udata_openChoice(const char *path, const char *type, const char *name, + UDataMemoryIsAcceptable *isAcceptable, void *context, + UErrorCode *pErrorCode); + +/** + * Close the data memory. + * This function must be called to allow the system to + * release resources associated with this data memory. + * @param pData The pointer to data memory object + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +udata_close(UDataMemory *pData); + +/** + * Get the pointer to the actual data inside the data memory. + * The data is read-only. + * + * ICU data must be at least 8-aligned, and should be 16-aligned. + * + * @param pData The pointer to data memory object + * @stable ICU 2.0 + */ +U_CAPI const void * U_EXPORT2 +udata_getMemory(UDataMemory *pData); + +/** + * Get the information from the data memory header. + * This allows to get access to the header containing + * platform data properties etc. which is not part of + * the data itself and can therefore not be accessed + * via the pointer that udata_getMemory() returns. + * + * @param pData pointer to the data memory object + * @param pInfo pointer to a UDataInfo object; + * its size field must be set correctly, + * typically to sizeof(UDataInfo). + * + * *pInfo will be filled with the UDataInfo structure + * in the data memory object. If this structure is smaller than + * pInfo->size, then the size will be + * adjusted and only part of the structure will be filled. + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +udata_getInfo(UDataMemory *pData, UDataInfo *pInfo); + +/** + * This function bypasses the normal ICU data loading process and + * allows you to force ICU's system data to come out of a user-specified + * area in memory. + * + * ICU data must be at least 8-aligned, and should be 16-aligned. + * See https://unicode-org.github.io/icu/userguide/icu_data + * + * The format of this data is that of the icu common data file, as is + * generated by the pkgdata tool with mode=common or mode=dll. + * You can read in a whole common mode file and pass the address to the start of the + * data, or (with the appropriate link options) pass in the pointer to + * the data that has been loaded from a dll by the operating system, + * as shown in this code: + * + * extern const char U_IMPORT U_ICUDATA_ENTRY_POINT []; + * // U_ICUDATA_ENTRY_POINT is same as entry point specified to pkgdata tool + * UErrorCode status = U_ZERO_ERROR; + * + * udata_setCommonData(&U_ICUDATA_ENTRY_POINT, &status); + * + * It is important that the declaration be as above. The entry point + * must not be declared as an extern void*. + * + * Starting with ICU 4.4, it is possible to set several data packages, + * one per call to this function. + * udata_open() will look for data in the multiple data packages in the order + * in which they were set. + * The position of the linked-in or default-name ICU .data package in the + * search list depends on when the first data item is loaded that is not contained + * in the already explicitly set packages. + * If data was loaded implicitly before the first call to this function + * (for example, via opening a converter, constructing a UnicodeString + * from default-codepage data, using formatting or collation APIs, etc.), + * then the default data will be first in the list. + * + * This function has no effect on application (non ICU) data. See udata_setAppData() + * for similar functionality for application data. + * + * @param data pointer to ICU common data + * @param err outgoing error status U_USING_DEFAULT_WARNING, U_UNSUPPORTED_ERROR + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +udata_setCommonData(const void *data, UErrorCode *err); + + +/** + * This function bypasses the normal ICU data loading process for application-specific + * data and allows you to force the it to come out of a user-specified + * pointer. + * + * ICU data must be at least 8-aligned, and should be 16-aligned. + * See https://unicode-org.github.io/icu/userguide/icu_data + * + * The format of this data is that of the icu common data file, like 'icudt26l.dat' + * or the corresponding shared library (DLL) file. + * The application must read in or otherwise construct an image of the data and then + * pass the address of it to this function. + * + * + * Warning: setAppData will set a U_USING_DEFAULT_WARNING code if + * data with the specified path that has already been opened, or + * if setAppData with the same path has already been called. + * Any such calls to setAppData will have no effect. + * + * + * @param packageName the package name by which the application will refer + * to (open) this data + * @param data pointer to the data + * @param err outgoing error status U_USING_DEFAULT_WARNING, U_UNSUPPORTED_ERROR + * @see udata_setCommonData + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +udata_setAppData(const char *packageName, const void *data, UErrorCode *err); + +/** + * Possible settings for udata_setFileAccess() + * @see udata_setFileAccess + * @stable ICU 3.4 + */ +typedef enum UDataFileAccess { + /** ICU looks for data in single files first, then in packages. (default) @stable ICU 3.4 */ + UDATA_FILES_FIRST, + /** An alias for the default access mode. @stable ICU 3.4 */ + UDATA_DEFAULT_ACCESS = UDATA_FILES_FIRST, + /** ICU only loads data from packages, not from single files. @stable ICU 3.4 */ + UDATA_ONLY_PACKAGES, + /** ICU loads data from packages first, and only from single files + if the data cannot be found in a package. @stable ICU 3.4 */ + UDATA_PACKAGES_FIRST, + /** ICU does not access the file system for data loading. @stable ICU 3.4 */ + UDATA_NO_FILES, +#ifndef U_HIDE_DEPRECATED_API + /** + * Number of real UDataFileAccess values. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + UDATA_FILE_ACCESS_COUNT +#endif // U_HIDE_DEPRECATED_API +} UDataFileAccess; + +/** + * This function may be called to control how ICU loads data. It must be called + * before any ICU data is loaded, including application data loaded with + * ures/ResourceBundle or udata APIs. This function is not multithread safe. + * The results of calling it while other threads are loading data are undefined. + * @param access The type of file access to be used + * @param status Error code. + * @see UDataFileAccess + * @stable ICU 3.4 + */ +U_CAPI void U_EXPORT2 +udata_setFileAccess(UDataFileAccess access, UErrorCode *status); + +U_CDECL_END + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUDataMemoryPointer + * "Smart pointer" class, closes a UDataMemory via udata_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUDataMemoryPointer, UDataMemory, udata_close); + +U_NAMESPACE_END + +#endif // U_SHOW_CPLUSPLUS_API + +#endif diff --git a/intl/icu/source/common/unicode/udisplaycontext.h b/intl/icu/source/common/unicode/udisplaycontext.h new file mode 100644 index 0000000000..6e14217980 --- /dev/null +++ b/intl/icu/source/common/unicode/udisplaycontext.h @@ -0,0 +1,173 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +***************************************************************************************** +* Copyright (C) 2014-2016, International Business Machines +* Corporation and others. All Rights Reserved. +***************************************************************************************** +*/ + +#ifndef UDISPLAYCONTEXT_H +#define UDISPLAYCONTEXT_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_FORMATTING + +/** + * \file + * \brief C API: Display context types (enum values) + */ + +/** + * Display context types, for getting values of a particular setting. + * Note, the specific numeric values are internal and may change. + * @stable ICU 51 + */ +enum UDisplayContextType { + /** + * Type to retrieve the dialect handling setting, e.g. + * UDISPCTX_STANDARD_NAMES or UDISPCTX_DIALECT_NAMES. + * @stable ICU 51 + */ + UDISPCTX_TYPE_DIALECT_HANDLING = 0, + /** + * Type to retrieve the capitalization context setting, e.g. + * UDISPCTX_CAPITALIZATION_NONE, UDISPCTX_CAPITALIZATION_FOR_MIDDLE_OF_SENTENCE, + * UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE, etc. + * @stable ICU 51 + */ + UDISPCTX_TYPE_CAPITALIZATION = 1, + /** + * Type to retrieve the display length setting, e.g. + * UDISPCTX_LENGTH_FULL, UDISPCTX_LENGTH_SHORT. + * @stable ICU 54 + */ + UDISPCTX_TYPE_DISPLAY_LENGTH = 2, + /** + * Type to retrieve the substitute handling setting, e.g. + * UDISPCTX_SUBSTITUTE, UDISPCTX_NO_SUBSTITUTE. + * @stable ICU 58 + */ + UDISPCTX_TYPE_SUBSTITUTE_HANDLING = 3 +}; +/** +* @stable ICU 51 +*/ +typedef enum UDisplayContextType UDisplayContextType; + +/** + * Display context settings. + * Note, the specific numeric values are internal and may change. + * @stable ICU 51 + */ +enum UDisplayContext { + /** + * ================================ + * DIALECT_HANDLING can be set to one of UDISPCTX_STANDARD_NAMES or + * UDISPCTX_DIALECT_NAMES. Use UDisplayContextType UDISPCTX_TYPE_DIALECT_HANDLING + * to get the value. + */ + /** + * A possible setting for DIALECT_HANDLING: + * use standard names when generating a locale name, + * e.g. en_GB displays as 'English (United Kingdom)'. + * @stable ICU 51 + */ + UDISPCTX_STANDARD_NAMES = (UDISPCTX_TYPE_DIALECT_HANDLING<<8) + 0, + /** + * A possible setting for DIALECT_HANDLING: + * use dialect names, when generating a locale name, + * e.g. en_GB displays as 'British English'. + * @stable ICU 51 + */ + UDISPCTX_DIALECT_NAMES = (UDISPCTX_TYPE_DIALECT_HANDLING<<8) + 1, + /** + * ================================ + * CAPITALIZATION can be set to one of UDISPCTX_CAPITALIZATION_NONE, + * UDISPCTX_CAPITALIZATION_FOR_MIDDLE_OF_SENTENCE, + * UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE, + * UDISPCTX_CAPITALIZATION_FOR_UI_LIST_OR_MENU, or + * UDISPCTX_CAPITALIZATION_FOR_STANDALONE. + * Use UDisplayContextType UDISPCTX_TYPE_CAPITALIZATION to get the value. + */ + /** + * The capitalization context to be used is unknown (this is the default value). + * @stable ICU 51 + */ + UDISPCTX_CAPITALIZATION_NONE = (UDISPCTX_TYPE_CAPITALIZATION<<8) + 0, + /** + * The capitalization context if a date, date symbol or display name is to be + * formatted with capitalization appropriate for the middle of a sentence. + * @stable ICU 51 + */ + UDISPCTX_CAPITALIZATION_FOR_MIDDLE_OF_SENTENCE = (UDISPCTX_TYPE_CAPITALIZATION<<8) + 1, + /** + * The capitalization context if a date, date symbol or display name is to be + * formatted with capitalization appropriate for the beginning of a sentence. + * @stable ICU 51 + */ + UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE = (UDISPCTX_TYPE_CAPITALIZATION<<8) + 2, + /** + * The capitalization context if a date, date symbol or display name is to be + * formatted with capitalization appropriate for a user-interface list or menu item. + * @stable ICU 51 + */ + UDISPCTX_CAPITALIZATION_FOR_UI_LIST_OR_MENU = (UDISPCTX_TYPE_CAPITALIZATION<<8) + 3, + /** + * The capitalization context if a date, date symbol or display name is to be + * formatted with capitalization appropriate for stand-alone usage such as an + * isolated name on a calendar page. + * @stable ICU 51 + */ + UDISPCTX_CAPITALIZATION_FOR_STANDALONE = (UDISPCTX_TYPE_CAPITALIZATION<<8) + 4, + /** + * ================================ + * DISPLAY_LENGTH can be set to one of UDISPCTX_LENGTH_FULL or + * UDISPCTX_LENGTH_SHORT. Use UDisplayContextType UDISPCTX_TYPE_DISPLAY_LENGTH + * to get the value. + */ + /** + * A possible setting for DISPLAY_LENGTH: + * use full names when generating a locale name, + * e.g. "United States" for US. + * @stable ICU 54 + */ + UDISPCTX_LENGTH_FULL = (UDISPCTX_TYPE_DISPLAY_LENGTH<<8) + 0, + /** + * A possible setting for DISPLAY_LENGTH: + * use short names when generating a locale name, + * e.g. "U.S." for US. + * @stable ICU 54 + */ + UDISPCTX_LENGTH_SHORT = (UDISPCTX_TYPE_DISPLAY_LENGTH<<8) + 1, + /** + * ================================ + * SUBSTITUTE_HANDLING can be set to one of UDISPCTX_SUBSTITUTE or + * UDISPCTX_NO_SUBSTITUTE. Use UDisplayContextType UDISPCTX_TYPE_SUBSTITUTE_HANDLING + * to get the value. + */ + /** + * A possible setting for SUBSTITUTE_HANDLING: + * Returns a fallback value (e.g., the input code) when no data is available. + * This is the default value. + * @stable ICU 58 + */ + UDISPCTX_SUBSTITUTE = (UDISPCTX_TYPE_SUBSTITUTE_HANDLING<<8) + 0, + /** + * A possible setting for SUBSTITUTE_HANDLING: + * Returns a null value with error code set to U_ILLEGAL_ARGUMENT_ERROR when no + * data is available. + * @stable ICU 58 + */ + UDISPCTX_NO_SUBSTITUTE = (UDISPCTX_TYPE_SUBSTITUTE_HANDLING<<8) + 1 + +}; +/** +* @stable ICU 51 +*/ +typedef enum UDisplayContext UDisplayContext; + +#endif /* #if !UCONFIG_NO_FORMATTING */ + +#endif diff --git a/intl/icu/source/common/unicode/uenum.h b/intl/icu/source/common/unicode/uenum.h new file mode 100644 index 0000000000..d9c893e06d --- /dev/null +++ b/intl/icu/source/common/unicode/uenum.h @@ -0,0 +1,209 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2002-2013, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: uenum.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:2 +* +* created on: 2002jul08 +* created by: Vladimir Weinstein +*/ + +#ifndef __UENUM_H +#define __UENUM_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" + +U_NAMESPACE_BEGIN +class StringEnumeration; +U_NAMESPACE_END +#endif // U_SHOW_CPLUSPLUS_API + +/** + * \file + * \brief C API: String Enumeration + */ + +/** + * An enumeration object. + * For usage in C programs. + * @stable ICU 2.2 + */ +struct UEnumeration; +/** structure representing an enumeration object instance @stable ICU 2.2 */ +typedef struct UEnumeration UEnumeration; + +/** + * Disposes of resources in use by the iterator. If en is NULL, + * does nothing. After this call, any char* or UChar* pointer + * returned by uenum_unext() or uenum_next() is invalid. + * @param en UEnumeration structure pointer + * @stable ICU 2.2 + */ +U_CAPI void U_EXPORT2 +uenum_close(UEnumeration* en); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUEnumerationPointer + * "Smart pointer" class, closes a UEnumeration via uenum_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUEnumerationPointer, UEnumeration, uenum_close); + +U_NAMESPACE_END + +#endif + +/** + * Returns the number of elements that the iterator traverses. If + * the iterator is out-of-sync with its service, status is set to + * U_ENUM_OUT_OF_SYNC_ERROR. + * This is a convenience function. It can end up being very + * expensive as all the items might have to be pre-fetched (depending + * on the type of data being traversed). Use with caution and only + * when necessary. + * @param en UEnumeration structure pointer + * @param status error code, can be U_ENUM_OUT_OF_SYNC_ERROR if the + * iterator is out of sync. + * @return number of elements in the iterator + * @stable ICU 2.2 + */ +U_CAPI int32_t U_EXPORT2 +uenum_count(UEnumeration* en, UErrorCode* status); + +/** + * Returns the next element in the iterator's list. If there are + * no more elements, returns NULL. If the iterator is out-of-sync + * with its service, status is set to U_ENUM_OUT_OF_SYNC_ERROR and + * NULL is returned. If the native service string is a char* string, + * it is converted to UChar* with the invariant converter. + * The result is terminated by (UChar)0. + * @param en the iterator object + * @param resultLength pointer to receive the length of the result + * (not including the terminating \\0). + * If the pointer is NULL it is ignored. + * @param status the error code, set to U_ENUM_OUT_OF_SYNC_ERROR if + * the iterator is out of sync with its service. + * @return a pointer to the string. The string will be + * zero-terminated. The return pointer is owned by this iterator + * and must not be deleted by the caller. The pointer is valid + * until the next call to any uenum_... method, including + * uenum_next() or uenum_unext(). When all strings have been + * traversed, returns NULL. + * @stable ICU 2.2 + */ +U_CAPI const UChar* U_EXPORT2 +uenum_unext(UEnumeration* en, + int32_t* resultLength, + UErrorCode* status); + +/** + * Returns the next element in the iterator's list. If there are + * no more elements, returns NULL. If the iterator is out-of-sync + * with its service, status is set to U_ENUM_OUT_OF_SYNC_ERROR and + * NULL is returned. If the native service string is a UChar* + * string, it is converted to char* with the invariant converter. + * The result is terminated by (char)0. If the conversion fails + * (because a character cannot be converted) then status is set to + * U_INVARIANT_CONVERSION_ERROR and the return value is undefined + * (but non-NULL). + * @param en the iterator object + * @param resultLength pointer to receive the length of the result + * (not including the terminating \\0). + * If the pointer is NULL it is ignored. + * @param status the error code, set to U_ENUM_OUT_OF_SYNC_ERROR if + * the iterator is out of sync with its service. Set to + * U_INVARIANT_CONVERSION_ERROR if the underlying native string is + * UChar* and conversion to char* with the invariant converter + * fails. This error pertains only to current string, so iteration + * might be able to continue successfully. + * @return a pointer to the string. The string will be + * zero-terminated. The return pointer is owned by this iterator + * and must not be deleted by the caller. The pointer is valid + * until the next call to any uenum_... method, including + * uenum_next() or uenum_unext(). When all strings have been + * traversed, returns NULL. + * @stable ICU 2.2 + */ +U_CAPI const char* U_EXPORT2 +uenum_next(UEnumeration* en, + int32_t* resultLength, + UErrorCode* status); + +/** + * Resets the iterator to the current list of service IDs. This + * re-establishes sync with the service and rewinds the iterator + * to start at the first element. + * @param en the iterator object + * @param status the error code, set to U_ENUM_OUT_OF_SYNC_ERROR if + * the iterator is out of sync with its service. + * @stable ICU 2.2 + */ +U_CAPI void U_EXPORT2 +uenum_reset(UEnumeration* en, UErrorCode* status); + +#if U_SHOW_CPLUSPLUS_API + +/** + * Given a StringEnumeration, wrap it in a UEnumeration. The + * StringEnumeration is adopted; after this call, the caller must not + * delete it (regardless of error status). + * @param adopted the C++ StringEnumeration to be wrapped in a UEnumeration. + * @param ec the error code. + * @return a UEnumeration wrapping the adopted StringEnumeration. + * @stable ICU 4.2 + */ +U_CAPI UEnumeration* U_EXPORT2 +uenum_openFromStringEnumeration(icu::StringEnumeration* adopted, UErrorCode* ec); + +#endif + +/** + * Given an array of const UChar* strings, return a UEnumeration. String pointers from 0..count-1 must not be null. + * Do not free or modify either the string array or the characters it points to until this object has been destroyed with uenum_close. + * \snippet test/cintltst/uenumtst.c uenum_openUCharStringsEnumeration + * @param strings array of const UChar* strings (each null terminated). All storage is owned by the caller. + * @param count length of the array + * @param ec error code + * @return the new UEnumeration object. Caller is responsible for calling uenum_close to free memory. + * @see uenum_close + * @stable ICU 50 + */ +U_CAPI UEnumeration* U_EXPORT2 +uenum_openUCharStringsEnumeration(const UChar* const strings[], int32_t count, + UErrorCode* ec); + +/** + * Given an array of const char* strings (invariant chars only), return a UEnumeration. String pointers from 0..count-1 must not be null. + * Do not free or modify either the string array or the characters it points to until this object has been destroyed with uenum_close. + * \snippet test/cintltst/uenumtst.c uenum_openCharStringsEnumeration + * @param strings array of char* strings (each null terminated). All storage is owned by the caller. + * @param count length of the array + * @param ec error code + * @return the new UEnumeration object. Caller is responsible for calling uenum_close to free memory + * @see uenum_close + * @stable ICU 50 + */ +U_CAPI UEnumeration* U_EXPORT2 +uenum_openCharStringsEnumeration(const char* const strings[], int32_t count, + UErrorCode* ec); + +#endif diff --git a/intl/icu/source/common/unicode/uidna.h b/intl/icu/source/common/unicode/uidna.h new file mode 100644 index 0000000000..24a81ceadd --- /dev/null +++ b/intl/icu/source/common/unicode/uidna.h @@ -0,0 +1,776 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ******************************************************************************* + * + * Copyright (C) 2003-2014, International Business Machines + * Corporation and others. All Rights Reserved. + * + ******************************************************************************* + * file name: uidna.h + * encoding: UTF-8 + * tab size: 8 (not used) + * indentation:4 + * + * created on: 2003feb1 + * created by: Ram Viswanadha + */ + +#ifndef __UIDNA_H__ +#define __UIDNA_H__ + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_IDNA + +#include +#include "unicode/parseerr.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +/** + * \file + * \brief C API: Internationalizing Domain Names in Applications (IDNA) + * + * IDNA2008 is implemented according to UTS #46, see the IDNA C++ class in idna.h. + * + * The C API functions which do take a UIDNA * service object pointer + * implement UTS #46 and IDNA2008. + * + * IDNA2003 is obsolete. + * The C API functions which do not take a service object pointer + * implement IDNA2003. They are all deprecated. + */ + +/* + * IDNA option bit set values. + */ +enum { + /** + * Default options value: None of the other options are set. + * For use in static worker and factory methods. + * @stable ICU 2.6 + */ + UIDNA_DEFAULT=0, +#ifndef U_HIDE_DEPRECATED_API + /** + * Option to allow unassigned code points in domain names and labels. + * For use in static worker and factory methods. + *

This option is ignored by the UTS46 implementation. + * (UTS #46 disallows unassigned code points.) + * @deprecated ICU 55 Use UTS #46 instead via uidna_openUTS46() or class IDNA. + */ + UIDNA_ALLOW_UNASSIGNED=1, +#endif /* U_HIDE_DEPRECATED_API */ + /** + * Option to check whether the input conforms to the STD3 ASCII rules, + * for example the restriction of labels to LDH characters + * (ASCII Letters, Digits and Hyphen-Minus). + * For use in static worker and factory methods. + * @stable ICU 2.6 + */ + UIDNA_USE_STD3_RULES=2, + /** + * IDNA option to check for whether the input conforms to the BiDi rules. + * For use in static worker and factory methods. + *

This option is ignored by the IDNA2003 implementation. + * (IDNA2003 always performs a BiDi check.) + * @stable ICU 4.6 + */ + UIDNA_CHECK_BIDI=4, + /** + * IDNA option to check for whether the input conforms to the CONTEXTJ rules. + * For use in static worker and factory methods. + *

This option is ignored by the IDNA2003 implementation. + * (The CONTEXTJ check is new in IDNA2008.) + * @stable ICU 4.6 + */ + UIDNA_CHECK_CONTEXTJ=8, + /** + * IDNA option for nontransitional processing in ToASCII(). + * For use in static worker and factory methods. + *

By default, ToASCII() uses transitional processing. + *

This option is ignored by the IDNA2003 implementation. + * (This is only relevant for compatibility of newer IDNA implementations with IDNA2003.) + * @stable ICU 4.6 + */ + UIDNA_NONTRANSITIONAL_TO_ASCII=0x10, + /** + * IDNA option for nontransitional processing in ToUnicode(). + * For use in static worker and factory methods. + *

By default, ToUnicode() uses transitional processing. + *

This option is ignored by the IDNA2003 implementation. + * (This is only relevant for compatibility of newer IDNA implementations with IDNA2003.) + * @stable ICU 4.6 + */ + UIDNA_NONTRANSITIONAL_TO_UNICODE=0x20, + /** + * IDNA option to check for whether the input conforms to the CONTEXTO rules. + * For use in static worker and factory methods. + *

This option is ignored by the IDNA2003 implementation. + * (The CONTEXTO check is new in IDNA2008.) + *

This is for use by registries for IDNA2008 conformance. + * UTS #46 does not require the CONTEXTO check. + * @stable ICU 49 + */ + UIDNA_CHECK_CONTEXTO=0x40 +}; + +/** + * Opaque C service object type for the new IDNA API. + * @stable ICU 4.6 + */ +struct UIDNA; +typedef struct UIDNA UIDNA; /**< C typedef for struct UIDNA. @stable ICU 4.6 */ + +/** + * Returns a UIDNA instance which implements UTS #46. + * Returns an unmodifiable instance, owned by the caller. + * Cache it for multiple operations, and uidna_close() it when done. + * The instance is thread-safe, that is, it can be used concurrently. + * + * For details about the UTS #46 implementation see the IDNA C++ class in idna.h. + * + * @param options Bit set to modify the processing and error checking. + * See option bit set values in uidna.h. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the UTS #46 UIDNA instance, if successful + * @stable ICU 4.6 + */ +U_CAPI UIDNA * U_EXPORT2 +uidna_openUTS46(uint32_t options, UErrorCode *pErrorCode); + +/** + * Closes a UIDNA instance. + * @param idna UIDNA instance to be closed + * @stable ICU 4.6 + */ +U_CAPI void U_EXPORT2 +uidna_close(UIDNA *idna); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUIDNAPointer + * "Smart pointer" class, closes a UIDNA via uidna_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.6 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUIDNAPointer, UIDNA, uidna_close); + +U_NAMESPACE_END + +#endif + +/** + * Output container for IDNA processing errors. + * Initialize with UIDNA_INFO_INITIALIZER: + * \code + * UIDNAInfo info = UIDNA_INFO_INITIALIZER; + * int32_t length = uidna_nameToASCII(..., &info, &errorCode); + * if(U_SUCCESS(errorCode) && info.errors!=0) { ... } + * \endcode + * @stable ICU 4.6 + */ +typedef struct UIDNAInfo { + /** sizeof(UIDNAInfo) @stable ICU 4.6 */ + int16_t size; + /** + * Set to true if transitional and nontransitional processing produce different results. + * For details see C++ IDNAInfo::isTransitionalDifferent(). + * @stable ICU 4.6 + */ + UBool isTransitionalDifferent; + UBool reservedB3; /**< Reserved field, do not use. @internal */ + /** + * Bit set indicating IDNA processing errors. 0 if no errors. + * See UIDNA_ERROR_... constants. + * @stable ICU 4.6 + */ + uint32_t errors; + int32_t reservedI2; /**< Reserved field, do not use. @internal */ + int32_t reservedI3; /**< Reserved field, do not use. @internal */ +} UIDNAInfo; + +/** + * Static initializer for a UIDNAInfo struct. + * @stable ICU 4.6 + */ +#define UIDNA_INFO_INITIALIZER { \ + (int16_t)sizeof(UIDNAInfo), \ + false, false, \ + 0, 0, 0 } + +/** + * Converts a single domain name label into its ASCII form for DNS lookup. + * If any processing step fails, then pInfo->errors will be non-zero and + * the result might not be an ASCII string. + * The label might be modified according to the types of errors. + * Labels with severe errors will be left in (or turned into) their Unicode form. + * + * The UErrorCode indicates an error only in exceptional cases, + * such as a U_MEMORY_ALLOCATION_ERROR. + * + * @param idna UIDNA instance + * @param label Input domain name label + * @param length Label length, or -1 if NUL-terminated + * @param dest Destination string buffer + * @param capacity Destination buffer capacity + * @param pInfo Output container of IDNA processing details. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return destination string length + * @stable ICU 4.6 + */ +U_CAPI int32_t U_EXPORT2 +uidna_labelToASCII(const UIDNA *idna, + const UChar *label, int32_t length, + UChar *dest, int32_t capacity, + UIDNAInfo *pInfo, UErrorCode *pErrorCode); + +/** + * Converts a single domain name label into its Unicode form for human-readable display. + * If any processing step fails, then pInfo->errors will be non-zero. + * The label might be modified according to the types of errors. + * + * The UErrorCode indicates an error only in exceptional cases, + * such as a U_MEMORY_ALLOCATION_ERROR. + * + * @param idna UIDNA instance + * @param label Input domain name label + * @param length Label length, or -1 if NUL-terminated + * @param dest Destination string buffer + * @param capacity Destination buffer capacity + * @param pInfo Output container of IDNA processing details. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return destination string length + * @stable ICU 4.6 + */ +U_CAPI int32_t U_EXPORT2 +uidna_labelToUnicode(const UIDNA *idna, + const UChar *label, int32_t length, + UChar *dest, int32_t capacity, + UIDNAInfo *pInfo, UErrorCode *pErrorCode); + +/** + * Converts a whole domain name into its ASCII form for DNS lookup. + * If any processing step fails, then pInfo->errors will be non-zero and + * the result might not be an ASCII string. + * The domain name might be modified according to the types of errors. + * Labels with severe errors will be left in (or turned into) their Unicode form. + * + * The UErrorCode indicates an error only in exceptional cases, + * such as a U_MEMORY_ALLOCATION_ERROR. + * + * @param idna UIDNA instance + * @param name Input domain name + * @param length Domain name length, or -1 if NUL-terminated + * @param dest Destination string buffer + * @param capacity Destination buffer capacity + * @param pInfo Output container of IDNA processing details. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return destination string length + * @stable ICU 4.6 + */ +U_CAPI int32_t U_EXPORT2 +uidna_nameToASCII(const UIDNA *idna, + const UChar *name, int32_t length, + UChar *dest, int32_t capacity, + UIDNAInfo *pInfo, UErrorCode *pErrorCode); + +/** + * Converts a whole domain name into its Unicode form for human-readable display. + * If any processing step fails, then pInfo->errors will be non-zero. + * The domain name might be modified according to the types of errors. + * + * The UErrorCode indicates an error only in exceptional cases, + * such as a U_MEMORY_ALLOCATION_ERROR. + * + * @param idna UIDNA instance + * @param name Input domain name + * @param length Domain name length, or -1 if NUL-terminated + * @param dest Destination string buffer + * @param capacity Destination buffer capacity + * @param pInfo Output container of IDNA processing details. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return destination string length + * @stable ICU 4.6 + */ +U_CAPI int32_t U_EXPORT2 +uidna_nameToUnicode(const UIDNA *idna, + const UChar *name, int32_t length, + UChar *dest, int32_t capacity, + UIDNAInfo *pInfo, UErrorCode *pErrorCode); + +/* UTF-8 versions of the processing methods --------------------------------- */ + +/** + * Converts a single domain name label into its ASCII form for DNS lookup. + * UTF-8 version of uidna_labelToASCII(), same behavior. + * + * @param idna UIDNA instance + * @param label Input domain name label + * @param length Label length, or -1 if NUL-terminated + * @param dest Destination string buffer + * @param capacity Destination buffer capacity + * @param pInfo Output container of IDNA processing details. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return destination string length + * @stable ICU 4.6 + */ +U_CAPI int32_t U_EXPORT2 +uidna_labelToASCII_UTF8(const UIDNA *idna, + const char *label, int32_t length, + char *dest, int32_t capacity, + UIDNAInfo *pInfo, UErrorCode *pErrorCode); + +/** + * Converts a single domain name label into its Unicode form for human-readable display. + * UTF-8 version of uidna_labelToUnicode(), same behavior. + * + * @param idna UIDNA instance + * @param label Input domain name label + * @param length Label length, or -1 if NUL-terminated + * @param dest Destination string buffer + * @param capacity Destination buffer capacity + * @param pInfo Output container of IDNA processing details. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return destination string length + * @stable ICU 4.6 + */ +U_CAPI int32_t U_EXPORT2 +uidna_labelToUnicodeUTF8(const UIDNA *idna, + const char *label, int32_t length, + char *dest, int32_t capacity, + UIDNAInfo *pInfo, UErrorCode *pErrorCode); + +/** + * Converts a whole domain name into its ASCII form for DNS lookup. + * UTF-8 version of uidna_nameToASCII(), same behavior. + * + * @param idna UIDNA instance + * @param name Input domain name + * @param length Domain name length, or -1 if NUL-terminated + * @param dest Destination string buffer + * @param capacity Destination buffer capacity + * @param pInfo Output container of IDNA processing details. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return destination string length + * @stable ICU 4.6 + */ +U_CAPI int32_t U_EXPORT2 +uidna_nameToASCII_UTF8(const UIDNA *idna, + const char *name, int32_t length, + char *dest, int32_t capacity, + UIDNAInfo *pInfo, UErrorCode *pErrorCode); + +/** + * Converts a whole domain name into its Unicode form for human-readable display. + * UTF-8 version of uidna_nameToUnicode(), same behavior. + * + * @param idna UIDNA instance + * @param name Input domain name + * @param length Domain name length, or -1 if NUL-terminated + * @param dest Destination string buffer + * @param capacity Destination buffer capacity + * @param pInfo Output container of IDNA processing details. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return destination string length + * @stable ICU 4.6 + */ +U_CAPI int32_t U_EXPORT2 +uidna_nameToUnicodeUTF8(const UIDNA *idna, + const char *name, int32_t length, + char *dest, int32_t capacity, + UIDNAInfo *pInfo, UErrorCode *pErrorCode); + +/* + * IDNA error bit set values. + * When a domain name or label fails a processing step or does not meet the + * validity criteria, then one or more of these error bits are set. + */ +enum { + /** + * A non-final domain name label (or the whole domain name) is empty. + * @stable ICU 4.6 + */ + UIDNA_ERROR_EMPTY_LABEL=1, + /** + * A domain name label is longer than 63 bytes. + * (See STD13/RFC1034 3.1. Name space specifications and terminology.) + * This is only checked in ToASCII operations, and only if the output label is all-ASCII. + * @stable ICU 4.6 + */ + UIDNA_ERROR_LABEL_TOO_LONG=2, + /** + * A domain name is longer than 255 bytes in its storage form. + * (See STD13/RFC1034 3.1. Name space specifications and terminology.) + * This is only checked in ToASCII operations, and only if the output domain name is all-ASCII. + * @stable ICU 4.6 + */ + UIDNA_ERROR_DOMAIN_NAME_TOO_LONG=4, + /** + * A label starts with a hyphen-minus ('-'). + * @stable ICU 4.6 + */ + UIDNA_ERROR_LEADING_HYPHEN=8, + /** + * A label ends with a hyphen-minus ('-'). + * @stable ICU 4.6 + */ + UIDNA_ERROR_TRAILING_HYPHEN=0x10, + /** + * A label contains hyphen-minus ('-') in the third and fourth positions. + * @stable ICU 4.6 + */ + UIDNA_ERROR_HYPHEN_3_4=0x20, + /** + * A label starts with a combining mark. + * @stable ICU 4.6 + */ + UIDNA_ERROR_LEADING_COMBINING_MARK=0x40, + /** + * A label or domain name contains disallowed characters. + * @stable ICU 4.6 + */ + UIDNA_ERROR_DISALLOWED=0x80, + /** + * A label starts with "xn--" but does not contain valid Punycode. + * That is, an xn-- label failed Punycode decoding. + * @stable ICU 4.6 + */ + UIDNA_ERROR_PUNYCODE=0x100, + /** + * A label contains a dot=full stop. + * This can occur in an input string for a single-label function. + * @stable ICU 4.6 + */ + UIDNA_ERROR_LABEL_HAS_DOT=0x200, + /** + * An ACE label does not contain a valid label string. + * The label was successfully ACE (Punycode) decoded but the resulting + * string had severe validation errors. For example, + * it might contain characters that are not allowed in ACE labels, + * or it might not be normalized. + * @stable ICU 4.6 + */ + UIDNA_ERROR_INVALID_ACE_LABEL=0x400, + /** + * A label does not meet the IDNA BiDi requirements (for right-to-left characters). + * @stable ICU 4.6 + */ + UIDNA_ERROR_BIDI=0x800, + /** + * A label does not meet the IDNA CONTEXTJ requirements. + * @stable ICU 4.6 + */ + UIDNA_ERROR_CONTEXTJ=0x1000, + /** + * A label does not meet the IDNA CONTEXTO requirements for punctuation characters. + * Some punctuation characters "Would otherwise have been DISALLOWED" + * but are allowed in certain contexts. (RFC 5892) + * @stable ICU 49 + */ + UIDNA_ERROR_CONTEXTO_PUNCTUATION=0x2000, + /** + * A label does not meet the IDNA CONTEXTO requirements for digits. + * Arabic-Indic Digits (U+066x) must not be mixed with Extended Arabic-Indic Digits (U+06Fx). + * @stable ICU 49 + */ + UIDNA_ERROR_CONTEXTO_DIGITS=0x4000 +}; + +#ifndef U_HIDE_DEPRECATED_API + +/* IDNA2003 API ------------------------------------------------------------- */ + +/** + * IDNA2003: This function implements the ToASCII operation as defined in the IDNA RFC. + * This operation is done on single labels before sending it to something that expects + * ASCII names. A label is an individual part of a domain name. Labels are usually + * separated by dots; e.g. "www.example.com" is composed of 3 labels "www","example", and "com". + * + * IDNA2003 API Overview: + * + * The uidna_ API implements the IDNA protocol as defined in the IDNA RFC + * (http://www.ietf.org/rfc/rfc3490.txt). + * The RFC defines 2 operations: ToASCII and ToUnicode. Domain name labels + * containing non-ASCII code points are processed by the + * ToASCII operation before passing it to resolver libraries. Domain names + * that are obtained from resolver libraries are processed by the + * ToUnicode operation before displaying the domain name to the user. + * IDNA requires that implementations process input strings with Nameprep + * (http://www.ietf.org/rfc/rfc3491.txt), + * which is a profile of Stringprep (http://www.ietf.org/rfc/rfc3454.txt), + * and then with Punycode (http://www.ietf.org/rfc/rfc3492.txt). + * Implementations of IDNA MUST fully implement Nameprep and Punycode; + * neither Nameprep nor Punycode are optional. + * The input and output of ToASCII and ToUnicode operations are Unicode + * and are designed to be chainable, i.e., applying ToASCII or ToUnicode operations + * multiple times to an input string will yield the same result as applying the operation + * once. + * ToUnicode(ToUnicode(ToUnicode...(ToUnicode(string)))) == ToUnicode(string) + * ToASCII(ToASCII(ToASCII...(ToASCII(string))) == ToASCII(string). + * + * @param src Input UChar array containing label in Unicode. + * @param srcLength Number of UChars in src, or -1 if NUL-terminated. + * @param dest Output UChar array with ASCII (ACE encoded) label. + * @param destCapacity Size of dest. + * @param options A bit set of options: + * + * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points + * and do not use STD3 ASCII rules + * If unassigned code points are found the operation fails with + * U_UNASSIGNED_ERROR error code. + * + * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations + * If this option is set, the unassigned code points are in the input + * are treated as normal Unicode code points. + * + * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions + * If this option is set and the input does not satisfy STD3 rules, + * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR + * + * @param parseError Pointer to UParseError struct to receive information on position + * of error if an error is encountered. Can be NULL. + * @param status ICU in/out error code parameter. + * U_INVALID_CHAR_FOUND if src contains + * unmatched single surrogates. + * U_INDEX_OUTOFBOUNDS_ERROR if src contains + * too many code points. + * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough + * @return The length of the result string, if successful - or in case of a buffer overflow, + * in which case it will be greater than destCapacity. + * @deprecated ICU 55 Use UTS #46 instead via uidna_openUTS46() or class IDNA. + */ +U_DEPRECATED int32_t U_EXPORT2 +uidna_toASCII(const UChar* src, int32_t srcLength, + UChar* dest, int32_t destCapacity, + int32_t options, + UParseError* parseError, + UErrorCode* status); + + +/** + * IDNA2003: This function implements the ToUnicode operation as defined in the IDNA RFC. + * This operation is done on single labels before sending it to something that expects + * Unicode names. A label is an individual part of a domain name. Labels are usually + * separated by dots; for e.g. "www.example.com" is composed of 3 labels "www","example", and "com". + * + * @param src Input UChar array containing ASCII (ACE encoded) label. + * @param srcLength Number of UChars in src, or -1 if NUL-terminated. + * @param dest Output Converted UChar array containing Unicode equivalent of label. + * @param destCapacity Size of dest. + * @param options A bit set of options: + * + * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points + * and do not use STD3 ASCII rules + * If unassigned code points are found the operation fails with + * U_UNASSIGNED_ERROR error code. + * + * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations + * If this option is set, the unassigned code points are in the input + * are treated as normal Unicode code points. Note: This option is + * required on toUnicode operation because the RFC mandates + * verification of decoded ACE input by applying toASCII and comparing + * its output with source + * + * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions + * If this option is set and the input does not satisfy STD3 rules, + * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR + * + * @param parseError Pointer to UParseError struct to receive information on position + * of error if an error is encountered. Can be NULL. + * @param status ICU in/out error code parameter. + * U_INVALID_CHAR_FOUND if src contains + * unmatched single surrogates. + * U_INDEX_OUTOFBOUNDS_ERROR if src contains + * too many code points. + * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough + * @return The length of the result string, if successful - or in case of a buffer overflow, + * in which case it will be greater than destCapacity. + * @deprecated ICU 55 Use UTS #46 instead via uidna_openUTS46() or class IDNA. + */ +U_DEPRECATED int32_t U_EXPORT2 +uidna_toUnicode(const UChar* src, int32_t srcLength, + UChar* dest, int32_t destCapacity, + int32_t options, + UParseError* parseError, + UErrorCode* status); + + +/** + * IDNA2003: Convenience function that implements the IDNToASCII operation as defined in the IDNA RFC. + * This operation is done on complete domain names, e.g: "www.example.com". + * It is important to note that this operation can fail. If it fails, then the input + * domain name cannot be used as an Internationalized Domain Name and the application + * should have methods defined to deal with the failure. + * + * Note: IDNA RFC specifies that a conformant application should divide a domain name + * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each, + * and then convert. This function does not offer that level of granularity. The options once + * set will apply to all labels in the domain name + * + * @param src Input UChar array containing IDN in Unicode. + * @param srcLength Number of UChars in src, or -1 if NUL-terminated. + * @param dest Output UChar array with ASCII (ACE encoded) IDN. + * @param destCapacity Size of dest. + * @param options A bit set of options: + * + * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points + * and do not use STD3 ASCII rules + * If unassigned code points are found the operation fails with + * U_UNASSIGNED_CODE_POINT_FOUND error code. + * + * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations + * If this option is set, the unassigned code points are in the input + * are treated as normal Unicode code points. + * + * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions + * If this option is set and the input does not satisfy STD3 rules, + * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR + * + * @param parseError Pointer to UParseError struct to receive information on position + * of error if an error is encountered. Can be NULL. + * @param status ICU in/out error code parameter. + * U_INVALID_CHAR_FOUND if src contains + * unmatched single surrogates. + * U_INDEX_OUTOFBOUNDS_ERROR if src contains + * too many code points. + * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough + * @return The length of the result string, if successful - or in case of a buffer overflow, + * in which case it will be greater than destCapacity. + * @deprecated ICU 55 Use UTS #46 instead via uidna_openUTS46() or class IDNA. + */ +U_DEPRECATED int32_t U_EXPORT2 +uidna_IDNToASCII( const UChar* src, int32_t srcLength, + UChar* dest, int32_t destCapacity, + int32_t options, + UParseError* parseError, + UErrorCode* status); + +/** + * IDNA2003: Convenience function that implements the IDNToUnicode operation as defined in the IDNA RFC. + * This operation is done on complete domain names, e.g: "www.example.com". + * + * Note: IDNA RFC specifies that a conformant application should divide a domain name + * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each, + * and then convert. This function does not offer that level of granularity. The options once + * set will apply to all labels in the domain name + * + * @param src Input UChar array containing IDN in ASCII (ACE encoded) form. + * @param srcLength Number of UChars in src, or -1 if NUL-terminated. + * @param dest Output UChar array containing Unicode equivalent of source IDN. + * @param destCapacity Size of dest. + * @param options A bit set of options: + * + * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points + * and do not use STD3 ASCII rules + * If unassigned code points are found the operation fails with + * U_UNASSIGNED_CODE_POINT_FOUND error code. + * + * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations + * If this option is set, the unassigned code points are in the input + * are treated as normal Unicode code points. + * + * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions + * If this option is set and the input does not satisfy STD3 rules, + * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR + * + * @param parseError Pointer to UParseError struct to receive information on position + * of error if an error is encountered. Can be NULL. + * @param status ICU in/out error code parameter. + * U_INVALID_CHAR_FOUND if src contains + * unmatched single surrogates. + * U_INDEX_OUTOFBOUNDS_ERROR if src contains + * too many code points. + * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough + * @return The length of the result string, if successful - or in case of a buffer overflow, + * in which case it will be greater than destCapacity. + * @deprecated ICU 55 Use UTS #46 instead via uidna_openUTS46() or class IDNA. + */ +U_DEPRECATED int32_t U_EXPORT2 +uidna_IDNToUnicode( const UChar* src, int32_t srcLength, + UChar* dest, int32_t destCapacity, + int32_t options, + UParseError* parseError, + UErrorCode* status); + +/** + * IDNA2003: Compare two IDN strings for equivalence. + * This function splits the domain names into labels and compares them. + * According to IDN RFC, whenever two labels are compared, they are + * considered equal if and only if their ASCII forms (obtained by + * applying toASCII) match using an case-insensitive ASCII comparison. + * Two domain names are considered a match if and only if all labels + * match regardless of whether label separators match. + * + * @param s1 First source string. + * @param length1 Length of first source string, or -1 if NUL-terminated. + * + * @param s2 Second source string. + * @param length2 Length of second source string, or -1 if NUL-terminated. + * @param options A bit set of options: + * + * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points + * and do not use STD3 ASCII rules + * If unassigned code points are found the operation fails with + * U_UNASSIGNED_CODE_POINT_FOUND error code. + * + * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations + * If this option is set, the unassigned code points are in the input + * are treated as normal Unicode code points. + * + * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions + * If this option is set and the input does not satisfy STD3 rules, + * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR + * + * @param status ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return <0 or 0 or >0 as usual for string comparisons + * @deprecated ICU 55 Use UTS #46 instead via uidna_openUTS46() or class IDNA. + */ +U_DEPRECATED int32_t U_EXPORT2 +uidna_compare( const UChar *s1, int32_t length1, + const UChar *s2, int32_t length2, + int32_t options, + UErrorCode* status); + +#endif /* U_HIDE_DEPRECATED_API */ + +#endif /* #if !UCONFIG_NO_IDNA */ + +#endif diff --git a/intl/icu/source/common/unicode/uiter.h b/intl/icu/source/common/unicode/uiter.h new file mode 100644 index 0000000000..be232c774d --- /dev/null +++ b/intl/icu/source/common/unicode/uiter.h @@ -0,0 +1,709 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2002-2011 International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: uiter.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002jan18 +* created by: Markus W. Scherer +*/ + +#ifndef __UITER_H__ +#define __UITER_H__ + +/** + * \file + * \brief C API: Unicode Character Iteration + * + * @see UCharIterator + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + U_NAMESPACE_BEGIN + + class CharacterIterator; + class Replaceable; + + U_NAMESPACE_END +#endif + +U_CDECL_BEGIN + +struct UCharIterator; +typedef struct UCharIterator UCharIterator; /**< C typedef for struct UCharIterator. @stable ICU 2.1 */ + +/** + * Origin constants for UCharIterator.getIndex() and UCharIterator.move(). + * @see UCharIteratorMove + * @see UCharIterator + * @stable ICU 2.1 + */ +typedef enum UCharIteratorOrigin { + UITER_START, UITER_CURRENT, UITER_LIMIT, UITER_ZERO, UITER_LENGTH +} UCharIteratorOrigin; + +/** Constants for UCharIterator. @stable ICU 2.6 */ +enum { + /** + * Constant value that may be returned by UCharIteratorMove + * indicating that the final UTF-16 index is not known, but that the move succeeded. + * This can occur when moving relative to limit or length, or + * when moving relative to the current index after a setState() + * when the current UTF-16 index is not known. + * + * It would be very inefficient to have to count from the beginning of the text + * just to get the current/limit/length index after moving relative to it. + * The actual index can be determined with getIndex(UITER_CURRENT) + * which will count the UChars if necessary. + * + * @stable ICU 2.6 + */ + UITER_UNKNOWN_INDEX=-2 +}; + + +/** + * Constant for UCharIterator getState() indicating an error or + * an unknown state. + * Returned by uiter_getState()/UCharIteratorGetState + * when an error occurs. + * Also, some UCharIterator implementations may not be able to return + * a valid state for each position. This will be clearly documented + * for each such iterator (none of the public ones here). + * + * @stable ICU 2.6 + */ +#define UITER_NO_STATE ((uint32_t)0xffffffff) + +/** + * Function type declaration for UCharIterator.getIndex(). + * + * Gets the current position, or the start or limit of the + * iteration range. + * + * This function may perform slowly for UITER_CURRENT after setState() was called, + * or for UITER_LENGTH, because an iterator implementation may have to count + * UChars if the underlying storage is not UTF-16. + * + * @param iter the UCharIterator structure ("this pointer") + * @param origin get the 0, start, limit, length, or current index + * @return the requested index, or U_SENTINEL in an error condition + * + * @see UCharIteratorOrigin + * @see UCharIterator + * @stable ICU 2.1 + */ +typedef int32_t U_CALLCONV +UCharIteratorGetIndex(UCharIterator *iter, UCharIteratorOrigin origin); + +/** + * Function type declaration for UCharIterator.move(). + * + * Use iter->move(iter, index, UITER_ZERO) like CharacterIterator::setIndex(index). + * + * Moves the current position relative to the start or limit of the + * iteration range, or relative to the current position itself. + * The movement is expressed in numbers of code units forward + * or backward by specifying a positive or negative delta. + * Out of bounds movement will be pinned to the start or limit. + * + * This function may perform slowly for moving relative to UITER_LENGTH + * because an iterator implementation may have to count the rest of the + * UChars if the native storage is not UTF-16. + * + * When moving relative to the limit or length, or + * relative to the current position after setState() was called, + * move() may return UITER_UNKNOWN_INDEX (-2) to avoid an inefficient + * determination of the actual UTF-16 index. + * The actual index can be determined with getIndex(UITER_CURRENT) + * which will count the UChars if necessary. + * See UITER_UNKNOWN_INDEX for details. + * + * @param iter the UCharIterator structure ("this pointer") + * @param delta can be positive, zero, or negative + * @param origin move relative to the 0, start, limit, length, or current index + * @return the new index, or U_SENTINEL on an error condition, + * or UITER_UNKNOWN_INDEX when the index is not known. + * + * @see UCharIteratorOrigin + * @see UCharIterator + * @see UITER_UNKNOWN_INDEX + * @stable ICU 2.1 + */ +typedef int32_t U_CALLCONV +UCharIteratorMove(UCharIterator *iter, int32_t delta, UCharIteratorOrigin origin); + +/** + * Function type declaration for UCharIterator.hasNext(). + * + * Check if current() and next() can still + * return another code unit. + * + * @param iter the UCharIterator structure ("this pointer") + * @return boolean value for whether current() and next() can still return another code unit + * + * @see UCharIterator + * @stable ICU 2.1 + */ +typedef UBool U_CALLCONV +UCharIteratorHasNext(UCharIterator *iter); + +/** + * Function type declaration for UCharIterator.hasPrevious(). + * + * Check if previous() can still return another code unit. + * + * @param iter the UCharIterator structure ("this pointer") + * @return boolean value for whether previous() can still return another code unit + * + * @see UCharIterator + * @stable ICU 2.1 + */ +typedef UBool U_CALLCONV +UCharIteratorHasPrevious(UCharIterator *iter); + +/** + * Function type declaration for UCharIterator.current(). + * + * Return the code unit at the current position, + * or U_SENTINEL if there is none (index is at the limit). + * + * @param iter the UCharIterator structure ("this pointer") + * @return the current code unit + * + * @see UCharIterator + * @stable ICU 2.1 + */ +typedef UChar32 U_CALLCONV +UCharIteratorCurrent(UCharIterator *iter); + +/** + * Function type declaration for UCharIterator.next(). + * + * Return the code unit at the current index and increment + * the index (post-increment, like s[i++]), + * or return U_SENTINEL if there is none (index is at the limit). + * + * @param iter the UCharIterator structure ("this pointer") + * @return the current code unit (and post-increment the current index) + * + * @see UCharIterator + * @stable ICU 2.1 + */ +typedef UChar32 U_CALLCONV +UCharIteratorNext(UCharIterator *iter); + +/** + * Function type declaration for UCharIterator.previous(). + * + * Decrement the index and return the code unit from there + * (pre-decrement, like s[--i]), + * or return U_SENTINEL if there is none (index is at the start). + * + * @param iter the UCharIterator structure ("this pointer") + * @return the previous code unit (after pre-decrementing the current index) + * + * @see UCharIterator + * @stable ICU 2.1 + */ +typedef UChar32 U_CALLCONV +UCharIteratorPrevious(UCharIterator *iter); + +/** + * Function type declaration for UCharIterator.reservedFn(). + * Reserved for future use. + * + * @param iter the UCharIterator structure ("this pointer") + * @param something some integer argument + * @return some integer + * + * @see UCharIterator + * @stable ICU 2.1 + */ +typedef int32_t U_CALLCONV +UCharIteratorReserved(UCharIterator *iter, int32_t something); + +/** + * Function type declaration for UCharIterator.getState(). + * + * Get the "state" of the iterator in the form of a single 32-bit word. + * It is recommended that the state value be calculated to be as small as + * is feasible. For strings with limited lengths, fewer than 32 bits may + * be sufficient. + * + * This is used together with setState()/UCharIteratorSetState + * to save and restore the iterator position more efficiently than with + * getIndex()/move(). + * + * The iterator state is defined as a uint32_t value because it is designed + * for use in ucol_nextSortKeyPart() which provides 32 bits to store the state + * of the character iterator. + * + * With some UCharIterator implementations (e.g., UTF-8), + * getting and setting the UTF-16 index with existing functions + * (getIndex(UITER_CURRENT) followed by move(pos, UITER_ZERO)) is possible but + * relatively slow because the iterator has to "walk" from a known index + * to the requested one. + * This takes more time the farther it needs to go. + * + * An opaque state value allows an iterator implementation to provide + * an internal index (UTF-8: the source byte array index) for + * fast, constant-time restoration. + * + * After calling setState(), a getIndex(UITER_CURRENT) may be slow because + * the UTF-16 index may not be restored as well, but the iterator can deliver + * the correct text contents and move relative to the current position + * without performance degradation. + * + * Some UCharIterator implementations may not be able to return + * a valid state for each position, in which case they return UITER_NO_STATE instead. + * This will be clearly documented for each such iterator (none of the public ones here). + * + * @param iter the UCharIterator structure ("this pointer") + * @return the state word + * + * @see UCharIterator + * @see UCharIteratorSetState + * @see UITER_NO_STATE + * @stable ICU 2.6 + */ +typedef uint32_t U_CALLCONV +UCharIteratorGetState(const UCharIterator *iter); + +/** + * Function type declaration for UCharIterator.setState(). + * + * Restore the "state" of the iterator using a state word from a getState() call. + * The iterator object need not be the same one as for which getState() was called, + * but it must be of the same type (set up using the same uiter_setXYZ function) + * and it must iterate over the same string + * (binary identical regardless of memory address). + * For more about the state word see UCharIteratorGetState. + * + * After calling setState(), a getIndex(UITER_CURRENT) may be slow because + * the UTF-16 index may not be restored as well, but the iterator can deliver + * the correct text contents and move relative to the current position + * without performance degradation. + * + * @param iter the UCharIterator structure ("this pointer") + * @param state the state word from a getState() call + * on a same-type, same-string iterator + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * + * @see UCharIterator + * @see UCharIteratorGetState + * @stable ICU 2.6 + */ +typedef void U_CALLCONV +UCharIteratorSetState(UCharIterator *iter, uint32_t state, UErrorCode *pErrorCode); + + +/** + * C API for code unit iteration. + * This can be used as a C wrapper around + * CharacterIterator, Replaceable, or implemented using simple strings, etc. + * + * There are two roles for using UCharIterator: + * + * A "provider" sets the necessary function pointers and controls the "protected" + * fields of the UCharIterator structure. A "provider" passes a UCharIterator + * into C APIs that need a UCharIterator as an abstract, flexible string interface. + * + * Implementations of such C APIs are "callers" of UCharIterator functions; + * they only use the "public" function pointers and never access the "protected" + * fields directly. + * + * The current() and next() functions only check the current index against the + * limit, and previous() only checks the current index against the start, + * to see if the iterator already reached the end of the iteration range. + * + * The assumption - in all iterators - is that the index is moved via the API, + * which means it won't go out of bounds, or the index is modified by + * user code that knows enough about the iterator implementation to set valid + * index values. + * + * UCharIterator functions return code unit values 0..0xffff, + * or U_SENTINEL if the iteration bounds are reached. + * + * @stable ICU 2.1 + */ +struct UCharIterator { + /** + * (protected) Pointer to string or wrapped object or similar. + * Not used by caller. + * @stable ICU 2.1 + */ + const void *context; + + /** + * (protected) Length of string or similar. + * Not used by caller. + * @stable ICU 2.1 + */ + int32_t length; + + /** + * (protected) Start index or similar. + * Not used by caller. + * @stable ICU 2.1 + */ + int32_t start; + + /** + * (protected) Current index or similar. + * Not used by caller. + * @stable ICU 2.1 + */ + int32_t index; + + /** + * (protected) Limit index or similar. + * Not used by caller. + * @stable ICU 2.1 + */ + int32_t limit; + + /** + * (protected) Used by UTF-8 iterators and possibly others. + * @stable ICU 2.1 + */ + int32_t reservedField; + + /** + * (public) Returns the current position or the + * start or limit index of the iteration range. + * + * @see UCharIteratorGetIndex + * @stable ICU 2.1 + */ + UCharIteratorGetIndex *getIndex; + + /** + * (public) Moves the current position relative to the start or limit of the + * iteration range, or relative to the current position itself. + * The movement is expressed in numbers of code units forward + * or backward by specifying a positive or negative delta. + * + * @see UCharIteratorMove + * @stable ICU 2.1 + */ + UCharIteratorMove *move; + + /** + * (public) Check if current() and next() can still + * return another code unit. + * + * @see UCharIteratorHasNext + * @stable ICU 2.1 + */ + UCharIteratorHasNext *hasNext; + + /** + * (public) Check if previous() can still return another code unit. + * + * @see UCharIteratorHasPrevious + * @stable ICU 2.1 + */ + UCharIteratorHasPrevious *hasPrevious; + + /** + * (public) Return the code unit at the current position, + * or U_SENTINEL if there is none (index is at the limit). + * + * @see UCharIteratorCurrent + * @stable ICU 2.1 + */ + UCharIteratorCurrent *current; + + /** + * (public) Return the code unit at the current index and increment + * the index (post-increment, like s[i++]), + * or return U_SENTINEL if there is none (index is at the limit). + * + * @see UCharIteratorNext + * @stable ICU 2.1 + */ + UCharIteratorNext *next; + + /** + * (public) Decrement the index and return the code unit from there + * (pre-decrement, like s[--i]), + * or return U_SENTINEL if there is none (index is at the start). + * + * @see UCharIteratorPrevious + * @stable ICU 2.1 + */ + UCharIteratorPrevious *previous; + + /** + * (public) Reserved for future use. Currently NULL. + * + * @see UCharIteratorReserved + * @stable ICU 2.1 + */ + UCharIteratorReserved *reservedFn; + + /** + * (public) Return the state of the iterator, to be restored later with setState(). + * This function pointer is NULL if the iterator does not implement it. + * + * @see UCharIteratorGet + * @stable ICU 2.6 + */ + UCharIteratorGetState *getState; + + /** + * (public) Restore the iterator state from the state word from a call + * to getState(). + * This function pointer is NULL if the iterator does not implement it. + * + * @see UCharIteratorSet + * @stable ICU 2.6 + */ + UCharIteratorSetState *setState; +}; + +/** + * Helper function for UCharIterator to get the code point + * at the current index. + * + * Return the code point that includes the code unit at the current position, + * or U_SENTINEL if there is none (index is at the limit). + * If the current code unit is a lead or trail surrogate, + * then the following or preceding surrogate is used to form + * the code point value. + * + * @param iter the UCharIterator structure ("this pointer") + * @return the current code point + * + * @see UCharIterator + * @see U16_GET + * @see UnicodeString::char32At() + * @stable ICU 2.1 + */ +U_CAPI UChar32 U_EXPORT2 +uiter_current32(UCharIterator *iter); + +/** + * Helper function for UCharIterator to get the next code point. + * + * Return the code point at the current index and increment + * the index (post-increment, like s[i++]), + * or return U_SENTINEL if there is none (index is at the limit). + * + * @param iter the UCharIterator structure ("this pointer") + * @return the current code point (and post-increment the current index) + * + * @see UCharIterator + * @see U16_NEXT + * @stable ICU 2.1 + */ +U_CAPI UChar32 U_EXPORT2 +uiter_next32(UCharIterator *iter); + +/** + * Helper function for UCharIterator to get the previous code point. + * + * Decrement the index and return the code point from there + * (pre-decrement, like s[--i]), + * or return U_SENTINEL if there is none (index is at the start). + * + * @param iter the UCharIterator structure ("this pointer") + * @return the previous code point (after pre-decrementing the current index) + * + * @see UCharIterator + * @see U16_PREV + * @stable ICU 2.1 + */ +U_CAPI UChar32 U_EXPORT2 +uiter_previous32(UCharIterator *iter); + +/** + * Get the "state" of the iterator in the form of a single 32-bit word. + * This is a convenience function that calls iter->getState(iter) + * if iter->getState is not NULL; + * if it is NULL or any other error occurs, then UITER_NO_STATE is returned. + * + * Some UCharIterator implementations may not be able to return + * a valid state for each position, in which case they return UITER_NO_STATE instead. + * This will be clearly documented for each such iterator (none of the public ones here). + * + * @param iter the UCharIterator structure ("this pointer") + * @return the state word + * + * @see UCharIterator + * @see UCharIteratorGetState + * @see UITER_NO_STATE + * @stable ICU 2.6 + */ +U_CAPI uint32_t U_EXPORT2 +uiter_getState(const UCharIterator *iter); + +/** + * Restore the "state" of the iterator using a state word from a getState() call. + * This is a convenience function that calls iter->setState(iter, state, pErrorCode) + * if iter->setState is not NULL; if it is NULL, then U_UNSUPPORTED_ERROR is set. + * + * @param iter the UCharIterator structure ("this pointer") + * @param state the state word from a getState() call + * on a same-type, same-string iterator + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * + * @see UCharIterator + * @see UCharIteratorSetState + * @stable ICU 2.6 + */ +U_CAPI void U_EXPORT2 +uiter_setState(UCharIterator *iter, uint32_t state, UErrorCode *pErrorCode); + +/** + * Set up a UCharIterator to iterate over a string. + * + * Sets the UCharIterator function pointers for iteration over the string s + * with iteration boundaries start=index=0 and length=limit=string length. + * The "provider" may set the start, index, and limit values at any time + * within the range 0..length. + * The length field will be ignored. + * + * The string pointer s is set into UCharIterator.context without copying + * or reallocating the string contents. + * + * getState() simply returns the current index. + * move() will always return the final index. + * + * @param iter UCharIterator structure to be set for iteration + * @param s String to iterate over + * @param length Length of s, or -1 if NUL-terminated + * + * @see UCharIterator + * @stable ICU 2.1 + */ +U_CAPI void U_EXPORT2 +uiter_setString(UCharIterator *iter, const UChar *s, int32_t length); + +/** + * Set up a UCharIterator to iterate over a UTF-16BE string + * (byte vector with a big-endian pair of bytes per UChar). + * + * Everything works just like with a normal UChar iterator (uiter_setString), + * except that UChars are assembled from byte pairs, + * and that the length argument here indicates an even number of bytes. + * + * getState() simply returns the current index. + * move() will always return the final index. + * + * @param iter UCharIterator structure to be set for iteration + * @param s UTF-16BE string to iterate over + * @param length Length of s as an even number of bytes, or -1 if NUL-terminated + * (NUL means pair of 0 bytes at even index from s) + * + * @see UCharIterator + * @see uiter_setString + * @stable ICU 2.6 + */ +U_CAPI void U_EXPORT2 +uiter_setUTF16BE(UCharIterator *iter, const char *s, int32_t length); + +/** + * Set up a UCharIterator to iterate over a UTF-8 string. + * + * Sets the UCharIterator function pointers for iteration over the UTF-8 string s + * with UTF-8 iteration boundaries 0 and length. + * The implementation counts the UTF-16 index on the fly and + * lazily evaluates the UTF-16 length of the text. + * + * The start field is used as the UTF-8 offset, the limit field as the UTF-8 length. + * When the reservedField is not 0, then it contains a supplementary code point + * and the UTF-16 index is between the two corresponding surrogates. + * At that point, the UTF-8 index is behind that code point. + * + * The UTF-8 string pointer s is set into UCharIterator.context without copying + * or reallocating the string contents. + * + * getState() returns a state value consisting of + * - the current UTF-8 source byte index (bits 31..1) + * - a flag (bit 0) that indicates whether the UChar position is in the middle + * of a surrogate pair + * (from a 4-byte UTF-8 sequence for the corresponding supplementary code point) + * + * getState() cannot also encode the UTF-16 index in the state value. + * move(relative to limit or length), or + * move(relative to current) after setState(), may return UITER_UNKNOWN_INDEX. + * + * @param iter UCharIterator structure to be set for iteration + * @param s UTF-8 string to iterate over + * @param length Length of s in bytes, or -1 if NUL-terminated + * + * @see UCharIterator + * @stable ICU 2.6 + */ +U_CAPI void U_EXPORT2 +uiter_setUTF8(UCharIterator *iter, const char *s, int32_t length); + +#if U_SHOW_CPLUSPLUS_API + +/** + * Set up a UCharIterator to wrap around a C++ CharacterIterator. + * + * Sets the UCharIterator function pointers for iteration using the + * CharacterIterator charIter. + * + * The CharacterIterator pointer charIter is set into UCharIterator.context + * without copying or cloning the CharacterIterator object. + * The other "protected" UCharIterator fields are set to 0 and will be ignored. + * The iteration index and boundaries are controlled by the CharacterIterator. + * + * getState() simply returns the current index. + * move() will always return the final index. + * + * @param iter UCharIterator structure to be set for iteration + * @param charIter CharacterIterator to wrap + * + * @see UCharIterator + * @stable ICU 2.1 + */ +U_CAPI void U_EXPORT2 +uiter_setCharacterIterator(UCharIterator *iter, icu::CharacterIterator *charIter); + +/** + * Set up a UCharIterator to iterate over a C++ Replaceable. + * + * Sets the UCharIterator function pointers for iteration over the + * Replaceable rep with iteration boundaries start=index=0 and + * length=limit=rep->length(). + * The "provider" may set the start, index, and limit values at any time + * within the range 0..length=rep->length(). + * The length field will be ignored. + * + * The Replaceable pointer rep is set into UCharIterator.context without copying + * or cloning/reallocating the Replaceable object. + * + * getState() simply returns the current index. + * move() will always return the final index. + * + * @param iter UCharIterator structure to be set for iteration + * @param rep Replaceable to iterate over + * + * @see UCharIterator + * @stable ICU 2.1 + */ +U_CAPI void U_EXPORT2 +uiter_setReplaceable(UCharIterator *iter, const icu::Replaceable *rep); + +#endif + +U_CDECL_END + +#endif diff --git a/intl/icu/source/common/unicode/uldnames.h b/intl/icu/source/common/unicode/uldnames.h new file mode 100644 index 0000000000..47b047ece9 --- /dev/null +++ b/intl/icu/source/common/unicode/uldnames.h @@ -0,0 +1,307 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2016, International Business Machines Corporation and +* others. All Rights Reserved. +******************************************************************************* +*/ + +#ifndef __ULDNAMES_H__ +#define __ULDNAMES_H__ + +/** + * \file + * \brief C API: Provides display names of Locale ids and their components. + */ + +#include "unicode/utypes.h" +#include "unicode/uscript.h" +#include "unicode/udisplaycontext.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +/** + * Enum used in LocaleDisplayNames::createInstance. + * @stable ICU 4.4 + */ +typedef enum { + /** + * Use standard names when generating a locale name, + * e.g. en_GB displays as 'English (United Kingdom)'. + * @stable ICU 4.4 + */ + ULDN_STANDARD_NAMES = 0, + /** + * Use dialect names, when generating a locale name, + * e.g. en_GB displays as 'British English'. + * @stable ICU 4.4 + */ + ULDN_DIALECT_NAMES +} UDialectHandling; + +/** + * Opaque C service object type for the locale display names API + * @stable ICU 4.4 + */ +struct ULocaleDisplayNames; + +/** + * C typedef for struct ULocaleDisplayNames. + * @stable ICU 4.4 + */ +typedef struct ULocaleDisplayNames ULocaleDisplayNames; + +#if !UCONFIG_NO_FORMATTING + +/** + * Returns an instance of LocaleDisplayNames that returns names + * formatted for the provided locale, using the provided + * dialectHandling. The usual value for dialectHandling is + * ULOC_STANDARD_NAMES. + * + * @param locale the display locale + * @param dialectHandling how to select names for locales + * @return a ULocaleDisplayNames instance + * @param pErrorCode the status code + * @stable ICU 4.4 + */ +U_CAPI ULocaleDisplayNames * U_EXPORT2 +uldn_open(const char * locale, + UDialectHandling dialectHandling, + UErrorCode *pErrorCode); + +/** + * Closes a ULocaleDisplayNames instance obtained from uldn_open(). + * @param ldn the ULocaleDisplayNames instance to be closed + * @stable ICU 4.4 + */ +U_CAPI void U_EXPORT2 +uldn_close(ULocaleDisplayNames *ldn); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalULocaleDisplayNamesPointer + * "Smart pointer" class, closes a ULocaleDisplayNames via uldn_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalULocaleDisplayNamesPointer, ULocaleDisplayNames, uldn_close); + +U_NAMESPACE_END + +#endif + +/* getters for state */ + +/** + * Returns the locale used to determine the display names. This is + * not necessarily the same locale passed to {@link #uldn_open}. + * @param ldn the LocaleDisplayNames instance + * @return the display locale + * @stable ICU 4.4 + */ +U_CAPI const char * U_EXPORT2 +uldn_getLocale(const ULocaleDisplayNames *ldn); + +/** + * Returns the dialect handling used in the display names. + * @param ldn the LocaleDisplayNames instance + * @return the dialect handling enum + * @stable ICU 4.4 + */ +U_CAPI UDialectHandling U_EXPORT2 +uldn_getDialectHandling(const ULocaleDisplayNames *ldn); + +/* names for entire locales */ + +/** + * Returns the display name of the provided locale. + * @param ldn the LocaleDisplayNames instance + * @param locale the locale whose display name to return + * @param result receives the display name + * @param maxResultSize the size of the result buffer + * @param pErrorCode the status code + * @return the actual buffer size needed for the display name. If it's + * greater than maxResultSize, the returned name will be truncated. + * @stable ICU 4.4 + */ +U_CAPI int32_t U_EXPORT2 +uldn_localeDisplayName(const ULocaleDisplayNames *ldn, + const char *locale, + UChar *result, + int32_t maxResultSize, + UErrorCode *pErrorCode); + +/* names for components of a locale */ + +/** + * Returns the display name of the provided language code. + * @param ldn the LocaleDisplayNames instance + * @param lang the language code whose display name to return + * @param result receives the display name + * @param maxResultSize the size of the result buffer + * @param pErrorCode the status code + * @return the actual buffer size needed for the display name. If it's + * greater than maxResultSize, the returned name will be truncated. + * @stable ICU 4.4 + */ +U_CAPI int32_t U_EXPORT2 +uldn_languageDisplayName(const ULocaleDisplayNames *ldn, + const char *lang, + UChar *result, + int32_t maxResultSize, + UErrorCode *pErrorCode); + +/** + * Returns the display name of the provided script. + * @param ldn the LocaleDisplayNames instance + * @param script the script whose display name to return + * @param result receives the display name + * @param maxResultSize the size of the result buffer + * @param pErrorCode the status code + * @return the actual buffer size needed for the display name. If it's + * greater than maxResultSize, the returned name will be truncated. + * @stable ICU 4.4 + */ +U_CAPI int32_t U_EXPORT2 +uldn_scriptDisplayName(const ULocaleDisplayNames *ldn, + const char *script, + UChar *result, + int32_t maxResultSize, + UErrorCode *pErrorCode); + +/** + * Returns the display name of the provided script code. + * @param ldn the LocaleDisplayNames instance + * @param scriptCode the script code whose display name to return + * @param result receives the display name + * @param maxResultSize the size of the result buffer + * @param pErrorCode the status code + * @return the actual buffer size needed for the display name. If it's + * greater than maxResultSize, the returned name will be truncated. + * @stable ICU 4.4 + */ +U_CAPI int32_t U_EXPORT2 +uldn_scriptCodeDisplayName(const ULocaleDisplayNames *ldn, + UScriptCode scriptCode, + UChar *result, + int32_t maxResultSize, + UErrorCode *pErrorCode); + +/** + * Returns the display name of the provided region code. + * @param ldn the LocaleDisplayNames instance + * @param region the region code whose display name to return + * @param result receives the display name + * @param maxResultSize the size of the result buffer + * @param pErrorCode the status code + * @return the actual buffer size needed for the display name. If it's + * greater than maxResultSize, the returned name will be truncated. + * @stable ICU 4.4 + */ +U_CAPI int32_t U_EXPORT2 +uldn_regionDisplayName(const ULocaleDisplayNames *ldn, + const char *region, + UChar *result, + int32_t maxResultSize, + UErrorCode *pErrorCode); + +/** + * Returns the display name of the provided variant + * @param ldn the LocaleDisplayNames instance + * @param variant the variant whose display name to return + * @param result receives the display name + * @param maxResultSize the size of the result buffer + * @param pErrorCode the status code + * @return the actual buffer size needed for the display name. If it's + * greater than maxResultSize, the returned name will be truncated. + * @stable ICU 4.4 + */ +U_CAPI int32_t U_EXPORT2 +uldn_variantDisplayName(const ULocaleDisplayNames *ldn, + const char *variant, + UChar *result, + int32_t maxResultSize, + UErrorCode *pErrorCode); + +/** + * Returns the display name of the provided locale key + * @param ldn the LocaleDisplayNames instance + * @param key the locale key whose display name to return + * @param result receives the display name + * @param maxResultSize the size of the result buffer + * @param pErrorCode the status code + * @return the actual buffer size needed for the display name. If it's + * greater than maxResultSize, the returned name will be truncated. + * @stable ICU 4.4 + */ +U_CAPI int32_t U_EXPORT2 +uldn_keyDisplayName(const ULocaleDisplayNames *ldn, + const char *key, + UChar *result, + int32_t maxResultSize, + UErrorCode *pErrorCode); + +/** + * Returns the display name of the provided value (used with the provided key). + * @param ldn the LocaleDisplayNames instance + * @param key the locale key + * @param value the locale key's value + * @param result receives the display name + * @param maxResultSize the size of the result buffer + * @param pErrorCode the status code + * @return the actual buffer size needed for the display name. If it's + * greater than maxResultSize, the returned name will be truncated. + * @stable ICU 4.4 + */ +U_CAPI int32_t U_EXPORT2 +uldn_keyValueDisplayName(const ULocaleDisplayNames *ldn, + const char *key, + const char *value, + UChar *result, + int32_t maxResultSize, + UErrorCode *pErrorCode); + +/** +* Returns an instance of LocaleDisplayNames that returns names formatted +* for the provided locale, using the provided UDisplayContext settings. +* +* @param locale The display locale +* @param contexts List of one or more context settings (e.g. for dialect +* handling, capitalization, etc. +* @param length Number of items in the contexts list +* @param pErrorCode Pointer to UErrorCode input/output status. If at entry this indicates +* a failure status, the function will do nothing; otherwise this will be +* updated with any new status from the function. +* @return a ULocaleDisplayNames instance +* @stable ICU 51 +*/ +U_CAPI ULocaleDisplayNames * U_EXPORT2 +uldn_openForContext(const char * locale, UDisplayContext *contexts, + int32_t length, UErrorCode *pErrorCode); + +/** +* Returns the UDisplayContext value for the specified UDisplayContextType. +* @param ldn the ULocaleDisplayNames instance +* @param type the UDisplayContextType whose value to return +* @param pErrorCode Pointer to UErrorCode input/output status. If at entry this indicates +* a failure status, the function will do nothing; otherwise this will be +* updated with any new status from the function. +* @return the UDisplayContextValue for the specified type. +* @stable ICU 51 +*/ +U_CAPI UDisplayContext U_EXPORT2 +uldn_getContext(const ULocaleDisplayNames *ldn, UDisplayContextType type, + UErrorCode *pErrorCode); + +#endif /* !UCONFIG_NO_FORMATTING */ +#endif /* __ULDNAMES_H__ */ diff --git a/intl/icu/source/common/unicode/uloc.h b/intl/icu/source/common/unicode/uloc.h new file mode 100644 index 0000000000..21179c1b62 --- /dev/null +++ b/intl/icu/source/common/unicode/uloc.h @@ -0,0 +1,1393 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1997-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* +* File ULOC.H +* +* Modification History: +* +* Date Name Description +* 04/01/97 aliu Creation. +* 08/22/98 stephen JDK 1.2 sync. +* 12/08/98 rtg New C API for Locale +* 03/30/99 damiba overhaul +* 03/31/99 helena Javadoc for uloc functions. +* 04/15/99 Madhu Updated Javadoc +******************************************************************************** +*/ + +#ifndef ULOC_H +#define ULOC_H + +#include "unicode/utypes.h" +#include "unicode/uenum.h" + +/** + * \file + * \brief C API: Locale ID functionality similar to C++ class Locale + * + *

ULoc C API for Locale

+ * A Locale represents a specific geographical, political, + * or cultural region. An operation that requires a Locale to perform + * its task is called locale-sensitive and uses the Locale + * to tailor information for the user. For example, displaying a number + * is a locale-sensitive operation--the number should be formatted + * according to the customs/conventions of the user's native country, + * region, or culture. In the C APIs, a locales is simply a const char string. + * + *

+ * You create a Locale with one of the three options listed below. + * Each of the component is separated by '_' in the locale string. + * \htmlonly

\endhtmlonly + * 
+ * \code
+ *       newLanguage
+ * 
+ *       newLanguage + newCountry
+ * 
+ *       newLanguage + newCountry + newVariant
+ * \endcode
+ *
+ * \htmlonly
\endhtmlonly + * The first option is a valid ISO + * Language Code. These codes are the lower-case two-letter + * codes as defined by ISO-639. + * You can find a full list of these codes at a number of sites, such as: + *
+ * http://www.ics.uci.edu/pub/ietf/http/related/iso639.txt + * + *

+ * The second option includes an additional ISO Country + * Code. These codes are the upper-case two-letter codes + * as defined by ISO-3166. + * You can find a full list of these codes at a number of sites, such as: + *
+ * http://www.chemie.fu-berlin.de/diverse/doc/ISO_3166.html + * + *

+ * The third option requires another additional information--the + * Variant. + * The Variant codes are vendor and browser-specific. + * For example, use WIN for Windows, MAC for Macintosh, and POSIX for POSIX. + * Where there are two variants, separate them with an underscore, and + * put the most important one first. For + * example, a Traditional Spanish collation might be referenced, with + * "ES", "ES", "Traditional_WIN". + * + *

+ * Because a Locale is just an identifier for a region, + * no validity check is performed when you specify a Locale. + * If you want to see whether particular resources are available for the + * Locale you asked for, you must query those resources. For + * example, ask the UNumberFormat for the locales it supports + * using its getAvailable method. + *
Note: When you ask for a resource for a particular + * locale, you get back the best available match, not necessarily + * precisely what you asked for. For more information, look at + * UResourceBundle. + * + *

+ * The Locale provides a number of convenient constants + * that you can use to specify the commonly used + * locales. For example, the following refers to a locale + * for the United States: + * \htmlonly

\endhtmlonly + * 
+ * \code
+ *       ULOC_US
+ * \endcode
+ *
+ * \htmlonly
\endhtmlonly + * + *

+ * Once you've specified a locale you can query it for information about + * itself. Use uloc_getCountry to get the ISO Country Code and + * uloc_getLanguage to get the ISO Language Code. You can + * use uloc_getDisplayCountry to get the + * name of the country suitable for displaying to the user. Similarly, + * you can use uloc_getDisplayLanguage to get the name of + * the language suitable for displaying to the user. Interestingly, + * the uloc_getDisplayXXX methods are themselves locale-sensitive + * and have two versions: one that uses the default locale and one + * that takes a locale as an argument and displays the name or country in + * a language appropriate to that locale. + * + *

+ * The ICU provides a number of services that perform locale-sensitive + * operations. For example, the unum_xxx functions format + * numbers, currency, or percentages in a locale-sensitive manner. + *

+ * \htmlonly
\endhtmlonly + * 
+ * \code
+ *     UErrorCode success = U_ZERO_ERROR;
+ *     UNumberFormat *nf;
+ *     const char* myLocale = "fr_FR";
+ * 
+ *     nf = unum_open( UNUM_DEFAULT, NULL, success );          
+ *     unum_close(nf);
+ *     nf = unum_open( UNUM_CURRENCY, NULL, success );
+ *     unum_close(nf);
+ *     nf = unum_open( UNUM_PERCENT, NULL, success );   
+ *     unum_close(nf);
+ * \endcode
+ *
+ * \htmlonly
\endhtmlonly + * Each of these methods has two variants; one with an explicit locale + * and one without; the latter using the default locale. + * \htmlonly
\endhtmlonly + * 
+ * \code 
+ * 
+ *     nf = unum_open( UNUM_DEFAULT, myLocale, success );          
+ *     unum_close(nf);
+ *     nf = unum_open( UNUM_CURRENCY, myLocale, success );
+ *     unum_close(nf);
+ *     nf = unum_open( UNUM_PERCENT, myLocale, success );   
+ *     unum_close(nf);
+ * \endcode
+ *
+ * \htmlonly
\endhtmlonly + * A Locale is the mechanism for identifying the kind of services + * (UNumberFormat) that you would like to get. The locale is + * just a mechanism for identifying these services. + * + *

+ * Each international service that performs locale-sensitive operations + * allows you + * to get all the available objects of that type. You can sift + * through these objects by language, country, or variant, + * and use the display names to present a menu to the user. + * For example, you can create a menu of all the collation objects + * suitable for a given language. Such classes implement these + * three class methods: + * \htmlonly

\endhtmlonly + * 
+ * \code
+ *       const char* uloc_getAvailable(int32_t index);
+ *       int32_t uloc_countAvailable();
+ *       int32_t
+ *       uloc_getDisplayName(const char* localeID,
+ *                 const char* inLocaleID, 
+ *                 UChar* result,
+ *                 int32_t maxResultSize,
+ *                  UErrorCode* err);
+ * 
+ * \endcode
+ *
+ * \htmlonly
\endhtmlonly + *

+ * Concerning POSIX/RFC1766 Locale IDs, + * the getLanguage/getCountry/getVariant/getName functions do understand + * the POSIX type form of language_COUNTRY.ENCODING\@VARIANT + * and if there is not an ICU-stype variant, uloc_getVariant() for example + * will return the one listed after the \@at sign. As well, the hyphen + * "-" is recognized as a country/variant separator similarly to RFC1766. + * So for example, "en-us" will be interpreted as en_US. + * As a result, uloc_getName() is far from a no-op, and will have the + * effect of converting POSIX/RFC1766 IDs into ICU form, although it does + * NOT map any of the actual codes (i.e. russian->ru) in any way. + * Applications should call uloc_getName() at the point where a locale ID + * is coming from an external source (user entry, OS, web browser) + * and pass the resulting string to other ICU functions. For example, + * don't use de-de\@EURO as an argument to resourcebundle. + * + * @see UResourceBundle + */ + +/** Useful constant for this language. @stable ICU 2.0 */ +#define ULOC_CHINESE "zh" +/** Useful constant for this language. @stable ICU 2.0 */ +#define ULOC_ENGLISH "en" +/** Useful constant for this language. @stable ICU 2.0 */ +#define ULOC_FRENCH "fr" +/** Useful constant for this language. @stable ICU 2.0 */ +#define ULOC_GERMAN "de" +/** Useful constant for this language. @stable ICU 2.0 */ +#define ULOC_ITALIAN "it" +/** Useful constant for this language. @stable ICU 2.0 */ +#define ULOC_JAPANESE "ja" +/** Useful constant for this language. @stable ICU 2.0 */ +#define ULOC_KOREAN "ko" +/** Useful constant for this language. @stable ICU 2.0 */ +#define ULOC_SIMPLIFIED_CHINESE "zh_CN" +/** Useful constant for this language. @stable ICU 2.0 */ +#define ULOC_TRADITIONAL_CHINESE "zh_TW" + +/** Useful constant for this country/region. @stable ICU 2.0 */ +#define ULOC_CANADA "en_CA" +/** Useful constant for this country/region. @stable ICU 2.0 */ +#define ULOC_CANADA_FRENCH "fr_CA" +/** Useful constant for this country/region. @stable ICU 2.0 */ +#define ULOC_CHINA "zh_CN" +/** Useful constant for this country/region. @stable ICU 2.0 */ +#define ULOC_PRC "zh_CN" +/** Useful constant for this country/region. @stable ICU 2.0 */ +#define ULOC_FRANCE "fr_FR" +/** Useful constant for this country/region. @stable ICU 2.0 */ +#define ULOC_GERMANY "de_DE" +/** Useful constant for this country/region. @stable ICU 2.0 */ +#define ULOC_ITALY "it_IT" +/** Useful constant for this country/region. @stable ICU 2.0 */ +#define ULOC_JAPAN "ja_JP" +/** Useful constant for this country/region. @stable ICU 2.0 */ +#define ULOC_KOREA "ko_KR" +/** Useful constant for this country/region. @stable ICU 2.0 */ +#define ULOC_TAIWAN "zh_TW" +/** Useful constant for this country/region. @stable ICU 2.0 */ +#define ULOC_UK "en_GB" +/** Useful constant for this country/region. @stable ICU 2.0 */ +#define ULOC_US "en_US" + +/** + * Useful constant for the maximum size of the language part of a locale ID. + * (including the terminating NULL). + * @stable ICU 2.0 + */ +#define ULOC_LANG_CAPACITY 12 + +/** + * Useful constant for the maximum size of the country part of a locale ID + * (including the terminating NULL). + * @stable ICU 2.0 + */ +#define ULOC_COUNTRY_CAPACITY 4 +/** + * Useful constant for the maximum size of the whole locale ID + * (including the terminating NULL and all keywords). + * @stable ICU 2.0 + */ +#define ULOC_FULLNAME_CAPACITY 157 + +/** + * Useful constant for the maximum size of the script part of a locale ID + * (including the terminating NULL). + * @stable ICU 2.8 + */ +#define ULOC_SCRIPT_CAPACITY 6 + +/** + * Useful constant for the maximum size of keywords in a locale + * @stable ICU 2.8 + */ +#define ULOC_KEYWORDS_CAPACITY 96 + +/** + * Useful constant for the maximum total size of keywords and their values in a locale + * @stable ICU 2.8 + */ +#define ULOC_KEYWORD_AND_VALUES_CAPACITY 100 + +/** + * Invariant character separating keywords from the locale string + * @stable ICU 2.8 + */ +#define ULOC_KEYWORD_SEPARATOR '@' + +/** + * Unicode code point for '@' separating keywords from the locale string. + * @see ULOC_KEYWORD_SEPARATOR + * @stable ICU 4.6 + */ +#define ULOC_KEYWORD_SEPARATOR_UNICODE 0x40 + +/** + * Invariant character for assigning value to a keyword + * @stable ICU 2.8 + */ +#define ULOC_KEYWORD_ASSIGN '=' + +/** + * Unicode code point for '=' for assigning value to a keyword. + * @see ULOC_KEYWORD_ASSIGN + * @stable ICU 4.6 + */ +#define ULOC_KEYWORD_ASSIGN_UNICODE 0x3D + +/** + * Invariant character separating keywords + * @stable ICU 2.8 + */ +#define ULOC_KEYWORD_ITEM_SEPARATOR ';' + +/** + * Unicode code point for ';' separating keywords + * @see ULOC_KEYWORD_ITEM_SEPARATOR + * @stable ICU 4.6 + */ +#define ULOC_KEYWORD_ITEM_SEPARATOR_UNICODE 0x3B + +/** + * Constants for *_getLocale() + * Allow user to select whether she wants information on + * requested, valid or actual locale. + * For example, a collator for "en_US_CALIFORNIA" was + * requested. In the current state of ICU (2.0), + * the requested locale is "en_US_CALIFORNIA", + * the valid locale is "en_US" (most specific locale supported by ICU) + * and the actual locale is "root" (the collation data comes unmodified + * from the UCA) + * The locale is considered supported by ICU if there is a core ICU bundle + * for that locale (although it may be empty). + * @stable ICU 2.1 + */ +typedef enum { + /** This is locale the data actually comes from + * @stable ICU 2.1 + */ + ULOC_ACTUAL_LOCALE = 0, + /** This is the most specific locale supported by ICU + * @stable ICU 2.1 + */ + ULOC_VALID_LOCALE = 1, + +#ifndef U_HIDE_DEPRECATED_API + /** This is the requested locale + * @deprecated ICU 2.8 + */ + ULOC_REQUESTED_LOCALE = 2, + + /** + * One more than the highest normal ULocDataLocaleType value. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + ULOC_DATA_LOCALE_TYPE_LIMIT = 3 +#endif // U_HIDE_DEPRECATED_API +} ULocDataLocaleType; + +#ifndef U_HIDE_SYSTEM_API +/** + * Gets ICU's default locale. + * The returned string is a snapshot in time, and will remain valid + * and unchanged even when uloc_setDefault() is called. + * The returned storage is owned by ICU, and must not be altered or deleted + * by the caller. + * + * @return the ICU default locale + * @system + * @stable ICU 2.0 + */ +U_CAPI const char* U_EXPORT2 +uloc_getDefault(void); + +/** + * Sets ICU's default locale. + * By default (without calling this function), ICU's default locale will be based + * on information obtained from the underlying system environment. + *

+ * Changes to ICU's default locale do not propagate back to the + * system environment. + *

+ * Changes to ICU's default locale to not affect any ICU services that + * may already be open based on the previous default locale value. + * + * @param localeID the new ICU default locale. A value of NULL will try to get + * the system's default locale. + * @param status the error information if the setting of default locale fails + * @system + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +uloc_setDefault(const char* localeID, + UErrorCode* status); +#endif /* U_HIDE_SYSTEM_API */ + +/** + * Gets the language code for the specified locale. + * + * @param localeID the locale to get the ISO language code with + * @param language the language code for localeID + * @param languageCapacity the size of the language buffer to store the + * language code with + * @param err error information if retrieving the language code failed + * @return the actual buffer size needed for the language code. If it's greater + * than languageCapacity, the returned language code will be truncated. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getLanguage(const char* localeID, + char* language, + int32_t languageCapacity, + UErrorCode* err); + +/** + * Gets the script code for the specified locale. + * + * @param localeID the locale to get the ISO language code with + * @param script the language code for localeID + * @param scriptCapacity the size of the language buffer to store the + * language code with + * @param err error information if retrieving the language code failed + * @return the actual buffer size needed for the language code. If it's greater + * than scriptCapacity, the returned language code will be truncated. + * @stable ICU 2.8 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getScript(const char* localeID, + char* script, + int32_t scriptCapacity, + UErrorCode* err); + +/** + * Gets the country code for the specified locale. + * + * @param localeID the locale to get the country code with + * @param country the country code for localeID + * @param countryCapacity the size of the country buffer to store the + * country code with + * @param err error information if retrieving the country code failed + * @return the actual buffer size needed for the country code. If it's greater + * than countryCapacity, the returned country code will be truncated. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getCountry(const char* localeID, + char* country, + int32_t countryCapacity, + UErrorCode* err); + +/** + * Gets the variant code for the specified locale. + * + * @param localeID the locale to get the variant code with + * @param variant the variant code for localeID + * @param variantCapacity the size of the variant buffer to store the + * variant code with + * @param err error information if retrieving the variant code failed + * @return the actual buffer size needed for the variant code. If it's greater + * than variantCapacity, the returned variant code will be truncated. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getVariant(const char* localeID, + char* variant, + int32_t variantCapacity, + UErrorCode* err); + + +/** + * Gets the full name for the specified locale. + * Note: This has the effect of 'canonicalizing' the ICU locale ID to + * a certain extent. Upper and lower case are set as needed. + * It does NOT map aliased names in any way. + * See the top of this header file. + * This API supports preflighting. + * + * @param localeID the locale to get the full name with + * @param name fill in buffer for the name without keywords. + * @param nameCapacity capacity of the fill in buffer. + * @param err error information if retrieving the full name failed + * @return the actual buffer size needed for the full name. If it's greater + * than nameCapacity, the returned full name will be truncated. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getName(const char* localeID, + char* name, + int32_t nameCapacity, + UErrorCode* err); + +/** + * Gets the full name for the specified locale. + * Note: This has the effect of 'canonicalizing' the string to + * a certain extent. Upper and lower case are set as needed, + * and if the components were in 'POSIX' format they are changed to + * ICU format. It does NOT map aliased names in any way. + * See the top of this header file. + * + * @param localeID the locale to get the full name with + * @param name the full name for localeID + * @param nameCapacity the size of the name buffer to store the + * full name with + * @param err error information if retrieving the full name failed + * @return the actual buffer size needed for the full name. If it's greater + * than nameCapacity, the returned full name will be truncated. + * @stable ICU 2.8 + */ +U_CAPI int32_t U_EXPORT2 +uloc_canonicalize(const char* localeID, + char* name, + int32_t nameCapacity, + UErrorCode* err); + +/** + * Gets the ISO language code for the specified locale. + * + * @param localeID the locale to get the ISO language code with + * @return language the ISO language code for localeID + * @stable ICU 2.0 + */ +U_CAPI const char* U_EXPORT2 +uloc_getISO3Language(const char* localeID); + + +/** + * Gets the ISO country code for the specified locale. + * + * @param localeID the locale to get the ISO country code with + * @return country the ISO country code for localeID + * @stable ICU 2.0 + */ +U_CAPI const char* U_EXPORT2 +uloc_getISO3Country(const char* localeID); + +/** + * Gets the Win32 LCID value for the specified locale. + * If the ICU locale is not recognized by Windows, 0 will be returned. + * + * LCIDs were deprecated with Windows Vista and Microsoft recommends + * that developers use BCP47 style tags instead (uloc_toLanguageTag). + * + * @param localeID the locale to get the Win32 LCID value with + * @return country the Win32 LCID for localeID + * @stable ICU 2.0 + */ +U_CAPI uint32_t U_EXPORT2 +uloc_getLCID(const char* localeID); + +/** + * Gets the language name suitable for display for the specified locale. + * + * @param locale the locale to get the ISO language code with + * @param displayLocale Specifies the locale to be used to display the name. In + * other words, if the locale's language code is "en", passing + * Locale::getFrench() for inLocale would result in "Anglais", + * while passing Locale::getGerman() for inLocale would result + * in "Englisch". + * @param language the displayable language code for localeID + * @param languageCapacity the size of the language buffer to store the + * displayable language code with. + * @param status error information if retrieving the displayable language code + * failed. U_USING_DEFAULT_WARNING indicates that no data was + * found from the locale resources and a case canonicalized + * language code is placed into language as fallback. + * @return the actual buffer size needed for the displayable language code. If + * it's greater than languageCapacity, the returned language + * code will be truncated. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayLanguage(const char* locale, + const char* displayLocale, + UChar* language, + int32_t languageCapacity, + UErrorCode* status); + +/** + * Gets the script name suitable for display for the specified locale. + * + * @param locale the locale to get the displayable script code with. NULL may be + * used to specify the default. + * @param displayLocale Specifies the locale to be used to display the name. In + * other words, if the locale's language code is "en", passing + * Locale::getFrench() for inLocale would result in "", while + * passing Locale::getGerman() for inLocale would result in "". + * NULL may be used to specify the default. + * @param script the displayable script for the localeID. + * @param scriptCapacity the size of the script buffer to store the displayable + * script code with. + * @param status error information if retrieving the displayable script code + * failed. U_USING_DEFAULT_WARNING indicates that no data was + * found from the locale resources and a case canonicalized + * script code is placed into script as fallback. + * @return the actual buffer size needed for the displayable script code. If + * it's greater than scriptCapacity, the returned displayable + * script code will be truncated. + * @stable ICU 2.8 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayScript(const char* locale, + const char* displayLocale, + UChar* script, + int32_t scriptCapacity, + UErrorCode* status); + +/** + * Gets the country name suitable for display for the specified locale. + * Warning: this is for the region part of a valid locale ID; it cannot just be + * the region code (like "FR"). To get the display name for a region alone, or + * for other options, use ULocaleDisplayNames instead. + * + * @param locale the locale to get the displayable country code with. NULL may + * be used to specify the default. + * @param displayLocale Specifies the locale to be used to display the name. In + * other words, if the locale's language code is "en", passing + * Locale::getFrench() for inLocale would result in "Anglais", + * while passing Locale::getGerman() for inLocale would result + * in "Englisch". NULL may be used to specify the default. + * @param country the displayable country code for localeID. + * @param countryCapacity the size of the country buffer to store the + * displayable country code with. + * @param status error information if retrieving the displayable country code + * failed. U_USING_DEFAULT_WARNING indicates that no data was + * found from the locale resources and a case canonicalized + * country code is placed into country as fallback. + * @return the actual buffer size needed for the displayable country code. If + * it's greater than countryCapacity, the returned displayable + * country code will be truncated. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayCountry(const char* locale, + const char* displayLocale, + UChar* country, + int32_t countryCapacity, + UErrorCode* status); + + +/** + * Gets the variant name suitable for display for the specified locale. + * + * @param locale the locale to get the displayable variant code with. NULL may + * be used to specify the default. + * @param displayLocale Specifies the locale to be used to display the name. In + * other words, if the locale's language code is "en", passing + * Locale::getFrench() for inLocale would result in "Anglais", + * while passing Locale::getGerman() for inLocale would result + * in "Englisch". NULL may be used to specify the default. + * @param variant the displayable variant code for localeID. + * @param variantCapacity the size of the variant buffer to store the + * displayable variant code with. + * @param status error information if retrieving the displayable variant code + * failed. U_USING_DEFAULT_WARNING indicates that no data was + * found from the locale resources and a case canonicalized + * variant code is placed into variant as fallback. + * @return the actual buffer size needed for the displayable variant code. If + * it's greater than variantCapacity, the returned displayable + * variant code will be truncated. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayVariant(const char* locale, + const char* displayLocale, + UChar* variant, + int32_t variantCapacity, + UErrorCode* status); + +/** + * Gets the keyword name suitable for display for the specified locale. E.g: + * for the locale string de_DE\@collation=PHONEBOOK, this API gets the display + * string for the keyword collation. + * Usage: + * + * UErrorCode status = U_ZERO_ERROR; + * const char* keyword =NULL; + * int32_t keywordLen = 0; + * int32_t keywordCount = 0; + * UChar displayKeyword[256]; + * int32_t displayKeywordLen = 0; + * UEnumeration* keywordEnum = uloc_openKeywords("de_DE@collation=PHONEBOOK;calendar=TRADITIONAL", &status); + * for(keywordCount = uenum_count(keywordEnum, &status); keywordCount > 0 ; keywordCount--){ + * if(U_FAILURE(status)){ + * ...something went wrong so handle the error... + * break; + * } + * // the uenum_next returns NUL terminated string + * keyword = uenum_next(keywordEnum, &keywordLen, &status); + * displayKeywordLen = uloc_getDisplayKeyword(keyword, "en_US", displayKeyword, 256); + * ... do something interesting ..... + * } + * uenum_close(keywordEnum); + * + * @param keyword The keyword whose display string needs to be returned. + * @param displayLocale Specifies the locale to be used to display the name. In other words, + * if the locale's language code is "en", passing Locale::getFrench() for + * inLocale would result in "Anglais", while passing Locale::getGerman() + * for inLocale would result in "Englisch". NULL may be used to specify the default. + * @param dest the buffer to which the displayable keyword should be written. + * @param destCapacity The size of the buffer (number of UChars). If it is 0, then + * dest may be NULL and the function will only return the length of the + * result without writing any of the result string (pre-flighting). + * @param status error information if retrieving the displayable string failed. + * Should not be NULL and should not indicate failure on entry. + * U_USING_DEFAULT_WARNING indicates that no data was found from the locale + * resources and the keyword is placed into dest as fallback. + * @return the actual buffer size needed for the displayable variant code. + * @see #uloc_openKeywords + * @stable ICU 2.8 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayKeyword(const char* keyword, + const char* displayLocale, + UChar* dest, + int32_t destCapacity, + UErrorCode* status); +/** + * Gets the value of the keyword suitable for display for the specified locale. + * E.g: for the locale string de_DE\@collation=PHONEBOOK, this API gets the display + * string for PHONEBOOK, in the display locale, when "collation" is specified as the keyword. + * + * @param locale The locale to get the displayable variant code with. NULL may be used to specify the default. + * @param keyword The keyword for whose value should be used. + * @param displayLocale Specifies the locale to be used to display the name. In other words, + * if the locale's language code is "en", passing Locale::getFrench() for + * inLocale would result in "Anglais", while passing Locale::getGerman() + * for inLocale would result in "Englisch". NULL may be used to specify the default. + * @param dest the buffer to which the displayable keyword should be written. + * @param destCapacity The size of the buffer (number of UChars). If it is 0, then + * dest may be NULL and the function will only return the length of the + * result without writing any of the result string (pre-flighting). + * @param status error information if retrieving the displayable string failed. + * Should not be NULL and must not indicate failure on entry. + * U_USING_DEFAULT_WARNING indicates that no data was found from the locale + * resources and the value of the keyword is placed into dest as fallback. + * @return the actual buffer size needed for the displayable variant code. + * @stable ICU 2.8 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayKeywordValue( const char* locale, + const char* keyword, + const char* displayLocale, + UChar* dest, + int32_t destCapacity, + UErrorCode* status); +/** + * Gets the full name suitable for display for the specified locale. + * + * @param localeID the locale to get the displayable name with. NULL may be used to specify the default. + * @param inLocaleID Specifies the locale to be used to display the name. In other words, + * if the locale's language code is "en", passing Locale::getFrench() for + * inLocale would result in "Anglais", while passing Locale::getGerman() + * for inLocale would result in "Englisch". NULL may be used to specify the default. + * @param result the displayable name for localeID + * @param maxResultSize the size of the name buffer to store the + * displayable full name with + * @param err error information if retrieving the displayable name failed + * @return the actual buffer size needed for the displayable name. If it's greater + * than maxResultSize, the returned displayable name will be truncated. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getDisplayName(const char* localeID, + const char* inLocaleID, + UChar* result, + int32_t maxResultSize, + UErrorCode* err); + + +/** + * Gets the specified locale from a list of available locales. + * + * This method corresponds to uloc_openAvailableByType called with the + * ULOC_AVAILABLE_DEFAULT type argument. + * + * The return value is a pointer to an item of a locale name array. Both this + * array and the pointers it contains are owned by ICU and should not be + * deleted or written through by the caller. The locale name is terminated by + * a null pointer. + * + * @param n the specific locale name index of the available locale list; + * should not exceed the number returned by uloc_countAvailable. + * @return a specified locale name of all available locales + * @stable ICU 2.0 + */ +U_CAPI const char* U_EXPORT2 +uloc_getAvailable(int32_t n); + +/** + * Gets the size of the all available locale list. + * + * @return the size of the locale list + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 uloc_countAvailable(void); + +/** + * Types for uloc_getAvailableByType and uloc_countAvailableByType. + * + * @stable ICU 65 + */ +typedef enum ULocAvailableType { + /** + * Locales that return data when passed to ICU APIs, + * but not including legacy or alias locales. + * + * @stable ICU 65 + */ + ULOC_AVAILABLE_DEFAULT, + + /** + * Legacy or alias locales that return data when passed to ICU APIs. + * Examples of supported legacy or alias locales: + * + * - iw (alias to he) + * - mo (alias to ro) + * - zh_CN (alias to zh_Hans_CN) + * - sr_BA (alias to sr_Cyrl_BA) + * - ars (alias to ar_SA) + * + * The locales in this set are disjoint from the ones in + * ULOC_AVAILABLE_DEFAULT. To get both sets at the same time, use + * ULOC_AVAILABLE_WITH_LEGACY_ALIASES. + * + * @stable ICU 65 + */ + ULOC_AVAILABLE_ONLY_LEGACY_ALIASES, + + /** + * The union of the locales in ULOC_AVAILABLE_DEFAULT and + * ULOC_AVAILABLE_ONLY_LEGACY_ALIAS. + * + * @stable ICU 65 + */ + ULOC_AVAILABLE_WITH_LEGACY_ALIASES, + +#ifndef U_HIDE_INTERNAL_API + /** + * @internal + */ + ULOC_AVAILABLE_COUNT +#endif /* U_HIDE_INTERNAL_API */ +} ULocAvailableType; + +/** + * Gets a list of available locales according to the type argument, allowing + * the user to access different sets of supported locales in ICU. + * + * The returned UEnumeration must be closed by the caller. + * + * @param type Type choice from ULocAvailableType. + * @param status Set if an error occurred. + * @return a UEnumeration owned by the caller, or nullptr on failure. + * @stable ICU 65 + */ +U_CAPI UEnumeration* U_EXPORT2 +uloc_openAvailableByType(ULocAvailableType type, UErrorCode* status); + +/** + * + * Gets a list of all available 2-letter language codes defined in ISO 639, + * plus additional 3-letter codes determined to be useful for locale generation as + * defined by Unicode CLDR. This is a pointer + * to an array of pointers to arrays of char. All of these pointers are owned + * by ICU-- do not delete them, and do not write through them. The array is + * terminated with a null pointer. + * @return a list of all available language codes + * @stable ICU 2.0 + */ +U_CAPI const char* const* U_EXPORT2 +uloc_getISOLanguages(void); + +/** + * + * Gets a list of all available 2-letter country codes defined in ISO 639. This is a + * pointer to an array of pointers to arrays of char. All of these pointers are + * owned by ICU-- do not delete them, and do not write through them. The array is + * terminated with a null pointer. + * @return a list of all available country codes + * @stable ICU 2.0 + */ +U_CAPI const char* const* U_EXPORT2 +uloc_getISOCountries(void); + +/** + * Truncate the locale ID string to get the parent locale ID. + * Copies the part of the string before the last underscore. + * The parent locale ID will be an empty string if there is no + * underscore, or if there is only one underscore at localeID[0]. + * + * @param localeID Input locale ID string. + * @param parent Output string buffer for the parent locale ID. + * @param parentCapacity Size of the output buffer. + * @param err A UErrorCode value. + * @return The length of the parent locale ID. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getParent(const char* localeID, + char* parent, + int32_t parentCapacity, + UErrorCode* err); + + + + +/** + * Gets the full name for the specified locale, like uloc_getName(), + * but without keywords. + * + * Note: This has the effect of 'canonicalizing' the string to + * a certain extent. Upper and lower case are set as needed, + * and if the components were in 'POSIX' format they are changed to + * ICU format. It does NOT map aliased names in any way. + * See the top of this header file. + * + * This API strips off the keyword part, so "de_DE\@collation=phonebook" + * will become "de_DE". + * This API supports preflighting. + * + * @param localeID the locale to get the full name with + * @param name fill in buffer for the name without keywords. + * @param nameCapacity capacity of the fill in buffer. + * @param err error information if retrieving the full name failed + * @return the actual buffer size needed for the full name. If it's greater + * than nameCapacity, the returned full name will be truncated. + * @stable ICU 2.8 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getBaseName(const char* localeID, + char* name, + int32_t nameCapacity, + UErrorCode* err); + +/** + * Gets an enumeration of keywords for the specified locale. Enumeration + * must get disposed of by the client using uenum_close function. + * + * @param localeID the locale to get the variant code with + * @param status error information if retrieving the keywords failed + * @return enumeration of keywords or NULL if there are no keywords. + * @stable ICU 2.8 + */ +U_CAPI UEnumeration* U_EXPORT2 +uloc_openKeywords(const char* localeID, + UErrorCode* status); + +/** + * Get the value for a keyword. Locale name does not need to be normalized. + * + * @param localeID locale name containing the keyword ("de_DE@currency=EURO;collation=PHONEBOOK") + * @param keywordName name of the keyword for which we want the value; must not be + * NULL or empty, and must consist only of [A-Za-z0-9]. Case insensitive. + * @param buffer receiving buffer + * @param bufferCapacity capacity of receiving buffer + * @param status containing error code: e.g. buffer not big enough or ill-formed localeID + * or keywordName parameters. + * @return the length of keyword value + * @stable ICU 2.8 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getKeywordValue(const char* localeID, + const char* keywordName, + char* buffer, int32_t bufferCapacity, + UErrorCode* status); + + +/** + * Sets or removes the value of the specified keyword. + * + * For removing all keywords, use uloc_getBaseName(). + * + * NOTE: Unlike almost every other ICU function which takes a + * buffer, this function will NOT truncate the output text, and will + * not update the buffer with unterminated text setting a status of + * U_STRING_NOT_TERMINATED_WARNING. If a BUFFER_OVERFLOW_ERROR is received, + * it means a terminated version of the updated locale ID would not fit + * in the buffer, and the original buffer is untouched. This is done to + * prevent incorrect or possibly even malformed locales from being generated + * and used. + * + * @param keywordName name of the keyword to be set; must not be + * NULL or empty, and must consist only of [A-Za-z0-9]. Case insensitive. + * @param keywordValue value of the keyword to be set. If 0-length or + * NULL, will result in the keyword being removed; no error is given if + * that keyword does not exist. Otherwise, must consist only of + * [A-Za-z0-9] and [/_+-]. + * @param buffer input buffer containing well-formed locale ID to be + * modified. + * @param bufferCapacity capacity of receiving buffer + * @param status containing error code: e.g. buffer not big enough + * or ill-formed keywordName or keywordValue parameters, or ill-formed + * locale ID in buffer on input. + * @return the length needed for the buffer + * @see uloc_getKeywordValue + * @stable ICU 3.2 + */ +U_CAPI int32_t U_EXPORT2 +uloc_setKeywordValue(const char* keywordName, + const char* keywordValue, + char* buffer, int32_t bufferCapacity, + UErrorCode* status); + +/** + * Returns whether the locale's script is written right-to-left. + * If there is no script subtag, then the likely script is used, see uloc_addLikelySubtags(). + * If no likely script is known, then false is returned. + * + * A script is right-to-left according to the CLDR script metadata + * which corresponds to whether the script's letters have Bidi_Class=R or AL. + * + * Returns true for "ar" and "en-Hebr", false for "zh" and "fa-Cyrl". + * + * @param locale input locale ID + * @return true if the locale's script is written right-to-left + * @stable ICU 54 + */ +U_CAPI UBool U_EXPORT2 +uloc_isRightToLeft(const char *locale); + +/** + * enums for the return value for the character and line orientation + * functions. + * @stable ICU 4.0 + */ +typedef enum { + ULOC_LAYOUT_LTR = 0, /* left-to-right. */ + ULOC_LAYOUT_RTL = 1, /* right-to-left. */ + ULOC_LAYOUT_TTB = 2, /* top-to-bottom. */ + ULOC_LAYOUT_BTT = 3, /* bottom-to-top. */ + ULOC_LAYOUT_UNKNOWN +} ULayoutType; + +/** + * Get the layout character orientation for the specified locale. + * + * @param localeId locale name + * @param status Error status + * @return an enum indicating the layout orientation for characters. + * @stable ICU 4.0 + */ +U_CAPI ULayoutType U_EXPORT2 +uloc_getCharacterOrientation(const char* localeId, + UErrorCode *status); + +/** + * Get the layout line orientation for the specified locale. + * + * @param localeId locale name + * @param status Error status + * @return an enum indicating the layout orientation for lines. + * @stable ICU 4.0 + */ +U_CAPI ULayoutType U_EXPORT2 +uloc_getLineOrientation(const char* localeId, + UErrorCode *status); + +/** + * Output values which uloc_acceptLanguage() writes to the 'outResult' parameter. + * + * @see uloc_acceptLanguageFromHTTP + * @see uloc_acceptLanguage + * @stable ICU 3.2 + */ +typedef enum { + /** + * No exact match was found. + * @stable ICU 3.2 + */ + ULOC_ACCEPT_FAILED = 0, + /** + * An exact match was found. + * @stable ICU 3.2 + */ + ULOC_ACCEPT_VALID = 1, + /** + * A fallback was found. For example, the Accept-Language list includes 'ja_JP' + * and is matched with available locale 'ja'. + * @stable ICU 3.2 + */ + ULOC_ACCEPT_FALLBACK = 2 /* */ +} UAcceptResult; + +/** + * Based on a HTTP header from a web browser and a list of available locales, + * determine an acceptable locale for the user. + * + * This is a thin wrapper over C++ class LocaleMatcher. + * + * @param result - buffer to accept the result locale + * @param resultAvailable the size of the result buffer. + * @param outResult - An out parameter that contains the fallback status + * @param httpAcceptLanguage - "Accept-Language:" header as per HTTP. + * @param availableLocales - list of available locales to match + * @param status ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return length needed for the locale. + * @stable ICU 3.2 + */ +U_CAPI int32_t U_EXPORT2 +uloc_acceptLanguageFromHTTP(char *result, int32_t resultAvailable, + UAcceptResult *outResult, + const char *httpAcceptLanguage, + UEnumeration* availableLocales, + UErrorCode *status); + +/** + * Based on a list of available locales, + * determine an acceptable locale for the user. + * + * This is a thin wrapper over C++ class LocaleMatcher. + * + * @param result - buffer to accept the result locale + * @param resultAvailable the size of the result buffer. + * @param outResult - An out parameter that contains the fallback status + * @param acceptList - list of acceptable languages + * @param acceptListCount - count of acceptList items + * @param availableLocales - list of available locales to match + * @param status ICU error code. Its input value must pass the U_SUCCESS() test, + * or else the function returns immediately. Check for U_FAILURE() + * on output or use with function chaining. (See User Guide for details.) + * @return length needed for the locale. + * @stable ICU 3.2 + */ +U_CAPI int32_t U_EXPORT2 +uloc_acceptLanguage(char *result, int32_t resultAvailable, + UAcceptResult *outResult, const char **acceptList, + int32_t acceptListCount, + UEnumeration* availableLocales, + UErrorCode *status); + + +/** + * Gets the ICU locale ID for the specified Win32 LCID value. + * + * @param hostID the Win32 LCID to translate + * @param locale the output buffer for the ICU locale ID, which will be NUL-terminated + * if there is room. + * @param localeCapacity the size of the output buffer + * @param status an error is returned if the LCID is unrecognized or the output buffer + * is too small + * @return actual the actual size of the locale ID, not including NUL-termination + * @stable ICU 3.8 + */ +U_CAPI int32_t U_EXPORT2 +uloc_getLocaleForLCID(uint32_t hostID, char *locale, int32_t localeCapacity, + UErrorCode *status); + + +/** + * Add the likely subtags for a provided locale ID, per the algorithm described + * in the following CLDR technical report: + * + * http://www.unicode.org/reports/tr35/#Likely_Subtags + * + * If localeID is already in the maximal form, or there is no data available + * for maximization, it will be copied to the output buffer. For example, + * "und-Zzzz" cannot be maximized, since there is no reasonable maximization. + * + * Examples: + * + * "en" maximizes to "en_Latn_US" + * + * "de" maximizes to "de_Latn_US" + * + * "sr" maximizes to "sr_Cyrl_RS" + * + * "sh" maximizes to "sr_Latn_RS" (Note this will not reverse.) + * + * "zh_Hani" maximizes to "zh_Hans_CN" (Note this will not reverse.) + * + * @param localeID The locale to maximize + * @param maximizedLocaleID The maximized locale + * @param maximizedLocaleIDCapacity The capacity of the maximizedLocaleID buffer + * @param err Error information if maximizing the locale failed. If the length + * of the localeID and the null-terminator is greater than the maximum allowed size, + * or the localeId is not well-formed, the error code is U_ILLEGAL_ARGUMENT_ERROR. + * @return The actual buffer size needed for the maximized locale. If it's + * greater than maximizedLocaleIDCapacity, the returned ID will be truncated. + * On error, the return value is -1. + * @stable ICU 4.0 + */ +U_CAPI int32_t U_EXPORT2 +uloc_addLikelySubtags(const char* localeID, + char* maximizedLocaleID, + int32_t maximizedLocaleIDCapacity, + UErrorCode* err); + + +/** + * Minimize the subtags for a provided locale ID, per the algorithm described + * in the following CLDR technical report: + * + * http://www.unicode.org/reports/tr35/#Likely_Subtags + * + * If localeID is already in the minimal form, or there is no data available + * for minimization, it will be copied to the output buffer. Since the + * minimization algorithm relies on proper maximization, see the comments + * for uloc_addLikelySubtags for reasons why there might not be any data. + * + * Examples: + * + * "en_Latn_US" minimizes to "en" + * + * "de_Latn_US" minimizes to "de" + * + * "sr_Cyrl_RS" minimizes to "sr" + * + * "zh_Hant_TW" minimizes to "zh_TW" (The region is preferred to the + * script, and minimizing to "zh" would imply "zh_Hans_CN".) + * + * @param localeID The locale to minimize + * @param minimizedLocaleID The minimized locale + * @param minimizedLocaleIDCapacity The capacity of the minimizedLocaleID buffer + * @param err Error information if minimizing the locale failed. If the length + * of the localeID and the null-terminator is greater than the maximum allowed size, + * or the localeId is not well-formed, the error code is U_ILLEGAL_ARGUMENT_ERROR. + * @return The actual buffer size needed for the minimized locale. If it's + * greater than minimizedLocaleIDCapacity, the returned ID will be truncated. + * On error, the return value is -1. + * @stable ICU 4.0 + */ +U_CAPI int32_t U_EXPORT2 +uloc_minimizeSubtags(const char* localeID, + char* minimizedLocaleID, + int32_t minimizedLocaleIDCapacity, + UErrorCode* err); + +/** + * Returns a locale ID for the specified BCP47 language tag string. + * If the specified language tag contains any ill-formed subtags, + * the first such subtag and all following subtags are ignored. + *

+ * This implements the 'Language-Tag' production of BCP 47, and so + * supports legacy language tags (marked as “Type: grandfathered” in BCP 47) + * (regular and irregular) as well as private use language tags. + * + * Private use tags are represented as 'x-whatever', + * and legacy tags are converted to their canonical replacements where they exist. + * + * Note that a few legacy tags have no modern replacement; + * these will be converted using the fallback described in + * the first paragraph, so some information might be lost. + * + * @param langtag the input BCP47 language tag. + * @param localeID the output buffer receiving a locale ID for the + * specified BCP47 language tag. + * @param localeIDCapacity the size of the locale ID output buffer. + * @param parsedLength if not NULL, successfully parsed length + * for the input language tag is set. + * @param err error information if receiving the locald ID + * failed. + * @return the length of the locale ID. + * @stable ICU 4.2 + */ +U_CAPI int32_t U_EXPORT2 +uloc_forLanguageTag(const char* langtag, + char* localeID, + int32_t localeIDCapacity, + int32_t* parsedLength, + UErrorCode* err); + +/** + * Returns a well-formed language tag for this locale ID. + *

+ * Note: When strict is false, any locale + * fields which do not satisfy the BCP47 syntax requirement will + * be omitted from the result. When strict is + * true, this function sets U_ILLEGAL_ARGUMENT_ERROR to the + * err if any locale fields do not satisfy the + * BCP47 syntax requirement. + * @param localeID the input locale ID + * @param langtag the output buffer receiving BCP47 language + * tag for the locale ID. + * @param langtagCapacity the size of the BCP47 language tag + * output buffer. + * @param strict boolean value indicating if the function returns + * an error for an ill-formed input locale ID. + * @param err error information if receiving the language + * tag failed. + * @return The length of the BCP47 language tag. + * @stable ICU 4.2 + */ +U_CAPI int32_t U_EXPORT2 +uloc_toLanguageTag(const char* localeID, + char* langtag, + int32_t langtagCapacity, + UBool strict, + UErrorCode* err); + +/** + * Converts the specified keyword (legacy key, or BCP 47 Unicode locale + * extension key) to the equivalent BCP 47 Unicode locale extension key. + * For example, BCP 47 Unicode locale extension key "co" is returned for + * the input keyword "collation". + *

+ * When the specified keyword is unknown, but satisfies the BCP syntax, + * then the pointer to the input keyword itself will be returned. + * For example, + * uloc_toUnicodeLocaleKey("ZZ") returns "ZZ". + * + * @param keyword the input locale keyword (either legacy key + * such as "collation" or BCP 47 Unicode locale extension + * key such as "co"). + * @return the well-formed BCP 47 Unicode locale extension key, + * or NULL if the specified locale keyword cannot be + * mapped to a well-formed BCP 47 Unicode locale extension + * key. + * @see uloc_toLegacyKey + * @stable ICU 54 + */ +U_CAPI const char* U_EXPORT2 +uloc_toUnicodeLocaleKey(const char* keyword); + +/** + * Converts the specified keyword value (legacy type, or BCP 47 + * Unicode locale extension type) to the well-formed BCP 47 Unicode locale + * extension type for the specified keyword (category). For example, BCP 47 + * Unicode locale extension type "phonebk" is returned for the input + * keyword value "phonebook", with the keyword "collation" (or "co"). + *

+ * When the specified keyword is not recognized, but the specified value + * satisfies the syntax of the BCP 47 Unicode locale extension type, + * or when the specified keyword allows 'variable' type and the specified + * value satisfies the syntax, then the pointer to the input type value itself + * will be returned. + * For example, + * uloc_toUnicodeLocaleType("Foo", "Bar") returns "Bar", + * uloc_toUnicodeLocaleType("variableTop", "00A4") returns "00A4". + * + * @param keyword the locale keyword (either legacy key such as + * "collation" or BCP 47 Unicode locale extension + * key such as "co"). + * @param value the locale keyword value (either legacy type + * such as "phonebook" or BCP 47 Unicode locale extension + * type such as "phonebk"). + * @return the well-formed BCP47 Unicode locale extension type, + * or NULL if the locale keyword value cannot be mapped to + * a well-formed BCP 47 Unicode locale extension type. + * @see uloc_toLegacyType + * @stable ICU 54 + */ +U_CAPI const char* U_EXPORT2 +uloc_toUnicodeLocaleType(const char* keyword, const char* value); + +/** + * Converts the specified keyword (BCP 47 Unicode locale extension key, or + * legacy key) to the legacy key. For example, legacy key "collation" is + * returned for the input BCP 47 Unicode locale extension key "co". + * + * @param keyword the input locale keyword (either BCP 47 Unicode locale + * extension key or legacy key). + * @return the well-formed legacy key, or NULL if the specified + * keyword cannot be mapped to a well-formed legacy key. + * @see toUnicodeLocaleKey + * @stable ICU 54 + */ +U_CAPI const char* U_EXPORT2 +uloc_toLegacyKey(const char* keyword); + +/** + * Converts the specified keyword value (BCP 47 Unicode locale extension type, + * or legacy type or type alias) to the canonical legacy type. For example, + * the legacy type "phonebook" is returned for the input BCP 47 Unicode + * locale extension type "phonebk" with the keyword "collation" (or "co"). + *

+ * When the specified keyword is not recognized, but the specified value + * satisfies the syntax of legacy key, or when the specified keyword + * allows 'variable' type and the specified value satisfies the syntax, + * then the pointer to the input type value itself will be returned. + * For example, + * uloc_toLegacyType("Foo", "Bar") returns "Bar", + * uloc_toLegacyType("vt", "00A4") returns "00A4". + * + * @param keyword the locale keyword (either legacy keyword such as + * "collation" or BCP 47 Unicode locale extension + * key such as "co"). + * @param value the locale keyword value (either BCP 47 Unicode locale + * extension type such as "phonebk" or legacy keyword value + * such as "phonebook"). + * @return the well-formed legacy type, or NULL if the specified + * keyword value cannot be mapped to a well-formed legacy + * type. + * @see toUnicodeLocaleType + * @stable ICU 54 + */ +U_CAPI const char* U_EXPORT2 +uloc_toLegacyType(const char* keyword, const char* value); + +#endif /*_ULOC*/ diff --git a/intl/icu/source/common/unicode/umachine.h b/intl/icu/source/common/unicode/umachine.h new file mode 100644 index 0000000000..545abef595 --- /dev/null +++ b/intl/icu/source/common/unicode/umachine.h @@ -0,0 +1,459 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 1999-2015, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: umachine.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 1999sep13 +* created by: Markus W. Scherer +* +* This file defines basic types and constants for ICU to be +* platform-independent. umachine.h and utf.h are included into +* utypes.h to provide all the general definitions for ICU. +* All of these definitions used to be in utypes.h before +* the UTF-handling macros made this unmaintainable. +*/ + +#ifndef __UMACHINE_H__ +#define __UMACHINE_H__ + + +/** + * \file + * \brief Basic types and constants for UTF + * + *

Basic types and constants for UTF

+ * This file defines basic types and constants for utf.h to be + * platform-independent. umachine.h and utf.h are included into + * utypes.h to provide all the general definitions for ICU. + * All of these definitions used to be in utypes.h before + * the UTF-handling macros made this unmaintainable. + * + */ +/*==========================================================================*/ +/* Include platform-dependent definitions */ +/* which are contained in the platform-specific file platform.h */ +/*==========================================================================*/ + +#include "unicode/ptypes.h" /* platform.h is included in ptypes.h */ + +/* + * ANSI C headers: + * stddef.h defines wchar_t + */ +#include +#include + +/*==========================================================================*/ +/* For C wrappers, we use the symbol U_CAPI. */ +/* This works properly if the includer is C or C++. */ +/* Functions are declared U_CAPI return-type U_EXPORT2 function-name()... */ +/*==========================================================================*/ + +/** + * \def U_CFUNC + * This is used in a declaration of a library private ICU C function. + * @stable ICU 2.4 + */ + +/** + * \def U_CDECL_BEGIN + * This is used to begin a declaration of a library private ICU C API. + * @stable ICU 2.4 + */ + +/** + * \def U_CDECL_END + * This is used to end a declaration of a library private ICU C API + * @stable ICU 2.4 + */ + +#ifdef __cplusplus +# define U_CFUNC extern "C" +# define U_CDECL_BEGIN extern "C" { +# define U_CDECL_END } +#else +# define U_CFUNC extern +# define U_CDECL_BEGIN +# define U_CDECL_END +#endif + +#ifndef U_ATTRIBUTE_DEPRECATED +/** + * \def U_ATTRIBUTE_DEPRECATED + * This is used for GCC specific attributes + * @internal + */ +#if U_GCC_MAJOR_MINOR >= 302 +# define U_ATTRIBUTE_DEPRECATED __attribute__ ((deprecated)) +/** + * \def U_ATTRIBUTE_DEPRECATED + * This is used for Visual C++ specific attributes + * @internal + */ +#elif defined(_MSC_VER) && (_MSC_VER >= 1400) +# define U_ATTRIBUTE_DEPRECATED __declspec(deprecated) +#else +# define U_ATTRIBUTE_DEPRECATED +#endif +#endif + +/** This is used to declare a function as a public ICU C API @stable ICU 2.0*/ +#define U_CAPI U_CFUNC U_EXPORT +/** Obsolete/same as U_CAPI; was used to declare a function as a stable public ICU C API*/ +#define U_STABLE U_CAPI +/** Obsolete/same as U_CAPI; was used to declare a function as a draft public ICU C API */ +#define U_DRAFT U_CAPI +/** This is used to declare a function as a deprecated public ICU C API */ +#define U_DEPRECATED U_CAPI U_ATTRIBUTE_DEPRECATED +/** Obsolete/same as U_CAPI; was used to declare a function as an obsolete public ICU C API */ +#define U_OBSOLETE U_CAPI +/** Obsolete/same as U_CAPI; was used to declare a function as an internal ICU C API */ +#define U_INTERNAL U_CAPI + +// Before ICU 65, function-like, multi-statement ICU macros were just defined as +// series of statements wrapped in { } blocks and the caller could choose to +// either treat them as if they were actual functions and end the invocation +// with a trailing ; creating an empty statement after the block or else omit +// this trailing ; using the knowledge that the macro would expand to { }. +// +// But doing so doesn't work well with macros that look like functions and +// compiler warnings about empty statements (ICU-20601) and ICU 65 therefore +// switches to the standard solution of wrapping such macros in do { } while. +// +// This will however break existing code that depends on being able to invoke +// these macros without a trailing ; so to be able to remain compatible with +// such code the wrapper is itself defined as macros so that it's possible to +// build ICU 65 and later with the old macro behaviour, like this: +// +// export CPPFLAGS='-DUPRV_BLOCK_MACRO_BEGIN="" -DUPRV_BLOCK_MACRO_END=""' +// runConfigureICU ... +// + +/** + * \def UPRV_BLOCK_MACRO_BEGIN + * Defined as the "do" keyword by default. + * @internal + */ +#ifndef UPRV_BLOCK_MACRO_BEGIN +#define UPRV_BLOCK_MACRO_BEGIN do +#endif + +/** + * \def UPRV_BLOCK_MACRO_END + * Defined as "while (false)" by default. + * @internal + */ +#ifndef UPRV_BLOCK_MACRO_END +#define UPRV_BLOCK_MACRO_END while (false) +#endif + +/*==========================================================================*/ +/* limits for int32_t etc., like in POSIX inttypes.h */ +/*==========================================================================*/ + +#ifndef INT8_MIN +/** The smallest value an 8 bit signed integer can hold @stable ICU 2.0 */ +# define INT8_MIN ((int8_t)(-128)) +#endif +#ifndef INT16_MIN +/** The smallest value a 16 bit signed integer can hold @stable ICU 2.0 */ +# define INT16_MIN ((int16_t)(-32767-1)) +#endif +#ifndef INT32_MIN +/** The smallest value a 32 bit signed integer can hold @stable ICU 2.0 */ +# define INT32_MIN ((int32_t)(-2147483647-1)) +#endif + +#ifndef INT8_MAX +/** The largest value an 8 bit signed integer can hold @stable ICU 2.0 */ +# define INT8_MAX ((int8_t)(127)) +#endif +#ifndef INT16_MAX +/** The largest value a 16 bit signed integer can hold @stable ICU 2.0 */ +# define INT16_MAX ((int16_t)(32767)) +#endif +#ifndef INT32_MAX +/** The largest value a 32 bit signed integer can hold @stable ICU 2.0 */ +# define INT32_MAX ((int32_t)(2147483647)) +#endif + +#ifndef UINT8_MAX +/** The largest value an 8 bit unsigned integer can hold @stable ICU 2.0 */ +# define UINT8_MAX ((uint8_t)(255U)) +#endif +#ifndef UINT16_MAX +/** The largest value a 16 bit unsigned integer can hold @stable ICU 2.0 */ +# define UINT16_MAX ((uint16_t)(65535U)) +#endif +#ifndef UINT32_MAX +/** The largest value a 32 bit unsigned integer can hold @stable ICU 2.0 */ +# define UINT32_MAX ((uint32_t)(4294967295U)) +#endif + +#if defined(U_INT64_T_UNAVAILABLE) +# error int64_t is required for decimal format and rule-based number format. +#else +# ifndef INT64_C +/** + * Provides a platform independent way to specify a signed 64-bit integer constant. + * note: may be wrong for some 64 bit platforms - ensure your compiler provides INT64_C + * @stable ICU 2.8 + */ +# define INT64_C(c) c ## LL +# endif +# ifndef UINT64_C +/** + * Provides a platform independent way to specify an unsigned 64-bit integer constant. + * note: may be wrong for some 64 bit platforms - ensure your compiler provides UINT64_C + * @stable ICU 2.8 + */ +# define UINT64_C(c) c ## ULL +# endif +# ifndef U_INT64_MIN +/** The smallest value a 64 bit signed integer can hold @stable ICU 2.8 */ +# define U_INT64_MIN ((int64_t)(INT64_C(-9223372036854775807)-1)) +# endif +# ifndef U_INT64_MAX +/** The largest value a 64 bit signed integer can hold @stable ICU 2.8 */ +# define U_INT64_MAX ((int64_t)(INT64_C(9223372036854775807))) +# endif +# ifndef U_UINT64_MAX +/** The largest value a 64 bit unsigned integer can hold @stable ICU 2.8 */ +# define U_UINT64_MAX ((uint64_t)(UINT64_C(18446744073709551615))) +# endif +#endif + +/*==========================================================================*/ +/* Boolean data type */ +/*==========================================================================*/ + +/** + * The ICU boolean type, a signed-byte integer. + * ICU-specific for historical reasons: The C and C++ standards used to not define type bool. + * Also provides a fixed type definition, as opposed to + * type bool whose details (e.g., sizeof) may vary by compiler and between C and C++. + * + * @stable ICU 2.0 + */ +typedef int8_t UBool; + +/** + * \def U_DEFINE_FALSE_AND_TRUE + * Normally turns off defining macros FALSE=0 & TRUE=1 in public ICU headers. + * These obsolete macros sometimes break compilation of other code that + * defines enum constants or similar with these names. + * C++ has long defined bool/false/true. + * C99 also added definitions for these, although as macros; see stdbool.h. + * + * You may transitionally define U_DEFINE_FALSE_AND_TRUE=1 if you need time to migrate code. + * + * @internal ICU 68 + */ +#ifdef U_DEFINE_FALSE_AND_TRUE + // Use the predefined value. +#else + // Default to avoiding collision with non-macro definitions of FALSE & TRUE. +# define U_DEFINE_FALSE_AND_TRUE 0 +#endif + +#if U_DEFINE_FALSE_AND_TRUE || defined(U_IN_DOXYGEN) +#ifndef TRUE +/** + * The TRUE value of a UBool. + * + * @deprecated ICU 68 Use standard "true" instead. + */ +# define TRUE 1 +#endif +#ifndef FALSE +/** + * The FALSE value of a UBool. + * + * @deprecated ICU 68 Use standard "false" instead. + */ +# define FALSE 0 +#endif +#endif // U_DEFINE_FALSE_AND_TRUE + +/*==========================================================================*/ +/* Unicode data types */ +/*==========================================================================*/ + +/* wchar_t-related definitions -------------------------------------------- */ + +/* + * \def U_WCHAR_IS_UTF16 + * Defined if wchar_t uses UTF-16. + * + * @stable ICU 2.0 + */ +/* + * \def U_WCHAR_IS_UTF32 + * Defined if wchar_t uses UTF-32. + * + * @stable ICU 2.0 + */ +#if !defined(U_WCHAR_IS_UTF16) && !defined(U_WCHAR_IS_UTF32) +# ifdef __STDC_ISO_10646__ +# if (U_SIZEOF_WCHAR_T==2) +# define U_WCHAR_IS_UTF16 +# elif (U_SIZEOF_WCHAR_T==4) +# define U_WCHAR_IS_UTF32 +# endif +# elif defined __UCS2__ +# if (U_PF_OS390 <= U_PLATFORM && U_PLATFORM <= U_PF_OS400) && (U_SIZEOF_WCHAR_T==2) +# define U_WCHAR_IS_UTF16 +# endif +# elif defined(__UCS4__) || (U_PLATFORM == U_PF_OS400 && defined(__UTF32__)) +# if (U_SIZEOF_WCHAR_T==4) +# define U_WCHAR_IS_UTF32 +# endif +# elif U_PLATFORM_IS_DARWIN_BASED || (U_SIZEOF_WCHAR_T==4 && U_PLATFORM_IS_LINUX_BASED) +# define U_WCHAR_IS_UTF32 +# elif U_PLATFORM_HAS_WIN32_API +# define U_WCHAR_IS_UTF16 +# endif +#endif + +/* UChar and UChar32 definitions -------------------------------------------- */ + +/** Number of bytes in a UChar (always 2). @stable ICU 2.0 */ +#define U_SIZEOF_UCHAR 2 + +/** + * \def U_CHAR16_IS_TYPEDEF + * If 1, then char16_t is a typedef and not a real type (yet) + * @internal + */ +#if defined(_MSC_VER) && (_MSC_VER < 1900) +// Versions of Visual Studio/MSVC below 2015 do not support char16_t as a real type, +// and instead use a typedef. https://msdn.microsoft.com/library/bb531344.aspx +# define U_CHAR16_IS_TYPEDEF 1 +#else +# define U_CHAR16_IS_TYPEDEF 0 +#endif + + +/** + * \var UChar + * + * The base type for UTF-16 code units and pointers. + * Unsigned 16-bit integer. + * Starting with ICU 59, C++ API uses char16_t directly, while C API continues to use UChar. + * + * UChar is configurable by defining the macro UCHAR_TYPE + * on the preprocessor or compiler command line: + * -DUCHAR_TYPE=uint16_t or -DUCHAR_TYPE=wchar_t (if U_SIZEOF_WCHAR_T==2) etc. + * (The UCHAR_TYPE can also be \#defined earlier in this file, for outside the ICU library code.) + * This is for transitional use from application code that uses uint16_t or wchar_t for UTF-16. + * + * The default is UChar=char16_t. + * + * C++11 defines char16_t as bit-compatible with uint16_t, but as a distinct type. + * + * In C, char16_t is a simple typedef of uint_least16_t. + * ICU requires uint_least16_t=uint16_t for data memory mapping. + * On macOS, char16_t is not available because the uchar.h standard header is missing. + * + * @stable ICU 4.4 + */ + +#if 1 + // #if 1 is normal. UChar defaults to char16_t in C++. + // For configuration testing of UChar=uint16_t temporarily change this to #if 0. + // The intltest Makefile #defines UCHAR_TYPE=char16_t, + // so we only #define it to uint16_t if it is undefined so far. +#elif !defined(UCHAR_TYPE) +# define UCHAR_TYPE uint16_t +#endif + +#if defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || \ + defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION) + // Inside the ICU library code, never configurable. + typedef char16_t UChar; +#elif defined(UCHAR_TYPE) + typedef UCHAR_TYPE UChar; +#elif U_CPLUSPLUS_VERSION != 0 + typedef char16_t UChar; // C++ +#else + typedef uint16_t UChar; // C +#endif + +/** + * \var OldUChar + * Default ICU 58 definition of UChar. + * A base type for UTF-16 code units and pointers. + * Unsigned 16-bit integer. + * + * Define OldUChar to be wchar_t if that is 16 bits wide. + * If wchar_t is not 16 bits wide, then define UChar to be uint16_t. + * + * This makes the definition of OldUChar platform-dependent + * but allows direct string type compatibility with platforms with + * 16-bit wchar_t types. + * + * This is how UChar was defined in ICU 58, for transition convenience. + * Exception: ICU 58 UChar was defined to UCHAR_TYPE if that macro was defined. + * The current UChar responds to UCHAR_TYPE but OldUChar does not. + * + * @stable ICU 59 + */ +#if U_SIZEOF_WCHAR_T==2 + typedef wchar_t OldUChar; +#elif defined(__CHAR16_TYPE__) + typedef __CHAR16_TYPE__ OldUChar; +#else + typedef uint16_t OldUChar; +#endif + +/** + * Define UChar32 as a type for single Unicode code points. + * UChar32 is a signed 32-bit integer (same as int32_t). + * + * The Unicode code point range is 0..0x10ffff. + * All other values (negative or >=0x110000) are illegal as Unicode code points. + * They may be used as sentinel values to indicate "done", "error" + * or similar non-code point conditions. + * + * Before ICU 2.4 (Jitterbug 2146), UChar32 was defined + * to be wchar_t if that is 32 bits wide (wchar_t may be signed or unsigned) + * or else to be uint32_t. + * That is, the definition of UChar32 was platform-dependent. + * + * @see U_SENTINEL + * @stable ICU 2.4 + */ +typedef int32_t UChar32; + +/** + * This value is intended for sentinel values for APIs that + * (take or) return single code points (UChar32). + * It is outside of the Unicode code point range 0..0x10ffff. + * + * For example, a "done" or "error" value in a new API + * could be indicated with U_SENTINEL. + * + * ICU APIs designed before ICU 2.4 usually define service-specific "done" + * values, mostly 0xffff. + * Those may need to be distinguished from + * actual U+ffff text contents by calling functions like + * CharacterIterator::hasNext() or UnicodeString::length(). + * + * @return -1 + * @see UChar32 + * @stable ICU 2.4 + */ +#define U_SENTINEL (-1) + +#include "unicode/urename.h" + +#endif diff --git a/intl/icu/source/common/unicode/umisc.h b/intl/icu/source/common/unicode/umisc.h new file mode 100644 index 0000000000..4e9dda7450 --- /dev/null +++ b/intl/icu/source/common/unicode/umisc.h @@ -0,0 +1,62 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1999-2006, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* file name: umisc.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 1999oct15 +* created by: Markus W. Scherer +*/ + +#ifndef UMISC_H +#define UMISC_H + +#include "unicode/utypes.h" + +/** + * \file + * \brief C API: Miscellaneous definitions + * + * This file contains miscellaneous definitions for the C APIs. + */ + +U_CDECL_BEGIN + +/** A struct representing a range of text containing a specific field + * @stable ICU 2.0 + */ +typedef struct UFieldPosition { + /** + * The field + * @stable ICU 2.0 + */ + int32_t field; + /** + * The start of the text range containing field + * @stable ICU 2.0 + */ + int32_t beginIndex; + /** + * The limit of the text range containing field + * @stable ICU 2.0 + */ + int32_t endIndex; +} UFieldPosition; + +#if !UCONFIG_NO_SERVICE +/** + * Opaque type returned by registerInstance, registerFactory and unregister for service registration. + * @stable ICU 2.6 + */ +typedef const void* URegistryKey; +#endif + +U_CDECL_END + +#endif diff --git a/intl/icu/source/common/unicode/umutablecptrie.h b/intl/icu/source/common/unicode/umutablecptrie.h new file mode 100644 index 0000000000..d60fd61819 --- /dev/null +++ b/intl/icu/source/common/unicode/umutablecptrie.h @@ -0,0 +1,240 @@ +// © 2017 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html + +// umutablecptrie.h (split out of ucptrie.h) +// created: 2018jan24 Markus W. Scherer + +#ifndef __UMUTABLECPTRIE_H__ +#define __UMUTABLECPTRIE_H__ + +#include "unicode/utypes.h" + +#include "unicode/ucpmap.h" +#include "unicode/ucptrie.h" +#include "unicode/utf8.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +U_CDECL_BEGIN + +/** + * \file + * \brief C API: This file defines a mutable Unicode code point trie. + * + * @see UCPTrie + * @see UMutableCPTrie + */ + +/** + * Mutable Unicode code point trie. + * Fast map from Unicode code points (U+0000..U+10FFFF) to 32-bit integer values. + * For details see https://icu.unicode.org/design/struct/utrie + * + * Setting values (especially ranges) and lookup is fast. + * The mutable trie is only somewhat space-efficient. + * It builds a compacted, immutable UCPTrie. + * + * This trie can be modified while iterating over its contents. + * For example, it is possible to merge its values with those from another + * set of ranges (e.g., another mutable or immutable trie): + * Iterate over those source ranges; for each of them iterate over this trie; + * add the source value into the value of each trie range. + * + * @see UCPTrie + * @see umutablecptrie_buildImmutable + * @stable ICU 63 + */ +typedef struct UMutableCPTrie UMutableCPTrie; + +/** + * Creates a mutable trie that initially maps each Unicode code point to the same value. + * It uses 32-bit data values until umutablecptrie_buildImmutable() is called. + * umutablecptrie_buildImmutable() takes a valueWidth parameter which + * determines the number of bits in the data value in the resulting UCPTrie. + * You must umutablecptrie_close() the trie once you are done using it. + * + * @param initialValue the initial value that is set for all code points + * @param errorValue the value for out-of-range code points and ill-formed UTF-8/16 + * @param pErrorCode an in/out ICU UErrorCode + * @return the trie + * @stable ICU 63 + */ +U_CAPI UMutableCPTrie * U_EXPORT2 +umutablecptrie_open(uint32_t initialValue, uint32_t errorValue, UErrorCode *pErrorCode); + +/** + * Clones a mutable trie. + * You must umutablecptrie_close() the clone once you are done using it. + * + * @param other the trie to clone + * @param pErrorCode an in/out ICU UErrorCode + * @return the trie clone + * @stable ICU 63 + */ +U_CAPI UMutableCPTrie * U_EXPORT2 +umutablecptrie_clone(const UMutableCPTrie *other, UErrorCode *pErrorCode); + +/** + * Closes a mutable trie and releases associated memory. + * + * @param trie the trie + * @stable ICU 63 + */ +U_CAPI void U_EXPORT2 +umutablecptrie_close(UMutableCPTrie *trie); + +/** + * Creates a mutable trie with the same contents as the UCPMap. + * You must umutablecptrie_close() the mutable trie once you are done using it. + * + * @param map the source map + * @param pErrorCode an in/out ICU UErrorCode + * @return the mutable trie + * @stable ICU 63 + */ +U_CAPI UMutableCPTrie * U_EXPORT2 +umutablecptrie_fromUCPMap(const UCPMap *map, UErrorCode *pErrorCode); + +/** + * Creates a mutable trie with the same contents as the immutable one. + * You must umutablecptrie_close() the mutable trie once you are done using it. + * + * @param trie the immutable trie + * @param pErrorCode an in/out ICU UErrorCode + * @return the mutable trie + * @stable ICU 63 + */ +U_CAPI UMutableCPTrie * U_EXPORT2 +umutablecptrie_fromUCPTrie(const UCPTrie *trie, UErrorCode *pErrorCode); + +/** + * Returns the value for a code point as stored in the trie. + * + * @param trie the trie + * @param c the code point + * @return the value + * @stable ICU 63 + */ +U_CAPI uint32_t U_EXPORT2 +umutablecptrie_get(const UMutableCPTrie *trie, UChar32 c); + +/** + * Returns the last code point such that all those from start to there have the same value. + * Can be used to efficiently iterate over all same-value ranges in a trie. + * (This is normally faster than iterating over code points and get()ting each value, + * but much slower than a data structure that stores ranges directly.) + * + * The trie can be modified between calls to this function. + * + * If the UCPMapValueFilter function pointer is not NULL, then + * the value to be delivered is passed through that function, and the return value is the end + * of the range where all values are modified to the same actual value. + * The value is unchanged if that function pointer is NULL. + * + * See the same-signature ucptrie_getRange() for a code sample. + * + * @param trie the trie + * @param start range start + * @param option defines whether surrogates are treated normally, + * or as having the surrogateValue; usually UCPMAP_RANGE_NORMAL + * @param surrogateValue value for surrogates; ignored if option==UCPMAP_RANGE_NORMAL + * @param filter a pointer to a function that may modify the trie data value, + * or NULL if the values from the trie are to be used unmodified + * @param context an opaque pointer that is passed on to the filter function + * @param pValue if not NULL, receives the value that every code point start..end has; + * may have been modified by filter(context, trie value) + * if that function pointer is not NULL + * @return the range end code point, or -1 if start is not a valid code point + * @stable ICU 63 + */ +U_CAPI UChar32 U_EXPORT2 +umutablecptrie_getRange(const UMutableCPTrie *trie, UChar32 start, + UCPMapRangeOption option, uint32_t surrogateValue, + UCPMapValueFilter *filter, const void *context, uint32_t *pValue); + +/** + * Sets a value for a code point. + * + * @param trie the trie + * @param c the code point + * @param value the value + * @param pErrorCode an in/out ICU UErrorCode + * @stable ICU 63 + */ +U_CAPI void U_EXPORT2 +umutablecptrie_set(UMutableCPTrie *trie, UChar32 c, uint32_t value, UErrorCode *pErrorCode); + +/** + * Sets a value for each code point [start..end]. + * Faster and more space-efficient than setting the value for each code point separately. + * + * @param trie the trie + * @param start the first code point to get the value + * @param end the last code point to get the value (inclusive) + * @param value the value + * @param pErrorCode an in/out ICU UErrorCode + * @stable ICU 63 + */ +U_CAPI void U_EXPORT2 +umutablecptrie_setRange(UMutableCPTrie *trie, + UChar32 start, UChar32 end, + uint32_t value, UErrorCode *pErrorCode); + +/** + * Compacts the data and builds an immutable UCPTrie according to the parameters. + * After this, the mutable trie will be empty. + * + * The mutable trie stores 32-bit values until buildImmutable() is called. + * If values shorter than 32 bits are to be stored in the immutable trie, + * then the upper bits are discarded. + * For example, when the mutable trie contains values 0x81, -0x7f, and 0xa581, + * and the value width is 8 bits, then each of these is stored as 0x81 + * and the immutable trie will return that as an unsigned value. + * (Some implementations may want to make productive temporary use of the upper bits + * until buildImmutable() discards them.) + * + * Not every possible set of mappings can be built into a UCPTrie, + * because of limitations resulting from speed and space optimizations. + * Every Unicode assigned character can be mapped to a unique value. + * Typical data yields data structures far smaller than the limitations. + * + * It is possible to construct extremely unusual mappings that exceed the data structure limits. + * In such a case this function will fail with a U_INDEX_OUTOFBOUNDS_ERROR. + * + * @param trie the trie trie + * @param type selects the trie type + * @param valueWidth selects the number of bits in a trie data value; if smaller than 32 bits, + * then the values stored in the trie will be truncated first + * @param pErrorCode an in/out ICU UErrorCode + * + * @see umutablecptrie_fromUCPTrie + * @stable ICU 63 + */ +U_CAPI UCPTrie * U_EXPORT2 +umutablecptrie_buildImmutable(UMutableCPTrie *trie, UCPTrieType type, UCPTrieValueWidth valueWidth, + UErrorCode *pErrorCode); + +U_CDECL_END + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUMutableCPTriePointer + * "Smart pointer" class, closes a UMutableCPTrie via umutablecptrie_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 63 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUMutableCPTriePointer, UMutableCPTrie, umutablecptrie_close); + +U_NAMESPACE_END + +#endif + +#endif diff --git a/intl/icu/source/common/unicode/unifilt.h b/intl/icu/source/common/unicode/unifilt.h new file mode 100644 index 0000000000..0fcaf4b789 --- /dev/null +++ b/intl/icu/source/common/unicode/unifilt.h @@ -0,0 +1,136 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1999-2010, International Business Machines Corporation and others. +* All Rights Reserved. +********************************************************************** +* Date Name Description +* 11/17/99 aliu Creation. +********************************************************************** +*/ +#ifndef UNIFILT_H +#define UNIFILT_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/unifunct.h" +#include "unicode/unimatch.h" + +/** + * \file + * \brief C++ API: Unicode Filter + */ + +U_NAMESPACE_BEGIN + +/** + * U_ETHER is used to represent character values for positions outside + * a range. For example, transliterator uses this to represent + * characters outside the range contextStart..contextLimit-1. This + * allows explicit matching by rules and UnicodeSets of text outside a + * defined range. + * @stable ICU 3.0 + */ +#define U_ETHER ((char16_t)0xFFFF) + +/** + * + * UnicodeFilter defines a protocol for selecting a + * subset of the full range (U+0000 to U+10FFFF) of Unicode characters. + * Currently, filters are used in conjunction with classes like + * {@link Transliterator} to only process selected characters through a + * transformation. + * + *

Note: UnicodeFilter currently stubs out two pure virtual methods + * of its base class, UnicodeMatcher. These methods are toPattern() + * and matchesIndexValue(). This is done so that filter classes that + * are not actually used as matchers -- specifically, those in the + * UnicodeFilterLogic component, and those in tests -- can continue to + * work without defining these methods. As long as a filter is not + * used in an RBT during real transliteration, these methods will not + * be called. However, this breaks the UnicodeMatcher base class + * protocol, and it is not a correct solution. + * + *

In the future we may revisit the UnicodeMatcher / UnicodeFilter + * hierarchy and either redesign it, or simply remove the stubs in + * UnicodeFilter and force subclasses to implement the full + * UnicodeMatcher protocol. + * + * @see UnicodeFilterLogic + * @stable ICU 2.0 + */ +class U_COMMON_API UnicodeFilter : public UnicodeFunctor, public UnicodeMatcher { + +public: + /** + * Destructor + * @stable ICU 2.0 + */ + virtual ~UnicodeFilter(); + + /** + * Clones this object polymorphically. + * The caller owns the result and should delete it when done. + * @return clone, or nullptr if an error occurred + * @stable ICU 2.4 + */ + virtual UnicodeFilter* clone() const override = 0; + + /** + * Returns true for characters that are in the selected + * subset. In other words, if a character is to be + * filtered, then contains() returns + * false. + * @stable ICU 2.0 + */ + virtual UBool contains(UChar32 c) const = 0; + + /** + * UnicodeFunctor API. Cast 'this' to a UnicodeMatcher* pointer + * and return the pointer. + * @stable ICU 2.4 + */ + virtual UnicodeMatcher* toMatcher() const override; + + /** + * Implement UnicodeMatcher API. + * @stable ICU 2.4 + */ + virtual UMatchDegree matches(const Replaceable& text, + int32_t& offset, + int32_t limit, + UBool incremental) override; + + /** + * UnicodeFunctor API. Nothing to do. + * @stable ICU 2.4 + */ + virtual void setData(const TransliterationRuleData*) override; + + /** + * ICU "poor man's RTTI", returns a UClassID for this class. + * + * @stable ICU 2.2 + */ + static UClassID U_EXPORT2 getStaticClassID(); + +protected: + + /* + * Since this class has pure virtual functions, + * a constructor can't be used. + * @stable ICU 2.0 + */ +/* UnicodeFilter();*/ +}; + +/*inline UnicodeFilter::UnicodeFilter() {}*/ + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/unifunct.h b/intl/icu/source/common/unicode/unifunct.h new file mode 100644 index 0000000000..8751302494 --- /dev/null +++ b/intl/icu/source/common/unicode/unifunct.h @@ -0,0 +1,132 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2002-2005, International Business Machines Corporation +* and others. All Rights Reserved. +********************************************************************** +* Date Name Description +* 01/14/2002 aliu Creation. +********************************************************************** +*/ +#ifndef UNIFUNCT_H +#define UNIFUNCT_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/uobject.h" + +/** + * \file + * \brief C++ API: Unicode Functor + */ + +U_NAMESPACE_BEGIN + +class UnicodeMatcher; +class UnicodeReplacer; +class TransliterationRuleData; + +/** + * UnicodeFunctor is an abstract base class for objects + * that perform match and/or replace operations on Unicode strings. + * @author Alan Liu + * @stable ICU 2.4 + */ +class U_COMMON_API UnicodeFunctor : public UObject { + +public: + + /** + * Destructor + * @stable ICU 2.4 + */ + virtual ~UnicodeFunctor(); + + /** + * Return a copy of this object. All UnicodeFunctor objects + * have to support cloning in order to allow classes using + * UnicodeFunctor to implement cloning. + * @stable ICU 2.4 + */ + virtual UnicodeFunctor* clone() const = 0; + + /** + * Cast 'this' to a UnicodeMatcher* pointer and return the + * pointer, or null if this is not a UnicodeMatcher*. Subclasses + * that mix in UnicodeMatcher as a base class must override this. + * This protocol is required because a pointer to a UnicodeFunctor + * cannot be cast to a pointer to a UnicodeMatcher, since + * UnicodeMatcher is a mixin that does not derive from + * UnicodeFunctor. + * @stable ICU 2.4 + */ + virtual UnicodeMatcher* toMatcher() const; + + /** + * Cast 'this' to a UnicodeReplacer* pointer and return the + * pointer, or null if this is not a UnicodeReplacer*. Subclasses + * that mix in UnicodeReplacer as a base class must override this. + * This protocol is required because a pointer to a UnicodeFunctor + * cannot be cast to a pointer to a UnicodeReplacer, since + * UnicodeReplacer is a mixin that does not derive from + * UnicodeFunctor. + * @stable ICU 2.4 + */ + virtual UnicodeReplacer* toReplacer() const; + + /** + * Return the class ID for this class. This is useful only for + * comparing to a return value from getDynamicClassID(). + * @return The class ID for all objects of this class. + * @stable ICU 2.0 + */ + static UClassID U_EXPORT2 getStaticClassID(void); + + /** + * Returns a unique class ID polymorphically. This method + * is to implement a simple version of RTTI, since not all C++ + * compilers support genuine RTTI. Polymorphic operator==() and + * clone() methods call this method. + * + *

Concrete subclasses of UnicodeFunctor should use the macro + * UOBJECT_DEFINE_RTTI_IMPLEMENTATION from uobject.h to + * provide definitions getStaticClassID and getDynamicClassID. + * + * @return The class ID for this object. All objects of a given + * class have the same class ID. Objects of other classes have + * different class IDs. + * @stable ICU 2.4 + */ + virtual UClassID getDynamicClassID(void) const override = 0; + + /** + * Set the data object associated with this functor. The data + * object provides context for functor-to-standin mapping. This + * method is required when assigning a functor to a different data + * object. This function MAY GO AWAY later if the architecture is + * changed to pass data object pointers through the API. + * @internal ICU 2.1 + */ + virtual void setData(const TransliterationRuleData*) = 0; + +protected: + + /** + * Since this class has pure virtual functions, + * a constructor can't be used. + * @stable ICU 2.0 + */ + /*UnicodeFunctor();*/ + +}; + +/*inline UnicodeFunctor::UnicodeFunctor() {}*/ + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/unimatch.h b/intl/icu/source/common/unicode/unimatch.h new file mode 100644 index 0000000000..302332f455 --- /dev/null +++ b/intl/icu/source/common/unicode/unimatch.h @@ -0,0 +1,168 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +* Copyright (C) 2001-2005, International Business Machines Corporation and others. All Rights Reserved. +********************************************************************** +* Date Name Description +* 07/18/01 aliu Creation. +********************************************************************** +*/ +#ifndef UNIMATCH_H +#define UNIMATCH_H + +#include "unicode/utypes.h" + +/** + * \file + * \brief C++ API: Unicode Matcher + */ + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +class Replaceable; +class UnicodeString; +class UnicodeSet; + +/** + * Constants returned by UnicodeMatcher::matches() + * indicating the degree of match. + * @stable ICU 2.4 + */ +enum UMatchDegree { + /** + * Constant returned by matches() indicating a + * mismatch between the text and this matcher. The text contains + * a character which does not match, or the text does not contain + * all desired characters for a non-incremental match. + * @stable ICU 2.4 + */ + U_MISMATCH, + + /** + * Constant returned by matches() indicating a + * partial match between the text and this matcher. This value is + * only returned for incremental match operations. All characters + * of the text match, but more characters are required for a + * complete match. Alternatively, for variable-length matchers, + * all characters of the text match, and if more characters were + * supplied at limit, they might also match. + * @stable ICU 2.4 + */ + U_PARTIAL_MATCH, + + /** + * Constant returned by matches() indicating a + * complete match between the text and this matcher. For an + * incremental variable-length match, this value is returned if + * the given text matches, and it is known that additional + * characters would not alter the extent of the match. + * @stable ICU 2.4 + */ + U_MATCH +}; + +/** + * UnicodeMatcher defines a protocol for objects that can + * match a range of characters in a Replaceable string. + * @stable ICU 2.4 + */ +class U_COMMON_API UnicodeMatcher /* not : public UObject because this is an interface/mixin class */ { + +public: + /** + * Destructor. + * @stable ICU 2.4 + */ + virtual ~UnicodeMatcher(); + + /** + * Return a UMatchDegree value indicating the degree of match for + * the given text at the given offset. Zero, one, or more + * characters may be matched. + * + * Matching in the forward direction is indicated by limit > + * offset. Characters from offset forwards to limit-1 will be + * considered for matching. + * + * Matching in the reverse direction is indicated by limit < + * offset. Characters from offset backwards to limit+1 will be + * considered for matching. + * + * If limit == offset then the only match possible is a zero + * character match (which subclasses may implement if desired). + * + * As a side effect, advance the offset parameter to the limit of + * the matched substring. In the forward direction, this will be + * the index of the last matched character plus one. In the + * reverse direction, this will be the index of the last matched + * character minus one. + * + *

Note: This method is not const because some classes may + * modify their state as the result of a match. + * + * @param text the text to be matched + * @param offset on input, the index into text at which to begin + * matching. On output, the limit of the matched text. The + * number of matched characters is the output value of offset + * minus the input value. Offset should always point to the + * HIGH SURROGATE (leading code unit) of a pair of surrogates, + * both on entry and upon return. + * @param limit the limit index of text to be matched. Greater + * than offset for a forward direction match, less than offset for + * a backward direction match. The last character to be + * considered for matching will be text.charAt(limit-1) in the + * forward direction or text.charAt(limit+1) in the backward + * direction. + * @param incremental if true, then assume further characters may + * be inserted at limit and check for partial matching. Otherwise + * assume the text as given is complete. + * @return a match degree value indicating a full match, a partial + * match, or a mismatch. If incremental is false then + * U_PARTIAL_MATCH should never be returned. + * @stable ICU 2.4 + */ + virtual UMatchDegree matches(const Replaceable& text, + int32_t& offset, + int32_t limit, + UBool incremental) = 0; + + /** + * Returns a string representation of this matcher. If the result of + * calling this function is passed to the appropriate parser, it + * will produce another matcher that is equal to this one. + * @param result the string to receive the pattern. Previous + * contents will be deleted. + * @param escapeUnprintable if true then convert unprintable + * character to their hex escape representations, \\uxxxx or + * \\Uxxxxxxxx. Unprintable characters are those other than + * U+000A, U+0020..U+007E. + * @stable ICU 2.4 + */ + virtual UnicodeString& toPattern(UnicodeString& result, + UBool escapeUnprintable = false) const = 0; + + /** + * Returns true if this matcher will match a character c, where c + * & 0xFF == v, at offset, in the forward direction (with limit > + * offset). This is used by RuleBasedTransliterator for + * indexing. + * @stable ICU 2.4 + */ + virtual UBool matchesIndexValue(uint8_t v) const = 0; + + /** + * Union the set of all characters that may be matched by this object + * into the given set. + * @param toUnionTo the set into which to union the source characters + * @stable ICU 2.4 + */ + virtual void addMatchSetTo(UnicodeSet& toUnionTo) const = 0; +}; + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/uniset.h b/intl/icu/source/common/unicode/uniset.h new file mode 100644 index 0000000000..84774d9f36 --- /dev/null +++ b/intl/icu/source/common/unicode/uniset.h @@ -0,0 +1,1793 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +*************************************************************************** +* Copyright (C) 1999-2016, International Business Machines Corporation +* and others. All Rights Reserved. +*************************************************************************** +* Date Name Description +* 10/20/99 alan Creation. +*************************************************************************** +*/ + +#ifndef UNICODESET_H +#define UNICODESET_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/ucpmap.h" +#include "unicode/unifilt.h" +#include "unicode/unistr.h" +#include "unicode/uset.h" + +/** + * \file + * \brief C++ API: Unicode Set + */ + +U_NAMESPACE_BEGIN + +// Forward Declarations. +class BMPSet; +class ParsePosition; +class RBBIRuleScanner; +class SymbolTable; +class UnicodeSetStringSpan; +class UVector; +class RuleCharacterIterator; + +/** + * A mutable set of Unicode characters and multicharacter strings. Objects of this class + * represent character classes used in regular expressions. + * A character specifies a subset of Unicode code points. Legal + * code points are U+0000 to U+10FFFF, inclusive. + * + *

The UnicodeSet class is not designed to be subclassed. + * + *

UnicodeSet supports two APIs. The first is the + * operand API that allows the caller to modify the value of + * a UnicodeSet object. It conforms to Java 2's + * java.util.Set interface, although + * UnicodeSet does not actually implement that + * interface. All methods of Set are supported, with the + * modification that they take a character range or single character + * instead of an Object, and they take a + * UnicodeSet instead of a Collection. The + * operand API may be thought of in terms of boolean logic: a boolean + * OR is implemented by add, a boolean AND is implemented + * by retain, a boolean XOR is implemented by + * complement taking an argument, and a boolean NOT is + * implemented by complement with no argument. In terms + * of traditional set theory function names, add is a + * union, retain is an intersection, remove + * is an asymmetric difference, and complement with no + * argument is a set complement with respect to the superset range + * MIN_VALUE-MAX_VALUE + * + *

The second API is the + * applyPattern()/toPattern() API from the + * java.text.Format-derived classes. Unlike the + * methods that add characters, add categories, and control the logic + * of the set, the method applyPattern() sets all + * attributes of a UnicodeSet at once, based on a + * string pattern. + * + *

Pattern syntax

+ * + * Patterns are accepted by the constructors and the + * applyPattern() methods and returned by the + * toPattern() method. These patterns follow a syntax + * similar to that employed by version 8 regular expression character + * classes. Here are some simple examples: + * + * \htmlonly
\endhtmlonly + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + *
[]No characters
[a]The character 'a'
[ae]The characters 'a' and 'e'
[a-e]The characters 'a' through 'e' inclusive, in Unicode code + * point order
[\\u4E01]The character U+4E01
[a{ab}{ac}]The character 'a' and the multicharacter strings "ab" and + * "ac"
[\\p{Lu}]All characters in the general category Uppercase Letter
+ * \htmlonly
\endhtmlonly + * + * Any character may be preceded by a backslash in order to remove any special + * meaning. White space characters, as defined by UCharacter.isWhitespace(), are + * ignored, unless they are escaped. + * + *

Property patterns specify a set of characters having a certain + * property as defined by the Unicode standard. Both the POSIX-like + * "[:Lu:]" and the Perl-like syntax "\\p{Lu}" are recognized. For a + * complete list of supported property patterns, see the User's Guide + * for UnicodeSet at + * + * https://unicode-org.github.io/icu/userguide/strings/unicodeset. + * Actual determination of property data is defined by the underlying + * Unicode database as implemented by UCharacter. + * + *

Patterns specify individual characters, ranges of characters, and + * Unicode property sets. When elements are concatenated, they + * specify their union. To complement a set, place a '^' immediately + * after the opening '['. Property patterns are inverted by modifying + * their delimiters; "[:^foo]" and "\\P{foo}". In any other location, + * '^' has no special meaning. + * + *

Since ICU 70, "[^...]", "[:^foo]", "\\P{foo}", and "[:binaryProperty=No:]" + * perform a “code point complement” (all code points minus the original set), + * removing all multicharacter strings, + * equivalent to .complement().removeAllStrings(). + * The complement() API function continues to perform a + * symmetric difference with all code points and thus retains all multicharacter strings. + * + *

Ranges are indicated by placing two a '-' between two + * characters, as in "a-z". This specifies the range of all + * characters from the left to the right, in Unicode order. If the + * left character is greater than or equal to the + * right character it is a syntax error. If a '-' occurs as the first + * character after the opening '[' or '[^', or if it occurs as the + * last character before the closing ']', then it is taken as a + * literal. Thus "[a\-b]", "[-ab]", and "[ab-]" all indicate the same + * set of three characters, 'a', 'b', and '-'. + * + *

Sets may be intersected using the '&' operator or the asymmetric + * set difference may be taken using the '-' operator, for example, + * "[[:L:]&[\\u0000-\\u0FFF]]" indicates the set of all Unicode letters + * with values less than 4096. Operators ('&' and '|') have equal + * precedence and bind left-to-right. Thus + * "[[:L:]-[a-z]-[\\u0100-\\u01FF]]" is equivalent to + * "[[[:L:]-[a-z]]-[\\u0100-\\u01FF]]". This only really matters for + * difference; intersection is commutative. + * + * + *
[a]The set containing 'a' + *
[a-z]The set containing 'a' + * through 'z' and all letters in between, in Unicode order + *
[^a-z]The set containing + * all characters but 'a' through 'z', + * that is, U+0000 through 'a'-1 and 'z'+1 through U+10FFFF + *
[[pat1][pat2]] + * The union of sets specified by pat1 and pat2 + *
[[pat1]&[pat2]] + * The intersection of sets specified by pat1 and pat2 + *
[[pat1]-[pat2]] + * The asymmetric difference of sets specified by pat1 and + * pat2 + *
[:Lu:] or \\p{Lu} + * The set of characters having the specified + * Unicode property; in + * this case, Unicode uppercase letters + *
[:^Lu:] or \\P{Lu} + * The set of characters not having the given + * Unicode property + *
+ * + *

Formal syntax

+ * + * \htmlonly
\endhtmlonly + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + *
pattern :=  ('[' '^'? item* ']') | + * property
item :=  char | (char '-' char) | pattern-expr
+ *
pattern-expr :=  pattern | pattern-expr pattern | + * pattern-expr op pattern
+ *
op :=  '&' | '-'
+ *
special :=  '[' | ']' | '-'
+ *
char :=  any character that is not special
+ * | ('\' any character)
+ * | ('\\u' hex hex hex hex)
+ *
hex :=  '0' | '1' | '2' | '3' | '4' | '5' | '6' | '7' | '8' | '9' |
+ *     'A' | 'B' | 'C' | 'D' | 'E' | 'F' | 'a' | 'b' | 'c' | 'd' | 'e' | 'f'
property :=  a Unicode property set pattern
+ *
+ * + * + * + * + *
Legend: + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + * + *
a := b  a may be replaced by b
a?zero or one instance of a
+ *
a*one or more instances of a
+ *
a | beither a or b
+ *
'a'the literal string between the quotes
+ *
+ * \htmlonly
\endhtmlonly + * + *

Note: + * - Most UnicodeSet methods do not take a UErrorCode parameter because + * there are usually very few opportunities for failure other than a shortage + * of memory, error codes in low-level C++ string methods would be inconvenient, + * and the error code as the last parameter (ICU convention) would prevent + * the use of default parameter values. + * Instead, such methods set the UnicodeSet into a "bogus" state + * (see isBogus()) if an error occurs. + * + * @author Alan Liu + * @stable ICU 2.0 + */ +class U_COMMON_API UnicodeSet final : public UnicodeFilter { +private: + /** + * Enough for sets with few ranges. + * For example, White_Space has 10 ranges, list length 21. + */ + static constexpr int32_t INITIAL_CAPACITY = 25; + // fFlags constant + static constexpr uint8_t kIsBogus = 1; // This set is bogus (i.e. not valid) + + UChar32* list = stackList; // MUST be terminated with HIGH + int32_t capacity = INITIAL_CAPACITY; // capacity of list + int32_t len = 1; // length of list used; 1 <= len <= capacity + uint8_t fFlags = 0; // Bit flag (see constants above) + + BMPSet *bmpSet = nullptr; // The set is frozen iff either bmpSet or stringSpan is not nullptr. + UChar32* buffer = nullptr; // internal buffer, may be nullptr + int32_t bufferCapacity = 0; // capacity of buffer + + /** + * The pattern representation of this set. This may not be the + * most economical pattern. It is the pattern supplied to + * applyPattern(), with variables substituted and whitespace + * removed. For sets constructed without applyPattern(), or + * modified using the non-pattern API, this string will be empty, + * indicating that toPattern() must generate a pattern + * representation from the inversion list. + */ + char16_t *pat = nullptr; + int32_t patLen = 0; + + UVector* strings = nullptr; // maintained in sorted order + UnicodeSetStringSpan *stringSpan = nullptr; + + /** + * Initial list array. + * Avoids some heap allocations, and list is never nullptr. + * Increases the object size a bit. + */ + UChar32 stackList[INITIAL_CAPACITY]; + +public: + /** + * Determine if this object contains a valid set. + * A bogus set has no value. It is different from an empty set. + * It can be used to indicate that no set value is available. + * + * @return true if the set is bogus/invalid, false otherwise + * @see setToBogus() + * @stable ICU 4.0 + */ + inline UBool isBogus(void) const; + + /** + * Make this UnicodeSet object invalid. + * The string will test true with isBogus(). + * + * A bogus set has no value. It is different from an empty set. + * It can be used to indicate that no set value is available. + * + * This utility function is used throughout the UnicodeSet + * implementation to indicate that a UnicodeSet operation failed, + * and may be used in other functions, + * especially but not exclusively when such functions do not + * take a UErrorCode for simplicity. + * + * @see isBogus() + * @stable ICU 4.0 + */ + void setToBogus(); + +public: + + enum { + /** + * Minimum value that can be stored in a UnicodeSet. + * @stable ICU 2.4 + */ + MIN_VALUE = 0, + + /** + * Maximum value that can be stored in a UnicodeSet. + * @stable ICU 2.4 + */ + MAX_VALUE = 0x10ffff + }; + + //---------------------------------------------------------------- + // Constructors &c + //---------------------------------------------------------------- + +public: + + /** + * Constructs an empty set. + * @stable ICU 2.0 + */ + UnicodeSet(); + + /** + * Constructs a set containing the given range. If end < + * start then an empty set is created. + * + * @param start first character, inclusive, of range + * @param end last character, inclusive, of range + * @stable ICU 2.4 + */ + UnicodeSet(UChar32 start, UChar32 end); + +#ifndef U_HIDE_INTERNAL_API + /** + * @internal + */ + enum ESerialization { + kSerialized /* result of serialize() */ + }; + + /** + * Constructs a set from the output of serialize(). + * + * @param buffer the 16 bit array + * @param bufferLen the original length returned from serialize() + * @param serialization the value 'kSerialized' + * @param status error code + * + * @internal + */ + UnicodeSet(const uint16_t buffer[], int32_t bufferLen, + ESerialization serialization, UErrorCode &status); +#endif /* U_HIDE_INTERNAL_API */ + + /** + * Constructs a set from the given pattern. See the class + * description for the syntax of the pattern language. + * @param pattern a string specifying what characters are in the set + * @param status returns U_ILLEGAL_ARGUMENT_ERROR if the pattern + * contains a syntax error. + * @stable ICU 2.0 + */ + UnicodeSet(const UnicodeString& pattern, + UErrorCode& status); + +#ifndef U_HIDE_INTERNAL_API + /** + * Constructs a set from the given pattern. See the class + * description for the syntax of the pattern language. + * @param pattern a string specifying what characters are in the set + * @param options bitmask for options to apply to the pattern. + * Valid options are USET_IGNORE_SPACE and + * at most one of USET_CASE_INSENSITIVE, USET_ADD_CASE_MAPPINGS, USET_SIMPLE_CASE_INSENSITIVE. + * These case options are mutually exclusive. + * @param symbols a symbol table mapping variable names to values + * and stand-in characters to UnicodeSets; may be nullptr + * @param status returns U_ILLEGAL_ARGUMENT_ERROR if the pattern + * contains a syntax error. + * @internal + */ + UnicodeSet(const UnicodeString& pattern, + uint32_t options, + const SymbolTable* symbols, + UErrorCode& status); +#endif /* U_HIDE_INTERNAL_API */ + + /** + * Constructs a set from the given pattern. See the class description + * for the syntax of the pattern language. + * @param pattern a string specifying what characters are in the set + * @param pos on input, the position in pattern at which to start parsing. + * On output, the position after the last character parsed. + * @param options bitmask for options to apply to the pattern. + * Valid options are USET_IGNORE_SPACE and + * at most one of USET_CASE_INSENSITIVE, USET_ADD_CASE_MAPPINGS, USET_SIMPLE_CASE_INSENSITIVE. + * These case options are mutually exclusive. + * @param symbols a symbol table mapping variable names to values + * and stand-in characters to UnicodeSets; may be nullptr + * @param status input-output error code + * @stable ICU 2.8 + */ + UnicodeSet(const UnicodeString& pattern, ParsePosition& pos, + uint32_t options, + const SymbolTable* symbols, + UErrorCode& status); + + /** + * Constructs a set that is identical to the given UnicodeSet. + * @stable ICU 2.0 + */ + UnicodeSet(const UnicodeSet& o); + + /** + * Destructs the set. + * @stable ICU 2.0 + */ + virtual ~UnicodeSet(); + + /** + * Assigns this object to be a copy of another. + * A frozen set will not be modified. + * @stable ICU 2.0 + */ + UnicodeSet& operator=(const UnicodeSet& o); + + /** + * Compares the specified object with this set for equality. Returns + * true if the two sets + * have the same size, and every member of the specified set is + * contained in this set (or equivalently, every member of this set is + * contained in the specified set). + * + * @param o set to be compared for equality with this set. + * @return true if the specified set is equal to this set. + * @stable ICU 2.0 + */ + virtual bool operator==(const UnicodeSet& o) const; + + /** + * Compares the specified object with this set for equality. Returns + * true if the specified set is not equal to this set. + * @stable ICU 2.0 + */ + inline bool operator!=(const UnicodeSet& o) const; + + /** + * Returns a copy of this object. All UnicodeFunctor objects have + * to support cloning in order to allow classes using + * UnicodeFunctors, such as Transliterator, to implement cloning. + * If this set is frozen, then the clone will be frozen as well. + * Use cloneAsThawed() for a mutable clone of a frozen set. + * @see cloneAsThawed + * @stable ICU 2.0 + */ + virtual UnicodeSet* clone() const override; + + /** + * Returns the hash code value for this set. + * + * @return the hash code value for this set. + * @see Object#hashCode() + * @stable ICU 2.0 + */ + virtual int32_t hashCode(void) const; + + /** + * Get a UnicodeSet pointer from a USet + * + * @param uset a USet (the ICU plain C type for UnicodeSet) + * @return the corresponding UnicodeSet pointer. + * + * @stable ICU 4.2 + */ + inline static UnicodeSet *fromUSet(USet *uset); + + /** + * Get a UnicodeSet pointer from a const USet + * + * @param uset a const USet (the ICU plain C type for UnicodeSet) + * @return the corresponding UnicodeSet pointer. + * + * @stable ICU 4.2 + */ + inline static const UnicodeSet *fromUSet(const USet *uset); + + /** + * Produce a USet * pointer for this UnicodeSet. + * USet is the plain C type for UnicodeSet + * + * @return a USet pointer for this UnicodeSet + * @stable ICU 4.2 + */ + inline USet *toUSet(); + + + /** + * Produce a const USet * pointer for this UnicodeSet. + * USet is the plain C type for UnicodeSet + * + * @return a const USet pointer for this UnicodeSet + * @stable ICU 4.2 + */ + inline const USet * toUSet() const; + + + //---------------------------------------------------------------- + // Freezable API + //---------------------------------------------------------------- + + /** + * Determines whether the set has been frozen (made immutable) or not. + * See the ICU4J Freezable interface for details. + * @return true/false for whether the set has been frozen + * @see freeze + * @see cloneAsThawed + * @stable ICU 3.8 + */ + inline UBool isFrozen() const; + + /** + * Freeze the set (make it immutable). + * Once frozen, it cannot be unfrozen and is therefore thread-safe + * until it is deleted. + * See the ICU4J Freezable interface for details. + * Freezing the set may also make some operations faster, for example + * contains() and span(). + * A frozen set will not be modified. (It remains frozen.) + * @return this set. + * @see isFrozen + * @see cloneAsThawed + * @stable ICU 3.8 + */ + UnicodeSet *freeze(); + + /** + * Clone the set and make the clone mutable. + * See the ICU4J Freezable interface for details. + * @return the mutable clone + * @see freeze + * @see isFrozen + * @stable ICU 3.8 + */ + UnicodeSet *cloneAsThawed() const; + + //---------------------------------------------------------------- + // Public API + //---------------------------------------------------------------- + + /** + * Make this object represent the range `start - end`. + * If `start > end` then this object is set to an empty range. + * A frozen set will not be modified. + * + * @param start first character in the set, inclusive + * @param end last character in the set, inclusive + * @stable ICU 2.4 + */ + UnicodeSet& set(UChar32 start, UChar32 end); + + /** + * Return true if the given position, in the given pattern, appears + * to be the start of a UnicodeSet pattern. + * @stable ICU 2.4 + */ + static UBool resemblesPattern(const UnicodeString& pattern, + int32_t pos); + + /** + * Modifies this set to represent the set specified by the given + * pattern, ignoring Unicode Pattern_White_Space characters. + * See the class description for the syntax of the pattern language. + * A frozen set will not be modified. + * @param pattern a string specifying what characters are in the set + * @param status returns U_ILLEGAL_ARGUMENT_ERROR if the pattern + * contains a syntax error. + * Empties the set passed before applying the pattern. + * @return a reference to this + * @stable ICU 2.0 + */ + UnicodeSet& applyPattern(const UnicodeString& pattern, + UErrorCode& status); + +#ifndef U_HIDE_INTERNAL_API + /** + * Modifies this set to represent the set specified by the given + * pattern, optionally ignoring Unicode Pattern_White_Space characters. + * See the class description for the syntax of the pattern language. + * A frozen set will not be modified. + * @param pattern a string specifying what characters are in the set + * @param options bitmask for options to apply to the pattern. + * Valid options are USET_IGNORE_SPACE and + * at most one of USET_CASE_INSENSITIVE, USET_ADD_CASE_MAPPINGS, USET_SIMPLE_CASE_INSENSITIVE. + * These case options are mutually exclusive. + * @param symbols a symbol table mapping variable names to + * values and stand-ins to UnicodeSets; may be nullptr + * @param status returns U_ILLEGAL_ARGUMENT_ERROR if the pattern + * contains a syntax error. + * Empties the set passed before applying the pattern. + * @return a reference to this + * @internal + */ + UnicodeSet& applyPattern(const UnicodeString& pattern, + uint32_t options, + const SymbolTable* symbols, + UErrorCode& status); +#endif /* U_HIDE_INTERNAL_API */ + + /** + * Parses the given pattern, starting at the given position. The + * character at pattern.charAt(pos.getIndex()) must be '[', or the + * parse fails. Parsing continues until the corresponding closing + * ']'. If a syntax error is encountered between the opening and + * closing brace, the parse fails. Upon return from a successful + * parse, the ParsePosition is updated to point to the character + * following the closing ']', and a StringBuffer containing a + * pairs list for the parsed pattern is returned. This method calls + * itself recursively to parse embedded subpatterns. + * Empties the set passed before applying the pattern. + * A frozen set will not be modified. + * + * @param pattern the string containing the pattern to be parsed. + * The portion of the string from pos.getIndex(), which must be a + * '[', to the corresponding closing ']', is parsed. + * @param pos upon entry, the position at which to being parsing. + * The character at pattern.charAt(pos.getIndex()) must be a '['. + * Upon return from a successful parse, pos.getIndex() is either + * the character after the closing ']' of the parsed pattern, or + * pattern.length() if the closing ']' is the last character of + * the pattern string. + * @param options bitmask for options to apply to the pattern. + * Valid options are USET_IGNORE_SPACE and + * at most one of USET_CASE_INSENSITIVE, USET_ADD_CASE_MAPPINGS, USET_SIMPLE_CASE_INSENSITIVE. + * These case options are mutually exclusive. + * @param symbols a symbol table mapping variable names to + * values and stand-ins to UnicodeSets; may be nullptr + * @param status returns U_ILLEGAL_ARGUMENT_ERROR if the pattern + * contains a syntax error. + * @return a reference to this + * @stable ICU 2.8 + */ + UnicodeSet& applyPattern(const UnicodeString& pattern, + ParsePosition& pos, + uint32_t options, + const SymbolTable* symbols, + UErrorCode& status); + + /** + * Returns a string representation of this set. If the result of + * calling this function is passed to a UnicodeSet constructor, it + * will produce another set that is equal to this one. + * A frozen set will not be modified. + * @param result the string to receive the rules. Previous + * contents will be deleted. + * @param escapeUnprintable if true then convert unprintable + * character to their hex escape representations, \\uxxxx or + * \\Uxxxxxxxx. Unprintable characters are those other than + * U+000A, U+0020..U+007E. + * @stable ICU 2.0 + */ + virtual UnicodeString& toPattern(UnicodeString& result, + UBool escapeUnprintable = false) const override; + + /** + * Modifies this set to contain those code points which have the given value + * for the given binary or enumerated property, as returned by + * u_getIntPropertyValue. Prior contents of this set are lost. + * A frozen set will not be modified. + * + * @param prop a property in the range UCHAR_BIN_START..UCHAR_BIN_LIMIT-1 + * or UCHAR_INT_START..UCHAR_INT_LIMIT-1 + * or UCHAR_MASK_START..UCHAR_MASK_LIMIT-1. + * + * @param value a value in the range u_getIntPropertyMinValue(prop).. + * u_getIntPropertyMaxValue(prop), with one exception. If prop is + * UCHAR_GENERAL_CATEGORY_MASK, then value should not be a UCharCategory, but + * rather a mask value produced by U_GET_GC_MASK(). This allows grouped + * categories such as [:L:] to be represented. + * + * @param ec error code input/output parameter + * + * @return a reference to this set + * + * @stable ICU 2.4 + */ + UnicodeSet& applyIntPropertyValue(UProperty prop, + int32_t value, + UErrorCode& ec); + + /** + * Modifies this set to contain those code points which have the + * given value for the given property. Prior contents of this + * set are lost. + * A frozen set will not be modified. + * + * @param prop a property alias, either short or long. The name is matched + * loosely. See PropertyAliases.txt for names and a description of loose + * matching. If the value string is empty, then this string is interpreted + * as either a General_Category value alias, a Script value alias, a binary + * property alias, or a special ID. Special IDs are matched loosely and + * correspond to the following sets: + * + * "ANY" = [\\u0000-\\U0010FFFF], + * "ASCII" = [\\u0000-\\u007F], + * "Assigned" = [:^Cn:]. + * + * @param value a value alias, either short or long. The name is matched + * loosely. See PropertyValueAliases.txt for names and a description of + * loose matching. In addition to aliases listed, numeric values and + * canonical combining classes may be expressed numerically, e.g., ("nv", + * "0.5") or ("ccc", "220"). The value string may also be empty. + * + * @param ec error code input/output parameter + * + * @return a reference to this set + * + * @stable ICU 2.4 + */ + UnicodeSet& applyPropertyAlias(const UnicodeString& prop, + const UnicodeString& value, + UErrorCode& ec); + + /** + * Returns the number of elements in this set (its cardinality). + * Note than the elements of a set may include both individual + * codepoints and strings. + * + * This is slower than getRangeCount() because + * it counts the code points of all ranges. + * + * @return the number of elements in this set (its cardinality). + * @stable ICU 2.0 + * @see getRangeCount + */ + virtual int32_t size(void) const; + + /** + * Returns true if this set contains no elements. + * + * @return true if this set contains no elements. + * @stable ICU 2.0 + */ + virtual UBool isEmpty(void) const; + + /** + * @return true if this set contains multi-character strings or the empty string. + * @stable ICU 70 + */ + UBool hasStrings() const; + + /** + * Returns true if this set contains the given character. + * This function works faster with a frozen set. + * @param c character to be checked for containment + * @return true if the test condition is met + * @stable ICU 2.0 + */ + virtual UBool contains(UChar32 c) const override; + + /** + * Returns true if this set contains every character + * of the given range. + * @param start first character, inclusive, of the range + * @param end last character, inclusive, of the range + * @return true if the test condition is met + * @stable ICU 2.0 + */ + virtual UBool contains(UChar32 start, UChar32 end) const; + + /** + * Returns true if this set contains the given + * multicharacter string. + * @param s string to be checked for containment + * @return true if this set contains the specified string + * @stable ICU 2.4 + */ + UBool contains(const UnicodeString& s) const; + + /** + * Returns true if this set contains all the characters and strings + * of the given set. + * @param c set to be checked for containment + * @return true if the test condition is met + * @stable ICU 2.4 + */ + virtual UBool containsAll(const UnicodeSet& c) const; + + /** + * Returns true if this set contains all the characters + * of the given string. + * @param s string containing characters to be checked for containment + * @return true if the test condition is met + * @stable ICU 2.4 + */ + UBool containsAll(const UnicodeString& s) const; + + /** + * Returns true if this set contains none of the characters + * of the given range. + * @param start first character, inclusive, of the range + * @param end last character, inclusive, of the range + * @return true if the test condition is met + * @stable ICU 2.4 + */ + UBool containsNone(UChar32 start, UChar32 end) const; + + /** + * Returns true if this set contains none of the characters and strings + * of the given set. + * @param c set to be checked for containment + * @return true if the test condition is met + * @stable ICU 2.4 + */ + UBool containsNone(const UnicodeSet& c) const; + + /** + * Returns true if this set contains none of the characters + * of the given string. + * @param s string containing characters to be checked for containment + * @return true if the test condition is met + * @stable ICU 2.4 + */ + UBool containsNone(const UnicodeString& s) const; + + /** + * Returns true if this set contains one or more of the characters + * in the given range. + * @param start first character, inclusive, of the range + * @param end last character, inclusive, of the range + * @return true if the condition is met + * @stable ICU 2.4 + */ + inline UBool containsSome(UChar32 start, UChar32 end) const; + + /** + * Returns true if this set contains one or more of the characters + * and strings of the given set. + * @param s The set to be checked for containment + * @return true if the condition is met + * @stable ICU 2.4 + */ + inline UBool containsSome(const UnicodeSet& s) const; + + /** + * Returns true if this set contains one or more of the characters + * of the given string. + * @param s string containing characters to be checked for containment + * @return true if the condition is met + * @stable ICU 2.4 + */ + inline UBool containsSome(const UnicodeString& s) const; + + /** + * Returns the length of the initial substring of the input string which + * consists only of characters and strings that are contained in this set + * (USET_SPAN_CONTAINED, USET_SPAN_SIMPLE), + * or only of characters and strings that are not contained + * in this set (USET_SPAN_NOT_CONTAINED). + * See USetSpanCondition for details. + * Similar to the strspn() C library function. + * Unpaired surrogates are treated according to contains() of their surrogate code points. + * This function works faster with a frozen set and with a non-negative string length argument. + * @param s start of the string + * @param length of the string; can be -1 for NUL-terminated + * @param spanCondition specifies the containment condition + * @return the length of the initial substring according to the spanCondition; + * 0 if the start of the string does not fit the spanCondition + * @stable ICU 3.8 + * @see USetSpanCondition + */ + int32_t span(const char16_t *s, int32_t length, USetSpanCondition spanCondition) const; + + /** + * Returns the end of the substring of the input string according to the USetSpanCondition. + * Same as start+span(s.getBuffer()+start, s.length()-start, spanCondition) + * after pinning start to 0<=start<=s.length(). + * @param s the string + * @param start the start index in the string for the span operation + * @param spanCondition specifies the containment condition + * @return the exclusive end of the substring according to the spanCondition; + * the substring s.tempSubStringBetween(start, end) fulfills the spanCondition + * @stable ICU 4.4 + * @see USetSpanCondition + */ + inline int32_t span(const UnicodeString &s, int32_t start, USetSpanCondition spanCondition) const; + + /** + * Returns the start of the trailing substring of the input string which + * consists only of characters and strings that are contained in this set + * (USET_SPAN_CONTAINED, USET_SPAN_SIMPLE), + * or only of characters and strings that are not contained + * in this set (USET_SPAN_NOT_CONTAINED). + * See USetSpanCondition for details. + * Unpaired surrogates are treated according to contains() of their surrogate code points. + * This function works faster with a frozen set and with a non-negative string length argument. + * @param s start of the string + * @param length of the string; can be -1 for NUL-terminated + * @param spanCondition specifies the containment condition + * @return the start of the trailing substring according to the spanCondition; + * the string length if the end of the string does not fit the spanCondition + * @stable ICU 3.8 + * @see USetSpanCondition + */ + int32_t spanBack(const char16_t *s, int32_t length, USetSpanCondition spanCondition) const; + + /** + * Returns the start of the substring of the input string according to the USetSpanCondition. + * Same as spanBack(s.getBuffer(), limit, spanCondition) + * after pinning limit to 0<=end<=s.length(). + * @param s the string + * @param limit the exclusive-end index in the string for the span operation + * (use s.length() or INT32_MAX for spanning back from the end of the string) + * @param spanCondition specifies the containment condition + * @return the start of the substring according to the spanCondition; + * the substring s.tempSubStringBetween(start, limit) fulfills the spanCondition + * @stable ICU 4.4 + * @see USetSpanCondition + */ + inline int32_t spanBack(const UnicodeString &s, int32_t limit, USetSpanCondition spanCondition) const; + + /** + * Returns the length of the initial substring of the input string which + * consists only of characters and strings that are contained in this set + * (USET_SPAN_CONTAINED, USET_SPAN_SIMPLE), + * or only of characters and strings that are not contained + * in this set (USET_SPAN_NOT_CONTAINED). + * See USetSpanCondition for details. + * Similar to the strspn() C library function. + * Malformed byte sequences are treated according to contains(0xfffd). + * This function works faster with a frozen set and with a non-negative string length argument. + * @param s start of the string (UTF-8) + * @param length of the string; can be -1 for NUL-terminated + * @param spanCondition specifies the containment condition + * @return the length of the initial substring according to the spanCondition; + * 0 if the start of the string does not fit the spanCondition + * @stable ICU 3.8 + * @see USetSpanCondition + */ + int32_t spanUTF8(const char *s, int32_t length, USetSpanCondition spanCondition) const; + + /** + * Returns the start of the trailing substring of the input string which + * consists only of characters and strings that are contained in this set + * (USET_SPAN_CONTAINED, USET_SPAN_SIMPLE), + * or only of characters and strings that are not contained + * in this set (USET_SPAN_NOT_CONTAINED). + * See USetSpanCondition for details. + * Malformed byte sequences are treated according to contains(0xfffd). + * This function works faster with a frozen set and with a non-negative string length argument. + * @param s start of the string (UTF-8) + * @param length of the string; can be -1 for NUL-terminated + * @param spanCondition specifies the containment condition + * @return the start of the trailing substring according to the spanCondition; + * the string length if the end of the string does not fit the spanCondition + * @stable ICU 3.8 + * @see USetSpanCondition + */ + int32_t spanBackUTF8(const char *s, int32_t length, USetSpanCondition spanCondition) const; + + /** + * Implement UnicodeMatcher::matches() + * @stable ICU 2.4 + */ + virtual UMatchDegree matches(const Replaceable& text, + int32_t& offset, + int32_t limit, + UBool incremental) override; + +private: + /** + * Returns the longest match for s in text at the given position. + * If limit > start then match forward from start+1 to limit + * matching all characters except s.charAt(0). If limit < start, + * go backward starting from start-1 matching all characters + * except s.charAt(s.length()-1). This method assumes that the + * first character, text.charAt(start), matches s, so it does not + * check it. + * @param text the text to match + * @param start the first character to match. In the forward + * direction, text.charAt(start) is matched against s.charAt(0). + * In the reverse direction, it is matched against + * s.charAt(s.length()-1). + * @param limit the limit offset for matching, either last+1 in + * the forward direction, or last-1 in the reverse direction, + * where last is the index of the last character to match. + * @param s + * @return If part of s matches up to the limit, return |limit - + * start|. If all of s matches before reaching the limit, return + * s.length(). If there is a mismatch between s and text, return + * 0 + */ + static int32_t matchRest(const Replaceable& text, + int32_t start, int32_t limit, + const UnicodeString& s); + + /** + * Returns the smallest value i such that c < list[i]. Caller + * must ensure that c is a legal value or this method will enter + * an infinite loop. This method performs a binary search. + * @param c a character in the range MIN_VALUE..MAX_VALUE + * inclusive + * @return the smallest integer i in the range 0..len-1, + * inclusive, such that c < list[i] + */ + int32_t findCodePoint(UChar32 c) const; + +public: + + /** + * Implementation of UnicodeMatcher API. Union the set of all + * characters that may be matched by this object into the given + * set. + * @param toUnionTo the set into which to union the source characters + * @stable ICU 2.4 + */ + virtual void addMatchSetTo(UnicodeSet& toUnionTo) const override; + + /** + * Returns the index of the given character within this set, where + * the set is ordered by ascending code point. If the character + * is not in this set, return -1. The inverse of this method is + * charAt(). + * @return an index from 0..size()-1, or -1 + * @stable ICU 2.4 + */ + int32_t indexOf(UChar32 c) const; + + /** + * Returns the character at the given index within this set, where + * the set is ordered by ascending code point. If the index is + * out of range for characters, returns (UChar32)-1. + * The inverse of this method is indexOf(). + * + * For iteration, this is slower than UnicodeSetIterator or + * getRangeCount()/getRangeStart()/getRangeEnd(), + * because for each call it skips linearly over index + * characters in the ranges. + * + * @param index an index from 0..size()-1 + * @return the character at the given index, or (UChar32)-1. + * @stable ICU 2.4 + */ + UChar32 charAt(int32_t index) const; + + /** + * Adds the specified range to this set if it is not already + * present. If this set already contains the specified range, + * the call leaves this set unchanged. If start > end + * then an empty range is added, leaving the set unchanged. + * This is equivalent to a boolean logic OR, or a set UNION. + * A frozen set will not be modified. + * + * @param start first character, inclusive, of range to be added + * to this set. + * @param end last character, inclusive, of range to be added + * to this set. + * @stable ICU 2.0 + */ + virtual UnicodeSet& add(UChar32 start, UChar32 end); + + /** + * Adds the specified character to this set if it is not already + * present. If this set already contains the specified character, + * the call leaves this set unchanged. + * A frozen set will not be modified. + * + * @param c the character (code point) + * @return this object, for chaining + * @stable ICU 2.0 + */ + UnicodeSet& add(UChar32 c); + + /** + * Adds the specified multicharacter to this set if it is not already + * present. If this set already contains the multicharacter, + * the call leaves this set unchanged. + * Thus "ch" => {"ch"} + * A frozen set will not be modified. + * + * @param s the source string + * @return this object, for chaining + * @stable ICU 2.4 + */ + UnicodeSet& add(const UnicodeString& s); + + private: + /** + * @return a code point IF the string consists of a single one. + * otherwise returns -1. + * @param s string to test + */ + static int32_t getSingleCP(const UnicodeString& s); + + void _add(const UnicodeString& s); + + public: + /** + * Adds each of the characters in this string to the set. Note: "ch" => {"c", "h"} + * If this set already contains any particular character, it has no effect on that character. + * A frozen set will not be modified. + * @param s the source string + * @return this object, for chaining + * @stable ICU 2.4 + */ + UnicodeSet& addAll(const UnicodeString& s); + + /** + * Retains EACH of the characters in this string. Note: "ch" == {"c", "h"} + * A frozen set will not be modified. + * @param s the source string + * @return this object, for chaining + * @stable ICU 2.4 + */ + UnicodeSet& retainAll(const UnicodeString& s); + + /** + * Complement EACH of the characters in this string. Note: "ch" == {"c", "h"} + * A frozen set will not be modified. + * @param s the source string + * @return this object, for chaining + * @stable ICU 2.4 + */ + UnicodeSet& complementAll(const UnicodeString& s); + + /** + * Remove EACH of the characters in this string. Note: "ch" == {"c", "h"} + * A frozen set will not be modified. + * @param s the source string + * @return this object, for chaining + * @stable ICU 2.4 + */ + UnicodeSet& removeAll(const UnicodeString& s); + + /** + * Makes a set from a multicharacter string. Thus "ch" => {"ch"} + * + * @param s the source string + * @return a newly created set containing the given string. + * The caller owns the return object and is responsible for deleting it. + * @stable ICU 2.4 + */ + static UnicodeSet* U_EXPORT2 createFrom(const UnicodeString& s); + + + /** + * Makes a set from each of the characters in the string. Thus "ch" => {"c", "h"} + * @param s the source string + * @return a newly created set containing the given characters + * The caller owns the return object and is responsible for deleting it. + * @stable ICU 2.4 + */ + static UnicodeSet* U_EXPORT2 createFromAll(const UnicodeString& s); + + /** + * Retain only the elements in this set that are contained in the + * specified range. If start > end then an empty range is + * retained, leaving the set empty. This is equivalent to + * a boolean logic AND, or a set INTERSECTION. + * A frozen set will not be modified. + * + * @param start first character, inclusive, of range + * @param end last character, inclusive, of range + * @stable ICU 2.0 + */ + virtual UnicodeSet& retain(UChar32 start, UChar32 end); + + + /** + * Retain the specified character from this set if it is present. + * A frozen set will not be modified. + * + * @param c the character (code point) + * @return this object, for chaining + * @stable ICU 2.0 + */ + UnicodeSet& retain(UChar32 c); + + /** + * Retains only the specified string from this set if it is present. + * Upon return this set will be empty if it did not contain s, or + * will only contain s if it did contain s. + * A frozen set will not be modified. + * + * @param s the source string + * @return this object, for chaining + * @stable ICU 69 + */ + UnicodeSet& retain(const UnicodeString &s); + + /** + * Removes the specified range from this set if it is present. + * The set will not contain the specified range once the call + * returns. If start > end then an empty range is + * removed, leaving the set unchanged. + * A frozen set will not be modified. + * + * @param start first character, inclusive, of range to be removed + * from this set. + * @param end last character, inclusive, of range to be removed + * from this set. + * @stable ICU 2.0 + */ + virtual UnicodeSet& remove(UChar32 start, UChar32 end); + + /** + * Removes the specified character from this set if it is present. + * The set will not contain the specified range once the call + * returns. + * A frozen set will not be modified. + * + * @param c the character (code point) + * @return this object, for chaining + * @stable ICU 2.0 + */ + UnicodeSet& remove(UChar32 c); + + /** + * Removes the specified string from this set if it is present. + * The set will not contain the specified character once the call + * returns. + * A frozen set will not be modified. + * @param s the source string + * @return this object, for chaining + * @stable ICU 2.4 + */ + UnicodeSet& remove(const UnicodeString& s); + + /** + * This is equivalent to + * complement(MIN_VALUE, MAX_VALUE). + * + * Note: This performs a symmetric difference with all code points + * and thus retains all multicharacter strings. + * In order to achieve a “code point complement” (all code points minus this set), + * the easiest is to .complement().removeAllStrings(). + * + * A frozen set will not be modified. + * @stable ICU 2.0 + */ + virtual UnicodeSet& complement(); + + /** + * Complements the specified range in this set. Any character in + * the range will be removed if it is in this set, or will be + * added if it is not in this set. If start > end + * then an empty range is complemented, leaving the set unchanged. + * This is equivalent to a boolean logic XOR. + * A frozen set will not be modified. + * + * @param start first character, inclusive, of range + * @param end last character, inclusive, of range + * @stable ICU 2.0 + */ + virtual UnicodeSet& complement(UChar32 start, UChar32 end); + + /** + * Complements the specified character in this set. The character + * will be removed if it is in this set, or will be added if it is + * not in this set. + * A frozen set will not be modified. + * + * @param c the character (code point) + * @return this object, for chaining + * @stable ICU 2.0 + */ + UnicodeSet& complement(UChar32 c); + + /** + * Complement the specified string in this set. + * The string will be removed if it is in this set, or will be added if it is not in this set. + * A frozen set will not be modified. + * + * @param s the string to complement + * @return this object, for chaining + * @stable ICU 2.4 + */ + UnicodeSet& complement(const UnicodeString& s); + + /** + * Adds all of the elements in the specified set to this set if + * they're not already present. This operation effectively + * modifies this set so that its value is the union of the two + * sets. The behavior of this operation is unspecified if the specified + * collection is modified while the operation is in progress. + * A frozen set will not be modified. + * + * @param c set whose elements are to be added to this set. + * @see #add(UChar32, UChar32) + * @stable ICU 2.0 + */ + virtual UnicodeSet& addAll(const UnicodeSet& c); + + /** + * Retains only the elements in this set that are contained in the + * specified set. In other words, removes from this set all of + * its elements that are not contained in the specified set. This + * operation effectively modifies this set so that its value is + * the intersection of the two sets. + * A frozen set will not be modified. + * + * @param c set that defines which elements this set will retain. + * @stable ICU 2.0 + */ + virtual UnicodeSet& retainAll(const UnicodeSet& c); + + /** + * Removes from this set all of its elements that are contained in the + * specified set. This operation effectively modifies this + * set so that its value is the asymmetric set difference of + * the two sets. + * A frozen set will not be modified. + * + * @param c set that defines which elements will be removed from + * this set. + * @stable ICU 2.0 + */ + virtual UnicodeSet& removeAll(const UnicodeSet& c); + + /** + * Complements in this set all elements contained in the specified + * set. Any character in the other set will be removed if it is + * in this set, or will be added if it is not in this set. + * A frozen set will not be modified. + * + * @param c set that defines which elements will be xor'ed from + * this set. + * @stable ICU 2.4 + */ + virtual UnicodeSet& complementAll(const UnicodeSet& c); + + /** + * Removes all of the elements from this set. This set will be + * empty after this call returns. + * A frozen set will not be modified. + * @stable ICU 2.0 + */ + virtual UnicodeSet& clear(void); + + /** + * Close this set over the given attribute. For the attribute + * USET_CASE_INSENSITIVE, the result is to modify this set so that: + * + * 1. For each character or string 'a' in this set, all strings or + * characters 'b' such that foldCase(a) == foldCase(b) are added + * to this set. + * + * 2. For each string 'e' in the resulting set, if e != + * foldCase(e), 'e' will be removed. + * + * Example: [aq\\u00DF{Bc}{bC}{Fi}] => [aAqQ\\u00DF\\uFB01{ss}{bc}{fi}] + * + * (Here foldCase(x) refers to the operation u_strFoldCase, and a + * == b denotes that the contents are the same, not pointer + * comparison.) + * + * A frozen set will not be modified. + * + * @param attribute bitmask for attributes to close over. + * Valid options: + * At most one of USET_CASE_INSENSITIVE, USET_ADD_CASE_MAPPINGS, USET_SIMPLE_CASE_INSENSITIVE. + * These case options are mutually exclusive. + * Unrelated options bits are ignored. + * @return a reference to this set. + * @stable ICU 4.2 + */ + UnicodeSet& closeOver(int32_t attribute); + + /** + * Remove all strings from this set. + * + * @return a reference to this set. + * @stable ICU 4.2 + */ + virtual UnicodeSet &removeAllStrings(); + + /** + * Iteration method that returns the number of ranges contained in + * this set. + * @see #getRangeStart + * @see #getRangeEnd + * @stable ICU 2.4 + */ + virtual int32_t getRangeCount(void) const; + + /** + * Iteration method that returns the first character in the + * specified range of this set. + * @see #getRangeCount + * @see #getRangeEnd + * @stable ICU 2.4 + */ + virtual UChar32 getRangeStart(int32_t index) const; + + /** + * Iteration method that returns the last character in the + * specified range of this set. + * @see #getRangeStart + * @see #getRangeEnd + * @stable ICU 2.4 + */ + virtual UChar32 getRangeEnd(int32_t index) const; + + /** + * Serializes this set into an array of 16-bit integers. Serialization + * (currently) only records the characters in the set; multicharacter + * strings are ignored. + * + * The array has following format (each line is one 16-bit + * integer): + * + * length = (n+2*m) | (m!=0?0x8000:0) + * bmpLength = n; present if m!=0 + * bmp[0] + * bmp[1] + * ... + * bmp[n-1] + * supp-high[0] + * supp-low[0] + * supp-high[1] + * supp-low[1] + * ... + * supp-high[m-1] + * supp-low[m-1] + * + * The array starts with a header. After the header are n bmp + * code points, then m supplementary code points. Either n or m + * or both may be zero. n+2*m is always <= 0x7FFF. + * + * If there are no supplementary characters (if m==0) then the + * header is one 16-bit integer, 'length', with value n. + * + * If there are supplementary characters (if m!=0) then the header + * is two 16-bit integers. The first, 'length', has value + * (n+2*m)|0x8000. The second, 'bmpLength', has value n. + * + * After the header the code points are stored in ascending order. + * Supplementary code points are stored as most significant 16 + * bits followed by least significant 16 bits. + * + * @param dest pointer to buffer of destCapacity 16-bit integers. + * May be nullptr only if destCapacity is zero. + * @param destCapacity size of dest, or zero. Must not be negative. + * @param ec error code. Will be set to U_INDEX_OUTOFBOUNDS_ERROR + * if n+2*m > 0x7FFF. Will be set to U_BUFFER_OVERFLOW_ERROR if + * n+2*m+(m!=0?2:1) > destCapacity. + * @return the total length of the serialized format, including + * the header, that is, n+2*m+(m!=0?2:1), or 0 on error other + * than U_BUFFER_OVERFLOW_ERROR. + * @stable ICU 2.4 + */ + int32_t serialize(uint16_t *dest, int32_t destCapacity, UErrorCode& ec) const; + + /** + * Reallocate this objects internal structures to take up the least + * possible space, without changing this object's value. + * A frozen set will not be modified. + * @stable ICU 2.4 + */ + virtual UnicodeSet& compact(); + + /** + * Return the class ID for this class. This is useful only for + * comparing to a return value from getDynamicClassID(). For example: + * 

+     * .      Base* polymorphic_pointer = createPolymorphicObject();
+     * .      if (polymorphic_pointer->getDynamicClassID() ==
+     * .          Derived::getStaticClassID()) ...
+     *
+ * @return The class ID for all objects of this class. + * @stable ICU 2.0 + */ + static UClassID U_EXPORT2 getStaticClassID(void); + + /** + * Implement UnicodeFunctor API. + * + * @return The class ID for this object. All objects of a given + * class have the same class ID. Objects of other classes have + * different class IDs. + * @stable ICU 2.4 + */ + virtual UClassID getDynamicClassID(void) const override; + +private: + + // Private API for the USet API + + friend class USetAccess; + + const UnicodeString* getString(int32_t index) const; + + //---------------------------------------------------------------- + // RuleBasedTransliterator support + //---------------------------------------------------------------- + +private: + + /** + * Returns true if this set contains any character whose low byte + * is the given value. This is used by RuleBasedTransliterator for + * indexing. + */ + virtual UBool matchesIndexValue(uint8_t v) const override; + +private: + friend class RBBIRuleScanner; + + //---------------------------------------------------------------- + // Implementation: Clone as thawed (see ICU4J Freezable) + //---------------------------------------------------------------- + + UnicodeSet(const UnicodeSet& o, UBool /* asThawed */); + UnicodeSet& copyFrom(const UnicodeSet& o, UBool asThawed); + + //---------------------------------------------------------------- + // Implementation: Pattern parsing + //---------------------------------------------------------------- + + void applyPatternIgnoreSpace(const UnicodeString& pattern, + ParsePosition& pos, + const SymbolTable* symbols, + UErrorCode& status); + + void applyPattern(RuleCharacterIterator& chars, + const SymbolTable* symbols, + UnicodeString& rebuiltPat, + uint32_t options, + UnicodeSet& (UnicodeSet::*caseClosure)(int32_t attribute), + int32_t depth, + UErrorCode& ec); + + void closeOverCaseInsensitive(bool simple); + void closeOverAddCaseMappings(); + + //---------------------------------------------------------------- + // Implementation: Utility methods + //---------------------------------------------------------------- + + static int32_t nextCapacity(int32_t minCapacity); + + bool ensureCapacity(int32_t newLen); + + bool ensureBufferCapacity(int32_t newLen); + + void swapBuffers(void); + + UBool allocateStrings(UErrorCode &status); + int32_t stringsSize() const; + UBool stringsContains(const UnicodeString &s) const; + + UnicodeString& _toPattern(UnicodeString& result, + UBool escapeUnprintable) const; + + UnicodeString& _generatePattern(UnicodeString& result, + UBool escapeUnprintable) const; + + static void _appendToPat(UnicodeString& buf, const UnicodeString& s, UBool escapeUnprintable); + + static void _appendToPat(UnicodeString& buf, UChar32 c, UBool escapeUnprintable); + + static void _appendToPat(UnicodeString &result, UChar32 start, UChar32 end, + UBool escapeUnprintable); + + //---------------------------------------------------------------- + // Implementation: Fundamental operators + //---------------------------------------------------------------- + + void exclusiveOr(const UChar32* other, int32_t otherLen, int8_t polarity); + + void add(const UChar32* other, int32_t otherLen, int8_t polarity); + + void retain(const UChar32* other, int32_t otherLen, int8_t polarity); + + /** + * Return true if the given position, in the given pattern, appears + * to be the start of a property set pattern [:foo:], \\p{foo}, or + * \\P{foo}, or \\N{name}. + */ + static UBool resemblesPropertyPattern(const UnicodeString& pattern, + int32_t pos); + + static UBool resemblesPropertyPattern(RuleCharacterIterator& chars, + int32_t iterOpts); + + /** + * Parse the given property pattern at the given parse position + * and set this UnicodeSet to the result. + * + * The original design document is out of date, but still useful. + * Ignore the property and value names: + * https://htmlpreview.github.io/?https://github.com/unicode-org/icu-docs/blob/main/design/unicodeset_properties.html + * + * Recognized syntax: + * + * [:foo:] [:^foo:] - white space not allowed within "[:" or ":]" + * \\p{foo} \\P{foo} - white space not allowed within "\\p" or "\\P" + * \\N{name} - white space not allowed within "\\N" + * + * Other than the above restrictions, Unicode Pattern_White_Space characters are ignored. + * Case is ignored except in "\\p" and "\\P" and "\\N". In 'name' leading + * and trailing space is deleted, and internal runs of whitespace + * are collapsed to a single space. + * + * We support binary properties, enumerated properties, and the + * following non-enumerated properties: + * + * Numeric_Value + * Name + * Unicode_1_Name + * + * @param pattern the pattern string + * @param ppos on entry, the position at which to begin parsing. + * This should be one of the locations marked '^': + * + * [:blah:] \\p{blah} \\P{blah} \\N{name} + * ^ % ^ % ^ % ^ % + * + * On return, the position after the last character parsed, that is, + * the locations marked '%'. If the parse fails, ppos is returned + * unchanged. + * @param ec status + * @return a reference to this. + */ + UnicodeSet& applyPropertyPattern(const UnicodeString& pattern, + ParsePosition& ppos, + UErrorCode &ec); + + void applyPropertyPattern(RuleCharacterIterator& chars, + UnicodeString& rebuiltPat, + UErrorCode& ec); + + /** + * A filter that returns true if the given code point should be + * included in the UnicodeSet being constructed. + */ + typedef UBool (*Filter)(UChar32 codePoint, void* context); + + /** + * Given a filter, set this UnicodeSet to the code points + * contained by that filter. The filter MUST be + * property-conformant. That is, if it returns value v for one + * code point, then it must return v for all affiliated code + * points, as defined by the inclusions list. See + * getInclusions(). + * src is a UPropertySource value. + */ + void applyFilter(Filter filter, + void* context, + const UnicodeSet* inclusions, + UErrorCode &status); + + /** + * Set the new pattern to cache. + */ + void setPattern(const UnicodeString& newPat) { + setPattern(newPat.getBuffer(), newPat.length()); + } + void setPattern(const char16_t *newPat, int32_t newPatLen); + /** + * Release existing cached pattern. + */ + void releasePattern(); + + friend class UnicodeSetIterator; +}; + + + +inline bool UnicodeSet::operator!=(const UnicodeSet& o) const { + return !operator==(o); +} + +inline UBool UnicodeSet::isFrozen() const { + return (UBool)(bmpSet!=nullptr || stringSpan!=nullptr); +} + +inline UBool UnicodeSet::containsSome(UChar32 start, UChar32 end) const { + return !containsNone(start, end); +} + +inline UBool UnicodeSet::containsSome(const UnicodeSet& s) const { + return !containsNone(s); +} + +inline UBool UnicodeSet::containsSome(const UnicodeString& s) const { + return !containsNone(s); +} + +inline UBool UnicodeSet::isBogus() const { + return (UBool)(fFlags & kIsBogus); +} + +inline UnicodeSet *UnicodeSet::fromUSet(USet *uset) { + return reinterpret_cast(uset); +} + +inline const UnicodeSet *UnicodeSet::fromUSet(const USet *uset) { + return reinterpret_cast(uset); +} + +inline USet *UnicodeSet::toUSet() { + return reinterpret_cast(this); +} + +inline const USet *UnicodeSet::toUSet() const { + return reinterpret_cast(this); +} + +inline int32_t UnicodeSet::span(const UnicodeString &s, int32_t start, USetSpanCondition spanCondition) const { + int32_t sLength=s.length(); + if(start<0) { + start=0; + } else if(start>sLength) { + start=sLength; + } + return start+span(s.getBuffer()+start, sLength-start, spanCondition); +} + +inline int32_t UnicodeSet::spanBack(const UnicodeString &s, int32_t limit, USetSpanCondition spanCondition) const { + int32_t sLength=s.length(); + if(limit<0) { + limit=0; + } else if(limit>sLength) { + limit=sLength; + } + return spanBack(s.getBuffer(), limit, spanCondition); +} + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/unistr.h b/intl/icu/source/common/unicode/unistr.h new file mode 100644 index 0000000000..4074e8d07b --- /dev/null +++ b/intl/icu/source/common/unicode/unistr.h @@ -0,0 +1,4784 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1998-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* +* File unistr.h +* +* Modification History: +* +* Date Name Description +* 09/25/98 stephen Creation. +* 11/11/98 stephen Changed per 11/9 code review. +* 04/20/99 stephen Overhauled per 4/16 code review. +* 11/18/99 aliu Made to inherit from Replaceable. Added method +* handleReplaceBetween(); other methods unchanged. +* 06/25/01 grhoten Remove dependency on iostream. +****************************************************************************** +*/ + +#ifndef UNISTR_H +#define UNISTR_H + +/** + * \file + * \brief C++ API: Unicode String + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include +#include "unicode/char16ptr.h" +#include "unicode/rep.h" +#include "unicode/std_string.h" +#include "unicode/stringpiece.h" +#include "unicode/bytestream.h" + +struct UConverter; // unicode/ucnv.h + +#ifndef USTRING_H +/** + * \ingroup ustring_ustrlen + * @param s Pointer to sequence of UChars. + * @return Length of sequence. + */ +U_CAPI int32_t U_EXPORT2 u_strlen(const UChar *s); +#endif + +U_NAMESPACE_BEGIN + +#if !UCONFIG_NO_BREAK_ITERATION +class BreakIterator; // unicode/brkiter.h +#endif +class Edits; + +U_NAMESPACE_END + +// Not #ifndef U_HIDE_INTERNAL_API because UnicodeString needs the UStringCaseMapper. +/** + * Internal string case mapping function type. + * All error checking must be done. + * src and dest must not overlap. + * @internal + */ +typedef int32_t U_CALLCONV +UStringCaseMapper(int32_t caseLocale, uint32_t options, +#if !UCONFIG_NO_BREAK_ITERATION + icu::BreakIterator *iter, +#endif + char16_t *dest, int32_t destCapacity, + const char16_t *src, int32_t srcLength, + icu::Edits *edits, + UErrorCode &errorCode); + +U_NAMESPACE_BEGIN + +class Locale; // unicode/locid.h +class StringCharacterIterator; +class UnicodeStringAppendable; // unicode/appendable.h + +/* The include has been moved to unicode/ustream.h */ + +/** + * Constant to be used in the UnicodeString(char *, int32_t, EInvariant) constructor + * which constructs a Unicode string from an invariant-character char * string. + * About invariant characters see utypes.h. + * This constructor has no runtime dependency on conversion code and is + * therefore recommended over ones taking a charset name string + * (where the empty string "" indicates invariant-character conversion). + * + * @stable ICU 3.2 + */ +#define US_INV icu::UnicodeString::kInvariant + +/** + * Unicode String literals in C++. + * + * Note: these macros are not recommended for new code. + * Prior to the availability of C++11 and u"unicode string literals", + * these macros were provided for portability and efficiency when + * initializing UnicodeStrings from literals. + * + * They work only for strings that contain "invariant characters", i.e., + * only latin letters, digits, and some punctuation. + * See utypes.h for details. + * + * The string parameter must be a C string literal. + * The length of the string, not including the terminating + * `NUL`, must be specified as a constant. + * @stable ICU 2.0 + */ +#if !U_CHAR16_IS_TYPEDEF +# define UNICODE_STRING(cs, _length) icu::UnicodeString(true, u ## cs, _length) +#else +# define UNICODE_STRING(cs, _length) icu::UnicodeString(true, (const char16_t*)u ## cs, _length) +#endif + +/** + * Unicode String literals in C++. + * Dependent on the platform properties, different UnicodeString + * constructors should be used to create a UnicodeString object from + * a string literal. + * The macros are defined for improved performance. + * They work only for strings that contain "invariant characters", i.e., + * only latin letters, digits, and some punctuation. + * See utypes.h for details. + * + * The string parameter must be a C string literal. + * @stable ICU 2.0 + */ +#define UNICODE_STRING_SIMPLE(cs) UNICODE_STRING(cs, -1) + +/** + * \def UNISTR_FROM_CHAR_EXPLICIT + * This can be defined to be empty or "explicit". + * If explicit, then the UnicodeString(char16_t) and UnicodeString(UChar32) + * constructors are marked as explicit, preventing their inadvertent use. + * @stable ICU 49 + */ +#ifndef UNISTR_FROM_CHAR_EXPLICIT +# if defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION) + // Auto-"explicit" in ICU library code. +# define UNISTR_FROM_CHAR_EXPLICIT explicit +# else + // Empty by default for source code compatibility. +# define UNISTR_FROM_CHAR_EXPLICIT +# endif +#endif + +/** + * \def UNISTR_FROM_STRING_EXPLICIT + * This can be defined to be empty or "explicit". + * If explicit, then the UnicodeString(const char *) and UnicodeString(const char16_t *) + * constructors are marked as explicit, preventing their inadvertent use. + * + * In particular, this helps prevent accidentally depending on ICU conversion code + * by passing a string literal into an API with a const UnicodeString & parameter. + * @stable ICU 49 + */ +#ifndef UNISTR_FROM_STRING_EXPLICIT +# if defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION) + // Auto-"explicit" in ICU library code. +# define UNISTR_FROM_STRING_EXPLICIT explicit +# else + // Empty by default for source code compatibility. +# define UNISTR_FROM_STRING_EXPLICIT +# endif +#endif + +/** + * \def UNISTR_OBJECT_SIZE + * Desired sizeof(UnicodeString) in bytes. + * It should be a multiple of sizeof(pointer) to avoid unusable space for padding. + * The object size may want to be a multiple of 16 bytes, + * which is a common granularity for heap allocation. + * + * Any space inside the object beyond sizeof(vtable pointer) + 2 + * is available for storing short strings inside the object. + * The bigger the object, the longer a string that can be stored inside the object, + * without additional heap allocation. + * + * Depending on a platform's pointer size, pointer alignment requirements, + * and struct padding, the compiler will usually round up sizeof(UnicodeString) + * to 4 * sizeof(pointer) (or 3 * sizeof(pointer) for P128 data models), + * to hold the fields for heap-allocated strings. + * Such a minimum size also ensures that the object is easily large enough + * to hold at least 2 char16_ts, for one supplementary code point (U16_MAX_LENGTH). + * + * sizeof(UnicodeString) >= 48 should work for all known platforms. + * + * For example, on a 64-bit machine where sizeof(vtable pointer) is 8, + * sizeof(UnicodeString) = 64 would leave space for + * (64 - sizeof(vtable pointer) - 2) / U_SIZEOF_UCHAR = (64 - 8 - 2) / 2 = 27 + * char16_ts stored inside the object. + * + * The minimum object size on a 64-bit machine would be + * 4 * sizeof(pointer) = 4 * 8 = 32 bytes, + * and the internal buffer would hold up to 11 char16_ts in that case. + * + * @see U16_MAX_LENGTH + * @stable ICU 56 + */ +#ifndef UNISTR_OBJECT_SIZE +# define UNISTR_OBJECT_SIZE 64 +#endif + +/** + * UnicodeString is a string class that stores Unicode characters directly and provides + * similar functionality as the Java String and StringBuffer/StringBuilder classes. + * It is a concrete implementation of the abstract class Replaceable (for transliteration). + * + * The UnicodeString equivalent of std::string’s clear() is remove(). + * + * A UnicodeString may "alias" an external array of characters + * (that is, point to it, rather than own the array) + * whose lifetime must then at least match the lifetime of the aliasing object. + * This aliasing may be preserved when returning a UnicodeString by value, + * depending on the compiler and the function implementation, + * via Return Value Optimization (RVO) or the move assignment operator. + * (However, the copy assignment operator does not preserve aliasing.) + * For details see the description of storage models at the end of the class API docs + * and in the User Guide chapter linked from there. + * + * The UnicodeString class is not suitable for subclassing. + * + * For an overview of Unicode strings in C and C++ see the + * [User Guide Strings chapter](https://unicode-org.github.io/icu/userguide/strings#strings-in-cc). + * + * In ICU, a Unicode string consists of 16-bit Unicode *code units*. + * A Unicode character may be stored with either one code unit + * (the most common case) or with a matched pair of special code units + * ("surrogates"). The data type for code units is char16_t. + * For single-character handling, a Unicode character code *point* is a value + * in the range 0..0x10ffff. ICU uses the UChar32 type for code points. + * + * Indexes and offsets into and lengths of strings always count code units, not code points. + * This is the same as with multi-byte char* strings in traditional string handling. + * Operations on partial strings typically do not test for code point boundaries. + * If necessary, the user needs to take care of such boundaries by testing for the code unit + * values or by using functions like + * UnicodeString::getChar32Start() and UnicodeString::getChar32Limit() + * (or, in C, the equivalent macros U16_SET_CP_START() and U16_SET_CP_LIMIT(), see utf.h). + * + * UnicodeString methods are more lenient with regard to input parameter values + * than other ICU APIs. In particular: + * - If indexes are out of bounds for a UnicodeString object + * (< 0 or > length()) then they are "pinned" to the nearest boundary. + * - If the buffer passed to an insert/append/replace operation is owned by the + * target object, e.g., calling str.append(str), an extra copy may take place + * to ensure safety. + * - If primitive string pointer values (e.g., const char16_t * or char *) + * for input strings are nullptr, then those input string parameters are treated + * as if they pointed to an empty string. + * However, this is *not* the case for char * parameters for charset names + * or other IDs. + * - Most UnicodeString methods do not take a UErrorCode parameter because + * there are usually very few opportunities for failure other than a shortage + * of memory, error codes in low-level C++ string methods would be inconvenient, + * and the error code as the last parameter (ICU convention) would prevent + * the use of default parameter values. + * Instead, such methods set the UnicodeString into a "bogus" state + * (see isBogus()) if an error occurs. + * + * In string comparisons, two UnicodeString objects that are both "bogus" + * compare equal (to be transitive and prevent endless loops in sorting), + * and a "bogus" string compares less than any non-"bogus" one. + * + * Const UnicodeString methods are thread-safe. Multiple threads can use + * const methods on the same UnicodeString object simultaneously, + * but non-const methods must not be called concurrently (in multiple threads) + * with any other (const or non-const) methods. + * + * Similarly, const UnicodeString & parameters are thread-safe. + * One object may be passed in as such a parameter concurrently in multiple threads. + * This includes the const UnicodeString & parameters for + * copy construction, assignment, and cloning. + * + * UnicodeString uses several storage methods. + * String contents can be stored inside the UnicodeString object itself, + * in an allocated and shared buffer, or in an outside buffer that is "aliased". + * Most of this is done transparently, but careful aliasing in particular provides + * significant performance improvements. + * Also, the internal buffer is accessible via special functions. + * For details see the + * [User Guide Strings chapter](https://unicode-org.github.io/icu/userguide/strings#maximizing-performance-with-the-unicodestring-storage-model). + * + * @see utf.h + * @see CharacterIterator + * @stable ICU 2.0 + */ +class U_COMMON_API UnicodeString : public Replaceable +{ +public: + + /** + * Constant to be used in the UnicodeString(char *, int32_t, EInvariant) constructor + * which constructs a Unicode string from an invariant-character char * string. + * Use the macro US_INV instead of the full qualification for this value. + * + * @see US_INV + * @stable ICU 3.2 + */ + enum EInvariant { + /** + * @see EInvariant + * @stable ICU 3.2 + */ + kInvariant + }; + + //======================================== + // Read-only operations + //======================================== + + /* Comparison - bitwise only - for international comparison use collation */ + + /** + * Equality operator. Performs only bitwise comparison. + * @param text The UnicodeString to compare to this one. + * @return true if `text` contains the same characters as this one, + * false otherwise. + * @stable ICU 2.0 + */ + inline bool operator== (const UnicodeString& text) const; + + /** + * Inequality operator. Performs only bitwise comparison. + * @param text The UnicodeString to compare to this one. + * @return false if `text` contains the same characters as this one, + * true otherwise. + * @stable ICU 2.0 + */ + inline bool operator!= (const UnicodeString& text) const; + + /** + * Greater than operator. Performs only bitwise comparison. + * @param text The UnicodeString to compare to this one. + * @return true if the characters in this are bitwise + * greater than the characters in `text`, false otherwise + * @stable ICU 2.0 + */ + inline UBool operator> (const UnicodeString& text) const; + + /** + * Less than operator. Performs only bitwise comparison. + * @param text The UnicodeString to compare to this one. + * @return true if the characters in this are bitwise + * less than the characters in `text`, false otherwise + * @stable ICU 2.0 + */ + inline UBool operator< (const UnicodeString& text) const; + + /** + * Greater than or equal operator. Performs only bitwise comparison. + * @param text The UnicodeString to compare to this one. + * @return true if the characters in this are bitwise + * greater than or equal to the characters in `text`, false otherwise + * @stable ICU 2.0 + */ + inline UBool operator>= (const UnicodeString& text) const; + + /** + * Less than or equal operator. Performs only bitwise comparison. + * @param text The UnicodeString to compare to this one. + * @return true if the characters in this are bitwise + * less than or equal to the characters in `text`, false otherwise + * @stable ICU 2.0 + */ + inline UBool operator<= (const UnicodeString& text) const; + + /** + * Compare the characters bitwise in this UnicodeString to + * the characters in `text`. + * @param text The UnicodeString to compare to this one. + * @return The result of bitwise character comparison: 0 if this + * contains the same characters as `text`, -1 if the characters in + * this are bitwise less than the characters in `text`, +1 if the + * characters in this are bitwise greater than the characters + * in `text`. + * @stable ICU 2.0 + */ + inline int8_t compare(const UnicodeString& text) const; + + /** + * Compare the characters bitwise in the range + * [`start`, `start + length`) with the characters + * in the **entire string** `text`. + * (The parameters "start" and "length" are not applied to the other text "text".) + * @param start the offset at which the compare operation begins + * @param length the number of characters of text to compare. + * @param text the other text to be compared against this string. + * @return The result of bitwise character comparison: 0 if this + * contains the same characters as `text`, -1 if the characters in + * this are bitwise less than the characters in `text`, +1 if the + * characters in this are bitwise greater than the characters + * in `text`. + * @stable ICU 2.0 + */ + inline int8_t compare(int32_t start, + int32_t length, + const UnicodeString& text) const; + + /** + * Compare the characters bitwise in the range + * [`start`, `start + length`) with the characters + * in `srcText` in the range + * [`srcStart`, `srcStart + srcLength`). + * @param start the offset at which the compare operation begins + * @param length the number of characters in this to compare. + * @param srcText the text to be compared + * @param srcStart the offset into `srcText` to start comparison + * @param srcLength the number of characters in `src` to compare + * @return The result of bitwise character comparison: 0 if this + * contains the same characters as `srcText`, -1 if the characters in + * this are bitwise less than the characters in `srcText`, +1 if the + * characters in this are bitwise greater than the characters + * in `srcText`. + * @stable ICU 2.0 + */ + inline int8_t compare(int32_t start, + int32_t length, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const; + + /** + * Compare the characters bitwise in this UnicodeString with the first + * `srcLength` characters in `srcChars`. + * @param srcChars The characters to compare to this UnicodeString. + * @param srcLength the number of characters in `srcChars` to compare + * @return The result of bitwise character comparison: 0 if this + * contains the same characters as `srcChars`, -1 if the characters in + * this are bitwise less than the characters in `srcChars`, +1 if the + * characters in this are bitwise greater than the characters + * in `srcChars`. + * @stable ICU 2.0 + */ + inline int8_t compare(ConstChar16Ptr srcChars, + int32_t srcLength) const; + + /** + * Compare the characters bitwise in the range + * [`start`, `start + length`) with the first + * `length` characters in `srcChars` + * @param start the offset at which the compare operation begins + * @param length the number of characters to compare. + * @param srcChars the characters to be compared + * @return The result of bitwise character comparison: 0 if this + * contains the same characters as `srcChars`, -1 if the characters in + * this are bitwise less than the characters in `srcChars`, +1 if the + * characters in this are bitwise greater than the characters + * in `srcChars`. + * @stable ICU 2.0 + */ + inline int8_t compare(int32_t start, + int32_t length, + const char16_t *srcChars) const; + + /** + * Compare the characters bitwise in the range + * [`start`, `start + length`) with the characters + * in `srcChars` in the range + * [`srcStart`, `srcStart + srcLength`). + * @param start the offset at which the compare operation begins + * @param length the number of characters in this to compare + * @param srcChars the characters to be compared + * @param srcStart the offset into `srcChars` to start comparison + * @param srcLength the number of characters in `srcChars` to compare + * @return The result of bitwise character comparison: 0 if this + * contains the same characters as `srcChars`, -1 if the characters in + * this are bitwise less than the characters in `srcChars`, +1 if the + * characters in this are bitwise greater than the characters + * in `srcChars`. + * @stable ICU 2.0 + */ + inline int8_t compare(int32_t start, + int32_t length, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength) const; + + /** + * Compare the characters bitwise in the range + * [`start`, `limit`) with the characters + * in `srcText` in the range + * [`srcStart`, `srcLimit`). + * @param start the offset at which the compare operation begins + * @param limit the offset immediately following the compare operation + * @param srcText the text to be compared + * @param srcStart the offset into `srcText` to start comparison + * @param srcLimit the offset into `srcText` to limit comparison + * @return The result of bitwise character comparison: 0 if this + * contains the same characters as `srcText`, -1 if the characters in + * this are bitwise less than the characters in `srcText`, +1 if the + * characters in this are bitwise greater than the characters + * in `srcText`. + * @stable ICU 2.0 + */ + inline int8_t compareBetween(int32_t start, + int32_t limit, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLimit) const; + + /** + * Compare two Unicode strings in code point order. + * The result may be different from the results of compare(), operator<, etc. + * if supplementary characters are present: + * + * In UTF-16, supplementary characters (with code points U+10000 and above) are + * stored with pairs of surrogate code units. These have values from 0xd800 to 0xdfff, + * which means that they compare as less than some other BMP characters like U+feff. + * This function compares Unicode strings in code point order. + * If either of the UTF-16 strings is malformed (i.e., it contains unpaired surrogates), then the result is not defined. + * + * @param text Another string to compare this one to. + * @return a negative/zero/positive integer corresponding to whether + * this string is less than/equal to/greater than the second one + * in code point order + * @stable ICU 2.0 + */ + inline int8_t compareCodePointOrder(const UnicodeString& text) const; + + /** + * Compare two Unicode strings in code point order. + * The result may be different from the results of compare(), operator<, etc. + * if supplementary characters are present: + * + * In UTF-16, supplementary characters (with code points U+10000 and above) are + * stored with pairs of surrogate code units. These have values from 0xd800 to 0xdfff, + * which means that they compare as less than some other BMP characters like U+feff. + * This function compares Unicode strings in code point order. + * If either of the UTF-16 strings is malformed (i.e., it contains unpaired surrogates), then the result is not defined. + * + * @param start The start offset in this string at which the compare operation begins. + * @param length The number of code units from this string to compare. + * @param srcText Another string to compare this one to. + * @return a negative/zero/positive integer corresponding to whether + * this string is less than/equal to/greater than the second one + * in code point order + * @stable ICU 2.0 + */ + inline int8_t compareCodePointOrder(int32_t start, + int32_t length, + const UnicodeString& srcText) const; + + /** + * Compare two Unicode strings in code point order. + * The result may be different from the results of compare(), operator<, etc. + * if supplementary characters are present: + * + * In UTF-16, supplementary characters (with code points U+10000 and above) are + * stored with pairs of surrogate code units. These have values from 0xd800 to 0xdfff, + * which means that they compare as less than some other BMP characters like U+feff. + * This function compares Unicode strings in code point order. + * If either of the UTF-16 strings is malformed (i.e., it contains unpaired surrogates), then the result is not defined. + * + * @param start The start offset in this string at which the compare operation begins. + * @param length The number of code units from this string to compare. + * @param srcText Another string to compare this one to. + * @param srcStart The start offset in that string at which the compare operation begins. + * @param srcLength The number of code units from that string to compare. + * @return a negative/zero/positive integer corresponding to whether + * this string is less than/equal to/greater than the second one + * in code point order + * @stable ICU 2.0 + */ + inline int8_t compareCodePointOrder(int32_t start, + int32_t length, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const; + + /** + * Compare two Unicode strings in code point order. + * The result may be different from the results of compare(), operator<, etc. + * if supplementary characters are present: + * + * In UTF-16, supplementary characters (with code points U+10000 and above) are + * stored with pairs of surrogate code units. These have values from 0xd800 to 0xdfff, + * which means that they compare as less than some other BMP characters like U+feff. + * This function compares Unicode strings in code point order. + * If either of the UTF-16 strings is malformed (i.e., it contains unpaired surrogates), then the result is not defined. + * + * @param srcChars A pointer to another string to compare this one to. + * @param srcLength The number of code units from that string to compare. + * @return a negative/zero/positive integer corresponding to whether + * this string is less than/equal to/greater than the second one + * in code point order + * @stable ICU 2.0 + */ + inline int8_t compareCodePointOrder(ConstChar16Ptr srcChars, + int32_t srcLength) const; + + /** + * Compare two Unicode strings in code point order. + * The result may be different from the results of compare(), operator<, etc. + * if supplementary characters are present: + * + * In UTF-16, supplementary characters (with code points U+10000 and above) are + * stored with pairs of surrogate code units. These have values from 0xd800 to 0xdfff, + * which means that they compare as less than some other BMP characters like U+feff. + * This function compares Unicode strings in code point order. + * If either of the UTF-16 strings is malformed (i.e., it contains unpaired surrogates), then the result is not defined. + * + * @param start The start offset in this string at which the compare operation begins. + * @param length The number of code units from this string to compare. + * @param srcChars A pointer to another string to compare this one to. + * @return a negative/zero/positive integer corresponding to whether + * this string is less than/equal to/greater than the second one + * in code point order + * @stable ICU 2.0 + */ + inline int8_t compareCodePointOrder(int32_t start, + int32_t length, + const char16_t *srcChars) const; + + /** + * Compare two Unicode strings in code point order. + * The result may be different from the results of compare(), operator<, etc. + * if supplementary characters are present: + * + * In UTF-16, supplementary characters (with code points U+10000 and above) are + * stored with pairs of surrogate code units. These have values from 0xd800 to 0xdfff, + * which means that they compare as less than some other BMP characters like U+feff. + * This function compares Unicode strings in code point order. + * If either of the UTF-16 strings is malformed (i.e., it contains unpaired surrogates), then the result is not defined. + * + * @param start The start offset in this string at which the compare operation begins. + * @param length The number of code units from this string to compare. + * @param srcChars A pointer to another string to compare this one to. + * @param srcStart The start offset in that string at which the compare operation begins. + * @param srcLength The number of code units from that string to compare. + * @return a negative/zero/positive integer corresponding to whether + * this string is less than/equal to/greater than the second one + * in code point order + * @stable ICU 2.0 + */ + inline int8_t compareCodePointOrder(int32_t start, + int32_t length, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength) const; + + /** + * Compare two Unicode strings in code point order. + * The result may be different from the results of compare(), operator<, etc. + * if supplementary characters are present: + * + * In UTF-16, supplementary characters (with code points U+10000 and above) are + * stored with pairs of surrogate code units. These have values from 0xd800 to 0xdfff, + * which means that they compare as less than some other BMP characters like U+feff. + * This function compares Unicode strings in code point order. + * If either of the UTF-16 strings is malformed (i.e., it contains unpaired surrogates), then the result is not defined. + * + * @param start The start offset in this string at which the compare operation begins. + * @param limit The offset after the last code unit from this string to compare. + * @param srcText Another string to compare this one to. + * @param srcStart The start offset in that string at which the compare operation begins. + * @param srcLimit The offset after the last code unit from that string to compare. + * @return a negative/zero/positive integer corresponding to whether + * this string is less than/equal to/greater than the second one + * in code point order + * @stable ICU 2.0 + */ + inline int8_t compareCodePointOrderBetween(int32_t start, + int32_t limit, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLimit) const; + + /** + * Compare two strings case-insensitively using full case folding. + * This is equivalent to this->foldCase(options).compare(text.foldCase(options)). + * + * @param text Another string to compare this one to. + * @param options A bit set of options: + * - U_FOLD_CASE_DEFAULT or 0 is used for default options: + * Comparison in code unit order with default case folding. + * + * - U_COMPARE_CODE_POINT_ORDER + * Set to choose code point order instead of code unit order + * (see u_strCompare for details). + * + * - U_FOLD_CASE_EXCLUDE_SPECIAL_I + * + * @return A negative, zero, or positive integer indicating the comparison result. + * @stable ICU 2.0 + */ + inline int8_t caseCompare(const UnicodeString& text, uint32_t options) const; + + /** + * Compare two strings case-insensitively using full case folding. + * This is equivalent to this->foldCase(options).compare(srcText.foldCase(options)). + * + * @param start The start offset in this string at which the compare operation begins. + * @param length The number of code units from this string to compare. + * @param srcText Another string to compare this one to. + * @param options A bit set of options: + * - U_FOLD_CASE_DEFAULT or 0 is used for default options: + * Comparison in code unit order with default case folding. + * + * - U_COMPARE_CODE_POINT_ORDER + * Set to choose code point order instead of code unit order + * (see u_strCompare for details). + * + * - U_FOLD_CASE_EXCLUDE_SPECIAL_I + * + * @return A negative, zero, or positive integer indicating the comparison result. + * @stable ICU 2.0 + */ + inline int8_t caseCompare(int32_t start, + int32_t length, + const UnicodeString& srcText, + uint32_t options) const; + + /** + * Compare two strings case-insensitively using full case folding. + * This is equivalent to this->foldCase(options).compare(srcText.foldCase(options)). + * + * @param start The start offset in this string at which the compare operation begins. + * @param length The number of code units from this string to compare. + * @param srcText Another string to compare this one to. + * @param srcStart The start offset in that string at which the compare operation begins. + * @param srcLength The number of code units from that string to compare. + * @param options A bit set of options: + * - U_FOLD_CASE_DEFAULT or 0 is used for default options: + * Comparison in code unit order with default case folding. + * + * - U_COMPARE_CODE_POINT_ORDER + * Set to choose code point order instead of code unit order + * (see u_strCompare for details). + * + * - U_FOLD_CASE_EXCLUDE_SPECIAL_I + * + * @return A negative, zero, or positive integer indicating the comparison result. + * @stable ICU 2.0 + */ + inline int8_t caseCompare(int32_t start, + int32_t length, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength, + uint32_t options) const; + + /** + * Compare two strings case-insensitively using full case folding. + * This is equivalent to this->foldCase(options).compare(srcChars.foldCase(options)). + * + * @param srcChars A pointer to another string to compare this one to. + * @param srcLength The number of code units from that string to compare. + * @param options A bit set of options: + * - U_FOLD_CASE_DEFAULT or 0 is used for default options: + * Comparison in code unit order with default case folding. + * + * - U_COMPARE_CODE_POINT_ORDER + * Set to choose code point order instead of code unit order + * (see u_strCompare for details). + * + * - U_FOLD_CASE_EXCLUDE_SPECIAL_I + * + * @return A negative, zero, or positive integer indicating the comparison result. + * @stable ICU 2.0 + */ + inline int8_t caseCompare(ConstChar16Ptr srcChars, + int32_t srcLength, + uint32_t options) const; + + /** + * Compare two strings case-insensitively using full case folding. + * This is equivalent to this->foldCase(options).compare(srcChars.foldCase(options)). + * + * @param start The start offset in this string at which the compare operation begins. + * @param length The number of code units from this string to compare. + * @param srcChars A pointer to another string to compare this one to. + * @param options A bit set of options: + * - U_FOLD_CASE_DEFAULT or 0 is used for default options: + * Comparison in code unit order with default case folding. + * + * - U_COMPARE_CODE_POINT_ORDER + * Set to choose code point order instead of code unit order + * (see u_strCompare for details). + * + * - U_FOLD_CASE_EXCLUDE_SPECIAL_I + * + * @return A negative, zero, or positive integer indicating the comparison result. + * @stable ICU 2.0 + */ + inline int8_t caseCompare(int32_t start, + int32_t length, + const char16_t *srcChars, + uint32_t options) const; + + /** + * Compare two strings case-insensitively using full case folding. + * This is equivalent to this->foldCase(options).compare(srcChars.foldCase(options)). + * + * @param start The start offset in this string at which the compare operation begins. + * @param length The number of code units from this string to compare. + * @param srcChars A pointer to another string to compare this one to. + * @param srcStart The start offset in that string at which the compare operation begins. + * @param srcLength The number of code units from that string to compare. + * @param options A bit set of options: + * - U_FOLD_CASE_DEFAULT or 0 is used for default options: + * Comparison in code unit order with default case folding. + * + * - U_COMPARE_CODE_POINT_ORDER + * Set to choose code point order instead of code unit order + * (see u_strCompare for details). + * + * - U_FOLD_CASE_EXCLUDE_SPECIAL_I + * + * @return A negative, zero, or positive integer indicating the comparison result. + * @stable ICU 2.0 + */ + inline int8_t caseCompare(int32_t start, + int32_t length, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength, + uint32_t options) const; + + /** + * Compare two strings case-insensitively using full case folding. + * This is equivalent to this->foldCase(options).compareBetween(text.foldCase(options)). + * + * @param start The start offset in this string at which the compare operation begins. + * @param limit The offset after the last code unit from this string to compare. + * @param srcText Another string to compare this one to. + * @param srcStart The start offset in that string at which the compare operation begins. + * @param srcLimit The offset after the last code unit from that string to compare. + * @param options A bit set of options: + * - U_FOLD_CASE_DEFAULT or 0 is used for default options: + * Comparison in code unit order with default case folding. + * + * - U_COMPARE_CODE_POINT_ORDER + * Set to choose code point order instead of code unit order + * (see u_strCompare for details). + * + * - U_FOLD_CASE_EXCLUDE_SPECIAL_I + * + * @return A negative, zero, or positive integer indicating the comparison result. + * @stable ICU 2.0 + */ + inline int8_t caseCompareBetween(int32_t start, + int32_t limit, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLimit, + uint32_t options) const; + + /** + * Determine if this starts with the characters in `text` + * @param text The text to match. + * @return true if this starts with the characters in `text`, + * false otherwise + * @stable ICU 2.0 + */ + inline UBool startsWith(const UnicodeString& text) const; + + /** + * Determine if this starts with the characters in `srcText` + * in the range [`srcStart`, `srcStart + srcLength`). + * @param srcText The text to match. + * @param srcStart the offset into `srcText` to start matching + * @param srcLength the number of characters in `srcText` to match + * @return true if this starts with the characters in `text`, + * false otherwise + * @stable ICU 2.0 + */ + inline UBool startsWith(const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const; + + /** + * Determine if this starts with the characters in `srcChars` + * @param srcChars The characters to match. + * @param srcLength the number of characters in `srcChars` + * @return true if this starts with the characters in `srcChars`, + * false otherwise + * @stable ICU 2.0 + */ + inline UBool startsWith(ConstChar16Ptr srcChars, + int32_t srcLength) const; + + /** + * Determine if this ends with the characters in `srcChars` + * in the range [`srcStart`, `srcStart + srcLength`). + * @param srcChars The characters to match. + * @param srcStart the offset into `srcText` to start matching + * @param srcLength the number of characters in `srcChars` to match + * @return true if this ends with the characters in `srcChars`, false otherwise + * @stable ICU 2.0 + */ + inline UBool startsWith(const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength) const; + + /** + * Determine if this ends with the characters in `text` + * @param text The text to match. + * @return true if this ends with the characters in `text`, + * false otherwise + * @stable ICU 2.0 + */ + inline UBool endsWith(const UnicodeString& text) const; + + /** + * Determine if this ends with the characters in `srcText` + * in the range [`srcStart`, `srcStart + srcLength`). + * @param srcText The text to match. + * @param srcStart the offset into `srcText` to start matching + * @param srcLength the number of characters in `srcText` to match + * @return true if this ends with the characters in `text`, + * false otherwise + * @stable ICU 2.0 + */ + inline UBool endsWith(const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const; + + /** + * Determine if this ends with the characters in `srcChars` + * @param srcChars The characters to match. + * @param srcLength the number of characters in `srcChars` + * @return true if this ends with the characters in `srcChars`, + * false otherwise + * @stable ICU 2.0 + */ + inline UBool endsWith(ConstChar16Ptr srcChars, + int32_t srcLength) const; + + /** + * Determine if this ends with the characters in `srcChars` + * in the range [`srcStart`, `srcStart + srcLength`). + * @param srcChars The characters to match. + * @param srcStart the offset into `srcText` to start matching + * @param srcLength the number of characters in `srcChars` to match + * @return true if this ends with the characters in `srcChars`, + * false otherwise + * @stable ICU 2.0 + */ + inline UBool endsWith(const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength) const; + + + /* Searching - bitwise only */ + + /** + * Locate in this the first occurrence of the characters in `text`, + * using bitwise comparison. + * @param text The text to search for. + * @return The offset into this of the start of `text`, + * or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t indexOf(const UnicodeString& text) const; + + /** + * Locate in this the first occurrence of the characters in `text` + * starting at offset `start`, using bitwise comparison. + * @param text The text to search for. + * @param start The offset at which searching will start. + * @return The offset into this of the start of `text`, + * or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t indexOf(const UnicodeString& text, + int32_t start) const; + + /** + * Locate in this the first occurrence in the range + * [`start`, `start + length`) of the characters + * in `text`, using bitwise comparison. + * @param text The text to search for. + * @param start The offset at which searching will start. + * @param length The number of characters to search + * @return The offset into this of the start of `text`, + * or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t indexOf(const UnicodeString& text, + int32_t start, + int32_t length) const; + + /** + * Locate in this the first occurrence in the range + * [`start`, `start + length`) of the characters + * in `srcText` in the range + * [`srcStart`, `srcStart + srcLength`), + * using bitwise comparison. + * @param srcText The text to search for. + * @param srcStart the offset into `srcText` at which + * to start matching + * @param srcLength the number of characters in `srcText` to match + * @param start the offset into this at which to start matching + * @param length the number of characters in this to search + * @return The offset into this of the start of `text`, + * or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t indexOf(const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength, + int32_t start, + int32_t length) const; + + /** + * Locate in this the first occurrence of the characters in + * `srcChars` + * starting at offset `start`, using bitwise comparison. + * @param srcChars The text to search for. + * @param srcLength the number of characters in `srcChars` to match + * @param start the offset into this at which to start matching + * @return The offset into this of the start of `text`, + * or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t indexOf(const char16_t *srcChars, + int32_t srcLength, + int32_t start) const; + + /** + * Locate in this the first occurrence in the range + * [`start`, `start + length`) of the characters + * in `srcChars`, using bitwise comparison. + * @param srcChars The text to search for. + * @param srcLength the number of characters in `srcChars` + * @param start The offset at which searching will start. + * @param length The number of characters to search + * @return The offset into this of the start of `srcChars`, + * or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t indexOf(ConstChar16Ptr srcChars, + int32_t srcLength, + int32_t start, + int32_t length) const; + + /** + * Locate in this the first occurrence in the range + * [`start`, `start + length`) of the characters + * in `srcChars` in the range + * [`srcStart`, `srcStart + srcLength`), + * using bitwise comparison. + * @param srcChars The text to search for. + * @param srcStart the offset into `srcChars` at which + * to start matching + * @param srcLength the number of characters in `srcChars` to match + * @param start the offset into this at which to start matching + * @param length the number of characters in this to search + * @return The offset into this of the start of `text`, + * or -1 if not found. + * @stable ICU 2.0 + */ + int32_t indexOf(const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength, + int32_t start, + int32_t length) const; + + /** + * Locate in this the first occurrence of the BMP code point `c`, + * using bitwise comparison. + * @param c The code unit to search for. + * @return The offset into this of `c`, or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t indexOf(char16_t c) const; + + /** + * Locate in this the first occurrence of the code point `c`, + * using bitwise comparison. + * + * @param c The code point to search for. + * @return The offset into this of `c`, or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t indexOf(UChar32 c) const; + + /** + * Locate in this the first occurrence of the BMP code point `c`, + * starting at offset `start`, using bitwise comparison. + * @param c The code unit to search for. + * @param start The offset at which searching will start. + * @return The offset into this of `c`, or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t indexOf(char16_t c, + int32_t start) const; + + /** + * Locate in this the first occurrence of the code point `c` + * starting at offset `start`, using bitwise comparison. + * + * @param c The code point to search for. + * @param start The offset at which searching will start. + * @return The offset into this of `c`, or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t indexOf(UChar32 c, + int32_t start) const; + + /** + * Locate in this the first occurrence of the BMP code point `c` + * in the range [`start`, `start + length`), + * using bitwise comparison. + * @param c The code unit to search for. + * @param start the offset into this at which to start matching + * @param length the number of characters in this to search + * @return The offset into this of `c`, or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t indexOf(char16_t c, + int32_t start, + int32_t length) const; + + /** + * Locate in this the first occurrence of the code point `c` + * in the range [`start`, `start + length`), + * using bitwise comparison. + * + * @param c The code point to search for. + * @param start the offset into this at which to start matching + * @param length the number of characters in this to search + * @return The offset into this of `c`, or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t indexOf(UChar32 c, + int32_t start, + int32_t length) const; + + /** + * Locate in this the last occurrence of the characters in `text`, + * using bitwise comparison. + * @param text The text to search for. + * @return The offset into this of the start of `text`, + * or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t lastIndexOf(const UnicodeString& text) const; + + /** + * Locate in this the last occurrence of the characters in `text` + * starting at offset `start`, using bitwise comparison. + * @param text The text to search for. + * @param start The offset at which searching will start. + * @return The offset into this of the start of `text`, + * or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t lastIndexOf(const UnicodeString& text, + int32_t start) const; + + /** + * Locate in this the last occurrence in the range + * [`start`, `start + length`) of the characters + * in `text`, using bitwise comparison. + * @param text The text to search for. + * @param start The offset at which searching will start. + * @param length The number of characters to search + * @return The offset into this of the start of `text`, + * or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t lastIndexOf(const UnicodeString& text, + int32_t start, + int32_t length) const; + + /** + * Locate in this the last occurrence in the range + * [`start`, `start + length`) of the characters + * in `srcText` in the range + * [`srcStart`, `srcStart + srcLength`), + * using bitwise comparison. + * @param srcText The text to search for. + * @param srcStart the offset into `srcText` at which + * to start matching + * @param srcLength the number of characters in `srcText` to match + * @param start the offset into this at which to start matching + * @param length the number of characters in this to search + * @return The offset into this of the start of `text`, + * or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t lastIndexOf(const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength, + int32_t start, + int32_t length) const; + + /** + * Locate in this the last occurrence of the characters in `srcChars` + * starting at offset `start`, using bitwise comparison. + * @param srcChars The text to search for. + * @param srcLength the number of characters in `srcChars` to match + * @param start the offset into this at which to start matching + * @return The offset into this of the start of `text`, + * or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t lastIndexOf(const char16_t *srcChars, + int32_t srcLength, + int32_t start) const; + + /** + * Locate in this the last occurrence in the range + * [`start`, `start + length`) of the characters + * in `srcChars`, using bitwise comparison. + * @param srcChars The text to search for. + * @param srcLength the number of characters in `srcChars` + * @param start The offset at which searching will start. + * @param length The number of characters to search + * @return The offset into this of the start of `srcChars`, + * or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t lastIndexOf(ConstChar16Ptr srcChars, + int32_t srcLength, + int32_t start, + int32_t length) const; + + /** + * Locate in this the last occurrence in the range + * [`start`, `start + length`) of the characters + * in `srcChars` in the range + * [`srcStart`, `srcStart + srcLength`), + * using bitwise comparison. + * @param srcChars The text to search for. + * @param srcStart the offset into `srcChars` at which + * to start matching + * @param srcLength the number of characters in `srcChars` to match + * @param start the offset into this at which to start matching + * @param length the number of characters in this to search + * @return The offset into this of the start of `text`, + * or -1 if not found. + * @stable ICU 2.0 + */ + int32_t lastIndexOf(const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength, + int32_t start, + int32_t length) const; + + /** + * Locate in this the last occurrence of the BMP code point `c`, + * using bitwise comparison. + * @param c The code unit to search for. + * @return The offset into this of `c`, or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t lastIndexOf(char16_t c) const; + + /** + * Locate in this the last occurrence of the code point `c`, + * using bitwise comparison. + * + * @param c The code point to search for. + * @return The offset into this of `c`, or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t lastIndexOf(UChar32 c) const; + + /** + * Locate in this the last occurrence of the BMP code point `c` + * starting at offset `start`, using bitwise comparison. + * @param c The code unit to search for. + * @param start The offset at which searching will start. + * @return The offset into this of `c`, or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t lastIndexOf(char16_t c, + int32_t start) const; + + /** + * Locate in this the last occurrence of the code point `c` + * starting at offset `start`, using bitwise comparison. + * + * @param c The code point to search for. + * @param start The offset at which searching will start. + * @return The offset into this of `c`, or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t lastIndexOf(UChar32 c, + int32_t start) const; + + /** + * Locate in this the last occurrence of the BMP code point `c` + * in the range [`start`, `start + length`), + * using bitwise comparison. + * @param c The code unit to search for. + * @param start the offset into this at which to start matching + * @param length the number of characters in this to search + * @return The offset into this of `c`, or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t lastIndexOf(char16_t c, + int32_t start, + int32_t length) const; + + /** + * Locate in this the last occurrence of the code point `c` + * in the range [`start`, `start + length`), + * using bitwise comparison. + * + * @param c The code point to search for. + * @param start the offset into this at which to start matching + * @param length the number of characters in this to search + * @return The offset into this of `c`, or -1 if not found. + * @stable ICU 2.0 + */ + inline int32_t lastIndexOf(UChar32 c, + int32_t start, + int32_t length) const; + + + /* Character access */ + + /** + * Return the code unit at offset `offset`. + * If the offset is not valid (0..length()-1) then U+ffff is returned. + * @param offset a valid offset into the text + * @return the code unit at offset `offset` + * or 0xffff if the offset is not valid for this string + * @stable ICU 2.0 + */ + inline char16_t charAt(int32_t offset) const; + + /** + * Return the code unit at offset `offset`. + * If the offset is not valid (0..length()-1) then U+ffff is returned. + * @param offset a valid offset into the text + * @return the code unit at offset `offset` + * @stable ICU 2.0 + */ + inline char16_t operator[] (int32_t offset) const; + + /** + * Return the code point that contains the code unit + * at offset `offset`. + * If the offset is not valid (0..length()-1) then U+ffff is returned. + * @param offset a valid offset into the text + * that indicates the text offset of any of the code units + * that will be assembled into a code point (21-bit value) and returned + * @return the code point of text at `offset` + * or 0xffff if the offset is not valid for this string + * @stable ICU 2.0 + */ + UChar32 char32At(int32_t offset) const; + + /** + * Adjust a random-access offset so that + * it points to the beginning of a Unicode character. + * The offset that is passed in points to + * any code unit of a code point, + * while the returned offset will point to the first code unit + * of the same code point. + * In UTF-16, if the input offset points to a second surrogate + * of a surrogate pair, then the returned offset will point + * to the first surrogate. + * @param offset a valid offset into one code point of the text + * @return offset of the first code unit of the same code point + * @see U16_SET_CP_START + * @stable ICU 2.0 + */ + int32_t getChar32Start(int32_t offset) const; + + /** + * Adjust a random-access offset so that + * it points behind a Unicode character. + * The offset that is passed in points behind + * any code unit of a code point, + * while the returned offset will point behind the last code unit + * of the same code point. + * In UTF-16, if the input offset points behind the first surrogate + * (i.e., to the second surrogate) + * of a surrogate pair, then the returned offset will point + * behind the second surrogate (i.e., to the first surrogate). + * @param offset a valid offset after any code unit of a code point of the text + * @return offset of the first code unit after the same code point + * @see U16_SET_CP_LIMIT + * @stable ICU 2.0 + */ + int32_t getChar32Limit(int32_t offset) const; + + /** + * Move the code unit index along the string by delta code points. + * Interpret the input index as a code unit-based offset into the string, + * move the index forward or backward by delta code points, and + * return the resulting index. + * The input index should point to the first code unit of a code point, + * if there is more than one. + * + * Both input and output indexes are code unit-based as for all + * string indexes/offsets in ICU (and other libraries, like MBCS char*). + * If delta<0 then the index is moved backward (toward the start of the string). + * If delta>0 then the index is moved forward (toward the end of the string). + * + * This behaves like CharacterIterator::move32(delta, kCurrent). + * + * Behavior for out-of-bounds indexes: + * `moveIndex32` pins the input index to 0..length(), i.e., + * if the input index<0 then it is pinned to 0; + * if it is index>length() then it is pinned to length(). + * Afterwards, the index is moved by `delta` code points + * forward or backward, + * but no further backward than to 0 and no further forward than to length(). + * The resulting index return value will be in between 0 and length(), inclusively. + * + * Examples: + * \code + * // s has code points 'a' U+10000 'b' U+10ffff U+2029 + * UnicodeString s(u"a\U00010000b\U0010ffff\u2029"); + * + * // initial index: position of U+10000 + * int32_t index=1; + * + * // the following examples will all result in index==4, position of U+10ffff + * + * // skip 2 code points from some position in the string + * index=s.moveIndex32(index, 2); // skips U+10000 and 'b' + * + * // go to the 3rd code point from the start of s (0-based) + * index=s.moveIndex32(0, 3); // skips 'a', U+10000, and 'b' + * + * // go to the next-to-last code point of s + * index=s.moveIndex32(s.length(), -2); // backward-skips U+2029 and U+10ffff + * \endcode + * + * @param index input code unit index + * @param delta (signed) code point count to move the index forward or backward + * in the string + * @return the resulting code unit index + * @stable ICU 2.0 + */ + int32_t moveIndex32(int32_t index, int32_t delta) const; + + /* Substring extraction */ + + /** + * Copy the characters in the range + * [`start`, `start + length`) into the array `dst`, + * beginning at `dstStart`. + * If the string aliases to `dst` itself as an external buffer, + * then extract() will not copy the contents. + * + * @param start offset of first character which will be copied into the array + * @param length the number of characters to extract + * @param dst array in which to copy characters. The length of `dst` + * must be at least (`dstStart + length`). + * @param dstStart the offset in `dst` where the first character + * will be extracted + * @stable ICU 2.0 + */ + inline void extract(int32_t start, + int32_t length, + Char16Ptr dst, + int32_t dstStart = 0) const; + + /** + * Copy the contents of the string into dest. + * This is a convenience function that + * checks if there is enough space in dest, + * extracts the entire string if possible, + * and NUL-terminates dest if possible. + * + * If the string fits into dest but cannot be NUL-terminated + * (length()==destCapacity) then the error code is set to U_STRING_NOT_TERMINATED_WARNING. + * If the string itself does not fit into dest + * (length()>destCapacity) then the error code is set to U_BUFFER_OVERFLOW_ERROR. + * + * If the string aliases to `dest` itself as an external buffer, + * then extract() will not copy the contents. + * + * @param dest Destination string buffer. + * @param destCapacity Number of char16_ts available at dest. + * @param errorCode ICU error code. + * @return length() + * @stable ICU 2.0 + */ + int32_t + extract(Char16Ptr dest, int32_t destCapacity, + UErrorCode &errorCode) const; + + /** + * Copy the characters in the range + * [`start`, `start + length`) into the UnicodeString + * `target`. + * @param start offset of first character which will be copied + * @param length the number of characters to extract + * @param target UnicodeString into which to copy characters. + * @stable ICU 2.0 + */ + inline void extract(int32_t start, + int32_t length, + UnicodeString& target) const; + + /** + * Copy the characters in the range [`start`, `limit`) + * into the array `dst`, beginning at `dstStart`. + * @param start offset of first character which will be copied into the array + * @param limit offset immediately following the last character to be copied + * @param dst array in which to copy characters. The length of `dst` + * must be at least (`dstStart + (limit - start)`). + * @param dstStart the offset in `dst` where the first character + * will be extracted + * @stable ICU 2.0 + */ + inline void extractBetween(int32_t start, + int32_t limit, + char16_t *dst, + int32_t dstStart = 0) const; + + /** + * Copy the characters in the range [`start`, `limit`) + * into the UnicodeString `target`. Replaceable API. + * @param start offset of first character which will be copied + * @param limit offset immediately following the last character to be copied + * @param target UnicodeString into which to copy characters. + * @stable ICU 2.0 + */ + virtual void extractBetween(int32_t start, + int32_t limit, + UnicodeString& target) const override; + + /** + * Copy the characters in the range + * [`start`, `start + startLength`) into an array of characters. + * All characters must be invariant (see utypes.h). + * Use US_INV as the last, signature-distinguishing parameter. + * + * This function does not write any more than `targetCapacity` + * characters but returns the length of the entire output string + * so that one can allocate a larger buffer and call the function again + * if necessary. + * The output string is NUL-terminated if possible. + * + * @param start offset of first character which will be copied + * @param startLength the number of characters to extract + * @param target the target buffer for extraction, can be nullptr + * if targetLength is 0 + * @param targetCapacity the length of the target buffer + * @param inv Signature-distinguishing parameter, use US_INV. + * @return the output string length, not including the terminating NUL + * @stable ICU 3.2 + */ + int32_t extract(int32_t start, + int32_t startLength, + char *target, + int32_t targetCapacity, + enum EInvariant inv) const; + +#if U_CHARSET_IS_UTF8 || !UCONFIG_NO_CONVERSION + + /** + * Copy the characters in the range + * [`start`, `start + length`) into an array of characters + * in the platform's default codepage. + * This function does not write any more than `targetLength` + * characters but returns the length of the entire output string + * so that one can allocate a larger buffer and call the function again + * if necessary. + * The output string is NUL-terminated if possible. + * + * @param start offset of first character which will be copied + * @param startLength the number of characters to extract + * @param target the target buffer for extraction + * @param targetLength the length of the target buffer + * If `target` is nullptr, then the number of bytes required for + * `target` is returned. + * @return the output string length, not including the terminating NUL + * @stable ICU 2.0 + */ + int32_t extract(int32_t start, + int32_t startLength, + char *target, + uint32_t targetLength) const; + +#endif + +#if !UCONFIG_NO_CONVERSION + + /** + * Copy the characters in the range + * [`start`, `start + length`) into an array of characters + * in a specified codepage. + * The output string is NUL-terminated. + * + * Recommendation: For invariant-character strings use + * extract(int32_t start, int32_t length, char *target, int32_t targetCapacity, enum EInvariant inv) const + * because it avoids object code dependencies of UnicodeString on + * the conversion code. + * + * @param start offset of first character which will be copied + * @param startLength the number of characters to extract + * @param target the target buffer for extraction + * @param codepage the desired codepage for the characters. 0 has + * the special meaning of the default codepage + * If `codepage` is an empty string (`""`), + * then a simple conversion is performed on the codepage-invariant + * subset ("invariant characters") of the platform encoding. See utypes.h. + * If `target` is nullptr, then the number of bytes required for + * `target` is returned. It is assumed that the target is big enough + * to fit all of the characters. + * @return the output string length, not including the terminating NUL + * @stable ICU 2.0 + */ + inline int32_t extract(int32_t start, + int32_t startLength, + char *target, + const char *codepage = 0) const; + + /** + * Copy the characters in the range + * [`start`, `start + length`) into an array of characters + * in a specified codepage. + * This function does not write any more than `targetLength` + * characters but returns the length of the entire output string + * so that one can allocate a larger buffer and call the function again + * if necessary. + * The output string is NUL-terminated if possible. + * + * Recommendation: For invariant-character strings use + * extract(int32_t start, int32_t length, char *target, int32_t targetCapacity, enum EInvariant inv) const + * because it avoids object code dependencies of UnicodeString on + * the conversion code. + * + * @param start offset of first character which will be copied + * @param startLength the number of characters to extract + * @param target the target buffer for extraction + * @param targetLength the length of the target buffer + * @param codepage the desired codepage for the characters. 0 has + * the special meaning of the default codepage + * If `codepage` is an empty string (`""`), + * then a simple conversion is performed on the codepage-invariant + * subset ("invariant characters") of the platform encoding. See utypes.h. + * If `target` is nullptr, then the number of bytes required for + * `target` is returned. + * @return the output string length, not including the terminating NUL + * @stable ICU 2.0 + */ + int32_t extract(int32_t start, + int32_t startLength, + char *target, + uint32_t targetLength, + const char *codepage) const; + + /** + * Convert the UnicodeString into a codepage string using an existing UConverter. + * The output string is NUL-terminated if possible. + * + * This function avoids the overhead of opening and closing a converter if + * multiple strings are extracted. + * + * @param dest destination string buffer, can be nullptr if destCapacity==0 + * @param destCapacity the number of chars available at dest + * @param cnv the converter object to be used (ucnv_resetFromUnicode() will be called), + * or nullptr for the default converter + * @param errorCode normal ICU error code + * @return the length of the output string, not counting the terminating NUL; + * if the length is greater than destCapacity, then the string will not fit + * and a buffer of the indicated length would need to be passed in + * @stable ICU 2.0 + */ + int32_t extract(char *dest, int32_t destCapacity, + UConverter *cnv, + UErrorCode &errorCode) const; + +#endif + + /** + * Create a temporary substring for the specified range. + * Unlike the substring constructor and setTo() functions, + * the object returned here will be a read-only alias (using getBuffer()) + * rather than copying the text. + * As a result, this substring operation is much faster but requires + * that the original string not be modified or deleted during the lifetime + * of the returned substring object. + * @param start offset of the first character visible in the substring + * @param length length of the substring + * @return a read-only alias UnicodeString object for the substring + * @stable ICU 4.4 + */ + UnicodeString tempSubString(int32_t start=0, int32_t length=INT32_MAX) const; + + /** + * Create a temporary substring for the specified range. + * Same as tempSubString(start, length) except that the substring range + * is specified as a (start, limit) pair (with an exclusive limit index) + * rather than a (start, length) pair. + * @param start offset of the first character visible in the substring + * @param limit offset immediately following the last character visible in the substring + * @return a read-only alias UnicodeString object for the substring + * @stable ICU 4.4 + */ + inline UnicodeString tempSubStringBetween(int32_t start, int32_t limit=INT32_MAX) const; + + /** + * Convert the UnicodeString to UTF-8 and write the result + * to a ByteSink. This is called by toUTF8String(). + * Unpaired surrogates are replaced with U+FFFD. + * Calls u_strToUTF8WithSub(). + * + * @param sink A ByteSink to which the UTF-8 version of the string is written. + * sink.Flush() is called at the end. + * @stable ICU 4.2 + * @see toUTF8String + */ + void toUTF8(ByteSink &sink) const; + + /** + * Convert the UnicodeString to UTF-8 and append the result + * to a standard string. + * Unpaired surrogates are replaced with U+FFFD. + * Calls toUTF8(). + * + * @param result A standard string (or a compatible object) + * to which the UTF-8 version of the string is appended. + * @return The string object. + * @stable ICU 4.2 + * @see toUTF8 + */ + template + StringClass &toUTF8String(StringClass &result) const { + StringByteSink sbs(&result, length()); + toUTF8(sbs); + return result; + } + + /** + * Convert the UnicodeString to UTF-32. + * Unpaired surrogates are replaced with U+FFFD. + * Calls u_strToUTF32WithSub(). + * + * @param utf32 destination string buffer, can be nullptr if capacity==0 + * @param capacity the number of UChar32s available at utf32 + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return The length of the UTF-32 string. + * @see fromUTF32 + * @stable ICU 4.2 + */ + int32_t toUTF32(UChar32 *utf32, int32_t capacity, UErrorCode &errorCode) const; + + /* Length operations */ + + /** + * Return the length of the UnicodeString object. + * The length is the number of char16_t code units are in the UnicodeString. + * If you want the number of code points, please use countChar32(). + * @return the length of the UnicodeString object + * @see countChar32 + * @stable ICU 2.0 + */ + inline int32_t length(void) const; + + /** + * Count Unicode code points in the length char16_t code units of the string. + * A code point may occupy either one or two char16_t code units. + * Counting code points involves reading all code units. + * + * This functions is basically the inverse of moveIndex32(). + * + * @param start the index of the first code unit to check + * @param length the number of char16_t code units to check + * @return the number of code points in the specified code units + * @see length + * @stable ICU 2.0 + */ + int32_t + countChar32(int32_t start=0, int32_t length=INT32_MAX) const; + + /** + * Check if the length char16_t code units of the string + * contain more Unicode code points than a certain number. + * This is more efficient than counting all code points in this part of the string + * and comparing that number with a threshold. + * This function may not need to scan the string at all if the length + * falls within a certain range, and + * never needs to count more than 'number+1' code points. + * Logically equivalent to (countChar32(start, length)>number). + * A Unicode code point may occupy either one or two char16_t code units. + * + * @param start the index of the first code unit to check (0 for the entire string) + * @param length the number of char16_t code units to check + * (use INT32_MAX for the entire string; remember that start/length + * values are pinned) + * @param number The number of code points in the (sub)string is compared against + * the 'number' parameter. + * @return Boolean value for whether the string contains more Unicode code points + * than 'number'. Same as (u_countChar32(s, length)>number). + * @see countChar32 + * @see u_strHasMoreChar32Than + * @stable ICU 2.4 + */ + UBool + hasMoreChar32Than(int32_t start, int32_t length, int32_t number) const; + + /** + * Determine if this string is empty. + * @return true if this string contains 0 characters, false otherwise. + * @stable ICU 2.0 + */ + inline UBool isEmpty(void) const; + + /** + * Return the capacity of the internal buffer of the UnicodeString object. + * This is useful together with the getBuffer functions. + * See there for details. + * + * @return the number of char16_ts available in the internal buffer + * @see getBuffer + * @stable ICU 2.0 + */ + inline int32_t getCapacity(void) const; + + /* Other operations */ + + /** + * Generate a hash code for this object. + * @return The hash code of this UnicodeString. + * @stable ICU 2.0 + */ + inline int32_t hashCode(void) const; + + /** + * Determine if this object contains a valid string. + * A bogus string has no value. It is different from an empty string, + * although in both cases isEmpty() returns true and length() returns 0. + * setToBogus() and isBogus() can be used to indicate that no string value is available. + * For a bogus string, getBuffer() and getTerminatedBuffer() return nullptr, and + * length() returns 0. + * + * @return true if the string is bogus/invalid, false otherwise + * @see setToBogus() + * @stable ICU 2.0 + */ + inline UBool isBogus(void) const; + + + //======================================== + // Write operations + //======================================== + + /* Assignment operations */ + + /** + * Assignment operator. Replace the characters in this UnicodeString + * with the characters from `srcText`. + * + * Starting with ICU 2.4, the assignment operator and the copy constructor + * allocate a new buffer and copy the buffer contents even for readonly aliases. + * By contrast, the fastCopyFrom() function implements the old, + * more efficient but less safe behavior + * of making this string also a readonly alias to the same buffer. + * + * If the source object has an "open" buffer from getBuffer(minCapacity), + * then the copy is an empty string. + * + * @param srcText The text containing the characters to replace + * @return a reference to this + * @stable ICU 2.0 + * @see fastCopyFrom + */ + UnicodeString &operator=(const UnicodeString &srcText); + + /** + * Almost the same as the assignment operator. + * Replace the characters in this UnicodeString + * with the characters from `srcText`. + * + * This function works the same as the assignment operator + * for all strings except for ones that are readonly aliases. + * + * Starting with ICU 2.4, the assignment operator and the copy constructor + * allocate a new buffer and copy the buffer contents even for readonly aliases. + * This function implements the old, more efficient but less safe behavior + * of making this string also a readonly alias to the same buffer. + * + * The fastCopyFrom function must be used only if it is known that the lifetime of + * this UnicodeString does not exceed the lifetime of the aliased buffer + * including its contents, for example for strings from resource bundles + * or aliases to string constants. + * + * If the source object has an "open" buffer from getBuffer(minCapacity), + * then the copy is an empty string. + * + * @param src The text containing the characters to replace. + * @return a reference to this + * @stable ICU 2.4 + */ + UnicodeString &fastCopyFrom(const UnicodeString &src); + + /** + * Move assignment operator; might leave src in bogus state. + * This string will have the same contents and state that the source string had. + * The behavior is undefined if *this and src are the same object. + * @param src source string + * @return *this + * @stable ICU 56 + */ + UnicodeString &operator=(UnicodeString &&src) noexcept; + + /** + * Swap strings. + * @param other other string + * @stable ICU 56 + */ + void swap(UnicodeString &other) noexcept; + + /** + * Non-member UnicodeString swap function. + * @param s1 will get s2's contents and state + * @param s2 will get s1's contents and state + * @stable ICU 56 + */ + friend inline void U_EXPORT2 + swap(UnicodeString &s1, UnicodeString &s2) noexcept { + s1.swap(s2); + } + + /** + * Assignment operator. Replace the characters in this UnicodeString + * with the code unit `ch`. + * @param ch the code unit to replace + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& operator= (char16_t ch); + + /** + * Assignment operator. Replace the characters in this UnicodeString + * with the code point `ch`. + * @param ch the code point to replace + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& operator= (UChar32 ch); + + /** + * Set the text in the UnicodeString object to the characters + * in `srcText` in the range + * [`srcStart`, `srcText.length()`). + * `srcText` is not modified. + * @param srcText the source for the new characters + * @param srcStart the offset into `srcText` where new characters + * will be obtained + * @return a reference to this + * @stable ICU 2.2 + */ + inline UnicodeString& setTo(const UnicodeString& srcText, + int32_t srcStart); + + /** + * Set the text in the UnicodeString object to the characters + * in `srcText` in the range + * [`srcStart`, `srcStart + srcLength`). + * `srcText` is not modified. + * @param srcText the source for the new characters + * @param srcStart the offset into `srcText` where new characters + * will be obtained + * @param srcLength the number of characters in `srcText` in the + * replace string. + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& setTo(const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength); + + /** + * Set the text in the UnicodeString object to the characters in + * `srcText`. + * `srcText` is not modified. + * @param srcText the source for the new characters + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& setTo(const UnicodeString& srcText); + + /** + * Set the characters in the UnicodeString object to the characters + * in `srcChars`. `srcChars` is not modified. + * @param srcChars the source for the new characters + * @param srcLength the number of Unicode characters in srcChars. + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& setTo(const char16_t *srcChars, + int32_t srcLength); + + /** + * Set the characters in the UnicodeString object to the code unit + * `srcChar`. + * @param srcChar the code unit which becomes the UnicodeString's character + * content + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& setTo(char16_t srcChar); + + /** + * Set the characters in the UnicodeString object to the code point + * `srcChar`. + * @param srcChar the code point which becomes the UnicodeString's character + * content + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& setTo(UChar32 srcChar); + + /** + * Aliasing setTo() function, analogous to the readonly-aliasing char16_t* constructor. + * The text will be used for the UnicodeString object, but + * it will not be released when the UnicodeString is destroyed. + * This has copy-on-write semantics: + * When the string is modified, then the buffer is first copied into + * newly allocated memory. + * The aliased buffer is never modified. + * + * In an assignment to another UnicodeString, when using the copy constructor + * or the assignment operator, the text will be copied. + * When using fastCopyFrom(), the text will be aliased again, + * so that both strings then alias the same readonly-text. + * + * @param isTerminated specifies if `text` is `NUL`-terminated. + * This must be true if `textLength==-1`. + * @param text The characters to alias for the UnicodeString. + * @param textLength The number of Unicode characters in `text` to alias. + * If -1, then this constructor will determine the length + * by calling `u_strlen()`. + * @return a reference to this + * @stable ICU 2.0 + */ + UnicodeString &setTo(UBool isTerminated, + ConstChar16Ptr text, + int32_t textLength); + + /** + * Aliasing setTo() function, analogous to the writable-aliasing char16_t* constructor. + * The text will be used for the UnicodeString object, but + * it will not be released when the UnicodeString is destroyed. + * This has write-through semantics: + * For as long as the capacity of the buffer is sufficient, write operations + * will directly affect the buffer. When more capacity is necessary, then + * a new buffer will be allocated and the contents copied as with regularly + * constructed strings. + * In an assignment to another UnicodeString, the buffer will be copied. + * The extract(Char16Ptr dst) function detects whether the dst pointer is the same + * as the string buffer itself and will in this case not copy the contents. + * + * @param buffer The characters to alias for the UnicodeString. + * @param buffLength The number of Unicode characters in `buffer` to alias. + * @param buffCapacity The size of `buffer` in char16_ts. + * @return a reference to this + * @stable ICU 2.0 + */ + UnicodeString &setTo(char16_t *buffer, + int32_t buffLength, + int32_t buffCapacity); + + /** + * Make this UnicodeString object invalid. + * The string will test true with isBogus(). + * + * A bogus string has no value. It is different from an empty string. + * It can be used to indicate that no string value is available. + * getBuffer() and getTerminatedBuffer() return nullptr, and + * length() returns 0. + * + * This utility function is used throughout the UnicodeString + * implementation to indicate that a UnicodeString operation failed, + * and may be used in other functions, + * especially but not exclusively when such functions do not + * take a UErrorCode for simplicity. + * + * The following methods, and no others, will clear a string object's bogus flag: + * - remove() + * - remove(0, INT32_MAX) + * - truncate(0) + * - operator=() (assignment operator) + * - setTo(...) + * + * The simplest ways to turn a bogus string into an empty one + * is to use the remove() function. + * Examples for other functions that are equivalent to "set to empty string": + * \code + * if(s.isBogus()) { + * s.remove(); // set to an empty string (remove all), or + * s.remove(0, INT32_MAX); // set to an empty string (remove all), or + * s.truncate(0); // set to an empty string (complete truncation), or + * s=UnicodeString(); // assign an empty string, or + * s.setTo((UChar32)-1); // set to a pseudo code point that is out of range, or + * s.setTo(u"", 0); // set to an empty C Unicode string + * } + * \endcode + * + * @see isBogus() + * @stable ICU 2.0 + */ + void setToBogus(); + + /** + * Set the character at the specified offset to the specified character. + * @param offset A valid offset into the text of the character to set + * @param ch The new character + * @return A reference to this + * @stable ICU 2.0 + */ + UnicodeString& setCharAt(int32_t offset, + char16_t ch); + + + /* Append operations */ + + /** + * Append operator. Append the code unit `ch` to the UnicodeString + * object. + * @param ch the code unit to be appended + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& operator+= (char16_t ch); + + /** + * Append operator. Append the code point `ch` to the UnicodeString + * object. + * @param ch the code point to be appended + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& operator+= (UChar32 ch); + + /** + * Append operator. Append the characters in `srcText` to the + * UnicodeString object. `srcText` is not modified. + * @param srcText the source for the new characters + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& operator+= (const UnicodeString& srcText); + + /** + * Append the characters + * in `srcText` in the range + * [`srcStart`, `srcStart + srcLength`) to the + * UnicodeString object at offset `start`. `srcText` + * is not modified. + * @param srcText the source for the new characters + * @param srcStart the offset into `srcText` where new characters + * will be obtained + * @param srcLength the number of characters in `srcText` in + * the append string + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& append(const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength); + + /** + * Append the characters in `srcText` to the UnicodeString object. + * `srcText` is not modified. + * @param srcText the source for the new characters + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& append(const UnicodeString& srcText); + + /** + * Append the characters in `srcChars` in the range + * [`srcStart`, `srcStart + srcLength`) to the UnicodeString + * object at offset + * `start`. `srcChars` is not modified. + * @param srcChars the source for the new characters + * @param srcStart the offset into `srcChars` where new characters + * will be obtained + * @param srcLength the number of characters in `srcChars` in + * the append string; can be -1 if `srcChars` is NUL-terminated + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& append(const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength); + + /** + * Append the characters in `srcChars` to the UnicodeString object + * at offset `start`. `srcChars` is not modified. + * @param srcChars the source for the new characters + * @param srcLength the number of Unicode characters in `srcChars`; + * can be -1 if `srcChars` is NUL-terminated + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& append(ConstChar16Ptr srcChars, + int32_t srcLength); + + /** + * Append the code unit `srcChar` to the UnicodeString object. + * @param srcChar the code unit to append + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& append(char16_t srcChar); + + /** + * Append the code point `srcChar` to the UnicodeString object. + * @param srcChar the code point to append + * @return a reference to this + * @stable ICU 2.0 + */ + UnicodeString& append(UChar32 srcChar); + + + /* Insert operations */ + + /** + * Insert the characters in `srcText` in the range + * [`srcStart`, `srcStart + srcLength`) into the UnicodeString + * object at offset `start`. `srcText` is not modified. + * @param start the offset where the insertion begins + * @param srcText the source for the new characters + * @param srcStart the offset into `srcText` where new characters + * will be obtained + * @param srcLength the number of characters in `srcText` in + * the insert string + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& insert(int32_t start, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength); + + /** + * Insert the characters in `srcText` into the UnicodeString object + * at offset `start`. `srcText` is not modified. + * @param start the offset where the insertion begins + * @param srcText the source for the new characters + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& insert(int32_t start, + const UnicodeString& srcText); + + /** + * Insert the characters in `srcChars` in the range + * [`srcStart`, `srcStart + srcLength`) into the UnicodeString + * object at offset `start`. `srcChars` is not modified. + * @param start the offset at which the insertion begins + * @param srcChars the source for the new characters + * @param srcStart the offset into `srcChars` where new characters + * will be obtained + * @param srcLength the number of characters in `srcChars` + * in the insert string + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& insert(int32_t start, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength); + + /** + * Insert the characters in `srcChars` into the UnicodeString object + * at offset `start`. `srcChars` is not modified. + * @param start the offset where the insertion begins + * @param srcChars the source for the new characters + * @param srcLength the number of Unicode characters in srcChars. + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& insert(int32_t start, + ConstChar16Ptr srcChars, + int32_t srcLength); + + /** + * Insert the code unit `srcChar` into the UnicodeString object at + * offset `start`. + * @param start the offset at which the insertion occurs + * @param srcChar the code unit to insert + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& insert(int32_t start, + char16_t srcChar); + + /** + * Insert the code point `srcChar` into the UnicodeString object at + * offset `start`. + * @param start the offset at which the insertion occurs + * @param srcChar the code point to insert + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& insert(int32_t start, + UChar32 srcChar); + + + /* Replace operations */ + + /** + * Replace the characters in the range + * [`start`, `start + length`) with the characters in + * `srcText` in the range + * [`srcStart`, `srcStart + srcLength`). + * `srcText` is not modified. + * @param start the offset at which the replace operation begins + * @param length the number of characters to replace. The character at + * `start + length` is not modified. + * @param srcText the source for the new characters + * @param srcStart the offset into `srcText` where new characters + * will be obtained + * @param srcLength the number of characters in `srcText` in + * the replace string + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& replace(int32_t start, + int32_t length, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength); + + /** + * Replace the characters in the range + * [`start`, `start + length`) + * with the characters in `srcText`. `srcText` is + * not modified. + * @param start the offset at which the replace operation begins + * @param length the number of characters to replace. The character at + * `start + length` is not modified. + * @param srcText the source for the new characters + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& replace(int32_t start, + int32_t length, + const UnicodeString& srcText); + + /** + * Replace the characters in the range + * [`start`, `start + length`) with the characters in + * `srcChars` in the range + * [`srcStart`, `srcStart + srcLength`). `srcChars` + * is not modified. + * @param start the offset at which the replace operation begins + * @param length the number of characters to replace. The character at + * `start + length` is not modified. + * @param srcChars the source for the new characters + * @param srcStart the offset into `srcChars` where new characters + * will be obtained + * @param srcLength the number of characters in `srcChars` + * in the replace string + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& replace(int32_t start, + int32_t length, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength); + + /** + * Replace the characters in the range + * [`start`, `start + length`) with the characters in + * `srcChars`. `srcChars` is not modified. + * @param start the offset at which the replace operation begins + * @param length number of characters to replace. The character at + * `start + length` is not modified. + * @param srcChars the source for the new characters + * @param srcLength the number of Unicode characters in srcChars + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& replace(int32_t start, + int32_t length, + ConstChar16Ptr srcChars, + int32_t srcLength); + + /** + * Replace the characters in the range + * [`start`, `start + length`) with the code unit + * `srcChar`. + * @param start the offset at which the replace operation begins + * @param length the number of characters to replace. The character at + * `start + length` is not modified. + * @param srcChar the new code unit + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& replace(int32_t start, + int32_t length, + char16_t srcChar); + + /** + * Replace the characters in the range + * [`start`, `start + length`) with the code point + * `srcChar`. + * @param start the offset at which the replace operation begins + * @param length the number of characters to replace. The character at + * `start + length` is not modified. + * @param srcChar the new code point + * @return a reference to this + * @stable ICU 2.0 + */ + UnicodeString& replace(int32_t start, int32_t length, UChar32 srcChar); + + /** + * Replace the characters in the range [`start`, `limit`) + * with the characters in `srcText`. `srcText` is not modified. + * @param start the offset at which the replace operation begins + * @param limit the offset immediately following the replace range + * @param srcText the source for the new characters + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& replaceBetween(int32_t start, + int32_t limit, + const UnicodeString& srcText); + + /** + * Replace the characters in the range [`start`, `limit`) + * with the characters in `srcText` in the range + * [`srcStart`, `srcLimit`). `srcText` is not modified. + * @param start the offset at which the replace operation begins + * @param limit the offset immediately following the replace range + * @param srcText the source for the new characters + * @param srcStart the offset into `srcChars` where new characters + * will be obtained + * @param srcLimit the offset immediately following the range to copy + * in `srcText` + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& replaceBetween(int32_t start, + int32_t limit, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLimit); + + /** + * Replace a substring of this object with the given text. + * @param start the beginning index, inclusive; `0 <= start <= limit`. + * @param limit the ending index, exclusive; `start <= limit <= length()`. + * @param text the text to replace characters `start` to `limit - 1` + * @stable ICU 2.0 + */ + virtual void handleReplaceBetween(int32_t start, + int32_t limit, + const UnicodeString& text) override; + + /** + * Replaceable API + * @return true if it has MetaData + * @stable ICU 2.4 + */ + virtual UBool hasMetaData() const override; + + /** + * Copy a substring of this object, retaining attribute (out-of-band) + * information. This method is used to duplicate or reorder substrings. + * The destination index must not overlap the source range. + * + * @param start the beginning index, inclusive; `0 <= start <= limit`. + * @param limit the ending index, exclusive; `start <= limit <= length()`. + * @param dest the destination index. The characters from + * `start..limit-1` will be copied to `dest`. + * Implementations of this method may assume that `dest <= start || + * dest >= limit`. + * @stable ICU 2.0 + */ + virtual void copy(int32_t start, int32_t limit, int32_t dest) override; + + /* Search and replace operations */ + + /** + * Replace all occurrences of characters in oldText with the characters + * in newText + * @param oldText the text containing the search text + * @param newText the text containing the replacement text + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& findAndReplace(const UnicodeString& oldText, + const UnicodeString& newText); + + /** + * Replace all occurrences of characters in oldText with characters + * in newText + * in the range [`start`, `start + length`). + * @param start the start of the range in which replace will performed + * @param length the length of the range in which replace will be performed + * @param oldText the text containing the search text + * @param newText the text containing the replacement text + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& findAndReplace(int32_t start, + int32_t length, + const UnicodeString& oldText, + const UnicodeString& newText); + + /** + * Replace all occurrences of characters in oldText in the range + * [`oldStart`, `oldStart + oldLength`) with the characters + * in newText in the range + * [`newStart`, `newStart + newLength`) + * in the range [`start`, `start + length`). + * @param start the start of the range in which replace will performed + * @param length the length of the range in which replace will be performed + * @param oldText the text containing the search text + * @param oldStart the start of the search range in `oldText` + * @param oldLength the length of the search range in `oldText` + * @param newText the text containing the replacement text + * @param newStart the start of the replacement range in `newText` + * @param newLength the length of the replacement range in `newText` + * @return a reference to this + * @stable ICU 2.0 + */ + UnicodeString& findAndReplace(int32_t start, + int32_t length, + const UnicodeString& oldText, + int32_t oldStart, + int32_t oldLength, + const UnicodeString& newText, + int32_t newStart, + int32_t newLength); + + + /* Remove operations */ + + /** + * Removes all characters from the UnicodeString object and clears the bogus flag. + * This is the UnicodeString equivalent of std::string’s clear(). + * + * @return a reference to this + * @see setToBogus + * @stable ICU 2.0 + */ + inline UnicodeString& remove(); + + /** + * Remove the characters in the range + * [`start`, `start + length`) from the UnicodeString object. + * @param start the offset of the first character to remove + * @param length the number of characters to remove + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& remove(int32_t start, + int32_t length = (int32_t)INT32_MAX); + + /** + * Remove the characters in the range + * [`start`, `limit`) from the UnicodeString object. + * @param start the offset of the first character to remove + * @param limit the offset immediately following the range to remove + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& removeBetween(int32_t start, + int32_t limit = (int32_t)INT32_MAX); + + /** + * Retain only the characters in the range + * [`start`, `limit`) from the UnicodeString object. + * Removes characters before `start` and at and after `limit`. + * @param start the offset of the first character to retain + * @param limit the offset immediately following the range to retain + * @return a reference to this + * @stable ICU 4.4 + */ + inline UnicodeString &retainBetween(int32_t start, int32_t limit = INT32_MAX); + + /* Length operations */ + + /** + * Pad the start of this UnicodeString with the character `padChar`. + * If the length of this UnicodeString is less than targetLength, + * length() - targetLength copies of padChar will be added to the + * beginning of this UnicodeString. + * @param targetLength the desired length of the string + * @param padChar the character to use for padding. Defaults to + * space (U+0020) + * @return true if the text was padded, false otherwise. + * @stable ICU 2.0 + */ + UBool padLeading(int32_t targetLength, + char16_t padChar = 0x0020); + + /** + * Pad the end of this UnicodeString with the character `padChar`. + * If the length of this UnicodeString is less than targetLength, + * length() - targetLength copies of padChar will be added to the + * end of this UnicodeString. + * @param targetLength the desired length of the string + * @param padChar the character to use for padding. Defaults to + * space (U+0020) + * @return true if the text was padded, false otherwise. + * @stable ICU 2.0 + */ + UBool padTrailing(int32_t targetLength, + char16_t padChar = 0x0020); + + /** + * Truncate this UnicodeString to the `targetLength`. + * @param targetLength the desired length of this UnicodeString. + * @return true if the text was truncated, false otherwise + * @stable ICU 2.0 + */ + inline UBool truncate(int32_t targetLength); + + /** + * Trims leading and trailing whitespace from this UnicodeString. + * @return a reference to this + * @stable ICU 2.0 + */ + UnicodeString& trim(void); + + + /* Miscellaneous operations */ + + /** + * Reverse this UnicodeString in place. + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& reverse(void); + + /** + * Reverse the range [`start`, `start + length`) in + * this UnicodeString. + * @param start the start of the range to reverse + * @param length the number of characters to to reverse + * @return a reference to this + * @stable ICU 2.0 + */ + inline UnicodeString& reverse(int32_t start, + int32_t length); + + /** + * Convert the characters in this to UPPER CASE following the conventions of + * the default locale. + * @return A reference to this. + * @stable ICU 2.0 + */ + UnicodeString& toUpper(void); + + /** + * Convert the characters in this to UPPER CASE following the conventions of + * a specific locale. + * @param locale The locale containing the conventions to use. + * @return A reference to this. + * @stable ICU 2.0 + */ + UnicodeString& toUpper(const Locale& locale); + + /** + * Convert the characters in this to lower case following the conventions of + * the default locale. + * @return A reference to this. + * @stable ICU 2.0 + */ + UnicodeString& toLower(void); + + /** + * Convert the characters in this to lower case following the conventions of + * a specific locale. + * @param locale The locale containing the conventions to use. + * @return A reference to this. + * @stable ICU 2.0 + */ + UnicodeString& toLower(const Locale& locale); + +#if !UCONFIG_NO_BREAK_ITERATION + + /** + * Titlecase this string, convenience function using the default locale. + * + * Casing is locale-dependent and context-sensitive. + * Titlecasing uses a break iterator to find the first characters of words + * that are to be titlecased. It titlecases those characters and lowercases + * all others. + * + * The titlecase break iterator can be provided to customize for arbitrary + * styles, using rules and dictionaries beyond the standard iterators. + * It may be more efficient to always provide an iterator to avoid + * opening and closing one for each string. + * The standard titlecase iterator for the root locale implements the + * algorithm of Unicode TR 21. + * + * This function uses only the setText(), first() and next() methods of the + * provided break iterator. + * + * @param titleIter A break iterator to find the first characters of words + * that are to be titlecased. + * If none is provided (0), then a standard titlecase + * break iterator is opened. + * Otherwise the provided iterator is set to the string's text. + * @return A reference to this. + * @stable ICU 2.1 + */ + UnicodeString &toTitle(BreakIterator *titleIter); + + /** + * Titlecase this string. + * + * Casing is locale-dependent and context-sensitive. + * Titlecasing uses a break iterator to find the first characters of words + * that are to be titlecased. It titlecases those characters and lowercases + * all others. + * + * The titlecase break iterator can be provided to customize for arbitrary + * styles, using rules and dictionaries beyond the standard iterators. + * It may be more efficient to always provide an iterator to avoid + * opening and closing one for each string. + * The standard titlecase iterator for the root locale implements the + * algorithm of Unicode TR 21. + * + * This function uses only the setText(), first() and next() methods of the + * provided break iterator. + * + * @param titleIter A break iterator to find the first characters of words + * that are to be titlecased. + * If none is provided (0), then a standard titlecase + * break iterator is opened. + * Otherwise the provided iterator is set to the string's text. + * @param locale The locale to consider. + * @return A reference to this. + * @stable ICU 2.1 + */ + UnicodeString &toTitle(BreakIterator *titleIter, const Locale &locale); + + /** + * Titlecase this string, with options. + * + * Casing is locale-dependent and context-sensitive. + * Titlecasing uses a break iterator to find the first characters of words + * that are to be titlecased. It titlecases those characters and lowercases + * all others. (This can be modified with options.) + * + * The titlecase break iterator can be provided to customize for arbitrary + * styles, using rules and dictionaries beyond the standard iterators. + * It may be more efficient to always provide an iterator to avoid + * opening and closing one for each string. + * The standard titlecase iterator for the root locale implements the + * algorithm of Unicode TR 21. + * + * This function uses only the setText(), first() and next() methods of the + * provided break iterator. + * + * @param titleIter A break iterator to find the first characters of words + * that are to be titlecased. + * If none is provided (0), then a standard titlecase + * break iterator is opened. + * Otherwise the provided iterator is set to the string's text. + * @param locale The locale to consider. + * @param options Options bit set, usually 0. See U_TITLECASE_NO_LOWERCASE, + * U_TITLECASE_NO_BREAK_ADJUSTMENT, U_TITLECASE_ADJUST_TO_CASED, + * U_TITLECASE_WHOLE_STRING, U_TITLECASE_SENTENCES. + * @return A reference to this. + * @stable ICU 3.8 + */ + UnicodeString &toTitle(BreakIterator *titleIter, const Locale &locale, uint32_t options); + +#endif + + /** + * Case-folds the characters in this string. + * + * Case-folding is locale-independent and not context-sensitive, + * but there is an option for whether to include or exclude mappings for dotted I + * and dotless i that are marked with 'T' in CaseFolding.txt. + * + * The result may be longer or shorter than the original. + * + * @param options Either U_FOLD_CASE_DEFAULT or U_FOLD_CASE_EXCLUDE_SPECIAL_I + * @return A reference to this. + * @stable ICU 2.0 + */ + UnicodeString &foldCase(uint32_t options=0 /*U_FOLD_CASE_DEFAULT*/); + + //======================================== + // Access to the internal buffer + //======================================== + + /** + * Get a read/write pointer to the internal buffer. + * The buffer is guaranteed to be large enough for at least minCapacity char16_ts, + * writable, and is still owned by the UnicodeString object. + * Calls to getBuffer(minCapacity) must not be nested, and + * must be matched with calls to releaseBuffer(newLength). + * If the string buffer was read-only or shared, + * then it will be reallocated and copied. + * + * An attempted nested call will return 0, and will not further modify the + * state of the UnicodeString object. + * It also returns 0 if the string is bogus. + * + * The actual capacity of the string buffer may be larger than minCapacity. + * getCapacity() returns the actual capacity. + * For many operations, the full capacity should be used to avoid reallocations. + * + * While the buffer is "open" between getBuffer(minCapacity) + * and releaseBuffer(newLength), the following applies: + * - The string length is set to 0. + * - Any read API call on the UnicodeString object will behave like on a 0-length string. + * - Any write API call on the UnicodeString object is disallowed and will have no effect. + * - You can read from and write to the returned buffer. + * - The previous string contents will still be in the buffer; + * if you want to use it, then you need to call length() before getBuffer(minCapacity). + * If the length() was greater than minCapacity, then any contents after minCapacity + * may be lost. + * The buffer contents is not NUL-terminated by getBuffer(). + * If length() < getCapacity() then you can terminate it by writing a NUL + * at index length(). + * - You must call releaseBuffer(newLength) before and in order to + * return to normal UnicodeString operation. + * + * @param minCapacity the minimum number of char16_ts that are to be available + * in the buffer, starting at the returned pointer; + * default to the current string capacity if minCapacity==-1 + * @return a writable pointer to the internal string buffer, + * or nullptr if an error occurs (nested calls, out of memory) + * + * @see releaseBuffer + * @see getTerminatedBuffer() + * @stable ICU 2.0 + */ + char16_t *getBuffer(int32_t minCapacity); + + /** + * Release a read/write buffer on a UnicodeString object with an + * "open" getBuffer(minCapacity). + * This function must be called in a matched pair with getBuffer(minCapacity). + * releaseBuffer(newLength) must be called if and only if a getBuffer(minCapacity) is "open". + * + * It will set the string length to newLength, at most to the current capacity. + * If newLength==-1 then it will set the length according to the + * first NUL in the buffer, or to the capacity if there is no NUL. + * + * After calling releaseBuffer(newLength) the UnicodeString is back to normal operation. + * + * @param newLength the new length of the UnicodeString object; + * defaults to the current capacity if newLength is greater than that; + * if newLength==-1, it defaults to u_strlen(buffer) but not more than + * the current capacity of the string + * + * @see getBuffer(int32_t minCapacity) + * @stable ICU 2.0 + */ + void releaseBuffer(int32_t newLength=-1); + + /** + * Get a read-only pointer to the internal buffer. + * This can be called at any time on a valid UnicodeString. + * + * It returns 0 if the string is bogus, or + * during an "open" getBuffer(minCapacity). + * + * It can be called as many times as desired. + * The pointer that it returns will remain valid until the UnicodeString object is modified, + * at which time the pointer is semantically invalidated and must not be used any more. + * + * The capacity of the buffer can be determined with getCapacity(). + * The part after length() may or may not be initialized and valid, + * depending on the history of the UnicodeString object. + * + * The buffer contents is (probably) not NUL-terminated. + * You can check if it is with + * `(s.length() < s.getCapacity() && buffer[s.length()]==0)`. + * (See getTerminatedBuffer().) + * + * The buffer may reside in read-only memory. Its contents must not + * be modified. + * + * @return a read-only pointer to the internal string buffer, + * or nullptr if the string is empty or bogus + * + * @see getBuffer(int32_t minCapacity) + * @see getTerminatedBuffer() + * @stable ICU 2.0 + */ + inline const char16_t *getBuffer() const; + + /** + * Get a read-only pointer to the internal buffer, + * making sure that it is NUL-terminated. + * This can be called at any time on a valid UnicodeString. + * + * It returns 0 if the string is bogus, or + * during an "open" getBuffer(minCapacity), or if the buffer cannot + * be NUL-terminated (because memory allocation failed). + * + * It can be called as many times as desired. + * The pointer that it returns will remain valid until the UnicodeString object is modified, + * at which time the pointer is semantically invalidated and must not be used any more. + * + * The capacity of the buffer can be determined with getCapacity(). + * The part after length()+1 may or may not be initialized and valid, + * depending on the history of the UnicodeString object. + * + * The buffer contents is guaranteed to be NUL-terminated. + * getTerminatedBuffer() may reallocate the buffer if a terminating NUL + * is written. + * For this reason, this function is not const, unlike getBuffer(). + * Note that a UnicodeString may also contain NUL characters as part of its contents. + * + * The buffer may reside in read-only memory. Its contents must not + * be modified. + * + * @return a read-only pointer to the internal string buffer, + * or 0 if the string is empty or bogus + * + * @see getBuffer(int32_t minCapacity) + * @see getBuffer() + * @stable ICU 2.2 + */ + const char16_t *getTerminatedBuffer(); + + //======================================== + // Constructors + //======================================== + + /** Construct an empty UnicodeString. + * @stable ICU 2.0 + */ + inline UnicodeString(); + + /** + * Construct a UnicodeString with capacity to hold `capacity` char16_ts + * @param capacity the number of char16_ts this UnicodeString should hold + * before a resize is necessary; if count is greater than 0 and count + * code points c take up more space than capacity, then capacity is adjusted + * accordingly. + * @param c is used to initially fill the string + * @param count specifies how many code points c are to be written in the + * string + * @stable ICU 2.0 + */ + UnicodeString(int32_t capacity, UChar32 c, int32_t count); + + /** + * Single char16_t (code unit) constructor. + * + * It is recommended to mark this constructor "explicit" by + * `-DUNISTR_FROM_CHAR_EXPLICIT=explicit` + * on the compiler command line or similar. + * @param ch the character to place in the UnicodeString + * @stable ICU 2.0 + */ + UNISTR_FROM_CHAR_EXPLICIT UnicodeString(char16_t ch); + + /** + * Single UChar32 (code point) constructor. + * + * It is recommended to mark this constructor "explicit" by + * `-DUNISTR_FROM_CHAR_EXPLICIT=explicit` + * on the compiler command line or similar. + * @param ch the character to place in the UnicodeString + * @stable ICU 2.0 + */ + UNISTR_FROM_CHAR_EXPLICIT UnicodeString(UChar32 ch); + + /** + * char16_t* constructor. + * + * It is recommended to mark this constructor "explicit" by + * `-DUNISTR_FROM_STRING_EXPLICIT=explicit` + * on the compiler command line or similar. + * @param text The characters to place in the UnicodeString. `text` + * must be NUL (U+0000) terminated. + * @stable ICU 2.0 + */ + UNISTR_FROM_STRING_EXPLICIT UnicodeString(const char16_t *text); + +#if !U_CHAR16_IS_TYPEDEF + /** + * uint16_t * constructor. + * Delegates to UnicodeString(const char16_t *). + * + * It is recommended to mark this constructor "explicit" by + * `-DUNISTR_FROM_STRING_EXPLICIT=explicit` + * on the compiler command line or similar. + * @param text NUL-terminated UTF-16 string + * @stable ICU 59 + */ + UNISTR_FROM_STRING_EXPLICIT UnicodeString(const uint16_t *text) : + UnicodeString(ConstChar16Ptr(text)) {} +#endif + +#if U_SIZEOF_WCHAR_T==2 || defined(U_IN_DOXYGEN) + /** + * wchar_t * constructor. + * (Only defined if U_SIZEOF_WCHAR_T==2.) + * Delegates to UnicodeString(const char16_t *). + * + * It is recommended to mark this constructor "explicit" by + * `-DUNISTR_FROM_STRING_EXPLICIT=explicit` + * on the compiler command line or similar. + * @param text NUL-terminated UTF-16 string + * @stable ICU 59 + */ + UNISTR_FROM_STRING_EXPLICIT UnicodeString(const wchar_t *text) : + UnicodeString(ConstChar16Ptr(text)) {} +#endif + + /** + * nullptr_t constructor. + * Effectively the same as the default constructor, makes an empty string object. + * + * It is recommended to mark this constructor "explicit" by + * `-DUNISTR_FROM_STRING_EXPLICIT=explicit` + * on the compiler command line or similar. + * @param text nullptr + * @stable ICU 59 + */ + UNISTR_FROM_STRING_EXPLICIT inline UnicodeString(const std::nullptr_t text); + + /** + * char16_t* constructor. + * @param text The characters to place in the UnicodeString. + * @param textLength The number of Unicode characters in `text` + * to copy. + * @stable ICU 2.0 + */ + UnicodeString(const char16_t *text, + int32_t textLength); + +#if !U_CHAR16_IS_TYPEDEF + /** + * uint16_t * constructor. + * Delegates to UnicodeString(const char16_t *, int32_t). + * @param text UTF-16 string + * @param textLength string length + * @stable ICU 59 + */ + UnicodeString(const uint16_t *text, int32_t textLength) : + UnicodeString(ConstChar16Ptr(text), textLength) {} +#endif + +#if U_SIZEOF_WCHAR_T==2 || defined(U_IN_DOXYGEN) + /** + * wchar_t * constructor. + * (Only defined if U_SIZEOF_WCHAR_T==2.) + * Delegates to UnicodeString(const char16_t *, int32_t). + * @param text NUL-terminated UTF-16 string + * @param textLength string length + * @stable ICU 59 + */ + UnicodeString(const wchar_t *text, int32_t textLength) : + UnicodeString(ConstChar16Ptr(text), textLength) {} +#endif + + /** + * nullptr_t constructor. + * Effectively the same as the default constructor, makes an empty string object. + * @param text nullptr + * @param textLength ignored + * @stable ICU 59 + */ + inline UnicodeString(const std::nullptr_t text, int32_t textLength); + + /** + * Readonly-aliasing char16_t* constructor. + * The text will be used for the UnicodeString object, but + * it will not be released when the UnicodeString is destroyed. + * This has copy-on-write semantics: + * When the string is modified, then the buffer is first copied into + * newly allocated memory. + * The aliased buffer is never modified. + * + * In an assignment to another UnicodeString, when using the copy constructor + * or the assignment operator, the text will be copied. + * When using fastCopyFrom(), the text will be aliased again, + * so that both strings then alias the same readonly-text. + * + * @param isTerminated specifies if `text` is `NUL`-terminated. + * This must be true if `textLength==-1`. + * @param text The characters to alias for the UnicodeString. + * @param textLength The number of Unicode characters in `text` to alias. + * If -1, then this constructor will determine the length + * by calling `u_strlen()`. + * @stable ICU 2.0 + */ + UnicodeString(UBool isTerminated, + ConstChar16Ptr text, + int32_t textLength); + + /** + * Writable-aliasing char16_t* constructor. + * The text will be used for the UnicodeString object, but + * it will not be released when the UnicodeString is destroyed. + * This has write-through semantics: + * For as long as the capacity of the buffer is sufficient, write operations + * will directly affect the buffer. When more capacity is necessary, then + * a new buffer will be allocated and the contents copied as with regularly + * constructed strings. + * In an assignment to another UnicodeString, the buffer will be copied. + * The extract(Char16Ptr dst) function detects whether the dst pointer is the same + * as the string buffer itself and will in this case not copy the contents. + * + * @param buffer The characters to alias for the UnicodeString. + * @param buffLength The number of Unicode characters in `buffer` to alias. + * @param buffCapacity The size of `buffer` in char16_ts. + * @stable ICU 2.0 + */ + UnicodeString(char16_t *buffer, int32_t buffLength, int32_t buffCapacity); + +#if !U_CHAR16_IS_TYPEDEF + /** + * Writable-aliasing uint16_t * constructor. + * Delegates to UnicodeString(const char16_t *, int32_t, int32_t). + * @param buffer writable buffer of/for UTF-16 text + * @param buffLength length of the current buffer contents + * @param buffCapacity buffer capacity + * @stable ICU 59 + */ + UnicodeString(uint16_t *buffer, int32_t buffLength, int32_t buffCapacity) : + UnicodeString(Char16Ptr(buffer), buffLength, buffCapacity) {} +#endif + +#if U_SIZEOF_WCHAR_T==2 || defined(U_IN_DOXYGEN) + /** + * Writable-aliasing wchar_t * constructor. + * (Only defined if U_SIZEOF_WCHAR_T==2.) + * Delegates to UnicodeString(const char16_t *, int32_t, int32_t). + * @param buffer writable buffer of/for UTF-16 text + * @param buffLength length of the current buffer contents + * @param buffCapacity buffer capacity + * @stable ICU 59 + */ + UnicodeString(wchar_t *buffer, int32_t buffLength, int32_t buffCapacity) : + UnicodeString(Char16Ptr(buffer), buffLength, buffCapacity) {} +#endif + + /** + * Writable-aliasing nullptr_t constructor. + * Effectively the same as the default constructor, makes an empty string object. + * @param buffer nullptr + * @param buffLength ignored + * @param buffCapacity ignored + * @stable ICU 59 + */ + inline UnicodeString(std::nullptr_t buffer, int32_t buffLength, int32_t buffCapacity); + +#if U_CHARSET_IS_UTF8 || !UCONFIG_NO_CONVERSION + + /** + * char* constructor. + * Uses the default converter (and thus depends on the ICU conversion code) + * unless U_CHARSET_IS_UTF8 is set to 1. + * + * For ASCII (really "invariant character") strings it is more efficient to use + * the constructor that takes a US_INV (for its enum EInvariant). + * For ASCII (invariant-character) string literals, see UNICODE_STRING and + * UNICODE_STRING_SIMPLE. + * + * It is recommended to mark this constructor "explicit" by + * `-DUNISTR_FROM_STRING_EXPLICIT=explicit` + * on the compiler command line or similar. + * @param codepageData an array of bytes, null-terminated, + * in the platform's default codepage. + * @stable ICU 2.0 + * @see UNICODE_STRING + * @see UNICODE_STRING_SIMPLE + */ + UNISTR_FROM_STRING_EXPLICIT UnicodeString(const char *codepageData); + + /** + * char* constructor. + * Uses the default converter (and thus depends on the ICU conversion code) + * unless U_CHARSET_IS_UTF8 is set to 1. + * @param codepageData an array of bytes in the platform's default codepage. + * @param dataLength The number of bytes in `codepageData`. + * @stable ICU 2.0 + */ + UnicodeString(const char *codepageData, int32_t dataLength); + +#endif + +#if !UCONFIG_NO_CONVERSION + + /** + * char* constructor. + * @param codepageData an array of bytes, null-terminated + * @param codepage the encoding of `codepageData`. The special + * value 0 for `codepage` indicates that the text is in the + * platform's default codepage. + * + * If `codepage` is an empty string (`""`), + * then a simple conversion is performed on the codepage-invariant + * subset ("invariant characters") of the platform encoding. See utypes.h. + * Recommendation: For invariant-character strings use the constructor + * UnicodeString(const char *src, int32_t length, enum EInvariant inv) + * because it avoids object code dependencies of UnicodeString on + * the conversion code. + * + * @stable ICU 2.0 + */ + UnicodeString(const char *codepageData, const char *codepage); + + /** + * char* constructor. + * @param codepageData an array of bytes. + * @param dataLength The number of bytes in `codepageData`. + * @param codepage the encoding of `codepageData`. The special + * value 0 for `codepage` indicates that the text is in the + * platform's default codepage. + * If `codepage` is an empty string (`""`), + * then a simple conversion is performed on the codepage-invariant + * subset ("invariant characters") of the platform encoding. See utypes.h. + * Recommendation: For invariant-character strings use the constructor + * UnicodeString(const char *src, int32_t length, enum EInvariant inv) + * because it avoids object code dependencies of UnicodeString on + * the conversion code. + * + * @stable ICU 2.0 + */ + UnicodeString(const char *codepageData, int32_t dataLength, const char *codepage); + + /** + * char * / UConverter constructor. + * This constructor uses an existing UConverter object to + * convert the codepage string to Unicode and construct a UnicodeString + * from that. + * + * The converter is reset at first. + * If the error code indicates a failure before this constructor is called, + * or if an error occurs during conversion or construction, + * then the string will be bogus. + * + * This function avoids the overhead of opening and closing a converter if + * multiple strings are constructed. + * + * @param src input codepage string + * @param srcLength length of the input string, can be -1 for NUL-terminated strings + * @param cnv converter object (ucnv_resetToUnicode() will be called), + * can be nullptr for the default converter + * @param errorCode normal ICU error code + * @stable ICU 2.0 + */ + UnicodeString( + const char *src, int32_t srcLength, + UConverter *cnv, + UErrorCode &errorCode); + +#endif + + /** + * Constructs a Unicode string from an invariant-character char * string. + * About invariant characters see utypes.h. + * This constructor has no runtime dependency on conversion code and is + * therefore recommended over ones taking a charset name string + * (where the empty string "" indicates invariant-character conversion). + * + * Use the macro US_INV as the third, signature-distinguishing parameter. + * + * For example: + * \code + * void fn(const char *s) { + * UnicodeString ustr(s, -1, US_INV); + * // use ustr ... + * } + * \endcode + * @param src String using only invariant characters. + * @param textLength Length of src, or -1 if NUL-terminated. + * @param inv Signature-distinguishing parameter, use US_INV. + * + * @see US_INV + * @stable ICU 3.2 + */ + UnicodeString(const char *src, int32_t textLength, enum EInvariant inv); + + + /** + * Copy constructor. + * + * Starting with ICU 2.4, the assignment operator and the copy constructor + * allocate a new buffer and copy the buffer contents even for readonly aliases. + * By contrast, the fastCopyFrom() function implements the old, + * more efficient but less safe behavior + * of making this string also a readonly alias to the same buffer. + * + * If the source object has an "open" buffer from getBuffer(minCapacity), + * then the copy is an empty string. + * + * @param that The UnicodeString object to copy. + * @stable ICU 2.0 + * @see fastCopyFrom + */ + UnicodeString(const UnicodeString& that); + + /** + * Move constructor; might leave src in bogus state. + * This string will have the same contents and state that the source string had. + * @param src source string + * @stable ICU 56 + */ + UnicodeString(UnicodeString &&src) noexcept; + + /** + * 'Substring' constructor from tail of source string. + * @param src The UnicodeString object to copy. + * @param srcStart The offset into `src` at which to start copying. + * @stable ICU 2.2 + */ + UnicodeString(const UnicodeString& src, int32_t srcStart); + + /** + * 'Substring' constructor from subrange of source string. + * @param src The UnicodeString object to copy. + * @param srcStart The offset into `src` at which to start copying. + * @param srcLength The number of characters from `src` to copy. + * @stable ICU 2.2 + */ + UnicodeString(const UnicodeString& src, int32_t srcStart, int32_t srcLength); + + /** + * Clone this object, an instance of a subclass of Replaceable. + * Clones can be used concurrently in multiple threads. + * If a subclass does not implement clone(), or if an error occurs, + * then nullptr is returned. + * The caller must delete the clone. + * + * @return a clone of this object + * + * @see Replaceable::clone + * @see getDynamicClassID + * @stable ICU 2.6 + */ + virtual UnicodeString *clone() const override; + + /** Destructor. + * @stable ICU 2.0 + */ + virtual ~UnicodeString(); + + /** + * Create a UnicodeString from a UTF-8 string. + * Illegal input is replaced with U+FFFD. Otherwise, errors result in a bogus string. + * Calls u_strFromUTF8WithSub(). + * + * @param utf8 UTF-8 input string. + * Note that a StringPiece can be implicitly constructed + * from a std::string or a NUL-terminated const char * string. + * @return A UnicodeString with equivalent UTF-16 contents. + * @see toUTF8 + * @see toUTF8String + * @stable ICU 4.2 + */ + static UnicodeString fromUTF8(StringPiece utf8); + + /** + * Create a UnicodeString from a UTF-32 string. + * Illegal input is replaced with U+FFFD. Otherwise, errors result in a bogus string. + * Calls u_strFromUTF32WithSub(). + * + * @param utf32 UTF-32 input string. Must not be nullptr. + * @param length Length of the input string, or -1 if NUL-terminated. + * @return A UnicodeString with equivalent UTF-16 contents. + * @see toUTF32 + * @stable ICU 4.2 + */ + static UnicodeString fromUTF32(const UChar32 *utf32, int32_t length); + + /* Miscellaneous operations */ + + /** + * Unescape a string of characters and return a string containing + * the result. The following escape sequences are recognized: + * + * \\uhhhh 4 hex digits; h in [0-9A-Fa-f] + * \\Uhhhhhhhh 8 hex digits + * \\xhh 1-2 hex digits + * \\ooo 1-3 octal digits; o in [0-7] + * \\cX control-X; X is masked with 0x1F + * + * as well as the standard ANSI C escapes: + * + * \\a => U+0007, \\b => U+0008, \\t => U+0009, \\n => U+000A, + * \\v => U+000B, \\f => U+000C, \\r => U+000D, \\e => U+001B, + * \\" => U+0022, \\' => U+0027, \\? => U+003F, \\\\ => U+005C + * + * Anything else following a backslash is generically escaped. For + * example, "[a\\-z]" returns "[a-z]". + * + * If an escape sequence is ill-formed, this method returns an empty + * string. An example of an ill-formed sequence is "\\u" followed by + * fewer than 4 hex digits. + * + * This function is similar to u_unescape() but not identical to it. + * The latter takes a source char*, so it does escape recognition + * and also invariant conversion. + * + * @return a string with backslash escapes interpreted, or an + * empty string on error. + * @see UnicodeString#unescapeAt() + * @see u_unescape() + * @see u_unescapeAt() + * @stable ICU 2.0 + */ + UnicodeString unescape() const; + + /** + * Unescape a single escape sequence and return the represented + * character. See unescape() for a listing of the recognized escape + * sequences. The character at offset-1 is assumed (without + * checking) to be a backslash. If the escape sequence is + * ill-formed, or the offset is out of range, U_SENTINEL=-1 is + * returned. + * + * @param offset an input output parameter. On input, it is the + * offset into this string where the escape sequence is located, + * after the initial backslash. On output, it is advanced after the + * last character parsed. On error, it is not advanced at all. + * @return the character represented by the escape sequence at + * offset, or U_SENTINEL=-1 on error. + * @see UnicodeString#unescape() + * @see u_unescape() + * @see u_unescapeAt() + * @stable ICU 2.0 + */ + UChar32 unescapeAt(int32_t &offset) const; + + /** + * ICU "poor man's RTTI", returns a UClassID for this class. + * + * @stable ICU 2.2 + */ + static UClassID U_EXPORT2 getStaticClassID(); + + /** + * ICU "poor man's RTTI", returns a UClassID for the actual class. + * + * @stable ICU 2.2 + */ + virtual UClassID getDynamicClassID() const override; + + //======================================== + // Implementation methods + //======================================== + +protected: + /** + * Implement Replaceable::getLength() (see jitterbug 1027). + * @stable ICU 2.4 + */ + virtual int32_t getLength() const override; + + /** + * The change in Replaceable to use virtual getCharAt() allows + * UnicodeString::charAt() to be inline again (see jitterbug 709). + * @stable ICU 2.4 + */ + virtual char16_t getCharAt(int32_t offset) const override; + + /** + * The change in Replaceable to use virtual getChar32At() allows + * UnicodeString::char32At() to be inline again (see jitterbug 709). + * @stable ICU 2.4 + */ + virtual UChar32 getChar32At(int32_t offset) const override; + +private: + // For char* constructors. Could be made public. + UnicodeString &setToUTF8(StringPiece utf8); + // For extract(char*). + // We could make a toUTF8(target, capacity, errorCode) public but not + // this version: New API will be cleaner if we make callers create substrings + // rather than having start+length on every method, + // and it should take a UErrorCode&. + int32_t + toUTF8(int32_t start, int32_t len, + char *target, int32_t capacity) const; + + /** + * Internal string contents comparison, called by operator==. + * Requires: this & text not bogus and have same lengths. + */ + UBool doEquals(const UnicodeString &text, int32_t len) const; + + inline UBool + doEqualsSubstring(int32_t start, + int32_t length, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const; + + UBool doEqualsSubstring(int32_t start, + int32_t length, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength) const; + + inline int8_t + doCompare(int32_t start, + int32_t length, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const; + + int8_t doCompare(int32_t start, + int32_t length, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength) const; + + inline int8_t + doCompareCodePointOrder(int32_t start, + int32_t length, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const; + + int8_t doCompareCodePointOrder(int32_t start, + int32_t length, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength) const; + + inline int8_t + doCaseCompare(int32_t start, + int32_t length, + const UnicodeString &srcText, + int32_t srcStart, + int32_t srcLength, + uint32_t options) const; + + int8_t + doCaseCompare(int32_t start, + int32_t length, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength, + uint32_t options) const; + + int32_t doIndexOf(char16_t c, + int32_t start, + int32_t length) const; + + int32_t doIndexOf(UChar32 c, + int32_t start, + int32_t length) const; + + int32_t doLastIndexOf(char16_t c, + int32_t start, + int32_t length) const; + + int32_t doLastIndexOf(UChar32 c, + int32_t start, + int32_t length) const; + + void doExtract(int32_t start, + int32_t length, + char16_t *dst, + int32_t dstStart) const; + + inline void doExtract(int32_t start, + int32_t length, + UnicodeString& target) const; + + inline char16_t doCharAt(int32_t offset) const; + + UnicodeString& doReplace(int32_t start, + int32_t length, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength); + + UnicodeString& doReplace(int32_t start, + int32_t length, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength); + + UnicodeString& doAppend(const UnicodeString& src, int32_t srcStart, int32_t srcLength); + UnicodeString& doAppend(const char16_t *srcChars, int32_t srcStart, int32_t srcLength); + + UnicodeString& doReverse(int32_t start, + int32_t length); + + // calculate hash code + int32_t doHashCode(void) const; + + // get pointer to start of array + // these do not check for kOpenGetBuffer, unlike the public getBuffer() function + inline char16_t* getArrayStart(void); + inline const char16_t* getArrayStart(void) const; + + inline UBool hasShortLength() const; + inline int32_t getShortLength() const; + + // A UnicodeString object (not necessarily its current buffer) + // is writable unless it isBogus() or it has an "open" getBuffer(minCapacity). + inline UBool isWritable() const; + + // Is the current buffer writable? + inline UBool isBufferWritable() const; + + // None of the following does releaseArray(). + inline void setZeroLength(); + inline void setShortLength(int32_t len); + inline void setLength(int32_t len); + inline void setToEmpty(); + inline void setArray(char16_t *array, int32_t len, int32_t capacity); // sets length but not flags + + // allocate the array; result may be the stack buffer + // sets refCount to 1 if appropriate + // sets fArray, fCapacity, and flags + // sets length to 0 + // returns boolean for success or failure + UBool allocate(int32_t capacity); + + // release the array if owned + void releaseArray(void); + + // turn a bogus string into an empty one + void unBogus(); + + // implements assignment operator, copy constructor, and fastCopyFrom() + UnicodeString ©From(const UnicodeString &src, UBool fastCopy=false); + + // Copies just the fields without memory management. + void copyFieldsFrom(UnicodeString &src, UBool setSrcToBogus) noexcept; + + // Pin start and limit to acceptable values. + inline void pinIndex(int32_t& start) const; + inline void pinIndices(int32_t& start, + int32_t& length) const; + +#if !UCONFIG_NO_CONVERSION + + /* Internal extract() using UConverter. */ + int32_t doExtract(int32_t start, int32_t length, + char *dest, int32_t destCapacity, + UConverter *cnv, + UErrorCode &errorCode) const; + + /* + * Real constructor for converting from codepage data. + * It assumes that it is called with !fRefCounted. + * + * If `codepage==0`, then the default converter + * is used for the platform encoding. + * If `codepage` is an empty string (`""`), + * then a simple conversion is performed on the codepage-invariant + * subset ("invariant characters") of the platform encoding. See utypes.h. + */ + void doCodepageCreate(const char *codepageData, + int32_t dataLength, + const char *codepage); + + /* + * Worker function for creating a UnicodeString from + * a codepage string using a UConverter. + */ + void + doCodepageCreate(const char *codepageData, + int32_t dataLength, + UConverter *converter, + UErrorCode &status); + +#endif + + /* + * This function is called when write access to the array + * is necessary. + * + * We need to make a copy of the array if + * the buffer is read-only, or + * the buffer is refCounted (shared), and refCount>1, or + * the buffer is too small. + * + * Return false if memory could not be allocated. + */ + UBool cloneArrayIfNeeded(int32_t newCapacity = -1, + int32_t growCapacity = -1, + UBool doCopyArray = true, + int32_t **pBufferToDelete = 0, + UBool forceClone = false); + + /** + * Common function for UnicodeString case mappings. + * The stringCaseMapper has the same type UStringCaseMapper + * as in ustr_imp.h for ustrcase_map(). + */ + UnicodeString & + caseMap(int32_t caseLocale, uint32_t options, +#if !UCONFIG_NO_BREAK_ITERATION + BreakIterator *iter, +#endif + UStringCaseMapper *stringCaseMapper); + + // ref counting + void addRef(void); + int32_t removeRef(void); + int32_t refCount(void) const; + + // constants + enum { + /** + * Size of stack buffer for short strings. + * Must be at least U16_MAX_LENGTH for the single-code point constructor to work. + * @see UNISTR_OBJECT_SIZE + */ + US_STACKBUF_SIZE=(int32_t)(UNISTR_OBJECT_SIZE-sizeof(void *)-2)/U_SIZEOF_UCHAR, + kInvalidUChar=0xffff, // U+FFFF returned by charAt(invalid index) + kInvalidHashCode=0, // invalid hash code + kEmptyHashCode=1, // hash code for empty string + + // bit flag values for fLengthAndFlags + kIsBogus=1, // this string is bogus, i.e., not valid or nullptr + kUsingStackBuffer=2,// using fUnion.fStackFields instead of fUnion.fFields + kRefCounted=4, // there is a refCount field before the characters in fArray + kBufferIsReadonly=8,// do not write to this buffer + kOpenGetBuffer=16, // getBuffer(minCapacity) was called (is "open"), + // and releaseBuffer(newLength) must be called + kAllStorageFlags=0x1f, + + kLengthShift=5, // remaining 11 bits for non-negative short length, or negative if long + kLength1=1<127; else undefined + int32_t fCapacity; // capacity of fArray (in char16_ts) + // array pointer last to minimize padding for machines with P128 data model + // or pointer sizes that are not a power of 2 + char16_t *fArray; // the Unicode data + } fFields; + } fUnion; +}; + +/** + * Create a new UnicodeString with the concatenation of two others. + * + * @param s1 The first string to be copied to the new one. + * @param s2 The second string to be copied to the new one, after s1. + * @return UnicodeString(s1).append(s2) + * @stable ICU 2.8 + */ +U_COMMON_API UnicodeString U_EXPORT2 +operator+ (const UnicodeString &s1, const UnicodeString &s2); + +//======================================== +// Inline members +//======================================== + +//======================================== +// Privates +//======================================== + +inline void +UnicodeString::pinIndex(int32_t& start) const +{ + // pin index + if(start < 0) { + start = 0; + } else if(start > length()) { + start = length(); + } +} + +inline void +UnicodeString::pinIndices(int32_t& start, + int32_t& _length) const +{ + // pin indices + int32_t len = length(); + if(start < 0) { + start = 0; + } else if(start > len) { + start = len; + } + if(_length < 0) { + _length = 0; + } else if(_length > (len - start)) { + _length = (len - start); + } +} + +inline char16_t* +UnicodeString::getArrayStart() { + return (fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) ? + fUnion.fStackFields.fBuffer : fUnion.fFields.fArray; +} + +inline const char16_t* +UnicodeString::getArrayStart() const { + return (fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) ? + fUnion.fStackFields.fBuffer : fUnion.fFields.fArray; +} + +//======================================== +// Default constructor +//======================================== + +inline +UnicodeString::UnicodeString() { + fUnion.fStackFields.fLengthAndFlags=kShortString; +} + +inline UnicodeString::UnicodeString(const std::nullptr_t /*text*/) { + fUnion.fStackFields.fLengthAndFlags=kShortString; +} + +inline UnicodeString::UnicodeString(const std::nullptr_t /*text*/, int32_t /*length*/) { + fUnion.fStackFields.fLengthAndFlags=kShortString; +} + +inline UnicodeString::UnicodeString(std::nullptr_t /*buffer*/, int32_t /*buffLength*/, int32_t /*buffCapacity*/) { + fUnion.fStackFields.fLengthAndFlags=kShortString; +} + +//======================================== +// Read-only implementation methods +//======================================== +inline UBool +UnicodeString::hasShortLength() const { + return fUnion.fFields.fLengthAndFlags>=0; +} + +inline int32_t +UnicodeString::getShortLength() const { + // fLengthAndFlags must be non-negative -> short length >= 0 + // and arithmetic or logical shift does not matter. + return fUnion.fFields.fLengthAndFlags>>kLengthShift; +} + +inline int32_t +UnicodeString::length() const { + return hasShortLength() ? getShortLength() : fUnion.fFields.fLength; +} + +inline int32_t +UnicodeString::getCapacity() const { + return (fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) ? + US_STACKBUF_SIZE : fUnion.fFields.fCapacity; +} + +inline int32_t +UnicodeString::hashCode() const +{ return doHashCode(); } + +inline UBool +UnicodeString::isBogus() const +{ return (UBool)(fUnion.fFields.fLengthAndFlags & kIsBogus); } + +inline UBool +UnicodeString::isWritable() const +{ return (UBool)!(fUnion.fFields.fLengthAndFlags&(kOpenGetBuffer|kIsBogus)); } + +inline UBool +UnicodeString::isBufferWritable() const +{ + return (UBool)( + !(fUnion.fFields.fLengthAndFlags&(kOpenGetBuffer|kIsBogus|kBufferIsReadonly)) && + (!(fUnion.fFields.fLengthAndFlags&kRefCounted) || refCount()==1)); +} + +inline const char16_t * +UnicodeString::getBuffer() const { + if(fUnion.fFields.fLengthAndFlags&(kIsBogus|kOpenGetBuffer)) { + return nullptr; + } else if(fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) { + return fUnion.fStackFields.fBuffer; + } else { + return fUnion.fFields.fArray; + } +} + +//======================================== +// Read-only alias methods +//======================================== +inline int8_t +UnicodeString::doCompare(int32_t start, + int32_t thisLength, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const +{ + if(srcText.isBogus()) { + return (int8_t)!isBogus(); // 0 if both are bogus, 1 otherwise + } else { + srcText.pinIndices(srcStart, srcLength); + return doCompare(start, thisLength, srcText.getArrayStart(), srcStart, srcLength); + } +} + +inline UBool +UnicodeString::doEqualsSubstring(int32_t start, + int32_t thisLength, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const +{ + if(srcText.isBogus()) { + return isBogus(); + } else { + srcText.pinIndices(srcStart, srcLength); + return !isBogus() && doEqualsSubstring(start, thisLength, srcText.getArrayStart(), srcStart, srcLength); + } +} + +inline bool +UnicodeString::operator== (const UnicodeString& text) const +{ + if(isBogus()) { + return text.isBogus(); + } else { + int32_t len = length(), textLength = text.length(); + return !text.isBogus() && len == textLength && doEquals(text, len); + } +} + +inline bool +UnicodeString::operator!= (const UnicodeString& text) const +{ return (! operator==(text)); } + +inline UBool +UnicodeString::operator> (const UnicodeString& text) const +{ return doCompare(0, length(), text, 0, text.length()) == 1; } + +inline UBool +UnicodeString::operator< (const UnicodeString& text) const +{ return doCompare(0, length(), text, 0, text.length()) == -1; } + +inline UBool +UnicodeString::operator>= (const UnicodeString& text) const +{ return doCompare(0, length(), text, 0, text.length()) != -1; } + +inline UBool +UnicodeString::operator<= (const UnicodeString& text) const +{ return doCompare(0, length(), text, 0, text.length()) != 1; } + +inline int8_t +UnicodeString::compare(const UnicodeString& text) const +{ return doCompare(0, length(), text, 0, text.length()); } + +inline int8_t +UnicodeString::compare(int32_t start, + int32_t _length, + const UnicodeString& srcText) const +{ return doCompare(start, _length, srcText, 0, srcText.length()); } + +inline int8_t +UnicodeString::compare(ConstChar16Ptr srcChars, + int32_t srcLength) const +{ return doCompare(0, length(), srcChars, 0, srcLength); } + +inline int8_t +UnicodeString::compare(int32_t start, + int32_t _length, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const +{ return doCompare(start, _length, srcText, srcStart, srcLength); } + +inline int8_t +UnicodeString::compare(int32_t start, + int32_t _length, + const char16_t *srcChars) const +{ return doCompare(start, _length, srcChars, 0, _length); } + +inline int8_t +UnicodeString::compare(int32_t start, + int32_t _length, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength) const +{ return doCompare(start, _length, srcChars, srcStart, srcLength); } + +inline int8_t +UnicodeString::compareBetween(int32_t start, + int32_t limit, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLimit) const +{ return doCompare(start, limit - start, + srcText, srcStart, srcLimit - srcStart); } + +inline int8_t +UnicodeString::doCompareCodePointOrder(int32_t start, + int32_t thisLength, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const +{ + if(srcText.isBogus()) { + return (int8_t)!isBogus(); // 0 if both are bogus, 1 otherwise + } else { + srcText.pinIndices(srcStart, srcLength); + return doCompareCodePointOrder(start, thisLength, srcText.getArrayStart(), srcStart, srcLength); + } +} + +inline int8_t +UnicodeString::compareCodePointOrder(const UnicodeString& text) const +{ return doCompareCodePointOrder(0, length(), text, 0, text.length()); } + +inline int8_t +UnicodeString::compareCodePointOrder(int32_t start, + int32_t _length, + const UnicodeString& srcText) const +{ return doCompareCodePointOrder(start, _length, srcText, 0, srcText.length()); } + +inline int8_t +UnicodeString::compareCodePointOrder(ConstChar16Ptr srcChars, + int32_t srcLength) const +{ return doCompareCodePointOrder(0, length(), srcChars, 0, srcLength); } + +inline int8_t +UnicodeString::compareCodePointOrder(int32_t start, + int32_t _length, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const +{ return doCompareCodePointOrder(start, _length, srcText, srcStart, srcLength); } + +inline int8_t +UnicodeString::compareCodePointOrder(int32_t start, + int32_t _length, + const char16_t *srcChars) const +{ return doCompareCodePointOrder(start, _length, srcChars, 0, _length); } + +inline int8_t +UnicodeString::compareCodePointOrder(int32_t start, + int32_t _length, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength) const +{ return doCompareCodePointOrder(start, _length, srcChars, srcStart, srcLength); } + +inline int8_t +UnicodeString::compareCodePointOrderBetween(int32_t start, + int32_t limit, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLimit) const +{ return doCompareCodePointOrder(start, limit - start, + srcText, srcStart, srcLimit - srcStart); } + +inline int8_t +UnicodeString::doCaseCompare(int32_t start, + int32_t thisLength, + const UnicodeString &srcText, + int32_t srcStart, + int32_t srcLength, + uint32_t options) const +{ + if(srcText.isBogus()) { + return (int8_t)!isBogus(); // 0 if both are bogus, 1 otherwise + } else { + srcText.pinIndices(srcStart, srcLength); + return doCaseCompare(start, thisLength, srcText.getArrayStart(), srcStart, srcLength, options); + } +} + +inline int8_t +UnicodeString::caseCompare(const UnicodeString &text, uint32_t options) const { + return doCaseCompare(0, length(), text, 0, text.length(), options); +} + +inline int8_t +UnicodeString::caseCompare(int32_t start, + int32_t _length, + const UnicodeString &srcText, + uint32_t options) const { + return doCaseCompare(start, _length, srcText, 0, srcText.length(), options); +} + +inline int8_t +UnicodeString::caseCompare(ConstChar16Ptr srcChars, + int32_t srcLength, + uint32_t options) const { + return doCaseCompare(0, length(), srcChars, 0, srcLength, options); +} + +inline int8_t +UnicodeString::caseCompare(int32_t start, + int32_t _length, + const UnicodeString &srcText, + int32_t srcStart, + int32_t srcLength, + uint32_t options) const { + return doCaseCompare(start, _length, srcText, srcStart, srcLength, options); +} + +inline int8_t +UnicodeString::caseCompare(int32_t start, + int32_t _length, + const char16_t *srcChars, + uint32_t options) const { + return doCaseCompare(start, _length, srcChars, 0, _length, options); +} + +inline int8_t +UnicodeString::caseCompare(int32_t start, + int32_t _length, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength, + uint32_t options) const { + return doCaseCompare(start, _length, srcChars, srcStart, srcLength, options); +} + +inline int8_t +UnicodeString::caseCompareBetween(int32_t start, + int32_t limit, + const UnicodeString &srcText, + int32_t srcStart, + int32_t srcLimit, + uint32_t options) const { + return doCaseCompare(start, limit - start, srcText, srcStart, srcLimit - srcStart, options); +} + +inline int32_t +UnicodeString::indexOf(const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength, + int32_t start, + int32_t _length) const +{ + if(!srcText.isBogus()) { + srcText.pinIndices(srcStart, srcLength); + if(srcLength > 0) { + return indexOf(srcText.getArrayStart(), srcStart, srcLength, start, _length); + } + } + return -1; +} + +inline int32_t +UnicodeString::indexOf(const UnicodeString& text) const +{ return indexOf(text, 0, text.length(), 0, length()); } + +inline int32_t +UnicodeString::indexOf(const UnicodeString& text, + int32_t start) const { + pinIndex(start); + return indexOf(text, 0, text.length(), start, length() - start); +} + +inline int32_t +UnicodeString::indexOf(const UnicodeString& text, + int32_t start, + int32_t _length) const +{ return indexOf(text, 0, text.length(), start, _length); } + +inline int32_t +UnicodeString::indexOf(const char16_t *srcChars, + int32_t srcLength, + int32_t start) const { + pinIndex(start); + return indexOf(srcChars, 0, srcLength, start, length() - start); +} + +inline int32_t +UnicodeString::indexOf(ConstChar16Ptr srcChars, + int32_t srcLength, + int32_t start, + int32_t _length) const +{ return indexOf(srcChars, 0, srcLength, start, _length); } + +inline int32_t +UnicodeString::indexOf(char16_t c, + int32_t start, + int32_t _length) const +{ return doIndexOf(c, start, _length); } + +inline int32_t +UnicodeString::indexOf(UChar32 c, + int32_t start, + int32_t _length) const +{ return doIndexOf(c, start, _length); } + +inline int32_t +UnicodeString::indexOf(char16_t c) const +{ return doIndexOf(c, 0, length()); } + +inline int32_t +UnicodeString::indexOf(UChar32 c) const +{ return indexOf(c, 0, length()); } + +inline int32_t +UnicodeString::indexOf(char16_t c, + int32_t start) const { + pinIndex(start); + return doIndexOf(c, start, length() - start); +} + +inline int32_t +UnicodeString::indexOf(UChar32 c, + int32_t start) const { + pinIndex(start); + return indexOf(c, start, length() - start); +} + +inline int32_t +UnicodeString::lastIndexOf(ConstChar16Ptr srcChars, + int32_t srcLength, + int32_t start, + int32_t _length) const +{ return lastIndexOf(srcChars, 0, srcLength, start, _length); } + +inline int32_t +UnicodeString::lastIndexOf(const char16_t *srcChars, + int32_t srcLength, + int32_t start) const { + pinIndex(start); + return lastIndexOf(srcChars, 0, srcLength, start, length() - start); +} + +inline int32_t +UnicodeString::lastIndexOf(const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength, + int32_t start, + int32_t _length) const +{ + if(!srcText.isBogus()) { + srcText.pinIndices(srcStart, srcLength); + if(srcLength > 0) { + return lastIndexOf(srcText.getArrayStart(), srcStart, srcLength, start, _length); + } + } + return -1; +} + +inline int32_t +UnicodeString::lastIndexOf(const UnicodeString& text, + int32_t start, + int32_t _length) const +{ return lastIndexOf(text, 0, text.length(), start, _length); } + +inline int32_t +UnicodeString::lastIndexOf(const UnicodeString& text, + int32_t start) const { + pinIndex(start); + return lastIndexOf(text, 0, text.length(), start, length() - start); +} + +inline int32_t +UnicodeString::lastIndexOf(const UnicodeString& text) const +{ return lastIndexOf(text, 0, text.length(), 0, length()); } + +inline int32_t +UnicodeString::lastIndexOf(char16_t c, + int32_t start, + int32_t _length) const +{ return doLastIndexOf(c, start, _length); } + +inline int32_t +UnicodeString::lastIndexOf(UChar32 c, + int32_t start, + int32_t _length) const { + return doLastIndexOf(c, start, _length); +} + +inline int32_t +UnicodeString::lastIndexOf(char16_t c) const +{ return doLastIndexOf(c, 0, length()); } + +inline int32_t +UnicodeString::lastIndexOf(UChar32 c) const { + return lastIndexOf(c, 0, length()); +} + +inline int32_t +UnicodeString::lastIndexOf(char16_t c, + int32_t start) const { + pinIndex(start); + return doLastIndexOf(c, start, length() - start); +} + +inline int32_t +UnicodeString::lastIndexOf(UChar32 c, + int32_t start) const { + pinIndex(start); + return lastIndexOf(c, start, length() - start); +} + +inline UBool +UnicodeString::startsWith(const UnicodeString& text) const +{ return doEqualsSubstring(0, text.length(), text, 0, text.length()); } + +inline UBool +UnicodeString::startsWith(const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const +{ return doEqualsSubstring(0, srcLength, srcText, srcStart, srcLength); } + +inline UBool +UnicodeString::startsWith(ConstChar16Ptr srcChars, int32_t srcLength) const { + if(srcLength < 0) { + srcLength = u_strlen(toUCharPtr(srcChars)); + } + return doEqualsSubstring(0, srcLength, srcChars, 0, srcLength); +} + +inline UBool +UnicodeString::startsWith(const char16_t *srcChars, int32_t srcStart, int32_t srcLength) const { + if(srcLength < 0) { + srcLength = u_strlen(toUCharPtr(srcChars)); + } + return doEqualsSubstring(0, srcLength, srcChars, srcStart, srcLength); +} + +inline UBool +UnicodeString::endsWith(const UnicodeString& text) const +{ return doEqualsSubstring(length() - text.length(), text.length(), + text, 0, text.length()); } + +inline UBool +UnicodeString::endsWith(const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) const { + srcText.pinIndices(srcStart, srcLength); + return doEqualsSubstring(length() - srcLength, srcLength, + srcText, srcStart, srcLength); +} + +inline UBool +UnicodeString::endsWith(ConstChar16Ptr srcChars, + int32_t srcLength) const { + if(srcLength < 0) { + srcLength = u_strlen(toUCharPtr(srcChars)); + } + return doEqualsSubstring(length() - srcLength, srcLength, srcChars, 0, srcLength); +} + +inline UBool +UnicodeString::endsWith(const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength) const { + if(srcLength < 0) { + srcLength = u_strlen(toUCharPtr(srcChars + srcStart)); + } + return doEqualsSubstring(length() - srcLength, srcLength, + srcChars, srcStart, srcLength); +} + +//======================================== +// replace +//======================================== +inline UnicodeString& +UnicodeString::replace(int32_t start, + int32_t _length, + const UnicodeString& srcText) +{ return doReplace(start, _length, srcText, 0, srcText.length()); } + +inline UnicodeString& +UnicodeString::replace(int32_t start, + int32_t _length, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) +{ return doReplace(start, _length, srcText, srcStart, srcLength); } + +inline UnicodeString& +UnicodeString::replace(int32_t start, + int32_t _length, + ConstChar16Ptr srcChars, + int32_t srcLength) +{ return doReplace(start, _length, srcChars, 0, srcLength); } + +inline UnicodeString& +UnicodeString::replace(int32_t start, + int32_t _length, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength) +{ return doReplace(start, _length, srcChars, srcStart, srcLength); } + +inline UnicodeString& +UnicodeString::replace(int32_t start, + int32_t _length, + char16_t srcChar) +{ return doReplace(start, _length, &srcChar, 0, 1); } + +inline UnicodeString& +UnicodeString::replaceBetween(int32_t start, + int32_t limit, + const UnicodeString& srcText) +{ return doReplace(start, limit - start, srcText, 0, srcText.length()); } + +inline UnicodeString& +UnicodeString::replaceBetween(int32_t start, + int32_t limit, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLimit) +{ return doReplace(start, limit - start, srcText, srcStart, srcLimit - srcStart); } + +inline UnicodeString& +UnicodeString::findAndReplace(const UnicodeString& oldText, + const UnicodeString& newText) +{ return findAndReplace(0, length(), oldText, 0, oldText.length(), + newText, 0, newText.length()); } + +inline UnicodeString& +UnicodeString::findAndReplace(int32_t start, + int32_t _length, + const UnicodeString& oldText, + const UnicodeString& newText) +{ return findAndReplace(start, _length, oldText, 0, oldText.length(), + newText, 0, newText.length()); } + +// ============================ +// extract +// ============================ +inline void +UnicodeString::doExtract(int32_t start, + int32_t _length, + UnicodeString& target) const +{ target.replace(0, target.length(), *this, start, _length); } + +inline void +UnicodeString::extract(int32_t start, + int32_t _length, + Char16Ptr target, + int32_t targetStart) const +{ doExtract(start, _length, target, targetStart); } + +inline void +UnicodeString::extract(int32_t start, + int32_t _length, + UnicodeString& target) const +{ doExtract(start, _length, target); } + +#if !UCONFIG_NO_CONVERSION + +inline int32_t +UnicodeString::extract(int32_t start, + int32_t _length, + char *dst, + const char *codepage) const + +{ + // This dstSize value will be checked explicitly + return extract(start, _length, dst, dst!=0 ? 0xffffffff : 0, codepage); +} + +#endif + +inline void +UnicodeString::extractBetween(int32_t start, + int32_t limit, + char16_t *dst, + int32_t dstStart) const { + pinIndex(start); + pinIndex(limit); + doExtract(start, limit - start, dst, dstStart); +} + +inline UnicodeString +UnicodeString::tempSubStringBetween(int32_t start, int32_t limit) const { + return tempSubString(start, limit - start); +} + +inline char16_t +UnicodeString::doCharAt(int32_t offset) const +{ + if((uint32_t)offset < (uint32_t)length()) { + return getArrayStart()[offset]; + } else { + return kInvalidUChar; + } +} + +inline char16_t +UnicodeString::charAt(int32_t offset) const +{ return doCharAt(offset); } + +inline char16_t +UnicodeString::operator[] (int32_t offset) const +{ return doCharAt(offset); } + +inline UBool +UnicodeString::isEmpty() const { + // Arithmetic or logical right shift does not matter: only testing for 0. + return (fUnion.fFields.fLengthAndFlags>>kLengthShift) == 0; +} + +//======================================== +// Write implementation methods +//======================================== +inline void +UnicodeString::setZeroLength() { + fUnion.fFields.fLengthAndFlags &= kAllStorageFlags; +} + +inline void +UnicodeString::setShortLength(int32_t len) { + // requires 0 <= len <= kMaxShortLength + fUnion.fFields.fLengthAndFlags = + (int16_t)((fUnion.fFields.fLengthAndFlags & kAllStorageFlags) | (len << kLengthShift)); +} + +inline void +UnicodeString::setLength(int32_t len) { + if(len <= kMaxShortLength) { + setShortLength(len); + } else { + fUnion.fFields.fLengthAndFlags |= kLengthIsLarge; + fUnion.fFields.fLength = len; + } +} + +inline void +UnicodeString::setToEmpty() { + fUnion.fFields.fLengthAndFlags = kShortString; +} + +inline void +UnicodeString::setArray(char16_t *array, int32_t len, int32_t capacity) { + setLength(len); + fUnion.fFields.fArray = array; + fUnion.fFields.fCapacity = capacity; +} + +inline UnicodeString& +UnicodeString::operator= (char16_t ch) +{ return doReplace(0, length(), &ch, 0, 1); } + +inline UnicodeString& +UnicodeString::operator= (UChar32 ch) +{ return replace(0, length(), ch); } + +inline UnicodeString& +UnicodeString::setTo(const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) +{ + unBogus(); + return doReplace(0, length(), srcText, srcStart, srcLength); +} + +inline UnicodeString& +UnicodeString::setTo(const UnicodeString& srcText, + int32_t srcStart) +{ + unBogus(); + srcText.pinIndex(srcStart); + return doReplace(0, length(), srcText, srcStart, srcText.length() - srcStart); +} + +inline UnicodeString& +UnicodeString::setTo(const UnicodeString& srcText) +{ + return copyFrom(srcText); +} + +inline UnicodeString& +UnicodeString::setTo(const char16_t *srcChars, + int32_t srcLength) +{ + unBogus(); + return doReplace(0, length(), srcChars, 0, srcLength); +} + +inline UnicodeString& +UnicodeString::setTo(char16_t srcChar) +{ + unBogus(); + return doReplace(0, length(), &srcChar, 0, 1); +} + +inline UnicodeString& +UnicodeString::setTo(UChar32 srcChar) +{ + unBogus(); + return replace(0, length(), srcChar); +} + +inline UnicodeString& +UnicodeString::append(const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) +{ return doAppend(srcText, srcStart, srcLength); } + +inline UnicodeString& +UnicodeString::append(const UnicodeString& srcText) +{ return doAppend(srcText, 0, srcText.length()); } + +inline UnicodeString& +UnicodeString::append(const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength) +{ return doAppend(srcChars, srcStart, srcLength); } + +inline UnicodeString& +UnicodeString::append(ConstChar16Ptr srcChars, + int32_t srcLength) +{ return doAppend(srcChars, 0, srcLength); } + +inline UnicodeString& +UnicodeString::append(char16_t srcChar) +{ return doAppend(&srcChar, 0, 1); } + +inline UnicodeString& +UnicodeString::operator+= (char16_t ch) +{ return doAppend(&ch, 0, 1); } + +inline UnicodeString& +UnicodeString::operator+= (UChar32 ch) { + return append(ch); +} + +inline UnicodeString& +UnicodeString::operator+= (const UnicodeString& srcText) +{ return doAppend(srcText, 0, srcText.length()); } + +inline UnicodeString& +UnicodeString::insert(int32_t start, + const UnicodeString& srcText, + int32_t srcStart, + int32_t srcLength) +{ return doReplace(start, 0, srcText, srcStart, srcLength); } + +inline UnicodeString& +UnicodeString::insert(int32_t start, + const UnicodeString& srcText) +{ return doReplace(start, 0, srcText, 0, srcText.length()); } + +inline UnicodeString& +UnicodeString::insert(int32_t start, + const char16_t *srcChars, + int32_t srcStart, + int32_t srcLength) +{ return doReplace(start, 0, srcChars, srcStart, srcLength); } + +inline UnicodeString& +UnicodeString::insert(int32_t start, + ConstChar16Ptr srcChars, + int32_t srcLength) +{ return doReplace(start, 0, srcChars, 0, srcLength); } + +inline UnicodeString& +UnicodeString::insert(int32_t start, + char16_t srcChar) +{ return doReplace(start, 0, &srcChar, 0, 1); } + +inline UnicodeString& +UnicodeString::insert(int32_t start, + UChar32 srcChar) +{ return replace(start, 0, srcChar); } + + +inline UnicodeString& +UnicodeString::remove() +{ + // remove() of a bogus string makes the string empty and non-bogus + if(isBogus()) { + setToEmpty(); + } else { + setZeroLength(); + } + return *this; +} + +inline UnicodeString& +UnicodeString::remove(int32_t start, + int32_t _length) +{ + if(start <= 0 && _length == INT32_MAX) { + // remove(guaranteed everything) of a bogus string makes the string empty and non-bogus + return remove(); + } + return doReplace(start, _length, nullptr, 0, 0); +} + +inline UnicodeString& +UnicodeString::removeBetween(int32_t start, + int32_t limit) +{ return doReplace(start, limit - start, nullptr, 0, 0); } + +inline UnicodeString & +UnicodeString::retainBetween(int32_t start, int32_t limit) { + truncate(limit); + return doReplace(0, start, nullptr, 0, 0); +} + +inline UBool +UnicodeString::truncate(int32_t targetLength) +{ + if(isBogus() && targetLength == 0) { + // truncate(0) of a bogus string makes the string empty and non-bogus + unBogus(); + return false; + } else if((uint32_t)targetLength < (uint32_t)length()) { + setLength(targetLength); + return true; + } else { + return false; + } +} + +inline UnicodeString& +UnicodeString::reverse() +{ return doReverse(0, length()); } + +inline UnicodeString& +UnicodeString::reverse(int32_t start, + int32_t _length) +{ return doReverse(start, _length); } + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/unorm.h b/intl/icu/source/common/unicode/unorm.h new file mode 100644 index 0000000000..38fb895155 --- /dev/null +++ b/intl/icu/source/common/unicode/unorm.h @@ -0,0 +1,476 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (c) 1996-2016, International Business Machines Corporation +* and others. All Rights Reserved. +******************************************************************************* +* File unorm.h +* +* Created by: Vladimir Weinstein 12052000 +* +* Modification history : +* +* Date Name Description +* 02/01/01 synwee Added normalization quickcheck enum and method. +*/ +#ifndef UNORM_H +#define UNORM_H + +#include "unicode/utypes.h" + +#if !UCONFIG_NO_NORMALIZATION + +#include "unicode/uiter.h" +#include "unicode/unorm2.h" + +/** + * \file + * \brief C API: Unicode Normalization + * + * Old Unicode normalization API. + * + * This API has been replaced by the unorm2.h API and is only available + * for backward compatibility. The functions here simply delegate to the + * unorm2.h functions, for example unorm2_getInstance() and unorm2_normalize(). + * There is one exception: The new API does not provide a replacement for unorm_compare(). + * Its declaration has been moved to unorm2.h. + * + * unorm_normalize transforms Unicode text into an equivalent composed or + * decomposed form, allowing for easier sorting and searching of text. + * unorm_normalize supports the standard normalization forms described in + * + * Unicode Standard Annex #15: Unicode Normalization Forms. + * + * Characters with accents or other adornments can be encoded in + * several different ways in Unicode. For example, take the character A-acute. + * In Unicode, this can be encoded as a single character (the + * "composed" form): + * + * \code + * 00C1 LATIN CAPITAL LETTER A WITH ACUTE + * \endcode + * + * or as two separate characters (the "decomposed" form): + * + * \code + * 0041 LATIN CAPITAL LETTER A + * 0301 COMBINING ACUTE ACCENT + * \endcode + * + * To a user of your program, however, both of these sequences should be + * treated as the same "user-level" character "A with acute accent". When you are searching or + * comparing text, you must ensure that these two sequences are treated + * equivalently. In addition, you must handle characters with more than one + * accent. Sometimes the order of a character's combining accents is + * significant, while in other cases accent sequences in different orders are + * really equivalent. + * + * Similarly, the string "ffi" can be encoded as three separate letters: + * + * \code + * 0066 LATIN SMALL LETTER F + * 0066 LATIN SMALL LETTER F + * 0069 LATIN SMALL LETTER I + * \endcode + * + * or as the single character + * + * \code + * FB03 LATIN SMALL LIGATURE FFI + * \endcode + * + * The ffi ligature is not a distinct semantic character, and strictly speaking + * it shouldn't be in Unicode at all, but it was included for compatibility + * with existing character sets that already provided it. The Unicode standard + * identifies such characters by giving them "compatibility" decompositions + * into the corresponding semantic characters. When sorting and searching, you + * will often want to use these mappings. + * + * unorm_normalize helps solve these problems by transforming text into the + * canonical composed and decomposed forms as shown in the first example above. + * In addition, you can have it perform compatibility decompositions so that + * you can treat compatibility characters the same as their equivalents. + * Finally, unorm_normalize rearranges accents into the proper canonical + * order, so that you do not have to worry about accent rearrangement on your + * own. + * + * Form FCD, "Fast C or D", is also designed for collation. + * It allows to work on strings that are not necessarily normalized + * with an algorithm (like in collation) that works under "canonical closure", i.e., it treats precomposed + * characters and their decomposed equivalents the same. + * + * It is not a normalization form because it does not provide for uniqueness of representation. Multiple strings + * may be canonically equivalent (their NFDs are identical) and may all conform to FCD without being identical + * themselves. + * + * The form is defined such that the "raw decomposition", the recursive canonical decomposition of each character, + * results in a string that is canonically ordered. This means that precomposed characters are allowed for as long + * as their decompositions do not need canonical reordering. + * + * Its advantage for a process like collation is that all NFD and most NFC texts - and many unnormalized texts - + * already conform to FCD and do not need to be normalized (NFD) for such a process. The FCD quick check will + * return UNORM_YES for most strings in practice. + * + * unorm_normalize(UNORM_FCD) may be implemented with UNORM_NFD. + * + * For more details on FCD see the collation design document: + * https://htmlpreview.github.io/?https://github.com/unicode-org/icu-docs/blob/main/design/collation/ICU_collation_design.htm + * + * ICU collation performs either NFD or FCD normalization automatically if normalization + * is turned on for the collator object. + * Beyond collation and string search, normalized strings may be useful for string equivalence comparisons, + * transliteration/transcription, unique representations, etc. + * + * The W3C generally recommends to exchange texts in NFC. + * Note also that most legacy character encodings use only precomposed forms and often do not + * encode any combining marks by themselves. For conversion to such character encodings the + * Unicode text needs to be normalized to NFC. + * For more usage examples, see the Unicode Standard Annex. + */ + +// Do not conditionalize the following enum with #ifndef U_HIDE_DEPRECATED_API, +// it is needed for layout of Normalizer object. +#ifndef U_FORCE_HIDE_DEPRECATED_API + +/** + * Constants for normalization modes. + * @deprecated ICU 56 Use unorm2.h instead. + */ +typedef enum { + /** No decomposition/composition. @deprecated ICU 56 Use unorm2.h instead. */ + UNORM_NONE = 1, + /** Canonical decomposition. @deprecated ICU 56 Use unorm2.h instead. */ + UNORM_NFD = 2, + /** Compatibility decomposition. @deprecated ICU 56 Use unorm2.h instead. */ + UNORM_NFKD = 3, + /** Canonical decomposition followed by canonical composition. @deprecated ICU 56 Use unorm2.h instead. */ + UNORM_NFC = 4, + /** Default normalization. @deprecated ICU 56 Use unorm2.h instead. */ + UNORM_DEFAULT = UNORM_NFC, + /** Compatibility decomposition followed by canonical composition. @deprecated ICU 56 Use unorm2.h instead. */ + UNORM_NFKC =5, + /** "Fast C or D" form. @deprecated ICU 56 Use unorm2.h instead. */ + UNORM_FCD = 6, + + /** One more than the highest normalization mode constant. @deprecated ICU 56 Use unorm2.h instead. */ + UNORM_MODE_COUNT +} UNormalizationMode; + +#endif // U_FORCE_HIDE_DEPRECATED_API + +#ifndef U_HIDE_DEPRECATED_API + +/** + * Constants for options flags for normalization. + * Use 0 for default options, + * including normalization according to the Unicode version + * that is currently supported by ICU (see u_getUnicodeVersion). + * @deprecated ICU 56 Use unorm2.h instead. + */ +enum { + /** + * Options bit set value to select Unicode 3.2 normalization + * (except NormalizationCorrections). + * At most one Unicode version can be selected at a time. + * @deprecated ICU 56 Use unorm2.h instead. + */ + UNORM_UNICODE_3_2=0x20 +}; + +/** + * Lowest-order bit number of unorm_compare() options bits corresponding to + * normalization options bits. + * + * The options parameter for unorm_compare() uses most bits for + * itself and for various comparison and folding flags. + * The most significant bits, however, are shifted down and passed on + * to the normalization implementation. + * (That is, from unorm_compare(..., options, ...), + * options>>UNORM_COMPARE_NORM_OPTIONS_SHIFT will be passed on to the + * internal normalization functions.) + * + * @see unorm_compare + * @deprecated ICU 56 Use unorm2.h instead. + */ +#define UNORM_COMPARE_NORM_OPTIONS_SHIFT 20 + +/** + * Normalize a string. + * The string will be normalized according the specified normalization mode + * and options. + * The source and result buffers must not be the same, nor overlap. + * + * @param source The string to normalize. + * @param sourceLength The length of source, or -1 if NUL-terminated. + * @param mode The normalization mode; one of UNORM_NONE, + * UNORM_NFD, UNORM_NFC, UNORM_NFKC, UNORM_NFKD, UNORM_DEFAULT. + * @param options The normalization options, ORed together (0 for no options). + * @param result A pointer to a buffer to receive the result string. + * The result string is NUL-terminated if possible. + * @param resultLength The maximum size of result. + * @param status A pointer to a UErrorCode to receive any errors. + * @return The total buffer size needed; if greater than resultLength, + * the output was truncated, and the error code is set to U_BUFFER_OVERFLOW_ERROR. + * @deprecated ICU 56 Use unorm2.h instead. + */ +U_DEPRECATED int32_t U_EXPORT2 +unorm_normalize(const UChar *source, int32_t sourceLength, + UNormalizationMode mode, int32_t options, + UChar *result, int32_t resultLength, + UErrorCode *status); + +/** + * Performing quick check on a string, to quickly determine if the string is + * in a particular normalization format. + * Three types of result can be returned UNORM_YES, UNORM_NO or + * UNORM_MAYBE. Result UNORM_YES indicates that the argument + * string is in the desired normalized format, UNORM_NO determines that + * argument string is not in the desired normalized format. A + * UNORM_MAYBE result indicates that a more thorough check is required, + * the user may have to put the string in its normalized form and compare the + * results. + * + * @param source string for determining if it is in a normalized format + * @param sourcelength length of source to test, or -1 if NUL-terminated + * @param mode which normalization form to test for + * @param status a pointer to a UErrorCode to receive any errors + * @return UNORM_YES, UNORM_NO or UNORM_MAYBE + * + * @see unorm_isNormalized + * @deprecated ICU 56 Use unorm2.h instead. + */ +U_DEPRECATED UNormalizationCheckResult U_EXPORT2 +unorm_quickCheck(const UChar *source, int32_t sourcelength, + UNormalizationMode mode, + UErrorCode *status); + +/** + * Performing quick check on a string; same as unorm_quickCheck but + * takes an extra options parameter like most normalization functions. + * + * @param src String that is to be tested if it is in a normalization format. + * @param srcLength Length of source to test, or -1 if NUL-terminated. + * @param mode Which normalization form to test for. + * @param options The normalization options, ORed together (0 for no options). + * @param pErrorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return UNORM_YES, UNORM_NO or UNORM_MAYBE + * + * @see unorm_quickCheck + * @see unorm_isNormalized + * @deprecated ICU 56 Use unorm2.h instead. + */ +U_DEPRECATED UNormalizationCheckResult U_EXPORT2 +unorm_quickCheckWithOptions(const UChar *src, int32_t srcLength, + UNormalizationMode mode, int32_t options, + UErrorCode *pErrorCode); + +/** + * Test if a string is in a given normalization form. + * This is semantically equivalent to source.equals(normalize(source, mode)) . + * + * Unlike unorm_quickCheck(), this function returns a definitive result, + * never a "maybe". + * For NFD, NFKD, and FCD, both functions work exactly the same. + * For NFC and NFKC where quickCheck may return "maybe", this function will + * perform further tests to arrive at a true/false result. + * + * @param src String that is to be tested if it is in a normalization format. + * @param srcLength Length of source to test, or -1 if NUL-terminated. + * @param mode Which normalization form to test for. + * @param pErrorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return Boolean value indicating whether the source string is in the + * "mode" normalization form. + * + * @see unorm_quickCheck + * @deprecated ICU 56 Use unorm2.h instead. + */ +U_DEPRECATED UBool U_EXPORT2 +unorm_isNormalized(const UChar *src, int32_t srcLength, + UNormalizationMode mode, + UErrorCode *pErrorCode); + +/** + * Test if a string is in a given normalization form; same as unorm_isNormalized but + * takes an extra options parameter like most normalization functions. + * + * @param src String that is to be tested if it is in a normalization format. + * @param srcLength Length of source to test, or -1 if NUL-terminated. + * @param mode Which normalization form to test for. + * @param options The normalization options, ORed together (0 for no options). + * @param pErrorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return Boolean value indicating whether the source string is in the + * "mode/options" normalization form. + * + * @see unorm_quickCheck + * @see unorm_isNormalized + * @deprecated ICU 56 Use unorm2.h instead. + */ +U_DEPRECATED UBool U_EXPORT2 +unorm_isNormalizedWithOptions(const UChar *src, int32_t srcLength, + UNormalizationMode mode, int32_t options, + UErrorCode *pErrorCode); + +/** + * Iterative normalization forward. + * This function (together with unorm_previous) is somewhat + * similar to the C++ Normalizer class (see its non-static functions). + * + * Iterative normalization is useful when only a small portion of a longer + * string/text needs to be processed. + * + * For example, the likelihood may be high that processing the first 10% of some + * text will be sufficient to find certain data. + * Another example: When one wants to concatenate two normalized strings and get a + * normalized result, it is much more efficient to normalize just a small part of + * the result around the concatenation place instead of re-normalizing everything. + * + * The input text is an instance of the C character iteration API UCharIterator. + * It may wrap around a simple string, a CharacterIterator, a Replaceable, or any + * other kind of text object. + * + * If a buffer overflow occurs, then the caller needs to reset the iterator to the + * old index and call the function again with a larger buffer - if the caller cares + * for the actual output. + * Regardless of the output buffer, the iterator will always be moved to the next + * normalization boundary. + * + * This function (like unorm_previous) serves two purposes: + * + * 1) To find the next boundary so that the normalization of the part of the text + * from the current position to that boundary does not affect and is not affected + * by the part of the text beyond that boundary. + * + * 2) To normalize the text up to the boundary. + * + * The second step is optional, per the doNormalize parameter. + * It is omitted for operations like string concatenation, where the two adjacent + * string ends need to be normalized together. + * In such a case, the output buffer will just contain a copy of the text up to the + * boundary. + * + * pNeededToNormalize is an output-only parameter. Its output value is only defined + * if normalization was requested (doNormalize) and successful (especially, no + * buffer overflow). + * It is useful for operations like a normalizing transliterator, where one would + * not want to replace a piece of text if it is not modified. + * + * If doNormalize==true and pNeededToNormalize!=NULL then *pNeeded... is set true + * if the normalization was necessary. + * + * If doNormalize==false then *pNeededToNormalize will be set to false. + * + * If the buffer overflows, then *pNeededToNormalize will be undefined; + * essentially, whenever U_FAILURE is true (like in buffer overflows), this result + * will be undefined. + * + * @param src The input text in the form of a C character iterator. + * @param dest The output buffer; can be NULL if destCapacity==0 for pure preflighting. + * @param destCapacity The number of UChars that fit into dest. + * @param mode The normalization mode. + * @param options The normalization options, ORed together (0 for no options). + * @param doNormalize Indicates if the source text up to the next boundary + * is to be normalized (true) or just copied (false). + * @param pNeededToNormalize Output flag indicating if the normalization resulted in + * different text from the input. + * Not defined if an error occurs including buffer overflow. + * Always false if !doNormalize. + * @param pErrorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return Length of output (number of UChars) when successful or buffer overflow. + * + * @see unorm_previous + * @see unorm_normalize + * + * @deprecated ICU 56 Use unorm2.h instead. + */ +U_DEPRECATED int32_t U_EXPORT2 +unorm_next(UCharIterator *src, + UChar *dest, int32_t destCapacity, + UNormalizationMode mode, int32_t options, + UBool doNormalize, UBool *pNeededToNormalize, + UErrorCode *pErrorCode); + +/** + * Iterative normalization backward. + * This function (together with unorm_next) is somewhat + * similar to the C++ Normalizer class (see its non-static functions). + * For all details see unorm_next. + * + * @param src The input text in the form of a C character iterator. + * @param dest The output buffer; can be NULL if destCapacity==0 for pure preflighting. + * @param destCapacity The number of UChars that fit into dest. + * @param mode The normalization mode. + * @param options The normalization options, ORed together (0 for no options). + * @param doNormalize Indicates if the source text up to the next boundary + * is to be normalized (true) or just copied (false). + * @param pNeededToNormalize Output flag indicating if the normalization resulted in + * different text from the input. + * Not defined if an error occurs including buffer overflow. + * Always false if !doNormalize. + * @param pErrorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return Length of output (number of UChars) when successful or buffer overflow. + * + * @see unorm_next + * @see unorm_normalize + * + * @deprecated ICU 56 Use unorm2.h instead. + */ +U_DEPRECATED int32_t U_EXPORT2 +unorm_previous(UCharIterator *src, + UChar *dest, int32_t destCapacity, + UNormalizationMode mode, int32_t options, + UBool doNormalize, UBool *pNeededToNormalize, + UErrorCode *pErrorCode); + +/** + * Concatenate normalized strings, making sure that the result is normalized as well. + * + * If both the left and the right strings are in + * the normalization form according to "mode/options", + * then the result will be + * + * \code + * dest=normalize(left+right, mode, options) + * \endcode + * + * With the input strings already being normalized, + * this function will use unorm_next() and unorm_previous() + * to find the adjacent end pieces of the input strings. + * Only the concatenation of these end pieces will be normalized and + * then concatenated with the remaining parts of the input strings. + * + * It is allowed to have dest==left to avoid copying the entire left string. + * + * @param left Left source string, may be same as dest. + * @param leftLength Length of left source string, or -1 if NUL-terminated. + * @param right Right source string. Must not be the same as dest, nor overlap. + * @param rightLength Length of right source string, or -1 if NUL-terminated. + * @param dest The output buffer; can be NULL if destCapacity==0 for pure preflighting. + * @param destCapacity The number of UChars that fit into dest. + * @param mode The normalization mode. + * @param options The normalization options, ORed together (0 for no options). + * @param pErrorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return Length of output (number of UChars) when successful or buffer overflow. + * + * @see unorm_normalize + * @see unorm_next + * @see unorm_previous + * + * @deprecated ICU 56 Use unorm2.h instead. + */ +U_DEPRECATED int32_t U_EXPORT2 +unorm_concatenate(const UChar *left, int32_t leftLength, + const UChar *right, int32_t rightLength, + UChar *dest, int32_t destCapacity, + UNormalizationMode mode, int32_t options, + UErrorCode *pErrorCode); + +#endif /* U_HIDE_DEPRECATED_API */ +#endif /* #if !UCONFIG_NO_NORMALIZATION */ +#endif diff --git a/intl/icu/source/common/unicode/unorm2.h b/intl/icu/source/common/unicode/unorm2.h new file mode 100644 index 0000000000..24417b7103 --- /dev/null +++ b/intl/icu/source/common/unicode/unorm2.h @@ -0,0 +1,606 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2009-2015, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: unorm2.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2009dec15 +* created by: Markus W. Scherer +*/ + +#ifndef __UNORM2_H__ +#define __UNORM2_H__ + +/** + * \file + * \brief C API: New API for Unicode Normalization. + * + * Unicode normalization functionality for standard Unicode normalization or + * for using custom mapping tables. + * All instances of UNormalizer2 are unmodifiable/immutable. + * Instances returned by unorm2_getInstance() are singletons that must not be deleted by the caller. + * For more details see the Normalizer2 C++ class. + */ + +#include "unicode/utypes.h" +#include "unicode/stringoptions.h" +#include "unicode/uset.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +/** + * Constants for normalization modes. + * For details about standard Unicode normalization forms + * and about the algorithms which are also used with custom mapping tables + * see http://www.unicode.org/unicode/reports/tr15/ + * @stable ICU 4.4 + */ +typedef enum { + /** + * Decomposition followed by composition. + * Same as standard NFC when using an "nfc" instance. + * Same as standard NFKC when using an "nfkc" instance. + * For details about standard Unicode normalization forms + * see http://www.unicode.org/unicode/reports/tr15/ + * @stable ICU 4.4 + */ + UNORM2_COMPOSE, + /** + * Map, and reorder canonically. + * Same as standard NFD when using an "nfc" instance. + * Same as standard NFKD when using an "nfkc" instance. + * For details about standard Unicode normalization forms + * see http://www.unicode.org/unicode/reports/tr15/ + * @stable ICU 4.4 + */ + UNORM2_DECOMPOSE, + /** + * "Fast C or D" form. + * If a string is in this form, then further decomposition without reordering + * would yield the same form as DECOMPOSE. + * Text in "Fast C or D" form can be processed efficiently with data tables + * that are "canonically closed", that is, that provide equivalent data for + * equivalent text, without having to be fully normalized. + * Not a standard Unicode normalization form. + * Not a unique form: Different FCD strings can be canonically equivalent. + * For details see http://www.unicode.org/notes/tn5/#FCD + * @stable ICU 4.4 + */ + UNORM2_FCD, + /** + * Compose only contiguously. + * Also known as "FCC" or "Fast C Contiguous". + * The result will often but not always be in NFC. + * The result will conform to FCD which is useful for processing. + * Not a standard Unicode normalization form. + * For details see http://www.unicode.org/notes/tn5/#FCC + * @stable ICU 4.4 + */ + UNORM2_COMPOSE_CONTIGUOUS +} UNormalization2Mode; + +/** + * Result values for normalization quick check functions. + * For details see http://www.unicode.org/reports/tr15/#Detecting_Normalization_Forms + * @stable ICU 2.0 + */ +typedef enum UNormalizationCheckResult { + /** + * The input string is not in the normalization form. + * @stable ICU 2.0 + */ + UNORM_NO, + /** + * The input string is in the normalization form. + * @stable ICU 2.0 + */ + UNORM_YES, + /** + * The input string may or may not be in the normalization form. + * This value is only returned for composition forms like NFC and FCC, + * when a backward-combining character is found for which the surrounding text + * would have to be analyzed further. + * @stable ICU 2.0 + */ + UNORM_MAYBE +} UNormalizationCheckResult; + +/** + * Opaque C service object type for the new normalization API. + * @stable ICU 4.4 + */ +struct UNormalizer2; +typedef struct UNormalizer2 UNormalizer2; /**< C typedef for struct UNormalizer2. @stable ICU 4.4 */ + +#if !UCONFIG_NO_NORMALIZATION + +/** + * Returns a UNormalizer2 instance for Unicode NFC normalization. + * Same as unorm2_getInstance(NULL, "nfc", UNORM2_COMPOSE, pErrorCode). + * Returns an unmodifiable singleton instance. Do not delete it. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the requested Normalizer2, if successful + * @stable ICU 49 + */ +U_CAPI const UNormalizer2 * U_EXPORT2 +unorm2_getNFCInstance(UErrorCode *pErrorCode); + +/** + * Returns a UNormalizer2 instance for Unicode NFD normalization. + * Same as unorm2_getInstance(NULL, "nfc", UNORM2_DECOMPOSE, pErrorCode). + * Returns an unmodifiable singleton instance. Do not delete it. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the requested Normalizer2, if successful + * @stable ICU 49 + */ +U_CAPI const UNormalizer2 * U_EXPORT2 +unorm2_getNFDInstance(UErrorCode *pErrorCode); + +/** + * Returns a UNormalizer2 instance for Unicode NFKC normalization. + * Same as unorm2_getInstance(NULL, "nfkc", UNORM2_COMPOSE, pErrorCode). + * Returns an unmodifiable singleton instance. Do not delete it. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the requested Normalizer2, if successful + * @stable ICU 49 + */ +U_CAPI const UNormalizer2 * U_EXPORT2 +unorm2_getNFKCInstance(UErrorCode *pErrorCode); + +/** + * Returns a UNormalizer2 instance for Unicode NFKD normalization. + * Same as unorm2_getInstance(NULL, "nfkc", UNORM2_DECOMPOSE, pErrorCode). + * Returns an unmodifiable singleton instance. Do not delete it. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the requested Normalizer2, if successful + * @stable ICU 49 + */ +U_CAPI const UNormalizer2 * U_EXPORT2 +unorm2_getNFKDInstance(UErrorCode *pErrorCode); + +/** + * Returns a UNormalizer2 instance for Unicode NFKC_Casefold normalization. + * Same as unorm2_getInstance(NULL, "nfkc_cf", UNORM2_COMPOSE, pErrorCode). + * Returns an unmodifiable singleton instance. Do not delete it. + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the requested Normalizer2, if successful + * @stable ICU 49 + */ +U_CAPI const UNormalizer2 * U_EXPORT2 +unorm2_getNFKCCasefoldInstance(UErrorCode *pErrorCode); + +/** + * Returns a UNormalizer2 instance which uses the specified data file + * (packageName/name similar to ucnv_openPackage() and ures_open()/ResourceBundle) + * and which composes or decomposes text according to the specified mode. + * Returns an unmodifiable singleton instance. Do not delete it. + * + * Use packageName=NULL for data files that are part of ICU's own data. + * Use name="nfc" and UNORM2_COMPOSE/UNORM2_DECOMPOSE for Unicode standard NFC/NFD. + * Use name="nfkc" and UNORM2_COMPOSE/UNORM2_DECOMPOSE for Unicode standard NFKC/NFKD. + * Use name="nfkc_cf" and UNORM2_COMPOSE for Unicode standard NFKC_CF=NFKC_Casefold. + * + * @param packageName NULL for ICU built-in data, otherwise application data package name + * @param name "nfc" or "nfkc" or "nfkc_cf" or name of custom data file + * @param mode normalization mode (compose or decompose etc.) + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the requested UNormalizer2, if successful + * @stable ICU 4.4 + */ +U_CAPI const UNormalizer2 * U_EXPORT2 +unorm2_getInstance(const char *packageName, + const char *name, + UNormalization2Mode mode, + UErrorCode *pErrorCode); + +/** + * Constructs a filtered normalizer wrapping any UNormalizer2 instance + * and a filter set. + * Both are aliased and must not be modified or deleted while this object + * is used. + * The filter set should be frozen; otherwise the performance will suffer greatly. + * @param norm2 wrapped UNormalizer2 instance + * @param filterSet USet which determines the characters to be normalized + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the requested UNormalizer2, if successful + * @stable ICU 4.4 + */ +U_CAPI UNormalizer2 * U_EXPORT2 +unorm2_openFiltered(const UNormalizer2 *norm2, const USet *filterSet, UErrorCode *pErrorCode); + +/** + * Closes a UNormalizer2 instance from unorm2_openFiltered(). + * Do not close instances from unorm2_getInstance()! + * @param norm2 UNormalizer2 instance to be closed + * @stable ICU 4.4 + */ +U_CAPI void U_EXPORT2 +unorm2_close(UNormalizer2 *norm2); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUNormalizer2Pointer + * "Smart pointer" class, closes a UNormalizer2 via unorm2_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUNormalizer2Pointer, UNormalizer2, unorm2_close); + +U_NAMESPACE_END + +#endif + +/** + * Writes the normalized form of the source string to the destination string + * (replacing its contents) and returns the length of the destination string. + * The source and destination strings must be different buffers. + * @param norm2 UNormalizer2 instance + * @param src source string + * @param length length of the source string, or -1 if NUL-terminated + * @param dest destination string; its contents is replaced with normalized src + * @param capacity number of UChars that can be written to dest + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return dest + * @stable ICU 4.4 + */ +U_CAPI int32_t U_EXPORT2 +unorm2_normalize(const UNormalizer2 *norm2, + const UChar *src, int32_t length, + UChar *dest, int32_t capacity, + UErrorCode *pErrorCode); +/** + * Appends the normalized form of the second string to the first string + * (merging them at the boundary) and returns the length of the first string. + * The result is normalized if the first string was normalized. + * The first and second strings must be different buffers. + * @param norm2 UNormalizer2 instance + * @param first string, should be normalized + * @param firstLength length of the first string, or -1 if NUL-terminated + * @param firstCapacity number of UChars that can be written to first + * @param second string, will be normalized + * @param secondLength length of the source string, or -1 if NUL-terminated + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return first + * @stable ICU 4.4 + */ +U_CAPI int32_t U_EXPORT2 +unorm2_normalizeSecondAndAppend(const UNormalizer2 *norm2, + UChar *first, int32_t firstLength, int32_t firstCapacity, + const UChar *second, int32_t secondLength, + UErrorCode *pErrorCode); +/** + * Appends the second string to the first string + * (merging them at the boundary) and returns the length of the first string. + * The result is normalized if both the strings were normalized. + * The first and second strings must be different buffers. + * @param norm2 UNormalizer2 instance + * @param first string, should be normalized + * @param firstLength length of the first string, or -1 if NUL-terminated + * @param firstCapacity number of UChars that can be written to first + * @param second string, should be normalized + * @param secondLength length of the source string, or -1 if NUL-terminated + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return first + * @stable ICU 4.4 + */ +U_CAPI int32_t U_EXPORT2 +unorm2_append(const UNormalizer2 *norm2, + UChar *first, int32_t firstLength, int32_t firstCapacity, + const UChar *second, int32_t secondLength, + UErrorCode *pErrorCode); + +/** + * Gets the decomposition mapping of c. + * Roughly equivalent to normalizing the String form of c + * on a UNORM2_DECOMPOSE UNormalizer2 instance, but much faster, and except that this function + * returns a negative value and does not write a string + * if c does not have a decomposition mapping in this instance's data. + * This function is independent of the mode of the UNormalizer2. + * @param norm2 UNormalizer2 instance + * @param c code point + * @param decomposition String buffer which will be set to c's + * decomposition mapping, if there is one. + * @param capacity number of UChars that can be written to decomposition + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the non-negative length of c's decomposition, if there is one; otherwise a negative value + * @stable ICU 4.6 + */ +U_CAPI int32_t U_EXPORT2 +unorm2_getDecomposition(const UNormalizer2 *norm2, + UChar32 c, UChar *decomposition, int32_t capacity, + UErrorCode *pErrorCode); + +/** + * Gets the raw decomposition mapping of c. + * + * This is similar to the unorm2_getDecomposition() function but returns the + * raw decomposition mapping as specified in UnicodeData.txt or + * (for custom data) in the mapping files processed by the gennorm2 tool. + * By contrast, unorm2_getDecomposition() returns the processed, + * recursively-decomposed version of this mapping. + * + * When used on a standard NFKC Normalizer2 instance, + * unorm2_getRawDecomposition() returns the Unicode Decomposition_Mapping (dm) property. + * + * When used on a standard NFC Normalizer2 instance, + * it returns the Decomposition_Mapping only if the Decomposition_Type (dt) is Canonical (Can); + * in this case, the result contains either one or two code points (=1..4 UChars). + * + * This function is independent of the mode of the UNormalizer2. + * @param norm2 UNormalizer2 instance + * @param c code point + * @param decomposition String buffer which will be set to c's + * raw decomposition mapping, if there is one. + * @param capacity number of UChars that can be written to decomposition + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return the non-negative length of c's raw decomposition, if there is one; otherwise a negative value + * @stable ICU 49 + */ +U_CAPI int32_t U_EXPORT2 +unorm2_getRawDecomposition(const UNormalizer2 *norm2, + UChar32 c, UChar *decomposition, int32_t capacity, + UErrorCode *pErrorCode); + +/** + * Performs pairwise composition of a & b and returns the composite if there is one. + * + * Returns a composite code point c only if c has a two-way mapping to a+b. + * In standard Unicode normalization, this means that + * c has a canonical decomposition to a+b + * and c does not have the Full_Composition_Exclusion property. + * + * This function is independent of the mode of the UNormalizer2. + * @param norm2 UNormalizer2 instance + * @param a A (normalization starter) code point. + * @param b Another code point. + * @return The non-negative composite code point if there is one; otherwise a negative value. + * @stable ICU 49 + */ +U_CAPI UChar32 U_EXPORT2 +unorm2_composePair(const UNormalizer2 *norm2, UChar32 a, UChar32 b); + +/** + * Gets the combining class of c. + * The default implementation returns 0 + * but all standard implementations return the Unicode Canonical_Combining_Class value. + * @param norm2 UNormalizer2 instance + * @param c code point + * @return c's combining class + * @stable ICU 49 + */ +U_CAPI uint8_t U_EXPORT2 +unorm2_getCombiningClass(const UNormalizer2 *norm2, UChar32 c); + +/** + * Tests if the string is normalized. + * Internally, in cases where the quickCheck() method would return "maybe" + * (which is only possible for the two COMPOSE modes) this method + * resolves to "yes" or "no" to provide a definitive result, + * at the cost of doing more work in those cases. + * @param norm2 UNormalizer2 instance + * @param s input string + * @param length length of the string, or -1 if NUL-terminated + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return true if s is normalized + * @stable ICU 4.4 + */ +U_CAPI UBool U_EXPORT2 +unorm2_isNormalized(const UNormalizer2 *norm2, + const UChar *s, int32_t length, + UErrorCode *pErrorCode); + +/** + * Tests if the string is normalized. + * For the two COMPOSE modes, the result could be "maybe" in cases that + * would take a little more work to resolve definitively. + * Use spanQuickCheckYes() and normalizeSecondAndAppend() for a faster + * combination of quick check + normalization, to avoid + * re-checking the "yes" prefix. + * @param norm2 UNormalizer2 instance + * @param s input string + * @param length length of the string, or -1 if NUL-terminated + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return UNormalizationCheckResult + * @stable ICU 4.4 + */ +U_CAPI UNormalizationCheckResult U_EXPORT2 +unorm2_quickCheck(const UNormalizer2 *norm2, + const UChar *s, int32_t length, + UErrorCode *pErrorCode); + +/** + * Returns the end of the normalized substring of the input string. + * In other words, with end=spanQuickCheckYes(s, ec); + * the substring UnicodeString(s, 0, end) + * will pass the quick check with a "yes" result. + * + * The returned end index is usually one or more characters before the + * "no" or "maybe" character: The end index is at a normalization boundary. + * (See the class documentation for more about normalization boundaries.) + * + * When the goal is a normalized string and most input strings are expected + * to be normalized already, then call this method, + * and if it returns a prefix shorter than the input string, + * copy that prefix and use normalizeSecondAndAppend() for the remainder. + * @param norm2 UNormalizer2 instance + * @param s input string + * @param length length of the string, or -1 if NUL-terminated + * @param pErrorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return "yes" span end index + * @stable ICU 4.4 + */ +U_CAPI int32_t U_EXPORT2 +unorm2_spanQuickCheckYes(const UNormalizer2 *norm2, + const UChar *s, int32_t length, + UErrorCode *pErrorCode); + +/** + * Tests if the character always has a normalization boundary before it, + * regardless of context. + * For details see the Normalizer2 base class documentation. + * @param norm2 UNormalizer2 instance + * @param c character to test + * @return true if c has a normalization boundary before it + * @stable ICU 4.4 + */ +U_CAPI UBool U_EXPORT2 +unorm2_hasBoundaryBefore(const UNormalizer2 *norm2, UChar32 c); + +/** + * Tests if the character always has a normalization boundary after it, + * regardless of context. + * For details see the Normalizer2 base class documentation. + * @param norm2 UNormalizer2 instance + * @param c character to test + * @return true if c has a normalization boundary after it + * @stable ICU 4.4 + */ +U_CAPI UBool U_EXPORT2 +unorm2_hasBoundaryAfter(const UNormalizer2 *norm2, UChar32 c); + +/** + * Tests if the character is normalization-inert. + * For details see the Normalizer2 base class documentation. + * @param norm2 UNormalizer2 instance + * @param c character to test + * @return true if c is normalization-inert + * @stable ICU 4.4 + */ +U_CAPI UBool U_EXPORT2 +unorm2_isInert(const UNormalizer2 *norm2, UChar32 c); + +/** + * Compares two strings for canonical equivalence. + * Further options include case-insensitive comparison and + * code point order (as opposed to code unit order). + * + * Canonical equivalence between two strings is defined as their normalized + * forms (NFD or NFC) being identical. + * This function compares strings incrementally instead of normalizing + * (and optionally case-folding) both strings entirely, + * improving performance significantly. + * + * Bulk normalization is only necessary if the strings do not fulfill the FCD + * conditions. Only in this case, and only if the strings are relatively long, + * is memory allocated temporarily. + * For FCD strings and short non-FCD strings there is no memory allocation. + * + * Semantically, this is equivalent to + * strcmp[CodePointOrder](NFD(foldCase(NFD(s1))), NFD(foldCase(NFD(s2)))) + * where code point order and foldCase are all optional. + * + * UAX 21 2.5 Caseless Matching specifies that for a canonical caseless match + * the case folding must be performed first, then the normalization. + * + * @param s1 First source string. + * @param length1 Length of first source string, or -1 if NUL-terminated. + * + * @param s2 Second source string. + * @param length2 Length of second source string, or -1 if NUL-terminated. + * + * @param options A bit set of options: + * - U_FOLD_CASE_DEFAULT or 0 is used for default options: + * Case-sensitive comparison in code unit order, and the input strings + * are quick-checked for FCD. + * + * - UNORM_INPUT_IS_FCD + * Set if the caller knows that both s1 and s2 fulfill the FCD conditions. + * If not set, the function will quickCheck for FCD + * and normalize if necessary. + * + * - U_COMPARE_CODE_POINT_ORDER + * Set to choose code point order instead of code unit order + * (see u_strCompare for details). + * + * - U_COMPARE_IGNORE_CASE + * Set to compare strings case-insensitively using case folding, + * instead of case-sensitively. + * If set, then the following case folding options are used. + * + * - Options as used with case-insensitive comparisons, currently: + * + * - U_FOLD_CASE_EXCLUDE_SPECIAL_I + * (see u_strCaseCompare for details) + * + * - regular normalization options shifted left by UNORM_COMPARE_NORM_OPTIONS_SHIFT + * + * @param pErrorCode ICU error code in/out parameter. + * Must fulfill U_SUCCESS before the function call. + * @return <0 or 0 or >0 as usual for string comparisons + * + * @see unorm_normalize + * @see UNORM_FCD + * @see u_strCompare + * @see u_strCaseCompare + * + * @stable ICU 2.2 + */ +U_CAPI int32_t U_EXPORT2 +unorm_compare(const UChar *s1, int32_t length1, + const UChar *s2, int32_t length2, + uint32_t options, + UErrorCode *pErrorCode); + +#endif /* !UCONFIG_NO_NORMALIZATION */ +#endif /* __UNORM2_H__ */ diff --git a/intl/icu/source/common/unicode/uobject.h b/intl/icu/source/common/unicode/uobject.h new file mode 100644 index 0000000000..f53ec1d111 --- /dev/null +++ b/intl/icu/source/common/unicode/uobject.h @@ -0,0 +1,324 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2002-2012, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: uobject.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002jun26 +* created by: Markus W. Scherer +*/ + +#ifndef __UOBJECT_H__ +#define __UOBJECT_H__ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/platform.h" + +/** + * \file + * \brief C++ API: Common ICU base class UObject. + */ + +/** + * \def U_NO_THROW + * Since ICU 64, use noexcept instead. + * + * Previously, define this to define the throw() specification so + * certain functions do not throw any exceptions + * + * UMemory operator new methods should have the throw() specification + * appended to them, so that the compiler adds the additional nullptr check + * before calling constructors. Without, if operator new returns nullptr the + * constructor is still called, and if the constructor references member + * data, (which it typically does), the result is a segmentation violation. + * + * @stable ICU 4.2. Since ICU 64, Use noexcept instead. See ICU-20422. + */ +#ifndef U_NO_THROW +#define U_NO_THROW noexcept +#endif + +/*===========================================================================*/ +/* UClassID-based RTTI */ +/*===========================================================================*/ + +/** + * UClassID is used to identify classes without using the compiler's RTTI. + * This was used before C++ compilers consistently supported RTTI. + * ICU 4.6 requires compiler RTTI to be turned on. + * + * Each class hierarchy which needs + * to implement polymorphic clone() or operator==() defines two methods, + * described in detail below. UClassID values can be compared using + * operator==(). Nothing else should be done with them. + * + * \par + * In class hierarchies that implement "poor man's RTTI", + * each concrete subclass implements getDynamicClassID() in the same way: + * + * \code + * class Derived { + * public: + * virtual UClassID getDynamicClassID() const + * { return Derived::getStaticClassID(); } + * } + * \endcode + * + * Each concrete class implements getStaticClassID() as well, which allows + * clients to test for a specific type. + * + * \code + * class Derived { + * public: + * static UClassID U_EXPORT2 getStaticClassID(); + * private: + * static char fgClassID; + * } + * + * // In Derived.cpp: + * UClassID Derived::getStaticClassID() + * { return (UClassID)&Derived::fgClassID; } + * char Derived::fgClassID = 0; // Value is irrelevant + * \endcode + * @stable ICU 2.0 + */ +typedef void* UClassID; + +U_NAMESPACE_BEGIN + +/** + * UMemory is the common ICU base class. + * All other ICU C++ classes are derived from UMemory (starting with ICU 2.4). + * + * This is primarily to make it possible and simple to override the + * C++ memory management by adding new/delete operators to this base class. + * + * To override ALL ICU memory management, including that from plain C code, + * replace the allocation functions declared in cmemory.h + * + * UMemory does not contain any virtual functions. + * Common "boilerplate" functions are defined in UObject. + * + * @stable ICU 2.4 + */ +class U_COMMON_API UMemory { +public: + +/* test versions for debugging shaper heap memory problems */ +#ifdef SHAPER_MEMORY_DEBUG + static void * NewArray(int size, int count); + static void * GrowArray(void * array, int newSize ); + static void FreeArray(void * array ); +#endif + +#if U_OVERRIDE_CXX_ALLOCATION + /** + * Override for ICU4C C++ memory management. + * simple, non-class types are allocated using the macros in common/cmemory.h + * (uprv_malloc(), uprv_free(), uprv_realloc()); + * they or something else could be used here to implement C++ new/delete + * for ICU4C C++ classes + * @stable ICU 2.4 + */ + static void * U_EXPORT2 operator new(size_t size) noexcept; + + /** + * Override for ICU4C C++ memory management. + * See new(). + * @stable ICU 2.4 + */ + static void * U_EXPORT2 operator new[](size_t size) noexcept; + + /** + * Override for ICU4C C++ memory management. + * simple, non-class types are allocated using the macros in common/cmemory.h + * (uprv_malloc(), uprv_free(), uprv_realloc()); + * they or something else could be used here to implement C++ new/delete + * for ICU4C C++ classes + * @stable ICU 2.4 + */ + static void U_EXPORT2 operator delete(void *p) noexcept; + + /** + * Override for ICU4C C++ memory management. + * See delete(). + * @stable ICU 2.4 + */ + static void U_EXPORT2 operator delete[](void *p) noexcept; + +#if U_HAVE_PLACEMENT_NEW + /** + * Override for ICU4C C++ memory management for STL. + * See new(). + * @stable ICU 2.6 + */ + static inline void * U_EXPORT2 operator new(size_t, void *ptr) noexcept { return ptr; } + + /** + * Override for ICU4C C++ memory management for STL. + * See delete(). + * @stable ICU 2.6 + */ + static inline void U_EXPORT2 operator delete(void *, void *) noexcept {} +#endif /* U_HAVE_PLACEMENT_NEW */ +#if U_HAVE_DEBUG_LOCATION_NEW + /** + * This method overrides the MFC debug version of the operator new + * + * @param size The requested memory size + * @param file The file where the allocation was requested + * @param line The line where the allocation was requested + */ + static void * U_EXPORT2 operator new(size_t size, const char* file, int line) noexcept; + /** + * This method provides a matching delete for the MFC debug new + * + * @param p The pointer to the allocated memory + * @param file The file where the allocation was requested + * @param line The line where the allocation was requested + */ + static void U_EXPORT2 operator delete(void* p, const char* file, int line) noexcept; +#endif /* U_HAVE_DEBUG_LOCATION_NEW */ +#endif /* U_OVERRIDE_CXX_ALLOCATION */ + + /* + * Assignment operator not declared. The compiler will provide one + * which does nothing since this class does not contain any data members. + * API/code coverage may show the assignment operator as present and + * untested - ignore. + * Subclasses need this assignment operator if they use compiler-provided + * assignment operators of their own. An alternative to not declaring one + * here would be to declare and empty-implement a protected or public one. + UMemory &UMemory::operator=(const UMemory &); + */ +}; + +/** + * UObject is the common ICU "boilerplate" class. + * UObject inherits UMemory (starting with ICU 2.4), + * and all other public ICU C++ classes + * are derived from UObject (starting with ICU 2.2). + * + * UObject contains common virtual functions, in particular a virtual destructor. + * + * The clone() function is not available in UObject because it is not + * implemented by all ICU classes. + * Many ICU services provide a clone() function for their class trees, + * defined on the service's C++ base class + * (which itself is a subclass of UObject). + * + * @stable ICU 2.2 + */ +class U_COMMON_API UObject : public UMemory { +public: + /** + * Destructor. + * + * @stable ICU 2.2 + */ + virtual ~UObject(); + + /** + * ICU4C "poor man's RTTI", returns a UClassID for the actual ICU class. + * The base class implementation returns a dummy value. + * + * Use compiler RTTI rather than ICU's "poor man's RTTI". + * Since ICU 4.6, new ICU C++ class hierarchies do not implement "poor man's RTTI". + * + * @stable ICU 2.2 + */ + virtual UClassID getDynamicClassID() const; + +protected: + // the following functions are protected to prevent instantiation and + // direct use of UObject itself + + // default constructor + // inline UObject() {} + + // copy constructor + // inline UObject(const UObject &other) {} + +#if 0 + // TODO Sometime in the future. Implement operator==(). + // (This comment inserted in 2.2) + // some or all of the following "boilerplate" functions may be made public + // in a future ICU4C release when all subclasses implement them + + // assignment operator + // (not virtual, see "Taligent's Guide to Designing Programs" pp.73..74) + // commented out because the implementation is the same as a compiler's default + // UObject &operator=(const UObject &other) { return *this; } + + // comparison operators + virtual inline bool operator==(const UObject &other) const { return this==&other; } + inline bool operator!=(const UObject &other) const { return !operator==(other); } + + // clone() commented out from the base class: + // some compilers do not support co-variant return types + // (i.e., subclasses would have to return UObject * as well, instead of SubClass *) + // see also UObject class documentation. + // virtual UObject *clone() const; +#endif + + /* + * Assignment operator not declared. The compiler will provide one + * which does nothing since this class does not contain any data members. + * API/code coverage may show the assignment operator as present and + * untested - ignore. + * Subclasses need this assignment operator if they use compiler-provided + * assignment operators of their own. An alternative to not declaring one + * here would be to declare and empty-implement a protected or public one. + UObject &UObject::operator=(const UObject &); + */ +}; + +#ifndef U_HIDE_INTERNAL_API +/** + * This is a simple macro to add ICU RTTI to an ICU object implementation. + * This does not go into the header. This should only be used in *.cpp files. + * + * @param myClass The name of the class that needs RTTI defined. + * @internal + */ +#define UOBJECT_DEFINE_RTTI_IMPLEMENTATION(myClass) \ + UClassID U_EXPORT2 myClass::getStaticClassID() { \ + static char classID = 0; \ + return (UClassID)&classID; \ + } \ + UClassID myClass::getDynamicClassID() const \ + { return myClass::getStaticClassID(); } + + +/** + * This macro adds ICU RTTI to an ICU abstract class implementation. + * This macro should be invoked in *.cpp files. The corresponding + * header should declare getStaticClassID. + * + * @param myClass The name of the class that needs RTTI defined. + * @internal + */ +#define UOBJECT_DEFINE_ABSTRACT_RTTI_IMPLEMENTATION(myClass) \ + UClassID U_EXPORT2 myClass::getStaticClassID() { \ + static char classID = 0; \ + return (UClassID)&classID; \ + } + +#endif /* U_HIDE_INTERNAL_API */ + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/urename.h b/intl/icu/source/common/unicode/urename.h new file mode 100644 index 0000000000..b35df45380 --- /dev/null +++ b/intl/icu/source/common/unicode/urename.h @@ -0,0 +1,1975 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2002-2016, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* +* file name: urename.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* Created by: Perl script tools/genren.pl written by Vladimir Weinstein +* +* Contains data for renaming ICU exports. +* Gets included by umachine.h +* +* THIS FILE IS MACHINE-GENERATED, DON'T PLAY WITH IT IF YOU DON'T KNOW WHAT +* YOU ARE DOING, OTHERWISE VERY BAD THINGS WILL HAPPEN! +*/ + +#ifndef URENAME_H +#define URENAME_H + +/* U_DISABLE_RENAMING can be defined in the following ways: + * - when running configure, e.g. + * runConfigureICU Linux --disable-renaming + * - by changing the default setting of U_DISABLE_RENAMING in uconfig.h + */ + +#include "unicode/uconfig.h" + +#if !U_DISABLE_RENAMING + +// Disable Renaming for Visual Studio's IntelliSense feature, so that 'Go-to-Definition' (F12) will work. +#if !(defined(_MSC_VER) && defined(__INTELLISENSE__)) + +/* We need the U_ICU_ENTRY_POINT_RENAME definition. There's a default one in unicode/uvernum.h we can use, but we will give + the platform a chance to define it first. + Normally (if utypes.h or umachine.h was included first) this will not be necessary as it will already be defined. + */ + +#ifndef U_ICU_ENTRY_POINT_RENAME +#include "unicode/umachine.h" +#endif + +/* If we still don't have U_ICU_ENTRY_POINT_RENAME use the default. */ +#ifndef U_ICU_ENTRY_POINT_RENAME +#include "unicode/uvernum.h" +#endif + +/* Error out before the following defines cause very strange and unexpected code breakage */ +#ifndef U_ICU_ENTRY_POINT_RENAME +#error U_ICU_ENTRY_POINT_RENAME is not defined - cannot continue. Consider defining U_DISABLE_RENAMING if renaming should not be used. +#endif + + +/* C exports renaming data */ + +#define CreateLSTMBreakEngine U_ICU_ENTRY_POINT_RENAME(CreateLSTMBreakEngine) +#define CreateLSTMData U_ICU_ENTRY_POINT_RENAME(CreateLSTMData) +#define CreateLSTMDataForScript U_ICU_ENTRY_POINT_RENAME(CreateLSTMDataForScript) +#define DeleteLSTMData U_ICU_ENTRY_POINT_RENAME(DeleteLSTMData) +#define LSTMDataName U_ICU_ENTRY_POINT_RENAME(LSTMDataName) +#define T_CString_int64ToString U_ICU_ENTRY_POINT_RENAME(T_CString_int64ToString) +#define T_CString_integerToString U_ICU_ENTRY_POINT_RENAME(T_CString_integerToString) +#define T_CString_stringToInteger U_ICU_ENTRY_POINT_RENAME(T_CString_stringToInteger) +#define T_CString_toLowerCase U_ICU_ENTRY_POINT_RENAME(T_CString_toLowerCase) +#define T_CString_toUpperCase U_ICU_ENTRY_POINT_RENAME(T_CString_toUpperCase) +#define UCNV_FROM_U_CALLBACK_ESCAPE U_ICU_ENTRY_POINT_RENAME(UCNV_FROM_U_CALLBACK_ESCAPE) +#define UCNV_FROM_U_CALLBACK_SKIP U_ICU_ENTRY_POINT_RENAME(UCNV_FROM_U_CALLBACK_SKIP) +#define UCNV_FROM_U_CALLBACK_STOP U_ICU_ENTRY_POINT_RENAME(UCNV_FROM_U_CALLBACK_STOP) +#define UCNV_FROM_U_CALLBACK_SUBSTITUTE U_ICU_ENTRY_POINT_RENAME(UCNV_FROM_U_CALLBACK_SUBSTITUTE) +#define UCNV_TO_U_CALLBACK_ESCAPE U_ICU_ENTRY_POINT_RENAME(UCNV_TO_U_CALLBACK_ESCAPE) +#define UCNV_TO_U_CALLBACK_SKIP U_ICU_ENTRY_POINT_RENAME(UCNV_TO_U_CALLBACK_SKIP) +#define UCNV_TO_U_CALLBACK_STOP U_ICU_ENTRY_POINT_RENAME(UCNV_TO_U_CALLBACK_STOP) +#define UCNV_TO_U_CALLBACK_SUBSTITUTE U_ICU_ENTRY_POINT_RENAME(UCNV_TO_U_CALLBACK_SUBSTITUTE) +#define UDataMemory_createNewInstance U_ICU_ENTRY_POINT_RENAME(UDataMemory_createNewInstance) +#define UDataMemory_init U_ICU_ENTRY_POINT_RENAME(UDataMemory_init) +#define UDataMemory_isLoaded U_ICU_ENTRY_POINT_RENAME(UDataMemory_isLoaded) +#define UDataMemory_normalizeDataPointer U_ICU_ENTRY_POINT_RENAME(UDataMemory_normalizeDataPointer) +#define UDataMemory_setData U_ICU_ENTRY_POINT_RENAME(UDataMemory_setData) +#define UDatamemory_assign U_ICU_ENTRY_POINT_RENAME(UDatamemory_assign) +#define _ASCIIData U_ICU_ENTRY_POINT_RENAME(_ASCIIData) +#define _Bocu1Data U_ICU_ENTRY_POINT_RENAME(_Bocu1Data) +#define _CESU8Data U_ICU_ENTRY_POINT_RENAME(_CESU8Data) +#define _CompoundTextData U_ICU_ENTRY_POINT_RENAME(_CompoundTextData) +#define _HZData U_ICU_ENTRY_POINT_RENAME(_HZData) +#define _IMAPData U_ICU_ENTRY_POINT_RENAME(_IMAPData) +#define _ISCIIData U_ICU_ENTRY_POINT_RENAME(_ISCIIData) +#define _ISO2022Data U_ICU_ENTRY_POINT_RENAME(_ISO2022Data) +#define _LMBCSData1 U_ICU_ENTRY_POINT_RENAME(_LMBCSData1) +#define _LMBCSData11 U_ICU_ENTRY_POINT_RENAME(_LMBCSData11) +#define _LMBCSData16 U_ICU_ENTRY_POINT_RENAME(_LMBCSData16) +#define _LMBCSData17 U_ICU_ENTRY_POINT_RENAME(_LMBCSData17) +#define _LMBCSData18 U_ICU_ENTRY_POINT_RENAME(_LMBCSData18) +#define _LMBCSData19 U_ICU_ENTRY_POINT_RENAME(_LMBCSData19) +#define _LMBCSData2 U_ICU_ENTRY_POINT_RENAME(_LMBCSData2) +#define _LMBCSData3 U_ICU_ENTRY_POINT_RENAME(_LMBCSData3) +#define _LMBCSData4 U_ICU_ENTRY_POINT_RENAME(_LMBCSData4) +#define _LMBCSData5 U_ICU_ENTRY_POINT_RENAME(_LMBCSData5) +#define _LMBCSData6 U_ICU_ENTRY_POINT_RENAME(_LMBCSData6) +#define _LMBCSData8 U_ICU_ENTRY_POINT_RENAME(_LMBCSData8) +#define _Latin1Data U_ICU_ENTRY_POINT_RENAME(_Latin1Data) +#define _MBCSData U_ICU_ENTRY_POINT_RENAME(_MBCSData) +#define _SCSUData U_ICU_ENTRY_POINT_RENAME(_SCSUData) +#define _UTF16BEData U_ICU_ENTRY_POINT_RENAME(_UTF16BEData) +#define _UTF16Data U_ICU_ENTRY_POINT_RENAME(_UTF16Data) +#define _UTF16LEData U_ICU_ENTRY_POINT_RENAME(_UTF16LEData) +#define _UTF16v2Data U_ICU_ENTRY_POINT_RENAME(_UTF16v2Data) +#define _UTF32BEData U_ICU_ENTRY_POINT_RENAME(_UTF32BEData) +#define _UTF32Data U_ICU_ENTRY_POINT_RENAME(_UTF32Data) +#define _UTF32LEData U_ICU_ENTRY_POINT_RENAME(_UTF32LEData) +#define _UTF7Data U_ICU_ENTRY_POINT_RENAME(_UTF7Data) +#define _UTF8Data U_ICU_ENTRY_POINT_RENAME(_UTF8Data) +#define _isUnicodeLocaleTypeSubtag U_ICU_ENTRY_POINT_RENAME(_isUnicodeLocaleTypeSubtag) +#define allowedHourFormatsCleanup U_ICU_ENTRY_POINT_RENAME(allowedHourFormatsCleanup) +#define cmemory_cleanup U_ICU_ENTRY_POINT_RENAME(cmemory_cleanup) +#define dayPeriodRulesCleanup U_ICU_ENTRY_POINT_RENAME(dayPeriodRulesCleanup) +#define deleteAllowedHourFormats U_ICU_ENTRY_POINT_RENAME(deleteAllowedHourFormats) +#define gTimeZoneFilesInitOnce U_ICU_ENTRY_POINT_RENAME(gTimeZoneFilesInitOnce) +#define initNumsysNames U_ICU_ENTRY_POINT_RENAME(initNumsysNames) +#define izrule_clone U_ICU_ENTRY_POINT_RENAME(izrule_clone) +#define izrule_close U_ICU_ENTRY_POINT_RENAME(izrule_close) +#define izrule_equals U_ICU_ENTRY_POINT_RENAME(izrule_equals) +#define izrule_getDSTSavings U_ICU_ENTRY_POINT_RENAME(izrule_getDSTSavings) +#define izrule_getDynamicClassID U_ICU_ENTRY_POINT_RENAME(izrule_getDynamicClassID) +#define izrule_getFinalStart U_ICU_ENTRY_POINT_RENAME(izrule_getFinalStart) +#define izrule_getFirstStart U_ICU_ENTRY_POINT_RENAME(izrule_getFirstStart) +#define izrule_getName U_ICU_ENTRY_POINT_RENAME(izrule_getName) +#define izrule_getNextStart U_ICU_ENTRY_POINT_RENAME(izrule_getNextStart) +#define izrule_getPreviousStart U_ICU_ENTRY_POINT_RENAME(izrule_getPreviousStart) +#define izrule_getRawOffset U_ICU_ENTRY_POINT_RENAME(izrule_getRawOffset) +#define izrule_getStaticClassID U_ICU_ENTRY_POINT_RENAME(izrule_getStaticClassID) +#define izrule_isEquivalentTo U_ICU_ENTRY_POINT_RENAME(izrule_isEquivalentTo) +#define izrule_open U_ICU_ENTRY_POINT_RENAME(izrule_open) +#define locale_getKeywordsStart U_ICU_ENTRY_POINT_RENAME(locale_getKeywordsStart) +#define locale_get_default U_ICU_ENTRY_POINT_RENAME(locale_get_default) +#define locale_set_default U_ICU_ENTRY_POINT_RENAME(locale_set_default) +#define numSysCleanup U_ICU_ENTRY_POINT_RENAME(numSysCleanup) +#define rbbi_cleanup U_ICU_ENTRY_POINT_RENAME(rbbi_cleanup) +#define pl_addFontRun U_ICU_ENTRY_POINT_RENAME(pl_addFontRun) +#define pl_addLocaleRun U_ICU_ENTRY_POINT_RENAME(pl_addLocaleRun) +#define pl_addValueRun U_ICU_ENTRY_POINT_RENAME(pl_addValueRun) +#define pl_close U_ICU_ENTRY_POINT_RENAME(pl_close) +#define pl_closeFontRuns U_ICU_ENTRY_POINT_RENAME(pl_closeFontRuns) +#define pl_closeLine U_ICU_ENTRY_POINT_RENAME(pl_closeLine) +#define pl_closeLocaleRuns U_ICU_ENTRY_POINT_RENAME(pl_closeLocaleRuns) +#define pl_closeValueRuns U_ICU_ENTRY_POINT_RENAME(pl_closeValueRuns) +#define pl_countLineRuns U_ICU_ENTRY_POINT_RENAME(pl_countLineRuns) +#define pl_create U_ICU_ENTRY_POINT_RENAME(pl_create) +#define pl_getAscent U_ICU_ENTRY_POINT_RENAME(pl_getAscent) +#define pl_getDescent U_ICU_ENTRY_POINT_RENAME(pl_getDescent) +#define pl_getFontRunCount U_ICU_ENTRY_POINT_RENAME(pl_getFontRunCount) +#define pl_getFontRunFont U_ICU_ENTRY_POINT_RENAME(pl_getFontRunFont) +#define pl_getFontRunLastLimit U_ICU_ENTRY_POINT_RENAME(pl_getFontRunLastLimit) +#define pl_getFontRunLimit U_ICU_ENTRY_POINT_RENAME(pl_getFontRunLimit) +#define pl_getLeading U_ICU_ENTRY_POINT_RENAME(pl_getLeading) +#define pl_getLineAscent U_ICU_ENTRY_POINT_RENAME(pl_getLineAscent) +#define pl_getLineDescent U_ICU_ENTRY_POINT_RENAME(pl_getLineDescent) +#define pl_getLineLeading U_ICU_ENTRY_POINT_RENAME(pl_getLineLeading) +#define pl_getLineVisualRun U_ICU_ENTRY_POINT_RENAME(pl_getLineVisualRun) +#define pl_getLineWidth U_ICU_ENTRY_POINT_RENAME(pl_getLineWidth) +#define pl_getLocaleRunCount U_ICU_ENTRY_POINT_RENAME(pl_getLocaleRunCount) +#define pl_getLocaleRunLastLimit U_ICU_ENTRY_POINT_RENAME(pl_getLocaleRunLastLimit) +#define pl_getLocaleRunLimit U_ICU_ENTRY_POINT_RENAME(pl_getLocaleRunLimit) +#define pl_getLocaleRunLocale U_ICU_ENTRY_POINT_RENAME(pl_getLocaleRunLocale) +#define pl_getParagraphLevel U_ICU_ENTRY_POINT_RENAME(pl_getParagraphLevel) +#define pl_getTextDirection U_ICU_ENTRY_POINT_RENAME(pl_getTextDirection) +#define pl_getValueRunCount U_ICU_ENTRY_POINT_RENAME(pl_getValueRunCount) +#define pl_getValueRunLastLimit U_ICU_ENTRY_POINT_RENAME(pl_getValueRunLastLimit) +#define pl_getValueRunLimit U_ICU_ENTRY_POINT_RENAME(pl_getValueRunLimit) +#define pl_getValueRunValue U_ICU_ENTRY_POINT_RENAME(pl_getValueRunValue) +#define pl_getVisualRunAscent U_ICU_ENTRY_POINT_RENAME(pl_getVisualRunAscent) +#define pl_getVisualRunDescent U_ICU_ENTRY_POINT_RENAME(pl_getVisualRunDescent) +#define pl_getVisualRunDirection U_ICU_ENTRY_POINT_RENAME(pl_getVisualRunDirection) +#define pl_getVisualRunFont U_ICU_ENTRY_POINT_RENAME(pl_getVisualRunFont) +#define pl_getVisualRunGlyphCount U_ICU_ENTRY_POINT_RENAME(pl_getVisualRunGlyphCount) +#define pl_getVisualRunGlyphToCharMap U_ICU_ENTRY_POINT_RENAME(pl_getVisualRunGlyphToCharMap) +#define pl_getVisualRunGlyphs U_ICU_ENTRY_POINT_RENAME(pl_getVisualRunGlyphs) +#define pl_getVisualRunLeading U_ICU_ENTRY_POINT_RENAME(pl_getVisualRunLeading) +#define pl_getVisualRunPositions U_ICU_ENTRY_POINT_RENAME(pl_getVisualRunPositions) +#define pl_isComplex U_ICU_ENTRY_POINT_RENAME(pl_isComplex) +#define pl_nextLine U_ICU_ENTRY_POINT_RENAME(pl_nextLine) +#define pl_openEmptyFontRuns U_ICU_ENTRY_POINT_RENAME(pl_openEmptyFontRuns) +#define pl_openEmptyLocaleRuns U_ICU_ENTRY_POINT_RENAME(pl_openEmptyLocaleRuns) +#define pl_openEmptyValueRuns U_ICU_ENTRY_POINT_RENAME(pl_openEmptyValueRuns) +#define pl_openFontRuns U_ICU_ENTRY_POINT_RENAME(pl_openFontRuns) +#define pl_openLocaleRuns U_ICU_ENTRY_POINT_RENAME(pl_openLocaleRuns) +#define pl_openValueRuns U_ICU_ENTRY_POINT_RENAME(pl_openValueRuns) +#define pl_reflow U_ICU_ENTRY_POINT_RENAME(pl_reflow) +#define pl_resetFontRuns U_ICU_ENTRY_POINT_RENAME(pl_resetFontRuns) +#define pl_resetLocaleRuns U_ICU_ENTRY_POINT_RENAME(pl_resetLocaleRuns) +#define pl_resetValueRuns U_ICU_ENTRY_POINT_RENAME(pl_resetValueRuns) +#define res_countArrayItems U_ICU_ENTRY_POINT_RENAME(res_countArrayItems) +#define res_findResource U_ICU_ENTRY_POINT_RENAME(res_findResource) +#define res_getAlias U_ICU_ENTRY_POINT_RENAME(res_getAlias) +#define res_getArrayItem U_ICU_ENTRY_POINT_RENAME(res_getArrayItem) +#define res_getBinaryNoTrace U_ICU_ENTRY_POINT_RENAME(res_getBinaryNoTrace) +#define res_getIntVectorNoTrace U_ICU_ENTRY_POINT_RENAME(res_getIntVectorNoTrace) +#define res_getPublicType U_ICU_ENTRY_POINT_RENAME(res_getPublicType) +#define res_getResource U_ICU_ENTRY_POINT_RENAME(res_getResource) +#define res_getStringNoTrace U_ICU_ENTRY_POINT_RENAME(res_getStringNoTrace) +#define res_getTableItemByIndex U_ICU_ENTRY_POINT_RENAME(res_getTableItemByIndex) +#define res_getTableItemByKey U_ICU_ENTRY_POINT_RENAME(res_getTableItemByKey) +#define res_load U_ICU_ENTRY_POINT_RENAME(res_load) +#define res_read U_ICU_ENTRY_POINT_RENAME(res_read) +#define res_unload U_ICU_ENTRY_POINT_RENAME(res_unload) +#define u_UCharsToChars U_ICU_ENTRY_POINT_RENAME(u_UCharsToChars) +#define u_asciiToUpper U_ICU_ENTRY_POINT_RENAME(u_asciiToUpper) +#define u_austrcpy U_ICU_ENTRY_POINT_RENAME(u_austrcpy) +#define u_austrncpy U_ICU_ENTRY_POINT_RENAME(u_austrncpy) +#define u_caseInsensitivePrefixMatch U_ICU_ENTRY_POINT_RENAME(u_caseInsensitivePrefixMatch) +#define u_catclose U_ICU_ENTRY_POINT_RENAME(u_catclose) +#define u_catgets U_ICU_ENTRY_POINT_RENAME(u_catgets) +#define u_catopen U_ICU_ENTRY_POINT_RENAME(u_catopen) +#define u_charAge U_ICU_ENTRY_POINT_RENAME(u_charAge) +#define u_charDigitValue U_ICU_ENTRY_POINT_RENAME(u_charDigitValue) +#define u_charDirection U_ICU_ENTRY_POINT_RENAME(u_charDirection) +#define u_charFromName U_ICU_ENTRY_POINT_RENAME(u_charFromName) +#define u_charMirror U_ICU_ENTRY_POINT_RENAME(u_charMirror) +#define u_charName U_ICU_ENTRY_POINT_RENAME(u_charName) +#define u_charType U_ICU_ENTRY_POINT_RENAME(u_charType) +#define u_charsToUChars U_ICU_ENTRY_POINT_RENAME(u_charsToUChars) +#define u_cleanup U_ICU_ENTRY_POINT_RENAME(u_cleanup) +#define u_countChar32 U_ICU_ENTRY_POINT_RENAME(u_countChar32) +#define u_digit U_ICU_ENTRY_POINT_RENAME(u_digit) +#define u_enumCharNames U_ICU_ENTRY_POINT_RENAME(u_enumCharNames) +#define u_enumCharTypes U_ICU_ENTRY_POINT_RENAME(u_enumCharTypes) +#define u_errorName U_ICU_ENTRY_POINT_RENAME(u_errorName) +#define u_fadopt U_ICU_ENTRY_POINT_RENAME(u_fadopt) +#define u_fclose U_ICU_ENTRY_POINT_RENAME(u_fclose) +#define u_feof U_ICU_ENTRY_POINT_RENAME(u_feof) +#define u_fflush U_ICU_ENTRY_POINT_RENAME(u_fflush) +#define u_fgetConverter U_ICU_ENTRY_POINT_RENAME(u_fgetConverter) +#define u_fgetNumberFormat U_ICU_ENTRY_POINT_RENAME(u_fgetNumberFormat) +#define u_fgetc U_ICU_ENTRY_POINT_RENAME(u_fgetc) +#define u_fgetcodepage U_ICU_ENTRY_POINT_RENAME(u_fgetcodepage) +#define u_fgetcx U_ICU_ENTRY_POINT_RENAME(u_fgetcx) +#define u_fgetfile U_ICU_ENTRY_POINT_RENAME(u_fgetfile) +#define u_fgetlocale U_ICU_ENTRY_POINT_RENAME(u_fgetlocale) +#define u_fgets U_ICU_ENTRY_POINT_RENAME(u_fgets) +#define u_file_read U_ICU_ENTRY_POINT_RENAME(u_file_read) +#define u_file_write U_ICU_ENTRY_POINT_RENAME(u_file_write) +#define u_file_write_flush U_ICU_ENTRY_POINT_RENAME(u_file_write_flush) +#define u_finit U_ICU_ENTRY_POINT_RENAME(u_finit) +#define u_flushDefaultConverter U_ICU_ENTRY_POINT_RENAME(u_flushDefaultConverter) +#define u_foldCase U_ICU_ENTRY_POINT_RENAME(u_foldCase) +#define u_fopen U_ICU_ENTRY_POINT_RENAME(u_fopen) +#define u_fopen_u U_ICU_ENTRY_POINT_RENAME(u_fopen_u) +#define u_forDigit U_ICU_ENTRY_POINT_RENAME(u_forDigit) +#define u_formatMessage U_ICU_ENTRY_POINT_RENAME(u_formatMessage) +#define u_formatMessageWithError U_ICU_ENTRY_POINT_RENAME(u_formatMessageWithError) +#define u_fprintf U_ICU_ENTRY_POINT_RENAME(u_fprintf) +#define u_fprintf_u U_ICU_ENTRY_POINT_RENAME(u_fprintf_u) +#define u_fputc U_ICU_ENTRY_POINT_RENAME(u_fputc) +#define u_fputs U_ICU_ENTRY_POINT_RENAME(u_fputs) +#define u_frewind U_ICU_ENTRY_POINT_RENAME(u_frewind) +#define u_fscanf U_ICU_ENTRY_POINT_RENAME(u_fscanf) +#define u_fscanf_u U_ICU_ENTRY_POINT_RENAME(u_fscanf_u) +#define u_fsetcodepage U_ICU_ENTRY_POINT_RENAME(u_fsetcodepage) +#define u_fsetlocale U_ICU_ENTRY_POINT_RENAME(u_fsetlocale) +#define u_fsettransliterator U_ICU_ENTRY_POINT_RENAME(u_fsettransliterator) +#define u_fstropen U_ICU_ENTRY_POINT_RENAME(u_fstropen) +#define u_fungetc U_ICU_ENTRY_POINT_RENAME(u_fungetc) +#define u_getBidiPairedBracket U_ICU_ENTRY_POINT_RENAME(u_getBidiPairedBracket) +#define u_getBinaryPropertySet U_ICU_ENTRY_POINT_RENAME(u_getBinaryPropertySet) +#define u_getCombiningClass U_ICU_ENTRY_POINT_RENAME(u_getCombiningClass) +#define u_getDataDirectory U_ICU_ENTRY_POINT_RENAME(u_getDataDirectory) +#define u_getDataVersion U_ICU_ENTRY_POINT_RENAME(u_getDataVersion) +#define u_getDefaultConverter U_ICU_ENTRY_POINT_RENAME(u_getDefaultConverter) +#define u_getFC_NFKC_Closure U_ICU_ENTRY_POINT_RENAME(u_getFC_NFKC_Closure) +#define u_getISOComment U_ICU_ENTRY_POINT_RENAME(u_getISOComment) +#define u_getIntPropertyMap U_ICU_ENTRY_POINT_RENAME(u_getIntPropertyMap) +#define u_getIntPropertyMaxValue U_ICU_ENTRY_POINT_RENAME(u_getIntPropertyMaxValue) +#define u_getIntPropertyMinValue U_ICU_ENTRY_POINT_RENAME(u_getIntPropertyMinValue) +#define u_getIntPropertyValue U_ICU_ENTRY_POINT_RENAME(u_getIntPropertyValue) +#define u_getMainProperties U_ICU_ENTRY_POINT_RENAME(u_getMainProperties) +#define u_getNumericValue U_ICU_ENTRY_POINT_RENAME(u_getNumericValue) +#define u_getPropertyEnum U_ICU_ENTRY_POINT_RENAME(u_getPropertyEnum) +#define u_getPropertyName U_ICU_ENTRY_POINT_RENAME(u_getPropertyName) +#define u_getPropertyValueEnum U_ICU_ENTRY_POINT_RENAME(u_getPropertyValueEnum) +#define u_getPropertyValueName U_ICU_ENTRY_POINT_RENAME(u_getPropertyValueName) +#define u_getTimeZoneFilesDirectory U_ICU_ENTRY_POINT_RENAME(u_getTimeZoneFilesDirectory) +#define u_getUnicodeProperties U_ICU_ENTRY_POINT_RENAME(u_getUnicodeProperties) +#define u_getUnicodeVersion U_ICU_ENTRY_POINT_RENAME(u_getUnicodeVersion) +#define u_getVersion U_ICU_ENTRY_POINT_RENAME(u_getVersion) +#define u_get_stdout U_ICU_ENTRY_POINT_RENAME(u_get_stdout) +#define u_hasBinaryProperty U_ICU_ENTRY_POINT_RENAME(u_hasBinaryProperty) +#define u_init U_ICU_ENTRY_POINT_RENAME(u_init) +#define u_isIDIgnorable U_ICU_ENTRY_POINT_RENAME(u_isIDIgnorable) +#define u_isIDPart U_ICU_ENTRY_POINT_RENAME(u_isIDPart) +#define u_isIDStart U_ICU_ENTRY_POINT_RENAME(u_isIDStart) +#define u_isISOControl U_ICU_ENTRY_POINT_RENAME(u_isISOControl) +#define u_isJavaIDPart U_ICU_ENTRY_POINT_RENAME(u_isJavaIDPart) +#define u_isJavaIDStart U_ICU_ENTRY_POINT_RENAME(u_isJavaIDStart) +#define u_isJavaSpaceChar U_ICU_ENTRY_POINT_RENAME(u_isJavaSpaceChar) +#define u_isMirrored U_ICU_ENTRY_POINT_RENAME(u_isMirrored) +#define u_isUAlphabetic U_ICU_ENTRY_POINT_RENAME(u_isUAlphabetic) +#define u_isULowercase U_ICU_ENTRY_POINT_RENAME(u_isULowercase) +#define u_isUUppercase U_ICU_ENTRY_POINT_RENAME(u_isUUppercase) +#define u_isUWhiteSpace U_ICU_ENTRY_POINT_RENAME(u_isUWhiteSpace) +#define u_isWhitespace U_ICU_ENTRY_POINT_RENAME(u_isWhitespace) +#define u_isalnum U_ICU_ENTRY_POINT_RENAME(u_isalnum) +#define u_isalnumPOSIX U_ICU_ENTRY_POINT_RENAME(u_isalnumPOSIX) +#define u_isalpha U_ICU_ENTRY_POINT_RENAME(u_isalpha) +#define u_isbase U_ICU_ENTRY_POINT_RENAME(u_isbase) +#define u_isblank U_ICU_ENTRY_POINT_RENAME(u_isblank) +#define u_iscntrl U_ICU_ENTRY_POINT_RENAME(u_iscntrl) +#define u_isdefined U_ICU_ENTRY_POINT_RENAME(u_isdefined) +#define u_isdigit U_ICU_ENTRY_POINT_RENAME(u_isdigit) +#define u_isgraph U_ICU_ENTRY_POINT_RENAME(u_isgraph) +#define u_isgraphPOSIX U_ICU_ENTRY_POINT_RENAME(u_isgraphPOSIX) +#define u_islower U_ICU_ENTRY_POINT_RENAME(u_islower) +#define u_isprint U_ICU_ENTRY_POINT_RENAME(u_isprint) +#define u_isprintPOSIX U_ICU_ENTRY_POINT_RENAME(u_isprintPOSIX) +#define u_ispunct U_ICU_ENTRY_POINT_RENAME(u_ispunct) +#define u_isspace U_ICU_ENTRY_POINT_RENAME(u_isspace) +#define u_istitle U_ICU_ENTRY_POINT_RENAME(u_istitle) +#define u_isupper U_ICU_ENTRY_POINT_RENAME(u_isupper) +#define u_isxdigit U_ICU_ENTRY_POINT_RENAME(u_isxdigit) +#define u_locbund_close U_ICU_ENTRY_POINT_RENAME(u_locbund_close) +#define u_locbund_getNumberFormat U_ICU_ENTRY_POINT_RENAME(u_locbund_getNumberFormat) +#define u_locbund_init U_ICU_ENTRY_POINT_RENAME(u_locbund_init) +#define u_memcasecmp U_ICU_ENTRY_POINT_RENAME(u_memcasecmp) +#define u_memchr U_ICU_ENTRY_POINT_RENAME(u_memchr) +#define u_memchr32 U_ICU_ENTRY_POINT_RENAME(u_memchr32) +#define u_memcmp U_ICU_ENTRY_POINT_RENAME(u_memcmp) +#define u_memcmpCodePointOrder U_ICU_ENTRY_POINT_RENAME(u_memcmpCodePointOrder) +#define u_memcpy U_ICU_ENTRY_POINT_RENAME(u_memcpy) +#define u_memmove U_ICU_ENTRY_POINT_RENAME(u_memmove) +#define u_memrchr U_ICU_ENTRY_POINT_RENAME(u_memrchr) +#define u_memrchr32 U_ICU_ENTRY_POINT_RENAME(u_memrchr32) +#define u_memset U_ICU_ENTRY_POINT_RENAME(u_memset) +#define u_parseMessage U_ICU_ENTRY_POINT_RENAME(u_parseMessage) +#define u_parseMessageWithError U_ICU_ENTRY_POINT_RENAME(u_parseMessageWithError) +#define u_printf U_ICU_ENTRY_POINT_RENAME(u_printf) +#define u_printf_parse U_ICU_ENTRY_POINT_RENAME(u_printf_parse) +#define u_printf_u U_ICU_ENTRY_POINT_RENAME(u_printf_u) +#define u_releaseDefaultConverter U_ICU_ENTRY_POINT_RENAME(u_releaseDefaultConverter) +#define u_scanf_parse U_ICU_ENTRY_POINT_RENAME(u_scanf_parse) +#define u_setAtomicIncDecFunctions U_ICU_ENTRY_POINT_RENAME(u_setAtomicIncDecFunctions) +#define u_setDataDirectory U_ICU_ENTRY_POINT_RENAME(u_setDataDirectory) +#define u_setMemoryFunctions U_ICU_ENTRY_POINT_RENAME(u_setMemoryFunctions) +#define u_setMutexFunctions U_ICU_ENTRY_POINT_RENAME(u_setMutexFunctions) +#define u_setTimeZoneFilesDirectory U_ICU_ENTRY_POINT_RENAME(u_setTimeZoneFilesDirectory) +#define u_shapeArabic U_ICU_ENTRY_POINT_RENAME(u_shapeArabic) +#define u_snprintf U_ICU_ENTRY_POINT_RENAME(u_snprintf) +#define u_snprintf_u U_ICU_ENTRY_POINT_RENAME(u_snprintf_u) +#define u_sprintf U_ICU_ENTRY_POINT_RENAME(u_sprintf) +#define u_sprintf_u U_ICU_ENTRY_POINT_RENAME(u_sprintf_u) +#define u_sscanf U_ICU_ENTRY_POINT_RENAME(u_sscanf) +#define u_sscanf_u U_ICU_ENTRY_POINT_RENAME(u_sscanf_u) +#define u_strCaseCompare U_ICU_ENTRY_POINT_RENAME(u_strCaseCompare) +#define u_strCompare U_ICU_ENTRY_POINT_RENAME(u_strCompare) +#define u_strCompareIter U_ICU_ENTRY_POINT_RENAME(u_strCompareIter) +#define u_strFindFirst U_ICU_ENTRY_POINT_RENAME(u_strFindFirst) +#define u_strFindLast U_ICU_ENTRY_POINT_RENAME(u_strFindLast) +#define u_strFoldCase U_ICU_ENTRY_POINT_RENAME(u_strFoldCase) +#define u_strFromJavaModifiedUTF8WithSub U_ICU_ENTRY_POINT_RENAME(u_strFromJavaModifiedUTF8WithSub) +#define u_strFromPunycode U_ICU_ENTRY_POINT_RENAME(u_strFromPunycode) +#define u_strFromUTF32 U_ICU_ENTRY_POINT_RENAME(u_strFromUTF32) +#define u_strFromUTF32WithSub U_ICU_ENTRY_POINT_RENAME(u_strFromUTF32WithSub) +#define u_strFromUTF8 U_ICU_ENTRY_POINT_RENAME(u_strFromUTF8) +#define u_strFromUTF8Lenient U_ICU_ENTRY_POINT_RENAME(u_strFromUTF8Lenient) +#define u_strFromUTF8WithSub U_ICU_ENTRY_POINT_RENAME(u_strFromUTF8WithSub) +#define u_strFromWCS U_ICU_ENTRY_POINT_RENAME(u_strFromWCS) +#define u_strHasMoreChar32Than U_ICU_ENTRY_POINT_RENAME(u_strHasMoreChar32Than) +#define u_strToJavaModifiedUTF8 U_ICU_ENTRY_POINT_RENAME(u_strToJavaModifiedUTF8) +#define u_strToLower U_ICU_ENTRY_POINT_RENAME(u_strToLower) +#define u_strToPunycode U_ICU_ENTRY_POINT_RENAME(u_strToPunycode) +#define u_strToTitle U_ICU_ENTRY_POINT_RENAME(u_strToTitle) +#define u_strToUTF32 U_ICU_ENTRY_POINT_RENAME(u_strToUTF32) +#define u_strToUTF32WithSub U_ICU_ENTRY_POINT_RENAME(u_strToUTF32WithSub) +#define u_strToUTF8 U_ICU_ENTRY_POINT_RENAME(u_strToUTF8) +#define u_strToUTF8WithSub U_ICU_ENTRY_POINT_RENAME(u_strToUTF8WithSub) +#define u_strToUpper U_ICU_ENTRY_POINT_RENAME(u_strToUpper) +#define u_strToWCS U_ICU_ENTRY_POINT_RENAME(u_strToWCS) +#define u_strcasecmp U_ICU_ENTRY_POINT_RENAME(u_strcasecmp) +#define u_strcat U_ICU_ENTRY_POINT_RENAME(u_strcat) +#define u_strchr U_ICU_ENTRY_POINT_RENAME(u_strchr) +#define u_strchr32 U_ICU_ENTRY_POINT_RENAME(u_strchr32) +#define u_strcmp U_ICU_ENTRY_POINT_RENAME(u_strcmp) +#define u_strcmpCodePointOrder U_ICU_ENTRY_POINT_RENAME(u_strcmpCodePointOrder) +#define u_strcmpFold U_ICU_ENTRY_POINT_RENAME(u_strcmpFold) +#define u_strcpy U_ICU_ENTRY_POINT_RENAME(u_strcpy) +#define u_strcspn U_ICU_ENTRY_POINT_RENAME(u_strcspn) +#define u_stringHasBinaryProperty U_ICU_ENTRY_POINT_RENAME(u_stringHasBinaryProperty) +#define u_strlen U_ICU_ENTRY_POINT_RENAME(u_strlen) +#define u_strncasecmp U_ICU_ENTRY_POINT_RENAME(u_strncasecmp) +#define u_strncat U_ICU_ENTRY_POINT_RENAME(u_strncat) +#define u_strncmp U_ICU_ENTRY_POINT_RENAME(u_strncmp) +#define u_strncmpCodePointOrder U_ICU_ENTRY_POINT_RENAME(u_strncmpCodePointOrder) +#define u_strncpy U_ICU_ENTRY_POINT_RENAME(u_strncpy) +#define u_strpbrk U_ICU_ENTRY_POINT_RENAME(u_strpbrk) +#define u_strrchr U_ICU_ENTRY_POINT_RENAME(u_strrchr) +#define u_strrchr32 U_ICU_ENTRY_POINT_RENAME(u_strrchr32) +#define u_strrstr U_ICU_ENTRY_POINT_RENAME(u_strrstr) +#define u_strspn U_ICU_ENTRY_POINT_RENAME(u_strspn) +#define u_strstr U_ICU_ENTRY_POINT_RENAME(u_strstr) +#define u_strtok_r U_ICU_ENTRY_POINT_RENAME(u_strtok_r) +#define u_terminateChars U_ICU_ENTRY_POINT_RENAME(u_terminateChars) +#define u_terminateUChar32s U_ICU_ENTRY_POINT_RENAME(u_terminateUChar32s) +#define u_terminateUChars U_ICU_ENTRY_POINT_RENAME(u_terminateUChars) +#define u_terminateWChars U_ICU_ENTRY_POINT_RENAME(u_terminateWChars) +#define u_tolower U_ICU_ENTRY_POINT_RENAME(u_tolower) +#define u_totitle U_ICU_ENTRY_POINT_RENAME(u_totitle) +#define u_toupper U_ICU_ENTRY_POINT_RENAME(u_toupper) +#define u_uastrcpy U_ICU_ENTRY_POINT_RENAME(u_uastrcpy) +#define u_uastrncpy U_ICU_ENTRY_POINT_RENAME(u_uastrncpy) +#define u_unescape U_ICU_ENTRY_POINT_RENAME(u_unescape) +#define u_unescapeAt U_ICU_ENTRY_POINT_RENAME(u_unescapeAt) +#define u_versionFromString U_ICU_ENTRY_POINT_RENAME(u_versionFromString) +#define u_versionFromUString U_ICU_ENTRY_POINT_RENAME(u_versionFromUString) +#define u_versionToString U_ICU_ENTRY_POINT_RENAME(u_versionToString) +#define u_vformatMessage U_ICU_ENTRY_POINT_RENAME(u_vformatMessage) +#define u_vformatMessageWithError U_ICU_ENTRY_POINT_RENAME(u_vformatMessageWithError) +#define u_vfprintf U_ICU_ENTRY_POINT_RENAME(u_vfprintf) +#define u_vfprintf_u U_ICU_ENTRY_POINT_RENAME(u_vfprintf_u) +#define u_vfscanf U_ICU_ENTRY_POINT_RENAME(u_vfscanf) +#define u_vfscanf_u U_ICU_ENTRY_POINT_RENAME(u_vfscanf_u) +#define u_vparseMessage U_ICU_ENTRY_POINT_RENAME(u_vparseMessage) +#define u_vparseMessageWithError U_ICU_ENTRY_POINT_RENAME(u_vparseMessageWithError) +#define u_vsnprintf U_ICU_ENTRY_POINT_RENAME(u_vsnprintf) +#define u_vsnprintf_u U_ICU_ENTRY_POINT_RENAME(u_vsnprintf_u) +#define u_vsprintf U_ICU_ENTRY_POINT_RENAME(u_vsprintf) +#define u_vsprintf_u U_ICU_ENTRY_POINT_RENAME(u_vsprintf_u) +#define u_vsscanf U_ICU_ENTRY_POINT_RENAME(u_vsscanf) +#define u_vsscanf_u U_ICU_ENTRY_POINT_RENAME(u_vsscanf_u) +#define u_writeIdenticalLevelRun U_ICU_ENTRY_POINT_RENAME(u_writeIdenticalLevelRun) +#define ubidi_addPropertyStarts U_ICU_ENTRY_POINT_RENAME(ubidi_addPropertyStarts) +#define ubidi_close U_ICU_ENTRY_POINT_RENAME(ubidi_close) +#define ubidi_countParagraphs U_ICU_ENTRY_POINT_RENAME(ubidi_countParagraphs) +#define ubidi_countRuns U_ICU_ENTRY_POINT_RENAME(ubidi_countRuns) +#define ubidi_getBaseDirection U_ICU_ENTRY_POINT_RENAME(ubidi_getBaseDirection) +#define ubidi_getClass U_ICU_ENTRY_POINT_RENAME(ubidi_getClass) +#define ubidi_getClassCallback U_ICU_ENTRY_POINT_RENAME(ubidi_getClassCallback) +#define ubidi_getCustomizedClass U_ICU_ENTRY_POINT_RENAME(ubidi_getCustomizedClass) +#define ubidi_getDirection U_ICU_ENTRY_POINT_RENAME(ubidi_getDirection) +#define ubidi_getJoiningGroup U_ICU_ENTRY_POINT_RENAME(ubidi_getJoiningGroup) +#define ubidi_getJoiningType U_ICU_ENTRY_POINT_RENAME(ubidi_getJoiningType) +#define ubidi_getLength U_ICU_ENTRY_POINT_RENAME(ubidi_getLength) +#define ubidi_getLevelAt U_ICU_ENTRY_POINT_RENAME(ubidi_getLevelAt) +#define ubidi_getLevels U_ICU_ENTRY_POINT_RENAME(ubidi_getLevels) +#define ubidi_getLogicalIndex U_ICU_ENTRY_POINT_RENAME(ubidi_getLogicalIndex) +#define ubidi_getLogicalMap U_ICU_ENTRY_POINT_RENAME(ubidi_getLogicalMap) +#define ubidi_getLogicalRun U_ICU_ENTRY_POINT_RENAME(ubidi_getLogicalRun) +#define ubidi_getMaxValue U_ICU_ENTRY_POINT_RENAME(ubidi_getMaxValue) +#define ubidi_getMemory U_ICU_ENTRY_POINT_RENAME(ubidi_getMemory) +#define ubidi_getMirror U_ICU_ENTRY_POINT_RENAME(ubidi_getMirror) +#define ubidi_getPairedBracket U_ICU_ENTRY_POINT_RENAME(ubidi_getPairedBracket) +#define ubidi_getPairedBracketType U_ICU_ENTRY_POINT_RENAME(ubidi_getPairedBracketType) +#define ubidi_getParaLevel U_ICU_ENTRY_POINT_RENAME(ubidi_getParaLevel) +#define ubidi_getParaLevelAtIndex U_ICU_ENTRY_POINT_RENAME(ubidi_getParaLevelAtIndex) +#define ubidi_getParagraph U_ICU_ENTRY_POINT_RENAME(ubidi_getParagraph) +#define ubidi_getParagraphByIndex U_ICU_ENTRY_POINT_RENAME(ubidi_getParagraphByIndex) +#define ubidi_getProcessedLength U_ICU_ENTRY_POINT_RENAME(ubidi_getProcessedLength) +#define ubidi_getReorderingMode U_ICU_ENTRY_POINT_RENAME(ubidi_getReorderingMode) +#define ubidi_getReorderingOptions U_ICU_ENTRY_POINT_RENAME(ubidi_getReorderingOptions) +#define ubidi_getResultLength U_ICU_ENTRY_POINT_RENAME(ubidi_getResultLength) +#define ubidi_getRuns U_ICU_ENTRY_POINT_RENAME(ubidi_getRuns) +#define ubidi_getText U_ICU_ENTRY_POINT_RENAME(ubidi_getText) +#define ubidi_getVisualIndex U_ICU_ENTRY_POINT_RENAME(ubidi_getVisualIndex) +#define ubidi_getVisualMap U_ICU_ENTRY_POINT_RENAME(ubidi_getVisualMap) +#define ubidi_getVisualRun U_ICU_ENTRY_POINT_RENAME(ubidi_getVisualRun) +#define ubidi_invertMap U_ICU_ENTRY_POINT_RENAME(ubidi_invertMap) +#define ubidi_isBidiControl U_ICU_ENTRY_POINT_RENAME(ubidi_isBidiControl) +#define ubidi_isInverse U_ICU_ENTRY_POINT_RENAME(ubidi_isInverse) +#define ubidi_isJoinControl U_ICU_ENTRY_POINT_RENAME(ubidi_isJoinControl) +#define ubidi_isMirrored U_ICU_ENTRY_POINT_RENAME(ubidi_isMirrored) +#define ubidi_isOrderParagraphsLTR U_ICU_ENTRY_POINT_RENAME(ubidi_isOrderParagraphsLTR) +#define ubidi_open U_ICU_ENTRY_POINT_RENAME(ubidi_open) +#define ubidi_openSized U_ICU_ENTRY_POINT_RENAME(ubidi_openSized) +#define ubidi_orderParagraphsLTR U_ICU_ENTRY_POINT_RENAME(ubidi_orderParagraphsLTR) +#define ubidi_reorderLogical U_ICU_ENTRY_POINT_RENAME(ubidi_reorderLogical) +#define ubidi_reorderVisual U_ICU_ENTRY_POINT_RENAME(ubidi_reorderVisual) +#define ubidi_setClassCallback U_ICU_ENTRY_POINT_RENAME(ubidi_setClassCallback) +#define ubidi_setContext U_ICU_ENTRY_POINT_RENAME(ubidi_setContext) +#define ubidi_setInverse U_ICU_ENTRY_POINT_RENAME(ubidi_setInverse) +#define ubidi_setLine U_ICU_ENTRY_POINT_RENAME(ubidi_setLine) +#define ubidi_setPara U_ICU_ENTRY_POINT_RENAME(ubidi_setPara) +#define ubidi_setReorderingMode U_ICU_ENTRY_POINT_RENAME(ubidi_setReorderingMode) +#define ubidi_setReorderingOptions U_ICU_ENTRY_POINT_RENAME(ubidi_setReorderingOptions) +#define ubidi_writeReordered U_ICU_ENTRY_POINT_RENAME(ubidi_writeReordered) +#define ubidi_writeReverse U_ICU_ENTRY_POINT_RENAME(ubidi_writeReverse) +#define ubiditransform_close U_ICU_ENTRY_POINT_RENAME(ubiditransform_close) +#define ubiditransform_open U_ICU_ENTRY_POINT_RENAME(ubiditransform_open) +#define ubiditransform_transform U_ICU_ENTRY_POINT_RENAME(ubiditransform_transform) +#define ublock_getCode U_ICU_ENTRY_POINT_RENAME(ublock_getCode) +#define ubrk_clone U_ICU_ENTRY_POINT_RENAME(ubrk_clone) +#define ubrk_close U_ICU_ENTRY_POINT_RENAME(ubrk_close) +#define ubrk_countAvailable U_ICU_ENTRY_POINT_RENAME(ubrk_countAvailable) +#define ubrk_current U_ICU_ENTRY_POINT_RENAME(ubrk_current) +#define ubrk_first U_ICU_ENTRY_POINT_RENAME(ubrk_first) +#define ubrk_following U_ICU_ENTRY_POINT_RENAME(ubrk_following) +#define ubrk_getAvailable U_ICU_ENTRY_POINT_RENAME(ubrk_getAvailable) +#define ubrk_getBinaryRules U_ICU_ENTRY_POINT_RENAME(ubrk_getBinaryRules) +#define ubrk_getLocaleByType U_ICU_ENTRY_POINT_RENAME(ubrk_getLocaleByType) +#define ubrk_getRuleStatus U_ICU_ENTRY_POINT_RENAME(ubrk_getRuleStatus) +#define ubrk_getRuleStatusVec U_ICU_ENTRY_POINT_RENAME(ubrk_getRuleStatusVec) +#define ubrk_isBoundary U_ICU_ENTRY_POINT_RENAME(ubrk_isBoundary) +#define ubrk_last U_ICU_ENTRY_POINT_RENAME(ubrk_last) +#define ubrk_next U_ICU_ENTRY_POINT_RENAME(ubrk_next) +#define ubrk_open U_ICU_ENTRY_POINT_RENAME(ubrk_open) +#define ubrk_openBinaryRules U_ICU_ENTRY_POINT_RENAME(ubrk_openBinaryRules) +#define ubrk_openRules U_ICU_ENTRY_POINT_RENAME(ubrk_openRules) +#define ubrk_preceding U_ICU_ENTRY_POINT_RENAME(ubrk_preceding) +#define ubrk_previous U_ICU_ENTRY_POINT_RENAME(ubrk_previous) +#define ubrk_refreshUText U_ICU_ENTRY_POINT_RENAME(ubrk_refreshUText) +#define ubrk_safeClone U_ICU_ENTRY_POINT_RENAME(ubrk_safeClone) +#define ubrk_setText U_ICU_ENTRY_POINT_RENAME(ubrk_setText) +#define ubrk_setUText U_ICU_ENTRY_POINT_RENAME(ubrk_setUText) +#define ubrk_swap U_ICU_ENTRY_POINT_RENAME(ubrk_swap) +#define ucache_compareKeys U_ICU_ENTRY_POINT_RENAME(ucache_compareKeys) +#define ucache_deleteKey U_ICU_ENTRY_POINT_RENAME(ucache_deleteKey) +#define ucache_hashKeys U_ICU_ENTRY_POINT_RENAME(ucache_hashKeys) +#define ucal_add U_ICU_ENTRY_POINT_RENAME(ucal_add) +#define ucal_clear U_ICU_ENTRY_POINT_RENAME(ucal_clear) +#define ucal_clearField U_ICU_ENTRY_POINT_RENAME(ucal_clearField) +#define ucal_clone U_ICU_ENTRY_POINT_RENAME(ucal_clone) +#define ucal_close U_ICU_ENTRY_POINT_RENAME(ucal_close) +#define ucal_countAvailable U_ICU_ENTRY_POINT_RENAME(ucal_countAvailable) +#define ucal_equivalentTo U_ICU_ENTRY_POINT_RENAME(ucal_equivalentTo) +#define ucal_get U_ICU_ENTRY_POINT_RENAME(ucal_get) +#define ucal_getAttribute U_ICU_ENTRY_POINT_RENAME(ucal_getAttribute) +#define ucal_getAvailable U_ICU_ENTRY_POINT_RENAME(ucal_getAvailable) +#define ucal_getCanonicalTimeZoneID U_ICU_ENTRY_POINT_RENAME(ucal_getCanonicalTimeZoneID) +#define ucal_getDSTSavings U_ICU_ENTRY_POINT_RENAME(ucal_getDSTSavings) +#define ucal_getDayOfWeekType U_ICU_ENTRY_POINT_RENAME(ucal_getDayOfWeekType) +#define ucal_getDefaultTimeZone U_ICU_ENTRY_POINT_RENAME(ucal_getDefaultTimeZone) +#define ucal_getFieldDifference U_ICU_ENTRY_POINT_RENAME(ucal_getFieldDifference) +#define ucal_getGregorianChange U_ICU_ENTRY_POINT_RENAME(ucal_getGregorianChange) +#define ucal_getHostTimeZone U_ICU_ENTRY_POINT_RENAME(ucal_getHostTimeZone) +#define ucal_getKeywordValuesForLocale U_ICU_ENTRY_POINT_RENAME(ucal_getKeywordValuesForLocale) +#define ucal_getLimit U_ICU_ENTRY_POINT_RENAME(ucal_getLimit) +#define ucal_getLocaleByType U_ICU_ENTRY_POINT_RENAME(ucal_getLocaleByType) +#define ucal_getMillis U_ICU_ENTRY_POINT_RENAME(ucal_getMillis) +#define ucal_getNow U_ICU_ENTRY_POINT_RENAME(ucal_getNow) +#define ucal_getTZDataVersion U_ICU_ENTRY_POINT_RENAME(ucal_getTZDataVersion) +#define ucal_getTimeZoneDisplayName U_ICU_ENTRY_POINT_RENAME(ucal_getTimeZoneDisplayName) +#define ucal_getTimeZoneID U_ICU_ENTRY_POINT_RENAME(ucal_getTimeZoneID) +#define ucal_getTimeZoneIDForWindowsID U_ICU_ENTRY_POINT_RENAME(ucal_getTimeZoneIDForWindowsID) +#define ucal_getTimeZoneOffsetFromLocal U_ICU_ENTRY_POINT_RENAME(ucal_getTimeZoneOffsetFromLocal) +#define ucal_getTimeZoneTransitionDate U_ICU_ENTRY_POINT_RENAME(ucal_getTimeZoneTransitionDate) +#define ucal_getType U_ICU_ENTRY_POINT_RENAME(ucal_getType) +#define ucal_getWeekendTransition U_ICU_ENTRY_POINT_RENAME(ucal_getWeekendTransition) +#define ucal_getWindowsTimeZoneID U_ICU_ENTRY_POINT_RENAME(ucal_getWindowsTimeZoneID) +#define ucal_inDaylightTime U_ICU_ENTRY_POINT_RENAME(ucal_inDaylightTime) +#define ucal_isSet U_ICU_ENTRY_POINT_RENAME(ucal_isSet) +#define ucal_isWeekend U_ICU_ENTRY_POINT_RENAME(ucal_isWeekend) +#define ucal_open U_ICU_ENTRY_POINT_RENAME(ucal_open) +#define ucal_openCountryTimeZones U_ICU_ENTRY_POINT_RENAME(ucal_openCountryTimeZones) +#define ucal_openTimeZoneIDEnumeration U_ICU_ENTRY_POINT_RENAME(ucal_openTimeZoneIDEnumeration) +#define ucal_openTimeZones U_ICU_ENTRY_POINT_RENAME(ucal_openTimeZones) +#define ucal_roll U_ICU_ENTRY_POINT_RENAME(ucal_roll) +#define ucal_set U_ICU_ENTRY_POINT_RENAME(ucal_set) +#define ucal_setAttribute U_ICU_ENTRY_POINT_RENAME(ucal_setAttribute) +#define ucal_setDate U_ICU_ENTRY_POINT_RENAME(ucal_setDate) +#define ucal_setDateTime U_ICU_ENTRY_POINT_RENAME(ucal_setDateTime) +#define ucal_setDefaultTimeZone U_ICU_ENTRY_POINT_RENAME(ucal_setDefaultTimeZone) +#define ucal_setGregorianChange U_ICU_ENTRY_POINT_RENAME(ucal_setGregorianChange) +#define ucal_setMillis U_ICU_ENTRY_POINT_RENAME(ucal_setMillis) +#define ucal_setTimeZone U_ICU_ENTRY_POINT_RENAME(ucal_setTimeZone) +#define ucase_addCaseClosure U_ICU_ENTRY_POINT_RENAME(ucase_addCaseClosure) +#define ucase_addPropertyStarts U_ICU_ENTRY_POINT_RENAME(ucase_addPropertyStarts) +#define ucase_addSimpleCaseClosure U_ICU_ENTRY_POINT_RENAME(ucase_addSimpleCaseClosure) +#define ucase_addStringCaseClosure U_ICU_ENTRY_POINT_RENAME(ucase_addStringCaseClosure) +#define ucase_fold U_ICU_ENTRY_POINT_RENAME(ucase_fold) +#define ucase_getCaseLocale U_ICU_ENTRY_POINT_RENAME(ucase_getCaseLocale) +#define ucase_getSingleton U_ICU_ENTRY_POINT_RENAME(ucase_getSingleton) +#define ucase_getTrie U_ICU_ENTRY_POINT_RENAME(ucase_getTrie) +#define ucase_getType U_ICU_ENTRY_POINT_RENAME(ucase_getType) +#define ucase_getTypeOrIgnorable U_ICU_ENTRY_POINT_RENAME(ucase_getTypeOrIgnorable) +#define ucase_hasBinaryProperty U_ICU_ENTRY_POINT_RENAME(ucase_hasBinaryProperty) +#define ucase_isCaseSensitive U_ICU_ENTRY_POINT_RENAME(ucase_isCaseSensitive) +#define ucase_isSoftDotted U_ICU_ENTRY_POINT_RENAME(ucase_isSoftDotted) +#define ucase_toFullFolding U_ICU_ENTRY_POINT_RENAME(ucase_toFullFolding) +#define ucase_toFullLower U_ICU_ENTRY_POINT_RENAME(ucase_toFullLower) +#define ucase_toFullTitle U_ICU_ENTRY_POINT_RENAME(ucase_toFullTitle) +#define ucase_toFullUpper U_ICU_ENTRY_POINT_RENAME(ucase_toFullUpper) +#define ucase_tolower U_ICU_ENTRY_POINT_RENAME(ucase_tolower) +#define ucase_totitle U_ICU_ENTRY_POINT_RENAME(ucase_totitle) +#define ucase_toupper U_ICU_ENTRY_POINT_RENAME(ucase_toupper) +#define ucasemap_close U_ICU_ENTRY_POINT_RENAME(ucasemap_close) +#define ucasemap_getBreakIterator U_ICU_ENTRY_POINT_RENAME(ucasemap_getBreakIterator) +#define ucasemap_getLocale U_ICU_ENTRY_POINT_RENAME(ucasemap_getLocale) +#define ucasemap_getOptions U_ICU_ENTRY_POINT_RENAME(ucasemap_getOptions) +#define ucasemap_internalUTF8ToTitle U_ICU_ENTRY_POINT_RENAME(ucasemap_internalUTF8ToTitle) +#define ucasemap_open U_ICU_ENTRY_POINT_RENAME(ucasemap_open) +#define ucasemap_setBreakIterator U_ICU_ENTRY_POINT_RENAME(ucasemap_setBreakIterator) +#define ucasemap_setLocale U_ICU_ENTRY_POINT_RENAME(ucasemap_setLocale) +#define ucasemap_setOptions U_ICU_ENTRY_POINT_RENAME(ucasemap_setOptions) +#define ucasemap_toTitle U_ICU_ENTRY_POINT_RENAME(ucasemap_toTitle) +#define ucasemap_utf8FoldCase U_ICU_ENTRY_POINT_RENAME(ucasemap_utf8FoldCase) +#define ucasemap_utf8ToLower U_ICU_ENTRY_POINT_RENAME(ucasemap_utf8ToLower) +#define ucasemap_utf8ToTitle U_ICU_ENTRY_POINT_RENAME(ucasemap_utf8ToTitle) +#define ucasemap_utf8ToUpper U_ICU_ENTRY_POINT_RENAME(ucasemap_utf8ToUpper) +#define ucfpos_close U_ICU_ENTRY_POINT_RENAME(ucfpos_close) +#define ucfpos_constrainCategory U_ICU_ENTRY_POINT_RENAME(ucfpos_constrainCategory) +#define ucfpos_constrainField U_ICU_ENTRY_POINT_RENAME(ucfpos_constrainField) +#define ucfpos_getCategory U_ICU_ENTRY_POINT_RENAME(ucfpos_getCategory) +#define ucfpos_getField U_ICU_ENTRY_POINT_RENAME(ucfpos_getField) +#define ucfpos_getIndexes U_ICU_ENTRY_POINT_RENAME(ucfpos_getIndexes) +#define ucfpos_getInt64IterationContext U_ICU_ENTRY_POINT_RENAME(ucfpos_getInt64IterationContext) +#define ucfpos_matchesField U_ICU_ENTRY_POINT_RENAME(ucfpos_matchesField) +#define ucfpos_open U_ICU_ENTRY_POINT_RENAME(ucfpos_open) +#define ucfpos_reset U_ICU_ENTRY_POINT_RENAME(ucfpos_reset) +#define ucfpos_setInt64IterationContext U_ICU_ENTRY_POINT_RENAME(ucfpos_setInt64IterationContext) +#define ucfpos_setState U_ICU_ENTRY_POINT_RENAME(ucfpos_setState) +#define uchar_addPropertyStarts U_ICU_ENTRY_POINT_RENAME(uchar_addPropertyStarts) +#define uchar_swapNames U_ICU_ENTRY_POINT_RENAME(uchar_swapNames) +#define ucln_cleanupOne U_ICU_ENTRY_POINT_RENAME(ucln_cleanupOne) +#define ucln_common_registerCleanup U_ICU_ENTRY_POINT_RENAME(ucln_common_registerCleanup) +#define ucln_i18n_registerCleanup U_ICU_ENTRY_POINT_RENAME(ucln_i18n_registerCleanup) +#define ucln_io_registerCleanup U_ICU_ENTRY_POINT_RENAME(ucln_io_registerCleanup) +#define ucln_lib_cleanup U_ICU_ENTRY_POINT_RENAME(ucln_lib_cleanup) +#define ucln_registerCleanup U_ICU_ENTRY_POINT_RENAME(ucln_registerCleanup) +#define ucnv_MBCSFromUChar32 U_ICU_ENTRY_POINT_RENAME(ucnv_MBCSFromUChar32) +#define ucnv_MBCSFromUnicodeWithOffsets U_ICU_ENTRY_POINT_RENAME(ucnv_MBCSFromUnicodeWithOffsets) +#define ucnv_MBCSGetFilteredUnicodeSetForUnicode U_ICU_ENTRY_POINT_RENAME(ucnv_MBCSGetFilteredUnicodeSetForUnicode) +#define ucnv_MBCSGetType U_ICU_ENTRY_POINT_RENAME(ucnv_MBCSGetType) +#define ucnv_MBCSGetUnicodeSetForUnicode U_ICU_ENTRY_POINT_RENAME(ucnv_MBCSGetUnicodeSetForUnicode) +#define ucnv_MBCSIsLeadByte U_ICU_ENTRY_POINT_RENAME(ucnv_MBCSIsLeadByte) +#define ucnv_MBCSSimpleGetNextUChar U_ICU_ENTRY_POINT_RENAME(ucnv_MBCSSimpleGetNextUChar) +#define ucnv_MBCSToUnicodeWithOffsets U_ICU_ENTRY_POINT_RENAME(ucnv_MBCSToUnicodeWithOffsets) +#define ucnv_bld_countAvailableConverters U_ICU_ENTRY_POINT_RENAME(ucnv_bld_countAvailableConverters) +#define ucnv_bld_getAvailableConverter U_ICU_ENTRY_POINT_RENAME(ucnv_bld_getAvailableConverter) +#define ucnv_canCreateConverter U_ICU_ENTRY_POINT_RENAME(ucnv_canCreateConverter) +#define ucnv_cbFromUWriteBytes U_ICU_ENTRY_POINT_RENAME(ucnv_cbFromUWriteBytes) +#define ucnv_cbFromUWriteSub U_ICU_ENTRY_POINT_RENAME(ucnv_cbFromUWriteSub) +#define ucnv_cbFromUWriteUChars U_ICU_ENTRY_POINT_RENAME(ucnv_cbFromUWriteUChars) +#define ucnv_cbToUWriteSub U_ICU_ENTRY_POINT_RENAME(ucnv_cbToUWriteSub) +#define ucnv_cbToUWriteUChars U_ICU_ENTRY_POINT_RENAME(ucnv_cbToUWriteUChars) +#define ucnv_clone U_ICU_ENTRY_POINT_RENAME(ucnv_clone) +#define ucnv_close U_ICU_ENTRY_POINT_RENAME(ucnv_close) +#define ucnv_compareNames U_ICU_ENTRY_POINT_RENAME(ucnv_compareNames) +#define ucnv_convert U_ICU_ENTRY_POINT_RENAME(ucnv_convert) +#define ucnv_convertEx U_ICU_ENTRY_POINT_RENAME(ucnv_convertEx) +#define ucnv_countAliases U_ICU_ENTRY_POINT_RENAME(ucnv_countAliases) +#define ucnv_countAvailable U_ICU_ENTRY_POINT_RENAME(ucnv_countAvailable) +#define ucnv_countStandards U_ICU_ENTRY_POINT_RENAME(ucnv_countStandards) +#define ucnv_createAlgorithmicConverter U_ICU_ENTRY_POINT_RENAME(ucnv_createAlgorithmicConverter) +#define ucnv_createConverter U_ICU_ENTRY_POINT_RENAME(ucnv_createConverter) +#define ucnv_createConverterFromPackage U_ICU_ENTRY_POINT_RENAME(ucnv_createConverterFromPackage) +#define ucnv_createConverterFromSharedData U_ICU_ENTRY_POINT_RENAME(ucnv_createConverterFromSharedData) +#define ucnv_detectUnicodeSignature U_ICU_ENTRY_POINT_RENAME(ucnv_detectUnicodeSignature) +#define ucnv_enableCleanup U_ICU_ENTRY_POINT_RENAME(ucnv_enableCleanup) +#define ucnv_extContinueMatchFromU U_ICU_ENTRY_POINT_RENAME(ucnv_extContinueMatchFromU) +#define ucnv_extContinueMatchToU U_ICU_ENTRY_POINT_RENAME(ucnv_extContinueMatchToU) +#define ucnv_extGetUnicodeSet U_ICU_ENTRY_POINT_RENAME(ucnv_extGetUnicodeSet) +#define ucnv_extInitialMatchFromU U_ICU_ENTRY_POINT_RENAME(ucnv_extInitialMatchFromU) +#define ucnv_extInitialMatchToU U_ICU_ENTRY_POINT_RENAME(ucnv_extInitialMatchToU) +#define ucnv_extSimpleMatchFromU U_ICU_ENTRY_POINT_RENAME(ucnv_extSimpleMatchFromU) +#define ucnv_extSimpleMatchToU U_ICU_ENTRY_POINT_RENAME(ucnv_extSimpleMatchToU) +#define ucnv_fixFileSeparator U_ICU_ENTRY_POINT_RENAME(ucnv_fixFileSeparator) +#define ucnv_flushCache U_ICU_ENTRY_POINT_RENAME(ucnv_flushCache) +#define ucnv_fromAlgorithmic U_ICU_ENTRY_POINT_RENAME(ucnv_fromAlgorithmic) +#define ucnv_fromUChars U_ICU_ENTRY_POINT_RENAME(ucnv_fromUChars) +#define ucnv_fromUCountPending U_ICU_ENTRY_POINT_RENAME(ucnv_fromUCountPending) +#define ucnv_fromUWriteBytes U_ICU_ENTRY_POINT_RENAME(ucnv_fromUWriteBytes) +#define ucnv_fromUnicode U_ICU_ENTRY_POINT_RENAME(ucnv_fromUnicode) +#define ucnv_fromUnicode_UTF8 U_ICU_ENTRY_POINT_RENAME(ucnv_fromUnicode_UTF8) +#define ucnv_fromUnicode_UTF8_OFFSETS_LOGIC U_ICU_ENTRY_POINT_RENAME(ucnv_fromUnicode_UTF8_OFFSETS_LOGIC) +#define ucnv_getAlias U_ICU_ENTRY_POINT_RENAME(ucnv_getAlias) +#define ucnv_getAliases U_ICU_ENTRY_POINT_RENAME(ucnv_getAliases) +#define ucnv_getAvailableName U_ICU_ENTRY_POINT_RENAME(ucnv_getAvailableName) +#define ucnv_getCCSID U_ICU_ENTRY_POINT_RENAME(ucnv_getCCSID) +#define ucnv_getCanonicalName U_ICU_ENTRY_POINT_RENAME(ucnv_getCanonicalName) +#define ucnv_getCompleteUnicodeSet U_ICU_ENTRY_POINT_RENAME(ucnv_getCompleteUnicodeSet) +#define ucnv_getDefaultName U_ICU_ENTRY_POINT_RENAME(ucnv_getDefaultName) +#define ucnv_getDisplayName U_ICU_ENTRY_POINT_RENAME(ucnv_getDisplayName) +#define ucnv_getFromUCallBack U_ICU_ENTRY_POINT_RENAME(ucnv_getFromUCallBack) +#define ucnv_getInvalidChars U_ICU_ENTRY_POINT_RENAME(ucnv_getInvalidChars) +#define ucnv_getInvalidUChars U_ICU_ENTRY_POINT_RENAME(ucnv_getInvalidUChars) +#define ucnv_getMaxCharSize U_ICU_ENTRY_POINT_RENAME(ucnv_getMaxCharSize) +#define ucnv_getMinCharSize U_ICU_ENTRY_POINT_RENAME(ucnv_getMinCharSize) +#define ucnv_getName U_ICU_ENTRY_POINT_RENAME(ucnv_getName) +#define ucnv_getNextUChar U_ICU_ENTRY_POINT_RENAME(ucnv_getNextUChar) +#define ucnv_getNonSurrogateUnicodeSet U_ICU_ENTRY_POINT_RENAME(ucnv_getNonSurrogateUnicodeSet) +#define ucnv_getPlatform U_ICU_ENTRY_POINT_RENAME(ucnv_getPlatform) +#define ucnv_getStandard U_ICU_ENTRY_POINT_RENAME(ucnv_getStandard) +#define ucnv_getStandardName U_ICU_ENTRY_POINT_RENAME(ucnv_getStandardName) +#define ucnv_getStarters U_ICU_ENTRY_POINT_RENAME(ucnv_getStarters) +#define ucnv_getSubstChars U_ICU_ENTRY_POINT_RENAME(ucnv_getSubstChars) +#define ucnv_getToUCallBack U_ICU_ENTRY_POINT_RENAME(ucnv_getToUCallBack) +#define ucnv_getType U_ICU_ENTRY_POINT_RENAME(ucnv_getType) +#define ucnv_getUnicodeSet U_ICU_ENTRY_POINT_RENAME(ucnv_getUnicodeSet) +#define ucnv_incrementRefCount U_ICU_ENTRY_POINT_RENAME(ucnv_incrementRefCount) +#define ucnv_io_countKnownConverters U_ICU_ENTRY_POINT_RENAME(ucnv_io_countKnownConverters) +#define ucnv_io_getConverterName U_ICU_ENTRY_POINT_RENAME(ucnv_io_getConverterName) +#define ucnv_io_stripASCIIForCompare U_ICU_ENTRY_POINT_RENAME(ucnv_io_stripASCIIForCompare) +#define ucnv_io_stripEBCDICForCompare U_ICU_ENTRY_POINT_RENAME(ucnv_io_stripEBCDICForCompare) +#define ucnv_isAmbiguous U_ICU_ENTRY_POINT_RENAME(ucnv_isAmbiguous) +#define ucnv_isFixedWidth U_ICU_ENTRY_POINT_RENAME(ucnv_isFixedWidth) +#define ucnv_load U_ICU_ENTRY_POINT_RENAME(ucnv_load) +#define ucnv_loadSharedData U_ICU_ENTRY_POINT_RENAME(ucnv_loadSharedData) +#define ucnv_open U_ICU_ENTRY_POINT_RENAME(ucnv_open) +#define ucnv_openAllNames U_ICU_ENTRY_POINT_RENAME(ucnv_openAllNames) +#define ucnv_openCCSID U_ICU_ENTRY_POINT_RENAME(ucnv_openCCSID) +#define ucnv_openPackage U_ICU_ENTRY_POINT_RENAME(ucnv_openPackage) +#define ucnv_openStandardNames U_ICU_ENTRY_POINT_RENAME(ucnv_openStandardNames) +#define ucnv_openU U_ICU_ENTRY_POINT_RENAME(ucnv_openU) +#define ucnv_reset U_ICU_ENTRY_POINT_RENAME(ucnv_reset) +#define ucnv_resetFromUnicode U_ICU_ENTRY_POINT_RENAME(ucnv_resetFromUnicode) +#define ucnv_resetToUnicode U_ICU_ENTRY_POINT_RENAME(ucnv_resetToUnicode) +#define ucnv_safeClone U_ICU_ENTRY_POINT_RENAME(ucnv_safeClone) +#define ucnv_setDefaultName U_ICU_ENTRY_POINT_RENAME(ucnv_setDefaultName) +#define ucnv_setFallback U_ICU_ENTRY_POINT_RENAME(ucnv_setFallback) +#define ucnv_setFromUCallBack U_ICU_ENTRY_POINT_RENAME(ucnv_setFromUCallBack) +#define ucnv_setSubstChars U_ICU_ENTRY_POINT_RENAME(ucnv_setSubstChars) +#define ucnv_setSubstString U_ICU_ENTRY_POINT_RENAME(ucnv_setSubstString) +#define ucnv_setToUCallBack U_ICU_ENTRY_POINT_RENAME(ucnv_setToUCallBack) +#define ucnv_swap U_ICU_ENTRY_POINT_RENAME(ucnv_swap) +#define ucnv_swapAliases U_ICU_ENTRY_POINT_RENAME(ucnv_swapAliases) +#define ucnv_toAlgorithmic U_ICU_ENTRY_POINT_RENAME(ucnv_toAlgorithmic) +#define ucnv_toUChars U_ICU_ENTRY_POINT_RENAME(ucnv_toUChars) +#define ucnv_toUCountPending U_ICU_ENTRY_POINT_RENAME(ucnv_toUCountPending) +#define ucnv_toUWriteCodePoint U_ICU_ENTRY_POINT_RENAME(ucnv_toUWriteCodePoint) +#define ucnv_toUWriteUChars U_ICU_ENTRY_POINT_RENAME(ucnv_toUWriteUChars) +#define ucnv_toUnicode U_ICU_ENTRY_POINT_RENAME(ucnv_toUnicode) +#define ucnv_unload U_ICU_ENTRY_POINT_RENAME(ucnv_unload) +#define ucnv_unloadSharedDataIfReady U_ICU_ENTRY_POINT_RENAME(ucnv_unloadSharedDataIfReady) +#define ucnv_usesFallback U_ICU_ENTRY_POINT_RENAME(ucnv_usesFallback) +#define ucnvsel_close U_ICU_ENTRY_POINT_RENAME(ucnvsel_close) +#define ucnvsel_open U_ICU_ENTRY_POINT_RENAME(ucnvsel_open) +#define ucnvsel_openFromSerialized U_ICU_ENTRY_POINT_RENAME(ucnvsel_openFromSerialized) +#define ucnvsel_selectForString U_ICU_ENTRY_POINT_RENAME(ucnvsel_selectForString) +#define ucnvsel_selectForUTF8 U_ICU_ENTRY_POINT_RENAME(ucnvsel_selectForUTF8) +#define ucnvsel_serialize U_ICU_ENTRY_POINT_RENAME(ucnvsel_serialize) +#define ucol_clone U_ICU_ENTRY_POINT_RENAME(ucol_clone) +#define ucol_cloneBinary U_ICU_ENTRY_POINT_RENAME(ucol_cloneBinary) +#define ucol_close U_ICU_ENTRY_POINT_RENAME(ucol_close) +#define ucol_closeElements U_ICU_ENTRY_POINT_RENAME(ucol_closeElements) +#define ucol_countAvailable U_ICU_ENTRY_POINT_RENAME(ucol_countAvailable) +#define ucol_equal U_ICU_ENTRY_POINT_RENAME(ucol_equal) +#define ucol_equals U_ICU_ENTRY_POINT_RENAME(ucol_equals) +#define ucol_getAttribute U_ICU_ENTRY_POINT_RENAME(ucol_getAttribute) +#define ucol_getAvailable U_ICU_ENTRY_POINT_RENAME(ucol_getAvailable) +#define ucol_getBound U_ICU_ENTRY_POINT_RENAME(ucol_getBound) +#define ucol_getContractions U_ICU_ENTRY_POINT_RENAME(ucol_getContractions) +#define ucol_getContractionsAndExpansions U_ICU_ENTRY_POINT_RENAME(ucol_getContractionsAndExpansions) +#define ucol_getDisplayName U_ICU_ENTRY_POINT_RENAME(ucol_getDisplayName) +#define ucol_getEquivalentReorderCodes U_ICU_ENTRY_POINT_RENAME(ucol_getEquivalentReorderCodes) +#define ucol_getFunctionalEquivalent U_ICU_ENTRY_POINT_RENAME(ucol_getFunctionalEquivalent) +#define ucol_getKeywordValues U_ICU_ENTRY_POINT_RENAME(ucol_getKeywordValues) +#define ucol_getKeywordValuesForLocale U_ICU_ENTRY_POINT_RENAME(ucol_getKeywordValuesForLocale) +#define ucol_getKeywords U_ICU_ENTRY_POINT_RENAME(ucol_getKeywords) +#define ucol_getLocale U_ICU_ENTRY_POINT_RENAME(ucol_getLocale) +#define ucol_getLocaleByType U_ICU_ENTRY_POINT_RENAME(ucol_getLocaleByType) +#define ucol_getMaxExpansion U_ICU_ENTRY_POINT_RENAME(ucol_getMaxExpansion) +#define ucol_getMaxVariable U_ICU_ENTRY_POINT_RENAME(ucol_getMaxVariable) +#define ucol_getOffset U_ICU_ENTRY_POINT_RENAME(ucol_getOffset) +#define ucol_getReorderCodes U_ICU_ENTRY_POINT_RENAME(ucol_getReorderCodes) +#define ucol_getRules U_ICU_ENTRY_POINT_RENAME(ucol_getRules) +#define ucol_getRulesEx U_ICU_ENTRY_POINT_RENAME(ucol_getRulesEx) +#define ucol_getShortDefinitionString U_ICU_ENTRY_POINT_RENAME(ucol_getShortDefinitionString) +#define ucol_getSortKey U_ICU_ENTRY_POINT_RENAME(ucol_getSortKey) +#define ucol_getStrength U_ICU_ENTRY_POINT_RENAME(ucol_getStrength) +#define ucol_getTailoredSet U_ICU_ENTRY_POINT_RENAME(ucol_getTailoredSet) +#define ucol_getUCAVersion U_ICU_ENTRY_POINT_RENAME(ucol_getUCAVersion) +#define ucol_getUnsafeSet U_ICU_ENTRY_POINT_RENAME(ucol_getUnsafeSet) +#define ucol_getVariableTop U_ICU_ENTRY_POINT_RENAME(ucol_getVariableTop) +#define ucol_getVersion U_ICU_ENTRY_POINT_RENAME(ucol_getVersion) +#define ucol_greater U_ICU_ENTRY_POINT_RENAME(ucol_greater) +#define ucol_greaterOrEqual U_ICU_ENTRY_POINT_RENAME(ucol_greaterOrEqual) +#define ucol_keyHashCode U_ICU_ENTRY_POINT_RENAME(ucol_keyHashCode) +#define ucol_looksLikeCollationBinary U_ICU_ENTRY_POINT_RENAME(ucol_looksLikeCollationBinary) +#define ucol_mergeSortkeys U_ICU_ENTRY_POINT_RENAME(ucol_mergeSortkeys) +#define ucol_next U_ICU_ENTRY_POINT_RENAME(ucol_next) +#define ucol_nextSortKeyPart U_ICU_ENTRY_POINT_RENAME(ucol_nextSortKeyPart) +#define ucol_normalizeShortDefinitionString U_ICU_ENTRY_POINT_RENAME(ucol_normalizeShortDefinitionString) +#define ucol_open U_ICU_ENTRY_POINT_RENAME(ucol_open) +#define ucol_openAvailableLocales U_ICU_ENTRY_POINT_RENAME(ucol_openAvailableLocales) +#define ucol_openBinary U_ICU_ENTRY_POINT_RENAME(ucol_openBinary) +#define ucol_openElements U_ICU_ENTRY_POINT_RENAME(ucol_openElements) +#define ucol_openFromShortString U_ICU_ENTRY_POINT_RENAME(ucol_openFromShortString) +#define ucol_openRules U_ICU_ENTRY_POINT_RENAME(ucol_openRules) +#define ucol_prepareShortStringOpen U_ICU_ENTRY_POINT_RENAME(ucol_prepareShortStringOpen) +#define ucol_previous U_ICU_ENTRY_POINT_RENAME(ucol_previous) +#define ucol_primaryOrder U_ICU_ENTRY_POINT_RENAME(ucol_primaryOrder) +#define ucol_reset U_ICU_ENTRY_POINT_RENAME(ucol_reset) +#define ucol_restoreVariableTop U_ICU_ENTRY_POINT_RENAME(ucol_restoreVariableTop) +#define ucol_safeClone U_ICU_ENTRY_POINT_RENAME(ucol_safeClone) +#define ucol_secondaryOrder U_ICU_ENTRY_POINT_RENAME(ucol_secondaryOrder) +#define ucol_setAttribute U_ICU_ENTRY_POINT_RENAME(ucol_setAttribute) +#define ucol_setMaxVariable U_ICU_ENTRY_POINT_RENAME(ucol_setMaxVariable) +#define ucol_setOffset U_ICU_ENTRY_POINT_RENAME(ucol_setOffset) +#define ucol_setReorderCodes U_ICU_ENTRY_POINT_RENAME(ucol_setReorderCodes) +#define ucol_setStrength U_ICU_ENTRY_POINT_RENAME(ucol_setStrength) +#define ucol_setText U_ICU_ENTRY_POINT_RENAME(ucol_setText) +#define ucol_setVariableTop U_ICU_ENTRY_POINT_RENAME(ucol_setVariableTop) +#define ucol_strcoll U_ICU_ENTRY_POINT_RENAME(ucol_strcoll) +#define ucol_strcollIter U_ICU_ENTRY_POINT_RENAME(ucol_strcollIter) +#define ucol_strcollUTF8 U_ICU_ENTRY_POINT_RENAME(ucol_strcollUTF8) +#define ucol_swap U_ICU_ENTRY_POINT_RENAME(ucol_swap) +#define ucol_swapInverseUCA U_ICU_ENTRY_POINT_RENAME(ucol_swapInverseUCA) +#define ucol_tertiaryOrder U_ICU_ENTRY_POINT_RENAME(ucol_tertiaryOrder) +#define ucpmap_get U_ICU_ENTRY_POINT_RENAME(ucpmap_get) +#define ucpmap_getRange U_ICU_ENTRY_POINT_RENAME(ucpmap_getRange) +#define ucptrie_close U_ICU_ENTRY_POINT_RENAME(ucptrie_close) +#define ucptrie_get U_ICU_ENTRY_POINT_RENAME(ucptrie_get) +#define ucptrie_getRange U_ICU_ENTRY_POINT_RENAME(ucptrie_getRange) +#define ucptrie_getType U_ICU_ENTRY_POINT_RENAME(ucptrie_getType) +#define ucptrie_getValueWidth U_ICU_ENTRY_POINT_RENAME(ucptrie_getValueWidth) +#define ucptrie_internalGetRange U_ICU_ENTRY_POINT_RENAME(ucptrie_internalGetRange) +#define ucptrie_internalSmallIndex U_ICU_ENTRY_POINT_RENAME(ucptrie_internalSmallIndex) +#define ucptrie_internalSmallU8Index U_ICU_ENTRY_POINT_RENAME(ucptrie_internalSmallU8Index) +#define ucptrie_internalU8PrevIndex U_ICU_ENTRY_POINT_RENAME(ucptrie_internalU8PrevIndex) +#define ucptrie_openFromBinary U_ICU_ENTRY_POINT_RENAME(ucptrie_openFromBinary) +#define ucptrie_swap U_ICU_ENTRY_POINT_RENAME(ucptrie_swap) +#define ucptrie_toBinary U_ICU_ENTRY_POINT_RENAME(ucptrie_toBinary) +#define ucsdet_close U_ICU_ENTRY_POINT_RENAME(ucsdet_close) +#define ucsdet_detect U_ICU_ENTRY_POINT_RENAME(ucsdet_detect) +#define ucsdet_detectAll U_ICU_ENTRY_POINT_RENAME(ucsdet_detectAll) +#define ucsdet_enableInputFilter U_ICU_ENTRY_POINT_RENAME(ucsdet_enableInputFilter) +#define ucsdet_getAllDetectableCharsets U_ICU_ENTRY_POINT_RENAME(ucsdet_getAllDetectableCharsets) +#define ucsdet_getConfidence U_ICU_ENTRY_POINT_RENAME(ucsdet_getConfidence) +#define ucsdet_getDetectableCharsets U_ICU_ENTRY_POINT_RENAME(ucsdet_getDetectableCharsets) +#define ucsdet_getLanguage U_ICU_ENTRY_POINT_RENAME(ucsdet_getLanguage) +#define ucsdet_getName U_ICU_ENTRY_POINT_RENAME(ucsdet_getName) +#define ucsdet_getUChars U_ICU_ENTRY_POINT_RENAME(ucsdet_getUChars) +#define ucsdet_isInputFilterEnabled U_ICU_ENTRY_POINT_RENAME(ucsdet_isInputFilterEnabled) +#define ucsdet_open U_ICU_ENTRY_POINT_RENAME(ucsdet_open) +#define ucsdet_setDeclaredEncoding U_ICU_ENTRY_POINT_RENAME(ucsdet_setDeclaredEncoding) +#define ucsdet_setDetectableCharset U_ICU_ENTRY_POINT_RENAME(ucsdet_setDetectableCharset) +#define ucsdet_setText U_ICU_ENTRY_POINT_RENAME(ucsdet_setText) +#define ucurr_countCurrencies U_ICU_ENTRY_POINT_RENAME(ucurr_countCurrencies) +#define ucurr_forLocale U_ICU_ENTRY_POINT_RENAME(ucurr_forLocale) +#define ucurr_forLocaleAndDate U_ICU_ENTRY_POINT_RENAME(ucurr_forLocaleAndDate) +#define ucurr_getDefaultFractionDigits U_ICU_ENTRY_POINT_RENAME(ucurr_getDefaultFractionDigits) +#define ucurr_getDefaultFractionDigitsForUsage U_ICU_ENTRY_POINT_RENAME(ucurr_getDefaultFractionDigitsForUsage) +#define ucurr_getKeywordValuesForLocale U_ICU_ENTRY_POINT_RENAME(ucurr_getKeywordValuesForLocale) +#define ucurr_getName U_ICU_ENTRY_POINT_RENAME(ucurr_getName) +#define ucurr_getNumericCode U_ICU_ENTRY_POINT_RENAME(ucurr_getNumericCode) +#define ucurr_getPluralName U_ICU_ENTRY_POINT_RENAME(ucurr_getPluralName) +#define ucurr_getRoundingIncrement U_ICU_ENTRY_POINT_RENAME(ucurr_getRoundingIncrement) +#define ucurr_getRoundingIncrementForUsage U_ICU_ENTRY_POINT_RENAME(ucurr_getRoundingIncrementForUsage) +#define ucurr_isAvailable U_ICU_ENTRY_POINT_RENAME(ucurr_isAvailable) +#define ucurr_openISOCurrencies U_ICU_ENTRY_POINT_RENAME(ucurr_openISOCurrencies) +#define ucurr_register U_ICU_ENTRY_POINT_RENAME(ucurr_register) +#define ucurr_unregister U_ICU_ENTRY_POINT_RENAME(ucurr_unregister) +#define udat_adoptNumberFormat U_ICU_ENTRY_POINT_RENAME(udat_adoptNumberFormat) +#define udat_adoptNumberFormatForFields U_ICU_ENTRY_POINT_RENAME(udat_adoptNumberFormatForFields) +#define udat_applyPattern U_ICU_ENTRY_POINT_RENAME(udat_applyPattern) +#define udat_applyPatternRelative U_ICU_ENTRY_POINT_RENAME(udat_applyPatternRelative) +#define udat_clone U_ICU_ENTRY_POINT_RENAME(udat_clone) +#define udat_close U_ICU_ENTRY_POINT_RENAME(udat_close) +#define udat_countAvailable U_ICU_ENTRY_POINT_RENAME(udat_countAvailable) +#define udat_countSymbols U_ICU_ENTRY_POINT_RENAME(udat_countSymbols) +#define udat_format U_ICU_ENTRY_POINT_RENAME(udat_format) +#define udat_formatCalendar U_ICU_ENTRY_POINT_RENAME(udat_formatCalendar) +#define udat_formatCalendarForFields U_ICU_ENTRY_POINT_RENAME(udat_formatCalendarForFields) +#define udat_formatForFields U_ICU_ENTRY_POINT_RENAME(udat_formatForFields) +#define udat_get2DigitYearStart U_ICU_ENTRY_POINT_RENAME(udat_get2DigitYearStart) +#define udat_getAvailable U_ICU_ENTRY_POINT_RENAME(udat_getAvailable) +#define udat_getBooleanAttribute U_ICU_ENTRY_POINT_RENAME(udat_getBooleanAttribute) +#define udat_getCalendar U_ICU_ENTRY_POINT_RENAME(udat_getCalendar) +#define udat_getContext U_ICU_ENTRY_POINT_RENAME(udat_getContext) +#define udat_getLocaleByType U_ICU_ENTRY_POINT_RENAME(udat_getLocaleByType) +#define udat_getNumberFormat U_ICU_ENTRY_POINT_RENAME(udat_getNumberFormat) +#define udat_getNumberFormatForField U_ICU_ENTRY_POINT_RENAME(udat_getNumberFormatForField) +#define udat_getSymbols U_ICU_ENTRY_POINT_RENAME(udat_getSymbols) +#define udat_isLenient U_ICU_ENTRY_POINT_RENAME(udat_isLenient) +#define udat_open U_ICU_ENTRY_POINT_RENAME(udat_open) +#define udat_parse U_ICU_ENTRY_POINT_RENAME(udat_parse) +#define udat_parseCalendar U_ICU_ENTRY_POINT_RENAME(udat_parseCalendar) +#define udat_registerOpener U_ICU_ENTRY_POINT_RENAME(udat_registerOpener) +#define udat_set2DigitYearStart U_ICU_ENTRY_POINT_RENAME(udat_set2DigitYearStart) +#define udat_setBooleanAttribute U_ICU_ENTRY_POINT_RENAME(udat_setBooleanAttribute) +#define udat_setCalendar U_ICU_ENTRY_POINT_RENAME(udat_setCalendar) +#define udat_setContext U_ICU_ENTRY_POINT_RENAME(udat_setContext) +#define udat_setLenient U_ICU_ENTRY_POINT_RENAME(udat_setLenient) +#define udat_setNumberFormat U_ICU_ENTRY_POINT_RENAME(udat_setNumberFormat) +#define udat_setSymbols U_ICU_ENTRY_POINT_RENAME(udat_setSymbols) +#define udat_toCalendarDateField U_ICU_ENTRY_POINT_RENAME(udat_toCalendarDateField) +#define udat_toPattern U_ICU_ENTRY_POINT_RENAME(udat_toPattern) +#define udat_toPatternRelativeDate U_ICU_ENTRY_POINT_RENAME(udat_toPatternRelativeDate) +#define udat_toPatternRelativeTime U_ICU_ENTRY_POINT_RENAME(udat_toPatternRelativeTime) +#define udat_unregisterOpener U_ICU_ENTRY_POINT_RENAME(udat_unregisterOpener) +#define udata_checkCommonData U_ICU_ENTRY_POINT_RENAME(udata_checkCommonData) +#define udata_close U_ICU_ENTRY_POINT_RENAME(udata_close) +#define udata_closeSwapper U_ICU_ENTRY_POINT_RENAME(udata_closeSwapper) +#define udata_getHeaderSize U_ICU_ENTRY_POINT_RENAME(udata_getHeaderSize) +#define udata_getInfo U_ICU_ENTRY_POINT_RENAME(udata_getInfo) +#define udata_getInfoSize U_ICU_ENTRY_POINT_RENAME(udata_getInfoSize) +#define udata_getLength U_ICU_ENTRY_POINT_RENAME(udata_getLength) +#define udata_getMemory U_ICU_ENTRY_POINT_RENAME(udata_getMemory) +#define udata_getRawMemory U_ICU_ENTRY_POINT_RENAME(udata_getRawMemory) +#define udata_open U_ICU_ENTRY_POINT_RENAME(udata_open) +#define udata_openChoice U_ICU_ENTRY_POINT_RENAME(udata_openChoice) +#define udata_openSwapper U_ICU_ENTRY_POINT_RENAME(udata_openSwapper) +#define udata_openSwapperForInputData U_ICU_ENTRY_POINT_RENAME(udata_openSwapperForInputData) +#define udata_printError U_ICU_ENTRY_POINT_RENAME(udata_printError) +#define udata_readInt16 U_ICU_ENTRY_POINT_RENAME(udata_readInt16) +#define udata_readInt32 U_ICU_ENTRY_POINT_RENAME(udata_readInt32) +#define udata_setAppData U_ICU_ENTRY_POINT_RENAME(udata_setAppData) +#define udata_setCommonData U_ICU_ENTRY_POINT_RENAME(udata_setCommonData) +#define udata_setFileAccess U_ICU_ENTRY_POINT_RENAME(udata_setFileAccess) +#define udata_swapDataHeader U_ICU_ENTRY_POINT_RENAME(udata_swapDataHeader) +#define udata_swapInvStringBlock U_ICU_ENTRY_POINT_RENAME(udata_swapInvStringBlock) +#define udatpg_addPattern U_ICU_ENTRY_POINT_RENAME(udatpg_addPattern) +#define udatpg_clone U_ICU_ENTRY_POINT_RENAME(udatpg_clone) +#define udatpg_close U_ICU_ENTRY_POINT_RENAME(udatpg_close) +#define udatpg_getAppendItemFormat U_ICU_ENTRY_POINT_RENAME(udatpg_getAppendItemFormat) +#define udatpg_getAppendItemName U_ICU_ENTRY_POINT_RENAME(udatpg_getAppendItemName) +#define udatpg_getBaseSkeleton U_ICU_ENTRY_POINT_RENAME(udatpg_getBaseSkeleton) +#define udatpg_getBestPattern U_ICU_ENTRY_POINT_RENAME(udatpg_getBestPattern) +#define udatpg_getBestPatternWithOptions U_ICU_ENTRY_POINT_RENAME(udatpg_getBestPatternWithOptions) +#define udatpg_getDateTimeFormat U_ICU_ENTRY_POINT_RENAME(udatpg_getDateTimeFormat) +#define udatpg_getDateTimeFormatForStyle U_ICU_ENTRY_POINT_RENAME(udatpg_getDateTimeFormatForStyle) +#define udatpg_getDecimal U_ICU_ENTRY_POINT_RENAME(udatpg_getDecimal) +#define udatpg_getDefaultHourCycle U_ICU_ENTRY_POINT_RENAME(udatpg_getDefaultHourCycle) +#define udatpg_getFieldDisplayName U_ICU_ENTRY_POINT_RENAME(udatpg_getFieldDisplayName) +#define udatpg_getPatternForSkeleton U_ICU_ENTRY_POINT_RENAME(udatpg_getPatternForSkeleton) +#define udatpg_getSkeleton U_ICU_ENTRY_POINT_RENAME(udatpg_getSkeleton) +#define udatpg_open U_ICU_ENTRY_POINT_RENAME(udatpg_open) +#define udatpg_openBaseSkeletons U_ICU_ENTRY_POINT_RENAME(udatpg_openBaseSkeletons) +#define udatpg_openEmpty U_ICU_ENTRY_POINT_RENAME(udatpg_openEmpty) +#define udatpg_openSkeletons U_ICU_ENTRY_POINT_RENAME(udatpg_openSkeletons) +#define udatpg_replaceFieldTypes U_ICU_ENTRY_POINT_RENAME(udatpg_replaceFieldTypes) +#define udatpg_replaceFieldTypesWithOptions U_ICU_ENTRY_POINT_RENAME(udatpg_replaceFieldTypesWithOptions) +#define udatpg_setAppendItemFormat U_ICU_ENTRY_POINT_RENAME(udatpg_setAppendItemFormat) +#define udatpg_setAppendItemName U_ICU_ENTRY_POINT_RENAME(udatpg_setAppendItemName) +#define udatpg_setDateTimeFormat U_ICU_ENTRY_POINT_RENAME(udatpg_setDateTimeFormat) +#define udatpg_setDateTimeFormatForStyle U_ICU_ENTRY_POINT_RENAME(udatpg_setDateTimeFormatForStyle) +#define udatpg_setDecimal U_ICU_ENTRY_POINT_RENAME(udatpg_setDecimal) +#define udict_swap U_ICU_ENTRY_POINT_RENAME(udict_swap) +#define udispopt_fromGrammaticalCaseIdentifier U_ICU_ENTRY_POINT_RENAME(udispopt_fromGrammaticalCaseIdentifier) +#define udispopt_fromNounClassIdentifier U_ICU_ENTRY_POINT_RENAME(udispopt_fromNounClassIdentifier) +#define udispopt_fromPluralCategoryIdentifier U_ICU_ENTRY_POINT_RENAME(udispopt_fromPluralCategoryIdentifier) +#define udispopt_getGrammaticalCaseIdentifier U_ICU_ENTRY_POINT_RENAME(udispopt_getGrammaticalCaseIdentifier) +#define udispopt_getNounClassIdentifier U_ICU_ENTRY_POINT_RENAME(udispopt_getNounClassIdentifier) +#define udispopt_getPluralCategoryIdentifier U_ICU_ENTRY_POINT_RENAME(udispopt_getPluralCategoryIdentifier) +#define udtitvfmt_close U_ICU_ENTRY_POINT_RENAME(udtitvfmt_close) +#define udtitvfmt_closeResult U_ICU_ENTRY_POINT_RENAME(udtitvfmt_closeResult) +#define udtitvfmt_format U_ICU_ENTRY_POINT_RENAME(udtitvfmt_format) +#define udtitvfmt_formatCalendarToResult U_ICU_ENTRY_POINT_RENAME(udtitvfmt_formatCalendarToResult) +#define udtitvfmt_formatToResult U_ICU_ENTRY_POINT_RENAME(udtitvfmt_formatToResult) +#define udtitvfmt_getContext U_ICU_ENTRY_POINT_RENAME(udtitvfmt_getContext) +#define udtitvfmt_open U_ICU_ENTRY_POINT_RENAME(udtitvfmt_open) +#define udtitvfmt_openResult U_ICU_ENTRY_POINT_RENAME(udtitvfmt_openResult) +#define udtitvfmt_resultAsValue U_ICU_ENTRY_POINT_RENAME(udtitvfmt_resultAsValue) +#define udtitvfmt_setContext U_ICU_ENTRY_POINT_RENAME(udtitvfmt_setContext) +#define uenum_close U_ICU_ENTRY_POINT_RENAME(uenum_close) +#define uenum_count U_ICU_ENTRY_POINT_RENAME(uenum_count) +#define uenum_next U_ICU_ENTRY_POINT_RENAME(uenum_next) +#define uenum_nextDefault U_ICU_ENTRY_POINT_RENAME(uenum_nextDefault) +#define uenum_openCharStringsEnumeration U_ICU_ENTRY_POINT_RENAME(uenum_openCharStringsEnumeration) +#define uenum_openFromStringEnumeration U_ICU_ENTRY_POINT_RENAME(uenum_openFromStringEnumeration) +#define uenum_openUCharStringsEnumeration U_ICU_ENTRY_POINT_RENAME(uenum_openUCharStringsEnumeration) +#define uenum_reset U_ICU_ENTRY_POINT_RENAME(uenum_reset) +#define uenum_unext U_ICU_ENTRY_POINT_RENAME(uenum_unext) +#define uenum_unextDefault U_ICU_ENTRY_POINT_RENAME(uenum_unextDefault) +#define ufieldpositer_close U_ICU_ENTRY_POINT_RENAME(ufieldpositer_close) +#define ufieldpositer_next U_ICU_ENTRY_POINT_RENAME(ufieldpositer_next) +#define ufieldpositer_open U_ICU_ENTRY_POINT_RENAME(ufieldpositer_open) +#define ufile_getch U_ICU_ENTRY_POINT_RENAME(ufile_getch) +#define ufile_getch32 U_ICU_ENTRY_POINT_RENAME(ufile_getch32) +#define ufmt_close U_ICU_ENTRY_POINT_RENAME(ufmt_close) +#define ufmt_getArrayItemByIndex U_ICU_ENTRY_POINT_RENAME(ufmt_getArrayItemByIndex) +#define ufmt_getArrayLength U_ICU_ENTRY_POINT_RENAME(ufmt_getArrayLength) +#define ufmt_getDate U_ICU_ENTRY_POINT_RENAME(ufmt_getDate) +#define ufmt_getDecNumChars U_ICU_ENTRY_POINT_RENAME(ufmt_getDecNumChars) +#define ufmt_getDouble U_ICU_ENTRY_POINT_RENAME(ufmt_getDouble) +#define ufmt_getInt64 U_ICU_ENTRY_POINT_RENAME(ufmt_getInt64) +#define ufmt_getLong U_ICU_ENTRY_POINT_RENAME(ufmt_getLong) +#define ufmt_getObject U_ICU_ENTRY_POINT_RENAME(ufmt_getObject) +#define ufmt_getType U_ICU_ENTRY_POINT_RENAME(ufmt_getType) +#define ufmt_getUChars U_ICU_ENTRY_POINT_RENAME(ufmt_getUChars) +#define ufmt_isNumeric U_ICU_ENTRY_POINT_RENAME(ufmt_isNumeric) +#define ufmt_open U_ICU_ENTRY_POINT_RENAME(ufmt_open) +#define ufmtval_getString U_ICU_ENTRY_POINT_RENAME(ufmtval_getString) +#define ufmtval_nextPosition U_ICU_ENTRY_POINT_RENAME(ufmtval_nextPosition) +#define ugender_getInstance U_ICU_ENTRY_POINT_RENAME(ugender_getInstance) +#define ugender_getListGender U_ICU_ENTRY_POINT_RENAME(ugender_getListGender) +#define uhash_close U_ICU_ENTRY_POINT_RENAME(uhash_close) +#define uhash_compareCaselessUnicodeString U_ICU_ENTRY_POINT_RENAME(uhash_compareCaselessUnicodeString) +#define uhash_compareChars U_ICU_ENTRY_POINT_RENAME(uhash_compareChars) +#define uhash_compareIChars U_ICU_ENTRY_POINT_RENAME(uhash_compareIChars) +#define uhash_compareLong U_ICU_ENTRY_POINT_RENAME(uhash_compareLong) +#define uhash_compareScriptSet U_ICU_ENTRY_POINT_RENAME(uhash_compareScriptSet) +#define uhash_compareUChars U_ICU_ENTRY_POINT_RENAME(uhash_compareUChars) +#define uhash_compareUnicodeString U_ICU_ENTRY_POINT_RENAME(uhash_compareUnicodeString) +#define uhash_containsKey U_ICU_ENTRY_POINT_RENAME(uhash_containsKey) +#define uhash_count U_ICU_ENTRY_POINT_RENAME(uhash_count) +#define uhash_deleteHashtable U_ICU_ENTRY_POINT_RENAME(uhash_deleteHashtable) +#define uhash_deleteScriptSet U_ICU_ENTRY_POINT_RENAME(uhash_deleteScriptSet) +#define uhash_equals U_ICU_ENTRY_POINT_RENAME(uhash_equals) +#define uhash_equalsScriptSet U_ICU_ENTRY_POINT_RENAME(uhash_equalsScriptSet) +#define uhash_find U_ICU_ENTRY_POINT_RENAME(uhash_find) +#define uhash_get U_ICU_ENTRY_POINT_RENAME(uhash_get) +#define uhash_geti U_ICU_ENTRY_POINT_RENAME(uhash_geti) +#define uhash_getiAndFound U_ICU_ENTRY_POINT_RENAME(uhash_getiAndFound) +#define uhash_hashCaselessUnicodeString U_ICU_ENTRY_POINT_RENAME(uhash_hashCaselessUnicodeString) +#define uhash_hashChars U_ICU_ENTRY_POINT_RENAME(uhash_hashChars) +#define uhash_hashIChars U_ICU_ENTRY_POINT_RENAME(uhash_hashIChars) +#define uhash_hashLong U_ICU_ENTRY_POINT_RENAME(uhash_hashLong) +#define uhash_hashScriptSet U_ICU_ENTRY_POINT_RENAME(uhash_hashScriptSet) +#define uhash_hashUChars U_ICU_ENTRY_POINT_RENAME(uhash_hashUChars) +#define uhash_hashUnicodeString U_ICU_ENTRY_POINT_RENAME(uhash_hashUnicodeString) +#define uhash_icontainsKey U_ICU_ENTRY_POINT_RENAME(uhash_icontainsKey) +#define uhash_iget U_ICU_ENTRY_POINT_RENAME(uhash_iget) +#define uhash_igeti U_ICU_ENTRY_POINT_RENAME(uhash_igeti) +#define uhash_igetiAndFound U_ICU_ENTRY_POINT_RENAME(uhash_igetiAndFound) +#define uhash_init U_ICU_ENTRY_POINT_RENAME(uhash_init) +#define uhash_initSize U_ICU_ENTRY_POINT_RENAME(uhash_initSize) +#define uhash_iput U_ICU_ENTRY_POINT_RENAME(uhash_iput) +#define uhash_iputi U_ICU_ENTRY_POINT_RENAME(uhash_iputi) +#define uhash_iputiAllowZero U_ICU_ENTRY_POINT_RENAME(uhash_iputiAllowZero) +#define uhash_iremove U_ICU_ENTRY_POINT_RENAME(uhash_iremove) +#define uhash_iremovei U_ICU_ENTRY_POINT_RENAME(uhash_iremovei) +#define uhash_nextElement U_ICU_ENTRY_POINT_RENAME(uhash_nextElement) +#define uhash_open U_ICU_ENTRY_POINT_RENAME(uhash_open) +#define uhash_openSize U_ICU_ENTRY_POINT_RENAME(uhash_openSize) +#define uhash_put U_ICU_ENTRY_POINT_RENAME(uhash_put) +#define uhash_puti U_ICU_ENTRY_POINT_RENAME(uhash_puti) +#define uhash_putiAllowZero U_ICU_ENTRY_POINT_RENAME(uhash_putiAllowZero) +#define uhash_remove U_ICU_ENTRY_POINT_RENAME(uhash_remove) +#define uhash_removeAll U_ICU_ENTRY_POINT_RENAME(uhash_removeAll) +#define uhash_removeElement U_ICU_ENTRY_POINT_RENAME(uhash_removeElement) +#define uhash_removei U_ICU_ENTRY_POINT_RENAME(uhash_removei) +#define uhash_setKeyComparator U_ICU_ENTRY_POINT_RENAME(uhash_setKeyComparator) +#define uhash_setKeyDeleter U_ICU_ENTRY_POINT_RENAME(uhash_setKeyDeleter) +#define uhash_setKeyHasher U_ICU_ENTRY_POINT_RENAME(uhash_setKeyHasher) +#define uhash_setResizePolicy U_ICU_ENTRY_POINT_RENAME(uhash_setResizePolicy) +#define uhash_setValueComparator U_ICU_ENTRY_POINT_RENAME(uhash_setValueComparator) +#define uhash_setValueDeleter U_ICU_ENTRY_POINT_RENAME(uhash_setValueDeleter) +#define uidna_IDNToASCII U_ICU_ENTRY_POINT_RENAME(uidna_IDNToASCII) +#define uidna_IDNToUnicode U_ICU_ENTRY_POINT_RENAME(uidna_IDNToUnicode) +#define uidna_close U_ICU_ENTRY_POINT_RENAME(uidna_close) +#define uidna_compare U_ICU_ENTRY_POINT_RENAME(uidna_compare) +#define uidna_labelToASCII U_ICU_ENTRY_POINT_RENAME(uidna_labelToASCII) +#define uidna_labelToASCII_UTF8 U_ICU_ENTRY_POINT_RENAME(uidna_labelToASCII_UTF8) +#define uidna_labelToUnicode U_ICU_ENTRY_POINT_RENAME(uidna_labelToUnicode) +#define uidna_labelToUnicodeUTF8 U_ICU_ENTRY_POINT_RENAME(uidna_labelToUnicodeUTF8) +#define uidna_nameToASCII U_ICU_ENTRY_POINT_RENAME(uidna_nameToASCII) +#define uidna_nameToASCII_UTF8 U_ICU_ENTRY_POINT_RENAME(uidna_nameToASCII_UTF8) +#define uidna_nameToUnicode U_ICU_ENTRY_POINT_RENAME(uidna_nameToUnicode) +#define uidna_nameToUnicodeUTF8 U_ICU_ENTRY_POINT_RENAME(uidna_nameToUnicodeUTF8) +#define uidna_openUTS46 U_ICU_ENTRY_POINT_RENAME(uidna_openUTS46) +#define uidna_toASCII U_ICU_ENTRY_POINT_RENAME(uidna_toASCII) +#define uidna_toUnicode U_ICU_ENTRY_POINT_RENAME(uidna_toUnicode) +#define uiter_current32 U_ICU_ENTRY_POINT_RENAME(uiter_current32) +#define uiter_getState U_ICU_ENTRY_POINT_RENAME(uiter_getState) +#define uiter_next32 U_ICU_ENTRY_POINT_RENAME(uiter_next32) +#define uiter_previous32 U_ICU_ENTRY_POINT_RENAME(uiter_previous32) +#define uiter_setCharacterIterator U_ICU_ENTRY_POINT_RENAME(uiter_setCharacterIterator) +#define uiter_setReplaceable U_ICU_ENTRY_POINT_RENAME(uiter_setReplaceable) +#define uiter_setState U_ICU_ENTRY_POINT_RENAME(uiter_setState) +#define uiter_setString U_ICU_ENTRY_POINT_RENAME(uiter_setString) +#define uiter_setUTF16BE U_ICU_ENTRY_POINT_RENAME(uiter_setUTF16BE) +#define uiter_setUTF8 U_ICU_ENTRY_POINT_RENAME(uiter_setUTF8) +#define uldn_close U_ICU_ENTRY_POINT_RENAME(uldn_close) +#define uldn_getContext U_ICU_ENTRY_POINT_RENAME(uldn_getContext) +#define uldn_getDialectHandling U_ICU_ENTRY_POINT_RENAME(uldn_getDialectHandling) +#define uldn_getLocale U_ICU_ENTRY_POINT_RENAME(uldn_getLocale) +#define uldn_keyDisplayName U_ICU_ENTRY_POINT_RENAME(uldn_keyDisplayName) +#define uldn_keyValueDisplayName U_ICU_ENTRY_POINT_RENAME(uldn_keyValueDisplayName) +#define uldn_languageDisplayName U_ICU_ENTRY_POINT_RENAME(uldn_languageDisplayName) +#define uldn_localeDisplayName U_ICU_ENTRY_POINT_RENAME(uldn_localeDisplayName) +#define uldn_open U_ICU_ENTRY_POINT_RENAME(uldn_open) +#define uldn_openForContext U_ICU_ENTRY_POINT_RENAME(uldn_openForContext) +#define uldn_regionDisplayName U_ICU_ENTRY_POINT_RENAME(uldn_regionDisplayName) +#define uldn_scriptCodeDisplayName U_ICU_ENTRY_POINT_RENAME(uldn_scriptCodeDisplayName) +#define uldn_scriptDisplayName U_ICU_ENTRY_POINT_RENAME(uldn_scriptDisplayName) +#define uldn_variantDisplayName U_ICU_ENTRY_POINT_RENAME(uldn_variantDisplayName) +#define ulist_addItemBeginList U_ICU_ENTRY_POINT_RENAME(ulist_addItemBeginList) +#define ulist_addItemEndList U_ICU_ENTRY_POINT_RENAME(ulist_addItemEndList) +#define ulist_close_keyword_values_iterator U_ICU_ENTRY_POINT_RENAME(ulist_close_keyword_values_iterator) +#define ulist_containsString U_ICU_ENTRY_POINT_RENAME(ulist_containsString) +#define ulist_count_keyword_values U_ICU_ENTRY_POINT_RENAME(ulist_count_keyword_values) +#define ulist_createEmptyList U_ICU_ENTRY_POINT_RENAME(ulist_createEmptyList) +#define ulist_deleteList U_ICU_ENTRY_POINT_RENAME(ulist_deleteList) +#define ulist_getListFromEnum U_ICU_ENTRY_POINT_RENAME(ulist_getListFromEnum) +#define ulist_getListSize U_ICU_ENTRY_POINT_RENAME(ulist_getListSize) +#define ulist_getNext U_ICU_ENTRY_POINT_RENAME(ulist_getNext) +#define ulist_next_keyword_value U_ICU_ENTRY_POINT_RENAME(ulist_next_keyword_value) +#define ulist_removeString U_ICU_ENTRY_POINT_RENAME(ulist_removeString) +#define ulist_resetList U_ICU_ENTRY_POINT_RENAME(ulist_resetList) +#define ulist_reset_keyword_values_iterator U_ICU_ENTRY_POINT_RENAME(ulist_reset_keyword_values_iterator) +#define ulistfmt_close U_ICU_ENTRY_POINT_RENAME(ulistfmt_close) +#define ulistfmt_closeResult U_ICU_ENTRY_POINT_RENAME(ulistfmt_closeResult) +#define ulistfmt_format U_ICU_ENTRY_POINT_RENAME(ulistfmt_format) +#define ulistfmt_formatStringsToResult U_ICU_ENTRY_POINT_RENAME(ulistfmt_formatStringsToResult) +#define ulistfmt_open U_ICU_ENTRY_POINT_RENAME(ulistfmt_open) +#define ulistfmt_openForType U_ICU_ENTRY_POINT_RENAME(ulistfmt_openForType) +#define ulistfmt_openResult U_ICU_ENTRY_POINT_RENAME(ulistfmt_openResult) +#define ulistfmt_resultAsValue U_ICU_ENTRY_POINT_RENAME(ulistfmt_resultAsValue) +#define uloc_acceptLanguage U_ICU_ENTRY_POINT_RENAME(uloc_acceptLanguage) +#define uloc_acceptLanguageFromHTTP U_ICU_ENTRY_POINT_RENAME(uloc_acceptLanguageFromHTTP) +#define uloc_addLikelySubtags U_ICU_ENTRY_POINT_RENAME(uloc_addLikelySubtags) +#define uloc_canonicalize U_ICU_ENTRY_POINT_RENAME(uloc_canonicalize) +#define uloc_countAvailable U_ICU_ENTRY_POINT_RENAME(uloc_countAvailable) +#define uloc_forLanguageTag U_ICU_ENTRY_POINT_RENAME(uloc_forLanguageTag) +#define uloc_getAvailable U_ICU_ENTRY_POINT_RENAME(uloc_getAvailable) +#define uloc_getBaseName U_ICU_ENTRY_POINT_RENAME(uloc_getBaseName) +#define uloc_getCharacterOrientation U_ICU_ENTRY_POINT_RENAME(uloc_getCharacterOrientation) +#define uloc_getCountry U_ICU_ENTRY_POINT_RENAME(uloc_getCountry) +#define uloc_getCurrentCountryID U_ICU_ENTRY_POINT_RENAME(uloc_getCurrentCountryID) +#define uloc_getCurrentLanguageID U_ICU_ENTRY_POINT_RENAME(uloc_getCurrentLanguageID) +#define uloc_getDefault U_ICU_ENTRY_POINT_RENAME(uloc_getDefault) +#define uloc_getDisplayCountry U_ICU_ENTRY_POINT_RENAME(uloc_getDisplayCountry) +#define uloc_getDisplayKeyword U_ICU_ENTRY_POINT_RENAME(uloc_getDisplayKeyword) +#define uloc_getDisplayKeywordValue U_ICU_ENTRY_POINT_RENAME(uloc_getDisplayKeywordValue) +#define uloc_getDisplayLanguage U_ICU_ENTRY_POINT_RENAME(uloc_getDisplayLanguage) +#define uloc_getDisplayName U_ICU_ENTRY_POINT_RENAME(uloc_getDisplayName) +#define uloc_getDisplayScript U_ICU_ENTRY_POINT_RENAME(uloc_getDisplayScript) +#define uloc_getDisplayVariant U_ICU_ENTRY_POINT_RENAME(uloc_getDisplayVariant) +#define uloc_getISO3Country U_ICU_ENTRY_POINT_RENAME(uloc_getISO3Country) +#define uloc_getISO3Language U_ICU_ENTRY_POINT_RENAME(uloc_getISO3Language) +#define uloc_getISOCountries U_ICU_ENTRY_POINT_RENAME(uloc_getISOCountries) +#define uloc_getISOLanguages U_ICU_ENTRY_POINT_RENAME(uloc_getISOLanguages) +#define uloc_getKeywordValue U_ICU_ENTRY_POINT_RENAME(uloc_getKeywordValue) +#define uloc_getLCID U_ICU_ENTRY_POINT_RENAME(uloc_getLCID) +#define uloc_getLanguage U_ICU_ENTRY_POINT_RENAME(uloc_getLanguage) +#define uloc_getLineOrientation U_ICU_ENTRY_POINT_RENAME(uloc_getLineOrientation) +#define uloc_getLocaleForLCID U_ICU_ENTRY_POINT_RENAME(uloc_getLocaleForLCID) +#define uloc_getName U_ICU_ENTRY_POINT_RENAME(uloc_getName) +#define uloc_getParent U_ICU_ENTRY_POINT_RENAME(uloc_getParent) +#define uloc_getScript U_ICU_ENTRY_POINT_RENAME(uloc_getScript) +#define uloc_getTableStringWithFallback U_ICU_ENTRY_POINT_RENAME(uloc_getTableStringWithFallback) +#define uloc_getVariant U_ICU_ENTRY_POINT_RENAME(uloc_getVariant) +#define uloc_isRightToLeft U_ICU_ENTRY_POINT_RENAME(uloc_isRightToLeft) +#define uloc_minimizeSubtags U_ICU_ENTRY_POINT_RENAME(uloc_minimizeSubtags) +#define uloc_openAvailableByType U_ICU_ENTRY_POINT_RENAME(uloc_openAvailableByType) +#define uloc_openKeywordList U_ICU_ENTRY_POINT_RENAME(uloc_openKeywordList) +#define uloc_openKeywords U_ICU_ENTRY_POINT_RENAME(uloc_openKeywords) +#define uloc_setDefault U_ICU_ENTRY_POINT_RENAME(uloc_setDefault) +#define uloc_setKeywordValue U_ICU_ENTRY_POINT_RENAME(uloc_setKeywordValue) +#define uloc_toLanguageTag U_ICU_ENTRY_POINT_RENAME(uloc_toLanguageTag) +#define uloc_toLegacyKey U_ICU_ENTRY_POINT_RENAME(uloc_toLegacyKey) +#define uloc_toLegacyType U_ICU_ENTRY_POINT_RENAME(uloc_toLegacyType) +#define uloc_toUnicodeLocaleKey U_ICU_ENTRY_POINT_RENAME(uloc_toUnicodeLocaleKey) +#define uloc_toUnicodeLocaleType U_ICU_ENTRY_POINT_RENAME(uloc_toUnicodeLocaleType) +#define ulocdata_close U_ICU_ENTRY_POINT_RENAME(ulocdata_close) +#define ulocdata_getCLDRVersion U_ICU_ENTRY_POINT_RENAME(ulocdata_getCLDRVersion) +#define ulocdata_getDelimiter U_ICU_ENTRY_POINT_RENAME(ulocdata_getDelimiter) +#define ulocdata_getExemplarSet U_ICU_ENTRY_POINT_RENAME(ulocdata_getExemplarSet) +#define ulocdata_getLocaleDisplayPattern U_ICU_ENTRY_POINT_RENAME(ulocdata_getLocaleDisplayPattern) +#define ulocdata_getLocaleSeparator U_ICU_ENTRY_POINT_RENAME(ulocdata_getLocaleSeparator) +#define ulocdata_getMeasurementSystem U_ICU_ENTRY_POINT_RENAME(ulocdata_getMeasurementSystem) +#define ulocdata_getNoSubstitute U_ICU_ENTRY_POINT_RENAME(ulocdata_getNoSubstitute) +#define ulocdata_getPaperSize U_ICU_ENTRY_POINT_RENAME(ulocdata_getPaperSize) +#define ulocdata_open U_ICU_ENTRY_POINT_RENAME(ulocdata_open) +#define ulocdata_setNoSubstitute U_ICU_ENTRY_POINT_RENAME(ulocdata_setNoSubstitute) +#define ulocimp_addLikelySubtags U_ICU_ENTRY_POINT_RENAME(ulocimp_addLikelySubtags) +#define ulocimp_canonicalize U_ICU_ENTRY_POINT_RENAME(ulocimp_canonicalize) +#define ulocimp_forLanguageTag U_ICU_ENTRY_POINT_RENAME(ulocimp_forLanguageTag) +#define ulocimp_getBaseName U_ICU_ENTRY_POINT_RENAME(ulocimp_getBaseName) +#define ulocimp_getCountry U_ICU_ENTRY_POINT_RENAME(ulocimp_getCountry) +#define ulocimp_getKeywordValue U_ICU_ENTRY_POINT_RENAME(ulocimp_getKeywordValue) +#define ulocimp_getKeywords U_ICU_ENTRY_POINT_RENAME(ulocimp_getKeywords) +#define ulocimp_getKnownCanonicalizedLocaleForTest U_ICU_ENTRY_POINT_RENAME(ulocimp_getKnownCanonicalizedLocaleForTest) +#define ulocimp_getLanguage U_ICU_ENTRY_POINT_RENAME(ulocimp_getLanguage) +#define ulocimp_getName U_ICU_ENTRY_POINT_RENAME(ulocimp_getName) +#define ulocimp_getRegionForSupplementalData U_ICU_ENTRY_POINT_RENAME(ulocimp_getRegionForSupplementalData) +#define ulocimp_getScript U_ICU_ENTRY_POINT_RENAME(ulocimp_getScript) +#define ulocimp_isCanonicalizedLocaleForTest U_ICU_ENTRY_POINT_RENAME(ulocimp_isCanonicalizedLocaleForTest) +#define ulocimp_minimizeSubtags U_ICU_ENTRY_POINT_RENAME(ulocimp_minimizeSubtags) +#define ulocimp_toBcpKey U_ICU_ENTRY_POINT_RENAME(ulocimp_toBcpKey) +#define ulocimp_toBcpType U_ICU_ENTRY_POINT_RENAME(ulocimp_toBcpType) +#define ulocimp_toLanguageTag U_ICU_ENTRY_POINT_RENAME(ulocimp_toLanguageTag) +#define ulocimp_toLegacyKey U_ICU_ENTRY_POINT_RENAME(ulocimp_toLegacyKey) +#define ulocimp_toLegacyType U_ICU_ENTRY_POINT_RENAME(ulocimp_toLegacyType) +#define ultag_getTKeyStart U_ICU_ENTRY_POINT_RENAME(ultag_getTKeyStart) +#define ultag_isExtensionSubtags U_ICU_ENTRY_POINT_RENAME(ultag_isExtensionSubtags) +#define ultag_isLanguageSubtag U_ICU_ENTRY_POINT_RENAME(ultag_isLanguageSubtag) +#define ultag_isPrivateuseValueSubtags U_ICU_ENTRY_POINT_RENAME(ultag_isPrivateuseValueSubtags) +#define ultag_isRegionSubtag U_ICU_ENTRY_POINT_RENAME(ultag_isRegionSubtag) +#define ultag_isScriptSubtag U_ICU_ENTRY_POINT_RENAME(ultag_isScriptSubtag) +#define ultag_isTransformedExtensionSubtags U_ICU_ENTRY_POINT_RENAME(ultag_isTransformedExtensionSubtags) +#define ultag_isUnicodeExtensionSubtags U_ICU_ENTRY_POINT_RENAME(ultag_isUnicodeExtensionSubtags) +#define ultag_isUnicodeLocaleAttribute U_ICU_ENTRY_POINT_RENAME(ultag_isUnicodeLocaleAttribute) +#define ultag_isUnicodeLocaleAttributes U_ICU_ENTRY_POINT_RENAME(ultag_isUnicodeLocaleAttributes) +#define ultag_isUnicodeLocaleKey U_ICU_ENTRY_POINT_RENAME(ultag_isUnicodeLocaleKey) +#define ultag_isUnicodeLocaleType U_ICU_ENTRY_POINT_RENAME(ultag_isUnicodeLocaleType) +#define ultag_isVariantSubtags U_ICU_ENTRY_POINT_RENAME(ultag_isVariantSubtags) +#define umeas_getPrefixBase U_ICU_ENTRY_POINT_RENAME(umeas_getPrefixBase) +#define umeas_getPrefixPower U_ICU_ENTRY_POINT_RENAME(umeas_getPrefixPower) +#define umsg_applyPattern U_ICU_ENTRY_POINT_RENAME(umsg_applyPattern) +#define umsg_autoQuoteApostrophe U_ICU_ENTRY_POINT_RENAME(umsg_autoQuoteApostrophe) +#define umsg_clone U_ICU_ENTRY_POINT_RENAME(umsg_clone) +#define umsg_close U_ICU_ENTRY_POINT_RENAME(umsg_close) +#define umsg_format U_ICU_ENTRY_POINT_RENAME(umsg_format) +#define umsg_getLocale U_ICU_ENTRY_POINT_RENAME(umsg_getLocale) +#define umsg_open U_ICU_ENTRY_POINT_RENAME(umsg_open) +#define umsg_parse U_ICU_ENTRY_POINT_RENAME(umsg_parse) +#define umsg_setLocale U_ICU_ENTRY_POINT_RENAME(umsg_setLocale) +#define umsg_toPattern U_ICU_ENTRY_POINT_RENAME(umsg_toPattern) +#define umsg_vformat U_ICU_ENTRY_POINT_RENAME(umsg_vformat) +#define umsg_vparse U_ICU_ENTRY_POINT_RENAME(umsg_vparse) +#define umtx_lock U_ICU_ENTRY_POINT_RENAME(umtx_lock) +#define umtx_unlock U_ICU_ENTRY_POINT_RENAME(umtx_unlock) +#define umutablecptrie_buildImmutable U_ICU_ENTRY_POINT_RENAME(umutablecptrie_buildImmutable) +#define umutablecptrie_clone U_ICU_ENTRY_POINT_RENAME(umutablecptrie_clone) +#define umutablecptrie_close U_ICU_ENTRY_POINT_RENAME(umutablecptrie_close) +#define umutablecptrie_fromUCPMap U_ICU_ENTRY_POINT_RENAME(umutablecptrie_fromUCPMap) +#define umutablecptrie_fromUCPTrie U_ICU_ENTRY_POINT_RENAME(umutablecptrie_fromUCPTrie) +#define umutablecptrie_get U_ICU_ENTRY_POINT_RENAME(umutablecptrie_get) +#define umutablecptrie_getRange U_ICU_ENTRY_POINT_RENAME(umutablecptrie_getRange) +#define umutablecptrie_open U_ICU_ENTRY_POINT_RENAME(umutablecptrie_open) +#define umutablecptrie_set U_ICU_ENTRY_POINT_RENAME(umutablecptrie_set) +#define umutablecptrie_setRange U_ICU_ENTRY_POINT_RENAME(umutablecptrie_setRange) +#define uniset_getUnicode32Instance U_ICU_ENTRY_POINT_RENAME(uniset_getUnicode32Instance) +#define unorm2_append U_ICU_ENTRY_POINT_RENAME(unorm2_append) +#define unorm2_close U_ICU_ENTRY_POINT_RENAME(unorm2_close) +#define unorm2_composePair U_ICU_ENTRY_POINT_RENAME(unorm2_composePair) +#define unorm2_getCombiningClass U_ICU_ENTRY_POINT_RENAME(unorm2_getCombiningClass) +#define unorm2_getDecomposition U_ICU_ENTRY_POINT_RENAME(unorm2_getDecomposition) +#define unorm2_getInstance U_ICU_ENTRY_POINT_RENAME(unorm2_getInstance) +#define unorm2_getNFCInstance U_ICU_ENTRY_POINT_RENAME(unorm2_getNFCInstance) +#define unorm2_getNFDInstance U_ICU_ENTRY_POINT_RENAME(unorm2_getNFDInstance) +#define unorm2_getNFKCCasefoldInstance U_ICU_ENTRY_POINT_RENAME(unorm2_getNFKCCasefoldInstance) +#define unorm2_getNFKCInstance U_ICU_ENTRY_POINT_RENAME(unorm2_getNFKCInstance) +#define unorm2_getNFKDInstance U_ICU_ENTRY_POINT_RENAME(unorm2_getNFKDInstance) +#define unorm2_getRawDecomposition U_ICU_ENTRY_POINT_RENAME(unorm2_getRawDecomposition) +#define unorm2_hasBoundaryAfter U_ICU_ENTRY_POINT_RENAME(unorm2_hasBoundaryAfter) +#define unorm2_hasBoundaryBefore U_ICU_ENTRY_POINT_RENAME(unorm2_hasBoundaryBefore) +#define unorm2_isInert U_ICU_ENTRY_POINT_RENAME(unorm2_isInert) +#define unorm2_isNormalized U_ICU_ENTRY_POINT_RENAME(unorm2_isNormalized) +#define unorm2_normalize U_ICU_ENTRY_POINT_RENAME(unorm2_normalize) +#define unorm2_normalizeSecondAndAppend U_ICU_ENTRY_POINT_RENAME(unorm2_normalizeSecondAndAppend) +#define unorm2_openFiltered U_ICU_ENTRY_POINT_RENAME(unorm2_openFiltered) +#define unorm2_quickCheck U_ICU_ENTRY_POINT_RENAME(unorm2_quickCheck) +#define unorm2_spanQuickCheckYes U_ICU_ENTRY_POINT_RENAME(unorm2_spanQuickCheckYes) +#define unorm2_swap U_ICU_ENTRY_POINT_RENAME(unorm2_swap) +#define unorm_compare U_ICU_ENTRY_POINT_RENAME(unorm_compare) +#define unorm_concatenate U_ICU_ENTRY_POINT_RENAME(unorm_concatenate) +#define unorm_getFCD16 U_ICU_ENTRY_POINT_RENAME(unorm_getFCD16) +#define unorm_getQuickCheck U_ICU_ENTRY_POINT_RENAME(unorm_getQuickCheck) +#define unorm_isNormalized U_ICU_ENTRY_POINT_RENAME(unorm_isNormalized) +#define unorm_isNormalizedWithOptions U_ICU_ENTRY_POINT_RENAME(unorm_isNormalizedWithOptions) +#define unorm_next U_ICU_ENTRY_POINT_RENAME(unorm_next) +#define unorm_normalize U_ICU_ENTRY_POINT_RENAME(unorm_normalize) +#define unorm_previous U_ICU_ENTRY_POINT_RENAME(unorm_previous) +#define unorm_quickCheck U_ICU_ENTRY_POINT_RENAME(unorm_quickCheck) +#define unorm_quickCheckWithOptions U_ICU_ENTRY_POINT_RENAME(unorm_quickCheckWithOptions) +#define unum_applyPattern U_ICU_ENTRY_POINT_RENAME(unum_applyPattern) +#define unum_clone U_ICU_ENTRY_POINT_RENAME(unum_clone) +#define unum_close U_ICU_ENTRY_POINT_RENAME(unum_close) +#define unum_countAvailable U_ICU_ENTRY_POINT_RENAME(unum_countAvailable) +#define unum_format U_ICU_ENTRY_POINT_RENAME(unum_format) +#define unum_formatDecimal U_ICU_ENTRY_POINT_RENAME(unum_formatDecimal) +#define unum_formatDouble U_ICU_ENTRY_POINT_RENAME(unum_formatDouble) +#define unum_formatDoubleCurrency U_ICU_ENTRY_POINT_RENAME(unum_formatDoubleCurrency) +#define unum_formatDoubleForFields U_ICU_ENTRY_POINT_RENAME(unum_formatDoubleForFields) +#define unum_formatInt64 U_ICU_ENTRY_POINT_RENAME(unum_formatInt64) +#define unum_formatUFormattable U_ICU_ENTRY_POINT_RENAME(unum_formatUFormattable) +#define unum_getAttribute U_ICU_ENTRY_POINT_RENAME(unum_getAttribute) +#define unum_getAvailable U_ICU_ENTRY_POINT_RENAME(unum_getAvailable) +#define unum_getContext U_ICU_ENTRY_POINT_RENAME(unum_getContext) +#define unum_getDoubleAttribute U_ICU_ENTRY_POINT_RENAME(unum_getDoubleAttribute) +#define unum_getLocaleByType U_ICU_ENTRY_POINT_RENAME(unum_getLocaleByType) +#define unum_getSymbol U_ICU_ENTRY_POINT_RENAME(unum_getSymbol) +#define unum_getTextAttribute U_ICU_ENTRY_POINT_RENAME(unum_getTextAttribute) +#define unum_hasAttribute U_ICU_ENTRY_POINT_RENAME(unum_hasAttribute) +#define unum_open U_ICU_ENTRY_POINT_RENAME(unum_open) +#define unum_parse U_ICU_ENTRY_POINT_RENAME(unum_parse) +#define unum_parseDecimal U_ICU_ENTRY_POINT_RENAME(unum_parseDecimal) +#define unum_parseDouble U_ICU_ENTRY_POINT_RENAME(unum_parseDouble) +#define unum_parseDoubleCurrency U_ICU_ENTRY_POINT_RENAME(unum_parseDoubleCurrency) +#define unum_parseInt64 U_ICU_ENTRY_POINT_RENAME(unum_parseInt64) +#define unum_parseToUFormattable U_ICU_ENTRY_POINT_RENAME(unum_parseToUFormattable) +#define unum_setAttribute U_ICU_ENTRY_POINT_RENAME(unum_setAttribute) +#define unum_setContext U_ICU_ENTRY_POINT_RENAME(unum_setContext) +#define unum_setDoubleAttribute U_ICU_ENTRY_POINT_RENAME(unum_setDoubleAttribute) +#define unum_setSymbol U_ICU_ENTRY_POINT_RENAME(unum_setSymbol) +#define unum_setTextAttribute U_ICU_ENTRY_POINT_RENAME(unum_setTextAttribute) +#define unum_toPattern U_ICU_ENTRY_POINT_RENAME(unum_toPattern) +#define unumf_close U_ICU_ENTRY_POINT_RENAME(unumf_close) +#define unumf_closeResult U_ICU_ENTRY_POINT_RENAME(unumf_closeResult) +#define unumf_formatDecimal U_ICU_ENTRY_POINT_RENAME(unumf_formatDecimal) +#define unumf_formatDouble U_ICU_ENTRY_POINT_RENAME(unumf_formatDouble) +#define unumf_formatInt U_ICU_ENTRY_POINT_RENAME(unumf_formatInt) +#define unumf_openForSkeletonAndLocale U_ICU_ENTRY_POINT_RENAME(unumf_openForSkeletonAndLocale) +#define unumf_openForSkeletonAndLocaleWithError U_ICU_ENTRY_POINT_RENAME(unumf_openForSkeletonAndLocaleWithError) +#define unumf_openResult U_ICU_ENTRY_POINT_RENAME(unumf_openResult) +#define unumf_resultAsValue U_ICU_ENTRY_POINT_RENAME(unumf_resultAsValue) +#define unumf_resultGetAllFieldPositions U_ICU_ENTRY_POINT_RENAME(unumf_resultGetAllFieldPositions) +#define unumf_resultNextFieldPosition U_ICU_ENTRY_POINT_RENAME(unumf_resultNextFieldPosition) +#define unumf_resultToDecimalNumber U_ICU_ENTRY_POINT_RENAME(unumf_resultToDecimalNumber) +#define unumf_resultToString U_ICU_ENTRY_POINT_RENAME(unumf_resultToString) +#define unumrf_close U_ICU_ENTRY_POINT_RENAME(unumrf_close) +#define unumrf_closeResult U_ICU_ENTRY_POINT_RENAME(unumrf_closeResult) +#define unumrf_formatDecimalRange U_ICU_ENTRY_POINT_RENAME(unumrf_formatDecimalRange) +#define unumrf_formatDoubleRange U_ICU_ENTRY_POINT_RENAME(unumrf_formatDoubleRange) +#define unumrf_openForSkeletonWithCollapseAndIdentityFallback U_ICU_ENTRY_POINT_RENAME(unumrf_openForSkeletonWithCollapseAndIdentityFallback) +#define unumrf_openResult U_ICU_ENTRY_POINT_RENAME(unumrf_openResult) +#define unumrf_resultAsValue U_ICU_ENTRY_POINT_RENAME(unumrf_resultAsValue) +#define unumrf_resultGetFirstDecimalNumber U_ICU_ENTRY_POINT_RENAME(unumrf_resultGetFirstDecimalNumber) +#define unumrf_resultGetIdentityResult U_ICU_ENTRY_POINT_RENAME(unumrf_resultGetIdentityResult) +#define unumrf_resultGetSecondDecimalNumber U_ICU_ENTRY_POINT_RENAME(unumrf_resultGetSecondDecimalNumber) +#define unumsys_close U_ICU_ENTRY_POINT_RENAME(unumsys_close) +#define unumsys_getDescription U_ICU_ENTRY_POINT_RENAME(unumsys_getDescription) +#define unumsys_getName U_ICU_ENTRY_POINT_RENAME(unumsys_getName) +#define unumsys_getRadix U_ICU_ENTRY_POINT_RENAME(unumsys_getRadix) +#define unumsys_isAlgorithmic U_ICU_ENTRY_POINT_RENAME(unumsys_isAlgorithmic) +#define unumsys_open U_ICU_ENTRY_POINT_RENAME(unumsys_open) +#define unumsys_openAvailableNames U_ICU_ENTRY_POINT_RENAME(unumsys_openAvailableNames) +#define unumsys_openByName U_ICU_ENTRY_POINT_RENAME(unumsys_openByName) +#define uplrules_close U_ICU_ENTRY_POINT_RENAME(uplrules_close) +#define uplrules_getKeywords U_ICU_ENTRY_POINT_RENAME(uplrules_getKeywords) +#define uplrules_open U_ICU_ENTRY_POINT_RENAME(uplrules_open) +#define uplrules_openForType U_ICU_ENTRY_POINT_RENAME(uplrules_openForType) +#define uplrules_select U_ICU_ENTRY_POINT_RENAME(uplrules_select) +#define uplrules_selectForRange U_ICU_ENTRY_POINT_RENAME(uplrules_selectForRange) +#define uplrules_selectFormatted U_ICU_ENTRY_POINT_RENAME(uplrules_selectFormatted) +#define uplrules_selectWithFormat U_ICU_ENTRY_POINT_RENAME(uplrules_selectWithFormat) +#define uplug_closeLibrary U_ICU_ENTRY_POINT_RENAME(uplug_closeLibrary) +#define uplug_findLibrary U_ICU_ENTRY_POINT_RENAME(uplug_findLibrary) +#define uplug_getConfiguration U_ICU_ENTRY_POINT_RENAME(uplug_getConfiguration) +#define uplug_getContext U_ICU_ENTRY_POINT_RENAME(uplug_getContext) +#define uplug_getCurrentLevel U_ICU_ENTRY_POINT_RENAME(uplug_getCurrentLevel) +#define uplug_getLibrary U_ICU_ENTRY_POINT_RENAME(uplug_getLibrary) +#define uplug_getLibraryName U_ICU_ENTRY_POINT_RENAME(uplug_getLibraryName) +#define uplug_getPlugInternal U_ICU_ENTRY_POINT_RENAME(uplug_getPlugInternal) +#define uplug_getPlugLevel U_ICU_ENTRY_POINT_RENAME(uplug_getPlugLevel) +#define uplug_getPlugLoadStatus U_ICU_ENTRY_POINT_RENAME(uplug_getPlugLoadStatus) +#define uplug_getPlugName U_ICU_ENTRY_POINT_RENAME(uplug_getPlugName) +#define uplug_getPluginFile U_ICU_ENTRY_POINT_RENAME(uplug_getPluginFile) +#define uplug_getSymbolName U_ICU_ENTRY_POINT_RENAME(uplug_getSymbolName) +#define uplug_init U_ICU_ENTRY_POINT_RENAME(uplug_init) +#define uplug_loadPlugFromEntrypoint U_ICU_ENTRY_POINT_RENAME(uplug_loadPlugFromEntrypoint) +#define uplug_loadPlugFromLibrary U_ICU_ENTRY_POINT_RENAME(uplug_loadPlugFromLibrary) +#define uplug_nextPlug U_ICU_ENTRY_POINT_RENAME(uplug_nextPlug) +#define uplug_openLibrary U_ICU_ENTRY_POINT_RENAME(uplug_openLibrary) +#define uplug_removePlug U_ICU_ENTRY_POINT_RENAME(uplug_removePlug) +#define uplug_setContext U_ICU_ENTRY_POINT_RENAME(uplug_setContext) +#define uplug_setPlugLevel U_ICU_ENTRY_POINT_RENAME(uplug_setPlugLevel) +#define uplug_setPlugName U_ICU_ENTRY_POINT_RENAME(uplug_setPlugName) +#define uplug_setPlugNoUnload U_ICU_ENTRY_POINT_RENAME(uplug_setPlugNoUnload) +#define uprops_addPropertyStarts U_ICU_ENTRY_POINT_RENAME(uprops_addPropertyStarts) +#define uprops_getSource U_ICU_ENTRY_POINT_RENAME(uprops_getSource) +#define upropsvec_addPropertyStarts U_ICU_ENTRY_POINT_RENAME(upropsvec_addPropertyStarts) +#define uprv_add32_overflow U_ICU_ENTRY_POINT_RENAME(uprv_add32_overflow) +#define uprv_aestrncpy U_ICU_ENTRY_POINT_RENAME(uprv_aestrncpy) +#define uprv_asciiFromEbcdic U_ICU_ENTRY_POINT_RENAME(uprv_asciiFromEbcdic) +#define uprv_asciitolower U_ICU_ENTRY_POINT_RENAME(uprv_asciitolower) +#define uprv_calloc U_ICU_ENTRY_POINT_RENAME(uprv_calloc) +#define uprv_ceil U_ICU_ENTRY_POINT_RENAME(uprv_ceil) +#define uprv_compareASCIIPropertyNames U_ICU_ENTRY_POINT_RENAME(uprv_compareASCIIPropertyNames) +#define uprv_compareEBCDICPropertyNames U_ICU_ENTRY_POINT_RENAME(uprv_compareEBCDICPropertyNames) +#define uprv_compareInvAscii U_ICU_ENTRY_POINT_RENAME(uprv_compareInvAscii) +#define uprv_compareInvEbcdic U_ICU_ENTRY_POINT_RENAME(uprv_compareInvEbcdic) +#define uprv_compareInvEbcdicAsAscii U_ICU_ENTRY_POINT_RENAME(uprv_compareInvEbcdicAsAscii) +#define uprv_convertToLCID U_ICU_ENTRY_POINT_RENAME(uprv_convertToLCID) +#define uprv_convertToLCIDPlatform U_ICU_ENTRY_POINT_RENAME(uprv_convertToLCIDPlatform) +#define uprv_convertToPosix U_ICU_ENTRY_POINT_RENAME(uprv_convertToPosix) +#define uprv_copyAscii U_ICU_ENTRY_POINT_RENAME(uprv_copyAscii) +#define uprv_copyEbcdic U_ICU_ENTRY_POINT_RENAME(uprv_copyEbcdic) +#define uprv_decContextClearStatus U_ICU_ENTRY_POINT_RENAME(uprv_decContextClearStatus) +#define uprv_decContextDefault U_ICU_ENTRY_POINT_RENAME(uprv_decContextDefault) +#define uprv_decContextGetRounding U_ICU_ENTRY_POINT_RENAME(uprv_decContextGetRounding) +#define uprv_decContextGetStatus U_ICU_ENTRY_POINT_RENAME(uprv_decContextGetStatus) +#define uprv_decContextRestoreStatus U_ICU_ENTRY_POINT_RENAME(uprv_decContextRestoreStatus) +#define uprv_decContextSaveStatus U_ICU_ENTRY_POINT_RENAME(uprv_decContextSaveStatus) +#define uprv_decContextSetRounding U_ICU_ENTRY_POINT_RENAME(uprv_decContextSetRounding) +#define uprv_decContextSetStatus U_ICU_ENTRY_POINT_RENAME(uprv_decContextSetStatus) +#define uprv_decContextSetStatusFromString U_ICU_ENTRY_POINT_RENAME(uprv_decContextSetStatusFromString) +#define uprv_decContextSetStatusFromStringQuiet U_ICU_ENTRY_POINT_RENAME(uprv_decContextSetStatusFromStringQuiet) +#define uprv_decContextSetStatusQuiet U_ICU_ENTRY_POINT_RENAME(uprv_decContextSetStatusQuiet) +#define uprv_decContextStatusToString U_ICU_ENTRY_POINT_RENAME(uprv_decContextStatusToString) +#define uprv_decContextTestSavedStatus U_ICU_ENTRY_POINT_RENAME(uprv_decContextTestSavedStatus) +#define uprv_decContextTestStatus U_ICU_ENTRY_POINT_RENAME(uprv_decContextTestStatus) +#define uprv_decContextZeroStatus U_ICU_ENTRY_POINT_RENAME(uprv_decContextZeroStatus) +#define uprv_decNumberAbs U_ICU_ENTRY_POINT_RENAME(uprv_decNumberAbs) +#define uprv_decNumberAdd U_ICU_ENTRY_POINT_RENAME(uprv_decNumberAdd) +#define uprv_decNumberAnd U_ICU_ENTRY_POINT_RENAME(uprv_decNumberAnd) +#define uprv_decNumberClassToString U_ICU_ENTRY_POINT_RENAME(uprv_decNumberClassToString) +#define uprv_decNumberCompare U_ICU_ENTRY_POINT_RENAME(uprv_decNumberCompare) +#define uprv_decNumberCompareSignal U_ICU_ENTRY_POINT_RENAME(uprv_decNumberCompareSignal) +#define uprv_decNumberCompareTotal U_ICU_ENTRY_POINT_RENAME(uprv_decNumberCompareTotal) +#define uprv_decNumberCompareTotalMag U_ICU_ENTRY_POINT_RENAME(uprv_decNumberCompareTotalMag) +#define uprv_decNumberCopy U_ICU_ENTRY_POINT_RENAME(uprv_decNumberCopy) +#define uprv_decNumberCopyAbs U_ICU_ENTRY_POINT_RENAME(uprv_decNumberCopyAbs) +#define uprv_decNumberCopyNegate U_ICU_ENTRY_POINT_RENAME(uprv_decNumberCopyNegate) +#define uprv_decNumberCopySign U_ICU_ENTRY_POINT_RENAME(uprv_decNumberCopySign) +#define uprv_decNumberDivide U_ICU_ENTRY_POINT_RENAME(uprv_decNumberDivide) +#define uprv_decNumberDivideInteger U_ICU_ENTRY_POINT_RENAME(uprv_decNumberDivideInteger) +#define uprv_decNumberExp U_ICU_ENTRY_POINT_RENAME(uprv_decNumberExp) +#define uprv_decNumberFMA U_ICU_ENTRY_POINT_RENAME(uprv_decNumberFMA) +#define uprv_decNumberFromInt32 U_ICU_ENTRY_POINT_RENAME(uprv_decNumberFromInt32) +#define uprv_decNumberFromString U_ICU_ENTRY_POINT_RENAME(uprv_decNumberFromString) +#define uprv_decNumberFromUInt32 U_ICU_ENTRY_POINT_RENAME(uprv_decNumberFromUInt32) +#define uprv_decNumberGetBCD U_ICU_ENTRY_POINT_RENAME(uprv_decNumberGetBCD) +#define uprv_decNumberInvert U_ICU_ENTRY_POINT_RENAME(uprv_decNumberInvert) +#define uprv_decNumberIsNormal U_ICU_ENTRY_POINT_RENAME(uprv_decNumberIsNormal) +#define uprv_decNumberIsSubnormal U_ICU_ENTRY_POINT_RENAME(uprv_decNumberIsSubnormal) +#define uprv_decNumberLn U_ICU_ENTRY_POINT_RENAME(uprv_decNumberLn) +#define uprv_decNumberLog10 U_ICU_ENTRY_POINT_RENAME(uprv_decNumberLog10) +#define uprv_decNumberLogB U_ICU_ENTRY_POINT_RENAME(uprv_decNumberLogB) +#define uprv_decNumberMax U_ICU_ENTRY_POINT_RENAME(uprv_decNumberMax) +#define uprv_decNumberMaxMag U_ICU_ENTRY_POINT_RENAME(uprv_decNumberMaxMag) +#define uprv_decNumberMin U_ICU_ENTRY_POINT_RENAME(uprv_decNumberMin) +#define uprv_decNumberMinMag U_ICU_ENTRY_POINT_RENAME(uprv_decNumberMinMag) +#define uprv_decNumberMinus U_ICU_ENTRY_POINT_RENAME(uprv_decNumberMinus) +#define uprv_decNumberMultiply U_ICU_ENTRY_POINT_RENAME(uprv_decNumberMultiply) +#define uprv_decNumberNextMinus U_ICU_ENTRY_POINT_RENAME(uprv_decNumberNextMinus) +#define uprv_decNumberNextPlus U_ICU_ENTRY_POINT_RENAME(uprv_decNumberNextPlus) +#define uprv_decNumberNextToward U_ICU_ENTRY_POINT_RENAME(uprv_decNumberNextToward) +#define uprv_decNumberNormalize U_ICU_ENTRY_POINT_RENAME(uprv_decNumberNormalize) +#define uprv_decNumberOr U_ICU_ENTRY_POINT_RENAME(uprv_decNumberOr) +#define uprv_decNumberPlus U_ICU_ENTRY_POINT_RENAME(uprv_decNumberPlus) +#define uprv_decNumberPower U_ICU_ENTRY_POINT_RENAME(uprv_decNumberPower) +#define uprv_decNumberQuantize U_ICU_ENTRY_POINT_RENAME(uprv_decNumberQuantize) +#define uprv_decNumberReduce U_ICU_ENTRY_POINT_RENAME(uprv_decNumberReduce) +#define uprv_decNumberRemainder U_ICU_ENTRY_POINT_RENAME(uprv_decNumberRemainder) +#define uprv_decNumberRemainderNear U_ICU_ENTRY_POINT_RENAME(uprv_decNumberRemainderNear) +#define uprv_decNumberRescale U_ICU_ENTRY_POINT_RENAME(uprv_decNumberRescale) +#define uprv_decNumberRotate U_ICU_ENTRY_POINT_RENAME(uprv_decNumberRotate) +#define uprv_decNumberSameQuantum U_ICU_ENTRY_POINT_RENAME(uprv_decNumberSameQuantum) +#define uprv_decNumberScaleB U_ICU_ENTRY_POINT_RENAME(uprv_decNumberScaleB) +#define uprv_decNumberSetBCD U_ICU_ENTRY_POINT_RENAME(uprv_decNumberSetBCD) +#define uprv_decNumberShift U_ICU_ENTRY_POINT_RENAME(uprv_decNumberShift) +#define uprv_decNumberSquareRoot U_ICU_ENTRY_POINT_RENAME(uprv_decNumberSquareRoot) +#define uprv_decNumberSubtract U_ICU_ENTRY_POINT_RENAME(uprv_decNumberSubtract) +#define uprv_decNumberToEngString U_ICU_ENTRY_POINT_RENAME(uprv_decNumberToEngString) +#define uprv_decNumberToInt32 U_ICU_ENTRY_POINT_RENAME(uprv_decNumberToInt32) +#define uprv_decNumberToIntegralExact U_ICU_ENTRY_POINT_RENAME(uprv_decNumberToIntegralExact) +#define uprv_decNumberToIntegralValue U_ICU_ENTRY_POINT_RENAME(uprv_decNumberToIntegralValue) +#define uprv_decNumberToString U_ICU_ENTRY_POINT_RENAME(uprv_decNumberToString) +#define uprv_decNumberToUInt32 U_ICU_ENTRY_POINT_RENAME(uprv_decNumberToUInt32) +#define uprv_decNumberTrim U_ICU_ENTRY_POINT_RENAME(uprv_decNumberTrim) +#define uprv_decNumberVersion U_ICU_ENTRY_POINT_RENAME(uprv_decNumberVersion) +#define uprv_decNumberXor U_ICU_ENTRY_POINT_RENAME(uprv_decNumberXor) +#define uprv_decNumberZero U_ICU_ENTRY_POINT_RENAME(uprv_decNumberZero) +#define uprv_deleteConditionalCE32 U_ICU_ENTRY_POINT_RENAME(uprv_deleteConditionalCE32) +#define uprv_deleteUObject U_ICU_ENTRY_POINT_RENAME(uprv_deleteUObject) +#define uprv_dl_close U_ICU_ENTRY_POINT_RENAME(uprv_dl_close) +#define uprv_dl_open U_ICU_ENTRY_POINT_RENAME(uprv_dl_open) +#define uprv_dlsym_func U_ICU_ENTRY_POINT_RENAME(uprv_dlsym_func) +#define uprv_eastrncpy U_ICU_ENTRY_POINT_RENAME(uprv_eastrncpy) +#define uprv_ebcdicFromAscii U_ICU_ENTRY_POINT_RENAME(uprv_ebcdicFromAscii) +#define uprv_ebcdicToAscii U_ICU_ENTRY_POINT_RENAME(uprv_ebcdicToAscii) +#define uprv_ebcdicToLowercaseAscii U_ICU_ENTRY_POINT_RENAME(uprv_ebcdicToLowercaseAscii) +#define uprv_ebcdictolower U_ICU_ENTRY_POINT_RENAME(uprv_ebcdictolower) +#define uprv_fabs U_ICU_ENTRY_POINT_RENAME(uprv_fabs) +#define uprv_floor U_ICU_ENTRY_POINT_RENAME(uprv_floor) +#define uprv_fmax U_ICU_ENTRY_POINT_RENAME(uprv_fmax) +#define uprv_fmin U_ICU_ENTRY_POINT_RENAME(uprv_fmin) +#define uprv_fmod U_ICU_ENTRY_POINT_RENAME(uprv_fmod) +#define uprv_free U_ICU_ENTRY_POINT_RENAME(uprv_free) +#define uprv_getCharNameCharacters U_ICU_ENTRY_POINT_RENAME(uprv_getCharNameCharacters) +#define uprv_getDefaultLocaleID U_ICU_ENTRY_POINT_RENAME(uprv_getDefaultLocaleID) +#define uprv_getInfinity U_ICU_ENTRY_POINT_RENAME(uprv_getInfinity) +#define uprv_getMaxCharNameLength U_ICU_ENTRY_POINT_RENAME(uprv_getMaxCharNameLength) +#define uprv_getMaxValues U_ICU_ENTRY_POINT_RENAME(uprv_getMaxValues) +#define uprv_getNaN U_ICU_ENTRY_POINT_RENAME(uprv_getNaN) +#define uprv_getRawUTCtime U_ICU_ENTRY_POINT_RENAME(uprv_getRawUTCtime) +#define uprv_getStaticCurrencyName U_ICU_ENTRY_POINT_RENAME(uprv_getStaticCurrencyName) +#define uprv_getUTCtime U_ICU_ENTRY_POINT_RENAME(uprv_getUTCtime) +#define uprv_int32Comparator U_ICU_ENTRY_POINT_RENAME(uprv_int32Comparator) +#define uprv_isASCIILetter U_ICU_ENTRY_POINT_RENAME(uprv_isASCIILetter) +#define uprv_isEbcdicAtSign U_ICU_ENTRY_POINT_RENAME(uprv_isEbcdicAtSign) +#define uprv_isInfinite U_ICU_ENTRY_POINT_RENAME(uprv_isInfinite) +#define uprv_isInvariantString U_ICU_ENTRY_POINT_RENAME(uprv_isInvariantString) +#define uprv_isInvariantUString U_ICU_ENTRY_POINT_RENAME(uprv_isInvariantUString) +#define uprv_isNaN U_ICU_ENTRY_POINT_RENAME(uprv_isNaN) +#define uprv_isNegativeInfinity U_ICU_ENTRY_POINT_RENAME(uprv_isNegativeInfinity) +#define uprv_isPositiveInfinity U_ICU_ENTRY_POINT_RENAME(uprv_isPositiveInfinity) +#define uprv_itou U_ICU_ENTRY_POINT_RENAME(uprv_itou) +#define uprv_log U_ICU_ENTRY_POINT_RENAME(uprv_log) +#define uprv_malloc U_ICU_ENTRY_POINT_RENAME(uprv_malloc) +#define uprv_mapFile U_ICU_ENTRY_POINT_RENAME(uprv_mapFile) +#define uprv_max U_ICU_ENTRY_POINT_RENAME(uprv_max) +#define uprv_maxMantissa U_ICU_ENTRY_POINT_RENAME(uprv_maxMantissa) +#define uprv_maximumPtr U_ICU_ENTRY_POINT_RENAME(uprv_maximumPtr) +#define uprv_min U_ICU_ENTRY_POINT_RENAME(uprv_min) +#define uprv_modf U_ICU_ENTRY_POINT_RENAME(uprv_modf) +#define uprv_mul32_overflow U_ICU_ENTRY_POINT_RENAME(uprv_mul32_overflow) +#define uprv_parseCurrency U_ICU_ENTRY_POINT_RENAME(uprv_parseCurrency) +#define uprv_pathIsAbsolute U_ICU_ENTRY_POINT_RENAME(uprv_pathIsAbsolute) +#define uprv_pow U_ICU_ENTRY_POINT_RENAME(uprv_pow) +#define uprv_pow10 U_ICU_ENTRY_POINT_RENAME(uprv_pow10) +#define uprv_realloc U_ICU_ENTRY_POINT_RENAME(uprv_realloc) +#define uprv_round U_ICU_ENTRY_POINT_RENAME(uprv_round) +#define uprv_sortArray U_ICU_ENTRY_POINT_RENAME(uprv_sortArray) +#define uprv_stableBinarySearch U_ICU_ENTRY_POINT_RENAME(uprv_stableBinarySearch) +#define uprv_strCompare U_ICU_ENTRY_POINT_RENAME(uprv_strCompare) +#define uprv_strdup U_ICU_ENTRY_POINT_RENAME(uprv_strdup) +#define uprv_stricmp U_ICU_ENTRY_POINT_RENAME(uprv_stricmp) +#define uprv_strndup U_ICU_ENTRY_POINT_RENAME(uprv_strndup) +#define uprv_strnicmp U_ICU_ENTRY_POINT_RENAME(uprv_strnicmp) +#define uprv_syntaxError U_ICU_ENTRY_POINT_RENAME(uprv_syntaxError) +#define uprv_timezone U_ICU_ENTRY_POINT_RENAME(uprv_timezone) +#define uprv_toupper U_ICU_ENTRY_POINT_RENAME(uprv_toupper) +#define uprv_trunc U_ICU_ENTRY_POINT_RENAME(uprv_trunc) +#define uprv_tzname U_ICU_ENTRY_POINT_RENAME(uprv_tzname) +#define uprv_tzname_clear_cache U_ICU_ENTRY_POINT_RENAME(uprv_tzname_clear_cache) +#define uprv_tzset U_ICU_ENTRY_POINT_RENAME(uprv_tzset) +#define uprv_uint16Comparator U_ICU_ENTRY_POINT_RENAME(uprv_uint16Comparator) +#define uprv_uint32Comparator U_ICU_ENTRY_POINT_RENAME(uprv_uint32Comparator) +#define uprv_unmapFile U_ICU_ENTRY_POINT_RENAME(uprv_unmapFile) +#define upvec_cloneArray U_ICU_ENTRY_POINT_RENAME(upvec_cloneArray) +#define upvec_close U_ICU_ENTRY_POINT_RENAME(upvec_close) +#define upvec_compact U_ICU_ENTRY_POINT_RENAME(upvec_compact) +#define upvec_compactToUTrie2Handler U_ICU_ENTRY_POINT_RENAME(upvec_compactToUTrie2Handler) +#define upvec_compactToUTrie2WithRowIndexes U_ICU_ENTRY_POINT_RENAME(upvec_compactToUTrie2WithRowIndexes) +#define upvec_getArray U_ICU_ENTRY_POINT_RENAME(upvec_getArray) +#define upvec_getRow U_ICU_ENTRY_POINT_RENAME(upvec_getRow) +#define upvec_getValue U_ICU_ENTRY_POINT_RENAME(upvec_getValue) +#define upvec_open U_ICU_ENTRY_POINT_RENAME(upvec_open) +#define upvec_setValue U_ICU_ENTRY_POINT_RENAME(upvec_setValue) +#define uregex_appendReplacement U_ICU_ENTRY_POINT_RENAME(uregex_appendReplacement) +#define uregex_appendReplacementUText U_ICU_ENTRY_POINT_RENAME(uregex_appendReplacementUText) +#define uregex_appendTail U_ICU_ENTRY_POINT_RENAME(uregex_appendTail) +#define uregex_appendTailUText U_ICU_ENTRY_POINT_RENAME(uregex_appendTailUText) +#define uregex_clone U_ICU_ENTRY_POINT_RENAME(uregex_clone) +#define uregex_close U_ICU_ENTRY_POINT_RENAME(uregex_close) +#define uregex_end U_ICU_ENTRY_POINT_RENAME(uregex_end) +#define uregex_end64 U_ICU_ENTRY_POINT_RENAME(uregex_end64) +#define uregex_find U_ICU_ENTRY_POINT_RENAME(uregex_find) +#define uregex_find64 U_ICU_ENTRY_POINT_RENAME(uregex_find64) +#define uregex_findNext U_ICU_ENTRY_POINT_RENAME(uregex_findNext) +#define uregex_flags U_ICU_ENTRY_POINT_RENAME(uregex_flags) +#define uregex_getFindProgressCallback U_ICU_ENTRY_POINT_RENAME(uregex_getFindProgressCallback) +#define uregex_getMatchCallback U_ICU_ENTRY_POINT_RENAME(uregex_getMatchCallback) +#define uregex_getStackLimit U_ICU_ENTRY_POINT_RENAME(uregex_getStackLimit) +#define uregex_getText U_ICU_ENTRY_POINT_RENAME(uregex_getText) +#define uregex_getTimeLimit U_ICU_ENTRY_POINT_RENAME(uregex_getTimeLimit) +#define uregex_getUText U_ICU_ENTRY_POINT_RENAME(uregex_getUText) +#define uregex_group U_ICU_ENTRY_POINT_RENAME(uregex_group) +#define uregex_groupCount U_ICU_ENTRY_POINT_RENAME(uregex_groupCount) +#define uregex_groupNumberFromCName U_ICU_ENTRY_POINT_RENAME(uregex_groupNumberFromCName) +#define uregex_groupNumberFromName U_ICU_ENTRY_POINT_RENAME(uregex_groupNumberFromName) +#define uregex_groupUText U_ICU_ENTRY_POINT_RENAME(uregex_groupUText) +#define uregex_hasAnchoringBounds U_ICU_ENTRY_POINT_RENAME(uregex_hasAnchoringBounds) +#define uregex_hasTransparentBounds U_ICU_ENTRY_POINT_RENAME(uregex_hasTransparentBounds) +#define uregex_hitEnd U_ICU_ENTRY_POINT_RENAME(uregex_hitEnd) +#define uregex_lookingAt U_ICU_ENTRY_POINT_RENAME(uregex_lookingAt) +#define uregex_lookingAt64 U_ICU_ENTRY_POINT_RENAME(uregex_lookingAt64) +#define uregex_matches U_ICU_ENTRY_POINT_RENAME(uregex_matches) +#define uregex_matches64 U_ICU_ENTRY_POINT_RENAME(uregex_matches64) +#define uregex_open U_ICU_ENTRY_POINT_RENAME(uregex_open) +#define uregex_openC U_ICU_ENTRY_POINT_RENAME(uregex_openC) +#define uregex_openUText U_ICU_ENTRY_POINT_RENAME(uregex_openUText) +#define uregex_pattern U_ICU_ENTRY_POINT_RENAME(uregex_pattern) +#define uregex_patternUText U_ICU_ENTRY_POINT_RENAME(uregex_patternUText) +#define uregex_refreshUText U_ICU_ENTRY_POINT_RENAME(uregex_refreshUText) +#define uregex_regionEnd U_ICU_ENTRY_POINT_RENAME(uregex_regionEnd) +#define uregex_regionEnd64 U_ICU_ENTRY_POINT_RENAME(uregex_regionEnd64) +#define uregex_regionStart U_ICU_ENTRY_POINT_RENAME(uregex_regionStart) +#define uregex_regionStart64 U_ICU_ENTRY_POINT_RENAME(uregex_regionStart64) +#define uregex_replaceAll U_ICU_ENTRY_POINT_RENAME(uregex_replaceAll) +#define uregex_replaceAllUText U_ICU_ENTRY_POINT_RENAME(uregex_replaceAllUText) +#define uregex_replaceFirst U_ICU_ENTRY_POINT_RENAME(uregex_replaceFirst) +#define uregex_replaceFirstUText U_ICU_ENTRY_POINT_RENAME(uregex_replaceFirstUText) +#define uregex_requireEnd U_ICU_ENTRY_POINT_RENAME(uregex_requireEnd) +#define uregex_reset U_ICU_ENTRY_POINT_RENAME(uregex_reset) +#define uregex_reset64 U_ICU_ENTRY_POINT_RENAME(uregex_reset64) +#define uregex_setFindProgressCallback U_ICU_ENTRY_POINT_RENAME(uregex_setFindProgressCallback) +#define uregex_setMatchCallback U_ICU_ENTRY_POINT_RENAME(uregex_setMatchCallback) +#define uregex_setRegion U_ICU_ENTRY_POINT_RENAME(uregex_setRegion) +#define uregex_setRegion64 U_ICU_ENTRY_POINT_RENAME(uregex_setRegion64) +#define uregex_setRegionAndStart U_ICU_ENTRY_POINT_RENAME(uregex_setRegionAndStart) +#define uregex_setStackLimit U_ICU_ENTRY_POINT_RENAME(uregex_setStackLimit) +#define uregex_setText U_ICU_ENTRY_POINT_RENAME(uregex_setText) +#define uregex_setTimeLimit U_ICU_ENTRY_POINT_RENAME(uregex_setTimeLimit) +#define uregex_setUText U_ICU_ENTRY_POINT_RENAME(uregex_setUText) +#define uregex_split U_ICU_ENTRY_POINT_RENAME(uregex_split) +#define uregex_splitUText U_ICU_ENTRY_POINT_RENAME(uregex_splitUText) +#define uregex_start U_ICU_ENTRY_POINT_RENAME(uregex_start) +#define uregex_start64 U_ICU_ENTRY_POINT_RENAME(uregex_start64) +#define uregex_ucstr_unescape_charAt U_ICU_ENTRY_POINT_RENAME(uregex_ucstr_unescape_charAt) +#define uregex_useAnchoringBounds U_ICU_ENTRY_POINT_RENAME(uregex_useAnchoringBounds) +#define uregex_useTransparentBounds U_ICU_ENTRY_POINT_RENAME(uregex_useTransparentBounds) +#define uregex_utext_unescape_charAt U_ICU_ENTRY_POINT_RENAME(uregex_utext_unescape_charAt) +#define uregion_areEqual U_ICU_ENTRY_POINT_RENAME(uregion_areEqual) +#define uregion_contains U_ICU_ENTRY_POINT_RENAME(uregion_contains) +#define uregion_getAvailable U_ICU_ENTRY_POINT_RENAME(uregion_getAvailable) +#define uregion_getContainedRegions U_ICU_ENTRY_POINT_RENAME(uregion_getContainedRegions) +#define uregion_getContainedRegionsOfType U_ICU_ENTRY_POINT_RENAME(uregion_getContainedRegionsOfType) +#define uregion_getContainingRegion U_ICU_ENTRY_POINT_RENAME(uregion_getContainingRegion) +#define uregion_getContainingRegionOfType U_ICU_ENTRY_POINT_RENAME(uregion_getContainingRegionOfType) +#define uregion_getNumericCode U_ICU_ENTRY_POINT_RENAME(uregion_getNumericCode) +#define uregion_getPreferredValues U_ICU_ENTRY_POINT_RENAME(uregion_getPreferredValues) +#define uregion_getRegionCode U_ICU_ENTRY_POINT_RENAME(uregion_getRegionCode) +#define uregion_getRegionFromCode U_ICU_ENTRY_POINT_RENAME(uregion_getRegionFromCode) +#define uregion_getRegionFromNumericCode U_ICU_ENTRY_POINT_RENAME(uregion_getRegionFromNumericCode) +#define uregion_getType U_ICU_ENTRY_POINT_RENAME(uregion_getType) +#define ureldatefmt_close U_ICU_ENTRY_POINT_RENAME(ureldatefmt_close) +#define ureldatefmt_closeResult U_ICU_ENTRY_POINT_RENAME(ureldatefmt_closeResult) +#define ureldatefmt_combineDateAndTime U_ICU_ENTRY_POINT_RENAME(ureldatefmt_combineDateAndTime) +#define ureldatefmt_format U_ICU_ENTRY_POINT_RENAME(ureldatefmt_format) +#define ureldatefmt_formatNumeric U_ICU_ENTRY_POINT_RENAME(ureldatefmt_formatNumeric) +#define ureldatefmt_formatNumericToResult U_ICU_ENTRY_POINT_RENAME(ureldatefmt_formatNumericToResult) +#define ureldatefmt_formatToResult U_ICU_ENTRY_POINT_RENAME(ureldatefmt_formatToResult) +#define ureldatefmt_open U_ICU_ENTRY_POINT_RENAME(ureldatefmt_open) +#define ureldatefmt_openResult U_ICU_ENTRY_POINT_RENAME(ureldatefmt_openResult) +#define ureldatefmt_resultAsValue U_ICU_ENTRY_POINT_RENAME(ureldatefmt_resultAsValue) +#define ures_close U_ICU_ENTRY_POINT_RENAME(ures_close) +#define ures_copyResb U_ICU_ENTRY_POINT_RENAME(ures_copyResb) +#define ures_countArrayItems U_ICU_ENTRY_POINT_RENAME(ures_countArrayItems) +#define ures_findResource U_ICU_ENTRY_POINT_RENAME(ures_findResource) +#define ures_findSubResource U_ICU_ENTRY_POINT_RENAME(ures_findSubResource) +#define ures_getAllChildrenWithFallback U_ICU_ENTRY_POINT_RENAME(ures_getAllChildrenWithFallback) +#define ures_getAllItemsWithFallback U_ICU_ENTRY_POINT_RENAME(ures_getAllItemsWithFallback) +#define ures_getBinary U_ICU_ENTRY_POINT_RENAME(ures_getBinary) +#define ures_getByIndex U_ICU_ENTRY_POINT_RENAME(ures_getByIndex) +#define ures_getByKey U_ICU_ENTRY_POINT_RENAME(ures_getByKey) +#define ures_getByKeyWithFallback U_ICU_ENTRY_POINT_RENAME(ures_getByKeyWithFallback) +#define ures_getFunctionalEquivalent U_ICU_ENTRY_POINT_RENAME(ures_getFunctionalEquivalent) +#define ures_getInt U_ICU_ENTRY_POINT_RENAME(ures_getInt) +#define ures_getIntVector U_ICU_ENTRY_POINT_RENAME(ures_getIntVector) +#define ures_getKey U_ICU_ENTRY_POINT_RENAME(ures_getKey) +#define ures_getKeywordValues U_ICU_ENTRY_POINT_RENAME(ures_getKeywordValues) +#define ures_getLocale U_ICU_ENTRY_POINT_RENAME(ures_getLocale) +#define ures_getLocaleByType U_ICU_ENTRY_POINT_RENAME(ures_getLocaleByType) +#define ures_getLocaleInternal U_ICU_ENTRY_POINT_RENAME(ures_getLocaleInternal) +#define ures_getName U_ICU_ENTRY_POINT_RENAME(ures_getName) +#define ures_getNextResource U_ICU_ENTRY_POINT_RENAME(ures_getNextResource) +#define ures_getNextString U_ICU_ENTRY_POINT_RENAME(ures_getNextString) +#define ures_getSize U_ICU_ENTRY_POINT_RENAME(ures_getSize) +#define ures_getString U_ICU_ENTRY_POINT_RENAME(ures_getString) +#define ures_getStringByIndex U_ICU_ENTRY_POINT_RENAME(ures_getStringByIndex) +#define ures_getStringByKey U_ICU_ENTRY_POINT_RENAME(ures_getStringByKey) +#define ures_getStringByKeyWithFallback U_ICU_ENTRY_POINT_RENAME(ures_getStringByKeyWithFallback) +#define ures_getType U_ICU_ENTRY_POINT_RENAME(ures_getType) +#define ures_getUInt U_ICU_ENTRY_POINT_RENAME(ures_getUInt) +#define ures_getUTF8String U_ICU_ENTRY_POINT_RENAME(ures_getUTF8String) +#define ures_getUTF8StringByIndex U_ICU_ENTRY_POINT_RENAME(ures_getUTF8StringByIndex) +#define ures_getUTF8StringByKey U_ICU_ENTRY_POINT_RENAME(ures_getUTF8StringByKey) +#define ures_getValueWithFallback U_ICU_ENTRY_POINT_RENAME(ures_getValueWithFallback) +#define ures_getVersion U_ICU_ENTRY_POINT_RENAME(ures_getVersion) +#define ures_getVersionByKey U_ICU_ENTRY_POINT_RENAME(ures_getVersionByKey) +#define ures_getVersionNumber U_ICU_ENTRY_POINT_RENAME(ures_getVersionNumber) +#define ures_getVersionNumberInternal U_ICU_ENTRY_POINT_RENAME(ures_getVersionNumberInternal) +#define ures_hasNext U_ICU_ENTRY_POINT_RENAME(ures_hasNext) +#define ures_initStackObject U_ICU_ENTRY_POINT_RENAME(ures_initStackObject) +#define ures_open U_ICU_ENTRY_POINT_RENAME(ures_open) +#define ures_openAvailableLocales U_ICU_ENTRY_POINT_RENAME(ures_openAvailableLocales) +#define ures_openDirect U_ICU_ENTRY_POINT_RENAME(ures_openDirect) +#define ures_openDirectFillIn U_ICU_ENTRY_POINT_RENAME(ures_openDirectFillIn) +#define ures_openFillIn U_ICU_ENTRY_POINT_RENAME(ures_openFillIn) +#define ures_openNoDefault U_ICU_ENTRY_POINT_RENAME(ures_openNoDefault) +#define ures_openU U_ICU_ENTRY_POINT_RENAME(ures_openU) +#define ures_resetIterator U_ICU_ENTRY_POINT_RENAME(ures_resetIterator) +#define ures_swap U_ICU_ENTRY_POINT_RENAME(ures_swap) +#define uscript_breaksBetweenLetters U_ICU_ENTRY_POINT_RENAME(uscript_breaksBetweenLetters) +#define uscript_closeRun U_ICU_ENTRY_POINT_RENAME(uscript_closeRun) +#define uscript_getCode U_ICU_ENTRY_POINT_RENAME(uscript_getCode) +#define uscript_getName U_ICU_ENTRY_POINT_RENAME(uscript_getName) +#define uscript_getSampleString U_ICU_ENTRY_POINT_RENAME(uscript_getSampleString) +#define uscript_getSampleUnicodeString U_ICU_ENTRY_POINT_RENAME(uscript_getSampleUnicodeString) +#define uscript_getScript U_ICU_ENTRY_POINT_RENAME(uscript_getScript) +#define uscript_getScriptExtensions U_ICU_ENTRY_POINT_RENAME(uscript_getScriptExtensions) +#define uscript_getShortName U_ICU_ENTRY_POINT_RENAME(uscript_getShortName) +#define uscript_getUsage U_ICU_ENTRY_POINT_RENAME(uscript_getUsage) +#define uscript_hasScript U_ICU_ENTRY_POINT_RENAME(uscript_hasScript) +#define uscript_isCased U_ICU_ENTRY_POINT_RENAME(uscript_isCased) +#define uscript_isRightToLeft U_ICU_ENTRY_POINT_RENAME(uscript_isRightToLeft) +#define uscript_nextRun U_ICU_ENTRY_POINT_RENAME(uscript_nextRun) +#define uscript_openRun U_ICU_ENTRY_POINT_RENAME(uscript_openRun) +#define uscript_resetRun U_ICU_ENTRY_POINT_RENAME(uscript_resetRun) +#define uscript_setRunText U_ICU_ENTRY_POINT_RENAME(uscript_setRunText) +#define usearch_close U_ICU_ENTRY_POINT_RENAME(usearch_close) +#define usearch_first U_ICU_ENTRY_POINT_RENAME(usearch_first) +#define usearch_following U_ICU_ENTRY_POINT_RENAME(usearch_following) +#define usearch_getAttribute U_ICU_ENTRY_POINT_RENAME(usearch_getAttribute) +#define usearch_getBreakIterator U_ICU_ENTRY_POINT_RENAME(usearch_getBreakIterator) +#define usearch_getCollator U_ICU_ENTRY_POINT_RENAME(usearch_getCollator) +#define usearch_getMatchedLength U_ICU_ENTRY_POINT_RENAME(usearch_getMatchedLength) +#define usearch_getMatchedStart U_ICU_ENTRY_POINT_RENAME(usearch_getMatchedStart) +#define usearch_getMatchedText U_ICU_ENTRY_POINT_RENAME(usearch_getMatchedText) +#define usearch_getOffset U_ICU_ENTRY_POINT_RENAME(usearch_getOffset) +#define usearch_getPattern U_ICU_ENTRY_POINT_RENAME(usearch_getPattern) +#define usearch_getText U_ICU_ENTRY_POINT_RENAME(usearch_getText) +#define usearch_handleNextCanonical U_ICU_ENTRY_POINT_RENAME(usearch_handleNextCanonical) +#define usearch_handleNextExact U_ICU_ENTRY_POINT_RENAME(usearch_handleNextExact) +#define usearch_handlePreviousCanonical U_ICU_ENTRY_POINT_RENAME(usearch_handlePreviousCanonical) +#define usearch_handlePreviousExact U_ICU_ENTRY_POINT_RENAME(usearch_handlePreviousExact) +#define usearch_last U_ICU_ENTRY_POINT_RENAME(usearch_last) +#define usearch_next U_ICU_ENTRY_POINT_RENAME(usearch_next) +#define usearch_open U_ICU_ENTRY_POINT_RENAME(usearch_open) +#define usearch_openFromCollator U_ICU_ENTRY_POINT_RENAME(usearch_openFromCollator) +#define usearch_preceding U_ICU_ENTRY_POINT_RENAME(usearch_preceding) +#define usearch_previous U_ICU_ENTRY_POINT_RENAME(usearch_previous) +#define usearch_reset U_ICU_ENTRY_POINT_RENAME(usearch_reset) +#define usearch_search U_ICU_ENTRY_POINT_RENAME(usearch_search) +#define usearch_searchBackwards U_ICU_ENTRY_POINT_RENAME(usearch_searchBackwards) +#define usearch_setAttribute U_ICU_ENTRY_POINT_RENAME(usearch_setAttribute) +#define usearch_setBreakIterator U_ICU_ENTRY_POINT_RENAME(usearch_setBreakIterator) +#define usearch_setCollator U_ICU_ENTRY_POINT_RENAME(usearch_setCollator) +#define usearch_setOffset U_ICU_ENTRY_POINT_RENAME(usearch_setOffset) +#define usearch_setPattern U_ICU_ENTRY_POINT_RENAME(usearch_setPattern) +#define usearch_setText U_ICU_ENTRY_POINT_RENAME(usearch_setText) +#define uset_add U_ICU_ENTRY_POINT_RENAME(uset_add) +#define uset_addAll U_ICU_ENTRY_POINT_RENAME(uset_addAll) +#define uset_addAllCodePoints U_ICU_ENTRY_POINT_RENAME(uset_addAllCodePoints) +#define uset_addRange U_ICU_ENTRY_POINT_RENAME(uset_addRange) +#define uset_addString U_ICU_ENTRY_POINT_RENAME(uset_addString) +#define uset_applyIntPropertyValue U_ICU_ENTRY_POINT_RENAME(uset_applyIntPropertyValue) +#define uset_applyPattern U_ICU_ENTRY_POINT_RENAME(uset_applyPattern) +#define uset_applyPropertyAlias U_ICU_ENTRY_POINT_RENAME(uset_applyPropertyAlias) +#define uset_charAt U_ICU_ENTRY_POINT_RENAME(uset_charAt) +#define uset_clear U_ICU_ENTRY_POINT_RENAME(uset_clear) +#define uset_clone U_ICU_ENTRY_POINT_RENAME(uset_clone) +#define uset_cloneAsThawed U_ICU_ENTRY_POINT_RENAME(uset_cloneAsThawed) +#define uset_close U_ICU_ENTRY_POINT_RENAME(uset_close) +#define uset_closeOver U_ICU_ENTRY_POINT_RENAME(uset_closeOver) +#define uset_compact U_ICU_ENTRY_POINT_RENAME(uset_compact) +#define uset_complement U_ICU_ENTRY_POINT_RENAME(uset_complement) +#define uset_complementAll U_ICU_ENTRY_POINT_RENAME(uset_complementAll) +#define uset_complementAllCodePoints U_ICU_ENTRY_POINT_RENAME(uset_complementAllCodePoints) +#define uset_complementRange U_ICU_ENTRY_POINT_RENAME(uset_complementRange) +#define uset_complementString U_ICU_ENTRY_POINT_RENAME(uset_complementString) +#define uset_contains U_ICU_ENTRY_POINT_RENAME(uset_contains) +#define uset_containsAll U_ICU_ENTRY_POINT_RENAME(uset_containsAll) +#define uset_containsAllCodePoints U_ICU_ENTRY_POINT_RENAME(uset_containsAllCodePoints) +#define uset_containsNone U_ICU_ENTRY_POINT_RENAME(uset_containsNone) +#define uset_containsRange U_ICU_ENTRY_POINT_RENAME(uset_containsRange) +#define uset_containsSome U_ICU_ENTRY_POINT_RENAME(uset_containsSome) +#define uset_containsString U_ICU_ENTRY_POINT_RENAME(uset_containsString) +#define uset_equals U_ICU_ENTRY_POINT_RENAME(uset_equals) +#define uset_freeze U_ICU_ENTRY_POINT_RENAME(uset_freeze) +#define uset_getItem U_ICU_ENTRY_POINT_RENAME(uset_getItem) +#define uset_getItemCount U_ICU_ENTRY_POINT_RENAME(uset_getItemCount) +#define uset_getRangeCount U_ICU_ENTRY_POINT_RENAME(uset_getRangeCount) +#define uset_getSerializedRange U_ICU_ENTRY_POINT_RENAME(uset_getSerializedRange) +#define uset_getSerializedRangeCount U_ICU_ENTRY_POINT_RENAME(uset_getSerializedRangeCount) +#define uset_getSerializedSet U_ICU_ENTRY_POINT_RENAME(uset_getSerializedSet) +#define uset_hasStrings U_ICU_ENTRY_POINT_RENAME(uset_hasStrings) +#define uset_indexOf U_ICU_ENTRY_POINT_RENAME(uset_indexOf) +#define uset_isEmpty U_ICU_ENTRY_POINT_RENAME(uset_isEmpty) +#define uset_isFrozen U_ICU_ENTRY_POINT_RENAME(uset_isFrozen) +#define uset_open U_ICU_ENTRY_POINT_RENAME(uset_open) +#define uset_openEmpty U_ICU_ENTRY_POINT_RENAME(uset_openEmpty) +#define uset_openPattern U_ICU_ENTRY_POINT_RENAME(uset_openPattern) +#define uset_openPatternOptions U_ICU_ENTRY_POINT_RENAME(uset_openPatternOptions) +#define uset_remove U_ICU_ENTRY_POINT_RENAME(uset_remove) +#define uset_removeAll U_ICU_ENTRY_POINT_RENAME(uset_removeAll) +#define uset_removeAllCodePoints U_ICU_ENTRY_POINT_RENAME(uset_removeAllCodePoints) +#define uset_removeAllStrings U_ICU_ENTRY_POINT_RENAME(uset_removeAllStrings) +#define uset_removeRange U_ICU_ENTRY_POINT_RENAME(uset_removeRange) +#define uset_removeString U_ICU_ENTRY_POINT_RENAME(uset_removeString) +#define uset_resemblesPattern U_ICU_ENTRY_POINT_RENAME(uset_resemblesPattern) +#define uset_retain U_ICU_ENTRY_POINT_RENAME(uset_retain) +#define uset_retainAll U_ICU_ENTRY_POINT_RENAME(uset_retainAll) +#define uset_retainAllCodePoints U_ICU_ENTRY_POINT_RENAME(uset_retainAllCodePoints) +#define uset_retainString U_ICU_ENTRY_POINT_RENAME(uset_retainString) +#define uset_serialize U_ICU_ENTRY_POINT_RENAME(uset_serialize) +#define uset_serializedContains U_ICU_ENTRY_POINT_RENAME(uset_serializedContains) +#define uset_set U_ICU_ENTRY_POINT_RENAME(uset_set) +#define uset_setSerializedToOne U_ICU_ENTRY_POINT_RENAME(uset_setSerializedToOne) +#define uset_size U_ICU_ENTRY_POINT_RENAME(uset_size) +#define uset_span U_ICU_ENTRY_POINT_RENAME(uset_span) +#define uset_spanBack U_ICU_ENTRY_POINT_RENAME(uset_spanBack) +#define uset_spanBackUTF8 U_ICU_ENTRY_POINT_RENAME(uset_spanBackUTF8) +#define uset_spanUTF8 U_ICU_ENTRY_POINT_RENAME(uset_spanUTF8) +#define uset_toPattern U_ICU_ENTRY_POINT_RENAME(uset_toPattern) +#define usnum_close U_ICU_ENTRY_POINT_RENAME(usnum_close) +#define usnum_multiplyByPowerOfTen U_ICU_ENTRY_POINT_RENAME(usnum_multiplyByPowerOfTen) +#define usnum_openForInt64 U_ICU_ENTRY_POINT_RENAME(usnum_openForInt64) +#define usnum_roundTo U_ICU_ENTRY_POINT_RENAME(usnum_roundTo) +#define usnum_setMinimumFractionDigits U_ICU_ENTRY_POINT_RENAME(usnum_setMinimumFractionDigits) +#define usnum_setMinimumIntegerDigits U_ICU_ENTRY_POINT_RENAME(usnum_setMinimumIntegerDigits) +#define usnum_setSign U_ICU_ENTRY_POINT_RENAME(usnum_setSign) +#define usnum_setToInt64 U_ICU_ENTRY_POINT_RENAME(usnum_setToInt64) +#define usnum_truncateStart U_ICU_ENTRY_POINT_RENAME(usnum_truncateStart) +#define usnumf_close U_ICU_ENTRY_POINT_RENAME(usnumf_close) +#define usnumf_format U_ICU_ENTRY_POINT_RENAME(usnumf_format) +#define usnumf_formatInt64 U_ICU_ENTRY_POINT_RENAME(usnumf_formatInt64) +#define usnumf_openForLocale U_ICU_ENTRY_POINT_RENAME(usnumf_openForLocale) +#define usnumf_openForLocaleAndGroupingStrategy U_ICU_ENTRY_POINT_RENAME(usnumf_openForLocaleAndGroupingStrategy) +#define uspoof_areConfusable U_ICU_ENTRY_POINT_RENAME(uspoof_areConfusable) +#define uspoof_areConfusableUTF8 U_ICU_ENTRY_POINT_RENAME(uspoof_areConfusableUTF8) +#define uspoof_areConfusableUnicodeString U_ICU_ENTRY_POINT_RENAME(uspoof_areConfusableUnicodeString) +#define uspoof_check U_ICU_ENTRY_POINT_RENAME(uspoof_check) +#define uspoof_check2 U_ICU_ENTRY_POINT_RENAME(uspoof_check2) +#define uspoof_check2UTF8 U_ICU_ENTRY_POINT_RENAME(uspoof_check2UTF8) +#define uspoof_check2UnicodeString U_ICU_ENTRY_POINT_RENAME(uspoof_check2UnicodeString) +#define uspoof_checkUTF8 U_ICU_ENTRY_POINT_RENAME(uspoof_checkUTF8) +#define uspoof_checkUnicodeString U_ICU_ENTRY_POINT_RENAME(uspoof_checkUnicodeString) +#define uspoof_clone U_ICU_ENTRY_POINT_RENAME(uspoof_clone) +#define uspoof_close U_ICU_ENTRY_POINT_RENAME(uspoof_close) +#define uspoof_closeCheckResult U_ICU_ENTRY_POINT_RENAME(uspoof_closeCheckResult) +#define uspoof_getAllowedChars U_ICU_ENTRY_POINT_RENAME(uspoof_getAllowedChars) +#define uspoof_getAllowedLocales U_ICU_ENTRY_POINT_RENAME(uspoof_getAllowedLocales) +#define uspoof_getAllowedUnicodeSet U_ICU_ENTRY_POINT_RENAME(uspoof_getAllowedUnicodeSet) +#define uspoof_getCheckResultChecks U_ICU_ENTRY_POINT_RENAME(uspoof_getCheckResultChecks) +#define uspoof_getCheckResultNumerics U_ICU_ENTRY_POINT_RENAME(uspoof_getCheckResultNumerics) +#define uspoof_getCheckResultRestrictionLevel U_ICU_ENTRY_POINT_RENAME(uspoof_getCheckResultRestrictionLevel) +#define uspoof_getChecks U_ICU_ENTRY_POINT_RENAME(uspoof_getChecks) +#define uspoof_getInclusionSet U_ICU_ENTRY_POINT_RENAME(uspoof_getInclusionSet) +#define uspoof_getInclusionUnicodeSet U_ICU_ENTRY_POINT_RENAME(uspoof_getInclusionUnicodeSet) +#define uspoof_getRecommendedSet U_ICU_ENTRY_POINT_RENAME(uspoof_getRecommendedSet) +#define uspoof_getRecommendedUnicodeSet U_ICU_ENTRY_POINT_RENAME(uspoof_getRecommendedUnicodeSet) +#define uspoof_getRestrictionLevel U_ICU_ENTRY_POINT_RENAME(uspoof_getRestrictionLevel) +#define uspoof_getSkeleton U_ICU_ENTRY_POINT_RENAME(uspoof_getSkeleton) +#define uspoof_getSkeletonUTF8 U_ICU_ENTRY_POINT_RENAME(uspoof_getSkeletonUTF8) +#define uspoof_getSkeletonUnicodeString U_ICU_ENTRY_POINT_RENAME(uspoof_getSkeletonUnicodeString) +#define uspoof_internalInitStatics U_ICU_ENTRY_POINT_RENAME(uspoof_internalInitStatics) +#define uspoof_open U_ICU_ENTRY_POINT_RENAME(uspoof_open) +#define uspoof_openCheckResult U_ICU_ENTRY_POINT_RENAME(uspoof_openCheckResult) +#define uspoof_openFromSerialized U_ICU_ENTRY_POINT_RENAME(uspoof_openFromSerialized) +#define uspoof_openFromSource U_ICU_ENTRY_POINT_RENAME(uspoof_openFromSource) +#define uspoof_serialize U_ICU_ENTRY_POINT_RENAME(uspoof_serialize) +#define uspoof_setAllowedChars U_ICU_ENTRY_POINT_RENAME(uspoof_setAllowedChars) +#define uspoof_setAllowedLocales U_ICU_ENTRY_POINT_RENAME(uspoof_setAllowedLocales) +#define uspoof_setAllowedUnicodeSet U_ICU_ENTRY_POINT_RENAME(uspoof_setAllowedUnicodeSet) +#define uspoof_setChecks U_ICU_ENTRY_POINT_RENAME(uspoof_setChecks) +#define uspoof_setRestrictionLevel U_ICU_ENTRY_POINT_RENAME(uspoof_setRestrictionLevel) +#define uspoof_swap U_ICU_ENTRY_POINT_RENAME(uspoof_swap) +#define usprep_close U_ICU_ENTRY_POINT_RENAME(usprep_close) +#define usprep_open U_ICU_ENTRY_POINT_RENAME(usprep_open) +#define usprep_openByType U_ICU_ENTRY_POINT_RENAME(usprep_openByType) +#define usprep_prepare U_ICU_ENTRY_POINT_RENAME(usprep_prepare) +#define usprep_swap U_ICU_ENTRY_POINT_RENAME(usprep_swap) +#define ustr_hashCharsN U_ICU_ENTRY_POINT_RENAME(ustr_hashCharsN) +#define ustr_hashICharsN U_ICU_ENTRY_POINT_RENAME(ustr_hashICharsN) +#define ustr_hashUCharsN U_ICU_ENTRY_POINT_RENAME(ustr_hashUCharsN) +#define ustrcase_getCaseLocale U_ICU_ENTRY_POINT_RENAME(ustrcase_getCaseLocale) +#define ustrcase_getTitleBreakIterator U_ICU_ENTRY_POINT_RENAME(ustrcase_getTitleBreakIterator) +#define ustrcase_internalFold U_ICU_ENTRY_POINT_RENAME(ustrcase_internalFold) +#define ustrcase_internalToLower U_ICU_ENTRY_POINT_RENAME(ustrcase_internalToLower) +#define ustrcase_internalToTitle U_ICU_ENTRY_POINT_RENAME(ustrcase_internalToTitle) +#define ustrcase_internalToUpper U_ICU_ENTRY_POINT_RENAME(ustrcase_internalToUpper) +#define ustrcase_map U_ICU_ENTRY_POINT_RENAME(ustrcase_map) +#define ustrcase_mapWithOverlap U_ICU_ENTRY_POINT_RENAME(ustrcase_mapWithOverlap) +#define utext_char32At U_ICU_ENTRY_POINT_RENAME(utext_char32At) +#define utext_clone U_ICU_ENTRY_POINT_RENAME(utext_clone) +#define utext_close U_ICU_ENTRY_POINT_RENAME(utext_close) +#define utext_copy U_ICU_ENTRY_POINT_RENAME(utext_copy) +#define utext_current32 U_ICU_ENTRY_POINT_RENAME(utext_current32) +#define utext_equals U_ICU_ENTRY_POINT_RENAME(utext_equals) +#define utext_extract U_ICU_ENTRY_POINT_RENAME(utext_extract) +#define utext_freeze U_ICU_ENTRY_POINT_RENAME(utext_freeze) +#define utext_getNativeIndex U_ICU_ENTRY_POINT_RENAME(utext_getNativeIndex) +#define utext_getPreviousNativeIndex U_ICU_ENTRY_POINT_RENAME(utext_getPreviousNativeIndex) +#define utext_hasMetaData U_ICU_ENTRY_POINT_RENAME(utext_hasMetaData) +#define utext_isLengthExpensive U_ICU_ENTRY_POINT_RENAME(utext_isLengthExpensive) +#define utext_isWritable U_ICU_ENTRY_POINT_RENAME(utext_isWritable) +#define utext_moveIndex32 U_ICU_ENTRY_POINT_RENAME(utext_moveIndex32) +#define utext_nativeLength U_ICU_ENTRY_POINT_RENAME(utext_nativeLength) +#define utext_next32 U_ICU_ENTRY_POINT_RENAME(utext_next32) +#define utext_next32From U_ICU_ENTRY_POINT_RENAME(utext_next32From) +#define utext_openCharacterIterator U_ICU_ENTRY_POINT_RENAME(utext_openCharacterIterator) +#define utext_openConstUnicodeString U_ICU_ENTRY_POINT_RENAME(utext_openConstUnicodeString) +#define utext_openReplaceable U_ICU_ENTRY_POINT_RENAME(utext_openReplaceable) +#define utext_openUChars U_ICU_ENTRY_POINT_RENAME(utext_openUChars) +#define utext_openUTF8 U_ICU_ENTRY_POINT_RENAME(utext_openUTF8) +#define utext_openUnicodeString U_ICU_ENTRY_POINT_RENAME(utext_openUnicodeString) +#define utext_previous32 U_ICU_ENTRY_POINT_RENAME(utext_previous32) +#define utext_previous32From U_ICU_ENTRY_POINT_RENAME(utext_previous32From) +#define utext_replace U_ICU_ENTRY_POINT_RENAME(utext_replace) +#define utext_setNativeIndex U_ICU_ENTRY_POINT_RENAME(utext_setNativeIndex) +#define utext_setup U_ICU_ENTRY_POINT_RENAME(utext_setup) +#define utf8_appendCharSafeBody U_ICU_ENTRY_POINT_RENAME(utf8_appendCharSafeBody) +#define utf8_back1SafeBody U_ICU_ENTRY_POINT_RENAME(utf8_back1SafeBody) +#define utf8_countTrailBytes U_ICU_ENTRY_POINT_RENAME(utf8_countTrailBytes) +#define utf8_nextCharSafeBody U_ICU_ENTRY_POINT_RENAME(utf8_nextCharSafeBody) +#define utf8_prevCharSafeBody U_ICU_ENTRY_POINT_RENAME(utf8_prevCharSafeBody) +#define utmscale_fromInt64 U_ICU_ENTRY_POINT_RENAME(utmscale_fromInt64) +#define utmscale_getTimeScaleValue U_ICU_ENTRY_POINT_RENAME(utmscale_getTimeScaleValue) +#define utmscale_toInt64 U_ICU_ENTRY_POINT_RENAME(utmscale_toInt64) +#define utrace_cleanup U_ICU_ENTRY_POINT_RENAME(utrace_cleanup) +#define utrace_data U_ICU_ENTRY_POINT_RENAME(utrace_data) +#define utrace_entry U_ICU_ENTRY_POINT_RENAME(utrace_entry) +#define utrace_exit U_ICU_ENTRY_POINT_RENAME(utrace_exit) +#define utrace_format U_ICU_ENTRY_POINT_RENAME(utrace_format) +#define utrace_functionName U_ICU_ENTRY_POINT_RENAME(utrace_functionName) +#define utrace_getFunctions U_ICU_ENTRY_POINT_RENAME(utrace_getFunctions) +#define utrace_getLevel U_ICU_ENTRY_POINT_RENAME(utrace_getLevel) +#define utrace_setFunctions U_ICU_ENTRY_POINT_RENAME(utrace_setFunctions) +#define utrace_setLevel U_ICU_ENTRY_POINT_RENAME(utrace_setLevel) +#define utrace_vformat U_ICU_ENTRY_POINT_RENAME(utrace_vformat) +#define utrans_clone U_ICU_ENTRY_POINT_RENAME(utrans_clone) +#define utrans_close U_ICU_ENTRY_POINT_RENAME(utrans_close) +#define utrans_countAvailableIDs U_ICU_ENTRY_POINT_RENAME(utrans_countAvailableIDs) +#define utrans_getAvailableID U_ICU_ENTRY_POINT_RENAME(utrans_getAvailableID) +#define utrans_getID U_ICU_ENTRY_POINT_RENAME(utrans_getID) +#define utrans_getSourceSet U_ICU_ENTRY_POINT_RENAME(utrans_getSourceSet) +#define utrans_getUnicodeID U_ICU_ENTRY_POINT_RENAME(utrans_getUnicodeID) +#define utrans_open U_ICU_ENTRY_POINT_RENAME(utrans_open) +#define utrans_openIDs U_ICU_ENTRY_POINT_RENAME(utrans_openIDs) +#define utrans_openInverse U_ICU_ENTRY_POINT_RENAME(utrans_openInverse) +#define utrans_openU U_ICU_ENTRY_POINT_RENAME(utrans_openU) +#define utrans_register U_ICU_ENTRY_POINT_RENAME(utrans_register) +#define utrans_rep_caseContextIterator U_ICU_ENTRY_POINT_RENAME(utrans_rep_caseContextIterator) +#define utrans_setFilter U_ICU_ENTRY_POINT_RENAME(utrans_setFilter) +#define utrans_stripRules U_ICU_ENTRY_POINT_RENAME(utrans_stripRules) +#define utrans_toRules U_ICU_ENTRY_POINT_RENAME(utrans_toRules) +#define utrans_trans U_ICU_ENTRY_POINT_RENAME(utrans_trans) +#define utrans_transIncremental U_ICU_ENTRY_POINT_RENAME(utrans_transIncremental) +#define utrans_transIncrementalUChars U_ICU_ENTRY_POINT_RENAME(utrans_transIncrementalUChars) +#define utrans_transUChars U_ICU_ENTRY_POINT_RENAME(utrans_transUChars) +#define utrans_transliterator_cleanup U_ICU_ENTRY_POINT_RENAME(utrans_transliterator_cleanup) +#define utrans_unregister U_ICU_ENTRY_POINT_RENAME(utrans_unregister) +#define utrans_unregisterID U_ICU_ENTRY_POINT_RENAME(utrans_unregisterID) +#define utrie2_clone U_ICU_ENTRY_POINT_RENAME(utrie2_clone) +#define utrie2_cloneAsThawed U_ICU_ENTRY_POINT_RENAME(utrie2_cloneAsThawed) +#define utrie2_close U_ICU_ENTRY_POINT_RENAME(utrie2_close) +#define utrie2_enum U_ICU_ENTRY_POINT_RENAME(utrie2_enum) +#define utrie2_enumForLeadSurrogate U_ICU_ENTRY_POINT_RENAME(utrie2_enumForLeadSurrogate) +#define utrie2_freeze U_ICU_ENTRY_POINT_RENAME(utrie2_freeze) +#define utrie2_fromUTrie U_ICU_ENTRY_POINT_RENAME(utrie2_fromUTrie) +#define utrie2_get32 U_ICU_ENTRY_POINT_RENAME(utrie2_get32) +#define utrie2_get32FromLeadSurrogateCodeUnit U_ICU_ENTRY_POINT_RENAME(utrie2_get32FromLeadSurrogateCodeUnit) +#define utrie2_internalU8NextIndex U_ICU_ENTRY_POINT_RENAME(utrie2_internalU8NextIndex) +#define utrie2_internalU8PrevIndex U_ICU_ENTRY_POINT_RENAME(utrie2_internalU8PrevIndex) +#define utrie2_isFrozen U_ICU_ENTRY_POINT_RENAME(utrie2_isFrozen) +#define utrie2_open U_ICU_ENTRY_POINT_RENAME(utrie2_open) +#define utrie2_openDummy U_ICU_ENTRY_POINT_RENAME(utrie2_openDummy) +#define utrie2_openFromSerialized U_ICU_ENTRY_POINT_RENAME(utrie2_openFromSerialized) +#define utrie2_serialize U_ICU_ENTRY_POINT_RENAME(utrie2_serialize) +#define utrie2_set32 U_ICU_ENTRY_POINT_RENAME(utrie2_set32) +#define utrie2_set32ForLeadSurrogateCodeUnit U_ICU_ENTRY_POINT_RENAME(utrie2_set32ForLeadSurrogateCodeUnit) +#define utrie2_setRange32 U_ICU_ENTRY_POINT_RENAME(utrie2_setRange32) +#define utrie2_swap U_ICU_ENTRY_POINT_RENAME(utrie2_swap) +#define utrie_clone U_ICU_ENTRY_POINT_RENAME(utrie_clone) +#define utrie_close U_ICU_ENTRY_POINT_RENAME(utrie_close) +#define utrie_defaultGetFoldingOffset U_ICU_ENTRY_POINT_RENAME(utrie_defaultGetFoldingOffset) +#define utrie_enum U_ICU_ENTRY_POINT_RENAME(utrie_enum) +#define utrie_get32 U_ICU_ENTRY_POINT_RENAME(utrie_get32) +#define utrie_getData U_ICU_ENTRY_POINT_RENAME(utrie_getData) +#define utrie_open U_ICU_ENTRY_POINT_RENAME(utrie_open) +#define utrie_serialize U_ICU_ENTRY_POINT_RENAME(utrie_serialize) +#define utrie_set32 U_ICU_ENTRY_POINT_RENAME(utrie_set32) +#define utrie_setRange32 U_ICU_ENTRY_POINT_RENAME(utrie_setRange32) +#define utrie_swap U_ICU_ENTRY_POINT_RENAME(utrie_swap) +#define utrie_swapAnyVersion U_ICU_ENTRY_POINT_RENAME(utrie_swapAnyVersion) +#define utrie_unserialize U_ICU_ENTRY_POINT_RENAME(utrie_unserialize) +#define utrie_unserializeDummy U_ICU_ENTRY_POINT_RENAME(utrie_unserializeDummy) +#define vzone_clone U_ICU_ENTRY_POINT_RENAME(vzone_clone) +#define vzone_close U_ICU_ENTRY_POINT_RENAME(vzone_close) +#define vzone_countTransitionRules U_ICU_ENTRY_POINT_RENAME(vzone_countTransitionRules) +#define vzone_equals U_ICU_ENTRY_POINT_RENAME(vzone_equals) +#define vzone_getDynamicClassID U_ICU_ENTRY_POINT_RENAME(vzone_getDynamicClassID) +#define vzone_getLastModified U_ICU_ENTRY_POINT_RENAME(vzone_getLastModified) +#define vzone_getNextTransition U_ICU_ENTRY_POINT_RENAME(vzone_getNextTransition) +#define vzone_getOffset U_ICU_ENTRY_POINT_RENAME(vzone_getOffset) +#define vzone_getOffset2 U_ICU_ENTRY_POINT_RENAME(vzone_getOffset2) +#define vzone_getOffset3 U_ICU_ENTRY_POINT_RENAME(vzone_getOffset3) +#define vzone_getPreviousTransition U_ICU_ENTRY_POINT_RENAME(vzone_getPreviousTransition) +#define vzone_getRawOffset U_ICU_ENTRY_POINT_RENAME(vzone_getRawOffset) +#define vzone_getStaticClassID U_ICU_ENTRY_POINT_RENAME(vzone_getStaticClassID) +#define vzone_getTZURL U_ICU_ENTRY_POINT_RENAME(vzone_getTZURL) +#define vzone_hasSameRules U_ICU_ENTRY_POINT_RENAME(vzone_hasSameRules) +#define vzone_inDaylightTime U_ICU_ENTRY_POINT_RENAME(vzone_inDaylightTime) +#define vzone_openData U_ICU_ENTRY_POINT_RENAME(vzone_openData) +#define vzone_openID U_ICU_ENTRY_POINT_RENAME(vzone_openID) +#define vzone_setLastModified U_ICU_ENTRY_POINT_RENAME(vzone_setLastModified) +#define vzone_setRawOffset U_ICU_ENTRY_POINT_RENAME(vzone_setRawOffset) +#define vzone_setTZURL U_ICU_ENTRY_POINT_RENAME(vzone_setTZURL) +#define vzone_useDaylightTime U_ICU_ENTRY_POINT_RENAME(vzone_useDaylightTime) +#define vzone_write U_ICU_ENTRY_POINT_RENAME(vzone_write) +#define vzone_writeFromStart U_ICU_ENTRY_POINT_RENAME(vzone_writeFromStart) +#define vzone_writeSimple U_ICU_ENTRY_POINT_RENAME(vzone_writeSimple) +#define zrule_close U_ICU_ENTRY_POINT_RENAME(zrule_close) +#define zrule_equals U_ICU_ENTRY_POINT_RENAME(zrule_equals) +#define zrule_getDSTSavings U_ICU_ENTRY_POINT_RENAME(zrule_getDSTSavings) +#define zrule_getName U_ICU_ENTRY_POINT_RENAME(zrule_getName) +#define zrule_getRawOffset U_ICU_ENTRY_POINT_RENAME(zrule_getRawOffset) +#define zrule_isEquivalentTo U_ICU_ENTRY_POINT_RENAME(zrule_isEquivalentTo) +#define ztrans_adoptFrom U_ICU_ENTRY_POINT_RENAME(ztrans_adoptFrom) +#define ztrans_adoptTo U_ICU_ENTRY_POINT_RENAME(ztrans_adoptTo) +#define ztrans_clone U_ICU_ENTRY_POINT_RENAME(ztrans_clone) +#define ztrans_close U_ICU_ENTRY_POINT_RENAME(ztrans_close) +#define ztrans_equals U_ICU_ENTRY_POINT_RENAME(ztrans_equals) +#define ztrans_getDynamicClassID U_ICU_ENTRY_POINT_RENAME(ztrans_getDynamicClassID) +#define ztrans_getFrom U_ICU_ENTRY_POINT_RENAME(ztrans_getFrom) +#define ztrans_getStaticClassID U_ICU_ENTRY_POINT_RENAME(ztrans_getStaticClassID) +#define ztrans_getTime U_ICU_ENTRY_POINT_RENAME(ztrans_getTime) +#define ztrans_getTo U_ICU_ENTRY_POINT_RENAME(ztrans_getTo) +#define ztrans_open U_ICU_ENTRY_POINT_RENAME(ztrans_open) +#define ztrans_openEmpty U_ICU_ENTRY_POINT_RENAME(ztrans_openEmpty) +#define ztrans_setFrom U_ICU_ENTRY_POINT_RENAME(ztrans_setFrom) +#define ztrans_setTime U_ICU_ENTRY_POINT_RENAME(ztrans_setTime) +#define ztrans_setTo U_ICU_ENTRY_POINT_RENAME(ztrans_setTo) + +#endif /* !(defined(_MSC_VER) && defined(__INTELLISENSE__)) */ +#endif /* U_DISABLE_RENAMING */ +#endif /* URENAME_H */ + diff --git a/intl/icu/source/common/unicode/urep.h b/intl/icu/source/common/unicode/urep.h new file mode 100644 index 0000000000..932202ddb0 --- /dev/null +++ b/intl/icu/source/common/unicode/urep.h @@ -0,0 +1,157 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* Copyright (C) 1997-2010, International Business Machines +* Corporation and others. All Rights Reserved. +****************************************************************************** +* Date Name Description +* 06/23/00 aliu Creation. +****************************************************************************** +*/ + +#ifndef __UREP_H +#define __UREP_H + +#include "unicode/utypes.h" + +U_CDECL_BEGIN + +/******************************************************************** + * General Notes + ******************************************************************** + * TODO + * Add usage scenario + * Add test code + * Talk about pinning + * Talk about "can truncate result if out of memory" + */ + +/******************************************************************** + * Data Structures + ********************************************************************/ +/** + * \file + * \brief C API: Callbacks for UReplaceable + */ +/** + * An opaque replaceable text object. This will be manipulated only + * through the caller-supplied UReplaceableFunctor struct. Related + * to the C++ class Replaceable. + * This is currently only used in the Transliterator C API, see utrans.h . + * @stable ICU 2.0 + */ +typedef void* UReplaceable; + +/** + * A set of function pointers that transliterators use to manipulate a + * UReplaceable. The caller should supply the required functions to + * manipulate their text appropriately. Related to the C++ class + * Replaceable. + * @stable ICU 2.0 + */ +typedef struct UReplaceableCallbacks { + + /** + * Function pointer that returns the number of UChar code units in + * this text. + * + * @param rep A pointer to "this" UReplaceable object. + * @return The length of the text. + * @stable ICU 2.0 + */ + int32_t (*length)(const UReplaceable* rep); + + /** + * Function pointer that returns a UChar code units at the given + * offset into this text; 0 <= offset < n, where n is the value + * returned by (*length)(rep). See unistr.h for a description of + * charAt() vs. char32At(). + * + * @param rep A pointer to "this" UReplaceable object. + * @param offset The index at which to fetch the UChar (code unit). + * @return The UChar (code unit) at offset, or U+FFFF if the offset is out of bounds. + * @stable ICU 2.0 + */ + UChar (*charAt)(const UReplaceable* rep, + int32_t offset); + + /** + * Function pointer that returns a UChar32 code point at the given + * offset into this text. See unistr.h for a description of + * charAt() vs. char32At(). + * + * @param rep A pointer to "this" UReplaceable object. + * @param offset The index at which to fetch the UChar32 (code point). + * @return The UChar32 (code point) at offset, or U+FFFF if the offset is out of bounds. + * @stable ICU 2.0 + */ + UChar32 (*char32At)(const UReplaceable* rep, + int32_t offset); + + /** + * Function pointer that replaces text between start and limit in + * this text with the given text. Attributes (out of band info) + * should be retained. + * + * @param rep A pointer to "this" UReplaceable object. + * @param start the starting index of the text to be replaced, + * inclusive. + * @param limit the ending index of the text to be replaced, + * exclusive. + * @param text the new text to replace the UChars from + * start..limit-1. + * @param textLength the number of UChars at text, or -1 if text + * is null-terminated. + * @stable ICU 2.0 + */ + void (*replace)(UReplaceable* rep, + int32_t start, + int32_t limit, + const UChar* text, + int32_t textLength); + + /** + * Function pointer that copies the characters in the range + * [start, limit) into the array dst. + * + * @param rep A pointer to "this" UReplaceable object. + * @param start offset of first character which will be copied + * into the array + * @param limit offset immediately following the last character to + * be copied + * @param dst array in which to copy characters. The length of + * dst must be at least (limit - start). + * @stable ICU 2.1 + */ + void (*extract)(UReplaceable* rep, + int32_t start, + int32_t limit, + UChar* dst); + + /** + * Function pointer that copies text between start and limit in + * this text to another index in the text. Attributes (out of + * band info) should be retained. After this call, there will be + * (at least) two copies of the characters originally located at + * start..limit-1. + * + * @param rep A pointer to "this" UReplaceable object. + * @param start the starting index of the text to be copied, + * inclusive. + * @param limit the ending index of the text to be copied, + * exclusive. + * @param dest the index at which the copy of the UChars should be + * inserted. + * @stable ICU 2.0 + */ + void (*copy)(UReplaceable* rep, + int32_t start, + int32_t limit, + int32_t dest); + +} UReplaceableCallbacks; + +U_CDECL_END + +#endif diff --git a/intl/icu/source/common/unicode/ures.h b/intl/icu/source/common/unicode/ures.h new file mode 100644 index 0000000000..cc25b6e49c --- /dev/null +++ b/intl/icu/source/common/unicode/ures.h @@ -0,0 +1,911 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1997-2016, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* +* File URES.H (formerly CRESBUND.H) +* +* Modification History: +* +* Date Name Description +* 04/01/97 aliu Creation. +* 02/22/99 damiba overhaul. +* 04/04/99 helena Fixed internal header inclusion. +* 04/15/99 Madhu Updated Javadoc +* 06/14/99 stephen Removed functions taking a filename suffix. +* 07/20/99 stephen Language-independent typedef to void* +* 11/09/99 weiv Added ures_getLocale() +* 06/24/02 weiv Added support for resource sharing +****************************************************************************** +*/ + +#ifndef URES_H +#define URES_H + +#include "unicode/utypes.h" +#include "unicode/uloc.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +/** + * \file + * \brief C API: Resource Bundle + * + *

C API: Resource Bundle

+ * + * C API representing a collection of resource information pertaining to a given + * locale. A resource bundle provides a way of accessing locale- specific information in + * a data file. You create a resource bundle that manages the resources for a given + * locale and then ask it for individual resources. + *

+ * Resource bundles in ICU4C are currently defined using text files which conform to the following + * BNF definition. + * More on resource bundle concepts and syntax can be found in the + * Users Guide. + *

+ */ + +/** + * UResourceBundle is an opaque type for handles for resource bundles in C APIs. + * @stable ICU 2.0 + */ +struct UResourceBundle; + +/** + * @stable ICU 2.0 + */ +typedef struct UResourceBundle UResourceBundle; + +/** + * Numeric constants for types of resource items. + * @see ures_getType + * @stable ICU 2.0 + */ +typedef enum { + /** Resource type constant for "no resource". @stable ICU 2.6 */ + URES_NONE=-1, + + /** Resource type constant for 16-bit Unicode strings. @stable ICU 2.6 */ + URES_STRING=0, + + /** Resource type constant for binary data. @stable ICU 2.6 */ + URES_BINARY=1, + + /** Resource type constant for tables of key-value pairs. @stable ICU 2.6 */ + URES_TABLE=2, + + /** + * Resource type constant for aliases; + * internally stores a string which identifies the actual resource + * storing the data (can be in a different resource bundle). + * Resolved internally before delivering the actual resource through the API. + * @stable ICU 2.6 + */ + URES_ALIAS=3, + + /** + * Resource type constant for a single 28-bit integer, interpreted as + * signed or unsigned by the ures_getInt() or ures_getUInt() function. + * @see ures_getInt + * @see ures_getUInt + * @stable ICU 2.6 + */ + URES_INT=7, + + /** Resource type constant for arrays of resources. @stable ICU 2.6 */ + URES_ARRAY=8, + + /** + * Resource type constant for vectors of 32-bit integers. + * @see ures_getIntVector + * @stable ICU 2.6 + */ + URES_INT_VECTOR = 14, +#ifndef U_HIDE_DEPRECATED_API + /** @deprecated ICU 2.6 Use the URES_ constant instead. */ + RES_NONE=URES_NONE, + /** @deprecated ICU 2.6 Use the URES_ constant instead. */ + RES_STRING=URES_STRING, + /** @deprecated ICU 2.6 Use the URES_ constant instead. */ + RES_BINARY=URES_BINARY, + /** @deprecated ICU 2.6 Use the URES_ constant instead. */ + RES_TABLE=URES_TABLE, + /** @deprecated ICU 2.6 Use the URES_ constant instead. */ + RES_ALIAS=URES_ALIAS, + /** @deprecated ICU 2.6 Use the URES_ constant instead. */ + RES_INT=URES_INT, + /** @deprecated ICU 2.6 Use the URES_ constant instead. */ + RES_ARRAY=URES_ARRAY, + /** @deprecated ICU 2.6 Use the URES_ constant instead. */ + RES_INT_VECTOR=URES_INT_VECTOR, + /** @deprecated ICU 2.6 Not used. */ + RES_RESERVED=15, + + /** + * One more than the highest normal UResType value. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + URES_LIMIT = 16 +#endif // U_HIDE_DEPRECATED_API +} UResType; + +/* + * Functions to create and destroy resource bundles. + */ + +/** + * Opens a UResourceBundle, from which users can extract strings by using + * their corresponding keys. + * Note that the caller is responsible of calling ures_close on each successfully + * opened resource bundle. + * @param packageName The packageName and locale together point to an ICU udata object, + * as defined by udata_open( packageName, "res", locale, err) + * or equivalent. Typically, packageName will refer to a (.dat) file, or to + * a package registered with udata_setAppData(). Using a full file or directory + * pathname for packageName is deprecated. If NULL, ICU data will be used. + * @param locale specifies the locale for which we want to open the resource + * if NULL, the default locale will be used. If strlen(locale) == 0 + * root locale will be used. + * + * @param status fills in the outgoing error code. + * The UErrorCode err parameter is used to return status information to the user. To + * check whether the construction succeeded or not, you should check the value of + * U_SUCCESS(err). If you wish more detailed information, you can check for + * informational status results which still indicate success. U_USING_FALLBACK_WARNING + * indicates that a fall back locale was used. For example, 'de_CH' was requested, + * but nothing was found there, so 'de' was used. U_USING_DEFAULT_WARNING indicates that + * the default locale data or root locale data was used; neither the requested locale + * nor any of its fall back locales could be found. Please see the users guide for more + * information on this topic. + * @return a newly allocated resource bundle. + * @see ures_close + * @stable ICU 2.0 + */ +U_CAPI UResourceBundle* U_EXPORT2 +ures_open(const char* packageName, + const char* locale, + UErrorCode* status); + + +/** This function does not care what kind of localeID is passed in. It simply opens a bundle with + * that name. Fallback mechanism is disabled for the new bundle. If the requested bundle contains + * an %%ALIAS directive, the results are undefined. + * @param packageName The packageName and locale together point to an ICU udata object, + * as defined by udata_open( packageName, "res", locale, err) + * or equivalent. Typically, packageName will refer to a (.dat) file, or to + * a package registered with udata_setAppData(). Using a full file or directory + * pathname for packageName is deprecated. If NULL, ICU data will be used. + * @param locale specifies the locale for which we want to open the resource + * if NULL, the default locale will be used. If strlen(locale) == 0 + * root locale will be used. + * + * @param status fills in the outgoing error code. Either U_ZERO_ERROR or U_MISSING_RESOURCE_ERROR + * @return a newly allocated resource bundle or NULL if it doesn't exist. + * @see ures_close + * @stable ICU 2.0 + */ +U_CAPI UResourceBundle* U_EXPORT2 +ures_openDirect(const char* packageName, + const char* locale, + UErrorCode* status); + +/** + * Same as ures_open() but takes a const UChar *path. + * This path will be converted to char * using the default converter, + * then ures_open() is called. + * + * @param packageName The packageName and locale together point to an ICU udata object, + * as defined by udata_open( packageName, "res", locale, err) + * or equivalent. Typically, packageName will refer to a (.dat) file, or to + * a package registered with udata_setAppData(). Using a full file or directory + * pathname for packageName is deprecated. If NULL, ICU data will be used. + * @param locale specifies the locale for which we want to open the resource + * if NULL, the default locale will be used. If strlen(locale) == 0 + * root locale will be used. + * @param status fills in the outgoing error code. + * @return a newly allocated resource bundle. + * @see ures_open + * @stable ICU 2.0 + */ +U_CAPI UResourceBundle* U_EXPORT2 +ures_openU(const UChar* packageName, + const char* locale, + UErrorCode* status); + +#ifndef U_HIDE_DEPRECATED_API +/** + * Returns the number of strings/arrays in resource bundles. + * Better to use ures_getSize, as this function will be deprecated. + * + *@param resourceBundle resource bundle containing the desired strings + *@param resourceKey key tagging the resource + *@param err fills in the outgoing error code + * could be U_MISSING_RESOURCE_ERROR if the key is not found + * could be a non-failing error + * e.g.: U_USING_FALLBACK_WARNING,U_USING_FALLBACK_WARNING + *@return: for Arrays: returns the number of resources in the array + * Tables: returns the number of resources in the table + * single string: returns 1 + *@see ures_getSize + * @deprecated ICU 2.8 User ures_getSize instead + */ +U_DEPRECATED int32_t U_EXPORT2 +ures_countArrayItems(const UResourceBundle* resourceBundle, + const char* resourceKey, + UErrorCode* err); +#endif /* U_HIDE_DEPRECATED_API */ + +/** + * Close a resource bundle, all pointers returned from the various ures_getXXX calls + * on this particular bundle should be considered invalid henceforth. + * + * @param resourceBundle a pointer to a resourceBundle struct. Can be NULL. + * @see ures_open + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ures_close(UResourceBundle* resourceBundle); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUResourceBundlePointer + * "Smart pointer" class, closes a UResourceBundle via ures_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUResourceBundlePointer, UResourceBundle, ures_close); + +U_NAMESPACE_END + +#endif + +#ifndef U_HIDE_DEPRECATED_API +/** + * Return the version number associated with this ResourceBundle as a string. Please + * use ures_getVersion as this function is going to be deprecated. + * + * @param resourceBundle The resource bundle for which the version is checked. + * @return A version number string as specified in the resource bundle or its parent. + * The caller does not own this string. + * @see ures_getVersion + * @deprecated ICU 2.8 Use ures_getVersion instead. + */ +U_DEPRECATED const char* U_EXPORT2 +ures_getVersionNumber(const UResourceBundle* resourceBundle); +#endif /* U_HIDE_DEPRECATED_API */ + +/** + * Return the version number associated with this ResourceBundle as an + * UVersionInfo array. + * + * @param resB The resource bundle for which the version is checked. + * @param versionInfo A UVersionInfo array that is filled with the version number + * as specified in the resource bundle or its parent. + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ures_getVersion(const UResourceBundle* resB, + UVersionInfo versionInfo); + +#ifndef U_HIDE_DEPRECATED_API +/** + * Return the name of the Locale associated with this ResourceBundle. This API allows + * you to query for the real locale of the resource. For example, if you requested + * "en_US_CALIFORNIA" and only "en_US" bundle exists, "en_US" will be returned. + * For subresources, the locale where this resource comes from will be returned. + * If fallback has occurred, getLocale will reflect this. + * + * @param resourceBundle resource bundle in question + * @param status just for catching illegal arguments + * @return A Locale name + * @deprecated ICU 2.8 Use ures_getLocaleByType instead. + */ +U_DEPRECATED const char* U_EXPORT2 +ures_getLocale(const UResourceBundle* resourceBundle, + UErrorCode* status); +#endif /* U_HIDE_DEPRECATED_API */ + +/** + * Return the name of the Locale associated with this ResourceBundle. + * You can choose between requested, valid and real locale. + * + * @param resourceBundle resource bundle in question + * @param type You can choose between requested, valid and actual + * locale. For description see the definition of + * ULocDataLocaleType in uloc.h + * @param status just for catching illegal arguments + * @return A Locale name + * @stable ICU 2.8 + */ +U_CAPI const char* U_EXPORT2 +ures_getLocaleByType(const UResourceBundle* resourceBundle, + ULocDataLocaleType type, + UErrorCode* status); + + +#ifndef U_HIDE_INTERNAL_API +/** + * Same as ures_open() but uses the fill-in parameter instead of allocating a new bundle. + * + * TODO need to revisit usefulness of this function + * and usage model for fillIn parameters without knowing sizeof(UResourceBundle) + * @param r The existing UResourceBundle to fill in. If NULL then status will be + * set to U_ILLEGAL_ARGUMENT_ERROR. + * @param packageName The packageName and locale together point to an ICU udata object, + * as defined by udata_open( packageName, "res", locale, err) + * or equivalent. Typically, packageName will refer to a (.dat) file, or to + * a package registered with udata_setAppData(). Using a full file or directory + * pathname for packageName is deprecated. If NULL, ICU data will be used. + * @param localeID specifies the locale for which we want to open the resource + * @param status The error code. + * @internal + */ +U_CAPI void U_EXPORT2 +ures_openFillIn(UResourceBundle *r, + const char* packageName, + const char* localeID, + UErrorCode* status); +#endif /* U_HIDE_INTERNAL_API */ + +/** + * Returns a string from a string resource type + * + * @param resourceBundle a string resource + * @param len fills in the length of resulting string + * @param status fills in the outgoing error code + * could be U_MISSING_RESOURCE_ERROR if the key is not found + * Always check the value of status. Don't count on returning NULL. + * could be a non-failing error + * e.g.: U_USING_FALLBACK_WARNING,U_USING_DEFAULT_WARNING + * @return a pointer to a zero-terminated UChar array which lives in a memory mapped/DLL file. + * @see ures_getBinary + * @see ures_getIntVector + * @see ures_getInt + * @see ures_getUInt + * @stable ICU 2.0 + */ +U_CAPI const UChar* U_EXPORT2 +ures_getString(const UResourceBundle* resourceBundle, + int32_t* len, + UErrorCode* status); + +/** + * Returns a UTF-8 string from a string resource. + * The UTF-8 string may be returnable directly as a pointer, or + * it may need to be copied, or transformed from UTF-16 using u_strToUTF8() + * or equivalent. + * + * If forceCopy==true, then the string is always written to the dest buffer + * and dest is returned. + * + * If forceCopy==false, then the string is returned as a pointer if possible, + * without needing a dest buffer (it can be NULL). If the string needs to be + * copied or transformed, then it may be placed into dest at an arbitrary offset. + * + * If the string is to be written to dest, then U_BUFFER_OVERFLOW_ERROR and + * U_STRING_NOT_TERMINATED_WARNING are set if appropriate, as usual. + * + * If the string is transformed from UTF-16, then a conversion error may occur + * if an unpaired surrogate is encountered. If the function is successful, then + * the output UTF-8 string is always well-formed. + * + * @param resB Resource bundle. + * @param dest Destination buffer. Can be NULL only if capacity=*length==0. + * @param length Input: Capacity of destination buffer. + * Output: Actual length of the UTF-8 string, not counting the + * terminating NUL, even in case of U_BUFFER_OVERFLOW_ERROR. + * Can be NULL, meaning capacity=0 and the string length is not + * returned to the caller. + * @param forceCopy If true, then the output string will always be written to + * dest, with U_BUFFER_OVERFLOW_ERROR and + * U_STRING_NOT_TERMINATED_WARNING set if appropriate. + * If false, then the dest buffer may or may not contain a + * copy of the string. dest may or may not be modified. + * If a copy needs to be written, then the UErrorCode parameter + * indicates overflow etc. as usual. + * @param status Pointer to a standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return The pointer to the UTF-8 string. It may be dest, or at some offset + * from dest (only if !forceCopy), or in unrelated memory. + * Always NUL-terminated unless the string was written to dest and + * length==capacity (in which case U_STRING_NOT_TERMINATED_WARNING is set). + * + * @see ures_getString + * @see u_strToUTF8 + * @stable ICU 3.6 + */ +U_CAPI const char * U_EXPORT2 +ures_getUTF8String(const UResourceBundle *resB, + char *dest, int32_t *length, + UBool forceCopy, + UErrorCode *status); + +/** + * Returns a binary data from a binary resource. + * + * @param resourceBundle a string resource + * @param len fills in the length of resulting byte chunk + * @param status fills in the outgoing error code + * could be U_MISSING_RESOURCE_ERROR if the key is not found + * Always check the value of status. Don't count on returning NULL. + * could be a non-failing error + * e.g.: U_USING_FALLBACK_WARNING,U_USING_DEFAULT_WARNING + * @return a pointer to a chunk of unsigned bytes which live in a memory mapped/DLL file. + * @see ures_getString + * @see ures_getIntVector + * @see ures_getInt + * @see ures_getUInt + * @stable ICU 2.0 + */ +U_CAPI const uint8_t* U_EXPORT2 +ures_getBinary(const UResourceBundle* resourceBundle, + int32_t* len, + UErrorCode* status); + +/** + * Returns a 32 bit integer array from a resource. + * + * @param resourceBundle an int vector resource + * @param len fills in the length of resulting byte chunk + * @param status fills in the outgoing error code + * could be U_MISSING_RESOURCE_ERROR if the key is not found + * Always check the value of status. Don't count on returning NULL. + * could be a non-failing error + * e.g.: U_USING_FALLBACK_WARNING,U_USING_DEFAULT_WARNING + * @return a pointer to a chunk of integers which live in a memory mapped/DLL file. + * @see ures_getBinary + * @see ures_getString + * @see ures_getInt + * @see ures_getUInt + * @stable ICU 2.0 + */ +U_CAPI const int32_t* U_EXPORT2 +ures_getIntVector(const UResourceBundle* resourceBundle, + int32_t* len, + UErrorCode* status); + +/** + * Returns an unsigned integer from a resource. + * This integer is originally 28 bits. + * + * @param resourceBundle a string resource + * @param status fills in the outgoing error code + * could be U_MISSING_RESOURCE_ERROR if the key is not found + * could be a non-failing error + * e.g.: U_USING_FALLBACK_WARNING,U_USING_DEFAULT_WARNING + * @return an integer value + * @see ures_getInt + * @see ures_getIntVector + * @see ures_getBinary + * @see ures_getString + * @stable ICU 2.0 + */ +U_CAPI uint32_t U_EXPORT2 +ures_getUInt(const UResourceBundle* resourceBundle, + UErrorCode *status); + +/** + * Returns a signed integer from a resource. + * This integer is originally 28 bit and the sign gets propagated. + * + * @param resourceBundle a string resource + * @param status fills in the outgoing error code + * could be U_MISSING_RESOURCE_ERROR if the key is not found + * could be a non-failing error + * e.g.: U_USING_FALLBACK_WARNING,U_USING_DEFAULT_WARNING + * @return an integer value + * @see ures_getUInt + * @see ures_getIntVector + * @see ures_getBinary + * @see ures_getString + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ures_getInt(const UResourceBundle* resourceBundle, + UErrorCode *status); + +/** + * Returns the size of a resource. Size for scalar types is always 1, + * and for vector/table types is the number of child resources. + * @warning Integer array is treated as a scalar type. There are no + * APIs to access individual members of an integer array. It + * is always returned as a whole. + * @param resourceBundle a resource + * @return number of resources in a given resource. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +ures_getSize(const UResourceBundle *resourceBundle); + +/** + * Returns the type of a resource. Available types are defined in enum UResType + * + * @param resourceBundle a resource + * @return type of the given resource. + * @see UResType + * @stable ICU 2.0 + */ +U_CAPI UResType U_EXPORT2 +ures_getType(const UResourceBundle *resourceBundle); + +/** + * Returns the key associated with a given resource. Not all the resources have a key - only + * those that are members of a table. + * + * @param resourceBundle a resource + * @return a key associated to this resource, or NULL if it doesn't have a key + * @stable ICU 2.0 + */ +U_CAPI const char * U_EXPORT2 +ures_getKey(const UResourceBundle *resourceBundle); + +/* ITERATION API + This API provides means for iterating through a resource +*/ + +/** + * Resets the internal context of a resource so that iteration starts from the first element. + * + * @param resourceBundle a resource + * @stable ICU 2.0 + */ +U_CAPI void U_EXPORT2 +ures_resetIterator(UResourceBundle *resourceBundle); + +/** + * Checks whether the given resource has another element to iterate over. + * + * @param resourceBundle a resource + * @return true if there are more elements, false if there is no more elements + * @stable ICU 2.0 + */ +U_CAPI UBool U_EXPORT2 +ures_hasNext(const UResourceBundle *resourceBundle); + +/** + * Returns the next resource in a given resource or NULL if there are no more resources + * to iterate over. Features a fill-in parameter. + * + * @param resourceBundle a resource + * @param fillIn if NULL a new UResourceBundle struct is allocated and must be closed by the caller. + * Alternatively, you can supply a struct to be filled by this function. + * @param status fills in the outgoing error code. You may still get a non NULL result even if an + * error occurred. Check status instead. + * @return a pointer to a UResourceBundle struct. If fill in param was NULL, caller must close it + * @stable ICU 2.0 + */ +U_CAPI UResourceBundle* U_EXPORT2 +ures_getNextResource(UResourceBundle *resourceBundle, + UResourceBundle *fillIn, + UErrorCode *status); + +/** + * Returns the next string in a given resource or NULL if there are no more resources + * to iterate over. + * + * @param resourceBundle a resource + * @param len fill in length of the string + * @param key fill in for key associated with this string. NULL if no key + * @param status fills in the outgoing error code. If an error occurred, we may return NULL, but don't + * count on it. Check status instead! + * @return a pointer to a zero-terminated UChar array which lives in a memory mapped/DLL file. + * @stable ICU 2.0 + */ +U_CAPI const UChar* U_EXPORT2 +ures_getNextString(UResourceBundle *resourceBundle, + int32_t* len, + const char ** key, + UErrorCode *status); + +/** + * Returns the resource in a given resource at the specified index. Features a fill-in parameter. + * + * @param resourceBundle the resource bundle from which to get a sub-resource + * @param indexR an index to the wanted resource. + * @param fillIn if NULL a new UResourceBundle struct is allocated and must be closed by the caller. + * Alternatively, you can supply a struct to be filled by this function. + * @param status fills in the outgoing error code. Don't count on NULL being returned if an error has + * occurred. Check status instead. + * @return a pointer to a UResourceBundle struct. If fill in param was NULL, caller must close it + * @stable ICU 2.0 + */ +U_CAPI UResourceBundle* U_EXPORT2 +ures_getByIndex(const UResourceBundle *resourceBundle, + int32_t indexR, + UResourceBundle *fillIn, + UErrorCode *status); + +/** + * Returns the string in a given resource at the specified index. + * + * @param resourceBundle a resource + * @param indexS an index to the wanted string. + * @param len fill in length of the string + * @param status fills in the outgoing error code. If an error occurred, we may return NULL, but don't + * count on it. Check status instead! + * @return a pointer to a zero-terminated UChar array which lives in a memory mapped/DLL file. + * @stable ICU 2.0 + */ +U_CAPI const UChar* U_EXPORT2 +ures_getStringByIndex(const UResourceBundle *resourceBundle, + int32_t indexS, + int32_t* len, + UErrorCode *status); + +/** + * Returns a UTF-8 string from a resource at the specified index. + * The UTF-8 string may be returnable directly as a pointer, or + * it may need to be copied, or transformed from UTF-16 using u_strToUTF8() + * or equivalent. + * + * If forceCopy==true, then the string is always written to the dest buffer + * and dest is returned. + * + * If forceCopy==false, then the string is returned as a pointer if possible, + * without needing a dest buffer (it can be NULL). If the string needs to be + * copied or transformed, then it may be placed into dest at an arbitrary offset. + * + * If the string is to be written to dest, then U_BUFFER_OVERFLOW_ERROR and + * U_STRING_NOT_TERMINATED_WARNING are set if appropriate, as usual. + * + * If the string is transformed from UTF-16, then a conversion error may occur + * if an unpaired surrogate is encountered. If the function is successful, then + * the output UTF-8 string is always well-formed. + * + * @param resB Resource bundle. + * @param stringIndex An index to the wanted string. + * @param dest Destination buffer. Can be NULL only if capacity=*length==0. + * @param pLength Input: Capacity of destination buffer. + * Output: Actual length of the UTF-8 string, not counting the + * terminating NUL, even in case of U_BUFFER_OVERFLOW_ERROR. + * Can be NULL, meaning capacity=0 and the string length is not + * returned to the caller. + * @param forceCopy If true, then the output string will always be written to + * dest, with U_BUFFER_OVERFLOW_ERROR and + * U_STRING_NOT_TERMINATED_WARNING set if appropriate. + * If false, then the dest buffer may or may not contain a + * copy of the string. dest may or may not be modified. + * If a copy needs to be written, then the UErrorCode parameter + * indicates overflow etc. as usual. + * @param status Pointer to a standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return The pointer to the UTF-8 string. It may be dest, or at some offset + * from dest (only if !forceCopy), or in unrelated memory. + * Always NUL-terminated unless the string was written to dest and + * length==capacity (in which case U_STRING_NOT_TERMINATED_WARNING is set). + * + * @see ures_getStringByIndex + * @see u_strToUTF8 + * @stable ICU 3.6 + */ +U_CAPI const char * U_EXPORT2 +ures_getUTF8StringByIndex(const UResourceBundle *resB, + int32_t stringIndex, + char *dest, int32_t *pLength, + UBool forceCopy, + UErrorCode *status); + +/** + * Returns a resource in a given resource that has a given key. This procedure works only with table + * resources. Features a fill-in parameter. + * + * @param resourceBundle a resource + * @param key a key associated with the wanted resource + * @param fillIn if NULL a new UResourceBundle struct is allocated and must be closed by the caller. + * Alternatively, you can supply a struct to be filled by this function. + * @param status fills in the outgoing error code. + * @return a pointer to a UResourceBundle struct. If fill in param was NULL, caller must close it + * @stable ICU 2.0 + */ +U_CAPI UResourceBundle* U_EXPORT2 +ures_getByKey(const UResourceBundle *resourceBundle, + const char* key, + UResourceBundle *fillIn, + UErrorCode *status); + +/** + * Returns a string in a given resource that has a given key. This procedure works only with table + * resources. + * + * @param resB a resource + * @param key a key associated with the wanted string + * @param len fill in length of the string + * @param status fills in the outgoing error code. If an error occurred, we may return NULL, but don't + * count on it. Check status instead! + * @return a pointer to a zero-terminated UChar array which lives in a memory mapped/DLL file. + * @stable ICU 2.0 + */ +U_CAPI const UChar* U_EXPORT2 +ures_getStringByKey(const UResourceBundle *resB, + const char* key, + int32_t* len, + UErrorCode *status); + +/** + * Returns a UTF-8 string from a resource and a key. + * This function works only with table resources. + * + * The UTF-8 string may be returnable directly as a pointer, or + * it may need to be copied, or transformed from UTF-16 using u_strToUTF8() + * or equivalent. + * + * If forceCopy==true, then the string is always written to the dest buffer + * and dest is returned. + * + * If forceCopy==false, then the string is returned as a pointer if possible, + * without needing a dest buffer (it can be NULL). If the string needs to be + * copied or transformed, then it may be placed into dest at an arbitrary offset. + * + * If the string is to be written to dest, then U_BUFFER_OVERFLOW_ERROR and + * U_STRING_NOT_TERMINATED_WARNING are set if appropriate, as usual. + * + * If the string is transformed from UTF-16, then a conversion error may occur + * if an unpaired surrogate is encountered. If the function is successful, then + * the output UTF-8 string is always well-formed. + * + * @param resB Resource bundle. + * @param key A key associated with the wanted resource + * @param dest Destination buffer. Can be NULL only if capacity=*length==0. + * @param pLength Input: Capacity of destination buffer. + * Output: Actual length of the UTF-8 string, not counting the + * terminating NUL, even in case of U_BUFFER_OVERFLOW_ERROR. + * Can be NULL, meaning capacity=0 and the string length is not + * returned to the caller. + * @param forceCopy If true, then the output string will always be written to + * dest, with U_BUFFER_OVERFLOW_ERROR and + * U_STRING_NOT_TERMINATED_WARNING set if appropriate. + * If false, then the dest buffer may or may not contain a + * copy of the string. dest may or may not be modified. + * If a copy needs to be written, then the UErrorCode parameter + * indicates overflow etc. as usual. + * @param status Pointer to a standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return The pointer to the UTF-8 string. It may be dest, or at some offset + * from dest (only if !forceCopy), or in unrelated memory. + * Always NUL-terminated unless the string was written to dest and + * length==capacity (in which case U_STRING_NOT_TERMINATED_WARNING is set). + * + * @see ures_getStringByKey + * @see u_strToUTF8 + * @stable ICU 3.6 + */ +U_CAPI const char * U_EXPORT2 +ures_getUTF8StringByKey(const UResourceBundle *resB, + const char *key, + char *dest, int32_t *pLength, + UBool forceCopy, + UErrorCode *status); + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/unistr.h" + +U_NAMESPACE_BEGIN +/** + * Returns the string value from a string resource bundle. + * + * @param resB a resource, should have type URES_STRING + * @param status: fills in the outgoing error code + * could be U_MISSING_RESOURCE_ERROR if the key is not found + * could be a non-failing error + * e.g.: U_USING_FALLBACK_WARNING,U_USING_DEFAULT_WARNING + * @return The string value, or a bogus string if there is a failure UErrorCode. + * @stable ICU 2.0 + */ +inline UnicodeString +ures_getUnicodeString(const UResourceBundle *resB, UErrorCode* status) { + UnicodeString result; + int32_t len = 0; + const char16_t *r = ures_getString(resB, &len, status); + if(U_SUCCESS(*status)) { + result.setTo(true, r, len); + } else { + result.setToBogus(); + } + return result; +} + +/** + * Returns the next string in a resource, or an empty string if there are no more resources + * to iterate over. + * Use ures_getNextString() instead to distinguish between + * the end of the iteration and a real empty string value. + * + * @param resB a resource + * @param key fill in for key associated with this string + * @param status fills in the outgoing error code + * @return The string value, or a bogus string if there is a failure UErrorCode. + * @stable ICU 2.0 + */ +inline UnicodeString +ures_getNextUnicodeString(UResourceBundle *resB, const char ** key, UErrorCode* status) { + UnicodeString result; + int32_t len = 0; + const char16_t* r = ures_getNextString(resB, &len, key, status); + if(U_SUCCESS(*status)) { + result.setTo(true, r, len); + } else { + result.setToBogus(); + } + return result; +} + +/** + * Returns the string in a given resource array or table at the specified index. + * + * @param resB a resource + * @param indexS an index to the wanted string. + * @param status fills in the outgoing error code + * @return The string value, or a bogus string if there is a failure UErrorCode. + * @stable ICU 2.0 + */ +inline UnicodeString +ures_getUnicodeStringByIndex(const UResourceBundle *resB, int32_t indexS, UErrorCode* status) { + UnicodeString result; + int32_t len = 0; + const char16_t* r = ures_getStringByIndex(resB, indexS, &len, status); + if(U_SUCCESS(*status)) { + result.setTo(true, r, len); + } else { + result.setToBogus(); + } + return result; +} + +/** + * Returns a string in a resource that has a given key. + * This procedure works only with table resources. + * + * @param resB a resource + * @param key a key associated with the wanted string + * @param status fills in the outgoing error code + * @return The string value, or a bogus string if there is a failure UErrorCode. + * @stable ICU 2.0 + */ +inline UnicodeString +ures_getUnicodeStringByKey(const UResourceBundle *resB, const char* key, UErrorCode* status) { + UnicodeString result; + int32_t len = 0; + const char16_t* r = ures_getStringByKey(resB, key, &len, status); + if(U_SUCCESS(*status)) { + result.setTo(true, r, len); + } else { + result.setToBogus(); + } + return result; +} + +U_NAMESPACE_END + +#endif + +/** + * Create a string enumerator, owned by the caller, of all locales located within + * the specified resource tree. + * @param packageName name of the tree, such as (NULL) or U_ICUDATA_ALIAS or or "ICUDATA-coll" + * This call is similar to uloc_getAvailable(). + * @param status error code + * @stable ICU 3.2 + */ +U_CAPI UEnumeration* U_EXPORT2 +ures_openAvailableLocales(const char *packageName, UErrorCode *status); + + +#endif /*_URES*/ +/*eof*/ diff --git a/intl/icu/source/common/unicode/uscript.h b/intl/icu/source/common/unicode/uscript.h new file mode 100644 index 0000000000..dc97ab2ba5 --- /dev/null +++ b/intl/icu/source/common/unicode/uscript.h @@ -0,0 +1,724 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ********************************************************************** + * Copyright (C) 1997-2016, International Business Machines + * Corporation and others. All Rights Reserved. + ********************************************************************** + * + * File USCRIPT.H + * + * Modification History: + * + * Date Name Description + * 07/06/2001 Ram Creation. + ****************************************************************************** + */ + +#ifndef USCRIPT_H +#define USCRIPT_H +#include "unicode/utypes.h" + +/** + * \file + * \brief C API: Unicode Script Information + */ + +/** + * Constants for ISO 15924 script codes. + * + * The current set of script code constants supports at least all scripts + * that are encoded in the version of Unicode which ICU currently supports. + * The names of the constants are usually derived from the + * Unicode script property value aliases. + * See UAX #24 Unicode Script Property (http://www.unicode.org/reports/tr24/) + * and http://www.unicode.org/Public/UCD/latest/ucd/PropertyValueAliases.txt . + * + * In addition, constants for many ISO 15924 script codes + * are included, for use with language tags, CLDR data, and similar. + * Some of those codes are not used in the Unicode Character Database (UCD). + * For example, there are no characters that have a UCD script property value of + * Hans or Hant. All Han ideographs have the Hani script property value in Unicode. + * + * Private-use codes Qaaa..Qabx are not included, except as used in the UCD or in CLDR. + * + * Starting with ICU 55, script codes are only added when their scripts + * have been or will certainly be encoded in Unicode, + * and have been assigned Unicode script property value aliases, + * to ensure that their script names are stable and match the names of the constants. + * Script codes like Latf and Aran that are not subject to separate encoding + * may be added at any time. + * + * @stable ICU 2.2 + */ +typedef enum UScriptCode { + /* + * Note: UScriptCode constants and their ISO script code comments + * are parsed by preparseucd.py. + * It matches lines like + * USCRIPT_ = , / * * / + */ + + /** @stable ICU 2.2 */ + USCRIPT_INVALID_CODE = -1, + /** @stable ICU 2.2 */ + USCRIPT_COMMON = 0, /* Zyyy */ + /** @stable ICU 2.2 */ + USCRIPT_INHERITED = 1, /* Zinh */ /* "Code for inherited script", for non-spacing combining marks; also Qaai */ + /** @stable ICU 2.2 */ + USCRIPT_ARABIC = 2, /* Arab */ + /** @stable ICU 2.2 */ + USCRIPT_ARMENIAN = 3, /* Armn */ + /** @stable ICU 2.2 */ + USCRIPT_BENGALI = 4, /* Beng */ + /** @stable ICU 2.2 */ + USCRIPT_BOPOMOFO = 5, /* Bopo */ + /** @stable ICU 2.2 */ + USCRIPT_CHEROKEE = 6, /* Cher */ + /** @stable ICU 2.2 */ + USCRIPT_COPTIC = 7, /* Copt */ + /** @stable ICU 2.2 */ + USCRIPT_CYRILLIC = 8, /* Cyrl */ + /** @stable ICU 2.2 */ + USCRIPT_DESERET = 9, /* Dsrt */ + /** @stable ICU 2.2 */ + USCRIPT_DEVANAGARI = 10, /* Deva */ + /** @stable ICU 2.2 */ + USCRIPT_ETHIOPIC = 11, /* Ethi */ + /** @stable ICU 2.2 */ + USCRIPT_GEORGIAN = 12, /* Geor */ + /** @stable ICU 2.2 */ + USCRIPT_GOTHIC = 13, /* Goth */ + /** @stable ICU 2.2 */ + USCRIPT_GREEK = 14, /* Grek */ + /** @stable ICU 2.2 */ + USCRIPT_GUJARATI = 15, /* Gujr */ + /** @stable ICU 2.2 */ + USCRIPT_GURMUKHI = 16, /* Guru */ + /** @stable ICU 2.2 */ + USCRIPT_HAN = 17, /* Hani */ + /** @stable ICU 2.2 */ + USCRIPT_HANGUL = 18, /* Hang */ + /** @stable ICU 2.2 */ + USCRIPT_HEBREW = 19, /* Hebr */ + /** @stable ICU 2.2 */ + USCRIPT_HIRAGANA = 20, /* Hira */ + /** @stable ICU 2.2 */ + USCRIPT_KANNADA = 21, /* Knda */ + /** @stable ICU 2.2 */ + USCRIPT_KATAKANA = 22, /* Kana */ + /** @stable ICU 2.2 */ + USCRIPT_KHMER = 23, /* Khmr */ + /** @stable ICU 2.2 */ + USCRIPT_LAO = 24, /* Laoo */ + /** @stable ICU 2.2 */ + USCRIPT_LATIN = 25, /* Latn */ + /** @stable ICU 2.2 */ + USCRIPT_MALAYALAM = 26, /* Mlym */ + /** @stable ICU 2.2 */ + USCRIPT_MONGOLIAN = 27, /* Mong */ + /** @stable ICU 2.2 */ + USCRIPT_MYANMAR = 28, /* Mymr */ + /** @stable ICU 2.2 */ + USCRIPT_OGHAM = 29, /* Ogam */ + /** @stable ICU 2.2 */ + USCRIPT_OLD_ITALIC = 30, /* Ital */ + /** @stable ICU 2.2 */ + USCRIPT_ORIYA = 31, /* Orya */ + /** @stable ICU 2.2 */ + USCRIPT_RUNIC = 32, /* Runr */ + /** @stable ICU 2.2 */ + USCRIPT_SINHALA = 33, /* Sinh */ + /** @stable ICU 2.2 */ + USCRIPT_SYRIAC = 34, /* Syrc */ + /** @stable ICU 2.2 */ + USCRIPT_TAMIL = 35, /* Taml */ + /** @stable ICU 2.2 */ + USCRIPT_TELUGU = 36, /* Telu */ + /** @stable ICU 2.2 */ + USCRIPT_THAANA = 37, /* Thaa */ + /** @stable ICU 2.2 */ + USCRIPT_THAI = 38, /* Thai */ + /** @stable ICU 2.2 */ + USCRIPT_TIBETAN = 39, /* Tibt */ + /** Canadian_Aboriginal script. @stable ICU 2.6 */ + USCRIPT_CANADIAN_ABORIGINAL = 40, /* Cans */ + /** Canadian_Aboriginal script (alias). @stable ICU 2.2 */ + USCRIPT_UCAS = USCRIPT_CANADIAN_ABORIGINAL, + /** @stable ICU 2.2 */ + USCRIPT_YI = 41, /* Yiii */ + /* New scripts in Unicode 3.2 */ + /** @stable ICU 2.2 */ + USCRIPT_TAGALOG = 42, /* Tglg */ + /** @stable ICU 2.2 */ + USCRIPT_HANUNOO = 43, /* Hano */ + /** @stable ICU 2.2 */ + USCRIPT_BUHID = 44, /* Buhd */ + /** @stable ICU 2.2 */ + USCRIPT_TAGBANWA = 45, /* Tagb */ + + /* New scripts in Unicode 4 */ + /** @stable ICU 2.6 */ + USCRIPT_BRAILLE = 46, /* Brai */ + /** @stable ICU 2.6 */ + USCRIPT_CYPRIOT = 47, /* Cprt */ + /** @stable ICU 2.6 */ + USCRIPT_LIMBU = 48, /* Limb */ + /** @stable ICU 2.6 */ + USCRIPT_LINEAR_B = 49, /* Linb */ + /** @stable ICU 2.6 */ + USCRIPT_OSMANYA = 50, /* Osma */ + /** @stable ICU 2.6 */ + USCRIPT_SHAVIAN = 51, /* Shaw */ + /** @stable ICU 2.6 */ + USCRIPT_TAI_LE = 52, /* Tale */ + /** @stable ICU 2.6 */ + USCRIPT_UGARITIC = 53, /* Ugar */ + + /** New script code in Unicode 4.0.1 @stable ICU 3.0 */ + USCRIPT_KATAKANA_OR_HIRAGANA = 54,/*Hrkt */ + + /* New scripts in Unicode 4.1 */ + /** @stable ICU 3.4 */ + USCRIPT_BUGINESE = 55, /* Bugi */ + /** @stable ICU 3.4 */ + USCRIPT_GLAGOLITIC = 56, /* Glag */ + /** @stable ICU 3.4 */ + USCRIPT_KHAROSHTHI = 57, /* Khar */ + /** @stable ICU 3.4 */ + USCRIPT_SYLOTI_NAGRI = 58, /* Sylo */ + /** @stable ICU 3.4 */ + USCRIPT_NEW_TAI_LUE = 59, /* Talu */ + /** @stable ICU 3.4 */ + USCRIPT_TIFINAGH = 60, /* Tfng */ + /** @stable ICU 3.4 */ + USCRIPT_OLD_PERSIAN = 61, /* Xpeo */ + + /* New script codes from Unicode and ISO 15924 */ + /** @stable ICU 3.6 */ + USCRIPT_BALINESE = 62, /* Bali */ + /** @stable ICU 3.6 */ + USCRIPT_BATAK = 63, /* Batk */ + /** @stable ICU 3.6 */ + USCRIPT_BLISSYMBOLS = 64, /* Blis */ + /** @stable ICU 3.6 */ + USCRIPT_BRAHMI = 65, /* Brah */ + /** @stable ICU 3.6 */ + USCRIPT_CHAM = 66, /* Cham */ + /** @stable ICU 3.6 */ + USCRIPT_CIRTH = 67, /* Cirt */ + /** @stable ICU 3.6 */ + USCRIPT_OLD_CHURCH_SLAVONIC_CYRILLIC = 68, /* Cyrs */ + /** @stable ICU 3.6 */ + USCRIPT_DEMOTIC_EGYPTIAN = 69, /* Egyd */ + /** @stable ICU 3.6 */ + USCRIPT_HIERATIC_EGYPTIAN = 70, /* Egyh */ + /** @stable ICU 3.6 */ + USCRIPT_EGYPTIAN_HIEROGLYPHS = 71, /* Egyp */ + /** @stable ICU 3.6 */ + USCRIPT_KHUTSURI = 72, /* Geok */ + /** @stable ICU 3.6 */ + USCRIPT_SIMPLIFIED_HAN = 73, /* Hans */ + /** @stable ICU 3.6 */ + USCRIPT_TRADITIONAL_HAN = 74, /* Hant */ + /** @stable ICU 3.6 */ + USCRIPT_PAHAWH_HMONG = 75, /* Hmng */ + /** @stable ICU 3.6 */ + USCRIPT_OLD_HUNGARIAN = 76, /* Hung */ + /** @stable ICU 3.6 */ + USCRIPT_HARAPPAN_INDUS = 77, /* Inds */ + /** @stable ICU 3.6 */ + USCRIPT_JAVANESE = 78, /* Java */ + /** @stable ICU 3.6 */ + USCRIPT_KAYAH_LI = 79, /* Kali */ + /** @stable ICU 3.6 */ + USCRIPT_LATIN_FRAKTUR = 80, /* Latf */ + /** @stable ICU 3.6 */ + USCRIPT_LATIN_GAELIC = 81, /* Latg */ + /** @stable ICU 3.6 */ + USCRIPT_LEPCHA = 82, /* Lepc */ + /** @stable ICU 3.6 */ + USCRIPT_LINEAR_A = 83, /* Lina */ + /** @stable ICU 4.6 */ + USCRIPT_MANDAIC = 84, /* Mand */ + /** @stable ICU 3.6 */ + USCRIPT_MANDAEAN = USCRIPT_MANDAIC, + /** @stable ICU 3.6 */ + USCRIPT_MAYAN_HIEROGLYPHS = 85, /* Maya */ + /** @stable ICU 4.6 */ + USCRIPT_MEROITIC_HIEROGLYPHS = 86, /* Mero */ + /** @stable ICU 3.6 */ + USCRIPT_MEROITIC = USCRIPT_MEROITIC_HIEROGLYPHS, + /** @stable ICU 3.6 */ + USCRIPT_NKO = 87, /* Nkoo */ + /** @stable ICU 3.6 */ + USCRIPT_ORKHON = 88, /* Orkh */ + /** @stable ICU 3.6 */ + USCRIPT_OLD_PERMIC = 89, /* Perm */ + /** @stable ICU 3.6 */ + USCRIPT_PHAGS_PA = 90, /* Phag */ + /** @stable ICU 3.6 */ + USCRIPT_PHOENICIAN = 91, /* Phnx */ + /** @stable ICU 52 */ + USCRIPT_MIAO = 92, /* Plrd */ + /** @stable ICU 3.6 */ + USCRIPT_PHONETIC_POLLARD = USCRIPT_MIAO, + /** @stable ICU 3.6 */ + USCRIPT_RONGORONGO = 93, /* Roro */ + /** @stable ICU 3.6 */ + USCRIPT_SARATI = 94, /* Sara */ + /** @stable ICU 3.6 */ + USCRIPT_ESTRANGELO_SYRIAC = 95, /* Syre */ + /** @stable ICU 3.6 */ + USCRIPT_WESTERN_SYRIAC = 96, /* Syrj */ + /** @stable ICU 3.6 */ + USCRIPT_EASTERN_SYRIAC = 97, /* Syrn */ + /** @stable ICU 3.6 */ + USCRIPT_TENGWAR = 98, /* Teng */ + /** @stable ICU 3.6 */ + USCRIPT_VAI = 99, /* Vaii */ + /** @stable ICU 3.6 */ + USCRIPT_VISIBLE_SPEECH = 100,/* Visp */ + /** @stable ICU 3.6 */ + USCRIPT_CUNEIFORM = 101,/* Xsux */ + /** @stable ICU 3.6 */ + USCRIPT_UNWRITTEN_LANGUAGES = 102,/* Zxxx */ + /** @stable ICU 3.6 */ + USCRIPT_UNKNOWN = 103,/* Zzzz */ /* Unknown="Code for uncoded script", for unassigned code points */ + + /** @stable ICU 3.8 */ + USCRIPT_CARIAN = 104,/* Cari */ + /** @stable ICU 3.8 */ + USCRIPT_JAPANESE = 105,/* Jpan */ + /** @stable ICU 3.8 */ + USCRIPT_LANNA = 106,/* Lana */ + /** @stable ICU 3.8 */ + USCRIPT_LYCIAN = 107,/* Lyci */ + /** @stable ICU 3.8 */ + USCRIPT_LYDIAN = 108,/* Lydi */ + /** @stable ICU 3.8 */ + USCRIPT_OL_CHIKI = 109,/* Olck */ + /** @stable ICU 3.8 */ + USCRIPT_REJANG = 110,/* Rjng */ + /** @stable ICU 3.8 */ + USCRIPT_SAURASHTRA = 111,/* Saur */ + /** Sutton SignWriting @stable ICU 3.8 */ + USCRIPT_SIGN_WRITING = 112,/* Sgnw */ + /** @stable ICU 3.8 */ + USCRIPT_SUNDANESE = 113,/* Sund */ + /** @stable ICU 3.8 */ + USCRIPT_MOON = 114,/* Moon */ + /** @stable ICU 3.8 */ + USCRIPT_MEITEI_MAYEK = 115,/* Mtei */ + + /** @stable ICU 4.0 */ + USCRIPT_IMPERIAL_ARAMAIC = 116,/* Armi */ + /** @stable ICU 4.0 */ + USCRIPT_AVESTAN = 117,/* Avst */ + /** @stable ICU 4.0 */ + USCRIPT_CHAKMA = 118,/* Cakm */ + /** @stable ICU 4.0 */ + USCRIPT_KOREAN = 119,/* Kore */ + /** @stable ICU 4.0 */ + USCRIPT_KAITHI = 120,/* Kthi */ + /** @stable ICU 4.0 */ + USCRIPT_MANICHAEAN = 121,/* Mani */ + /** @stable ICU 4.0 */ + USCRIPT_INSCRIPTIONAL_PAHLAVI = 122,/* Phli */ + /** @stable ICU 4.0 */ + USCRIPT_PSALTER_PAHLAVI = 123,/* Phlp */ + /** @stable ICU 4.0 */ + USCRIPT_BOOK_PAHLAVI = 124,/* Phlv */ + /** @stable ICU 4.0 */ + USCRIPT_INSCRIPTIONAL_PARTHIAN = 125,/* Prti */ + /** @stable ICU 4.0 */ + USCRIPT_SAMARITAN = 126,/* Samr */ + /** @stable ICU 4.0 */ + USCRIPT_TAI_VIET = 127,/* Tavt */ + /** @stable ICU 4.0 */ + USCRIPT_MATHEMATICAL_NOTATION = 128,/* Zmth */ + /** @stable ICU 4.0 */ + USCRIPT_SYMBOLS = 129,/* Zsym */ + + /** @stable ICU 4.4 */ + USCRIPT_BAMUM = 130,/* Bamu */ + /** @stable ICU 4.4 */ + USCRIPT_LISU = 131,/* Lisu */ + /** @stable ICU 4.4 */ + USCRIPT_NAKHI_GEBA = 132,/* Nkgb */ + /** @stable ICU 4.4 */ + USCRIPT_OLD_SOUTH_ARABIAN = 133,/* Sarb */ + + /** @stable ICU 4.6 */ + USCRIPT_BASSA_VAH = 134,/* Bass */ + /** @stable ICU 54 */ + USCRIPT_DUPLOYAN = 135,/* Dupl */ +#ifndef U_HIDE_DEPRECATED_API + /** @deprecated ICU 54 Typo, use USCRIPT_DUPLOYAN */ + USCRIPT_DUPLOYAN_SHORTAND = USCRIPT_DUPLOYAN, +#endif /* U_HIDE_DEPRECATED_API */ + /** @stable ICU 4.6 */ + USCRIPT_ELBASAN = 136,/* Elba */ + /** @stable ICU 4.6 */ + USCRIPT_GRANTHA = 137,/* Gran */ + /** @stable ICU 4.6 */ + USCRIPT_KPELLE = 138,/* Kpel */ + /** @stable ICU 4.6 */ + USCRIPT_LOMA = 139,/* Loma */ + /** Mende Kikakui @stable ICU 4.6 */ + USCRIPT_MENDE = 140,/* Mend */ + /** @stable ICU 4.6 */ + USCRIPT_MEROITIC_CURSIVE = 141,/* Merc */ + /** @stable ICU 4.6 */ + USCRIPT_OLD_NORTH_ARABIAN = 142,/* Narb */ + /** @stable ICU 4.6 */ + USCRIPT_NABATAEAN = 143,/* Nbat */ + /** @stable ICU 4.6 */ + USCRIPT_PALMYRENE = 144,/* Palm */ + /** @stable ICU 54 */ + USCRIPT_KHUDAWADI = 145,/* Sind */ + /** @stable ICU 4.6 */ + USCRIPT_SINDHI = USCRIPT_KHUDAWADI, + /** @stable ICU 4.6 */ + USCRIPT_WARANG_CITI = 146,/* Wara */ + + /** @stable ICU 4.8 */ + USCRIPT_AFAKA = 147,/* Afak */ + /** @stable ICU 4.8 */ + USCRIPT_JURCHEN = 148,/* Jurc */ + /** @stable ICU 4.8 */ + USCRIPT_MRO = 149,/* Mroo */ + /** @stable ICU 4.8 */ + USCRIPT_NUSHU = 150,/* Nshu */ + /** @stable ICU 4.8 */ + USCRIPT_SHARADA = 151,/* Shrd */ + /** @stable ICU 4.8 */ + USCRIPT_SORA_SOMPENG = 152,/* Sora */ + /** @stable ICU 4.8 */ + USCRIPT_TAKRI = 153,/* Takr */ + /** @stable ICU 4.8 */ + USCRIPT_TANGUT = 154,/* Tang */ + /** @stable ICU 4.8 */ + USCRIPT_WOLEAI = 155,/* Wole */ + + /** @stable ICU 49 */ + USCRIPT_ANATOLIAN_HIEROGLYPHS = 156,/* Hluw */ + /** @stable ICU 49 */ + USCRIPT_KHOJKI = 157,/* Khoj */ + /** @stable ICU 49 */ + USCRIPT_TIRHUTA = 158,/* Tirh */ + + /** @stable ICU 52 */ + USCRIPT_CAUCASIAN_ALBANIAN = 159,/* Aghb */ + /** @stable ICU 52 */ + USCRIPT_MAHAJANI = 160,/* Mahj */ + + /** @stable ICU 54 */ + USCRIPT_AHOM = 161,/* Ahom */ + /** @stable ICU 54 */ + USCRIPT_HATRAN = 162,/* Hatr */ + /** @stable ICU 54 */ + USCRIPT_MODI = 163,/* Modi */ + /** @stable ICU 54 */ + USCRIPT_MULTANI = 164,/* Mult */ + /** @stable ICU 54 */ + USCRIPT_PAU_CIN_HAU = 165,/* Pauc */ + /** @stable ICU 54 */ + USCRIPT_SIDDHAM = 166,/* Sidd */ + + /** @stable ICU 58 */ + USCRIPT_ADLAM = 167,/* Adlm */ + /** @stable ICU 58 */ + USCRIPT_BHAIKSUKI = 168,/* Bhks */ + /** @stable ICU 58 */ + USCRIPT_MARCHEN = 169,/* Marc */ + /** @stable ICU 58 */ + USCRIPT_NEWA = 170,/* Newa */ + /** @stable ICU 58 */ + USCRIPT_OSAGE = 171,/* Osge */ + + /** @stable ICU 58 */ + USCRIPT_HAN_WITH_BOPOMOFO = 172,/* Hanb */ + /** @stable ICU 58 */ + USCRIPT_JAMO = 173,/* Jamo */ + /** @stable ICU 58 */ + USCRIPT_SYMBOLS_EMOJI = 174,/* Zsye */ + + /** @stable ICU 60 */ + USCRIPT_MASARAM_GONDI = 175,/* Gonm */ + /** @stable ICU 60 */ + USCRIPT_SOYOMBO = 176,/* Soyo */ + /** @stable ICU 60 */ + USCRIPT_ZANABAZAR_SQUARE = 177,/* Zanb */ + + /** @stable ICU 62 */ + USCRIPT_DOGRA = 178,/* Dogr */ + /** @stable ICU 62 */ + USCRIPT_GUNJALA_GONDI = 179,/* Gong */ + /** @stable ICU 62 */ + USCRIPT_MAKASAR = 180,/* Maka */ + /** @stable ICU 62 */ + USCRIPT_MEDEFAIDRIN = 181,/* Medf */ + /** @stable ICU 62 */ + USCRIPT_HANIFI_ROHINGYA = 182,/* Rohg */ + /** @stable ICU 62 */ + USCRIPT_SOGDIAN = 183,/* Sogd */ + /** @stable ICU 62 */ + USCRIPT_OLD_SOGDIAN = 184,/* Sogo */ + + /** @stable ICU 64 */ + USCRIPT_ELYMAIC = 185,/* Elym */ + /** @stable ICU 64 */ + USCRIPT_NYIAKENG_PUACHUE_HMONG = 186,/* Hmnp */ + /** @stable ICU 64 */ + USCRIPT_NANDINAGARI = 187,/* Nand */ + /** @stable ICU 64 */ + USCRIPT_WANCHO = 188,/* Wcho */ + + /** @stable ICU 66 */ + USCRIPT_CHORASMIAN = 189,/* Chrs */ + /** @stable ICU 66 */ + USCRIPT_DIVES_AKURU = 190,/* Diak */ + /** @stable ICU 66 */ + USCRIPT_KHITAN_SMALL_SCRIPT = 191,/* Kits */ + /** @stable ICU 66 */ + USCRIPT_YEZIDI = 192,/* Yezi */ + + /** @stable ICU 70 */ + USCRIPT_CYPRO_MINOAN = 193,/* Cpmn */ + /** @stable ICU 70 */ + USCRIPT_OLD_UYGHUR = 194,/* Ougr */ + /** @stable ICU 70 */ + USCRIPT_TANGSA = 195,/* Tnsa */ + /** @stable ICU 70 */ + USCRIPT_TOTO = 196,/* Toto */ + /** @stable ICU 70 */ + USCRIPT_VITHKUQI = 197,/* Vith */ + + /** @stable ICU 72 */ + USCRIPT_KAWI = 198,/* Kawi */ + /** @stable ICU 72 */ + USCRIPT_NAG_MUNDARI = 199,/* Nagm */ + +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the highest normal UScriptCode value. + * The highest value is available via u_getIntPropertyMaxValue(UCHAR_SCRIPT). + * + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + USCRIPT_CODE_LIMIT = 200 +#endif // U_HIDE_DEPRECATED_API +} UScriptCode; + +/** + * Gets the script codes associated with the given locale or ISO 15924 abbreviation or name. + * Fills in USCRIPT_MALAYALAM given "Malayam" OR "Mlym". + * Fills in USCRIPT_LATIN given "en" OR "en_US" + * If the required capacity is greater than the capacity of the destination buffer, + * then the error code is set to U_BUFFER_OVERFLOW_ERROR and the required capacity is returned. + * + *

Note: To search by short or long script alias only, use + * u_getPropertyValueEnum(UCHAR_SCRIPT, alias) instead. That does + * a fast lookup with no access of the locale data. + * + * @param nameOrAbbrOrLocale name of the script, as given in + * PropertyValueAliases.txt, or ISO 15924 code or locale + * @param fillIn the UScriptCode buffer to fill in the script code + * @param capacity the capacity (size) of UScriptCode buffer passed in. + * @param err the error status code. + * @return The number of script codes filled in the buffer passed in + * @stable ICU 2.4 + */ +U_CAPI int32_t U_EXPORT2 +uscript_getCode(const char* nameOrAbbrOrLocale,UScriptCode* fillIn,int32_t capacity,UErrorCode *err); + +/** + * Returns the long Unicode script name, if there is one. + * Otherwise returns the 4-letter ISO 15924 script code. + * Returns "Malayam" given USCRIPT_MALAYALAM. + * + * @param scriptCode UScriptCode enum + * @return long script name as given in PropertyValueAliases.txt, or the 4-letter code, + * or NULL if scriptCode is invalid + * @stable ICU 2.4 + */ +U_CAPI const char* U_EXPORT2 +uscript_getName(UScriptCode scriptCode); + +/** + * Returns the 4-letter ISO 15924 script code, + * which is the same as the short Unicode script name if Unicode has names for the script. + * Returns "Mlym" given USCRIPT_MALAYALAM. + * + * @param scriptCode UScriptCode enum + * @return short script name (4-letter code), or NULL if scriptCode is invalid + * @stable ICU 2.4 + */ +U_CAPI const char* U_EXPORT2 +uscript_getShortName(UScriptCode scriptCode); + +/** + * Gets the script code associated with the given codepoint. + * Returns USCRIPT_MALAYALAM given 0x0D02 + * @param codepoint UChar32 codepoint + * @param err the error status code. + * @return The UScriptCode, or 0 if codepoint is invalid + * @stable ICU 2.4 + */ +U_CAPI UScriptCode U_EXPORT2 +uscript_getScript(UChar32 codepoint, UErrorCode *err); + +/** + * Do the Script_Extensions of code point c contain script sc? + * If c does not have explicit Script_Extensions, then this tests whether + * c has the Script property value sc. + * + * Some characters are commonly used in multiple scripts. + * For more information, see UAX #24: http://www.unicode.org/reports/tr24/. + * @param c code point + * @param sc script code + * @return true if sc is in Script_Extensions(c) + * @stable ICU 49 + */ +U_CAPI UBool U_EXPORT2 +uscript_hasScript(UChar32 c, UScriptCode sc); + +/** + * Writes code point c's Script_Extensions as a list of UScriptCode values + * to the output scripts array and returns the number of script codes. + * - If c does have Script_Extensions, then the Script property value + * (normally Common or Inherited) is not included. + * - If c does not have Script_Extensions, then the one Script code is written to the output array. + * - If c is not a valid code point, then the one USCRIPT_UNKNOWN code is written. + * In other words, if the return value is 1, + * then the output array contains exactly c's single Script code. + * If the return value is n>=2, then the output array contains c's n Script_Extensions script codes. + * + * Some characters are commonly used in multiple scripts. + * For more information, see UAX #24: http://www.unicode.org/reports/tr24/. + * + * If there are more than capacity script codes to be written, then + * U_BUFFER_OVERFLOW_ERROR is set and the number of Script_Extensions is returned. + * (Usual ICU buffer handling behavior.) + * + * @param c code point + * @param scripts output script code array + * @param capacity capacity of the scripts array + * @param errorCode Standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return number of script codes in c's Script_Extensions, or 1 for the single Script value, + * written to scripts unless U_BUFFER_OVERFLOW_ERROR indicates insufficient capacity + * @stable ICU 49 + */ +U_CAPI int32_t U_EXPORT2 +uscript_getScriptExtensions(UChar32 c, + UScriptCode *scripts, int32_t capacity, + UErrorCode *errorCode); + +/** + * Script usage constants. + * See UAX #31 Unicode Identifier and Pattern Syntax. + * http://www.unicode.org/reports/tr31/#Table_Candidate_Characters_for_Exclusion_from_Identifiers + * + * @stable ICU 51 + */ +typedef enum UScriptUsage { + /** Not encoded in Unicode. @stable ICU 51 */ + USCRIPT_USAGE_NOT_ENCODED, + /** Unknown script usage. @stable ICU 51 */ + USCRIPT_USAGE_UNKNOWN, + /** Candidate for Exclusion from Identifiers. @stable ICU 51 */ + USCRIPT_USAGE_EXCLUDED, + /** Limited Use script. @stable ICU 51 */ + USCRIPT_USAGE_LIMITED_USE, + /** Aspirational Use script. @stable ICU 51 */ + USCRIPT_USAGE_ASPIRATIONAL, + /** Recommended script. @stable ICU 51 */ + USCRIPT_USAGE_RECOMMENDED +} UScriptUsage; + +/** + * Writes the script sample character string. + * This string normally consists of one code point but might be longer. + * The string is empty if the script is not encoded. + * + * @param script script code + * @param dest output string array + * @param capacity number of UChars in the dest array + * @param pErrorCode standard ICU in/out error code, must pass U_SUCCESS() on input + * @return the string length, even if U_BUFFER_OVERFLOW_ERROR + * @stable ICU 51 + */ +U_CAPI int32_t U_EXPORT2 +uscript_getSampleString(UScriptCode script, UChar *dest, int32_t capacity, UErrorCode *pErrorCode); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN +class UnicodeString; +U_NAMESPACE_END + +/** + * Returns the script sample character string. + * This string normally consists of one code point but might be longer. + * The string is empty if the script is not encoded. + * + * @param script script code + * @return the sample character string + * @stable ICU 51 + */ +U_COMMON_API icu::UnicodeString U_EXPORT2 +uscript_getSampleUnicodeString(UScriptCode script); + +#endif + +/** + * Returns the script usage according to UAX #31 Unicode Identifier and Pattern Syntax. + * Returns USCRIPT_USAGE_NOT_ENCODED if the script is not encoded in Unicode. + * + * @param script script code + * @return script usage + * @see UScriptUsage + * @stable ICU 51 + */ +U_CAPI UScriptUsage U_EXPORT2 +uscript_getUsage(UScriptCode script); + +/** + * Returns true if the script is written right-to-left. + * For example, Arab and Hebr. + * + * @param script script code + * @return true if the script is right-to-left + * @stable ICU 51 + */ +U_CAPI UBool U_EXPORT2 +uscript_isRightToLeft(UScriptCode script); + +/** + * Returns true if the script allows line breaks between letters (excluding hyphenation). + * Such a script typically requires dictionary-based line breaking. + * For example, Hani and Thai. + * + * @param script script code + * @return true if the script allows line breaks between letters + * @stable ICU 51 + */ +U_CAPI UBool U_EXPORT2 +uscript_breaksBetweenLetters(UScriptCode script); + +/** + * Returns true if in modern (or most recent) usage of the script case distinctions are customary. + * For example, Latn and Cyrl. + * + * @param script script code + * @return true if the script is cased + * @stable ICU 51 + */ +U_CAPI UBool U_EXPORT2 +uscript_isCased(UScriptCode script); + +#endif diff --git a/intl/icu/source/common/unicode/uset.h b/intl/icu/source/common/unicode/uset.h new file mode 100644 index 0000000000..ee4e0036d2 --- /dev/null +++ b/intl/icu/source/common/unicode/uset.h @@ -0,0 +1,1290 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2002-2014, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: uset.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2002mar07 +* created by: Markus W. Scherer +* +* C version of UnicodeSet. +*/ + + +/** + * \file + * \brief C API: Unicode Set + * + *

This is a C wrapper around the C++ UnicodeSet class.

+ */ + +#ifndef __USET_H__ +#define __USET_H__ + +#include "unicode/utypes.h" +#include "unicode/uchar.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +#ifndef USET_DEFINED + +#ifndef U_IN_DOXYGEN +#define USET_DEFINED +#endif +/** + * USet is the C API type corresponding to C++ class UnicodeSet. + * Use the uset_* API to manipulate. Create with + * uset_open*, and destroy with uset_close. + * @stable ICU 2.4 + */ +typedef struct USet USet; +#endif + +/** + * Bitmask values to be passed to uset_openPatternOptions() or + * uset_applyPattern() taking an option parameter. + * + * Use at most one of USET_CASE_INSENSITIVE, USET_ADD_CASE_MAPPINGS, USET_SIMPLE_CASE_INSENSITIVE. + * These case options are mutually exclusive. + * + * Undefined options bits are ignored, and reserved for future use. + * + * @stable ICU 2.4 + */ +enum { + /** + * Ignore white space within patterns unless quoted or escaped. + * @stable ICU 2.4 + */ + USET_IGNORE_SPACE = 1, + + /** + * Enable case insensitive matching. E.g., "[ab]" with this flag + * will match 'a', 'A', 'b', and 'B'. "[^ab]" with this flag will + * match all except 'a', 'A', 'b', and 'B'. This performs a full + * closure over case mappings, e.g. 'ſ' (U+017F long s) for 's'. + * + * The resulting set is a superset of the input for the code points but + * not for the strings. + * It performs a case mapping closure of the code points and adds + * full case folding strings for the code points, and reduces strings of + * the original set to their full case folding equivalents. + * + * This is designed for case-insensitive matches, for example + * in regular expressions. The full code point case closure allows checking of + * an input character directly against the closure set. + * Strings are matched by comparing the case-folded form from the closure + * set with an incremental case folding of the string in question. + * + * The closure set will also contain single code points if the original + * set contained case-equivalent strings (like U+00DF for "ss" or "Ss" etc.). + * This is not necessary (that is, redundant) for the above matching method + * but results in the same closure sets regardless of whether the original + * set contained the code point or a string. + * + * @stable ICU 2.4 + */ + USET_CASE_INSENSITIVE = 2, + + /** + * Adds all case mappings for each element in the set. + * This adds the full lower-, title-, and uppercase mappings as well as the full case folding + * of each existing element in the set. + * + * Unlike the “case insensitive” options, this does not perform a closure. + * For example, it does not add 'ſ' (U+017F long s) for 's', + * 'K' (U+212A Kelvin sign) for 'k', or replace set strings by their case-folded versions. + * + * @stable ICU 3.2 + */ + USET_ADD_CASE_MAPPINGS = 4, + +#ifndef U_HIDE_DRAFT_API + /** + * Enable case insensitive matching. + * Same as USET_CASE_INSENSITIVE but using only Simple_Case_Folding (scf) mappings, + * which map each code point to one code point, + * not full Case_Folding (cf) mappings, which map some code points to multiple code points. + * + * This is designed for case-insensitive matches, for example in certain + * regular expression implementations where only Simple_Case_Folding mappings are used, + * such as in ECMAScript (JavaScript) regular expressions. + * + * @draft ICU 73 + */ + USET_SIMPLE_CASE_INSENSITIVE = 6 +#endif // U_HIDE_DRAFT_API +}; + +/** + * Argument values for whether span() and similar functions continue while + * the current character is contained vs. not contained in the set. + * + * The functionality is straightforward for sets with only single code points, + * without strings (which is the common case): + * - USET_SPAN_CONTAINED and USET_SPAN_SIMPLE work the same. + * - USET_SPAN_CONTAINED and USET_SPAN_SIMPLE are inverses of USET_SPAN_NOT_CONTAINED. + * - span() and spanBack() partition any string the same way when + * alternating between span(USET_SPAN_NOT_CONTAINED) and + * span(either "contained" condition). + * - Using a complemented (inverted) set and the opposite span conditions + * yields the same results. + * + * When a set contains multi-code point strings, then these statements may not + * be true, depending on the strings in the set (for example, whether they + * overlap with each other) and the string that is processed. + * For a set with strings: + * - The complement of the set contains the opposite set of code points, + * but the same set of strings. + * Therefore, complementing both the set and the span conditions + * may yield different results. + * - When starting spans at different positions in a string + * (span(s, ...) vs. span(s+1, ...)) the ends of the spans may be different + * because a set string may start before the later position. + * - span(USET_SPAN_SIMPLE) may be shorter than + * span(USET_SPAN_CONTAINED) because it will not recursively try + * all possible paths. + * For example, with a set which contains the three strings "xy", "xya" and "ax", + * span("xyax", USET_SPAN_CONTAINED) will return 4 but + * span("xyax", USET_SPAN_SIMPLE) will return 3. + * span(USET_SPAN_SIMPLE) will never be longer than + * span(USET_SPAN_CONTAINED). + * - With either "contained" condition, span() and spanBack() may partition + * a string in different ways. + * For example, with a set which contains the two strings "ab" and "ba", + * and when processing the string "aba", + * span() will yield contained/not-contained boundaries of { 0, 2, 3 } + * while spanBack() will yield boundaries of { 0, 1, 3 }. + * + * Note: If it is important to get the same boundaries whether iterating forward + * or backward through a string, then either only span() should be used and + * the boundaries cached for backward operation, or an ICU BreakIterator + * could be used. + * + * Note: Unpaired surrogates are treated like surrogate code points. + * Similarly, set strings match only on code point boundaries, + * never in the middle of a surrogate pair. + * Illegal UTF-8 sequences are treated like U+FFFD. + * When processing UTF-8 strings, malformed set strings + * (strings with unpaired surrogates which cannot be converted to UTF-8) + * are ignored. + * + * @stable ICU 3.8 + */ +typedef enum USetSpanCondition { + /** + * Continues a span() while there is no set element at the current position. + * Increments by one code point at a time. + * Stops before the first set element (character or string). + * (For code points only, this is like while contains(current)==false). + * + * When span() returns, the substring between where it started and the position + * it returned consists only of characters that are not in the set, + * and none of its strings overlap with the span. + * + * @stable ICU 3.8 + */ + USET_SPAN_NOT_CONTAINED = 0, + /** + * Spans the longest substring that is a concatenation of set elements (characters or strings). + * (For characters only, this is like while contains(current)==true). + * + * When span() returns, the substring between where it started and the position + * it returned consists only of set elements (characters or strings) that are in the set. + * + * If a set contains strings, then the span will be the longest substring for which there + * exists at least one non-overlapping concatenation of set elements (characters or strings). + * This is equivalent to a POSIX regular expression for (OR of each set element)*. + * (Java/ICU/Perl regex stops at the first match of an OR.) + * + * @stable ICU 3.8 + */ + USET_SPAN_CONTAINED = 1, + /** + * Continues a span() while there is a set element at the current position. + * Increments by the longest matching element at each position. + * (For characters only, this is like while contains(current)==true). + * + * When span() returns, the substring between where it started and the position + * it returned consists only of set elements (characters or strings) that are in the set. + * + * If a set only contains single characters, then this is the same + * as USET_SPAN_CONTAINED. + * + * If a set contains strings, then the span will be the longest substring + * with a match at each position with the longest single set element (character or string). + * + * Use this span condition together with other longest-match algorithms, + * such as ICU converters (ucnv_getUnicodeSet()). + * + * @stable ICU 3.8 + */ + USET_SPAN_SIMPLE = 2, +#ifndef U_HIDE_DEPRECATED_API + /** + * One more than the last span condition. + * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420. + */ + USET_SPAN_CONDITION_COUNT +#endif // U_HIDE_DEPRECATED_API +} USetSpanCondition; + +enum { + /** + * Capacity of USerializedSet::staticArray. + * Enough for any single-code point set. + * Also provides padding for nice sizeof(USerializedSet). + * @stable ICU 2.4 + */ + USET_SERIALIZED_STATIC_ARRAY_CAPACITY=8 +}; + +/** + * A serialized form of a Unicode set. Limited manipulations are + * possible directly on a serialized set. See below. + * @stable ICU 2.4 + */ +typedef struct USerializedSet { + /** + * The serialized Unicode Set. + * @stable ICU 2.4 + */ + const uint16_t *array; + /** + * The length of the array that contains BMP characters. + * @stable ICU 2.4 + */ + int32_t bmpLength; + /** + * The total length of the array. + * @stable ICU 2.4 + */ + int32_t length; + /** + * A small buffer for the array to reduce memory allocations. + * @stable ICU 2.4 + */ + uint16_t staticArray[USET_SERIALIZED_STATIC_ARRAY_CAPACITY]; +} USerializedSet; + +/********************************************************************* + * USet API + *********************************************************************/ + +/** + * Create an empty USet object. + * Equivalent to uset_open(1, 0). + * @return a newly created USet. The caller must call uset_close() on + * it when done. + * @stable ICU 4.2 + */ +U_CAPI USet* U_EXPORT2 +uset_openEmpty(void); + +/** + * Creates a USet object that contains the range of characters + * start..end, inclusive. If start > end + * then an empty set is created (same as using uset_openEmpty()). + * @param start first character of the range, inclusive + * @param end last character of the range, inclusive + * @return a newly created USet. The caller must call uset_close() on + * it when done. + * @stable ICU 2.4 + */ +U_CAPI USet* U_EXPORT2 +uset_open(UChar32 start, UChar32 end); + +/** + * Creates a set from the given pattern. See the UnicodeSet class + * description for the syntax of the pattern language. + * @param pattern a string specifying what characters are in the set + * @param patternLength the length of the pattern, or -1 if null + * terminated + * @param ec the error code + * @stable ICU 2.4 + */ +U_CAPI USet* U_EXPORT2 +uset_openPattern(const UChar* pattern, int32_t patternLength, + UErrorCode* ec); + +/** + * Creates a set from the given pattern. See the UnicodeSet class + * description for the syntax of the pattern language. + * @param pattern a string specifying what characters are in the set + * @param patternLength the length of the pattern, or -1 if null + * terminated + * @param options bitmask for options to apply to the pattern. + * Valid options are USET_IGNORE_SPACE and + * at most one of USET_CASE_INSENSITIVE, USET_ADD_CASE_MAPPINGS, USET_SIMPLE_CASE_INSENSITIVE. + * These case options are mutually exclusive. + * @param ec the error code + * @stable ICU 2.4 + */ +U_CAPI USet* U_EXPORT2 +uset_openPatternOptions(const UChar* pattern, int32_t patternLength, + uint32_t options, + UErrorCode* ec); + +/** + * Disposes of the storage used by a USet object. This function should + * be called exactly once for objects returned by uset_open(). + * @param set the object to dispose of + * @stable ICU 2.4 + */ +U_CAPI void U_EXPORT2 +uset_close(USet* set); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUSetPointer + * "Smart pointer" class, closes a USet via uset_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUSetPointer, USet, uset_close); + +U_NAMESPACE_END + +#endif + +/** + * Returns a copy of this object. + * If this set is frozen, then the clone will be frozen as well. + * Use uset_cloneAsThawed() for a mutable clone of a frozen set. + * @param set the original set + * @return the newly allocated copy of the set + * @see uset_cloneAsThawed + * @stable ICU 3.8 + */ +U_CAPI USet * U_EXPORT2 +uset_clone(const USet *set); + +/** + * Determines whether the set has been frozen (made immutable) or not. + * See the ICU4J Freezable interface for details. + * @param set the set + * @return true/false for whether the set has been frozen + * @see uset_freeze + * @see uset_cloneAsThawed + * @stable ICU 3.8 + */ +U_CAPI UBool U_EXPORT2 +uset_isFrozen(const USet *set); + +/** + * Freeze the set (make it immutable). + * Once frozen, it cannot be unfrozen and is therefore thread-safe + * until it is deleted. + * See the ICU4J Freezable interface for details. + * Freezing the set may also make some operations faster, for example + * uset_contains() and uset_span(). + * A frozen set will not be modified. (It remains frozen.) + * @param set the set + * @return the same set, now frozen + * @see uset_isFrozen + * @see uset_cloneAsThawed + * @stable ICU 3.8 + */ +U_CAPI void U_EXPORT2 +uset_freeze(USet *set); + +/** + * Clone the set and make the clone mutable. + * See the ICU4J Freezable interface for details. + * @param set the set + * @return the mutable clone + * @see uset_freeze + * @see uset_isFrozen + * @see uset_clone + * @stable ICU 3.8 + */ +U_CAPI USet * U_EXPORT2 +uset_cloneAsThawed(const USet *set); + +/** + * Causes the USet object to represent the range start - end. + * If start > end then this USet is set to an empty range. + * A frozen set will not be modified. + * @param set the object to set to the given range + * @param start first character in the set, inclusive + * @param end last character in the set, inclusive + * @stable ICU 3.2 + */ +U_CAPI void U_EXPORT2 +uset_set(USet* set, + UChar32 start, UChar32 end); + +/** + * Modifies the set to represent the set specified by the given + * pattern. See the UnicodeSet class description for the syntax of + * the pattern language. See also the User Guide chapter about UnicodeSet. + * Empties the set passed before applying the pattern. + * A frozen set will not be modified. + * @param set The set to which the pattern is to be applied. + * @param pattern A pointer to UChar string specifying what characters are in the set. + * The character at pattern[0] must be a '['. + * @param patternLength The length of the UChar string. -1 if NUL terminated. + * @param options A bitmask for options to apply to the pattern. + * Valid options are USET_IGNORE_SPACE and + * at most one of USET_CASE_INSENSITIVE, USET_ADD_CASE_MAPPINGS, + * USET_SIMPLE_CASE_INSENSITIVE. + * These case options are mutually exclusive. + * @param status Returns an error if the pattern cannot be parsed. + * @return Upon successful parse, the value is either + * the index of the character after the closing ']' + * of the parsed pattern. + * If the status code indicates failure, then the return value + * is the index of the error in the source. + * + * @stable ICU 2.8 + */ +U_CAPI int32_t U_EXPORT2 +uset_applyPattern(USet *set, + const UChar *pattern, int32_t patternLength, + uint32_t options, + UErrorCode *status); + +/** + * Modifies the set to contain those code points which have the given value + * for the given binary or enumerated property, as returned by + * u_getIntPropertyValue. Prior contents of this set are lost. + * A frozen set will not be modified. + * + * @param set the object to contain the code points defined by the property + * + * @param prop a property in the range UCHAR_BIN_START..UCHAR_BIN_LIMIT-1 + * or UCHAR_INT_START..UCHAR_INT_LIMIT-1 + * or UCHAR_MASK_START..UCHAR_MASK_LIMIT-1. + * + * @param value a value in the range u_getIntPropertyMinValue(prop).. + * u_getIntPropertyMaxValue(prop), with one exception. If prop is + * UCHAR_GENERAL_CATEGORY_MASK, then value should not be a UCharCategory, but + * rather a mask value produced by U_GET_GC_MASK(). This allows grouped + * categories such as [:L:] to be represented. + * + * @param ec error code input/output parameter + * + * @stable ICU 3.2 + */ +U_CAPI void U_EXPORT2 +uset_applyIntPropertyValue(USet* set, + UProperty prop, int32_t value, UErrorCode* ec); + +/** + * Modifies the set to contain those code points which have the + * given value for the given property. Prior contents of this + * set are lost. + * A frozen set will not be modified. + * + * @param set the object to contain the code points defined by the given + * property and value alias + * + * @param prop a string specifying a property alias, either short or long. + * The name is matched loosely. See PropertyAliases.txt for names and a + * description of loose matching. If the value string is empty, then this + * string is interpreted as either a General_Category value alias, a Script + * value alias, a binary property alias, or a special ID. Special IDs are + * matched loosely and correspond to the following sets: + * + * "ANY" = [\\u0000-\\U0010FFFF], + * "ASCII" = [\\u0000-\\u007F], + * "Assigned" = [:^Cn:]. + * + * @param propLength the length of the prop, or -1 if NULL + * + * @param value a string specifying a value alias, either short or long. + * The name is matched loosely. See PropertyValueAliases.txt for names + * and a description of loose matching. In addition to aliases listed, + * numeric values and canonical combining classes may be expressed + * numerically, e.g., ("nv", "0.5") or ("ccc", "220"). The value string + * may also be empty. + * + * @param valueLength the length of the value, or -1 if NULL + * + * @param ec error code input/output parameter + * + * @stable ICU 3.2 + */ +U_CAPI void U_EXPORT2 +uset_applyPropertyAlias(USet* set, + const UChar *prop, int32_t propLength, + const UChar *value, int32_t valueLength, + UErrorCode* ec); + +/** + * Return true if the given position, in the given pattern, appears + * to be the start of a UnicodeSet pattern. + * + * @param pattern a string specifying the pattern + * @param patternLength the length of the pattern, or -1 if NULL + * @param pos the given position + * @stable ICU 3.2 + */ +U_CAPI UBool U_EXPORT2 +uset_resemblesPattern(const UChar *pattern, int32_t patternLength, + int32_t pos); + +/** + * Returns a string representation of this set. If the result of + * calling this function is passed to a uset_openPattern(), it + * will produce another set that is equal to this one. + * @param set the set + * @param result the string to receive the rules, may be NULL + * @param resultCapacity the capacity of result, may be 0 if result is NULL + * @param escapeUnprintable if true then convert unprintable + * character to their hex escape representations, \\uxxxx or + * \\Uxxxxxxxx. Unprintable characters are those other than + * U+000A, U+0020..U+007E. + * @param ec error code. + * @return length of string, possibly larger than resultCapacity + * @stable ICU 2.4 + */ +U_CAPI int32_t U_EXPORT2 +uset_toPattern(const USet* set, + UChar* result, int32_t resultCapacity, + UBool escapeUnprintable, + UErrorCode* ec); + +/** + * Adds the given character to the given USet. After this call, + * uset_contains(set, c) will return true. + * A frozen set will not be modified. + * @param set the object to which to add the character + * @param c the character to add + * @stable ICU 2.4 + */ +U_CAPI void U_EXPORT2 +uset_add(USet* set, UChar32 c); + +/** + * Adds all of the elements in the specified set to this set if + * they're not already present. This operation effectively + * modifies this set so that its value is the union of the two + * sets. The behavior of this operation is unspecified if the specified + * collection is modified while the operation is in progress. + * A frozen set will not be modified. + * + * @param set the object to which to add the set + * @param additionalSet the source set whose elements are to be added to this set. + * @stable ICU 2.6 + */ +U_CAPI void U_EXPORT2 +uset_addAll(USet* set, const USet *additionalSet); + +/** + * Adds the given range of characters to the given USet. After this call, + * uset_contains(set, start, end) will return true. + * A frozen set will not be modified. + * @param set the object to which to add the character + * @param start the first character of the range to add, inclusive + * @param end the last character of the range to add, inclusive + * @stable ICU 2.2 + */ +U_CAPI void U_EXPORT2 +uset_addRange(USet* set, UChar32 start, UChar32 end); + +/** + * Adds the given string to the given USet. After this call, + * uset_containsString(set, str, strLen) will return true. + * A frozen set will not be modified. + * @param set the object to which to add the character + * @param str the string to add + * @param strLen the length of the string or -1 if null terminated. + * @stable ICU 2.4 + */ +U_CAPI void U_EXPORT2 +uset_addString(USet* set, const UChar* str, int32_t strLen); + +/** + * Adds each of the characters in this string to the set. Note: "ch" => {"c", "h"} + * If this set already contains any particular character, it has no effect on that character. + * A frozen set will not be modified. + * @param set the object to which to add the character + * @param str the source string + * @param strLen the length of the string or -1 if null terminated. + * @stable ICU 3.4 + */ +U_CAPI void U_EXPORT2 +uset_addAllCodePoints(USet* set, const UChar *str, int32_t strLen); + +/** + * Removes the given character from the given USet. After this call, + * uset_contains(set, c) will return false. + * A frozen set will not be modified. + * @param set the object from which to remove the character + * @param c the character to remove + * @stable ICU 2.4 + */ +U_CAPI void U_EXPORT2 +uset_remove(USet* set, UChar32 c); + +/** + * Removes the given range of characters from the given USet. After this call, + * uset_contains(set, start, end) will return false. + * A frozen set will not be modified. + * @param set the object to which to add the character + * @param start the first character of the range to remove, inclusive + * @param end the last character of the range to remove, inclusive + * @stable ICU 2.2 + */ +U_CAPI void U_EXPORT2 +uset_removeRange(USet* set, UChar32 start, UChar32 end); + +/** + * Removes the given string to the given USet. After this call, + * uset_containsString(set, str, strLen) will return false. + * A frozen set will not be modified. + * @param set the object to which to add the character + * @param str the string to remove + * @param strLen the length of the string or -1 if null terminated. + * @stable ICU 2.4 + */ +U_CAPI void U_EXPORT2 +uset_removeString(USet* set, const UChar* str, int32_t strLen); + +/** + * Removes EACH of the characters in this string. Note: "ch" == {"c", "h"} + * A frozen set will not be modified. + * + * @param set the object to be modified + * @param str the string + * @param length the length of the string, or -1 if NUL-terminated + * @stable ICU 69 + */ +U_CAPI void U_EXPORT2 +uset_removeAllCodePoints(USet *set, const UChar *str, int32_t length); + +/** + * Removes from this set all of its elements that are contained in the + * specified set. This operation effectively modifies this + * set so that its value is the asymmetric set difference of + * the two sets. + * A frozen set will not be modified. + * @param set the object from which the elements are to be removed + * @param removeSet the object that defines which elements will be + * removed from this set + * @stable ICU 3.2 + */ +U_CAPI void U_EXPORT2 +uset_removeAll(USet* set, const USet* removeSet); + +/** + * Retain only the elements in this set that are contained in the + * specified range. If start > end then an empty range is + * retained, leaving the set empty. This is equivalent to + * a boolean logic AND, or a set INTERSECTION. + * A frozen set will not be modified. + * + * @param set the object for which to retain only the specified range + * @param start first character, inclusive, of range + * @param end last character, inclusive, of range + * @stable ICU 3.2 + */ +U_CAPI void U_EXPORT2 +uset_retain(USet* set, UChar32 start, UChar32 end); + +/** + * Retains only the specified string from this set if it is present. + * Upon return this set will be empty if it did not contain s, or + * will only contain s if it did contain s. + * A frozen set will not be modified. + * + * @param set the object to be modified + * @param str the string + * @param length the length of the string, or -1 if NUL-terminated + * @stable ICU 69 + */ +U_CAPI void U_EXPORT2 +uset_retainString(USet *set, const UChar *str, int32_t length); + +/** + * Retains EACH of the characters in this string. Note: "ch" == {"c", "h"} + * A frozen set will not be modified. + * + * @param set the object to be modified + * @param str the string + * @param length the length of the string, or -1 if NUL-terminated + * @stable ICU 69 + */ +U_CAPI void U_EXPORT2 +uset_retainAllCodePoints(USet *set, const UChar *str, int32_t length); + +/** + * Retains only the elements in this set that are contained in the + * specified set. In other words, removes from this set all of + * its elements that are not contained in the specified set. This + * operation effectively modifies this set so that its value is + * the intersection of the two sets. + * A frozen set will not be modified. + * + * @param set the object on which to perform the retain + * @param retain set that defines which elements this set will retain + * @stable ICU 3.2 + */ +U_CAPI void U_EXPORT2 +uset_retainAll(USet* set, const USet* retain); + +/** + * Reallocate this objects internal structures to take up the least + * possible space, without changing this object's value. + * A frozen set will not be modified. + * + * @param set the object on which to perform the compact + * @stable ICU 3.2 + */ +U_CAPI void U_EXPORT2 +uset_compact(USet* set); + +/** + * This is equivalent to + * uset_complementRange(set, 0, 0x10FFFF). + * + * Note: This performs a symmetric difference with all code points + * and thus retains all multicharacter strings. + * In order to achieve a “code point complement” (all code points minus this set), + * the easiest is to uset_complement(set); uset_removeAllStrings(set);. + * + * A frozen set will not be modified. + * @param set the set + * @stable ICU 2.4 + */ +U_CAPI void U_EXPORT2 +uset_complement(USet* set); + +/** + * Complements the specified range in this set. Any character in + * the range will be removed if it is in this set, or will be + * added if it is not in this set. If start > end + * then an empty range is complemented, leaving the set unchanged. + * This is equivalent to a boolean logic XOR. + * A frozen set will not be modified. + * + * @param set the object to be modified + * @param start first character, inclusive, of range + * @param end last character, inclusive, of range + * @stable ICU 69 + */ +U_CAPI void U_EXPORT2 +uset_complementRange(USet *set, UChar32 start, UChar32 end); + +/** + * Complements the specified string in this set. + * The string will be removed if it is in this set, or will be added if it is not in this set. + * A frozen set will not be modified. + * + * @param set the object to be modified + * @param str the string + * @param length the length of the string, or -1 if NUL-terminated + * @stable ICU 69 + */ +U_CAPI void U_EXPORT2 +uset_complementString(USet *set, const UChar *str, int32_t length); + +/** + * Complements EACH of the characters in this string. Note: "ch" == {"c", "h"} + * A frozen set will not be modified. + * + * @param set the object to be modified + * @param str the string + * @param length the length of the string, or -1 if NUL-terminated + * @stable ICU 69 + */ +U_CAPI void U_EXPORT2 +uset_complementAllCodePoints(USet *set, const UChar *str, int32_t length); + +/** + * Complements in this set all elements contained in the specified + * set. Any character in the other set will be removed if it is + * in this set, or will be added if it is not in this set. + * A frozen set will not be modified. + * + * @param set the set with which to complement + * @param complement set that defines which elements will be xor'ed + * from this set. + * @stable ICU 3.2 + */ +U_CAPI void U_EXPORT2 +uset_complementAll(USet* set, const USet* complement); + +/** + * Removes all of the elements from this set. This set will be + * empty after this call returns. + * A frozen set will not be modified. + * @param set the set + * @stable ICU 2.4 + */ +U_CAPI void U_EXPORT2 +uset_clear(USet* set); + +/** + * Close this set over the given attribute. For the attribute + * USET_CASE_INSENSITIVE, the result is to modify this set so that: + * + * 1. For each character or string 'a' in this set, all strings or + * characters 'b' such that foldCase(a) == foldCase(b) are added + * to this set. + * + * 2. For each string 'e' in the resulting set, if e != + * foldCase(e), 'e' will be removed. + * + * Example: [aq\\u00DF{Bc}{bC}{Fi}] => [aAqQ\\u00DF\\uFB01{ss}{bc}{fi}] + * + * (Here foldCase(x) refers to the operation u_strFoldCase, and a + * == b denotes that the contents are the same, not pointer + * comparison.) + * + * A frozen set will not be modified. + * + * @param set the set + * + * @param attributes bitmask for attributes to close over. + * Valid options: + * At most one of USET_CASE_INSENSITIVE, USET_ADD_CASE_MAPPINGS, USET_SIMPLE_CASE_INSENSITIVE. + * These case options are mutually exclusive. + * Unrelated options bits are ignored. + * @stable ICU 4.2 + */ +U_CAPI void U_EXPORT2 +uset_closeOver(USet* set, int32_t attributes); + +/** + * Remove all strings from this set. + * + * @param set the set + * @stable ICU 4.2 + */ +U_CAPI void U_EXPORT2 +uset_removeAllStrings(USet* set); + +/** + * Returns true if the given USet contains no characters and no + * strings. + * @param set the set + * @return true if set is empty + * @stable ICU 2.4 + */ +U_CAPI UBool U_EXPORT2 +uset_isEmpty(const USet* set); + +/** + * @param set the set + * @return true if this set contains multi-character strings or the empty string. + * @stable ICU 70 + */ +U_CAPI UBool U_EXPORT2 +uset_hasStrings(const USet *set); + +/** + * Returns true if the given USet contains the given character. + * This function works faster with a frozen set. + * @param set the set + * @param c The codepoint to check for within the set + * @return true if set contains c + * @stable ICU 2.4 + */ +U_CAPI UBool U_EXPORT2 +uset_contains(const USet* set, UChar32 c); + +/** + * Returns true if the given USet contains all characters c + * where start <= c && c <= end. + * @param set the set + * @param start the first character of the range to test, inclusive + * @param end the last character of the range to test, inclusive + * @return true if set contains the range + * @stable ICU 2.2 + */ +U_CAPI UBool U_EXPORT2 +uset_containsRange(const USet* set, UChar32 start, UChar32 end); + +/** + * Returns true if the given USet contains the given string. + * @param set the set + * @param str the string + * @param strLen the length of the string or -1 if null terminated. + * @return true if set contains str + * @stable ICU 2.4 + */ +U_CAPI UBool U_EXPORT2 +uset_containsString(const USet* set, const UChar* str, int32_t strLen); + +/** + * Returns the index of the given character within this set, where + * the set is ordered by ascending code point. If the character + * is not in this set, return -1. The inverse of this method is + * charAt(). + * @param set the set + * @param c the character to obtain the index for + * @return an index from 0..size()-1, or -1 + * @stable ICU 3.2 + */ +U_CAPI int32_t U_EXPORT2 +uset_indexOf(const USet* set, UChar32 c); + +/** + * Returns the character at the given index within this set, where + * the set is ordered by ascending code point. If the index is + * out of range for characters, returns (UChar32)-1. + * The inverse of this method is indexOf(). + * + * For iteration, this is slower than uset_getRangeCount()/uset_getItemCount() + * with uset_getItem(), because for each call it skips linearly over index + * characters in the ranges. + * + * @param set the set + * @param charIndex an index from 0..size()-1 to obtain the char for + * @return the character at the given index, or (UChar32)-1. + * @stable ICU 3.2 + */ +U_CAPI UChar32 U_EXPORT2 +uset_charAt(const USet* set, int32_t charIndex); + +/** + * Returns the number of characters and strings contained in this set. + * The last (uset_getItemCount() - uset_getRangeCount()) items are strings. + * + * This is slower than uset_getRangeCount() and uset_getItemCount() because + * it counts the code points of all ranges. + * + * @param set the set + * @return a non-negative integer counting the characters and strings + * contained in set + * @stable ICU 2.4 + * @see uset_getRangeCount + */ +U_CAPI int32_t U_EXPORT2 +uset_size(const USet* set); + +/** + * @param set the set + * @return the number of ranges in this set. + * @stable ICU 70 + * @see uset_getItemCount + * @see uset_getItem + * @see uset_size + */ +U_CAPI int32_t U_EXPORT2 +uset_getRangeCount(const USet *set); + +/** + * Returns the number of items in this set. An item is either a range + * of characters or a single multicharacter string. + * @param set the set + * @return a non-negative integer counting the character ranges + * and/or strings contained in set + * @stable ICU 2.4 + */ +U_CAPI int32_t U_EXPORT2 +uset_getItemCount(const USet* set); + +/** + * Returns an item of this set. An item is either a range of + * characters or a single multicharacter string (which can be the empty string). + * + * If itemIndex is less than uset_getRangeCount(), then this function returns 0, + * and the range is *start..*end. + * + * If itemIndex is at least uset_getRangeCount() and less than uset_getItemCount(), then + * this function copies the string into str[strCapacity] and + * returns the length of the string (0 for the empty string). + * + * If itemIndex is out of range, then this function returns -1. + * + * Note that 0 is returned for each range as well as for the empty string. + * + * @param set the set + * @param itemIndex a non-negative integer in the range 0..uset_getItemCount(set)-1 + * @param start pointer to variable to receive first character in range, inclusive; + * can be NULL for a string item + * @param end pointer to variable to receive last character in range, inclusive; + * can be NULL for a string item + * @param str buffer to receive the string, may be NULL + * @param strCapacity capacity of str, or 0 if str is NULL + * @param ec error code; U_INDEX_OUTOFBOUNDS_ERROR if the itemIndex is out of range + * @return the length of the string (0 or >= 2), or 0 if the item is a range, + * or -1 if the itemIndex is out of range + * @stable ICU 2.4 + */ +U_CAPI int32_t U_EXPORT2 +uset_getItem(const USet* set, int32_t itemIndex, + UChar32* start, UChar32* end, + UChar* str, int32_t strCapacity, + UErrorCode* ec); + +/** + * Returns true if set1 contains all the characters and strings + * of set2. It answers the question, 'Is set1 a superset of set2?' + * @param set1 set to be checked for containment + * @param set2 set to be checked for containment + * @return true if the test condition is met + * @stable ICU 3.2 + */ +U_CAPI UBool U_EXPORT2 +uset_containsAll(const USet* set1, const USet* set2); + +/** + * Returns true if this set contains all the characters + * of the given string. This is does not check containment of grapheme + * clusters, like uset_containsString. + * @param set set of characters to be checked for containment + * @param str string containing codepoints to be checked for containment + * @param strLen the length of the string or -1 if null terminated. + * @return true if the test condition is met + * @stable ICU 3.4 + */ +U_CAPI UBool U_EXPORT2 +uset_containsAllCodePoints(const USet* set, const UChar *str, int32_t strLen); + +/** + * Returns true if set1 contains none of the characters and strings + * of set2. It answers the question, 'Is set1 a disjoint set of set2?' + * @param set1 set to be checked for containment + * @param set2 set to be checked for containment + * @return true if the test condition is met + * @stable ICU 3.2 + */ +U_CAPI UBool U_EXPORT2 +uset_containsNone(const USet* set1, const USet* set2); + +/** + * Returns true if set1 contains some of the characters and strings + * of set2. It answers the question, 'Does set1 and set2 have an intersection?' + * @param set1 set to be checked for containment + * @param set2 set to be checked for containment + * @return true if the test condition is met + * @stable ICU 3.2 + */ +U_CAPI UBool U_EXPORT2 +uset_containsSome(const USet* set1, const USet* set2); + +/** + * Returns the length of the initial substring of the input string which + * consists only of characters and strings that are contained in this set + * (USET_SPAN_CONTAINED, USET_SPAN_SIMPLE), + * or only of characters and strings that are not contained + * in this set (USET_SPAN_NOT_CONTAINED). + * See USetSpanCondition for details. + * Similar to the strspn() C library function. + * Unpaired surrogates are treated according to contains() of their surrogate code points. + * This function works faster with a frozen set and with a non-negative string length argument. + * @param set the set + * @param s start of the string + * @param length of the string; can be -1 for NUL-terminated + * @param spanCondition specifies the containment condition + * @return the length of the initial substring according to the spanCondition; + * 0 if the start of the string does not fit the spanCondition + * @stable ICU 3.8 + * @see USetSpanCondition + */ +U_CAPI int32_t U_EXPORT2 +uset_span(const USet *set, const UChar *s, int32_t length, USetSpanCondition spanCondition); + +/** + * Returns the start of the trailing substring of the input string which + * consists only of characters and strings that are contained in this set + * (USET_SPAN_CONTAINED, USET_SPAN_SIMPLE), + * or only of characters and strings that are not contained + * in this set (USET_SPAN_NOT_CONTAINED). + * See USetSpanCondition for details. + * Unpaired surrogates are treated according to contains() of their surrogate code points. + * This function works faster with a frozen set and with a non-negative string length argument. + * @param set the set + * @param s start of the string + * @param length of the string; can be -1 for NUL-terminated + * @param spanCondition specifies the containment condition + * @return the start of the trailing substring according to the spanCondition; + * the string length if the end of the string does not fit the spanCondition + * @stable ICU 3.8 + * @see USetSpanCondition + */ +U_CAPI int32_t U_EXPORT2 +uset_spanBack(const USet *set, const UChar *s, int32_t length, USetSpanCondition spanCondition); + +/** + * Returns the length of the initial substring of the input string which + * consists only of characters and strings that are contained in this set + * (USET_SPAN_CONTAINED, USET_SPAN_SIMPLE), + * or only of characters and strings that are not contained + * in this set (USET_SPAN_NOT_CONTAINED). + * See USetSpanCondition for details. + * Similar to the strspn() C library function. + * Malformed byte sequences are treated according to contains(0xfffd). + * This function works faster with a frozen set and with a non-negative string length argument. + * @param set the set + * @param s start of the string (UTF-8) + * @param length of the string; can be -1 for NUL-terminated + * @param spanCondition specifies the containment condition + * @return the length of the initial substring according to the spanCondition; + * 0 if the start of the string does not fit the spanCondition + * @stable ICU 3.8 + * @see USetSpanCondition + */ +U_CAPI int32_t U_EXPORT2 +uset_spanUTF8(const USet *set, const char *s, int32_t length, USetSpanCondition spanCondition); + +/** + * Returns the start of the trailing substring of the input string which + * consists only of characters and strings that are contained in this set + * (USET_SPAN_CONTAINED, USET_SPAN_SIMPLE), + * or only of characters and strings that are not contained + * in this set (USET_SPAN_NOT_CONTAINED). + * See USetSpanCondition for details. + * Malformed byte sequences are treated according to contains(0xfffd). + * This function works faster with a frozen set and with a non-negative string length argument. + * @param set the set + * @param s start of the string (UTF-8) + * @param length of the string; can be -1 for NUL-terminated + * @param spanCondition specifies the containment condition + * @return the start of the trailing substring according to the spanCondition; + * the string length if the end of the string does not fit the spanCondition + * @stable ICU 3.8 + * @see USetSpanCondition + */ +U_CAPI int32_t U_EXPORT2 +uset_spanBackUTF8(const USet *set, const char *s, int32_t length, USetSpanCondition spanCondition); + +/** + * Returns true if set1 contains all of the characters and strings + * of set2, and vis versa. It answers the question, 'Is set1 equal to set2?' + * @param set1 set to be checked for containment + * @param set2 set to be checked for containment + * @return true if the test condition is met + * @stable ICU 3.2 + */ +U_CAPI UBool U_EXPORT2 +uset_equals(const USet* set1, const USet* set2); + +/********************************************************************* + * Serialized set API + *********************************************************************/ + +/** + * Serializes this set into an array of 16-bit integers. Serialization + * (currently) only records the characters in the set; multicharacter + * strings are ignored. + * + * The array + * has following format (each line is one 16-bit integer): + * + * length = (n+2*m) | (m!=0?0x8000:0) + * bmpLength = n; present if m!=0 + * bmp[0] + * bmp[1] + * ... + * bmp[n-1] + * supp-high[0] + * supp-low[0] + * supp-high[1] + * supp-low[1] + * ... + * supp-high[m-1] + * supp-low[m-1] + * + * The array starts with a header. After the header are n bmp + * code points, then m supplementary code points. Either n or m + * or both may be zero. n+2*m is always <= 0x7FFF. + * + * If there are no supplementary characters (if m==0) then the + * header is one 16-bit integer, 'length', with value n. + * + * If there are supplementary characters (if m!=0) then the header + * is two 16-bit integers. The first, 'length', has value + * (n+2*m)|0x8000. The second, 'bmpLength', has value n. + * + * After the header the code points are stored in ascending order. + * Supplementary code points are stored as most significant 16 + * bits followed by least significant 16 bits. + * + * @param set the set + * @param dest pointer to buffer of destCapacity 16-bit integers. + * May be NULL only if destCapacity is zero. + * @param destCapacity size of dest, or zero. Must not be negative. + * @param pErrorCode pointer to the error code. Will be set to + * U_INDEX_OUTOFBOUNDS_ERROR if n+2*m > 0x7FFF. Will be set to + * U_BUFFER_OVERFLOW_ERROR if n+2*m+(m!=0?2:1) > destCapacity. + * @return the total length of the serialized format, including + * the header, that is, n+2*m+(m!=0?2:1), or 0 on error other + * than U_BUFFER_OVERFLOW_ERROR. + * @stable ICU 2.4 + */ +U_CAPI int32_t U_EXPORT2 +uset_serialize(const USet* set, uint16_t* dest, int32_t destCapacity, UErrorCode* pErrorCode); + +/** + * Given a serialized array, fill in the given serialized set object. + * @param fillSet pointer to result + * @param src pointer to start of array + * @param srcLength length of array + * @return true if the given array is valid, otherwise false + * @stable ICU 2.4 + */ +U_CAPI UBool U_EXPORT2 +uset_getSerializedSet(USerializedSet* fillSet, const uint16_t* src, int32_t srcLength); + +/** + * Set the USerializedSet to contain the given character (and nothing + * else). + * @param fillSet pointer to result + * @param c The codepoint to set + * @stable ICU 2.4 + */ +U_CAPI void U_EXPORT2 +uset_setSerializedToOne(USerializedSet* fillSet, UChar32 c); + +/** + * Returns true if the given USerializedSet contains the given + * character. + * @param set the serialized set + * @param c The codepoint to check for within the set + * @return true if set contains c + * @stable ICU 2.4 + */ +U_CAPI UBool U_EXPORT2 +uset_serializedContains(const USerializedSet* set, UChar32 c); + +/** + * Returns the number of disjoint ranges of characters contained in + * the given serialized set. Ignores any strings contained in the + * set. + * @param set the serialized set + * @return a non-negative integer counting the character ranges + * contained in set + * @stable ICU 2.4 + */ +U_CAPI int32_t U_EXPORT2 +uset_getSerializedRangeCount(const USerializedSet* set); + +/** + * Returns a range of characters contained in the given serialized + * set. + * @param set the serialized set + * @param rangeIndex a non-negative integer in the range 0.. + * uset_getSerializedRangeCount(set)-1 + * @param pStart pointer to variable to receive first character + * in range, inclusive + * @param pEnd pointer to variable to receive last character in range, + * inclusive + * @return true if rangeIndex is valid, otherwise false + * @stable ICU 2.4 + */ +U_CAPI UBool U_EXPORT2 +uset_getSerializedRange(const USerializedSet* set, int32_t rangeIndex, + UChar32* pStart, UChar32* pEnd); + +#endif diff --git a/intl/icu/source/common/unicode/usetiter.h b/intl/icu/source/common/unicode/usetiter.h new file mode 100644 index 0000000000..3168d3b0f6 --- /dev/null +++ b/intl/icu/source/common/unicode/usetiter.h @@ -0,0 +1,323 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (c) 2002-2014, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +*/ +#ifndef USETITER_H +#define USETITER_H + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API + +#include "unicode/uobject.h" +#include "unicode/unistr.h" + +/** + * \file + * \brief C++ API: UnicodeSetIterator iterates over the contents of a UnicodeSet. + */ + +U_NAMESPACE_BEGIN + +class UnicodeSet; +class UnicodeString; + +/** + * + * UnicodeSetIterator iterates over the contents of a UnicodeSet. It + * iterates over either code points or code point ranges. After all + * code points or ranges have been returned, it returns the + * multicharacter strings of the UnicodeSet, if any. + * + * This class is not intended for public subclassing. + * + *

To iterate over code points and strings, use a loop like this: + * 

+ * UnicodeSetIterator it(set);
+ * while (it.next()) {
+ *     processItem(it.getString());
+ * }
+ *
+ *

Each item in the set is accessed as a string. Set elements + * consisting of single code points are returned as strings containing + * just the one code point. + * + *

To iterate over code point ranges, instead of individual code points, + * use a loop like this: + * 

+ * UnicodeSetIterator it(set);
+ * while (it.nextRange()) {
+ *   if (it.isString()) {
+ *     processString(it.getString());
+ *   } else {
+ *     processCodepointRange(it.getCodepoint(), it.getCodepointEnd());
+ *   }
+ * }
+ *
+ * + * To iterate over only the strings, start with skipToStrings(). + * + * @author M. Davis + * @stable ICU 2.4 + */ +class U_COMMON_API UnicodeSetIterator final : public UObject { + /** + * Value of codepoint if the iterator points to a string. + * If codepoint == IS_STRING, then examine + * string for the current iteration result. + */ + enum { IS_STRING = -1 }; + + /** + * Current code point, or the special value IS_STRING, if + * the iterator points to a string. + */ + UChar32 codepoint; + + /** + * When iterating over ranges using nextRange(), + * codepointEnd contains the inclusive end of the + * iteration range, if codepoint != IS_STRING. If + * iterating over code points using next(), or if + * codepoint == IS_STRING, then the value of + * codepointEnd is undefined. + */ + UChar32 codepointEnd; + + /** + * If codepoint == IS_STRING, then string points + * to the current string. If codepoint != IS_STRING, the + * value of string is undefined. + */ + const UnicodeString* string; + + public: + + /** + * Create an iterator over the given set. The iterator is valid + * only so long as set is valid. + * @param set set to iterate over + * @stable ICU 2.4 + */ + UnicodeSetIterator(const UnicodeSet& set); + + /** + * Create an iterator over nothing. next() and + * nextRange() return false. This is a convenience + * constructor allowing the target to be set later. + * @stable ICU 2.4 + */ + UnicodeSetIterator(); + + /** + * Destructor. + * @stable ICU 2.4 + */ + virtual ~UnicodeSetIterator(); + + /** + * Returns true if the current element is a string. If so, the + * caller can retrieve it with getString(). If this + * method returns false, the current element is a code point or + * code point range, depending on whether next() or + * nextRange() was called. + * Elements of types string and codepoint can both be retrieved + * with the function getString(). + * Elements of type codepoint can also be retrieved with + * getCodepoint(). + * For ranges, getCodepoint() returns the starting codepoint + * of the range, and getCodepointEnd() returns the end + * of the range. + * @stable ICU 2.4 + */ + inline UBool isString() const; + + /** + * Returns the current code point, if isString() returned + * false. Otherwise returns an undefined result. + * @stable ICU 2.4 + */ + inline UChar32 getCodepoint() const; + + /** + * Returns the end of the current code point range, if + * isString() returned false and nextRange() was + * called. Otherwise returns an undefined result. + * @stable ICU 2.4 + */ + inline UChar32 getCodepointEnd() const; + + /** + * Returns the current string, if isString() returned + * true. If the current iteration item is a code point, a UnicodeString + * containing that single code point is returned. + * + * Ownership of the returned string remains with the iterator. + * The string is guaranteed to remain valid only until the iterator is + * advanced to the next item, or until the iterator is deleted. + * + * @stable ICU 2.4 + */ + const UnicodeString& getString(); + + /** + * Skips over the remaining code points/ranges, if any. + * A following call to next() or nextRange() will yield a string, if there is one. + * No-op if next() would return false, or if it would yield a string anyway. + * + * @return *this + * @stable ICU 70 + * @see UnicodeSet#strings() + */ + inline UnicodeSetIterator &skipToStrings() { + // Finish code point/range iteration. + range = endRange; + endElement = -1; + nextElement = 0; + return *this; + } + + /** + * Advances the iteration position to the next element in the set, + * which can be either a single code point or a string. + * If there are no more elements in the set, return false. + * + *

+ * If isString() == true, the value is a + * string, otherwise the value is a + * single code point. Elements of either type can be retrieved + * with the function getString(), while elements of + * consisting of a single code point can be retrieved with + * getCodepoint() + * + *

The order of iteration is all code points in sorted order, + * followed by all strings sorted order. Do not mix + * calls to next() and nextRange() without + * calling reset() between them. The results of doing so + * are undefined. + * + * @return true if there was another element in the set. + * @stable ICU 2.4 + */ + UBool next(); + + /** + * Returns the next element in the set, either a code point range + * or a string. If there are no more elements in the set, return + * false. If isString() == true, the value is a + * string and can be accessed with getString(). Otherwise the value is a + * range of one or more code points from getCodepoint() to + * getCodepointeEnd() inclusive. + * + *

The order of iteration is all code points ranges in sorted + * order, followed by all strings sorted order. Ranges are + * disjoint and non-contiguous. The value returned from getString() + * is undefined unless isString() == true. Do not mix calls to + * next() and nextRange() without calling + * reset() between them. The results of doing so are + * undefined. + * + * @return true if there was another element in the set. + * @stable ICU 2.4 + */ + UBool nextRange(); + + /** + * Sets this iterator to visit the elements of the given set and + * resets it to the start of that set. The iterator is valid only + * so long as set is valid. + * @param set the set to iterate over. + * @stable ICU 2.4 + */ + void reset(const UnicodeSet& set); + + /** + * Resets this iterator to the start of the set. + * @stable ICU 2.4 + */ + void reset(); + + /** + * ICU "poor man's RTTI", returns a UClassID for this class. + * + * @stable ICU 2.4 + */ + static UClassID U_EXPORT2 getStaticClassID(); + + /** + * ICU "poor man's RTTI", returns a UClassID for the actual class. + * + * @stable ICU 2.4 + */ + virtual UClassID getDynamicClassID() const override; + + // ======================= PRIVATES =========================== + +private: + + // endElement and nextElements are really UChar32's, but we keep + // them as signed int32_t's so we can do comparisons with + // endElement set to -1. Leave them as int32_t's. + /** The set + */ + const UnicodeSet* set; + /** End range + */ + int32_t endRange; + /** Range + */ + int32_t range; + /** End element + */ + int32_t endElement; + /** Next element + */ + int32_t nextElement; + /** Next string + */ + int32_t nextString; + /** String count + */ + int32_t stringCount; + + /** + * Points to the string to use when the caller asks for a + * string and the current iteration item is a code point, not a string. + */ + UnicodeString *cpString; + + /** Copy constructor. Disallowed. + */ + UnicodeSetIterator(const UnicodeSetIterator&) = delete; + + /** Assignment operator. Disallowed. + */ + UnicodeSetIterator& operator=(const UnicodeSetIterator&) = delete; + + /** Load range + */ + void loadRange(int32_t range); +}; + +inline UBool UnicodeSetIterator::isString() const { + return codepoint < 0; +} + +inline UChar32 UnicodeSetIterator::getCodepoint() const { + return codepoint; +} + +inline UChar32 UnicodeSetIterator::getCodepointEnd() const { + return codepointEnd; +} + + +U_NAMESPACE_END + +#endif /* U_SHOW_CPLUSPLUS_API */ + +#endif diff --git a/intl/icu/source/common/unicode/ushape.h b/intl/icu/source/common/unicode/ushape.h new file mode 100644 index 0000000000..14371edc8f --- /dev/null +++ b/intl/icu/source/common/unicode/ushape.h @@ -0,0 +1,476 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +****************************************************************************** +* +* Copyright (C) 2000-2012, International Business Machines +* Corporation and others. All Rights Reserved. +* +****************************************************************************** +* file name: ushape.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2000jun29 +* created by: Markus W. Scherer +*/ + +#ifndef __USHAPE_H__ +#define __USHAPE_H__ + +#include "unicode/utypes.h" + +/** + * \file + * \brief C API: Arabic shaping + * + */ + +/** + * Shape Arabic text on a character basis. + * + *

This function performs basic operations for "shaping" Arabic text. It is most + * useful for use with legacy data formats and legacy display technology + * (simple terminals). All operations are performed on Unicode characters.

+ * + *

Text-based shaping means that some character code points in the text are + * replaced by others depending on the context. It transforms one kind of text + * into another. In comparison, modern displays for Arabic text select + * appropriate, context-dependent font glyphs for each text element, which means + * that they transform text into a glyph vector.

+ * + *

Text transformations are necessary when modern display technology is not + * available or when text needs to be transformed to or from legacy formats that + * use "shaped" characters. Since the Arabic script is cursive, connecting + * adjacent letters to each other, computers select images for each letter based + * on the surrounding letters. This usually results in four images per Arabic + * letter: initial, middle, final, and isolated forms. In Unicode, on the other + * hand, letters are normally stored abstract, and a display system is expected + * to select the necessary glyphs. (This makes searching and other text + * processing easier because the same letter has only one code.) It is possible + * to mimic this with text transformations because there are characters in + * Unicode that are rendered as letters with a specific shape + * (or cursive connectivity). They were included for interoperability with + * legacy systems and codepages, and for unsophisticated display systems.

+ * + *

A second kind of text transformations is supported for Arabic digits: + * For compatibility with legacy codepages that only include European digits, + * it is possible to replace one set of digits by another, changing the + * character code points. These operations can be performed for either + * Arabic-Indic Digits (U+0660...U+0669) or Eastern (Extended) Arabic-Indic + * digits (U+06f0...U+06f9).

+ * + *

Some replacements may result in more or fewer characters (code points). + * By default, this means that the destination buffer may receive text with a + * length different from the source length. Some legacy systems rely on the + * length of the text to be constant. They expect extra spaces to be added + * or consumed either next to the affected character or at the end of the + * text.

+ * + *

For details about the available operations, see the description of the + * U_SHAPE_... options.

+ * + * @param source The input text. + * + * @param sourceLength The number of UChars in source. + * + * @param dest The destination buffer that will receive the results of the + * requested operations. It may be NULL only if + * destSize is 0. The source and destination must not + * overlap. + * + * @param destSize The size (capacity) of the destination buffer in UChars. + * If destSize is 0, then no output is produced, + * but the necessary buffer size is returned ("preflighting"). + * + * @param options This is a 32-bit set of flags that specify the operations + * that are performed on the input text. If no error occurs, + * then the result will always be written to the destination + * buffer. + * + * @param pErrorCode must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * + * @return The number of UChars written to the destination buffer. + * If an error occurred, then no output was written, or it may be + * incomplete. If U_BUFFER_OVERFLOW_ERROR is set, then + * the return value indicates the necessary destination buffer size. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_shapeArabic(const UChar *source, int32_t sourceLength, + UChar *dest, int32_t destSize, + uint32_t options, + UErrorCode *pErrorCode); + +/** + * Memory option: allow the result to have a different length than the source. + * Affects: LamAlef options + * @stable ICU 2.0 + */ +#define U_SHAPE_LENGTH_GROW_SHRINK 0 + +/** + * Memory option: allow the result to have a different length than the source. + * Affects: LamAlef options + * This option is an alias to U_SHAPE_LENGTH_GROW_SHRINK + * @stable ICU 4.2 + */ +#define U_SHAPE_LAMALEF_RESIZE 0 + +/** + * Memory option: the result must have the same length as the source. + * If more room is necessary, then try to consume spaces next to modified characters. + * @stable ICU 2.0 + */ +#define U_SHAPE_LENGTH_FIXED_SPACES_NEAR 1 + +/** + * Memory option: the result must have the same length as the source. + * If more room is necessary, then try to consume spaces next to modified characters. + * Affects: LamAlef options + * This option is an alias to U_SHAPE_LENGTH_FIXED_SPACES_NEAR + * @stable ICU 4.2 + */ +#define U_SHAPE_LAMALEF_NEAR 1 + +/** + * Memory option: the result must have the same length as the source. + * If more room is necessary, then try to consume spaces at the end of the text. + * @stable ICU 2.0 + */ +#define U_SHAPE_LENGTH_FIXED_SPACES_AT_END 2 + +/** + * Memory option: the result must have the same length as the source. + * If more room is necessary, then try to consume spaces at the end of the text. + * Affects: LamAlef options + * This option is an alias to U_SHAPE_LENGTH_FIXED_SPACES_AT_END + * @stable ICU 4.2 + */ +#define U_SHAPE_LAMALEF_END 2 + +/** + * Memory option: the result must have the same length as the source. + * If more room is necessary, then try to consume spaces at the beginning of the text. + * @stable ICU 2.0 + */ +#define U_SHAPE_LENGTH_FIXED_SPACES_AT_BEGINNING 3 + +/** + * Memory option: the result must have the same length as the source. + * If more room is necessary, then try to consume spaces at the beginning of the text. + * Affects: LamAlef options + * This option is an alias to U_SHAPE_LENGTH_FIXED_SPACES_AT_BEGINNING + * @stable ICU 4.2 + */ +#define U_SHAPE_LAMALEF_BEGIN 3 + + +/** + * Memory option: the result must have the same length as the source. + * Shaping Mode: For each LAMALEF character found, expand LAMALEF using space at end. + * If there is no space at end, use spaces at beginning of the buffer. If there + * is no space at beginning of the buffer, use spaces at the near (i.e. the space + * after the LAMALEF character). + * If there are no spaces found, an error U_NO_SPACE_AVAILABLE (as defined in utypes.h) + * will be set in pErrorCode + * + * Deshaping Mode: Perform the same function as the flag equals U_SHAPE_LAMALEF_END. + * Affects: LamAlef options + * @stable ICU 4.2 + */ +#define U_SHAPE_LAMALEF_AUTO 0x10000 + +/** Bit mask for memory options. @stable ICU 2.0 */ +#define U_SHAPE_LENGTH_MASK 0x10003 /* Changed old value 3 */ + + +/** + * Bit mask for LamAlef memory options. + * @stable ICU 4.2 + */ +#define U_SHAPE_LAMALEF_MASK 0x10003 /* updated */ + +/** Direction indicator: the source is in logical (keyboard) order. @stable ICU 2.0 */ +#define U_SHAPE_TEXT_DIRECTION_LOGICAL 0 + +/** + * Direction indicator: + * the source is in visual RTL order, + * the rightmost displayed character stored first. + * This option is an alias to U_SHAPE_TEXT_DIRECTION_LOGICAL + * @stable ICU 4.2 + */ +#define U_SHAPE_TEXT_DIRECTION_VISUAL_RTL 0 + +/** + * Direction indicator: + * the source is in visual LTR order, + * the leftmost displayed character stored first. + * @stable ICU 2.0 + */ +#define U_SHAPE_TEXT_DIRECTION_VISUAL_LTR 4 + +/** Bit mask for direction indicators. @stable ICU 2.0 */ +#define U_SHAPE_TEXT_DIRECTION_MASK 4 + + +/** Letter shaping option: do not perform letter shaping. @stable ICU 2.0 */ +#define U_SHAPE_LETTERS_NOOP 0 + +/** Letter shaping option: replace abstract letter characters by "shaped" ones. @stable ICU 2.0 */ +#define U_SHAPE_LETTERS_SHAPE 8 + +/** Letter shaping option: replace "shaped" letter characters by abstract ones. @stable ICU 2.0 */ +#define U_SHAPE_LETTERS_UNSHAPE 0x10 + +/** + * Letter shaping option: replace abstract letter characters by "shaped" ones. + * The only difference with U_SHAPE_LETTERS_SHAPE is that Tashkeel letters + * are always "shaped" into the isolated form instead of the medial form + * (selecting code points from the Arabic Presentation Forms-B block). + * @stable ICU 2.0 + */ +#define U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED 0x18 + + +/** Bit mask for letter shaping options. @stable ICU 2.0 */ +#define U_SHAPE_LETTERS_MASK 0x18 + + +/** Digit shaping option: do not perform digit shaping. @stable ICU 2.0 */ +#define U_SHAPE_DIGITS_NOOP 0 + +/** + * Digit shaping option: + * Replace European digits (U+0030...) by Arabic-Indic digits. + * @stable ICU 2.0 + */ +#define U_SHAPE_DIGITS_EN2AN 0x20 + +/** + * Digit shaping option: + * Replace Arabic-Indic digits by European digits (U+0030...). + * @stable ICU 2.0 + */ +#define U_SHAPE_DIGITS_AN2EN 0x40 + +/** + * Digit shaping option: + * Replace European digits (U+0030...) by Arabic-Indic digits if the most recent + * strongly directional character is an Arabic letter + * (u_charDirection() result U_RIGHT_TO_LEFT_ARABIC [AL]).
+ * The direction of "preceding" depends on the direction indicator option. + * For the first characters, the preceding strongly directional character + * (initial state) is assumed to be not an Arabic letter + * (it is U_LEFT_TO_RIGHT [L] or U_RIGHT_TO_LEFT [R]). + * @stable ICU 2.0 + */ +#define U_SHAPE_DIGITS_ALEN2AN_INIT_LR 0x60 + +/** + * Digit shaping option: + * Replace European digits (U+0030...) by Arabic-Indic digits if the most recent + * strongly directional character is an Arabic letter + * (u_charDirection() result U_RIGHT_TO_LEFT_ARABIC [AL]).
+ * The direction of "preceding" depends on the direction indicator option. + * For the first characters, the preceding strongly directional character + * (initial state) is assumed to be an Arabic letter. + * @stable ICU 2.0 + */ +#define U_SHAPE_DIGITS_ALEN2AN_INIT_AL 0x80 + +/** Not a valid option value. May be replaced by a new option. @stable ICU 2.0 */ +#define U_SHAPE_DIGITS_RESERVED 0xa0 + +/** Bit mask for digit shaping options. @stable ICU 2.0 */ +#define U_SHAPE_DIGITS_MASK 0xe0 + + +/** Digit type option: Use Arabic-Indic digits (U+0660...U+0669). @stable ICU 2.0 */ +#define U_SHAPE_DIGIT_TYPE_AN 0 + +/** Digit type option: Use Eastern (Extended) Arabic-Indic digits (U+06f0...U+06f9). @stable ICU 2.0 */ +#define U_SHAPE_DIGIT_TYPE_AN_EXTENDED 0x100 + +/** Not a valid option value. May be replaced by a new option. @stable ICU 2.0 */ +#define U_SHAPE_DIGIT_TYPE_RESERVED 0x200 + +/** Bit mask for digit type options. @stable ICU 2.0 */ +#define U_SHAPE_DIGIT_TYPE_MASK 0x300 /* I need to change this from 0x3f00 to 0x300 */ + +/** + * Tashkeel aggregation option: + * Replaces any combination of U+0651 with one of + * U+064C, U+064D, U+064E, U+064F, U+0650 with + * U+FC5E, U+FC5F, U+FC60, U+FC61, U+FC62 consecutively. + * @stable ICU 3.6 + */ +#define U_SHAPE_AGGREGATE_TASHKEEL 0x4000 +/** Tashkeel aggregation option: do not aggregate tashkeels. @stable ICU 3.6 */ +#define U_SHAPE_AGGREGATE_TASHKEEL_NOOP 0 +/** Bit mask for tashkeel aggregation. @stable ICU 3.6 */ +#define U_SHAPE_AGGREGATE_TASHKEEL_MASK 0x4000 + +/** + * Presentation form option: + * Don't replace Arabic Presentation Forms-A and Arabic Presentation Forms-B + * characters with 0+06xx characters, before shaping. + * @stable ICU 3.6 + */ +#define U_SHAPE_PRESERVE_PRESENTATION 0x8000 +/** Presentation form option: + * Replace Arabic Presentation Forms-A and Arabic Presentationo Forms-B with + * their unshaped correspondents in range 0+06xx, before shaping. + * @stable ICU 3.6 + */ +#define U_SHAPE_PRESERVE_PRESENTATION_NOOP 0 +/** Bit mask for preserve presentation form. @stable ICU 3.6 */ +#define U_SHAPE_PRESERVE_PRESENTATION_MASK 0x8000 + +/* Seen Tail option */ +/** + * Memory option: the result must have the same length as the source. + * Shaping mode: The SEEN family character will expand into two characters using space near + * the SEEN family character(i.e. the space after the character). + * If there are no spaces found, an error U_NO_SPACE_AVAILABLE (as defined in utypes.h) + * will be set in pErrorCode + * + * De-shaping mode: Any Seen character followed by Tail character will be + * replaced by one cell Seen and a space will replace the Tail. + * Affects: Seen options + * @stable ICU 4.2 + */ +#define U_SHAPE_SEEN_TWOCELL_NEAR 0x200000 + +/** + * Bit mask for Seen memory options. + * @stable ICU 4.2 + */ +#define U_SHAPE_SEEN_MASK 0x700000 + +/* YehHamza option */ +/** + * Memory option: the result must have the same length as the source. + * Shaping mode: The YEHHAMZA character will expand into two characters using space near it + * (i.e. the space after the character + * If there are no spaces found, an error U_NO_SPACE_AVAILABLE (as defined in utypes.h) + * will be set in pErrorCode + * + * De-shaping mode: Any Yeh (final or isolated) character followed by Hamza character will be + * replaced by one cell YehHamza and space will replace the Hamza. + * Affects: YehHamza options + * @stable ICU 4.2 + */ +#define U_SHAPE_YEHHAMZA_TWOCELL_NEAR 0x1000000 + + +/** + * Bit mask for YehHamza memory options. + * @stable ICU 4.2 + */ +#define U_SHAPE_YEHHAMZA_MASK 0x3800000 + +/* New Tashkeel options */ +/** + * Memory option: the result must have the same length as the source. + * Shaping mode: Tashkeel characters will be replaced by spaces. + * Spaces will be placed at beginning of the buffer + * + * De-shaping mode: N/A + * Affects: Tashkeel options + * @stable ICU 4.2 + */ +#define U_SHAPE_TASHKEEL_BEGIN 0x40000 + +/** + * Memory option: the result must have the same length as the source. + * Shaping mode: Tashkeel characters will be replaced by spaces. + * Spaces will be placed at end of the buffer + * + * De-shaping mode: N/A + * Affects: Tashkeel options + * @stable ICU 4.2 + */ +#define U_SHAPE_TASHKEEL_END 0x60000 + +/** + * Memory option: allow the result to have a different length than the source. + * Shaping mode: Tashkeel characters will be removed, buffer length will shrink. + * De-shaping mode: N/A + * + * Affect: Tashkeel options + * @stable ICU 4.2 + */ +#define U_SHAPE_TASHKEEL_RESIZE 0x80000 + +/** + * Memory option: the result must have the same length as the source. + * Shaping mode: Tashkeel characters will be replaced by Tatweel if it is connected to adjacent + * characters (i.e. shaped on Tatweel) or replaced by space if it is not connected. + * + * De-shaping mode: N/A + * Affects: YehHamza options + * @stable ICU 4.2 + */ +#define U_SHAPE_TASHKEEL_REPLACE_BY_TATWEEL 0xC0000 + +/** + * Bit mask for Tashkeel replacement with Space or Tatweel memory options. + * @stable ICU 4.2 + */ +#define U_SHAPE_TASHKEEL_MASK 0xE0000 + + +/* Space location Control options */ +/** + * This option affect the meaning of BEGIN and END options. if this option is not used the default + * for BEGIN and END will be as following: + * The Default (for both Visual LTR, Visual RTL and Logical Text) + * 1. BEGIN always refers to the start address of physical memory. + * 2. END always refers to the end address of physical memory. + * + * If this option is used it will swap the meaning of BEGIN and END only for Visual LTR text. + * + * The effect on BEGIN and END Memory Options will be as following: + * A. BEGIN For Visual LTR text: This will be the beginning (right side) of the visual text( + * corresponding to the physical memory address end for Visual LTR text, Same as END in + * default behavior) + * B. BEGIN For Logical text: Same as BEGIN in default behavior. + * C. END For Visual LTR text: This will be the end (left side) of the visual text (corresponding + * to the physical memory address beginning for Visual LTR text, Same as BEGIN in default behavior. + * D. END For Logical text: Same as END in default behavior). + * Affects: All LamAlef BEGIN, END and AUTO options. + * @stable ICU 4.2 + */ +#define U_SHAPE_SPACES_RELATIVE_TO_TEXT_BEGIN_END 0x4000000 + +/** + * Bit mask for swapping BEGIN and END for Visual LTR text + * @stable ICU 4.2 + */ +#define U_SHAPE_SPACES_RELATIVE_TO_TEXT_MASK 0x4000000 + +/** + * If this option is used, shaping will use the new Unicode code point for TAIL (i.e. 0xFE73). + * If this option is not specified (Default), old unofficial Unicode TAIL code point is used (i.e. 0x200B) + * De-shaping will not use this option as it will always search for both the new Unicode code point for the + * TAIL (i.e. 0xFE73) or the old unofficial Unicode TAIL code point (i.e. 0x200B) and de-shape the + * Seen-Family letter accordingly. + * + * Shaping Mode: Only shaping. + * De-shaping Mode: N/A. + * Affects: All Seen options + * @stable ICU 4.8 + */ +#define U_SHAPE_TAIL_NEW_UNICODE 0x8000000 + +/** + * Bit mask for new Unicode Tail option + * @stable ICU 4.8 + */ +#define U_SHAPE_TAIL_TYPE_MASK 0x8000000 + +#endif diff --git a/intl/icu/source/common/unicode/usprep.h b/intl/icu/source/common/unicode/usprep.h new file mode 100644 index 0000000000..f8a0f58e0d --- /dev/null +++ b/intl/icu/source/common/unicode/usprep.h @@ -0,0 +1,274 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* + ******************************************************************************* + * + * Copyright (C) 2003-2014, International Business Machines + * Corporation and others. All Rights Reserved. + * + ******************************************************************************* + * file name: usprep.h + * encoding: UTF-8 + * tab size: 8 (not used) + * indentation:4 + * + * created on: 2003jul2 + * created by: Ram Viswanadha + */ + +#ifndef __USPREP_H__ +#define __USPREP_H__ + +/** + * \file + * \brief C API: Implements the StringPrep algorithm. + */ + +#include "unicode/utypes.h" + +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#endif // U_SHOW_CPLUSPLUS_API + +/** + * + * StringPrep API implements the StingPrep framework as described by RFC 3454. + * StringPrep prepares Unicode strings for use in network protocols. + * Profiles of StingPrep are set of rules and data according to with the + * Unicode Strings are prepared. Each profiles contains tables which describe + * how a code point should be treated. The tables are broadly classified into + *
    + *
  • Unassigned Table: Contains code points that are unassigned + * in the Unicode Version supported by StringPrep. Currently + * RFC 3454 supports Unicode 3.2.
    • + *
  • Prohibited Table: Contains code points that are prohibited from + * the output of the StringPrep processing function.
    • + *
  • Mapping Table: Contains code points that are deleted from the output or case mapped.
    • + *
+ * + * The procedure for preparing Unicode strings: + *
    + *
  1. Map: For each character in the input, check if it has a mapping + * and, if so, replace it with its mapping.
    2. + *
  2. Normalize: Possibly normalize the result of step 1 using Unicode + * normalization.
    3. + *
  3. Prohibit: Check for any characters that are not allowed in the + * output. If any are found, return an error.
    4. + *
  4. Check bidi: Possibly check for right-to-left characters, and if + * any are found, make sure that the whole string satisfies the + * requirements for bidirectional strings. If the string does not + * satisfy the requirements for bidirectional strings, return an + * error.
    5. + *
+ * @author Ram Viswanadha + */ +#if !UCONFIG_NO_IDNA + +#include "unicode/parseerr.h" + +/** + * The StringPrep profile + * @stable ICU 2.8 + */ +typedef struct UStringPrepProfile UStringPrepProfile; + + +/** + * Option to prohibit processing of unassigned code points in the input + * + * @see usprep_prepare + * @stable ICU 2.8 + */ +#define USPREP_DEFAULT 0x0000 + +/** + * Option to allow processing of unassigned code points in the input + * + * @see usprep_prepare + * @stable ICU 2.8 + */ +#define USPREP_ALLOW_UNASSIGNED 0x0001 + +/** + * enums for the standard stringprep profile types + * supported by usprep_openByType. + * @see usprep_openByType + * @stable ICU 4.2 + */ +typedef enum UStringPrepProfileType { + /** + * RFC3491 Nameprep + * @stable ICU 4.2 + */ + USPREP_RFC3491_NAMEPREP, + /** + * RFC3530 nfs4_cs_prep + * @stable ICU 4.2 + */ + USPREP_RFC3530_NFS4_CS_PREP, + /** + * RFC3530 nfs4_cs_prep with case insensitive option + * @stable ICU 4.2 + */ + USPREP_RFC3530_NFS4_CS_PREP_CI, + /** + * RFC3530 nfs4_cis_prep + * @stable ICU 4.2 + */ + USPREP_RFC3530_NFS4_CIS_PREP, + /** + * RFC3530 nfs4_mixed_prep for prefix + * @stable ICU 4.2 + */ + USPREP_RFC3530_NFS4_MIXED_PREP_PREFIX, + /** + * RFC3530 nfs4_mixed_prep for suffix + * @stable ICU 4.2 + */ + USPREP_RFC3530_NFS4_MIXED_PREP_SUFFIX, + /** + * RFC3722 iSCSI + * @stable ICU 4.2 + */ + USPREP_RFC3722_ISCSI, + /** + * RFC3920 XMPP Nodeprep + * @stable ICU 4.2 + */ + USPREP_RFC3920_NODEPREP, + /** + * RFC3920 XMPP Resourceprep + * @stable ICU 4.2 + */ + USPREP_RFC3920_RESOURCEPREP, + /** + * RFC4011 Policy MIB Stringprep + * @stable ICU 4.2 + */ + USPREP_RFC4011_MIB, + /** + * RFC4013 SASLprep + * @stable ICU 4.2 + */ + USPREP_RFC4013_SASLPREP, + /** + * RFC4505 trace + * @stable ICU 4.2 + */ + USPREP_RFC4505_TRACE, + /** + * RFC4518 LDAP + * @stable ICU 4.2 + */ + USPREP_RFC4518_LDAP, + /** + * RFC4518 LDAP for case ignore, numeric and stored prefix + * matching rules + * @stable ICU 4.2 + */ + USPREP_RFC4518_LDAP_CI +} UStringPrepProfileType; + +/** + * Creates a StringPrep profile from the data file. + * + * @param path string containing the full path pointing to the directory + * where the profile reside followed by the package name + * e.g. "/usr/resource/my_app/profiles/mydata" on a Unix system. + * if NULL, ICU default data files will be used. + * @param fileName name of the profile file to be opened + * @param status ICU error code in/out parameter. Must not be NULL. + * Must fulfill U_SUCCESS before the function call. + * @return Pointer to UStringPrepProfile that is opened. Should be closed by + * calling usprep_close() + * @see usprep_close() + * @stable ICU 2.8 + */ +U_CAPI UStringPrepProfile* U_EXPORT2 +usprep_open(const char* path, + const char* fileName, + UErrorCode* status); + +/** + * Creates a StringPrep profile for the specified profile type. + * + * @param type The profile type + * @param status ICU error code in/out parameter. Must not be NULL. + * Must fulfill U_SUCCESS before the function call. + * @return Pointer to UStringPrepProfile that is opened. Should be closed by + * calling usprep_close() + * @see usprep_close() + * @stable ICU 4.2 + */ +U_CAPI UStringPrepProfile* U_EXPORT2 +usprep_openByType(UStringPrepProfileType type, + UErrorCode* status); + +/** + * Closes the profile + * @param profile The profile to close + * @stable ICU 2.8 + */ +U_CAPI void U_EXPORT2 +usprep_close(UStringPrepProfile* profile); + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUStringPrepProfilePointer + * "Smart pointer" class, closes a UStringPrepProfile via usprep_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUStringPrepProfilePointer, UStringPrepProfile, usprep_close); + +U_NAMESPACE_END + +#endif + +/** + * Prepare the input buffer for use in applications with the given profile. This operation maps, normalizes(NFKC), + * checks for prohibited and BiDi characters in the order defined by RFC 3454 + * depending on the options specified in the profile. + * + * @param prep The profile to use + * @param src Pointer to UChar buffer containing the string to prepare + * @param srcLength Number of characters in the source string + * @param dest Pointer to the destination buffer to receive the output + * @param destCapacity The capacity of destination array + * @param options A bit set of options: + * + * - USPREP_DEFAULT Prohibit processing of unassigned code points in the input + * + * - USPREP_ALLOW_UNASSIGNED Treat the unassigned code points are in the input + * as normal Unicode code points. + * + * @param parseError Pointer to UParseError struct to receive information on position + * of error if an error is encountered. Can be NULL. + * @param status ICU in/out error code parameter. + * U_INVALID_CHAR_FOUND if src contains + * unmatched single surrogates. + * U_INDEX_OUTOFBOUNDS_ERROR if src contains + * too many code points. + * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough + * @return The number of UChars in the destination buffer + * @stable ICU 2.8 + */ + +U_CAPI int32_t U_EXPORT2 +usprep_prepare( const UStringPrepProfile* prep, + const UChar* src, int32_t srcLength, + UChar* dest, int32_t destCapacity, + int32_t options, + UParseError* parseError, + UErrorCode* status ); + + +#endif /* #if !UCONFIG_NO_IDNA */ + +#endif diff --git a/intl/icu/source/common/unicode/ustring.h b/intl/icu/source/common/unicode/ustring.h new file mode 100644 index 0000000000..03c697c722 --- /dev/null +++ b/intl/icu/source/common/unicode/ustring.h @@ -0,0 +1,1685 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +********************************************************************** +* Copyright (C) 1998-2014, International Business Machines +* Corporation and others. All Rights Reserved. +********************************************************************** +* +* File ustring.h +* +* Modification History: +* +* Date Name Description +* 12/07/98 bertrand Creation. +****************************************************************************** +*/ + +#ifndef USTRING_H +#define USTRING_H + +#include "unicode/utypes.h" +#include "unicode/putil.h" +#include "unicode/uiter.h" + +/** + * \def UBRK_TYPEDEF_UBREAK_ITERATOR + * @internal + */ + +#ifndef UBRK_TYPEDEF_UBREAK_ITERATOR +# define UBRK_TYPEDEF_UBREAK_ITERATOR +/** Simple declaration for u_strToTitle() to avoid including unicode/ubrk.h. @stable ICU 2.1*/ + typedef struct UBreakIterator UBreakIterator; +#endif + +/** + * \file + * \brief C API: Unicode string handling functions + * + * These C API functions provide general Unicode string handling. + * + * Some functions are equivalent in name, signature, and behavior to the ANSI C + * functions. (For example, they do not check for bad arguments like NULL string pointers.) + * In some cases, only the thread-safe variant of such a function is implemented here + * (see u_strtok_r()). + * + * Other functions provide more Unicode-specific functionality like locale-specific + * upper/lower-casing and string comparison in code point order. + * + * ICU uses 16-bit Unicode (UTF-16) in the form of arrays of UChar code units. + * UTF-16 encodes each Unicode code point with either one or two UChar code units. + * (This is the default form of Unicode, and a forward-compatible extension of the original, + * fixed-width form that was known as UCS-2. UTF-16 superseded UCS-2 with Unicode 2.0 + * in 1996.) + * + * Some APIs accept a 32-bit UChar32 value for a single code point. + * + * ICU also handles 16-bit Unicode text with unpaired surrogates. + * Such text is not well-formed UTF-16. + * Code-point-related functions treat unpaired surrogates as surrogate code points, + * i.e., as separate units. + * + * Although UTF-16 is a variable-width encoding form (like some legacy multi-byte encodings), + * it is much more efficient even for random access because the code unit values + * for single-unit characters vs. lead units vs. trail units are completely disjoint. + * This means that it is easy to determine character (code point) boundaries from + * random offsets in the string. + * + * Unicode (UTF-16) string processing is optimized for the single-unit case. + * Although it is important to support supplementary characters + * (which use pairs of lead/trail code units called "surrogates"), + * their occurrence is rare. Almost all characters in modern use require only + * a single UChar code unit (i.e., their code point values are <=0xffff). + * + * For more details see the User Guide Strings chapter (https://unicode-org.github.io/icu/userguide/strings/). + * For a discussion of the handling of unpaired surrogates see also + * Jitterbug 2145 and its icu mailing list proposal on 2002-sep-18. + */ + +/** + * \defgroup ustring_ustrlen String Length + * \ingroup ustring_strlen + */ +/*@{*/ +/** + * Determine the length of an array of UChar. + * + * @param s The array of UChars, NULL (U+0000) terminated. + * @return The number of UChars in chars, minus the terminator. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_strlen(const UChar *s); +/*@}*/ + +/** + * Count Unicode code points in the length UChar code units of the string. + * A code point may occupy either one or two UChar code units. + * Counting code points involves reading all code units. + * + * This functions is basically the inverse of the U16_FWD_N() macro (see utf.h). + * + * @param s The input string. + * @param length The number of UChar code units to be checked, or -1 to count all + * code points before the first NUL (U+0000). + * @return The number of code points in the specified code units. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_countChar32(const UChar *s, int32_t length); + +/** + * Check if the string contains more Unicode code points than a certain number. + * This is more efficient than counting all code points in the entire string + * and comparing that number with a threshold. + * This function may not need to scan the string at all if the length is known + * (not -1 for NUL-termination) and falls within a certain range, and + * never needs to count more than 'number+1' code points. + * Logically equivalent to (u_countChar32(s, length)>number). + * A Unicode code point may occupy either one or two UChar code units. + * + * @param s The input string. + * @param length The length of the string, or -1 if it is NUL-terminated. + * @param number The number of code points in the string is compared against + * the 'number' parameter. + * @return Boolean value for whether the string contains more Unicode code points + * than 'number'. Same as (u_countChar32(s, length)>number). + * @stable ICU 2.4 + */ +U_CAPI UBool U_EXPORT2 +u_strHasMoreChar32Than(const UChar *s, int32_t length, int32_t number); + +/** + * Concatenate two ustrings. Appends a copy of src, + * including the null terminator, to dst. The initial copied + * character from src overwrites the null terminator in dst. + * + * @param dst The destination string. + * @param src The source string. + * @return A pointer to dst. + * @stable ICU 2.0 + */ +U_CAPI UChar* U_EXPORT2 +u_strcat(UChar *dst, + const UChar *src); + +/** + * Concatenate two ustrings. + * Appends at most n characters from src to dst. + * Adds a terminating NUL. + * If src is too long, then only n-1 characters will be copied + * before the terminating NUL. + * If n<=0 then dst is not modified. + * + * @param dst The destination string. + * @param src The source string (can be NULL/invalid if n<=0). + * @param n The maximum number of characters to append; no-op if <=0. + * @return A pointer to dst. + * @stable ICU 2.0 + */ +U_CAPI UChar* U_EXPORT2 +u_strncat(UChar *dst, + const UChar *src, + int32_t n); + +/** + * Find the first occurrence of a substring in a string. + * The substring is found at code point boundaries. + * That means that if the substring begins with + * a trail surrogate or ends with a lead surrogate, + * then it is found only if these surrogates stand alone in the text. + * Otherwise, the substring edge units would be matched against + * halves of surrogate pairs. + * + * @param s The string to search (NUL-terminated). + * @param substring The substring to find (NUL-terminated). + * @return A pointer to the first occurrence of substring in s, + * or s itself if the substring is empty, + * or NULL if substring is not in s. + * @stable ICU 2.0 + * + * @see u_strrstr + * @see u_strFindFirst + * @see u_strFindLast + */ +U_CAPI UChar * U_EXPORT2 +u_strstr(const UChar *s, const UChar *substring); + +/** + * Find the first occurrence of a substring in a string. + * The substring is found at code point boundaries. + * That means that if the substring begins with + * a trail surrogate or ends with a lead surrogate, + * then it is found only if these surrogates stand alone in the text. + * Otherwise, the substring edge units would be matched against + * halves of surrogate pairs. + * + * @param s The string to search. + * @param length The length of s (number of UChars), or -1 if it is NUL-terminated. + * @param substring The substring to find (NUL-terminated). + * @param subLength The length of substring (number of UChars), or -1 if it is NUL-terminated. + * @return A pointer to the first occurrence of substring in s, + * or s itself if the substring is empty, + * or NULL if substring is not in s. + * @stable ICU 2.4 + * + * @see u_strstr + * @see u_strFindLast + */ +U_CAPI UChar * U_EXPORT2 +u_strFindFirst(const UChar *s, int32_t length, const UChar *substring, int32_t subLength); + +/** + * Find the first occurrence of a BMP code point in a string. + * A surrogate code point is found only if its match in the text is not + * part of a surrogate pair. + * A NUL character is found at the string terminator. + * + * @param s The string to search (NUL-terminated). + * @param c The BMP code point to find. + * @return A pointer to the first occurrence of c in s + * or NULL if c is not in s. + * @stable ICU 2.0 + * + * @see u_strchr32 + * @see u_memchr + * @see u_strstr + * @see u_strFindFirst + */ +U_CAPI UChar * U_EXPORT2 +u_strchr(const UChar *s, UChar c); + +/** + * Find the first occurrence of a code point in a string. + * A surrogate code point is found only if its match in the text is not + * part of a surrogate pair. + * A NUL character is found at the string terminator. + * + * @param s The string to search (NUL-terminated). + * @param c The code point to find. + * @return A pointer to the first occurrence of c in s + * or NULL if c is not in s. + * @stable ICU 2.0 + * + * @see u_strchr + * @see u_memchr32 + * @see u_strstr + * @see u_strFindFirst + */ +U_CAPI UChar * U_EXPORT2 +u_strchr32(const UChar *s, UChar32 c); + +/** + * Find the last occurrence of a substring in a string. + * The substring is found at code point boundaries. + * That means that if the substring begins with + * a trail surrogate or ends with a lead surrogate, + * then it is found only if these surrogates stand alone in the text. + * Otherwise, the substring edge units would be matched against + * halves of surrogate pairs. + * + * @param s The string to search (NUL-terminated). + * @param substring The substring to find (NUL-terminated). + * @return A pointer to the last occurrence of substring in s, + * or s itself if the substring is empty, + * or NULL if substring is not in s. + * @stable ICU 2.4 + * + * @see u_strstr + * @see u_strFindFirst + * @see u_strFindLast + */ +U_CAPI UChar * U_EXPORT2 +u_strrstr(const UChar *s, const UChar *substring); + +/** + * Find the last occurrence of a substring in a string. + * The substring is found at code point boundaries. + * That means that if the substring begins with + * a trail surrogate or ends with a lead surrogate, + * then it is found only if these surrogates stand alone in the text. + * Otherwise, the substring edge units would be matched against + * halves of surrogate pairs. + * + * @param s The string to search. + * @param length The length of s (number of UChars), or -1 if it is NUL-terminated. + * @param substring The substring to find (NUL-terminated). + * @param subLength The length of substring (number of UChars), or -1 if it is NUL-terminated. + * @return A pointer to the last occurrence of substring in s, + * or s itself if the substring is empty, + * or NULL if substring is not in s. + * @stable ICU 2.4 + * + * @see u_strstr + * @see u_strFindLast + */ +U_CAPI UChar * U_EXPORT2 +u_strFindLast(const UChar *s, int32_t length, const UChar *substring, int32_t subLength); + +/** + * Find the last occurrence of a BMP code point in a string. + * A surrogate code point is found only if its match in the text is not + * part of a surrogate pair. + * A NUL character is found at the string terminator. + * + * @param s The string to search (NUL-terminated). + * @param c The BMP code point to find. + * @return A pointer to the last occurrence of c in s + * or NULL if c is not in s. + * @stable ICU 2.4 + * + * @see u_strrchr32 + * @see u_memrchr + * @see u_strrstr + * @see u_strFindLast + */ +U_CAPI UChar * U_EXPORT2 +u_strrchr(const UChar *s, UChar c); + +/** + * Find the last occurrence of a code point in a string. + * A surrogate code point is found only if its match in the text is not + * part of a surrogate pair. + * A NUL character is found at the string terminator. + * + * @param s The string to search (NUL-terminated). + * @param c The code point to find. + * @return A pointer to the last occurrence of c in s + * or NULL if c is not in s. + * @stable ICU 2.4 + * + * @see u_strrchr + * @see u_memchr32 + * @see u_strrstr + * @see u_strFindLast + */ +U_CAPI UChar * U_EXPORT2 +u_strrchr32(const UChar *s, UChar32 c); + +/** + * Locates the first occurrence in the string string of any of the characters + * in the string matchSet. + * Works just like C's strpbrk but with Unicode. + * + * @param string The string in which to search, NUL-terminated. + * @param matchSet A NUL-terminated string defining a set of code points + * for which to search in the text string. + * @return A pointer to the character in string that matches one of the + * characters in matchSet, or NULL if no such character is found. + * @stable ICU 2.0 + */ +U_CAPI UChar * U_EXPORT2 +u_strpbrk(const UChar *string, const UChar *matchSet); + +/** + * Returns the number of consecutive characters in string, + * beginning with the first, that do not occur somewhere in matchSet. + * Works just like C's strcspn but with Unicode. + * + * @param string The string in which to search, NUL-terminated. + * @param matchSet A NUL-terminated string defining a set of code points + * for which to search in the text string. + * @return The number of initial characters in string that do not + * occur in matchSet. + * @see u_strspn + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_strcspn(const UChar *string, const UChar *matchSet); + +/** + * Returns the number of consecutive characters in string, + * beginning with the first, that occur somewhere in matchSet. + * Works just like C's strspn but with Unicode. + * + * @param string The string in which to search, NUL-terminated. + * @param matchSet A NUL-terminated string defining a set of code points + * for which to search in the text string. + * @return The number of initial characters in string that do + * occur in matchSet. + * @see u_strcspn + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_strspn(const UChar *string, const UChar *matchSet); + +/** + * The string tokenizer API allows an application to break a string into + * tokens. Unlike strtok(), the saveState (the current pointer within the + * original string) is maintained in saveState. In the first call, the + * argument src is a pointer to the string. In subsequent calls to + * return successive tokens of that string, src must be specified as + * NULL. The value saveState is set by this function to maintain the + * function's position within the string, and on each subsequent call + * you must give this argument the same variable. This function does + * handle surrogate pairs. This function is similar to the strtok_r() + * the POSIX Threads Extension (1003.1c-1995) version. + * + * @param src String containing token(s). This string will be modified. + * After the first call to u_strtok_r(), this argument must + * be NULL to get to the next token. + * @param delim Set of delimiter characters (Unicode code points). + * @param saveState The current pointer within the original string, + * which is set by this function. The saveState + * parameter should the address of a local variable of type + * UChar *. (i.e. defined "UChar *myLocalSaveState" and use + * &myLocalSaveState for this parameter). + * @return A pointer to the next token found in src, or NULL + * when there are no more tokens. + * @stable ICU 2.0 + */ +U_CAPI UChar * U_EXPORT2 +u_strtok_r(UChar *src, + const UChar *delim, + UChar **saveState); + +/** + * Compare two Unicode strings for bitwise equality (code unit order). + * + * @param s1 A string to compare. + * @param s2 A string to compare. + * @return 0 if s1 and s2 are bitwise equal; a negative + * value if s1 is bitwise less than s2,; a positive + * value if s1 is bitwise greater than s2. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_strcmp(const UChar *s1, + const UChar *s2); + +/** + * Compare two Unicode strings in code point order. + * See u_strCompare for details. + * + * @param s1 A string to compare. + * @param s2 A string to compare. + * @return a negative/zero/positive integer corresponding to whether + * the first string is less than/equal to/greater than the second one + * in code point order + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_strcmpCodePointOrder(const UChar *s1, const UChar *s2); + +/** + * Compare two Unicode strings (binary order). + * + * The comparison can be done in code unit order or in code point order. + * They differ only in UTF-16 when + * comparing supplementary code points (U+10000..U+10ffff) + * to BMP code points near the end of the BMP (i.e., U+e000..U+ffff). + * In code unit order, high BMP code points sort after supplementary code points + * because they are stored as pairs of surrogates which are at U+d800..U+dfff. + * + * This functions works with strings of different explicitly specified lengths + * unlike the ANSI C-like u_strcmp() and u_memcmp() etc. + * NUL-terminated strings are possible with length arguments of -1. + * + * @param s1 First source string. + * @param length1 Length of first source string, or -1 if NUL-terminated. + * + * @param s2 Second source string. + * @param length2 Length of second source string, or -1 if NUL-terminated. + * + * @param codePointOrder Choose between code unit order (false) + * and code point order (true). + * + * @return <0 or 0 or >0 as usual for string comparisons + * + * @stable ICU 2.2 + */ +U_CAPI int32_t U_EXPORT2 +u_strCompare(const UChar *s1, int32_t length1, + const UChar *s2, int32_t length2, + UBool codePointOrder); + +/** + * Compare two Unicode strings (binary order) + * as presented by UCharIterator objects. + * Works otherwise just like u_strCompare(). + * + * Both iterators are reset to their start positions. + * When the function returns, it is undefined where the iterators + * have stopped. + * + * @param iter1 First source string iterator. + * @param iter2 Second source string iterator. + * @param codePointOrder Choose between code unit order (false) + * and code point order (true). + * + * @return <0 or 0 or >0 as usual for string comparisons + * + * @see u_strCompare + * + * @stable ICU 2.6 + */ +U_CAPI int32_t U_EXPORT2 +u_strCompareIter(UCharIterator *iter1, UCharIterator *iter2, UBool codePointOrder); + +/** + * Compare two strings case-insensitively using full case folding. + * This is equivalent to + * u_strCompare(u_strFoldCase(s1, options), + * u_strFoldCase(s2, options), + * (options&U_COMPARE_CODE_POINT_ORDER)!=0). + * + * The comparison can be done in UTF-16 code unit order or in code point order. + * They differ only when comparing supplementary code points (U+10000..U+10ffff) + * to BMP code points near the end of the BMP (i.e., U+e000..U+ffff). + * In code unit order, high BMP code points sort after supplementary code points + * because they are stored as pairs of surrogates which are at U+d800..U+dfff. + * + * This functions works with strings of different explicitly specified lengths + * unlike the ANSI C-like u_strcmp() and u_memcmp() etc. + * NUL-terminated strings are possible with length arguments of -1. + * + * @param s1 First source string. + * @param length1 Length of first source string, or -1 if NUL-terminated. + * + * @param s2 Second source string. + * @param length2 Length of second source string, or -1 if NUL-terminated. + * + * @param options A bit set of options: + * - U_FOLD_CASE_DEFAULT or 0 is used for default options: + * Comparison in code unit order with default case folding. + * + * - U_COMPARE_CODE_POINT_ORDER + * Set to choose code point order instead of code unit order + * (see u_strCompare for details). + * + * - U_FOLD_CASE_EXCLUDE_SPECIAL_I + * + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * + * @return <0 or 0 or >0 as usual for string comparisons + * + * @stable ICU 2.2 + */ +U_CAPI int32_t U_EXPORT2 +u_strCaseCompare(const UChar *s1, int32_t length1, + const UChar *s2, int32_t length2, + uint32_t options, + UErrorCode *pErrorCode); + +/** + * Compare two ustrings for bitwise equality. + * Compares at most n characters. + * + * @param ucs1 A string to compare (can be NULL/invalid if n<=0). + * @param ucs2 A string to compare (can be NULL/invalid if n<=0). + * @param n The maximum number of characters to compare; always returns 0 if n<=0. + * @return 0 if s1 and s2 are bitwise equal; a negative + * value if s1 is bitwise less than s2; a positive + * value if s1 is bitwise greater than s2. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_strncmp(const UChar *ucs1, + const UChar *ucs2, + int32_t n); + +/** + * Compare two Unicode strings in code point order. + * This is different in UTF-16 from u_strncmp() if supplementary characters are present. + * For details, see u_strCompare(). + * + * @param s1 A string to compare. + * @param s2 A string to compare. + * @param n The maximum number of characters to compare. + * @return a negative/zero/positive integer corresponding to whether + * the first string is less than/equal to/greater than the second one + * in code point order + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_strncmpCodePointOrder(const UChar *s1, const UChar *s2, int32_t n); + +/** + * Compare two strings case-insensitively using full case folding. + * This is equivalent to u_strcmp(u_strFoldCase(s1, options), u_strFoldCase(s2, options)). + * + * @param s1 A string to compare. + * @param s2 A string to compare. + * @param options A bit set of options: + * - U_FOLD_CASE_DEFAULT or 0 is used for default options: + * Comparison in code unit order with default case folding. + * + * - U_COMPARE_CODE_POINT_ORDER + * Set to choose code point order instead of code unit order + * (see u_strCompare for details). + * + * - U_FOLD_CASE_EXCLUDE_SPECIAL_I + * + * @return A negative, zero, or positive integer indicating the comparison result. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_strcasecmp(const UChar *s1, const UChar *s2, uint32_t options); + +/** + * Compare two strings case-insensitively using full case folding. + * This is equivalent to u_strcmp(u_strFoldCase(s1, at most n, options), + * u_strFoldCase(s2, at most n, options)). + * + * @param s1 A string to compare. + * @param s2 A string to compare. + * @param n The maximum number of characters each string to case-fold and then compare. + * @param options A bit set of options: + * - U_FOLD_CASE_DEFAULT or 0 is used for default options: + * Comparison in code unit order with default case folding. + * + * - U_COMPARE_CODE_POINT_ORDER + * Set to choose code point order instead of code unit order + * (see u_strCompare for details). + * + * - U_FOLD_CASE_EXCLUDE_SPECIAL_I + * + * @return A negative, zero, or positive integer indicating the comparison result. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_strncasecmp(const UChar *s1, const UChar *s2, int32_t n, uint32_t options); + +/** + * Compare two strings case-insensitively using full case folding. + * This is equivalent to u_strcmp(u_strFoldCase(s1, n, options), + * u_strFoldCase(s2, n, options)). + * + * @param s1 A string to compare. + * @param s2 A string to compare. + * @param length The number of characters in each string to case-fold and then compare. + * @param options A bit set of options: + * - U_FOLD_CASE_DEFAULT or 0 is used for default options: + * Comparison in code unit order with default case folding. + * + * - U_COMPARE_CODE_POINT_ORDER + * Set to choose code point order instead of code unit order + * (see u_strCompare for details). + * + * - U_FOLD_CASE_EXCLUDE_SPECIAL_I + * + * @return A negative, zero, or positive integer indicating the comparison result. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_memcasecmp(const UChar *s1, const UChar *s2, int32_t length, uint32_t options); + +/** + * Copy a ustring. Adds a null terminator. + * + * @param dst The destination string. + * @param src The source string. + * @return A pointer to dst. + * @stable ICU 2.0 + */ +U_CAPI UChar* U_EXPORT2 +u_strcpy(UChar *dst, + const UChar *src); + +/** + * Copy a ustring. + * Copies at most n characters. The result will be null terminated + * if the length of src is less than n. + * + * @param dst The destination string. + * @param src The source string (can be NULL/invalid if n<=0). + * @param n The maximum number of characters to copy; no-op if <=0. + * @return A pointer to dst. + * @stable ICU 2.0 + */ +U_CAPI UChar* U_EXPORT2 +u_strncpy(UChar *dst, + const UChar *src, + int32_t n); + +#if !UCONFIG_NO_CONVERSION + +/** + * Copy a byte string encoded in the default codepage to a ustring. + * Adds a null terminator. + * Performs a host byte to UChar conversion + * + * @param dst The destination string. + * @param src The source string. + * @return A pointer to dst. + * @stable ICU 2.0 + */ +U_CAPI UChar* U_EXPORT2 u_uastrcpy(UChar *dst, + const char *src ); + +/** + * Copy a byte string encoded in the default codepage to a ustring. + * Copies at most n characters. The result will be null terminated + * if the length of src is less than n. + * Performs a host byte to UChar conversion + * + * @param dst The destination string. + * @param src The source string. + * @param n The maximum number of characters to copy. + * @return A pointer to dst. + * @stable ICU 2.0 + */ +U_CAPI UChar* U_EXPORT2 u_uastrncpy(UChar *dst, + const char *src, + int32_t n); + +/** + * Copy ustring to a byte string encoded in the default codepage. + * Adds a null terminator. + * Performs a UChar to host byte conversion + * + * @param dst The destination string. + * @param src The source string. + * @return A pointer to dst. + * @stable ICU 2.0 + */ +U_CAPI char* U_EXPORT2 u_austrcpy(char *dst, + const UChar *src ); + +/** + * Copy ustring to a byte string encoded in the default codepage. + * Copies at most n characters. The result will be null terminated + * if the length of src is less than n. + * Performs a UChar to host byte conversion + * + * @param dst The destination string. + * @param src The source string. + * @param n The maximum number of characters to copy. + * @return A pointer to dst. + * @stable ICU 2.0 + */ +U_CAPI char* U_EXPORT2 u_austrncpy(char *dst, + const UChar *src, + int32_t n ); + +#endif + +/** + * Synonym for memcpy(), but with UChars only. + * @param dest The destination string + * @param src The source string (can be NULL/invalid if count<=0) + * @param count The number of characters to copy; no-op if <=0 + * @return A pointer to dest + * @stable ICU 2.0 + */ +U_CAPI UChar* U_EXPORT2 +u_memcpy(UChar *dest, const UChar *src, int32_t count); + +/** + * Synonym for memmove(), but with UChars only. + * @param dest The destination string + * @param src The source string (can be NULL/invalid if count<=0) + * @param count The number of characters to move; no-op if <=0 + * @return A pointer to dest + * @stable ICU 2.0 + */ +U_CAPI UChar* U_EXPORT2 +u_memmove(UChar *dest, const UChar *src, int32_t count); + +/** + * Initialize count characters of dest to c. + * + * @param dest The destination string. + * @param c The character to initialize the string. + * @param count The maximum number of characters to set. + * @return A pointer to dest. + * @stable ICU 2.0 + */ +U_CAPI UChar* U_EXPORT2 +u_memset(UChar *dest, UChar c, int32_t count); + +/** + * Compare the first count UChars of each buffer. + * + * @param buf1 The first string to compare. + * @param buf2 The second string to compare. + * @param count The maximum number of UChars to compare. + * @return When buf1 < buf2, a negative number is returned. + * When buf1 == buf2, 0 is returned. + * When buf1 > buf2, a positive number is returned. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_memcmp(const UChar *buf1, const UChar *buf2, int32_t count); + +/** + * Compare two Unicode strings in code point order. + * This is different in UTF-16 from u_memcmp() if supplementary characters are present. + * For details, see u_strCompare(). + * + * @param s1 A string to compare. + * @param s2 A string to compare. + * @param count The maximum number of characters to compare. + * @return a negative/zero/positive integer corresponding to whether + * the first string is less than/equal to/greater than the second one + * in code point order + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_memcmpCodePointOrder(const UChar *s1, const UChar *s2, int32_t count); + +/** + * Find the first occurrence of a BMP code point in a string. + * A surrogate code point is found only if its match in the text is not + * part of a surrogate pair. + * A NUL character is found at the string terminator. + * + * @param s The string to search (contains count UChars). + * @param c The BMP code point to find. + * @param count The length of the string. + * @return A pointer to the first occurrence of c in s + * or NULL if c is not in s. + * @stable ICU 2.0 + * + * @see u_strchr + * @see u_memchr32 + * @see u_strFindFirst + */ +U_CAPI UChar* U_EXPORT2 +u_memchr(const UChar *s, UChar c, int32_t count); + +/** + * Find the first occurrence of a code point in a string. + * A surrogate code point is found only if its match in the text is not + * part of a surrogate pair. + * A NUL character is found at the string terminator. + * + * @param s The string to search (contains count UChars). + * @param c The code point to find. + * @param count The length of the string. + * @return A pointer to the first occurrence of c in s + * or NULL if c is not in s. + * @stable ICU 2.0 + * + * @see u_strchr32 + * @see u_memchr + * @see u_strFindFirst + */ +U_CAPI UChar* U_EXPORT2 +u_memchr32(const UChar *s, UChar32 c, int32_t count); + +/** + * Find the last occurrence of a BMP code point in a string. + * A surrogate code point is found only if its match in the text is not + * part of a surrogate pair. + * A NUL character is found at the string terminator. + * + * @param s The string to search (contains count UChars). + * @param c The BMP code point to find. + * @param count The length of the string. + * @return A pointer to the last occurrence of c in s + * or NULL if c is not in s. + * @stable ICU 2.4 + * + * @see u_strrchr + * @see u_memrchr32 + * @see u_strFindLast + */ +U_CAPI UChar* U_EXPORT2 +u_memrchr(const UChar *s, UChar c, int32_t count); + +/** + * Find the last occurrence of a code point in a string. + * A surrogate code point is found only if its match in the text is not + * part of a surrogate pair. + * A NUL character is found at the string terminator. + * + * @param s The string to search (contains count UChars). + * @param c The code point to find. + * @param count The length of the string. + * @return A pointer to the last occurrence of c in s + * or NULL if c is not in s. + * @stable ICU 2.4 + * + * @see u_strrchr32 + * @see u_memrchr + * @see u_strFindLast + */ +U_CAPI UChar* U_EXPORT2 +u_memrchr32(const UChar *s, UChar32 c, int32_t count); + +/** + * Unicode String literals in C. + * We need one macro to declare a variable for the string + * and to statically preinitialize it if possible, + * and a second macro to dynamically initialize such a string variable if necessary. + * + * The macros are defined for maximum performance. + * They work only for strings that contain "invariant characters", i.e., + * only latin letters, digits, and some punctuation. + * See utypes.h for details. + * + * A pair of macros for a single string must be used with the same + * parameters. + * The string parameter must be a C string literal. + * The length of the string, not including the terminating + * `NUL`, must be specified as a constant. + * The U_STRING_DECL macro should be invoked exactly once for one + * such string variable before it is used. + * + * Usage: + * + * U_STRING_DECL(ustringVar1, "Quick-Fox 2", 11); + * U_STRING_DECL(ustringVar2, "jumps 5%", 8); + * static UBool didInit=false; + * + * int32_t function() { + * if(!didInit) { + * U_STRING_INIT(ustringVar1, "Quick-Fox 2", 11); + * U_STRING_INIT(ustringVar2, "jumps 5%", 8); + * didInit=true; + * } + * return u_strcmp(ustringVar1, ustringVar2); + * } + * + * Note that the macros will NOT consistently work if their argument is another #`define`. + * The following will not work on all platforms, don't use it. + * + * #define GLUCK "Mr. Gluck" + * U_STRING_DECL(var, GLUCK, 9) + * U_STRING_INIT(var, GLUCK, 9) + * + * Instead, use the string literal "Mr. Gluck" as the argument to both macro + * calls. + * + * + * @stable ICU 2.0 + */ +#if defined(U_DECLARE_UTF16) +# define U_STRING_DECL(var, cs, length) static const UChar *var=(const UChar *)U_DECLARE_UTF16(cs) + /**@stable ICU 2.0 */ +# define U_STRING_INIT(var, cs, length) +#elif U_SIZEOF_WCHAR_T==U_SIZEOF_UCHAR && (U_CHARSET_FAMILY==U_ASCII_FAMILY || defined(U_WCHAR_IS_UTF16)) +# define U_STRING_DECL(var, cs, length) static const UChar var[(length)+1]=L ## cs + /**@stable ICU 2.0 */ +# define U_STRING_INIT(var, cs, length) +#else +# define U_STRING_DECL(var, cs, length) static UChar var[(length)+1] + /**@stable ICU 2.0 */ +# define U_STRING_INIT(var, cs, length) u_charsToUChars(cs, var, length+1) +#endif + +/** + * Unescape a string of characters and write the resulting + * Unicode characters to the destination buffer. The following escape + * sequences are recognized: + * + * \\uhhhh 4 hex digits; h in [0-9A-Fa-f] + * \\Uhhhhhhhh 8 hex digits + * \\xhh 1-2 hex digits + * \\x{h...} 1-8 hex digits + * \\ooo 1-3 octal digits; o in [0-7] + * \\cX control-X; X is masked with 0x1F + * + * as well as the standard ANSI C escapes: + * + * \\a => U+0007, \\b => U+0008, \\t => U+0009, \\n => U+000A, + * \\v => U+000B, \\f => U+000C, \\r => U+000D, \\e => U+001B, + * \\" => U+0022, \\' => U+0027, \\? => U+003F, \\\\ => U+005C + * + * Anything else following a backslash is generically escaped. For + * example, "[a\\-z]" returns "[a-z]". + * + * If an escape sequence is ill-formed, this method returns an empty + * string. An example of an ill-formed sequence is "\\u" followed by + * fewer than 4 hex digits. + * + * The above characters are recognized in the compiler's codepage, + * that is, they are coded as 'u', '\\', etc. Characters that are + * not parts of escape sequences are converted using u_charsToUChars(). + * + * This function is similar to UnicodeString::unescape() but not + * identical to it. The latter takes a source UnicodeString, so it + * does escape recognition but no conversion. + * + * @param src a zero-terminated string of invariant characters + * @param dest pointer to buffer to receive converted and unescaped + * text and, if there is room, a zero terminator. May be NULL for + * preflighting, in which case no UChars will be written, but the + * return value will still be valid. On error, an empty string is + * stored here (if possible). + * @param destCapacity the number of UChars that may be written at + * dest. Ignored if dest == NULL. + * @return the length of unescaped string. + * @see u_unescapeAt + * @see UnicodeString#unescape() + * @see UnicodeString#unescapeAt() + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_unescape(const char *src, + UChar *dest, int32_t destCapacity); + +U_CDECL_BEGIN +/** + * Callback function for u_unescapeAt() that returns a character of + * the source text given an offset and a context pointer. The context + * pointer will be whatever is passed into u_unescapeAt(). + * + * @param offset pointer to the offset that will be passed to u_unescapeAt(). + * @param context an opaque pointer passed directly into u_unescapeAt() + * @return the character represented by the escape sequence at + * offset + * @see u_unescapeAt + * @stable ICU 2.0 + */ +typedef UChar (U_CALLCONV *UNESCAPE_CHAR_AT)(int32_t offset, void *context); +U_CDECL_END + +/** + * Unescape a single sequence. The character at offset-1 is assumed + * (without checking) to be a backslash. This method takes a callback + * pointer to a function that returns the UChar at a given offset. By + * varying this callback, ICU functions are able to unescape char* + * strings, UnicodeString objects, and UFILE pointers. + * + * If offset is out of range, or if the escape sequence is ill-formed, + * (UChar32)0xFFFFFFFF is returned. See documentation of u_unescape() + * for a list of recognized sequences. + * + * @param charAt callback function that returns a UChar of the source + * text given an offset and a context pointer. + * @param offset pointer to the offset that will be passed to charAt. + * The offset value will be updated upon return to point after the + * last parsed character of the escape sequence. On error the offset + * is unchanged. + * @param length the number of characters in the source text. The + * last character of the source text is considered to be at offset + * length-1. + * @param context an opaque pointer passed directly into charAt. + * @return the character represented by the escape sequence at + * offset, or (UChar32)0xFFFFFFFF on error. + * @see u_unescape() + * @see UnicodeString#unescape() + * @see UnicodeString#unescapeAt() + * @stable ICU 2.0 + */ +U_CAPI UChar32 U_EXPORT2 +u_unescapeAt(UNESCAPE_CHAR_AT charAt, + int32_t *offset, + int32_t length, + void *context); + +/** + * Uppercase the characters in a string. + * Casing is locale-dependent and context-sensitive. + * The result may be longer or shorter than the original. + * The source string and the destination buffer are allowed to overlap. + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of UChars). If it is 0, then + * dest may be NULL and the function will only return the length of the result + * without writing any of the result string. + * @param src The original string + * @param srcLength The length of the original string. If -1, then src must be zero-terminated. + * @param locale The locale to consider, or "" for the root locale or NULL for the default locale. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return The length of the result string. It may be greater than destCapacity. In that case, + * only some of the result was written to the destination buffer. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_strToUpper(UChar *dest, int32_t destCapacity, + const UChar *src, int32_t srcLength, + const char *locale, + UErrorCode *pErrorCode); + +/** + * Lowercase the characters in a string. + * Casing is locale-dependent and context-sensitive. + * The result may be longer or shorter than the original. + * The source string and the destination buffer are allowed to overlap. + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of UChars). If it is 0, then + * dest may be NULL and the function will only return the length of the result + * without writing any of the result string. + * @param src The original string + * @param srcLength The length of the original string. If -1, then src must be zero-terminated. + * @param locale The locale to consider, or "" for the root locale or NULL for the default locale. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return The length of the result string. It may be greater than destCapacity. In that case, + * only some of the result was written to the destination buffer. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_strToLower(UChar *dest, int32_t destCapacity, + const UChar *src, int32_t srcLength, + const char *locale, + UErrorCode *pErrorCode); + +#if !UCONFIG_NO_BREAK_ITERATION + +/** + * Titlecase a string. + * Casing is locale-dependent and context-sensitive. + * Titlecasing uses a break iterator to find the first characters of words + * that are to be titlecased. It titlecases those characters and lowercases + * all others. + * + * The titlecase break iterator can be provided to customize for arbitrary + * styles, using rules and dictionaries beyond the standard iterators. + * It may be more efficient to always provide an iterator to avoid + * opening and closing one for each string. + * The standard titlecase iterator for the root locale implements the + * algorithm of Unicode TR 21. + * + * This function uses only the setText(), first() and next() methods of the + * provided break iterator. + * + * The result may be longer or shorter than the original. + * The source string and the destination buffer are allowed to overlap. + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of UChars). If it is 0, then + * dest may be NULL and the function will only return the length of the result + * without writing any of the result string. + * @param src The original string + * @param srcLength The length of the original string. If -1, then src must be zero-terminated. + * @param titleIter A break iterator to find the first characters of words + * that are to be titlecased. + * If none is provided (NULL), then a standard titlecase + * break iterator is opened. + * @param locale The locale to consider, or "" for the root locale or NULL for the default locale. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return The length of the result string. It may be greater than destCapacity. In that case, + * only some of the result was written to the destination buffer. + * @stable ICU 2.1 + */ +U_CAPI int32_t U_EXPORT2 +u_strToTitle(UChar *dest, int32_t destCapacity, + const UChar *src, int32_t srcLength, + UBreakIterator *titleIter, + const char *locale, + UErrorCode *pErrorCode); + +#endif + +/** + * Case-folds the characters in a string. + * + * Case-folding is locale-independent and not context-sensitive, + * but there is an option for whether to include or exclude mappings for dotted I + * and dotless i that are marked with 'T' in CaseFolding.txt. + * + * The result may be longer or shorter than the original. + * The source string and the destination buffer are allowed to overlap. + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of UChars). If it is 0, then + * dest may be NULL and the function will only return the length of the result + * without writing any of the result string. + * @param src The original string + * @param srcLength The length of the original string. If -1, then src must be zero-terminated. + * @param options Either U_FOLD_CASE_DEFAULT or U_FOLD_CASE_EXCLUDE_SPECIAL_I + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return The length of the result string. It may be greater than destCapacity. In that case, + * only some of the result was written to the destination buffer. + * @stable ICU 2.0 + */ +U_CAPI int32_t U_EXPORT2 +u_strFoldCase(UChar *dest, int32_t destCapacity, + const UChar *src, int32_t srcLength, + uint32_t options, + UErrorCode *pErrorCode); + +#if defined(U_WCHAR_IS_UTF16) || defined(U_WCHAR_IS_UTF32) || !UCONFIG_NO_CONVERSION +/** + * Convert a UTF-16 string to a wchar_t string. + * If it is known at compile time that wchar_t strings are in UTF-16 or UTF-32, then + * this function simply calls the fast, dedicated function for that. + * Otherwise, two conversions UTF-16 -> default charset -> wchar_t* are performed. + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of wchar_t's). If it is 0, then + * dest may be NULL and the function will only return the length of the + * result without writing any of the result string (pre-flighting). + * @param pDestLength A pointer to receive the number of units written to the destination. If + * pDestLength!=NULL then *pDestLength is always set to the + * number of output units corresponding to the transformation of + * all the input units, even in case of a buffer overflow. + * @param src The original source string + * @param srcLength The length of the original string. If -1, then src must be zero-terminated. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return The pointer to destination buffer. + * @stable ICU 2.0 + */ +U_CAPI wchar_t* U_EXPORT2 +u_strToWCS(wchar_t *dest, + int32_t destCapacity, + int32_t *pDestLength, + const UChar *src, + int32_t srcLength, + UErrorCode *pErrorCode); +/** + * Convert a wchar_t string to UTF-16. + * If it is known at compile time that wchar_t strings are in UTF-16 or UTF-32, then + * this function simply calls the fast, dedicated function for that. + * Otherwise, two conversions wchar_t* -> default charset -> UTF-16 are performed. + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of UChars). If it is 0, then + * dest may be NULL and the function will only return the length of the + * result without writing any of the result string (pre-flighting). + * @param pDestLength A pointer to receive the number of units written to the destination. If + * pDestLength!=NULL then *pDestLength is always set to the + * number of output units corresponding to the transformation of + * all the input units, even in case of a buffer overflow. + * @param src The original source string + * @param srcLength The length of the original string. If -1, then src must be zero-terminated. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return The pointer to destination buffer. + * @stable ICU 2.0 + */ +U_CAPI UChar* U_EXPORT2 +u_strFromWCS(UChar *dest, + int32_t destCapacity, + int32_t *pDestLength, + const wchar_t *src, + int32_t srcLength, + UErrorCode *pErrorCode); +#endif /* defined(U_WCHAR_IS_UTF16) || defined(U_WCHAR_IS_UTF32) || !UCONFIG_NO_CONVERSION */ + +/** + * Convert a UTF-16 string to UTF-8. + * If the input string is not well-formed, then the U_INVALID_CHAR_FOUND error code is set. + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of chars). If it is 0, then + * dest may be NULL and the function will only return the length of the + * result without writing any of the result string (pre-flighting). + * @param pDestLength A pointer to receive the number of units written to the destination. If + * pDestLength!=NULL then *pDestLength is always set to the + * number of output units corresponding to the transformation of + * all the input units, even in case of a buffer overflow. + * @param src The original source string + * @param srcLength The length of the original string. If -1, then src must be zero-terminated. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return The pointer to destination buffer. + * @stable ICU 2.0 + * @see u_strToUTF8WithSub + * @see u_strFromUTF8 + */ +U_CAPI char* U_EXPORT2 +u_strToUTF8(char *dest, + int32_t destCapacity, + int32_t *pDestLength, + const UChar *src, + int32_t srcLength, + UErrorCode *pErrorCode); + +/** + * Convert a UTF-8 string to UTF-16. + * If the input string is not well-formed, then the U_INVALID_CHAR_FOUND error code is set. + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of UChars). If it is 0, then + * dest may be NULL and the function will only return the length of the + * result without writing any of the result string (pre-flighting). + * @param pDestLength A pointer to receive the number of units written to the destination. If + * pDestLength!=NULL then *pDestLength is always set to the + * number of output units corresponding to the transformation of + * all the input units, even in case of a buffer overflow. + * @param src The original source string + * @param srcLength The length of the original string. If -1, then src must be zero-terminated. + * @param pErrorCode Must be a valid pointer to an error code value, + * which must not indicate a failure before the function call. + * @return The pointer to destination buffer. + * @stable ICU 2.0 + * @see u_strFromUTF8WithSub + * @see u_strFromUTF8Lenient + */ +U_CAPI UChar* U_EXPORT2 +u_strFromUTF8(UChar *dest, + int32_t destCapacity, + int32_t *pDestLength, + const char *src, + int32_t srcLength, + UErrorCode *pErrorCode); + +/** + * Convert a UTF-16 string to UTF-8. + * + * Same as u_strToUTF8() except for the additional subchar which is output for + * illegal input sequences, instead of stopping with the U_INVALID_CHAR_FOUND error code. + * With subchar==U_SENTINEL, this function behaves exactly like u_strToUTF8(). + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of chars). If it is 0, then + * dest may be NULL and the function will only return the length of the + * result without writing any of the result string (pre-flighting). + * @param pDestLength A pointer to receive the number of units written to the destination. If + * pDestLength!=NULL then *pDestLength is always set to the + * number of output units corresponding to the transformation of + * all the input units, even in case of a buffer overflow. + * @param src The original source string + * @param srcLength The length of the original string. If -1, then src must be zero-terminated. + * @param subchar The substitution character to use in place of an illegal input sequence, + * or U_SENTINEL if the function is to return with U_INVALID_CHAR_FOUND instead. + * A substitution character can be any valid Unicode code point (up to U+10FFFF) + * except for surrogate code points (U+D800..U+DFFF). + * The recommended value is U+FFFD "REPLACEMENT CHARACTER". + * @param pNumSubstitutions Output parameter receiving the number of substitutions if subchar>=0. + * Set to 0 if no substitutions occur or subchar<0. + * pNumSubstitutions can be NULL. + * @param pErrorCode Pointer to a standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return The pointer to destination buffer. + * @see u_strToUTF8 + * @see u_strFromUTF8WithSub + * @stable ICU 3.6 + */ +U_CAPI char* U_EXPORT2 +u_strToUTF8WithSub(char *dest, + int32_t destCapacity, + int32_t *pDestLength, + const UChar *src, + int32_t srcLength, + UChar32 subchar, int32_t *pNumSubstitutions, + UErrorCode *pErrorCode); + +/** + * Convert a UTF-8 string to UTF-16. + * + * Same as u_strFromUTF8() except for the additional subchar which is output for + * illegal input sequences, instead of stopping with the U_INVALID_CHAR_FOUND error code. + * With subchar==U_SENTINEL, this function behaves exactly like u_strFromUTF8(). + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of UChars). If it is 0, then + * dest may be NULL and the function will only return the length of the + * result without writing any of the result string (pre-flighting). + * @param pDestLength A pointer to receive the number of units written to the destination. If + * pDestLength!=NULL then *pDestLength is always set to the + * number of output units corresponding to the transformation of + * all the input units, even in case of a buffer overflow. + * @param src The original source string + * @param srcLength The length of the original string. If -1, then src must be zero-terminated. + * @param subchar The substitution character to use in place of an illegal input sequence, + * or U_SENTINEL if the function is to return with U_INVALID_CHAR_FOUND instead. + * A substitution character can be any valid Unicode code point (up to U+10FFFF) + * except for surrogate code points (U+D800..U+DFFF). + * The recommended value is U+FFFD "REPLACEMENT CHARACTER". + * @param pNumSubstitutions Output parameter receiving the number of substitutions if subchar>=0. + * Set to 0 if no substitutions occur or subchar<0. + * pNumSubstitutions can be NULL. + * @param pErrorCode Pointer to a standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return The pointer to destination buffer. + * @see u_strFromUTF8 + * @see u_strFromUTF8Lenient + * @see u_strToUTF8WithSub + * @stable ICU 3.6 + */ +U_CAPI UChar* U_EXPORT2 +u_strFromUTF8WithSub(UChar *dest, + int32_t destCapacity, + int32_t *pDestLength, + const char *src, + int32_t srcLength, + UChar32 subchar, int32_t *pNumSubstitutions, + UErrorCode *pErrorCode); + +/** + * Convert a UTF-8 string to UTF-16. + * + * Same as u_strFromUTF8() except that this function is designed to be very fast, + * which it achieves by being lenient about malformed UTF-8 sequences. + * This function is intended for use in environments where UTF-8 text is + * expected to be well-formed. + * + * Its semantics are: + * - Well-formed UTF-8 text is correctly converted to well-formed UTF-16 text. + * - The function will not read beyond the input string, nor write beyond + * the destCapacity. + * - Malformed UTF-8 results in "garbage" 16-bit Unicode strings which may not + * be well-formed UTF-16. + * The function will resynchronize to valid code point boundaries + * within a small number of code points after an illegal sequence. + * - Non-shortest forms are not detected and will result in "spoofing" output. + * + * For further performance improvement, if srcLength is given (>=0), + * then it must be destCapacity>=srcLength. + * + * There is no inverse u_strToUTF8Lenient() function because there is practically + * no performance gain from not checking that a UTF-16 string is well-formed. + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of UChars). If it is 0, then + * dest may be NULL and the function will only return the length of the + * result without writing any of the result string (pre-flighting). + * Unlike for other ICU functions, if srcLength>=0 then it + * must be destCapacity>=srcLength. + * @param pDestLength A pointer to receive the number of units written to the destination. If + * pDestLength!=NULL then *pDestLength is always set to the + * number of output units corresponding to the transformation of + * all the input units, even in case of a buffer overflow. + * Unlike for other ICU functions, if srcLength>=0 but + * destCapacity=0. + * Set to 0 if no substitutions occur or subchar<0. + * pNumSubstitutions can be NULL. + * @param pErrorCode Pointer to a standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return The pointer to destination buffer. + * @see u_strToUTF32 + * @see u_strFromUTF32WithSub + * @stable ICU 4.2 + */ +U_CAPI UChar32* U_EXPORT2 +u_strToUTF32WithSub(UChar32 *dest, + int32_t destCapacity, + int32_t *pDestLength, + const UChar *src, + int32_t srcLength, + UChar32 subchar, int32_t *pNumSubstitutions, + UErrorCode *pErrorCode); + +/** + * Convert a UTF-32 string to UTF-16. + * + * Same as u_strFromUTF32() except for the additional subchar which is output for + * illegal input sequences, instead of stopping with the U_INVALID_CHAR_FOUND error code. + * With subchar==U_SENTINEL, this function behaves exactly like u_strFromUTF32(). + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of UChars). If it is 0, then + * dest may be NULL and the function will only return the length of the + * result without writing any of the result string (pre-flighting). + * @param pDestLength A pointer to receive the number of units written to the destination. If + * pDestLength!=NULL then *pDestLength is always set to the + * number of output units corresponding to the transformation of + * all the input units, even in case of a buffer overflow. + * @param src The original source string + * @param srcLength The length of the original string. If -1, then src must be zero-terminated. + * @param subchar The substitution character to use in place of an illegal input sequence, + * or U_SENTINEL if the function is to return with U_INVALID_CHAR_FOUND instead. + * A substitution character can be any valid Unicode code point (up to U+10FFFF) + * except for surrogate code points (U+D800..U+DFFF). + * The recommended value is U+FFFD "REPLACEMENT CHARACTER". + * @param pNumSubstitutions Output parameter receiving the number of substitutions if subchar>=0. + * Set to 0 if no substitutions occur or subchar<0. + * pNumSubstitutions can be NULL. + * @param pErrorCode Pointer to a standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return The pointer to destination buffer. + * @see u_strFromUTF32 + * @see u_strToUTF32WithSub + * @stable ICU 4.2 + */ +U_CAPI UChar* U_EXPORT2 +u_strFromUTF32WithSub(UChar *dest, + int32_t destCapacity, + int32_t *pDestLength, + const UChar32 *src, + int32_t srcLength, + UChar32 subchar, int32_t *pNumSubstitutions, + UErrorCode *pErrorCode); + +/** + * Convert a 16-bit Unicode string to Java Modified UTF-8. + * See http://java.sun.com/javase/6/docs/api/java/io/DataInput.html#modified-utf-8 + * + * This function behaves according to the documentation for Java DataOutput.writeUTF() + * except that it does not encode the output length in the destination buffer + * and does not have an output length restriction. + * See http://java.sun.com/javase/6/docs/api/java/io/DataOutput.html#writeUTF(java.lang.String) + * + * The input string need not be well-formed UTF-16. + * (Therefore there is no subchar parameter.) + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of chars). If it is 0, then + * dest may be NULL and the function will only return the length of the + * result without writing any of the result string (pre-flighting). + * @param pDestLength A pointer to receive the number of units written to the destination. If + * pDestLength!=NULL then *pDestLength is always set to the + * number of output units corresponding to the transformation of + * all the input units, even in case of a buffer overflow. + * @param src The original source string + * @param srcLength The length of the original string. If -1, then src must be zero-terminated. + * @param pErrorCode Pointer to a standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return The pointer to destination buffer. + * @stable ICU 4.4 + * @see u_strToUTF8WithSub + * @see u_strFromJavaModifiedUTF8WithSub + */ +U_CAPI char* U_EXPORT2 +u_strToJavaModifiedUTF8( + char *dest, + int32_t destCapacity, + int32_t *pDestLength, + const UChar *src, + int32_t srcLength, + UErrorCode *pErrorCode); + +/** + * Convert a Java Modified UTF-8 string to a 16-bit Unicode string. + * If the input string is not well-formed and no substitution char is specified, + * then the U_INVALID_CHAR_FOUND error code is set. + * + * This function behaves according to the documentation for Java DataInput.readUTF() + * except that it takes a length parameter rather than + * interpreting the first two input bytes as the length. + * See http://java.sun.com/javase/6/docs/api/java/io/DataInput.html#readUTF() + * + * The output string may not be well-formed UTF-16. + * + * @param dest A buffer for the result string. The result will be zero-terminated if + * the buffer is large enough. + * @param destCapacity The size of the buffer (number of UChars). If it is 0, then + * dest may be NULL and the function will only return the length of the + * result without writing any of the result string (pre-flighting). + * @param pDestLength A pointer to receive the number of units written to the destination. If + * pDestLength!=NULL then *pDestLength is always set to the + * number of output units corresponding to the transformation of + * all the input units, even in case of a buffer overflow. + * @param src The original source string + * @param srcLength The length of the original string. If -1, then src must be zero-terminated. + * @param subchar The substitution character to use in place of an illegal input sequence, + * or U_SENTINEL if the function is to return with U_INVALID_CHAR_FOUND instead. + * A substitution character can be any valid Unicode code point (up to U+10FFFF) + * except for surrogate code points (U+D800..U+DFFF). + * The recommended value is U+FFFD "REPLACEMENT CHARACTER". + * @param pNumSubstitutions Output parameter receiving the number of substitutions if subchar>=0. + * Set to 0 if no substitutions occur or subchar<0. + * pNumSubstitutions can be NULL. + * @param pErrorCode Pointer to a standard ICU error code. Its input value must + * pass the U_SUCCESS() test, or else the function returns + * immediately. Check for U_FAILURE() on output or use with + * function chaining. (See User Guide for details.) + * @return The pointer to destination buffer. + * @see u_strFromUTF8WithSub + * @see u_strFromUTF8Lenient + * @see u_strToJavaModifiedUTF8 + * @stable ICU 4.4 + */ +U_CAPI UChar* U_EXPORT2 +u_strFromJavaModifiedUTF8WithSub( + UChar *dest, + int32_t destCapacity, + int32_t *pDestLength, + const char *src, + int32_t srcLength, + UChar32 subchar, int32_t *pNumSubstitutions, + UErrorCode *pErrorCode); + +#endif diff --git a/intl/icu/source/common/unicode/ustringtrie.h b/intl/icu/source/common/unicode/ustringtrie.h new file mode 100644 index 0000000000..fd85648225 --- /dev/null +++ b/intl/icu/source/common/unicode/ustringtrie.h @@ -0,0 +1,97 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* Copyright (C) 2010-2012, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +* file name: udicttrie.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2010dec17 +* created by: Markus W. Scherer +*/ + +#ifndef __USTRINGTRIE_H__ +#define __USTRINGTRIE_H__ + +/** + * \file + * \brief C API: Helper definitions for dictionary trie APIs. + */ + +#include "unicode/utypes.h" + + +/** + * Return values for BytesTrie::next(), UCharsTrie::next() and similar methods. + * @see USTRINGTRIE_MATCHES + * @see USTRINGTRIE_HAS_VALUE + * @see USTRINGTRIE_HAS_NEXT + * @stable ICU 4.8 + */ +enum UStringTrieResult { + /** + * The input unit(s) did not continue a matching string. + * Once current()/next() return USTRINGTRIE_NO_MATCH, + * all further calls to current()/next() will also return USTRINGTRIE_NO_MATCH, + * until the trie is reset to its original state or to a saved state. + * @stable ICU 4.8 + */ + USTRINGTRIE_NO_MATCH, + /** + * The input unit(s) continued a matching string + * but there is no value for the string so far. + * (It is a prefix of a longer string.) + * @stable ICU 4.8 + */ + USTRINGTRIE_NO_VALUE, + /** + * The input unit(s) continued a matching string + * and there is a value for the string so far. + * This value will be returned by getValue(). + * No further input byte/unit can continue a matching string. + * @stable ICU 4.8 + */ + USTRINGTRIE_FINAL_VALUE, + /** + * The input unit(s) continued a matching string + * and there is a value for the string so far. + * This value will be returned by getValue(). + * Another input byte/unit can continue a matching string. + * @stable ICU 4.8 + */ + USTRINGTRIE_INTERMEDIATE_VALUE +}; + +/** + * Same as (result!=USTRINGTRIE_NO_MATCH). + * @param result A result from BytesTrie::first(), UCharsTrie::next() etc. + * @return true if the input bytes/units so far are part of a matching string/byte sequence. + * @stable ICU 4.8 + */ +#define USTRINGTRIE_MATCHES(result) ((result)!=USTRINGTRIE_NO_MATCH) + +/** + * Equivalent to (result==USTRINGTRIE_INTERMEDIATE_VALUE || result==USTRINGTRIE_FINAL_VALUE) but + * this macro evaluates result exactly once. + * @param result A result from BytesTrie::first(), UCharsTrie::next() etc. + * @return true if there is a value for the input bytes/units so far. + * @see BytesTrie::getValue + * @see UCharsTrie::getValue + * @stable ICU 4.8 + */ +#define USTRINGTRIE_HAS_VALUE(result) ((result)>=USTRINGTRIE_FINAL_VALUE) + +/** + * Equivalent to (result==USTRINGTRIE_NO_VALUE || result==USTRINGTRIE_INTERMEDIATE_VALUE) but + * this macro evaluates result exactly once. + * @param result A result from BytesTrie::first(), UCharsTrie::next() etc. + * @return true if another input byte/unit can continue a matching string. + * @stable ICU 4.8 + */ +#define USTRINGTRIE_HAS_NEXT(result) ((result)&1) + +#endif /* __USTRINGTRIE_H__ */ diff --git a/intl/icu/source/common/unicode/utext.h b/intl/icu/source/common/unicode/utext.h new file mode 100644 index 0000000000..423b281631 --- /dev/null +++ b/intl/icu/source/common/unicode/utext.h @@ -0,0 +1,1603 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 2004-2012, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: utext.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 2004oct06 +* created by: Markus W. Scherer +*/ + +#ifndef __UTEXT_H__ +#define __UTEXT_H__ + +/** + * \file + * \brief C API: Abstract Unicode Text API + * + * The Text Access API provides a means to allow text that is stored in alternative + * formats to work with ICU services. ICU normally operates on text that is + * stored in UTF-16 format, in (UChar *) arrays for the C APIs or as type + * UnicodeString for C++ APIs. + * + * ICU Text Access allows other formats, such as UTF-8 or non-contiguous + * UTF-16 strings, to be placed in a UText wrapper and then passed to ICU services. + * + * There are three general classes of usage for UText: + * + * Application Level Use. This is the simplest usage - applications would + * use one of the utext_open() functions on their input text, and pass + * the resulting UText to the desired ICU service. + * + * Second is usage in ICU Services, such as break iteration, that will need to + * operate on input presented to them as a UText. These implementations + * will need to use the iteration and related UText functions to gain + * access to the actual text. + * + * The third class of UText users are "text providers." These are the + * UText implementations for the various text storage formats. An application + * or system with a unique text storage format can implement a set of + * UText provider functions for that format, which will then allow + * ICU services to operate on that format. + * + * + * Iterating over text + * + * Here is sample code for a forward iteration over the contents of a UText + * + * \code + * UChar32 c; + * UText *ut = whatever(); + * + * for (c=utext_next32From(ut, 0); c>=0; c=utext_next32(ut)) { + * // do whatever with the codepoint c here. + * } + * \endcode + * + * And here is similar code to iterate in the reverse direction, from the end + * of the text towards the beginning. + * + * \code + * UChar32 c; + * UText *ut = whatever(); + * int textLength = utext_nativeLength(ut); + * for (c=utext_previous32From(ut, textLength); c>=0; c=utext_previous32(ut)) { + * // do whatever with the codepoint c here. + * } + * \endcode + * + * Characters and Indexing + * + * Indexing into text by UText functions is nearly always in terms of the native + * indexing of the underlying text storage. The storage format could be UTF-8 + * or UTF-32, for example. When coding to the UText access API, no assumptions + * can be made regarding the size of characters, or how far an index + * may move when iterating between characters. + * + * All indices supplied to UText functions are pinned to the length of the + * text. An out-of-bounds index is not considered to be an error, but is + * adjusted to be in the range 0 <= index <= length of input text. + * + * + * When an index position is returned from a UText function, it will be + * a native index to the underlying text. In the case of multi-unit characters, + * it will always refer to the first position of the character, + * never to the interior. This is essentially the same thing as saying that + * a returned index will always point to a boundary between characters. + * + * When a native index is supplied to a UText function, all indices that + * refer to any part of a multi-unit character representation are considered + * to be equivalent. In the case of multi-unit characters, an incoming index + * will be logically normalized to refer to the start of the character. + * + * It is possible to test whether a native index is on a code point boundary + * by doing a utext_setNativeIndex() followed by a utext_getNativeIndex(). + * If the index is returned unchanged, it was on a code point boundary. If + * an adjusted index is returned, the original index referred to the + * interior of a character. + * + * Conventions for calling UText functions + * + * Most UText access functions have as their first parameter a (UText *) pointer, + * which specifies the UText to be used. Unless otherwise noted, the + * pointer must refer to a valid, open UText. Attempting to + * use a closed UText or passing a NULL pointer is a programming error and + * will produce undefined results or NULL pointer exceptions. + * + * The UText_Open family of functions can either open an existing (closed) + * UText, or heap allocate a new UText. Here is sample code for creating + * a stack-allocated UText. + * + * \code + * char *s = whatever(); // A utf-8 string + * U_ErrorCode status = U_ZERO_ERROR; + * UText ut = UTEXT_INITIALIZER; + * utext_openUTF8(ut, s, -1, &status); + * if (U_FAILURE(status)) { + * // error handling + * } else { + * // work with the UText + * } + * \endcode + * + * Any existing UText passed to an open function _must_ have been initialized, + * either by the UTEXT_INITIALIZER, or by having been originally heap-allocated + * by an open function. Passing NULL will cause the open function to + * heap-allocate and fully initialize a new UText. + * + */ + + + +#include "unicode/utypes.h" +#include "unicode/uchar.h" +#if U_SHOW_CPLUSPLUS_API +#include "unicode/localpointer.h" +#include "unicode/rep.h" +#include "unicode/unistr.h" +#include "unicode/chariter.h" +#endif + + +U_CDECL_BEGIN + +struct UText; +typedef struct UText UText; /**< C typedef for struct UText. @stable ICU 3.6 */ + + +/*************************************************************************************** + * + * C Functions for creating UText wrappers around various kinds of text strings. + * + ****************************************************************************************/ + + +/** + * Close function for UText instances. + * Cleans up, releases any resources being held by an open UText. + *

+ * If the UText was originally allocated by one of the utext_open functions, + * the storage associated with the utext will also be freed. + * If the UText storage originated with the application, as it would with + * a local or static instance, the storage will not be deleted. + * + * An open UText can be reset to refer to new string by using one of the utext_open() + * functions without first closing the UText. + * + * @param ut The UText to be closed. + * @return NULL if the UText struct was deleted by the close. If the UText struct + * was originally provided by the caller to the open function, it is + * returned by this function, and may be safely used again in + * a subsequent utext_open. + * + * @stable ICU 3.4 + */ +U_CAPI UText * U_EXPORT2 +utext_close(UText *ut); + +/** + * Open a read-only UText implementation for UTF-8 strings. + * + * \htmlonly + * Any invalid UTF-8 in the input will be handled in this way: + * a sequence of bytes that has the form of a truncated, but otherwise valid, + * UTF-8 sequence will be replaced by a single unicode replacement character, \uFFFD. + * Any other illegal bytes will each be replaced by a \uFFFD. + * \endhtmlonly + * + * @param ut Pointer to a UText struct. If NULL, a new UText will be created. + * If non-NULL, must refer to an initialized UText struct, which will then + * be reset to reference the specified UTF-8 string. + * @param s A UTF-8 string. Must not be NULL. + * @param length The length of the UTF-8 string in bytes, or -1 if the string is + * zero terminated. + * @param status Errors are returned here. + * @return A pointer to the UText. If a pre-allocated UText was provided, it + * will always be used and returned. + * @stable ICU 3.4 + */ +U_CAPI UText * U_EXPORT2 +utext_openUTF8(UText *ut, const char *s, int64_t length, UErrorCode *status); + + +/** + * Open a read-only UText for UChar * string. + * + * @param ut Pointer to a UText struct. If NULL, a new UText will be created. + * If non-NULL, must refer to an initialized UText struct, which will then + * be reset to reference the specified UChar string. + * @param s A UChar (UTF-16) string + * @param length The number of UChars in the input string, or -1 if the string is + * zero terminated. + * @param status Errors are returned here. + * @return A pointer to the UText. If a pre-allocated UText was provided, it + * will always be used and returned. + * @stable ICU 3.4 + */ +U_CAPI UText * U_EXPORT2 +utext_openUChars(UText *ut, const UChar *s, int64_t length, UErrorCode *status); + + +#if U_SHOW_CPLUSPLUS_API +/** + * Open a writable UText for a non-const UnicodeString. + * + * @param ut Pointer to a UText struct. If nullptr, a new UText will be created. + * If non-nullptr, must refer to an initialized UText struct, which will then + * be reset to reference the specified input string. + * @param s A UnicodeString. + * @param status Errors are returned here. + * @return Pointer to the UText. If a UText was supplied as input, this + * will always be used and returned. + * @stable ICU 3.4 + */ +U_CAPI UText * U_EXPORT2 +utext_openUnicodeString(UText *ut, icu::UnicodeString *s, UErrorCode *status); + + +/** + * Open a UText for a const UnicodeString. The resulting UText will not be writable. + * + * @param ut Pointer to a UText struct. If nullptr, a new UText will be created. + * If non-nullptr, must refer to an initialized UText struct, which will then + * be reset to reference the specified input string. + * @param s A const UnicodeString to be wrapped. + * @param status Errors are returned here. + * @return Pointer to the UText. If a UText was supplied as input, this + * will always be used and returned. + * @stable ICU 3.4 + */ +U_CAPI UText * U_EXPORT2 +utext_openConstUnicodeString(UText *ut, const icu::UnicodeString *s, UErrorCode *status); + + +/** + * Open a writable UText implementation for an ICU Replaceable object. + * @param ut Pointer to a UText struct. If nullptr, a new UText will be created. + * If non-nullptr, must refer to an already existing UText, which will then + * be reset to reference the specified replaceable text. + * @param rep A Replaceable text object. + * @param status Errors are returned here. + * @return Pointer to the UText. If a UText was supplied as input, this + * will always be used and returned. + * @see Replaceable + * @stable ICU 3.4 + */ +U_CAPI UText * U_EXPORT2 +utext_openReplaceable(UText *ut, icu::Replaceable *rep, UErrorCode *status); + +/** + * Open a UText implementation over an ICU CharacterIterator. + * @param ut Pointer to a UText struct. If nullptr, a new UText will be created. + * If non-nullptr, must refer to an already existing UText, which will then + * be reset to reference the specified replaceable text. + * @param ci A Character Iterator. + * @param status Errors are returned here. + * @return Pointer to the UText. If a UText was supplied as input, this + * will always be used and returned. + * @see Replaceable + * @stable ICU 3.4 + */ +U_CAPI UText * U_EXPORT2 +utext_openCharacterIterator(UText *ut, icu::CharacterIterator *ci, UErrorCode *status); + +#endif + + +/** + * Clone a UText. This is much like opening a UText where the source text is itself + * another UText. + * + * A deep clone will copy both the UText data structures and the underlying text. + * The original and cloned UText will operate completely independently; modifications + * made to the text in one will not affect the other. Text providers are not + * required to support deep clones. The user of clone() must check the status return + * and be prepared to handle failures. + * + * The standard UText implementations for UTF8, UChar *, UnicodeString and + * Replaceable all support deep cloning. + * + * The UText returned from a deep clone will be writable, assuming that the text + * provider is able to support writing, even if the source UText had been made + * non-writable by means of UText_freeze(). + * + * A shallow clone replicates only the UText data structures; it does not make + * a copy of the underlying text. Shallow clones can be used as an efficient way to + * have multiple iterators active in a single text string that is not being + * modified. + * + * A shallow clone operation will not fail, barring truly exceptional conditions such + * as memory allocation failures. + * + * Shallow UText clones should be avoided if the UText functions that modify the + * text are expected to be used, either on the original or the cloned UText. + * Any such modifications can cause unpredictable behavior. Read Only + * shallow clones provide some protection against errors of this type by + * disabling text modification via the cloned UText. + * + * A shallow clone made with the readOnly parameter == false will preserve the + * utext_isWritable() state of the source object. Note, however, that + * write operations must be avoided while more than one UText exists that refer + * to the same underlying text. + * + * A UText and its clone may be safely concurrently accessed by separate threads. + * This is true for read access only with shallow clones, and for both read and + * write access with deep clones. + * It is the responsibility of the Text Provider to ensure that this thread safety + * constraint is met. + * + * @param dest A UText struct to be filled in with the result of the clone operation, + * or NULL if the clone function should heap-allocate a new UText struct. + * If non-NULL, must refer to an already existing UText, which will then + * be reset to become the clone. + * @param src The UText to be cloned. + * @param deep true to request a deep clone, false for a shallow clone. + * @param readOnly true to request that the cloned UText have read only access to the + * underlying text. + + * @param status Errors are returned here. For deep clones, U_UNSUPPORTED_ERROR + * will be returned if the text provider is unable to clone the + * original text. + * @return The newly created clone, or NULL if the clone operation failed. + * @stable ICU 3.4 + */ +U_CAPI UText * U_EXPORT2 +utext_clone(UText *dest, const UText *src, UBool deep, UBool readOnly, UErrorCode *status); + + +/** + * Compare two UText objects for equality. + * UTexts are equal if they are iterating over the same text, and + * have the same iteration position within the text. + * If either or both of the parameters are NULL, the comparison is false. + * + * @param a The first of the two UTexts to compare. + * @param b The other UText to be compared. + * @return true if the two UTexts are equal. + * @stable ICU 3.6 + */ +U_CAPI UBool U_EXPORT2 +utext_equals(const UText *a, const UText *b); + + +/***************************************************************************** + * + * Functions to work with the text represented by a UText wrapper + * + *****************************************************************************/ + +/** + * Get the length of the text. Depending on the characteristics + * of the underlying text representation, this may be expensive. + * @see utext_isLengthExpensive() + * + * + * @param ut the text to be accessed. + * @return the length of the text, expressed in native units. + * + * @stable ICU 3.4 + */ +U_CAPI int64_t U_EXPORT2 +utext_nativeLength(UText *ut); + +/** + * Return true if calculating the length of the text could be expensive. + * Finding the length of NUL terminated strings is considered to be expensive. + * + * Note that the value of this function may change + * as the result of other operations on a UText. + * Once the length of a string has been discovered, it will no longer + * be expensive to report it. + * + * @param ut the text to be accessed. + * @return true if determining the length of the text could be time consuming. + * @stable ICU 3.4 + */ +U_CAPI UBool U_EXPORT2 +utext_isLengthExpensive(const UText *ut); + +/** + * Returns the code point at the requested index, + * or U_SENTINEL (-1) if it is out of bounds. + * + * If the specified index points to the interior of a multi-unit + * character - one of the trail bytes of a UTF-8 sequence, for example - + * the complete code point will be returned. + * + * The iteration position will be set to the start of the returned code point. + * + * This function is roughly equivalent to the sequence + * utext_setNativeIndex(index); + * utext_current32(); + * (There is a subtle difference if the index is out of bounds by being less than zero - + * utext_setNativeIndex(negative value) sets the index to zero, after which utext_current() + * will return the char at zero. utext_char32At(negative index), on the other hand, will + * return the U_SENTINEL value of -1.) + * + * @param ut the text to be accessed + * @param nativeIndex the native index of the character to be accessed. If the index points + * to other than the first unit of a multi-unit character, it will be adjusted + * to the start of the character. + * @return the code point at the specified index. + * @stable ICU 3.4 + */ +U_CAPI UChar32 U_EXPORT2 +utext_char32At(UText *ut, int64_t nativeIndex); + + +/** + * + * Get the code point at the current iteration position, + * or U_SENTINEL (-1) if the iteration has reached the end of + * the input text. + * + * @param ut the text to be accessed. + * @return the Unicode code point at the current iterator position. + * @stable ICU 3.4 + */ +U_CAPI UChar32 U_EXPORT2 +utext_current32(UText *ut); + + +/** + * Get the code point at the current iteration position of the UText, and + * advance the position to the first index following the character. + * + * If the position is at the end of the text (the index following + * the last character, which is also the length of the text), + * return U_SENTINEL (-1) and do not advance the index. + * + * This is a post-increment operation. + * + * An inline macro version of this function, UTEXT_NEXT32(), + * is available for performance critical use. + * + * @param ut the text to be accessed. + * @return the Unicode code point at the iteration position. + * @see UTEXT_NEXT32 + * @stable ICU 3.4 + */ +U_CAPI UChar32 U_EXPORT2 +utext_next32(UText *ut); + + +/** + * Move the iterator position to the character (code point) whose + * index precedes the current position, and return that character. + * This is a pre-decrement operation. + * + * If the initial position is at the start of the text (index of 0) + * return U_SENTINEL (-1), and leave the position unchanged. + * + * An inline macro version of this function, UTEXT_PREVIOUS32(), + * is available for performance critical use. + * + * @param ut the text to be accessed. + * @return the previous UChar32 code point, or U_SENTINEL (-1) + * if the iteration has reached the start of the text. + * @see UTEXT_PREVIOUS32 + * @stable ICU 3.4 + */ +U_CAPI UChar32 U_EXPORT2 +utext_previous32(UText *ut); + + +/** + * Set the iteration index and return the code point at that index. + * Leave the iteration index at the start of the following code point. + * + * This function is the most efficient and convenient way to + * begin a forward iteration. The results are identical to the those + * from the sequence + * \code + * utext_setIndex(); + * utext_next32(); + * \endcode + * + * @param ut the text to be accessed. + * @param nativeIndex Iteration index, in the native units of the text provider. + * @return Code point which starts at or before index, + * or U_SENTINEL (-1) if it is out of bounds. + * @stable ICU 3.4 + */ +U_CAPI UChar32 U_EXPORT2 +utext_next32From(UText *ut, int64_t nativeIndex); + + + +/** + * Set the iteration index, and return the code point preceding the + * one specified by the initial index. Leave the iteration position + * at the start of the returned code point. + * + * This function is the most efficient and convenient way to + * begin a backwards iteration. + * + * @param ut the text to be accessed. + * @param nativeIndex Iteration index in the native units of the text provider. + * @return Code point preceding the one at the initial index, + * or U_SENTINEL (-1) if it is out of bounds. + * + * @stable ICU 3.4 + */ +U_CAPI UChar32 U_EXPORT2 +utext_previous32From(UText *ut, int64_t nativeIndex); + +/** + * Get the current iterator position, which can range from 0 to + * the length of the text. + * The position is a native index into the input text, in whatever format it + * may have (possibly UTF-8 for example), and may not always be the same as + * the corresponding UChar (UTF-16) index. + * The returned position will always be aligned to a code point boundary. + * + * @param ut the text to be accessed. + * @return the current index position, in the native units of the text provider. + * @stable ICU 3.4 + */ +U_CAPI int64_t U_EXPORT2 +utext_getNativeIndex(const UText *ut); + +/** + * Set the current iteration position to the nearest code point + * boundary at or preceding the specified index. + * The index is in the native units of the original input text. + * If the index is out of range, it will be pinned to be within + * the range of the input text. + *

+ * It will usually be more efficient to begin an iteration + * using the functions utext_next32From() or utext_previous32From() + * rather than setIndex(). + *

+ * Moving the index position to an adjacent character is best done + * with utext_next32(), utext_previous32() or utext_moveIndex32(). + * Attempting to do direct arithmetic on the index position is + * complicated by the fact that the size (in native units) of a + * character depends on the underlying representation of the character + * (UTF-8, UTF-16, UTF-32, arbitrary codepage), and is not + * easily knowable. + * + * @param ut the text to be accessed. + * @param nativeIndex the native unit index of the new iteration position. + * @stable ICU 3.4 + */ +U_CAPI void U_EXPORT2 +utext_setNativeIndex(UText *ut, int64_t nativeIndex); + +/** + * Move the iterator position by delta code points. The number of code points + * is a signed number; a negative delta will move the iterator backwards, + * towards the start of the text. + *

+ * The index is moved by delta code points + * forward or backward, but no further backward than to 0 and + * no further forward than to utext_nativeLength(). + * The resulting index value will be in between 0 and length, inclusive. + * + * @param ut the text to be accessed. + * @param delta the signed number of code points to move the iteration position. + * @return true if the position could be moved the requested number of positions while + * staying within the range [0 - text length]. + * @stable ICU 3.4 + */ +U_CAPI UBool U_EXPORT2 +utext_moveIndex32(UText *ut, int32_t delta); + +/** + * Get the native index of the character preceding the current position. + * If the iteration position is already at the start of the text, zero + * is returned. + * The value returned is the same as that obtained from the following sequence, + * but without the side effect of changing the iteration position. + * + * \code + * UText *ut = whatever; + * ... + * utext_previous(ut) + * utext_getNativeIndex(ut); + * \endcode + * + * This function is most useful during forwards iteration, where it will get the + * native index of the character most recently returned from utext_next(). + * + * @param ut the text to be accessed + * @return the native index of the character preceding the current index position, + * or zero if the current position is at the start of the text. + * @stable ICU 3.6 + */ +U_CAPI int64_t U_EXPORT2 +utext_getPreviousNativeIndex(UText *ut); + + +/** + * + * Extract text from a UText into a UChar buffer. The range of text to be extracted + * is specified in the native indices of the UText provider. These may not necessarily + * be UTF-16 indices. + *

+ * The size (number of 16 bit UChars) of the data to be extracted is returned. The + * full number of UChars is returned, even when the extracted text is truncated + * because the specified buffer size is too small. + *

+ * The extracted string will (if you are a user) / must (if you are a text provider) + * be NUL-terminated if there is sufficient space in the destination buffer. This + * terminating NUL is not included in the returned length. + *

+ * The iteration index is left at the position following the last extracted character. + * + * @param ut the UText from which to extract data. + * @param nativeStart the native index of the first character to extract.\ + * If the specified index is out of range, + * it will be pinned to be within 0 <= index <= textLength + * @param nativeLimit the native string index of the position following the last + * character to extract. If the specified index is out of range, + * it will be pinned to be within 0 <= index <= textLength. + * nativeLimit must be >= nativeStart. + * @param dest the UChar (UTF-16) buffer into which the extracted text is placed + * @param destCapacity The size, in UChars, of the destination buffer. May be zero + * for precomputing the required size. + * @param status receives any error status. + * U_BUFFER_OVERFLOW_ERROR: the extracted text was truncated because the + * buffer was too small. Returns number of UChars for preflighting. + * @return Number of UChars in the data to be extracted. Does not include a trailing NUL. + * + * @stable ICU 3.4 + */ +U_CAPI int32_t U_EXPORT2 +utext_extract(UText *ut, + int64_t nativeStart, int64_t nativeLimit, + UChar *dest, int32_t destCapacity, + UErrorCode *status); + + + +/************************************************************************************ + * + * #define inline versions of selected performance-critical text access functions + * Caution: do not use auto increment++ or decrement-- expressions + * as parameters to these macros. + * + * For most use, where there is no extreme performance constraint, the + * normal, non-inline functions are a better choice. The resulting code + * will be smaller, and, if the need ever arises, easier to debug. + * + * These are implemented as #defines rather than real functions + * because there is no fully portable way to do inline functions in plain C. + * + ************************************************************************************/ + +#ifndef U_HIDE_INTERNAL_API +/** + * inline version of utext_current32(), for performance-critical situations. + * + * Get the code point at the current iteration position of the UText. + * Returns U_SENTINEL (-1) if the position is at the end of the + * text. + * + * @internal ICU 4.4 technology preview + */ +#define UTEXT_CURRENT32(ut) \ + ((ut)->chunkOffset < (ut)->chunkLength && ((ut)->chunkContents)[(ut)->chunkOffset]<0xd800 ? \ + ((ut)->chunkContents)[((ut)->chunkOffset)] : utext_current32(ut)) +#endif /* U_HIDE_INTERNAL_API */ + +/** + * inline version of utext_next32(), for performance-critical situations. + * + * Get the code point at the current iteration position of the UText, and + * advance the position to the first index following the character. + * This is a post-increment operation. + * Returns U_SENTINEL (-1) if the position is at the end of the + * text. + * + * @stable ICU 3.4 + */ +#define UTEXT_NEXT32(ut) \ + ((ut)->chunkOffset < (ut)->chunkLength && ((ut)->chunkContents)[(ut)->chunkOffset]<0xd800 ? \ + ((ut)->chunkContents)[((ut)->chunkOffset)++] : utext_next32(ut)) + +/** + * inline version of utext_previous32(), for performance-critical situations. + * + * Move the iterator position to the character (code point) whose + * index precedes the current position, and return that character. + * This is a pre-decrement operation. + * Returns U_SENTINEL (-1) if the position is at the start of the text. + * + * @stable ICU 3.4 + */ +#define UTEXT_PREVIOUS32(ut) \ + ((ut)->chunkOffset > 0 && \ + (ut)->chunkContents[(ut)->chunkOffset-1] < 0xd800 ? \ + (ut)->chunkContents[--((ut)->chunkOffset)] : utext_previous32(ut)) + +/** + * inline version of utext_getNativeIndex(), for performance-critical situations. + * + * Get the current iterator position, which can range from 0 to + * the length of the text. + * The position is a native index into the input text, in whatever format it + * may have (possibly UTF-8 for example), and may not always be the same as + * the corresponding UChar (UTF-16) index. + * The returned position will always be aligned to a code point boundary. + * + * @stable ICU 3.6 + */ +#define UTEXT_GETNATIVEINDEX(ut) \ + ((ut)->chunkOffset <= (ut)->nativeIndexingLimit? \ + (ut)->chunkNativeStart+(ut)->chunkOffset : \ + (ut)->pFuncs->mapOffsetToNative(ut)) + +/** + * inline version of utext_setNativeIndex(), for performance-critical situations. + * + * Set the current iteration position to the nearest code point + * boundary at or preceding the specified index. + * The index is in the native units of the original input text. + * If the index is out of range, it will be pinned to be within + * the range of the input text. + * + * @stable ICU 3.8 + */ +#define UTEXT_SETNATIVEINDEX(ut, ix) UPRV_BLOCK_MACRO_BEGIN { \ + int64_t __offset = (ix) - (ut)->chunkNativeStart; \ + if (__offset>=0 && __offset<(int64_t)(ut)->nativeIndexingLimit && (ut)->chunkContents[__offset]<0xdc00) { \ + (ut)->chunkOffset=(int32_t)__offset; \ + } else { \ + utext_setNativeIndex((ut), (ix)); \ + } \ +} UPRV_BLOCK_MACRO_END + + + +/************************************************************************************ + * + * Functions related to writing or modifying the text. + * These will work only with modifiable UTexts. Attempting to + * modify a read-only UText will return an error status. + * + ************************************************************************************/ + + +/** + * Return true if the text can be written (modified) with utext_replace() or + * utext_copy(). For the text to be writable, the text provider must + * be of a type that supports writing and the UText must not be frozen. + * + * Attempting to modify text when utext_isWriteable() is false will fail - + * the text will not be modified, and an error will be returned from the function + * that attempted the modification. + * + * @param ut the UText to be tested. + * @return true if the text is modifiable. + * + * @see utext_freeze() + * @see utext_replace() + * @see utext_copy() + * @stable ICU 3.4 + * + */ +U_CAPI UBool U_EXPORT2 +utext_isWritable(const UText *ut); + + +/** + * Test whether there is meta data associated with the text. + * @see Replaceable::hasMetaData() + * + * @param ut The UText to be tested + * @return true if the underlying text includes meta data. + * @stable ICU 3.4 + */ +U_CAPI UBool U_EXPORT2 +utext_hasMetaData(const UText *ut); + + +/** + * Replace a range of the original text with a replacement text. + * + * Leaves the current iteration position at the position following the + * newly inserted replacement text. + * + * This function is only available on UText types that support writing, + * that is, ones where utext_isWritable() returns true. + * + * When using this function, there should be only a single UText opened onto the + * underlying native text string. Behavior after a replace operation + * on a UText is undefined for any other additional UTexts that refer to the + * modified string. + * + * @param ut the UText representing the text to be operated on. + * @param nativeStart the native index of the start of the region to be replaced + * @param nativeLimit the native index of the character following the region to be replaced. + * @param replacementText pointer to the replacement text + * @param replacementLength length of the replacement text, or -1 if the text is NUL terminated. + * @param status receives any error status. Possible errors include + * U_NO_WRITE_PERMISSION + * + * @return The signed number of (native) storage units by which + * the length of the text expanded or contracted. + * + * @stable ICU 3.4 + */ +U_CAPI int32_t U_EXPORT2 +utext_replace(UText *ut, + int64_t nativeStart, int64_t nativeLimit, + const UChar *replacementText, int32_t replacementLength, + UErrorCode *status); + + + +/** + * + * Copy or move a substring from one position to another within the text, + * while retaining any metadata associated with the text. + * This function is used to duplicate or reorder substrings. + * The destination index must not overlap the source range. + * + * The text to be copied or moved is inserted at destIndex; + * it does not replace or overwrite any existing text. + * + * The iteration position is left following the newly inserted text + * at the destination position. + * + * This function is only available on UText types that support writing, + * that is, ones where utext_isWritable() returns true. + * + * When using this function, there should be only a single UText opened onto the + * underlying native text string. Behavior after a copy operation + * on a UText is undefined in any other additional UTexts that refer to the + * modified string. + * + * @param ut The UText representing the text to be operated on. + * @param nativeStart The native index of the start of the region to be copied or moved + * @param nativeLimit The native index of the character position following the region + * to be copied. + * @param destIndex The native destination index to which the source substring is + * copied or moved. + * @param move If true, then the substring is moved, not copied/duplicated. + * @param status receives any error status. Possible errors include U_NO_WRITE_PERMISSION + * + * @stable ICU 3.4 + */ +U_CAPI void U_EXPORT2 +utext_copy(UText *ut, + int64_t nativeStart, int64_t nativeLimit, + int64_t destIndex, + UBool move, + UErrorCode *status); + + +/** + *

+ * Freeze a UText. This prevents any modification to the underlying text itself + * by means of functions operating on this UText. + *

+ *

+ * Once frozen, a UText can not be unfrozen. The intent is to ensure + * that a the text underlying a frozen UText wrapper cannot be modified via that UText. + *

+ *

+ * Caution: freezing a UText will disable changes made via the specific + * frozen UText wrapper only; it will not have any effect on the ability to + * directly modify the text by bypassing the UText. Any such backdoor modifications + * are always an error while UText access is occurring because the underlying + * text can get out of sync with UText's buffering. + *

+ * + * @param ut The UText to be frozen. + * @see utext_isWritable() + * @stable ICU 3.6 + */ +U_CAPI void U_EXPORT2 +utext_freeze(UText *ut); + + +/** + * UText provider properties (bit field indexes). + * + * @see UText + * @stable ICU 3.4 + */ +enum { + /** + * It is potentially time consuming for the provider to determine the length of the text. + * @stable ICU 3.4 + */ + UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE = 1, + /** + * Text chunks remain valid and usable until the text object is modified or + * deleted, not just until the next time the access() function is called + * (which is the default). + * @stable ICU 3.4 + */ + UTEXT_PROVIDER_STABLE_CHUNKS = 2, + /** + * The provider supports modifying the text via the replace() and copy() + * functions. + * @see Replaceable + * @stable ICU 3.4 + */ + UTEXT_PROVIDER_WRITABLE = 3, + /** + * There is meta data associated with the text. + * @see Replaceable::hasMetaData() + * @stable ICU 3.4 + */ + UTEXT_PROVIDER_HAS_META_DATA = 4, + /** + * Text provider owns the text storage. + * Generally occurs as the result of a deep clone of the UText. + * When closing the UText, the associated text must + * also be closed/deleted/freed/ whatever is appropriate. + * @stable ICU 3.6 + */ + UTEXT_PROVIDER_OWNS_TEXT = 5 +}; + +/** + * Function type declaration for UText.clone(). + * + * clone a UText. Much like opening a UText where the source text is itself + * another UText. + * + * A deep clone will copy both the UText data structures and the underlying text. + * The original and cloned UText will operate completely independently; modifications + * made to the text in one will not effect the other. Text providers are not + * required to support deep clones. The user of clone() must check the status return + * and be prepared to handle failures. + * + * A shallow clone replicates only the UText data structures; it does not make + * a copy of the underlying text. Shallow clones can be used as an efficient way to + * have multiple iterators active in a single text string that is not being + * modified. + * + * A shallow clone operation must not fail except for truly exceptional conditions such + * as memory allocation failures. + * + * A UText and its clone may be safely concurrently accessed by separate threads. + * This is true for both shallow and deep clones. + * It is the responsibility of the Text Provider to ensure that this thread safety + * constraint is met. + + * + * @param dest A UText struct to be filled in with the result of the clone operation, + * or NULL if the clone function should heap-allocate a new UText struct. + * @param src The UText to be cloned. + * @param deep true to request a deep clone, false for a shallow clone. + * @param status Errors are returned here. For deep clones, U_UNSUPPORTED_ERROR + * should be returned if the text provider is unable to clone the + * original text. + * @return The newly created clone, or NULL if the clone operation failed. + * + * @stable ICU 3.4 + */ +typedef UText * U_CALLCONV +UTextClone(UText *dest, const UText *src, UBool deep, UErrorCode *status); + + +/** + * Function type declaration for UText.nativeLength(). + * + * @param ut the UText to get the length of. + * @return the length, in the native units of the original text string. + * @see UText + * @stable ICU 3.4 + */ +typedef int64_t U_CALLCONV +UTextNativeLength(UText *ut); + +/** + * Function type declaration for UText.access(). Get the description of the text chunk + * containing the text at a requested native index. The UText's iteration + * position will be left at the requested index. If the index is out + * of bounds, the iteration position will be left at the start or end + * of the string, as appropriate. + * + * Chunks must begin and end on code point boundaries. A single code point + * comprised of multiple storage units must never span a chunk boundary. + * + * + * @param ut the UText being accessed. + * @param nativeIndex Requested index of the text to be accessed. + * @param forward If true, then the returned chunk must contain text + * starting from the index, so that start<=index + * The size (number of 16 bit UChars) in the data to be extracted is returned. The + * full amount is returned, even when the specified buffer size is smaller. + *

+ * The extracted string will (if you are a user) / must (if you are a text provider) + * be NUL-terminated if there is sufficient space in the destination buffer. + * + * @param ut the UText from which to extract data. + * @param nativeStart the native index of the first character to extract. + * @param nativeLimit the native string index of the position following the last + * character to extract. + * @param dest the UChar (UTF-16) buffer into which the extracted text is placed + * @param destCapacity The size, in UChars, of the destination buffer. May be zero + * for precomputing the required size. + * @param status receives any error status. + * If U_BUFFER_OVERFLOW_ERROR: Returns number of UChars for + * preflighting. + * @return Number of UChars in the data. Does not include a trailing NUL. + * + * @stable ICU 3.4 + */ +typedef int32_t U_CALLCONV +UTextExtract(UText *ut, + int64_t nativeStart, int64_t nativeLimit, + UChar *dest, int32_t destCapacity, + UErrorCode *status); + +/** + * Function type declaration for UText.replace(). + * + * Replace a range of the original text with a replacement text. + * + * Leaves the current iteration position at the position following the + * newly inserted replacement text. + * + * This function need only be implemented on UText types that support writing. + * + * When using this function, there should be only a single UText opened onto the + * underlying native text string. The function is responsible for updating the + * text chunk within the UText to reflect the updated iteration position, + * taking into account any changes to the underlying string's structure caused + * by the replace operation. + * + * @param ut the UText representing the text to be operated on. + * @param nativeStart the index of the start of the region to be replaced + * @param nativeLimit the index of the character following the region to be replaced. + * @param replacementText pointer to the replacement text + * @param replacmentLength length of the replacement text in UChars, or -1 if the text is NUL terminated. + * @param status receives any error status. Possible errors include + * U_NO_WRITE_PERMISSION + * + * @return The signed number of (native) storage units by which + * the length of the text expanded or contracted. + * + * @stable ICU 3.4 + */ +typedef int32_t U_CALLCONV +UTextReplace(UText *ut, + int64_t nativeStart, int64_t nativeLimit, + const UChar *replacementText, int32_t replacmentLength, + UErrorCode *status); + +/** + * Function type declaration for UText.copy(). + * + * Copy or move a substring from one position to another within the text, + * while retaining any metadata associated with the text. + * This function is used to duplicate or reorder substrings. + * The destination index must not overlap the source range. + * + * The text to be copied or moved is inserted at destIndex; + * it does not replace or overwrite any existing text. + * + * This function need only be implemented for UText types that support writing. + * + * When using this function, there should be only a single UText opened onto the + * underlying native text string. The function is responsible for updating the + * text chunk within the UText to reflect the updated iteration position, + * taking into account any changes to the underlying string's structure caused + * by the replace operation. + * + * @param ut The UText representing the text to be operated on. + * @param nativeStart The index of the start of the region to be copied or moved + * @param nativeLimit The index of the character following the region to be replaced. + * @param nativeDest The destination index to which the source substring is copied or moved. + * @param move If true, then the substring is moved, not copied/duplicated. + * @param status receives any error status. Possible errors include U_NO_WRITE_PERMISSION + * + * @stable ICU 3.4 + */ +typedef void U_CALLCONV +UTextCopy(UText *ut, + int64_t nativeStart, int64_t nativeLimit, + int64_t nativeDest, + UBool move, + UErrorCode *status); + +/** + * Function type declaration for UText.mapOffsetToNative(). + * Map from the current UChar offset within the current text chunk to + * the corresponding native index in the original source text. + * + * This is required only for text providers that do not use native UTF-16 indexes. + * + * @param ut the UText. + * @return Absolute (native) index corresponding to chunkOffset in the current chunk. + * The returned native index should always be to a code point boundary. + * + * @stable ICU 3.4 + */ +typedef int64_t U_CALLCONV +UTextMapOffsetToNative(const UText *ut); + +/** + * Function type declaration for UText.mapIndexToUTF16(). + * Map from a native index to a UChar offset within a text chunk. + * Behavior is undefined if the native index does not fall within the + * current chunk. + * + * This function is required only for text providers that do not use native UTF-16 indexes. + * + * @param ut The UText containing the text chunk. + * @param nativeIndex Absolute (native) text index, chunk->start<=index<=chunk->limit. + * @return Chunk-relative UTF-16 offset corresponding to the specified native + * index. + * + * @stable ICU 3.4 + */ +typedef int32_t U_CALLCONV +UTextMapNativeIndexToUTF16(const UText *ut, int64_t nativeIndex); + + +/** + * Function type declaration for UText.utextClose(). + * + * A Text Provider close function is only required for provider types that make + * allocations in their open function (or other functions) that must be + * cleaned when the UText is closed. + * + * The allocation of the UText struct itself and any "extra" storage + * associated with the UText is handled by the common UText implementation + * and does not require provider specific cleanup in a close function. + * + * Most UText provider implementations do not need to implement this function. + * + * @param ut A UText object to be closed. + * + * @stable ICU 3.4 + */ +typedef void U_CALLCONV +UTextClose(UText *ut); + + +/** + * (public) Function dispatch table for UText. + * Conceptually very much like a C++ Virtual Function Table. + * This struct defines the organization of the table. + * Each text provider implementation must provide an + * actual table that is initialized with the appropriate functions + * for the type of text being handled. + * @stable ICU 3.6 + */ +struct UTextFuncs { + /** + * (public) Function table size, sizeof(UTextFuncs) + * Intended for use should the table grow to accommodate added + * functions in the future, to allow tests for older format + * function tables that do not contain the extensions. + * + * Fields are placed for optimal alignment on + * 32/64/128-bit-pointer machines, by normally grouping together + * 4 32-bit fields, + * 4 pointers, + * 2 64-bit fields + * in sequence. + * @stable ICU 3.6 + */ + int32_t tableSize; + + /** + * (private) Alignment padding. + * Do not use, reserved for use by the UText framework only. + * @internal + */ + int32_t reserved1, /** @internal */ reserved2, /** @internal */ reserved3; + + + /** + * (public) Function pointer for UTextClone + * + * @see UTextClone + * @stable ICU 3.6 + */ + UTextClone *clone; + + /** + * (public) function pointer for UTextLength + * May be expensive to compute! + * + * @see UTextLength + * @stable ICU 3.6 + */ + UTextNativeLength *nativeLength; + + /** + * (public) Function pointer for UTextAccess. + * + * @see UTextAccess + * @stable ICU 3.6 + */ + UTextAccess *access; + + /** + * (public) Function pointer for UTextExtract. + * + * @see UTextExtract + * @stable ICU 3.6 + */ + UTextExtract *extract; + + /** + * (public) Function pointer for UTextReplace. + * + * @see UTextReplace + * @stable ICU 3.6 + */ + UTextReplace *replace; + + /** + * (public) Function pointer for UTextCopy. + * + * @see UTextCopy + * @stable ICU 3.6 + */ + UTextCopy *copy; + + /** + * (public) Function pointer for UTextMapOffsetToNative. + * + * @see UTextMapOffsetToNative + * @stable ICU 3.6 + */ + UTextMapOffsetToNative *mapOffsetToNative; + + /** + * (public) Function pointer for UTextMapNativeIndexToUTF16. + * + * @see UTextMapNativeIndexToUTF16 + * @stable ICU 3.6 + */ + UTextMapNativeIndexToUTF16 *mapNativeIndexToUTF16; + + /** + * (public) Function pointer for UTextClose. + * + * @see UTextClose + * @stable ICU 3.6 + */ + UTextClose *close; + + /** + * (private) Spare function pointer + * @internal + */ + UTextClose *spare1; + + /** + * (private) Spare function pointer + * @internal + */ + UTextClose *spare2; + + /** + * (private) Spare function pointer + * @internal + */ + UTextClose *spare3; + +}; +/** + * Function dispatch table for UText + * @see UTextFuncs + */ +typedef struct UTextFuncs UTextFuncs; + + /** + * UText struct. Provides the interface between the generic UText access code + * and the UText provider code that works on specific kinds of + * text (UTF-8, noncontiguous UTF-16, whatever.) + * + * Applications that are using predefined types of text providers + * to pass text data to ICU services will have no need to view the + * internals of the UText structs that they open. + * + * @stable ICU 3.6 + */ +struct UText { + /** + * (private) Magic. Used to help detect when UText functions are handed + * invalid or uninitialized UText structs. + * utext_openXYZ() functions take an initialized, + * but not necessarily open, UText struct as an + * optional fill-in parameter. This magic field + * is used to check for that initialization. + * Text provider close functions must NOT clear + * the magic field because that would prevent + * reuse of the UText struct. + * @internal + */ + uint32_t magic; + + + /** + * (private) Flags for managing the allocation and freeing of + * memory associated with this UText. + * @internal + */ + int32_t flags; + + + /** + * Text provider properties. This set of flags is maintained by the + * text provider implementation. + * @stable ICU 3.4 + */ + int32_t providerProperties; + + /** + * (public) sizeOfStruct=sizeof(UText) + * Allows possible backward compatible extension. + * + * @stable ICU 3.4 + */ + int32_t sizeOfStruct; + + /* ------ 16 byte alignment boundary ----------- */ + + + /** + * (protected) Native index of the first character position following + * the current chunk. + * @stable ICU 3.6 + */ + int64_t chunkNativeLimit; + + /** + * (protected) Size in bytes of the extra space (pExtra). + * @stable ICU 3.4 + */ + int32_t extraSize; + + /** + * (protected) The highest chunk offset where native indexing and + * chunk (UTF-16) indexing correspond. For UTF-16 sources, value + * will be equal to chunkLength. + * + * @stable ICU 3.6 + */ + int32_t nativeIndexingLimit; + + /* ---- 16 byte alignment boundary------ */ + + /** + * (protected) Native index of the first character in the text chunk. + * @stable ICU 3.6 + */ + int64_t chunkNativeStart; + + /** + * (protected) Current iteration position within the text chunk (UTF-16 buffer). + * This is the index to the character that will be returned by utext_next32(). + * @stable ICU 3.6 + */ + int32_t chunkOffset; + + /** + * (protected) Length the text chunk (UTF-16 buffer), in UChars. + * @stable ICU 3.6 + */ + int32_t chunkLength; + + /* ---- 16 byte alignment boundary-- */ + + + /** + * (protected) pointer to a chunk of text in UTF-16 format. + * May refer either to original storage of the source of the text, or + * if conversion was required, to a buffer owned by the UText. + * @stable ICU 3.6 + */ + const UChar *chunkContents; + + /** + * (public) Pointer to Dispatch table for accessing functions for this UText. + * @stable ICU 3.6 + */ + const UTextFuncs *pFuncs; + + /** + * (protected) Pointer to additional space requested by the + * text provider during the utext_open operation. + * @stable ICU 3.4 + */ + void *pExtra; + + /** + * (protected) Pointer to string or text-containing object or similar. + * This is the source of the text that this UText is wrapping, in a format + * that is known to the text provider functions. + * @stable ICU 3.4 + */ + const void *context; + + /* --- 16 byte alignment boundary--- */ + + /** + * (protected) Pointer fields available for use by the text provider. + * Not used by UText common code. + * @stable ICU 3.6 + */ + const void *p; + /** + * (protected) Pointer fields available for use by the text provider. + * Not used by UText common code. + * @stable ICU 3.6 + */ + const void *q; + /** + * (protected) Pointer fields available for use by the text provider. + * Not used by UText common code. + * @stable ICU 3.6 + */ + const void *r; + + /** + * Private field reserved for future use by the UText framework + * itself. This is not to be touched by the text providers. + * @internal ICU 3.4 + */ + void *privP; + + + /* --- 16 byte alignment boundary--- */ + + + /** + * (protected) Integer field reserved for use by the text provider. + * Not used by the UText framework, or by the client (user) of the UText. + * @stable ICU 3.4 + */ + int64_t a; + + /** + * (protected) Integer field reserved for use by the text provider. + * Not used by the UText framework, or by the client (user) of the UText. + * @stable ICU 3.4 + */ + int32_t b; + + /** + * (protected) Integer field reserved for use by the text provider. + * Not used by the UText framework, or by the client (user) of the UText. + * @stable ICU 3.4 + */ + int32_t c; + + /* ---- 16 byte alignment boundary---- */ + + + /** + * Private field reserved for future use by the UText framework + * itself. This is not to be touched by the text providers. + * @internal ICU 3.4 + */ + int64_t privA; + /** + * Private field reserved for future use by the UText framework + * itself. This is not to be touched by the text providers. + * @internal ICU 3.4 + */ + int32_t privB; + /** + * Private field reserved for future use by the UText framework + * itself. This is not to be touched by the text providers. + * @internal ICU 3.4 + */ + int32_t privC; +}; + + +/** + * Common function for use by Text Provider implementations to allocate and/or initialize + * a new UText struct. To be called in the implementation of utext_open() functions. + * If the supplied UText parameter is null, a new UText struct will be allocated on the heap. + * If the supplied UText is already open, the provider's close function will be called + * so that the struct can be reused by the open that is in progress. + * + * @param ut pointer to a UText struct to be re-used, or null if a new UText + * should be allocated. + * @param extraSpace The amount of additional space to be allocated as part + * of this UText, for use by types of providers that require + * additional storage. + * @param status Errors are returned here. + * @return pointer to the UText, allocated if necessary, with extra space set up if requested. + * @stable ICU 3.4 + */ +U_CAPI UText * U_EXPORT2 +utext_setup(UText *ut, int32_t extraSpace, UErrorCode *status); + +// do not use #ifndef U_HIDE_INTERNAL_API around the following! +/** + * @internal + * Value used to help identify correctly initialized UText structs. + * Note: must be publicly visible so that UTEXT_INITIALIZER can access it. + */ +enum { + UTEXT_MAGIC = 0x345ad82c +}; + +/** + * initializer to be used with local (stack) instances of a UText + * struct. UText structs must be initialized before passing + * them to one of the utext_open functions. + * + * @stable ICU 3.6 + */ +#define UTEXT_INITIALIZER { \ + UTEXT_MAGIC, /* magic */ \ + 0, /* flags */ \ + 0, /* providerProps */ \ + sizeof(UText), /* sizeOfStruct */ \ + 0, /* chunkNativeLimit */ \ + 0, /* extraSize */ \ + 0, /* nativeIndexingLimit */ \ + 0, /* chunkNativeStart */ \ + 0, /* chunkOffset */ \ + 0, /* chunkLength */ \ + NULL, /* chunkContents */ \ + NULL, /* pFuncs */ \ + NULL, /* pExtra */ \ + NULL, /* context */ \ + NULL, NULL, NULL, /* p, q, r */ \ + NULL, /* privP */ \ + 0, 0, 0, /* a, b, c */ \ + 0, 0, 0 /* privA,B,C, */ \ + } + + +U_CDECL_END + + +#if U_SHOW_CPLUSPLUS_API + +U_NAMESPACE_BEGIN + +/** + * \class LocalUTextPointer + * "Smart pointer" class, closes a UText via utext_close(). + * For most methods see the LocalPointerBase base class. + * + * @see LocalPointerBase + * @see LocalPointer + * @stable ICU 4.4 + */ +U_DEFINE_LOCAL_OPEN_POINTER(LocalUTextPointer, UText, utext_close); + +U_NAMESPACE_END + +#endif + + +#endif diff --git a/intl/icu/source/common/unicode/utf.h b/intl/icu/source/common/unicode/utf.h new file mode 100644 index 0000000000..c9d5f5785c --- /dev/null +++ b/intl/icu/source/common/unicode/utf.h @@ -0,0 +1,225 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html +/* +******************************************************************************* +* +* Copyright (C) 1999-2011, International Business Machines +* Corporation and others. All Rights Reserved. +* +******************************************************************************* +* file name: utf.h +* encoding: UTF-8 +* tab size: 8 (not used) +* indentation:4 +* +* created on: 1999sep09 +* created by: Markus W. Scherer +*/ + +/** + * \file + * \brief C API: Code point macros + * + * This file defines macros for checking whether a code point is + * a surrogate or a non-character etc. + * + * If U_NO_DEFAULT_INCLUDE_UTF_HEADERS is 0 then utf.h is included by utypes.h + * and itself includes utf8.h and utf16.h after some + * common definitions. + * If U_NO_DEFAULT_INCLUDE_UTF_HEADERS is 1 then each of these headers must be + * included explicitly if their definitions are used. + * + * utf8.h and utf16.h define macros for efficiently getting code points + * in and out of UTF-8/16 strings. + * utf16.h macros have "U16_" prefixes. + * utf8.h defines similar macros with "U8_" prefixes for UTF-8 string handling. + * + * ICU mostly processes 16-bit Unicode strings. + * Most of the time, such strings are well-formed UTF-16. + * Single, unpaired surrogates must be handled as well, and are treated in ICU + * like regular code points where possible. + * (Pairs of surrogate code points are indistinguishable from supplementary + * code points encoded as pairs of supplementary code units.) + * + * In fact, almost all Unicode code points in normal text (>99%) + * are on the BMP (<=U+ffff) and even <=U+d7ff. + * ICU functions handle supplementary code points (U+10000..U+10ffff) + * but are optimized for the much more frequently occurring BMP code points. + * + * umachine.h defines UChar to be an unsigned 16-bit integer. + * Since ICU 59, ICU uses char16_t in C++, UChar only in C, + * and defines UChar=char16_t by default. See the UChar API docs for details. + * + * UChar32 is defined to be a signed 32-bit integer (int32_t), large enough for a 21-bit + * Unicode code point (Unicode scalar value, 0..0x10ffff) and U_SENTINEL (-1). + * Before ICU 2.4, the definition of UChar32 was similarly platform-dependent as + * the definition of UChar. For details see the documentation for UChar32 itself. + * + * utf.h defines a small number of C macros for single Unicode code points. + * These are simple checks for surrogates and non-characters. + * For actual Unicode character properties see uchar.h. + * + * By default, string operations must be done with error checking in case + * a string is not well-formed UTF-16 or UTF-8. + * + * The U16_ macros detect if a surrogate code unit is unpaired + * (lead unit without trail unit or vice versa) and just return the unit itself + * as the code point. + * + * The U8_ macros detect illegal byte sequences and return a negative value. + * Starting with ICU 60, the observable length of a single illegal byte sequence + * skipped by one of these macros follows the Unicode 6+ recommendation + * which is consistent with the W3C Encoding Standard. + * + * There are ..._OR_FFFD versions of both U16_ and U8_ macros + * that return U+FFFD for illegal code unit sequences. + * + * The regular "safe" macros require that the initial, passed-in string index + * is within bounds. They only check the index when they read more than one + * code unit. This is usually done with code similar to the following loop: + * 

while(i
+ *
+ * When it is safe to assume that text is well-formed UTF-16
+ * (does not contain single, unpaired surrogates), then one can use
+ * U16_..._UNSAFE macros.
+ * These do not check for proper code unit sequences or truncated text and may
+ * yield wrong results or even cause a crash if they are used with "malformed"
+ * text.
+ * In practice, U16_..._UNSAFE macros will produce slightly less code but
+ * should not be faster because the processing is only different when a
+ * surrogate code unit is detected, which will be rare.
+ *
+ * Similarly for UTF-8, there are "safe" macros without a suffix,
+ * and U8_..._UNSAFE versions.
+ * The performance differences are much larger here because UTF-8 provides so
+ * many opportunities for malformed sequences.
+ * The unsafe UTF-8 macros are entirely implemented inside the macro definitions
+ * and are fast, while the safe UTF-8 macros call functions for some complicated cases.
+ *
+ * Unlike with UTF-16, malformed sequences cannot be expressed with distinct
+ * code point values (0..U+10ffff). They are indicated with negative values instead.
+ *
+ * For more information see the ICU User Guide Strings chapter
+ * (https://unicode-org.github.io/icu/userguide/strings).
+ *
+ * Usage:
+ * ICU coding guidelines for if() statements should be followed when using these macros.
+ * Compound statements (curly braces {}) must be used  for if-else-while... 
+ * bodies and all macro statements should be terminated with semicolon.
+ *
+ * @stable ICU 2.4
+ */
+
+#ifndef __UTF_H__
+#define __UTF_H__
+
+#include "unicode/umachine.h"
+/* include the utfXX.h after the following definitions */
+
+/* single-code point definitions -------------------------------------------- */
+
+/**
+ * Is this code point a Unicode noncharacter?
+ * @param c 32-bit code point
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U_IS_UNICODE_NONCHAR(c) \
+    ((c)>=0xfdd0 && \
+     ((c)<=0xfdef || ((c)&0xfffe)==0xfffe) && (c)<=0x10ffff)
+
+/**
+ * Is c a Unicode code point value (0..U+10ffff)
+ * that can be assigned a character?
+ *
+ * Code points that are not characters include:
+ * - single surrogate code points (U+d800..U+dfff, 2048 code points)
+ * - the last two code points on each plane (U+__fffe and U+__ffff, 34 code points)
+ * - U+fdd0..U+fdef (new with Unicode 3.1, 32 code points)
+ * - the highest Unicode code point value is U+10ffff
+ *
+ * This means that all code points below U+d800 are character code points,
+ * and that boundary is tested first for performance.
+ *
+ * @param c 32-bit code point
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U_IS_UNICODE_CHAR(c) \
+    ((uint32_t)(c)<0xd800 || \
+        (0xdfff<(c) && (c)<=0x10ffff && !U_IS_UNICODE_NONCHAR(c)))
+
+/**
+ * Is this code point a BMP code point (U+0000..U+ffff)?
+ * @param c 32-bit code point
+ * @return true or false
+ * @stable ICU 2.8
+ */
+#define U_IS_BMP(c) ((uint32_t)(c)<=0xffff)
+
+/**
+ * Is this code point a supplementary code point (U+10000..U+10ffff)?
+ * @param c 32-bit code point
+ * @return true or false
+ * @stable ICU 2.8
+ */
+#define U_IS_SUPPLEMENTARY(c) ((uint32_t)((c)-0x10000)<=0xfffff)
+ 
+/**
+ * Is this code point a lead surrogate (U+d800..U+dbff)?
+ * @param c 32-bit code point
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U_IS_LEAD(c) (((c)&0xfffffc00)==0xd800)
+
+/**
+ * Is this code point a trail surrogate (U+dc00..U+dfff)?
+ * @param c 32-bit code point
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U_IS_TRAIL(c) (((c)&0xfffffc00)==0xdc00)
+
+/**
+ * Is this code point a surrogate (U+d800..U+dfff)?
+ * @param c 32-bit code point
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U_IS_SURROGATE(c) (((c)&0xfffff800)==0xd800)
+
+/**
+ * Assuming c is a surrogate code point (U_IS_SURROGATE(c)),
+ * is it a lead surrogate?
+ * @param c 32-bit code point
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U_IS_SURROGATE_LEAD(c) (((c)&0x400)==0)
+
+/**
+ * Assuming c is a surrogate code point (U_IS_SURROGATE(c)),
+ * is it a trail surrogate?
+ * @param c 32-bit code point
+ * @return true or false
+ * @stable ICU 4.2
+ */
+#define U_IS_SURROGATE_TRAIL(c) (((c)&0x400)!=0)
+
+/* include the utfXX.h ------------------------------------------------------ */
+
+#if !U_NO_DEFAULT_INCLUDE_UTF_HEADERS
+
+#include "unicode/utf8.h"
+#include "unicode/utf16.h"
+
+/* utf_old.h contains deprecated, pre-ICU 2.4 definitions */
+#include "unicode/utf_old.h"
+
+#endif  /* !U_NO_DEFAULT_INCLUDE_UTF_HEADERS */
+
+#endif  /* __UTF_H__ */
diff --git a/intl/icu/source/common/unicode/utf16.h b/intl/icu/source/common/unicode/utf16.h
new file mode 100644
index 0000000000..3902c60e95
--- /dev/null
+++ b/intl/icu/source/common/unicode/utf16.h
@@ -0,0 +1,734 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 1999-2012, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  utf16.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 1999sep09
+*   created by: Markus W. Scherer
+*/
+
+/**
+ * \file
+ * \brief C API: 16-bit Unicode handling macros
+ * 
+ * This file defines macros to deal with 16-bit Unicode (UTF-16) code units and strings.
+ *
+ * For more information see utf.h and the ICU User Guide Strings chapter
+ * (https://unicode-org.github.io/icu/userguide/strings).
+ *
+ * Usage:
+ * ICU coding guidelines for if() statements should be followed when using these macros.
+ * Compound statements (curly braces {}) must be used  for if-else-while... 
+ * bodies and all macro statements should be terminated with semicolon.
+ */
+
+#ifndef __UTF16_H__
+#define __UTF16_H__
+
+#include 
+#include "unicode/umachine.h"
+#ifndef __UTF_H__
+#   include "unicode/utf.h"
+#endif
+
+/* single-code point definitions -------------------------------------------- */
+
+/**
+ * Does this code unit alone encode a code point (BMP, not a surrogate)?
+ * @param c 16-bit code unit
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U16_IS_SINGLE(c) !U_IS_SURROGATE(c)
+
+/**
+ * Is this code unit a lead surrogate (U+d800..U+dbff)?
+ * @param c 16-bit code unit
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U16_IS_LEAD(c) (((c)&0xfffffc00)==0xd800)
+
+/**
+ * Is this code unit a trail surrogate (U+dc00..U+dfff)?
+ * @param c 16-bit code unit
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U16_IS_TRAIL(c) (((c)&0xfffffc00)==0xdc00)
+
+/**
+ * Is this code unit a surrogate (U+d800..U+dfff)?
+ * @param c 16-bit code unit
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U16_IS_SURROGATE(c) U_IS_SURROGATE(c)
+
+/**
+ * Assuming c is a surrogate code point (U16_IS_SURROGATE(c)),
+ * is it a lead surrogate?
+ * @param c 16-bit code unit
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U16_IS_SURROGATE_LEAD(c) (((c)&0x400)==0)
+
+/**
+ * Assuming c is a surrogate code point (U16_IS_SURROGATE(c)),
+ * is it a trail surrogate?
+ * @param c 16-bit code unit
+ * @return true or false
+ * @stable ICU 4.2
+ */
+#define U16_IS_SURROGATE_TRAIL(c) (((c)&0x400)!=0)
+
+/**
+ * Helper constant for U16_GET_SUPPLEMENTARY.
+ * @internal
+ */
+#define U16_SURROGATE_OFFSET ((0xd800<<10UL)+0xdc00-0x10000)
+
+/**
+ * Get a supplementary code point value (U+10000..U+10ffff)
+ * from its lead and trail surrogates.
+ * The result is undefined if the input values are not
+ * lead and trail surrogates.
+ *
+ * @param lead lead surrogate (U+d800..U+dbff)
+ * @param trail trail surrogate (U+dc00..U+dfff)
+ * @return supplementary code point (U+10000..U+10ffff)
+ * @stable ICU 2.4
+ */
+#define U16_GET_SUPPLEMENTARY(lead, trail) \
+    (((UChar32)(lead)<<10UL)+(UChar32)(trail)-U16_SURROGATE_OFFSET)
+
+
+/**
+ * Get the lead surrogate (0xd800..0xdbff) for a
+ * supplementary code point (0x10000..0x10ffff).
+ * @param supplementary 32-bit code point (U+10000..U+10ffff)
+ * @return lead surrogate (U+d800..U+dbff) for supplementary
+ * @stable ICU 2.4
+ */
+#define U16_LEAD(supplementary) (UChar)(((supplementary)>>10)+0xd7c0)
+
+/**
+ * Get the trail surrogate (0xdc00..0xdfff) for a
+ * supplementary code point (0x10000..0x10ffff).
+ * @param supplementary 32-bit code point (U+10000..U+10ffff)
+ * @return trail surrogate (U+dc00..U+dfff) for supplementary
+ * @stable ICU 2.4
+ */
+#define U16_TRAIL(supplementary) (UChar)(((supplementary)&0x3ff)|0xdc00)
+
+/**
+ * How many 16-bit code units are used to encode this Unicode code point? (1 or 2)
+ * The result is not defined if c is not a Unicode code point (U+0000..U+10ffff).
+ * @param c 32-bit code point
+ * @return 1 or 2
+ * @stable ICU 2.4
+ */
+#define U16_LENGTH(c) ((uint32_t)(c)<=0xffff ? 1 : 2)
+
+/**
+ * The maximum number of 16-bit code units per Unicode code point (U+0000..U+10ffff).
+ * @return 2
+ * @stable ICU 2.4
+ */
+#define U16_MAX_LENGTH 2
+
+/**
+ * Get a code point from a string at a random-access offset,
+ * without changing the offset.
+ * "Unsafe" macro, assumes well-formed UTF-16.
+ *
+ * The offset may point to either the lead or trail surrogate unit
+ * for a supplementary code point, in which case the macro will read
+ * the adjacent matching surrogate as well.
+ * The result is undefined if the offset points to a single, unpaired surrogate.
+ * Iteration through a string is more efficient with U16_NEXT_UNSAFE or U16_NEXT.
+ *
+ * @param s const UChar * string
+ * @param i string offset
+ * @param c output UChar32 variable
+ * @see U16_GET
+ * @stable ICU 2.4
+ */
+#define U16_GET_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[i]; \
+    if(U16_IS_SURROGATE(c)) { \
+        if(U16_IS_SURROGATE_LEAD(c)) { \
+            (c)=U16_GET_SUPPLEMENTARY((c), (s)[(i)+1]); \
+        } else { \
+            (c)=U16_GET_SUPPLEMENTARY((s)[(i)-1], (c)); \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Get a code point from a string at a random-access offset,
+ * without changing the offset.
+ * "Safe" macro, handles unpaired surrogates and checks for string boundaries.
+ *
+ * The offset may point to either the lead or trail surrogate unit
+ * for a supplementary code point, in which case the macro will read
+ * the adjacent matching surrogate as well.
+ *
+ * The length can be negative for a NUL-terminated string.
+ *
+ * If the offset points to a single, unpaired surrogate, then
+ * c is set to that unpaired surrogate.
+ * Iteration through a string is more efficient with U16_NEXT_UNSAFE or U16_NEXT.
+ *
+ * @param s const UChar * string
+ * @param start starting string offset (usually 0)
+ * @param i string offset, must be start<=i(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
+                (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
+            } \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Get a code point from a string at a random-access offset,
+ * without changing the offset.
+ * "Safe" macro, handles unpaired surrogates and checks for string boundaries.
+ *
+ * The offset may point to either the lead or trail surrogate unit
+ * for a supplementary code point, in which case the macro will read
+ * the adjacent matching surrogate as well.
+ *
+ * The length can be negative for a NUL-terminated string.
+ *
+ * If the offset points to a single, unpaired surrogate, then
+ * c is set to U+FFFD.
+ * Iteration through a string is more efficient with U16_NEXT_UNSAFE or U16_NEXT_OR_FFFD.
+ *
+ * @param s const UChar * string
+ * @param start starting string offset (usually 0)
+ * @param i string offset, must be start<=i(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
+                (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
+            } else { \
+                (c)=0xfffd; \
+            } \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/* definitions with forward iteration --------------------------------------- */
+
+/**
+ * Get a code point from a string at a code point boundary offset,
+ * and advance the offset to the next code point boundary.
+ * (Post-incrementing forward iteration.)
+ * "Unsafe" macro, assumes well-formed UTF-16.
+ *
+ * The offset may point to the lead surrogate unit
+ * for a supplementary code point, in which case the macro will read
+ * the following trail surrogate as well.
+ * If the offset points to a trail surrogate, then that itself
+ * will be returned as the code point.
+ * The result is undefined if the offset points to a single, unpaired lead surrogate.
+ *
+ * @param s const UChar * string
+ * @param i string offset
+ * @param c output UChar32 variable
+ * @see U16_NEXT
+ * @stable ICU 2.4
+ */
+#define U16_NEXT_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[(i)++]; \
+    if(U16_IS_LEAD(c)) { \
+        (c)=U16_GET_SUPPLEMENTARY((c), (s)[(i)++]); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Get a code point from a string at a code point boundary offset,
+ * and advance the offset to the next code point boundary.
+ * (Post-incrementing forward iteration.)
+ * "Safe" macro, handles unpaired surrogates and checks for string boundaries.
+ *
+ * The length can be negative for a NUL-terminated string.
+ *
+ * The offset may point to the lead surrogate unit
+ * for a supplementary code point, in which case the macro will read
+ * the following trail surrogate as well.
+ * If the offset points to a trail surrogate or
+ * to a single, unpaired lead surrogate, then c is set to that unpaired surrogate.
+ *
+ * @param s const UChar * string
+ * @param i string offset, must be i>10)+0xd7c0); \
+        (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Append a code point to a string, overwriting 1 or 2 code units.
+ * The offset points to the current end of the string contents
+ * and is advanced (post-increment).
+ * "Safe" macro, checks for a valid code point.
+ * If a surrogate pair is written, checks for sufficient space in the string.
+ * If the code point is not valid or a trail surrogate does not fit,
+ * then isError is set to true.
+ *
+ * @param s const UChar * string buffer
+ * @param i string offset, must be i>10)+0xd7c0); \
+        (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
+    } else /* c>0x10ffff or not enough space */ { \
+        (isError)=true; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Advance the string offset from one code point boundary to the next.
+ * (Post-incrementing iteration.)
+ * "Unsafe" macro, assumes well-formed UTF-16.
+ *
+ * @param s const UChar * string
+ * @param i string offset
+ * @see U16_FWD_1
+ * @stable ICU 2.4
+ */
+#define U16_FWD_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    if(U16_IS_LEAD((s)[(i)++])) { \
+        ++(i); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Advance the string offset from one code point boundary to the next.
+ * (Post-incrementing iteration.)
+ * "Safe" macro, handles unpaired surrogates and checks for string boundaries.
+ *
+ * The length can be negative for a NUL-terminated string.
+ *
+ * @param s const UChar * string
+ * @param i string offset, must be i0) { \
+        U16_FWD_1_UNSAFE(s, i); \
+        --__N; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Advance the string offset from one code point boundary to the n-th next one,
+ * i.e., move forward by n code points.
+ * (Post-incrementing iteration.)
+ * "Safe" macro, handles unpaired surrogates and checks for string boundaries.
+ *
+ * The length can be negative for a NUL-terminated string.
+ *
+ * @param s const UChar * string
+ * @param i int32_t string offset, must be i0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
+        U16_FWD_1(s, i, length); \
+        --__N; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Adjust a random-access offset to a code point boundary
+ * at the start of a code point.
+ * If the offset points to the trail surrogate of a surrogate pair,
+ * then the offset is decremented.
+ * Otherwise, it is not modified.
+ * "Unsafe" macro, assumes well-formed UTF-16.
+ *
+ * @param s const UChar * string
+ * @param i string offset
+ * @see U16_SET_CP_START
+ * @stable ICU 2.4
+ */
+#define U16_SET_CP_START_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    if(U16_IS_TRAIL((s)[i])) { \
+        --(i); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Adjust a random-access offset to a code point boundary
+ * at the start of a code point.
+ * If the offset points to the trail surrogate of a surrogate pair,
+ * then the offset is decremented.
+ * Otherwise, it is not modified.
+ * "Safe" macro, handles unpaired surrogates and checks for string boundaries.
+ *
+ * @param s const UChar * string
+ * @param start starting string offset (usually 0)
+ * @param i string offset, must be start<=i
+ * @see U16_SET_CP_START_UNSAFE
+ * @stable ICU 2.4
+ */
+#define U16_SET_CP_START(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
+    if(U16_IS_TRAIL((s)[i]) && (i)>(start) && U16_IS_LEAD((s)[(i)-1])) { \
+        --(i); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/* definitions with backward iteration -------------------------------------- */
+
+/**
+ * Move the string offset from one code point boundary to the previous one
+ * and get the code point between them.
+ * (Pre-decrementing backward iteration.)
+ * "Unsafe" macro, assumes well-formed UTF-16.
+ *
+ * The input offset may be the same as the string length.
+ * If the offset is behind a trail surrogate unit
+ * for a supplementary code point, then the macro will read
+ * the preceding lead surrogate as well.
+ * If the offset is behind a lead surrogate, then that itself
+ * will be returned as the code point.
+ * The result is undefined if the offset is behind a single, unpaired trail surrogate.
+ *
+ * @param s const UChar * string
+ * @param i string offset
+ * @param c output UChar32 variable
+ * @see U16_PREV
+ * @stable ICU 2.4
+ */
+#define U16_PREV_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[--(i)]; \
+    if(U16_IS_TRAIL(c)) { \
+        (c)=U16_GET_SUPPLEMENTARY((s)[--(i)], (c)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Move the string offset from one code point boundary to the previous one
+ * and get the code point between them.
+ * (Pre-decrementing backward iteration.)
+ * "Safe" macro, handles unpaired surrogates and checks for string boundaries.
+ *
+ * The input offset may be the same as the string length.
+ * If the offset is behind a trail surrogate unit
+ * for a supplementary code point, then the macro will read
+ * the preceding lead surrogate as well.
+ * If the offset is behind a lead surrogate or behind a single, unpaired
+ * trail surrogate, then c is set to that unpaired surrogate.
+ *
+ * @param s const UChar * string
+ * @param start starting string offset (usually 0)
+ * @param i string offset, must be start(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
+            --(i); \
+            (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Move the string offset from one code point boundary to the previous one
+ * and get the code point between them.
+ * (Pre-decrementing backward iteration.)
+ * "Safe" macro, handles unpaired surrogates and checks for string boundaries.
+ *
+ * The input offset may be the same as the string length.
+ * If the offset is behind a trail surrogate unit
+ * for a supplementary code point, then the macro will read
+ * the preceding lead surrogate as well.
+ * If the offset is behind a lead surrogate or behind a single, unpaired
+ * trail surrogate, then c is set to U+FFFD.
+ *
+ * @param s const UChar * string
+ * @param start starting string offset (usually 0)
+ * @param i string offset, must be start(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
+            --(i); \
+            (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
+        } else { \
+            (c)=0xfffd; \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Move the string offset from one code point boundary to the previous one.
+ * (Pre-decrementing backward iteration.)
+ * The input offset may be the same as the string length.
+ * "Unsafe" macro, assumes well-formed UTF-16.
+ *
+ * @param s const UChar * string
+ * @param i string offset
+ * @see U16_BACK_1
+ * @stable ICU 2.4
+ */
+#define U16_BACK_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    if(U16_IS_TRAIL((s)[--(i)])) { \
+        --(i); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Move the string offset from one code point boundary to the previous one.
+ * (Pre-decrementing backward iteration.)
+ * The input offset may be the same as the string length.
+ * "Safe" macro, handles unpaired surrogates and checks for string boundaries.
+ *
+ * @param s const UChar * string
+ * @param start starting string offset (usually 0)
+ * @param i string offset, must be start(start) && U16_IS_LEAD((s)[(i)-1])) { \
+        --(i); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Move the string offset from one code point boundary to the n-th one before it,
+ * i.e., move backward by n code points.
+ * (Pre-decrementing backward iteration.)
+ * The input offset may be the same as the string length.
+ * "Unsafe" macro, assumes well-formed UTF-16.
+ *
+ * @param s const UChar * string
+ * @param i string offset
+ * @param n number of code points to skip
+ * @see U16_BACK_N
+ * @stable ICU 2.4
+ */
+#define U16_BACK_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
+    int32_t __N=(n); \
+    while(__N>0) { \
+        U16_BACK_1_UNSAFE(s, i); \
+        --__N; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Move the string offset from one code point boundary to the n-th one before it,
+ * i.e., move backward by n code points.
+ * (Pre-decrementing backward iteration.)
+ * The input offset may be the same as the string length.
+ * "Safe" macro, handles unpaired surrogates and checks for string boundaries.
+ *
+ * @param s const UChar * string
+ * @param start start of string
+ * @param i string offset, must be start0 && (i)>(start)) { \
+        U16_BACK_1(s, start, i); \
+        --__N; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Adjust a random-access offset to a code point boundary after a code point.
+ * If the offset is behind the lead surrogate of a surrogate pair,
+ * then the offset is incremented.
+ * Otherwise, it is not modified.
+ * The input offset may be the same as the string length.
+ * "Unsafe" macro, assumes well-formed UTF-16.
+ *
+ * @param s const UChar * string
+ * @param i string offset
+ * @see U16_SET_CP_LIMIT
+ * @stable ICU 2.4
+ */
+#define U16_SET_CP_LIMIT_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    if(U16_IS_LEAD((s)[(i)-1])) { \
+        ++(i); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Adjust a random-access offset to a code point boundary after a code point.
+ * If the offset is behind the lead surrogate of a surrogate pair,
+ * then the offset is incremented.
+ * Otherwise, it is not modified.
+ * The input offset may be the same as the string length.
+ * "Safe" macro, handles unpaired surrogates and checks for string boundaries.
+ *
+ * The length can be negative for a NUL-terminated string.
+ *
+ * @param s const UChar * string
+ * @param start int32_t starting string offset (usually 0)
+ * @param i int32_t string offset, start<=i<=length
+ * @param length int32_t string length
+ * @see U16_SET_CP_LIMIT_UNSAFE
+ * @stable ICU 2.4
+ */
+#define U16_SET_CP_LIMIT(s, start, i, length) UPRV_BLOCK_MACRO_BEGIN { \
+    if((start)<(i) && ((i)<(length) || (length)<0) && U16_IS_LEAD((s)[(i)-1]) && U16_IS_TRAIL((s)[i])) { \
+        ++(i); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+#endif
diff --git a/intl/icu/source/common/unicode/utf32.h b/intl/icu/source/common/unicode/utf32.h
new file mode 100644
index 0000000000..8822c4dd09
--- /dev/null
+++ b/intl/icu/source/common/unicode/utf32.h
@@ -0,0 +1,25 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 1999-2001, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  utf32.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 1999sep20
+*   created by: Markus W. Scherer
+*/
+/**
+ * \file
+ * \brief C API: UTF-32 macros
+ *
+ * This file is obsolete and its contents moved to utf_old.h.
+ * See utf_old.h and Jitterbug 2150 and its discussion on the ICU mailing list
+ * in September 2002.
+ */
diff --git a/intl/icu/source/common/unicode/utf8.h b/intl/icu/source/common/unicode/utf8.h
new file mode 100644
index 0000000000..5a07435fcf
--- /dev/null
+++ b/intl/icu/source/common/unicode/utf8.h
@@ -0,0 +1,882 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 1999-2015, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  utf8.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 1999sep13
+*   created by: Markus W. Scherer
+*/
+
+/**
+ * \file
+ * \brief C API: 8-bit Unicode handling macros
+ * 
+ * This file defines macros to deal with 8-bit Unicode (UTF-8) code units (bytes) and strings.
+ *
+ * For more information see utf.h and the ICU User Guide Strings chapter
+ * (https://unicode-org.github.io/icu/userguide/strings).
+ *
+ * Usage:
+ * ICU coding guidelines for if() statements should be followed when using these macros.
+ * Compound statements (curly braces {}) must be used  for if-else-while... 
+ * bodies and all macro statements should be terminated with semicolon.
+ */
+
+#ifndef __UTF8_H__
+#define __UTF8_H__
+
+#include 
+#include "unicode/umachine.h"
+#ifndef __UTF_H__
+#   include "unicode/utf.h"
+#endif
+
+/* internal definitions ----------------------------------------------------- */
+
+/**
+ * Counts the trail bytes for a UTF-8 lead byte.
+ * Returns 0 for 0..0xc1 as well as for 0xf5..0xff.
+ * leadByte might be evaluated multiple times.
+ *
+ * This is internal since it is not meant to be called directly by external clients;
+ * however it is called by public macros in this file and thus must remain stable.
+ *
+ * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff.
+ * @internal
+ */
+#define U8_COUNT_TRAIL_BYTES(leadByte) \
+    (U8_IS_LEAD(leadByte) ? \
+        ((uint8_t)(leadByte)>=0xe0)+((uint8_t)(leadByte)>=0xf0)+1 : 0)
+
+/**
+ * Counts the trail bytes for a UTF-8 lead byte of a valid UTF-8 sequence.
+ * Returns 0 for 0..0xc1. Undefined for 0xf5..0xff.
+ * leadByte might be evaluated multiple times.
+ *
+ * This is internal since it is not meant to be called directly by external clients;
+ * however it is called by public macros in this file and thus must remain stable.
+ *
+ * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff.
+ * @internal
+ */
+#define U8_COUNT_TRAIL_BYTES_UNSAFE(leadByte) \
+    (((uint8_t)(leadByte)>=0xc2)+((uint8_t)(leadByte)>=0xe0)+((uint8_t)(leadByte)>=0xf0))
+
+/**
+ * Mask a UTF-8 lead byte, leave only the lower bits that form part of the code point value.
+ *
+ * This is internal since it is not meant to be called directly by external clients;
+ * however it is called by public macros in this file and thus must remain stable.
+ * @internal
+ */
+#define U8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1)
+
+/**
+ * Internal bit vector for 3-byte UTF-8 validity check, for use in U8_IS_VALID_LEAD3_AND_T1.
+ * Each bit indicates whether one lead byte + first trail byte pair starts a valid sequence.
+ * Lead byte E0..EF bits 3..0 are used as byte index,
+ * first trail byte bits 7..5 are used as bit index into that byte.
+ * @see U8_IS_VALID_LEAD3_AND_T1
+ * @internal
+ */
+#define U8_LEAD3_T1_BITS "\x20\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x10\x30\x30"
+
+/**
+ * Internal 3-byte UTF-8 validity check.
+ * Non-zero if lead byte E0..EF and first trail byte 00..FF start a valid sequence.
+ * @internal
+ */
+#define U8_IS_VALID_LEAD3_AND_T1(lead, t1) (U8_LEAD3_T1_BITS[(lead)&0xf]&(1<<((uint8_t)(t1)>>5)))
+
+/**
+ * Internal bit vector for 4-byte UTF-8 validity check, for use in U8_IS_VALID_LEAD4_AND_T1.
+ * Each bit indicates whether one lead byte + first trail byte pair starts a valid sequence.
+ * First trail byte bits 7..4 are used as byte index,
+ * lead byte F0..F4 bits 2..0 are used as bit index into that byte.
+ * @see U8_IS_VALID_LEAD4_AND_T1
+ * @internal
+ */
+#define U8_LEAD4_T1_BITS "\x00\x00\x00\x00\x00\x00\x00\x00\x1E\x0F\x0F\x0F\x00\x00\x00\x00"
+
+/**
+ * Internal 4-byte UTF-8 validity check.
+ * Non-zero if lead byte F0..F4 and first trail byte 00..FF start a valid sequence.
+ * @internal
+ */
+#define U8_IS_VALID_LEAD4_AND_T1(lead, t1) (U8_LEAD4_T1_BITS[(uint8_t)(t1)>>4]&(1<<((lead)&7)))
+
+/**
+ * Function for handling "next code point" with error-checking.
+ *
+ * This is internal since it is not meant to be called directly by external clients;
+ * however it is called by public macros in this
+ * file and thus must remain stable, and should not be hidden when other internal
+ * functions are hidden (otherwise public macros would fail to compile).
+ * @internal
+ */
+U_CAPI UChar32 U_EXPORT2
+utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict);
+
+/**
+ * Function for handling "append code point" with error-checking.
+ *
+ * This is internal since it is not meant to be called directly by external clients;
+ * however it is called by public macros in this
+ * file and thus must remain stable, and should not be hidden when other internal
+ * functions are hidden (otherwise public macros would fail to compile).
+ * @internal
+ */
+U_CAPI int32_t U_EXPORT2
+utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError);
+
+/**
+ * Function for handling "previous code point" with error-checking.
+ *
+ * This is internal since it is not meant to be called directly by external clients;
+ * however it is called by public macros in this
+ * file and thus must remain stable, and should not be hidden when other internal
+ * functions are hidden (otherwise public macros would fail to compile).
+ * @internal
+ */
+U_CAPI UChar32 U_EXPORT2
+utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict);
+
+/**
+ * Function for handling "skip backward one code point" with error-checking.
+ *
+ * This is internal since it is not meant to be called directly by external clients;
+ * however it is called by public macros in this
+ * file and thus must remain stable, and should not be hidden when other internal
+ * functions are hidden (otherwise public macros would fail to compile).
+ * @internal
+ */
+U_CAPI int32_t U_EXPORT2
+utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i);
+
+/* single-code point definitions -------------------------------------------- */
+
+/**
+ * Does this code unit (byte) encode a code point by itself (US-ASCII 0..0x7f)?
+ * @param c 8-bit code unit (byte)
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U8_IS_SINGLE(c) (((c)&0x80)==0)
+
+/**
+ * Is this code unit (byte) a UTF-8 lead byte? (0xC2..0xF4)
+ * @param c 8-bit code unit (byte)
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U8_IS_LEAD(c) ((uint8_t)((c)-0xc2)<=0x32)
+// 0x32=0xf4-0xc2
+
+/**
+ * Is this code unit (byte) a UTF-8 trail byte? (0x80..0xBF)
+ * @param c 8-bit code unit (byte)
+ * @return true or false
+ * @stable ICU 2.4
+ */
+#define U8_IS_TRAIL(c) ((int8_t)(c)<-0x40)
+
+/**
+ * How many code units (bytes) are used for the UTF-8 encoding
+ * of this Unicode code point?
+ * @param c 32-bit code point
+ * @return 1..4, or 0 if c is a surrogate or not a Unicode code point
+ * @stable ICU 2.4
+ */
+#define U8_LENGTH(c) \
+    ((uint32_t)(c)<=0x7f ? 1 : \
+        ((uint32_t)(c)<=0x7ff ? 2 : \
+            ((uint32_t)(c)<=0xd7ff ? 3 : \
+                ((uint32_t)(c)<=0xdfff || (uint32_t)(c)>0x10ffff ? 0 : \
+                    ((uint32_t)(c)<=0xffff ? 3 : 4)\
+                ) \
+            ) \
+        ) \
+    )
+
+/**
+ * The maximum number of UTF-8 code units (bytes) per Unicode code point (U+0000..U+10ffff).
+ * @return 4
+ * @stable ICU 2.4
+ */
+#define U8_MAX_LENGTH 4
+
+/**
+ * Get a code point from a string at a random-access offset,
+ * without changing the offset.
+ * The offset may point to either the lead byte or one of the trail bytes
+ * for a code point, in which case the macro will read all of the bytes
+ * for the code point.
+ * The result is undefined if the offset points to an illegal UTF-8
+ * byte sequence.
+ * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
+ *
+ * @param s const uint8_t * string
+ * @param i string offset
+ * @param c output UChar32 variable
+ * @see U8_GET
+ * @stable ICU 2.4
+ */
+#define U8_GET_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    int32_t _u8_get_unsafe_index=(int32_t)(i); \
+    U8_SET_CP_START_UNSAFE(s, _u8_get_unsafe_index); \
+    U8_NEXT_UNSAFE(s, _u8_get_unsafe_index, c); \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Get a code point from a string at a random-access offset,
+ * without changing the offset.
+ * The offset may point to either the lead byte or one of the trail bytes
+ * for a code point, in which case the macro will read all of the bytes
+ * for the code point.
+ *
+ * The length can be negative for a NUL-terminated string.
+ *
+ * If the offset points to an illegal UTF-8 byte sequence, then
+ * c is set to a negative value.
+ * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
+ *
+ * @param s const uint8_t * string
+ * @param start int32_t starting string offset
+ * @param i int32_t string offset, must be start<=i=0xe0 ? \
+                ((c)<0xf0 ?  /* U+0800..U+FFFF except surrogates */ \
+                    U8_LEAD3_T1_BITS[(c)&=0xf]&(1<<((__t=(s)[i])>>5)) && \
+                    (__t&=0x3f, 1) \
+                :  /* U+10000..U+10FFFF */ \
+                    ((c)-=0xf0)<=4 && \
+                    U8_LEAD4_T1_BITS[(__t=(s)[i])>>4]&(1<<(c)) && \
+                    ((c)=((c)<<6)|(__t&0x3f), ++(i)!=(length)) && \
+                    (__t=(s)[i]-0x80)<=0x3f) && \
+                /* valid second-to-last trail byte */ \
+                ((c)=((c)<<6)|__t, ++(i)!=(length)) \
+            :  /* U+0080..U+07FF */ \
+                (c)>=0xc2 && ((c)&=0x1f, 1)) && \
+            /* last trail byte */ \
+            (__t=(s)[i]-0x80)<=0x3f && \
+            ((c)=((c)<<6)|__t, ++(i), 1)) { \
+        } else { \
+            (c)=(sub);  /* ill-formed*/ \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Append a code point to a string, overwriting 1 to 4 bytes.
+ * The offset points to the current end of the string contents
+ * and is advanced (post-increment).
+ * "Unsafe" macro, assumes a valid code point and sufficient space in the string.
+ * Otherwise, the result is undefined.
+ *
+ * @param s const uint8_t * string buffer
+ * @param i string offset
+ * @param c code point to append
+ * @see U8_APPEND
+ * @stable ICU 2.4
+ */
+#define U8_APPEND_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    uint32_t __uc=(c); \
+    if(__uc<=0x7f) { \
+        (s)[(i)++]=(uint8_t)__uc; \
+    } else { \
+        if(__uc<=0x7ff) { \
+            (s)[(i)++]=(uint8_t)((__uc>>6)|0xc0); \
+        } else { \
+            if(__uc<=0xffff) { \
+                (s)[(i)++]=(uint8_t)((__uc>>12)|0xe0); \
+            } else { \
+                (s)[(i)++]=(uint8_t)((__uc>>18)|0xf0); \
+                (s)[(i)++]=(uint8_t)(((__uc>>12)&0x3f)|0x80); \
+            } \
+            (s)[(i)++]=(uint8_t)(((__uc>>6)&0x3f)|0x80); \
+        } \
+        (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Append a code point to a string, overwriting 1 to 4 bytes.
+ * The offset points to the current end of the string contents
+ * and is advanced (post-increment).
+ * "Safe" macro, checks for a valid code point.
+ * If a non-ASCII code point is written, checks for sufficient space in the string.
+ * If the code point is not valid or trail bytes do not fit,
+ * then isError is set to true.
+ *
+ * @param s const uint8_t * string buffer
+ * @param i int32_t string offset, must be i>6)|0xc0); \
+        (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
+    } else if((__uc<=0xd7ff || (0xe000<=__uc && __uc<=0xffff)) && (i)+2<(capacity)) { \
+        (s)[(i)++]=(uint8_t)((__uc>>12)|0xe0); \
+        (s)[(i)++]=(uint8_t)(((__uc>>6)&0x3f)|0x80); \
+        (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
+    } else if(0xffff<__uc && __uc<=0x10ffff && (i)+3<(capacity)) { \
+        (s)[(i)++]=(uint8_t)((__uc>>18)|0xf0); \
+        (s)[(i)++]=(uint8_t)(((__uc>>12)&0x3f)|0x80); \
+        (s)[(i)++]=(uint8_t)(((__uc>>6)&0x3f)|0x80); \
+        (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
+    } else { \
+        (isError)=true; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Advance the string offset from one code point boundary to the next.
+ * (Post-incrementing iteration.)
+ * "Unsafe" macro, assumes well-formed UTF-8.
+ *
+ * @param s const uint8_t * string
+ * @param i string offset
+ * @see U8_FWD_1
+ * @stable ICU 2.4
+ */
+#define U8_FWD_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    (i)+=1+U8_COUNT_TRAIL_BYTES_UNSAFE((s)[i]); \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Advance the string offset from one code point boundary to the next.
+ * (Post-incrementing iteration.)
+ * "Safe" macro, checks for illegal sequences and for string boundaries.
+ *
+ * The length can be negative for a NUL-terminated string.
+ *
+ * @param s const uint8_t * string
+ * @param i int32_t string offset, must be i=0xf0 */ { \
+            if(U8_IS_VALID_LEAD4_AND_T1(__b, __t1) && \
+                    ++(i)!=(length) && U8_IS_TRAIL((s)[i]) && \
+                    ++(i)!=(length) && U8_IS_TRAIL((s)[i])) { \
+                ++(i); \
+            } \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Advance the string offset from one code point boundary to the n-th next one,
+ * i.e., move forward by n code points.
+ * (Post-incrementing iteration.)
+ * "Unsafe" macro, assumes well-formed UTF-8.
+ *
+ * @param s const uint8_t * string
+ * @param i string offset
+ * @param n number of code points to skip
+ * @see U8_FWD_N
+ * @stable ICU 2.4
+ */
+#define U8_FWD_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
+    int32_t __N=(n); \
+    while(__N>0) { \
+        U8_FWD_1_UNSAFE(s, i); \
+        --__N; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Advance the string offset from one code point boundary to the n-th next one,
+ * i.e., move forward by n code points.
+ * (Post-incrementing iteration.)
+ * "Safe" macro, checks for illegal sequences and for string boundaries.
+ *
+ * The length can be negative for a NUL-terminated string.
+ *
+ * @param s const uint8_t * string
+ * @param i int32_t string offset, must be i0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
+        U8_FWD_1(s, i, length); \
+        --__N; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Adjust a random-access offset to a code point boundary
+ * at the start of a code point.
+ * If the offset points to a UTF-8 trail byte,
+ * then the offset is moved backward to the corresponding lead byte.
+ * Otherwise, it is not modified.
+ * "Unsafe" macro, assumes well-formed UTF-8.
+ *
+ * @param s const uint8_t * string
+ * @param i string offset
+ * @see U8_SET_CP_START
+ * @stable ICU 2.4
+ */
+#define U8_SET_CP_START_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    while(U8_IS_TRAIL((s)[i])) { --(i); } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Adjust a random-access offset to a code point boundary
+ * at the start of a code point.
+ * If the offset points to a UTF-8 trail byte,
+ * then the offset is moved backward to the corresponding lead byte.
+ * Otherwise, it is not modified.
+ *
+ * "Safe" macro, checks for illegal sequences and for string boundaries.
+ * Unlike U8_TRUNCATE_IF_INCOMPLETE(), this macro always reads s[i].
+ *
+ * @param s const uint8_t * string
+ * @param start int32_t starting string offset (usually 0)
+ * @param i int32_t string offset, must be start<=i
+ * @see U8_SET_CP_START_UNSAFE
+ * @see U8_TRUNCATE_IF_INCOMPLETE
+ * @stable ICU 2.4
+ */
+#define U8_SET_CP_START(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
+    if(U8_IS_TRAIL((s)[(i)])) { \
+        (i)=utf8_back1SafeBody(s, start, (i)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * If the string ends with a UTF-8 byte sequence that is valid so far
+ * but incomplete, then reduce the length of the string to end before
+ * the lead byte of that incomplete sequence.
+ * For example, if the string ends with E1 80, the length is reduced by 2.
+ *
+ * In all other cases (the string ends with a complete sequence, or it is not
+ * possible for any further trail byte to extend the trailing sequence)
+ * the length remains unchanged.
+ *
+ * Useful for processing text split across multiple buffers
+ * (save the incomplete sequence for later)
+ * and for optimizing iteration
+ * (check for string length only once per character).
+ *
+ * "Safe" macro, checks for illegal sequences and for string boundaries.
+ * Unlike U8_SET_CP_START(), this macro never reads s[length].
+ *
+ * (In UTF-16, simply check for U16_IS_LEAD(last code unit).)
+ *
+ * @param s const uint8_t * string
+ * @param start int32_t starting string offset (usually 0)
+ * @param length int32_t string length (usually start<=length)
+ * @see U8_SET_CP_START
+ * @stable ICU 61
+ */
+#define U8_TRUNCATE_IF_INCOMPLETE(s, start, length) UPRV_BLOCK_MACRO_BEGIN { \
+    if((length)>(start)) { \
+        uint8_t __b1=s[(length)-1]; \
+        if(U8_IS_SINGLE(__b1)) { \
+            /* common ASCII character */ \
+        } else if(U8_IS_LEAD(__b1)) { \
+            --(length); \
+        } else if(U8_IS_TRAIL(__b1) && ((length)-2)>=(start)) { \
+            uint8_t __b2=s[(length)-2]; \
+            if(0xe0<=__b2 && __b2<=0xf4) { \
+                if(__b2<0xf0 ? U8_IS_VALID_LEAD3_AND_T1(__b2, __b1) : \
+                        U8_IS_VALID_LEAD4_AND_T1(__b2, __b1)) { \
+                    (length)-=2; \
+                } \
+            } else if(U8_IS_TRAIL(__b2) && ((length)-3)>=(start)) { \
+                uint8_t __b3=s[(length)-3]; \
+                if(0xf0<=__b3 && __b3<=0xf4 && U8_IS_VALID_LEAD4_AND_T1(__b3, __b2)) { \
+                    (length)-=3; \
+                } \
+            } \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/* definitions with backward iteration -------------------------------------- */
+
+/**
+ * Move the string offset from one code point boundary to the previous one
+ * and get the code point between them.
+ * (Pre-decrementing backward iteration.)
+ * "Unsafe" macro, assumes well-formed UTF-8.
+ *
+ * The input offset may be the same as the string length.
+ * If the offset is behind a multi-byte sequence, then the macro will read
+ * the whole sequence.
+ * If the offset is behind a lead byte, then that itself
+ * will be returned as the code point.
+ * The result is undefined if the offset is behind an illegal UTF-8 sequence.
+ *
+ * @param s const uint8_t * string
+ * @param i string offset
+ * @param c output UChar32 variable
+ * @see U8_PREV
+ * @stable ICU 2.4
+ */
+#define U8_PREV_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(uint8_t)(s)[--(i)]; \
+    if(U8_IS_TRAIL(c)) { \
+        uint8_t __b, __count=1, __shift=6; \
+\
+        /* c is a trail byte */ \
+        (c)&=0x3f; \
+        for(;;) { \
+            __b=(s)[--(i)]; \
+            if(__b>=0xc0) { \
+                U8_MASK_LEAD_BYTE(__b, __count); \
+                (c)|=(UChar32)__b<<__shift; \
+                break; \
+            } else { \
+                (c)|=(UChar32)(__b&0x3f)<<__shift; \
+                ++__count; \
+                __shift+=6; \
+            } \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Move the string offset from one code point boundary to the previous one
+ * and get the code point between them.
+ * (Pre-decrementing backward iteration.)
+ * "Safe" macro, checks for illegal sequences and for string boundaries.
+ *
+ * The input offset may be the same as the string length.
+ * If the offset is behind a multi-byte sequence, then the macro will read
+ * the whole sequence.
+ * If the offset is behind a lead byte, then that itself
+ * will be returned as the code point.
+ * If the offset is behind an illegal UTF-8 sequence, then c is set to a negative value.
+ *
+ * @param s const uint8_t * string
+ * @param start int32_t starting string offset (usually 0)
+ * @param i int32_t string offset, must be start0) { \
+        U8_BACK_1_UNSAFE(s, i); \
+        --__N; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Move the string offset from one code point boundary to the n-th one before it,
+ * i.e., move backward by n code points.
+ * (Pre-decrementing backward iteration.)
+ * The input offset may be the same as the string length.
+ * "Safe" macro, checks for illegal sequences and for string boundaries.
+ *
+ * @param s const uint8_t * string
+ * @param start int32_t index of the start of the string
+ * @param i int32_t string offset, must be start0 && (i)>(start)) { \
+        U8_BACK_1(s, start, i); \
+        --__N; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Adjust a random-access offset to a code point boundary after a code point.
+ * If the offset is behind a partial multi-byte sequence,
+ * then the offset is incremented to behind the whole sequence.
+ * Otherwise, it is not modified.
+ * The input offset may be the same as the string length.
+ * "Unsafe" macro, assumes well-formed UTF-8.
+ *
+ * @param s const uint8_t * string
+ * @param i string offset
+ * @see U8_SET_CP_LIMIT
+ * @stable ICU 2.4
+ */
+#define U8_SET_CP_LIMIT_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    U8_BACK_1_UNSAFE(s, i); \
+    U8_FWD_1_UNSAFE(s, i); \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Adjust a random-access offset to a code point boundary after a code point.
+ * If the offset is behind a partial multi-byte sequence,
+ * then the offset is incremented to behind the whole sequence.
+ * Otherwise, it is not modified.
+ * The input offset may be the same as the string length.
+ * "Safe" macro, checks for illegal sequences and for string boundaries.
+ *
+ * The length can be negative for a NUL-terminated string.
+ *
+ * @param s const uint8_t * string
+ * @param start int32_t starting string offset (usually 0)
+ * @param i int32_t string offset, must be start<=i<=length
+ * @param length int32_t string length
+ * @see U8_SET_CP_LIMIT_UNSAFE
+ * @stable ICU 2.4
+ */
+#define U8_SET_CP_LIMIT(s, start, i, length) UPRV_BLOCK_MACRO_BEGIN { \
+    if((start)<(i) && ((i)<(length) || (length)<0)) { \
+        U8_BACK_1(s, start, i); \
+        U8_FWD_1(s, i, length); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+#endif
diff --git a/intl/icu/source/common/unicode/utf_old.h b/intl/icu/source/common/unicode/utf_old.h
new file mode 100644
index 0000000000..6b868c7280
--- /dev/null
+++ b/intl/icu/source/common/unicode/utf_old.h
@@ -0,0 +1,1201 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2002-2012, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  utf_old.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2002sep21
+*   created by: Markus W. Scherer
+*/
+
+/**
+ * \file
+ * \brief C API: Deprecated macros for Unicode string handling
+ *
+ * The macros in utf_old.h are all deprecated and their use discouraged.
+ * Some of the design principles behind the set of UTF macros
+ * have changed or proved impractical.
+ * Almost all of the old "UTF macros" are at least renamed.
+ * If you are looking for a new equivalent to an old macro, please see the
+ * comment at the old one.
+ *
+ * Brief summary of reasons for deprecation:
+ * - Switch on UTF_SIZE (selection of UTF-8/16/32 default string processing)
+ *   was impractical.
+ * - Switch on UTF_SAFE etc. (selection of unsafe/safe/strict default string processing)
+ *   was of little use and impractical.
+ * - Whole classes of macros became obsolete outside of the UTF_SIZE/UTF_SAFE
+ *   selection framework: UTF32_ macros (all trivial)
+ *   and UTF_ default and intermediate macros (all aliases).
+ * - The selection framework also caused many macro aliases.
+ * - Change in Unicode standard: "irregular" sequences (3.0) became illegal (3.2).
+ * - Change of language in Unicode standard:
+ *   Growing distinction between internal x-bit Unicode strings and external UTF-x
+ *   forms, with the former more lenient.
+ *   Suggests renaming of UTF16_ macros to U16_.
+ * - The prefix "UTF_" without a width number confused some users.
+ * - "Safe" append macros needed the addition of an error indicator output.
+ * - "Safe" UTF-8 macros used legitimate (if rarely used) code point values
+ *   to indicate error conditions.
+ * - The use of the "_CHAR" infix for code point operations confused some users.
+ *
+ * More details:
+ *
+ * Until ICU 2.2, utf.h theoretically allowed to choose among UTF-8/16/32
+ * for string processing, and among unsafe/safe/strict default macros for that.
+ *
+ * It proved nearly impossible to write non-trivial, high-performance code
+ * that is UTF-generic.
+ * Unsafe default macros would be dangerous for default string processing,
+ * and the main reason for the "strict" versions disappeared:
+ * Between Unicode 3.0 and 3.2 all "irregular" UTF-8 sequences became illegal.
+ * The only other conditions that "strict" checked for were non-characters,
+ * which are valid during processing. Only during text input/output should they
+ * be checked, and at that time other well-formedness checks may be
+ * necessary or useful as well.
+ * This can still be done by using U16_NEXT and U_IS_UNICODE_NONCHAR
+ * or U_IS_UNICODE_CHAR.
+ *
+ * The old UTF8_..._SAFE macros also used some normal Unicode code points
+ * to indicate malformed sequences.
+ * The new UTF8_ macros without suffix use negative values instead.
+ *
+ * The entire contents of utf32.h was moved here without replacement
+ * because all those macros were trivial and
+ * were meaningful only in the framework of choosing the UTF size.
+ *
+ * See Jitterbug 2150 and its discussion on the ICU mailing list
+ * in September 2002.
+ *
+ * 

+ *
+ * Obsolete part of pre-ICU 2.4 utf.h file documentation:
+ *
+ * 
The original concept for these files was for ICU to allow
+ * in principle to set which UTF (UTF-8/16/32) is used internally
+ * by defining UTF_SIZE to either 8, 16, or 32. utf.h would then define the UChar type
+ * accordingly. UTF-16 was the default.

+ *
+ * 
This concept has been abandoned.
+ * A lot of the ICU source code assumes UChar strings are in UTF-16.
+ * This is especially true for low-level code like
+ * conversion, normalization, and collation.
+ * The utf.h header enforces the default of UTF-16.
+ * The UTF-8 and UTF-32 macros remain for now for completeness and backward compatibility.

+ *
+ * 
Accordingly, utf.h defines UChar to be an unsigned 16-bit integer. If this matches wchar_t, then
+ * UChar is defined to be exactly wchar_t, otherwise uint16_t.

+ *
+ * 
UChar32 is defined to be a signed 32-bit integer (int32_t), large enough for a 21-bit
+ * Unicode code point (Unicode scalar value, 0..0x10ffff).
+ * Before ICU 2.4, the definition of UChar32 was similarly platform-dependent as
+ * the definition of UChar. For details see the documentation for UChar32 itself.

+ *
+ * 
utf.h also defines a number of C macros for handling single Unicode code points and
+ * for using UTF Unicode strings. It includes utf8.h, utf16.h, and utf32.h for the actual
+ * implementations of those macros and then aliases one set of them (for UTF-16) for general use.
+ * The UTF-specific macros have the UTF size in the macro name prefixes (UTF16_...), while
+ * the general alias macros always begin with UTF_...

+ *
+ * 
Many string operations can be done with or without error checking.
+ * Where such a distinction is useful, there are two versions of the macros, "unsafe" and "safe"
+ * ones with ..._UNSAFE and ..._SAFE suffixes. The unsafe macros are fast but may cause
+ * program failures if the strings are not well-formed. The safe macros have an additional, boolean
+ * parameter "strict". If strict is false, then only illegal sequences are detected.
+ * Otherwise, irregular sequences and non-characters are detected as well (like single surrogates).
+ * Safe macros return special error code points for illegal/irregular sequences:
+ * Typically, U+ffff, or values that would result in a code unit sequence of the same length
+ * as the erroneous input sequence.

+ * Note that _UNSAFE macros have fewer parameters: They do not have the strictness parameter, and
+ * they do not have start/length parameters for boundary checking.

+ *
+ * 
Here, the macros are aliased in two steps:
+ * In the first step, the UTF-specific macros with UTF16_ prefix and _UNSAFE and _SAFE suffixes are
+ * aliased according to the UTF_SIZE to macros with UTF_ prefix and the same suffixes and signatures.
+ * Then, in a second step, the default, general alias macros are set to use either the unsafe or
+ * the safe/not strict (default) or the safe/strict macro;
+ * these general macros do not have a strictness parameter.

+ *
+ * 
It is possible to change the default choice for the general alias macros to be unsafe, safe/not strict or safe/strict.
+ * The default is safe/not strict. It is not recommended to select the unsafe macros as the basis for
+ * Unicode string handling in ICU! To select this, define UTF_SAFE, UTF_STRICT, or UTF_UNSAFE.

+ *
+ * 
For general use, one should use the default, general macros with UTF_ prefix and no _SAFE/_UNSAFE suffix.
+ * Only in some cases it may be necessary to control the choice of macro directly and use a less generic alias.
+ * For example, if it can be assumed that a string is well-formed and the index will stay within the bounds,
+ * then the _UNSAFE version may be used.
+ * If a UTF-8 string is to be processed, then the macros with UTF8_ prefixes need to be used.

+ *
+ * 

+ *
+ * Deprecated ICU 2.4. Use the macros in utf.h, utf16.h, utf8.h instead.
+ */
+
+#ifndef __UTF_OLD_H__
+#define __UTF_OLD_H__
+
+#include "unicode/utf.h"
+#include "unicode/utf8.h"
+#include "unicode/utf16.h"
+
+/**
+ * \def U_HIDE_OBSOLETE_UTF_OLD_H
+ *
+ * Hides the obsolete definitions in unicode/utf_old.h.
+ * Recommended to be set to 1 at compile time to make sure
+ * the long-deprecated macros are no longer used.
+ *
+ * For reasons for the deprecation see the utf_old.h file comments.
+ *
+ * @internal
+ */
+#ifndef U_HIDE_OBSOLETE_UTF_OLD_H
+#   define U_HIDE_OBSOLETE_UTF_OLD_H 0
+#endif
+
+#if !defined(U_HIDE_DEPRECATED_API) && !U_HIDE_OBSOLETE_UTF_OLD_H
+
+/* Formerly utf.h, part 1 --------------------------------------------------- */
+
+#ifdef U_USE_UTF_DEPRECATES
+/**
+ * Unicode string and array offset and index type.
+ * ICU always counts Unicode code units (UChars) for
+ * string offsets, indexes, and lengths, not Unicode code points.
+ *
+ * @obsolete ICU 2.6. Use int32_t directly instead since this API will be removed in that release.
+ */
+typedef int32_t UTextOffset;
+#endif
+
+/** Number of bits in a Unicode string code unit - ICU uses 16-bit Unicode. @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF_SIZE 16
+
+/**
+ * The default choice for general Unicode string macros is to use the ..._SAFE macro implementations
+ * with strict=false.
+ *
+ * @deprecated ICU 2.4. Obsolete, see utf_old.h.
+ */
+#define UTF_SAFE
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#undef UTF_UNSAFE
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#undef UTF_STRICT
+
+/**
+ * UTF8_ERROR_VALUE_1 and UTF8_ERROR_VALUE_2 are special error values for UTF-8,
+ * which need 1 or 2 bytes in UTF-8:
+ * \code
+ * U+0015 = NAK = Negative Acknowledge, C0 control character
+ * U+009f = highest C1 control character
+ * \endcode
+ *
+ * These are used by UTF8_..._SAFE macros so that they can return an error value
+ * that needs the same number of code units (bytes) as were seen by
+ * a macro. They should be tested with UTF_IS_ERROR() or UTF_IS_VALID().
+ *
+ * @deprecated ICU 2.4. Obsolete, see utf_old.h.
+ */
+#define UTF8_ERROR_VALUE_1 0x15
+
+/**
+ * See documentation on UTF8_ERROR_VALUE_1 for details.
+ *
+ * @deprecated ICU 2.4. Obsolete, see utf_old.h.
+ */
+#define UTF8_ERROR_VALUE_2 0x9f
+
+/**
+ * Error value for all UTFs. This code point value will be set by macros with error
+ * checking if an error is detected.
+ *
+ * @deprecated ICU 2.4. Obsolete, see utf_old.h.
+ */
+#define UTF_ERROR_VALUE 0xffff
+
+/**
+ * Is a given 32-bit code an error value
+ * as returned by one of the macros for any UTF?
+ *
+ * @deprecated ICU 2.4. Obsolete, see utf_old.h.
+ */
+#define UTF_IS_ERROR(c) \
+    (((c)&0xfffe)==0xfffe || (c)==UTF8_ERROR_VALUE_1 || (c)==UTF8_ERROR_VALUE_2)
+
+/**
+ * This is a combined macro: Is c a valid Unicode value _and_ not an error code?
+ *
+ * @deprecated ICU 2.4. Obsolete, see utf_old.h.
+ */
+#define UTF_IS_VALID(c) \
+    (UTF_IS_UNICODE_CHAR(c) && \
+     (c)!=UTF8_ERROR_VALUE_1 && (c)!=UTF8_ERROR_VALUE_2)
+
+/**
+ * Is this code unit or code point a surrogate (U+d800..U+dfff)?
+ * @deprecated ICU 2.4. Renamed to U_IS_SURROGATE and U16_IS_SURROGATE, see utf_old.h.
+ */
+#define UTF_IS_SURROGATE(uchar) (((uchar)&0xfffff800)==0xd800)
+
+/**
+ * Is a given 32-bit code point a Unicode noncharacter?
+ *
+ * @deprecated ICU 2.4. Renamed to U_IS_UNICODE_NONCHAR, see utf_old.h.
+ */
+#define UTF_IS_UNICODE_NONCHAR(c) \
+    ((c)>=0xfdd0 && \
+     ((uint32_t)(c)<=0xfdef || ((c)&0xfffe)==0xfffe) && \
+     (uint32_t)(c)<=0x10ffff)
+
+/**
+ * Is a given 32-bit value a Unicode code point value (0..U+10ffff)
+ * that can be assigned a character?
+ *
+ * Code points that are not characters include:
+ * - single surrogate code points (U+d800..U+dfff, 2048 code points)
+ * - the last two code points on each plane (U+__fffe and U+__ffff, 34 code points)
+ * - U+fdd0..U+fdef (new with Unicode 3.1, 32 code points)
+ * - the highest Unicode code point value is U+10ffff
+ *
+ * This means that all code points below U+d800 are character code points,
+ * and that boundary is tested first for performance.
+ *
+ * @deprecated ICU 2.4. Renamed to U_IS_UNICODE_CHAR, see utf_old.h.
+ */
+#define UTF_IS_UNICODE_CHAR(c) \
+    ((uint32_t)(c)<0xd800 || \
+        ((uint32_t)(c)>0xdfff && \
+         (uint32_t)(c)<=0x10ffff && \
+         !UTF_IS_UNICODE_NONCHAR(c)))
+
+/* Formerly utf8.h ---------------------------------------------------------- */
+
+/**
+* \var utf8_countTrailBytes
+* Internal array with numbers of trail bytes for any given byte used in
+* lead byte position.
+*
+* This is internal since it is not meant to be called directly by external clients;
+* however it is called by public macros in this file and thus must remain stable,
+* and should not be hidden when other internal functions are hidden (otherwise
+* public macros would fail to compile).
+* @internal
+*/
+#ifdef U_UTF8_IMPL
+// No forward declaration if compiling utf_impl.cpp, which defines utf8_countTrailBytes.
+#elif defined(U_STATIC_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION)
+U_CAPI const uint8_t utf8_countTrailBytes[];
+#else
+U_CFUNC U_IMPORT const uint8_t utf8_countTrailBytes[];
+#endif 
+
+/**
+ * Count the trail bytes for a UTF-8 lead byte.
+ * @deprecated ICU 2.4. Renamed to U8_COUNT_TRAIL_BYTES, see utf_old.h.
+ */
+#define UTF8_COUNT_TRAIL_BYTES(leadByte) (utf8_countTrailBytes[(uint8_t)leadByte])
+
+/**
+ * Mask a UTF-8 lead byte, leave only the lower bits that form part of the code point value.
+ * @deprecated ICU 2.4. Renamed to U8_MASK_LEAD_BYTE, see utf_old.h.
+ */
+#define UTF8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1)
+
+/** Is this this code point a single code unit (byte)? @deprecated ICU 2.4. Renamed to U8_IS_SINGLE, see utf_old.h. */
+#define UTF8_IS_SINGLE(uchar) (((uchar)&0x80)==0)
+/** Is this this code unit the lead code unit (byte) of a code point? @deprecated ICU 2.4. Renamed to U8_IS_LEAD, see utf_old.h. */
+#define UTF8_IS_LEAD(uchar) ((uint8_t)((uchar)-0xc0)<0x3e)
+/** Is this this code unit a trailing code unit (byte) of a code point? @deprecated ICU 2.4. Renamed to U8_IS_TRAIL, see utf_old.h. */
+#define UTF8_IS_TRAIL(uchar) (((uchar)&0xc0)==0x80)
+
+/** Does this scalar Unicode value need multiple code units for storage? @deprecated ICU 2.4. Use U8_LENGTH or test ((uint32_t)(c)>0x7f) instead, see utf_old.h. */
+#define UTF8_NEED_MULTIPLE_UCHAR(c) ((uint32_t)(c)>0x7f)
+
+/**
+ * Given the lead character, how many bytes are taken by this code point.
+ * ICU does not deal with code points >0x10ffff
+ * unless necessary for advancing in the byte stream.
+ *
+ * These length macros take into account that for values >0x10ffff
+ * the UTF8_APPEND_CHAR_SAFE macros would write the error code point 0xffff
+ * with 3 bytes.
+ * Code point comparisons need to be in uint32_t because UChar32
+ * may be a signed type, and negative values must be recognized.
+ *
+ * @deprecated ICU 2.4. Use U8_LENGTH instead, see utf.h.
+ */
+#if 1
+#   define UTF8_CHAR_LENGTH(c) \
+        ((uint32_t)(c)<=0x7f ? 1 : \
+            ((uint32_t)(c)<=0x7ff ? 2 : \
+                ((uint32_t)((c)-0x10000)>0xfffff ? 3 : 4) \
+            ) \
+        )
+#else
+#   define UTF8_CHAR_LENGTH(c) \
+        ((uint32_t)(c)<=0x7f ? 1 : \
+            ((uint32_t)(c)<=0x7ff ? 2 : \
+                ((uint32_t)(c)<=0xffff ? 3 : \
+                    ((uint32_t)(c)<=0x10ffff ? 4 : \
+                        ((uint32_t)(c)<=0x3ffffff ? 5 : \
+                            ((uint32_t)(c)<=0x7fffffff ? 6 : 3) \
+                        ) \
+                    ) \
+                ) \
+            ) \
+        )
+#endif
+
+/** The maximum number of bytes per code point. @deprecated ICU 2.4. Renamed to U8_MAX_LENGTH, see utf_old.h. */
+#define UTF8_MAX_CHAR_LENGTH 4
+
+/** Average number of code units compared to UTF-16. @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF8_ARRAY_SIZE(size) ((5*(size))/2)
+
+/** @deprecated ICU 2.4. Renamed to U8_GET_UNSAFE, see utf_old.h. */
+#define UTF8_GET_CHAR_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    int32_t _utf8_get_char_unsafe_index=(int32_t)(i); \
+    UTF8_SET_CHAR_START_UNSAFE(s, _utf8_get_char_unsafe_index); \
+    UTF8_NEXT_CHAR_UNSAFE(s, _utf8_get_char_unsafe_index, c); \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Use U8_GET instead, see utf_old.h. */
+#define UTF8_GET_CHAR_SAFE(s, start, i, length, c, strict) UPRV_BLOCK_MACRO_BEGIN { \
+    int32_t _utf8_get_char_safe_index=(int32_t)(i); \
+    UTF8_SET_CHAR_START_SAFE(s, start, _utf8_get_char_safe_index); \
+    UTF8_NEXT_CHAR_SAFE(s, _utf8_get_char_safe_index, length, c, strict); \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U8_NEXT_UNSAFE, see utf_old.h. */
+#define UTF8_NEXT_CHAR_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[(i)++]; \
+    if((uint8_t)((c)-0xc0)<0x35) { \
+        uint8_t __count=UTF8_COUNT_TRAIL_BYTES(c); \
+        UTF8_MASK_LEAD_BYTE(c, __count); \
+        switch(__count) { \
+        /* each following branch falls through to the next one */ \
+        case 3: \
+            (c)=((c)<<6)|((s)[(i)++]&0x3f); \
+        case 2: \
+            (c)=((c)<<6)|((s)[(i)++]&0x3f); \
+        case 1: \
+            (c)=((c)<<6)|((s)[(i)++]&0x3f); \
+        /* no other branches to optimize switch() */ \
+            break; \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U8_APPEND_UNSAFE, see utf_old.h. */
+#define UTF8_APPEND_CHAR_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    if((uint32_t)(c)<=0x7f) { \
+        (s)[(i)++]=(uint8_t)(c); \
+    } else { \
+        if((uint32_t)(c)<=0x7ff) { \
+            (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
+        } else { \
+            if((uint32_t)(c)<=0xffff) { \
+                (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
+            } else { \
+                (s)[(i)++]=(uint8_t)(((c)>>18)|0xf0); \
+                (s)[(i)++]=(uint8_t)((((c)>>12)&0x3f)|0x80); \
+            } \
+            (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
+        } \
+        (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U8_FWD_1_UNSAFE, see utf_old.h. */
+#define UTF8_FWD_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    (i)+=1+UTF8_COUNT_TRAIL_BYTES((s)[i]); \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U8_FWD_N_UNSAFE, see utf_old.h. */
+#define UTF8_FWD_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
+    int32_t __N=(n); \
+    while(__N>0) { \
+        UTF8_FWD_1_UNSAFE(s, i); \
+        --__N; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U8_SET_CP_START_UNSAFE, see utf_old.h. */
+#define UTF8_SET_CHAR_START_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    while(UTF8_IS_TRAIL((s)[i])) { --(i); } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Use U8_NEXT instead, see utf_old.h. */
+#define UTF8_NEXT_CHAR_SAFE(s, i, length, c, strict) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[(i)++]; \
+    if((c)>=0x80) { \
+        if(UTF8_IS_LEAD(c)) { \
+            (c)=utf8_nextCharSafeBody(s, &(i), (int32_t)(length), c, strict); \
+        } else { \
+            (c)=UTF8_ERROR_VALUE_1; \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Use U8_APPEND instead, see utf_old.h. */
+#define UTF8_APPEND_CHAR_SAFE(s, i, length, c)  UPRV_BLOCK_MACRO_BEGIN { \
+    if((uint32_t)(c)<=0x7f) { \
+        (s)[(i)++]=(uint8_t)(c); \
+    } else { \
+        (i)=utf8_appendCharSafeBody(s, (int32_t)(i), (int32_t)(length), c, NULL); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U8_FWD_1, see utf_old.h. */
+#define UTF8_FWD_1_SAFE(s, i, length) U8_FWD_1(s, i, length)
+
+/** @deprecated ICU 2.4. Renamed to U8_FWD_N, see utf_old.h. */
+#define UTF8_FWD_N_SAFE(s, i, length, n) U8_FWD_N(s, i, length, n)
+
+/** @deprecated ICU 2.4. Renamed to U8_SET_CP_START, see utf_old.h. */
+#define UTF8_SET_CHAR_START_SAFE(s, start, i) U8_SET_CP_START(s, start, i)
+
+/** @deprecated ICU 2.4. Renamed to U8_PREV_UNSAFE, see utf_old.h. */
+#define UTF8_PREV_CHAR_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[--(i)]; \
+    if(UTF8_IS_TRAIL(c)) { \
+        uint8_t __b, __count=1, __shift=6; \
+\
+        /* c is a trail byte */ \
+        (c)&=0x3f; \
+        for(;;) { \
+            __b=(s)[--(i)]; \
+            if(__b>=0xc0) { \
+                UTF8_MASK_LEAD_BYTE(__b, __count); \
+                (c)|=(UChar32)__b<<__shift; \
+                break; \
+            } else { \
+                (c)|=(UChar32)(__b&0x3f)<<__shift; \
+                ++__count; \
+                __shift+=6; \
+            } \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U8_BACK_1_UNSAFE, see utf_old.h. */
+#define UTF8_BACK_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    while(UTF8_IS_TRAIL((s)[--(i)])) {} \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U8_BACK_N_UNSAFE, see utf_old.h. */
+#define UTF8_BACK_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
+    int32_t __N=(n); \
+    while(__N>0) { \
+        UTF8_BACK_1_UNSAFE(s, i); \
+        --__N; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U8_SET_CP_LIMIT_UNSAFE, see utf_old.h. */
+#define UTF8_SET_CHAR_LIMIT_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    UTF8_BACK_1_UNSAFE(s, i); \
+    UTF8_FWD_1_UNSAFE(s, i); \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Use U8_PREV instead, see utf_old.h. */
+#define UTF8_PREV_CHAR_SAFE(s, start, i, c, strict) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[--(i)]; \
+    if((c)>=0x80) { \
+        if((c)<=0xbf) { \
+            (c)=utf8_prevCharSafeBody(s, start, &(i), c, strict); \
+        } else { \
+            (c)=UTF8_ERROR_VALUE_1; \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U8_BACK_1, see utf_old.h. */
+#define UTF8_BACK_1_SAFE(s, start, i) U8_BACK_1(s, start, i)
+
+/** @deprecated ICU 2.4. Renamed to U8_BACK_N, see utf_old.h. */
+#define UTF8_BACK_N_SAFE(s, start, i, n) U8_BACK_N(s, start, i, n)
+
+/** @deprecated ICU 2.4. Renamed to U8_SET_CP_LIMIT, see utf_old.h. */
+#define UTF8_SET_CHAR_LIMIT_SAFE(s, start, i, length) U8_SET_CP_LIMIT(s, start, i, length)
+
+/* Formerly utf16.h --------------------------------------------------------- */
+
+/** Is uchar a first/lead surrogate? @deprecated ICU 2.4. Renamed to U_IS_LEAD and U16_IS_LEAD, see utf_old.h. */
+#define UTF_IS_FIRST_SURROGATE(uchar) (((uchar)&0xfffffc00)==0xd800)
+
+/** Is uchar a second/trail surrogate? @deprecated ICU 2.4. Renamed to U_IS_TRAIL and U16_IS_TRAIL, see utf_old.h. */
+#define UTF_IS_SECOND_SURROGATE(uchar) (((uchar)&0xfffffc00)==0xdc00)
+
+/** Assuming c is a surrogate, is it a first/lead surrogate? @deprecated ICU 2.4. Renamed to U_IS_SURROGATE_LEAD and U16_IS_SURROGATE_LEAD, see utf_old.h. */
+#define UTF_IS_SURROGATE_FIRST(c) (((c)&0x400)==0)
+
+/** Helper constant for UTF16_GET_PAIR_VALUE. @deprecated ICU 2.4. Renamed to U16_SURROGATE_OFFSET, see utf_old.h. */
+#define UTF_SURROGATE_OFFSET ((0xd800<<10UL)+0xdc00-0x10000)
+
+/** Get the UTF-32 value from the surrogate code units. @deprecated ICU 2.4. Renamed to U16_GET_SUPPLEMENTARY, see utf_old.h. */
+#define UTF16_GET_PAIR_VALUE(first, second) \
+    (((first)<<10UL)+(second)-UTF_SURROGATE_OFFSET)
+
+/** @deprecated ICU 2.4. Renamed to U16_LEAD, see utf_old.h. */
+#define UTF_FIRST_SURROGATE(supplementary) (UChar)(((supplementary)>>10)+0xd7c0)
+
+/** @deprecated ICU 2.4. Renamed to U16_TRAIL, see utf_old.h. */
+#define UTF_SECOND_SURROGATE(supplementary) (UChar)(((supplementary)&0x3ff)|0xdc00)
+
+/** @deprecated ICU 2.4. Renamed to U16_LEAD, see utf_old.h. */
+#define UTF16_LEAD(supplementary) UTF_FIRST_SURROGATE(supplementary)
+
+/** @deprecated ICU 2.4. Renamed to U16_TRAIL, see utf_old.h. */
+#define UTF16_TRAIL(supplementary) UTF_SECOND_SURROGATE(supplementary)
+
+/** @deprecated ICU 2.4. Renamed to U16_IS_SINGLE, see utf_old.h. */
+#define UTF16_IS_SINGLE(uchar) !UTF_IS_SURROGATE(uchar)
+
+/** @deprecated ICU 2.4. Renamed to U16_IS_LEAD, see utf_old.h. */
+#define UTF16_IS_LEAD(uchar) UTF_IS_FIRST_SURROGATE(uchar)
+
+/** @deprecated ICU 2.4. Renamed to U16_IS_TRAIL, see utf_old.h. */
+#define UTF16_IS_TRAIL(uchar) UTF_IS_SECOND_SURROGATE(uchar)
+
+/** Does this scalar Unicode value need multiple code units for storage? @deprecated ICU 2.4. Use U16_LENGTH or test ((uint32_t)(c)>0xffff) instead, see utf_old.h. */
+#define UTF16_NEED_MULTIPLE_UCHAR(c) ((uint32_t)(c)>0xffff)
+
+/** @deprecated ICU 2.4. Renamed to U16_LENGTH, see utf_old.h. */
+#define UTF16_CHAR_LENGTH(c) ((uint32_t)(c)<=0xffff ? 1 : 2)
+
+/** @deprecated ICU 2.4. Renamed to U16_MAX_LENGTH, see utf_old.h. */
+#define UTF16_MAX_CHAR_LENGTH 2
+
+/** Average number of code units compared to UTF-16. @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF16_ARRAY_SIZE(size) (size)
+
+/**
+ * Get a single code point from an offset that points to any
+ * of the code units that belong to that code point.
+ * Assume 0<=i=(start) && UTF_IS_FIRST_SURROGATE(__c2=(s)[(i)-1])) { \
+                (c)=UTF16_GET_PAIR_VALUE(__c2, (c)); \
+                /* strict: ((c)&0xfffe)==0xfffe is caught by UTF_IS_ERROR() and UTF_IS_UNICODE_CHAR() */ \
+            } else if(strict) {\
+                /* unmatched second surrogate */ \
+                (c)=UTF_ERROR_VALUE; \
+            } \
+        } \
+    } else if((strict) && !UTF_IS_UNICODE_CHAR(c)) { \
+        (c)=UTF_ERROR_VALUE; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U16_NEXT_UNSAFE, see utf_old.h. */
+#define UTF16_NEXT_CHAR_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[(i)++]; \
+    if(UTF_IS_FIRST_SURROGATE(c)) { \
+        (c)=UTF16_GET_PAIR_VALUE((c), (s)[(i)++]); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U16_APPEND_UNSAFE, see utf_old.h. */
+#define UTF16_APPEND_CHAR_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    if((uint32_t)(c)<=0xffff) { \
+        (s)[(i)++]=(uint16_t)(c); \
+    } else { \
+        (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \
+        (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U16_FWD_1_UNSAFE, see utf_old.h. */
+#define UTF16_FWD_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTF_IS_FIRST_SURROGATE((s)[(i)++])) { \
+        ++(i); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U16_FWD_N_UNSAFE, see utf_old.h. */
+#define UTF16_FWD_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
+    int32_t __N=(n); \
+    while(__N>0) { \
+        UTF16_FWD_1_UNSAFE(s, i); \
+        --__N; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U16_SET_CP_START_UNSAFE, see utf_old.h. */
+#define UTF16_SET_CHAR_START_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTF_IS_SECOND_SURROGATE((s)[i])) { \
+        --(i); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Use U16_NEXT instead, see utf_old.h. */
+#define UTF16_NEXT_CHAR_SAFE(s, i, length, c, strict) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[(i)++]; \
+    if(UTF_IS_FIRST_SURROGATE(c)) { \
+        uint16_t __c2; \
+        if((i)<(length) && UTF_IS_SECOND_SURROGATE(__c2=(s)[(i)])) { \
+            ++(i); \
+            (c)=UTF16_GET_PAIR_VALUE((c), __c2); \
+            /* strict: ((c)&0xfffe)==0xfffe is caught by UTF_IS_ERROR() and UTF_IS_UNICODE_CHAR() */ \
+        } else if(strict) {\
+            /* unmatched first surrogate */ \
+            (c)=UTF_ERROR_VALUE; \
+        } \
+    } else if((strict) && !UTF_IS_UNICODE_CHAR(c)) { \
+        /* unmatched second surrogate or other non-character */ \
+        (c)=UTF_ERROR_VALUE; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Use U16_APPEND instead, see utf_old.h. */
+#define UTF16_APPEND_CHAR_SAFE(s, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
+    if((uint32_t)(c)<=0xffff) { \
+        (s)[(i)++]=(uint16_t)(c); \
+    } else if((uint32_t)(c)<=0x10ffff) { \
+        if((i)+1<(length)) { \
+            (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \
+            (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
+        } else /* not enough space */ { \
+            (s)[(i)++]=UTF_ERROR_VALUE; \
+        } \
+    } else /* c>0x10ffff, write error value */ { \
+        (s)[(i)++]=UTF_ERROR_VALUE; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U16_FWD_1, see utf_old.h. */
+#define UTF16_FWD_1_SAFE(s, i, length) U16_FWD_1(s, i, length)
+
+/** @deprecated ICU 2.4. Renamed to U16_FWD_N, see utf_old.h. */
+#define UTF16_FWD_N_SAFE(s, i, length, n) U16_FWD_N(s, i, length, n)
+
+/** @deprecated ICU 2.4. Renamed to U16_SET_CP_START, see utf_old.h. */
+#define UTF16_SET_CHAR_START_SAFE(s, start, i) U16_SET_CP_START(s, start, i)
+
+/** @deprecated ICU 2.4. Renamed to U16_PREV_UNSAFE, see utf_old.h. */
+#define UTF16_PREV_CHAR_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[--(i)]; \
+    if(UTF_IS_SECOND_SURROGATE(c)) { \
+        (c)=UTF16_GET_PAIR_VALUE((s)[--(i)], (c)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U16_BACK_1_UNSAFE, see utf_old.h. */
+#define UTF16_BACK_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTF_IS_SECOND_SURROGATE((s)[--(i)])) { \
+        --(i); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U16_BACK_N_UNSAFE, see utf_old.h. */
+#define UTF16_BACK_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
+    int32_t __N=(n); \
+    while(__N>0) { \
+        UTF16_BACK_1_UNSAFE(s, i); \
+        --__N; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U16_SET_CP_LIMIT_UNSAFE, see utf_old.h. */
+#define UTF16_SET_CHAR_LIMIT_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTF_IS_FIRST_SURROGATE((s)[(i)-1])) { \
+        ++(i); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Use U16_PREV instead, see utf_old.h. */
+#define UTF16_PREV_CHAR_SAFE(s, start, i, c, strict) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[--(i)]; \
+    if(UTF_IS_SECOND_SURROGATE(c)) { \
+        uint16_t __c2; \
+        if((i)>(start) && UTF_IS_FIRST_SURROGATE(__c2=(s)[(i)-1])) { \
+            --(i); \
+            (c)=UTF16_GET_PAIR_VALUE(__c2, (c)); \
+            /* strict: ((c)&0xfffe)==0xfffe is caught by UTF_IS_ERROR() and UTF_IS_UNICODE_CHAR() */ \
+        } else if(strict) {\
+            /* unmatched second surrogate */ \
+            (c)=UTF_ERROR_VALUE; \
+        } \
+    } else if((strict) && !UTF_IS_UNICODE_CHAR(c)) { \
+        /* unmatched first surrogate or other non-character */ \
+        (c)=UTF_ERROR_VALUE; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Renamed to U16_BACK_1, see utf_old.h. */
+#define UTF16_BACK_1_SAFE(s, start, i) U16_BACK_1(s, start, i)
+
+/** @deprecated ICU 2.4. Renamed to U16_BACK_N, see utf_old.h. */
+#define UTF16_BACK_N_SAFE(s, start, i, n) U16_BACK_N(s, start, i, n)
+
+/** @deprecated ICU 2.4. Renamed to U16_SET_CP_LIMIT, see utf_old.h. */
+#define UTF16_SET_CHAR_LIMIT_SAFE(s, start, i, length) U16_SET_CP_LIMIT(s, start, i, length)
+
+/* Formerly utf32.h --------------------------------------------------------- */
+
+/*
+* Old documentation:
+*
+*   This file defines macros to deal with UTF-32 code units and code points.
+*   Signatures and semantics are the same as for the similarly named macros
+*   in utf16.h.
+*   utf32.h is included by utf.h after unicode/umachine.h

+*   and some common definitions.
+*   
Usage:  ICU coding guidelines for if() statements should be followed when using these macros.
+*                  Compound statements (curly braces {}) must be used  for if-else-while...
+*                  bodies and all macro statements should be terminated with semicolon.

+*/
+
+/* internal definitions ----------------------------------------------------- */
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_IS_SAFE(c, strict) \
+    (!(strict) ? \
+        (uint32_t)(c)<=0x10ffff : \
+        UTF_IS_UNICODE_CHAR(c))
+
+/*
+ * For the semantics of all of these macros, see utf16.h.
+ * The UTF-32 versions are trivial because any code point is
+ * encoded using exactly one code unit.
+ */
+
+/* single-code point definitions -------------------------------------------- */
+
+/* classes of code unit values */
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_IS_SINGLE(uchar) 1
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_IS_LEAD(uchar) 0
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_IS_TRAIL(uchar) 0
+
+/* number of code units per code point */
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_NEED_MULTIPLE_UCHAR(c) 0
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_CHAR_LENGTH(c) 1
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_MAX_CHAR_LENGTH 1
+
+/* average number of code units compared to UTF-16 */
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_ARRAY_SIZE(size) (size)
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_GET_CHAR_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[i]; \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_GET_CHAR_SAFE(s, start, i, length, c, strict) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[i]; \
+    if(!UTF32_IS_SAFE(c, strict)) { \
+        (c)=UTF_ERROR_VALUE; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/* definitions with forward iteration --------------------------------------- */
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_NEXT_CHAR_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[(i)++]; \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_APPEND_CHAR_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    (s)[(i)++]=(c); \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_FWD_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    ++(i); \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_FWD_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
+    (i)+=(n); \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_SET_CHAR_START_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_NEXT_CHAR_SAFE(s, i, length, c, strict) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[(i)++]; \
+    if(!UTF32_IS_SAFE(c, strict)) { \
+        (c)=UTF_ERROR_VALUE; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_APPEND_CHAR_SAFE(s, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
+    if((uint32_t)(c)<=0x10ffff) { \
+        (s)[(i)++]=(c); \
+    } else /* c>0x10ffff, write 0xfffd */ { \
+        (s)[(i)++]=0xfffd; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_FWD_1_SAFE(s, i, length) UPRV_BLOCK_MACRO_BEGIN { \
+    ++(i); \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_FWD_N_SAFE(s, i, length, n) UPRV_BLOCK_MACRO_BEGIN { \
+    if(((i)+=(n))>(length)) { \
+        (i)=(length); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_SET_CHAR_START_SAFE(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
+} UPRV_BLOCK_MACRO_END
+
+/* definitions with backward iteration -------------------------------------- */
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_PREV_CHAR_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[--(i)]; \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_BACK_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+    --(i); \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_BACK_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
+    (i)-=(n); \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_SET_CHAR_LIMIT_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_PREV_CHAR_SAFE(s, start, i, c, strict) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=(s)[--(i)]; \
+    if(!UTF32_IS_SAFE(c, strict)) { \
+        (c)=UTF_ERROR_VALUE; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_BACK_1_SAFE(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
+    --(i); \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_BACK_N_SAFE(s, start, i, n) UPRV_BLOCK_MACRO_BEGIN { \
+    (i)-=(n); \
+    if((i)<(start)) { \
+        (i)=(start); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** @deprecated ICU 2.4. Obsolete, see utf_old.h. */
+#define UTF32_SET_CHAR_LIMIT_SAFE(s, i, length) UPRV_BLOCK_MACRO_BEGIN { \
+} UPRV_BLOCK_MACRO_END
+
+/* Formerly utf.h, part 2 --------------------------------------------------- */
+
+/**
+ * Estimate the number of code units for a string based on the number of UTF-16 code units.
+ *
+ * @deprecated ICU 2.4. Obsolete, see utf_old.h.
+ */
+#define UTF_ARRAY_SIZE(size) UTF16_ARRAY_SIZE(size)
+
+/** @deprecated ICU 2.4. Renamed to U16_GET_UNSAFE, see utf_old.h. */
+#define UTF_GET_CHAR_UNSAFE(s, i, c)                 UTF16_GET_CHAR_UNSAFE(s, i, c)
+
+/** @deprecated ICU 2.4. Use U16_GET instead, see utf_old.h. */
+#define UTF_GET_CHAR_SAFE(s, start, i, length, c, strict) UTF16_GET_CHAR_SAFE(s, start, i, length, c, strict)
+
+
+/** @deprecated ICU 2.4. Renamed to U16_NEXT_UNSAFE, see utf_old.h. */
+#define UTF_NEXT_CHAR_UNSAFE(s, i, c)                UTF16_NEXT_CHAR_UNSAFE(s, i, c)
+
+/** @deprecated ICU 2.4. Use U16_NEXT instead, see utf_old.h. */
+#define UTF_NEXT_CHAR_SAFE(s, i, length, c, strict)  UTF16_NEXT_CHAR_SAFE(s, i, length, c, strict)
+
+
+/** @deprecated ICU 2.4. Renamed to U16_APPEND_UNSAFE, see utf_old.h. */
+#define UTF_APPEND_CHAR_UNSAFE(s, i, c)              UTF16_APPEND_CHAR_UNSAFE(s, i, c)
+
+/** @deprecated ICU 2.4. Use U16_APPEND instead, see utf_old.h. */
+#define UTF_APPEND_CHAR_SAFE(s, i, length, c)        UTF16_APPEND_CHAR_SAFE(s, i, length, c)
+
+
+/** @deprecated ICU 2.4. Renamed to U16_FWD_1_UNSAFE, see utf_old.h. */
+#define UTF_FWD_1_UNSAFE(s, i)                       UTF16_FWD_1_UNSAFE(s, i)
+
+/** @deprecated ICU 2.4. Renamed to U16_FWD_1, see utf_old.h. */
+#define UTF_FWD_1_SAFE(s, i, length)                 UTF16_FWD_1_SAFE(s, i, length)
+
+
+/** @deprecated ICU 2.4. Renamed to U16_FWD_N_UNSAFE, see utf_old.h. */
+#define UTF_FWD_N_UNSAFE(s, i, n)                    UTF16_FWD_N_UNSAFE(s, i, n)
+
+/** @deprecated ICU 2.4. Renamed to U16_FWD_N, see utf_old.h. */
+#define UTF_FWD_N_SAFE(s, i, length, n)              UTF16_FWD_N_SAFE(s, i, length, n)
+
+
+/** @deprecated ICU 2.4. Renamed to U16_SET_CP_START_UNSAFE, see utf_old.h. */
+#define UTF_SET_CHAR_START_UNSAFE(s, i)              UTF16_SET_CHAR_START_UNSAFE(s, i)
+
+/** @deprecated ICU 2.4. Renamed to U16_SET_CP_START, see utf_old.h. */
+#define UTF_SET_CHAR_START_SAFE(s, start, i)         UTF16_SET_CHAR_START_SAFE(s, start, i)
+
+
+/** @deprecated ICU 2.4. Renamed to U16_PREV_UNSAFE, see utf_old.h. */
+#define UTF_PREV_CHAR_UNSAFE(s, i, c)                UTF16_PREV_CHAR_UNSAFE(s, i, c)
+
+/** @deprecated ICU 2.4. Use U16_PREV instead, see utf_old.h. */
+#define UTF_PREV_CHAR_SAFE(s, start, i, c, strict)   UTF16_PREV_CHAR_SAFE(s, start, i, c, strict)
+
+
+/** @deprecated ICU 2.4. Renamed to U16_BACK_1_UNSAFE, see utf_old.h. */
+#define UTF_BACK_1_UNSAFE(s, i)                      UTF16_BACK_1_UNSAFE(s, i)
+
+/** @deprecated ICU 2.4. Renamed to U16_BACK_1, see utf_old.h. */
+#define UTF_BACK_1_SAFE(s, start, i)                 UTF16_BACK_1_SAFE(s, start, i)
+
+
+/** @deprecated ICU 2.4. Renamed to U16_BACK_N_UNSAFE, see utf_old.h. */
+#define UTF_BACK_N_UNSAFE(s, i, n)                   UTF16_BACK_N_UNSAFE(s, i, n)
+
+/** @deprecated ICU 2.4. Renamed to U16_BACK_N, see utf_old.h. */
+#define UTF_BACK_N_SAFE(s, start, i, n)              UTF16_BACK_N_SAFE(s, start, i, n)
+
+
+/** @deprecated ICU 2.4. Renamed to U16_SET_CP_LIMIT_UNSAFE, see utf_old.h. */
+#define UTF_SET_CHAR_LIMIT_UNSAFE(s, i)              UTF16_SET_CHAR_LIMIT_UNSAFE(s, i)
+
+/** @deprecated ICU 2.4. Renamed to U16_SET_CP_LIMIT, see utf_old.h. */
+#define UTF_SET_CHAR_LIMIT_SAFE(s, start, i, length) UTF16_SET_CHAR_LIMIT_SAFE(s, start, i, length)
+
+/* Define default macros (UTF-16 "safe") ------------------------------------ */
+
+/**
+ * Does this code unit alone encode a code point (BMP, not a surrogate)?
+ * Same as UTF16_IS_SINGLE.
+ * @deprecated ICU 2.4. Renamed to U_IS_SINGLE and U16_IS_SINGLE, see utf_old.h.
+ */
+#define UTF_IS_SINGLE(uchar) U16_IS_SINGLE(uchar)
+
+/**
+ * Is this code unit the first one of several (a lead surrogate)?
+ * Same as UTF16_IS_LEAD.
+ * @deprecated ICU 2.4. Renamed to U_IS_LEAD and U16_IS_LEAD, see utf_old.h.
+ */
+#define UTF_IS_LEAD(uchar) U16_IS_LEAD(uchar)
+
+/**
+ * Is this code unit one of several but not the first one (a trail surrogate)?
+ * Same as UTF16_IS_TRAIL.
+ * @deprecated ICU 2.4. Renamed to U_IS_TRAIL and U16_IS_TRAIL, see utf_old.h.
+ */
+#define UTF_IS_TRAIL(uchar) U16_IS_TRAIL(uchar)
+
+/**
+ * Does this code point require multiple code units (is it a supplementary code point)?
+ * Same as UTF16_NEED_MULTIPLE_UCHAR.
+ * @deprecated ICU 2.4. Use U16_LENGTH or test ((uint32_t)(c)>0xffff) instead.
+ */
+#define UTF_NEED_MULTIPLE_UCHAR(c) UTF16_NEED_MULTIPLE_UCHAR(c)
+
+/**
+ * How many code units are used to encode this code point (1 or 2)?
+ * Same as UTF16_CHAR_LENGTH.
+ * @deprecated ICU 2.4. Renamed to U16_LENGTH, see utf_old.h.
+ */
+#define UTF_CHAR_LENGTH(c) U16_LENGTH(c)
+
+/**
+ * How many code units are used at most for any Unicode code point (2)?
+ * Same as UTF16_MAX_CHAR_LENGTH.
+ * @deprecated ICU 2.4. Renamed to U16_MAX_LENGTH, see utf_old.h.
+ */
+#define UTF_MAX_CHAR_LENGTH U16_MAX_LENGTH
+
+/**
+ * Set c to the code point that contains the code unit i.
+ * i could point to the lead or the trail surrogate for the code point.
+ * i is not modified.
+ * Same as UTF16_GET_CHAR.
+ * \pre 0<=i
+#include "unicode/utypes.h"
+
+/**
+ * \file
+ * \brief C API:  Definitions for ICU tracing/logging. 
+ *
+ * This provides API for debugging the internals of ICU without the use of
+ * a traditional debugger.
+ *
+ * By default, tracing is disabled in ICU. If you need to debug ICU with 
+ * tracing, please compile ICU with the --enable-tracing configure option.
+ */
+ 
+U_CDECL_BEGIN
+
+/**
+ * Trace severity levels.  Higher levels increase the verbosity of the trace output.
+ * @see utrace_setLevel
+ * @stable ICU 2.8
+ */
+typedef enum UTraceLevel {
+    /** Disable all tracing  @stable ICU 2.8*/
+    UTRACE_OFF=-1,
+    /** Trace error conditions only  @stable ICU 2.8*/
+    UTRACE_ERROR=0,
+    /** Trace errors and warnings  @stable ICU 2.8*/
+    UTRACE_WARNING=3,
+    /** Trace opens and closes of ICU services  @stable ICU 2.8*/
+    UTRACE_OPEN_CLOSE=5,
+    /** Trace an intermediate number of ICU operations  @stable ICU 2.8*/
+    UTRACE_INFO=7,
+    /** Trace the maximum number of ICU operations  @stable ICU 2.8*/
+    UTRACE_VERBOSE=9
+} UTraceLevel;
+
+/**
+ *  These are the ICU functions that will be traced when tracing is enabled.
+ *  @stable ICU 2.8
+ */
+typedef enum UTraceFunctionNumber {
+    UTRACE_FUNCTION_START=0,
+    UTRACE_U_INIT=UTRACE_FUNCTION_START,
+    UTRACE_U_CLEANUP,
+
+#ifndef U_HIDE_DEPRECATED_API
+    /**
+     * One more than the highest normal collation trace location.
+     * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
+     */
+    UTRACE_FUNCTION_LIMIT,
+#endif  // U_HIDE_DEPRECATED_API
+
+    UTRACE_CONVERSION_START=0x1000,
+    UTRACE_UCNV_OPEN=UTRACE_CONVERSION_START,
+    UTRACE_UCNV_OPEN_PACKAGE,
+    UTRACE_UCNV_OPEN_ALGORITHMIC,
+    UTRACE_UCNV_CLONE,
+    UTRACE_UCNV_CLOSE,
+    UTRACE_UCNV_FLUSH_CACHE,
+    UTRACE_UCNV_LOAD,
+    UTRACE_UCNV_UNLOAD,
+
+#ifndef U_HIDE_DEPRECATED_API
+    /**
+     * One more than the highest normal collation trace location.
+     * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
+     */
+    UTRACE_CONVERSION_LIMIT,
+#endif  // U_HIDE_DEPRECATED_API
+
+    UTRACE_COLLATION_START=0x2000,
+    UTRACE_UCOL_OPEN=UTRACE_COLLATION_START,
+    UTRACE_UCOL_CLOSE,
+    UTRACE_UCOL_STRCOLL,
+    UTRACE_UCOL_GET_SORTKEY,
+    UTRACE_UCOL_GETLOCALE,
+    UTRACE_UCOL_NEXTSORTKEYPART,
+    UTRACE_UCOL_STRCOLLITER,
+    UTRACE_UCOL_OPEN_FROM_SHORT_STRING,
+    UTRACE_UCOL_STRCOLLUTF8, /**< @stable ICU 50 */
+
+#ifndef U_HIDE_DEPRECATED_API
+    /**
+     * One more than the highest normal collation trace location.
+     * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
+     */
+    UTRACE_COLLATION_LIMIT,
+#endif  // U_HIDE_DEPRECATED_API
+
+    /**
+     * The lowest resource/data location.
+     * @stable ICU 65
+     */
+    UTRACE_UDATA_START=0x3000,
+
+    /**
+     * Indicates that a value was read from a resource bundle. Provides three
+     * C-style strings to UTraceData: type, file name, and resource path. The
+     * possible types are:
+     *
+     * - "string" (a string value was accessed)
+     * - "binary" (a binary value was accessed)
+     * - "intvector" (a integer vector value was accessed)
+     * - "int" (a signed integer value was accessed)
+     * - "uint" (a unsigned integer value was accessed)
+     * - "get" (a path was loaded, but the value was not accessed)
+     * - "getalias" (a path was loaded, and an alias was resolved)
+     *
+     * @stable ICU 65
+     */
+    UTRACE_UDATA_RESOURCE=UTRACE_UDATA_START,
+
+    /**
+     * Indicates that a resource bundle was opened.
+     *
+     * Provides one C-style string to UTraceData: file name.
+     * @stable ICU 65
+     */
+    UTRACE_UDATA_BUNDLE,
+
+    /**
+     * Indicates that a data file was opened, but not *.res files.
+     *
+     * Provides one C-style string to UTraceData: file name.
+     *
+     * @stable ICU 65
+     */
+    UTRACE_UDATA_DATA_FILE,
+
+    /**
+     * Indicates that a *.res file was opened.
+     *
+     * This differs from UTRACE_UDATA_BUNDLE because a res file is typically
+     * opened only once per application runtime, but the bundle corresponding
+     * to that res file may be opened many times.
+     *
+     * Provides one C-style string to UTraceData: file name.
+     *
+     * @stable ICU 65
+     */
+    UTRACE_UDATA_RES_FILE,
+
+#ifndef U_HIDE_INTERNAL_API
+    /**
+     * One more than the highest normal resource/data trace location.
+     * @internal The numeric value may change over time, see ICU ticket #12420.
+     */
+    UTRACE_RES_DATA_LIMIT,
+#endif  // U_HIDE_INTERNAL_API
+
+    /**
+     * The lowest break iterator location.
+     * @stable ICU 67
+     */
+    UTRACE_UBRK_START=0x4000,
+
+    /**
+     * Indicates that a character instance of break iterator was created.
+     *
+     * @stable ICU 67
+     */
+    UTRACE_UBRK_CREATE_CHARACTER = UTRACE_UBRK_START,
+
+    /**
+     * Indicates that a word instance of break iterator was created.
+     *
+     * @stable ICU 67
+     */
+    UTRACE_UBRK_CREATE_WORD,
+
+    /**
+     * Indicates that a line instance of break iterator was created.
+     *
+     * Provides one C-style string to UTraceData: the lb value ("",
+     * "loose", "strict", or "normal").
+     *
+     * @stable ICU 67
+     */
+    UTRACE_UBRK_CREATE_LINE,
+
+    /**
+     * Indicates that a sentence instance of break iterator was created.
+     *
+     * @stable ICU 67
+     */
+    UTRACE_UBRK_CREATE_SENTENCE,
+
+    /**
+     * Indicates that a title instance of break iterator was created.
+     *
+     * @stable ICU 67
+     */
+    UTRACE_UBRK_CREATE_TITLE,
+
+    /**
+     * Indicates that an internal dictionary break engine was created.
+     *
+     * Provides one C-style string to UTraceData: the script code of what
+     * the break engine cover ("Hani", "Khmr", "Laoo", "Mymr", or "Thai").
+     *
+     * @stable ICU 67
+     */
+    UTRACE_UBRK_CREATE_BREAK_ENGINE,
+
+#ifndef U_HIDE_INTERNAL_API
+    /**
+     * One more than the highest normal break iterator trace location.
+     * @internal The numeric value may change over time, see ICU ticket #12420.
+     */
+    UTRACE_UBRK_LIMIT,
+#endif  // U_HIDE_INTERNAL_API
+
+} UTraceFunctionNumber;
+
+/**
+ * Setter for the trace level.
+ * @param traceLevel A UTraceLevel value.
+ * @stable ICU 2.8
+ */
+U_CAPI void U_EXPORT2
+utrace_setLevel(int32_t traceLevel);
+
+/**
+ * Getter for the trace level.
+ * @return The UTraceLevel value being used by ICU.
+ * @stable ICU 2.8
+ */
+U_CAPI int32_t U_EXPORT2
+utrace_getLevel(void);
+
+/* Trace function pointers types  ----------------------------- */
+
+/**
+  *  Type signature for the trace function to be called when entering a function.
+  *  @param context value supplied at the time the trace functions are set.
+  *  @param fnNumber Enum value indicating the ICU function being entered.
+  *  @stable ICU 2.8
+  */
+typedef void U_CALLCONV
+UTraceEntry(const void *context, int32_t fnNumber);
+
+/**
+  *  Type signature for the trace function to be called when exiting from a function.
+  *  @param context value supplied at the time the trace functions are set.
+  *  @param fnNumber Enum value indicating the ICU function being exited.
+  *  @param fmt     A formatting string that describes the number and types
+  *                 of arguments included with the variable args.  The fmt
+  *                 string has the same form as the utrace_vformat format
+  *                 string.
+  *  @param args    A variable arguments list.  Contents are described by
+  *                 the fmt parameter.
+  *  @see   utrace_vformat
+  *  @stable ICU 2.8
+  */
+typedef void U_CALLCONV
+UTraceExit(const void *context, int32_t fnNumber, 
+           const char *fmt, va_list args);
+
+/**
+  *  Type signature for the trace function to be called from within an ICU function
+  *  to display data or messages.
+  *  @param context  value supplied at the time the trace functions are set.
+  *  @param fnNumber Enum value indicating the ICU function being exited.
+  *  @param level    The current tracing level
+  *  @param fmt      A format string describing the tracing data that is supplied
+  *                  as variable args
+  *  @param args     The data being traced, passed as variable args.
+  *  @stable ICU 2.8
+  */
+typedef void U_CALLCONV
+UTraceData(const void *context, int32_t fnNumber, int32_t level,
+           const char *fmt, va_list args);
+
+/**
+  *  Set ICU Tracing functions.  Installs application-provided tracing
+  *  functions into ICU.  After doing this, subsequent ICU operations
+  *  will call back to the installed functions, providing a trace
+  *  of the use of ICU.  Passing a NULL pointer for a tracing function
+  *  is allowed, and inhibits tracing action at points where that function
+  *  would be called.
+  *  

+  *  Tracing and Threads:  Tracing functions are global to a process, and
+  *  will be called in response to ICU operations performed by any
+  *  thread.  If tracing of an individual thread is desired, the
+  *  tracing functions must themselves filter by checking that the
+  *  current thread is the desired thread.
+  *
+  *  @param context an uninterpreted pointer.  Whatever is passed in
+  *                 here will in turn be passed to each of the tracing
+  *                 functions UTraceEntry, UTraceExit and UTraceData.
+  *                 ICU does not use or alter this pointer.
+  *  @param e       Callback function to be called on entry to a 
+  *                 a traced ICU function.
+  *  @param x       Callback function to be called on exit from a
+  *                 traced ICU function.
+  *  @param d       Callback function to be called from within a 
+  *                 traced ICU function, for the purpose of providing
+  *                 data to the trace.
+  *
+  *  @stable ICU 2.8
+  */
+U_CAPI void U_EXPORT2
+utrace_setFunctions(const void *context,
+                    UTraceEntry *e, UTraceExit *x, UTraceData *d);
+
+/**
+  * Get the currently installed ICU tracing functions.   Note that a null function
+  *   pointer will be returned if no trace function has been set.
+  *
+  * @param context  The currently installed tracing context.
+  * @param e        The currently installed UTraceEntry function.
+  * @param x        The currently installed UTraceExit function.
+  * @param d        The currently installed UTraceData function.
+  * @stable ICU 2.8
+  */
+U_CAPI void U_EXPORT2
+utrace_getFunctions(const void **context,
+                    UTraceEntry **e, UTraceExit **x, UTraceData **d);
+
+
+
+/*
+ *
+ * ICU trace format string syntax
+ *
+ * Format Strings are passed to UTraceData functions, and define the
+ * number and types of the trace data being passed on each call.
+ *
+ * The UTraceData function, which is supplied by the application,
+ * not by ICU, can either forward the trace data (passed via
+ * varargs) and the format string back to ICU for formatting into
+ * a displayable string, or it can interpret the format itself,
+ * and do as it wishes with the trace data.
+ *
+ *
+ * Goals for the format string
+ * - basic data output
+ * - easy to use for trace programmer
+ * - sufficient provision for data types for trace output readability
+ * - well-defined types and binary portable APIs
+ *
+ * Non-goals
+ * - printf compatibility
+ * - fancy formatting
+ * - argument reordering and other internationalization features
+ *
+ * ICU trace format strings contain plain text with argument inserts,
+ * much like standard printf format strings.
+ * Each insert begins with a '%', then optionally contains a 'v',
+ * then exactly one type character.
+ * Two '%' in a row represent a '%' instead of an insert.
+ * The trace format strings need not have \n at the end.
+ *
+ *
+ * Types
+ * -----
+ *
+ * Type characters:
+ * - c A char character in the default codepage.
+ * - s A NUL-terminated char * string in the default codepage.
+ * - S A UChar * string.  Requires two params, (ptr, length).  Length=-1 for nul term.
+ * - b A byte (8-bit integer).
+ * - h A 16-bit integer.  Also a 16 bit Unicode code unit.
+ * - d A 32-bit integer.  Also a 20 bit Unicode code point value. 
+ * - l A 64-bit integer.
+ * - p A data pointer.
+ *
+ * Vectors
+ * -------
+ *
+ * If the 'v' is not specified, then one item of the specified type
+ * is passed in.
+ * If the 'v' (for "vector") is specified, then a vector of items of the
+ * specified type is passed in, via a pointer to the first item
+ * and an int32_t value for the length of the vector.
+ * Length==-1 means zero or NUL termination.  Works for vectors of all types.
+ *
+ * Note:  %vS is a vector of (UChar *) strings.  The strings must
+ *        be nul terminated as there is no way to provide a
+ *        separate length parameter for each string.  The length
+ *        parameter (required for all vectors) is the number of
+ *        strings, not the length of the strings.
+ *
+ * Examples
+ * --------
+ *
+ * These examples show the parameters that will be passed to an application's
+ *   UTraceData() function for various formats.
+ *
+ * - the precise formatting is up to the application!
+ * - the examples use type casts for arguments only to _show_ the types of
+ *   arguments without needing variable declarations in the examples;
+ *   the type casts will not be necessary in actual code
+ *
+ * UTraceDataFunc(context, fnNumber, level,
+ *              "There is a character %c in the string %s.",   // Format String 
+ *              (char)c, (const char *)s);                     // varargs parameters
+ * ->   There is a character 0x42 'B' in the string "Bravo".
+ *
+ * UTraceDataFunc(context, fnNumber, level,
+ *              "Vector of bytes %vb vector of chars %vc",
+ *              (const uint8_t *)bytes, (int32_t)bytesLength,
+ *              (const char *)chars, (int32_t)charsLength);
+ * ->  Vector of bytes
+ *      42 63 64 3f [4]
+ *     vector of chars
+ *      "Bcd?"[4]
+ *
+ * UTraceDataFunc(context, fnNumber, level,
+ *              "An int32_t %d and a whole bunch of them %vd",
+ *              (int32_t)-5, (const int32_t *)ints, (int32_t)intsLength);
+ * ->   An int32_t 0xfffffffb and a whole bunch of them
+ *      fffffffb 00000005 0000010a [3]
+ *
+ */
+
+
+
+/**
+  *  Trace output Formatter.  An application's UTraceData tracing functions may call
+  *                 back to this function to format the trace output in a
+  *                 human readable form.  Note that a UTraceData function may choose
+  *                 to not format the data;  it could, for example, save it in
+  *                 in the raw form it was received (more compact), leaving
+  *                 formatting for a later trace analysis tool.
+  *  @param outBuf  pointer to a buffer to receive the formatted output.  Output
+  *                 will be nul terminated if there is space in the buffer -
+  *                 if the length of the requested output < the output buffer size.
+  *  @param capacity  Length of the output buffer.
+  *  @param indent  Number of spaces to indent the output.  Intended to allow
+  *                 data displayed from nested functions to be indented for readability.
+  *  @param fmt     Format specification for the data to output
+  *  @param args    Data to be formatted.
+  *  @return        Length of formatted output, including the terminating NUL.
+  *                 If buffer capacity is insufficient, the required capacity is returned. 
+  *  @stable ICU 2.8
+  */
+U_CAPI int32_t U_EXPORT2
+utrace_vformat(char *outBuf, int32_t capacity,
+              int32_t indent, const char *fmt,  va_list args);
+
+/**
+  *  Trace output Formatter.  An application's UTraceData tracing functions may call
+  *                 this function to format any additional trace data, beyond that
+  *                 provided by default, in human readable form with the same
+  *                 formatting conventions used by utrace_vformat().
+  *  @param outBuf  pointer to a buffer to receive the formatted output.  Output
+  *                 will be nul terminated if there is space in the buffer -
+  *                 if the length of the requested output < the output buffer size.
+  *  @param capacity  Length of the output buffer.
+  *  @param indent  Number of spaces to indent the output.  Intended to allow
+  *                 data displayed from nested functions to be indented for readability.
+  *  @param fmt     Format specification for the data to output
+  *  @param ...     Data to be formatted.
+  *  @return        Length of formatted output, including the terminating NUL.
+  *                 If buffer capacity is insufficient, the required capacity is returned. 
+  *  @stable ICU 2.8
+  */
+U_CAPI int32_t U_EXPORT2
+utrace_format(char *outBuf, int32_t capacity,
+              int32_t indent, const char *fmt,  ...);
+
+
+
+/* Trace function numbers --------------------------------------------------- */
+
+/**
+ * Get the name of a function from its trace function number.
+ *
+ * @param fnNumber The trace number for an ICU function.
+ * @return The name string for the function.
+ *
+ * @see UTraceFunctionNumber
+ * @stable ICU 2.8
+ */
+U_CAPI const char * U_EXPORT2
+utrace_functionName(int32_t fnNumber);
+
+U_CDECL_END
+
+#endif
diff --git a/intl/icu/source/common/unicode/utypes.h b/intl/icu/source/common/unicode/utypes.h
new file mode 100644
index 0000000000..f890d5d1db
--- /dev/null
+++ b/intl/icu/source/common/unicode/utypes.h
@@ -0,0 +1,730 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+*   Copyright (C) 1996-2016, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*
+*  FILE NAME : UTYPES.H (formerly ptypes.h)
+*
+*   Date        Name        Description
+*   12/11/96    helena      Creation.
+*   02/27/97    aliu        Added typedefs for UClassID, int8, int16, int32,
+*                           uint8, uint16, and uint32.
+*   04/01/97    aliu        Added XP_CPLUSPLUS and modified to work under C as
+*                            well as C++.
+*                           Modified to use memcpy() for uprv_arrayCopy() fns.
+*   04/14/97    aliu        Added TPlatformUtilities.
+*   05/07/97    aliu        Added import/export specifiers (replacing the old
+*                           broken EXT_CLASS).  Added version number for our
+*                           code.  Cleaned up header.
+*    6/20/97    helena      Java class name change.
+*   08/11/98    stephen     UErrorCode changed from typedef to enum
+*   08/12/98    erm         Changed T_ANALYTIC_PACKAGE_VERSION to 3
+*   08/14/98    stephen     Added uprv_arrayCopy() for int8_t, int16_t, int32_t
+*   12/09/98    jfitz       Added BUFFER_OVERFLOW_ERROR (bug 1100066)
+*   04/20/99    stephen     Cleaned up & reworked for autoconf.
+*                           Renamed to utypes.h.
+*   05/05/99    stephen     Changed to use 
+*   12/07/99    helena      Moved copyright notice string from ucnv_bld.h here.
+*******************************************************************************
+*/
+
+#ifndef UTYPES_H
+#define UTYPES_H
+
+
+#include "unicode/umachine.h"
+#include "unicode/uversion.h"
+#include "unicode/uconfig.h"
+#include 
+
+#if !U_NO_DEFAULT_INCLUDE_UTF_HEADERS
+#   include "unicode/utf.h"
+#endif
+
+/*!
+ * \file
+ * \brief Basic definitions for ICU, for both C and C++ APIs
+ *
+ * This file defines basic types, constants, and enumerations directly or
+ * indirectly by including other header files, especially utf.h for the
+ * basic character and string definitions and umachine.h for consistent
+ * integer and other types.
+ */
+
+
+/**
+ * \def U_SHOW_CPLUSPLUS_API
+ * @internal
+ */
+#ifdef __cplusplus
+#   ifndef U_SHOW_CPLUSPLUS_API
+#       define U_SHOW_CPLUSPLUS_API 1
+#   endif
+#else
+#   undef U_SHOW_CPLUSPLUS_API
+#   define U_SHOW_CPLUSPLUS_API 0
+#endif
+
+/** @{ API visibility control */
+
+/**
+ * \def U_HIDE_DRAFT_API
+ * Define this to 1 to request that draft API be "hidden"
+ * @internal
+ */
+/**
+ * \def U_HIDE_INTERNAL_API
+ * Define this to 1 to request that internal API be "hidden"
+ * @internal
+ */
+#if !U_DEFAULT_SHOW_DRAFT && !defined(U_SHOW_DRAFT_API)
+#define U_HIDE_DRAFT_API 1
+#endif
+#if !U_DEFAULT_SHOW_DRAFT && !defined(U_SHOW_INTERNAL_API)
+#define U_HIDE_INTERNAL_API 1
+#endif
+
+/** @} */
+
+/*===========================================================================*/
+/* ICUDATA naming scheme                                                     */
+/*===========================================================================*/
+
+/**
+ * \def U_ICUDATA_TYPE_LETTER
+ *
+ * This is a platform-dependent string containing one letter:
+ * - b for big-endian, ASCII-family platforms
+ * - l for little-endian, ASCII-family platforms
+ * - e for big-endian, EBCDIC-family platforms
+ * This letter is part of the common data file name.
+ * @stable ICU 2.0
+ */
+
+/**
+ * \def U_ICUDATA_TYPE_LITLETTER
+ * The non-string form of U_ICUDATA_TYPE_LETTER
+ * @stable ICU 2.0
+ */
+#if U_CHARSET_FAMILY
+#   if U_IS_BIG_ENDIAN
+   /* EBCDIC - should always be BE */
+#     define U_ICUDATA_TYPE_LETTER "e"
+#     define U_ICUDATA_TYPE_LITLETTER e
+#   else
+#     error "Don't know what to do with little endian EBCDIC!"
+#     define U_ICUDATA_TYPE_LETTER "x"
+#     define U_ICUDATA_TYPE_LITLETTER x
+#   endif
+#else
+#   if U_IS_BIG_ENDIAN
+      /* Big-endian ASCII */
+#     define U_ICUDATA_TYPE_LETTER "b"
+#     define U_ICUDATA_TYPE_LITLETTER b
+#   else
+      /* Little-endian ASCII */
+#     define U_ICUDATA_TYPE_LETTER "l"
+#     define U_ICUDATA_TYPE_LITLETTER l
+#   endif
+#endif
+
+/**
+ * A single string literal containing the icudata stub name. i.e. 'icudt18e' for
+ * ICU 1.8.x on EBCDIC, etc..
+ * @stable ICU 2.0
+ */
+#define U_ICUDATA_NAME    "icudt" U_ICU_VERSION_SHORT U_ICUDATA_TYPE_LETTER
+#ifndef U_HIDE_INTERNAL_API
+#define U_USRDATA_NAME    "usrdt" U_ICU_VERSION_SHORT U_ICUDATA_TYPE_LETTER  /**< @internal */
+#define U_USE_USRDATA     0  /**< @internal */
+#endif  /* U_HIDE_INTERNAL_API */
+
+/**
+ *  U_ICU_ENTRY_POINT is the name of the DLL entry point to the ICU data library.
+ *    Defined as a literal, not a string.
+ *    Tricky Preprocessor use - ## operator replaces macro parameters with the literal string
+ *                              from the corresponding macro invocation, _before_ other macro substitutions.
+ *                              Need a nested \#defines to get the actual version numbers rather than
+ *                              the literal text U_ICU_VERSION_MAJOR_NUM into the name.
+ *                              The net result will be something of the form
+ *                                  \#define U_ICU_ENTRY_POINT icudt19_dat
+ * @stable ICU 2.4
+ */
+#define U_ICUDATA_ENTRY_POINT  U_DEF2_ICUDATA_ENTRY_POINT(U_ICU_VERSION_MAJOR_NUM,U_LIB_SUFFIX_C_NAME)
+
+#ifndef U_HIDE_INTERNAL_API
+/**
+ * Do not use. Note that it's OK for the 2nd argument to be undefined (literal).
+ * @internal
+ */
+#define U_DEF2_ICUDATA_ENTRY_POINT(major,suff) U_DEF_ICUDATA_ENTRY_POINT(major,suff)
+
+/**
+ * Do not use.
+ * @internal
+ */
+#ifndef U_DEF_ICUDATA_ENTRY_POINT
+/* affected by symbol renaming. See platform.h */
+#ifndef U_LIB_SUFFIX_C_NAME
+#define U_DEF_ICUDATA_ENTRY_POINT(major, suff) icudt##major##_dat
+#else
+#define U_DEF_ICUDATA_ENTRY_POINT(major, suff) icudt##suff ## major##_dat
+#endif
+#endif
+#endif  /* U_HIDE_INTERNAL_API */
+
+/**
+ * \def NULL
+ * Define NULL if necessary, to nullptr for C++ and to ((void *)0) for C.
+ * @stable ICU 2.0
+ */
+#ifndef NULL
+#ifdef __cplusplus
+#define NULL    nullptr
+#else
+#define NULL    ((void *)0)
+#endif
+#endif
+
+/*===========================================================================*/
+/* Calendar/TimeZone data types                                              */
+/*===========================================================================*/
+
+/**
+ * Date and Time data type.
+ * This is a primitive data type that holds the date and time
+ * as the number of milliseconds since 1970-jan-01, 00:00 UTC.
+ * UTC leap seconds are ignored.
+ * @stable ICU 2.0
+ */
+typedef double UDate;
+
+/** The number of milliseconds per second @stable ICU 2.0 */
+#define U_MILLIS_PER_SECOND        (1000)
+/** The number of milliseconds per minute @stable ICU 2.0 */
+#define U_MILLIS_PER_MINUTE       (60000)
+/** The number of milliseconds per hour @stable ICU 2.0 */
+#define U_MILLIS_PER_HOUR       (3600000)
+/** The number of milliseconds per day @stable ICU 2.0 */
+#define U_MILLIS_PER_DAY       (86400000)
+
+/** 
+ * Maximum UDate value 
+ * @stable ICU 4.8 
+ */ 
+#define U_DATE_MAX DBL_MAX
+
+/**
+ * Minimum UDate value 
+ * @stable ICU 4.8 
+ */ 
+#define U_DATE_MIN -U_DATE_MAX
+
+/*===========================================================================*/
+/* Shared library/DLL import-export API control                              */
+/*===========================================================================*/
+
+/*
+ * Control of symbol import/export.
+ * ICU is separated into three libraries.
+ */
+
+/**
+ * \def U_COMBINED_IMPLEMENTATION
+ * Set to export library symbols from inside the ICU library
+ * when all of ICU is in a single library.
+ * This can be set as a compiler option while building ICU, and it
+ * needs to be the first one tested to override U_COMMON_API, U_I18N_API, etc.
+ * @stable ICU 2.0
+ */
+
+/**
+ * \def U_DATA_API
+ * Set to export library symbols from inside the stubdata library,
+ * and to import them from outside.
+ * @stable ICU 3.0
+ */
+
+/**
+ * \def U_COMMON_API
+ * Set to export library symbols from inside the common library,
+ * and to import them from outside.
+ * @stable ICU 2.0
+ */
+
+/**
+ * \def U_I18N_API
+ * Set to export library symbols from inside the i18n library,
+ * and to import them from outside.
+ * @stable ICU 2.0
+ */
+
+/**
+ * \def U_LAYOUT_API
+ * Set to export library symbols from inside the layout engine library,
+ * and to import them from outside.
+ * @stable ICU 2.0
+ */
+
+/**
+ * \def U_LAYOUTEX_API
+ * Set to export library symbols from inside the layout extensions library,
+ * and to import them from outside.
+ * @stable ICU 2.6
+ */
+
+/**
+ * \def U_IO_API
+ * Set to export library symbols from inside the ustdio library,
+ * and to import them from outside.
+ * @stable ICU 2.0
+ */
+
+/**
+ * \def U_TOOLUTIL_API
+ * Set to export library symbols from inside the toolutil library,
+ * and to import them from outside.
+ * @stable ICU 3.4
+ */
+
+#ifdef U_IN_DOXYGEN
+// This definition is required when generating the API docs.
+#define U_COMBINED_IMPLEMENTATION 1
+#endif
+
+#if defined(U_COMBINED_IMPLEMENTATION)
+#define U_DATA_API     U_EXPORT
+#define U_COMMON_API   U_EXPORT
+#define U_I18N_API     U_EXPORT
+#define U_LAYOUT_API   U_EXPORT
+#define U_LAYOUTEX_API U_EXPORT
+#define U_IO_API       U_EXPORT
+#define U_TOOLUTIL_API U_EXPORT
+#elif defined(U_STATIC_IMPLEMENTATION)
+#define U_DATA_API
+#define U_COMMON_API
+#define U_I18N_API
+#define U_LAYOUT_API
+#define U_LAYOUTEX_API
+#define U_IO_API
+#define U_TOOLUTIL_API
+#elif defined(U_COMMON_IMPLEMENTATION)
+#define U_DATA_API     U_IMPORT
+#define U_COMMON_API   U_EXPORT
+#define U_I18N_API     U_IMPORT
+#define U_LAYOUT_API   U_IMPORT
+#define U_LAYOUTEX_API U_IMPORT
+#define U_IO_API       U_IMPORT
+#define U_TOOLUTIL_API U_IMPORT
+#elif defined(U_I18N_IMPLEMENTATION)
+#define U_DATA_API     U_IMPORT
+#define U_COMMON_API   U_IMPORT
+#define U_I18N_API     U_EXPORT
+#define U_LAYOUT_API   U_IMPORT
+#define U_LAYOUTEX_API U_IMPORT
+#define U_IO_API       U_IMPORT
+#define U_TOOLUTIL_API U_IMPORT
+#elif defined(U_LAYOUT_IMPLEMENTATION)
+#define U_DATA_API     U_IMPORT
+#define U_COMMON_API   U_IMPORT
+#define U_I18N_API     U_IMPORT
+#define U_LAYOUT_API   U_EXPORT
+#define U_LAYOUTEX_API U_IMPORT
+#define U_IO_API       U_IMPORT
+#define U_TOOLUTIL_API U_IMPORT
+#elif defined(U_LAYOUTEX_IMPLEMENTATION)
+#define U_DATA_API     U_IMPORT
+#define U_COMMON_API   U_IMPORT
+#define U_I18N_API     U_IMPORT
+#define U_LAYOUT_API   U_IMPORT
+#define U_LAYOUTEX_API U_EXPORT
+#define U_IO_API       U_IMPORT
+#define U_TOOLUTIL_API U_IMPORT
+#elif defined(U_IO_IMPLEMENTATION)
+#define U_DATA_API     U_IMPORT
+#define U_COMMON_API   U_IMPORT
+#define U_I18N_API     U_IMPORT
+#define U_LAYOUT_API   U_IMPORT
+#define U_LAYOUTEX_API U_IMPORT
+#define U_IO_API       U_EXPORT
+#define U_TOOLUTIL_API U_IMPORT
+#elif defined(U_TOOLUTIL_IMPLEMENTATION)
+#define U_DATA_API     U_IMPORT
+#define U_COMMON_API   U_IMPORT
+#define U_I18N_API     U_IMPORT
+#define U_LAYOUT_API   U_IMPORT
+#define U_LAYOUTEX_API U_IMPORT
+#define U_IO_API       U_IMPORT
+#define U_TOOLUTIL_API U_EXPORT
+#else
+#define U_DATA_API     U_IMPORT
+#define U_COMMON_API   U_IMPORT
+#define U_I18N_API     U_IMPORT
+#define U_LAYOUT_API   U_IMPORT
+#define U_LAYOUTEX_API U_IMPORT
+#define U_IO_API       U_IMPORT
+#define U_TOOLUTIL_API U_IMPORT
+#endif
+
+/**
+ * \def U_STANDARD_CPP_NAMESPACE
+ * Control of C++ Namespace
+ * @stable ICU 2.0
+ */
+#ifdef __cplusplus
+#define U_STANDARD_CPP_NAMESPACE        ::
+#else
+#define U_STANDARD_CPP_NAMESPACE
+#endif
+
+/*===========================================================================*/
+/* UErrorCode */
+/*===========================================================================*/
+
+/**
+ * Standard ICU4C error code type, a substitute for exceptions.
+ *
+ * Initialize the UErrorCode with U_ZERO_ERROR, and check for success or
+ * failure using U_SUCCESS() or U_FAILURE():
+ *
+ *     UErrorCode errorCode = U_ZERO_ERROR;
+ *     // call ICU API that needs an error code parameter.
+ *     if (U_FAILURE(errorCode)) {
+ *         // An error occurred. Handle it here.
+ *     }
+ *
+ * C++ code should use icu::ErrorCode, available in unicode/errorcode.h, or a
+ * suitable subclass.
+ *
+ * For more information, see:
+ * https://unicode-org.github.io/icu/userguide/dev/codingguidelines#details-about-icu-error-codes
+ *
+ * Note: By convention, ICU functions that take a reference (C++) or a pointer
+ * (C) to a UErrorCode first test:
+ *
+ *     if (U_FAILURE(errorCode)) { return immediately; }
+ *
+ * so that in a chain of such functions the first one that sets an error code
+ * causes the following ones to not perform any operations.
+ *
+ * @stable ICU 2.0
+ */
+typedef enum UErrorCode {
+    /* The ordering of U_ERROR_INFO_START Vs U_USING_FALLBACK_WARNING looks weird
+     * and is that way because VC++ debugger displays first encountered constant,
+     * which is not the what the code is used for
+     */
+
+    U_USING_FALLBACK_WARNING  = -128,   /**< A resource bundle lookup returned a fallback result (not an error) */
+
+    U_ERROR_WARNING_START     = -128,   /**< Start of information results (semantically successful) */
+
+    U_USING_DEFAULT_WARNING   = -127,   /**< A resource bundle lookup returned a result from the root locale (not an error) */
+
+    U_SAFECLONE_ALLOCATED_WARNING = -126, /**< A SafeClone operation required allocating memory (informational only) */
+
+    U_STATE_OLD_WARNING       = -125,   /**< ICU has to use compatibility layer to construct the service. Expect performance/memory usage degradation. Consider upgrading */
+
+    U_STRING_NOT_TERMINATED_WARNING = -124,/**< An output string could not be NUL-terminated because output length==destCapacity. */
+
+    U_SORT_KEY_TOO_SHORT_WARNING = -123, /**< Number of levels requested in getBound is higher than the number of levels in the sort key */
+
+    U_AMBIGUOUS_ALIAS_WARNING = -122,   /**< This converter alias can go to different converter implementations */
+
+    U_DIFFERENT_UCA_VERSION = -121,     /**< ucol_open encountered a mismatch between UCA version and collator image version, so the collator was constructed from rules. No impact to further function */
+    
+    U_PLUGIN_CHANGED_LEVEL_WARNING = -120, /**< A plugin caused a level change. May not be an error, but later plugins may not load. */
+
+#ifndef U_HIDE_DEPRECATED_API
+    /**
+     * One more than the highest normal UErrorCode warning value.
+     * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
+     */
+    U_ERROR_WARNING_LIMIT,
+#endif  // U_HIDE_DEPRECATED_API
+
+    U_ZERO_ERROR              =  0,     /**< No error, no warning. */
+
+    U_ILLEGAL_ARGUMENT_ERROR  =  1,     /**< Start of codes indicating failure */
+    U_MISSING_RESOURCE_ERROR  =  2,     /**< The requested resource cannot be found */
+    U_INVALID_FORMAT_ERROR    =  3,     /**< Data format is not what is expected */
+    U_FILE_ACCESS_ERROR       =  4,     /**< The requested file cannot be found */
+    U_INTERNAL_PROGRAM_ERROR  =  5,     /**< Indicates a bug in the library code */
+    U_MESSAGE_PARSE_ERROR     =  6,     /**< Unable to parse a message (message format) */
+    U_MEMORY_ALLOCATION_ERROR =  7,     /**< Memory allocation error */
+    U_INDEX_OUTOFBOUNDS_ERROR =  8,     /**< Trying to access the index that is out of bounds */
+    U_PARSE_ERROR             =  9,     /**< Equivalent to Java ParseException */
+    U_INVALID_CHAR_FOUND      = 10,     /**< Character conversion: Unmappable input sequence. In other APIs: Invalid character. */
+    U_TRUNCATED_CHAR_FOUND    = 11,     /**< Character conversion: Incomplete input sequence. */
+    U_ILLEGAL_CHAR_FOUND      = 12,     /**< Character conversion: Illegal input sequence/combination of input units. */
+    U_INVALID_TABLE_FORMAT    = 13,     /**< Conversion table file found, but corrupted */
+    U_INVALID_TABLE_FILE      = 14,     /**< Conversion table file not found */
+    U_BUFFER_OVERFLOW_ERROR   = 15,     /**< A result would not fit in the supplied buffer */
+    U_UNSUPPORTED_ERROR       = 16,     /**< Requested operation not supported in current context */
+    U_RESOURCE_TYPE_MISMATCH  = 17,     /**< an operation is requested over a resource that does not support it */
+    U_ILLEGAL_ESCAPE_SEQUENCE = 18,     /**< ISO-2022 illegal escape sequence */
+    U_UNSUPPORTED_ESCAPE_SEQUENCE = 19, /**< ISO-2022 unsupported escape sequence */
+    U_NO_SPACE_AVAILABLE      = 20,     /**< No space available for in-buffer expansion for Arabic shaping */
+    U_CE_NOT_FOUND_ERROR      = 21,     /**< Currently used only while setting variable top, but can be used generally */
+    U_PRIMARY_TOO_LONG_ERROR  = 22,     /**< User tried to set variable top to a primary that is longer than two bytes */
+    U_STATE_TOO_OLD_ERROR     = 23,     /**< ICU cannot construct a service from this state, as it is no longer supported */
+    U_TOO_MANY_ALIASES_ERROR  = 24,     /**< There are too many aliases in the path to the requested resource.
+                                             It is very possible that a circular alias definition has occurred */
+    U_ENUM_OUT_OF_SYNC_ERROR  = 25,     /**< UEnumeration out of sync with underlying collection */
+    U_INVARIANT_CONVERSION_ERROR = 26,  /**< Unable to convert a UChar* string to char* with the invariant converter. */
+    U_INVALID_STATE_ERROR     = 27,     /**< Requested operation can not be completed with ICU in its current state */
+    U_COLLATOR_VERSION_MISMATCH = 28,   /**< Collator version is not compatible with the base version */
+    U_USELESS_COLLATOR_ERROR  = 29,     /**< Collator is options only and no base is specified */
+    U_NO_WRITE_PERMISSION     = 30,     /**< Attempt to modify read-only or constant data. */
+    /**
+     * The input is impractically long for an operation.
+     * It is rejected because it may lead to problems such as excessive
+     * processing time, stack depth, or heap memory requirements.
+     *
+     * @stable ICU 68
+     */
+    U_INPUT_TOO_LONG_ERROR = 31,
+
+#ifndef U_HIDE_DEPRECATED_API
+    /**
+     * One more than the highest standard error code.
+     * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
+     */
+    U_STANDARD_ERROR_LIMIT = 32,
+#endif  // U_HIDE_DEPRECATED_API
+
+    /*
+     * Error codes in the range 0x10000 0x10100 are reserved for Transliterator.
+     */
+    U_BAD_VARIABLE_DEFINITION=0x10000,/**< Missing '$' or duplicate variable name */
+    U_PARSE_ERROR_START = 0x10000,    /**< Start of Transliterator errors */
+    U_MALFORMED_RULE,                 /**< Elements of a rule are misplaced */
+    U_MALFORMED_SET,                  /**< A UnicodeSet pattern is invalid*/
+    U_MALFORMED_SYMBOL_REFERENCE,     /**< UNUSED as of ICU 2.4 */
+    U_MALFORMED_UNICODE_ESCAPE,       /**< A Unicode escape pattern is invalid*/
+    U_MALFORMED_VARIABLE_DEFINITION,  /**< A variable definition is invalid */
+    U_MALFORMED_VARIABLE_REFERENCE,   /**< A variable reference is invalid */
+    U_MISMATCHED_SEGMENT_DELIMITERS,  /**< UNUSED as of ICU 2.4 */
+    U_MISPLACED_ANCHOR_START,         /**< A start anchor appears at an illegal position */
+    U_MISPLACED_CURSOR_OFFSET,        /**< A cursor offset occurs at an illegal position */
+    U_MISPLACED_QUANTIFIER,           /**< A quantifier appears after a segment close delimiter */
+    U_MISSING_OPERATOR,               /**< A rule contains no operator */
+    U_MISSING_SEGMENT_CLOSE,          /**< UNUSED as of ICU 2.4 */
+    U_MULTIPLE_ANTE_CONTEXTS,         /**< More than one ante context */
+    U_MULTIPLE_CURSORS,               /**< More than one cursor */
+    U_MULTIPLE_POST_CONTEXTS,         /**< More than one post context */
+    U_TRAILING_BACKSLASH,             /**< A dangling backslash */
+    U_UNDEFINED_SEGMENT_REFERENCE,    /**< A segment reference does not correspond to a defined segment */
+    U_UNDEFINED_VARIABLE,             /**< A variable reference does not correspond to a defined variable */
+    U_UNQUOTED_SPECIAL,               /**< A special character was not quoted or escaped */
+    U_UNTERMINATED_QUOTE,             /**< A closing single quote is missing */
+    U_RULE_MASK_ERROR,                /**< A rule is hidden by an earlier more general rule */
+    U_MISPLACED_COMPOUND_FILTER,      /**< A compound filter is in an invalid location */
+    U_MULTIPLE_COMPOUND_FILTERS,      /**< More than one compound filter */
+    U_INVALID_RBT_SYNTAX,             /**< A "::id" rule was passed to the RuleBasedTransliterator parser */
+    U_INVALID_PROPERTY_PATTERN,       /**< UNUSED as of ICU 2.4 */
+    U_MALFORMED_PRAGMA,               /**< A 'use' pragma is invalid */
+    U_UNCLOSED_SEGMENT,               /**< A closing ')' is missing */
+    U_ILLEGAL_CHAR_IN_SEGMENT,        /**< UNUSED as of ICU 2.4 */
+    U_VARIABLE_RANGE_EXHAUSTED,       /**< Too many stand-ins generated for the given variable range */
+    U_VARIABLE_RANGE_OVERLAP,         /**< The variable range overlaps characters used in rules */
+    U_ILLEGAL_CHARACTER,              /**< A special character is outside its allowed context */
+    U_INTERNAL_TRANSLITERATOR_ERROR,  /**< Internal transliterator system error */
+    U_INVALID_ID,                     /**< A "::id" rule specifies an unknown transliterator */
+    U_INVALID_FUNCTION,               /**< A "&fn()" rule specifies an unknown transliterator */
+#ifndef U_HIDE_DEPRECATED_API
+    /**
+     * One more than the highest normal Transliterator error code.
+     * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
+     */
+    U_PARSE_ERROR_LIMIT,
+#endif  // U_HIDE_DEPRECATED_API
+
+    /*
+     * Error codes in the range 0x10100 0x10200 are reserved for the formatting API.
+     */
+    U_UNEXPECTED_TOKEN=0x10100,       /**< Syntax error in format pattern */
+    U_FMT_PARSE_ERROR_START=0x10100,  /**< Start of format library errors */
+    U_MULTIPLE_DECIMAL_SEPARATORS,    /**< More than one decimal separator in number pattern */
+    U_MULTIPLE_DECIMAL_SEPERATORS = U_MULTIPLE_DECIMAL_SEPARATORS, /**< Typo: kept for backward compatibility. Use U_MULTIPLE_DECIMAL_SEPARATORS */
+    U_MULTIPLE_EXPONENTIAL_SYMBOLS,   /**< More than one exponent symbol in number pattern */
+    U_MALFORMED_EXPONENTIAL_PATTERN,  /**< Grouping symbol in exponent pattern */
+    U_MULTIPLE_PERCENT_SYMBOLS,       /**< More than one percent symbol in number pattern */
+    U_MULTIPLE_PERMILL_SYMBOLS,       /**< More than one permill symbol in number pattern */
+    U_MULTIPLE_PAD_SPECIFIERS,        /**< More than one pad symbol in number pattern */
+    U_PATTERN_SYNTAX_ERROR,           /**< Syntax error in format pattern */
+    U_ILLEGAL_PAD_POSITION,           /**< Pad symbol misplaced in number pattern */
+    U_UNMATCHED_BRACES,               /**< Braces do not match in message pattern */
+    U_UNSUPPORTED_PROPERTY,           /**< UNUSED as of ICU 2.4 */
+    U_UNSUPPORTED_ATTRIBUTE,          /**< UNUSED as of ICU 2.4 */
+    U_ARGUMENT_TYPE_MISMATCH,         /**< Argument name and argument index mismatch in MessageFormat functions */
+    U_DUPLICATE_KEYWORD,              /**< Duplicate keyword in PluralFormat */
+    U_UNDEFINED_KEYWORD,              /**< Undefined Plural keyword */
+    U_DEFAULT_KEYWORD_MISSING,        /**< Missing DEFAULT rule in plural rules */
+    U_DECIMAL_NUMBER_SYNTAX_ERROR,    /**< Decimal number syntax error */
+    U_FORMAT_INEXACT_ERROR,           /**< Cannot format a number exactly and rounding mode is ROUND_UNNECESSARY @stable ICU 4.8 */
+    U_NUMBER_ARG_OUTOFBOUNDS_ERROR,   /**< The argument to a NumberFormatter helper method was out of bounds; the bounds are usually 0 to 999. @stable ICU 61 */
+    U_NUMBER_SKELETON_SYNTAX_ERROR,   /**< The number skeleton passed to C++ NumberFormatter or C UNumberFormatter was invalid or contained a syntax error. @stable ICU 62 */
+#ifndef U_HIDE_DEPRECATED_API
+    /**
+     * One more than the highest normal formatting API error code.
+     * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
+     */
+    U_FMT_PARSE_ERROR_LIMIT = 0x10114,
+#endif  // U_HIDE_DEPRECATED_API
+
+    /*
+     * Error codes in the range 0x10200 0x102ff are reserved for BreakIterator.
+     */
+    U_BRK_INTERNAL_ERROR=0x10200,          /**< An internal error (bug) was detected.             */
+    U_BRK_ERROR_START=0x10200,             /**< Start of codes indicating Break Iterator failures */
+    U_BRK_HEX_DIGITS_EXPECTED,             /**< Hex digits expected as part of a escaped char in a rule. */
+    U_BRK_SEMICOLON_EXPECTED,              /**< Missing ';' at the end of a RBBI rule.            */
+    U_BRK_RULE_SYNTAX,                     /**< Syntax error in RBBI rule.                        */
+    U_BRK_UNCLOSED_SET,                    /**< UnicodeSet writing an RBBI rule missing a closing ']'. */
+    U_BRK_ASSIGN_ERROR,                    /**< Syntax error in RBBI rule assignment statement.   */
+    U_BRK_VARIABLE_REDFINITION,            /**< RBBI rule $Variable redefined.                    */
+    U_BRK_MISMATCHED_PAREN,                /**< Mis-matched parentheses in an RBBI rule.          */
+    U_BRK_NEW_LINE_IN_QUOTED_STRING,       /**< Missing closing quote in an RBBI rule.            */
+    U_BRK_UNDEFINED_VARIABLE,              /**< Use of an undefined $Variable in an RBBI rule.    */
+    U_BRK_INIT_ERROR,                      /**< Initialization failure.  Probable missing ICU Data. */
+    U_BRK_RULE_EMPTY_SET,                  /**< Rule contains an empty Unicode Set.               */
+    U_BRK_UNRECOGNIZED_OPTION,             /**< !!option in RBBI rules not recognized.            */
+    U_BRK_MALFORMED_RULE_TAG,              /**< The {nnn} tag on a rule is malformed              */
+#ifndef U_HIDE_DEPRECATED_API
+    /**
+     * One more than the highest normal BreakIterator error code.
+     * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
+     */
+    U_BRK_ERROR_LIMIT,
+#endif  // U_HIDE_DEPRECATED_API
+
+    /*
+     * Error codes in the range 0x10300-0x103ff are reserved for regular expression related errors.
+     */
+    U_REGEX_INTERNAL_ERROR=0x10300,       /**< An internal error (bug) was detected.              */
+    U_REGEX_ERROR_START=0x10300,          /**< Start of codes indicating Regexp failures          */
+    U_REGEX_RULE_SYNTAX,                  /**< Syntax error in regexp pattern.                    */
+    U_REGEX_INVALID_STATE,                /**< RegexMatcher in invalid state for requested operation */
+    U_REGEX_BAD_ESCAPE_SEQUENCE,          /**< Unrecognized backslash escape sequence in pattern  */
+    U_REGEX_PROPERTY_SYNTAX,              /**< Incorrect Unicode property                         */
+    U_REGEX_UNIMPLEMENTED,                /**< Use of regexp feature that is not yet implemented. */
+    U_REGEX_MISMATCHED_PAREN,             /**< Incorrectly nested parentheses in regexp pattern.  */
+    U_REGEX_NUMBER_TOO_BIG,               /**< Decimal number is too large.                       */
+    U_REGEX_BAD_INTERVAL,                 /**< Error in {min,max} interval                        */
+    U_REGEX_MAX_LT_MIN,                   /**< In {min,max}, max is less than min.                */
+    U_REGEX_INVALID_BACK_REF,             /**< Back-reference to a non-existent capture group.    */
+    U_REGEX_INVALID_FLAG,                 /**< Invalid value for match mode flags.                */
+    U_REGEX_LOOK_BEHIND_LIMIT,            /**< Look-Behind pattern matches must have a bounded maximum length.    */
+    U_REGEX_SET_CONTAINS_STRING,          /**< Regexps cannot have UnicodeSets containing strings.*/
+#ifndef U_HIDE_DEPRECATED_API
+    U_REGEX_OCTAL_TOO_BIG,                /**< Octal character constants must be <= 0377. @deprecated ICU 54. This error cannot occur. */
+#endif  /* U_HIDE_DEPRECATED_API */
+    U_REGEX_MISSING_CLOSE_BRACKET=U_REGEX_SET_CONTAINS_STRING+2, /**< Missing closing bracket on a bracket expression. */
+    U_REGEX_INVALID_RANGE,                /**< In a character range [x-y], x is greater than y.   */
+    U_REGEX_STACK_OVERFLOW,               /**< Regular expression backtrack stack overflow.       */
+    U_REGEX_TIME_OUT,                     /**< Maximum allowed match time exceeded                */
+    U_REGEX_STOPPED_BY_CALLER,            /**< Matching operation aborted by user callback fn.    */
+    U_REGEX_PATTERN_TOO_BIG,              /**< Pattern exceeds limits on size or complexity. @stable ICU 55 */
+    U_REGEX_INVALID_CAPTURE_GROUP_NAME,   /**< Invalid capture group name. @stable ICU 55 */
+#ifndef U_HIDE_DEPRECATED_API
+    /**
+     * One more than the highest normal regular expression error code.
+     * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
+     */
+    U_REGEX_ERROR_LIMIT=U_REGEX_STOPPED_BY_CALLER+3,
+#endif  // U_HIDE_DEPRECATED_API
+
+    /*
+     * Error codes in the range 0x10400-0x104ff are reserved for IDNA related error codes.
+     */
+    U_IDNA_PROHIBITED_ERROR=0x10400,
+    U_IDNA_ERROR_START=0x10400,
+    U_IDNA_UNASSIGNED_ERROR,
+    U_IDNA_CHECK_BIDI_ERROR,
+    U_IDNA_STD3_ASCII_RULES_ERROR,
+    U_IDNA_ACE_PREFIX_ERROR,
+    U_IDNA_VERIFICATION_ERROR,
+    U_IDNA_LABEL_TOO_LONG_ERROR,
+    U_IDNA_ZERO_LENGTH_LABEL_ERROR,
+    U_IDNA_DOMAIN_NAME_TOO_LONG_ERROR,
+#ifndef U_HIDE_DEPRECATED_API
+    /**
+     * One more than the highest normal IDNA error code.
+     * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
+     */
+    U_IDNA_ERROR_LIMIT,
+#endif  // U_HIDE_DEPRECATED_API
+    /*
+     * Aliases for StringPrep
+     */
+    U_STRINGPREP_PROHIBITED_ERROR = U_IDNA_PROHIBITED_ERROR,
+    U_STRINGPREP_UNASSIGNED_ERROR = U_IDNA_UNASSIGNED_ERROR,
+    U_STRINGPREP_CHECK_BIDI_ERROR = U_IDNA_CHECK_BIDI_ERROR,
+    
+    /*
+     * Error codes in the range 0x10500-0x105ff are reserved for Plugin related error codes.
+     */
+    U_PLUGIN_ERROR_START=0x10500,         /**< Start of codes indicating plugin failures */
+    U_PLUGIN_TOO_HIGH=0x10500,            /**< The plugin's level is too high to be loaded right now. */
+    U_PLUGIN_DIDNT_SET_LEVEL,             /**< The plugin didn't call uplug_setPlugLevel in response to a QUERY */
+#ifndef U_HIDE_DEPRECATED_API
+    /**
+     * One more than the highest normal plug-in error code.
+     * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
+     */
+    U_PLUGIN_ERROR_LIMIT,
+#endif  // U_HIDE_DEPRECATED_API
+
+#ifndef U_HIDE_DEPRECATED_API
+    /**
+     * One more than the highest normal error code.
+     * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
+     */
+    U_ERROR_LIMIT=U_PLUGIN_ERROR_LIMIT
+#endif  // U_HIDE_DEPRECATED_API
+} UErrorCode;
+
+/* Use the following to determine if an UErrorCode represents */
+/* operational success or failure. */
+
+#ifdef __cplusplus
+    /**
+     * Does the error code indicate success?
+     * @stable ICU 2.0
+     */
+    static
+    inline UBool U_SUCCESS(UErrorCode code) { return (UBool)(code<=U_ZERO_ERROR); }
+    /**
+     * Does the error code indicate a failure?
+     * @stable ICU 2.0
+     */
+    static
+    inline UBool U_FAILURE(UErrorCode code) { return (UBool)(code>U_ZERO_ERROR); }
+#else
+    /**
+     * Does the error code indicate success?
+     * @stable ICU 2.0
+     */
+#   define U_SUCCESS(x) ((x)<=U_ZERO_ERROR)
+    /**
+     * Does the error code indicate a failure?
+     * @stable ICU 2.0
+     */
+#   define U_FAILURE(x) ((x)>U_ZERO_ERROR)
+#endif
+
+/**
+ * Return a string for a UErrorCode value.
+ * The string will be the same as the name of the error code constant
+ * in the UErrorCode enum above.
+ * @stable ICU 2.0
+ */
+U_CAPI const char * U_EXPORT2
+u_errorName(UErrorCode code);
+
+
+#endif /* _UTYPES */
diff --git a/intl/icu/source/common/unicode/uvernum.h b/intl/icu/source/common/unicode/uvernum.h
new file mode 100644
index 0000000000..f0fc671b4b
--- /dev/null
+++ b/intl/icu/source/common/unicode/uvernum.h
@@ -0,0 +1,191 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*   Copyright (C) 2000-2016, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*******************************************************************************
+*
+*   file name:  uvernum.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   Created by: Vladimir Weinstein
+*   Updated by: Steven R. Loomis
+*
+*/
+
+/**
+ * \file
+ * \brief C API: definitions of ICU version numbers
+ *
+ * This file is included by uversion.h and other files. This file contains only
+ * macros and definitions. The actual version numbers are defined here.
+ */
+
+ /*
+  * IMPORTANT: When updating version, the following things need to be done:
+  * source/common/unicode/uvernum.h - this file: update major, minor,
+  *        patchlevel, suffix, version, short version constants, namespace,
+  *                    renaming macro, and copyright
+  *
+  * The following files need to be updated as well, which can be done
+  *  by running the UNIX makefile target 'update-windows-makefiles' in icu4c/source.
+  *
+  * source/allinone/Build.Windows.IcuVersion.props - Update the IcuMajorVersion
+  * source/data/makedata.mak - change U_ICUDATA_NAME so that it contains
+  *                            the new major/minor combination, and UNICODE_VERSION
+  *                            for the Unicode version.
+  */
+
+#ifndef UVERNUM_H
+#define UVERNUM_H
+
+/** The standard copyright notice that gets compiled into each library.
+ *  This value will change in the subsequent releases of ICU
+ *  @stable ICU 2.4
+ */
+#define U_COPYRIGHT_STRING \
+  " Copyright (C) 2016 and later: Unicode, Inc. and others. License & terms of use: http://www.unicode.org/copyright.html "
+
+/** The current ICU major version as an integer.
+ *  This value will change in the subsequent releases of ICU
+ *  @stable ICU 2.4
+ */
+#define U_ICU_VERSION_MAJOR_NUM 73
+
+/** The current ICU minor version as an integer.
+ *  This value will change in the subsequent releases of ICU
+ *  @stable ICU 2.6
+ */
+#define U_ICU_VERSION_MINOR_NUM 1
+
+/** The current ICU patchlevel version as an integer.
+ *  This value will change in the subsequent releases of ICU
+ *  @stable ICU 2.4
+ */
+#define U_ICU_VERSION_PATCHLEVEL_NUM 0
+
+/** The current ICU build level version as an integer.
+ *  This value is for use by ICU clients. It defaults to 0.
+ *  @stable ICU 4.0
+ */
+#ifndef U_ICU_VERSION_BUILDLEVEL_NUM
+#define U_ICU_VERSION_BUILDLEVEL_NUM 0
+#endif
+
+/** Glued version suffix for renamers
+ *  This value will change in the subsequent releases of ICU
+ *  @stable ICU 2.6
+ */
+#define U_ICU_VERSION_SUFFIX _73
+
+/**
+ * \def U_DEF2_ICU_ENTRY_POINT_RENAME
+ * @internal
+ */
+/**
+ * \def U_DEF_ICU_ENTRY_POINT_RENAME
+ * @internal
+ */
+/** Glued version suffix function for renamers
+ *  This value will change in the subsequent releases of ICU.
+ *  If a custom suffix (such as matching library suffixes) is desired, this can be modified.
+ *  Note that if present, platform.h may contain an earlier definition of this macro.
+ *  \def U_ICU_ENTRY_POINT_RENAME
+ *  @stable ICU 4.2
+ */
+/**
+ * Disable the version suffix. Use the custom suffix if exists.
+ * \def U_DISABLE_VERSION_SUFFIX
+ * @internal
+ */
+#ifndef U_DISABLE_VERSION_SUFFIX
+#define U_DISABLE_VERSION_SUFFIX 0
+#endif
+
+#ifndef U_ICU_ENTRY_POINT_RENAME
+#ifdef U_HAVE_LIB_SUFFIX
+#   if !U_DISABLE_VERSION_SUFFIX
+#       define U_DEF_ICU_ENTRY_POINT_RENAME(x,y,z) x ## y ##  z
+#       define U_DEF2_ICU_ENTRY_POINT_RENAME(x,y,z) U_DEF_ICU_ENTRY_POINT_RENAME(x,y,z)
+#       define U_ICU_ENTRY_POINT_RENAME(x)    U_DEF2_ICU_ENTRY_POINT_RENAME(x,U_ICU_VERSION_SUFFIX,U_LIB_SUFFIX_C_NAME)
+#   else
+#       define U_DEF_ICU_ENTRY_POINT_RENAME(x,y) x ## y
+#       define U_DEF2_ICU_ENTRY_POINT_RENAME(x,y) U_DEF_ICU_ENTRY_POINT_RENAME(x,y)
+#       define U_ICU_ENTRY_POINT_RENAME(x)    U_DEF2_ICU_ENTRY_POINT_RENAME(x,U_LIB_SUFFIX_C_NAME)
+#   endif
+#else
+#   if !U_DISABLE_VERSION_SUFFIX
+#       define U_DEF_ICU_ENTRY_POINT_RENAME(x,y) x ## y
+#       define U_DEF2_ICU_ENTRY_POINT_RENAME(x,y) U_DEF_ICU_ENTRY_POINT_RENAME(x,y)
+#       define U_ICU_ENTRY_POINT_RENAME(x)    U_DEF2_ICU_ENTRY_POINT_RENAME(x,U_ICU_VERSION_SUFFIX)
+#   else
+#       define U_ICU_ENTRY_POINT_RENAME(x)    x
+#   endif
+#endif
+#endif
+
+/** The current ICU library version as a dotted-decimal string. The patchlevel
+ *  only appears in this string if it non-zero.
+ *  This value will change in the subsequent releases of ICU
+ *  @stable ICU 2.4
+ */
+#define U_ICU_VERSION "73.1"
+
+/**
+ * The current ICU library major version number as a string, for library name suffixes.
+ * This value will change in subsequent releases of ICU.
+ *
+ * Until ICU 4.8, this was the combination of the single-digit major and minor ICU version numbers
+ * into one string without dots ("48").
+ * Since ICU 49, it is the double-digit major ICU version number.
+ * See https://unicode-org.github.io/icu/userguide/design#version-numbers-in-icu
+ *
+ * @stable ICU 2.6
+ */
+#define U_ICU_VERSION_SHORT "73"
+
+#ifndef U_HIDE_INTERNAL_API
+/** Data version in ICU4C.
+ * @internal ICU 4.4 Internal Use Only
+ **/
+#define U_ICU_DATA_VERSION "73.1"
+#endif  /* U_HIDE_INTERNAL_API */
+
+/*===========================================================================
+ * ICU collation framework version information
+ * Version info that can be obtained from a collator is affected by these
+ * numbers in a secret and magic way. Please use collator version as whole
+ *===========================================================================
+ */
+
+/**
+ * Collation runtime version (sort key generator, strcoll).
+ * If the version is different, sort keys for the same string could be different.
+ * This value may change in subsequent releases of ICU.
+ * @stable ICU 2.4
+ */
+#define UCOL_RUNTIME_VERSION 9
+
+/**
+ * Collation builder code version.
+ * When this is different, the same tailoring might result
+ * in assigning different collation elements to code points.
+ * This value may change in subsequent releases of ICU.
+ * @stable ICU 2.4
+ */
+#define UCOL_BUILDER_VERSION 9
+
+#ifndef U_HIDE_DEPRECATED_API
+/**
+ * Constant 1.
+ * This was intended to be the version of collation tailorings,
+ * but instead the tailoring data carries a version number.
+ * @deprecated ICU 54
+ */
+#define UCOL_TAILORINGS_VERSION 1
+#endif  /* U_HIDE_DEPRECATED_API */
+
+#endif
diff --git a/intl/icu/source/common/unicode/uversion.h b/intl/icu/source/common/unicode/uversion.h
new file mode 100644
index 0000000000..113568df8c
--- /dev/null
+++ b/intl/icu/source/common/unicode/uversion.h
@@ -0,0 +1,187 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*   Copyright (C) 2000-2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*******************************************************************************
+*
+*   file name:  uversion.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   Created by: Vladimir Weinstein
+*
+*  Gets included by utypes.h and Windows .rc files
+*/
+
+/**
+ * \file
+ * \brief C API: API for accessing ICU version numbers. 
+ */
+/*===========================================================================*/
+/* Main ICU version information                                              */
+/*===========================================================================*/
+
+#ifndef UVERSION_H
+#define UVERSION_H
+
+#include "unicode/umachine.h"
+
+/* Actual version info lives in uvernum.h */
+#include "unicode/uvernum.h"
+
+/** Maximum length of the copyright string.
+ *  @stable ICU 2.4
+ */
+#define U_COPYRIGHT_STRING_LENGTH  128
+
+/** An ICU version consists of up to 4 numbers from 0..255.
+ *  @stable ICU 2.4
+ */
+#define U_MAX_VERSION_LENGTH 4
+
+/** In a string, ICU version fields are delimited by dots.
+ *  @stable ICU 2.4
+ */
+#define U_VERSION_DELIMITER '.'
+
+/** The maximum length of an ICU version string.
+ *  @stable ICU 2.4
+ */
+#define U_MAX_VERSION_STRING_LENGTH 20
+
+/** The binary form of a version on ICU APIs is an array of 4 uint8_t.
+ *  To compare two versions, use memcmp(v1,v2,sizeof(UVersionInfo)).
+ *  @stable ICU 2.4
+ */
+typedef uint8_t UVersionInfo[U_MAX_VERSION_LENGTH];
+
+/*===========================================================================*/
+/* C++ namespace if supported. Versioned unless versioning is disabled.      */
+/*===========================================================================*/
+
+/* Define C++ namespace symbols. */
+#ifdef __cplusplus
+
+/**
+ * \def U_NAMESPACE_BEGIN
+ * This is used to begin a declaration of a public ICU C++ API within
+ * versioned-ICU-namespace block.
+ *
+ * @stable ICU 2.4
+ */
+
+/**
+ * \def U_NAMESPACE_END
+ * This is used to end a declaration of a public ICU C++ API.
+ * It ends the versioned-ICU-namespace block begun by U_NAMESPACE_BEGIN.
+ *
+ * @stable ICU 2.4
+ */
+
+/**
+ * \def U_NAMESPACE_USE
+ * This is used to specify that the rest of the code uses the
+ * public ICU C++ API namespace.
+ * @stable ICU 2.4
+ */
+
+/**
+ * \def U_NAMESPACE_QUALIFIER
+ * This is used to qualify that a function or class is part of
+ * the public ICU C++ API namespace.
+ *
+ * This macro is unnecessary since ICU 49 requires namespace support.
+ * You can just use "icu::" instead.
+ * @stable ICU 2.4
+ */
+
+#   if U_DISABLE_RENAMING
+#       define U_ICU_NAMESPACE icu
+        namespace U_ICU_NAMESPACE { }
+#   else
+#       define U_ICU_NAMESPACE U_ICU_ENTRY_POINT_RENAME(icu)
+        namespace U_ICU_NAMESPACE { }
+        namespace icu = U_ICU_NAMESPACE;
+#   endif
+
+#   define U_NAMESPACE_BEGIN namespace U_ICU_NAMESPACE {
+#   define U_NAMESPACE_END }
+#   define U_NAMESPACE_USE using namespace U_ICU_NAMESPACE;
+#   define U_NAMESPACE_QUALIFIER U_ICU_NAMESPACE::
+
+#   ifndef U_USING_ICU_NAMESPACE
+#       if defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || \
+                defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION) || \
+                defined(U_LAYOUTEX_IMPLEMENTATION) || defined(U_TOOLUTIL_IMPLEMENTATION)
+#           define U_USING_ICU_NAMESPACE 0
+#       else
+#           define U_USING_ICU_NAMESPACE 0
+#       endif
+#   endif
+#   if U_USING_ICU_NAMESPACE
+        U_NAMESPACE_USE
+#   endif
+#endif /* __cplusplus */
+
+/*===========================================================================*/
+/* General version helper functions. Definitions in putil.c                  */
+/*===========================================================================*/
+
+/**
+ * Parse a string with dotted-decimal version information and
+ * fill in a UVersionInfo structure with the result.
+ * Definition of this function lives in putil.c
+ *
+ * @param versionArray The destination structure for the version information.
+ * @param versionString A string with dotted-decimal version information,
+ *                      with up to four non-negative number fields with
+ *                      values of up to 255 each.
+ * @stable ICU 2.4
+ */
+U_CAPI void U_EXPORT2
+u_versionFromString(UVersionInfo versionArray, const char *versionString);
+
+/**
+ * Parse a Unicode string with dotted-decimal version information and
+ * fill in a UVersionInfo structure with the result.
+ * Definition of this function lives in putil.c
+ *
+ * @param versionArray The destination structure for the version information.
+ * @param versionString A Unicode string with dotted-decimal version
+ *                      information, with up to four non-negative number
+ *                      fields with values of up to 255 each.
+ * @stable ICU 4.2
+ */
+U_CAPI void U_EXPORT2
+u_versionFromUString(UVersionInfo versionArray, const UChar *versionString);
+
+
+/**
+ * Write a string with dotted-decimal version information according
+ * to the input UVersionInfo.
+ * Definition of this function lives in putil.c
+ *
+ * @param versionArray The version information to be written as a string.
+ * @param versionString A string buffer that will be filled in with
+ *                      a string corresponding to the numeric version
+ *                      information in versionArray.
+ *                      The buffer size must be at least U_MAX_VERSION_STRING_LENGTH.
+ * @stable ICU 2.4
+ */
+U_CAPI void U_EXPORT2
+u_versionToString(const UVersionInfo versionArray, char *versionString);
+
+/**
+ * Gets the ICU release version.  The version array stores the version information
+ * for ICU.  For example, release "1.3.31.2" is then represented as 0x01031F02.
+ * Definition of this function lives in putil.c
+ *
+ * @param versionArray the version # information, the result will be filled in
+ * @stable ICU 2.0
+ */
+U_CAPI void U_EXPORT2
+u_getVersion(UVersionInfo versionArray);
+#endif
diff --git a/intl/icu/source/common/unifiedcache.cpp b/intl/icu/source/common/unifiedcache.cpp
new file mode 100644
index 0000000000..6004111d6a
--- /dev/null
+++ b/intl/icu/source/common/unifiedcache.cpp
@@ -0,0 +1,554 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+* Copyright (C) 2015, International Business Machines Corporation and
+* others. All Rights Reserved.
+******************************************************************************
+*
+* File unifiedcache.cpp
+******************************************************************************
+*/
+
+#include "unifiedcache.h"
+
+#include       // For std::max()
+#ifndef __wasi__
+ #include 
+#endif
+
+#include "uassert.h"
+#include "uhash.h"
+#include "ucln_cmn.h"
+
+static icu::UnifiedCache *gCache = nullptr;
+#ifndef __wasi__
+static std::mutex *gCacheMutex = nullptr;
+static std::condition_variable *gInProgressValueAddedCond;
+#endif
+static icu::UInitOnce gCacheInitOnce {};
+
+static const int32_t MAX_EVICT_ITERATIONS = 10;
+static const int32_t DEFAULT_MAX_UNUSED = 1000;
+static const int32_t DEFAULT_PERCENTAGE_OF_IN_USE = 100;
+
+
+U_CDECL_BEGIN
+static UBool U_CALLCONV unifiedcache_cleanup() {
+    gCacheInitOnce.reset();
+    delete gCache;
+    gCache = nullptr;
+#ifndef __wasi__
+    gCacheMutex->~mutex();
+    gCacheMutex = nullptr;
+    gInProgressValueAddedCond->~condition_variable();
+    gInProgressValueAddedCond = nullptr;
+#endif
+    return true;
+}
+U_CDECL_END
+
+
+U_NAMESPACE_BEGIN
+
+int32_t U_EXPORT2
+ucache_hashKeys(const UHashTok key) {
+    const CacheKeyBase *ckey = (const CacheKeyBase *) key.pointer;
+    return ckey->hashCode();
+}
+
+UBool U_EXPORT2
+ucache_compareKeys(const UHashTok key1, const UHashTok key2) {
+    const CacheKeyBase *p1 = (const CacheKeyBase *) key1.pointer;
+    const CacheKeyBase *p2 = (const CacheKeyBase *) key2.pointer;
+    return *p1 == *p2;
+}
+
+void U_EXPORT2
+ucache_deleteKey(void *obj) {
+    CacheKeyBase *p = (CacheKeyBase *) obj;
+    delete p;
+}
+
+CacheKeyBase::~CacheKeyBase() {
+}
+
+static void U_CALLCONV cacheInit(UErrorCode &status) {
+    U_ASSERT(gCache == nullptr);
+    ucln_common_registerCleanup(
+            UCLN_COMMON_UNIFIED_CACHE, unifiedcache_cleanup);
+
+#ifndef __wasi__
+    gCacheMutex = STATIC_NEW(std::mutex);
+    gInProgressValueAddedCond = STATIC_NEW(std::condition_variable);
+#endif
+    gCache = new UnifiedCache(status);
+    if (gCache == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+    }
+    if (U_FAILURE(status)) {
+        delete gCache;
+        gCache = nullptr;
+        return;
+    }
+}
+
+UnifiedCache *UnifiedCache::getInstance(UErrorCode &status) {
+    umtx_initOnce(gCacheInitOnce, &cacheInit, status);
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+    U_ASSERT(gCache != nullptr);
+    return gCache;
+}
+
+UnifiedCache::UnifiedCache(UErrorCode &status) :
+        fHashtable(nullptr),
+        fEvictPos(UHASH_FIRST),
+        fNumValuesTotal(0),
+        fNumValuesInUse(0),
+        fMaxUnused(DEFAULT_MAX_UNUSED),
+        fMaxPercentageOfInUse(DEFAULT_PERCENTAGE_OF_IN_USE),
+        fAutoEvictedCount(0),
+        fNoValue(nullptr) {
+    if (U_FAILURE(status)) {
+        return;
+    }
+    fNoValue = new SharedObject();
+    if (fNoValue == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return;
+    }
+    fNoValue->softRefCount = 1;  // Add fake references to prevent fNoValue from being deleted
+    fNoValue->hardRefCount = 1;  // when other references to it are removed.
+    fNoValue->cachePtr = this;
+
+    fHashtable = uhash_open(
+            &ucache_hashKeys,
+            &ucache_compareKeys,
+            nullptr,
+            &status);
+    if (U_FAILURE(status)) {
+        return;
+    }
+    uhash_setKeyDeleter(fHashtable, &ucache_deleteKey);
+}
+
+void UnifiedCache::setEvictionPolicy(
+        int32_t count, int32_t percentageOfInUseItems, UErrorCode &status) {
+    if (U_FAILURE(status)) {
+        return;
+    }
+    if (count < 0 || percentageOfInUseItems < 0) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+#ifndef __wasi__
+    std::lock_guard lock(*gCacheMutex);
+#endif
+    fMaxUnused = count;
+    fMaxPercentageOfInUse = percentageOfInUseItems;
+}
+
+int32_t UnifiedCache::unusedCount() const {
+#ifndef __wasi__
+    std::lock_guard lock(*gCacheMutex);
+#endif
+    return uhash_count(fHashtable) - fNumValuesInUse;
+}
+
+int64_t UnifiedCache::autoEvictedCount() const {
+#ifndef __wasi__
+    std::lock_guard lock(*gCacheMutex);
+#endif
+    return fAutoEvictedCount;
+}
+
+int32_t UnifiedCache::keyCount() const {
+#ifndef __wasi__
+    std::lock_guard lock(*gCacheMutex);
+#endif
+    return uhash_count(fHashtable);
+}
+
+void UnifiedCache::flush() const {
+#ifndef __wasi__
+    std::lock_guard lock(*gCacheMutex);
+#endif
+
+    // Use a loop in case cache items that are flushed held hard references to
+    // other cache items making those additional cache items eligible for
+    // flushing.
+    while (_flush(false));
+}
+
+void UnifiedCache::handleUnreferencedObject() const {
+#ifndef __wasi__
+    std::lock_guard lock(*gCacheMutex);
+#endif
+    --fNumValuesInUse;
+    _runEvictionSlice();
+}
+
+#ifdef UNIFIED_CACHE_DEBUG
+#include 
+
+void UnifiedCache::dump() {
+    UErrorCode status = U_ZERO_ERROR;
+    const UnifiedCache *cache = getInstance(status);
+    if (U_FAILURE(status)) {
+        fprintf(stderr, "Unified Cache: Error fetching cache.\n");
+        return;
+    }
+    cache->dumpContents();
+}
+
+void UnifiedCache::dumpContents() const {
+#ifndef __wasi__
+    std::lock_guard lock(*gCacheMutex);
+#endif
+    _dumpContents();
+}
+
+// Dumps content of cache.
+// On entry, gCacheMutex must be held.
+// On exit, cache contents dumped to stderr.
+void UnifiedCache::_dumpContents() const {
+    int32_t pos = UHASH_FIRST;
+    const UHashElement *element = uhash_nextElement(fHashtable, &pos);
+    char buffer[256];
+    int32_t cnt = 0;
+    for (; element != nullptr; element = uhash_nextElement(fHashtable, &pos)) {
+        const SharedObject *sharedObject =
+                (const SharedObject *) element->value.pointer;
+        const CacheKeyBase *key =
+                (const CacheKeyBase *) element->key.pointer;
+        if (sharedObject->hasHardReferences()) {
+            ++cnt;
+            fprintf(
+                    stderr,
+                    "Unified Cache: Key '%s', error %d, value %p, total refcount %d, soft refcount %d\n",
+                    key->writeDescription(buffer, 256),
+                    key->creationStatus,
+                    sharedObject == fNoValue ? nullptr :sharedObject,
+                    sharedObject->getRefCount(),
+                    sharedObject->getSoftRefCount());
+        }
+    }
+    fprintf(stderr, "Unified Cache: %d out of a total of %d still have hard references\n", cnt, uhash_count(fHashtable));
+}
+#endif
+
+UnifiedCache::~UnifiedCache() {
+    // Try our best to clean up first.
+    flush();
+    {
+        // Now all that should be left in the cache are entries that refer to
+        // each other and entries with hard references from outside the cache.
+        // Nothing we can do about these so proceed to wipe out the cache.
+#ifndef __wasi__
+        std::lock_guard lock(*gCacheMutex);
+#endif
+        _flush(true);
+    }
+    uhash_close(fHashtable);
+    fHashtable = nullptr;
+    delete fNoValue;
+    fNoValue = nullptr;
+}
+
+const UHashElement *
+UnifiedCache::_nextElement() const {
+    const UHashElement *element = uhash_nextElement(fHashtable, &fEvictPos);
+    if (element == nullptr) {
+        fEvictPos = UHASH_FIRST;
+        return uhash_nextElement(fHashtable, &fEvictPos);
+    }
+    return element;
+}
+
+UBool UnifiedCache::_flush(UBool all) const {
+    UBool result = false;
+    int32_t origSize = uhash_count(fHashtable);
+    for (int32_t i = 0; i < origSize; ++i) {
+        const UHashElement *element = _nextElement();
+        if (element == nullptr) {
+            break;
+        }
+        if (all || _isEvictable(element)) {
+            const SharedObject *sharedObject =
+                    (const SharedObject *) element->value.pointer;
+            U_ASSERT(sharedObject->cachePtr == this);
+            uhash_removeElement(fHashtable, element);
+            removeSoftRef(sharedObject);    // Deletes the sharedObject when softRefCount goes to zero.
+            result = true;
+        }
+    }
+    return result;
+}
+
+int32_t UnifiedCache::_computeCountOfItemsToEvict() const {
+    int32_t totalItems = uhash_count(fHashtable);
+    int32_t evictableItems = totalItems - fNumValuesInUse;
+
+    int32_t unusedLimitByPercentage = fNumValuesInUse * fMaxPercentageOfInUse / 100;
+    int32_t unusedLimit = std::max(unusedLimitByPercentage, fMaxUnused);
+    int32_t countOfItemsToEvict = std::max(0, evictableItems - unusedLimit);
+    return countOfItemsToEvict;
+}
+
+void UnifiedCache::_runEvictionSlice() const {
+    int32_t maxItemsToEvict = _computeCountOfItemsToEvict();
+    if (maxItemsToEvict <= 0) {
+        return;
+    }
+    for (int32_t i = 0; i < MAX_EVICT_ITERATIONS; ++i) {
+        const UHashElement *element = _nextElement();
+        if (element == nullptr) {
+            break;
+        }
+        if (_isEvictable(element)) {
+            const SharedObject *sharedObject =
+                    (const SharedObject *) element->value.pointer;
+            uhash_removeElement(fHashtable, element);
+            removeSoftRef(sharedObject);   // Deletes sharedObject when SoftRefCount goes to zero.
+            ++fAutoEvictedCount;
+            if (--maxItemsToEvict == 0) {
+                break;
+            }
+        }
+    }
+}
+
+void UnifiedCache::_putNew(
+        const CacheKeyBase &key,
+        const SharedObject *value,
+        const UErrorCode creationStatus,
+        UErrorCode &status) const {
+    if (U_FAILURE(status)) {
+        return;
+    }
+    CacheKeyBase *keyToAdopt = key.clone();
+    if (keyToAdopt == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return;
+    }
+    keyToAdopt->fCreationStatus = creationStatus;
+    if (value->softRefCount == 0) {
+        _registerPrimary(keyToAdopt, value);
+    }
+    void *oldValue = uhash_put(fHashtable, keyToAdopt, (void *) value, &status);
+    U_ASSERT(oldValue == nullptr);
+    (void)oldValue;
+    if (U_SUCCESS(status)) {
+        value->softRefCount++;
+    }
+}
+
+void UnifiedCache::_putIfAbsentAndGet(
+        const CacheKeyBase &key,
+        const SharedObject *&value,
+        UErrorCode &status) const {
+#ifndef __wasi__
+    std::lock_guard lock(*gCacheMutex);
+#endif
+    const UHashElement *element = uhash_find(fHashtable, &key);
+    if (element != nullptr && !_inProgress(element)) {
+        _fetch(element, value, status);
+        return;
+    }
+    if (element == nullptr) {
+        UErrorCode putError = U_ZERO_ERROR;
+        // best-effort basis only.
+        _putNew(key, value, status, putError);
+    } else {
+        _put(element, value, status);
+    }
+    // Run an eviction slice. This will run even if we added a primary entry
+    // which doesn't increase the unused count, but that is still o.k
+    _runEvictionSlice();
+}
+
+
+UBool UnifiedCache::_poll(
+        const CacheKeyBase &key,
+        const SharedObject *&value,
+        UErrorCode &status) const {
+    U_ASSERT(value == nullptr);
+    U_ASSERT(status == U_ZERO_ERROR);
+#ifndef __wasi__
+    std::unique_lock lock(*gCacheMutex);
+#endif
+    const UHashElement *element = uhash_find(fHashtable, &key);
+
+    // If the hash table contains an inProgress placeholder entry for this key,
+    // this means that another thread is currently constructing the value object.
+    // Loop, waiting for that construction to complete.
+     while (element != nullptr && _inProgress(element)) {
+#ifndef __wasi__
+         gInProgressValueAddedCond->wait(lock);
+#endif
+         element = uhash_find(fHashtable, &key);
+    }
+
+    // If the hash table contains an entry for the key,
+    // fetch out the contents and return them.
+    if (element != nullptr) {
+         _fetch(element, value, status);
+        return true;
+    }
+
+    // The hash table contained nothing for this key.
+    // Insert an inProgress place holder value.
+    // Our caller will create the final value and update the hash table.
+    _putNew(key, fNoValue, U_ZERO_ERROR, status);
+    return false;
+}
+
+void UnifiedCache::_get(
+        const CacheKeyBase &key,
+        const SharedObject *&value,
+        const void *creationContext,
+        UErrorCode &status) const {
+    U_ASSERT(value == nullptr);
+    U_ASSERT(status == U_ZERO_ERROR);
+    if (_poll(key, value, status)) {
+        if (value == fNoValue) {
+            SharedObject::clearPtr(value);
+        }
+        return;
+    }
+    if (U_FAILURE(status)) {
+        return;
+    }
+    value = key.createObject(creationContext, status);
+    U_ASSERT(value == nullptr || value->hasHardReferences());
+    U_ASSERT(value != nullptr || status != U_ZERO_ERROR);
+    if (value == nullptr) {
+        SharedObject::copyPtr(fNoValue, value);
+    }
+    _putIfAbsentAndGet(key, value, status);
+    if (value == fNoValue) {
+        SharedObject::clearPtr(value);
+    }
+}
+
+void UnifiedCache::_registerPrimary(
+            const CacheKeyBase *theKey, const SharedObject *value) const {
+    theKey->fIsPrimary = true;
+    value->cachePtr = this;
+    ++fNumValuesTotal;
+    ++fNumValuesInUse;
+}
+
+void UnifiedCache::_put(
+        const UHashElement *element,
+        const SharedObject *value,
+        const UErrorCode status) const {
+    U_ASSERT(_inProgress(element));
+    const CacheKeyBase *theKey = (const CacheKeyBase *) element->key.pointer;
+    const SharedObject *oldValue = (const SharedObject *) element->value.pointer;
+    theKey->fCreationStatus = status;
+    if (value->softRefCount == 0) {
+        _registerPrimary(theKey, value);
+    }
+    value->softRefCount++;
+    UHashElement *ptr = const_cast(element);
+    ptr->value.pointer = (void *) value;
+    U_ASSERT(oldValue == fNoValue);
+    removeSoftRef(oldValue);
+
+#ifndef __wasi__
+    // Tell waiting threads that we replace in-progress status with
+    // an error.
+    gInProgressValueAddedCond->notify_all();
+#endif
+}
+
+void UnifiedCache::_fetch(
+        const UHashElement *element,
+        const SharedObject *&value,
+        UErrorCode &status) const {
+    const CacheKeyBase *theKey = (const CacheKeyBase *) element->key.pointer;
+    status = theKey->fCreationStatus;
+
+    // Since we have the cache lock, calling regular SharedObject add/removeRef
+    // could cause us to deadlock on ourselves since they may need to lock
+    // the cache mutex.
+    removeHardRef(value);
+    value = static_cast(element->value.pointer);
+    addHardRef(value);
+}
+
+
+UBool UnifiedCache::_inProgress(const UHashElement* element) const {
+    UErrorCode status = U_ZERO_ERROR;
+    const SharedObject * value = nullptr;
+    _fetch(element, value, status);
+    UBool result = _inProgress(value, status);
+    removeHardRef(value);
+    return result;
+}
+
+UBool UnifiedCache::_inProgress(
+        const SharedObject* theValue, UErrorCode creationStatus) const {
+    return (theValue == fNoValue && creationStatus == U_ZERO_ERROR);
+}
+
+UBool UnifiedCache::_isEvictable(const UHashElement *element) const
+{
+    const CacheKeyBase *theKey = (const CacheKeyBase *) element->key.pointer;
+    const SharedObject *theValue =
+            (const SharedObject *) element->value.pointer;
+
+    // Entries that are under construction are never evictable
+    if (_inProgress(theValue, theKey->fCreationStatus)) {
+        return false;
+    }
+
+    // We can evict entries that are either not a primary or have just
+    // one reference (The one reference being from the cache itself).
+    return (!theKey->fIsPrimary || (theValue->softRefCount == 1 && theValue->noHardReferences()));
+}
+
+void UnifiedCache::removeSoftRef(const SharedObject *value) const {
+    U_ASSERT(value->cachePtr == this);
+    U_ASSERT(value->softRefCount > 0);
+    if (--value->softRefCount == 0) {
+        --fNumValuesTotal;
+        if (value->noHardReferences()) {
+            delete value;
+        } else {
+            // This path only happens from flush(all). Which only happens from the
+            // UnifiedCache destructor.  Nulling out value.cacheptr changes the behavior
+            // of value.removeRef(), causing the deletion to be done there.
+            value->cachePtr = nullptr;
+        }
+    }
+}
+
+int32_t UnifiedCache::removeHardRef(const SharedObject *value) const {
+    int refCount = 0;
+    if (value) {
+        refCount = umtx_atomic_dec(&value->hardRefCount);
+        U_ASSERT(refCount >= 0);
+        if (refCount == 0) {
+            --fNumValuesInUse;
+        }
+    }
+    return refCount;
+}
+
+int32_t UnifiedCache::addHardRef(const SharedObject *value) const {
+    int refCount = 0;
+    if (value) {
+        refCount = umtx_atomic_inc(&value->hardRefCount);
+        U_ASSERT(refCount >= 1);
+        if (refCount == 1) {
+            fNumValuesInUse++;
+        }
+    }
+    return refCount;
+}
+
+U_NAMESPACE_END
diff --git a/intl/icu/source/common/unifiedcache.h b/intl/icu/source/common/unifiedcache.h
new file mode 100644
index 0000000000..217031a27d
--- /dev/null
+++ b/intl/icu/source/common/unifiedcache.h
@@ -0,0 +1,556 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+* Copyright (C) 2015, International Business Machines Corporation and
+* others. All Rights Reserved.
+******************************************************************************
+*
+* File UNIFIEDCACHE.H - The ICU Unified cache.
+******************************************************************************
+*/
+
+#ifndef __UNIFIED_CACHE_H__
+#define __UNIFIED_CACHE_H__
+
+#include "utypeinfo.h"  // for 'typeid' to work
+
+#include "unicode/uobject.h"
+#include "unicode/locid.h"
+#include "sharedobject.h"
+#include "unicode/unistr.h"
+#include "cstring.h"
+#include "ustr_imp.h"
+
+struct UHashtable;
+struct UHashElement;
+
+U_NAMESPACE_BEGIN
+
+class UnifiedCache;
+
+/**
+ * A base class for all cache keys.
+ */
+class U_COMMON_API CacheKeyBase : public UObject {
+ public:
+   CacheKeyBase() : fCreationStatus(U_ZERO_ERROR), fIsPrimary(false) {}
+
+   /**
+    * Copy constructor. Needed to support cloning.
+    */
+   CacheKeyBase(const CacheKeyBase &other) 
+           : UObject(other), fCreationStatus(other.fCreationStatus), fIsPrimary(false) { }
+   virtual ~CacheKeyBase();
+
+   /**
+    * Returns the hash code for this object.
+    */
+   virtual int32_t hashCode() const = 0;
+
+   /**
+    * Clones this object polymorphically. Caller owns returned value.
+    */
+   virtual CacheKeyBase *clone() const = 0;
+
+   /**
+    * Create a new object for this key. Called by cache on cache miss.
+    * createObject must add a reference to the object it returns. Note
+    * that getting an object from the cache and returning it without calling
+    * removeRef on it satisfies this requirement. It can also return nullptr
+    * and set status to an error.
+    *
+    * @param creationContext the context in which the object is being
+    *                        created. May be nullptr.
+    * @param status          Implementations can return a failure here.
+    *                        In addition, implementations may return a
+    *                        non nullptr object and set a warning status.
+    */
+   virtual const SharedObject *createObject(
+           const void *creationContext, UErrorCode &status) const = 0;
+
+   /**
+    * Writes a description of this key to buffer and returns buffer. Written
+    * description is nullptr terminated.
+    */
+   virtual char *writeDescription(char *buffer, int32_t bufSize) const = 0;
+
+   friend inline bool operator==(const CacheKeyBase& lhs,
+                                 const CacheKeyBase& rhs) {
+       return lhs.equals(rhs);
+   }
+
+   friend inline bool operator!=(const CacheKeyBase& lhs,
+                                 const CacheKeyBase& rhs) {
+       return !lhs.equals(rhs);
+   }
+
+ protected:
+   virtual bool equals(const CacheKeyBase& other) const = 0;
+
+ private:
+   mutable UErrorCode fCreationStatus;
+   mutable UBool fIsPrimary;
+   friend class UnifiedCache;
+};
+
+
+
+/**
+ * Templated version of CacheKeyBase. 
+ * A key of type LocaleCacheKey maps to a value of type T.
+ */
+template
+class CacheKey : public CacheKeyBase {
+ public:
+   virtual ~CacheKey() { }
+   /**
+    * The template parameter, T, determines the hash code returned.
+    */
+   virtual int32_t hashCode() const override {
+       const char *s = typeid(T).name();
+       return ustr_hashCharsN(s, static_cast(uprv_strlen(s)));
+   }
+
+   /**
+    * Use the value type, T,  as the description.
+    */
+   virtual char *writeDescription(char *buffer, int32_t bufLen) const override {
+       const char *s = typeid(T).name();
+       uprv_strncpy(buffer, s, bufLen);
+       buffer[bufLen - 1] = 0;
+       return buffer;
+   }
+
+ protected:
+   /**
+    * Two objects are equal if they are of the same type.
+    */
+   virtual bool equals(const CacheKeyBase &other) const override {
+       return this == &other || typeid(*this) == typeid(other);
+   }
+};
+
+/**
+ * Cache key based on locale.
+ * A key of type LocaleCacheKey maps to a value of type T.
+ */
+template
+class LocaleCacheKey : public CacheKey {
+ protected:
+   Locale   fLoc;
+   virtual bool equals(const CacheKeyBase &other) const override {
+       if (!CacheKey::equals(other)) {
+           return false;
+       }
+       // We know this and other are of same class because equals() on
+       // CacheKey returned true.
+       return operator==(static_cast &>(other));
+   }
+ public:
+   LocaleCacheKey(const Locale &loc) : fLoc(loc) {}
+   LocaleCacheKey(const LocaleCacheKey &other)
+           : CacheKey(other), fLoc(other.fLoc) { }
+   virtual ~LocaleCacheKey() { }
+   virtual int32_t hashCode() const override {
+       return (int32_t)(37u * (uint32_t)CacheKey::hashCode() + (uint32_t)fLoc.hashCode());
+   }
+   inline bool operator == (const LocaleCacheKey &other) const {
+       return fLoc == other.fLoc;
+   }
+   virtual CacheKeyBase *clone() const override {
+       return new LocaleCacheKey(*this);
+   }
+   virtual const T *createObject(
+           const void *creationContext, UErrorCode &status) const override;
+   /**
+    * Use the locale id as the description.
+    */
+   virtual char *writeDescription(char *buffer, int32_t bufLen) const override {
+       const char *s = fLoc.getName();
+       uprv_strncpy(buffer, s, bufLen);
+       buffer[bufLen - 1] = 0;
+       return buffer;
+   }
+
+};
+
+/**
+ * The unified cache. A singleton type.
+ * Design doc here:
+ * https://docs.google.com/document/d/1RwGQJs4N4tawNbf809iYDRCvXoMKqDJihxzYt1ysmd8/edit?usp=sharing
+ */
+class U_COMMON_API UnifiedCache : public UnifiedCacheBase {
+ public:
+   /**
+    * @internal
+    * Do not call directly. Instead use UnifiedCache::getInstance() as
+    * there should be only one UnifiedCache in an application.
+    */
+   UnifiedCache(UErrorCode &status);
+
+   /**
+    * Return a pointer to the global cache instance.
+    */
+   static UnifiedCache *getInstance(UErrorCode &status);
+
+   /**
+    * Fetches a value from the cache by key. Equivalent to
+    * get(key, nullptr, ptr, status);
+    */
+   template
+   void get(
+           const CacheKey& key,
+           const T *&ptr,
+           UErrorCode &status) const {
+       get(key, nullptr, ptr, status);
+   }
+
+   /**
+    * Fetches value from the cache by key.
+    *
+    * @param key             the cache key.
+    * @param creationContext passed verbatim to createObject method of key
+    * @param ptr             On entry, ptr must be nullptr or be included if
+    *                        the reference count of the object it points
+    *                        to. On exit, ptr points to the fetched object
+    *                        from the cache or is left unchanged on
+    *                        failure. Caller must call removeRef on ptr
+    *                        if set to a non nullptr value.
+    * @param status          Any error returned here. May be set to a
+    *                        warning value even if ptr is set.
+    */
+   template
+   void get(
+           const CacheKey& key,
+           const void *creationContext,
+           const T *&ptr,
+           UErrorCode &status) const {
+       if (U_FAILURE(status)) {
+           return;
+       }
+       UErrorCode creationStatus = U_ZERO_ERROR;
+       const SharedObject *value = nullptr;
+       _get(key, value, creationContext, creationStatus);
+       const T *tvalue = (const T *) value;
+       if (U_SUCCESS(creationStatus)) {
+           SharedObject::copyPtr(tvalue, ptr);
+       }
+       SharedObject::clearPtr(tvalue);
+       // Take care not to overwrite a warning status passed in with
+       // another warning or U_ZERO_ERROR.
+       if (status == U_ZERO_ERROR || U_FAILURE(creationStatus)) {
+           status = creationStatus;
+       }
+   }
+
+#ifdef UNIFIED_CACHE_DEBUG
+   /**
+    * Dumps the contents of this cache to standard error. Used for testing of
+    * cache only.
+    */
+   void dumpContents() const;
+#endif
+
+   /**
+    * Convenience method to get a value of type T from cache for a
+    * particular locale with creationContext == nullptr.
+    * @param loc    the locale
+    * @param ptr    On entry, must be nullptr or included in the ref count
+    *               of the object to which it points.
+    *               On exit, fetched value stored here or is left
+    *               unchanged on failure. Caller must call removeRef on
+    *               ptr if set to a non nullptr value.
+    * @param status Any error returned here. May be set to a
+    *               warning value even if ptr is set.
+    */
+   template
+   static void getByLocale(
+           const Locale &loc, const T *&ptr, UErrorCode &status) {
+       const UnifiedCache *cache = getInstance(status);
+       if (U_FAILURE(status)) {
+           return;
+       }
+       cache->get(LocaleCacheKey(loc), ptr, status);
+   }
+
+#ifdef UNIFIED_CACHE_DEBUG
+   /**
+    * Dumps the cache contents to stderr. For testing only.
+    */
+   static void dump();
+#endif
+
+   /**
+    * Returns the number of keys in this cache. For testing only.
+    */
+   int32_t keyCount() const;
+
+   /**
+    * Removes any values from cache that are not referenced outside
+    * the cache.
+    */
+   void flush() const;
+
+   /**
+    * Configures at what point eviction of unused entries will begin.
+    * Eviction is triggered whenever the number of evictable keys exceeds
+    * BOTH count AND (number of in-use items) * (percentageOfInUseItems / 100).
+    * Once the number of unused entries drops below one of these,
+    * eviction ceases. Because eviction happens incrementally,
+    * the actual unused entry count may exceed both these numbers
+    * from time to time.
+    *
+    * A cache entry is defined as unused if it is not essential to guarantee
+    * that for a given key X, the cache returns the same reference to the
+    * same value as long as the client already holds a reference to that
+    * value.
+    *
+    * If this method is never called, the default settings are 1000 and 100%.
+    *
+    * Although this method is thread-safe, it is designed to be called at
+    * application startup. If it is called in the middle of execution, it
+    * will have no immediate effect on the cache. However over time, the
+    * cache will perform eviction slices in an attempt to honor the new
+    * settings.
+    *
+    * If a client already holds references to many different unique values
+    * in the cache such that the number of those unique values far exceeds
+    * "count" then the cache may not be able to maintain this maximum.
+    * However, if this happens, the cache still guarantees that the number of
+    * unused entries will remain only a small percentage of the total cache
+    * size.
+    *
+    * If the parameters passed are negative, setEvctionPolicy sets status to
+    * U_ILLEGAL_ARGUMENT_ERROR.
+    */
+   void setEvictionPolicy(
+           int32_t count, int32_t percentageOfInUseItems, UErrorCode &status);
+
+
+   /**
+    * Returns how many entries have been auto evicted during the lifetime
+    * of this cache. This only includes auto evicted entries, not
+    * entries evicted because of a call to flush().
+    */
+   int64_t autoEvictedCount() const;
+
+   /**
+    * Returns the unused entry count in this cache. For testing only,
+    * Regular clients will not need this.
+    */
+   int32_t unusedCount() const;
+
+   virtual void handleUnreferencedObject() const override;
+   virtual ~UnifiedCache();
+   
+ private:
+   UHashtable *fHashtable;
+   mutable int32_t fEvictPos;
+   mutable int32_t fNumValuesTotal;
+   mutable int32_t fNumValuesInUse;
+   int32_t fMaxUnused;
+   int32_t fMaxPercentageOfInUse;
+   mutable int64_t fAutoEvictedCount;
+   SharedObject *fNoValue;
+   
+   UnifiedCache(const UnifiedCache &other) = delete;
+   UnifiedCache &operator=(const UnifiedCache &other) = delete;
+   
+   /**
+    * Flushes the contents of the cache. If cache values hold references to other
+    * cache values then _flush should be called in a loop until it returns false.
+    * 
+    * On entry, gCacheMutex must be held.
+    * On exit, those values with are evictable are flushed.
+    * 
+    *  @param all if false flush evictable items only, which are those with no external
+    *                    references, plus those that can be safely recreated.

+    *            if true, flush all elements. Any values (sharedObjects) with remaining
+    *                     hard (external) references are not deleted, but are detached from
+    *                     the cache, so that a subsequent removeRefs can delete them.
+    *                     _flush is not thread safe when all is true.
+    *   @return true if any value in cache was flushed or false otherwise.
+    */
+   UBool _flush(UBool all) const;
+   
+   /**
+    * Gets value out of cache.
+    * On entry. gCacheMutex must not be held. value must be nullptr. status
+    * must be U_ZERO_ERROR.
+    * On exit. value and status set to what is in cache at key or on cache
+    * miss the key's createObject() is called and value and status are set to
+    * the result of that. In this latter case, best effort is made to add the
+    * value and status to the cache. If createObject() fails to create a value,
+    * fNoValue is stored in cache, and value is set to nullptr. Caller must call
+    * removeRef on value if non nullptr.
+    */
+   void _get(
+           const CacheKeyBase &key,
+           const SharedObject *&value,
+           const void *creationContext,
+           UErrorCode &status) const;
+
+    /**
+     * Attempts to fetch value and status for key from cache.
+     * On entry, gCacheMutex must not be held value must be nullptr and status must
+     * be U_ZERO_ERROR.
+     * On exit, either returns false (In this
+     * case caller should try to create the object) or returns true with value
+     * pointing to the fetched value and status set to fetched status. When
+     * false is returned status may be set to failure if an in progress hash
+     * entry could not be made but value will remain unchanged. When true is
+     * returned, caller must call removeRef() on value.
+     */
+    UBool _poll(
+            const CacheKeyBase &key,
+            const SharedObject *&value,
+            UErrorCode &status) const;
+    
+    /**
+     * Places a new value and creationStatus in the cache for the given key.
+     * On entry, gCacheMutex must be held. key must not exist in the cache. 
+     * On exit, value and creation status placed under key. Soft reference added
+     * to value on successful add. On error sets status.
+     */
+    void _putNew(
+        const CacheKeyBase &key,
+        const SharedObject *value,
+        const UErrorCode creationStatus,
+        UErrorCode &status) const;
+           
+    /**
+     * Places value and status at key if there is no value at key or if cache
+     * entry for key is in progress. Otherwise, it leaves the current value and
+     * status there.
+     * 
+     * On entry. gCacheMutex must not be held. Value must be
+     * included in the reference count of the object to which it points.
+     * 
+     * On exit, value and status are changed to what was already in the cache if
+     * something was there and not in progress. Otherwise, value and status are left
+     * unchanged in which case they are placed in the cache on a best-effort basis.
+     * Caller must call removeRef() on value.
+     */
+   void _putIfAbsentAndGet(
+           const CacheKeyBase &key,
+           const SharedObject *&value,
+           UErrorCode &status) const;
+
+    /**
+     * Returns the next element in the cache round robin style.
+     * Returns nullptr if the cache is empty.
+     * On entry, gCacheMutex must be held.
+     */
+    const UHashElement *_nextElement() const;
+   
+   /**
+    * Return the number of cache items that would need to be evicted
+    * to bring usage into conformance with eviction policy.
+    * 
+    * An item corresponds to an entry in the hash table, a hash table element.
+    * 
+    * On entry, gCacheMutex must be held.
+    */
+   int32_t _computeCountOfItemsToEvict() const;
+   
+   /**
+    * Run an eviction slice.
+    * On entry, gCacheMutex must be held.
+    * _runEvictionSlice runs a slice of the evict pipeline by examining the next
+    * 10 entries in the cache round robin style evicting them if they are eligible.
+    */
+   void _runEvictionSlice() const;
+ 
+   /**
+    * Register a primary cache entry. A primary key is the first key to create
+    * a given  SharedObject value. Subsequent keys whose create function
+    * produce references to an already existing SharedObject are not primary -
+    * they can be evicted and subsequently recreated.
+    * 
+    * On entry, gCacheMutex must be held.
+    * On exit, items in use count incremented, entry is marked as a primary
+    * entry, and value registered with cache so that subsequent calls to
+    * addRef() and removeRef() on it correctly interact with the cache.
+    */
+   void _registerPrimary(const CacheKeyBase *theKey, const SharedObject *value) const;
+        
+   /**
+    * Store a value and creation error status in given hash entry.
+    * On entry, gCacheMutex must be held. Hash entry element must be in progress.
+    * value must be non nullptr.
+    * On Exit, soft reference added to value. value and status stored in hash
+    * entry. Soft reference removed from previous stored value. Waiting
+    * threads notified.
+    */
+   void _put(
+           const UHashElement *element,
+           const SharedObject *value,
+           const UErrorCode status) const;
+    /**
+     * Remove a soft reference, and delete the SharedObject if no references remain.
+     * To be used from within the UnifiedCache implementation only.
+     * gCacheMutex must be held by caller.
+     * @param value the SharedObject to be acted on.
+     */
+   void removeSoftRef(const SharedObject *value) const;
+   
+   /**
+    * Increment the hard reference count of the given SharedObject.
+    * gCacheMutex must be held by the caller.
+    * Update numValuesEvictable on transitions between zero and one reference.
+    * 
+    * @param value The SharedObject to be referenced.
+    * @return the hard reference count after the addition.
+    */
+   int32_t addHardRef(const SharedObject *value) const;
+   
+  /**
+    * Decrement the hard reference count of the given SharedObject.
+    * gCacheMutex must be held by the caller.
+    * Update numValuesEvictable on transitions between one and zero reference.
+    * 
+    * @param value The SharedObject to be referenced.
+    * @return the hard reference count after the removal.
+    */
+   int32_t removeHardRef(const SharedObject *value) const;
+
+   
+#ifdef UNIFIED_CACHE_DEBUG
+   void _dumpContents() const;
+#endif
+   
+   /**
+    *  Fetch value and error code from a particular hash entry.
+    *  On entry, gCacheMutex must be held. value must be either nullptr or must be
+    *  included in the ref count of the object to which it points.
+    *  On exit, value and status set to what is in the hash entry. Caller must
+    *  eventually call removeRef on value.
+    *  If hash entry is in progress, value will be set to gNoValue and status will
+    *  be set to U_ZERO_ERROR.
+    */
+   void _fetch(const UHashElement *element, const SharedObject *&value,
+                       UErrorCode &status) const;
+                       
+    /**
+     * Determine if given hash entry is in progress.
+     * On entry, gCacheMutex must be held.
+     */
+   UBool _inProgress(const UHashElement *element) const;
+   
+   /**
+    * Determine if given hash entry is in progress.
+    * On entry, gCacheMutex must be held.
+    */
+   UBool _inProgress(const SharedObject *theValue, UErrorCode creationStatus) const;
+   
+   /**
+    * Determine if given hash entry is eligible for eviction.
+    * On entry, gCacheMutex must be held.
+    */
+   UBool _isEvictable(const UHashElement *element) const;
+};
+
+U_NAMESPACE_END
+
+#endif
diff --git a/intl/icu/source/common/unifilt.cpp b/intl/icu/source/common/unifilt.cpp
new file mode 100644
index 0000000000..4ab0d9b5f9
--- /dev/null
+++ b/intl/icu/source/common/unifilt.cpp
@@ -0,0 +1,71 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+* Copyright (c) 2001-2012, International Business Machines
+* Corporation and others.  All Rights Reserved.
+**********************************************************************
+*   Date        Name        Description
+*   07/18/01    aliu        Creation.
+**********************************************************************
+*/
+
+#include "unicode/unifilt.h"
+#include "unicode/rep.h"
+#include "unicode/utf16.h"
+
+U_NAMESPACE_BEGIN
+UOBJECT_DEFINE_ABSTRACT_RTTI_IMPLEMENTATION(UnicodeFilter)
+
+
+/* Define this here due to the lack of another file.
+   It can't be defined in the header */
+UnicodeMatcher::~UnicodeMatcher() {}
+
+UnicodeFilter::~UnicodeFilter() {}
+
+/**
+ * UnicodeFunctor API.
+ *   Note that UnicodeMatcher is a base class of UnicodeFilter.
+ */
+UnicodeMatcher* UnicodeFilter::toMatcher() const {
+  return const_cast(this);
+}
+
+void UnicodeFilter::setData(const TransliterationRuleData*) {}
+
+/**
+ * Default implementation of UnicodeMatcher::matches() for Unicode
+ * filters.  Matches a single code point at offset (either one or
+ * two 16-bit code units).
+ */
+UMatchDegree UnicodeFilter::matches(const Replaceable& text,
+                                    int32_t& offset,
+                                    int32_t limit,
+                                    UBool incremental) {
+    UChar32 c;
+    if (offset < limit &&
+        contains(c = text.char32At(offset))) {
+        offset += U16_LENGTH(c);
+        return U_MATCH;
+    }
+    if (offset > limit &&
+        contains(c = text.char32At(offset))) {
+        // Backup offset by 1, unless the preceding character is a
+        // surrogate pair -- then backup by 2 (keep offset pointing at
+        // the lead surrogate).
+        --offset;
+        if (offset >= 0) {
+            offset -= U16_LENGTH(text.char32At(offset)) - 1;
+        }
+        return U_MATCH;
+    }
+    if (incremental && offset == limit) {
+        return U_PARTIAL_MATCH;
+    }
+    return U_MISMATCH;
+}
+
+U_NAMESPACE_END
+
+//eof
diff --git a/intl/icu/source/common/unifunct.cpp b/intl/icu/source/common/unifunct.cpp
new file mode 100644
index 0000000000..f3995b298d
--- /dev/null
+++ b/intl/icu/source/common/unifunct.cpp
@@ -0,0 +1,28 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+* Copyright (c) 2002-2004, International Business Machines
+* Corporation and others.  All Rights Reserved.
+**********************************************************************
+*/
+
+#include "unicode/unifunct.h"
+
+U_NAMESPACE_BEGIN
+
+UOBJECT_DEFINE_ABSTRACT_RTTI_IMPLEMENTATION(UnicodeFunctor)
+
+UnicodeFunctor::~UnicodeFunctor() {}
+
+UnicodeMatcher* UnicodeFunctor::toMatcher() const {
+    return 0;
+}
+
+UnicodeReplacer* UnicodeFunctor::toReplacer() const {
+    return 0;
+}
+
+U_NAMESPACE_END
+
+//eof
diff --git a/intl/icu/source/common/uniquecharstr.h b/intl/icu/source/common/uniquecharstr.h
new file mode 100644
index 0000000000..10cc924f7f
--- /dev/null
+++ b/intl/icu/source/common/uniquecharstr.h
@@ -0,0 +1,98 @@
+// © 2020 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+
+// uniquecharstr.h
+// created: 2020sep01 Frank Yung-Fong Tang
+
+#ifndef __UNIQUECHARSTR_H__
+#define __UNIQUECHARSTR_H__
+
+#include "charstr.h"
+#include "uassert.h"
+#include "uhash.h"
+
+U_NAMESPACE_BEGIN
+
+/**
+ * Stores NUL-terminated strings with duplicate elimination.
+ * Checks for unique UTF-16 string pointers and converts to invariant characters.
+ *
+ * Intended to be stack-allocated. Add strings, get a unique number for each,
+ * freeze the object, get a char * pointer for each string,
+ * call orphanCharStrings() to capture the string storage, and let this object go out of scope.
+ */
+class UniqueCharStrings {
+public:
+    UniqueCharStrings(UErrorCode &errorCode) : strings(nullptr) {
+        // Note: We hash on string contents but store stable char16_t * pointers.
+        // If the strings are stored in resource bundles which should be built with
+        // duplicate elimination, then we should be able to hash on just the pointer values.
+        uhash_init(&map, uhash_hashUChars, uhash_compareUChars, uhash_compareLong, &errorCode);
+        if (U_FAILURE(errorCode)) { return; }
+        strings = new CharString();
+        if (strings == nullptr) {
+            errorCode = U_MEMORY_ALLOCATION_ERROR;
+        }
+    }
+    ~UniqueCharStrings() {
+        uhash_close(&map);
+        delete strings;
+    }
+
+    /** Returns/orphans the CharString that contains all strings. */
+    CharString *orphanCharStrings() {
+        CharString *result = strings;
+        strings = nullptr;
+        return result;
+    }
+
+    /**
+     * Adds a string and returns a unique number for it.
+     * The string's buffer contents must not change, nor move around in memory,
+     * while this UniqueCharStrings is in use.
+     * The string contents must be NUL-terminated exactly at s.length().
+     *
+     * Best used with read-only-alias UnicodeString objects that point to
+     * stable storage, such as strings returned by resource bundle functions.
+     */
+    int32_t add(const UnicodeString &s, UErrorCode &errorCode) {
+        if (U_FAILURE(errorCode)) { return 0; }
+        if (isFrozen) {
+            errorCode = U_NO_WRITE_PERMISSION;
+            return 0;
+        }
+        // The string points into the resource bundle.
+        const char16_t *p = s.getBuffer();
+        int32_t oldIndex = uhash_geti(&map, p);
+        if (oldIndex != 0) {  // found duplicate
+            return oldIndex;
+        }
+        // Explicit NUL terminator for the previous string.
+        // The strings object is also terminated with one implicit NUL.
+        strings->append(0, errorCode);
+        int32_t newIndex = strings->length();
+        strings->appendInvariantChars(s, errorCode);
+        uhash_puti(&map, const_cast(p), newIndex, &errorCode);
+        return newIndex;
+    }
+
+    void freeze() { isFrozen = true; }
+
+    /**
+     * Returns a string pointer for its unique number, if this object is frozen.
+     * Otherwise nullptr.
+     */
+    const char *get(int32_t i) const {
+        U_ASSERT(isFrozen);
+        return isFrozen && i > 0 ? strings->data() + i : nullptr;
+    }
+
+private:
+    UHashtable map;
+    CharString *strings;
+    bool isFrozen = false;
+};
+
+U_NAMESPACE_END
+
+#endif  // __UNIQUECHARSTR_H__
diff --git a/intl/icu/source/common/uniset.cpp b/intl/icu/source/common/uniset.cpp
new file mode 100644
index 0000000000..71f57fc12d
--- /dev/null
+++ b/intl/icu/source/common/uniset.cpp
@@ -0,0 +1,2355 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+*   Copyright (C) 1999-2015, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*   Date        Name        Description
+*   10/20/99    alan        Creation.
+**********************************************************************
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/parsepos.h"
+#include "unicode/symtable.h"
+#include "unicode/uniset.h"
+#include "unicode/ustring.h"
+#include "unicode/utf8.h"
+#include "unicode/utf16.h"
+#include "ruleiter.h"
+#include "cmemory.h"
+#include "cstring.h"
+#include "patternprops.h"
+#include "uelement.h"
+#include "util.h"
+#include "uvector.h"
+#include "charstr.h"
+#include "ustrfmt.h"
+#include "uassert.h"
+#include "bmpset.h"
+#include "unisetspan.h"
+
+// HIGH_VALUE > all valid values. 110000 for codepoints
+#define UNICODESET_HIGH 0x0110000
+
+// LOW <= all valid values. ZERO for codepoints
+#define UNICODESET_LOW 0x000000
+
+/** Max list [0, 1, 2, ..., max code point, HIGH] */
+constexpr int32_t MAX_LENGTH = UNICODESET_HIGH + 1;
+
+U_NAMESPACE_BEGIN
+
+SymbolTable::~SymbolTable() {}
+
+UOBJECT_DEFINE_RTTI_IMPLEMENTATION(UnicodeSet)
+
+/**
+ * Modify the given UChar32 variable so that it is in range, by
+ * pinning values < UNICODESET_LOW to UNICODESET_LOW, and
+ * pinning values > UNICODESET_HIGH-1 to UNICODESET_HIGH-1.
+ * It modifies its argument in-place and also returns it.
+ */
+static inline UChar32 pinCodePoint(UChar32& c) {
+    if (c < UNICODESET_LOW) {
+        c = UNICODESET_LOW;
+    } else if (c > (UNICODESET_HIGH-1)) {
+        c = (UNICODESET_HIGH-1);
+    }
+    return c;
+}
+
+//----------------------------------------------------------------
+// Debugging
+//----------------------------------------------------------------
+
+// DO NOT DELETE THIS CODE.  This code is used to debug memory leaks.
+// To enable the debugging, define the symbol DEBUG_MEM in the line
+// below.  This will result in text being sent to stdout that looks
+// like this:
+//   DEBUG UnicodeSet: ct 0x00A39B20; 397 [\u0A81-\u0A83\u0A85-
+//   DEBUG UnicodeSet: dt 0x00A39B20; 396 [\u0A81-\u0A83\u0A85-
+// Each line lists a construction (ct) or destruction (dt) event, the
+// object address, the number of outstanding objects after the event,
+// and the pattern of the object in question.
+
+// #define DEBUG_MEM
+
+#ifdef DEBUG_MEM
+#include 
+static int32_t _dbgCount = 0;
+
+static inline void _dbgct(UnicodeSet* set) {
+    UnicodeString str;
+    set->toPattern(str, true);
+    char buf[40];
+    str.extract(0, 39, buf, "");
+    printf("DEBUG UnicodeSet: ct 0x%08X; %d %s\n", set, ++_dbgCount, buf);
+}
+
+static inline void _dbgdt(UnicodeSet* set) {
+    UnicodeString str;
+    set->toPattern(str, true);
+    char buf[40];
+    str.extract(0, 39, buf, "");
+    printf("DEBUG UnicodeSet: dt 0x%08X; %d %s\n", set, --_dbgCount, buf);
+}
+
+#else
+
+#define _dbgct(set)
+#define _dbgdt(set)
+
+#endif
+
+//----------------------------------------------------------------
+// UnicodeString in UVector support
+//----------------------------------------------------------------
+
+static void U_CALLCONV cloneUnicodeString(UElement *dst, UElement *src) {
+    dst->pointer = new UnicodeString(*(UnicodeString*)src->pointer);
+}
+
+static int32_t U_CALLCONV compareUnicodeString(UElement t1, UElement t2) {
+    const UnicodeString &a = *(const UnicodeString*)t1.pointer;
+    const UnicodeString &b = *(const UnicodeString*)t2.pointer;
+    return a.compare(b);
+}
+
+UBool UnicodeSet::hasStrings() const {
+    return strings != nullptr && !strings->isEmpty();
+}
+
+int32_t UnicodeSet::stringsSize() const {
+    return strings == nullptr ? 0 : strings->size();
+}
+
+UBool UnicodeSet::stringsContains(const UnicodeString &s) const {
+    return strings != nullptr && strings->contains((void*) &s);
+}
+
+//----------------------------------------------------------------
+// Constructors &c
+//----------------------------------------------------------------
+
+/**
+ * Constructs an empty set.
+ */
+UnicodeSet::UnicodeSet() {
+    list[0] = UNICODESET_HIGH;
+    _dbgct(this);
+}
+
+/**
+ * Constructs a set containing the given range. If end >
+ * start then an empty set is created.
+ *
+ * @param start first character, inclusive, of range
+ * @param end last character, inclusive, of range
+ */
+UnicodeSet::UnicodeSet(UChar32 start, UChar32 end) {
+    list[0] = UNICODESET_HIGH;
+    add(start, end);
+    _dbgct(this);
+}
+
+/**
+ * Constructs a set that is identical to the given UnicodeSet.
+ */
+UnicodeSet::UnicodeSet(const UnicodeSet& o) : UnicodeFilter(o) {
+    *this = o;
+    _dbgct(this);
+}
+
+// Copy-construct as thawed.
+UnicodeSet::UnicodeSet(const UnicodeSet& o, UBool /* asThawed */) : UnicodeFilter(o) {
+    if (ensureCapacity(o.len)) {
+        // *this = o except for bmpSet and stringSpan
+        len = o.len;
+        uprv_memcpy(list, o.list, (size_t)len*sizeof(UChar32));
+        if (o.hasStrings()) {
+            UErrorCode status = U_ZERO_ERROR;
+            if (!allocateStrings(status) ||
+                    (strings->assign(*o.strings, cloneUnicodeString, status), U_FAILURE(status))) {
+                setToBogus();
+                return;
+            }
+        }
+        if (o.pat) {
+            setPattern(o.pat, o.patLen);
+        }
+        _dbgct(this);
+    }
+}
+
+/**
+ * Destructs the set.
+ */
+UnicodeSet::~UnicodeSet() {
+    _dbgdt(this); // first!
+    if (list != stackList) {
+        uprv_free(list);
+    }
+    delete bmpSet;
+    if (buffer != stackList) {
+        uprv_free(buffer);
+    }
+    delete strings;
+    delete stringSpan;
+    releasePattern();
+}
+
+/**
+ * Assigns this object to be a copy of another.
+ */
+UnicodeSet& UnicodeSet::operator=(const UnicodeSet& o) {
+    return copyFrom(o, false);
+}
+
+UnicodeSet& UnicodeSet::copyFrom(const UnicodeSet& o, UBool asThawed) {
+    if (this == &o) {
+        return *this;
+    }
+    if (isFrozen()) {
+        return *this;
+    }
+    if (o.isBogus()) {
+        setToBogus();
+        return *this;
+    }
+    if (!ensureCapacity(o.len)) {
+        // ensureCapacity will mark the UnicodeSet as Bogus if OOM failure happens.
+        return *this;
+    }
+    len = o.len;
+    uprv_memcpy(list, o.list, (size_t)len*sizeof(UChar32));
+    if (o.bmpSet != nullptr && !asThawed) {
+        bmpSet = new BMPSet(*o.bmpSet, list, len);
+        if (bmpSet == nullptr) { // Check for memory allocation error.
+            setToBogus();
+            return *this;
+        }
+    }
+    if (o.hasStrings()) {
+        UErrorCode status = U_ZERO_ERROR;
+        if ((strings == nullptr && !allocateStrings(status)) ||
+                (strings->assign(*o.strings, cloneUnicodeString, status), U_FAILURE(status))) {
+            setToBogus();
+            return *this;
+        }
+    } else if (hasStrings()) {
+        strings->removeAllElements();
+    }
+    if (o.stringSpan != nullptr && !asThawed) {
+        stringSpan = new UnicodeSetStringSpan(*o.stringSpan, *strings);
+        if (stringSpan == nullptr) { // Check for memory allocation error.
+            setToBogus();
+            return *this;
+        }
+    }
+    releasePattern();
+    if (o.pat) {
+        setPattern(o.pat, o.patLen);
+    }
+    return *this;
+}
+
+/**
+ * Returns a copy of this object.  All UnicodeMatcher objects have
+ * to support cloning in order to allow classes using
+ * UnicodeMatchers, such as Transliterator, to implement cloning.
+ */
+UnicodeSet* UnicodeSet::clone() const {
+    return new UnicodeSet(*this);
+}
+
+UnicodeSet *UnicodeSet::cloneAsThawed() const {
+    return new UnicodeSet(*this, true);
+}
+
+/**
+ * Compares the specified object with this set for equality.  Returns
+ * true if the two sets
+ * have the same size, and every member of the specified set is
+ * contained in this set (or equivalently, every member of this set is
+ * contained in the specified set).
+ *
+ * @param o set to be compared for equality with this set.
+ * @return true if the specified set is equal to this set.
+ */
+bool UnicodeSet::operator==(const UnicodeSet& o) const {
+    if (len != o.len) return false;
+    for (int32_t i = 0; i < len; ++i) {
+        if (list[i] != o.list[i]) return false;
+    }
+    if (hasStrings() != o.hasStrings()) { return false; }
+    if (hasStrings() && *strings != *o.strings) return false;
+    return true;
+}
+
+/**
+ * Returns the hash code value for this set.
+ *
+ * @return the hash code value for this set.
+ * @see Object#hashCode()
+ */
+int32_t UnicodeSet::hashCode() const {
+    uint32_t result = static_cast(len);
+    for (int32_t i = 0; i < len; ++i) {
+        result *= 1000003u;
+        result += list[i];
+    }
+    return static_cast(result);
+}
+
+//----------------------------------------------------------------
+// Public API
+//----------------------------------------------------------------
+
+/**
+ * Returns the number of elements in this set (its cardinality),
+ * Note than the elements of a set may include both individual
+ * codepoints and strings.
+ *
+ * @return the number of elements in this set (its cardinality).
+ */
+int32_t UnicodeSet::size() const {
+    int32_t n = 0;
+    int32_t count = getRangeCount();
+    for (int32_t i = 0; i < count; ++i) {
+        n += getRangeEnd(i) - getRangeStart(i) + 1;
+    }
+    return n + stringsSize();
+}
+
+/**
+ * Returns true if this set contains no elements.
+ *
+ * @return true if this set contains no elements.
+ */
+UBool UnicodeSet::isEmpty() const {
+    return len == 1 && !hasStrings();
+}
+
+/**
+ * Returns true if this set contains the given character.
+ * @param c character to be checked for containment
+ * @return true if the test condition is met
+ */
+UBool UnicodeSet::contains(UChar32 c) const {
+    // Set i to the index of the start item greater than ch
+    // We know we will terminate without length test!
+    // LATER: for large sets, add binary search
+    //int32_t i = -1;
+    //for (;;) {
+    //    if (c < list[++i]) break;
+    //}
+    if (bmpSet != nullptr) {
+        return bmpSet->contains(c);
+    }
+    if (stringSpan != nullptr) {
+        return stringSpan->contains(c);
+    }
+    if (c >= UNICODESET_HIGH) { // Don't need to check LOW bound
+        return false;
+    }
+    int32_t i = findCodePoint(c);
+    return (UBool)(i & 1); // return true if odd
+}
+
+/**
+ * Returns the smallest value i such that c < list[i].  Caller
+ * must ensure that c is a legal value or this method will enter
+ * an infinite loop.  This method performs a binary search.
+ * @param c a character in the range MIN_VALUE..MAX_VALUE
+ * inclusive
+ * @return the smallest integer i in the range 0..len-1,
+ * inclusive, such that c < list[i]
+ */
+int32_t UnicodeSet::findCodePoint(UChar32 c) const {
+    /* Examples:
+                                       findCodePoint(c)
+       set              list[]         c=0 1 3 4 7 8
+       ===              ==============   ===========
+       []               [110000]         0 0 0 0 0 0
+       [\u0000-\u0003]  [0, 4, 110000]   1 1 1 2 2 2
+       [\u0004-\u0007]  [4, 8, 110000]   0 0 0 1 1 2
+       [:Any:]          [0, 110000]      1 1 1 1 1 1
+     */
+
+    // Return the smallest i such that c < list[i].  Assume
+    // list[len - 1] == HIGH and that c is legal (0..HIGH-1).
+    if (c < list[0])
+        return 0;
+    // High runner test.  c is often after the last range, so an
+    // initial check for this condition pays off.
+    int32_t lo = 0;
+    int32_t hi = len - 1;
+    if (lo >= hi || c >= list[hi-1])
+        return hi;
+    // invariant: c >= list[lo]
+    // invariant: c < list[hi]
+    for (;;) {
+        int32_t i = (lo + hi) >> 1;
+        if (i == lo) {
+            break; // Found!
+        } else if (c < list[i]) {
+            hi = i;
+        } else {
+            lo = i;
+        }
+    }
+    return hi;
+}
+
+/**
+ * Returns true if this set contains every character
+ * of the given range.
+ * @param start first character, inclusive, of the range
+ * @param end last character, inclusive, of the range
+ * @return true if the test condition is met
+ */
+UBool UnicodeSet::contains(UChar32 start, UChar32 end) const {
+    //int32_t i = -1;
+    //for (;;) {
+    //    if (start < list[++i]) break;
+    //}
+    int32_t i = findCodePoint(start);
+    return ((i & 1) != 0 && end < list[i]);
+}
+
+/**
+ * Returns true if this set contains the given
+ * multicharacter string.
+ * @param s string to be checked for containment
+ * @return true if this set contains the specified string
+ */
+UBool UnicodeSet::contains(const UnicodeString& s) const {
+    int32_t cp = getSingleCP(s);
+    if (cp < 0) {
+        return stringsContains(s);
+    } else {
+        return contains((UChar32) cp);
+    }
+}
+
+/**
+ * Returns true if this set contains all the characters and strings
+ * of the given set.
+ * @param c set to be checked for containment
+ * @return true if the test condition is met
+ */
+UBool UnicodeSet::containsAll(const UnicodeSet& c) const {
+    // The specified set is a subset if all of its pairs are contained in
+    // this set.  It's possible to code this more efficiently in terms of
+    // direct manipulation of the inversion lists if the need arises.
+    int32_t n = c.getRangeCount();
+    for (int i=0; icontainsAll(*c.strings));
+}
+
+/**
+ * Returns true if this set contains all the characters
+ * of the given string.
+ * @param s string containing characters to be checked for containment
+ * @return true if the test condition is met
+ */
+UBool UnicodeSet::containsAll(const UnicodeString& s) const {
+    return (UBool)(span(s.getBuffer(), s.length(), USET_SPAN_CONTAINED) ==
+                   s.length());
+}
+
+/**
+ * Returns true if this set contains none of the characters
+ * of the given range.
+ * @param start first character, inclusive, of the range
+ * @param end last character, inclusive, of the range
+ * @return true if the test condition is met
+ */
+UBool UnicodeSet::containsNone(UChar32 start, UChar32 end) const {
+    //int32_t i = -1;
+    //for (;;) {
+    //    if (start < list[++i]) break;
+    //}
+    int32_t i = findCodePoint(start);
+    return ((i & 1) == 0 && end < list[i]);
+}
+
+/**
+ * Returns true if this set contains none of the characters and strings
+ * of the given set.
+ * @param c set to be checked for containment
+ * @return true if the test condition is met
+ */
+UBool UnicodeSet::containsNone(const UnicodeSet& c) const {
+    // The specified set is a subset if all of its pairs are contained in
+    // this set.  It's possible to code this more efficiently in terms of
+    // direct manipulation of the inversion lists if the need arises.
+    int32_t n = c.getRangeCount();
+    for (int32_t i=0; icontainsNone(*c.strings);
+}
+
+/**
+ * Returns true if this set contains none of the characters
+ * of the given string.
+ * @param s string containing characters to be checked for containment
+ * @return true if the test condition is met
+ */
+UBool UnicodeSet::containsNone(const UnicodeString& s) const {
+    return (UBool)(span(s.getBuffer(), s.length(), USET_SPAN_NOT_CONTAINED) ==
+                   s.length());
+}
+
+/**
+ * Returns true if this set contains any character whose low byte
+ * is the given value.  This is used by RuleBasedTransliterator for
+ * indexing.
+ */
+UBool UnicodeSet::matchesIndexValue(uint8_t v) const {
+    /* The index value v, in the range [0,255], is contained in this set if
+     * it is contained in any pair of this set.  Pairs either have the high
+     * bytes equal, or unequal.  If the high bytes are equal, then we have
+     * aaxx..aayy, where aa is the high byte.  Then v is contained if xx <=
+     * v <= yy.  If the high bytes are unequal we have aaxx..bbyy, bb>aa.
+     * Then v is contained if xx <= v || v <= yy.  (This is identical to the
+     * time zone month containment logic.)
+     */
+    int32_t i;
+    int32_t rangeCount=getRangeCount();
+    for (i=0; isize(); ++i) {
+            const UnicodeString& s = *(const UnicodeString*)strings->elementAt(i);
+            if (s.isEmpty()) {
+                continue;  // skip the empty string
+            }
+            UChar32 c = s.char32At(0);
+            if ((c & 0xFF) == v) {
+                return true;
+            }
+        }
+    }
+    return false;
+}
+
+/**
+ * Implementation of UnicodeMatcher::matches().  Always matches the
+ * longest possible multichar string.
+ */
+UMatchDegree UnicodeSet::matches(const Replaceable& text,
+                                 int32_t& offset,
+                                 int32_t limit,
+                                 UBool incremental) {
+    if (offset == limit) {
+        if (contains(U_ETHER)) {
+            return incremental ? U_PARTIAL_MATCH : U_MATCH;
+        } else {
+            return U_MISMATCH;
+        }
+    } else {
+        if (hasStrings()) { // try strings first
+
+            // might separate forward and backward loops later
+            // for now they are combined
+
+            // TODO Improve efficiency of this, at least in the forward
+            // direction, if not in both.  In the forward direction we
+            // can assume the strings are sorted.
+
+            int32_t i;
+            UBool forward = offset < limit;
+
+            // firstChar is the leftmost char to match in the
+            // forward direction or the rightmost char to match in
+            // the reverse direction.
+            char16_t firstChar = text.charAt(offset);
+
+            // If there are multiple strings that can match we
+            // return the longest match.
+            int32_t highWaterLength = 0;
+
+            for (i=0; isize(); ++i) {
+                const UnicodeString& trial = *(const UnicodeString*)strings->elementAt(i);
+                if (trial.isEmpty()) {
+                    continue;  // skip the empty string
+                }
+
+                char16_t c = trial.charAt(forward ? 0 : trial.length() - 1);
+
+                // Strings are sorted, so we can optimize in the
+                // forward direction.
+                if (forward && c > firstChar) break;
+                if (c != firstChar) continue;
+
+                int32_t matchLen = matchRest(text, offset, limit, trial);
+
+                if (incremental) {
+                    int32_t maxLen = forward ? limit-offset : offset-limit;
+                    if (matchLen == maxLen) {
+                        // We have successfully matched but only up to limit.
+                        return U_PARTIAL_MATCH;
+                    }
+                }
+
+                if (matchLen == trial.length()) {
+                    // We have successfully matched the whole string.
+                    if (matchLen > highWaterLength) {
+                        highWaterLength = matchLen;
+                    }
+                    // In the forward direction we know strings
+                    // are sorted so we can bail early.
+                    if (forward && matchLen < highWaterLength) {
+                        break;
+                    }
+                    continue;
+                }
+            }
+
+            // We've checked all strings without a partial match.
+            // If we have full matches, return the longest one.
+            if (highWaterLength != 0) {
+                offset += forward ? highWaterLength : -highWaterLength;
+                return U_MATCH;
+            }
+        }
+        return UnicodeFilter::matches(text, offset, limit, incremental);
+    }
+}
+
+/**
+ * Returns the longest match for s in text at the given position.
+ * If limit > start then match forward from start+1 to limit
+ * matching all characters except s.charAt(0).  If limit < start,
+ * go backward starting from start-1 matching all characters
+ * except s.charAt(s.length()-1).  This method assumes that the
+ * first character, text.charAt(start), matches s, so it does not
+ * check it.
+ * @param text the text to match
+ * @param start the first character to match.  In the forward
+ * direction, text.charAt(start) is matched against s.charAt(0).
+ * In the reverse direction, it is matched against
+ * s.charAt(s.length()-1).
+ * @param limit the limit offset for matching, either last+1 in
+ * the forward direction, or last-1 in the reverse direction,
+ * where last is the index of the last character to match.
+ * @return If part of s matches up to the limit, return |limit -
+ * start|.  If all of s matches before reaching the limit, return
+ * s.length().  If there is a mismatch between s and text, return
+ * 0
+ */
+int32_t UnicodeSet::matchRest(const Replaceable& text,
+                              int32_t start, int32_t limit,
+                              const UnicodeString& s) {
+    int32_t i;
+    int32_t maxLen;
+    int32_t slen = s.length();
+    if (start < limit) {
+        maxLen = limit - start;
+        if (maxLen > slen) maxLen = slen;
+        for (i = 1; i < maxLen; ++i) {
+            if (text.charAt(start + i) != s.charAt(i)) return 0;
+        }
+    } else {
+        maxLen = start - limit;
+        if (maxLen > slen) maxLen = slen;
+        --slen; // <=> slen = s.length() - 1;
+        for (i = 1; i < maxLen; ++i) {
+            if (text.charAt(start - i) != s.charAt(slen - i)) return 0;
+        }
+    }
+    return maxLen;
+}
+
+/**
+ * Implement of UnicodeMatcher
+ */
+void UnicodeSet::addMatchSetTo(UnicodeSet& toUnionTo) const {
+    toUnionTo.addAll(*this);
+}
+
+/**
+ * Returns the index of the given character within this set, where
+ * the set is ordered by ascending code point.  If the character
+ * is not in this set, return -1.  The inverse of this method is
+ * charAt().
+ * @return an index from 0..size()-1, or -1
+ */
+int32_t UnicodeSet::indexOf(UChar32 c) const {
+    if (c < MIN_VALUE || c > MAX_VALUE) {
+        return -1;
+    }
+    int32_t i = 0;
+    int32_t n = 0;
+    for (;;) {
+        UChar32 start = list[i++];
+        if (c < start) {
+            return -1;
+        }
+        UChar32 limit = list[i++];
+        if (c < limit) {
+            return n + c - start;
+        }
+        n += limit - start;
+    }
+}
+
+/**
+ * Returns the character at the given index within this set, where
+ * the set is ordered by ascending code point.  If the index is
+ * out of range, return (UChar32)-1.  The inverse of this method is
+ * indexOf().
+ * @param index an index from 0..size()-1
+ * @return the character at the given index, or (UChar32)-1.
+ */
+UChar32 UnicodeSet::charAt(int32_t index) const {
+    if (index >= 0) {
+        // len2 is the largest even integer <= len, that is, it is len
+        // for even values and len-1 for odd values.  With odd values
+        // the last entry is UNICODESET_HIGH.
+        int32_t len2 = len & ~1;
+        for (int32_t i=0; i < len2;) {
+            UChar32 start = list[i++];
+            int32_t count = list[i++] - start;
+            if (index < count) {
+                return (UChar32)(start + index);
+            }
+            index -= count;
+        }
+    }
+    return (UChar32)-1;
+}
+
+/**
+ * Make this object represent the range start - end.
+ * If end > start then this object is set to an
+ * an empty range.
+ *
+ * @param start first character in the set, inclusive
+ * @rparam end last character in the set, inclusive
+ */
+UnicodeSet& UnicodeSet::set(UChar32 start, UChar32 end) {
+    clear();
+    complement(start, end);
+    return *this;
+}
+
+/**
+ * Adds the specified range to this set if it is not already
+ * present.  If this set already contains the specified range,
+ * the call leaves this set unchanged.  If end > start
+ * then an empty range is added, leaving the set unchanged.
+ *
+ * @param start first character, inclusive, of range to be added
+ * to this set.
+ * @param end last character, inclusive, of range to be added
+ * to this set.
+ */
+UnicodeSet& UnicodeSet::add(UChar32 start, UChar32 end) {
+    if (pinCodePoint(start) < pinCodePoint(end)) {
+        UChar32 limit = end + 1;
+        // Fast path for adding a new range after the last one.
+        // Odd list length: [..., lastStart, lastLimit, HIGH]
+        if ((len & 1) != 0) {
+            // If the list is empty, set lastLimit low enough to not be adjacent to 0.
+            UChar32 lastLimit = len == 1 ? -2 : list[len - 2];
+            if (lastLimit <= start && !isFrozen() && !isBogus()) {
+                if (lastLimit == start) {
+                    // Extend the last range.
+                    list[len - 2] = limit;
+                    if (limit == UNICODESET_HIGH) {
+                        --len;
+                    }
+                } else {
+                    list[len - 1] = start;
+                    if (limit < UNICODESET_HIGH) {
+                        if (ensureCapacity(len + 2)) {
+                            list[len++] = limit;
+                            list[len++] = UNICODESET_HIGH;
+                        }
+                    } else {  // limit == UNICODESET_HIGH
+                        if (ensureCapacity(len + 1)) {
+                            list[len++] = UNICODESET_HIGH;
+                        }
+                    }
+                }
+                releasePattern();
+                return *this;
+            }
+        }
+        // This is slow. Could be much faster using findCodePoint(start)
+        // and modifying the list, dealing with adjacent & overlapping ranges.
+        UChar32 range[3] = { start, limit, UNICODESET_HIGH };
+        add(range, 2, 0);
+    } else if (start == end) {
+        add(start);
+    }
+    return *this;
+}
+
+// #define DEBUG_US_ADD
+
+#ifdef DEBUG_US_ADD
+#include 
+void dump(UChar32 c) {
+    if (c <= 0xFF) {
+        printf("%c", (char)c);
+    } else {
+        printf("U+%04X", c);
+    }
+}
+void dump(const UChar32* list, int32_t len) {
+    printf("[");
+    for (int32_t i=0; i ");
+#endif
+
+    if (c == list[i]-1) {
+        // c is before start of next range
+        list[i] = c;
+        // if we touched the HIGH mark, then add a new one
+        if (c == (UNICODESET_HIGH - 1)) {
+            if (!ensureCapacity(len+1)) {
+                // ensureCapacity will mark the object as Bogus if OOM failure happens.
+                return *this;
+            }
+            list[len++] = UNICODESET_HIGH;
+        }
+        if (i > 0 && c == list[i-1]) {
+            // collapse adjacent ranges
+
+            // [..., start_k-1, c, c, limit_k, ..., HIGH]
+            //                     ^
+            //                     list[i]
+
+            //for (int32_t k=i-1; k 0 && c == list[i-1]) {
+        // c is after end of prior range
+        list[i-1]++;
+        // no need to check for collapse here
+    }
+
+    else {
+        // At this point we know the new char is not adjacent to
+        // any existing ranges, and it is not 10FFFF.
+
+
+        // [..., start_k-1, limit_k-1, start_k, limit_k, ..., HIGH]
+        //                             ^
+        //                             list[i]
+
+        // [..., start_k-1, limit_k-1, c, c+1, start_k, limit_k, ..., HIGH]
+        //                             ^
+        //                             list[i]
+
+        if (!ensureCapacity(len+2)) {
+            // ensureCapacity will mark the object as Bogus if OOM failure happens.
+            return *this;
+        }
+
+        UChar32 *p = list + i;
+        uprv_memmove(p + 2, p, (len - i) * sizeof(*p));
+        list[i] = c;
+        list[i+1] = c+1;
+        len += 2;
+    }
+
+#ifdef DEBUG_US_ADD
+    dump(list, len);
+    printf("\n");
+
+    for (i=1; i {"ch"}
+ *
+ * @param s the source string
+ * @return the modified set, for chaining
+ */
+UnicodeSet& UnicodeSet::add(const UnicodeString& s) {
+    if (isFrozen() || isBogus()) return *this;
+    int32_t cp = getSingleCP(s);
+    if (cp < 0) {
+        if (!stringsContains(s)) {
+            _add(s);
+            releasePattern();
+        }
+    } else {
+        add((UChar32)cp);
+    }
+    return *this;
+}
+
+/**
+ * Adds the given string, in order, to 'strings'.  The given string
+ * must have been checked by the caller to not already be in 'strings'.
+ */
+void UnicodeSet::_add(const UnicodeString& s) {
+    if (isFrozen() || isBogus()) {
+        return;
+    }
+    UErrorCode ec = U_ZERO_ERROR;
+    if (strings == nullptr && !allocateStrings(ec)) {
+        setToBogus();
+        return;
+    }
+    UnicodeString* t = new UnicodeString(s);
+    if (t == nullptr) { // Check for memory allocation error.
+        setToBogus();
+        return;
+    }
+    strings->sortedInsert(t, compareUnicodeString, ec);
+    if (U_FAILURE(ec)) {
+        setToBogus();
+    }
+}
+
+/**
+ * @return a code point IF the string consists of a single one.
+ * otherwise returns -1.
+ * @param string to test
+ */
+int32_t UnicodeSet::getSingleCP(const UnicodeString& s) {
+    int32_t sLength = s.length();
+    if (sLength == 1) return s.charAt(0);
+    if (sLength == 2) {
+        UChar32 cp = s.char32At(0);
+        if (cp > 0xFFFF) { // is surrogate pair
+            return cp;
+        }
+    }
+    return -1;
+}
+
+/**
+ * Adds each of the characters in this string to the set. Thus "ch" => {"c", "h"}
+ * If this set already any particular character, it has no effect on that character.
+ * @param the source string
+ * @return the modified set, for chaining
+ */
+UnicodeSet& UnicodeSet::addAll(const UnicodeString& s) {
+    UChar32 cp;
+    for (int32_t i = 0; i < s.length(); i += U16_LENGTH(cp)) {
+        cp = s.char32At(i);
+        add(cp);
+    }
+    return *this;
+}
+
+/**
+ * Retains EACH of the characters in this string. Note: "ch" == {"c", "h"}
+ * If this set already any particular character, it has no effect on that character.
+ * @param the source string
+ * @return the modified set, for chaining
+ */
+UnicodeSet& UnicodeSet::retainAll(const UnicodeString& s) {
+    UnicodeSet set;
+    set.addAll(s);
+    retainAll(set);
+    return *this;
+}
+
+/**
+ * Complement EACH of the characters in this string. Note: "ch" == {"c", "h"}
+ * If this set already any particular character, it has no effect on that character.
+ * @param the source string
+ * @return the modified set, for chaining
+ */
+UnicodeSet& UnicodeSet::complementAll(const UnicodeString& s) {
+    UnicodeSet set;
+    set.addAll(s);
+    complementAll(set);
+    return *this;
+}
+
+/**
+ * Remove EACH of the characters in this string. Note: "ch" == {"c", "h"}
+ * If this set already any particular character, it has no effect on that character.
+ * @param the source string
+ * @return the modified set, for chaining
+ */
+UnicodeSet& UnicodeSet::removeAll(const UnicodeString& s) {
+    UnicodeSet set;
+    set.addAll(s);
+    removeAll(set);
+    return *this;
+}
+
+UnicodeSet& UnicodeSet::removeAllStrings() {
+    if (!isFrozen() && hasStrings()) {
+        strings->removeAllElements();
+        releasePattern();
+    }
+    return *this;
+}
+
+
+/**
+ * Makes a set from a multicharacter string. Thus "ch" => {"ch"}
+ * 
Warning: you cannot add an empty string ("") to a UnicodeSet.
+ * @param the source string
+ * @return a newly created set containing the given string
+ */
+UnicodeSet* U_EXPORT2 UnicodeSet::createFrom(const UnicodeString& s) {
+    UnicodeSet *set = new UnicodeSet();
+    if (set != nullptr) { // Check for memory allocation error.
+        set->add(s);
+    }
+    return set;
+}
+
+
+/**
+ * Makes a set from each of the characters in the string. Thus "ch" => {"c", "h"}
+ * @param the source string
+ * @return a newly created set containing the given characters
+ */
+UnicodeSet* U_EXPORT2 UnicodeSet::createFromAll(const UnicodeString& s) {
+    UnicodeSet *set = new UnicodeSet();
+    if (set != nullptr) { // Check for memory allocation error.
+        set->addAll(s);
+    }
+    return set;
+}
+
+/**
+ * Retain only the elements in this set that are contained in the
+ * specified range.  If end > start then an empty range is
+ * retained, leaving the set empty.
+ *
+ * @param start first character, inclusive, of range to be retained
+ * to this set.
+ * @param end last character, inclusive, of range to be retained
+ * to this set.
+ */
+UnicodeSet& UnicodeSet::retain(UChar32 start, UChar32 end) {
+    if (pinCodePoint(start) <= pinCodePoint(end)) {
+        UChar32 range[3] = { start, end+1, UNICODESET_HIGH };
+        retain(range, 2, 0);
+    } else {
+        clear();
+    }
+    return *this;
+}
+
+UnicodeSet& UnicodeSet::retain(UChar32 c) {
+    return retain(c, c);
+}
+
+UnicodeSet& UnicodeSet::retain(const UnicodeString &s) {
+    if (isFrozen() || isBogus()) { return *this; }
+    UChar32 cp = getSingleCP(s);
+    if (cp < 0) {
+        bool isIn = stringsContains(s);
+        // Check for getRangeCount() first to avoid somewhat-expensive size()
+        // when there are single code points.
+        if (isIn && getRangeCount() == 0 && size() == 1) {
+            return *this;
+        }
+        clear();
+        if (isIn) {
+            _add(s);
+        }
+    } else {
+        retain(cp, cp);
+    }
+    return *this;
+}
+
+/**
+ * Removes the specified range from this set if it is present.
+ * The set will not contain the specified range once the call
+ * returns.  If end > start then an empty range is
+ * removed, leaving the set unchanged.
+ *
+ * @param start first character, inclusive, of range to be removed
+ * from this set.
+ * @param end last character, inclusive, of range to be removed
+ * from this set.
+ */
+UnicodeSet& UnicodeSet::remove(UChar32 start, UChar32 end) {
+    if (pinCodePoint(start) <= pinCodePoint(end)) {
+        UChar32 range[3] = { start, end+1, UNICODESET_HIGH };
+        retain(range, 2, 2);
+    }
+    return *this;
+}
+
+/**
+ * Removes the specified character from this set if it is present.
+ * The set will not contain the specified range once the call
+ * returns.
+ */
+UnicodeSet& UnicodeSet::remove(UChar32 c) {
+    return remove(c, c);
+}
+
+/**
+ * Removes the specified string from this set if it is present.
+ * The set will not contain the specified character once the call
+ * returns.
+ * @param the source string
+ * @return the modified set, for chaining
+ */
+UnicodeSet& UnicodeSet::remove(const UnicodeString& s) {
+    if (isFrozen() || isBogus()) return *this;
+    int32_t cp = getSingleCP(s);
+    if (cp < 0) {
+        if (strings != nullptr && strings->removeElement((void*) &s)) {
+            releasePattern();
+        }
+    } else {
+        remove((UChar32)cp, (UChar32)cp);
+    }
+    return *this;
+}
+
+/**
+ * Complements the specified range in this set.  Any character in
+ * the range will be removed if it is in this set, or will be
+ * added if it is not in this set.  If end > start
+ * then an empty range is xor'ed, leaving the set unchanged.
+ *
+ * @param start first character, inclusive, of range to be removed
+ * from this set.
+ * @param end last character, inclusive, of range to be removed
+ * from this set.
+ */
+UnicodeSet& UnicodeSet::complement(UChar32 start, UChar32 end) {
+    if (isFrozen() || isBogus()) {
+        return *this;
+    }
+    if (pinCodePoint(start) <= pinCodePoint(end)) {
+        UChar32 range[3] = { start, end+1, UNICODESET_HIGH };
+        exclusiveOr(range, 2, 0);
+    }
+    releasePattern();
+    return *this;
+}
+
+UnicodeSet& UnicodeSet::complement(UChar32 c) {
+    return complement(c, c);
+}
+
+/**
+ * This is equivalent to
+ * complement(MIN_VALUE, MAX_VALUE).
+ */
+UnicodeSet& UnicodeSet::complement() {
+    if (isFrozen() || isBogus()) {
+        return *this;
+    }
+    if (list[0] == UNICODESET_LOW) {
+        uprv_memmove(list, list + 1, (size_t)(len-1)*sizeof(UChar32));
+        --len;
+    } else {
+        if (!ensureCapacity(len+1)) {
+            return *this;
+        }
+        uprv_memmove(list + 1, list, (size_t)len*sizeof(UChar32));
+        list[0] = UNICODESET_LOW;
+        ++len;
+    }
+    releasePattern();
+    return *this;
+}
+
+/**
+ * Complement the specified string in this set.
+ * The set will not contain the specified string once the call
+ * returns.
+ *
+ * @param s the string to complement
+ * @return this object, for chaining
+ */
+UnicodeSet& UnicodeSet::complement(const UnicodeString& s) {
+    if (isFrozen() || isBogus()) return *this;
+    int32_t cp = getSingleCP(s);
+    if (cp < 0) {
+        if (stringsContains(s)) {
+            strings->removeElement((void*) &s);
+        } else {
+            _add(s);
+        }
+        releasePattern();
+    } else {
+        complement((UChar32)cp, (UChar32)cp);
+    }
+    return *this;
+}
+
+/**
+ * Adds all of the elements in the specified set to this set if
+ * they're not already present.  This operation effectively
+ * modifies this set so that its value is the union of the two
+ * sets.  The behavior of this operation is unspecified if the specified
+ * collection is modified while the operation is in progress.
+ *
+ * @param c set whose elements are to be added to this set.
+ * @see #add(char, char)
+ */
+UnicodeSet& UnicodeSet::addAll(const UnicodeSet& c) {
+    if ( c.len>0 && c.list!=nullptr ) {
+        add(c.list, c.len, 0);
+    }
+
+    // Add strings in order
+    if ( c.strings!=nullptr ) {
+        for (int32_t i=0; isize(); ++i) {
+            const UnicodeString* s = (const UnicodeString*)c.strings->elementAt(i);
+            if (!stringsContains(*s)) {
+                _add(*s);
+            }
+        }
+    }
+    return *this;
+}
+
+/**
+ * Retains only the elements in this set that are contained in the
+ * specified set.  In other words, removes from this set all of
+ * its elements that are not contained in the specified set.  This
+ * operation effectively modifies this set so that its value is
+ * the intersection of the two sets.
+ *
+ * @param c set that defines which elements this set will retain.
+ */
+UnicodeSet& UnicodeSet::retainAll(const UnicodeSet& c) {
+    if (isFrozen() || isBogus()) {
+        return *this;
+    }
+    retain(c.list, c.len, 0);
+    if (hasStrings()) {
+        if (!c.hasStrings()) {
+            strings->removeAllElements();
+        } else {
+            strings->retainAll(*c.strings);
+        }
+    }
+    return *this;
+}
+
+/**
+ * Removes from this set all of its elements that are contained in the
+ * specified set.  This operation effectively modifies this
+ * set so that its value is the asymmetric set difference of
+ * the two sets.
+ *
+ * @param c set that defines which elements will be removed from
+ *          this set.
+ */
+UnicodeSet& UnicodeSet::removeAll(const UnicodeSet& c) {
+    if (isFrozen() || isBogus()) {
+        return *this;
+    }
+    retain(c.list, c.len, 2);
+    if (hasStrings() && c.hasStrings()) {
+        strings->removeAll(*c.strings);
+    }
+    return *this;
+}
+
+/**
+ * Complements in this set all elements contained in the specified
+ * set.  Any character in the other set will be removed if it is
+ * in this set, or will be added if it is not in this set.
+ *
+ * @param c set that defines which elements will be xor'ed from
+ *          this set.
+ */
+UnicodeSet& UnicodeSet::complementAll(const UnicodeSet& c) {
+    if (isFrozen() || isBogus()) {
+        return *this;
+    }
+    exclusiveOr(c.list, c.len, 0);
+
+    if (c.strings != nullptr) {
+        for (int32_t i=0; isize(); ++i) {
+            void* e = c.strings->elementAt(i);
+            if (strings == nullptr || !strings->removeElement(e)) {
+                _add(*(const UnicodeString*)e);
+            }
+        }
+    }
+    return *this;
+}
+
+/**
+ * Removes all of the elements from this set.  This set will be
+ * empty after this call returns.
+ */
+UnicodeSet& UnicodeSet::clear() {
+    if (isFrozen()) {
+        return *this;
+    }
+    list[0] = UNICODESET_HIGH;
+    len = 1;
+    releasePattern();
+    if (strings != nullptr) {
+        strings->removeAllElements();
+    }
+    // Remove bogus
+    fFlags = 0;
+    return *this;
+}
+
+/**
+ * Iteration method that returns the number of ranges contained in
+ * this set.
+ * @see #getRangeStart
+ * @see #getRangeEnd
+ */
+int32_t UnicodeSet::getRangeCount() const {
+    return len/2;
+}
+
+/**
+ * Iteration method that returns the first character in the
+ * specified range of this set.
+ * @see #getRangeCount
+ * @see #getRangeEnd
+ */
+UChar32 UnicodeSet::getRangeStart(int32_t index) const {
+    return list[index*2];
+}
+
+/**
+ * Iteration method that returns the last character in the
+ * specified range of this set.
+ * @see #getRangeStart
+ * @see #getRangeEnd
+ */
+UChar32 UnicodeSet::getRangeEnd(int32_t index) const {
+    return list[index*2 + 1] - 1;
+}
+
+const UnicodeString* UnicodeSet::getString(int32_t index) const {
+    return (const UnicodeString*) strings->elementAt(index);
+}
+
+/**
+ * Reallocate this objects internal structures to take up the least
+ * possible space, without changing this object's value.
+ */
+UnicodeSet& UnicodeSet::compact() {
+    if (isFrozen() || isBogus()) {
+        return *this;
+    }
+    // Delete buffer first to defragment memory less.
+    if (buffer != stackList) {
+        uprv_free(buffer);
+        buffer = nullptr;
+        bufferCapacity = 0;
+    }
+    if (list == stackList) {
+        // pass
+    } else if (len <= INITIAL_CAPACITY) {
+        uprv_memcpy(stackList, list, len * sizeof(UChar32));
+        uprv_free(list);
+        list = stackList;
+        capacity = INITIAL_CAPACITY;
+    } else if ((len + 7) < capacity) {
+        // If we have more than a little unused capacity, shrink it to len.
+        UChar32* temp = (UChar32*) uprv_realloc(list, sizeof(UChar32) * len);
+        if (temp) {
+            list = temp;
+            capacity = len;
+        }
+        // else what the heck happened?! We allocated less memory!
+        // Oh well. We'll keep our original array.
+    }
+    if (strings != nullptr && strings->isEmpty()) {
+        delete strings;
+        strings = nullptr;
+    }
+    return *this;
+}
+
+#ifdef DEBUG_SERIALIZE
+#include 
+#endif
+
+/**
+ * Deserialize constructor.
+ */
+UnicodeSet::UnicodeSet(const uint16_t data[], int32_t dataLen, ESerialization serialization,
+                       UErrorCode &ec) {
+
+  if(U_FAILURE(ec)) {
+    setToBogus();
+    return;
+  }
+
+  if( (serialization != kSerialized)
+      || (data==nullptr)
+      || (dataLen < 1)) {
+    ec = U_ILLEGAL_ARGUMENT_ERROR;
+    setToBogus();
+    return;
+  }
+
+  // bmp?
+  int32_t headerSize = ((data[0]&0x8000)) ?2:1;
+  int32_t bmpLength = (headerSize==1)?data[0]:data[1];
+
+  int32_t newLength = (((data[0]&0x7FFF)-bmpLength)/2)+bmpLength;
+#ifdef DEBUG_SERIALIZE
+  printf("dataLen %d headerSize %d bmpLen %d len %d. data[0]=%X/%X/%X/%X\n", dataLen,headerSize,bmpLength,newLength, data[0],data[1],data[2],data[3]);
+#endif
+  if(!ensureCapacity(newLength + 1)) {  // +1 for HIGH
+    return;
+  }
+  // copy bmp
+  int32_t i;
+  for(i = 0; i< bmpLength;i++) {
+    list[i] = data[i+headerSize];
+#ifdef DEBUG_SERIALIZE
+    printf("<<16@%d[%d] %X\n", i+headerSize, i, list[i]);
+#endif
+  }
+  // copy smp
+  for(i=bmpLength;i0 && dest==nullptr)) {
+        ec=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    /* count necessary 16-bit units */
+    length=this->len-1; // Subtract 1 to ignore final UNICODESET_HIGH
+    // assert(length>=0);
+    if (length==0) {
+        /* empty set */
+        if (destCapacity>0) {
+            *dest=0;
+        } else {
+            ec=U_BUFFER_OVERFLOW_ERROR;
+        }
+        return 1;
+    }
+    /* now length>0 */
+
+    if (this->list[length-1]<=0xffff) {
+        /* all BMP */
+        bmpLength=length;
+    } else if (this->list[0]>=0x10000) {
+        /* all supplementary */
+        bmpLength=0;
+        length*=2;
+    } else {
+        /* some BMP, some supplementary */
+        for (bmpLength=0; bmpLengthlist[bmpLength]<=0xffff; ++bmpLength) {}
+        length=bmpLength+2*(length-bmpLength);
+    }
+#ifdef DEBUG_SERIALIZE
+    printf(">> bmpLength%d length%d len%d\n", bmpLength, length, len);
+#endif
+    /* length: number of 16-bit array units */
+    if (length>0x7fff) {
+        /* there are only 15 bits for the length in the first serialized word */
+        ec=U_INDEX_OUTOFBOUNDS_ERROR;
+        return 0;
+    }
+
+    /*
+     * total serialized length:
+     * number of 16-bit array units (length) +
+     * 1 length unit (always) +
+     * 1 bmpLength unit (if there are supplementary values)
+     */
+    destLength=length+((length>bmpLength)?2:1);
+    if (destLength<=destCapacity) {
+        const UChar32 *p;
+        int32_t i;
+
+#ifdef DEBUG_SERIALIZE
+        printf("writeHdr\n");
+#endif
+        *dest=(uint16_t)length;
+        if (length>bmpLength) {
+            *dest|=0x8000;
+            *++dest=(uint16_t)bmpLength;
+        }
+        ++dest;
+
+        /* write the BMP part of the array */
+        p=this->list;
+        for (i=0; i>16);
+            *dest++=(uint16_t)*p++;
+        }
+    } else {
+        ec=U_BUFFER_OVERFLOW_ERROR;
+    }
+    return destLength;
+}
+
+//----------------------------------------------------------------
+// Implementation: Utility methods
+//----------------------------------------------------------------
+
+/**
+ * Allocate our strings vector and return true if successful.
+ */
+UBool UnicodeSet::allocateStrings(UErrorCode &status) {
+    if (U_FAILURE(status)) {
+        return false;
+    }
+    strings = new UVector(uprv_deleteUObject,
+                          uhash_compareUnicodeString, 1, status);
+    if (strings == nullptr) { // Check for memory allocation error.
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return false;
+    }
+    if (U_FAILURE(status)) {
+        delete strings;
+        strings = nullptr;
+        return false;
+    } 
+    return true;
+}
+
+int32_t UnicodeSet::nextCapacity(int32_t minCapacity) {
+    // Grow exponentially to reduce the frequency of allocations.
+    if (minCapacity < INITIAL_CAPACITY) {
+        return minCapacity + INITIAL_CAPACITY;
+    } else if (minCapacity <= 2500) {
+        return 5 * minCapacity;
+    } else {
+        int32_t newCapacity = 2 * minCapacity;
+        if (newCapacity > MAX_LENGTH) {
+            newCapacity = MAX_LENGTH;
+        }
+        return newCapacity;
+    }
+}
+
+bool UnicodeSet::ensureCapacity(int32_t newLen) {
+    if (newLen > MAX_LENGTH) {
+        newLen = MAX_LENGTH;
+    }
+    if (newLen <= capacity) {
+        return true;
+    }
+    int32_t newCapacity = nextCapacity(newLen);
+    UChar32* temp = (UChar32*) uprv_malloc(newCapacity * sizeof(UChar32));
+    if (temp == nullptr) {
+        setToBogus(); // set the object to bogus state if an OOM failure occurred.
+        return false;
+    }
+    // Copy only the actual contents.
+    uprv_memcpy(temp, list, len * sizeof(UChar32));
+    if (list != stackList) {
+        uprv_free(list);
+    }
+    list = temp;
+    capacity = newCapacity;
+    return true;
+}
+
+bool UnicodeSet::ensureBufferCapacity(int32_t newLen) {
+    if (newLen > MAX_LENGTH) {
+        newLen = MAX_LENGTH;
+    }
+    if (newLen <= bufferCapacity) {
+        return true;
+    }
+    int32_t newCapacity = nextCapacity(newLen);
+    UChar32* temp = (UChar32*) uprv_malloc(newCapacity * sizeof(UChar32));
+    if (temp == nullptr) {
+        setToBogus();
+        return false;
+    }
+    // The buffer has no contents to be copied.
+    // It is always filled from scratch after this call.
+    if (buffer != stackList) {
+        uprv_free(buffer);
+    }
+    buffer = temp;
+    bufferCapacity = newCapacity;
+    return true;
+}
+
+/**
+ * Swap list and buffer.
+ */
+void UnicodeSet::swapBuffers() {
+    // swap list and buffer
+    UChar32* temp = list;
+    list = buffer;
+    buffer = temp;
+
+    int32_t c = capacity;
+    capacity = bufferCapacity;
+    bufferCapacity = c;
+}
+
+void UnicodeSet::setToBogus() {
+    clear(); // Remove everything in the set.
+    fFlags = kIsBogus;
+}
+
+//----------------------------------------------------------------
+// Implementation: Fundamental operators
+//----------------------------------------------------------------
+
+static inline UChar32 max(UChar32 a, UChar32 b) {
+    return (a > b) ? a : b;
+}
+
+// polarity = 0, 3 is normal: x xor y
+// polarity = 1, 2: x xor ~y == x === y
+
+void UnicodeSet::exclusiveOr(const UChar32* other, int32_t otherLen, int8_t polarity) {
+    if (isFrozen() || isBogus()) {
+        return;
+    }
+    if (!ensureBufferCapacity(len + otherLen)) {
+        return;
+    }
+
+    int32_t i = 0, j = 0, k = 0;
+    UChar32 a = list[i++];
+    UChar32 b;
+    if (polarity == 1 || polarity == 2) {
+        b = UNICODESET_LOW;
+        if (other[j] == UNICODESET_LOW) { // skip base if already LOW
+            ++j;
+            b = other[j];
+        }
+    } else {
+        b = other[j++];
+    }
+    // simplest of all the routines
+    // sort the values, discarding identicals!
+    for (;;) {
+        if (a < b) {
+            buffer[k++] = a;
+            a = list[i++];
+        } else if (b < a) {
+            buffer[k++] = b;
+            b = other[j++];
+        } else if (a != UNICODESET_HIGH) { // at this point, a == b
+            // discard both values!
+            a = list[i++];
+            b = other[j++];
+        } else { // DONE!
+            buffer[k++] = UNICODESET_HIGH;
+            len = k;
+            break;
+        }
+    }
+    swapBuffers();
+    releasePattern();
+}
+
+// polarity = 0 is normal: x union y
+// polarity = 2: x union ~y
+// polarity = 1: ~x union y
+// polarity = 3: ~x union ~y
+
+void UnicodeSet::add(const UChar32* other, int32_t otherLen, int8_t polarity) {
+    if (isFrozen() || isBogus() || other==nullptr) {
+        return;
+    }
+    if (!ensureBufferCapacity(len + otherLen)) {
+        return;
+    }
+
+    int32_t i = 0, j = 0, k = 0;
+    UChar32 a = list[i++];
+    UChar32 b = other[j++];
+    // change from xor is that we have to check overlapping pairs
+    // polarity bit 1 means a is second, bit 2 means b is.
+    for (;;) {
+        switch (polarity) {
+          case 0: // both first; take lower if unequal
+            if (a < b) { // take a
+                // Back up over overlapping ranges in buffer[]
+                if (k > 0 && a <= buffer[k-1]) {
+                    // Pick latter end value in buffer[] vs. list[]
+                    a = max(list[i], buffer[--k]);
+                } else {
+                    // No overlap
+                    buffer[k++] = a;
+                    a = list[i];
+                }
+                i++; // Common if/else code factored out
+                polarity ^= 1;
+            } else if (b < a) { // take b
+                if (k > 0 && b <= buffer[k-1]) {
+                    b = max(other[j], buffer[--k]);
+                } else {
+                    buffer[k++] = b;
+                    b = other[j];
+                }
+                j++;
+                polarity ^= 2;
+            } else { // a == b, take a, drop b
+                if (a == UNICODESET_HIGH) goto loop_end;
+                // This is symmetrical; it doesn't matter if
+                // we backtrack with a or b. - liu
+                if (k > 0 && a <= buffer[k-1]) {
+                    a = max(list[i], buffer[--k]);
+                } else {
+                    // No overlap
+                    buffer[k++] = a;
+                    a = list[i];
+                }
+                i++;
+                polarity ^= 1;
+                b = other[j++];
+                polarity ^= 2;
+            }
+            break;
+          case 3: // both second; take higher if unequal, and drop other
+            if (b <= a) { // take a
+                if (a == UNICODESET_HIGH) goto loop_end;
+                buffer[k++] = a;
+            } else { // take b
+                if (b == UNICODESET_HIGH) goto loop_end;
+                buffer[k++] = b;
+            }
+            a = list[i++];
+            polarity ^= 1;   // factored common code
+            b = other[j++];
+            polarity ^= 2;
+            break;
+          case 1: // a second, b first; if b < a, overlap
+            if (a < b) { // no overlap, take a
+                buffer[k++] = a; a = list[i++]; polarity ^= 1;
+            } else if (b < a) { // OVERLAP, drop b
+                b = other[j++];
+                polarity ^= 2;
+            } else { // a == b, drop both!
+                if (a == UNICODESET_HIGH) goto loop_end;
+                a = list[i++];
+                polarity ^= 1;
+                b = other[j++];
+                polarity ^= 2;
+            }
+            break;
+          case 2: // a first, b second; if a < b, overlap
+            if (b < a) { // no overlap, take b
+                buffer[k++] = b;
+                b = other[j++];
+                polarity ^= 2;
+            } else  if (a < b) { // OVERLAP, drop a
+                a = list[i++];
+                polarity ^= 1;
+            } else { // a == b, drop both!
+                if (a == UNICODESET_HIGH) goto loop_end;
+                a = list[i++];
+                polarity ^= 1;
+                b = other[j++];
+                polarity ^= 2;
+            }
+            break;
+        }
+    }
+ loop_end:
+    buffer[k++] = UNICODESET_HIGH;    // terminate
+    len = k;
+    swapBuffers();
+    releasePattern();
+}
+
+// polarity = 0 is normal: x intersect y
+// polarity = 2: x intersect ~y == set-minus
+// polarity = 1: ~x intersect y
+// polarity = 3: ~x intersect ~y
+
+void UnicodeSet::retain(const UChar32* other, int32_t otherLen, int8_t polarity) {
+    if (isFrozen() || isBogus()) {
+        return;
+    }
+    if (!ensureBufferCapacity(len + otherLen)) {
+        return;
+    }
+
+    int32_t i = 0, j = 0, k = 0;
+    UChar32 a = list[i++];
+    UChar32 b = other[j++];
+    // change from xor is that we have to check overlapping pairs
+    // polarity bit 1 means a is second, bit 2 means b is.
+    for (;;) {
+        switch (polarity) {
+          case 0: // both first; drop the smaller
+            if (a < b) { // drop a
+                a = list[i++];
+                polarity ^= 1;
+            } else if (b < a) { // drop b
+                b = other[j++];
+                polarity ^= 2;
+            } else { // a == b, take one, drop other
+                if (a == UNICODESET_HIGH) goto loop_end;
+                buffer[k++] = a;
+                a = list[i++];
+                polarity ^= 1;
+                b = other[j++];
+                polarity ^= 2;
+            }
+            break;
+          case 3: // both second; take lower if unequal
+            if (a < b) { // take a
+                buffer[k++] = a;
+                a = list[i++];
+                polarity ^= 1;
+            } else if (b < a) { // take b
+                buffer[k++] = b;
+                b = other[j++];
+                polarity ^= 2;
+            } else { // a == b, take one, drop other
+                if (a == UNICODESET_HIGH) goto loop_end;
+                buffer[k++] = a;
+                a = list[i++];
+                polarity ^= 1;
+                b = other[j++];
+                polarity ^= 2;
+            }
+            break;
+          case 1: // a second, b first;
+            if (a < b) { // NO OVERLAP, drop a
+                a = list[i++];
+                polarity ^= 1;
+            } else if (b < a) { // OVERLAP, take b
+                buffer[k++] = b;
+                b = other[j++];
+                polarity ^= 2;
+            } else { // a == b, drop both!
+                if (a == UNICODESET_HIGH) goto loop_end;
+                a = list[i++];
+                polarity ^= 1;
+                b = other[j++];
+                polarity ^= 2;
+            }
+            break;
+          case 2: // a first, b second; if a < b, overlap
+            if (b < a) { // no overlap, drop b
+                b = other[j++];
+                polarity ^= 2;
+            } else  if (a < b) { // OVERLAP, take a
+                buffer[k++] = a;
+                a = list[i++];
+                polarity ^= 1;
+            } else { // a == b, drop both!
+                if (a == UNICODESET_HIGH) goto loop_end;
+                a = list[i++];
+                polarity ^= 1;
+                b = other[j++];
+                polarity ^= 2;
+            }
+            break;
+        }
+    }
+ loop_end:
+    buffer[k++] = UNICODESET_HIGH;    // terminate
+    len = k;
+    swapBuffers();
+    releasePattern();
+}
+
+/**
+ * Append the toPattern() representation of a
+ * string to the given StringBuffer.
+ */
+void UnicodeSet::_appendToPat(UnicodeString& buf, const UnicodeString& s, UBool escapeUnprintable) {
+    UChar32 cp;
+    for (int32_t i = 0; i < s.length(); i += U16_LENGTH(cp)) {
+        _appendToPat(buf, cp = s.char32At(i), escapeUnprintable);
+    }
+}
+
+/**
+ * Append the toPattern() representation of a
+ * character to the given StringBuffer.
+ */
+void UnicodeSet::_appendToPat(UnicodeString& buf, UChar32 c, UBool escapeUnprintable) {
+    if (escapeUnprintable ? ICU_Utility::isUnprintable(c) : ICU_Utility::shouldAlwaysBeEscaped(c)) {
+        // Use hex escape notation (\uxxxx or \Uxxxxxxxx) for anything
+        // unprintable
+        ICU_Utility::escape(buf, c);
+        return;
+    }
+    // Okay to let ':' pass through
+    switch (c) {
+    case u'[':
+    case u']':
+    case u'-':
+    case u'^':
+    case u'&':
+    case u'\\':
+    case u'{':
+    case u'}':
+    case u':':
+    case SymbolTable::SYMBOL_REF:
+        buf.append(u'\\');
+        break;
+    default:
+        // Escape whitespace
+        if (PatternProps::isWhiteSpace(c)) {
+            buf.append(u'\\');
+        }
+        break;
+    }
+    buf.append(c);
+}
+
+void UnicodeSet::_appendToPat(UnicodeString &result, UChar32 start, UChar32 end,
+                              UBool escapeUnprintable) {
+    _appendToPat(result, start, escapeUnprintable);
+    if (start != end) {
+        if ((start+1) != end ||
+                // Avoid writing what looks like a lead+trail surrogate pair.
+                start == 0xdbff) {
+            result.append(u'-');
+        }
+        _appendToPat(result, end, escapeUnprintable);
+    }
+}
+
+/**
+ * Append a string representation of this set to result.  This will be
+ * a cleaned version of the string passed to applyPattern(), if there
+ * is one.  Otherwise it will be generated.
+ */
+UnicodeString& UnicodeSet::_toPattern(UnicodeString& result,
+                                      UBool escapeUnprintable) const
+{
+    if (pat != nullptr) {
+        int32_t i;
+        int32_t backslashCount = 0;
+        for (i=0; i= 2 &&
+    //             getRangeStart(0) == MIN_VALUE &&
+    //             getRangeEnd(last) == MAX_VALUE)
+    // Invariant: list[len-1] == HIGH == MAX_VALUE + 1
+    // If limit == len then len is even and the last range ends with MAX_VALUE.
+    //
+    // *But* do not write the inverse (complement) if there are strings.
+    // Since ICU 70, the '^' performs a code point complement which removes all strings.
+    if (len >= 4 && list[0] == 0 && limit == len && !hasStrings()) {
+        // Emit the inverse
+        result.append(u'^');
+        // Offsetting the inversion list index by one lets us
+        // iterate over the ranges of the set complement.
+        i = 1;
+        --limit;
+    }
+
+    // Emit the ranges as pairs.
+    while (i < limit) {
+        UChar32 start = list[i];  // getRangeStart()
+        UChar32 end = list[i + 1] - 1;  // getRangeEnd() = range limit minus one
+        if (!(0xd800 <= end && end <= 0xdbff)) {
+            _appendToPat(result, start, end, escapeUnprintable);
+            i += 2;
+        } else {
+            // The range ends with a lead surrogate.
+            // Avoid writing what looks like a lead+trail surrogate pair.
+            // 1. Postpone ranges that start with a lead surrogate code point.
+            int32_t firstLead = i;
+            while ((i += 2) < limit && list[i] <= 0xdbff) {}
+            int32_t firstAfterLead = i;
+            // 2. Write following ranges that start with a trail surrogate code point.
+            while (i < limit && (start = list[i]) <= 0xdfff) {
+                _appendToPat(result, start, list[i + 1] - 1, escapeUnprintable);
+                i += 2;
+            }
+            // 3. Now write the postponed ranges.
+            for (int j = firstLead; j < firstAfterLead; j += 2) {
+                _appendToPat(result, list[j], list[j + 1] - 1, escapeUnprintable);
+            }
+        }
+    }
+
+    if (strings != nullptr) {
+        for (int32_t i = 0; isize(); ++i) {
+            result.append(u'{');
+            _appendToPat(result,
+                         *(const UnicodeString*) strings->elementAt(i),
+                         escapeUnprintable);
+            result.append(u'}');
+        }
+    }
+    return result.append(u']');
+}
+
+/**
+* Release existing cached pattern
+*/
+void UnicodeSet::releasePattern() {
+    if (pat) {
+        uprv_free(pat);
+        pat = nullptr;
+        patLen = 0;
+    }
+}
+
+/**
+* Set the new pattern to cache.
+*/
+void UnicodeSet::setPattern(const char16_t *newPat, int32_t newPatLen) {
+    releasePattern();
+    pat = (char16_t *)uprv_malloc((newPatLen + 1) * sizeof(char16_t));
+    if (pat) {
+        patLen = newPatLen;
+        u_memcpy(pat, newPat, patLen);
+        pat[patLen] = 0;
+    }
+    // else we don't care if malloc failed. This was just a nice cache.
+    // We can regenerate an equivalent pattern later when requested.
+}
+
+UnicodeSet *UnicodeSet::freeze() {
+    if(!isFrozen() && !isBogus()) {
+        compact();
+
+        // Optimize contains() and span() and similar functions.
+        if (hasStrings()) {
+            stringSpan = new UnicodeSetStringSpan(*this, *strings, UnicodeSetStringSpan::ALL);
+            if (stringSpan == nullptr) {
+                setToBogus();
+                return this;
+            } else if (!stringSpan->needsStringSpanUTF16()) {
+                // All strings are irrelevant for span() etc. because
+                // all of each string's code points are contained in this set.
+                // Do not check needsStringSpanUTF8() because UTF-8 has at most as
+                // many relevant strings as UTF-16.
+                // (Thus needsStringSpanUTF8() implies needsStringSpanUTF16().)
+                delete stringSpan;
+                stringSpan = nullptr;
+            }
+        }
+        if (stringSpan == nullptr) {
+            // No span-relevant strings: Optimize for code point spans.
+            bmpSet=new BMPSet(list, len);
+            if (bmpSet == nullptr) { // Check for memory allocation error.
+                setToBogus();
+            }
+        }
+    }
+    return this;
+}
+
+int32_t UnicodeSet::span(const char16_t *s, int32_t length, USetSpanCondition spanCondition) const {
+    if(length>0 && bmpSet!=nullptr) {
+        return (int32_t)(bmpSet->span(s, s+length, spanCondition)-s);
+    }
+    if(length<0) {
+        length=u_strlen(s);
+    }
+    if(length==0) {
+        return 0;
+    }
+    if(stringSpan!=nullptr) {
+        return stringSpan->span(s, length, spanCondition);
+    } else if(hasStrings()) {
+        uint32_t which= spanCondition==USET_SPAN_NOT_CONTAINED ?
+                            UnicodeSetStringSpan::FWD_UTF16_NOT_CONTAINED :
+                            UnicodeSetStringSpan::FWD_UTF16_CONTAINED;
+        UnicodeSetStringSpan strSpan(*this, *strings, which);
+        if(strSpan.needsStringSpanUTF16()) {
+            return strSpan.span(s, length, spanCondition);
+        }
+    }
+
+    if(spanCondition!=USET_SPAN_NOT_CONTAINED) {
+        spanCondition=USET_SPAN_CONTAINED;  // Pin to 0/1 values.
+    }
+
+    UChar32 c;
+    int32_t start=0, prev=0;
+    do {
+        U16_NEXT(s, start, length, c);
+        if(spanCondition!=contains(c)) {
+            break;
+        }
+    } while((prev=start)0 && bmpSet!=nullptr) {
+        return (int32_t)(bmpSet->spanBack(s, s+length, spanCondition)-s);
+    }
+    if(length<0) {
+        length=u_strlen(s);
+    }
+    if(length==0) {
+        return 0;
+    }
+    if(stringSpan!=nullptr) {
+        return stringSpan->spanBack(s, length, spanCondition);
+    } else if(hasStrings()) {
+        uint32_t which= spanCondition==USET_SPAN_NOT_CONTAINED ?
+                            UnicodeSetStringSpan::BACK_UTF16_NOT_CONTAINED :
+                            UnicodeSetStringSpan::BACK_UTF16_CONTAINED;
+        UnicodeSetStringSpan strSpan(*this, *strings, which);
+        if(strSpan.needsStringSpanUTF16()) {
+            return strSpan.spanBack(s, length, spanCondition);
+        }
+    }
+
+    if(spanCondition!=USET_SPAN_NOT_CONTAINED) {
+        spanCondition=USET_SPAN_CONTAINED;  // Pin to 0/1 values.
+    }
+
+    UChar32 c;
+    int32_t prev=length;
+    do {
+        U16_PREV(s, 0, length, c);
+        if(spanCondition!=contains(c)) {
+            break;
+        }
+    } while((prev=length)>0);
+    return prev;
+}
+
+int32_t UnicodeSet::spanUTF8(const char *s, int32_t length, USetSpanCondition spanCondition) const {
+    if(length>0 && bmpSet!=nullptr) {
+        const uint8_t *s0=(const uint8_t *)s;
+        return (int32_t)(bmpSet->spanUTF8(s0, length, spanCondition)-s0);
+    }
+    if(length<0) {
+        length=(int32_t)uprv_strlen(s);
+    }
+    if(length==0) {
+        return 0;
+    }
+    if(stringSpan!=nullptr) {
+        return stringSpan->spanUTF8((const uint8_t *)s, length, spanCondition);
+    } else if(hasStrings()) {
+        uint32_t which= spanCondition==USET_SPAN_NOT_CONTAINED ?
+                            UnicodeSetStringSpan::FWD_UTF8_NOT_CONTAINED :
+                            UnicodeSetStringSpan::FWD_UTF8_CONTAINED;
+        UnicodeSetStringSpan strSpan(*this, *strings, which);
+        if(strSpan.needsStringSpanUTF8()) {
+            return strSpan.spanUTF8((const uint8_t *)s, length, spanCondition);
+        }
+    }
+
+    if(spanCondition!=USET_SPAN_NOT_CONTAINED) {
+        spanCondition=USET_SPAN_CONTAINED;  // Pin to 0/1 values.
+    }
+
+    UChar32 c;
+    int32_t start=0, prev=0;
+    do {
+        U8_NEXT_OR_FFFD(s, start, length, c);
+        if(spanCondition!=contains(c)) {
+            break;
+        }
+    } while((prev=start)0 && bmpSet!=nullptr) {
+        const uint8_t *s0=(const uint8_t *)s;
+        return bmpSet->spanBackUTF8(s0, length, spanCondition);
+    }
+    if(length<0) {
+        length=(int32_t)uprv_strlen(s);
+    }
+    if(length==0) {
+        return 0;
+    }
+    if(stringSpan!=nullptr) {
+        return stringSpan->spanBackUTF8((const uint8_t *)s, length, spanCondition);
+    } else if(hasStrings()) {
+        uint32_t which= spanCondition==USET_SPAN_NOT_CONTAINED ?
+                            UnicodeSetStringSpan::BACK_UTF8_NOT_CONTAINED :
+                            UnicodeSetStringSpan::BACK_UTF8_CONTAINED;
+        UnicodeSetStringSpan strSpan(*this, *strings, which);
+        if(strSpan.needsStringSpanUTF8()) {
+            return strSpan.spanBackUTF8((const uint8_t *)s, length, spanCondition);
+        }
+    }
+
+    if(spanCondition!=USET_SPAN_NOT_CONTAINED) {
+        spanCondition=USET_SPAN_CONTAINED;  // Pin to 0/1 values.
+    }
+
+    UChar32 c;
+    int32_t prev=length;
+    do {
+        U8_PREV_OR_FFFD(s, 0, length, c);
+        if(spanCondition!=contains(c)) {
+            break;
+        }
+    } while((prev=length)>0);
+    return prev;
+}
+
+U_NAMESPACE_END
diff --git a/intl/icu/source/common/uniset_closure.cpp b/intl/icu/source/common/uniset_closure.cpp
new file mode 100644
index 0000000000..173a5cbaae
--- /dev/null
+++ b/intl/icu/source/common/uniset_closure.cpp
@@ -0,0 +1,358 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  uniset_closure.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2011may30
+*   created by: Markus W. Scherer
+*
+*   UnicodeSet::closeOver() and related methods moved here from uniset_props.cpp
+*   to simplify dependencies.
+*   In particular, this depends on the BreakIterator, but the BreakIterator
+*   code also builds UnicodeSets from patterns and needs uniset_props.
+*/
+
+#include "unicode/brkiter.h"
+#include "unicode/locid.h"
+#include "unicode/parsepos.h"
+#include "unicode/uniset.h"
+#include "unicode/utf16.h"
+#include "cmemory.h"
+#include "ruleiter.h"
+#include "ucase.h"
+#include "uprops.h"
+#include "util.h"
+#include "uvector.h"
+
+U_NAMESPACE_BEGIN
+
+// TODO memory debugging provided inside uniset.cpp
+// could be made available here but probably obsolete with use of modern
+// memory leak checker tools
+#define _dbgct(me)
+
+//----------------------------------------------------------------
+// Constructors &c
+//----------------------------------------------------------------
+
+UnicodeSet::UnicodeSet(const UnicodeString& pattern,
+                       uint32_t options,
+                       const SymbolTable* symbols,
+                       UErrorCode& status) {
+    applyPattern(pattern, options, symbols, status);
+    _dbgct(this);
+}
+
+UnicodeSet::UnicodeSet(const UnicodeString& pattern, ParsePosition& pos,
+                       uint32_t options,
+                       const SymbolTable* symbols,
+                       UErrorCode& status) {
+    applyPattern(pattern, pos, options, symbols, status);
+    _dbgct(this);
+}
+
+//----------------------------------------------------------------
+// Public API
+//----------------------------------------------------------------
+
+UnicodeSet& UnicodeSet::applyPattern(const UnicodeString& pattern,
+                                     uint32_t options,
+                                     const SymbolTable* symbols,
+                                     UErrorCode& status) {
+    ParsePosition pos(0);
+    applyPattern(pattern, pos, options, symbols, status);
+    if (U_FAILURE(status)) return *this;
+
+    int32_t i = pos.getIndex();
+
+    if (options & USET_IGNORE_SPACE) {
+        // Skip over trailing whitespace
+        ICU_Utility::skipWhitespace(pattern, i, true);
+    }
+
+    if (i != pattern.length()) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+    }
+    return *this;
+}
+
+UnicodeSet& UnicodeSet::applyPattern(const UnicodeString& pattern,
+                              ParsePosition& pos,
+                              uint32_t options,
+                              const SymbolTable* symbols,
+                              UErrorCode& status) {
+    if (U_FAILURE(status)) {
+        return *this;
+    }
+    if (isFrozen()) {
+        status = U_NO_WRITE_PERMISSION;
+        return *this;
+    }
+    // Need to build the pattern in a temporary string because
+    // _applyPattern calls add() etc., which set pat to empty.
+    UnicodeString rebuiltPat;
+    RuleCharacterIterator chars(pattern, symbols, pos);
+    applyPattern(chars, symbols, rebuiltPat, options, &UnicodeSet::closeOver, 0, status);
+    if (U_FAILURE(status)) return *this;
+    if (chars.inVariable()) {
+        // syntaxError(chars, "Extra chars in variable value");
+        status = U_MALFORMED_SET;
+        return *this;
+    }
+    setPattern(rebuiltPat);
+    return *this;
+}
+
+// USetAdder implementation
+// Does not use uset.h to reduce code dependencies
+static void U_CALLCONV
+_set_add(USet *set, UChar32 c) {
+    ((UnicodeSet *)set)->add(c);
+}
+
+static void U_CALLCONV
+_set_addRange(USet *set, UChar32 start, UChar32 end) {
+    ((UnicodeSet *)set)->add(start, end);
+}
+
+static void U_CALLCONV
+_set_addString(USet *set, const char16_t *str, int32_t length) {
+    ((UnicodeSet *)set)->add(UnicodeString((UBool)(length<0), str, length));
+}
+
+//----------------------------------------------------------------
+// Case folding API
+//----------------------------------------------------------------
+
+// add the result of a full case mapping to the set
+// use str as a temporary string to avoid constructing one
+static inline void
+addCaseMapping(UnicodeSet &set, int32_t result, const char16_t *full, UnicodeString &str) {
+    if(result >= 0) {
+        if(result > UCASE_MAX_STRING_LENGTH) {
+            // add a single-code point case mapping
+            set.add(result);
+        } else {
+            // add a string case mapping from full with length result
+            str.setTo((UBool)false, full, result);
+            set.add(str);
+        }
+    }
+    // result < 0: the code point mapped to itself, no need to add it
+    // see ucase.h
+}
+
+namespace {
+
+/** For case closure on a large set, look only at code points with relevant properties. */
+const UnicodeSet &maybeOnlyCaseSensitive(const UnicodeSet &src, UnicodeSet &subset) {
+    // The subset must have been constructed with all code points,
+    // so that the retainAll() intersection effectively copies all single code points from src.
+    U_ASSERT(subset.contains(0, 0x10ffff));
+    if (src.size() < 30) {
+        return src;
+    }
+    // Return the intersection of the src code points with Case_Sensitive ones.
+    UErrorCode errorCode = U_ZERO_ERROR;
+    const UnicodeSet *sensitive =
+        CharacterProperties::getBinaryPropertySet(UCHAR_CASE_SENSITIVE, errorCode);
+    if (U_FAILURE(errorCode)) {
+        return src;
+    }
+    // Start by copying the "smaller" set.
+    // (We "copy" by intersecting all Unicode *code points* with the first set,
+    // which omits any strings.)
+    if (src.getRangeCount() > sensitive->getRangeCount()) {
+        subset.retainAll(*sensitive);
+        subset.retainAll(src);
+    } else {
+        subset.retainAll(src);
+        subset.retainAll(*sensitive);
+    }
+    return subset;
+}
+
+// Per-character scf = Simple_Case_Folding of a string.
+// (Normally when we case-fold a string we use full case foldings.)
+bool scfString(const UnicodeString &s, UnicodeString &scf) {
+    // Iterate over the raw buffer for best performance.
+    const char16_t *p = s.getBuffer();
+    int32_t length = s.length();
+    // Loop while not needing modification.
+    for (int32_t i = 0; i < length;) {
+        UChar32 c;
+        U16_NEXT(p, i, length, c);  // post-increments i
+        UChar32 scfChar = u_foldCase(c, U_FOLD_CASE_DEFAULT);
+        if (scfChar != c) {
+            // Copy the characters before c.
+            scf.setTo(p, i - U16_LENGTH(c));
+            // Loop over the rest of the string and keep case-folding.
+            for (;;) {
+                scf.append(scfChar);
+                if (i == length) {
+                    return true;
+                }
+                U16_NEXT(p, i, length, c);  // post-increments i
+                scfChar = u_foldCase(c, U_FOLD_CASE_DEFAULT);
+            }
+        }
+    }
+    return false;
+}
+
+}  // namespace
+
+UnicodeSet& UnicodeSet::closeOver(int32_t attribute) {
+    if (isFrozen() || isBogus()) {
+        return *this;
+    }
+    switch (attribute & USET_CASE_MASK) {
+    case 0:
+        break;
+    case USET_CASE_INSENSITIVE:
+        closeOverCaseInsensitive(/* simple= */ false);
+        break;
+    case USET_ADD_CASE_MAPPINGS:
+        closeOverAddCaseMappings();
+        break;
+    case USET_SIMPLE_CASE_INSENSITIVE:
+        closeOverCaseInsensitive(/* simple= */ true);
+        break;
+    default:
+        // bad option (unreachable)
+        break;
+    }
+    return *this;
+}
+
+void UnicodeSet::closeOverCaseInsensitive(bool simple) {
+    // Start with input set to guarantee inclusion.
+    UnicodeSet foldSet(*this);
+    // Full case mappings closure:
+    // Remove strings because the strings will actually be reduced (folded);
+    // therefore, start with no strings and add only those needed.
+    // Do this before processing code points, because they may add strings.
+    if (!simple && foldSet.hasStrings()) {
+        foldSet.strings->removeAllElements();
+    }
+
+    USetAdder sa = {
+        foldSet.toUSet(),
+        _set_add,
+        _set_addRange,
+        _set_addString,
+        nullptr, // don't need remove()
+        nullptr // don't need removeRange()
+    };
+
+    UnicodeSet subset(0, 0x10ffff);
+    const UnicodeSet &codePoints = maybeOnlyCaseSensitive(*this, subset);
+
+    // Iterate over the ranges of single code points. Nested loop for each code point.
+    int32_t n = codePoints.getRangeCount();
+
+    for (int32_t i=0; isize(); ++j) {
+            const UnicodeString *pStr = (const UnicodeString *) strings->elementAt(j);
+            if (simple) {
+                if (scfString(*pStr, str)) {
+                    foldSet.remove(*pStr).add(str);
+                }
+            } else {
+                str = *pStr;
+                str.foldCase();
+                if(!ucase_addStringCaseClosure(str.getBuffer(), str.length(), &sa)) {
+                    foldSet.add(str); // does not map to code points: add the folded string itself
+                }
+            }
+        }
+    }
+    *this = foldSet;
+}
+
+void UnicodeSet::closeOverAddCaseMappings() {
+    // Start with input set to guarantee inclusion.
+    UnicodeSet foldSet(*this);
+
+    UnicodeSet subset(0, 0x10ffff);
+    const UnicodeSet &codePoints = maybeOnlyCaseSensitive(*this, subset);
+
+    // Iterate over the ranges of single code points. Nested loop for each code point.
+    int32_t n = codePoints.getRangeCount();
+    UChar32 result;
+    const char16_t *full;
+    UnicodeString str;
+
+    for (int32_t i=0; isize(); ++j) {
+                const UnicodeString *pStr = (const UnicodeString *) strings->elementAt(j);
+                (str = *pStr).toLower(root);
+                foldSet.add(str);
+#if !UCONFIG_NO_BREAK_ITERATION
+                (str = *pStr).toTitle(bi, root);
+                foldSet.add(str);
+#endif
+                (str = *pStr).toUpper(root);
+                foldSet.add(str);
+                (str = *pStr).foldCase();
+                foldSet.add(str);
+            }
+#if !UCONFIG_NO_BREAK_ITERATION
+        }
+        delete bi;
+#endif
+    }
+    *this = foldSet;
+}
+
+U_NAMESPACE_END
diff --git a/intl/icu/source/common/uniset_props.cpp b/intl/icu/source/common/uniset_props.cpp
new file mode 100644
index 0000000000..b3dbdf93c8
--- /dev/null
+++ b/intl/icu/source/common/uniset_props.cpp
@@ -0,0 +1,1139 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 1999-2014, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  uniset_props.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2004aug25
+*   created by: Markus W. Scherer
+*
+*   Character property dependent functions moved here from uniset.cpp
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/uniset.h"
+#include "unicode/parsepos.h"
+#include "unicode/uchar.h"
+#include "unicode/uscript.h"
+#include "unicode/symtable.h"
+#include "unicode/uset.h"
+#include "unicode/locid.h"
+#include "unicode/brkiter.h"
+#include "uset_imp.h"
+#include "ruleiter.h"
+#include "cmemory.h"
+#include "ucln_cmn.h"
+#include "util.h"
+#include "uvector.h"
+#include "uprops.h"
+#include "propname.h"
+#include "normalizer2impl.h"
+#include "uinvchar.h"
+#include "uprops.h"
+#include "charstr.h"
+#include "cstring.h"
+#include "mutex.h"
+#include "umutex.h"
+#include "uassert.h"
+#include "hash.h"
+
+U_NAMESPACE_USE
+
+// Special property set IDs
+static const char ANY[]   = "ANY";   // [\u0000-\U0010FFFF]
+static const char ASCII[] = "ASCII"; // [\u0000-\u007F]
+static const char ASSIGNED[] = "Assigned"; // [:^Cn:]
+
+// Unicode name property alias
+#define NAME_PROP "na"
+#define NAME_PROP_LENGTH 2
+
+// Cached sets ------------------------------------------------------------- ***
+
+U_CDECL_BEGIN
+static UBool U_CALLCONV uset_cleanup();
+
+static UnicodeSet *uni32Singleton;
+static icu::UInitOnce uni32InitOnce {};
+
+/**
+ * Cleanup function for UnicodeSet
+ */
+static UBool U_CALLCONV uset_cleanup() {
+    delete uni32Singleton;
+    uni32Singleton = nullptr;
+    uni32InitOnce.reset();
+    return true;
+}
+
+U_CDECL_END
+
+U_NAMESPACE_BEGIN
+
+namespace {
+
+// Cache some sets for other services -------------------------------------- ***
+void U_CALLCONV createUni32Set(UErrorCode &errorCode) {
+    U_ASSERT(uni32Singleton == nullptr);
+    uni32Singleton = new UnicodeSet(UNICODE_STRING_SIMPLE("[:age=3.2:]"), errorCode);
+    if(uni32Singleton==nullptr) {
+        errorCode=U_MEMORY_ALLOCATION_ERROR;
+    } else {
+        uni32Singleton->freeze();
+    }
+    ucln_common_registerCleanup(UCLN_COMMON_USET, uset_cleanup);
+}
+
+
+U_CFUNC UnicodeSet *
+uniset_getUnicode32Instance(UErrorCode &errorCode) {
+    umtx_initOnce(uni32InitOnce, &createUni32Set, errorCode);
+    return uni32Singleton;
+}
+
+// helper functions for matching of pattern syntax pieces ------------------ ***
+// these functions are parallel to the PERL_OPEN etc. strings above
+
+// using these functions is not only faster than UnicodeString::compare() and
+// caseCompare(), but they also make UnicodeSet work for simple patterns when
+// no Unicode properties data is available - when caseCompare() fails
+
+static inline UBool
+isPerlOpen(const UnicodeString &pattern, int32_t pos) {
+    char16_t c;
+    return pattern.charAt(pos)==u'\\' && ((c=pattern.charAt(pos+1))==u'p' || c==u'P');
+}
+
+/*static inline UBool
+isPerlClose(const UnicodeString &pattern, int32_t pos) {
+    return pattern.charAt(pos)==u'}';
+}*/
+
+static inline UBool
+isNameOpen(const UnicodeString &pattern, int32_t pos) {
+    return pattern.charAt(pos)==u'\\' && pattern.charAt(pos+1)==u'N';
+}
+
+static inline UBool
+isPOSIXOpen(const UnicodeString &pattern, int32_t pos) {
+    return pattern.charAt(pos)==u'[' && pattern.charAt(pos+1)==u':';
+}
+
+/*static inline UBool
+isPOSIXClose(const UnicodeString &pattern, int32_t pos) {
+    return pattern.charAt(pos)==u':' && pattern.charAt(pos+1)==u']';
+}*/
+
+// TODO memory debugging provided inside uniset.cpp
+// could be made available here but probably obsolete with use of modern
+// memory leak checker tools
+#define _dbgct(me)
+
+}  // namespace
+
+//----------------------------------------------------------------
+// Constructors &c
+//----------------------------------------------------------------
+
+/**
+ * Constructs a set from the given pattern, optionally ignoring
+ * white space.  See the class description for the syntax of the
+ * pattern language.
+ * @param pattern a string specifying what characters are in the set
+ */
+UnicodeSet::UnicodeSet(const UnicodeString& pattern,
+                       UErrorCode& status) {
+    applyPattern(pattern, status);
+    _dbgct(this);
+}
+
+//----------------------------------------------------------------
+// Public API
+//----------------------------------------------------------------
+
+UnicodeSet& UnicodeSet::applyPattern(const UnicodeString& pattern,
+                                     UErrorCode& status) {
+    // Equivalent to
+    //   return applyPattern(pattern, USET_IGNORE_SPACE, nullptr, status);
+    // but without dependency on closeOver().
+    ParsePosition pos(0);
+    applyPatternIgnoreSpace(pattern, pos, nullptr, status);
+    if (U_FAILURE(status)) return *this;
+
+    int32_t i = pos.getIndex();
+    // Skip over trailing whitespace
+    ICU_Utility::skipWhitespace(pattern, i, true);
+    if (i != pattern.length()) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+    }
+    return *this;
+}
+
+void
+UnicodeSet::applyPatternIgnoreSpace(const UnicodeString& pattern,
+                                    ParsePosition& pos,
+                                    const SymbolTable* symbols,
+                                    UErrorCode& status) {
+    if (U_FAILURE(status)) {
+        return;
+    }
+    if (isFrozen()) {
+        status = U_NO_WRITE_PERMISSION;
+        return;
+    }
+    // Need to build the pattern in a temporary string because
+    // _applyPattern calls add() etc., which set pat to empty.
+    UnicodeString rebuiltPat;
+    RuleCharacterIterator chars(pattern, symbols, pos);
+    applyPattern(chars, symbols, rebuiltPat, USET_IGNORE_SPACE, nullptr, 0, status);
+    if (U_FAILURE(status)) return;
+    if (chars.inVariable()) {
+        // syntaxError(chars, "Extra chars in variable value");
+        status = U_MALFORMED_SET;
+        return;
+    }
+    setPattern(rebuiltPat);
+}
+
+/**
+ * Return true if the given position, in the given pattern, appears
+ * to be the start of a UnicodeSet pattern.
+ */
+UBool UnicodeSet::resemblesPattern(const UnicodeString& pattern, int32_t pos) {
+    return ((pos+1) < pattern.length() &&
+            pattern.charAt(pos) == (char16_t)91/*[*/) ||
+        resemblesPropertyPattern(pattern, pos);
+}
+
+//----------------------------------------------------------------
+// Implementation: Pattern parsing
+//----------------------------------------------------------------
+
+namespace {
+
+/**
+ * A small all-inline class to manage a UnicodeSet pointer.  Add
+ * operator->() etc. as needed.
+ */
+class UnicodeSetPointer {
+    UnicodeSet* p;
+public:
+    inline UnicodeSetPointer() : p(0) {}
+    inline ~UnicodeSetPointer() { delete p; }
+    inline UnicodeSet* pointer() { return p; }
+    inline UBool allocate() {
+        if (p == 0) {
+            p = new UnicodeSet();
+        }
+        return p != 0;
+    }
+};
+
+constexpr int32_t MAX_DEPTH = 100;
+
+}  // namespace
+
+/**
+ * Parse the pattern from the given RuleCharacterIterator.  The
+ * iterator is advanced over the parsed pattern.
+ * @param chars iterator over the pattern characters.  Upon return
+ * it will be advanced to the first character after the parsed
+ * pattern, or the end of the iteration if all characters are
+ * parsed.
+ * @param symbols symbol table to use to parse and dereference
+ * variables, or null if none.
+ * @param rebuiltPat the pattern that was parsed, rebuilt or
+ * copied from the input pattern, as appropriate.
+ * @param options a bit mask of zero or more of the following:
+ * IGNORE_SPACE, CASE.
+ */
+void UnicodeSet::applyPattern(RuleCharacterIterator& chars,
+                              const SymbolTable* symbols,
+                              UnicodeString& rebuiltPat,
+                              uint32_t options,
+                              UnicodeSet& (UnicodeSet::*caseClosure)(int32_t attribute),
+                              int32_t depth,
+                              UErrorCode& ec) {
+    if (U_FAILURE(ec)) return;
+    if (depth > MAX_DEPTH) {
+        ec = U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+
+    // Syntax characters: [ ] ^ - & { }
+
+    // Recognized special forms for chars, sets: c-c s-s s&s
+
+    int32_t opts = RuleCharacterIterator::PARSE_VARIABLES |
+                   RuleCharacterIterator::PARSE_ESCAPES;
+    if ((options & USET_IGNORE_SPACE) != 0) {
+        opts |= RuleCharacterIterator::SKIP_WHITESPACE;
+    }
+
+    UnicodeString patLocal, buf;
+    UBool usePat = false;
+    UnicodeSetPointer scratch;
+    RuleCharacterIterator::Pos backup;
+
+    // mode: 0=before [, 1=between [...], 2=after ]
+    // lastItem: 0=none, 1=char, 2=set
+    int8_t lastItem = 0, mode = 0;
+    UChar32 lastChar = 0;
+    char16_t op = 0;
+
+    UBool invert = false;
+
+    clear();
+
+    while (mode != 2 && !chars.atEnd()) {
+        U_ASSERT((lastItem == 0 && op == 0) ||
+                 (lastItem == 1 && (op == 0 || op == u'-')) ||
+                 (lastItem == 2 && (op == 0 || op == u'-' || op == u'&')));
+
+        UChar32 c = 0;
+        UBool literal = false;
+        UnicodeSet* nested = 0; // alias - do not delete
+
+        // -------- Check for property pattern
+
+        // setMode: 0=none, 1=unicodeset, 2=propertypat, 3=preparsed
+        int8_t setMode = 0;
+        if (resemblesPropertyPattern(chars, opts)) {
+            setMode = 2;
+        }
+
+        // -------- Parse '[' of opening delimiter OR nested set.
+        // If there is a nested set, use `setMode' to define how
+        // the set should be parsed.  If the '[' is part of the
+        // opening delimiter for this pattern, parse special
+        // strings "[", "[^", "[-", and "[^-".  Check for stand-in
+        // characters representing a nested set in the symbol
+        // table.
+
+        else {
+            // Prepare to backup if necessary
+            chars.getPos(backup);
+            c = chars.next(opts, literal, ec);
+            if (U_FAILURE(ec)) return;
+
+            if (c == u'[' && !literal) {
+                if (mode == 1) {
+                    chars.setPos(backup); // backup
+                    setMode = 1;
+                } else {
+                    // Handle opening '[' delimiter
+                    mode = 1;
+                    patLocal.append(u'[');
+                    chars.getPos(backup); // prepare to backup
+                    c = chars.next(opts, literal, ec); 
+                    if (U_FAILURE(ec)) return;
+                    if (c == u'^' && !literal) {
+                        invert = true;
+                        patLocal.append(u'^');
+                        chars.getPos(backup); // prepare to backup
+                        c = chars.next(opts, literal, ec);
+                        if (U_FAILURE(ec)) return;
+                    }
+                    // Fall through to handle special leading '-';
+                    // otherwise restart loop for nested [], \p{}, etc.
+                    if (c == u'-') {
+                        literal = true;
+                        // Fall through to handle literal '-' below
+                    } else {
+                        chars.setPos(backup); // backup
+                        continue;
+                    }
+                }
+            } else if (symbols != 0) {
+                const UnicodeFunctor *m = symbols->lookupMatcher(c);
+                if (m != 0) {
+                    const UnicodeSet *ms = dynamic_cast(m);
+                    if (ms == nullptr) {
+                        ec = U_MALFORMED_SET;
+                        return;
+                    }
+                    // casting away const, but `nested' won't be modified
+                    // (important not to modify stored set)
+                    nested = const_cast(ms);
+                    setMode = 3;
+                }
+            }
+        }
+
+        // -------- Handle a nested set.  This either is inline in
+        // the pattern or represented by a stand-in that has
+        // previously been parsed and was looked up in the symbol
+        // table.
+
+        if (setMode != 0) {
+            if (lastItem == 1) {
+                if (op != 0) {
+                    // syntaxError(chars, "Char expected after operator");
+                    ec = U_MALFORMED_SET;
+                    return;
+                }
+                add(lastChar, lastChar);
+                _appendToPat(patLocal, lastChar, false);
+                lastItem = 0;
+                op = 0;
+            }
+
+            if (op == u'-' || op == u'&') {
+                patLocal.append(op);
+            }
+
+            if (nested == 0) {
+                // lazy allocation
+                if (!scratch.allocate()) {
+                    ec = U_MEMORY_ALLOCATION_ERROR;
+                    return;
+                }
+                nested = scratch.pointer();
+            }
+            switch (setMode) {
+            case 1:
+                nested->applyPattern(chars, symbols, patLocal, options, caseClosure, depth + 1, ec);
+                break;
+            case 2:
+                chars.skipIgnored(opts);
+                nested->applyPropertyPattern(chars, patLocal, ec);
+                if (U_FAILURE(ec)) return;
+                break;
+            case 3: // `nested' already parsed
+                nested->_toPattern(patLocal, false);
+                break;
+            }
+
+            usePat = true;
+
+            if (mode == 0) {
+                // Entire pattern is a category; leave parse loop
+                *this = *nested;
+                mode = 2;
+                break;
+            }
+
+            switch (op) {
+            case u'-':
+                removeAll(*nested);
+                break;
+            case u'&':
+                retainAll(*nested);
+                break;
+            case 0:
+                addAll(*nested);
+                break;
+            }
+
+            op = 0;
+            lastItem = 2;
+
+            continue;
+        }
+
+        if (mode == 0) {
+            // syntaxError(chars, "Missing '['");
+            ec = U_MALFORMED_SET;
+            return;
+        }
+
+        // -------- Parse special (syntax) characters.  If the
+        // current character is not special, or if it is escaped,
+        // then fall through and handle it below.
+
+        if (!literal) {
+            switch (c) {
+            case u']':
+                if (lastItem == 1) {
+                    add(lastChar, lastChar);
+                    _appendToPat(patLocal, lastChar, false);
+                }
+                // Treat final trailing '-' as a literal
+                if (op == u'-') {
+                    add(op, op);
+                    patLocal.append(op);
+                } else if (op == u'&') {
+                    // syntaxError(chars, "Trailing '&'");
+                    ec = U_MALFORMED_SET;
+                    return;
+                }
+                patLocal.append(u']');
+                mode = 2;
+                continue;
+            case u'-':
+                if (op == 0) {
+                    if (lastItem != 0) {
+                        op = (char16_t) c;
+                        continue;
+                    } else {
+                        // Treat final trailing '-' as a literal
+                        add(c, c);
+                        c = chars.next(opts, literal, ec);
+                        if (U_FAILURE(ec)) return;
+                        if (c == u']' && !literal) {
+                            patLocal.append(u"-]", 2);
+                            mode = 2;
+                            continue;
+                        }
+                    }
+                }
+                // syntaxError(chars, "'-' not after char or set");
+                ec = U_MALFORMED_SET;
+                return;
+            case u'&':
+                if (lastItem == 2 && op == 0) {
+                    op = (char16_t) c;
+                    continue;
+                }
+                // syntaxError(chars, "'&' not after set");
+                ec = U_MALFORMED_SET;
+                return;
+            case u'^':
+                // syntaxError(chars, "'^' not after '['");
+                ec = U_MALFORMED_SET;
+                return;
+            case u'{':
+                if (op != 0) {
+                    // syntaxError(chars, "Missing operand after operator");
+                    ec = U_MALFORMED_SET;
+                    return;
+                }
+                if (lastItem == 1) {
+                    add(lastChar, lastChar);
+                    _appendToPat(patLocal, lastChar, false);
+                }
+                lastItem = 0;
+                buf.truncate(0);
+                {
+                    UBool ok = false;
+                    while (!chars.atEnd()) {
+                        c = chars.next(opts, literal, ec);
+                        if (U_FAILURE(ec)) return;
+                        if (c == u'}' && !literal) {
+                            ok = true;
+                            break;
+                        }
+                        buf.append(c);
+                    }
+                    if (!ok) {
+                        // syntaxError(chars, "Invalid multicharacter string");
+                        ec = U_MALFORMED_SET;
+                        return;
+                    }
+                }
+                // We have new string. Add it to set and continue;
+                // we don't need to drop through to the further
+                // processing
+                add(buf);
+                patLocal.append(u'{');
+                _appendToPat(patLocal, buf, false);
+                patLocal.append(u'}');
+                continue;
+            case SymbolTable::SYMBOL_REF:
+                //         symbols  nosymbols
+                // [a-$]   error    error (ambiguous)
+                // [a$]    anchor   anchor
+                // [a-$x]  var "x"* literal '$'
+                // [a-$.]  error    literal '$'
+                // *We won't get here in the case of var "x"
+                {
+                    chars.getPos(backup);
+                    c = chars.next(opts, literal, ec);
+                    if (U_FAILURE(ec)) return;
+                    UBool anchor = (c == u']' && !literal);
+                    if (symbols == 0 && !anchor) {
+                        c = SymbolTable::SYMBOL_REF;
+                        chars.setPos(backup);
+                        break; // literal '$'
+                    }
+                    if (anchor && op == 0) {
+                        if (lastItem == 1) {
+                            add(lastChar, lastChar);
+                            _appendToPat(patLocal, lastChar, false);
+                        }
+                        add(U_ETHER);
+                        usePat = true;
+                        patLocal.append((char16_t) SymbolTable::SYMBOL_REF);
+                        patLocal.append(u']');
+                        mode = 2;
+                        continue;
+                    }
+                    // syntaxError(chars, "Unquoted '$'");
+                    ec = U_MALFORMED_SET;
+                    return;
+                }
+            default:
+                break;
+            }
+        }
+
+        // -------- Parse literal characters.  This includes both
+        // escaped chars ("\u4E01") and non-syntax characters
+        // ("a").
+
+        switch (lastItem) {
+        case 0:
+            lastItem = 1;
+            lastChar = c;
+            break;
+        case 1:
+            if (op == u'-') {
+                if (lastChar >= c) {
+                    // Don't allow redundant (a-a) or empty (b-a) ranges;
+                    // these are most likely typos.
+                    // syntaxError(chars, "Invalid range");
+                    ec = U_MALFORMED_SET;
+                    return;
+                }
+                add(lastChar, c);
+                _appendToPat(patLocal, lastChar, false);
+                patLocal.append(op);
+                _appendToPat(patLocal, c, false);
+                lastItem = 0;
+                op = 0;
+            } else {
+                add(lastChar, lastChar);
+                _appendToPat(patLocal, lastChar, false);
+                lastChar = c;
+            }
+            break;
+        case 2:
+            if (op != 0) {
+                // syntaxError(chars, "Set expected after operator");
+                ec = U_MALFORMED_SET;
+                return;
+            }
+            lastChar = c;
+            lastItem = 1;
+            break;
+        }
+    }
+
+    if (mode != 2) {
+        // syntaxError(chars, "Missing ']'");
+        ec = U_MALFORMED_SET;
+        return;
+    }
+
+    chars.skipIgnored(opts);
+
+    /**
+     * Handle global flags (invert, case insensitivity).  If this
+     * pattern should be compiled case-insensitive, then we need
+     * to close over case BEFORE COMPLEMENTING.  This makes
+     * patterns like /[^abc]/i work.
+     */
+    if ((options & USET_CASE_MASK) != 0) {
+        (this->*caseClosure)(options);
+    }
+    if (invert) {
+        complement().removeAllStrings();  // code point complement
+    }
+
+    // Use the rebuilt pattern (patLocal) only if necessary.  Prefer the
+    // generated pattern.
+    if (usePat) {
+        rebuiltPat.append(patLocal);
+    } else {
+        _generatePattern(rebuiltPat, false);
+    }
+    if (isBogus() && U_SUCCESS(ec)) {
+        // We likely ran out of memory. AHHH!
+        ec = U_MEMORY_ALLOCATION_ERROR;
+    }
+}
+
+//----------------------------------------------------------------
+// Property set implementation
+//----------------------------------------------------------------
+
+namespace {
+
+static UBool numericValueFilter(UChar32 ch, void* context) {
+    return u_getNumericValue(ch) == *(double*)context;
+}
+
+static UBool generalCategoryMaskFilter(UChar32 ch, void* context) {
+    int32_t value = *(int32_t*)context;
+    return (U_GET_GC_MASK((UChar32) ch) & value) != 0;
+}
+
+static UBool versionFilter(UChar32 ch, void* context) {
+    static const UVersionInfo none = { 0, 0, 0, 0 };
+    UVersionInfo v;
+    u_charAge(ch, v);
+    UVersionInfo* version = (UVersionInfo*)context;
+    return uprv_memcmp(&v, &none, sizeof(v)) > 0 && uprv_memcmp(&v, version, sizeof(v)) <= 0;
+}
+
+typedef struct {
+    UProperty prop;
+    int32_t value;
+} IntPropertyContext;
+
+static UBool intPropertyFilter(UChar32 ch, void* context) {
+    IntPropertyContext* c = (IntPropertyContext*)context;
+    return u_getIntPropertyValue((UChar32) ch, c->prop) == c->value;
+}
+
+static UBool scriptExtensionsFilter(UChar32 ch, void* context) {
+    return uscript_hasScript(ch, *(UScriptCode*)context);
+}
+
+}  // namespace
+
+/**
+ * Generic filter-based scanning code for UCD property UnicodeSets.
+ */
+void UnicodeSet::applyFilter(UnicodeSet::Filter filter,
+                             void* context,
+                             const UnicodeSet* inclusions,
+                             UErrorCode &status) {
+    if (U_FAILURE(status)) return;
+
+    // Logically, walk through all Unicode characters, noting the start
+    // and end of each range for which filter.contain(c) is
+    // true.  Add each range to a set.
+    //
+    // To improve performance, use an inclusions set which
+    // encodes information about character ranges that are known
+    // to have identical properties.
+    // inclusions contains the first characters of
+    // same-value ranges for the given property.
+
+    clear();
+
+    UChar32 startHasProperty = -1;
+    int32_t limitRange = inclusions->getRangeCount();
+
+    for (int j=0; jgetRangeStart(j);
+        UChar32 end = inclusions->getRangeEnd(j);
+
+        // for all the code points in the range, process
+        for (UChar32 ch = start; ch <= end; ++ch) {
+            // only add to this UnicodeSet on inflection points --
+            // where the hasProperty value changes to false
+            if ((*filter)(ch, context)) {
+                if (startHasProperty < 0) {
+                    startHasProperty = ch;
+                }
+            } else if (startHasProperty >= 0) {
+                add(startHasProperty, ch-1);
+                startHasProperty = -1;
+            }
+        }
+    }
+    if (startHasProperty >= 0) {
+        add((UChar32)startHasProperty, (UChar32)0x10FFFF);
+    }
+    if (isBogus() && U_SUCCESS(status)) {
+        // We likely ran out of memory. AHHH!
+        status = U_MEMORY_ALLOCATION_ERROR;
+    }
+}
+
+namespace {
+
+static UBool mungeCharName(char* dst, const char* src, int32_t dstCapacity) {
+    /* Note: we use ' ' in compiler code page */
+    int32_t j = 0;
+    char ch;
+    --dstCapacity; /* make room for term. zero */
+    while ((ch = *src++) != 0) {
+        if (ch == ' ' && (j==0 || (j>0 && dst[j-1]==' '))) {
+            continue;
+        }
+        if (j >= dstCapacity) return false;
+        dst[j++] = ch;
+    }
+    if (j > 0 && dst[j-1] == ' ') --j;
+    dst[j] = 0;
+    return true;
+}
+
+}  // namespace
+
+//----------------------------------------------------------------
+// Property set API
+//----------------------------------------------------------------
+
+#define FAIL(ec) UPRV_BLOCK_MACRO_BEGIN { \
+    ec=U_ILLEGAL_ARGUMENT_ERROR; \
+    return *this; \
+} UPRV_BLOCK_MACRO_END
+
+UnicodeSet&
+UnicodeSet::applyIntPropertyValue(UProperty prop, int32_t value, UErrorCode& ec) {
+    if (U_FAILURE(ec) || isFrozen()) { return *this; }
+    if (prop == UCHAR_GENERAL_CATEGORY_MASK) {
+        const UnicodeSet* inclusions = CharacterProperties::getInclusionsForProperty(prop, ec);
+        applyFilter(generalCategoryMaskFilter, &value, inclusions, ec);
+    } else if (prop == UCHAR_SCRIPT_EXTENSIONS) {
+        const UnicodeSet* inclusions = CharacterProperties::getInclusionsForProperty(prop, ec);
+        UScriptCode script = (UScriptCode)value;
+        applyFilter(scriptExtensionsFilter, &script, inclusions, ec);
+    } else if (0 <= prop && prop < UCHAR_BINARY_LIMIT) {
+        if (value == 0 || value == 1) {
+            const USet *set = u_getBinaryPropertySet(prop, &ec);
+            if (U_FAILURE(ec)) { return *this; }
+            copyFrom(*UnicodeSet::fromUSet(set), true);
+            if (value == 0) {
+                complement().removeAllStrings();  // code point complement
+            }
+        } else {
+            clear();
+        }
+    } else if (UCHAR_INT_START <= prop && prop < UCHAR_INT_LIMIT) {
+        const UnicodeSet* inclusions = CharacterProperties::getInclusionsForProperty(prop, ec);
+        IntPropertyContext c = {prop, value};
+        applyFilter(intPropertyFilter, &c, inclusions, ec);
+    } else {
+        ec = U_ILLEGAL_ARGUMENT_ERROR;
+    }
+    return *this;
+}
+
+UnicodeSet&
+UnicodeSet::applyPropertyAlias(const UnicodeString& prop,
+                               const UnicodeString& value,
+                               UErrorCode& ec) {
+    if (U_FAILURE(ec) || isFrozen()) return *this;
+
+    // prop and value used to be converted to char * using the default
+    // converter instead of the invariant conversion.
+    // This should not be necessary because all Unicode property and value
+    // names use only invariant characters.
+    // If there are any variant characters, then we won't find them anyway.
+    // Checking first avoids assertion failures in the conversion.
+    if( !uprv_isInvariantUString(prop.getBuffer(), prop.length()) ||
+        !uprv_isInvariantUString(value.getBuffer(), value.length())
+    ) {
+        FAIL(ec);
+    }
+    CharString pname, vname;
+    pname.appendInvariantChars(prop, ec);
+    vname.appendInvariantChars(value, ec);
+    if (U_FAILURE(ec)) return *this;
+
+    UProperty p;
+    int32_t v;
+    UBool invert = false;
+
+    if (value.length() > 0) {
+        p = u_getPropertyEnum(pname.data());
+        if (p == UCHAR_INVALID_CODE) FAIL(ec);
+
+        // Treat gc as gcm
+        if (p == UCHAR_GENERAL_CATEGORY) {
+            p = UCHAR_GENERAL_CATEGORY_MASK;
+        }
+
+        if ((p >= UCHAR_BINARY_START && p < UCHAR_BINARY_LIMIT) ||
+            (p >= UCHAR_INT_START && p < UCHAR_INT_LIMIT) ||
+            (p >= UCHAR_MASK_START && p < UCHAR_MASK_LIMIT)) {
+            v = u_getPropertyValueEnum(p, vname.data());
+            if (v == UCHAR_INVALID_CODE) {
+                // Handle numeric CCC
+                if (p == UCHAR_CANONICAL_COMBINING_CLASS ||
+                    p == UCHAR_TRAIL_CANONICAL_COMBINING_CLASS ||
+                    p == UCHAR_LEAD_CANONICAL_COMBINING_CLASS) {
+                    char* end;
+                    double val = uprv_strtod(vname.data(), &end);
+                    // Anything between 0 and 255 is valid even if unused.
+                    // Cast double->int only after range check.
+                    // We catch NaN here because comparing it with both 0 and 255 will be false
+                    // (as are all comparisons with NaN).
+                    if (*end != 0 || !(0 <= val && val <= 255) ||
+                            (v = (int32_t)val) != val) {
+                        // non-integral value or outside 0..255, or trailing junk
+                        FAIL(ec);
+                    }
+                } else {
+                    FAIL(ec);
+                }
+            }
+        }
+
+        else {
+
+            switch (p) {
+            case UCHAR_NUMERIC_VALUE:
+                {
+                    char* end;
+                    double val = uprv_strtod(vname.data(), &end);
+                    if (*end != 0) {
+                        FAIL(ec);
+                    }
+                    applyFilter(numericValueFilter, &val,
+                                CharacterProperties::getInclusionsForProperty(p, ec), ec);
+                    return *this;
+                }
+            case UCHAR_NAME:
+                {
+                    // Must munge name, since u_charFromName() does not do
+                    // 'loose' matching.
+                    char buf[128]; // it suffices that this be > uprv_getMaxCharNameLength
+                    if (!mungeCharName(buf, vname.data(), sizeof(buf))) FAIL(ec);
+                    UChar32 ch = u_charFromName(U_EXTENDED_CHAR_NAME, buf, &ec);
+                    if (U_SUCCESS(ec)) {
+                        clear();
+                        add(ch);
+                        return *this;
+                    } else {
+                        FAIL(ec);
+                    }
+                }
+            case UCHAR_UNICODE_1_NAME:
+                // ICU 49 deprecates the Unicode_1_Name property APIs.
+                FAIL(ec);
+            case UCHAR_AGE:
+                {
+                    // Must munge name, since u_versionFromString() does not do
+                    // 'loose' matching.
+                    char buf[128];
+                    if (!mungeCharName(buf, vname.data(), sizeof(buf))) FAIL(ec);
+                    UVersionInfo version;
+                    u_versionFromString(version, buf);
+                    applyFilter(versionFilter, &version,
+                                CharacterProperties::getInclusionsForProperty(p, ec), ec);
+                    return *this;
+                }
+            case UCHAR_SCRIPT_EXTENSIONS:
+                v = u_getPropertyValueEnum(UCHAR_SCRIPT, vname.data());
+                if (v == UCHAR_INVALID_CODE) {
+                    FAIL(ec);
+                }
+                // fall through to calling applyIntPropertyValue()
+                break;
+            default:
+                // p is a non-binary, non-enumerated property that we
+                // don't support (yet).
+                FAIL(ec);
+            }
+        }
+    }
+
+    else {
+        // value is empty.  Interpret as General Category, Script, or
+        // Binary property.
+        p = UCHAR_GENERAL_CATEGORY_MASK;
+        v = u_getPropertyValueEnum(p, pname.data());
+        if (v == UCHAR_INVALID_CODE) {
+            p = UCHAR_SCRIPT;
+            v = u_getPropertyValueEnum(p, pname.data());
+            if (v == UCHAR_INVALID_CODE) {
+                p = u_getPropertyEnum(pname.data());
+                if (p >= UCHAR_BINARY_START && p < UCHAR_BINARY_LIMIT) {
+                    v = 1;
+                } else if (0 == uprv_comparePropertyNames(ANY, pname.data())) {
+                    set(MIN_VALUE, MAX_VALUE);
+                    return *this;
+                } else if (0 == uprv_comparePropertyNames(ASCII, pname.data())) {
+                    set(0, 0x7F);
+                    return *this;
+                } else if (0 == uprv_comparePropertyNames(ASSIGNED, pname.data())) {
+                    // [:Assigned:]=[:^Cn:]
+                    p = UCHAR_GENERAL_CATEGORY_MASK;
+                    v = U_GC_CN_MASK;
+                    invert = true;
+                } else {
+                    FAIL(ec);
+                }
+            }
+        }
+    }
+
+    applyIntPropertyValue(p, v, ec);
+    if(invert) {
+        complement().removeAllStrings();  // code point complement
+    }
+
+    if (isBogus() && U_SUCCESS(ec)) {
+        // We likely ran out of memory. AHHH!
+        ec = U_MEMORY_ALLOCATION_ERROR;
+    }
+    return *this;
+}
+
+//----------------------------------------------------------------
+// Property set patterns
+//----------------------------------------------------------------
+
+/**
+ * Return true if the given position, in the given pattern, appears
+ * to be the start of a property set pattern.
+ */
+UBool UnicodeSet::resemblesPropertyPattern(const UnicodeString& pattern,
+                                           int32_t pos) {
+    // Patterns are at least 5 characters long
+    if ((pos+5) > pattern.length()) {
+        return false;
+    }
+
+    // Look for an opening [:, [:^, \p, or \P
+    return isPOSIXOpen(pattern, pos) || isPerlOpen(pattern, pos) || isNameOpen(pattern, pos);
+}
+
+/**
+ * Return true if the given iterator appears to point at a
+ * property pattern.  Regardless of the result, return with the
+ * iterator unchanged.
+ * @param chars iterator over the pattern characters.  Upon return
+ * it will be unchanged.
+ * @param iterOpts RuleCharacterIterator options
+ */
+UBool UnicodeSet::resemblesPropertyPattern(RuleCharacterIterator& chars,
+                                           int32_t iterOpts) {
+    // NOTE: literal will always be false, because we don't parse escapes.
+    UBool result = false, literal;
+    UErrorCode ec = U_ZERO_ERROR;
+    iterOpts &= ~RuleCharacterIterator::PARSE_ESCAPES;
+    RuleCharacterIterator::Pos pos;
+    chars.getPos(pos);
+    UChar32 c = chars.next(iterOpts, literal, ec);
+    if (c == u'[' || c == u'\\') {
+        UChar32 d = chars.next(iterOpts & ~RuleCharacterIterator::SKIP_WHITESPACE,
+                               literal, ec);
+        result = (c == u'[') ? (d == u':') :
+                               (d == u'N' || d == u'p' || d == u'P');
+    }
+    chars.setPos(pos);
+    return result && U_SUCCESS(ec);
+}
+
+/**
+ * Parse the given property pattern at the given parse position.
+ */
+UnicodeSet& UnicodeSet::applyPropertyPattern(const UnicodeString& pattern,
+                                             ParsePosition& ppos,
+                                             UErrorCode &ec) {
+    int32_t pos = ppos.getIndex();
+
+    UBool posix = false; // true for [:pat:], false for \p{pat} \P{pat} \N{pat}
+    UBool isName = false; // true for \N{pat}, o/w false
+    UBool invert = false;
+
+    if (U_FAILURE(ec)) return *this;
+
+    // Minimum length is 5 characters, e.g. \p{L}
+    if ((pos+5) > pattern.length()) {
+        FAIL(ec);
+    }
+
+    // On entry, ppos should point to one of the following locations:
+    // Look for an opening [:, [:^, \p, or \P
+    if (isPOSIXOpen(pattern, pos)) {
+        posix = true;
+        pos += 2;
+        pos = ICU_Utility::skipWhitespace(pattern, pos);
+        if (pos < pattern.length() && pattern.charAt(pos) == u'^') {
+            ++pos;
+            invert = true;
+        }
+    } else if (isPerlOpen(pattern, pos) || isNameOpen(pattern, pos)) {
+        char16_t c = pattern.charAt(pos+1);
+        invert = (c == u'P');
+        isName = (c == u'N');
+        pos += 2;
+        pos = ICU_Utility::skipWhitespace(pattern, pos);
+        if (pos == pattern.length() || pattern.charAt(pos++) != u'{') {
+            // Syntax error; "\p" or "\P" not followed by "{"
+            FAIL(ec);
+        }
+    } else {
+        // Open delimiter not seen
+        FAIL(ec);
+    }
+
+    // Look for the matching close delimiter, either :] or }
+    int32_t close;
+    if (posix) {
+      close = pattern.indexOf(u":]", 2, pos);
+    } else {
+      close = pattern.indexOf(u'}', pos);
+    }
+    if (close < 0) {
+        // Syntax error; close delimiter missing
+        FAIL(ec);
+    }
+
+    // Look for an '=' sign.  If this is present, we will parse a
+    // medium \p{gc=Cf} or long \p{GeneralCategory=Format}
+    // pattern.
+    int32_t equals = pattern.indexOf(u'=', pos);
+    UnicodeString propName, valueName;
+    if (equals >= 0 && equals < close && !isName) {
+        // Equals seen; parse medium/long pattern
+        pattern.extractBetween(pos, equals, propName);
+        pattern.extractBetween(equals+1, close, valueName);
+    }
+
+    else {
+        // Handle case where no '=' is seen, and \N{}
+        pattern.extractBetween(pos, close, propName);
+            
+        // Handle \N{name}
+        if (isName) {
+            // This is a little inefficient since it means we have to
+            // parse NAME_PROP back to UCHAR_NAME even though we already
+            // know it's UCHAR_NAME.  If we refactor the API to
+            // support args of (UProperty, char*) then we can remove
+            // NAME_PROP and make this a little more efficient.
+            valueName = propName;
+            propName = UnicodeString(NAME_PROP, NAME_PROP_LENGTH, US_INV);
+        }
+    }
+
+    applyPropertyAlias(propName, valueName, ec);
+
+    if (U_SUCCESS(ec)) {
+        if (invert) {
+            complement().removeAllStrings();  // code point complement
+        }
+
+        // Move to the limit position after the close delimiter if the
+        // parse succeeded.
+        ppos.setIndex(close + (posix ? 2 : 1));
+    }
+
+    return *this;
+}
+
+/**
+ * Parse a property pattern.
+ * @param chars iterator over the pattern characters.  Upon return
+ * it will be advanced to the first character after the parsed
+ * pattern, or the end of the iteration if all characters are
+ * parsed.
+ * @param rebuiltPat the pattern that was parsed, rebuilt or
+ * copied from the input pattern, as appropriate.
+ */
+void UnicodeSet::applyPropertyPattern(RuleCharacterIterator& chars,
+                                      UnicodeString& rebuiltPat,
+                                      UErrorCode& ec) {
+    if (U_FAILURE(ec)) return;
+    UnicodeString pattern;
+    chars.lookahead(pattern);
+    ParsePosition pos(0);
+    applyPropertyPattern(pattern, pos, ec);
+    if (U_FAILURE(ec)) return;
+    if (pos.getIndex() == 0) {
+        // syntaxError(chars, "Invalid property pattern");
+        ec = U_MALFORMED_SET;
+        return;
+    }
+    chars.jumpahead(pos.getIndex());
+    rebuiltPat.append(pattern, 0, pos.getIndex());
+}
+
+U_NAMESPACE_END
diff --git a/intl/icu/source/common/unisetspan.cpp b/intl/icu/source/common/unisetspan.cpp
new file mode 100644
index 0000000000..b7256d8fd1
--- /dev/null
+++ b/intl/icu/source/common/unisetspan.cpp
@@ -0,0 +1,1522 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+*
+*   Copyright (C) 2007-2012, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+******************************************************************************
+*   file name:  unisetspan.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2007mar01
+*   created by: Markus W. Scherer
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/uniset.h"
+#include "unicode/ustring.h"
+#include "unicode/utf8.h"
+#include "unicode/utf16.h"
+#include "cmemory.h"
+#include "uvector.h"
+#include "unisetspan.h"
+
+U_NAMESPACE_BEGIN
+
+/*
+ * List of offsets from the current position from where to try matching
+ * a code point or a string.
+ * Store offsets rather than indexes to simplify the code and use the same list
+ * for both increments (in span()) and decrements (in spanBack()).
+ *
+ * Assumption: The maximum offset is limited, and the offsets that are stored
+ * at any one time are relatively dense, that is, there are normally no gaps of
+ * hundreds or thousands of offset values.
+ *
+ * The implementation uses a circular buffer of byte flags,
+ * each indicating whether the corresponding offset is in the list.
+ * This avoids inserting into a sorted list of offsets (or absolute indexes) and
+ * physically moving part of the list.
+ *
+ * Note: In principle, the caller should setMaxLength() to the maximum of the
+ * max string length and U16_LENGTH/U8_LENGTH to account for
+ * "long" single code points.
+ * However, this implementation uses at least a staticList with more than
+ * U8_LENGTH entries anyway.
+ *
+ * Note: If maxLength were guaranteed to be no more than 32 or 64,
+ * the list could be stored as bit flags in a single integer.
+ * Rather than handling a circular buffer with a start list index,
+ * the integer would simply be shifted when lower offsets are removed.
+ * UnicodeSet does not have a limit on the lengths of strings.
+ */
+class OffsetList {  // Only ever stack-allocated, does not need to inherit UMemory.
+public:
+    OffsetList() : list(staticList), capacity(0), length(0), start(0) {}
+
+    ~OffsetList() {
+        if(list!=staticList) {
+            uprv_free(list);
+        }
+    }
+
+    // Call exactly once if the list is to be used.
+    void setMaxLength(int32_t maxLength) {
+        if(maxLength<=(int32_t)sizeof(staticList)) {
+            capacity=(int32_t)sizeof(staticList);
+        } else {
+            UBool *l=(UBool *)uprv_malloc(maxLength);
+            if(l!=nullptr) {
+                list=l;
+                capacity=maxLength;
+            }
+        }
+        uprv_memset(list, 0, capacity);
+    }
+
+    void clear() {
+        uprv_memset(list, 0, capacity);
+        start=length=0;
+    }
+
+    UBool isEmpty() const {
+        return (UBool)(length==0);
+    }
+
+    // Reduce all stored offsets by delta, used when the current position
+    // moves by delta.
+    // There must not be any offsets lower than delta.
+    // If there is an offset equal to delta, it is removed.
+    // delta=[1..maxLength]
+    void shift(int32_t delta) {
+        int32_t i=start+delta;
+        if(i>=capacity) {
+            i-=capacity;
+        }
+        if(list[i]) {
+            list[i]=false;
+            --length;
+        }
+        start=i;
+    }
+
+    // Add an offset. The list must not contain it yet.
+    // offset=[1..maxLength]
+    void addOffset(int32_t offset) {
+        int32_t i=start+offset;
+        if(i>=capacity) {
+            i-=capacity;
+        }
+        list[i]=true;
+        ++length;
+    }
+
+    // offset=[1..maxLength]
+    UBool containsOffset(int32_t offset) const {
+        int32_t i=start+offset;
+        if(i>=capacity) {
+            i-=capacity;
+        }
+        return list[i];
+    }
+
+    // Find the lowest stored offset from a non-empty list, remove it,
+    // and reduce all other offsets by this minimum.
+    // Returns [1..maxLength].
+    int32_t popMinimum() {
+        // Look for the next offset in list[start+1..capacity-1].
+        int32_t i=start, result;
+        while(++imaxLength16) {
+            maxLength16=length16;
+        }
+        if((which&UTF8) && (thisRelevant || (which&CONTAINED))) {
+            int32_t length8=getUTF8Length(s16, length16);
+            utf8Length+=length8;
+            if(length8>maxLength8) {
+                maxLength8=length8;
+            }
+        }
+    }
+    if(!someRelevant) {
+        maxLength16=maxLength8=0;
+        return;
+    }
+
+    // Freeze after checking for the need to use strings at all because freezing
+    // a set takes some time and memory which are wasted if there are no relevant strings.
+    if(all) {
+        spanSet.freeze();
+    }
+
+    uint8_t *spanBackLengths;
+    uint8_t *spanUTF8Lengths;
+    uint8_t *spanBackUTF8Lengths;
+
+    // Allocate a block of meta data.
+    int32_t allocSize;
+    if(all) {
+        // UTF-8 lengths, 4 sets of span lengths, UTF-8 strings.
+        allocSize=stringsLength*(4+1+1+1+1)+utf8Length;
+    } else {
+        allocSize=stringsLength;  // One set of span lengths.
+        if(which&UTF8) {
+            // UTF-8 lengths and UTF-8 strings.
+            allocSize+=stringsLength*4+utf8Length;
+        }
+    }
+    if(allocSize<=(int32_t)sizeof(staticLengths)) {
+        utf8Lengths=staticLengths;
+    } else {
+        utf8Lengths=(int32_t *)uprv_malloc(allocSize);
+        if(utf8Lengths==nullptr) {
+            maxLength16=maxLength8=0;  // Prevent usage by making needsStringSpanUTF16/8() return false.
+            return;  // Out of memory.
+        }
+    }
+
+    if(all) {
+        // Store span lengths for all span() variants.
+        spanLengths=(uint8_t *)(utf8Lengths+stringsLength);
+        spanBackLengths=spanLengths+stringsLength;
+        spanUTF8Lengths=spanBackLengths+stringsLength;
+        spanBackUTF8Lengths=spanUTF8Lengths+stringsLength;
+        utf8=spanBackUTF8Lengths+stringsLength;
+    } else {
+        // Store span lengths for only one span() variant.
+        if(which&UTF8) {
+            spanLengths=(uint8_t *)(utf8Lengths+stringsLength);
+            utf8=spanLengths+stringsLength;
+        } else {
+            spanLengths=(uint8_t *)utf8Lengths;
+        }
+        spanBackLengths=spanUTF8Lengths=spanBackUTF8Lengths=spanLengths;
+    }
+
+    // Set the meta data and pSpanNotSet and write the UTF-8 strings.
+    int32_t utf8Count=0;  // Count UTF-8 bytes written so far.
+
+    for(i=0; i0) {  // Relevant string.
+            if(which&UTF16) {
+                if(which&CONTAINED) {
+                    if(which&FWD) {
+                        spanLengths[i]=makeSpanLengthByte(spanLength);
+                    }
+                    if(which&BACK) {
+                        spanLength=length16-spanSet.spanBack(s16, length16, USET_SPAN_CONTAINED);
+                        spanBackLengths[i]=makeSpanLengthByte(spanLength);
+                    }
+                } else /* not CONTAINED, not all, but NOT_CONTAINED */ {
+                    spanLengths[i]=spanBackLengths[i]=0;  // Only store a relevant/irrelevant flag.
+                }
+            }
+            if(which&UTF8) {
+                uint8_t *s8=utf8+utf8Count;
+                int32_t length8=appendUTF8(s16, length16, s8, utf8Length-utf8Count);
+                utf8Count+=utf8Lengths[i]=length8;
+                if(length8==0) {  // Irrelevant for UTF-8 because not representable in UTF-8.
+                    spanUTF8Lengths[i]=spanBackUTF8Lengths[i]=(uint8_t)ALL_CP_CONTAINED;
+                } else {  // Relevant for UTF-8.
+                    if(which&CONTAINED) {
+                        if(which&FWD) {
+                            spanLength=spanSet.spanUTF8((const char *)s8, length8, USET_SPAN_CONTAINED);
+                            spanUTF8Lengths[i]=makeSpanLengthByte(spanLength);
+                        }
+                        if(which&BACK) {
+                            spanLength=length8-spanSet.spanBackUTF8((const char *)s8, length8, USET_SPAN_CONTAINED);
+                            spanBackUTF8Lengths[i]=makeSpanLengthByte(spanLength);
+                        }
+                    } else /* not CONTAINED, not all, but NOT_CONTAINED */ {
+                        spanUTF8Lengths[i]=spanBackUTF8Lengths[i]=0;  // Only store a relevant/irrelevant flag.
+                    }
+                }
+            }
+            if(which&NOT_CONTAINED) {
+                // Add string start and end code points to the spanNotSet so that
+                // a span(while not contained) stops before any string.
+                UChar32 c;
+                if(which&FWD) {
+                    int32_t len=0;
+                    U16_NEXT(s16, len, length16, c);
+                    addToSpanNotSet(c);
+                }
+                if(which&BACK) {
+                    int32_t len=length16;
+                    U16_PREV(s16, 0, len, c);
+                    addToSpanNotSet(c);
+                }
+            }
+        } else {  // Irrelevant string. (Also the empty string.)
+            if(which&UTF8) {
+                if(which&CONTAINED) {  // Only necessary for LONGEST_MATCH.
+                    uint8_t *s8=utf8+utf8Count;
+                    int32_t length8=appendUTF8(s16, length16, s8, utf8Length-utf8Count);
+                    utf8Count+=utf8Lengths[i]=length8;
+                } else {
+                    utf8Lengths[i]=0;
+                }
+            }
+            if(all) {
+                spanLengths[i]=spanBackLengths[i]=
+                    spanUTF8Lengths[i]=spanBackUTF8Lengths[i]=
+                        (uint8_t)ALL_CP_CONTAINED;
+            } else {
+                // All spanXYZLengths pointers contain the same address.
+                spanLengths[i]=(uint8_t)ALL_CP_CONTAINED;
+            }
+        }
+    }
+
+    // Finish.
+    if(all) {
+        pSpanNotSet->freeze();
+    }
+}
+
+// Copy constructor. Assumes which==ALL for a frozen set.
+UnicodeSetStringSpan::UnicodeSetStringSpan(const UnicodeSetStringSpan &otherStringSpan,
+                                           const UVector &newParentSetStrings)
+        : spanSet(otherStringSpan.spanSet), pSpanNotSet(nullptr), strings(newParentSetStrings),
+          utf8Lengths(nullptr), spanLengths(nullptr), utf8(nullptr),
+          utf8Length(otherStringSpan.utf8Length),
+          maxLength16(otherStringSpan.maxLength16), maxLength8(otherStringSpan.maxLength8),
+          all(true) {
+    if(otherStringSpan.pSpanNotSet==&otherStringSpan.spanSet) {
+        pSpanNotSet=&spanSet;
+    } else {
+        pSpanNotSet=otherStringSpan.pSpanNotSet->clone();
+    }
+
+    // Allocate a block of meta data.
+    // UTF-8 lengths, 4 sets of span lengths, UTF-8 strings.
+    int32_t stringsLength=strings.size();
+    int32_t allocSize=stringsLength*(4+1+1+1+1)+utf8Length;
+    if(allocSize<=(int32_t)sizeof(staticLengths)) {
+        utf8Lengths=staticLengths;
+    } else {
+        utf8Lengths=(int32_t *)uprv_malloc(allocSize);
+        if(utf8Lengths==nullptr) {
+            maxLength16=maxLength8=0;  // Prevent usage by making needsStringSpanUTF16/8() return false.
+            return;  // Out of memory.
+        }
+    }
+
+    spanLengths=(uint8_t *)(utf8Lengths+stringsLength);
+    utf8=spanLengths+stringsLength*4;
+    uprv_memcpy(utf8Lengths, otherStringSpan.utf8Lengths, allocSize);
+}
+
+UnicodeSetStringSpan::~UnicodeSetStringSpan() {
+    if(pSpanNotSet!=nullptr && pSpanNotSet!=&spanSet) {
+        delete pSpanNotSet;
+    }
+    if(utf8Lengths!=nullptr && utf8Lengths!=staticLengths) {
+        uprv_free(utf8Lengths);
+    }
+}
+
+void UnicodeSetStringSpan::addToSpanNotSet(UChar32 c) {
+    if(pSpanNotSet==nullptr || pSpanNotSet==&spanSet) {
+        if(spanSet.contains(c)) {
+            return;  // Nothing to do.
+        }
+        UnicodeSet *newSet=spanSet.cloneAsThawed();
+        if(newSet==nullptr) {
+            return;  // Out of memory.
+        } else {
+            pSpanNotSet=newSet;
+        }
+    }
+    pSpanNotSet->add(c);
+}
+
+// Compare strings without any argument checks. Requires length>0.
+static inline UBool
+matches16(const char16_t *s, const char16_t *t, int32_t length) {
+    do {
+        if(*s++!=*t++) {
+            return false;
+        }
+    } while(--length>0);
+    return true;
+}
+
+static inline UBool
+matches8(const uint8_t *s, const uint8_t *t, int32_t length) {
+    do {
+        if(*s++!=*t++) {
+            return false;
+        }
+    } while(--length>0);
+    return true;
+}
+
+// Compare 16-bit Unicode strings (which may be malformed UTF-16)
+// at code point boundaries.
+// That is, each edge of a match must not be in the middle of a surrogate pair.
+static inline UBool
+matches16CPB(const char16_t *s, int32_t start, int32_t limit, const char16_t *t, int32_t length) {
+    s+=start;
+    limit-=start;
+    return matches16(s, t, length) &&
+           !(0=0xd800 && c<=0xdbff && length>=2 && U16_IS_TRAIL(c2=s[1])) {
+        return set.contains(U16_GET_SUPPLEMENTARY(c, c2)) ? 2 : -2;
+    }
+    return set.contains(c) ? 1 : -1;
+}
+
+static inline int32_t
+spanOneBack(const UnicodeSet &set, const char16_t *s, int32_t length) {
+    char16_t c=s[length-1], c2;
+    if(c>=0xdc00 && c<=0xdfff && length>=2 && U16_IS_LEAD(c2=s[length-2])) {
+        return set.contains(U16_GET_SUPPLEMENTARY(c2, c)) ? 2 : -2;
+    }
+    return set.contains(c) ? 1 : -1;
+}
+
+static inline int32_t
+spanOneUTF8(const UnicodeSet &set, const uint8_t *s, int32_t length) {
+    UChar32 c=*s;
+    if(U8_IS_SINGLE(c)) {
+        return set.contains(c) ? 1 : -1;
+    }
+    // Take advantage of non-ASCII fastpaths in U8_NEXT_OR_FFFD().
+    int32_t i=0;
+    U8_NEXT_OR_FFFD(s, i, length, c);
+    return set.contains(c) ? i : -i;
+}
+
+static inline int32_t
+spanOneBackUTF8(const UnicodeSet &set, const uint8_t *s, int32_t length) {
+    UChar32 c=s[length-1];
+    if(U8_IS_SINGLE(c)) {
+        return set.contains(c) ? 1 : -1;
+    }
+    int32_t i=length-1;
+    c=utf8_prevCharSafeBody(s, 0, &i, c, -3);
+    length-=i;
+    return set.contains(c) ? length : -length;
+}
+
+/*
+ * Note: In span() when spanLength==0 (after a string match, or at the beginning
+ * after an empty code point span) and in spanNot() and spanNotUTF8(),
+ * string matching could use a binary search
+ * because all string matches are done from the same start index.
+ *
+ * For UTF-8, this would require a comparison function that returns UTF-16 order.
+ *
+ * This optimization should not be necessary for normal UnicodeSets because
+ * most sets have no strings, and most sets with strings have
+ * very few very short strings.
+ * For cases with many strings, it might be better to use a different API
+ * and implementation with a DFA (state machine).
+ */
+
+/*
+ * Algorithm for span(USET_SPAN_CONTAINED)
+ *
+ * Theoretical algorithm:
+ * - Iterate through the string, and at each code point boundary:
+ *   + If the code point there is in the set, then remember to continue after it.
+ *   + If a set string matches at the current position, then remember to continue after it.
+ *   + Either recursively span for each code point or string match,
+ *     or recursively span for all but the shortest one and
+ *     iteratively continue the span with the shortest local match.
+ *   + Remember the longest recursive span (the farthest end point).
+ *   + If there is no match at the current position, neither for the code point there
+ *     nor for any set string, then stop and return the longest recursive span length.
+ *
+ * Optimized implementation:
+ *
+ * (We assume that most sets will have very few very short strings.
+ * A span using a string-less set is extremely fast.)
+ *
+ * Create and cache a spanSet which contains all of the single code points
+ * of the original set but none of its strings.
+ *
+ * - Start with spanLength=spanSet.span(USET_SPAN_CONTAINED).
+ * - Loop:
+ *   + Try to match each set string at the end of the spanLength.
+ *     ~ Set strings that start with set-contained code points must be matched
+ *       with a partial overlap because the recursive algorithm would have tried
+ *       to match them at every position.
+ *     ~ Set strings that entirely consist of set-contained code points
+ *       are irrelevant for span(USET_SPAN_CONTAINED) because the
+ *       recursive algorithm would continue after them anyway
+ *       and find the longest recursive match from their end.
+ *     ~ Rather than recursing, note each end point of a set string match.
+ *   + If no set string matched after spanSet.span(), then return
+ *     with where the spanSet.span() ended.
+ *   + If at least one set string matched after spanSet.span(), then
+ *     pop the shortest string match end point and continue
+ *     the loop, trying to match all set strings from there.
+ *   + If at least one more set string matched after a previous string match,
+ *     then test if the code point after the previous string match is also
+ *     contained in the set.
+ *     Continue the loop with the shortest end point of either this code point
+ *     or a matching set string.
+ *   + If no more set string matched after a previous string match,
+ *     then try another spanLength=spanSet.span(USET_SPAN_CONTAINED).
+ *     Stop if spanLength==0, otherwise continue the loop.
+ *
+ * By noting each end point of a set string match,
+ * the function visits each string position at most once and finishes
+ * in linear time.
+ *
+ * The recursive algorithm may visit the same string position many times
+ * if multiple paths lead to it and finishes in exponential time.
+ */
+
+/*
+ * Algorithm for span(USET_SPAN_SIMPLE)
+ *
+ * Theoretical algorithm:
+ * - Iterate through the string, and at each code point boundary:
+ *   + If the code point there is in the set, then remember to continue after it.
+ *   + If a set string matches at the current position, then remember to continue after it.
+ *   + Continue from the farthest match position and ignore all others.
+ *   + If there is no match at the current position,
+ *     then stop and return the current position.
+ *
+ * Optimized implementation:
+ *
+ * (Same assumption and spanSet as above.)
+ *
+ * - Start with spanLength=spanSet.span(USET_SPAN_CONTAINED).
+ * - Loop:
+ *   + Try to match each set string at the end of the spanLength.
+ *     ~ Set strings that start with set-contained code points must be matched
+ *       with a partial overlap because the standard algorithm would have tried
+ *       to match them earlier.
+ *     ~ Set strings that entirely consist of set-contained code points
+ *       must be matched with a full overlap because the longest-match algorithm
+ *       would hide set string matches that end earlier.
+ *       Such set strings need not be matched earlier inside the code point span
+ *       because the standard algorithm would then have continued after
+ *       the set string match anyway.
+ *     ~ Remember the longest set string match (farthest end point) from the earliest
+ *       starting point.
+ *   + If no set string matched after spanSet.span(), then return
+ *     with where the spanSet.span() ended.
+ *   + If at least one set string matched, then continue the loop after the
+ *     longest match from the earliest position.
+ *   + If no more set string matched after a previous string match,
+ *     then try another spanLength=spanSet.span(USET_SPAN_CONTAINED).
+ *     Stop if spanLength==0, otherwise continue the loop.
+ */
+
+int32_t UnicodeSetStringSpan::span(const char16_t *s, int32_t length, USetSpanCondition spanCondition) const {
+    if(spanCondition==USET_SPAN_NOT_CONTAINED) {
+        return spanNot(s, length);
+    }
+    int32_t spanLength=spanSet.span(s, length, USET_SPAN_CONTAINED);
+    if(spanLength==length) {
+        return length;
+    }
+
+    // Consider strings; they may overlap with the span.
+    OffsetList offsets;
+    if(spanCondition==USET_SPAN_CONTAINED) {
+        // Use offset list to try all possibilities.
+        offsets.setMaxLength(maxLength16);
+    }
+    int32_t pos=spanLength, rest=length-pos;
+    int32_t i, stringsLength=strings.size();
+    for(;;) {
+        if(spanCondition==USET_SPAN_CONTAINED) {
+            for(i=0; i0);
+
+                // Try to match this string at pos-overlap..pos.
+                if(overlap>=LONG_SPAN) {
+                    overlap=length16;
+                    // While contained: No point matching fully inside the code point span.
+                    U16_BACK_1(s16, 0, overlap);  // Length of the string minus the last code point.
+                }
+                if(overlap>spanLength) {
+                    overlap=spanLength;
+                }
+                int32_t inc=length16-overlap;  // Keep overlap+inc==length16.
+                for(;;) {
+                    if(inc>rest) {
+                        break;
+                    }
+                    // Try to match if the increment is not listed already.
+                    if(!offsets.containsOffset(inc) && matches16CPB(s, pos-overlap, length, s16, length16)) {
+                        if(inc==rest) {
+                            return length;  // Reached the end of the string.
+                        }
+                        offsets.addOffset(inc);
+                    }
+                    if(overlap==0) {
+                        break;
+                    }
+                    --overlap;
+                    ++inc;
+                }
+            }
+        } else /* USET_SPAN_SIMPLE */ {
+            int32_t maxInc=0, maxOverlap=0;
+            for(i=0; i=LONG_SPAN) {
+                    overlap=length16;
+                    // Longest match: Need to match fully inside the code point span
+                    // to find the match from the earliest start.
+                }
+                if(overlap>spanLength) {
+                    overlap=spanLength;
+                }
+                int32_t inc=length16-overlap;  // Keep overlap+inc==length16.
+                for(;;) {
+                    if(inc>rest || overlapmaxOverlap || /* redundant overlap==maxOverlap && */ inc>maxInc) &&
+                        matches16CPB(s, pos-overlap, length, s16, length16)
+                    ) {
+                        maxInc=inc;  // Longest match from earliest start.
+                        maxOverlap=overlap;
+                        break;
+                    }
+                    --overlap;
+                    ++inc;
+                }
+            }
+
+            if(maxInc!=0 || maxOverlap!=0) {
+                // Longest-match algorithm, and there was a string match.
+                // Simply continue after it.
+                pos+=maxInc;
+                rest-=maxInc;
+                if(rest==0) {
+                    return length;  // Reached the end of the string.
+                }
+                spanLength=0;  // Match strings from after a string match.
+                continue;
+            }
+        }
+        // Finished trying to match all strings at pos.
+
+        if(spanLength!=0 || pos==0) {
+            // The position is after an unlimited code point span (spanLength!=0),
+            // not after a string match.
+            // The only position where spanLength==0 after a span is pos==0.
+            // Otherwise, an unlimited code point span is only tried again when no
+            // strings match, and if such a non-initial span fails we stop.
+            if(offsets.isEmpty()) {
+                return pos;  // No strings matched after a span.
+            }
+            // Match strings from after the next string match.
+        } else {
+            // The position is after a string match (or a single code point).
+            if(offsets.isEmpty()) {
+                // No more strings matched after a previous string match.
+                // Try another code point span from after the last string match.
+                spanLength=spanSet.span(s+pos, rest, USET_SPAN_CONTAINED);
+                if( spanLength==rest || // Reached the end of the string, or
+                    spanLength==0       // neither strings nor span progressed.
+                ) {
+                    return pos+spanLength;
+                }
+                pos+=spanLength;
+                rest-=spanLength;
+                continue;  // spanLength>0: Match strings from after a span.
+            } else {
+                // Try to match only one code point from after a string match if some
+                // string matched beyond it, so that we try all possible positions
+                // and don't overshoot.
+                spanLength=spanOne(spanSet, s+pos, rest);
+                if(spanLength>0) {
+                    if(spanLength==rest) {
+                        return length;  // Reached the end of the string.
+                    }
+                    // Match strings after this code point.
+                    // There cannot be any increments below it because UnicodeSet strings
+                    // contain multiple code points.
+                    pos+=spanLength;
+                    rest-=spanLength;
+                    offsets.shift(spanLength);
+                    spanLength=0;
+                    continue;  // Match strings from after a single code point.
+                }
+                // Match strings from after the next string match.
+            }
+        }
+        int32_t minOffset=offsets.popMinimum();
+        pos+=minOffset;
+        rest-=minOffset;
+        spanLength=0;  // Match strings from after a string match.
+    }
+}
+
+int32_t UnicodeSetStringSpan::spanBack(const char16_t *s, int32_t length, USetSpanCondition spanCondition) const {
+    if(spanCondition==USET_SPAN_NOT_CONTAINED) {
+        return spanNotBack(s, length);
+    }
+    int32_t pos=spanSet.spanBack(s, length, USET_SPAN_CONTAINED);
+    if(pos==0) {
+        return 0;
+    }
+    int32_t spanLength=length-pos;
+
+    // Consider strings; they may overlap with the span.
+    OffsetList offsets;
+    if(spanCondition==USET_SPAN_CONTAINED) {
+        // Use offset list to try all possibilities.
+        offsets.setMaxLength(maxLength16);
+    }
+    int32_t i, stringsLength=strings.size();
+    uint8_t *spanBackLengths=spanLengths;
+    if(all) {
+        spanBackLengths+=stringsLength;
+    }
+    for(;;) {
+        if(spanCondition==USET_SPAN_CONTAINED) {
+            for(i=0; i0);
+
+                // Try to match this string at pos-(length16-overlap)..pos-length16.
+                if(overlap>=LONG_SPAN) {
+                    overlap=length16;
+                    // While contained: No point matching fully inside the code point span.
+                    int32_t len1=0;
+                    U16_FWD_1(s16, len1, overlap);
+                    overlap-=len1;  // Length of the string minus the first code point.
+                }
+                if(overlap>spanLength) {
+                    overlap=spanLength;
+                }
+                int32_t dec=length16-overlap;  // Keep dec+overlap==length16.
+                for(;;) {
+                    if(dec>pos) {
+                        break;
+                    }
+                    // Try to match if the decrement is not listed already.
+                    if(!offsets.containsOffset(dec) && matches16CPB(s, pos-dec, length, s16, length16)) {
+                        if(dec==pos) {
+                            return 0;  // Reached the start of the string.
+                        }
+                        offsets.addOffset(dec);
+                    }
+                    if(overlap==0) {
+                        break;
+                    }
+                    --overlap;
+                    ++dec;
+                }
+            }
+        } else /* USET_SPAN_SIMPLE */ {
+            int32_t maxDec=0, maxOverlap=0;
+            for(i=0; i=LONG_SPAN) {
+                    overlap=length16;
+                    // Longest match: Need to match fully inside the code point span
+                    // to find the match from the latest end.
+                }
+                if(overlap>spanLength) {
+                    overlap=spanLength;
+                }
+                int32_t dec=length16-overlap;  // Keep dec+overlap==length16.
+                for(;;) {
+                    if(dec>pos || overlapmaxOverlap || /* redundant overlap==maxOverlap && */ dec>maxDec) &&
+                        matches16CPB(s, pos-dec, length, s16, length16)
+                    ) {
+                        maxDec=dec;  // Longest match from latest end.
+                        maxOverlap=overlap;
+                        break;
+                    }
+                    --overlap;
+                    ++dec;
+                }
+            }
+
+            if(maxDec!=0 || maxOverlap!=0) {
+                // Longest-match algorithm, and there was a string match.
+                // Simply continue before it.
+                pos-=maxDec;
+                if(pos==0) {
+                    return 0;  // Reached the start of the string.
+                }
+                spanLength=0;  // Match strings from before a string match.
+                continue;
+            }
+        }
+        // Finished trying to match all strings at pos.
+
+        if(spanLength!=0 || pos==length) {
+            // The position is before an unlimited code point span (spanLength!=0),
+            // not before a string match.
+            // The only position where spanLength==0 before a span is pos==length.
+            // Otherwise, an unlimited code point span is only tried again when no
+            // strings match, and if such a non-initial span fails we stop.
+            if(offsets.isEmpty()) {
+                return pos;  // No strings matched before a span.
+            }
+            // Match strings from before the next string match.
+        } else {
+            // The position is before a string match (or a single code point).
+            if(offsets.isEmpty()) {
+                // No more strings matched before a previous string match.
+                // Try another code point span from before the last string match.
+                int32_t oldPos=pos;
+                pos=spanSet.spanBack(s, oldPos, USET_SPAN_CONTAINED);
+                spanLength=oldPos-pos;
+                if( pos==0 ||           // Reached the start of the string, or
+                    spanLength==0       // neither strings nor span progressed.
+                ) {
+                    return pos;
+                }
+                continue;  // spanLength>0: Match strings from before a span.
+            } else {
+                // Try to match only one code point from before a string match if some
+                // string matched beyond it, so that we try all possible positions
+                // and don't overshoot.
+                spanLength=spanOneBack(spanSet, s, pos);
+                if(spanLength>0) {
+                    if(spanLength==pos) {
+                        return 0;  // Reached the start of the string.
+                    }
+                    // Match strings before this code point.
+                    // There cannot be any decrements below it because UnicodeSet strings
+                    // contain multiple code points.
+                    pos-=spanLength;
+                    offsets.shift(spanLength);
+                    spanLength=0;
+                    continue;  // Match strings from before a single code point.
+                }
+                // Match strings from before the next string match.
+            }
+        }
+        pos-=offsets.popMinimum();
+        spanLength=0;  // Match strings from before a string match.
+    }
+}
+
+int32_t UnicodeSetStringSpan::spanUTF8(const uint8_t *s, int32_t length, USetSpanCondition spanCondition) const {
+    if(spanCondition==USET_SPAN_NOT_CONTAINED) {
+        return spanNotUTF8(s, length);
+    }
+    int32_t spanLength=spanSet.spanUTF8((const char *)s, length, USET_SPAN_CONTAINED);
+    if(spanLength==length) {
+        return length;
+    }
+
+    // Consider strings; they may overlap with the span.
+    OffsetList offsets;
+    if(spanCondition==USET_SPAN_CONTAINED) {
+        // Use offset list to try all possibilities.
+        offsets.setMaxLength(maxLength8);
+    }
+    int32_t pos=spanLength, rest=length-pos;
+    int32_t i, stringsLength=strings.size();
+    uint8_t *spanUTF8Lengths=spanLengths;
+    if(all) {
+        spanUTF8Lengths+=2*stringsLength;
+    }
+    for(;;) {
+        const uint8_t *s8=utf8;
+        int32_t length8;
+        if(spanCondition==USET_SPAN_CONTAINED) {
+            for(i=0; i=LONG_SPAN) {
+                    overlap=length8;
+                    // While contained: No point matching fully inside the code point span.
+                    U8_BACK_1(s8, 0, overlap);  // Length of the string minus the last code point.
+                }
+                if(overlap>spanLength) {
+                    overlap=spanLength;
+                }
+                int32_t inc=length8-overlap;  // Keep overlap+inc==length8.
+                for(;;) {
+                    if(inc>rest) {
+                        break;
+                    }
+                    // Try to match if the increment is not listed already.
+                    // Match at code point boundaries. (The UTF-8 strings were converted
+                    // from UTF-16 and are guaranteed to be well-formed.)
+                    if(!U8_IS_TRAIL(s[pos-overlap]) &&
+                            !offsets.containsOffset(inc) &&
+                            matches8(s+pos-overlap, s8, length8)) {
+                        if(inc==rest) {
+                            return length;  // Reached the end of the string.
+                        }
+                        offsets.addOffset(inc);
+                    }
+                    if(overlap==0) {
+                        break;
+                    }
+                    --overlap;
+                    ++inc;
+                }
+                s8+=length8;
+            }
+        } else /* USET_SPAN_SIMPLE */ {
+            int32_t maxInc=0, maxOverlap=0;
+            for(i=0; i=LONG_SPAN) {
+                    overlap=length8;
+                    // Longest match: Need to match fully inside the code point span
+                    // to find the match from the earliest start.
+                }
+                if(overlap>spanLength) {
+                    overlap=spanLength;
+                }
+                int32_t inc=length8-overlap;  // Keep overlap+inc==length8.
+                for(;;) {
+                    if(inc>rest || overlapmaxOverlap ||
+                                /* redundant overlap==maxOverlap && */ inc>maxInc) &&
+                            matches8(s+pos-overlap, s8, length8)) {
+                        maxInc=inc;  // Longest match from earliest start.
+                        maxOverlap=overlap;
+                        break;
+                    }
+                    --overlap;
+                    ++inc;
+                }
+                s8+=length8;
+            }
+
+            if(maxInc!=0 || maxOverlap!=0) {
+                // Longest-match algorithm, and there was a string match.
+                // Simply continue after it.
+                pos+=maxInc;
+                rest-=maxInc;
+                if(rest==0) {
+                    return length;  // Reached the end of the string.
+                }
+                spanLength=0;  // Match strings from after a string match.
+                continue;
+            }
+        }
+        // Finished trying to match all strings at pos.
+
+        if(spanLength!=0 || pos==0) {
+            // The position is after an unlimited code point span (spanLength!=0),
+            // not after a string match.
+            // The only position where spanLength==0 after a span is pos==0.
+            // Otherwise, an unlimited code point span is only tried again when no
+            // strings match, and if such a non-initial span fails we stop.
+            if(offsets.isEmpty()) {
+                return pos;  // No strings matched after a span.
+            }
+            // Match strings from after the next string match.
+        } else {
+            // The position is after a string match (or a single code point).
+            if(offsets.isEmpty()) {
+                // No more strings matched after a previous string match.
+                // Try another code point span from after the last string match.
+                spanLength=spanSet.spanUTF8((const char *)s+pos, rest, USET_SPAN_CONTAINED);
+                if( spanLength==rest || // Reached the end of the string, or
+                    spanLength==0       // neither strings nor span progressed.
+                ) {
+                    return pos+spanLength;
+                }
+                pos+=spanLength;
+                rest-=spanLength;
+                continue;  // spanLength>0: Match strings from after a span.
+            } else {
+                // Try to match only one code point from after a string match if some
+                // string matched beyond it, so that we try all possible positions
+                // and don't overshoot.
+                spanLength=spanOneUTF8(spanSet, s+pos, rest);
+                if(spanLength>0) {
+                    if(spanLength==rest) {
+                        return length;  // Reached the end of the string.
+                    }
+                    // Match strings after this code point.
+                    // There cannot be any increments below it because UnicodeSet strings
+                    // contain multiple code points.
+                    pos+=spanLength;
+                    rest-=spanLength;
+                    offsets.shift(spanLength);
+                    spanLength=0;
+                    continue;  // Match strings from after a single code point.
+                }
+                // Match strings from after the next string match.
+            }
+        }
+        int32_t minOffset=offsets.popMinimum();
+        pos+=minOffset;
+        rest-=minOffset;
+        spanLength=0;  // Match strings from after a string match.
+    }
+}
+
+int32_t UnicodeSetStringSpan::spanBackUTF8(const uint8_t *s, int32_t length, USetSpanCondition spanCondition) const {
+    if(spanCondition==USET_SPAN_NOT_CONTAINED) {
+        return spanNotBackUTF8(s, length);
+    }
+    int32_t pos=spanSet.spanBackUTF8((const char *)s, length, USET_SPAN_CONTAINED);
+    if(pos==0) {
+        return 0;
+    }
+    int32_t spanLength=length-pos;
+
+    // Consider strings; they may overlap with the span.
+    OffsetList offsets;
+    if(spanCondition==USET_SPAN_CONTAINED) {
+        // Use offset list to try all possibilities.
+        offsets.setMaxLength(maxLength8);
+    }
+    int32_t i, stringsLength=strings.size();
+    uint8_t *spanBackUTF8Lengths=spanLengths;
+    if(all) {
+        spanBackUTF8Lengths+=3*stringsLength;
+    }
+    for(;;) {
+        const uint8_t *s8=utf8;
+        int32_t length8;
+        if(spanCondition==USET_SPAN_CONTAINED) {
+            for(i=0; i=LONG_SPAN) {
+                    overlap=length8;
+                    // While contained: No point matching fully inside the code point span.
+                    int32_t len1=0;
+                    U8_FWD_1(s8, len1, overlap);
+                    overlap-=len1;  // Length of the string minus the first code point.
+                }
+                if(overlap>spanLength) {
+                    overlap=spanLength;
+                }
+                int32_t dec=length8-overlap;  // Keep dec+overlap==length8.
+                for(;;) {
+                    if(dec>pos) {
+                        break;
+                    }
+                    // Try to match if the decrement is not listed already.
+                    // Match at code point boundaries. (The UTF-8 strings were converted
+                    // from UTF-16 and are guaranteed to be well-formed.)
+                    if( !U8_IS_TRAIL(s[pos-dec]) &&
+                        !offsets.containsOffset(dec) &&
+                        matches8(s+pos-dec, s8, length8)
+                    ) {
+                        if(dec==pos) {
+                            return 0;  // Reached the start of the string.
+                        }
+                        offsets.addOffset(dec);
+                    }
+                    if(overlap==0) {
+                        break;
+                    }
+                    --overlap;
+                    ++dec;
+                }
+                s8+=length8;
+            }
+        } else /* USET_SPAN_SIMPLE */ {
+            int32_t maxDec=0, maxOverlap=0;
+            for(i=0; i=LONG_SPAN) {
+                    overlap=length8;
+                    // Longest match: Need to match fully inside the code point span
+                    // to find the match from the latest end.
+                }
+                if(overlap>spanLength) {
+                    overlap=spanLength;
+                }
+                int32_t dec=length8-overlap;  // Keep dec+overlap==length8.
+                for(;;) {
+                    if(dec>pos || overlapmaxOverlap || /* redundant overlap==maxOverlap && */ dec>maxDec) &&
+                        matches8(s+pos-dec, s8, length8)
+                    ) {
+                        maxDec=dec;  // Longest match from latest end.
+                        maxOverlap=overlap;
+                        break;
+                    }
+                    --overlap;
+                    ++dec;
+                }
+                s8+=length8;
+            }
+
+            if(maxDec!=0 || maxOverlap!=0) {
+                // Longest-match algorithm, and there was a string match.
+                // Simply continue before it.
+                pos-=maxDec;
+                if(pos==0) {
+                    return 0;  // Reached the start of the string.
+                }
+                spanLength=0;  // Match strings from before a string match.
+                continue;
+            }
+        }
+        // Finished trying to match all strings at pos.
+
+        if(spanLength!=0 || pos==length) {
+            // The position is before an unlimited code point span (spanLength!=0),
+            // not before a string match.
+            // The only position where spanLength==0 before a span is pos==length.
+            // Otherwise, an unlimited code point span is only tried again when no
+            // strings match, and if such a non-initial span fails we stop.
+            if(offsets.isEmpty()) {
+                return pos;  // No strings matched before a span.
+            }
+            // Match strings from before the next string match.
+        } else {
+            // The position is before a string match (or a single code point).
+            if(offsets.isEmpty()) {
+                // No more strings matched before a previous string match.
+                // Try another code point span from before the last string match.
+                int32_t oldPos=pos;
+                pos=spanSet.spanBackUTF8((const char *)s, oldPos, USET_SPAN_CONTAINED);
+                spanLength=oldPos-pos;
+                if( pos==0 ||           // Reached the start of the string, or
+                    spanLength==0       // neither strings nor span progressed.
+                ) {
+                    return pos;
+                }
+                continue;  // spanLength>0: Match strings from before a span.
+            } else {
+                // Try to match only one code point from before a string match if some
+                // string matched beyond it, so that we try all possible positions
+                // and don't overshoot.
+                spanLength=spanOneBackUTF8(spanSet, s, pos);
+                if(spanLength>0) {
+                    if(spanLength==pos) {
+                        return 0;  // Reached the start of the string.
+                    }
+                    // Match strings before this code point.
+                    // There cannot be any decrements below it because UnicodeSet strings
+                    // contain multiple code points.
+                    pos-=spanLength;
+                    offsets.shift(spanLength);
+                    spanLength=0;
+                    continue;  // Match strings from before a single code point.
+                }
+                // Match strings from before the next string match.
+            }
+        }
+        pos-=offsets.popMinimum();
+        spanLength=0;  // Match strings from before a string match.
+    }
+}
+
+/*
+ * Algorithm for spanNot()==span(USET_SPAN_NOT_CONTAINED)
+ *
+ * Theoretical algorithm:
+ * - Iterate through the string, and at each code point boundary:
+ *   + If the code point there is in the set, then return with the current position.
+ *   + If a set string matches at the current position, then return with the current position.
+ *
+ * Optimized implementation:
+ *
+ * (Same assumption as for span() above.)
+ *
+ * Create and cache a spanNotSet which contains all of the single code points
+ * of the original set but none of its strings.
+ * For each set string add its initial code point to the spanNotSet.
+ * (Also add its final code point for spanNotBack().)
+ *
+ * - Loop:
+ *   + Do spanLength=spanNotSet.span(USET_SPAN_NOT_CONTAINED).
+ *   + If the current code point is in the original set, then
+ *     return the current position.
+ *   + If any set string matches at the current position, then
+ *     return the current position.
+ *   + If there is no match at the current position, neither for the code point there
+ *     nor for any set string, then skip this code point and continue the loop.
+ *     This happens for set-string-initial code points that were added to spanNotSet
+ *     when there is not actually a match for such a set string.
+ */
+
+int32_t UnicodeSetStringSpan::spanNot(const char16_t *s, int32_t length) const {
+    int32_t pos=0, rest=length;
+    int32_t i, stringsLength=strings.size();
+    do {
+        // Span until we find a code point from the set,
+        // or a code point that starts or ends some string.
+        i=pSpanNotSet->span(s+pos, rest, USET_SPAN_NOT_CONTAINED);
+        if(i==rest) {
+            return length;  // Reached the end of the string.
+        }
+        pos+=i;
+        rest-=i;
+
+        // Check whether the current code point is in the original set,
+        // without the string starts and ends.
+        int32_t cpLength=spanOne(spanSet, s+pos, rest);
+        if(cpLength>0) {
+            return pos;  // There is a set element at pos.
+        }
+
+        // Try to match the strings at pos.
+        for(i=0; i0);
+            if(length16<=rest && matches16CPB(s, pos, length, s16, length16)) {
+                return pos;  // There is a set element at pos.
+            }
+        }
+
+        // The span(while not contained) ended on a string start/end which is
+        // not in the original set. Skip this code point and continue.
+        // cpLength<0
+        pos-=cpLength;
+        rest+=cpLength;
+    } while(rest!=0);
+    return length;  // Reached the end of the string.
+}
+
+int32_t UnicodeSetStringSpan::spanNotBack(const char16_t *s, int32_t length) const {
+    int32_t pos=length;
+    int32_t i, stringsLength=strings.size();
+    do {
+        // Span until we find a code point from the set,
+        // or a code point that starts or ends some string.
+        pos=pSpanNotSet->spanBack(s, pos, USET_SPAN_NOT_CONTAINED);
+        if(pos==0) {
+            return 0;  // Reached the start of the string.
+        }
+
+        // Check whether the current code point is in the original set,
+        // without the string starts and ends.
+        int32_t cpLength=spanOneBack(spanSet, s, pos);
+        if(cpLength>0) {
+            return pos;  // There is a set element at pos.
+        }
+
+        // Try to match the strings at pos.
+        for(i=0; i0);
+            if(length16<=pos && matches16CPB(s, pos-length16, length, s16, length16)) {
+                return pos;  // There is a set element at pos.
+            }
+        }
+
+        // The span(while not contained) ended on a string start/end which is
+        // not in the original set. Skip this code point and continue.
+        // cpLength<0
+        pos+=cpLength;
+    } while(pos!=0);
+    return 0;  // Reached the start of the string.
+}
+
+int32_t UnicodeSetStringSpan::spanNotUTF8(const uint8_t *s, int32_t length) const {
+    int32_t pos=0, rest=length;
+    int32_t i, stringsLength=strings.size();
+    uint8_t *spanUTF8Lengths=spanLengths;
+    if(all) {
+        spanUTF8Lengths+=2*stringsLength;
+    }
+    do {
+        // Span until we find a code point from the set,
+        // or a code point that starts or ends some string.
+        i=pSpanNotSet->spanUTF8((const char *)s+pos, rest, USET_SPAN_NOT_CONTAINED);
+        if(i==rest) {
+            return length;  // Reached the end of the string.
+        }
+        pos+=i;
+        rest-=i;
+
+        // Check whether the current code point is in the original set,
+        // without the string starts and ends.
+        int32_t cpLength=spanOneUTF8(spanSet, s+pos, rest);
+        if(cpLength>0) {
+            return pos;  // There is a set element at pos.
+        }
+
+        // Try to match the strings at pos.
+        const uint8_t *s8=utf8;
+        int32_t length8;
+        for(i=0; ispanBackUTF8((const char *)s, pos, USET_SPAN_NOT_CONTAINED);
+        if(pos==0) {
+            return 0;  // Reached the start of the string.
+        }
+
+        // Check whether the current code point is in the original set,
+        // without the string starts and ends.
+        int32_t cpLength=spanOneBackUTF8(spanSet, s, pos);
+        if(cpLength>0) {
+            return pos;  // There is a set element at pos.
+        }
+
+        // Try to match the strings at pos.
+        const uint8_t *s8=utf8;
+        int32_t length8;
+        for(i=0; i=0xfe.
+        LONG_SPAN=0xfe,
+        // All code points in the string are contained in the parent set.
+        ALL_CP_CONTAINED=0xff
+    };
+
+    // Add a starting or ending string character to the spanNotSet
+    // so that a character span ends before any string.
+    void addToSpanNotSet(UChar32 c);
+
+    int32_t spanNot(const char16_t *s, int32_t length) const;
+    int32_t spanNotBack(const char16_t *s, int32_t length) const;
+    int32_t spanNotUTF8(const uint8_t *s, int32_t length) const;
+    int32_t spanNotBackUTF8(const uint8_t *s, int32_t length) const;
+
+    // Set for span(). Same as parent but without strings.
+    UnicodeSet spanSet;
+
+    // Set for span(not contained).
+    // Same as spanSet, plus characters that start or end strings.
+    UnicodeSet *pSpanNotSet;
+
+    // The strings of the parent set.
+    const UVector &strings;
+
+    // Pointer to the UTF-8 string lengths.
+    // Also pointer to further allocated storage for meta data and
+    // UTF-8 string contents as necessary.
+    int32_t *utf8Lengths;
+
+    // Pointer to the part of the (utf8Lengths) memory block that stores
+    // the lengths of span(), spanBack() etc. for each string.
+    uint8_t *spanLengths;
+
+    // Pointer to the part of the (utf8Lengths) memory block that stores
+    // the UTF-8 versions of the parent set's strings.
+    uint8_t *utf8;
+
+    // Number of bytes for all UTF-8 versions of strings together.
+    int32_t utf8Length;
+
+    // Maximum lengths of relevant strings.
+    int32_t maxLength16;
+    int32_t maxLength8;
+
+    // Set up for all variants of span()?
+    UBool all;
+
+    // Memory for small numbers and lengths of strings.
+    // For example, for 8 strings:
+    // 8 UTF-8 lengths, 8*4 bytes span lengths, 8*2 3-byte UTF-8 characters
+    // = 112 bytes = int32_t[28].
+    int32_t staticLengths[32];
+};
+
+UBool UnicodeSetStringSpan::needsStringSpanUTF16() {
+    return (UBool)(maxLength16!=0);
+}
+
+UBool UnicodeSetStringSpan::needsStringSpanUTF8() {
+    return (UBool)(maxLength8!=0);
+}
+
+UBool UnicodeSetStringSpan::contains(UChar32 c) const {
+    return spanSet.contains(c);
+}
+
+U_NAMESPACE_END
+
+#endif
diff --git a/intl/icu/source/common/unistr.cpp b/intl/icu/source/common/unistr.cpp
new file mode 100644
index 0000000000..04f01cfa16
--- /dev/null
+++ b/intl/icu/source/common/unistr.cpp
@@ -0,0 +1,2030 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+* Copyright (C) 1999-2016, International Business Machines Corporation and
+* others. All Rights Reserved.
+******************************************************************************
+*
+* File unistr.cpp
+*
+* Modification History:
+*
+*   Date        Name        Description
+*   09/25/98    stephen     Creation.
+*   04/20/99    stephen     Overhauled per 4/16 code review.
+*   07/09/99    stephen     Renamed {hi,lo},{byte,word} to icu_X for HP/UX
+*   11/18/99    aliu        Added handleReplaceBetween() to make inherit from
+*                           Replaceable.
+*   06/25/01    grhoten     Removed the dependency on iostream
+******************************************************************************
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/appendable.h"
+#include "unicode/putil.h"
+#include "cstring.h"
+#include "cmemory.h"
+#include "unicode/ustring.h"
+#include "unicode/unistr.h"
+#include "unicode/utf.h"
+#include "unicode/utf16.h"
+#include "uelement.h"
+#include "ustr_imp.h"
+#include "umutex.h"
+#include "uassert.h"
+
+#if 0
+
+#include 
+using namespace std;
+
+//DEBUGGING
+void
+print(const UnicodeString& s,
+      const char *name)
+{
+  char16_t c;
+  cout << name << ":|";
+  for(int i = 0; i < s.length(); ++i) {
+    c = s[i];
+    if(c>= 0x007E || c < 0x0020)
+      cout << "[0x" << hex << s[i] << "]";
+    else
+      cout << (char) s[i];
+  }
+  cout << '|' << endl;
+}
+
+void
+print(const char16_t *s,
+      int32_t len,
+      const char *name)
+{
+  char16_t c;
+  cout << name << ":|";
+  for(int i = 0; i < len; ++i) {
+    c = s[i];
+    if(c>= 0x007E || c < 0x0020)
+      cout << "[0x" << hex << s[i] << "]";
+    else
+      cout << (char) s[i];
+  }
+  cout << '|' << endl;
+}
+// END DEBUGGING
+#endif
+
+// Local function definitions for now
+
+// need to copy areas that may overlap
+static
+inline void
+us_arrayCopy(const char16_t *src, int32_t srcStart,
+         char16_t *dst, int32_t dstStart, int32_t count)
+{
+  if(count>0) {
+    uprv_memmove(dst+dstStart, src+srcStart, (size_t)count*sizeof(*src));
+  }
+}
+
+// u_unescapeAt() callback to get a char16_t from a UnicodeString
+U_CDECL_BEGIN
+static char16_t U_CALLCONV
+UnicodeString_charAt(int32_t offset, void *context) {
+    return ((icu::UnicodeString*) context)->charAt(offset);
+}
+U_CDECL_END
+
+U_NAMESPACE_BEGIN
+
+/* The Replaceable virtual destructor can't be defined in the header
+   due to how AIX works with multiple definitions of virtual functions.
+*/
+Replaceable::~Replaceable() {}
+
+UOBJECT_DEFINE_RTTI_IMPLEMENTATION(UnicodeString)
+
+UnicodeString U_EXPORT2
+operator+ (const UnicodeString &s1, const UnicodeString &s2) {
+    return
+        UnicodeString(s1.length()+s2.length()+1, (UChar32)0, 0).
+            append(s1).
+                append(s2);
+}
+
+//========================================
+// Reference Counting functions, put at top of file so that optimizing compilers
+//                               have a chance to automatically inline.
+//========================================
+
+void
+UnicodeString::addRef() {
+  umtx_atomic_inc((u_atomic_int32_t *)fUnion.fFields.fArray - 1);
+}
+
+int32_t
+UnicodeString::removeRef() {
+  return umtx_atomic_dec((u_atomic_int32_t *)fUnion.fFields.fArray - 1);
+}
+
+int32_t
+UnicodeString::refCount() const {
+  return umtx_loadAcquire(*((u_atomic_int32_t *)fUnion.fFields.fArray - 1));
+}
+
+void
+UnicodeString::releaseArray() {
+  if((fUnion.fFields.fLengthAndFlags & kRefCounted) && removeRef() == 0) {
+    uprv_free((int32_t *)fUnion.fFields.fArray - 1);
+  }
+}
+
+
+
+//========================================
+// Constructors
+//========================================
+
+// The default constructor is inline in unistr.h.
+
+UnicodeString::UnicodeString(int32_t capacity, UChar32 c, int32_t count) {
+  fUnion.fFields.fLengthAndFlags = 0;
+  if(count <= 0 || (uint32_t)c > 0x10ffff) {
+    // just allocate and do not do anything else
+    allocate(capacity);
+  } else if(c <= 0xffff) {
+    int32_t length = count;
+    if(capacity < length) {
+      capacity = length;
+    }
+    if(allocate(capacity)) {
+      char16_t *array = getArrayStart();
+      char16_t unit = (char16_t)c;
+      for(int32_t i = 0; i < length; ++i) {
+        array[i] = unit;
+      }
+      setLength(length);
+    }
+  } else {  // supplementary code point, write surrogate pairs
+    if(count > (INT32_MAX / 2)) {
+      // We would get more than 2G UChars.
+      allocate(capacity);
+      return;
+    }
+    int32_t length = count * 2;
+    if(capacity < length) {
+      capacity = length;
+    }
+    if(allocate(capacity)) {
+      char16_t *array = getArrayStart();
+      char16_t lead = U16_LEAD(c);
+      char16_t trail = U16_TRAIL(c);
+      for(int32_t i = 0; i < length; i += 2) {
+        array[i] = lead;
+        array[i + 1] = trail;
+      }
+      setLength(length);
+    }
+  }
+}
+
+UnicodeString::UnicodeString(char16_t ch) {
+  fUnion.fFields.fLengthAndFlags = kLength1 | kShortString;
+  fUnion.fStackFields.fBuffer[0] = ch;
+}
+
+UnicodeString::UnicodeString(UChar32 ch) {
+  fUnion.fFields.fLengthAndFlags = kShortString;
+  int32_t i = 0;
+  UBool isError = false;
+  U16_APPEND(fUnion.fStackFields.fBuffer, i, US_STACKBUF_SIZE, ch, isError);
+  // We test isError so that the compiler does not complain that we don't.
+  // If isError then i==0 which is what we want anyway.
+  if(!isError) {
+    setShortLength(i);
+  }
+}
+
+UnicodeString::UnicodeString(const char16_t *text) {
+  fUnion.fFields.fLengthAndFlags = kShortString;
+  doAppend(text, 0, -1);
+}
+
+UnicodeString::UnicodeString(const char16_t *text,
+                             int32_t textLength) {
+  fUnion.fFields.fLengthAndFlags = kShortString;
+  doAppend(text, 0, textLength);
+}
+
+UnicodeString::UnicodeString(UBool isTerminated,
+                             ConstChar16Ptr textPtr,
+                             int32_t textLength) {
+  fUnion.fFields.fLengthAndFlags = kReadonlyAlias;
+  const char16_t *text = textPtr;
+  if(text == nullptr) {
+    // treat as an empty string, do not alias
+    setToEmpty();
+  } else if(textLength < -1 ||
+            (textLength == -1 && !isTerminated) ||
+            (textLength >= 0 && isTerminated && text[textLength] != 0)
+  ) {
+    setToBogus();
+  } else {
+    if(textLength == -1) {
+      // text is terminated, or else it would have failed the above test
+      textLength = u_strlen(text);
+    }
+    setArray(const_cast(text), textLength,
+             isTerminated ? textLength + 1 : textLength);
+  }
+}
+
+UnicodeString::UnicodeString(char16_t *buff,
+                             int32_t buffLength,
+                             int32_t buffCapacity) {
+  fUnion.fFields.fLengthAndFlags = kWritableAlias;
+  if(buff == nullptr) {
+    // treat as an empty string, do not alias
+    setToEmpty();
+  } else if(buffLength < -1 || buffCapacity < 0 || buffLength > buffCapacity) {
+    setToBogus();
+  } else {
+    if(buffLength == -1) {
+      // fLength = u_strlen(buff); but do not look beyond buffCapacity
+      const char16_t *p = buff, *limit = buff + buffCapacity;
+      while(p != limit && *p != 0) {
+        ++p;
+      }
+      buffLength = (int32_t)(p - buff);
+    }
+    setArray(buff, buffLength, buffCapacity);
+  }
+}
+
+UnicodeString::UnicodeString(const char *src, int32_t length, EInvariant) {
+  fUnion.fFields.fLengthAndFlags = kShortString;
+  if(src==nullptr) {
+    // treat as an empty string
+  } else {
+    if(length<0) {
+      length=(int32_t)uprv_strlen(src);
+    }
+    if(cloneArrayIfNeeded(length, length, false)) {
+      u_charsToUChars(src, getArrayStart(), length);
+      setLength(length);
+    } else {
+      setToBogus();
+    }
+  }
+}
+
+#if U_CHARSET_IS_UTF8
+
+UnicodeString::UnicodeString(const char *codepageData) {
+  fUnion.fFields.fLengthAndFlags = kShortString;
+  if(codepageData != 0) {
+    setToUTF8(codepageData);
+  }
+}
+
+UnicodeString::UnicodeString(const char *codepageData, int32_t dataLength) {
+  fUnion.fFields.fLengthAndFlags = kShortString;
+  // if there's nothing to convert, do nothing
+  if(codepageData == 0 || dataLength == 0 || dataLength < -1) {
+    return;
+  }
+  if(dataLength == -1) {
+    dataLength = (int32_t)uprv_strlen(codepageData);
+  }
+  setToUTF8(StringPiece(codepageData, dataLength));
+}
+
+// else see unistr_cnv.cpp
+#endif
+
+UnicodeString::UnicodeString(const UnicodeString& that) {
+  fUnion.fFields.fLengthAndFlags = kShortString;
+  copyFrom(that);
+}
+
+UnicodeString::UnicodeString(UnicodeString &&src) noexcept {
+  copyFieldsFrom(src, true);
+}
+
+UnicodeString::UnicodeString(const UnicodeString& that,
+                             int32_t srcStart) {
+  fUnion.fFields.fLengthAndFlags = kShortString;
+  setTo(that, srcStart);
+}
+
+UnicodeString::UnicodeString(const UnicodeString& that,
+                             int32_t srcStart,
+                             int32_t srcLength) {
+  fUnion.fFields.fLengthAndFlags = kShortString;
+  setTo(that, srcStart, srcLength);
+}
+
+// Replaceable base class clone() default implementation, does not clone
+Replaceable *
+Replaceable::clone() const {
+  return nullptr;
+}
+
+// UnicodeString overrides clone() with a real implementation
+UnicodeString *
+UnicodeString::clone() const {
+  LocalPointer clonedString(new UnicodeString(*this));
+  return clonedString.isValid() && !clonedString->isBogus() ? clonedString.orphan() : nullptr;
+}
+
+//========================================
+// array allocation
+//========================================
+
+namespace {
+
+const int32_t kGrowSize = 128;
+
+// The number of bytes for one int32_t reference counter and capacity UChars
+// must fit into a 32-bit size_t (at least when on a 32-bit platform).
+// We also add one for the NUL terminator, to avoid reallocation in getTerminatedBuffer(),
+// and round up to a multiple of 16 bytes.
+// This means that capacity must be at most (0xfffffff0 - 4) / 2 - 1 = 0x7ffffff5.
+// (With more complicated checks we could go up to 0x7ffffffd without rounding up,
+// but that does not seem worth it.)
+const int32_t kMaxCapacity = 0x7ffffff5;
+
+int32_t getGrowCapacity(int32_t newLength) {
+  int32_t growSize = (newLength >> 2) + kGrowSize;
+  if(growSize <= (kMaxCapacity - newLength)) {
+    return newLength + growSize;
+  } else {
+    return kMaxCapacity;
+  }
+}
+
+}  // namespace
+
+UBool
+UnicodeString::allocate(int32_t capacity) {
+  if(capacity <= US_STACKBUF_SIZE) {
+    fUnion.fFields.fLengthAndFlags = kShortString;
+    return true;
+  }
+  if(capacity <= kMaxCapacity) {
+    ++capacity;  // for the NUL
+    // Switch to size_t which is unsigned so that we can allocate up to 4GB.
+    // Reference counter + UChars.
+    size_t numBytes = sizeof(int32_t) + (size_t)capacity * U_SIZEOF_UCHAR;
+    // Round up to a multiple of 16.
+    numBytes = (numBytes + 15) & ~15;
+    int32_t *array = (int32_t *) uprv_malloc(numBytes);
+    if(array != nullptr) {
+      // set initial refCount and point behind the refCount
+      *array++ = 1;
+      numBytes -= sizeof(int32_t);
+
+      // have fArray point to the first char16_t
+      fUnion.fFields.fArray = (char16_t *)array;
+      fUnion.fFields.fCapacity = (int32_t)(numBytes / U_SIZEOF_UCHAR);
+      fUnion.fFields.fLengthAndFlags = kLongString;
+      return true;
+    }
+  }
+  fUnion.fFields.fLengthAndFlags = kIsBogus;
+  fUnion.fFields.fArray = 0;
+  fUnion.fFields.fCapacity = 0;
+  return false;
+}
+
+//========================================
+// Destructor
+//========================================
+
+#ifdef UNISTR_COUNT_FINAL_STRING_LENGTHS
+static u_atomic_int32_t finalLengthCounts[0x400];  // UnicodeString::kMaxShortLength+1
+static u_atomic_int32_t beyondCount(0);
+
+U_CAPI void unistr_printLengths() {
+  int32_t i;
+  for(i = 0; i <= 59; ++i) {
+    printf("%2d,  %9d\n", i, (int32_t)finalLengthCounts[i]);
+  }
+  int32_t beyond = beyondCount;
+  for(; i < UPRV_LENGTHOF(finalLengthCounts); ++i) {
+    beyond += finalLengthCounts[i];
+  }
+  printf(">59, %9d\n", beyond);
+}
+#endif
+
+UnicodeString::~UnicodeString()
+{
+#ifdef UNISTR_COUNT_FINAL_STRING_LENGTHS
+  // Count lengths of strings at the end of their lifetime.
+  // Useful for discussion of a desirable stack buffer size.
+  // Count the contents length, not the optional NUL terminator nor further capacity.
+  // Ignore open-buffer strings and strings which alias external storage.
+  if((fUnion.fFields.fLengthAndFlags&(kOpenGetBuffer|kReadonlyAlias|kWritableAlias)) == 0) {
+    if(hasShortLength()) {
+      umtx_atomic_inc(finalLengthCounts + getShortLength());
+    } else {
+      umtx_atomic_inc(&beyondCount);
+    }
+  }
+#endif
+
+  releaseArray();
+}
+
+//========================================
+// Factory methods
+//========================================
+
+UnicodeString UnicodeString::fromUTF8(StringPiece utf8) {
+  UnicodeString result;
+  result.setToUTF8(utf8);
+  return result;
+}
+
+UnicodeString UnicodeString::fromUTF32(const UChar32 *utf32, int32_t length) {
+  UnicodeString result;
+  int32_t capacity;
+  // Most UTF-32 strings will be BMP-only and result in a same-length
+  // UTF-16 string. We overestimate the capacity just slightly,
+  // just in case there are a few supplementary characters.
+  if(length <= US_STACKBUF_SIZE) {
+    capacity = US_STACKBUF_SIZE;
+  } else {
+    capacity = length + (length >> 4) + 4;
+  }
+  do {
+    char16_t *utf16 = result.getBuffer(capacity);
+    int32_t length16;
+    UErrorCode errorCode = U_ZERO_ERROR;
+    u_strFromUTF32WithSub(utf16, result.getCapacity(), &length16,
+        utf32, length,
+        0xfffd,  // Substitution character.
+        nullptr,    // Don't care about number of substitutions.
+        &errorCode);
+    result.releaseBuffer(length16);
+    if(errorCode == U_BUFFER_OVERFLOW_ERROR) {
+      capacity = length16 + 1;  // +1 for the terminating NUL.
+      continue;
+    } else if(U_FAILURE(errorCode)) {
+      result.setToBogus();
+    }
+    break;
+  } while(true);
+  return result;
+}
+
+//========================================
+// Assignment
+//========================================
+
+UnicodeString &
+UnicodeString::operator=(const UnicodeString &src) {
+  return copyFrom(src);
+}
+
+UnicodeString &
+UnicodeString::fastCopyFrom(const UnicodeString &src) {
+  return copyFrom(src, true);
+}
+
+UnicodeString &
+UnicodeString::copyFrom(const UnicodeString &src, UBool fastCopy) {
+  // if assigning to ourselves, do nothing
+  if(this == &src) {
+    return *this;
+  }
+
+  // is the right side bogus?
+  if(src.isBogus()) {
+    setToBogus();
+    return *this;
+  }
+
+  // delete the current contents
+  releaseArray();
+
+  if(src.isEmpty()) {
+    // empty string - use the stack buffer
+    setToEmpty();
+    return *this;
+  }
+
+  // fLength>0 and not an "open" src.getBuffer(minCapacity)
+  fUnion.fFields.fLengthAndFlags = src.fUnion.fFields.fLengthAndFlags;
+  switch(src.fUnion.fFields.fLengthAndFlags & kAllStorageFlags) {
+  case kShortString:
+    // short string using the stack buffer, do the same
+    uprv_memcpy(fUnion.fStackFields.fBuffer, src.fUnion.fStackFields.fBuffer,
+                getShortLength() * U_SIZEOF_UCHAR);
+    break;
+  case kLongString:
+    // src uses a refCounted string buffer, use that buffer with refCount
+    // src is const, use a cast - we don't actually change it
+    const_cast(src).addRef();
+    // copy all fields, share the reference-counted buffer
+    fUnion.fFields.fArray = src.fUnion.fFields.fArray;
+    fUnion.fFields.fCapacity = src.fUnion.fFields.fCapacity;
+    if(!hasShortLength()) {
+      fUnion.fFields.fLength = src.fUnion.fFields.fLength;
+    }
+    break;
+  case kReadonlyAlias:
+    if(fastCopy) {
+      // src is a readonly alias, do the same
+      // -> maintain the readonly alias as such
+      fUnion.fFields.fArray = src.fUnion.fFields.fArray;
+      fUnion.fFields.fCapacity = src.fUnion.fFields.fCapacity;
+      if(!hasShortLength()) {
+        fUnion.fFields.fLength = src.fUnion.fFields.fLength;
+      }
+      break;
+    }
+    // else if(!fastCopy) fall through to case kWritableAlias
+    // -> allocate a new buffer and copy the contents
+    U_FALLTHROUGH;
+  case kWritableAlias: {
+    // src is a writable alias; we make a copy of that instead
+    int32_t srcLength = src.length();
+    if(allocate(srcLength)) {
+      u_memcpy(getArrayStart(), src.getArrayStart(), srcLength);
+      setLength(srcLength);
+      break;
+    }
+    // if there is not enough memory, then fall through to setting to bogus
+    U_FALLTHROUGH;
+  }
+  default:
+    // if src is bogus, set ourselves to bogus
+    // do not call setToBogus() here because fArray and flags are not consistent here
+    fUnion.fFields.fLengthAndFlags = kIsBogus;
+    fUnion.fFields.fArray = 0;
+    fUnion.fFields.fCapacity = 0;
+    break;
+  }
+
+  return *this;
+}
+
+UnicodeString &UnicodeString::operator=(UnicodeString &&src) noexcept {
+  // No explicit check for self move assignment, consistent with standard library.
+  // Self move assignment causes no crash nor leak but might make the object bogus.
+  releaseArray();
+  copyFieldsFrom(src, true);
+  return *this;
+}
+
+// Same as move assignment except without memory management.
+void UnicodeString::copyFieldsFrom(UnicodeString &src, UBool setSrcToBogus) noexcept {
+  int16_t lengthAndFlags = fUnion.fFields.fLengthAndFlags = src.fUnion.fFields.fLengthAndFlags;
+  if(lengthAndFlags & kUsingStackBuffer) {
+    // Short string using the stack buffer, copy the contents.
+    // Check for self assignment to prevent "overlap in memcpy" warnings,
+    // although it should be harmless to copy a buffer to itself exactly.
+    if(this != &src) {
+      uprv_memcpy(fUnion.fStackFields.fBuffer, src.fUnion.fStackFields.fBuffer,
+                  getShortLength() * U_SIZEOF_UCHAR);
+    }
+  } else {
+    // In all other cases, copy all fields.
+    fUnion.fFields.fArray = src.fUnion.fFields.fArray;
+    fUnion.fFields.fCapacity = src.fUnion.fFields.fCapacity;
+    if(!hasShortLength()) {
+      fUnion.fFields.fLength = src.fUnion.fFields.fLength;
+    }
+    if(setSrcToBogus) {
+      // Set src to bogus without releasing any memory.
+      src.fUnion.fFields.fLengthAndFlags = kIsBogus;
+      src.fUnion.fFields.fArray = nullptr;
+      src.fUnion.fFields.fCapacity = 0;
+    }
+  }
+}
+
+void UnicodeString::swap(UnicodeString &other) noexcept {
+  UnicodeString temp;  // Empty short string: Known not to need releaseArray().
+  // Copy fields without resetting source values in between.
+  temp.copyFieldsFrom(*this, false);
+  this->copyFieldsFrom(other, false);
+  other.copyFieldsFrom(temp, false);
+  // Set temp to an empty string so that other's memory is not released twice.
+  temp.fUnion.fFields.fLengthAndFlags = kShortString;
+}
+
+//========================================
+// Miscellaneous operations
+//========================================
+
+UnicodeString UnicodeString::unescape() const {
+    UnicodeString result(length(), (UChar32)0, (int32_t)0); // construct with capacity
+    if (result.isBogus()) {
+        return result;
+    }
+    const char16_t *array = getBuffer();
+    int32_t len = length();
+    int32_t prev = 0;
+    for (int32_t i=0;;) {
+        if (i == len) {
+            result.append(array, prev, len - prev);
+            break;
+        }
+        if (array[i++] == 0x5C /*'\\'*/) {
+            result.append(array, prev, (i - 1) - prev);
+            UChar32 c = unescapeAt(i); // advances i
+            if (c < 0) {
+                result.remove(); // return empty string
+                break; // invalid escape sequence
+            }
+            result.append(c);
+            prev = i;
+        }
+    }
+    return result;
+}
+
+UChar32 UnicodeString::unescapeAt(int32_t &offset) const {
+    return u_unescapeAt(UnicodeString_charAt, &offset, length(), (void*)this);
+}
+
+//========================================
+// Read-only implementation
+//========================================
+UBool
+UnicodeString::doEquals(const UnicodeString &text, int32_t len) const {
+  // Requires: this & text not bogus and have same lengths.
+  // Byte-wise comparison works for equality regardless of endianness.
+  return uprv_memcmp(getArrayStart(), text.getArrayStart(), len * U_SIZEOF_UCHAR) == 0;
+}
+
+UBool
+UnicodeString::doEqualsSubstring( int32_t start,
+              int32_t length,
+              const char16_t *srcChars,
+              int32_t srcStart,
+              int32_t srcLength) const
+{
+  // compare illegal string values
+  if(isBogus()) {
+    return false;
+  }
+  
+  // pin indices to legal values
+  pinIndices(start, length);
+
+  if(srcChars == nullptr) {
+    // treat const char16_t *srcChars==nullptr as an empty string
+    return length == 0 ? true : false;
+  }
+
+  // get the correct pointer
+  const char16_t *chars = getArrayStart();
+
+  chars += start;
+  srcChars += srcStart;
+
+  // get the srcLength if necessary
+  if(srcLength < 0) {
+    srcLength = u_strlen(srcChars + srcStart);
+  }
+
+  if (length != srcLength) {
+    return false;
+  }
+
+  if(length == 0 || chars == srcChars) {
+    return true;
+  }
+
+  return u_memcmp(chars, srcChars, srcLength) == 0;
+}
+
+int8_t
+UnicodeString::doCompare( int32_t start,
+              int32_t length,
+              const char16_t *srcChars,
+              int32_t srcStart,
+              int32_t srcLength) const
+{
+  // compare illegal string values
+  if(isBogus()) {
+    return -1;
+  }
+  
+  // pin indices to legal values
+  pinIndices(start, length);
+
+  if(srcChars == nullptr) {
+    // treat const char16_t *srcChars==nullptr as an empty string
+    return length == 0 ? 0 : 1;
+  }
+
+  // get the correct pointer
+  const char16_t *chars = getArrayStart();
+
+  chars += start;
+  srcChars += srcStart;
+
+  int32_t minLength;
+  int8_t lengthResult;
+
+  // get the srcLength if necessary
+  if(srcLength < 0) {
+    srcLength = u_strlen(srcChars + srcStart);
+  }
+
+  // are we comparing different lengths?
+  if(length != srcLength) {
+    if(length < srcLength) {
+      minLength = length;
+      lengthResult = -1;
+    } else {
+      minLength = srcLength;
+      lengthResult = 1;
+    }
+  } else {
+    minLength = length;
+    lengthResult = 0;
+  }
+
+  /*
+   * note that uprv_memcmp() returns an int but we return an int8_t;
+   * we need to take care not to truncate the result -
+   * one way to do this is to right-shift the value to
+   * move the sign bit into the lower 8 bits and making sure that this
+   * does not become 0 itself
+   */
+
+  if(minLength > 0 && chars != srcChars) {
+    int32_t result;
+
+#   if U_IS_BIG_ENDIAN 
+      // big-endian: byte comparison works
+      result = uprv_memcmp(chars, srcChars, minLength * sizeof(char16_t));
+      if(result != 0) {
+        return (int8_t)(result >> 15 | 1);
+      }
+#   else
+      // little-endian: compare char16_t units
+      do {
+        result = ((int32_t)*(chars++) - (int32_t)*(srcChars++));
+        if(result != 0) {
+          return (int8_t)(result >> 15 | 1);
+        }
+      } while(--minLength > 0);
+#   endif
+  }
+  return lengthResult;
+}
+
+/* String compare in code point order - doCompare() compares in code unit order. */
+int8_t
+UnicodeString::doCompareCodePointOrder(int32_t start,
+                                       int32_t length,
+                                       const char16_t *srcChars,
+                                       int32_t srcStart,
+                                       int32_t srcLength) const
+{
+  // compare illegal string values
+  // treat const char16_t *srcChars==nullptr as an empty string
+  if(isBogus()) {
+    return -1;
+  }
+
+  // pin indices to legal values
+  pinIndices(start, length);
+
+  if(srcChars == nullptr) {
+    srcStart = srcLength = 0;
+  }
+
+  int32_t diff = uprv_strCompare(getArrayStart() + start, length, (srcChars!=nullptr)?(srcChars + srcStart):nullptr, srcLength, false, true);
+  /* translate the 32-bit result into an 8-bit one */
+  if(diff!=0) {
+    return (int8_t)(diff >> 15 | 1);
+  } else {
+    return 0;
+  }
+}
+
+int32_t
+UnicodeString::getLength() const {
+    return length();
+}
+
+char16_t
+UnicodeString::getCharAt(int32_t offset) const {
+  return charAt(offset);
+}
+
+UChar32
+UnicodeString::getChar32At(int32_t offset) const {
+  return char32At(offset);
+}
+
+UChar32
+UnicodeString::char32At(int32_t offset) const
+{
+  int32_t len = length();
+  if((uint32_t)offset < (uint32_t)len) {
+    const char16_t *array = getArrayStart();
+    UChar32 c;
+    U16_GET(array, 0, offset, len, c);
+    return c;
+  } else {
+    return kInvalidUChar;
+  }
+}
+
+int32_t
+UnicodeString::getChar32Start(int32_t offset) const {
+  if((uint32_t)offset < (uint32_t)length()) {
+    const char16_t *array = getArrayStart();
+    U16_SET_CP_START(array, 0, offset);
+    return offset;
+  } else {
+    return 0;
+  }
+}
+
+int32_t
+UnicodeString::getChar32Limit(int32_t offset) const {
+  int32_t len = length();
+  if((uint32_t)offset < (uint32_t)len) {
+    const char16_t *array = getArrayStart();
+    U16_SET_CP_LIMIT(array, 0, offset, len);
+    return offset;
+  } else {
+    return len;
+  }
+}
+
+int32_t
+UnicodeString::countChar32(int32_t start, int32_t length) const {
+  pinIndices(start, length);
+  // if(isBogus()) then fArray==0 and start==0 - u_countChar32() checks for nullptr
+  return u_countChar32(getArrayStart()+start, length);
+}
+
+UBool
+UnicodeString::hasMoreChar32Than(int32_t start, int32_t length, int32_t number) const {
+  pinIndices(start, length);
+  // if(isBogus()) then fArray==0 and start==0 - u_strHasMoreChar32Than() checks for nullptr
+  return u_strHasMoreChar32Than(getArrayStart()+start, length, number);
+}
+
+int32_t
+UnicodeString::moveIndex32(int32_t index, int32_t delta) const {
+  // pin index
+  int32_t len = length();
+  if(index<0) {
+    index=0;
+  } else if(index>len) {
+    index=len;
+  }
+
+  const char16_t *array = getArrayStart();
+  if(delta>0) {
+    U16_FWD_N(array, index, len, delta);
+  } else {
+    U16_BACK_N(array, 0, index, -delta);
+  }
+
+  return index;
+}
+
+void
+UnicodeString::doExtract(int32_t start,
+             int32_t length,
+             char16_t *dst,
+             int32_t dstStart) const
+{
+  // pin indices to legal values
+  pinIndices(start, length);
+
+  // do not copy anything if we alias dst itself
+  const char16_t *array = getArrayStart();
+  if(array + start != dst + dstStart) {
+    us_arrayCopy(array, start, dst, dstStart, length);
+  }
+}
+
+int32_t
+UnicodeString::extract(Char16Ptr dest, int32_t destCapacity,
+                       UErrorCode &errorCode) const {
+  int32_t len = length();
+  if(U_SUCCESS(errorCode)) {
+    if(isBogus() || destCapacity<0 || (destCapacity>0 && dest==0)) {
+      errorCode=U_ILLEGAL_ARGUMENT_ERROR;
+    } else {
+      const char16_t *array = getArrayStart();
+      if(len>0 && len<=destCapacity && array!=dest) {
+        u_memcpy(dest, array, len);
+      }
+      return u_terminateUChars(dest, destCapacity, len, &errorCode);
+    }
+  }
+
+  return len;
+}
+
+int32_t
+UnicodeString::extract(int32_t start,
+                       int32_t length,
+                       char *target,
+                       int32_t targetCapacity,
+                       enum EInvariant) const
+{
+  // if the arguments are illegal, then do nothing
+  if(targetCapacity < 0 || (targetCapacity > 0 && target == nullptr)) {
+    return 0;
+  }
+
+  // pin the indices to legal values
+  pinIndices(start, length);
+
+  if(length <= targetCapacity) {
+    u_UCharsToChars(getArrayStart() + start, target, length);
+  }
+  UErrorCode status = U_ZERO_ERROR;
+  return u_terminateChars(target, targetCapacity, length, &status);
+}
+
+UnicodeString
+UnicodeString::tempSubString(int32_t start, int32_t len) const {
+  pinIndices(start, len);
+  const char16_t *array = getBuffer();  // not getArrayStart() to check kIsBogus & kOpenGetBuffer
+  if(array==nullptr) {
+    array=fUnion.fStackFields.fBuffer;  // anything not nullptr because that would make an empty string
+    len=-2;  // bogus result string
+  }
+  return UnicodeString(false, array + start, len);
+}
+
+int32_t
+UnicodeString::toUTF8(int32_t start, int32_t len,
+                      char *target, int32_t capacity) const {
+  pinIndices(start, len);
+  int32_t length8;
+  UErrorCode errorCode = U_ZERO_ERROR;
+  u_strToUTF8WithSub(target, capacity, &length8,
+                     getBuffer() + start, len,
+                     0xFFFD,  // Standard substitution character.
+                     nullptr,    // Don't care about number of substitutions.
+                     &errorCode);
+  return length8;
+}
+
+#if U_CHARSET_IS_UTF8
+
+int32_t
+UnicodeString::extract(int32_t start, int32_t len,
+                       char *target, uint32_t dstSize) const {
+  // if the arguments are illegal, then do nothing
+  if(/*dstSize < 0 || */(dstSize > 0 && target == 0)) {
+    return 0;
+  }
+  return toUTF8(start, len, target, dstSize <= 0x7fffffff ? (int32_t)dstSize : 0x7fffffff);
+}
+
+// else see unistr_cnv.cpp
+#endif
+
+void 
+UnicodeString::extractBetween(int32_t start,
+                  int32_t limit,
+                  UnicodeString& target) const {
+  pinIndex(start);
+  pinIndex(limit);
+  doExtract(start, limit - start, target);
+}
+
+// When converting from UTF-16 to UTF-8, the result will have at most 3 times
+// as many bytes as the source has UChars.
+// The "worst cases" are writing systems like Indic, Thai and CJK with
+// 3:1 bytes:UChars.
+void
+UnicodeString::toUTF8(ByteSink &sink) const {
+  int32_t length16 = length();
+  if(length16 != 0) {
+    char stackBuffer[1024];
+    int32_t capacity = (int32_t)sizeof(stackBuffer);
+    UBool utf8IsOwned = false;
+    char *utf8 = sink.GetAppendBuffer(length16 < capacity ? length16 : capacity,
+                                      3*length16,
+                                      stackBuffer, capacity,
+                                      &capacity);
+    int32_t length8 = 0;
+    UErrorCode errorCode = U_ZERO_ERROR;
+    u_strToUTF8WithSub(utf8, capacity, &length8,
+                       getBuffer(), length16,
+                       0xFFFD,  // Standard substitution character.
+                       nullptr,    // Don't care about number of substitutions.
+                       &errorCode);
+    if(errorCode == U_BUFFER_OVERFLOW_ERROR) {
+      utf8 = (char *)uprv_malloc(length8);
+      if(utf8 != nullptr) {
+        utf8IsOwned = true;
+        errorCode = U_ZERO_ERROR;
+        u_strToUTF8WithSub(utf8, length8, &length8,
+                           getBuffer(), length16,
+                           0xFFFD,  // Standard substitution character.
+                           nullptr,    // Don't care about number of substitutions.
+                           &errorCode);
+      } else {
+        errorCode = U_MEMORY_ALLOCATION_ERROR;
+      }
+    }
+    if(U_SUCCESS(errorCode)) {
+      sink.Append(utf8, length8);
+      sink.Flush();
+    }
+    if(utf8IsOwned) {
+      uprv_free(utf8);
+    }
+  }
+}
+
+int32_t
+UnicodeString::toUTF32(UChar32 *utf32, int32_t capacity, UErrorCode &errorCode) const {
+  int32_t length32=0;
+  if(U_SUCCESS(errorCode)) {
+    // getBuffer() and u_strToUTF32WithSub() check for illegal arguments.
+    u_strToUTF32WithSub(utf32, capacity, &length32,
+        getBuffer(), length(),
+        0xfffd,  // Substitution character.
+        nullptr,    // Don't care about number of substitutions.
+        &errorCode);
+  }
+  return length32;
+}
+
+int32_t 
+UnicodeString::indexOf(const char16_t *srcChars,
+               int32_t srcStart,
+               int32_t srcLength,
+               int32_t start,
+               int32_t length) const
+{
+  if(isBogus() || srcChars == 0 || srcStart < 0 || srcLength == 0) {
+    return -1;
+  }
+
+  // UnicodeString does not find empty substrings
+  if(srcLength < 0 && srcChars[srcStart] == 0) {
+    return -1;
+  }
+
+  // get the indices within bounds
+  pinIndices(start, length);
+
+  // find the first occurrence of the substring
+  const char16_t *array = getArrayStart();
+  const char16_t *match = u_strFindFirst(array + start, length, srcChars + srcStart, srcLength);
+  if(match == nullptr) {
+    return -1;
+  } else {
+    return (int32_t)(match - array);
+  }
+}
+
+int32_t
+UnicodeString::doIndexOf(char16_t c,
+             int32_t start,
+             int32_t length) const
+{
+  // pin indices
+  pinIndices(start, length);
+
+  // find the first occurrence of c
+  const char16_t *array = getArrayStart();
+  const char16_t *match = u_memchr(array + start, c, length);
+  if(match == nullptr) {
+    return -1;
+  } else {
+    return (int32_t)(match - array);
+  }
+}
+
+int32_t
+UnicodeString::doIndexOf(UChar32 c,
+                         int32_t start,
+                         int32_t length) const {
+  // pin indices
+  pinIndices(start, length);
+
+  // find the first occurrence of c
+  const char16_t *array = getArrayStart();
+  const char16_t *match = u_memchr32(array + start, c, length);
+  if(match == nullptr) {
+    return -1;
+  } else {
+    return (int32_t)(match - array);
+  }
+}
+
+int32_t 
+UnicodeString::lastIndexOf(const char16_t *srcChars,
+               int32_t srcStart,
+               int32_t srcLength,
+               int32_t start,
+               int32_t length) const
+{
+  if(isBogus() || srcChars == 0 || srcStart < 0 || srcLength == 0) {
+    return -1;
+  }
+
+  // UnicodeString does not find empty substrings
+  if(srcLength < 0 && srcChars[srcStart] == 0) {
+    return -1;
+  }
+
+  // get the indices within bounds
+  pinIndices(start, length);
+
+  // find the last occurrence of the substring
+  const char16_t *array = getArrayStart();
+  const char16_t *match = u_strFindLast(array + start, length, srcChars + srcStart, srcLength);
+  if(match == nullptr) {
+    return -1;
+  } else {
+    return (int32_t)(match - array);
+  }
+}
+
+int32_t
+UnicodeString::doLastIndexOf(char16_t c,
+                 int32_t start,
+                 int32_t length) const
+{
+  if(isBogus()) {
+    return -1;
+  }
+
+  // pin indices
+  pinIndices(start, length);
+
+  // find the last occurrence of c
+  const char16_t *array = getArrayStart();
+  const char16_t *match = u_memrchr(array + start, c, length);
+  if(match == nullptr) {
+    return -1;
+  } else {
+    return (int32_t)(match - array);
+  }
+}
+
+int32_t
+UnicodeString::doLastIndexOf(UChar32 c,
+                             int32_t start,
+                             int32_t length) const {
+  // pin indices
+  pinIndices(start, length);
+
+  // find the last occurrence of c
+  const char16_t *array = getArrayStart();
+  const char16_t *match = u_memrchr32(array + start, c, length);
+  if(match == nullptr) {
+    return -1;
+  } else {
+    return (int32_t)(match - array);
+  }
+}
+
+//========================================
+// Write implementation
+//========================================
+
+UnicodeString& 
+UnicodeString::findAndReplace(int32_t start,
+                  int32_t length,
+                  const UnicodeString& oldText,
+                  int32_t oldStart,
+                  int32_t oldLength,
+                  const UnicodeString& newText,
+                  int32_t newStart,
+                  int32_t newLength)
+{
+  if(isBogus() || oldText.isBogus() || newText.isBogus()) {
+    return *this;
+  }
+
+  pinIndices(start, length);
+  oldText.pinIndices(oldStart, oldLength);
+  newText.pinIndices(newStart, newLength);
+
+  if(oldLength == 0) {
+    return *this;
+  }
+
+  while(length > 0 && length >= oldLength) {
+    int32_t pos = indexOf(oldText, oldStart, oldLength, start, length);
+    if(pos < 0) {
+      // no more oldText's here: done
+      break;
+    } else {
+      // we found oldText, replace it by newText and go beyond it
+      replace(pos, oldLength, newText, newStart, newLength);
+      length -= pos + oldLength - start;
+      start = pos + newLength;
+    }
+  }
+
+  return *this;
+}
+
+
+void
+UnicodeString::setToBogus()
+{
+  releaseArray();
+
+  fUnion.fFields.fLengthAndFlags = kIsBogus;
+  fUnion.fFields.fArray = 0;
+  fUnion.fFields.fCapacity = 0;
+}
+
+// turn a bogus string into an empty one
+void
+UnicodeString::unBogus() {
+  if(fUnion.fFields.fLengthAndFlags & kIsBogus) {
+    setToEmpty();
+  }
+}
+
+const char16_t *
+UnicodeString::getTerminatedBuffer() {
+  if(!isWritable()) {
+    return nullptr;
+  }
+  char16_t *array = getArrayStart();
+  int32_t len = length();
+  if(len < getCapacity()) {
+    if(fUnion.fFields.fLengthAndFlags & kBufferIsReadonly) {
+      // If len= 0 && isTerminated && text[textLength] != 0)
+  ) {
+    setToBogus();
+    return *this;
+  }
+
+  releaseArray();
+
+  if(textLength == -1) {
+    // text is terminated, or else it would have failed the above test
+    textLength = u_strlen(text);
+  }
+  fUnion.fFields.fLengthAndFlags = kReadonlyAlias;
+  setArray((char16_t *)text, textLength, isTerminated ? textLength + 1 : textLength);
+  return *this;
+}
+
+// setTo() analogous to the writable-aliasing constructor with the same signature
+UnicodeString &
+UnicodeString::setTo(char16_t *buffer,
+                     int32_t buffLength,
+                     int32_t buffCapacity) {
+  if(fUnion.fFields.fLengthAndFlags & kOpenGetBuffer) {
+    // do not modify a string that has an "open" getBuffer(minCapacity)
+    return *this;
+  }
+
+  if(buffer == nullptr) {
+    // treat as an empty string, do not alias
+    releaseArray();
+    setToEmpty();
+    return *this;
+  }
+
+  if(buffLength < -1 || buffCapacity < 0 || buffLength > buffCapacity) {
+    setToBogus();
+    return *this;
+  } else if(buffLength == -1) {
+    // buffLength = u_strlen(buff); but do not look beyond buffCapacity
+    const char16_t *p = buffer, *limit = buffer + buffCapacity;
+    while(p != limit && *p != 0) {
+      ++p;
+    }
+    buffLength = (int32_t)(p - buffer);
+  }
+
+  releaseArray();
+
+  fUnion.fFields.fLengthAndFlags = kWritableAlias;
+  setArray(buffer, buffLength, buffCapacity);
+  return *this;
+}
+
+UnicodeString &UnicodeString::setToUTF8(StringPiece utf8) {
+  unBogus();
+  int32_t length = utf8.length();
+  int32_t capacity;
+  // The UTF-16 string will be at most as long as the UTF-8 string.
+  if(length <= US_STACKBUF_SIZE) {
+    capacity = US_STACKBUF_SIZE;
+  } else {
+    capacity = length + 1;  // +1 for the terminating NUL.
+  }
+  char16_t *utf16 = getBuffer(capacity);
+  int32_t length16;
+  UErrorCode errorCode = U_ZERO_ERROR;
+  u_strFromUTF8WithSub(utf16, getCapacity(), &length16,
+      utf8.data(), length,
+      0xfffd,  // Substitution character.
+      nullptr,    // Don't care about number of substitutions.
+      &errorCode);
+  releaseBuffer(length16);
+  if(U_FAILURE(errorCode)) {
+    setToBogus();
+  }
+  return *this;
+}
+
+UnicodeString&
+UnicodeString::setCharAt(int32_t offset,
+             char16_t c)
+{
+  int32_t len = length();
+  if(cloneArrayIfNeeded() && len > 0) {
+    if(offset < 0) {
+      offset = 0;
+    } else if(offset >= len) {
+      offset = len - 1;
+    }
+
+    getArrayStart()[offset] = c;
+  }
+  return *this;
+}
+
+UnicodeString&
+UnicodeString::replace(int32_t start,
+               int32_t _length,
+               UChar32 srcChar) {
+  char16_t buffer[U16_MAX_LENGTH];
+  int32_t count = 0;
+  UBool isError = false;
+  U16_APPEND(buffer, count, U16_MAX_LENGTH, srcChar, isError);
+  // We test isError so that the compiler does not complain that we don't.
+  // If isError (srcChar is not a valid code point) then count==0 which means
+  // we remove the source segment rather than replacing it with srcChar.
+  return doReplace(start, _length, buffer, 0, isError ? 0 : count);
+}
+
+UnicodeString&
+UnicodeString::append(UChar32 srcChar) {
+  char16_t buffer[U16_MAX_LENGTH];
+  int32_t _length = 0;
+  UBool isError = false;
+  U16_APPEND(buffer, _length, U16_MAX_LENGTH, srcChar, isError);
+  // We test isError so that the compiler does not complain that we don't.
+  // If isError then _length==0 which turns the doAppend() into a no-op anyway.
+  return isError ? *this : doAppend(buffer, 0, _length);
+}
+
+UnicodeString&
+UnicodeString::doReplace( int32_t start,
+              int32_t length,
+              const UnicodeString& src,
+              int32_t srcStart,
+              int32_t srcLength)
+{
+  // pin the indices to legal values
+  src.pinIndices(srcStart, srcLength);
+
+  // get the characters from src
+  // and replace the range in ourselves with them
+  return doReplace(start, length, src.getArrayStart(), srcStart, srcLength);
+}
+
+UnicodeString&
+UnicodeString::doReplace(int32_t start,
+             int32_t length,
+             const char16_t *srcChars,
+             int32_t srcStart,
+             int32_t srcLength)
+{
+  if(!isWritable()) {
+    return *this;
+  }
+
+  int32_t oldLength = this->length();
+
+  // optimize (read-only alias).remove(0, start) and .remove(start, end)
+  if((fUnion.fFields.fLengthAndFlags&kBufferIsReadonly) && srcLength == 0) {
+    if(start == 0) {
+      // remove prefix by adjusting the array pointer
+      pinIndex(length);
+      fUnion.fFields.fArray += length;
+      fUnion.fFields.fCapacity -= length;
+      setLength(oldLength - length);
+      return *this;
+    } else {
+      pinIndex(start);
+      if(length >= (oldLength - start)) {
+        // remove suffix by reducing the length (like truncate())
+        setLength(start);
+        fUnion.fFields.fCapacity = start;  // not NUL-terminated any more
+        return *this;
+      }
+    }
+  }
+
+  if(start == oldLength) {
+    return doAppend(srcChars, srcStart, srcLength);
+  }
+
+  if(srcChars == 0) {
+    srcLength = 0;
+  } else {
+    // Perform all remaining operations relative to srcChars + srcStart.
+    // From this point forward, do not use srcStart.
+    srcChars += srcStart;
+    if (srcLength < 0) {
+      // get the srcLength if necessary
+      srcLength = u_strlen(srcChars);
+    }
+  }
+
+  // pin the indices to legal values
+  pinIndices(start, length);
+
+  // Calculate the size of the string after the replace.
+  // Avoid int32_t overflow.
+  int32_t newLength = oldLength - length;
+  if(srcLength > (INT32_MAX - newLength)) {
+    setToBogus();
+    return *this;
+  }
+  newLength += srcLength;
+
+  // Check for insertion into ourself
+  const char16_t *oldArray = getArrayStart();
+  if (isBufferWritable() &&
+      oldArray < srcChars + srcLength &&
+      srcChars < oldArray + oldLength) {
+    // Copy into a new UnicodeString and start over
+    UnicodeString copy(srcChars, srcLength);
+    if (copy.isBogus()) {
+      setToBogus();
+      return *this;
+    }
+    return doReplace(start, length, copy.getArrayStart(), 0, srcLength);
+  }
+
+  // cloneArrayIfNeeded(doCopyArray=false) may change fArray but will not copy the current contents;
+  // therefore we need to keep the current fArray
+  char16_t oldStackBuffer[US_STACKBUF_SIZE];
+  if((fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) && (newLength > US_STACKBUF_SIZE)) {
+    // copy the stack buffer contents because it will be overwritten with
+    // fUnion.fFields values
+    u_memcpy(oldStackBuffer, oldArray, oldLength);
+    oldArray = oldStackBuffer;
+  }
+
+  // clone our array and allocate a bigger array if needed
+  int32_t *bufferToDelete = 0;
+  if(!cloneArrayIfNeeded(newLength, getGrowCapacity(newLength),
+                         false, &bufferToDelete)
+  ) {
+    return *this;
+  }
+
+  // now do the replace
+
+  char16_t *newArray = getArrayStart();
+  if(newArray != oldArray) {
+    // if fArray changed, then we need to copy everything except what will change
+    us_arrayCopy(oldArray, 0, newArray, 0, start);
+    us_arrayCopy(oldArray, start + length,
+                 newArray, start + srcLength,
+                 oldLength - (start + length));
+  } else if(length != srcLength) {
+    // fArray did not change; copy only the portion that isn't changing, leaving a hole
+    us_arrayCopy(oldArray, start + length,
+                 newArray, start + srcLength,
+                 oldLength - (start + length));
+  }
+
+  // now fill in the hole with the new string
+  us_arrayCopy(srcChars, 0, newArray, start, srcLength);
+
+  setLength(newLength);
+
+  // delayed delete in case srcChars == fArray when we started, and
+  // to keep oldArray alive for the above operations
+  if (bufferToDelete) {
+    uprv_free(bufferToDelete);
+  }
+
+  return *this;
+}
+
+// Versions of doReplace() only for append() variants.
+// doReplace() and doAppend() optimize for different cases.
+
+UnicodeString&
+UnicodeString::doAppend(const UnicodeString& src, int32_t srcStart, int32_t srcLength) {
+  if(srcLength == 0) {
+    return *this;
+  }
+
+  // pin the indices to legal values
+  src.pinIndices(srcStart, srcLength);
+  return doAppend(src.getArrayStart(), srcStart, srcLength);
+}
+
+UnicodeString&
+UnicodeString::doAppend(const char16_t *srcChars, int32_t srcStart, int32_t srcLength) {
+  if(!isWritable() || srcLength == 0 || srcChars == nullptr) {
+    return *this;
+  }
+
+  // Perform all remaining operations relative to srcChars + srcStart.
+  // From this point forward, do not use srcStart.
+  srcChars += srcStart;
+
+  if(srcLength < 0) {
+    // get the srcLength if necessary
+    if((srcLength = u_strlen(srcChars)) == 0) {
+      return *this;
+    }
+  }
+
+  int32_t oldLength = length();
+  int32_t newLength;
+  if (uprv_add32_overflow(oldLength, srcLength, &newLength)) {
+    setToBogus();
+    return *this;
+  }
+
+  // Check for append onto ourself
+  const char16_t* oldArray = getArrayStart();
+  if (isBufferWritable() &&
+      oldArray < srcChars + srcLength &&
+      srcChars < oldArray + oldLength) {
+    // Copy into a new UnicodeString and start over
+    UnicodeString copy(srcChars, srcLength);
+    if (copy.isBogus()) {
+      setToBogus();
+      return *this;
+    }
+    return doAppend(copy.getArrayStart(), 0, srcLength);
+  }
+
+  // optimize append() onto a large-enough, owned string
+  if((newLength <= getCapacity() && isBufferWritable()) ||
+      cloneArrayIfNeeded(newLength, getGrowCapacity(newLength))) {
+    char16_t *newArray = getArrayStart();
+    // Do not copy characters when
+    //   char16_t *buffer=str.getAppendBuffer(...);
+    // is followed by
+    //   str.append(buffer, length);
+    // or
+    //   str.appendString(buffer, length)
+    // or similar.
+    if(srcChars != newArray + oldLength) {
+      us_arrayCopy(srcChars, 0, newArray, oldLength, srcLength);
+    }
+    setLength(newLength);
+  }
+  return *this;
+}
+
+/**
+ * Replaceable API
+ */
+void
+UnicodeString::handleReplaceBetween(int32_t start,
+                                    int32_t limit,
+                                    const UnicodeString& text) {
+    replaceBetween(start, limit, text);
+}
+
+/**
+ * Replaceable API
+ */
+void 
+UnicodeString::copy(int32_t start, int32_t limit, int32_t dest) {
+    if (limit <= start) {
+        return; // Nothing to do; avoid bogus malloc call
+    }
+    char16_t* text = (char16_t*) uprv_malloc( sizeof(char16_t) * (limit - start) );
+    // Check to make sure text is not null.
+    if (text != nullptr) {
+	    extractBetween(start, limit, text, 0);
+	    insert(dest, text, 0, limit - start);    
+	    uprv_free(text);
+    }
+}
+
+/**
+ * Replaceable API
+ *
+ * NOTE: This is for the Replaceable class.  There is no rep.cpp,
+ * so we implement this function here.
+ */
+UBool Replaceable::hasMetaData() const {
+    return true;
+}
+
+/**
+ * Replaceable API
+ */
+UBool UnicodeString::hasMetaData() const {
+    return false;
+}
+
+UnicodeString&
+UnicodeString::doReverse(int32_t start, int32_t length) {
+  if(length <= 1 || !cloneArrayIfNeeded()) {
+    return *this;
+  }
+
+  // pin the indices to legal values
+  pinIndices(start, length);
+  if(length <= 1) {  // pinIndices() might have shrunk the length
+    return *this;
+  }
+
+  char16_t *left = getArrayStart() + start;
+  char16_t *right = left + length - 1;  // -1 for inclusive boundary (length>=2)
+  char16_t swap;
+  UBool hasSupplementary = false;
+
+  // Before the loop we know left=2.
+  do {
+    hasSupplementary |= (UBool)U16_IS_LEAD(swap = *left);
+    hasSupplementary |= (UBool)U16_IS_LEAD(*left++ = *right);
+    *right-- = swap;
+  } while(left < right);
+  // Make sure to test the middle code unit of an odd-length string.
+  // Redundant if the length is even.
+  hasSupplementary |= (UBool)U16_IS_LEAD(*left);
+
+  /* if there are supplementary code points in the reversed range, then re-swap their surrogates */
+  if(hasSupplementary) {
+    char16_t swap2;
+
+    left = getArrayStart() + start;
+    right = left + length - 1; // -1 so that we can look at *(left+1) if left= targetLength || !cloneArrayIfNeeded(targetLength)) {
+    return false;
+  } else {
+    // move contents up by padding width
+    char16_t *array = getArrayStart();
+    int32_t start = targetLength - oldLength;
+    us_arrayCopy(array, 0, array, start, oldLength);
+
+    // fill in padding character
+    while(--start >= 0) {
+      array[start] = padChar;
+    }
+    setLength(targetLength);
+    return true;
+  }
+}
+
+UBool 
+UnicodeString::padTrailing(int32_t targetLength,
+                           char16_t padChar)
+{
+  int32_t oldLength = length();
+  if(oldLength >= targetLength || !cloneArrayIfNeeded(targetLength)) {
+    return false;
+  } else {
+    // fill in padding character
+    char16_t *array = getArrayStart();
+    int32_t length = targetLength;
+    while(--length >= oldLength) {
+      array[length] = padChar;
+    }
+    setLength(targetLength);
+    return true;
+  }
+}
+
+//========================================
+// Hashing
+//========================================
+int32_t
+UnicodeString::doHashCode() const
+{
+    /* Delegate hash computation to uhash.  This makes UnicodeString
+     * hashing consistent with char16_t* hashing.  */
+    int32_t hashCode = ustr_hashUCharsN(getArrayStart(), length());
+    if (hashCode == kInvalidHashCode) {
+        hashCode = kEmptyHashCode;
+    }
+    return hashCode;
+}
+
+//========================================
+// External Buffer
+//========================================
+
+char16_t *
+UnicodeString::getBuffer(int32_t minCapacity) {
+  if(minCapacity>=-1 && cloneArrayIfNeeded(minCapacity)) {
+    fUnion.fFields.fLengthAndFlags|=kOpenGetBuffer;
+    setZeroLength();
+    return getArrayStart();
+  } else {
+    return nullptr;
+  }
+}
+
+void
+UnicodeString::releaseBuffer(int32_t newLength) {
+  if(fUnion.fFields.fLengthAndFlags&kOpenGetBuffer && newLength>=-1) {
+    // set the new fLength
+    int32_t capacity=getCapacity();
+    if(newLength==-1) {
+      // the new length is the string length, capped by fCapacity
+      const char16_t *array=getArrayStart(), *p=array, *limit=array+capacity;
+      while(pcapacity) {
+      newLength=capacity;
+    }
+    setLength(newLength);
+    fUnion.fFields.fLengthAndFlags&=~kOpenGetBuffer;
+  }
+}
+
+//========================================
+// Miscellaneous
+//========================================
+UBool
+UnicodeString::cloneArrayIfNeeded(int32_t newCapacity,
+                                  int32_t growCapacity,
+                                  UBool doCopyArray,
+                                  int32_t **pBufferToDelete,
+                                  UBool forceClone) {
+  // default parameters need to be static, therefore
+  // the defaults are -1 to have convenience defaults
+  if(newCapacity == -1) {
+    newCapacity = getCapacity();
+  }
+
+  // while a getBuffer(minCapacity) is "open",
+  // prevent any modifications of the string by returning false here
+  // if the string is bogus, then only an assignment or similar can revive it
+  if(!isWritable()) {
+    return false;
+  }
+
+  /*
+   * We need to make a copy of the array if
+   * the buffer is read-only, or
+   * the buffer is refCounted (shared), and refCount>1, or
+   * the buffer is too small.
+   * Return false if memory could not be allocated.
+   */
+  if(forceClone ||
+     fUnion.fFields.fLengthAndFlags & kBufferIsReadonly ||
+     (fUnion.fFields.fLengthAndFlags & kRefCounted && refCount() > 1) ||
+     newCapacity > getCapacity()
+  ) {
+    // check growCapacity for default value and use of the stack buffer
+    if(growCapacity < 0) {
+      growCapacity = newCapacity;
+    } else if(newCapacity <= US_STACKBUF_SIZE && growCapacity > US_STACKBUF_SIZE) {
+      growCapacity = US_STACKBUF_SIZE;
+    }
+
+    // save old values
+    char16_t oldStackBuffer[US_STACKBUF_SIZE];
+    char16_t *oldArray;
+    int32_t oldLength = length();
+    int16_t flags = fUnion.fFields.fLengthAndFlags;
+
+    if(flags&kUsingStackBuffer) {
+      U_ASSERT(!(flags&kRefCounted)); /* kRefCounted and kUsingStackBuffer are mutally exclusive */
+      if(doCopyArray && growCapacity > US_STACKBUF_SIZE) {
+        // copy the stack buffer contents because it will be overwritten with
+        // fUnion.fFields values
+        us_arrayCopy(fUnion.fStackFields.fBuffer, 0, oldStackBuffer, 0, oldLength);
+        oldArray = oldStackBuffer;
+      } else {
+        oldArray = nullptr; // no need to copy from the stack buffer to itself
+      }
+    } else {
+      oldArray = fUnion.fFields.fArray;
+      U_ASSERT(oldArray!=nullptr); /* when stack buffer is not used, oldArray must have a non-nullptr reference */
+    }
+
+    // allocate a new array
+    if(allocate(growCapacity) ||
+       (newCapacity < growCapacity && allocate(newCapacity))
+    ) {
+      if(doCopyArray) {
+        // copy the contents
+        // do not copy more than what fits - it may be smaller than before
+        int32_t minLength = oldLength;
+        newCapacity = getCapacity();
+        if(newCapacity < minLength) {
+          minLength = newCapacity;
+        }
+        if(oldArray != nullptr) {
+          us_arrayCopy(oldArray, 0, getArrayStart(), 0, minLength);
+        }
+        setLength(minLength);
+      } else {
+        setZeroLength();
+      }
+
+      // release the old array
+      if(flags & kRefCounted) {
+        // the array is refCounted; decrement and release if 0
+        u_atomic_int32_t *pRefCount = ((u_atomic_int32_t *)oldArray - 1);
+        if(umtx_atomic_dec(pRefCount) == 0) {
+          if(pBufferToDelete == 0) {
+              // Note: cast to (void *) is needed with MSVC, where u_atomic_int32_t
+              // is defined as volatile. (Volatile has useful non-standard behavior
+              //   with this compiler.)
+            uprv_free((void *)pRefCount);
+          } else {
+            // the caller requested to delete it himself
+            *pBufferToDelete = (int32_t *)pRefCount;
+          }
+        }
+      }
+    } else {
+      // not enough memory for growCapacity and not even for the smaller newCapacity
+      // reset the old values for setToBogus() to release the array
+      if(!(flags&kUsingStackBuffer)) {
+        fUnion.fFields.fArray = oldArray;
+      }
+      fUnion.fFields.fLengthAndFlags = flags;
+      setToBogus();
+      return false;
+    }
+  }
+  return true;
+}
+
+// UnicodeStringAppendable ------------------------------------------------- ***
+
+UnicodeStringAppendable::~UnicodeStringAppendable() {}
+
+UBool
+UnicodeStringAppendable::appendCodeUnit(char16_t c) {
+  return str.doAppend(&c, 0, 1).isWritable();
+}
+
+UBool
+UnicodeStringAppendable::appendCodePoint(UChar32 c) {
+  char16_t buffer[U16_MAX_LENGTH];
+  int32_t cLength = 0;
+  UBool isError = false;
+  U16_APPEND(buffer, cLength, U16_MAX_LENGTH, c, isError);
+  return !isError && str.doAppend(buffer, 0, cLength).isWritable();
+}
+
+UBool
+UnicodeStringAppendable::appendString(const char16_t *s, int32_t length) {
+  return str.doAppend(s, 0, length).isWritable();
+}
+
+UBool
+UnicodeStringAppendable::reserveAppendCapacity(int32_t appendCapacity) {
+  return str.cloneArrayIfNeeded(str.length() + appendCapacity);
+}
+
+char16_t *
+UnicodeStringAppendable::getAppendBuffer(int32_t minCapacity,
+                                         int32_t desiredCapacityHint,
+                                         char16_t *scratch, int32_t scratchCapacity,
+                                         int32_t *resultCapacity) {
+  if(minCapacity < 1 || scratchCapacity < minCapacity) {
+    *resultCapacity = 0;
+    return nullptr;
+  }
+  int32_t oldLength = str.length();
+  if(minCapacity <= (kMaxCapacity - oldLength) &&
+      desiredCapacityHint <= (kMaxCapacity - oldLength) &&
+      str.cloneArrayIfNeeded(oldLength + minCapacity, oldLength + desiredCapacityHint)) {
+    *resultCapacity = str.getCapacity() - oldLength;
+    return str.getArrayStart() + oldLength;
+  }
+  *resultCapacity = scratchCapacity;
+  return scratch;
+}
+
+U_NAMESPACE_END
+
+U_NAMESPACE_USE
+
+U_CAPI int32_t U_EXPORT2
+uhash_hashUnicodeString(const UElement key) {
+    const UnicodeString *str = (const UnicodeString*) key.pointer;
+    return (str == nullptr) ? 0 : str->hashCode();
+}
+
+// Moved here from uhash_us.cpp so that using a UVector of UnicodeString*
+// does not depend on hashtable code.
+U_CAPI UBool U_EXPORT2
+uhash_compareUnicodeString(const UElement key1, const UElement key2) {
+    const UnicodeString *str1 = (const UnicodeString*) key1.pointer;
+    const UnicodeString *str2 = (const UnicodeString*) key2.pointer;
+    if (str1 == str2) {
+        return true;
+    }
+    if (str1 == nullptr || str2 == nullptr) {
+        return false;
+    }
+    return *str1 == *str2;
+}
+
+#ifdef U_STATIC_IMPLEMENTATION
+/*
+This should never be called. It is defined here to make sure that the
+virtual vector deleting destructor is defined within unistr.cpp.
+The vector deleting destructor is already a part of UObject,
+but defining it here makes sure that it is included with this object file.
+This makes sure that static library dependencies are kept to a minimum.
+*/
+#if defined(__clang__) || U_GCC_MAJOR_MINOR >= 1100
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wunused-function"
+static void uprv_UnicodeStringDummy() {
+    delete [] (new UnicodeString[2]);
+}
+#pragma GCC diagnostic pop
+#endif
+#endif
diff --git a/intl/icu/source/common/unistr_case.cpp b/intl/icu/source/common/unistr_case.cpp
new file mode 100644
index 0000000000..57f307c203
--- /dev/null
+++ b/intl/icu/source/common/unistr_case.cpp
@@ -0,0 +1,250 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 1999-2014, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  unistr_case.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:2
+*
+*   created on: 2004aug19
+*   created by: Markus W. Scherer
+*
+*   Case-mapping functions moved here from unistr.cpp
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/brkiter.h"
+#include "unicode/casemap.h"
+#include "unicode/edits.h"
+#include "unicode/putil.h"
+#include "cstring.h"
+#include "cmemory.h"
+#include "unicode/ustring.h"
+#include "unicode/unistr.h"
+#include "unicode/uchar.h"
+#include "uassert.h"
+#include "ucasemap_imp.h"
+#include "uelement.h"
+
+U_NAMESPACE_BEGIN
+
+//========================================
+// Read-only implementation
+//========================================
+
+int8_t
+UnicodeString::doCaseCompare(int32_t start,
+                             int32_t length,
+                             const char16_t *srcChars,
+                             int32_t srcStart,
+                             int32_t srcLength,
+                             uint32_t options) const
+{
+  // compare illegal string values
+  // treat const char16_t *srcChars==nullptr as an empty string
+  if(isBogus()) {
+    return -1;
+  }
+
+  // pin indices to legal values
+  pinIndices(start, length);
+
+  if(srcChars == nullptr) {
+    srcStart = srcLength = 0;
+  }
+
+  // get the correct pointer
+  const char16_t *chars = getArrayStart();
+
+  chars += start;
+  if(srcStart!=0) {
+    srcChars += srcStart;
+  }
+
+  if(chars != srcChars) {
+    UErrorCode errorCode=U_ZERO_ERROR;
+    int32_t result=u_strcmpFold(chars, length, srcChars, srcLength,
+                                options|U_COMPARE_IGNORE_CASE, &errorCode);
+    if(result!=0) {
+      return (int8_t)(result >> 24 | 1);
+    }
+  } else {
+    // get the srcLength if necessary
+    if(srcLength < 0) {
+      srcLength = u_strlen(srcChars + srcStart);
+    }
+    if(length != srcLength) {
+      return (int8_t)((length - srcLength) >> 24 | 1);
+    }
+  }
+  return 0;
+}
+
+//========================================
+// Write implementation
+//========================================
+
+UnicodeString &
+UnicodeString::caseMap(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM
+                       UStringCaseMapper *stringCaseMapper) {
+  if(isEmpty() || !isWritable()) {
+    // nothing to do
+    return *this;
+  }
+
+  char16_t oldBuffer[2 * US_STACKBUF_SIZE];
+  char16_t *oldArray;
+  int32_t oldLength = length();
+  int32_t newLength;
+  UBool writable = isBufferWritable();
+  UErrorCode errorCode = U_ZERO_ERROR;
+
+#if !UCONFIG_NO_BREAK_ITERATION
+  // Read-only alias to the original string contents for the titlecasing BreakIterator.
+  // We cannot set the iterator simply to *this because *this is being modified.
+  UnicodeString oldString;
+#endif
+
+  // Try to avoid heap-allocating a new character array for this string.
+  if (writable ? oldLength <= UPRV_LENGTHOF(oldBuffer) : oldLength < US_STACKBUF_SIZE) {
+    // Short string: Copy the contents into a temporary buffer and
+    // case-map back into the current array, or into the stack buffer.
+    char16_t *buffer = getArrayStart();
+    int32_t capacity;
+    oldArray = oldBuffer;
+    u_memcpy(oldBuffer, buffer, oldLength);
+    if (writable) {
+      capacity = getCapacity();
+    } else {
+      // Switch from the read-only alias or shared heap buffer to the stack buffer.
+      if (!cloneArrayIfNeeded(US_STACKBUF_SIZE, US_STACKBUF_SIZE, /* doCopyArray= */ false)) {
+        return *this;
+      }
+      U_ASSERT(fUnion.fFields.fLengthAndFlags & kUsingStackBuffer);
+      buffer = fUnion.fStackFields.fBuffer;
+      capacity = US_STACKBUF_SIZE;
+    }
+#if !UCONFIG_NO_BREAK_ITERATION
+    if (iter != nullptr) {
+      oldString.setTo(false, oldArray, oldLength);
+      iter->setText(oldString);
+    }
+#endif
+    newLength = stringCaseMapper(caseLocale, options, UCASEMAP_BREAK_ITERATOR
+                                 buffer, capacity,
+                                 oldArray, oldLength, nullptr, errorCode);
+    if (U_SUCCESS(errorCode)) {
+      setLength(newLength);
+      return *this;
+    } else if (errorCode == U_BUFFER_OVERFLOW_ERROR) {
+      // common overflow handling below
+    } else {
+      setToBogus();
+      return *this;
+    }
+  } else {
+    // Longer string or read-only buffer:
+    // Collect only changes and then apply them to this string.
+    // Case mapping often changes only small parts of a string,
+    // and often does not change its length.
+    oldArray = getArrayStart();
+    Edits edits;
+    char16_t replacementChars[200];
+#if !UCONFIG_NO_BREAK_ITERATION
+    if (iter != nullptr) {
+      oldString.setTo(false, oldArray, oldLength);
+      iter->setText(oldString);
+    }
+#endif
+    stringCaseMapper(caseLocale, options | U_OMIT_UNCHANGED_TEXT, UCASEMAP_BREAK_ITERATOR
+                     replacementChars, UPRV_LENGTHOF(replacementChars),
+                     oldArray, oldLength, &edits, errorCode);
+    if (U_SUCCESS(errorCode)) {
+      // Grow the buffer at most once, not for multiple doReplace() calls.
+      newLength = oldLength + edits.lengthDelta();
+      if (newLength > oldLength && !cloneArrayIfNeeded(newLength, newLength)) {
+        return *this;
+      }
+      for (Edits::Iterator ei = edits.getCoarseChangesIterator(); ei.next(errorCode);) {
+        doReplace(ei.destinationIndex(), ei.oldLength(),
+                  replacementChars, ei.replacementIndex(), ei.newLength());
+      }
+      if (U_FAILURE(errorCode)) {
+        setToBogus();
+      }
+      return *this;
+    } else if (errorCode == U_BUFFER_OVERFLOW_ERROR) {
+      // common overflow handling below
+      newLength = oldLength + edits.lengthDelta();
+    } else {
+      setToBogus();
+      return *this;
+    }
+  }
+
+  // Handle buffer overflow, newLength is known.
+  // We need to allocate a new buffer for the internal string case mapping function.
+  // This is very similar to how doReplace() keeps the old array pointer
+  // and deletes the old array itself after it is done.
+  // In addition, we are forcing cloneArrayIfNeeded() to always allocate a new array.
+  int32_t *bufferToDelete = 0;
+  if (!cloneArrayIfNeeded(newLength, newLength, false, &bufferToDelete, true)) {
+    return *this;
+  }
+  errorCode = U_ZERO_ERROR;
+  // No need to iter->setText() again: The case mapper restarts via iter->first().
+  newLength = stringCaseMapper(caseLocale, options, UCASEMAP_BREAK_ITERATOR
+                               getArrayStart(), getCapacity(),
+                               oldArray, oldLength, nullptr, errorCode);
+  if (bufferToDelete) {
+    uprv_free(bufferToDelete);
+  }
+  if (U_SUCCESS(errorCode)) {
+    setLength(newLength);
+  } else {
+    setToBogus();
+  }
+  return *this;
+}
+
+UnicodeString &
+UnicodeString::foldCase(uint32_t options) {
+  return caseMap(UCASE_LOC_ROOT, options, UCASEMAP_BREAK_ITERATOR_NULL ustrcase_internalFold);
+}
+
+U_NAMESPACE_END
+
+// Defined here to reduce dependencies on break iterator
+U_CAPI int32_t U_EXPORT2
+uhash_hashCaselessUnicodeString(const UElement key) {
+    U_NAMESPACE_USE
+    const UnicodeString *str = (const UnicodeString*) key.pointer;
+    if (str == nullptr) {
+        return 0;
+    }
+    // Inefficient; a better way would be to have a hash function in
+    // UnicodeString that does case folding on the fly.
+    UnicodeString copy(*str);
+    return copy.foldCase().hashCode();
+}
+
+// Defined here to reduce dependencies on break iterator
+U_CAPI UBool U_EXPORT2
+uhash_compareCaselessUnicodeString(const UElement key1, const UElement key2) {
+    U_NAMESPACE_USE
+    const UnicodeString *str1 = (const UnicodeString*) key1.pointer;
+    const UnicodeString *str2 = (const UnicodeString*) key2.pointer;
+    if (str1 == str2) {
+        return true;
+    }
+    if (str1 == nullptr || str2 == nullptr) {
+        return false;
+    }
+    return str1->caseCompare(*str2, U_FOLD_CASE_DEFAULT) == 0;
+}
diff --git a/intl/icu/source/common/unistr_case_locale.cpp b/intl/icu/source/common/unistr_case_locale.cpp
new file mode 100644
index 0000000000..8b6a9ca0ca
--- /dev/null
+++ b/intl/icu/source/common/unistr_case_locale.cpp
@@ -0,0 +1,56 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*   Copyright (C) 2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*******************************************************************************
+*   file name:  unistr_case_locale.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2011may31
+*   created by: Markus W. Scherer
+*
+*   Locale-sensitive case mapping functions (ones that call uloc_getDefault())
+*   were moved here to break dependency cycles among parts of the common library.
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/locid.h"
+#include "unicode/ucasemap.h"
+#include "unicode/unistr.h"
+#include "ucasemap_imp.h"
+
+U_NAMESPACE_BEGIN
+
+//========================================
+// Write implementation
+//========================================
+
+UnicodeString &
+UnicodeString::toLower() {
+  return caseMap(ustrcase_getCaseLocale(nullptr), 0,
+                 UCASEMAP_BREAK_ITERATOR_NULL ustrcase_internalToLower);
+}
+
+UnicodeString &
+UnicodeString::toLower(const Locale &locale) {
+  return caseMap(ustrcase_getCaseLocale(locale.getBaseName()), 0,
+                 UCASEMAP_BREAK_ITERATOR_NULL ustrcase_internalToLower);
+}
+
+UnicodeString &
+UnicodeString::toUpper() {
+  return caseMap(ustrcase_getCaseLocale(nullptr), 0,
+                 UCASEMAP_BREAK_ITERATOR_NULL ustrcase_internalToUpper);
+}
+
+UnicodeString &
+UnicodeString::toUpper(const Locale &locale) {
+  return caseMap(ustrcase_getCaseLocale(locale.getBaseName()), 0,
+                 UCASEMAP_BREAK_ITERATOR_NULL ustrcase_internalToUpper);
+}
+
+U_NAMESPACE_END
diff --git a/intl/icu/source/common/unistr_cnv.cpp b/intl/icu/source/common/unistr_cnv.cpp
new file mode 100644
index 0000000000..2d649b2d51
--- /dev/null
+++ b/intl/icu/source/common/unistr_cnv.cpp
@@ -0,0 +1,417 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 1999-2014, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  unistr_cnv.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:2
+*
+*   created on: 2004aug19
+*   created by: Markus W. Scherer
+*
+*   Character conversion functions moved here from unistr.cpp
+*/
+
+#include "unicode/utypes.h"
+
+#if !UCONFIG_NO_CONVERSION
+
+#include "unicode/putil.h"
+#include "cstring.h"
+#include "cmemory.h"
+#include "unicode/ustring.h"
+#include "unicode/unistr.h"
+#include "unicode/ucnv.h"
+#include "ucnv_imp.h"
+#include "putilimp.h"
+#include "ustr_cnv.h"
+#include "ustr_imp.h"
+
+U_NAMESPACE_BEGIN
+
+//========================================
+// Constructors
+//========================================
+
+#if !U_CHARSET_IS_UTF8
+
+UnicodeString::UnicodeString(const char *codepageData) {
+    fUnion.fFields.fLengthAndFlags = kShortString;
+    if(codepageData != 0) {
+        doCodepageCreate(codepageData, (int32_t)uprv_strlen(codepageData), 0);
+    }
+}
+
+UnicodeString::UnicodeString(const char *codepageData,
+                             int32_t dataLength) {
+    fUnion.fFields.fLengthAndFlags = kShortString;
+    if(codepageData != 0) {
+        doCodepageCreate(codepageData, dataLength, 0);
+    }
+}
+
+// else see unistr.cpp
+#endif
+
+UnicodeString::UnicodeString(const char *codepageData,
+                             const char *codepage) {
+    fUnion.fFields.fLengthAndFlags = kShortString;
+    if(codepageData != 0) {
+        doCodepageCreate(codepageData, (int32_t)uprv_strlen(codepageData), codepage);
+    }
+}
+
+UnicodeString::UnicodeString(const char *codepageData,
+                             int32_t dataLength,
+                             const char *codepage) {
+    fUnion.fFields.fLengthAndFlags = kShortString;
+    if(codepageData != 0) {
+        doCodepageCreate(codepageData, dataLength, codepage);
+    }
+}
+
+UnicodeString::UnicodeString(const char *src, int32_t srcLength,
+                             UConverter *cnv,
+                             UErrorCode &errorCode) {
+    fUnion.fFields.fLengthAndFlags = kShortString;
+    if(U_SUCCESS(errorCode)) {
+        // check arguments
+        if(src==nullptr) {
+            // treat as an empty string, do nothing more
+        } else if(srcLength<-1) {
+            errorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        } else {
+            // get input length
+            if(srcLength==-1) {
+                srcLength=(int32_t)uprv_strlen(src);
+            }
+            if(srcLength>0) {
+                if(cnv!=0) {
+                    // use the provided converter
+                    ucnv_resetToUnicode(cnv);
+                    doCodepageCreate(src, srcLength, cnv, errorCode);
+                } else {
+                    // use the default converter
+                    cnv=u_getDefaultConverter(&errorCode);
+                    doCodepageCreate(src, srcLength, cnv, errorCode);
+                    u_releaseDefaultConverter(cnv);
+                }
+            }
+        }
+
+        if(U_FAILURE(errorCode)) {
+            setToBogus();
+        }
+    }
+}
+
+//========================================
+// Codeset conversion
+//========================================
+
+#if !U_CHARSET_IS_UTF8
+
+int32_t
+UnicodeString::extract(int32_t start,
+                       int32_t length,
+                       char *target,
+                       uint32_t dstSize) const {
+    return extract(start, length, target, dstSize, 0);
+}
+
+// else see unistr.cpp
+#endif
+
+int32_t
+UnicodeString::extract(int32_t start,
+                       int32_t length,
+                       char *target,
+                       uint32_t dstSize,
+                       const char *codepage) const
+{
+    // if the arguments are illegal, then do nothing
+    if(/*dstSize < 0 || */(dstSize > 0 && target == 0)) {
+        return 0;
+    }
+
+    // pin the indices to legal values
+    pinIndices(start, length);
+
+    // We need to cast dstSize to int32_t for all subsequent code.
+    // I don't know why the API was defined with uint32_t but we are stuck with it.
+    // Also, dstSize==0xffffffff means "unlimited" but if we use target+dstSize
+    // as a limit in some functions, it may wrap around and yield a pointer
+    // that compares less-than target.
+    int32_t capacity;
+    if(dstSize < 0x7fffffff) {
+        // Assume that the capacity is real and a limit pointer won't wrap around.
+        capacity = (int32_t)dstSize;
+    } else {
+        // Pin the capacity so that a limit pointer does not wrap around.
+        char *targetLimit = (char *)U_MAX_PTR(target);
+        // U_MAX_PTR(target) returns a targetLimit that is at most 0x7fffffff
+        // greater than target and does not wrap around the top of the address space.
+        capacity = (int32_t)(targetLimit - target);
+    }
+
+    // create the converter
+    UConverter *converter;
+    UErrorCode status = U_ZERO_ERROR;
+
+    // just write the NUL if the string length is 0
+    if(length == 0) {
+        return u_terminateChars(target, capacity, 0, &status);
+    }
+
+    // if the codepage is the default, use our cache
+    // if it is an empty string, then use the "invariant character" conversion
+    if (codepage == 0) {
+        const char *defaultName = ucnv_getDefaultName();
+        if(UCNV_FAST_IS_UTF8(defaultName)) {
+            return toUTF8(start, length, target, capacity);
+        }
+        converter = u_getDefaultConverter(&status);
+    } else if (*codepage == 0) {
+        // use the "invariant characters" conversion
+        int32_t destLength;
+        if(length <= capacity) {
+            destLength = length;
+        } else {
+            destLength = capacity;
+        }
+        u_UCharsToChars(getArrayStart() + start, target, destLength);
+        return u_terminateChars(target, capacity, length, &status);
+    } else {
+        converter = ucnv_open(codepage, &status);
+    }
+
+    length = doExtract(start, length, target, capacity, converter, status);
+
+    // close the converter
+    if (codepage == 0) {
+        u_releaseDefaultConverter(converter);
+    } else {
+        ucnv_close(converter);
+    }
+
+    return length;
+}
+
+int32_t
+UnicodeString::extract(char *dest, int32_t destCapacity,
+                       UConverter *cnv,
+                       UErrorCode &errorCode) const
+{
+    if(U_FAILURE(errorCode)) {
+        return 0;
+    }
+
+    if(isBogus() || destCapacity<0 || (destCapacity>0 && dest==0)) {
+        errorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    // nothing to do?
+    if(isEmpty()) {
+        return u_terminateChars(dest, destCapacity, 0, &errorCode);
+    }
+
+    // get the converter
+    UBool isDefaultConverter;
+    if(cnv==0) {
+        isDefaultConverter=true;
+        cnv=u_getDefaultConverter(&errorCode);
+        if(U_FAILURE(errorCode)) {
+            return 0;
+        }
+    } else {
+        isDefaultConverter=false;
+        ucnv_resetFromUnicode(cnv);
+    }
+
+    // convert
+    int32_t len=doExtract(0, length(), dest, destCapacity, cnv, errorCode);
+
+    // release the converter
+    if(isDefaultConverter) {
+        u_releaseDefaultConverter(cnv);
+    }
+
+    return len;
+}
+
+int32_t
+UnicodeString::doExtract(int32_t start, int32_t length,
+                         char *dest, int32_t destCapacity,
+                         UConverter *cnv,
+                         UErrorCode &errorCode) const
+{
+    if(U_FAILURE(errorCode)) {
+        if(destCapacity!=0) {
+            *dest=0;
+        }
+        return 0;
+    }
+
+    const char16_t *src=getArrayStart()+start, *srcLimit=src+length;
+    char *originalDest=dest;
+    const char *destLimit;
+
+    if(destCapacity==0) {
+        destLimit=dest=0;
+    } else if(destCapacity==-1) {
+        // Pin the limit to U_MAX_PTR if the "magic" destCapacity is used.
+        destLimit=(char*)U_MAX_PTR(dest);
+        // for NUL-termination, translate into highest int32_t
+        destCapacity=0x7fffffff;
+    } else {
+        destLimit=dest+destCapacity;
+    }
+
+    // perform the conversion
+    ucnv_fromUnicode(cnv, &dest, destLimit, &src, srcLimit, 0, true, &errorCode);
+    length=(int32_t)(dest-originalDest);
+
+    // if an overflow occurs, then get the preflighting length
+    if(errorCode==U_BUFFER_OVERFLOW_ERROR) {
+        char buffer[1024];
+
+        destLimit=buffer+sizeof(buffer);
+        do {
+            dest=buffer;
+            errorCode=U_ZERO_ERROR;
+            ucnv_fromUnicode(cnv, &dest, destLimit, &src, srcLimit, 0, true, &errorCode);
+            length+=(int32_t)(dest-buffer);
+        } while(errorCode==U_BUFFER_OVERFLOW_ERROR);
+    }
+
+    return u_terminateChars(originalDest, destCapacity, length, &errorCode);
+}
+
+void
+UnicodeString::doCodepageCreate(const char *codepageData,
+                                int32_t dataLength,
+                                const char *codepage)
+{
+    // if there's nothing to convert, do nothing
+    if(codepageData == 0 || dataLength == 0 || dataLength < -1) {
+        return;
+    }
+    if(dataLength == -1) {
+        dataLength = (int32_t)uprv_strlen(codepageData);
+    }
+
+    UErrorCode status = U_ZERO_ERROR;
+
+    // create the converter
+    // if the codepage is the default, use our cache
+    // if it is an empty string, then use the "invariant character" conversion
+    UConverter *converter;
+    if (codepage == 0) {
+        const char *defaultName = ucnv_getDefaultName();
+        if(UCNV_FAST_IS_UTF8(defaultName)) {
+            setToUTF8(StringPiece(codepageData, dataLength));
+            return;
+        }
+        converter = u_getDefaultConverter(&status);
+    } else if(*codepage == 0) {
+        // use the "invariant characters" conversion
+        if(cloneArrayIfNeeded(dataLength, dataLength, false)) {
+            u_charsToUChars(codepageData, getArrayStart(), dataLength);
+            setLength(dataLength);
+        } else {
+            setToBogus();
+        }
+        return;
+    } else {
+        converter = ucnv_open(codepage, &status);
+    }
+
+    // if we failed, set the appropriate flags and return
+    if(U_FAILURE(status)) {
+        setToBogus();
+        return;
+    }
+
+    // perform the conversion
+    doCodepageCreate(codepageData, dataLength, converter, status);
+    if(U_FAILURE(status)) {
+        setToBogus();
+    }
+
+    // close the converter
+    if(codepage == 0) {
+        u_releaseDefaultConverter(converter);
+    } else {
+        ucnv_close(converter);
+    }
+}
+
+void
+UnicodeString::doCodepageCreate(const char *codepageData,
+                                int32_t dataLength,
+                                UConverter *converter,
+                                UErrorCode &status)
+{
+    if(U_FAILURE(status)) {
+        return;
+    }
+
+    // set up the conversion parameters
+    const char *mySource     = codepageData;
+    const char *mySourceEnd  = mySource + dataLength;
+    char16_t *array, *myTarget;
+
+    // estimate the size needed:
+    int32_t arraySize;
+    if(dataLength <= US_STACKBUF_SIZE) {
+        // try to use the stack buffer
+        arraySize = US_STACKBUF_SIZE;
+    } else {
+        // 1.25 char16_t's per source byte should cover most cases
+        arraySize = dataLength + (dataLength >> 2);
+    }
+
+    // we do not care about the current contents
+    UBool doCopyArray = false;
+    for(;;) {
+        if(!cloneArrayIfNeeded(arraySize, arraySize, doCopyArray)) {
+            setToBogus();
+            break;
+        }
+
+        // perform the conversion
+        array = getArrayStart();
+        myTarget = array + length();
+        ucnv_toUnicode(converter, &myTarget,  array + getCapacity(),
+            &mySource, mySourceEnd, 0, true, &status);
+
+        // update the conversion parameters
+        setLength((int32_t)(myTarget - array));
+
+        // allocate more space and copy data, if needed
+        if(status == U_BUFFER_OVERFLOW_ERROR) {
+            // reset the error code
+            status = U_ZERO_ERROR;
+
+            // keep the previous conversion results
+            doCopyArray = true;
+
+            // estimate the new size needed, larger than before
+            // try 2 char16_t's per remaining source byte
+            arraySize = (int32_t)(length() + 2 * (mySourceEnd - mySource));
+        } else {
+            break;
+        }
+    }
+}
+
+U_NAMESPACE_END
+
+#endif
diff --git a/intl/icu/source/common/unistr_props.cpp b/intl/icu/source/common/unistr_props.cpp
new file mode 100644
index 0000000000..3d05233bb9
--- /dev/null
+++ b/intl/icu/source/common/unistr_props.cpp
@@ -0,0 +1,77 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 1999-2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  unistr_props.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:2
+*
+*   created on: 2004aug25
+*   created by: Markus W. Scherer
+*
+*   Character property dependent functions moved here from unistr.cpp
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/uchar.h"
+#include "unicode/unistr.h"
+#include "unicode/utf16.h"
+
+U_NAMESPACE_BEGIN
+
+UnicodeString& 
+UnicodeString::trim()
+{
+  if(isBogus()) {
+    return *this;
+  }
+
+  char16_t *array = getArrayStart();
+  UChar32 c;
+  int32_t oldLength = this->length();
+  int32_t i = oldLength, length;
+
+  // first cut off trailing white space
+  for(;;) {
+    length = i;
+    if(i <= 0) {
+      break;
+    }
+    U16_PREV(array, 0, i, c);
+    if(!(c == 0x20 || u_isWhitespace(c))) {
+      break;
+    }
+  }
+  if(length < oldLength) {
+    setLength(length);
+  }
+
+  // find leading white space
+  int32_t start;
+  i = 0;
+  for(;;) {
+    start = i;
+    if(i >= length) {
+      break;
+    }
+    U16_NEXT(array, i, length, c);
+    if(!(c == 0x20 || u_isWhitespace(c))) {
+      break;
+    }
+  }
+
+  // move string forward over leading white space
+  if(start > 0) {
+    doReplace(0, start, 0, 0, 0);
+  }
+
+  return *this;
+}
+
+U_NAMESPACE_END
diff --git a/intl/icu/source/common/unistr_titlecase_brkiter.cpp b/intl/icu/source/common/unistr_titlecase_brkiter.cpp
new file mode 100644
index 0000000000..4969884b0d
--- /dev/null
+++ b/intl/icu/source/common/unistr_titlecase_brkiter.cpp
@@ -0,0 +1,57 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*   Copyright (C) 2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*******************************************************************************
+*   file name:  unistr_titlecase_brkiter.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:2
+*
+*   created on: 2011may30
+*   created by: Markus W. Scherer
+*
+*   Titlecasing functions that are based on BreakIterator
+*   were moved here to break dependency cycles among parts of the common library.
+*/
+
+#include "unicode/utypes.h"
+
+#if !UCONFIG_NO_BREAK_ITERATION
+
+#include "unicode/brkiter.h"
+#include "unicode/locid.h"
+#include "unicode/ucasemap.h"
+#include "unicode/unistr.h"
+#include "ucasemap_imp.h"
+
+U_NAMESPACE_BEGIN
+
+UnicodeString &
+UnicodeString::toTitle(BreakIterator *iter) {
+    return toTitle(iter, Locale::getDefault(), 0);
+}
+
+UnicodeString &
+UnicodeString::toTitle(BreakIterator *iter, const Locale &locale) {
+    return toTitle(iter, locale, 0);
+}
+
+UnicodeString &
+UnicodeString::toTitle(BreakIterator *iter, const Locale &locale, uint32_t options) {
+    LocalPointer ownedIter;
+    UErrorCode errorCode = U_ZERO_ERROR;
+    iter = ustrcase_getTitleBreakIterator(&locale, "", options, iter, ownedIter, errorCode);
+    if (iter == nullptr) {
+        setToBogus();
+        return *this;
+    }
+    caseMap(ustrcase_getCaseLocale(locale.getBaseName()), options, iter, ustrcase_internalToTitle);
+    return *this;
+}
+
+U_NAMESPACE_END
+
+#endif  // !UCONFIG_NO_BREAK_ITERATION
diff --git a/intl/icu/source/common/unistrappender.h b/intl/icu/source/common/unistrappender.h
new file mode 100644
index 0000000000..ac49b5f2f9
--- /dev/null
+++ b/intl/icu/source/common/unistrappender.h
@@ -0,0 +1,90 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+* Copyright (C) 2015, International Business Machines Corporation and
+* others. All Rights Reserved.
+******************************************************************************
+*
+* File unistrappender.h
+******************************************************************************
+*/
+
+#ifndef __UNISTRAPPENDER_H__
+#define __UNISTRAPPENDER_H__
+
+#include "unicode/unistr.h"
+#include "unicode/uobject.h"
+#include "unicode/utf16.h"
+#include "unicode/utypes.h"
+#include "cmemory.h"
+
+U_NAMESPACE_BEGIN
+
+/**
+ * An optimization for the slowness of calling UnicodeString::append()
+ * one character at a time in a loop. It stores appends in a buffer while
+ * never actually calling append on the unicode string unless the buffer
+ * fills up or is flushed.
+ * 
+ * proper usage:
+ * {
+ *     UnicodeStringAppender appender(astring);
+ *     for (int32_t i = 0; i < 100; ++i) {
+ *        appender.append((char16_t) i);
+ *     }
+ *     // appender flushed automatically when it goes out of scope.
+ * }
+ */
+class UnicodeStringAppender : public UMemory {
+public:
+    
+    /**
+     * dest is the UnicodeString being appended to. It must always
+     * exist while this instance exists.
+     */
+    UnicodeStringAppender(UnicodeString &dest) : fDest(&dest), fIdx(0) { }
+
+    inline void append(char16_t x) {
+        if (fIdx == UPRV_LENGTHOF(fBuffer)) {
+            fDest->append(fBuffer, 0, fIdx);
+            fIdx = 0;
+        }
+        fBuffer[fIdx++] = x;
+    }
+
+    inline void append(UChar32 x) {
+        if (fIdx >= UPRV_LENGTHOF(fBuffer) - 1) {
+            fDest->append(fBuffer, 0, fIdx);
+            fIdx = 0;
+        }
+        U16_APPEND_UNSAFE(fBuffer, fIdx, x);
+    }
+
+    /**
+     * Ensures that all appended characters have been written out to dest.
+     */
+    inline void flush() {
+        if (fIdx) {
+            fDest->append(fBuffer, 0, fIdx);
+        }
+        fIdx = 0;
+    }
+
+    /**
+     * flush the buffer when we go out of scope.
+     */
+    ~UnicodeStringAppender() {
+        flush();
+    }
+private:
+    UnicodeString *fDest;
+    int32_t fIdx;
+    char16_t fBuffer[32];
+    UnicodeStringAppender(const UnicodeStringAppender &other);
+    UnicodeStringAppender &operator=(const UnicodeStringAppender &other);
+};
+
+U_NAMESPACE_END
+
+#endif
diff --git a/intl/icu/source/common/unorm.cpp b/intl/icu/source/common/unorm.cpp
new file mode 100644
index 0000000000..f2ef7fa664
--- /dev/null
+++ b/intl/icu/source/common/unorm.cpp
@@ -0,0 +1,280 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+* Copyright (c) 1996-2014, International Business Machines
+* Corporation and others. All Rights Reserved.
+******************************************************************************
+* File unorm.cpp
+*
+* Created by: Vladimir Weinstein 12052000
+*
+* Modification history :
+*
+* Date        Name        Description
+* 02/01/01    synwee      Added normalization quickcheck enum and method.
+* 02/12/01    synwee      Commented out quickcheck util api has been approved
+*                         Added private method for doing FCD checks
+* 02/23/01    synwee      Modified quickcheck and checkFCE to run through 
+*                         string for codepoints < 0x300 for the normalization 
+*                         mode NFC.
+* 05/25/01+   Markus Scherer total rewrite, implement all normalization here
+*                         instead of just wrappers around normlzr.cpp,
+*                         load unorm.dat, support Unicode 3.1 with
+*                         supplementary code points, etc.
+* 2009-nov..2010-jan  Markus Scherer  total rewrite, new Normalizer2 API & code
+*/
+
+#include "unicode/utypes.h"
+
+#if !UCONFIG_NO_NORMALIZATION
+
+#include "unicode/udata.h"
+#include "unicode/ustring.h"
+#include "unicode/uiter.h"
+#include "unicode/unorm.h"
+#include "unicode/unorm2.h"
+#include "normalizer2impl.h"
+#include "unormimp.h"
+#include "uprops.h"
+#include "ustr_imp.h"
+
+U_NAMESPACE_USE
+
+/* quick check functions ---------------------------------------------------- */
+
+U_CAPI UNormalizationCheckResult U_EXPORT2
+unorm_quickCheck(const char16_t *src,
+                 int32_t srcLength, 
+                 UNormalizationMode mode,
+                 UErrorCode *pErrorCode) {
+    const Normalizer2 *n2=Normalizer2Factory::getInstance(mode, *pErrorCode);
+    return unorm2_quickCheck((const UNormalizer2 *)n2, src, srcLength, pErrorCode);
+}
+
+U_CAPI UNormalizationCheckResult U_EXPORT2
+unorm_quickCheckWithOptions(const char16_t *src, int32_t srcLength,
+                            UNormalizationMode mode, int32_t options,
+                            UErrorCode *pErrorCode) {
+    const Normalizer2 *n2=Normalizer2Factory::getInstance(mode, *pErrorCode);
+    if(options&UNORM_UNICODE_3_2) {
+        FilteredNormalizer2 fn2(*n2, *uniset_getUnicode32Instance(*pErrorCode));
+        return unorm2_quickCheck(
+            reinterpret_cast(static_cast(&fn2)),
+            src, srcLength, pErrorCode);
+    } else {
+        return unorm2_quickCheck((const UNormalizer2 *)n2, src, srcLength, pErrorCode);
+    }
+}
+
+U_CAPI UBool U_EXPORT2
+unorm_isNormalized(const char16_t *src, int32_t srcLength,
+                   UNormalizationMode mode,
+                   UErrorCode *pErrorCode) {
+    const Normalizer2 *n2=Normalizer2Factory::getInstance(mode, *pErrorCode);
+    return unorm2_isNormalized((const UNormalizer2 *)n2, src, srcLength, pErrorCode);
+}
+
+U_CAPI UBool U_EXPORT2
+unorm_isNormalizedWithOptions(const char16_t *src, int32_t srcLength,
+                              UNormalizationMode mode, int32_t options,
+                              UErrorCode *pErrorCode) {
+    const Normalizer2 *n2=Normalizer2Factory::getInstance(mode, *pErrorCode);
+    if(options&UNORM_UNICODE_3_2) {
+        FilteredNormalizer2 fn2(*n2, *uniset_getUnicode32Instance(*pErrorCode));
+        return unorm2_isNormalized(
+            reinterpret_cast(static_cast(&fn2)),
+            src, srcLength, pErrorCode);
+    } else {
+        return unorm2_isNormalized((const UNormalizer2 *)n2, src, srcLength, pErrorCode);
+    }
+}
+
+/* normalize() API ---------------------------------------------------------- */
+
+/** Public API for normalizing. */
+U_CAPI int32_t U_EXPORT2
+unorm_normalize(const char16_t *src, int32_t srcLength,
+                UNormalizationMode mode, int32_t options,
+                char16_t *dest, int32_t destCapacity,
+                UErrorCode *pErrorCode) {
+    const Normalizer2 *n2=Normalizer2Factory::getInstance(mode, *pErrorCode);
+    if(options&UNORM_UNICODE_3_2) {
+        FilteredNormalizer2 fn2(*n2, *uniset_getUnicode32Instance(*pErrorCode));
+        return unorm2_normalize(
+            reinterpret_cast(static_cast(&fn2)),
+            src, srcLength, dest, destCapacity, pErrorCode);
+    } else {
+        return unorm2_normalize((const UNormalizer2 *)n2,
+            src, srcLength, dest, destCapacity, pErrorCode);
+    }
+}
+
+
+/* iteration functions ------------------------------------------------------ */
+
+static int32_t
+_iterate(UCharIterator *src, UBool forward,
+              char16_t *dest, int32_t destCapacity,
+              const Normalizer2 *n2,
+              UBool doNormalize, UBool *pNeededToNormalize,
+              UErrorCode *pErrorCode) {
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+    if(destCapacity<0 || (dest==nullptr && destCapacity>0) || src==nullptr) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    if(pNeededToNormalize!=nullptr) {
+        *pNeededToNormalize=false;
+    }
+    if(!(forward ? src->hasNext(src) : src->hasPrevious(src))) {
+        return u_terminateUChars(dest, destCapacity, 0, pErrorCode);
+    }
+
+    UnicodeString buffer;
+    UChar32 c;
+    if(forward) {
+        /* get one character and ignore its properties */
+        buffer.append(uiter_next32(src));
+        /* get all following characters until we see a boundary */
+        while((c=uiter_next32(src))>=0) {
+            if(n2->hasBoundaryBefore(c)) {
+                /* back out the latest movement to stop at the boundary */
+                src->move(src, -U16_LENGTH(c), UITER_CURRENT);
+                break;
+            } else {
+                buffer.append(c);
+            }
+        }
+    } else {
+        while((c=uiter_previous32(src))>=0) {
+            /* always write this character to the front of the buffer */
+            buffer.insert(0, c);
+            /* stop if this just-copied character is a boundary */
+            if(n2->hasBoundaryBefore(c)) {
+                break;
+            }
+        }
+    }
+
+    UnicodeString destString(dest, 0, destCapacity);
+    if(buffer.length()>0 && doNormalize) {
+        n2->normalize(buffer, destString, *pErrorCode).extract(dest, destCapacity, *pErrorCode);
+        if(pNeededToNormalize!=nullptr && U_SUCCESS(*pErrorCode)) {
+            *pNeededToNormalize= destString!=buffer;
+        }
+        return destString.length();
+    } else {
+        /* just copy the source characters */
+        return buffer.extract(dest, destCapacity, *pErrorCode);
+    }
+}
+
+static int32_t
+unorm_iterate(UCharIterator *src, UBool forward,
+              char16_t *dest, int32_t destCapacity,
+              UNormalizationMode mode, int32_t options,
+              UBool doNormalize, UBool *pNeededToNormalize,
+              UErrorCode *pErrorCode) {
+    const Normalizer2 *n2=Normalizer2Factory::getInstance(mode, *pErrorCode);
+    if(options&UNORM_UNICODE_3_2) {
+        const UnicodeSet *uni32 = uniset_getUnicode32Instance(*pErrorCode);
+        if(U_FAILURE(*pErrorCode)) {
+            return 0;
+        }
+        FilteredNormalizer2 fn2(*n2, *uni32);
+        return _iterate(src, forward, dest, destCapacity,
+            &fn2, doNormalize, pNeededToNormalize, pErrorCode);
+    }
+    return _iterate(src, forward, dest, destCapacity,
+            n2, doNormalize, pNeededToNormalize, pErrorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+unorm_previous(UCharIterator *src,
+               char16_t *dest, int32_t destCapacity,
+               UNormalizationMode mode, int32_t options,
+               UBool doNormalize, UBool *pNeededToNormalize,
+               UErrorCode *pErrorCode) {
+    return unorm_iterate(src, false,
+                         dest, destCapacity,
+                         mode, options,
+                         doNormalize, pNeededToNormalize,
+                         pErrorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+unorm_next(UCharIterator *src,
+           char16_t *dest, int32_t destCapacity,
+           UNormalizationMode mode, int32_t options,
+           UBool doNormalize, UBool *pNeededToNormalize,
+           UErrorCode *pErrorCode) {
+    return unorm_iterate(src, true,
+                         dest, destCapacity,
+                         mode, options,
+                         doNormalize, pNeededToNormalize,
+                         pErrorCode);
+}
+
+/* Concatenation of normalized strings -------------------------------------- */
+
+static int32_t
+_concatenate(const char16_t *left, int32_t leftLength,
+                  const char16_t *right, int32_t rightLength,
+                  char16_t *dest, int32_t destCapacity,
+                  const Normalizer2 *n2,
+                  UErrorCode *pErrorCode) {
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+    if(destCapacity<0 || (dest==nullptr && destCapacity>0) ||
+        left==nullptr || leftLength<-1 || right==nullptr || rightLength<-1) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    /* check for overlapping right and destination */
+    if( dest!=nullptr &&
+        ((right>=dest && right<(dest+destCapacity)) ||
+         (rightLength>0 && dest>=right && dest<(right+rightLength)))
+    ) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    /* allow left==dest */
+    UnicodeString destString;
+    if(left==dest) {
+        destString.setTo(dest, leftLength, destCapacity);
+    } else {
+        destString.setTo(dest, 0, destCapacity);
+        destString.append(left, leftLength);
+    }
+    return n2->append(destString, UnicodeString(rightLength<0, right, rightLength), *pErrorCode).
+           extract(dest, destCapacity, *pErrorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+unorm_concatenate(const char16_t *left, int32_t leftLength,
+                  const char16_t *right, int32_t rightLength,
+                  char16_t *dest, int32_t destCapacity,
+                  UNormalizationMode mode, int32_t options,
+                  UErrorCode *pErrorCode) {
+    const Normalizer2 *n2=Normalizer2Factory::getInstance(mode, *pErrorCode);
+    if(options&UNORM_UNICODE_3_2) {
+        const UnicodeSet *uni32 = uniset_getUnicode32Instance(*pErrorCode);
+        if(U_FAILURE(*pErrorCode)) {
+            return 0;
+        }
+        FilteredNormalizer2 fn2(*n2, *uni32);
+        return _concatenate(left, leftLength, right, rightLength,
+            dest, destCapacity, &fn2, pErrorCode);
+    }
+    return _concatenate(left, leftLength, right, rightLength,
+        dest, destCapacity, n2, pErrorCode);
+}
+
+#endif /* #if !UCONFIG_NO_NORMALIZATION */
diff --git a/intl/icu/source/common/unormcmp.cpp b/intl/icu/source/common/unormcmp.cpp
new file mode 100644
index 0000000000..e11e716c8d
--- /dev/null
+++ b/intl/icu/source/common/unormcmp.cpp
@@ -0,0 +1,640 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2001-2014, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  unormcmp.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2004sep13
+*   created by: Markus W. Scherer
+*
+*   unorm_compare() function moved here from unorm.cpp for better modularization.
+*   Depends on both normalization and case folding.
+*   Allows unorm.cpp to not depend on any character properties code.
+*/
+
+#include "unicode/utypes.h"
+
+#if !UCONFIG_NO_NORMALIZATION
+
+#include "unicode/unorm.h"
+#include "unicode/ustring.h"
+#include "cmemory.h"
+#include "normalizer2impl.h"
+#include "ucase.h"
+#include "uprops.h"
+#include "ustr_imp.h"
+
+U_NAMESPACE_USE
+
+/* compare canonically equivalent ------------------------------------------- */
+
+/*
+ * Compare two strings for canonical equivalence.
+ * Further options include case-insensitive comparison and
+ * code point order (as opposed to code unit order).
+ *
+ * In this function, canonical equivalence is optional as well.
+ * If canonical equivalence is tested, then both strings must fulfill
+ * the FCD check.
+ *
+ * Semantically, this is equivalent to
+ *   strcmp[CodePointOrder](NFD(foldCase(s1)), NFD(foldCase(s2)))
+ * where code point order, NFD and foldCase are all optional.
+ *
+ * String comparisons almost always yield results before processing both strings
+ * completely.
+ * They are generally more efficient working incrementally instead of
+ * performing the sub-processing (strlen, normalization, case-folding)
+ * on the entire strings first.
+ *
+ * It is also unnecessary to not normalize identical characters.
+ *
+ * This function works in principle as follows:
+ *
+ * loop {
+ *   get one code unit c1 from s1 (-1 if end of source)
+ *   get one code unit c2 from s2 (-1 if end of source)
+ *
+ *   if(either string finished) {
+ *     return result;
+ *   }
+ *   if(c1==c2) {
+ *     continue;
+ *   }
+ *
+ *   // c1!=c2
+ *   try to decompose/case-fold c1/c2, and continue if one does;
+ *
+ *   // still c1!=c2 and neither decomposes/case-folds, return result
+ *   return c1-c2;
+ * }
+ *
+ * When a character decomposes, then the pointer for that source changes to
+ * the decomposition, pushing the previous pointer onto a stack.
+ * When the end of the decomposition is reached, then the code unit reader
+ * pops the previous source from the stack.
+ * (Same for case-folding.)
+ *
+ * This is complicated further by operating on variable-width UTF-16.
+ * The top part of the loop works on code units, while lookups for decomposition
+ * and case-folding need code points.
+ * Code points are assembled after the equality/end-of-source part.
+ * The source pointer is only advanced beyond all code units when the code point
+ * actually decomposes/case-folds.
+ *
+ * If we were on a trail surrogate unit when assembling a code point,
+ * and the code point decomposes/case-folds, then the decomposition/folding
+ * result must be compared with the part of the other string that corresponds to
+ * this string's lead surrogate.
+ * Since we only assemble a code point when hitting a trail unit when the
+ * preceding lead units were identical, we back up the other string by one unit
+ * in such a case.
+ *
+ * The optional code point order comparison at the end works with
+ * the same fix-up as the other code point order comparison functions.
+ * See ustring.c and the comment near the end of this function.
+ *
+ * Assumption: A decomposition or case-folding result string never contains
+ * a single surrogate. This is a safe assumption in the Unicode Standard.
+ * Therefore, we do not need to check for surrogate pairs across
+ * decomposition/case-folding boundaries.
+ *
+ * Further assumptions (see verifications tstnorm.cpp):
+ * The API function checks for FCD first, while the core function
+ * first case-folds and then decomposes. This requires that case-folding does not
+ * un-FCD any strings.
+ *
+ * The API function may also NFD the input and turn off decomposition.
+ * This requires that case-folding does not un-NFD strings either.
+ *
+ * TODO If any of the above two assumptions is violated,
+ * then this entire code must be re-thought.
+ * If this happens, then a simple solution is to case-fold both strings up front
+ * and to turn off UNORM_INPUT_IS_FCD.
+ * We already do this when not both strings are in FCD because makeFCD
+ * would be a partial NFD before the case folding, which does not work.
+ * Note that all of this is only a problem when case-folding _and_
+ * canonical equivalence come together.
+ * (Comments in unorm_compare() are more up to date than this TODO.)
+ */
+
+/* stack element for previous-level source/decomposition pointers */
+struct CmpEquivLevel {
+    const char16_t *start, *s, *limit;
+};
+typedef struct CmpEquivLevel CmpEquivLevel;
+
+/**
+ * Internal option for unorm_cmpEquivFold() for decomposing.
+ * If not set, just do strcasecmp().
+ */
+#define _COMPARE_EQUIV 0x80000
+
+/* internal function */
+static int32_t
+unorm_cmpEquivFold(const char16_t *s1, int32_t length1,
+                   const char16_t *s2, int32_t length2,
+                   uint32_t options,
+                   UErrorCode *pErrorCode) {
+    const Normalizer2Impl *nfcImpl;
+
+    /* current-level start/limit - s1/s2 as current */
+    const char16_t *start1, *start2, *limit1, *limit2;
+
+    /* decomposition and case folding variables */
+    const char16_t *p;
+    int32_t length;
+
+    /* stacks of previous-level start/current/limit */
+    CmpEquivLevel stack1[2], stack2[2];
+
+    /* buffers for algorithmic decompositions */
+    char16_t decomp1[4], decomp2[4];
+
+    /* case folding buffers, only use current-level start/limit */
+    char16_t fold1[UCASE_MAX_STRING_LENGTH+1], fold2[UCASE_MAX_STRING_LENGTH+1];
+
+    /* track which is the current level per string */
+    int32_t level1, level2;
+
+    /* current code units, and code points for lookups */
+    UChar32 c1, c2, cp1, cp2;
+
+    /* no argument error checking because this itself is not an API */
+
+    /*
+     * assume that at least one of the options _COMPARE_EQUIV and U_COMPARE_IGNORE_CASE is set
+     * otherwise this function must behave exactly as uprv_strCompare()
+     * not checking for that here makes testing this function easier
+     */
+
+    /* normalization/properties data loaded? */
+    if((options&_COMPARE_EQUIV)!=0) {
+        nfcImpl=Normalizer2Factory::getNFCImpl(*pErrorCode);
+    } else {
+        nfcImpl=nullptr;
+    }
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+
+    /* initialize */
+    start1=s1;
+    if(length1==-1) {
+        limit1=nullptr;
+    } else {
+        limit1=s1+length1;
+    }
+
+    start2=s2;
+    if(length2==-1) {
+        limit2=nullptr;
+    } else {
+        limit2=s2+length2;
+    }
+
+    level1=level2=0;
+    c1=c2=-1;
+
+    /* comparison loop */
+    for(;;) {
+        /*
+         * here a code unit value of -1 means "get another code unit"
+         * below it will mean "this source is finished"
+         */
+
+        if(c1<0) {
+            /* get next code unit from string 1, post-increment */
+            for(;;) {
+                if(s1==limit1 || ((c1=*s1)==0 && (limit1==nullptr || (options&_STRNCMP_STYLE)))) {
+                    if(level1==0) {
+                        c1=-1;
+                        break;
+                    }
+                } else {
+                    ++s1;
+                    break;
+                }
+
+                /* reached end of level buffer, pop one level */
+                do {
+                    --level1;
+                    start1=stack1[level1].start;    /*Not uninitialized*/
+                } while(start1==nullptr);
+                s1=stack1[level1].s;                /*Not uninitialized*/
+                limit1=stack1[level1].limit;        /*Not uninitialized*/
+            }
+        }
+
+        if(c2<0) {
+            /* get next code unit from string 2, post-increment */
+            for(;;) {
+                if(s2==limit2 || ((c2=*s2)==0 && (limit2==nullptr || (options&_STRNCMP_STYLE)))) {
+                    if(level2==0) {
+                        c2=-1;
+                        break;
+                    }
+                } else {
+                    ++s2;
+                    break;
+                }
+
+                /* reached end of level buffer, pop one level */
+                do {
+                    --level2;
+                    start2=stack2[level2].start;    /*Not uninitialized*/
+                } while(start2==nullptr);
+                s2=stack2[level2].s;                /*Not uninitialized*/
+                limit2=stack2[level2].limit;        /*Not uninitialized*/
+            }
+        }
+
+        /*
+         * compare c1 and c2
+         * either variable c1, c2 is -1 only if the corresponding string is finished
+         */
+        if(c1==c2) {
+            if(c1<0) {
+                return 0;   /* c1==c2==-1 indicating end of strings */
+            }
+            c1=c2=-1;       /* make us fetch new code units */
+            continue;
+        } else if(c1<0) {
+            return -1;      /* string 1 ends before string 2 */
+        } else if(c2<0) {
+            return 1;       /* string 2 ends before string 1 */
+        }
+        /* c1!=c2 && c1>=0 && c2>=0 */
+
+        /* get complete code points for c1, c2 for lookups if either is a surrogate */
+        cp1=c1;
+        if(U_IS_SURROGATE(c1)) {
+            char16_t c;
+
+            if(U_IS_SURROGATE_LEAD(c1)) {
+                if(s1!=limit1 && U16_IS_TRAIL(c=*s1)) {
+                    /* advance ++s1; only below if cp1 decomposes/case-folds */
+                    cp1=U16_GET_SUPPLEMENTARY(c1, c);
+                }
+            } else /* isTrail(c1) */ {
+                if(start1<=(s1-2) && U16_IS_LEAD(c=*(s1-2))) {
+                    cp1=U16_GET_SUPPLEMENTARY(c, c1);
+                }
+            }
+        }
+
+        cp2=c2;
+        if(U_IS_SURROGATE(c2)) {
+            char16_t c;
+
+            if(U_IS_SURROGATE_LEAD(c2)) {
+                if(s2!=limit2 && U16_IS_TRAIL(c=*s2)) {
+                    /* advance ++s2; only below if cp2 decomposes/case-folds */
+                    cp2=U16_GET_SUPPLEMENTARY(c2, c);
+                }
+            } else /* isTrail(c2) */ {
+                if(start2<=(s2-2) && U16_IS_LEAD(c=*(s2-2))) {
+                    cp2=U16_GET_SUPPLEMENTARY(c, c2);
+                }
+            }
+        }
+
+        /*
+         * go down one level for each string
+         * continue with the main loop as soon as there is a real change
+         */
+
+        if( level1==0 && (options&U_COMPARE_IGNORE_CASE) &&
+            (length=ucase_toFullFolding((UChar32)cp1, &p, options))>=0
+        ) {
+            /* cp1 case-folds to the code point "length" or to p[length] */
+            if(U_IS_SURROGATE(c1)) {
+                if(U_IS_SURROGATE_LEAD(c1)) {
+                    /* advance beyond source surrogate pair if it case-folds */
+                    ++s1;
+                } else /* isTrail(c1) */ {
+                    /*
+                     * we got a supplementary code point when hitting its trail surrogate,
+                     * therefore the lead surrogate must have been the same as in the other string;
+                     * compare this decomposition with the lead surrogate in the other string
+                     * remember that this simulates bulk text replacement:
+                     * the decomposition would replace the entire code point
+                     */
+                    --s2;
+                    c2=*(s2-1);
+                }
+            }
+
+            /* push current level pointers */
+            stack1[0].start=start1;
+            stack1[0].s=s1;
+            stack1[0].limit=limit1;
+            ++level1;
+
+            /* copy the folding result to fold1[] */
+            if(length<=UCASE_MAX_STRING_LENGTH) {
+                u_memcpy(fold1, p, length);
+            } else {
+                int32_t i=0;
+                U16_APPEND_UNSAFE(fold1, i, length);
+                length=i;
+            }
+
+            /* set next level pointers to case folding */
+            start1=s1=fold1;
+            limit1=fold1+length;
+
+            /* get ready to read from decomposition, continue with loop */
+            c1=-1;
+            continue;
+        }
+
+        if( level2==0 && (options&U_COMPARE_IGNORE_CASE) &&
+            (length=ucase_toFullFolding((UChar32)cp2, &p, options))>=0
+        ) {
+            /* cp2 case-folds to the code point "length" or to p[length] */
+            if(U_IS_SURROGATE(c2)) {
+                if(U_IS_SURROGATE_LEAD(c2)) {
+                    /* advance beyond source surrogate pair if it case-folds */
+                    ++s2;
+                } else /* isTrail(c2) */ {
+                    /*
+                     * we got a supplementary code point when hitting its trail surrogate,
+                     * therefore the lead surrogate must have been the same as in the other string;
+                     * compare this decomposition with the lead surrogate in the other string
+                     * remember that this simulates bulk text replacement:
+                     * the decomposition would replace the entire code point
+                     */
+                    --s1;
+                    c1=*(s1-1);
+                }
+            }
+
+            /* push current level pointers */
+            stack2[0].start=start2;
+            stack2[0].s=s2;
+            stack2[0].limit=limit2;
+            ++level2;
+
+            /* copy the folding result to fold2[] */
+            if(length<=UCASE_MAX_STRING_LENGTH) {
+                u_memcpy(fold2, p, length);
+            } else {
+                int32_t i=0;
+                U16_APPEND_UNSAFE(fold2, i, length);
+                length=i;
+            }
+
+            /* set next level pointers to case folding */
+            start2=s2=fold2;
+            limit2=fold2+length;
+
+            /* get ready to read from decomposition, continue with loop */
+            c2=-1;
+            continue;
+        }
+
+        if( level1<2 && (options&_COMPARE_EQUIV) &&
+            0!=(p=nfcImpl->getDecomposition((UChar32)cp1, decomp1, length))
+        ) {
+            /* cp1 decomposes into p[length] */
+            if(U_IS_SURROGATE(c1)) {
+                if(U_IS_SURROGATE_LEAD(c1)) {
+                    /* advance beyond source surrogate pair if it decomposes */
+                    ++s1;
+                } else /* isTrail(c1) */ {
+                    /*
+                     * we got a supplementary code point when hitting its trail surrogate,
+                     * therefore the lead surrogate must have been the same as in the other string;
+                     * compare this decomposition with the lead surrogate in the other string
+                     * remember that this simulates bulk text replacement:
+                     * the decomposition would replace the entire code point
+                     */
+                    --s2;
+                    c2=*(s2-1);
+                }
+            }
+
+            /* push current level pointers */
+            stack1[level1].start=start1;
+            stack1[level1].s=s1;
+            stack1[level1].limit=limit1;
+            ++level1;
+
+            /* set empty intermediate level if skipped */
+            if(level1<2) {
+                stack1[level1++].start=nullptr;
+            }
+
+            /* set next level pointers to decomposition */
+            start1=s1=p;
+            limit1=p+length;
+
+            /* get ready to read from decomposition, continue with loop */
+            c1=-1;
+            continue;
+        }
+
+        if( level2<2 && (options&_COMPARE_EQUIV) &&
+            0!=(p=nfcImpl->getDecomposition((UChar32)cp2, decomp2, length))
+        ) {
+            /* cp2 decomposes into p[length] */
+            if(U_IS_SURROGATE(c2)) {
+                if(U_IS_SURROGATE_LEAD(c2)) {
+                    /* advance beyond source surrogate pair if it decomposes */
+                    ++s2;
+                } else /* isTrail(c2) */ {
+                    /*
+                     * we got a supplementary code point when hitting its trail surrogate,
+                     * therefore the lead surrogate must have been the same as in the other string;
+                     * compare this decomposition with the lead surrogate in the other string
+                     * remember that this simulates bulk text replacement:
+                     * the decomposition would replace the entire code point
+                     */
+                    --s1;
+                    c1=*(s1-1);
+                }
+            }
+
+            /* push current level pointers */
+            stack2[level2].start=start2;
+            stack2[level2].s=s2;
+            stack2[level2].limit=limit2;
+            ++level2;
+
+            /* set empty intermediate level if skipped */
+            if(level2<2) {
+                stack2[level2++].start=nullptr;
+            }
+
+            /* set next level pointers to decomposition */
+            start2=s2=p;
+            limit2=p+length;
+
+            /* get ready to read from decomposition, continue with loop */
+            c2=-1;
+            continue;
+        }
+
+        /*
+         * no decomposition/case folding, max level for both sides:
+         * return difference result
+         *
+         * code point order comparison must not just return cp1-cp2
+         * because when single surrogates are present then the surrogate pairs
+         * that formed cp1 and cp2 may be from different string indexes
+         *
+         * example: { d800 d800 dc01 } vs. { d800 dc00 }, compare at second code units
+         * c1=d800 cp1=10001 c2=dc00 cp2=10000
+         * cp1-cp2>0 but c1-c2<0 and in fact in UTF-32 it is { d800 10001 } < { 10000 }
+         *
+         * therefore, use same fix-up as in ustring.c/uprv_strCompare()
+         * except: uprv_strCompare() fetches c=*s while this functions fetches c=*s++
+         * so we have slightly different pointer/start/limit comparisons here
+         */
+
+        if(c1>=0xd800 && c2>=0xd800 && (options&U_COMPARE_CODE_POINT_ORDER)) {
+            /* subtract 0x2800 from BMP code points to make them smaller than supplementary ones */
+            if(
+                (c1<=0xdbff && s1!=limit1 && U16_IS_TRAIL(*s1)) ||
+                (U16_IS_TRAIL(c1) && start1!=(s1-1) && U16_IS_LEAD(*(s1-2)))
+            ) {
+                /* part of a surrogate pair, leave >=d800 */
+            } else {
+                /* BMP code point - may be surrogate code point - make =d800 */
+            } else {
+                /* BMP code point - may be surrogate code point - make spanQuickCheckYes(str, *pErrorCode);
+    if (U_FAILURE(*pErrorCode)) {
+        return false;
+    }
+    /*
+     * ICU 2.4 had a further optimization:
+     * If both strings were not in FCD, then they were both NFD'ed,
+     * and the _COMPARE_EQUIV option was turned off.
+     * It is not entirely clear that this is valid with the current
+     * definition of the canonical caseless match.
+     * Therefore, ICU 2.6 removes that optimization.
+     */
+    if(spanQCYesnormalizeSecondAndAppend(normalized, unnormalized, *pErrorCode);
+        if (U_SUCCESS(*pErrorCode)) {
+            return true;
+        }
+    }
+    return false;
+}
+
+U_CAPI int32_t U_EXPORT2
+unorm_compare(const char16_t *s1, int32_t length1,
+              const char16_t *s2, int32_t length2,
+              uint32_t options,
+              UErrorCode *pErrorCode) {
+    /* argument checking */
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+    if(s1==0 || length1<-1 || s2==0 || length2<-1) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    UnicodeString fcd1, fcd2;
+    int32_t normOptions=(int32_t)(options>>UNORM_COMPARE_NORM_OPTIONS_SHIFT);
+    options|=_COMPARE_EQUIV;
+
+    /*
+     * UAX #21 Case Mappings, as fixed for Unicode version 4
+     * (see Jitterbug 2021), defines a canonical caseless match as
+     *
+     * A string X is a canonical caseless match
+     * for a string Y if and only if
+     * NFD(toCasefold(NFD(X))) = NFD(toCasefold(NFD(Y)))
+     *
+     * For better performance, we check for FCD (or let the caller tell us that
+     * both strings are in FCD) for the inner normalization.
+     * BasicNormalizerTest::FindFoldFCDExceptions() makes sure that
+     * case-folding preserves the FCD-ness of a string.
+     * The outer normalization is then only performed by unorm_cmpEquivFold()
+     * when there is a difference.
+     *
+     * Exception: When using the Turkic case-folding option, we do perform
+     * full NFD first. This is because in the Turkic case precomposed characters
+     * with 0049 capital I or 0069 small i fold differently whether they
+     * are first decomposed or not, so an FCD check - a check only for
+     * canonical order - is not sufficient.
+     */
+    if(!(options&UNORM_INPUT_IS_FCD) || (options&U_FOLD_CASE_EXCLUDE_SPECIAL_I)) {
+        const Normalizer2 *n2;
+        if(options&U_FOLD_CASE_EXCLUDE_SPECIAL_I) {
+            n2=Normalizer2::getNFDInstance(*pErrorCode);
+        } else {
+            n2=Normalizer2Factory::getFCDInstance(*pErrorCode);
+        }
+        if (U_FAILURE(*pErrorCode)) {
+            return 0;
+        }
+
+        if(normOptions&UNORM_UNICODE_3_2) {
+            const UnicodeSet *uni32=uniset_getUnicode32Instance(*pErrorCode);
+            FilteredNormalizer2 fn2(*n2, *uni32);
+            if(_normalize(&fn2, s1, length1, fcd1, pErrorCode)) {
+                s1=fcd1.getBuffer();
+                length1=fcd1.length();
+            }
+            if(_normalize(&fn2, s2, length2, fcd2, pErrorCode)) {
+                s2=fcd2.getBuffer();
+                length2=fcd2.length();
+            }
+        } else {
+            if(_normalize(n2, s1, length1, fcd1, pErrorCode)) {
+                s1=fcd1.getBuffer();
+                length1=fcd1.length();
+            }
+            if(_normalize(n2, s2, length2, fcd2, pErrorCode)) {
+                s2=fcd2.getBuffer();
+                length2=fcd2.length();
+            }
+        }
+    }
+
+    if(U_SUCCESS(*pErrorCode)) {
+        return unorm_cmpEquivFold(s1, length1, s2, length2, options, pErrorCode);
+    } else {
+        return 0;
+    }
+}
+
+#endif /* #if !UCONFIG_NO_NORMALIZATION */
diff --git a/intl/icu/source/common/unormimp.h b/intl/icu/source/common/unormimp.h
new file mode 100644
index 0000000000..f107eac67f
--- /dev/null
+++ b/intl/icu/source/common/unormimp.h
@@ -0,0 +1,488 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2001-2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  unormimp.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2001may25
+*   created by: Markus W. Scherer
+*/
+
+#ifndef __UNORMIMP_H__
+#define __UNORMIMP_H__
+
+#include "unicode/utypes.h"
+
+#if !UCONFIG_NO_NORMALIZATION
+
+#include "udataswp.h"
+
+/*
+ * The 2001-2010 implementation of the normalization code loads its data from
+ * unorm.icu, which is generated with the gennorm tool.
+ * The format of that file is described at the end of this file.
+ */
+
+/* norm32 value constants */
+enum {
+    /* quick check flags 0..3 set mean "no" for their forms */
+    _NORM_QC_NFC=0x11,          /* no|maybe */
+    _NORM_QC_NFKC=0x22,         /* no|maybe */
+    _NORM_QC_NFD=4,             /* no */
+    _NORM_QC_NFKD=8,            /* no */
+
+    _NORM_QC_ANY_NO=0xf,
+
+    /* quick check flags 4..5 mean "maybe" for their forms; test flags>=_NORM_QC_MAYBE */
+    _NORM_QC_MAYBE=0x10,
+    _NORM_QC_ANY_MAYBE=0x30,
+
+    _NORM_QC_MASK=0x3f,
+
+    _NORM_COMBINES_FWD=0x40,
+    _NORM_COMBINES_BACK=0x80,
+    _NORM_COMBINES_ANY=0xc0,
+
+    _NORM_CC_SHIFT=8,           /* UnicodeData.txt combining class in bits 15..8 */
+    _NORM_CC_MASK=0xff00,
+
+    _NORM_EXTRA_SHIFT=16,               /* 16 bits for the index to UChars and other extra data */
+    _NORM_EXTRA_INDEX_TOP=0xfc00,       /* start of surrogate specials after shift */
+
+    _NORM_EXTRA_SURROGATE_MASK=0x3ff,
+    _NORM_EXTRA_SURROGATE_TOP=0x3f0,    /* hangul etc. */
+
+    _NORM_EXTRA_HANGUL=_NORM_EXTRA_SURROGATE_TOP,
+    _NORM_EXTRA_JAMO_L,
+    _NORM_EXTRA_JAMO_V,
+    _NORM_EXTRA_JAMO_T
+};
+
+/* norm32 value constants using >16 bits */
+#define _NORM_MIN_SPECIAL       0xfc000000
+#define _NORM_SURROGATES_TOP    0xfff00000
+#define _NORM_MIN_HANGUL        0xfff00000
+#define _NORM_MIN_JAMO_V        0xfff20000
+#define _NORM_JAMO_V_TOP        0xfff30000
+
+/* value constants for auxTrie */
+enum {
+    _NORM_AUX_COMP_EX_SHIFT=10,
+    _NORM_AUX_UNSAFE_SHIFT=11,
+    _NORM_AUX_NFC_SKIPPABLE_F_SHIFT=12
+};
+
+#define _NORM_AUX_MAX_FNC           ((int32_t)1<<_NORM_AUX_COMP_EX_SHIFT)
+
+#define _NORM_AUX_FNC_MASK          (uint32_t)(_NORM_AUX_MAX_FNC-1)
+#define _NORM_AUX_COMP_EX_MASK      ((uint32_t)1<<_NORM_AUX_COMP_EX_SHIFT)
+#define _NORM_AUX_UNSAFE_MASK       ((uint32_t)1<<_NORM_AUX_UNSAFE_SHIFT)
+#define _NORM_AUX_NFC_SKIP_F_MASK   ((uint32_t)1<<_NORM_AUX_NFC_SKIPPABLE_F_SHIFT)
+
+/* canonStartSets[0..31] contains indexes for what is in the array */
+enum {
+    _NORM_SET_INDEX_CANON_SETS_LENGTH,      /* number of uint16_t in canonical starter sets */
+    _NORM_SET_INDEX_CANON_BMP_TABLE_LENGTH, /* number of uint16_t in the BMP search table (contains pairs) */
+    _NORM_SET_INDEX_CANON_SUPP_TABLE_LENGTH,/* number of uint16_t in the supplementary search table (contains triplets) */
+
+    /* from formatVersion 2.3: */
+    _NORM_SET_INDEX_NX_CJK_COMPAT_OFFSET,   /* uint16_t offset from canonStartSets[0] to the
+                                               exclusion set for CJK compatibility characters */
+    _NORM_SET_INDEX_NX_UNICODE32_OFFSET,    /* uint16_t offset from canonStartSets[0] to the
+                                               exclusion set for Unicode 3.2 characters */
+    _NORM_SET_INDEX_NX_RESERVED_OFFSET,     /* uint16_t offset from canonStartSets[0] to the
+                                               end of the previous exclusion set */
+
+    _NORM_SET_INDEX_TOP=32                  /* changing this requires a new formatVersion */
+};
+
+/* more constants for canonical starter sets */
+
+/* 14 bit indexes to canonical USerializedSets */
+#define _NORM_MAX_CANON_SETS            0x4000
+
+/* single-code point BMP sets are encoded directly in the search table except if result=0x4000..0x7fff */
+#define _NORM_CANON_SET_BMP_MASK        0xc000
+#define _NORM_CANON_SET_BMP_IS_INDEX    0x4000
+
+/* indexes[] value names */
+enum {
+    _NORM_INDEX_TRIE_SIZE,              /* number of bytes in normalization trie */
+    _NORM_INDEX_UCHAR_COUNT,            /* number of UChars in extra data */
+
+    _NORM_INDEX_COMBINE_DATA_COUNT,     /* number of uint16_t words for combining data */
+    _NORM_INDEX_COMBINE_FWD_COUNT,      /* number of code points that combine forward */
+    _NORM_INDEX_COMBINE_BOTH_COUNT,     /* number of code points that combine forward and backward */
+    _NORM_INDEX_COMBINE_BACK_COUNT,     /* number of code points that combine backward */
+
+    _NORM_INDEX_MIN_NFC_NO_MAYBE,       /* first code point with quick check NFC NO/MAYBE */
+    _NORM_INDEX_MIN_NFKC_NO_MAYBE,      /* first code point with quick check NFKC NO/MAYBE */
+    _NORM_INDEX_MIN_NFD_NO_MAYBE,       /* first code point with quick check NFD NO/MAYBE */
+    _NORM_INDEX_MIN_NFKD_NO_MAYBE,      /* first code point with quick check NFKD NO/MAYBE */
+
+    _NORM_INDEX_FCD_TRIE_SIZE,          /* number of bytes in FCD trie */
+
+    _NORM_INDEX_AUX_TRIE_SIZE,          /* number of bytes in the auxiliary trie */
+    _NORM_INDEX_CANON_SET_COUNT,        /* number of uint16_t in the array of serialized USet */
+
+    _NORM_INDEX_TOP=32                  /* changing this requires a new formatVersion */
+};
+
+enum {
+    /* FCD check: everything below this code point is known to have a 0 lead combining class */
+    _NORM_MIN_WITH_LEAD_CC=0x300
+};
+
+enum {
+    /**
+     * Bit 7 of the length byte for a decomposition string in extra data is
+     * a flag indicating whether the decomposition string is
+     * preceded by a 16-bit word with the leading and trailing cc
+     * of the decomposition (like for A-umlaut);
+     * if not, then both cc's are zero (like for compatibility ideographs).
+     */
+    _NORM_DECOMP_FLAG_LENGTH_HAS_CC=0x80,
+    /**
+     * Bits 6..0 of the length byte contain the actual length.
+     */
+    _NORM_DECOMP_LENGTH_MASK=0x7f
+};
+
+/** Constants for options flags for normalization. */
+enum {
+    /** Options bit 0, do not decompose Hangul syllables. */
+    UNORM_NX_HANGUL=1,
+    /** Options bit 1, do not decompose CJK compatibility characters. */
+    UNORM_NX_CJK_COMPAT=2
+};
+
+/**
+ * Description of the format of unorm.icu version 2.3.
+ *
+ * Main change from version 1 to version 2:
+ * Use of new, common UTrie instead of normalization-specific tries.
+ * Change to version 2.1: add third/auxiliary trie with associated data.
+ * Change to version 2.2: add skippable (f) flag data (_NORM_AUX_NFC_SKIP_F_MASK).
+ * Change to version 2.3: add serialized sets for normalization exclusions
+ *                        stored inside canonStartSets[]
+ *
+ * For more details of how to use the data structures see the code
+ * in unorm.cpp (runtime normalization code) and
+ * in gennorm.c and gennorm/store.c (build-time data generation).
+ *
+ * For the serialized format of UTrie see utrie.c/UTrieHeader.
+ *
+ * - Overall partition
+ *
+ * unorm.dat customarily begins with a UDataInfo structure, see udata.h and .c.
+ * After that there are the following structures:
+ *
+ * int32_t indexes[_NORM_INDEX_TOP];            -- _NORM_INDEX_TOP=32, see enum in this file
+ *
+ * UTrie normTrie;                              -- size in bytes=indexes[_NORM_INDEX_TRIE_SIZE]
+ * 
+ * uint16_t extraData[extraDataTop];            -- extraDataTop=indexes[_NORM_INDEX_UCHAR_COUNT]
+ *                                                 extraData[0] contains the number of units for
+ *                                                 FC_NFKC_Closure (formatVersion>=2.1)
+ *
+ * uint16_t combiningTable[combiningTableTop];  -- combiningTableTop=indexes[_NORM_INDEX_COMBINE_DATA_COUNT]
+ *                                                 combiningTableTop may include one 16-bit padding unit
+ *                                                 to make sure that fcdTrie is 32-bit-aligned
+ *
+ * UTrie fcdTrie;                               -- size in bytes=indexes[_NORM_INDEX_FCD_TRIE_SIZE]
+ *
+ * UTrie auxTrie;                               -- size in bytes=indexes[_NORM_INDEX_AUX_TRIE_SIZE]
+ *
+ * uint16_t canonStartSets[canonStartSetsTop]   -- canonStartSetsTop=indexes[_NORM_INDEX_CANON_SET_COUNT]
+ *                                                 serialized USets and binary search tables, see below
+ *
+ *
+ * The indexes array contains lengths and sizes of the following arrays and structures
+ * as well as the following values:
+ *  indexes[_NORM_INDEX_COMBINE_FWD_COUNT]=combineFwdTop
+ *      -- one more than the highest combining index computed for forward-only-combining characters
+ *  indexes[_NORM_INDEX_COMBINE_BOTH_COUNT]=combineBothTop-combineFwdTop
+ *      -- number of combining indexes computed for both-ways-combining characters
+ *  indexes[_NORM_INDEX_COMBINE_BACK_COUNT]=combineBackTop-combineBothTop
+ *      -- number of combining indexes computed for backward-only-combining characters
+ *
+ *  indexes[_NORM_INDEX_MIN_NF*_NO_MAYBE] (where *={ C, D, KC, KD })
+ *      -- first code point with a quick check NF* value of NO/MAYBE
+ *
+ *
+ * - Tries
+ *
+ * The main structures are two UTrie tables ("compact arrays"),
+ * each with one index array and one data array.
+ * See utrie.h and utrie.c.
+ *
+ *
+ * - Tries in unorm.dat
+ *
+ * The first trie (normTrie above)
+ * provides data for the NF* quick checks and normalization.
+ * The second trie (fcdTrie above) provides data just for FCD checks.
+ *
+ *
+ * - norm32 data words from the first trie
+ *
+ * The norm32Table contains one 32-bit word "norm32" per code point.
+ * It contains the following bit fields:
+ * 31..16   extra data index, _NORM_EXTRA_SHIFT is used to shift this field down
+ *          if this index is <_NORM_EXTRA_INDEX_TOP then it is an index into
+ *              extraData[] where variable-length normalization data for this
+ *              code point is found
+ *          if this index is <_NORM_EXTRA_INDEX_TOP+_NORM_EXTRA_SURROGATE_TOP
+ *              then this is a norm32 for a leading surrogate, and the index
+ *              value is used together with the following trailing surrogate
+ *              code unit in the second trie access
+ *          if this index is >=_NORM_EXTRA_INDEX_TOP+_NORM_EXTRA_SURROGATE_TOP
+ *              then this is a norm32 for a "special" character,
+ *              i.e., the character is a Hangul syllable or a Jamo
+ *              see _NORM_EXTRA_HANGUL etc.
+ *          generally, instead of extracting this index from the norm32 and
+ *              comparing it with the above constants,
+ *              the normalization code compares the entire norm32 value
+ *              with _NORM_MIN_SPECIAL, _NORM_SURROGATES_TOP, _NORM_MIN_HANGUL etc.
+ *
+ * 15..8    combining class (cc) according to UnicodeData.txt
+ *
+ *  7..6    _NORM_COMBINES_ANY flags, used in composition to see if a character
+ *              combines with any following or preceding character(s)
+ *              at all
+ *     7    _NORM_COMBINES_BACK
+ *     6    _NORM_COMBINES_FWD
+ *
+ *  5..0    quick check flags, set for "no" or "maybe", with separate flags for
+ *              each normalization form
+ *              the higher bits are "maybe" flags; for NF*D there are no such flags
+ *              the lower bits are "no" flags for all forms, in the same order
+ *              as the "maybe" flags,
+ *              which is (MSB to LSB): NFKD NFD NFKC NFC
+ *  5..4    _NORM_QC_ANY_MAYBE
+ *  3..0    _NORM_QC_ANY_NO
+ *              see further related constants
+ *
+ *
+ * - Extra data per code point
+ *
+ * "Extra data" is referenced by the index in norm32.
+ * It is variable-length data. It is only present, and only those parts
+ * of it are, as needed for a given character.
+ * The norm32 extra data index is added to the beginning of extraData[]
+ * to get to a vector of 16-bit words with data at the following offsets:
+ *
+ * [-1]     Combining index for composition.
+ *              Stored only if norm32&_NORM_COMBINES_ANY .
+ * [0]      Lengths of the canonical and compatibility decomposition strings.
+ *              Stored only if there are decompositions, i.e.,
+ *              if norm32&(_NORM_QC_NFD|_NORM_QC_NFKD)
+ *          High byte: length of NFKD, or 0 if none
+ *          Low byte: length of NFD, or 0 if none
+ *          Each length byte also has another flag:
+ *              Bit 7 of a length byte is set if there are non-zero
+ *              combining classes (cc's) associated with the respective
+ *              decomposition. If this flag is set, then the decomposition
+ *              is preceded by a 16-bit word that contains the
+ *              leading and trailing cc's.
+ *              Bits 6..0 of a length byte are the length of the
+ *              decomposition string, not counting the cc word.
+ * [1..n]   NFD
+ * [n+1..]  NFKD
+ *
+ * Each of the two decompositions consists of up to two parts:
+ * - The 16-bit words with the leading and trailing cc's.
+ *   This is only stored if bit 7 of the corresponding length byte
+ *   is set. In this case, at least one of the cc's is not zero.
+ *   High byte: leading cc==cc of the first code point in the decomposition string
+ *   Low byte: trailing cc==cc of the last code point in the decomposition string
+ * - The decomposition string in UTF-16, with length code units.
+ *
+ *
+ * - Combining indexes and combiningTable[]
+ *
+ * Combining indexes are stored at the [-1] offset of the extra data
+ * if the character combines forward or backward with any other characters.
+ * They are used for (re)composition in NF*C.
+ * Values of combining indexes are arranged according to whether a character
+ * combines forward, backward, or both ways:
+ *    forward-only < both ways < backward-only
+ *
+ * The index values for forward-only and both-ways combining characters
+ * are indexes into the combiningTable[].
+ * The index values for backward-only combining characters are simply
+ * incremented from the preceding index values to be unique.
+ *
+ * In the combiningTable[], a variable-length list
+ * of variable-length (back-index, code point) pair entries is stored
+ * for each forward-combining character.
+ *
+ * These back-indexes are the combining indexes of both-ways or backward-only
+ * combining characters that the forward-combining character combines with.
+ *
+ * Each list is sorted in ascending order of back-indexes.
+ * Each list is terminated with the last back-index having bit 15 set.
+ *
+ * Each pair (back-index, code point) takes up either 2 or 3
+ * 16-bit words.
+ * The first word of a list entry is the back-index, with its bit 15 set if
+ * this is the last pair in the list.
+ *
+ * The second word contains flags in bits 15..13 that determine
+ * if there is a third word and how the combined character is encoded:
+ * 15   set if there is a third word in this list entry
+ * 14   set if the result is a supplementary character
+ * 13   set if the result itself combines forward
+ *
+ * According to these bits 15..14 of the second word,
+ * the result character is encoded as follows:
+ * 00 or 01 The result is <=0x1fff and stored in bits 12..0 of
+ *          the second word.
+ * 10       The result is 0x2000..0xffff and stored in the third word.
+ *          Bits 12..0 of the second word are not used.
+ * 11       The result is a supplementary character.
+ *          Bits 9..0 of the leading surrogate are in bits 9..0 of
+ *          the second word.
+ *          Add 0xd800 to these bits to get the complete surrogate.
+ *          Bits 12..10 of the second word are not used.
+ *          The trailing surrogate is stored in the third word.
+ *
+ *
+ * - FCD trie
+ *
+ * The FCD trie is very simple.
+ * It is a folded trie with 16-bit data words.
+ * In each word, the high byte contains the leading cc of the character,
+ * and the low byte contains the trailing cc of the character.
+ * These cc's are the cc's of the first and last code points in the
+ * canonical decomposition of the character.
+ *
+ * Since all 16 bits are used for cc's, lead surrogates must be tested
+ * by checking the code unit instead of the trie data.
+ * This is done only if the 16-bit data word is not zero.
+ * If the code unit is a leading surrogate and the data word is not zero,
+ * then instead of cc's it contains the offset for the second trie lookup.
+ *
+ *
+ * - Auxiliary trie and data
+ *
+ * The auxiliary 16-bit trie contains data for additional properties.
+ * Bits
+ * 15..13   reserved
+ *     12   not NFC_Skippable (f) (formatVersion>=2.2)
+ *     11   flag: not a safe starter for canonical closure
+ *     10   composition exclusion
+ *  9.. 0   index into extraData[] to FC_NFKC_Closure string
+ *          (not for lead surrogate),
+ *          or lead surrogate offset (for lead surrogate, if 9..0 not zero)
+ *
+ * - FC_NFKC_Closure strings in extraData[]
+ *
+ * Strings are either stored as a single code unit or as the length
+ * followed by that many units.
+ *   const char16_t *s=extraData+(index from auxTrie data bits 9..0);
+ *   int32_t length;
+ *   if(*s<0xff00) {
+ *     // s points to the single-unit string
+ *     length=1;
+ *   } else {
+ *     length=*s&0xff;
+ *     ++s;
+ *   }
+ *
+ * Conditions for "NF* Skippable" from Mark Davis' com.ibm.text.UCD.NFSkippable:
+ * (used in NormalizerTransliterator)
+ *
+ * A skippable character is
+ * a) unassigned, or ALL of the following:
+ * b) of combining class 0.
+ * c) not decomposed by this normalization form.
+ * AND if NFC or NFKC,
+ * d) can never compose with a previous character.
+ * e) can never compose with a following character.
+ * f) can never change if another character is added.
+ *    Example: a-breve might satisfy all but f, but if you
+ *    add an ogonek it changes to a-ogonek + breve
+ *
+ * a)..e) must be tested from norm32.
+ * Since f) is more complicated, the (not-)NFC_Skippable flag (f) is built
+ * into the auxiliary trie.
+ * The same bit is used for NFC and NFKC; (c) differs for them.
+ * As usual, we build the "not skippable" flags so that unassigned
+ * code points get a 0 bit.
+ * This bit is only valid after (a)..(e) test false; test NFD_NO before (f) as well.
+ * Test Hangul LV syllables entirely in code.
+ *
+ *
+ * - structure inside canonStartSets[]
+ *
+ * This array maps from code points c to sets of code points (USerializedSet).
+ * The result sets are the code points whose canonical decompositions start
+ * with c.
+ *
+ * canonStartSets[] contains the following sub-arrays:
+ *
+ * indexes[_NORM_SET_INDEX_TOP]
+ *   - contains lengths of sub-arrays etc.
+ *
+ * startSets[indexes[_NORM_SET_INDEX_CANON_SETS_LENGTH]-_NORM_SET_INDEX_TOP]
+ *   - contains serialized sets (USerializedSet) of canonical starters for
+ *     enumerating canonically equivalent strings
+ *     indexes[_NORM_SET_INDEX_CANON_SETS_LENGTH] includes _NORM_SET_INDEX_TOP
+ *     for details about the structure see uset.c
+ *
+ * bmpTable[indexes[_NORM_SET_INDEX_CANON_BMP_TABLE_LENGTH]]
+ *   - a sorted search table for BMP code points whose results are
+ *     either indexes to USerializedSets or single code points for
+ *     single-code point sets;
+ *     each entry is a pair of { code point, result } with result=(binary) yy xxxxxx xxxxxxxx
+ *     if yy==01 then there is a USerializedSet at canonStartSets+x
+ *     else build a USerializedSet with result as the single code point
+ *
+ * suppTable[indexes[_NORM_SET_INDEX_CANON_SUPP_TABLE_LENGTH]]
+ *   - a sorted search table for supplementary code points whose results are
+ *     either indexes to USerializedSets or single code points for
+ *     single-code point sets;
+ *     each entry is a triplet of { high16(cp), low16(cp), result }
+ *     each code point's high-word may contain extra data in bits 15..5:
+ *     if the high word has bit 15 set, then build a set with a single code point
+ *     which is (((high16(cp)&0x1f00)<<8)|result;
+ *     else there is a USerializedSet at canonStartSets+result
+ *
+ * FormatVersion 2.3 adds 2 serialized sets for normalization exclusions.
+ * They are stored in the data file so that the runtime normalization code need
+ * not depend on other properties and their data and implementation files.
+ * The _NORM_SET_INDEX_NX_..._OFFSET offsets in the canonStartSets index table
+ * give the location for each set.
+ * There is no set stored for UNORM_NX_HANGUL because it's trivial to create
+ * without using properties.
+ *
+ * Set contents:
+ *
+ * _NORM_SET_INDEX_NX_CJK_COMPAT_OFFSET (for UNORM_NX_CJK_COMPAT)
+ *     [[:Ideographic:]&[:NFD_QC=No:]]
+ *     =[CJK Ideographs]&[has canonical decomposition]
+ *
+ * _NORM_SET_INDEX_NX_UNICODE32_OFFSET (for UNORM_UNICODE_3_2)
+ *     [:^Age=3.2:]
+ *     =set with all code points that were not designated by the specified Unicode version
+ *
+ * _NORM_SET_INDEX_NX_RESERVED_OFFSET
+ *     This is an offset that points to where the next, future set would start.
+ *     Currently it indicates where the previous set ends, and thus its length.
+ *     The name for this enum constant may in the future be applied to different
+ *     index slots. In order to get the limit of a set, use its index slot and
+ *     the immediately following one regardless of that one's enum name.
+ */
+
+#endif /* #if !UCONFIG_NO_NORMALIZATION */
+
+#endif
diff --git a/intl/icu/source/common/uobject.cpp b/intl/icu/source/common/uobject.cpp
new file mode 100644
index 0000000000..c9e435ceca
--- /dev/null
+++ b/intl/icu/source/common/uobject.cpp
@@ -0,0 +1,105 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+*
+*   Copyright (C) 2002-2012, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+******************************************************************************
+*   file name:  uobject.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2002jun26
+*   created by: Markus W. Scherer
+*/
+
+#include "unicode/uobject.h"
+#include "cmemory.h"
+
+U_NAMESPACE_BEGIN
+
+#if U_OVERRIDE_CXX_ALLOCATION
+
+/*
+ * Default implementation of UMemory::new/delete
+ * using uprv_malloc() and uprv_free().
+ *
+ * For testing, this is used together with a list of imported symbols to verify
+ * that ICU is not using the global ::new and ::delete operators.
+ *
+ * These operators can be implemented like this or any other appropriate way
+ * when customizing ICU for certain environments.
+ * Whenever ICU is customized in binary incompatible ways please be sure
+ * to use library name suffixes to distinguish such libraries from
+ * the standard build.
+ *
+ * Instead of just modifying these C++ new/delete operators, it is usually best
+ * to modify the uprv_malloc()/uprv_free()/uprv_realloc() functions in cmemory.c.
+ *
+ * Memory test on Windows/MSVC 6:
+ * The global operators new and delete look as follows:
+ *   04F 00000000 UNDEF  notype ()    External     | ??2@YAPAXI@Z (void * __cdecl operator new(unsigned int))
+ *   03F 00000000 UNDEF  notype ()    External     | ??3@YAXPAX@Z (void __cdecl operator delete(void *))
+ *
+ * These lines are from output generated by the MSVC 6 tool dumpbin with
+ * dumpbin /symbols *.obj
+ *
+ * ??2@YAPAXI@Z and ??3@YAXPAX@Z are the linker symbols in the .obj
+ * files and are imported from msvcrtd.dll (in a debug build).
+ *
+ * Make sure that with the UMemory operators new and delete defined these two symbols
+ * do not appear in the dumpbin /symbols output for the ICU libraries!
+ *
+ * If such a symbol appears in the output then look in the preceding lines in the output
+ * for which file and function calls the global new or delete operator,
+ * and replace with uprv_malloc/uprv_free.
+ */
+
+void * U_EXPORT2 UMemory::operator new(size_t size) noexcept {
+    return uprv_malloc(size);
+}
+
+void U_EXPORT2 UMemory::operator delete(void *p) noexcept {
+    if(p!=nullptr) {
+        uprv_free(p);
+    }
+}
+
+void * U_EXPORT2 UMemory::operator new[](size_t size) noexcept {
+    return uprv_malloc(size);
+}
+
+void U_EXPORT2 UMemory::operator delete[](void *p) noexcept {
+    if(p!=nullptr) {
+        uprv_free(p);
+    }
+}
+
+#if U_HAVE_DEBUG_LOCATION_NEW
+void * U_EXPORT2 UMemory::operator new(size_t size, const char* /*file*/, int /*line*/) noexcept {
+    return UMemory::operator new(size);
+}
+
+void U_EXPORT2 UMemory::operator delete(void* p, const char* /*file*/, int /*line*/) noexcept {
+    UMemory::operator delete(p);
+}
+#endif /* U_HAVE_DEBUG_LOCATION_NEW */
+
+
+#endif
+
+UObject::~UObject() {}
+
+UClassID UObject::getDynamicClassID() const { return nullptr; }
+
+U_NAMESPACE_END
+
+U_NAMESPACE_USE
+
+U_CAPI void U_EXPORT2
+uprv_deleteUObject(void *obj) {
+    delete static_cast(obj);
+}
diff --git a/intl/icu/source/common/uposixdefs.h b/intl/icu/source/common/uposixdefs.h
new file mode 100644
index 0000000000..826c9bb47a
--- /dev/null
+++ b/intl/icu/source/common/uposixdefs.h
@@ -0,0 +1,77 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*   Copyright (C) 2011-2015, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*******************************************************************************
+*   file name:  uposixdefs.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2011jul25
+*   created by: Markus W. Scherer
+*
+*   Common definitions for implementation files working with POSIX functions.
+*   *Important*: #include this file before any other header files!
+*/
+
+#ifndef __UPOSIXDEFS_H__
+#define __UPOSIXDEFS_H__
+
+/*
+ * Define _XOPEN_SOURCE for access to POSIX functions.
+ *
+ * We cannot use U_PLATFORM from platform.h/utypes.h because
+ * "The Open Group Base Specifications"
+ * chapter "2.2 The Compilation Environment" says:
+ * "In the compilation of an application that #defines a feature test macro
+ * specified by IEEE Std 1003.1-2001,
+ * no header defined by IEEE Std 1003.1-2001 shall be included prior to
+ * the definition of the feature test macro."
+ */
+#ifdef _XOPEN_SOURCE
+    /* Use the predefined value. */
+#else
+    /*
+     * Version 6.0:
+     * The Open Group Base Specifications Issue 6 (IEEE Std 1003.1, 2004 Edition)
+     * also known as
+     * SUSv3 = Open Group Single UNIX Specification, Version 3 (UNIX03)
+     *
+     * Note: This definition used to be in C source code (e.g., putil.c)
+     * and define _XOPEN_SOURCE to different values depending on __STDC_VERSION__.
+     * In C++ source code (e.g., putil.cpp), __STDC_VERSION__ is not defined at all.
+     */
+#   define _XOPEN_SOURCE 600
+#endif
+
+/*
+ * Make sure things like realpath and such functions work.
+ * Poorly upgraded Solaris machines can't have this defined.
+ * Cleanly installed Solaris can use this #define.
+ *
+ * z/OS needs this definition for timeval and to get usleep.
+ */
+#if !defined(_XOPEN_SOURCE_EXTENDED) && defined(__TOS_MVS__)
+#   define _XOPEN_SOURCE_EXTENDED 1
+#endif
+
+/**
+ * Solaris says:
+ *   "...it is invalid to compile an XPG6 or a POSIX.1-2001 application with anything other
+ *   than a c99 or later compiler."
+ * Apparently C++11 is not "or later". Work around this.
+ */
+#if defined(__cplusplus) && (defined(sun) || defined(__sun)) && !defined (_STDC_C99)
+#   define _STDC_C99
+#endif
+
+#if !defined _POSIX_C_SOURCE && \
+    defined(__APPLE__) && defined(__MACH__) && !defined(__clang__)
+// Needed to prevent EOWNERDEAD issues with GCC on Mac
+#define _POSIX_C_SOURCE 200809L
+#endif
+
+#endif  /* __UPOSIXDEFS_H__ */
diff --git a/intl/icu/source/common/uprops.cpp b/intl/icu/source/common/uprops.cpp
new file mode 100644
index 0000000000..28540186c6
--- /dev/null
+++ b/intl/icu/source/common/uprops.cpp
@@ -0,0 +1,843 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2002-2016, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  uprops.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2002feb24
+*   created by: Markus W. Scherer
+*
+*   Implementations for mostly non-core Unicode character properties
+*   stored in uprops.icu.
+*
+*   With the APIs implemented here, almost all properties files and
+*   their associated implementation files are used from this file,
+*   including those for normalization and case mappings.
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/uchar.h"
+#include "unicode/ucptrie.h"
+#include "unicode/udata.h"
+#include "unicode/unorm2.h"
+#include "unicode/uscript.h"
+#include "unicode/ustring.h"
+#include "unicode/utf16.h"
+#include "cstring.h"
+#include "emojiprops.h"
+#include "mutex.h"
+#include "normalizer2impl.h"
+#include "umutex.h"
+#include "ubidi_props.h"
+#include "uprops.h"
+#include "ucase.h"
+#include "ucln_cmn.h"
+#include "ulayout_props.h"
+#include "ustr_imp.h"
+
+U_NAMESPACE_USE
+
+// Unicode text layout properties data -----------------------------------------
+
+namespace {
+
+icu::UInitOnce gLayoutInitOnce {};
+UDataMemory *gLayoutMemory = nullptr;
+
+UCPTrie *gInpcTrie = nullptr;  // Indic_Positional_Category
+UCPTrie *gInscTrie = nullptr;  // Indic_Syllabic_Category
+UCPTrie *gVoTrie = nullptr;  // Vertical_Orientation
+
+int32_t gMaxInpcValue = 0;
+int32_t gMaxInscValue = 0;
+int32_t gMaxVoValue = 0;
+
+UBool U_CALLCONV uprops_cleanup() {
+    udata_close(gLayoutMemory);
+    gLayoutMemory = nullptr;
+
+    ucptrie_close(gInpcTrie);
+    gInpcTrie = nullptr;
+    ucptrie_close(gInscTrie);
+    gInscTrie = nullptr;
+    ucptrie_close(gVoTrie);
+    gVoTrie = nullptr;
+
+    gMaxInpcValue = 0;
+    gMaxInscValue = 0;
+    gMaxVoValue = 0;
+
+    gLayoutInitOnce.reset();
+    return true;
+}
+
+UBool U_CALLCONV
+ulayout_isAcceptable(void * /*context*/,
+                     const char * /* type */, const char * /*name*/,
+                     const UDataInfo *pInfo) {
+    return pInfo->size >= 20 &&
+        pInfo->isBigEndian == U_IS_BIG_ENDIAN &&
+        pInfo->charsetFamily == U_CHARSET_FAMILY &&
+        pInfo->dataFormat[0] == ULAYOUT_FMT_0 &&
+        pInfo->dataFormat[1] == ULAYOUT_FMT_1 &&
+        pInfo->dataFormat[2] == ULAYOUT_FMT_2 &&
+        pInfo->dataFormat[3] == ULAYOUT_FMT_3 &&
+        pInfo->formatVersion[0] == 1;
+}
+
+// UInitOnce singleton initialization function
+void U_CALLCONV ulayout_load(UErrorCode &errorCode) {
+    gLayoutMemory = udata_openChoice(
+        nullptr, ULAYOUT_DATA_TYPE, ULAYOUT_DATA_NAME,
+        ulayout_isAcceptable, nullptr, &errorCode);
+    if (U_FAILURE(errorCode)) { return; }
+
+    const uint8_t *inBytes = (const uint8_t *)udata_getMemory(gLayoutMemory);
+    const int32_t *inIndexes = (const int32_t *)inBytes;
+    int32_t indexesLength = inIndexes[ULAYOUT_IX_INDEXES_LENGTH];
+    if (indexesLength < 12) {
+        errorCode = U_INVALID_FORMAT_ERROR;  // Not enough indexes.
+        return;
+    }
+    int32_t offset = indexesLength * 4;
+    int32_t top = inIndexes[ULAYOUT_IX_INPC_TRIE_TOP];
+    int32_t trieSize = top - offset;
+    if (trieSize >= 16) {
+        gInpcTrie = ucptrie_openFromBinary(
+            UCPTRIE_TYPE_ANY, UCPTRIE_VALUE_BITS_ANY,
+            inBytes + offset, trieSize, nullptr, &errorCode);
+    }
+    offset = top;
+    top = inIndexes[ULAYOUT_IX_INSC_TRIE_TOP];
+    trieSize = top - offset;
+    if (trieSize >= 16) {
+        gInscTrie = ucptrie_openFromBinary(
+            UCPTRIE_TYPE_ANY, UCPTRIE_VALUE_BITS_ANY,
+            inBytes + offset, trieSize, nullptr, &errorCode);
+    }
+    offset = top;
+    top = inIndexes[ULAYOUT_IX_VO_TRIE_TOP];
+    trieSize = top - offset;
+    if (trieSize >= 16) {
+        gVoTrie = ucptrie_openFromBinary(
+            UCPTRIE_TYPE_ANY, UCPTRIE_VALUE_BITS_ANY,
+            inBytes + offset, trieSize, nullptr, &errorCode);
+    }
+
+    uint32_t maxValues = inIndexes[ULAYOUT_IX_MAX_VALUES];
+    gMaxInpcValue = maxValues >> ULAYOUT_MAX_INPC_SHIFT;
+    gMaxInscValue = (maxValues >> ULAYOUT_MAX_INSC_SHIFT) & 0xff;
+    gMaxVoValue = (maxValues >> ULAYOUT_MAX_VO_SHIFT) & 0xff;
+
+    ucln_common_registerCleanup(UCLN_COMMON_UPROPS, uprops_cleanup);
+}
+
+UBool ulayout_ensureData(UErrorCode &errorCode) {
+    if (U_FAILURE(errorCode)) { return false; }
+    umtx_initOnce(gLayoutInitOnce, &ulayout_load, errorCode);
+    return U_SUCCESS(errorCode);
+}
+
+UBool ulayout_ensureData() {
+    UErrorCode errorCode = U_ZERO_ERROR;
+    return ulayout_ensureData(errorCode);
+}
+
+}  // namespace
+
+/* general properties API functions ----------------------------------------- */
+
+struct BinaryProperty;
+
+typedef UBool BinaryPropertyContains(const BinaryProperty &prop, UChar32 c, UProperty which);
+
+struct BinaryProperty {
+    int32_t column;  // SRC_PROPSVEC column, or "source" if mask==0
+    uint32_t mask;
+    BinaryPropertyContains *contains;
+};
+
+static UBool defaultContains(const BinaryProperty &prop, UChar32 c, UProperty /*which*/) {
+    /* systematic, directly stored properties */
+    return (u_getUnicodeProperties(c, prop.column)&prop.mask)!=0;
+}
+
+static UBool caseBinaryPropertyContains(const BinaryProperty &/*prop*/, UChar32 c, UProperty which) {
+    return static_cast(ucase_hasBinaryProperty(c, which));
+}
+
+static UBool isBidiControl(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return ubidi_isBidiControl(c);
+}
+
+static UBool isMirrored(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return ubidi_isMirrored(c);
+}
+
+static UBool isJoinControl(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return ubidi_isJoinControl(c);
+}
+
+#if UCONFIG_NO_NORMALIZATION
+static UBool hasFullCompositionExclusion(const BinaryProperty &, UChar32, UProperty) {
+    return false;
+}
+#else
+static UBool hasFullCompositionExclusion(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    // By definition, Full_Composition_Exclusion is the same as NFC_QC=No.
+    UErrorCode errorCode=U_ZERO_ERROR;
+    const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode);
+    return U_SUCCESS(errorCode) && impl->isCompNo(impl->getNorm16(c));
+}
+#endif
+
+// UCHAR_NF*_INERT properties
+#if UCONFIG_NO_NORMALIZATION
+static UBool isNormInert(const BinaryProperty &, UChar32, UProperty) {
+    return false;
+}
+#else
+static UBool isNormInert(const BinaryProperty &/*prop*/, UChar32 c, UProperty which) {
+    UErrorCode errorCode=U_ZERO_ERROR;
+    const Normalizer2 *norm2=Normalizer2Factory::getInstance(
+        (UNormalizationMode)(which-UCHAR_NFD_INERT+UNORM_NFD), errorCode);
+    return U_SUCCESS(errorCode) && norm2->isInert(c);
+}
+#endif
+
+#if UCONFIG_NO_NORMALIZATION
+static UBool changesWhenCasefolded(const BinaryProperty &, UChar32, UProperty) {
+    return false;
+}
+#else
+static UBool changesWhenCasefolded(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    UnicodeString nfd;
+    UErrorCode errorCode=U_ZERO_ERROR;
+    const Normalizer2 *nfcNorm2=Normalizer2::getNFCInstance(errorCode);
+    if(U_FAILURE(errorCode)) {
+        return false;
+    }
+    if(nfcNorm2->getDecomposition(c, nfd)) {
+        /* c has a decomposition */
+        if(nfd.length()==1) {
+            c=nfd[0];  /* single BMP code point */
+        } else if(nfd.length()<=U16_MAX_LENGTH &&
+                  nfd.length()==U16_LENGTH(c=nfd.char32At(0))
+        ) {
+            /* single supplementary code point */
+        } else {
+            c=U_SENTINEL;
+        }
+    } else if(c<0) {
+        return false;  /* protect against bad input */
+    }
+    if(c>=0) {
+        /* single code point */
+        const char16_t *resultString;
+        return (UBool)(ucase_toFullFolding(c, &resultString, U_FOLD_CASE_DEFAULT)>=0);
+    } else {
+        /* guess some large but stack-friendly capacity */
+        char16_t dest[2*UCASE_MAX_STRING_LENGTH];
+        int32_t destLength;
+        destLength=u_strFoldCase(dest, UPRV_LENGTHOF(dest),
+                                  nfd.getBuffer(), nfd.length(),
+                                  U_FOLD_CASE_DEFAULT, &errorCode);
+        return (UBool)(U_SUCCESS(errorCode) &&
+                       0!=u_strCompare(nfd.getBuffer(), nfd.length(),
+                                       dest, destLength, false));
+    }
+}
+#endif
+
+#if UCONFIG_NO_NORMALIZATION
+static UBool changesWhenNFKC_Casefolded(const BinaryProperty &, UChar32, UProperty) {
+    return false;
+}
+#else
+static UBool changesWhenNFKC_Casefolded(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    UErrorCode errorCode=U_ZERO_ERROR;
+    const Normalizer2Impl *kcf=Normalizer2Factory::getNFKC_CFImpl(errorCode);
+    if(U_FAILURE(errorCode)) {
+        return false;
+    }
+    UnicodeString src(c);
+    UnicodeString dest;
+    {
+        // The ReorderingBuffer must be in a block because its destructor
+        // needs to release dest's buffer before we look at its contents.
+        ReorderingBuffer buffer(*kcf, dest);
+        // Small destCapacity for NFKC_CF(c).
+        if(buffer.init(5, errorCode)) {
+            const char16_t *srcArray=src.getBuffer();
+            kcf->compose(srcArray, srcArray+src.length(), false,
+                          true, buffer, errorCode);
+        }
+    }
+    return U_SUCCESS(errorCode) && dest!=src;
+}
+#endif
+
+#if UCONFIG_NO_NORMALIZATION
+static UBool isCanonSegmentStarter(const BinaryProperty &, UChar32, UProperty) {
+    return false;
+}
+#else
+static UBool isCanonSegmentStarter(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    UErrorCode errorCode=U_ZERO_ERROR;
+    const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode);
+    return
+        U_SUCCESS(errorCode) && impl->ensureCanonIterData(errorCode) &&
+        impl->isCanonSegmentStarter(c);
+}
+#endif
+
+static UBool isPOSIX_alnum(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return u_isalnumPOSIX(c);
+}
+
+static UBool isPOSIX_blank(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return u_isblank(c);
+}
+
+static UBool isPOSIX_graph(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return u_isgraphPOSIX(c);
+}
+
+static UBool isPOSIX_print(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return u_isprintPOSIX(c);
+}
+
+static UBool isPOSIX_xdigit(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return u_isxdigit(c);
+}
+
+static UBool isRegionalIndicator(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    // Property starts are a subset of lb=RI etc.
+    return 0x1F1E6<=c && c<=0x1F1FF;
+}
+
+static UBool hasEmojiProperty(const BinaryProperty &/*prop*/, UChar32 c, UProperty which) {
+    return EmojiProps::hasBinaryProperty(c, which);
+}
+
+static const BinaryProperty binProps[UCHAR_BINARY_LIMIT]={
+    /*
+     * column and mask values for binary properties from u_getUnicodeProperties().
+     * Must be in order of corresponding UProperty,
+     * and there must be exactly one entry per binary UProperty.
+     *
+     * Properties with mask==0 are handled in code.
+     * For them, column is the UPropertySource value.
+     */
+    { 1,                U_MASK(UPROPS_ALPHABETIC), defaultContains },
+    { 1,                U_MASK(UPROPS_ASCII_HEX_DIGIT), defaultContains },
+    { UPROPS_SRC_BIDI,  0, isBidiControl },
+    { UPROPS_SRC_BIDI,  0, isMirrored },
+    { 1,                U_MASK(UPROPS_DASH), defaultContains },
+    { 1,                U_MASK(UPROPS_DEFAULT_IGNORABLE_CODE_POINT), defaultContains },
+    { 1,                U_MASK(UPROPS_DEPRECATED), defaultContains },
+    { 1,                U_MASK(UPROPS_DIACRITIC), defaultContains },
+    { 1,                U_MASK(UPROPS_EXTENDER), defaultContains },
+    { UPROPS_SRC_NFC,   0, hasFullCompositionExclusion },
+    { 1,                U_MASK(UPROPS_GRAPHEME_BASE), defaultContains },
+    { 1,                U_MASK(UPROPS_GRAPHEME_EXTEND), defaultContains },
+    { 1,                U_MASK(UPROPS_GRAPHEME_LINK), defaultContains },
+    { 1,                U_MASK(UPROPS_HEX_DIGIT), defaultContains },
+    { 1,                U_MASK(UPROPS_HYPHEN), defaultContains },
+    { 1,                U_MASK(UPROPS_ID_CONTINUE), defaultContains },
+    { 1,                U_MASK(UPROPS_ID_START), defaultContains },
+    { 1,                U_MASK(UPROPS_IDEOGRAPHIC), defaultContains },
+    { 1,                U_MASK(UPROPS_IDS_BINARY_OPERATOR), defaultContains },
+    { 1,                U_MASK(UPROPS_IDS_TRINARY_OPERATOR), defaultContains },
+    { UPROPS_SRC_BIDI,  0, isJoinControl },
+    { 1,                U_MASK(UPROPS_LOGICAL_ORDER_EXCEPTION), defaultContains },
+    { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_LOWERCASE
+    { 1,                U_MASK(UPROPS_MATH), defaultContains },
+    { 1,                U_MASK(UPROPS_NONCHARACTER_CODE_POINT), defaultContains },
+    { 1,                U_MASK(UPROPS_QUOTATION_MARK), defaultContains },
+    { 1,                U_MASK(UPROPS_RADICAL), defaultContains },
+    { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_SOFT_DOTTED
+    { 1,                U_MASK(UPROPS_TERMINAL_PUNCTUATION), defaultContains },
+    { 1,                U_MASK(UPROPS_UNIFIED_IDEOGRAPH), defaultContains },
+    { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_UPPERCASE
+    { 1,                U_MASK(UPROPS_WHITE_SPACE), defaultContains },
+    { 1,                U_MASK(UPROPS_XID_CONTINUE), defaultContains },
+    { 1,                U_MASK(UPROPS_XID_START), defaultContains },
+    { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CASE_SENSITIVE
+    { 1,                U_MASK(UPROPS_S_TERM), defaultContains },
+    { 1,                U_MASK(UPROPS_VARIATION_SELECTOR), defaultContains },
+    { UPROPS_SRC_NFC,   0, isNormInert },  // UCHAR_NFD_INERT
+    { UPROPS_SRC_NFKC,  0, isNormInert },  // UCHAR_NFKD_INERT
+    { UPROPS_SRC_NFC,   0, isNormInert },  // UCHAR_NFC_INERT
+    { UPROPS_SRC_NFKC,  0, isNormInert },  // UCHAR_NFKC_INERT
+    { UPROPS_SRC_NFC_CANON_ITER, 0, isCanonSegmentStarter },
+    { 1,                U_MASK(UPROPS_PATTERN_SYNTAX), defaultContains },
+    { 1,                U_MASK(UPROPS_PATTERN_WHITE_SPACE), defaultContains },
+    { UPROPS_SRC_CHAR_AND_PROPSVEC,  0, isPOSIX_alnum },
+    { UPROPS_SRC_CHAR,  0, isPOSIX_blank },
+    { UPROPS_SRC_CHAR,  0, isPOSIX_graph },
+    { UPROPS_SRC_CHAR,  0, isPOSIX_print },
+    { UPROPS_SRC_CHAR,  0, isPOSIX_xdigit },
+    { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CASED
+    { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CASE_IGNORABLE
+    { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CHANGES_WHEN_LOWERCASED
+    { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CHANGES_WHEN_UPPERCASED
+    { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CHANGES_WHEN_TITLECASED
+    { UPROPS_SRC_CASE_AND_NORM,  0, changesWhenCasefolded },
+    { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CHANGES_WHEN_CASEMAPPED
+    { UPROPS_SRC_NFKC_CF, 0, changesWhenNFKC_Casefolded },
+    { UPROPS_SRC_EMOJI, 0, hasEmojiProperty },  // UCHAR_EMOJI
+    { UPROPS_SRC_EMOJI, 0, hasEmojiProperty },  // UCHAR_EMOJI_PRESENTATION
+    { UPROPS_SRC_EMOJI, 0, hasEmojiProperty },  // UCHAR_EMOJI_MODIFIER
+    { UPROPS_SRC_EMOJI, 0, hasEmojiProperty },  // UCHAR_EMOJI_MODIFIER_BASE
+    { UPROPS_SRC_EMOJI, 0, hasEmojiProperty },  // UCHAR_EMOJI_COMPONENT
+    { 2,                0, isRegionalIndicator },
+    { 1,                U_MASK(UPROPS_PREPENDED_CONCATENATION_MARK), defaultContains },
+    { UPROPS_SRC_EMOJI, 0, hasEmojiProperty },  // UCHAR_EXTENDED_PICTOGRAPHIC
+    { UPROPS_SRC_EMOJI, 0, hasEmojiProperty },  // UCHAR_BASIC_EMOJI
+    { UPROPS_SRC_EMOJI, 0, hasEmojiProperty },  // UCHAR_EMOJI_KEYCAP_SEQUENCE
+    { UPROPS_SRC_EMOJI, 0, hasEmojiProperty },  // UCHAR_RGI_EMOJI_MODIFIER_SEQUENCE
+    { UPROPS_SRC_EMOJI, 0, hasEmojiProperty },  // UCHAR_RGI_EMOJI_FLAG_SEQUENCE
+    { UPROPS_SRC_EMOJI, 0, hasEmojiProperty },  // UCHAR_RGI_EMOJI_TAG_SEQUENCE
+    { UPROPS_SRC_EMOJI, 0, hasEmojiProperty },  // UCHAR_RGI_EMOJI_ZWJ_SEQUENCE
+    { UPROPS_SRC_EMOJI, 0, hasEmojiProperty },  // UCHAR_RGI_EMOJI
+};
+
+U_CAPI UBool U_EXPORT2
+u_hasBinaryProperty(UChar32 c, UProperty which) {
+    /* c is range-checked in the functions that are called from here */
+    if(which 0 ? i == length : s[i] == 0) {
+            return u_hasBinaryProperty(c, which);  // single code point
+        }
+    }
+    // Only call into EmojiProps for a relevant property,
+    // so that we not unnecessarily try to load its data file.
+    return UCHAR_BASIC_EMOJI <= which && which <= UCHAR_RGI_EMOJI &&
+        EmojiProps::hasBinaryProperty(s, length, which);
+}
+
+struct IntProperty;
+
+typedef int32_t IntPropertyGetValue(const IntProperty &prop, UChar32 c, UProperty which);
+typedef int32_t IntPropertyGetMaxValue(const IntProperty &prop, UProperty which);
+
+struct IntProperty {
+    int32_t column;  // SRC_PROPSVEC column, or "source" if mask==0
+    uint32_t mask;
+    int32_t shift;  // =maxValue if getMaxValueFromShift() is used
+    IntPropertyGetValue *getValue;
+    IntPropertyGetMaxValue *getMaxValue;
+};
+
+static int32_t defaultGetValue(const IntProperty &prop, UChar32 c, UProperty /*which*/) {
+    /* systematic, directly stored properties */
+    return (int32_t)(u_getUnicodeProperties(c, prop.column)&prop.mask)>>prop.shift;
+}
+
+static int32_t defaultGetMaxValue(const IntProperty &prop, UProperty /*which*/) {
+    return (uprv_getMaxValues(prop.column)&prop.mask)>>prop.shift;
+}
+
+static int32_t getMaxValueFromShift(const IntProperty &prop, UProperty /*which*/) {
+    return prop.shift;
+}
+
+static int32_t getBiDiClass(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return (int32_t)u_charDirection(c);
+}
+
+static int32_t getBiDiPairedBracketType(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return (int32_t)ubidi_getPairedBracketType(c);
+}
+
+static int32_t biDiGetMaxValue(const IntProperty &/*prop*/, UProperty which) {
+    return ubidi_getMaxValue(which);
+}
+
+#if UCONFIG_NO_NORMALIZATION
+static int32_t getCombiningClass(const IntProperty &, UChar32, UProperty) {
+    return 0;
+}
+#else
+static int32_t getCombiningClass(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return u_getCombiningClass(c);
+}
+#endif
+
+static int32_t getGeneralCategory(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return (int32_t)u_charType(c);
+}
+
+static int32_t getJoiningGroup(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return ubidi_getJoiningGroup(c);
+}
+
+static int32_t getJoiningType(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return ubidi_getJoiningType(c);
+}
+
+static int32_t getNumericType(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    int32_t ntv=(int32_t)GET_NUMERIC_TYPE_VALUE(u_getMainProperties(c));
+    return UPROPS_NTV_GET_TYPE(ntv);
+}
+
+static int32_t getScript(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    UErrorCode errorCode=U_ZERO_ERROR;
+    return (int32_t)uscript_getScript(c, &errorCode);
+}
+
+static int32_t scriptGetMaxValue(const IntProperty &/*prop*/, UProperty /*which*/) {
+    uint32_t scriptX=uprv_getMaxValues(0)&UPROPS_SCRIPT_X_MASK;
+    return uprops_mergeScriptCodeOrIndex(scriptX);
+}
+
+/*
+ * Map some of the Grapheme Cluster Break values to Hangul Syllable Types.
+ * Hangul_Syllable_Type is fully redundant with a subset of Grapheme_Cluster_Break.
+ */
+static const UHangulSyllableType gcbToHst[]={
+    U_HST_NOT_APPLICABLE,   /* U_GCB_OTHER */
+    U_HST_NOT_APPLICABLE,   /* U_GCB_CONTROL */
+    U_HST_NOT_APPLICABLE,   /* U_GCB_CR */
+    U_HST_NOT_APPLICABLE,   /* U_GCB_EXTEND */
+    U_HST_LEADING_JAMO,     /* U_GCB_L */
+    U_HST_NOT_APPLICABLE,   /* U_GCB_LF */
+    U_HST_LV_SYLLABLE,      /* U_GCB_LV */
+    U_HST_LVT_SYLLABLE,     /* U_GCB_LVT */
+    U_HST_TRAILING_JAMO,    /* U_GCB_T */
+    U_HST_VOWEL_JAMO        /* U_GCB_V */
+    /*
+     * Omit GCB values beyond what we need for hst.
+     * The code below checks for the array length.
+     */
+};
+
+static int32_t getHangulSyllableType(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    /* see comments on gcbToHst[] above */
+    int32_t gcb=(int32_t)(u_getUnicodeProperties(c, 2)&UPROPS_GCB_MASK)>>UPROPS_GCB_SHIFT;
+    if(gcb>8;
+}
+#endif
+
+#if UCONFIG_NO_NORMALIZATION
+static int32_t getTrailCombiningClass(const IntProperty &, UChar32, UProperty) {
+    return 0;
+}
+#else
+static int32_t getTrailCombiningClass(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
+    return unorm_getFCD16(c)&0xff;
+}
+#endif
+
+static int32_t getInPC(const IntProperty &, UChar32 c, UProperty) {
+    return ulayout_ensureData() && gInpcTrie != nullptr ? ucptrie_get(gInpcTrie, c) : 0;
+}
+
+static int32_t getInSC(const IntProperty &, UChar32 c, UProperty) {
+    return ulayout_ensureData() && gInscTrie != nullptr ? ucptrie_get(gInscTrie, c) : 0;
+}
+
+static int32_t getVo(const IntProperty &, UChar32 c, UProperty) {
+    return ulayout_ensureData() && gVoTrie != nullptr ? ucptrie_get(gVoTrie, c) : 0;
+}
+
+static int32_t layoutGetMaxValue(const IntProperty &/*prop*/, UProperty which) {
+    if (!ulayout_ensureData()) { return 0; }
+    switch (which) {
+    case UCHAR_INDIC_POSITIONAL_CATEGORY:
+        return gMaxInpcValue;
+    case UCHAR_INDIC_SYLLABIC_CATEGORY:
+        return gMaxInscValue;
+    case UCHAR_VERTICAL_ORIENTATION:
+        return gMaxVoValue;
+    default:
+        return 0;
+    }
+}
+
+static const IntProperty intProps[UCHAR_INT_LIMIT-UCHAR_INT_START]={
+    /*
+     * column, mask and shift values for int-value properties from u_getUnicodeProperties().
+     * Must be in order of corresponding UProperty,
+     * and there must be exactly one entry per int UProperty.
+     *
+     * Properties with mask==0 are handled in code.
+     * For them, column is the UPropertySource value.
+     */
+    { UPROPS_SRC_BIDI,  0, 0,                               getBiDiClass, biDiGetMaxValue },
+    { 0,                UPROPS_BLOCK_MASK, UPROPS_BLOCK_SHIFT, defaultGetValue, defaultGetMaxValue },
+    { UPROPS_SRC_NFC,   0, 0xff,                            getCombiningClass, getMaxValueFromShift },
+    { 2,                UPROPS_DT_MASK, 0,                  defaultGetValue, defaultGetMaxValue },
+    { 0,                UPROPS_EA_MASK, UPROPS_EA_SHIFT,    defaultGetValue, defaultGetMaxValue },
+    { UPROPS_SRC_CHAR,  0, (int32_t)U_CHAR_CATEGORY_COUNT-1,getGeneralCategory, getMaxValueFromShift },
+    { UPROPS_SRC_BIDI,  0, 0,                               getJoiningGroup, biDiGetMaxValue },
+    { UPROPS_SRC_BIDI,  0, 0,                               getJoiningType, biDiGetMaxValue },
+    { 2,                UPROPS_LB_MASK, UPROPS_LB_SHIFT,    defaultGetValue, defaultGetMaxValue },
+    { UPROPS_SRC_CHAR,  0, (int32_t)U_NT_COUNT-1,           getNumericType, getMaxValueFromShift },
+    { UPROPS_SRC_PROPSVEC, 0, 0,                            getScript, scriptGetMaxValue },
+    { UPROPS_SRC_PROPSVEC, 0, (int32_t)U_HST_COUNT-1,       getHangulSyllableType, getMaxValueFromShift },
+    // UCHAR_NFD_QUICK_CHECK: max=1=YES -- never "maybe", only "no" or "yes"
+    { UPROPS_SRC_NFC,   0, (int32_t)UNORM_YES,              getNormQuickCheck, getMaxValueFromShift },
+    // UCHAR_NFKD_QUICK_CHECK: max=1=YES -- never "maybe", only "no" or "yes"
+    { UPROPS_SRC_NFKC,  0, (int32_t)UNORM_YES,              getNormQuickCheck, getMaxValueFromShift },
+    // UCHAR_NFC_QUICK_CHECK: max=2=MAYBE
+    { UPROPS_SRC_NFC,   0, (int32_t)UNORM_MAYBE,            getNormQuickCheck, getMaxValueFromShift },
+    // UCHAR_NFKC_QUICK_CHECK: max=2=MAYBE
+    { UPROPS_SRC_NFKC,  0, (int32_t)UNORM_MAYBE,            getNormQuickCheck, getMaxValueFromShift },
+    { UPROPS_SRC_NFC,   0, 0xff,                            getLeadCombiningClass, getMaxValueFromShift },
+    { UPROPS_SRC_NFC,   0, 0xff,                            getTrailCombiningClass, getMaxValueFromShift },
+    { 2,                UPROPS_GCB_MASK, UPROPS_GCB_SHIFT,  defaultGetValue, defaultGetMaxValue },
+    { 2,                UPROPS_SB_MASK, UPROPS_SB_SHIFT,    defaultGetValue, defaultGetMaxValue },
+    { 2,                UPROPS_WB_MASK, UPROPS_WB_SHIFT,    defaultGetValue, defaultGetMaxValue },
+    { UPROPS_SRC_BIDI,  0, 0,                               getBiDiPairedBracketType, biDiGetMaxValue },
+    { UPROPS_SRC_INPC,  0, 0,                               getInPC, layoutGetMaxValue },
+    { UPROPS_SRC_INSC,  0, 0,                               getInSC, layoutGetMaxValue },
+    { UPROPS_SRC_VO,    0, 0,                               getVo, layoutGetMaxValue },
+};
+
+U_CAPI int32_t U_EXPORT2
+u_getIntPropertyValue(UChar32 c, UProperty which) {
+    if(which= 0) {
+        sa->add(sa->set, start);
+        start = end + 1;
+    }
+}
+
+#if !UCONFIG_NO_NORMALIZATION
+
+U_CAPI int32_t U_EXPORT2
+u_getFC_NFKC_Closure(UChar32 c, char16_t *dest, int32_t destCapacity, UErrorCode *pErrorCode) {
+    if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+    if(destCapacity<0 || (dest==nullptr && destCapacity>0)) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    // Compute the FC_NFKC_Closure on the fly:
+    // We have the API for complete coverage of Unicode properties, although
+    // this value by itself is not useful via API.
+    // (What could be useful is a custom normalization table that combines
+    // case folding and NFKC.)
+    // For the derivation, see Unicode's DerivedNormalizationProps.txt.
+    const Normalizer2 *nfkc=Normalizer2::getNFKCInstance(*pErrorCode);
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+    // first: b = NFKC(Fold(a))
+    UnicodeString folded1String;
+    const char16_t *folded1;
+    int32_t folded1Length=ucase_toFullFolding(c, &folded1, U_FOLD_CASE_DEFAULT);
+    if(folded1Length<0) {
+        const Normalizer2Impl *nfkcImpl=Normalizer2Factory::getImpl(nfkc);
+        if(nfkcImpl->getCompQuickCheck(nfkcImpl->getNorm16(c))!=UNORM_NO) {
+            return u_terminateUChars(dest, destCapacity, 0, pErrorCode);  // c does not change at all under CaseFolding+NFKC
+        }
+        folded1String.setTo(c);
+    } else {
+        if(folded1Length>UCASE_MAX_STRING_LENGTH) {
+            folded1String.setTo(folded1Length);
+        } else {
+            folded1String.setTo(false, folded1, folded1Length);
+        }
+    }
+    UnicodeString kc1=nfkc->normalize(folded1String, *pErrorCode);
+    // second: c = NFKC(Fold(b))
+    UnicodeString folded2String(kc1);
+    UnicodeString kc2=nfkc->normalize(folded2String.foldCase(), *pErrorCode);
+    // if (c != b) add the mapping from a to c
+    if(U_FAILURE(*pErrorCode) || kc1==kc2) {
+        return u_terminateUChars(dest, destCapacity, 0, pErrorCode);
+    } else {
+        return kc2.extract(dest, destCapacity, *pErrorCode);
+    }
+}
+
+#endif
diff --git a/intl/icu/source/common/uprops.h b/intl/icu/source/common/uprops.h
new file mode 100644
index 0000000000..1e06d03519
--- /dev/null
+++ b/intl/icu/source/common/uprops.h
@@ -0,0 +1,455 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2002-2016, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  uprops.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2002feb24
+*   created by: Markus W. Scherer
+*
+*   Constants for mostly non-core Unicode character properties
+*   stored in uprops.icu.
+*/
+
+#ifndef __UPROPS_H__
+#define __UPROPS_H__
+
+#include "unicode/utypes.h"
+#include "unicode/uset.h"
+#include "uset_imp.h"
+#include "udataswp.h"
+
+/* indexes[] entries */
+enum {
+    UPROPS_PROPS32_INDEX,
+    UPROPS_EXCEPTIONS_INDEX,
+    UPROPS_EXCEPTIONS_TOP_INDEX,
+
+    UPROPS_ADDITIONAL_TRIE_INDEX,
+    UPROPS_ADDITIONAL_VECTORS_INDEX,
+    UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX,
+
+    UPROPS_SCRIPT_EXTENSIONS_INDEX,
+
+    UPROPS_RESERVED_INDEX_7,
+    UPROPS_RESERVED_INDEX_8,
+
+    /* size of the data file (number of 32-bit units after the header) */
+    UPROPS_DATA_TOP_INDEX,
+
+    /* maximum values for code values in vector word 0 */
+    UPROPS_MAX_VALUES_INDEX=10,
+    /* maximum values for code values in vector word 2 */
+    UPROPS_MAX_VALUES_2_INDEX,
+
+    UPROPS_INDEX_COUNT=16
+};
+
+/* definitions for the main properties words */
+enum {
+    /* general category shift==0                                0 (5 bits) */
+    /* reserved                                                 5 (1 bit) */
+    UPROPS_NUMERIC_TYPE_VALUE_SHIFT=6                       /*  6 (10 bits) */
+};
+
+#define GET_CATEGORY(props) ((props)&0x1f)
+#define CAT_MASK(props) U_MASK(GET_CATEGORY(props))
+
+#define GET_NUMERIC_TYPE_VALUE(props) ((props)>>UPROPS_NUMERIC_TYPE_VALUE_SHIFT)
+
+/* constants for the storage form of numeric types and values */
+enum {
+    /** No numeric value. */
+    UPROPS_NTV_NONE=0,
+    /** Decimal digits: nv=0..9 */
+    UPROPS_NTV_DECIMAL_START=1,
+    /** Other digits: nv=0..9 */
+    UPROPS_NTV_DIGIT_START=11,
+    /** Small integers: nv=0..154 */
+    UPROPS_NTV_NUMERIC_START=21,
+    /** Fractions: ((ntv>>4)-12) / ((ntv&0xf)+1) = -1..17 / 1..16 */
+    UPROPS_NTV_FRACTION_START=0xb0,
+    /**
+     * Large integers:
+     * ((ntv>>5)-14) * 10^((ntv&0x1f)+2) = (1..9)*(10^2..10^33)
+     * (only one significant decimal digit)
+     */
+    UPROPS_NTV_LARGE_START=0x1e0,
+    /**
+     * Sexagesimal numbers:
+     * ((ntv>>2)-0xbf) * 60^((ntv&3)+1) = (1..9)*(60^1..60^4)
+     */
+    UPROPS_NTV_BASE60_START=0x300,
+    /**
+     * Fraction-20 values:
+     * frac20 = ntv-0x324 = 0..0x17 -> 1|3|5|7 / 20|40|80|160|320|640
+     * numerator: num = 2*(frac20&3)+1
+     * denominator: den = 20<<(frac20>>2)
+     */
+    UPROPS_NTV_FRACTION20_START=UPROPS_NTV_BASE60_START+36,  // 0x300+9*4=0x324
+    /**
+     * Fraction-32 values:
+     * frac32 = ntv-0x34c = 0..15 -> 1|3|5|7 / 32|64|128|256
+     * numerator: num = 2*(frac32&3)+1
+     * denominator: den = 32<<(frac32>>2)
+     */
+    UPROPS_NTV_FRACTION32_START=UPROPS_NTV_FRACTION20_START+24,  // 0x324+6*4=0x34c
+    /** No numeric value (yet). */
+    UPROPS_NTV_RESERVED_START=UPROPS_NTV_FRACTION32_START+16,  // 0x34c+4*4=0x35c
+
+    UPROPS_NTV_MAX_SMALL_INT=UPROPS_NTV_FRACTION_START-UPROPS_NTV_NUMERIC_START-1
+};
+
+#define UPROPS_NTV_GET_TYPE(ntv) \
+    ((ntv==UPROPS_NTV_NONE) ? U_NT_NONE : \
+    (ntv> UPROPS_SCRIPT_HIGH_SHIFT) |
+        (scriptX & UPROPS_SCRIPT_LOW_MASK);
+}
+
+}  // namespace
+
+#endif  // __cplusplus
+
+/*
+ * Properties in vector word 1
+ * Each bit encodes one binary property.
+ * The following constants represent the bit number, use 1<=(int32_t)sizeof(pathBuffer)) {
+            *status=U_ILLEGAL_ARGUMENT_ERROR;
+            return nullptr;
+        } else if(uprv_isInvariantUString(myPath, length)) {
+            /*
+             * the invariant converter is sufficient for package and tree names
+             * and is more efficient
+             */
+            u_UCharsToChars(myPath, path, length+1); /* length+1 to include the NUL */
+        } else {
+#if !UCONFIG_NO_CONVERSION
+            /* use the default converter to support variant-character paths */
+            UConverter *cnv=u_getDefaultConverter(status);
+            length=ucnv_fromUChars(cnv, path, (int32_t)sizeof(pathBuffer), myPath, length, status);
+            u_releaseDefaultConverter(cnv);
+            if(U_FAILURE(*status)) {
+                return nullptr;
+            }
+            if(length>=(int32_t)sizeof(pathBuffer)) {
+                /* not NUL-terminated - path too long */
+                *status=U_ILLEGAL_ARGUMENT_ERROR;
+                return nullptr;
+            }
+#else
+            /* the default converter is not available */
+            *status=U_UNSUPPORTED_ERROR;
+            return nullptr;
+#endif
+        }
+    }
+
+    return ures_open(path, localeID, status);
+}
diff --git a/intl/icu/source/common/uresbund.cpp b/intl/icu/source/common/uresbund.cpp
new file mode 100644
index 0000000000..d02bba8921
--- /dev/null
+++ b/intl/icu/source/common/uresbund.cpp
@@ -0,0 +1,3462 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+* Copyright (C) 1997-2016, International Business Machines Corporation and
+* others. All Rights Reserved.
+******************************************************************************
+*
+* File uresbund.cpp
+*
+* Modification History:
+*
+*   Date        Name        Description
+*   04/01/97    aliu        Creation.
+*   06/14/99    stephen     Removed functions taking a filename suffix.
+*   07/20/99    stephen     Changed for UResourceBundle typedef'd to void*
+*   11/09/99    weiv            Added ures_getLocale()
+*   March 2000  weiv        Total overhaul - using data in DLLs
+*   06/20/2000  helena      OS/400 port changes; mostly typecast.
+*   06/24/02    weiv        Added support for resource sharing
+******************************************************************************
+*/
+
+#include "unicode/ures.h"
+#include "unicode/ustring.h"
+#include "unicode/ucnv.h"
+#include "charstr.h"
+#include "uresimp.h"
+#include "ustr_imp.h"
+#include "cwchar.h"
+#include "ucln_cmn.h"
+#include "cmemory.h"
+#include "cstring.h"
+#include "mutex.h"
+#include "uhash.h"
+#include "unicode/uenum.h"
+#include "uenumimp.h"
+#include "ulocimp.h"
+#include "umutex.h"
+#include "putilimp.h"
+#include "uassert.h"
+#include "uresdata.h"
+
+using namespace icu;
+
+/*
+Static cache for already opened resource bundles - mostly for keeping fallback info
+TODO: This cache should probably be removed when the deprecated code is
+      completely removed.
+*/
+static UHashtable *cache = nullptr;
+static icu::UInitOnce gCacheInitOnce {};
+
+static UMutex resbMutex;
+
+/* INTERNAL: hashes an entry  */
+static int32_t U_CALLCONV hashEntry(const UHashTok parm) {
+    UResourceDataEntry *b = (UResourceDataEntry *)parm.pointer;
+    UHashTok namekey, pathkey;
+    namekey.pointer = b->fName;
+    pathkey.pointer = b->fPath;
+    return uhash_hashChars(namekey)+37u*uhash_hashChars(pathkey);
+}
+
+/* INTERNAL: compares two entries */
+static UBool U_CALLCONV compareEntries(const UHashTok p1, const UHashTok p2) {
+    UResourceDataEntry *b1 = (UResourceDataEntry *)p1.pointer;
+    UResourceDataEntry *b2 = (UResourceDataEntry *)p2.pointer;
+    UHashTok name1, name2, path1, path2;
+    name1.pointer = b1->fName;
+    name2.pointer = b2->fName;
+    path1.pointer = b1->fPath;
+    path2.pointer = b2->fPath;
+    return (UBool)(uhash_compareChars(name1, name2) &&
+        uhash_compareChars(path1, path2));
+}
+
+
+/**
+ *  Internal function, gets parts of locale name according 
+ *  to the position of '_' character
+ */
+static UBool chopLocale(char *name) {
+    char *i = uprv_strrchr(name, '_');
+
+    if(i != nullptr) {
+        *i = '\0';
+        return true;
+    }
+
+    return false;
+}
+
+static UBool hasVariant(const char* localeID) {
+    UErrorCode err = U_ZERO_ERROR;
+    int32_t variantLength = uloc_getVariant(localeID, nullptr, 0, &err);
+    return variantLength != 0;
+}
+
+// This file contains the tables for doing locale fallback, which are generated
+// by the CLDR-to-ICU process directly from the CLDR data.  This file should only
+// ever be included from here.
+#define INCLUDED_FROM_URESBUND_CPP
+#include "localefallback_data.h"
+
+static const char* performFallbackLookup(const char* key,
+                                         const char* keyStrs,
+                                         const char* valueStrs,
+                                         const int32_t* lookupTable,
+                                         int32_t lookupTableLength) {
+    const int32_t* bottom = lookupTable;
+    const int32_t* top = lookupTable + lookupTableLength;
+
+    while (bottom < top) {
+        // Effectively, divide by 2 and round down to an even index
+        const int32_t* middle = bottom + (((top - bottom) / 4) * 2);
+        const char* entryKey = &(keyStrs[*middle]);
+        int32_t strcmpResult = uprv_strcmp(key, entryKey);
+        if (strcmpResult == 0) {
+            return &(valueStrs[middle[1]]);
+        } else if (strcmpResult < 0) {
+            top = middle;
+        } else {
+            bottom = middle + 2;
+        }
+    }
+    return nullptr;
+}
+
+static CharString getDefaultScript(const CharString& language, const CharString& region) {
+    const char* defaultScript = nullptr;
+    UErrorCode err = U_ZERO_ERROR;
+    
+    // the default script will be "Latn" if we don't find the locale ID in the tables
+    CharString result("Latn", err);
+    
+    // if we were passed both language and region, make them into a locale ID and look that up in the default
+    // script table
+    if (!region.isEmpty()) {
+        CharString localeID;
+        localeID.append(language, err).append("_", err).append(region, err);
+        if (U_FAILURE(err)) {
+            return result;
+        }
+        defaultScript = performFallbackLookup(localeID.data(), dsLocaleIDChars, scriptCodeChars, defaultScriptTable, UPRV_LENGTHOF(defaultScriptTable));
+    }
+    
+    // if we didn't find anything, look up just the language in the default script table
+    if (defaultScript == nullptr) {
+        defaultScript = performFallbackLookup(language.data(), dsLocaleIDChars, scriptCodeChars, defaultScriptTable, UPRV_LENGTHOF(defaultScriptTable));
+    }
+    
+    // if either lookup above succeeded, copy the result from "defaultScript" into "result"; otherwise, return "Latn"
+    if (defaultScript != nullptr) {
+        result.clear();
+        result.append(defaultScript, err);
+    }
+    return result;
+}
+
+enum UResOpenType {
+    /**
+     * Open a resource bundle for the locale;
+     * if there is not even a base language bundle, then fall back to the default locale;
+     * if there is no bundle for that either, then load the root bundle.
+     *
+     * This is the default bundle loading behavior.
+     */
+    URES_OPEN_LOCALE_DEFAULT_ROOT,
+    // TODO: ICU ticket #11271 "consistent default locale across locale trees"
+    // Add an option to look at the main locale tree for whether to
+    // fall back to root directly (if the locale has main data) or
+    // fall back to the default locale first (if the locale does not even have main data).
+    /**
+     * Open a resource bundle for the locale;
+     * if there is not even a base language bundle, then load the root bundle;
+     * never fall back to the default locale.
+     *
+     * This is used for algorithms that have good pan-Unicode default behavior,
+     * such as case mappings, collation, and segmentation (BreakIterator).
+     */
+    URES_OPEN_LOCALE_ROOT,
+    /**
+     * Open a resource bundle for the exact bundle name as requested;
+     * no fallbacks, do not load parent bundles.
+     *
+     * This is used for supplemental (non-locale) data.
+     */
+    URES_OPEN_DIRECT
+};
+typedef enum UResOpenType UResOpenType;
+
+/**
+ *  Internal function, determines the search path for resource bundle files.
+ *  Currently, this function is used only by findFirstExisting() to help search for resource bundle files when a bundle for the specified
+ *  locale doesn't exist.  The code that supports inheritance of resources between existing resource bundle files continues to
+ *  use chopLocale() below.
+ *  @param name In-out parameter: On input, the locale ID to get a parent locale ID for (this is a locale's base name, without keywords); on output, the
+ *  requested parent locale ID.
+ *  @param origName The original locale ID the caller of findFirstExisting() requested.  This is the same as `name` on the first call to this function,
+ *  but as findFirstExisting() ascends the resource bundle's parent tree, this parameter will continue to be the original locale ID requested.
+ */
+static bool getParentLocaleID(char *name, const char *origName, UResOpenType openType) {
+    // early out if the locale ID has a variant code or ends with _
+    size_t nameLen = uprv_strlen(name);
+    if (!nameLen || name[nameLen - 1] == '_' || hasVariant(name)) {
+        return chopLocale(name);
+    }
+    
+    UErrorCode err = U_ZERO_ERROR;
+    const char* tempNamePtr = name;
+    CharString language = ulocimp_getLanguage(tempNamePtr, &tempNamePtr, err);
+    if (*tempNamePtr == '_') {
+        ++tempNamePtr;
+    }
+    CharString script = ulocimp_getScript(tempNamePtr, &tempNamePtr, err);
+    if (*tempNamePtr == '_') {
+        ++tempNamePtr;
+    }
+    CharString region = ulocimp_getCountry(tempNamePtr, &tempNamePtr, err);
+    CharString workingLocale;
+    if (U_FAILURE(err)) {
+        // hopefully this never happens...
+        return chopLocale(name);
+    }
+    
+    // if the open type is URES_OPEN_LOCALE_DEFAULT_ROOT, first look the locale ID up in the parent locale table;
+    // if that table specifies a parent for it, return that  (we don't do this for the other open types-- if we're not
+    // falling back through the system default locale, we also want to do straight truncation fallback instead
+    // of looking things up in the parent locale table-- see https://www.unicode.org/reports/tr35/tr35.html#Parent_Locales:
+    // "Collation data, however, is an exception...")
+    if (openType == URES_OPEN_LOCALE_DEFAULT_ROOT) {
+        const char* parentID = performFallbackLookup(name, parentLocaleChars, parentLocaleChars, parentLocaleTable, UPRV_LENGTHOF(parentLocaleTable));
+        if (parentID != nullptr) {
+            uprv_strcpy(name, parentID);
+            return true;
+        }
+    }
+
+    // if it's not in the parent locale table, figure out the fallback script algorithmically
+    // (see CLDR-15265 for an explanation of the algorithm)
+    if (!script.isEmpty() && !region.isEmpty()) {
+        // if "name" has both script and region, is the script the default script?
+        // - if so, remove it and keep the region
+        // - if not, remove the region and keep the script
+        if (getDefaultScript(language, region) == script.toStringPiece()) {
+            workingLocale.append(language, err).append("_", err).append(region, err);
+        } else {
+            workingLocale.append(language, err).append("_", err).append(script, err);
+        }
+    } else if (!region.isEmpty()) {
+        // if "name" has region but not script, did the original locale ID specify a script?
+        // - if yes, replace the region with the script from the original locale ID
+        // - if no, replace the region with the default script for that language and region
+        UErrorCode err = U_ZERO_ERROR;
+        tempNamePtr = origName;
+        CharString origNameLanguage = ulocimp_getLanguage(tempNamePtr, &tempNamePtr, err);
+        if (*tempNamePtr == '_') {
+            ++tempNamePtr;
+        }
+        CharString origNameScript = ulocimp_getScript(origName, nullptr, err);
+        if (!origNameScript.isEmpty()) {
+            workingLocale.append(language, err).append("_", err).append(origNameScript, err);
+        } else {
+            workingLocale.append(language, err).append("_", err).append(getDefaultScript(language, region), err);
+        }
+    } else if (!script.isEmpty()) {
+        // if "name" has script but not region (and our open type if URES_OPEN_LOCALE_DEFAULT_ROOT), is the script
+        // the default script for the language?
+        // - if so, remove it from the locale ID
+        // - if not, return false to continue up the chain
+        // (we don't do this for other open types for the same reason we don't look things up in the parent
+        // locale table for other open types-- see the reference to UTS #35 above)
+        if (openType != URES_OPEN_LOCALE_DEFAULT_ROOT || getDefaultScript(language, CharString()) == script.toStringPiece()) {
+            workingLocale.append(language, err);
+        } else {
+            return false;
+        }
+    } else {
+        // if "name" just contains a language code, return false so the calling code falls back to "root"
+        return false;
+    }
+    if (U_SUCCESS(err) && !workingLocale.isEmpty()) {
+        uprv_strcpy(name, workingLocale.data());
+        return true;
+    } else {
+        return false;
+    }
+}
+
+/**
+ *  Called to check whether a name without '_' needs to be checked for a parent.
+ *  Some code had assumed that locale IDs with '_' could not have a non-root parent.
+ *  We may want a better way of doing this.
+ */
+static UBool mayHaveParent(char *name) {
+    return (name[0] != 0 && uprv_strstr("nb nn",name) != nullptr);
+}
+
+/**
+ *  Internal function
+ */
+static void entryIncrease(UResourceDataEntry *entry) {
+    Mutex lock(&resbMutex);
+    entry->fCountExisting++;
+    while(entry->fParent != nullptr) {
+      entry = entry->fParent;
+      entry->fCountExisting++;
+    }
+}
+
+/**
+ *  Internal function. Tries to find a resource in given Resource
+ *  Bundle, as well as in its parents
+ */
+static UResourceDataEntry *getFallbackData(
+        const UResourceBundle *resBundle,
+        const char **resTag, Resource *res, UErrorCode *status) {
+    UResourceDataEntry *dataEntry = resBundle->fData;
+    int32_t indexR = -1;
+    int32_t i = 0;
+    *res = RES_BOGUS;
+    if(dataEntry == nullptr) {
+        *status = U_MISSING_RESOURCE_ERROR;
+        return nullptr;
+    }
+    if(dataEntry->fBogus == U_ZERO_ERROR) { /* if this resource is real, */
+        *res = res_getTableItemByKey(&(dataEntry->fData), dataEntry->fData.rootRes, &indexR, resTag); /* try to get data from there */
+        i++;
+    }
+    if(resBundle->fHasFallback) {
+        // Otherwise, we'll look in parents.
+        while(*res == RES_BOGUS && dataEntry->fParent != nullptr) {
+            dataEntry = dataEntry->fParent;
+            if(dataEntry->fBogus == U_ZERO_ERROR) {
+                i++;
+                *res = res_getTableItemByKey(&(dataEntry->fData), dataEntry->fData.rootRes, &indexR, resTag);
+            }
+        }
+    }
+
+    if(*res == RES_BOGUS) {
+        // If the resource is not found, we need to give an error.
+        *status = U_MISSING_RESOURCE_ERROR;
+        return nullptr;
+    }
+    // If the resource is found in parents, we need to adjust the error.
+    if(i>1) {
+        if(uprv_strcmp(dataEntry->fName, uloc_getDefault())==0 || uprv_strcmp(dataEntry->fName, kRootLocaleName)==0) {
+            *status = U_USING_DEFAULT_WARNING;
+        } else {
+            *status = U_USING_FALLBACK_WARNING;
+        }
+    }
+    return dataEntry;
+}
+
+static void
+free_entry(UResourceDataEntry *entry) {
+    UResourceDataEntry *alias;
+    res_unload(&(entry->fData));
+    if(entry->fName != nullptr && entry->fName != entry->fNameBuffer) {
+        uprv_free(entry->fName);
+    }
+    if(entry->fPath != nullptr) {
+        uprv_free(entry->fPath);
+    }
+    if(entry->fPool != nullptr) {
+        --entry->fPool->fCountExisting;
+    }
+    alias = entry->fAlias;
+    if(alias != nullptr) {
+        while(alias->fAlias != nullptr) {
+            alias = alias->fAlias;
+        }
+        --alias->fCountExisting;
+    }
+    uprv_free(entry);
+}
+
+/* Works just like ucnv_flushCache() */
+static int32_t ures_flushCache()
+{
+    UResourceDataEntry *resB;
+    int32_t pos;
+    int32_t rbDeletedNum = 0;
+    const UHashElement *e;
+    UBool deletedMore;
+
+    /*if shared data hasn't even been lazy evaluated yet
+    * return 0
+    */
+    Mutex lock(&resbMutex);
+    if (cache == nullptr) {
+        return 0;
+    }
+
+    do {
+        deletedMore = false;
+        /*creates an enumeration to iterate through every element in the table */
+        pos = UHASH_FIRST;
+        while ((e = uhash_nextElement(cache, &pos)) != nullptr)
+        {
+            resB = (UResourceDataEntry *) e->value.pointer;
+            /* Deletes only if reference counter == 0
+             * Don't worry about the children of this node.
+             * Those will eventually get deleted too, if not already.
+             * Don't worry about the parents of this node.
+             * Those will eventually get deleted too, if not already.
+             */
+            /* 04/05/2002 [weiv] fCountExisting should now be accurate. If it's not zero, that means that    */
+            /* some resource bundles are still open somewhere. */
+
+            if (resB->fCountExisting == 0) {
+                rbDeletedNum++;
+                deletedMore = true;
+                uhash_removeElement(cache, e);
+                free_entry(resB);
+            }
+        }
+        /*
+         * Do it again to catch bundles (aliases, pool bundle) whose fCountExisting
+         * got decremented by free_entry().
+         */
+    } while(deletedMore);
+
+    return rbDeletedNum;
+}
+
+#ifdef URES_DEBUG
+#include 
+
+U_CAPI UBool U_EXPORT2 ures_dumpCacheContents() {
+  UBool cacheNotEmpty = false;
+  int32_t pos = UHASH_FIRST;
+  const UHashElement *e;
+  UResourceDataEntry *resB;
+  
+    Mutex lock(&resbMutex);
+    if (cache == nullptr) {
+      fprintf(stderr,"%s:%d: RB Cache is nullptr.\n", __FILE__, __LINE__);
+      return false;
+    }
+
+    while ((e = uhash_nextElement(cache, &pos)) != nullptr) {
+      cacheNotEmpty=true;
+      resB = (UResourceDataEntry *) e->value.pointer;
+      fprintf(stderr,"%s:%d: RB Cache: Entry @0x%p, refcount %d, name %s:%s.  Pool 0x%p, alias 0x%p, parent 0x%p\n",
+              __FILE__, __LINE__,
+              (void*)resB, resB->fCountExisting,
+              resB->fName?resB->fName:"nullptr",
+              resB->fPath?resB->fPath:"nullptr",
+              (void*)resB->fPool,
+              (void*)resB->fAlias,
+              (void*)resB->fParent);       
+    }
+    
+    fprintf(stderr,"%s:%d: RB Cache still contains %d items.\n", __FILE__, __LINE__, uhash_count(cache));
+    return cacheNotEmpty;
+}
+
+#endif
+
+static UBool U_CALLCONV ures_cleanup()
+{
+    if (cache != nullptr) {
+        ures_flushCache();
+        uhash_close(cache);
+        cache = nullptr;
+    }
+    gCacheInitOnce.reset();
+    return true;
+}
+
+/** INTERNAL: Initializes the cache for resources */
+static void U_CALLCONV createCache(UErrorCode &status) {
+    U_ASSERT(cache == nullptr);
+    cache = uhash_open(hashEntry, compareEntries, nullptr, &status);
+    ucln_common_registerCleanup(UCLN_COMMON_URES, ures_cleanup);
+}
+     
+static void initCache(UErrorCode *status) {
+    umtx_initOnce(gCacheInitOnce, &createCache, *status);
+}
+
+/** INTERNAL: sets the name (locale) of the resource bundle to given name */
+
+static void setEntryName(UResourceDataEntry *res, const char *name, UErrorCode *status) {
+    int32_t len = (int32_t)uprv_strlen(name);
+    if(res->fName != nullptr && res->fName != res->fNameBuffer) {
+        uprv_free(res->fName);
+    }
+    if (len < (int32_t)sizeof(res->fNameBuffer)) {
+        res->fName = res->fNameBuffer;
+    }
+    else {
+        res->fName = (char *)uprv_malloc(len+1);
+    }
+    if(res->fName == nullptr) {
+        *status = U_MEMORY_ALLOCATION_ERROR;
+    } else {
+        uprv_strcpy(res->fName, name);
+    }
+}
+
+static UResourceDataEntry *
+getPoolEntry(const char *path, UErrorCode *status);
+
+/**
+ *  INTERNAL: Inits and opens an entry from a data DLL.
+ *    CAUTION:  resbMutex must be locked when calling this function.
+ */
+static UResourceDataEntry *init_entry(const char *localeID, const char *path, UErrorCode *status) {
+    UResourceDataEntry *r = nullptr;
+    UResourceDataEntry find;
+    /*int32_t hashValue;*/
+    const char *name;
+    char aliasName[100] = { 0 };
+    int32_t aliasLen = 0;
+    /*UBool isAlias = false;*/
+    /*UHashTok hashkey; */
+
+    if(U_FAILURE(*status)) {
+        return nullptr;
+    }
+
+    /* here we try to deduce the right locale name */
+    if(localeID == nullptr) { /* if localeID is nullptr, we're trying to open default locale */
+        name = uloc_getDefault();
+    } else if(*localeID == 0) { /* if localeID is "" then we try to open root locale */
+        name = kRootLocaleName;
+    } else { /* otherwise, we'll open what we're given */
+        name = localeID;
+    }
+
+    find.fName = (char *)name;
+    find.fPath = (char *)path;
+
+    /* calculate the hash value of the entry */
+    /*hashkey.pointer = (void *)&find;*/
+    /*hashValue = hashEntry(hashkey);*/
+
+    /* check to see if we already have this entry */
+    r = (UResourceDataEntry *)uhash_get(cache, &find);
+    if(r == nullptr) {
+        /* if the entry is not yet in the hash table, we'll try to construct a new one */
+        r = (UResourceDataEntry *) uprv_malloc(sizeof(UResourceDataEntry));
+        if(r == nullptr) {
+            *status = U_MEMORY_ALLOCATION_ERROR;
+            return nullptr;
+        }
+
+        uprv_memset(r, 0, sizeof(UResourceDataEntry));
+        /*r->fHashKey = hashValue;*/
+
+        setEntryName(r, name, status);
+        if (U_FAILURE(*status)) {
+            uprv_free(r);
+            return nullptr;
+        }
+
+        if(path != nullptr) {
+            r->fPath = (char *)uprv_strdup(path);
+            if(r->fPath == nullptr) {
+                *status = U_MEMORY_ALLOCATION_ERROR;
+                uprv_free(r);
+                return nullptr;
+            }
+        }
+
+        /* this is the actual loading */
+        res_load(&(r->fData), r->fPath, r->fName, status);
+
+        if (U_FAILURE(*status)) {
+            /* if we failed to load due to an out-of-memory error, exit early. */
+            if (*status == U_MEMORY_ALLOCATION_ERROR) {
+                uprv_free(r);
+                return nullptr;
+            }
+            /* we have no such entry in dll, so it will always use fallback */
+            *status = U_USING_FALLBACK_WARNING;
+            r->fBogus = U_USING_FALLBACK_WARNING;
+        } else { /* if we have a regular entry */
+            Resource aliasres;
+            if (r->fData.usesPoolBundle) {
+                r->fPool = getPoolEntry(r->fPath, status);
+                if (U_SUCCESS(*status)) {
+                    const int32_t *poolIndexes = r->fPool->fData.pRoot + 1;
+                    if(r->fData.pRoot[1 + URES_INDEX_POOL_CHECKSUM] == poolIndexes[URES_INDEX_POOL_CHECKSUM]) {
+                        r->fData.poolBundleKeys = (const char *)(poolIndexes + (poolIndexes[URES_INDEX_LENGTH] & 0xff));
+                        r->fData.poolBundleStrings = r->fPool->fData.p16BitUnits;
+                    } else {
+                        r->fBogus = *status = U_INVALID_FORMAT_ERROR;
+                    }
+                } else {
+                    r->fBogus = *status;
+                }
+            }
+            if (U_SUCCESS(*status)) {
+                /* handle the alias by trying to get out the %%Alias tag.*/
+                /* We'll try to get alias string from the bundle */
+                aliasres = res_getResource(&(r->fData), "%%ALIAS");
+                if (aliasres != RES_BOGUS) {
+                    // No tracing: called during initial data loading
+                    const char16_t *alias = res_getStringNoTrace(&(r->fData), aliasres, &aliasLen);
+                    if(alias != nullptr && aliasLen > 0) { /* if there is actual alias - unload and load new data */
+                        u_UCharsToChars(alias, aliasName, aliasLen+1);
+                        r->fAlias = init_entry(aliasName, path, status);
+                    }
+                }
+            }
+        }
+
+        {
+            UResourceDataEntry *oldR = nullptr;
+            if((oldR = (UResourceDataEntry *)uhash_get(cache, r)) == nullptr) { /* if the data is not cached */
+                /* just insert it in the cache */
+                UErrorCode cacheStatus = U_ZERO_ERROR;
+                uhash_put(cache, (void *)r, r, &cacheStatus);
+                if (U_FAILURE(cacheStatus)) {
+                    *status = cacheStatus;
+                    free_entry(r);
+                    r = nullptr;
+                }
+            } else {
+                /* somebody have already inserted it while we were working, discard newly opened data */
+                /* Also, we could get here IF we opened an alias */
+                free_entry(r);
+                r = oldR;
+            }
+        }
+
+    }
+    if(r != nullptr) {
+        /* return the real bundle */
+        while(r->fAlias != nullptr) {
+            r = r->fAlias;
+        }
+        r->fCountExisting++; /* we increase its reference count */
+        /* if the resource has a warning */
+        /* we don't want to overwrite a status with no error */
+        if(r->fBogus != U_ZERO_ERROR && U_SUCCESS(*status)) {
+             *status = r->fBogus; /* set the returning status */
+        }
+    }
+    return r;
+}
+
+static UResourceDataEntry *
+getPoolEntry(const char *path, UErrorCode *status) {
+    UResourceDataEntry *poolBundle = init_entry(kPoolBundleName, path, status);
+    if( U_SUCCESS(*status) &&
+        (poolBundle == nullptr || poolBundle->fBogus != U_ZERO_ERROR || !poolBundle->fData.isPoolBundle)
+    ) {
+        *status = U_INVALID_FORMAT_ERROR;
+    }
+    return poolBundle;
+}
+
+/* INTERNAL: */
+/*   CAUTION:  resbMutex must be locked when calling this function! */
+static UResourceDataEntry *
+findFirstExisting(const char* path, char* name, const char* defaultLocale, UResOpenType openType,
+                  UBool *isRoot, UBool *foundParent, UBool *isDefault, UErrorCode* status) {
+    UResourceDataEntry *r = nullptr;
+    UBool hasRealData = false;
+    *foundParent = true; /* we're starting with a fresh name */
+    char origName[ULOC_FULLNAME_CAPACITY];
+
+    uprv_strcpy(origName, name);
+    while(*foundParent && !hasRealData) {
+        r = init_entry(name, path, status);
+        /* Null pointer test */
+        if (U_FAILURE(*status)) {
+            return nullptr;
+        }
+        *isDefault = (UBool)(uprv_strncmp(name, defaultLocale, uprv_strlen(name)) == 0);
+        hasRealData = (UBool)(r->fBogus == U_ZERO_ERROR);
+        if(!hasRealData) {
+            /* this entry is not real. We will discard it. */
+            /* However, the parent line for this entry is  */
+            /* not to be used - as there might be parent   */
+            /* lines in cache from previous openings that  */
+            /* are not updated yet. */
+            r->fCountExisting--;
+            /*entryCloseInt(r);*/
+            r = nullptr;
+            *status = U_USING_FALLBACK_WARNING;
+        } else {
+            uprv_strcpy(name, r->fName); /* this is needed for supporting aliases */
+        }
+
+        *isRoot = (UBool)(uprv_strcmp(name, kRootLocaleName) == 0);
+
+        /*Fallback data stuff*/
+        if (!hasRealData) {
+            *foundParent = getParentLocaleID(name, origName, openType);
+        } else {
+            // we've already found a real resource file; what we return to the caller is the parent
+            // locale ID for inheritance, which should come from chopLocale(), not getParentLocaleID()
+            *foundParent = chopLocale(name);
+        }
+        if (*foundParent && *name == '\0') {
+            uprv_strcpy(name, "und");
+        }
+    }
+    return r;
+}
+
+static void ures_setIsStackObject( UResourceBundle* resB, UBool state) {
+    if(state) {
+        resB->fMagic1 = 0;
+        resB->fMagic2 = 0;
+    } else {
+        resB->fMagic1 = MAGIC1;
+        resB->fMagic2 = MAGIC2;
+    }
+}
+
+static UBool ures_isStackObject(const UResourceBundle* resB) {
+  return((resB->fMagic1 == MAGIC1 && resB->fMagic2 == MAGIC2)?false:true);
+}
+
+
+U_CFUNC void ures_initStackObject(UResourceBundle* resB) {
+  uprv_memset(resB, 0, sizeof(UResourceBundle));
+  ures_setIsStackObject(resB, true);
+}
+
+U_NAMESPACE_BEGIN
+
+StackUResourceBundle::StackUResourceBundle() {
+    ures_initStackObject(&bundle);
+}
+
+StackUResourceBundle::~StackUResourceBundle() {
+    ures_close(&bundle);
+}
+
+U_NAMESPACE_END
+
+static UBool  // returns U_SUCCESS(*status)
+loadParentsExceptRoot(UResourceDataEntry *&t1,
+                      char name[], int32_t nameCapacity,
+                      UBool usingUSRData, char usrDataPath[], UErrorCode *status) {
+    if (U_FAILURE(*status)) { return false; }
+    UBool checkParent = true;
+    while (checkParent && t1->fParent == nullptr && !t1->fData.noFallback &&
+            res_getResource(&t1->fData,"%%ParentIsRoot") == RES_BOGUS) {
+        Resource parentRes = res_getResource(&t1->fData, "%%Parent");
+        if (parentRes != RES_BOGUS) {  // An explicit parent was found.
+            int32_t parentLocaleLen = 0;
+            // No tracing: called during initial data loading
+            const char16_t *parentLocaleName = res_getStringNoTrace(&(t1->fData), parentRes, &parentLocaleLen);
+            if(parentLocaleName != nullptr && 0 < parentLocaleLen && parentLocaleLen < nameCapacity) {
+                u_UCharsToChars(parentLocaleName, name, parentLocaleLen + 1);
+                if (uprv_strcmp(name, kRootLocaleName) == 0) {
+                    return true;
+                }
+            }
+        }
+        // Insert regular parents.
+        UErrorCode parentStatus = U_ZERO_ERROR;
+        UResourceDataEntry *t2 = init_entry(name, t1->fPath, &parentStatus);
+        if (U_FAILURE(parentStatus)) {
+            *status = parentStatus;
+            return false;
+        }
+        UResourceDataEntry *u2 = nullptr;
+        UErrorCode usrStatus = U_ZERO_ERROR;
+        if (usingUSRData) {  // This code inserts user override data into the inheritance chain.
+            u2 = init_entry(name, usrDataPath, &usrStatus);
+            // If we failed due to out-of-memory, report that to the caller and exit early.
+            if (usrStatus == U_MEMORY_ALLOCATION_ERROR) {
+                *status = usrStatus;
+                return false;
+            }
+        }
+
+        if (usingUSRData && U_SUCCESS(usrStatus) && u2->fBogus == U_ZERO_ERROR) {
+            t1->fParent = u2;
+            u2->fParent = t2;
+        } else {
+            t1->fParent = t2;
+            if (usingUSRData) {
+                // The USR override data wasn't found, set it to be deleted.
+                u2->fCountExisting = 0;
+            }
+        }
+        t1 = t2;
+        checkParent = chopLocale(name) || mayHaveParent(name);
+    }
+    return true;
+}
+
+static UBool  // returns U_SUCCESS(*status)
+insertRootBundle(UResourceDataEntry *&t1, UErrorCode *status) {
+    if (U_FAILURE(*status)) { return false; }
+    UErrorCode parentStatus = U_ZERO_ERROR;
+    UResourceDataEntry *t2 = init_entry(kRootLocaleName, t1->fPath, &parentStatus);
+    if (U_FAILURE(parentStatus)) {
+        *status = parentStatus;
+        return false;
+    }
+    t1->fParent = t2;
+    t1 = t2;
+    return true;
+}
+
+static UResourceDataEntry *entryOpen(const char* path, const char* localeID,
+                                     UResOpenType openType, UErrorCode* status) {
+    U_ASSERT(openType != URES_OPEN_DIRECT);
+    UErrorCode intStatus = U_ZERO_ERROR;
+    UResourceDataEntry *r = nullptr;
+    UResourceDataEntry *t1 = nullptr;
+    UBool isDefault = false;
+    UBool isRoot = false;
+    UBool hasRealData = false;
+    UBool hasChopped = true;
+    UBool usingUSRData = U_USE_USRDATA && ( path == nullptr || uprv_strncmp(path,U_ICUDATA_NAME,8) == 0);
+
+    char name[ULOC_FULLNAME_CAPACITY];
+    char usrDataPath[96];
+
+    initCache(status);
+
+    if(U_FAILURE(*status)) {
+        return nullptr;
+    }
+
+    uprv_strncpy(name, localeID, sizeof(name) - 1);
+    name[sizeof(name) - 1] = 0;
+
+    if ( usingUSRData ) {
+        if ( path == nullptr ) {
+            uprv_strcpy(usrDataPath, U_USRDATA_NAME);
+        } else {
+            uprv_strncpy(usrDataPath, path, sizeof(usrDataPath) - 1);
+            usrDataPath[0] = 'u';
+            usrDataPath[1] = 's';
+            usrDataPath[2] = 'r';
+            usrDataPath[sizeof(usrDataPath) - 1] = 0;
+        }
+    }
+ 
+    // Note: We need to query the default locale *before* locking resbMutex.
+    const char *defaultLocale = uloc_getDefault();
+
+    Mutex lock(&resbMutex);    // Lock resbMutex until the end of this function.
+
+    /* We're going to skip all the locales that do not have any data */
+    r = findFirstExisting(path, name, defaultLocale, openType, &isRoot, &hasChopped, &isDefault, &intStatus);
+
+    // If we failed due to out-of-memory, report the failure and exit early.
+    if (intStatus == U_MEMORY_ALLOCATION_ERROR) {
+        *status = intStatus;
+        goto finish;
+    }
+
+    if(r != nullptr) { /* if there is one real locale, we can look for parents. */
+        t1 = r;
+        hasRealData = true;
+        if ( usingUSRData ) {  /* This code inserts user override data into the inheritance chain */
+            UErrorCode usrStatus = U_ZERO_ERROR;
+            UResourceDataEntry *u1 = init_entry(t1->fName, usrDataPath, &usrStatus);
+            // If we failed due to out-of-memory, report the failure and exit early.
+            if (intStatus == U_MEMORY_ALLOCATION_ERROR) {
+                *status = intStatus;
+                goto finish;
+            }
+            if ( u1 != nullptr ) {
+                if(u1->fBogus == U_ZERO_ERROR) {
+                    u1->fParent = t1;
+                    r = u1;
+                } else {
+                    /* the USR override data wasn't found, set it to be deleted */
+                    u1->fCountExisting = 0;
+                }
+            }
+        }
+        if ((hasChopped || mayHaveParent(name)) && !isRoot) {
+            if (!loadParentsExceptRoot(t1, name, UPRV_LENGTHOF(name), usingUSRData, usrDataPath, status)) {
+                goto finish;
+            }
+        }
+    }
+
+    /* we could have reached this point without having any real data */
+    /* if that is the case, we need to chain in the default locale   */
+    if(r==nullptr && openType == URES_OPEN_LOCALE_DEFAULT_ROOT && !isDefault && !isRoot) {
+        /* insert default locale */
+        uprv_strcpy(name, defaultLocale);
+        r = findFirstExisting(path, name, defaultLocale, openType, &isRoot, &hasChopped, &isDefault, &intStatus);
+        // If we failed due to out-of-memory, report the failure and exit early.
+        if (intStatus == U_MEMORY_ALLOCATION_ERROR) {
+            *status = intStatus;
+            goto finish;
+        }
+        intStatus = U_USING_DEFAULT_WARNING;
+        if(r != nullptr) { /* the default locale exists */
+            t1 = r;
+            hasRealData = true;
+            isDefault = true;
+            // TODO: Why not if (usingUSRData) { ... } like in the non-default-locale code path?
+            if ((hasChopped || mayHaveParent(name)) && !isRoot) {
+                if (!loadParentsExceptRoot(t1, name, UPRV_LENGTHOF(name), usingUSRData, usrDataPath, status)) {
+                    goto finish;
+                }
+            }
+        }
+    }
+
+    /* we could still have r == nullptr at this point - maybe even default locale is not */
+    /* present */
+    if(r == nullptr) {
+        uprv_strcpy(name, kRootLocaleName);
+        r = findFirstExisting(path, name, defaultLocale, openType, &isRoot, &hasChopped, &isDefault, &intStatus);
+        // If we failed due to out-of-memory, report the failure and exit early.
+        if (intStatus == U_MEMORY_ALLOCATION_ERROR) {
+            *status = intStatus;
+            goto finish;
+        }
+        if(r != nullptr) {
+            t1 = r;
+            intStatus = U_USING_DEFAULT_WARNING;
+            hasRealData = true;
+        } else { /* we don't even have the root locale */
+            *status = U_MISSING_RESOURCE_ERROR;
+            goto finish;
+        }
+    } else if(!isRoot && uprv_strcmp(t1->fName, kRootLocaleName) != 0 &&
+            t1->fParent == nullptr && !r->fData.noFallback) {
+        if (!insertRootBundle(t1, status)) {
+            goto finish;
+        }
+        if(!hasRealData) {
+            r->fBogus = U_USING_DEFAULT_WARNING;
+        }
+    }
+
+    // TODO: Does this ever loop?
+    while(r != nullptr && !isRoot && t1->fParent != nullptr) {
+        t1->fParent->fCountExisting++;
+        t1 = t1->fParent;
+    }
+
+finish:
+    if(U_SUCCESS(*status)) {
+        if(intStatus != U_ZERO_ERROR) {
+            *status = intStatus;  
+        }
+        return r;
+    } else {
+        return nullptr;
+    }
+}
+
+/**
+ * Version of entryOpen() and findFirstExisting() for ures_openDirect(),
+ * with no fallbacks.
+ * Parent and root locale bundles are loaded if
+ * the requested bundle does not have the "nofallback" flag.
+ */
+static UResourceDataEntry *
+entryOpenDirect(const char* path, const char* localeID, UErrorCode* status) {
+    initCache(status);
+    if(U_FAILURE(*status)) {
+        return nullptr;
+    }
+
+    // Note: We need to query the default locale *before* locking resbMutex.
+    // If the localeID is nullptr, then we want to use the default locale.
+    if (localeID == nullptr) {
+        localeID = uloc_getDefault();
+    } else if (*localeID == 0) {
+        // If the localeID is "", then we want to use the root locale.
+        localeID = kRootLocaleName;
+    }
+
+    Mutex lock(&resbMutex);
+
+    // findFirstExisting() without fallbacks.
+    UResourceDataEntry *r = init_entry(localeID, path, status);
+    if(U_SUCCESS(*status)) {
+        if(r->fBogus != U_ZERO_ERROR) {
+            r->fCountExisting--;
+            r = nullptr;
+        }
+    } else {
+        r = nullptr;
+    }
+
+    // Some code depends on the ures_openDirect() bundle to have a parent bundle chain,
+    // unless it is marked with "nofallback".
+    UResourceDataEntry *t1 = r;
+    if(r != nullptr && uprv_strcmp(localeID, kRootLocaleName) != 0 &&  // not root
+            r->fParent == nullptr && !r->fData.noFallback &&
+            uprv_strlen(localeID) < ULOC_FULLNAME_CAPACITY) {
+        char name[ULOC_FULLNAME_CAPACITY];
+        uprv_strcpy(name, localeID);
+        if(!chopLocale(name) || uprv_strcmp(name, kRootLocaleName) == 0 ||
+                loadParentsExceptRoot(t1, name, UPRV_LENGTHOF(name), false, nullptr, status)) {
+            if(uprv_strcmp(t1->fName, kRootLocaleName) != 0 && t1->fParent == nullptr) {
+                insertRootBundle(t1, status);
+            }
+        }
+        if(U_FAILURE(*status)) {
+            r = nullptr;
+        }
+    }
+
+    if(r != nullptr) {
+        // TODO: Does this ever loop?
+        while(t1->fParent != nullptr) {
+            t1->fParent->fCountExisting++;
+            t1 = t1->fParent;
+        }
+    }
+    return r;
+}
+
+/**
+ * Functions to create and destroy resource bundles.
+ *     CAUTION:  resbMutex must be locked when calling this function.
+ */
+/* INTERNAL: */
+static void entryCloseInt(UResourceDataEntry *resB) {
+    UResourceDataEntry *p = resB;
+
+    while(resB != nullptr) {
+        p = resB->fParent;
+        resB->fCountExisting--;
+
+        /* Entries are left in the cache. TODO: add ures_flushCache() to force a flush
+         of the cache. */
+/*
+        if(resB->fCountExisting <= 0) {
+            uhash_remove(cache, resB);
+            if(resB->fBogus == U_ZERO_ERROR) {
+                res_unload(&(resB->fData));
+            }
+            if(resB->fName != nullptr) {
+                uprv_free(resB->fName);
+            }
+            if(resB->fPath != nullptr) {
+                uprv_free(resB->fPath);
+            }
+            uprv_free(resB);
+        }
+*/
+
+        resB = p;
+    }
+}
+
+/** 
+ *  API: closes a resource bundle and cleans up.
+ */
+
+static void entryClose(UResourceDataEntry *resB) {
+  Mutex lock(&resbMutex);
+  entryCloseInt(resB);
+}
+
+/*
+U_CFUNC void ures_setResPath(UResourceBundle *resB, const char* toAdd) {
+  if(resB->fResPath == nullptr) {
+    resB->fResPath = resB->fResBuf;
+    *(resB->fResPath) = 0;
+  } 
+  resB->fResPathLen = uprv_strlen(toAdd);
+  if(RES_BUFSIZE <= resB->fResPathLen+1) {
+    if(resB->fResPath == resB->fResBuf) {
+      resB->fResPath = (char *)uprv_malloc((resB->fResPathLen+1)*sizeof(char));
+    } else {
+      resB->fResPath = (char *)uprv_realloc(resB->fResPath, (resB->fResPathLen+1)*sizeof(char));
+    }
+  }
+  uprv_strcpy(resB->fResPath, toAdd);
+}
+*/
+static void ures_appendResPath(UResourceBundle *resB, const char* toAdd, int32_t lenToAdd, UErrorCode *status) {
+    int32_t resPathLenOrig = resB->fResPathLen;
+    if(resB->fResPath == nullptr) {
+        resB->fResPath = resB->fResBuf;
+        *(resB->fResPath) = 0;
+        resB->fResPathLen = 0;
+    } 
+    resB->fResPathLen += lenToAdd;
+    if(RES_BUFSIZE <= resB->fResPathLen+1) {
+        if(resB->fResPath == resB->fResBuf) {
+            resB->fResPath = (char *)uprv_malloc((resB->fResPathLen+1)*sizeof(char));
+            /* Check that memory was allocated correctly. */
+            if (resB->fResPath == nullptr) {
+                *status = U_MEMORY_ALLOCATION_ERROR;
+                return;
+            }
+            uprv_strcpy(resB->fResPath, resB->fResBuf);
+        } else {
+            char *temp = (char *)uprv_realloc(resB->fResPath, (resB->fResPathLen+1)*sizeof(char));
+            /* Check that memory was reallocated correctly. */
+            if (temp == nullptr) {
+                *status = U_MEMORY_ALLOCATION_ERROR;
+                return;
+            }
+            resB->fResPath = temp;
+        }
+    }
+    uprv_strcpy(resB->fResPath + resPathLenOrig, toAdd);
+}
+
+static void ures_freeResPath(UResourceBundle *resB) {
+    if (resB->fResPath && resB->fResPath != resB->fResBuf) {
+        uprv_free(resB->fResPath);
+    }
+    resB->fResPath = nullptr;
+    resB->fResPathLen = 0;
+}
+
+static void
+ures_closeBundle(UResourceBundle* resB, UBool freeBundleObj)
+{
+    if(resB != nullptr) {
+        if(resB->fData != nullptr) {
+            entryClose(resB->fData);
+        }
+        if(resB->fVersion != nullptr) {
+            uprv_free(resB->fVersion);
+        }
+        ures_freeResPath(resB);
+
+        if(ures_isStackObject(resB) == false && freeBundleObj) {
+            uprv_free(resB);
+        }
+#if 0 /*U_DEBUG*/
+        else {
+            /* poison the data */
+            uprv_memset(resB, -1, sizeof(UResourceBundle));
+        }
+#endif
+    }
+}
+
+U_CAPI void  U_EXPORT2
+ures_close(UResourceBundle* resB)
+{
+    ures_closeBundle(resB, true);
+}
+
+namespace {
+
+UResourceBundle *init_resb_result(
+        UResourceDataEntry *dataEntry, Resource r, const char *key, int32_t idx,
+        UResourceDataEntry *validLocaleDataEntry, const char *containerResPath,
+        int32_t recursionDepth,
+        UResourceBundle *resB, UErrorCode *status);
+
+// TODO: Try to refactor further, so that we output a dataEntry + Resource + (optionally) resPath,
+// rather than a UResourceBundle.
+// May need to entryIncrease() the resulting dataEntry.
+UResourceBundle *getAliasTargetAsResourceBundle(
+        const ResourceData &resData, Resource r, const char *key, int32_t idx,
+        UResourceDataEntry *validLocaleDataEntry, const char *containerResPath,
+        int32_t recursionDepth,
+        UResourceBundle *resB, UErrorCode *status) {
+    // TODO: When an error occurs: Should we return nullptr vs. resB?
+    if (U_FAILURE(*status)) { return resB; }
+    U_ASSERT(RES_GET_TYPE(r) == URES_ALIAS);
+    int32_t len = 0;
+    const char16_t *alias = res_getAlias(&resData, r, &len);
+    if(len <= 0) {
+        // bad alias
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return resB;
+    }
+
+    // Copy the UTF-16 alias string into an invariant-character string.
+    //
+    // We do this so that res_findResource() can modify the path,
+    // which allows us to remove redundant _res_findResource() variants
+    // in uresdata.c.
+    // res_findResource() now NUL-terminates each segment so that table keys
+    // can always be compared with strcmp() instead of strncmp().
+    // Saves code there and simplifies testing and code coverage.
+    //
+    // markus 2003oct17
+    CharString chAlias;
+    chAlias.appendInvariantChars(alias, len, *status);
+    if (U_FAILURE(*status)) {
+        return nullptr;
+    }
+
+    // We have an alias, now let's cut it up.
+    const char *path = nullptr, *locale = nullptr, *keyPath = nullptr;
+    if(chAlias[0] == RES_PATH_SEPARATOR) {
+        // There is a path included.
+        char *chAliasData = chAlias.data();
+        char *sep = chAliasData + 1;
+        path = sep;
+        sep = uprv_strchr(sep, RES_PATH_SEPARATOR);
+        if(sep != nullptr) {
+            *sep++ = 0;
+        }
+        if(uprv_strcmp(path, "LOCALE") == 0) {
+            // This is an XPath alias, starting with "/LOCALE/".
+            // It contains the path to a resource which should be looked up
+            // starting in the valid locale.
+            // TODO: Can/should we forbid a /LOCALE alias without key path?
+            //   It seems weird to alias to the same path, just starting from the valid locale.
+            //   That will often yield an infinite loop.
+            keyPath = sep;
+            // Read from the valid locale which we already have.
+            path = locale = nullptr;
+        } else {
+            if(uprv_strcmp(path, "ICUDATA") == 0) { /* want ICU data */
+                path = nullptr;
+            }
+            if (sep == nullptr) {
+                // TODO: This ends up using the root bundle. Can/should we forbid this?
+                locale = "";
+            } else {
+                locale = sep;
+                sep = uprv_strchr(sep, RES_PATH_SEPARATOR);
+                if(sep != nullptr) {
+                    *sep++ = 0;
+                }
+                keyPath = sep;
+            }
+        }
+    } else {
+        // No path, start with a locale.
+        char *sep = chAlias.data();
+        locale = sep;
+        sep = uprv_strchr(sep, RES_PATH_SEPARATOR);
+        if(sep != nullptr) {
+            *sep++ = 0;
+        }
+        keyPath = sep;
+        path = validLocaleDataEntry->fPath;
+    }
+
+    // Got almost everything, let's try to open.
+    // First, open the bundle with real data.
+    LocalUResourceBundlePointer mainRes;
+    UResourceDataEntry *dataEntry;
+    if (locale == nullptr) {
+        // alias = /LOCALE/keyPath
+        // Read from the valid locale which we already have.
+        dataEntry = validLocaleDataEntry;
+    } else {
+        UErrorCode intStatus = U_ZERO_ERROR;
+        // TODO: Shouldn't we use ures_open() for locale data bundles (!noFallback)?
+        mainRes.adoptInstead(ures_openDirect(path, locale, &intStatus));
+        if(U_FAILURE(intStatus)) {
+            // We failed to open the resource bundle we're aliasing to.
+            *status = intStatus;
+            return resB;
+        }
+        dataEntry = mainRes->fData;
+    }
+
+    const char* temp = nullptr;
+    if(keyPath == nullptr) {
+        // No key path. This means that we are going to to use the corresponding resource from
+        // another bundle.
+        // TODO: Why the special code path?
+        //   Why not put together a key path from containerResPath + key or idx,
+        //   as a comment below suggests, and go into the regular code branch?
+        // First, we are going to get a corresponding container
+        // resource to the one we are searching.
+        r = dataEntry->fData.rootRes;
+        if(containerResPath) {
+            chAlias.clear().append(containerResPath, *status);
+            if (U_FAILURE(*status)) {
+                return nullptr;
+            }
+            char *aKey = chAlias.data();
+            // TODO: should res_findResource() return a new dataEntry, too?
+            r = res_findResource(&dataEntry->fData, r, &aKey, &temp);
+        }
+        if(key) {
+            // We need to make keyPath from the containerResPath and
+            // current key, if there is a key associated.
+            chAlias.clear().append(key, *status);
+            if (U_FAILURE(*status)) {
+                return nullptr;
+            }
+            char *aKey = chAlias.data();
+            r = res_findResource(&dataEntry->fData, r, &aKey, &temp);
+        } else if(idx != -1) {
+            // If there is no key, but there is an index, try to get by the index.
+            // Here we have either a table or an array, so get the element.
+            int32_t type = RES_GET_TYPE(r);
+            if(URES_IS_TABLE(type)) {
+                const char *aKey;
+                r = res_getTableItemByIndex(&dataEntry->fData, r, idx, &aKey);
+            } else { /* array */
+                r = res_getArrayItem(&dataEntry->fData, r, idx);
+            }
+        }
+        if(r != RES_BOGUS) {
+            resB = init_resb_result(
+                dataEntry, r, temp, -1, validLocaleDataEntry, nullptr, recursionDepth+1,
+                resB, status);
+        } else {
+            *status = U_MISSING_RESOURCE_ERROR;
+        }
+    } else {
+        // This one is a bit trickier.
+        // We start finding keys, but after we resolve one alias, the path might continue.
+        // Consider:
+        //     aliastest:alias { "testtypes/anotheralias/Sequence" }
+        //     anotheralias:alias { "/ICUDATA/sh/CollationElements" }
+        // aliastest resource should finally have the sequence, not collation elements.
+        CharString pathBuf(keyPath, *status);
+        if (U_FAILURE(*status)) {
+            return nullptr;
+        }
+        char *myPath = pathBuf.data();
+        containerResPath = nullptr;
+        // Now we have fallback following here.
+        for(;;) {
+            r = dataEntry->fData.rootRes;
+            // TODO: Move  containerResPath = nullptr  to here,
+            // consistent with restarting from the rootRes of another bundle?!
+
+            // This loop handles 'found' resources over several levels.
+            while(*myPath && U_SUCCESS(*status)) {
+                r = res_findResource(&(dataEntry->fData), r, &myPath, &temp);
+                if(r == RES_BOGUS) {
+                    // No resource found, we don't really want to look anymore on this level.
+                    break;
+                }
+                // Found a resource, but it might be an indirection.
+                resB = init_resb_result(
+                    dataEntry, r, temp, -1,
+                    validLocaleDataEntry, containerResPath, recursionDepth+1,
+                    resB, status);
+                if (U_FAILURE(*status)) {
+                    break;
+                }
+                if (temp == nullptr || uprv_strcmp(keyPath, temp) != 0) {
+                    // The call to init_resb_result() above will set resB->fKeyPath to be
+                    // the same as resB->fKey,
+                    // throwing away any additional path elements if we had them --
+                    // if the key path wasn't just a single resource ID, clear out
+                    // the bundle's key path and re-set it to be equal to keyPath.
+                    ures_freeResPath(resB);
+                    ures_appendResPath(resB, keyPath, (int32_t)uprv_strlen(keyPath), status);
+                    if(resB->fResPath[resB->fResPathLen-1] != RES_PATH_SEPARATOR) {
+                        ures_appendResPath(resB, RES_PATH_SEPARATOR_S, 1, status);
+                    }
+                    if (U_FAILURE(*status)) {
+                        break;
+                    }
+                }
+                r = resB->fRes; /* switch to a new resource, possibly a new tree */
+                dataEntry = resB->fData;
+                containerResPath = resB->fResPath;
+            }
+            if (U_FAILURE(*status) || r != RES_BOGUS) {
+                break;
+            }
+            // Fall back to the parent bundle, if there is one.
+            dataEntry = dataEntry->fParent;
+            if (dataEntry == nullptr) {
+                *status = U_MISSING_RESOURCE_ERROR;
+                break;
+            }
+            // Copy the same keyPath again.
+            myPath = pathBuf.data();
+            uprv_strcpy(myPath, keyPath);
+        }
+    }
+    if(mainRes.getAlias() == resB) {
+        mainRes.orphan();
+    }
+    ResourceTracer(resB).maybeTrace("getalias");
+    return resB;
+}
+
+// Recursive function, should be called only by itself, by its simpler wrapper,
+// or by getAliasTargetAsResourceBundle().
+UResourceBundle *init_resb_result(
+        UResourceDataEntry *dataEntry, Resource r, const char *key, int32_t idx,
+        UResourceDataEntry *validLocaleDataEntry, const char *containerResPath,
+        int32_t recursionDepth,
+        UResourceBundle *resB, UErrorCode *status) {
+    // TODO: When an error occurs: Should we return nullptr vs. resB?
+    if(status == nullptr || U_FAILURE(*status)) {
+        return resB;
+    }
+    if (validLocaleDataEntry == nullptr) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+    if(RES_GET_TYPE(r) == URES_ALIAS) {
+        // This is an alias, need to exchange with real data.
+        if(recursionDepth >= URES_MAX_ALIAS_LEVEL) {
+            *status = U_TOO_MANY_ALIASES_ERROR;
+            return resB;
+        }
+        return getAliasTargetAsResourceBundle(
+            dataEntry->fData, r, key, idx,
+            validLocaleDataEntry, containerResPath, recursionDepth, resB, status);
+    }
+    if(resB == nullptr) {
+        resB = (UResourceBundle *)uprv_malloc(sizeof(UResourceBundle));
+        if (resB == nullptr) {
+            *status = U_MEMORY_ALLOCATION_ERROR;
+            return nullptr;
+        }
+        ures_setIsStackObject(resB, false);
+        resB->fResPath = nullptr;
+        resB->fResPathLen = 0;
+    } else {
+        if(resB->fData != nullptr) {
+            entryClose(resB->fData);
+        }
+        if(resB->fVersion != nullptr) {
+            uprv_free(resB->fVersion);
+        }
+        /* 
+        weiv: if stack object was passed in, it doesn't really need to be reinited,
+        since the purpose of initing is to remove stack junk. However, at this point 
+        we would not do anything to an allocated object, so stack object should be
+        treated the same
+        */
+        /*
+        if(ures_isStackObject(resB) != false) {
+        ures_initStackObject(resB);
+        }
+        */
+        if(containerResPath != resB->fResPath) {
+            ures_freeResPath(resB);
+        }
+    }
+    resB->fData = dataEntry;
+    entryIncrease(resB->fData);
+    resB->fHasFallback = false;
+    resB->fIsTopLevel = false;
+    resB->fIndex = -1;
+    resB->fKey = key; 
+    resB->fValidLocaleDataEntry = validLocaleDataEntry;
+    if(containerResPath != resB->fResPath) {
+        ures_appendResPath(
+            resB, containerResPath, static_cast(uprv_strlen(containerResPath)), status);
+    }
+    if(key != nullptr) {
+        ures_appendResPath(resB, key, (int32_t)uprv_strlen(key), status);
+        if(resB->fResPath[resB->fResPathLen-1] != RES_PATH_SEPARATOR) {
+            ures_appendResPath(resB, RES_PATH_SEPARATOR_S, 1, status);
+        }
+    } else if(idx >= 0) {
+        char buf[256];
+        int32_t len = T_CString_integerToString(buf, idx, 10);
+        ures_appendResPath(resB, buf, len, status);
+        if(resB->fResPath[resB->fResPathLen-1] != RES_PATH_SEPARATOR) {
+            ures_appendResPath(resB, RES_PATH_SEPARATOR_S, 1, status);
+        }
+    }
+    /* Make sure that Purify doesn't complain about uninitialized memory copies. */
+    {
+        int32_t usedLen = ((resB->fResBuf == resB->fResPath) ? resB->fResPathLen : 0);
+        uprv_memset(resB->fResBuf + usedLen, 0, sizeof(resB->fResBuf) - usedLen);
+    }
+
+    resB->fVersion = nullptr;
+    resB->fRes = r;
+    resB->fSize = res_countArrayItems(&resB->getResData(), resB->fRes);
+    ResourceTracer(resB).trace("get");
+    return resB;
+}
+
+UResourceBundle *init_resb_result(
+        UResourceDataEntry *dataEntry, Resource r, const char *key, int32_t idx,
+        // validLocaleDataEntry + containerResPath
+        const UResourceBundle *container,
+        UResourceBundle *resB, UErrorCode *status) {
+    return init_resb_result(
+        dataEntry, r, key, idx,
+        container->fValidLocaleDataEntry, container->fResPath, 0, resB, status);
+}
+
+}  // namespace
+
+UResourceBundle *ures_copyResb(UResourceBundle *r, const UResourceBundle *original, UErrorCode *status) {
+    UBool isStackObject;
+    if(U_FAILURE(*status) || r == original) {
+        return r;
+    }
+    if(original != nullptr) {
+        if(r == nullptr) {
+            isStackObject = false;
+            r = (UResourceBundle *)uprv_malloc(sizeof(UResourceBundle));
+            /* test for nullptr */
+            if (r == nullptr) {
+                *status = U_MEMORY_ALLOCATION_ERROR;
+                return nullptr;
+            }
+        } else {
+            isStackObject = ures_isStackObject(r);
+            ures_closeBundle(r, false);
+        }
+        uprv_memcpy(r, original, sizeof(UResourceBundle));
+        r->fResPath = nullptr;
+        r->fResPathLen = 0;
+        if(original->fResPath) {
+            ures_appendResPath(r, original->fResPath, original->fResPathLen, status);
+        }
+        ures_setIsStackObject(r, isStackObject);
+        if(r->fData != nullptr) {
+            entryIncrease(r->fData);
+        }
+    }
+    return r;
+}
+
+/**
+ * Functions to retrieve data from resource bundles.
+ */
+
+U_CAPI const char16_t* U_EXPORT2 ures_getString(const UResourceBundle* resB, int32_t* len, UErrorCode* status) {
+    const char16_t *s;
+    if (status==nullptr || U_FAILURE(*status)) {
+        return nullptr;
+    }
+    if(resB == nullptr) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+    s = res_getString({resB}, &resB->getResData(), resB->fRes, len);
+    if (s == nullptr) {
+        *status = U_RESOURCE_TYPE_MISMATCH;
+    }
+    return s;
+}
+
+static const char *
+ures_toUTF8String(const char16_t *s16, int32_t length16,
+                  char *dest, int32_t *pLength,
+                  UBool forceCopy,
+                  UErrorCode *status) {
+    int32_t capacity;
+
+    if (U_FAILURE(*status)) {
+        return nullptr;
+    }
+    if (pLength != nullptr) {
+        capacity = *pLength;
+    } else {
+        capacity = 0;
+    }
+    if (capacity < 0 || (capacity > 0 && dest == nullptr)) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+
+    if (length16 == 0) {
+        /* empty string, return as read-only pointer */
+        if (pLength != nullptr) {
+            *pLength = 0;
+        }
+        if (forceCopy) {
+            u_terminateChars(dest, capacity, 0, status);
+            return dest;
+        } else {
+            return "";
+        }
+    } else {
+        /* We need to transform the string to the destination buffer. */
+        if (capacity < length16) {
+            /* No chance for the string to fit. Pure preflighting. */
+            return u_strToUTF8(nullptr, 0, pLength, s16, length16, status);
+        }
+        if (!forceCopy && (length16 <= 0x2aaaaaaa)) {
+            /*
+             * We know the string will fit into dest because each char16_t turns
+             * into at most three UTF-8 bytes. Fill the latter part of dest
+             * so that callers do not expect to use dest as a string pointer,
+             * hopefully leading to more robust code for when resource bundles
+             * may store UTF-8 natively.
+             * (In which case dest would not be used at all.)
+             *
+             * We do not do this if forceCopy=true because then the caller
+             * expects the string to start exactly at dest.
+             *
+             * The test above for <= 0x2aaaaaaa prevents overflows.
+             * The +1 is for the NUL terminator.
+             */
+            int32_t maxLength = 3 * length16 + 1;
+            if (capacity > maxLength) {
+                dest += capacity - maxLength;
+                capacity = maxLength;
+            }
+        }
+        return u_strToUTF8(dest, capacity, pLength, s16, length16, status);
+    }
+}
+
+U_CAPI const char * U_EXPORT2
+ures_getUTF8String(const UResourceBundle *resB,
+                   char *dest, int32_t *pLength,
+                   UBool forceCopy,
+                   UErrorCode *status) {
+    int32_t length16;
+    const char16_t *s16 = ures_getString(resB, &length16, status);
+    return ures_toUTF8String(s16, length16, dest, pLength, forceCopy, status);
+}
+
+U_CAPI const uint8_t* U_EXPORT2 ures_getBinary(const UResourceBundle* resB, int32_t* len, 
+                                               UErrorCode*               status) {
+  const uint8_t *p;
+  if (status==nullptr || U_FAILURE(*status)) {
+    return nullptr;
+  }
+  if(resB == nullptr) {
+    *status = U_ILLEGAL_ARGUMENT_ERROR;
+    return nullptr;
+  }
+  p = res_getBinary({resB}, &resB->getResData(), resB->fRes, len);
+  if (p == nullptr) {
+    *status = U_RESOURCE_TYPE_MISMATCH;
+  }
+  return p;
+}
+
+U_CAPI const int32_t* U_EXPORT2 ures_getIntVector(const UResourceBundle* resB, int32_t* len, 
+                                                   UErrorCode*               status) {
+  const int32_t *p;
+  if (status==nullptr || U_FAILURE(*status)) {
+    return nullptr;
+  }
+  if(resB == nullptr) {
+    *status = U_ILLEGAL_ARGUMENT_ERROR;
+    return nullptr;
+  }
+  p = res_getIntVector({resB}, &resB->getResData(), resB->fRes, len);
+  if (p == nullptr) {
+    *status = U_RESOURCE_TYPE_MISMATCH;
+  }
+  return p;
+}
+
+/* this function returns a signed integer */ 
+/* it performs sign extension */
+U_CAPI int32_t U_EXPORT2 ures_getInt(const UResourceBundle* resB, UErrorCode *status) {
+  if (status==nullptr || U_FAILURE(*status)) {
+    return 0xffffffff;
+  }
+  if(resB == nullptr) {
+    *status = U_ILLEGAL_ARGUMENT_ERROR;
+    return 0xffffffff;
+  }
+  if(RES_GET_TYPE(resB->fRes) != URES_INT) {
+    *status = U_RESOURCE_TYPE_MISMATCH;
+    return 0xffffffff;
+  }
+  return res_getInt({resB}, resB->fRes);
+}
+
+U_CAPI uint32_t U_EXPORT2 ures_getUInt(const UResourceBundle* resB, UErrorCode *status) {
+  if (status==nullptr || U_FAILURE(*status)) {
+    return 0xffffffff;
+  }
+  if(resB == nullptr) {
+    *status = U_ILLEGAL_ARGUMENT_ERROR;
+    return 0xffffffff;
+  }
+  if(RES_GET_TYPE(resB->fRes) != URES_INT) {
+    *status = U_RESOURCE_TYPE_MISMATCH;
+    return 0xffffffff;
+  }
+  return res_getUInt({resB}, resB->fRes);
+}
+
+U_CAPI UResType U_EXPORT2 ures_getType(const UResourceBundle *resB) {
+  if(resB == nullptr) {
+    return URES_NONE;
+  }
+  return res_getPublicType(resB->fRes);
+}
+
+U_CAPI const char * U_EXPORT2 ures_getKey(const UResourceBundle *resB) {
+  //
+  // TODO: Trace ures_getKey? I guess not usually.
+  //
+  // We usually get the key string to decide whether we want the value, or to
+  // make a key-value pair. Tracing the value should suffice.
+  //
+  // However, I believe we have some data (e.g., in res_index) where the key
+  // strings are the data. Tracing the enclosing table should suffice.
+  //
+  if(resB == nullptr) {
+    return nullptr;
+  }
+  return(resB->fKey);
+}
+
+U_CAPI int32_t U_EXPORT2 ures_getSize(const UResourceBundle *resB) {
+  if(resB == nullptr) {
+    return 0;
+  }
+  
+  return resB->fSize;
+}
+
+static const char16_t* ures_getStringWithAlias(const UResourceBundle *resB, Resource r, int32_t sIndex, int32_t *len, UErrorCode *status) {
+  if(RES_GET_TYPE(r) == URES_ALIAS) {
+    const char16_t* result = 0;
+    UResourceBundle *tempRes = ures_getByIndex(resB, sIndex, nullptr, status);
+    result = ures_getString(tempRes, len, status);
+    ures_close(tempRes);
+    return result;
+  } else {
+    return res_getString({resB, sIndex}, &resB->getResData(), r, len); 
+  }
+}
+
+U_CAPI void U_EXPORT2 ures_resetIterator(UResourceBundle *resB){
+  if(resB == nullptr) {
+    return;
+  }
+  resB->fIndex = -1;
+}
+
+U_CAPI UBool U_EXPORT2 ures_hasNext(const UResourceBundle *resB) {
+  if(resB == nullptr) {
+    return false;
+  }
+  return (UBool)(resB->fIndex < resB->fSize-1);
+}
+
+U_CAPI const char16_t* U_EXPORT2 ures_getNextString(UResourceBundle *resB, int32_t* len, const char ** key, UErrorCode *status) {
+  Resource r = RES_BOGUS;
+  
+  if (status==nullptr || U_FAILURE(*status)) {
+    return nullptr;
+  }
+  if(resB == nullptr) {
+    *status = U_ILLEGAL_ARGUMENT_ERROR;
+    return nullptr;
+  }
+  
+  if(resB->fIndex == resB->fSize-1) {
+    *status = U_INDEX_OUTOFBOUNDS_ERROR;
+  } else {
+    resB->fIndex++;
+    switch(RES_GET_TYPE(resB->fRes)) {
+    case URES_STRING:
+    case URES_STRING_V2:
+      return res_getString({resB}, &resB->getResData(), resB->fRes, len);
+    case URES_TABLE:
+    case URES_TABLE16:
+    case URES_TABLE32:
+      r = res_getTableItemByIndex(&resB->getResData(), resB->fRes, resB->fIndex, key);
+      if(r == RES_BOGUS && resB->fHasFallback) {
+        /* TODO: do the fallback */
+      }
+      return ures_getStringWithAlias(resB, r, resB->fIndex, len, status);
+    case URES_ARRAY:
+    case URES_ARRAY16:
+      r = res_getArrayItem(&resB->getResData(), resB->fRes, resB->fIndex);
+      if(r == RES_BOGUS && resB->fHasFallback) {
+        /* TODO: do the fallback */
+      }
+      return ures_getStringWithAlias(resB, r, resB->fIndex, len, status);
+    case URES_ALIAS:
+      return ures_getStringWithAlias(resB, resB->fRes, resB->fIndex, len, status);
+    case URES_INT:
+    case URES_BINARY:
+    case URES_INT_VECTOR:
+        *status = U_RESOURCE_TYPE_MISMATCH;
+        U_FALLTHROUGH;
+    default:
+      return nullptr;
+    }
+  }
+
+  return nullptr;
+}
+
+U_CAPI UResourceBundle* U_EXPORT2 ures_getNextResource(UResourceBundle *resB, UResourceBundle *fillIn, UErrorCode *status) {
+    const char *key = nullptr;
+    Resource r = RES_BOGUS;
+
+    if (status==nullptr || U_FAILURE(*status)) {
+            /*return nullptr;*/
+            return fillIn;
+    }
+    if(resB == nullptr) {
+            *status = U_ILLEGAL_ARGUMENT_ERROR;
+            /*return nullptr;*/
+            return fillIn;
+    }
+
+    if(resB->fIndex == resB->fSize-1) {
+      *status = U_INDEX_OUTOFBOUNDS_ERROR;
+      /*return nullptr;*/
+    } else {
+        resB->fIndex++;
+        switch(RES_GET_TYPE(resB->fRes)) {
+        case URES_INT:
+        case URES_BINARY:
+        case URES_STRING:
+        case URES_STRING_V2:
+        case URES_INT_VECTOR:
+            return ures_copyResb(fillIn, resB, status);
+        case URES_TABLE:
+        case URES_TABLE16:
+        case URES_TABLE32:
+            r = res_getTableItemByIndex(&resB->getResData(), resB->fRes, resB->fIndex, &key);
+            if(r == RES_BOGUS && resB->fHasFallback) {
+                /* TODO: do the fallback */
+            }
+            return init_resb_result(resB->fData, r, key, resB->fIndex, resB, fillIn, status);
+        case URES_ARRAY:
+        case URES_ARRAY16:
+            r = res_getArrayItem(&resB->getResData(), resB->fRes, resB->fIndex);
+            if(r == RES_BOGUS && resB->fHasFallback) {
+                /* TODO: do the fallback */
+            }
+            return init_resb_result(resB->fData, r, key, resB->fIndex, resB, fillIn, status);
+        default:
+            /*return nullptr;*/
+            return fillIn;
+        }
+    }
+    /*return nullptr;*/
+    return fillIn;
+}
+
+U_CAPI UResourceBundle* U_EXPORT2 ures_getByIndex(const UResourceBundle *resB, int32_t indexR, UResourceBundle *fillIn, UErrorCode *status) {
+    const char* key = nullptr;
+    Resource r = RES_BOGUS;
+
+    if (status==nullptr || U_FAILURE(*status)) {
+        /*return nullptr;*/
+        return fillIn;
+    }
+    if(resB == nullptr) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        /*return nullptr;*/
+        return fillIn;
+    }
+
+    if(indexR >= 0 && resB->fSize > indexR) {
+        switch(RES_GET_TYPE(resB->fRes)) {
+        case URES_INT:
+        case URES_BINARY:
+        case URES_STRING:
+        case URES_STRING_V2:
+        case URES_INT_VECTOR:
+            return ures_copyResb(fillIn, resB, status);
+        case URES_TABLE:
+        case URES_TABLE16:
+        case URES_TABLE32:
+            r = res_getTableItemByIndex(&resB->getResData(), resB->fRes, indexR, &key);
+            if(r == RES_BOGUS && resB->fHasFallback) {
+                /* TODO: do the fallback */
+            }
+            return init_resb_result(resB->fData, r, key, indexR, resB, fillIn, status);
+        case URES_ARRAY:
+        case URES_ARRAY16:
+            r = res_getArrayItem(&resB->getResData(), resB->fRes, indexR);
+            if(r == RES_BOGUS && resB->fHasFallback) {
+                /* TODO: do the fallback */
+            }
+            return init_resb_result(resB->fData, r, key, indexR, resB, fillIn, status);
+        default:
+            /*return nullptr;*/
+            return fillIn;
+        }
+    } else {
+        *status = U_MISSING_RESOURCE_ERROR;
+    }
+    /*return nullptr;*/
+    return fillIn;
+}
+
+U_CAPI const char16_t* U_EXPORT2 ures_getStringByIndex(const UResourceBundle *resB, int32_t indexS, int32_t* len, UErrorCode *status) {
+    const char* key = nullptr;
+    Resource r = RES_BOGUS;
+
+    if (status==nullptr || U_FAILURE(*status)) {
+        return nullptr;
+    }
+    if(resB == nullptr) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+
+    if(indexS >= 0 && resB->fSize > indexS) {
+        switch(RES_GET_TYPE(resB->fRes)) {
+        case URES_STRING:
+        case URES_STRING_V2:
+            return res_getString({resB}, &resB->getResData(), resB->fRes, len);
+        case URES_TABLE:
+        case URES_TABLE16:
+        case URES_TABLE32:
+            r = res_getTableItemByIndex(&resB->getResData(), resB->fRes, indexS, &key);
+            if(r == RES_BOGUS && resB->fHasFallback) {
+                /* TODO: do the fallback */
+            }
+            return ures_getStringWithAlias(resB, r, indexS, len, status);
+        case URES_ARRAY:
+        case URES_ARRAY16:
+            r = res_getArrayItem(&resB->getResData(), resB->fRes, indexS);
+            if(r == RES_BOGUS && resB->fHasFallback) {
+                /* TODO: do the fallback */
+            }
+            return ures_getStringWithAlias(resB, r, indexS, len, status);
+        case URES_ALIAS:
+            return ures_getStringWithAlias(resB, resB->fRes, indexS, len, status);
+        case URES_INT:
+        case URES_BINARY:
+        case URES_INT_VECTOR:
+            *status = U_RESOURCE_TYPE_MISMATCH;
+            break;
+        default:
+          /* must not occur */
+          *status = U_INTERNAL_PROGRAM_ERROR;
+          break;
+        }
+    } else {
+        *status = U_MISSING_RESOURCE_ERROR;
+    }
+    return nullptr;
+}
+
+U_CAPI const char * U_EXPORT2
+ures_getUTF8StringByIndex(const UResourceBundle *resB,
+                          int32_t idx,
+                          char *dest, int32_t *pLength,
+                          UBool forceCopy,
+                          UErrorCode *status) {
+    int32_t length16;
+    const char16_t *s16 = ures_getStringByIndex(resB, idx, &length16, status);
+    return ures_toUTF8String(s16, length16, dest, pLength, forceCopy, status);
+}
+
+/*U_CAPI const char *ures_getResPath(UResourceBundle *resB) {
+  return resB->fResPath;
+}*/
+
+U_CAPI UResourceBundle* U_EXPORT2
+ures_findResource(const char* path, UResourceBundle *fillIn, UErrorCode *status) 
+{
+  UResourceBundle *first = nullptr; 
+  UResourceBundle *result = fillIn;
+  char *packageName = nullptr;
+  char *pathToResource = nullptr, *save = nullptr;
+  char *locale = nullptr, *localeEnd = nullptr;
+  int32_t length;
+
+  if(status == nullptr || U_FAILURE(*status)) {
+    return result;
+  }
+
+  length = (int32_t)(uprv_strlen(path)+1);
+  save = pathToResource = (char *)uprv_malloc(length*sizeof(char));
+  /* test for nullptr */
+  if(pathToResource == nullptr) {
+    *status = U_MEMORY_ALLOCATION_ERROR;
+    return result;
+  }
+  uprv_memcpy(pathToResource, path, length);
+
+  locale = pathToResource;
+  if(*pathToResource == RES_PATH_SEPARATOR) { /* there is a path specification */
+    pathToResource++;
+    packageName = pathToResource;
+    pathToResource = uprv_strchr(pathToResource, RES_PATH_SEPARATOR);
+    if(pathToResource == nullptr) {
+      *status = U_ILLEGAL_ARGUMENT_ERROR;
+    } else {
+      *pathToResource = 0;
+      locale = pathToResource+1;
+    }
+  }
+
+  localeEnd = uprv_strchr(locale, RES_PATH_SEPARATOR);
+  if(localeEnd != nullptr) {
+    *localeEnd = 0;
+  }
+
+  first = ures_open(packageName, locale, status);
+
+  if(U_SUCCESS(*status)) {
+    if(localeEnd) {
+      result = ures_findSubResource(first, localeEnd+1, fillIn, status);
+    } else {
+      result = ures_copyResb(fillIn, first, status);
+    }
+    ures_close(first);
+  }
+  uprv_free(save);
+  return result;
+}
+
+U_CAPI UResourceBundle* U_EXPORT2
+ures_findSubResource(const UResourceBundle *resB, char* path, UResourceBundle *fillIn, UErrorCode *status) 
+{
+  Resource res = RES_BOGUS;
+  UResourceBundle *result = fillIn;
+  const char *key;
+
+  if(status == nullptr || U_FAILURE(*status)) {
+    return result;
+  }
+
+  /* here we do looping and circular alias checking */
+  /* this loop is here because aliasing is resolved on this level, not on res level */
+  /* so, when we encounter an alias, it is not an aggregate resource, so we return */
+  do {
+    res = res_findResource(&resB->getResData(), resB->fRes, &path, &key); 
+    if(res != RES_BOGUS) {
+        result = init_resb_result(resB->fData, res, key, -1, resB, fillIn, status);
+        resB = result;
+    } else {
+        *status = U_MISSING_RESOURCE_ERROR;
+        break;
+    }
+  } while(*path); /* there is more stuff in the path */
+
+  return result;
+}
+U_CAPI const char16_t* U_EXPORT2
+ures_getStringByKeyWithFallback(const UResourceBundle *resB, 
+                                const char* inKey, 
+                                int32_t* len,
+                                UErrorCode *status) {
+
+    UResourceBundle stack;
+    const char16_t* retVal = nullptr;
+    ures_initStackObject(&stack);
+    ures_getByKeyWithFallback(resB, inKey, &stack, status);
+    int32_t length;
+    retVal = ures_getString(&stack, &length, status);
+    ures_close(&stack);
+    if (U_FAILURE(*status)) {
+        return nullptr;
+    }
+    if (length == 3 && retVal[0] == EMPTY_SET && retVal[1] == EMPTY_SET && retVal[2] == EMPTY_SET ) {
+        retVal = nullptr;
+        length = 0;
+        *status = U_MISSING_RESOURCE_ERROR;
+    }
+    if (len != nullptr) {
+        *len = length;
+    }
+    return retVal;
+}
+
+/*
+  Like res_getTableItemByKey but accepts full paths like "NumberElements/latn/patternsShort".
+*/  
+static Resource getTableItemByKeyPath(const ResourceData *pResData, Resource table, const char *key) {
+  Resource resource = table;  /* The current resource */
+  icu::CharString path;
+  UErrorCode errorCode = U_ZERO_ERROR;
+  path.append(key, errorCode);
+  if (U_FAILURE(errorCode)) { return RES_BOGUS; }
+  char *pathPart = path.data();  /* Path from current resource to desired resource */
+  UResType type = (UResType)RES_GET_TYPE(resource);  /* the current resource type */
+  while (*pathPart && resource != RES_BOGUS && URES_IS_CONTAINER(type)) {
+    char *nextPathPart = uprv_strchr(pathPart, RES_PATH_SEPARATOR);
+    if (nextPathPart != nullptr) {
+      *nextPathPart = 0;  /* Terminating null for this part of path. */
+      nextPathPart++;
+    } else {
+      nextPathPart = uprv_strchr(pathPart, 0);
+    }
+    int32_t t;
+    const char *pathP = pathPart;
+    resource = res_getTableItemByKey(pResData, resource, &t, &pathP);
+    type = (UResType)RES_GET_TYPE(resource);
+    pathPart = nextPathPart; 
+  }
+  if (*pathPart) {
+    return RES_BOGUS;
+  }
+  return resource;
+}
+
+static void createPath(const char* origResPath,
+                       int32_t     origResPathLen,
+                       const char* resPath,
+                       int32_t     resPathLen,
+                       const char* inKey,
+                       CharString& path,
+                       UErrorCode* status) {
+    // This is a utility function used by ures_getByKeyWithFallback() below.  This function builds a path from
+    // resPath and inKey, returning the result in `path`.  Originally, this function just cleared `path` and
+    // appended resPath and inKey to it, but that caused problems for horizontal inheritance.
+    //
+    // In normal cases, resPath is the same as origResPath, but if ures_getByKeyWithFallback() has followed an
+    // alias, resPath may be different from origResPath.  Not only may the existing path elements be different,
+    // but resPath may also have MORE path elements than origResPath did.  If it does, those additional path
+    // elements SUPERSEDE the corresponding elements of inKey.  So this code counts the number of elements in
+    // resPath and origResPath and, for each path element in resPath that doesn't have a counterpart in origResPath,
+    // deletes a path element from the beginning of inKey.  The remainder of inKey is then appended to
+    // resPath to form the result.  (We're not using uprv_strchr() here because resPath and origResPath may
+    // not be zero-terminated.)
+    path.clear();
+    const char* key = inKey;
+    if (resPathLen > 0) {
+        path.append(resPath, resPathLen, *status);
+        if (U_SUCCESS(*status)) {
+            const char* resPathLimit = resPath + resPathLen;
+            const char* origResPathLimit = origResPath + origResPathLen;
+            const char* resPathPtr = resPath;
+            const char* origResPathPtr = origResPath;
+            
+            // Remove from the beginning of resPath the number of segments that are contained in origResPath.
+            // If origResPath has MORE segments than resPath, this will leave resPath as the empty string.
+            while (origResPathPtr < origResPathLimit && resPathPtr < resPathLimit) {
+                while (origResPathPtr < origResPathLimit && *origResPathPtr != RES_PATH_SEPARATOR) {
+                    ++origResPathPtr;
+                }
+                if (origResPathPtr < origResPathLimit && *origResPathPtr == RES_PATH_SEPARATOR) {
+                    ++origResPathPtr;
+                }
+                while (resPathPtr < resPathLimit && *resPathPtr != RES_PATH_SEPARATOR) {
+                    ++resPathPtr;
+                }
+                if (resPathPtr < resPathLimit && *resPathPtr == RES_PATH_SEPARATOR) {
+                    ++resPathPtr;
+                }
+            }
+            
+            // New remove from the beginning of `key` the number of segments remaining in resPath.
+            // If resPath has more segments than `key` does, `key` will end up empty.
+            while (resPathPtr < resPathLimit && *key != '\0') {
+                while (resPathPtr < resPathLimit && *resPathPtr != RES_PATH_SEPARATOR) {
+                    ++resPathPtr;
+                }
+                if (resPathPtr < resPathLimit && *resPathPtr == RES_PATH_SEPARATOR) {
+                    ++resPathPtr;
+                }
+                while (*key != '\0' && *key != RES_PATH_SEPARATOR) {
+                    ++key;
+                }
+                if (*key == RES_PATH_SEPARATOR) {
+                    ++key;
+                }
+            }
+        }
+        // Finally, append what's left of `key` to `path`.  What you end up with here is `resPath`, plus
+        // any pieces of `key` that aren't superseded by `resPath`.
+        // Or, to put it another way, calculate <#-segments-in-key> - (<#-segments-in-resPath> - <#-segments-in-origResPath>),
+        // and append that many segments from the end of `key` to `resPath` to produce the result.
+        path.append(key, *status);
+    } else {
+        path.append(inKey, *status);
+    }
+}
+
+U_CAPI UResourceBundle* U_EXPORT2
+ures_getByKeyWithFallback(const UResourceBundle *resB,
+                          const char* inKey,
+                          UResourceBundle *fillIn,
+                          UErrorCode *status) {
+    Resource res = RES_BOGUS, rootRes = RES_BOGUS;
+    UResourceBundle *helper = nullptr;
+
+    if (status==nullptr || U_FAILURE(*status)) {
+        return fillIn;
+    }
+    if(resB == nullptr) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return fillIn;
+    }
+
+    int32_t type = RES_GET_TYPE(resB->fRes);
+    if(URES_IS_TABLE(type)) {
+        const char* origResPath = resB->fResPath;
+        int32_t origResPathLen = resB->fResPathLen;
+        res = getTableItemByKeyPath(&resB->getResData(), resB->fRes, inKey);
+        const char* key = inKey;
+        bool didRootOnce = false;
+        if(res == RES_BOGUS) {
+            UResourceDataEntry *dataEntry = resB->fData;
+            CharString path;
+            char *myPath = nullptr;
+            const char* resPath = resB->fResPath;
+            int32_t len = resB->fResPathLen;
+            while(res == RES_BOGUS && (dataEntry->fParent != nullptr || !didRootOnce)) { /* Otherwise, we'll look in parents */
+                if (dataEntry->fParent != nullptr) {
+                    dataEntry = dataEntry->fParent;
+                } else {
+                    // We can't just stop when we get to a bundle whose fParent is nullptr.  That'll work most of the time,
+                    // but if the bundle that the caller passed to us was "root" (which happens in getAllItemsWithFallback(),
+                    // this function will drop right out without doing anything if "root" doesn't contain the exact key path
+                    // specified.  In that case, we need one extra time through this loop to make sure we follow any
+                    // applicable aliases at the root level.
+                    didRootOnce = true;
+                }
+                rootRes = dataEntry->fData.rootRes;
+
+                if(dataEntry->fBogus == U_ZERO_ERROR) {
+                    createPath(origResPath, origResPathLen, resPath, len, inKey, path, status);
+                    if (U_FAILURE(*status)) {
+                        ures_close(helper);
+                        return fillIn;
+                    }
+                    myPath = path.data();
+                    key = inKey;
+                    do {
+                        res = res_findResource(&(dataEntry->fData), rootRes, &myPath, &key);
+                        if (RES_GET_TYPE(res) == URES_ALIAS && *myPath) {
+                            /* We hit an alias, but we didn't finish following the path. */
+                            helper = init_resb_result(dataEntry, res, nullptr, -1, resB, helper, status);
+                            /*helper = init_resb_result(dataEntry, res, inKey, -1, resB, helper, status);*/
+                            if(helper) {
+                              dataEntry = helper->fData;
+                              rootRes = helper->fRes;
+                              resPath = helper->fResPath;
+                              len = helper->fResPathLen;
+
+                            } else {
+                              break;
+                            }
+                        } else if (res == RES_BOGUS) {
+                            break;
+                        }
+                    } while(*myPath); /* Continue until the whole path is consumed */
+                }
+            }
+            /*dataEntry = getFallbackData(resB, &key, &res, status);*/
+            if(res != RES_BOGUS) {
+              /* check if resB->fResPath gives the right name here */
+                if(uprv_strcmp(dataEntry->fName, uloc_getDefault())==0 || uprv_strcmp(dataEntry->fName, kRootLocaleName)==0) {
+                    *status = U_USING_DEFAULT_WARNING;
+                } else {
+                    *status = U_USING_FALLBACK_WARNING;
+                }
+
+                fillIn = init_resb_result(dataEntry, res, key, -1, resB, fillIn, status);
+                if (resPath != nullptr) {
+                    createPath(origResPath, origResPathLen, resPath, len, inKey, path, status);
+                } else {
+                    const char* separator = nullptr;
+                    if (fillIn->fResPath != nullptr) {
+                        separator = uprv_strchr(fillIn->fResPath, RES_PATH_SEPARATOR);
+                    }
+                    if (separator != nullptr && separator[1] != '\0') {
+                        createPath(origResPath, origResPathLen, fillIn->fResPath,
+                                   static_cast(uprv_strlen(fillIn->fResPath)), inKey, path, status);
+                    } else {
+                        createPath(origResPath, origResPathLen, "", 0, inKey, path, status);
+                    }
+                }
+                ures_freeResPath(fillIn);
+                ures_appendResPath(fillIn, path.data(), path.length(), status);
+                if(fillIn->fResPath[fillIn->fResPathLen-1] != RES_PATH_SEPARATOR) {
+                    ures_appendResPath(fillIn, RES_PATH_SEPARATOR_S, 1, status);
+                }
+            } else {
+                *status = U_MISSING_RESOURCE_ERROR;
+            }
+        } else {
+            fillIn = init_resb_result(resB->fData, res, key, -1, resB, fillIn, status);
+        }
+    } 
+    else {
+        *status = U_RESOURCE_TYPE_MISMATCH;
+    }
+    ures_close(helper);
+    return fillIn;
+}
+
+namespace {
+
+void getAllItemsWithFallback(
+        const UResourceBundle *bundle, ResourceDataValue &value,
+        ResourceSink &sink, UErrorCode &errorCode) {
+    if (U_FAILURE(errorCode)) { return; }
+    // We recursively enumerate child-first,
+    // only storing parent items in the absence of child items.
+    // The sink needs to store a placeholder value for the no-fallback/no-inheritance marker
+    // to prevent a parent item from being stored.
+    //
+    // It would be possible to recursively enumerate parent-first,
+    // overriding parent items with child items.
+    // When the sink sees the no-fallback/no-inheritance marker,
+    // then it would remove the parent's item.
+    // We would deserialize parent values even though they are overridden in a child bundle.
+    value.setData(bundle->getResData());
+    value.setValidLocaleDataEntry(bundle->fValidLocaleDataEntry);
+    UResourceDataEntry *parentEntry = bundle->fData->fParent;
+    UBool hasParent = parentEntry != nullptr && U_SUCCESS(parentEntry->fBogus);
+    value.setResource(bundle->fRes, ResourceTracer(bundle));
+    sink.put(bundle->fKey, value, !hasParent, errorCode);
+    if (hasParent) {
+        // We might try to query the sink whether
+        // any fallback from the parent bundle is still possible.
+
+        // Turn the parent UResourceDataEntry into a UResourceBundle,
+        // much like in ures_openWithType().
+        // TODO: See if we can refactor ures_getByKeyWithFallback()
+        // and pull out an inner function that takes and returns a UResourceDataEntry
+        // so that we need not create UResourceBundle objects.
+        StackUResourceBundle parentBundle;
+        UResourceBundle &parentRef = parentBundle.ref();
+        parentRef.fData = parentEntry;
+        parentRef.fValidLocaleDataEntry = bundle->fValidLocaleDataEntry;
+        parentRef.fHasFallback = !parentRef.getResData().noFallback;
+        parentRef.fIsTopLevel = true;
+        parentRef.fRes = parentRef.getResData().rootRes;
+        parentRef.fSize = res_countArrayItems(&parentRef.getResData(), parentRef.fRes);
+        parentRef.fIndex = -1;
+        entryIncrease(parentEntry);
+
+        // Look up the container item in the parent bundle.
+        StackUResourceBundle containerBundle;
+        const UResourceBundle *rb;
+        UErrorCode pathErrorCode = U_ZERO_ERROR;  // Ignore if parents up to root do not have this path.
+        if (bundle->fResPath == nullptr || *bundle->fResPath == 0) {
+            rb = parentBundle.getAlias();
+        } else {
+            rb = ures_getByKeyWithFallback(parentBundle.getAlias(), bundle->fResPath,
+                                           containerBundle.getAlias(), &pathErrorCode);
+        }
+        if (U_SUCCESS(pathErrorCode)) {
+            getAllItemsWithFallback(rb, value, sink, errorCode);
+        }
+    }
+}
+
+struct GetAllChildrenSink : public ResourceSink {
+    // Destination sink
+    ResourceSink& dest;
+
+    GetAllChildrenSink(ResourceSink& dest)
+        : dest(dest) {}
+    virtual ~GetAllChildrenSink() override;
+    virtual void put(const char *key, ResourceValue &value, UBool isRoot,
+           UErrorCode &errorCode) override {
+        ResourceTable itemsTable = value.getTable(errorCode);
+        if (U_FAILURE(errorCode)) { return; }
+        for (int32_t i = 0; itemsTable.getKeyAndValue(i, key, value); ++i) {
+            if (value.getType() == URES_ALIAS) {
+                ResourceDataValue& rdv = static_cast(value);
+                StackUResourceBundle stackTempBundle;
+                UResourceBundle* aliasRB = getAliasTargetAsResourceBundle(rdv.getData(), rdv.getResource(), nullptr, -1,
+                                                                          rdv.getValidLocaleDataEntry(), nullptr, 0,
+                                                                          stackTempBundle.getAlias(), &errorCode);
+                if (U_SUCCESS(errorCode)) {
+                    ResourceDataValue aliasedValue;
+                    aliasedValue.setData(aliasRB->getResData());
+                    aliasedValue.setValidLocaleDataEntry(aliasRB->fValidLocaleDataEntry);
+                    aliasedValue.setResource(aliasRB->fRes, ResourceTracer(aliasRB));
+                    dest.put(key, aliasedValue, isRoot, errorCode);
+                }
+            } else {
+                dest.put(key, value, isRoot, errorCode);
+            }
+            if (U_FAILURE(errorCode)) { return; }
+        }
+    }
+};
+
+// Virtual destructors must be defined out of line.
+GetAllChildrenSink::~GetAllChildrenSink() {}
+
+U_CAPI void U_EXPORT2
+ures_getAllChildrenWithFallback(const UResourceBundle *bundle, const char *path,
+                                icu::ResourceSink &sink, UErrorCode &errorCode) {
+    GetAllChildrenSink allChildrenSink(sink);
+    ures_getAllItemsWithFallback(bundle, path, allChildrenSink, errorCode);
+}
+
+}  // namespace
+
+// Requires a ResourceDataValue fill-in, so that we need not cast from a ResourceValue.
+// Unfortunately, the caller must know which subclass to make and pass in.
+// Alternatively, we could make it as polymorphic as in Java by
+// returning a ResourceValue pointer (possibly wrapped into a LocalPointer)
+// that the caller then owns.
+//
+// Also requires a UResourceBundle fill-in, so that the value's ResourceTracer
+// can point to a non-local bundle.
+// Without tracing, the child bundle could be a function-local object.
+U_CAPI void U_EXPORT2
+ures_getValueWithFallback(const UResourceBundle *bundle, const char *path,
+                          UResourceBundle *tempFillIn,
+                          ResourceDataValue &value, UErrorCode &errorCode) {
+    if (U_FAILURE(errorCode)) { return; }
+    if (path == nullptr) {
+        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+    const UResourceBundle *rb;
+    if (*path == 0) {
+        // empty path
+        rb = bundle;
+    } else {
+        rb = ures_getByKeyWithFallback(bundle, path, tempFillIn, &errorCode);
+        if (U_FAILURE(errorCode)) {
+            return;
+        }
+    }
+    value.setData(rb->getResData());
+    value.setValidLocaleDataEntry(rb->fValidLocaleDataEntry);
+    value.setResource(rb->fRes, ResourceTracer(rb));
+}
+
+U_CAPI void U_EXPORT2
+ures_getAllItemsWithFallback(const UResourceBundle *bundle, const char *path,
+                             icu::ResourceSink &sink, UErrorCode &errorCode) {
+    if (U_FAILURE(errorCode)) { return; }
+    if (path == nullptr) {
+        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+    StackUResourceBundle stackBundle;
+    const UResourceBundle *rb;
+    if (*path == 0) {
+        // empty path
+        rb = bundle;
+    } else {
+        rb = ures_getByKeyWithFallback(bundle, path, stackBundle.getAlias(), &errorCode);
+        if (U_FAILURE(errorCode)) {
+            return;
+        }
+    }
+    // Get all table items with fallback.
+    ResourceDataValue value;
+    getAllItemsWithFallback(rb, value, sink, errorCode);
+}
+
+U_CAPI UResourceBundle* U_EXPORT2 ures_getByKey(const UResourceBundle *resB, const char* inKey, UResourceBundle *fillIn, UErrorCode *status) {
+    Resource res = RES_BOGUS;
+    UResourceDataEntry *dataEntry = nullptr;
+    const char *key = inKey;
+
+    if (status==nullptr || U_FAILURE(*status)) {
+        return fillIn;
+    }
+    if(resB == nullptr) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return fillIn;
+    }
+
+    int32_t type = RES_GET_TYPE(resB->fRes);
+    if(URES_IS_TABLE(type)) {
+        int32_t t;
+        res = res_getTableItemByKey(&resB->getResData(), resB->fRes, &t, &key);
+        if(res == RES_BOGUS) {
+            key = inKey;
+            if(resB->fHasFallback) {
+                dataEntry = getFallbackData(resB, &key, &res, status);
+                if(U_SUCCESS(*status)) {
+                    /* check if resB->fResPath gives the right name here */
+                    return init_resb_result(dataEntry, res, key, -1, resB, fillIn, status);
+                } else {
+                    *status = U_MISSING_RESOURCE_ERROR;
+                }
+            } else {
+                *status = U_MISSING_RESOURCE_ERROR;
+            }
+        } else {
+            return init_resb_result(resB->fData, res, key, -1, resB, fillIn, status);
+        }
+    } 
+#if 0
+    /* this is a kind of TODO item. If we have an array with an index table, we could do this. */
+    /* not currently */
+    else if(RES_GET_TYPE(resB->fRes) == URES_ARRAY && resB->fHasFallback == true) {
+        /* here should go a first attempt to locate the key using index table */
+        dataEntry = getFallbackData(resB, &key, &res, status);
+        if(U_SUCCESS(*status)) {
+            return init_resb_result(dataEntry, res, key, resB, fillIn, status);
+        } else {
+            *status = U_MISSING_RESOURCE_ERROR;
+        }
+    }
+#endif    
+    else {
+        *status = U_RESOURCE_TYPE_MISMATCH;
+    }
+    return fillIn;
+}
+
+U_CAPI const char16_t* U_EXPORT2 ures_getStringByKey(const UResourceBundle *resB, const char* inKey, int32_t* len, UErrorCode *status) {
+    Resource res = RES_BOGUS;
+    UResourceDataEntry *dataEntry = nullptr;
+    const char* key = inKey;
+
+    if (status==nullptr || U_FAILURE(*status)) {
+        return nullptr;
+    }
+    if(resB == nullptr) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+
+    int32_t type = RES_GET_TYPE(resB->fRes);
+    if(URES_IS_TABLE(type)) {
+        int32_t t=0;
+
+        res = res_getTableItemByKey(&resB->getResData(), resB->fRes, &t, &key);
+
+        if(res == RES_BOGUS) {
+            key = inKey;
+            if(resB->fHasFallback) {
+                dataEntry = getFallbackData(resB, &key, &res, status);
+                if(U_SUCCESS(*status)) {
+                    switch (RES_GET_TYPE(res)) {
+                    case URES_STRING:
+                    case URES_STRING_V2:
+                        return res_getString({resB, key}, &dataEntry->fData, res, len);
+                    case URES_ALIAS:
+                      {
+                        const char16_t* result = 0;
+                        UResourceBundle *tempRes = ures_getByKey(resB, inKey, nullptr, status);
+                        result = ures_getString(tempRes, len, status);
+                        ures_close(tempRes);
+                        return result;
+                      }
+                    default:
+                        *status = U_RESOURCE_TYPE_MISMATCH;
+                    }
+                } else {
+                    *status = U_MISSING_RESOURCE_ERROR;
+                }
+            } else {
+                *status = U_MISSING_RESOURCE_ERROR;
+            }
+        } else {
+            switch (RES_GET_TYPE(res)) {
+            case URES_STRING:
+            case URES_STRING_V2:
+                return res_getString({resB, key}, &resB->getResData(), res, len);
+            case URES_ALIAS:
+              {
+                const char16_t* result = 0;
+                UResourceBundle *tempRes = ures_getByKey(resB, inKey, nullptr, status);
+                result = ures_getString(tempRes, len, status);
+                ures_close(tempRes);
+                return result;
+              }
+            default:
+                *status = U_RESOURCE_TYPE_MISMATCH;
+            }
+        }
+    } 
+#if 0 
+    /* this is a kind of TODO item. If we have an array with an index table, we could do this. */
+    /* not currently */   
+    else if(RES_GET_TYPE(resB->fRes) == URES_ARRAY && resB->fHasFallback == true) {
+        /* here should go a first attempt to locate the key using index table */
+        dataEntry = getFallbackData(resB, &key, &res, status);
+        if(U_SUCCESS(*status)) {
+            // TODO: Tracing
+            return res_getString(rd, res, len);
+        } else {
+            *status = U_MISSING_RESOURCE_ERROR;
+        }
+    } 
+#endif    
+    else {
+        *status = U_RESOURCE_TYPE_MISMATCH;
+    }
+    return nullptr;
+}
+
+U_CAPI const char * U_EXPORT2
+ures_getUTF8StringByKey(const UResourceBundle *resB,
+                        const char *key,
+                        char *dest, int32_t *pLength,
+                        UBool forceCopy,
+                        UErrorCode *status) {
+    int32_t length16;
+    const char16_t *s16 = ures_getStringByKey(resB, key, &length16, status);
+    return ures_toUTF8String(s16, length16, dest, pLength, forceCopy, status);
+}
+
+/* TODO: clean from here down */
+
+/**
+ *  INTERNAL: Get the name of the first real locale (not placeholder) 
+ *  that has resource bundle data.
+ */
+U_CAPI const char*  U_EXPORT2
+ures_getLocaleInternal(const UResourceBundle* resourceBundle, UErrorCode* status)
+{
+    if (status==nullptr || U_FAILURE(*status)) {
+        return nullptr;
+    }
+    if (!resourceBundle) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    } else {
+      return resourceBundle->fData->fName;
+    }
+}
+
+U_CAPI const char* U_EXPORT2 
+ures_getLocale(const UResourceBundle* resourceBundle, 
+               UErrorCode* status)
+{
+  return ures_getLocaleInternal(resourceBundle, status);
+}
+
+
+U_CAPI const char* U_EXPORT2 
+ures_getLocaleByType(const UResourceBundle* resourceBundle, 
+                     ULocDataLocaleType type, 
+                     UErrorCode* status) {
+    if (status==nullptr || U_FAILURE(*status)) {
+        return nullptr;
+    }
+    if (!resourceBundle) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    } else {
+        switch(type) {
+        case ULOC_ACTUAL_LOCALE:
+            return resourceBundle->fData->fName;
+        case ULOC_VALID_LOCALE:
+            return resourceBundle->fValidLocaleDataEntry->fName;
+        case ULOC_REQUESTED_LOCALE:
+        default:
+            *status = U_ILLEGAL_ARGUMENT_ERROR;
+            return nullptr;
+        }
+    }
+}
+
+U_CFUNC const char* ures_getName(const UResourceBundle* resB) {
+  if(resB == nullptr) {
+    return nullptr;
+  }
+
+  return resB->fData->fName;
+}
+
+#ifdef URES_DEBUG
+U_CFUNC const char* ures_getPath(const UResourceBundle* resB) {
+  if(resB == nullptr) {
+    return nullptr;
+  }
+
+  return resB->fData->fPath;
+}
+#endif
+
+static UResourceBundle*
+ures_openWithType(UResourceBundle *r, const char* path, const char* localeID,
+                  UResOpenType openType, UErrorCode* status) {
+    if(U_FAILURE(*status)) {
+        return nullptr;
+    }
+
+    UResourceDataEntry *entry;
+    if(openType != URES_OPEN_DIRECT) {
+        /* first "canonicalize" the locale ID */
+        char canonLocaleID[ULOC_FULLNAME_CAPACITY];
+        uloc_getBaseName(localeID, canonLocaleID, UPRV_LENGTHOF(canonLocaleID), status);
+        if(U_FAILURE(*status) || *status == U_STRING_NOT_TERMINATED_WARNING) {
+            *status = U_ILLEGAL_ARGUMENT_ERROR;
+            return nullptr;
+        }
+        entry = entryOpen(path, canonLocaleID, openType, status);
+    } else {
+        entry = entryOpenDirect(path, localeID, status);
+    }
+    if(U_FAILURE(*status)) {
+        return nullptr;
+    }
+    if(entry == nullptr) {
+        *status = U_MISSING_RESOURCE_ERROR;
+        return nullptr;
+    }
+
+    UBool isStackObject;
+    if(r == nullptr) {
+        r = (UResourceBundle *)uprv_malloc(sizeof(UResourceBundle));
+        if(r == nullptr) {
+            entryClose(entry);
+            *status = U_MEMORY_ALLOCATION_ERROR;
+            return nullptr;
+        }
+        isStackObject = false;
+    } else {  // fill-in
+        isStackObject = ures_isStackObject(r);
+        ures_closeBundle(r, false);
+    }
+    uprv_memset(r, 0, sizeof(UResourceBundle));
+    ures_setIsStackObject(r, isStackObject);
+
+    r->fValidLocaleDataEntry = r->fData = entry;
+    r->fHasFallback = openType != URES_OPEN_DIRECT && !r->getResData().noFallback;
+    r->fIsTopLevel = true;
+    r->fRes = r->getResData().rootRes;
+    r->fSize = res_countArrayItems(&r->getResData(), r->fRes);
+    r->fIndex = -1;
+
+    ResourceTracer(r).traceOpen();
+
+    return r;
+}
+
+U_CAPI UResourceBundle* U_EXPORT2
+ures_open(const char* path, const char* localeID, UErrorCode* status) {
+    return ures_openWithType(nullptr, path, localeID, URES_OPEN_LOCALE_DEFAULT_ROOT, status);
+}
+
+U_CAPI UResourceBundle* U_EXPORT2
+ures_openNoDefault(const char* path, const char* localeID, UErrorCode* status) {
+    return ures_openWithType(nullptr, path, localeID, URES_OPEN_LOCALE_ROOT, status);
+}
+
+/**
+ *  Opens a resource bundle without "canonicalizing" the locale name. No fallback will be performed 
+ *  or sought. However, alias substitution will happen!
+ */
+U_CAPI UResourceBundle*  U_EXPORT2
+ures_openDirect(const char* path, const char* localeID, UErrorCode* status) {
+    return ures_openWithType(nullptr, path, localeID, URES_OPEN_DIRECT, status);
+}
+
+/**
+ *  Internal API: This function is used to open a resource bundle 
+ *  proper fallback chaining is executed while initialization. 
+ *  The result is stored in cache for later fallback search.
+ * 
+ * Same as ures_open(), but uses the fill-in parameter and does not allocate a new bundle.
+ */
+U_CAPI void U_EXPORT2
+ures_openFillIn(UResourceBundle *r, const char* path,
+                const char* localeID, UErrorCode* status) {
+    if(U_SUCCESS(*status) && r == nullptr) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+    ures_openWithType(r, path, localeID, URES_OPEN_LOCALE_DEFAULT_ROOT, status);
+}
+
+/**
+ * Same as ures_openDirect(), but uses the fill-in parameter and does not allocate a new bundle.
+ */
+U_CAPI void U_EXPORT2
+ures_openDirectFillIn(UResourceBundle *r, const char* path, const char* localeID, UErrorCode* status) {
+    if(U_SUCCESS(*status) && r == nullptr) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+    ures_openWithType(r, path, localeID, URES_OPEN_DIRECT, status);
+}
+
+/**
+ *  API: Counts members. For arrays and tables, returns number of resources.
+ *  For strings, returns 1.
+ */
+U_CAPI int32_t  U_EXPORT2
+ures_countArrayItems(const UResourceBundle* resourceBundle,
+                  const char* resourceKey,
+                  UErrorCode* status)
+{
+    UResourceBundle resData;
+    ures_initStackObject(&resData);
+    if (status==nullptr || U_FAILURE(*status)) {
+        return 0;
+    }
+    if(resourceBundle == nullptr) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    ures_getByKey(resourceBundle, resourceKey, &resData, status);
+    
+    if(resData.getResData().data != nullptr) {
+        int32_t result = res_countArrayItems(&resData.getResData(), resData.fRes);
+        ures_close(&resData);
+        return result;
+    } else {
+        *status = U_MISSING_RESOURCE_ERROR;
+        ures_close(&resData);
+        return 0;
+    }
+}
+
+/**
+ * Internal function.
+ * Return the version number associated with this ResourceBundle as a string.
+ *
+ * @param resourceBundle The resource bundle for which the version is checked.
+ * @return  A version number string as specified in the resource bundle or its parent.
+ *          The caller does not own this string.
+ * @see ures_getVersion
+ * @internal
+ */
+U_CAPI const char* U_EXPORT2 
+ures_getVersionNumberInternal(const UResourceBundle *resourceBundle)
+{
+    if (!resourceBundle) return nullptr;
+
+    if(resourceBundle->fVersion == nullptr) {
+
+        /* If the version ID has not been built yet, then do so.  Retrieve */
+        /* the minor version from the file. */
+        UErrorCode status = U_ZERO_ERROR;
+        int32_t minor_len = 0;
+        int32_t len;
+
+        const char16_t* minor_version = ures_getStringByKey(resourceBundle, kVersionTag, &minor_len, &status);
+        
+        /* Determine the length of of the final version string.  This is */
+        /* the length of the major part + the length of the separator */
+        /* (==1) + the length of the minor part (+ 1 for the zero byte at */
+        /* the end). */
+
+        len = (minor_len > 0) ? minor_len : 1;
+    
+        /* Allocate the string, and build it up. */
+        /* + 1 for zero byte */
+
+
+        ((UResourceBundle *)resourceBundle)->fVersion = (char *)uprv_malloc(1 + len); 
+        /* Check for null pointer. */
+        if (((UResourceBundle *)resourceBundle)->fVersion == nullptr) {
+            return nullptr;
+        }
+       
+        if(minor_len > 0) {
+            u_UCharsToChars(minor_version, resourceBundle->fVersion , minor_len);
+            resourceBundle->fVersion[len] =  '\0';
+        }
+        else {
+            uprv_strcpy(resourceBundle->fVersion, kDefaultMinorVersion);
+        }
+    }
+
+    return resourceBundle->fVersion;
+}
+
+U_CAPI const char*  U_EXPORT2
+ures_getVersionNumber(const UResourceBundle*   resourceBundle)
+{
+    return ures_getVersionNumberInternal(resourceBundle);
+}
+
+U_CAPI void U_EXPORT2 ures_getVersion(const UResourceBundle* resB, UVersionInfo versionInfo) {
+    if (!resB) return;
+
+    u_versionFromString(versionInfo, ures_getVersionNumberInternal(resB));
+}
+
+/** Tree support functions *******************************/
+#define INDEX_LOCALE_NAME "res_index"
+#define INDEX_TAG         "InstalledLocales"
+#define DEFAULT_TAG       "default"
+
+#if defined(URES_TREE_DEBUG)
+#include 
+#endif
+
+typedef struct ULocalesContext {
+    UResourceBundle installed;
+    UResourceBundle curr;
+} ULocalesContext;
+
+static void U_CALLCONV
+ures_loc_closeLocales(UEnumeration *enumerator) {
+    ULocalesContext *ctx = (ULocalesContext *)enumerator->context;
+    ures_close(&ctx->curr);
+    ures_close(&ctx->installed);
+    uprv_free(ctx);
+    uprv_free(enumerator);
+}
+
+static int32_t U_CALLCONV
+ures_loc_countLocales(UEnumeration *en, UErrorCode * /*status*/) {
+    ULocalesContext *ctx = (ULocalesContext *)en->context;
+    return ures_getSize(&ctx->installed);
+}
+
+U_CDECL_BEGIN
+
+
+static const char * U_CALLCONV
+ures_loc_nextLocale(UEnumeration* en,
+                    int32_t* resultLength,
+                    UErrorCode* status) {
+    ULocalesContext *ctx = (ULocalesContext *)en->context;
+    UResourceBundle *res = &(ctx->installed);
+    UResourceBundle *k = nullptr;
+    const char *result = nullptr;
+    int32_t len = 0;
+    if(ures_hasNext(res) && (k = ures_getNextResource(res, &ctx->curr, status)) != 0) {
+        result = ures_getKey(k);
+        len = (int32_t)uprv_strlen(result);
+    }
+    if (resultLength) {
+        *resultLength = len;
+    }
+    return result;
+}
+
+static void U_CALLCONV 
+ures_loc_resetLocales(UEnumeration* en, 
+                      UErrorCode* /*status*/) {
+    UResourceBundle *res = &((ULocalesContext *)en->context)->installed;
+    ures_resetIterator(res);
+}
+
+U_CDECL_END
+
+static const UEnumeration gLocalesEnum = {
+    nullptr,
+        nullptr,
+        ures_loc_closeLocales,
+        ures_loc_countLocales,
+        uenum_unextDefault,
+        ures_loc_nextLocale,
+        ures_loc_resetLocales
+};
+
+
+U_CAPI UEnumeration* U_EXPORT2
+ures_openAvailableLocales(const char *path, UErrorCode *status)
+{
+    UResourceBundle *idx = nullptr;
+    UEnumeration *en = nullptr;
+    ULocalesContext *myContext = nullptr;
+
+    if(U_FAILURE(*status)) {
+        return nullptr;
+    }
+    myContext = static_cast(uprv_malloc(sizeof(ULocalesContext)));
+    en =  (UEnumeration *)uprv_malloc(sizeof(UEnumeration));
+    if(!en || !myContext) {
+        *status = U_MEMORY_ALLOCATION_ERROR;
+        uprv_free(en);
+        uprv_free(myContext);
+        return nullptr;
+    }
+    uprv_memcpy(en, &gLocalesEnum, sizeof(UEnumeration));
+
+    ures_initStackObject(&myContext->installed);
+    ures_initStackObject(&myContext->curr);
+    idx = ures_openDirect(path, INDEX_LOCALE_NAME, status);
+    ures_getByKey(idx, INDEX_TAG, &myContext->installed, status);
+    if(U_SUCCESS(*status)) {
+#if defined(URES_TREE_DEBUG)
+        fprintf(stderr, "Got %s::%s::[%s] : %s\n", 
+            path, INDEX_LOCALE_NAME, INDEX_TAG, ures_getKey(&myContext->installed));
+#endif
+        en->context = myContext;
+    } else {
+#if defined(URES_TREE_DEBUG)
+        fprintf(stderr, "%s open failed - %s\n", path, u_errorName(*status));
+#endif
+        ures_close(&myContext->installed);
+        uprv_free(myContext);
+        uprv_free(en);
+        en = nullptr;
+    }
+    
+    ures_close(idx);
+    
+    return en;
+}
+
+static UBool isLocaleInList(UEnumeration *locEnum, const char *locToSearch, UErrorCode *status) {
+    const char *loc;
+    while ((loc = uenum_next(locEnum, nullptr, status)) != nullptr) {
+        if (uprv_strcmp(loc, locToSearch) == 0) {
+            return true;
+        }
+    }
+    return false;
+}
+
+U_CAPI int32_t U_EXPORT2
+ures_getFunctionalEquivalent(char *result, int32_t resultCapacity,
+                             const char *path, const char *resName, const char *keyword, const char *locid,
+                             UBool *isAvailable, UBool omitDefault, UErrorCode *status)
+{
+    char kwVal[1024] = ""; /* value of keyword 'keyword' */
+    char defVal[1024] = ""; /* default value for given locale */
+    char defLoc[1024] = ""; /* default value for given locale */
+    char base[1024] = ""; /* base locale */
+    char found[1024] = "";
+    char parent[1024] = "";
+    char full[1024] = "";
+    UResourceBundle bund1, bund2;
+    UResourceBundle *res = nullptr;
+    UErrorCode subStatus = U_ZERO_ERROR;
+    int32_t length = 0;
+    if(U_FAILURE(*status)) return 0;
+    uloc_getKeywordValue(locid, keyword, kwVal, 1024-1,&subStatus);
+    if(!uprv_strcmp(kwVal, DEFAULT_TAG)) {
+        kwVal[0]=0;
+    }
+    uloc_getBaseName(locid, base, 1024-1,&subStatus);
+#if defined(URES_TREE_DEBUG)
+    fprintf(stderr, "getFunctionalEquivalent: \"%s\" [%s=%s] in %s - %s\n", 
+            locid, keyword, kwVal, base, u_errorName(subStatus));
+#endif
+    ures_initStackObject(&bund1);
+    ures_initStackObject(&bund2);
+    
+    
+    uprv_strcpy(parent, base);
+    uprv_strcpy(found, base);
+
+    if(isAvailable) { 
+        UEnumeration *locEnum = ures_openAvailableLocales(path, &subStatus);
+        *isAvailable = true;
+        if (U_SUCCESS(subStatus)) {
+            *isAvailable = isLocaleInList(locEnum, parent, &subStatus);
+        }
+        uenum_close(locEnum);
+    }
+
+    if(U_FAILURE(subStatus)) {
+        *status = subStatus;
+        return 0;
+    }
+    
+    do {
+        subStatus = U_ZERO_ERROR;
+        res = ures_open(path, parent, &subStatus);
+        if(((subStatus == U_USING_FALLBACK_WARNING) ||
+            (subStatus == U_USING_DEFAULT_WARNING)) && isAvailable)
+        {
+            *isAvailable = false;
+        }
+        isAvailable = nullptr; /* only want to set this the first time around */
+        
+#if defined(URES_TREE_DEBUG)
+        fprintf(stderr, "%s;%s -> %s [%s]\n", path?path:"ICUDATA", parent, u_errorName(subStatus), ures_getLocale(res, &subStatus));
+#endif
+        if(U_FAILURE(subStatus)) {
+            *status = subStatus;
+        } else if(subStatus == U_ZERO_ERROR) {
+            ures_getByKey(res,resName,&bund1, &subStatus);
+            if(subStatus == U_ZERO_ERROR) {
+                const char16_t *defUstr;
+                int32_t defLen;
+                /* look for default item */
+#if defined(URES_TREE_DEBUG)
+                fprintf(stderr, "%s;%s : loaded default -> %s\n",
+                    path?path:"ICUDATA", parent, u_errorName(subStatus));
+#endif
+                defUstr = ures_getStringByKey(&bund1, DEFAULT_TAG, &defLen, &subStatus);
+                if(U_SUCCESS(subStatus) && defLen) {
+                    u_UCharsToChars(defUstr, defVal, u_strlen(defUstr));
+#if defined(URES_TREE_DEBUG)
+                    fprintf(stderr, "%s;%s -> default %s=%s,  %s\n", 
+                        path?path:"ICUDATA", parent, keyword, defVal, u_errorName(subStatus));
+#endif
+                    uprv_strcpy(defLoc, parent);
+                    if(kwVal[0]==0) {
+                        uprv_strcpy(kwVal, defVal);
+#if defined(URES_TREE_DEBUG)
+                        fprintf(stderr, "%s;%s -> kwVal =  %s\n", 
+                            path?path:"ICUDATA", parent, keyword, kwVal);
+#endif
+                    }
+                }
+            }
+        }
+        
+        subStatus = U_ZERO_ERROR;
+
+        if (res != nullptr) {
+            uprv_strcpy(found, ures_getLocaleByType(res, ULOC_VALID_LOCALE, &subStatus));
+        }
+
+        uloc_getParent(found,parent,sizeof(parent),&subStatus);
+        ures_close(res);
+    } while(!defVal[0] && *found && uprv_strcmp(found, "root") != 0 && U_SUCCESS(*status));
+    
+    /* Now, see if we can find the kwVal collator.. start the search over.. */
+    uprv_strcpy(parent, base);
+    uprv_strcpy(found, base);
+    
+    do {
+        subStatus = U_ZERO_ERROR;
+        res = ures_open(path, parent, &subStatus);
+        if((subStatus == U_USING_FALLBACK_WARNING) && isAvailable) {
+            *isAvailable = false;
+        }
+        isAvailable = nullptr; /* only want to set this the first time around */
+        
+#if defined(URES_TREE_DEBUG)
+        fprintf(stderr, "%s;%s -> %s (looking for %s)\n", 
+            path?path:"ICUDATA", parent, u_errorName(subStatus), kwVal);
+#endif
+        if(U_FAILURE(subStatus)) {
+            *status = subStatus;
+        } else if(subStatus == U_ZERO_ERROR) {
+            ures_getByKey(res,resName,&bund1, &subStatus);
+#if defined(URES_TREE_DEBUG)
+/**/ fprintf(stderr,"@%d [%s] %s\n", __LINE__, resName, u_errorName(subStatus));
+#endif
+            if(subStatus == U_ZERO_ERROR) {
+                ures_getByKey(&bund1, kwVal, &bund2, &subStatus);
+#if defined(URES_TREE_DEBUG)
+/**/ fprintf(stderr,"@%d [%s] %s\n", __LINE__, kwVal, u_errorName(subStatus));
+#endif
+                if(subStatus == U_ZERO_ERROR) {
+#if defined(URES_TREE_DEBUG)
+                    fprintf(stderr, "%s;%s -> full0 %s=%s,  %s\n", 
+                        path?path:"ICUDATA", parent, keyword, kwVal, u_errorName(subStatus));
+#endif
+                    uprv_strcpy(full, parent);
+                    if(*full == 0) {
+                        uprv_strcpy(full, "root");
+                    }
+                        /* now, recalculate default kw if need be */
+                        if(uprv_strlen(defLoc) > uprv_strlen(full)) {
+                          const char16_t *defUstr;
+                          int32_t defLen;
+                          /* look for default item */
+#if defined(URES_TREE_DEBUG)
+                            fprintf(stderr, "%s;%s -> recalculating Default0\n", 
+                                    path?path:"ICUDATA", full);
+#endif
+                          defUstr = ures_getStringByKey(&bund1, DEFAULT_TAG, &defLen, &subStatus);
+                          if(U_SUCCESS(subStatus) && defLen) {
+                            u_UCharsToChars(defUstr, defVal, u_strlen(defUstr));
+#if defined(URES_TREE_DEBUG)
+                            fprintf(stderr, "%s;%s -> default0 %s=%s,  %s\n", 
+                                    path?path:"ICUDATA", full, keyword, defVal, u_errorName(subStatus));
+#endif
+                            uprv_strcpy(defLoc, full);
+                          }
+                        } /* end of recalculate default KW */
+#if defined(URES_TREE_DEBUG)
+                        else {
+                          fprintf(stderr, "No trim0,  %s <= %s\n", defLoc, full);
+                        }
+#endif
+                } else {
+#if defined(URES_TREE_DEBUG)
+                    fprintf(stderr, "err=%s in %s looking for %s\n", 
+                        u_errorName(subStatus), parent, kwVal);
+#endif
+                }
+            }
+        }
+        
+        subStatus = U_ZERO_ERROR;
+        
+        uprv_strcpy(found, parent);
+        uloc_getParent(found,parent,1023,&subStatus);
+        ures_close(res);
+    } while(!full[0] && *found && U_SUCCESS(*status));
+    
+    if((full[0]==0) && uprv_strcmp(kwVal, defVal)) {
+#if defined(URES_TREE_DEBUG)
+        fprintf(stderr, "Failed to locate kw %s - try default %s\n", kwVal, defVal);
+#endif
+        uprv_strcpy(kwVal, defVal);
+        uprv_strcpy(parent, base);
+        uprv_strcpy(found, base);
+        
+        do { /* search for 'default' named item */
+            subStatus = U_ZERO_ERROR;
+            res = ures_open(path, parent, &subStatus);
+            if((subStatus == U_USING_FALLBACK_WARNING) && isAvailable) {
+                *isAvailable = false;
+            }
+            isAvailable = nullptr; /* only want to set this the first time around */
+            
+#if defined(URES_TREE_DEBUG)
+            fprintf(stderr, "%s;%s -> %s (looking for default %s)\n",
+                path?path:"ICUDATA", parent, u_errorName(subStatus), kwVal);
+#endif
+            if(U_FAILURE(subStatus)) {
+                *status = subStatus;
+            } else if(subStatus == U_ZERO_ERROR) {
+                ures_getByKey(res,resName,&bund1, &subStatus);
+                if(subStatus == U_ZERO_ERROR) {
+                    ures_getByKey(&bund1, kwVal, &bund2, &subStatus);
+                    if(subStatus == U_ZERO_ERROR) {
+#if defined(URES_TREE_DEBUG)
+                        fprintf(stderr, "%s;%s -> full1 %s=%s,  %s\n", path?path:"ICUDATA",
+                            parent, keyword, kwVal, u_errorName(subStatus));
+#endif
+                        uprv_strcpy(full, parent);
+                        if(*full == 0) {
+                            uprv_strcpy(full, "root");
+                        }
+                        
+                        /* now, recalculate default kw if need be */
+                        if(uprv_strlen(defLoc) > uprv_strlen(full)) {
+                          const char16_t *defUstr;
+                          int32_t defLen;
+                          /* look for default item */
+#if defined(URES_TREE_DEBUG)
+                            fprintf(stderr, "%s;%s -> recalculating Default1\n", 
+                                    path?path:"ICUDATA", full);
+#endif
+                          defUstr = ures_getStringByKey(&bund1, DEFAULT_TAG, &defLen, &subStatus);
+                          if(U_SUCCESS(subStatus) && defLen) {
+                            u_UCharsToChars(defUstr, defVal, u_strlen(defUstr));
+#if defined(URES_TREE_DEBUG)
+                            fprintf(stderr, "%s;%s -> default %s=%s,  %s\n", 
+                                    path?path:"ICUDATA", full, keyword, defVal, u_errorName(subStatus));
+#endif
+                            uprv_strcpy(defLoc, full);
+                          }
+                        } /* end of recalculate default KW */
+#if defined(URES_TREE_DEBUG)
+                        else {
+                          fprintf(stderr, "No trim1,  %s <= %s\n", defLoc, full);
+                        }
+#endif
+                    }
+                }
+            }
+            subStatus = U_ZERO_ERROR;
+            
+            uprv_strcpy(found, parent);
+            uloc_getParent(found,parent,1023,&subStatus);
+            ures_close(res);
+        } while(!full[0] && *found && U_SUCCESS(*status));
+    }
+    
+    if(U_SUCCESS(*status)) {
+        if(!full[0]) {
+#if defined(URES_TREE_DEBUG)
+          fprintf(stderr, "Still could not load keyword %s=%s\n", keyword, kwVal);
+#endif
+          *status = U_MISSING_RESOURCE_ERROR;
+        } else if(omitDefault) {
+#if defined(URES_TREE_DEBUG)
+          fprintf(stderr,"Trim? full=%s, defLoc=%s, found=%s\n", full, defLoc, found);
+#endif        
+          if(uprv_strlen(defLoc) <= uprv_strlen(full)) {
+            /* found the keyword in a *child* of where the default tag was present. */
+            if(!uprv_strcmp(kwVal, defVal)) { /* if the requested kw is default, */
+              /* and the default is in or in an ancestor of the current locale */
+#if defined(URES_TREE_DEBUG)
+              fprintf(stderr, "Removing unneeded var %s=%s\n", keyword, kwVal);
+#endif
+              kwVal[0]=0;
+            }
+          }
+        }
+        uprv_strcpy(found, full);
+        if(kwVal[0]) {
+            uprv_strcat(found, "@");
+            uprv_strcat(found, keyword);
+            uprv_strcat(found, "=");
+            uprv_strcat(found, kwVal);
+        } else if(!omitDefault) {
+            uprv_strcat(found, "@");
+            uprv_strcat(found, keyword);
+            uprv_strcat(found, "=");
+            uprv_strcat(found, defVal);
+        }
+    }
+    /* we found the default locale - no need to repeat it.*/
+    
+    ures_close(&bund1);
+    ures_close(&bund2);
+    
+    length = (int32_t)uprv_strlen(found);
+
+    if(U_SUCCESS(*status)) {
+        int32_t copyLength = uprv_min(length, resultCapacity);
+        if(copyLength>0) {
+            uprv_strncpy(result, found, copyLength);
+        }
+        if(length == 0) {
+          *status = U_MISSING_RESOURCE_ERROR; 
+        }
+    } else {
+        length = 0;
+        result[0]=0;
+    }
+    return u_terminateChars(result, resultCapacity, length, status);
+}
+
+U_CAPI UEnumeration* U_EXPORT2
+ures_getKeywordValues(const char *path, const char *keyword, UErrorCode *status)
+{
+#define VALUES_BUF_SIZE 2048
+#define VALUES_LIST_SIZE 512
+    
+    char       valuesBuf[VALUES_BUF_SIZE];
+    int32_t    valuesIndex = 0;
+    const char *valuesList[VALUES_LIST_SIZE];
+    int32_t    valuesCount = 0;
+    
+    const char *locale;
+    int32_t     locLen;
+    
+    UEnumeration *locs = nullptr;
+    
+    UResourceBundle    item;
+    UResourceBundle    subItem;
+    
+    ures_initStackObject(&item);
+    ures_initStackObject(&subItem);
+    locs = ures_openAvailableLocales(path, status);
+    
+    if(U_FAILURE(*status)) {
+        ures_close(&item);
+        ures_close(&subItem);
+        return nullptr;
+    }
+    
+    valuesBuf[0]=0;
+    valuesBuf[1]=0;
+    
+    while((locale = uenum_next(locs, &locLen, status)) != 0) {
+        UResourceBundle   *bund = nullptr;
+        UResourceBundle   *subPtr = nullptr;
+        UErrorCode subStatus = U_ZERO_ERROR; /* don't fail if a bundle is unopenable */
+        bund = ures_open(path, locale, &subStatus);
+        
+#if defined(URES_TREE_DEBUG)
+        if(!bund || U_FAILURE(subStatus)) {
+            fprintf(stderr, "%s-%s values: Can't open %s locale - skipping. (%s)\n", 
+                path?path:"", keyword, locale, u_errorName(subStatus));
+        }
+#endif
+        
+        ures_getByKey(bund, keyword, &item, &subStatus);
+        
+        if(!bund || U_FAILURE(subStatus)) {
+#if defined(URES_TREE_DEBUG)
+            fprintf(stderr, "%s-%s values: Can't find in %s - skipping. (%s)\n", 
+                path?path:"", keyword, locale, u_errorName(subStatus));
+#endif
+            ures_close(bund);
+            bund = nullptr;
+            continue;
+        }
+        
+        while((subPtr = ures_getNextResource(&item,&subItem,&subStatus)) != 0
+            && U_SUCCESS(subStatus)) {
+            const char *k;
+            int32_t i;
+            k = ures_getKey(subPtr);
+
+#if defined(URES_TREE_DEBUG)
+            /* fprintf(stderr, "%s | %s | %s | %s\n", path?path:"", keyword, locale, k); */
+#endif
+            if(k == nullptr || *k == 0 ||
+                    uprv_strcmp(k, DEFAULT_TAG) == 0 || uprv_strncmp(k, "private-", 8) == 0) {
+                // empty or "default" or unlisted type
+                continue;
+            }
+            for(i=0; i= (VALUES_LIST_SIZE-1)) ||       /* no more space in list .. */
+                    ((valuesIndex+kLen+1+1) >= VALUES_BUF_SIZE)) { /* no more space in buffer (string + 2 nulls) */
+                    *status = U_ILLEGAL_ARGUMENT_ERROR; /* out of space.. */
+                } else {
+                    uprv_strcpy(valuesBuf+valuesIndex, k);
+                    valuesList[valuesCount++] = valuesBuf+valuesIndex;
+                    valuesIndex += kLen;
+#if defined(URES_TREE_DEBUG)
+                    fprintf(stderr, "%s | %s | %s | [%s]   (UNIQUE)\n",
+                        path?path:"", keyword, locale, k);
+#endif
+                    valuesBuf[valuesIndex++] = 0; /* terminate */
+                }
+            }
+        }
+        ures_close(bund);
+    }
+    valuesBuf[valuesIndex++] = 0; /* terminate */
+    
+    ures_close(&item);
+    ures_close(&subItem);
+    uenum_close(locs);
+#if defined(URES_TREE_DEBUG)
+    fprintf(stderr, "%s:  size %d, #%d\n", u_errorName(*status), 
+        valuesIndex, valuesCount);
+#endif
+    return uloc_openKeywordList(valuesBuf, valuesIndex, status);
+}
+#if 0
+/* This code isn't needed, and given the documentation warnings the implementation is suspect */
+U_CAPI UBool U_EXPORT2
+ures_equal(const UResourceBundle* res1, const UResourceBundle* res2){
+    if(res1==nullptr || res2==nullptr){
+        return res1==res2; /* pointer comparison */
+    }
+    if(res1->fKey==nullptr||  res2->fKey==nullptr){
+        return (res1->fKey==res2->fKey);
+    }else{
+        if(uprv_strcmp(res1->fKey, res2->fKey)!=0){
+            return false;
+        }
+    }
+    if(uprv_strcmp(res1->fData->fName, res2->fData->fName)!=0){
+        return false;
+    }
+    if(res1->fData->fPath == nullptr||  res2->fData->fPath==nullptr){
+        return (res1->fData->fPath == res2->fData->fPath);
+    }else{
+        if(uprv_strcmp(res1->fData->fPath, res2->fData->fPath)!=0){
+            return false;
+        }
+    }
+    if(uprv_strcmp(res1->fData->fParent->fName, res2->fData->fParent->fName)!=0){
+        return false;
+    }
+    if(uprv_strcmp(res1->fData->fParent->fPath, res2->fData->fParent->fPath)!=0){
+        return false;
+    }
+    if(uprv_strncmp(res1->fResPath, res2->fResPath, res1->fResPathLen)!=0){
+        return false;
+    }
+    if(res1->fRes != res2->fRes){
+        return false;
+    }
+    return true;
+}
+U_CAPI UResourceBundle* U_EXPORT2
+ures_clone(const UResourceBundle* res, UErrorCode* status){
+    UResourceBundle* bundle = nullptr;
+    UResourceBundle* ret = nullptr;
+    if(U_FAILURE(*status) || res == nullptr){
+        return nullptr;
+    }
+    bundle = ures_open(res->fData->fPath, res->fData->fName, status);
+    if(res->fResPath!=nullptr){
+        ret = ures_findSubResource(bundle, res->fResPath, nullptr, status);
+        ures_close(bundle);
+    }else{
+        ret = bundle;
+    }
+    return ret;
+}
+U_CAPI const UResourceBundle* U_EXPORT2
+ures_getParentBundle(const UResourceBundle* res){
+    if(res==nullptr){
+        return nullptr;
+    }
+    return res->fParentRes;
+}
+#endif
+
+U_CAPI void U_EXPORT2
+ures_getVersionByKey(const UResourceBundle* res, const char *key, UVersionInfo ver, UErrorCode *status) {
+  const char16_t *str;
+  int32_t len;
+  str = ures_getStringByKey(res, key, &len, status);
+  if(U_SUCCESS(*status)) {
+    u_versionFromUString(ver, str);
+  } 
+}
+
+/* eof */
diff --git a/intl/icu/source/common/uresdata.cpp b/intl/icu/source/common/uresdata.cpp
new file mode 100644
index 0000000000..b219e40e21
--- /dev/null
+++ b/intl/icu/source/common/uresdata.cpp
@@ -0,0 +1,1510 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+* Copyright (C) 1999-2016, International Business Machines Corporation
+*               and others. All Rights Reserved.
+*******************************************************************************
+*   file name:  uresdata.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 1999dec08
+*   created by: Markus W. Scherer
+* Modification History:
+*
+*   Date        Name        Description
+*   06/20/2000  helena      OS/400 port changes; mostly typecast.
+*   06/24/02    weiv        Added support for resource sharing
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/udata.h"
+#include "unicode/ustring.h"
+#include "unicode/utf16.h"
+#include "cmemory.h"
+#include "cstring.h"
+#include "resource.h"
+#include "uarrsort.h"
+#include "uassert.h"
+#include "ucol_swp.h"
+#include "udataswp.h"
+#include "uinvchar.h"
+#include "uresdata.h"
+#include "uresimp.h"
+#include "utracimp.h"
+
+/*
+ * Resource access helpers
+ */
+
+/* get a const char* pointer to the key with the keyOffset byte offset from pRoot */
+#define RES_GET_KEY16(pResData, keyOffset) \
+    ((keyOffset)<(pResData)->localKeyLimit ? \
+        (const char *)(pResData)->pRoot+(keyOffset) : \
+        (pResData)->poolBundleKeys+(keyOffset)-(pResData)->localKeyLimit)
+
+#define RES_GET_KEY32(pResData, keyOffset) \
+    ((keyOffset)>=0 ? \
+        (const char *)(pResData)->pRoot+(keyOffset) : \
+        (pResData)->poolBundleKeys+((keyOffset)&0x7fffffff))
+
+#define URESDATA_ITEM_NOT_FOUND -1
+
+/* empty resources, returned when the resource offset is 0 */
+static const uint16_t gEmpty16=0;
+
+static const struct {
+    int32_t length;
+    int32_t res;
+} gEmpty32={ 0, 0 };
+
+static const struct {
+    int32_t length;
+    char16_t nul;
+    char16_t pad;
+} gEmptyString={ 0, 0, 0 };
+
+/*
+ * All the type-access functions assume that
+ * the resource is of the expected type.
+ */
+
+static int32_t
+_res_findTableItem(const ResourceData *pResData, const uint16_t *keyOffsets, int32_t length,
+                   const char *key, const char **realKey) {
+    const char *tableKey;
+    int32_t mid, start, limit;
+    int result;
+
+    /* do a binary search for the key */
+    start=0;
+    limit=length;
+    while(startuseNativeStrcmp) {
+            result = uprv_strcmp(key, tableKey);
+        } else {
+            result = uprv_compareInvCharsAsAscii(key, tableKey);
+        }
+        if (result < 0) {
+            limit = mid;
+        } else if (result > 0) {
+            start = mid + 1;
+        } else {
+            /* We found it! */
+            *realKey=tableKey;
+            return mid;
+        }
+    }
+    return URESDATA_ITEM_NOT_FOUND;  /* not found or table is empty. */
+}
+
+static int32_t
+_res_findTable32Item(const ResourceData *pResData, const int32_t *keyOffsets, int32_t length,
+                     const char *key, const char **realKey) {
+    const char *tableKey;
+    int32_t mid, start, limit;
+    int result;
+
+    /* do a binary search for the key */
+    start=0;
+    limit=length;
+    while(startuseNativeStrcmp) {
+            result = uprv_strcmp(key, tableKey);
+        } else {
+            result = uprv_compareInvCharsAsAscii(key, tableKey);
+        }
+        if (result < 0) {
+            limit = mid;
+        } else if (result > 0) {
+            start = mid + 1;
+        } else {
+            /* We found it! */
+            *realKey=tableKey;
+            return mid;
+        }
+    }
+    return URESDATA_ITEM_NOT_FOUND;  /* not found or table is empty. */
+}
+
+/* helper for res_load() ---------------------------------------------------- */
+
+static UBool U_CALLCONV
+isAcceptable(void *context,
+             const char * /*type*/, const char * /*name*/,
+             const UDataInfo *pInfo) {
+    uprv_memcpy(context, pInfo->formatVersion, 4);
+    return (UBool)(
+        pInfo->size>=20 &&
+        pInfo->isBigEndian==U_IS_BIG_ENDIAN &&
+        pInfo->charsetFamily==U_CHARSET_FAMILY &&
+        pInfo->sizeofUChar==U_SIZEOF_UCHAR &&
+        pInfo->dataFormat[0]==0x52 &&   /* dataFormat="ResB" */
+        pInfo->dataFormat[1]==0x65 &&
+        pInfo->dataFormat[2]==0x73 &&
+        pInfo->dataFormat[3]==0x42 &&
+        (1<=pInfo->formatVersion[0] && pInfo->formatVersion[0]<=3));
+}
+
+/* semi-public functions ---------------------------------------------------- */
+
+static void
+res_init(ResourceData *pResData,
+         UVersionInfo formatVersion, const void *inBytes, int32_t length,
+         UErrorCode *errorCode) {
+    UResType rootType;
+
+    /* get the root resource */
+    pResData->pRoot=(const int32_t *)inBytes;
+    pResData->rootRes=(Resource)*pResData->pRoot;
+    pResData->p16BitUnits=&gEmpty16;
+
+    /* formatVersion 1.1 must have a root item and at least 5 indexes */
+    if(length>=0 && (length/4)<((formatVersion[0]==1 && formatVersion[1]==0) ? 1 : 1+5)) {
+        *errorCode=U_INVALID_FORMAT_ERROR;
+        res_unload(pResData);
+        return;
+    }
+
+    /* currently, we accept only resources that have a Table as their roots */
+    rootType=(UResType)RES_GET_TYPE(pResData->rootRes);
+    if(!URES_IS_TABLE(rootType)) {
+        *errorCode=U_INVALID_FORMAT_ERROR;
+        res_unload(pResData);
+        return;
+    }
+
+    if(formatVersion[0]==1 && formatVersion[1]==0) {
+        pResData->localKeyLimit=0x10000;  /* greater than any 16-bit key string offset */
+    } else {
+        /* bundles with formatVersion 1.1 and later contain an indexes[] array */
+        const int32_t *indexes=pResData->pRoot+1;
+        int32_t indexLength=indexes[URES_INDEX_LENGTH]&0xff;
+        if(indexLength<=URES_INDEX_MAX_TABLE_LENGTH) {
+            *errorCode=U_INVALID_FORMAT_ERROR;
+            res_unload(pResData);
+            return;
+        }
+        if( length>=0 &&
+            (length<((1+indexLength)<<2) ||
+             length<(indexes[URES_INDEX_BUNDLE_TOP]<<2))
+        ) {
+            *errorCode=U_INVALID_FORMAT_ERROR;
+            res_unload(pResData);
+            return;
+        }
+        if(indexes[URES_INDEX_KEYS_TOP]>(1+indexLength)) {
+            pResData->localKeyLimit=indexes[URES_INDEX_KEYS_TOP]<<2;
+        }
+        if(formatVersion[0]>=3) {
+            // In formatVersion 1, the indexLength took up this whole int.
+            // In version 2, bits 31..8 were reserved and always 0.
+            // In version 3, they contain bits 23..0 of the poolStringIndexLimit.
+            // Bits 27..24 are in indexes[URES_INDEX_ATTRIBUTES] bits 15..12.
+            pResData->poolStringIndexLimit=(int32_t)((uint32_t)indexes[URES_INDEX_LENGTH]>>8);
+        }
+        if(indexLength>URES_INDEX_ATTRIBUTES) {
+            int32_t att=indexes[URES_INDEX_ATTRIBUTES];
+            pResData->noFallback=(UBool)(att&URES_ATT_NO_FALLBACK);
+            pResData->isPoolBundle=(UBool)((att&URES_ATT_IS_POOL_BUNDLE)!=0);
+            pResData->usesPoolBundle=(UBool)((att&URES_ATT_USES_POOL_BUNDLE)!=0);
+            pResData->poolStringIndexLimit|=(att&0xf000)<<12;  // bits 15..12 -> 27..24
+            pResData->poolStringIndex16Limit=(int32_t)((uint32_t)att>>16);
+        }
+        if((pResData->isPoolBundle || pResData->usesPoolBundle) && indexLength<=URES_INDEX_POOL_CHECKSUM) {
+            *errorCode=U_INVALID_FORMAT_ERROR;
+            res_unload(pResData);
+            return;
+        }
+        if( indexLength>URES_INDEX_16BIT_TOP &&
+            indexes[URES_INDEX_16BIT_TOP]>indexes[URES_INDEX_KEYS_TOP]
+        ) {
+            pResData->p16BitUnits=(const uint16_t *)(pResData->pRoot+indexes[URES_INDEX_KEYS_TOP]);
+        }
+    }
+
+    if(formatVersion[0]==1 || U_CHARSET_FAMILY==U_ASCII_FAMILY) {
+        /*
+         * formatVersion 1: compare key strings in native-charset order
+         * formatVersion 2 and up: compare key strings in ASCII order
+         */
+        pResData->useNativeStrcmp=true;
+    }
+}
+
+U_CAPI void U_EXPORT2
+res_read(ResourceData *pResData,
+         const UDataInfo *pInfo, const void *inBytes, int32_t length,
+         UErrorCode *errorCode) {
+    UVersionInfo formatVersion;
+
+    uprv_memset(pResData, 0, sizeof(ResourceData));
+    if(U_FAILURE(*errorCode)) {
+        return;
+    }
+    if(!isAcceptable(formatVersion, nullptr, nullptr, pInfo)) {
+        *errorCode=U_INVALID_FORMAT_ERROR;
+        return;
+    }
+    res_init(pResData, formatVersion, inBytes, length, errorCode);
+}
+
+U_CFUNC void
+res_load(ResourceData *pResData,
+         const char *path, const char *name, UErrorCode *errorCode) {
+    UVersionInfo formatVersion;
+
+    uprv_memset(pResData, 0, sizeof(ResourceData));
+
+    /* load the ResourceBundle file */
+    pResData->data=udata_openChoice(path, "res", name, isAcceptable, formatVersion, errorCode);
+    if(U_FAILURE(*errorCode)) {
+        return;
+    }
+
+    /* get its memory and initialize *pResData */
+    res_init(pResData, formatVersion, udata_getMemory(pResData->data), -1, errorCode);
+}
+
+U_CFUNC void
+res_unload(ResourceData *pResData) {
+    if(pResData->data!=nullptr) {
+        udata_close(pResData->data);
+        pResData->data=nullptr;
+    }
+}
+
+static const int8_t gPublicTypes[URES_LIMIT] = {
+    URES_STRING,
+    URES_BINARY,
+    URES_TABLE,
+    URES_ALIAS,
+
+    URES_TABLE,     /* URES_TABLE32 */
+    URES_TABLE,     /* URES_TABLE16 */
+    URES_STRING,    /* URES_STRING_V2 */
+    URES_INT,
+
+    URES_ARRAY,
+    URES_ARRAY,     /* URES_ARRAY16 */
+    URES_NONE,
+    URES_NONE,
+
+    URES_NONE,
+    URES_NONE,
+    URES_INT_VECTOR,
+    URES_NONE
+};
+
+U_CAPI UResType U_EXPORT2
+res_getPublicType(Resource res) {
+    return (UResType)gPublicTypes[RES_GET_TYPE(res)];
+}
+
+U_CAPI const char16_t * U_EXPORT2
+res_getStringNoTrace(const ResourceData *pResData, Resource res, int32_t *pLength) {
+    const char16_t *p;
+    uint32_t offset=RES_GET_OFFSET(res);
+    int32_t length;
+    if(RES_GET_TYPE(res)==URES_STRING_V2) {
+        int32_t first;
+        if((int32_t)offsetpoolStringIndexLimit) {
+            p=(const char16_t *)pResData->poolBundleStrings+offset;
+        } else {
+            p=(const char16_t *)pResData->p16BitUnits+(offset-pResData->poolStringIndexLimit);
+        }
+        first=*p;
+        if(!U16_IS_TRAIL(first)) {
+            length=u_strlen(p);
+        } else if(first<0xdfef) {
+            length=first&0x3ff;
+            ++p;
+        } else if(first<0xdfff) {
+            length=((first-0xdfef)<<16)|p[1];
+            p+=2;
+        } else {
+            length=((int32_t)p[1]<<16)|p[2];
+            p+=3;
+        }
+    } else if(res==offset) /* RES_GET_TYPE(res)==URES_STRING */ {
+        const int32_t *p32= res==0 ? &gEmptyString.length : pResData->pRoot+res;
+        length=*p32++;
+        p=(const char16_t *)p32;
+    } else {
+        p=nullptr;
+        length=0;
+    }
+    if(pLength) {
+        *pLength=length;
+    }
+    return p;
+}
+
+namespace {
+
+/**
+ * CLDR string value (three empty-set symbols)=={2205, 2205, 2205}
+ * prevents fallback to the parent bundle.
+ * TODO: combine with other code that handles this marker, use EMPTY_SET constant.
+ * TODO: maybe move to uresbund.cpp?
+ */
+UBool isNoInheritanceMarker(const ResourceData *pResData, Resource res) {
+    uint32_t offset=RES_GET_OFFSET(res);
+    if (offset == 0) {
+        // empty string
+    } else if (res == offset) {
+        const int32_t *p32=pResData->pRoot+res;
+        int32_t length=*p32;
+        const char16_t *p=(const char16_t *)p32;
+        return length == 3 && p[2] == 0x2205 && p[3] == 0x2205 && p[4] == 0x2205;
+    } else if (RES_GET_TYPE(res) == URES_STRING_V2) {
+        const char16_t *p;
+        if((int32_t)offsetpoolStringIndexLimit) {
+            p=(const char16_t *)pResData->poolBundleStrings+offset;
+        } else {
+            p=(const char16_t *)pResData->p16BitUnits+(offset-pResData->poolStringIndexLimit);
+        }
+        int32_t first=*p;
+        if (first == 0x2205) {  // implicit length
+            return p[1] == 0x2205 && p[2] == 0x2205 && p[3] == 0;
+        } else if (first == 0xdc03) {  // explicit length 3 (should not occur)
+            return p[1] == 0x2205 && p[2] == 0x2205 && p[3] == 0x2205;
+        } else {
+            // Assume that the string has not been stored with more length units than necessary.
+            return false;
+        }
+    }
+    return false;
+}
+
+int32_t getStringArray(const ResourceData *pResData, const icu::ResourceArray &array,
+                       icu::UnicodeString *dest, int32_t capacity,
+                       UErrorCode &errorCode) {
+    if(U_FAILURE(errorCode)) {
+        return 0;
+    }
+    if(dest == nullptr ? capacity != 0 : capacity < 0) {
+        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    int32_t length = array.getSize();
+    if(length == 0) {
+        return 0;
+    }
+    if(length > capacity) {
+        errorCode = U_BUFFER_OVERFLOW_ERROR;
+        return length;
+    }
+    for(int32_t i = 0; i < length; ++i) {
+        int32_t sLength;
+        // No tracing: handled by the caller
+        const char16_t *s = res_getStringNoTrace(pResData, array.internalGetResource(pResData, i), &sLength);
+        if(s == nullptr) {
+            errorCode = U_RESOURCE_TYPE_MISMATCH;
+            return 0;
+        }
+        dest[i].setTo(true, s, sLength);
+    }
+    return length;
+}
+
+}  // namespace
+
+U_CAPI const char16_t * U_EXPORT2
+res_getAlias(const ResourceData *pResData, Resource res, int32_t *pLength) {
+    const char16_t *p;
+    uint32_t offset=RES_GET_OFFSET(res);
+    int32_t length;
+    if(RES_GET_TYPE(res)==URES_ALIAS) {
+        const int32_t *p32= offset==0 ? &gEmptyString.length : pResData->pRoot+offset;
+        length=*p32++;
+        p=(const char16_t *)p32;
+    } else {
+        p=nullptr;
+        length=0;
+    }
+    if(pLength) {
+        *pLength=length;
+    }
+    return p;
+}
+
+U_CAPI const uint8_t * U_EXPORT2
+res_getBinaryNoTrace(const ResourceData *pResData, Resource res, int32_t *pLength) {
+    const uint8_t *p;
+    uint32_t offset=RES_GET_OFFSET(res);
+    int32_t length;
+    if(RES_GET_TYPE(res)==URES_BINARY) {
+        const int32_t *p32= offset==0 ? (const int32_t*)&gEmpty32 : pResData->pRoot+offset;
+        length=*p32++;
+        p=(const uint8_t *)p32;
+    } else {
+        p=nullptr;
+        length=0;
+    }
+    if(pLength) {
+        *pLength=length;
+    }
+    return p;
+}
+
+
+U_CAPI const int32_t * U_EXPORT2
+res_getIntVectorNoTrace(const ResourceData *pResData, Resource res, int32_t *pLength) {
+    const int32_t *p;
+    uint32_t offset=RES_GET_OFFSET(res);
+    int32_t length;
+    if(RES_GET_TYPE(res)==URES_INT_VECTOR) {
+        p= offset==0 ? (const int32_t *)&gEmpty32 : pResData->pRoot+offset;
+        length=*p++;
+    } else {
+        p=nullptr;
+        length=0;
+    }
+    if(pLength) {
+        *pLength=length;
+    }
+    return p;
+}
+
+U_CAPI int32_t U_EXPORT2
+res_countArrayItems(const ResourceData *pResData, Resource res) {
+    uint32_t offset=RES_GET_OFFSET(res);
+    switch(RES_GET_TYPE(res)) {
+    case URES_STRING:
+    case URES_STRING_V2:
+    case URES_BINARY:
+    case URES_ALIAS:
+    case URES_INT:
+    case URES_INT_VECTOR:
+        return 1;
+    case URES_ARRAY:
+    case URES_TABLE32:
+        return offset==0 ? 0 : *(pResData->pRoot+offset);
+    case URES_TABLE:
+        return offset==0 ? 0 : *((const uint16_t *)(pResData->pRoot+offset));
+    case URES_ARRAY16:
+    case URES_TABLE16:
+        return pResData->p16BitUnits[offset];
+    default:
+        return 0;
+    }
+}
+
+U_NAMESPACE_BEGIN
+
+ResourceDataValue::~ResourceDataValue() {}
+
+UResType ResourceDataValue::getType() const {
+    return res_getPublicType(res);
+}
+
+const char16_t *ResourceDataValue::getString(int32_t &length, UErrorCode &errorCode) const {
+    if(U_FAILURE(errorCode)) {
+        return nullptr;
+    }
+    const char16_t *s = res_getString(fTraceInfo, &getData(), res, &length);
+    if(s == nullptr) {
+        errorCode = U_RESOURCE_TYPE_MISMATCH;
+    }
+    return s;
+}
+
+const char16_t *ResourceDataValue::getAliasString(int32_t &length, UErrorCode &errorCode) const {
+    if(U_FAILURE(errorCode)) {
+        return nullptr;
+    }
+    const char16_t *s = res_getAlias(&getData(), res, &length);
+    if(s == nullptr) {
+        errorCode = U_RESOURCE_TYPE_MISMATCH;
+    }
+    return s;
+}
+
+int32_t ResourceDataValue::getInt(UErrorCode &errorCode) const {
+    if(U_FAILURE(errorCode)) {
+        return 0;
+    }
+    if(RES_GET_TYPE(res) != URES_INT) {
+        errorCode = U_RESOURCE_TYPE_MISMATCH;
+    }
+    return res_getInt(fTraceInfo, res);
+}
+
+uint32_t ResourceDataValue::getUInt(UErrorCode &errorCode) const {
+    if(U_FAILURE(errorCode)) {
+        return 0;
+    }
+    if(RES_GET_TYPE(res) != URES_INT) {
+        errorCode = U_RESOURCE_TYPE_MISMATCH;
+    }
+    return res_getUInt(fTraceInfo, res);
+}
+
+const int32_t *ResourceDataValue::getIntVector(int32_t &length, UErrorCode &errorCode) const {
+    if(U_FAILURE(errorCode)) {
+        return nullptr;
+    }
+    const int32_t *iv = res_getIntVector(fTraceInfo, &getData(), res, &length);
+    if(iv == nullptr) {
+        errorCode = U_RESOURCE_TYPE_MISMATCH;
+    }
+    return iv;
+}
+
+const uint8_t *ResourceDataValue::getBinary(int32_t &length, UErrorCode &errorCode) const {
+    if(U_FAILURE(errorCode)) {
+        return nullptr;
+    }
+    const uint8_t *b = res_getBinary(fTraceInfo, &getData(), res, &length);
+    if(b == nullptr) {
+        errorCode = U_RESOURCE_TYPE_MISMATCH;
+    }
+    return b;
+}
+
+ResourceArray ResourceDataValue::getArray(UErrorCode &errorCode) const {
+    if(U_FAILURE(errorCode)) {
+        return ResourceArray();
+    }
+    const uint16_t *items16 = nullptr;
+    const Resource *items32 = nullptr;
+    uint32_t offset=RES_GET_OFFSET(res);
+    int32_t length = 0;
+    switch(RES_GET_TYPE(res)) {
+    case URES_ARRAY:
+        if (offset!=0) {  // empty if offset==0
+            items32 = (const Resource *)getData().pRoot+offset;
+            length = *items32++;
+        }
+        break;
+    case URES_ARRAY16:
+        items16 = getData().p16BitUnits+offset;
+        length = *items16++;
+        break;
+    default:
+        errorCode = U_RESOURCE_TYPE_MISMATCH;
+        return ResourceArray();
+    }
+    return ResourceArray(items16, items32, length, fTraceInfo);
+}
+
+ResourceTable ResourceDataValue::getTable(UErrorCode &errorCode) const {
+    if(U_FAILURE(errorCode)) {
+        return ResourceTable();
+    }
+    const uint16_t *keys16 = nullptr;
+    const int32_t *keys32 = nullptr;
+    const uint16_t *items16 = nullptr;
+    const Resource *items32 = nullptr;
+    uint32_t offset = RES_GET_OFFSET(res);
+    int32_t length = 0;
+    switch(RES_GET_TYPE(res)) {
+    case URES_TABLE:
+        if (offset != 0) {  // empty if offset==0
+            keys16 = (const uint16_t *)(getData().pRoot+offset);
+            length = *keys16++;
+            items32 = (const Resource *)(keys16+length+(~length&1));
+        }
+        break;
+    case URES_TABLE16:
+        keys16 = getData().p16BitUnits+offset;
+        length = *keys16++;
+        items16 = keys16 + length;
+        break;
+    case URES_TABLE32:
+        if (offset != 0) {  // empty if offset==0
+            keys32 = getData().pRoot+offset;
+            length = *keys32++;
+            items32 = (const Resource *)keys32 + length;
+        }
+        break;
+    default:
+        errorCode = U_RESOURCE_TYPE_MISMATCH;
+        return ResourceTable();
+    }
+    return ResourceTable(keys16, keys32, items16, items32, length, fTraceInfo);
+}
+
+UBool ResourceDataValue::isNoInheritanceMarker() const {
+    return ::isNoInheritanceMarker(&getData(), res);
+}
+
+int32_t ResourceDataValue::getStringArray(UnicodeString *dest, int32_t capacity,
+                                          UErrorCode &errorCode) const {
+    return ::getStringArray(&getData(), getArray(errorCode), dest, capacity, errorCode);
+}
+
+int32_t ResourceDataValue::getStringArrayOrStringAsArray(UnicodeString *dest, int32_t capacity,
+                                                         UErrorCode &errorCode) const {
+    if(URES_IS_ARRAY(res)) {
+        return ::getStringArray(&getData(), getArray(errorCode), dest, capacity, errorCode);
+    }
+    if(U_FAILURE(errorCode)) {
+        return 0;
+    }
+    if(dest == nullptr ? capacity != 0 : capacity < 0) {
+        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    if(capacity < 1) {
+        errorCode = U_BUFFER_OVERFLOW_ERROR;
+        return 1;
+    }
+    int32_t sLength;
+    const char16_t *s = res_getString(fTraceInfo, &getData(), res, &sLength);
+    if(s != nullptr) {
+        dest[0].setTo(true, s, sLength);
+        return 1;
+    }
+    errorCode = U_RESOURCE_TYPE_MISMATCH;
+    return 0;
+}
+
+UnicodeString ResourceDataValue::getStringOrFirstOfArray(UErrorCode &errorCode) const {
+    UnicodeString us;
+    if(U_FAILURE(errorCode)) {
+        return us;
+    }
+    int32_t sLength;
+    const char16_t *s = res_getString(fTraceInfo, &getData(), res, &sLength);
+    if(s != nullptr) {
+        us.setTo(true, s, sLength);
+        return us;
+    }
+    ResourceArray array = getArray(errorCode);
+    if(U_FAILURE(errorCode)) {
+        return us;
+    }
+    if(array.getSize() > 0) {
+        // Tracing is already performed above (unimportant for trace that this is an array)
+        s = res_getStringNoTrace(&getData(), array.internalGetResource(&getData(), 0), &sLength);
+        if(s != nullptr) {
+            us.setTo(true, s, sLength);
+            return us;
+        }
+    }
+    errorCode = U_RESOURCE_TYPE_MISMATCH;
+    return us;
+}
+
+U_NAMESPACE_END
+
+static Resource
+makeResourceFrom16(const ResourceData *pResData, int32_t res16) {
+    if(res16poolStringIndex16Limit) {
+        // Pool string, nothing to do.
+    } else {
+        // Local string, adjust the 16-bit offset to a regular one,
+        // with a larger pool string index limit.
+        res16=res16-pResData->poolStringIndex16Limit+pResData->poolStringIndexLimit;
+    }
+    return URES_MAKE_RESOURCE(URES_STRING_V2, res16);
+}
+
+U_CAPI Resource U_EXPORT2
+res_getTableItemByKey(const ResourceData *pResData, Resource table,
+                      int32_t *indexR, const char **key) {
+    uint32_t offset=RES_GET_OFFSET(table);
+    int32_t length;
+    int32_t idx;
+    if(key == nullptr || *key == nullptr) {
+        return RES_BOGUS;
+    }
+    switch(RES_GET_TYPE(table)) {
+    case URES_TABLE: {
+        if (offset!=0) { /* empty if offset==0 */
+            const uint16_t *p= (const uint16_t *)(pResData->pRoot+offset);
+            length=*p++;
+            *indexR=idx=_res_findTableItem(pResData, p, length, *key, key);
+            if(idx>=0) {
+                const Resource *p32=(const Resource *)(p+length+(~length&1));
+                return p32[idx];
+            }
+        }
+        break;
+    }
+    case URES_TABLE16: {
+        const uint16_t *p=pResData->p16BitUnits+offset;
+        length=*p++;
+        *indexR=idx=_res_findTableItem(pResData, p, length, *key, key);
+        if(idx>=0) {
+            return makeResourceFrom16(pResData, p[length+idx]);
+        }
+        break;
+    }
+    case URES_TABLE32: {
+        if (offset!=0) { /* empty if offset==0 */
+            const int32_t *p= pResData->pRoot+offset;
+            length=*p++;
+            *indexR=idx=_res_findTable32Item(pResData, p, length, *key, key);
+            if(idx>=0) {
+                return (Resource)p[length+idx];
+            }
+        }
+        break;
+    }
+    default:
+        break;
+    }
+    return RES_BOGUS;
+}
+
+U_CAPI Resource U_EXPORT2
+res_getTableItemByIndex(const ResourceData *pResData, Resource table,
+                        int32_t indexR, const char **key) {
+    uint32_t offset=RES_GET_OFFSET(table);
+    int32_t length;
+    if (indexR < 0) {
+        return RES_BOGUS;
+    }
+    switch(RES_GET_TYPE(table)) {
+    case URES_TABLE: {
+        if (offset != 0) { /* empty if offset==0 */
+            const uint16_t *p= (const uint16_t *)(pResData->pRoot+offset);
+            length=*p++;
+            if(indexRp16BitUnits+offset;
+        length=*p++;
+        if(indexRpRoot+offset;
+            length=*p++;
+            if(indexRrootRes, &idx, &realKey);
+}
+
+
+UBool icu::ResourceTable::getKeyAndValue(int32_t i,
+                                         const char *&key, icu::ResourceValue &value) const {
+    if(0 <= i && i < length) {
+        icu::ResourceDataValue &rdValue = static_cast(value);
+        if (keys16 != nullptr) {
+            key = RES_GET_KEY16(&rdValue.getData(), keys16[i]);
+        } else {
+            key = RES_GET_KEY32(&rdValue.getData(), keys32[i]);
+        }
+        Resource res;
+        if (items16 != nullptr) {
+            res = makeResourceFrom16(&rdValue.getData(), items16[i]);
+        } else {
+            res = items32[i];
+        }
+        // Note: the ResourceTracer keeps a reference to the field of this
+        // ResourceTable. This is OK because the ResourceTable should remain
+        // alive for the duration that fields are being read from it
+        // (including nested fields).
+        rdValue.setResource(res, ResourceTracer(fTraceInfo, key));
+        return true;
+    }
+    return false;
+}
+
+UBool icu::ResourceTable::findValue(const char *key, ResourceValue &value) const {
+    icu::ResourceDataValue &rdValue = static_cast(value);
+    const char *realKey = nullptr;
+    int32_t i;
+    if (keys16 != nullptr) {
+        i = _res_findTableItem(&rdValue.getData(), keys16, length, key, &realKey);
+    } else {
+        i = _res_findTable32Item(&rdValue.getData(), keys32, length, key, &realKey);
+    }
+    if (i >= 0) {
+        Resource res;
+        if (items16 != nullptr) {
+            res = makeResourceFrom16(&rdValue.getData(), items16[i]);
+        } else {
+            res = items32[i];
+        }
+        // Same note about lifetime as in getKeyAndValue().
+        rdValue.setResource(res, ResourceTracer(fTraceInfo, key));
+        return true;
+    }
+    return false;
+}
+
+U_CAPI Resource U_EXPORT2
+res_getArrayItem(const ResourceData *pResData, Resource array, int32_t indexR) {
+    uint32_t offset=RES_GET_OFFSET(array);
+    if (indexR < 0) {
+        return RES_BOGUS;
+    }
+    switch(RES_GET_TYPE(array)) {
+    case URES_ARRAY: {
+        if (offset!=0) { /* empty if offset==0 */
+            const int32_t *p= pResData->pRoot+offset;
+            if(indexR<*p) {
+                return (Resource)p[1+indexR];
+            }
+        }
+        break;
+    }
+    case URES_ARRAY16: {
+        const uint16_t *p=pResData->p16BitUnits+offset;
+        if(indexR<*p) {
+            return makeResourceFrom16(pResData, p[1+indexR]);
+        }
+        break;
+    }
+    default:
+        break;
+    }
+    return RES_BOGUS;
+}
+
+uint32_t icu::ResourceArray::internalGetResource(const ResourceData *pResData, int32_t i) const {
+    if (items16 != nullptr) {
+        return makeResourceFrom16(pResData, items16[i]);
+    } else {
+        return items32[i];
+    }
+}
+
+UBool icu::ResourceArray::getValue(int32_t i, icu::ResourceValue &value) const {
+    if(0 <= i && i < length) {
+        icu::ResourceDataValue &rdValue = static_cast(value);
+        // Note: the ResourceTracer keeps a reference to the field of this
+        // ResourceArray. This is OK because the ResourceArray should remain
+        // alive for the duration that fields are being read from it
+        // (including nested fields).
+        rdValue.setResource(
+            internalGetResource(&rdValue.getData(), i),
+            ResourceTracer(fTraceInfo, i));
+        return true;
+    }
+    return false;
+}
+
+U_CFUNC Resource
+res_findResource(const ResourceData *pResData, Resource r, char** path, const char** key) {
+  char *pathP = *path, *nextSepP = *path;
+  char *closeIndex = nullptr;
+  Resource t1 = r;
+  Resource t2;
+  int32_t indexR = 0;
+  UResType type = (UResType)RES_GET_TYPE(t1);
+
+  /* if you come in with an empty path, you'll be getting back the same resource */
+  if(!uprv_strlen(pathP)) {
+      return r;
+  }
+
+  /* one needs to have an aggregate resource in order to search in it */
+  if(!URES_IS_CONTAINER(type)) {
+      return RES_BOGUS;
+  }
+  
+  while(nextSepP && *pathP && t1 != RES_BOGUS && URES_IS_CONTAINER(type)) {
+    /* Iteration stops if: the path has been consumed, we found a non-existing
+     * resource (t1 == RES_BOGUS) or we found a scalar resource (including alias)
+     */
+    nextSepP = uprv_strchr(pathP, RES_PATH_SEPARATOR);
+    /* if there are more separators, terminate string 
+     * and set path to the remaining part of the string
+     */
+    if(nextSepP != nullptr) {
+      if(nextSepP == pathP) {
+        // Empty key string.
+        return RES_BOGUS;
+      }
+      *nextSepP = 0; /* overwrite the separator with a NUL to terminate the key */
+      *path = nextSepP+1;
+    } else {
+      *path = uprv_strchr(pathP, 0);
+    }
+
+    /* if the resource is a table */
+    /* try the key based access */
+    if(URES_IS_TABLE(type)) {
+      *key = pathP;
+      t2 = res_getTableItemByKey(pResData, t1, &indexR, key);
+    } else if(URES_IS_ARRAY(type)) {
+      indexR = uprv_strtol(pathP, &closeIndex, 10);
+      if(indexR >= 0 && *closeIndex == 0) {
+        t2 = res_getArrayItem(pResData, t1, indexR);
+      } else {
+        t2 = RES_BOGUS; /* have an array, but don't have a valid index */
+      }
+      *key = nullptr;
+    } else { /* can't do much here, except setting t2 to bogus */
+      t2 = RES_BOGUS;
+    }
+    t1 = t2;
+    type = (UResType)RES_GET_TYPE(t1);
+    /* position pathP to next resource key/index */
+    pathP = *path;
+  }
+
+  return t1;
+}
+
+/* resource bundle swapping ------------------------------------------------- */
+
+/*
+ * Need to always enumerate the entire item tree,
+ * track the lowest address of any item to use as the limit for char keys[],
+ * track the highest address of any item to return the size of the data.
+ *
+ * We should have thought of storing those in the data...
+ * It is possible to extend the data structure by putting additional values
+ * in places that are inaccessible by ordinary enumeration of the item tree.
+ * For example, additional integers could be stored at the beginning or
+ * end of the key strings; this could be indicated by a minor version number,
+ * and the data swapping would have to know about these values.
+ *
+ * The data structure does not forbid keys to be shared, so we must swap
+ * all keys once instead of each key when it is referenced.
+ *
+ * These swapping functions assume that a resource bundle always has a length
+ * that is a multiple of 4 bytes.
+ * Currently, this is trivially true because genrb writes bundle tree leaves
+ * physically first, before their branches, so that the root table with its
+ * array of resource items (uint32_t values) is always last.
+ */
+
+/* definitions for table sorting ------------------------ */
+
+/*
+ * row of a temporary array
+ *
+ * gets platform-endian key string indexes and sorting indexes;
+ * after sorting this array by keys, the actual key/value arrays are permutated
+ * according to the sorting indexes
+ */
+typedef struct Row {
+    int32_t keyIndex, sortIndex;
+} Row;
+
+static int32_t U_CALLCONV
+ures_compareRows(const void *context, const void *left, const void *right) {
+    const char *keyChars=(const char *)context;
+    return (int32_t)uprv_strcmp(keyChars+((const Row *)left)->keyIndex,
+                                keyChars+((const Row *)right)->keyIndex);
+}
+
+typedef struct TempTable {
+    const char *keyChars;
+    Row *rows;
+    int32_t *resort;
+    uint32_t *resFlags;
+    int32_t localKeyLimit;
+    uint8_t majorFormatVersion;
+} TempTable;
+
+enum {
+    STACK_ROW_CAPACITY=200
+};
+
+/* The table item key string is not locally available. */
+static const char *const gUnknownKey="";
+
+/* resource table key for collation binaries: "%%CollationBin" */
+static const char16_t gCollationBinKey[]={
+    0x25, 0x25,
+    0x43, 0x6f, 0x6c, 0x6c, 0x61, 0x74, 0x69, 0x6f, 0x6e,
+    0x42, 0x69, 0x6e,
+    0
+};
+
+/*
+ * swap one resource item
+ */
+static void
+ures_swapResource(const UDataSwapper *ds,
+                  const Resource *inBundle, Resource *outBundle,
+                  Resource res, /* caller swaps res itself */
+                  const char *key,
+                  TempTable *pTempTable,
+                  UErrorCode *pErrorCode) {
+    const Resource *p;
+    Resource *q;
+    int32_t offset, count;
+
+    switch(RES_GET_TYPE(res)) {
+    case URES_TABLE16:
+    case URES_STRING_V2:
+    case URES_INT:
+    case URES_ARRAY16:
+        /* integer, or points to 16-bit units, nothing to do here */
+        return;
+    default:
+        break;
+    }
+
+    /* all other types use an offset to point to their data */
+    offset=(int32_t)RES_GET_OFFSET(res);
+    if(offset==0) {
+        /* special offset indicating an empty item */
+        return;
+    }
+    if(pTempTable->resFlags[offset>>5]&((uint32_t)1<<(offset&0x1f))) {
+        /* we already swapped this resource item */
+        return;
+    } else {
+        /* mark it as swapped now */
+        pTempTable->resFlags[offset>>5]|=((uint32_t)1<<(offset&0x1f));
+    }
+
+    p=inBundle+offset;
+    q=outBundle+offset;
+
+    switch(RES_GET_TYPE(res)) {
+    case URES_ALIAS:
+        /* physically same value layout as string, fall through */
+        U_FALLTHROUGH;
+    case URES_STRING:
+        count=udata_readInt32(ds, (int32_t)*p);
+        /* swap length */
+        ds->swapArray32(ds, p, 4, q, pErrorCode);
+        /* swap each char16_t (the terminating NUL would not change) */
+        ds->swapArray16(ds, p+1, 2*count, q+1, pErrorCode);
+        break;
+    case URES_BINARY:
+        count=udata_readInt32(ds, (int32_t)*p);
+        /* swap length */
+        ds->swapArray32(ds, p, 4, q, pErrorCode);
+        /* no need to swap or copy bytes - ures_swap() copied them all */
+
+        /* swap known formats */
+#if !UCONFIG_NO_COLLATION
+        if( key!=nullptr &&  /* the binary is in a table */
+            (key!=gUnknownKey ?
+                /* its table key string is "%%CollationBin" */
+                0==ds->compareInvChars(ds, key, -1,
+                                       gCollationBinKey, UPRV_LENGTHOF(gCollationBinKey)-1) :
+                /* its table key string is unknown but it looks like a collation binary */
+                ucol_looksLikeCollationBinary(ds, p+1, count))
+        ) {
+            ucol_swap(ds, p+1, count, q+1, pErrorCode);
+        }
+#endif
+        break;
+    case URES_TABLE:
+    case URES_TABLE32:
+        {
+            const uint16_t *pKey16;
+            uint16_t *qKey16;
+
+            const int32_t *pKey32;
+            int32_t *qKey32;
+
+            Resource item;
+            int32_t i, oldIndex;
+
+            if(RES_GET_TYPE(res)==URES_TABLE) {
+                /* get table item count */
+                pKey16=(const uint16_t *)p;
+                qKey16=(uint16_t *)q;
+                count=ds->readUInt16(*pKey16);
+
+                pKey32=qKey32=nullptr;
+
+                /* swap count */
+                ds->swapArray16(ds, pKey16++, 2, qKey16++, pErrorCode);
+
+                offset+=((1+count)+1)/2;
+            } else {
+                /* get table item count */
+                pKey32=(const int32_t *)p;
+                qKey32=(int32_t *)q;
+                count=udata_readInt32(ds, *pKey32);
+
+                pKey16=qKey16=nullptr;
+
+                /* swap count */
+                ds->swapArray32(ds, pKey32++, 4, qKey32++, pErrorCode);
+
+                offset+=1+count;
+            }
+
+            if(count==0) {
+                break;
+            }
+
+            p=inBundle+offset; /* pointer to table resources */
+            q=outBundle+offset;
+
+            /* recurse */
+            for(i=0; ireadUInt16(pKey16[i]);
+                    if(keyOffsetlocalKeyLimit) {
+                        itemKey=(const char *)outBundle+keyOffset;
+                    }
+                } else {
+                    int32_t keyOffset=udata_readInt32(ds, pKey32[i]);
+                    if(keyOffset>=0) {
+                        itemKey=(const char *)outBundle+keyOffset;
+                    }
+                }
+                item=ds->readUInt32(p[i]);
+                ures_swapResource(ds, inBundle, outBundle, item, itemKey, pTempTable, pErrorCode);
+                if(U_FAILURE(*pErrorCode)) {
+                    udata_printError(ds, "ures_swapResource(table res=%08x)[%d].recurse(%08x) failed\n",
+                                     res, i, item);
+                    return;
+                }
+            }
+
+            if(pTempTable->majorFormatVersion>1 || ds->inCharset==ds->outCharset) {
+                /* no need to sort, just swap the offset/value arrays */
+                if(pKey16!=nullptr) {
+                    ds->swapArray16(ds, pKey16, count*2, qKey16, pErrorCode);
+                    ds->swapArray32(ds, p, count*4, q, pErrorCode);
+                } else {
+                    /* swap key offsets and items as one array */
+                    ds->swapArray32(ds, pKey32, count*2*4, qKey32, pErrorCode);
+                }
+                break;
+            }
+
+            /*
+             * We need to sort tables by outCharset key strings because they
+             * sort differently for different charset families.
+             * ures_swap() already set pTempTable->keyChars appropriately.
+             * First we set up a temporary table with the key indexes and
+             * sorting indexes and sort that.
+             * Then we permutate and copy/swap the actual values.
+             */
+            if(pKey16!=nullptr) {
+                for(i=0; irows[i].keyIndex=ds->readUInt16(pKey16[i]);
+                    pTempTable->rows[i].sortIndex=i;
+                }
+            } else {
+                for(i=0; irows[i].keyIndex=udata_readInt32(ds, pKey32[i]);
+                    pTempTable->rows[i].sortIndex=i;
+                }
+            }
+            uprv_sortArray(pTempTable->rows, count, sizeof(Row),
+                           ures_compareRows, pTempTable->keyChars,
+                           false, pErrorCode);
+            if(U_FAILURE(*pErrorCode)) {
+                udata_printError(ds, "ures_swapResource(table res=%08x).uprv_sortArray(%d items) failed\n",
+                                 res, count);
+                return;
+            }
+
+            /*
+             * copy/swap/permutate items
+             *
+             * If we swap in-place, then the permutation must use another
+             * temporary array (pTempTable->resort)
+             * before the results are copied to the outBundle.
+             */
+            /* keys */
+            if(pKey16!=nullptr) {
+                uint16_t *rKey16;
+
+                if(pKey16!=qKey16) {
+                    rKey16=qKey16;
+                } else {
+                    rKey16=(uint16_t *)pTempTable->resort;
+                }
+                for(i=0; irows[i].sortIndex;
+                    ds->swapArray16(ds, pKey16+oldIndex, 2, rKey16+i, pErrorCode);
+                }
+                if(qKey16!=rKey16) {
+                    uprv_memcpy(qKey16, rKey16, 2*count);
+                }
+            } else {
+                int32_t *rKey32;
+
+                if(pKey32!=qKey32) {
+                    rKey32=qKey32;
+                } else {
+                    rKey32=pTempTable->resort;
+                }
+                for(i=0; irows[i].sortIndex;
+                    ds->swapArray32(ds, pKey32+oldIndex, 4, rKey32+i, pErrorCode);
+                }
+                if(qKey32!=rKey32) {
+                    uprv_memcpy(qKey32, rKey32, 4*count);
+                }
+            }
+
+            /* resources */
+            {
+                Resource *r;
+
+
+                if(p!=q) {
+                    r=q;
+                } else {
+                    r=(Resource *)pTempTable->resort;
+                }
+                for(i=0; irows[i].sortIndex;
+                    ds->swapArray32(ds, p+oldIndex, 4, r+i, pErrorCode);
+                }
+                if(q!=r) {
+                    uprv_memcpy(q, r, 4*count);
+                }
+            }
+        }
+        break;
+    case URES_ARRAY:
+        {
+            Resource item;
+            int32_t i;
+
+            count=udata_readInt32(ds, (int32_t)*p);
+            /* swap length */
+            ds->swapArray32(ds, p++, 4, q++, pErrorCode);
+
+            /* recurse */
+            for(i=0; ireadUInt32(p[i]);
+                ures_swapResource(ds, inBundle, outBundle, item, nullptr, pTempTable, pErrorCode);
+                if(U_FAILURE(*pErrorCode)) {
+                    udata_printError(ds, "ures_swapResource(array res=%08x)[%d].recurse(%08x) failed\n",
+                                     res, i, item);
+                    return;
+                }
+            }
+
+            /* swap items */
+            ds->swapArray32(ds, p, 4*count, q, pErrorCode);
+        }
+        break;
+    case URES_INT_VECTOR:
+        count=udata_readInt32(ds, (int32_t)*p);
+        /* swap length and each integer */
+        ds->swapArray32(ds, p, 4*(1+count), q, pErrorCode);
+        break;
+    default:
+        /* also catches RES_BOGUS */
+        *pErrorCode=U_UNSUPPORTED_ERROR;
+        break;
+    }
+}
+
+U_CAPI int32_t U_EXPORT2
+ures_swap(const UDataSwapper *ds,
+          const void *inData, int32_t length, void *outData,
+          UErrorCode *pErrorCode) {
+    const UDataInfo *pInfo;
+    const Resource *inBundle;
+    Resource rootRes;
+    int32_t headerSize, maxTableLength;
+
+    Row rows[STACK_ROW_CAPACITY];
+    int32_t resort[STACK_ROW_CAPACITY];
+    TempTable tempTable;
+
+    const int32_t *inIndexes;
+
+    /* the following integers count Resource item offsets (4 bytes each), not bytes */
+    int32_t bundleLength, indexLength, keysBottom, keysTop, resBottom, top;
+
+    /* udata_swapDataHeader checks the arguments */
+    headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
+    if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+
+    /* check data format and format version */
+    pInfo=(const UDataInfo *)((const char *)inData+4);
+    if(!(
+        pInfo->dataFormat[0]==0x52 &&   /* dataFormat="ResB" */
+        pInfo->dataFormat[1]==0x65 &&
+        pInfo->dataFormat[2]==0x73 &&
+        pInfo->dataFormat[3]==0x42 &&
+        /* formatVersion 1.1+ or 2.x or 3.x */
+        ((pInfo->formatVersion[0]==1 && pInfo->formatVersion[1]>=1) ||
+            pInfo->formatVersion[0]==2 || pInfo->formatVersion[0]==3)
+    )) {
+        udata_printError(ds, "ures_swap(): data format %02x.%02x.%02x.%02x (format version %02x.%02x) is not a resource bundle\n",
+                         pInfo->dataFormat[0], pInfo->dataFormat[1],
+                         pInfo->dataFormat[2], pInfo->dataFormat[3],
+                         pInfo->formatVersion[0], pInfo->formatVersion[1]);
+        *pErrorCode=U_UNSUPPORTED_ERROR;
+        return 0;
+    }
+    tempTable.majorFormatVersion=pInfo->formatVersion[0];
+
+    /* a resource bundle must contain at least one resource item */
+    if(length<0) {
+        bundleLength=-1;
+    } else {
+        bundleLength=(length-headerSize)/4;
+
+        /* formatVersion 1.1 must have a root item and at least 5 indexes */
+        if(bundleLength<(1+5)) {
+            udata_printError(ds, "ures_swap(): too few bytes (%d after header) for a resource bundle\n",
+                             length-headerSize);
+            *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+            return 0;
+        }
+    }
+
+    inBundle=(const Resource *)((const char *)inData+headerSize);
+    rootRes=ds->readUInt32(*inBundle);
+
+    /* formatVersion 1.1 adds the indexes[] array */
+    inIndexes=(const int32_t *)(inBundle+1);
+
+    indexLength=udata_readInt32(ds, inIndexes[URES_INDEX_LENGTH])&0xff;
+    if(indexLength<=URES_INDEX_MAX_TABLE_LENGTH) {
+        udata_printError(ds, "ures_swap(): too few indexes for a 1.1+ resource bundle\n");
+        *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+        return 0;
+    }
+    keysBottom=1+indexLength;
+    keysTop=udata_readInt32(ds, inIndexes[URES_INDEX_KEYS_TOP]);
+    if(indexLength>URES_INDEX_16BIT_TOP) {
+        resBottom=udata_readInt32(ds, inIndexes[URES_INDEX_16BIT_TOP]);
+    } else {
+        resBottom=keysTop;
+    }
+    top=udata_readInt32(ds, inIndexes[URES_INDEX_BUNDLE_TOP]);
+    maxTableLength=udata_readInt32(ds, inIndexes[URES_INDEX_MAX_TABLE_LENGTH]);
+
+    if(0<=bundleLength && bundleLength(1+indexLength)) {
+        tempTable.localKeyLimit=keysTop<<2;
+    } else {
+        tempTable.localKeyLimit=0;
+    }
+
+    if(length>=0) {
+        Resource *outBundle=(Resource *)((char *)outData+headerSize);
+
+        /* track which resources we have already swapped */
+        uint32_t stackResFlags[STACK_ROW_CAPACITY];
+        int32_t resFlagsLength;
+
+        /*
+         * We need one bit per 4 resource bundle bytes so that we can track
+         * every possible Resource for whether we have swapped it already.
+         * Multiple Resource words can refer to the same bundle offsets
+         * for sharing identical values.
+         * We could optimize this by allocating only for locations above
+         * where Resource values are stored (above keys & strings).
+         */
+        resFlagsLength=(length+31)>>5;          /* number of bytes needed */
+        resFlagsLength=(resFlagsLength+3)&~3;   /* multiple of 4 bytes for uint32_t */
+        if(resFlagsLength<=(int32_t)sizeof(stackResFlags)) {
+            tempTable.resFlags=stackResFlags;
+        } else {
+            tempTable.resFlags=(uint32_t *)uprv_malloc(resFlagsLength);
+            if(tempTable.resFlags==nullptr) {
+                udata_printError(ds, "ures_swap(): unable to allocate memory for tracking resources\n");
+                *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+                return 0;
+            }
+        }
+        uprv_memset(tempTable.resFlags, 0, resFlagsLength);
+
+        /* copy the bundle for binary and inaccessible data */
+        if(inData!=outData) {
+            uprv_memcpy(outBundle, inBundle, 4*top);
+        }
+
+        /* swap the key strings, but not the padding bytes (0xaa) after the last string and its NUL */
+        udata_swapInvStringBlock(ds, inBundle+keysBottom, 4*(keysTop-keysBottom),
+                                    outBundle+keysBottom, pErrorCode);
+        if(U_FAILURE(*pErrorCode)) {
+            udata_printError(ds, "ures_swap().udata_swapInvStringBlock(keys[%d]) failed\n", 4*(keysTop-keysBottom));
+            return 0;
+        }
+
+        /* swap the 16-bit units (strings, table16, array16) */
+        if(keysTopswapArray16(ds, inBundle+keysTop, (resBottom-keysTop)*4, outBundle+keysTop, pErrorCode);
+            if(U_FAILURE(*pErrorCode)) {
+                udata_printError(ds, "ures_swap().swapArray16(16-bit units[%d]) failed\n", 2*(resBottom-keysTop));
+                return 0;
+            }
+        }
+
+        /* allocate the temporary table for sorting resource tables */
+        tempTable.keyChars=(const char *)outBundle; /* sort by outCharset */
+        if(tempTable.majorFormatVersion>1 || maxTableLength<=STACK_ROW_CAPACITY) {
+            tempTable.rows=rows;
+            tempTable.resort=resort;
+        } else {
+            tempTable.rows=(Row *)uprv_malloc(maxTableLength*sizeof(Row)+maxTableLength*4);
+            if(tempTable.rows==nullptr) {
+                udata_printError(ds, "ures_swap(): unable to allocate memory for sorting tables (max length: %d)\n",
+                                 maxTableLength);
+                *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+                if(tempTable.resFlags!=stackResFlags) {
+                    uprv_free(tempTable.resFlags);
+                }
+                return 0;
+            }
+            tempTable.resort=(int32_t *)(tempTable.rows+maxTableLength);
+        }
+
+        /* swap the resources */
+        ures_swapResource(ds, inBundle, outBundle, rootRes, nullptr, &tempTable, pErrorCode);
+        if(U_FAILURE(*pErrorCode)) {
+            udata_printError(ds, "ures_swapResource(root res=%08x) failed\n",
+                             rootRes);
+        }
+
+        if(tempTable.rows!=rows) {
+            uprv_free(tempTable.rows);
+        }
+        if(tempTable.resFlags!=stackResFlags) {
+            uprv_free(tempTable.resFlags);
+        }
+
+        /* swap the root resource and indexes */
+        ds->swapArray32(ds, inBundle, keysBottom*4, outBundle, pErrorCode);
+    }
+
+    return headerSize+4*top;
+}
diff --git a/intl/icu/source/common/uresdata.h b/intl/icu/source/common/uresdata.h
new file mode 100644
index 0000000000..171189051f
--- /dev/null
+++ b/intl/icu/source/common/uresdata.h
@@ -0,0 +1,574 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+* Copyright (C) 1999-2016, International Business Machines
+*                Corporation and others. All Rights Reserved.
+******************************************************************************
+*   file name:  uresdata.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 1999dec08
+*   created by: Markus W. Scherer
+*   06/24/02    weiv        Added support for resource sharing
+*/
+
+#ifndef __RESDATA_H__
+#define __RESDATA_H__
+
+#include "unicode/utypes.h"
+#include "unicode/udata.h"
+#include "unicode/ures.h"
+#include "putilimp.h"
+#include "udataswp.h"
+
+/**
+ * Numeric constants for internal-only types of resource items.
+ * These must use different numeric values than UResType constants
+ * because they are used together.
+ * Internal types are never returned by ures_getType().
+ */
+typedef enum {
+    /** Include a negative value so that the compiler uses the same int type as for UResType. */
+    URES_INTERNAL_NONE=-1,
+
+    /** Resource type constant for tables with 32-bit count, key offsets and values. */
+    URES_TABLE32=4,
+
+    /**
+     * Resource type constant for tables with 16-bit count, key offsets and values.
+     * All values are URES_STRING_V2 strings.
+     */
+    URES_TABLE16=5,
+
+    /** Resource type constant for 16-bit Unicode strings in formatVersion 2. */
+    URES_STRING_V2=6,
+
+    /**
+     * Resource type constant for arrays with 16-bit count and values.
+     * All values are URES_STRING_V2 strings.
+     */
+    URES_ARRAY16=9
+
+    /* Resource type 15 is not defined but effectively used by RES_BOGUS=0xffffffff. */
+} UResInternalType;
+
+/*
+ * A Resource is a 32-bit value that has 2 bit fields:
+ * 31..28   4-bit type, see enum below
+ * 27..0    28-bit four-byte-offset or value according to the type
+ */
+typedef uint32_t Resource;
+
+#define RES_BOGUS 0xffffffff
+#define RES_MAX_OFFSET 0x0fffffff
+
+#define RES_GET_TYPE(res) ((int32_t)((res)>>28UL))
+#define RES_GET_OFFSET(res) ((res)&0x0fffffff)
+#define RES_GET_POINTER(pRoot, res) ((pRoot)+RES_GET_OFFSET(res))
+
+/* get signed and unsigned integer values directly from the Resource handle
+ * NOTE: For proper logging, please use the res_getInt() constexpr
+ */
+#if U_SIGNED_RIGHT_SHIFT_IS_ARITHMETIC
+#   define RES_GET_INT_NO_TRACE(res) (((int32_t)((res)<<4L))>>4L)
+#else
+#   define RES_GET_INT_NO_TRACE(res) (int32_t)(((res)&0x08000000) ? (res)|0xf0000000 : (res)&0x07ffffff)
+#endif
+
+#define RES_GET_UINT_NO_TRACE(res) ((res)&0x0fffffff)
+
+#define URES_IS_ARRAY(type) ((int32_t)(type)==URES_ARRAY || (int32_t)(type)==URES_ARRAY16)
+#define URES_IS_TABLE(type) ((int32_t)(type)==URES_TABLE || (int32_t)(type)==URES_TABLE16 || (int32_t)(type)==URES_TABLE32)
+#define URES_IS_CONTAINER(type) (URES_IS_TABLE(type) || URES_IS_ARRAY(type))
+
+#define URES_MAKE_RESOURCE(type, offset) (((Resource)(type)<<28)|(Resource)(offset))
+#define URES_MAKE_EMPTY_RESOURCE(type) ((Resource)(type)<<28)
+
+/* indexes[] value names; indexes are generally 32-bit (Resource) indexes */
+enum {
+    /**
+     * [0] contains the length of indexes[]
+     * which is at most URES_INDEX_TOP of the latest format version
+     *
+     * formatVersion==1: all bits contain the length of indexes[]
+     *   but the length is much less than 0xff;
+     * formatVersion>1:
+     *   only bits  7..0 contain the length of indexes[],
+     *        bits 31..8 are reserved and set to 0
+     * formatVersion>=3:
+     *        bits 31..8 poolStringIndexLimit bits 23..0
+     */
+    URES_INDEX_LENGTH,
+    /**
+     * [1] contains the top of the key strings,
+     *     same as the bottom of resources or UTF-16 strings, rounded up
+     */
+    URES_INDEX_KEYS_TOP,
+    /** [2] contains the top of all resources */
+    URES_INDEX_RESOURCES_TOP,
+    /**
+     * [3] contains the top of the bundle,
+     *     in case it were ever different from [2]
+     */
+    URES_INDEX_BUNDLE_TOP,
+    /** [4] max. length of any table */
+    URES_INDEX_MAX_TABLE_LENGTH,
+    /**
+     * [5] attributes bit set, see URES_ATT_* (new in formatVersion 1.2)
+     *
+     * formatVersion>=3:
+     *   bits 31..16 poolStringIndex16Limit
+     *   bits 15..12 poolStringIndexLimit bits 27..24
+     */
+    URES_INDEX_ATTRIBUTES,
+    /**
+     * [6] top of the 16-bit units (UTF-16 string v2 UChars, URES_TABLE16, URES_ARRAY16),
+     *     rounded up (new in formatVersion 2.0, ICU 4.4)
+     */
+    URES_INDEX_16BIT_TOP,
+    /** [7] checksum of the pool bundle (new in formatVersion 2.0, ICU 4.4) */
+    URES_INDEX_POOL_CHECKSUM,
+    URES_INDEX_TOP
+};
+
+/*
+ * Nofallback attribute, attribute bit 0 in indexes[URES_INDEX_ATTRIBUTES].
+ * New in formatVersion 1.2 (ICU 3.6).
+ *
+ * If set, then this resource bundle is a standalone bundle.
+ * If not set, then the bundle participates in locale fallback, eventually
+ * all the way to the root bundle.
+ * If indexes[] is missing or too short, then the attribute cannot be determined
+ * reliably. Dependency checking should ignore such bundles, and loading should
+ * use fallbacks.
+ */
+#define URES_ATT_NO_FALLBACK 1
+
+/*
+ * Attributes for bundles that are, or use, a pool bundle.
+ * A pool bundle provides key strings that are shared among several other bundles
+ * to reduce their total size.
+ * New in formatVersion 2 (ICU 4.4).
+ */
+#define URES_ATT_IS_POOL_BUNDLE 2
+#define URES_ATT_USES_POOL_BUNDLE 4
+
+/*
+ * File format for .res resource bundle files
+ *
+ * ICU 56: New in formatVersion 3 compared with 2: -------------
+ *
+ * Resource bundles can optionally use shared string-v2 values
+ * stored in the pool bundle.
+ * If so, then the indexes[] contain two new values
+ * in previously-unused bits of existing indexes[] slots:
+ * - poolStringIndexLimit:
+ *     String-v2 offsets (in 32-bit Resource words) below this limit
+ *     point to pool bundle string-v2 values.
+ * - poolStringIndex16Limit:
+ *     Resource16 string-v2 offsets below this limit
+ *     point to pool bundle string-v2 values.
+ * Guarantee: poolStringIndex16Limit <= poolStringIndexLimit
+ *
+ * The local bundle's poolStringIndexLimit is greater than
+ * any pool bundle string index used in the local bundle.
+ * The poolStringIndexLimit should not be greater than
+ * the maximum possible pool bundle string index.
+ *
+ * The maximum possible pool bundle string index is the index to the last non-NUL
+ * pool string character, due to suffix sharing.
+ *
+ * In the pool bundle, there is no structure that lists the strings.
+ * (The root resource is an empty Table.)
+ * If the strings need to be enumerated (as genrb --usePoolBundle does),
+ * then iterate through the pool bundle's 16-bit-units array from the beginning.
+ * Stop at the end of the array, or when an explicit or implicit string length
+ * would lead beyond the end of the array,
+ * or when an apparent string is not NUL-terminated.
+ * (Future genrb version might terminate the strings with
+ * what looks like a large explicit string length.)
+ *
+ * ICU 4.4: New in formatVersion 2 compared with 1.3: -------------
+ *
+ * Three new resource types -- String-v2, Table16 and Array16 -- have their
+ * values stored in a new array of 16-bit units between the table key strings
+ * and the start of the other resources.
+ *
+ * genrb eliminates duplicates among Unicode string-v2 values.
+ * Multiple Unicode strings may use the same offset and string data,
+ * or a short string may point to the suffix of a longer string. ("Suffix sharing")
+ * For example, one string "abc" may be reused for another string "bc" by pointing
+ * to the second character. (Short strings-v2 are NUL-terminated
+ * and not preceded by an explicit length value.)
+ *
+ * It is allowed for all resource types to share values.
+ * The swapper code (ures_swap()) has been modified so that it swaps each item
+ * exactly once.
+ *
+ * A resource bundle may use a special pool bundle. Some or all of the table key strings
+ * of the using-bundle are omitted, and the key string offsets for such key strings refer
+ * to offsets in the pool bundle.
+ * The using-bundle's and the pool-bundle's indexes[URES_INDEX_POOL_CHECKSUM] values
+ * must match.
+ * Two bits in indexes[URES_INDEX_ATTRIBUTES] indicate whether a resource bundle
+ * is or uses a pool bundle.
+ *
+ * Table key strings must be compared in ASCII order, even if they are not
+ * stored in ASCII.
+ *
+ * New in formatVersion 1.3 compared with 1.2: -------------
+ *
+ * genrb eliminates duplicates among key strings.
+ * Multiple table items may share one key string, or one item may point
+ * to the suffix of another's key string. ("Suffix sharing")
+ * For example, one key "abc" may be reused for another key "bc" by pointing
+ * to the second character. (Key strings are NUL-terminated.)
+ *
+ * -------------
+ *
+ * An ICU4C resource bundle file (.res) is a binary, memory-mappable file
+ * with nested, hierarchical data structures.
+ * It physically contains the following:
+ *
+ *   Resource root; -- 32-bit Resource item, root item for this bundle's tree;
+ *                     currently, the root item must be a table or table32 resource item
+ *   int32_t indexes[indexes[0]]; -- array of indexes for friendly
+ *                                   reading and swapping; see URES_INDEX_* above
+ *                                   new in formatVersion 1.1 (ICU 2.8)
+ *   char keys[]; -- characters for key strings
+ *                   (formatVersion 1.0: up to 65k of characters; 1.1: <2G)
+ *                   (minus the space for root and indexes[]),
+ *                   which consist of invariant characters (ASCII/EBCDIC) and are NUL-terminated;
+ *                   padded to multiple of 4 bytes for 4-alignment of the following data
+ *   uint16_t 16BitUnits[]; -- resources that are stored entirely as sequences of 16-bit units
+ *                             (new in formatVersion 2/ICU 4.4)
+ *                             data is indexed by the offset values in 16-bit resource types,
+ *                             with offset 0 pointing to the beginning of this array;
+ *                             there is a 0 at offset 0, for empty resources;
+ *                             padded to multiple of 4 bytes for 4-alignment of the following data
+ *   data; -- data directly and indirectly indexed by the root item;
+ *            the structure is determined by walking the tree
+ *
+ * Each resource bundle item has a 32-bit Resource handle (see typedef above)
+ * which contains the item type number in its upper 4 bits (31..28) and either
+ * an offset or a direct value in its lower 28 bits (27..0).
+ * The order of items is undefined and only determined by walking the tree.
+ * Leaves of the tree may be stored first or last or anywhere in between,
+ * and it is in theory possible to have unreferenced holes in the file.
+ *
+ * 16-bit-unit values:
+ * Starting with formatVersion 2/ICU 4.4, some resources are stored in a special
+ * array of 16-bit units. Each resource value is a sequence of 16-bit units,
+ * with no per-resource padding to a 4-byte boundary.
+ * 16-bit container types (Table16 and Array16) contain Resource16 values
+ * which are offsets to String-v2 resources in the same 16-bit-units array.
+ *
+ * Direct values:
+ * - Empty Unicode strings have an offset value of 0 in the Resource handle itself.
+ * - Starting with formatVersion 2/ICU 4.4, an offset value of 0 for
+ *   _any_ resource type indicates an empty value.
+ * - Integer values are 28-bit values stored in the Resource handle itself;
+ *   the interpretation of unsigned vs. signed integers is up to the application.
+ *
+ * All other types and values use 28-bit offsets to point to the item's data.
+ * The offset is an index to the first 32-bit word of the value, relative to the
+ * start of the resource data (i.e., the root item handle is at offset 0).
+ * To get byte offsets, the offset is multiplied by 4 (or shifted left by 2 bits).
+ * All resource item values are 4-aligned.
+ *
+ * New in formatVersion 2/ICU 4.4: Some types use offsets into the 16-bit-units array,
+ * indexing 16-bit units in that array.
+ *
+ * The structures (memory layouts) for the values for each item type are listed
+ * in the table below.
+ *
+ * Nested, hierarchical structures: -------------
+ *
+ * Table items contain key-value pairs where the keys are offsets to char * key strings.
+ * The values of these pairs are either Resource handles or
+ * offsets into the 16-bit-units array, depending on the table type.
+ *
+ * Array items are simple vectors of Resource handles,
+ * or of offsets into the 16-bit-units array, depending on the array type.
+ *
+ * Table key string offsets: -------
+ *
+ * Key string offsets are relative to the start of the resource data (of the root handle),
+ * i.e., the first string has an offset of 4+sizeof(indexes).
+ * (After the 4-byte root handle and after the indexes array.)
+ *
+ * If the resource bundle uses a pool bundle, then some key strings are stored
+ * in the pool bundle rather than in the local bundle itself.
+ * - In a Table or Table16, the 16-bit key string offset is local if it is
+ *   less than indexes[URES_INDEX_KEYS_TOP]<<2.
+ *   Otherwise, subtract indexes[URES_INDEX_KEYS_TOP]<<2 to get the offset into
+ *   the pool bundle key strings.
+ * - In a Table32, the 32-bit key string offset is local if it is non-negative.
+ *   Otherwise, reset bit 31 to get the pool key string offset.
+ *
+ * Unlike the local offset, the pool key offset is relative to
+ * the start of the key strings, not to the start of the bundle.
+ *
+ * An alias item is special (and new in ICU 2.4): --------------
+ *
+ * Its memory layout is just like for a UnicodeString, but at runtime it resolves to
+ * another resource bundle's item according to the path in the string.
+ * This is used to share items across bundles that are in different lookup/fallback
+ * chains (e.g., large collation data among zh_TW and zh_HK).
+ * This saves space (for large items) and maintenance effort (less duplication of data).
+ *
+ * --------------------------------------------------------------------------
+ *
+ * Resource types:
+ *
+ * Most resources have their values stored at four-byte offsets from the start
+ * of the resource data. These values are at least 4-aligned.
+ * Some resource values are stored directly in the offset field of the Resource itself.
+ * See UResType in unicode/ures.h for enumeration constants for Resource types.
+ *
+ * Some resources have their values stored as sequences of 16-bit units,
+ * at 2-byte offsets from the start of a contiguous 16-bit-unit array between
+ * the table key strings and the other resources. (new in formatVersion 2/ICU 4.4)
+ * At offset 0 of that array is a 16-bit zero value for empty 16-bit resources.
+ *
+ * Resource16 values in Table16 and Array16 are 16-bit offsets to String-v2
+ * resources, with the offsets relative to the start of the 16-bit-units array.
+ * Starting with formatVersion 3/ICU 56, if offset(URES_NONE)),
+        fTraceInfo() {}
+    virtual ~ResourceDataValue();
+
+    void setData(const ResourceData &data) {
+        pResData = &data;
+    }
+    
+    void setValidLocaleDataEntry(UResourceDataEntry *entry) {
+        validLocaleDataEntry = entry;
+    }
+
+    void setResource(Resource r, ResourceTracer&& traceInfo) {
+        res = r;
+        fTraceInfo = traceInfo;
+    }
+
+    const ResourceData &getData() const { return *pResData; }
+    UResourceDataEntry *getValidLocaleDataEntry() const { return validLocaleDataEntry; }
+    Resource getResource() const { return res; }
+    virtual UResType getType() const override;
+    virtual const char16_t *getString(int32_t &length, UErrorCode &errorCode) const override;
+    virtual const char16_t *getAliasString(int32_t &length, UErrorCode &errorCode) const override;
+    virtual int32_t getInt(UErrorCode &errorCode) const override;
+    virtual uint32_t getUInt(UErrorCode &errorCode) const override;
+    virtual const int32_t *getIntVector(int32_t &length, UErrorCode &errorCode) const override;
+    virtual const uint8_t *getBinary(int32_t &length, UErrorCode &errorCode) const override;
+    virtual ResourceArray getArray(UErrorCode &errorCode) const override;
+    virtual ResourceTable getTable(UErrorCode &errorCode) const override;
+    virtual UBool isNoInheritanceMarker() const override;
+    virtual int32_t getStringArray(UnicodeString *dest, int32_t capacity,
+                                   UErrorCode &errorCode) const override;
+    virtual int32_t getStringArrayOrStringAsArray(UnicodeString *dest, int32_t capacity,
+                                                  UErrorCode &errorCode) const override;
+    virtual UnicodeString getStringOrFirstOfArray(UErrorCode &errorCode) const override;
+
+private:
+    const ResourceData *pResData;
+    UResourceDataEntry *validLocaleDataEntry;
+    Resource res;
+    ResourceTracer fTraceInfo;
+};
+
+U_NAMESPACE_END
+
+#endif  /* __cplusplus */
+
+/**
+ * Swap an ICU resource bundle. See udataswp.h.
+ * @internal
+ */
+U_CAPI int32_t U_EXPORT2
+ures_swap(const UDataSwapper *ds,
+          const void *inData, int32_t length, void *outData,
+          UErrorCode *pErrorCode);
+
+#endif
diff --git a/intl/icu/source/common/uresimp.h b/intl/icu/source/common/uresimp.h
new file mode 100644
index 0000000000..8a1679717e
--- /dev/null
+++ b/intl/icu/source/common/uresimp.h
@@ -0,0 +1,397 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+*   Copyright (C) 2000-2016, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*/
+
+#ifndef URESIMP_H
+#define URESIMP_H
+
+#include "unicode/ures.h"
+#include "unicode/utypes.h"
+
+#include "uresdata.h"
+
+#define kRootLocaleName         "root"
+#define kPoolBundleName         "pool"
+
+/*
+ The default minor version and the version separator must be exactly one
+ character long.
+*/
+
+#define kDefaultMinorVersion    "0"
+#define kVersionSeparator       "."
+#define kVersionTag             "Version"
+
+#define MAGIC1 19700503
+#define MAGIC2 19641227
+
+#define URES_MAX_ALIAS_LEVEL 256
+
+#define EMPTY_SET 0x2205
+
+struct UResourceDataEntry;
+typedef struct UResourceDataEntry UResourceDataEntry;
+
+/*
+ * Note: If we wanted to make this structure smaller, then we could try
+ * to use one UResourceDataEntry pointer for fAlias and fPool, with a separate
+ * flag to distinguish whether this struct is for a real bundle with a pool,
+ * or for an alias entry for which we won't use the pool after loading.
+ */
+struct UResourceDataEntry {
+    char *fName; /* name of the locale for bundle - still to decide whether it is original or fallback */
+    char *fPath; /* path to bundle - used for distinguishing between resources with the same name */
+    UResourceDataEntry *fParent; /*next resource in fallback chain*/
+    UResourceDataEntry *fAlias;
+    UResourceDataEntry *fPool;
+    ResourceData fData; /* data for low level access */
+    char fNameBuffer[3]; /* A small buffer of free space for fName. The free space is due to struct padding. */
+    uint32_t fCountExisting; /* how much is this resource used */
+    UErrorCode fBogus;
+    /* int32_t fHashKey;*/ /* for faster access in the hashtable */
+};
+
+#define RES_BUFSIZE 64
+#define RES_PATH_SEPARATOR   '/'
+#define RES_PATH_SEPARATOR_S   "/"
+
+U_CAPI void U_EXPORT2 ures_initStackObject(UResourceBundle* resB);
+
+#ifdef __cplusplus
+
+struct UResourceBundle {
+    const char *fKey; /*tag*/
+    /**
+     * The dataEntry for the actual locale in which this item lives.
+     * Used for accessing the item's data.
+     * Non-const pointer for reference counting via entryIncrease().
+     */
+    UResourceDataEntry *fData; /*for low-level access*/
+    char *fVersion;
+    /**
+     * The dataEntry for the valid locale.
+     * Used for /LOCALE/path alias resolution that starts back from the valid locale,
+     * rather than from the actual locale of this item which might live in
+     * an ancestor bundle.
+     */
+    UResourceDataEntry *fValidLocaleDataEntry;
+    char *fResPath; /* full path to the resource: "zh_TW/CollationElements/Sequence" */
+    char fResBuf[RES_BUFSIZE];
+    int32_t fResPathLen;
+    Resource fRes;
+    UBool fHasFallback;
+    UBool fIsTopLevel;
+    uint32_t fMagic1;   /* For determining if it's a stack object */
+    uint32_t fMagic2;   /* For determining if it's a stack object */
+    int32_t fIndex;
+    int32_t fSize;
+
+    inline const ResourceData &getResData() const { return fData->fData; }
+};
+
+U_NAMESPACE_BEGIN
+
+/**
+ * \class StackUResourceBundle
+ * "Smart pointer" like class, closes a UResourceBundle via ures_close().
+ *
+ * This code:
+ *
+ *   StackUResourceBundle bundle;
+ *   foo(bundle.getAlias());
+ *
+ * Is equivalent to this code:
+ *
+ *   UResourceBundle bundle;
+ *   ures_initStackObject(&bundle);
+ *   foo(&bundle);
+ *   ures_close(&bundle);
+ *
+ * @see LocalUResourceBundlePointer
+ * @internal
+ */
+class U_COMMON_API StackUResourceBundle {
+public:
+    // No heap allocation. Use only on the stack.
+    static void* U_EXPORT2 operator new(size_t) noexcept = delete;
+    static void* U_EXPORT2 operator new[](size_t) noexcept = delete;
+#if U_HAVE_PLACEMENT_NEW
+    static void* U_EXPORT2 operator new(size_t, void*) noexcept = delete;
+#endif
+
+    StackUResourceBundle();
+    ~StackUResourceBundle();
+
+    UResourceBundle* getAlias() { return &bundle; }
+
+    UResourceBundle& ref() { return bundle; }
+    const UResourceBundle& ref() const { return bundle; }
+
+    StackUResourceBundle(const StackUResourceBundle&) = delete;
+    StackUResourceBundle& operator=(const StackUResourceBundle&) = delete;
+
+    StackUResourceBundle(StackUResourceBundle&&) = delete;
+    StackUResourceBundle& operator=(StackUResourceBundle&&) = delete;
+
+private:
+    UResourceBundle bundle;
+};
+
+U_NAMESPACE_END
+
+#endif  /* __cplusplus */
+
+/**
+ * Opens a resource bundle for the locale;
+ * if there is not even a base language bundle, then loads the root bundle;
+ * never falls back to the default locale.
+ *
+ * This is used for algorithms that have good pan-Unicode default behavior,
+ * such as case mappings, collation, and segmentation (BreakIterator).
+ */
+U_CAPI UResourceBundle* U_EXPORT2
+ures_openNoDefault(const char* path, const char* localeID, UErrorCode* status);
+
+/* Some getters used by the copy constructor */
+U_CFUNC const char* ures_getName(const UResourceBundle* resB);
+#ifdef URES_DEBUG
+U_CFUNC const char* ures_getPath(const UResourceBundle* resB);
+/**
+ * If anything was in the RB cache, dump it to the screen.
+ * @return true if there was anything into the cache
+ */
+U_CAPI UBool U_EXPORT2 ures_dumpCacheContents(void);
+#endif
+
+/* Candidates for export */
+U_CFUNC UResourceBundle *ures_copyResb(UResourceBundle *r, const UResourceBundle *original, UErrorCode *status);
+
+/**
+ * Returns a resource that can be located using the pathToResource argument. One needs optional package, locale
+ * and path inside the locale, for example: "/myData/en/zoneStrings/3". Keys and indexes are supported. Keys
+ * need to reference data in named structures, while indexes can reference both named and anonymous resources.
+ * Features a fill-in parameter. 
+ * 
+ * Note, this function does NOT have a syntax for specifying items within a tree.  May want to consider a
+ * syntax that delineates between package/tree and resource.  
+ *
+ * @param pathToResource    a path that will lead to the requested resource
+ * @param fillIn            if NULL a new UResourceBundle struct is allocated and must be deleted by the caller.
+ *                          Alternatively, you can supply a struct to be filled by this function.
+ * @param status            fills in the outgoing error code.
+ * @return                  a pointer to a UResourceBundle struct. If fill in param was NULL, caller must delete it
+ */
+U_CAPI UResourceBundle* U_EXPORT2
+ures_findResource(const char* pathToResource, 
+                  UResourceBundle *fillIn, UErrorCode *status); 
+
+/**
+ * Returns a sub resource that can be located using the pathToResource argument. One needs a path inside 
+ * the supplied resource, for example, if you have "en_US" resource bundle opened, you might ask for
+ * "zoneStrings/3". Keys and indexes are supported. Keys
+ * need to reference data in named structures, while indexes can reference both 
+ * named and anonymous resources.
+ * Features a fill-in parameter. 
+ *
+ * @param resourceBundle    a resource
+ * @param pathToResource    a path that will lead to the requested resource
+ * @param fillIn            if NULL a new UResourceBundle struct is allocated and must be deleted by the caller.
+ *                          Alternatively, you can supply a struct to be filled by this function.
+ * @param status            fills in the outgoing error code.
+ * @return                  a pointer to a UResourceBundle struct. If fill in param was NULL, caller must delete it
+ */
+U_CAPI UResourceBundle* U_EXPORT2
+ures_findSubResource(const UResourceBundle *resB, 
+                     char* pathToResource, 
+                     UResourceBundle *fillIn, UErrorCode *status);
+
+/**
+ * Returns a functionally equivalent locale (considering keywords) for the specified keyword.
+ * @param result fillin for the equivalent locale
+ * @param resultCapacity capacity of the fillin buffer
+ * @param path path to the tree, or NULL for ICU data
+ * @param resName top level resource. Example: "collations"
+ * @param keyword locale keyword. Example: "collation"
+ * @param locid The requested locale
+ * @param isAvailable If non-null, pointer to fillin parameter that indicates whether the 
+ * requested locale was available. The locale is defined as 'available' if it physically 
+ * exists within the specified tree.
+ * @param omitDefault if true, omit keyword and value if default. 'de_DE\@collation=standard' -> 'de_DE'
+ * @param status error code
+ * @return  the actual buffer size needed for the full locale.  If it's greater 
+ * than resultCapacity, the returned full name will be truncated and an error code will be returned.
+ */
+U_CAPI int32_t U_EXPORT2
+ures_getFunctionalEquivalent(char *result, int32_t resultCapacity, 
+                             const char *path, const char *resName, const char *keyword, const char *locid,
+                             UBool *isAvailable, UBool omitDefault, UErrorCode *status);
+
+/**
+ * Given a tree path and keyword, return a string enumeration of all possible values for that keyword.
+ * @param path path to the tree, or NULL for ICU data
+ * @param keyword a particular keyword to consider, must match a top level resource name 
+ * within the tree.
+ * @param status error code
+ */
+U_CAPI UEnumeration* U_EXPORT2
+ures_getKeywordValues(const char *path, const char *keyword, UErrorCode *status);
+
+
+/**
+ * Get a resource with multi-level fallback. Normally only the top level resources will
+ * fallback to its parent. This performs fallback on subresources. For example, when a table
+ * is defined in a resource bundle and a parent resource bundle, normally no fallback occurs
+ * on the sub-resources because the table is defined in the current resource bundle, but this
+ * function can perform fallback on the sub-resources of the table.
+ * @param resB              a resource
+ * @param inKey             a key associated with the requested resource
+ * @param fillIn            if NULL a new UResourceBundle struct is allocated and must be deleted by the caller.
+ *                          Alternatively, you can supply a struct to be filled by this function.
+ * @param status: fills in the outgoing error code
+ *                could be U_MISSING_RESOURCE_ERROR if the key is not found
+ *                could be a non-failing error 
+ *                e.g.: U_USING_FALLBACK_WARNING,U_USING_DEFAULT_WARNING 
+ * @return                  a pointer to a UResourceBundle struct. If fill in param was NULL, caller must delete it
+ */
+U_CAPI UResourceBundle* U_EXPORT2 
+ures_getByKeyWithFallback(const UResourceBundle *resB, 
+                          const char* inKey, 
+                          UResourceBundle *fillIn, 
+                          UErrorCode *status);
+
+/**
+ * Get a String with multi-level fallback. Normally only the top level resources will
+ * fallback to its parent. This performs fallback on subresources. For example, when a table
+ * is defined in a resource bundle and a parent resource bundle, normally no fallback occurs
+ * on the sub-resources because the table is defined in the current resource bundle, but this
+ * function can perform fallback on the sub-resources of the table.
+ * @param resB              a resource
+ * @param inKey             a key associated with the requested resource
+ * @param len               if not NULL, used to return the length of the string
+ * @param status: fills in the outgoing error code
+ *                could be U_MISSING_RESOURCE_ERROR if the key is not found
+ *                could be a non-failing error 
+ *                e.g.: U_USING_FALLBACK_WARNING,U_USING_DEFAULT_WARNING 
+ * @return returns a pointer to a zero-terminated UChar array which lives in a
+ *         memory mapped/DLL file.
+ */
+U_CAPI const UChar* U_EXPORT2 
+ures_getStringByKeyWithFallback(const UResourceBundle *resB, 
+                          const char* inKey,  
+                          int32_t* len,
+                          UErrorCode *status);
+
+#ifdef __cplusplus
+
+U_CAPI void U_EXPORT2
+ures_getValueWithFallback(const UResourceBundle *bundle, const char *path,
+                          UResourceBundle *tempFillIn,
+                          icu::ResourceDataValue &value, UErrorCode &errorCode);
+
+/**
+ * Locates the resource specified by `path` in the resource bundle specified by `bundle` (performing any
+ * necessary fallback and following any aliases) and calls the specified `sink`'s `put()` method with that
+ * resource.  Then walks the bundle's parent chain, calling `put()` on the sink for each item in the
+ * parent chain.
+ * @param bundle The bundle to search
+ * @param path The path of the desired resource
+ * @param sink A `ResourceSink` that gets called for each resource in the parent chain
+ * @param errorCode The error code
+ */
+U_CAPI void U_EXPORT2
+ures_getAllItemsWithFallback(const UResourceBundle *bundle, const char *path,
+                             icu::ResourceSink &sink, UErrorCode &errorCode);
+
+/**
+ * Locates the resource specified by `path` in the resource bundle specified by `bundle` (performing any
+ * necessary fallback and following any aliases) and, if the resource is a table resource, iterates over its
+ * immediate child resources (again, following any aliases to get the individual resource values), and calls the specified
+ * `sink`'s `put()` method for each child resource (passing it that resource's key and either its actual value or,
+ * if that value is an alias, the value you get by following the alias).  Then walks back over the bundle's
+ * parent chain, similarly iterating over each parent table resource's child resources.
+ * Does not descend beyond one level of table children.
+ * @param bundle The bundle to search
+ * @param path The path of the desired resource
+ * @param sink A `ResourceSink` that gets called for each child resource of the specified resource (and each child
+ * of the resources in its parent chain).
+ * @param errorCode The error code.  This will be U_RESOURCE_TYPE_MISMATCH if the resource the caller
+ * is asking for isn't a table resource.
+ */
+U_CAPI void U_EXPORT2
+ures_getAllChildrenWithFallback(const UResourceBundle *bundle, const char *path,
+                                icu::ResourceSink &sink, UErrorCode &errorCode);
+
+#endif  /* __cplusplus */
+
+/**
+ * Get a version number by key
+ * @param resB bundle containing version number
+ * @param key the key for the version number
+ * @param ver fillin for the version number
+ * @param status error code
+ */
+U_CAPI void U_EXPORT2
+ures_getVersionByKey(const UResourceBundle *resB,
+                     const char *key,
+                     UVersionInfo ver,
+                     UErrorCode *status);
+
+
+/**
+ * Internal function.
+ * Return the version number associated with this ResourceBundle as a string.
+ *
+ * @param resourceBundle The resource bundle for which the version is checked.
+ * @return  A version number string as specified in the resource bundle or its parent.
+ *          The caller does not own this string.
+ * @see ures_getVersion
+ */
+U_CAPI const char* U_EXPORT2 
+ures_getVersionNumberInternal(const UResourceBundle *resourceBundle);
+
+/**
+ * Return the name of the Locale associated with this ResourceBundle. This API allows
+ * you to query for the real locale of the resource. For example, if you requested 
+ * "en_US_CALIFORNIA" and only "en_US" bundle exists, "en_US" will be returned. 
+ * For subresources, the locale where this resource comes from will be returned.
+ * If fallback has occurred, getLocale will reflect this.
+ *
+ * This internal version avoids deprecated-warnings in ICU code.
+ *
+ * @param resourceBundle resource bundle in question
+ * @param status just for catching illegal arguments
+ * @return  A Locale name
+ */
+U_CAPI const char* U_EXPORT2 
+ures_getLocaleInternal(const UResourceBundle* resourceBundle, 
+               UErrorCode* status);
+
+/**
+ * Same as ures_openDirect() but uses the fill-in parameter instead of allocating a new bundle.
+ * 
+ * @param r The existing UResourceBundle to fill in. If NULL then status will be
+ *          set to U_ILLEGAL_ARGUMENT_ERROR.
+ * @param packageName   The packageName and locale together point to an ICU udata object,
+ *                      as defined by  udata_open( packageName, "res", locale, err) 
+ *                      or equivalent.  Typically, packageName will refer to a (.dat) file, or to
+ *                      a package registered with udata_setAppData(). Using a full file or directory
+ *                      pathname for packageName is deprecated. If NULL, ICU data will be used.
+ * @param locale  specifies the locale for which we want to open the resource
+ *                if NULL, the default locale will be used. If strlen(locale) == 0
+ *                root locale will be used.
+ * @param status The error code.
+ * @see ures_openDirect
+ * @internal
+ */
+U_CAPI void U_EXPORT2
+ures_openDirectFillIn(UResourceBundle *r,
+                      const char *packageName,
+                      const char *locale,
+                      UErrorCode *status);
+
+#endif /*URESIMP_H*/
diff --git a/intl/icu/source/common/ureslocs.h b/intl/icu/source/common/ureslocs.h
new file mode 100644
index 0000000000..f7c3344ef2
--- /dev/null
+++ b/intl/icu/source/common/ureslocs.h
@@ -0,0 +1,27 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+*   Copyright (C) 2009-2014 International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*/
+
+#ifndef __URESLOCS_H__
+#define __URESLOCS_H__
+
+#include "unicode/utypes.h"
+#include "unicode/udata.h"
+
+U_CDECL_BEGIN
+
+
+#define U_ICUDATA_LANG U_ICUDATA_NAME U_TREE_SEPARATOR_STRING "lang"
+#define U_ICUDATA_REGION U_ICUDATA_NAME U_TREE_SEPARATOR_STRING "region"
+#define U_ICUDATA_CURR U_ICUDATA_NAME U_TREE_SEPARATOR_STRING "curr"
+#define U_ICUDATA_ZONE U_ICUDATA_NAME U_TREE_SEPARATOR_STRING "zone"
+#define U_ICUDATA_UNIT U_ICUDATA_NAME U_TREE_SEPARATOR_STRING "unit"
+
+U_CDECL_END
+
+#endif
diff --git a/intl/icu/source/common/usc_impl.cpp b/intl/icu/source/common/usc_impl.cpp
new file mode 100644
index 0000000000..cd5a4e5b8d
--- /dev/null
+++ b/intl/icu/source/common/usc_impl.cpp
@@ -0,0 +1,361 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+*   Copyright (C) 1999-2016, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*
+* File USC_IMPL.C
+*
+* Modification History:
+*
+*   Date        Name        Description
+*   07/08/2002  Eric Mader  Creation.
+******************************************************************************
+*/
+
+#include "unicode/uscript.h"
+#include "usc_impl.h"
+#include "cmemory.h"
+
+#define PAREN_STACK_DEPTH 32
+
+#define MOD(sp) ((sp) % PAREN_STACK_DEPTH)
+#define LIMIT_INC(sp) (((sp) < PAREN_STACK_DEPTH)? (sp) + 1 : PAREN_STACK_DEPTH)
+#define INC(sp,count) (MOD((sp) + (count)))
+#define INC1(sp) (INC(sp, 1))
+#define DEC(sp,count) (MOD((sp) + PAREN_STACK_DEPTH - (count)))
+#define DEC1(sp) (DEC(sp, 1))
+#define STACK_IS_EMPTY(scriptRun) ((scriptRun)->pushCount <= 0)
+#define STACK_IS_NOT_EMPTY(scriptRun) (! STACK_IS_EMPTY(scriptRun))
+#define TOP(scriptRun) ((scriptRun)->parenStack[(scriptRun)->parenSP])
+#define SYNC_FIXUP(scriptRun) ((scriptRun)->fixupCount = 0)
+
+struct ParenStackEntry
+{
+    int32_t pairIndex;
+    UScriptCode scriptCode;
+};
+
+struct UScriptRun
+{
+    int32_t textLength;
+    const char16_t *textArray;
+
+    int32_t scriptStart;
+    int32_t scriptLimit;
+    UScriptCode scriptCode;
+
+    struct ParenStackEntry parenStack[PAREN_STACK_DEPTH];
+    int32_t parenSP;
+    int32_t pushCount;
+    int32_t fixupCount;
+};
+
+static int8_t highBit(int32_t value);
+
+static const UChar32 pairedChars[] = {
+    0x0028, 0x0029, /* ascii paired punctuation */
+    0x003c, 0x003e,
+    0x005b, 0x005d,
+    0x007b, 0x007d,
+    0x00ab, 0x00bb, /* guillemets */
+    0x2018, 0x2019, /* general punctuation */
+    0x201c, 0x201d,
+    0x2039, 0x203a,
+    0x3008, 0x3009, /* chinese paired punctuation */
+    0x300a, 0x300b,
+    0x300c, 0x300d,
+    0x300e, 0x300f,
+    0x3010, 0x3011,
+    0x3014, 0x3015,
+    0x3016, 0x3017,
+    0x3018, 0x3019,
+    0x301a, 0x301b
+};
+
+static void push(UScriptRun *scriptRun, int32_t pairIndex, UScriptCode scriptCode)
+{
+    scriptRun->pushCount  = LIMIT_INC(scriptRun->pushCount);
+    scriptRun->fixupCount = LIMIT_INC(scriptRun->fixupCount);
+    
+    scriptRun->parenSP = INC1(scriptRun->parenSP);
+    scriptRun->parenStack[scriptRun->parenSP].pairIndex  = pairIndex;
+    scriptRun->parenStack[scriptRun->parenSP].scriptCode = scriptCode;
+}
+
+static void pop(UScriptRun *scriptRun)
+{
+    if (STACK_IS_EMPTY(scriptRun)) {
+        return;
+    }
+    
+    if (scriptRun->fixupCount > 0) {
+        scriptRun->fixupCount -= 1;
+    }
+    
+    scriptRun->pushCount -= 1;
+    scriptRun->parenSP = DEC1(scriptRun->parenSP);
+    
+    /* If the stack is now empty, reset the stack
+       pointers to their initial values.
+     */
+    if (STACK_IS_EMPTY(scriptRun)) {
+        scriptRun->parenSP = -1;
+    }
+}
+
+static void fixup(UScriptRun *scriptRun, UScriptCode scriptCode)
+{
+    int32_t fixupSP = DEC(scriptRun->parenSP, scriptRun->fixupCount);
+    
+    while (scriptRun->fixupCount-- > 0) {
+        fixupSP = INC1(fixupSP);
+        scriptRun->parenStack[fixupSP].scriptCode = scriptCode;
+    }
+}
+
+static int8_t
+highBit(int32_t value)
+{
+    int8_t bit = 0;
+
+    if (value <= 0) {
+        return -32;
+    }
+
+    if (value >= 1 << 16) {
+        value >>= 16;
+        bit += 16;
+    }
+
+    if (value >= 1 << 8) {
+        value >>= 8;
+        bit += 8;
+    }
+
+    if (value >= 1 << 4) {
+        value >>= 4;
+        bit += 4;
+    }
+
+    if (value >= 1 << 2) {
+        value >>= 2;
+        bit += 2;
+    }
+
+    if (value >= 1 << 1) {
+        //value >>= 1;
+        bit += 1;
+    }
+
+    return bit;
+}
+
+static int32_t
+getPairIndex(UChar32 ch)
+{
+    int32_t pairedCharCount = UPRV_LENGTHOF(pairedChars);
+    int32_t pairedCharPower = 1 << highBit(pairedCharCount);
+    int32_t pairedCharExtra = pairedCharCount - pairedCharPower;
+
+    int32_t probe = pairedCharPower;
+    int32_t pairIndex = 0;
+
+    if (ch >= pairedChars[pairedCharExtra]) {
+        pairIndex = pairedCharExtra;
+    }
+
+    while (probe > (1 << 0)) {
+        probe >>= 1;
+
+        if (ch >= pairedChars[pairIndex + probe]) {
+            pairIndex += probe;
+        }
+    }
+
+    if (pairedChars[pairIndex] != ch) {
+        pairIndex = -1;
+    }
+
+    return pairIndex;
+}
+
+static UBool
+sameScript(UScriptCode scriptOne, UScriptCode scriptTwo)
+{
+    return scriptOne <= USCRIPT_INHERITED || scriptTwo <= USCRIPT_INHERITED || scriptOne == scriptTwo;
+}
+
+U_CAPI UScriptRun * U_EXPORT2
+uscript_openRun(const char16_t *src, int32_t length, UErrorCode *pErrorCode)
+{
+    UScriptRun *result = nullptr;
+
+    if (pErrorCode == nullptr || U_FAILURE(*pErrorCode)) {
+        return nullptr;
+    }
+
+    result = (UScriptRun *)uprv_malloc(sizeof (UScriptRun));
+
+    if (result == nullptr) {
+        *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
+        return nullptr;
+    }
+
+    uscript_setRunText(result, src, length, pErrorCode);
+
+    /* Release the UScriptRun if uscript_setRunText() returns an error */
+    if (U_FAILURE(*pErrorCode)) {
+        uprv_free(result);
+        result = nullptr;
+    }
+
+    return result;
+}
+
+U_CAPI void U_EXPORT2
+uscript_closeRun(UScriptRun *scriptRun)
+{
+    if (scriptRun != nullptr) {
+        uprv_free(scriptRun);
+    }
+}
+
+U_CAPI void U_EXPORT2
+uscript_resetRun(UScriptRun *scriptRun)
+{
+    if (scriptRun != nullptr) {
+        scriptRun->scriptStart = 0;
+        scriptRun->scriptLimit = 0;
+        scriptRun->scriptCode  = USCRIPT_INVALID_CODE;
+        scriptRun->parenSP     = -1;
+        scriptRun->pushCount   =  0;
+        scriptRun->fixupCount  =  0;
+    }
+}
+
+U_CAPI void U_EXPORT2
+uscript_setRunText(UScriptRun *scriptRun, const char16_t *src, int32_t length, UErrorCode *pErrorCode)
+{
+    if (pErrorCode == nullptr || U_FAILURE(*pErrorCode)) {
+        return;
+    }
+
+    if (scriptRun == nullptr || length < 0 || ((src == nullptr) != (length == 0))) {
+        *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+
+    scriptRun->textArray  = src;
+    scriptRun->textLength = length;
+
+    uscript_resetRun(scriptRun);
+}
+
+U_CAPI UBool U_EXPORT2
+uscript_nextRun(UScriptRun *scriptRun, int32_t *pRunStart, int32_t *pRunLimit, UScriptCode *pRunScript)
+{
+    UErrorCode error = U_ZERO_ERROR;
+
+    /* if we've fallen off the end of the text, we're done */
+    if (scriptRun == nullptr || scriptRun->scriptLimit >= scriptRun->textLength) {
+        return false;
+    }
+    
+    SYNC_FIXUP(scriptRun);
+    scriptRun->scriptCode = USCRIPT_COMMON;
+
+    for (scriptRun->scriptStart = scriptRun->scriptLimit; scriptRun->scriptLimit < scriptRun->textLength; scriptRun->scriptLimit += 1) {
+        char16_t high = scriptRun->textArray[scriptRun->scriptLimit];
+        UChar32  ch   = high;
+        UScriptCode sc;
+        int32_t pairIndex;
+
+        /*
+         * if the character is a high surrogate and it's not the last one
+         * in the text, see if it's followed by a low surrogate
+         */
+        if (high >= 0xD800 && high <= 0xDBFF && scriptRun->scriptLimit < scriptRun->textLength - 1) {
+            char16_t low = scriptRun->textArray[scriptRun->scriptLimit + 1];
+
+            /*
+             * if it is followed by a low surrogate,
+             * consume it and form the full character
+             */
+            if (low >= 0xDC00 && low <= 0xDFFF) {
+                ch = (high - 0xD800) * 0x0400 + low - 0xDC00 + 0x10000;
+                scriptRun->scriptLimit += 1;
+            }
+        }
+
+        sc = uscript_getScript(ch, &error);
+        pairIndex = getPairIndex(ch);
+
+        /*
+         * Paired character handling:
+         *
+         * if it's an open character, push it onto the stack.
+         * if it's a close character, find the matching open on the
+         * stack, and use that script code. Any non-matching open
+         * characters above it on the stack will be poped.
+         */
+        if (pairIndex >= 0) {
+            if ((pairIndex & 1) == 0) {
+                push(scriptRun, pairIndex, scriptRun->scriptCode);
+            } else {
+                int32_t pi = pairIndex & ~1;
+
+                while (STACK_IS_NOT_EMPTY(scriptRun) && TOP(scriptRun).pairIndex != pi) {
+                    pop(scriptRun);
+                }
+
+                if (STACK_IS_NOT_EMPTY(scriptRun)) {
+                    sc = TOP(scriptRun).scriptCode;
+                }
+            }
+        }
+
+        if (sameScript(scriptRun->scriptCode, sc)) {
+            if (scriptRun->scriptCode <= USCRIPT_INHERITED && sc > USCRIPT_INHERITED) {
+                scriptRun->scriptCode = sc;
+
+                fixup(scriptRun, scriptRun->scriptCode);
+            }
+
+            /*
+             * if this character is a close paired character,
+             * pop the matching open character from the stack
+             */
+            if (pairIndex >= 0 && (pairIndex & 1) != 0) {
+                pop(scriptRun);
+            }
+        } else {
+            /*
+             * if the run broke on a surrogate pair,
+             * end it before the high surrogate
+             */
+            if (ch >= 0x10000) {
+                scriptRun->scriptLimit -= 1;
+            }
+
+            break;
+        }
+    }
+
+
+    if (pRunStart != nullptr) {
+        *pRunStart = scriptRun->scriptStart;
+    }
+
+    if (pRunLimit != nullptr) {
+        *pRunLimit = scriptRun->scriptLimit;
+    }
+
+    if (pRunScript != nullptr) {
+        *pRunScript = scriptRun->scriptCode;
+    }
+
+    return true;
+}
diff --git a/intl/icu/source/common/usc_impl.h b/intl/icu/source/common/usc_impl.h
new file mode 100644
index 0000000000..cd76990501
--- /dev/null
+++ b/intl/icu/source/common/usc_impl.h
@@ -0,0 +1,139 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+*   Copyright (C) 1999-2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*
+* File USC_IMPL.H
+*
+* Modification History:
+*
+*   Date        Name        Description
+*   07/08/2002  Eric Mader  Creation.
+******************************************************************************
+*/
+
+#ifndef USC_IMPL_H
+#define USC_IMPL_H
+#include "unicode/utypes.h"
+#include "unicode/uscript.h"
+
+/**
+ * UScriptRun is used to find runs of characters in
+ * the same script. It implements a simple iterator over an array
+ * of characters. The iterator will resolve script-neutral characters
+ * like punctuation into the script of the surrounding characters.
+ *
+ * The iterator will try to match paired punctuation. If it sees an
+ * opening punctuation character, it will remember the script that
+ * was assigned to that character, and assign the same script to the
+ * matching closing punctuation.
+ *
+ * Scripts are chosen based on the UScriptCode enumeration.
+ * No attempt is made to combine related scripts into a single run. In
+ * particular, Hiragana, Katakana, and Han characters will appear in separate
+ * runs.
+
+ * Here is an example of how to iterate over script runs:
+ * 
+ * \code
+ * void printScriptRuns(const UChar *text, int32_t length)
+ * {
+ *     UErrorCode error = U_ZERO_ERROR;
+ *     UScriptRun *scriptRun = uscript_openRun(text, testLength, &error);
+ *     int32_t start = 0, limit = 0;
+ *     UScriptCode code = USCRIPT_INVALID_CODE;
+ *
+ *     while (uscript_nextRun(&start, &limit, &code)) {
+ *         printf("Script '%s' from %d to %d.\n", uscript_getName(code), start, limit);
+ *     }
+ *
+ *     uscript_closeRun(scriptRun);
+ *  }
+ * 

+ */
+struct UScriptRun;
+
+typedef struct UScriptRun UScriptRun;
+
+/**
+ * Create a UScriptRun object for iterating over the given text. This object must
+ * be freed using uscript_closeRun(). Note that this object does not copy the source text,
+ * only the pointer to it. You must make sure that the pointer remains valid until you call
+ * uscript_closeRun() or uscript_setRunText().
+ *
+ * @param src is the address of the array of characters over which to iterate.
+ *        if src == NULL and length == 0,
+ *        an empty UScriptRun object will be returned.
+ *
+ * @param length is the number of characters over which to iterate.
+ *
+ * @param pErrorCode is a pointer to a valid UErrorCode value. If this value
+ *        indicates a failure on entry, the function will immediately return.
+ *        On exit the value will indicate the success of the operation.
+ *
+ * @return the address of UScriptRun object which will iterate over the text,
+ *         or NULL if the operation failed.
+ */
+U_CAPI UScriptRun * U_EXPORT2
+uscript_openRun(const UChar *src, int32_t length, UErrorCode *pErrorCode);
+
+/**
+ * Frees the given UScriptRun object and any storage associated with it.
+ * On return, scriptRun no longer points to a valid UScriptRun object.
+ *
+ * @param scriptRun is the UScriptRun object which will be freed.
+ */
+U_CAPI void U_EXPORT2
+uscript_closeRun(UScriptRun *scriptRun);
+
+/**
+ * Reset the UScriptRun object so that it will start iterating from
+ * the beginning.
+ *
+ * @param scriptRun is the address of the UScriptRun object to be reset.
+ */
+U_CAPI void U_EXPORT2
+uscript_resetRun(UScriptRun *scriptRun);
+
+/**
+ * Change the text over which the given UScriptRun object iterates.
+ *
+ * @param scriptRun is the UScriptRun object which will be changed.
+ *
+ * @param src is the address of the new array of characters over which to iterate.
+ *        If src == NULL and length == 0,
+ *        the UScriptRun object will become empty.
+ *
+ * @param length is the new number of characters over which to iterate
+ *
+ * @param pErrorCode is a pointer to a valid UErrorCode value. If this value
+ *        indicates a failure on entry, the function will immediately return.
+ *        On exit the value will indicate the success of the operation.
+ */
+U_CAPI void U_EXPORT2
+uscript_setRunText(UScriptRun *scriptRun, const UChar *src, int32_t length, UErrorCode *pErrorCode);
+
+/**
+ * Advance the UScriptRun object to the next script run, return the start and limit
+ * offsets, and the script of the run.
+ *
+ * @param scriptRun is the address of the UScriptRun object.
+ *
+ * @param pRunStart is a pointer to the variable to receive the starting offset of the next run.
+ *        This pointer can be NULL if the value is not needed.
+ *
+ * @param pRunLimit is a pointer to the variable to receive the limit offset of the next run.
+ *        This pointer can be NULL if the value is not needed.
+ *
+ * @param pRunScript is a pointer to the variable to receive the UScriptCode for the
+ *        script of the current run. This pointer can be NULL if the value is not needed.
+ *
+ * @return true if there was another script run.
+ */
+U_CAPI UBool U_EXPORT2
+uscript_nextRun(UScriptRun *scriptRun, int32_t *pRunStart, int32_t *pRunLimit, UScriptCode *pRunScript);
+
+#endif
diff --git a/intl/icu/source/common/uscript.cpp b/intl/icu/source/common/uscript.cpp
new file mode 100644
index 0000000000..3cc2b6675c
--- /dev/null
+++ b/intl/icu/source/common/uscript.cpp
@@ -0,0 +1,162 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+*   Copyright (C) 1997-2014, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*
+* File USCRIPT.C
+*
+* Modification History:
+*
+*   Date        Name        Description
+*   07/06/2001    Ram         Creation.
+******************************************************************************
+*/
+
+#include "unicode/uchar.h"
+#include "unicode/uscript.h"
+#include "unicode/uloc.h"
+#include "bytesinkutil.h"
+#include "charstr.h"
+#include "cmemory.h"
+#include "cstring.h"
+#include "ulocimp.h"
+
+static const UScriptCode JAPANESE[3] = { USCRIPT_KATAKANA, USCRIPT_HIRAGANA, USCRIPT_HAN };
+static const UScriptCode KOREAN[2] = { USCRIPT_HANGUL, USCRIPT_HAN };
+static const UScriptCode HAN_BOPO[2] = { USCRIPT_HAN, USCRIPT_BOPOMOFO };
+
+static int32_t
+setCodes(const UScriptCode *src, int32_t length,
+         UScriptCode *dest, int32_t capacity, UErrorCode *err) {
+    int32_t i;
+    if(U_FAILURE(*err)) { return 0; }
+    if(length > capacity) {
+        *err = U_BUFFER_OVERFLOW_ERROR;
+        return length;
+    }
+    for(i = 0; i < length; ++i) {
+        dest[i] = src[i];
+    }
+    return length;
+}
+
+static int32_t
+setOneCode(UScriptCode script, UScriptCode *scripts, int32_t capacity, UErrorCode *err) {
+    if(U_FAILURE(*err)) { return 0; }
+    if(1 > capacity) {
+        *err = U_BUFFER_OVERFLOW_ERROR;
+        return 1;
+    }
+    scripts[0] = script;
+    return 1;
+}
+
+static int32_t
+getCodesFromLocale(const char *locale,
+                   UScriptCode *scripts, int32_t capacity, UErrorCode *err) {
+    UErrorCode internalErrorCode = U_ZERO_ERROR;
+    char lang[8] = {0};
+    char script[8] = {0};
+    int32_t scriptLength;
+    if(U_FAILURE(*err)) { return 0; }
+    // Multi-script languages, equivalent to the LocaleScript data
+    // that we used to load from locale resource bundles.
+    /*length = */ uloc_getLanguage(locale, lang, UPRV_LENGTHOF(lang), &internalErrorCode);
+    if(U_FAILURE(internalErrorCode) || internalErrorCode == U_STRING_NOT_TERMINATED_WARNING) {
+        return 0;
+    }
+    if(0 == uprv_strcmp(lang, "ja")) {
+        return setCodes(JAPANESE, UPRV_LENGTHOF(JAPANESE), scripts, capacity, err);
+    }
+    if(0 == uprv_strcmp(lang, "ko")) {
+        return setCodes(KOREAN, UPRV_LENGTHOF(KOREAN), scripts, capacity, err);
+    }
+    scriptLength = uloc_getScript(locale, script, UPRV_LENGTHOF(script), &internalErrorCode);
+    if(U_FAILURE(internalErrorCode) || internalErrorCode == U_STRING_NOT_TERMINATED_WARNING) {
+        return 0;
+    }
+    if(0 == uprv_strcmp(lang, "zh") && 0 == uprv_strcmp(script, "Hant")) {
+        return setCodes(HAN_BOPO, UPRV_LENGTHOF(HAN_BOPO), scripts, capacity, err);
+    }
+    // Explicit script code.
+    if(scriptLength != 0) {
+        UScriptCode scriptCode = (UScriptCode)u_getPropertyValueEnum(UCHAR_SCRIPT, script);
+        if(scriptCode != USCRIPT_INVALID_CODE) {
+            if(scriptCode == USCRIPT_SIMPLIFIED_HAN || scriptCode == USCRIPT_TRADITIONAL_HAN) {
+                scriptCode = USCRIPT_HAN;
+            }
+            return setOneCode(scriptCode, scripts, capacity, err);
+        }
+    }
+    return 0;
+}
+
+/* TODO: this is a bad API and should be deprecated, ticket #11141 */
+U_CAPI int32_t  U_EXPORT2
+uscript_getCode(const char* nameOrAbbrOrLocale,
+                UScriptCode* fillIn,
+                int32_t capacity,
+                UErrorCode* err){
+    UBool triedCode;
+    UErrorCode internalErrorCode;
+    int32_t length;
+
+    if(U_FAILURE(*err)) {
+        return 0;
+    }
+    if(nameOrAbbrOrLocale==nullptr ||
+            (fillIn == nullptr ? capacity != 0 : capacity < 0)) {
+        *err = U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    triedCode = false;
+    const char* lastSepPtr = uprv_strrchr(nameOrAbbrOrLocale, '-');
+    if (lastSepPtr==nullptr) {
+        lastSepPtr = uprv_strrchr(nameOrAbbrOrLocale, '_');
+    }
+    // Favor interpretation of nameOrAbbrOrLocale as a script alias if either
+    // 1. nameOrAbbrOrLocale does not contain -/_. Handles Han, Mro, Nko, etc.
+    // 2. The last instance of -/_ is at offset 3, and the portion after that is
+    //    longer than 4 characters (i.e. not a script or region code). This handles
+    //    Old_Hungarian, Old_Italic, etc. ("old" is a valid language code)
+    // 3. The last instance of -/_ is at offset 7, and the portion after that is
+    //    3 characters. This handles New_Tai_Lue ("new" is a valid language code).
+    if (lastSepPtr==nullptr
+            || (lastSepPtr-nameOrAbbrOrLocale == 3 && uprv_strlen(nameOrAbbrOrLocale) > 8)
+            || (lastSepPtr-nameOrAbbrOrLocale == 7 && uprv_strlen(nameOrAbbrOrLocale) == 11) ) {
+        /* try long and abbreviated script names first */
+        UScriptCode code = (UScriptCode) u_getPropertyValueEnum(UCHAR_SCRIPT, nameOrAbbrOrLocale);
+        if(code!=USCRIPT_INVALID_CODE) {
+            return setOneCode(code, fillIn, capacity, err);
+        }
+        triedCode = true;
+    }
+    internalErrorCode = U_ZERO_ERROR;
+    length = getCodesFromLocale(nameOrAbbrOrLocale, fillIn, capacity, err);
+    if(U_FAILURE(*err) || length != 0) {
+        return length;
+    }
+    icu::CharString likely;
+    {
+        icu::CharStringByteSink sink(&likely);
+        ulocimp_addLikelySubtags(nameOrAbbrOrLocale, sink, &internalErrorCode);
+    }
+    if(U_SUCCESS(internalErrorCode) && internalErrorCode != U_STRING_NOT_TERMINATED_WARNING) {
+        length = getCodesFromLocale(likely.data(), fillIn, capacity, err);
+        if(U_FAILURE(*err) || length != 0) {
+            return length;
+        }
+    }
+    if(!triedCode) {
+        /* still not found .. try long and abbreviated script names again */
+        UScriptCode code = (UScriptCode) u_getPropertyValueEnum(UCHAR_SCRIPT, nameOrAbbrOrLocale);
+        if(code!=USCRIPT_INVALID_CODE) {
+            return setOneCode(code, fillIn, capacity, err);
+        }
+    }
+    return 0;
+}
diff --git a/intl/icu/source/common/uscript_props.cpp b/intl/icu/source/common/uscript_props.cpp
new file mode 100644
index 0000000000..b26164408a
--- /dev/null
+++ b/intl/icu/source/common/uscript_props.cpp
@@ -0,0 +1,309 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*   Copyright (C) 2013-2016, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*******************************************************************************
+*   file name:  uscript_props.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2013feb16
+*   created by: Markus W. Scherer
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/unistr.h"
+#include "unicode/uscript.h"
+#include "unicode/utf16.h"
+#include "ustr_imp.h"
+#include "cmemory.h"
+
+namespace {
+
+// Script metadata (script properties).
+// See http://unicode.org/cldr/trac/browser/trunk/common/properties/scriptMetadata.txt
+
+// 0 = NOT_ENCODED, no sample character, default false script properties.
+// Bits 20.. 0: sample character
+
+// Bits 23..21: usage
+const int32_t UNKNOWN = 1 << 21;
+const int32_t EXCLUSION = 2 << 21;
+const int32_t LIMITED_USE = 3 << 21;
+// st int32_t ASPIRATIONAL = 4 << 21; -- not used any more since Unicode 10
+const int32_t RECOMMENDED = 5 << 21;
+
+// Bits 31..24: Single-bit flags
+const int32_t RTL = 1 << 24;
+const int32_t LB_LETTERS = 1 << 25;
+const int32_t CASED = 1 << 26;
+
+const int32_t SCRIPT_PROPS[] = {
+    // Begin copy-paste output from
+    // tools/trunk/unicode/py/parsescriptmetadata.py
+    0x0040 | RECOMMENDED,  // Zyyy
+    0x0308 | RECOMMENDED,  // Zinh
+    0x0628 | RECOMMENDED | RTL,  // Arab
+    0x0531 | RECOMMENDED | CASED,  // Armn
+    0x0995 | RECOMMENDED,  // Beng
+    0x3105 | RECOMMENDED | LB_LETTERS,  // Bopo
+    0x13C4 | LIMITED_USE | CASED,  // Cher
+    0x03E2 | EXCLUSION | CASED,  // Copt
+    0x042F | RECOMMENDED | CASED,  // Cyrl
+    0x10414 | EXCLUSION | CASED,  // Dsrt
+    0x0905 | RECOMMENDED,  // Deva
+    0x12A0 | RECOMMENDED,  // Ethi
+    0x10D3 | RECOMMENDED,  // Geor
+    0x10330 | EXCLUSION,  // Goth
+    0x03A9 | RECOMMENDED | CASED,  // Grek
+    0x0A95 | RECOMMENDED,  // Gujr
+    0x0A15 | RECOMMENDED,  // Guru
+    0x5B57 | RECOMMENDED | LB_LETTERS,  // Hani
+    0xAC00 | RECOMMENDED,  // Hang
+    0x05D0 | RECOMMENDED | RTL,  // Hebr
+    0x304B | RECOMMENDED | LB_LETTERS,  // Hira
+    0x0C95 | RECOMMENDED,  // Knda
+    0x30AB | RECOMMENDED | LB_LETTERS,  // Kana
+    0x1780 | RECOMMENDED | LB_LETTERS,  // Khmr
+    0x0EA5 | RECOMMENDED | LB_LETTERS,  // Laoo
+    0x004C | RECOMMENDED | CASED,  // Latn
+    0x0D15 | RECOMMENDED,  // Mlym
+    0x1826 | EXCLUSION,  // Mong
+    0x1000 | RECOMMENDED | LB_LETTERS,  // Mymr
+    0x168F | EXCLUSION,  // Ogam
+    0x10300 | EXCLUSION,  // Ital
+    0x0B15 | RECOMMENDED,  // Orya
+    0x16A0 | EXCLUSION,  // Runr
+    0x0D85 | RECOMMENDED,  // Sinh
+    0x0710 | LIMITED_USE | RTL,  // Syrc
+    0x0B95 | RECOMMENDED,  // Taml
+    0x0C15 | RECOMMENDED,  // Telu
+    0x078C | RECOMMENDED | RTL,  // Thaa
+    0x0E17 | RECOMMENDED | LB_LETTERS,  // Thai
+    0x0F40 | RECOMMENDED,  // Tibt
+    0x14C0 | LIMITED_USE,  // Cans
+    0xA288 | LIMITED_USE | LB_LETTERS,  // Yiii
+    0x1703 | EXCLUSION,  // Tglg
+    0x1723 | EXCLUSION,  // Hano
+    0x1743 | EXCLUSION,  // Buhd
+    0x1763 | EXCLUSION,  // Tagb
+    0x280E | UNKNOWN,  // Brai
+    0x10800 | EXCLUSION | RTL,  // Cprt
+    0x1900 | LIMITED_USE,  // Limb
+    0x10000 | EXCLUSION,  // Linb
+    0x10480 | EXCLUSION,  // Osma
+    0x10450 | EXCLUSION,  // Shaw
+    0x1950 | LIMITED_USE | LB_LETTERS,  // Tale
+    0x10380 | EXCLUSION,  // Ugar
+    0,
+    0x1A00 | EXCLUSION,  // Bugi
+    0x2C00 | EXCLUSION | CASED,  // Glag
+    0x10A00 | EXCLUSION | RTL,  // Khar
+    0xA800 | LIMITED_USE,  // Sylo
+    0x1980 | LIMITED_USE | LB_LETTERS,  // Talu
+    0x2D30 | LIMITED_USE,  // Tfng
+    0x103A0 | EXCLUSION,  // Xpeo
+    0x1B05 | LIMITED_USE,  // Bali
+    0x1BC0 | LIMITED_USE,  // Batk
+    0,
+    0x11005 | EXCLUSION,  // Brah
+    0xAA00 | LIMITED_USE,  // Cham
+    0,
+    0,
+    0,
+    0,
+    0x13153 | EXCLUSION,  // Egyp
+    0,
+    0x5B57 | RECOMMENDED | LB_LETTERS,  // Hans
+    0x5B57 | RECOMMENDED | LB_LETTERS,  // Hant
+    0x16B1C | EXCLUSION,  // Hmng
+    0x10CA1 | EXCLUSION | RTL | CASED,  // Hung
+    0,
+    0xA984 | LIMITED_USE,  // Java
+    0xA90A | LIMITED_USE,  // Kali
+    0,
+    0,
+    0x1C00 | LIMITED_USE,  // Lepc
+    0x10647 | EXCLUSION,  // Lina
+    0x0840 | LIMITED_USE | RTL,  // Mand
+    0,
+    0x10980 | EXCLUSION | RTL,  // Mero
+    0x07CA | LIMITED_USE | RTL,  // Nkoo
+    0x10C00 | EXCLUSION | RTL,  // Orkh
+    0x1036B | EXCLUSION,  // Perm
+    0xA840 | EXCLUSION,  // Phag
+    0x10900 | EXCLUSION | RTL,  // Phnx
+    0x16F00 | LIMITED_USE,  // Plrd
+    0,
+    0,
+    0,
+    0,
+    0,
+    0,
+    0xA549 | LIMITED_USE,  // Vaii
+    0,
+    0x12000 | EXCLUSION,  // Xsux
+    0,
+    0xFDD0 | UNKNOWN,  // Zzzz
+    0x102A0 | EXCLUSION,  // Cari
+    0x304B | RECOMMENDED | LB_LETTERS,  // Jpan
+    0x1A20 | LIMITED_USE | LB_LETTERS,  // Lana
+    0x10280 | EXCLUSION,  // Lyci
+    0x10920 | EXCLUSION | RTL,  // Lydi
+    0x1C5A | LIMITED_USE,  // Olck
+    0xA930 | EXCLUSION,  // Rjng
+    0xA882 | LIMITED_USE,  // Saur
+    0x1D850 | EXCLUSION,  // Sgnw
+    0x1B83 | LIMITED_USE,  // Sund
+    0,
+    0xABC0 | LIMITED_USE,  // Mtei
+    0x10840 | EXCLUSION | RTL,  // Armi
+    0x10B00 | EXCLUSION | RTL,  // Avst
+    0x11103 | LIMITED_USE,  // Cakm
+    0xAC00 | RECOMMENDED,  // Kore
+    0x11083 | EXCLUSION,  // Kthi
+    0x10AD8 | EXCLUSION | RTL,  // Mani
+    0x10B60 | EXCLUSION | RTL,  // Phli
+    0x10B8F | EXCLUSION | RTL,  // Phlp
+    0,
+    0x10B40 | EXCLUSION | RTL,  // Prti
+    0x0800 | EXCLUSION | RTL,  // Samr
+    0xAA80 | LIMITED_USE | LB_LETTERS,  // Tavt
+    0,
+    0,
+    0xA6A0 | LIMITED_USE,  // Bamu
+    0xA4D0 | LIMITED_USE,  // Lisu
+    0,
+    0x10A60 | EXCLUSION | RTL,  // Sarb
+    0x16AE6 | EXCLUSION,  // Bass
+    0x1BC20 | EXCLUSION,  // Dupl
+    0x10500 | EXCLUSION,  // Elba
+    0x11315 | EXCLUSION,  // Gran
+    0,
+    0,
+    0x1E802 | EXCLUSION | RTL,  // Mend
+    0x109A0 | EXCLUSION | RTL,  // Merc
+    0x10A95 | EXCLUSION | RTL,  // Narb
+    0x10896 | EXCLUSION | RTL,  // Nbat
+    0x10873 | EXCLUSION | RTL,  // Palm
+    0x112BE | EXCLUSION,  // Sind
+    0x118B4 | EXCLUSION | CASED,  // Wara
+    0,
+    0,
+    0x16A4F | EXCLUSION,  // Mroo
+    0x1B1C4 | EXCLUSION | LB_LETTERS,  // Nshu
+    0x11183 | EXCLUSION,  // Shrd
+    0x110D0 | EXCLUSION,  // Sora
+    0x11680 | EXCLUSION,  // Takr
+    0x18229 | EXCLUSION | LB_LETTERS,  // Tang
+    0,
+    0x14400 | EXCLUSION,  // Hluw
+    0x11208 | EXCLUSION,  // Khoj
+    0x11484 | EXCLUSION,  // Tirh
+    0x10537 | EXCLUSION,  // Aghb
+    0x11152 | EXCLUSION,  // Mahj
+    0x11717 | EXCLUSION | LB_LETTERS,  // Ahom
+    0x108F4 | EXCLUSION | RTL,  // Hatr
+    0x1160E | EXCLUSION,  // Modi
+    0x1128F | EXCLUSION,  // Mult
+    0x11AC0 | EXCLUSION,  // Pauc
+    0x1158E | EXCLUSION,  // Sidd
+    0x1E909 | LIMITED_USE | RTL | CASED,  // Adlm
+    0x11C0E | EXCLUSION,  // Bhks
+    0x11C72 | EXCLUSION,  // Marc
+    0x11412 | LIMITED_USE,  // Newa
+    0x104B5 | LIMITED_USE | CASED,  // Osge
+    0x5B57 | RECOMMENDED | LB_LETTERS,  // Hanb
+    0x1112 | RECOMMENDED,  // Jamo
+    0,
+    0x11D10 | EXCLUSION,  // Gonm
+    0x11A5C | EXCLUSION,  // Soyo
+    0x11A0B | EXCLUSION,  // Zanb
+    0x1180B | EXCLUSION,  // Dogr
+    0x11D71 | LIMITED_USE,  // Gong
+    0x11EE5 | EXCLUSION,  // Maka
+    0x16E40 | EXCLUSION | CASED,  // Medf
+    0x10D12 | LIMITED_USE | RTL,  // Rohg
+    0x10F42 | EXCLUSION | RTL,  // Sogd
+    0x10F19 | EXCLUSION | RTL,  // Sogo
+    0x10FF1 | EXCLUSION | RTL,  // Elym
+    0x1E108 | LIMITED_USE,  // Hmnp
+    0x119CE | EXCLUSION,  // Nand
+    0x1E2E1 | LIMITED_USE,  // Wcho
+    0x10FBF | EXCLUSION | RTL,  // Chrs
+    0x1190C | EXCLUSION,  // Diak
+    0x18C65 | EXCLUSION | LB_LETTERS,  // Kits
+    0x10E88 | EXCLUSION | RTL,  // Yezi
+    0x12FE5 | EXCLUSION,  // Cpmn
+    0x10F7C | EXCLUSION | RTL,  // Ougr
+    0x16ABC | EXCLUSION,  // Tnsa
+    0x1E290 | EXCLUSION,  // Toto
+    0x10582 | EXCLUSION | CASED,  // Vith
+    0x11F1B | EXCLUSION | LB_LETTERS,  // Kawi
+    0x1E4E6 | EXCLUSION,  // Nagm
+    // End copy-paste from parsescriptmetadata.py
+};
+
+int32_t getScriptProps(UScriptCode script) {
+    if (0 <= script && script < UPRV_LENGTHOF(SCRIPT_PROPS)) {
+        return SCRIPT_PROPS[script];
+    } else {
+        return 0;
+    }
+}
+
+}  // namespace
+
+U_CAPI int32_t U_EXPORT2
+uscript_getSampleString(UScriptCode script, char16_t *dest, int32_t capacity, UErrorCode *pErrorCode) {
+    if(U_FAILURE(*pErrorCode)) { return 0; }
+    if(capacity < 0 || (capacity > 0 && dest == nullptr)) {
+        *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    int32_t sampleChar = getScriptProps(script) & 0x1fffff;
+    int32_t length;
+    if(sampleChar == 0) {
+        length = 0;
+    } else {
+        length = U16_LENGTH(sampleChar);
+        if(length <= capacity) {
+            int32_t i = 0;
+            U16_APPEND_UNSAFE(dest, i, sampleChar);
+        }
+    }
+    return u_terminateUChars(dest, capacity, length, pErrorCode);
+}
+
+U_COMMON_API icu::UnicodeString U_EXPORT2
+uscript_getSampleUnicodeString(UScriptCode script) {
+    icu::UnicodeString sample;
+    int32_t sampleChar = getScriptProps(script) & 0x1fffff;
+    if(sampleChar != 0) {
+        sample.append(sampleChar);
+    }
+    return sample;
+}
+
+U_CAPI UScriptUsage U_EXPORT2
+uscript_getUsage(UScriptCode script) {
+    return (UScriptUsage)((getScriptProps(script) >> 21) & 7);
+}
+
+U_CAPI UBool U_EXPORT2
+uscript_isRightToLeft(UScriptCode script) {
+    return (getScriptProps(script) & RTL) != 0;
+}
+
+U_CAPI UBool U_EXPORT2
+uscript_breaksBetweenLetters(UScriptCode script) {
+    return (getScriptProps(script) & LB_LETTERS) != 0;
+}
+
+U_CAPI UBool U_EXPORT2
+uscript_isCased(UScriptCode script) {
+    return (getScriptProps(script) & CASED) != 0;
+}
diff --git a/intl/icu/source/common/uset.cpp b/intl/icu/source/common/uset.cpp
new file mode 100644
index 0000000000..b2d0b91d4b
--- /dev/null
+++ b/intl/icu/source/common/uset.cpp
@@ -0,0 +1,681 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2002-2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  uset.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2002mar07
+*   created by: Markus W. Scherer
+*
+*   There are functions to efficiently serialize a USet into an array of uint16_t
+*   and functions to use such a serialized form efficiently without
+*   instantiating a new USet.
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/uobject.h"
+#include "unicode/uset.h"
+#include "unicode/uniset.h"
+#include "cmemory.h"
+#include "unicode/ustring.h"
+#include "unicode/parsepos.h"
+
+U_NAMESPACE_USE
+
+U_CAPI USet* U_EXPORT2
+uset_openEmpty() {
+    return (USet*) new UnicodeSet();
+}
+
+U_CAPI USet* U_EXPORT2
+uset_open(UChar32 start, UChar32 end) {
+    return (USet*) new UnicodeSet(start, end);
+}
+
+U_CAPI void U_EXPORT2
+uset_close(USet* set) {
+    delete (UnicodeSet*) set;
+}
+
+U_CAPI USet * U_EXPORT2
+uset_clone(const USet *set) {
+    return (USet*) (((UnicodeSet*) set)->UnicodeSet::clone());
+}
+
+U_CAPI UBool U_EXPORT2
+uset_isFrozen(const USet *set) {
+    return ((UnicodeSet*) set)->UnicodeSet::isFrozen();
+}
+
+U_CAPI void U_EXPORT2
+uset_freeze(USet *set) {
+    ((UnicodeSet*) set)->UnicodeSet::freeze();
+}
+
+U_CAPI USet * U_EXPORT2
+uset_cloneAsThawed(const USet *set) {
+    return (USet*) (((UnicodeSet*) set)->UnicodeSet::cloneAsThawed());
+}
+
+U_CAPI void U_EXPORT2
+uset_set(USet* set,
+     UChar32 start, UChar32 end) {
+    ((UnicodeSet*) set)->UnicodeSet::set(start, end);
+}
+
+U_CAPI void U_EXPORT2
+uset_addAll(USet* set, const USet *additionalSet) {
+    ((UnicodeSet*) set)->UnicodeSet::addAll(*((const UnicodeSet*)additionalSet));
+}
+
+U_CAPI void U_EXPORT2
+uset_add(USet* set, UChar32 c) {
+    ((UnicodeSet*) set)->UnicodeSet::add(c);
+}
+
+U_CAPI void U_EXPORT2
+uset_addRange(USet* set, UChar32 start, UChar32 end) {
+    ((UnicodeSet*) set)->UnicodeSet::add(start, end);    
+}
+
+U_CAPI void U_EXPORT2
+uset_addString(USet* set, const char16_t* str, int32_t strLen) {
+    // UnicodeString handles -1 for strLen
+    UnicodeString s(strLen<0, str, strLen);
+    ((UnicodeSet*) set)->UnicodeSet::add(s);
+}
+
+U_CAPI void U_EXPORT2
+uset_addAllCodePoints(USet* set, const char16_t *str, int32_t strLen) {
+    // UnicodeString handles -1 for strLen
+    UnicodeString s(str, strLen);
+    ((UnicodeSet*) set)->UnicodeSet::addAll(s);
+}
+
+U_CAPI void U_EXPORT2
+uset_remove(USet* set, UChar32 c) {
+    ((UnicodeSet*) set)->UnicodeSet::remove(c);
+}
+
+U_CAPI void U_EXPORT2
+uset_removeRange(USet* set, UChar32 start, UChar32 end) {
+    ((UnicodeSet*) set)->UnicodeSet::remove(start, end);
+}
+
+U_CAPI void U_EXPORT2
+uset_removeString(USet* set, const char16_t* str, int32_t strLen) {
+    UnicodeString s(strLen==-1, str, strLen);
+    ((UnicodeSet*) set)->UnicodeSet::remove(s);
+}
+
+U_CAPI void U_EXPORT2
+uset_removeAllCodePoints(USet *set, const char16_t *str, int32_t length) {
+    UnicodeString s(length==-1, str, length);
+    ((UnicodeSet*) set)->UnicodeSet::removeAll(s);
+}
+
+U_CAPI void U_EXPORT2
+uset_removeAll(USet* set, const USet* remove) {
+    ((UnicodeSet*) set)->UnicodeSet::removeAll(*(const UnicodeSet*)remove);
+}
+
+U_CAPI void U_EXPORT2
+uset_retain(USet* set, UChar32 start, UChar32 end) {
+    ((UnicodeSet*) set)->UnicodeSet::retain(start, end);
+}
+
+U_CAPI void U_EXPORT2
+uset_retainString(USet *set, const char16_t *str, int32_t length) {
+    UnicodeString s(length==-1, str, length);
+    ((UnicodeSet*) set)->UnicodeSet::retain(s);
+}
+
+U_CAPI void U_EXPORT2
+uset_retainAllCodePoints(USet *set, const char16_t *str, int32_t length) {
+    UnicodeString s(length==-1, str, length);
+    ((UnicodeSet*) set)->UnicodeSet::retainAll(s);
+}
+
+U_CAPI void U_EXPORT2
+uset_retainAll(USet* set, const USet* retain) {
+    ((UnicodeSet*) set)->UnicodeSet::retainAll(*(const UnicodeSet*)retain);
+}
+
+U_CAPI void U_EXPORT2
+uset_compact(USet* set) {
+    ((UnicodeSet*) set)->UnicodeSet::compact();
+}
+
+U_CAPI void U_EXPORT2
+uset_complement(USet* set) {
+    ((UnicodeSet*) set)->UnicodeSet::complement();
+}
+
+U_CAPI void U_EXPORT2
+uset_complementRange(USet *set, UChar32 start, UChar32 end) {
+    ((UnicodeSet*) set)->UnicodeSet::complement(start, end);
+}
+
+U_CAPI void U_EXPORT2
+uset_complementString(USet *set, const char16_t *str, int32_t length) {
+    UnicodeString s(length==-1, str, length);
+    ((UnicodeSet*) set)->UnicodeSet::complement(s);
+}
+
+U_CAPI void U_EXPORT2
+uset_complementAllCodePoints(USet *set, const char16_t *str, int32_t length) {
+    UnicodeString s(length==-1, str, length);
+    ((UnicodeSet*) set)->UnicodeSet::complementAll(s);
+}
+
+U_CAPI void U_EXPORT2
+uset_complementAll(USet* set, const USet* complement) {
+    ((UnicodeSet*) set)->UnicodeSet::complementAll(*(const UnicodeSet*)complement);
+}
+
+U_CAPI void U_EXPORT2
+uset_clear(USet* set) {
+    ((UnicodeSet*) set)->UnicodeSet::clear();
+}
+
+U_CAPI void U_EXPORT2
+uset_removeAllStrings(USet* set) {
+    ((UnicodeSet*) set)->UnicodeSet::removeAllStrings();
+}
+
+U_CAPI UBool U_EXPORT2
+uset_isEmpty(const USet* set) {
+    return ((const UnicodeSet*) set)->UnicodeSet::isEmpty();
+}
+
+U_CAPI UBool U_EXPORT2
+uset_hasStrings(const USet* set) {
+    return ((const UnicodeSet*) set)->UnicodeSet::hasStrings();
+}
+
+U_CAPI UBool U_EXPORT2
+uset_contains(const USet* set, UChar32 c) {
+    return ((const UnicodeSet*) set)->UnicodeSet::contains(c);
+}
+
+U_CAPI UBool U_EXPORT2
+uset_containsRange(const USet* set, UChar32 start, UChar32 end) {
+    return ((const UnicodeSet*) set)->UnicodeSet::contains(start, end);
+}
+
+U_CAPI UBool U_EXPORT2
+uset_containsString(const USet* set, const char16_t* str, int32_t strLen) {
+    UnicodeString s(strLen==-1, str, strLen);
+    return ((const UnicodeSet*) set)->UnicodeSet::contains(s);
+}
+
+U_CAPI UBool U_EXPORT2
+uset_containsAll(const USet* set1, const USet* set2) {
+    return ((const UnicodeSet*) set1)->UnicodeSet::containsAll(* (const UnicodeSet*) set2);
+}
+
+U_CAPI UBool U_EXPORT2
+uset_containsAllCodePoints(const USet* set, const char16_t *str, int32_t strLen) {
+    // Create a string alias, since nothing is being added to the set.
+    UnicodeString s(strLen==-1, str, strLen);
+    return ((const UnicodeSet*) set)->UnicodeSet::containsAll(s);
+}
+
+U_CAPI UBool U_EXPORT2
+uset_containsNone(const USet* set1, const USet* set2) {
+    return ((const UnicodeSet*) set1)->UnicodeSet::containsNone(* (const UnicodeSet*) set2);
+}
+
+U_CAPI UBool U_EXPORT2
+uset_containsSome(const USet* set1, const USet* set2) {
+    return ((const UnicodeSet*) set1)->UnicodeSet::containsSome(* (const UnicodeSet*) set2);
+}
+
+U_CAPI int32_t U_EXPORT2
+uset_span(const USet *set, const char16_t *s, int32_t length, USetSpanCondition spanCondition) {
+    return ((UnicodeSet*) set)->UnicodeSet::span(s, length, spanCondition);
+}
+
+U_CAPI int32_t U_EXPORT2
+uset_spanBack(const USet *set, const char16_t *s, int32_t length, USetSpanCondition spanCondition) {
+    return ((UnicodeSet*) set)->UnicodeSet::spanBack(s, length, spanCondition);
+}
+
+U_CAPI int32_t U_EXPORT2
+uset_spanUTF8(const USet *set, const char *s, int32_t length, USetSpanCondition spanCondition) {
+    return ((UnicodeSet*) set)->UnicodeSet::spanUTF8(s, length, spanCondition);
+}
+
+U_CAPI int32_t U_EXPORT2
+uset_spanBackUTF8(const USet *set, const char *s, int32_t length, USetSpanCondition spanCondition) {
+    return ((UnicodeSet*) set)->UnicodeSet::spanBackUTF8(s, length, spanCondition);
+}
+
+U_CAPI UBool U_EXPORT2
+uset_equals(const USet* set1, const USet* set2) {
+    return *(const UnicodeSet*)set1 == *(const UnicodeSet*)set2;
+}
+
+U_CAPI int32_t U_EXPORT2
+uset_indexOf(const USet* set, UChar32 c) {
+    return ((UnicodeSet*) set)->UnicodeSet::indexOf(c);
+}
+
+U_CAPI UChar32 U_EXPORT2
+uset_charAt(const USet* set, int32_t index) {
+    return ((UnicodeSet*) set)->UnicodeSet::charAt(index);
+}
+
+U_CAPI int32_t U_EXPORT2
+uset_size(const USet* set) {
+    return ((const UnicodeSet*) set)->UnicodeSet::size();
+}
+
+U_NAMESPACE_BEGIN
+/**
+ * This class only exists to provide access to the UnicodeSet private
+ * USet support API.  Declaring a class a friend is more portable than
+ * trying to declare extern "C" functions as friends.
+ */
+class USetAccess /* not : public UObject because all methods are static */ {
+public:
+    /* Try to have the compiler inline these*/
+    inline static int32_t getStringCount(const UnicodeSet& set) {
+        return set.stringsSize();
+    }
+    inline static const UnicodeString* getString(const UnicodeSet& set,
+                                                 int32_t i) {
+        return set.getString(i);
+    }
+private:
+    /* do not instantiate*/
+    USetAccess();
+};
+U_NAMESPACE_END
+
+U_CAPI int32_t U_EXPORT2
+uset_getRangeCount(const USet *set) {
+    return ((const UnicodeSet *)set)->UnicodeSet::getRangeCount();
+}
+
+U_CAPI int32_t U_EXPORT2
+uset_getItemCount(const USet* uset) {
+    const UnicodeSet& set = *(const UnicodeSet*)uset;
+    return set.getRangeCount() + USetAccess::getStringCount(set);
+}
+
+U_CAPI int32_t U_EXPORT2
+uset_getItem(const USet* uset, int32_t itemIndex,
+             UChar32* start, UChar32* end,
+             char16_t* str, int32_t strCapacity,
+             UErrorCode* ec) {
+    if (U_FAILURE(*ec)) return 0;
+    const UnicodeSet& set = *(const UnicodeSet*)uset;
+    int32_t rangeCount;
+
+    if (itemIndex < 0) {
+        *ec = U_ILLEGAL_ARGUMENT_ERROR;
+        return -1;
+    } else if (itemIndex < (rangeCount = set.getRangeCount())) {
+        *start = set.getRangeStart(itemIndex);
+        *end = set.getRangeEnd(itemIndex);
+        return 0;
+    } else {
+        itemIndex -= rangeCount;
+        if (itemIndex < USetAccess::getStringCount(set)) {
+            const UnicodeString* s = USetAccess::getString(set, itemIndex);
+            return s->extract(str, strCapacity, *ec);
+        } else {
+            *ec = U_INDEX_OUTOFBOUNDS_ERROR;
+            return -1;
+        }
+    }
+}
+
+//U_CAPI UBool U_EXPORT2
+//uset_getRange(const USet* set, int32_t rangeIndex,
+//              UChar32* pStart, UChar32* pEnd) {
+//    if ((uint32_t) rangeIndex >= (uint32_t) uset_getRangeCount(set)) {
+//        return false;
+//    }
+//    const UnicodeSet* us = (const UnicodeSet*) set;
+//    *pStart = us->getRangeStart(rangeIndex);
+//    *pEnd = us->getRangeEnd(rangeIndex);
+//    return true;
+//}
+
+/*
+ * Serialize a USet into 16-bit units.
+ * Store BMP code points as themselves with one 16-bit unit each.
+ *
+ * Important: the code points in the array are in ascending order,
+ * therefore all BMP code points precede all supplementary code points.
+ *
+ * Store each supplementary code point in 2 16-bit units,
+ * simply with higher-then-lower 16-bit halves.
+ *
+ * Precede the entire list with the length.
+ * If there are supplementary code points, then set bit 15 in the length
+ * and add the bmpLength between it and the array.
+ *
+ * In other words:
+ * - all BMP:            (length=bmpLength) BMP, .., BMP
+ * - some supplementary: (length|0x8000) (bmpLengthUnicodeSet::serialize(dest, destCapacity,* ec);
+}
+
+U_CAPI UBool U_EXPORT2
+uset_getSerializedSet(USerializedSet* fillSet, const uint16_t* src, int32_t srcLength) {
+    int32_t length;
+
+    if(fillSet==nullptr) {
+        return false;
+    }
+    if(src==nullptr || srcLength<=0) {
+        fillSet->length=fillSet->bmpLength=0;
+        return false;
+    }
+
+    length=*src++;
+    if(length&0x8000) {
+        /* there are supplementary values */
+        length&=0x7fff;
+        if(srcLength<(2+length)) {
+            fillSet->length=fillSet->bmpLength=0;
+            return false;
+        }
+        fillSet->bmpLength=*src++;
+    } else {
+        /* only BMP values */
+        if(srcLength<(1+length)) {
+            fillSet->length=fillSet->bmpLength=0;
+            return false;
+        }
+        fillSet->bmpLength=length;
+    }
+    fillSet->array=src;
+    fillSet->length=length;
+    return true;
+}
+
+U_CAPI void U_EXPORT2
+uset_setSerializedToOne(USerializedSet* fillSet, UChar32 c) {
+    if(fillSet==nullptr || (uint32_t)c>0x10ffff) {
+        return;
+    }
+
+    fillSet->array=fillSet->staticArray;
+    if(c<0xffff) {
+        fillSet->bmpLength=fillSet->length=2;
+        fillSet->staticArray[0]=(uint16_t)c;
+        fillSet->staticArray[1]=(uint16_t)c+1;
+    } else if(c==0xffff) {
+        fillSet->bmpLength=1;
+        fillSet->length=3;
+        fillSet->staticArray[0]=0xffff;
+        fillSet->staticArray[1]=1;
+        fillSet->staticArray[2]=0;
+    } else if(c<0x10ffff) {
+        fillSet->bmpLength=0;
+        fillSet->length=4;
+        fillSet->staticArray[0]=(uint16_t)(c>>16);
+        fillSet->staticArray[1]=(uint16_t)c;
+        ++c;
+        fillSet->staticArray[2]=(uint16_t)(c>>16);
+        fillSet->staticArray[3]=(uint16_t)c;
+    } else /* c==0x10ffff */ {
+        fillSet->bmpLength=0;
+        fillSet->length=2;
+        fillSet->staticArray[0]=0x10;
+        fillSet->staticArray[1]=0xffff;
+    }
+}
+
+U_CAPI UBool U_EXPORT2
+uset_serializedContains(const USerializedSet* set, UChar32 c) {
+    const uint16_t* array;
+
+    if(set==nullptr || (uint32_t)c>0x10ffff) {
+        return false;
+    }
+
+    array=set->array;
+    if(c<=0xffff) {
+        /* find c in the BMP part */
+        int32_t lo = 0;
+        int32_t hi = set->bmpLength-1;
+        if (c < array[0]) {
+            hi = 0;
+        } else if (c < array[hi]) {
+            for(;;) {
+                int32_t i = (lo + hi) >> 1;
+                if (i == lo) {
+                    break;  // Done!
+                } else if (c < array[i]) {
+                    hi = i;
+                } else {
+                    lo = i;
+                }
+            }
+        } else {
+            hi += 1;
+        }
+        return (UBool)(hi&1);
+    } else {
+        /* find c in the supplementary part */
+        uint16_t high=(uint16_t)(c>>16), low=(uint16_t)c;
+        int32_t base = set->bmpLength;
+        int32_t lo = 0;
+        int32_t hi = set->length - 2 - base;
+        if (high < array[base] || (high==array[base] && low> 1) & ~1;  // Guarantee even result
+                int32_t iabs = i + base;
+                if (i == lo) {
+                    break;  // Done!
+                } else if (high < array[iabs] || (high==array[iabs] && lowbmpLength+(set->length-set->bmpLength)/2+1)/2;
+}
+
+U_CAPI UBool U_EXPORT2
+uset_getSerializedRange(const USerializedSet* set, int32_t rangeIndex,
+                        UChar32* pStart, UChar32* pEnd) {
+    const uint16_t* array;
+    int32_t bmpLength, length;
+
+    if(set==nullptr || rangeIndex<0 || pStart==nullptr || pEnd==nullptr) {
+        return false;
+    }
+
+    array=set->array;
+    length=set->length;
+    bmpLength=set->bmpLength;
+
+    rangeIndex*=2; /* address start/limit pairs */
+    if(rangeIndex0) {
+//         if(c>=array[length-1]) {
+//             return length;
+//         }
+// 
+//         /* do not check the last range limit again in the loop below */
+//         --length;
+//     }
+// 
+//     for(i=0; i=array[i]; ++i) {}
+//     return i;
+// }
+// 
+// static UBool
+// addRemove(USet* set, UChar32 c, int32_t doRemove) {
+//     int32_t i, length, more;
+// 
+//     if(set==nullptr || (uint32_t)c>0x10ffff) {
+//         return false;
+//     }
+// 
+//     length=set->length;
+//     i=findChar(set->array, length, c);
+//     if((i&1)^doRemove) {
+//         /* c is already in the set */
+//         return true;
+//     }
+// 
+//     /* how many more array items do we need? */
+//     if(iarray[i]) {
+//         /* c is just before the following range, extend that in-place by one */
+//         set->array[i]=c;
+//         if(i>0) {
+//             --i;
+//             if(c==set->array[i]) {
+//                 /* the previous range collapsed, remove it */
+//                 set->length=length-=2;
+//                 if(iarray+i, set->array+i+2, (length-i)*4);
+//                 }
+//             }
+//         }
+//         return true;
+//     } else if(i>0 && c==set->array[i-1]) {
+//         /* c is just after the previous range, extend that in-place by one */
+//         if(++c<=0x10ffff) {
+//             set->array[i-1]=c;
+//             if(iarray[i]) {
+//                 /* the following range collapsed, remove it */
+//                 --i;
+//                 set->length=length-=2;
+//                 if(iarray+i, set->array+i+2, (length-i)*4);
+//                 }
+//             }
+//         } else {
+//             /* extend the previous range (had limit 0x10ffff) to the end of Unicode */
+//             set->length=i-1;
+//         }
+//         return true;
+//     } else if(i==length && c==0x10ffff) {
+//         /* insert one range limit c */
+//         more=1;
+//     } else {
+//         /* insert two range limits c, c+1 */
+//         more=2;
+//     }
+// 
+//     /* insert  range limits */
+//     if(length+more>set->capacity) {
+//         /* reallocate */
+//         int32_t newCapacity=set->capacity+set->capacity/2+USET_GROW_DELTA;
+//         UChar32* newArray=(UChar32* )uprv_malloc(newCapacity*4);
+//         if(newArray==nullptr) {
+//             return false;
+//         }
+//         set->capacity=newCapacity;
+//         uprv_memcpy(newArray, set->array, length*4);
+// 
+//         if(set->array!=set->staticBuffer) {
+//             uprv_free(set->array);
+//         }
+//         set->array=newArray;
+//     }
+// 
+//     if(iarray+i+more, set->array+i, (length-i)*4);
+//     }
+//     set->array[i]=c;
+//     if(more==2) {
+//         set->array[i+1]=c+1;
+//     }
+//     set->length+=more;
+// 
+//     return true;
+// }
+// 
+// U_CAPI UBool U_EXPORT2
+// uset_add(USet* set, UChar32 c) {
+//     return addRemove(set, c, 0);
+// }
+// 
+// U_CAPI void U_EXPORT2
+// uset_remove(USet* set, UChar32 c) {
+//     addRemove(set, c, 1);
+// }
diff --git a/intl/icu/source/common/uset_imp.h b/intl/icu/source/common/uset_imp.h
new file mode 100644
index 0000000000..77197aaed7
--- /dev/null
+++ b/intl/icu/source/common/uset_imp.h
@@ -0,0 +1,71 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2004-2007, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  uset_imp.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2004sep07
+*   created by: Markus W. Scherer
+*
+*   Internal USet definitions.
+*/
+
+#ifndef __USET_IMP_H__
+#define __USET_IMP_H__
+
+#include "unicode/utypes.h"
+#include "unicode/uset.h"
+
+U_CDECL_BEGIN
+
+typedef void U_CALLCONV
+USetAdd(USet *set, UChar32 c);
+
+typedef void U_CALLCONV
+USetAddRange(USet *set, UChar32 start, UChar32 end);
+
+typedef void U_CALLCONV
+USetAddString(USet *set, const UChar *str, int32_t length);
+
+typedef void U_CALLCONV
+USetRemove(USet *set, UChar32 c);
+
+typedef void U_CALLCONV
+USetRemoveRange(USet *set, UChar32 start, UChar32 end);
+
+/**
+ * Interface for adding items to a USet, to keep low-level code from
+ * statically depending on the USet implementation.
+ * Calls will look like sa->add(sa->set, c);
+ */
+struct USetAdder {
+    USet *set;
+    USetAdd *add;
+    USetAddRange *addRange;
+    USetAddString *addString;
+    USetRemove *remove;
+    USetRemoveRange *removeRange;
+};
+typedef struct USetAdder USetAdder;
+
+U_CDECL_END
+
+#ifdef __cplusplus
+
+namespace {
+
+constexpr int32_t USET_CASE_MASK = USET_CASE_INSENSITIVE | USET_ADD_CASE_MAPPINGS;
+
+}  // namespace
+
+#endif  // __cplusplus
+
+#endif
diff --git a/intl/icu/source/common/uset_props.cpp b/intl/icu/source/common/uset_props.cpp
new file mode 100644
index 0000000000..6f6e0c550f
--- /dev/null
+++ b/intl/icu/source/common/uset_props.cpp
@@ -0,0 +1,143 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2002-2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  uset_props.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2004aug30
+*   created by: Markus W. Scherer
+*
+*   C wrappers around UnicodeSet functions that are implemented in
+*   uniset_props.cpp, split off for modularization.
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/uobject.h"
+#include "unicode/uset.h"
+#include "unicode/uniset.h"
+#include "cmemory.h"
+#include "unicode/ustring.h"
+#include "unicode/parsepos.h"
+
+U_NAMESPACE_USE
+
+U_CAPI USet* U_EXPORT2
+uset_openPattern(const char16_t* pattern, int32_t patternLength,
+                 UErrorCode* ec)
+{
+    UnicodeString pat(patternLength==-1, pattern, patternLength);
+    UnicodeSet* set = new UnicodeSet(pat, *ec);
+    /* test for nullptr */
+    if(set == 0) {
+        *ec = U_MEMORY_ALLOCATION_ERROR;
+        return 0;
+    }
+
+    if (U_FAILURE(*ec)) {
+        delete set;
+        set = nullptr;
+    }
+    return (USet*) set;
+}
+
+U_CAPI USet* U_EXPORT2
+uset_openPatternOptions(const char16_t* pattern, int32_t patternLength,
+                 uint32_t options,
+                 UErrorCode* ec)
+{
+    UnicodeString pat(patternLength==-1, pattern, patternLength);
+    UnicodeSet* set = new UnicodeSet(pat, options, nullptr, *ec);
+    /* test for nullptr */
+    if(set == 0) {
+        *ec = U_MEMORY_ALLOCATION_ERROR;
+        return 0;
+    }
+
+    if (U_FAILURE(*ec)) {
+        delete set;
+        set = nullptr;
+    }
+    return (USet*) set;
+}
+
+
+U_CAPI int32_t U_EXPORT2 
+uset_applyPattern(USet *set,
+                  const char16_t *pattern, int32_t patternLength,
+                  uint32_t options,
+                  UErrorCode *status){
+
+    // status code needs to be checked since we 
+    // dereference it
+    if(status == nullptr || U_FAILURE(*status)){
+        return 0;
+    }
+
+    // check only the set paramenter
+    // if pattern is nullptr or NUL terminated
+    // UnicodeString constructor takes care of it
+    if(set == nullptr){
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    UnicodeString pat(pattern, patternLength);
+
+    ParsePosition pos;
+   
+    ((UnicodeSet*) set)->applyPattern(pat, pos, options, nullptr, *status);
+    
+    return pos.getIndex();
+}
+
+U_CAPI void U_EXPORT2
+uset_applyIntPropertyValue(USet* set,
+               UProperty prop, int32_t value, UErrorCode* ec) {
+    ((UnicodeSet*) set)->applyIntPropertyValue(prop, value, *ec);
+}
+
+U_CAPI void U_EXPORT2
+uset_applyPropertyAlias(USet* set,
+                        const char16_t *prop, int32_t propLength,
+                        const char16_t *value, int32_t valueLength,
+            UErrorCode* ec) {
+
+    UnicodeString p(prop, propLength);
+    UnicodeString v(value, valueLength);
+
+    ((UnicodeSet*) set)->applyPropertyAlias(p, v, *ec);
+}
+
+U_CAPI UBool U_EXPORT2
+uset_resemblesPattern(const char16_t *pattern, int32_t patternLength,
+                      int32_t pos) {
+
+    UnicodeString pat(pattern, patternLength);
+
+    return ((pos+1) < pat.length() &&
+            pat.charAt(pos) == (char16_t)91/*[*/) ||
+            UnicodeSet::resemblesPattern(pat, pos);
+}
+
+U_CAPI int32_t U_EXPORT2
+uset_toPattern(const USet* set,
+               char16_t* result, int32_t resultCapacity,
+               UBool escapeUnprintable,
+               UErrorCode* ec) {
+    UnicodeString pat;
+    ((const UnicodeSet*) set)->toPattern(pat, escapeUnprintable);
+    return pat.extract(result, resultCapacity, *ec);
+}
+
+U_CAPI void U_EXPORT2
+uset_closeOver(USet* set, int32_t attributes) {
+    ((UnicodeSet*) set)->UnicodeSet::closeOver(attributes);
+}
diff --git a/intl/icu/source/common/usetiter.cpp b/intl/icu/source/common/usetiter.cpp
new file mode 100644
index 0000000000..d24a15ab2d
--- /dev/null
+++ b/intl/icu/source/common/usetiter.cpp
@@ -0,0 +1,152 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+* Copyright (c) 2002-2006, International Business Machines
+* Corporation and others.  All Rights Reserved.
+**********************************************************************
+*/
+#include "unicode/usetiter.h"
+#include "unicode/uniset.h"
+#include "unicode/unistr.h"
+#include "uvector.h"
+
+U_NAMESPACE_BEGIN
+
+UOBJECT_DEFINE_RTTI_IMPLEMENTATION(UnicodeSetIterator)
+
+/**
+ * Create an iterator
+ * @param set set to iterate over
+ */
+UnicodeSetIterator::UnicodeSetIterator(const UnicodeSet& uSet) {
+    cpString  = nullptr;
+    reset(uSet);
+}
+
+/**
+ * Create an iterator. Convenience for when the contents are to be set later.
+ */
+UnicodeSetIterator::UnicodeSetIterator() {
+    this->set = nullptr;
+    cpString  = nullptr;
+    reset();
+}
+
+UnicodeSetIterator::~UnicodeSetIterator() {
+    delete cpString;
+}
+
+/**
+ * Returns the next element in the set.
+ * @return true if there was another element in the set.
+ * if so, if codepoint == IS_STRING, the value is a string in the string field
+ * else the value is a single code point in the codepoint field.
+ * 
You are guaranteed that the codepoints are in sorted order, and the strings are in sorted order,
+ * and that all code points are returned before any strings are returned.
+ * 
Note also that the codepointEnd is undefined after calling this method.
+ */
+UBool UnicodeSetIterator::next() {
+    if (nextElement <= endElement) {
+        codepoint = codepointEnd = nextElement++;
+        string = nullptr;
+        return true;
+    }
+    if (range < endRange) {
+        loadRange(++range);
+        codepoint = codepointEnd = nextElement++;
+        string = nullptr;
+        return true;
+    }
+
+    if (nextString >= stringCount) return false;
+    codepoint = (UChar32)IS_STRING; // signal that value is actually a string
+    string = (const UnicodeString*) set->strings->elementAt(nextString++);
+    return true;
+}
+
+/**
+ * @return true if there was another element in the set.
+ * if so, if codepoint == IS_STRING, the value is a string in the string field
+ * else the value is a range of codepoints in the  fields.
+ * 
Note that the codepoints are in sorted order, and the strings are in sorted order,
+ * and that all code points are returned before any strings are returned.
+ * 
You are guaranteed that the ranges are in sorted order, and the strings are in sorted order,
+ * and that all ranges are returned before any strings are returned.
+ * 
You are also guaranteed that ranges are disjoint and non-contiguous.
+ * 
Note also that the codepointEnd is undefined after calling this method.
+ */
+UBool UnicodeSetIterator::nextRange() {
+    string = nullptr;
+    if (nextElement <= endElement) {
+        codepointEnd = endElement;
+        codepoint = nextElement;
+        nextElement = endElement+1;
+        return true;
+    }
+    if (range < endRange) {
+        loadRange(++range);
+        codepointEnd = endElement;
+        codepoint = nextElement;
+        nextElement = endElement+1;
+        return true;
+    }
+
+    if (nextString >= stringCount) return false;
+    codepoint = (UChar32)IS_STRING; // signal that value is actually a string
+    string = (const UnicodeString*) set->strings->elementAt(nextString++);
+    return true;
+}
+
+/**
+ *@param set the set to iterate over. This allows reuse of the iterator.
+ */
+void UnicodeSetIterator::reset(const UnicodeSet& uSet) {
+    this->set = &uSet;
+    reset();
+}
+
+/**
+ * Resets to the start, to allow the iteration to start over again.
+ */
+void UnicodeSetIterator::reset() {
+    if (set == nullptr) {
+        // Set up indices to empty iteration
+        endRange = -1;
+        stringCount = 0;
+    } else {
+        endRange = set->getRangeCount() - 1;
+        stringCount = set->stringsSize();
+    }
+    range = 0;
+    endElement = -1;
+    nextElement = 0;            
+    if (endRange >= 0) {
+        loadRange(range);
+    }
+    nextString = 0;
+    string = nullptr;
+}
+
+void UnicodeSetIterator::loadRange(int32_t iRange) {
+    nextElement = set->getRangeStart(iRange);
+    endElement = set->getRangeEnd(iRange);
+}
+
+
+const UnicodeString& UnicodeSetIterator::getString()  {
+    if (string==nullptr && codepoint!=(UChar32)IS_STRING) {
+       if (cpString == nullptr) {
+          cpString = new UnicodeString();
+       }
+       if (cpString != nullptr) {
+          cpString->setTo((UChar32)codepoint);
+       }
+       string = cpString;
+    }
+    return *string;
+}
+
+U_NAMESPACE_END
+
+//eof
diff --git a/intl/icu/source/common/ushape.cpp b/intl/icu/source/common/ushape.cpp
new file mode 100644
index 0000000000..d0ac95e0b2
--- /dev/null
+++ b/intl/icu/source/common/ushape.cpp
@@ -0,0 +1,1728 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+ ******************************************************************************
+ *
+ *   Copyright (C) 2000-2016, International Business Machines
+ *   Corporation and others.  All Rights Reserved.
+ *
+ ******************************************************************************
+ *   file name:  ushape.cpp
+ *   encoding:   UTF-8
+ *   tab size:   8 (not used)
+ *   indentation:4
+ *
+ *   created on: 2000jun29
+ *   created by: Markus W. Scherer
+ *
+ *   Arabic letter shaping implemented by Ayman Roshdy
+ */
+
+#include "unicode/utypes.h"
+#include "unicode/uchar.h"
+#include "unicode/ustring.h"
+#include "unicode/ushape.h"
+#include "cmemory.h"
+#include "putilimp.h"
+#include "ustr_imp.h"
+#include "ubidi_props.h"
+#include "uassert.h"
+
+/*
+ * This implementation is designed for 16-bit Unicode strings.
+ * The main assumption is that the Arabic characters and their
+ * presentation forms each fit into a single char16_t.
+ * With UTF-8, they occupy 2 or 3 bytes, and more than the ASCII
+ * characters.
+ */
+
+/*
+ * ### TODO in general for letter shaping:
+ * - the letter shaping code is UTF-16-unaware; needs update
+ *   + especially invertBuffer()?!
+ * - needs to handle the "Arabic Tail" that is used in some legacy codepages
+ *   as a glyph fragment of wide-glyph letters
+ *   + IBM Unicode conversion tables map it to U+200B (ZWSP)
+ *   + IBM Egypt has proposed to encode the tail in Unicode among Arabic Presentation Forms
+ *   + Unicode 3.2 added U+FE73 ARABIC TAIL FRAGMENT
+ */
+
+/* definitions for Arabic letter shaping ------------------------------------ */
+
+#define IRRELEVANT 4
+#define LAMTYPE    16
+#define ALEFTYPE   32
+#define LINKR      1
+#define LINKL      2
+#define APRESENT   8
+#define SHADDA     64
+#define CSHADDA    128
+#define COMBINE    (SHADDA+CSHADDA)
+
+#define HAMZAFE_CHAR       0xfe80
+#define HAMZA06_CHAR       0x0621
+#define YEH_HAMZA_CHAR     0x0626
+#define YEH_HAMZAFE_CHAR   0xFE89
+#define LAMALEF_SPACE_SUB  0xFFFF
+#define TASHKEEL_SPACE_SUB 0xFFFE
+#define NEW_TAIL_CHAR      0xFE73
+#define OLD_TAIL_CHAR      0x200B
+#define LAM_CHAR           0x0644
+#define SPACE_CHAR         0x0020
+#define SHADDA_CHAR        0xFE7C
+#define TATWEEL_CHAR       0x0640
+#define SHADDA_TATWEEL_CHAR  0xFE7D
+#define SHADDA06_CHAR      0x0651
+
+#define SHAPE_MODE   0
+#define DESHAPE_MODE 1
+
+struct uShapeVariables {
+     char16_t tailChar;
+     uint32_t uShapeLamalefBegin;
+     uint32_t uShapeLamalefEnd;
+     uint32_t uShapeTashkeelBegin;
+     uint32_t uShapeTashkeelEnd;
+     int spacesRelativeToTextBeginEnd;
+};
+
+static const uint8_t tailFamilyIsolatedFinal[] = {
+    /* FEB1 */ 1,
+    /* FEB2 */ 1,
+    /* FEB3 */ 0,
+    /* FEB4 */ 0,
+    /* FEB5 */ 1,
+    /* FEB6 */ 1,
+    /* FEB7 */ 0,
+    /* FEB8 */ 0,
+    /* FEB9 */ 1,
+    /* FEBA */ 1,
+    /* FEBB */ 0,
+    /* FEBC */ 0,
+    /* FEBD */ 1,
+    /* FEBE */ 1
+};
+
+static const uint8_t tashkeelMedial[] = {
+    /* FE70 */ 0,
+    /* FE71 */ 1,
+    /* FE72 */ 0,
+    /* FE73 */ 0,
+    /* FE74 */ 0,
+    /* FE75 */ 0,
+    /* FE76 */ 0,
+    /* FE77 */ 1,
+    /* FE78 */ 0,
+    /* FE79 */ 1,
+    /* FE7A */ 0,
+    /* FE7B */ 1,
+    /* FE7C */ 0,
+    /* FE7D */ 1,
+    /* FE7E */ 0,
+    /* FE7F */ 1
+};
+
+static const char16_t yehHamzaToYeh[] =
+{
+/* isolated*/ 0xFEEF,
+/* final   */ 0xFEF0
+};
+
+static const uint8_t IrrelevantPos[] = {
+    0x0, 0x2, 0x4, 0x6,
+    0x8, 0xA, 0xC, 0xE
+};
+
+
+static const char16_t convertLamAlef[] =
+{
+/*FEF5*/    0x0622,
+/*FEF6*/    0x0622,
+/*FEF7*/    0x0623,
+/*FEF8*/    0x0623,
+/*FEF9*/    0x0625,
+/*FEFA*/    0x0625,
+/*FEFB*/    0x0627,
+/*FEFC*/    0x0627
+};
+
+static const char16_t araLink[178]=
+{
+  1           + 32 + 256 * 0x11,/*0x0622*/
+  1           + 32 + 256 * 0x13,/*0x0623*/
+  1                + 256 * 0x15,/*0x0624*/
+  1           + 32 + 256 * 0x17,/*0x0625*/
+  1 + 2            + 256 * 0x19,/*0x0626*/
+  1           + 32 + 256 * 0x1D,/*0x0627*/
+  1 + 2            + 256 * 0x1F,/*0x0628*/
+  1                + 256 * 0x23,/*0x0629*/
+  1 + 2            + 256 * 0x25,/*0x062A*/
+  1 + 2            + 256 * 0x29,/*0x062B*/
+  1 + 2            + 256 * 0x2D,/*0x062C*/
+  1 + 2            + 256 * 0x31,/*0x062D*/
+  1 + 2            + 256 * 0x35,/*0x062E*/
+  1                + 256 * 0x39,/*0x062F*/
+  1                + 256 * 0x3B,/*0x0630*/
+  1                + 256 * 0x3D,/*0x0631*/
+  1                + 256 * 0x3F,/*0x0632*/
+  1 + 2            + 256 * 0x41,/*0x0633*/
+  1 + 2            + 256 * 0x45,/*0x0634*/
+  1 + 2            + 256 * 0x49,/*0x0635*/
+  1 + 2            + 256 * 0x4D,/*0x0636*/
+  1 + 2            + 256 * 0x51,/*0x0637*/
+  1 + 2            + 256 * 0x55,/*0x0638*/
+  1 + 2            + 256 * 0x59,/*0x0639*/
+  1 + 2            + 256 * 0x5D,/*0x063A*/
+  0, 0, 0, 0, 0,                /*0x063B-0x063F*/
+  1 + 2,                        /*0x0640*/
+  1 + 2            + 256 * 0x61,/*0x0641*/
+  1 + 2            + 256 * 0x65,/*0x0642*/
+  1 + 2            + 256 * 0x69,/*0x0643*/
+  1 + 2       + 16 + 256 * 0x6D,/*0x0644*/
+  1 + 2            + 256 * 0x71,/*0x0645*/
+  1 + 2            + 256 * 0x75,/*0x0646*/
+  1 + 2            + 256 * 0x79,/*0x0647*/
+  1                + 256 * 0x7D,/*0x0648*/
+  1                + 256 * 0x7F,/*0x0649*/
+  1 + 2            + 256 * 0x81,/*0x064A*/
+         4         + 256 * 1,   /*0x064B*/
+         4 + 128   + 256 * 1,   /*0x064C*/
+         4 + 128   + 256 * 1,   /*0x064D*/
+         4 + 128   + 256 * 1,   /*0x064E*/
+         4 + 128   + 256 * 1,   /*0x064F*/
+         4 + 128   + 256 * 1,   /*0x0650*/
+         4 + 64    + 256 * 3,   /*0x0651*/
+         4         + 256 * 1,   /*0x0652*/
+         4         + 256 * 7,   /*0x0653*/
+         4         + 256 * 8,   /*0x0654*/
+         4         + 256 * 8,   /*0x0655*/
+         4         + 256 * 1,   /*0x0656*/
+  0, 0, 0, 0, 0,                /*0x0657-0x065B*/
+  1                + 256 * 0x85,/*0x065C*/
+  1                + 256 * 0x87,/*0x065D*/
+  1                + 256 * 0x89,/*0x065E*/
+  1                + 256 * 0x8B,/*0x065F*/
+  0, 0, 0, 0, 0,                /*0x0660-0x0664*/
+  0, 0, 0, 0, 0,                /*0x0665-0x0669*/
+  0, 0, 0, 0, 0, 0,             /*0x066A-0x066F*/
+         4         + 256 * 6,   /*0x0670*/
+  1        + 8     + 256 * 0x00,/*0x0671*/
+  1            + 32,            /*0x0672*/
+  1            + 32,            /*0x0673*/
+  0,                            /*0x0674*/
+  1            + 32,            /*0x0675*/
+  1, 1,                         /*0x0676-0x0677*/
+  1 + 2,                        /*0x0678*/
+  1 + 2 + 8        + 256 * 0x16,/*0x0679*/
+  1 + 2 + 8        + 256 * 0x0E,/*0x067A*/
+  1 + 2 + 8        + 256 * 0x02,/*0x067B*/
+  1+2, 1+2,                     /*0x67C-0x067D*/
+  1+2+8+256 * 0x06, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x067E-0x0683*/
+  1+2, 1+2, 1+2+8+256 * 0x2A, 1+2,           /*0x0684-0x0687*/
+  1     + 8        + 256 * 0x38,/*0x0688*/
+  1, 1, 1,                      /*0x0689-0x068B*/
+  1     + 8        + 256 * 0x34,/*0x068C*/
+  1     + 8        + 256 * 0x32,/*0x068D*/
+  1     + 8        + 256 * 0x36,/*0x068E*/
+  1, 1,                         /*0x068F-0x0690*/
+  1     + 8        + 256 * 0x3C,/*0x0691*/
+  1, 1, 1, 1, 1, 1, 1+8+256 * 0x3A, 1,       /*0x0692-0x0699*/
+  1+2, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x069A-0x06A3*/
+  1+2, 1+2, 1+2, 1+2,           /*0x069A-0x06A3*/
+  1+2, 1+2, 1+2, 1+2, 1+2, 1+2+8+256 * 0x3E, /*0x06A4-0x06AD*/
+  1+2, 1+2, 1+2, 1+2,           /*0x06A4-0x06AD*/
+  1+2, 1+2+8+256 * 0x42, 1+2, 1+2, 1+2, 1+2, /*0x06AE-0x06B7*/
+  1+2, 1+2, 1+2, 1+2,           /*0x06AE-0x06B7*/
+  1+2, 1+2,                     /*0x06B8-0x06B9*/
+  1     + 8        + 256 * 0x4E,/*0x06BA*/
+  1 + 2 + 8        + 256 * 0x50,/*0x06BB*/
+  1+2, 1+2,                     /*0x06BC-0x06BD*/
+  1 + 2 + 8        + 256 * 0x5A,/*0x06BE*/
+  1+2,                          /*0x06BF*/
+  1     + 8        + 256 * 0x54,/*0x06C0*/
+  1 + 2 + 8        + 256 * 0x56,/*0x06C1*/
+  1, 1, 1,                      /*0x06C2-0x06C4*/
+  1     + 8        + 256 * 0x90,/*0x06C5*/
+  1     + 8        + 256 * 0x89,/*0x06C6*/
+  1     + 8        + 256 * 0x87,/*0x06C7*/
+  1     + 8        + 256 * 0x8B,/*0x06C8*/
+  1     + 8        + 256 * 0x92,/*0x06C9*/
+  1,                            /*0x06CA*/
+  1     + 8        + 256 * 0x8E,/*0x06CB*/
+  1 + 2 + 8        + 256 * 0xAC,/*0x06CC*/
+  1,                            /*0x06CD*/
+  1+2, 1+2,                     /*0x06CE-0x06CF*/
+  1 + 2 + 8        + 256 * 0x94,/*0x06D0*/
+  1+2,                          /*0x06D1*/
+  1     + 8        + 256 * 0x5E,/*0x06D2*/
+  1     + 8        + 256 * 0x60 /*0x06D3*/
+};
+
+static const uint8_t presALink[] = {
+/***********0*****1*****2*****3*****4*****5*****6*****7*****8*****9*****A*****B*****C*****D*****E*****F*/
+/*FB5*/    0,    1,    0,    0,    0,    0,    0,    1,    2,1 + 2,    0,    0,    0,    0,    0,    0,
+/*FB6*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+/*FB7*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    1,    2,1 + 2,    0,    0,
+/*FB8*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    1,    0,    0,    0,    1,
+/*FB9*/    2,1 + 2,    0,    1,    2,1 + 2,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+/*FBA*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+/*FBB*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+/*FBC*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+/*FBD*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+/*FBE*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+/*FBF*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    1,    2,1 + 2,
+/*FC0*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+/*FC1*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+/*FC2*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+/*FC3*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+/*FC4*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,
+/*FC5*/    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    4,    4,
+/*FC6*/    4,    4,    4
+};
+
+static const uint8_t presBLink[]=
+{
+/***********0*****1*****2*****3*****4*****5*****6*****7*****8*****9*****A*****B*****C*****D*****E*****F*/
+/*FE7*/1 + 2,1 + 2,1 + 2,    0,1 + 2,    0,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,
+/*FE8*/    0,    0,    1,    0,    1,    0,    1,    0,    1,    0,    1,    2,1 + 2,    0,    1,    0,
+/*FE9*/    1,    2,1 + 2,    0,    1,    0,    1,    2,1 + 2,    0,    1,    2,1 + 2,    0,    1,    2,
+/*FEA*/1 + 2,    0,    1,    2,1 + 2,    0,    1,    2,1 + 2,    0,    1,    0,    1,    0,    1,    0,
+/*FEB*/    1,    0,    1,    2,1 + 2,    0,    1,    2,1 + 2,    0,    1,    2,1 + 2,    0,    1,    2,
+/*FEC*/1 + 2,    0,    1,    2,1 + 2,    0,    1,    2,1 + 2,    0,    1,    2,1 + 2,    0,    1,    2,
+/*FED*/1 + 2,    0,    1,    2,1 + 2,    0,    1,    2,1 + 2,    0,    1,    2,1 + 2,    0,    1,    2,
+/*FEE*/1 + 2,    0,    1,    2,1 + 2,    0,    1,    2,1 + 2,    0,    1,    2,1 + 2,    0,    1,    0,
+/*FEF*/    1,    0,    1,    2,1 + 2,    0,    1,    0,    1,    0,    1,    0,    1,    0,    0,    0
+};
+
+static const char16_t convertFBto06[] =
+{
+/***********0******1******2******3******4******5******6******7******8******9******A******B******C******D******E******F***/
+/*FB5*/   0x671, 0x671, 0x67B, 0x67B, 0x67B, 0x67B, 0x67E, 0x67E, 0x67E, 0x67E,     0,     0,     0,     0, 0x67A, 0x67A,
+/*FB6*/   0x67A, 0x67A,     0,     0,     0,     0, 0x679, 0x679, 0x679, 0x679,     0,     0,     0,     0,     0,     0,
+/*FB7*/       0,     0,     0,     0,     0,     0,     0,     0,     0,     0, 0x686, 0x686, 0x686, 0x686,     0,     0,
+/*FB8*/       0,     0, 0x68D, 0x68D, 0x68C, 0x68C, 0x68E, 0x68E, 0x688, 0x688, 0x698, 0x698, 0x691, 0x691, 0x6A9, 0x6A9,
+/*FB9*/   0x6A9, 0x6A9, 0x6AF, 0x6AF, 0x6AF, 0x6AF,     0,     0,     0,     0,     0,     0,     0,     0, 0x6BA, 0x6BA,
+/*FBA*/   0x6BB, 0x6BB, 0x6BB, 0x6BB, 0x6C0, 0x6C0, 0x6C1, 0x6C1, 0x6C1, 0x6C1, 0x6BE, 0x6BE, 0x6BE, 0x6BE, 0x6d2, 0x6D2,
+/*FBB*/   0x6D3, 0x6D3,     0,     0,     0,     0,     0,     0,     0,     0,     0,     0,     0,     0,     0,     0,
+/*FBC*/       0,     0,     0,     0,     0,     0,     0,     0,     0,     0,     0,     0,     0,     0,     0,     0,
+/*FBD*/       0,     0,     0,     0,     0,     0,     0, 0x6C7, 0x6C7, 0x6C6, 0x6C6, 0x6C8, 0x6C8,     0, 0x6CB, 0x6CB,
+/*FBE*/   0x6C5, 0x6C5, 0x6C9, 0x6C9, 0x6D0, 0x6D0, 0x6D0, 0x6D0,     0,     0,     0,     0,     0,     0,     0,     0,
+/*FBF*/       0,     0,     0,     0,     0,     0,     0,     0,     0,     0,     0,     0, 0x6CC, 0x6CC, 0x6CC, 0x6CC
+};
+
+static const char16_t convertFEto06[] =
+{
+/***********0******1******2******3******4******5******6******7******8******9******A******B******C******D******E******F***/
+/*FE7*/   0x64B, 0x64B, 0x64C, 0x64C, 0x64D, 0x64D, 0x64E, 0x64E, 0x64F, 0x64F, 0x650, 0x650, 0x651, 0x651, 0x652, 0x652,
+/*FE8*/   0x621, 0x622, 0x622, 0x623, 0x623, 0x624, 0x624, 0x625, 0x625, 0x626, 0x626, 0x626, 0x626, 0x627, 0x627, 0x628,
+/*FE9*/   0x628, 0x628, 0x628, 0x629, 0x629, 0x62A, 0x62A, 0x62A, 0x62A, 0x62B, 0x62B, 0x62B, 0x62B, 0x62C, 0x62C, 0x62C,
+/*FEA*/   0x62C, 0x62D, 0x62D, 0x62D, 0x62D, 0x62E, 0x62E, 0x62E, 0x62E, 0x62F, 0x62F, 0x630, 0x630, 0x631, 0x631, 0x632,
+/*FEB*/   0x632, 0x633, 0x633, 0x633, 0x633, 0x634, 0x634, 0x634, 0x634, 0x635, 0x635, 0x635, 0x635, 0x636, 0x636, 0x636,
+/*FEC*/   0x636, 0x637, 0x637, 0x637, 0x637, 0x638, 0x638, 0x638, 0x638, 0x639, 0x639, 0x639, 0x639, 0x63A, 0x63A, 0x63A,
+/*FED*/   0x63A, 0x641, 0x641, 0x641, 0x641, 0x642, 0x642, 0x642, 0x642, 0x643, 0x643, 0x643, 0x643, 0x644, 0x644, 0x644,
+/*FEE*/   0x644, 0x645, 0x645, 0x645, 0x645, 0x646, 0x646, 0x646, 0x646, 0x647, 0x647, 0x647, 0x647, 0x648, 0x648, 0x649,
+/*FEF*/   0x649, 0x64A, 0x64A, 0x64A, 0x64A, 0x65C, 0x65C, 0x65D, 0x65D, 0x65E, 0x65E, 0x65F, 0x65F
+};
+
+static const uint8_t shapeTable[4][4][4]=
+{
+  { {0,0,0,0}, {0,0,0,0}, {0,1,0,3}, {0,1,0,1} },
+  { {0,0,2,2}, {0,0,1,2}, {0,1,1,2}, {0,1,1,3} },
+  { {0,0,0,0}, {0,0,0,0}, {0,1,0,3}, {0,1,0,3} },
+  { {0,0,1,2}, {0,0,1,2}, {0,1,1,2}, {0,1,1,3} }
+};
+
+/*
+ * This function shapes European digits to Arabic-Indic digits
+ * in-place, writing over the input characters.
+ * Since we know that we are only looking for BMP code points,
+ * we can safely just work with code units (again, at least UTF-16).
+ */
+static void
+_shapeToArabicDigitsWithContext(char16_t *s, int32_t length,
+                                char16_t digitBase,
+                                UBool isLogical, UBool lastStrongWasAL) {
+    int32_t i;
+    char16_t c;
+
+    digitBase-=0x30;
+
+    /* the iteration direction depends on the type of input */
+    if(isLogical) {
+        for(i=0; i0; /* pre-decrement in the body */) {
+            c=s[--i];
+            switch(ubidi_getClass(c)) {
+            case U_LEFT_TO_RIGHT: /* L */
+            case U_RIGHT_TO_LEFT: /* R */
+                lastStrongWasAL=false;
+                break;
+            case U_RIGHT_TO_LEFT_ARABIC: /* AL */
+                lastStrongWasAL=true;
+                break;
+            case U_EUROPEAN_NUMBER: /* EN */
+                if(lastStrongWasAL && (uint32_t)(c-0x30)<10) {
+                    s[i]=(char16_t)(digitBase+c); /* digitBase+(c-0x30) - digitBase was modified above */
+                }
+                break;
+            default :
+                break;
+            }
+        }
+    }
+}
+
+/*
+ *Name     : invertBuffer
+ *Function : This function inverts the buffer, it's used
+ *           in case the user specifies the buffer to be
+ *           U_SHAPE_TEXT_DIRECTION_LOGICAL
+ */
+static void
+invertBuffer(char16_t *buffer, int32_t size, uint32_t /*options*/, int32_t lowlimit, int32_t highlimit) {
+    char16_t temp;
+    int32_t i=0,j=0;
+    for(i=lowlimit,j=size-highlimit-1;i= 0x0622 && ch <= 0x06D3) {
+        return(araLink[ch-0x0622]);
+    } else if(ch == 0x200D) {
+        return(3);
+    } else if(ch >= 0x206D && ch <= 0x206F) {
+        return(4);
+    }else if(ch >= 0xFB50 && ch <= 0xFC62) {
+        return(presALink[ch-0xFB50]);
+    } else if(ch >= 0xFE70 && ch <= 0xFEFC) {
+        return(presBLink[ch-0xFE70]);
+    }else {
+        return(0);
+    }
+}
+
+/*
+ *Name     : countSpaces
+ *Function : Counts the number of spaces
+ *           at each end of the logical buffer
+ */
+static void
+countSpaces(char16_t *dest, int32_t size, uint32_t /*options*/, int32_t *spacesCountl, int32_t *spacesCountr) {
+    int32_t i = 0;
+    int32_t countl = 0,countr = 0;
+    while((dest[i] == SPACE_CHAR) && (countl < size)) {
+       countl++;
+       i++;
+    }
+    if (countl < size) {  /* the entire buffer is not all space */
+        while(dest[size-1] == SPACE_CHAR) {
+            countr++;
+            size--;
+        }
+    }
+    *spacesCountl = countl;
+    *spacesCountr = countr;
+}
+
+/*
+ *Name     : isTashkeelChar
+ *Function : Returns 1 for Tashkeel characters in 06 range else return 0
+ */
+static inline int32_t
+isTashkeelChar(char16_t ch) {
+    return (int32_t)( ch>=0x064B && ch<= 0x0652 );
+}
+
+/*
+ *Name     : isTashkeelCharFE
+ *Function : Returns 1 for Tashkeel characters in FE range else return 0
+ */
+static inline int32_t
+isTashkeelCharFE(char16_t ch) {
+    return (int32_t)( ch>=0xFE70 && ch<= 0xFE7F );
+}
+
+/*
+ *Name     : isAlefChar
+ *Function : Returns 1 for Alef characters else return 0
+ */
+static inline int32_t
+isAlefChar(char16_t ch) {
+    return (int32_t)( (ch==0x0622)||(ch==0x0623)||(ch==0x0625)||(ch==0x0627) );
+}
+
+/*
+ *Name     : isLamAlefChar
+ *Function : Returns 1 for LamAlef characters else return 0
+ */
+static inline int32_t
+isLamAlefChar(char16_t ch) {
+    return (int32_t)((ch>=0xFEF5)&&(ch<=0xFEFC) );
+}
+
+/*BIDI
+ *Name     : isTailChar
+ *Function : returns 1 if the character matches one of the tail characters (0xfe73 or 0x200b) otherwise returns 0
+ */
+
+static inline int32_t
+isTailChar(char16_t ch) {
+    if(ch == OLD_TAIL_CHAR || ch == NEW_TAIL_CHAR){
+            return 1;
+    }else{
+            return 0;
+    }
+}
+
+/*BIDI
+ *Name     : isSeenTailFamilyChar
+ *Function : returns 1 if the character is a seen family isolated character
+ *           in the FE range otherwise returns 0
+ */
+
+static inline int32_t
+isSeenTailFamilyChar(char16_t ch) {
+    if(ch >= 0xfeb1 && ch < 0xfebf){
+            return tailFamilyIsolatedFinal [ch - 0xFEB1];
+    }else{
+            return 0;
+    }
+}
+
+ /* Name     : isSeenFamilyChar
+  * Function : returns 1 if the character is a seen family character in the Unicode
+  *            06 range otherwise returns 0
+ */
+
+static inline int32_t
+isSeenFamilyChar(char16_t  ch){
+    if(ch >= 0x633 && ch <= 0x636){
+        return 1;
+    }else {
+        return 0;
+    }
+}
+
+/*Start of BIDI*/
+/*
+ *Name     : isAlefMaksouraChar
+ *Function : returns 1 if the character is a Alef Maksoura Final or isolated
+ *           otherwise returns 0
+ */
+static inline int32_t
+isAlefMaksouraChar(char16_t ch) {
+    return (int32_t)( (ch == 0xFEEF) || ( ch == 0xFEF0) || (ch == 0x0649));
+}
+
+/*
+ * Name     : isYehHamzaChar
+ * Function : returns 1 if the character is a yehHamza isolated or yehhamza
+ *            final is found otherwise returns 0
+ */
+static inline int32_t
+isYehHamzaChar(char16_t ch) {
+    if((ch==0xFE89)||(ch==0xFE8A)){
+        return 1;
+    }else{
+        return 0;
+    }
+}
+
+ /*
+ * Name: isTashkeelOnTatweelChar
+ * Function: Checks if the Tashkeel Character is on Tatweel or not,if the
+ *           Tashkeel on tatweel (FE range), it returns 1 else if the
+ *           Tashkeel with shadda on tatweel (FC range)return 2 otherwise
+ *           returns 0
+ */
+static inline int32_t
+isTashkeelOnTatweelChar(char16_t ch){
+    if(ch >= 0xfe70 && ch <= 0xfe7f && ch != NEW_TAIL_CHAR && ch != 0xFE75 && ch != SHADDA_TATWEEL_CHAR)
+    {
+        return tashkeelMedial [ch - 0xFE70];
+    }else if( (ch >= 0xfcf2 && ch <= 0xfcf4) || (ch == SHADDA_TATWEEL_CHAR)) {
+        return 2;
+    }else{
+        return 0;
+    }
+}
+
+/*
+ * Name: isIsolatedTashkeelChar
+ * Function: Checks if the Tashkeel Character is in the isolated form
+ *           (i.e. Unicode FE range) returns 1 else if the Tashkeel
+ *           with shadda is in the isolated form (i.e. Unicode FC range)
+ *           returns 2 otherwise returns 0
+ */
+static inline int32_t
+isIsolatedTashkeelChar(char16_t ch){
+    if(ch >= 0xfe70 && ch <= 0xfe7f && ch != NEW_TAIL_CHAR && ch != 0xFE75){
+        return (1 - tashkeelMedial [ch - 0xFE70]);
+    }else if(ch >= 0xfc5e && ch <= 0xfc63){
+        return 1;
+    }else{
+        return 0;
+    }
+}
+
+
+
+
+/*
+ *Name     : calculateSize
+ *Function : This function calculates the destSize to be used in preflighting
+ *           when the destSize is equal to 0
+ *           It is used also to calculate the new destsize in case the
+ *           destination buffer will be resized.
+ */
+
+static int32_t
+calculateSize(const char16_t *source, int32_t sourceLength,
+int32_t destSize,uint32_t options) {
+    int32_t i = 0;
+
+    int lamAlefOption = 0;
+    int tashkeelOption = 0;
+
+    destSize = sourceLength;
+
+    if (((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_SHAPE ||
+        ((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED )) &&
+        ((options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_RESIZE )){
+            lamAlefOption = 1;
+    }
+    if((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_SHAPE &&
+       ((options&U_SHAPE_TASHKEEL_MASK) == U_SHAPE_TASHKEEL_RESIZE ) ){
+            tashkeelOption = 1;
+        }
+
+    if(lamAlefOption || tashkeelOption){
+        if((options&U_SHAPE_TEXT_DIRECTION_MASK)==U_SHAPE_TEXT_DIRECTION_VISUAL_LTR) {
+            for(i=0;i= 0) {
+            tempbuffer[i] = 0x0000;
+            i--;
+            count--;
+        }
+
+        u_memcpy(dest, tempbuffer, sourceLength);
+        destSize = u_strlen(dest);
+    }
+
+      lamAlefOption = 0;
+
+    if (shapingMode == 0){
+        if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_NEAR ){
+            lamAlefOption = 1;
+        }
+    }
+
+    if (lamAlefOption){
+        /* Lam+Alef is already shaped into LamAlef + FFFF */
+        i = 0;
+        while(i < sourceLength) {
+            if(lamAlefOption&&dest[i] == LAMALEF_SPACE_SUB){
+                dest[i] = SPACE_CHAR;
+            }
+            i++;
+        }
+        destSize = sourceLength;
+    }
+    lamAlefOption = 0;
+    tashkeelOption = 0;
+
+    if (shapingMode == 0) {
+        if ( ((options&U_SHAPE_LAMALEF_MASK) == shapeVars.uShapeLamalefBegin) ||
+              (((options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_AUTO )
+              && (shapeVars.spacesRelativeToTextBeginEnd==1)) ) {
+            lamAlefOption = 1;
+        }
+        if ( (options&U_SHAPE_TASHKEEL_MASK) == shapeVars.uShapeTashkeelBegin ) {
+            tashkeelOption = 1;
+        }
+    }
+
+    if(lamAlefOption || tashkeelOption){
+        uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
+        
+        i = j = sourceLength; count = 0;
+        
+        while(i >= 0) {
+            if ( (lamAlefOption && dest[i] == LAMALEF_SPACE_SUB) ||
+                 (tashkeelOption && dest[i] == TASHKEEL_SPACE_SUB) ){
+                j++;
+                count++;
+            }else {
+                tempbuffer[j] = dest[i];
+            }
+            i--;
+            j--;
+        }
+
+        for(i=0 ;i < count; i++){
+                tempbuffer[i] = SPACE_CHAR;
+        }
+
+        u_memcpy(dest, tempbuffer, sourceLength);
+        destSize = sourceLength;
+    }
+
+
+
+    lamAlefOption = 0;
+    tashkeelOption = 0;
+
+    if (shapingMode == 0) {
+        if ( ((options&U_SHAPE_LAMALEF_MASK) == shapeVars.uShapeLamalefEnd) ||
+              (((options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_AUTO )
+              && (shapeVars.spacesRelativeToTextBeginEnd==0)) ) {
+            lamAlefOption = 1;
+        }
+        if ( (options&U_SHAPE_TASHKEEL_MASK) == shapeVars.uShapeTashkeelEnd ){
+            tashkeelOption = 1;
+        }
+    }
+
+    if(lamAlefOption || tashkeelOption){
+        uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
+
+        i = j = 0; count = 0;
+        while(i < sourceLength) {
+            if ( (lamAlefOption && dest[i] == LAMALEF_SPACE_SUB) ||
+                 (tashkeelOption && dest[i] == TASHKEEL_SPACE_SUB) ){
+                j--;
+                count++;
+            }else {
+                tempbuffer[j] = dest[i];
+            }
+            i++;
+            j++;
+        }
+
+        while(count >= 0) {
+            tempbuffer[i] = SPACE_CHAR;
+            i--;
+            count--;
+        }
+
+        u_memcpy(dest, tempbuffer, sourceLength);
+        destSize = sourceLength;
+    }
+
+
+    if(tempbuffer){
+        uprv_free(tempbuffer);
+    }
+
+    return destSize;
+}
+
+/*
+ *Name     :expandCompositCharAtBegin
+ *Function :Expands the LamAlef character to Lam and Alef consuming the required
+ *         space from beginning of the buffer. If the text type was visual_LTR
+ *         and the option U_SHAPE_SPACES_RELATIVE_TO_TEXT_BEGIN_END was selected
+ *         the spaces will be located at end of buffer.
+ *         If there are no spaces to expand the LamAlef, an error
+ *         will be set to U_NO_SPACE_AVAILABLE as defined in utypes.h
+ */
+
+static int32_t
+expandCompositCharAtBegin(char16_t *dest, int32_t sourceLength, int32_t destSize,UErrorCode *pErrorCode) {
+    int32_t      i = 0,j = 0;
+    int32_t      countl = 0;
+    char16_t *tempbuffer=nullptr;
+
+    tempbuffer = (char16_t *)uprv_malloc((sourceLength+1)*U_SIZEOF_UCHAR);
+
+    /* Test for nullptr */
+    if(tempbuffer == nullptr) {
+        *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
+        return 0;
+    }
+
+        uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
+
+        i = 0;
+        while(dest[i] == SPACE_CHAR) {
+            countl++;
+            i++;
+        }
+
+        i = j = sourceLength-1;
+
+        while(i >= 0 && j >= 0) {
+            if( countl>0 && isLamAlefChar(dest[i])) {
+                tempbuffer[j] = LAM_CHAR;
+                /* to ensure the array index is within the range */
+                U_ASSERT(dest[i] >= 0xFEF5u
+                    && dest[i]-0xFEF5u < UPRV_LENGTHOF(convertLamAlef));
+                tempbuffer[j-1] = convertLamAlef[ dest[i] - 0xFEF5 ];
+                j--;
+                countl--;
+            }else {
+                 if( countl == 0 && isLamAlefChar(dest[i]) ) {
+                     *pErrorCode=U_NO_SPACE_AVAILABLE;
+                     }
+                 tempbuffer[j] = dest[i];
+            }
+            i--;
+            j--;
+        }
+        u_memcpy(dest, tempbuffer, sourceLength);
+
+        uprv_free(tempbuffer);
+
+        destSize = sourceLength;
+        return destSize;
+}
+
+/*
+ *Name     : expandCompositCharAtEnd
+ *Function : Expands the LamAlef character to Lam and Alef consuming the
+ *           required space from end of the buffer. If the text type was
+ *           Visual LTR and the option U_SHAPE_SPACES_RELATIVE_TO_TEXT_BEGIN_END
+ *           was used, the spaces will be consumed from begin of buffer. If
+ *           there are no spaces to expand the LamAlef, an error
+ *           will be set to U_NO_SPACE_AVAILABLE as defined in utypes.h
+ */
+
+static int32_t
+expandCompositCharAtEnd(char16_t *dest, int32_t sourceLength, int32_t destSize,UErrorCode *pErrorCode) {
+    int32_t      i = 0,j = 0;
+
+    int32_t      countr = 0;
+    int32_t  inpsize = sourceLength;
+
+    char16_t *tempbuffer=nullptr;
+    tempbuffer = (char16_t *)uprv_malloc((sourceLength+1)*U_SIZEOF_UCHAR);
+
+    /* Test for nullptr */
+    if(tempbuffer == nullptr) {
+        *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
+         return 0;
+    }
+
+    uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
+
+    while(dest[inpsize-1] == SPACE_CHAR) {
+        countr++;
+        inpsize--;
+    }
+
+    i = sourceLength - countr - 1;
+    j = sourceLength - 1;
+
+    while(i >= 0 && j >= 0) {
+        if( countr>0 && isLamAlefChar(dest[i]) ) {
+            tempbuffer[j] = LAM_CHAR;
+            tempbuffer[j-1] = convertLamAlef[ dest[i] - 0xFEF5 ];
+            j--;
+            countr--;
+        }else {
+            if ((countr == 0) && isLamAlefChar(dest[i]) ) {
+                *pErrorCode=U_NO_SPACE_AVAILABLE;
+            }
+            tempbuffer[j] = dest[i];
+        }
+        i--;
+        j--;
+    }
+
+    if(countr > 0) {
+        u_memmove(tempbuffer, tempbuffer+countr, sourceLength);
+        if(u_strlen(tempbuffer) < sourceLength) {
+            for(i=sourceLength-1;i>=sourceLength-countr;i--) {
+                tempbuffer[i] = SPACE_CHAR;
+            }
+        }
+    }
+    u_memcpy(dest, tempbuffer, sourceLength);
+
+    uprv_free(tempbuffer);
+
+    destSize = sourceLength;
+    return destSize;
+}
+
+/*
+ *Name     : expandCompositCharAtNear
+ *Function : Expands the LamAlef character into Lam + Alef, YehHamza character
+ *           into Yeh + Hamza, SeenFamily character into SeenFamily character
+ *           + Tail, while consuming the space next to the character.
+ *           If there are no spaces next to the character, an error
+ *           will be set to U_NO_SPACE_AVAILABLE as defined in utypes.h
+ */
+
+static int32_t
+expandCompositCharAtNear(char16_t *dest, int32_t sourceLength, int32_t destSize,UErrorCode *pErrorCode,
+                         int yehHamzaOption, int seenTailOption, int lamAlefOption, struct uShapeVariables shapeVars) {
+    int32_t      i = 0;
+
+
+    char16_t lamalefChar, yehhamzaChar;
+
+    for(i = 0 ;i<=sourceLength-1;i++) {
+            if (seenTailOption && isSeenTailFamilyChar(dest[i])) {
+                if ((i>0) && (dest[i-1] == SPACE_CHAR) ) {
+                    dest[i-1] = shapeVars.tailChar;
+                }else {
+                    *pErrorCode=U_NO_SPACE_AVAILABLE;
+                }
+            }else if(yehHamzaOption && (isYehHamzaChar(dest[i])) ) {
+                if ((i>0) && (dest[i-1] == SPACE_CHAR) ) {
+                    yehhamzaChar = dest[i];
+                    dest[i] = yehHamzaToYeh[yehhamzaChar - YEH_HAMZAFE_CHAR];
+                    dest[i-1] = HAMZAFE_CHAR;
+                }else {
+
+                    *pErrorCode=U_NO_SPACE_AVAILABLE;
+                }
+            }else if(lamAlefOption && isLamAlefChar(dest[i+1])) {
+                if(dest[i] == SPACE_CHAR){
+                    lamalefChar = dest[i+1];
+                    dest[i+1] = LAM_CHAR;
+                    dest[i] = convertLamAlef[ lamalefChar - 0xFEF5 ];
+                }else {
+                    *pErrorCode=U_NO_SPACE_AVAILABLE;
+                }
+            }
+       }
+       destSize = sourceLength;
+       return destSize;
+}
+ /*
+ * Name     : expandCompositChar
+ * Function : LamAlef, need special handling, since it expands from one
+ *            character into two characters while shaping or deshaping.
+ *            In order to expand it, near or far spaces according to the
+ *            options user specifies. Also buffer size can be increased.
+ *
+ *            For SeenFamily characters and YehHamza only the near option is
+ *            supported, while for LamAlef we can take spaces from begin, end,
+ *            near or even increase the buffer size.
+ *            There is also the Auto option for LamAlef only, which will first
+ *            search for a space at end, begin then near, respectively.
+ *            If there are no spaces to expand these characters, an error will be set to
+ *            U_NO_SPACE_AVAILABLE as defined in utypes.h
+ */
+
+static int32_t
+expandCompositChar(char16_t *dest, int32_t sourceLength,
+              int32_t destSize,uint32_t options,
+              UErrorCode *pErrorCode, int shapingMode,struct uShapeVariables shapeVars) {
+
+    int32_t      i = 0,j = 0;
+
+    char16_t *tempbuffer=nullptr;
+    int yehHamzaOption = 0;
+    int seenTailOption = 0;
+    int lamAlefOption = 0;
+
+    if (shapingMode == 1){
+        if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_AUTO){
+
+            if(shapeVars.spacesRelativeToTextBeginEnd == 0) {
+                destSize = expandCompositCharAtEnd(dest, sourceLength, destSize, pErrorCode);
+
+                if(*pErrorCode == U_NO_SPACE_AVAILABLE) {
+                    *pErrorCode = U_ZERO_ERROR;
+                    destSize = expandCompositCharAtBegin(dest, sourceLength, destSize, pErrorCode);
+                }
+            }else {
+                destSize = expandCompositCharAtBegin(dest, sourceLength, destSize, pErrorCode);
+
+                if(*pErrorCode == U_NO_SPACE_AVAILABLE) {
+                    *pErrorCode = U_ZERO_ERROR;
+                    destSize = expandCompositCharAtEnd(dest, sourceLength, destSize, pErrorCode);
+                }
+            }
+
+            if(*pErrorCode == U_NO_SPACE_AVAILABLE) {
+                *pErrorCode = U_ZERO_ERROR;
+                destSize = expandCompositCharAtNear(dest, sourceLength, destSize, pErrorCode, yehHamzaOption,
+                                                seenTailOption, 1,shapeVars);
+            }
+        }
+    }
+
+    if (shapingMode == 1){
+        if ( (options&U_SHAPE_LAMALEF_MASK) == shapeVars.uShapeLamalefEnd){
+            destSize = expandCompositCharAtEnd(dest, sourceLength, destSize, pErrorCode);
+        }
+    }
+
+    if (shapingMode == 1){
+        if ( (options&U_SHAPE_LAMALEF_MASK) == shapeVars.uShapeLamalefBegin){
+            destSize = expandCompositCharAtBegin(dest, sourceLength, destSize, pErrorCode);
+        }
+    }
+
+    if (shapingMode == 0){
+         if ((options&U_SHAPE_YEHHAMZA_MASK) == U_SHAPE_YEHHAMZA_TWOCELL_NEAR){
+             yehHamzaOption = 1;
+         }
+         if ((options&U_SHAPE_SEEN_MASK) == U_SHAPE_SEEN_TWOCELL_NEAR){
+            seenTailOption = 1;
+         }
+    }
+    if (shapingMode == 1) {
+        if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_NEAR) {
+            lamAlefOption = 1;
+        }
+    }
+
+
+    if (yehHamzaOption || seenTailOption || lamAlefOption){
+        destSize = expandCompositCharAtNear(dest, sourceLength, destSize, pErrorCode, yehHamzaOption,
+                                            seenTailOption,lamAlefOption,shapeVars);
+    }
+
+
+    if (shapingMode == 1){
+        if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_RESIZE){
+            destSize = calculateSize(dest,sourceLength,destSize,options);
+            tempbuffer = (char16_t *)uprv_malloc((destSize+1)*U_SIZEOF_UCHAR);
+
+            /* Test for nullptr */
+            if(tempbuffer == nullptr) {
+                *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
+                return 0;
+            }
+
+            uprv_memset(tempbuffer, 0, (destSize+1)*U_SIZEOF_UCHAR);
+
+            i = j = 0;
+            while(i < destSize && j < destSize) {
+                if(isLamAlefChar(dest[i]) ) {
+                    tempbuffer[j] = convertLamAlef[ dest[i] - 0xFEF5 ];
+                    tempbuffer[j+1] = LAM_CHAR;
+                    j++;
+                }else {
+                    tempbuffer[j] = dest[i];
+                }
+                i++;
+                j++;
+            }
+
+            u_memcpy(dest, tempbuffer, destSize);
+        }
+    }
+
+    if(tempbuffer) {
+        uprv_free(tempbuffer);
+    }
+    return destSize;
+}
+
+/*
+ *Name     : shapeUnicode
+ *Function : Converts an Arabic Unicode buffer in 06xx Range into a shaped
+ *           arabic Unicode buffer in FExx Range
+ */
+static int32_t
+shapeUnicode(char16_t *dest, int32_t sourceLength,
+             int32_t destSize,uint32_t options,
+             UErrorCode *pErrorCode,
+             int tashkeelFlag, struct uShapeVariables shapeVars) {
+
+    int32_t          i, iend;
+    int32_t          step;
+    int32_t          lastPos,Nx, Nw;
+    unsigned int     Shape;
+    int32_t          lamalef_found = 0;
+    int32_t seenfamFound = 0, yehhamzaFound =0, tashkeelFound  = 0;
+    char16_t         prevLink = 0, lastLink = 0, currLink, nextLink = 0;
+    char16_t         wLamalef;
+
+    /*
+     * Converts the input buffer from FExx Range into 06xx Range
+     * to make sure that all characters are in the 06xx range
+     * even the lamalef is converted to the special region in
+     * the 06xx range
+     */
+    if ((options & U_SHAPE_PRESERVE_PRESENTATION_MASK)  == U_SHAPE_PRESERVE_PRESENTATION_NOOP) {
+        for (i = 0; i < sourceLength; i++) {
+            char16_t inputChar  = dest[i];
+            if ( (inputChar >= 0xFB50) && (inputChar <= 0xFBFF)) {
+                char16_t c = convertFBto06 [ (inputChar - 0xFB50) ];
+                if (c != 0)
+                    dest[i] = c;
+            } else if ( (inputChar >= 0xFE70) && (inputChar <= 0xFEFC)) {
+                dest[i] = convertFEto06 [ (inputChar - 0xFE70) ] ;
+            } else {
+                dest[i] = inputChar ;
+            }
+        }
+    }
+
+
+    /* sets the index to the end of the buffer, together with the step point to -1 */
+    i = sourceLength - 1;
+    iend = -1;
+    step = -1;
+
+    /*
+     * This function resolves the link between the characters .
+     * Arabic characters have four forms :
+     * Isolated Form, Initial Form, Middle Form and Final Form
+     */
+    currLink = getLink(dest[i]);
+
+    lastPos = i;
+    Nx = -2, Nw = 0;
+
+    while (i != iend) {
+        /* If high byte of currLink > 0 then more than one shape */
+        if ((currLink & 0xFF00) > 0 || (getLink(dest[i]) & IRRELEVANT) != 0) {
+            Nw = i + step;
+            while (Nx < 0) {         /* we need to know about next char */
+                if(Nw == iend) {
+                    nextLink = 0;
+                    Nx = 3000;
+                } else {
+                    nextLink = getLink(dest[Nw]);
+                    if((nextLink & IRRELEVANT) == 0) {
+                        Nx = Nw;
+                    } else {
+                        Nw = Nw + step;
+                    }
+                }
+            }
+
+            if ( ((currLink & ALEFTYPE) > 0)  &&  ((lastLink & LAMTYPE) > 0) ) {
+                lamalef_found = 1;
+                wLamalef = changeLamAlef(dest[i]); /*get from 0x065C-0x065f */
+                if ( wLamalef != 0) {
+                    dest[i] = LAMALEF_SPACE_SUB;            /* The default case is to drop the Alef and replace */
+                    dest[lastPos] =wLamalef;     /* it by LAMALEF_SPACE_SUB which is the last character in the  */
+                    i=lastPos;                   /* unicode private use area, this is done to make   */
+                }                                /* sure that removeLamAlefSpaces() handles only the */
+                lastLink = prevLink;             /* spaces generated during lamalef generation.      */
+                currLink = getLink(wLamalef);    /* LAMALEF_SPACE_SUB is added here and is replaced by spaces   */
+            }                                    /* in removeLamAlefSpaces()                         */
+
+            if ((i > 0) && (dest[i-1] == SPACE_CHAR)){
+                if ( isSeenFamilyChar(dest[i])) {
+                    seenfamFound = 1;
+                } else if (dest[i] == YEH_HAMZA_CHAR) {
+                    yehhamzaFound = 1;
+                }
+            }
+            else if(i==0){
+                if ( isSeenFamilyChar(dest[i])){
+                    seenfamFound = 1;
+                } else if (dest[i] == YEH_HAMZA_CHAR) {
+                    yehhamzaFound = 1;
+                }
+            }
+
+            /*
+             * get the proper shape according to link ability of neighbors
+             * and of character; depends on the order of the shapes
+             * (isolated, initial, middle, final) in the compatibility area
+             */
+            Shape = shapeTable[nextLink & (LINKR + LINKL)]
+                              [lastLink & (LINKR + LINKL)]
+                              [currLink & (LINKR + LINKL)];
+
+            if ((currLink & (LINKR+LINKL)) == 1) {
+                Shape &= 1;
+            } else if(isTashkeelChar(dest[i])) {
+                if( (lastLink & LINKL) && (nextLink & LINKR) && (tashkeelFlag == 1) &&
+                     dest[i] != 0x064C && dest[i] != 0x064D )
+                {
+                    Shape = 1;
+                    if( (nextLink&ALEFTYPE) == ALEFTYPE && (lastLink&LAMTYPE) == LAMTYPE ) {
+                        Shape = 0;
+                    }
+                } else if(tashkeelFlag == 2 && dest[i] == SHADDA06_CHAR){
+                    Shape = 1;
+                } else {
+                    Shape = 0;
+                }
+            }
+            if ((dest[i] ^ 0x0600) < 0x100) {
+                if ( isTashkeelChar(dest[i]) ){
+                    if (tashkeelFlag == 2  && dest[i] != SHADDA06_CHAR){
+                        dest[i] = TASHKEEL_SPACE_SUB;
+                        tashkeelFound  = 1;
+                    } else {
+                        /* to ensure the array index is within the range */
+                        U_ASSERT(dest[i] >= 0x064Bu
+                            && dest[i]-0x064Bu < UPRV_LENGTHOF(IrrelevantPos));
+                        dest[i] =  0xFE70 + IrrelevantPos[(dest[i] - 0x064B)] + static_cast(Shape);
+                    }
+                }else if ((currLink & APRESENT) > 0) {
+                    dest[i] = (char16_t)(0xFB50 + (currLink >> 8) + Shape);
+                }else if ((currLink >> 8) > 0 && (currLink & IRRELEVANT) == 0) {
+                    dest[i] = (char16_t)(0xFE70 + (currLink >> 8) + Shape);
+                }
+            }
+        }
+
+        /* move one notch forward */
+        if ((currLink & IRRELEVANT) == 0) {
+            prevLink = lastLink;
+            lastLink = currLink;
+            lastPos = i;
+        }
+
+        i = i + step;
+        if (i == Nx) {
+            currLink = nextLink;
+            Nx = -2;
+        } else if(i != iend) {
+            currLink = getLink(dest[i]);
+        }
+    }
+    destSize = sourceLength;
+    if ( (lamalef_found != 0 ) || (tashkeelFound  != 0) ){
+        destSize = handleGeneratedSpaces(dest,sourceLength,destSize,options,pErrorCode, shapeVars);
+    }
+
+    if ( (seenfamFound != 0) || (yehhamzaFound != 0) ) {
+        destSize = expandCompositChar(dest, sourceLength,destSize,options,pErrorCode, SHAPE_MODE,shapeVars);
+    }
+    return destSize;
+}
+
+/*
+ *Name     : deShapeUnicode
+ *Function : Converts an Arabic Unicode buffer in FExx Range into unshaped
+ *           arabic Unicode buffer in 06xx Range
+ */
+static int32_t
+deShapeUnicode(char16_t *dest, int32_t sourceLength,
+               int32_t destSize,uint32_t options,
+               UErrorCode *pErrorCode, struct uShapeVariables shapeVars) {
+    int32_t i = 0;
+    int32_t lamalef_found = 0;
+    int32_t yehHamzaComposeEnabled = 0;
+    int32_t seenComposeEnabled = 0;
+
+    yehHamzaComposeEnabled = ((options&U_SHAPE_YEHHAMZA_MASK) == U_SHAPE_YEHHAMZA_TWOCELL_NEAR) ? 1 : 0;
+    seenComposeEnabled = ((options&U_SHAPE_SEEN_MASK) == U_SHAPE_SEEN_TWOCELL_NEAR)? 1 : 0;
+
+    /*
+     *This for loop changes the buffer from the Unicode FE range to
+     *the Unicode 06 range
+     */
+
+    for(i = 0; i < sourceLength; i++) {
+        char16_t  inputChar = dest[i];
+        if ( (inputChar >= 0xFB50) && (inputChar <= 0xFBFF)) { /* FBxx Arabic range */
+            char16_t c = convertFBto06 [ (inputChar - 0xFB50) ];
+            if (c != 0)
+                dest[i] = c;
+        } else if( (yehHamzaComposeEnabled == 1) && ((inputChar == HAMZA06_CHAR) || (inputChar == HAMZAFE_CHAR))
+                && (i < (sourceLength - 1)) && isAlefMaksouraChar(dest[i+1] )) {
+                 dest[i] = SPACE_CHAR;
+                 dest[i+1] = YEH_HAMZA_CHAR;
+        } else if ( (seenComposeEnabled == 1) && (isTailChar(inputChar)) && (i< (sourceLength - 1))
+                        && (isSeenTailFamilyChar(dest[i+1])) ) {
+                dest[i] = SPACE_CHAR;
+        } else if (( inputChar >= 0xFE70) && (inputChar <= 0xFEF4 )) { /* FExx Arabic range */
+                dest[i] = convertFEto06 [ (inputChar - 0xFE70) ];
+        } else {
+            dest[i] = inputChar ;
+        }
+
+        if( isLamAlefChar(dest[i]) )
+            lamalef_found = 1;
+    }
+
+   destSize = sourceLength;
+   if (lamalef_found != 0){
+          destSize = expandCompositChar(dest,sourceLength,destSize,options,pErrorCode,DESHAPE_MODE, shapeVars);
+   }
+   return destSize;
+}
+
+/*
+ ****************************************
+ * u_shapeArabic
+ ****************************************
+ */
+
+U_CAPI int32_t U_EXPORT2
+u_shapeArabic(const char16_t *source, int32_t sourceLength,
+              char16_t *dest, int32_t destCapacity,
+              uint32_t options,
+              UErrorCode *pErrorCode) {
+
+    int32_t destLength;
+    struct  uShapeVariables shapeVars = { OLD_TAIL_CHAR,U_SHAPE_LAMALEF_BEGIN,U_SHAPE_LAMALEF_END,U_SHAPE_TASHKEEL_BEGIN,U_SHAPE_TASHKEEL_END,0};
+
+    /* usual error checking */
+    if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+
+    /* make sure that no reserved options values are used; allow dest==nullptr only for preflighting */
+    if( source==nullptr || sourceLength<-1 || (dest==nullptr && destCapacity!=0) || destCapacity<0 ||
+                (((options&U_SHAPE_TASHKEEL_MASK) > 0) &&
+                 ((options&U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED) == U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED) ) ||
+                (((options&U_SHAPE_TASHKEEL_MASK) > 0) &&
+                 ((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_UNSHAPE)) ||
+                (options&U_SHAPE_DIGIT_TYPE_RESERVED)==U_SHAPE_DIGIT_TYPE_RESERVED ||
+                (options&U_SHAPE_DIGITS_MASK)==U_SHAPE_DIGITS_RESERVED ||
+                ((options&U_SHAPE_LAMALEF_MASK) != U_SHAPE_LAMALEF_RESIZE  &&
+                (options&U_SHAPE_AGGREGATE_TASHKEEL_MASK) != 0) ||
+                ((options&U_SHAPE_AGGREGATE_TASHKEEL_MASK) == U_SHAPE_AGGREGATE_TASHKEEL &&
+                (options&U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED) != U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED)
+    )
+    {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    /* Validate  lamalef options */
+    if(((options&U_SHAPE_LAMALEF_MASK) > 0)&&
+              !(((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_BEGIN) ||
+                ((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_END ) ||
+                ((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_RESIZE )||
+                 ((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_AUTO) ||
+                 ((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_NEAR)))
+    {
+         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    /* Validate  Tashkeel options */
+    if(((options&U_SHAPE_TASHKEEL_MASK) > 0)&&
+                   !(((options & U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_BEGIN) ||
+                     ((options & U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_END )
+                    ||((options & U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_RESIZE )||
+                    ((options & U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_REPLACE_BY_TATWEEL)))
+    {
+         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    /* determine the source length */
+    if(sourceLength==-1) {
+        sourceLength=u_strlen(source);
+    }
+    if(sourceLength<=0) {
+        return u_terminateUChars(dest, destCapacity, 0, pErrorCode);
+    }
+
+    /* check that source and destination do not overlap */
+    if( dest!=nullptr &&
+        ((source<=dest && dest0) {
+            int32_t logical_order = (options&U_SHAPE_TEXT_DIRECTION_MASK) == U_SHAPE_TEXT_DIRECTION_LOGICAL;
+            int32_t aggregate_tashkeel =
+                        (options&(U_SHAPE_AGGREGATE_TASHKEEL_MASK+U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED)) ==
+                        (U_SHAPE_AGGREGATE_TASHKEEL+U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED);
+            int step=logical_order?1:-1;
+            int j=logical_order?-1:2*sourceLength;
+            int i=logical_order?-1:sourceLength;
+            int end=logical_order?sourceLength:-1;
+            int aggregation_possible = 1;
+            char16_t prev = 0;
+            char16_t prevLink, currLink = 0;
+            int newSourceLength = 0;
+            tempsource = (char16_t *)uprv_malloc(2*sourceLength*U_SIZEOF_UCHAR);
+            if(tempsource == nullptr) {
+                *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
+                return 0;
+            }
+
+            while ((i+=step) != end) {
+                prevLink = currLink;
+                currLink = getLink(source[i]);
+                if (aggregate_tashkeel && ((prevLink|currLink)&COMBINE) == COMBINE && aggregation_possible) {
+                    aggregation_possible = 0;
+                    tempsource[j] = (prevdestCapacity) {
+            *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
+                if (tempsource != nullptr) uprv_free(tempsource);
+            return outputSize;
+        }
+
+        /*
+         * need a temporary buffer of size max(outputSize, sourceLength)
+         * because at first we copy source->temp
+         */
+        if(sourceLength>outputSize) {
+            outputSize=sourceLength;
+        }
+
+        /* Start of Arabic letter shaping part */
+        if(outputSize<=UPRV_LENGTHOF(buffer)) {
+            outputSize=UPRV_LENGTHOF(buffer);
+            tempbuffer=buffer;
+        } else {
+            tempbuffer = (char16_t *)uprv_malloc(outputSize*U_SIZEOF_UCHAR);
+
+            /*Test for nullptr*/
+            if(tempbuffer == nullptr) {
+                *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
+                if (tempsource != nullptr) uprv_free(tempsource);
+                return 0;
+            }
+        }
+        u_memcpy(tempbuffer, source, sourceLength);
+        if (tempsource != nullptr){
+            uprv_free(tempsource);
+        }
+
+        if(sourceLength 0
+                 && ((options&U_SHAPE_TASHKEEL_MASK) !=U_SHAPE_TASHKEEL_REPLACE_BY_TATWEEL)) {
+                /* Call the shaping function with tashkeel flag == 2 for removal of tashkeel */
+                destLength = shapeUnicode(tempbuffer,sourceLength,destCapacity,options,pErrorCode,2,shapeVars);
+             }else {
+                /* default Call the shaping function with tashkeel flag == 1 */
+                destLength = shapeUnicode(tempbuffer,sourceLength,destCapacity,options,pErrorCode,1,shapeVars);
+
+                /*After shaping text check if user wants to remove tashkeel and replace it with tatweel*/
+                if( (options&U_SHAPE_TASHKEEL_MASK) == U_SHAPE_TASHKEEL_REPLACE_BY_TATWEEL){
+                  destLength = handleTashkeelWithTatweel(tempbuffer,destLength,destCapacity,options,pErrorCode);
+                }
+            }
+            break;
+        case U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED :
+            /* Call the shaping function with tashkeel flag == 0 */
+            destLength = shapeUnicode(tempbuffer,sourceLength,destCapacity,options,pErrorCode,0,shapeVars);
+            break;
+
+        case U_SHAPE_LETTERS_UNSHAPE :
+            /* Call the deshaping function */
+            destLength = deShapeUnicode(tempbuffer,sourceLength,destCapacity,options,pErrorCode,shapeVars);
+            break;
+        default :
+            /* will never occur because of validity checks above */
+            destLength = 0;
+            break;
+        }
+
+        /*
+         * TODO: (markus 2002aug01)
+         * For as long as we always preflight the outputSize above
+         * we should U_ASSERT(outputSize==destLength)
+         * except for the adjustment above before the tempbuffer allocation
+         */
+
+        if((options&U_SHAPE_TEXT_DIRECTION_MASK) == U_SHAPE_TEXT_DIRECTION_LOGICAL) {
+            countSpaces(tempbuffer,destLength,options,&spacesCountl,&spacesCountr);
+            invertBuffer(tempbuffer,destLength,options,spacesCountl,spacesCountr);
+        }
+        u_memcpy(dest, tempbuffer, uprv_min(destLength, destCapacity));
+
+        if(tempbuffer!=buffer) {
+            uprv_free(tempbuffer);
+        }
+
+        if(destLength>destCapacity) {
+            *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
+            return destLength;
+        }
+
+        /* End of Arabic letter shaping part */
+    } else {
+        /*
+         * No letter shaping:
+         * just make sure the destination is large enough and copy the string.
+         */
+        if(destCapacitysize>=20 &&
+        pInfo->isBigEndian==U_IS_BIG_ENDIAN &&
+        pInfo->charsetFamily==U_CHARSET_FAMILY &&
+        pInfo->dataFormat[0]==0x53 &&   /* dataFormat="SPRP" */
+        pInfo->dataFormat[1]==0x50 &&
+        pInfo->dataFormat[2]==0x52 &&
+        pInfo->dataFormat[3]==0x50 &&
+        pInfo->formatVersion[0]==3 &&
+        pInfo->formatVersion[2]==UTRIE_SHIFT &&
+        pInfo->formatVersion[3]==UTRIE_INDEX_SHIFT
+    ) {
+        //uprv_memcpy(formatVersion, pInfo->formatVersion, 4);
+        uprv_memcpy(dataVersion, pInfo->dataVersion, 4);
+        return true;
+    } else {
+        return false;
+    }
+}
+
+static int32_t U_CALLCONV
+getSPrepFoldingOffset(uint32_t data) {
+
+    return (int32_t)data;
+
+}
+
+/* hashes an entry  */
+static int32_t U_CALLCONV 
+hashEntry(const UHashTok parm) {
+    UStringPrepKey *b = (UStringPrepKey *)parm.pointer;
+    UHashTok namekey, pathkey;
+    namekey.pointer = b->name;
+    pathkey.pointer = b->path;
+    uint32_t unsignedHash = static_cast(uhash_hashChars(namekey)) +
+            37u * static_cast(uhash_hashChars(pathkey));
+    return static_cast(unsignedHash);
+}
+
+/* compares two entries */
+static UBool U_CALLCONV 
+compareEntries(const UHashTok p1, const UHashTok p2) {
+    UStringPrepKey *b1 = (UStringPrepKey *)p1.pointer;
+    UStringPrepKey *b2 = (UStringPrepKey *)p2.pointer;
+    UHashTok name1, name2, path1, path2;
+    name1.pointer = b1->name;
+    name2.pointer = b2->name;
+    path1.pointer = b1->path;
+    path2.pointer = b2->path;
+    return ((UBool)(uhash_compareChars(name1, name2) & 
+        uhash_compareChars(path1, path2)));
+}
+
+static void 
+usprep_unload(UStringPrepProfile* data){
+    udata_close(data->sprepData);
+}
+
+static int32_t 
+usprep_internal_flushCache(UBool noRefCount){
+    UStringPrepProfile *profile = nullptr;
+    UStringPrepKey  *key  = nullptr;
+    int32_t pos = UHASH_FIRST;
+    int32_t deletedNum = 0;
+    const UHashElement *e;
+
+    /*
+     * if shared data hasn't even been lazy evaluated yet
+     * return 0
+     */
+    umtx_lock(&usprepMutex);
+    if (SHARED_DATA_HASHTABLE == nullptr) {
+        umtx_unlock(&usprepMutex);
+        return 0;
+    }
+
+    /*creates an enumeration to iterate through every element in the table */
+    while ((e = uhash_nextElement(SHARED_DATA_HASHTABLE, &pos)) != nullptr)
+    {
+        profile = (UStringPrepProfile *) e->value.pointer;
+        key  = (UStringPrepKey *) e->key.pointer;
+
+        if ((noRefCount== false && profile->refCount == 0) || 
+             noRefCount) {
+            deletedNum++;
+            uhash_removeElement(SHARED_DATA_HASHTABLE, e);
+
+            /* unload the data */
+            usprep_unload(profile);
+
+            if(key->name != nullptr) {
+                uprv_free(key->name);
+                key->name=nullptr;
+            }
+            if(key->path != nullptr) {
+                uprv_free(key->path);
+                key->path=nullptr;
+            }
+            uprv_free(profile);
+            uprv_free(key);
+        }
+       
+    }
+    umtx_unlock(&usprepMutex);
+
+    return deletedNum;
+}
+
+/* Works just like ucnv_flushCache() 
+static int32_t 
+usprep_flushCache(){
+    return usprep_internal_flushCache(false);
+}
+*/
+
+static UBool U_CALLCONV usprep_cleanup(){
+    if (SHARED_DATA_HASHTABLE != nullptr) {
+        usprep_internal_flushCache(true);
+        if (SHARED_DATA_HASHTABLE != nullptr && uhash_count(SHARED_DATA_HASHTABLE) == 0) {
+            uhash_close(SHARED_DATA_HASHTABLE);
+            SHARED_DATA_HASHTABLE = nullptr;
+        }
+    }
+    gSharedDataInitOnce.reset();
+    return (SHARED_DATA_HASHTABLE == nullptr);
+}
+U_CDECL_END
+
+
+/** Initializes the cache for resources */
+static void U_CALLCONV
+createCache(UErrorCode &status) {
+    SHARED_DATA_HASHTABLE = uhash_open(hashEntry, compareEntries, nullptr, &status);
+    if (U_FAILURE(status)) {
+        SHARED_DATA_HASHTABLE = nullptr;
+    }
+    ucln_common_registerCleanup(UCLN_COMMON_USPREP, usprep_cleanup);
+}
+
+static void 
+initCache(UErrorCode *status) {
+    umtx_initOnce(gSharedDataInitOnce, &createCache, *status);
+}
+
+static UBool U_CALLCONV
+loadData(UStringPrepProfile* profile, 
+         const char* path, 
+         const char* name, 
+         const char* type, 
+         UErrorCode* errorCode) {
+    /* load Unicode SPREP data from file */    
+    UTrie _sprepTrie={ 0,0,0,0,0,0,0 };
+    UDataMemory *dataMemory;
+    const int32_t *p=nullptr;
+    const uint8_t *pb;
+    UVersionInfo normUnicodeVersion;
+    int32_t normUniVer, sprepUniVer, normCorrVer;
+
+    if(errorCode==nullptr || U_FAILURE(*errorCode)) {
+        return 0;
+    }
+
+    /* open the data outside the mutex block */
+    //TODO: change the path
+    dataMemory=udata_openChoice(path, type, name, isSPrepAcceptable, nullptr, errorCode);
+    if(U_FAILURE(*errorCode)) {
+        return false;
+    }
+
+    p=(const int32_t *)udata_getMemory(dataMemory);
+    pb=(const uint8_t *)(p+_SPREP_INDEX_TOP);
+    utrie_unserialize(&_sprepTrie, pb, p[_SPREP_INDEX_TRIE_SIZE], errorCode);
+    _sprepTrie.getFoldingOffset=getSPrepFoldingOffset;
+
+
+    if(U_FAILURE(*errorCode)) {
+        udata_close(dataMemory);
+        return false;
+    }
+
+    /* in the mutex block, set the data for this process */
+    umtx_lock(&usprepMutex);
+    if(profile->sprepData==nullptr) {
+        profile->sprepData=dataMemory;
+        dataMemory=nullptr;
+        uprv_memcpy(&profile->indexes, p, sizeof(profile->indexes));
+        uprv_memcpy(&profile->sprepTrie, &_sprepTrie, sizeof(UTrie));
+    } else {
+        p=(const int32_t *)udata_getMemory(profile->sprepData);
+    }
+    umtx_unlock(&usprepMutex);
+    /* initialize some variables */
+    profile->mappingData=(uint16_t *)((uint8_t *)(p+_SPREP_INDEX_TOP)+profile->indexes[_SPREP_INDEX_TRIE_SIZE]);
+    
+    u_getUnicodeVersion(normUnicodeVersion);
+    normUniVer = (normUnicodeVersion[0] << 24) + (normUnicodeVersion[1] << 16) + 
+                 (normUnicodeVersion[2] << 8 ) + (normUnicodeVersion[3]);
+    sprepUniVer = (dataVersion[0] << 24) + (dataVersion[1] << 16) + 
+                  (dataVersion[2] << 8 ) + (dataVersion[3]);
+    normCorrVer = profile->indexes[_SPREP_NORM_CORRECTNS_LAST_UNI_VERSION];
+    
+    if(U_FAILURE(*errorCode)){
+        udata_close(dataMemory);
+        return false;
+    }
+    if( normUniVer < sprepUniVer && /* the Unicode version of SPREP file must be less than the Unicode Version of the normalization data */
+        normUniVer < normCorrVer && /* the Unicode version of the NormalizationCorrections.txt file should be less than the Unicode Version of the normalization data */
+        ((profile->indexes[_SPREP_OPTIONS] & _SPREP_NORMALIZATION_ON) > 0) /* normalization turned on*/
+      ){
+        *errorCode = U_INVALID_FORMAT_ERROR;
+        udata_close(dataMemory);
+        return false;
+    }
+    profile->isDataLoaded = true;
+
+    /* if a different thread set it first, then close the extra data */
+    if(dataMemory!=nullptr) {
+        udata_close(dataMemory); /* nullptr if it was set correctly */
+    }
+
+
+    return profile->isDataLoaded;
+}
+
+static UStringPrepProfile* 
+usprep_getProfile(const char* path, 
+                  const char* name,
+                  UErrorCode *status){
+
+    UStringPrepProfile* profile = nullptr;
+
+    initCache(status);
+
+    if(U_FAILURE(*status)){
+        return nullptr;
+    }
+
+    UStringPrepKey stackKey;
+    /* 
+     * const is cast way to save malloc, strcpy and free calls 
+     * we use the passed in pointers for fetching the data from the 
+     * hash table which is safe
+     */
+    stackKey.name = (char*) name;
+    stackKey.path = (char*) path;
+
+    /* fetch the data from the cache */
+    umtx_lock(&usprepMutex);
+    profile = (UStringPrepProfile*) (uhash_get(SHARED_DATA_HASHTABLE,&stackKey));
+    if(profile != nullptr) {
+        profile->refCount++;
+    }
+    umtx_unlock(&usprepMutex);
+    
+    if(profile == nullptr) {
+        /* else load the data and put the data in the cache */
+        LocalMemory newProfile;
+        if(newProfile.allocateInsteadAndReset() == nullptr) {
+            *status = U_MEMORY_ALLOCATION_ERROR;
+            return nullptr;
+        }
+
+        /* load the data */
+        if(!loadData(newProfile.getAlias(), path, name, _SPREP_DATA_TYPE, status) || U_FAILURE(*status) ){
+            return nullptr;
+        }
+
+        /* get the options */
+        newProfile->doNFKC = (UBool)((newProfile->indexes[_SPREP_OPTIONS] & _SPREP_NORMALIZATION_ON) > 0);
+        newProfile->checkBiDi = (UBool)((newProfile->indexes[_SPREP_OPTIONS] & _SPREP_CHECK_BIDI_ON) > 0);
+
+        LocalMemory key;
+        LocalMemory keyName;
+        LocalMemory keyPath;
+        if( key.allocateInsteadAndReset() == nullptr ||
+            keyName.allocateInsteadAndCopy(static_cast(uprv_strlen(name)+1)) == nullptr ||
+            (path != nullptr &&
+             keyPath.allocateInsteadAndCopy(static_cast(uprv_strlen(path)+1)) == nullptr)
+         ) {
+            *status = U_MEMORY_ALLOCATION_ERROR;
+            usprep_unload(newProfile.getAlias());
+            return nullptr;
+        }
+
+        umtx_lock(&usprepMutex);
+        // If another thread already inserted the same key/value, refcount and cleanup our thread data
+        profile = (UStringPrepProfile*) (uhash_get(SHARED_DATA_HASHTABLE,&stackKey));
+        if(profile != nullptr) {
+            profile->refCount++;
+            usprep_unload(newProfile.getAlias());
+        }
+        else {
+            /* initialize the key members */
+            key->name = keyName.orphan();
+            uprv_strcpy(key->name, name);
+            if(path != nullptr){
+                key->path = keyPath.orphan();
+                uprv_strcpy(key->path, path);
+            }        
+            profile = newProfile.orphan();
+    
+            /* add the data object to the cache */
+            profile->refCount = 1;
+            uhash_put(SHARED_DATA_HASHTABLE, key.orphan(), profile, status);
+        }
+        umtx_unlock(&usprepMutex);
+    }
+
+    return profile;
+}
+
+U_CAPI UStringPrepProfile* U_EXPORT2
+usprep_open(const char* path, 
+            const char* name,
+            UErrorCode* status){
+
+    if(status == nullptr || U_FAILURE(*status)){
+        return nullptr;
+    }
+       
+    /* initialize the profile struct members */
+    return usprep_getProfile(path,name,status);
+}
+
+U_CAPI UStringPrepProfile* U_EXPORT2
+usprep_openByType(UStringPrepProfileType type,
+				  UErrorCode* status) {
+    if(status == nullptr || U_FAILURE(*status)){
+        return nullptr;
+    }
+    int32_t index = (int32_t)type;
+    if (index < 0 || index >= UPRV_LENGTHOF(PROFILE_NAMES)) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+    return usprep_open(nullptr, PROFILE_NAMES[index], status);
+}
+
+U_CAPI void U_EXPORT2
+usprep_close(UStringPrepProfile* profile){
+    if(profile==nullptr){
+        return;
+    }
+
+    umtx_lock(&usprepMutex);
+    /* decrement the ref count*/
+    if(profile->refCount > 0){
+        profile->refCount--;
+    }
+    umtx_unlock(&usprepMutex);
+    
+}
+
+U_CFUNC void 
+uprv_syntaxError(const char16_t* rules,
+                 int32_t pos,
+                 int32_t rulesLen,
+                 UParseError* parseError){
+    if(parseError == nullptr){
+        return;
+    }
+    parseError->offset = pos;
+    parseError->line = 0 ; // we are not using line numbers 
+    
+    // for pre-context
+    int32_t start = (pos < U_PARSE_CONTEXT_LEN)? 0 : (pos - (U_PARSE_CONTEXT_LEN-1));
+    int32_t limit = pos;
+    
+    u_memcpy(parseError->preContext,rules+start,limit-start);
+    //null terminate the buffer
+    parseError->preContext[limit-start] = 0;
+    
+    // for post-context; include error rules[pos]
+    start = pos;
+    limit = start + (U_PARSE_CONTEXT_LEN-1);
+    if (limit > rulesLen) {
+        limit = rulesLen;
+    }
+    if (start < rulesLen) {
+        u_memcpy(parseError->postContext,rules+start,limit-start);
+    }
+    //null terminate the buffer
+    parseError->postContext[limit-start]= 0;
+}
+
+
+static inline UStringPrepType
+getValues(uint16_t trieWord, int16_t& value, UBool& isIndex){
+
+    UStringPrepType type;
+    if(trieWord == 0){
+        /* 
+         * Initial value stored in the mapping table 
+         * just return USPREP_TYPE_LIMIT .. so that
+         * the source codepoint is copied to the destination
+         */
+        type = USPREP_TYPE_LIMIT;
+        isIndex =false;
+        value = 0;
+    }else if(trieWord >= _SPREP_TYPE_THRESHOLD){
+        type = (UStringPrepType) (trieWord - _SPREP_TYPE_THRESHOLD);
+        isIndex =false;
+        value = 0;
+    }else{
+        /* get the type */
+        type = USPREP_MAP;
+        /* ascertain if the value is index or delta */
+        if(trieWord & 0x02){
+            isIndex = true;
+            value = trieWord  >> 2; //mask off the lower 2 bits and shift
+        }else{
+            isIndex = false;
+            value = (int16_t)trieWord;
+            value =  (value >> 2);
+        }
+ 
+        if((trieWord>>2) == _SPREP_MAX_INDEX_VALUE){
+            type = USPREP_DELETE;
+            isIndex =false;
+            value = 0;
+        }
+    }
+    return type;
+}
+
+// TODO: change to writing to UnicodeString not char16_t *
+static int32_t 
+usprep_map(  const UStringPrepProfile* profile, 
+             const char16_t* src, int32_t srcLength,
+             char16_t* dest, int32_t destCapacity,
+             int32_t options,
+             UParseError* parseError,
+             UErrorCode* status ){
+    
+    uint16_t result;
+    int32_t destIndex=0;
+    int32_t srcIndex;
+    UBool allowUnassigned = (UBool) ((options & USPREP_ALLOW_UNASSIGNED)>0);
+    UStringPrepType type;
+    int16_t value;
+    UBool isIndex;
+    const int32_t* indexes = profile->indexes;
+
+    // no error checking the caller check for error and arguments
+    // no string length check the caller finds out the string length
+
+    for(srcIndex=0;srcIndexsprepTrie,ch,result);
+        
+        type = getValues(result, value, isIndex);
+
+        // check if the source codepoint is unassigned
+        if(type == USPREP_UNASSIGNED && allowUnassigned == false){
+
+            uprv_syntaxError(src,srcIndex-U16_LENGTH(ch), srcLength,parseError);
+            *status = U_STRINGPREP_UNASSIGNED_ERROR;
+            return 0;
+            
+        }else if(type == USPREP_MAP){
+            
+            int32_t index, length;
+
+            if(isIndex){
+                index = value;
+                if(index >= indexes[_SPREP_ONE_UCHAR_MAPPING_INDEX_START] &&
+                         index < indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START]){
+                    length = 1;
+                }else if(index >= indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START] &&
+                         index < indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START]){
+                    length = 2;
+                }else if(index >= indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START] &&
+                         index < indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START]){
+                    length = 3;
+                }else{
+                    length = profile->mappingData[index++];
+         
+                }
+
+                /* copy mapping to destination */
+                for(int32_t i=0; i< length; i++){
+                    if(destIndex < destCapacity  ){
+                        dest[destIndex] = profile->mappingData[index+i];
+                    }
+                    destIndex++; /* for pre-flighting */
+                }  
+                continue;
+            }else{
+                // subtract the delta to arrive at the code point
+                ch -= value;
+            }
+
+        }else if(type==USPREP_DELETE){
+             // just consume the codepoint and continue
+            continue;
+        }
+        //copy the code point into destination
+        if(ch <= 0xFFFF){
+            if(destIndex < destCapacity ){
+                dest[destIndex] = (char16_t)ch;
+            }
+            destIndex++;
+        }else{
+            if(destIndex+1 < destCapacity ){
+                dest[destIndex]   = U16_LEAD(ch);
+                dest[destIndex+1] = U16_TRAIL(ch);
+            }
+            destIndex +=2;
+        }
+       
+    }
+        
+    return u_terminateUChars(dest, destCapacity, destIndex, status);
+}
+
+/*
+   1) Map -- For each character in the input, check if it has a mapping
+      and, if so, replace it with its mapping.  
+
+   2) Normalize -- Possibly normalize the result of step 1 using Unicode
+      normalization. 
+
+   3) Prohibit -- Check for any characters that are not allowed in the
+      output.  If any are found, return an error.  
+
+   4) Check bidi -- Possibly check for right-to-left characters, and if
+      any are found, make sure that the whole string satisfies the
+      requirements for bidirectional strings.  If the string does not
+      satisfy the requirements for bidirectional strings, return an
+      error.  
+      [Unicode3.2] defines several bidirectional categories; each character
+       has one bidirectional category assigned to it.  For the purposes of
+       the requirements below, an "RandALCat character" is a character that
+       has Unicode bidirectional categories "R" or "AL"; an "LCat character"
+       is a character that has Unicode bidirectional category "L".  Note
+
+
+       that there are many characters which fall in neither of the above
+       definitions; Latin digits ( through ) are examples of
+       this because they have bidirectional category "EN".
+
+       In any profile that specifies bidirectional character handling, all
+       three of the following requirements MUST be met:
+
+       1) The characters in section 5.8 MUST be prohibited.
+
+       2) If a string contains any RandALCat character, the string MUST NOT
+          contain any LCat character.
+
+       3) If a string contains any RandALCat character, a RandALCat
+          character MUST be the first character of the string, and a
+          RandALCat character MUST be the last character of the string.
+*/
+U_CAPI int32_t U_EXPORT2
+usprep_prepare(   const UStringPrepProfile* profile,
+                  const char16_t* src, int32_t srcLength,
+                  char16_t* dest, int32_t destCapacity,
+                  int32_t options,
+                  UParseError* parseError,
+                  UErrorCode* status ){
+
+    // check error status
+    if(U_FAILURE(*status)){
+        return 0;
+    }
+
+    //check arguments
+    if(profile==nullptr ||
+            (src==nullptr ? srcLength!=0 : srcLength<-1) ||
+            (dest==nullptr ? destCapacity!=0 : destCapacity<0)) {
+        *status=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    //get the string length
+    if(srcLength < 0){
+        srcLength = u_strlen(src);
+    }
+    // map
+    UnicodeString s1;
+    char16_t *b1 = s1.getBuffer(srcLength);
+    if(b1==nullptr){
+        *status = U_MEMORY_ALLOCATION_ERROR;
+        return 0;
+    }
+    int32_t b1Len = usprep_map(profile, src, srcLength,
+                               b1, s1.getCapacity(), options, parseError, status);
+    s1.releaseBuffer(U_SUCCESS(*status) ? b1Len : 0);
+
+    if(*status == U_BUFFER_OVERFLOW_ERROR){
+        // redo processing of string
+        /* we do not have enough room so grow the buffer*/
+        b1 = s1.getBuffer(b1Len);
+        if(b1==nullptr){
+            *status = U_MEMORY_ALLOCATION_ERROR;
+            return 0;
+        }
+
+        *status = U_ZERO_ERROR; // reset error
+        b1Len = usprep_map(profile, src, srcLength,
+                           b1, s1.getCapacity(), options, parseError, status);
+        s1.releaseBuffer(U_SUCCESS(*status) ? b1Len : 0);
+    }
+    if(U_FAILURE(*status)){
+        return 0;
+    }
+
+    // normalize
+    UnicodeString s2;
+    if(profile->doNFKC){
+        const Normalizer2 *n2 = Normalizer2::getNFKCInstance(*status);
+        FilteredNormalizer2 fn2(*n2, *uniset_getUnicode32Instance(*status));
+        if(U_FAILURE(*status)){
+            return 0;
+        }
+        fn2.normalize(s1, s2, *status);
+    }else{
+        s2.fastCopyFrom(s1);
+    }
+    if(U_FAILURE(*status)){
+        return 0;
+    }
+
+    // Prohibit and checkBiDi in one pass
+    const char16_t *b2 = s2.getBuffer();
+    int32_t b2Len = s2.length();
+    UCharDirection direction=U_CHAR_DIRECTION_COUNT, firstCharDir=U_CHAR_DIRECTION_COUNT;
+    UBool leftToRight=false, rightToLeft=false;
+    int32_t rtlPos =-1, ltrPos =-1;
+
+    for(int32_t b2Index=0; b2IndexsprepTrie,ch,result);
+
+        int16_t value;
+        UBool isIndex;
+        UStringPrepType type = getValues(result, value, isIndex);
+
+        if( type == USPREP_PROHIBITED || 
+            ((result < _SPREP_TYPE_THRESHOLD) && (result & 0x01) /* first bit says it the code point is prohibited*/)
+           ){
+            *status = U_STRINGPREP_PROHIBITED_ERROR;
+            uprv_syntaxError(b2, b2Index-U16_LENGTH(ch), b2Len, parseError);
+            return 0;
+        }
+
+        if(profile->checkBiDi) {
+            direction = ubidi_getClass(ch);
+            if(firstCharDir == U_CHAR_DIRECTION_COUNT){
+                firstCharDir = direction;
+            }
+            if(direction == U_LEFT_TO_RIGHT){
+                leftToRight = true;
+                ltrPos = b2Index-1;
+            }
+            if(direction == U_RIGHT_TO_LEFT || direction == U_RIGHT_TO_LEFT_ARABIC){
+                rightToLeft = true;
+                rtlPos = b2Index-1;
+            }
+        }
+    }
+    if(profile->checkBiDi){
+        // satisfy 2
+        if( leftToRight && rightToLeft){
+            *status = U_STRINGPREP_CHECK_BIDI_ERROR;
+            uprv_syntaxError(b2,(rtlPos>ltrPos) ? rtlPos : ltrPos, b2Len, parseError);
+            return 0;
+        }
+
+        //satisfy 3
+        if( rightToLeft && 
+            !((firstCharDir == U_RIGHT_TO_LEFT || firstCharDir == U_RIGHT_TO_LEFT_ARABIC) &&
+              (direction == U_RIGHT_TO_LEFT || direction == U_RIGHT_TO_LEFT_ARABIC))
+           ){
+            *status = U_STRINGPREP_CHECK_BIDI_ERROR;
+            uprv_syntaxError(b2, rtlPos, b2Len, parseError);
+            return false;
+        }
+    }
+    return s2.extract(dest, destCapacity, *status);
+}
+
+
+/* data swapping ------------------------------------------------------------ */
+
+U_CAPI int32_t U_EXPORT2
+usprep_swap(const UDataSwapper *ds,
+            const void *inData, int32_t length, void *outData,
+            UErrorCode *pErrorCode) {
+    const UDataInfo *pInfo;
+    int32_t headerSize;
+
+    const uint8_t *inBytes;
+    uint8_t *outBytes;
+
+    const int32_t *inIndexes;
+    int32_t indexes[16];
+
+    int32_t i, offset, count, size;
+
+    /* udata_swapDataHeader checks the arguments */
+    headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
+    if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+
+    /* check data format and format version */
+    pInfo=(const UDataInfo *)((const char *)inData+4);
+    if(!(
+        pInfo->dataFormat[0]==0x53 &&   /* dataFormat="SPRP" */
+        pInfo->dataFormat[1]==0x50 &&
+        pInfo->dataFormat[2]==0x52 &&
+        pInfo->dataFormat[3]==0x50 &&
+        pInfo->formatVersion[0]==3
+    )) {
+        udata_printError(ds, "usprep_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized as StringPrep .spp data\n",
+                         pInfo->dataFormat[0], pInfo->dataFormat[1],
+                         pInfo->dataFormat[2], pInfo->dataFormat[3],
+                         pInfo->formatVersion[0]);
+        *pErrorCode=U_UNSUPPORTED_ERROR;
+        return 0;
+    }
+
+    inBytes=(const uint8_t *)inData+headerSize;
+    outBytes= (outData == nullptr ) ? nullptr : (uint8_t *)outData+headerSize;
+
+    inIndexes=(const int32_t *)inBytes;
+
+    if(length>=0) {
+        length-=headerSize;
+        if(length<16*4) {
+            udata_printError(ds, "usprep_swap(): too few bytes (%d after header) for StringPrep .spp data\n",
+                             length);
+            *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+            return 0;
+        }
+    }
+
+    /* read the first 16 indexes (ICU 2.8/format version 3: _SPREP_INDEX_TOP==16, might grow) */
+    for(i=0; i<16; ++i) {
+        indexes[i]=udata_readInt32(ds, inIndexes[i]);
+    }
+
+    /* calculate the total length of the data */
+    size=
+        16*4+ /* size of indexes[] */
+        indexes[_SPREP_INDEX_TRIE_SIZE]+
+        indexes[_SPREP_INDEX_MAPPING_DATA_SIZE];
+
+    if(length>=0) {
+        if(lengthswapArray32(ds, inBytes, count, outBytes, pErrorCode);
+        offset+=count;
+
+        /* swap the UTrie */
+        count=indexes[_SPREP_INDEX_TRIE_SIZE];
+        utrie_swap(ds, inBytes+offset, count, outBytes+offset, pErrorCode);
+        offset+=count;
+
+        /* swap the uint16_t mappingTable[] */
+        count=indexes[_SPREP_INDEX_MAPPING_DATA_SIZE];
+        ds->swapArray16(ds, inBytes+offset, count, outBytes+offset, pErrorCode);
+        //offset+=count;
+    }
+
+    return headerSize+size;
+}
+
+#endif /* #if !UCONFIG_NO_IDNA */
diff --git a/intl/icu/source/common/ustack.cpp b/intl/icu/source/common/ustack.cpp
new file mode 100644
index 0000000000..5b15efcd7c
--- /dev/null
+++ b/intl/icu/source/common/ustack.cpp
@@ -0,0 +1,62 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+*   Copyright (C) 2003-2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*/
+
+#include "uvector.h"
+
+U_NAMESPACE_BEGIN
+
+UOBJECT_DEFINE_RTTI_IMPLEMENTATION(UStack)
+
+UStack::UStack(UErrorCode &status) :
+    UVector(status)
+{
+}
+
+UStack::UStack(int32_t initialCapacity, UErrorCode &status) :
+    UVector(initialCapacity, status)
+{
+}
+
+UStack::UStack(UObjectDeleter *d, UElementsAreEqual *c, UErrorCode &status) :
+    UVector(d, c, status)
+{
+}
+
+UStack::UStack(UObjectDeleter *d, UElementsAreEqual *c, int32_t initialCapacity, UErrorCode &status) :
+    UVector(d, c, initialCapacity, status)
+{
+}
+
+UStack::~UStack() {}
+
+void* UStack::pop() {
+    int32_t n = size() - 1;
+    void* result = nullptr;
+    if (n >= 0) {
+        result = orphanElementAt(n);
+    }
+    return result;
+}
+
+int32_t UStack::popi() {
+    int32_t n = size() - 1;
+    int32_t result = 0;
+    if (n >= 0) {
+        result = elementAti(n);
+        removeElementAt(n);
+    }
+    return result;
+}
+
+int32_t UStack::search(void* obj) const {
+    int32_t i = indexOf(obj);
+    return (i >= 0) ? size() - i : i;
+}
+
+U_NAMESPACE_END
diff --git a/intl/icu/source/common/ustr_cnv.cpp b/intl/icu/source/common/ustr_cnv.cpp
new file mode 100644
index 0000000000..abcc12f683
--- /dev/null
+++ b/intl/icu/source/common/ustr_cnv.cpp
@@ -0,0 +1,256 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 1998-2014, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  ustr_cnv.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2004aug24
+*   created by: Markus W. Scherer
+*
+*   Character conversion functions moved here from ustring.c
+*/
+
+#include "unicode/utypes.h"
+
+#if !UCONFIG_NO_CONVERSION
+
+#include "unicode/ustring.h"
+#include "unicode/ucnv.h"
+#include "cstring.h"
+#include "cmemory.h"
+#include "umutex.h"
+#include "ustr_cnv.h"
+#include "ucnv_bld.h"
+
+/* mutexed access to a shared default converter ----------------------------- */
+
+static UConverter *gDefaultConverter = nullptr;
+
+U_CAPI UConverter* U_EXPORT2
+u_getDefaultConverter(UErrorCode *status)
+{
+    UConverter *converter = nullptr;
+    
+    if (gDefaultConverter != nullptr) {
+        icu::umtx_lock(nullptr);
+        
+        /* need to check to make sure it wasn't taken out from under us */
+        if (gDefaultConverter != nullptr) {
+            converter = gDefaultConverter;
+            gDefaultConverter = nullptr;
+        }
+        icu::umtx_unlock(nullptr);
+    }
+
+    /* if the cache was empty, create a converter */
+    if(converter == nullptr) {
+        converter = ucnv_open(nullptr, status);
+        if(U_FAILURE(*status)) {
+            ucnv_close(converter);
+            converter = nullptr;
+        }
+    }
+
+    return converter;
+}
+
+U_CAPI void U_EXPORT2
+u_releaseDefaultConverter(UConverter *converter)
+{
+    if(gDefaultConverter == nullptr) {
+        if (converter != nullptr) {
+            ucnv_reset(converter);
+        }
+        ucnv_enableCleanup();
+        icu::umtx_lock(nullptr);
+        if(gDefaultConverter == nullptr) {
+            gDefaultConverter = converter;
+            converter = nullptr;
+        }
+        icu::umtx_unlock(nullptr);
+    }
+
+    if(converter != nullptr) {
+        ucnv_close(converter);
+    }
+}
+
+U_CAPI void U_EXPORT2
+u_flushDefaultConverter()
+{
+    UConverter *converter = nullptr;
+    
+    if (gDefaultConverter != nullptr) {
+        icu::umtx_lock(nullptr);
+        
+        /* need to check to make sure it wasn't taken out from under us */
+        if (gDefaultConverter != nullptr) {
+            converter = gDefaultConverter;
+            gDefaultConverter = nullptr;
+        }
+        icu::umtx_unlock(nullptr);
+    }
+
+    /* if the cache was populated, flush it */
+    if(converter != nullptr) {
+         ucnv_close(converter);
+    }
+}
+
+
+/* conversions between char* and char16_t* ------------------------------------- */
+
+/* maximum string length for u_uastrcpy() and u_austrcpy() implementations */
+#define MAX_STRLEN 0x0FFFFFFF
+
+/*
+ returns the minimum of (the length of the null-terminated string) and n.
+*/
+static int32_t u_astrnlen(const char *s1, int32_t n)
+{
+    int32_t len = 0;
+
+    if (s1)
+    {
+        while (n-- && *(s1++))
+        {
+            len++;
+        }
+    }
+    return len;
+}
+
+U_CAPI char16_t*  U_EXPORT2
+u_uastrncpy(char16_t *ucs1,
+           const char *s2,
+           int32_t n)
+{
+  char16_t *target = ucs1;
+  UErrorCode err = U_ZERO_ERROR;
+  UConverter *cnv = u_getDefaultConverter(&err);
+  if(U_SUCCESS(err) && cnv != nullptr) {
+    ucnv_reset(cnv);
+    ucnv_toUnicode(cnv,
+                   &target,
+                   ucs1+n,
+                   &s2,
+                   s2+u_astrnlen(s2, n),
+                   nullptr,
+                   true,
+                   &err);
+    ucnv_reset(cnv); /* be good citizens */
+    u_releaseDefaultConverter(cnv);
+    if(U_FAILURE(err) && (err != U_BUFFER_OVERFLOW_ERROR) ) {
+      *ucs1 = 0; /* failure */
+    }
+    if(target < (ucs1+n)) { /* U_BUFFER_OVERFLOW_ERROR isn't an err, just means no termination will happen. */
+      *target = 0;  /* terminate */
+    }
+  } else {
+    *ucs1 = 0;
+  }
+  return ucs1;
+}
+
+U_CAPI char16_t*  U_EXPORT2
+u_uastrcpy(char16_t *ucs1,
+          const char *s2 )
+{
+  UErrorCode err = U_ZERO_ERROR;
+  UConverter *cnv = u_getDefaultConverter(&err);
+  if(U_SUCCESS(err) && cnv != nullptr) {
+    ucnv_toUChars(cnv,
+                    ucs1,
+                    MAX_STRLEN,
+                    s2,
+                    (int32_t)uprv_strlen(s2),
+                    &err);
+    u_releaseDefaultConverter(cnv);
+    if(U_FAILURE(err)) {
+      *ucs1 = 0;
+    }
+  } else {
+    *ucs1 = 0;
+  }
+  return ucs1;
+}
+
+/*
+ returns the minimum of (the length of the null-terminated string) and n.
+*/
+static int32_t u_ustrnlen(const char16_t *ucs1, int32_t n)
+{
+    int32_t len = 0;
+
+    if (ucs1)
+    {
+        while (n-- && *(ucs1++))
+        {
+            len++;
+        }
+    }
+    return len;
+}
+
+U_CAPI char*  U_EXPORT2
+u_austrncpy(char *s1,
+        const char16_t *ucs2,
+        int32_t n)
+{
+  char *target = s1;
+  UErrorCode err = U_ZERO_ERROR;
+  UConverter *cnv = u_getDefaultConverter(&err);
+  if(U_SUCCESS(err) && cnv != nullptr) {
+    ucnv_reset(cnv);
+    ucnv_fromUnicode(cnv,
+                  &target,
+                  s1+n,
+                  &ucs2,
+                  ucs2+u_ustrnlen(ucs2, n),
+                  nullptr,
+                  true,
+                  &err);
+    ucnv_reset(cnv); /* be good citizens */
+    u_releaseDefaultConverter(cnv);
+    if(U_FAILURE(err) && (err != U_BUFFER_OVERFLOW_ERROR) ) {
+      *s1 = 0; /* failure */
+    }
+    if(target < (s1+n)) { /* U_BUFFER_OVERFLOW_ERROR isn't an err, just means no termination will happen. */
+      *target = 0;  /* terminate */
+    }
+  } else {
+    *s1 = 0;
+  }
+  return s1;
+}
+
+U_CAPI char*  U_EXPORT2
+u_austrcpy(char *s1,
+         const char16_t *ucs2 )
+{
+  UErrorCode err = U_ZERO_ERROR;
+  UConverter *cnv = u_getDefaultConverter(&err);
+  if(U_SUCCESS(err) && cnv != nullptr) {
+    int32_t len = ucnv_fromUChars(cnv,
+                  s1,
+                  MAX_STRLEN,
+                  ucs2,
+                  -1,
+                  &err);
+    u_releaseDefaultConverter(cnv);
+    s1[len] = 0;
+  } else {
+    *s1 = 0;
+  }
+  return s1;
+}
+
+#endif
diff --git a/intl/icu/source/common/ustr_cnv.h b/intl/icu/source/common/ustr_cnv.h
new file mode 100644
index 0000000000..861e3ebff0
--- /dev/null
+++ b/intl/icu/source/common/ustr_cnv.h
@@ -0,0 +1,51 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*  
+**********************************************************************
+*   Copyright (C) 1999-2010, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*   file name:  ustr_cnv.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2004Aug27
+*   created by: George Rhoten
+*/
+
+#ifndef USTR_CNV_IMP_H
+#define USTR_CNV_IMP_H
+
+#include "unicode/utypes.h"
+#include "unicode/ucnv.h"
+
+#if !UCONFIG_NO_CONVERSION
+
+/**
+ * Get the default converter. This is a commonly used converter
+ * that is used for the ustring and UnicodeString API.
+ * Remember to use the u_releaseDefaultConverter when you are done.
+ * @internal
+ */
+U_CAPI UConverter* U_EXPORT2
+u_getDefaultConverter(UErrorCode *status);
+
+
+/**
+ * Release the default converter to the converter cache.
+ * @internal
+ */
+U_CAPI void U_EXPORT2
+u_releaseDefaultConverter(UConverter *converter);
+
+/**
+ * Flush the default converter, if cached. 
+ * @internal
+ */
+U_CAPI void U_EXPORT2
+u_flushDefaultConverter(void);
+
+#endif
+
+#endif
diff --git a/intl/icu/source/common/ustr_imp.h b/intl/icu/source/common/ustr_imp.h
new file mode 100644
index 0000000000..3c4b9cc2a5
--- /dev/null
+++ b/intl/icu/source/common/ustr_imp.h
@@ -0,0 +1,155 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*  
+**********************************************************************
+*   Copyright (C) 1999-2015, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*   file name:  ustr_imp.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2001jan30
+*   created by: Markus W. Scherer
+*/
+
+#ifndef __USTR_IMP_H__
+#define __USTR_IMP_H__
+
+#include "unicode/utypes.h"
+#include "unicode/utf8.h"
+
+/**
+ * Internal option for unorm_cmpEquivFold() for strncmp style.
+ * If set, checks for both string length and terminating NUL.
+ */
+#define _STRNCMP_STYLE 0x1000
+
+/**
+ * Compare two strings in code point order or code unit order.
+ * Works in strcmp style (both lengths -1),
+ * strncmp style (lengths equal and >=0, flag true),
+ * and memcmp/UnicodeString style (at least one length >=0).
+ */
+U_CFUNC int32_t U_EXPORT2
+uprv_strCompare(const UChar *s1, int32_t length1,
+                const UChar *s2, int32_t length2,
+                UBool strncmpStyle, UBool codePointOrder);
+
+U_CAPI int32_t U_EXPORT2 
+ustr_hashUCharsN(const UChar *str, int32_t length);
+
+U_CAPI int32_t U_EXPORT2 
+ustr_hashCharsN(const char *str, int32_t length);
+
+U_CAPI int32_t U_EXPORT2
+ustr_hashICharsN(const char *str, int32_t length);
+
+/**
+ * Convert an ASCII-range lowercase character to uppercase.
+ * 
+ * @param c A UChar.
+ * @return If UChar is a lowercase ASCII character, returns the uppercase version.
+ *         Otherwise, returns the input character.
+ */
+U_CAPI UChar U_EXPORT2
+u_asciiToUpper(UChar c);
+
+// TODO: Add u_asciiToLower if/when there is a need for it.
+
+/**
+ * NUL-terminate a UChar * string if possible.
+ * If length  < destCapacity then NUL-terminate.
+ * If length == destCapacity then do not terminate but set U_STRING_NOT_TERMINATED_WARNING.
+ * If length  > destCapacity then do not terminate but set U_BUFFER_OVERFLOW_ERROR.
+ *
+ * @param dest Destination buffer, can be NULL if destCapacity==0.
+ * @param destCapacity Number of UChars available at dest.
+ * @param length Number of UChars that were (to be) written to dest.
+ * @param pErrorCode ICU error code.
+ * @return length
+ */
+U_CAPI int32_t U_EXPORT2
+u_terminateUChars(UChar *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode);
+
+/**
+ * NUL-terminate a char * string if possible.
+ * Same as u_terminateUChars() but for a different string type.
+ */
+U_CAPI int32_t U_EXPORT2
+u_terminateChars(char *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode);
+
+/**
+ * NUL-terminate a UChar32 * string if possible.
+ * Same as u_terminateUChars() but for a different string type.
+ */
+U_CAPI int32_t U_EXPORT2
+u_terminateUChar32s(UChar32 *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode);
+
+/**
+ * NUL-terminate a wchar_t * string if possible.
+ * Same as u_terminateUChars() but for a different string type.
+ */
+U_CAPI int32_t U_EXPORT2
+u_terminateWChars(wchar_t *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode);
+
+/**
+ * Counts the bytes of any whole valid sequence for a UTF-8 lead byte.
+ * Returns 1 for ASCII 0..0x7f.
+ * Returns 0 for 0x80..0xc1 as well as for 0xf5..0xff.
+ * leadByte might be evaluated multiple times.
+ *
+ * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff.
+ * @return 0..4
+ */
+#define U8_COUNT_BYTES(leadByte) \
+    (U8_IS_SINGLE(leadByte) ? 1 : U8_COUNT_BYTES_NON_ASCII(leadByte))
+
+/**
+ * Counts the bytes of any whole valid sequence for a UTF-8 lead byte.
+ * Returns 0 for 0x00..0xc1 as well as for 0xf5..0xff.
+ * leadByte might be evaluated multiple times.
+ *
+ * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff.
+ * @return 0 or 2..4
+ */
+#define U8_COUNT_BYTES_NON_ASCII(leadByte) \
+    (U8_IS_LEAD(leadByte) ? ((uint8_t)(leadByte)>=0xe0)+((uint8_t)(leadByte)>=0xf0)+2 : 0)
+
+#ifdef __cplusplus
+
+U_NAMESPACE_BEGIN
+
+class UTF8 {
+public:
+    UTF8() = delete;  // all static
+
+    /**
+     * Is t a valid UTF-8 trail byte?
+     *
+     * @param prev Must be the preceding lead byte if i==1 and length>=3;
+     *             otherwise ignored.
+     * @param t The i-th byte following the lead byte.
+     * @param i The index (1..3) of byte t in the byte sequence. 0 1) {
+            return U8_IS_TRAIL(t);
+        } else if (length == 3) {
+            return U8_IS_VALID_LEAD3_AND_T1(prev, t);
+        } else {  // length == 4
+            return U8_IS_VALID_LEAD4_AND_T1(prev, t);
+        }
+    }
+};
+
+U_NAMESPACE_END
+
+#endif  // __cplusplus
+
+#endif
diff --git a/intl/icu/source/common/ustr_titlecase_brkiter.cpp b/intl/icu/source/common/ustr_titlecase_brkiter.cpp
new file mode 100644
index 0000000000..82beaca65b
--- /dev/null
+++ b/intl/icu/source/common/ustr_titlecase_brkiter.cpp
@@ -0,0 +1,237 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*   Copyright (C) 2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*******************************************************************************
+*   file name:  ustr_titlecase_brkiter.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2011may30
+*   created by: Markus W. Scherer
+*
+*   Titlecasing functions that are based on BreakIterator
+*   were moved here to break dependency cycles among parts of the common library.
+*/
+
+#include "unicode/utypes.h"
+
+#if !UCONFIG_NO_BREAK_ITERATION
+
+#include "unicode/brkiter.h"
+#include "unicode/casemap.h"
+#include "unicode/chariter.h"
+#include "unicode/localpointer.h"
+#include "unicode/ubrk.h"
+#include "unicode/ucasemap.h"
+#include "unicode/utext.h"
+#include "cmemory.h"
+#include "uassert.h"
+#include "ucase.h"
+#include "ucasemap_imp.h"
+
+U_NAMESPACE_BEGIN
+
+/**
+ * Whole-string BreakIterator.
+ * Titlecasing only calls setText(), first(), and next().
+ * We implement the rest only to satisfy the abstract interface.
+ */
+class WholeStringBreakIterator : public BreakIterator {
+public:
+    WholeStringBreakIterator() : BreakIterator(), length(0) {}
+    ~WholeStringBreakIterator() override;
+    bool operator==(const BreakIterator&) const override;
+    WholeStringBreakIterator *clone() const override;
+    static UClassID U_EXPORT2 getStaticClassID();
+    UClassID getDynamicClassID() const override;
+    CharacterIterator &getText() const override;
+    UText *getUText(UText *fillIn, UErrorCode &errorCode) const override;
+    void  setText(const UnicodeString &text) override;
+    void  setText(UText *text, UErrorCode &errorCode) override;
+    void  adoptText(CharacterIterator* it) override;
+    int32_t first() override;
+    int32_t last() override;
+    int32_t previous() override;
+    int32_t next() override;
+    int32_t current() const override;
+    int32_t following(int32_t offset) override;
+    int32_t preceding(int32_t offset) override;
+    UBool isBoundary(int32_t offset) override;
+    int32_t next(int32_t n) override;
+    WholeStringBreakIterator *createBufferClone(void *stackBuffer, int32_t &BufferSize,
+                                                UErrorCode &errorCode) override;
+    WholeStringBreakIterator &refreshInputText(UText *input, UErrorCode &errorCode) override;
+
+private:
+    int32_t length;
+};
+
+UOBJECT_DEFINE_RTTI_IMPLEMENTATION(WholeStringBreakIterator)
+
+WholeStringBreakIterator::~WholeStringBreakIterator() {}
+bool WholeStringBreakIterator::operator==(const BreakIterator&) const { return false; }
+WholeStringBreakIterator *WholeStringBreakIterator::clone() const { return nullptr; }
+
+CharacterIterator &WholeStringBreakIterator::getText() const {
+    UPRV_UNREACHABLE_EXIT;  // really should not be called
+}
+UText *WholeStringBreakIterator::getUText(UText * /*fillIn*/, UErrorCode &errorCode) const {
+    if (U_SUCCESS(errorCode)) {
+        errorCode = U_UNSUPPORTED_ERROR;
+    }
+    return nullptr;
+}
+
+void  WholeStringBreakIterator::setText(const UnicodeString &text) {
+    length = text.length();
+}
+void  WholeStringBreakIterator::setText(UText *text, UErrorCode &errorCode) {
+    if (U_SUCCESS(errorCode)) {
+        int64_t length64 = utext_nativeLength(text);
+        if (length64 <= INT32_MAX) {
+            length = (int32_t)length64;
+        } else {
+            errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
+        }
+    }
+}
+void  WholeStringBreakIterator::adoptText(CharacterIterator*) {
+    UPRV_UNREACHABLE_EXIT;  // should not be called
+}
+
+int32_t WholeStringBreakIterator::first() { return 0; }
+int32_t WholeStringBreakIterator::last() { return length; }
+int32_t WholeStringBreakIterator::previous() { return 0; }
+int32_t WholeStringBreakIterator::next() { return length; }
+int32_t WholeStringBreakIterator::current() const { return 0; }
+int32_t WholeStringBreakIterator::following(int32_t /*offset*/) { return length; }
+int32_t WholeStringBreakIterator::preceding(int32_t /*offset*/) { return 0; }
+UBool WholeStringBreakIterator::isBoundary(int32_t /*offset*/) { return false; }
+int32_t WholeStringBreakIterator::next(int32_t /*n*/) { return length; }
+
+WholeStringBreakIterator *WholeStringBreakIterator::createBufferClone(
+        void * /*stackBuffer*/, int32_t & /*BufferSize*/, UErrorCode &errorCode) {
+    if (U_SUCCESS(errorCode)) {
+        errorCode = U_UNSUPPORTED_ERROR;
+    }
+    return nullptr;
+}
+WholeStringBreakIterator &WholeStringBreakIterator::refreshInputText(
+        UText * /*input*/, UErrorCode &errorCode) {
+    if (U_SUCCESS(errorCode)) {
+        errorCode = U_UNSUPPORTED_ERROR;
+    }
+    return *this;
+}
+
+U_CFUNC
+BreakIterator *ustrcase_getTitleBreakIterator(
+        const Locale *locale, const char *locID, uint32_t options, BreakIterator *iter,
+        LocalPointer &ownedIter, UErrorCode &errorCode) {
+    if (U_FAILURE(errorCode)) { return nullptr; }
+    options &= U_TITLECASE_ITERATOR_MASK;
+    if (options != 0 && iter != nullptr) {
+        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+    if (iter == nullptr) {
+        switch (options) {
+        case 0:
+            iter = BreakIterator::createWordInstance(
+                locale != nullptr ? *locale : Locale(locID), errorCode);
+            break;
+        case U_TITLECASE_WHOLE_STRING:
+            iter = new WholeStringBreakIterator();
+            if (iter == nullptr) {
+                errorCode = U_MEMORY_ALLOCATION_ERROR;
+            }
+            break;
+        case U_TITLECASE_SENTENCES:
+            iter = BreakIterator::createSentenceInstance(
+                locale != nullptr ? *locale : Locale(locID), errorCode);
+            break;
+        default:
+            errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+            break;
+        }
+        ownedIter.adoptInstead(iter);
+    }
+    return iter;
+}
+
+int32_t CaseMap::toTitle(
+        const char *locale, uint32_t options, BreakIterator *iter,
+        const char16_t *src, int32_t srcLength,
+        char16_t *dest, int32_t destCapacity, Edits *edits,
+        UErrorCode &errorCode) {
+    LocalPointer ownedIter;
+    iter = ustrcase_getTitleBreakIterator(nullptr, locale, options, iter, ownedIter, errorCode);
+    if(iter==nullptr) {
+        return 0;
+    }
+    UnicodeString s(srcLength<0, src, srcLength);
+    iter->setText(s);
+    return ustrcase_map(
+        ustrcase_getCaseLocale(locale), options, iter,
+        dest, destCapacity,
+        src, srcLength,
+        ustrcase_internalToTitle, edits, errorCode);
+}
+
+U_NAMESPACE_END
+
+U_NAMESPACE_USE
+
+U_CAPI int32_t U_EXPORT2
+u_strToTitle(char16_t *dest, int32_t destCapacity,
+             const char16_t *src, int32_t srcLength,
+             UBreakIterator *titleIter,
+             const char *locale,
+             UErrorCode *pErrorCode) {
+    LocalPointer ownedIter;
+    BreakIterator *iter = ustrcase_getTitleBreakIterator(
+        nullptr, locale, 0, reinterpret_cast(titleIter),
+        ownedIter, *pErrorCode);
+    if (iter == nullptr) {
+        return 0;
+    }
+    UnicodeString s(srcLength<0, src, srcLength);
+    iter->setText(s);
+    return ustrcase_mapWithOverlap(
+        ustrcase_getCaseLocale(locale), 0, iter,
+        dest, destCapacity,
+        src, srcLength,
+        ustrcase_internalToTitle, *pErrorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+ucasemap_toTitle(UCaseMap *csm,
+                 char16_t *dest, int32_t destCapacity,
+                 const char16_t *src, int32_t srcLength,
+                 UErrorCode *pErrorCode) {
+    if (U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+    if (csm->iter == nullptr) {
+        LocalPointer ownedIter;
+        BreakIterator *iter = ustrcase_getTitleBreakIterator(
+            nullptr, csm->locale, csm->options, nullptr, ownedIter, *pErrorCode);
+        if (iter == nullptr) {
+            return 0;
+        }
+        csm->iter = ownedIter.orphan();
+    }
+    UnicodeString s(srcLength<0, src, srcLength);
+    csm->iter->setText(s);
+    return ustrcase_map(
+        csm->caseLocale, csm->options, csm->iter,
+        dest, destCapacity,
+        src, srcLength,
+        ustrcase_internalToTitle, nullptr, *pErrorCode);
+}
+
+#endif  // !UCONFIG_NO_BREAK_ITERATION
diff --git a/intl/icu/source/common/ustr_wcs.cpp b/intl/icu/source/common/ustr_wcs.cpp
new file mode 100644
index 0000000000..efbbbc2f18
--- /dev/null
+++ b/intl/icu/source/common/ustr_wcs.cpp
@@ -0,0 +1,535 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2001-2012, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  ustr_wcs.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2004sep07
+*   created by: Markus W. Scherer
+*
+*   u_strToWCS() and u_strFromWCS() functions
+*   moved here from ustrtrns.c for better modularization.
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/ustring.h"
+#include "cstring.h"
+#include "cwchar.h"
+#include "cmemory.h"
+#include "ustr_imp.h"
+#include "ustr_cnv.h"
+
+#if defined(U_WCHAR_IS_UTF16) || defined(U_WCHAR_IS_UTF32) || !UCONFIG_NO_CONVERSION
+
+#define _STACK_BUFFER_CAPACITY 1000
+#define _BUFFER_CAPACITY_MULTIPLIER 2
+
+#if !defined(U_WCHAR_IS_UTF16) && !defined(U_WCHAR_IS_UTF32)
+// TODO: We should use CharString for char buffers and UnicodeString for char16_t buffers.
+// Then we could change this to work only with wchar_t buffers.
+static inline UBool 
+u_growAnyBufferFromStatic(void *context,
+                       void **pBuffer, int32_t *pCapacity, int32_t reqCapacity,
+                       int32_t length, int32_t size) {
+    // Use char* not void* to avoid the compiler's strict-aliasing assumptions
+    // and related warnings.
+    char *newBuffer=(char *)uprv_malloc(reqCapacity*size);
+    if(newBuffer!=nullptr) {
+        if(length>0) {
+            uprv_memcpy(newBuffer, *pBuffer, (size_t)length*size);
+        }
+        *pCapacity=reqCapacity;
+    } else {
+        *pCapacity=0;
+    }
+
+    /* release the old pBuffer if it was not statically allocated */
+    if(*pBuffer!=(char *)context) {
+        uprv_free(*pBuffer);
+    }
+
+    *pBuffer=newBuffer;
+    return (UBool)(newBuffer!=nullptr);
+}
+
+/* helper function */
+static wchar_t* 
+_strToWCS(wchar_t *dest, 
+           int32_t destCapacity,
+           int32_t *pDestLength,
+           const char16_t *src,
+           int32_t srcLength,
+           UErrorCode *pErrorCode){
+
+    char stackBuffer [_STACK_BUFFER_CAPACITY];
+    char* tempBuf = stackBuffer;
+    int32_t tempBufCapacity = _STACK_BUFFER_CAPACITY;
+    char* tempBufLimit = stackBuffer + tempBufCapacity;
+    UConverter* conv = nullptr;
+    char* saveBuf = tempBuf;
+    wchar_t* intTarget=nullptr;
+    int32_t intTargetCapacity=0;
+    int count=0,retVal=0;
+    
+    const char16_t *pSrcLimit =nullptr;
+    const char16_t *pSrc = src;
+
+    conv = u_getDefaultConverter(pErrorCode);
+    
+    if(U_FAILURE(*pErrorCode)){
+        return nullptr;
+    }
+    
+    if(srcLength == -1){
+        srcLength = u_strlen(pSrc);
+    }
+    
+    pSrcLimit = pSrc + srcLength;
+
+    for(;;) {
+        /* reset the error state */
+        *pErrorCode = U_ZERO_ERROR;
+
+        /* convert to chars using default converter */
+        ucnv_fromUnicode(conv,&tempBuf,tempBufLimit,&pSrc,pSrcLimit,nullptr,(UBool)(pSrc==pSrcLimit),pErrorCode);
+        count =(tempBuf - saveBuf);
+        
+        /* This should rarely occur */
+        if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR){
+            tempBuf = saveBuf;
+            
+            /* we don't have enough room on the stack grow the buffer */
+            int32_t newCapacity = 2 * srcLength;
+            if(newCapacity <= tempBufCapacity) {
+                newCapacity = _BUFFER_CAPACITY_MULTIPLIER * tempBufCapacity;
+            }
+            if(!u_growAnyBufferFromStatic(stackBuffer,(void**) &tempBuf, &tempBufCapacity,
+                    newCapacity, count, 1)) {
+                goto cleanup;
+            }
+          
+           saveBuf = tempBuf;
+           tempBufLimit = tempBuf + tempBufCapacity;
+           tempBuf = tempBuf + count;
+
+        } else {
+            break;
+        }
+    }
+
+    if(U_FAILURE(*pErrorCode)){
+        goto cleanup;
+    }
+
+    /* done with conversion null terminate the char buffer */
+    if(count>=tempBufCapacity){
+        tempBuf = saveBuf;
+        /* we don't have enough room on the stack grow the buffer */
+        if(!u_growAnyBufferFromStatic(stackBuffer,(void**) &tempBuf, &tempBufCapacity, 
+                count+1, count, 1)) {
+            goto cleanup;
+        }              
+       saveBuf = tempBuf;
+    }
+    
+    saveBuf[count]=0;
+      
+
+    /* allocate more space than required 
+     * here we assume that every char requires 
+     * no more than 2 wchar_ts
+     */
+    intTargetCapacity =  (count * _BUFFER_CAPACITY_MULTIPLIER + 1) /*for null termination */;
+    intTarget = (wchar_t*)uprv_malloc( intTargetCapacity * sizeof(wchar_t) );
+
+    if(intTarget){
+
+        int32_t nulLen = 0;
+        int32_t remaining = intTargetCapacity;
+        wchar_t* pIntTarget=intTarget;
+        tempBuf = saveBuf;
+        
+        /* now convert the mbs to wcs */
+        for(;;){
+            
+            /* we can call the system API since we are sure that
+             * there is atleast 1 null in the input
+             */
+            retVal = uprv_mbstowcs(pIntTarget,(tempBuf+nulLen),remaining);
+            
+            if(retVal==-1){
+                *pErrorCode = U_INVALID_CHAR_FOUND;
+                break;
+            }else if(retVal== remaining){/* should never occur */
+                int numWritten = (pIntTarget-intTarget);
+                u_growAnyBufferFromStatic(nullptr,(void**) &intTarget,
+                                          &intTargetCapacity,
+                                          intTargetCapacity * _BUFFER_CAPACITY_MULTIPLIER,
+                                          numWritten,
+                                          sizeof(wchar_t));
+                pIntTarget = intTarget;
+                remaining=intTargetCapacity;
+
+                if(nulLen!=count){ /*there are embedded nulls*/
+                    pIntTarget+=numWritten;
+                    remaining-=numWritten;
+                }
+
+            }else{
+                int32_t nulVal;
+                /*scan for nulls */
+                /* we donot check for limit since tempBuf is null terminated */
+                while(tempBuf[nulLen++] != 0){
+                }
+                nulVal = (nulLen < srcLength) ? 1 : 0; 
+                pIntTarget = pIntTarget + retVal+nulVal;
+                remaining -=(retVal+nulVal);
+            
+                /* check if we have reached the source limit*/
+                if(nulLen>=(count)){
+                    break;
+                }
+            }
+        }
+        count = (int32_t)(pIntTarget-intTarget);
+       
+        if(0 < count && count <= destCapacity){
+            uprv_memcpy(dest, intTarget, (size_t)count*sizeof(wchar_t));
+        }  
+
+        if(pDestLength){
+            *pDestLength = count;
+        }
+
+        /* free the allocated memory */
+        uprv_free(intTarget);
+
+    }else{
+        *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
+    }
+cleanup:
+    /* are we still using stack buffer */
+    if(stackBuffer != saveBuf){
+        uprv_free(saveBuf);
+    }
+    u_terminateWChars(dest,destCapacity,count,pErrorCode);
+
+    u_releaseDefaultConverter(conv);
+
+    return dest;
+}
+#endif
+
+U_CAPI wchar_t* U_EXPORT2
+u_strToWCS(wchar_t *dest, 
+           int32_t destCapacity,
+           int32_t *pDestLength,
+           const char16_t *src,
+           int32_t srcLength,
+           UErrorCode *pErrorCode){
+
+    /* args check */
+    if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)){
+        return nullptr;
+    }
+        
+    if( (src==nullptr && srcLength!=0) || srcLength < -1 ||
+        (destCapacity<0) || (dest == nullptr && destCapacity > 0)
+    ) {
+        *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+    
+#ifdef U_WCHAR_IS_UTF16
+    /* wchar_t is UTF-16 just do a memcpy */
+    if(srcLength == -1){
+        srcLength = u_strlen(src);
+    }
+    if(0 < srcLength && srcLength <= destCapacity){
+        u_memcpy((char16_t *)dest, src, srcLength);
+    }
+    if(pDestLength){
+       *pDestLength = srcLength;
+    }
+
+    u_terminateUChars((char16_t *)dest,destCapacity,srcLength,pErrorCode);
+
+    return dest;
+
+#elif defined U_WCHAR_IS_UTF32
+    
+    return (wchar_t*)u_strToUTF32((UChar32*)dest, destCapacity, pDestLength,
+                                  src, srcLength, pErrorCode);
+
+#else
+    
+    return _strToWCS(dest,destCapacity,pDestLength,src,srcLength, pErrorCode);
+    
+#endif
+
+}
+
+#if !defined(U_WCHAR_IS_UTF16) && !defined(U_WCHAR_IS_UTF32)
+/* helper function */
+static char16_t*
+_strFromWCS( char16_t   *dest,
+             int32_t destCapacity, 
+             int32_t *pDestLength,
+             const wchar_t *src,
+             int32_t srcLength,
+             UErrorCode *pErrorCode)
+{
+    int32_t retVal =0, count =0 ;
+    UConverter* conv = nullptr;
+    char16_t* pTarget = nullptr;
+    char16_t* pTargetLimit = nullptr;
+    char16_t* target = nullptr;
+    
+    char16_t uStack [_STACK_BUFFER_CAPACITY];
+
+    wchar_t wStack[_STACK_BUFFER_CAPACITY];
+    wchar_t* pWStack = wStack;
+
+
+    char cStack[_STACK_BUFFER_CAPACITY];
+    int32_t cStackCap = _STACK_BUFFER_CAPACITY;
+    char* pCSrc=cStack;
+    char* pCSave=pCSrc;
+    char* pCSrcLimit=nullptr;
+
+    const wchar_t* pSrc = src;
+    const wchar_t* pSrcLimit = nullptr;
+
+    if(srcLength ==-1){
+        /* if the wchar_t source is null terminated we can safely
+         * assume that there are no embedded nulls, this is a fast
+         * path for null terminated strings.
+         */
+        for(;;){
+            /* convert wchars  to chars */
+            retVal = uprv_wcstombs(pCSrc,src, cStackCap);
+    
+            if(retVal == -1){
+                *pErrorCode = U_ILLEGAL_CHAR_FOUND;
+                goto cleanup;
+            }else if(retVal >= (cStackCap-1)){
+                /* Should rarely occur */
+                u_growAnyBufferFromStatic(cStack,(void**)&pCSrc,&cStackCap,
+                    cStackCap * _BUFFER_CAPACITY_MULTIPLIER, 0, sizeof(char));
+                pCSave = pCSrc;
+            }else{
+                /* converted every thing */
+                pCSrc = pCSrc+retVal;
+                break;
+            }
+        }
+        
+    }else{
+        /* here the source is not null terminated 
+         * so it may have nulls embedded and we need to
+         * do some extra processing 
+         */
+        int32_t remaining =cStackCap;
+        
+        pSrcLimit = src + srcLength;
+
+        for(;;){
+            int32_t nulLen = 0;
+
+            /* find nulls in the string */
+            while(nulLen= _STACK_BUFFER_CAPACITY){
+                    /* Should rarely occur */
+                    /* allocate new buffer buffer */
+                    pWStack =(wchar_t*) uprv_malloc(sizeof(wchar_t) * (nulLen + 1));
+                    if(pWStack==nullptr){
+                        *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
+                        goto cleanup;
+                    }
+                }
+                if(nulLen>0){
+                    /* copy the contents to tempStack */
+                    uprv_memcpy(pWStack, pSrc, (size_t)nulLen*sizeof(wchar_t));
+                }
+            
+                /* null terminate the tempBuffer */
+                pWStack[nulLen] =0 ;
+            
+                if(remaining < (nulLen * MB_CUR_MAX)){
+                    /* Should rarely occur */
+                    int32_t len = (pCSrc-pCSave);
+                    pCSrc = pCSave;
+                    /* we do not have enough room so grow the buffer*/
+                    u_growAnyBufferFromStatic(cStack,(void**)&pCSrc,&cStackCap,
+                           cStackCap+(nulLen*MB_CUR_MAX),len,sizeof(char));
+
+                    pCSave = pCSrc;
+                    pCSrc = pCSave+len;
+                    remaining = cStackCap-(pCSrc - pCSave);
+                }
+                /* convert to chars */
+                retVal = uprv_wcstombs(pCSrc,pWStack,remaining);
+            
+                pCSrc += retVal;
+                pSrc  += nulLen;
+                srcLength-=nulLen; /* decrement the srcLength */
+                break;
+            }
+        }
+    }
+
+    /* OK..now we have converted from wchar_ts to chars now 
+     * convert chars to UChars 
+     */
+    pCSrcLimit = pCSrc;
+    pCSrc = pCSave;
+    pTarget = target= dest;
+    pTargetLimit = dest + destCapacity;    
+    
+    conv= u_getDefaultConverter(pErrorCode);
+    
+    if(U_FAILURE(*pErrorCode)|| conv==nullptr){
+        goto cleanup;
+    }
+    
+    for(;;) {
+        
+        *pErrorCode = U_ZERO_ERROR;
+        
+        /* convert to stack buffer*/
+        ucnv_toUnicode(conv,&pTarget,pTargetLimit,(const char**)&pCSrc,pCSrcLimit,nullptr,(UBool)(pCSrc==pCSrcLimit),pErrorCode);
+        
+        /* increment count to number written to stack */
+        count+= pTarget - target;
+        
+        if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR){
+            target = uStack;
+            pTarget = uStack;
+            pTargetLimit = uStack + _STACK_BUFFER_CAPACITY;
+        } else {
+            break;
+        }
+        
+    }
+    
+    if(pDestLength){
+        *pDestLength =count;
+    }
+
+    u_terminateUChars(dest,destCapacity,count,pErrorCode);
+    
+cleanup:
+ 
+    if(cStack != pCSave){
+        uprv_free(pCSave);
+    }
+
+    if(wStack != pWStack){
+        uprv_free(pWStack);
+    }
+    
+    u_releaseDefaultConverter(conv);
+
+    return dest;
+}
+#endif
+
+U_CAPI char16_t* U_EXPORT2
+u_strFromWCS(char16_t   *dest,
+             int32_t destCapacity, 
+             int32_t *pDestLength,
+             const wchar_t *src,
+             int32_t srcLength,
+             UErrorCode *pErrorCode)
+{
+
+    /* args check */
+    if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)){
+        return nullptr;
+    }
+
+    if( (src==nullptr && srcLength!=0) || srcLength < -1 ||
+        (destCapacity<0) || (dest == nullptr && destCapacity > 0)
+    ) {
+        *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+
+#ifdef U_WCHAR_IS_UTF16
+    /* wchar_t is UTF-16 just do a memcpy */
+    if(srcLength == -1){
+        srcLength = u_strlen((const char16_t *)src);
+    }
+    if(0 < srcLength && srcLength <= destCapacity){
+        u_memcpy(dest, (const char16_t *)src, srcLength);
+    }
+    if(pDestLength){
+       *pDestLength = srcLength;
+    }
+
+    u_terminateUChars(dest,destCapacity,srcLength,pErrorCode);
+
+    return dest;
+
+#elif defined U_WCHAR_IS_UTF32
+    
+    return u_strFromUTF32(dest, destCapacity, pDestLength,
+                          (UChar32*)src, srcLength, pErrorCode);
+
+#else
+
+    return _strFromWCS(dest,destCapacity,pDestLength,src,srcLength,pErrorCode);  
+
+#endif
+
+}
+
+#endif /* #if !defined(U_WCHAR_IS_UTF16) && !defined(U_WCHAR_IS_UTF32) && !UCONFIG_NO_CONVERSION */
diff --git a/intl/icu/source/common/ustrcase.cpp b/intl/icu/source/common/ustrcase.cpp
new file mode 100644
index 0000000000..537b5ba857
--- /dev/null
+++ b/intl/icu/source/common/ustrcase.cpp
@@ -0,0 +1,1900 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2001-2015, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  ustrcase.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2002feb20
+*   created by: Markus W. Scherer
+*
+*   Implementation file for string casing C API functions.
+*   Uses functions from uchar.c for basic functionality that requires access
+*   to the Unicode Character Database (uprops.dat).
+*/
+
+#include "unicode/utypes.h"
+#include "unicode/brkiter.h"
+#include "unicode/casemap.h"
+#include "unicode/edits.h"
+#include "unicode/stringoptions.h"
+#include "unicode/ustring.h"
+#include "unicode/ucasemap.h"
+#include "unicode/ubrk.h"
+#include "unicode/utf.h"
+#include "unicode/utf16.h"
+#include "cmemory.h"
+#include "ucase.h"
+#include "ucasemap_imp.h"
+#include "ustr_imp.h"
+#include "uassert.h"
+
+/**
+ * Code point for COMBINING ACUTE ACCENT
+ * @internal
+ */
+#define ACUTE u'\u0301'
+
+U_NAMESPACE_BEGIN
+
+namespace {
+
+int32_t checkOverflowAndEditsError(int32_t destIndex, int32_t destCapacity,
+                                   Edits *edits, UErrorCode &errorCode) {
+    if (U_SUCCESS(errorCode)) {
+        if (destIndex > destCapacity) {
+            errorCode = U_BUFFER_OVERFLOW_ERROR;
+        } else if (edits != nullptr) {
+            edits->copyErrorTo(errorCode);
+        }
+    }
+    return destIndex;
+}
+
+/* Appends a full case mapping result, see UCASE_MAX_STRING_LENGTH. */
+inline int32_t
+appendResult(char16_t *dest, int32_t destIndex, int32_t destCapacity,
+             int32_t result, const char16_t *s,
+             int32_t cpLength, uint32_t options, icu::Edits *edits) {
+    UChar32 c;
+    int32_t length;
+
+    /* decode the result */
+    if(result<0) {
+        /* (not) original code point */
+        if(edits!=nullptr) {
+            edits->addUnchanged(cpLength);
+        }
+        if(options & U_OMIT_UNCHANGED_TEXT) {
+            return destIndex;
+        }
+        c=~result;
+        if(destIndexaddReplace(cpLength, 1);
+            }
+            return destIndex;
+        } else {
+            c=result;
+            length=U16_LENGTH(c);
+        }
+        if(edits!=nullptr) {
+            edits->addReplace(cpLength, length);
+        }
+    }
+    if(length>(INT32_MAX-destIndex)) {
+        return -1;  // integer overflow
+    }
+
+    if(destIndex=0) {
+            /* code point */
+            UBool isError=false;
+            U16_APPEND(dest, destIndex, destCapacity, c, isError);
+            if(isError) {
+                /* overflow, nothing written */
+                destIndex+=length;
+            }
+        } else {
+            /* string */
+            if((destIndex+length)<=destCapacity) {
+                while(length>0) {
+                    dest[destIndex++]=*s++;
+                    --length;
+                }
+            } else {
+                /* overflow */
+                destIndex+=length;
+            }
+        }
+    } else {
+        /* preflight */
+        destIndex+=length;
+    }
+    return destIndex;
+}
+
+inline int32_t
+appendUChar(char16_t *dest, int32_t destIndex, int32_t destCapacity, char16_t c) {
+    if(destIndexaddUnchanged(length);
+    }
+    if(options & U_OMIT_UNCHANGED_TEXT) {
+        return destIndex;
+    }
+    if(length>(INT32_MAX-destIndex)) {
+        return -1;  // integer overflow
+    }
+    if((destIndex+length)<=destCapacity) {
+        u_memcpy(dest+destIndex, s, length);
+    }
+    return destIndex + length;
+}
+
+inline int32_t
+appendUnchanged(char16_t *dest, int32_t destIndex, int32_t destCapacity,
+                const char16_t *s, int32_t length, uint32_t options, icu::Edits *edits) {
+    if (length <= 0) {
+        return destIndex;
+    }
+    return appendNonEmptyUnchanged(dest, destIndex, destCapacity, s, length, options, edits);
+}
+
+UChar32 U_CALLCONV
+utf16_caseContextIterator(void *context, int8_t dir) {
+    UCaseContext *csc=(UCaseContext *)context;
+    UChar32 c;
+
+    if(dir<0) {
+        /* reset for backward iteration */
+        csc->index=csc->cpStart;
+        csc->dir=dir;
+    } else if(dir>0) {
+        /* reset for forward iteration */
+        csc->index=csc->cpLimit;
+        csc->dir=dir;
+    } else {
+        /* continue current iteration direction */
+        dir=csc->dir;
+    }
+
+    if(dir<0) {
+        if(csc->startindex) {
+            U16_PREV((const char16_t *)csc->p, csc->start, csc->index, c);
+            return c;
+        }
+    } else {
+        if(csc->indexlimit) {
+            U16_NEXT((const char16_t *)csc->p, csc->index, csc->limit, c);
+            return c;
+        }
+    }
+    return U_SENTINEL;
+}
+
+/**
+ * caseLocale >= 0: Lowercases [srcStart..srcLimit[ but takes context [0..srcLength[ into account.
+ * caseLocale < 0: Case-folds [srcStart..srcLimit[.
+ */
+int32_t toLower(int32_t caseLocale, uint32_t options,
+                char16_t *dest, int32_t destCapacity,
+                const char16_t *src, UCaseContext *csc, int32_t srcStart, int32_t srcLimit,
+                icu::Edits *edits, UErrorCode &errorCode) {
+    const int8_t *latinToLower;
+    if (caseLocale == UCASE_LOC_ROOT ||
+            (caseLocale >= 0 ?
+                !(caseLocale == UCASE_LOC_TURKISH || caseLocale == UCASE_LOC_LITHUANIAN) :
+                (options & _FOLD_CASE_OPTIONS_MASK) == U_FOLD_CASE_DEFAULT)) {
+        latinToLower = LatinCase::TO_LOWER_NORMAL;
+    } else {
+        latinToLower = LatinCase::TO_LOWER_TR_LT;
+    }
+    const UTrie2 *trie = ucase_getTrie();
+    int32_t destIndex = 0;
+    int32_t prev = srcStart;
+    int32_t srcIndex = srcStart;
+    for (;;) {
+        // fast path for simple cases
+        char16_t lead = 0;
+        while (srcIndex < srcLimit) {
+            lead = src[srcIndex];
+            int32_t delta;
+            if (lead < LatinCase::LONG_S) {
+                int8_t d = latinToLower[lead];
+                if (d == LatinCase::EXC) { break; }
+                ++srcIndex;
+                if (d == 0) { continue; }
+                delta = d;
+            } else if (lead >= 0xd800) {
+                break;  // surrogate or higher
+            } else {
+                uint16_t props = UTRIE2_GET16_FROM_U16_SINGLE_LEAD(trie, lead);
+                if (UCASE_HAS_EXCEPTION(props)) { break; }
+                ++srcIndex;
+                if (!UCASE_IS_UPPER_OR_TITLE(props) || (delta = UCASE_GET_DELTA(props)) == 0) {
+                    continue;
+                }
+            }
+            lead += static_cast(delta);
+            destIndex = appendUnchanged(dest, destIndex, destCapacity,
+                                        src + prev, srcIndex - 1 - prev, options, edits);
+            if (destIndex >= 0) {
+                destIndex = appendUChar(dest, destIndex, destCapacity, lead);
+                if (edits != nullptr) {
+                    edits->addReplace(1, 1);
+                }
+            }
+            if (destIndex < 0) {
+                errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
+                return 0;
+            }
+            prev = srcIndex;
+        }
+        if (srcIndex >= srcLimit) {
+            break;
+        }
+        // slow path
+        int32_t cpStart = srcIndex++;
+        char16_t trail;
+        UChar32 c;
+        if (U16_IS_LEAD(lead) && srcIndex < srcLimit && U16_IS_TRAIL(trail = src[srcIndex])) {
+            c = U16_GET_SUPPLEMENTARY(lead, trail);
+            ++srcIndex;
+        } else {
+            c = lead;
+        }
+        const char16_t *s;
+        if (caseLocale >= 0) {
+            csc->cpStart = cpStart;
+            csc->cpLimit = srcIndex;
+            c = ucase_toFullLower(c, utf16_caseContextIterator, csc, &s, caseLocale);
+        } else {
+            c = ucase_toFullFolding(c, &s, options);
+        }
+        if (c >= 0) {
+            destIndex = appendUnchanged(dest, destIndex, destCapacity,
+                                        src + prev, cpStart - prev, options, edits);
+            if (destIndex >= 0) {
+                destIndex = appendResult(dest, destIndex, destCapacity, c, s,
+                                         srcIndex - cpStart, options, edits);
+            }
+            if (destIndex < 0) {
+                errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
+                return 0;
+            }
+            prev = srcIndex;
+        }
+    }
+    destIndex = appendUnchanged(dest, destIndex, destCapacity,
+                                src + prev, srcIndex - prev, options, edits);
+    if (destIndex < 0) {
+        errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
+        return 0;
+    }
+    return destIndex;
+}
+
+int32_t toUpper(int32_t caseLocale, uint32_t options,
+                char16_t *dest, int32_t destCapacity,
+                const char16_t *src, UCaseContext *csc, int32_t srcLength,
+                icu::Edits *edits, UErrorCode &errorCode) {
+    const int8_t *latinToUpper;
+    if (caseLocale == UCASE_LOC_TURKISH) {
+        latinToUpper = LatinCase::TO_UPPER_TR;
+    } else {
+        latinToUpper = LatinCase::TO_UPPER_NORMAL;
+    }
+    const UTrie2 *trie = ucase_getTrie();
+    int32_t destIndex = 0;
+    int32_t prev = 0;
+    int32_t srcIndex = 0;
+    for (;;) {
+        // fast path for simple cases
+        char16_t lead = 0;
+        while (srcIndex < srcLength) {
+            lead = src[srcIndex];
+            int32_t delta;
+            if (lead < LatinCase::LONG_S) {
+                int8_t d = latinToUpper[lead];
+                if (d == LatinCase::EXC) { break; }
+                ++srcIndex;
+                if (d == 0) { continue; }
+                delta = d;
+            } else if (lead >= 0xd800) {
+                break;  // surrogate or higher
+            } else {
+                uint16_t props = UTRIE2_GET16_FROM_U16_SINGLE_LEAD(trie, lead);
+                if (UCASE_HAS_EXCEPTION(props)) { break; }
+                ++srcIndex;
+                if (UCASE_GET_TYPE(props) != UCASE_LOWER || (delta = UCASE_GET_DELTA(props)) == 0) {
+                    continue;
+                }
+            }
+            lead += static_cast(delta);
+            destIndex = appendUnchanged(dest, destIndex, destCapacity,
+                                        src + prev, srcIndex - 1 - prev, options, edits);
+            if (destIndex >= 0) {
+                destIndex = appendUChar(dest, destIndex, destCapacity, lead);
+                if (edits != nullptr) {
+                    edits->addReplace(1, 1);
+                }
+            }
+            if (destIndex < 0) {
+                errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
+                return 0;
+            }
+            prev = srcIndex;
+        }
+        if (srcIndex >= srcLength) {
+            break;
+        }
+        // slow path
+        int32_t cpStart;
+        csc->cpStart = cpStart = srcIndex++;
+        char16_t trail;
+        UChar32 c;
+        if (U16_IS_LEAD(lead) && srcIndex < srcLength && U16_IS_TRAIL(trail = src[srcIndex])) {
+            c = U16_GET_SUPPLEMENTARY(lead, trail);
+            ++srcIndex;
+        } else {
+            c = lead;
+        }
+        csc->cpLimit = srcIndex;
+        const char16_t *s;
+        c = ucase_toFullUpper(c, utf16_caseContextIterator, csc, &s, caseLocale);
+        if (c >= 0) {
+            destIndex = appendUnchanged(dest, destIndex, destCapacity,
+                                        src + prev, cpStart - prev, options, edits);
+            if (destIndex >= 0) {
+                destIndex = appendResult(dest, destIndex, destCapacity, c, s,
+                                         srcIndex - cpStart, options, edits);
+            }
+            if (destIndex < 0) {
+                errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
+                return 0;
+            }
+            prev = srcIndex;
+        }
+    }
+    destIndex = appendUnchanged(dest, destIndex, destCapacity,
+                                src + prev, srcIndex - prev, options, edits);
+    if (destIndex < 0) {
+        errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
+        return 0;
+    }
+    return destIndex;
+}
+
+}  // namespace
+
+U_NAMESPACE_END
+
+U_NAMESPACE_USE
+
+#if !UCONFIG_NO_BREAK_ITERATION
+
+namespace {
+
+/**
+ * Input: c is a letter I with or without acute accent.
+ * start is the index in src after c, and is less than segmentLimit.
+ * If a plain i/I is followed by a plain j/J,
+ * or an i/I with acute (precomposed or decomposed) is followed by a j/J with acute,
+ * then we output accordingly.
+ *
+ * @return the src index after the titlecased sequence, or the start index if no Dutch IJ
+ */
+int32_t maybeTitleDutchIJ(const char16_t *src, UChar32 c, int32_t start, int32_t segmentLimit,
+                          char16_t *dest, int32_t &destIndex, int32_t destCapacity, uint32_t options,
+                          icu::Edits *edits) {
+    U_ASSERT(start < segmentLimit);
+
+    int32_t index = start;
+    bool withAcute = false;
+
+    // If the conditions are met, then the following variables tell us what to output.
+    int32_t unchanged1 = 0;  // code units before the j, or the whole sequence (0..3)
+    bool doTitleJ = false;  // true if the j needs to be titlecased
+    int32_t unchanged2 = 0;  // after the j (0 or 1)
+
+    // next character after the first letter
+    char16_t c2 = src[index++];
+
+    // Is the first letter an i/I with accent?
+    if (c == u'I') {
+        if (c2 == ACUTE) {
+            withAcute = true;
+            unchanged1 = 1;
+            if (index == segmentLimit) { return start; }
+            c2 = src[index++];
+        }
+    } else {  // Í
+        withAcute = true;
+    }
+
+    // Is the next character a j/J?
+    if (c2 == u'j') {
+        doTitleJ = true;
+    } else if (c2 == u'J') {
+        ++unchanged1;
+    } else {
+        return start;
+    }
+
+    // A plain i/I must be followed by a plain j/J.
+    // An i/I with acute must be followed by a j/J with acute.
+    if (withAcute) {
+        if (index == segmentLimit || src[index++] != ACUTE) { return start; }
+        if (doTitleJ) {
+            unchanged2 = 1;
+        } else {
+            ++unchanged1;
+        }
+    }
+
+    // There must not be another combining mark.
+    if (index < segmentLimit) {
+        int32_t cp;
+        int32_t i = index;
+        U16_NEXT(src, i, segmentLimit, cp);
+        uint32_t typeMask = U_GET_GC_MASK(cp);
+        if ((typeMask & U_GC_M_MASK) != 0) {
+            return start;
+        }
+    }
+
+    // Output the rest of the Dutch IJ.
+    destIndex = appendUnchanged(dest, destIndex, destCapacity, src + start, unchanged1, options, edits);
+    start += unchanged1;
+    if (doTitleJ) {
+        destIndex = appendUChar(dest, destIndex, destCapacity, u'J');
+        if (edits != nullptr) {
+            edits->addReplace(1, 1);
+        }
+        ++start;
+    }
+    destIndex = appendUnchanged(dest, destIndex, destCapacity, src + start, unchanged2, options, edits);
+
+    U_ASSERT(start + unchanged2 == index);
+    return index;
+}
+
+}  // namespace
+
+U_CFUNC int32_t U_CALLCONV
+ustrcase_internalToTitle(int32_t caseLocale, uint32_t options, BreakIterator *iter,
+                         char16_t *dest, int32_t destCapacity,
+                         const char16_t *src, int32_t srcLength,
+                         icu::Edits *edits,
+                         UErrorCode &errorCode) {
+    if (!ustrcase_checkTitleAdjustmentOptions(options, errorCode)) {
+        return 0;
+    }
+
+    /* set up local variables */
+    UCaseContext csc=UCASECONTEXT_INITIALIZER;
+    csc.p=(void *)src;
+    csc.limit=srcLength;
+    int32_t destIndex=0;
+    int32_t prev=0;
+    bool isFirstIndex=true;
+
+    /* titlecasing loop */
+    while(prevfirst();
+        } else {
+            index=iter->next();
+        }
+        if(index==UBRK_DONE || index>srcLength) {
+            index=srcLength;
+        }
+
+        /*
+         * Segment [prev..index[ into 3 parts:
+         * a) skipped characters (copy as-is) [prev..titleStart[
+         * b) first letter (titlecase)              [titleStart..titleLimit[
+         * c) subsequent characters (lowercase)                 [titleLimit..index[
+         */
+        if(prev 0) {
+            uint32_t upper = data & UPPER_MASK;
+            // Add a dialytika to this iota or ypsilon vowel
+            // if we removed a tonos from the previous vowel,
+            // and that previous vowel did not also have (or gain) a dialytika.
+            // Adding one only to the final vowel in a longer sequence
+            // (which does not occur in normal writing) would require lookahead.
+            // Set the same flag as for preserving an existing dialytika.
+            if ((data & HAS_VOWEL) != 0 && (state & AFTER_VOWEL_WITH_ACCENT) != 0 &&
+                    (upper == 0x399 || upper == 0x3A5)) {
+                data |= HAS_DIALYTIKA;
+            }
+            int32_t numYpogegrammeni = 0;  // Map each one to a trailing, spacing, capital iota.
+            if ((data & HAS_YPOGEGRAMMENI) != 0) {
+                numYpogegrammeni = 1;
+            }
+            // Skip combining diacritics after this Greek letter.
+            while (nextIndex < srcLength) {
+                uint32_t diacriticData = getDiacriticData(src[nextIndex]);
+                if (diacriticData != 0) {
+                    data |= diacriticData;
+                    if ((diacriticData & HAS_YPOGEGRAMMENI) != 0) {
+                        ++numYpogegrammeni;
+                    }
+                    ++nextIndex;
+                } else {
+                    break;  // not a Greek diacritic
+                }
+            }
+            if ((data & HAS_VOWEL_AND_ACCENT_AND_DIALYTIKA) == HAS_VOWEL_AND_ACCENT) {
+                nextState |= AFTER_VOWEL_WITH_ACCENT;
+            }
+            // Map according to Greek rules.
+            UBool addTonos = false;
+            if (upper == 0x397 &&
+                    (data & HAS_ACCENT) != 0 &&
+                    numYpogegrammeni == 0 &&
+                    (state & AFTER_CASED) == 0 &&
+                    !isFollowedByCasedLetter(src, nextIndex, srcLength)) {
+                // Keep disjunctive "or" with (only) a tonos.
+                // We use the same "word boundary" conditions as for the Final_Sigma test.
+                if (i == nextIndex) {
+                    upper = 0x389;  // Preserve the precomposed form.
+                } else {
+                    addTonos = true;
+                }
+            } else if ((data & HAS_DIALYTIKA) != 0) {
+                // Preserve a vowel with dialytika in precomposed form if it exists.
+                if (upper == 0x399) {
+                    upper = 0x3AA;
+                    data &= ~HAS_EITHER_DIALYTIKA;
+                } else if (upper == 0x3A5) {
+                    upper = 0x3AB;
+                    data &= ~HAS_EITHER_DIALYTIKA;
+                }
+            }
+
+            UBool change;
+            if (edits == nullptr && (options & U_OMIT_UNCHANGED_TEXT) == 0) {
+                change = true;  // common, simple usage
+            } else {
+                // Find out first whether we are changing the text.
+                change = src[i] != upper || numYpogegrammeni > 0;
+                int32_t i2 = i + 1;
+                if ((data & HAS_EITHER_DIALYTIKA) != 0) {
+                    change |= i2 >= nextIndex || src[i2] != 0x308;
+                    ++i2;
+                }
+                if (addTonos) {
+                    change |= i2 >= nextIndex || src[i2] != 0x301;
+                    ++i2;
+                }
+                int32_t oldLength = nextIndex - i;
+                int32_t newLength = (i2 - i) + numYpogegrammeni;
+                change |= oldLength != newLength;
+                if (change) {
+                    if (edits != nullptr) {
+                        edits->addReplace(oldLength, newLength);
+                    }
+                } else {
+                    if (edits != nullptr) {
+                        edits->addUnchanged(oldLength);
+                    }
+                    // Write unchanged text?
+                    change = (options & U_OMIT_UNCHANGED_TEXT) == 0;
+                }
+            }
+
+            if (change) {
+                destIndex=appendUChar(dest, destIndex, destCapacity, (char16_t)upper);
+                if (destIndex >= 0 && (data & HAS_EITHER_DIALYTIKA) != 0) {
+                    destIndex=appendUChar(dest, destIndex, destCapacity, 0x308);  // restore or add a dialytika
+                }
+                if (destIndex >= 0 && addTonos) {
+                    destIndex=appendUChar(dest, destIndex, destCapacity, 0x301);
+                }
+                while (destIndex >= 0 && numYpogegrammeni > 0) {
+                    destIndex=appendUChar(dest, destIndex, destCapacity, 0x399);
+                    --numYpogegrammeni;
+                }
+                if(destIndex<0) {
+                    errorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+                    return 0;
+                }
+            }
+        } else {
+            const char16_t *s;
+            c=ucase_toFullUpper(c, nullptr, nullptr, &s, UCASE_LOC_GREEK);
+            destIndex = appendResult(dest, destIndex, destCapacity, c, s,
+                                     nextIndex - i, options, edits);
+            if (destIndex < 0) {
+                errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
+                return 0;
+            }
+        }
+        i = nextIndex;
+        state = nextState;
+    }
+
+    return destIndex;
+}
+
+}  // namespace GreekUpper
+U_NAMESPACE_END
+
+/* functions available in the common library (for unistr_case.cpp) */
+
+U_CFUNC int32_t U_CALLCONV
+ustrcase_internalToLower(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED
+                         char16_t *dest, int32_t destCapacity,
+                         const char16_t *src, int32_t srcLength,
+                         icu::Edits *edits,
+                         UErrorCode &errorCode) {
+    UCaseContext csc=UCASECONTEXT_INITIALIZER;
+    csc.p=(void *)src;
+    csc.limit=srcLength;
+    int32_t destIndex = toLower(
+        caseLocale, options,
+        dest, destCapacity,
+        src, &csc, 0, srcLength,
+        edits, errorCode);
+    return checkOverflowAndEditsError(destIndex, destCapacity, edits, errorCode);
+}
+
+U_CFUNC int32_t U_CALLCONV
+ustrcase_internalToUpper(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED
+                         char16_t *dest, int32_t destCapacity,
+                         const char16_t *src, int32_t srcLength,
+                         icu::Edits *edits,
+                         UErrorCode &errorCode) {
+    int32_t destIndex;
+    if (caseLocale == UCASE_LOC_GREEK) {
+        destIndex = GreekUpper::toUpper(options, dest, destCapacity,
+                                        src, srcLength, edits, errorCode);
+    } else {
+        UCaseContext csc=UCASECONTEXT_INITIALIZER;
+        csc.p=(void *)src;
+        csc.limit=srcLength;
+        destIndex = toUpper(
+            caseLocale, options,
+            dest, destCapacity,
+            src, &csc, srcLength,
+            edits, errorCode);
+    }
+    return checkOverflowAndEditsError(destIndex, destCapacity, edits, errorCode);
+}
+
+U_CFUNC int32_t U_CALLCONV
+ustrcase_internalFold(int32_t /* caseLocale */, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED
+                      char16_t *dest, int32_t destCapacity,
+                      const char16_t *src, int32_t srcLength,
+                      icu::Edits *edits,
+                      UErrorCode &errorCode) {
+    int32_t destIndex = toLower(
+        -1, options,
+        dest, destCapacity,
+        src, nullptr, 0, srcLength,
+        edits, errorCode);
+    return checkOverflowAndEditsError(destIndex, destCapacity, edits, errorCode);
+}
+
+U_CFUNC int32_t
+ustrcase_map(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM
+             char16_t *dest, int32_t destCapacity,
+             const char16_t *src, int32_t srcLength,
+             UStringCaseMapper *stringCaseMapper,
+             icu::Edits *edits,
+             UErrorCode &errorCode) {
+    int32_t destLength;
+
+    /* check argument values */
+    if(U_FAILURE(errorCode)) {
+        return 0;
+    }
+    if( destCapacity<0 ||
+        (dest==nullptr && destCapacity>0) ||
+        src==nullptr ||
+        srcLength<-1
+    ) {
+        errorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    /* get the string length */
+    if(srcLength==-1) {
+        srcLength=u_strlen(src);
+    }
+
+    /* check for overlapping source and destination */
+    if( dest!=nullptr &&
+        ((src>=dest && src<(dest+destCapacity)) ||
+         (dest>=src && dest<(src+srcLength)))
+    ) {
+        errorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    if (edits != nullptr && (options & U_EDITS_NO_RESET) == 0) {
+        edits->reset();
+    }
+    destLength=stringCaseMapper(caseLocale, options, UCASEMAP_BREAK_ITERATOR
+                                dest, destCapacity, src, srcLength, edits, errorCode);
+    return u_terminateUChars(dest, destCapacity, destLength, &errorCode);
+}
+
+U_CFUNC int32_t
+ustrcase_mapWithOverlap(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM
+                        char16_t *dest, int32_t destCapacity,
+                        const char16_t *src, int32_t srcLength,
+                        UStringCaseMapper *stringCaseMapper,
+                        UErrorCode &errorCode) {
+    char16_t buffer[300];
+    char16_t *temp;
+
+    int32_t destLength;
+
+    /* check argument values */
+    if(U_FAILURE(errorCode)) {
+        return 0;
+    }
+    if( destCapacity<0 ||
+        (dest==nullptr && destCapacity>0) ||
+        src==nullptr ||
+        srcLength<-1
+    ) {
+        errorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    /* get the string length */
+    if(srcLength==-1) {
+        srcLength=u_strlen(src);
+    }
+
+    /* check for overlapping source and destination */
+    if( dest!=nullptr &&
+        ((src>=dest && src<(dest+destCapacity)) ||
+         (dest>=src && dest<(src+srcLength)))
+    ) {
+        /* overlap: provide a temporary destination buffer and later copy the result */
+        if(destCapacity<=UPRV_LENGTHOF(buffer)) {
+            /* the stack buffer is large enough */
+            temp=buffer;
+        } else {
+            /* allocate a buffer */
+            temp=(char16_t *)uprv_malloc(destCapacity*U_SIZEOF_UCHAR);
+            if(temp==nullptr) {
+                errorCode=U_MEMORY_ALLOCATION_ERROR;
+                return 0;
+            }
+        }
+    } else {
+        temp=dest;
+    }
+
+    destLength=stringCaseMapper(caseLocale, options, UCASEMAP_BREAK_ITERATOR
+                                temp, destCapacity, src, srcLength, nullptr, errorCode);
+    if(temp!=dest) {
+        /* copy the result string to the destination buffer */
+        if (U_SUCCESS(errorCode) && 0 < destLength && destLength <= destCapacity) {
+            u_memmove(dest, temp, destLength);
+        }
+        if(temp!=buffer) {
+            uprv_free(temp);
+        }
+    }
+
+    return u_terminateUChars(dest, destCapacity, destLength, &errorCode);
+}
+
+/* public API functions */
+
+U_CAPI int32_t U_EXPORT2
+u_strFoldCase(char16_t *dest, int32_t destCapacity,
+              const char16_t *src, int32_t srcLength,
+              uint32_t options,
+              UErrorCode *pErrorCode) {
+    return ustrcase_mapWithOverlap(
+        UCASE_LOC_ROOT, options, UCASEMAP_BREAK_ITERATOR_NULL
+        dest, destCapacity,
+        src, srcLength,
+        ustrcase_internalFold, *pErrorCode);
+}
+
+U_NAMESPACE_BEGIN
+
+int32_t CaseMap::fold(
+        uint32_t options,
+        const char16_t *src, int32_t srcLength,
+        char16_t *dest, int32_t destCapacity, Edits *edits,
+        UErrorCode &errorCode) {
+    return ustrcase_map(
+        UCASE_LOC_ROOT, options, UCASEMAP_BREAK_ITERATOR_NULL
+        dest, destCapacity,
+        src, srcLength,
+        ustrcase_internalFold, edits, errorCode);
+}
+
+U_NAMESPACE_END
+
+/* case-insensitive string comparisons -------------------------------------- */
+
+/*
+ * This function is a copy of unorm_cmpEquivFold() minus the parts for
+ * canonical equivalence.
+ * Keep the functions in sync, and see there for how this works.
+ * The duplication is for modularization:
+ * It makes caseless (but not canonical caseless) matches independent of
+ * the normalization code.
+ */
+
+/* stack element for previous-level source/decomposition pointers */
+struct CmpEquivLevel {
+    const char16_t *start, *s, *limit;
+};
+typedef struct CmpEquivLevel CmpEquivLevel;
+
+/**
+ * Internal implementation code comparing string with case fold.
+ * This function is called from u_strcmpFold() and u_caseInsensitivePrefixMatch().
+ *
+ * @param s1            input string 1
+ * @param length1       length of string 1, or -1 (NUL terminated)
+ * @param s2            input string 2
+ * @param length2       length of string 2, or -1 (NUL terminated)
+ * @param options       compare options
+ * @param matchLen1     (output) length of partial prefix match in s1
+ * @param matchLen2     (output) length of partial prefix match in s2
+ * @param pErrorCode    receives error status
+ * @return The result of comparison
+ */
+static int32_t _cmpFold(
+            const char16_t *s1, int32_t length1,
+            const char16_t *s2, int32_t length2,
+            uint32_t options,
+            int32_t *matchLen1, int32_t *matchLen2,
+            UErrorCode *pErrorCode) {
+    int32_t cmpRes = 0;
+
+    /* current-level start/limit - s1/s2 as current */
+    const char16_t *start1, *start2, *limit1, *limit2;
+
+    /* points to the original start address */
+    const char16_t *org1, *org2;
+
+    /* points to the end of match + 1 */
+    const char16_t *m1, *m2;
+
+    /* case folding variables */
+    const char16_t *p;
+    int32_t length;
+
+    /* stacks of previous-level start/current/limit */
+    CmpEquivLevel stack1[2], stack2[2];
+
+    /* case folding buffers, only use current-level start/limit */
+    char16_t fold1[UCASE_MAX_STRING_LENGTH+1], fold2[UCASE_MAX_STRING_LENGTH+1];
+
+    /* track which is the current level per string */
+    int32_t level1, level2;
+
+    /* current code units, and code points for lookups */
+    UChar32 c1, c2, cp1, cp2;
+
+    /* no argument error checking because this itself is not an API */
+
+    /*
+     * assume that at least the option U_COMPARE_IGNORE_CASE is set
+     * otherwise this function would have to behave exactly as uprv_strCompare()
+     */
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+
+    /* initialize */
+    if(matchLen1) {
+        U_ASSERT(matchLen2 !=nullptr);
+        *matchLen1=0;
+        *matchLen2=0;
+    }
+
+    start1=m1=org1=s1;
+    if(length1==-1) {
+        limit1=nullptr;
+    } else {
+        limit1=s1+length1;
+    }
+
+    start2=m2=org2=s2;
+    if(length2==-1) {
+        limit2=nullptr;
+    } else {
+        limit2=s2+length2;
+    }
+
+    level1=level2=0;
+    c1=c2=-1;
+
+    /* comparison loop */
+    for(;;) {
+        /*
+         * here a code unit value of -1 means "get another code unit"
+         * below it will mean "this source is finished"
+         */
+
+        if(c1<0) {
+            /* get next code unit from string 1, post-increment */
+            for(;;) {
+                if(s1==limit1 || ((c1=*s1)==0 && (limit1==nullptr || (options&_STRNCMP_STYLE)))) {
+                    if(level1==0) {
+                        c1=-1;
+                        break;
+                    }
+                } else {
+                    ++s1;
+                    break;
+                }
+
+                /* reached end of level buffer, pop one level */
+                do {
+                    --level1;
+                    start1=stack1[level1].start;    /*Not uninitialized*/
+                } while(start1==nullptr);
+                s1=stack1[level1].s;                /*Not uninitialized*/
+                limit1=stack1[level1].limit;        /*Not uninitialized*/
+            }
+        }
+
+        if(c2<0) {
+            /* get next code unit from string 2, post-increment */
+            for(;;) {
+                if(s2==limit2 || ((c2=*s2)==0 && (limit2==nullptr || (options&_STRNCMP_STYLE)))) {
+                    if(level2==0) {
+                        c2=-1;
+                        break;
+                    }
+                } else {
+                    ++s2;
+                    break;
+                }
+
+                /* reached end of level buffer, pop one level */
+                do {
+                    --level2;
+                    start2=stack2[level2].start;    /*Not uninitialized*/
+                } while(start2==nullptr);
+                s2=stack2[level2].s;                /*Not uninitialized*/
+                limit2=stack2[level2].limit;        /*Not uninitialized*/
+            }
+        }
+
+        /*
+         * compare c1 and c2
+         * either variable c1, c2 is -1 only if the corresponding string is finished
+         */
+        if(c1==c2) {
+            const char16_t *next1, *next2;
+
+            if(c1<0) {
+                cmpRes=0;   /* c1==c2==-1 indicating end of strings */
+                break;
+            }
+
+            /*
+             * Note: Move the match positions in both strings at the same time
+             *      only when corresponding code point(s) in the original strings
+             *      are fully consumed. For example, when comparing s1="Fust" and
+             *      s2="Fu\u00dfball", s2[2] is folded into "ss", and s1[2] matches
+             *      the first code point in the case-folded data. But the second "s"
+             *      has no matching code point in s1, so this implementation returns
+             *      2 as the prefix match length ("Fu").
+             */
+            next1=next2=nullptr;
+            if(level1==0) {
+                next1=s1;
+            } else if(s1==limit1) {
+                /* Note: This implementation only use a single level of stack.
+                 *      If this code needs to be changed to use multiple levels
+                 *      of stacks, the code above should check if the current
+                 *      code is at the end of all stacks.
+                 */
+                U_ASSERT(level1==1);
+
+                /* is s1 at the end of the current stack? */
+                next1=stack1[0].s;
+            }
+
+            if (next1!=nullptr) {
+                if(level2==0) {
+                    next2=s2;
+                } else if(s2==limit2) {
+                    U_ASSERT(level2==1);
+
+                    /* is s2 at the end of the current stack? */
+                    next2=stack2[0].s;
+                }
+                if(next2!=nullptr) {
+                    m1=next1;
+                    m2=next2;
+                }
+            }
+            c1=c2=-1;       /* make us fetch new code units */
+            continue;
+        } else if(c1<0) {
+            cmpRes=-1;      /* string 1 ends before string 2 */
+            break;
+        } else if(c2<0) {
+            cmpRes=1;       /* string 2 ends before string 1 */
+            break;
+        }
+        /* c1!=c2 && c1>=0 && c2>=0 */
+
+        /* get complete code points for c1, c2 for lookups if either is a surrogate */
+        cp1=c1;
+        if(U_IS_SURROGATE(c1)) {
+            char16_t c;
+
+            if(U_IS_SURROGATE_LEAD(c1)) {
+                if(s1!=limit1 && U16_IS_TRAIL(c=*s1)) {
+                    /* advance ++s1; only below if cp1 decomposes/case-folds */
+                    cp1=U16_GET_SUPPLEMENTARY(c1, c);
+                }
+            } else /* isTrail(c1) */ {
+                if(start1<=(s1-2) && U16_IS_LEAD(c=*(s1-2))) {
+                    cp1=U16_GET_SUPPLEMENTARY(c, c1);
+                }
+            }
+        }
+
+        cp2=c2;
+        if(U_IS_SURROGATE(c2)) {
+            char16_t c;
+
+            if(U_IS_SURROGATE_LEAD(c2)) {
+                if(s2!=limit2 && U16_IS_TRAIL(c=*s2)) {
+                    /* advance ++s2; only below if cp2 decomposes/case-folds */
+                    cp2=U16_GET_SUPPLEMENTARY(c2, c);
+                }
+            } else /* isTrail(c2) */ {
+                if(start2<=(s2-2) && U16_IS_LEAD(c=*(s2-2))) {
+                    cp2=U16_GET_SUPPLEMENTARY(c, c2);
+                }
+            }
+        }
+
+        /*
+         * go down one level for each string
+         * continue with the main loop as soon as there is a real change
+         */
+
+        if( level1==0 &&
+            (length=ucase_toFullFolding((UChar32)cp1, &p, options))>=0
+        ) {
+            /* cp1 case-folds to the code point "length" or to p[length] */
+            if(U_IS_SURROGATE(c1)) {
+                if(U_IS_SURROGATE_LEAD(c1)) {
+                    /* advance beyond source surrogate pair if it case-folds */
+                    ++s1;
+                } else /* isTrail(c1) */ {
+                    /*
+                     * we got a supplementary code point when hitting its trail surrogate,
+                     * therefore the lead surrogate must have been the same as in the other string;
+                     * compare this decomposition with the lead surrogate in the other string
+                     * remember that this simulates bulk text replacement:
+                     * the decomposition would replace the entire code point
+                     */
+                    --s2;
+                    --m2;
+                    c2=*(s2-1);
+                }
+            }
+
+            /* push current level pointers */
+            stack1[0].start=start1;
+            stack1[0].s=s1;
+            stack1[0].limit=limit1;
+            ++level1;
+
+            /* copy the folding result to fold1[] */
+            if(length<=UCASE_MAX_STRING_LENGTH) {
+                u_memcpy(fold1, p, length);
+            } else {
+                int32_t i=0;
+                U16_APPEND_UNSAFE(fold1, i, length);
+                length=i;
+            }
+
+            /* set next level pointers to case folding */
+            start1=s1=fold1;
+            limit1=fold1+length;
+
+            /* get ready to read from decomposition, continue with loop */
+            c1=-1;
+            continue;
+        }
+
+        if( level2==0 &&
+            (length=ucase_toFullFolding((UChar32)cp2, &p, options))>=0
+        ) {
+            /* cp2 case-folds to the code point "length" or to p[length] */
+            if(U_IS_SURROGATE(c2)) {
+                if(U_IS_SURROGATE_LEAD(c2)) {
+                    /* advance beyond source surrogate pair if it case-folds */
+                    ++s2;
+                } else /* isTrail(c2) */ {
+                    /*
+                     * we got a supplementary code point when hitting its trail surrogate,
+                     * therefore the lead surrogate must have been the same as in the other string;
+                     * compare this decomposition with the lead surrogate in the other string
+                     * remember that this simulates bulk text replacement:
+                     * the decomposition would replace the entire code point
+                     */
+                    --s1;
+                    --m2;
+                    c1=*(s1-1);
+                }
+            }
+
+            /* push current level pointers */
+            stack2[0].start=start2;
+            stack2[0].s=s2;
+            stack2[0].limit=limit2;
+            ++level2;
+
+            /* copy the folding result to fold2[] */
+            if(length<=UCASE_MAX_STRING_LENGTH) {
+                u_memcpy(fold2, p, length);
+            } else {
+                int32_t i=0;
+                U16_APPEND_UNSAFE(fold2, i, length);
+                length=i;
+            }
+
+            /* set next level pointers to case folding */
+            start2=s2=fold2;
+            limit2=fold2+length;
+
+            /* get ready to read from decomposition, continue with loop */
+            c2=-1;
+            continue;
+        }
+
+        /*
+         * no decomposition/case folding, max level for both sides:
+         * return difference result
+         *
+         * code point order comparison must not just return cp1-cp2
+         * because when single surrogates are present then the surrogate pairs
+         * that formed cp1 and cp2 may be from different string indexes
+         *
+         * example: { d800 d800 dc01 } vs. { d800 dc00 }, compare at second code units
+         * c1=d800 cp1=10001 c2=dc00 cp2=10000
+         * cp1-cp2>0 but c1-c2<0 and in fact in UTF-32 it is { d800 10001 } < { 10000 }
+         *
+         * therefore, use same fix-up as in ustring.c/uprv_strCompare()
+         * except: uprv_strCompare() fetches c=*s while this functions fetches c=*s++
+         * so we have slightly different pointer/start/limit comparisons here
+         */
+
+        if(c1>=0xd800 && c2>=0xd800 && (options&U_COMPARE_CODE_POINT_ORDER)) {
+            /* subtract 0x2800 from BMP code points to make them smaller than supplementary ones */
+            if(
+                (c1<=0xdbff && s1!=limit1 && U16_IS_TRAIL(*s1)) ||
+                (U16_IS_TRAIL(c1) && start1!=(s1-1) && U16_IS_LEAD(*(s1-2)))
+            ) {
+                /* part of a surrogate pair, leave >=d800 */
+            } else {
+                /* BMP code point - may be surrogate code point - make =d800 */
+            } else {
+                /* BMP code point - may be surrogate code point - make (m1-org1);
+        *matchLen2=static_cast(m2-org2);
+    }
+    return cmpRes;
+}
+
+/* internal function */
+U_CFUNC int32_t
+u_strcmpFold(const char16_t *s1, int32_t length1,
+             const char16_t *s2, int32_t length2,
+             uint32_t options,
+             UErrorCode *pErrorCode) {
+    return _cmpFold(s1, length1, s2, length2, options, nullptr, nullptr, pErrorCode);
+}
+
+/* public API functions */
+
+U_CAPI int32_t U_EXPORT2
+u_strCaseCompare(const char16_t *s1, int32_t length1,
+                 const char16_t *s2, int32_t length2,
+                 uint32_t options,
+                 UErrorCode *pErrorCode) {
+    /* argument checking */
+    if(pErrorCode==0 || U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+    if(s1==nullptr || length1<-1 || s2==nullptr || length2<-1) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    return u_strcmpFold(s1, length1, s2, length2,
+                        options|U_COMPARE_IGNORE_CASE,
+                        pErrorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+u_strcasecmp(const char16_t *s1, const char16_t *s2, uint32_t options) {
+    UErrorCode errorCode=U_ZERO_ERROR;
+    return u_strcmpFold(s1, -1, s2, -1,
+                        options|U_COMPARE_IGNORE_CASE,
+                        &errorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+u_memcasecmp(const char16_t *s1, const char16_t *s2, int32_t length, uint32_t options) {
+    UErrorCode errorCode=U_ZERO_ERROR;
+    return u_strcmpFold(s1, length, s2, length,
+                        options|U_COMPARE_IGNORE_CASE,
+                        &errorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+u_strncasecmp(const char16_t *s1, const char16_t *s2, int32_t n, uint32_t options) {
+    UErrorCode errorCode=U_ZERO_ERROR;
+    return u_strcmpFold(s1, n, s2, n,
+                        options|(U_COMPARE_IGNORE_CASE|_STRNCMP_STYLE),
+                        &errorCode);
+}
+
+/* internal API - detect length of shared prefix */
+U_CAPI void
+u_caseInsensitivePrefixMatch(const char16_t *s1, int32_t length1,
+                             const char16_t *s2, int32_t length2,
+                             uint32_t options,
+                             int32_t *matchLen1, int32_t *matchLen2,
+                             UErrorCode *pErrorCode) {
+    _cmpFold(s1, length1, s2, length2, options,
+        matchLen1, matchLen2, pErrorCode);
+}
diff --git a/intl/icu/source/common/ustrcase_locale.cpp b/intl/icu/source/common/ustrcase_locale.cpp
new file mode 100644
index 0000000000..176add9c51
--- /dev/null
+++ b/intl/icu/source/common/ustrcase_locale.cpp
@@ -0,0 +1,94 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*   Copyright (C) 2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*******************************************************************************
+*   file name:  ustrcase_locale.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2011may31
+*   created by: Markus W. Scherer
+*
+*   Locale-sensitive case mapping functions (ones that call uloc_getDefault())
+*   were moved here to break dependency cycles among parts of the common library.
+*/
+
+#include "unicode/utypes.h"
+#include "uassert.h"
+#include "unicode/brkiter.h"
+#include "unicode/casemap.h"
+#include "unicode/ucasemap.h"
+#include "unicode/uloc.h"
+#include "unicode/ustring.h"
+#include "ucase.h"
+#include "ucasemap_imp.h"
+
+U_CFUNC int32_t
+ustrcase_getCaseLocale(const char *locale) {
+    if (locale == nullptr) {
+        locale = uloc_getDefault();
+    }
+    if (*locale == 0) {
+        return UCASE_LOC_ROOT;
+    } else {
+        return ucase_getCaseLocale(locale);
+    }
+}
+
+/* public API functions */
+
+U_CAPI int32_t U_EXPORT2
+u_strToLower(char16_t *dest, int32_t destCapacity,
+             const char16_t *src, int32_t srcLength,
+             const char *locale,
+             UErrorCode *pErrorCode) {
+    return ustrcase_mapWithOverlap(
+        ustrcase_getCaseLocale(locale), 0, UCASEMAP_BREAK_ITERATOR_NULL
+        dest, destCapacity,
+        src, srcLength,
+        ustrcase_internalToLower, *pErrorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+u_strToUpper(char16_t *dest, int32_t destCapacity,
+             const char16_t *src, int32_t srcLength,
+             const char *locale,
+             UErrorCode *pErrorCode) {
+    return ustrcase_mapWithOverlap(
+        ustrcase_getCaseLocale(locale), 0, UCASEMAP_BREAK_ITERATOR_NULL
+        dest, destCapacity,
+        src, srcLength,
+        ustrcase_internalToUpper, *pErrorCode);
+}
+
+U_NAMESPACE_BEGIN
+
+int32_t CaseMap::toLower(
+        const char *locale, uint32_t options,
+        const char16_t *src, int32_t srcLength,
+        char16_t *dest, int32_t destCapacity, Edits *edits,
+        UErrorCode &errorCode) {
+    return ustrcase_map(
+        ustrcase_getCaseLocale(locale), options, UCASEMAP_BREAK_ITERATOR_NULL
+        dest, destCapacity,
+        src, srcLength,
+        ustrcase_internalToLower, edits, errorCode);
+}
+
+int32_t CaseMap::toUpper(
+        const char *locale, uint32_t options,
+        const char16_t *src, int32_t srcLength,
+        char16_t *dest, int32_t destCapacity, Edits *edits,
+        UErrorCode &errorCode) {
+    return ustrcase_map(
+        ustrcase_getCaseLocale(locale), options, UCASEMAP_BREAK_ITERATOR_NULL
+        dest, destCapacity,
+        src, srcLength,
+        ustrcase_internalToUpper, edits, errorCode);
+}
+
+U_NAMESPACE_END
diff --git a/intl/icu/source/common/ustrenum.cpp b/intl/icu/source/common/ustrenum.cpp
new file mode 100644
index 0000000000..a60b2208f6
--- /dev/null
+++ b/intl/icu/source/common/ustrenum.cpp
@@ -0,0 +1,398 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+* Copyright (c) 2002-2014, International Business Machines
+* Corporation and others.  All Rights Reserved.
+**********************************************************************
+* Author: Alan Liu
+* Created: November 11 2002
+* Since: ICU 2.4
+**********************************************************************
+*/
+#include "utypeinfo.h"  // for 'typeid' to work 
+
+#include "unicode/ustring.h"
+#include "unicode/strenum.h"
+#include "unicode/putil.h"
+#include "uenumimp.h"
+#include "ustrenum.h"
+#include "cstring.h"
+#include "cmemory.h"
+#include "uassert.h"
+
+U_NAMESPACE_BEGIN
+// StringEnumeration implementation ---------------------------------------- ***
+
+StringEnumeration::StringEnumeration()
+    : chars(charsBuffer), charsCapacity(sizeof(charsBuffer)) {
+}
+
+StringEnumeration::~StringEnumeration() {
+    if (chars != nullptr && chars != charsBuffer) {
+        uprv_free(chars);
+    }
+}
+
+// StringEnumeration base class clone() default implementation, does not clone
+StringEnumeration *
+StringEnumeration::clone() const {
+  return nullptr;
+}
+
+const char *
+StringEnumeration::next(int32_t *resultLength, UErrorCode &status) {
+    const UnicodeString *s=snext(status);
+    if(U_SUCCESS(status) && s!=nullptr) {
+        unistr=*s;
+        ensureCharsCapacity(unistr.length()+1, status);
+        if(U_SUCCESS(status)) {
+            if(resultLength!=nullptr) {
+                *resultLength=unistr.length();
+            }
+            unistr.extract(0, INT32_MAX, chars, charsCapacity, US_INV);
+            return chars;
+        }
+    }
+
+    return nullptr;
+}
+
+const char16_t *
+StringEnumeration::unext(int32_t *resultLength, UErrorCode &status) {
+    const UnicodeString *s=snext(status);
+    if(U_SUCCESS(status) && s!=nullptr) {
+        unistr=*s;
+        if(resultLength!=nullptr) {
+            *resultLength=unistr.length();
+        }
+        return unistr.getTerminatedBuffer();
+    }
+
+    return nullptr;
+}
+
+const UnicodeString *
+StringEnumeration::snext(UErrorCode &status) {
+    int32_t length;
+    const char *s=next(&length, status);
+    return setChars(s, length, status);
+}
+
+void
+StringEnumeration::ensureCharsCapacity(int32_t capacity, UErrorCode &status) {
+    if(U_SUCCESS(status) && capacity>charsCapacity) {
+        if(capacity<(charsCapacity+charsCapacity/2)) {
+            // avoid allocation thrashing
+            capacity=charsCapacity+charsCapacity/2;
+        }
+        if(chars!=charsBuffer) {
+            uprv_free(chars);
+        }
+        chars=(char *)uprv_malloc(capacity);
+        if(chars==nullptr) {
+            chars=charsBuffer;
+            charsCapacity=sizeof(charsBuffer);
+            status=U_MEMORY_ALLOCATION_ERROR;
+        } else {
+            charsCapacity=capacity;
+        }
+    }
+}
+
+UnicodeString *
+StringEnumeration::setChars(const char *s, int32_t length, UErrorCode &status) {
+    if(U_SUCCESS(status) && s!=nullptr) {
+        if(length<0) {
+            length=(int32_t)uprv_strlen(s);
+        }
+
+        char16_t *buffer=unistr.getBuffer(length+1);
+        if(buffer!=nullptr) {
+            u_charsToUChars(s, buffer, length);
+            buffer[length]=0;
+            unistr.releaseBuffer(length);
+            return &unistr;
+        } else {
+            status=U_MEMORY_ALLOCATION_ERROR;
+        }
+    }
+
+    return nullptr;
+}
+bool
+StringEnumeration::operator==(const StringEnumeration& that)const {
+    return typeid(*this) == typeid(that); 
+}
+
+bool
+StringEnumeration::operator!=(const StringEnumeration& that)const {
+    return !operator==(that);
+}
+
+// UStringEnumeration implementation --------------------------------------- ***
+
+UStringEnumeration * U_EXPORT2
+UStringEnumeration::fromUEnumeration(
+        UEnumeration *uenumToAdopt, UErrorCode &status) {
+    if (U_FAILURE(status)) {
+        uenum_close(uenumToAdopt);
+        return nullptr;
+    }
+    UStringEnumeration *result = new UStringEnumeration(uenumToAdopt);
+    if (result == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+        uenum_close(uenumToAdopt);
+        return nullptr;
+    }
+    return result;
+}
+
+UStringEnumeration::UStringEnumeration(UEnumeration* _uenum) :
+    uenum(_uenum) {
+    U_ASSERT(_uenum != 0);
+}
+
+UStringEnumeration::~UStringEnumeration() {
+    uenum_close(uenum);
+}
+
+int32_t UStringEnumeration::count(UErrorCode& status) const {
+    return uenum_count(uenum, &status);
+}
+
+const char *UStringEnumeration::next(int32_t *resultLength, UErrorCode &status) {
+    return uenum_next(uenum, resultLength, &status);
+}
+
+const UnicodeString* UStringEnumeration::snext(UErrorCode& status) {
+    int32_t length;
+    const char16_t* str = uenum_unext(uenum, &length, &status);
+    if (str == 0 || U_FAILURE(status)) {
+        return 0;
+    }
+    return &unistr.setTo(str, length);
+}
+
+void UStringEnumeration::reset(UErrorCode& status) {
+    uenum_reset(uenum, &status);
+}
+
+UOBJECT_DEFINE_RTTI_IMPLEMENTATION(UStringEnumeration)
+U_NAMESPACE_END
+
+// C wrapper --------------------------------------------------------------- ***
+
+#define THIS(en) ((icu::StringEnumeration*)(en->context))
+
+U_CDECL_BEGIN
+
+/**
+ * Wrapper API to make StringEnumeration look like UEnumeration.
+ */
+static void U_CALLCONV
+ustrenum_close(UEnumeration* en) {
+    delete THIS(en);
+    uprv_free(en);
+}
+
+/**
+ * Wrapper API to make StringEnumeration look like UEnumeration.
+ */
+static int32_t U_CALLCONV
+ustrenum_count(UEnumeration* en,
+               UErrorCode* ec)
+{
+    return THIS(en)->count(*ec);
+}
+
+/**
+ * Wrapper API to make StringEnumeration look like UEnumeration.
+ */
+static const char16_t* U_CALLCONV
+ustrenum_unext(UEnumeration* en,
+               int32_t* resultLength,
+               UErrorCode* ec)
+{
+    return THIS(en)->unext(resultLength, *ec);
+}
+
+/**
+ * Wrapper API to make StringEnumeration look like UEnumeration.
+ */
+static const char* U_CALLCONV
+ustrenum_next(UEnumeration* en,
+              int32_t* resultLength,
+              UErrorCode* ec)
+{
+    return THIS(en)->next(resultLength, *ec);
+}
+
+/**
+ * Wrapper API to make StringEnumeration look like UEnumeration.
+ */
+static void U_CALLCONV
+ustrenum_reset(UEnumeration* en,
+               UErrorCode* ec)
+{
+    THIS(en)->reset(*ec);
+}
+
+/**
+ * Pseudo-vtable for UEnumeration wrapper around StringEnumeration.
+ * The StringEnumeration pointer will be stored in 'context'.
+ */
+static const UEnumeration USTRENUM_VT = {
+    nullptr,
+    nullptr, // store StringEnumeration pointer here
+    ustrenum_close,
+    ustrenum_count,
+    ustrenum_unext,
+    ustrenum_next,
+    ustrenum_reset
+};
+
+U_CDECL_END
+
+/**
+ * Given a StringEnumeration, wrap it in a UEnumeration.  The
+ * StringEnumeration is adopted; after this call, the caller must not
+ * delete it (regardless of error status).
+ */
+U_CAPI UEnumeration* U_EXPORT2
+uenum_openFromStringEnumeration(icu::StringEnumeration* adopted, UErrorCode* ec) { 
+    UEnumeration* result = nullptr;
+    if (U_SUCCESS(*ec) && adopted != nullptr) {
+        result = (UEnumeration*) uprv_malloc(sizeof(UEnumeration));
+        if (result == nullptr) {
+            *ec = U_MEMORY_ALLOCATION_ERROR;
+        } else {
+            uprv_memcpy(result, &USTRENUM_VT, sizeof(USTRENUM_VT));
+            result->context = adopted;
+        }
+    }
+    if (result == nullptr) {
+        delete adopted;
+    }
+    return result;
+}
+
+// C wrapper --------------------------------------------------------------- ***
+
+U_CDECL_BEGIN
+
+typedef struct UCharStringEnumeration {
+    UEnumeration uenum;
+    int32_t index, count;
+} UCharStringEnumeration;
+
+static void U_CALLCONV
+ucharstrenum_close(UEnumeration* en) {
+    uprv_free(en);
+}
+
+static int32_t U_CALLCONV
+ucharstrenum_count(UEnumeration* en,
+                   UErrorCode* /*ec*/) {
+    return ((UCharStringEnumeration*)en)->count;
+}
+
+static const char16_t* U_CALLCONV
+ucharstrenum_unext(UEnumeration* en,
+                  int32_t* resultLength,
+                  UErrorCode* /*ec*/) {
+    UCharStringEnumeration *e = (UCharStringEnumeration*) en;
+    if (e->index >= e->count) {
+        return nullptr;
+    }
+    const char16_t* result = ((const char16_t**)e->uenum.context)[e->index++];
+    if (resultLength) {
+        *resultLength = (int32_t)u_strlen(result);
+    }
+    return result;
+}
+
+
+static const char* U_CALLCONV
+ucharstrenum_next(UEnumeration* en,
+                  int32_t* resultLength,
+                  UErrorCode* /*ec*/) {
+    UCharStringEnumeration *e = (UCharStringEnumeration*) en;
+    if (e->index >= e->count) {
+        return nullptr;
+    }
+    const char* result = ((const char**)e->uenum.context)[e->index++];
+    if (resultLength) {
+        *resultLength = (int32_t)uprv_strlen(result);
+    }
+    return result;
+}
+
+static void U_CALLCONV
+ucharstrenum_reset(UEnumeration* en,
+                   UErrorCode* /*ec*/) {
+    ((UCharStringEnumeration*)en)->index = 0;
+}
+
+static const UEnumeration UCHARSTRENUM_VT = {
+    nullptr,
+    nullptr, // store StringEnumeration pointer here
+    ucharstrenum_close,
+    ucharstrenum_count,
+    uenum_unextDefault,
+    ucharstrenum_next,
+    ucharstrenum_reset
+};
+
+static const UEnumeration UCHARSTRENUM_U_VT = {
+    nullptr,
+    nullptr, // store StringEnumeration pointer here
+    ucharstrenum_close,
+    ucharstrenum_count,
+    ucharstrenum_unext,
+    uenum_nextDefault,
+    ucharstrenum_reset
+};
+
+U_CDECL_END
+
+U_CAPI UEnumeration* U_EXPORT2
+uenum_openCharStringsEnumeration(const char* const strings[], int32_t count,
+                                 UErrorCode* ec) {
+    UCharStringEnumeration* result = nullptr;
+    if (U_SUCCESS(*ec) && count >= 0 && (count == 0 || strings != 0)) {
+        result = (UCharStringEnumeration*) uprv_malloc(sizeof(UCharStringEnumeration));
+        if (result == nullptr) {
+            *ec = U_MEMORY_ALLOCATION_ERROR;
+        } else {
+            U_ASSERT((char*)result==(char*)(&result->uenum));
+            uprv_memcpy(result, &UCHARSTRENUM_VT, sizeof(UCHARSTRENUM_VT));
+            result->uenum.context = (void*)strings;
+            result->index = 0;
+            result->count = count;
+        }
+    }
+    return (UEnumeration*) result;
+}
+
+U_CAPI UEnumeration* U_EXPORT2
+uenum_openUCharStringsEnumeration(const char16_t* const strings[], int32_t count,
+                                 UErrorCode* ec) {
+    UCharStringEnumeration* result = nullptr;
+    if (U_SUCCESS(*ec) && count >= 0 && (count == 0 || strings != 0)) {
+        result = (UCharStringEnumeration*) uprv_malloc(sizeof(UCharStringEnumeration));
+        if (result == nullptr) {
+            *ec = U_MEMORY_ALLOCATION_ERROR;
+        } else {
+            U_ASSERT((char*)result==(char*)(&result->uenum));
+            uprv_memcpy(result, &UCHARSTRENUM_U_VT, sizeof(UCHARSTRENUM_U_VT));
+            result->uenum.context = (void*)strings;
+            result->index = 0;
+            result->count = count;
+        }
+    }
+    return (UEnumeration*) result;
+}
+
+
+// end C Wrapper
diff --git a/intl/icu/source/common/ustrenum.h b/intl/icu/source/common/ustrenum.h
new file mode 100644
index 0000000000..ff546ca0a1
--- /dev/null
+++ b/intl/icu/source/common/ustrenum.h
@@ -0,0 +1,87 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+* Copyright (c) 2002-2014, International Business Machines
+* Corporation and others.  All Rights Reserved.
+**********************************************************************
+* Author: Alan Liu
+* Created: November 11 2002
+* Since: ICU 2.4
+**********************************************************************
+*/
+#ifndef _USTRENUM_H_
+#define _USTRENUM_H_
+
+#include "unicode/uenum.h"
+#include "unicode/strenum.h"
+
+//----------------------------------------------------------------------
+U_NAMESPACE_BEGIN
+
+/**
+ * A wrapper to make a UEnumeration into a StringEnumeration.  The
+ * wrapper adopts the UEnumeration is wraps.
+ */
+class U_COMMON_API UStringEnumeration : public StringEnumeration {
+
+public:
+    /**
+     * Constructor.  This constructor adopts its UEnumeration
+     * argument.
+     * @param uenum a UEnumeration object.  This object takes
+     * ownership of 'uenum' and will close it in its destructor.  The
+     * caller must not call uenum_close on 'uenum' after calling this
+     * constructor.
+     */
+    UStringEnumeration(UEnumeration* uenum);
+
+    /**
+     * Destructor.  This closes the UEnumeration passed in to the
+     * constructor.
+     */
+    virtual ~UStringEnumeration();
+
+    /**
+     * Return the number of elements that the iterator traverses.
+     * @param status the error code.
+     * @return number of elements in the iterator.
+     */
+    virtual int32_t count(UErrorCode& status) const override;
+
+    virtual const char* next(int32_t *resultLength, UErrorCode& status) override;
+
+    /**
+     * Returns the next element a UnicodeString*.  If there are no
+     * more elements, returns nullptr.
+     * @param status the error code.
+     * @return a pointer to the string, or nullptr.
+     */
+    virtual const UnicodeString* snext(UErrorCode& status) override;
+
+    /**
+     * Resets the iterator.
+     * @param status the error code.
+     */
+    virtual void reset(UErrorCode& status) override;
+
+    /**
+     * ICU4C "poor man's RTTI", returns a UClassID for the actual ICU class.
+     */
+    virtual UClassID getDynamicClassID() const override;
+
+    /**
+     * ICU4C "poor man's RTTI", returns a UClassID for this ICU class.
+     */
+    static UClassID U_EXPORT2 getStaticClassID();
+
+    static UStringEnumeration * U_EXPORT2 fromUEnumeration(
+            UEnumeration *enumToAdopt, UErrorCode &status);
+private:
+    UEnumeration *uenum; // owned
+};
+
+U_NAMESPACE_END
+
+#endif
+
diff --git a/intl/icu/source/common/ustrfmt.cpp b/intl/icu/source/common/ustrfmt.cpp
new file mode 100644
index 0000000000..3db9f35821
--- /dev/null
+++ b/intl/icu/source/common/ustrfmt.cpp
@@ -0,0 +1,59 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+*   Copyright (C) 2001-2006, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*/
+
+#include "cstring.h"
+#include "ustrfmt.h"
+
+
+/***
+ * Fills in a char16_t* string with the radix-based representation of a
+ * uint32_t number padded with zeroes to minwidth.  The result
+ * will be null terminated if there is room.
+ *
+ * @param buffer char16_t buffer to receive result
+ * @param capacity capacity of buffer
+ * @param i the unsigned number to be formatted
+ * @param radix the radix from 2..36
+ * @param minwidth the minimum width.  If the result is narrower than
+ *        this, '0's will be added on the left.  Must be <=
+ *        capacity.
+ * @return the length of the result, not including any terminating
+ *        null
+ */
+U_CAPI int32_t U_EXPORT2
+uprv_itou (char16_t * buffer, int32_t capacity,
+           uint32_t i, uint32_t radix, int32_t minwidth)
+{
+    int32_t length = 0;
+    int digit;
+    int32_t j;
+    char16_t temp;
+
+    do{
+        digit = (int)(i % radix);
+        buffer[length++]=(char16_t)(digit<=9?(0x0030+digit):(0x0030+digit+7));
+        i=i/radix;
+    } while(i && length= 0) {
+        return (char16_t *)string + idx;
+    } else {
+        return nullptr;
+    }
+}
+
+/* Search for a codepoint in a string that matches one of the matchSet codepoints. */
+U_CAPI int32_t U_EXPORT2
+u_strcspn(const char16_t *string, const char16_t *matchSet)
+{
+    int32_t idx = _matchFromSet(string, matchSet, true);
+    if(idx >= 0) {
+        return idx;
+    } else {
+        return -idx - 1; /* == u_strlen(string) */
+    }
+}
+
+/* Search for a codepoint in a string that does not match one of the matchSet codepoints. */
+U_CAPI int32_t U_EXPORT2
+u_strspn(const char16_t *string, const char16_t *matchSet)
+{
+    int32_t idx = _matchFromSet(string, matchSet, false);
+    if(idx >= 0) {
+        return idx;
+    } else {
+        return -idx - 1; /* == u_strlen(string) */
+    }
+}
+
+/* ----- Text manipulation functions --- */
+
+U_CAPI char16_t* U_EXPORT2
+u_strtok_r(char16_t *src,
+     const char16_t *delim,
+           char16_t   **saveState)
+{
+    char16_t *tokSource;
+    char16_t *nextToken;
+    uint32_t nonDelimIdx;
+
+    /* If saveState is nullptr, the user messed up. */
+    if (src != nullptr) {
+        tokSource = src;
+        *saveState = src; /* Set to "src" in case there are no delimiters */
+    }
+    else if (*saveState) {
+        tokSource = *saveState;
+    }
+    else {
+        /* src == nullptr && *saveState == nullptr */
+        /* This shouldn't happen. We already finished tokenizing. */
+        return nullptr;
+    }
+
+    /* Skip initial delimiters */
+    nonDelimIdx = u_strspn(tokSource, delim);
+    tokSource = &tokSource[nonDelimIdx];
+
+    if (*tokSource) {
+        nextToken = u_strpbrk(tokSource, delim);
+        if (nextToken != nullptr) {
+            /* Create a token */
+            *(nextToken++) = 0;
+            *saveState = nextToken;
+            return tokSource;
+        }
+        else if (*saveState) {
+            /* Return the last token */
+            *saveState = nullptr;
+            return tokSource;
+        }
+    }
+    else {
+        /* No tokens were found. Only delimiters were left. */
+        *saveState = nullptr;
+    }
+    return nullptr;
+}
+
+/* Miscellaneous functions -------------------------------------------------- */
+
+U_CAPI char16_t* U_EXPORT2
+u_strcat(char16_t  *dst,
+    const char16_t  *src)
+{
+    char16_t *anchor = dst;            /* save a pointer to start of dst */
+
+    while(*dst != 0) {              /* To end of first string          */
+        ++dst;
+    }
+    while((*(dst++) = *(src++)) != 0) {     /* copy string 2 over              */
+    }
+
+    return anchor;
+}
+
+U_CAPI char16_t*  U_EXPORT2
+u_strncat(char16_t  *dst,
+     const char16_t  *src,
+     int32_t     n ) 
+{
+    if(n > 0) {
+        char16_t *anchor = dst;            /* save a pointer to start of dst */
+
+        while(*dst != 0) {              /* To end of first string          */
+            ++dst;
+        }
+        while((*dst = *src) != 0) {     /* copy string 2 over              */
+            ++dst;
+            if(--n == 0) {
+                *dst = 0;
+                break;
+            }
+            ++src;
+        }
+
+        return anchor;
+    } else {
+        return dst;
+    }
+}
+
+/* ----- Text property functions --- */
+
+U_CAPI int32_t   U_EXPORT2
+u_strcmp(const char16_t *s1,
+    const char16_t *s2)
+{
+    char16_t  c1, c2;
+
+    for(;;) {
+        c1=*s1++;
+        c2=*s2++;
+        if (c1 != c2 || c1 == 0) {
+            break;
+        }
+    }
+    return (int32_t)c1 - (int32_t)c2;
+}
+
+U_CFUNC int32_t U_EXPORT2
+uprv_strCompare(const char16_t *s1, int32_t length1,
+                const char16_t *s2, int32_t length2,
+                UBool strncmpStyle, UBool codePointOrder) {
+    const char16_t *start1, *start2, *limit1, *limit2;
+    char16_t c1, c2;
+
+    /* setup for fix-up */
+    start1=s1;
+    start2=s2;
+
+    /* compare identical prefixes - they do not need to be fixed up */
+    if(length1<0 && length2<0) {
+        /* strcmp style, both NUL-terminated */
+        if(s1==s2) {
+            return 0;
+        }
+
+        for(;;) {
+            c1=*s1;
+            c2=*s2;
+            if(c1!=c2) {
+                break;
+            }
+            if(c1==0) {
+                return 0;
+            }
+            ++s1;
+            ++s2;
+        }
+
+        /* setup for fix-up */
+        limit1=limit2=nullptr;
+    } else if(strncmpStyle) {
+        /* special handling for strncmp, assume length1==length2>=0 but also check for NUL */
+        if(s1==s2) {
+            return 0;
+        }
+
+        limit1=start1+length1;
+
+        for(;;) {
+            /* both lengths are same, check only one limit */
+            if(s1==limit1) {
+                return 0;
+            }
+
+            c1=*s1;
+            c2=*s2;
+            if(c1!=c2) {
+                break;
+            }
+            if(c1==0) {
+                return 0;
+            }
+            ++s1;
+            ++s2;
+        }
+
+        /* setup for fix-up */
+        limit2=start2+length1; /* use length1 here, too, to enforce assumption */
+    } else {
+        /* memcmp/UnicodeString style, both length-specified */
+        int32_t lengthResult;
+
+        if(length1<0) {
+            length1=u_strlen(s1);
+        }
+        if(length2<0) {
+            length2=u_strlen(s2);
+        }
+
+        /* limit1=start1+min(length1, length2) */
+        if(length1length2 */ {
+            lengthResult=1;
+            limit1=start1+length2;
+        }
+
+        if(s1==s2) {
+            return lengthResult;
+        }
+
+        for(;;) {
+            /* check pseudo-limit */
+            if(s1==limit1) {
+                return lengthResult;
+            }
+
+            c1=*s1;
+            c2=*s2;
+            if(c1!=c2) {
+                break;
+            }
+            ++s1;
+            ++s2;
+        }
+
+        /* setup for fix-up */
+        limit1=start1+length1;
+        limit2=start2+length2;
+    }
+
+    /* if both values are in or above the surrogate range, fix them up */
+    if(c1>=0xd800 && c2>=0xd800 && codePointOrder) {
+        /* subtract 0x2800 from BMP code points to make them smaller than supplementary ones */
+        if(
+            (c1<=0xdbff && (s1+1)!=limit1 && U16_IS_TRAIL(*(s1+1))) ||
+            (U16_IS_TRAIL(c1) && start1!=s1 && U16_IS_LEAD(*(s1-1)))
+        ) {
+            /* part of a surrogate pair, leave >=d800 */
+        } else {
+            /* BMP code point - may be surrogate code point - make =d800 */
+        } else {
+            /* BMP code point - may be surrogate code point - make move(iter1, 0, UITER_START);
+    iter2->move(iter2, 0, UITER_START);
+
+    /* compare identical prefixes - they do not need to be fixed up */
+    for(;;) {
+        c1=iter1->next(iter1);
+        c2=iter2->next(iter2);
+        if(c1!=c2) {
+            break;
+        }
+        if(c1==-1) {
+            return 0;
+        }
+    }
+
+    /* if both values are in or above the surrogate range, fix them up */
+    if(c1>=0xd800 && c2>=0xd800 && codePointOrder) {
+        /* subtract 0x2800 from BMP code points to make them smaller than supplementary ones */
+        if(
+            (c1<=0xdbff && U16_IS_TRAIL(iter1->current(iter1))) ||
+            (U16_IS_TRAIL(c1) && (iter1->previous(iter1), U16_IS_LEAD(iter1->previous(iter1))))
+        ) {
+            /* part of a surrogate pair, leave >=d800 */
+        } else {
+            /* BMP code point - may be surrogate code point - make current(iter2))) ||
+            (U16_IS_TRAIL(c2) && (iter2->previous(iter2), U16_IS_LEAD(iter2->previous(iter2))))
+        ) {
+            /* part of a surrogate pair, leave >=d800 */
+        } else {
+            /* BMP code point - may be surrogate code point - make =0.
+ *
+ * Consistently leaving them _behind_ the different units is not an option
+ * because the current "unit" is the end of the string if that is reached,
+ * and in such a case the iterator does not move.
+ * For example, when comparing "ab" with "abc", both iterators rest _on_ the end
+ * of their strings. Calling previous() on each does not move them to where
+ * the comparison fails.
+ *
+ * So the simplest semantics is to not define where the iterators end up.
+ *
+ * The following fragment is part of what would need to be done for backing up.
+ */
+void fragment {
+        /* iff a surrogate is part of a surrogate pair, leave >=d800 */
+        if(c1<=0xdbff) {
+            if(!U16_IS_TRAIL(iter1->current(iter1))) {
+                /* lead surrogate code point - make getIndex(iter1, UITER_CURRENT);
+            iter1->previous(iter1); /* ==c1 */
+            if(!U16_IS_LEAD(iter1->previous(iter1))) {
+                /* trail surrogate code point - make move(iter1, idx, UITER_ZERO);
+        } else /* 0xe000<=c1<=0xffff */ {
+            /* BMP code point - make  0) {
+        int32_t rc;
+        for(;;) {
+            rc = (int32_t)*s1 - (int32_t)*s2;
+            if(rc != 0 || *s1 == 0 || --n == 0) {
+                return rc;
+            }
+            ++s1;
+            ++s2;
+        }
+    } else {
+        return 0;
+    }
+}
+
+U_CAPI int32_t U_EXPORT2
+u_strncmpCodePointOrder(const char16_t *s1, const char16_t *s2, int32_t n) {
+    return uprv_strCompare(s1, n, s2, n, true, true);
+}
+
+U_CAPI char16_t* U_EXPORT2
+u_strcpy(char16_t  *dst,
+    const char16_t  *src)
+{
+    char16_t *anchor = dst;            /* save a pointer to start of dst */
+
+    while((*(dst++) = *(src++)) != 0) {     /* copy string 2 over              */
+    }
+
+    return anchor;
+}
+
+U_CAPI char16_t*  U_EXPORT2
+u_strncpy(char16_t  *dst,
+     const char16_t  *src,
+     int32_t     n) 
+{
+    char16_t *anchor = dst;            /* save a pointer to start of dst */
+
+    /* copy string 2 over */
+    while(n > 0 && (*(dst++) = *(src++)) != 0) {
+        --n;
+    }
+
+    return anchor;
+}
+
+U_CAPI int32_t   U_EXPORT2
+u_strlen(const char16_t *s)
+{
+#if U_SIZEOF_WCHAR_T == U_SIZEOF_UCHAR
+    return (int32_t)uprv_wcslen((const wchar_t *)s);
+#else
+    const char16_t *t = s;
+    while(*t != 0) {
+      ++t;
+    }
+    return t - s;
+#endif
+}
+
+U_CAPI int32_t U_EXPORT2
+u_countChar32(const char16_t *s, int32_t length) {
+    int32_t count;
+
+    if(s==nullptr || length<-1) {
+        return 0;
+    }
+
+    count=0;
+    if(length>=0) {
+        while(length>0) {
+            ++count;
+            if(U16_IS_LEAD(*s) && length>=2 && U16_IS_TRAIL(*(s+1))) {
+                s+=2;
+                length-=2;
+            } else {
+                ++s;
+                --length;
+            }
+        }
+    } else /* length==-1 */ {
+        char16_t c;
+
+        for(;;) {
+            if((c=*s++)==0) {
+                break;
+            }
+            ++count;
+
+            /*
+             * sufficient to look ahead one because of UTF-16;
+             * safe to look ahead one because at worst that would be the terminating NUL
+             */
+            if(U16_IS_LEAD(c) && U16_IS_TRAIL(*s)) {
+                ++s;
+            }
+        }
+    }
+    return count;
+}
+
+U_CAPI UBool U_EXPORT2
+u_strHasMoreChar32Than(const char16_t *s, int32_t length, int32_t number) {
+
+    if(number<0) {
+        return true;
+    }
+    if(s==nullptr || length<-1) {
+        return false;
+    }
+
+    if(length==-1) {
+        /* s is NUL-terminated */
+        char16_t c;
+
+        /* count code points until they exceed */
+        for(;;) {
+            if((c=*s++)==0) {
+                return false;
+            }
+            if(number==0) {
+                return true;
+            }
+            if(U16_IS_LEAD(c) && U16_IS_TRAIL(*s)) {
+                ++s;
+            }
+            --number;
+        }
+    } else {
+        /* length>=0 known */
+        const char16_t *limit;
+        int32_t maxSupplementary;
+
+        /* s contains at least (length+1)/2 code points: <=2 UChars per cp */
+        if(((length+1)/2)>number) {
+            return true;
+        }
+
+        /* check if s does not even contain enough UChars */
+        maxSupplementary=length-number;
+        if(maxSupplementary<=0) {
+            return false;
+        }
+        /* there are maxSupplementary=length-number more UChars than asked-for code points */
+
+        /*
+         * count code points until they exceed and also check that there are
+         * no more than maxSupplementary supplementary code points (char16_t pairs)
+         */
+        limit=s+length;
+        for(;;) {
+            if(s==limit) {
+                return false;
+            }
+            if(number==0) {
+                return true;
+            }
+            if(U16_IS_LEAD(*s++) && s!=limit && U16_IS_TRAIL(*s)) {
+                ++s;
+                if(--maxSupplementary<=0) {
+                    /* too many pairs - too few code points */
+                    return false;
+                }
+            }
+            --number;
+        }
+    }
+}
+
+U_CAPI char16_t * U_EXPORT2
+u_memcpy(char16_t *dest, const char16_t *src, int32_t count) {
+    if(count > 0) {
+        uprv_memcpy(dest, src, (size_t)count*U_SIZEOF_UCHAR);
+    }
+    return dest;
+}
+
+U_CAPI char16_t * U_EXPORT2
+u_memmove(char16_t *dest, const char16_t *src, int32_t count) {
+    if(count > 0) {
+        uprv_memmove(dest, src, (size_t)count*U_SIZEOF_UCHAR);
+    }
+    return dest;
+}
+
+U_CAPI char16_t * U_EXPORT2
+u_memset(char16_t *dest, char16_t c, int32_t count) {
+    if(count > 0) {
+        char16_t *ptr = dest;
+        char16_t *limit = dest + count;
+
+        while (ptr < limit) {
+            *(ptr++) = c;
+        }
+    }
+    return dest;
+}
+
+U_CAPI int32_t U_EXPORT2
+u_memcmp(const char16_t *buf1, const char16_t *buf2, int32_t count) {
+    if(count > 0) {
+        const char16_t *limit = buf1 + count;
+        int32_t result;
+
+        while (buf1 < limit) {
+            result = (int32_t)(uint16_t)*buf1 - (int32_t)(uint16_t)*buf2;
+            if (result != 0) {
+                return result;
+            }
+            buf1++;
+            buf2++;
+        }
+    }
+    return 0;
+}
+
+U_CAPI int32_t U_EXPORT2
+u_memcmpCodePointOrder(const char16_t *s1, const char16_t *s2, int32_t count) {
+    return uprv_strCompare(s1, count, s2, count, false, true);
+}
+
+/* u_unescape & support fns ------------------------------------------------- */
+
+/* This map must be in ASCENDING ORDER OF THE ESCAPE CODE */
+static const char16_t UNESCAPE_MAP[] = {
+    /*"   0x22, 0x22 */
+    /*'   0x27, 0x27 */
+    /*?   0x3F, 0x3F */
+    /*\   0x5C, 0x5C */
+    /*a*/ 0x61, 0x07,
+    /*b*/ 0x62, 0x08,
+    /*e*/ 0x65, 0x1b,
+    /*f*/ 0x66, 0x0c,
+    /*n*/ 0x6E, 0x0a,
+    /*r*/ 0x72, 0x0d,
+    /*t*/ 0x74, 0x09,
+    /*v*/ 0x76, 0x0b
+};
+enum { UNESCAPE_MAP_LENGTH = UPRV_LENGTHOF(UNESCAPE_MAP) };
+
+/* Convert one octal digit to a numeric value 0..7, or -1 on failure */
+static int32_t _digit8(char16_t c) {
+    if (c >= u'0' && c <= u'7') {
+        return c - u'0';
+    }
+    return -1;
+}
+
+/* Convert one hex digit to a numeric value 0..F, or -1 on failure */
+static int32_t _digit16(char16_t c) {
+    if (c >= u'0' && c <= u'9') {
+        return c - u'0';
+    }
+    if (c >= u'A' && c <= u'F') {
+        return c - (u'A' - 10);
+    }
+    if (c >= u'a' && c <= u'f') {
+        return c - (u'a' - 10);
+    }
+    return -1;
+}
+
+/* Parse a single escape sequence.  Although this method deals in
+ * UChars, it does not use C++ or UnicodeString.  This allows it to
+ * be used from C contexts. */
+U_CAPI UChar32 U_EXPORT2
+u_unescapeAt(UNESCAPE_CHAR_AT charAt,
+             int32_t *offset,
+             int32_t length,
+             void *context) {
+
+    int32_t start = *offset;
+    UChar32 c;
+    UChar32 result = 0;
+    int8_t n = 0;
+    int8_t minDig = 0;
+    int8_t maxDig = 0;
+    int8_t bitsPerDigit = 4; 
+    int32_t dig;
+    UBool braces = false;
+
+    /* Check that offset is in range */
+    if (*offset < 0 || *offset >= length) {
+        goto err;
+    }
+
+    /* Fetch first char16_t after '\\' */
+    c = charAt((*offset)++, context);
+
+    /* Convert hexadecimal and octal escapes */
+    switch (c) {
+    case u'u':
+        minDig = maxDig = 4;
+        break;
+    case u'U':
+        minDig = maxDig = 8;
+        break;
+    case u'x':
+        minDig = 1;
+        if (*offset < length && charAt(*offset, context) == u'{') {
+            ++(*offset);
+            braces = true;
+            maxDig = 8;
+        } else {
+            maxDig = 2;
+        }
+        break;
+    default:
+        dig = _digit8(c);
+        if (dig >= 0) {
+            minDig = 1;
+            maxDig = 3;
+            n = 1; /* Already have first octal digit */
+            bitsPerDigit = 3;
+            result = dig;
+        }
+        break;
+    }
+    if (minDig != 0) {
+        while (*offset < length && n < maxDig) {
+            c = charAt(*offset, context);
+            dig = (bitsPerDigit == 3) ? _digit8(c) : _digit16(c);
+            if (dig < 0) {
+                break;
+            }
+            result = (result << bitsPerDigit) | dig;
+            ++(*offset);
+            ++n;
+        }
+        if (n < minDig) {
+            goto err;
+        }
+        if (braces) {
+            if (c != u'}') {
+                goto err;
+            }
+            ++(*offset);
+        }
+        if (result < 0 || result >= 0x110000) {
+            goto err;
+        }
+        /* If an escape sequence specifies a lead surrogate, see if
+         * there is a trail surrogate after it, either as an escape or
+         * as a literal.  If so, join them up into a supplementary.
+         */
+        if (*offset < length && U16_IS_LEAD(result)) {
+            int32_t ahead = *offset + 1;
+            c = charAt(*offset, context);
+            if (c == u'\\' && ahead < length) {
+                // Calling ourselves recursively may cause a stack overflow if
+                // we have repeated escaped lead surrogates.
+                // Limit the length to 11 ("x{0000DFFF}") after ahead.
+                int32_t tailLimit = ahead + 11;
+                if (tailLimit > length) {
+                    tailLimit = length;
+                }
+                c = u_unescapeAt(charAt, &ahead, tailLimit, context);
+            }
+            if (U16_IS_TRAIL(c)) {
+                *offset = ahead;
+                result = U16_GET_SUPPLEMENTARY(result, c);
+            }
+        }
+        return result;
+    }
+
+    /* Convert C-style escapes in table */
+    for (int32_t i=0; i destCapacity) {
+        srcLen = destCapacity;
+    }
+    u_charsToUChars(src, dest, srcLen);
+}
+
+/* Do an invariant conversion of char* -> char16_t*, with escape parsing */
+U_CAPI int32_t U_EXPORT2
+u_unescape(const char *src, char16_t *dest, int32_t destCapacity) {
+    const char *segment = src;
+    int32_t i = 0;
+    char c;
+
+    while ((c=*src) != 0) {
+        /* '\\' intentionally written as compiler-specific
+         * character constant to correspond to compiler-specific
+         * char* constants. */
+        if (c == '\\') {
+            int32_t lenParsed = 0;
+            UChar32 c32;
+            if (src != segment) {
+                if (dest != nullptr) {
+                    _appendUChars(dest + i, destCapacity - i,
+                                  segment, (int32_t)(src - segment));
+                }
+                i += (int32_t)(src - segment);
+            }
+            ++src; /* advance past '\\' */
+            c32 = (UChar32)u_unescapeAt(_charPtr_charAt, &lenParsed, (int32_t)uprv_strlen(src), (void*)src);
+            if (lenParsed == 0) {
+                goto err;
+            }
+            src += lenParsed; /* advance past escape seq. */
+            if (dest != nullptr && U16_LENGTH(c32) <= (destCapacity - i)) {
+                U16_APPEND_UNSAFE(dest, i, c32);
+            } else {
+                i += U16_LENGTH(c32);
+            }
+            segment = src;
+        } else {
+            ++src;
+        }
+    }
+    if (src != segment) {
+        if (dest != nullptr) {
+            _appendUChars(dest + i, destCapacity - i,
+                          segment, (int32_t)(src - segment));
+        }
+        i += (int32_t)(src - segment);
+    }
+    if (dest != nullptr && i < destCapacity) {
+        dest[i] = 0;
+    }
+    return i;
+
+ err:
+    if (dest != nullptr && destCapacity > 0) {
+        *dest = 0;
+    }
+    return 0;
+}
+
+/* NUL-termination of strings ----------------------------------------------- */
+
+/**
+ * NUL-terminate a string no matter what its type.
+ * Set warning and error codes accordingly.
+ */
+#define __TERMINATE_STRING(dest, destCapacity, length, pErrorCode) UPRV_BLOCK_MACRO_BEGIN { \
+    if(pErrorCode!=nullptr && U_SUCCESS(*pErrorCode)) {                    \
+        /* not a public function, so no complete argument checking */   \
+                                                                        \
+        if(length<0) {                                                  \
+            /* assume that the caller handles this */                   \
+        } else if(lengthdestCapacity */ {                              \
+            /* even the string itself did not fit - set an error code */ \
+            *pErrorCode=U_BUFFER_OVERFLOW_ERROR;                        \
+        }                                                               \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+U_CAPI char16_t U_EXPORT2
+u_asciiToUpper(char16_t c) {
+    if (u'a' <= c && c <= u'z') {
+        c = c + u'A' - u'a';
+    }
+    return c;
+}
+
+U_CAPI int32_t U_EXPORT2
+u_terminateUChars(char16_t *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode) {
+    __TERMINATE_STRING(dest, destCapacity, length, pErrorCode);
+    return length;
+}
+
+U_CAPI int32_t U_EXPORT2
+u_terminateChars(char *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode) {
+    __TERMINATE_STRING(dest, destCapacity, length, pErrorCode);
+    return length;
+}
+
+U_CAPI int32_t U_EXPORT2
+u_terminateUChar32s(UChar32 *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode) {
+    __TERMINATE_STRING(dest, destCapacity, length, pErrorCode);
+    return length;
+}
+
+U_CAPI int32_t U_EXPORT2
+u_terminateWChars(wchar_t *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode) {
+    __TERMINATE_STRING(dest, destCapacity, length, pErrorCode);
+    return length;
+}
+
+// Compute the hash code for a string -------------------------------------- ***
+
+// Moved here from uhash.c so that UnicodeString::hashCode() does not depend
+// on UHashtable code.
+
+/*
+  Compute the hash by iterating sparsely over about 32 (up to 63)
+  characters spaced evenly through the string.  For each character,
+  multiply the previous hash value by a prime number and add the new
+  character in, like a linear congruential random number generator,
+  producing a pseudorandom deterministic value well distributed over
+  the output range. [LIU]
+*/
+
+#define STRING_HASH(TYPE, STR, STRLEN, DEREF) UPRV_BLOCK_MACRO_BEGIN { \
+    uint32_t hash = 0;                        \
+    const TYPE *p = (const TYPE*) STR;        \
+    if (p != nullptr) {                          \
+        int32_t len = (int32_t)(STRLEN);      \
+        int32_t inc = ((len - 32) / 32) + 1;  \
+        const TYPE *limit = p + len;          \
+        while (p(hash);        \
+} UPRV_BLOCK_MACRO_END
+
+/* Used by UnicodeString to compute its hashcode - Not public API. */
+U_CAPI int32_t U_EXPORT2
+ustr_hashUCharsN(const char16_t *str, int32_t length) {
+    STRING_HASH(char16_t, str, length, *p);
+}
+
+U_CAPI int32_t U_EXPORT2
+ustr_hashCharsN(const char *str, int32_t length) {
+    STRING_HASH(uint8_t, str, length, *p);
+}
+
+U_CAPI int32_t U_EXPORT2
+ustr_hashICharsN(const char *str, int32_t length) {
+    STRING_HASH(char, str, length, (uint8_t)uprv_tolower(*p));
+}
diff --git a/intl/icu/source/common/ustrtrns.cpp b/intl/icu/source/common/ustrtrns.cpp
new file mode 100644
index 0000000000..244c111fc4
--- /dev/null
+++ b/intl/icu/source/common/ustrtrns.cpp
@@ -0,0 +1,1451 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+*
+*   Copyright (C) 2001-2016, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+******************************************************************************
+*
+* File ustrtrns.cpp
+*
+* Modification History:
+*
+*   Date        Name        Description
+*   9/10/2001    Ram    Creation.
+******************************************************************************
+*/
+
+/*******************************************************************************
+ *
+ * u_strTo* and u_strFrom* APIs
+ * WCS functions moved to ustr_wcs.c for better modularization
+ *
+ *******************************************************************************
+ */
+
+
+#include "unicode/putil.h"
+#include "unicode/ustring.h"
+#include "unicode/utf.h"
+#include "unicode/utf8.h"
+#include "unicode/utf16.h"
+#include "cstring.h"
+#include "cmemory.h"
+#include "ustr_imp.h"
+#include "uassert.h"
+
+U_CAPI char16_t* U_EXPORT2
+u_strFromUTF32WithSub(char16_t *dest,
+               int32_t destCapacity,
+               int32_t *pDestLength,
+               const UChar32 *src,
+               int32_t srcLength,
+               UChar32 subchar, int32_t *pNumSubstitutions,
+               UErrorCode *pErrorCode) {
+    const UChar32 *srcLimit;
+    UChar32 ch;
+    char16_t *destLimit;
+    char16_t *pDest;
+    int32_t reqLength;
+    int32_t numSubstitutions;
+
+    /* args check */
+    if(U_FAILURE(*pErrorCode)){
+        return nullptr;
+    }
+    if( (src==nullptr && srcLength!=0) || srcLength < -1 ||
+        (destCapacity<0) || (dest == nullptr && destCapacity > 0) ||
+        subchar > 0x10ffff || U_IS_SURROGATE(subchar)
+    ) {
+        *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+
+    if(pNumSubstitutions != nullptr) {
+        *pNumSubstitutions = 0;
+    }
+
+    pDest = dest;
+    destLimit = (dest!=nullptr)?(dest + destCapacity):nullptr;
+    reqLength = 0;
+    numSubstitutions = 0;
+
+    if(srcLength < 0) {
+        /* simple loop for conversion of a NUL-terminated BMP string */
+        while((ch=*src) != 0 &&
+              ((uint32_t)ch < 0xd800 || (0xe000 <= ch && ch <= 0xffff))) {
+            ++src;
+            if(pDest < destLimit) {
+                *pDest++ = (char16_t)ch;
+            } else {
+                ++reqLength;
+            }
+        }
+        srcLimit = src;
+        if(ch != 0) {
+            /* "complicated" case, find the end of the remaining string */
+            while(*++srcLimit != 0) {}
+        }
+    } else {
+      srcLimit = (src!=nullptr)?(src + srcLength):nullptr;
+    }
+
+    /* convert with length */
+    while(src < srcLimit) {
+        ch = *src++;
+        do {
+            /* usually "loops" once; twice only for writing subchar */
+            if((uint32_t)ch < 0xd800 || (0xe000 <= ch && ch <= 0xffff)) {
+                if(pDest < destLimit) {
+                    *pDest++ = (char16_t)ch;
+                } else {
+                    ++reqLength;
+                }
+                break;
+            } else if(0x10000 <= ch && ch <= 0x10ffff) {
+                if(pDest!=nullptr && ((pDest + 2) <= destLimit)) {
+                    *pDest++ = U16_LEAD(ch);
+                    *pDest++ = U16_TRAIL(ch);
+                } else {
+                    reqLength += 2;
+                }
+                break;
+            } else if((ch = subchar) < 0) {
+                /* surrogate code point, or not a Unicode code point at all */
+                *pErrorCode = U_INVALID_CHAR_FOUND;
+                return nullptr;
+            } else {
+                ++numSubstitutions;
+            }
+        } while(true);
+    }
+
+    reqLength += (int32_t)(pDest - dest);
+    if(pDestLength) {
+        *pDestLength = reqLength;
+    }
+    if(pNumSubstitutions != nullptr) {
+        *pNumSubstitutions = numSubstitutions;
+    }
+
+    /* Terminate the buffer */
+    u_terminateUChars(dest, destCapacity, reqLength, pErrorCode);
+    
+    return dest;
+}
+
+U_CAPI char16_t* U_EXPORT2
+u_strFromUTF32(char16_t *dest,
+               int32_t destCapacity, 
+               int32_t *pDestLength,
+               const UChar32 *src,
+               int32_t srcLength,
+               UErrorCode *pErrorCode) {
+    return u_strFromUTF32WithSub(
+            dest, destCapacity, pDestLength,
+            src, srcLength,
+            U_SENTINEL, nullptr,
+            pErrorCode);
+}
+
+U_CAPI UChar32* U_EXPORT2 
+u_strToUTF32WithSub(UChar32 *dest,
+             int32_t destCapacity,
+             int32_t *pDestLength,
+             const char16_t *src,
+             int32_t srcLength,
+             UChar32 subchar, int32_t *pNumSubstitutions,
+             UErrorCode *pErrorCode) {
+    const char16_t *srcLimit;
+    UChar32 ch;
+    char16_t ch2;
+    UChar32 *destLimit;
+    UChar32 *pDest;
+    int32_t reqLength;
+    int32_t numSubstitutions;
+
+    /* args check */
+    if(U_FAILURE(*pErrorCode)){
+        return nullptr;
+    }
+    if( (src==nullptr && srcLength!=0) || srcLength < -1 ||
+        (destCapacity<0) || (dest == nullptr && destCapacity > 0) ||
+        subchar > 0x10ffff || U_IS_SURROGATE(subchar)
+    ) {
+        *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+
+    if(pNumSubstitutions != nullptr) {
+        *pNumSubstitutions = 0;
+    }
+
+    pDest = dest;
+    destLimit = (dest!=nullptr)?(dest + destCapacity):nullptr;
+    reqLength = 0;
+    numSubstitutions = 0;
+
+    if(srcLength < 0) {
+        /* simple loop for conversion of a NUL-terminated BMP string */
+        while((ch=*src) != 0 && !U16_IS_SURROGATE(ch)) {
+            ++src;
+            if(pDest < destLimit) {
+                *pDest++ = ch;
+            } else {
+                ++reqLength;
+            }
+        }
+        srcLimit = src;
+        if(ch != 0) {
+            /* "complicated" case, find the end of the remaining string */
+            while(*++srcLimit != 0) {}
+        }
+    } else {
+        srcLimit = (src!=nullptr)?(src + srcLength):nullptr;
+    }
+
+    /* convert with length */
+    while(src < srcLimit) {
+        ch = *src++;
+        if(!U16_IS_SURROGATE(ch)) {
+            /* write or count ch below */
+        } else if(U16_IS_SURROGATE_LEAD(ch) && src < srcLimit && U16_IS_TRAIL(ch2 = *src)) {
+            ++src;
+            ch = U16_GET_SUPPLEMENTARY(ch, ch2);
+        } else if((ch = subchar) < 0) {
+            /* unpaired surrogate */
+            *pErrorCode = U_INVALID_CHAR_FOUND;
+            return nullptr;
+        } else {
+            ++numSubstitutions;
+        }
+        if(pDest < destLimit) {
+            *pDest++ = ch;
+        } else {
+            ++reqLength;
+        }
+    }
+
+    reqLength += (int32_t)(pDest - dest);
+    if(pDestLength) {
+        *pDestLength = reqLength;
+    }
+    if(pNumSubstitutions != nullptr) {
+        *pNumSubstitutions = numSubstitutions;
+    }
+
+    /* Terminate the buffer */
+    u_terminateUChar32s(dest, destCapacity, reqLength, pErrorCode);
+
+    return dest;
+}
+
+U_CAPI UChar32* U_EXPORT2 
+u_strToUTF32(UChar32 *dest, 
+             int32_t destCapacity,
+             int32_t *pDestLength,
+             const char16_t *src,
+             int32_t srcLength,
+             UErrorCode *pErrorCode) {
+    return u_strToUTF32WithSub(
+            dest, destCapacity, pDestLength,
+            src, srcLength,
+            U_SENTINEL, nullptr,
+            pErrorCode);
+}
+
+U_CAPI char16_t* U_EXPORT2
+u_strFromUTF8WithSub(char16_t *dest,
+              int32_t destCapacity,
+              int32_t *pDestLength,
+              const char* src,
+              int32_t srcLength,
+              UChar32 subchar, int32_t *pNumSubstitutions,
+              UErrorCode *pErrorCode){
+    /* args check */
+    if(U_FAILURE(*pErrorCode)) {
+        return nullptr;
+    }
+    if( (src==nullptr && srcLength!=0) || srcLength < -1 ||
+        (destCapacity<0) || (dest == nullptr && destCapacity > 0) ||
+        subchar > 0x10ffff || U_IS_SURROGATE(subchar)
+    ) {
+        *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+
+    if(pNumSubstitutions!=nullptr) {
+        *pNumSubstitutions=0;
+    }
+    char16_t *pDest = dest;
+    char16_t *pDestLimit = dest+destCapacity;
+    int32_t reqLength = 0;
+    int32_t numSubstitutions=0;
+
+    /*
+     * Inline processing of UTF-8 byte sequences:
+     *
+     * Byte sequences for the most common characters are handled inline in
+     * the conversion loops. In order to reduce the path lengths for those
+     * characters, the tests are arranged in a kind of binary search.
+     * ASCII (<=0x7f) is checked first, followed by the dividing point
+     * between 2- and 3-byte sequences (0xe0).
+     * The 3-byte branch is tested first to speed up CJK text.
+     * The compiler should combine the subtractions for the two tests for 0xe0.
+     * Each branch then tests for the other end of its range.
+     */
+
+    if(srcLength < 0){
+        /*
+         * Transform a NUL-terminated string.
+         * The code explicitly checks for NULs only in the lead byte position.
+         * A NUL byte in the trail byte position fails the trail byte range check anyway.
+         */
+        int32_t i;
+        UChar32 c;
+        for(i = 0; (c = (uint8_t)src[i]) != 0 && (pDest < pDestLimit);) {
+            // modified copy of U8_NEXT()
+            ++i;
+            if(U8_IS_SINGLE(c)) {
+                *pDest++=(char16_t)c;
+            } else {
+                uint8_t __t1, __t2;
+                if( /* handle U+0800..U+FFFF inline */
+                        (0xe0<=(c) && (c)<0xf0) &&
+                        U8_IS_VALID_LEAD3_AND_T1((c), src[i]) &&
+                        (__t2=src[(i)+1]-0x80)<=0x3f) {
+                    *pDest++ = (((c)&0xf)<<12)|((src[i]&0x3f)<<6)|__t2;
+                    i+=2;
+                } else if( /* handle U+0080..U+07FF inline */
+                        ((c)<0xe0 && (c)>=0xc2) &&
+                        (__t1=src[i]-0x80)<=0x3f) {
+                    *pDest++ = (((c)&0x1f)<<6)|__t1;
+                    ++(i);
+                } else {
+                    /* function call for "complicated" and error cases */
+                    (c)=utf8_nextCharSafeBody((const uint8_t *)src, &(i), -1, c, -1);
+                    if(c<0 && (++numSubstitutions, c = subchar) < 0) {
+                        *pErrorCode = U_INVALID_CHAR_FOUND;
+                        return nullptr;
+                    } else if(c<=0xFFFF) {
+                        *(pDest++)=(char16_t)c;
+                    } else {
+                        *(pDest++)=U16_LEAD(c);
+                        if(pDest=0xc2) &&
+                        (__t1=src[i]-0x80)<=0x3f) {
+                    ++reqLength;
+                    ++(i);
+                } else {
+                    /* function call for "complicated" and error cases */
+                    (c)=utf8_nextCharSafeBody((const uint8_t *)src, &(i), -1, c, -1);
+                    if(c<0 && (++numSubstitutions, c = subchar) < 0) {
+                        *pErrorCode = U_INVALID_CHAR_FOUND;
+                        return nullptr;
+                    }
+                    reqLength += U16_LENGTH(c);
+                }
+            }
+        }
+    } else /* srcLength >= 0 */ {
+        /* Faster loop without ongoing checking for srcLength and pDestLimit. */
+        int32_t i = 0;
+        UChar32 c;
+        for(;;) {
+            /*
+             * Each iteration of the inner loop progresses by at most 3 UTF-8
+             * bytes and one char16_t, for most characters.
+             * For supplementary code points (4 & 2), which are rare,
+             * there is an additional adjustment.
+             */
+            int32_t count = (int32_t)(pDestLimit - pDest);
+            int32_t count2 = (srcLength - i) / 3;
+            if(count > count2) {
+                count = count2; /* min(remaining dest, remaining src/3) */
+            }
+            if(count < 3) {
+                /*
+                 * Too much overhead if we get near the end of the string,
+                 * continue with the next loop.
+                 */
+                break;
+            }
+
+            do {
+                // modified copy of U8_NEXT()
+                c = (uint8_t)src[i++];
+                if(U8_IS_SINGLE(c)) {
+                    *pDest++=(char16_t)c;
+                } else {
+                    uint8_t __t1, __t2;
+                    if( /* handle U+0800..U+FFFF inline */
+                            (0xe0<=(c) && (c)<0xf0) &&
+                            ((i)+1)=0xc2) &&
+                            ((i)!=srcLength) &&
+                            (__t1=src[i]-0x80)<=0x3f) {
+                        *pDest++ = (((c)&0x1f)<<6)|__t1;
+                        ++(i);
+                    } else {
+                        if(c >= 0xf0 || subchar > 0xffff) {
+                            // We may read up to four bytes and write up to two UChars,
+                            // which we didn't account for with computing count,
+                            // so we adjust it here.
+                            if(--count == 0) {
+                                --i;  // back out byte c
+                                break;
+                            }
+                        }
+
+                        /* function call for "complicated" and error cases */
+                        (c)=utf8_nextCharSafeBody((const uint8_t *)src, &(i), srcLength, c, -1);
+                        if(c<0 && (++numSubstitutions, c = subchar) < 0) {
+                            *pErrorCode = U_INVALID_CHAR_FOUND;
+                            return nullptr;
+                        } else if(c<=0xFFFF) {
+                            *(pDest++)=(char16_t)c;
+                        } else {
+                            *(pDest++)=U16_LEAD(c);
+                            *(pDest++)=U16_TRAIL(c);
+                        }
+                    }
+                }
+            } while(--count > 0);
+        }
+
+        while(i < srcLength && (pDest < pDestLimit)) {
+            // modified copy of U8_NEXT()
+            c = (uint8_t)src[i++];
+            if(U8_IS_SINGLE(c)) {
+                *pDest++=(char16_t)c;
+            } else {
+                uint8_t __t1, __t2;
+                if( /* handle U+0800..U+FFFF inline */
+                        (0xe0<=(c) && (c)<0xf0) &&
+                        ((i)+1)=0xc2) &&
+                        ((i)!=srcLength) &&
+                        (__t1=src[i]-0x80)<=0x3f) {
+                    *pDest++ = (((c)&0x1f)<<6)|__t1;
+                    ++(i);
+                } else {
+                    /* function call for "complicated" and error cases */
+                    (c)=utf8_nextCharSafeBody((const uint8_t *)src, &(i), srcLength, c, -1);
+                    if(c<0 && (++numSubstitutions, c = subchar) < 0) {
+                        *pErrorCode = U_INVALID_CHAR_FOUND;
+                        return nullptr;
+                    } else if(c<=0xFFFF) {
+                        *(pDest++)=(char16_t)c;
+                    } else {
+                        *(pDest++)=U16_LEAD(c);
+                        if(pDest=0xc2) &&
+                        ((i)!=srcLength) &&
+                        (__t1=src[i]-0x80)<=0x3f) {
+                    ++reqLength;
+                    ++(i);
+                } else {
+                    /* function call for "complicated" and error cases */
+                    (c)=utf8_nextCharSafeBody((const uint8_t *)src, &(i), srcLength, c, -1);
+                    if(c<0 && (++numSubstitutions, c = subchar) < 0) {
+                        *pErrorCode = U_INVALID_CHAR_FOUND;
+                        return nullptr;
+                    }
+                    reqLength += U16_LENGTH(c);
+                }
+            }
+        }
+    }
+
+    reqLength+=(int32_t)(pDest - dest);
+
+    if(pNumSubstitutions!=nullptr) {
+        *pNumSubstitutions=numSubstitutions;
+    }
+
+    if(pDestLength){
+        *pDestLength = reqLength;
+    }
+
+    /* Terminate the buffer */
+    u_terminateUChars(dest,destCapacity,reqLength,pErrorCode);
+
+    return dest;
+}
+
+U_CAPI char16_t* U_EXPORT2
+u_strFromUTF8(char16_t *dest,
+              int32_t destCapacity,
+              int32_t *pDestLength,
+              const char* src,
+              int32_t srcLength,
+              UErrorCode *pErrorCode){
+    return u_strFromUTF8WithSub(
+            dest, destCapacity, pDestLength,
+            src, srcLength,
+            U_SENTINEL, nullptr,
+            pErrorCode);
+}
+
+U_CAPI char16_t * U_EXPORT2
+u_strFromUTF8Lenient(char16_t *dest,
+                     int32_t destCapacity,
+                     int32_t *pDestLength,
+                     const char *src,
+                     int32_t srcLength,
+                     UErrorCode *pErrorCode) {
+    char16_t *pDest = dest;
+    UChar32 ch;
+    int32_t reqLength = 0;
+    uint8_t* pSrc = (uint8_t*) src;
+
+    /* args check */
+    if(U_FAILURE(*pErrorCode)){
+        return nullptr;
+    }
+        
+    if( (src==nullptr && srcLength!=0) || srcLength < -1 ||
+        (destCapacity<0) || (dest == nullptr && destCapacity > 0)
+    ) {
+        *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+
+    if(srcLength < 0) {
+        /* Transform a NUL-terminated string. */
+        char16_t *pDestLimit = (dest!=nullptr)?(dest+destCapacity):nullptr;
+        uint8_t t1, t2, t3; /* trail bytes */
+
+        while(((ch = *pSrc) != 0) && (pDest < pDestLimit)) {
+            if(ch < 0xc0) {
+                /*
+                 * ASCII, or a trail byte in lead position which is treated like
+                 * a single-byte sequence for better character boundary
+                 * resynchronization after illegal sequences.
+                 */
+                *pDest++=(char16_t)ch;
+                ++pSrc;
+                continue;
+            } else if(ch < 0xe0) { /* U+0080..U+07FF */
+                if((t1 = pSrc[1]) != 0) {
+                    /* 0x3080 = (0xc0 << 6) + 0x80 */
+                    *pDest++ = (char16_t)((ch << 6) + t1 - 0x3080);
+                    pSrc += 2;
+                    continue;
+                }
+            } else if(ch < 0xf0) { /* U+0800..U+FFFF */
+                if((t1 = pSrc[1]) != 0 && (t2 = pSrc[2]) != 0) {
+                    /* no need for (ch & 0xf) because the upper bits are truncated after <<12 in the cast to (char16_t) */
+                    /* 0x2080 = (0x80 << 6) + 0x80 */
+                    *pDest++ = (char16_t)((ch << 12) + (t1 << 6) + t2 - 0x2080);
+                    pSrc += 3;
+                    continue;
+                }
+            } else /* f0..f4 */ { /* U+10000..U+10FFFF */
+                if((t1 = pSrc[1]) != 0 && (t2 = pSrc[2]) != 0 && (t3 = pSrc[3]) != 0) {
+                    pSrc += 4;
+                    /* 0x3c82080 = (0xf0 << 18) + (0x80 << 12) + (0x80 << 6) + 0x80 */
+                    ch = (ch << 18) + (t1 << 12) + (t2 << 6) + t3 - 0x3c82080;
+                    *(pDest++) = U16_LEAD(ch);
+                    if(pDest < pDestLimit) {
+                        *(pDest++) = U16_TRAIL(ch);
+                    } else {
+                        reqLength = 1;
+                        break;
+                    }
+                    continue;
+                }
+            }
+
+            /* truncated character at the end */
+            *pDest++ = 0xfffd;
+            while(*++pSrc != 0) {}
+            break;
+        }
+
+        /* Pre-flight the rest of the string. */
+        while((ch = *pSrc) != 0) {
+            if(ch < 0xc0) {
+                /*
+                 * ASCII, or a trail byte in lead position which is treated like
+                 * a single-byte sequence for better character boundary
+                 * resynchronization after illegal sequences.
+                 */
+                ++reqLength;
+                ++pSrc;
+                continue;
+            } else if(ch < 0xe0) { /* U+0080..U+07FF */
+                if(pSrc[1] != 0) {
+                    ++reqLength;
+                    pSrc += 2;
+                    continue;
+                }
+            } else if(ch < 0xf0) { /* U+0800..U+FFFF */
+                if(pSrc[1] != 0 && pSrc[2] != 0) {
+                    ++reqLength;
+                    pSrc += 3;
+                    continue;
+                }
+            } else /* f0..f4 */ { /* U+10000..U+10FFFF */
+                if(pSrc[1] != 0 && pSrc[2] != 0 && pSrc[3] != 0) {
+                    reqLength += 2;
+                    pSrc += 4;
+                    continue;
+                }
+            }
+
+            /* truncated character at the end */
+            ++reqLength;
+            break;
+        }
+    } else /* srcLength >= 0 */ {
+      const uint8_t *pSrcLimit = (pSrc!=nullptr)?(pSrc + srcLength):nullptr;
+
+        /*
+         * This function requires that if srcLength is given, then it must be
+         * destCapatity >= srcLength so that we need not check for
+         * destination buffer overflow in the loop.
+         */
+        if(destCapacity < srcLength) {
+            if(pDestLength != nullptr) {
+                *pDestLength = srcLength; /* this likely overestimates the true destLength! */
+            }
+            *pErrorCode = U_BUFFER_OVERFLOW_ERROR;
+            return nullptr;
+        }
+
+        if((pSrcLimit - pSrc) >= 4) {
+            pSrcLimit -= 3; /* temporarily reduce pSrcLimit */
+
+            /* in this loop, we can always access at least 4 bytes, up to pSrc+3 */
+            do {
+                ch = *pSrc++;
+                if(ch < 0xc0) {
+                    /*
+                     * ASCII, or a trail byte in lead position which is treated like
+                     * a single-byte sequence for better character boundary
+                     * resynchronization after illegal sequences.
+                     */
+                    *pDest++=(char16_t)ch;
+                } else if(ch < 0xe0) { /* U+0080..U+07FF */
+                    /* 0x3080 = (0xc0 << 6) + 0x80 */
+                    *pDest++ = (char16_t)((ch << 6) + *pSrc++ - 0x3080);
+                } else if(ch < 0xf0) { /* U+0800..U+FFFF */
+                    /* no need for (ch & 0xf) because the upper bits are truncated after <<12 in the cast to (char16_t) */
+                    /* 0x2080 = (0x80 << 6) + 0x80 */
+                    ch = (ch << 12) + (*pSrc++ << 6);
+                    *pDest++ = (char16_t)(ch + *pSrc++ - 0x2080);
+                } else /* f0..f4 */ { /* U+10000..U+10FFFF */
+                    /* 0x3c82080 = (0xf0 << 18) + (0x80 << 12) + (0x80 << 6) + 0x80 */
+                    ch = (ch << 18) + (*pSrc++ << 12);
+                    ch += *pSrc++ << 6;
+                    ch += *pSrc++ - 0x3c82080;
+                    *(pDest++) = U16_LEAD(ch);
+                    *(pDest++) = U16_TRAIL(ch);
+                }
+            } while(pSrc < pSrcLimit);
+
+            pSrcLimit += 3; /* restore original pSrcLimit */
+        }
+
+        while(pSrc < pSrcLimit) {
+            ch = *pSrc++;
+            if(ch < 0xc0) {
+                /*
+                 * ASCII, or a trail byte in lead position which is treated like
+                 * a single-byte sequence for better character boundary
+                 * resynchronization after illegal sequences.
+                 */
+                *pDest++=(char16_t)ch;
+                continue;
+            } else if(ch < 0xe0) { /* U+0080..U+07FF */
+                if(pSrc < pSrcLimit) {
+                    /* 0x3080 = (0xc0 << 6) + 0x80 */
+                    *pDest++ = (char16_t)((ch << 6) + *pSrc++ - 0x3080);
+                    continue;
+                }
+            } else if(ch < 0xf0) { /* U+0800..U+FFFF */
+                if((pSrcLimit - pSrc) >= 2) {
+                    /* no need for (ch & 0xf) because the upper bits are truncated after <<12 in the cast to (char16_t) */
+                    /* 0x2080 = (0x80 << 6) + 0x80 */
+                    ch = (ch << 12) + (*pSrc++ << 6);
+                    *pDest++ = (char16_t)(ch + *pSrc++ - 0x2080);
+                    pSrc += 3;
+                    continue;
+                }
+            } else /* f0..f4 */ { /* U+10000..U+10FFFF */
+                if((pSrcLimit - pSrc) >= 3) {
+                    /* 0x3c82080 = (0xf0 << 18) + (0x80 << 12) + (0x80 << 6) + 0x80 */
+                    ch = (ch << 18) + (*pSrc++ << 12);
+                    ch += *pSrc++ << 6;
+                    ch += *pSrc++ - 0x3c82080;
+                    *(pDest++) = U16_LEAD(ch);
+                    *(pDest++) = U16_TRAIL(ch);
+                    pSrc += 4;
+                    continue;
+                }
+            }
+
+            /* truncated character at the end */
+            *pDest++ = 0xfffd;
+            break;
+        }
+    }
+
+    reqLength+=(int32_t)(pDest - dest);
+
+    if(pDestLength){
+        *pDestLength = reqLength;
+    }
+
+    /* Terminate the buffer */
+    u_terminateUChars(dest,destCapacity,reqLength,pErrorCode);
+
+    return dest;
+}
+
+static inline uint8_t *
+_appendUTF8(uint8_t *pDest, UChar32 c) {
+    /* it is 0<=c<=0x10ffff and not a surrogate if called by a validating function */
+    if((c)<=0x7f) {
+        *pDest++=(uint8_t)c;
+    } else if(c<=0x7ff) {
+        *pDest++=(uint8_t)((c>>6)|0xc0);
+        *pDest++=(uint8_t)((c&0x3f)|0x80);
+    } else if(c<=0xffff) {
+        *pDest++=(uint8_t)((c>>12)|0xe0);
+        *pDest++=(uint8_t)(((c>>6)&0x3f)|0x80);
+        *pDest++=(uint8_t)(((c)&0x3f)|0x80);
+    } else /* if((uint32_t)(c)<=0x10ffff) */ {
+        *pDest++=(uint8_t)(((c)>>18)|0xf0);
+        *pDest++=(uint8_t)((((c)>>12)&0x3f)|0x80);
+        *pDest++=(uint8_t)((((c)>>6)&0x3f)|0x80);
+        *pDest++=(uint8_t)(((c)&0x3f)|0x80);
+    }
+    return pDest;
+}
+
+   
+U_CAPI char* U_EXPORT2 
+u_strToUTF8WithSub(char *dest,
+            int32_t destCapacity,
+            int32_t *pDestLength,
+            const char16_t *pSrc,
+            int32_t srcLength,
+            UChar32 subchar, int32_t *pNumSubstitutions,
+            UErrorCode *pErrorCode){
+    int32_t reqLength=0;
+    uint32_t ch=0,ch2=0;
+    uint8_t *pDest = (uint8_t *)dest;
+    uint8_t *pDestLimit = (pDest!=nullptr)?(pDest + destCapacity):nullptr;
+    int32_t numSubstitutions;
+
+    /* args check */
+    if(U_FAILURE(*pErrorCode)){
+        return nullptr;
+    }
+        
+    if( (pSrc==nullptr && srcLength!=0) || srcLength < -1 ||
+        (destCapacity<0) || (dest == nullptr && destCapacity > 0) ||
+        subchar > 0x10ffff || U_IS_SURROGATE(subchar)
+    ) {
+        *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+
+    if(pNumSubstitutions!=nullptr) {
+        *pNumSubstitutions=0;
+    }
+    numSubstitutions=0;
+
+    if(srcLength==-1) {
+        while((ch=*pSrc)!=0) {
+            ++pSrc;
+            if(ch <= 0x7f) {
+                if(pDest= 2) {
+                    *pDest++=(uint8_t)((ch>>6)|0xc0);
+                    *pDest++=(uint8_t)((ch&0x3f)|0x80);
+                } else {
+                    reqLength = 2;
+                    break;
+                }
+            } else if(ch <= 0xd7ff || ch >= 0xe000) {
+                if((pDestLimit - pDest) >= 3) {
+                    *pDest++=(uint8_t)((ch>>12)|0xe0);
+                    *pDest++=(uint8_t)(((ch>>6)&0x3f)|0x80);
+                    *pDest++=(uint8_t)((ch&0x3f)|0x80);
+                } else {
+                    reqLength = 3;
+                    break;
+                }
+            } else /* ch is a surrogate */ {
+                int32_t length;
+
+                /*need not check for NUL because NUL fails U16_IS_TRAIL() anyway*/
+                if(U16_IS_SURROGATE_LEAD(ch) && U16_IS_TRAIL(ch2=*pSrc)) { 
+                    ++pSrc;
+                    ch=U16_GET_SUPPLEMENTARY(ch, ch2);
+                } else if(subchar>=0) {
+                    ch=subchar;
+                    ++numSubstitutions;
+                } else {
+                    /* Unicode 3.2 forbids surrogate code points in UTF-8 */
+                    *pErrorCode = U_INVALID_CHAR_FOUND;
+                    return nullptr;
+                }
+
+                length = U8_LENGTH(ch);
+                if((pDestLimit - pDest) >= length) {
+                    /* convert and append*/
+                    pDest=_appendUTF8(pDest, ch);
+                } else {
+                    reqLength = length;
+                    break;
+                }
+            }
+        }
+        while((ch=*pSrc++)!=0) {
+            if(ch<=0x7f) {
+                ++reqLength;
+            } else if(ch<=0x7ff) {
+                reqLength+=2;
+            } else if(!U16_IS_SURROGATE(ch)) {
+                reqLength+=3;
+            } else if(U16_IS_SURROGATE_LEAD(ch) && U16_IS_TRAIL(ch2=*pSrc)) {
+                ++pSrc;
+                reqLength+=4;
+            } else if(subchar>=0) {
+                reqLength+=U8_LENGTH(subchar);
+                ++numSubstitutions;
+            } else {
+                /* Unicode 3.2 forbids surrogate code points in UTF-8 */
+                *pErrorCode = U_INVALID_CHAR_FOUND;
+                return nullptr;
+            }
+        }
+    } else {
+        const char16_t *pSrcLimit = (pSrc!=nullptr)?(pSrc+srcLength):nullptr;
+        int32_t count;
+
+        /* Faster loop without ongoing checking for pSrcLimit and pDestLimit. */
+        for(;;) {
+            /*
+             * Each iteration of the inner loop progresses by at most 3 UTF-8
+             * bytes and one char16_t, for most characters.
+             * For supplementary code points (4 & 2), which are rare,
+             * there is an additional adjustment.
+             */
+            count = (int32_t)((pDestLimit - pDest) / 3);
+            srcLength = (int32_t)(pSrcLimit - pSrc);
+            if(count > srcLength) {
+                count = srcLength; /* min(remaining dest/3, remaining src) */
+            }
+            if(count < 3) {
+                /*
+                 * Too much overhead if we get near the end of the string,
+                 * continue with the next loop.
+                 */
+                break;
+            }
+            do {
+                ch=*pSrc++;
+                if(ch <= 0x7f) {
+                    *pDest++ = (uint8_t)ch;
+                } else if(ch <= 0x7ff) {
+                    *pDest++=(uint8_t)((ch>>6)|0xc0);
+                    *pDest++=(uint8_t)((ch&0x3f)|0x80);
+                } else if(ch <= 0xd7ff || ch >= 0xe000) {
+                    *pDest++=(uint8_t)((ch>>12)|0xe0);
+                    *pDest++=(uint8_t)(((ch>>6)&0x3f)|0x80);
+                    *pDest++=(uint8_t)((ch&0x3f)|0x80);
+                } else /* ch is a surrogate */ {
+                    /*
+                     * We will read two UChars and probably output four bytes,
+                     * which we didn't account for with computing count,
+                     * so we adjust it here.
+                     */
+                    if(--count == 0) {
+                        --pSrc; /* undo ch=*pSrc++ for the lead surrogate */
+                        break;  /* recompute count */
+                    }
+
+                    if(U16_IS_SURROGATE_LEAD(ch) && U16_IS_TRAIL(ch2=*pSrc)) { 
+                        ++pSrc;
+                        ch=U16_GET_SUPPLEMENTARY(ch, ch2);
+
+                        /* writing 4 bytes per 2 UChars is ok */
+                        *pDest++=(uint8_t)((ch>>18)|0xf0);
+                        *pDest++=(uint8_t)(((ch>>12)&0x3f)|0x80);
+                        *pDest++=(uint8_t)(((ch>>6)&0x3f)|0x80);
+                        *pDest++=(uint8_t)((ch&0x3f)|0x80);
+                    } else  {
+                        /* Unicode 3.2 forbids surrogate code points in UTF-8 */
+                        if(subchar>=0) {
+                            ch=subchar;
+                            ++numSubstitutions;
+                        } else {
+                            *pErrorCode = U_INVALID_CHAR_FOUND;
+                            return nullptr;
+                        }
+
+                        /* convert and append*/
+                        pDest=_appendUTF8(pDest, ch);
+                    }
+                }
+            } while(--count > 0);
+        }
+
+        while(pSrc= 2) {
+                    *pDest++=(uint8_t)((ch>>6)|0xc0);
+                    *pDest++=(uint8_t)((ch&0x3f)|0x80);
+                } else {
+                    reqLength = 2;
+                    break;
+                }
+            } else if(ch <= 0xd7ff || ch >= 0xe000) {
+                if((pDestLimit - pDest) >= 3) {
+                    *pDest++=(uint8_t)((ch>>12)|0xe0);
+                    *pDest++=(uint8_t)(((ch>>6)&0x3f)|0x80);
+                    *pDest++=(uint8_t)((ch&0x3f)|0x80);
+                } else {
+                    reqLength = 3;
+                    break;
+                }
+            } else /* ch is a surrogate */ {
+                int32_t length;
+
+                if(U16_IS_SURROGATE_LEAD(ch) && pSrc=0) {
+                    ch=subchar;
+                    ++numSubstitutions;
+                } else {
+                    /* Unicode 3.2 forbids surrogate code points in UTF-8 */
+                    *pErrorCode = U_INVALID_CHAR_FOUND;
+                    return nullptr;
+                }
+
+                length = U8_LENGTH(ch);
+                if((pDestLimit - pDest) >= length) {
+                    /* convert and append*/
+                    pDest=_appendUTF8(pDest, ch);
+                } else {
+                    reqLength = length;
+                    break;
+                }
+            }
+        }
+        while(pSrc=0) {
+                reqLength+=U8_LENGTH(subchar);
+                ++numSubstitutions;
+            } else {
+                /* Unicode 3.2 forbids surrogate code points in UTF-8 */
+                *pErrorCode = U_INVALID_CHAR_FOUND;
+                return nullptr;
+            }
+        }
+    }
+
+    reqLength+=(int32_t)(pDest - (uint8_t *)dest);
+
+    if(pNumSubstitutions!=nullptr) {
+        *pNumSubstitutions=numSubstitutions;
+    }
+
+    if(pDestLength){
+        *pDestLength = reqLength;
+    }
+
+    /* Terminate the buffer */
+    u_terminateChars(dest, destCapacity, reqLength, pErrorCode);
+    return dest;
+}
+
+U_CAPI char* U_EXPORT2 
+u_strToUTF8(char *dest,
+            int32_t destCapacity,
+            int32_t *pDestLength,
+            const char16_t *pSrc,
+            int32_t srcLength,
+            UErrorCode *pErrorCode){
+    return u_strToUTF8WithSub(
+            dest, destCapacity, pDestLength,
+            pSrc, srcLength,
+            U_SENTINEL, nullptr,
+            pErrorCode);
+}
+
+U_CAPI char16_t* U_EXPORT2
+u_strFromJavaModifiedUTF8WithSub(
+        char16_t *dest,
+        int32_t destCapacity,
+        int32_t *pDestLength,
+        const char *src,
+        int32_t srcLength,
+        UChar32 subchar, int32_t *pNumSubstitutions,
+        UErrorCode *pErrorCode) {
+    /* args check */
+    if(U_FAILURE(*pErrorCode)) {
+        return nullptr;
+    }
+    if( (src==nullptr && srcLength!=0) || srcLength < -1 ||
+        (dest==nullptr && destCapacity!=0) || destCapacity<0 ||
+        subchar > 0x10ffff || U_IS_SURROGATE(subchar)
+    ) {
+        *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+
+    if(pNumSubstitutions!=nullptr) {
+        *pNumSubstitutions=0;
+    }
+    char16_t *pDest = dest;
+    char16_t *pDestLimit = dest+destCapacity;
+    int32_t reqLength = 0;
+    int32_t numSubstitutions=0;
+
+    if(srcLength < 0) {
+        /*
+         * Transform a NUL-terminated ASCII string.
+         * Handle non-ASCII strings with slower code.
+         */
+        UChar32 c;
+        while(((c = (uint8_t)*src) != 0) && c <= 0x7f && (pDest < pDestLimit)) {
+            *pDest++=(char16_t)c;
+            ++src;
+        }
+        if(c == 0) {
+            reqLength=(int32_t)(pDest - dest);
+            if(pDestLength) {
+                *pDestLength = reqLength;
+            }
+
+            /* Terminate the buffer */
+            u_terminateUChars(dest, destCapacity, reqLength, pErrorCode);
+            return dest;
+        }
+        srcLength = static_cast(uprv_strlen(src));
+    }
+
+    /* Faster loop without ongoing checking for srcLength and pDestLimit. */
+    UChar32 ch;
+    uint8_t t1, t2;
+    int32_t i = 0;
+    for(;;) {
+        int32_t count = (int32_t)(pDestLimit - pDest);
+        int32_t count2 = srcLength - i;
+        if(count >= count2 && srcLength > 0 && U8_IS_SINGLE(*src)) {
+            /* fast ASCII loop */
+            int32_t start = i;
+            uint8_t b;
+            while(i < srcLength && U8_IS_SINGLE(b = src[i])) {
+                *pDest++=b;
+                ++i;
+            }
+            int32_t delta = i - start;
+            count -= delta;
+            count2 -= delta;
+        }
+        /*
+         * Each iteration of the inner loop progresses by at most 3 UTF-8
+         * bytes and one char16_t.
+         */
+        if(subchar > 0xFFFF) {
+            break;
+        }
+        count2 /= 3;
+        if(count > count2) {
+            count = count2; /* min(remaining dest, remaining src/3) */
+        }
+        if(count < 3) {
+            /*
+             * Too much overhead if we get near the end of the string,
+             * continue with the next loop.
+             */
+            break;
+        }
+        do {
+            ch = (uint8_t)src[i++];
+            if(U8_IS_SINGLE(ch)) {
+                *pDest++=(char16_t)ch;
+            } else {
+                if(ch >= 0xe0) {
+                    if( /* handle U+0000..U+FFFF inline */
+                        ch <= 0xef &&
+                        (t1 = (uint8_t)(src[i] - 0x80)) <= 0x3f &&
+                        (t2 = (uint8_t)(src[i+1] - 0x80)) <= 0x3f
+                    ) {
+                        /* no need for (ch & 0xf) because the upper bits are truncated after <<12 in the cast to (char16_t) */
+                        *pDest++ = (char16_t)((ch << 12) | (t1 << 6) | t2);
+                        i += 2;
+                        continue;
+                    }
+                } else {
+                    if( /* handle U+0000..U+07FF inline */
+                        ch >= 0xc0 &&
+                        (t1 = (uint8_t)(src[i] - 0x80)) <= 0x3f
+                    ) {
+                        *pDest++ = (char16_t)(((ch & 0x1f) << 6) | t1);
+                        ++i;
+                        continue;
+                    }
+                }
+
+                if(subchar < 0) {
+                    *pErrorCode = U_INVALID_CHAR_FOUND;
+                    return nullptr;
+                } else if(subchar > 0xffff && --count == 0) {
+                    /*
+                     * We need to write two UChars, adjusted count for that,
+                     * and ran out of space.
+                     */
+                    --i;  // back out byte ch
+                    break;
+                } else {
+                    /* function call for error cases */
+                    utf8_nextCharSafeBody((const uint8_t *)src, &(i), srcLength, ch, -1);
+                    ++numSubstitutions;
+                    *(pDest++)=(char16_t)subchar;
+                }
+            }
+        } while(--count > 0);
+    }
+
+    while(i < srcLength && (pDest < pDestLimit)) {
+        ch = (uint8_t)src[i++];
+        if(U8_IS_SINGLE(ch)){
+            *pDest++=(char16_t)ch;
+        } else {
+            if(ch >= 0xe0) {
+                if( /* handle U+0000..U+FFFF inline */
+                    ch <= 0xef &&
+                    (i+1) < srcLength &&
+                    (t1 = (uint8_t)(src[i] - 0x80)) <= 0x3f &&
+                    (t2 = (uint8_t)(src[i+1] - 0x80)) <= 0x3f
+                ) {
+                    /* no need for (ch & 0xf) because the upper bits are truncated after <<12 in the cast to (char16_t) */
+                    *pDest++ = (char16_t)((ch << 12) | (t1 << 6) | t2);
+                    i += 2;
+                    continue;
+                }
+            } else {
+                if( /* handle U+0000..U+07FF inline */
+                    ch >= 0xc0 &&
+                    i < srcLength &&
+                    (t1 = (uint8_t)(src[i] - 0x80)) <= 0x3f
+                ) {
+                    *pDest++ = (char16_t)(((ch & 0x1f) << 6) | t1);
+                    ++i;
+                    continue;
+                }
+            }
+
+            if(subchar < 0) {
+                *pErrorCode = U_INVALID_CHAR_FOUND;
+                return nullptr;
+            } else {
+                /* function call for error cases */
+                utf8_nextCharSafeBody((const uint8_t *)src, &(i), srcLength, ch, -1);
+                ++numSubstitutions;
+                if(subchar<=0xFFFF) {
+                    *(pDest++)=(char16_t)subchar;
+                } else {
+                    *(pDest++)=U16_LEAD(subchar);
+                    if(pDest= 0xe0) {
+                if( /* handle U+0000..U+FFFF inline */
+                    ch <= 0xef &&
+                    (i+1) < srcLength &&
+                    (uint8_t)(src[i] - 0x80) <= 0x3f &&
+                    (uint8_t)(src[i+1] - 0x80) <= 0x3f
+                ) {
+                    reqLength++;
+                    i += 2;
+                    continue;
+                }
+            } else {
+                if( /* handle U+0000..U+07FF inline */
+                    ch >= 0xc0 &&
+                    i < srcLength &&
+                    (uint8_t)(src[i] - 0x80) <= 0x3f
+                ) {
+                    reqLength++;
+                    ++i;
+                    continue;
+                }
+            }
+
+            if(subchar < 0) {
+                *pErrorCode = U_INVALID_CHAR_FOUND;
+                return nullptr;
+            } else {
+                /* function call for error cases */
+                utf8_nextCharSafeBody((const uint8_t *)src, &(i), srcLength, ch, -1);
+                ++numSubstitutions;
+                reqLength+=U16_LENGTH(ch);
+            }
+        }
+    }
+
+    if(pNumSubstitutions!=nullptr) {
+        *pNumSubstitutions=numSubstitutions;
+    }
+
+    reqLength+=(int32_t)(pDest - dest);
+    if(pDestLength) {
+        *pDestLength = reqLength;
+    }
+
+    /* Terminate the buffer */
+    u_terminateUChars(dest, destCapacity, reqLength, pErrorCode);
+    return dest;
+}
+
+U_CAPI char* U_EXPORT2 
+u_strToJavaModifiedUTF8(
+        char *dest,
+        int32_t destCapacity,
+        int32_t *pDestLength,
+        const char16_t *src,
+        int32_t srcLength,
+        UErrorCode *pErrorCode) {
+    int32_t reqLength=0;
+    uint32_t ch=0;
+    const char16_t *pSrcLimit;
+    int32_t count;
+
+    /* args check */
+    if(U_FAILURE(*pErrorCode)){
+        return nullptr;
+    }
+    if( (src==nullptr && srcLength!=0) || srcLength < -1 ||
+        (dest==nullptr && destCapacity!=0) || destCapacity<0
+    ) {
+        *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+    uint8_t *pDest = (uint8_t *)dest;
+    uint8_t *pDestLimit = pDest + destCapacity;
+
+    if(srcLength==-1) {
+        /* Convert NUL-terminated ASCII, then find the string length. */
+        while((ch=*src)<=0x7f && ch != 0 && pDest= srcLength && srcLength > 0 && *src <= 0x7f) {
+            /* fast ASCII loop */
+            const char16_t *prevSrc = src;
+            int32_t delta;
+            while(src < pSrcLimit && (ch = *src) <= 0x7f && ch != 0) {
+                *pDest++=(uint8_t)ch;
+                ++src;
+            }
+            delta = (int32_t)(src - prevSrc);
+            count -= delta;
+            srcLength -= delta;
+        }
+        /*
+         * Each iteration of the inner loop progresses by at most 3 UTF-8
+         * bytes and one char16_t.
+         */
+        count /= 3;
+        if(count > srcLength) {
+            count = srcLength; /* min(remaining dest/3, remaining src) */
+        }
+        if(count < 3) {
+            /*
+             * Too much overhead if we get near the end of the string,
+             * continue with the next loop.
+             */
+            break;
+        }
+        do {
+            ch=*src++;
+            if(ch <= 0x7f && ch != 0) {
+                *pDest++ = (uint8_t)ch;
+            } else if(ch <= 0x7ff) {
+                *pDest++=(uint8_t)((ch>>6)|0xc0);
+                *pDest++=(uint8_t)((ch&0x3f)|0x80);
+            } else {
+                *pDest++=(uint8_t)((ch>>12)|0xe0);
+                *pDest++=(uint8_t)(((ch>>6)&0x3f)|0x80);
+                *pDest++=(uint8_t)((ch&0x3f)|0x80);
+            }
+        } while(--count > 0);
+    }
+
+    while(src= 2) {
+                *pDest++=(uint8_t)((ch>>6)|0xc0);
+                *pDest++=(uint8_t)((ch&0x3f)|0x80);
+            } else {
+                reqLength = 2;
+                break;
+            }
+        } else {
+            if((pDestLimit - pDest) >= 3) {
+                *pDest++=(uint8_t)((ch>>12)|0xe0);
+                *pDest++=(uint8_t)(((ch>>6)&0x3f)|0x80);
+                *pDest++=(uint8_t)((ch&0x3f)|0x80);
+            } else {
+                reqLength = 3;
+                break;
+            }
+        }
+    }
+    while(src
+
+#include "unicode/utypes.h"
+#include "unicode/ustring.h"
+#include "unicode/unistr.h"
+#include "unicode/chariter.h"
+#include "unicode/utext.h"
+#include "unicode/utf.h"
+#include "unicode/utf8.h"
+#include "unicode/utf16.h"
+#include "ustr_imp.h"
+#include "cmemory.h"
+#include "cstring.h"
+#include "uassert.h"
+#include "putilimp.h"
+
+U_NAMESPACE_USE
+
+#define I32_FLAG(bitIndex) ((int32_t)1<<(bitIndex))
+
+
+static UBool
+utext_access(UText *ut, int64_t index, UBool forward) {
+    return ut->pFuncs->access(ut, index, forward);
+}
+
+
+
+U_CAPI UBool U_EXPORT2
+utext_moveIndex32(UText *ut, int32_t delta) {
+    UChar32  c;
+    if (delta > 0) {
+        do {
+            if(ut->chunkOffset>=ut->chunkLength && !utext_access(ut, ut->chunkNativeLimit, true)) {
+                return false;
+            }
+            c = ut->chunkContents[ut->chunkOffset];
+            if (U16_IS_SURROGATE(c)) {
+                c = utext_next32(ut);
+                if (c == U_SENTINEL) {
+                    return false;
+                }
+            } else {
+                ut->chunkOffset++;
+            }
+        } while(--delta>0);
+
+    } else if (delta<0) {
+        do {
+            if(ut->chunkOffset<=0 && !utext_access(ut, ut->chunkNativeStart, false)) {
+                return false;
+            }
+            c = ut->chunkContents[ut->chunkOffset-1];
+            if (U16_IS_SURROGATE(c)) {
+                c = utext_previous32(ut);
+                if (c == U_SENTINEL) {
+                    return false;
+                }
+            } else {
+                ut->chunkOffset--;
+            }
+        } while(++delta<0);
+    }
+
+    return true;
+}
+
+
+U_CAPI int64_t U_EXPORT2
+utext_nativeLength(UText *ut) {
+    return ut->pFuncs->nativeLength(ut);
+}
+
+
+U_CAPI UBool U_EXPORT2
+utext_isLengthExpensive(const UText *ut) {
+    UBool r = (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)) != 0;
+    return r;
+}
+
+
+U_CAPI int64_t U_EXPORT2
+utext_getNativeIndex(const UText *ut) {
+    if(ut->chunkOffset <= ut->nativeIndexingLimit) {
+        return ut->chunkNativeStart+ut->chunkOffset;
+    } else {
+        return ut->pFuncs->mapOffsetToNative(ut);
+    }
+}
+
+
+U_CAPI void U_EXPORT2
+utext_setNativeIndex(UText *ut, int64_t index) {
+    if(indexchunkNativeStart || index>=ut->chunkNativeLimit) {
+        // The desired position is outside of the current chunk.
+        // Access the new position.  Assume a forward iteration from here,
+        // which will also be optimimum for a single random access.
+        // Reverse iterations may suffer slightly.
+        ut->pFuncs->access(ut, index, true);
+    } else if((int32_t)(index - ut->chunkNativeStart) <= ut->nativeIndexingLimit) {
+        // utf-16 indexing.
+        ut->chunkOffset=(int32_t)(index-ut->chunkNativeStart);
+    } else {
+         ut->chunkOffset=ut->pFuncs->mapNativeIndexToUTF16(ut, index);
+    }
+    // The convention is that the index must always be on a code point boundary.
+    // Adjust the index position if it is in the middle of a surrogate pair.
+    if (ut->chunkOffsetchunkLength) {
+        char16_t c= ut->chunkContents[ut->chunkOffset];
+        if (U16_IS_TRAIL(c)) {
+            if (ut->chunkOffset==0) {
+                ut->pFuncs->access(ut, ut->chunkNativeStart, false);
+            }
+            if (ut->chunkOffset>0) {
+                char16_t lead = ut->chunkContents[ut->chunkOffset-1];
+                if (U16_IS_LEAD(lead)) {
+                    ut->chunkOffset--;
+                }
+            }
+        }
+    }
+}
+
+
+
+U_CAPI int64_t U_EXPORT2
+utext_getPreviousNativeIndex(UText *ut) {
+    //
+    //  Fast-path the common case.
+    //     Common means current position is not at the beginning of a chunk
+    //     and the preceding character is not supplementary.
+    //
+    int32_t i = ut->chunkOffset - 1;
+    int64_t result;
+    if (i >= 0) {
+        char16_t c = ut->chunkContents[i];
+        if (U16_IS_TRAIL(c) == false) {
+            if (i <= ut->nativeIndexingLimit) {
+                result = ut->chunkNativeStart + i;
+            } else {
+                ut->chunkOffset = i;
+                result = ut->pFuncs->mapOffsetToNative(ut);
+                ut->chunkOffset++;
+            }
+            return result;
+        }
+    }
+
+    // If at the start of text, simply return 0.
+    if (ut->chunkOffset==0 && ut->chunkNativeStart==0) {
+        return 0;
+    }
+
+    // Harder, less common cases.  We are at a chunk boundary, or on a surrogate.
+    //    Keep it simple, use other functions to handle the edges.
+    //
+    utext_previous32(ut);
+    result = UTEXT_GETNATIVEINDEX(ut);
+    utext_next32(ut);
+    return result;
+}
+
+
+//
+//  utext_current32.  Get the UChar32 at the current position.
+//                    UText iteration position is always on a code point boundary,
+//                    never on the trail half of a surrogate pair.
+//
+U_CAPI UChar32 U_EXPORT2
+utext_current32(UText *ut) {
+    UChar32  c;
+    if (ut->chunkOffset==ut->chunkLength) {
+        // Current position is just off the end of the chunk.
+        if (ut->pFuncs->access(ut, ut->chunkNativeLimit, true) == false) {
+            // Off the end of the text.
+            return U_SENTINEL;
+        }
+    }
+
+    c = ut->chunkContents[ut->chunkOffset];
+    if (U16_IS_LEAD(c) == false) {
+        // Normal, non-supplementary case.
+        return c;
+    }
+
+    //
+    //  Possible supplementary char.
+    //
+    UChar32   trail = 0;
+    UChar32   supplementaryC = c;
+    if ((ut->chunkOffset+1) < ut->chunkLength) {
+        // The trail surrogate is in the same chunk.
+        trail = ut->chunkContents[ut->chunkOffset+1];
+    } else {
+        //  The trail surrogate is in a different chunk.
+        //     Because we must maintain the iteration position, we need to switch forward
+        //     into the new chunk, get the trail surrogate, then revert the chunk back to the
+        //     original one.
+        //     An edge case to be careful of:  the entire text may end with an unpaired
+        //        leading surrogate.  The attempt to access the trail will fail, but
+        //        the original position before the unpaired lead still needs to be restored.
+        int64_t  nativePosition = ut->chunkNativeLimit;
+        if (ut->pFuncs->access(ut, nativePosition, true)) {
+            trail = ut->chunkContents[ut->chunkOffset];
+        }
+        UBool r = ut->pFuncs->access(ut, nativePosition, false);  // reverse iteration flag loads preceding chunk
+        U_ASSERT(r);
+        // Here we need to restore chunkOffset since the access functions were called with
+        // chunkNativeLimit but that is not where we were (we were 1 code unit before the
+        // limit). Restoring was originally added in ICU-4669 but did not support access
+        // functions that changed the chunk size, the following does.
+        ut->chunkOffset = ut->chunkLength - 1;
+        if(!r) {
+            return U_SENTINEL;
+        }
+    }
+
+    if (U16_IS_TRAIL(trail)) {
+        supplementaryC = U16_GET_SUPPLEMENTARY(c, trail);
+    }
+    return supplementaryC;
+
+}
+
+
+U_CAPI UChar32 U_EXPORT2
+utext_char32At(UText *ut, int64_t nativeIndex) {
+    UChar32 c = U_SENTINEL;
+
+    // Fast path the common case.
+    if (nativeIndex>=ut->chunkNativeStart && nativeIndex < ut->chunkNativeStart + ut->nativeIndexingLimit) {
+        ut->chunkOffset = (int32_t)(nativeIndex - ut->chunkNativeStart);
+        c = ut->chunkContents[ut->chunkOffset];
+        if (U16_IS_SURROGATE(c) == false) {
+            return c;
+        }
+    }
+
+
+    utext_setNativeIndex(ut, nativeIndex);
+    if (nativeIndex>=ut->chunkNativeStart && ut->chunkOffsetchunkLength) {
+        c = ut->chunkContents[ut->chunkOffset];
+        if (U16_IS_SURROGATE(c)) {
+            // For surrogates, let current32() deal with the complications
+            //    of supplementaries that may span chunk boundaries.
+            c = utext_current32(ut);
+        }
+    }
+    return c;
+}
+
+
+U_CAPI UChar32 U_EXPORT2
+utext_next32(UText *ut) {
+    UChar32       c;
+
+    if (ut->chunkOffset >= ut->chunkLength) {
+        if (ut->pFuncs->access(ut, ut->chunkNativeLimit, true) == false) {
+            return U_SENTINEL;
+        }
+    }
+
+    c = ut->chunkContents[ut->chunkOffset++];
+    if (U16_IS_LEAD(c) == false) {
+        // Normal case, not supplementary.
+        //   (A trail surrogate seen here is just returned as is, as a surrogate value.
+        //    It cannot be part of a pair.)
+        return c;
+    }
+
+    if (ut->chunkOffset >= ut->chunkLength) {
+        if (ut->pFuncs->access(ut, ut->chunkNativeLimit, true) == false) {
+            // c is an unpaired lead surrogate at the end of the text.
+            // return it as it is.
+            return c;
+        }
+    }
+    UChar32 trail = ut->chunkContents[ut->chunkOffset];
+    if (U16_IS_TRAIL(trail) == false) {
+        // c was an unpaired lead surrogate, not at the end of the text.
+        // return it as it is (unpaired).  Iteration position is on the
+        // following character, possibly in the next chunk, where the
+        //  trail surrogate would have been if it had existed.
+        return c;
+    }
+
+    UChar32 supplementary = U16_GET_SUPPLEMENTARY(c, trail);
+    ut->chunkOffset++;   // move iteration position over the trail surrogate.
+    return supplementary;
+    }
+
+
+U_CAPI UChar32 U_EXPORT2
+utext_previous32(UText *ut) {
+    UChar32       c;
+
+    if (ut->chunkOffset <= 0) {
+        if (ut->pFuncs->access(ut, ut->chunkNativeStart, false) == false) {
+            return U_SENTINEL;
+        }
+    }
+    ut->chunkOffset--;
+    c = ut->chunkContents[ut->chunkOffset];
+    if (U16_IS_TRAIL(c) == false) {
+        // Normal case, not supplementary.
+        //   (A lead surrogate seen here is just returned as is, as a surrogate value.
+        //    It cannot be part of a pair.)
+        return c;
+    }
+
+    if (ut->chunkOffset <= 0) {
+        if (ut->pFuncs->access(ut, ut->chunkNativeStart, false) == false) {
+            // c is an unpaired trail surrogate at the start of the text.
+            // return it as it is.
+            return c;
+        }
+    }
+
+    UChar32 lead = ut->chunkContents[ut->chunkOffset-1];
+    if (U16_IS_LEAD(lead) == false) {
+        // c was an unpaired trail surrogate, not at the end of the text.
+        // return it as it is (unpaired).  Iteration position is at c
+        return c;
+    }
+
+    UChar32 supplementary = U16_GET_SUPPLEMENTARY(lead, c);
+    ut->chunkOffset--;   // move iteration position over the lead surrogate.
+    return supplementary;
+}
+
+
+
+U_CAPI UChar32 U_EXPORT2
+utext_next32From(UText *ut, int64_t index) {
+    UChar32       c      = U_SENTINEL;
+
+    if(indexchunkNativeStart || index>=ut->chunkNativeLimit) {
+        // Desired position is outside of the current chunk.
+        if(!ut->pFuncs->access(ut, index, true)) {
+            // no chunk available here
+            return U_SENTINEL;
+        }
+    } else if (index - ut->chunkNativeStart  <= (int64_t)ut->nativeIndexingLimit) {
+        // Desired position is in chunk, with direct 1:1 native to UTF16 indexing
+        ut->chunkOffset = (int32_t)(index - ut->chunkNativeStart);
+    } else {
+        // Desired position is in chunk, with non-UTF16 indexing.
+        ut->chunkOffset = ut->pFuncs->mapNativeIndexToUTF16(ut, index);
+    }
+
+    c = ut->chunkContents[ut->chunkOffset++];
+    if (U16_IS_SURROGATE(c)) {
+        // Surrogates.  Many edge cases.  Use other functions that already
+        //              deal with the problems.
+        utext_setNativeIndex(ut, index);
+        c = utext_next32(ut);
+    }
+    return c;
+}
+
+
+U_CAPI UChar32 U_EXPORT2
+utext_previous32From(UText *ut, int64_t index) {
+    //
+    //  Return the character preceding the specified index.
+    //  Leave the iteration position at the start of the character that was returned.
+    //
+    UChar32     cPrev;    // The character preceding cCurr, which is what we will return.
+
+    // Address the chunk containing the position preceding the incoming index
+    // A tricky edge case:
+    //   We try to test the requested native index against the chunkNativeStart to determine
+    //    whether the character preceding the one at the index is in the current chunk.
+    //    BUT, this test can fail with UTF-8 (or any other multibyte encoding), when the
+    //    requested index is on something other than the first position of the first char.
+    //
+    if(index<=ut->chunkNativeStart || index>ut->chunkNativeLimit) {
+        // Requested native index is outside of the current chunk.
+        if(!ut->pFuncs->access(ut, index, false)) {
+            // no chunk available here
+            return U_SENTINEL;
+        }
+    } else if(index - ut->chunkNativeStart <= (int64_t)ut->nativeIndexingLimit) {
+        // Direct UTF-16 indexing.
+        ut->chunkOffset = (int32_t)(index - ut->chunkNativeStart);
+    } else {
+        ut->chunkOffset=ut->pFuncs->mapNativeIndexToUTF16(ut, index);
+        if (ut->chunkOffset==0 && !ut->pFuncs->access(ut, index, false)) {
+            // no chunk available here
+            return U_SENTINEL;
+        }
+    }
+
+    //
+    // Simple case with no surrogates.
+    //
+    ut->chunkOffset--;
+    cPrev = ut->chunkContents[ut->chunkOffset];
+
+    if (U16_IS_SURROGATE(cPrev)) {
+        // Possible supplementary.  Many edge cases.
+        // Let other functions do the heavy lifting.
+        utext_setNativeIndex(ut, index);
+        cPrev = utext_previous32(ut);
+    }
+    return cPrev;
+}
+
+
+U_CAPI int32_t U_EXPORT2
+utext_extract(UText *ut,
+             int64_t start, int64_t limit,
+             char16_t *dest, int32_t destCapacity,
+             UErrorCode *status) {
+                 return ut->pFuncs->extract(ut, start, limit, dest, destCapacity, status);
+             }
+
+
+
+U_CAPI UBool U_EXPORT2
+utext_equals(const UText *a, const UText *b) {
+    if (a==nullptr || b==nullptr ||
+        a->magic != UTEXT_MAGIC ||
+        b->magic != UTEXT_MAGIC) {
+            // Null or invalid arguments don't compare equal to anything.
+            return false;
+    }
+
+    if (a->pFuncs != b->pFuncs) {
+        // Different types of text providers.
+        return false;
+    }
+
+    if (a->context != b->context) {
+        // Different sources (different strings)
+        return false;
+    }
+    if (utext_getNativeIndex(a) != utext_getNativeIndex(b)) {
+        // Different current position in the string.
+        return false;
+    }
+
+    return true;
+}
+
+U_CAPI UBool U_EXPORT2
+utext_isWritable(const UText *ut)
+{
+    UBool b = (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_WRITABLE)) != 0;
+    return b;
+}
+
+
+U_CAPI void U_EXPORT2
+utext_freeze(UText *ut) {
+    // Zero out the WRITABLE flag.
+    ut->providerProperties &= ~(I32_FLAG(UTEXT_PROVIDER_WRITABLE));
+}
+
+
+U_CAPI UBool U_EXPORT2
+utext_hasMetaData(const UText *ut)
+{
+    UBool b = (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_HAS_META_DATA)) != 0;
+    return b;
+}
+
+
+
+U_CAPI int32_t U_EXPORT2
+utext_replace(UText *ut,
+             int64_t nativeStart, int64_t nativeLimit,
+             const char16_t *replacementText, int32_t replacementLength,
+             UErrorCode *status)
+{
+    if (U_FAILURE(*status)) {
+        return 0;
+    }
+    if ((ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_WRITABLE)) == 0) {
+        *status = U_NO_WRITE_PERMISSION;
+        return 0;
+    }
+    int32_t i = ut->pFuncs->replace(ut, nativeStart, nativeLimit, replacementText, replacementLength, status);
+    return i;
+}
+
+U_CAPI void U_EXPORT2
+utext_copy(UText *ut,
+          int64_t nativeStart, int64_t nativeLimit,
+          int64_t destIndex,
+          UBool move,
+          UErrorCode *status)
+{
+    if (U_FAILURE(*status)) {
+        return;
+    }
+    if ((ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_WRITABLE)) == 0) {
+        *status = U_NO_WRITE_PERMISSION;
+        return;
+    }
+    ut->pFuncs->copy(ut, nativeStart, nativeLimit, destIndex, move, status);
+}
+
+
+
+U_CAPI UText * U_EXPORT2
+utext_clone(UText *dest, const UText *src, UBool deep, UBool readOnly, UErrorCode *status) {
+    if (U_FAILURE(*status)) {
+        return dest;
+    }
+    UText *result = src->pFuncs->clone(dest, src, deep, status);
+    if (U_FAILURE(*status)) {
+        return result;
+    }
+    if (result == nullptr) {
+        *status = U_MEMORY_ALLOCATION_ERROR;
+        return result;
+    }
+    if (readOnly) {
+        utext_freeze(result);
+    }
+    return result;
+}
+
+
+
+//------------------------------------------------------------------------------
+//
+//   UText common functions implementation
+//
+//------------------------------------------------------------------------------
+
+//
+//  UText.flags bit definitions
+//
+enum {
+    UTEXT_HEAP_ALLOCATED  = 1,      //  1 if ICU has allocated this UText struct on the heap.
+                                    //  0 if caller provided storage for the UText.
+
+    UTEXT_EXTRA_HEAP_ALLOCATED = 2, //  1 if ICU has allocated extra storage as a separate
+                                    //     heap block.
+                                    //  0 if there is no separate allocation.  Either no extra
+                                    //     storage was requested, or it is appended to the end
+                                    //     of the main UText storage.
+
+    UTEXT_OPEN = 4                  //  1 if this UText is currently open
+                                    //  0 if this UText is not open.
+};
+
+
+//
+//  Extended form of a UText.  The purpose is to aid in computing the total size required
+//    when a provider asks for a UText to be allocated with extra storage.
+
+struct ExtendedUText {
+    UText               ut;
+    std::max_align_t    extension;
+};
+
+static const UText emptyText = UTEXT_INITIALIZER;
+
+U_CAPI UText * U_EXPORT2
+utext_setup(UText *ut, int32_t extraSpace, UErrorCode *status) {
+    if (U_FAILURE(*status)) {
+        return ut;
+    }
+
+    if (ut == nullptr) {
+        // We need to heap-allocate storage for the new UText
+        int32_t spaceRequired = sizeof(UText);
+        if (extraSpace > 0) {
+            spaceRequired = sizeof(ExtendedUText) + extraSpace - sizeof(std::max_align_t);
+        }
+        ut = (UText *)uprv_malloc(spaceRequired);
+        if (ut == nullptr) {
+            *status = U_MEMORY_ALLOCATION_ERROR;
+            return nullptr;
+        } else {
+            *ut = emptyText;
+            ut->flags |= UTEXT_HEAP_ALLOCATED;
+            if (spaceRequired>0) {
+                ut->extraSize = extraSpace;
+                ut->pExtra    = &((ExtendedUText *)ut)->extension;
+            }
+        }
+    } else {
+        // We have been supplied with an already existing UText.
+        // Verify that it really appears to be a UText.
+        if (ut->magic != UTEXT_MAGIC) {
+            *status = U_ILLEGAL_ARGUMENT_ERROR;
+            return ut;
+        }
+        // If the ut is already open and there's a provider supplied close
+        //   function, call it.
+        if ((ut->flags & UTEXT_OPEN) && ut->pFuncs->close != nullptr)  {
+            ut->pFuncs->close(ut);
+        }
+        ut->flags &= ~UTEXT_OPEN;
+
+        // If extra space was requested by our caller, check whether
+        //   sufficient already exists, and allocate new if needed.
+        if (extraSpace > ut->extraSize) {
+            // Need more space.  If there is existing separately allocated space,
+            //   delete it first, then allocate new space.
+            if (ut->flags & UTEXT_EXTRA_HEAP_ALLOCATED) {
+                uprv_free(ut->pExtra);
+                ut->extraSize = 0;
+            }
+            ut->pExtra = uprv_malloc(extraSpace);
+            if (ut->pExtra == nullptr) {
+                *status = U_MEMORY_ALLOCATION_ERROR;
+            } else {
+                ut->extraSize = extraSpace;
+                ut->flags |= UTEXT_EXTRA_HEAP_ALLOCATED;
+            }
+        }
+    }
+    if (U_SUCCESS(*status)) {
+        ut->flags |= UTEXT_OPEN;
+
+        // Initialize all remaining fields of the UText.
+        //
+        ut->context             = nullptr;
+        ut->chunkContents       = nullptr;
+        ut->p                   = nullptr;
+        ut->q                   = nullptr;
+        ut->r                   = nullptr;
+        ut->a                   = 0;
+        ut->b                   = 0;
+        ut->c                   = 0;
+        ut->chunkOffset         = 0;
+        ut->chunkLength         = 0;
+        ut->chunkNativeStart    = 0;
+        ut->chunkNativeLimit    = 0;
+        ut->nativeIndexingLimit = 0;
+        ut->providerProperties  = 0;
+        ut->privA               = 0;
+        ut->privB               = 0;
+        ut->privC               = 0;
+        ut->privP               = nullptr;
+        if (ut->pExtra!=nullptr && ut->extraSize>0)
+            uprv_memset(ut->pExtra, 0, ut->extraSize);
+
+    }
+    return ut;
+}
+
+
+U_CAPI UText * U_EXPORT2
+utext_close(UText *ut) {
+    if (ut==nullptr ||
+        ut->magic != UTEXT_MAGIC ||
+        (ut->flags & UTEXT_OPEN) == 0)
+    {
+        // The supplied ut is not an open UText.
+        // Do nothing.
+        return ut;
+    }
+
+    // If the provider gave us a close function, call it now.
+    // This will clean up anything allocated specifically by the provider.
+    if (ut->pFuncs->close != nullptr) {
+        ut->pFuncs->close(ut);
+    }
+    ut->flags &= ~UTEXT_OPEN;
+
+    // If we (the framework) allocated the UText or subsidiary storage,
+    //   delete it.
+    if (ut->flags & UTEXT_EXTRA_HEAP_ALLOCATED) {
+        uprv_free(ut->pExtra);
+        ut->pExtra = nullptr;
+        ut->flags &= ~UTEXT_EXTRA_HEAP_ALLOCATED;
+        ut->extraSize = 0;
+    }
+
+    // Zero out function table of the closed UText.  This is a defensive move,
+    //   intended to cause applications that inadvertently use a closed
+    //   utext to crash with null pointer errors.
+    ut->pFuncs        = nullptr;
+
+    if (ut->flags & UTEXT_HEAP_ALLOCATED) {
+        // This UText was allocated by UText setup.  We need to free it.
+        // Clear magic, so we can detect if the user messes up and immediately
+        //  tries to reopen another UText using the deleted storage.
+        ut->magic = 0;
+        uprv_free(ut);
+        ut = nullptr;
+    }
+    return ut;
+}
+
+
+
+
+//
+// invalidateChunk   Reset a chunk to have no contents, so that the next call
+//                   to access will cause new data to load.
+//                   This is needed when copy/move/replace operate directly on the
+//                   backing text, potentially putting it out of sync with the
+//                   contents in the chunk.
+//
+static void
+invalidateChunk(UText *ut) {
+    ut->chunkLength = 0;
+    ut->chunkNativeLimit = 0;
+    ut->chunkNativeStart = 0;
+    ut->chunkOffset = 0;
+    ut->nativeIndexingLimit = 0;
+}
+
+//
+// pinIndex        Do range pinning on a native index parameter.
+//                 64 bit pinning is done in place.
+//                 32 bit truncated result is returned as a convenience for
+//                        use in providers that don't need 64 bits.
+static int32_t
+pinIndex(int64_t &index, int64_t limit) {
+    if (index<0) {
+        index = 0;
+    } else if (index > limit) {
+        index = limit;
+    }
+    return (int32_t)index;
+}
+
+
+U_CDECL_BEGIN
+
+//
+// Pointer relocation function,
+//   a utility used by shallow clone.
+//   Adjust a pointer that refers to something within one UText (the source)
+//   to refer to the same relative offset within a another UText (the target)
+//
+static void adjustPointer(UText *dest, const void **destPtr, const UText *src) {
+    // convert all pointers to (char *) so that byte address arithmetic will work.
+    char  *dptr = (char *)*destPtr;
+    char  *dUText = (char *)dest;
+    char  *sUText = (char *)src;
+
+    if (dptr >= (char *)src->pExtra && dptr < ((char*)src->pExtra)+src->extraSize) {
+        // target ptr was to something within the src UText's pExtra storage.
+        //   relocate it into the target UText's pExtra region.
+        *destPtr = ((char *)dest->pExtra) + (dptr - (char *)src->pExtra);
+    } else if (dptr>=sUText && dptr < sUText+src->sizeOfStruct) {
+        // target ptr was pointing to somewhere within the source UText itself.
+        //   Move it to the same offset within the target UText.
+        *destPtr = dUText + (dptr-sUText);
+    }
+}
+
+
+//
+//  Clone.  This is a generic copy-the-utext-by-value clone function that can be
+//          used as-is with some utext types, and as a helper by other clones.
+//
+static UText * U_CALLCONV
+shallowTextClone(UText * dest, const UText * src, UErrorCode * status) {
+    if (U_FAILURE(*status)) {
+        return nullptr;
+    }
+    int32_t  srcExtraSize = src->extraSize;
+
+    //
+    // Use the generic text_setup to allocate storage if required.
+    //
+    dest = utext_setup(dest, srcExtraSize, status);
+    if (U_FAILURE(*status)) {
+        return dest;
+    }
+
+    //
+    //  flags (how the UText was allocated) and the pointer to the
+    //   extra storage must retain the values in the cloned utext that
+    //   were set up by utext_setup.  Save them separately before
+    //   copying the whole struct.
+    //
+    void *destExtra = dest->pExtra;
+    int32_t flags   = dest->flags;
+
+
+    //
+    //  Copy the whole UText struct by value.
+    //  Any "Extra" storage is copied also.
+    //
+    int sizeToCopy = src->sizeOfStruct;
+    if (sizeToCopy > dest->sizeOfStruct) {
+        sizeToCopy = dest->sizeOfStruct;
+    }
+    uprv_memcpy(dest, src, sizeToCopy);
+    dest->pExtra = destExtra;
+    dest->flags  = flags;
+    if (srcExtraSize > 0) {
+        uprv_memcpy(dest->pExtra, src->pExtra, srcExtraSize);
+    }
+
+    //
+    // Relocate any pointers in the target that refer to the UText itself
+    //   to point to the cloned copy rather than the original source.
+    //
+    adjustPointer(dest, &dest->context, src);
+    adjustPointer(dest, &dest->p, src);
+    adjustPointer(dest, &dest->q, src);
+    adjustPointer(dest, &dest->r, src);
+    adjustPointer(dest, (const void **)&dest->chunkContents, src);
+
+    // The newly shallow-cloned UText does _not_ own the underlying storage for the text.
+    // (The source for the clone may or may not have owned the text.)
+
+    dest->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT);
+
+    return dest;
+}
+
+
+U_CDECL_END
+
+
+
+//------------------------------------------------------------------------------
+//
+//     UText implementation for UTF-8 char * strings (read-only)
+//     Limitation:  string length must be <= 0x7fffffff in length.
+//                  (length must for in an int32_t variable)
+//
+//         Use of UText data members:
+//              context    pointer to UTF-8 string
+//              utext.b    is the input string length (bytes).
+//              utext.c    Length scanned so far in string
+//                           (for optimizing finding length of zero terminated strings.)
+//              utext.p    pointer to the current buffer
+//              utext.q    pointer to the other buffer.
+//
+//------------------------------------------------------------------------------
+
+// Chunk size.
+//     Must be less than 85 (256/3), because of byte mapping from char16_t indexes to native indexes.
+//     Worst case is three native bytes to one char16_t.  (Supplemenaries are 4 native bytes
+//     to two UChars.)
+//     The longest illegal byte sequence treated as a single error (and converted to U+FFFD)
+//     is a three-byte sequence (truncated four-byte sequence).
+//
+enum { UTF8_TEXT_CHUNK_SIZE=32 };
+
+//
+// UTF8Buf  Two of these structs will be set up in the UText's extra allocated space.
+//          Each contains the char16_t chunk buffer, the to and from native maps, and
+//          header info.
+//
+//     because backwards iteration fills the buffers starting at the end and
+//     working towards the front, the filled part of the buffers may not begin
+//     at the start of the available storage for the buffers.
+//
+//     Buffer size is one bigger than the specified UTF8_TEXT_CHUNK_SIZE to allow for
+//     the last character added being a supplementary, and thus requiring a surrogate
+//     pair.  Doing this is simpler than checking for the edge case.
+//
+
+struct UTF8Buf {
+    int32_t   bufNativeStart;                        // Native index of first char in char16_t buf
+    int32_t   bufNativeLimit;                        // Native index following last char in buf.
+    int32_t   bufStartIdx;                           // First filled position in buf.
+    int32_t   bufLimitIdx;                           // Limit of filled range in buf.
+    int32_t   bufNILimit;                            // Limit of native indexing part of buf
+    int32_t   toUCharsMapStart;                      // Native index corresponding to
+                                                     //   mapToUChars[0].
+                                                     //   Set to bufNativeStart when filling forwards.
+                                                     //   Set to computed value when filling backwards.
+
+    char16_t  buf[UTF8_TEXT_CHUNK_SIZE+4];           // The char16_t buffer.  Requires one extra position beyond the
+                                                     //   the chunk size, to allow for surrogate at the end.
+                                                     //   Length must be identical to mapToNative array, below,
+                                                     //   because of the way indexing works when the array is
+                                                     //   filled backwards during a reverse iteration.  Thus,
+                                                     //   the additional extra size.
+    uint8_t   mapToNative[UTF8_TEXT_CHUNK_SIZE+4];   // map char16_t index in buf to
+                                                     //  native offset from bufNativeStart.
+                                                     //  Requires two extra slots,
+                                                     //    one for a supplementary starting in the last normal position,
+                                                     //    and one for an entry for the buffer limit position.
+    uint8_t   mapToUChars[UTF8_TEXT_CHUNK_SIZE*3+6]; // Map native offset from bufNativeStart to
+                                                     //   corresponding offset in filled part of buf.
+    int32_t   align;
+};
+
+U_CDECL_BEGIN
+
+//
+//   utf8TextLength
+//
+//        Get the length of the string.  If we don't already know it,
+//              we'll need to scan for the trailing  nul.
+//
+static int64_t U_CALLCONV
+utf8TextLength(UText *ut) {
+    if (ut->b < 0) {
+        // Zero terminated string, and we haven't scanned to the end yet.
+        // Scan it now.
+        const char *r = (const char *)ut->context + ut->c;
+        while (*r != 0) {
+            r++;
+        }
+        if ((r - (const char *)ut->context) < 0x7fffffff) {
+            ut->b = (int32_t)(r - (const char *)ut->context);
+        } else {
+            // Actual string was bigger (more than 2 gig) than we
+            //   can handle.  Clip it to 2 GB.
+            ut->b = 0x7fffffff;
+        }
+        ut->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE);
+    }
+    return ut->b;
+}
+
+
+
+
+
+
+static UBool U_CALLCONV
+utf8TextAccess(UText *ut, int64_t index, UBool forward) {
+    //
+    //  Apologies to those who are allergic to goto statements.
+    //    Consider each goto to a labelled block to be the equivalent of
+    //         call the named block as if it were a function();
+    //         return;
+    //
+    const uint8_t *s8=(const uint8_t *)ut->context;
+    UTF8Buf *u8b = nullptr;
+    int32_t  length = ut->b;         // Length of original utf-8
+    int32_t  ix= (int32_t)index;     // Requested index, trimmed to 32 bits.
+    int32_t  mapIndex = 0;
+    if (index<0) {
+        ix=0;
+    } else if (index > 0x7fffffff) {
+        // Strings with 64 bit lengths not supported by this UTF-8 provider.
+        ix = 0x7fffffff;
+    }
+
+    // Pin requested index to the string length.
+    if (ix>length) {
+        if (length>=0) {
+            ix=length;
+        } else if (ix>=ut->c) {
+            // Zero terminated string, and requested index is beyond
+            //   the region that has already been scanned.
+            //   Scan up to either the end of the string or to the
+            //   requested position, whichever comes first.
+            while (ut->cc]!=0) {
+                ut->c++;
+            }
+            //  TODO:  support for null terminated string length > 32 bits.
+            if (s8[ut->c] == 0) {
+                // We just found the actual length of the string.
+                //  Trim the requested index back to that.
+                ix     = ut->c;
+                ut->b  = ut->c;
+                length = ut->c;
+                ut->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE);
+            }
+        }
+    }
+
+    //
+    // Dispatch to the appropriate action for a forward iteration request.
+    //
+    if (forward) {
+        if (ix==ut->chunkNativeLimit) {
+            // Check for normal sequential iteration cases first.
+            if (ix==length) {
+                // Just reached end of string
+                // Don't swap buffers, but do set the
+                //   current buffer position.
+                ut->chunkOffset = ut->chunkLength;
+                return false;
+            } else {
+                // End of current buffer.
+                //   check whether other buffer already has what we need.
+                UTF8Buf *altB = (UTF8Buf *)ut->q;
+                if (ix>=altB->bufNativeStart && ixbufNativeLimit) {
+                    goto swapBuffers;
+                }
+            }
+        }
+
+        // A random access.  Desired index could be in either or niether buf.
+        // For optimizing the order of testing, first check for the index
+        //    being in the other buffer.  This will be the case for uses that
+        //    move back and forth over a fairly limited range
+        {
+            u8b = (UTF8Buf *)ut->q;   // the alternate buffer
+            if (ix>=u8b->bufNativeStart && ixbufNativeLimit) {
+                // Requested index is in the other buffer.
+                goto swapBuffers;
+            }
+            if (ix == length) {
+                // Requested index is end-of-string.
+                //   (this is the case of randomly seeking to the end.
+                //    The case of iterating off the end is handled earlier.)
+                if (ix == ut->chunkNativeLimit) {
+                    // Current buffer extends up to the end of the string.
+                    //   Leave it as the current buffer.
+                    ut->chunkOffset = ut->chunkLength;
+                    return false;
+                }
+                if (ix == u8b->bufNativeLimit) {
+                    // Alternate buffer extends to the end of string.
+                    //   Swap it in as the current buffer.
+                    goto swapBuffersAndFail;
+                }
+
+                // Neither existing buffer extends to the end of the string.
+                goto makeStubBuffer;
+            }
+
+            if (ixchunkNativeStart || ix>=ut->chunkNativeLimit) {
+                // Requested index is in neither buffer.
+                goto fillForward;
+            }
+
+            // Requested index is in this buffer.
+            u8b = (UTF8Buf *)ut->p;   // the current buffer
+            mapIndex = ix - u8b->toUCharsMapStart;
+            U_ASSERT(mapIndex < (int32_t)sizeof(UTF8Buf::mapToUChars));
+            ut->chunkOffset = u8b->mapToUChars[mapIndex] - u8b->bufStartIdx;
+            return true;
+
+        }
+    }
+
+
+    //
+    // Dispatch to the appropriate action for a
+    //   Backwards Direction iteration request.
+    //
+    if (ix==ut->chunkNativeStart) {
+        // Check for normal sequential iteration cases first.
+        if (ix==0) {
+            // Just reached the start of string
+            // Don't swap buffers, but do set the
+            //   current buffer position.
+            ut->chunkOffset = 0;
+            return false;
+        } else {
+            // Start of current buffer.
+            //   check whether other buffer already has what we need.
+            UTF8Buf *altB = (UTF8Buf *)ut->q;
+            if (ix>altB->bufNativeStart && ix<=altB->bufNativeLimit) {
+                goto swapBuffers;
+            }
+        }
+    }
+
+    // A random access.  Desired index could be in either or niether buf.
+    // For optimizing the order of testing,
+    //    Most likely case:  in the other buffer.
+    //    Second most likely: in neither buffer.
+    //    Unlikely, but must work:  in the current buffer.
+    u8b = (UTF8Buf *)ut->q;   // the alternate buffer
+    if (ix>u8b->bufNativeStart && ix<=u8b->bufNativeLimit) {
+        // Requested index is in the other buffer.
+        goto swapBuffers;
+    }
+    // Requested index is start-of-string.
+    //   (this is the case of randomly seeking to the start.
+    //    The case of iterating off the start is handled earlier.)
+    if (ix==0) {
+        if (u8b->bufNativeStart==0) {
+            // Alternate buffer contains the data for the start string.
+            // Make it be the current buffer.
+            goto swapBuffersAndFail;
+        } else {
+            // Request for data before the start of string,
+            //   neither buffer is usable.
+            //   set up a zero-length buffer.
+            goto makeStubBuffer;
+        }
+    }
+
+    if (ix<=ut->chunkNativeStart || ix>ut->chunkNativeLimit) {
+        // Requested index is in neither buffer.
+        goto fillReverse;
+    }
+
+    // Requested index is in this buffer.
+    //   Set the utf16 buffer index.
+    u8b = (UTF8Buf *)ut->p;
+    mapIndex = ix - u8b->toUCharsMapStart;
+    ut->chunkOffset = u8b->mapToUChars[mapIndex] - u8b->bufStartIdx;
+    if (ut->chunkOffset==0) {
+        // This occurs when the first character in the text is
+        //   a multi-byte UTF-8 char, and the requested index is to
+        //   one of the trailing bytes.  Because there is no preceding ,
+        //   character, this access fails.  We can't pick up on the
+        //   situation sooner because the requested index is not zero.
+        return false;
+    } else {
+        return true;
+    }
+
+
+
+swapBuffers:
+    //  The alternate buffer (ut->q) has the string data that was requested.
+    //  Swap the primary and alternate buffers, and set the
+    //   chunk index into the new primary buffer.
+    {
+        u8b   = (UTF8Buf *)ut->q;
+        ut->q = ut->p;
+        ut->p = u8b;
+        ut->chunkContents       = &u8b->buf[u8b->bufStartIdx];
+        ut->chunkLength         = u8b->bufLimitIdx - u8b->bufStartIdx;
+        ut->chunkNativeStart    = u8b->bufNativeStart;
+        ut->chunkNativeLimit    = u8b->bufNativeLimit;
+        ut->nativeIndexingLimit = u8b->bufNILimit;
+
+        // Index into the (now current) chunk
+        // Use the map to set the chunk index.  It's more trouble than it's worth
+        //    to check whether native indexing can be used.
+        U_ASSERT(ix>=u8b->bufNativeStart);
+        U_ASSERT(ix<=u8b->bufNativeLimit);
+        mapIndex = ix - u8b->toUCharsMapStart;
+        U_ASSERT(mapIndex>=0);
+        U_ASSERT(mapIndex<(int32_t)sizeof(u8b->mapToUChars));
+        ut->chunkOffset = u8b->mapToUChars[mapIndex] - u8b->bufStartIdx;
+
+        return true;
+    }
+
+
+ swapBuffersAndFail:
+    // We got a request for either the start or end of the string,
+    //  with iteration continuing in the out-of-bounds direction.
+    // The alternate buffer already contains the data up to the
+    //  start/end.
+    // Swap the buffers, then return failure, indicating that we couldn't
+    //  make things correct for continuing the iteration in the requested
+    //  direction.  The position & buffer are correct should the
+    //  user decide to iterate in the opposite direction.
+    u8b   = (UTF8Buf *)ut->q;
+    ut->q = ut->p;
+    ut->p = u8b;
+    ut->chunkContents       = &u8b->buf[u8b->bufStartIdx];
+    ut->chunkLength         = u8b->bufLimitIdx - u8b->bufStartIdx;
+    ut->chunkNativeStart    = u8b->bufNativeStart;
+    ut->chunkNativeLimit    = u8b->bufNativeLimit;
+    ut->nativeIndexingLimit = u8b->bufNILimit;
+
+    // Index into the (now current) chunk
+    //  For this function  (swapBuffersAndFail), the requested index
+    //    will always be at either the start or end of the chunk.
+    if (ix==u8b->bufNativeLimit) {
+        ut->chunkOffset = ut->chunkLength;
+    } else  {
+        ut->chunkOffset = 0;
+        U_ASSERT(ix == u8b->bufNativeStart);
+    }
+    return false;
+
+makeStubBuffer:
+    //   The user has done a seek/access past the start or end
+    //   of the string.  Rather than loading data that is likely
+    //   to never be used, just set up a zero-length buffer at
+    //   the position.
+    u8b = (UTF8Buf *)ut->q;
+    u8b->bufNativeStart   = ix;
+    u8b->bufNativeLimit   = ix;
+    u8b->bufStartIdx      = 0;
+    u8b->bufLimitIdx      = 0;
+    u8b->bufNILimit       = 0;
+    u8b->toUCharsMapStart = ix;
+    u8b->mapToNative[0]   = 0;
+    u8b->mapToUChars[0]   = 0;
+    goto swapBuffersAndFail;
+
+
+
+fillForward:
+    {
+        // Move the incoming index to a code point boundary.
+        U8_SET_CP_START(s8, 0, ix);
+
+        // Swap the UText buffers.
+        //  We want to fill what was previously the alternate buffer,
+        //  and make what was the current buffer be the new alternate.
+        UTF8Buf *u8b_swap = (UTF8Buf *)ut->q;
+        ut->q = ut->p;
+        ut->p = u8b_swap;
+
+        int32_t strLen = ut->b;
+        UBool   nulTerminated = false;
+        if (strLen < 0) {
+            strLen = 0x7fffffff;
+            nulTerminated = true;
+        }
+
+        char16_t   *buf = u8b_swap->buf;
+        uint8_t *mapToNative  = u8b_swap->mapToNative;
+        uint8_t *mapToUChars  = u8b_swap->mapToUChars;
+        int32_t  destIx       = 0;
+        int32_t  srcIx        = ix;
+        UBool    seenNonAscii = false;
+        UChar32  c = 0;
+
+        // Fill the chunk buffer and mapping arrays.
+        while (destIx0 && c<0x80) {
+                // Special case ASCII range for speed.
+                //   zero is excluded to simplify bounds checking.
+                buf[destIx] = (char16_t)c;
+                mapToNative[destIx]    = (uint8_t)(srcIx - ix);
+                mapToUChars[srcIx-ix]  = (uint8_t)destIx;
+                srcIx++;
+                destIx++;
+            } else {
+                // General case, handle everything.
+                if (seenNonAscii == false) {
+                    seenNonAscii = true;
+                    u8b_swap->bufNILimit = destIx;
+                }
+
+                int32_t  cIx      = srcIx;
+                int32_t  dIx      = destIx;
+                int32_t  dIxSaved = destIx;
+                U8_NEXT_OR_FFFD(s8, srcIx, strLen, c);
+                if (c==0 && nulTerminated) {
+                    srcIx--;
+                    break;
+                }
+
+                U16_APPEND_UNSAFE(buf, destIx, c);
+                do {
+                    mapToNative[dIx++] = (uint8_t)(cIx - ix);
+                } while (dIx < destIx);
+
+                do {
+                    mapToUChars[cIx++ - ix] = (uint8_t)dIxSaved;
+                } while (cIx < srcIx);
+            }
+            if (srcIx>=strLen) {
+                break;
+            }
+
+        }
+
+        //  store Native <--> Chunk Map entries for the end of the buffer.
+        //    There is no actual character here, but the index position is valid.
+        mapToNative[destIx]     = (uint8_t)(srcIx - ix);
+        mapToUChars[srcIx - ix] = (uint8_t)destIx;
+
+        //  fill in Buffer descriptor
+        u8b_swap->bufNativeStart     = ix;
+        u8b_swap->bufNativeLimit     = srcIx;
+        u8b_swap->bufStartIdx        = 0;
+        u8b_swap->bufLimitIdx        = destIx;
+        if (seenNonAscii == false) {
+            u8b_swap->bufNILimit     = destIx;
+        }
+        u8b_swap->toUCharsMapStart   = u8b_swap->bufNativeStart;
+
+        // Set UText chunk to refer to this buffer.
+        ut->chunkContents       = buf;
+        ut->chunkOffset         = 0;
+        ut->chunkLength         = u8b_swap->bufLimitIdx;
+        ut->chunkNativeStart    = u8b_swap->bufNativeStart;
+        ut->chunkNativeLimit    = u8b_swap->bufNativeLimit;
+        ut->nativeIndexingLimit = u8b_swap->bufNILimit;
+
+        // For zero terminated strings, keep track of the maximum point
+        //   scanned so far.
+        if (nulTerminated && srcIx>ut->c) {
+            ut->c = srcIx;
+            if (c==0) {
+                // We scanned to the end.
+                //   Remember the actual length.
+                ut->b = srcIx;
+                ut->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE);
+            }
+        }
+        return true;
+    }
+
+
+fillReverse:
+    {
+        // Move the incoming index to a code point boundary.
+        // Can only do this if the incoming index is somewhere in the interior of the string.
+        //   If index is at the end, there is no character there to look at.
+        if (ix != ut->b) {
+            // Note: this function will only move the index back if it is on a trail byte
+            //       and there is a preceding lead byte and the sequence from the lead 
+            //       through this trail could be part of a valid UTF-8 sequence
+            //       Otherwise the index remains unchanged.
+            U8_SET_CP_START(s8, 0, ix);
+        }
+
+        // Swap the UText buffers.
+        //  We want to fill what was previously the alternate buffer,
+        //  and make what was the current buffer be the new alternate.
+        UTF8Buf *u8b_swap = (UTF8Buf *)ut->q;
+        ut->q = ut->p;
+        ut->p = u8b_swap;
+
+        char16_t   *buf = u8b_swap->buf;
+        uint8_t *mapToNative = u8b_swap->mapToNative;
+        uint8_t *mapToUChars = u8b_swap->mapToUChars;
+        int32_t  toUCharsMapStart = ix - sizeof(UTF8Buf::mapToUChars) + 1;
+        // Note that toUCharsMapStart can be negative. Happens when the remaining
+        // text from current position to the beginning is less than the buffer size.
+        // + 1 because mapToUChars must have a slot at the end for the bufNativeLimit entry.
+        int32_t  destIx = UTF8_TEXT_CHUNK_SIZE+2;   // Start in the overflow region
+                                                    //   at end of buffer to leave room
+                                                    //   for a surrogate pair at the
+                                                    //   buffer start.
+        int32_t  srcIx  = ix;
+        int32_t  bufNILimit = destIx;
+        UChar32   c;
+
+        // Map to/from Native Indexes, fill in for the position at the end of
+        //   the buffer.
+        //
+        mapToNative[destIx] = (uint8_t)(srcIx - toUCharsMapStart);
+        mapToUChars[srcIx - toUCharsMapStart] = (uint8_t)destIx;
+
+        // Fill the chunk buffer
+        // Work backwards, filling from the end of the buffer towards the front.
+        //
+        while (destIx>2 && (srcIx - toUCharsMapStart > 5) && (srcIx > 0)) {
+            srcIx--;
+            destIx--;
+
+            // Get last byte of the UTF-8 character
+            c = s8[srcIx];
+            if (c<0x80) {
+                // Special case ASCII range for speed.
+                buf[destIx] = (char16_t)c;
+                U_ASSERT(toUCharsMapStart <= srcIx);
+                mapToUChars[srcIx - toUCharsMapStart] = (uint8_t)destIx;
+                mapToNative[destIx] = (uint8_t)(srcIx - toUCharsMapStart);
+            } else {
+                // General case, handle everything non-ASCII.
+
+                int32_t  sIx      = srcIx;  // ix of last byte of multi-byte u8 char
+
+                // Get the full character from the UTF8 string.
+                //   use code derived from the macros in utf8.h
+                //   Leaves srcIx pointing at the first byte of the UTF-8 char.
+                //
+                c=utf8_prevCharSafeBody(s8, 0, &srcIx, c, -3);
+                // leaves srcIx at first byte of the multi-byte char.
+
+                // Store the character in UTF-16 buffer.
+                if (c<0x10000) {
+                    buf[destIx] = (char16_t)c;
+                    mapToNative[destIx] = (uint8_t)(srcIx - toUCharsMapStart);
+                } else {
+                    buf[destIx]         = U16_TRAIL(c);
+                    mapToNative[destIx] = (uint8_t)(srcIx - toUCharsMapStart);
+                    buf[--destIx]       = U16_LEAD(c);
+                    mapToNative[destIx] = (uint8_t)(srcIx - toUCharsMapStart);
+                }
+
+                // Fill in the map from native indexes to UChars buf index.
+                do {
+                    mapToUChars[sIx-- - toUCharsMapStart] = (uint8_t)destIx;
+                } while (sIx >= srcIx);
+                U_ASSERT(toUCharsMapStart <= (srcIx+1));
+
+                // Set native indexing limit to be the current position.
+                //   We are processing a non-ascii, non-native-indexing char now;
+                //     the limit will be here if the rest of the chars to be
+                //     added to this buffer are ascii.
+                bufNILimit = destIx;
+            }
+        }
+        u8b_swap->bufNativeStart     = srcIx;
+        u8b_swap->bufNativeLimit     = ix;
+        u8b_swap->bufStartIdx        = destIx;
+        u8b_swap->bufLimitIdx        = UTF8_TEXT_CHUNK_SIZE+2;
+        u8b_swap->bufNILimit         = bufNILimit - u8b_swap->bufStartIdx;
+        u8b_swap->toUCharsMapStart   = toUCharsMapStart;
+
+        ut->chunkContents       = &buf[u8b_swap->bufStartIdx];
+        ut->chunkLength         = u8b_swap->bufLimitIdx - u8b_swap->bufStartIdx;
+        ut->chunkOffset         = ut->chunkLength;
+        ut->chunkNativeStart    = u8b_swap->bufNativeStart;
+        ut->chunkNativeLimit    = u8b_swap->bufNativeLimit;
+        ut->nativeIndexingLimit = u8b_swap->bufNILimit;
+        return true;
+    }
+
+}
+
+
+
+//
+//  This is a slightly modified copy of u_strFromUTF8,
+//     Inserts a Replacement Char rather than failing on invalid UTF-8
+//     Removes unnecessary features.
+//
+static char16_t*
+utext_strFromUTF8(char16_t *dest,
+              int32_t destCapacity,
+              int32_t *pDestLength,
+              const char* src,
+              int32_t srcLength,        // required.  NUL terminated not supported.
+              UErrorCode *pErrorCode
+              )
+{
+
+    char16_t *pDest = dest;
+    char16_t *pDestLimit = (dest!=nullptr)?(dest+destCapacity):nullptr;
+    UChar32 ch=0;
+    int32_t index = 0;
+    int32_t reqLength = 0;
+    uint8_t* pSrc = (uint8_t*) src;
+
+
+    while((index < srcLength)&&(pDest0)) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    int32_t  length  = ut->b;
+    int32_t  start32 = pinIndex(start, length);
+    int32_t  limit32 = pinIndex(limit, length);
+
+    if(start32>limit32) {
+        *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+        return 0;
+    }
+
+
+    // adjust the incoming indexes to land on code point boundaries if needed.
+    //    adjust by no more than three, because that is the largest number of trail bytes
+    //    in a well formed UTF8 character.
+    const uint8_t *buf = (const uint8_t *)ut->context;
+    int i;
+    if (start32 < ut->chunkNativeLimit) {
+        for (i=0; i<3; i++) {
+            if (U8_IS_SINGLE(buf[start32]) || U8_IS_LEAD(buf[start32]) || start32==0) {
+                break;
+            }
+            start32--;
+        }
+    }
+
+    if (limit32 < ut->chunkNativeLimit) {
+        for (i=0; i<3; i++) {
+            if (U8_IS_SINGLE(buf[limit32]) || U8_IS_LEAD(buf[limit32]) || limit32==0) {
+                break;
+            }
+            limit32--;
+        }
+    }
+
+    // Do the actual extract.
+    int32_t destLength=0;
+    utext_strFromUTF8(dest, destCapacity, &destLength,
+                    (const char *)ut->context+start32, limit32-start32,
+                    pErrorCode);
+    utf8TextAccess(ut, limit32, true);
+    return destLength;
+}
+
+//
+// utf8TextMapOffsetToNative
+//
+// Map a chunk (UTF-16) offset to a native index.
+static int64_t U_CALLCONV
+utf8TextMapOffsetToNative(const UText *ut) {
+    //
+    UTF8Buf *u8b = (UTF8Buf *)ut->p;
+    U_ASSERT(ut->chunkOffset>ut->nativeIndexingLimit && ut->chunkOffset<=ut->chunkLength);
+    int32_t nativeOffset = u8b->mapToNative[ut->chunkOffset + u8b->bufStartIdx] + u8b->toUCharsMapStart;
+    U_ASSERT(nativeOffset >= ut->chunkNativeStart && nativeOffset <= ut->chunkNativeLimit);
+    return nativeOffset;
+}
+
+//
+// Map a native index to the corresponding chunk offset
+//
+static int32_t U_CALLCONV
+utf8TextMapIndexToUTF16(const UText *ut, int64_t index64) {
+    U_ASSERT(index64 <= 0x7fffffff);
+    int32_t index = (int32_t)index64;
+    UTF8Buf *u8b = (UTF8Buf *)ut->p;
+    U_ASSERT(index>=ut->chunkNativeStart+ut->nativeIndexingLimit);
+    U_ASSERT(index<=ut->chunkNativeLimit);
+    int32_t mapIndex = index - u8b->toUCharsMapStart;
+    U_ASSERT(mapIndex < (int32_t)sizeof(UTF8Buf::mapToUChars));
+    int32_t offset = u8b->mapToUChars[mapIndex] - u8b->bufStartIdx;
+    U_ASSERT(offset>=0 && offset<=ut->chunkLength);
+    return offset;
+}
+
+static UText * U_CALLCONV
+utf8TextClone(UText *dest, const UText *src, UBool deep, UErrorCode *status)
+{
+    // First do a generic shallow clone.  Does everything needed for the UText struct itself.
+    dest = shallowTextClone(dest, src, status);
+
+    // For deep clones, make a copy of the string.
+    //  The copied storage is owned by the newly created clone.
+    //
+    // TODO:  There is an issue with using utext_nativeLength().
+    //        That function is non-const in cases where the input was NUL terminated
+    //          and the length has not yet been determined.
+    //        This function (clone()) is const.
+    //        There potentially a thread safety issue lurking here.
+    //
+    if (deep && U_SUCCESS(*status)) {
+        int32_t  len = (int32_t)utext_nativeLength((UText *)src);
+        char *copyStr = (char *)uprv_malloc(len+1);
+        if (copyStr == nullptr) {
+            *status = U_MEMORY_ALLOCATION_ERROR;
+        } else {
+            uprv_memcpy(copyStr, src->context, len+1);
+            dest->context = copyStr;
+            dest->providerProperties |= I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT);
+        }
+    }
+    return dest;
+}
+
+
+static void U_CALLCONV
+utf8TextClose(UText *ut) {
+    // Most of the work of close is done by the generic UText framework close.
+    // All that needs to be done here is to delete the UTF8 string if the UText
+    //  owns it.  This occurs if the UText was created by cloning.
+    if (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT)) {
+        char *s = (char *)ut->context;
+        uprv_free(s);
+        ut->context = nullptr;
+    }
+}
+
+U_CDECL_END
+
+
+static const struct UTextFuncs utf8Funcs =
+{
+    sizeof(UTextFuncs),
+    0, 0, 0,             // Reserved alignment padding
+    utf8TextClone,
+    utf8TextLength,
+    utf8TextAccess,
+    utf8TextExtract,
+    nullptr,                /* replace*/
+    nullptr,                /* copy   */
+    utf8TextMapOffsetToNative,
+    utf8TextMapIndexToUTF16,
+    utf8TextClose,
+    nullptr,                // spare 1
+    nullptr,                // spare 2
+    nullptr                 // spare 3
+};
+
+
+static const char gEmptyString[] = {0};
+
+U_CAPI UText * U_EXPORT2
+utext_openUTF8(UText *ut, const char *s, int64_t length, UErrorCode *status) {
+    if(U_FAILURE(*status)) {
+        return nullptr;
+    }
+    if(s==nullptr && length==0) {
+        s = gEmptyString;
+    }
+
+    if(s==nullptr || length<-1 || length>INT32_MAX) {
+        *status=U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+
+    ut = utext_setup(ut, sizeof(UTF8Buf) * 2, status);
+    if (U_FAILURE(*status)) {
+        return ut;
+    }
+
+    ut->pFuncs  = &utf8Funcs;
+    ut->context = s;
+    ut->b       = (int32_t)length;
+    ut->c       = (int32_t)length;
+    if (ut->c < 0) {
+        ut->c = 0;
+        ut->providerProperties |= I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE);
+    }
+    ut->p = ut->pExtra;
+    ut->q = (char *)ut->pExtra + sizeof(UTF8Buf);
+    return ut;
+
+}
+
+
+
+
+
+
+
+
+//------------------------------------------------------------------------------
+//
+//     UText implementation wrapper for Replaceable (read/write)
+//
+//         Use of UText data members:
+//            context    pointer to Replaceable.
+//            p          pointer to Replaceable if it is owned by the UText.
+//
+//------------------------------------------------------------------------------
+
+
+
+// minimum chunk size for this implementation: 3
+// to allow for possible trimming for code point boundaries
+enum { REP_TEXT_CHUNK_SIZE=10 };
+
+struct ReplExtra {
+    /*
+     * Chunk UChars.
+     * +1 to simplify filling with surrogate pair at the end.
+     */
+    char16_t s[REP_TEXT_CHUNK_SIZE+1];
+};
+
+
+U_CDECL_BEGIN
+
+static UText * U_CALLCONV
+repTextClone(UText *dest, const UText *src, UBool deep, UErrorCode *status) {
+    // First do a generic shallow clone.  Does everything needed for the UText struct itself.
+    dest = shallowTextClone(dest, src, status);
+
+    // For deep clones, make a copy of the Replaceable.
+    //  The copied Replaceable storage is owned by the newly created UText clone.
+    //  A non-nullptr pointer in UText.p is the signal to the close() function to delete
+    //    it.
+    //
+    if (deep && U_SUCCESS(*status)) {
+        const Replaceable *replSrc = (const Replaceable *)src->context;
+        dest->context = replSrc->clone();
+        dest->providerProperties |= I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT);
+
+        // with deep clone, the copy is writable, even when the source is not.
+        dest->providerProperties |= I32_FLAG(UTEXT_PROVIDER_WRITABLE);
+    }
+    return dest;
+}
+
+
+static void U_CALLCONV
+repTextClose(UText *ut) {
+    // Most of the work of close is done by the generic UText framework close.
+    // All that needs to be done here is delete the Replaceable if the UText
+    //  owns it.  This occurs if the UText was created by cloning.
+    if (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT)) {
+        Replaceable *rep = (Replaceable *)ut->context;
+        delete rep;
+        ut->context = nullptr;
+    }
+}
+
+
+static int64_t U_CALLCONV
+repTextLength(UText *ut) {
+    const Replaceable *replSrc = (const Replaceable *)ut->context;
+    int32_t  len = replSrc->length();
+    return len;
+}
+
+
+static UBool U_CALLCONV
+repTextAccess(UText *ut, int64_t index, UBool forward) {
+    const Replaceable *rep=(const Replaceable *)ut->context;
+    int32_t length=rep->length();   // Full length of the input text (bigger than a chunk)
+
+    // clip the requested index to the limits of the text.
+    int32_t index32 = pinIndex(index, length);
+    U_ASSERT(index<=INT32_MAX);
+
+
+    /*
+     * Compute start/limit boundaries around index, for a segment of text
+     * to be extracted.
+     * To allow for the possibility that our user gave an index to the trailing
+     * half of a surrogate pair, we must request one extra preceding char16_t when
+     * going in the forward direction.  This will ensure that the buffer has the
+     * entire code point at the specified index.
+     */
+    if(forward) {
+
+        if (index32>=ut->chunkNativeStart && index32chunkNativeLimit) {
+            // Buffer already contains the requested position.
+            ut->chunkOffset = (int32_t)(index - ut->chunkNativeStart);
+            return true;
+        }
+        if (index32>=length && ut->chunkNativeLimit==length) {
+            // Request for end of string, and buffer already extends up to it.
+            // Can't get the data, but don't change the buffer.
+            ut->chunkOffset = length - (int32_t)ut->chunkNativeStart;
+            return false;
+        }
+
+        ut->chunkNativeLimit = index + REP_TEXT_CHUNK_SIZE - 1;
+        // Going forward, so we want to have the buffer with stuff at and beyond
+        //   the requested index.  The -1 gets us one code point before the
+        //   requested index also, to handle the case of the index being on
+        //   a trail surrogate of a surrogate pair.
+        if(ut->chunkNativeLimit > length) {
+            ut->chunkNativeLimit = length;
+        }
+        // unless buffer ran off end, start is index-1.
+        ut->chunkNativeStart = ut->chunkNativeLimit - REP_TEXT_CHUNK_SIZE;
+        if(ut->chunkNativeStart < 0) {
+            ut->chunkNativeStart = 0;
+        }
+    } else {
+        // Reverse iteration.  Fill buffer with data preceding the requested index.
+        if (index32>ut->chunkNativeStart && index32<=ut->chunkNativeLimit) {
+            // Requested position already in buffer.
+            ut->chunkOffset = index32 - (int32_t)ut->chunkNativeStart;
+            return true;
+        }
+        if (index32==0 && ut->chunkNativeStart==0) {
+            // Request for start, buffer already begins at start.
+            //  No data, but keep the buffer as is.
+            ut->chunkOffset = 0;
+            return false;
+        }
+
+        // Figure out the bounds of the chunk to extract for reverse iteration.
+        // Need to worry about chunk not splitting surrogate pairs, and while still
+        // containing the data we need.
+        // Fix by requesting a chunk that includes an extra char16_t at the end.
+        // If this turns out to be a lead surrogate, we can lop it off and still have
+        //   the data we wanted.
+        ut->chunkNativeStart = index32 + 1 - REP_TEXT_CHUNK_SIZE;
+        if (ut->chunkNativeStart < 0) {
+            ut->chunkNativeStart = 0;
+        }
+
+        ut->chunkNativeLimit = index32 + 1;
+        if (ut->chunkNativeLimit > length) {
+            ut->chunkNativeLimit = length;
+        }
+    }
+
+    // Extract the new chunk of text from the Replaceable source.
+    ReplExtra *ex = (ReplExtra *)ut->pExtra;
+    // UnicodeString with its buffer a writable alias to the chunk buffer
+    UnicodeString buffer(ex->s, 0 /*buffer length*/, REP_TEXT_CHUNK_SIZE /*buffer capacity*/);
+    rep->extractBetween((int32_t)ut->chunkNativeStart, (int32_t)ut->chunkNativeLimit, buffer);
+
+    ut->chunkContents  = ex->s;
+    ut->chunkLength    = (int32_t)(ut->chunkNativeLimit - ut->chunkNativeStart);
+    ut->chunkOffset    = (int32_t)(index32 - ut->chunkNativeStart);
+
+    // Surrogate pairs from the input text must not span chunk boundaries.
+    // If end of chunk could be the start of a surrogate, trim it off.
+    if (ut->chunkNativeLimit < length &&
+        U16_IS_LEAD(ex->s[ut->chunkLength-1])) {
+            ut->chunkLength--;
+            ut->chunkNativeLimit--;
+            if (ut->chunkOffset > ut->chunkLength) {
+                ut->chunkOffset = ut->chunkLength;
+            }
+        }
+
+    // if the first char16_t in the chunk could be the trailing half of a surrogate pair,
+    // trim it off.
+    if(ut->chunkNativeStart>0 && U16_IS_TRAIL(ex->s[0])) {
+        ++(ut->chunkContents);
+        ++(ut->chunkNativeStart);
+        --(ut->chunkLength);
+        --(ut->chunkOffset);
+    }
+
+    // adjust the index/chunkOffset to a code point boundary
+    U16_SET_CP_START(ut->chunkContents, 0, ut->chunkOffset);
+
+    // Use fast indexing for get/setNativeIndex()
+    ut->nativeIndexingLimit = ut->chunkLength;
+
+    return true;
+}
+
+
+
+static int32_t U_CALLCONV
+repTextExtract(UText *ut,
+               int64_t start, int64_t limit,
+               char16_t *dest, int32_t destCapacity,
+               UErrorCode *status) {
+    const Replaceable *rep=(const Replaceable *)ut->context;
+    int32_t  length=rep->length();
+
+    if(U_FAILURE(*status)) {
+        return 0;
+    }
+    if(destCapacity<0 || (dest==nullptr && destCapacity>0)) {
+        *status=U_ILLEGAL_ARGUMENT_ERROR;
+    }
+    if(start>limit) {
+        *status=U_INDEX_OUTOFBOUNDS_ERROR;
+        return 0;
+    }
+
+    int32_t  start32 = pinIndex(start, length);
+    int32_t  limit32 = pinIndex(limit, length);
+
+    // adjust start, limit if they point to trail half of surrogates
+    if (start32charAt(start32)) &&
+        U_IS_SUPPLEMENTARY(rep->char32At(start32))){
+            start32--;
+    }
+    if (limit32charAt(limit32)) &&
+        U_IS_SUPPLEMENTARY(rep->char32At(limit32))){
+            limit32--;
+    }
+
+    length=limit32-start32;
+    if(length>destCapacity) {
+        limit32 = start32 + destCapacity;
+    }
+    UnicodeString buffer(dest, 0, destCapacity); // writable alias
+    rep->extractBetween(start32, limit32, buffer);
+    repTextAccess(ut, limit32, true);
+
+    return u_terminateUChars(dest, destCapacity, length, status);
+}
+
+static int32_t U_CALLCONV
+repTextReplace(UText *ut,
+               int64_t start, int64_t limit,
+               const char16_t *src, int32_t length,
+               UErrorCode *status) {
+    Replaceable *rep=(Replaceable *)ut->context;
+    int32_t oldLength;
+
+    if(U_FAILURE(*status)) {
+        return 0;
+    }
+    if(src==nullptr && length!=0) {
+        *status=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    oldLength=rep->length(); // will subtract from new length
+    if(start>limit ) {
+        *status=U_INDEX_OUTOFBOUNDS_ERROR;
+        return 0;
+    }
+
+    int32_t start32 = pinIndex(start, oldLength);
+    int32_t limit32 = pinIndex(limit, oldLength);
+
+    // Snap start & limit to code point boundaries.
+    if (start32charAt(start32)) &&
+        start32>0 && U16_IS_LEAD(rep->charAt(start32-1)))
+    {
+            start32--;
+    }
+    if (limit32charAt(limit32-1)) &&
+        U16_IS_TRAIL(rep->charAt(limit32)))
+    {
+            limit32++;
+    }
+
+    // Do the actual replace operation using methods of the Replaceable class
+    UnicodeString replStr((UBool)(length<0), src, length); // read-only alias
+    rep->handleReplaceBetween(start32, limit32, replStr);
+    int32_t newLength = rep->length();
+    int32_t lengthDelta = newLength - oldLength;
+
+    // Is the UText chunk buffer OK?
+    if (ut->chunkNativeLimit > start32) {
+        // this replace operation may have impacted the current chunk.
+        // invalidate it, which will force a reload on the next access.
+        invalidateChunk(ut);
+    }
+
+    // set the iteration position to the end of the newly inserted replacement text.
+    int32_t newIndexPos = limit32 + lengthDelta;
+    repTextAccess(ut, newIndexPos, true);
+
+    return lengthDelta;
+}
+
+
+static void U_CALLCONV
+repTextCopy(UText *ut,
+                int64_t start, int64_t limit,
+                int64_t destIndex,
+                UBool move,
+                UErrorCode *status)
+{
+    Replaceable *rep=(Replaceable *)ut->context;
+    int32_t length=rep->length();
+
+    if(U_FAILURE(*status)) {
+        return;
+    }
+    if (start>limit || (startcopy(start32, limit32, destIndex32);
+        if(destIndex32handleReplaceBetween(start32, limit32, UnicodeString());
+    } else {
+        // copy
+        rep->copy(start32, limit32, destIndex32);
+    }
+
+    // If the change to the text touched the region in the chunk buffer,
+    //  invalidate the buffer.
+    int32_t firstAffectedIndex = destIndex32;
+    if (move && start32chunkNativeLimit) {
+        // changes may have affected range covered by the chunk
+        invalidateChunk(ut);
+    }
+
+    // Put iteration position at the newly inserted (moved) block,
+    int32_t  nativeIterIndex = destIndex32 + limit32 - start32;
+    if (move && destIndex32>start32) {
+        // moved a block of text towards the end of the string.
+        nativeIterIndex = destIndex32;
+    }
+
+    // Set position, reload chunk if needed.
+    repTextAccess(ut, nativeIterIndex, true);
+}
+
+static const struct UTextFuncs repFuncs =
+{
+    sizeof(UTextFuncs),
+    0, 0, 0,           // Reserved alignment padding
+    repTextClone,
+    repTextLength,
+    repTextAccess,
+    repTextExtract,
+    repTextReplace,
+    repTextCopy,
+    nullptr,              // MapOffsetToNative,
+    nullptr,              // MapIndexToUTF16,
+    repTextClose,
+    nullptr,              // spare 1
+    nullptr,              // spare 2
+    nullptr               // spare 3
+};
+
+
+U_CAPI UText * U_EXPORT2
+utext_openReplaceable(UText *ut, Replaceable *rep, UErrorCode *status)
+{
+    if(U_FAILURE(*status)) {
+        return nullptr;
+    }
+    if(rep==nullptr) {
+        *status=U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+    ut = utext_setup(ut, sizeof(ReplExtra), status);
+    if(U_FAILURE(*status)) {
+        return ut;
+    }
+
+    ut->providerProperties = I32_FLAG(UTEXT_PROVIDER_WRITABLE);
+    if(rep->hasMetaData()) {
+        ut->providerProperties |=I32_FLAG(UTEXT_PROVIDER_HAS_META_DATA);
+    }
+
+    ut->pFuncs  = &repFuncs;
+    ut->context =  rep;
+    return ut;
+}
+
+U_CDECL_END
+
+
+
+
+
+
+
+
+//------------------------------------------------------------------------------
+//
+//     UText implementation for UnicodeString (read/write)  and
+//                    for const UnicodeString (read only)
+//             (same implementation, only the flags are different)
+//
+//         Use of UText data members:
+//            context    pointer to UnicodeString
+//            p          pointer to UnicodeString IF this UText owns the string
+//                       and it must be deleted on close().  nullptr otherwise.
+//
+//------------------------------------------------------------------------------
+
+U_CDECL_BEGIN
+
+
+static UText * U_CALLCONV
+unistrTextClone(UText *dest, const UText *src, UBool deep, UErrorCode *status) {
+    // First do a generic shallow clone.  Does everything needed for the UText struct itself.
+    dest = shallowTextClone(dest, src, status);
+
+    // For deep clones, make a copy of the UnicodeSring.
+    //  The copied UnicodeString storage is owned by the newly created UText clone.
+    //  A non-nullptr pointer in UText.p is the signal to the close() function to delete
+    //    the UText.
+    //
+    if (deep && U_SUCCESS(*status)) {
+        const UnicodeString *srcString = (const UnicodeString *)src->context;
+        dest->context = new UnicodeString(*srcString);
+        dest->providerProperties |= I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT);
+
+        // with deep clone, the copy is writable, even when the source is not.
+        dest->providerProperties |= I32_FLAG(UTEXT_PROVIDER_WRITABLE);
+    }
+    return dest;
+}
+
+static void U_CALLCONV
+unistrTextClose(UText *ut) {
+    // Most of the work of close is done by the generic UText framework close.
+    // All that needs to be done here is delete the UnicodeString if the UText
+    //  owns it.  This occurs if the UText was created by cloning.
+    if (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT)) {
+        UnicodeString *str = (UnicodeString *)ut->context;
+        delete str;
+        ut->context = nullptr;
+    }
+}
+
+
+static int64_t U_CALLCONV
+unistrTextLength(UText *t) {
+    return ((const UnicodeString *)t->context)->length();
+}
+
+
+static UBool U_CALLCONV
+unistrTextAccess(UText *ut, int64_t index, UBool  forward) {
+    int32_t length  = ut->chunkLength;
+    ut->chunkOffset = pinIndex(index, length);
+
+    // Check whether request is at the start or end
+    UBool retVal = (forward && index0);
+    return retVal;
+}
+
+
+
+static int32_t U_CALLCONV
+unistrTextExtract(UText *t,
+                  int64_t start, int64_t limit,
+                  char16_t *dest, int32_t destCapacity,
+                  UErrorCode *pErrorCode) {
+    const UnicodeString *us=(const UnicodeString *)t->context;
+    int32_t length=us->length();
+
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+    if(destCapacity<0 || (dest==nullptr && destCapacity>0)) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+    }
+    if(start<0 || start>limit) {
+        *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+        return 0;
+    }
+
+    int32_t start32 = startgetChar32Start((int32_t)start) : length;
+    int32_t limit32 = limitgetChar32Start((int32_t)limit) : length;
+
+    length=limit32-start32;
+    if (destCapacity>0 && dest!=nullptr) {
+        int32_t trimmedLength = length;
+        if(trimmedLength>destCapacity) {
+            trimmedLength=destCapacity;
+        }
+        us->extract(start32, trimmedLength, dest);
+        t->chunkOffset = start32+trimmedLength;
+    } else {
+        t->chunkOffset = start32;
+    }
+    u_terminateUChars(dest, destCapacity, length, pErrorCode);
+    return length;
+}
+
+static int32_t U_CALLCONV
+unistrTextReplace(UText *ut,
+                  int64_t start, int64_t limit,
+                  const char16_t *src, int32_t length,
+                  UErrorCode *pErrorCode) {
+    UnicodeString *us=(UnicodeString *)ut->context;
+    int32_t oldLength;
+
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+    if(src==nullptr && length!=0) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+    }
+    if(start>limit) {
+        *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+        return 0;
+    }
+    oldLength=us->length();
+    int32_t start32 = pinIndex(start, oldLength);
+    int32_t limit32 = pinIndex(limit, oldLength);
+    if (start32 < oldLength) {
+        start32 = us->getChar32Start(start32);
+    }
+    if (limit32 < oldLength) {
+        limit32 = us->getChar32Start(limit32);
+    }
+
+    // replace
+    us->replace(start32, limit32-start32, src, length);
+    int32_t newLength = us->length();
+
+    // Update the chunk description.
+    ut->chunkContents    = us->getBuffer();
+    ut->chunkLength      = newLength;
+    ut->chunkNativeLimit = newLength;
+    ut->nativeIndexingLimit = newLength;
+
+    // Set iteration position to the point just following the newly inserted text.
+    int32_t lengthDelta = newLength - oldLength;
+    ut->chunkOffset = limit32 + lengthDelta;
+
+    return lengthDelta;
+}
+
+static void U_CALLCONV
+unistrTextCopy(UText *ut,
+               int64_t start, int64_t limit,
+               int64_t destIndex,
+               UBool move,
+               UErrorCode *pErrorCode) {
+    UnicodeString *us=(UnicodeString *)ut->context;
+    int32_t length=us->length();
+
+    if(U_FAILURE(*pErrorCode)) {
+        return;
+    }
+    int32_t start32 = pinIndex(start, length);
+    int32_t limit32 = pinIndex(limit, length);
+    int32_t destIndex32 = pinIndex(destIndex, length);
+
+    if( start32>limit32 || (start32copy(start32, limit32, destIndex32);
+        if(destIndex32remove(start32, segLength);
+    } else {
+        // copy
+        us->copy(start32, limit32, destIndex32);
+    }
+
+    // update chunk description, set iteration position.
+    ut->chunkContents = us->getBuffer();
+    if (move==false) {
+        // copy operation, string length grows
+        ut->chunkLength += limit32-start32;
+        ut->chunkNativeLimit = ut->chunkLength;
+        ut->nativeIndexingLimit = ut->chunkLength;
+    }
+
+    // Iteration position to end of the newly inserted text.
+    ut->chunkOffset = destIndex32+limit32-start32;
+    if (move && destIndex32>start32) {
+        ut->chunkOffset = destIndex32;
+    }
+
+}
+
+static const struct UTextFuncs unistrFuncs =
+{
+    sizeof(UTextFuncs),
+    0, 0, 0,             // Reserved alignment padding
+    unistrTextClone,
+    unistrTextLength,
+    unistrTextAccess,
+    unistrTextExtract,
+    unistrTextReplace,
+    unistrTextCopy,
+    nullptr,                // MapOffsetToNative,
+    nullptr,                // MapIndexToUTF16,
+    unistrTextClose,
+    nullptr,                // spare 1
+    nullptr,                // spare 2
+    nullptr                 // spare 3
+};
+
+
+
+U_CDECL_END
+
+
+U_CAPI UText * U_EXPORT2
+utext_openUnicodeString(UText *ut, UnicodeString *s, UErrorCode *status) {
+    ut = utext_openConstUnicodeString(ut, s, status);
+    if (U_SUCCESS(*status)) {
+        ut->providerProperties |= I32_FLAG(UTEXT_PROVIDER_WRITABLE);
+    }
+    return ut;
+}
+
+
+
+U_CAPI UText * U_EXPORT2
+utext_openConstUnicodeString(UText *ut, const UnicodeString *s, UErrorCode *status) {
+    if (U_SUCCESS(*status) && s->isBogus()) {
+        // The UnicodeString is bogus, but we still need to detach the UText
+        //   from whatever it was hooked to before, if anything.
+        utext_openUChars(ut, nullptr, 0, status);
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return ut;
+    }
+    ut = utext_setup(ut, 0, status);
+    //    note:  use the standard (writable) function table for UnicodeString.
+    //           The flag settings disable writing, so having the functions in
+    //           the table is harmless.
+    if (U_SUCCESS(*status)) {
+        ut->pFuncs              = &unistrFuncs;
+        ut->context             = s;
+        ut->providerProperties  = I32_FLAG(UTEXT_PROVIDER_STABLE_CHUNKS);
+        ut->chunkContents       = s->getBuffer();
+        ut->chunkLength         = s->length();
+        ut->chunkNativeStart    = 0;
+        ut->chunkNativeLimit    = ut->chunkLength;
+        ut->nativeIndexingLimit = ut->chunkLength;
+    }
+    return ut;
+}
+
+//------------------------------------------------------------------------------
+//
+//     UText implementation for const char16_t * strings
+//
+//         Use of UText data members:
+//            context    pointer to UnicodeString
+//            a          length.  -1 if not yet known.
+//
+//         TODO:  support 64 bit lengths.
+//
+//------------------------------------------------------------------------------
+
+U_CDECL_BEGIN
+
+
+static UText * U_CALLCONV
+ucstrTextClone(UText *dest, const UText * src, UBool deep, UErrorCode * status) {
+    // First do a generic shallow clone.
+    dest = shallowTextClone(dest, src, status);
+
+    // For deep clones, make a copy of the string.
+    //  The copied storage is owned by the newly created clone.
+    //  A non-nullptr pointer in UText.p is the signal to the close() function to delete
+    //    it.
+    //
+    if (deep && U_SUCCESS(*status)) {
+        U_ASSERT(utext_nativeLength(dest) < INT32_MAX);
+        int32_t  len = (int32_t)utext_nativeLength(dest);
+
+        // The cloned string IS going to be NUL terminated, whether or not the original was.
+        const char16_t *srcStr = (const char16_t *)src->context;
+        char16_t *copyStr = (char16_t *)uprv_malloc((len+1) * sizeof(char16_t));
+        if (copyStr == nullptr) {
+            *status = U_MEMORY_ALLOCATION_ERROR;
+        } else {
+            int64_t i;
+            for (i=0; icontext = copyStr;
+            dest->providerProperties |= I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT);
+        }
+    }
+    return dest;
+}
+
+
+static void U_CALLCONV
+ucstrTextClose(UText *ut) {
+    // Most of the work of close is done by the generic UText framework close.
+    // All that needs to be done here is delete the string if the UText
+    //  owns it.  This occurs if the UText was created by cloning.
+    if (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT)) {
+        char16_t *s = (char16_t *)ut->context;
+        uprv_free(s);
+        ut->context = nullptr;
+    }
+}
+
+
+
+static int64_t U_CALLCONV
+ucstrTextLength(UText *ut) {
+    if (ut->a < 0) {
+        // null terminated, we don't yet know the length. Scan for it.
+        //    Access is not convenient for doing this
+        //    because the current iteration position can't be changed.
+        const char16_t  *str = (const char16_t *)ut->context;
+        for (;;) {
+            if (str[ut->chunkNativeLimit] == 0) {
+                break;
+            }
+            ut->chunkNativeLimit++;
+        }
+        ut->a = ut->chunkNativeLimit;
+        ut->chunkLength = (int32_t)ut->chunkNativeLimit;
+        ut->nativeIndexingLimit = ut->chunkLength;
+        ut->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE);
+    }
+    return ut->a;
+}
+
+
+static UBool U_CALLCONV
+ucstrTextAccess(UText *ut, int64_t index, UBool  forward) {
+    const char16_t *str   = (const char16_t *)ut->context;
+
+    // pin the requested index to the bounds of the string,
+    //  and set current iteration position.
+    if (index<0) {
+        index = 0;
+    } else if (index < ut->chunkNativeLimit) {
+        // The request data is within the chunk as it is known so far.
+        // Put index on a code point boundary.
+        U16_SET_CP_START(str, 0, index);
+    } else if (ut->a >= 0) {
+        // We know the length of this string, and the user is requesting something
+        // at or beyond the length.  Pin the requested index to the length.
+        index = ut->a;
+    } else {
+        // Null terminated string, length not yet known, and the requested index
+        //  is beyond where we have scanned so far.
+        //  Scan to 32 UChars beyond the requested index.  The strategy here is
+        //  to avoid fully scanning a long string when the caller only wants to
+        //  see a few characters at its beginning.
+        int32_t scanLimit = (int32_t)index + 32;
+        if ((index + 32)>INT32_MAX || (index + 32)<0 ) {   // note: int64 expression
+            scanLimit = INT32_MAX;
+        }
+
+        int32_t chunkLimit = (int32_t)ut->chunkNativeLimit;
+        for (; chunkLimita = chunkLimit;
+                ut->chunkLength = chunkLimit;
+                ut->nativeIndexingLimit = chunkLimit;
+                if (index >= chunkLimit) {
+                    index = chunkLimit;
+                } else {
+                    U16_SET_CP_START(str, 0, index);
+                }
+
+                ut->chunkNativeLimit = chunkLimit;
+                ut->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE);
+                goto breakout;
+            }
+        }
+        // We scanned through the next batch of UChars without finding the end.
+        U16_SET_CP_START(str, 0, index);
+        if (chunkLimit == INT32_MAX) {
+            // Scanned to the limit of a 32 bit length.
+            // Forceably trim the overlength string back so length fits in int32
+            //  TODO:  add support for 64 bit strings.
+            ut->a = chunkLimit;
+            ut->chunkLength = chunkLimit;
+            ut->nativeIndexingLimit = chunkLimit;
+            if (index > chunkLimit) {
+                index = chunkLimit;
+            }
+            ut->chunkNativeLimit = chunkLimit;
+            ut->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE);
+        } else {
+            // The endpoint of a chunk must not be left in the middle of a surrogate pair.
+            // If the current end is on a lead surrogate, back the end up by one.
+            // It doesn't matter if the end char happens to be an unpaired surrogate,
+            //    and it's simpler not to worry about it.
+            if (U16_IS_LEAD(str[chunkLimit-1])) {
+                --chunkLimit;
+            }
+            // Null-terminated chunk with end still unknown.
+            // Update the chunk length to reflect what has been scanned thus far.
+            // That the full length is still unknown is (still) flagged by
+            //    ut->a being < 0.
+            ut->chunkNativeLimit = chunkLimit;
+            ut->nativeIndexingLimit = chunkLimit;
+            ut->chunkLength = chunkLimit;
+        }
+
+    }
+breakout:
+    U_ASSERT(index<=INT32_MAX);
+    ut->chunkOffset = (int32_t)index;
+
+    // Check whether request is at the start or end
+    UBool retVal = (forward && indexchunkNativeLimit) || (!forward && index>0);
+    return retVal;
+}
+
+
+
+static int32_t U_CALLCONV
+ucstrTextExtract(UText *ut,
+                  int64_t start, int64_t limit,
+                  char16_t *dest, int32_t destCapacity,
+                  UErrorCode *pErrorCode)
+{
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+    if(destCapacity<0 || (dest==nullptr && destCapacity>0) || start>limit) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    //const char16_t *s=(const char16_t *)ut->context;
+    int32_t si, di;
+
+    int32_t start32;
+    int32_t limit32;
+
+    // Access the start.  Does two things we need:
+    //   Pins 'start' to the length of the string, if it came in out-of-bounds.
+    //   Snaps 'start' to the beginning of a code point.
+    ucstrTextAccess(ut, start, true);
+    const char16_t *s=ut->chunkContents;
+    start32 = ut->chunkOffset;
+
+    int32_t strLength=(int32_t)ut->a;
+    if (strLength >= 0) {
+        limit32 = pinIndex(limit, strLength);
+    } else {
+        limit32 = pinIndex(limit, INT32_MAX);
+    }
+    di = 0;
+    for (si=start32; sia = si;               // set string length for this UText
+            ut->chunkNativeLimit    = si;
+            ut->chunkLength         = si;
+            ut->nativeIndexingLimit = si;
+            strLength               = si;
+            limit32                 = si;
+            break;
+        }
+        U_ASSERT(di>=0); /* to ensure di never exceeds INT32_MAX, which must not happen logically */
+        if (di=0) {
+                // We have filled the destination buffer, and the string length is known.
+                //  Cut the loop short.  There is no need to scan string termination.
+                di = limit32 - start32;
+                si = limit32;
+                break;
+            }
+        }
+        di++;
+    }
+
+    // If the limit index points to a lead surrogate of a pair,
+    //   add the corresponding trail surrogate to the destination.
+    if (si>0 && U16_IS_LEAD(s[si-1]) &&
+            ((sichunkNativeLimit) {
+        ut->chunkOffset = si;
+    } else {
+        ucstrTextAccess(ut, si, true);
+    }
+
+    // Add a terminating NUL if space in the buffer permits,
+    // and set the error status as required.
+    u_terminateUChars(dest, destCapacity, di, pErrorCode);
+    return di;
+}
+
+static const struct UTextFuncs ucstrFuncs =
+{
+    sizeof(UTextFuncs),
+    0, 0, 0,           // Reserved alignment padding
+    ucstrTextClone,
+    ucstrTextLength,
+    ucstrTextAccess,
+    ucstrTextExtract,
+    nullptr,              // Replace
+    nullptr,              // Copy
+    nullptr,              // MapOffsetToNative,
+    nullptr,              // MapIndexToUTF16,
+    ucstrTextClose,
+    nullptr,              // spare 1
+    nullptr,              // spare 2
+    nullptr,              // spare 3
+};
+
+U_CDECL_END
+
+static const char16_t gEmptyUString[] = {0};
+
+U_CAPI UText * U_EXPORT2
+utext_openUChars(UText *ut, const char16_t *s, int64_t length, UErrorCode *status) {
+    if (U_FAILURE(*status)) {
+        return nullptr;
+    }
+    if(s==nullptr && length==0) {
+        s = gEmptyUString;
+    }
+    if (s==nullptr || length < -1 || length>INT32_MAX) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+    ut = utext_setup(ut, 0, status);
+    if (U_SUCCESS(*status)) {
+        ut->pFuncs               = &ucstrFuncs;
+        ut->context              = s;
+        ut->providerProperties   = I32_FLAG(UTEXT_PROVIDER_STABLE_CHUNKS);
+        if (length==-1) {
+            ut->providerProperties |= I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE);
+        }
+        ut->a                    = length;
+        ut->chunkContents        = s;
+        ut->chunkNativeStart     = 0;
+        ut->chunkNativeLimit     = length>=0? length : 0;
+        ut->chunkLength          = (int32_t)ut->chunkNativeLimit;
+        ut->chunkOffset          = 0;
+        ut->nativeIndexingLimit  = ut->chunkLength;
+    }
+    return ut;
+}
+
+
+//------------------------------------------------------------------------------
+//
+//     UText implementation for text from ICU CharacterIterators
+//
+//         Use of UText data members:
+//            context    pointer to the CharacterIterator
+//            a          length of the full text.
+//            p          pointer to  buffer 1
+//            b          start index of local buffer 1 contents
+//            q          pointer to buffer 2
+//            c          start index of local buffer 2 contents
+//            r          pointer to the character iterator if the UText owns it.
+//                       Null otherwise.
+//
+//------------------------------------------------------------------------------
+#define CIBufSize 16
+
+U_CDECL_BEGIN
+static void U_CALLCONV
+charIterTextClose(UText *ut) {
+    // Most of the work of close is done by the generic UText framework close.
+    // All that needs to be done here is delete the CharacterIterator if the UText
+    //  owns it.  This occurs if the UText was created by cloning.
+    CharacterIterator *ci = (CharacterIterator *)ut->r;
+    delete ci;
+    ut->r = nullptr;
+}
+
+static int64_t U_CALLCONV
+charIterTextLength(UText *ut) {
+    return (int32_t)ut->a;
+}
+
+static UBool U_CALLCONV
+charIterTextAccess(UText *ut, int64_t index, UBool  forward) {
+    CharacterIterator *ci   = (CharacterIterator *)ut->context;
+
+    int32_t clippedIndex = (int32_t)index;
+    if (clippedIndex<0) {
+        clippedIndex=0;
+    } else if (clippedIndex>=ut->a) {
+        clippedIndex=(int32_t)ut->a;
+    }
+    int32_t neededIndex = clippedIndex;
+    if (!forward && neededIndex>0) {
+        // reverse iteration, want the position just before what was asked for.
+        neededIndex--;
+    } else if (forward && neededIndex==ut->a && neededIndex>0) {
+        // Forward iteration, don't ask for something past the end of the text.
+        neededIndex--;
+    }
+
+    // Find the native index of the start of the buffer containing what we want.
+    neededIndex -= neededIndex % CIBufSize;
+
+    char16_t *buf = nullptr;
+    UBool  needChunkSetup = true;
+    int    i;
+    if (ut->chunkNativeStart == neededIndex) {
+        // The buffer we want is already the current chunk.
+        needChunkSetup = false;
+    } else if (ut->b == neededIndex) {
+        // The first buffer (buffer p) has what we need.
+        buf = (char16_t *)ut->p;
+    } else if (ut->c == neededIndex) {
+        // The second buffer (buffer q) has what we need.
+        buf = (char16_t *)ut->q;
+    } else {
+        // Neither buffer already has what we need.
+        // Load new data from the character iterator.
+        // Use the buf that is not the current buffer.
+        buf = (char16_t *)ut->p;
+        if (ut->p == ut->chunkContents) {
+            buf = (char16_t *)ut->q;
+        }
+        ci->setIndex(neededIndex);
+        for (i=0; inextPostInc();
+            if (i+neededIndex > ut->a) {
+                break;
+            }
+        }
+    }
+
+    // We have a buffer with the data we need.
+    // Set it up as the current chunk, if it wasn't already.
+    if (needChunkSetup) {
+        ut->chunkContents = buf;
+        ut->chunkLength   = CIBufSize;
+        ut->chunkNativeStart = neededIndex;
+        ut->chunkNativeLimit = neededIndex + CIBufSize;
+        if (ut->chunkNativeLimit > ut->a) {
+            ut->chunkNativeLimit = ut->a;
+            ut->chunkLength  = (int32_t)(ut->chunkNativeLimit)-(int32_t)(ut->chunkNativeStart);
+        }
+        ut->nativeIndexingLimit = ut->chunkLength;
+        U_ASSERT(ut->chunkOffset>=0 && ut->chunkOffset<=CIBufSize);
+    }
+    ut->chunkOffset = clippedIndex - (int32_t)ut->chunkNativeStart;
+    UBool success = (forward? ut->chunkOffsetchunkLength : ut->chunkOffset>0);
+    return success;
+}
+
+static UText * U_CALLCONV
+charIterTextClone(UText *dest, const UText *src, UBool deep, UErrorCode * status) {
+    if (U_FAILURE(*status)) {
+        return nullptr;
+    }
+
+    if (deep) {
+        // There is no CharacterIterator API for cloning the underlying text storage.
+        *status = U_UNSUPPORTED_ERROR;
+        return nullptr;
+    } else {
+        CharacterIterator *srcCI =(CharacterIterator *)src->context;
+        srcCI = srcCI->clone();
+        dest = utext_openCharacterIterator(dest, srcCI, status);
+        if (U_FAILURE(*status)) {
+            return dest;
+        }
+        // cast off const on getNativeIndex.
+        //   For CharacterIterator based UTexts, this is safe, the operation is const.
+        int64_t  ix = utext_getNativeIndex((UText *)src);
+        utext_setNativeIndex(dest, ix);
+        dest->r = srcCI;    // flags that this UText owns the CharacterIterator
+    }
+    return dest;
+}
+
+static int32_t U_CALLCONV
+charIterTextExtract(UText *ut,
+                  int64_t start, int64_t limit,
+                  char16_t *dest, int32_t destCapacity,
+                  UErrorCode *status)
+{
+    if(U_FAILURE(*status)) {
+        return 0;
+    }
+    if(destCapacity<0 || (dest==nullptr && destCapacity>0) || start>limit) {
+        *status=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    int32_t  length  = (int32_t)ut->a;
+    int32_t  start32 = pinIndex(start, length);
+    int32_t  limit32 = pinIndex(limit, length);
+    int32_t  desti   = 0;
+    int32_t  srci;
+    int32_t  copyLimit;
+
+    CharacterIterator *ci = (CharacterIterator *)ut->context;
+    ci->setIndex32(start32);   // Moves ix to lead of surrogate pair, if needed.
+    srci = ci->getIndex();
+    copyLimit = srci;
+    while (srcinext32PostInc();
+        int32_t  len = U16_LENGTH(c);
+        U_ASSERT(desti+len>0); /* to ensure desti+len never exceeds MAX_INT32, which must not happen logically */
+        if (desti+len <= destCapacity) {
+            U16_APPEND_UNSAFE(dest, desti, c);
+            copyLimit = srci+len;
+        } else {
+            desti += len;
+            *status = U_BUFFER_OVERFLOW_ERROR;
+        }
+        srci += len;
+    }
+
+    charIterTextAccess(ut, copyLimit, true);
+
+    u_terminateUChars(dest, destCapacity, desti, status);
+    return desti;
+}
+
+static const struct UTextFuncs charIterFuncs =
+{
+    sizeof(UTextFuncs),
+    0, 0, 0,             // Reserved alignment padding
+    charIterTextClone,
+    charIterTextLength,
+    charIterTextAccess,
+    charIterTextExtract,
+    nullptr,                // Replace
+    nullptr,                // Copy
+    nullptr,                // MapOffsetToNative,
+    nullptr,                // MapIndexToUTF16,
+    charIterTextClose,
+    nullptr,                // spare 1
+    nullptr,                // spare 2
+    nullptr                 // spare 3
+};
+U_CDECL_END
+
+
+U_CAPI UText * U_EXPORT2
+utext_openCharacterIterator(UText *ut, CharacterIterator *ci, UErrorCode *status) {
+    if (U_FAILURE(*status)) {
+        return nullptr;
+    }
+
+    if (ci->startIndex() > 0) {
+        // No support for CharacterIterators that do not start indexing from zero.
+        *status = U_UNSUPPORTED_ERROR;
+        return nullptr;
+    }
+
+    // Extra space in UText for 2 buffers of CIBufSize UChars each.
+    int32_t  extraSpace = 2 * CIBufSize * sizeof(char16_t);
+    ut = utext_setup(ut, extraSpace, status);
+    if (U_SUCCESS(*status)) {
+        ut->pFuncs                = &charIterFuncs;
+        ut->context              = ci;
+        ut->providerProperties   = 0;
+        ut->a                    = ci->endIndex();        // Length of text
+        ut->p                    = ut->pExtra;            // First buffer
+        ut->b                    = -1;                    // Native index of first buffer contents
+        ut->q                    = (char16_t*)ut->pExtra+CIBufSize;  // Second buffer
+        ut->c                    = -1;                    // Native index of second buffer contents
+
+        // Initialize current chunk contents to be empty.
+        //   First access will fault something in.
+        //   Note:  The initial nativeStart and chunkOffset must sum to zero
+        //          so that getNativeIndex() will correctly compute to zero
+        //          if no call to Access() has ever been made.  They can't be both
+        //          zero without Access() thinking that the chunk is valid.
+        ut->chunkContents        = (char16_t *)ut->p;
+        ut->chunkNativeStart     = -1;
+        ut->chunkOffset          = 1;
+        ut->chunkNativeLimit     = 0;
+        ut->chunkLength          = 0;
+        ut->nativeIndexingLimit  = ut->chunkOffset;  // enables native indexing
+    }
+    return ut;
+}
diff --git a/intl/icu/source/common/utf_impl.cpp b/intl/icu/source/common/utf_impl.cpp
new file mode 100644
index 0000000000..827a82daf4
--- /dev/null
+++ b/intl/icu/source/common/utf_impl.cpp
@@ -0,0 +1,329 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+*
+*   Copyright (C) 1999-2012, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+******************************************************************************
+*   file name:  utf_impl.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 1999sep13
+*   created by: Markus W. Scherer
+*
+*   This file provides implementation functions for macros in the utfXX.h
+*   that would otherwise be too long as macros.
+*/
+
+/* set import/export definitions */
+#ifndef U_UTF8_IMPL
+#   define U_UTF8_IMPL
+#endif
+
+#include "unicode/utypes.h"
+#include "unicode/utf.h"
+#include "unicode/utf8.h"
+#include "uassert.h"
+
+/*
+ * Table of the number of utf8 trail bytes, indexed by the lead byte.
+ * Used by the deprecated macro UTF8_COUNT_TRAIL_BYTES, defined in utf_old.h
+ *
+ * The current macro, U8_COUNT_TRAIL_BYTES, does _not_ use this table.
+ *
+ * Note that this table cannot be removed, even if UTF8_COUNT_TRAIL_BYTES were
+ * changed to no longer use it. References to the table from expansions of UTF8_COUNT_TRAIL_BYTES
+ * may exist in old client code that must continue to run with newer icu library versions.
+ *
+ * This table could be replaced on many machines by
+ * a few lines of assembler code using an
+ * "index of first 0-bit from msb" instruction and
+ * one or two more integer instructions.
+ *
+ * For example, on an i386, do something like
+ * - MOV AL, leadByte
+ * - NOT AL         (8-bit, leave b15..b8==0..0, reverse only b7..b0)
+ * - MOV AH, 0
+ * - BSR BX, AX     (16-bit)
+ * - MOV AX, 6      (result)
+ * - JZ finish      (ZF==1 if leadByte==0xff)
+ * - SUB AX, BX (result)
+ * -finish:
+ * (BSR: Bit Scan Reverse, scans for a 1-bit, starting from the MSB)
+ */
+U_CAPI const uint8_t
+utf8_countTrailBytes[256]={
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+
+    // illegal C0 & C1
+    // 2-byte lead bytes C2..DF
+    0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+
+    // 3-byte lead bytes E0..EF
+    2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
+    // 4-byte lead bytes F0..F4
+    // illegal F5..FF
+    3, 3, 3, 3, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+};
+
+static const UChar32
+utf8_errorValue[6]={
+    // Same values as UTF8_ERROR_VALUE_1, UTF8_ERROR_VALUE_2, UTF_ERROR_VALUE,
+    // but without relying on the obsolete unicode/utf_old.h.
+    0x15, 0x9f, 0xffff,
+    0x10ffff
+};
+
+static UChar32
+errorValue(int32_t count, int8_t strict) {
+    if(strict>=0) {
+        return utf8_errorValue[count];
+    } else if(strict==-3) {
+        return 0xfffd;
+    } else {
+        return U_SENTINEL;
+    }
+}
+
+/*
+ * Handle the non-inline part of the U8_NEXT() and U8_NEXT_FFFD() macros
+ * and their obsolete sibling UTF8_NEXT_CHAR_SAFE().
+ *
+ * U8_NEXT() supports NUL-terminated strings indicated via length<0.
+ *
+ * The "strict" parameter controls the error behavior:
+ * <0  "Safe" behavior of U8_NEXT():
+ *     -1: All illegal byte sequences yield U_SENTINEL=-1.
+ *     -2: Same as -1, except for lenient treatment of surrogate code points as legal.
+ *         Some implementations use this for roundtripping of
+ *         Unicode 16-bit strings that are not well-formed UTF-16, that is, they
+ *         contain unpaired surrogates.
+ *     -3: All illegal byte sequences yield U+FFFD.
+ *  0  Obsolete "safe" behavior of UTF8_NEXT_CHAR_SAFE(..., false):
+ *     All illegal byte sequences yield a positive code point such that this
+ *     result code point would be encoded with the same number of bytes as
+ *     the illegal sequence.
+ * >0  Obsolete "strict" behavior of UTF8_NEXT_CHAR_SAFE(..., true):
+ *     Same as the obsolete "safe" behavior, but non-characters are also treated
+ *     like illegal sequences.
+ *
+ * Note that a UBool is the same as an int8_t.
+ */
+U_CAPI UChar32 U_EXPORT2
+utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict) {
+    // *pi is one after byte c.
+    int32_t i=*pi;
+    // length can be negative for NUL-terminated strings: Read and validate one byte at a time.
+    if(i==length || c>0xf4) {
+        // end of string, or not a lead byte
+    } else if(c>=0xf0) {
+        // Test for 4-byte sequences first because
+        // U8_NEXT() handles shorter valid sequences inline.
+        uint8_t t1=s[i], t2, t3;
+        c&=7;
+        if(U8_IS_VALID_LEAD4_AND_T1(c, t1) &&
+                ++i!=length && (t2=s[i]-0x80)<=0x3f &&
+                ++i!=length && (t3=s[i]-0x80)<=0x3f) {
+            ++i;
+            c=(c<<18)|((t1&0x3f)<<12)|(t2<<6)|t3;
+            // strict: forbid non-characters like U+fffe
+            if(strict<=0 || !U_IS_UNICODE_NONCHAR(c)) {
+                *pi=i;
+                return c;
+            }
+        }
+    } else if(c>=0xe0) {
+        c&=0xf;
+        if(strict!=-2) {
+            uint8_t t1=s[i], t2;
+            if(U8_IS_VALID_LEAD3_AND_T1(c, t1) &&
+                    ++i!=length && (t2=s[i]-0x80)<=0x3f) {
+                ++i;
+                c=(c<<12)|((t1&0x3f)<<6)|t2;
+                // strict: forbid non-characters like U+fffe
+                if(strict<=0 || !U_IS_UNICODE_NONCHAR(c)) {
+                    *pi=i;
+                    return c;
+                }
+            }
+        } else {
+            // strict=-2 -> lenient: allow surrogates
+            uint8_t t1=s[i]-0x80, t2;
+            if(t1<=0x3f && (c>0 || t1>=0x20) &&
+                    ++i!=length && (t2=s[i]-0x80)<=0x3f) {
+                *pi=i+1;
+                return (c<<12)|(t1<<6)|t2;
+            }
+        }
+    } else if(c>=0xc2) {
+        uint8_t t1=s[i]-0x80;
+        if(t1<=0x3f) {
+            *pi=i+1;
+            return ((c-0xc0)<<6)|t1;
+        }
+    }  // else 0x80<=c<0xc2 is not a lead byte
+
+    /* error handling */
+    c=errorValue(i-*pi, strict);
+    *pi=i;
+    return c;
+}
+
+U_CAPI int32_t U_EXPORT2
+utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError) {
+    if((uint32_t)(c)<=0x7ff) {
+        if((i)+1<(length)) {
+            (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0);
+            (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80);
+            return i;
+        }
+    } else if((uint32_t)(c)<=0xffff) {
+        /* Starting with Unicode 3.2, surrogate code points must not be encoded in UTF-8. */
+        if((i)+2<(length) && !U_IS_SURROGATE(c)) {
+            (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0);
+            (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80);
+            (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80);
+            return i;
+        }
+    } else if((uint32_t)(c)<=0x10ffff) {
+        if((i)+3<(length)) {
+            (s)[(i)++]=(uint8_t)(((c)>>18)|0xf0);
+            (s)[(i)++]=(uint8_t)((((c)>>12)&0x3f)|0x80);
+            (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80);
+            (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80);
+            return i;
+        }
+    }
+    /* c>0x10ffff or not enough space, write an error value */
+    if(pIsError!=nullptr) {
+        *pIsError=true;
+    } else {
+        length-=i;
+        if(length>0) {
+            int32_t offset;
+            if(length>3) {
+                length=3;
+            }
+            s+=i;
+            offset=0;
+            c=utf8_errorValue[length-1];
+            U8_APPEND_UNSAFE(s, offset, c);
+            i=i+offset;
+        }
+    }
+    return i;
+}
+
+U_CAPI UChar32 U_EXPORT2
+utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict) {
+    // *pi is the index of byte c.
+    int32_t i=*pi;
+    if(U8_IS_TRAIL(c) && i>start) {
+        uint8_t b1=s[--i];
+        if(U8_IS_LEAD(b1)) {
+            if(b1<0xe0) {
+                *pi=i;
+                return ((b1-0xc0)<<6)|(c&0x3f);
+            } else if(b1<0xf0 ? U8_IS_VALID_LEAD3_AND_T1(b1, c) : U8_IS_VALID_LEAD4_AND_T1(b1, c)) {
+                // Truncated 3- or 4-byte sequence.
+                *pi=i;
+                return errorValue(1, strict);
+            }
+        } else if(U8_IS_TRAIL(b1) && i>start) {
+            // Extract the value bits from the last trail byte.
+            c&=0x3f;
+            uint8_t b2=s[--i];
+            if(0xe0<=b2 && b2<=0xf4) {
+                if(b2<0xf0) {
+                    b2&=0xf;
+                    if(strict!=-2) {
+                        if(U8_IS_VALID_LEAD3_AND_T1(b2, b1)) {
+                            *pi=i;
+                            c=(b2<<12)|((b1&0x3f)<<6)|c;
+                            if(strict<=0 || !U_IS_UNICODE_NONCHAR(c)) {
+                                return c;
+                            } else {
+                                // strict: forbid non-characters like U+fffe
+                                return errorValue(2, strict);
+                            }
+                        }
+                    } else {
+                        // strict=-2 -> lenient: allow surrogates
+                        b1-=0x80;
+                        if((b2>0 || b1>=0x20)) {
+                            *pi=i;
+                            return (b2<<12)|(b1<<6)|c;
+                        }
+                    }
+                } else if(U8_IS_VALID_LEAD4_AND_T1(b2, b1)) {
+                    // Truncated 4-byte sequence.
+                    *pi=i;
+                    return errorValue(2, strict);
+                }
+            } else if(U8_IS_TRAIL(b2) && i>start) {
+                uint8_t b3=s[--i];
+                if(0xf0<=b3 && b3<=0xf4) {
+                    b3&=7;
+                    if(U8_IS_VALID_LEAD4_AND_T1(b3, b2)) {
+                        *pi=i;
+                        c=(b3<<18)|((b2&0x3f)<<12)|((b1&0x3f)<<6)|c;
+                        if(strict<=0 || !U_IS_UNICODE_NONCHAR(c)) {
+                            return c;
+                        } else {
+                            // strict: forbid non-characters like U+fffe
+                            return errorValue(3, strict);
+                        }
+                    }
+                }
+            }
+        }
+    }
+    return errorValue(0, strict);
+}
+
+U_CAPI int32_t U_EXPORT2
+utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i) {
+    // Same as utf8_prevCharSafeBody(..., strict=-1) minus assembling code points.
+    int32_t orig_i=i;
+    uint8_t c=s[i];
+    if(U8_IS_TRAIL(c) && i>start) {
+        uint8_t b1=s[--i];
+        if(U8_IS_LEAD(b1)) {
+            if(b1<0xe0 ||
+                    (b1<0xf0 ? U8_IS_VALID_LEAD3_AND_T1(b1, c) : U8_IS_VALID_LEAD4_AND_T1(b1, c))) {
+                return i;
+            }
+        } else if(U8_IS_TRAIL(b1) && i>start) {
+            uint8_t b2=s[--i];
+            if(0xe0<=b2 && b2<=0xf4) {
+                if(b2<0xf0 ? U8_IS_VALID_LEAD3_AND_T1(b2, b1) : U8_IS_VALID_LEAD4_AND_T1(b2, b1)) {
+                    return i;
+                }
+            } else if(U8_IS_TRAIL(b2) && i>start) {
+                uint8_t b3=s[--i];
+                if(0xf0<=b3 && b3<=0xf4 && U8_IS_VALID_LEAD4_AND_T1(b3, b2)) {
+                    return i;
+                }
+            }
+        }
+    }
+    return orig_i;
+}
diff --git a/intl/icu/source/common/util.cpp b/intl/icu/source/common/util.cpp
new file mode 100644
index 0000000000..f6a13aa445
--- /dev/null
+++ b/intl/icu/source/common/util.cpp
@@ -0,0 +1,435 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+*   Copyright (c) 2001-2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*   Date        Name        Description
+*   11/19/2001  aliu        Creation.
+**********************************************************************
+*/
+
+#include "unicode/unimatch.h"
+#include "unicode/utf16.h"
+#include "patternprops.h"
+#include "util.h"
+
+// Define char16_t constants using hex for EBCDIC compatibility
+
+static const char16_t BACKSLASH  = 0x005C; /*\*/
+static const char16_t UPPER_U    = 0x0055; /*U*/
+static const char16_t LOWER_U    = 0x0075; /*u*/
+static const char16_t APOSTROPHE = 0x0027; // '\''
+static const char16_t SPACE      = 0x0020; // ' '
+
+// "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+static const char16_t DIGITS[] = {
+    48,49,50,51,52,53,54,55,56,57,
+    65,66,67,68,69,70,71,72,73,74,
+    75,76,77,78,79,80,81,82,83,84,
+    85,86,87,88,89,90
+};
+
+U_NAMESPACE_BEGIN
+
+UnicodeString& ICU_Utility::appendNumber(UnicodeString& result, int32_t n,
+                                     int32_t radix, int32_t minDigits) {
+    if (radix < 2 || radix > 36) {
+        // Bogus radix
+        return result.append((char16_t)63/*?*/);
+    }
+    // Handle negatives
+    if (n < 0) {
+        n = -n;
+        result.append((char16_t)45/*-*/);
+    }
+    // First determine the number of digits
+    int32_t nn = n;
+    int32_t r = 1;
+    while (nn >= radix) {
+        nn /= radix;
+        r *= radix;
+        --minDigits;
+    }
+    // Now generate the digits
+    while (--minDigits > 0) {
+        result.append(DIGITS[0]);
+    }
+    while (r > 0) {
+        int32_t digit = n / r;
+        result.append(DIGITS[digit]);
+        n -= digit * r;
+        r /= radix;
+    }
+    return result;
+}
+
+UBool ICU_Utility::isUnprintable(UChar32 c) {
+    return !(c >= 0x20 && c <= 0x7E);
+}
+
+UBool ICU_Utility::shouldAlwaysBeEscaped(UChar32 c) {
+    if (c < 0x20) {
+        return true;  // C0 control codes
+    } else if (c <= 0x7e) {
+        return false;  // printable ASCII
+    } else if (c <= 0x9f) {
+        return true;  // C1 control codes
+    } else if (c < 0xd800) {
+        return false;  // most of the BMP
+    } else if (c <= 0xdfff || (0xfdd0 <= c && c <= 0xfdef) || (c & 0xfffe) == 0xfffe) {
+        return true;  // surrogate or noncharacter code points
+    } else if (c <= 0x10ffff) {
+        return false;  // all else
+    } else {
+        return true;  // not a code point
+    }
+}
+
+UBool ICU_Utility::escapeUnprintable(UnicodeString& result, UChar32 c) {
+    if (isUnprintable(c)) {
+        escape(result, c);
+        return true;
+    }
+    return false;
+}
+
+UnicodeString &ICU_Utility::escape(UnicodeString& result, UChar32 c) {
+    result.append(BACKSLASH);
+    if (c & ~0xFFFF) {
+        result.append(UPPER_U);
+        result.append(DIGITS[0xF&(c>>28)]);
+        result.append(DIGITS[0xF&(c>>24)]);
+        result.append(DIGITS[0xF&(c>>20)]);
+        result.append(DIGITS[0xF&(c>>16)]);
+    } else {
+        result.append(LOWER_U);
+    }
+    result.append(DIGITS[0xF&(c>>12)]);
+    result.append(DIGITS[0xF&(c>>8)]);
+    result.append(DIGITS[0xF&(c>>4)]);
+    result.append(DIGITS[0xF&c]);
+    return result;
+}
+
+/**
+ * Returns the index of a character, ignoring quoted text.
+ * For example, in the string "abc'hide'h", the 'h' in "hide" will not be
+ * found by a search for 'h'.
+ */
+// FOR FUTURE USE.  DISABLE FOR NOW for coverage reasons.
+/*
+int32_t ICU_Utility::quotedIndexOf(const UnicodeString& text,
+                               int32_t start, int32_t limit,
+                               char16_t charToFind) {
+    for (int32_t i=start; i= pos);
+//?
+//?    if (!isForward) {
+//?        --pos; // pos is a limit, so back up by one
+//?    }
+//?    
+//?    while (pos != stop &&
+//?           PatternProps::isWhiteSpace(c = text.char32At(pos))) {
+//?        if (isForward) {
+//?            pos += U16_LENGTH(c);
+//?        } else {
+//?            pos -= U16_LENGTH(c);
+//?        }
+//?    }
+//?
+//?    if (!isForward) {
+//?        ++pos; // make pos back into a limit
+//?    }
+//?
+//?    return pos;
+//?}
+
+/**
+ * Parse a single non-whitespace character 'ch', optionally
+ * preceded by whitespace.
+ * @param id the string to be parsed
+ * @param pos INPUT-OUTPUT parameter.  On input, pos[0] is the
+ * offset of the first character to be parsed.  On output, pos[0]
+ * is the index after the last parsed character.  If the parse
+ * fails, pos[0] will be unchanged.
+ * @param ch the non-whitespace character to be parsed.
+ * @return true if 'ch' is seen preceded by zero or more
+ * whitespace characters.
+ */
+UBool ICU_Utility::parseChar(const UnicodeString& id, int32_t& pos, char16_t ch) {
+    int32_t start = pos;
+    skipWhitespace(id, pos, true);
+    if (pos == id.length() ||
+        id.charAt(pos) != ch) {
+        pos = start;
+        return false;
+    }
+    ++pos;
+    return true;
+}
+
+/**
+ * Parse a pattern string within the given Replaceable and a parsing
+ * pattern.  Characters are matched literally and case-sensitively
+ * except for the following special characters:
+ *
+ * ~  zero or more Pattern_White_Space chars
+ *
+ * If end of pattern is reached with all matches along the way,
+ * pos is advanced to the first unparsed index and returned.
+ * Otherwise -1 is returned.
+ * @param pat pattern that controls parsing
+ * @param text text to be parsed, starting at index
+ * @param index offset to first character to parse
+ * @param limit offset after last character to parse
+ * @return index after last parsed character, or -1 on parse failure.
+ */
+int32_t ICU_Utility::parsePattern(const UnicodeString& pat,
+                                  const Replaceable& text,
+                                  int32_t index,
+                                  int32_t limit) {
+    int32_t ipat = 0;
+
+    // empty pattern matches immediately
+    if (ipat == pat.length()) {
+        return index;
+    }
+
+    UChar32 cpat = pat.char32At(ipat);
+
+    while (index < limit) {
+        UChar32 c = text.char32At(index);
+
+        // parse \s*
+        if (cpat == 126 /*~*/) {
+            if (PatternProps::isWhiteSpace(c)) {
+                index += U16_LENGTH(c);
+                continue;
+            } else {
+                if (++ipat == pat.length()) {
+                    return index; // success; c unparsed
+                }
+                // fall thru; process c again with next cpat
+            }
+        }
+
+        // parse literal
+        else if (c == cpat) {
+            index += U16_LENGTH(c);
+            ipat += U16_LENGTH(cpat);
+            if (ipat == pat.length()) {
+                return index; // success; c parsed
+            }
+            // fall thru; get next cpat
+        }
+
+        // match failure of literal
+        else {
+            return -1;
+        }
+
+        cpat = pat.char32At(ipat);
+    }
+
+    return -1; // text ended before end of pat
+}
+
+int32_t ICU_Utility::parseAsciiInteger(const UnicodeString& str, int32_t& pos) {
+    int32_t result = 0;
+    char16_t c;
+    while (pos < str.length() && (c = str.charAt(pos)) >= u'0' && c <= u'9') {
+        result = result * 10 + (c - u'0');
+        pos++;
+    }
+    return result;
+}
+
+/**
+ * Append a character to a rule that is being built up.  To flush
+ * the quoteBuf to rule, make one final call with isLiteral == true.
+ * If there is no final character, pass in (UChar32)-1 as c.
+ * @param rule the string to append the character to
+ * @param c the character to append, or (UChar32)-1 if none.
+ * @param isLiteral if true, then the given character should not be
+ * quoted or escaped.  Usually this means it is a syntactic element
+ * such as > or $
+ * @param escapeUnprintable if true, then unprintable characters
+ * should be escaped using \uxxxx or \Uxxxxxxxx.  These escapes will
+ * appear outside of quotes.
+ * @param quoteBuf a buffer which is used to build up quoted
+ * substrings.  The caller should initially supply an empty buffer,
+ * and thereafter should not modify the buffer.  The buffer should be
+ * cleared out by, at the end, calling this method with a literal
+ * character.
+ */
+void ICU_Utility::appendToRule(UnicodeString& rule,
+                               UChar32 c,
+                               UBool isLiteral,
+                               UBool escapeUnprintable,
+                               UnicodeString& quoteBuf) {
+    // If we are escaping unprintables, then escape them outside
+    // quotes.  \u and \U are not recognized within quotes.  The same
+    // logic applies to literals, but literals are never escaped.
+    if (isLiteral ||
+        (escapeUnprintable && ICU_Utility::isUnprintable(c))) {
+        if (quoteBuf.length() > 0) {
+            // We prefer backslash APOSTROPHE to double APOSTROPHE
+            // (more readable, less similar to ") so if there are
+            // double APOSTROPHEs at the ends, we pull them outside
+            // of the quote.
+
+            // If the first thing in the quoteBuf is APOSTROPHE
+            // (doubled) then pull it out.
+            while (quoteBuf.length() >= 2 &&
+                   quoteBuf.charAt(0) == APOSTROPHE &&
+                   quoteBuf.charAt(1) == APOSTROPHE) {
+                rule.append(BACKSLASH).append(APOSTROPHE);
+                quoteBuf.remove(0, 2);
+            }
+            // If the last thing in the quoteBuf is APOSTROPHE
+            // (doubled) then remove and count it and add it after.
+            int32_t trailingCount = 0;
+            while (quoteBuf.length() >= 2 &&
+                   quoteBuf.charAt(quoteBuf.length()-2) == APOSTROPHE &&
+                   quoteBuf.charAt(quoteBuf.length()-1) == APOSTROPHE) {
+                quoteBuf.truncate(quoteBuf.length()-2);
+                ++trailingCount;
+            }
+            if (quoteBuf.length() > 0) {
+                rule.append(APOSTROPHE);
+                rule.append(quoteBuf);
+                rule.append(APOSTROPHE);
+                quoteBuf.truncate(0);
+            }
+            while (trailingCount-- > 0) {
+                rule.append(BACKSLASH).append(APOSTROPHE);
+            }
+        }
+        if (c != (UChar32)-1) {
+            /* Since spaces are ignored during parsing, they are
+             * emitted only for readability.  We emit one here
+             * only if there isn't already one at the end of the
+             * rule.
+             */
+            if (c == SPACE) {
+                int32_t len = rule.length();
+                if (len > 0 && rule.charAt(len-1) != c) {
+                    rule.append(c);
+                }
+            } else if (!escapeUnprintable || !ICU_Utility::escapeUnprintable(rule, c)) {
+                rule.append(c);
+            }
+        }
+    }
+
+    // Escape ' and '\' and don't begin a quote just for them
+    else if (quoteBuf.length() == 0 &&
+             (c == APOSTROPHE || c == BACKSLASH)) {
+        rule.append(BACKSLASH);
+        rule.append(c);
+    }
+
+    // Specials (printable ascii that isn't [0-9a-zA-Z]) and
+    // whitespace need quoting.  Also append stuff to quotes if we are
+    // building up a quoted substring already.
+    else if (quoteBuf.length() > 0 ||
+             (c >= 0x0021 && c <= 0x007E &&
+              !((c >= 0x0030/*'0'*/ && c <= 0x0039/*'9'*/) ||
+                (c >= 0x0041/*'A'*/ && c <= 0x005A/*'Z'*/) ||
+                (c >= 0x0061/*'a'*/ && c <= 0x007A/*'z'*/))) ||
+             PatternProps::isWhiteSpace(c)) {
+        quoteBuf.append(c);
+        // Double ' within a quote
+        if (c == APOSTROPHE) {
+            quoteBuf.append(c);
+        }
+    }
+    
+    // Otherwise just append
+    else {
+        rule.append(c);
+    }
+}
+
+void ICU_Utility::appendToRule(UnicodeString& rule,
+                               const UnicodeString& text,
+                               UBool isLiteral,
+                               UBool escapeUnprintable,
+                               UnicodeString& quoteBuf) {
+    for (int32_t i=0; itoPattern(pat, escapeUnprintable),
+                     true, escapeUnprintable, quoteBuf);
+    }
+}
+
+U_NAMESPACE_END
diff --git a/intl/icu/source/common/util.h b/intl/icu/source/common/util.h
new file mode 100644
index 0000000000..cf71b9623e
--- /dev/null
+++ b/intl/icu/source/common/util.h
@@ -0,0 +1,267 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+ **********************************************************************
+ *   Copyright (c) 2001-2011, International Business Machines
+ *   Corporation and others.  All Rights Reserved.
+ **********************************************************************
+ *   Date        Name        Description
+ *   11/19/2001  aliu        Creation.
+ **********************************************************************
+ */
+
+#ifndef ICU_UTIL_H
+#define ICU_UTIL_H
+
+#include "charstr.h"
+#include "unicode/unistr.h"
+#include "unicode/uobject.h"
+#include "unicode/utypes.h"
+//--------------------------------------------------------------------
+// class ICU_Utility
+// i18n utility functions, scoped into the class ICU_Utility.
+//--------------------------------------------------------------------
+
+U_NAMESPACE_BEGIN
+
+class UnicodeMatcher;
+
+class U_COMMON_API ICU_Utility /* not : public UObject because all methods are static */ {
+ public:
+
+    /**
+     * Append a number to the given UnicodeString in the given radix.
+     * Standard digits '0'-'9' are used and letters 'A'-'Z' for
+     * radices 11 through 36.
+     * @param result the digits of the number are appended here
+     * @param n the number to be converted to digits; may be negative.
+     * If negative, a '-' is prepended to the digits.
+     * @param radix a radix from 2 to 36 inclusive.
+     * @param minDigits the minimum number of digits, not including
+     * any '-', to produce.  Values less than 2 have no effect.  One
+     * digit is always emitted regardless of this parameter.
+     * @return a reference to result
+     */
+    static UnicodeString& appendNumber(UnicodeString& result, int32_t n,
+                                       int32_t radix = 10,
+                                       int32_t minDigits = 1);
+
+    /** Returns a bogus UnicodeString by value. */
+    static inline UnicodeString makeBogusString() {
+        UnicodeString result;
+        result.setToBogus();
+        return result;
+    }
+
+    /**
+     * Return true if the character is NOT printable ASCII.
+     * The tab, newline and linefeed characters are considered unprintable.
+     */
+    static UBool isUnprintable(UChar32 c);
+
+    /**
+     * @return true for control codes and for surrogate and noncharacter code points
+     */
+    static UBool shouldAlwaysBeEscaped(UChar32 c);
+
+    /**
+     * Escapes one unprintable code point using \uxxxx notation for U+0000 to
+     * U+FFFF and \Uxxxxxxxx for U+10000 and above.  If the character is
+     * printable ASCII, then do nothing and return false.  Otherwise,
+     * append the escaped notation and return true.
+     */
+    static UBool escapeUnprintable(UnicodeString& result, UChar32 c);
+
+    /**
+     * Escapes one code point using \uxxxx notation
+     * for U+0000 to U+FFFF and \Uxxxxxxxx for U+10000 and above.
+     * @return result
+     */
+    static UnicodeString &escape(UnicodeString& result, UChar32 c);
+
+    /**
+     * Returns the index of a character, ignoring quoted text.
+     * For example, in the string "abc'hide'h", the 'h' in "hide" will not be
+     * found by a search for 'h'.
+     * @param text text to be searched
+     * @param start the beginning index, inclusive; 0 <= start
+     * <= limit.
+     * @param limit the ending index, exclusive; start <= limit
+     * <= text.length().
+     * @param c character to search for
+     * @return Offset of the first instance of c, or -1 if not found.
+     */
+//?FOR FUTURE USE.  DISABLE FOR NOW for coverage reasons.
+//    static int32_t quotedIndexOf(const UnicodeString& text,
+//                                 int32_t start, int32_t limit,
+//                                 char16_t c);
+
+    /**
+     * Skip over a sequence of zero or more white space characters at pos.
+     * @param advance if true, advance pos to the first non-white-space
+     * character at or after pos, or str.length(), if there is none.
+     * Otherwise leave pos unchanged.
+     * @return the index of the first non-white-space character at or
+     * after pos, or str.length(), if there is none.
+     */
+    static int32_t skipWhitespace(const UnicodeString& str, int32_t& pos,
+                                  UBool advance = false);
+
+    /**
+     * Skip over Pattern_White_Space in a Replaceable.
+     * Skipping may be done in the forward or
+     * reverse direction.  In either case, the leftmost index will be
+     * inclusive, and the rightmost index will be exclusive.  That is,
+     * given a range defined as [start, limit), the call
+     * skipWhitespace(text, start, limit) will advance start past leading
+     * whitespace, whereas the call skipWhitespace(text, limit, start),
+     * will back up limit past trailing whitespace.
+     * @param text the text to be analyzed
+     * @param pos either the start or limit of a range of 'text', to skip
+     * leading or trailing whitespace, respectively
+     * @param stop either the limit or start of a range of 'text', to skip
+     * leading or trailing whitespace, respectively
+     * @return the new start or limit, depending on what was passed in to
+     * 'pos'
+     */
+//?FOR FUTURE USE.  DISABLE FOR NOW for coverage reasons.
+//?    static int32_t skipWhitespace(const Replaceable& text,
+//?                                  int32_t pos, int32_t stop);
+
+    /**
+     * Parse a single non-whitespace character 'ch', optionally
+     * preceded by whitespace.
+     * @param id the string to be parsed
+     * @param pos INPUT-OUTPUT parameter.  On input, pos[0] is the
+     * offset of the first character to be parsed.  On output, pos[0]
+     * is the index after the last parsed character.  If the parse
+     * fails, pos[0] will be unchanged.
+     * @param ch the non-whitespace character to be parsed.
+     * @return true if 'ch' is seen preceded by zero or more
+     * whitespace characters.
+     */
+    static UBool parseChar(const UnicodeString& id, int32_t& pos, char16_t ch);
+
+    /**
+     * Parse a pattern string starting at offset pos.  Keywords are
+     * matched case-insensitively.  Spaces may be skipped and may be
+     * optional or required.  Integer values may be parsed, and if
+     * they are, they will be returned in the given array.  If
+     * successful, the offset of the next non-space character is
+     * returned.  On failure, -1 is returned.
+     * @param pattern must only contain lowercase characters, which
+     * will match their uppercase equivalents as well.  A space
+     * character matches one or more required spaces.  A '~' character
+     * matches zero or more optional spaces.  A '#' character matches
+     * an integer and stores it in parsedInts, which the caller must
+     * ensure has enough capacity.
+     * @param parsedInts array to receive parsed integers.  Caller
+     * must ensure that parsedInts.length is >= the number of '#'
+     * signs in 'pattern'.
+     * @return the position after the last character parsed, or -1 if
+     * the parse failed
+     */
+    static int32_t parsePattern(const UnicodeString& rule, int32_t pos, int32_t limit,
+                                const UnicodeString& pattern, int32_t* parsedInts);
+        
+    /**
+     * Parse a pattern string within the given Replaceable and a parsing
+     * pattern.  Characters are matched literally and case-sensitively
+     * except for the following special characters:
+     *
+     * ~  zero or more Pattern_White_Space chars
+     *
+     * If end of pattern is reached with all matches along the way,
+     * pos is advanced to the first unparsed index and returned.
+     * Otherwise -1 is returned.
+     * @param pat pattern that controls parsing
+     * @param text text to be parsed, starting at index
+     * @param index offset to first character to parse
+     * @param limit offset after last character to parse
+     * @return index after last parsed character, or -1 on parse failure.
+     */
+    static int32_t parsePattern(const UnicodeString& pat,
+                                const Replaceable& text,
+                                int32_t index,
+                                int32_t limit);
+
+    /**
+     * Parse an integer at pos, either of the form \d+ or of the form
+     * 0x[0-9A-Fa-f]+ or 0[0-7]+, that is, in standard decimal, hex,
+     * or octal format.
+     * @param pos INPUT-OUTPUT parameter.  On input, the index of the first
+     * character to parse.  On output, the index of the character after the
+     * last parsed character.
+     */
+    static int32_t parseInteger(const UnicodeString& rule, int32_t& pos, int32_t limit);
+
+    /**
+     * Parse an integer at pos using only ASCII digits.
+     * Base 10 only.
+     * @param pos INPUT-OUTPUT parameter.  On input, the index of the first
+     * character to parse.  On output, the index of the character after the
+     * last parsed character.
+     */
+    static int32_t parseAsciiInteger(const UnicodeString& str, int32_t& pos);
+
+    /**
+     * Parse a Unicode identifier from the given string at the given
+     * position.  Return the identifier, or an empty string if there
+     * is no identifier.
+     * @param str the string to parse
+     * @param pos INPUT-OUTPUT parameter.  On INPUT, pos is the
+     * first character to examine.  It must be less than str.length(),
+     * and it must not point to a whitespace character.  That is, must
+     * have pos < str.length() and
+     * !UCharacter::isWhitespace(str.char32At(pos)).  On
+     * OUTPUT, the position after the last parsed character.
+     * @return the Unicode identifier, or an empty string if there is
+     * no valid identifier at pos.
+     */
+    static UnicodeString parseUnicodeIdentifier(const UnicodeString& str, int32_t& pos);
+
+    /**
+     * Parse an unsigned 31-bit integer at the given offset.  Use
+     * UCharacter.digit() to parse individual characters into digits.
+     * @param text the text to be parsed
+     * @param pos INPUT-OUTPUT parameter.  On entry, pos is the
+     * offset within text at which to start parsing; it should point
+     * to a valid digit.  On exit, pos is the offset after the last
+     * parsed character.  If the parse failed, it will be unchanged on
+     * exit.  Must be >= 0 on entry.
+     * @param radix the radix in which to parse; must be >= 2 and <=
+     * 36.
+     * @return a non-negative parsed number, or -1 upon parse failure.
+     * Parse fails if there are no digits, that is, if pos does not
+     * point to a valid digit on entry, or if the number to be parsed
+     * does not fit into a 31-bit unsigned integer.
+     */
+    static int32_t parseNumber(const UnicodeString& text,
+                               int32_t& pos, int8_t radix);
+
+    static void appendToRule(UnicodeString& rule,
+                             UChar32 c,
+                             UBool isLiteral,
+                             UBool escapeUnprintable,
+                             UnicodeString& quoteBuf);
+    
+    static void appendToRule(UnicodeString& rule,
+                             const UnicodeString& text,
+                             UBool isLiteral,
+                             UBool escapeUnprintable,
+                             UnicodeString& quoteBuf);
+
+    static void appendToRule(UnicodeString& rule,
+                             const UnicodeMatcher* matcher,
+                             UBool escapeUnprintable,
+                             UnicodeString& quoteBuf);
+
+private:
+    // do not instantiate
+    ICU_Utility() = delete;
+};
+
+U_NAMESPACE_END
+
+#endif
+//eof
diff --git a/intl/icu/source/common/util_props.cpp b/intl/icu/source/common/util_props.cpp
new file mode 100644
index 0000000000..5991769d06
--- /dev/null
+++ b/intl/icu/source/common/util_props.cpp
@@ -0,0 +1,217 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+*   Copyright (c) 2001-2016, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*   Date        Name        Description
+*   11/19/2001  aliu        Creation.
+**********************************************************************
+*/
+
+#include "unicode/uchar.h"
+#include "unicode/utf16.h"
+#include "patternprops.h"
+#include "util.h"
+
+U_NAMESPACE_BEGIN
+
+/**
+ * Parse an integer at pos, either of the form \d+ or of the form
+ * 0x[0-9A-Fa-f]+ or 0[0-7]+, that is, in standard decimal, hex,
+ * or octal format.
+ * @param pos INPUT-OUTPUT parameter.  On input, the first
+ * character to parse.  On output, the character after the last
+ * parsed character.
+ */
+int32_t ICU_Utility::parseInteger(const UnicodeString& rule, int32_t& pos, int32_t limit) {
+    int32_t count = 0;
+    int32_t value = 0;
+    int32_t p = pos;
+    int8_t radix = 10;
+
+    if (p < limit && rule.charAt(p) == 48 /*0*/) {
+        if (p+1 < limit && (rule.charAt(p+1) == 0x78 /*x*/ || rule.charAt(p+1) == 0x58 /*X*/)) {
+            p += 2;
+            radix = 16;
+        }
+        else {
+            p++;
+            count = 1;
+            radix = 8;
+        }
+    }
+
+    while (p < limit) {
+        int32_t d = u_digit(rule.charAt(p++), radix);
+        if (d < 0) {
+            --p;
+            break;
+        }
+        ++count;
+        int32_t v = (value * radix) + d;
+        if (v <= value) {
+            // If there are too many input digits, at some point
+            // the value will go negative, e.g., if we have seen
+            // "0x8000000" already and there is another '0', when
+            // we parse the next 0 the value will go negative.
+            return 0;
+        }
+        value = v;
+    }
+    if (count > 0) {
+        pos = p;
+    }
+    return value;
+}
+
+/**
+ * Parse a pattern string starting at offset pos.  Keywords are
+ * matched case-insensitively.  Spaces may be skipped and may be
+ * optional or required.  Integer values may be parsed, and if
+ * they are, they will be returned in the given array.  If
+ * successful, the offset of the next non-space character is
+ * returned.  On failure, -1 is returned.
+ * @param pattern must only contain lowercase characters, which
+ * will match their uppercase equivalents as well.  A space
+ * character matches one or more required spaces.  A '~' character
+ * matches zero or more optional spaces.  A '#' character matches
+ * an integer and stores it in parsedInts, which the caller must
+ * ensure has enough capacity.
+ * @param parsedInts array to receive parsed integers.  Caller
+ * must ensure that parsedInts.length is >= the number of '#'
+ * signs in 'pattern'.
+ * @return the position after the last character parsed, or -1 if
+ * the parse failed
+ */
+int32_t ICU_Utility::parsePattern(const UnicodeString& rule, int32_t pos, int32_t limit,
+                              const UnicodeString& pattern, int32_t* parsedInts) {
+    // TODO Update this to handle surrogates
+    int32_t p;
+    int32_t intCount = 0; // number of integers parsed
+    for (int32_t i=0; i= limit) {
+                return -1;
+            }
+            c = rule.charAt(pos++);
+            if (!PatternProps::isWhiteSpace(c)) {
+                return -1;
+            }
+            // FALL THROUGH to skipWhitespace
+            U_FALLTHROUGH;
+        case 126 /*'~'*/:
+            pos = skipWhitespace(rule, pos);
+            break;
+        case 35 /*'#'*/:
+            p = pos;
+            parsedInts[intCount++] = parseInteger(rule, p, limit);
+            if (p == pos) {
+                // Syntax error; failed to parse integer
+                return -1;
+            }
+            pos = p;
+            break;
+        default:
+            if (pos >= limit) {
+                return -1;
+            }
+            c = (char16_t) u_tolower(rule.charAt(pos++));
+            if (c != cpat) {
+                return -1;
+            }
+            break;
+        }
+    }
+    return pos;
+}
+
+/**
+ * Parse a Unicode identifier from the given string at the given
+ * position.  Return the identifier, or an empty string if there
+ * is no identifier.
+ * @param str the string to parse
+ * @param pos INPUT-OUTPUT parameter.  On INPUT, pos is the
+ * first character to examine.  It must be less than str.length(),
+ * and it must not point to a whitespace character.  That is, must
+ * have pos < str.length().  On
+ * OUTPUT, the position after the last parsed character.
+ * @return the Unicode identifier, or an empty string if there is
+ * no valid identifier at pos.
+ */
+UnicodeString ICU_Utility::parseUnicodeIdentifier(const UnicodeString& str, int32_t& pos) {
+    // assert(pos < str.length());
+    UnicodeString buf;
+    int p = pos;
+    while (p < str.length()) {
+        UChar32 ch = str.char32At(p);
+        if (buf.length() == 0) {
+            if (u_isIDStart(ch)) {
+                buf.append(ch);
+            } else {
+                buf.truncate(0);
+                return buf;
+            }
+        } else {
+            if (u_isIDPart(ch)) {
+                buf.append(ch);
+            } else {
+                break;
+            }
+        }
+        p += U16_LENGTH(ch);
+    }
+    pos = p;
+    return buf;
+}
+
+/**
+ * Parse an unsigned 31-bit integer at the given offset.  Use
+ * UCharacter.digit() to parse individual characters into digits.
+ * @param text the text to be parsed
+ * @param pos INPUT-OUTPUT parameter.  On entry, pos[0] is the
+ * offset within text at which to start parsing; it should point
+ * to a valid digit.  On exit, pos[0] is the offset after the last
+ * parsed character.  If the parse failed, it will be unchanged on
+ * exit.  Must be >= 0 on entry.
+ * @param radix the radix in which to parse; must be >= 2 and <=
+ * 36.
+ * @return a non-negative parsed number, or -1 upon parse failure.
+ * Parse fails if there are no digits, that is, if pos[0] does not
+ * point to a valid digit on entry, or if the number to be parsed
+ * does not fit into a 31-bit unsigned integer.
+ */
+int32_t ICU_Utility::parseNumber(const UnicodeString& text,
+                                 int32_t& pos, int8_t radix) {
+    // assert(pos[0] >= 0);
+    // assert(radix >= 2);
+    // assert(radix <= 36);
+    int32_t n = 0;
+    int32_t p = pos;
+    while (p < text.length()) {
+        UChar32 ch = text.char32At(p);
+        int32_t d = u_digit(ch, radix);
+        if (d < 0) {
+            break;
+        }
+        n = radix*n + d;
+        // ASSUME that when a 32-bit integer overflows it becomes
+        // negative.  E.g., 214748364 * 10 + 8 => negative value.
+        if (n < 0) {
+            return -1;
+        }
+        ++p;
+    }
+    if (p == pos) {
+        return -1;
+    }
+    pos = p;
+    return n;
+}
+
+U_NAMESPACE_END
+
diff --git a/intl/icu/source/common/utrace.cpp b/intl/icu/source/common/utrace.cpp
new file mode 100644
index 0000000000..1f536d5643
--- /dev/null
+++ b/intl/icu/source/common/utrace.cpp
@@ -0,0 +1,504 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*   Copyright (C) 2003-2014, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*******************************************************************************
+*   file name:  utrace.c
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*/
+
+#include "unicode/utrace.h"
+#include "utracimp.h"
+#include "cstring.h"
+#include "uassert.h"
+#include "ucln_cmn.h"
+
+
+static UTraceEntry     *pTraceEntryFunc = nullptr;
+static UTraceExit      *pTraceExitFunc  = nullptr;
+static UTraceData      *pTraceDataFunc  = nullptr;
+static const void      *gTraceContext   = nullptr;
+
+/**
+ * \var utrace_level
+ * Trace level variable. Negative for "off".
+ */
+static int32_t
+utrace_level = UTRACE_ERROR;
+
+U_CAPI void U_EXPORT2
+utrace_entry(int32_t fnNumber) {
+    if (pTraceEntryFunc != nullptr) {
+        (*pTraceEntryFunc)(gTraceContext, fnNumber);
+    }
+}
+
+
+static const char gExitFmt[]             = "Returns.";
+static const char gExitFmtValue[]        = "Returns %d.";
+static const char gExitFmtStatus[]       = "Returns.  Status = %d.";
+static const char gExitFmtValueStatus[]  = "Returns %d.  Status = %d.";
+static const char gExitFmtPtrStatus[]    = "Returns %d.  Status = %p.";
+
+U_CAPI void U_EXPORT2
+utrace_exit(int32_t fnNumber, int32_t returnType, ...) {
+    if (pTraceExitFunc != nullptr) {
+        va_list     args;
+        const char *fmt;
+
+        switch (returnType) {
+        case 0:
+            fmt = gExitFmt;
+            break;
+        case UTRACE_EXITV_I32:
+            fmt = gExitFmtValue;
+            break;
+        case UTRACE_EXITV_STATUS:
+            fmt = gExitFmtStatus;
+            break;
+        case UTRACE_EXITV_I32 | UTRACE_EXITV_STATUS:
+            fmt = gExitFmtValueStatus;
+            break;
+        case UTRACE_EXITV_PTR | UTRACE_EXITV_STATUS:
+            fmt = gExitFmtPtrStatus;
+            break;
+        default:
+            UPRV_UNREACHABLE_EXIT;
+        }
+
+        va_start(args, returnType);
+        (*pTraceExitFunc)(gTraceContext, fnNumber, fmt, args);
+        va_end(args);
+    }
+}
+ 
+
+ 
+U_CAPI void U_EXPORT2 
+utrace_data(int32_t fnNumber, int32_t level, const char *fmt, ...) {
+    if (pTraceDataFunc != nullptr) {
+           va_list args;
+           va_start(args, fmt ); 
+           (*pTraceDataFunc)(gTraceContext, fnNumber, level, fmt, args);
+           va_end(args);
+    }
+}
+
+
+static void outputChar(char c, char *outBuf, int32_t *outIx, int32_t capacity, int32_t indent) {
+    int32_t i;
+    /* Check whether a start of line indenting is needed.  Three cases:
+     *   1.  At the start of the first line  (output index == 0).
+     *   2.  At the start of subsequent lines  (preceding char in buffer == '\n')
+     *   3.  When preflighting buffer len (buffer capacity is exceeded), when
+     *       a \n is output.  Ideally we wouldn't do the indent until the following char
+     *       is received, but that won't work because there's no place to remember that
+     *       the preceding char was \n.  Meaning that we may overstimate the
+     *       buffer size needed.  No harm done.
+     */
+    if (*outIx==0 ||   /* case 1. */
+        (c!='\n' && c!=0 && *outIx < capacity && outBuf[(*outIx)-1]=='\n') ||  /* case 2. */
+        (c=='\n' && *outIx>=capacity))    /* case 3 */
+    {
+        /* At the start of a line.  Indent. */
+        for(i=0; i= 0; shiftCount-=4) {
+        char c = gHexChars[(val >> shiftCount) & 0xf];
+        outputChar(c, outBuf, outIx, capacity, 0);
+    }
+}
+
+/* Output a pointer value in hex.  Work with any size of pointer   */
+static void outputPtrBytes(void *val, char *outBuf, int32_t *outIx, int32_t capacity) {
+    uint32_t  i;
+    int32_t  incVal = 1;              /* +1 for big endian, -1 for little endian          */
+    char     *p     = (char *)&val;   /* point to current byte to output in the ptr val  */
+
+#if !U_IS_BIG_ENDIAN
+    /* Little Endian.  Move p to most significant end of the value      */
+    incVal = -1;
+    p += sizeof(void *) - 1;
+#endif
+
+    /* Loop through the bytes of the ptr as it sits in memory, from 
+     * most significant to least significant end                    */
+    for (i=0; i 0) {
+                            outputHexBytes(longArg, charsToOutput, outBuf, &outIx, capacity);
+                            outputChar(' ', outBuf, &outIx, capacity, indent);
+                        }
+                        if (vectorLen == -1 && longArg == 0) {
+                            break;
+                        }
+                    }
+                }
+                outputChar('[', outBuf, &outIx, capacity, indent);
+                outputHexBytes(vectorLen, 8, outBuf, &outIx, capacity);
+                outputChar(']', outBuf, &outIx, capacity, indent);
+            }
+            break;
+
+
+        default:
+            /* %. in format string, where . is some character not in the set
+             *    of recognized format chars.  Just output it as if % wasn't there.
+             *    (Covers "%%" outputting a single '%')
+             */
+             outputChar(fmtC, outBuf, &outIx, capacity, indent);
+        }
+    }
+    outputChar(0, outBuf, &outIx, capacity, indent);  /* Make sure that output is NUL terminated   */
+    return outIx + 1;     /* outIx + 1 because outIx does not increment when outputting final NUL. */
+}
+
+
+
+
+U_CAPI int32_t U_EXPORT2
+utrace_format(char *outBuf, int32_t capacity,
+                int32_t indent, const char *fmt,  ...) {
+    int32_t retVal;
+    va_list args;
+    va_start(args, fmt ); 
+    retVal = utrace_vformat(outBuf, capacity, indent, fmt, args);
+    va_end(args);
+    return retVal;
+}
+
+
+U_CAPI void U_EXPORT2
+utrace_setFunctions(const void *context,
+                    UTraceEntry *e, UTraceExit *x, UTraceData *d) {
+    pTraceEntryFunc = e;
+    pTraceExitFunc  = x;
+    pTraceDataFunc  = d;
+    gTraceContext   = context;
+}
+
+
+U_CAPI void U_EXPORT2
+utrace_getFunctions(const void **context,
+                    UTraceEntry **e, UTraceExit **x, UTraceData **d) {
+    *e = pTraceEntryFunc;
+    *x = pTraceExitFunc;
+    *d = pTraceDataFunc;
+    *context = gTraceContext;
+}
+
+U_CAPI void U_EXPORT2
+utrace_setLevel(int32_t level) {
+    if (level < UTRACE_OFF) {
+        level = UTRACE_OFF;
+    }
+    if (level > UTRACE_VERBOSE) {
+        level = UTRACE_VERBOSE;
+    }
+    utrace_level = level;
+}
+
+U_CAPI int32_t U_EXPORT2
+utrace_getLevel() {
+    return utrace_level;
+}
+
+
+U_CFUNC UBool 
+utrace_cleanup() {
+    pTraceEntryFunc = nullptr;
+    pTraceExitFunc  = nullptr;
+    pTraceDataFunc  = nullptr;
+    utrace_level    = UTRACE_OFF;
+    gTraceContext   = nullptr;
+    return true;
+}
+
+
+static const char * const
+trFnName[] = {
+    "u_init",
+    "u_cleanup",
+    nullptr
+};
+
+
+static const char * const
+trConvNames[] = {
+    "ucnv_open",
+    "ucnv_openPackage",
+    "ucnv_openAlgorithmic",
+    "ucnv_clone",
+    "ucnv_close",
+    "ucnv_flushCache",
+    "ucnv_load",
+    "ucnv_unload",
+    nullptr
+};
+
+    
+static const char * const
+trCollNames[] = {
+    "ucol_open",
+    "ucol_close",
+    "ucol_strcoll",
+    "ucol_getSortKey",
+    "ucol_getLocale",
+    "ucol_nextSortKeyPart",
+    "ucol_strcollIter",
+    "ucol_openFromShortString",
+    "ucol_strcollUTF8",
+    nullptr
+};
+
+
+static const char* const
+trResDataNames[] = {
+    "resc",
+    "bundle-open",
+    "file-open",
+    "res-open",
+    nullptr
+};
+
+                
+U_CAPI const char * U_EXPORT2
+utrace_functionName(int32_t fnNumber) {
+    if(UTRACE_FUNCTION_START <= fnNumber && fnNumber < UTRACE_FUNCTION_LIMIT) {
+        return trFnName[fnNumber];
+    } else if(UTRACE_CONVERSION_START <= fnNumber && fnNumber < UTRACE_CONVERSION_LIMIT) {
+        return trConvNames[fnNumber - UTRACE_CONVERSION_START];
+    } else if(UTRACE_COLLATION_START <= fnNumber && fnNumber < UTRACE_COLLATION_LIMIT){
+        return trCollNames[fnNumber - UTRACE_COLLATION_START];
+    } else if(UTRACE_UDATA_START <= fnNumber && fnNumber < UTRACE_RES_DATA_LIMIT){
+        return trResDataNames[fnNumber - UTRACE_UDATA_START];
+    } else {
+        return "[BOGUS Trace Function Number]";
+    }
+}
+
diff --git a/intl/icu/source/common/utracimp.h b/intl/icu/source/common/utracimp.h
new file mode 100644
index 0000000000..945540d25a
--- /dev/null
+++ b/intl/icu/source/common/utracimp.h
@@ -0,0 +1,391 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2003-2009, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  utracimp.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2003aug06
+*   created by: Markus W. Scherer
+*
+*   Internal header for ICU tracing/logging.
+*
+*
+*   Various notes:
+*   - using a trace level variable to only call trace functions
+*     when the level is sufficient
+*   - using the same variable for tracing on/off to never make a function
+*     call when off
+*   - the function number is put into a local variable by the entry macro
+*     and used implicitly to avoid copy&paste/typing mistakes by the developer
+*   - the application must call utrace_setFunctions() and pass in
+*     implementations for the trace functions
+*   - ICU trace macros call ICU functions that route through the function
+*     pointers if they have been set;
+*     this avoids an indirection at the call site
+*     (which would cost more code for another check and for the indirection)
+*
+*   ### TODO Issues:
+*   - Verify that va_list is portable among compilers for the same platform.
+*     va_list should be portable because printf() would fail otherwise!
+*   - Should enum values like UTraceLevel be passed into int32_t-type arguments,
+*     or should enum types be used?
+*/
+
+#ifndef __UTRACIMP_H__
+#define __UTRACIMP_H__
+
+#include "unicode/utrace.h"
+#include 
+
+U_CDECL_BEGIN
+
+/** 
+ *   Traced Function Exit return types.  
+ *   Flags indicating the number and types of varargs included in a call
+ *   to a UTraceExit function.
+ *   Bits 0-3:  The function return type.  First variable param.
+ *   Bit    4:  Flag for presence of U_ErrorCode status param.
+ *   @internal
+ */
+typedef enum UTraceExitVal {
+    /** The traced function returns no value  @internal */
+    UTRACE_EXITV_NONE   = 0,
+    /** The traced function returns an int32_t, or compatible, type.  @internal */
+    UTRACE_EXITV_I32    = 1,
+    /** The traced function returns a pointer  @internal */
+    UTRACE_EXITV_PTR    = 2,
+    /** The traced function returns a UBool  @internal */
+    UTRACE_EXITV_BOOL   = 3,
+    /** Mask to extract the return type values from a UTraceExitVal  @internal */
+    UTRACE_EXITV_MASK   = 0xf,
+    /** Bit indicating that the traced function includes a UErrorCode parameter  @internal */
+    UTRACE_EXITV_STATUS = 0x10
+} UTraceExitVal;
+
+/**
+ * Trace function for the entry point of a function.
+ * Do not use directly, use UTRACE_ENTRY instead.
+ * @param fnNumber The UTraceFunctionNumber for the current function.
+ * @internal
+ */
+U_CAPI void U_EXPORT2
+utrace_entry(int32_t fnNumber);
+
+/**
+ * Trace function for each exit point of a function.
+ * Do not use directly, use UTRACE_EXIT* instead.
+ * @param fnNumber The UTraceFunctionNumber for the current function.
+ * @param returnType The type of the value returned by the function.
+ * @param errorCode The UErrorCode value at function exit. See UTRACE_EXIT.
+ * @internal
+ */
+U_CAPI void U_EXPORT2
+utrace_exit(int32_t fnNumber, int32_t returnType, ...);
+
+
+/**
+ * Trace function used inside functions that have a UTRACE_ENTRY() statement.
+ * Do not use directly, use UTRACE_DATAX() macros instead.
+ *
+ * @param utraceFnNumber The number of the current function, from the local
+ *        variable of the same name.
+ * @param level The trace level for this message.
+ * @param fmt The trace format string.
+ *
+ * @internal
+ */
+U_CAPI void U_EXPORT2
+utrace_data(int32_t utraceFnNumber, int32_t level, const char *fmt, ...);
+
+U_CDECL_END
+
+#if U_ENABLE_TRACING
+
+/**
+ * Boolean expression to see if ICU tracing is turned on
+ * to at least the specified level.
+ * @internal
+ */
+#define UTRACE_LEVEL(level) (utrace_getLevel()>=(level))
+
+/**
+  *  Flag bit in utraceFnNumber, the local variable added to each function 
+  *  with tracing code to contains the function number.
+  *
+  *  Set the flag if the function's entry is traced, which will cause the
+  *  function's exit to also be traced.  utraceFnNumber is uncoditionally 
+  *  set at entry, whether or not the entry is traced, so that it will
+  *  always be available for error trace output.
+  *  @internal
+  */            
+#define UTRACE_TRACED_ENTRY 0x80000000
+
+/**
+ * Trace statement for the entry point of a function.
+ * Stores the function number in a local variable.
+ * In C code, must be placed immediately after the last variable declaration.
+ * Must be matched with UTRACE_EXIT() at all function exit points.
+ *
+ * Tracing should start with UTRACE_ENTRY after checking for
+ * U_FAILURE at function entry, so that if a function returns immediately
+ * because of a pre-existing error condition, it does not show up in the trace,
+ * consistent with ICU's error handling model.
+ *
+ * @param fnNumber The UTraceFunctionNumber for the current function.
+ * @internal
+ */
+#define UTRACE_ENTRY(fnNumber) \
+    int32_t utraceFnNumber=(fnNumber); \
+UPRV_BLOCK_MACRO_BEGIN { \
+    if(utrace_getLevel()>=UTRACE_INFO) { \
+        utrace_entry(fnNumber); \
+        utraceFnNumber |= UTRACE_TRACED_ENTRY; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+
+/**
+ * Trace statement for the entry point of open and close functions.
+ * Produces trace output at a less verbose setting than plain UTRACE_ENTRY
+ * Stores the function number in a local variable.
+ * In C code, must be placed immediately after the last variable declaration.
+ * Must be matched with UTRACE_EXIT() at all function exit points.
+ *
+ * @param fnNumber The UTraceFunctionNumber for the current function.
+ * @internal
+ */
+#define UTRACE_ENTRY_OC(fnNumber) \
+    int32_t utraceFnNumber=(fnNumber); \
+UPRV_BLOCK_MACRO_BEGIN { \
+    if(utrace_getLevel()>=UTRACE_OPEN_CLOSE) { \
+        utrace_entry(fnNumber); \
+        utraceFnNumber |= UTRACE_TRACED_ENTRY; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Trace statement for each exit point of a function that has a UTRACE_ENTRY()
+ * statement.
+ *
+ * @param errorCode The function's ICU UErrorCode value at function exit,
+ *                  or U_ZERO_ERROR if the function does not use a UErrorCode.
+ *                  0==U_ZERO_ERROR indicates success,
+ *                  positive values an error (see u_errorName()),
+ *                  negative values an informational status.
+ *
+ * @internal
+ */
+#define UTRACE_EXIT() UPRV_BLOCK_MACRO_BEGIN { \
+    if(utraceFnNumber & UTRACE_TRACED_ENTRY) { \
+        utrace_exit(utraceFnNumber & ~UTRACE_TRACED_ENTRY, UTRACE_EXITV_NONE); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Trace statement for each exit point of a function that has a UTRACE_ENTRY()
+ * statement, and that returns a value.
+ *
+ * @param val       The function's return value, int32_t or compatible type.
+ *
+ * @internal 
+ */
+#define UTRACE_EXIT_VALUE(val) UPRV_BLOCK_MACRO_BEGIN { \
+    if(utraceFnNumber & UTRACE_TRACED_ENTRY) { \
+        utrace_exit(utraceFnNumber & ~UTRACE_TRACED_ENTRY, UTRACE_EXITV_I32, val); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+#define UTRACE_EXIT_STATUS(status) UPRV_BLOCK_MACRO_BEGIN { \
+    if(utraceFnNumber & UTRACE_TRACED_ENTRY) { \
+        utrace_exit(utraceFnNumber & ~UTRACE_TRACED_ENTRY, UTRACE_EXITV_STATUS, status); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+#define UTRACE_EXIT_VALUE_STATUS(val, status) UPRV_BLOCK_MACRO_BEGIN { \
+    if(utraceFnNumber & UTRACE_TRACED_ENTRY) { \
+        utrace_exit(utraceFnNumber & ~UTRACE_TRACED_ENTRY, (UTRACE_EXITV_I32 | UTRACE_EXITV_STATUS), val, status); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+#define UTRACE_EXIT_PTR_STATUS(ptr, status) UPRV_BLOCK_MACRO_BEGIN { \
+    if(utraceFnNumber & UTRACE_TRACED_ENTRY) { \
+        utrace_exit(utraceFnNumber & ~UTRACE_TRACED_ENTRY, (UTRACE_EXITV_PTR | UTRACE_EXITV_STATUS), ptr, status); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Trace statement used inside functions that have a UTRACE_ENTRY() statement.
+ * Takes no data arguments.
+ * The number of arguments for this macro must match the number of inserts
+ * in the format string. Vector inserts count as two arguments.
+ * Calls utrace_data() if the level is high enough.
+ * @internal
+ */
+#define UTRACE_DATA0(level, fmt) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTRACE_LEVEL(level)) { \
+        utrace_data(utraceFnNumber & ~UTRACE_TRACED_ENTRY, (level), (fmt)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Trace statement used inside functions that have a UTRACE_ENTRY() statement.
+ * Takes one data argument.
+ * The number of arguments for this macro must match the number of inserts
+ * in the format string. Vector inserts count as two arguments.
+ * Calls utrace_data() if the level is high enough.
+ * @internal
+ */
+#define UTRACE_DATA1(level, fmt, a) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTRACE_LEVEL(level)) { \
+        utrace_data(utraceFnNumber & ~UTRACE_TRACED_ENTRY , (level), (fmt), (a)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Trace statement used inside functions that have a UTRACE_ENTRY() statement.
+ * Takes two data arguments.
+ * The number of arguments for this macro must match the number of inserts
+ * in the format string. Vector inserts count as two arguments.
+ * Calls utrace_data() if the level is high enough.
+ * @internal
+ */
+#define UTRACE_DATA2(level, fmt, a, b) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTRACE_LEVEL(level)) { \
+        utrace_data(utraceFnNumber & ~UTRACE_TRACED_ENTRY , (level), (fmt), (a), (b)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Trace statement used inside functions that have a UTRACE_ENTRY() statement.
+ * Takes three data arguments.
+ * The number of arguments for this macro must match the number of inserts
+ * in the format string. Vector inserts count as two arguments.
+ * Calls utrace_data() if the level is high enough.
+ * @internal
+ */
+#define UTRACE_DATA3(level, fmt, a, b, c) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTRACE_LEVEL(level)) { \
+        utrace_data(utraceFnNumber & ~UTRACE_TRACED_ENTRY, (level), (fmt), (a), (b), (c)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Trace statement used inside functions that have a UTRACE_ENTRY() statement.
+ * Takes four data arguments.
+ * The number of arguments for this macro must match the number of inserts
+ * in the format string. Vector inserts count as two arguments.
+ * Calls utrace_data() if the level is high enough.
+ * @internal
+ */
+#define UTRACE_DATA4(level, fmt, a, b, c, d) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTRACE_LEVEL(level)) { \
+        utrace_data(utraceFnNumber & ~UTRACE_TRACED_ENTRY, (level), (fmt), (a), (b), (c), (d)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Trace statement used inside functions that have a UTRACE_ENTRY() statement.
+ * Takes five data arguments.
+ * The number of arguments for this macro must match the number of inserts
+ * in the format string. Vector inserts count as two arguments.
+ * Calls utrace_data() if the level is high enough.
+ * @internal
+ */
+#define UTRACE_DATA5(level, fmt, a, b, c, d, e) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTRACE_LEVEL(level)) { \
+        utrace_data(utraceFnNumber & ~UTRACE_TRACED_ENTRY, (level), (fmt), (a), (b), (c), (d), (e)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Trace statement used inside functions that have a UTRACE_ENTRY() statement.
+ * Takes six data arguments.
+ * The number of arguments for this macro must match the number of inserts
+ * in the format string. Vector inserts count as two arguments.
+ * Calls utrace_data() if the level is high enough.
+ * @internal
+ */
+#define UTRACE_DATA6(level, fmt, a, b, c, d, e, f) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTRACE_LEVEL(level)) { \
+        utrace_data(utraceFnNumber & ~UTRACE_TRACED_ENTRY, (level), (fmt), (a), (b), (c), (d), (e), (f)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Trace statement used inside functions that have a UTRACE_ENTRY() statement.
+ * Takes seven data arguments.
+ * The number of arguments for this macro must match the number of inserts
+ * in the format string. Vector inserts count as two arguments.
+ * Calls utrace_data() if the level is high enough.
+ * @internal
+ */
+#define UTRACE_DATA7(level, fmt, a, b, c, d, e, f, g) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTRACE_LEVEL(level)) { \
+        utrace_data(utraceFnNumber & ~UTRACE_TRACED_ENTRY, (level), (fmt), (a), (b), (c), (d), (e), (f), (g)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Trace statement used inside functions that have a UTRACE_ENTRY() statement.
+ * Takes eight data arguments.
+ * The number of arguments for this macro must match the number of inserts
+ * in the format string. Vector inserts count as two arguments.
+ * Calls utrace_data() if the level is high enough.
+ * @internal
+ */
+#define UTRACE_DATA8(level, fmt, a, b, c, d, e, f, g, h) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTRACE_LEVEL(level)) { \
+        utrace_data(utraceFnNumber & ~UTRACE_TRACED_ENTRY, (level), (fmt), (a), (b), (c), (d), (e), (f), (g), (h)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/**
+ * Trace statement used inside functions that have a UTRACE_ENTRY() statement.
+ * Takes nine data arguments.
+ * The number of arguments for this macro must match the number of inserts
+ * in the format string. Vector inserts count as two arguments.
+ * Calls utrace_data() if the level is high enough.
+ * @internal
+ */
+#define UTRACE_DATA9(level, fmt, a, b, c, d, e, f, g, h, i) UPRV_BLOCK_MACRO_BEGIN { \
+    if(UTRACE_LEVEL(level)) { \
+        utrace_data(utraceFnNumber & ~UTRACE_TRACED_ENTRY, (level), (fmt), (a), (b), (c), (d), (e), (f), (g), (h), (i)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+#else
+
+/*
+ * When tracing is disabled, the following macros become empty
+ */
+
+#define UTRACE_LEVEL(level) 0
+#define UTRACE_ENTRY(fnNumber)
+#define UTRACE_ENTRY_OC(fnNumber)
+#define UTRACE_EXIT()
+#define UTRACE_EXIT_VALUE(val)
+#define UTRACE_EXIT_STATUS(status)
+#define UTRACE_EXIT_VALUE_STATUS(val, status)
+#define UTRACE_EXIT_PTR_STATUS(ptr, status)
+#define UTRACE_DATA0(level, fmt)
+#define UTRACE_DATA1(level, fmt, a)
+#define UTRACE_DATA2(level, fmt, a, b)
+#define UTRACE_DATA3(level, fmt, a, b, c)
+#define UTRACE_DATA4(level, fmt, a, b, c, d)
+#define UTRACE_DATA5(level, fmt, a, b, c, d, e)
+#define UTRACE_DATA6(level, fmt, a, b, c, d, e, f)
+#define UTRACE_DATA7(level, fmt, a, b, c, d, e, f, g)
+#define UTRACE_DATA8(level, fmt, a, b, c, d, e, f, g, h)
+#define UTRACE_DATA9(level, fmt, a, b, c, d, e, f, g, h, i)
+
+#endif
+
+#endif
diff --git a/intl/icu/source/common/utrie.cpp b/intl/icu/source/common/utrie.cpp
new file mode 100644
index 0000000000..83be0e3407
--- /dev/null
+++ b/intl/icu/source/common/utrie.cpp
@@ -0,0 +1,1234 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+*
+*   Copyright (C) 2001-2012, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+******************************************************************************
+*   file name:  utrie.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2001oct20
+*   created by: Markus W. Scherer
+*
+*   This is a common implementation of a "folded" trie.
+*   It is a kind of compressed, serializable table of 16- or 32-bit values associated with
+*   Unicode code points (0..0x10ffff).
+*/
+
+#ifdef UTRIE_DEBUG
+#   include 
+#endif
+
+#include "unicode/utypes.h"
+#include "cmemory.h"
+#include "utrie.h"
+
+/* miscellaneous ------------------------------------------------------------ */
+
+#undef ABS
+#define ABS(x) ((x)>=0 ? (x) : -(x))
+
+static inline UBool
+equal_uint32(const uint32_t *s, const uint32_t *t, int32_t length) {
+    while(length>0 && *s==*t) {
+        ++s;
+        ++t;
+        --length;
+    }
+    return (UBool)(length==0);
+}
+
+/* Building a trie ----------------------------------------------------------*/
+
+U_CAPI UNewTrie * U_EXPORT2
+utrie_open(UNewTrie *fillIn,
+           uint32_t *aliasData, int32_t maxDataLength,
+           uint32_t initialValue, uint32_t leadUnitValue,
+           UBool latin1Linear) {
+    UNewTrie *trie;
+    int32_t i, j;
+
+    if( maxDataLengthisAllocated= (UBool)(fillIn==nullptr);
+
+    if(aliasData!=nullptr) {
+        trie->data=aliasData;
+        trie->isDataAllocated=false;
+    } else {
+        trie->data=(uint32_t *)uprv_malloc(maxDataLength*4);
+        if(trie->data==nullptr) {
+            uprv_free(trie);
+            return nullptr;
+        }
+        trie->isDataAllocated=true;
+    }
+
+    /* preallocate and reset the first data block (block index 0) */
+    j=UTRIE_DATA_BLOCK_LENGTH;
+
+    if(latin1Linear) {
+        /* preallocate and reset the first block (number 0) and Latin-1 (U+0000..U+00ff) after that */
+        /* made sure above that maxDataLength>=1024 */
+
+        /* set indexes to point to consecutive data blocks */
+        i=0;
+        do {
+            /* do this at least for trie->index[0] even if that block is only partly used for Latin-1 */
+            trie->index[i++]=j;
+            j+=UTRIE_DATA_BLOCK_LENGTH;
+        } while(i<(256>>UTRIE_SHIFT));
+    }
+
+    /* reset the initially allocated blocks to the initial value */
+    trie->dataLength=j;
+    while(j>0) {
+        trie->data[--j]=initialValue;
+    }
+
+    trie->leadUnitValue=leadUnitValue;
+    trie->indexLength=UTRIE_MAX_INDEX_LENGTH;
+    trie->dataCapacity=maxDataLength;
+    trie->isLatin1Linear=latin1Linear;
+    trie->isCompacted=false;
+    return trie;
+}
+
+U_CAPI UNewTrie * U_EXPORT2
+utrie_clone(UNewTrie *fillIn, const UNewTrie *other, uint32_t *aliasData, int32_t aliasDataCapacity) {
+    UNewTrie *trie;
+    UBool isDataAllocated;
+
+    /* do not clone if other is not valid or already compacted */
+    if(other==nullptr || other->data==nullptr || other->isCompacted) {
+        return nullptr;
+    }
+
+    /* clone data */
+    if(aliasData!=nullptr && aliasDataCapacity>=other->dataCapacity) {
+        isDataAllocated=false;
+    } else {
+        aliasDataCapacity=other->dataCapacity;
+        aliasData=(uint32_t *)uprv_malloc(other->dataCapacity*4);
+        if(aliasData==nullptr) {
+            return nullptr;
+        }
+        isDataAllocated=true;
+    }
+
+    trie=utrie_open(fillIn, aliasData, aliasDataCapacity,
+                    other->data[0], other->leadUnitValue,
+                    other->isLatin1Linear);
+    if(trie==nullptr) {
+        uprv_free(aliasData);
+    } else {
+        uprv_memcpy(trie->index, other->index, sizeof(trie->index));
+        uprv_memcpy(trie->data, other->data, (size_t)other->dataLength*4);
+        trie->dataLength=other->dataLength;
+        trie->isDataAllocated=isDataAllocated;
+    }
+
+    return trie;
+}
+
+U_CAPI void U_EXPORT2
+utrie_close(UNewTrie *trie) {
+    if(trie!=nullptr) {
+        if(trie->isDataAllocated) {
+            uprv_free(trie->data);
+            trie->data=nullptr;
+        }
+        if(trie->isAllocated) {
+            uprv_free(trie);
+        }
+    }
+}
+
+U_CAPI uint32_t * U_EXPORT2
+utrie_getData(UNewTrie *trie, int32_t *pLength) {
+    if(trie==nullptr || pLength==nullptr) {
+        return nullptr;
+    }
+
+    *pLength=trie->dataLength;
+    return trie->data;
+}
+
+static int32_t
+utrie_allocDataBlock(UNewTrie *trie) {
+    int32_t newBlock, newTop;
+
+    newBlock=trie->dataLength;
+    newTop=newBlock+UTRIE_DATA_BLOCK_LENGTH;
+    if(newTop>trie->dataCapacity) {
+        /* out of memory in the data array */
+        return -1;
+    }
+    trie->dataLength=newTop;
+    return newBlock;
+}
+
+/**
+ * No error checking for illegal arguments.
+ *
+ * @return -1 if no new data block available (out of memory in data array)
+ * @internal
+ */
+static int32_t
+utrie_getDataBlock(UNewTrie *trie, UChar32 c) {
+    int32_t indexValue, newBlock;
+
+    c>>=UTRIE_SHIFT;
+    indexValue=trie->index[c];
+    if(indexValue>0) {
+        return indexValue;
+    }
+
+    /* allocate a new data block */
+    newBlock=utrie_allocDataBlock(trie);
+    if(newBlock<0) {
+        /* out of memory in the data array */
+        return -1;
+    }
+    trie->index[c]=newBlock;
+
+    /* copy-on-write for a block from a setRange() */
+    uprv_memcpy(trie->data+newBlock, trie->data-indexValue, 4*UTRIE_DATA_BLOCK_LENGTH);
+    return newBlock;
+}
+
+/**
+ * @return true if the value was successfully set
+ */
+U_CAPI UBool U_EXPORT2
+utrie_set32(UNewTrie *trie, UChar32 c, uint32_t value) {
+    int32_t block;
+
+    /* valid, uncompacted trie and valid c? */
+    if(trie==nullptr || trie->isCompacted || (uint32_t)c>0x10ffff) {
+        return false;
+    }
+
+    block=utrie_getDataBlock(trie, c);
+    if(block<0) {
+        return false;
+    }
+
+    trie->data[block+(c&UTRIE_MASK)]=value;
+    return true;
+}
+
+U_CAPI uint32_t U_EXPORT2
+utrie_get32(UNewTrie *trie, UChar32 c, UBool *pInBlockZero) {
+    int32_t block;
+
+    /* valid, uncompacted trie and valid c? */
+    if(trie==nullptr || trie->isCompacted || (uint32_t)c>0x10ffff) {
+        if(pInBlockZero!=nullptr) {
+            *pInBlockZero=true;
+        }
+        return 0;
+    }
+
+    block=trie->index[c>>UTRIE_SHIFT];
+    if(pInBlockZero!=nullptr) {
+        *pInBlockZero= (UBool)(block==0);
+    }
+
+    return trie->data[ABS(block)+(c&UTRIE_MASK)];
+}
+
+/**
+ * @internal
+ */
+static void
+utrie_fillBlock(uint32_t *block, UChar32 start, UChar32 limit,
+                uint32_t value, uint32_t initialValue, UBool overwrite) {
+    uint32_t *pLimit;
+
+    pLimit=block+limit;
+    block+=start;
+    if(overwrite) {
+        while(blockisCompacted ||
+        (uint32_t)start>0x10ffff || (uint32_t)limit>0x110000 || start>limit
+    ) {
+        return false;
+    }
+    if(start==limit) {
+        return true; /* nothing to do */
+    }
+
+    initialValue=trie->data[0];
+    if(start&UTRIE_MASK) {
+        UChar32 nextStart;
+
+        /* set partial block at [start..following block boundary[ */
+        block=utrie_getDataBlock(trie, start);
+        if(block<0) {
+            return false;
+        }
+
+        nextStart=(start+UTRIE_DATA_BLOCK_LENGTH)&~UTRIE_MASK;
+        if(nextStart<=limit) {
+            utrie_fillBlock(trie->data+block, start&UTRIE_MASK, UTRIE_DATA_BLOCK_LENGTH,
+                            value, initialValue, overwrite);
+            start=nextStart;
+        } else {
+            utrie_fillBlock(trie->data+block, start&UTRIE_MASK, limit&UTRIE_MASK,
+                            value, initialValue, overwrite);
+            return true;
+        }
+    }
+
+    /* number of positions in the last, partial block */
+    rest=limit&UTRIE_MASK;
+
+    /* round down limit to a block boundary */
+    limit&=~UTRIE_MASK;
+
+    /* iterate over all-value blocks */
+    if(value==initialValue) {
+        repeatBlock=0;
+    } else {
+        repeatBlock=-1;
+    }
+    while(startindex[start>>UTRIE_SHIFT];
+        if(block>0) {
+            /* already allocated, fill in value */
+            utrie_fillBlock(trie->data+block, 0, UTRIE_DATA_BLOCK_LENGTH, value, initialValue, overwrite);
+        } else if(trie->data[-block]!=value && (block==0 || overwrite)) {
+            /* set the repeatBlock instead of the current block 0 or range block */
+            if(repeatBlock>=0) {
+                trie->index[start>>UTRIE_SHIFT]=-repeatBlock;
+            } else {
+                /* create and set and fill the repeatBlock */
+                repeatBlock=utrie_getDataBlock(trie, start);
+                if(repeatBlock<0) {
+                    return false;
+                }
+
+                /* set the negative block number to indicate that it is a repeat block */
+                trie->index[start>>UTRIE_SHIFT]=-repeatBlock;
+                utrie_fillBlock(trie->data+repeatBlock, 0, UTRIE_DATA_BLOCK_LENGTH, value, initialValue, true);
+            }
+        }
+
+        start+=UTRIE_DATA_BLOCK_LENGTH;
+    }
+
+    if(rest>0) {
+        /* set partial block at [last block boundary..limit[ */
+        block=utrie_getDataBlock(trie, start);
+        if(block<0) {
+            return false;
+        }
+
+        utrie_fillBlock(trie->data+block, 0, rest, value, initialValue, overwrite);
+    }
+
+    return true;
+}
+
+static int32_t
+_findSameIndexBlock(const int32_t *idx, int32_t indexLength,
+                    int32_t otherBlock) {
+    int32_t block, i;
+
+    for(block=UTRIE_BMP_INDEX_LENGTH; blockindex;
+
+    /* copy the lead surrogate indexes into a temporary array */
+    uprv_memcpy(leadIndexes, idx+(0xd800>>UTRIE_SHIFT), 4*UTRIE_SURROGATE_BLOCK_COUNT);
+
+    /*
+     * set all values for lead surrogate code *units* to leadUnitValue
+     * so that, by default, runtime lookups will find no data for associated
+     * supplementary code points, unless there is data for such code points
+     * which will result in a non-zero folding value below that is set for
+     * the respective lead units
+     *
+     * the above saved the indexes for surrogate code *points*
+     * fill the indexes with simplified code from utrie_setRange32()
+     */
+    if(trie->leadUnitValue==trie->data[0]) {
+        block=0; /* leadUnitValue==initialValue, use all-initial-value block */
+    } else {
+        /* create and fill the repeatBlock */
+        block=utrie_allocDataBlock(trie);
+        if(block<0) {
+            /* data table overflow */
+            *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+            return;
+        }
+        utrie_fillBlock(trie->data+block, 0, UTRIE_DATA_BLOCK_LENGTH, trie->leadUnitValue, trie->data[0], true);
+        block=-block; /* negative block number to indicate that it is a repeat block */
+    }
+    for(c=(0xd800>>UTRIE_SHIFT); c<(0xdc00>>UTRIE_SHIFT); ++c) {
+        trie->index[c]=block;
+    }
+
+    /*
+     * Fold significant index values into the area just after the BMP indexes.
+     * In case the first lead surrogate has significant data,
+     * its index block must be used first (in which case the folding is a no-op).
+     * Later all folded index blocks are moved up one to insert the copied
+     * lead surrogate indexes.
+     */
+    indexLength=UTRIE_BMP_INDEX_LENGTH;
+
+    /* search for any index (stage 1) entries for supplementary code points */
+    for(c=0x10000; c<0x110000;) {
+        if(idx[c>>UTRIE_SHIFT]!=0) {
+            /* there is data, treat the full block for a lead surrogate */
+            c&=~0x3ff;
+
+#ifdef UTRIE_DEBUG
+            ++countLeadCUWithData;
+            /* printf("supplementary data for lead surrogate U+%04lx\n", (long)(0xd7c0+(c>>10))); */
+#endif
+
+            /* is there an identical index block? */
+            block=_findSameIndexBlock(idx, indexLength, c>>UTRIE_SHIFT);
+
+            /*
+             * get a folded value for [c..c+0x400[ and,
+             * if different from the value for the lead surrogate code point,
+             * set it for the lead surrogate code unit
+             */
+            value=getFoldedValue(trie, c, block+UTRIE_SURROGATE_BLOCK_COUNT);
+            if(value!=utrie_get32(trie, U16_LEAD(c), nullptr)) {
+                if(!utrie_set32(trie, U16_LEAD(c), value)) {
+                    /* data table overflow */
+                    *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+                    return;
+                }
+
+                /* if we did not find an identical index block... */
+                if(block==indexLength) {
+                    /* move the actual index (stage 1) entries from the supplementary position to the new one */
+                    uprv_memmove(idx+indexLength,
+                                 idx+(c>>UTRIE_SHIFT),
+                                 4*UTRIE_SURROGATE_BLOCK_COUNT);
+                    indexLength+=UTRIE_SURROGATE_BLOCK_COUNT;
+                }
+            }
+            c+=0x400;
+        } else {
+            c+=UTRIE_DATA_BLOCK_LENGTH;
+        }
+    }
+#ifdef UTRIE_DEBUG
+    if(countLeadCUWithData>0) {
+        printf("supplementary data for %d lead surrogates\n", countLeadCUWithData);
+    }
+#endif
+
+    /*
+     * index array overflow?
+     * This is to guarantee that a folding offset is of the form
+     * UTRIE_BMP_INDEX_LENGTH+n*UTRIE_SURROGATE_BLOCK_COUNT with n=0..1023.
+     * If the index is too large, then n>=1024 and more than 10 bits are necessary.
+     *
+     * In fact, it can only ever become n==1024 with completely unfoldable data and
+     * the additional block of duplicated values for lead surrogates.
+     */
+    if(indexLength>=UTRIE_MAX_INDEX_LENGTH) {
+        *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+        return;
+    }
+
+    /*
+     * make space for the lead surrogate index block and
+     * insert it between the BMP indexes and the folded ones
+     */
+    uprv_memmove(idx+UTRIE_BMP_INDEX_LENGTH+UTRIE_SURROGATE_BLOCK_COUNT,
+                 idx+UTRIE_BMP_INDEX_LENGTH,
+                 4*(indexLength-UTRIE_BMP_INDEX_LENGTH));
+    uprv_memcpy(idx+UTRIE_BMP_INDEX_LENGTH,
+                leadIndexes,
+                4*UTRIE_SURROGATE_BLOCK_COUNT);
+    indexLength+=UTRIE_SURROGATE_BLOCK_COUNT;
+
+#ifdef UTRIE_DEBUG
+    printf("trie index count: BMP %ld  all Unicode %ld  folded %ld\n",
+           UTRIE_BMP_INDEX_LENGTH, (long)UTRIE_MAX_INDEX_LENGTH, indexLength);
+#endif
+
+    trie->indexLength=indexLength;
+}
+
+/*
+ * Set a value in the trie index map to indicate which data block
+ * is referenced and which one is not.
+ * utrie_compact() will remove data blocks that are not used at all.
+ * Set
+ * - 0 if it is used
+ * - -1 if it is not used
+ */
+static void
+_findUnusedBlocks(UNewTrie *trie) {
+    int32_t i;
+
+    /* fill the entire map with "not used" */
+    uprv_memset(trie->map, 0xff, (UTRIE_MAX_BUILD_TIME_DATA_LENGTH>>UTRIE_SHIFT)*4);
+
+    /* mark each block that _is_ used with 0 */
+    for(i=0; iindexLength; ++i) {
+        trie->map[ABS(trie->index[i])>>UTRIE_SHIFT]=0;
+    }
+
+    /* never move the all-initial-value block 0 */
+    trie->map[0]=0;
+}
+
+static int32_t
+_findSameDataBlock(const uint32_t *data, int32_t dataLength,
+                   int32_t otherBlock, int32_t step) {
+    int32_t block;
+
+    /* ensure that we do not even partially get past dataLength */
+    dataLength-=UTRIE_DATA_BLOCK_LENGTH;
+
+    for(block=0; block<=dataLength; block+=step) {
+        if(equal_uint32(data+block, data+otherBlock, UTRIE_DATA_BLOCK_LENGTH)) {
+            return block;
+        }
+    }
+    return -1;
+}
+
+/*
+ * Compact a folded build-time trie.
+ *
+ * The compaction
+ * - removes blocks that are identical with earlier ones
+ * - overlaps adjacent blocks as much as possible (if overlap==true)
+ * - moves blocks in steps of the data granularity
+ * - moves and overlaps blocks that overlap with multiple values in the overlap region
+ *
+ * It does not
+ * - try to move and overlap blocks that are not already adjacent
+ */
+static void
+utrie_compact(UNewTrie *trie, UBool overlap, UErrorCode *pErrorCode) {
+    int32_t i, start, newStart, overlapStart;
+
+    if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) {
+        return;
+    }
+
+    /* valid, uncompacted trie? */
+    if(trie==nullptr) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+    if(trie->isCompacted) {
+        return; /* nothing left to do */
+    }
+
+    /* compaction */
+
+    /* initialize the index map with "block is used/unused" flags */
+    _findUnusedBlocks(trie);
+
+    /* if Latin-1 is preallocated and linear, then do not compact Latin-1 data */
+    if(trie->isLatin1Linear && UTRIE_SHIFT<=8) {
+        overlapStart=UTRIE_DATA_BLOCK_LENGTH+256;
+    } else {
+        overlapStart=UTRIE_DATA_BLOCK_LENGTH;
+    }
+
+    newStart=UTRIE_DATA_BLOCK_LENGTH;
+    for(start=newStart; startdataLength;) {
+        /*
+         * start: index of first entry of current block
+         * newStart: index where the current block is to be moved
+         *           (right after current end of already-compacted data)
+         */
+
+        /* skip blocks that are not used */
+        if(trie->map[start>>UTRIE_SHIFT]<0) {
+            /* advance start to the next block */
+            start+=UTRIE_DATA_BLOCK_LENGTH;
+
+            /* leave newStart with the previous block! */
+            continue;
+        }
+
+        /* search for an identical block */
+        if( start>=overlapStart &&
+            (i=_findSameDataBlock(trie->data, newStart, start,
+                            overlap ? UTRIE_DATA_GRANULARITY : UTRIE_DATA_BLOCK_LENGTH))
+             >=0
+        ) {
+            /* found an identical block, set the other block's index value for the current block */
+            trie->map[start>>UTRIE_SHIFT]=i;
+
+            /* advance start to the next block */
+            start+=UTRIE_DATA_BLOCK_LENGTH;
+
+            /* leave newStart with the previous block! */
+            continue;
+        }
+
+        /* see if the beginning of this block can be overlapped with the end of the previous block */
+        if(overlap && start>=overlapStart) {
+            /* look for maximum overlap (modulo granularity) with the previous, adjacent block */
+            for(i=UTRIE_DATA_BLOCK_LENGTH-UTRIE_DATA_GRANULARITY;
+                i>0 && !equal_uint32(trie->data+(newStart-i), trie->data+start, i);
+                i-=UTRIE_DATA_GRANULARITY) {}
+        } else {
+            i=0;
+        }
+
+        if(i>0) {
+            /* some overlap */
+            trie->map[start>>UTRIE_SHIFT]=newStart-i;
+
+            /* move the non-overlapping indexes to their new positions */
+            start+=i;
+            for(i=UTRIE_DATA_BLOCK_LENGTH-i; i>0; --i) {
+                trie->data[newStart++]=trie->data[start++];
+            }
+        } else if(newStartmap[start>>UTRIE_SHIFT]=newStart;
+            for(i=UTRIE_DATA_BLOCK_LENGTH; i>0; --i) {
+                trie->data[newStart++]=trie->data[start++];
+            }
+        } else /* no overlap && newStart==start */ {
+            trie->map[start>>UTRIE_SHIFT]=start;
+            newStart+=UTRIE_DATA_BLOCK_LENGTH;
+            start=newStart;
+        }
+    }
+
+    /* now adjust the index (stage 1) table */
+    for(i=0; iindexLength; ++i) {
+        trie->index[i]=trie->map[ABS(trie->index[i])>>UTRIE_SHIFT];
+    }
+
+#ifdef UTRIE_DEBUG
+    /* we saved some space */
+    printf("compacting trie: count of 32-bit words %lu->%lu\n",
+            (long)trie->dataLength, (long)newStart);
+#endif
+
+    trie->dataLength=newStart;
+}
+
+/* serialization ------------------------------------------------------------ */
+
+/*
+ * Default function for the folding value:
+ * Just store the offset (16 bits) if there is any non-initial-value entry.
+ *
+ * The offset parameter is never 0.
+ * Returning the offset itself is safe for UTRIE_SHIFT>=5 because
+ * for UTRIE_SHIFT==5 the maximum index length is UTRIE_MAX_INDEX_LENGTH==0x8800
+ * which fits into 16-bit trie values;
+ * for higher UTRIE_SHIFT, UTRIE_MAX_INDEX_LENGTH decreases.
+ *
+ * Theoretically, it would be safer for all possible UTRIE_SHIFT including
+ * those of 4 and lower to return offset>>UTRIE_SURROGATE_BLOCK_BITS
+ * which would always result in a value of 0x40..0x43f
+ * (start/end 1k blocks of supplementary Unicode code points).
+ * However, this would be uglier, and would not work for some existing
+ * binary data file formats.
+ *
+ * Also, we do not plan to change UTRIE_SHIFT because it would change binary
+ * data file formats, and we would probably not make it smaller because of
+ * the then even larger BMP index length even for empty tries.
+ */
+static uint32_t U_CALLCONV
+defaultGetFoldedValue(UNewTrie *trie, UChar32 start, int32_t offset) {
+    uint32_t value, initialValue;
+    UChar32 limit;
+    UBool inBlockZero;
+
+    initialValue=trie->data[0];
+    limit=start+0x400;
+    while(start0 && dt==nullptr)) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    if(getFoldedValue==nullptr) {
+        getFoldedValue=defaultGetFoldedValue;
+    }
+
+    data = (uint8_t*)dt;
+    /* fold and compact if necessary, also checks that indexLength is within limits */
+    if(!trie->isCompacted) {
+        /* compact once without overlap to improve folding */
+        utrie_compact(trie, false, pErrorCode);
+
+        /* fold the supplementary part of the index array */
+        utrie_fold(trie, getFoldedValue, pErrorCode);
+
+        /* compact again with overlap for minimum data array length */
+        utrie_compact(trie, true, pErrorCode);
+
+        trie->isCompacted=true;
+        if(U_FAILURE(*pErrorCode)) {
+            return 0;
+        }
+    }
+
+    /* is dataLength within limits? */
+    if( (reduceTo16Bits ? (trie->dataLength+trie->indexLength) : trie->dataLength) >= UTRIE_MAX_DATA_LENGTH) {
+        *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+    }
+
+    length=sizeof(UTrieHeader)+2*trie->indexLength;
+    if(reduceTo16Bits) {
+        length+=2*trie->dataLength;
+    } else {
+        length+=4*trie->dataLength;
+    }
+
+    if(length>capacity) {
+        return length; /* preflighting */
+    }
+
+#ifdef UTRIE_DEBUG
+    printf("**UTrieLengths(serialize)** index:%6ld  data:%6ld  serialized:%6ld\n",
+           (long)trie->indexLength, (long)trie->dataLength, (long)length);
+#endif
+
+    /* set the header fields */
+    header=(UTrieHeader *)data;
+    data+=sizeof(UTrieHeader);
+
+    header->signature=0x54726965; /* "Trie" */
+    header->options=UTRIE_SHIFT | (UTRIE_INDEX_SHIFT<options|=UTRIE_OPTIONS_DATA_IS_32_BIT;
+    }
+    if(trie->isLatin1Linear) {
+        header->options|=UTRIE_OPTIONS_LATIN1_IS_LINEAR;
+    }
+
+    header->indexLength=trie->indexLength;
+    header->dataLength=trie->dataLength;
+
+    /* write the index (stage 1) array and the 16/32-bit data (stage 2) array */
+    if(reduceTo16Bits) {
+        /* write 16-bit index values shifted right by UTRIE_INDEX_SHIFT, after adding indexLength */
+        p=(uint32_t *)trie->index;
+        dest16=(uint16_t *)data;
+        for(i=trie->indexLength; i>0; --i) {
+            *dest16++=(uint16_t)((*p++ + trie->indexLength)>>UTRIE_INDEX_SHIFT);
+        }
+
+        /* write 16-bit data values */
+        p=trie->data;
+        for(i=trie->dataLength; i>0; --i) {
+            *dest16++=(uint16_t)*p++;
+        }
+    } else {
+        /* write 16-bit index values shifted right by UTRIE_INDEX_SHIFT */
+        p=(uint32_t *)trie->index;
+        dest16=(uint16_t *)data;
+        for(i=trie->indexLength; i>0; --i) {
+            *dest16++=(uint16_t)(*p++ >> UTRIE_INDEX_SHIFT);
+        }
+
+        /* write 32-bit data values */
+        uprv_memcpy(dest16, trie->data, 4*(size_t)trie->dataLength);
+    }
+
+    return length;
+}
+
+/* inverse to defaultGetFoldedValue() */
+U_CAPI int32_t U_EXPORT2
+utrie_defaultGetFoldingOffset(uint32_t data) {
+    return (int32_t)data;
+}
+
+U_CAPI int32_t U_EXPORT2
+utrie_unserialize(UTrie *trie, const void *data, int32_t length, UErrorCode *pErrorCode) {
+    const UTrieHeader *header;
+    const uint16_t *p16;
+    uint32_t options;
+
+    if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) {
+        return -1;
+    }
+
+    /* enough data for a trie header? */
+    if(length<(int32_t)sizeof(UTrieHeader)) {
+        *pErrorCode=U_INVALID_FORMAT_ERROR;
+        return -1;
+    }
+
+    /* check the signature */
+    header=(const UTrieHeader *)data;
+    if(header->signature!=0x54726965) {
+        *pErrorCode=U_INVALID_FORMAT_ERROR;
+        return -1;
+    }
+
+    /* get the options and check the shift values */
+    options=header->options;
+    if( (options&UTRIE_OPTIONS_SHIFT_MASK)!=UTRIE_SHIFT ||
+        ((options>>UTRIE_OPTIONS_INDEX_SHIFT)&UTRIE_OPTIONS_SHIFT_MASK)!=UTRIE_INDEX_SHIFT
+    ) {
+        *pErrorCode=U_INVALID_FORMAT_ERROR;
+        return -1;
+    }
+    trie->isLatin1Linear= (UBool)((options&UTRIE_OPTIONS_LATIN1_IS_LINEAR)!=0);
+
+    /* get the length values */
+    trie->indexLength=header->indexLength;
+    trie->dataLength=header->dataLength;
+
+    length-=(int32_t)sizeof(UTrieHeader);
+
+    /* enough data for the index? */
+    if(length<2*trie->indexLength) {
+        *pErrorCode=U_INVALID_FORMAT_ERROR;
+        return -1;
+    }
+    p16=(const uint16_t *)(header+1);
+    trie->index=p16;
+    p16+=trie->indexLength;
+    length-=2*trie->indexLength;
+
+    /* get the data */
+    if(options&UTRIE_OPTIONS_DATA_IS_32_BIT) {
+        if(length<4*trie->dataLength) {
+            *pErrorCode=U_INVALID_FORMAT_ERROR;
+            return -1;
+        }
+        trie->data32=(const uint32_t *)p16;
+        trie->initialValue=trie->data32[0];
+        length=(int32_t)sizeof(UTrieHeader)+2*trie->indexLength+4*trie->dataLength;
+    } else {
+        if(length<2*trie->dataLength) {
+            *pErrorCode=U_INVALID_FORMAT_ERROR;
+            return -1;
+        }
+
+        /* the "data16" data is used via the index pointer */
+        trie->data32=nullptr;
+        trie->initialValue=trie->index[trie->indexLength];
+        length=(int32_t)sizeof(UTrieHeader)+2*trie->indexLength+2*trie->dataLength;
+    }
+
+    trie->getFoldingOffset=utrie_defaultGetFoldingOffset;
+
+    return length;
+}
+
+U_CAPI int32_t U_EXPORT2
+utrie_unserializeDummy(UTrie *trie,
+                       void *data, int32_t length,
+                       uint32_t initialValue, uint32_t leadUnitValue,
+                       UBool make16BitTrie,
+                       UErrorCode *pErrorCode) {
+    uint16_t *p16;
+    int32_t actualLength, latin1Length, i, limit;
+    uint16_t block;
+
+    if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) {
+        return -1;
+    }
+
+    /* calculate the actual size of the dummy trie data */
+
+    /* max(Latin-1, block 0) */
+    latin1Length= 256; /*UTRIE_SHIFT<=8 ? 256 : UTRIE_DATA_BLOCK_LENGTH;*/
+
+    trie->indexLength=UTRIE_BMP_INDEX_LENGTH+UTRIE_SURROGATE_BLOCK_COUNT;
+    trie->dataLength=latin1Length;
+    if(leadUnitValue!=initialValue) {
+        trie->dataLength+=UTRIE_DATA_BLOCK_LENGTH;
+    }
+
+    actualLength=trie->indexLength*2;
+    if(make16BitTrie) {
+        actualLength+=trie->dataLength*2;
+    } else {
+        actualLength+=trie->dataLength*4;
+    }
+
+    /* enough space for the dummy trie? */
+    if(lengthisLatin1Linear=true;
+    trie->initialValue=initialValue;
+
+    /* fill the index and data arrays */
+    p16=(uint16_t *)data;
+    trie->index=p16;
+
+    if(make16BitTrie) {
+        /* indexes to block 0 */
+        block=(uint16_t)(trie->indexLength>>UTRIE_INDEX_SHIFT);
+        limit=trie->indexLength;
+        for(i=0; i>UTRIE_INDEX_SHIFT);
+            i=0xd800>>UTRIE_SHIFT;
+            limit=0xdc00>>UTRIE_SHIFT;
+            for(; idata32=nullptr;
+
+        /* Latin-1 data */
+        p16+=trie->indexLength;
+        for(i=0; iindexLength*2);
+
+        if(leadUnitValue!=initialValue) {
+            /* indexes for lead surrogate code units to the block after Latin-1 */
+            block=(uint16_t)(latin1Length>>UTRIE_INDEX_SHIFT);
+            i=0xd800>>UTRIE_SHIFT;
+            limit=0xdc00>>UTRIE_SHIFT;
+            for(; idata32=p32=(uint32_t *)(p16+trie->indexLength);
+
+        /* Latin-1 data */
+        for(i=0; igetFoldingOffset=utrie_defaultGetFoldingOffset;
+
+    return actualLength;
+}
+
+/* enumeration -------------------------------------------------------------- */
+
+/* default UTrieEnumValue() returns the input value itself */
+static uint32_t U_CALLCONV
+enumSameValue(const void * /*context*/, uint32_t value) {
+    return value;
+}
+
+/**
+ * Enumerate all ranges of code points with the same relevant values.
+ * The values are transformed from the raw trie entries by the enumValue function.
+ */
+U_CAPI void U_EXPORT2
+utrie_enum(const UTrie *trie,
+           UTrieEnumValue *enumValue, UTrieEnumRange *enumRange, const void *context) {
+    const uint32_t *data32;
+    const uint16_t *idx;
+
+    uint32_t value, prevValue, initialValue;
+    UChar32 c, prev;
+    int32_t l, i, j, block, prevBlock, nullBlock, offset;
+
+    /* check arguments */
+    if(trie==nullptr || trie->index==nullptr || enumRange==nullptr) {
+        return;
+    }
+    if(enumValue==nullptr) {
+        enumValue=enumSameValue;
+    }
+
+    idx=trie->index;
+    data32=trie->data32;
+
+    /* get the enumeration value that corresponds to an initial-value trie data entry */
+    initialValue=enumValue(context, trie->initialValue);
+
+    if(data32==nullptr) {
+        nullBlock=trie->indexLength;
+    } else {
+        nullBlock=0;
+    }
+
+    /* set variables for previous range */
+    prevBlock=nullBlock;
+    prev=0;
+    prevValue=initialValue;
+
+    /* enumerate BMP - the main loop enumerates data blocks */
+    for(i=0, c=0; c<=0xffff; ++i) {
+        if(c==0xd800) {
+            /* skip lead surrogate code _units_, go to lead surr. code _points_ */
+            i=UTRIE_BMP_INDEX_LENGTH;
+        } else if(c==0xdc00) {
+            /* go back to regular BMP code points */
+            i=c>>UTRIE_SHIFT;
+        }
+
+        block=idx[i]<0) {
+                        /* the block is not filled with all the same value */
+                        prevBlock=-1;
+                    }
+                    prev=c;
+                    prevValue=value;
+                }
+                ++c;
+            }
+        }
+    }
+
+    /* enumerate supplementary code points */
+    for(l=0xd800; l<0xdc00;) {
+        /* lead surrogate access */
+        offset=idx[l>>UTRIE_SHIFT]<getFoldingOffset(value);
+        if(offset<=0) {
+            /* no data for this lead surrogate */
+            if(prevValue!=initialValue) {
+                if(prev0) {
+                                /* the block is not filled with all the same value */
+                                prevBlock=-1;
+                            }
+                            prev=c;
+                            prevValue=value;
+                        }
+                        ++c;
+                    }
+                }
+            } while(++i>UTRIE_SHIFT,
+
+    /**
+     * Shift size for shifting left the index array values.
+     * Increases possible data size with 16-bit index values at the cost
+     * of compactability.
+     * This requires blocks of stage 2 data to be aligned by UTRIE_DATA_GRANULARITY.
+     * 0..UTRIE_SHIFT
+     */
+    UTRIE_INDEX_SHIFT=2,
+
+    /** The alignment size of a stage 2 data block. Also the granularity for compaction. */
+    UTRIE_DATA_GRANULARITY=1<>UTRIE_SHIFT
+     */
+    UTRIE_SURROGATE_BLOCK_COUNT=(1<>UTRIE_SHIFT
+};
+
+/**
+ * Length of the index (stage 1) array before folding.
+ * Maximum number of Unicode code points (0x110000) shifted right by UTRIE_SHIFT.
+ */
+#define UTRIE_MAX_INDEX_LENGTH (0x110000>>UTRIE_SHIFT)
+
+/**
+ * Maximum length of the runtime data (stage 2) array.
+ * Limited by 16-bit index values that are left-shifted by UTRIE_INDEX_SHIFT.
+ */
+#define UTRIE_MAX_DATA_LENGTH (0x10000<=UTRIE_BMP_INDEX_LENGTH, or 0 if there is no data for the lead surrogate
+ */
+typedef int32_t U_CALLCONV
+UTrieGetFoldingOffset(uint32_t data);
+
+/**
+ * Run-time Trie structure.
+ *
+ * Either the data table is 16 bits wide and accessed via the index
+ * pointer, with each index item increased by indexLength;
+ * in this case, data32==NULL.
+ *
+ * Or the data table is 32 bits wide and accessed via the data32 pointer.
+ */
+struct UTrie {
+    const uint16_t *index;
+    const uint32_t *data32; /* NULL if 16b data is used via index */
+
+    /**
+     * This function is not used in _FROM_LEAD, _FROM_BMP, and _FROM_OFFSET_TRAIL macros.
+     * If convenience macros like _GET16 or _NEXT32 are used, this function must be set.
+     *
+     * utrie_unserialize() sets a default function which simply returns
+     * the lead surrogate's value itself - which is the inverse of the default
+     * folding function used by utrie_serialize().
+     *
+     * @see UTrieGetFoldingOffset
+     */
+    UTrieGetFoldingOffset *getFoldingOffset;
+
+    int32_t indexLength, dataLength;
+    uint32_t initialValue;
+    UBool isLatin1Linear;
+};
+
+#ifndef __UTRIE2_H__
+typedef struct UTrie UTrie;
+#endif
+
+/** Internal trie getter from an offset (0 if c16 is a BMP/lead units) and a 16-bit unit */
+#define _UTRIE_GET_RAW(trie, data, offset, c16) \
+    (trie)->data[ \
+        ((int32_t)((trie)->index[(offset)+((c16)>>UTRIE_SHIFT)])<getFoldingOffset(result); \
+\
+    /* get the real data from the folded lead/trail units */ \
+    if(__offset>0) { \
+        (result)=_UTRIE_GET_RAW((trie), data, __offset, (c2)&0x3ff); \
+    } else { \
+        (result)=(resultType)((trie)->initialValue); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** Internal trie getter from a BMP code point, treating a lead surrogate as a normal code point */
+#define _UTRIE_GET_FROM_BMP(trie, data, c16) \
+    _UTRIE_GET_RAW(trie, data, 0xd800<=(c16) && (c16)<=0xdbff ? UTRIE_LEAD_INDEX_DISP : 0, c16)
+
+/**
+ * Internal trie getter from a code point.
+ * Could be faster(?) but longer with
+ *   if((c32)<=0xd7ff) { (result)=_UTRIE_GET_RAW(trie, data, 0, c32); }
+ */
+#define _UTRIE_GET(trie, data, c32, result, resultType) UPRV_BLOCK_MACRO_BEGIN { \
+    if((uint32_t)(c32)<=0xffff) { \
+        /* BMP code points */ \
+        (result)=_UTRIE_GET_FROM_BMP(trie, data, c32); \
+    } else if((uint32_t)(c32)<=0x10ffff) { \
+        /* supplementary code point */ \
+        UChar __lead16=U16_LEAD(c32); \
+        _UTRIE_GET_FROM_PAIR(trie, data, __lead16, c32, result, resultType); \
+    } else { \
+        /* out of range */ \
+        (result)=(resultType)((trie)->initialValue); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** Internal next-post-increment: get the next code point (c, c2) and its data */
+#define _UTRIE_NEXT(trie, data, src, limit, c, c2, result, resultType) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=*(src)++; \
+    if(!U16_IS_LEAD(c)) { \
+        (c2)=0; \
+        (result)=_UTRIE_GET_RAW((trie), data, 0, (c)); \
+    } else if((src)!=(limit) && U16_IS_TRAIL((c2)=*(src))) { \
+        ++(src); \
+        _UTRIE_GET_FROM_PAIR((trie), data, (c), (c2), (result), resultType); \
+    } else { \
+        /* unpaired lead surrogate code point */ \
+        (c2)=0; \
+        (result)=_UTRIE_GET_RAW((trie), data, UTRIE_LEAD_INDEX_DISP, (c)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** Internal previous: get the previous code point (c, c2) and its data */
+#define _UTRIE_PREVIOUS(trie, data, start, src, c, c2, result, resultType) UPRV_BLOCK_MACRO_BEGIN { \
+    (c)=*--(src); \
+    if(!U16_IS_SURROGATE(c)) { \
+        (c2)=0; \
+        (result)=_UTRIE_GET_RAW((trie), data, 0, (c)); \
+    } else if(!U16_IS_SURROGATE_LEAD(c)) { \
+        /* trail surrogate */ \
+        if((start)!=(src) && U16_IS_LEAD((c2)=*((src)-1))) { \
+            --(src); \
+            (result)=(c); (c)=(c2); (c2)=(UChar)(result); /* swap c, c2 */ \
+            _UTRIE_GET_FROM_PAIR((trie), data, (c), (c2), (result), resultType); \
+        } else { \
+            /* unpaired trail surrogate code point */ \
+            (c2)=0; \
+            (result)=_UTRIE_GET_RAW((trie), data, 0, (c)); \
+        } \
+    } else { \
+        /* unpaired lead surrogate code point */ \
+        (c2)=0; \
+        (result)=_UTRIE_GET_RAW((trie), data, UTRIE_LEAD_INDEX_DISP, (c)); \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/* Public UTrie API ---------------------------------------------------------*/
+
+/**
+ * Get a pointer to the contiguous part of the data array
+ * for the Latin-1 range (U+0000..U+00ff).
+ * Must be used only if the Latin-1 range is in fact linear
+ * (trie->isLatin1Linear).
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @return (const uint16_t *) pointer to values for Latin-1 code points
+ */
+#define UTRIE_GET16_LATIN1(trie) ((trie)->index+(trie)->indexLength+UTRIE_DATA_BLOCK_LENGTH)
+
+/**
+ * Get a pointer to the contiguous part of the data array
+ * for the Latin-1 range (U+0000..U+00ff).
+ * Must be used only if the Latin-1 range is in fact linear
+ * (trie->isLatin1Linear).
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @return (const uint32_t *) pointer to values for Latin-1 code points
+ */
+#define UTRIE_GET32_LATIN1(trie) ((trie)->data32+UTRIE_DATA_BLOCK_LENGTH)
+
+/**
+ * Get a 16-bit trie value from a BMP code point (UChar, <=U+ffff).
+ * c16 may be a lead surrogate, which may have a value including a folding offset.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param c16 (UChar, in) the input BMP code point
+ * @return (uint16_t) trie lookup result
+ */
+#define UTRIE_GET16_FROM_LEAD(trie, c16) _UTRIE_GET_RAW(trie, index, 0, c16)
+
+/**
+ * Get a 32-bit trie value from a BMP code point (UChar, <=U+ffff).
+ * c16 may be a lead surrogate, which may have a value including a folding offset.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param c16 (UChar, in) the input BMP code point
+ * @return (uint32_t) trie lookup result
+ */
+#define UTRIE_GET32_FROM_LEAD(trie, c16) _UTRIE_GET_RAW(trie, data32, 0, c16)
+
+/**
+ * Get a 16-bit trie value from a BMP code point (UChar, <=U+ffff).
+ * Even lead surrogate code points are treated as normal code points,
+ * with unfolded values that may differ from _FROM_LEAD() macro results for them.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param c16 (UChar, in) the input BMP code point
+ * @return (uint16_t) trie lookup result
+ */
+#define UTRIE_GET16_FROM_BMP(trie, c16) _UTRIE_GET_FROM_BMP(trie, index, c16)
+
+/**
+ * Get a 32-bit trie value from a BMP code point (UChar, <=U+ffff).
+ * Even lead surrogate code points are treated as normal code points,
+ * with unfolded values that may differ from _FROM_LEAD() macro results for them.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param c16 (UChar, in) the input BMP code point
+ * @return (uint32_t) trie lookup result
+ */
+#define UTRIE_GET32_FROM_BMP(trie, c16) _UTRIE_GET_FROM_BMP(trie, data32, c16)
+
+/**
+ * Get a 16-bit trie value from a code point.
+ * Even lead surrogate code points are treated as normal code points,
+ * with unfolded values that may differ from _FROM_LEAD() macro results for them.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param c32 (UChar32, in) the input code point
+ * @param result (uint16_t, out) uint16_t variable for the trie lookup result
+ */
+#define UTRIE_GET16(trie, c32, result) _UTRIE_GET(trie, index, c32, result, uint16_t)
+
+/**
+ * Get a 32-bit trie value from a code point.
+ * Even lead surrogate code points are treated as normal code points,
+ * with unfolded values that may differ from _FROM_LEAD() macro results for them.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param c32 (UChar32, in) the input code point
+ * @param result (uint32_t, out) uint32_t variable for the trie lookup result
+ */
+#define UTRIE_GET32(trie, c32, result) _UTRIE_GET(trie, data32, c32, result, uint32_t)
+
+/**
+ * Get the next code point (c, c2), post-increment src,
+ * and get a 16-bit value from the trie.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param src (const UChar *, in/out) the source text pointer
+ * @param limit (const UChar *, in) the limit pointer for the text, or NULL
+ * @param c (UChar, out) variable for the BMP or lead code unit
+ * @param c2 (UChar, out) variable for 0 or the trail code unit
+ * @param result (uint16_t, out) uint16_t variable for the trie lookup result
+ */
+#define UTRIE_NEXT16(trie, src, limit, c, c2, result) _UTRIE_NEXT(trie, index, src, limit, c, c2, result, uint16_t)
+
+/**
+ * Get the next code point (c, c2), post-increment src,
+ * and get a 32-bit value from the trie.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param src (const UChar *, in/out) the source text pointer
+ * @param limit (const UChar *, in) the limit pointer for the text, or NULL
+ * @param c (UChar, out) variable for the BMP or lead code unit
+ * @param c2 (UChar, out) variable for 0 or the trail code unit
+ * @param result (uint32_t, out) uint32_t variable for the trie lookup result
+ */
+#define UTRIE_NEXT32(trie, src, limit, c, c2, result) _UTRIE_NEXT(trie, data32, src, limit, c, c2, result, uint32_t)
+
+/**
+ * Get the previous code point (c, c2), pre-decrement src,
+ * and get a 16-bit value from the trie.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param start (const UChar *, in) the start pointer for the text, or NULL
+ * @param src (const UChar *, in/out) the source text pointer
+ * @param c (UChar, out) variable for the BMP or lead code unit
+ * @param c2 (UChar, out) variable for 0 or the trail code unit
+ * @param result (uint16_t, out) uint16_t variable for the trie lookup result
+ */
+#define UTRIE_PREVIOUS16(trie, start, src, c, c2, result) _UTRIE_PREVIOUS(trie, index, start, src, c, c2, result, uint16_t)
+
+/**
+ * Get the previous code point (c, c2), pre-decrement src,
+ * and get a 32-bit value from the trie.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param start (const UChar *, in) the start pointer for the text, or NULL
+ * @param src (const UChar *, in/out) the source text pointer
+ * @param c (UChar, out) variable for the BMP or lead code unit
+ * @param c2 (UChar, out) variable for 0 or the trail code unit
+ * @param result (uint32_t, out) uint32_t variable for the trie lookup result
+ */
+#define UTRIE_PREVIOUS32(trie, start, src, c, c2, result) _UTRIE_PREVIOUS(trie, data32, start, src, c, c2, result, uint32_t)
+
+/**
+ * Get a 16-bit trie value from a pair of surrogates.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param c (UChar, in) a lead surrogate
+ * @param c2 (UChar, in) a trail surrogate
+ * @param result (uint16_t, out) uint16_t variable for the trie lookup result
+ */
+#define UTRIE_GET16_FROM_PAIR(trie, c, c2, result) _UTRIE_GET_FROM_PAIR(trie, index, c, c2, result, uint16_t)
+
+/**
+ * Get a 32-bit trie value from a pair of surrogates.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param c (UChar, in) a lead surrogate
+ * @param c2 (UChar, in) a trail surrogate
+ * @param result (uint32_t, out) uint32_t variable for the trie lookup result
+ */
+#define UTRIE_GET32_FROM_PAIR(trie, c, c2, result) _UTRIE_GET_FROM_PAIR(trie, data32, c, c2, result, uint32_t)
+
+/**
+ * Get a 16-bit trie value from a folding offset (from the value of a lead surrogate)
+ * and a trail surrogate.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param offset (int32_t, in) the folding offset from the value of a lead surrogate
+ * @param c2 (UChar, in) a trail surrogate (only the 10 low bits are significant)
+ * @return (uint16_t) trie lookup result
+ */
+#define UTRIE_GET16_FROM_OFFSET_TRAIL(trie, offset, c2) _UTRIE_GET_RAW(trie, index, offset, (c2)&0x3ff)
+
+/**
+ * Get a 32-bit trie value from a folding offset (from the value of a lead surrogate)
+ * and a trail surrogate.
+ *
+ * @param trie (const UTrie *, in) a pointer to the runtime trie structure
+ * @param offset (int32_t, in) the folding offset from the value of a lead surrogate
+ * @param c2 (UChar, in) a trail surrogate (only the 10 low bits are significant)
+ * @return (uint32_t) trie lookup result
+ */
+#define UTRIE_GET32_FROM_OFFSET_TRAIL(trie, offset, c2) _UTRIE_GET_RAW(trie, data32, offset, (c2)&0x3ff)
+
+/* enumeration callback types */
+
+/**
+ * Callback from utrie_enum(), extracts a uint32_t value from a
+ * trie value. This value will be passed on to the UTrieEnumRange function.
+ *
+ * @param context an opaque pointer, as passed into utrie_enum()
+ * @param value a value from the trie
+ * @return the value that is to be passed on to the UTrieEnumRange function
+ */
+typedef uint32_t U_CALLCONV
+UTrieEnumValue(const void *context, uint32_t value);
+
+/**
+ * Callback from utrie_enum(), is called for each contiguous range
+ * of code points with the same value as retrieved from the trie and
+ * transformed by the UTrieEnumValue function.
+ *
+ * The callback function can stop the enumeration by returning false.
+ *
+ * @param context an opaque pointer, as passed into utrie_enum()
+ * @param start the first code point in a contiguous range with value
+ * @param limit one past the last code point in a contiguous range with value
+ * @param value the value that is set for all code points in [start..limit[
+ * @return false to stop the enumeration
+ */
+typedef UBool U_CALLCONV
+UTrieEnumRange(const void *context, UChar32 start, UChar32 limit, uint32_t value);
+
+/**
+ * Enumerate efficiently all values in a trie.
+ * For each entry in the trie, the value to be delivered is passed through
+ * the UTrieEnumValue function.
+ * The value is unchanged if that function pointer is NULL.
+ *
+ * For each contiguous range of code points with a given value,
+ * the UTrieEnumRange function is called.
+ *
+ * @param trie a pointer to the runtime trie structure
+ * @param enumValue a pointer to a function that may transform the trie entry value,
+ *                  or NULL if the values from the trie are to be used directly
+ * @param enumRange a pointer to a function that is called for each contiguous range
+ *                  of code points with the same value
+ * @param context an opaque pointer that is passed on to the callback functions
+ */
+U_CAPI void U_EXPORT2
+utrie_enum(const UTrie *trie,
+           UTrieEnumValue *enumValue, UTrieEnumRange *enumRange, const void *context);
+
+/**
+ * Unserialize a trie from 32-bit-aligned memory.
+ * Inverse of utrie_serialize().
+ * Fills the UTrie runtime trie structure with the settings for the trie data.
+ *
+ * @param trie a pointer to the runtime trie structure
+ * @param data a pointer to 32-bit-aligned memory containing trie data
+ * @param length the number of bytes available at data
+ * @param pErrorCode an in/out ICU UErrorCode
+ * @return the number of bytes at data taken up by the trie data
+ */
+U_CAPI int32_t U_EXPORT2
+utrie_unserialize(UTrie *trie, const void *data, int32_t length, UErrorCode *pErrorCode);
+
+/**
+ * "Unserialize" a dummy trie.
+ * A dummy trie is an empty runtime trie, used when a real data trie cannot
+ * be loaded.
+ *
+ * The input memory is filled so that the trie always returns the initialValue,
+ * or the leadUnitValue for lead surrogate code points.
+ * The Latin-1 part is always set up to be linear.
+ *
+ * @param trie a pointer to the runtime trie structure
+ * @param data a pointer to 32-bit-aligned memory to be filled with the dummy trie data
+ * @param length the number of bytes available at data (recommended to use UTRIE_DUMMY_SIZE)
+ * @param initialValue the initial value that is set for all code points
+ * @param leadUnitValue the value for lead surrogate code _units_ that do not
+ *                      have associated supplementary data
+ * @param pErrorCode an in/out ICU UErrorCode
+ *
+ * @see UTRIE_DUMMY_SIZE
+ * @see utrie_open
+ */
+U_CAPI int32_t U_EXPORT2
+utrie_unserializeDummy(UTrie *trie,
+                       void *data, int32_t length,
+                       uint32_t initialValue, uint32_t leadUnitValue,
+                       UBool make16BitTrie,
+                       UErrorCode *pErrorCode);
+
+/**
+ * Default implementation for UTrie.getFoldingOffset, set automatically by
+ * utrie_unserialize().
+ * Simply returns the lead surrogate's value itself - which is the inverse
+ * of the default folding function used by utrie_serialize().
+ * Exported for static const UTrie structures.
+ *
+ * @see UTrieGetFoldingOffset
+ */
+U_CAPI int32_t U_EXPORT2
+utrie_defaultGetFoldingOffset(uint32_t data);
+
+/* Building a trie ----------------------------------------------------------*/
+
+/**
+ * Build-time trie structure.
+ * Opaque definition, here only to make fillIn parameters possible
+ * for utrie_open() and utrie_clone().
+ */
+struct UNewTrie {
+    /**
+     * Index values at build-time are 32 bits wide for easier processing.
+     * Bit 31 is set if the data block is used by multiple index values (from utrie_setRange()).
+     */
+    int32_t index[UTRIE_MAX_INDEX_LENGTH+UTRIE_SURROGATE_BLOCK_COUNT];
+    uint32_t *data;
+
+    uint32_t leadUnitValue;
+    int32_t indexLength, dataCapacity, dataLength;
+    UBool isAllocated, isDataAllocated;
+    UBool isLatin1Linear, isCompacted;
+
+    /**
+     * Map of adjusted indexes, used in utrie_compact().
+     * Maps from original indexes to new ones.
+     */
+    int32_t map[UTRIE_MAX_BUILD_TIME_DATA_LENGTH>>UTRIE_SHIFT];
+};
+
+typedef struct UNewTrie UNewTrie;
+
+/**
+ * Build-time trie callback function, used with utrie_serialize().
+ * This function calculates a lead surrogate's value including a folding offset
+ * from the 1024 supplementary code points [start..start+1024[ .
+ * It is U+10000 <= start <= U+10fc00 and (start&0x3ff)==0.
+ *
+ * The folding offset is provided by the caller.
+ * It is offset=UTRIE_BMP_INDEX_LENGTH+n*UTRIE_SURROGATE_BLOCK_COUNT with n=0..1023.
+ * Instead of the offset itself, n can be stored in 10 bits -
+ * or fewer if it can be assumed that few lead surrogates have associated data.
+ *
+ * The returned value must be
+ * - not zero if and only if there is relevant data
+ *   for the corresponding 1024 supplementary code points
+ * - such that UTrie.getFoldingOffset(UNewTrieGetFoldedValue(..., offset))==offset
+ *
+ * @return a folded value, or 0 if there is no relevant data for the lead surrogate.
+ */
+typedef uint32_t U_CALLCONV
+UNewTrieGetFoldedValue(UNewTrie *trie, UChar32 start, int32_t offset);
+
+/**
+ * Open a build-time trie structure.
+ * The size of the build-time data array is specified to avoid allocating a large
+ * array in all cases. The array itself can also be passed in.
+ *
+ * Although the trie is never fully expanded to a linear array, especially when
+ * utrie_setRange32() is used, the data array could be large during build time.
+ * The maximum length is
+ * UTRIE_MAX_BUILD_TIME_DATA_LENGTH=0x110000+UTRIE_DATA_BLOCK_LENGTH+0x400.
+ * (Number of Unicode code points + one all-initial-value block +
+ *  possible duplicate entries for 1024 lead surrogates.)
+ * (UTRIE_DATA_BLOCK_LENGTH<=0x200 in all cases.)
+ *
+ * @param fillIn a pointer to a UNewTrie structure to be initialized (will not be released), or
+ *               NULL if one is to be allocated
+ * @param aliasData a pointer to a data array to be used (will not be released), or
+ *                  NULL if one is to be allocated
+ * @param maxDataLength the capacity of aliasData (if not NULL) or
+ *                      the length of the data array to be allocated
+ * @param initialValue the initial value that is set for all code points
+ * @param leadUnitValue the value for lead surrogate code _units_ that do not
+ *                      have associated supplementary data
+ * @param latin1Linear a flag indicating whether the Latin-1 range is to be allocated and
+ *                     kept in a linear, contiguous part of the data array
+ * @return a pointer to the initialized fillIn or the allocated and initialized new UNewTrie
+ */
+U_CAPI UNewTrie * U_EXPORT2
+utrie_open(UNewTrie *fillIn,
+           uint32_t *aliasData, int32_t maxDataLength,
+           uint32_t initialValue, uint32_t leadUnitValue,
+           UBool latin1Linear);
+
+/**
+ * Clone a build-time trie structure with all entries.
+ *
+ * @param fillIn like in utrie_open()
+ * @param other the build-time trie structure to clone
+ * @param aliasData like in utrie_open(),
+ *                  used if aliasDataLength>=(capacity of other's data array)
+ * @param aliasDataLength the length of aliasData
+ * @return a pointer to the initialized fillIn or the allocated and initialized new UNewTrie
+ */
+U_CAPI UNewTrie * U_EXPORT2
+utrie_clone(UNewTrie *fillIn, const UNewTrie *other, uint32_t *aliasData, int32_t aliasDataLength);
+
+/**
+ * Close a build-time trie structure, and release memory
+ * that was allocated by utrie_open() or utrie_clone().
+ *
+ * @param trie the build-time trie
+ */
+U_CAPI void U_EXPORT2
+utrie_close(UNewTrie *trie);
+
+/**
+ * Get the data array of a build-time trie.
+ * The data may be modified, but entries that are equal before
+ * must still be equal after modification.
+ *
+ * @param trie the build-time trie
+ * @param pLength (out) a pointer to a variable that receives the number
+ *                of entries in the data array
+ * @return the data array
+ */
+U_CAPI uint32_t * U_EXPORT2
+utrie_getData(UNewTrie *trie, int32_t *pLength);
+
+/**
+ * Set a value for a code point.
+ *
+ * @param trie the build-time trie
+ * @param c the code point
+ * @param value the value
+ * @return false if a failure occurred (illegal argument or data array overrun)
+ */
+U_CAPI UBool U_EXPORT2
+utrie_set32(UNewTrie *trie, UChar32 c, uint32_t value);
+
+/**
+ * Get a value from a code point as stored in the build-time trie.
+ *
+ * @param trie the build-time trie
+ * @param c the code point
+ * @param pInBlockZero if not NULL, then *pInBlockZero is set to true
+ *                     iff the value is retrieved from block 0;
+ *                     block 0 is the all-initial-value initial block
+ * @return the value
+ */
+U_CAPI uint32_t U_EXPORT2
+utrie_get32(UNewTrie *trie, UChar32 c, UBool *pInBlockZero);
+
+/**
+ * Set a value in a range of code points [start..limit[.
+ * All code points c with start<=c=UTRIE_DATA_BLOCK_LENGTH */
+    int32_t dataLength;
+} UTrieHeader;
+
+/**
+ * Constants for use with UTrieHeader.options.
+ * @internal
+ */
+enum {
+    /** Mask to get the UTRIE_SHIFT value from options. */
+    UTRIE_OPTIONS_SHIFT_MASK=0xf,
+
+    /** Shift options right this much to get the UTRIE_INDEX_SHIFT value. */
+    UTRIE_OPTIONS_INDEX_SHIFT=4,
+
+    /** If set, then the data (stage 2) array is 32 bits wide. */
+    UTRIE_OPTIONS_DATA_IS_32_BIT=0x100,
+
+    /**
+     * If set, then Latin-1 data (for U+0000..U+00ff) is stored in the data (stage 2) array
+     * as a simple, linear array at data+UTRIE_DATA_BLOCK_LENGTH.
+     */
+    UTRIE_OPTIONS_LATIN1_IS_LINEAR=0x200
+};
+
+U_CDECL_END
+
+#endif
diff --git a/intl/icu/source/common/utrie2.cpp b/intl/icu/source/common/utrie2.cpp
new file mode 100644
index 0000000000..74f91bc813
--- /dev/null
+++ b/intl/icu/source/common/utrie2.cpp
@@ -0,0 +1,663 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+*
+*   Copyright (C) 2001-2014, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+******************************************************************************
+*   file name:  utrie2.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2008aug16 (starting from a copy of utrie.c)
+*   created by: Markus W. Scherer
+*
+*   This is a common implementation of a Unicode trie.
+*   It is a kind of compressed, serializable table of 16- or 32-bit values associated with
+*   Unicode code points (0..0x10ffff).
+*   This is the second common version of a Unicode trie (hence the name UTrie2).
+*   See utrie2.h for a comparison.
+*
+*   This file contains only the runtime and enumeration code, for read-only access.
+*   See utrie2_builder.c for the builder code.
+*/
+#include "unicode/utypes.h"
+#ifdef UCPTRIE_DEBUG
+#include "unicode/umutablecptrie.h"
+#endif
+#include "unicode/utf.h"
+#include "unicode/utf8.h"
+#include "unicode/utf16.h"
+#include "cmemory.h"
+#include "utrie2.h"
+#include "utrie2_impl.h"
+#include "uassert.h"
+
+/* Public UTrie2 API implementation ----------------------------------------- */
+
+static uint32_t
+get32(const UNewTrie2 *trie, UChar32 c, UBool fromLSCP) {
+    int32_t i2, block;
+
+    if(c>=trie->highStart && (!U_IS_LEAD(c) || fromLSCP)) {
+        return trie->data[trie->dataLength-UTRIE2_DATA_GRANULARITY];
+    }
+
+    if(U_IS_LEAD(c) && fromLSCP) {
+        i2=(UTRIE2_LSCP_INDEX_2_OFFSET-(0xd800>>UTRIE2_SHIFT_2))+
+            (c>>UTRIE2_SHIFT_2);
+    } else {
+        i2=trie->index1[c>>UTRIE2_SHIFT_1]+
+            ((c>>UTRIE2_SHIFT_2)&UTRIE2_INDEX_2_MASK);
+    }
+    block=trie->index2[i2];
+    return trie->data[block+(c&UTRIE2_DATA_MASK)];
+}
+
+U_CAPI uint32_t U_EXPORT2
+utrie2_get32(const UTrie2 *trie, UChar32 c) {
+    if(trie->data16!=nullptr) {
+        return UTRIE2_GET16(trie, c);
+    } else if(trie->data32!=nullptr) {
+        return UTRIE2_GET32(trie, c);
+    } else if((uint32_t)c>0x10ffff) {
+        return trie->errorValue;
+    } else {
+        return get32(trie->newTrie, c, true);
+    }
+}
+
+U_CAPI uint32_t U_EXPORT2
+utrie2_get32FromLeadSurrogateCodeUnit(const UTrie2 *trie, UChar32 c) {
+    if(!U_IS_LEAD(c)) {
+        return trie->errorValue;
+    }
+    if(trie->data16!=nullptr) {
+        return UTRIE2_GET16_FROM_U16_SINGLE_LEAD(trie, c);
+    } else if(trie->data32!=nullptr) {
+        return UTRIE2_GET32_FROM_U16_SINGLE_LEAD(trie, c);
+    } else {
+        return get32(trie->newTrie, c, false);
+    }
+}
+
+static inline int32_t
+u8Index(const UTrie2 *trie, UChar32 c, int32_t i) {
+    int32_t idx=
+        _UTRIE2_INDEX_FROM_CP(
+            trie,
+            trie->data32==nullptr ? trie->indexLength : 0,
+            c);
+    return (idx<<3)|i;
+}
+
+U_CAPI int32_t U_EXPORT2
+utrie2_internalU8NextIndex(const UTrie2 *trie, UChar32 c,
+                           const uint8_t *src, const uint8_t *limit) {
+    int32_t i, length;
+    i=0;
+    /* support 64-bit pointers by avoiding cast of arbitrary difference */
+    if((limit-src)<=7) {
+        length=(int32_t)(limit-src);
+    } else {
+        length=7;
+    }
+    c=utf8_nextCharSafeBody(src, &i, length, c, -1);
+    return u8Index(trie, c, i);
+}
+
+U_CAPI int32_t U_EXPORT2
+utrie2_internalU8PrevIndex(const UTrie2 *trie, UChar32 c,
+                           const uint8_t *start, const uint8_t *src) {
+    int32_t i, length;
+    /* support 64-bit pointers by avoiding cast of arbitrary difference */
+    if((src-start)<=7) {
+        i=length=(int32_t)(src-start);
+    } else {
+        i=length=7;
+        start=src-7;
+    }
+    c=utf8_prevCharSafeBody(start, 0, &i, c, -1);
+    i=length-i;  /* number of bytes read backward from src */
+    return u8Index(trie, c, i);
+}
+
+U_CAPI UTrie2 * U_EXPORT2
+utrie2_openFromSerialized(UTrie2ValueBits valueBits,
+                          const void *data, int32_t length, int32_t *pActualLength,
+                          UErrorCode *pErrorCode) {
+    const UTrie2Header *header;
+    const uint16_t *p16;
+    int32_t actualLength;
+
+    UTrie2 tempTrie;
+    UTrie2 *trie;
+
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+
+    if( length<=0 || (U_POINTER_MASK_LSB(data, 3)!=0) ||
+        valueBits<0 || UTRIE2_COUNT_VALUE_BITS<=valueBits
+    ) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    /* enough data for a trie header? */
+    if(length<(int32_t)sizeof(UTrie2Header)) {
+        *pErrorCode=U_INVALID_FORMAT_ERROR;
+        return 0;
+    }
+
+    /* check the signature */
+    header=(const UTrie2Header *)data;
+    if(header->signature!=UTRIE2_SIG) {
+        *pErrorCode=U_INVALID_FORMAT_ERROR;
+        return 0;
+    }
+
+    /* get the options */
+    if(valueBits!=(UTrie2ValueBits)(header->options&UTRIE2_OPTIONS_VALUE_BITS_MASK)) {
+        *pErrorCode=U_INVALID_FORMAT_ERROR;
+        return 0;
+    }
+
+    /* get the length values and offsets */
+    uprv_memset(&tempTrie, 0, sizeof(tempTrie));
+    tempTrie.indexLength=header->indexLength;
+    tempTrie.dataLength=header->shiftedDataLength<index2NullOffset;
+    tempTrie.dataNullOffset=header->dataNullOffset;
+
+    tempTrie.highStart=header->shiftedHighStart<memory=(uint32_t *)data;
+    trie->length=actualLength;
+    trie->isMemoryOwned=false;
+#ifdef UTRIE2_DEBUG
+    trie->name="fromSerialized";
+#endif
+
+    /* set the pointers to its index and data arrays */
+    p16=(const uint16_t *)(header+1);
+    trie->index=p16;
+    p16+=trie->indexLength;
+
+    /* get the data */
+    switch(valueBits) {
+    case UTRIE2_16_VALUE_BITS:
+        trie->data16=p16;
+        trie->data32=nullptr;
+        trie->initialValue=trie->index[trie->dataNullOffset];
+        trie->errorValue=trie->data16[UTRIE2_BAD_UTF8_DATA_OFFSET];
+        break;
+    case UTRIE2_32_VALUE_BITS:
+        trie->data16=nullptr;
+        trie->data32=(const uint32_t *)p16;
+        trie->initialValue=trie->data32[trie->dataNullOffset];
+        trie->errorValue=trie->data32[UTRIE2_BAD_UTF8_DATA_OFFSET];
+        break;
+    default:
+        *pErrorCode=U_INVALID_FORMAT_ERROR;
+        return 0;
+    }
+
+    if(pActualLength!=nullptr) {
+        *pActualLength=actualLength;
+    }
+    return trie;
+}
+
+U_CAPI UTrie2 * U_EXPORT2
+utrie2_openDummy(UTrie2ValueBits valueBits,
+                 uint32_t initialValue, uint32_t errorValue,
+                 UErrorCode *pErrorCode) {
+    UTrie2 *trie;
+    UTrie2Header *header;
+    uint32_t *p;
+    uint16_t *dest16;
+    int32_t indexLength, dataLength, length, i;
+    int32_t dataMove;  /* >0 if the data is moved to the end of the index array */
+
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+
+    if(valueBits<0 || UTRIE2_COUNT_VALUE_BITS<=valueBits) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    /* calculate the total length of the dummy trie data */
+    indexLength=UTRIE2_INDEX_1_OFFSET;
+    dataLength=UTRIE2_DATA_START_OFFSET+UTRIE2_DATA_GRANULARITY;
+    length=(int32_t)sizeof(UTrie2Header)+indexLength*2;
+    if(valueBits==UTRIE2_16_VALUE_BITS) {
+        length+=dataLength*2;
+    } else {
+        length+=dataLength*4;
+    }
+
+    /* allocate the trie */
+    trie=(UTrie2 *)uprv_malloc(sizeof(UTrie2));
+    if(trie==nullptr) {
+        *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+        return 0;
+    }
+    uprv_memset(trie, 0, sizeof(UTrie2));
+    trie->memory=uprv_malloc(length);
+    if(trie->memory==nullptr) {
+        uprv_free(trie);
+        *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+        return 0;
+    }
+    trie->length=length;
+    trie->isMemoryOwned=true;
+
+    /* set the UTrie2 fields */
+    if(valueBits==UTRIE2_16_VALUE_BITS) {
+        dataMove=indexLength;
+    } else {
+        dataMove=0;
+    }
+
+    trie->indexLength=indexLength;
+    trie->dataLength=dataLength;
+    trie->index2NullOffset=UTRIE2_INDEX_2_OFFSET;
+    trie->dataNullOffset=(uint16_t)dataMove;
+    trie->initialValue=initialValue;
+    trie->errorValue=errorValue;
+    trie->highStart=0;
+    trie->highValueIndex=dataMove+UTRIE2_DATA_START_OFFSET;
+#ifdef UTRIE2_DEBUG
+    trie->name="dummy";
+#endif
+
+    /* set the header fields */
+    header=(UTrie2Header *)trie->memory;
+
+    header->signature=UTRIE2_SIG; /* "Tri2" */
+    header->options=(uint16_t)valueBits;
+
+    header->indexLength=(uint16_t)indexLength;
+    header->shiftedDataLength=(uint16_t)(dataLength>>UTRIE2_INDEX_SHIFT);
+    header->index2NullOffset=(uint16_t)UTRIE2_INDEX_2_OFFSET;
+    header->dataNullOffset=(uint16_t)dataMove;
+    header->shiftedHighStart=0;
+
+    /* fill the index and data arrays */
+    dest16=(uint16_t *)(header+1);
+    trie->index=dest16;
+
+    /* write the index-2 array values shifted right by UTRIE2_INDEX_SHIFT */
+    for(i=0; i>UTRIE2_INDEX_SHIFT);  /* null data block */
+    }
+
+    /* write UTF-8 2-byte index-2 values, not right-shifted */
+    for(i=0; i<(0xc2-0xc0); ++i) {                                  /* C0..C1 */
+        *dest16++=(uint16_t)(dataMove+UTRIE2_BAD_UTF8_DATA_OFFSET);
+    }
+    for(; i<(0xe0-0xc0); ++i) {                                     /* C2..DF */
+        *dest16++=(uint16_t)dataMove;
+    }
+
+    /* write the 16/32-bit data array */
+    switch(valueBits) {
+    case UTRIE2_16_VALUE_BITS:
+        /* write 16-bit data values */
+        trie->data16=dest16;
+        trie->data32=nullptr;
+        for(i=0; i<0x80; ++i) {
+            *dest16++=(uint16_t)initialValue;
+        }
+        for(; i<0xc0; ++i) {
+            *dest16++=(uint16_t)errorValue;
+        }
+        /* highValue and reserved values */
+        for(i=0; idata16=nullptr;
+        trie->data32=p;
+        for(i=0; i<0x80; ++i) {
+            *p++=initialValue;
+        }
+        for(; i<0xc0; ++i) {
+            *p++=errorValue;
+        }
+        /* highValue and reserved values */
+        for(i=0; iisMemoryOwned) {
+            uprv_free(trie->memory);
+        }
+        if(trie->newTrie!=nullptr) {
+            uprv_free(trie->newTrie->data);
+#ifdef UCPTRIE_DEBUG
+            umutablecptrie_close(trie->newTrie->t3);
+#endif
+            uprv_free(trie->newTrie);
+        }
+        uprv_free(trie);
+    }
+}
+
+U_CAPI UBool U_EXPORT2
+utrie2_isFrozen(const UTrie2 *trie) {
+    return (UBool)(trie->newTrie==nullptr);
+}
+
+U_CAPI int32_t U_EXPORT2
+utrie2_serialize(const UTrie2 *trie,
+                 void *data, int32_t capacity,
+                 UErrorCode *pErrorCode) {
+    /* argument check */
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+
+    if( trie==nullptr || trie->memory==nullptr || trie->newTrie!=nullptr ||
+        capacity<0 || (capacity>0 && (data==nullptr || (U_POINTER_MASK_LSB(data, 3)!=0)))
+    ) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    if(capacity>=trie->length) {
+        uprv_memcpy(data, trie->memory, trie->length);
+    } else {
+        *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
+    }
+    return trie->length;
+}
+
+/* enumeration -------------------------------------------------------------- */
+
+#define MIN_VALUE(a, b) ((a)<(b) ? (a) : (b))
+
+/* default UTrie2EnumValue() returns the input value itself */
+static uint32_t U_CALLCONV
+enumSameValue(const void * /*context*/, uint32_t value) {
+    return value;
+}
+
+/**
+ * Enumerate all ranges of code points with the same relevant values.
+ * The values are transformed from the raw trie entries by the enumValue function.
+ *
+ * Currently requires startnewTrie==nullptr) {
+        /* frozen trie */
+        idx=trie->index;
+        U_ASSERT(idx!=nullptr); /* the following code assumes trie->newTrie is not nullptr when idx is nullptr */
+        data32=trie->data32;
+
+        index2NullOffset=trie->index2NullOffset;
+        nullBlock=trie->dataNullOffset;
+    } else {
+        /* unfrozen, mutable trie */
+        idx=nullptr;
+        data32=trie->newTrie->data;
+        U_ASSERT(data32!=nullptr); /* the following code assumes idx is not nullptr when data32 is nullptr */
+
+        index2NullOffset=trie->newTrie->index2NullOffset;
+        nullBlock=trie->newTrie->dataNullOffset;
+    }
+
+    highStart=trie->highStart;
+
+    /* get the enumeration value that corresponds to an initial-value trie data entry */
+    initialValue=enumValue(context, trie->initialValue);
+
+    /* set variables for previous range */
+    prevI2Block=-1;
+    prevBlock=-1;
+    prev=start;
+    prevValue=0;
+
+    /* enumerate index-2 blocks */
+    for(c=start; c>UTRIE2_SHIFT_2;
+            } else if(U_IS_SURROGATE_LEAD(c)) {
+                /*
+                 * Enumerate values for lead surrogate code points, not code units:
+                 * This special block has half the normal length.
+                 */
+                i2Block=UTRIE2_LSCP_INDEX_2_OFFSET;
+                tempLimit=MIN_VALUE(0xdc00, limit);
+            } else {
+                /*
+                 * Switch back to the normal part of the index-2 table.
+                 * Enumerate the second half of the surrogates block.
+                 */
+                i2Block=0xd800>>UTRIE2_SHIFT_2;
+                tempLimit=MIN_VALUE(0xe000, limit);
+            }
+        } else {
+            /* supplementary code points */
+            if(idx!=nullptr) {
+                i2Block=idx[(UTRIE2_INDEX_1_OFFSET-UTRIE2_OMITTED_BMP_INDEX_1_LENGTH)+
+                              (c>>UTRIE2_SHIFT_1)];
+            } else {
+                i2Block=trie->newTrie->index1[c>>UTRIE2_SHIFT_1];
+            }
+            if(i2Block==prevI2Block && (c-prev)>=UTRIE2_CP_PER_INDEX_1_ENTRY) {
+                /*
+                 * The index-2 block is the same as the previous one, and filled with prevValue.
+                 * Only possible for supplementary code points because the linear-BMP index-2
+                 * table creates unique i2Block values.
+                 */
+                c+=UTRIE2_CP_PER_INDEX_1_ENTRY;
+                continue;
+            }
+        }
+        prevI2Block=i2Block;
+        if(i2Block==index2NullOffset) {
+            /* this is the null index-2 block */
+            if(prevValue!=initialValue) {
+                if(prev>UTRIE2_SHIFT_2)&UTRIE2_INDEX_2_MASK;
+            if((c>>UTRIE2_SHIFT_1)==(tempLimit>>UTRIE2_SHIFT_1)) {
+                i2Limit=(tempLimit>>UTRIE2_SHIFT_2)&UTRIE2_INDEX_2_MASK;
+            } else {
+                i2Limit=UTRIE2_INDEX_2_BLOCK_LENGTH;
+            }
+            for(; i2newTrie->index2[i2Block+i2];
+                }
+                if(block==prevBlock && (c-prev)>=UTRIE2_DATA_BLOCK_LENGTH) {
+                    /* the block is the same as the previous one, and filled with prevValue */
+                    c+=UTRIE2_DATA_BLOCK_LENGTH;
+                    continue;
+                }
+                prevBlock=block;
+                if(block==nullBlock) {
+                    /* this is the null data block */
+                    if(prevValue!=initialValue) {
+                        if(prevlimit) {
+        c=limit;  /* could be higher if in the index2NullOffset */
+    } else if(chighValueIndex] :
+                    idx[trie->highValueIndex];
+        } else {
+            highValue=trie->newTrie->data[trie->newTrie->dataLength-UTRIE2_DATA_GRANULARITY];
+        }
+        value=enumValue(context, highValue);
+        if(value!=prevValue) {
+            if(prev=codePointStart) {
+        codePoint=U_SENTINEL;
+        return static_cast(trie->errorValue);
+    }
+    uint16_t result;
+    UTRIE2_U16_PREV16(trie, start, codePointStart, codePoint, result);
+    return result;
+}
+
+uint16_t ForwardUTrie2StringIterator::next16() {
+    codePointStart=codePointLimit;
+    if(codePointLimit==limit) {
+        codePoint=U_SENTINEL;
+        return static_cast(trie->errorValue);
+    }
+    uint16_t result;
+    UTRIE2_U16_NEXT16(trie, codePointLimit, limit, codePoint, result);
+    return result;
+}
+
+U_NAMESPACE_END
diff --git a/intl/icu/source/common/utrie2.h b/intl/icu/source/common/utrie2.h
new file mode 100644
index 0000000000..431f11568e
--- /dev/null
+++ b/intl/icu/source/common/utrie2.h
@@ -0,0 +1,955 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+*
+*   Copyright (C) 2001-2014, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+******************************************************************************
+*   file name:  utrie2.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2008aug16 (starting from a copy of utrie.h)
+*   created by: Markus W. Scherer
+*/
+
+#ifndef __UTRIE2_H__
+#define __UTRIE2_H__
+
+#include "unicode/utypes.h"
+#include "unicode/utf8.h"
+#include "putilimp.h"
+
+U_CDECL_BEGIN
+
+struct UTrie;  /* forward declaration */
+#ifndef __UTRIE_H__
+typedef struct UTrie UTrie;
+#endif
+
+/**
+ * \file
+ *
+ * This is a common implementation of a Unicode trie.
+ * It is a kind of compressed, serializable table of 16- or 32-bit values associated with
+ * Unicode code points (0..0x10ffff). (A map from code points to integers.)
+ *
+ * This is the second common version of a Unicode trie (hence the name UTrie2).
+ * Compared with UTrie version 1:
+ * - Still splitting BMP code points 11:5 bits for index and data table lookups.
+ * - Still separate data for lead surrogate code _units_ vs. code _points_,
+ *   but the lead surrogate code unit values are not required any more
+ *   for data lookup for supplementary code points.
+ * - The "folding" mechanism is removed. In UTrie version 1, this somewhat
+ *   hard-to-explain mechanism was meant to be used for optimized UTF-16
+ *   processing, with application-specific encoding of indexing bits
+ *   in the lead surrogate data for the associated supplementary code points.
+ * - For the last single-value code point range (ending with U+10ffff),
+ *   the starting code point ("highStart") and the value are stored.
+ * - For supplementary code points U+10000..highStart-1 a three-table lookup
+ *   (two index tables and one data table) is used. The first index
+ *   is truncated, omitting both the BMP portion and the high range.
+ * - There is a special small index for 2-byte UTF-8, and the initial data
+ *   entries are designed for fast 1/2-byte UTF-8 lookup.
+ *   Starting with ICU 60, C0 and C1 are not recognized as UTF-8 lead bytes any more at all,
+ *   and the associated 2-byte indexes are unused.
+ */
+
+/**
+ * Trie structure.
+ * Use only with public API macros and functions.
+ */
+struct UTrie2;
+typedef struct UTrie2 UTrie2;
+
+/* Public UTrie2 API functions: read-only access ---------------------------- */
+
+/**
+ * Selectors for the width of a UTrie2 data value.
+ */
+enum UTrie2ValueBits {
+    /** 16 bits per UTrie2 data value. */
+    UTRIE2_16_VALUE_BITS,
+    /** 32 bits per UTrie2 data value. */
+    UTRIE2_32_VALUE_BITS,
+    /** Number of selectors for the width of UTrie2 data values. */
+    UTRIE2_COUNT_VALUE_BITS
+};
+typedef enum UTrie2ValueBits UTrie2ValueBits;
+
+/**
+ * Open a frozen trie from its serialized from, stored in 32-bit-aligned memory.
+ * Inverse of utrie2_serialize().
+ * The memory must remain valid and unchanged as long as the trie is used.
+ * You must utrie2_close() the trie once you are done using it.
+ *
+ * @param valueBits selects the data entry size; results in an
+ *                  U_INVALID_FORMAT_ERROR if it does not match the serialized form
+ * @param data a pointer to 32-bit-aligned memory containing the serialized form of a UTrie2
+ * @param length the number of bytes available at data;
+ *               can be more than necessary
+ * @param pActualLength receives the actual number of bytes at data taken up by the trie data;
+ *                      can be NULL
+ * @param pErrorCode an in/out ICU UErrorCode
+ * @return the unserialized trie
+ *
+ * @see utrie2_open
+ * @see utrie2_serialize
+ */
+U_CAPI UTrie2 * U_EXPORT2
+utrie2_openFromSerialized(UTrie2ValueBits valueBits,
+                          const void *data, int32_t length, int32_t *pActualLength,
+                          UErrorCode *pErrorCode);
+
+/**
+ * Open a frozen, empty "dummy" trie.
+ * A dummy trie is an empty trie, used when a real data trie cannot
+ * be loaded. Equivalent to calling utrie2_open() and utrie2_freeze(),
+ * but without internally creating and compacting/serializing the
+ * builder data structure.
+ *
+ * The trie always returns the initialValue,
+ * or the errorValue for out-of-range code points and illegal UTF-8.
+ *
+ * You must utrie2_close() the trie once you are done using it.
+ *
+ * @param valueBits selects the data entry size
+ * @param initialValue the initial value that is set for all code points
+ * @param errorValue the value for out-of-range code points and illegal UTF-8
+ * @param pErrorCode an in/out ICU UErrorCode
+ * @return the dummy trie
+ *
+ * @see utrie2_openFromSerialized
+ * @see utrie2_open
+ */
+U_CAPI UTrie2 * U_EXPORT2
+utrie2_openDummy(UTrie2ValueBits valueBits,
+                 uint32_t initialValue, uint32_t errorValue,
+                 UErrorCode *pErrorCode);
+
+/**
+ * Get a value from a code point as stored in the trie.
+ * Easier to use than UTRIE2_GET16() and UTRIE2_GET32() but slower.
+ * Easier to use because, unlike the macros, this function works on all UTrie2
+ * objects, frozen or not, holding 16-bit or 32-bit data values.
+ *
+ * @param trie the trie
+ * @param c the code point
+ * @return the value
+ */
+U_CAPI uint32_t U_EXPORT2
+utrie2_get32(const UTrie2 *trie, UChar32 c);
+
+/* enumeration callback types */
+
+/**
+ * Callback from utrie2_enum(), extracts a uint32_t value from a
+ * trie value. This value will be passed on to the UTrie2EnumRange function.
+ *
+ * @param context an opaque pointer, as passed into utrie2_enum()
+ * @param value a value from the trie
+ * @return the value that is to be passed on to the UTrie2EnumRange function
+ */
+typedef uint32_t U_CALLCONV
+UTrie2EnumValue(const void *context, uint32_t value);
+
+/**
+ * Callback from utrie2_enum(), is called for each contiguous range
+ * of code points with the same value as retrieved from the trie and
+ * transformed by the UTrie2EnumValue function.
+ *
+ * The callback function can stop the enumeration by returning false.
+ *
+ * @param context an opaque pointer, as passed into utrie2_enum()
+ * @param start the first code point in a contiguous range with value
+ * @param end the last code point in a contiguous range with value (inclusive)
+ * @param value the value that is set for all code points in [start..end]
+ * @return false to stop the enumeration
+ */
+typedef UBool U_CALLCONV
+UTrie2EnumRange(const void *context, UChar32 start, UChar32 end, uint32_t value);
+
+/**
+ * Enumerate efficiently all values in a trie.
+ * Do not modify the trie during the enumeration.
+ *
+ * For each entry in the trie, the value to be delivered is passed through
+ * the UTrie2EnumValue function.
+ * The value is unchanged if that function pointer is NULL.
+ *
+ * For each contiguous range of code points with a given (transformed) value,
+ * the UTrie2EnumRange function is called.
+ *
+ * @param trie a pointer to the trie
+ * @param enumValue a pointer to a function that may transform the trie entry value,
+ *                  or NULL if the values from the trie are to be used directly
+ * @param enumRange a pointer to a function that is called for each contiguous range
+ *                  of code points with the same (transformed) value
+ * @param context an opaque pointer that is passed on to the callback functions
+ */
+U_CAPI void U_EXPORT2
+utrie2_enum(const UTrie2 *trie,
+            UTrie2EnumValue *enumValue, UTrie2EnumRange *enumRange, const void *context);
+
+/* Building a trie ---------------------------------------------------------- */
+
+/**
+ * Open an empty, writable trie. At build time, 32-bit data values are used.
+ * utrie2_freeze() takes a valueBits parameter
+ * which determines the data value width in the serialized and frozen forms.
+ * You must utrie2_close() the trie once you are done using it.
+ *
+ * @param initialValue the initial value that is set for all code points
+ * @param errorValue the value for out-of-range code points and illegal UTF-8
+ * @param pErrorCode an in/out ICU UErrorCode
+ * @return a pointer to the allocated and initialized new trie
+ */
+U_CAPI UTrie2 * U_EXPORT2
+utrie2_open(uint32_t initialValue, uint32_t errorValue, UErrorCode *pErrorCode);
+
+/**
+ * Clone a trie.
+ * You must utrie2_close() the clone once you are done using it.
+ *
+ * @param other the trie to clone
+ * @param pErrorCode an in/out ICU UErrorCode
+ * @return a pointer to the new trie clone
+ */
+U_CAPI UTrie2 * U_EXPORT2
+utrie2_clone(const UTrie2 *other, UErrorCode *pErrorCode);
+
+/**
+ * Clone a trie. The clone will be mutable/writable even if the other trie
+ * is frozen. (See utrie2_freeze().)
+ * You must utrie2_close() the clone once you are done using it.
+ *
+ * @param other the trie to clone
+ * @param pErrorCode an in/out ICU UErrorCode
+ * @return a pointer to the new trie clone
+ */
+U_CAPI UTrie2 * U_EXPORT2
+utrie2_cloneAsThawed(const UTrie2 *other, UErrorCode *pErrorCode);
+
+/**
+ * Close a trie and release associated memory.
+ *
+ * @param trie the trie
+ */
+U_CAPI void U_EXPORT2
+utrie2_close(UTrie2 *trie);
+
+/**
+ * Set a value for a code point.
+ *
+ * @param trie the unfrozen trie
+ * @param c the code point
+ * @param value the value
+ * @param pErrorCode an in/out ICU UErrorCode; among other possible error codes:
+ * - U_NO_WRITE_PERMISSION if the trie is frozen
+ */
+U_CAPI void U_EXPORT2
+utrie2_set32(UTrie2 *trie, UChar32 c, uint32_t value, UErrorCode *pErrorCode);
+
+/**
+ * Set a value in a range of code points [start..end].
+ * All code points c with start<=c<=end will get the value if
+ * overwrite is true or if the old value is the initial value.
+ *
+ * @param trie the unfrozen trie
+ * @param start the first code point to get the value
+ * @param end the last code point to get the value (inclusive)
+ * @param value the value
+ * @param overwrite flag for whether old non-initial values are to be overwritten
+ * @param pErrorCode an in/out ICU UErrorCode; among other possible error codes:
+ * - U_NO_WRITE_PERMISSION if the trie is frozen
+ */
+U_CAPI void U_EXPORT2
+utrie2_setRange32(UTrie2 *trie,
+                  UChar32 start, UChar32 end,
+                  uint32_t value, UBool overwrite,
+                  UErrorCode *pErrorCode);
+
+/**
+ * Freeze a trie. Make it immutable (read-only) and compact it,
+ * ready for serialization and for use with fast macros.
+ * Functions to set values will fail after serializing.
+ *
+ * A trie can be frozen only once. If this function is called again with different
+ * valueBits then it will set a U_ILLEGAL_ARGUMENT_ERROR.
+ *
+ * @param trie the trie
+ * @param valueBits selects the data entry size; if smaller than 32 bits, then
+ *                  the values stored in the trie will be truncated
+ * @param pErrorCode an in/out ICU UErrorCode; among other possible error codes:
+ * - U_INDEX_OUTOFBOUNDS_ERROR if the compacted index or data arrays are too long
+ *                             for serialization
+ *                             (the trie will be immutable and usable,
+ *                             but not frozen and not usable with the fast macros)
+ *
+ * @see utrie2_cloneAsThawed
+ */
+U_CAPI void U_EXPORT2
+utrie2_freeze(UTrie2 *trie, UTrie2ValueBits valueBits, UErrorCode *pErrorCode);
+
+/**
+ * Test if the trie is frozen. (See utrie2_freeze().)
+ *
+ * @param trie the trie
+ * @return true if the trie is frozen, that is, immutable, ready for serialization
+ *         and for use with fast macros
+ */
+U_CAPI UBool U_EXPORT2
+utrie2_isFrozen(const UTrie2 *trie);
+
+/**
+ * Serialize a frozen trie into 32-bit aligned memory.
+ * If the trie is not frozen, then the function returns with a U_ILLEGAL_ARGUMENT_ERROR.
+ * A trie can be serialized multiple times.
+ *
+ * @param trie the frozen trie
+ * @param data a pointer to 32-bit-aligned memory to be filled with the trie data,
+ *             can be NULL if capacity==0
+ * @param capacity the number of bytes available at data,
+ *                 or 0 for preflighting
+ * @param pErrorCode an in/out ICU UErrorCode; among other possible error codes:
+ * - U_BUFFER_OVERFLOW_ERROR if the data storage block is too small for serialization
+ * - U_ILLEGAL_ARGUMENT_ERROR if the trie is not frozen or the data and capacity
+ *                            parameters are bad
+ * @return the number of bytes written or needed for the trie
+ *
+ * @see utrie2_openFromSerialized()
+ */
+U_CAPI int32_t U_EXPORT2
+utrie2_serialize(const UTrie2 *trie,
+                 void *data, int32_t capacity,
+                 UErrorCode *pErrorCode);
+
+/* Public UTrie2 API: miscellaneous functions ------------------------------- */
+
+/**
+ * Build a UTrie2 (version 2) from a UTrie (version 1).
+ * Enumerates all values in the UTrie and builds a UTrie2 with the same values.
+ * The resulting UTrie2 will be frozen.
+ *
+ * @param trie1 the runtime UTrie structure to be enumerated
+ * @param errorValue the value for out-of-range code points and illegal UTF-8
+ * @param pErrorCode an in/out ICU UErrorCode
+ * @return The frozen UTrie2 with the same values as the UTrie.
+ */
+U_CAPI UTrie2 * U_EXPORT2
+utrie2_fromUTrie(const UTrie *trie1, uint32_t errorValue, UErrorCode *pErrorCode);
+
+/* Public UTrie2 API macros ------------------------------------------------- */
+
+/*
+ * These macros provide fast data lookup from a frozen trie.
+ * They will crash when used on an unfrozen trie.
+ */
+
+/**
+ * Return a 16-bit trie value from a code point, with range checking.
+ * Returns trie->errorValue if c is not in the range 0..U+10ffff.
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param c (UChar32, in) the input code point
+ * @return (uint16_t) The code point's trie value.
+ */
+#define UTRIE2_GET16(trie, c) _UTRIE2_GET((trie), index, (trie)->indexLength, (c))
+
+/**
+ * Return a 32-bit trie value from a code point, with range checking.
+ * Returns trie->errorValue if c is not in the range 0..U+10ffff.
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param c (UChar32, in) the input code point
+ * @return (uint32_t) The code point's trie value.
+ */
+#define UTRIE2_GET32(trie, c) _UTRIE2_GET((trie), data32, 0, (c))
+
+/**
+ * UTF-16: Get the next code point (UChar32 c, out), post-increment src,
+ * and get a 16-bit value from the trie.
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param src (const UChar *, in/out) the source text pointer
+ * @param limit (const UChar *, in) the limit pointer for the text, or NULL if NUL-terminated
+ * @param c (UChar32, out) variable for the code point
+ * @param result (uint16_t, out) uint16_t variable for the trie lookup result
+ */
+#define UTRIE2_U16_NEXT16(trie, src, limit, c, result) _UTRIE2_U16_NEXT(trie, index, src, limit, c, result)
+
+/**
+ * UTF-16: Get the next code point (UChar32 c, out), post-increment src,
+ * and get a 32-bit value from the trie.
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param src (const UChar *, in/out) the source text pointer
+ * @param limit (const UChar *, in) the limit pointer for the text, or NULL if NUL-terminated
+ * @param c (UChar32, out) variable for the code point
+ * @param result (uint32_t, out) uint32_t variable for the trie lookup result
+ */
+#define UTRIE2_U16_NEXT32(trie, src, limit, c, result) _UTRIE2_U16_NEXT(trie, data32, src, limit, c, result)
+
+/**
+ * UTF-16: Get the previous code point (UChar32 c, out), pre-decrement src,
+ * and get a 16-bit value from the trie.
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param start (const UChar *, in) the start pointer for the text
+ * @param src (const UChar *, in/out) the source text pointer
+ * @param c (UChar32, out) variable for the code point
+ * @param result (uint16_t, out) uint16_t variable for the trie lookup result
+ */
+#define UTRIE2_U16_PREV16(trie, start, src, c, result) _UTRIE2_U16_PREV(trie, index, start, src, c, result)
+
+/**
+ * UTF-16: Get the previous code point (UChar32 c, out), pre-decrement src,
+ * and get a 32-bit value from the trie.
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param start (const UChar *, in) the start pointer for the text
+ * @param src (const UChar *, in/out) the source text pointer
+ * @param c (UChar32, out) variable for the code point
+ * @param result (uint32_t, out) uint32_t variable for the trie lookup result
+ */
+#define UTRIE2_U16_PREV32(trie, start, src, c, result) _UTRIE2_U16_PREV(trie, data32, start, src, c, result)
+
+/**
+ * UTF-8: Post-increment src and get a 16-bit value from the trie.
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param src (const char *, in/out) the source text pointer
+ * @param limit (const char *, in) the limit pointer for the text (must not be NULL)
+ * @param result (uint16_t, out) uint16_t variable for the trie lookup result
+ */
+#define UTRIE2_U8_NEXT16(trie, src, limit, result)\
+    _UTRIE2_U8_NEXT(trie, data16, index, src, limit, result)
+
+/**
+ * UTF-8: Post-increment src and get a 32-bit value from the trie.
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param src (const char *, in/out) the source text pointer
+ * @param limit (const char *, in) the limit pointer for the text (must not be NULL)
+ * @param result (uint16_t, out) uint32_t variable for the trie lookup result
+ */
+#define UTRIE2_U8_NEXT32(trie, src, limit, result) \
+    _UTRIE2_U8_NEXT(trie, data32, data32, src, limit, result)
+
+/**
+ * UTF-8: Pre-decrement src and get a 16-bit value from the trie.
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param start (const char *, in) the start pointer for the text
+ * @param src (const char *, in/out) the source text pointer
+ * @param result (uint16_t, out) uint16_t variable for the trie lookup result
+ */
+#define UTRIE2_U8_PREV16(trie, start, src, result) \
+    _UTRIE2_U8_PREV(trie, data16, index, start, src, result)
+
+/**
+ * UTF-8: Pre-decrement src and get a 32-bit value from the trie.
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param start (const char *, in) the start pointer for the text
+ * @param src (const char *, in/out) the source text pointer
+ * @param result (uint16_t, out) uint32_t variable for the trie lookup result
+ */
+#define UTRIE2_U8_PREV32(trie, start, src, result) \
+    _UTRIE2_U8_PREV(trie, data32, data32, start, src, result)
+
+/* Public UTrie2 API: optimized UTF-16 access ------------------------------- */
+
+/*
+ * The following functions and macros are used for highly optimized UTF-16
+ * text processing. The UTRIE2_U16_NEXTxy() macros do not depend on these.
+ *
+ * A UTrie2 stores separate values for lead surrogate code _units_ vs. code _points_.
+ * UTF-16 text processing can be optimized by detecting surrogate pairs and
+ * assembling supplementary code points only when there is non-trivial data
+ * available.
+ *
+ * At build-time, use utrie2_enumForLeadSurrogate() to see if there
+ * is non-trivial (non-initialValue) data for any of the supplementary
+ * code points associated with a lead surrogate.
+ * If so, then set a special (application-specific) value for the
+ * lead surrogate code _unit_, with utrie2_set32ForLeadSurrogateCodeUnit().
+ *
+ * At runtime, use UTRIE2_GET16_FROM_U16_SINGLE_LEAD() or
+ * UTRIE2_GET32_FROM_U16_SINGLE_LEAD() per code unit. If there is non-trivial
+ * data and the code unit is a lead surrogate, then check if a trail surrogate
+ * follows. If so, assemble the supplementary code point with
+ * U16_GET_SUPPLEMENTARY() and look up its value with UTRIE2_GET16_FROM_SUPP()
+ * or UTRIE2_GET32_FROM_SUPP(); otherwise reset the lead
+ * surrogate's value or do a code point lookup for it.
+ *
+ * If there is only trivial data for lead and trail surrogates, then processing
+ * can often skip them. For example, in normalization or case mapping
+ * all characters that do not have any mappings are simply copied as is.
+ */
+
+/**
+ * Get a value from a lead surrogate code unit as stored in the trie.
+ *
+ * @param trie the trie
+ * @param c the code unit (U+D800..U+DBFF)
+ * @return the value
+ */
+U_CAPI uint32_t U_EXPORT2
+utrie2_get32FromLeadSurrogateCodeUnit(const UTrie2 *trie, UChar32 c);
+
+/**
+ * Enumerate the trie values for the 1024=0x400 code points
+ * corresponding to a given lead surrogate.
+ * For example, for the lead surrogate U+D87E it will enumerate the values
+ * for [U+2F800..U+2FC00[.
+ * Used by data builder code that sets special lead surrogate code unit values
+ * for optimized UTF-16 string processing.
+ *
+ * Do not modify the trie during the enumeration.
+ *
+ * Except for the limited code point range, this functions just like utrie2_enum():
+ * For each entry in the trie, the value to be delivered is passed through
+ * the UTrie2EnumValue function.
+ * The value is unchanged if that function pointer is NULL.
+ *
+ * For each contiguous range of code points with a given (transformed) value,
+ * the UTrie2EnumRange function is called.
+ *
+ * @param trie a pointer to the trie
+ * @param enumValue a pointer to a function that may transform the trie entry value,
+ *                  or NULL if the values from the trie are to be used directly
+ * @param enumRange a pointer to a function that is called for each contiguous range
+ *                  of code points with the same (transformed) value
+ * @param context an opaque pointer that is passed on to the callback functions
+ */
+U_CAPI void U_EXPORT2
+utrie2_enumForLeadSurrogate(const UTrie2 *trie, UChar32 lead,
+                            UTrie2EnumValue *enumValue, UTrie2EnumRange *enumRange,
+                            const void *context);
+
+/**
+ * Set a value for a lead surrogate code unit.
+ *
+ * @param trie the unfrozen trie
+ * @param lead the lead surrogate code unit (U+D800..U+DBFF)
+ * @param value the value
+ * @param pErrorCode an in/out ICU UErrorCode; among other possible error codes:
+ * - U_NO_WRITE_PERMISSION if the trie is frozen
+ */
+U_CAPI void U_EXPORT2
+utrie2_set32ForLeadSurrogateCodeUnit(UTrie2 *trie,
+                                     UChar32 lead, uint32_t value,
+                                     UErrorCode *pErrorCode);
+
+/**
+ * Return a 16-bit trie value from a UTF-16 single/lead code unit (<=U+ffff).
+ * Same as UTRIE2_GET16() if c is a BMP code point except for lead surrogates,
+ * but smaller and faster.
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param c (UChar32, in) the input code unit, must be 0<=c<=U+ffff
+ * @return (uint16_t) The code unit's trie value.
+ */
+#define UTRIE2_GET16_FROM_U16_SINGLE_LEAD(trie, c) _UTRIE2_GET_FROM_U16_SINGLE_LEAD((trie), index, c)
+
+/**
+ * Return a 32-bit trie value from a UTF-16 single/lead code unit (<=U+ffff).
+ * Same as UTRIE2_GET32() if c is a BMP code point except for lead surrogates,
+ * but smaller and faster.
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param c (UChar32, in) the input code unit, must be 0<=c<=U+ffff
+ * @return (uint32_t) The code unit's trie value.
+ */
+#define UTRIE2_GET32_FROM_U16_SINGLE_LEAD(trie, c) _UTRIE2_GET_FROM_U16_SINGLE_LEAD((trie), data32, c)
+
+/**
+ * Return a 16-bit trie value from a supplementary code point (U+10000..U+10ffff).
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param c (UChar32, in) the input code point, must be U+10000<=c<=U+10ffff
+ * @return (uint16_t) The code point's trie value.
+ */
+#define UTRIE2_GET16_FROM_SUPP(trie, c) _UTRIE2_GET_FROM_SUPP((trie), index, c)
+
+/**
+ * Return a 32-bit trie value from a supplementary code point (U+10000..U+10ffff).
+ *
+ * @param trie (const UTrie2 *, in) a frozen trie
+ * @param c (UChar32, in) the input code point, must be U+10000<=c<=U+10ffff
+ * @return (uint32_t) The code point's trie value.
+ */
+#define UTRIE2_GET32_FROM_SUPP(trie, c) _UTRIE2_GET_FROM_SUPP((trie), data32, c)
+
+U_CDECL_END
+
+/* C++ convenience wrappers ------------------------------------------------- */
+
+#ifdef __cplusplus
+
+#include "unicode/uobject.h"
+#include "unicode/utf.h"
+
+U_NAMESPACE_BEGIN
+
+// Use the Forward/Backward subclasses below.
+class UTrie2StringIterator : public UMemory {
+public:
+    UTrie2StringIterator(const UTrie2 *t, const char16_t *p) :
+        trie(t), codePointStart(p), codePointLimit(p), codePoint(U_SENTINEL) {}
+
+    const UTrie2 *trie;
+    const char16_t *codePointStart, *codePointLimit;
+    UChar32 codePoint;
+};
+
+class BackwardUTrie2StringIterator : public UTrie2StringIterator {
+public:
+    BackwardUTrie2StringIterator(const UTrie2 *t, const char16_t *s, const char16_t *p) :
+        UTrie2StringIterator(t, p), start(s) {}
+
+    uint16_t previous16();
+
+    const char16_t *start;
+};
+
+class ForwardUTrie2StringIterator : public UTrie2StringIterator {
+public:
+    // Iteration limit l can be nullptr.
+    // In that case, the caller must detect c==0 and stop.
+    ForwardUTrie2StringIterator(const UTrie2 *t, const char16_t *p, const char16_t *l) :
+        UTrie2StringIterator(t, p), limit(l) {}
+
+    uint16_t next16();
+
+    const char16_t *limit;
+};
+
+U_NAMESPACE_END
+
+#endif
+
+/* Internal definitions ----------------------------------------------------- */
+
+U_CDECL_BEGIN
+
+/** Build-time trie structure. */
+struct UNewTrie2;
+typedef struct UNewTrie2 UNewTrie2;
+
+/*
+ * Trie structure definition.
+ *
+ * Either the data table is 16 bits wide and accessed via the index
+ * pointer, with each index item increased by indexLength;
+ * in this case, data32==NULL, and data16 is used for direct ASCII access.
+ *
+ * Or the data table is 32 bits wide and accessed via the data32 pointer.
+ */
+struct UTrie2 {
+    /* protected: used by macros and functions for reading values */
+    const uint16_t *index;
+    const uint16_t *data16;     /* for fast UTF-8 ASCII access, if 16b data */
+    const uint32_t *data32;     /* NULL if 16b data is used via index */
+
+    int32_t indexLength, dataLength;
+    uint16_t index2NullOffset;  /* 0xffff if there is no dedicated index-2 null block */
+    uint16_t dataNullOffset;
+    uint32_t initialValue;
+    /** Value returned for out-of-range code points and illegal UTF-8. */
+    uint32_t errorValue;
+
+    /* Start of the last range which ends at U+10ffff, and its value. */
+    UChar32 highStart;
+    int32_t highValueIndex;
+
+    /* private: used by builder and unserialization functions */
+    void *memory;           /* serialized bytes; NULL if not frozen yet */
+    int32_t length;         /* number of serialized bytes at memory; 0 if not frozen yet */
+    UBool isMemoryOwned;    /* true if the trie owns the memory */
+    UBool padding1;
+    int16_t padding2;
+    UNewTrie2 *newTrie;     /* builder object; NULL when frozen */
+
+#ifdef UTRIE2_DEBUG
+    const char *name;
+#endif
+};
+
+/**
+ * Trie constants, defining shift widths, index array lengths, etc.
+ *
+ * These are needed for the runtime macros but users can treat these as
+ * implementation details and skip to the actual public API further below.
+ */
+enum {
+    /** Shift size for getting the index-1 table offset. */
+    UTRIE2_SHIFT_1=6+5,
+
+    /** Shift size for getting the index-2 table offset. */
+    UTRIE2_SHIFT_2=5,
+
+    /**
+     * Difference between the two shift sizes,
+     * for getting an index-1 offset from an index-2 offset. 6=11-5
+     */
+    UTRIE2_SHIFT_1_2=UTRIE2_SHIFT_1-UTRIE2_SHIFT_2,
+
+    /**
+     * Number of index-1 entries for the BMP. 32=0x20
+     * This part of the index-1 table is omitted from the serialized form.
+     */
+    UTRIE2_OMITTED_BMP_INDEX_1_LENGTH=0x10000>>UTRIE2_SHIFT_1,
+
+    /** Number of code points per index-1 table entry. 2048=0x800 */
+    UTRIE2_CP_PER_INDEX_1_ENTRY=1<>UTRIE2_SHIFT_2.
+     */
+    UTRIE2_INDEX_2_OFFSET=0,
+
+    /**
+     * The part of the index-2 table for U+D800..U+DBFF stores values for
+     * lead surrogate code _units_ not code _points_.
+     * Values for lead surrogate code _points_ are indexed with this portion of the table.
+     * Length=32=0x20=0x400>>UTRIE2_SHIFT_2. (There are 1024=0x400 lead surrogates.)
+     */
+    UTRIE2_LSCP_INDEX_2_OFFSET=0x10000>>UTRIE2_SHIFT_2,
+    UTRIE2_LSCP_INDEX_2_LENGTH=0x400>>UTRIE2_SHIFT_2,
+
+    /** Count the lengths of both BMP pieces. 2080=0x820 */
+    UTRIE2_INDEX_2_BMP_LENGTH=UTRIE2_LSCP_INDEX_2_OFFSET+UTRIE2_LSCP_INDEX_2_LENGTH,
+
+    /**
+     * The 2-byte UTF-8 version of the index-2 table follows at offset 2080=0x820.
+     * Length 32=0x20 for lead bytes C0..DF, regardless of UTRIE2_SHIFT_2.
+     */
+    UTRIE2_UTF8_2B_INDEX_2_OFFSET=UTRIE2_INDEX_2_BMP_LENGTH,
+    UTRIE2_UTF8_2B_INDEX_2_LENGTH=0x800>>6,  /* U+0800 is the first code point after 2-byte UTF-8 */
+
+    /**
+     * The index-1 table, only used for supplementary code points, at offset 2112=0x840.
+     * Variable length, for code points up to highStart, where the last single-value range starts.
+     * Maximum length 512=0x200=0x100000>>UTRIE2_SHIFT_1.
+     * (For 0x100000 supplementary code points U+10000..U+10ffff.)
+     *
+     * The part of the index-2 table for supplementary code points starts
+     * after this index-1 table.
+     *
+     * Both the index-1 table and the following part of the index-2 table
+     * are omitted completely if there is only BMP data.
+     */
+    UTRIE2_INDEX_1_OFFSET=UTRIE2_UTF8_2B_INDEX_2_OFFSET+UTRIE2_UTF8_2B_INDEX_2_LENGTH,
+    UTRIE2_MAX_INDEX_1_LENGTH=0x100000>>UTRIE2_SHIFT_1,
+
+    /*
+     * Fixed layout of the first part of the data array. -----------------------
+     * Starts with 4 blocks (128=0x80 entries) for ASCII.
+     */
+
+    /**
+     * The illegal-UTF-8 data block follows the ASCII block, at offset 128=0x80.
+     * Used with linear access for single bytes 0..0xbf for simple error handling.
+     * Length 64=0x40, not UTRIE2_DATA_BLOCK_LENGTH.
+     */
+    UTRIE2_BAD_UTF8_DATA_OFFSET=0x80,
+
+    /** The start of non-linear-ASCII data blocks, at offset 192=0xc0. */
+    UTRIE2_DATA_START_OFFSET=0xc0
+};
+
+/* Internal functions and macros -------------------------------------------- */
+
+/**
+ * Internal function for part of the UTRIE2_U8_NEXTxx() macro implementations.
+ * Do not call directly.
+ * @internal
+ */
+U_CAPI int32_t U_EXPORT2
+utrie2_internalU8NextIndex(const UTrie2 *trie, UChar32 c,
+                           const uint8_t *src, const uint8_t *limit);
+
+/**
+ * Internal function for part of the UTRIE2_U8_PREVxx() macro implementations.
+ * Do not call directly.
+ * @internal
+ */
+U_CAPI int32_t U_EXPORT2
+utrie2_internalU8PrevIndex(const UTrie2 *trie, UChar32 c,
+                           const uint8_t *start, const uint8_t *src);
+
+
+/** Internal low-level trie getter. Returns a data index. */
+#define _UTRIE2_INDEX_RAW(offset, trieIndex, c) \
+    (((int32_t)((trieIndex)[(offset)+((c)>>UTRIE2_SHIFT_2)]) \
+    <>UTRIE2_SHIFT_2), trieIndex, c)
+
+/** Internal trie getter from a BMP code point. Returns the data index. */
+#define _UTRIE2_INDEX_FROM_BMP(trieIndex, c) \
+    _UTRIE2_INDEX_RAW(U_IS_LEAD(c) ? UTRIE2_LSCP_INDEX_2_OFFSET-(0xd800>>UTRIE2_SHIFT_2) : 0, \
+                      trieIndex, c)
+
+/** Internal trie getter from a supplementary code point below highStart. Returns the data index. */
+#define _UTRIE2_INDEX_FROM_SUPP(trieIndex, c) \
+    (((int32_t)((trieIndex)[ \
+        (trieIndex)[(UTRIE2_INDEX_1_OFFSET-UTRIE2_OMITTED_BMP_INDEX_1_LENGTH)+ \
+                      ((c)>>UTRIE2_SHIFT_1)]+ \
+        (((c)>>UTRIE2_SHIFT_2)&UTRIE2_INDEX_2_MASK)]) \
+    <index, c) : \
+        (uint32_t)(c)<=0xffff ? \
+            _UTRIE2_INDEX_RAW( \
+                (c)<=0xdbff ? UTRIE2_LSCP_INDEX_2_OFFSET-(0xd800>>UTRIE2_SHIFT_2) : 0, \
+                (trie)->index, c) : \
+            (uint32_t)(c)>0x10ffff ? \
+                (asciiOffset)+UTRIE2_BAD_UTF8_DATA_OFFSET : \
+                (c)>=(trie)->highStart ? \
+                    (trie)->highValueIndex : \
+                    _UTRIE2_INDEX_FROM_SUPP((trie)->index, c))
+
+/** Internal trie getter from a UTF-16 single/lead code unit. Returns the data. */
+#define _UTRIE2_GET_FROM_U16_SINGLE_LEAD(trie, data, c) \
+    (trie)->data[_UTRIE2_INDEX_FROM_U16_SINGLE_LEAD((trie)->index, c)]
+
+/** Internal trie getter from a supplementary code point. Returns the data. */
+#define _UTRIE2_GET_FROM_SUPP(trie, data, c) \
+    (trie)->data[(c)>=(trie)->highStart ? (trie)->highValueIndex : \
+                 _UTRIE2_INDEX_FROM_SUPP((trie)->index, c)]
+
+/**
+ * Internal trie getter from a code point, with checking that c is in 0..10FFFF.
+ * Returns the data.
+ */
+#define _UTRIE2_GET(trie, data, asciiOffset, c) \
+    (trie)->data[_UTRIE2_INDEX_FROM_CP(trie, asciiOffset, c)]
+
+/** Internal next-post-increment: get the next code point (c) and its data. */
+#define _UTRIE2_U16_NEXT(trie, data, src, limit, c, result) UPRV_BLOCK_MACRO_BEGIN { \
+    { \
+        uint16_t __c2; \
+        (c)=*(src)++; \
+        if(!U16_IS_LEAD(c)) { \
+            (result)=_UTRIE2_GET_FROM_U16_SINGLE_LEAD(trie, data, c); \
+        } else if((src)==(limit) || !U16_IS_TRAIL(__c2=*(src))) { \
+            (result)=(trie)->data[_UTRIE2_INDEX_FROM_LSCP((trie)->index, c)]; \
+        } else { \
+            ++(src); \
+            (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
+            (result)=_UTRIE2_GET_FROM_SUPP((trie), data, (c)); \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** Internal pre-decrement-previous: get the previous code point (c) and its data */
+#define _UTRIE2_U16_PREV(trie, data, start, src, c, result) UPRV_BLOCK_MACRO_BEGIN { \
+    { \
+        uint16_t __c2; \
+        (c)=*--(src); \
+        if(!U16_IS_TRAIL(c) || (src)==(start) || !U16_IS_LEAD(__c2=*((src)-1))) { \
+            (result)=(trie)->data[_UTRIE2_INDEX_FROM_BMP((trie)->index, c)]; \
+        } else { \
+            --(src); \
+            (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
+            (result)=_UTRIE2_GET_FROM_SUPP((trie), data, (c)); \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** Internal UTF-8 next-post-increment: get the next code point's data. */
+#define _UTRIE2_U8_NEXT(trie, ascii, data, src, limit, result) UPRV_BLOCK_MACRO_BEGIN { \
+    uint8_t __lead=(uint8_t)*(src)++; \
+    if(U8_IS_SINGLE(__lead)) { \
+        (result)=(trie)->ascii[__lead]; \
+    } else { \
+        uint8_t __t1, __t2; \
+        if( /* handle U+0800..U+FFFF inline */ \
+            0xe0<=__lead && __lead<0xf0 && ((src)+1)<(limit) && \
+            U8_IS_VALID_LEAD3_AND_T1(__lead, __t1=(uint8_t)*(src)) && \
+            (__t2=(uint8_t)(*((src)+1)-0x80))<= 0x3f \
+        ) { \
+            (src)+=2; \
+            (result)=(trie)->data[ \
+                ((int32_t)((trie)->index[((__lead-0xe0)<<(12-UTRIE2_SHIFT_2))+ \
+                                         ((__t1&0x3f)<<(6-UTRIE2_SHIFT_2))+(__t2>>UTRIE2_SHIFT_2)]) \
+                <=0xc2 && (src)<(limit) && \
+            (__t1=(uint8_t)(*(src)-0x80))<=0x3f \
+        ) { \
+            ++(src); \
+            (result)=(trie)->data[ \
+                (trie)->index[(UTRIE2_UTF8_2B_INDEX_2_OFFSET-0xc0)+__lead]+ \
+                __t1]; \
+        } else { \
+            int32_t __index=utrie2_internalU8NextIndex((trie), __lead, (const uint8_t *)(src), \
+                                                                       (const uint8_t *)(limit)); \
+            (src)+=__index&7; \
+            (result)=(trie)->data[__index>>3]; \
+        } \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+/** Internal UTF-8 pre-decrement-previous: get the previous code point's data. */
+#define _UTRIE2_U8_PREV(trie, ascii, data, start, src, result) UPRV_BLOCK_MACRO_BEGIN { \
+    uint8_t __b=(uint8_t)*--(src); \
+    if(U8_IS_SINGLE(__b)) { \
+        (result)=(trie)->ascii[__b]; \
+    } else { \
+        int32_t __index=utrie2_internalU8PrevIndex((trie), __b, (const uint8_t *)(start), \
+                                                                (const uint8_t *)(src)); \
+        (src)-=__index&7; \
+        (result)=(trie)->data[__index>>3]; \
+    } \
+} UPRV_BLOCK_MACRO_END
+
+U_CDECL_END
+
+#endif
diff --git a/intl/icu/source/common/utrie2_builder.cpp b/intl/icu/source/common/utrie2_builder.cpp
new file mode 100644
index 0000000000..1254e8cba6
--- /dev/null
+++ b/intl/icu/source/common/utrie2_builder.cpp
@@ -0,0 +1,1483 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+*
+*   Copyright (C) 2001-2014, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+******************************************************************************
+*   file name:  utrie2_builder.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2008sep26 (split off from utrie2.c)
+*   created by: Markus W. Scherer
+*
+*   This is a common implementation of a Unicode trie.
+*   It is a kind of compressed, serializable table of 16- or 32-bit values associated with
+*   Unicode code points (0..0x10ffff).
+*   This is the second common version of a Unicode trie (hence the name UTrie2).
+*   See utrie2.h for a comparison.
+*
+*   This file contains only the builder code.
+*   See utrie2.c for the runtime and enumeration code.
+*/
+// #define UTRIE2_DEBUG
+#ifdef UTRIE2_DEBUG
+#   include 
+#endif
+// #define UCPTRIE_DEBUG
+
+#include "unicode/utypes.h"
+#ifdef UCPTRIE_DEBUG
+#include "unicode/ucptrie.h"
+#include "unicode/umutablecptrie.h"
+#include "ucptrie_impl.h"
+#endif
+#include "cmemory.h"
+#include "utrie2.h"
+#include "utrie2_impl.h"
+
+#include "utrie.h"  // for utrie2_fromUTrie()
+
+/* Implementation notes ----------------------------------------------------- */
+
+/*
+ * The UTRIE2_SHIFT_1, UTRIE2_SHIFT_2, UTRIE2_INDEX_SHIFT and other values
+ * have been chosen to minimize trie sizes overall.
+ * Most of the code is flexible enough to work with a range of values,
+ * within certain limits.
+ *
+ * Exception: Support for separate values for lead surrogate code _units_
+ * vs. code _points_ was added after the constants were fixed,
+ * and has not been tested nor particularly designed for different constant values.
+ * (Especially the utrie2_enum() code that jumps to the special LSCP index-2
+ * part and back.)
+ *
+ * Requires UTRIE2_SHIFT_2<=6. Otherwise 0xc0 which is the top of the ASCII-linear data
+ * including the bad-UTF-8-data block is not a multiple of UTRIE2_DATA_BLOCK_LENGTH
+ * and map[block>>UTRIE2_SHIFT_2] (used in reference counting and compaction
+ * remapping) stops working.
+ *
+ * Requires UTRIE2_SHIFT_1>=10 because utrie2_enumForLeadSurrogate()
+ * assumes that a single index-2 block is used for 0x400 code points
+ * corresponding to one lead surrogate.
+ *
+ * Requires UTRIE2_SHIFT_1<=16. Otherwise one single index-2 block contains
+ * more than one Unicode plane, and the split of the index-2 table into a BMP
+ * part and a supplementary part, with a gap in between, would not work.
+ *
+ * Requires UTRIE2_INDEX_SHIFT>=1 not because of the code but because
+ * there is data with more than 64k distinct values,
+ * for example for Unihan collation with a separate collation weight per
+ * Han character.
+ */
+
+/* Building a trie ----------------------------------------------------------*/
+
+enum {
+    /** The null index-2 block, following the gap in the index-2 table. */
+    UNEWTRIE2_INDEX_2_NULL_OFFSET=UNEWTRIE2_INDEX_GAP_OFFSET+UNEWTRIE2_INDEX_GAP_LENGTH,
+
+    /** The start of allocated index-2 blocks. */
+    UNEWTRIE2_INDEX_2_START_OFFSET=UNEWTRIE2_INDEX_2_NULL_OFFSET+UTRIE2_INDEX_2_BLOCK_LENGTH,
+
+    /**
+     * The null data block.
+     * Length 64=0x40 even if UTRIE2_DATA_BLOCK_LENGTH is smaller,
+     * to work with 6-bit trail bytes from 2-byte UTF-8.
+     */
+    UNEWTRIE2_DATA_NULL_OFFSET=UTRIE2_DATA_START_OFFSET,
+
+    /** The start of allocated data blocks. */
+    UNEWTRIE2_DATA_START_OFFSET=UNEWTRIE2_DATA_NULL_OFFSET+0x40,
+
+    /**
+     * The start of data blocks for U+0800 and above.
+     * Below, compaction uses a block length of 64 for 2-byte UTF-8.
+     * From here on, compaction uses UTRIE2_DATA_BLOCK_LENGTH.
+     * Data values for 0x780 code points beyond ASCII.
+     */
+    UNEWTRIE2_DATA_0800_OFFSET=UNEWTRIE2_DATA_START_OFFSET+0x780
+};
+
+/* Start with allocation of 16k data entries. */
+#define UNEWTRIE2_INITIAL_DATA_LENGTH ((int32_t)1<<14)
+
+/* Grow about 8x each time. */
+#define UNEWTRIE2_MEDIUM_DATA_LENGTH ((int32_t)1<<17)
+
+static int32_t
+allocIndex2Block(UNewTrie2 *trie);
+
+U_CAPI UTrie2 * U_EXPORT2
+utrie2_open(uint32_t initialValue, uint32_t errorValue, UErrorCode *pErrorCode) {
+    UTrie2 *trie;
+    UNewTrie2 *newTrie;
+    uint32_t *data;
+    int32_t i, j;
+
+    if(U_FAILURE(*pErrorCode)) {
+        return nullptr;
+    }
+
+    trie=(UTrie2 *)uprv_malloc(sizeof(UTrie2));
+    newTrie=(UNewTrie2 *)uprv_malloc(sizeof(UNewTrie2));
+    data=(uint32_t *)uprv_malloc(UNEWTRIE2_INITIAL_DATA_LENGTH*4);
+    if(trie==nullptr || newTrie==nullptr || data==nullptr) {
+        uprv_free(trie);
+        uprv_free(newTrie);
+        uprv_free(data);
+        *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+        return 0;
+    }
+
+    uprv_memset(trie, 0, sizeof(UTrie2));
+    trie->initialValue=initialValue;
+    trie->errorValue=errorValue;
+    trie->highStart=0x110000;
+    trie->newTrie=newTrie;
+#ifdef UTRIE2_DEBUG
+    trie->name="open";
+#endif
+
+    newTrie->data=data;
+#ifdef UCPTRIE_DEBUG
+    newTrie->t3=umutablecptrie_open(initialValue, errorValue, pErrorCode);
+#endif
+    newTrie->dataCapacity=UNEWTRIE2_INITIAL_DATA_LENGTH;
+    newTrie->initialValue=initialValue;
+    newTrie->errorValue=errorValue;
+    newTrie->highStart=0x110000;
+    newTrie->firstFreeBlock=0;  /* no free block in the list */
+    newTrie->isCompacted=false;
+
+    /*
+     * preallocate and reset
+     * - ASCII
+     * - the bad-UTF-8-data block
+     * - the null data block
+     */
+    for(i=0; i<0x80; ++i) {
+        newTrie->data[i]=initialValue;
+    }
+    for(; i<0xc0; ++i) {
+        newTrie->data[i]=errorValue;
+    }
+    for(i=UNEWTRIE2_DATA_NULL_OFFSET; idata[i]=initialValue;
+    }
+    newTrie->dataNullOffset=UNEWTRIE2_DATA_NULL_OFFSET;
+    newTrie->dataLength=UNEWTRIE2_DATA_START_OFFSET;
+
+    /* set the index-2 indexes for the 2=0x80>>UTRIE2_SHIFT_2 ASCII data blocks */
+    for(i=0, j=0; j<0x80; ++i, j+=UTRIE2_DATA_BLOCK_LENGTH) {
+        newTrie->index2[i]=j;
+        newTrie->map[i]=1;
+    }
+    /* reference counts for the bad-UTF-8-data block */
+    for(; j<0xc0; ++i, j+=UTRIE2_DATA_BLOCK_LENGTH) {
+        newTrie->map[i]=0;
+    }
+    /*
+     * Reference counts for the null data block: all blocks except for the ASCII blocks.
+     * Plus 1 so that we don't drop this block during compaction.
+     * Plus as many as needed for lead surrogate code points.
+     */
+    /* i==newTrie->dataNullOffset */
+    newTrie->map[i++]=
+        (0x110000>>UTRIE2_SHIFT_2)-
+        (0x80>>UTRIE2_SHIFT_2)+
+        1+
+        UTRIE2_LSCP_INDEX_2_LENGTH;
+    j+=UTRIE2_DATA_BLOCK_LENGTH;
+    for(; jmap[i]=0;
+    }
+
+    /*
+     * set the remaining indexes in the BMP index-2 block
+     * to the null data block
+     */
+    for(i=0x80>>UTRIE2_SHIFT_2; iindex2[i]=UNEWTRIE2_DATA_NULL_OFFSET;
+    }
+
+    /*
+     * Fill the index gap with impossible values so that compaction
+     * does not overlap other index-2 blocks with the gap.
+     */
+    for(i=0; iindex2[UNEWTRIE2_INDEX_GAP_OFFSET+i]=-1;
+    }
+
+    /* set the indexes in the null index-2 block */
+    for(i=0; iindex2[UNEWTRIE2_INDEX_2_NULL_OFFSET+i]=UNEWTRIE2_DATA_NULL_OFFSET;
+    }
+    newTrie->index2NullOffset=UNEWTRIE2_INDEX_2_NULL_OFFSET;
+    newTrie->index2Length=UNEWTRIE2_INDEX_2_START_OFFSET;
+
+    /* set the index-1 indexes for the linear index-2 block */
+    for(i=0, j=0;
+        iindex1[i]=j;
+    }
+
+    /* set the remaining index-1 indexes to the null index-2 block */
+    for(; iindex1[i]=UNEWTRIE2_INDEX_2_NULL_OFFSET;
+    }
+
+    /*
+     * Preallocate and reset data for U+0080..U+07ff,
+     * for 2-byte UTF-8 which will be compacted in 64-blocks
+     * even if UTRIE2_DATA_BLOCK_LENGTH is smaller.
+     */
+    for(i=0x80; i<0x800; i+=UTRIE2_DATA_BLOCK_LENGTH) {
+        utrie2_set32(trie, i, initialValue, pErrorCode);
+    }
+
+    return trie;
+}
+
+static UNewTrie2 *
+cloneBuilder(const UNewTrie2 *other) {
+    UNewTrie2 *trie;
+
+    trie=(UNewTrie2 *)uprv_malloc(sizeof(UNewTrie2));
+    if(trie==nullptr) {
+        return nullptr;
+    }
+
+    trie->data=(uint32_t *)uprv_malloc(other->dataCapacity*4);
+    if(trie->data==nullptr) {
+        uprv_free(trie);
+        return nullptr;
+    }
+#ifdef UCPTRIE_DEBUG
+    if(other->t3==nullptr) {
+        trie->t3=nullptr;
+    } else {
+        UErrorCode errorCode=U_ZERO_ERROR;
+        trie->t3=umutablecptrie_clone(other->t3, &errorCode);
+    }
+#endif
+    trie->dataCapacity=other->dataCapacity;
+
+    /* clone data */
+    uprv_memcpy(trie->index1, other->index1, sizeof(trie->index1));
+    uprv_memcpy(trie->index2, other->index2, (size_t)other->index2Length*4);
+    trie->index2NullOffset=other->index2NullOffset;
+    trie->index2Length=other->index2Length;
+
+    uprv_memcpy(trie->data, other->data, (size_t)other->dataLength*4);
+    trie->dataNullOffset=other->dataNullOffset;
+    trie->dataLength=other->dataLength;
+
+    /* reference counters */
+    if(other->isCompacted) {
+        trie->firstFreeBlock=0;
+    } else {
+        uprv_memcpy(trie->map, other->map, ((size_t)other->dataLength>>UTRIE2_SHIFT_2)*4);
+        trie->firstFreeBlock=other->firstFreeBlock;
+    }
+
+    trie->initialValue=other->initialValue;
+    trie->errorValue=other->errorValue;
+    trie->highStart=other->highStart;
+    trie->isCompacted=other->isCompacted;
+
+    return trie;
+}
+
+U_CAPI UTrie2 * U_EXPORT2
+utrie2_clone(const UTrie2 *other, UErrorCode *pErrorCode) {
+    UTrie2 *trie;
+
+    if(U_FAILURE(*pErrorCode)) {
+        return nullptr;
+    }
+    if(other==nullptr || (other->memory==nullptr && other->newTrie==nullptr)) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+
+    trie=(UTrie2 *)uprv_malloc(sizeof(UTrie2));
+    if(trie==nullptr) {
+        *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+        return nullptr;
+    }
+    uprv_memcpy(trie, other, sizeof(UTrie2));
+
+    if(other->memory!=nullptr) {
+        trie->memory=uprv_malloc(other->length);
+        if(trie->memory!=nullptr) {
+            trie->isMemoryOwned=true;
+            uprv_memcpy(trie->memory, other->memory, other->length);
+
+            /* make the clone's pointers point to its own memory */
+            trie->index=(uint16_t *)trie->memory+(other->index-(uint16_t *)other->memory);
+            if(other->data16!=nullptr) {
+                trie->data16=(uint16_t *)trie->memory+(other->data16-(uint16_t *)other->memory);
+            }
+            if(other->data32!=nullptr) {
+                trie->data32=(uint32_t *)trie->memory+(other->data32-(uint32_t *)other->memory);
+            }
+        }
+    } else /* other->newTrie!=nullptr */ {
+        trie->newTrie=cloneBuilder(other->newTrie);
+    }
+
+    if(trie->memory==nullptr && trie->newTrie==nullptr) {
+        *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+        uprv_free(trie);
+        trie=nullptr;
+    }
+    return trie;
+}
+
+typedef struct NewTrieAndStatus {
+    UTrie2 *trie;
+    UErrorCode errorCode;
+    UBool exclusiveLimit;  /* rather than inclusive range end */
+} NewTrieAndStatus;
+
+static UBool U_CALLCONV
+copyEnumRange(const void *context, UChar32 start, UChar32 end, uint32_t value) {
+    NewTrieAndStatus *nt=(NewTrieAndStatus *)context;
+    if(value!=nt->trie->initialValue) {
+        if(nt->exclusiveLimit) {
+            --end;
+        }
+        if(start==end) {
+            utrie2_set32(nt->trie, start, value, &nt->errorCode);
+        } else {
+            utrie2_setRange32(nt->trie, start, end, value, true, &nt->errorCode);
+        }
+        return U_SUCCESS(nt->errorCode);
+    } else {
+        return true;
+    }
+}
+
+#ifdef UTRIE2_DEBUG
+static long countInitial(const UTrie2 *trie) {
+    uint32_t initialValue=trie->initialValue;
+    int32_t length=trie->dataLength;
+    long count=0;
+    if(trie->data16!=nullptr) {
+        for(int32_t i=0; idata16[i]==initialValue) { ++count; }
+        }
+    } else {
+        for(int32_t i=0; idata32[i]==initialValue) { ++count; }
+        }
+    }
+    return count;
+}
+
+static void
+utrie_printLengths(const UTrie *trie) {
+    long indexLength=trie->indexLength;
+    long dataLength=(long)trie->dataLength;
+    long totalLength=(long)sizeof(UTrieHeader)+indexLength*2+dataLength*(trie->data32!=nullptr ? 4 : 2);
+    printf("**UTrieLengths** index:%6ld  data:%6ld  serialized:%6ld\n",
+           indexLength, dataLength, totalLength);
+}
+
+static void
+utrie2_printLengths(const UTrie2 *trie, const char *which) {
+    long indexLength=trie->indexLength;
+    long dataLength=(long)trie->dataLength;
+    long totalLength=(long)sizeof(UTrie2Header)+indexLength*2+dataLength*(trie->data32!=nullptr ? 4 : 2);
+    printf("**UTrie2Lengths(%s %s)** index:%6ld  data:%6ld  countInitial:%6ld  serialized:%6ld\n",
+           which, trie->name, indexLength, dataLength, countInitial(trie), totalLength);
+}
+#endif
+
+U_CAPI UTrie2 * U_EXPORT2
+utrie2_cloneAsThawed(const UTrie2 *other, UErrorCode *pErrorCode) {
+    NewTrieAndStatus context;
+    char16_t lead;
+
+    if(U_FAILURE(*pErrorCode)) {
+        return nullptr;
+    }
+    if(other==nullptr || (other->memory==nullptr && other->newTrie==nullptr)) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+    if(other->newTrie!=nullptr && !other->newTrie->isCompacted) {
+        return utrie2_clone(other, pErrorCode);  /* clone an unfrozen trie */
+    }
+
+    /* Clone the frozen trie by enumerating it and building a new one. */
+    context.trie=utrie2_open(other->initialValue, other->errorValue, pErrorCode);
+    if(U_FAILURE(*pErrorCode)) {
+        return nullptr;
+    }
+    context.exclusiveLimit=false;
+    context.errorCode=*pErrorCode;
+    utrie2_enum(other, nullptr, copyEnumRange, &context);
+    *pErrorCode=context.errorCode;
+    for(lead=0xd800; lead<0xdc00; ++lead) {
+        uint32_t value;
+        if(other->data32==nullptr) {
+            value=UTRIE2_GET16_FROM_U16_SINGLE_LEAD(other, lead);
+        } else {
+            value=UTRIE2_GET32_FROM_U16_SINGLE_LEAD(other, lead);
+        }
+        if(value!=other->initialValue) {
+            utrie2_set32ForLeadSurrogateCodeUnit(context.trie, lead, value, pErrorCode);
+        }
+    }
+    if(U_FAILURE(*pErrorCode)) {
+        utrie2_close(context.trie);
+        context.trie=nullptr;
+    }
+    return context.trie;
+}
+
+/* Almost the same as utrie2_cloneAsThawed() but copies a UTrie and freezes the clone. */
+U_CAPI UTrie2 * U_EXPORT2
+utrie2_fromUTrie(const UTrie *trie1, uint32_t errorValue, UErrorCode *pErrorCode) {
+    NewTrieAndStatus context;
+    char16_t lead;
+
+    if(U_FAILURE(*pErrorCode)) {
+        return nullptr;
+    }
+    if(trie1==nullptr) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return nullptr;
+    }
+    context.trie=utrie2_open(trie1->initialValue, errorValue, pErrorCode);
+    if(U_FAILURE(*pErrorCode)) {
+        return nullptr;
+    }
+    context.exclusiveLimit=true;
+    context.errorCode=*pErrorCode;
+    utrie_enum(trie1, nullptr, copyEnumRange, &context);
+    *pErrorCode=context.errorCode;
+    for(lead=0xd800; lead<0xdc00; ++lead) {
+        uint32_t value;
+        if(trie1->data32==nullptr) {
+            value=UTRIE_GET16_FROM_LEAD(trie1, lead);
+        } else {
+            value=UTRIE_GET32_FROM_LEAD(trie1, lead);
+        }
+        if(value!=trie1->initialValue) {
+            utrie2_set32ForLeadSurrogateCodeUnit(context.trie, lead, value, pErrorCode);
+        }
+    }
+    if(U_SUCCESS(*pErrorCode)) {
+        utrie2_freeze(context.trie,
+                      trie1->data32!=nullptr ? UTRIE2_32_VALUE_BITS : UTRIE2_16_VALUE_BITS,
+                      pErrorCode);
+    }
+#ifdef UTRIE2_DEBUG
+    if(U_SUCCESS(*pErrorCode)) {
+        utrie_printLengths(trie1);
+        utrie2_printLengths(context.trie, "fromUTrie");
+    }
+#endif
+    if(U_FAILURE(*pErrorCode)) {
+        utrie2_close(context.trie);
+        context.trie=nullptr;
+    }
+    return context.trie;
+}
+
+static inline UBool
+isInNullBlock(UNewTrie2 *trie, UChar32 c, UBool forLSCP) {
+    int32_t i2, block;
+
+    if(U_IS_LEAD(c) && forLSCP) {
+        i2=(UTRIE2_LSCP_INDEX_2_OFFSET-(0xd800>>UTRIE2_SHIFT_2))+
+            (c>>UTRIE2_SHIFT_2);
+    } else {
+        i2=trie->index1[c>>UTRIE2_SHIFT_1]+
+            ((c>>UTRIE2_SHIFT_2)&UTRIE2_INDEX_2_MASK);
+    }
+    block=trie->index2[i2];
+    return (UBool)(block==trie->dataNullOffset);
+}
+
+static int32_t
+allocIndex2Block(UNewTrie2 *trie) {
+    int32_t newBlock, newTop;
+
+    newBlock=trie->index2Length;
+    newTop=newBlock+UTRIE2_INDEX_2_BLOCK_LENGTH;
+    if(newTop>UPRV_LENGTHOF(trie->index2)) {
+        /*
+         * Should never occur.
+         * Either UTRIE2_MAX_BUILD_TIME_INDEX_LENGTH is incorrect,
+         * or the code writes more values than should be possible.
+         */
+        return -1;
+    }
+    trie->index2Length=newTop;
+    uprv_memcpy(trie->index2+newBlock, trie->index2+trie->index2NullOffset, UTRIE2_INDEX_2_BLOCK_LENGTH*4);
+    return newBlock;
+}
+
+static int32_t
+getIndex2Block(UNewTrie2 *trie, UChar32 c, UBool forLSCP) {
+    int32_t i1, i2;
+
+    if(U_IS_LEAD(c) && forLSCP) {
+        return UTRIE2_LSCP_INDEX_2_OFFSET;
+    }
+
+    i1=c>>UTRIE2_SHIFT_1;
+    i2=trie->index1[i1];
+    if(i2==trie->index2NullOffset) {
+        i2=allocIndex2Block(trie);
+        if(i2<0) {
+            return -1;  /* program error */
+        }
+        trie->index1[i1]=i2;
+    }
+    return i2;
+}
+
+static int32_t
+allocDataBlock(UNewTrie2 *trie, int32_t copyBlock) {
+    int32_t newBlock, newTop;
+
+    if(trie->firstFreeBlock!=0) {
+        /* get the first free block */
+        newBlock=trie->firstFreeBlock;
+        trie->firstFreeBlock=-trie->map[newBlock>>UTRIE2_SHIFT_2];
+    } else {
+        /* get a new block from the high end */
+        newBlock=trie->dataLength;
+        newTop=newBlock+UTRIE2_DATA_BLOCK_LENGTH;
+        if(newTop>trie->dataCapacity) {
+            /* out of memory in the data array */
+            int32_t capacity;
+            uint32_t *data;
+
+            if(trie->dataCapacitydataCapacitydata, (size_t)trie->dataLength*4);
+            uprv_free(trie->data);
+            trie->data=data;
+            trie->dataCapacity=capacity;
+        }
+        trie->dataLength=newTop;
+    }
+    uprv_memcpy(trie->data+newBlock, trie->data+copyBlock, UTRIE2_DATA_BLOCK_LENGTH*4);
+    trie->map[newBlock>>UTRIE2_SHIFT_2]=0;
+    return newBlock;
+}
+
+/* call when the block's reference counter reaches 0 */
+static void
+releaseDataBlock(UNewTrie2 *trie, int32_t block) {
+    /* put this block at the front of the free-block chain */
+    trie->map[block>>UTRIE2_SHIFT_2]=-trie->firstFreeBlock;
+    trie->firstFreeBlock=block;
+}
+
+static inline UBool
+isWritableBlock(UNewTrie2 *trie, int32_t block) {
+    return (UBool)(block!=trie->dataNullOffset && 1==trie->map[block>>UTRIE2_SHIFT_2]);
+}
+
+static inline void
+setIndex2Entry(UNewTrie2 *trie, int32_t i2, int32_t block) {
+    int32_t oldBlock;
+    ++trie->map[block>>UTRIE2_SHIFT_2];  /* increment first, in case block==oldBlock! */
+    oldBlock=trie->index2[i2];
+    if(0 == --trie->map[oldBlock>>UTRIE2_SHIFT_2]) {
+        releaseDataBlock(trie, oldBlock);
+    }
+    trie->index2[i2]=block;
+}
+
+/**
+ * No error checking for illegal arguments.
+ *
+ * @return -1 if no new data block available (out of memory in data array)
+ * @internal
+ */
+static int32_t
+getDataBlock(UNewTrie2 *trie, UChar32 c, UBool forLSCP) {
+    int32_t i2, oldBlock, newBlock;
+
+    i2=getIndex2Block(trie, c, forLSCP);
+    if(i2<0) {
+        return -1;  /* program error */
+    }
+
+    i2+=(c>>UTRIE2_SHIFT_2)&UTRIE2_INDEX_2_MASK;
+    oldBlock=trie->index2[i2];
+    if(isWritableBlock(trie, oldBlock)) {
+        return oldBlock;
+    }
+
+    /* allocate a new data block */
+    newBlock=allocDataBlock(trie, oldBlock);
+    if(newBlock<0) {
+        /* out of memory in the data array */
+        return -1;
+    }
+    setIndex2Entry(trie, i2, newBlock);
+    return newBlock;
+}
+
+/**
+ * @return true if the value was successfully set
+ */
+static void
+set32(UNewTrie2 *trie,
+      UChar32 c, UBool forLSCP, uint32_t value,
+      UErrorCode *pErrorCode) {
+    int32_t block;
+
+    if(trie==nullptr || trie->isCompacted) {
+        *pErrorCode=U_NO_WRITE_PERMISSION;
+        return;
+    }
+#ifdef UCPTRIE_DEBUG
+    umutablecptrie_set(trie->t3, c, value, pErrorCode);
+#endif
+
+    block=getDataBlock(trie, c, forLSCP);
+    if(block<0) {
+        *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+        return;
+    }
+
+    trie->data[block+(c&UTRIE2_DATA_MASK)]=value;
+}
+
+U_CAPI void U_EXPORT2
+utrie2_set32(UTrie2 *trie, UChar32 c, uint32_t value, UErrorCode *pErrorCode) {
+    if(U_FAILURE(*pErrorCode)) {
+        return;
+    }
+    if((uint32_t)c>0x10ffff) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+    set32(trie->newTrie, c, true, value, pErrorCode);
+}
+
+U_CAPI void U_EXPORT2
+utrie2_set32ForLeadSurrogateCodeUnit(UTrie2 *trie,
+                                     UChar32 c, uint32_t value,
+                                     UErrorCode *pErrorCode) {
+    if(U_FAILURE(*pErrorCode)) {
+        return;
+    }
+    if(!U_IS_LEAD(c)) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+    set32(trie->newTrie, c, false, value, pErrorCode);
+}
+
+static void
+writeBlock(uint32_t *block, uint32_t value) {
+    uint32_t *limit=block+UTRIE2_DATA_BLOCK_LENGTH;
+    while(block0x10ffff || (uint32_t)end>0x10ffff || start>end) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+    newTrie=trie->newTrie;
+    if(newTrie==nullptr || newTrie->isCompacted) {
+        *pErrorCode=U_NO_WRITE_PERMISSION;
+        return;
+    }
+#ifdef UCPTRIE_DEBUG
+    umutablecptrie_setRange(newTrie->t3, start, end, value, pErrorCode);
+#endif
+    if(!overwrite && value==newTrie->initialValue) {
+        return; /* nothing to do */
+    }
+
+    limit=end+1;
+    if(start&UTRIE2_DATA_MASK) {
+        UChar32 nextStart;
+
+        /* set partial block at [start..following block boundary[ */
+        block=getDataBlock(newTrie, start, true);
+        if(block<0) {
+            *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+            return;
+        }
+
+        nextStart=(start+UTRIE2_DATA_MASK)&~UTRIE2_DATA_MASK;
+        if(nextStart<=limit) {
+            fillBlock(newTrie->data+block, start&UTRIE2_DATA_MASK, UTRIE2_DATA_BLOCK_LENGTH,
+                      value, newTrie->initialValue, overwrite);
+            start=nextStart;
+        } else {
+            fillBlock(newTrie->data+block, start&UTRIE2_DATA_MASK, limit&UTRIE2_DATA_MASK,
+                      value, newTrie->initialValue, overwrite);
+            return;
+        }
+    }
+
+    /* number of positions in the last, partial block */
+    rest=limit&UTRIE2_DATA_MASK;
+
+    /* round down limit to a block boundary */
+    limit&=~UTRIE2_DATA_MASK;
+
+    /* iterate over all-value blocks */
+    if(value==newTrie->initialValue) {
+        repeatBlock=newTrie->dataNullOffset;
+    } else {
+        repeatBlock=-1;
+    }
+
+    while(startinitialValue && isInNullBlock(newTrie, start, true)) {
+            start+=UTRIE2_DATA_BLOCK_LENGTH; /* nothing to do */
+            continue;
+        }
+
+        /* get index value */
+        i2=getIndex2Block(newTrie, start, true);
+        if(i2<0) {
+            *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
+            return;
+        }
+        i2+=(start>>UTRIE2_SHIFT_2)&UTRIE2_INDEX_2_MASK;
+        block=newTrie->index2[i2];
+        if(isWritableBlock(newTrie, block)) {
+            /* already allocated */
+            if(overwrite && block>=UNEWTRIE2_DATA_0800_OFFSET) {
+                /*
+                 * We overwrite all values, and it's not a
+                 * protected (ASCII-linear or 2-byte UTF-8) block:
+                 * replace with the repeatBlock.
+                 */
+                setRepeatBlock=true;
+            } else {
+                /* !overwrite, or protected block: just write the values into this block */
+                fillBlock(newTrie->data+block,
+                          0, UTRIE2_DATA_BLOCK_LENGTH,
+                          value, newTrie->initialValue, overwrite);
+            }
+        } else if(newTrie->data[block]!=value && (overwrite || block==newTrie->dataNullOffset)) {
+            /*
+             * Set the repeatBlock instead of the null block or previous repeat block:
+             *
+             * If !isWritableBlock() then all entries in the block have the same value
+             * because it's the null block or a range block (the repeatBlock from a previous
+             * call to utrie2_setRange32()).
+             * No other blocks are used multiple times before compacting.
+             *
+             * The null block is the only non-writable block with the initialValue because
+             * of the repeatBlock initialization above. (If value==initialValue, then
+             * the repeatBlock will be the null data block.)
+             *
+             * We set our repeatBlock if the desired value differs from the block's value,
+             * and if we overwrite any data or if the data is all initial values
+             * (which is the same as the block being the null block, see above).
+             */
+            setRepeatBlock=true;
+        }
+        if(setRepeatBlock) {
+            if(repeatBlock>=0) {
+                setIndex2Entry(newTrie, i2, repeatBlock);
+            } else {
+                /* create and set and fill the repeatBlock */
+                repeatBlock=getDataBlock(newTrie, start, true);
+                if(repeatBlock<0) {
+                    *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+                    return;
+                }
+                writeBlock(newTrie->data+repeatBlock, value);
+            }
+        }
+
+        start+=UTRIE2_DATA_BLOCK_LENGTH;
+    }
+
+    if(rest>0) {
+        /* set partial block at [last block boundary..limit[ */
+        block=getDataBlock(newTrie, start, true);
+        if(block<0) {
+            *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+            return;
+        }
+
+        fillBlock(newTrie->data+block, 0, rest, value, newTrie->initialValue, overwrite);
+    }
+
+    return;
+}
+
+/* compaction --------------------------------------------------------------- */
+
+static inline UBool
+equal_int32(const int32_t *s, const int32_t *t, int32_t length) {
+    while(length>0 && *s==*t) {
+        ++s;
+        ++t;
+        --length;
+    }
+    return (UBool)(length==0);
+}
+
+static inline UBool
+equal_uint32(const uint32_t *s, const uint32_t *t, int32_t length) {
+    while(length>0 && *s==*t) {
+        ++s;
+        ++t;
+        --length;
+    }
+    return (UBool)(length==0);
+}
+
+static int32_t
+findSameIndex2Block(const int32_t *idx, int32_t index2Length, int32_t otherBlock) {
+    int32_t block;
+
+    /* ensure that we do not even partially get past index2Length */
+    index2Length-=UTRIE2_INDEX_2_BLOCK_LENGTH;
+
+    for(block=0; block<=index2Length; ++block) {
+        if(equal_int32(idx+block, idx+otherBlock, UTRIE2_INDEX_2_BLOCK_LENGTH)) {
+            return block;
+        }
+    }
+    return -1;
+}
+
+static int32_t
+findSameDataBlock(const uint32_t *data, int32_t dataLength, int32_t otherBlock, int32_t blockLength) {
+    int32_t block;
+
+    /* ensure that we do not even partially get past dataLength */
+    dataLength-=blockLength;
+
+    for(block=0; block<=dataLength; block+=UTRIE2_DATA_GRANULARITY) {
+        if(equal_uint32(data+block, data+otherBlock, blockLength)) {
+            return block;
+        }
+    }
+    return -1;
+}
+
+/*
+ * Find the start of the last range in the trie by enumerating backward.
+ * Indexes for supplementary code points higher than this will be omitted.
+ */
+static UChar32
+findHighStart(UNewTrie2 *trie, uint32_t highValue) {
+    const uint32_t *data32;
+
+    uint32_t value, initialValue;
+    UChar32 c, prev;
+    int32_t i1, i2, j, i2Block, prevI2Block, index2NullOffset, block, prevBlock, nullBlock;
+
+    data32=trie->data;
+    initialValue=trie->initialValue;
+
+    index2NullOffset=trie->index2NullOffset;
+    nullBlock=trie->dataNullOffset;
+
+    /* set variables for previous range */
+    if(highValue==initialValue) {
+        prevI2Block=index2NullOffset;
+        prevBlock=nullBlock;
+    } else {
+        prevI2Block=-1;
+        prevBlock=-1;
+    }
+    prev=0x110000;
+
+    /* enumerate index-2 blocks */
+    i1=UNEWTRIE2_INDEX_1_LENGTH;
+    c=prev;
+    while(c>0) {
+        i2Block=trie->index1[--i1];
+        if(i2Block==prevI2Block) {
+            /* the index-2 block is the same as the previous one, and filled with highValue */
+            c-=UTRIE2_CP_PER_INDEX_1_ENTRY;
+            continue;
+        }
+        prevI2Block=i2Block;
+        if(i2Block==index2NullOffset) {
+            /* this is the null index-2 block */
+            if(highValue!=initialValue) {
+                return c;
+            }
+            c-=UTRIE2_CP_PER_INDEX_1_ENTRY;
+        } else {
+            /* enumerate data blocks for one index-2 block */
+            for(i2=UTRIE2_INDEX_2_BLOCK_LENGTH; i2>0;) {
+                block=trie->index2[i2Block+ --i2];
+                if(block==prevBlock) {
+                    /* the block is the same as the previous one, and filled with highValue */
+                    c-=UTRIE2_DATA_BLOCK_LENGTH;
+                    continue;
+                }
+                prevBlock=block;
+                if(block==nullBlock) {
+                    /* this is the null data block */
+                    if(highValue!=initialValue) {
+                        return c;
+                    }
+                    c-=UTRIE2_DATA_BLOCK_LENGTH;
+                } else {
+                    for(j=UTRIE2_DATA_BLOCK_LENGTH; j>0;) {
+                        value=data32[block+ --j];
+                        if(value!=highValue) {
+                            return c;
+                        }
+                        --c;
+                    }
+                }
+            }
+        }
+    }
+
+    /* deliver last range */
+    return 0;
+}
+
+/*
+ * Compact a build-time trie.
+ *
+ * The compaction
+ * - removes blocks that are identical with earlier ones
+ * - overlaps adjacent blocks as much as possible (if overlap==true)
+ * - moves blocks in steps of the data granularity
+ * - moves and overlaps blocks that overlap with multiple values in the overlap region
+ *
+ * It does not
+ * - try to move and overlap blocks that are not already adjacent
+ */
+static void
+compactData(UNewTrie2 *trie) {
+#ifdef UTRIE2_DEBUG
+    int32_t countSame=0, sumOverlaps=0;
+#endif
+
+    int32_t start, newStart, movedStart;
+    int32_t blockLength, overlap;
+    int32_t i, mapIndex, blockCount;
+
+    /* do not compact linear-ASCII data */
+    newStart=UTRIE2_DATA_START_OFFSET;
+    for(start=0, i=0; startmap[i]=start;
+    }
+
+    /*
+     * Start with a block length of 64 for 2-byte UTF-8,
+     * then switch to UTRIE2_DATA_BLOCK_LENGTH.
+     */
+    blockLength=64;
+    blockCount=blockLength>>UTRIE2_SHIFT_2;
+    for(start=newStart; startdataLength;) {
+        /*
+         * start: index of first entry of current block
+         * newStart: index where the current block is to be moved
+         *           (right after current end of already-compacted data)
+         */
+        if(start==UNEWTRIE2_DATA_0800_OFFSET) {
+            blockLength=UTRIE2_DATA_BLOCK_LENGTH;
+            blockCount=1;
+        }
+
+        /* skip blocks that are not used */
+        if(trie->map[start>>UTRIE2_SHIFT_2]<=0) {
+            /* advance start to the next block */
+            start+=blockLength;
+
+            /* leave newStart with the previous block! */
+            continue;
+        }
+
+        /* search for an identical block */
+        if( (movedStart=findSameDataBlock(trie->data, newStart, start, blockLength))
+             >=0
+        ) {
+#ifdef UTRIE2_DEBUG
+            ++countSame;
+#endif
+            /* found an identical block, set the other block's index value for the current block */
+            for(i=blockCount, mapIndex=start>>UTRIE2_SHIFT_2; i>0; --i) {
+                trie->map[mapIndex++]=movedStart;
+                movedStart+=UTRIE2_DATA_BLOCK_LENGTH;
+            }
+
+            /* advance start to the next block */
+            start+=blockLength;
+
+            /* leave newStart with the previous block! */
+            continue;
+        }
+
+        /* see if the beginning of this block can be overlapped with the end of the previous block */
+        /* look for maximum overlap (modulo granularity) with the previous, adjacent block */
+        for(overlap=blockLength-UTRIE2_DATA_GRANULARITY;
+            overlap>0 && !equal_uint32(trie->data+(newStart-overlap), trie->data+start, overlap);
+            overlap-=UTRIE2_DATA_GRANULARITY) {}
+
+#ifdef UTRIE2_DEBUG
+            sumOverlaps+=overlap;
+#endif
+        if(overlap>0 || newStart>UTRIE2_SHIFT_2; i>0; --i) {
+                trie->map[mapIndex++]=movedStart;
+                movedStart+=UTRIE2_DATA_BLOCK_LENGTH;
+            }
+
+            /* move the non-overlapping indexes to their new positions */
+            start+=overlap;
+            for(i=blockLength-overlap; i>0; --i) {
+                trie->data[newStart++]=trie->data[start++];
+            }
+        } else /* no overlap && newStart==start */ {
+            for(i=blockCount, mapIndex=start>>UTRIE2_SHIFT_2; i>0; --i) {
+                trie->map[mapIndex++]=start;
+                start+=UTRIE2_DATA_BLOCK_LENGTH;
+            }
+            newStart=start;
+        }
+    }
+
+    /* now adjust the index-2 table */
+    for(i=0; iindex2Length; ++i) {
+        if(i==UNEWTRIE2_INDEX_GAP_OFFSET) {
+            /* Gap indexes are invalid (-1). Skip over the gap. */
+            i+=UNEWTRIE2_INDEX_GAP_LENGTH;
+        }
+        trie->index2[i]=trie->map[trie->index2[i]>>UTRIE2_SHIFT_2];
+    }
+    trie->dataNullOffset=trie->map[trie->dataNullOffset>>UTRIE2_SHIFT_2];
+
+    /* ensure dataLength alignment */
+    while((newStart&(UTRIE2_DATA_GRANULARITY-1))!=0) {
+        trie->data[newStart++]=trie->initialValue;
+    }
+
+#ifdef UTRIE2_DEBUG
+    /* we saved some space */
+    printf("compacting UTrie2: count of 32-bit data words %lu->%lu  countSame=%ld  sumOverlaps=%ld\n",
+            (long)trie->dataLength, (long)newStart, (long)countSame, (long)sumOverlaps);
+#endif
+
+    trie->dataLength=newStart;
+}
+
+static void
+compactIndex2(UNewTrie2 *trie) {
+    int32_t i, start, newStart, movedStart, overlap;
+
+    /* do not compact linear-BMP index-2 blocks */
+    newStart=UTRIE2_INDEX_2_BMP_LENGTH;
+    for(start=0, i=0; startmap[i]=start;
+    }
+
+    /* Reduce the index table gap to what will be needed at runtime. */
+    newStart+=UTRIE2_UTF8_2B_INDEX_2_LENGTH+((trie->highStart-0x10000)>>UTRIE2_SHIFT_1);
+
+    for(start=UNEWTRIE2_INDEX_2_NULL_OFFSET; startindex2Length;) {
+        /*
+         * start: index of first entry of current block
+         * newStart: index where the current block is to be moved
+         *           (right after current end of already-compacted data)
+         */
+
+        /* search for an identical block */
+        if( (movedStart=findSameIndex2Block(trie->index2, newStart, start))
+             >=0
+        ) {
+            /* found an identical block, set the other block's index value for the current block */
+            trie->map[start>>UTRIE2_SHIFT_1_2]=movedStart;
+
+            /* advance start to the next block */
+            start+=UTRIE2_INDEX_2_BLOCK_LENGTH;
+
+            /* leave newStart with the previous block! */
+            continue;
+        }
+
+        /* see if the beginning of this block can be overlapped with the end of the previous block */
+        /* look for maximum overlap with the previous, adjacent block */
+        for(overlap=UTRIE2_INDEX_2_BLOCK_LENGTH-1;
+            overlap>0 && !equal_int32(trie->index2+(newStart-overlap), trie->index2+start, overlap);
+            --overlap) {}
+
+        if(overlap>0 || newStartmap[start>>UTRIE2_SHIFT_1_2]=newStart-overlap;
+
+            /* move the non-overlapping indexes to their new positions */
+            start+=overlap;
+            for(i=UTRIE2_INDEX_2_BLOCK_LENGTH-overlap; i>0; --i) {
+                trie->index2[newStart++]=trie->index2[start++];
+            }
+        } else /* no overlap && newStart==start */ {
+            trie->map[start>>UTRIE2_SHIFT_1_2]=start;
+            start+=UTRIE2_INDEX_2_BLOCK_LENGTH;
+            newStart=start;
+        }
+    }
+
+    /* now adjust the index-1 table */
+    for(i=0; iindex1[i]=trie->map[trie->index1[i]>>UTRIE2_SHIFT_1_2];
+    }
+    trie->index2NullOffset=trie->map[trie->index2NullOffset>>UTRIE2_SHIFT_1_2];
+
+    /*
+     * Ensure data table alignment:
+     * Needs to be granularity-aligned for 16-bit trie
+     * (so that dataMove will be down-shiftable),
+     * and 2-aligned for uint32_t data.
+     */
+    while((newStart&((UTRIE2_DATA_GRANULARITY-1)|1))!=0) {
+        /* Arbitrary value: 0x3fffc not possible for real data. */
+        trie->index2[newStart++]=(int32_t)0xffff<%lu\n",
+            (long)trie->index2Length, (long)newStart);
+#endif
+
+    trie->index2Length=newStart;
+}
+
+static void
+compactTrie(UTrie2 *trie, UErrorCode *pErrorCode) {
+    UNewTrie2 *newTrie;
+    UChar32 highStart, suppHighStart;
+    uint32_t highValue;
+
+    newTrie=trie->newTrie;
+
+    /* find highStart and round it up */
+    highValue=utrie2_get32(trie, 0x10ffff);
+    highStart=findHighStart(newTrie, highValue);
+    highStart=(highStart+(UTRIE2_CP_PER_INDEX_1_ENTRY-1))&~(UTRIE2_CP_PER_INDEX_1_ENTRY-1);
+    if(highStart==0x110000) {
+        highValue=trie->errorValue;
+    }
+
+    /*
+     * Set trie->highStart only after utrie2_get32(trie, highStart).
+     * Otherwise utrie2_get32(trie, highStart) would try to read the highValue.
+     */
+    trie->highStart=newTrie->highStart=highStart;
+
+#ifdef UTRIE2_DEBUG
+    printf("UTrie2: highStart U+%06lx  highValue 0x%lx  initialValue 0x%lx\n",
+            (long)highStart, (long)highValue, (long)trie->initialValue);
+#endif
+
+    if(highStart<0x110000) {
+        /* Blank out [highStart..10ffff] to release associated data blocks. */
+        suppHighStart= highStart<=0x10000 ? 0x10000 : highStart;
+        utrie2_setRange32(trie, suppHighStart, 0x10ffff, trie->initialValue, true, pErrorCode);
+        if(U_FAILURE(*pErrorCode)) {
+            return;
+        }
+    }
+
+    compactData(newTrie);
+    if(highStart>0x10000) {
+        compactIndex2(newTrie);
+#ifdef UTRIE2_DEBUG
+    } else {
+        printf("UTrie2: highStart U+%04lx  count of 16-bit index words %lu->%lu\n",
+                (long)highStart, (long)trie->newTrie->index2Length, (long)UTRIE2_INDEX_1_OFFSET);
+#endif
+    }
+
+    /*
+     * Store the highValue in the data array and round up the dataLength.
+     * Must be done after compactData() because that assumes that dataLength
+     * is a multiple of UTRIE2_DATA_BLOCK_LENGTH.
+     */
+    newTrie->data[newTrie->dataLength++]=highValue;
+    while((newTrie->dataLength&(UTRIE2_DATA_GRANULARITY-1))!=0) {
+        newTrie->data[newTrie->dataLength++]=trie->initialValue;
+    }
+
+    newTrie->isCompacted=true;
+}
+
+/* serialization ------------------------------------------------------------ */
+
+/**
+ * Maximum length of the runtime index array.
+ * Limited by its own 16-bit index values, and by uint16_t UTrie2Header.indexLength.
+ * (The actual maximum length is lower,
+ * (0x110000>>UTRIE2_SHIFT_2)+UTRIE2_UTF8_2B_INDEX_2_LENGTH+UTRIE2_MAX_INDEX_1_LENGTH.)
+ */
+#define UTRIE2_MAX_INDEX_LENGTH 0xffff
+
+/**
+ * Maximum length of the runtime data array.
+ * Limited by 16-bit index values that are left-shifted by UTRIE2_INDEX_SHIFT,
+ * and by uint16_t UTrie2Header.shiftedDataLength.
+ */
+#define UTRIE2_MAX_DATA_LENGTH (0xffff<0 if the data is moved to the end of the index array */
+    UChar32 highStart;
+
+    /* argument check */
+    if(U_FAILURE(*pErrorCode)) {
+        return;
+    }
+    if( trie==nullptr ||
+        valueBits<0 || UTRIE2_COUNT_VALUE_BITS<=valueBits
+    ) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+    newTrie=trie->newTrie;
+    if(newTrie==nullptr) {
+        /* already frozen */
+        UTrie2ValueBits frozenValueBits=
+            trie->data16!=nullptr ? UTRIE2_16_VALUE_BITS : UTRIE2_32_VALUE_BITS;
+        if(valueBits!=frozenValueBits) {
+            *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        }
+        return;
+    }
+
+    /* compact if necessary */
+    if(!newTrie->isCompacted) {
+        compactTrie(trie, pErrorCode);
+        if(U_FAILURE(*pErrorCode)) {
+            return;
+        }
+    }
+    highStart=trie->highStart;
+
+    if(highStart<=0x10000) {
+        allIndexesLength=UTRIE2_INDEX_1_OFFSET;
+    } else {
+        allIndexesLength=newTrie->index2Length;
+    }
+    if(valueBits==UTRIE2_16_VALUE_BITS) {
+        dataMove=allIndexesLength;
+    } else {
+        dataMove=0;
+    }
+
+    /* are indexLength and dataLength within limits? */
+    if( /* for unshifted indexLength */
+        allIndexesLength>UTRIE2_MAX_INDEX_LENGTH ||
+        /* for unshifted dataNullOffset */
+        (dataMove+newTrie->dataNullOffset)>0xffff ||
+        /* for unshifted 2-byte UTF-8 index-2 values */
+        (dataMove+UNEWTRIE2_DATA_0800_OFFSET)>0xffff ||
+        /* for shiftedDataLength */
+        (dataMove+newTrie->dataLength)>UTRIE2_MAX_DATA_LENGTH
+    ) {
+        *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+        return;
+    }
+
+    /* calculate the total serialized length */
+    length=sizeof(UTrie2Header)+allIndexesLength*2;
+    if(valueBits==UTRIE2_16_VALUE_BITS) {
+        length+=newTrie->dataLength*2;
+    } else {
+        length+=newTrie->dataLength*4;
+    }
+
+    trie->memory=uprv_malloc(length);
+    if(trie->memory==nullptr) {
+        *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
+        return;
+    }
+    trie->length=length;
+    trie->isMemoryOwned=true;
+
+    trie->indexLength=allIndexesLength;
+    trie->dataLength=newTrie->dataLength;
+    if(highStart<=0x10000) {
+        trie->index2NullOffset=0xffff;
+    } else {
+        trie->index2NullOffset=static_cast(UTRIE2_INDEX_2_OFFSET+newTrie->index2NullOffset);
+    }
+    trie->dataNullOffset=(uint16_t)(dataMove+newTrie->dataNullOffset);
+    trie->highValueIndex=dataMove+trie->dataLength-UTRIE2_DATA_GRANULARITY;
+
+    /* set the header fields */
+    header=(UTrie2Header *)trie->memory;
+
+    header->signature=UTRIE2_SIG; /* "Tri2" */
+    header->options=(uint16_t)valueBits;
+
+    header->indexLength=(uint16_t)trie->indexLength;
+    header->shiftedDataLength=(uint16_t)(trie->dataLength>>UTRIE2_INDEX_SHIFT);
+    header->index2NullOffset=trie->index2NullOffset;
+    header->dataNullOffset=trie->dataNullOffset;
+    header->shiftedHighStart=(uint16_t)(highStart>>UTRIE2_SHIFT_1);
+
+    /* fill the index and data arrays */
+    dest16=(uint16_t *)(header+1);
+    trie->index=dest16;
+
+    /* write the index-2 array values shifted right by UTRIE2_INDEX_SHIFT, after adding dataMove */
+    p=(uint32_t *)newTrie->index2;
+    for(i=UTRIE2_INDEX_2_BMP_LENGTH; i>0; --i) {
+        *dest16++=(uint16_t)((dataMove + *p++)>>UTRIE2_INDEX_SHIFT);
+    }
+
+    /* write UTF-8 2-byte index-2 values, not right-shifted */
+    for(i=0; i<(0xc2-0xc0); ++i) {                                  /* C0..C1 */
+        *dest16++=(uint16_t)(dataMove+UTRIE2_BAD_UTF8_DATA_OFFSET);
+    }
+    for(; i<(0xe0-0xc0); ++i) {                                     /* C2..DF */
+        *dest16++=(uint16_t)(dataMove+newTrie->index2[i<<(6-UTRIE2_SHIFT_2)]);
+    }
+
+    if(highStart>0x10000) {
+        int32_t index1Length=(highStart-0x10000)>>UTRIE2_SHIFT_1;
+        int32_t index2Offset=UTRIE2_INDEX_2_BMP_LENGTH+UTRIE2_UTF8_2B_INDEX_2_LENGTH+index1Length;
+
+        /* write 16-bit index-1 values for supplementary code points */
+        p=(uint32_t *)newTrie->index1+UTRIE2_OMITTED_BMP_INDEX_1_LENGTH;
+        for(i=index1Length; i>0; --i) {
+            *dest16++=(uint16_t)(UTRIE2_INDEX_2_OFFSET + *p++);
+        }
+
+        /*
+         * write the index-2 array values for supplementary code points,
+         * shifted right by UTRIE2_INDEX_SHIFT, after adding dataMove
+         */
+        p=(uint32_t *)newTrie->index2+index2Offset;
+        for(i=newTrie->index2Length-index2Offset; i>0; --i) {
+            *dest16++=(uint16_t)((dataMove + *p++)>>UTRIE2_INDEX_SHIFT);
+        }
+    }
+
+    /* write the 16/32-bit data array */
+    switch(valueBits) {
+    case UTRIE2_16_VALUE_BITS:
+        /* write 16-bit data values */
+        trie->data16=dest16;
+        trie->data32=nullptr;
+        p=newTrie->data;
+        for(i=newTrie->dataLength; i>0; --i) {
+            *dest16++=(uint16_t)*p++;
+        }
+        break;
+    case UTRIE2_32_VALUE_BITS:
+        /* write 32-bit data values */
+        trie->data16=nullptr;
+        trie->data32=(uint32_t *)dest16;
+        uprv_memcpy(dest16, newTrie->data, (size_t)newTrie->dataLength*4);
+        break;
+    default:
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+
+#ifdef UTRIE2_DEBUG
+    utrie2_printLengths(trie, "");
+#endif
+
+#ifdef UCPTRIE_DEBUG
+    umutablecptrie_setName(newTrie->t3, trie->name);
+    ucptrie_close(
+        umutablecptrie_buildImmutable(
+            newTrie->t3, UCPTRIE_TYPE_FAST, (UCPTrieValueWidth)valueBits, pErrorCode));
+#endif
+    /* Delete the UNewTrie2. */
+    uprv_free(newTrie->data);
+    uprv_free(newTrie);
+    trie->newTrie=nullptr;
+}
diff --git a/intl/icu/source/common/utrie2_impl.h b/intl/icu/source/common/utrie2_impl.h
new file mode 100644
index 0000000000..2a14db3a6b
--- /dev/null
+++ b/intl/icu/source/common/utrie2_impl.h
@@ -0,0 +1,175 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+*
+*   Copyright (C) 2001-2008, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+******************************************************************************
+*   file name:  utrie2_impl.h
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2008sep26 (split off from utrie2.c)
+*   created by: Markus W. Scherer
+*
+*   Definitions needed for both runtime and builder code for UTrie2,
+*   used by utrie2.c and utrie2_builder.c.
+*/
+
+#ifndef __UTRIE2_IMPL_H__
+#define __UTRIE2_IMPL_H__
+
+#ifdef UCPTRIE_DEBUG
+#include "unicode/umutablecptrie.h"
+#endif
+#include "utrie2.h"
+
+/* Public UTrie2 API implementation ----------------------------------------- */
+
+/*
+ * These definitions are mostly needed by utrie2.cpp,
+ * but also by utrie2_serialize() and utrie2_swap().
+ */
+
+// UTrie2 signature values, in platform endianness and opposite endianness.
+// The UTrie2 signature ASCII byte values spell "Tri2".
+#define UTRIE2_SIG      0x54726932
+#define UTRIE2_OE_SIG   0x32697254
+
+/**
+ * Trie data structure in serialized form:
+ *
+ * UTrie2Header header;
+ * uint16_t index[header.index2Length];
+ * uint16_t data[header.shiftedDataLength<<2];  -- or uint32_t data[...]
+ * @internal
+ */
+typedef struct UTrie2Header {
+    /** "Tri2" in big-endian US-ASCII (0x54726932) */
+    uint32_t signature;
+
+    /**
+     * options bit field:
+     * 15.. 4   reserved (0)
+     *  3.. 0   UTrie2ValueBits valueBits
+     */
+    uint16_t options;
+
+    /** UTRIE2_INDEX_1_OFFSET..UTRIE2_MAX_INDEX_LENGTH */
+    uint16_t indexLength;
+
+    /** (UTRIE2_DATA_START_OFFSET..UTRIE2_MAX_DATA_LENGTH)>>UTRIE2_INDEX_SHIFT */
+    uint16_t shiftedDataLength;
+
+    /** Null index and data blocks, not shifted. */
+    uint16_t index2NullOffset, dataNullOffset;
+
+    /**
+     * First code point of the single-value range ending with U+10ffff,
+     * rounded up and then shifted right by UTRIE2_SHIFT_1.
+     */
+    uint16_t shiftedHighStart;
+} UTrie2Header;
+
+/**
+ * Constants for use with UTrie2Header.options.
+ * @internal
+ */
+enum {
+    /** Mask to get the UTrie2ValueBits valueBits from options. */
+    UTRIE2_OPTIONS_VALUE_BITS_MASK=0xf
+};
+
+/* Building a trie ---------------------------------------------------------- */
+
+/*
+ * These definitions are mostly needed by utrie2_builder.c, but also by
+ * utrie2_get32() and utrie2_enum().
+ */
+
+enum {
+    /**
+     * At build time, leave a gap in the index-2 table,
+     * at least as long as the maximum lengths of the 2-byte UTF-8 index-2 table
+     * and the supplementary index-1 table.
+     * Round up to UTRIE2_INDEX_2_BLOCK_LENGTH for proper compacting.
+     */
+    UNEWTRIE2_INDEX_GAP_OFFSET=UTRIE2_INDEX_2_BMP_LENGTH,
+    UNEWTRIE2_INDEX_GAP_LENGTH=
+        ((UTRIE2_UTF8_2B_INDEX_2_LENGTH+UTRIE2_MAX_INDEX_1_LENGTH)+UTRIE2_INDEX_2_MASK)&
+        ~UTRIE2_INDEX_2_MASK,
+
+    /**
+     * Maximum length of the build-time index-2 array.
+     * Maximum number of Unicode code points (0x110000) shifted right by UTRIE2_SHIFT_2,
+     * plus the part of the index-2 table for lead surrogate code points,
+     * plus the build-time index gap,
+     * plus the null index-2 block.
+     */
+    UNEWTRIE2_MAX_INDEX_2_LENGTH=
+        (0x110000>>UTRIE2_SHIFT_2)+
+        UTRIE2_LSCP_INDEX_2_LENGTH+
+        UNEWTRIE2_INDEX_GAP_LENGTH+
+        UTRIE2_INDEX_2_BLOCK_LENGTH,
+
+    UNEWTRIE2_INDEX_1_LENGTH=0x110000>>UTRIE2_SHIFT_1
+};
+
+/**
+ * Maximum length of the build-time data array.
+ * One entry per 0x110000 code points, plus the illegal-UTF-8 block and the null block,
+ * plus values for the 0x400 surrogate code units.
+ */
+#define UNEWTRIE2_MAX_DATA_LENGTH (0x110000+0x40+0x40+0x400)
+
+/*
+ * Build-time trie structure.
+ *
+ * Just using a boolean flag for "repeat use" could lead to data array overflow
+ * because we would not be able to detect when a data block becomes unused.
+ * It also leads to orphan data blocks that are kept through serialization.
+ *
+ * Need to use reference counting for data blocks,
+ * and allocDataBlock() needs to look for a free block before increasing dataLength.
+ *
+ * This scheme seems like overkill for index-2 blocks since the whole index array is
+ * preallocated anyway (unlike the growable data array).
+ * Just allocating multiple index-2 blocks as needed.
+ */
+struct UNewTrie2 {
+    int32_t index1[UNEWTRIE2_INDEX_1_LENGTH];
+    int32_t index2[UNEWTRIE2_MAX_INDEX_2_LENGTH];
+    uint32_t *data;
+#ifdef UCPTRIE_DEBUG
+    UMutableCPTrie *t3;
+#endif
+
+    uint32_t initialValue, errorValue;
+    int32_t index2Length, dataCapacity, dataLength;
+    int32_t firstFreeBlock;
+    int32_t index2NullOffset, dataNullOffset;
+    UChar32 highStart;
+    UBool isCompacted;
+
+    /**
+     * Multi-purpose per-data-block table.
+     *
+     * Before compacting:
+     *
+     * Per-data-block reference counters/free-block list.
+     *  0: unused
+     * >0: reference counter (number of index-2 entries pointing here)
+     * <0: next free data block in free-block list
+     *
+     * While compacting:
+     *
+     * Map of adjusted indexes, used in compactData() and compactIndex2().
+     * Maps from original indexes to new ones.
+     */
+    int32_t map[UNEWTRIE2_MAX_DATA_LENGTH>>UTRIE2_SHIFT_2];
+};
+
+#endif
diff --git a/intl/icu/source/common/utrie_swap.cpp b/intl/icu/source/common/utrie_swap.cpp
new file mode 100644
index 0000000000..83d183699d
--- /dev/null
+++ b/intl/icu/source/common/utrie_swap.cpp
@@ -0,0 +1,348 @@
+// © 2018 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+
+// utrie_swap.cpp
+// created: 2018aug08 Markus W. Scherer
+
+#include "unicode/utypes.h"
+#include "cmemory.h"
+#include "ucptrie_impl.h"
+#include "udataswp.h"
+#include "utrie.h"
+#include "utrie2_impl.h"
+
+// These functions for swapping different generations of ICU code point tries are here
+// so that their implementation files need not depend on swapper code,
+// need not depend on each other, and so that other swapper code
+// need not depend on other trie code.
+
+namespace {
+
+constexpr int32_t ASCII_LIMIT = 0x80;
+
+}  // namespace
+
+U_CAPI int32_t U_EXPORT2
+utrie_swap(const UDataSwapper *ds,
+           const void *inData, int32_t length, void *outData,
+           UErrorCode *pErrorCode) {
+    const UTrieHeader *inTrie;
+    UTrieHeader trie;
+    int32_t size;
+    UBool dataIs32;
+
+    if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+    if(ds==nullptr || inData==nullptr || (length>=0 && outData==nullptr)) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    /* setup and swapping */
+    if(length>=0 && (uint32_t)lengthreadUInt32(inTrie->signature);
+    trie.options=ds->readUInt32(inTrie->options);
+    trie.indexLength=udata_readInt32(ds, inTrie->indexLength);
+    trie.dataLength=udata_readInt32(ds, inTrie->dataLength);
+
+    if( trie.signature!=0x54726965 ||
+        (trie.options&UTRIE_OPTIONS_SHIFT_MASK)!=UTRIE_SHIFT ||
+        ((trie.options>>UTRIE_OPTIONS_INDEX_SHIFT)&UTRIE_OPTIONS_SHIFT_MASK)!=UTRIE_INDEX_SHIFT ||
+        trie.indexLength=0) {
+        UTrieHeader *outTrie;
+
+        if(lengthswapArray32(ds, inTrie, sizeof(UTrieHeader), outTrie, pErrorCode);
+
+        /* swap the index and the data */
+        if(dataIs32) {
+            ds->swapArray16(ds, inTrie+1, trie.indexLength*2, outTrie+1, pErrorCode);
+            ds->swapArray32(ds, (const uint16_t *)(inTrie+1)+trie.indexLength, trie.dataLength*4,
+                                     (uint16_t *)(outTrie+1)+trie.indexLength, pErrorCode);
+        } else {
+            ds->swapArray16(ds, inTrie+1, (trie.indexLength+trie.dataLength)*2, outTrie+1, pErrorCode);
+        }
+    }
+
+    return size;
+}
+
+U_CAPI int32_t U_EXPORT2
+utrie2_swap(const UDataSwapper *ds,
+            const void *inData, int32_t length, void *outData,
+            UErrorCode *pErrorCode) {
+    const UTrie2Header *inTrie;
+    UTrie2Header trie;
+    int32_t dataLength, size;
+    UTrie2ValueBits valueBits;
+
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+    if(ds==nullptr || inData==nullptr || (length>=0 && outData==nullptr)) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    /* setup and swapping */
+    if(length>=0 && length<(int32_t)sizeof(UTrie2Header)) {
+        *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+        return 0;
+    }
+
+    inTrie=(const UTrie2Header *)inData;
+    trie.signature=ds->readUInt32(inTrie->signature);
+    trie.options=ds->readUInt16(inTrie->options);
+    trie.indexLength=ds->readUInt16(inTrie->indexLength);
+    trie.shiftedDataLength=ds->readUInt16(inTrie->shiftedDataLength);
+
+    valueBits=(UTrie2ValueBits)(trie.options&UTRIE2_OPTIONS_VALUE_BITS_MASK);
+    dataLength=(int32_t)trie.shiftedDataLength<=0) {
+        UTrie2Header *outTrie;
+
+        if(lengthswapArray32(ds, &inTrie->signature, 4, &outTrie->signature, pErrorCode);
+        ds->swapArray16(ds, &inTrie->options, 12, &outTrie->options, pErrorCode);
+
+        /* swap the index and the data */
+        switch(valueBits) {
+        case UTRIE2_16_VALUE_BITS:
+            ds->swapArray16(ds, inTrie+1, (trie.indexLength+dataLength)*2, outTrie+1, pErrorCode);
+            break;
+        case UTRIE2_32_VALUE_BITS:
+            ds->swapArray16(ds, inTrie+1, trie.indexLength*2, outTrie+1, pErrorCode);
+            ds->swapArray32(ds, (const uint16_t *)(inTrie+1)+trie.indexLength, dataLength*4,
+                                     (uint16_t *)(outTrie+1)+trie.indexLength, pErrorCode);
+            break;
+        default:
+            *pErrorCode=U_INVALID_FORMAT_ERROR;
+            return 0;
+        }
+    }
+
+    return size;
+}
+
+U_CAPI int32_t U_EXPORT2
+ucptrie_swap(const UDataSwapper *ds,
+             const void *inData, int32_t length, void *outData,
+             UErrorCode *pErrorCode) {
+    const UCPTrieHeader *inTrie;
+    UCPTrieHeader trie;
+    int32_t dataLength, size;
+    UCPTrieValueWidth valueWidth;
+
+    if(U_FAILURE(*pErrorCode)) {
+        return 0;
+    }
+    if(ds==nullptr || inData==nullptr || (length>=0 && outData==nullptr)) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    /* setup and swapping */
+    if(length>=0 && length<(int32_t)sizeof(UCPTrieHeader)) {
+        *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+        return 0;
+    }
+
+    inTrie=(const UCPTrieHeader *)inData;
+    trie.signature=ds->readUInt32(inTrie->signature);
+    trie.options=ds->readUInt16(inTrie->options);
+    trie.indexLength=ds->readUInt16(inTrie->indexLength);
+    trie.dataLength = ds->readUInt16(inTrie->dataLength);
+
+    UCPTrieType type = (UCPTrieType)((trie.options >> 6) & 3);
+    valueWidth = (UCPTrieValueWidth)(trie.options & UCPTRIE_OPTIONS_VALUE_BITS_MASK);
+    dataLength = ((int32_t)(trie.options & UCPTRIE_OPTIONS_DATA_LENGTH_MASK) << 4) | trie.dataLength;
+
+    int32_t minIndexLength = type == UCPTRIE_TYPE_FAST ?
+        UCPTRIE_BMP_INDEX_LENGTH : UCPTRIE_SMALL_INDEX_LENGTH;
+    if( trie.signature!=UCPTRIE_SIG ||
+        type > UCPTRIE_TYPE_SMALL ||
+        (trie.options & UCPTRIE_OPTIONS_RESERVED_MASK) != 0 ||
+        valueWidth > UCPTRIE_VALUE_BITS_8 ||
+        trie.indexLength < minIndexLength ||
+        dataLength < ASCII_LIMIT
+    ) {
+        *pErrorCode=U_INVALID_FORMAT_ERROR; /* not a UCPTrie */
+        return 0;
+    }
+
+    size=sizeof(UCPTrieHeader)+trie.indexLength*2;
+    switch(valueWidth) {
+    case UCPTRIE_VALUE_BITS_16:
+        size+=dataLength*2;
+        break;
+    case UCPTRIE_VALUE_BITS_32:
+        size+=dataLength*4;
+        break;
+    case UCPTRIE_VALUE_BITS_8:
+        size+=dataLength;
+        break;
+    default:
+        *pErrorCode=U_INVALID_FORMAT_ERROR;
+        return 0;
+    }
+
+    if(length>=0) {
+        UCPTrieHeader *outTrie;
+
+        if(lengthswapArray32(ds, &inTrie->signature, 4, &outTrie->signature, pErrorCode);
+        ds->swapArray16(ds, &inTrie->options, 12, &outTrie->options, pErrorCode);
+
+        /* swap the index */
+        const uint16_t *inIndex=reinterpret_cast(inTrie+1);
+        uint16_t *outIndex=reinterpret_cast(outTrie+1);
+        ds->swapArray16(ds, inIndex, trie.indexLength*2, outIndex, pErrorCode);
+
+        /* swap the data */
+        const uint16_t *inData=inIndex+trie.indexLength;
+        uint16_t *outData=outIndex+trie.indexLength;
+        switch(valueWidth) {
+        case UCPTRIE_VALUE_BITS_16:
+            ds->swapArray16(ds, inData, dataLength*2, outData, pErrorCode);
+            break;
+        case UCPTRIE_VALUE_BITS_32:
+            ds->swapArray32(ds, inData, dataLength*4, outData, pErrorCode);
+            break;
+        case UCPTRIE_VALUE_BITS_8:
+            if(inTrie!=outTrie) {
+                uprv_memmove(outData, inData, dataLength);
+            }
+            break;
+        default:
+            *pErrorCode=U_INVALID_FORMAT_ERROR;
+            return 0;
+        }
+    }
+
+    return size;
+}
+
+namespace {
+
+/**
+ * Gets the trie version from 32-bit-aligned memory containing the serialized form
+ * of a UTrie (version 1), a UTrie2 (version 2), or a UCPTrie (version 3).
+ *
+ * @param data a pointer to 32-bit-aligned memory containing the serialized form of a trie
+ * @param length the number of bytes available at data;
+ *               can be more than necessary (see return value)
+ * @param anyEndianOk If false, only platform-endian serialized forms are recognized.
+ *                    If true, opposite-endian serialized forms are recognized as well.
+ * @return the trie version of the serialized form, or 0 if it is not
+ *         recognized as a serialized trie
+ */
+int32_t
+getVersion(const void *data, int32_t length, UBool anyEndianOk) {
+    uint32_t signature;
+    if(length<16 || data==nullptr || (U_POINTER_MASK_LSB(data, 3)!=0)) {
+        return 0;
+    }
+    signature=*(const uint32_t *)data;
+    if(signature==UCPTRIE_SIG) {
+        return 3;
+    }
+    if(anyEndianOk && signature==UCPTRIE_OE_SIG) {
+        return 3;
+    }
+    if(signature==UTRIE2_SIG) {
+        return 2;
+    }
+    if(anyEndianOk && signature==UTRIE2_OE_SIG) {
+        return 2;
+    }
+    if(signature==UTRIE_SIG) {
+        return 1;
+    }
+    if(anyEndianOk && signature==UTRIE_OE_SIG) {
+        return 1;
+    }
+    return 0;
+}
+
+}  // namespace
+
+U_CAPI int32_t U_EXPORT2
+utrie_swapAnyVersion(const UDataSwapper *ds,
+                     const void *inData, int32_t length, void *outData,
+                     UErrorCode *pErrorCode) {
+    if(U_FAILURE(*pErrorCode)) { return 0; }
+    switch(getVersion(inData, length, true)) {
+    case 1:
+        return utrie_swap(ds, inData, length, outData, pErrorCode);
+    case 2:
+        return utrie2_swap(ds, inData, length, outData, pErrorCode);
+    case 3:
+        return ucptrie_swap(ds, inData, length, outData, pErrorCode);
+    default:
+        *pErrorCode=U_INVALID_FORMAT_ERROR;
+        return 0;
+    }
+}
diff --git a/intl/icu/source/common/uts46.cpp b/intl/icu/source/common/uts46.cpp
new file mode 100644
index 0000000000..fce2af4130
--- /dev/null
+++ b/intl/icu/source/common/uts46.cpp
@@ -0,0 +1,1494 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*   Copyright (C) 2010-2015, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*******************************************************************************
+*   file name:  uts46.cpp
+*   encoding:   UTF-8
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2010mar09
+*   created by: Markus W. Scherer
+*/
+
+#include "unicode/utypes.h"
+
+#if !UCONFIG_NO_IDNA
+
+#include "unicode/idna.h"
+#include "unicode/normalizer2.h"
+#include "unicode/uscript.h"
+#include "unicode/ustring.h"
+#include "unicode/utf16.h"
+#include "cmemory.h"
+#include "cstring.h"
+#include "punycode.h"
+#include "ubidi_props.h"
+#include "ustr_imp.h"
+
+// Note about tests for UIDNA_ERROR_DOMAIN_NAME_TOO_LONG:
+//
+// The domain name length limit is 255 octets in an internal DNS representation
+// where the last ("root") label is the empty label
+// represented by length byte 0 alone.
+// In a conventional string, this translates to 253 characters, or 254
+// if there is a trailing dot for the root label.
+
+U_NAMESPACE_BEGIN
+
+// Severe errors which usually result in a U+FFFD replacement character in the result string.
+const uint32_t severeErrors=
+    UIDNA_ERROR_LEADING_COMBINING_MARK|
+    UIDNA_ERROR_DISALLOWED|
+    UIDNA_ERROR_PUNYCODE|
+    UIDNA_ERROR_LABEL_HAS_DOT|
+    UIDNA_ERROR_INVALID_ACE_LABEL;
+
+static inline UBool
+isASCIIString(const UnicodeString &dest) {
+    const char16_t *s=dest.getBuffer();
+    const char16_t *limit=s+dest.length();
+    while(s0x7f) {
+            return false;
+        }
+    }
+    return true;
+}
+
+static UBool
+isASCIIOkBiDi(const char16_t *s, int32_t length);
+
+static UBool
+isASCIIOkBiDi(const char *s, int32_t length);
+
+// IDNA class default implementations -------------------------------------- ***
+
+IDNA::~IDNA() {}
+
+void
+IDNA::labelToASCII_UTF8(StringPiece label, ByteSink &dest,
+                        IDNAInfo &info, UErrorCode &errorCode) const {
+    if(U_SUCCESS(errorCode)) {
+        UnicodeString destString;
+        labelToASCII(UnicodeString::fromUTF8(label), destString,
+                     info, errorCode).toUTF8(dest);
+    }
+}
+
+void
+IDNA::labelToUnicodeUTF8(StringPiece label, ByteSink &dest,
+                         IDNAInfo &info, UErrorCode &errorCode) const {
+    if(U_SUCCESS(errorCode)) {
+        UnicodeString destString;
+        labelToUnicode(UnicodeString::fromUTF8(label), destString,
+                       info, errorCode).toUTF8(dest);
+    }
+}
+
+void
+IDNA::nameToASCII_UTF8(StringPiece name, ByteSink &dest,
+                       IDNAInfo &info, UErrorCode &errorCode) const {
+    if(U_SUCCESS(errorCode)) {
+        UnicodeString destString;
+        nameToASCII(UnicodeString::fromUTF8(name), destString,
+                    info, errorCode).toUTF8(dest);
+    }
+}
+
+void
+IDNA::nameToUnicodeUTF8(StringPiece name, ByteSink &dest,
+                        IDNAInfo &info, UErrorCode &errorCode) const {
+    if(U_SUCCESS(errorCode)) {
+        UnicodeString destString;
+        nameToUnicode(UnicodeString::fromUTF8(name), destString,
+                      info, errorCode).toUTF8(dest);
+    }
+}
+
+// UTS46 class declaration ------------------------------------------------- ***
+
+class UTS46 : public IDNA {
+public:
+    UTS46(uint32_t options, UErrorCode &errorCode);
+    virtual ~UTS46();
+
+    virtual UnicodeString &
+    labelToASCII(const UnicodeString &label, UnicodeString &dest,
+                 IDNAInfo &info, UErrorCode &errorCode) const override;
+
+    virtual UnicodeString &
+    labelToUnicode(const UnicodeString &label, UnicodeString &dest,
+                   IDNAInfo &info, UErrorCode &errorCode) const override;
+
+    virtual UnicodeString &
+    nameToASCII(const UnicodeString &name, UnicodeString &dest,
+                IDNAInfo &info, UErrorCode &errorCode) const override;
+
+    virtual UnicodeString &
+    nameToUnicode(const UnicodeString &name, UnicodeString &dest,
+                  IDNAInfo &info, UErrorCode &errorCode) const override;
+
+    virtual void
+    labelToASCII_UTF8(StringPiece label, ByteSink &dest,
+                      IDNAInfo &info, UErrorCode &errorCode) const override;
+
+    virtual void
+    labelToUnicodeUTF8(StringPiece label, ByteSink &dest,
+                       IDNAInfo &info, UErrorCode &errorCode) const override;
+
+    virtual void
+    nameToASCII_UTF8(StringPiece name, ByteSink &dest,
+                     IDNAInfo &info, UErrorCode &errorCode) const override;
+
+    virtual void
+    nameToUnicodeUTF8(StringPiece name, ByteSink &dest,
+                      IDNAInfo &info, UErrorCode &errorCode) const override;
+
+private:
+    UnicodeString &
+    process(const UnicodeString &src,
+            UBool isLabel, UBool toASCII,
+            UnicodeString &dest,
+            IDNAInfo &info, UErrorCode &errorCode) const;
+
+    void
+    processUTF8(StringPiece src,
+                UBool isLabel, UBool toASCII,
+                ByteSink &dest,
+                IDNAInfo &info, UErrorCode &errorCode) const;
+
+    UnicodeString &
+    processUnicode(const UnicodeString &src,
+                   int32_t labelStart, int32_t mappingStart,
+                   UBool isLabel, UBool toASCII,
+                   UnicodeString &dest,
+                   IDNAInfo &info, UErrorCode &errorCode) const;
+
+    // returns the new dest.length()
+    int32_t
+    mapDevChars(UnicodeString &dest, int32_t labelStart, int32_t mappingStart,
+                UErrorCode &errorCode) const;
+
+    // returns the new label length
+    int32_t
+    processLabel(UnicodeString &dest,
+                 int32_t labelStart, int32_t labelLength,
+                 UBool toASCII,
+                 IDNAInfo &info, UErrorCode &errorCode) const;
+    int32_t
+    markBadACELabel(UnicodeString &dest,
+                    int32_t labelStart, int32_t labelLength,
+                    UBool toASCII, IDNAInfo &info, UErrorCode &errorCode) const;
+
+    void
+    checkLabelBiDi(const char16_t *label, int32_t labelLength, IDNAInfo &info) const;
+
+    UBool
+    isLabelOkContextJ(const char16_t *label, int32_t labelLength) const;
+
+    void
+    checkLabelContextO(const char16_t *label, int32_t labelLength, IDNAInfo &info) const;
+
+    const Normalizer2 &uts46Norm2;  // uts46.nrm
+    uint32_t options;
+};
+
+IDNA *
+IDNA::createUTS46Instance(uint32_t options, UErrorCode &errorCode) {
+    if(U_SUCCESS(errorCode)) {
+        IDNA *idna=new UTS46(options, errorCode);
+        if(idna==nullptr) {
+            errorCode=U_MEMORY_ALLOCATION_ERROR;
+        } else if(U_FAILURE(errorCode)) {
+            delete idna;
+            idna=nullptr;
+        }
+        return idna;
+    } else {
+        return nullptr;
+    }
+}
+
+// UTS46 implementation ---------------------------------------------------- ***
+
+UTS46::UTS46(uint32_t opt, UErrorCode &errorCode)
+        : uts46Norm2(*Normalizer2::getInstance(nullptr, "uts46", UNORM2_COMPOSE, errorCode)),
+          options(opt) {}
+
+UTS46::~UTS46() {}
+
+UnicodeString &
+UTS46::labelToASCII(const UnicodeString &label, UnicodeString &dest,
+                    IDNAInfo &info, UErrorCode &errorCode) const {
+    return process(label, true, true, dest, info, errorCode);
+}
+
+UnicodeString &
+UTS46::labelToUnicode(const UnicodeString &label, UnicodeString &dest,
+                      IDNAInfo &info, UErrorCode &errorCode) const {
+    return process(label, true, false, dest, info, errorCode);
+}
+
+UnicodeString &
+UTS46::nameToASCII(const UnicodeString &name, UnicodeString &dest,
+                   IDNAInfo &info, UErrorCode &errorCode) const {
+    process(name, false, true, dest, info, errorCode);
+    if( dest.length()>=254 && (info.errors&UIDNA_ERROR_DOMAIN_NAME_TOO_LONG)==0 &&
+        isASCIIString(dest) &&
+        (dest.length()>254 || dest[253]!=0x2e)
+    ) {
+        info.errors|=UIDNA_ERROR_DOMAIN_NAME_TOO_LONG;
+    }
+    return dest;
+}
+
+UnicodeString &
+UTS46::nameToUnicode(const UnicodeString &name, UnicodeString &dest,
+                     IDNAInfo &info, UErrorCode &errorCode) const {
+    return process(name, false, false, dest, info, errorCode);
+}
+
+void
+UTS46::labelToASCII_UTF8(StringPiece label, ByteSink &dest,
+                         IDNAInfo &info, UErrorCode &errorCode) const {
+    processUTF8(label, true, true, dest, info, errorCode);
+}
+
+void
+UTS46::labelToUnicodeUTF8(StringPiece label, ByteSink &dest,
+                          IDNAInfo &info, UErrorCode &errorCode) const {
+    processUTF8(label, true, false, dest, info, errorCode);
+}
+
+void
+UTS46::nameToASCII_UTF8(StringPiece name, ByteSink &dest,
+                        IDNAInfo &info, UErrorCode &errorCode) const {
+    processUTF8(name, false, true, dest, info, errorCode);
+}
+
+void
+UTS46::nameToUnicodeUTF8(StringPiece name, ByteSink &dest,
+                         IDNAInfo &info, UErrorCode &errorCode) const {
+    processUTF8(name, false, false, dest, info, errorCode);
+}
+
+// UTS #46 data for ASCII characters.
+// The normalizer (using uts46.nrm) maps uppercase ASCII letters to lowercase
+// and passes through all other ASCII characters.
+// If UIDNA_USE_STD3_RULES is set, then non-LDH characters are disallowed
+// using this data.
+// The ASCII fastpath also uses this data.
+// Values: -1=disallowed  0==valid  1==mapped (lowercase)
+static const int8_t asciiData[128]={
+    -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
+    -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
+    // 002D..002E; valid  #  HYPHEN-MINUS..FULL STOP
+    -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,  0,  0, -1,
+    // 0030..0039; valid  #  DIGIT ZERO..DIGIT NINE
+     0,  0,  0,  0,  0,  0,  0,  0,  0,  0, -1, -1, -1, -1, -1, -1,
+    // 0041..005A; mapped  #  LATIN CAPITAL LETTER A..LATIN CAPITAL LETTER Z
+    -1,  1,  1,  1,  1,  1,  1,  1,  1,  1,  1,  1,  1,  1,  1,  1,
+     1,  1,  1,  1,  1,  1,  1,  1,  1,  1,  1, -1, -1, -1, -1, -1,
+    // 0061..007A; valid  #  LATIN SMALL LETTER A..LATIN SMALL LETTER Z
+    -1,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,
+     0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, -1, -1, -1, -1, -1
+};
+
+UnicodeString &
+UTS46::process(const UnicodeString &src,
+               UBool isLabel, UBool toASCII,
+               UnicodeString &dest,
+               IDNAInfo &info, UErrorCode &errorCode) const {
+    // uts46Norm2.normalize() would do all of this error checking and setup,
+    // but with the ASCII fastpath we do not always call it, and do not
+    // call it first.
+    if(U_FAILURE(errorCode)) {
+        dest.setToBogus();
+        return dest;
+    }
+    const char16_t *srcArray=src.getBuffer();
+    if(&dest==&src || srcArray==nullptr) {
+        errorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        dest.setToBogus();
+        return dest;
+    }
+    // Arguments are fine, reset output values.
+    dest.remove();
+    info.reset();
+    int32_t srcLength=src.length();
+    if(srcLength==0) {
+        info.errors|=UIDNA_ERROR_EMPTY_LABEL;
+        return dest;
+    }
+    char16_t *destArray=dest.getBuffer(srcLength);
+    if(destArray==nullptr) {
+        errorCode=U_MEMORY_ALLOCATION_ERROR;
+        return dest;
+    }
+    // ASCII fastpath
+    UBool disallowNonLDHDot=(options&UIDNA_USE_STD3_RULES)!=0;
+    int32_t labelStart=0;
+    int32_t i;
+    for(i=0;; ++i) {
+        if(i==srcLength) {
+            if(toASCII) {
+                if((i-labelStart)>63) {
+                    info.labelErrors|=UIDNA_ERROR_LABEL_TOO_LONG;
+                }
+                // There is a trailing dot if labelStart==i.
+                if(!isLabel && i>=254 && (i>254 || labelStart0x7f) {
+            break;
+        }
+        int cData=asciiData[c];
+        if(cData>0) {
+            destArray[i]=c+0x20;  // Lowercase an uppercase ASCII letter.
+        } else if(cData<0 && disallowNonLDHDot) {
+            break;  // Replacing with U+FFFD can be complicated for toASCII.
+        } else {
+            destArray[i]=c;
+            if(c==0x2d) {  // hyphen
+                if(i==(labelStart+3) && srcArray[i-1]==0x2d) {
+                    // "??--..." is Punycode or forbidden.
+                    ++i;  // '-' was copied to dest already
+                    break;
+                }
+                if(i==labelStart) {
+                    // label starts with "-"
+                    info.labelErrors|=UIDNA_ERROR_LEADING_HYPHEN;
+                }
+                if((i+1)==srcLength || srcArray[i+1]==0x2e) {
+                    // label ends with "-"
+                    info.labelErrors|=UIDNA_ERROR_TRAILING_HYPHEN;
+                }
+            } else if(c==0x2e) {  // dot
+                if(isLabel) {
+                    // Replacing with U+FFFD can be complicated for toASCII.
+                    ++i;  // '.' was copied to dest already
+                    break;
+                }
+                if(i==labelStart) {
+                    info.labelErrors|=UIDNA_ERROR_EMPTY_LABEL;
+                }
+                if(toASCII && (i-labelStart)>63) {
+                    info.labelErrors|=UIDNA_ERROR_LABEL_TOO_LONG;
+                }
+                info.errors|=info.labelErrors;
+                info.labelErrors=0;
+                labelStart=i+1;
+            }
+        }
+    }
+    info.errors|=info.labelErrors;
+    dest.releaseBuffer(i);
+    processUnicode(src, labelStart, i, isLabel, toASCII, dest, info, errorCode);
+    if( info.isBiDi && U_SUCCESS(errorCode) && (info.errors&severeErrors)==0 &&
+        (!info.isOkBiDi || (labelStart>0 && !isASCIIOkBiDi(dest.getBuffer(), labelStart)))
+    ) {
+        info.errors|=UIDNA_ERROR_BIDI;
+    }
+    return dest;
+}
+
+void
+UTS46::processUTF8(StringPiece src,
+                   UBool isLabel, UBool toASCII,
+                   ByteSink &dest,
+                   IDNAInfo &info, UErrorCode &errorCode) const {
+    if(U_FAILURE(errorCode)) {
+        return;
+    }
+    const char *srcArray=src.data();
+    int32_t srcLength=src.length();
+    if(srcArray==nullptr && srcLength!=0) {
+        errorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
+    // Arguments are fine, reset output values.
+    info.reset();
+    if(srcLength==0) {
+        info.errors|=UIDNA_ERROR_EMPTY_LABEL;
+        dest.Flush();
+        return;
+    }
+    UnicodeString destString;
+    int32_t labelStart=0;
+    if(srcLength<=256) {  // length of stackArray[]
+        // ASCII fastpath
+        char stackArray[256];
+        int32_t destCapacity;
+        char *destArray=dest.GetAppendBuffer(srcLength, srcLength+20,
+                                             stackArray, UPRV_LENGTHOF(stackArray), &destCapacity);
+        UBool disallowNonLDHDot=(options&UIDNA_USE_STD3_RULES)!=0;
+        int32_t i;
+        for(i=0;; ++i) {
+            if(i==srcLength) {
+                if(toASCII) {
+                    if((i-labelStart)>63) {
+                        info.labelErrors|=UIDNA_ERROR_LABEL_TOO_LONG;
+                    }
+                    // There is a trailing dot if labelStart==i.
+                    if(!isLabel && i>=254 && (i>254 || labelStart0x7f
+                break;
+            }
+            int cData=asciiData[(int)c];  // Cast: gcc warns about indexing with a char.
+            if(cData>0) {
+                destArray[i]=c+0x20;  // Lowercase an uppercase ASCII letter.
+            } else if(cData<0 && disallowNonLDHDot) {
+                break;  // Replacing with U+FFFD can be complicated for toASCII.
+            } else {
+                destArray[i]=c;
+                if(c==0x2d) {  // hyphen
+                    if(i==(labelStart+3) && srcArray[i-1]==0x2d) {
+                        // "??--..." is Punycode or forbidden.
+                        break;
+                    }
+                    if(i==labelStart) {
+                        // label starts with "-"
+                        info.labelErrors|=UIDNA_ERROR_LEADING_HYPHEN;
+                    }
+                    if((i+1)==srcLength || srcArray[i+1]==0x2e) {
+                        // label ends with "-"
+                        info.labelErrors|=UIDNA_ERROR_TRAILING_HYPHEN;
+                    }
+                } else if(c==0x2e) {  // dot
+                    if(isLabel) {
+                        break;  // Replacing with U+FFFD can be complicated for toASCII.
+                    }
+                    if(i==labelStart) {
+                        info.labelErrors|=UIDNA_ERROR_EMPTY_LABEL;
+                    }
+                    if(toASCII && (i-labelStart)>63) {
+                        info.labelErrors|=UIDNA_ERROR_LABEL_TOO_LONG;
+                    }
+                    info.errors|=info.labelErrors;
+                    info.labelErrors=0;
+                    labelStart=i+1;
+                }
+            }
+        }
+        info.errors|=info.labelErrors;
+        // Convert the processed ASCII prefix of the current label to UTF-16.
+        int32_t mappingStart=i-labelStart;
+        destString=UnicodeString::fromUTF8(StringPiece(destArray+labelStart, mappingStart));
+        // Output the previous ASCII labels and process the rest of src in UTF-16.
+        dest.Append(destArray, labelStart);
+        processUnicode(UnicodeString::fromUTF8(StringPiece(src, labelStart)), 0, mappingStart,
+                       isLabel, toASCII,
+                       destString, info, errorCode);
+    } else {
+        // src is too long for the ASCII fastpath implementation.
+        processUnicode(UnicodeString::fromUTF8(src), 0, 0,
+                       isLabel, toASCII,
+                       destString, info, errorCode);
+    }
+    destString.toUTF8(dest);  // calls dest.Flush()
+    if(toASCII && !isLabel) {
+        // length==labelStart==254 means that there is a trailing dot (ok) and
+        // destString is empty (do not index at 253-labelStart).
+        int32_t length=labelStart+destString.length();
+        if( length>=254 && isASCIIString(destString) &&
+            (length>254 ||
+             (labelStart<254 && destString[253-labelStart]!=0x2e))
+        ) {
+            info.errors|=UIDNA_ERROR_DOMAIN_NAME_TOO_LONG;
+        }
+    }
+    if( info.isBiDi && U_SUCCESS(errorCode) && (info.errors&severeErrors)==0 &&
+        (!info.isOkBiDi || (labelStart>0 && !isASCIIOkBiDi(srcArray, labelStart)))
+    ) {
+        info.errors|=UIDNA_ERROR_BIDI;
+    }
+}
+
+UnicodeString &
+UTS46::processUnicode(const UnicodeString &src,
+                      int32_t labelStart, int32_t mappingStart,
+                      UBool isLabel, UBool toASCII,
+                      UnicodeString &dest,
+                      IDNAInfo &info, UErrorCode &errorCode) const {
+    if(mappingStart==0) {
+        uts46Norm2.normalize(src, dest, errorCode);
+    } else {
+        uts46Norm2.normalizeSecondAndAppend(dest, src.tempSubString(mappingStart), errorCode);
+    }
+    if(U_FAILURE(errorCode)) {
+        return dest;
+    }
+    UBool doMapDevChars=
+        toASCII ? (options&UIDNA_NONTRANSITIONAL_TO_ASCII)==0 :
+                  (options&UIDNA_NONTRANSITIONAL_TO_UNICODE)==0;
+    const char16_t *destArray=dest.getBuffer();
+    int32_t destLength=dest.length();
+    int32_t labelLimit=labelStart;
+    while(labelLimit=0x200c)) {
+            info.isTransDiff=true;
+            if(doMapDevChars) {
+                destLength=mapDevChars(dest, labelStart, labelLimit, errorCode);
+                if(U_FAILURE(errorCode)) {
+                    return dest;
+                }
+                destArray=dest.getBuffer();
+                // All deviation characters have been mapped, no need to check for them again.
+                doMapDevChars=false;
+                // Do not increment labelLimit in case c was removed.
+                continue;
+            }
+        } else if(U16_IS_SURROGATE(c)) {
+            if(U16_IS_SURROGATE_LEAD(c) ?
+                    (labelLimit+1)==destLength || !U16_IS_TRAIL(destArray[labelLimit+1]) :
+                    labelLimit==labelStart || !U16_IS_LEAD(destArray[labelLimit-1])) {
+                // Map an unpaired surrogate to U+FFFD before normalization so that when
+                // that removes characters we do not turn two unpaired ones into a pair.
+                info.labelErrors|=UIDNA_ERROR_DISALLOWED;
+                dest.setCharAt(labelLimit, 0xfffd);
+                destArray=dest.getBuffer();
+            }
+        }
+        ++labelLimit;
+    }
+    // Permit an empty label at the end (0=4 && label[0]==0x78 && label[1]==0x6e && label[2]==0x2d && label[3]==0x2d) {
+        // Label starts with "xn--", try to un-Punycode it.
+        // In IDNA2008, labels like "xn--" (decodes to an empty string) and
+        // "xn--ASCII-" (decodes to just "ASCII") fail the round-trip validation from
+        // comparing the ToUnicode input with the back-to-ToASCII output.
+        // They are alternate encodings of the respective ASCII labels.
+        // Ignore "xn---" here: It will fail Punycode.decode() which logically comes before
+        // the round-trip verification.
+        if(labelLength==4 || (labelLength>5 && label[labelLength-1]==u'-')) {
+            info.labelErrors|=UIDNA_ERROR_INVALID_ACE_LABEL;
+            return markBadACELabel(dest, labelStart, labelLength, toASCII, info, errorCode);
+        }
+        wasPunycode=true;
+        char16_t *unicodeBuffer=fromPunycode.getBuffer(-1);  // capacity==-1: most labels should fit
+        if(unicodeBuffer==nullptr) {
+            // Should never occur if we used capacity==-1 which uses the internal buffer.
+            errorCode=U_MEMORY_ALLOCATION_ERROR;
+            return labelLength;
+        }
+        UErrorCode punycodeErrorCode=U_ZERO_ERROR;
+        int32_t unicodeLength=u_strFromPunycode(label+4, labelLength-4,
+                                                unicodeBuffer, fromPunycode.getCapacity(),
+                                                nullptr, &punycodeErrorCode);
+        if(punycodeErrorCode==U_BUFFER_OVERFLOW_ERROR) {
+            fromPunycode.releaseBuffer(0);
+            unicodeBuffer=fromPunycode.getBuffer(unicodeLength);
+            if(unicodeBuffer==nullptr) {
+                errorCode=U_MEMORY_ALLOCATION_ERROR;
+                return labelLength;
+            }
+            punycodeErrorCode=U_ZERO_ERROR;
+            unicodeLength=u_strFromPunycode(label+4, labelLength-4,
+                                            unicodeBuffer, fromPunycode.getCapacity(),
+                                            nullptr, &punycodeErrorCode);
+        }
+        fromPunycode.releaseBuffer(unicodeLength);
+        if(U_FAILURE(punycodeErrorCode)) {
+            info.labelErrors|=UIDNA_ERROR_PUNYCODE;
+            return markBadACELabel(dest, labelStart, labelLength, toASCII, info, errorCode);
+        }
+        // Check for NFC, and for characters that are not
+        // valid or deviation characters according to the normalizer.
+        // If there is something wrong, then the string will change.
+        // Note that the normalizer passes through non-LDH ASCII and deviation characters.
+        // Deviation characters are ok in Punycode even in transitional processing.
+        // In the code further below, if we find non-LDH ASCII and we have UIDNA_USE_STD3_RULES
+        // then we will set UIDNA_ERROR_INVALID_ACE_LABEL there too.
+        UBool isValid=uts46Norm2.isNormalized(fromPunycode, errorCode);
+        if(U_FAILURE(errorCode)) {
+            return labelLength;
+        }
+        if(!isValid) {
+            info.labelErrors|=UIDNA_ERROR_INVALID_ACE_LABEL;
+            return markBadACELabel(dest, labelStart, labelLength, toASCII, info, errorCode);
+        }
+        labelString=&fromPunycode;
+        label=fromPunycode.getBuffer();
+        labelStart=0;
+        labelLength=fromPunycode.length();
+    } else {
+        wasPunycode=false;
+        labelString=&dest;
+    }
+    // Validity check
+    if(labelLength==0) {
+        info.labelErrors|=UIDNA_ERROR_EMPTY_LABEL;
+        return replaceLabel(dest, destLabelStart, destLabelLength,
+                            *labelString, labelLength, errorCode);
+    }
+    // labelLength>0
+    if(labelLength>=4 && label[2]==0x2d && label[3]==0x2d) {
+        // label starts with "??--"
+        info.labelErrors|=UIDNA_ERROR_HYPHEN_3_4;
+    }
+    if(label[0]==0x2d) {
+        // label starts with "-"
+        info.labelErrors|=UIDNA_ERROR_LEADING_HYPHEN;
+    }
+    if(label[labelLength-1]==0x2d) {
+        // label ends with "-"
+        info.labelErrors|=UIDNA_ERROR_TRAILING_HYPHEN;
+    }
+    // If the label was not a Punycode label, then it was the result of
+    // mapping, normalization and label segmentation.
+    // If the label was in Punycode, then we mapped it again above
+    // and checked its validity.
+    // Now we handle the STD3 restriction to LDH characters (if set)
+    // and we look for U+FFFD which indicates disallowed characters
+    // in a non-Punycode label or U+FFFD itself in a Punycode label.
+    // We also check for dots which can come from the input to a single-label function.
+    // Ok to cast away const because we own the UnicodeString.
+    char16_t *s=(char16_t *)label;
+    const char16_t *limit=label+labelLength;
+    char16_t oredChars=0;
+    // If we enforce STD3 rules, then ASCII characters other than LDH and dot are disallowed.
+    UBool disallowNonLDHDot=(options&UIDNA_USE_STD3_RULES)!=0;
+    do {
+        char16_t c=*s;
+        if(c<=0x7f) {
+            if(c==0x2e) {
+                info.labelErrors|=UIDNA_ERROR_LABEL_HAS_DOT;
+                *s=0xfffd;
+            } else if(disallowNonLDHDot && asciiData[c]<0) {
+                info.labelErrors|=UIDNA_ERROR_DISALLOWED;
+                *s=0xfffd;
+            }
+        } else {
+            oredChars|=c;
+            if(disallowNonLDHDot && isNonASCIIDisallowedSTD3Valid(c)) {
+                info.labelErrors|=UIDNA_ERROR_DISALLOWED;
+                *s=0xfffd;
+            } else if(c==0xfffd) {
+                info.labelErrors|=UIDNA_ERROR_DISALLOWED;
+            }
+        }
+        ++s;
+    } while(sreplace(labelStart, cpLength, (char16_t)0xfffd);
+        label=labelString->getBuffer()+labelStart;
+        labelLength+=1-cpLength;
+        if(labelString==&dest) {
+            destLabelLength=labelLength;
+        }
+    }
+    if((info.labelErrors&severeErrors)==0) {
+        // Do contextual checks only if we do not have U+FFFD from a severe error
+        // because U+FFFD can make these checks fail.
+        if((options&UIDNA_CHECK_BIDI)!=0 && (!info.isBiDi || info.isOkBiDi)) {
+            checkLabelBiDi(label, labelLength, info);
+        }
+        if( (options&UIDNA_CHECK_CONTEXTJ)!=0 && (oredChars&0x200c)==0x200c &&
+            !isLabelOkContextJ(label, labelLength)
+        ) {
+            info.labelErrors|=UIDNA_ERROR_CONTEXTJ;
+        }
+        if((options&UIDNA_CHECK_CONTEXTO)!=0 && oredChars>=0xb7) {
+            checkLabelContextO(label, labelLength, info);
+        }
+        if(toASCII) {
+            if(wasPunycode) {
+                // Leave a Punycode label unchanged if it has no severe errors.
+                if(destLabelLength>63) {
+                    info.labelErrors|=UIDNA_ERROR_LABEL_TOO_LONG;
+                }
+                return destLabelLength;
+            } else if(oredChars>=0x80) {
+                // Contains non-ASCII characters.
+                UnicodeString punycode;
+                char16_t *buffer=punycode.getBuffer(63);  // 63==maximum DNS label length
+                if(buffer==nullptr) {
+                    errorCode=U_MEMORY_ALLOCATION_ERROR;
+                    return destLabelLength;
+                }
+                buffer[0]=0x78;  // Write "xn--".
+                buffer[1]=0x6e;
+                buffer[2]=0x2d;
+                buffer[3]=0x2d;
+                int32_t punycodeLength=u_strToPunycode(label, labelLength,
+                                                      buffer+4, punycode.getCapacity()-4,
+                                                      nullptr, &errorCode);
+                if(errorCode==U_BUFFER_OVERFLOW_ERROR) {
+                    errorCode=U_ZERO_ERROR;
+                    punycode.releaseBuffer(4);
+                    buffer=punycode.getBuffer(4+punycodeLength);
+                    if(buffer==nullptr) {
+                        errorCode=U_MEMORY_ALLOCATION_ERROR;
+                        return destLabelLength;
+                    }
+                    punycodeLength=u_strToPunycode(label, labelLength,
+                                                  buffer+4, punycode.getCapacity()-4,
+                                                  nullptr, &errorCode);
+                }
+                punycodeLength+=4;
+                punycode.releaseBuffer(punycodeLength);
+                if(U_FAILURE(errorCode)) {
+                    return destLabelLength;
+                }
+                if(punycodeLength>63) {
+                    info.labelErrors|=UIDNA_ERROR_LABEL_TOO_LONG;
+                }
+                return replaceLabel(dest, destLabelStart, destLabelLength,
+                                    punycode, punycodeLength, errorCode);
+            } else {
+                // all-ASCII label
+                if(labelLength>63) {
+                    info.labelErrors|=UIDNA_ERROR_LABEL_TOO_LONG;
+                }
+            }
+        }
+    } else {
+        // If a Punycode label has severe errors,
+        // then leave it but make sure it does not look valid.
+        if(wasPunycode) {
+            info.labelErrors|=UIDNA_ERROR_INVALID_ACE_LABEL;
+            return markBadACELabel(dest, destLabelStart, destLabelLength, toASCII, info, errorCode);
+        }
+    }
+    return replaceLabel(dest, destLabelStart, destLabelLength,
+                        *labelString, labelLength, errorCode);
+}
+
+// Make sure an ACE label does not look valid.
+// Append U+FFFD if the label has only LDH characters.
+// If UIDNA_USE_STD3_RULES, also replace disallowed ASCII characters with U+FFFD.
+int32_t
+UTS46::markBadACELabel(UnicodeString &dest,
+                       int32_t labelStart, int32_t labelLength,
+                       UBool toASCII, IDNAInfo &info, UErrorCode &errorCode) const {
+    if(U_FAILURE(errorCode)) {
+        return 0;
+    }
+    UBool disallowNonLDHDot=(options&UIDNA_USE_STD3_RULES)!=0;
+    UBool isASCII=true;
+    UBool onlyLDH=true;
+    const char16_t *label=dest.getBuffer()+labelStart;
+    const char16_t *limit=label+labelLength;
+    // Start after the initial "xn--".
+    // Ok to cast away const because we own the UnicodeString.
+    for(char16_t *s=const_cast(label+4); s63) {
+            info.labelErrors|=UIDNA_ERROR_LABEL_TOO_LONG;
+        }
+    }
+    return labelLength;
+}
+
+const uint32_t L_MASK=U_MASK(U_LEFT_TO_RIGHT);
+const uint32_t R_AL_MASK=U_MASK(U_RIGHT_TO_LEFT)|U_MASK(U_RIGHT_TO_LEFT_ARABIC);
+const uint32_t L_R_AL_MASK=L_MASK|R_AL_MASK;
+
+const uint32_t R_AL_AN_MASK=R_AL_MASK|U_MASK(U_ARABIC_NUMBER);
+
+const uint32_t EN_AN_MASK=U_MASK(U_EUROPEAN_NUMBER)|U_MASK(U_ARABIC_NUMBER);
+const uint32_t R_AL_EN_AN_MASK=R_AL_MASK|EN_AN_MASK;
+const uint32_t L_EN_MASK=L_MASK|U_MASK(U_EUROPEAN_NUMBER);
+
+const uint32_t ES_CS_ET_ON_BN_NSM_MASK=
+    U_MASK(U_EUROPEAN_NUMBER_SEPARATOR)|
+    U_MASK(U_COMMON_NUMBER_SEPARATOR)|
+    U_MASK(U_EUROPEAN_NUMBER_TERMINATOR)|
+    U_MASK(U_OTHER_NEUTRAL)|
+    U_MASK(U_BOUNDARY_NEUTRAL)|
+    U_MASK(U_DIR_NON_SPACING_MARK);
+const uint32_t L_EN_ES_CS_ET_ON_BN_NSM_MASK=L_EN_MASK|ES_CS_ET_ON_BN_NSM_MASK;
+const uint32_t R_AL_AN_EN_ES_CS_ET_ON_BN_NSM_MASK=R_AL_MASK|EN_AN_MASK|ES_CS_ET_ON_BN_NSM_MASK;
+
+// We scan the whole label and check both for whether it contains RTL characters
+// and whether it passes the BiDi Rule.
+// In a BiDi domain name, all labels must pass the BiDi Rule, but we might find
+// that a domain name is a BiDi domain name (has an RTL label) only after
+// processing several earlier labels.
+void
+UTS46::checkLabelBiDi(const char16_t *label, int32_t labelLength, IDNAInfo &info) const {
+    // IDNA2008 BiDi rule
+    // Get the directionality of the first character.
+    UChar32 c;
+    int32_t i=0;
+    U16_NEXT_UNSAFE(label, i, c);
+    uint32_t firstMask=U_MASK(u_charDirection(c));
+    // 1. The first character must be a character with BIDI property L, R
+    // or AL.  If it has the R or AL property, it is an RTL label; if it
+    // has the L property, it is an LTR label.
+    if((firstMask&~L_R_AL_MASK)!=0) {
+        info.isOkBiDi=false;
+    }
+    // Get the directionality of the last non-NSM character.
+    uint32_t lastMask;
+    for(;;) {
+        if(i>=labelLength) {
+            lastMask=firstMask;
+            break;
+        }
+        U16_PREV_UNSAFE(label, labelLength, c);
+        UCharDirection dir=u_charDirection(c);
+        if(dir!=U_DIR_NON_SPACING_MARK) {
+            lastMask=U_MASK(dir);
+            break;
+        }
+    }
+    // 3. In an RTL label, the end of the label must be a character with
+    // BIDI property R, AL, EN or AN, followed by zero or more
+    // characters with BIDI property NSM.
+    // 6. In an LTR label, the end of the label must be a character with
+    // BIDI property L or EN, followed by zero or more characters with
+    // BIDI property NSM.
+    if( (firstMask&L_MASK)!=0 ?
+            (lastMask&~L_EN_MASK)!=0 :
+            (lastMask&~R_AL_EN_AN_MASK)!=0
+    ) {
+        info.isOkBiDi=false;
+    }
+    // Add the directionalities of the intervening characters.
+    uint32_t mask=firstMask|lastMask;
+    while(ilabelStart) {
+                c=s[i-1];
+                if(!(0x61<=c && c<=0x7a) && !(0x30<=c && c<=0x39)) {
+                    // Last character in the label is not an L or EN.
+                    return false;
+                }
+            }
+            labelStart=i+1;
+        } else if(i==labelStart) {
+            if(!(0x61<=c && c<=0x7a)) {
+                // First character in the label is not an L.
+                return false;
+            }
+        } else {
+            if(c<=0x20 && (c>=0x1c || (9<=c && c<=0xd))) {
+                // Intermediate character in the label is a B, S or WS.
+                return false;
+            }
+        }
+    }
+    return true;
+}
+
+// UTF-8 version, called for source ASCII prefix.
+// Can contain uppercase A-Z.
+// s[length-1] must be the trailing dot.
+static UBool
+isASCIIOkBiDi(const char *s, int32_t length) {
+    int32_t labelStart=0;
+    for(int32_t i=0; ilabelStart) {
+                c=s[i-1];
+                if(!(0x61<=c && c<=0x7a) && !(0x41<=c && c<=0x5a) && !(0x30<=c && c<=0x39)) {
+                    // Last character in the label is not an L or EN.
+                    return false;
+                }
+            }
+            labelStart=i+1;
+        } else if(i==labelStart) {
+            if(!(0x61<=c && c<=0x7a) && !(0x41<=c && c<=0x5a)) {
+                // First character in the label is not an L.
+                return false;
+            }
+        } else {
+            if(c<=0x20 && (c>=0x1c || (9<=c && c<=0xd))) {
+                // Intermediate character in the label is a B, S or WS.
+                return false;
+            }
+        }
+    }
+    return true;
+}
+
+UBool
+UTS46::isLabelOkContextJ(const char16_t *label, int32_t labelLength) const {
+    // [IDNA2008-Tables]
+    // 200C..200D  ; CONTEXTJ    # ZERO WIDTH NON-JOINER..ZERO WIDTH JOINER
+    for(int32_t i=0; i0) {
+                        info.labelErrors|=UIDNA_ERROR_CONTEXTO_DIGITS;
+                    }
+                    arabicDigits=-1;
+                } else if(0x6f0<=c) {
+                    if(arabicDigits<0) {
+                        info.labelErrors|=UIDNA_ERROR_CONTEXTO_DIGITS;
+                    }
+                    arabicDigits=1;
+                }
+            }
+        } else if(c==0x30fb) {
+            // Appendix A.7. KATAKANA MIDDLE DOT (U+30FB)
+            // Rule Set:
+            //  False;
+            //  For All Characters:
+            //    If Script(cp) .in. {Hiragana, Katakana, Han} Then True;
+            //  End For;
+            UErrorCode errorCode=U_ZERO_ERROR;
+            for(int j=0;;) {
+                if(j>labelEnd) {
+                    info.labelErrors|=UIDNA_ERROR_CONTEXTO_PUNCTUATION;
+                    break;
+                }
+                U16_NEXT(label, j, labelLength, c);
+                UScriptCode script=uscript_getScript(c, &errorCode);
+                if(script==USCRIPT_HIRAGANA || script==USCRIPT_KATAKANA || script==USCRIPT_HAN) {
+                    break;
+                }
+            }
+        }
+    }
+}
+
+U_NAMESPACE_END
+
+// C API ------------------------------------------------------------------- ***
+
+U_NAMESPACE_USE
+
+U_CAPI UIDNA * U_EXPORT2
+uidna_openUTS46(uint32_t options, UErrorCode *pErrorCode) {
+    return reinterpret_cast(IDNA::createUTS46Instance(options, *pErrorCode));
+}
+
+U_CAPI void U_EXPORT2
+uidna_close(UIDNA *idna) {
+    delete reinterpret_cast(idna);
+}
+
+static UBool
+checkArgs(const void *label, int32_t length,
+          void *dest, int32_t capacity,
+          UIDNAInfo *pInfo, UErrorCode *pErrorCode) {
+    if(U_FAILURE(*pErrorCode)) {
+        return false;
+    }
+    // sizeof(UIDNAInfo)=16 in the first API version.
+    if(pInfo==nullptr || pInfo->size<16) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return false;
+    }
+    if( (label==nullptr ? length!=0 : length<-1) ||
+        (dest==nullptr ? capacity!=0 : capacity<0) ||
+        (dest==label && label!=nullptr)
+    ) {
+        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+        return false;
+    }
+    // Set all *pInfo bytes to 0 except for the size field itself.
+    uprv_memset(&pInfo->size+1, 0, pInfo->size-sizeof(pInfo->size));
+    return true;
+}
+
+static void
+idnaInfoToStruct(IDNAInfo &info, UIDNAInfo *pInfo) {
+    pInfo->isTransitionalDifferent=info.isTransitionalDifferent();
+    pInfo->errors=info.getErrors();
+}
+
+U_CAPI int32_t U_EXPORT2
+uidna_labelToASCII(const UIDNA *idna,
+                   const char16_t *label, int32_t length,
+                   char16_t *dest, int32_t capacity,
+                   UIDNAInfo *pInfo, UErrorCode *pErrorCode) {
+    if(!checkArgs(label, length, dest, capacity, pInfo, pErrorCode)) {
+        return 0;
+    }
+    UnicodeString src((UBool)(length<0), label, length);
+    UnicodeString destString(dest, 0, capacity);
+    IDNAInfo info;
+    reinterpret_cast(idna)->labelToASCII(src, destString, info, *pErrorCode);
+    idnaInfoToStruct(info, pInfo);
+    return destString.extract(dest, capacity, *pErrorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+uidna_labelToUnicode(const UIDNA *idna,
+                     const char16_t *label, int32_t length,
+                     char16_t *dest, int32_t capacity,
+                     UIDNAInfo *pInfo, UErrorCode *pErrorCode) {
+    if(!checkArgs(label, length, dest, capacity, pInfo, pErrorCode)) {
+        return 0;
+    }
+    UnicodeString src((UBool)(length<0), label, length);
+    UnicodeString destString(dest, 0, capacity);
+    IDNAInfo info;
+    reinterpret_cast(idna)->labelToUnicode(src, destString, info, *pErrorCode);
+    idnaInfoToStruct(info, pInfo);
+    return destString.extract(dest, capacity, *pErrorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+uidna_nameToASCII(const UIDNA *idna,
+                  const char16_t *name, int32_t length,
+                  char16_t *dest, int32_t capacity,
+                  UIDNAInfo *pInfo, UErrorCode *pErrorCode) {
+    if(!checkArgs(name, length, dest, capacity, pInfo, pErrorCode)) {
+        return 0;
+    }
+    UnicodeString src((UBool)(length<0), name, length);
+    UnicodeString destString(dest, 0, capacity);
+    IDNAInfo info;
+    reinterpret_cast(idna)->nameToASCII(src, destString, info, *pErrorCode);
+    idnaInfoToStruct(info, pInfo);
+    return destString.extract(dest, capacity, *pErrorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+uidna_nameToUnicode(const UIDNA *idna,
+                    const char16_t *name, int32_t length,
+                    char16_t *dest, int32_t capacity,
+                    UIDNAInfo *pInfo, UErrorCode *pErrorCode) {
+    if(!checkArgs(name, length, dest, capacity, pInfo, pErrorCode)) {
+        return 0;
+    }
+    UnicodeString src((UBool)(length<0), name, length);
+    UnicodeString destString(dest, 0, capacity);
+    IDNAInfo info;
+    reinterpret_cast(idna)->nameToUnicode(src, destString, info, *pErrorCode);
+    idnaInfoToStruct(info, pInfo);
+    return destString.extract(dest, capacity, *pErrorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+uidna_labelToASCII_UTF8(const UIDNA *idna,
+                        const char *label, int32_t length,
+                        char *dest, int32_t capacity,
+                        UIDNAInfo *pInfo, UErrorCode *pErrorCode) {
+    if(!checkArgs(label, length, dest, capacity, pInfo, pErrorCode)) {
+        return 0;
+    }
+    StringPiece src(label, length<0 ? static_cast(uprv_strlen(label)) : length);
+    CheckedArrayByteSink sink(dest, capacity);
+    IDNAInfo info;
+    reinterpret_cast(idna)->labelToASCII_UTF8(src, sink, info, *pErrorCode);
+    idnaInfoToStruct(info, pInfo);
+    return u_terminateChars(dest, capacity, sink.NumberOfBytesAppended(), pErrorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+uidna_labelToUnicodeUTF8(const UIDNA *idna,
+                         const char *label, int32_t length,
+                         char *dest, int32_t capacity,
+                         UIDNAInfo *pInfo, UErrorCode *pErrorCode) {
+    if(!checkArgs(label, length, dest, capacity, pInfo, pErrorCode)) {
+        return 0;
+    }
+    StringPiece src(label, length<0 ? static_cast(uprv_strlen(label)) : length);
+    CheckedArrayByteSink sink(dest, capacity);
+    IDNAInfo info;
+    reinterpret_cast(idna)->labelToUnicodeUTF8(src, sink, info, *pErrorCode);
+    idnaInfoToStruct(info, pInfo);
+    return u_terminateChars(dest, capacity, sink.NumberOfBytesAppended(), pErrorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+uidna_nameToASCII_UTF8(const UIDNA *idna,
+                       const char *name, int32_t length,
+                       char *dest, int32_t capacity,
+                       UIDNAInfo *pInfo, UErrorCode *pErrorCode) {
+    if(!checkArgs(name, length, dest, capacity, pInfo, pErrorCode)) {
+        return 0;
+    }
+    StringPiece src(name, length<0 ? static_cast(uprv_strlen(name)) : length);
+    CheckedArrayByteSink sink(dest, capacity);
+    IDNAInfo info;
+    reinterpret_cast(idna)->nameToASCII_UTF8(src, sink, info, *pErrorCode);
+    idnaInfoToStruct(info, pInfo);
+    return u_terminateChars(dest, capacity, sink.NumberOfBytesAppended(), pErrorCode);
+}
+
+U_CAPI int32_t U_EXPORT2
+uidna_nameToUnicodeUTF8(const UIDNA *idna,
+                        const char *name, int32_t length,
+                        char *dest, int32_t capacity,
+                        UIDNAInfo *pInfo, UErrorCode *pErrorCode) {
+    if(!checkArgs(name, length, dest, capacity, pInfo, pErrorCode)) {
+        return 0;
+    }
+    StringPiece src(name, length<0 ? static_cast(uprv_strlen(name)) : length);
+    CheckedArrayByteSink sink(dest, capacity);
+    IDNAInfo info;
+    reinterpret_cast(idna)->nameToUnicodeUTF8(src, sink, info, *pErrorCode);
+    idnaInfoToStruct(info, pInfo);
+    return u_terminateChars(dest, capacity, sink.NumberOfBytesAppended(), pErrorCode);
+}
+
+#endif  // UCONFIG_NO_IDNA
diff --git a/intl/icu/source/common/utypeinfo.h b/intl/icu/source/common/utypeinfo.h
new file mode 100644
index 0000000000..c6663734fc
--- /dev/null
+++ b/intl/icu/source/common/utypeinfo.h
@@ -0,0 +1,32 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+*
+*   Copyright (C) 2012-2016, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+******************************************************************************
+*/
+
+#ifndef __UTYPEINFO_H__
+#define __UTYPEINFO_H__
+
+// Windows header  does not define 'exception' in 'std' namespace.
+// Therefore, a project using ICU cannot be compiled with _HAS_EXCEPTIONS
+// set to 0 on Windows with Visual Studio. To work around that, we have to
+// include  explicitly and add using statement below.
+// Whenever 'typeid' is used, this header has to be included
+// instead of .
+// Visual Studio 10 emits warning 4275 with this change. If you compile
+// with exception disabled, you have to suppress warning 4275.
+#if defined(_MSC_VER) && _HAS_EXCEPTIONS == 0
+#include 
+using std::exception;
+#endif
+#if defined(__GLIBCXX__)
+namespace std { class type_info; } // WORKAROUND: http://llvm.org/bugs/show_bug.cgi?id=13364
+#endif
+#include   // for 'typeid' to work
+
+#endif
diff --git a/intl/icu/source/common/utypes.cpp b/intl/icu/source/common/utypes.cpp
new file mode 100644
index 0000000000..63e05b1249
--- /dev/null
+++ b/intl/icu/source/common/utypes.cpp
@@ -0,0 +1,227 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+*
+*   Copyright (C) 1997-2015, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+******************************************************************************
+*
+*  FILE NAME : utypes.c (previously putil.c)
+*
+*   Date        Name        Description
+*   10/07/2004  grhoten     split from putil.c
+******************************************************************************
+*/
+
+#include "unicode/utypes.h"
+
+/* u_errorName() ------------------------------------------------------------ */
+
+static const char * const
+_uErrorInfoName[U_ERROR_WARNING_LIMIT-U_ERROR_WARNING_START]={
+    "U_USING_FALLBACK_WARNING",
+    "U_USING_DEFAULT_WARNING",
+    "U_SAFECLONE_ALLOCATED_WARNING",
+    "U_STATE_OLD_WARNING",
+    "U_STRING_NOT_TERMINATED_WARNING",
+    "U_SORT_KEY_TOO_SHORT_WARNING",
+    "U_AMBIGUOUS_ALIAS_WARNING",
+    "U_DIFFERENT_UCA_VERSION",
+    "U_PLUGIN_CHANGED_LEVEL_WARNING",
+};
+
+static const char * const
+_uTransErrorName[U_PARSE_ERROR_LIMIT - U_PARSE_ERROR_START]={
+    "U_BAD_VARIABLE_DEFINITION",
+    "U_MALFORMED_RULE",
+    "U_MALFORMED_SET",
+    "U_MALFORMED_SYMBOL_REFERENCE",
+    "U_MALFORMED_UNICODE_ESCAPE",
+    "U_MALFORMED_VARIABLE_DEFINITION",
+    "U_MALFORMED_VARIABLE_REFERENCE",
+    "U_MISMATCHED_SEGMENT_DELIMITERS",
+    "U_MISPLACED_ANCHOR_START",
+    "U_MISPLACED_CURSOR_OFFSET",
+    "U_MISPLACED_QUANTIFIER",
+    "U_MISSING_OPERATOR",
+    "U_MISSING_SEGMENT_CLOSE",
+    "U_MULTIPLE_ANTE_CONTEXTS",
+    "U_MULTIPLE_CURSORS",
+    "U_MULTIPLE_POST_CONTEXTS",
+    "U_TRAILING_BACKSLASH",
+    "U_UNDEFINED_SEGMENT_REFERENCE",
+    "U_UNDEFINED_VARIABLE",
+    "U_UNQUOTED_SPECIAL",
+    "U_UNTERMINATED_QUOTE",
+    "U_RULE_MASK_ERROR",
+    "U_MISPLACED_COMPOUND_FILTER",
+    "U_MULTIPLE_COMPOUND_FILTERS",
+    "U_INVALID_RBT_SYNTAX",
+    "U_INVALID_PROPERTY_PATTERN",
+    "U_MALFORMED_PRAGMA",
+    "U_UNCLOSED_SEGMENT",
+    "U_ILLEGAL_CHAR_IN_SEGMENT",
+    "U_VARIABLE_RANGE_EXHAUSTED",
+    "U_VARIABLE_RANGE_OVERLAP",
+    "U_ILLEGAL_CHARACTER",
+    "U_INTERNAL_TRANSLITERATOR_ERROR",
+    "U_INVALID_ID",
+    "U_INVALID_FUNCTION"
+};
+
+static const char * const
+_uErrorName[U_STANDARD_ERROR_LIMIT]={
+    "U_ZERO_ERROR",
+
+    "U_ILLEGAL_ARGUMENT_ERROR",
+    "U_MISSING_RESOURCE_ERROR",
+    "U_INVALID_FORMAT_ERROR",
+    "U_FILE_ACCESS_ERROR",
+    "U_INTERNAL_PROGRAM_ERROR",
+    "U_MESSAGE_PARSE_ERROR",
+    "U_MEMORY_ALLOCATION_ERROR",
+    "U_INDEX_OUTOFBOUNDS_ERROR",
+    "U_PARSE_ERROR",
+    "U_INVALID_CHAR_FOUND",
+    "U_TRUNCATED_CHAR_FOUND",
+    "U_ILLEGAL_CHAR_FOUND",
+    "U_INVALID_TABLE_FORMAT",
+    "U_INVALID_TABLE_FILE",
+    "U_BUFFER_OVERFLOW_ERROR",
+    "U_UNSUPPORTED_ERROR",
+    "U_RESOURCE_TYPE_MISMATCH",
+    "U_ILLEGAL_ESCAPE_SEQUENCE",
+    "U_UNSUPPORTED_ESCAPE_SEQUENCE",
+    "U_NO_SPACE_AVAILABLE",
+    "U_CE_NOT_FOUND_ERROR",
+    "U_PRIMARY_TOO_LONG_ERROR",
+    "U_STATE_TOO_OLD_ERROR",
+    "U_TOO_MANY_ALIASES_ERROR",
+    "U_ENUM_OUT_OF_SYNC_ERROR",
+    "U_INVARIANT_CONVERSION_ERROR",
+    "U_INVALID_STATE_ERROR",
+    "U_COLLATOR_VERSION_MISMATCH",
+    "U_USELESS_COLLATOR_ERROR",
+    "U_NO_WRITE_PERMISSION",
+    "U_INPUT_TOO_LONG_ERROR"
+};
+static const char * const
+_uFmtErrorName[U_FMT_PARSE_ERROR_LIMIT - U_FMT_PARSE_ERROR_START] = {
+    "U_UNEXPECTED_TOKEN",
+    "U_MULTIPLE_DECIMAL_SEPARATORS",
+    "U_MULTIPLE_EXPONENTIAL_SYMBOLS",
+    "U_MALFORMED_EXPONENTIAL_PATTERN",
+    "U_MULTIPLE_PERCENT_SYMBOLS",
+    "U_MULTIPLE_PERMILL_SYMBOLS",
+    "U_MULTIPLE_PAD_SPECIFIERS",
+    "U_PATTERN_SYNTAX_ERROR",
+    "U_ILLEGAL_PAD_POSITION",
+    "U_UNMATCHED_BRACES",
+    "U_UNSUPPORTED_PROPERTY",
+    "U_UNSUPPORTED_ATTRIBUTE",
+    "U_ARGUMENT_TYPE_MISMATCH",
+    "U_DUPLICATE_KEYWORD",
+    "U_UNDEFINED_KEYWORD",
+    "U_DEFAULT_KEYWORD_MISSING",
+    "U_DECIMAL_NUMBER_SYNTAX_ERROR",
+    "U_FORMAT_INEXACT_ERROR",
+    "U_NUMBER_ARG_OUTOFBOUNDS_ERROR",
+    "U_NUMBER_SKELETON_SYNTAX_ERROR",
+};
+
+static const char * const
+_uBrkErrorName[U_BRK_ERROR_LIMIT - U_BRK_ERROR_START] = {
+    "U_BRK_INTERNAL_ERROR",
+    "U_BRK_HEX_DIGITS_EXPECTED",
+    "U_BRK_SEMICOLON_EXPECTED",
+    "U_BRK_RULE_SYNTAX",
+    "U_BRK_UNCLOSED_SET",
+    "U_BRK_ASSIGN_ERROR",
+    "U_BRK_VARIABLE_REDFINITION",
+    "U_BRK_MISMATCHED_PAREN",
+    "U_BRK_NEW_LINE_IN_QUOTED_STRING",
+    "U_BRK_UNDEFINED_VARIABLE",
+    "U_BRK_INIT_ERROR",
+    "U_BRK_RULE_EMPTY_SET",
+    "U_BRK_UNRECOGNIZED_OPTION",
+    "U_BRK_MALFORMED_RULE_TAG"
+};
+
+static const char * const
+_uRegexErrorName[U_REGEX_ERROR_LIMIT - U_REGEX_ERROR_START] = {
+    "U_REGEX_INTERNAL_ERROR",
+    "U_REGEX_RULE_SYNTAX",
+    "U_REGEX_INVALID_STATE",
+    "U_REGEX_BAD_ESCAPE_SEQUENCE",
+    "U_REGEX_PROPERTY_SYNTAX",
+    "U_REGEX_UNIMPLEMENTED",
+    "U_REGEX_MISMATCHED_PAREN",
+    "U_REGEX_NUMBER_TOO_BIG",
+    "U_REGEX_BAD_INTERVAL",
+    "U_REGEX_MAX_LT_MIN",
+    "U_REGEX_INVALID_BACK_REF",
+    "U_REGEX_INVALID_FLAG",
+    "U_REGEX_LOOK_BEHIND_LIMIT",
+    "U_REGEX_SET_CONTAINS_STRING",
+    "U_REGEX_OCTAL_TOO_BIG",
+    "U_REGEX_MISSING_CLOSE_BRACKET",
+    "U_REGEX_INVALID_RANGE",
+    "U_REGEX_STACK_OVERFLOW",
+    "U_REGEX_TIME_OUT",
+    "U_REGEX_STOPPED_BY_CALLER",
+    "U_REGEX_PATTERN_TOO_BIG",
+    "U_REGEX_INVALID_CAPTURE_GROUP_NAME"
+};
+
+static const char * const
+_uIDNAErrorName[U_IDNA_ERROR_LIMIT - U_IDNA_ERROR_START] = {
+     "U_STRINGPREP_PROHIBITED_ERROR",
+     "U_STRINGPREP_UNASSIGNED_ERROR",
+     "U_STRINGPREP_CHECK_BIDI_ERROR",
+     "U_IDNA_STD3_ASCII_RULES_ERROR",
+     "U_IDNA_ACE_PREFIX_ERROR",
+     "U_IDNA_VERIFICATION_ERROR",
+     "U_IDNA_LABEL_TOO_LONG_ERROR",
+     "U_IDNA_ZERO_LENGTH_LABEL_ERROR",
+     "U_IDNA_DOMAIN_NAME_TOO_LONG_ERROR"
+};
+
+static const char * const
+_uPluginErrorName[U_PLUGIN_ERROR_LIMIT - U_PLUGIN_ERROR_START] = {
+     "U_PLUGIN_TOO_HIGH",
+     "U_PLUGIN_DIDNT_SET_LEVEL",
+};
+
+U_CAPI const char * U_EXPORT2
+u_errorName(UErrorCode code) {
+    if(U_ZERO_ERROR <= code && code < U_STANDARD_ERROR_LIMIT) {
+        return _uErrorName[code];
+    } else if(U_ERROR_WARNING_START <= code && code < U_ERROR_WARNING_LIMIT) {
+        return _uErrorInfoName[code - U_ERROR_WARNING_START];
+    } else if(U_PARSE_ERROR_START <= code && code < U_PARSE_ERROR_LIMIT){
+        return _uTransErrorName[code - U_PARSE_ERROR_START];
+    } else if(U_FMT_PARSE_ERROR_START <= code && code < U_FMT_PARSE_ERROR_LIMIT){
+        return _uFmtErrorName[code - U_FMT_PARSE_ERROR_START];
+    } else if (U_BRK_ERROR_START <= code  && code < U_BRK_ERROR_LIMIT){
+        return _uBrkErrorName[code - U_BRK_ERROR_START];
+    } else if (U_REGEX_ERROR_START <= code && code < U_REGEX_ERROR_LIMIT) {
+        return _uRegexErrorName[code - U_REGEX_ERROR_START];
+    } else if(U_IDNA_ERROR_START <= code && code < U_IDNA_ERROR_LIMIT) {
+        return _uIDNAErrorName[code - U_IDNA_ERROR_START];
+    } else if(U_PLUGIN_ERROR_START <= code && code < U_PLUGIN_ERROR_LIMIT) {
+        return _uPluginErrorName[code - U_PLUGIN_ERROR_START];
+    } else {
+        return "[BOGUS UErrorCode]";
+    }
+}
+
+/*
+ * Hey, Emacs, please set the following:
+ *
+ * Local Variables:
+ * indent-tabs-mode: nil
+ * End:
+ *
+ */
diff --git a/intl/icu/source/common/uvector.cpp b/intl/icu/source/common/uvector.cpp
new file mode 100644
index 0000000000..f93d73a3c6
--- /dev/null
+++ b/intl/icu/source/common/uvector.cpp
@@ -0,0 +1,566 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+* Copyright (C) 1999-2013, International Business Machines Corporation and
+* others. All Rights Reserved.
+******************************************************************************
+*   Date        Name        Description
+*   10/22/99    alan        Creation.
+**********************************************************************
+*/
+
+#include "uvector.h"
+#include "cmemory.h"
+#include "uarrsort.h"
+#include "uelement.h"
+
+U_NAMESPACE_BEGIN
+
+constexpr int32_t DEFAULT_CAPACITY = 8;
+
+/*
+ * Constants for hinting whether a key is an integer
+ * or a pointer.  If a hint bit is zero, then the associated
+ * token is assumed to be an integer. This is needed for iSeries
+ */
+constexpr int8_t HINT_KEY_POINTER = 1;
+constexpr int8_t HINT_KEY_INTEGER = 0;
+ 
+UOBJECT_DEFINE_RTTI_IMPLEMENTATION(UVector)
+
+UVector::UVector(UErrorCode &status) :
+        UVector(nullptr, nullptr, DEFAULT_CAPACITY, status) {
+}
+
+UVector::UVector(int32_t initialCapacity, UErrorCode &status) :
+        UVector(nullptr, nullptr, initialCapacity, status) {
+}
+
+UVector::UVector(UObjectDeleter *d, UElementsAreEqual *c, UErrorCode &status) :
+        UVector(d, c, DEFAULT_CAPACITY, status) {
+}
+
+UVector::UVector(UObjectDeleter *d, UElementsAreEqual *c, int32_t initialCapacity, UErrorCode &status) :
+    deleter(d),
+    comparer(c)
+{
+    if (U_FAILURE(status)) {
+        return;
+    }
+    // Fix bogus initialCapacity values; avoid malloc(0) and integer overflow
+    if ((initialCapacity < 1) || (initialCapacity > (int32_t)(INT32_MAX / sizeof(UElement)))) {
+        initialCapacity = DEFAULT_CAPACITY;
+    }
+    elements = (UElement *)uprv_malloc(sizeof(UElement)*initialCapacity);
+    if (elements == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+    } else {
+        capacity = initialCapacity;
+    }
+}
+
+UVector::~UVector() {
+    removeAllElements();
+    uprv_free(elements);
+    elements = nullptr;
+}
+
+/**
+ * Assign this object to another (make this a copy of 'other').
+ * Use the 'assign' function to assign each element.
+ */
+void UVector::assign(const UVector& other, UElementAssigner *assign, UErrorCode &ec) {
+    if (ensureCapacity(other.count, ec)) {
+        setSize(other.count, ec);
+        if (U_SUCCESS(ec)) {
+            for (int32_t i=0; iindex; --i) {
+                elements[i] = elements[i-1];
+            }
+            elements[index].pointer = obj;
+            ++count;
+        } else {
+            /* index out of range */
+            status = U_ILLEGAL_ARGUMENT_ERROR;
+        }
+    }
+    if (U_FAILURE(status) && deleter != nullptr) {
+        (*deleter)(obj);
+    }
+}
+
+void UVector::insertElementAt(int32_t elem, int32_t index, UErrorCode &status) {
+    U_ASSERT(deleter == nullptr);  // Usage error. Mixing up ints and pointers.
+    // must have 0 <= index <= count
+    if (ensureCapacity(count + 1, status)) {
+        if (0 <= index && index <= count) {
+            for (int32_t i=count; i>index; --i) {
+                elements[i] = elements[i-1];
+            }
+            elements[index].pointer = nullptr;
+            elements[index].integer = elem;
+            ++count;
+        } else {
+            /* index out of range */
+            status = U_ILLEGAL_ARGUMENT_ERROR;
+        }
+    }
+}
+
+void* UVector::elementAt(int32_t index) const {
+    return (0 <= index && index < count) ? elements[index].pointer : 0;
+}
+
+int32_t UVector::elementAti(int32_t index) const {
+    return (0 <= index && index < count) ? elements[index].integer : 0;
+}
+
+UBool UVector::containsAll(const UVector& other) const {
+    for (int32_t i=0; i= 0) {
+            return false;
+        }
+    }
+    return true;
+}
+
+UBool UVector::removeAll(const UVector& other) {
+    UBool changed = false;
+    for (int32_t i=0; i= 0) {
+            removeElementAt(j);
+            changed = true;
+        }
+    }
+    return changed;
+}
+
+UBool UVector::retainAll(const UVector& other) {
+    UBool changed = false;
+    for (int32_t j=size()-1; j>=0; --j) {
+        int32_t i = other.indexOf(elements[j]);
+        if (i < 0) {
+            removeElementAt(j);
+            changed = true;
+        }
+    }
+    return changed;
+}
+
+void UVector::removeElementAt(int32_t index) {
+    void* e = orphanElementAt(index);
+    if (e != nullptr && deleter != nullptr) {
+        (*deleter)(e);
+    }
+}
+
+UBool UVector::removeElement(void* obj) {
+    int32_t i = indexOf(obj);
+    if (i >= 0) {
+        removeElementAt(i);
+        return true;
+    }
+    return false;
+}
+
+void UVector::removeAllElements() {
+    if (deleter != nullptr) {
+        for (int32_t i=0; icount != other.count) {
+        return false;
+    }
+    if (comparer == nullptr) {
+        for (i=0; i (INT32_MAX - 1) / 2) {        	// integer overflow check
+            status = U_ILLEGAL_ARGUMENT_ERROR;
+            return false;
+        }
+        int32_t newCap = capacity * 2;
+        if (newCap < minimumCapacity) {
+            newCap = minimumCapacity;
+        }
+        if (newCap > (int32_t)(INT32_MAX / sizeof(UElement))) {	// integer overflow check
+            // We keep the original memory contents on bad minimumCapacity.
+            status = U_ILLEGAL_ARGUMENT_ERROR;
+            return false;
+        }
+        UElement* newElems = (UElement *)uprv_realloc(elements, sizeof(UElement)*newCap);
+        if (newElems == nullptr) {
+            // We keep the original contents on the memory failure on realloc or bad minimumCapacity.
+            status = U_MEMORY_ALLOCATION_ERROR;
+            return false;
+        }
+        elements = newElems;
+        capacity = newCap;
+    }
+    return true;
+}
+
+/**
+ * Change the size of this vector as follows: If newSize is smaller,
+ * then truncate the array, possibly deleting held elements for i >=
+ * newSize.  If newSize is larger, grow the array, filling in new
+ * slots with nullptr.
+ */
+void UVector::setSize(int32_t newSize, UErrorCode &status) {
+    if (!ensureCapacity(newSize, status)) {
+        return;
+    }
+    if (newSize > count) {
+        UElement empty;
+        empty.pointer = nullptr;
+        empty.integer = 0;
+        for (int32_t i=count; i=newSize; --i) {
+            removeElementAt(i);
+        }
+    }
+    count = newSize;
+}
+
+/**
+ * Fill in the given array with all elements of this vector.
+ */
+void** UVector::toArray(void** result) const {
+    void** a = result;
+    for (int i=0; i 0) {
+            max = probe;
+        } else {
+            // assert(c <= 0);
+            min = probe + 1;
+        }
+    }
+    for (int32_t i=count; i>min; --i) {
+        elements[i] = elements[i-1];
+    }
+    elements[min] = e;
+    ++count;
+}
+
+/**
+  *  Array sort comparator function.
+  *  Used from UVector::sort()
+  *  Conforms to function signature required for uprv_sortArray().
+  *  This function is essentially just a wrapper, to make a
+  *  UVector style comparator function usable with uprv_sortArray().
+  *
+  *  The context pointer to this function is a pointer back
+  *  (with some extra indirection) to the user supplied comparator.
+  *  
+  */
+static int32_t U_CALLCONV
+sortComparator(const void *context, const void *left, const void *right) {
+    UElementComparator *compare = *static_cast(context);
+    UElement e1 = *static_cast(left);
+    UElement e2 = *static_cast(right);
+    int32_t result = (*compare)(e1, e2);
+    return result;
+}
+
+
+/**
+  *  Array sort comparison function for use from UVector::sorti()
+  *  Compares int32_t vector elements.
+  */
+static int32_t U_CALLCONV
+sortiComparator(const void * /*context */, const void *left, const void *right) {
+    const UElement *e1 = static_cast(left);
+    const UElement *e2 = static_cast(right);
+    int32_t result = e1->integer < e2->integer? -1 :
+                     e1->integer == e2->integer? 0 : 1;
+    return result;
+}
+
+/**
+  * Sort the vector, assuming it contains ints.
+  *     (A more general sort would take a comparison function, but it's
+  *     not clear whether UVector's UElementComparator or
+  *     UComparator from uprv_sortAray would be more appropriate.)
+  */
+void UVector::sorti(UErrorCode &ec) {
+    if (U_SUCCESS(ec)) {
+        uprv_sortArray(elements, count, sizeof(UElement),
+                       sortiComparator, nullptr,  false, &ec);
+    }
+}
+
+
+/**
+ *  Sort with a user supplied comparator.
+ *
+ *    The comparator function handling is confusing because the function type
+ *    for UVector  (as defined for sortedInsert()) is different from the signature
+ *    required by uprv_sortArray().  This is handled by passing the
+ *    the UVector sort function pointer via the context pointer to a
+ *    sortArray() comparator function, which can then call back to
+ *    the original user function.
+ *
+ *    An additional twist is that it's not safe to pass a pointer-to-function
+ *    as  a (void *) data pointer, so instead we pass a (data) pointer to a
+ *    pointer-to-function variable.
+ */
+void UVector::sort(UElementComparator *compare, UErrorCode &ec) {
+    if (U_SUCCESS(ec)) {
+        uprv_sortArray(elements, count, sizeof(UElement),
+                       sortComparator, &compare, false, &ec);
+    }
+}
+
+
+/**
+ *  Stable sort with a user supplied comparator of type UComparator.
+ */
+void UVector::sortWithUComparator(UComparator *compare, const void *context, UErrorCode &ec) {
+    if (U_SUCCESS(ec)) {
+        uprv_sortArray(elements, count, sizeof(UElement),
+                       compare, context, true, &ec);
+    }
+}
+
+U_NAMESPACE_END
+
diff --git a/intl/icu/source/common/uvector.h b/intl/icu/source/common/uvector.h
new file mode 100644
index 0000000000..1b2a58da86
--- /dev/null
+++ b/intl/icu/source/common/uvector.h
@@ -0,0 +1,386 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+*   Copyright (C) 1999-2016, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*   Date        Name        Description
+*   10/22/99    alan        Creation.  This is an internal header.
+*                           It should not be exported.
+**********************************************************************
+*/
+
+#ifndef UVECTOR_H
+#define UVECTOR_H
+
+#include "unicode/utypes.h"
+#include "unicode/uobject.h"
+#include "cmemory.h"
+#include "uarrsort.h"
+#include "uelement.h"
+
+U_NAMESPACE_BEGIN
+
+/**
+ * Ultralightweight C++ implementation of a `void*` vector
+ * that is (mostly) compatible with java.util.Vector.
+ *
+ * This is a very simple implementation, written to satisfy an
+ * immediate porting need.  As such, it is not completely fleshed out,
+ * and it aims for simplicity and conformity.  Nonetheless, it serves
+ * its purpose (porting code from java that uses java.util.Vector)
+ * well, and it could be easily made into a more robust vector class.
+ *
+ * *Design notes*
+ *
+ * There is index bounds checking, but little is done about it.  If
+ * indices are out of bounds, either nothing happens, or zero is
+ * returned.  We *do* avoid indexing off into the weeds.
+ *
+ * Since we don't have garbage collection, UVector was given the
+ * option to *own* its contents.  To employ this, set a deleter
+ * function.  The deleter is called on a `void *` pointer when that
+ * pointer is released by the vector, either when the vector itself is
+ * destructed, or when a call to `setElementAt()` overwrites an element,
+ * or when a call to remove()` or one of its variants explicitly
+ * removes an element.  If no deleter is set, or the deleter is set to
+ * zero, then it is assumed that the caller will delete elements as
+ * needed.
+ *
+ * *Error Handling* Functions that can fail, from out of memory conditions
+ * for example, include a UErrorCode parameter. Any function called
+ * with an error code already indicating a failure will not modify the
+ * vector in any way.
+ *
+ * For vectors that have a deleter function, any failure in inserting
+ * an element into the vector will instead delete the element that
+ * could not be adopted. This simplifies object ownership
+ * management around calls to `addElement()` and `insertElementAt()`;
+ * error or no, the function always takes ownership of an incoming object
+ * from the caller.
+ *
+ * In order to implement methods such as `contains()` and `indexOf()`,
+ * UVector needs a way to compare objects for equality.  To do so, it
+ * uses a comparison function, or "comparer."  If the comparer is not
+ * set, or is set to zero, then all such methods will act as if the
+ * vector contains no element.  That is, indexOf() will always return
+ * -1, contains() will always return false, etc.
+ *
+ * 
To do
+ *
+ * 
Improve the handling of index out of bounds errors.
+ *
+ * @author Alan Liu
+ */
+class U_COMMON_API UVector : public UObject {
+    // NOTE: UVector uses the UElement (union of void* and int32_t) as
+    // its basic storage type.  It uses UElementsAreEqual as its
+    // comparison function.  It uses UObjectDeleter as its deleter
+    // function.  This allows sharing of support functions with UHashtable.
+
+private:
+    int32_t count = 0;
+
+    int32_t capacity = 0;
+
+    UElement* elements = nullptr;
+
+    UObjectDeleter *deleter = nullptr;
+
+    UElementsAreEqual *comparer = nullptr;
+
+public:
+    UVector(UErrorCode &status);
+
+    UVector(int32_t initialCapacity, UErrorCode &status);
+
+    UVector(UObjectDeleter *d, UElementsAreEqual *c, UErrorCode &status);
+
+    UVector(UObjectDeleter *d, UElementsAreEqual *c, int32_t initialCapacity, UErrorCode &status);
+
+    virtual ~UVector();
+
+    /**
+     * Assign this object to another (make this a copy of 'other').
+     * Use the 'assign' function to assign each element.
+     */
+    void assign(const UVector& other, UElementAssigner *assign, UErrorCode &ec);
+
+    /**
+     * Compare this vector with another.  They will be considered
+     * equal if they are of the same size and all elements are equal,
+     * as compared using this object's comparer.
+     */
+    bool operator==(const UVector& other) const;
+
+    /**
+     * Equivalent to !operator==()
+     */
+    inline bool operator!=(const UVector& other) const {return !operator==(other);}
+
+    //------------------------------------------------------------
+    // java.util.Vector API
+    //------------------------------------------------------------
+
+    /**
+     * Add an element at the end of the vector.
+     * For use only with vectors that do not adopt their elements, which is to say,
+     * have not set an element deleter function. See `adoptElement()`.
+     */
+    void addElement(void *obj, UErrorCode &status);
+
+    /**
+     * Add an element at the end of the vector.
+     * For use only with vectors that adopt their elements, which is to say,
+     * have set an element deleter function. See `addElement()`.
+     *
+     * If the element cannot be successfully added, it will be deleted. This is
+     * normal ICU _adopt_ behavior - one way or another ownership of the incoming
+     * object is transferred from the caller.
+     *
+     * `addElement()` and `adoptElement()` are separate functions to make it easier
+     * to see what the function is doing at call sites. Having a single combined function,
+     * as in earlier versions of UVector, had proved to be error-prone.
+     */
+    void adoptElement(void *obj, UErrorCode &status);
+
+    void addElement(int32_t elem, UErrorCode &status);
+
+    void setElementAt(void* obj, int32_t index);
+
+    void setElementAt(int32_t elem, int32_t index);
+
+    void insertElementAt(void* obj, int32_t index, UErrorCode &status);
+
+    void insertElementAt(int32_t elem, int32_t index, UErrorCode &status);
+    
+    void* elementAt(int32_t index) const;
+
+    int32_t elementAti(int32_t index) const;
+
+    UBool equals(const UVector &other) const;
+
+    inline void* firstElement() const {return elementAt(0);}
+
+    inline void* lastElement() const {return elementAt(count-1);}
+
+    inline int32_t lastElementi() const {return elementAti(count-1);}
+
+    int32_t indexOf(void* obj, int32_t startIndex = 0) const;
+
+    int32_t indexOf(int32_t obj, int32_t startIndex = 0) const;
+
+    inline UBool contains(void* obj) const {return indexOf(obj) >= 0;}
+
+    inline UBool contains(int32_t obj) const {return indexOf(obj) >= 0;}
+
+    UBool containsAll(const UVector& other) const;
+
+    UBool removeAll(const UVector& other);
+
+    UBool retainAll(const UVector& other);
+
+    void removeElementAt(int32_t index);
+
+    UBool removeElement(void* obj);
+
+    void removeAllElements();
+
+    inline int32_t size() const {return count;}
+
+    inline UBool isEmpty() const {return count == 0;}
+
+    UBool ensureCapacity(int32_t minimumCapacity, UErrorCode &status);
+
+    /**
+     * Change the size of this vector as follows: If newSize is
+     * smaller, then truncate the array, possibly deleting held
+     * elements for i >= newSize.  If newSize is larger, grow the
+     * array, filling in new slots with nullptr.
+     */
+    void setSize(int32_t newSize, UErrorCode &status);
+
+    /**
+     * Fill in the given array with all elements of this vector.
+     */
+    void** toArray(void** result) const;
+
+    //------------------------------------------------------------
+    // New API
+    //------------------------------------------------------------
+
+    UObjectDeleter *setDeleter(UObjectDeleter *d);
+    bool hasDeleter() {return deleter != nullptr;}
+
+    UElementsAreEqual *setComparer(UElementsAreEqual *c);
+
+    inline void* operator[](int32_t index) const {return elementAt(index);}
+
+    /**
+     * Removes the element at the given index from this vector and
+     * transfer ownership of it to the caller.  After this call, the
+     * caller owns the result and must delete it and the vector entry
+     * at 'index' is removed, shifting all subsequent entries back by
+     * one index and shortening the size of the vector by one.  If the
+     * index is out of range or if there is no item at the given index
+     * then 0 is returned and the vector is unchanged.
+     */
+    void* orphanElementAt(int32_t index);
+
+    /**
+     * Returns true if this vector contains none of the elements
+     * of the given vector.
+     * @param other vector to be checked for containment
+     * @return true if the test condition is met
+     */
+    UBool containsNone(const UVector& other) const;
+
+    /**
+     * Insert the given object into this vector at its sorted position
+     * as defined by 'compare'.  The current elements are assumed to
+     * be sorted already.
+     */
+    void sortedInsert(void* obj, UElementComparator *compare, UErrorCode& ec);
+
+    /**
+     * Insert the given integer into this vector at its sorted position
+     * as defined by 'compare'.  The current elements are assumed to
+     * be sorted already.
+     */
+    void sortedInsert(int32_t obj, UElementComparator *compare, UErrorCode& ec);
+
+    /**
+     * Sort the contents of the vector, assuming that the contents of the
+     * vector are of type int32_t.
+     */
+    void sorti(UErrorCode &ec);
+
+    /**
+      * Sort the contents of this vector, using a caller-supplied function
+      * to do the comparisons.  (It's confusing that
+      *  UVector's UElementComparator function is different from the
+      *  UComparator function type defined in uarrsort.h)
+      */
+    void sort(UElementComparator *compare, UErrorCode &ec);
+
+    /**
+     * Stable sort the contents of this vector using a caller-supplied function
+     * of type UComparator to do the comparison.  Provides more flexibility
+     * than UVector::sort() because an additional user parameter can be passed to
+     * the comparison function.
+     */
+    void sortWithUComparator(UComparator *compare, const void *context, UErrorCode &ec);
+
+    /**
+     * ICU "poor man's RTTI", returns a UClassID for this class.
+     */
+    static UClassID U_EXPORT2 getStaticClassID();
+
+    /**
+     * ICU "poor man's RTTI", returns a UClassID for the actual class.
+     */
+    virtual UClassID getDynamicClassID() const override;
+
+private:
+    int32_t indexOf(UElement key, int32_t startIndex = 0, int8_t hint = 0) const;
+
+    void sortedInsert(UElement e, UElementComparator *compare, UErrorCode& ec);
+
+public:
+    // Disallow
+    UVector(const UVector&) = delete;
+
+    // Disallow
+    UVector& operator=(const UVector&) = delete;
+
+};
+
+
+/**
+ * Ultralightweight C++ implementation of a `void*` stack
+ * that is (mostly) compatible with java.util.Stack.  As in java, this
+ * is merely a paper thin layer around UVector.  See the UVector
+ * documentation for further information.
+ *
+ * *Design notes*
+ *
+ * The element at index `n-1` is (of course) the top of the
+ * stack.
+ *
+ * The poorly named `empty()` method doesn't empty the
+ * stack; it determines if the stack is empty.
+ *
+ * @author Alan Liu
+ */
+class U_COMMON_API UStack : public UVector {
+public:
+    UStack(UErrorCode &status);
+
+    UStack(int32_t initialCapacity, UErrorCode &status);
+
+    UStack(UObjectDeleter *d, UElementsAreEqual *c, UErrorCode &status);
+
+    UStack(UObjectDeleter *d, UElementsAreEqual *c, int32_t initialCapacity, UErrorCode &status);
+
+    virtual ~UStack();
+
+    // It's okay not to have a virtual destructor (in UVector)
+    // because UStack has no special cleanup to do.
+
+    inline UBool empty() const {return isEmpty();}
+
+    inline void* peek() const {return lastElement();}
+
+    inline int32_t peeki() const {return lastElementi();}
+    
+    /**
+     * Pop and return an element from the stack.
+     * For stacks with a deleter function, the caller takes ownership
+     * of the popped element.
+     */
+    void* pop();
+    
+    int32_t popi();
+    
+    inline void* push(void* obj, UErrorCode &status) {
+        if (hasDeleter()) {
+            adoptElement(obj, status);
+            return (U_SUCCESS(status)) ? obj : nullptr;
+        } else {
+            addElement(obj, status);
+            return obj;
+        }
+    }
+
+    inline int32_t push(int32_t i, UErrorCode &status) {
+        addElement(i, status);
+        return i;
+    }
+
+    /*
+    If the object o occurs as an item in this stack,
+    this method returns the 1-based distance from the top of the stack.
+    */
+    int32_t search(void* obj) const;
+
+    /**
+     * ICU "poor man's RTTI", returns a UClassID for this class.
+     */
+    static UClassID U_EXPORT2 getStaticClassID();
+
+    /**
+     * ICU "poor man's RTTI", returns a UClassID for the actual class.
+     */
+    virtual UClassID getDynamicClassID() const override;
+
+    // Disallow
+    UStack(const UStack&) = delete;
+
+    // Disallow
+    UStack& operator=(const UStack&) = delete;
+};
+
+U_NAMESPACE_END
+
+#endif
diff --git a/intl/icu/source/common/uvectr32.cpp b/intl/icu/source/common/uvectr32.cpp
new file mode 100644
index 0000000000..fb554729fe
--- /dev/null
+++ b/intl/icu/source/common/uvectr32.cpp
@@ -0,0 +1,335 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+* Copyright (C) 1999-2015, International Business Machines Corporation and
+* others. All Rights Reserved.
+******************************************************************************
+*   Date        Name        Description
+*   10/22/99    alan        Creation.
+**********************************************************************
+*/
+
+#include "uvectr32.h"
+#include "cmemory.h"
+#include "putilimp.h"
+
+U_NAMESPACE_BEGIN
+
+#define DEFAULT_CAPACITY 8
+
+/*
+ * Constants for hinting whether a key is an integer
+ * or a pointer.  If a hint bit is zero, then the associated
+ * token is assumed to be an integer. This is needed for iSeries
+ */
+ 
+UOBJECT_DEFINE_RTTI_IMPLEMENTATION(UVector32)
+
+UVector32::UVector32(UErrorCode &status) :
+    count(0),
+    capacity(0),
+    maxCapacity(0),
+    elements(nullptr)
+{
+    _init(DEFAULT_CAPACITY, status);
+}
+
+UVector32::UVector32(int32_t initialCapacity, UErrorCode &status) :
+    count(0),
+    capacity(0),
+    maxCapacity(0),
+    elements(0)
+{
+    _init(initialCapacity, status);
+}
+
+
+
+void UVector32::_init(int32_t initialCapacity, UErrorCode &status) {
+    // Fix bogus initialCapacity values; avoid malloc(0)
+    if (initialCapacity < 1) {
+        initialCapacity = DEFAULT_CAPACITY;
+    }
+    if (maxCapacity>0 && maxCapacity (int32_t)(INT32_MAX / sizeof(int32_t))) {
+        initialCapacity = uprv_min(DEFAULT_CAPACITY, maxCapacity);
+    }
+    elements = (int32_t *)uprv_malloc(sizeof(int32_t)*initialCapacity);
+    if (elements == 0) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+    } else {
+        capacity = initialCapacity;
+    }
+}
+
+UVector32::~UVector32() {
+    uprv_free(elements);
+    elements = 0;
+}
+
+/**
+ * Assign this object to another (make this a copy of 'other').
+ */
+void UVector32::assign(const UVector32& other, UErrorCode &ec) {
+    if (ensureCapacity(other.count, ec)) {
+        setSize(other.count);
+        for (int32_t i=0; iindex; --i) {
+            elements[i] = elements[i-1];
+        }
+        elements[index] = elem;
+        ++count;
+    }
+    /* else index out of range */
+}
+
+UBool UVector32::containsAll(const UVector32& other) const {
+    for (int32_t i=0; i= 0) {
+            return false;
+        }
+    }
+    return true;
+}
+
+UBool UVector32::removeAll(const UVector32& other) {
+    UBool changed = false;
+    for (int32_t i=0; i= 0) {
+            removeElementAt(j);
+            changed = true;
+        }
+    }
+    return changed;
+}
+
+UBool UVector32::retainAll(const UVector32& other) {
+    UBool changed = false;
+    for (int32_t j=size()-1; j>=0; --j) {
+        int32_t i = other.indexOf(elements[j]);
+        if (i < 0) {
+            removeElementAt(j);
+            changed = true;
+        }
+    }
+    return changed;
+}
+
+void UVector32::removeElementAt(int32_t index) {
+    if (index >= 0) {
+        for (int32_t i=index; icount != other.count) {
+        return false;
+    }
+    for (i=0; i= minimumCapacity) {
+        return true;
+    }
+    if (maxCapacity>0 && minimumCapacity>maxCapacity) {
+        status = U_BUFFER_OVERFLOW_ERROR;
+        return false;
+    }
+    if (capacity > (INT32_MAX - 1) / 2) {  // integer overflow check
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return false;
+    }
+    int32_t newCap = capacity * 2;
+    if (newCap < minimumCapacity) {
+        newCap = minimumCapacity;
+    }
+    if (maxCapacity > 0 && newCap > maxCapacity) {
+        newCap = maxCapacity;
+    }
+    if (newCap > (int32_t)(INT32_MAX / sizeof(int32_t))) {  // integer overflow check
+        // We keep the original memory contents on bad minimumCapacity/maxCapacity.
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return false;
+    }
+    int32_t* newElems = (int32_t *)uprv_realloc(elements, sizeof(int32_t)*newCap);
+    if (newElems == nullptr) {
+        // We keep the original contents on the memory failure on realloc.
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return false;
+    }
+    elements = newElems;
+    capacity = newCap;
+    return true;
+}
+
+void UVector32::setMaxCapacity(int32_t limit) {
+    U_ASSERT(limit >= 0);
+    if (limit < 0) {
+        limit = 0;
+    }
+    if (limit > (int32_t)(INT32_MAX / sizeof(int32_t))) {  // integer overflow check for realloc
+        //  Something is very wrong, don't realloc, leave capacity and maxCapacity unchanged
+        return;
+    }
+    maxCapacity = limit;
+    if (capacity <= maxCapacity || maxCapacity == 0) {
+        // Current capacity is within the new limit.
+        return;
+    }
+    
+    // New maximum capacity is smaller than the current size.
+    // Realloc the storage to the new, smaller size.
+    int32_t* newElems = (int32_t *)uprv_realloc(elements, sizeof(int32_t)*maxCapacity);
+    if (newElems == nullptr) {
+        // Realloc to smaller failed.
+        //   Just keep what we had.  No need to call it a failure.
+        return;
+    }
+    elements = newElems;
+    capacity = maxCapacity;
+    if (count > capacity) {
+        count = capacity;
+    }
+}
+
+/**
+ * Change the size of this vector as follows: If newSize is smaller,
+ * then truncate the array, possibly deleting held elements for i >=
+ * newSize.  If newSize is larger, grow the array, filling in new
+ * slots with nullptr.
+ */
+void UVector32::setSize(int32_t newSize) {
+    int32_t i;
+    if (newSize < 0) {
+        return;
+    }
+    if (newSize > count) {
+        UErrorCode ec = U_ZERO_ERROR;
+        if (!ensureCapacity(newSize, ec)) {
+            return;
+        }
+        for (i=count; i 0) {
+        if (elements[probe] > tok) {
+            max = probe;
+        } else {
+            // assert(c <= 0);
+            min = probe + 1;
+        }
+    }
+    if (ensureCapacity(count + 1, ec)) {
+        for (int32_t i=count; i>min; --i) {
+            elements[i] = elements[i-1];
+        }
+        elements[min] = tok;
+        ++count;
+    }
+}
+
+
+
+
+
+U_NAMESPACE_END
+
diff --git a/intl/icu/source/common/uvectr32.h b/intl/icu/source/common/uvectr32.h
new file mode 100644
index 0000000000..2841391f57
--- /dev/null
+++ b/intl/icu/source/common/uvectr32.h
@@ -0,0 +1,306 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+**********************************************************************
+*   Copyright (C) 1999-2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+**********************************************************************
+*/
+
+//
+//  UVector32 is a class implementing a vector of 32 bit integers.
+//            It is similar to UVector, but holds int32_t values rather than pointers.
+//            Most of the code is unchanged from UVector.
+//
+
+#ifndef UVECTOR32_H
+#define UVECTOR32_H
+
+#include "unicode/utypes.h"
+#include "unicode/uobject.h"
+#include "uhash.h"
+#include "uassert.h"
+
+U_NAMESPACE_BEGIN
+
+
+
+/**
+ * 
Ultralightweight C++ implementation of a void* vector
+ * that is (mostly) compatible with java.util.Vector.
+ *
+ * 
This is a very simple implementation, written to satisfy an
+ * immediate porting need.  As such, it is not completely fleshed out,
+ * and it aims for simplicity and conformity.  Nonetheless, it serves
+ * its purpose (porting code from java that uses java.util.Vector)
+ * well, and it could be easily made into a more robust vector class.
+ *
+ * 
Design notes
+ *
+ * 
There is index bounds checking, but little is done about it.  If
+ * indices are out of bounds, either nothing happens, or zero is
+ * returned.  We do avoid indexing off into the weeds.
+ *
+ * 
There is detection of out of memory, but the handling is very
+ * coarse-grained -- similar to UnicodeString's protocol, but even
+ * coarser.  The class contains one static flag that is set
+ * when any call to new returns zero.  This allows the caller
+ * to use several vectors and make just one check at the end to see if
+ * a memory failure occurred.  This is more efficient than making a
+ * check after each call on each vector when doing many operations on
+ * multiple vectors.  The single static flag works best when memory
+ * failures are infrequent, and when recovery options are limited or
+ * nonexistent.
+ *
+ * 
To do
+ *
+ * 
Improve the handling of index out of bounds errors.
+ *
+ * @author Alan Liu
+ */
+class U_COMMON_API UVector32 : public UObject {
+private:
+    int32_t   count;
+
+    int32_t   capacity;
+    
+    int32_t   maxCapacity;   // Limit beyond which capacity is not permitted to grow.
+
+    int32_t*  elements;
+
+public:
+    UVector32(UErrorCode &status);
+
+    UVector32(int32_t initialCapacity, UErrorCode &status);
+
+    virtual ~UVector32();
+
+    /**
+     * Assign this object to another (make this a copy of 'other').
+     * Use the 'assign' function to assign each element.
+     */
+    void assign(const UVector32& other, UErrorCode &ec);
+
+    /**
+     * Compare this vector with another.  They will be considered
+     * equal if they are of the same size and all elements are equal,
+     * as compared using this object's comparer.
+     */
+    bool operator==(const UVector32& other) const;
+
+    /**
+     * Equivalent to !operator==()
+     */
+    inline bool operator!=(const UVector32& other) const;
+
+    //------------------------------------------------------------
+    // java.util.Vector API
+    //------------------------------------------------------------
+
+    inline void addElement(int32_t elem, UErrorCode &status);
+
+    void setElementAt(int32_t elem, int32_t index);
+
+    void insertElementAt(int32_t elem, int32_t index, UErrorCode &status);
+    
+    inline int32_t elementAti(int32_t index) const;
+
+    UBool equals(const UVector32 &other) const;
+
+    inline int32_t lastElementi() const;
+
+    int32_t indexOf(int32_t elem, int32_t startIndex = 0) const;
+
+    inline UBool contains(int32_t elem) const;
+
+    UBool containsAll(const UVector32& other) const;
+
+    UBool removeAll(const UVector32& other);
+
+    UBool retainAll(const UVector32& other);
+
+    void removeElementAt(int32_t index);
+
+    void removeAllElements();
+
+    inline int32_t size() const;
+
+    inline UBool isEmpty() const;
+
+    // Inline.  Use this one for speedy size check.
+    inline UBool ensureCapacity(int32_t minimumCapacity, UErrorCode &status);
+
+    // Out-of-line, handles actual growth.  Called by ensureCapacity() when necessary.
+    UBool expandCapacity(int32_t minimumCapacity, UErrorCode &status);
+
+    /**
+     * Change the size of this vector as follows: If newSize is
+     * smaller, then truncate the array, possibly deleting held
+     * elements for i >= newSize.  If newSize is larger, grow the
+     * array, filling in new slows with zero.
+     */
+    void setSize(int32_t newSize);
+
+    //------------------------------------------------------------
+    // New API
+    //------------------------------------------------------------
+
+    /**
+     * Returns true if this vector contains none of the elements
+     * of the given vector.
+     * @param other vector to be checked for containment
+     * @return true if the test condition is met
+     */
+    UBool containsNone(const UVector32& other) const;
+
+
+    /**
+     * Insert the given integer into this vector at its sorted position.
+     * The current elements are assumed to be sorted already.
+     */
+    void sortedInsert(int32_t elem, UErrorCode& ec);
+
+    /**
+     * Returns a pointer to the internal array holding the vector.
+     */
+    inline int32_t *getBuffer() const;
+
+    /**
+     * Set the maximum allowed buffer capacity for this vector/stack.
+     * Default with no limit set is unlimited, go until malloc() fails.
+     * A Limit of zero means unlimited capacity.
+     * Units are vector elements (32 bits each), not bytes.
+     */
+    void setMaxCapacity(int32_t limit);
+
+    /**
+     * ICU "poor man's RTTI", returns a UClassID for this class.
+     */
+    static UClassID U_EXPORT2 getStaticClassID();
+
+    /**
+     * ICU "poor man's RTTI", returns a UClassID for the actual class.
+     */
+    virtual UClassID getDynamicClassID() const override;
+
+private:
+    void _init(int32_t initialCapacity, UErrorCode &status);
+
+    // Disallow
+    UVector32(const UVector32&) = delete;
+
+    // Disallow
+    UVector32& operator=(const UVector32&) = delete;
+
+
+    //  API Functions for Stack operations.
+    //  In the original UVector, these were in a separate derived class, UStack.
+    //  Here in UVector32, they are all together.
+public:
+    inline UBool empty() const;   // TODO:  redundant, same as empty().  Remove it?
+
+    inline int32_t peeki() const;
+    
+    inline int32_t popi();
+    
+    inline int32_t push(int32_t i, UErrorCode &status);
+
+    inline int32_t *reserveBlock(int32_t size, UErrorCode &status);
+    inline int32_t *popFrame(int32_t size);
+};
+
+
+// UVector32 inlines
+
+inline UBool UVector32::ensureCapacity(int32_t minimumCapacity, UErrorCode &status) {
+    if ((minimumCapacity >= 0) && (capacity >= minimumCapacity)) {
+        return true;
+    } else {
+        return expandCapacity(minimumCapacity, status);
+    }
+}
+
+inline int32_t UVector32::elementAti(int32_t index) const {
+    return (index >= 0 && count > 0 && count - index > 0) ? elements[index] : 0;
+}
+
+
+inline void UVector32::addElement(int32_t elem, UErrorCode &status) {
+    if (ensureCapacity(count + 1, status)) {
+        elements[count] = elem;
+        count++;
+    }
+}
+
+inline int32_t *UVector32::reserveBlock(int32_t size, UErrorCode &status) {
+    if (ensureCapacity(count+size, status) == false) {
+        return nullptr;
+    }
+    int32_t  *rp = elements+count;
+    count += size;
+    return rp;
+}
+
+inline int32_t *UVector32::popFrame(int32_t size) {
+    U_ASSERT(count >= size);
+    count -= size;
+    if (count < 0) {
+        count = 0;
+    }
+    return elements+count-size;
+}
+
+
+
+inline int32_t UVector32::size() const {
+    return count;
+}
+
+inline UBool UVector32::isEmpty() const {
+    return count == 0;
+}
+
+inline UBool UVector32::contains(int32_t obj) const {
+    return indexOf(obj) >= 0;
+}
+
+inline int32_t UVector32::lastElementi() const {
+    return elementAti(count-1);
+}
+
+inline bool UVector32::operator!=(const UVector32& other) const {
+    return !operator==(other);
+}
+
+inline int32_t *UVector32::getBuffer() const {
+    return elements;
+}
+
+
+// UStack inlines
+
+inline UBool UVector32::empty() const {
+    return isEmpty();
+}
+
+inline int32_t UVector32::peeki() const {
+    return lastElementi();
+}
+
+inline int32_t UVector32::push(int32_t i, UErrorCode &status) {
+    addElement(i, status);
+    return i;
+}
+
+inline int32_t UVector32::popi() {
+    int32_t result = 0;
+    if (count > 0) {
+        count--;
+        result = elements[count];
+    }
+    return result;
+}
+
+U_NAMESPACE_END
+
+#endif
diff --git a/intl/icu/source/common/uvectr64.cpp b/intl/icu/source/common/uvectr64.cpp
new file mode 100644
index 0000000000..05559dd833
--- /dev/null
+++ b/intl/icu/source/common/uvectr64.cpp
@@ -0,0 +1,214 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+******************************************************************************
+* Copyright (C) 1999-2015, International Business Machines Corporation and
+* others. All Rights Reserved.
+******************************************************************************
+*/
+
+#include "uvectr64.h"
+#include "cmemory.h"
+#include "putilimp.h"
+
+U_NAMESPACE_BEGIN
+
+#define DEFAULT_CAPACITY 8
+
+/*
+ * Constants for hinting whether a key is an integer
+ * or a pointer.  If a hint bit is zero, then the associated
+ * token is assumed to be an integer. This is needed for iSeries
+ */
+ 
+UOBJECT_DEFINE_RTTI_IMPLEMENTATION(UVector64)
+
+UVector64::UVector64(UErrorCode &status) :
+    count(0),
+    capacity(0),
+    maxCapacity(0),
+    elements(nullptr)
+{
+    _init(DEFAULT_CAPACITY, status);
+}
+
+UVector64::UVector64(int32_t initialCapacity, UErrorCode &status) :
+    count(0),
+    capacity(0),
+    maxCapacity(0),
+    elements(0)
+{
+    _init(initialCapacity, status);
+}
+
+
+
+void UVector64::_init(int32_t initialCapacity, UErrorCode &status) {
+    // Fix bogus initialCapacity values; avoid malloc(0)
+    if (initialCapacity < 1) {
+        initialCapacity = DEFAULT_CAPACITY;
+    }
+    if (maxCapacity>0 && maxCapacity (int32_t)(INT32_MAX / sizeof(int64_t))) {
+        initialCapacity = uprv_min(DEFAULT_CAPACITY, maxCapacity);
+    }
+    elements = (int64_t *)uprv_malloc(sizeof(int64_t)*initialCapacity);
+    if (elements == 0) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+    } else {
+        capacity = initialCapacity;
+    }
+}
+
+UVector64::~UVector64() {
+    uprv_free(elements);
+    elements = 0;
+}
+
+/**
+ * Assign this object to another (make this a copy of 'other').
+ */
+void UVector64::assign(const UVector64& other, UErrorCode &ec) {
+    if (ensureCapacity(other.count, ec)) {
+        setSize(other.count);
+        for (int32_t i=0; iindex; --i) {
+            elements[i] = elements[i-1];
+        }
+        elements[index] = elem;
+        ++count;
+    }
+    /* else index out of range */
+}
+
+void UVector64::removeAllElements() {
+    count = 0;
+}
+
+UBool UVector64::expandCapacity(int32_t minimumCapacity, UErrorCode &status) {
+    if (U_FAILURE(status)) {
+        return false;
+    }
+    if (minimumCapacity < 0) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return false;
+    }
+    if (capacity >= minimumCapacity) {
+        return true;
+    }
+    if (maxCapacity>0 && minimumCapacity>maxCapacity) {
+        status = U_BUFFER_OVERFLOW_ERROR;
+        return false;
+    }
+    if (capacity > (INT32_MAX - 1) / 2) {  // integer overflow check
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return false;
+    }
+    int32_t newCap = capacity * 2;
+    if (newCap < minimumCapacity) {
+        newCap = minimumCapacity;
+    }
+    if (maxCapacity > 0 && newCap > maxCapacity) {
+        newCap = maxCapacity;
+    }
+    if (newCap > (int32_t)(INT32_MAX / sizeof(int64_t))) {  // integer overflow check
+        // We keep the original memory contents on bad minimumCapacity/maxCapacity.
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return false;
+    }
+    int64_t* newElems = (int64_t *)uprv_realloc(elements, sizeof(int64_t)*newCap);
+    if (newElems == nullptr) {
+        // We keep the original contents on the memory failure on realloc.
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return false;
+    }
+    elements = newElems;
+    capacity = newCap;
+    return true;
+}
+
+void UVector64::setMaxCapacity(int32_t limit) {
+    U_ASSERT(limit >= 0);
+    if (limit < 0) {
+        limit = 0;
+    }
+    if (limit > (int32_t)(INT32_MAX / sizeof(int64_t))) {  // integer overflow check for realloc
+        //  Something is very wrong, don't realloc, leave capacity and maxCapacity unchanged
+        return;
+    }
+    maxCapacity = limit;
+    if (capacity <= maxCapacity || maxCapacity == 0) {
+        // Current capacity is within the new limit.
+        return;
+    }
+    
+    // New maximum capacity is smaller than the current size.
+    // Realloc the storage to the new, smaller size.
+    int64_t* newElems = (int64_t *)uprv_realloc(elements, sizeof(int64_t)*maxCapacity);
+    if (newElems == nullptr) {
+        // Realloc to smaller failed.
+        //   Just keep what we had.  No need to call it a failure.
+        return;
+    }
+    elements = newElems;
+    capacity = maxCapacity;
+    if (count > capacity) {
+        count = capacity;
+    }
+}
+
+/**
+ * Change the size of this vector as follows: If newSize is smaller,
+ * then truncate the array, possibly deleting held elements for i >=
+ * newSize.  If newSize is larger, grow the array, filling in new
+ * slots with nullptr.
+ */
+void UVector64::setSize(int32_t newSize) {
+    int32_t i;
+    if (newSize < 0) {
+        return;
+    }
+    if (newSize > count) {
+        UErrorCode ec = U_ZERO_ERROR;
+        if (!ensureCapacity(newSize, ec)) {
+            return;
+        }
+        for (i=count; iUltralightweight C++ implementation of an int64_t vector
+ * that has a subset of methods from UVector32
+ *
+ * 
This is a very simple implementation, written to satisfy an
+ * immediate porting need.  As such, it is not completely fleshed out,
+ * and it aims for simplicity and conformity.  Nonetheless, it serves
+ * its purpose (porting code from java that uses java.util.Vector)
+ * well, and it could be easily made into a more robust vector class.
+ *
+ * 
Design notes
+ *
+ * 
There is index bounds checking, but little is done about it.  If
+ * indices are out of bounds, either nothing happens, or zero is
+ * returned.  We do avoid indexing off into the weeds.
+ *
+ * 
There is detection of out of memory, but the handling is very
+ * coarse-grained -- similar to UnicodeString's protocol, but even
+ * coarser.  The class contains one static flag that is set
+ * when any call to new returns zero.  This allows the caller
+ * to use several vectors and make just one check at the end to see if
+ * a memory failure occurred.  This is more efficient than making a
+ * check after each call on each vector when doing many operations on
+ * multiple vectors.  The single static flag works best when memory
+ * failures are infrequent, and when recovery options are limited or
+ * nonexistent.
+ *
+ * 
To do
+ *
+ * 
Improve the handling of index out of bounds errors.
+ *
+ */
+class U_COMMON_API UVector64 : public UObject {
+private:
+    int32_t   count;
+
+    int32_t   capacity;
+    
+    int32_t   maxCapacity;   // Limit beyond which capacity is not permitted to grow.
+
+    int64_t*  elements;
+
+public:
+    UVector64(UErrorCode &status);
+
+    UVector64(int32_t initialCapacity, UErrorCode &status);
+
+    virtual ~UVector64();
+
+    /**
+     * Assign this object to another (make this a copy of 'other').
+     * Use the 'assign' function to assign each element.
+     */
+    void assign(const UVector64& other, UErrorCode &ec);
+
+    /**
+     * Compare this vector with another.  They will be considered
+     * equal if they are of the same size and all elements are equal,
+     * as compared using this object's comparer.
+     */
+    bool operator==(const UVector64& other);
+
+    /**
+     * Equivalent to !operator==()
+     */
+    inline bool operator!=(const UVector64& other);
+
+    //------------------------------------------------------------
+    // subset of java.util.Vector API
+    //------------------------------------------------------------
+
+    inline void addElement(int64_t elem, UErrorCode &status);
+
+    void setElementAt(int64_t elem, int32_t index);
+
+    void insertElementAt(int64_t elem, int32_t index, UErrorCode &status);
+    
+    inline int64_t elementAti(int32_t index) const;
+
+    //UBool equals(const UVector64 &other) const;
+
+    inline int64_t lastElementi() const;
+
+    //int32_t indexOf(int64_t elem, int32_t startIndex = 0) const;
+
+    //UBool contains(int64_t elem) const;
+
+    //UBool containsAll(const UVector64& other) const;
+
+    //UBool removeAll(const UVector64& other);
+
+    //UBool retainAll(const UVector64& other);
+
+    //void removeElementAt(int32_t index);
+
+    void removeAllElements();
+
+    inline int32_t size() const;
+
+    inline UBool isEmpty() const { return count == 0; }
+
+    // Inline.  Use this one for speedy size check.
+    inline UBool ensureCapacity(int32_t minimumCapacity, UErrorCode &status);
+
+    // Out-of-line, handles actual growth.  Called by ensureCapacity() when necessary.
+    UBool expandCapacity(int32_t minimumCapacity, UErrorCode &status);
+
+    /**
+     * Change the size of this vector as follows: If newSize is
+     * smaller, then truncate the array, possibly deleting held
+     * elements for i >= newSize.  If newSize is larger, grow the
+     * array, filling in new slows with zero.
+     */
+    void setSize(int32_t newSize);
+
+    //------------------------------------------------------------
+    // New API
+    //------------------------------------------------------------
+
+    //UBool containsNone(const UVector64& other) const;
+
+
+    //void sortedInsert(int64_t elem, UErrorCode& ec);
+
+    /**
+     * Returns a pointer to the internal array holding the vector.
+     */
+    inline int64_t *getBuffer() const;
+
+    /**
+     * Set the maximum allowed buffer capacity for this vector/stack.
+     * Default with no limit set is unlimited, go until malloc() fails.
+     * A Limit of zero means unlimited capacity.
+     * Units are vector elements (64 bits each), not bytes.
+     */
+    void setMaxCapacity(int32_t limit);
+
+    /**
+     * ICU "poor man's RTTI", returns a UClassID for this class.
+     */
+    static UClassID U_EXPORT2 getStaticClassID();
+
+    /**
+     * ICU "poor man's RTTI", returns a UClassID for the actual class.
+     */
+    virtual UClassID getDynamicClassID() const override;
+
+private:
+    void _init(int32_t initialCapacity, UErrorCode &status);
+
+    // Disallow
+    UVector64(const UVector64&) = delete;
+
+    // Disallow
+    UVector64& operator=(const UVector64&) = delete;
+
+
+    //  API Functions for Stack operations.
+    //  In the original UVector, these were in a separate derived class, UStack.
+    //  Here in UVector64, they are all together.
+public:
+    //UBool empty() const;   // TODO:  redundant, same as empty().  Remove it?
+
+    //int64_t peeki() const;
+    
+    inline int64_t popi();
+    
+    inline int64_t push(int64_t i, UErrorCode &status);
+
+    inline int64_t *reserveBlock(int32_t size, UErrorCode &status);
+    inline int64_t *popFrame(int32_t size);
+};
+
+
+// UVector64 inlines
+
+inline UBool UVector64::ensureCapacity(int32_t minimumCapacity, UErrorCode &status) {
+    if ((minimumCapacity >= 0) && (capacity >= minimumCapacity)) {
+        return true;
+    } else {
+        return expandCapacity(minimumCapacity, status);
+    }
+}
+
+inline int64_t UVector64::elementAti(int32_t index) const {
+    return (0 <= index && index < count) ? elements[index] : 0;
+}
+
+
+inline void UVector64::addElement(int64_t elem, UErrorCode &status) {
+    if (ensureCapacity(count + 1, status)) {
+        elements[count] = elem;
+        count++;
+    }
+}
+
+inline int64_t *UVector64::reserveBlock(int32_t size, UErrorCode &status) {
+    if (ensureCapacity(count+size, status) == false) {
+        return nullptr;
+    }
+    int64_t  *rp = elements+count;
+    count += size;
+    return rp;
+}
+
+inline int64_t *UVector64::popFrame(int32_t size) {
+    U_ASSERT(count >= size);
+    count -= size;
+    if (count < 0) {
+        count = 0;
+    }
+    return elements+count-size;
+}
+
+
+
+inline int32_t UVector64::size() const {
+    return count;
+}
+
+inline int64_t UVector64::lastElementi() const {
+    return elementAti(count-1);
+}
+
+inline bool UVector64::operator!=(const UVector64& other) {
+    return !operator==(other);
+}
+
+inline int64_t *UVector64::getBuffer() const {
+    return elements;
+}
+
+
+// UStack inlines
+
+inline int64_t UVector64::push(int64_t i, UErrorCode &status) {
+    addElement(i, status);
+    return i;
+}
+
+inline int64_t UVector64::popi() {
+    int64_t result = 0;
+    if (count > 0) {
+        count--;
+        result = elements[count];
+    }
+    return result;
+}
+
+U_NAMESPACE_END
+
+#endif
diff --git a/intl/icu/source/common/wintz.cpp b/intl/icu/source/common/wintz.cpp
new file mode 100644
index 0000000000..ef9cd68019
--- /dev/null
+++ b/intl/icu/source/common/wintz.cpp
@@ -0,0 +1,331 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+********************************************************************************
+*   Copyright (C) 2005-2015, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+********************************************************************************
+*
+* File WINTZ.CPP
+*
+********************************************************************************
+*/
+
+#include "unicode/utypes.h"
+
+#if U_PLATFORM_USES_ONLY_WIN32_API
+
+#include "wintz.h"
+#include "charstr.h"
+#include "cmemory.h"
+#include "cstring.h"
+
+#include "unicode/ures.h"
+#include "unicode/unistr.h"
+#include "uresimp.h"
+
+#ifndef WIN32_LEAN_AND_MEAN
+#   define WIN32_LEAN_AND_MEAN
+#endif
+#   define VC_EXTRALEAN
+#   define NOUSER
+#   define NOSERVICE
+#   define NOIME
+#   define NOMCX
+#include 
+
+U_NAMESPACE_BEGIN
+
+// Note these constants and the struct are only used when dealing with the fallback path for RDP sessions.
+
+// This is the location of the time zones in the registry on Vista+ systems.
+// See: https://docs.microsoft.com/windows/win32/api/timezoneapi/ns-timezoneapi-dynamic_time_zone_information
+#define WINDOWS_TIMEZONES_REG_KEY_PATH L"SOFTWARE\\Microsoft\\Windows NT\\CurrentVersion\\Time Zones"
+
+// Max length for a registry key is 255. +1 for null.
+// See: https://docs.microsoft.com/windows/win32/sysinfo/registry-element-size-limits
+#define WINDOWS_MAX_REG_KEY_LENGTH 256
+
+#if U_PLATFORM_HAS_WINUWP_API == 0
+
+// This is the layout of the TZI binary value in the registry.
+// See: https://docs.microsoft.com/windows/win32/api/timezoneapi/ns-timezoneapi-time_zone_information
+typedef struct _REG_TZI_FORMAT {
+    LONG Bias;
+    LONG StandardBias;
+    LONG DaylightBias;
+    SYSTEMTIME StandardDate;
+    SYSTEMTIME DaylightDate;
+} REG_TZI_FORMAT;
+
+#endif // U_PLATFORM_HAS_WINUWP_API
+
+/**
+* This is main Windows time zone detection function.
+* 
+* It returns the Windows time zone converted to an ICU time zone as a heap-allocated buffer, or nullptr upon failure.
+*
+* We use the Win32 API GetDynamicTimeZoneInformation (which is available since Vista) to get the current time zone info,
+* as this API returns a non-localized time zone name which can be then mapped to an ICU time zone.
+* 
+* However, in some RDP/terminal services situations, this struct isn't always fully complete, and the TimeZoneKeyName
+* field of the struct might be nullptr. This can happen with some 3rd party RDP clients, and also when using older versions
+* of the RDP protocol, which don't send the newer TimeZoneKeyNamei information and only send the StandardName and DaylightName.
+* 
+* Since these 3rd party clients and older RDP clients only send the pre-Vista time zone information to the server, this means that we 
+* need to fallback on using the pre-Vista methods to determine the time zone. This unfortunately requires examining the registry directly
+* in order to try and determine the current time zone.
+* 
+* Note that this can however still fail in some cases though if the client and server are using different languages, as the StandardName
+* that is sent by client is localized in the client's language. However, we must compare this to the names that are on the server, which
+* are localized in registry using the server's language. Despite that, this is the best we can do.
+* 
+* Note: This fallback method won't work for the UWP version though, as we can't use the registry APIs in UWP.
+* 
+* Once we have the current Windows time zone, then we can then map it to an ICU time zone ID (~ Olsen ID).
+*/
+U_CAPI const char* U_EXPORT2
+uprv_detectWindowsTimeZone()
+{
+    // We first try to obtain the time zone directly by using the TimeZoneKeyName field of the DYNAMIC_TIME_ZONE_INFORMATION struct.
+    DYNAMIC_TIME_ZONE_INFORMATION dynamicTZI;
+    uprv_memset(&dynamicTZI, 0, sizeof(dynamicTZI));
+    SYSTEMTIME systemTimeAllZero;
+    uprv_memset(&systemTimeAllZero, 0, sizeof(systemTimeAllZero));
+
+    if (GetDynamicTimeZoneInformation(&dynamicTZI) == TIME_ZONE_ID_INVALID) {
+        return nullptr;
+    }
+
+    // If the DST setting has been turned off in the Control Panel, then return "Etc/GMT".
+    //
+    // Note: This logic is based on how the Control Panel itself determines if DST is 'off' on Windows.
+    // The code is somewhat convoluted; in a sort of pseudo-code it looks like this:
+    // 
+    //   IF (GetDynamicTimeZoneInformation != TIME_ZONE_ID_INVALID) && (DynamicDaylightTimeDisabled != 0) &&
+    //      (StandardDate == DaylightDate) &&
+    //      (
+    //       (TimeZoneKeyName != Empty && StandardDate == 0) ||
+    //       (TimeZoneKeyName == Empty && StandardDate != 0)
+    //      )
+    //   THEN
+    //     DST setting is "Disabled".
+    //
+    if (dynamicTZI.DynamicDaylightTimeDisabled != 0 &&
+        uprv_memcmp(&dynamicTZI.StandardDate, &dynamicTZI.DaylightDate, sizeof(dynamicTZI.StandardDate)) == 0 &&
+        ((dynamicTZI.TimeZoneKeyName[0] != L'\0' && uprv_memcmp(&dynamicTZI.StandardDate, &systemTimeAllZero, sizeof(systemTimeAllZero)) == 0) ||
+         (dynamicTZI.TimeZoneKeyName[0] == L'\0' && uprv_memcmp(&dynamicTZI.StandardDate, &systemTimeAllZero, sizeof(systemTimeAllZero)) != 0)))
+    {
+        LONG utcOffsetMins = dynamicTZI.Bias;
+        if (utcOffsetMins == 0) {
+            return uprv_strdup("Etc/UTC");
+        }
+
+        // No way to support when DST is turned off and the offset in minutes is not a multiple of 60.
+        if (utcOffsetMins % 60 == 0) {
+            char gmtOffsetTz[11] = {}; // "Etc/GMT+dd" is 11-char long with a terminal null.
+            // Important note on the sign convention for zones:
+            //
+            // From https://en.wikipedia.org/wiki/Tz_database#Area
+            //   "In order to conform with the POSIX style, those zone names beginning with "Etc/GMT" have their sign reversed
+            //   from the standard ISO 8601 convention. In the "Etc" area, zones west of GMT have a positive sign and those
+            //   east have a negative sign in their name (e.g "Etc/GMT-14" is 14 hours ahead of GMT)."
+            //
+            // Regarding the POSIX style, from https://www.gnu.org/software/libc/manual/html_node/TZ-Variable.html
+            //   "The offset specifies the time value you must add to the local time to get a Coordinated Universal Time value."
+            //
+            // However, the Bias value in DYNAMIC_TIME_ZONE_INFORMATION *already* follows the POSIX convention.
+            // 
+            // From https://docs.microsoft.com/en-us/windows/win32/api/timezoneapi/ns-timezoneapi-dynamic_time_zone_information
+            //   "The bias is the difference, in minutes, between Coordinated Universal Time (UTC) and
+            //   local time. All translations between UTC and local time are based on the following formula:
+            //      UTC = local time + bias"
+            //
+            // For example, a time zone that is 3 hours ahead of UTC (UTC+03:00) would have a Bias value of -180, and the
+            // corresponding time zone ID would be "Etc/GMT-3". (So there is no need to negate utcOffsetMins below.)
+            int ret = snprintf(gmtOffsetTz, sizeof(gmtOffsetTz), "Etc/GMT%+ld", utcOffsetMins / 60);
+            if (ret > 0 && ret < UPRV_LENGTHOF(gmtOffsetTz)) {
+                return uprv_strdup(gmtOffsetTz);
+            }
+        }
+    }
+
+    // If DST is NOT disabled, but the TimeZoneKeyName field of the struct is nullptr, then we may be dealing with a
+    // RDP/terminal services session where the 'Time Zone Redirection' feature is enabled. However, either the RDP
+    // client sent the server incomplete info (some 3rd party RDP clients only send the StandardName and  DaylightName,
+    // but do not send the important TimeZoneKeyName), or if the RDP server has not appropriately populated the struct correctly.
+    //
+    // In this case we unfortunately have no choice but to fallback to using the pre-Vista method of determining the
+    // time zone, which requires examining the registry directly.
+    //
+    // Note that this can however still fail though if the client and server are using different languages, as the StandardName
+    // that is sent by client is *localized* in the client's language. However, we must compare this to the names that are
+    // on the server, which are *localized* in registry using the server's language.
+    //
+    // One other note is that this fallback method doesn't work for the UWP version, as we can't use the registry APIs.
+
+    // windowsTimeZoneName will point at timezoneSubKeyName if we had to fallback to using the registry, and we found a match.
+    WCHAR timezoneSubKeyName[WINDOWS_MAX_REG_KEY_LENGTH];
+    WCHAR *windowsTimeZoneName = dynamicTZI.TimeZoneKeyName;
+
+    if (dynamicTZI.TimeZoneKeyName[0] == 0) {
+
+// We can't use the registry APIs in the UWP version.
+#if U_PLATFORM_HAS_WINUWP_API == 1
+        (void)timezoneSubKeyName; // suppress unused variable warnings.
+        return nullptr;
+#else
+        // Open the path to the time zones in the Windows registry.
+        LONG ret;
+        HKEY hKeyAllTimeZones = nullptr;
+        ret = RegOpenKeyExW(HKEY_LOCAL_MACHINE, WINDOWS_TIMEZONES_REG_KEY_PATH, 0, KEY_READ,
+                            reinterpret_cast(&hKeyAllTimeZones));
+        
+        if (ret != ERROR_SUCCESS) {
+            // If we can't open the key, then we can't do much, so fail.
+            return nullptr;
+        }
+
+        // Read the number of subkeys under the time zone registry path.
+        DWORD numTimeZoneSubKeys;
+        ret = RegQueryInfoKeyW(hKeyAllTimeZones, nullptr, nullptr, nullptr, &numTimeZoneSubKeys,
+                               nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr);
+        
+        if (ret != ERROR_SUCCESS) {
+            RegCloseKey(hKeyAllTimeZones);
+            return nullptr;
+        }
+
+        // Examine each of the subkeys to try and find a match for the localized standard name ("Std").
+        //
+        // Note: The name of the time zone subkey itself is not localized, but the "Std" name is localized. This means
+        // that we could fail to find a match if the RDP client and RDP server are using different languages, but unfortunately
+        // there isn't much we can do about it.
+        HKEY hKeyTimeZoneSubKey = nullptr;
+        ULONG registryValueType;
+        WCHAR registryStandardName[WINDOWS_MAX_REG_KEY_LENGTH];
+
+        for (DWORD i = 0; i < numTimeZoneSubKeys; i++) {
+            // Note: RegEnumKeyExW wants the size of the buffer in characters.
+            DWORD size = UPRV_LENGTHOF(timezoneSubKeyName);
+            ret = RegEnumKeyExW(hKeyAllTimeZones, i, timezoneSubKeyName, &size, nullptr, nullptr, nullptr, nullptr);
+
+            if (ret != ERROR_SUCCESS) {
+                RegCloseKey(hKeyAllTimeZones);
+                return nullptr;
+            }
+            
+            ret = RegOpenKeyExW(hKeyAllTimeZones, timezoneSubKeyName, 0, KEY_READ,
+                                reinterpret_cast(&hKeyTimeZoneSubKey));
+            
+            if (ret != ERROR_SUCCESS) {
+                RegCloseKey(hKeyAllTimeZones);
+                return nullptr;
+            }
+
+            // Note: RegQueryValueExW wants the size of the buffer in bytes.
+            size = sizeof(registryStandardName);
+            ret = RegQueryValueExW(hKeyTimeZoneSubKey, L"Std", nullptr, ®istryValueType,
+                                   reinterpret_cast(registryStandardName), &size);
+            
+            if (ret != ERROR_SUCCESS || registryValueType != REG_SZ) {
+                RegCloseKey(hKeyTimeZoneSubKey);
+                RegCloseKey(hKeyAllTimeZones);
+                return nullptr;
+            }
+
+            // Note: wcscmp does an ordinal (byte) comparison.
+            if (wcscmp(reinterpret_cast(registryStandardName), dynamicTZI.StandardName) == 0) {
+                // Since we are comparing the *localized* time zone name, it's possible that some languages might use
+                // the same string for more than one time zone. Thus we need to examine the TZI data in the registry to
+                // compare the GMT offset (the bias), and the DST transition dates, to ensure it's the same time zone
+                // as the currently reported one.
+                REG_TZI_FORMAT registryTziValue;
+                uprv_memset(®istryTziValue, 0, sizeof(registryTziValue));
+
+                // Note: RegQueryValueExW wants the size of the buffer in bytes.
+                DWORD timezoneTziValueSize = sizeof(registryTziValue);
+                ret = RegQueryValueExW(hKeyTimeZoneSubKey, L"TZI", nullptr, ®istryValueType,
+                                     reinterpret_cast(®istryTziValue), &timezoneTziValueSize);
+
+                if (ret == ERROR_SUCCESS) {
+                    if ((dynamicTZI.Bias == registryTziValue.Bias) &&
+                        (memcmp((const void *)&dynamicTZI.StandardDate, (const void *)®istryTziValue.StandardDate, sizeof(SYSTEMTIME)) == 0) &&
+                        (memcmp((const void *)&dynamicTZI.DaylightDate, (const void *)®istryTziValue.DaylightDate, sizeof(SYSTEMTIME)) == 0))
+                    {
+                        // We found a matching time zone.
+                        windowsTimeZoneName = timezoneSubKeyName;
+                        break;
+                    }
+                }
+            }
+            RegCloseKey(hKeyTimeZoneSubKey);
+            hKeyTimeZoneSubKey = nullptr;
+        }
+
+        if (hKeyTimeZoneSubKey != nullptr) {
+            RegCloseKey(hKeyTimeZoneSubKey);
+        }
+        if (hKeyAllTimeZones != nullptr) {
+            RegCloseKey(hKeyAllTimeZones);
+        }
+#endif // U_PLATFORM_HAS_WINUWP_API
+    }
+
+    CharString winTZ;
+    UErrorCode status = U_ZERO_ERROR;
+    winTZ.appendInvariantChars(UnicodeString(true, windowsTimeZoneName, -1), status);
+
+    // Map Windows Timezone name (non-localized) to ICU timezone ID (~ Olson timezone id).
+    StackUResourceBundle winTZBundle;
+    ures_openDirectFillIn(winTZBundle.getAlias(), nullptr, "windowsZones", &status);
+    ures_getByKey(winTZBundle.getAlias(), "mapTimezones", winTZBundle.getAlias(), &status);
+    ures_getByKey(winTZBundle.getAlias(), winTZ.data(), winTZBundle.getAlias(), &status);
+
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+    
+    // Note: Since the ISO 3166 country/region codes are all invariant ASCII chars, we can
+    // directly downcast from wchar_t to do the conversion.
+    // We could call the A version of the GetGeoInfo API, but that would be slightly slower than calling the W API,
+    // as the A version of the API will end up calling MultiByteToWideChar anyways internally.
+    wchar_t regionCodeW[3] = {};
+    char regionCode[3] = {}; // 2 letter ISO 3166 country/region code made entirely of invariant chars.
+    int geoId = GetUserGeoID(GEOCLASS_NATION);
+    int regionCodeLen = GetGeoInfoW(geoId, GEO_ISO2, regionCodeW, UPRV_LENGTHOF(regionCodeW), 0);
+
+    const char16_t *icuTZ16 = nullptr;
+    int32_t tzListLen = 0;
+
+    if (regionCodeLen != 0) {
+        for (int i = 0; i < UPRV_LENGTHOF(regionCodeW); i++) {
+            regionCode[i] = static_cast(regionCodeW[i]);
+        }
+        icuTZ16 = ures_getStringByKey(winTZBundle.getAlias(), regionCode, &tzListLen, &status);
+    }
+    if (regionCodeLen == 0 || U_FAILURE(status)) {
+        // fallback to default "001" (world)
+        status = U_ZERO_ERROR;
+        icuTZ16 = ures_getStringByKey(winTZBundle.getAlias(), "001", &tzListLen, &status);
+    }
+
+    // Note: We want the first entry in the string returned by ures_getStringByKey.
+    // However this string can be a space delimited list of timezones:
+    //  Ex: "America/New_York America/Detroit America/Indiana/Petersburg ..."
+    // We need to stop at the first space, so we pass tzLen (instead of tzListLen) to appendInvariantChars below.
+    int32_t tzLen = 0;
+    if (tzListLen > 0) {
+        while (!(icuTZ16[tzLen] == u'\0' || icuTZ16[tzLen] == u' ')) {
+            tzLen++;
+        }
+    }
+
+    // Note: cloneData returns nullptr if the status is a failure, so this
+    // will return nullptr if the above look-up fails.
+    CharString icuTZStr;
+    return icuTZStr.appendInvariantChars(icuTZ16, tzLen, status).cloneData(status);
+}
+
+U_NAMESPACE_END
+#endif /* U_PLATFORM_USES_ONLY_WIN32_API  */
diff --git a/intl/icu/source/common/wintz.h b/intl/icu/source/common/wintz.h
new file mode 100644
index 0000000000..ce9c1e9019
--- /dev/null
+++ b/intl/icu/source/common/wintz.h
@@ -0,0 +1,36 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+********************************************************************************
+*   Copyright (C) 2005-2011, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+********************************************************************************
+*
+* File WINTZ.H
+*
+********************************************************************************
+*/
+
+#ifndef __WINTZ
+#define __WINTZ
+
+#include "unicode/utypes.h"
+
+#if U_PLATFORM_USES_ONLY_WIN32_API
+
+/**
+ * \file 
+ * \brief C API: Utilities for dealing w/ Windows time zones.
+ */
+
+U_CDECL_BEGIN
+/* Forward declarations for Windows types... */
+typedef struct _TIME_ZONE_INFORMATION TIME_ZONE_INFORMATION;
+U_CDECL_END
+
+U_CAPI const char* U_EXPORT2
+uprv_detectWindowsTimeZone();
+
+#endif /* U_PLATFORM_USES_ONLY_WIN32_API  */
+
+#endif /* __WINTZ */
-- 
cgit v1.2.3